Conformation-Activity Relationships of Cyclic Dermorphin Analogues BRIAN C. WILKES and PETER W. SCHILLER Laboratory of Chemical Biology and Peptide Research, Clinical Research Institute of Montreal, Montreal, Quebec, Canada H2W 1R7
SYNOPSIS
A theoretical conformational analysis (molecular mechanics study) of nine cyclic tetrapeptides, structurally related to the highly p-receptor-selective dermorphin analogue H-Tyr-D-Om-Phe-Asp -NH,, was performed. These compounds display considerable
u
diversity in their preceptor affinity and selectivity. A systematic search and subsequent energy minimization in absence of the exocycIic Tyr' residue and Phe3 side chain revealed the constrained nature of the 11-13-membered ring structures contained in these analogues. No more than four low-energy conformers (within 2 kcal/mol of the lowest energy conformation) were found in each case. After attachment of the Tyr' moiety and Phe:' side chain to the "bare" low-energy ring structures, a systematic search and energy minimization of these exocyclic moieties resulted in a limited number of low-energy confomationd families for dl compounds. Five analogues with high preceptor affinityH-Tyr-D- Om-Phe-Asp -NH, ,H-Tyr-D-Om-Phe-D-Asp -NH, ,
u
U
H-Tyr-D-Asp-Phe-Om -NH,, H-Tyr-D-Asp-Phe-A,bu-NH, (A, bu: a, y-diaminobutyric I
acid) and H-Tyr-D-Cys-Phe-Cys-NH,
u
-all showed a tilted stacking interaction between the Tyr' and Phe3 aromatic rings in the lowest or second lowest energy conformation found. The same kind of stacking was not possible in low-energy conformers of the four analogues with poor affinity for the p-receptor
[ H-Tyr-L-Om-PheAsp-NH, , u H-Tyr-D- Om-D-Phe-Asp-NH, ,
H-Tyr-D-Om-Phe(NMe)-Asp-NH, [Phe(NMe): Nu-methylphenylalanine],and I
I
H-Tyr-D-Om-Phg-Asp -NH, (Phg: phenylglycine) } . J
It is concluded that a tilted stacking arrangement of the two aromatic rings may represent a structural requirement for high preceptor affinity of the examined cyclic dermorphin analogues. T: 1990 John Wiley & Sons, Inc. CCC C~S-3525/90/010089-07 $04.00 Biopolymers, Vol. 29,89-95 (1990)
Abbreviations: Symbols and abbreviations are in accord with recommendations of the IUPAC-IUB Commi&on on Bio. chemical Nomenclature. (1984) Biochern. J . 219, 345-373.
INTRODUCTION
Since the discovery of the enkephalins [H-Tyr-GlyGly-Phe-Mewor Leu)-OH] numerous conformational studies On these pentapeptides have been performed with the goal of elucidating their bioac89
90
WILKES AND SCHILLEH
tive conformation(s) (for a review, see Ref. 1).These conformational studies made use of a variety of different methods, including theoretical conformational analysis, x-ray diffraction analysis, and various spectroscopic techniques, and led to the overall conclusion that the natural enkephalins can assume a number of both folded and extended conformations of comparatively low energy. Due to this structural flexibility, conclusive information about the bioactive (receptor-bound) conformation cannot be obtained from conformational analysis of the enkephalins and of other linear opioid peptides. The situation is further complicated by the demonstrated existence of a t least three different types of opioid receptors ( p , 3, and K), each of which has distinct structural and conformational requirements.2 The enkephalins interact preferentially with &receptors but also bind to p receptors, albeit with somewhat lower affinity. The lack of receptor selectivity of these linear pentapeptides is apparently due to their conformational flexibility, which permits structural adaptation to the receptor topography of both the 3 and the p receptor. In recent years, several investigators have prepared conformationally constrained enkephalin analogues in an attempt to improve the receptor selectivity of these peptides. Conformational restriction through various peptide cyclizations (side chain to end group or side chain to side chain) has been particularly successful. The first active cyclic enkephalin analogues were obtained through substitution of a D-diamino acid (Daa) in the 2-position and cyclization between the Daa side-chain amino group and the C-terminal carboxyl function.',' The resulting compounds, H-Tyr-cyclo[-D-Daa-Gly-Phe-Leu-] [Daa = a, pdiaminopropionic acid(A2pr), a ,y-diaminobutyric acid(A,bu), Om, or Lys], not only were potent and enzyme resistant, but also showed considerable preference for p receptors over 3 receptors. Several partial retro-inverso analogues of H-Tyr-cyclo[-DA,bu-Gly-Phe-Leu-] were also found to be very potent, extremely stable, and quite p selective."' An example of a side chain to side chain cyclized enkephalin analogue is [D- Pen2, D(or ~ ) - P e n ~ ] e n -
