Conformational properties of a cyclic peptide bradykinin B2 receptor antagonist using experimental and theoretical methods

Download Conformational properties of a cyclic peptide bradykinin B2 receptor antagonist using experimental and theoretical methods

Post on 06-Jul-2016




0 download

Embed Size (px)


<ul><li><p>K. KrauseL.F. PinedaR. PeteranderlS. Reissmann</p><p>Authors affiliations:</p><p>K. Krause and L.F. Pineda, Institut fur Biochemie</p><p>und Biophysik, Friedrich-Schiller-Universitat</p><p>Jena, Philosophenweg 12, D-07743 Jena,</p><p>Germany.</p><p>R. Peteranderl, Institut fur Organische Chemie</p><p>und Biochemie, Technische Universitat</p><p>Munchen, Lichtenbergstrae 4, D-85747</p><p>Garching, Germany.</p><p>S. Reissmann, Institut fur Biochemie und</p><p>Biophysik, Friedrich-Schiller-Universitat Jena,</p><p>Philosophenweg 12, D-07743 Jena, Germany.</p><p>Correspondence to:</p><p>Luis Felipe Pineda De Castro</p><p>Institut fur Molekulare Biotechnologie e.V.</p><p>Postfach 100813</p><p>D-07708 Jena</p><p>Germany</p><p>Tel.: 49-3641-656491</p><p>Fax: 49-3641-609818</p><p>E-mail:</p><p>Dates:</p><p>Received 9 April 1999</p><p>Revised 25 May 1999</p><p>Accepted 24 July 1999</p><p>To cite this article:</p><p>Krause, K., Pineda L.F., Peteranderl, R. &amp; Reissmann, S.</p><p>Conformational properties of a cyclic peptide bradykinin</p><p>B2 receptor agonist by experimental and theoretical</p><p>methods.</p><p>J. Peptide Res., 2000, 55, 6371</p><p>Copyright Munksgaard International Publishers Ltd, 2000</p><p>ISSN 1397002X</p><p>Conformational properties ofa cyclic peptide bradykinin B2receptor antagonist usingexperimental and theoreticalmethods</p><p>Key words: Molecular dynamics simulations; NOE; peptide</p><p>hormones; simulated annealing</p><p>Abstract: The solution conformation of the cyclic peptide J324</p><p>(cyclo0,6-[Lys0,Glu6,D-Phe7]BK), an antagonist targeted at the</p><p>bradykinin (BK) B2 receptor, has been investigated using</p><p>experimental and theoretical methods. In order to gain insight</p><p>into the structural requirements essential for BK antagonism, we</p><p>carried out molecular dynamics (MD) simulations using simulated</p><p>annealing as the sampling protocol. Following a free MD</p><p>simulation we performed simulations using nuclear Overhauser</p><p>enhancement (NOE) distance constraints determined by NMR</p><p>experiments. The low-energy structures obtained were compared</p><p>with each other, grouped into families and analyzed with</p><p>respect to the presence of secondary structural elements in their</p><p>backbone. We also introduced new ways of plotting structural</p><p>data for a more comprehensive analysis of large conformational</p><p>sets. Finally, the relationship between characteristic backbone</p><p>conformations and the spatial arrangement of specific</p><p>pharmacophore centers was investigated.</p><p>Abbreviations: BK, bradykinin; DMSO, dimethylsulfoxide; MD,</p><p>molecular dynamics; NOE, nuclear Overhauser effect; RMS, root</p><p>mean square; RMSD, root mean square deviation; SA, simulated</p><p>annealing; SPL, SYBYL programming language; TPPI, time</p><p>proportional phase incrementation.</p><p>Bradykinin (BK) is a peptide hormone with the sequence</p><p>RPPGFSPFR, found in tissues and plasma. It is implicated in</p><p>numerous pathophysiological processes, for example per-</p><p>ipheral pain and inflammation. Owing to these observa-</p><p>tions, great efforts have been undertaken in order to develop</p><p>63</p></li><li><p>BK antagonists as potential therapeutic agents. In the past a</p><p>large number of peptide antagonists has been synthesized by</p><p>various groups (1), while peptide design has gone through</p><p>several iterations.</p><p>In the last three decades several groups have begun</p><p>attempts to describe the conformation of BK and some of its</p><p>analogs. After initial studies focused on the conformation of</p><p>agonistic BK analogs and partial sequences, the situation</p><p>changed with the discovery of the first BK antagonist. The</p><p>goal of these investigations was to describe the bioactive</p><p>conformation(s) of BK antagonists and define the differences</p><p>between the agonists and antagonists. Whereas, in the past,</p><p>conformational analysis was performed by means of con-</p><p>ventional chemical and physical methods, such as thin film</p><p>dialysis, H-exchange, ORD-, CD-, Raman-, ESR- and</p><p>fluorescence spectroscopy, more recently NMR spectro-</p><p>scopy combined with molecular modeling has been increas-</p><p>ingly applied (2).