Drug Discovery and Design: Medical Aspects (Biomedical and Health Research, 55)

196

G.A. Spyroulias et al. / Conformational Analysis of Biomolecules in Solution

principal neuroregulator of the secretion of adrenocorticotropic hormone (ACTH), 0endorphin and other proopiomelanocortin products of the anterior pituitary gland considered now as the key-hormone controlling hypothalamic-pituitary-adrenal function [23]. The CRH acts at the nanomolar level by interacting with a membrane receptor belonging to the G protein-coupled family of receptors. CRH binds to membrane homogenates from the anterior pituitary, revealing a high-affinity component of 0.1 nM and a low-affinity binding site at 20nM [24]. The primary sequence of various organisms is presented below in Chart 2. DOG HUMAN RAT PIG SHEEP

5

10

15

SEEPP SEEPP SEEPP SEEPP SQEPP

ISLDL ISLDL ISLDL ISLDL ISLDL

TFHLL TFHLL TFHLL TFHLL TFHLL

20

REVLE REVLE REVLE REVLE REVLE

25

MPGAE MARAE MARAE MARAE MTKAD

30

QLAQQ QLAQQ QLAQQ QLAQQ QLAQQ

35

AHSNR AHSNR AHSNR AHSNR AHSNR

40

KLMEI KLMEI KLMEI KLMEI KLLDI

I I I I A

Chart 2. Sequence alignment of various Corticotropin Realing Hormones. Residues that are different in the sequences of other than human CRH are noted with different character format

The analogue of the human CRH studied is a synthetic polypeptide where in positions 12 and 15 of the sequence the native amino acids L-Phenylalanine and Leucine has been substituted by a D-Phenylalanine and by the a-aminoisobutyric acid (Aib), respectively. The CRH analogue was synthesized manually by solid phase Fmoc chemistry protocols [25] using 2-chlorotrityl resin [26] as solid support and 1-hydroxybenzotriazole (HOBt), N,N'-diisopropyl-carbodiimide (DIC) as activators of carboxylic residues [27]. The peptide was cleaved from the solid support by a TFA/DCM mixture using as ion scavengers anisol. 1,2- ethandithiol and water. The purification and characterization of peptide was achieved with an AKTA purifier RP-HPLC system and ESI-MS. Purity of CRH analogue was greater than 95%. Sample preparation. Deuterated trifluoroethanols, TFE-d2 and TFE-d3 (CORTEC, Paris, France), were used in mixtures with H2O and D2O, respectively. The peptide was dissolved to a concentration 2-2.5 mM in H2O/TFE-d2 (34%/66% v/v), while the pH was adjusted to 3.8. For amide exchange experiments, the peptide was lyophilized and redissolved in D2O/TFE-d3 (34%/66% v/v). NMR experiments carried out in 298K and 31 OK. NMR spectroscopy. NMR spectra were recorded on a Bruker 500, 600 and 700 MHz spectrometers. 1D spectra on 12-14 ppm spectral width were acquired using either the "presaturation" (presaturation during the relaxation delay and the mixing time) [28] or the "watergate" (gradient tailored excitation) [29] pulse sequence for the elimination of the H2O signal (Figure 4).

G.A. Spyroulias et al. / Conformational Analysis of Biomolecules in Solution

197

8.3O19/22" 25/22

9.00

8.50

8.00

a)

7.50

7.OO

5

ppm

Figure 4. Regions (A) 1H-1H 2D TOCSY 600 MHz NMR of the fingerprint region of NH-Ha protons. (B) 1H-1H 2D NOESY 600 MHz NMR of the NH-NH protons. Both spectra recorded on a 10.7 ppm spectral width (H2O/TFE-d2 34%/66% v/v, at pH=3.8, T= 310 K).

DQF-COSY [7] and TOCSY (using MLEV-17 spin-lock sequence, mixing time of 80-100 ms, 2K data points in F2 dimension, 16-32 transients and 1024 complex increments in the Fl dimension) [5] experiments were performed in order to facilitate the identification of the spin-systems of each individual aminoacid. TPPI NOESY [6] spectra were recorded using mixing times of 100 ms and 200 ms. The other acquisition parameters were the same as for the 1D spectra (pulse width - 90°- 11.28 us, relaxation delay 1.0-1.5 s and acquisition time 157.794 ms). The spectra consisted in 2K data points in the F2 dimension, 16-32 transients and 720-1024 complex increments in the Fl dimension. All 2D maps were acquired with a spectral width of 6410.26 Hz (10.7 ppm). Raw data were multiplied in both dimensions by a pure cosine-squared bell window function and Fourier-transformed to obtain 2048x1024 or 2048x2048 real data points. A polynomial base-line correction was applied in both directions. NMR data processing was performed using the standard Bruker software package and 2D maps were analyzed on a Silicon Graphics O2 workstation or on PentiumIII PC-Linux computers with the aid of the program XEASY (ETH, Zurich) [30].

198

G.A. Spyroulias et al. / Conformational Analysis of Biomolecules in Solution

Identification of slowly exchanging amide hydrogens. Slowly exchanging amides were identified after 4-5 hours in D2O/TFE-d3 and acquisition of NOESY where nonexchangeable amide protons are giving rise to detectable NH resonances after that period. 16 hydrogen bonds (13 in 20-41 region) involving non-exchangable NH amide protons, are occured to the final set of structures (with occurance > 70%) composing the family of 40 structures were used as structural constraints and introduced in the final structural calculation. Exchangeability of amide NHs is illustrated at the top of the Figure 5. NMR Constraints. The volumes of the 2D NOESY cross peaks between assigned resonances were obtained by manual integration. 1868 dipolar connectivities concerning sequential, short medium range connectivities Figure 6. The total number of unique NOESY cross peaks was 928 and the intensities were converted into upper limits of interatomic distances by following the methodology of the program CALIBA [31] and cross peaks of NOESY spectra were classified into five categories [11]. Appropriate pseudoatom corrections were applied to methylene and methyl hydrogens that were not stereospecifically assigned [32]. Determination of scalar coupling constants. 3JHNHa were determined using the method of D. S. Wishart [15] for measuring 3JHNHa coupling constants from NOESY spectra. 21 out of 28 measured coupling constants, were measured in the range of 4.8-5.6 Hz, equal or above 8.0 [33] and introduced to the calculations (Figure 5). Seven 3JHNHa were calculated in the range 6 to 7 Hz suggesting conformational avergaging and have not been used in structural calculations [11]. Structure Calculations and Refinement. A total of 1868 NOESY cross-peaks (90% derived from the NOESY map acquired with the "PRESATURATION" pulse sequence) were integrated and transformed into 928 unique upper distance limits, corresponding to almost 23.1 constraints per assigned residue (no resonances for Ser1 has been assigned) (Figure 6). 656 constraints (16,4 per assigned residue) were found meaningfull and used in structural calculations with DYANA program [34]. The Aib (a-aminoisobutyrilo) residue was generated from a Valine where the Ca. of the Aib corresponds to CB of a Valine and, CB1/CB2 of the Aib corresponds to Cy1/Cy2, respectively, of a Valine. DPhenylalanine is generated by modifying the coordinates in one axis of an LPhenylalanine. 14 pairs of diastereotopic protons (among them the two methyl groups of Aib15) have been stereospecifically assigned through the program GLOMSA [31]. The 40 best structures (out of 400 calculated) in terms of target function with NOE constraint violations below 0.2 A (larger violation 0.09 A) are included to the final familly. Each one of the 40 DYANA models is refined by Restrained Energy Minimization (REM) through programs included to AMBER 5.0 Package [35]. The whole proceedure is set up following protocols developed and applied on solution structure determination of larger biomolecules, such as metalloproteins [36]. A force constant of 133.76 kJ mol-1 A2 is applied for the distance constraints. Structural calculations have been performed on IBM RISC6000 or on PentiumIII PC-Linux computers.

199

G.A. Spyroulias et al. / Conformational Analysis ofBiomolecules in Solution

Quality Assessment of the Structure. The structure quality was analyzed in terms of backbone conformation through the PROCHECK program [22], which provides the elements of the secondary structure, Ramanchandran plot, and other statistical parameters. Secondary structure. Figure 5 shows the short and medium range NOEs observed for the backbone and ft protons in the NOESY maps. Sequential NH-NH connectivities for stretches up to two aminoacids were observed for residues 6-13, 14-19, 20-28, 29-31, and 31–40. Two prolines (in 4 and 5 position) are present in the sequence and interrupt the sequential connectivities. Strong NH-NH and medium range Ha-NH(i, i+3), Ha-Hp (i, i+3) and Ha-NH(i, i+4) NOEs, have been observed throughout the region from Pro5 to Ile40 indicating the extended a-helical structure of the CRH analogue. (Figure 4). 2

Figure 5. (A) Schematic representation of the sequential and medium-range NOE connectivities involving NH, Ha and H3 protons. Thickness of the bar indicates the intensity of NOEs. 3JHNHa coupling constants are illustrated by arrows ( for values in the range 4.2 -5.6, for values equal or above 8.0 Hz) or by filled circles (values in the range of 6.0-7.0 Hz) The average, over the 40 DYANA structures, secondary structure elements are also reported.

Valuable information for the secondary structure of a polypeptide chain could, also, be provided by the spin-spin coupling constants which in most cases are found typical for a- and 03/10 helical structures since experimental 3J values are measured in the range of 4.8 to 5.6 [33] somewhat larger than the ideal values of 3.9 and 4.2 Hz, respectively. In Figure 5 measured V values for 28 residues are illudstrated. From those, 18 3JHNHa are found to be equal or less than 5.6 Hz (range of 4.0 to 5.6 Hz) while 3 (Glu2, Glu3 and Val18) are found to be in the range of 8.0 to 8.5 Hz, one (DPhel2) is measured 6.9 Hz and 6 (Leu10, Thr11, Leu19, Gln26, Ile39 and Ile40) exhibited values in the range of 6.0 to 6.6 Hz. Tertiary structure. The resulting DYANA family of 40 structures has RMSD values (for residues 6-41) of 0.56 ± 0.24 A and 1.40 ± 0.20 A for backbone and heavy atoms, respectively. The target function lies in the range 0.10-0.12 A2 (average target function 0.11 ± 0.01 A 2 ). The final family resulting from REM refinment performed using both NOEs and Hydrogen bond constraints exhibits pairwise RMSD values for the 40

200

G.A. Spyroulias et al. / Confortnational Analysis of Biomolecules in Solution

structures 0.56 ± 0.23 A and 1.42 ± 0.19 A and 0.39 ± 0.16 A and 0.99 ± 0.13 A to the mean structure for backbone and heavy atoms, respectively. The RMSD per residue for the final family and the NOE-derived distance constraints are reported in Figure 6. Figure 7 shows the family of 40 structures for all heavy atoms (A) and for the backbone atoms (B). The polypeptide chains seems to fold to a rather linear helical structure. Features, in terms of NOEs, of the helix has been observed for the residue fragment from Ile6 to Ile41, while for the mobile, ill-defined N-terminus data (chemical shift index, 3J/HNHa, and 9 versus ^ plots) imply a conformational preference to extended, p sheet, secondary structure.

Figure 6. Number of all NOE constraints per residue (top) and number of meaningful NOE constraints (middle) per residue for [DPhe12, Aibl5]CRH used in the structural calculations. White, gray, and dark gray bars represent respectively intraresidue, sequential, and medium-range, connectivities. Long-range connectivities have not been observed. RMSD per residue (bottom) for the family of 40 structures for backbone (dark gray bars) and heavy atoms (light gray bars).

Beyond the structure : Biological Implications. A noteworthy characteristic of the [DPhe12, Aib15]CRH is the bend of the helical structure at the region 12-16 which is not observed for the native peptide [37]. This bend is supported by numerous NOEs at the 10-20 region and this distortion out of linearity calculated in the basis of helix dipole for

G.A. Spyroulias et al. / Conformational Analysis of Biomolecules in Solution

201

the peptide fragments 6 to 16 and 20 to 40 and found to be 34.2°. This is probably the result of the presence of a D- and an a,a-dialkyl aminoacids in close proximity in a helical polypeptide chain. Another interesting feature of this region of the threedimensional structure of the molecule, is that the chain at the region 12-15 folds in a helical structure in such a way that the turn formed comprised by three instead of four residues bringing the aromatic ring of the DPhel2 and the one methyl group of Aibl5 in close proximity while the subsequent four-residue turn in the region 15-19 allow the other methyl group of AiblS to approach the main- and side-chain of Leu 18 (Figure 7). Serl

Ile41

N terminus

terminus

Aibl5

Met21

Met38

Figure 7. The family of 40 structures calculated for the [DPhe12, Aib15]CRH analogue in line (all heavy atoms) and in saucage (backbone) representation. Figure was generated with MOLMOL [38].

The number, nature and density of charged residues in the surface of proteins or biomolecules in general, is of crucial importance, since electrostastic interactions possess a central role in numerous biological processes. The knoweledge of the nature and the strength of such intramolecular interactions requires the detailed analysis of the structural features of the biomolecules. In the synthetic CRH-analogue, as well as in the native polypeptide, there are eleven charged residues without taken into account the NH+3 group of the N-terminus. Seven, out of the eleven, residues are negatively charged while only four are positively charged resulting to a net negative charge of the molecule. Analysis of the charge distribution along the main axis of the almost linear and helical, peptide indicates that the 20-peptide of the N-terminus bears 5 negatively and 1 positively charged residues, while the second half of the molecule towards the C-

202

G.A. Spyroulias et al. / Conformational Analysis of Biomolecules in Solution

terminus exhibits a slight excess in positively charged residues (two Arg+ and one Lys+ versus two Glu-) giving the character of a dipole to the molecule with the negative pole at the N-terminus and a rather positive pole at the C-terminus. Calculation of the distribution of the electrostatic potential on the surface of the molecule reveals the amphipathic character of the helix and the non-equivalent distribution of the charges in each side of the helix. The five out of six negatively charged Glutamates existing in the sequence of CRH are oriented in a different way than the sixth, Glu20, residue. The three of them are clearly oriented in the opposite direction than that of Glu20 while the two others are found in the intermediate space between the two extreme orientations manifested by Glu20 on one side and by Glu3, 17 and 25 on the other, exhibiting though a tendecy to orient towards the side of the three Glutamates. This density of negative charges is slightly diminished by the presence of two consecutive, positively charged, residues at the C-terminus, Arg35 and Lys36, where the lysyl residue extent its long side chain bearing the amino group, to the same direction as the three Glutamates do. Arg35 exhibits rather the same orientation as that of its side chain which ends to a positively charged group and orients in direction that differs about 90° with that of the side chains of Glu3, 17, 25 and Lys36. On the other hand there are three additional residues that orients towards or similar to the Glu20. From these the two positevely charged arginines at positions 16 and 20 extent their side chains far from the helix axis, to the solvent, exactly on the opposite side form the position of the three glutamates (Glu3, 17, 25) and Lys36. Concluding Remarks The detailed conformational study of a represenative synthetic analogue of Corticotropin Releasing Hormone led to the solution structure determination of [D-Phe12 , Aib'5]CRH in high resolution. Detailed analysis of the structural features attempts to gain valuable insights of the impact of side-chains belong to residues foreseen to possess a key-role in molecular recognition process, binding and activity properties. Substitution of the natural amino acids at a close proximity to the peptide fragment 12-15 by non natural residues, seems to introduce a bend across the molecule which has not been observed in the solution structure of the native peptide. Although hard to quantitatively evaluate the outcome of the new structural features observed for each of the two new residues introduced to the polypeptide chain at positions 12 (DPhe) and 15 (Aib), the experimental results indicate that the DPhe seems to destabilize the helical secondary structury while Aib seems to lead to a more constrained helix with features between the a- and the 0 3 /10 helices. The distribution of the electrostatic potentials on the surface of the peptide could provide valuable information not only for the properties of the molecule and the intramolecular interaction with the CRH receptor, but also towards the design and the synthesis of new molecules with higher binding activity and possibly enhanced stability. Since it is reported that the side chains of the lipophilic part of the helix at the peptide fragment 6 to 20 are playing crucial role to binding and activation of the receptor, increase of the lipophilic character of this region would be designed with the extinction of the negative or positive charge density, or both. Rational design of new CRH analogues through substitution of natural residues by non natural is being carried out at our laboratory and synthesis of new peptides bearing residues that increase the action period of the molecule, and increase the binding and activity of the synthetic peptides are under way.

G.A. Spyroulias et al. / Conformational Analysis of Biomolecules in Solution

203

Acknowledgments P.C. acknowledges the financial support of General Secretariat of Research and Technology of Greece through the EKVAN (EPET II) Programme (Project 66, 19982001). G.A.S. wishes to acknowledge a European Union HP Fellowship (Marie Curie Research Training Grant, Contract No. HPMF-CT-1999-00344). We, also, wish to acknowledge European Community's Programme Access to Research Infrastructures Action of the Improving Human Potential Programme Contract No. HPRI-CT-199900009 and the NMR-PARABIO LSF at University of Florence for the provided access to NMR Instrumentation (500, 600 and 700 MHz), as well as for the access to powerful computational systems in order to run structural calculations. References [I] F.C. Bernstein, T.F. Koetzle, G.J.B. Williams, E.F. Meyer, M.D. Brice, J.R. Rodgers, O. Kennard, T. Simanoucghi, J. Mol. Biol, 112 (1977) 535. [2] G. Wider, K. Wiithrich, Curr. Opin. Struct. Biol., 9 (1999) 594. [3] R.R. Ernst, G. Bodenhausen, A. Wokaun, In: R. Breslow, J. Halpern Frs, J.S. Rowlinson Frs (eds.),Principles of Nuclear Magnetic Resonance in One and Two Dimensions. Clarendon Press, Oxford, 1987. [4] (a) J. Jeener, In: Ampere International Summer School. Basko Polje, Yugoslavia (1971); (b) W.P. Aue, E. Bartholdi, R. R. Ernst, J. Chen Phys. 64 (1975) 2229. [5] A. Bax, D.G. Davis, J Magn Reson 65 (1985) 355. [6] (a) D. Marion, K. Wiithrich, Biochem. Biophys. Res. Commun., 113 (1983) 967; (b) J. Jeener; B.H. Meier; P. Badimann; R. R. Ernst, J. Chem. Phys., 71 (1979) 4546. [7] U. Piantini; O. W. Sorensen; R. R. Ernst, J. Am. Chem. Soc., 104 (1982) 6800. [8] D. Wishart, B. Sykes, F. Richards, J. Mol. Biol., 222 (1991) 311. [9] D. Wishart, B. Sykes, F. Richards, Biochemistry, 31 (1992) 1647. [10] W. J. Chazin, P. E. Wright,J Mol. Biol., 202 (1988) 623. II1] K. Wuthrich, In: NMR of Proteins and Nucleic Acids. Wiley, New York, 1986. [12] K. Wuthrich, Biopolymers, 22 (1982) 131. [13] M. Billeter, W. Braun, K. Wuthrich, J. Mol. Biol, 155 (1982) 321. [14] T. Havel, In: V. Renugopalakrishnan, P.R. Carey, I.C.P. Smith, S. Huans and A.L. Storer (eds.), ESCOM Science Publishers, Leiden, Holland. 1991, pp. 110–115. [15] Y. Wang, A. M. Nip, D. S. Wishart, J. Biomol. NMR, 10 (1997) 373. [16] G.W. Vuister, A. Bax, J. Amer. Chem. Soc., 115 (1993) 7772. [17] M. Karplus, J. Phys. Chem. 30 (1959) 11. [18] L.J. Smith, K.A. Bolin, H. Schwalbe, M.W. MacArthur, J.M. Thorton, C.M. Dobson, J. Mol. Biol., 255(1996)494. [19] G.M. Crippen, T.F. Havel, In: Distance Geometry and Molecular Conformation. Research Studies Press, Taunton, UK, 1988. [20] (a) A.T. Brunger, G.M. Clore, A.M. Gronenborn, M. Karplus, Proc. Natl. Acad. Sci. (USA), 83 (1986) 3801; (b) G.M. Clore, A.T. Brunger, M. Karplus, A.M. Gronenborn, J. Mol. Biol, 191 (1986) 523. [21] B. Borgias, P.D. Thomas, T.L. James, Complete Relaxation Matrix Analysis (CORMA). U of California, San Francisco, USA 1989. [22] (a) R.A. Laskowski, M.W. MacArthur, D.S. Moss, J.M. Thornton, J. Appl. Crystallogr., 26 (1993) 283; (b) R.A. Laskowski, J.A.C. Rullmann, M.W. MacArthur, R. Kaptein, J.M. Thornton, J. Biomol. NMR, 8(1996)477. [23] (a) W. Vale, J. Spiess, C. Rivier, J. Rivier, Science, 213 (1981) 1394; (b) W. Vale, C. Rivier, M.R. Brown. J. Spiess, G. Koob, A. Swanson, Recent Prog. Horm. Res., 39 (1983) 245. [24] E.B. De Souza, T.R. Insel, M.H. Perrin, J. Rivier, W.W. Vale, M.J. Kuhar, Neurosci. Lett., 56 (1985) 121.

204

G.A. Spyroulias et al. / Conformational Analysis of Biomolecules in Solution

[25] (a) L.A. Carpino, G.Y. Han, J. Org. Chem., 37 (1972) 3404; (b) D. Sarantakis, J. Teichman, E.L. Lien, R.L. Fenichel, Biochem. Biophys. Res. Commun., 73 (1976) 366. [26] K. Barlos, D. Gatos, J. Kallitsis, G. Papaphotiou, P. Sotiriou, Y. Wenquing, W. Shafer, Tetraedron Lett., 30 (1989) 3943. [27] (a) D. Hudson, D. Kain, D. Ng, In: Y. Kiso (ed.), Peptide Chemistry. Protein Research Foundation, Osaka, 1986, pp. 4113; (b) W. Konig, R. Geiger, Chem. Ber., 103 (1970) 788; (c) Konig, R. Geiger, Chem. Ber., 103 (1970) 2024; (d) Konig, R. Geiger, Chem. Ber., 103 (1970) 2034. [28] S. Macura, K. Wuthrich, R.R. Ernst, J Magn. Reson., 47 (1982) 351. [29] M. Piotto, V. Saudek, V. Sklenar, J. Biomol. NMR. 2 (1992) 661. [30] C. Eccles, P. Guntert, M. Billeter, K. Wuthrich,J. Biomol. NMR, 1 (1991) 111. [31] P. Guntert, W. Braun, K. Wuthrich, J. Mol. BioL. 217 (1991) 517. [32] K. Wuthrich, M. Billeter, W. Brown, / Mol. Biol. 169 (1983) 949. [33] L.J. Smith, K.A. Bolin, H. Schwalbe, M.W. MacArthur, J.M. Thorton, C.M. Dobson, J. Mol. Biol., 255(1996)494. [34] P. Guntert, C. Mumenthaler, K. Wuthrich, J. Mol. Biol, 273 (1997) 283. [35] D.A. Pearlman, D.A. Case, J.W. Caldwell, W.S. Ross, T.E. Cheatham, D.M. Ferguson, G.L. Seibel, U.C. Singh, P.K. Weiner, P. A. Kollman, AMBER 5.0. U of California, San Francisco, USA. 1997. [36] L. Banci, I. Bertini, C. Luchinat, P. Turano, Solution structures of hemoproteins. In: K.M. Kadish, K.M. Smith, R. Guilard, (eds.), The Porphyrin Handbook. Academic Press, 2000, Volume 5, pp 323– 350. [37] C. Romier, J.-M. Bernassau, C. Cambillau, H. Darbon, Protein Eng., 6 (1993) 149. [38] R. Koradi, M. Billeter, K. Wuthrich, J. Mol. Graphics, 14 (1996) 51.

Drug Discovery and Design: Medical Aspects J, Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

205

Synthesis and Structural Investigation of a Synthetic LHRH Analogue in Solution A. A. Zompra, G. A. Spyroulias, V. Magafa and P. Cordopatis Department of Pharmacy, University of Patras, 26504 Patras, Greece

Abstract. The pivotal role that LHRH and its analogues play in the modulation of reproductive functions have attracted considerable scientific interest because of their usefulness in the treatment of endocrine-based diseases such as prostate cancer, breast cancer, endometriosis and precocious puberty. Several LHRH agonists, represented by Leuprolide ([DLeu6, desGly10]-LHRHethylamide), are currently used in the treatment of the above conditions. The goal of this research is the synthesis and conformational analysis of a LHRH analogue, such as [Aib6, desGly10]-LHRH-ethylamide in order to gain valuable insights on bioactive conformation and use these for the design of further analogues. This LHRH analogue was synthesized by the solid phase methology on a [3-((Ethyl-Fmoc-amino)-methyl)-l-indol-l-yl]-acetyl AM resin via Fmoc/tBu methology. ID and 2D NMR experiments were carried out in 2-2.5mM solution of [Aib6, desGly10]-LHRH ethylamide in DMSO-d6. The average structure of LHRH analogue was calculated from the family of the 20 models and refined using NOE-derived structural information.

Introduction The hypothalamic decapeptide luteinizing hormone-releasing hormone (LHRH) (pGluHis-Trp-Ser-Tyr-Gly-Leu-Arg-Pro-Gly-NH2) plays a key role in the control of mammalian reproduction. It is secreted in pulses from the hypothalamus and stimulates the anterior pituitary gland to release the gonadotropins luteinizing hormone (LH) and follicle-stimulating hormone (FSH). Ultimately, both of these hormones elicit gonadal production of sex steroids and gametogenesis [1-4]. By modification of the molecular structure of this decapeptide, analogues were synthesized with agonistic or antagonistic effects on the gonadotrophic cells of the anterior pituitary gland. The agonists, after an initial stimulatory effect ('flare up'), lead

206

A.A. Zompra et al. /A Synthetic LHRH Analogue in Solution

to desensitization of the gonadotrophic cells and a reduction in the number of LHRH receptors on the cell membrane ('down-regulation') [5], while the antagonists produce an immediate effect by competitive blockade of the LHRH receptors. Oncological uses of LHRH analogues are based on the inhibition of the pituitary gonadal function, the suppression of gonadal steroid hormone secretion leading to inhibition of the growth of sex hormone dependent tumors [6,7]. However, direct effects of LHRH analogues on tumor cells may also play a role [8]. There is evidence that specific membrane receptors for LHRH are present in various tumors, such as human prostate and breast cancer [811], ovarian [13, 14], endometrial [12-14], and pancreatic cancers [4, 15). In patients afflicted with these malignancies, conventional chemotherapy results in a varying degree of response and is associated with severe toxic side effects [6]. Agonistic analogues of LHRH, represented by Leuprolide ([DLeu6, desGly10]LHRH-ethylamide), have been widely used in oncology and gynecology for nearly two decades. Leuprolide (hCAS: 53714-56-0; Leuprolide or Leuprorelin) and Leuprolide Acetate (hCAS: 74381-53-6; Leuprorelin Acetate) are synthetic structural analogues of LHRH. Leuprolide presents a higher affinity for the LHRH receptors and a higher resistance to enzymatic degradation than the natural hormone, which increases its biological potential. It can thus be used for the treatment of a large number of diseases related with the regulation of sexual hormones, such as masculine and feminine infertility, uterine myomas and prostatic and mammalian tumors [16, 17]. It was revealed many years ago that substitution of the Gly amino acid residue at position 6 of LHRH by various D-amino acids results in very potent analogues of LHRH with high binding affinity [18]. In our early attempts to create agonistic analogue of LHRH, Aib (a-aminoisobutyric acid) residue, which is known to favor the formation of a turn when inserted in the peptide sequence [19-21], was incorporated at position 6 of LHRH. Key considerations in the synthesis of the above LHRH analogue were on the basis that the bioactive conformation of LHRH includes a type II ß-tum involving residues 5-8 {Tyr5-Gly6-Leu7-Arg8} [22]. Also, in previous studies, we had synthesized Leuprolide using the solid phase Fmoc/tBu methology on a Sieber Ethylamide resin (Ethylamino-xanthen-3-yloxy-Merrifield resin) in 30% overall yield. Conformational analysis of this molecule suggested the existence of a loop structure in the region of Tyr5-DLeu6 residues [23]. In this study, we report an improved synthesis (higher yield) and the conformational analysis in solution of the LHRH analogue namely [Aib6, desGly10]LHRH-ethylamide using NMR spectroscopy, in order to gain valuable insights on bioactive conformation and use these for the design of further synthetic analogues (agonists and/or antagonists) of LHRH. Experimental Section Materials: 9-Fluorenylmethoxycarbonyl-protected amino acids and peptide reagents were obtained from Bachem AG and Novabiochem. All solvents and reagents used for solid-phase synthesis were of analytical quality. Dimethylsulfoxide DMSO-d6 (MERCK) was used as deuterated solvent in NMR experiments and the peptides were dissolved to a final concentration of 2-2.5 mM in order to record ID and 2D NMR spectra.

A.A. Zompra et al. /A Synthetic LHRH Analogue in Solution

207

Peptide Synthesis and Purification: [Aib6, desGly'°]-LHRH-ethylamide was assembled on a [3-((Ethyl-Fmoc-amino)-methyl)-l-indol-l-yl]-acetyl AM resin [24] containing 0.85 mequiv. of amino group/g with the protecting groups shown in parentheses (pyroglutamic acid-His(Trt)-Tip(Boc)~Ser(tBu)-Tyr(tBu)-Aib-Leu-Arg(Pbf)-Pro-NHEt). Chain extension was carried out using standard Fmoc (9fluorenylmethyloxycarbony]) protocols [25]. In summary, the Fmoc group was removed with 25% piperidine in N,N-dimethylformamide (DMF). Activation of each amino acid was performed in situ using diisopropylcarbodiimide/l-hydroxy-7-azabenzotriazol DIC/HOAt in DMF [26]. Completeness of the reaction was monitored by Kaiser test [27]. Treatment of the peptidyl resin with trifluoroacetic acid/ methylene chloride/1,2ethanedithiol/anisole/water (90:5:1:2:2, v/v) (15 ml/g peptide resin) for 4 hours afforded the desired product. The peptide was precipitated upon evaporation in vacuo and addition of ether. The crude peptide was purified by gel filtration chromatography on Sephadex G-15 using 15% acetic acid as the eluent. Final purification was achieved by preparative HPLC (Pharmacia LKB-2250) on re versed-phase support C-18 with a linear gradient from 10 to 80% acetonitrile (0.1% TFA) for 40 min at a flow rate 2.0 ml/min and UV detection at 220 and 254 nm. The appropriate fractions were pooled and lyophilized. Analytical HPLC (Pharmacia LKB-2210) equipped with a C-18 Phase Sep column S3 ODS2 produced single peak with at least 98% of the total peptide peak integrals. Electro-spray MS was in agreement with the expected results. The desired product was in 44.5% overall yield. NMR Spectroscopy: ID NMR and 2D NMR spectra were recorded at the temperature of 298K on a Bruker Avance 400 spectrometer operating at 400.13 MHz. ID spectra over the full spectral width (12 ppm) were acquired with and without presaturation of the H2O signal and a recycle delay of 0.8 sec or 1.0 sec. DQF-COSY [28] and TOCSY [29] experiments were performed in order to facilitate the identification of the spin-systems of each individual amino acid. TOCSY experiments were carried out by using the MLEV-17 spin-lock sequence and a mixing time of 80-100 ms in both temperatures, consisted by 2K data points in F2 dimension, 16-32 transients and 1024 complex increments in the Fl dimension. TPPI NOESY [30] spectra were recorded using the same spectral width as in the case of ID spectra (12 ppm). The mixing times varied from 200 ms to 800 ms. Spectra recorded with mixing time above 400 ms were used in order to facilitate the resonance assignment. The suppression of the residual water signal was carried out by its presaturation during the relaxation delay and the mixing time using the [31] "PRESATURATION" pulse sequence of the standard BRUKER library. The other acquisition parameters were the same as for the ID spectra. The spectra consisted of 2K data points in the F2 dimension and IK experiments in the Fl dimension. Raw data were multiplied in both dimensions by a pure cosine-squared bell window function and Fourier-transformed to obtain 2024x1024 real data points. A polynomial base-line correction was applied in both directions. NMR data processing was performed using the standard Bruker software package on a Silicon Graphics O2 workstation. The 2D maps were analyzed on a Silicon Graphics O2 workstation or on Pentium III PC-Linux computers with the aid of the program XEASY (ETH, Zurich) [32].

208

A.A. Zompra et al. /A Synthetic LHRH Analogue in Solution

Results and Discussion NMR Constraints: The various protons are classified into 5 categories (according to classification by Wuthrich) [33] and their distance-to-volume correlation is calibrated according to the following scheme: intraresidue except HN, Ha, Hß- number of peaks 85 (8 methyl peaks) sequential and intraresidue HN, Ha, Hß - " 75 medium range " 14 long range backbone " 1 long range" 59 (10 methyl peaks) Total number of cross peaks 234 (18 methyl peaks). The above-mentioned upper distance limits were used as input data (structural constraints) for the structural calculations with program DYANA (Dynamics Algorithm for Nmr Applications) [34]. Structure Calculations and Refinement: A total of 456 NOESY cross-peaks (90% derived from the NOESY map acquired with the "PRESATURATION" pulse sequence) were integrated and transformed into 234 unique upper distance limits (Figure 1). From those, 149 constraints (14,9 per assigned residue) were found meaningful and used in structural calculations with DYANA program [35]. The Aib residue was generated from a Valine where the Ca of the Aib corresponds to Cß of a Valine and, Cß1/Cß2 of the Aib corresponds to Cyl/Cy2, respectively, of a Valine. The 20 best structures (out of 300 calculated) in terms of target function with NOE constraint violations below 0.2 A (largest violation 0.255 A) are included to the final family. Each one of 20 DYANA models, is refined by Restrained Energy Minimization (REM) through programs included to AMBER 5.0 Package [36]. A force constant of 133.76 kJxmol 1 A is applied for the distance constraints. Structural calculations have been performed on IBM RISC6000 or on Pentium ITJ PC-Linux computers.

A.A. Zompra et al. /A Synthetic LHRH Analogue in Solution

209

Secondary Structure: Figure 1 shows the short and medium range NOEs observed for the backbone and ß protons in the NOESY spectrum.

4

6 Sequence

Figure 1: (A) Sequential dipolar connectivities involving NH, Ha, and Hß protons in [Aib6, desGly10]LHRH-ethylamide), (B) The number of observed experimental NOEs per residue. White bars indicate intraresidue NOEs, light gray bars, sequential NOEs, and dark gray and black, medium and long range NOEs, respectively.

Sequential NH-NH are observed throughout the region of residues 1 to 8. HaHN connectivities are also observed in the peptide fragment 1 to 6, while interruption of this kind of connectivities is due to the presence of Aib residue at position 6, which lacks a protons. Additional Ha-HN connectivities are observed for the terminal tripeptide between residues Leu7, Arg8 and Pro9 with the NH-Et. Hß-HN are detected for the peptides fragments 1 to 6 and between the protons of Pro9 and NH-Et terminus. HN-HN couplings for stretches up to two residues are detected for the fragment of residues Ser - Pro9, while coupling of the same range (stretches up to two residues) of the type Ha-HN are detected for the region 1 to 7. Only one Ha-HN type of coupling

210

A.A. Zompra et at. /A Synthetic LHRH Analogue in Solution

between residues that sited three positions apart at the peptide sequences is observed and concerns the residues Tyr5 and Arg8. Tertiary Structure: Numerous NOE signals are detected between main- and side- chain protons in the peptide fragments consisted by residues 3-6 and 5-8. The fact that the pairs of residues Trp3/Aib6 and Tyr5/Arg8 are found in short distance apart giving rise to NOE contacts, could be justified by the existence of bend occurred in the middle of the sequence of the peptide chain. Indeed, the skeleton of the peptide, as calculated using NMR data, found to form aß turn at the region covered by residues Ser4 up to Arg8. Table 1: Restraint violations, structural and energetic statistics for the solution structure of [Aib6, desGly'°]-LHRH-ethylamide) 2 (20 structures)

RMS Violations per experimental distance constraint (A) Intra-residue-34 0.0304 ± 0.0046 0.0334 Sequential-76 0.0192 ± 0.0017 0.0240 Medium range-37 0.0128 ± 0.0053 0.0231 Long range-2 0.0416 ± 0.0250 0.1068 Total-149 0.0219 ± 0.0021 0.0289 Average number of Violations per structure Intra-residue 1.10 ± 0.30 1.00 Sequential 3.95 ± 0,67 6.00 Medium 1.10± 0.54 3.00 Long range 0.75 ± 0.43 1.00 Total 6.90 ± 0.83 11.00 Average number of NOE 0.00 ± 0.0 00.0 violations larger than 0.3

_A

Average number of NOE 2.15± of violations between 0.1 -0.3 A Largest NOE violation 0.26

0.7

5.00 0.20

Average distance penalty 2.75 ± 0.4 4.25 function (A2) REM indicates the energy-minimized family of 20 structures. 2 is the energy-minimize! energy-minimized average structure obtained from the coordinates of the 20 individual REM structures. Moreover, detected signals between the amide protons of the Trp3 and the terminal Ethylamide clearly indicates that the two termini of the peptide are in close proximity. Thus, the whole molecule is characterized by a U-shape tertiary structure. The resulting calculated DYANA family of 20 structures are characterized by high resolution and precision due to the large number of NOE constraints used. From the 234 upper distance limits (23.4 constraints per residue) resulted from the volume-todistance conversion of the NOE intensities, the 149 were found meaningful and used in molecular dynamic calculation. This number of constraints correspond to 14.9 meaningful distance constrain per residue of the LHRH analogue. The pairwise RMSD

21

A.A. Zompra et al. /A Synthetic LHRH Analogue in Solution

values are calculated for the region 2 - 9 are 0.31 ± 0.14 A for the backbone and 1.36 ± 0.59 A for the heavy atoms. The RMSD values for the mean structure compared to the 20 models of the family are 0.23 ± 0.07 A for the backbone and 1.02 ± 0.04 A for all the heavy atoms. At Table 1 statistical data concerning the calculation and the quality of the 20 models and the mean structure after the energy minimization procedure, are reported.

Leu7

NEt10"

NEtlO

Figure 2: Backbone of the best 20 structures of [Aib6, desGly'°]-LHRH-ethylamide. Side chains of Trp3, NEt10 (left) and Ser4, Leu7 (right) are also shown.

The backbone of the family, together with the side chains of the Trp3 and the ethylamide terminus, of 20 energy-minimized structures are presented at Figure 2. The conformation of the backbone and the orientation of the indole ring of Trp3 have been calculated with high resolution as indicated by the low RMSD values and illustrated at Figure 2. The bulky indole ring of Trp3 is found apart from the backbone of the peptide region 7-10, while the two termini of the peptide are in close distance. As shown at Figure 2 (left) the amide protons of Trp3 and NH-Et are found in distance that varies from 2.7 to 3.3 A among the family of 20 models. Additionally, the side chains of Ser4 and Leu7, which are sited at the ß turn region of the molecule, orient towards the same direction while approach one each other (Figure 2, right). Validation of the Structure: In order to validate our experimental results in terms of calculated structures based on NOE-derived distance constraints, we attempted to recalculate the structure in absence of two specific distance constraints. These constraints might play a fundamental role in the ß turn formation and the convergence of the two termini. In other words, we wish to elucidate whether the calculated structures are biased, or not, to form the characteristic ß-turn, by two structural constrains which are derived by NOE signals between protons of the two termini. More specifically, well-defined NOE cross-peaks are detected between the Hel imidazole 7 proton of His and the Ha backbone proton of Arg8 , in accordance with an intense NOE signal involving the two amide protons of Trp3 and NH-Et. Therefore, structural

212

A.A. Zompra et al. /A Synthetic LHRH Analogue in Solution

calculations were performed with a set of distance constraints that lacks only these two distances that concerns the above-mentioned proton NOEs. The resulted structures are of slightly lower resolution (higher RMSD values) due to the higher degree of conformational/orientational freedom that possesses the two termini, in the absence of these two distance constraints. The RMSD values are calculated 0.77 ± 0.35 A for the backbone and 1.76± 0.55 A for all the heavy atoms. The final 20 best models (in terms of dynamic energy and residual distance violations) are divided in two subfamilies; one consisted by 9 models and the other that consisted by 11 models (Figure 3).

Trp3

NEt10 pGluI Figure 3: Backbone of the best 20 models calculated without considering structural data for the long range NOEs.

Although the backbone atoms of the first and last residues of the peptide sequence in the subfamily of the 11 models might be in a somewhat larger distance than in the second subfamily (see also Figure 4) as well as in the originally calculated structures (using all the experimental distances) the resulting tertiary structure of the peptide is rather similar. These findings are leading to the following conclusions: Concerning the relative conformation of the two terminal parts of the peptide, the fact that two different conformational preferences are calculated without the use of the NOE constraints between the His2-Arg8 and Trp3-NHEt, can not rule out a possible convergence of the two termini, as found at the originally calculated structures. The tertiary structure of the peptide does not differ considerably in the absence of these strategic NOEs, and the molecule does not adopt a more linear structure while the ß turn observed in the region 4-8 is not collapsed. Surprisingly, the two fragments of the ß turn structure are found in a distance even smaller than in the original structure (see also Figure 4). This is manifested by the distance between the amide nitrogens of the residues Ser4 and Arg8 monitored for the members of each one of the two families of the 20 calculated models (Figure 4A). In 19 out of 20 models the HN-HN distance is found shorter for the models calculated without the two distance constraints.

A.A. Zompra et al. /A Synthetic LHRH Analogue in Solution

0

2

213

4 6 8 10 12 14 16 18 20 Structure number (20 best models DYANA) N H 3 - N H 1 0 — •— D Y A N A w i t h n o e HN3-HN10 NH3-NH10 D Y A N A w i t h o u t n o e HN3-HNH)

10,0 9,0 -

0

2

4 6 8 10 12 14 16 18 20 Structure number (20 best models DYANA)

Figure 4: The range of the distances between the nitrogens of the backbone for the Ser4 and Arg8 residues (above), and amide protons of the Trp3 and NEt10 (below) among the family of 20 models with or without the distance constraints derived from the NOE signals between residues 2/8 and 3/10.

The typical ß turn structure calculated for the peptide fragment 4-8 based on numerous interesidue NOEs is possibly the result of the introduction of Aib residue in the position 6 of the sequence. Additionally, it is noteworthy that this structure could also be stabilized by the existence of several hydrogen bonds found in the calculated structures, involving the amide HN protons of Leu7 and Arg8 with the carbonyl oxygen atoms of Ser4 and Aib6, respectively. Figure 5 presents the backbone of the mean, energy minimized, structure of the peptide [Aib6, desGly10]-LHRH-ethylamide, with the characteristic fi turn and the orientation of the side chains of all the residues.

