Vol. 51 No. 1/2004 33–49 QUARTERLY Review

In recent years, a massive effort has been directed towards designing potent and selective antagonists of neurohypophyseal hormones substituted at position 3. Modification of vasopressin at position 3 with 4,4'-biphenylalanine results in pharmacologically inactive analogues. Chemically, this substitution appears to vary only slightly from those previously made by us (1-Nal or 2-Nal), which afforded potent agonists of V(2) receptors. In this situation, it seemed worthwhile to study the structure of the analogues with 4,4'-biphenylalanine (BPhe) at position 3 in aqueous solution using NMR spectroscopy and total conformational analysis. This contribution is part of extensive studies aimed at understanding spatial structures of 3-substituted [Arg(8)]vasopressin analogues of different pharmacological properties. NMR data were used to calculate 3D structures for all the analogues using two methods, EDMC with the ECEPP/3 force field, and molecular dynamic with the simulated annealing (SA) algorithm. The structures obtained by the first method show a better fit between the NMR spectral evidence and the calculation for all the peptides.

]vasopressin (AVP) plays a primary role in the regulation of renal water excretion and a secondary role in the regulation of cardiovascular function in mammals.Agonists and antagonists of this hormone have been widely noted in studies aimed at clarifying the role of endogenous vasopressin in a variety of physiological and pathophysiological conditions.Moreover, one of the agonists of antidiuretic responses to AVP (agonist of V 2 receptors), namely 1-desamino-8-D-arginine vasopressin (dDAVP), is now a drug of choice for the treatment of diabetes insipidus.It is also worth mentioning that antagonists of V 2 receptors have a potential as therapeutic agents for the treatment of hyponatraemia, secondary to the syndrom of inappropriate secretion of the antidiuretic hormone (SIADH) (Langfeldt & Cooley, 2003).
The importance of AVP analogues has prompted many scientists to synthesize thousands of new peptides to establish a structure-activity relationship aimed at the development of either selectively active analogues or those with a sustained effect.However, the design of peptides that are truly selective for V 1a , V 2 and other AVP receptors remains a challenge.
Both NMR and theoretical methods are often very useful in examining the conformations of the hormones and their analogues, they can thus provide good tools in the effort to better correlate the relations between structure and activity of the peptides.
In recent years, a massive effort has been directed towards designing potent and selective antagonists of neurohypophyseal hormones substituted at position 3. Until recently, the Phe group at position 3 was thought to be crucial for producing agonism both towards the V 2 and V 1a receptors (Manning et al., 1997;Hlavacek, 1987).However, this assumption has been questioned by new reports indicating that substitution of the Phe 3 residue by conformationally restricted and bulky aromatic amino acids might result in analogues of interesting pharmacological properties (Manning et al., 1997;Stoev et al., 1999).An additional stimulus to continue these efforts was provided by the results of research carried out in our laboratory, which demonstrated that substitution at position 3 of dDAVP of either the L-3-(2-naphthyl)-alanine (2-Nal) or L-3-(1-naphthyl)-alanine (1-Nal) residues afforded potent agonists of V 2 receptors.In addition, the [Mpa 1 , (1-Nal) 3 , (D-Arg) 8 ]VP analogue was distinguished by its high selectivity (Lammek et al., 1997).These encouraging findings prompted us to design and synthesize three further analogues of vasopressin modified at position 3 with an aromatic residue restricting the conformational freedom of the molecule, namely with 4,4¢-biphenylalanine (BPhe) (Fig. 1).
Furthermore, in one of the peptides the modification was extended by changing the conformation of the Arg 8 residue, and in another, 3-mercapto-3,3-cyclopentamethylenepropionic acid (Cpa) was substituted at position 1 (Scheme 1).These modifications were based on the findings that substitution of the acid enhances the antagonistic efficacy of the analogues (Manning et al., 1987;László et al., 1991), while the change of configuration of Arg 8 evokes antidiuretic activity and/or selectivity of action (Manning et al., 1976).The results of biological experiments showed that modification of [Arg 8 ]AVP at position 3 with BPhe results in pharmacologically inactive analogues (Trzeciak, unpublished).Again, from the chemical standpoint, the substitution with BPhe appears to be only a slight variation of those previously made by us (1-Nal or 2-Nal).It has, however, a dramatic influence on the biological activity of the peptides.In this situation, it seemed worthwhile to study the structures of these analogues in aqueous solution, utilizing 2D NMR spectroscopy supplemented with two different computational methods.The present contribution is a part of extensive studies aimed at understanding the spatial structures of 3-substituted AVP analogues of different pharmacological properties.We believe that the understanding the differences in the structure would contribute to designing peptides with definite pharmacological properties.