</p><p>The results of those numerous investigations were rather</p><p>insufficient for an accurate description of the bioactive</p><p>conformations. Thus, different conformational shapes have</p><p>been proposed for agonists and antagonists, e.g. random</p><p>conformations (3, 4), conformations stabilized by up to three</p><p>hydrogen bonds (5) or quasicyclic structures (6, 7) have been</p><p>proposed for agonists. As a result of studies on antagonists</p><p>containing different amino acid substitutions at position 7</p><p>and additional replacements at other sequence positions,</p><p>turn structures have been postulated, either in the N-</p><p>terminal sequence (8), the C-terminal sequence (912) or</p><p>both (13, 14).</p><p>The conformation of the linear nonapeptide changes</p><p>through interaction with its receptor, a large membrane-</p><p>bound protein. Two approaches to characterizing the</p><p>receptor-bound conformation have been used considering</p><p>this conformational change and the lack of an isolated and</p><p>crystallized hormone receptor complex, suitable for X-ray</p><p>analysis. Ottleben et al. (15) used a monoclonal antibody to</p><p>mimic the receptor-binding area. The conformation of</p><p>antibody-bound BK was studied after labeling with 13C</p><p>and 15N by NMR spectroscopy. Nevertheless, this model fits</p><p>the binding requirements of the receptor to only a limited</p><p>extent. Thus, the C-terminal Arg is not necessary for</p><p>binding to the antibody, but is essential for binding to the B2</p><p>receptor. Therefore, the conformation found seems to</p><p>correlate only partially with the receptor-bound conforma-</p><p>tion. A second approach was used by Kyle et al. (16) to</p><p>describe the receptor-bound conformation. Based on the</p><p>results of structural homology modeling and computer</p><p>simulation, the authors generated a model of the ligand</p><p>bound to the rat B2 receptor, which, even though it is highly</p><p>speculative, is supported by mutagenesis data. Therefore,</p><p>conformational analysis of the hormonereceptor complex</p><p>remains a great challenge for biochemistry and also a</p><p>prerequisite for the real rational design of potent agonists</p><p>and antagonists.</p><p>Most conformational studies were performed on linear</p><p>agonists and antagonists, i.e. on peptides with high</p><p>conformational flexibility. In some analogs this flexibility</p><p>was reduced through the introduction of constrained</p><p>nonproteinogenic amino acids (17, 18) or by cyclization</p><p>between the N- and C-termini (19, 20). We attempted to</p><p>reduce the conformational flexibility by intramolecular</p><p>cyclization of the side chains. Owing to conformational</p><p>constraints, the solution conformation of a biologically</p><p>active cyclic peptide is expected to be close to the bioactive</p><p>conformation. For this reason we synthesized a series of BK</p><p>antagonists with lactam bridges in either the N- or C-</p><p>terminal part of the molecule. These bridges were built</p><p>between the side chains of Lys and Glu residues added to or</p><p>inserted into the sequence (21). A second series contained</p><p>backbone cyclisized (22) analogs with lactam bridges built</p><p>between peptide bonds modified with aminoalkyl and</p><p>carboxyalkyl residues (23), respectively.</p><p>In spite of the intramolecular cyclization, many of the</p><p>analogs are still too flexible for conformational analysis by</p><p>NMR spectroscopy and exist in solution as an equilibrium of</p><p>many conformers. Few peptides fulfilled both requirements:</p><p>biological potency and sufficiently reduced flexibility. In</p><p>this paper we report the results of our first conformational</p><p>studies on cyclic BK analogs. We started with the analog</p><p>cycloo,6-[Lys0,Glu6,d-Phe7]BK (J324), a first-generation</p><p>cyclic antagonist with a lactam bridge between the side</p><p>chains of a Lys residue that had been introduced as an N-</p><p>terminal extension and Glu residue that replaced the Ser</p><p>residue at position 6 in the native sequence. This antagonist</p><p>shows only modest antagonistic activity in the isolated rat</p><p>uterus assay (pA2 = 5.78 u 0.02). Nevertheless, its activityis higher than that of the corresponding linear peptide (rat</p><p>uterus: no activity; guinea-pig ileum: pA2 = 4.90 u 0.32;21).