214

A.A. Zompra et al. /A Synthetic LHRH Analogue in Solution

.Arg8

(C) Figure 5: (A) Backbone of the mean energy minimized structure of the [Aib6, desGly10]-LHRHethylamide, (B) Mean structure and conformation of the side chains of residues pGlu1,Tip3, Ser4, NEt10.Van der Waals radius of heavy atoms are also shown for these residues, (C) Mean structure of [Aib6, desGly10]-LHRH-ethylamide with all atoms. Figures are generated with MOLMOL [37].

Biological implications: The molecule of [Aib6, desGly10]-LHRH-ethylamide belongs to the family of Leuprolide synthetic analogues, which have demonstrated noticeable biological activity in the treatment of sex hormone dependent tumours. Consequently, the profound study of its structure-activity relationship is of immense interest and the detailed analysis of the structure and its features/properties could significantly contribute to the design and synthesis of new synthetic analogues. Such molecules, peptides or not, should have high binding affinity towards the target receptors and form stable receptor-peptide molecular complexes. For this reason the knowledge of the distribution of the electrostatic potentials on the surface of the molecule is essential. The present synthetic analogue has no charged end-residues due to the presence of pyroglutamic acid to the position 1 of the sequence and NH-Et to the carboxyl terminus of the peptide. The only charged residue in the sequence is the Arg8; thus the overall charge of the molecule is +1. At Figure 6 the distribution of the electrostatic charges on the surface of the peptide is illustrated. The positive charge density arisen from guanidine group is rather localized at the side chain of the Arginine residue and to the backbone of the proximal residues. The side chain of Arginine is oriented towards the exterior of the molecule and due to its flexibility could play a certain role in intra-molecular, receptor-substrate, recognition processes. The rest of the molecule is characterized mainly by its hydrophobic character (white colour = no charge, blue colour = positive charge) due to the presence of the aromatic residues, Histidine, Tryptophane and Tyrosine at positions 2, 3 and 5, respectively.

A.A. Zompra et al. /A Synthetic LHRH Analogue in Solution

215

Figure 6: (A) Mean, energy minimized, NMR structure and Van der Waals surfaces for His2, Trp3, Tyr5 and Arg8 residues, (B) Distribution of the electrostatic potentials on the surface of the mean structure of [Aib6, desGly'°]-LHRH-ethylamide. Figures are generated with MOLMOL [37].

Perspectives: [Aib6, desGly10]-LHRH-ethylamide has not been yet tested for the biological activity and binding affinity on the LHRH receptor, but having gone through design, synthesis and analysis of this molecule suitable model compounds that fulfil the structural requirements for a potent LHRH agonist or antagonist will be designed and synthesized. More specifically, the present conformational studies are expected to provide the structural basis for the development of new synthetic analogues (agonists and/or antagonists) through rational modification of the physical properties, such as hydrophobic/hydrophilic character, number and position of charged residues, while monitoring their binding affinity and biological activity. Synthesis of new analogues varying in positions 6, 8 and 10 are currently under way in our laboratory. References [1] H. Matsuo, Y. Baba, R.M.G. Nair, A. Arimura & A.V. Serially, Biochem. Biophys. Res. Commun., 43 (1971)1334. [2] Y. Baba, H. Matsuo & A.V. Schally, Biochem. Biophys. Res. Commun., 44 (1971) 459. [3] P.M. Conn, J.A. Janovick, D. Stanislaus, D. Kuphal & L. Jennes, In: Vitamins and Hormones, (G. Litwack, ed.). Academic Press, New York, 1995,Vol. 50, pp 151–214. [4] A.V. Schally & A.M. Comaru-Schally, Hypothalamic and other peptide hormones. In: Cancer Medicine, (J.R. Holland, E Frei III, R.R. Jr. Bast, D.E. Kufe, D.L. Morton & R.R. Weichselbaum, eds.), Williams and Wilkins, Baltimore, 1997, edn. 4, pp 1067–1086. [5] P.E. Belchetz, T.M. Plant, Y. Nakai, E.J. Keogh & E. Knobil, Science, 202 (1978) 631. [6] A.V. Schally, A. Nagy, K. Szepeshazi, J. Pinski, G. Halmos, P. Armatis, M. Miyazaki, A.M. ComaruSchally, T. Yano & G. Emons, In: Treatment with Gn-RH Analogs: Controversies and Perspectives, (M. Filicori & C. Flamigni eds.), Parthenon, Carnforth, U.K., 1996, pp. 33–44. [7] G. Emons & A.V. Schally, Human Reproduction, 9 (1994) 1364. [8J A. Qayum, W. Gullick, R.C. Clayton, K. Sikora & J. Waxman, fir. J. Cancer, 62 (1990) 96. [9] M. Fekete, T.W. Redding, A.M. Comaru-Schally, A.E. Pontes, R.W. Connelly, G. Srkalovic & A.V. Schally, Prostate, 14 (1989) 191. [10] A.V. Schally, A.M. Comaru-Schally & D. Gonzalez-Barcena. Biomed. Pharmacother., 46 (1992) 465. [11] M. Fekete, J.L. Wittliff. & A.V. Schally, J. din. Lab. Anal., 3 (1989) 137.

216

A.A. Zompra et al. /A Synthetic LHRH Analogue in Solution

[12] G. Emons & A.V. Schally, Hum. Reprod., 9 (1994) 1364. [13] G. Emons, O. Ortmann, M. Becker, G. Irmer, B. Springer, R. Laun, F. Holzel, K.D. Schulz & Schally, A.V., Cancer Res., 53 (1993) 5439. [14] G. Emons, B. Schroder, O. Ortmann, S. Westphalen, K.D. Schulz & A.V. Schally, J. Clin. Endocrinol. Metab., 77 (1993) 1458. [15] A.V. Schally, S. Radulovic & A.M. Comaru-Schally, In: Experimental and Clinical Studies in Hormone Dependent Cancers, (E. Mazzaferri & N. Samaan, eds.), Blackwell, Boston, 1993, pp. 4973. [16] Martindale, The Extra Pharmacopeia, 31th ed.. Royal Pharmaceutical Society, London, 1996. [ 17] G.L. Plosker &. R.N. Brogden, Drugs, 48 (1994) 930. [18] M.J. Karten & J.E. Rivier, Endocrine Reviews, 7 (1986) 44. [19] P. Cordopatis, E. Manessi-Zoupa, D. Theodoropoulos, R. Bosse, R. Bouley, S. Gagnon & E. Escher, Int. J. Pept. Prot. Res., 44 (1994) 320. [20] R.E. London, P.G. Schmidt, R.J. Vavrek & J.M. Stewart, Int. J. Pept. Protein. Res.. 4 (1982) 334. [21] T.S. Sudha & P. Balaram, Int. J. Pept. Protein. Res.. 21 (1983) 381. [22] F.A. Momany, J. Am. Chem. Soc.. 98 (1976), 2990. [23] (a) D. Benaki, E. Paxinou, V. Magafa, G.. Pairas, E. Manessi-Zoupa, P. Cordopatis & E. Mikros, 2nd Hellenic Forum on Bioactive Peptides, (P. Cordopatis, ed.), Typorama, 2001, p.p. 439–446, (b) D. Benaki, E. Paxinou, V. Magafa, G. Pairas, E. Manessi-Zoupa, P. Cordopatis & E. Mikros, Letters in Peptide Science, submitted. [24] K.G. Estep, C.E.E. Neipp, L.M. Stephens-Stramiello, M.D. Adam, M.P. Allen, S. Robinson & E.J. Roskamp, J. Org. Chem.. 63 (1998) 5300. [25] B.G. Fields & L.R. Noble, Int. J. Peptide Res., 35 (1990) 161. [26] L.A. Caprino, J. Am. Chem. Soc., 115 (1993) 4397. [27] E. Kaiser, R.L. Colescott, C.D. Bossinger & P.I. Cook, Anal. Biochem., 34 (1970) 595. [28] U. Piantini, O.W. Sorensen &. R.R Ernst, J. Am. Chem. Soc., 104 (1982) 6800. [29] A. Bax & D.G. Davis, J. Magn. Reson., 65 (1985) 355. [30] (a) D. Marion & K. Wuthrich, Biochem. Biophys. Res. Commun., 113 (1983) 967; (b) J. Jeener, B.H. Meier, P. Badimann & R.R. Ernst J. Chem. Phys., 71 (1979) 4546. [31] S. Macura, K. Wuthrich &. R.R. Ernst, J Magn. Reson., 47 (1982) 351. [32] C. Eccles, P. Guntert, M. Billeter & K. Wuthrich, J. Biomol. NMR, 1 (1991) 111. [33] K. Wuthrich, In: NMR of Proteins and Nucleic Acids, Wiley, New York, 1986. [34] P. Guntert, W. Braun & K. Wuthrich, J. Moi Biol., 217 (1991) 517. [35] P. Guntert, C. Mumenthaler & K. Wuthrich, J. Mol Biol., 273 (1997) 283. [36] D.A. Pearlman, D.A. Case, J.W. Caldwell, W.S. Ross, T.E. Cheatham, D.M. Ferguson, G.L. Seibel, U.C. Singh, P.K. Weiner & P.A. Kollman, AMBER 5.0. U of California, San Francisco, USA, 1997. [37] R. Koradi, M. Billeter & K. Wuthrich, J Mol. Graphics, 14 (1996) 51.

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

217

Synthesis, Biological Evaluation & NMR Solution Structural Models of New Oxytocin Analogues M. Fragiadaki1, S. Koumentakos1, D. Raptis2, G. A. Spyroulias1, V. Magafa1, J. Slaninova3 and P. Cordopatis 'Department of Pharmacy, University of Patras, GR-265 04 Patras, Greece 2 Technological Education Institute, Egaleo, GR-12210 Athens, Greece Department of Peptide Biochemistry, Institute of Organic Chemistry and Biochemistry, Academy of Sciences of Czech Republic, Flemingovo square 2, Prague 6, CZ-166-10, Czech Republic Abstract. We report the solid-phase synthesis of four oxytocin analogues in which the residues in sequence position 8 or 9 were replaced by L-a-t-butylglycine [Gly(Bu')]: [Gly(Bu')8]OT (re-synthesized), [Gly(But)9]OT, [Mpa1, DTyr(Et) 2 , Gly(But)8]OT, [Mpa1, D-Tyr(Et)2, Gly(But)9]OT. The analogues were tested in the in vitro oxytocic assay. In addition, the rat pressor activity was also examined. The analogues [Gly(But)8]OT and [Gly(But)9]OT were found to be weak agonists, while the peptides [Mpa1, D-Tyr(Et)2, Gly(But)8]OT and [Mpa1, D-Tyr(Et)2, Gly(But)9]OT act as very potent antagonists of oxytocin in the rat uterus assay with in vitro pA2 value of 8.2. The solution conformations of the analogues were also studied by a combined use of ID NMR and 2D NMR spectroscopy in DMSO-d6 through restrained molecular dynamics (MD).

218

M. Fragiadaki et al./ New Oxytocin Analogues

Introduction Oxytocin, [cyclo (Cys1-Tyr2-Ile3-Gln4-Asn5-Cys6)-Pro7-Leu8-Gly9-NH2] (OT), is a neurohypophyseal hormone which has been mainly associated with milk ejection and uterine contractions in mammals. Oxytocin was the first neuropeptide whose primary structure was determined and proved by total synthesis [1]. Ever since in several laboratories all over the word, hundreds of oxytocin analogues have been synthesized and studied in an important effort to obtain selective and active analogues, and to elucidate the requirements for peptide-receptor recognition and signal transduction. The first model of solution conformation of OT was proposed by Urry and Walter (cooperative model) [2] and consequently, a broad variety of models have been suggested. Most of them are in agreement that a conformation with a ß-turn at positions 3 and 4, and with a reverse turn at positions 7 and 8, might be a pre-requisite for the uterotonic activity of OT and its analogues [3], whereas the Tyr2 and Asn5 positions are thought to be important for initiation of the biological response [4]. Furthermore, the importance of conformational flexibility for agonist activity and relative conformational rigidity and steric constraints for antagonism has been emphasized (dynamic model) [5]. The oxytocin receptor (OTR) is a classical membrane receptor with seven transmembrane domains linked through a G protein complex to a phospholipase C/protein kinase C signal transduction system [6]. Using a series of OT antagonists, Chan [7] suggested the presence of two distinct subtypes of oxytocin receptors but, to date, there has been no evidence supporting the existence of more than one OTR gene or transcript [8]. The design of OT antagonists has been mainly based on data from structureactivity studies. Structural modifications of OT were studied for the changes in activity as determined by several in vitro and in vivo tests. These studies of oxytocin analogues have revealed that OT agonists and antagonists have different receptor-binding conformations [9]. With the cloning of the OT receptor the design of new analogues is based on structural modeling of the OT receptor complex [10]. Furthermore, recent studies establish that a small 12-residue segment in the distal portion of the N-terminus of the human OTR is required for agonist binding but it does not contribute to the binding site for antagonists [11]. Oxytocin antagonists, in contrast with currently used tocolytics, afford greater specificity and can be expected to exhibit improved efficacy and risk profiles. Such compounds will allow a more effective treatment of preterm labor, with a lower risk of side-effects. To date, although the structures of many oxytocin analogues have been reported, very few have been tested in animal models of parturition, probably because of their low receptor specificity or low receptor affinity and only one, atosiban [Mpa1, D-Tyr(Et)2, Thr4, Om8] has been administered to pregnant women [12]. Generally, inhibitors of the uterotonic activity of OT have been traditionally produced by introduction of bulky ß-carbon substituents into position 1 and/or by substitution of L-tyrosine in position 2 of OT with an aromatic D-amino acid. On the other hand previous structure-activity studies in our laboratory revealed that minimal structural changes of the oxytocin molecule provide weak antagonists [13]. Moreover, synthesis of analogues of biologically active peptides in which structural modifications should enhance resistance toward enzymatic cleavage of peptide bonds resulting in prolonged time course of action is of great importance. It has

M. Fragiadaki et al. /New Oxytocin Analogues

219

been reported that modification of the C-terminal tripeptide side chain influences the chymotryptic cleavage of Tyr-Ile peptide bond in the oxytocin molecule [14]. On the basis of these findings, we set out the synthesis of three new oxytocin analogues and one re-synthesized [15] which contain the conformation constraining amino acid residue L-a-t-butylglycine [Gly (But)] in positions 8 or 9. For these analogues were performed in vitro pharmacological tests and conformational analysis.

Experimental Section Materials: 9-Fluorenylmethoxycarbonyl-protected amino acids and peptide reagents were obtained from Bachem AG and Nova Biochern. The derivative Mpa(Trt) was prepared according to literature [16]. All solvents and reagents used for solid-phase synthesis were of analytical quality. Dimethylsulfoxide DMSO-d6 (MERCK) was used as deuterated solvent in NMR experiments and the peptides were dissolved to a final concentration of 2-2.5 mM in order to record ID and 2D NMR spectra. Peptide Synthesis and Purification: The analogues were synthesized by Fmoc solid phase methodology [17] utilizing a 2-chlorotrityl chloride resin [18] as solid support bearing a Rink-Bernatowitz linker to provide the peptidic amide [19]. Stepwise synthesis of the peptide analogue was achieved with diisopropylcarbodiimide/1hydroxy-benzotriazol (DIC/HOBt) in N,N-dimethylformamide (DMF) as coupling agent [20]. Couplings were performed with 3 molar excess of Fmoc-amino acid, 3.3 molar excess DIG and 4.5 molar excess HOBt as coupling agents in DMF for 2 h at room temperature. Completeness of the reaction was monitored by the Kaiser test [21]. The Fmoc groups were removed by treatment with 20% piperidine in DMF for 30 minutes. Cleavage of peptide-linker bond and removal of the protecting Trt-group from Cys, Asn, Gln and Mpa was accomplished in 4 h at room temperature using a solution of trifluoroacetic acid/ methylene chloride/1,2-ethanedithiol/anisole/water (80:10:5:3:2, v/v) (15 ml/g peptide resin). The obtained peptide was precipitated upon concentration of solvent and addition of ether. The formation of the disulfide bridge was performed in 25% DMSO/H2O for 48 h at room temperature [22]. The solution was frozen and lyophilized to give the final crude oxidized product. All the analogues were purified by gel filtration chromatography on Sephadex G-15 using 25% acetic acid as the eluent. Final purification was achieved by preparative HPLC (Pharmacia LKB-2250) on reversed-phase support C-18 with a linear gradient from 10 to 60% acetonitrile (0.1% TFA) for 30 min and 60 to 100% acetonitrile (0.1% TFA) for 10 min at a flow rate 1.5 ml/min and UV detection at 220 and 254 nm. The appropriate fractions were pooled and lyophilized. All products gave single spots on TLC (Merck precoated silica gel plates, type G60-F254) in the solvent systems: (A) 1butanol-acetic acid-water (4:1:5 upper phase) and (B) 1-butanol-acetic acid-waterpyridine (15:3:10:6). Analytical HPLC (Pharmacia LKB-2210) equipped with a C-18 Phase Sep column S3 ODS2 produced single peaks with at least 98% of the total peptide peak integrals. Electro-sprays MS were in agreement with the expected results.

220

M. Fragiadaki et al. /New Oxytocin Analogues

Biological Assay Methods: The uterotonic activity was determined in vitro on an isolated strip of rat uterus in the absence of magnesium [23]. In principle, cumulative dosing was applied in most experiments, i.e. doses of standard (in the presence or absence of analogues) or of the analogue were added successively to the uterus in the organ bath in doubling concentrations and in 1 min intervals without the fluid being changed until the maximal response was obtained. The agonistic activity was determined by comparing the threshold doses of oxytocin and the analogue. In the case that the analogues were not able to reach the same maximal response as oxytocin, the single dosing procedure was employed. The inhibitory potencies are expressed as their pA2 values. The pA2 values represent the negative logarithm to the base 10 of the average molar concentration [M] of an antagonist which will reduce the response of 2 x units of the agonist to the response with x units of the agonist [24]. Each analogue was tested on uteri from 4-8 different rats. Pressor activity was determined on phenoxybenzamine treated male rats [25]. NMR Spectroscopy: ID NMR and 2D NMR spectra were recorded at the temperature of 298K on a BRUKER A VANCE 400 spectrometer operating at 400.13 MHz. ID spectra over the full spectral width (12 ppm) were acquired with and without presaturation of the H2O signal and a recycle delay of 0.8 s or 1.0 s. DQF-COSY [26] and TOCSY [27] experiments were performed in order to facilitate the identification of the spin-systems of each individual amino acid. TOCSY experiments were carried out by using the MLEV-17 spin-lock sequence and a mixing time of 80-100 ms in both temperatures, consisted by 2K data points in F2 dimension, 16-32 transients and 1024 complex increments in the Fl dimension. TPPI [28] NOESY spectra were recorded using the same as for ID spectra, spectral width. The mixing times varied from 200 ms to 800 ms. Spectra recorded with mixing times 400 ms or longer, were used in order to facilitate the resonance assignment. The NOE intensity volume-to-distance conversion was performed for the NMR data derived by NOE maps acquired with mixing times of 200 or 400 ms. The suppression of the water signal was carried out by its presaturation during the relaxation delay and the mixing time using the "PRESATURATION" [29] pulse sequence of the standard BRUKER library. The other acquisition parameters were the same as for the ID spectra. The spectra consisted of 2K data points in the F2 dimension and IK experiments in the Fl dimension. Raw data were multiplied in both dimensions by a pure cosine-squared bell window function and Fourier-transformed to obtain 2024x1024 real data points. A polynomial base-line correction was applied in both directions. The assignment of TOCSY and NOESY spectra and the identification of dipolar intraproton connectivity for the analogues were performed with the use of sequence specific resonance assignment methodology [30]. NMR data processing was performed using the standard BRUKER software package on a Silicon Graphics O2 workstation. The 2D maps were analyzed on a Silicon Graphics O2 workstation or on Pentium III PC-Linux computers with the aid of the program XEASY (ETH, Zurich) [31]. Results and Discussion The new analogues [Gly(But)9]OT, [Mpa1, D-Tyr(Et)2, Gly(But)8]OT, [Mpa1, DTyr(Et)2, Gly(But)9]OT and the re-synthesized analogue [Gly(But)8]OT were prepared

M. Fragiadaki et al. /New Oxytocin Analogues

221

by Fmoc solid-phase procedure and tested for their biological potency in two pharmacological tests. The results are shown in Table 1 along with the activities of the parent hormone oxytocin, the deaminooxytocin and the oxytocin antagonist analogue Atosiban. As can be seen, replacement of Leu8 (analogue III) or Gly9 (analogue IV) by t Gly(Bu ) decreased biological activities significantly. The analogue [Gly(But)8]OT (III), with a shortened lipophilic bulky side-chain in position 8, was found to be a weak agonist in the uterotonic assay with a potency approximately 8% that of the native hormone. In addition, low potency was observed in the rat pressor assay. For comparison, previous reports concerning the biological activity of this analogue are in very good agreement with our results [15]. The analogue modified in position 9 (IV) with L-a-t-butylglycine (conformationally constrained and lipophilic residue) was found to be a partial agonist, weaker than the analogue III in the uterotonic assay (0.7% that of the oxytocin) and more selective due to the almost complete elimination of pressor activity. These findings are consistent with earlier observations in bibliography, which suggested that the C-terminal tripeptide sequence and especially the proper orientation of the C-terminal glycine carboxamide appear to be critical for obtaining high potency in oxytocin analogues [32, 33]. Therefore, structural modifications of the side-chain moieties in the C-terminal tripeptide oxytocin might lead to analogues with variable biological properties at different oxytocin target tissues. Such variations in biological activity could result from conformational changes influencing interactions of the analogues with the receptor, from steric demands, or from changed susceptibility of the molecules to enzymic cleavage [34]. Furthermore, the greater decrease in the agonistic potency of the analogue [Gly(But)9]OT comparing with the analogue [Gly(But)8]OT indicates that a lipophilic and bulky residue in position 9 induces major restrictions at the conformational freedom of the C-terminal part of the molecule than an identical residue in position 8. Most of the oxytocin antagonists described to date contain a residue without Nterminal amino group, i.e. ß-mercaptopropionic acid (Mpa) or ß-mercapto-ß, ßcyclopenta-methylenepropionic acid (Mca). This approach is the result of thorough study of data which indicate that vasopressin analogues lacking a terminal amino group are inactivated quite slowly [35] and also that the first amino acid has a decisive role in the receptor binding [36]. Furthermore, it has been reported that the configuration and the hydrophobicity of the aromatic amino acid in position 2 of oxytocin are important for the antagonistic activity [37]. In accordance with these observations we synthesized the analogues V and VI. Upon additional substitution of the residue in position 1 by ß-mercaptopropionic acid and the residue in position 2 by 0-ethyl-D-tyrosine, however, the weak agonist analogues (HI and IV) were converted to highly potent antagonist analogues in the in vitro uterine assay, having a pA2 value of 8.1. In comparison to [D-Tyr(OEt)2] OT it is almost 5 times higher inhibitory potency (pA2 = 7.36) [38]. Both analogues were completely inactive in the pressure test as was also the [D-Tyr(OEt)2] OT. The identical high antagonistic activity confirms the hypothesis that the basic character of the amino acid in position 8 is not important for the antagonism [12] of the uterotonic activity.

222

M. Fragiadaki et al. / New Oxytocin Analogues

Table 1. Biological Activities of Oxytocin and its Analogues Uterotonic activity in vitro (lU/mg)

Compound

Uterotonic activity in vitro pA2

Pressor activity lU/mg

I

Oxytocin

546

3.1

11

[Mpa1] oxytocin (deaminooxytocin) "

803

1.44

III

[Gly(Bul)8]OT

46.8 ± 7.2

1.1

IV

[Gly(But)9]OT

3.7+0.9

<0.04

V

[Mpa1, D-Tyr(Et)2, Gly(But)8]OT

8.1+0.2

0

VI

[Mpa1, D-Tyr(Et)2,Gly(But)9]OT

8.110.2

0

8.29

0.02+ 0.02

1

VII

2

4

8

[Mpa , D-Tyr(Et) , Thr , Arg ]OT (Atosiban)

The biological assays for the oxytocin analogues III, IV, V & VI were performed as outlined in the text. The biological activities of the other analogues reported here are taken from the literature: for oxytocin and deaminooxytocin see ref. [5] and for atosiban see ref. [12]. *If tested using cumulative dose-response procedure, the analogue is not able to reach the same maximal response as oxytocin; if tested using single dose procedure the response is reaching the same maximal contraction as oxytocin. Inactive up to the dose of 0.4 mg/kg of exp. animal.

The solution conformations of the analogues IV, V and VI were also studied by a combined use of ID NMR and 2D NMR spectroscopy in DMSO-d6. Upon completion of the identification and assignment of the NOE connectivities, the elements of secondary as well as the tertiary structure of the peptide have been identified. Structural information was extracted from signal intensities, (NOE[Hi-Hj]-l/r [Hi-Hj]6, where Hi, Hj are two protons, and r[Hi-Hj] is the distance between the two protons) and implemented to the structural calculations in order to determine the folding and the tertiary structure of each one of the analogues, in solution (Figure 1 shows representative spectra for analogue IV). Finally, it should be reported that 332, 210 and 243 unique NOE signals were detected for the analogues IV, V and VI respectively. The signal intensity was transformed into upper distance limits and used as structural constraints for the calculation of the 3D structure. From the NOEs observed, and reported above, 276, 154 and 187, respectively, are found meaningful and used in the calculations. After the end of the structural calculations through DYANA (Dynamics Algorithm for NMR Applications) program [39], the 20 best structures (in terms of minimum dynamic energy and smallest deviation of the experimental NOE distances with the same distances as found in the final set of calculated structures) are collected. The family of the 20 models is characterized by high resolution and accuracy as indicated by the low RMSD values for all analogues.

223

M. Fragiadaki et al. / New Oxytocin Analogues

8.50

8.00

3.80

3.80

4.00

4.00

4.20

4.20

4.40

4.40

4.60;

4.60

4.80

4.80

5.00

540

3.80

3.80

4.00

440

4.20

4.20

4.40

4.40

4.60

4.60

4.80

4.80

5.00

540

7.20 ',

7.20

7.40

c

4T

7.40

7.60

7.60

7.80

7.80

8.00

840

8.20

8,20

8.40

8.40

8.60

8.60 8.50

8.00

7.50

Figure 1: 2D !H NMR spectra of the analogue IV. At the A and B it is presented the characteristic area of Ha-NH conjugations as it is concerned the TOCSY and NOESY spectra respectively, while at the C it is presented the characteristic area of NH conjugations, as long as NH-NH NOE signals between the Cys1 and Asn5.

As can be seen, differences in the conformation of the final tripeptide are observed among the three analogues. This fact can be due to the expected high flexibility and mobility of this part of the molecule. High mobility of protein and peptide fragments results in a loss of an important number of NOE signals (constraints), and low precision of the calculated 3D structures. As far as the analogue IV is concerned the cyclic part of the molecule comprised by the amino acids 1-6 (which results by the disulfide bridge between the sulfur atoms of the residues at the positions 1 and 6) is well defined, as well as the aromatic ring of the Tyr at the position 2. In analogue VI the conformation of the C-terminal tripeptide is also well determined and it seems that this fragment folds in such a way that it adopts a ß-turn structure. This is manifested by the numerous NOE signals between the backbone and/or side chain protons of amino acid residues from which terminal of the molecule is consisted. In contrast, for analogue V with L-a-tbutylglycine in position 8, the final tripeptide fragment seems to adopt a different orientation in space but without loosing the characteristics of ß-turn structure observed in the case of analogue VI (L-a-r-butylglycine in position 9) [Figure 2A]. In addition, the tyrosine rings of the analogues IV and VI [Figure 2B] seem to have different orientation (the latter has Oethyl-D-tyrosine instead of the native tyrosine residue). The different conformation of the aromatic ring of the D-amino acid in position 2 could be discussed in conjunction with the antagonistic activity of the analogue VI, compared with the analogue bearing the natural L-amino acid at position 2 (IV).

224

M. Fragiadaki et al. / New Oxytocin Analogues

Figure 2: (A) Superposition of the mean structures of the analogues V and VI and (B) Superposition of the mean structures (as it concerns the part 1-6) of the analogues IV and VI

The combined approach including pharmacological tests, ID NMR, 2D NMR spectroscopy and 3D structures determined through molecular dynamics (MD) protocols allow us to conclude, that the C-terminal tripeptide has very different conformations and thus its conformation is relatively unimportant for the antagonistic character of the analogue in the oxytocin series and that the basic amino group in position 8 is also not a pre-requisite for the potency. On the other hand the position 2 plays a decisive role. References [ 1 ] V. du Vigneaud, C. Ressler, J. M. Swan, P. G. Katsoyannis and S. Gordon J. Am. Chem. Soc., 75 (1953) 4879; V. du Vigneaud, C. Ressler, J. M. Swan, C. W. Roberts and P. G. Katsoyannis, ibid., 76 (1954) 3115. [2] D. W. Urry and R. Walter, Proc. Natl. Acad. Sci. USA, 68 (1971) 956. [3] V. J. Hruby, M. Lebl, In CRC Handbook of Neurohypophyseal Hormone Analogs, K. Jost, M. Lebl, F. Brtnik (Eds), CRC Press, Boca Ralton, FL, Vol. 1 (1987) 105–155; P. Malon, ibid. Vol. 1(1987)211. [4] R. Walter, G. Stahl, Th. Caplaneris, P. Cordopatis and D. Theodoropoulos, J. Med. Chem., 22 (1979) 890; P. S. Hill. W.Y. Chan and V. J. Hruby, Int. J. Pept. Prot. Res., 38 (1991) 32. [5] P. S. Hill, D. D. Smith, J. Slaninova and V. J. Hruby, J Am. Chem. Soc., 112 (1990) 3110. [6] S. Phaneuf, N. Europe-Firmer, M. Varney, I. Z. Mackenzie, F. I. Watson and A. Lopez-Bernal, 7. Endocrinol, 136 (1993) 497. [7] W.Y. Chan, J. Pharmacol. Exp. Ther., 213 (1980) 575. [8] B. F. Mitchell, X. Fang and S. Wong, Reviews of Reproduction, 3 (1998) 113. [9] M. D. Shenderovich, K. E. Kover, S. Wilke, N. Collins and V. J. Hruby, J. Am. Chem. Soc.. 119 (1997)5833. [10]T. Kimura, O. Tanizawa, K. Mori, M. J. Brownstein and H. Okayama, Nature, 356 (1992) 526. [ 11]S. R. Hawtin, H. C. Howard and M. Wheatley, Biochem. J., 354 (2001) 465. [12]P. Melin, J. Trojnar, B. Johansson, H. Vilhardt and M. Akerlund, J. Endocrinol, 111 (1986) 125. [13]M. Lebl, In CRC Handbook of Neurohypophyseal Hormone Analogs, K Jost, M. Lebl, F. Brtnik (Eds), CRC Press, Boca Ralton, FL, Vol 2 (1987) pp. 17-74; N. Assimomytis, E. Manessi-Zoupa and P. Cordopatis, Collect. Czech. Chem. Commun., 59 (1994) 718; N. Assimomytis, V. Magafa. D. Theodoropoulos, P. Cordopatis and J. Slaninova, Lett. Pept. Sci, 3 (1996) 217. [14]J. Hlavacek and 1. Fric. Collect. Czech. Chem. Commun., 54 (1989) 2261. [15]M. Lebl, J. Pospisek. J. Hlavaek, T. Barth, P. Malon, L. Servitova, K. Hauzer and K. Jost. Collect. Czech. Chem. Commun., 47 (1982) 689.

M. Fragiadaki et al. /New Oxytocin Analogues

[16] I. Photaki,, C. Tsougraki, and C. Kotsira-Engonopoulos, Int. J. Pept. Prot. Res., 13 (1979) 426. [17]B. G Fields and L. R. Noble, Int. J. Pept. Prot Res., 35 (1990) 161. [18]K. Barlos, O. Chatzi, D. Gatos and G. Stavropoulos, Int. J Pept. Prot. Res., 37 (1991) 513. [19]M. S. Bernatowitz, S. B. Daniels and H. Koster, Tetrahedron Lett., 30 (1989) 4645. [20] D. Sarantakis, J. Teichnan, E. L. Lien and R.L. Fenichel, Biochem. Biophys. Res. Commun., 73 (1976) 336; W. Koning and R. Geiger, Chem. Ber., 103 (1970) 788. [21]E. Kaiser, R.L. Colescott, C.D. Bossinger. and P.I. Cook, Anal. Biochem., 34 (1970) 595. [22] J. P Tarn, C.-R. Wu, W Liu, and J.-W. Zhang, J. Am. Chem. Soc., 113 (1991) 6657. [23] P. Holton, Brit. J. PharmacoL, 3 (1948) 328-334; R. A. Munsick, Endocrinology, 66 (1960) 451. [24] J. Slaninova, In Handbook of Neurohypophysial Hormone Analogs, K. Jost, M. Lebl & F. Brtnik (Eds), CRC Press, Boca Raton, Vol I (1987) 83. [25] J. Dekanski, Br. J. PharmacoL, 1 (1952) 567. [26] U. Piantini, O.W. Sorensen &. R.R Ernst, J. Am. Chem. Soc., 104 (1982) 6800. [27]Bax & D.G. Davis, J. Magn. Reson., 65 (1985) 355. [28]D. Marion & K. Wuthrich, Biochem. Biophys. Res. Commun., 113 (1983) 967; J. Jeener, B.H. Meier, P. Badimann & R.R. Ernst, J. Chem. Phys., 71 (1979) 4546. [29] S. Macura, K. Wuthrich &. R.R. Ernst, J. Magn. Reson., 47 (1982) 351. [30] K. Wuthrich, In NMR of proteins and nucleic acids, Wiley, New York, 1986. [31]C. Eccles, P. Guntert, M. Billeter & K. Wuthrich, J. Biomol. NMR, 1 (1991) 111. [32] V. J Hruby, In the book: Biochemical Actions of Hormones, G. Litwack (Ed.), Academic Press, New York, (1986) 191. [33] Y. F. Ting, C. Smith, G. Stahl, R. Walter, P. Cordopatis and D. Theodoropoulos, J. Med. Chem., 23 (1980) 693; M. Lebl, J. Slaninova and R. L. Johnson, Int. J. Pept. Prot. Res., 33 (1989) 16. [34] J. Hlavacek, J. Pospisek, J. Slaninova, W.Y. Chan and V. J. Hruby, Collect. Czech. Chem. Commun., 52(1987)2317. [35] Z. Grzonka, F. Kasprzykowski, L. Lubkowska, K. Darlak, T. A. Halm and A. F. Spatola, Peptides Res., 4 (1991) 270. [36]G. K. Toth, K. Bakos, B. Penke, I. Pavo, C Varga, G. Torok, A. Peter and F. Fulop, Bioorg. & Med. Chem. Lett., 9 (1999) 667. [37] G. Flouret, W. Brieher, K. Mahan, J. Med. Chem., 34 (1991) 642; M. Lebl, G. Toth, J. Slaninova and V. J. Hruby. Int. J. Pept. Prot. Res., 40 (1992) 148. [38]R. Jezek, M. Zertova, J. Slaninova, P. Majer, Z. Prochazka, Coll. Czech. Chem. Commun., 59 (1994) 1430. [39] P. Giintert, C. Munenthaler and K. Wuthrich, J. Mol. Biol., 273 (1997) 289.

225

This page intentionally left blank

Part IV.

Molecular Targets and Drug Design

This page intentionally left blank

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

229

Molecular Aspects of Human Abdominal Aortic Aneurysm: Differential Expression of Genes Coding for Extracellular Matrix Proteoglycans A. D. Theocharis and N. K. Karamanos Department of Chemistry, Section of Organic Chemistry, Biochemistry and Natural Products, University of Patras, 261 10 Patras, Greece Abstract. Abdominal aortic aneurysm (AAA) is a common disease of human aorta with increased incidence the last decades. It is a complication to atherosclerosis and is closely associated with alterations in matrix macromolecules. Proteoglycans (PGs) are quantitatively minor components of aortic wall but are of importance for matrix assembly and the biomechanical properties of the tissue. The expression of the extracellular matrix arterial PGs (versican, biglycan and decorin) is reported. Analysis was performed by reverse-transcriptase polymerase chain reaction. The mRNAs coding for versican isoforms V0 and V1 were identified in both tissues, whereas V2 was absent. The expression of V0 was decreased 40% in aneurysmal vessel wall, whereas that for V1 remained constant. The level for the mRNA coding for biglycan was found to decreased 60% in AAA, while the biosynthesis of decorin remained constant in both tissues. It is suggested that the modulated biosynthesis of PGs in combination with the failure of elastin fibers are crucial factors contributing to the altered viscoelastic and compressive properties of the tissue and thereby to the deformity and dilatation of aorta.

Atherosclerosis and Abdominal Aortic Aneurysm: Atherosclerosis is the most common disease of blood vessels and the prime disorder leading to death and serious morbidity in the Western world [1]. The disease affects primarily the intima layer of both muscular and elastic arteries. The earliest type of atherosclerotic lesion, called fatty streak, is an inflammatory lesion consisting only of monocyte-derived macrophages and T-lymphocytes. Although different theories have been proposed for the atherogenesis, the hypothesis of the response to injury dominates. This theory proposes that endothelial denudation is the first

230

A.D. Theocharis andN.K. Karamanos / Human Abdominal Aortic Aneurysm

step of the disease, but the most recent version of this hypothesis emphasizes endothelial dysfunction rather than endothelial denudation. Several causes suspected to be responsible for endothelial dysfunction including elevated and modified low density lipoproteins (LDL), free radicals caused by cigarette smoking, elevated plasma homocysteine concentrations, hypertension, infections by microorganisms such as herpes virus or Chlamydia pneumoniae and other factors [2]. The chronicity of the inflammation results in emigration of numerous inflammatory cells from the blood within the lesion. Activation of these cells leads to release of several proteases, cytokines, chemokines and growth factors, which induce further damage and eventually lead to focal necrosis. The repeated cycles of emigration and proliferation of smooth muscle cells (SMCs) and accumulation of fibrous tissue and inflammatory cells lead to further enlargement and formation of advanced complicated lesion [3]. In this case intima layer is markedly thickened and consisted by a raised focal plaque, which has a lipid core covered by a fibrous cap and surrounded by increased number of SMCs, inflammatory cells and depositions of extracellular matrix components. The atherosclerotic lesion occludes the lumen of the vessel leading to various complications of the disease, such as ischemia of the heart, brain, or extremities and further, in infarction [2]. In these areas, where atherosclerotic occlusive disease (AOD) developed, aneurysmal dilatation often occurs with increasing age [4]. The most common type of aneurysm in human is the abdominal aortic aneurysm (AAA), which is constantly accompanied by severe atherosclerosis and often is referred as atherosclerotic aneurysm [5]. Zarins et al. [4] found that primates fed an atherogenic diet developed only occlusive atherosclerotic plaque during the first year, but aneurysmal dilatation occurred thereafter at sites of previous occlusive lesions. It is suggested that the chronicity of the disease is important for the development of the atherosclerotic aneurysm. This suggestion is consistent with the clinical observation that patients with AAA are approximately a decade older than those with AOD. Aneurysmal aorta is characterized by alterations of aortic wall architecture, which is mainly due to the destruction of media layer and the failure of elastic fibers, whereas the collagen fibers seems to affected only in ruptured aneurysms [6–8]. The decrease of the number of smooth muscle cells (SMCs) in aneurysmal aortic wall due to the induction of their apoptosis suggested that significantly contribute to the development of the disease [9]. Recent studies have also demonstrated that PGs and glycosaminoglycans (GAGs) present in aneurysmal aortas undergo marked and specific alteration in their contents and structures [10–12]. The overall GAG content significantly decreased (60%) in AAA as compared to normal aortas. A 90% decrease in heparan sulfate (HS) content and 65% and 73% in chondroitin sulfate (CS) and hyaluronan (HA), respectively was recorded. Dermatan sulfate (DS) was slightly decreased (8%) in AAA. Fine structural alterations in respect to disaccharide composition of DS and HS were also detected in AAA [10]. The differences in the composition of matrix molecules occurred in the diseased tissue may in part be attributed to the increased degradation of these molecules by several matrix metalloproteases released by SMCs and inflammatory cells [13–15]. Functional Roles of Proteoglycans in Aortic Wall:Although PGs / GAGs are quantitatively minor structural components of the aortic wall, they influence its physicochemical properties and participate in several physiological events. Versican, the large molecular size chondroitin sulfate proteoglycan (CSPG), is the most abundant PG of the arterial wall and is mainly localized in the arterial intima. This PG is synthesized predominantly by vascular SMCs and it is composed of a core protein, carrying several covalently bound CS chains. Versican, through its CS chains, is implicated in the entrapment of LDL in the arterial wall during the atherogenesis [16]. There are at least four possible splice variants of mammalian

A.D. Theocharis and N.K. Karamanos / Human Abdominal Aortic Aneurysm

231

versican - V0, V1, V2 and V3 - varying in length of their glycosaminoglycan attachment domain. Only the largest isoforms - V0 and V1 - have been identified in the arterial wall. Vo includes two such domains (GAG-a and GAG-J3), which provide attachment sites for GAG chains, while V1 include only the GAG-a domain [17]. Interactions of versican with hyaluronan and link proteins lead to formation of large aggregates that trap water creating viscoelastic and turgor pressures within the arterial wall. These aggregates are considered to create reversibly compressive structures necessary to avoid the deformity of blood vessels by the pulsatile pressures of the circulatory system [16]. Biglycan and decorin belong to a class of small leucine-rich proteoglycans (SLRPs), bearing two or one DS/CS side chains, respectively. They are usually found in close association with collagen type I and III and are thought to play important roles in stabilization and proper arrangement of collagen fibers [17]. SLRPs can also bind via their core proteins the transforming growth factor beta (TGF-P), regulating its bioavailability in this tissue. The DS chains of biglycan and decorin exhibit antithrombotic properties since can bind heparin cofactor II (HC II) accelerating the inhibition of thrombin by this complex [18]. Furthermore, several reports have demonstrated that chondroitin sulfated at the C-4 of galactosamine (C4S) and DS exhibit antiatherogenic properties, since protect LDL and high density lipoproteins (HDL) against copper induced oxidation [19, 20]. Heparan sulfate proteoglycans (HSPGs) is another class of arterial PGs present in basement membranes of endothelial cells playing important roles as tissue barrier. The HSPGs is present also on cell membrane of arterial cells and can act as low affinity receptors for basic fibroblast growth factor (bFGF) facilitating the internalisation of this growth factor. HSPGs are also powerful anticoagulant molecules and the basis for this activity lies in the ability of their GAG chains to bind serine protease inhibitors such as antithrombin III (AT III) and HC II [21]. Expression of Extracellular Matrix Proteoglycans: Aneurysmal abdominal aortic tissues (n = 12, males aged 55-80 years) were derived at aortic reconstruction surgery, performed in the University Hospital of Patras, Patras, Greece. Macroscopically normal abdominal aortas (n - 4 males aged 30-45) were obtained from normal organ transplant donors and were free of macroscopically observable disease. Outer adventitia and mural thrombi were removed from the remaining portions of aneurysmal aortas. Outer adventitia was also removed from normal aortas. Total RNA was isolated from aortic tissue samples by the quanidine thiocyanate / phenol / chloroform method [22] and the DNA impurities were removed following digestion with DNase I [12]. The RNA was isolated following addition of phenol / chloroform and consequently centrifugation at 10 000 x g for 20 min at 4 °C. The aqueous phase, containing the RNA, was precipitated by the addition of an equal volume of isopropanol. The yield and the purity of the RNA preparations were determined by measuring the absorbance at 260 nm and estimating the ratio A 260 nm/ A 280 nm , respectively. The expression of extracellular matrix proteoglycans was analysed by RT-PCR of the total RNA preparations. Equal amounts (2 u,g) of total RNA from each tissue samples were used for first strand cDNA. The entire reverse transcription products served as the template to PCR amplification of PGs transcripts using specific primers for human extracellular matrix PGs (versican isoforms, biglycan and decorin) and the stable expressed gene of GAPDH. PCR amplifications were performed for 40 cycles (Table 1), except of those of GAPDH, which performed for 28 cycles. PCR reaction includes 1 min denaturation at 94°C, 1 min in annealing temperature and 2 min at 72°C for extension. Aliquots (16 u.L) of the amplification products were electrophoresed in 2 % (w/v) agarose gels containing 1 |4,g/mL ethidium bromide, using low molecular weight DNA markers as standards. Bands were visualized on a UV lamp and gels were photographed with a Kodak

232

A.D. Theocharis and N.K. Karamanos /Human Abdominal Aortic Aneurysm

camera. The amounts of PCR products obtained were determined by measuring the fluorescence of bands, using the digital image analysis program [12]. Table 1: Sequence of oligonucleotide primer and PCR conditions.