NMR MEASUREMENTS
NMR spectra were recorded on a 500 MHz VARIAN and a 600 MHz DMX BRUKER spectrometer.The experiments were carried out in H 2 O/D 2 O (9:1).The sample concentration was approximately 7 mM in 0.5 ml of H 2 O/D 2 O, pH = 4.9 for the [BPhe 3 ]AVP and [BPhe 3 , D-AVP 8 ]VP peptides and pH = 5.1 for the [Cpa 1 , BPhe 3 ]AVP peptide.All spectra were recorded at 298 K except for those for the temperature coefficients of the amide protons chemical shifts, measured over the range of 275-323 K.
The analysis of the residual spin-coupling correlation systems was straightforward, being performed by a combination of the sequential-specific assignment procedure in the TOCSY and the sequential ROE network along the peptide backbone protons (Tables 1-3).The positions of H b and H g protons of Pro 7 for each analogue and the protons of Cpa 1 were found in the 1 H-13 C gHSQC spectra.On the basis of the 1 H-15 N gHSQC spectra, the chemical shifts of e-NH 2 protons of Gln 4 and d-NH 2 of Asn 5 for each peptide and H N of BPhe 3 of [Cpa 1 , BPhe 3 ]AVP were marked.All the ROE cross-peaks for all peptides were picked up on the ROESY spectra with mixing time of 250 ms.The value of the temperature dependence of H N proton chemical shifts (Dd/DT) was calculated from the 1D spectra recorded at 275, 283, 293, 303, 313 and 323 K.The coupling constants between H a and H N ( 3 J HNHa ) were found in the ACT-ct-COSY (KoŸmiñski, 1998) and 1D NMR spectra.The error delimitation of the values of coupling constants was 0.1-0.5 Hz.

CONFORMATIONAL CALCULATIONS
The conformational space of each peptide was determined using two methods, EDMC (Ripoll & Scheraga, 1988) with the ECEPP/3 force field (Némethy et al., 1992) and molecular dynamic with the simulated annealing (SA) algorithm.
In the first method, the applied algorithm consists of the following three steps: (i) search of the conformational space in order to find conformations with reasonably low energy, (ii) simulation of the NOE spectrum and vicinal coupling constants for each of the low energy conformations, and (iii) determining the statistical weights of the conformations, by means of the maximum entropy method in order to obtain the best fit of the averaged NOE intensities and coupling constants to the experimental quantities (Groth et al., 1999).
In the second method, the structure calculation was performed using the X-PLOR program (Brünger, 1992).The algorithm SA was used for each peptide.According to the NMR data, the inter-proton distances were calculated by the CALIBA algorithm and the torsion angles were computed by the HABAS algorithm of the DYANA program (Güntert & Wüthrich, 1991) on the basis of the Bystrov-Karplus equation (Bystrov, 1976).
The b-turn types and positions detected in the calculated conformations were defined according to Lewis et al. (1973).For displaying and analyzing the three-dimensional structure of vasopressin analogues the molecular graphics MOLMOL program (Koradi et al., 1996) was used.The hydrogen bonds were calculated using the HBPLUS (McDonald & Thornton, 1994) program (the hydrogen-acceptor distance (H-A) criterion was £2.5 C, the donor-acceptor (D-A) distance criterion was £3.9 C and the donor-hydrogen-acceptor angle (DHA) criterion was ³90°).