</p><p>Preliminary studies by means of energy minimization</p><p>calculations and MD simulations of J324, as well as of a</p><p>second antagonist that contains a disulfide bridge in the N-</p><p>terminal region, showed that a b-turn in this segment of the</p><p>peptide appears to be a defining structural feature, although</p><p>it is not the sole determinant of its antagonistic activity (21,</p><p>24).</p><p>Krause et al . Conformational properties of bradykinin B2 receptor agonist</p><p>64 | J. Peptide Res. 55, 2000 / 6371</p></li><li><p>An additional approach towards the elucidation of the</p><p>essential structural requirements for BK antagonism</p><p>focused, not on the backbone conformation, but rather on</p><p>the spatial arrangement of functional groups and their</p><p>specific properties (charge, hydrophobicity, aromaticity,</p><p>etc.). This led to the development of three-dimensional</p><p>pharmacophore models with the help of expert systems such</p><p>as Catalyst (25), and consensus MD simulations. Based on</p><p>these models, several proprietary three-dimensional data-</p><p>bases were screened for new potential BK antagonists (26).</p><p>As a direct result of these studies, a novel class of nonpeptide</p><p>lead structures for BK B2-receptor antagonists has already</p><p>been suggested (Pineda et al., unpublished results).</p><p>Continuing our previous studies on the conformational</p><p>properties of the cyclic BK B2-receptor antagonists synthe-</p><p>sized in one of our laboratories (SR), we report here the</p><p>results of the investigation into the structure of J324 in</p><p>dimethylsulfoxide (DMSO) solution using NMR spectro-</p><p>scopy and both restrained and free MD simulations, always</p><p>assuming that there is only a single preferred conformation.</p><p>The low-energy structures sampled were analyzed using</p><p>novel plots, and the relationship between the resulting</p><p>backbone conformations, which are characteristic for BK</p><p>antagonism, and the specific spatial arrangement of the</p><p>pharmacophore centers is addressed.</p><p>Experimental Procedures</p><p>Samples for NMR spectroscopy were prepared by dissolving</p><p>the HPLC-purified peptide in perdeuterated DMSO without</p><p>any further adjustment of the pH. The one- and two-</p><p>dimensional spectra were recorded on a Brucker AMX500</p><p>instrument at 300K. One-dimensional spectra were recorded</p><p>with 32 000 data points. Data for the TOCSY (27) or NOESY</p><p>(28) experiments were recorded with 4096 points in the</p><p>direct and 1024 points in the indirect dimension with four</p><p>scans. Time-proportional phase incrementation (TPPI) (29)</p><p>was used in the direct dimensions in both cases. Coupling</p><p>constants were extracted from a PE-COSY experiment (30)</p><p>with 4096 3 1024 data points. In all two-dimensional</p><p>spectra the sweep width in both dimensions was</p><p>4545.45 Hz. The different data sets were processed using</p><p>UXNMR xwin-nmr 1.3 on a Silicon Graphics Indy/4600</p><p>workstation.</p><p>The mixing time in the NOESY experiment was 200 ms,</p><p>and the volumes of the NOE cross-peaks were translated</p><p>into distances using a geminal proton pair as an internal</p><p>standard. In the TOCSY experiment an MLEV sequence was</p><p>used to transfer coherence. The mixing time was 62 ms,</p><p>leading to TOCSY as well as ROESY (31) peaks in the</p><p>resulting two-dimensional spectra. This combination of</p><p>through-space and through-bond transfer simplified the</p><p>spin-system assignment as well as the sequential assign-</p><p>ment of the spectra.