PGs Versican V0 Versican

v,

Versican V2 Decorin Biglycan GAPDH

Upstream primer GACCTCAGGCG CTTTC GCGCCACCCTG TGAC GACCTCAGGCG CTTTC GGGGCAAAAC ACCAGTCC TGCAGAACAAC GACATCTCCG ACATCATCCCT GCCTCTACTGG

Downstream primer CAGTGGTAACG AGATGCTTC CAGTGGTAACG AGATGCTTC TAGCACTGCCC TTGGA TATTTCCTAAG CCCAGCCTCTC GGAGGAGCTTG AGGTCTGGGA AGTGGGTGTCG CTGTTGAAGTC

4174–4524

Anneal. Temp. (°C) 50

Length (bp) 351

1178–1563

50

386

4174–4547

52

373

417-834

60

413

409-1027

60

618

608-868

-

261

Region

In both normal and aneurysmal aortas mRNA encoding for Vo and V1 versican isoforms was detected, whereas V2 isoform was not expressed in any of the preparations. As estimated by the semi-quantitative RT-PCR the cells from AAA expressed much less Vo isoform (40 %), as compared with normal aortas (fluorescence 0.58 ± 0.06 and 0.35 ± 0.04, respectively) (Fig 1). On contrary, V1 isoform was similarly expressed in both normal (0.24 ± 0.03) and aneurysmal aortas (0.25 ± 0.03) (Figs 1 and 2). The lower amounts of versican found in aneurysmal aortas thus correlate with a specifically lower expression of the V0 isoform.

Figure 1. Screening of gene expression coding for the extracellular matrix PGs at the mRNA level studied by RT-PCR. Lanes 1, 2 indicate two out of four representative cases of normal aortas, and lanes 3-7 five out of twelve cases of aneurismal aortas. The amplification of GAPDH reference gene was the internal standard for the method.

The analysis of both SLRPs (biglycan and decorin) in both tissues revealed that these two molecules were differentially regulated in AAA. Analysis showed that arterial cells in AAA

A.D. Theocharis and N.K. Karamanos /Human Abdominal Aortic Aneurysm

233

transcribed substantially less biglycan (fluorescence 0.71 ± 0.075) than those in normal aortas (fluorescence 1.78 ± 0.15) (Fig 2). In contrast, the biosynthesis of decorin was not affected in AAA, since decorin was found to express constantly in both tissues (fluorescence 0.31 ± 0.04 in normal aortas and 0.30 ± 0.04 in aneurysmal aortas) (Fig 2).

Figure 2. Variation of mRNA coding for matrix PGs estimated by RT-PCR analysis and determined by measuring the fluorescence in normal (n = 4) and aneurysmal aortas (n = 12). Asterisks indicate significantly different values (P < 0.001) as compared to normal aortic tissue.

Discussion - Concluding Remarks The architecture of the aortic wall is responsible for the biomechanical properties of the tissue. The extracellular matrix components elastin collagen and PGs provide the aortic wall with mechanical strength and elasticity, properties which needed to accommodate the pulsatile pressures of the circulatory system. Analysis for the mRNA encoding for versican by RT-PCR revealed that normal and aneurysmal aortas express only V0 and V1 versican isoforms, whereas V2 isoform was not detected in any case. A specific down regulation for the largest isoform V0 (40%) demonstrated in AAA in comparison to normal tissue. In contrast, the V1 isoform was constantly expressed in both tissues. The V0 includes both GAG-a and GAG-P domains providing attachment sites for up to 30 CS/DS chains, whereas the V1 isoform includes only the GAG-P domain with the fully glycanated form to contain up to 15 CS/DS chains [17]. Decreased concentration of versican with increased proportion of the V1 isoform, as seen in AAA would create an extracellular matrix with lower swelling pressure and decreased viscoelastic properties. It has been also proposed that the different isoforms of versican expressed in distinct and temporal patterns are related to specific properties of the tissues. The decreased expression of versican in AAA may be a response to various cytokines and growth factors present in diseased tissue. It is well known that several growth factors and cytokines influence the expression of versican and may also regulate the transcription of versican isoforms. For example, TGF-P and PDGF stimulate the expression of versican mRNA by SMCs in vitro [23], whereas interleukine-lbeta (IL-1P) and insulin growth factor (IGF-1) decrease the synthesis of this molecule [24, 25].

234

A. D. Theocharis and N. K. Karamanos / Human Abdominal Aortic Aneurysm

Significant decrease (60%) in the expression of biglycan mRNA was also observed in AAA. Our results are in agreement with a previous study [26], which demonstrated a more marked decrease in the expression of biglycan in AAA, whereas the expression of decorin remained unchanged. Biglycan binds a variety of matrix molecules, such as collagen types I, III and V, participating in the matrix assembly [27]. The decreased expression of biglycan in AAA may result in anomalous arrangement of collagen fibrils and in alterations in the mechanical properties of the fibrils. Furthermore, significant decrease in the antithrombotic properties in AAA is strongly suggested due to the limited expression of biglycan, since it is known that biglycan and deorin either bound to collagen type V or free in solution accelerate the HC II - thrombin inhibition reaction [18]. The elimination of the expression of biglycan in AAA may be regulated by the cytokines and growth factors found in the tissue. The maintenance of the expression of mRNA encoding for decorin in AAA in similar levels to that of normal aortas is in agreement with other investigators [26]. Decorin is directly implicated in collagen type I fibriilogenesis, assisting in the correct positioning of the collagen molecule in the staggered conformation of the fibril [17], The molar ratio of decorin to collagen is important for the proper formation of the matrix architecture. The elevated collagen amounts in aneurysmal aortas taking together to the unchanged biosynthesis of decorin suggest imbalance in this ratio, which probably results in disordered collagen deposition. The latter is also suggested by other studies, which demonstrate increased extractability of collagen and decorin in AAA in comparison to normal aortas [28, 29]. The alteration in conformation of collagen fibrils may further influence the mechanical properties of the aneurysmal aortic wall. In conclusion, in this article we demonstrate that the expression of the genes of matrix PGs is significantly affected in AAA. This modulated biosynthesis of PGs in combination with the failure of elastin fibbers, results to functional alterations and reduced mechanical resistance of the aortic wall, contributing in the development of aneurysmal dilatation.

References [1]TN Wight, Arteriosclerosis 9 (1989) 1. [2] R Ross, New Engl J Med 340 (1999) 15. [3] R Ross, Nature 362 (1993) 801. [4] CK Zarins, S Glagov, D Vesselinovitch and RW Wissler, J Vasc Surg 12 (1990) 246. [5] CB Ernst, N Engl J Med 328 (1993) 1167. [6] JM Reilly and MD Tilson, Incidence and etiology of abdominal aortic aneurysms. In GE Pierce (ed.) The surgical clinics of North America: Abdominal aortic aneurysms. W.B. Saunders Company, Vol 69 (4) 1989. [7] JS Campa, RM Greenhalgh and Powell JT, Atherosclerosis 65 (1987) 13. [8] S Menashi, JS Campa, RM Greenhalgh and JT Powell, J Vasc Surg 6(6) (1987) 578. [9] A Lopez Candales, DR Holmes, S Liao, MJ Scott, SA Wickline and RW Thompson, Am J Pathol 150 (1997)993. [ 10] AD Theocharis, I Tsolakis, T Tsegenidis and NK Karamanos, Atherosclerosis 145 (1999) 359. [11] F Lamari, AD Theocharis, A Hjerpe and NK Karamanos, J Chromatogr B 730 (1999) 129. [ 12] AD Theocharis, I Tsolakis, A Hjerpe and Karamanos NK, Atherosclerosis 154 (2001) 367. [ 13] RW Thompson, DR Holmes, RA Mertens, S Liao. MD Botney, RP Mecham, HG Welgus and WC Parks, J Clin Invest % (1995) 318. [14] I Halpert, UI Sires, JD Roby, S Potter-Perigo. TN Wight, SD Shapiro, HG Welgus, SA Wickhne and WC Parks, Proc Natl Acad Sci 93 (1996) 9748. [15] T Freestone, RJ Turner, A Coady. DJ Higman, RM Greenhalgh and JT Powell. Arterioscler Thromb Vasc Biol 15 (1995) 1145. [ 16] TN Wight, Arterial wall. In WD Comper (ed.) Extracellular Matrix. Tissue Function. OPA. Vol. 1. 1996, 175-202.

A.D. Theocharis and N.K. Karamanos /Human Abdominal Aortic Aneurysm

235

[ 17J RV lozzo, Ann Rev Biochem 61 (1998) 609. [ 18) RA Shirk, N Parthasarthy, JD San Antonio, FC Church and WD Wagner, J Biol Chem 275 (2000) 18085. [ 19] R Albertini, P Ramos, A Giessauf, A Passi, G De Luca and H Esterbauer, FEBS Letters 403 (1997) 154. (20] R Albertini, G De Luca, A Passi, M Moratti M and PM Abuja, Arch Biochem Biophys 365 (1999) 143 [21] L Kjellen and U Lindahl, Ann Rev Biochem 60 (1991) 443. [22] P Chomczynski and N Sacchi, Anal Biochem 162 (1987) 156. [23] E Schonherr, HT Jarvelainen, LJ Sandell and TN Wight, J Biol Chem 266 (1991) 17640. [24] EE Owarnstrom, HT Jarvelainen, MG Kinsella, CO Ostberg, LJ Sandell, RC Page and TN Wight, Biochem J 294 (1993) 613. [25] PY Davis, CR Frazier, JR Shapiro and NS Fedarko, Biochem J 324 (1997) 753. [26] NA Tamarina, MA Grassi, DA Johnson and WH Pearce, J Surg Res 74 (1998) 76. [27] E Schonherr, P Witsch-Prehm, B Harrach, H Robenek, J Rauterberg and H Kresse, J Biol Chem 270 (1995)2776. [28] M Spina, AM Degani, A Stella, M Gargiulo, G Guiducci and G Genacchi, Biology and Physiology of the Extracellular Matrix. In C Balduini, P Cherubino, G De Luca (eds.) La Goliardica Pavese, Pavia, 1996, pp. 219-224. [29] AD Theocharis, A Hjerpe, G De Luca and NK Karamanos, Submitted (2001).

236

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

Advances in Botulinum Neurotoxin Inhibitors based on the SNARE Motif G. J. Moore, D. Moore, S. Roy, L. J. Hayden and G. Murray Hamilton. Pepmetics Inc., 1762-1st Avenue N.W., Calgary, Alberta, Canada T2NOB1 and Therapy Group, Defence Research Establishment Suffield, Medicine Hat, Alberta, Canada T1A 8K6 Abstract. Botulinus neurotoxins are zinc proteases which cleave the SNARE proteins of synaptic vesicles thereby preventing neurotransmitter release. The SNARE complex is a four helix bundle made up of three proteins which each contain the repeating SNARE motif AAxxA, where A is Asp or Glu and x is any amino acid. In the present report 'hinge' peptide libraries were constructed based on the SNARE motif and shown to inhibit Botulinum neurotoxin.

Introduction Botulinum neurotoxin A (Botox A) is one of several protein toxins from clostridium botulinum which cause paralytic syndromes resulting from blockade of neurotransmitter release. These toxins are all zinc endopeptidases acting in the neuronal cytosol: Botox B, D, F and G as well as tetanus toxin attack specifically VAMP (also called synaptobrevin) - a protein of synaptic vesicles, whereas Botox A and E cleave SNAP-25 and Botox C acts on syntaxin - both proteins of the presynaptic membrane SNAP-25 is one component of the so-called SNARE complex which is responsible for docking synaptic vesicles and fusion to the cell membrane as the immediate precursor event to transmitter release. The SNARE complex is a four-helix bundle made up of two small proteins, namely VAMP and syntaxin, and a larger protein SNAP-25 which provides two of the four threads of the helix bundle. SNAP-25 also contains a lipid anchor region (amino acids 85-92) between the two helical threads, which lines up with transmembrane domains at the C-termini of both VAMP and syntaxin [1]. The botulinus neurotoxins each cleave selectively a different peptide bond within one of the three target proteins which comprise the SNARE complex.. Whereas Botox A selectively cleaves the Glnl87-Arg203 bond near the C-terminus of SNAP-25, Botox B cleaves Q -F77 of VAMP, Botox C cleaves K253-A254 of syntaxin, Botox D cleaves K59-L60 of VAMP. Botox E cleaves R180I181 of SNAP-25, Botox F cleaves Q58-K59 of VAMP, and Botox F cleaves A81-A82 of VAMP.

G.J. Moore et al. /Advances in Botulinum Neurotoxin Inhibitors

237

A repeating motif exists within the sequences of all three of the proteins of the SNARE complex which, when introduced in the form of synthetic peptides of about ten residues in length, inhibits the actions of botulinus toxins on the SNARE complex [2]. For example one version of the repeating motif present in VAMP, designated V2 and having the sequence E62LDDRADALQ71, blocks the neurotoxic actions of Botox A and B when the peptide is injected into cultured Aplysia neurons. The significance of the repeating SNARE motif, which appears twice on each of the four threads of the SNARE helix bundle upstream of the cleavage sites, is not well understood, although it presumably acts as a recognition site for binding of some other biomolecule(s), and may also be used by Botox as a binding-recognition element. In agreement with this, cross recognition of the target proteins by the various toxins occurs: Botox A inhibits VAMP proteolysis by Botox B, and Botox B and tetanus inhibit the cleavage of SNAP-25 by Botox A. Moreover serum albumin, which contains SNARE motifs within helical regions of its secondary structure (D183ELRD187 and D255DRAD259), inhibits the cleavage of synthetic substrate by Botox A (present report).. Thus although the mechanism of action of V2 (and other variants of the SNARE motif) is not proven, it could involve binding to a complementary recognition site on Botox resulting in inhibition of productive binding of Botox to the SNARE complex. Interestingly, possible complementary SNARE motifs (K592KVNK596 and K701RNEK705 in the heavy chain, and K335LKFDK340 and K359VLNRK364 in the light chain) exist in Botox A. It has been suggested that the distance between the SNARE motif and the cleavage site may be a determining factor for recognition/cleavage by the various Botox enzymes [3]. Strategy: Several possible strategies for inhibiting Botox exist: block the transmembrane domain of Botox and prevent cell entry, e.g. complementary peptides, antibodies; block the proteolytic site of Botox, e.g. substrate inhibitors, selective chelation of zinc; block the interaction of Botox with binding-recognition motif, e.g. mimics of SNARE motif which bind to complementary site(s) on Botox. The first strategy has potential application prior to exposure to toxin (e.g. immunization), as well as for the deactivation of toxin in body fluids after exposure. The latter strategies, on the other hand, have the potential for providing total body treatment after toxin exposure - assuming that the therapeutic agent is able to cross cell membranes and inactivate toxin which has already entered the cell. Whereas agents based on active site inhibitors will have to be tailored to individual variants of Botox, therapies based on the SNARE motif have the potential to treat poisoning by all forms of Botox as well as tetanus. Regardless of mechanism, one approach to producing improved inhibitors of Botox is to identify the structural elements of V2 (and other SNARE motifs) which make it an effective inhibitor and then reconstruct these components into a smaller and preferably nonpeptidic molecule which would be able to traverse membanes, thereby providing not only access to the inside of cells but also the potential for oral activity. A streamlined and efficient approach [4,5] to this goal is to create a library of small semimimetic peptides containing the essential elements of the repeating motif in the SNARE complex, identify the most active peptide in the mixture by iterative deconvolution of the library, and then restructure the best semimimetic peptide identified into a fully fledged nonpeptide mimetic using computer molecular modelling techniques based on structural details derived from Xray crystallography studies of Botox bound to SNARE motif. The important structural feature of the SNARE motif is comprised of an amino acid sequence made up of residues: A-A-x-x-A-x-x where A=acidic residue and x=nonpolar or polar residue.

238

G.J. Moore et al /Advances in Botulinum Neurotoxin Inhibitors

In the helical arrangement of these groups within the secondary structure of the SNARE protein complex from which they are derived, the end result is a cluster of three neighbouring negative charges juxtaposed by a nonpolar moiety. In other words, the required motif could be envisaged as three negatively charged groups mounted in close proximity on a hydrophobic template. A number of commercially available compounds which fit this general scheme are being investigated for inhibitory activity in our laboratory, together with several peptide libraries which have been designed and synthesized with these considerations in mind. On the other hand, recent evidence basd on X-ray crystallography studies on Botox B bound to its substrate VAMP/synaptobrevin, suggest that Botox enzymes attack their substrates, not when they are present in the four helix bundle formation of the SNARE complex, but rather when they are in random coil or extended structures [3]. Accordingly the triad of negative charges which comprise the SNARE motif would not necessarily exist in close proximity, and in this regard it is significant that our "hinge" peptide libraries are constructed so that they can assume extended as well as folded structures, capable of mimicking any of the possible relevant conformational scenarios for the SNARE motif. Design and Synthesis of Peptide Libraries: For example, the following library has been synthesized which is designed to mimic the SNARE motif: SN-1 = Ac -X1-X2-LINKER-X3-X4-NH2, where Xi, X2, X3 and X4 are mixtures of Asp, Glu, Gln and Arg, and the LINKER group is 4-aminobutyric acid The rationale for the design of the peptide minilibrary SN-1 is based on the preponderance of acidic residues in SNARE motifs, together with the occurrence of other residues in SNARE motifs which can provide solubility properties to the libraries - hence the inclusion of Arg and Gln in SN-1. The multiple negative charges represented by the Asp and Glu residues, when mounted around the flexible linker group 4-aminobutyric acid (Abu), should engender structural and conformational variations on the SNARE motif which will inhibit Botox Coincidentally the Gln and Arg residues in SN-1 are also representative of the scissile bond (Q-R) in the Botox A substrate, so that the peptide minilibrary SN-1 doubles in a rudimentary way as a substrate mimic. This parallels the classical approach to inhibitor discovery, which is usually done by minor modification of substrate structure leading to compounds which bind to the active site but are not cleaved by the enzyme. Accordingly, incorporating residues of the scissile bond (Gln-Arg) into peptide libraries such as SN-1 should provide structural and conformational characteristics which might result in substrate-based inhibitors. Clearly SN-1 contains the potential for containing inhibitors based not only on the SNARE motif but also on substrate structure. Another consideration for the design of peptide libraries with inhibitory potency is the inclusion of residues which would be expected to coordinate to the active site zinc atom of Botox, namely Cys and His as well as acidic residues. The approach here is to create a peptide library containing an array of zinc binding elements in a variety of different formats, wherein there should be individual peptides which recognise not only the zinc atom but also unique aspects of the active site of Botox A in which the zinc atom resides. The intent is to produce zinc targeted inhibitors which show selectivity for Botox over other zinc metalloenzymes. An example of a library targeted to produce active site zinc inhibitors is:SN-3 : Ac-X 1 -X 2 -Abu-X 3 -X 4 -NH2 where Ac = acetyl, Abu = 4-aminobutyric acid, and X1, X2, X3 and X4 are mixtures of the amino acids Asp, Glu, His and Cys.

G.J. Moore et al. /Advances in Botulinum Neurotoxin Inhibitors

239

This library contains both active site zinc directed probes and variations on the SNARE motif within the same basis set (library) of peptides. Inhibitors of Botox A: In our studies a number of compounds (Table 1) were found to inhibit the Botox A mediated cleavage of the 17-amino acid synthetic peptide substrate Ac-SNKTRIDQANQRATKML-NH2, which derives from the C-terminal part (residues 187-203) of SNAP-25 and contains the scissile Q-R peptide bond targeted by Botox A. Several thiol-containing compounds, namely dithiothreitol (widely used to reduce disulphide bonds in proteins), DMPS (a potent chelator of heavy metals) and Captopril (a clinical inhibitor of the zinc dipeptidase Angiotensin Converting Enzyme) block the cleavage of this substrate. However the non-thiol ACE inhibitor Lisinopril was ineffective, whereas the non-thiol prodrug ACE inhibitor enalapril was an effective inhibitor. The thiol compounds DMPS, captopril and DTT presumably act by a mechanism involving sequestration of the zinc atom at the active site of Botox A. The inhibitory activity of enalapril was unexpected but could suggest an affinity for the active site of Botox A which is not shared by lisinopril. Table 1. Inhibition of the Botox A cleavage of synthetic heptadecapeptide substrate

Inhibitor Dithiothreitol

Concentration (mM) % Inhibition I 26

-SH Concentration (mM)

DMPS*

0.1

72

0.2

Captopril

1

76

1

Lysinopril

5

7

0

Enalapril

5

40

0

V2(DDRADALQ)

5

40

0

Glu-Glu-Glu

5

22

0

Glycyrrizic acid

5

55

0

LibrarySN-1

5 (20uM each)

16

0

Library SN-3

0.5 (2uM each)

51

0.125

2

The synthetic peptide V2 was also able to inhibit Botox A, but at a higher concentration (5mM) than has been observed previously in vivo when SNAP-25 was the substrate (l00uM, see ref [2]). This may reflect the absence of SNARE motif recognition site in the short synthetic substrate used for the present studies. Interestingly serum albumin (l0uM), which contains the SNARE motif in duplicate, inhibited cleavage of SNAP-25(187–203) under our assay conditions which had been optimized for zinc (1 uM) and DTT (0.5 mM) concentrations. In contrast Schmidt & Bostian [7] observed that in the presence of 250uM zinc and 5mM DTT cleavage of this substrate by Botox A only occurred in the presence of BSA; our findings would suggest that this was probably due to sequestration of inhibitory levels of zinc/DTT by BSA. Investigations of SN-1 on the Botox A cleavage of substrate have shown that this peptide minilibrary does inhibit the enzyme activity (Table 1). This suggests that variations on the SNARE motif present in the library mixture may interfere with binding of Botox to the substrate. However a series of peptides which were selected for their potential to represent simple variations on the SNARE motif, i.e. Glu-Glu, Glu-Glu-Leu, Glu-Glu-Glu and GluPro-Glu-Thr, were generally inactive (data not shown), with the notable exception of Glu-

240

G.J. Moore et al. /Advances in Botulinum Neurotoxin Inhibitors

Glu-Glu. A number of peptides with potential complementary sequences of the SNARE motif, i.e. Lys-Arg, Lys-Lys, Orn-Om, Lys-Lys-Lys, Om-Orn-Om and Lys-Phe-Gly-Lys, were inactive as expected because the synthetic 17mer peptide substrate used for the assay lacks the repeating SNARE motifs that are present in the longer SNAP-25 natural substrate. However these basic peptides would be expected to inhibit the cleavage of SNARE proteins by Botox enzymes in general, if they bind well to the SNARE recognition motifs in these proteins. Interestingly glycyrrhizic acid, which is a steroid glycoside containing 3 carboxylate groups, was able to inhibit Botox (Table 1). This finding, as with Glu-Glu-Glu, suggests that certain configurations of negative charges are able to approximate the SNARE motif and inhibit Botox A. However in the final analysis it would appear that the most potent inhibitor of Botox containing a SNARE motif variation is likely to be found in the peptide library SN-1 because each peptide in the mixture was present at 20uM concentration (Table 1). Finally the present experimental findings also indicate that effective inhibitors of Botox A are likely to be derived from thiol containing compounds (the inhibitory activity of DMPS is particularly remarkable). In agreement with this, the peptide minilibrary SN-3 turned out to be an exceptionally potent inhibitor of Botox A (Table 1). When the overall thiol concentrations of the inhibitors are compared (Table 1) it is apparent that the peptide library SN-3 is the most potent thiol inhibitor of Botox. Furthermore it is very likely that SN-3 contains a peptide(s) which is a more potent inhibitor than DMPS. More recently we have found that an uncapped "hinge" peptide minilibrary XI-X2Abu-X3, where XI, X2 and X3 are mixtures of Asp, Glu and Arg, contains very potent inhibitors based on the SNARE motif (7). We are currently evaluating these minilibraries for inhibition of other botulinum neurotoxins, to determine if generic therapies based on the SNARE motif are present. In a much larger study, we are undertaking the synthesis of minilibrary subsets deriving from a more complete hinge peptide minilibrary comprising more than 20,000 hinge peptides, in order to ensure that optimal peptides are not overlooked. References [ 1 ] MA Poirier et al. Nature Struct Biol 5 (1998) 765. [2] O Rossetto et al, Nature 372 (1994) 416. [3] MA Hanson and RC Stevens, Nature Struct Biol 7 (2000) 687. [4] GJ Moore, Drug Devel Res 42 (1997) 157. [5] GJ Moore, Trends Pharmacol Sci 15 (1994)124. [6] JJ Schmidt & KA Bostian.,7 Protein Chem 16 (1997) 19. [7] LJ Hayden et al, J Appl Toxicol (2001) in press.

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

241

Effect of Linear and Cyclic Peptide Analogues of Myelin Basic Protein Epitope MBP72-85 on Human T-cell Activation S. Thymianou1, K. Chatzantoni1, M. Karakantza1, T. Tselios2, P. Papathanassopoulos3, J. Matsoukas2, A. Mouzaki1 'Laboratory Haematology and Transfusion Medicine, Medical School, University of Patras, GR-26110 Patras, 2 Department of Chemistry, University of Patras, GR-26500 Patras, 3 Neurology Clinic, Medical School, University of Patras, GR-26110 Patras, Greece Abstract. Multiple sclerosis (MS) is a chronic demyelinating disease of the central nervous system (CNS) mediated by helper (CD4) T-cells. We have previously shown that two linear and two cyclic peptide analogues of the guinea pig epitope 72–85 of the myelin basic protein (MBP) induce (agonists) or treat (antagonists) an animalmodel version of MS. In this report we studied whether these MBP72-85 are recognized as auto-antigens by normal peripheral blood (PB) T-cells. To this end we cultured PB T-cells in the presence of the petides at various concentrations and assessed their effect on cellular proliferation, phenotype and cytokine synthesis. The peptides, when administered separately, induced apoptosis in exvivo cells and acted synergistically with mitogens to stimulate the cells, acting through the T Cell Receptor (TCR). When administered together (agonist+antagonist), the peptides blocked T-cell proliferation completely without rendering the cells anergic. Cytokine secretion experiments showed that none of the peptides targets a specific Thl or Th2 type of T-cells. Overall, the peptides studied in this work show a promising pharmacological profile and a putative therapeutic use in the treatment of MS.

242

5. Thymianou et al. / Effect of Linear and Cyclic Peptide Analogues

Introduction Multiple sclerosis is a chronic CD4 T-cell mediated disease of the central nervous system characterized by local-cell and macrophage infiltrates, demyelination and loss of neurologic function [1,2]. MS is widely believed to be an autoimmune disease and to be triggered by CNS-specific CD4 T lymphocytes [3–6]. Candidate autoantigens include constituents of the myelin sheath such as myelin basic protein and proteolipid protein (PLP) [7]. Modem approaches towards the therapeutical management of MS involve the design and use of peptide analogues of disease-associated myelin epitopes to induce peripheral T cell tolerance [8]. Experimental autoimmune or allergic encephalomyelitis (EAE), one of the best studied experimental animal models of MS [1,9], is a useful in vivo system for the evaluation of such therapeutic approaches. In Lewis rats immunized with guinea pig MBP protein, encephalitogenic T cells which recognise the 72-85 amino acid sequence MBP772-85, dominate the immune response [10]. The linear analogue Gln-Lys-Ser-Gln-Arg-Ser-GlnAsp-Glu-Asn-Pro-Val (MBP72-85) has been found to induce EAE while substitution of the Asp residue at position 81 with Ala resulted in an analogue (Ala8lMBP72-85) which prevented the induction of EAE by its parent peptide [11,12]. Peptides that interfere with the formation of the trimolecular complex MHC-Peptide-TCR can actively inhibit disease through the activation of antigen-specific regulatory T cells [13,14]. Peptide therapy, however, is hindered due to the sensitivity of peptides to proteolytic enzymes. Continuous infusions and therefore prohibitive amounts of peptides are necessary to elicit the necessary biological response. Cyclization is known to restrict the number of possible conformations allowing to determine which of the conformations observed in solution best approximates the receptor bound conformation. To this end, we synthesized cyclic analogues [15,16] to address the need for more stable molecules, which could maintain or suppress the biological function of the original peptide and yet be able to elicit a response in pharmacological quantities. We recently showed that the cyclic MBP72-85 (c-MBP72-85) analogue has comparable potency to linear MBP72-85 in inducing EAE in Lewis rats and that the clinical and histopathological manifestations of disease induced by C-MBP72-85 are prevented by a linear antagonist (Ala8lMBP72-85) [16]. Furthermore, it was demonstrated that the encephalitogenic activity of MBP72-85 can be completely prevented by the co-injection with c-Ala8lMBP72-85 [17]. In the present study, the linear (MBP72-85 and Ala8lMBP72-85) and cyclic analogues (cMBP72-85 and c-Ala8lMBP72-85) were assessed for their biological activities in cultured human PB T-cells in order to investigate whether any of these peptides can be eventually used as an immunomodifying agent in MS.

Materials and Methods Peptides: In order to assess the biological activity of the MBP72-85 peptide analogues, the following peptides were used: Pl=Glu-Lys-Ser-Glu-Arg-Ser-Glu-Asp8l-Glu-Asn-Pro-Val Linear agonist MBP72-85. [16]; P2=Glu-Lys-Ser-Glu-Arg-Ser-Glu-Ala8l-Glu-Asn-Pro-Val Linear antagonist Ala8lMBP7285. [16]; P4=Glu-Lys-Ser-Glu-Arg-Ser-Glu-Asp8l-Glu-Asn-Pro-Val-NH2 Cyclic agonist cMBP72-85 [16]; I 1 P5= Glu-Lys-Ser-Glu-Arg-Ser-Glu-Ala81-Glu-Asn-Pro-Val Cyclic antagonist cAla81MBP72 85. [17].

I

1

As a negative control we used the peptide P3=Cyc[(D-F)LLReKDap]

S. Thymianou et al. /Effect of Linear and Cyclic Peptide Analogues

243

Cell cultivation: Heparinized venous blood was collected from three randomly selected blood donors (Blood Transfusion Center, Patras University Hospital). Mononuclear cells were prepared by centrifugation over a Ficoll-Paque gradient (Pharmacia, Sweden). The cells were cultured in RPMI1640 culture medium (CM, GIBCO BRL, Gaithersburg, MD, USA) containing 10% Fetal calf serum (PCS) (GIBCO BRL, Gaithersburg, MD, USA) and other supplements as described [18], in the presence or absence of 5 ug/ml/2xlO^ cells phytohaemaglutinine (PHA) (Sigma, St-Louis, MI, USA). When indicated, the cells were cultured in the presence of 1 ug/ml/2x!06 cells of anti-CD28 monoclonal antibody (mAb) (CALTAG Labs, Burlingame, CA, USA). When indicated the cells were cultured with varying doses of linear peptides MBP72-85 or Ala81MBP72-85 or the cyclic peptides C-MBP7285 or c-Ala81MBP72-85 or both linear or cyclic agonist and antagonist peptides at a ratio of 1:100 (see Figure legends). The number of cells was counted using a Sysmex NE-8000 counter (Japan) and their viability was estimated by the trypan blue exclusion method as described [18]. Cell surface immunophenotyping and FACS analysis: Cells obtained from 24 hours cultures were concentrated to 5xl06 cells/ml PBS (GIBCO BRL, Gaithersburg, MD, USA) and 100 ul of the cells were incubated with the following mAb: CD3-FITC, CD4-FTTC, CD8-FITC, CD4-FITC/CD25-PE, CD8-FITC/CD25-PE [19], for 30 min at 4°C. All mAbs were used at saturating conditions as instructed by the manufacturer (all mAb were from Daco A/S Denmark; the CD25-PE mAb was from Serotec ltd., Oxford, UK). Subsequently the cells were fixed using the Immunoprep reagent system of Coulter EPICS and the automated CoulterQ-Prep EPICS Immunology Workstation (Maiami, FL). Negative controls consisted of the sequential incubations as above, substituting specific antibodies with isotype-matched irrelevant controls at equivalent concentrations. The cells were analyzed using a Coulter EPICS XL-MCL. Cell debris were excluded from the analysis. For each sample 20,000 viable cells were acquired in list mode and data on these cells were processed using the EPICS-XL2 software. All the positive results were obtained by substracting the percentage of negative controls. ELISA assays: These were performed as instructed by the manufacturers. The following ELISA kits were used: Human (h) DL-2 ELISA from Boerhinger (Mannheim, Germany), hlFN-y ELISA from R&D Systems Quantikine™ (Minneapolis, USA), hIL-4 and hIL-10 from Endogen Inc. (Woburn, MA, USA). Results Effect of the peptides on human peripheral blood T-cells: The immunogenicity of the linear and cyclic peptides were tested in human adult peripheral blood mononuclear cell (PBMC) cultures in the presence or absence of the mitogen PHA. Dose response curves depicting the number of viable cells versus peptide concentration were constructed 48h later and the results are shown in Figures 1A and 1B. When the peptides were added to the cells without co-stimulation, the cells died reaching minimum numbers at 7nM concentration of any peptide (Fig. 1A and 1B, panels A and B). In contrast, when the peptides were added to the cells together with PHA they increased cellular proliferation rates reaching maximum numbers also at 7nM peptide concentration (Fig. 1A and 1B, panels D and E). These observations indicated that (i) both linear and cyclic analogues had a very similar to almost identical effect when administered to human PBMC and that (ii) all the peptides engage the TCR [20]. To show that this is the case, the cells were cultured in plain culture medium

244

5. Thymianou et al. /Effect of Linear and Cyclic Peptide Analogues

with the peptides in the presence of anti-CD28 mAb and the number of viable cells was estimated 48h later. In this case the number of cells remained constant (Fig. 1A and 1B, panels A and B). Thus, all the peptides tested can influence T cells with various Ag specificities and CD28 co-stimulation prevents cell death upon peptide administration to T cells [21]. Interestingly, when the peptides were added to the cell cultures together at a ratio of agonist:antagonist of 1:100, they had no effect on cellular proliferation regardless whether the cells were stimulated with PHA or not (Fig. 1A and 1B, panels C and F). In effect, the results are similar to those obtained when the cells were cultured in plain culture medium with a peptide alone in the presence of anti-CD28 mAb (Fig. 1A and 1B, panels A and B). To understand how this canceling-out effect was achieved, the following experiments were performed:

•Peptide concentration (nM)

Figure 1A: 1.5xl06 PBMC were cultured for 48h in plain culture medium (CM) (A, B, C) or with 5 ug/ml PHA (D, E, F) and various nM concentrations (as indicated on the X-axes) of the linear agonist peptide MBP72-85 (=P1) (A, D) or the linear antagonist peptide A1a81MBP72-85 (=P2) (B, E). In A and B, identical cultures were set up in the presence of 1 ug/ml anti-CD28 mAb. In C and F, the agonist (PI) and antagonist (P2) peptides were added together at a ratio of 1:100; on the X-axes the concentrations of PI are recorded. At the end of the culture period, the cells were harvested and counted by a Sysmex NE-8000 counter, their viability was estimated by the trypan blue exclusion method and the number of viable cells was calculated and plotted on the Y-axes.

5. Thymianou et al. /Effect of Linear and Cyclic Peptide Analogues

245

Peptide concentration (nM)

Resting PB T-cells

Activated PB T-cells Cyclic MBP72-85

Figure IB: Exactly as in Fig. 1A, with the difference that the cyclic forms of the agonist (c-MBP72-85;:::P4) and antagonist (c-Ala81MBP72-85=P5) peptides were used. A, B, C: Culture in plain CM. D, E, F: Culture in CM(PHA). In A and B identical cultures were set up with anti-CD28 mAb. In C and F the agonist (P4) and antagonist (P5) peptides were added together at a ratio of 1:100.

Effect of peptides on T-cell phenotypes: Cells were cultured with 7nM peptide concentration for 24h and phenotyped using mAb against the CDS, CD4, CDS, CD4/CD25 and CD8/CD25 markers. FACS analysis was performed and the percentage of the viable cells that carry these markers was calculated. The results are summarized in Table 1. Overall, the effect of the peptides on the percentage of the above cell types, compared to the control groups, was slight. An exeption was observed with the cyclic antagonist which increased significantly the CD25 [19] expression on unstimulated CD4 cells. When the agonist and antagonist were administered together, CD25 expression on CD4 cells was diminished. From these results it can be concluded that the canceling-out effect (cf. Fig. 1) observed when the agonist and antagonist peptides are added to the cultures together, is not

246

S. Thymianou et al. /Effect of Linear and Cyclic Peptidc Analogues

due to an alteration of the activity phase of the cells i.e. the resting cells remain so but do not die apoptotically and the activated cells remain so also but do not proliferate. Table 1: Effect of cyclic peptides on peripheral blood T cell phenotype cMBP72-85 cAIa81MBP72-85 Cellular Control group phenotype CM (%) + M ( % ) CM(%) +M(%) CM(%) +M(%) 76,7 82,4 86,7 CD3 67,5 68,8 85,3 CD4 52,7 42,5 58,8 53,2 60,3 43,1 25,7 CD8 24 25 23,6 25 32,8 22,6 10,8 23,6 CD4/CD25 0 2,5 21 10,7 13,2 CD8/CD25 0 0,5 0 11,1

c[Agonist+ Antagonist] CM(%) +M(%) 85,9 65 58,4 40 26,6 25 22,4 3,2

0

8

CM=Culture medium; +M=CM+5 ug/ml PHA; [cMBP72-85] and [cAla8lMBP72-85]=7 nM; c[Agonist+Anatgonist]=[ cMBP7285 + cAla8lMBP72-85]=7 nM + 700 nM

Effect of peptides on cytokine production by PB T-cells: The cells were cultured with or without PHA either alone or in the presence of 7nM of the agonist peptides MBP72-85 cMBP72-85 or 7nM of the antagonist peptides AIa8lMBP72-85, c-Ala8lMBP72-85 or with both agonist + antagonist peptides in linear and cyclic form, at a ratio of 1:100 respectively, for 48h. At the end of this period, supernatants were collected and were measured by ELISA for the Thl-type cytokines IL-2 and IFN-y and the Th2-type cytokines IL-4 and IL-10. The results are summarized in Figure 2 and show that: (i) The agonist or the antagonist peptides (linear or cyclic), when added to the cell cultures separately or together, do not induce the production of any of the above cytokines by the cells, (ii) The agonist and antagonist peptides when administered separately or together to PHA-stimulated cells, induce the production of both the Thl- and Th2-type cytokines tested, though in varying quantities. In all cases, the cyclic antagonist (P4) when administered alone induced the production of by far the highest amounts of IL-2, INF-y as well as IL-4 and IL-10. Interestingly, the simultaneous presence of cyclic agonist and antagonist did not inhibit the production of any of these cytokines by the cells but resulted in a reduction of the quantities of all the cytokines tested to levels similar or lower to the ones obtained when the cyclic agonist was added to the cells alone. Under the same culture conditions, the linear agonist and antagonist inhibited IL-2 production significantly, EL-4 and IL-10 production to a lesser degree, but had no effect on EFN-y production. In conclusion, both agonist and antagonist peptides do not seem to target any particular Th-type in healthy individuals and although cytokines are been produced by stimulated normal T-cells co-cultured with peptides, the cells do not proliferate (cf. Fig. 1A,F). It can be therefore deduced that, when administered together in activated cells, the peptides and especially the cyclic antagonist do not render the cells anergic, because they can still produce cytokines, but they block cellular proliferation.

5. Thymianou et al. /Effect of Linear and Cyclic Peptide Analogues

247

Figure 2: 2X100 PBMC/ml were cultured in CM (white columns) or with 5ug/ml PHA (black columns) for 48h. At the end of the culture period cellular supernatants were analyzed for their content of the cytokines IL2, IFN-y, IL-4 and IL-10 by ELISA. The results, plotted on the Y-axes, are [cytokine] in pg/ml/2x!06 cells. Placebo=cells cultured without peptides. Pl-P5=see M&M.