THE EDMC AND ANALYZE CALCULATIONS
A global conformational search of the peptides studied was carried out using the Electrostatically Driven Monte Carlo (EDMC) method (Ripoll & Scheraga, 1988) with the ECEPP/3 force field (Némethy et al., 1992), which assumes rigid valence geometry.The force field included a hydration contribution, which was evaluated in the SRFOPT solvent-accessible surface model (Vila et al., 1991), whose parameters pertain to solvation by water.According to the NMR data, the chirality of all C a atoms (except that of residue Arg 8 of the peptide [BPhe 3 , D-Arg]VP) was fixed to the L configuration.A total of 3000 energy-minimized conformations were generated for each peptide.The working temperature was 1000 K.The conformations were subsequently clustered into families using the minimum-variance algorithm (Späth, 1980).The root mean square deviation (RMSD) between heavy atoms at the optimal superposition was taken as the measure of the distance between conformations, and a cut-off value of 1.4 C was used to separate the families to afford 981, 414 and 783 families of conformations for [BPhe 3 ]AVP, [BPhe 3 , D-Arg 8 ]VP and [Cpa 1 , BPhe 3 ]AVP, respectively.
In the next step, for the lowest energy conformation of each family, a NOESY spectrum and vicinal constants coupling, 3 J HNHa , were calculated by using the MORASS (Meadows et al., 1994;Post et al., 1990) program.This program solves the system of Bloch differential equations (Massefski & Bolton, 1985) for the cross-relaxation of a system of interacting proton spins.The vicinal couplings, 3 J HNHa , were calculated from the empirical Bystrov-Karplus relationship (Bystrov, 1976).The NOE effects were generated using a correlation time of 0.45 ms (Yu et al., 1992), mix-ing time of 250 ms and cut-off value of 6 C. The weight of the coupling constant term was 0.5 in the minimized sum and the Marquardt convergence criterion (Marquardt, 1963) equal to 10 -5 was used.The entropy factor, a = 0.2, was used for all peptides.The populations of the families were determined by fitting a linear combination of the generated spectra and coupling values to the experimental data.

X-PLOR CALCULATIONS
The solution structures of the major species of each peptide studied were computed using the experimental: ROE inter-proton crosspeaks and vicinal coupling constants 3 J HNHa .According to the NMR data, the inter-proton distances were calculated by the CALIBA algorithm of the DYANA program (Güntert & Wüthrich, 1991), which performs a standard calibration of the current peaks using three different calibration classes: one for ROE assigned to backbone protons, one for ROE assigned to the more flexible side-chain protons and one for ROE assigned to methyl groups.In the next step, by using the macro GRIDSEARCH with the command "distance modify", irrelevant constraints (constraints that involve fixed distances and constraints which cannot be violated) were removed, retaining maximally one distance limit for each atom pair and introducing corrections for constraints with diastereotopic substituents for which stereospecific assignments are not available (Wüthrich et al., 1983;Güntert et al., 1991).Torsion angles were calculated by the HABAS algorithm of the DYANA program (Güntert & Wüthrich, 1991), on the basis of the Bystrov-Karplus equation (Bystrov, 1976).In the next step, the standard modules of the XPLOR program (Brünger, 1992) were used.The simulating annealing (SA) algorithm was applied for each peptide.The geometry of the peptide groups (all trans; f = 500 kcal/mol • rad 2 ) was kept fixed.For each peptide, the chirality of all C a atoms, except for the arginine residue in [BPhe 3 , D-Arg]VP, were fixed to L. A simulated annealing protocol incorporating ambiguous distance restraints and floating chirality was used.Force constants f = 50 kcal/mol • C 2 and f = 5 kcal/mol • rad 2 for the distance restraints and the dihedral angle restraints, respectively, were used during the simulation.The calculations were carried out using the CHARMM force field (Brooks et al., 1993) in vacuo.Simulated annealing started from the template coordinate set.For each molecule 300 cycles of SA were carried out.Electrostatic interactions and energy of hydrogen bonds were not directly included.In each cycle, the molecule was heated to 2000 K and equilibrated at this temperature for 40 ps, then cooled down to 1000 K for 30 ps and finally cooled from 1000 K to 100 K for 15 ps.Finally, 500 iterations of energy minimization with the use of the Powell algorithm were performed.As a result, 300 energy-minimized conformations were obtained.Then, the sets of the final conformations were clustered (using the minimal-tree algorithm) for each peptide.The RMSD of all heavy atoms at the optimum superposition was taken as the measure of the distance between conformations, and the cut-off values of 2.4 C for all peptides were used to separate the families.