</p><p>All force field calculations were performed on Silicon</p><p>Graphics (Crimson/R4000) and Evans &amp; Sutherland (ESV3/</p><p>32) graphics workstations operating under IRIX 5.3 and ES/</p><p>os 2.3, respectively. The molecular modeling software sybyl</p><p>V6.26.3 was used throughout the study (32). Calculations</p><p>were performed using the sybyl implementation of the</p><p>amber all-atom force field (33, 34) with default parameters.</p><p>Electrostatic interactions were included in all calculations</p><p>assuming a distance-dependent dielectric constant (e = r)</p><p>and completely ionized groups. This rather simplified</p><p>dielectric model seems to be appropriate for a first approach</p><p>and although it could lead to an overestimation of those</p><p>interactions in the absence of explicit solvent molecules, the</p><p>inclusion of a relatively large number of experimental</p><p>constraints into the calculations is expected to compensate</p><p>for this effect. The starting conformation was built and</p><p>minimized by consecutive steepest descent, conjugate</p><p>gradient (Powell) and quasi-Newton (BFGS) energy mini-</p><p>mization steps until a final root mean square (RMS) gradient</p><p>not larger than 0.05 kcal/mol.A was reached. A simulated</p><p>annealing (SA) MD protocol was used to sample the</p><p>conformational space accessible to the selected molecule</p><p>(NTV ensemble). One hundred and twenty SA cycles</p><p>consisting of a 5 ps high temperature interval at 900 K</p><p>followed by a 5 ps annealing interval in which the</p><p>temperature was decreased exponentially from 900 to</p><p>300 K were carried out during each simulation. An integra-</p><p>tion time step of 1 fs was used. Conformers were sampled at</p><p>the end of each cycle and stored in a database for further</p><p>minimization. The resulting 120 low-temperature confor-</p><p>mers were minimized following the same scheme as used for</p><p>the starting structure.</p><p>These calculations were performed both as free simula-</p><p>tions without any constraints and as restrained simulations</p><p>with NOE distance constraints. For the latter simulations</p><p>158 distance constraints were used. These distances</p><p>correspond to the unequivocally assigned NOE cross-peaks</p><p>of the main conformation of the peptide. Values of 50 and</p><p>1 kcal/mol.A2 were used for the force constant in the</p><p>additional distance range constraint energy terms.</p><p>Low-energy structures, i.e. those within an energy range of</p><p>20 kcal/mol above the lowest energy conformation, were</p><p>compared with each other and clustered into families prior</p><p>Krause et al . Conformational properties of bradykinin B2 receptor agonist</p><p>J. Peptide Res. 55, 2000 / 6371 | 65</p></li><li><p>to analysis. The classification into structure clusters was</p><p>based on the calculation of the RMS deviation (RMSD)</p><p>values for the best superposition of equivalent atomic</p><p>centers in two conformers. Custom-written sybyl program-</p><p>ming language (SPL) scripts allowed a graphical representa-</p><p>tion of RMSD values calculated for the set of sampled</p><p>conformers (clustergraph). The value of the RMSD threshold</p><p>(resolution level) was typically varied from 2.0 to 0.5 A until</p><p>we arrived at a manageable number of families (between 10</p><p>and 20).</p><p>Using this approach, the 120 conformers were grouped</p><p>into structural families according to the position of peptide</p><p>backbone atoms. These families were subsequently ana-</p><p>lyzed to clarify the presence of secondary structural</p><p>elements such as b- (35) and c-turns (36, 37). In addition to</p><p>the analysis of representative structures derived from the</p><p>clustering, we also used a novel approach of plotting the</p><p>structural data of all 120 conformers to arrive at a more</p><p>complete analysis of such large conformational sets.</p><p>Furthermore, we examined the general shape of the back-</p><p>bone and tried to identify characteristic hydrogen bond</p><p>patterns that stabilize the conformation of the molecule.</p><p>Shifting the focus of our investigation towards the role of</p><p>the side chains, we then used the clustergraph representa-</p><p>tions described above to compare the three-dimensional</p><p>arrangement of five specific pharmacophore centers, two...</p></li></ul>


View more >