Discussion EAE is an inflammatory CD4 T cell-mediated disease that can be induced by immunisation with MBP and PLP proteins as well as with peptide epitopes such as the 72–85 amino acid sequence of the guinea pig MBP. Altered peptide ligands, which differ from their parent encephalitogenic peptides by individual amino acid substitutions, have been described which are capable of inhibiting autoimmune-mediated disease. The binding of such altered ligands to the TCR can prevent disease through one of several different mechanisms including TCR antagonism [22], induction of anergy in responding cells [23], or the stimulation of immunoregulatory T cells which can actively modulate disease outcome through the secretion of Th2- and Th0- type cytokines [14,24]. However, the development of alternative molecules that will mimic the immunomodulatory activity of MBP epitope peptides and will maintain an advantage over regular peptides in terms of stability is a necessary step before these molecules can be used for therapeutic purposes. There are two approaches in the development of such molecules. One is based on the knowledge of the immunodominant amino acid residues of the MBP72-85 epitope and involves the construction of a chemical moiety in which the important pharmacophoric groups will be incorporated. Such approach may lead to nonpeptide mimetics or semimimetic peptides with immunomodulatory activity [15]. The other approach is the design and synthesis of potent cyclic MBP72-85 analogues which offer several advantages such as increased resistance to metabolic degradation and consequently

248

5. Thymianou et al. /Effect of Linear and Cyclic Peptide Analogues

longer duration of action. Due to restriction of conformational flexibility, constraint peptide analogues of high potency offer important structural information towards the bioactive conformation assumed by the peptide and could be very useful in drug design for the development of non-peptide mimetics [25,26]. Such approaches have been successful in the design and synthesis of non-peptide mimetics for peptide hormones such as Angiotensin II which is implicated in cardiovascular diseases [27-31] or for motifs of proteins such as Thrombin implicated in haemostasis and angiogenesis [32–34,25,26]. This computer-aided novel technology may eventually be applied in generating mimetic molecules of peptides with immunomodulatory ability. The guinea pig analogue of MBP72–85 peptide that induces transient EAE in Lewis rats was used as a reference peptide to synthesize the cyclic analogues which is an important intermediate step towards non-peptide mimetics. We have recently shown that a cyclic analogue of MBP72-85 (c-MBP72-85) is a potent inducer of EAE and that this cyclic molecule has a comparable potency to that of the linear analogue MBP72-85 [16]. These findings were extended to show that a cyclic analogue of the linear antagonist, c-Ala81MBP72-85, is a potent inhibitor of EAE when co-injected with MBP72-85 85 in Lewis rats: EAE induced by linear MBP72-85 was completely suppressed when it was co-injected with the cyclic antagonist c-Ala8lMBP72-85 [17]. EAE induced by cyclic analogue c-MBP72-85 was also completely suppressed when it was co-injected with the linear analogue Ala8lMBP72-85 [16] or the cyclic analogue c-Ala8lMBP72-85 [17]. To begin to assess the possibility that the cyclic antagonist c-Ala8lMBP72-85 can be eventually used in human MS patients, the effect of linear MBP72-85 and Ala8lMBP72-85 and cyclic c-MBP72-85 and c-Ala8lMBP72-85 peptides was analysed separately or in combination in human T-cell cultures. The rationale behind using peripheral blood cells obtained from randomly selected healthy individuals, was that first we had to establish whether these peptides work in the human system and then to deduce whether there is any point to test them in cells derived from MS patients. In addition, MS patients express a variety of TCRspecificities other than the autoreactive T-cell clones, and therefore it is important to know whether these peptides affect their survival and function. Furthermore, if the peptides cause a deleterious effect in normal T-cells their use becomes unlikely for therapeutic purposes in patients with an already established autoimmune disease with a Thl bias [35]. In MS patients, the majority of myelin Ag-specific T cells isolated during active disease secrete Thl-type cytokines while during remission the balance shifts to DL-4 and IL-10 producing cells [35]. The results shown here raise the following major points: (i) Both agonist and antagonist peptides act through the TCR. (ii) If the peptides under study are given to unstimulated cells they force them to die of apoptosis. (iii) If the peptides are administered together to cell cultures from PBMC without an additional stimulus, apoptosis is prevented completely but the cells are not stimulated. This fact can be explained by the hypotheses that the peptides either compete with each other for the same binding site and none can bind to the TCR or, otherwise, they render the cells anergic. This is to be clarified in future experiments, (iv) When the cells are PHA-activated, the simultaneous administration of the agonist and antagonist peptides to the cells seems to result to a qualitatively different activation of T cells so that cytokine production is induced without the concomitant proliferation of the cells. The above observations are reminiscent of previous studies performed with altered peptide ligands that have been shown to engage the TCR and induce cytokine production or cytolytic activity without the proliferation of the T cells induced by the wild-type agonist peptide [36,37]. An alternative explanation for the failure of activated T cells to proliferate in the presence of both agonist and antagonist peptides is that the simultaneous secretion of IL-2, IFN-y, IL-4 and IL-10 cytokines is in such quantities that apparently neutralize each other's action. Experiments are in progress aiming to show whether the peptides induce secretion of different cytokines from different PB T-cells or whether they alter the pattern

S. Thymianou et al. /Effect of Linear and Cyclic Peptide Analogues

249

of cytokine production in the same cells [38–40]. Finally, another parameter to be studied is the fact that this canceling out effect of the agonist and antagonist peptides on T-cell proliferation and also in the EAE animal model takes place when the ratio between agonist:antagonist is 1:100 (found to be the optimal ratio in preliminary experiments-data not shown). This could mean that indeed this is the required ratio for the antagonist to counteract the effect of the agonist that causes the disease (ie. the antagonist has a much lower affinity than the agonist). But it could also mean that it is the overall high peptide dose that causes a high dose supression effect or high zone tolerance observed also in other studies of EAE [41]. Acknowledgments This work was supported by the Greek Ministry of Energy and Technology (EPET II, PENED 99–115). T. Tselios is a recipient of a grant by the Leonidas Zervas foundation for MS. References [1] R. Martin, H. McFarland and D. McFarlin, Ann. Rev. Immunol. 10 (1992) 153. [2] L. Steinman, Cell 85 (1996) 299. [3] M. Pette, K. Fujita, D. Wilkinson, D. Altmann, J. Trowsdale, G. Giegerich, A. Hinkkanen, J. Epplen, L. Kappos and H. Wekerle, Proc. Natl. Acad. Sci. USA 87 (1990) 7968. [4] D. Hafler and H. Weiner, Immunol. Rev. 144 (1995) 75. [5] K. Ota, M. Matsui, E. Milford, G. Mackin, H. Weiner and D. Hafler, Nature (Lond) 346 (1990) 183. [6] R. Martin, D. Jaraquemada, M. Flerlage, J. Richert, J. Whitaker, E. Long, D. McFarlin and H. McFarland, J. Immunol. 145 (1990) 540. [7] J. Greer, P. Csurhes, K. Cameron, P. McCombe, M. Good and M. Pender, Brain 8 (1997) 1447. [8] R. Hohfeld, Brain 120 (1997) 865. [9] S. Zamvil and L. Steinman, Ann. Rev. Immunol. 8 (1990) 579. [10] Y. Chou, A. Vandenbark, R, Jones, G. Hashim and H. Offner, J. Neuros. Res. 22 (1989) 181. [11] M. Wauben, C. Boog, R. Zee, I. Joosten, A. Scihlief and W. Eden, J. Exp. Med. 176 (1992) 667. [ 12] R. Smeitz, M. Wauben, N. Wolf, and R. Swanborg, J. Immunol. 162(2) (1999) 829. [13]M. Vergelli, B. Hemmer, U. Utz, A. Vogt, M. Kalbus, L. Tranquill, P. Conlon, N. Ling, L. Steinman, H. H. McFarland and R. Martin, Eur. J. Immunol. 26 (1996) 2624. [14]S. Brocke, K. Gijbels, M. Allegretta, I. Ferber, C. Piercy, T. Blankenstein, R. Martin, U. Utz, N. Karin, D. Mitchell, T. Veroma, A. Waisman, A. Gaur, P. Conlon, N. Ling, P. Fairchild, D. Wraith, A. O'Carra, G. Farhman and L. Steinman, Nature 379 (1996) 343. [15]T. Tselios, L. Probert, G. Kollias, E. Matsoukas, P. Roumelioti, K. Alexopoulos, G. Moore and J. Matsoukas, Amino Acids 14 (1998) 333. [16]T. Tselios, L. Probert, I. Daliani, E. Matsoukas, A. Troganis, P. Gerothanassis, T. Mavromoustakos, G. Moore and J. Matsoukas, J. Med. Chem. 42 (1999) 1170. [17]T. Tselios, I. Daliani, S. Deraos, S. Thymianou, E. Matsoukas, A. Troganis, I. Gerothanassis, A. Mouzaki, T. Mavromoustakos, L. Probert and J. Matsoukas, Bioorganic & Medicinal Chemistry Letters 10 (2000) 2713. [18] Mouzaki, D. Rungger, A. D. Tucci, A. Doucet, R. H. Zubler, Eur. J. Immunol. 23 (1993) 1469. [19] Barclay, M. Brown, S. Law, A. McKnight, M. Tomlinson and P. van der Merwe, (Eds.), Leukocyte antigen Facts book. Academic Press, Harcoun Brace and Co, (1997) pp.486–489. [20]M. Alderson, T. Tough, T. Davis-smith, S. Braddy. B. Falk, K. Schooley, R. Goodwin, C. Smith. F. Ramsdell and D. Lynch, J. Exp. Med. 181 (1995) 71. [21] P. Noel, L. Boise, J. Green and C. Thompson, J. Immunol. 157 (1996) 636. [22] V. Kuchroo, J. Greer, D. Kaul, G. Ishioka, A. Franko, A. Sette, R. Sobel and M. Lees, J. Immunol. 153(1994)3326. [23] J. Sloan-Lancaster, B. Evavold and P. Allen, Nature 363 (1993) 156. [24] L. Nicholson. J. Greer, R, Sobel, M. Lees and V. Kuchroo, Immunity 3 (1995) 396. [25] K. Alexopoulos. P. Fatseas, E. Melissari. D. Vlahakos, J. Smith, T. Mavromoustakos, M. Saifeddine. G. Moore, M. Hollenberg and J. Matsoukas, Bioorg. Med. Chem. 7 (1999) 1.

250

S. Thymianou et al. / Effect of Linear and Cyclic Peptide Analogues

[26] K. Alexopoulos, J. Matsoukas, T. Tselios, P. Roumelioti, T. Mavromoustakos and K. Holada, Amino Acids 15 (1998) 211. [27] J. Matsoukas, J. Hondrelis, G. Agelis, K. Barlos, D. Gates, R. Ganter, D. Moore and G. Moore, J. Med. Chem. 37 (1994) 2958. [28] J. Matsoukas, R. Yamdagni and G. Moore, Peptides 11 (1990) 367. [29] J. Matsoukas, G. Bigam, N. Zhou and G. Moore, Peptides 11 (1990) 359. [30]J. Matsoukas, J. Hondrelis, M. Keramida, T. Mavromoustakos, A. Makriyiannis. R. Yamdagni, Q. Wu and G. Moore, J. Biol. Chem. 269 (1994) 5303. [31]D. Vlahakos, J. Matsoukas, J. Ancans, G. Moore, E. Iliodromitis, K. Marathia and D. Kremastinos, Let. Pept. Sci. (LiPS) 3 (1996) 191. [32] J. Matsoukas, J. Hondrelis, G. Agelis, K. Barlos, D. Gatos, R. Ganter, D. Moore and G. Moore, J. Med. Chem. 37 (1994) 2958. [33] D. Panagiotopoulos, J. Matsoukas, K. Alexopoulos, A. Zebeki, T. Mavromoustakos, M. Saifeddine and M. Hollenberg, Let. Pept. Sci. (LiPS) 3 (1996) 233. [34]J. Matsoukas, M. Hollenberg, T. Mavromoustakos, D. Panagiotopoulos, K. Alexopoulos, R. Yamdagni, Q. Wu and G. Moore, J. Protein Chem. 16 (1997) 113. [35] J. Correale, W. Gilmore, M. McMillan, S. Li, K. McCarty, T. Le and LP. Weiner, J. Immunol. 154 (1995)2959. [36]Evavold, J. Sloan-Lancaster and P. Allen, Immunol. Today 14 (1993) 602. [37] S. Valitutti and A. Lanzavecchia, Immunol. Today 18 (1997) 299. [38] V. Apostolopoulos, G. Pietersz and I. McKenzie, Vaccine 9 (1995) 930. [39] S. Lofthouse, V. Apostolopoulos, G. Pietersz, W. Li and I. McKenzie, Vaccine 14 (1997) 1586. [40] V. Apostolopoulos, G. Pietersz, B. Loveland, M. Sandrin and I. McKenzie, Proc. Natl. Acad. Sci. USA 92(1995)10128. [41]J. Critchfield, M. Racke, J. Zuniga-Pflucker, B. Cannella, C. Raine, J. Goverman and M. Lenardo, Science 263 (1994) 1139.

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

251

Use of Synthetic Peptides for the Identification of the Biologically Active Domains of the Growth Factor HARP E. Papadimitriou1, A. Polykratis2, E. Karestou1, J. Courty3, and P. Katsoris2 !

Lab of Molecular Pharmacology, Department of Pharmacy, University ofPatras, Greece, Lab of Cell Biology, Department of Biology, University ofPatras, Greece and 3 Lab de Recherche sur la Croissance Cellulaire, la Reparation et la Regeneration Tissulaires, Universite Paris XII, France.

Abstract. HARP (Heparin Affin Regulatory Peptide) is a growth/differentiation secreted protein with distinct lysine-rich clusters within both the NH2- and COOH-terminal domains and apparent molecular weight 18 kDa. It was initially purified from bovine uterus and neonatal rat brain and found to be highly conserved among species. It has a high affinity for heparin and is localized in the extracellular matrix through interactions with glycosaminoglycans. It is expressed in developing tissues and displays important function in the growth and differentiation processes. The purpose of the studies presented here was to define the effect of HARP on endothelial cell functions. We studied the effect of human recombinant HARP expressed in a bacterial expression system, as well as two HARP structurally related synthetic peptides (HARP residues 1–21 and residues 121–139), on different endothelial cell types derived from human umbilical vein (HUVEC), rat adrenal medulla (RAME), bovine retina capillaries (BREC) and bovine brain capillaries (BBC). We also tested the angiogenic effect of HARP and peptides on the in vivo system of the chicken embryo chorioallantoic membrane (CAM). Both peptides had a dose-response mitogenic effect whether added to the culture medium or adsorbed onto the tissue culture plate. The effect was at least two times more intense in the latter case. HARP seems to be mitogenic only when coated onto the cell culture plate, the only exception being the RAME cells. It is worth noticing that BBC cells, which are the most responsive to HARP, release the higher amounts of this growth factor into the culture medium, compared to the other types of cells. Both HUVEC and BBC endothelial cells express HARP mRNA and the expression changes at different levels of cell confluency. HARP, as well as both peptides, induce angiogenesis in vivo, exhibit chemotactic action and induce capillaries formation on collagen and fibrin gels, as well as on matrigel. These actions are dose dependent and different among studied cells, elucidating to a degree the implication of HARP terminal regions on endothelial cell biology and angiogenesis.

252

E. Papadimitriou et al. / Use of Synthetic Peptides

Introduction HARP (Heparin Affin Regulatory Peptide), also called pleiotrophin or HB-GAM (Heparin Binding - Growth Associated Molecule) is a new growth factor with high affinity to heparin [1, 2]. Along with Midkine, it belongs to a new family of heparin binding growth factors [3]. It consists of 168 amino acids -the first 30 corresponding to a signal peptide indicating a secretive protein- and has an approximate molecular weight of 18 kDa. There are no indications of post-translational modifications of the molecule. The mature form of the protein consists of 135 or 136 amino acids, depending on the species. It is highly basic, with 26 lysines forming two clusters at both termini of the molecule [1]. These amino acids contribute to the basicity of HARP and match with a consensus sequence involved in heparin-binding and angiogenic activity [4]. The presence of 10 cysteines at the central region of the molecule is characteristic, highly conserved and contributes to the formation of the tridimensional structure of the molecule through five intrachain disulphide bonds. HARP contains two p-sheet domains, homologous to the thrombospondin type I repeat (TSR), which are connected by a flexible linker. Each one of these domains contains three antiparallel p-strands (Figure I) [5].

Figure 1. The HARP molecule. The picture shows the two lysine-rich terminal domains, the two B -sheet domains and the five intrachain disulflde bonds.

HARP was originally purified from uterus [6], neonatal rat brain [1] and placental tissue [7]. It seems to be implicated in growth and differentiation processes and is expressed in several developing tissues [1, 8–10]. In adults, HARP is expressed in neuronal tissues, heart, uterus, cartilage and bone [3, 6, 9, 10], indicating that it also has important physiological role(s) during adulthood. This is further strengthened by the fact that HARP is highly conserved among species, with an overall homology reaching 98%. Although differences can be found at the signal peptide, no significant differences are observed in the mature molecule [11]. HARP regulation remains poorly documented. HARP mRNA is increased by the platelet-derived growth factor and basic fibroblast growth factor [7], retinoic acid [12], progesterone [13] dihydrotestosterone, testosterone and estrogen [14]. HARP may also be considered as a marker of neuronal injury [15]. Endothelial cells express HARP mRNA (Figure 2) and secrete the protein both in their conditioned medium and in the extracellular matrix [16]. Human recombinant HARP produced in a bacterial expression system downregulates the expression of HARP mRNA in endothelial cells (Figure 2). This is a strong indication that the molecule produced in bacteria is recognized by specific molecule(s) on the surface of endothelial cells and thus exerts its biolosical functions.

E. Papadimitriou et al. / Use of Synthetic Peptides

253

Figure 2. (A) RT-PCR reactions for human GAPDH and HARP mRNA levels in human umbilical vein endothelial cells (HUVEC) at several time points after stimulation with human recombinant HARP produced in a bacterial expression system. A representative image of 5 independent experiments is shown. (B) Expression of HARP mRNA as % ratio of HARP/GAPDH electrophoretic band values compared to controls (unstimulated cells). Results are mean ± S.E.M. from 5 independent experiments. * P<0,05.

Initially, HARP was described as a molecule that induces neurite outgrowth of neuronal cells [1] and its ability to promote neurite outgrowth is indisputable [8]. Furthermore, it is mitogenic for a variety of cells, although results concerning its mitogenic activity are controversial, depending on the cell type and the experimental conditions used [16–19]. Recombinant HARP expressed in eucaryotic cells is mitogenic, whereas HARP expressed in prokaryotic expression system, although promotes neurite outgrowth, was considered not to have any mitogenic effect [3]. We have recently shown that human recombinant HARP expressed in bacteria is mitogenic for endothelial cells only when presented to them as a substrate, while it does not influence endothelial cell proliferation when it is added to the medium of the cells [16]. The endothelial cells used are BBC (bovine brain capillary cells), RAME (rat adrenal medulla endothelial cells) and HUVEC (human umbilical vein endothelial cells).

Figure 3. Effect of human recombinant HARP expressed in bacteria and of two peptides, corresponding to its COOHand NH 2 -termini, on the proliferation of BBC cells. Different concentrations of the test agents were added to the cell culture medium of the cells (A) or coated onto the corresponding wells of the cell culture plates (B). Cells were further incubated for 48 h before counted, using the MTT assay [16). *P< 0.05, "P<0.0! and ***P<0.001 [16].

254

E. Papadimitriou et al. / Use of Synthetic Peptides

Two peptides that correspond to the NH2 and COOH termini of the protein (HARP residues 1-21 and 121-139 respectively) are mitogenic for all the types of endothelial cells used, in a concentration-dependent manner and the effect is more intense when the substances are coated onto the plate (Figures 3, 4 and 5). These two regions represent the lysine-rich domains of the molecule and are thought to be involved in multiple biological actions, since they are also required for successful transformation of NTH 3T3 cells by HARP residues 41–64 [20]. Recently, it was shown that the residues 111-136 of the lysine-rich COOHterminal domain are involved in the mitogenic and tumour formation activities of HARP and play a role in the binding of HARP to its high affinity receptor. The same residues are not implicated in the neurite-outgrowth activity [21], supporting the notion that the domains involved in HARP mitogenic and neurite outgrowth activities are different. The amino acids 111-128 seem to play a key role, since a mutant that lacks residues 129–136 is still mitogenic for Chinese hamster ovary cells [21].

icentrabon (ng/ml)

Figure 4. Effect of human recombinant HARP expressed in bacteria and of two peptides, corresponding to its COOHand NH2 -termini, on the proliferation of RAME cells. Different concentrations of the test agents were added to the cell culture medium of the cells (A) or coated onto the corresponding wells of the cell culture plates (B). Cells were further incubated for 48 h before counted, using the MTT assay [16]. *P<0.05, **P<0.01 and ***P
HARP may also be an important regulator of tumour transformation. It is detected in various carcinomas, such as human breast and prostate cancer, neuroblastomasr benign meningiomas, small cell lung cancer and rat mammary tumours [3, 22] and is involved in tumour growth and metastasis [22–24]. Many of the HARP expressing tumours become more invasive when HARP is overexpressed and reduction of HARP levels reduces the number of blood vessels in primary tumours, indicating a possible role of HARP in tumour angiogenesis [23, 25].

E. Papadimitriou et al. / Use of Synthetic Peptides

10

100

concentration (ng/well)

1000

255

Figure 5. Effect of human recombinant HARP expressed in bacteria and of two peptides, corresponding to its COOHand NH2 -termini, on the proliferation of HUVEC cells. Different concentrations of the test agents were added to the cell culture medium of the cells (A) or coated onto the corresponding wells of the cell culture plates (B). Cells were further incubated for 48 h before counted, using the MTT assay [16]. *P< 0.05 and ***P<0.001 [16].

Angiogenesis implicates many different actions, such as cell adhesion, migration, proliferation, differentiation and degradation of the basal lamina. HARP affects many endothelial cell processes involved in angiogenesis [26]. The two peptides that correspond to the NH2- and COOH- termini of the protein, have a similar effect in most of the systems studied. HARP stimulates the formation of cord-like structures by BBC and bovine retinal endothelial cells on collagen gels, while neither peptide has any significant effect. On matrigel, where both cell migration and differentiation take place, the two terminal peptides have a stimulating effect on tube formation by BBC and RAME and no effect on HUVEC, similarly to the effect of the whole molecule of HARP [26]. In the same line, HARP and both peptides stimulate migration of BBC and RAME cells, but have no significant effect on HUVEC (Figure 6). In vivo, in the chicken embryo chorioallantoic membrane (CAM) model, both HARP terminal peptides have an angiogenic effect, however, only in the case of the COOH peptide this effect is dose-dependent. The NH2 peptide has a weaker angiogenic effect, which was statistically significant only at one dose (Figure 7). The mechanism of action of HARP or its terminal peptides is not yet clarified and it is not known if it is a direct effect through binding to cell surface receptors or an indirect effect, potentiating the angiogenic effect of other growth factors.

256

E. Papadimitriou et al. / Use of Synthetic Peptides

Figure 6. Effect of human recombinant HARP and two peptides (loOng/mI), corresponding to its COOH- and NH2 - termini, on migration of HUVEC (A), R A M E (B) and BBC cells (C), using the Boyden chamber assay. Results are expressed as percent of the values obtained without stimulation. Data are the mean f SEM of at least three independent experiments. *Pc0.05,**P
In conclusion, HARP synthetic peptides representing native molecule residues 1-21 and 121-139 seem to be involved in multiple biological actions, since they stimulate endothelial cell proliferation 116, 211 and angiogenesis [26] and are required for successful transformation of cells by a central region of HARP [20]. These results suggest that minimal structures could be sufficient to trigger endothelial cell activation, if they are suitably presented to the cell. Whether these or similar peptides exist physiologically (e.g. after proteolysis of HARP) is not known at present and is under investigation.

Figure 7. Effect of human recombinant HARP and two peptides, corresponding to its COOH- and NH2- termini, on angiogenesis in the in vivo model of chicken embryo CAM. Data are the mean f SEM of three independent experiments, each one with 20 eggs tested for each concentration used. *P<0.05, **PcO.Oland ***P
E. Papadimitriou et al. / Use of Synthetic Peptides

257

References [ 1] H. Rauvala, EMBO J. 8 (1989) 2933. [2] F. Vacherot et al, J. Biol. Chem. 274 (1999) 7741. [3] N. Zhang and T. F. Deuel, Curr. Opin. Hematology 1 (1999) 44. [4] J. Courty et al., Heparin-Affin Regulatory Peptide, HARP. In: Comprehensive Vascular Biology and Pathology - An Encyclopedic Reference, Springer-Verlag, 2000, pp. 145-152. [5] I. Kilpelainen et al., J. Biol. Chem. 275 (2000) 13564. [61 P. G. Milner et al., Biochem. Biophys. Res. Commun. 165 (1989) 1096. [7] Y.-S. Li et al., J. Biol. Chem. 267 (1992) 26011. [8] Y.-S. Li et al., Science 250 (1990) 1690. [9] J. M. Vanderwinden et al.,Anat. Embryol 186 (1992) 387. [10] I. Silos-Santiago et al., J. Neurobiol. 32 (1996) 283. [ 11 ] B. S. Hampton et al, Mol. Biol Cell 3 (1992) 85. [12] P. Bohlen et al, Growth Factors 4 (1991) 97. [13] P. E. Milhieteral,J. Endocrinol. 158(1998) 389. [14] F. Vacherot et al, In vitro Cell Dev. Biol Anim. 31 (1995) 647. [15] A. Takeda et al, Neuroscience 68 (1995) 57. [16] E. Papadimitriou et al, Biochem. Biophys. Res. Commun. 274 (2000) 242. [17] J. Courty et al, Biochem. Bioph. Res. Commun. 180 (1991) 145. [ 18] W. Fang et al.J. Biol. Chem. 267 (1992) 25889. [19] E. Szabat and H. Rauvala, Dev. Biol. 178 (1996) 77. [20] N. Zhang et al, J. Biol. Chem. 274 (1999) 12959. [21] I. Bernard-Pierrot etal.J. Biol. Chem. 276(2001) 12228. [221 A. K. Chauhan et al, Proc. Natl. Acad. Sci. USA 90 (1993) 679. [23] F. Czubayko et al, Proc. Natl. Acad. Sci. USA 93 (1996) 14753. [24] N. Zhang et al, J. Biol Chem. 272 (1997) 16733. [25] R. Choudhuri et al.. Cancer Res. 57 (1997) 1814. [26] E. Papadimitriou et al, Biochem. Biophys. Res. Commun. 282 (2001) 306.

258

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

Unusual Interactions between MHC, peptide and the T cell Receptor V. Apostolopoulos1 and M. Plebanski2 'Head, Immunology and Vaccine Laboratory, Head, Vaccine Development and Infectious Disease Laboratory, The Austin Research Institute, Studley Road, Heidelberg, VIC 3084, Australia.

2

Abstract. The major histocompatibility complex on the surface of antigen presenting cells presents peptides to the T cell receptor. Recognition of peptide-MHC by T cells activates them as well as initiating a sequence of signals in T cells which enables a T cell dependent/specific immune response. For vaccine design, it is important to understand how peptides bind to MHC class I molecules. In this review, we will discuss, some unusual features of peptides binding to MHC class I, crystal structures of low and high affinity peptides lacking canonical anchor motifs in complex with H-2Kb, peptide mimics in cancer and the role of altered peptide ligands in the immune response in particular in cancer and in infectious diseases.

Role of the MHC in the cellular immune response The T cell receptor (TCR) interaction with major histocompatibility complex (MHC) molecules is a central event in T cell-mediated responses. During thymic development, MHC molecules educate T cells, guiding their positive and negative selection. Outside the thymus, protein antigens from viral or bacterial sources are presented in the context of MHC molecules to T cells [1] after intracellular processing to peptides [2]. TCRs on lymphocytes recognize foreign antigens bound to the MHC and can produce cytokines, proliferate or cause cytotoxic cell lysis in the periphery in response to these stimuli. The MHC loci encode Class I and Class II MHC molecules. Class I MHC are found on the surface of nucleated cells [3] where they form non-covalently linked heterodimers between an MHC-encoded, membranespanning variable heavy chain (44 kDa) and a non MHCencoded, invariant light chain, (52 microglobulin (P2M,

Peptide

Figure 1: Structure of MHC class (H-2Kb in complex with MUC1recently determined in our laboratory.

V. Apostolopoulos and M. Plebanski / Unusual Interactions

259

12 kDa) (Figure 1). Class I molecules bind short peptides (usually -8-10 amino acids), mostly derived from endogenously synthesised proteins but also from exogenous antigens, in the endoplasmic reticulum and present them to cytotoxic T lymphocytes (CTLs) that express the CDS co-receptor. Since an individual expresses only a limited number of MHC molecules, these antigen-presenting molecules must be able to interact with a large diversity of peptide sequences. Nevertheless, the binding of peptides cannot be overly "promiscuous" since individual MHC molecules have their own particular peptide-binding motifs [4]. Specificity of MHC class I: The peptide binding groove of MHC class I can be subdivided into various pockets (A-F) [5]. For MHC class I, peptide length varies from 8-9-mers (H2Kb, HLA-A2) and up to 13-mers (HLA-Aw68) [6]. Early biochemical studies indicated the presence of conserved (consensus) sequences in high affinity binding peptides eg. for H-2Kb, P2, P6, P9 for 9mers or P2, P5, P8 for Smers it was found that the preferred anchors are Phe/Tyr for the central P5/6 residues, Leu/Met at the C-terminus (P8/9), and, in some instances, Tyr at P3 (these were found by acid elution from purified Class I molecules). In this process, low affinity peptides were lost and do not feature in the considerations of Rammensee and colleagues who have set the initial set of useful "rules" for peptide binding for Class I based on the sequence findings [6]. Crystallographic studies indicated that the conserved region were hydrophobic "anchors" - which bound in "hydrophobic" pockets. The non-conserved (ie. variable region) of peptides were likely to be presented to T cells. Although, these anchor residues are required for stabilisation and high affinity binding of peptide to MHC, it has been shown that some peptides which do not contain the respective canonical anchor residues can still bind and be presented by MHC class I and cell lysis still occurs [7–9]. Until now, it has not been clear how such non-standard or low affinity peptides are presented by MHC class I molecules in order to serve as targets for CTL. We have recently determined the crystal structures of two non-canonical anchor motif peptides (one with low affinity and the other with high affinity binding) in complex with H-2K . Low affinity peptides binding to MHC class I: their use in immune responses: High affinity peptides which bind to MHC class I molecules usually induce high avidity CTL. However, the use of high affinity peptides from tumour antigens may not be effective since most tumour antigens are self antigens, and their specific CTL repertoire would most likely be deleted. Thus, for immunotherapy it is conceivable that the best candidate peptides for immunisation are the low affinity binding peptides. A major problem to the use of low affinity peptide epitopes for immunotherapy is the difficulty of their identification due to their poor immunogenicity. It has been shown that the affinity of peptide binding to MHC and the stability of the peptide-MHC complex can correlate with overall immunogenicity [10, 11]. Furthermore, low affinity peptides cannot be detected by elution studies or by prediction algorithms. Thus, peptide identification by systematic binding studies and CTL assay is required for their identification. It is likely that enhancement of the immunogenicity of low affinity MHC binding peptides, is possible by mutations of the peptide anchor side chains. Low affinity MUC1 peptides presented by H-2Kb (SAPDTRPAP) and H-2Db (APGSTAPPA) were not able to stabilise RMA-S cells as determined by FACS, however they were recognised by peptide specific CTL [9]. Mutations of SAPDTRPAP to SAPDTYPAL and APGSTAPPA to APGSNAPPA were now able to stabilise class I on RMA-S cells [9]. Furthermore, the H-2Kb low affinity peptide, MUT1 (a peptide from Lewis lung carcinoma), when mutated at P3, P5 and P8, has increased stability and affinity of peptide for RMA-S cells [12]. In addition, the low affinity MUC1 peptide

260

V. Apostolopoulos and M. Plebanski / Unusual Interactions

(SAPDTRPAP) that binds to HLA-A2 has been converted by mutation to high affinity and are currently being used in a clinical setting [8]. Until recently, no structural information was available on the binding of nonstandard peptides to MHC class I molecules, and whether they differed substantially in their interaction with the MHC and, therefore, with the T cell receptor. We recently determined the crystal structure of two peptides lacking canonical anchor motifs bound to MHC class I, H-2Kb; an 8-mer MUC1 peptide (MUC1-8) and the other a 9-mer yeast peptide [YEA9]. Crystal structure of a low affinity peptide in complex with MHC class I, H-2ff : MUC1-8 (SAPDTRPA) derived from Mucinl does not contain Phe/Tyr at P5 or Leu/Met at P8 which are usually associated with high affinity binding to H-2Kb [4, 6]. MUC1-8 has small residues at P2 (Ala), P5 (Thr) and P8 (Ala) [9] and has a relatively low affinity, compared to high affinity H-2Kb binding peptides. We recently determined the crystal structure of MUC1-8 in complex with H-2K at 1.6A resolution [13, 14]. The side chains of residues that point into the C and F pockets are small; hence, these anchor pockets are not completely filled and a large cavity is formed, as compared to high affinity OVA8 or VSV8 peptides. [13, 14]. Thus, this relatively low affinity peptide binds with the same overall features to MHC class I as high affinity peptides, in that, the peptide starts and finishes at the same position at the N- and C- termini; deviations occur within the central region of the peptide. The small peptide anchoring side chains at P2, P5 and P8, and a cavity at the side of the C pocket, appear to contribute to the lower affinity and of MUC1-8 with H-2Kb. Crystal structure of a relatively high affinity peptide lacking canonical anchor motifs bound to H-2K b : During expression of MHC class I, H-2Kb, in Drosophila melanogaster fly cells, yeast peptides added by the manufacturers to the culture medium, stabilise H-2Kb and aid in its purification. The peptide found in the peptide-binding groove is SRDHSRTPM, or YEA9 [15]. YEA9 contains large residues at Arg-P2, Arg-P6 and Met-P9 at the anchor sites as compared to the preferred Ala/Gly/Ile-P2, Phe/Tyr-P6 and Val/Met-P9. We recently determined the crystal structure of YEA9 in complex with H-2Kb at 1.5A resolution [13, 14]. The non-canonical anchor motif peptide, YEA9, bound with relatively high affinity by insertion of long non-canonical anchor residues into the B and E pockets. This is the first description of the use of alternative anchors for high affinity peptide binding. It is clear that peptides lacking canonical anchor motifs can bind to MHC class I with either low or relatively high affinity, and that all P2, P6 and P9 residues contribute in anchoring the peptide into the peptide-binding groove. These studies are important as they show that binding of peptide with high affinity can occur in two ways: (i) by canonical hydrophobic anchors binding in the C and F pockets and (ii) by non-canonical anchors binding in the B, E and F pockets. Other features of peptides binding to MHC class I Binding of longer peptides to MHC class I: MHC class I molecules preferentially bind peptides 8–10 amino acids long. MHC class I, H-2Kb binds peptides 8 or 9 amino acids long, and the mechanism of binding involves a bulge at the center of the peptide, leaving. the N and C termini unchanged [15, 16]. However, studies have indicated that longer peptides can be accommodated by MHC class I [11, 17]. It has been implicated that the mechanism of binding of longer peptide involves creating a bulge at the center of the peptide, leaving the N and C termini interactions unchanged. However, longer peptides may be accommodated by MHC class I by protrusion, primarily at the C-terminus. The

V. Apostolopoulos and M. Plebanski / Unusual Interactions

261

crystal structure of a 10-mer peptide bound to HLA-A2 showed a single glycine C-terminal extension extending from the binding groove [18]. Peptides with the addition, at the N- or C-termini, of two amino acids of the VSV8 and OVA8 peptides bind to H-2Kb [19]; four residues could be added at the C- but not at the N-terminus of a VSV8 peptide bound to H2Kb [17]. A peptide extended at the C-terminus by 10 amino acids (an 18-mer) binds to H2Kk as it could be eluted from the MHC class I molecule [20]. Recently, it was demonstrated that extensions could occur at either the N- or C-terminus for H-2Kk, H-2Kb and H-2Db molecules [11]. T cells were able to identify the extension; thus, Reptide extensions may play a role in the specificity of the T cells [11]. We have demonstrated that H-2Kb can bind SAPDTRPAP peptides that are extended by up to 5 residues at the Cterminus (SAPDTRPAPGSTAP; MUC1-14). How long peptides are presented by MHC and recognised by the TCR is of importance for the design of alternative peptides in vaccination protocols. In addition, it will be important to determine whether such peptides "loop" in the middle (as recently seen with the rat MHC) or "hang" out from the MHC at either the N- or C-terminal ends. Binding of shorter peptides to MHC class I: It has been demonstrated that immunisation of mice with mannan-MUCl generates CTL that recognise both peptide-pulsed and MUC1transfected target cells that protect mice against a MUC1+ tumour challenge [21]. MUC1-9 presented by H-2Kb, can also be presented and recognised by CTL as a 5-mer, 6-mer, 7-mer and 8-mer (deletion at the C-terminus) when pulsed on RMA-S target cells [Apostolopoulos et al., 2001, submitted, J. Molecular Biology]. The 4-mer, SAPD, is not presented nor lysed by CTL. Other short immunogenic peptides for class I have been demonstrated to bind to H-2Ld and to be recognized by CTL, including, a 3-mer (QNH), 4mers (QNHR, ALDL, PFDL) and 5-mers (RALDL, HFMPT) [22]. These observations are clearly outside the normal structural guidelines for tight binding of class I peptides, deduced from crystal structures of many peptide-MHC complexes. How short peptides are presented by MHC class I molecules and recognised by the TCR is currently unknown. Peptides looping out at the mid- and C- terminus when bound to MHC class I: We have demonstrated that H-2Kb molecules can present the 9-mer peptide SAPDTRPAP (MUC19Kb) which does not contain the appropriate anchors [8, 9, 23] and binds with very low affinity. Furthermore, the peptide APGSTAPPA (MUCl-9Db) was found to be presented by H-2Db. This peptide also does not contain the canonical anchor of Asn-P5. MUC1 VNTR peptides (MUCl-9Kb and MUCl-9Db) have other unusual features exemplified by antibody binding and modelling studies. It was previously demonstrated that MUCl-9Kb could loop out sufficiently at the mid and C-terminus to be detected by anti-MUCl peptide monoclonal antibodies when bound to H-2Kb [23]. The same antibodies to the mid and C terminus were also able to block CTL killing. In addition, monoclonal antibodies to the mid-region of APGSTAPPA peptide presented by H-2Db were able to bind to the peptide whilst bound to the groove. Molecular modelling demonstrated that the mid (in H-2Db) and C- (in H-2Kb) terminus of the peptide looped out and above the peptide binding groove (Figure 2). This is the first demonstration of a peptide bound to MHC class I, with the mid and C-terminus looping out of the peptide binding groove. A structure would give further new insights of peptides binding to MHC class I molecules.