NMR RESULTS
In this paper, the conformations of three analogues of vasopressin in aqueous solution were investigated by two-dimensional NMR spectroscopy.In particular, the influence of the substitution of Cys 1 , Phe 3 and Arg 8 with 3-mercapto-3,3-cyclopentamethylenepropionic acid, 4,4¢-biphenylalanine and D-Arg, respec-tively, on the conformation and flexibility of the peptide was examined.
The observable NMR parameters are averaged over the entire conformational ensemble, but the characteristic NOE connectivities and temperature coefficients or coupling constants, 3 J HNHa , which are diagnostic of reverse turns, are sufficient to identify even quite small populations of these structures (Dyson et al., 1988).The sequential H a (i) -H N (i + 1) NOE connectivity is always observed and is mostly very strong.Much weaker H b (i) -H N (i + 1) connectivities are often found.Sequential H N (i) -H N (i + 1) NOE connectivities are seen only in those regions of peptides which preferentially adopt folded conformations.Intraresidual H a (i) -H N (i) and H b (i) -H N (i) NOE connectivities are also present for most peptides (Dyson et al., 1988).In unfolded peptides, the H a (i) -H N (i) connectivity is much weaker than the sequential H a (i) -H N (i + 1) connectivity, despite the finding that the maximum value of the H a (i) -H N (i) distance is 2.8 C (Wüthrich et al., 1984).
Figure 2a shows the ROE effect patterns and vicinal couplings values, 3 J HNHa , for the [BPhe 3 ]AVP peptide.The presence of H N (4) -H N ( 5) and H N (5) -H N (6) cross-peaks suggests that the b-turn is at position 4,5.The cross-peaks H a (i) -H N (i + 1) indicate the presence of b-turns at positions 2,3; 3,4; 4,5 and 7,8.Additional cross-peaks H a (4) -H N (6) confirms the b-turn at position 4,5.The small value of the 3 J HNHa coupling of residue 4 (3.9Hz) and the higher value measured for residue 5 (8.5 Hz) suggest the presence of a b-turn at the 4,5 position.The small value of the temperature coefficient of the H N of Cys 6 may be caused by the hydrogen bonding of this proton to oxygen atom of the carbonyl group of BPhe 3 (b-turn) or to the oxygen atom of the carbonyl group of Gln 4 (g-turn).
Figure 2b summarizes the ROE patterns and vicinal couplings values 3 J HNHa for peptide [BPhe 3 , D-Arg]VP.In this peptide, the H a (i) -H N (i + 1) and H b (i) -H N (i + 1) crosspeaks suggest that the b-turn at position 3,4 is populated.On the basis of the 3 J HNHa coupling constants of residues 4 and 5 (3.8 Hz and 8.2 Hz, respectively) one can suggest the presence of a b-turn at position 4,5.The low temperature coefficient of the H N of Gln 4 may be caused by the hydrogen bonding of this proton either to the oxygen atom of the carbonyl group of Cys 1 (b-turn) or to that of the carbonyl group of Phe 2 (g-turn).A small coefficient was also found for the signal of the H N proton of Cys 6 , which may form a hydrogen bond to the oxygen atom of the carbonyl group of either BPhe 3 (b-turn) or Gln 4 (g-turn).
For the [Cpa 1 , BPhe 3 ]AVP analogue, a small number of ROE cross-peaks was noted.The H N (i) -H N (i + 1) cross-peaks were missing.The cross-peaks of H a (4) -H N ( 5) and H a (5) -H N (6) seem to confirm the presence of the b-turn at position 4,5.The magnitudes of the 3 J HNHa couplings do not permit confirmation of the occurrence of b-turns (Fig. 2c).Large temperature coefficients of the H N protons suggest an absence of hydrogen bonds.
For the [BPhe 3 ]AVP and [BPhe 3 , D-Arg8 ]VP analogues ROE cross-peaks between Cys 1 and Cys 6 residues were observed.The H a Cys 1 proton interacted with both H b proton sites of the Cys 6 residues, while the H a Cys 6 proton did not show any correlation with H b Cys 1 .This result unambiguously revealed the preference of the conformation in which the S-S dihedral angle was confined to a negative value of approximately -90° (Schmidt et al., 1991).