262

V. Apostolopoulos and M. Plcbanski / Unusual Interactions

Figure 2. Molecular model of high affinity canonical anchor motif peptide (SEV9, black) and low affinity non-canonical motif peptide (MUC1-9, grey) when in complex with MHC class I, H-2Kb

Binding of glycopeptides to MHC class I: The H-2Kb presented peptide, VSV8, with P6 linked to galabiose disaccharide is able to activate T cells and the crystal structure of H2Kb-VSV8 glycopeptide has been determined [24]. The crystal structure of a synthetic OGlcNAc bearing peptide in complex with MHC class I H-2Db was also determined at the same time. In both these structures the sugar points towards the TCR. MUC1 is highly glycosylated and, as result of aberrant glycosylation during cancer, MUC1 exposes certain carbohydrate, peptide and possibly glycopeptides that are not normally exposed on the normal mucin. In addition, due to the overexpression of MUC1 on cancer cells as compared to the normal cells, it is possible that such glycopeptides are presented by MHC class I or class II molecules for recognition by T cells. A number of studies have demonstrated potential glycosylation sites within the MUC1 VNTR region, reviewd in, [13] and studies are being pursued to identify specific glycopeptides presented by MHC class I or class II molecules. The identification of glycopeptides capable of stimulating an immune response, may be of potential value for immunotherapy studies in cancer and in human vaccination protocols [25]. Peptide mimetics for cancer The concept of molecular mimicry has been popular for many years, particularly in autoimmune disease where there is good evidence that mimicking epitopes present on viruses or bacteria to those on normal human tissue, give rise to autoimmune disease. Such concepts have been examined through the years and more recently the evidence is stronger in such diseases as multiple sclerosis, diabetes, rheumatoid arthritis and others. Mimics of

V. Apostolopoulos and M. Plebanski / Unusual Interactions

263

carbohydrates by proteins have also been described, wherein the shape of the protein and carbohydrate is sufficiently similar for them to be recognised by the Abs or CTLs. While molecular mimicry has mostly been discussed with regard to autoimmunity, there are now examples of such mimics occurring in cancer. Immunising humans (or mice) with an unrelated mimic peptide may have a therapeutic benefit to humans. A key question is whether a mimic, being different from the self molecule, will be more immunogenic and therefore more potent, but more likely to give rise to autoimmunity. The real answer to this, is by appropriate clinical trials and it would be of interest to see if mimics are more potent than attempting to induce immune responses to self cancer antigens. Peptide mimetics to mouse lymphoma and human melanoma: The identification of CTL epitopes of tumour antigens is a crucial first step for active tumour immunotherapy. One strategy is to use defined amino acids in two MHC anchor positions to search for epitopes that are recognized by CTL. One study used H2Db and H2K CTL for the mouse lymphoma EL4. Peptide libraries were screened and individual peptides were tested in an RMA-S cell assay. One H2Db and one H2Kb peptide mimics were deduced from the original library that contained more than 2xl0 8 potential peptides. Mice immunised with the synthetic peptide mimic induced specific EL4 CTL as well as inhibiting established EL4 tumours in mice [26]. In another study, an amino acid substitution in the MART-l(27–35) (AAGIGELTV) sequence led to related sequences which could sensitise cells for anti-MART-1 CTL lysis; no tumour studies were performed [27]. These studies mainly involved conventional amino acid substitutions in the basic framework of a peptide epitope. We have demonstrated one of the first examples of a completely different and unrelated peptide that elicits specific CTL and tumour protection to a cancer antigen. These methods represent an approach of identifying MHC class I epitopes and epitope mimics that can serve as immunogens for tumour immunotherapeutic vaccines. Peptide mimetic to human adenocarcinoma (MUC1 antigen): A peptide (DAHWESWL), isolated from a phage peptide library, mimics Gala(l,3)Gal expressed on pig endothelial cells [28]. In addition, DAHWESWL peptide also mimicked MUC1 peptides [29, 30], in particular, SAPDTRPAP within the MUC1 VNTR region (WESW mimicked PDTR). MUC1, a glycoprotein overexpressed on adenocarcinomas. However, could DAHWESWL mimic MUC1 peptides in association with MHC Class I molecules for recognition by T cells?. T cells recognise peptides presented in association with MHC-Class I molecules. To find a peptide mimic of MUC1, the mimic must be presented by the appropriate MHC molecule and the amino acids recognised by the T cell receptor must mimic that of the original peptide. DBA/2 (H2d) mice were injected with DAHWESWL peptide coupled to KLH. Mice generated Abs, and CTL responses to DAHWESWL peptide which also recognised MUC1 peptides. Mice were also protected against a MUC1+ tumour challenge [29]. C57BL/6 (H2b) and HLA-A*0201 mice could not generate CTLs. Using overlapping 9mer peptides spanning the 20 amino acid VNTR region of MUC1, the MUC1 epitope which was recognized by the anti-DAHWESWL CTLs was determined, which were SAPDTRPAP (H2Dd) and APDTRPAPG (H2Ld) from the MUC1 VNTR [29]. Using molecular modeling it was demonstrated that R5/R6 in SAPDTRPAP and APDTRPAPG and E5 in DAHWESWL projected downwards to the base of the MHC cleft; D4T5 from SAPDTRPAP, T4P6 from APDTRPAPG and W4 are located in the gap of the TCR. At the C-terminus of the peptide, A7P8 of APDTRPAPG and W7 of DAHWESWL also occupied similar space [29]. Although the amino acid composition of the peptide varied at the interface of the TCR, the shape and volume occupied by the mimic and MUC1 peptides is similar. Because of the lack of CTLs in HLA-A2 transgenic mice, the mimic DAHWESWL

264

V. Apostolopoulos and M. Plebanski / Unusual Interactions

peptide was mutated to a HLA-A2 compatible peptide (DLHWASWV) and CTLs were now induced [29]. Altered peptide ligands Recent evidence suggests that T cell epitopes with unusual properties can be maintained by populations of infectious pathogens to actively turn off protective T cell responses. The clearest example is malaria. Plasmodium falciparum malaria is a major cause of death in tropical regions [31]. Interferon gamma has emerged as one of the key mediators of protection [32]. Variability in specific parasite T cell epitopes can suppress IFNY secretion and cytotoxic activity by altered peptide ligand (APL) antagonism [33, 34]. Moreover, our recent findings suggest that the malaria parasite may actively switch Trl cells from IFNy to IL10 production by APL antagonism [34]. Trl cells are found frequently in malaria exposed individuals [35]. Thl cell secretion of pro-inflammatory lymphokines (IFNy and TNFa) can mediate malaria liver stage protection and early bloodstage immunity by nitrate based mechanisms, whereas Th2 and anti-inflammatory responses may promote antibody mediated protection and benefit the host by dampening potentially damaging TNFa responses [32, 36]. Interleukin 10 is a pleiotropic lymphokine which can directly or indirectly (via inhibition of accessory cell function) suppress T cell proliferation and interferon gamma (IFNY), tumour necrosis factor alpha (TNFa), granulocyte and monocyte colony stimulating factor (GM-CSF) and lymphotoxin (LT) production by T cells [37, 38]. Thus, although it can suppress proliferation of T cell in general, including T helper type 2 (Th2: cells secreting lymphokines like IL4, IL5, DL10 and IL13), it is mostly thought of as a specific inhibitor of T helper type 1 responses (Thl: cells secreting lymphokines like IL2. IFNY and TNFa) [39]. Interferon gamma, by contrast, up-regulates cellular immunity by a variety of mechanisms, for example by increasing the cytotoxic activity of macrophages, natural killer (NK) cells and T cells, the antigen presenting potential of macrophages and dendritic cells, the expression of MHC class I and II and adhesion molecules on somatic cells and by stimulating the production of other activating soluble pro-inflammatory factors such as TNFa [32, 39, 40]. Although it is now accepted that a spectrum of T cells with different lymphokine patterns may exist outside Thl/Th2, the existence of T cell clones capable of specifically secreting both IL10 and IFNY, which we will denominate T regulatory cells typel (Trl), is surprising. These cells are distinct from Th0, since they fail to secrete other Th2 type lymphokines such as IL4 [34, 35, 38, 41]. The fact that these cells can switch from IFNy to IL10 production suggests a potential for self-regulation [34]. This would radically alter current understanding of immune regulation as being mediated by distinct T cell types, since a cell of the same antigen specificity could perform a variety of opposed effector functions during its lifetime. It will be therefore important to characterise this novel T cell subset, both in homeostasis and in disease. Crystal structure ofMHC-agonist/antagonist peptide-TCR: We have determined the crystal structure of superagonist, agonist and antagonist peptides in complex with MHC and TCR. From these studies, it was evident that peptides which are involved in the trimolecular complex, (MHC-peptide-TCR), and cause an antagonistic effect (ie. loss of T cell activation) have a loss of hydrogen bond contacts of peptide side chains to the CDR3 loops of the TCR [42]. In such an event this loss of hydrogen bond contact causes an agonist peptide to become an antagonist peptide [42]. Mutation of large side chains of the peptide which interact with the TCR to small side chain amino acids (such as, Ala. Ser or Gly) can

V. Apostolopoulos and M. Plebanski / Unusual Interactions

265

cause antagonism, and thus, inhibit disease. These findings are important in designing antagonistic effects of an agonist peptide in infectious diseases, cancer and in autoimmunity. Future Prospects MHC on the surface of APCs to T cell receptors presents peptides. Recognition of peptide-MHC by T cells initiates a cascade of signals in T cells which maintains a T cell dependent immune response. An understanding of the how peptides bind to MHC class I molecules is important to determine the physiology of immune responses and how to manipulate peptides to affect an immune response. Thus far, only the crystal structures of high affinity binding peptides to MHC have been determined. We have determined structurally, features of non-canonical anchor motif peptides, one which binds with low affinity and the other with high affinity to MHC class I. Further structural studies in determining how long peptides (14-mer), short peptides (5-mer), glycopeptides and altered peptide ligands bind to the MHC and hence to the TCR are fundamental in defining additional immunogenic peptides from primary protein sequences and aid in the design of alternative approaches for vaccirtes. Acknowledgments This work was supported in part by, NH&MRC CJ Martin Fellowship, New Idea Breast Cancer Foundation and Hellenic Funds (V.A.), Howard Hughes Fellowship (M.P.) and The Austin Research Institute (V.A. and M.P.). References [I] Zinkernagel, RM and Doherty, PC Nature 248 (1974) 701. [2] Germain, RN Cell 1994; 76: 287–299,Townsend, AR, Gotch, FM and Davey, J Cell 42 (1985) 457. [3] Zinkernagel, RM and Doherty, PC Adv Immunol 27 (1979) 51. [4] Falk, K, Rotzschke, O, Stevanovic, S, Jung, G and Rammensee, HG Nature 351 (1991) 290. [5] Saper, MA, Bjorkman, PJ and Wiley, DC J Mol Biol 219 (1991) 277. [6] Rammensee, HG, Friede, T and Stevanoviic, S Immunogenetics 41 (1995) 178. [7] Mandelboim, O, Bar-Haim, E, Vadai, E, Fridkin, M and Eisenbach, L J Immunol 159 (1997) 6030, Gao, L, Walter, J, Travers, P, Stauss, H and Chain, BM J Immunol 155 (1995) 5519. [8] Apostolopoulos, V, Karanikas, V, Haurum, JS and McKenzie, IF J Immunol 159 (1997) 5211. [9] Apostolopoulos, V, Haurum, JS and McKenzie, IF Eur J Immunol 27 (1997) 2579. [10] Sette, A, Vitiello, A, Reherman, B, Fowler, P, Nayersina, R, Kast, WM, Melief, CJ, Oseroff, C, Yuan, L, Ruppert, J and et al. J Immunol 153 (1994) 5586. [ I I ] Stryhn, A, Pedersen, LO, Holm, A and Buus, S Eur J Immunol 30 (2000) 3089. [12] Tirosh, B, el-Shami, K, Vaisman, N, Carmon, L, Bar-Haim, E, Vadai, E, Feldman, M, Fridkin, M and Eisenbach, L Immunol Lett 70(1999) 21. [13] Apostolopoulos, V, McKenzie, IF and Wilson, IA Front Biosci 6 (2001) D1311. [14] Apostolopoulos, V, Yu, M, McKenzie, IF and Wilson, IA Curr Opin Mol Ther 2 (2000) 29. [15] Fremont, DH, Stura, EA, Matsumura, M, Peterson, PA and Wilson, IA Proc Natl Acad Sci U S A 92 (1995)2479. [16] Matsumura, M, Fremont, DH, Peterson, PA and Wilson, IA Science 257 (1992) 927, Fremont, DH, Matsumura, M. Stura, EA, Peterson, PA and Wilson, IA Science 257 (1992) 919. [17] Horig, H. Young, AC, Papadopoulos, NJ, DiLorenzo, TP and Nathenson, SG J Immunol 163 (1999) 4434. [18] Collins, EJ, Garboczi, DN and Wiley, DC Nature 371 (1994) 626 [ 19] Matsumura. M, Saito, Y, Jackson, MR, Song, ES and Peterson. PA J Biol Chem 267 (1992) 23589.

266

V. Apostolopoulos and M. Plebanski / Unusual Interactions

[20] Olsen, AC, Pedersen, LO, Hansen, AS, Nissen, MH, Olsen, M, Hansen, PR, Holm, A and Buus, S Ear J Immunol 24 (1994) 3K5. [21] Apostolopoulos, V, Pietersz, GA, Loveland, BE, Sandrin, MS and McKenzie, IF Proc Nail Acad Sci U S A 1995; 92: 10128-10132,Apostolopoulos, V, Pietersz, GA and McKenzie, IF Vaccine 14 (1996) 930. [22] Reddehase, MJ, Rothbard, JB and Koszinowski, UH Nature 337 (1989) 651, Eisen, HN, Sykulev, Y and Tsomides, TJ Adv Protein Chem 1996; 49: l-56,Gillanders, WE. Hanson. HL, Rubocki, RJ, Hansen. TH and Connolly, JM Int Immunol 9 (1997) 81. [23] Apostolopoulos, V, Chelvanayagam, G, Xing, PX and McKenzie. IF J Immunol 161 (1998) 767. [24] Speir, JA, Abdel-Motal, UM, Jondal, M and Wilson, IA Immunity 10 (1999) 51. [25]Muller, S, Goletz, S, Packer, N, Gooley, A, Lawson, AM and Hanisch, FG J Biol Chem 272 (1997) 24780. [26] Blake, J, Johnston, JV, Hellstrom, KE, Marquardt, H and Chen, L J Exp Med 184 (1996) 121. [27] Loftus, DJ, Castelli, C, Clay, TM, Squarcina, P, Marincola, FM, Nishimura, MI, Parmiani, G, Appella. E and Rivoltini, L J Exp Med 184 (1996) 647. [28] Vaughan, HA, Loveland, BE and Sandrin, MS Transplantation 58 (1994) 879. [29] Apostolopoulos, V, Lofthouse, SA, Popovski, V, Chelvanayagam, G, Sandrin, MS and McKenzie, IF Nat Biotechnol 16 (1998) 276. [30] Apostolopoulos, V, Osinski, C and McKenzie, IF Nat Med 4(1998) 315–320,Sandrin, MS, Vaughan, HA. Xing, PX and McKenzie, IF Glycoconj J 14 (1997) 97. [31] Snow, RW, Craig, M, Deichmann, U and Marsh, K Bull World Health Organ 77 (1999) 624. [32] Good, MF and Dooian, DL Curr Opin Immunol 11 (1999) 412. [33] Gilbert, SC, Plebanski, M, Gupta, S, Morris, J, Cox, M, Aidoo, M, Kwiatkowski, D, Greenwood, BM, Whittle, HC and Hill, AV Science 279 (1998) 1173,Plebanski, M, Lee, EA and Hill, AV Parasitology 115 (1997) S55,Plebanski, M, Lee, EA, Hannan, CM, Flanagan, KL, Gilbert, SC, Gravenor, MB and Hill. AV Nat Med 5 (1999) 565. [34] Plebanski, M, Flanagan, KL, Lee, EA, Reece, WH, Hart, K, Gelder. C. Gillespie, G, Pinder, M and Hill, AV Immunity 10 (1999) 651. [35] Winkler, S, Willheim, M, Baier, K, Schmid, D, Aichelburg, A, Graninger, W and Kremsner, PG Journal of Infectious Diseases 179 (1999) 209. [36] Stevenson, MM and Tarn, MF Clin Exp Immunol 92 (1993) 77, Taylor-Robinson, AW, Phillips. RS, Severn, A, Moncada, S and Liew, FY Science 260 (1993) 1931, von der Weid, T and Langhorne, J Immunobiology 189 (1993) 397, Kwiatkowski, D, Bate, CA, Scragg, IG, Beattie, P, Udalova, I and Knight, JC Ann Trap Med Parasitol 91 (1997) 533. [37] Cobbold, S and Waldmann, H Curr. Opin. Immunol. 1998; 10: 518–524,Moore, KW, O'Garra, A, de Waal Malefyt, R, Vieira, P and Mosmann, TR Annu Rev Immunol 11 (1993) 165. [38] Groux, H and Powrie, F Immunol Today 20 (1999) 442. [39] Mosmann, TR and Moore, KW Immunol Today 12 (1991) A49. [40] Boehm, U, Klamp, T, Groot, M and Howard, JC Annu Rev Immunol 15 (1997) 749. [41] Allen, JE and Maizels, RM Immunol Today 18 (1997) 387,Groux, H, O'Garra, A, Bigler, M, Rouleau, M, Antonenko, S, de Vries, JE and Roncarolo, MG Nature 389 (1997) 737, Asseman, C and Powrie, F Gut 42 (1998)157. [42] Degano, M, Garcia, KC, Apostolopoulos, V, Rudolph, MG, Teyton, L and Wilson, IA Immunity 12 (2000)251.

Drug Discovery and Design: Medical Aspects J. Matsoukasand T. Mavromoustakos (Eds.) 1OS Press, 2002

267

Effects of Angiotensin II, III and IV on Memory Retention of Rats: Interaction with Adenosine AI Receptor Related Ligands J. Tchekalarova and V. Georgiev Laboratory "Experimental Psychopharmacology", Institute of Physiology, Bulgarian Academy of Sciences, 1113 Sofia, Bulgaria Abstract. Angiotensins II, III and IV (Ang II, Ang III and Ang IV) exert a crucial role in learning and memory processes either alone or interacting with other neurotransmitters and neuromodulators. The present study examines the functional interaction between angiotensin II, III, IV and adenosine AI receptorrelated drugs on passive avoidance (step-through) conditioning in rats. Ang II and III facilitated while the heptapeptide decreased the retention response. Adenosine A, receptors could modulate the realization of angiotensins memory effects. Adenosine AI and Ang II receptors interact in adaptation to long-term theophylline administration in step-through memory task. Taken together, the obtained data suggest that angiotensin peptides are closely implicated in the regulation of memory retention processes and mutual interaction with adenosine A, receptors could play an important role in its realization.

Introduction Increasing evidence suggests that biologically active forms of angiotensins, II, in and IV (Ang II, Ang III and Ang IV) including their specific binding sites are present in the central nervous system (CNS). The accumulated data indicate that the angiotensin peptides play major role in the central neural excitability (memory processes included) either alone or in close relationship with certain neurotransmitters and neuromodulators. The effects of angiotensin ligands (Ang II, Ang III, Ang IV) are mediated through their interaction with specific angiotensin receptors. Due to the development of selective peptidic or nonpeptidic Ang II ligands several subtypes of angiotensin receptors have been identified. The AT] subtype binds with high affinity the imidazolic nonpeptidic compound DuP 753 or losartan but not the spinacme derivative PD 123319 or the pseudopeptidic Ang II analog CGP 42

268

J. Tchekalarova and V. Georgiev I Effects of Angiotensin II, III and IV

112 A. AT| receptor is a seven-transmembrane and G-protein-coupled receptor. It is present in high densities in areas closely related with the "classic functions" of the brain RAS (Renin Angiotensin System) including regulation of body fluid homeostasis, heart rate and blood pressure and control of vasopressin release. The AT2 subtype binds with high affinity to PD 123319 and CGP 42 112 A but not to DuP 753. AT2 is also a seven-transmembrane domain and G-protein coupled receptor. It is present in brain areas rather different of those for ATi site. The AT2 subtype is supposed to play a role in vascular growth, the autoregulation of cerebral blood flow, cGMP regulation and modification of signalling of other receptors. The AT4 subtype is distinct from the AT| and AT2 sites. It has been determined that [I25I] Ang IV binds to the AT4 site reversibly, saturably, and with high affinity. This subtype is not G-protein linked. The AT4 site is expressed in high density predominately in brain areas involved in the enhancement of memory acquisition and retrieval, regulation of blood flow, angiogenesis, etc. Divalinal- Ang IV is a specific antagonist of AT4 receptor [ 1 ]. The observed effects of ANG n on cognitive processes have been inconsistent. Thus, ANG n, infused into the dorsal striatum after passive avoidance conditioning, disrupted the retention 24 h later [2]. On the other hand, we have demonstrated that Ang n exerted retention-improving effect in shuttle-box active avoidance and facilitated retention in step-trough passive avoidance task [3, 4]. Angiotensin ATi receptor subtype has been implicated in the mediation of the octapeptide's Ang II effects on cognition. Braszko et al. first showed that i.c.v. infused Ang IV, in the equivalent dose of 1 nmol with Ang II, improved recall of passive avoidance conditioning and acquisition of active avoidance conditioning [5]. Wright et al. extended these finding revealing that ANG IV effect on passive avoidance response is dose-dependent and the selective AT4 receptor antagonist Divalinal-Ang IV (10 nmol), severely disrupted recall of this memory effect [6]. Furthermore, metabolically resistant analogs of Ang IV facilitated acquisition and retention in a circular water maze task in scopolamine treated rats, perforant path damaged rats and ischemia-induced hippocampal damaged gerbils [7]. Adenosine AI receptors play a significant role in the cognition. Thus, the selective agonists for AI receptor as N^-cyclopentyladenosine (CPA) and N^-U-methyl-2phenyl)adenosine (R-PIA) have been shown to impair the retention [8, 9] while the AI receptor antagonists as theophylline produce memory enhancement in passive avoidance tests [9]. There are several reports that have considered adenosinergic and angiotensinergic interactions in the periphery and the central nervous system (CNS). Thus, ANG n leads to a release of adenosine from the rat lung [10] and to an increase in the plasma adenosine concentration in renovascular hypertensive rats [11]. In the CNS, endogenous angiotensin modulates responses on blood pressure and heart rate to adenosine when microinjected into the area postrema of rats [12]. In our previous studies, we suggested that angiotensins n, in and IV as well as adenosine AI receptors could interact in the regulation of the seizure susceptibility [13, 14, 15]. The process of long-term potentiation (LTP) in hippocampus represents a basic physiological mechanism of memory storage in the brain. Accumulating data reveal that angiotensins play a major role in the excitability of hippocampus, including LTP. Wayner et al. has shown that Ang II microinjected in the hippocampus inhibited LTP in perforant path-stimulated dentate granule cells [16]. This Ang H-induced inhibition could be blocked with saralasin or DuP 753 but not by PD 123319. Ang III was found to be 40 to 50-fold less potent than Ang II at inhibition of LTP [16]. Further, AT4 receptor stimulation via native Ang IV or its more selective analog Nle'-Ang IV, enhanced synaptic transmission and LTP. while infusion of the AT4 receptor antagonist, Nle'-Llual^-Ang IV, disrupted stabilization of LTP [17]. These results are in agreement with the above mentioned behavioral work

J. Tchekalarova and V. Georgiev /Effects

of Angiotensin II, III and IV

269

indicating that AT| receptors mediated inhibition while AT4 receptors activation enhance the cognition in different memory paradigms. The present review aims to summarize the post-trial effects of bilogically active forms of angiotensin II, III and IV as well as of their interaction with adenosine AI receptor-related drugs on inhibitory avoidance (passive avoidance) task in rats. It is beyond the mission of this review to present detailed treatments of this topic, but rather to summarize the major findings of the pertinent literature and particularly those from our laboratory. Ang II-, HI-, IV- and adenosine Aj- receptor interactions: There is increasing evidence of interactions between Ang n and adenosine system in the CNS. Previously, in our laboratory we have found agonistic angiotensin-adenosine relationships in seizure susceptibility [13, 14, 15, 18]. Further, we extended this study on cognition processes, revealing antagonistic angiotensin-adenosine AI receptor-mediated interactions. For the purpose, we used steptrough method with punishment reinforcement where the rats were trained on a one-trial passive avoidance task. Retention was assessed 24 hours and 7 days later, using a procedure identical to that for training, except that footshock was not delivered. The latency to cross from the outer to the inner compartment with the four legs was recorded. Retentionenhancing effect exerted by Ang II was abolished by adenosine AI receptor agonist N6cyclohexyladenosine (CHA) upon re-testing of rats 24 but not 7 days after passive avoidance training (Table 1) [19]. Table 1. Latencies for the rats to enter the dark compartment in passive avoidance task measured 24 h and 7 days after training. The drugs were injected i.c.v. or i.p. immediately after training trial. n=10 animal/group; *p<0.05, ** p<0.01 vs. controls; °p<0.05, °°p<0.01 vs. angiotensin II, III or IV, respectively; +p<0.05, ++p<0.01 vs. CHA or theophylline, respectively.

Ang II 0. 1 u,g CHA 0.1 u.g Ang 0+CHA theoph.75mg/kg AngllO.l+theoph Ang III 0.1 ng

24th h Latency (%) 44% 70% 15% 40% 44% 30% 16%

Ang III+CHA

60%

Anglll+theoph Ang IV 0.1 [ig Ang IV+CPA AnglVl+theoph

2% 70% 34.3 % 72%

Drugs Controls

%2 3.8* 2.39* 7.2°° 3.83* 3.4* 4.34°° 3.83++ 3.86+ 5.79* 3.35° 6.99+0

7th day Latency (%)

30% 23% 22% 29% 20% 4%

*2 6.39**

26%

3.2*

2% 67% 42% 20%

4.5++ 4.78* 2.39°

Theophylline is a non-selective A]/A2 adenosine receptor antagonist that is known to improve cognitive performance. Prolonged exposure of adenosine AI receptors to their ligands lead to adaptive changes. Thus, chronic treatment with either adenosine agonist or antagonist results in cognitive responses that differ from those observed following acute administration of these compounds. Recently, we have demonstrated that the increasing effect of adenosine AI receptor antagonist 8-p-sulfophenyl-theophylline (8-p-SPT) on retention was potentiated by Ang II after chronic (14 days) systemic treatment of rats with

J. Tchekalarova and V. Georgiev /Effects

270

ofAngiotensin II, III and IV

theophylline (Table 2) [Tchekalarova, Georgiev, unpublished results]. On the contrary, the non-specific angiotensin receptor antagonist, sarilesin, showed opposite, memoryimproving effect, on theophylline-treated animals compared with the acute experiment where it demonstrated amnesic effect in passive avoidance task (Table 2). Table 2. Latencies for the rats to enter the dark compartment in passive avoidance task measured 24 h after training. Chronic theophylline was administered 14 days in dose of 75 mg/kg, i.p. The drugs were injected i.c.v. immediately after training trial. n=10 animal/group; *p<0.05 vs. respective controls.

Drugs controls 1 controls 2 (chronic theophylline) 8-p-SPT 50 ng (acute) 8-p-SPT 50 jig (chronic theophylline) sarilesin 1 pig (acute) sarilesin 1 |Xg (chronic theophylline)

24*11 Latency (%) 37%

X2

29%

-

46%

-

82%

3.8*

5%

4.89*

60%

-

The heptapeptide Ang HI, which impaired long-term memory upon re-testing of rats 24 h and 7 days later in step-through task, produced an opposite, enhancing effect on the retention when was co-administered with CHA (Table 1) [19]. Furthermore, Ang ID attenuated the memory-improving effect induced by theophylline after re-testing the rats 24 h and 7 days later. We confirmed and extended the findings reported by Braszko et al. and Wright et al. [5, 61] that when Ang IV is injected i.c.v., improved the recall of a passive avoidance response. Thus, this facilitatory effect of the hexapeptide was reduced by adenosine AI receptor agonist N6-cyclopentyladenosine (CPA) while combination of theophylline and Ang IV, both injected in ineffective doses, facilitated the retention 24 hours after training but not 7 days later (Table 1) [20]. Conclusions The present results demonstrate that Ang n and Ang ID produce opposite effects in passive avoidance conditioning as the octapeptide facilitated while the heptapeptide decreased the retention response. Furthermore, adenosine AI receptors could modulate the realization of angiotensin's memory effects. In addition with the proposed close relationships between Ang H/Ang HI and adenosine-related drugs, we have shown that the enhancing effect of Ang IV upon passive avoidance conditioning might be due to the concurrent interaction of AT4 and adenosine AI receptors. So far, most of the effects of angiotensins on cognition seem to be mediated by AT| or AT4 receptor subtypes. The role of AT2 receptor subtype is less clear although there are few studies suggesting that this receptor might mediate the cognitive enhancing action of Ang n [7]. Ang II has been recognized as the major peptide activating ATi receptors in the periphery while Ang HI could be the predominant effector peptide in the brain [7]. It is tempting to suggest that Ang II enhancing effects on cognition could be mediated by AT4 receptors. In this respect, AT4 receptors have to be expected to being excited indirectly by Ang II through its degradation to Ang IV. This point of view is supported by the observed failure of AT| and AT2 receptor antagonists to block subsequent

J. Tchekalarova and V. Georgiev / Effects of Angiotensin II, HI and IV

271

Ang II facilitation of acquisition of a conditioned response. The latter was explained by Wright et al. [7] with degradation of Ang II to Ang III, then to Ang IV, the ligand that by acting on AT4 receptors, improves the performance. On the other hand, the observed Ang HI decreasing effect on memory retention might be realized through interaction with ATi receptors. The intriguing finding of the present study is based upon the observation that the selective adenosine AI receptor agonist CHA modulates the memory effects of Ang II, III and IV. Furthermore, Ang III was able to convert memory-enhancing effect produced by theophylline not only 24 h but also 7 days after the trial test which fits well with the longterm memory effects observed by this heptapeptide in the present study. The persistence of increased retention in Ang IV-treated mice 7 days after training might be explained with stimulation of the protein synthesis thus influencing late consolidation and long-term memory storage [21, 22]. On the other hand, long-lasting action of sarilesin seems to be a sequence of its resistance to peptidases allowing it to remain in the vicinity of the receptors as proactive material for a longer period. The exact mechanism of Ang's n, HI, IV and adenosine AI receptor-related drugs relationships in cognition processes is not quite clear and needs further studies for its elucidation. Acknowledgements This work was supported by European Commission through the COPERNICUS programme, contract No. CIPA CT 94-0239, and by Grant VFP-L-6 from the National Fund for Scientific Research at the Bulgarian Ministry of Education and Science, Sofia. References [I] M. De Gaspare, K.J. Catt, T. Inagami, J.W. Wright J.W., T. Unger, Pharmacol. Rev. 52 (2000) 415. [2] M.R. Zarrindast, B. Shafahi, Eur. J. Pharmacol. 256 (1994) 233. [3] V. Georgiev, D. Yonkov, Meth. Find. Exptl. Clin. Pharmacol. 7 (1985) 415. [4] V. Georgiev, D. Yonkov, T. Kambourova, Neuropeptides 12 (1988) 155. [5] J.J. Braszko, B. Kupryszewski, B. Witczuk, K. Wisniewski, Neuroscience 27 (1988) 777. [6] J.W. Wright, L.T. Krebs, J.W. Stobb, J.W. Harding, Front. Neuroendocrinol. 16 (1995) 23. [7] J.W. Wright, J.W. Harding, Brain Res. Rev. 25 (1997) 96. [8] H.L. Normile, R.A. Barraco, Brain Res. Bull. 27 (1991) 101. [9] D.K.J.E Von Lubitz, K.A. Jacobson, Behavioral effects of adenosine receptor stimulation. In " Adenosine and Adenine Nucleotides: from Molecular Biology to Integrative Physiology", Edited by L.Belardinelli and A. Pelleg, Kluver Academic Publs., Boston, Dordrecht, London, 1995, pp. 489–498 [10] D.F. Bottiglieri, K. Curtis, K.E. Jackson, J. Cardiovasc. Pharmacol. 16 (1990) 101. II1] A. Ohnishi, P. Li, A.R. Brauch, O.I. Biaggioni, K.E. Jackson, Hypertension 12 (1988) 152. [12] H.C. Lin, F.J. Wan, C. Tung, C.J. Tseng, J. Auton. Nerv. Syst. 51 (1995) 19. [13] V. Georgiev, J. TchekalarovaBrain Res. 779 (1998) 259. [14] J. Tchekalarova, V. Georgiev, J. Physiol. (Paris) 93 (1999) 191. [15] J. Tchekalarova, T. Kambourova, V. Georgiev, Brain Res. Bull. 52 (2000) 13. [16] M.J. Wayner, D.L. Armstrong, J.L. Polan-Curtain, J.B. Denny, Pharmacol. Biochem. Behav. 45 (1993) 455. [17] E. Kramar, D. Armstrong, S. Ikeda, M. Wayner, J. Harding, J. Wright, Brain Res. 897 (2001) 114. [18] J. Tchekalarova, T. Kambourova, V. Georgiev, Brain Res. Bull, (in press) [19] J. Tchekalarova, T. Kambourova, V. GeorgievBehav. Brain Res. (in press) [20] J. Tchekalarova, T. Kambourova, V. Georgiev, Behav. Brain. Res. 123 (2001) 113. [21] T. Herdegen. J.D. Leah, Brain Res. Bull. 28 (1998) 370. [22] K.A. Robert, L.T Krebs, E.A. Kramar, M.J. Shaffer, J.W. Harding, J.W. Wright, Brain Res. 682 (1995) 13.

272

Drug Discovery and Design: Medical Aspects J. Matsoukasand T. Mavromoustakos (Eds.) IOS Press, 2002

Molecular Targets and Compounds for AntiHIV Therapy E. De Clercq Rega Institute for Medical Research, Katholieke Universiteit Leuven, Minderbroedersstraat 10, B-3000 Leuven, Belgium Abstract. Virtually all the compounds that are currently used, or under advanced clinical trial, for the treatment of HIV infections, belong to one of the following classes: (1) nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs): i.e., zidovudine (AZT), didanosine (ddl), zalcitabine (ddC), stavudine (d4T), lamivudine (3TC), abacavir (ABC), emtricitabine [(-)FTC], tenofovir (PMPA) disoproxil fumarate; (ii) non-nucleoside reverse transcriptase inhibitors (NNRTIs): i.e., nevirapine, delavirdme, efavirenz, emivirine (MKC-442); and (iii) protease inhibitors (Pls): i.e., saquinavir, ritonavir, indinavir, nelfinavir, amprenavir, and lopinavir. In addition to the reverse transcriptase and protease step, various other events in the HIV replicative cycle are potential targets for chemotherapeutic intervention: (i) viral adsorption, through binding to the viral envelope glycoprotein gp120 (polysulfates, polysulfonates, polyoxometalates, zintevir, negatively charged albumins, cosalane analogues); (ii) viral entry, through blockade of the viral coreceptors CXCR4 and CCR5 [bicyclams (i.e. AMD3100), polyphemusins (T22), TAK-779, MlP-lct LD780 isoform]; (iii) virus-cell fusion, through binding to the viral glycoprotein gp41 (T-20 (DP-178), T-1249 (DP-107), siamycins, betulinic acid derivatives]; (iv) viral assembly and disassembly, through NCp7 zinc finger-targeted agents [2,2'-dithiobisbenzamides (DIBAs), azadicarbonamide (ADA) and NCp7 peptide mimics]; (v) proviral DNA integration, through integrase inhibitors such as L-chicoric acid and diketo acids (i.e. L-731,988); (yj) viral mRNA transcription, through inhibitors of the transcription (transactivation) process (fluoroquinolone K-12, Streptomyces product EM2487, temacrazine, CGP64222). Also, in recent years new NRTIs, NNRTIs and Pis have been developed that possess respectively improved metabolic characteristics (i.e. phosphoramidate and cyc/osaligenyl pronucleotides of d4T), or increased activity against NNRTI-resistant HIV strains [second generation NNRTIs, such as capravirine and the novel quinoxaline, quinazolinone, phenylethylthiazoly-lthiourea (PETT) and emivirine (MKC-442) analogues], or, as in the case of Pis, a different, non-peptidic scaffold [i.e. cyclic urea (DMP 450), 4-hydroxy-2-pyrone (tipranavir)]. Given the multitude of molecular targets with which anti-HIV agents can interact, one should be cautious in extrapolating from cell-free enzymatic assays to the mode of action of these agents in intact cells. A number of compounds (i.e. zintevir and L-chicoric acid, on the one hand; and CGP64222 on the other hand) have recently been found to interact with virus-cell binding and viral entry in contrast to their proposed modes of action targeted at the integrase and transactivation process, respectively.

E. De Clercq /Molecular Targets and Compounds for Anti-HIV Therapy

273

Introduction Combination therapy, comprising at least three anti-HIV drugs, has become the standard treatment of AIDS or HIV-infected patients. Virtually all drugs that have been licensed for clinical use (or made available through expanded access programmes) for the treatment of HIV infections fall into one of the following three categories: (i), nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs), that, following two phosphorylation steps (tenofovir) or three phosphorylation steps (zidovudine, didanosine, zalcitabine, stavudine, lamivudine, abacavir), act, as chain terminators, at the substrate binding site of the reverse transcriptase; (ii), non-nucleoside reverse transcriptase inhibitors (NNRTIs), that interact with the reverse transcriptase at an allosteric, non-substrate binding site (nevirapine, delavirdine, efavirenz); and (iii), protease inhibitors (PIs), that specifically inhibit, as peptidomimetics, the virus-associated protease (saquinavir, ritonavir, indinavir, nelfinavir, amprenavir, lopinavir). Guidelines to the major clinical trials with these compounds have been recently published [1]. In numerous studies, combinations of NRTIs, NNRTIs and PIs have been found to decrease HIV viral load, increase CD4 count, decrease mortality and delay disease progression, particularly in AIDS patients with advanced immune suppression [2]. When initiated during early asymptomatic HIV infection, highly active antiretroviral (combination) therapy (HAART) initiates rapid reversal of disease-induced T-cell activation, while preserving pretherapy levels of immune function, suggesting that therapeutic benefit may be gained from early aggressive anti-HIV chemotherapy [3]. Combination therapy that produces sustained suppression of plasma HIV RNA may also be able to reduce the virus burden in the lymphoid tissues [4], although clearance of plasma viremia is not invariably associated with immune restoration [5] and not at all paralleled by a reduction in viral DNA burden in the peripheral blood mononuclear cells [6]. The introduction of HAART or combined anti-HIV drug regimens has had profound repercussions on various AIDS-associated diseases. While partial immune restitution induced by HAART in patients with advanced HIV infection can exacerbate clinically apparent cryptococcal meningitis [7] or CMV vitritis, HAART may reduce the prevalence of cervical squamous intra-epithelial lesions in HIV-seropositive women caused by HPV [8]. Several studies have indicated that HAART significantly improves the prognosis of AIDS patients with progressive multifocal leukoencephalopathy [9–11]. Likewise, HAART has been shown to favorably alter the prognosis of CMV retinitis in HIV-infected individuals, as attested by long-lasting remission of CMV retinitis without CMV maintenance therapy [12], lack of reactivation of CMV retinitis after stopping CMV maintenance therapy [13], decrease of CMV replication (viremia) [14,15] and increased survival [16,17]. In the era of combination antiretroviral therapy the management of CMV diseases in patients with AIDS has undergone dramatic modifications [18,19]. Although the long-term goal of eradicating the virus from latently and chronically infected cells remains forbidding [20] the advent of so many new compounds, other than those that have been formally approved, for the treatment of HIV infections, will undoubtedly improve the prognosis of patients with AIDS and AIDS-associated diseases. Here I will primarily address those new anti-HIV compounds that (i) have emerged as promising anti-HIV drug candidates during the last few years, that (ii) are in preclinical or early-clinical development, and that (iji) are targeted at well-defined steps in the HIV replicative cycle. In particular, reverse trascriptase (RT) targeted at the substrate and allosteric site will be covered.

274

E. De Clercq /Molecular Targets and Compounds for Anti-HIV Therapy

Reverse transcriptase inhibitors targeted at the substrate binding site The substrate (dNTP) binding site of the HIV-1 reverse transcriptase (RT) is the target for a large variety of NRTI analogues, which have for several years [21] been recognized as efficacious agents for the treatment of HIV infections: i.e., zidovudine (AZT), didanosine (ddl), zalcitabine (ddC), stavudine (d4T), lamivudine (3TC), abacavir (ABC). Other agents in clinical trial include tenofovir (PMPA) and emtricitabine [(-)FTC]. As a rule, all of these compounds must be phosphorylated to their 5'-triphosphate form, before they can act as competitive inhibitors/substrate analogues/chain terminators at the reverse transcriptase level (Figure 1). In contrast to the nucleoside analogues, the nucleotide analogues PMEA and PMPA (Figure 2) are already equipped with a phosphonate group, and thus only need two phosphorylation steps to be converted to the active metabolite [22]. From PMEA and PMPA the oral prodrug forms [bis(POM)-PMEA or adefovir dipivoxyl, and bis(POC)-PMPA or tenofovir disoproxil fumarate, respectively] have been prepared. The former is now in phase III clinical trials for the treatment of hepatitis B virus (HBV) infections, whereas the latter has gone through several phase III clinical trials for the treatment of HIV infections [23]. Tenofovir has proven to be antivirally active in heavily antiretroviral-experienced patients; it has a favorable activity/resistance profile in that it by itself leads to little, if any, resistance, while retaining full activity against HIV-1 RT mutations that have developed resistance to 3TC (Ml84V mutants) or multinucleosids (Ql51M mutants) [24]. NH,

bis(POM)-PMEA Adefovir dipivoxU

o

II

(CH&CH—O—C—O—CH2

O.

(CHjfeCH—O—C—O—CHj

O

O

bis(POC)-PMPA Tenofovir disoproxil Figure 1. Mechanism of anti-HFV action of 2',3'-dideoxynucleoside analogues, as exemplified forzidovudine (azidothymidine, AZT, AzddThd).

E. De Clercq / Molecular Targets and Compounds for Anti-HIV Therapy

275

CH,—C—S—CHj—CH2—O' II O Bis-S-acetyl-thioethylester

H,CO—C—CH— HN O

CH,

Aryloxyphosphoramidate

d4T

Cyclo saligenyl

Figure 2. Acyclic nucleoside phosphonates: PMEA (adefovir) and PMPA (tenofovir) in their prodrug forms, adefovir dipivoxil and tenofovir disoproxil. Prodrugs of ddNMPs: Bis(S-acetyl-thioethyl)ester of ddAMP, (Aryloxy)phosphoramidate of d4T, and cycloSal-d4TMP and cycloSal-ddAMP.

In addition to 3TC and (-)FTC, the structurally related 2'-deoxy-3'-oxa-4'thiocytidine (BCH-10652, dOTC) [25], the dioxolane purine nucleoside analogues [26] and the methylenecyclopropane nucleoside analogues (and their phosphoro-L-alaninate diesters) [27,28] have recently been described as new anti-HIV agents. Emtricitabine [(-) FTC] is in phase III trials for HIV and phase I/II trials for HBV; it is considered for use in the multidrug combination therapy of HIV-1 and HBV infections [29]. DAPD, (-)-|3-D-2,6diaminopurine dioxolane, which is converted by adenosine deaminase to (-)-p-D-dioxolane guanine (DXG) (which is then further metabolized intracellularly to its active metabolite DXG-TP, a strong alternative substrate inhibitor of the HIV RT) [30], has proven active against AZT- and 3TC-resistant HIV-1 strains and has proceeded to phase I/II clinical studies [31]. BCH-10652 (dOTC) has demonstrated activity against HIV-1 in the SCID-hu Thy/Liv model; despite its structural similarity to 3TC it proved also active against 3TCresistant HIV-1 (Ml84V), albeit at a relatively high dosage level (400 mg/kg/day) [32]. All the nucleoside analogues mentioned above may ultimately, upon their intracellular metabolism, act as chain terminators, akin to AZT (Fig. 9). Chain-terminating nucleotides inhibit HIV replication because their incorporation at the 3' end of the nascent DNA chain prevents further elongation. However, it has been recently shown that under physiological conditions RT can remove these chain terminators and thus unblock the primer termini [33,34]. Mutant RTs associated with AZT resistance would be capable of removing the chain terminating residue with much greater efficiency than wild-type RT and this should, in turn, facilitate rescue of DNA synthesis. On the other hand, the Ml84V mutation, which confers high-level resistance to 3TC, would severely impair the removal of chain-terminating nucleotides and hence suppress the rescue of AZT-terminated DNA synthesis [35]. Similarly, foscarnet resistance mutations would also suppress the removal of

276

E. De Clercq / Molecular Targets and Compounds for Anti-HIV Therapy

AZT monophosphate from the blocked primer/templates, thus providing a likely explanation for the resensitization of AZT-resistant HIV-1 strains to AZT [36]. The bottleneck in the metabolic pathway leading from AZT and the other 2',3'dideoxynucleoside (ddN) analogues to their active 5'-triphosphate form is the first phosphorylation step. Therefore attempts have been made at constructing 2',3'dideoxynucleotide (ddNMP) prodrugs, that, once they have been taken up by the cells, deliver the nucleotide (ddNMP) form. This approach has proven particularly successful for a number of NRTIs such as 2',3'-dideoxyadenosine (ddA) and d4T (Fig. 2). Thus, the bis(S-acetyl-2-thioethyl)phosphotriester of ddA [bis(SATE)ddAMP] was synthesized and found to be 1000-fold more potent against HIV than the parent compound [37]. By directly delivering bis(SATE)ddAMP into the cells, one circumvents the adenosine deaminase step that would otherwise rapidly degrade ddA to ddI. Similarly, aryloxyphosphoramidate derivatives (Fig. 2) of d4T [i.e., So324, a d4TMP prodrug containing at the phosphate moiety a phenyl group and the methylester of alanine linked to the phosphate group through a phosphoramidate linkage] have been constructed [38–40]. Once taken up by the cells, d4TMP is released intracellularly (via the alaninyl d4TMP intermediate) [41]. The latter could be considered as an intracellular depot form of d4TMP [41]. This phosphoramidate prodrug approach does not seem to work so well with zidovudine, where the main metabolite formed from the alaninyl AZTMP intermediate is AZT rather than AZTMP [42] thus explaining why d4T phosphoramidate prodrugs, but not AZT phosphoramidate prodrugs, retain anti-HIV activity in HlV-infected thymidine kinase-deficient cell cultures. In resting monocytes/macrophages (M/M) the aryloxyphosphoramidate derivatives of d4T, d4A and ddA provided an anti-HIV activity that was 25- to 625-fold greater than that of the parent nucleosides (d4T, d4A and ddA) [43]. The thymidine kinase (in the case of d4T) and the adenosine deaminase (in the case of ddA) can also be bypassed by using the cyclic saligenyl approach [44,45]. CycloSaligenyl pronucleotides of d4T and ddA (Fig. 10) deliver exclusively the nucleotides d4TMP and ddAMP, not only under chemical-simulated hydrolysis conditions but also under intracellular conditions [46,47]. The cycloSal approach has also been applied, with success, to F-ara-ddA (lodenosine, 2'-fluoro-ara-2',3'-dideoxyadenosine) [48]. Conclusions In recent years, an ever increasing number of compounds have been uncovered as anti-HIV agents targeted at virtually any step of the virus replicative cycle: adsorption, entry, fusion, uncoating, reverse transcription, integration, transcription (transactivation), and maturation. In addition to the "newer" NRTIs, NNRTIs and PIs, various other compounds, i.e. those that are targeted at viral entry (i.e. CXCR4 and CCR5 antagonists) and virus-cell adsorption/fusion (i.e. compounds interacting with either gp120 or gp41), offer great potential for the treatment of HIV infections. Most of the anti-HIV agents seem to accomplish their anti-HIV activity through an interaction with the presumed molecular target. However, quite a number of compounds are capable of interacting with more than one target so their true mechanism of action may not be what was expected. Three examples in point are the G-octet containing 17-mer oligonucleotide zintevir, the dicaffeoyltartaric acid L-chicoric acid, and the nonapeptoid CGP 64222. Zintevir and Lchicoric acid were originally identified as integrase inhibitors, and the nonapeptoid as a transactivation (Tat) antagonist, and their anti-HIV activity in acutely infected cells was ascribed to interference with the integration and transactivation process, respectively. As it now appears, zintevir and L-chicoric acid primarily interact as virus adsorption inhibitors, and the nonapeptoid as a CXCR4 antagonist, and thus these compounds owe their anti-HIV

E. De Clercq / Molecular Targets and Compounds for Anti-HIV Therapy

277

activity mainly to interference with an early event (adsorption, entry) of the HIV replicative cycle. Caution should therefore be exercised in postulating from the alleged interaction of a certain anti-HIV agent with a certain viral protein (or enzyme) that the anti-HIV activity is necessarily due to this interaction [49,50]. The compound(s) may well be able to interact with other targets that are more crucial in the anti-HIV activity that is eventually achieved. Acknowledgements Prof. Erik De Clercq holds the Professor P. De Somer Chair of Microbiology at the Katholieke Universiteit Leuven School of Medicine. I thank Christiane Callebaut for her invaluable editorial assistance. References [1] J. A. Tavel, K. D. Miller, and H. Masur, Clin. Infect. Dis. 28 (1999) 643. [2] K. Henry, A. Erice, C. Tierney, H. H. Jr Balfour, M. A. Fischl, A. Kmack, S.-H. Liou, A. Kenton, M. S. Hirsch, J. Phair, A. Martinez, and J. O. Kahn, J. Acquir. Immun. Defic. Syndr. Hum, Retrovirol 19 (1998) 339. [3] L. R. Bisset, R. W. Cone, W. Huber, M. Battegay, P. L. Vernazza, R. Weber, P. J. Grob, M. Opravil, and the Swiss HIV Cohort Study, AIDS 12 (1998) 2115. [4] M. Harris, P. Patenaude, P. Cooperberg, D. Filipenko, A. Thorne, J. Raboud, S. Rae, P. Dailey, D. Chernoff, J. Todd, B. Conway, J.S.G. Montaner, and the INCAS Study Group, J. Infect. Dis. 176 (1997) 1388. [5] N. G. Pakker, E.D.M.B. Kroon, M.T.L. Roos, A. S. Otto, D. Hall, F.W.N.M. Wit, D. Hamann, M. van der Ende, F.A.P. Claessen, H. R. Kauffmann, P. K. Koopmans, F. P. Kroon, C.H.H. ten Napel, H. G. Sprenger, H. M. Weigel, J.S.G. Montaner, J.M.A. Lange, P. Reiss, P.T.A. Schellekens, and F. Miedema, AIDS 13 (1999) 203. [6] J. J. Zaunders, H. P. Cunningham, D. A. Kelleher, R. G. Kaufmann, B. A. Jaramillo, R. Wright, D. Smith, P. Grey, J. Vizzard, A. Carr, and A. D. Cooper, J. Infect. Dis. 180 (1999) 320. L. M. Woods II, R. MacGinley, P. D. Eisen, and M. A. Allworth, AIDS 12 (1998) 1491. Heard, V. Schmitz, D. Costagliola, G. Orth, and M. D. Kazatchkine, AIDS 12 (1998) 1459. H. Albrecht, C. Hoffmann, O. Degen, A. Stoehr, A. Plettenberg, T. Mertenskotter, C. Eggers, and H.J. Stellbrink, AIDS 12 (1998) 1149. [ 10] P. Miralles, J. Berenguer, D. Garcia de Viedma, B. Padilla, J. Cosin, J. C. L6pez-Bernaldo de Quiros, L. Munoz, S. Moreno, and E. Bouza, AIDS 12 (1998) 2467. [ 11] M. S. Dworkin, P.-C. T. Wan, D. L. Hanson, J. L. Jones, and the Adult and Adolescent Spectrum of HIV Disease Project, J. Infect. Dis. 180 (1999) 621. [12]C. Tural, J. Romeu, G. Sirera, D. Andreu, M. Conejero, S. Ruiz, A. Jou, A. Bonjoch, L. Ruiz, A. Arno, and B. Clotet, J. Infect. Dis. 177 (1998) 1080. [ 13] J. C. Macdonald, F. J. Torriani, L. S. Morse, M. P. Karavellas, J. B. Reed, and W. R. Freeman, J. Infect. Dis. 177 (1998) 1182. [14] C. E. O'Sullivan, W. L. Drew, D. J. McMullen, R. Miner, J. Y. Lee, R. A. Kaslow, J. G. Lazar, and M. S. Saag, J. Infect. Dis. 180 (1999) 847. [15] J. Deayton, A. Mocroft, P. Wilson, V. C. Emery, M. A. Johnson, and P. D. Griffiths, AIDS 13 (1999) 1203. [16] J. L. Casado, M. J. Perez-Elias, P. Marti-Belda, A. Antela, M.,Suarez, E. Ciancas, B. Frutos, M. D. Perez, and A. Guerrero, J. Acquir. Immun. Defic. Syndr. Hum. Retrovirol 19 (1998) 130. [17] J. C. Walsh, C. D. Jones, E. A. Barnes, B. G. Gazzard, and S. M. Mitchell, AIDS 12 (1998) 613. [18] R. J. Whitley, M. A. Jacobson, N. D. Friedberg, N. G. Holland, D. A. Jabs, D. T. Dieterich, D. W. Hardy, A. M. Polis, T. A. Deutsch, J. Feinberg, A. S. Spector, S. Walmsley, W. L. Drew, G. W. Powderly, P. D. Griffiths, C. A. Benson, and H. A. Kessler, Arch. Intern. Med. 158 (1998) 957. [19] M. Jouan, and C. Katlama, Int. J. Antimicrob. Agents 13 (1999) 1. [20] Z.-Q. Zhang, T. Schuler, M. Zupancic, S. Wietgrefe, K. A. Staskus, A. K. Reimann, A. T. Reinhart, M. Rogan, W. Cavert, J. C. Miller, S. R. Veazey, D. Notermans, S. Little, S. A. Danner, D. D. Richman, D. Havlir, J. Wong, L. H. Jordan, W. T. Schacker, P. Racz, K. Tener-Racz, L. N. Letvin, S. Wolinsky, and T. A. Haase, Science 286 (1999) 1353. [21] E. De Clercq, J. Med. Chem. 38 (1995) 2491. [22] L. Naesens, R. Snoeck, G. Andrei, J. Balzarini, J. Neyts, and E. De Clercq, Antiviral Chem. Chemother. 8(1997) 1.

278

E. De Clercq / Molecular Targets and Compounds for Anti-HIV Therapy

[23] A. Fridland, Tenofovir, Curr. Opin. Anti-Infect. Invest. Drugs 2 (2000) 295. [24] O. M. Miller, N. A. Margot, K. Hertogs, B. Larder, and V. Miller, 4th International Workshop on Drug Resistance & Treatment Strategies, 12–16 June 2000, Sitges, Spain. Antiviral Ther. 5, Suppl. 3, p. 4, abstract 4. (2000) [25] J.-M. De Muys, H. Gourdeau, N. Nguyen-Ba, D. L. Taylor, P.S. Ahmed, T. Mansour, C. Locas, N. Richard, M. A. Wainberg, and R. F. Rando, Antimicrob. Agents Chemother. 43 (1999) 1835. [26] Z. Gu, M. A. Wainberg, N. Nguyen-Ba, L. L'Heureux, J.-M. De Muys, T. L. Bowlin, and R. F. Rando, Antimicrob. Agents Chemother. 43 (1999) 2376. [27] H. Uchida, E.N. Kodama, K. Yoshimura, Y. Maeda, P. Kosalaraksa, V. Maroun, Y.-L. Qiu, J. Zemlicka, and H. Mitsuya, Antimicrob. Agents Chemother. 43 (1999) 1487. [28] K. Yoshimura, R. Feldman, E. Kodama, M. F. Kavlick, Y.-L. Qiu, J. Zemlicka, and H. Mitsuya, Antimicrob. Agents Chemother. 43 (1999) 2479. [29] M. Daneshtalab, Emtricitabine, Curr. Opin. Anti-Infect. Invest. Drugs 2 (2000) 272. [30] P.A. Furman, J. Jeffrey, L. L. Kiefer, J. Y. Feng, K. S. Anderson, K. Borroto-Esoda, E. Hill, W. C. Copeland, C. K. Chu, J.-P. Sommadossi, I. Liberman, R. F. Schinazi, and G. R. Painter, Antimicrob. Agents Chemother. 45 (2001) 158. [31] P.A. Furman, R. Chin, K. Borroto-Esoda, B. Dent, A. Bartholomeusz, and S. Locarnini, (2000) 4th International Workshop on Drug Resistance & Treatment Strategies, 12–16 June 2000, Sitges, Spain. Antiviral Ther. 5, Suppl. 3, p. 5, abstract 5. [32] C. A. Stoddart, M. E. Moreno, V. D. Linquist-Stepps, C. Bare, M. R. Bogan, A. Gobbi, R. W. Jr. Buckheit, J. Bedard, R. F. Rando, and J. M. McCune, Antimicrob. Agents Chemother. 44 (2000) 783. [33] P. R. Meyer, S. E. Matsuura, A. G. So, and W. A. Scott, Proc. Natl. Acad. Sci. USA 95 (1998) 13471. [34] P. R. Meyer, A. G. So, and W. A. Scott, Int. Antiviral News 7 (1999) 124. [35] M. Gotte, D. Arion, M. A. Parniak, and M. A. Wainberg, J. Virol. 74 (2000) 3579. [36] P. R. Meyer, R. Chopra, E. Pendarvis, S. Matsuura, H. Bazmi, A. G. So, J. W. Mellors, and W. A. Scott, (2000) 4th International Workshop on Drug Resistance & Treatment Strategies, 12–16 June 2000, Sitges, Spain. Antiviral Ther. 5, Suppl. 3, p. 13, abstract 14. [37] C. Perigaud, A.-M. Aubertin, S. Benzaria, H. Pelicano, J.-L. Girardet, G. Maury, G. Gosselin, A. Kirn, and J.-L. Imbach, Biochem. Pharmacol. 48 (1994) 11. [38] C. McGuigan, H.-W. Tsang, D. Cahard, K. Turner, S. Velazquez, A. Salgado, L. Bidois, L. Naesens, E. De Clercq, and J. Balzarini, Antiviral Res. 35 (1997) 195. [39] A.Q. Siddiqui, C. Ballatore, C. McGuigan, E. De Clercq, and J. Balzarini, J. Med. Chem. 42 (1999) 393. [40] A.Q. Siddiqui, C. McGuigan, C. Ballatore, F. Zuccotto, I. H. Gilbert, E. De Clercq, and J. Balzarini, J. Med. Chem. 42 (1999) 4122. [41] J. Balzarini, A. Karlsson, S. Aquaro, C.-F. Perno, D. Cahard, L. Naesens, E. De Clercq, and C. McGuigan, Proc. Natl. Acad. Sci. USA 93 (1996) 7295. [42] D. Saboulard, L. Naesens, D. Cahard, A. Salgado, R. Pathirana, S. Velazquez, C. McGuigan, E. De Clercq, and J. Balzarini, Mol. Pharmacol. 56 (1999) 693. [43] S. Aquaro, O. Wedgwood, C. Yarnold, D. Cahard, R. Pathinara, C. McGuigan, R. Calio, E. De Clercq, J. Balzarini, and C. F. Perno, Antimicrob. Agents Chemother. 44 (2000) 173. [44] C. Meier, M. Lorey, E. De Clercq, and J. Balzarini, Bioorg. Med. Chem. Lett. 7 (1997) 99. [45] C. Meier, T. Knispel, E. De Clercq, and J. Balzarini, Bioorg. Med. Chem. Lett. 7 (1997) 1577. [46] C. Meier, M. Lorey, E. De Clercq, and J. Balzarini, J. Med. Chem. 41 (1998) 1417. [47] C. Meier, T. Knispel, E. De Clercq, and J. Balzarini, J. Med. Chem. 42 (1999) 1604. [48] C. Meier, T. Knispel, V. E. Marquez, M. A. Siddiqui, E. De Clercq, and J. Balzarini, J. Med. Chem. 42(1999) 1615. [49] E. De Clercq, Trends Pharmacol. Sci. 21 (2000) 167. [50] E. De Clercq, Int. Antiviral News 8 (2000) 53.

279

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

Proteomics in Drug Discovery: Potential and Limitations M. Fountoulakis F. Hoffmann-La Roche Ltd., Pharmaceutical Research, Genomics Technologies, Basel, Switzerland Abstract. Proteomics is the large-scale, high throughput analysis of proteins. Its goals are the detection of novel drug targets, diagnostic markers and the investigation of biological events. The bloom that proteomics enjoys today has resulted from corresponding developments in other sciences and technologies, such as computer and software, mass spectrometry, other analytical methods and in particular genomics, which delivered an enormous amount of information from the sequencing of the genome of many organisms. Proteomics is still under development and it possesses the potential of becoming a powerful tool in medicinal research. In this article, selected examples of application of proteomics in drug discovery, the potential of contribution in future discoveries and the limitations of the current technologies are discussed.

What is proteomics Proteomics is a relatively new term and refers to the technology-based science which studies the proteome, i.e. the proteins of an organism or an organ, their post-translational modifications, their interactions and in particular the changes in their levels and their modifications. The changes can be the result of specific diseases or the influence of various external factors, such as toxic agents. Proteomics has as goal the detection of novel drug targets, diagnostic markers and the investigation of biological events [1]. It is comprised of two steps, (i) the separation of protein mixtures, usually by two-dimensional (2-D) electrophoresis and (ii) the identification of the separated proteins by analytical techniques, mainly by mass spectrometry [2]. The process is facilitated by the use of highly sophisticated software for protein identification and advanced image analysis. The 2-D electrophoresis itself comprises two steps (dimensions): (i) separation of the proteins on the basis of differences in their net charge by isoelectric focusing (IEF) and (ii) separation of the focused proteins on the basis of differences in their molecular masses by sodium dodecyl sulfate (SDS) polyacrylamide gels [3–4]. For the performance of IEF, a pH gradient is required. As such, immobilized pH gradient (IPG) strips are usually employed on which the pH gradient has been already formed by acrylamide derivatives, called immobilines. They are weak acids and bases with a buffering capacity, copolymerized with acrylamide and the polyacrylamide gel is then dried on a plastic sheet [5]. The increased application of proteomics today is due to a large

280

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

extent to the introduction of the IPG strips, which are commercially available. The major advantage of using IPG strips is the ability to maintain a high reproducibility while a relatively large protein amount is applied [6]. The increase in reproducibility has allowed the high throughput analysis of proteomes and the application of larger sample quantities has enabled the subsequent identification of a large number of proteins. The use of narrow pH range strips (for example of 1 pH unit) provides a higher resolution, allows the detection of a larger number of protein isoforms and represents an additional advantage of employing IPG strips. Figure 1 shows the 2-D electrophoretic analysis of the soluble proteins of the bacterium Escherichia coli, separated first on a broad pH range IPG strip and then on a SDS polyacrylamide gel of large format.

Figure 1. Two-dimensional electrophoretic analysis of Escherichia coli proteins. One mg of cytosolic proteins were separated on a pH 3–10 nonlinear immobilized pH gradient strip, followed by an 11% sodium dodecyl sulfate polyacrylamide gel (dimensions of the full gel about 18 x 20 cm). The gel was stained with colloidal Coomassie blue.

Following separation by 2-D electrophoresis, proteins are usually identified by mass spectrometry. The proteins are in-gel digested with a protease, such as trypsin. For the proteins that have their genomic sequence in a database, the most efficient identification method presently available is the matrix-assisted laser desorption ionisation time of flight mass spectrometry (MALDI-TOF-MS) [7]. By this method more than 1000 proteins can be analysed daily by one person [8–10]. About 1000-3000 protein spots can be visualised on one 2-D gel, where 1 mg of the total protein amount has been applied and stained with Coomassie blue. Approximately half of the visible spots are available at sufficient quantities to be analysed for identity assignment. From one single gel, up to 1500 spots can be analysed by MALDI-MS and up to 1000 proteins are usually identified which however are the products of about 200-400 different genes. Following this approach, 2-D databases

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

281

for various proteomes have been constructed, usually including a few hundred different gene products [11–18]. The largest 2-D proteome databases, such as of the bacterium Haemophilus influenzae, include approximately 500 mapped proteins [19]. Many of the unidentified proteins are probably not expressed at sufficient amounts to be visualised and can not be detected on account of limitations of the technology (see below). Fig. 2 shows the workflow in proteomics in its classical definition as above, i.e. protein separation by 2-D electrophoresis and identification by mass spectrometry, as well as alternative approaches for a proteome analysis. Moreover, other proteomics technologies also exist, Protein Sample Preparation (cell disruption, protein extraction, fractionation)

Two-dimensional Electrophoresis (isoelectric focusing, SDS-gels) Protein Identification

j

j

Detection of Protein Changes

(construction of databases)

i

j

(comparison of protein levels, modifications)

Coomassie blue-stained gels

"1

In-gel digestion

!^

Electrotransfer to PVDF membrane "" T

N-terminal

Hydrolysis

sequencing

I Amino acid analysis

Mass spectrometry (MALDI-MS, electrospray)

Search in Databases (data storage)

Figure 2. Workflow in proteomics. Proteomics in its classical definition involves protein separation by two-dimensional electrophoresis and protein identification by mass spectrometry as well as detection of changes in the level and modifications of proteins. MALDI-MS, matrix-assisted laser desorption ionization mass spectrometry; PVDF, polyvinylidene difluoride;SDS, sodium dodecyl sulfate.

currently in a developmental stage. Their goal is also the high throughput proteome analysis, based on other separation and identification approaches, for example the use of protein chips. Proteins are often represented by more than one spot in 2-D gels. Examples of heterogeneity of brain, cerebrospinal fluid and liver proteins, which are represented by multiple spots, are shown in Fig. 3. We found that on average, 2-5 spots of brain and 10-20 spots of liver proteins correspond to one gene product [17-18 and unpublished results]. Presently, we do not know either the origin or the biological significance of most cases of observed heterogeneity. It may be the consequence of post-transational modifications, such as phosphorylation, glycosylation or deamidation, which result in the alteration of the pI of the molecule and its focusing position. Heterogeneity may also result from artifacts of the

282

M. Fountouhkis / Proteomics in Drug Discovery: Potential and Limitations

technology, such as carbamylation of the proteins upon prolonged contact of the sample with urea.

Figure 3. Partial 2-D gel images showing the heterogeneity of three proteins derived from the indicated proteomes. CSF,cerebrospinal fluid.

Drug discovery and potential of proteomics Drug dscovery is a lengthy and costly process. It takes approximately 10-15 years until a drug comes on the market, and the average costs for the development and introduction of a drug product can soar to 500 million dollars. Drugs are usually lowmolecular mass, organic compounds, synthesized in chemical laboratories. There are also many proteins, such as interferons, used as drugs, which are mainly prepared by biotechnological approaches. Their production costs are higher than those of the synthetic drugs, but the development and introduction costs are comparable. Many of the drugs currently in use have been discovered in an empirical way and some show strong side effects. The aim of the pharmaceutical science is to develop specific drugs against certain disorders with possibly few side effects. To reach such a goal, the contribution of new sciences is necessary.

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

283

In the development of a drug many basic sciences are involved, i.e. Biology, Biochemistry, Medicine and New Technologies. Proteomics and genomics belong to the New Technologies and represent alternative approaches to search for drug targets. The genomic analysis quickly provides us with data about the mRNA levels (here the corresponding terms are transcriptome and transcriptomics). However, knowledge of the transcription levels does not necessarily inform us about the protein levels or the number of modified protein forms, information, which can only result from the proteome analysis. Life is a dynamic process and not all-possible proteins are expressed at any time point. Moreover, the pattern of protein expression changes depending on various factors, as the stage of development of an organism and an organism's physiological state. It seems that the more complex the genome, the less of the total possible proteins will be expressed at any particular moment. Thus, proteomics supplements the genomics data with information about what gene products are being made, in what cell type, at what amounts and under what conditions and how protein levels and modifications change as a result of specific diseases or external challenges. Various expressions exist for essentially the same science, such as functional proteomics, large-scale proteomics, pharmaceutical proteomics, neuroproteomics etc. All of them have the same goal, the identification of proteins and the functions they exert in the various biological pathways [20-26]. In the drug discovery process, first a drug target needs to be identified, usually a protein that is involved in a biological pathway. Since proteomics is the study of proteins, their modifications, their changes and interactions, it represents the ideal method to search for protein drug targets. Its major advantage is that it enables the simultaneous observation of thousands of unknown proteins. No other method is as efficient at the present time. Thus, the search for unknown drug targets can be enormously facilitated, in particular the search for unknown species associated with other proteins, known to be involved in certain disorders. The identification of disease-related differences is usually a difficult task. Proteomics can relatively easily identify changes in bacterial proteomes and also in cell lines. The analysis of mammalian tissues, such as brain, is more difficult, as many factors are involved, such as differences in age [27], sex [28-29], possibly other diseases, treatment with medicines, as well as technical, disease-unrelated factors, such as post-mortem time [30], improper treatment of the samples etc., which all can affect a clear discrimination between healthy and diseased states of interest. Until now, the research in proteomics mainly resulted in development of the technologies, but the real, systematic application of the advanced proteomics methods in drug discovery has yet to come. Proteomics has been applied as a diagnostic tool in the investigation of infectious diseases, cancer, heart and diseases of the central nervous system, which are summarized in a series of review articles [31-38]. Here we shortly describe our findings of using proteomics technologies in the investigation of neurological disorders. Application of proteomics in neurodegeneration studies: We applied proteomics in the investigation of protein changes in disorders of the central nervous system, such as Alzheimer's disease (AD), Down syndrome (DS) and Pick's disease. Alzheimer's disease is a well-studied dementia affecting mainly elderly people, for which no reliable premortem diagnosis marker exists today. Down syndrome is the most frequent genetic cause of dementia. Although the trisomic state is responsible for the phenotype, the pathomechanisms are not well understood. Almost all subjects with Down syndrome over 40 years show neuropathological and neurochemical abnormalities on post-mortem brain examinations, indistinguishable from those seen in Alzheimer's disease [39-42]. Thus, the results from the DS may be useful in the AD studies. Pick's disease is also a neurodegenerative disorder not well defined presently.

284

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

Figure 4. Two-dimensional database of human brain proteins from the parietal cortex of a control individual. In the map, spots are indicated representing proteins with deranged levels in Alzheimer's disease and Down syndrome. The proteins were identified and the differences were quantified applying proteomics technologies.

We analyzed tissue from seven regions of the brain from the disease groups and from controls with no history of neurological disease. Fig. 4 shows a 2-D map of the human brain proteins, where species are indicated for which deranged levels in AD and DS were found. The proteins with deranged levels, some of which may be potential drug targets, have mainly neurotransmission, guidance, transport and signal transduction functions, are involved in metabolism and detoxification pathways and in conformational changes. We found changed levels for dihydropyrimidinase-related protein (DRP) 2 [43].

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

285

Dihydropyrimidinase-related proteins show homology to chicken collapsin 62 kDa and the rat turned on after division protein. DRPs are expressed in relatively large amounts in the brain, DRP-1, -3 and -4 mainly in the neonatal brain and DPR-2 in both neonatal and adult brain and show a high heterogeneity (Fig. 3) [27, 44]. The functions of DRPs are not clear. It is possible that these molecules are involved in neuronal migration during brain development. When the latter becomes less pronounced in the adult brain, DRP-2, which is almost equally expressed in both adult and neonatal brain, may serve as a guidance cue in repair and remodeling in adult brain. Reduced levels of DRP-2 in the Alzheimer and Down syndrome brain may have consequences in the functions of the protein [43]. We also observed reduced levels of synaptosomal-protein beta-soluble Nethylmaleimide-sensitive factor attachment protein (beta-SNAP) and synaptotagmin 1 in the AD and DS brain. SNAPs participate in vesicular transport between endoplasmic reticulum and Golgi apparatus. Synaptotagmin 1 is a calcium-binding protein, localized at synaptic vesicles and chromaffin granules, which forms complexes with SNAPs. It has probably a regulatory role in membrane interactions during trafficking of synaptic vesicles at the active zone of the synapse [45]. The levels of dynamin 1 were also decreased by about 50% in the disease groups. Dynamin 1 is a phosphoprotein with GTPase activity, abundant in nerve terminals, which is involved in endocytosis and neural transmission. Decreased dymanin levels may count for deficient wiring in the diseased brain (unpublished results). In the AD and DS groups, we found a manifold increase in the levels of the glial fibrillary acidic protein (GFAP), a known marker of neuronal decay and brain damage, which distinguishes astrocytes from other glial cells during development of the central nervous system [46]. We also observed increased levels for the 14-3-3y and 14-3-3e proteins in the disease groups by about 1.5-fold [47]. The 14-3-3 proteins exert complex functions in signal transduction pathways. Deranged levels may reflect impaired signaling and apoptosis in the brain. The levels of synaptosomal associated protein 25 kDa (snap-25) decreased in the two disease groups to approximately 40% [46]. Snap-25 is widely distributed in the brain, is an integral constituent of the synaptic core complex, participates in synaptic vesicle exocytosis and is involved in the formation of presynaptic sites. Decreased snap-25 levels may lead to deranged functions in exocytosis and neurotransmission. Changes were also observed in the levels of histamine-releasing factor [48] glyceraldehyde 3-phosphate dehydrogenase [49], NADH: ubiquinone oxidoreductase [50], superoxide dismutase, carbonyl reductase, alcohol dehydrogenase, voltage-dependent anion channel proteins VDAC1 and VDAC2 [51], heat shock proteins [52-53] and others. Further applications of proteomics: Proteomics has just started contributing to drug development. Various "Human Proteome" projects are currently running and the knowledge that will be accumulated is probably going to revolutionize the protein science in the future. On account of the large heterogeneity and variations, the analysis of a proteome is much more laborious than that of a genome. In the near future, proteomics can supplement the effort of genomics by detecting differences in protein levels and posttranslational modifications and species involved in biological events. Post-translational modifications resulting from various diseases may carry important biological information and proteomics will undertake the task for their identification and understanding of their meaning. Moreover, proteomics is already used and could be to a larger extent employed in the investigation of the toxicity of drug candidates [54-56]. In order to study their efficacy, these substances are subjected to three phases of human clinical trials, which generate enormous costs to the pharmaceutical companies. Early detection of eventual toxicity can save costs and effort and allocate resources in the efficient search of new drug candidates.

286

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

We have applied proteomics to study liver toxicity in mice treated with acetaminophen, the known antipyretic and analgesic agent, which is toxic at high doses. The mechanism of the toxicity is not well understood [57]. We found that the levels of about 30 proteins changed following acetaminophen administration. The changes were in the order of 10-50% of the control value and appeared mainly at a high dose of acetaminophen [16]. Administration of carbon tetrachloride in rats also resulted in significant changes in the liver protein levels (unpublished results). We further studied the effect of administration of kainic acid, a potent neurotoxin and excitatory amino acid, on the levels of brain proteins. The proteomic analysis revealed altered regulation of heat shock proteins, neuronal death, cytoskeletal disruption and mitochondrial derangement by systemic kainic acid administration [58]. Limitations The current proteomics technologies show certain limitations, which need to be overcome in order for the technology to develop its full power and capabilities. The limitations are of two kinds: (i) those related to the composition of the proteome to be analyzed, mainly concerning the protein expression levels and (ii) technical limitations of the technology. Proteomics is therefore simultaneously confronted with two problems, namely to increase the quantity of the low-abundance proteins so that they can be detected and to apply the proper analytical technologies to detect all species of a mixture. For the detection of a protein, three prerequisites must be fulfilled: (i) The protein should be available in a sufficient amount in the protein mixture prior to the 2-D electrophoretic analysis (otherwise the protein needs to be enriched, for example by chromatographic methods), (ii) The protein should be brought into solution with mild detergents and chaotropes, reagents that are compatible with the isoelectric focusing and kept in solution during the first dimensional separation (for proteins difficult to solubilize, strong solubilizing agents should be employed, such as sodium dodecyl sulfate). (iii) The protein should belong to the category of proteins that can be visualized by 2-D electrophoresis, i.e. it should have average pI (usually between pH 4 and 12) and molecular mass (between 10 and 150 kDa) values and should not include strong hydrophobic stretches (otherwise variations of the 2-D electrophoresis or alternative separation procedures should be applied) [59]. Here we shortly describe possibilities to overcome the limitations mentioned. Enrichment of low-abundance proteins: Not all-possible gene products of an organism are expressed at the same level at a certain time point. Some species, such as structural proteins, heat shock proteins or elongation factors, are produced at many thousands of copies per cell, whereas others, such as transcription factors, exert their functions at very low concentrations, maybe at a few copies per cell. In the small sample volume (about 10500 ul) usually used for the proteomic analysis, the majority of the expressed proteins are not present in sufficient quantities to be visualized and identified. Therefore, proteins present in low copy numbers can not be readily detected during the analysis of total proteins. The study of the low copy number gene products may be more interesting in investigating biological events than of their high abundance counterparts. To the present, samples representing the total protein mixture have been usually analyzed and mainly only the abundant, hydrophilic components have been visualized. These proteins could be solubilized with reagents compatible with isoelectric focusing. Such an analysis provides us with a limited image of the proteome and is insufficient for the study of biological events. For the enrichment of low-abundance proteins from crude extracts, biochemical protein enrichment methods are usually employed [60]. These methods separate the original protein mixture into simpler fractions, so that each contains a lower number of total proteins in

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

287

comparison with the starting material. This increases the likelihood of detection of lowabundance proteins manifold. Fig. 5 shows the general scheme followed for the enrichment of low-copy-number bacterial proteins, but similar pathways are valid for eukaryotic systems [59]. The first step is usually the separation of the mixture into organelles, i.e. cytosolic, membranal, mitochondrial etc. protein fractions. In bacteria, cell envelope and membrane proteins represent approximately 15% of the total cellular proteins, so that the probability of detection of a low-abundance membrane protein increases by about 10-fold if a pre-fractionation is employed. In eukaryotic systems, the separation of additional organelles, such as Golgi apparatus, increases the likelihood of detection of the species specific to that organelle. After the separation of the organelles, a further protein enrichment from larger volumes can be achieved by fractionation of the protein mixture, using chromatographic steps or preparative electrophoretic procedures [59-60].

Bacterial Cell Paste (cell disruption, centrifugation)

Pellet (cell envelopes, membranes)

Cytosol (soluble cytoplasmic proteins)

Solubilization (mild detergents and chaotropes)

Chromatography (affinity, hydrophobicity, ion exchange etc.)

Two-dimensional Electrophoresis (broad-, narrow-, basic-range 1PG strips, Tricine gels, 2-detergent system)

Figure 5. General scheme showing the steps that can be followed for the enrichment of low- abundance bacterial proteins. In eukaryotic systems, more "pellet" fractions can be prepared, for example mitochondria, membranes, Golgi apparatus etc.

Enrichment of low-abundance proteins from large volumes can be achieved by selective fractionation, prior or in addition to the chromatographic approaches, using precipitation agents, such as ammonium sulfate, polyethylene glycol or organic solvents. Chromatographic methods separate complex protein mixtures into simpler fractions on the basis of different binding principles and every approach adds a unique resolving power. The proteins are usually separated on the basis of affinity, charge, hydrophobicity, size etc.

288

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

[60]. The choice of the chromatographic method best suited to fulfill the experimental requirements is essential for the success of the experiment. Sequential chromatographic steps are often required. We used four chromatographic steps to enrich low-abundance cytosolic bacterial proteins from larger volumes: affinity chromatography on Heparin [6162], ion exchange chromatography on Polybuffer Exchanger (chromatofocusing) [63], hydrophobic interaction chromatography on a Phenyl column [64] and hydroxyapatite chromatography [65]. Heparin, a highly sulfated glycosaminoglycan, has an affinity for nucleic acidbinding proteins and also functions as a high capacity cation exchanger. Heparin chromatography is relatively easy to perform. We applied this step to enrich low-abundance proteins of the bacterium Haemophilus influenzae. Approximately 150 different proteins were identified in the pools collected from the column and 50 of them were low-abundance species, which had not been found in the 2-D gel of the starting material. Approximately 40% of the heparin-bound proteins were nucleic-acid binding proteins, including ribosomal proteins and t-RNA synthetases. Basic proteins, in particular ribosomal proteins, could be efficiently enriched using this step [66]. Heparin chromatography was also used to enrich cytosolic brain proteins [44]. Chromatofocusing is an ion exchange chromatographic step. The proteins are bound to the gel matrix, Polybuffer Exchanger, and are eluted with a specific buffer, Polybuffer, in the order of their decreasing isoelectric points. Proper choice of the pH of equilibration and elution buffers and of the dimensions of the column can result in an efficient protein concentration and high resolution. This step is usually chosen because the ion exchanger has a high protein binding capacity and can discriminate and enrich proteins with minor differences in their isoelectric points (pI values). We employed the chromatofocusing step to enrich low-abundance proteins of H. influenzae. Most fractions, in particular those eluted with salt, contained mainly acidic proteins. About 125 unique proteins were identified in the pools. Approximately 30 of them were low-abundance gene products not identified before [63]. Fig. 6 shows the enrichment of two enzymes of the microorganism using this step. In hydrophobic interaction chromatography, proteins are separated on the basis of differences in their hydrophobicity. The proteins are adsorbed onto an uncharged matrix carrying hydrophobic groups, in the presence of salts. Elution is achieved by lowering the salt concentration. Due to the various numbers of hydrophobic sites carried by proteins, the hydrophobic interaction chromatography can efficiently fractionate complex protein mixtures. We used hydrophobic interaction chromatography to enrich low-abundance proteins of H. influenzae. About 200 different proteins bound to the column were identified, 30 of which were low-abundance gene products. This step mainly enriched proteins with catalytic activities as well as hypothetical proteins of a wide spectrum of pI values, including basic proteins [64]. Hydroxyapatite chromatography represents a standard protein purification procedure today and the gel binds a wide range of proteins by different mechanisms from the other separation techniques. The hydroxyapatite matrix carries positively charged (calcium) and negatively charged (phosphate) sites. Proteins bind either by nonspecific, electrostatic interactions between their positive charges and the negative charge on the matrix, or by complexation of their carboxyl groups with the calcium sites. Due to the complexity of the protein-hydroxyapatite interactions and in order to achieve a high resolution, various steps for the development of the column are usually applied. We used this step to enrich low-abundance proteins of the bacterium Escherichia coli. Approximately 300 different proteins were identified in the pools collected from the column and about 100 of them were low-abundance proteins. Basic as well as many low

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

289

molecular mass proteins and the proteolytic products of many proteins as well as their fulllength counterparts were enriched by this method [65].

Figure 6. Partial 2-D images, showing the enrichment of two Haemophilus influenzae proteins by chromatofocusing. A, total soluble proteins; B, proteins of a fraction collected from the column.

In addition to the classical chromatographic steps, preparative electrophoresis methods can be applied to enrich proteins from large volumes. They comprise separation of protein mixtures by preparative polyacrylamide gel electrophoresis on the basis of protein size usually in the presence of ionic detergents and preparative isoelectrofocusing on the basis of protein charge in the presence of ampholines. Multicompartment electrolyzers operating with isoelectric membranes may be very useful in the fractionation process [67]. Changes in the levels of low-abundance proteins, resulting from diseases or the effect of external factors, can only be detected after enrichment of the corresponding gene products from large volumes. For example, we employed ion exchange chromatography prior to proteomic analysis to enrich low-copy number cytosolic rat brain proteins from controls and animals treated with kainic acid. We observed changes in the levels of four lowabundance brain proteins. Three out of the four proteins with changed levels had not been detected in the cytosolic fraction before and the detection of the changes was possible only after protein enrichment [68]. The deranged protein levels suggest for the involvement of an apoptotic pathway, the recruitment of the heat shock protein machinery, the generation of an antioxidant response and probably the induction of repair mechanisms in the brain. Chromatographic approaches for protein enrichment have certain limitations. We observed a significant overlap between the H, influenzae proteins enriched by heparin chromatography, chromatofocusing and hydrophobic interaction chromatography. An estimated 40% of the species enriched by hydrophobic interaction chromatography were also enriched by heparin chromatography, and about the same percentage was also enriched by chromatofocusing. About 20% of the proteins were enriched by all three methods. After having applied two or more fractionation methods, the use of additional chromatographic separations may not result in the detection of significant numbers of new proteins not detected or enriched by the previous methods. This points to the limits of the chromatographic enriching techniques and suggests that a large percentage of the proteins not detected are expressed at very low levels. The various chromatographic steps, which we employed, resulted in the enrichment of both low- and high-abundance proteins. This is to be expected, since the high- and low-

290

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

abundance proteins bind to the column matrix on the basis of the same principles. Certain enriched proteins may represent up to 50% of the protein content in a particular fraction, thereby suppressing the low-abundance proteins and hindering their detection in 2-D gels (Fig. 6). High-abundance proteins, for example heat shock proteins and certain housekeeping enzymes in bacteria or albumin in plasma, can be removed prior to 2-D electrophoresis by affinity chromatography using specific antibodies. Protein analysis by 2-D electrophoresis:Following enrichment by any of the methods described above or similar approaches, the visualization of the proteins is usually performed by 2-D electrophoresis [69]. Technical details on the performance of the 2-D electrophoresis can be found in Walsh and Herbert [70] or in Westermeier [71]. Often, it is not sufficient to enrich a protein for a successful detection. Protein detection and identification has certain limitations. The detection in 2-D gels and identification of lowabundance proteins, already successfully enriched from dilute solutions, might be technologically limited because of their charge and size. Fig. 7 shows approximately 400 brain proteins identified from 2-D gels in our laboratory sorted according to their molecular mass and pI values. The majority of the proteins have average molecular mass and pI values. No protein was detected with a pI below 4 and relatively few were detected with a pI value higher than 10. No proteins were found with a molecular mass below 10 kDa or above 160 kDa. In such cases, alternative separation techniques may need to be employed, for example Tricine gels for the detection of the low molecular proteins [72]. In Tricine gels, polypeptides with a mass of about 4 kDa can be detected. Another reason for an unsuccesful detection might be related to the hydrophobicity and low solubility of certain proteins. A protein can only be visualized and analyzed if it can be brought and kept in solution during the whole enrichment process and 2-D electrophoretic analysis. In general, the solubility problem arises at two time points along the performance of the 2-D electrophoresis: (i) during the initial extraction step to solubilize membrane proteins (the cytosolic proteins are already in solution) using agents that are compatible with isoelectric focusing, such as urea and CHAPS, and (ii) during the performance of the first dimensional separation, when hydrophobic proteins could precipitate at their application positions. Poorly soluble proteins can be brought in solution with the use of strong detergents and chaotropes, such as SDS [73]. Hydrophobic proteins with multiple transmembrane regions do not usually enter the immobilized pH gradient strips during the first dimensional separation and thus they can not be detected. It seems that not the hydrophobicity of the entire protein is decisive whether it will enter the IPG strip but the amino acid sequence of the hydrophobic stretches. For example, cytochrome P450 2D6, a hydrophilic protein with one transmembrane region, was not detected in 2-D gels, but at the sample application position (unpublished results). No single solubilizing agent is sufficient for the solubilization, resolution and visualization of all proteins of a proteome. Membrane proteins can also be separated in a discontinuous, two-detergent system [19]. In this system, the first dimensional separation is performed in polyacrylamide gels in the presence of benzyldimethyl-n-hexadecylammonium chloride (16-BAC) and the second dimensional separation in typical SDS-gels.

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

120

100

1 -

80 60

29!

_

a

T

•

40 20

- -

Molecular Mass (kDa) Figure 7. Distribution of about 400 fetal brain proteins identified from 2-D gels by mass spectrometry in our group in relation to their theoretical pI (A) and molecular mass (B) values. The bars indicate the number of proteins found in the pI and molecular mass intervals indicated. No protein with a pI value below 4 and no protein smaller than 10 kDa were found.

Conclusions Proteomics is still in a developmental stage. In spite of the current limitations of a proteomic study, this is the only method for a high throughput qualitative and quantitative analysis of a protein mixture. Moreover, with the sequencing of the genomes of more than 30 organisms to date, many new proteins with no known functions have been deduced. These proteins are referred to as "hypothetical" or as "predicted coding regions", and some of them might be of biological interest. Many of these proteins are produced in low-copy numbers and no literature information is available today for their function or isolation. In our hands, a large number of hypothetical or unknown proteins have been identified following enrichment using chromatographic steps. The analysis of the proteomes of higher organisms is anticipated to be very complex and will probably require a combination of many analytical approaches, comprising multiple enrichment steps and efficient separation and detection methods. The completion of the sequencing of more genomes together with improvements in the analytical techniques will add new search dimensions and will lead to a more widespread application of the technology in drug discovery and the investigation of biological events in general.