ANALYSIS OF THE STRUCTURES CALCULATED WITH EDMC AND ANALYZE
Tables 4, 5 and 6 summarize the statistical weights of the lowest energy conformations of each analogue obtained with EDMC and ANALYZE.Both the measured and computed vicinal couplings,3 J HNHa , of the peptides are shown in Table 7.
The In the peptide [Cpa 1 , BPhe 3 ]AVP, a type I b-turn appears at position 2,3 in almost all obtained conformations.An inverse g-turn at position 5 was found in the conformation with the highest statistical weight.The hydrogen bond between the H N proton of Arg 8 and the carbonyl group of Cys 6 is populated.A superposition of the conformations with the highest statistical weights for each peptide is presented in Fig. 3.The [Cpa 1 , BPhe 3 ]AVP peptide reveals the smallest structural difference within this part of the molecule, with RMSD 1-6 of 0.963 C for C a .

ANALYSIS OF THE STRUCTURES CALCULATED USING THE X-PLOR PROGRAM
The final conclusion as to the secondary-structure elements present in each peptide will be drawn based on the calculations using inter-proton distances and torsion angles as constraints.
The lowest energy conformation of peptide [BPhe 3 ]AVP is stabilized by neither b-turns nor hydrogen bonds.In the case of the other structures of this peptide, the b-turn at position 3,4 appears most frequently.Two calculated conformations have a g-turn at position 3, which is stabilized by the hydrogen bond between the H N proton of Gln 4 and the carbonyl group of Tyr 2 .
In the peptide [BPhe 3 , D-Arg 8 ]VP a b-turn of type VII at position 4,5 was found for the lowest energy conformation.One obtained conformation possesses inverse g-turns at positions 3 and 5.The hydrogen bond between H N proton of Cys 6 and the carbonyl group of Gln 4 stabilizes the inverse g-turn at position 5.In all the analogues studied here, the largest variability of the backbone conformation is seen for the disulfide bridge, the Tyr 2 and BPhe 3 residues and the non-cyclic tail.A superposition of the lowest energy conformations for each peptide is presented in Fig. 4.

COMPARISON OF THE RESULTS OBTAINED BY THE TWO CALCULATION METHODS
Figure 5 shows a comparison of the lowest energy conformations of each peptide obtained by the two different approaches, which reveals large structural differences with show a better fit between the NMR spectral evidence and the calculation for all the peptides.This is probably due to the fact that the computational method takes into account the conformational equilibrium of the peptides in solution as well as two parameters of considerable importance for short-chained peptides, namely the mixing and the correlation times.
It is know that cyclic peptides tend to be more structured than linear ones, but they seldom have a single rigid conformation (Petsko, 1996).The numerous conformations in a dynamic exchange produce averaged NMR parameters and it is important that all available data be examined with caution (Walse et al., 1998).In connection with this fact, the EDMC method with the ECEPP/3 force field seems to be more suitable for short peptides, such as analogues of vasopressin.