292

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

References [1] M. Fountoulakis, Amino Acids (2002) (in press). [2] M. Fountoulakis and H.-W. Lahm J. Chromatogr. 826 (1998) 109. [3] P. H. O'Farrell, J. Biol. Chem. 250 (1975) 4007. [4] M. Fountoulakis, Encyclopedia of Separation Science ll/Electrophoresis, Academic Press, London, 2000, pp. 1356-1363. [5] B. Bjellqvist, K. Ek, P. G. Righetti, E. Gianazza, A. Gorg, R. Westermeier and W. Postel J. Biochem. Biophys. Methods 6 (1982) 317. [6] A. Gorg, C. Obermaier, G. Boguth, A. Harder, B. Scheibe, R. Wildgruber and W. Weiss, Electrophoresis 21 (2000) 1037. [7] W. J. Henzel, T. M. Billed, J. T. Stults, S. C. Wong, C. Grimley and C. Watanabe, Proc. Natl. Acad. Sci. USA 90 (1993) 5011. [8] M. Fountoulakis and H. LangenAnal. Biochem. 250 (1997) 153. [9] P. Berndt, U. Hobohm and H. Langen, Electrophoresis 20 (1999) 3521. [10] H.-W. Lahm and H. Langen, Electrophoresis 21 (2000) 2105. [11] M. Fountoulakis, J.-F. Juranville and P. Berndt, Electrophoresis 18 (1997) 2968. [12] H. Langen, C. Gray, D. Roder, J.F. Juranville, B. Takacs and M. Fountoulakis, Electrophoresis 18 (1997) 1184. [13] G. Lubec, O. Labudova, N. Cairns, P. Berndt, H. Langen and M. Fountoulakis, J. Neural Transm. Suppl. 57 (1999) 21. [14] H. Langen, P. Berndt, D. Roder, N. Cairns, G. Lubec and M. Fountoulakis, Electrophoresis 20 (1999) 907. [15] M. Fountoulakis, E. Schuller, R. Hardmeier, P. Berndt and G. Lubec, Electrophoresis 20 (1999) 3572. [16] M. Fountoulakis, P. Berndt, U. A. Boelsterli, F. Crameri, M. Winter, S. Albertini and L. SuterElectrophoresis 21 (2000) 2148. [17] M. Fountoulakis, J.-F. Juranville, P. Berndt, H. Langen and L. Suter, Electrophoresis 22 (2001) 1747. [18] M. Fountoulakis, P. Berndt, H. Langen and L. Suter, Electrophoresis (2002) (in press). [19] H. Langen, B. Takacs, S. Evers, P. Berndt, H.-W. Lahm, B. Wipf, C. Gray and M. Fountoulakis, Electrophoresis 21 (2000) 411. [20] S. Borman, Chem. Engin. News, issue July 9 (2001) 43. [21] A. Abbott, Nature 402 (1999) 715. [22] S. Borman, Chem. Engin. News, issue July 31 (2001) 31. [23] A. Dove, Nat. Biotechn. 17 (1999) 233. [24] D. Eisenberg, E. M. Marcotte, I. Xenarios and T. Yeates, Nature 405 (2000) 823. [25] A. Pandey and M. Mann, Nature 405 (2000) 837. [26] J. H. Wang and R. M. Hewick, Drug Discovery Today 4 (1999) 129. [27] M. Fountoulakis, R. Hardmeier, E. Schuller and G. Lubec, Electrophoresis 21 (2000) 673. [28] I. Miller, P. Haynes, M. Gemeiner, R. Aebersold, C. Manzoli, M. R. Lovati, M. Vignati, I. Eberini and E. Gianazza, Electrophoresis 19 (1998) 1493. [29] S. Steiner, D. Wahl, M. C. Varela, L. Aicher and P. Prieto, Electrophoresis 16 (1995) 1969. [30] M. Fountoulakis, R. Hardmeier, H. Hoger and G. Lubec, Exp. Neurology 167 (2000) 86. [31] P. R. Jungblut, U. Zimny-Arndt, E. Zeindl-Eberhart, J. Stulik, K. Koupilova, K.-P. Pleissner, A. Otto, E.C. Muller, W. Sokolowska-K6hler, G. Grabher and G. Stoffler, Electrophoresis 20 (1999) 2100. [32] A. A. Alaiya, B. Franzen, G. Auer and S. Linder, Electrophoresis 21 (2000) 1210. [33] R. E. Banks, M. J. Dunn, D. F. Hochstrasser, J.-C. Sanchez, W. Blackstock, D. J. Pappin and P. J. Selby, Lancet 356 (2000) 1749. [34] P. Cash, Electrophoresis 21 (2000) 1187. [35] D. F. Hochstrasser, Proteomics. In M. R. Wilkins, K. L. Williams, R. D. Appel, D. F. Hochstrasser (Eds), Proteome research: New Frontiers in Functional Genomics. Springer, Berlin, 1997, pp. 187-219. [36] V. E. Bichsel, L. A. Liotta and E. F. Petricoin III, The Cancer Journal 7 (2001) 69. [37] S. J. Cordwell, A. S. Nouwens and B. J. Walsh, Proteomics 1 (2001) 461. [38] C. Rohlff, Electrophoresis 21 (2000) 1227. [39] S. E. Antonarakis, Genomics 51 (1998) 1. [40] N. J. Cairns, J. Neural Transm. Suppl. 57 (1999) 61. [41] S. M. de la Monte. J. Neural Transm. Suppl. 57 (1999) 1. [42] B. L. Shapiro. J. Neural Transm. Suppl. 57 (1999) 41. [43] G. Lubec, M. Nonaka. K. Krapfenbauer, M. Gratzer. N. Cairns and M. Fountoulakis. J. Neural Transm. Suppl. 57 (1999) 161. [44] K. Karlsson. N Cairns. G Lubec and M. Fountoulakis. Electrophoresis 20(1999) 2970.

M. Fountoulakis / Proteomics in Drug Discovery: Potential and Limitations

293

[45] B. C. Yoo, N. Cairns, M. Fountoulakis and G. Lubec, Dement. Geriatr. Cogn. Disord. 12 (2001) 219. [46] S. Greber, G. Lubec, N. Cairns and M. Fountoulakis, Electrophoresis 20 (1999) 928. [47] M. Fountoulakis, N. Cairns and G. Lubec, J. Neural Transm. Suppl. 57 (1999) 323. [48] S. H. Kim, N. Cairns, M. Fountoulakis and G. Lubec, Neurosci Lett. 300 (2001) 41. [49] G. Lubec, O. Labudova, N. Cairns and M. FountoulakisNeuroscience Lett. 260 (1999) 141. [50] S. H. Kim, R. Vlkolinsky, N. Cairns, M. Fountoulakis and G. Lubec, Life Science 68 (2001) 2741. [51] B. C. Yoo, M. Fountoulakis, N. Cairns and G. Lubec, Electrophoresis 22 (2001) 172. [52] B. C. Yoo, S. H. Kim, N. Cairns, M. Fountoulakis and G. Lubec, Biochem. Biophys. Res. Communications 280 (2001) 249. [53) B. C. Yoo, R. Vlkolinsky, E. Engidawork, N. Cairns, M. Fountoulakis and G. Lubec, Electrophoresis 22 (2001) 1233. [54] Y. Qiu, L. Z. Benet and A. L. BurlingameJ. Biol. Chem. 273 (1998) 17940. [55] S. Steiner and N. L. Anderson, Toxicol. Lett. 112-113 (2000) 467. [56] S. Steiner and F. A. Witzmann, Electrophoresis 21 (2000) 2099. [57] S. D. Cohen and E. A. Khairalaah, Acetaminophen. In: G. Sipes, C. A. McQueen and A. J. Gandolfi (Eds.), Comprehensive toxicology, Vol. 9: Hepatic and gastrointestinal toxicology. Pergamon, Cambridge, UK, 1997, pp. 329-343. [58] K. Krapfenbauer, M. Berger, G. Lubec and M. Fountoulakis, Electrophoresis 22 (2001) 2086. [59] M. Fountoulakis and B. Takacs, Methods Enzymology (2002) (in press). [60] B. J. Takacs, Protein purification: Theoretical and methodological considerations. In: R. A. Meyers (Ed.), Encyclopedia of Analytical Chemistry. John Wiley and Sons Ltd., Chichester, 2001, pp. 5955-5970. [61] M. Fountoulakis, H. Langen, S. Evers, C. Gray C and B. Takacs, Electrophoresis 18 (1997) 1193. [62] M. Fountoulakis and B. Takacs, Protein Expression Purification 14 (1998) 113. [63] M. Fountoulakis, H. Langen, C. Gray and B. Takacs, J. Chromatogr. 806 (1998) 279. [64] M. Fountoulakis, M.-F. Takacs and B. Takacs, J. Chromatogr. 833 (1999) 157. [65] M. Fountoulakis, M.-F. Takacs, P. Berndt, H. Langen and B. Takacs, Electrophoresis 20 (1999) 2181. [66] M. Fountoulakis, B. Takacs and H. Langen, Electrophoresis 19 (1998) 761. [67] P. G. Righetti, A. Castagna and B. Herbert, Anal. Chem. 73 (2001) 321. [68] K. Krapfenbauer, M. Berger, G. Lubec and M. Fountoulakis, Eur. J. Biochem. 268 (2001) 3532. [69] H. Langen, D. Roder, J.-F. Juranville and M. Fountoulakis, Electrophoresis 18 (1997) 2085. [70] B. J. Walsh and B. Herbert, (1998) http://rbams3115/Pages/2DPAGE/ABRFNews_2dpage.html [71] R. Westermeier, Electrophoresis in Practice. VCH Verlagsgesellschaft, Weinheim, 1993. [72] M. Fountoulakis, J.-F. Juranville, D. Roder, S. Evers, P. Berndt and H. LangenElectrophoresis 19 (1998) 1819. [73] M. Fountoulakis and B. Takacs, Electrophoresis 22 (2001) 1593.

294

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

Evaluation of Tumor Capabilities for Recurrence in Patients with Larynx and Pharynx Malignancies on the Basis of DNA Criterion S. Andreychenko O.Kolomiychenko Institute of Otolaryngology of Academy of Medical Sciences Kiev, 03057, Ukraine

Abstract. A special criterion adequate for prediction of tumor behavior as well as for an evaluation of post-operation recovery duration in patients with larynx and pharynx malignancies has been developed. The criterion in question was called a coefficient of tumor aggressiveness - A. It presents itself a proportion showing a relation of DNA concentration in tumor to that in nearby normal on convention tissue. A depends on tumor localization, varying in the range 1,15 - 2,02 for larynx and 1,55 - 2,62 for pharynx, respectively. Values of A 1,8 -1,9 are the most beneficial for post-operation recovery in patients with larynx malignancies, while for pharynx strong rehabilitation was observed in the diapason of 2,0 - 2,2. Beyond these borders a chance for tumor recurrence significantly increases. A correlates with the appearance of polyploid cell clones persistent to oncogenic transformation.

Introduction Nowadays cancer presents itself a most serious threat for a mankind because of its wide spreading in different races and populations all over the world. According to statistical data in Ukraine appear 2 new oncology patients among each 1000 of inhabitants, while in USA more than 40 000 new cases only of head & neck cancer are expected to be revealed in the coming year [1]. However, in Ukraine and several other countries of Eastern Europe the

5. Andreychenko / Evaluation of Tumor Capabilities for Recurrence

295

situation is being dramatically deteriorated by the consequences of Chernobyl catastrophe happened on April 25, 1986 on the Chernobyl atomic power station. This incident damaged a lot of people because of their continuous exposure to low-dose ionizing irradiation in the period of disaster and afterwards. As a result of the accident, cancer begun to grow in an unpredictable manner, causing either rapid metastases formation or sudden impetuous progressing to the ultimate stages of the disease. Apart from this, an obvious downfall in lower age limit for malignant transformation has been observed. In the connection to all aforesaid a search of an adequate criterion for tumor behavior evaluation appears to be of urgent need. This could be indicative as to the tumor capabilities in further recurrence after surgical intervention. Materials and Methods The examinations were carried out in 2 groups of patients with cancer of larynx and pharynx, respectively. Each group contained 30 persons. The malignant nature of tumors was confirmed on the basis of histopathological analyses executed on biopsies stained with hematoxylin and eosin after they had been primarily fixed in Holland's fluid, embedded in paraffin wax and sectioned at 5 um [2]. All the patients underwent surgery for extirpating malignancies. Under the operations samples of tumors as well as nearby normal on convention tissue were taken off to be processed both for total DNA concentration quantitative determination and nuclear DNA contents cytophotometry. The oncology patients in the age of 40 to 65 years were in stage 11/111 of the disease, subsequently followed the same standard scheme of treatment. Determination of total DNA concentration in tissue biopsies:The analyses were performed according to specially developed protocol. Thus, the minced biopsies were previously fixed in cold methanol (-18 C ) and stored in it for at least 24 h. Afterwards the samples were washed 3 times in 1 N perchloric acid (HC1O4) to extract free nucleotides and centrifuged at 1500 g , the supematants being discarded. The saved sediments were treated with 0,5 N HC1O4 twice at 70 °C for 20 min each time and then centrifuged at 1500 g, the supematants being united together and used in Dishe reaction with diphenylamine in Burton's modification [3]. According to the procedure to a sample aliquot 2 volumes of reagent consisting of 1,5 % diphenylamine, 1,5 % sulfuric acid (H2SO4) and 0,008% acetaldehyde in glacial acetic acid (CH3COOH) were added. Color intensity in reaction mixture was developing for 18 h at 30° C and then measured at 598 nm on spectrophotometer "SF-46" (Russia). In the samples DNA content was determined by the aid of calibration curve built beforehand for known quantities of similar hydrolyzed calf thymus DNA (Sigma). DNA cytophotometry:Tissue samples immediately after extirpation were put into physiological solution of 0,9% sodium chloride (NaCl) and then used for smear preparations. Following fixation in Carney's solution - ethanol: chloroform: acetic acid (6:3:1) - for at least 3 h the smears were passed through descending ethanol series (50 %, 30 %, 15 %) up to water, rinsed with cold 0,1 N hydrochloric acid (HC1) and then hydrolyzed in 1 N HC1 for 9 min at 60 ° C. Staining with basic fuchsin dye was performed during 2 h in the dark at room temperature [4]. For removing free dye and color fastening the preparations were placed in three changes of SO2 - water for 10 min in each, washed with water and passed through ascendant ethanol series (15 %, 30 %, 50 %, 70 %, 90 %, 96 %, absolute), ethanol: xylene mixture and xylene. After embedding in DPX mountant

296

S. Andreychenko / Evaluation of Tumor Capabilities for Recurrence

(Fluka) the preparations were analyzed in two rays automatic cytophotometer "MFTX 2M" (Russia) linked with computer IBM-486 at the wavelength of 546 nm. The specially developed program made us possible to record exclusively total nuclear DNA specific absorption. For control measurements really diploid mucous cells and haploid spermatozoa from volunteers were analyzed to establish properly the correlation coefficient between DNA absorption and ploidy level. The aforesaid technique permitted to specify cell distribution in tissue populations according to the value of their ploidy. Estimation of tumor aggressiveness coefficient - A: It is calculated on the basis of proportion showing the relation of DNA concentration in tumor (cDNAT) to that in nearby normal on convention epithelial tissue (cDNAN ): A = cDNAT/ cDNAN Statistical analysis: For statistical evaluation of data Student's t-test was used at the level of significance not less than 0,95 when the coincidence between control and experimental results is observed only at a value of p < 0,05 [5]. Results Analysis of DNA concentration in tumors in patients with larynx malignancies revealed the diapason of its variability ranging from 1,15 to 1,89 ug/mg (p>0,99) while for pharynx this parameter has shown a tendency for some increase reaching upper limit of 2,72 ug/mg. In the nearby normal on convention tissues of larynx and pharynx the values for DNA contents were 0,95 ± 0,16 ug/mg and 1,04 ± 0,17 ug/mg, respectively (Table 1). In connection to these data a tumor aggressiveness coefficient A for larynx malignancies ranged in the interval 1,15-1,91 and was some few different from that for pharynx, a latter one showing a broader range of variability and a tendency for manifestation at a higher level of significance - 2,04 ± 0,59. Table 1. Analysis of DNA concentration in malignant tissue in comparison to unaffected one in patients with tumors of larynx and pharynx*

Localization

Larynx Pharynx

DNA concentration variability range in tissue, ug/mg Malignant (cDNAT) 1.15–1,89 1,31–2,72

Normal on convention (cDNAN) 0,79–1,11 0,87–1,21

Tumor aggressiveness coefficient A variability range 1,15-2,02 1,55–2,62

* Full variability range of experimental data for each group of patients is presented Each patient was examined as to the duration of post-operation recovery. According to the data presented in Fig.1 evident correlation exists between A and patient's resistance to tumor recurrence. Thus, several patients after larynx surgery manifested explicit stability in recovery for more than 4 years long with A ranging from 1,8 to 1,9. Meanwhile for pharynx the same result was achieved at values of A equal to 2,15±0,20.

S. Andreychenko / Evaluation of Tumor Capabilities for Recurrence

297

For larynx rapid abatment was observed of recovery duration at A values below 1,5 with further little rise at 1,15. The plotted two curves (see fig.1) both for larynx and pharynx show similarities having obvious peak with two descending branches beyond it, although larynx manifests additional ascendant branch at low values. In both cases A value growth causes the appearance of recovery beneficial interval then replaced by pronounced downfall in recovery duration.

6,0 5,5 5,0 4,5 4,0 3,5 3,0 2,5 2,0 1,5 1,0 0,5 1,0

1,2

1,4

1,6

1,8

2,0

2,2

2,4

2,6

A (units) Figure 1. Dependence of recovery duration on the DNA ratio coefficient A in oncology patients after tumor removal.

The cytophotometric analysis of nuclear DNA contents revealed that cell distributions in larynx and pharynx normal on convention tissues were in the range of 2n to 4n showing usual realization of cell cycle in overall cell population. Meanwhile under strong recovery conditions the subpopulations of autopolyploid cells, i.e. tetraploids and octaploids were discovered. Apart from this, the accumulation of aneuploid cell clones with degraded genome (<2n) also was recorded. This reflects the depression of immune system as well as the reduction of macrophage activities. These data agree well with the present in literature information concerning head and neck cancer [6]. The patients having a tendency for short-term recovery were characterized by the appearance of intermediate ploidy cell clones, thus manifesting an inclination for selective gene amplification. In the case of pharynx malignancies partial DNA increase took place both in diploid and tetraploid cells while for pharynx malignancies selective DNA replication only at the diploid level was recorded.

298

S. Andreychenko / Evaluation of Tumor Capabilities for Recurrence

Discussion - Conclusion The obtained data show the possibility to evaluate tumor aggressiveness in relation to postoperation recurrence on the basis of DNA criterion that presents itself a ratio equal to proportion of DNA concentration in tumor to that in nearby normal on convention epithelial tissue. A is easy to determine in patients undergone surgical intervention by extirpating samples of corresponding tissues under the operation for their next processing in DNA quantitative analysis. A correlates well with the duration of post-operation recovery. However it depends on the cancer primary localization. At the range of A values most beneficial for long recovery A reflects the appearance in target tissue special cell clones undergone genome polyploidization while partial DNA amplification used to be incident to short-term recovery state. In the former case whole DNA repeated duplication might lead to the increase of genome stability and safety, on the contrary, in the latter one the sequence of events caused cell genome disbalance and casual proliferation of deleterious oncogenes. In this connection the ascendant short branch on larynx curve (see fig. 1) could be explained by diminution of oncogene dose below the crucial limit at low A. Launching of extra DNA replication in target tissue might be due to lowering of extracellular pH, activation of H+/Na+ antiport and concomitant growth in Ca2+ influx into cells, the latter one being a potent promoter of cell proliferation [7,8,9]. Moreover, it is known that DNA synthesis counteracts pH lowering and provides for normalization of intracellular pH as well as immobilization of superfluous Ca2+ in newly synthesized chromatin motifs [10,11]. Thus, when driving autopolyploidization program cells can acquire stability to oncogenic transformation and persist in unfavorable environment. This tendency can be registered by coefficient A measurement. The application of A criterion in clinic let us avoid a number of complications in post-operation period and outline properly the strategy for rehabilitation of oncology patients with aforementioned cancer localizations.

References [1] Buntze J., Basecke S., Weinaug R., Glatzel M., Kuttner K., Frohlich D., (2000). Liposomal doxorubicin in the palliative treatment of head and neck cancer. In: Jahnke K. and Fischer M. (ed.), "4th European Congress of Oto-Rhino-Laryngology Head and Neck Surgery", Berlin (Germany), May 13–18, 2000. Monduzzi Editore, Bologna, Italy, vol.2, p. 1377–1381. [2] Silverberg R.G., DeLellis R.A., Frable W.J. "Principles and Practice of Surgical Pathology and Cytopathology". 3rd edn. New York, Churchill-Livingstone, (1997). [3] Burton K.A., Biochem. J. 62 (1956) 315. [4] Darlington S.D., La Cour A.F. "Handling of Chromosomes". 6* edn. Atomizdat, Moscow, Russia, (1980). [5] Rokitsky P.F. "Biological Statistics". Higher School, Minsk, Belorussia, (1973). [6] Hadden J.W., Int. J. Immunopharmacol. 19 (1997) 629. [7] Rogers M.S., Strehler E.E., (1996). In: Celio R.C. (ed.), " Guidebook to the Calcium - Binding Proteins",A Sambrook & Tooze Publication AT Oxford University Press, New York, U.S.A., p. 34 - 40. [8] Ruas M., Peters G., Biocliim. Biophys. Acta 1378 (1998) 115. [9] Lee A.S., Tannock I.F, Cancer Research. 58 (1998) 1901. [10] L'Allemain P., Paris S., Poussegur J. J. Biol. Chem. 260 (1985) 4877. [11] Lewin B. "Genes", Oxford Univ. Press, New York, U.S.A. (1997)

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

Significance of G-protein-coupled receptor polymorphisms : The case of a2-adrenergic receptor A.S. Manolis1, A. Lymperopoulos, E.A. Bouga, M. Scheinin2 and CS. Flordellis Department of Pharmacology and1 Department of Cardiology, School of Medicine, University of Patras, 261 10 Rion, Patras Greece and 2Department of Pharmacology and Clinical Pharmacology, University of Turku, Finland Abstract. Until recently it was assumed that G-protein coupled receptors ( GPCRs ) were genetically invariant in the human population. This assumption has been challenged by the growing evidence demonstrating the existence of frequently occurring genetic variants of GPCRs. Using a2-adrenergic receptor subtypes as example, this work summarizes briefly current knowledge on the issue and describes the clinical and pharmacological significance of these polymorphisms in cardiovascular disease.

Polymorphism is defined as the occurrence in the same locality of two or more discontinuous genetically determined phenotypes in such proportions that the rarest of them cannot be maintained merely by recurrent mutations [1]. Simply stated, genetic polymorphisms are common genetic variants (allele frequency > 2%). In a medical context the advantage of dealing with genetic polymorphisms is that associations between genotype and phenotype can be expressed in terms of frequency. Their importance consists in the fact that such variations in receptors , drug metabolizing enzymes, transporters and other potential targets of drug interventions , in many cases entail functional consequences, which may provide the basis for observed inter-individual variability in clinical phenotypes and pharmacological responses [2]. The G-protein-coupled receptors ( GPCRs ) represent a very suitable system for investigating the potential contribution of receptor polymorphism to propensity for disease and response to drug therapy. These receptors are widely expressed and mediate the action of a variety of hormones, growth factors, neurotransmitters , peptide and local mediators involved in numerous cellular processes and consequently physiological functions, such as metabolism, neurotransmission and cardiovascular regulation among others. Their clinical importance is underlined by the fact that almost one third of all subscription medications are GPCR agonists or antagonists [3].

300

A.S. Manolis et al. / Significance of G-protein-coupled Receptor Polymorphisms

GPCRs are nearly devoid of intrinsic activity. Upon agonist stimulation they only catalyze the release from the Ga subunit of the tightly bound GDP , driving the heterotrimeric Gprotein into the process of G-protein GTPase cycle . The generated Ga and G (Jy subunits regulate downstream effectors, such as adenylate cyclase [4], PI3 Kinase [5], and MAPK ([6] among others. In addition , the Cfty complex activates GRKs, Ser/Thr kinases, which phosphorylate the agonist-bound receptor, thus initiating the process of receptor desensitization. The way genetic polymorphisms may modify the receptor's biochemical phenotype, becomes evident when considering the functional domains of GPCRs. Seven transmembrane receptors in general and alpha2-adrenergic receptors (ct2-ARs ) in particular have an extracellular aminoterminus, usually glycosylated , an intracellular carboxyl terminus and a peptide backbone that folds back and forth through the plasma membrane seven times forming the ligand-binding pocket (in the transmembrane domains ) and loops, notably the third cytoplasmatic loop, which has domains both for coupling to the cognate G-protein and for phosphorylation by G protein-coupled Receptor Kinases ( GRKs ) and subsequent desensitization [7]. In the case of <x2-ARs there is subtype-specific agonistpromoted phosphorylation and desensitization : a2A and tt2B undergo agonist-promoted phosphorylation and desensitization,whereas the o^c receptor does not, despite the fact that the aac has within the 3rd intracellular loop the sequence DESS, which is similar to the a2A sequence EESSS, shown to be an excellent substrate for GRK-induced phosphorylation [8]. Mutations of G-protein coupled receptors: They can cause a number of relatively rare diseases. In the majority of mutations that modify receptor function, the variant allele imparts constitutive activation (agonist -independent increased function) or decreased agonist-promoted function. Examples of the former are the familial male precocious puberty ( FMPP ), that results from a mutation in the luteinizing hormone receptor [9] and the hyperfunctioning thyroid adenomas due to somatic or germline mutation of the Gs-coupled TSH receptor [10]. Polymorphisms of GPCRs on the other can be minor risk factors for complex multifactorial diseases ( e.g cardiovascular disease, diabetes mellitus etc. ), modifiers of disease characteristics ( onset, progression or outcome ) or alter responses to drug therapies. Many of them have been described, but in the present context we mention representatively only some of them. The Gly16 form of the pVadrenergic receptor becomes more easily down regulated and desensitized after exposure to the agonist, compared with the wild-type Arg16 form. This genetic variant may predispose to some forms of asthma and hypertension in AfricanCaribbeans [11]. On the contrary the Glu27form of the receptor confers resistance to down regulation compared with the wild-type Gln27 form [12]. Consequently subjects with this variant receptor are predicted to have less reactive airways. Finally, the Deu164 pY adrenergic receptor polymorphism, displays a small reduction in agonist binding affinity and defective agonist-promoted coupling to Gs and has been found to affect adversely the outcome in congestive heart failure [13]. A mutation has been identified at the beginning of the first intracellular loop of (J 3adrenergic receptor resulting in a Trp64Arg substitution. The 0 3-AR is expressed mainly in brown fat adipocytes and impairement of its function could reduce adipocyte sensitivity to lipolysis. The allele frequency of the above variant was found similar in normal subjects and in patients with morbid obesity ( BMI > 40 kg*/m2 ) ( 0.1 vs. 0.08 respectively ), implying that the allele is not the cause of obesity. The presence of the allele may

A.S. Manolis et al. / Significance of G -protein-coupled Receptor Polymorphisms

301

predispose patients to abdominal obesity, decreased resting metabolic rate, insulin resistance and earlier onset of NIDDM ([14]. In recent studies several polymorphisms have been identified in the coding or the 5' and the 3' flanking regions of the ATl receptor for AngII [15]. The polymorphism A1166C ( an Adenine to Cytosine base substitution at position 1166, located in the 3' untranslated region of the ATl receptor gene), was recently found to be associated with myocardial infarction and essential hypertension. The C allele of this polymorphism is an independent determinant of aortic stiffness [16] and also shows a synergism with the D allele of the angiotensin converting enzyme insertion/deletion polymorphism on risk of myocardial infarction. A role of the renin-angiotensin system in the aetiology of coronary atherosclerosis has been indicated by the finding that the DD genotype of an insertion/deletion polymorphism (I/D) in the gene for the Angiotensin Converting Enzyme ( ACE ) is strongly associated significantly with the levels of circulating enzyme and is more frequent in patients with myocardial infarction [17]. The genetic analysis of aa-ARs is limited. Initial studies on small populations have indicated that all cta-AR genes are polymorphic, and because the mutations occur in the 3rd cytoplasmatic loop the variant alleles have altered biochemical properties in relation to the wild-type genes. The Del322-325 form of the human aac-adrenergic receptor results in impaired coupling'to multiple effectors [18]. In other cases however, the polymorphism confers enhanced agonist-promoted G-protein coupling. Thus, the Asn to Lys at position 251 in the third cytoplasmatic loop of the human aaA-AR imparts enhanced G-protein coupling [18]. Finally, a common genetic variant of ct2B- adrenergic receptor was identified , which consists in a deletion of three glutamic acid residues in the third cytoplasmatic loop [19]. This deletion results in reduced ( ~60 % ) agonist-promoted phosphorylation of the variant receptor relative to the wild-type receptor and in loss of short-term agonistpromoted desensitization and is associated with reduced metabolic rate in obese subjects of a Finnish population [19]. The clinical and pharmacological consequences of the 012-AR genetic variation can be predicted, if one considers the physiological function of the respective receptor subtype and the kind of biochemical modification resulting from the mutation. It has been anticipated for example, that the above mentioned (X2A-AR polymorphism would result in enhanced inhibition of depolarization-induced norepinephrine release and therefore predispose to hypotension and bradycardia and display a milder course of disease, in case of development of hypertension. For the same reason the polymorphism would result in potential protection of the myocardium from catecholamines in congestive heart failure. Similarly, the antihypertensive efficacy of (X2-AR agonists i.e. clonidine would be enhanced, whereas the central nervous system side-effects of such pharmacological agents would be pronounced limiting their therapeutic utility [18]. Role of alpha2-ARs in cardiovascular regulation: Three distinct and well characterized tt2AR subtypes ( v,2\, 0.23, «2C) have been described. They belong to the family of G-protein coupled receptors and they are linked to the inhibitory Gi/o signalling system modifying the activity of several effectors including adelylate cyclase, ion channels , MAPK and cPLA2 [20] among others. They differ in their structure, chromosomal localization tissue distribution and pharmacological properties and participate in the regulation of cardiovascular system [7]. Despite progress in dissecting the function of alpha2-ARs as a group, knowledge on the physiological roles of the individual a2-AR subtypes was very limited until recently due to the lack of subtype-selective antagonists. However, In the last five years, knock-out (KO) mouse lines generated by the employment of homologous recombination methodologies [21, 22] have allowed determination of the physiologic roles of the (X2-AR

302

A.S. Manolis et al. / Significance of G-protein-coupled Receptor Polymorphisms

subtypes in mice. Studies on such mice have indicated that central a2A-ARs mediate the long-known hypotensive effects of 02-AR agonists, such as clonidine [7], whereas 02adrenergic vasoconstriction is mainly mediated by postsynaptic or extrajunctional aie-ARs on vascular smooth muscle cells. Both <X2A- and c^c-AR mediate presynaptic inhibition of neurotransmitter release [23]. tt2-ARs exert important functions in renal physiology. It has been shown that ct2B-ARs participate in the development of salt-induced experimental hypertension and mice lacking the full complement of a2B receptor fail to develop effectively salt-induced hypertension [24]. Studying genetic polymorphisms of d2-ARs: Cardiovascular disease is a multifactorial disease that results apparently from the synergism of several genes interacting in a complex way with environmental factors. Recent clinical genetic studies have shown significant associations between a 2B-AR gene variant and basal metabolic rate and the risk for acute coronary events [19, 25]. We have been interested in the a2B-adrenergic receptor polymorphism and have started to study its association with cardiovascular disease and characterize the biochemical properties of the receptor in order to find : the clinical association of this alpha2 receptor polymorphism with coronary heart disease phenotypes in populations of Greek patients and the mechanisms by which genetic polymorphisms contribute to the development of cardiovascular diseases Determination of genotype: The study population used for genotype determination and association analysis consists of patients hospitalized for ischemic coronary events (unstable angina or AMI ) evaluated and classified in the Cardiology Department of the University Hospital, Patras-Greece. The a 2B-AR insertion/deletion genotypes are determined by electrophoretic separation of PCR-amplified DNA fragments. The size difference of the long and short alleles of this a 2B-AR polymorphism makes it possible to reliably and reproducibly distinguish them by their different electrophoretic migration on high density agarose gels [25]. To confirm the reliability of agarose gel electrophoresis in determining the genotype some of the DNA samples in our study were also analyzed by PAGE. A representative example of such analysis is shown in figure 1. So far we have analyzed 70 patients and in some cases their normal relatives. The frequencies for the I and D alleles in this population appears to be 0.8 and 0.2 respectively.

Figure 1. Determination of genotype by electrophoretic separation of PCR-amplified DNA fragments.

DNA for genotyping was axtracted from a small quantity of peripheral blood according to standard methods and amplified with the following pair of primers : 5'-CAA GCT GAG GCC GGA GAC ACT G-3' and 5'- AGG GTG TTT GTG GGG CAT CTC CT-3' yielding a DNA product of 112 bp for the long allele ( I ) and a DNA product of 103 bp for the short, deleted allele ( D ). The PCR reaction contained ~100 ng genomic DNA, 0.8 mM dNTPs, 0.3 nM of each primer and 0.5 units of Taq polymerase in a total volume of 20 ul. The conditions included an initial denaturation step at 94 oC for 2 minutes and 35 cycles of amplification as follows : 96 oC ( 40 sec ), 69 oC ( 30 sec ) and 72 oC ( 30 sec ) followed

A.S. Manolis et al. / Significance of G-protein-coupled Receptor Polymorphisms

303

by a final extension step at 72 oC for 6 minutes. After amplification the PCR products were separated electrophoretically on a 4% agarose gel with ethidium bromide staining. The long and short alleles were distinguished on agarose gels ( left panel ) based on their different electrophoretic migration rate as a result of their 9 bp size difference. The analytical power of this agarose gel system was verified by parallel analysis of PCR products in polyacrylamide gels ( right panel). Desensitization properties of polymorphic a.2B-adrenergic receptor Previously, it has been shown that an acidic stretch of amino acids in the 3rd cytoplasmatic loop of a 2B-adrenergic receptor regulates phosphorylation of the receptor by GRKs and experimentally induced deletion mutation in this region, is accompanied by reduced agonist-promoted phoshorylation and subsequent desensitization of the receptor [26]. In order to analyze the consequences of this deletion variant on short-term agonist-promoted receptor phosphorylation, we use Chinese Hamster Ovary ( CHO ) cells stably expressing the wild-type and the deleted form of a2B-AR. We have standardized the conditions for monitoring the state of receptor phosphorylation using a specific anti-human aas-AR monoclonal antibody (generously provided from the Turku laboratory ) to immunoprecipitate the receptor, after whole-cell phosphorylation. A representative result is shown in figure 2.

Figure 2. Detection of a2B-AR expressed in CHO cells. ( 1 ) Confluent cultures of CHO cells expessing the wild-type a2B-AR were solublized in RIPA buffer and the receptor was immunoprecipitated with a specific monoclonal anti-human a2u-AR , followed by Western blotting and visualization of the receptor using the same antibody and ECL detection. ( 2 ) Confluent cultures of CHO cells expessing the wild-type O2B-AR were incubated with [32P] orthophosphate ( 400uCi/ml) for 2h at 37oC, washed with ice-cold PBS 5 times, solublized in RIPA buffer and the receptor was immunoprecipitated with a specific monoclonal anti-human a2e-AR, followed by SDS-PAGE and autoradiography.

These genetic and biochemical studies are currently in progress. The rapid progress in functional genomics is expected to have a profound impact in the medical-pharmaceutical system. Detailed analysis of the genetic mechanisms of disease will clarify our understanding of multigenic diseases, will facilitate the assessment of individual risk for disease and will aid in the design of novel approaches to disease prevention and drug therapies. Pharmacogenomics promises the translation of functional genomics into rational drug therapy. Genetic tests for more rapid diagnosis will help classifying heterogenous populations of patient into genetically distinct groups. Genotyping before prescription of certain drugs will enable to predict how a patient will respond to a given drug. This will

304

allow on the one hand selection of appropriate therapies, from which patients are going to be benefited the most and on the other avoidance of the prescription of potentially toxic drugs, that are likely to yield dangerous side-effects. On the other side, the utilization of flexible and time-saving assays ( e.g. gene expression monitoring ) for predicting pharmacological and toxicological outcomes, will decrease the cost of clinical drug testing and may allow companies to develop drugs for small markets such as limited, genetically defined subpopulations and patients in poor countries. Furthermore, the new knowledge will facilitate the drug companies to sharpen the properties of their drugs, so that drug intervention is expected to become more individualized. These starting shifts in the existing paradigm of health care and drug development show that the medical-pharmaceutical system will be further challenged in the coming years. References [1] Ford EB. Ecological Genetics, 3rd edition, London Chapman & Hall. 1971. [2] Insel PA., Seminars in medicine of the Beth Israel Hospital, Boston. New Engl. J.Med. 334 (19%) 580. [3] Kallal L, Benovic JL, Trends Pharm.Sci. 21 (2000) 175. [4] Schaak S, Cayla C, Lymperopoulos A, Flordellis C, Cussac D, Denis C, Paris H., Mol. Pharmacology 58 (2000) 821. [5] Hawes BE, Luttrell LM, van Biesen T, Lefkowitz RJ., J.BioiChem. 271 (19%) 12133. [6] Flordellis CS, Berguerand M, Gouache P, Barbu V, Gavras H, Handy DE, Bereziat G, Masliah J., J.Biol.Chem. (1995) 3491. [7] MacDonald E, Kobilka BK, Scheinin M, Trends Pharmacol.Sci. 18 (1997) 211. [8] Jewell-Motz EA, Small KM, Theiss CT, Liggett SB., J.Biol. Chem. 275 (2000), 28989. [9] Shenker A, Laue L, Kosugi S, Merendino JJ Jr, Minegishi T, Cutler GB Jr., Nature 365 (1993 ) 652. [10] Parma J, Duprez L, Van Sande J, Cochaux P, Gervy C, Mockel J, Dumont J, Vassart G. , Nature 365, 649. [11] Reishaus et al. (1993 ) Am.J.Respir.Cell.Mol.Biol. 8 (1993) 334. [12] Green SA, Turki J, Bejarano P, Hall IP, Liggett SB., Am.J.Respir.Cell.Mol.Biol. 13 (1995) 25. [13] Liggett SB, Wagoner LE, Craft LL, Hornung RW, Hoit BD, Mclntosh TC, Walsh RA., J.Clin.Investigation 102 (1998) 1534. [14] Walston J, Silver K, Bogardus C, Knowler WC, Celi FS, Austin S, Manning B, Strosberg AD, Stem MP, Raben N, et al., New Engl. J Med. 333 (1995) 343. [15] Bonnardeaux A, Davies E, Jeunemaitre X, Fery I, Charm A, Clauser E, Tiret L, Cambien F, Corvol P, Soubrier F., Hypertension 24 (1994) 63. [16] Benetos A, Gautier S, Ricard S, Topouchian J, Asmar R, Poirier O, Larosa E, Guize L, Safar M, Soubrier F, Cambien F., Circulation (1996) 698. [17] Cambien F, Poirier O, Lecerf L, Evans A, Cambou JP, Arveiler D, Luc G, Bard JM, Bara L, Ricard S, et al. Nature 359 (1992) 641. [18] Small KM, Forbes SL, Rahman FF, Bridges KM, Liggett SB., J.Biol. Chem. (2000) 23059. [19] Heinonen P, Koulu M, Pesonen U, Karvonen MK, Rissanen A, Laakso M, Valve R, Uusitupa M, Scheinin M., J. Clin.Endocrinology & Metabolism, 84 (1999) 2429. [20] Audubert F, Klapisz E, Berguerand M, Gouache P, Jouniaux AM, Bereziat G, Masliah J., Bioch.Bioph.Acta 1437 (1999) 265. [21] Link RE, Desai K, Hein L, Stevens ME, Chruscinski A, Bernstein D, Barsh GS, Kobilka BK., Science 273 (1996) 803. [22] MacMillan LB, Hein L, Smith MS, Piascik MT, Limbird LE, Science 273 (1996) 801. [23] Hein L, Altman JD, Kobilka BK., Nature 402 (1999) 181. [24] Makaritsis KP, Handy DE, Johns C, Kobilka B, Gavras I, Gavras H., Hypertension 33 (1999) 14. [25] Snapir A, Heinonen P, Tuomainen TP, Alhopuro P, Karvonen MK, Lakka TA, Nyyssonen K, Salonen R, Kauhanen J, Valkonen VP, Pesonen U, Koulu M, Scheinin M, Salonen JT., J. Am.Coll. Cardiol. 37 (2001) 1516. [26] Jewell-Motz EA, Liggett SB., Biochemistry 34 (1995) 11946.

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

305

Computer Graphics Applications on Molecular Biology and Drug Design K. Perdikuri1, A. Tsakalidis1 1

Department of Computer Engineering and Informatics, University of Patras, 26500 Patras, Greece

Abstract. Molecular structure-based drug design is an art and a science. As graphics hardware has matured and software continues to mature, the applications of computer algorithms on molecular biology provides a rigorous basis for many drug design efforts. In this paper we try to present the basic principles in molecular graphics as long as it concerns molecular representations and modelling, and outline some important challenges and open problems in the field of computer aided drug design.

Introduction During the last 20 years, the process of drug discovery and drug design has been enormously affected from the increased available information, concerning 3D structures of biomolecules, determined by X-ray crystallography and NMR structure determination techniques. Exploiting this structural information, scientists can design novel pharmaceutical molecules (often referred to as ligands), which bind tightly and selectively to a target macromolecule, such as a protein (often referred to as receptor). The binding of a ligand to a specified receptor is based on the "lock-and-key" principle, which was firstly recognized by Emil Fischer more than 100 years ago. Along with this principle, the active site of a receptor binds with a ligand both spatially and chemically, just as only a specific key fits a given lock. The design of pharmaceuticals molecules based on this principle is usually referred to as structure-based drug design [1]. Modern methods for structure-based drug design can be divided into two categories. The first category, which is usually referred as Database Searching, searches ligands for a given receptor, through large-scale screens in databases of known 3D molecules. The goal of the search is to retrieve molecules, with shape and chemical complementarity to a given active site of a receptor. The second category of structurebased drug design, which is usually referred as de novo design, "builds" gradually the proper ligands [2]. In this case, ligands, are built up within the constraints (both geometrical

306

K. Perdikuri and A. Tsakalidis / Computer Graphics Applications

and chemical) of the binding receptor, by assembling atoms or small fragments in a stepwise manner. A number of such approaches have already been reported. Several ligands designed in this manner are now in clinical trials. The success of structure-based drug design has encouraged the development of various computational methods that can combine structural information and modem molecular graphics techniques to suggest novel structures, which may prove to be useful lead compounds. Ideally, these methods should be fast, objective and produce a set of diverse yet chemically reasonable structures. In this paper we will present and analyze the current methods used and the problems encountered in the field of Molecular Graphics and especially in the field of structure-based drug design.

Figure 1. A 3D molecule represented with ball and sticks Molecular Representations

Molecular Graphics deals with the representation and manipulation of biological molecules in a computer. It is a newly established scientific field encompassing the theoretical and application areas of computer science that deal with geometry and visualization. Among these areas are computer graphics, computer animation, computer vision, computational geometry and computer-aided geometric design. In this section we will try to introduce some definitions for the concept of molecular surface and molecular volume. The three-dimensional geometric structure of a molecule is often represented as a set of atoms (Figure 1). In this representation each atom is modeled as a "hard" sphere with a certain atomic radius. The three-dimensional placement of the spheres is expressed according to the internal distances between the centers of each pair of atoms (interatomic distances) and their relative angles (Table 1). The equilibrium configuration of this representation is defined as the "low energy conformation". In many representations, the spheres are allowed to interpenetrate one another. In structure-based drug design, one wishes to manipulate a molecule in the presence of another molecule to see whether they fit together. During such manipulation the molecules should have only limited interpenetration. Table 1. Table with internal coordinates of C2H6

1

C

2

C

1.54

1

3

H

1,0

1

109,5

2

4

H

1,0

2

109,5

1

180,0

3

5

H

1,0

1

109,5

2

60,0

4

6

H

1,0

2

109,5

1

-60,0

5

7

H

1,0

1

109,5

2

180,0

6

8

H

1,0

2

109,5

1

60,0

7

K. Perdikuri and A. Tsakalidis / Computer Graphics Applications

307

Molecular Surfaces The determination of the molecular surface of a collection of atom spheres of a molecule is of great importance in various applications in molecular biology for the interpretation of molecular properties, interactions and processes. Molecular surfaces are classified as contact surfaces and re-entrant surfaces. Various definitions have been proposed in the relative bibliography. The van der Waals surface (Figure 2), Lee and Richards' solvent accessible surface [4], and Richards' smooth molecular surface. In Lee and Richard's approach, a sphere is rolled on a reference surface, to obtain a new surface described by the center of the rolling sphere, which is assumed to be fairly small. In a simplified approach the molecular surface of a molecule can be described as the accessible parts of the modeled spheres (assuming that the molecules are modeled by the hard sphere model), or as the boundary of the union of the balls in the hard sphere model. Molecular Volumes A critical issue in most 3D molecular graphics packages is the choice of the model for representing, the three dimensional structure of a molecule. Obviously the molecular geometry is affected from the selected model. Generally the use of any model means that a significant level of detail is lost as molecules are visualized to have surfaces and volumes similar to our perception of surfaces and volumes of macroscopic objects. Although it is clear that the more accurate the model used, the better are the chances in predicting molecular interactions certain abstractions need to be made in order to calculate molecular interactions efficiently. The sphere model is the most popular model for representing approximately the volume of a molecule. A sphere is drawn around the centre of every atom of the molecule. The radius of each sphere reflects the space requirements of the corresponding atom and consequently the total volume of the molecule (the used radii have been determined by a combination of experimental observations and quantum mechanical calculations). Sticks models are also used to represent the bonds between two consecutive atoms and the angles they form in 2 dimensions. The bond lengths and bond angles are degrees of freedom (DOF) for each molecule. This way a conformation of a molecule is obtained by assigning values to all the DOF of the molecule.

Figure 2. Representation of Van der Waals surfaces

Color Coding and Texture Mapping Techniques The molecular surface concept is not only useful for a representation of the shape of molecules. These surfaces can be used as screens for the visualization of arbitrary properties using color coding techniques. Color coding is a popular means of displaying scalar information on a surface [3]. In interactive molecular graphics, high contrast color code variation can be realized by using texture mapping techniques which are available on

308

K. Perdikuri and A. Tsakalidis / Computer Graphics Applications

graphical workstations and PC's to represent a color ramp as a ID texture. Texture mapping is a technique that applies an image to an object's surface. Conformational Analysis Having represented the surface and volume of a molecule the next step in molecular modeling is to find the global energy minimum conformation of a molecule, regardless of the size of the molecule under consideration. This procedure is known as conformational analysis. A variety of conformational analysis methods exist and fall into two classes: deterministic and stochastic. Deterministic methods search all the rotatable bonds in a molecule. The time required for this type of search increases exponentially to the number of rotatable bonds. The success of a deterministic method relies on the selected granularity, which is difficult to define. For example it may be possible to perform a complete search if one limits the number of rotamers per bond to a small number. On the other hand the same search may become computationally infeasible if the number of scanned rotamers increases. Stochastic methods are based on algorithms that limit the conformational search space to the lowest energy conformations of a molecule. Some of the stochastic methods use molecular dynamics, Monte Carlo, genetic algorithms, random perturbations to the coordinates, or a combination in order to find and optimize local minima [9]. Unfortunately stochastic methods are not guaranteed to converge to the same set of low energy conformations as produced with a deterministic method. A slight deviation in a torsion angle may miss the lowest energy state, although we are very close to structurally to the global minimum. Open Challenges and Problems in Molecular Graphics Studying a hard sphere model from a computational geometry point of view, we have to take into consideration several properties such as: (i) the radii ranges in a fairly restricted area, and (ii) sphere centers cannot get too close to one another. One of the tasks of molecular modeling packages is to display molecules. This should preferably be done so fast that the user can interact with the model by turning it around to look at it from different directions or by moving different molecules with respect to each other, to see for example whether they fit together. In this paragraph we will refer to problems and challenges that arise in the field of molecular graphics and are crucial for application in computer-aided structure-based drug design. Reconstructing a 3D Model The reconstruction of a three-dimensional set of points using information about its inter-point distances is a task of great importance in determining molecular structure. In particular, spectroscopic methods such as two-dimensional NMR provide a mean for determining a labeled subset of the distances between atoms in large structures such as proteins and RNA, and such methods have therefore proved extremely valuable in conformational studies of such molecules. Determining the structure of n labeled points in R3 is easy if the exact value of every inter-point distance is given. However, this is far from what one obtains using NMR and its variants. In particular, not all distances can be measured, and the distance values that one does obtain are susceptible to a wide variety of sources of error.

K. Perdikuri and A. Tsakalidis / Computer Graphics Applications

309

NMR experiments are frequently based on measuring proton resonances, since hydrogen atoms are easily detected by these techniques; this is in general sufficient for conformational studies, since hydrogen atoms are abundant in organic molecules. To determine spatial information, one typically requires two NMR experiments: one based on spin-spin coupling to determine protons that are close in the covalent geometry of the molecule (COSY, Correlated Spectroscopy), and another to determine distance information for pairs of protons that are close in space but not necessarily close in the covalent structure (NOESY, Nuclear Overhauser Enhancement Spectroscopy). Combining information from these two types of experiments, one obtains labeled interatomic distance data- more precisely which distance corresponds to which pair of atoms. The nuclear Overhauser Effect (NOE), on which the latter technique is based, is manifested in cross-peaks in an NMR spectrum that arise from dipole- dipole coupling of one proton with nearby protons. NOE intensity is given by a formula with a term for dij-6, where dij is the Euclidean distance between two protons. If one has approximate knowledge of the other variables involved, then it is feasible to recover a value for dij. Distance calculations based on NOESY are subject to numerous sources of systematic error. Some basic examples are the following: i) Spin-diffusion is a phenomenon whereby NOE effects between protons i and j are in some sense "transferred" through a third proton k, resulting in spurious cross-peaks. These effects often result in a compression of the apparent distances derived from transferred NOE experiments. ii) For reasons of experimental efficiency, the relaxation delays between successive NMR scans are typically too short to allow for full recovery of the magnetization effects between nearby protons This is particularly a problem in RNA structure determination, where proton relaxation times are quite long and vary considerably [8]. iii) As the NMR spectrum becomes more and more dense, it is possible for two peaks to essentially coincide, and thus for one or more of these peaks to escape detection when the data is interpreted. This is a problem that becomes increasingly significant precisely when more of the inter-point distances are measurable. iv) As noted above, in order to produce labeled distance data from a NOESY experiment- in other words to determine that a given distance d is in fact associated with protons i and j— one must solve a complex assignment problem using information from an accompanying COSY experiment. Several error-correcting algorithms have been proposed in order to deal with arbitrary errors in distance measurements and reconstruct an accurate 3D model. Most of them belong in the area of distance geometry and use general model-building tool. Hidden Surface Removal Hidden surface raises a difficult problem among sets of intersecting spheres. In practice most 3D graphics workstations available nowadays, use a Z-buffer algorithm either in software or in their graphics hardware. Unfortunately, such an implementation can only handle polyhedral objects, which means that triangular meshes can approximate spheres. To get a reasonable approximation for a sphere one needs more than 100 triangles and these implementations lead to a large number of needed triangles in 3 dimensional space in order to model a molecule. Another common approach to hidden surface removal used in computer graphics is the painter's algorithm. Here one tries to define a depth order on the objects, sorting them from back to front. Next one draws the objects in this order on top of each other where each new object hides the parts of other objects that lie below it. Such an approach does not require special graphics hardware. The problem with this approach is that it requires a valid depth order on the objects. Such an order does not always exist, as in the case of intersecting objects. In such applications the combinatorial representation of visible pieces,

310

K. Perdikuri and A. Tsakalidis / Computer Graphics Applications

can be seen in the visibility map. The visibility map can be defined as the subdivision of the viewing plane into maximal connected regions in each of which a single object is seen, or no object is seen. Modelling Water What is important in a software package is the ability to simulate the presence of water molecules, in order to understand the effects of water on the shapes of biological molecules. Taking into consideration when building a cell that it contains billions of water molecules and the space not occupied by the atoms of biological molecules is filled with water, it is clear why water plays an important role in molecular interactions. A single water molecule (H2O) has a tetrahedral geometry, which gives water a loosely packed structure compared with that of most other liquids, such as oils. To construct a computer aided model of water, we need to take into account two different types of forces: intramolecular and intermolecular. Moreover there are two types of waters that must be considered when building a computer model of a biological molecule in aqueous solution: the "ordered waters" that surround and strongly interact with the molecule and the "bulk waters" that may be buried within the molecule [6]. Subsequent simulations of DNA in water have revealed that water molecules are able to interact with nearly every part of DNA double helix, including the base pairs that constitute the genetic code. In other words water help us to reveal the structure of several biomolecules. In contrast, water is not able to penetrate deeply into the structures of proteins. Experimental results have shown that water profoundly influences the interactions of proteins and DNA. The polarity and hydrogen-bonding capability of water make it a highly interacting molecule. Water can greatly weaken the electrostatic forces and H-bonds between polar molecules by competing for their attraction, play both roles of an H-bond donor and, by lending electron pairs of the oxygen atom, an acceptor, and form a variety of bridges between molecular donors and acceptors [7]. Molecular Docking One important application in the field of computer aided molecular modeling is related to the molecular docking process. An effective and interactive handling of the molecular complementarity is needed in order to investigate possible docking partners. Hidden surface problem is highly related to the protein docking process. Although the mechanisms of docking reactions are not well understood, two complementarity principles seem to be important for the recognition and binding of docking partners. The first principle is the shape complementarity principle: the shapes of the molecules that build a docking complex are complementary, that is, there is a large fit between the surfaces of the docking partners. The second complementarity principle is the chemistry principle: there is a strong chemical complementarity (with respect to hydrogen bonds, electrostatic interactions, hydrophobicity and so on), between the sites of docking partners. Although the second principle is the more important one, it is possibly to identify many docking sites solely with the help of the shape complementarity principle. In order to find these sites for two molecules A, B with n and m atoms, the following 3D. matching problem has to be solved: determine all transformations of B such that there is a large fit between the surface of A and the surface of B, and no penetration of B into the interior of A. For all the candidates with a good geometric fit the potential energy function of the docking conformation and the molecules A and B has to be computed. In the above description of the geometric 3D matching problem two strong assumptions were made. 1)

K. Perdikuri and A. Tsakalidis / Computer Graphics Applications

311

the molecules are considered rigid and 2) there is no penetration. Of course not all molecules are rigid and certain parts of the molecules are very flexible influencing their chemical activity. More over local changes of the shape of the docking sites happen during the docking reactions. So with the structure of even one target protein, and the knowledge of function of its receptor or active site, it is now possible to use computer tools to build and dock a ligand or inhibitor ("new leads") prior to investing time and resources for synthesis and testing. Thus molecular modeling is essential for understanding and exploring the structure-function relationship and evaluating novel compounds before being synthesized. Based on Fischer's "lock and key" principle, the mechanical view of molecular interactions can be easily understood and applied to biomolecules. However, even "rigid" molecules have local flexibility, as previously mentioned, and water molecules as previously described are usually a structural appendage of both the "lock" and the "key," which means the in vivo structure may differ significantly from that on the display screen. Conclusion Researchers in the areas of computer graphics, molecular biology and computational chemistry have long studied representations of molecular models as well as other computational problems related to the geometry of molecules. Current trends incorporate tools from computational geometry in order to provide algorithms that are provably efficient and at the same moment fast in practice. Taking as an input the information determined by X-ray crystallography and NMR structure determination techniques, scientists can move from sequence level to structure level, which is an important step for understanding the molecular details of biological processes. Various fields start to develop like genome modeling and annotation; comparative protein modeling and fold assignment; in silica drug design and modeling of cellular processes. Acknowledgments We would like to express our thanks to Prof. J. Matsoukas and T. Mavromoustakos for their support and useful guidelines. This work was partially supported by the General Secretariat of Research and Technology of Greece during the research project EPET II/ EKBAN 115. References [ 1 ] D. Kuntz, Science 257 (1992) 1078. [2] R. Wang, Y. Gao, L. Lai, LigBuilder, J. Mol. Modeling 6 (2000) 498. [3] J. Brickmann, T. Exner, M. Keil, R. Marhofer, J.Mol.Modeling, 6 (2000) 328. [4] F. Richards, Ann.Rev.Biophys.Bioeng 6 (1977) 151. [5] F. Richards, Methods in Enzymology 115 (1985) 440. [6] M. Gernstein, M. Levitt, Scientific American (November 1998). [7] Y. Deng, J. Glimm, Y. Wang, A. Korobka, M. Eisenberg, A. Grollman, J.Mol.Modeling 5 (1999) 125. [8] B. Berger, J. Kleinberg, T. Leighton, Journal of the ACM 46 (1999) 212. [9] Spellmeyer, D. Wong, A. Bower, M. Blaney, J., J.Mol.Gmphics Mod., Vol. 15, 1997. [10] H. Lenhof, An algorithm for the protein docking problem. From Nucleic Acid and Proteins to Cell Metabolism, D.Schomburg and U.Lessel (ed),1995 125-139.

312

Drug Discovery and Design: Medical Aspects J. Matsoukas and T. Mavromoustakos (Eds.) IOS Press, 2002

Investigation of Novel DNA Gyrase Inhibitors Using the High Resolution NMR Spectroscopy S.G. Grdadolnik, M. Oblak, T. Solmajer and R. Jerala Laboratory for Molecular Modeling and NMR Spectroscopy, National Institute of Chemistry, POB 660, Hajdrihova 19, 1001, Ljubljana, Slovenia

Abstract. DNA gyrase is a well-established antibacterial target, because it is restricted to prokaryotic organisms. Novel DNA gyrase inhibitors are still searched in order to overcome the limitations of the known inhibitors connected with resistance and cytotoxicity. In our efforts to find new lead compounds, the high resolution NMR Spectroscopy has been utilized, which has recently become a high-throughput screening method. The binding of new compound has been determined by observation of chemical shift changes of resonances of the 24 kDa N-terminal fragment of gyrase B, which contains the entire ATP binding site. The resonances of the protein have been sequentially assigned by application of the threedimensional heteronuclear correlation methods in order to identify the interactions of inhibitor with the protein binding site. The location and orientation of the new molecule in the protein binding site has been studied by molecular modeling methods and analysis of chemical shift differences.

Introduction High resolution NMR Spectroscopy is a powerful technique for studying the conformational and dynamic properties of drug molecules. The information derived from the conformational analysis studies aids in rationalizing the drug design of new synthetic molecules with better physicochemical properties. However, the relationship between bioactivity and conformation is indirect. In addition, conformational analysis in general follows the synthesis of novel drug molecules. Recently, NMR Spectroscopy has been applied very efficiently at an earlier stage of drug research, in screening for novel ligands (1-4). In particular, its ability in identification of weakly binding ligands has been utilized, which offers the identification of binding sites at unique resolution. Systems of proteins

S.G. Grdadolnik et al. / Investigation of Novel DNA Gyrase Inhibitors

313

with weakly binding ligands exist in a dynamic equilibrium between free and bound species, which usually can not be crystallized and the NMR is the method of choice for studying conformational and dynamic properties of such complexes. The search for the ligands with low binding affinity is important, because it reduces the number of compounds, which have to be synthesized. These molecules can be further improved on the basis of structural information and the knowledge of their interactions with the protein target. A number of low affinity ligands can be combined as building blocks for novel high binding affinity ligands. Thus the number of compounds, which have to be synthesized, is reduced and there is no need for investigation of a huge number of compounds by random screening, which can fail in searching for the new lead structures. Binding of new compounds to the protein target can be monitored by observation of chemical shift changes of the protein signals in the "fingerprint" 2D HSQC spectrum (5). When the molecule binds to the protein it changes the local chemical environment and causes the change of the local magnetic field, which manifests in the change of chemical shifts of the residues at the binding site. Chemical shifts can be rapidly measured even for larger proteins. The measuring time for NMR experiments mentioned above is approximately 30 minutes with conventional NMR equipment and approximately 10 minutes with the new cryogenic probe technology (2). In this article, we are presenting our efforts in investigation of novel DNA gyrase inhibitors. DNA gyrase is an essential prokaryotic type II topoisomerase, which is involved in vital processes of DNA replication, transcription and recombination (6,7). Its role as an attractive antibacterial target arises from the fact that there are no mammalian gyrases (8). The active enzyme consists of an A2B2 complex. The A subunit (97 kDa) is involved in direct interactions in DNA breakage and reunion, while the B subunit (90 kDa) catalyses the hydrolysis of ATP. Several classes of potent DNA gyrase inhibitors have been found e.g. quinolones, cyclothialidines and coumarins. Especially quinolones, which bind to the A subunit and inhibit the DNA-breakage-reunion cycle (9), are widely used in medicine as broad-spectrum antibiotics against bacteria such as B.anthracis, which causes anthrax (10). Cyclothialidines and coumarins bind to the ATP recognition site located in the subunit B and inhibit the ATP-ase activity, as well as dimerization of the protein (11,12). However, the known inhibitors have limitations due to bacterial resistance and cytotoxicity for human cells (13,14), which are stimulating the search for novel compounds. The methodology applied in our searching for novel high affinity DNA gyrase inhibitors is high resolution NMR spectroscopy and consists of the following steps. 1st step: Resonance assignment of the protein target. The assignment of resonances in the NMR spectra to specific atoms in the protein sequence enables to identify the residues of protein, which are influenced upon binding of ligands. The molecular weight of the DNA gyrase subunits is far beyond the capabilities of high-resolution NMR spectroscopy, which in nowadays are limited to around 40 kDa. Fortunately, the 24 kDa Nterminal part of the B subunit contains the entire ATP binding site (15,16) and can be separately overexpressed in bacteria, isotopically labeled and prepared in sufficient concentrations for NMR measurements to perform sequential resonance assignment of the protein. 2nd step: Exploring the binding of ligands. In the second step binding of low molecular weight ligands is determined by observation of chemical shifts changes in the 2D 15 N HSQC spectra of the protein. Detailed analysis of chemical shift differences allows the determination of location of the protein binding site. 3rd step: Optimization of the ligand binding. Information from chemical shift changes is used in combination with the results from the molecular modeling of ligands complexed with the protein to predict their location and orientation in the protein binding site, which is essential for the optimization of ligands and rational design of more potent

314

S.G. Grdadolnik et al. / Investigation of Novel DNA Gyrase Inhibitors

inhibitors. The fragments, which are identified as ligands and bind to the individual protein subsites are linked together in order to find more potent DNA gyrase inhibitors. The NMR methodology described above is using the principles of the so-called SAR-byNMR approach, which was developed in the group of S. Fesik at Abbot Laboratories (1–4). Materials and methods Sample preparation: The 24 kDa fragment of gyrase B from E.coli was overexpressed in the same organism and isolated by affinity chromatography on novobiocin-Sepharose based on the modification of a method of Staudenbauer et al. (17). Protein was eluted from the novobiocin-Sepharose column in a buffer containing 8 M urea in 20 mM Tris pH 7.5 and refolded by extensive dialysis against 50 mM Tris pH 7.5, 10% glycerol, 1 mM EDTA, 2 mM DTT and 0.02 % NaN3. The last buffer exchange contained 20 mM K-phosphate buffer at pH 7.2, ImM EDTA and 2 mM DTT. For NMR experiments protein sample was concentrated to approximately 0.4 mM concentration and 5% of D2O was added. For NMR experiments involving ligands, which are poorly soluble in water, DMSO was added to the sample up to the final concentration of 15%, at which concentration the protein still displayed the full dispersion of chemical shifts, indicating the preservation of defined tertiary structure. Stock of gyrase inhibitors were prepared at 10-100 mM concentration in DMSO and used for NMR titration of 24 kDa fragment. NMR spectroscopy: All NMR experiments were recorded at 303 K on a Varian INOVA 600 MHz NMR spectrometer using the pulse sequences provided in the ProteinPack Varian library. All experiments employed pulsed field gradients and gradient-selected sensitivity enhancement. For water suppression selective water flip-back pulses and gradient coherence selection were utilized (18). The 1H, 13Ca and I3C0/^ spectral widths were 10, 2.2, 4.8 and 12 kHz, respectively. The three-dimensional (3D) HNCA (18,19), HN(CO)CA (20) and HNCACB (18,19,21) spectra were recorded with the 512x64x32 complex data points, 24 scans and a relaxation delay of ls. The two-dimensional (2D) I5N HSQC spectra (5) were recorded with 512x64 complex points, 16 scans and a relaxation delay of ls. Data processing has included linear prediction in the incremented dimensions, apodization with sine-bell square function shifted by Ti/2 and zero filling. Molecular Modeling Molecular models of the 24kDa N-terminal fragment of gyrase subunit B in the complex with ligands novobiocin and FQ-115 were built by using the experimental X-ray structures. Two structures deposited in the Protein Data Bank were used to model the complex with novobiocin: N-terminal 43kDa fragment of E. coli GyrB subunit (g43) with ADPNP in the binding site (PDB entry 1EII) (22) and N-terminal 24kDa fragment of E. coli GyrB subunit (g24) with novobiocin in the binding site (PDB entry 1AJ6) (23). For building of the model of ligand FQ-115 in the complex with 24kDa N-terminal fragment of gyrase subunit B we used the experimental X-ray structure of the DNA gyrase (N-terminal 43kDa fragment of E. coli GyrB subunit) with ADPNP in the binding site (15) and, in addition, structures of ATP binding site in a structurally similar enzyme protein tyrosine kinase-substrate/inhibitor analogue complexes: protein tyrosine kinase in complex with a chromone derivative (PDB entry 2HCK) and tyrosine kinase with a nucleotide analog AMPPNP in the binding site(PDB entry 1AD5) (24). By superimposing the backbones of experimental structures of two protein tyrosine kinase complexes and ATP binding site of the gyrase the starting position of FQ-115 in the

S.G. Grdadolnik et al. / Investigation of Novel DNA Gyrase Inhibitors

315

DNA gyrase (subunit B) was determined. Next, molecular mechanics computations were performed in order to refine the initial structural models by using the Discover module as available in InsightII, Version 97 software package (MSI). The inhibitor binding sites were soaked with water molecules using radius of hydration R = 15A. All calculations were performed at pH 7.4 to determine the charge state of ionizable amino-acid residues, and consistent valence force field (CVFF) was used throughout. Potential energy of all systems described above was minimized using a combination of steepest descents (1000 steps) and conjugate gradients (5000 steps). The minimization was terminated when the energy gradients did not exceed 0.001 kcal/A.

Results Assignment of the protein: The assignment of the NMR signals to specific atoms in the protein sequence is done in a sequential way i.e. the signals of the neighboring amino acids are searched on the basis of inter- and intra-molecular correlations of the backbone nuclei. A variety of three-dimensional (3D) heteronuclear correlation experiments have been developed for sequential resonance assignment of the protein backbone (25,26). These methods are designed to overcome the problems of large signal overlap and short transverse relaxation times, which are connected with the large molecular weight of the proteins and prevent the application of conventional 2D NMR. Additional dimensions are implemented in the NMR experiments (3D or 4D NMR) to reduce signal overlap and larger heteronuclear coupling constants are used for magnetization transfer to overcome fast relaxation. For the proteins with molecular weight above ca. 20 kDa the use of partial deuteration to reduce relaxation is beneficial (27, 28). In addition, sample concentrations must be kept sufficiently low to prevent aggregation of the protein. Therefore, isotopic labeling of proteins with 15N and/or 13C nuclei is required to obtain the 3D spectra in a reasonable measuring time. The N and C uniformly labeled 24-kDa N-terminal fragment of gyrase B was prepared to perform the 3D heteronuclear correlation methods. The most stable condition of the protein is achieved during the complexation. Thus, the addition of novobiocin (Fig. 1), which belongs to the family of coumarin antibiotics (11) and is one of the most potent DNA gyrase inhibitors with binding affinity in nanomolar range caused stabilization of the complex with DNA gyrase protein. The novobiocin - gyrase B complex remained stable over several weeks, thus the measuring of a series of 3D experiments under stable conditions was possible. As shown bellow the experiments performed on uniformly 15N, C-doubly-labeled protein with combination of comparative studies of the free and bound protein were sufficient to assign the protein and the expensive procedure of the protein partial deuteration could be avoided.

H2N

w

OH

H \\ O

CH3

Figure 1 :The chemical structure of novobiocin, which is one of the most potent DNA gyrase inhibitors.

316

S.G. Grdadolnik el al. / Investigation of Novel DNA Gyrase Inhibitors

In order to select the most efficient way for sequential resonance assignment of the protein the NH signal comparison of the commonly used 3D methods was performed. The 24 kDa molecular weight is already critical for success of the high resolution NMR measurements and after sensitivity comparison it was realized that only a few of the 3D experiments can be successfully applied. For the sequential assignment HNCA and HN(CO)CA experiments were chosen. In the HNCA spectrum intra- and inter-residue correlations of amide 1HN and 15 N nuclei with Ca nuclei are observed (Fig.2, left), while in the HN(CO)CA spectrum only the inter-residue correlations between the same nuclei are present (Fig. 2, right). With the comparison of both spectra the intra- and inter-residue correlations can be distinguished and search for the resonances of the neighboring residues can be performed. Promising sensitivity was observed also for the HNCACB spectrum, where besides the correlations present in HNCA spectrum also the intra- and inter-residue correlations of the amide 1HN and 15N nuclei with I3C^ nuclei are observed. However, the sensitivity of the experiment was insufficient to be used for sequential assignment, because mostly only the intra-residue correlations resulted in the spectrum. The HNCACB spectrum contains helpful information of C chemical shifts, which have characteristic values for certain amino acids like alanine, serine or threonine and significantly facilitate the identification of the residues and thus the resonance assignment of the protein. In order to overcome the assignment problems due to the limited number of 3D experiments comparative NMR studies of the free and bound protein were performed and the information from the X-ray structure of the novobiocin-Gyrase B complex regarding the observed hydrogen bonds and hydrophobic interactions between the novobiocin and protein was used (16). The Ala47, Asn46, Asp73 and Arg136 form hydrogen bonds directly with novobiocin, while the Val43, Glu50, Gly77, Val71 and Thrl65 form hydrogen bonds across the ordered water molecules. The Arg76 and Pro79 form hydrophobic interaction with the coumarin ring of the novobiocin. HNCA

HN(CO)CA Q72

cu,,

El 8.1 056

" if

G77

F.I 83

Tt

F.I83? 1 LI.W t

*i

HiC

!<••„. 1 C '(i 1 •

» V7I Y184 >CW

12.0

II.0

10.0

H Ippm]

9.0

8.0 \ = i : s ~ < . ppin

12.0

II.0

10.0 H|ppm|

9.0

8.0 s = i:s^f>ppm

Figure 2: Planes from the 3D HNCA and HN(CO)CA spectra. With combination of both spectra inter (i-l) and intra (i) - residue correlations of amide nuclei ( 1 HN, I5 N) with 13Ca nuclei in HNCA spectrum can be distinguished. The sequential assignment is performed by searching for the amide resonances, which has the Ca ( i - l ) and C a (i) correlations at the same chemical shift. In the presented planes the amide resonance of the previous residue to Y184 was found in the same plane: the C" (i-l) correlation peak of amide Y184 has the same chemical shift as the Ca correlation peak of amide El83, thus identifying the signals of the two neighboring residues.

S.G. Grdadolnik et al. /Investigation of Novel DNA Gyrase Inhibitors

317

Figure 3: The model of novobiocin in complex with the 24 kDa N-terminal fragment of gyrase B. The residues participating in hydrogen bonds and hydrophobic interactions with novobiocin are labeled. The part of the protein, which could not be traced in the X-ray structure, is colored in black.

In the Brookhaven Protein Data Bank only the crystal structure of the N-terminal 24kDa fragment R136H mutant of DNA gyrase complexed with novobiocin is deposited. This structure lacks some structural information, which is vital for our study." Therefore, a structural model of g24-novobiocin complex (Fig. 3) has been constructed, which includes Argl36 and addition of missing amino acid residues not traced in the X-ray structure. Modeled positions of residues are in good agreement with previously published analyses (16,22). For example, Argl36 forms a direct hydrogen-bonding interaction with novobiocin, which is vital in directing the noviose sugar to the appropriate site. In the wild type structure the ester and carbonyl oxygens of the coumarin ring are 3.2 and 2.6A from the guanidinium nitrogens of Argl36, while the distances in our model are 3.18 and 2.92A, respectively (Fig. 4).

Figure 4: Detail from the model of novobiocin in complex with the 24 kDa N-terminal fragment of gyrase B: interaction of Argl36 with the coumarin ring of novobiocin.

318

S.G.Grdadolnik et al. /Investigation of Novel DNA Gyrase Inhibitors

Figure 5: The 2D 1SN-HSQCspectra of the free (left) and bound (right) protein. The signals of residues, which are involved in hydrogen bonds and hydrophobic interactions with novobiocin are labeled.

The chemical shifts of residues of the protein involved in intermolecular interactions with novobiocin are expected to be the most significantly affected. The "N HSQC spectra of the free and bound protein (Fig. 5) show that the chemical shifts of most of the residues, directly involved in..intermolecular interactions are significantly changed upon addition of novobiocin and served to check and complete the resonance assignment of the protein. Finally, almost the entire protein (90%) was assigned and the complete assignment of the residues of the protein active site was achieved, which is of the most importance for search for the new ligands. Identification of the protein binding site: The binding site of ligands in the protein can be identified on the basis of the most pronounced chemical shift differences upon addition of ligands. In the case of novobiocin the chemical shift differences clearly indicate binding of the antibiotic to the ATP binding site of the protein. Most of the residues with pronounced chemical shift differences are concentrated around the residues, which are involved in intermolecular interactions with the novobiocin (Fig. 6). Besides the interactions mentioned in the previous chapter, we can observe pronounced effect on the chemical shifts of the helix, which contains Val120 (Fig. 6). The side chain of Val120 is in a hydrophobic contact with the methyl groups of the noviose part of the novobiocin. The comparative NMR studies of the free protein and protein in complex with novobiocin were not important just for the assignment purposes but also enabled us to identify the residues of the protein, which are inflpenced upon binding of the novobiocin. This information can help us to determine the location of the new ligands in the ATP binding pocket.

S.G. Grdadolnik et al / Investigation of Novel DNA Gyrase Inhibitors

319

Figure 6: Presentation of residues with the most pronounced chemical shift differences upon addition of novobiocin. The residues, with the chemical shifts differences larger than 0.1 ppm for 1H or 0.4 ppm for 15N nuclei are colored in gray and black. The residues, which were mentioned in the X ray studies to participate in the interactions with novobiocin are colored in black.

Searching for the new ligands: The binding of various low molecular weight compounds, which were expected to bind to individual subsites of the ATP binding pocket according to our structural models, has been tested. Binding of a very promising compound namely FQ115 was characterized. This compound is structurally different than novobiocin. The effect on the chemical shifts of the protein upon binding of FQ-115 was observed in the 2D 15NHSQC spectra (Fig. 7).

H'[ppin]

H [ppm]

Figure 7: Parts of the 2D 15N-HSQC spectra of 24 kDa N-terminal fragment of gyrase B in complex with FQ115 (left) and novobiocin (right), which are overlapped with the corresponding region of 2D 15N-HSQC of the free protein (dotted contours) recorded under the same conditions. Generally lower chemical shift differences are observed for FQ-115 in comparison with novobiocin as shown for R76 and R136 residues.

320

S.G. Grdadolnik et al. / Investigation of Novel DNA Gyrase Inhibitors

The binding affinity of FQ-115 to 24 kDa fragment of gyrase B is in 10 micromolar range as determined by fluorescence measurements, which might account for lower chemical shift differences in comparison to novobiocin. Measurements of gyrase inhibition by FQ-115 confirmed that it indeed inhibits the activity of gyrase B (29). Location of the new ligand in the protein binding site: All chemical shift differences due to addition of the FQ-115 to the protein are located around the same residues as in the case of novobiocin (Fig. 8) indicating that the FQ-115 binds to the ATP binding site of DNA gyrase. The location of molecules in the ATP binding site can be predicted on the basis of induced chemical shift differences (Fig. 8, left) and comparison with the effect of novobiocin on the chemical shifts (Fig. 8 right). The NMR experimental data are in good agreement with the model of FQ-115 in complex with gyrase B, which was obtained using molecular modeling methods (Fig.8). By mapping the chemical shift differences of affected residues by different colors according to their size (Fig 8) the reduced effect of FQ-115 in comparison with novobiocin on the residues R136, and R76 (which interact with the coumarin ring of novobiocin) can be observed. This indicates that the FQ-115 is probably not located in the part of the binding pocket, which is occupied by the coumarin ring of novobiocin. On the other hand, the FQ-115 is significantly perturbing the chemical shifts of the residues, which are participating in the hydrogen bonding network with noviose moiety of novobiocin and are located close to the adenine part of ADPNP, as well as the V120, which forms a hydrophobic contact with methyl groups of noviose sugar. Thus the experimental NMR data support our model, in which the FQ-115 is located in the region of binding site occupied by ADPNP and is showing overlap with the noviose sugar moiety of novobiocin.

Figure 8: Models of 24 kDa N-terminal fragment of gyrase B in complex with FQ-115 (left) and novobiocin (right). In the model of complex with FQ-115 (in black) the locations of novobiocin (in white) and ADPNP (in gray) in the protein binding site are also shown. The influence of ligand FQ-115 on the chemical shifts of the protein residues is mapped on the left figure and for comparison the influence of novobiocin is presented on the right figure. The residues, which show the chemical shifts differences larger than 0.1 ppm for 1H or 0.4 ppm for 15N nuclei are colored in black. The residues with the less pronounced chemical shift differences i.e. smaller than 0.1 ppm for 1H nuclei or 0.4ppm for I5N nuclei are colored in gray.

S.G. Grdadolnik et al. / Investigation of Novel DNA Gyrase Inhibitors

321

Model of ligand docking to the gyrase B fragment will be improved and more information about intermolecular interactions will be obtained also from measurements of structurally dependent NMR parameters, especially NOE effects. However, the initial information about location and orientation of ligands as shown in the example of compound FQ-115 in the protein binding site can be rapidly obtained from the chemical shift analysis once the protein has been fully assigned. Conclusion The NMR approach is applied (1-4) to search for the new DNA gyrase inhibitors. In order to identify the binding site of new ligands and to determine their location and orientation in the binding site the sequential resonance assignment of the protein has been performed. With combination of 3D heteronuclear correlation methods and comparative studies of the free and bound protein the backbone of 90% of the residues has been assigned. Binding of new molecules to protein target is monitored by observation of chemical shift changes in the 2D 15N HSQC spectra of the protein. The binding of a new DNA gyrase ligand (FQ-115) is identified which was found to be biologically active. The detailed analysis of chemical shift differences upon addition of FQ-115 and molecular modeling of the 24kDa Gyrase B fragment in complex with FQ-115 provided initial information about the location and orientation of ligand in the ATP binding pocket. On the basis of this information we are currently optimizing this promising DNA gyrase inhibitor and searching for compounds, which would occupy the rest of the binding pocket. In the next stage the fragments, which bind to the individual protein subsites could be linked together in order to find more potent DNA gyrase inhibitors, which could serve us as leads for rational drug design. Acknowledgment This work was supported by the Ministry of Education, Science and Sport of the Republic of Slovenia. References: [ 1] P. J. Hajduk, R. P. Meadows and S. W. Fesik, Q. Rev. Biophys. 32 (1999) 211. [2] P. J. Hajduk, T. Gerfin, J. M. Boehlen, M. Haberli, D. Marek and S. W. Fesik, J. Med. Chem. 1 (1999) 2315. [3] P. J. Hajduk, R. P. Meadows and S. W. Fesik, Science 278 (1997) 497. [4] S. B. Shuker, P. J. Hajduk, R. P. Meadows and S. W. Fesik, Science 274 (1996) 1531. [5] L. E. Kay, P. Keifer and T. Saarinen, J. Am. Chem. Soc. 114 (1992) 10663. [6] J. C. Wang, Annu. Rev. Biochem. 54 (1985) 665. [7] R. J. Reece and A. Maxwell, Crit. Rev. Biochem. Mol. Biol. 26 (1991) 335. [8] A. Maxwell, Biochem. Soc. Trans. 27 (1999) 48. [9] A. Maxwell, J. Antimicrob. Chemother. 30 (1992) 409. [10] S. I. D'iakov. V. V. Katsalukha, I. K. Lebedeva, A. V. Lukashina, and V. Raiskaia, Antibiot. Khimioter. 39 (1994) 15. [11] A. Maxwell, Mol. Microbiol. 9 (1993) 681. [12] N. Nakada, H. Gmuender, T. Hirata and M. Arisawa, Antimicrob. Agents. Chemother. 38 (1994) 1966. [13] J. S. Wolfson, Etir. J. Clin. Microbiol. Infect. Dis. 8 (1989) 1080. [14] M. Stieger, P. Angehrn, B. Wohlgensinger and H Gmuender, Antimicrob. Agents. Chemother. 40 (1996) 1060. [15] D. B. Wigley, G. J. Davies, E. J. Doson, A. Maxwell and G. Dodson, Nature. 351 (1991) 624.

322

S.G. Grdadolnik et al. / Investigation of Novel DNA Gyrase Inhibitors

[16] R. J. Lewis, O. M. P. Singh, C. V. Smith, T. Skarzynski, A. Maxwell, A. J. Wonacott and, EMBO J. 15 (1996) 1412. [17] W. L. Staudenbauer and E. Orr, Nucleic.Acids.Res. 9 (1981) 3589. [18] D. R. Muhandiram and L. E. Kay, J. Magn. Reson. Ser. B, 103 (1994) 203. [19] L. E. Kay, G. Y. Xu and T. Yamazaki, J. Magn. Reson. Ser. A, 109 (1994) 129. [20] T. Yamazaki, W. Lee, C. H. Arrowsmith, D. R. Muhandiram and L. E. Kay, J. Am. Chem. Soc., 116 (1994) 11655. [21] M. Wittekind and L. Mueller, J. Magn. Reson. Ser. B, 101 (1993) 201. [22] L. Brino, A. Urzhumtsev, M. Mousli, C. Bronner, A. Mitschler, P. Oudet and D. Moras, J. Biol. Chem.. 275 (2000) 9468. [23] G. A. Holdgate, A. Tunnicliffe, W. H. J. Ward, S. A. Weston, G. Rosenbrock, P. T. Barth, I. W. F. Taylor, R. A. Pauptit and D. Timms, Biochemistry, 36 (1997) 9663. [24] F. Sicheri, I. Moarefi, and J. Kuriyan, Nature, 385 (1997) 602. [25] L. E. Kay, M. Ikura, R. Tschudin and A. Bax, J Magn. Reson., 89 (1990) 496. [26] L. E. Kay, M. Ikura and A. Bax, J. Magn. Reson., 91 (1991) 84. [27] R. A. Venters, C. C. Huang, B. T. Farmer R. Trolard, L. D. Spicer and C. A. Fierke, J. Biomol. NMR. 5 (1995) 339. [28] K. H. Gardner and L. E. Kay, Annu. Rev. Biophys. Biomol. Struct., 27 (1998) 357. [29] M. Golob, A. Plaper, S. Zalar, R. Jerala, In vitro study of effects of flavonoids on DNA gyrase activity. In: (T. Lah, T. Turk, M. Dolinar eds.) 3rd Meeting of the Slovenian Biochemical Society with International Participation, Portoroz, 25–29. September, 1999. Book of abstracts. [Ljubljana: Slovensko biokemijsko drutvo, 1999], p. 30.

323

Author Index Andreu-Vieyra, C.V. Andreychenko, S. Apostolopoulos, V. Athanasellis, G. Blagbrough, I.S. Bonvin, A. Bouga, E.A. Camoutsis, Ch. Chatzantoni, K. Constantinou-Kokotou, V. Cordopatis, P. Courty, J. Daliani, I. Damianakos, C. De Clercq, E. Demetzos, C. Deraos, S. Detsi, A. Flordellis, C.S. Foscolos, G.B. Fountoulakis, M. Fragiadaki, M. Fytas, G. Galanis, A. Gates, P. Gavrielatos, E. Georgiev, V. Gerothanassis, I.P. Giatas, N. Grdadolnik, S.G. Habibi, H.R. Hayden, L.J. lakovidou, Z. Igglessi-Markopoulou, O. Iliodromitis, E. Jerala, R. Kapou, A. Karakantza, M. Karamanos, N.K. Karayianni, V. Karestou, E. Katsoris, P. Keivish, T. Kidric, J.

41 294 258 25 64 180 299 97 241 76 188,205,217 251 116,125 53 103,272 125,131 116,180 25 299 103 279 217 103 188 53 25 267 180 3 167,174,312 41 236 20,97 25 3 312 150,167 241 229 97 251 251 3 161

Kokotos, G. 76 Kolocouris, A. 103 Kolocouris, N. 103 Koumentakos, S. 217 Kovala-Demertzi, D. 150 Kremastinos, D. 3 Kyriakidis, D.A. 33 Kyrikou, I. 142,150 Liakopoulou-Kyriakides, M. 13,20 Lymperopoulos, A. 299 Magafa, V. 205,217 Magrioti, V. 76 Makridis, T. 76 Manolis, A.S. 299 Markopoulos, J. 25 Maswadeh, H. 125 Matsoukas, J. 3,116,180,241 Mavromoustakos, T. 3,116,125,142, 150,167,174 Militsopoulou, M. 53 Mioglou, E. 20,97 Moore, D. 236 Moore, G.J. 236 Mourelatos, D. 97 Mouzaki, A. 241 Murray, G. 236 Mutule, I. 3 Neal, A.P. 64 Neyts, J. 103 Nikolaropoulos, S. 167 Nounesis, G. 125 Oblak, M. 312 Padalko, E. 103 Pantazaka, E. 53 Papadimitriou, E. 251 Papaioannou, D. 53 116,241 Papathanassopoulos, P. Papazacharias, S. 188 Perdikuri, K. 305 Plebanski, M. 258 Polevaya, L. 3 Polykratis, A. 251 Poulos, C. 142 Pristovsek, P. 161

324

Pritsa, A.A. Probert,L. Raptis, D. Roumelioti, P. Roy, S. Sarigiannis, Y. Scheinin, M. Slaninova,J. Solmajer,T. Soulakis, V. Spyroulias, G.A. Stavropoulos, G. Tchekalarova,J. Theocharis, A.D. Thymianou, S. Troganis, A.N.

33 116 217 3 236 13 299 217 312 97 188,205,217 13,20 267 229 241 116,180

Tsakalidis, A. Tselios.T. Tsiakopoulos, N. Tsortos, A. Tzakos, A. Tzougraki, C. Vakalopoulou, P. van Nuland, N. Vassis, S. Vlahakos, D. Voyiatzi, K. Zervou, M. Zoga, A. Zompra, A.A. Zoumpoulakis, P.

305 116,180,241 53 125 180 83 20 180 53 3 53 174 3 205 3,116,174