CONCLUSION
Conformational studies of the structure and activity of vasopressin and its analogues have provided information on which of the functional groups in the sequence are responsible for the interaction with the cellular membrane receptors.It is know that positions 1, 2,  3 and 8 are crucial for antidiuretic activity, whereas the pressor activity is sensitive to changes at positions 1, 2, 8 and 9 (Manning et al., 1987).The results of theoretical considerations on the interactions of AVP with lipids showed that only the side-chains of the Phe 3 , Gln 4 and Pro 7 residues and the disulfide bridge interact with the phospholipid membrane, whilst the side-chains of Tyr 2 , Asn 5 , Arg 8 and Gly 9 -NH 2 interact with the aqueous phase (Grzonka et al., 1991).Additionally, a typical relaxed complex of V2R/AVP was generated by using modelling and computational methods (Czaplewski et al., 1998).The results of that study showed that only the cyclic part vasopressin interacts with helices of the receptor, whilst the C-terminus is characterized by a considerable mobility.
The interaction of vasopressin with its receptors is determined by the correct orientation of some side-chains, including the side-chains of the residues in position 2 and 3.
In vasopressin, the orientation of Tyr 2 and Phe 3 side-chains is stabilized by a parallel p-p interaction between aromatic rings of these residues.Consequently, the side-chains of both aromatic residues extend away from the cyclic moiety.The greater hydrophobic properties of BPhe than Phe may be the reason for the change of interaction between the side-chains of the two neighbouring residues.As a result, the analogues with BPhe in position 3 are inactive, despite the fact that both aromatic residues lie outside the pressin ring as in vasopressin.
Conformational studies of native vasopressin in water (Hempel, 1987;Hruby & Lebl, 1987) and MeSO-d 6 (Schmidt et al., 1991) have shown that its dominant structural features are b-turns within the Tyr 2 -Asn 5 region.Similar results were obtained by computer-aided simulations of the [Arg 8 ]AVP molecule in vacuo (Liwo et al., 1995;KaŸmierkiewicz et al., 1997).Theoretical considerations revealed b-turns at positions 3,4 and 4,5.Also X-ray investigations of pressinoic acid (Langs et al., 1986) revealed type II' and type I b-turns at positions 3,4 and 4,5, respectively.In the peptides studied by us, b-turns generally occur at a position between the second and fifth residue and at position 7,8.The non-cyclic part of each analogue is characterized by increased flexibility.Among the conformations of the presented peptides, the lowest energy conformer of [BPhe 3 ]AVP obtained with X-PLOR reveals a similar backbone and side-chain conformation of Gln 4 to the crystal structure of pressinoic acid.Furthermore, the Asn 5 side-chain lies over the molecule as in the case of pressinoic acid, in contrast to the Asn 5 in vasopressin which extends away from the cyclic part of the peptide (Skala et al., 1984).
Analysis of the structures obtained with EDMC/ANALYZE reveals that the [BPhe 3 ]AVP and [BPhe 3 , D-Arg 8 ] peptides contain b-turns of type I at position 4,5 in most conformations, similar to pressinoic acid.
the ROESY spectra of the [BPhe 3 , D-Arg H d Pro 7 indicates the trans geometry of this peptide bond.In the case of izomerization across a X-Pro peptide bond, the signals of C b and C g of Pro are found at 30.5 ± 0.6 p.p.m. and at 25.1 ± 0.5 for the trans and at 32.2 ± 0.4 p.p.m. and 23.4 ± 0.3 for the cis isomer (Dorman, 1973).The carbon chemical shifts of C b and C g of Pro for all the peptides studied here indicate the trans Cys 6 -Pro 7 peptide bond (Tables lowest energy conformer of the [BPhe 3 ]AVP peptide forms b-turns at positions 3,4; 4,5 and 7,8 and a g-turn at position 9.This g-turn is stabilized by a hydrogen bond between the H N proton of C-NH 2 9 and the oxygen atom of the carbonyl group of Arg 8 .Two obtained conformations contain an inverse gturn at position 5, which is stabilized by appropriate hydrogen bonds.The conformers found for the peptide [BPhe 3 , D-Arg 8 ]VP contain mainly b-turns at positions 2,3; 3,4 and 4,5.Two of the obtained conformations have an inverse g-turn at either position 2 or 5.The conformer with the 40 E. Trzepałka and others 2004 Table 2. Proton and carbon chemical shifts [p.p.m.] and the amide proton temperature coefficients [ppb/K] of the [BPhe highest statistical weight forms a hydrogen bond between the H N proton of Cys 6 and the side-chain of Asn 5 .
in water at 25°C n.o.-not observedIn the case of peptide [Cpa 1 , BPhe 3 ]AVP, the lowest energy conformation has b-turns at positions 2,3 and 3,4.The majority of the calculated conformations form hydrogen bonds either between the H N proton of Arg 8 and the carbonyl group of Cys 6 or the dH N proton of Asn 5 and the carbonyl group of Gln 4 .No g-turn was observed in those conformations.
RMSD 1-6 = 1.483, 1.747 and 1.957 C for C a atoms, for peptides [BPhe 3 ]AVP, [BPhe 3 , D-Arg 8 ]VP and [Cpa 1 , BPhe 3 ]AVP, respectively.These differences resulted from the specificity of the methods themselves.The structures obtained by the EDMC/ANALYZE 42 E. Trzepałka and others 2004 Table 4. Statistical weights of the lowest energy conformations (with weights above 4%) obtained using EDMC and ANALYZE and the families of the low energy X-PLOR conformations from each family -handed disulfide bridge; R, right-handed disulfide bridge; g*, inverse g-turn