side chain bridging between appropriate residues substituted in positions 2 (D-configuration) and 4 has recently been Several cyclic lactam tetrapeptide amides of this type with small ring structures (12 or 13 membered) and a cystinebridged tetrapeptide amide (11-membered ring) showed high preference for p receptors over 6 receptors. Among these compounds, the highest preceptor selectivity was displayed by the analogue H-TJT-D-Om-Phe-Asp-NH,, the most selec1_1
tive cyclic opioid peptide analogue with p-agonist properties reported to date. Several studies on the conformational behavior of various cyclic opioid peptide analogues have been carried out (for a review, see Ref. lo). The major focus has been on the moderately p-selective prototype compound H-Tyr-cyclo[-D-A,bu-GlyPhe-Leu-]."-" These studies have not yet led to a consensus concerning a possible unique bioactive conformation of this analogue a t the p receptor. The most important finding of these endeavors was the realization that the 14-membered ring structure in H-Tyr-cycle[-~-A,bu-Gly-Phe-Leu-] still retains some flexibility, and that various intramolecular hydrogen bonds are constantly formed, broken, and reformed again, as shown most conclusively in a molecular dynamics study.I2 We recently performed a theoretical conformational analysis of the highly p-selective cyclic dermorphin analogue H-TV-D-Om-Phe-Asp-NH 2.14
-
The lowest energy conformer found in this molecular mechanics study was characterized by a lack of linear transannular hydrogen bonds and by a tilted stacking interaction of the two aromatic rings. In the present study we extend this type of analysis to several cyclic dermorphin analogues structurally related to H-Tyr-D-Om-Phe-Asp-NH,, and show-
u
ing great diversity with respect to their affinity for p- and 8-opioid receptors. We report a common low-energy conformation for analogues with high preceptor affinity, and propose that this conformation may be of relevance as the bioactive conformation a t the p receptor.
J
kephalin ('Pen: penicillamine), a compound displaying greatly improved 3-receptor selectivity? Whereas the enkephalins contain a phenylalanine residue in the 4-position of the peptide sequence, the opioid peptides dermorphin and pcasomorphin (morphiceptin) have that same residue in the 3-position. A series of cyclic dermorphin-(l-4) tetrapeptide analogues resulting from side chain to
METHODOLOGY
All calculations were performed using the molecular modeling software SYBYL (Tripos Associates, St. Louis, MO) on a VAX 11/750 mainframe using VMS version 4.2. Molecules were viewed with an Evans and Sutherland PS 300 computer graphics
CONFORMATION-ACTIVITYRELATIONSHIPS
display terminal. A Hewlett Packard H P 7475 plotter was used for the preparation of the figures. A comprehensive theoretical analysis of the parent compound H-TP-D- Om-Phe-Asp-NH, has been
u
reported e1~ewhere.l~The methodological approach taken in the latter study was used with some minor modifications to determine the lowenergy conformations of the eight analogues examined in the present study. The first step in ths stepwise procedure was the construction of the "bare" ring structure for each compound. Only atoms directly attached to the ring, with associated hydrogen atoms, were included in this part of the analysis. This approach has been used by other in~estigators'~ and was taken to permit the identification of the greatest possible number of solutions in the conformational search to be undertaken in the second step of the ana!ysis procedure. Each "crude" ring structure was minimized prior to the conformational search using the energy minimization program MINIM. The steepest descent approach was used and the potential energies were calculated from
3; =
WstrEstr
+
+
YorEtor
WangEmg
+
WvdwEvdw
where the Ws represent weight constants and the E s represent the energy terms for the bondstretching (str) energy, angle-bending (ang) energy, torsional (tor) energy, and van der Waal's (vdw) contact energy (including hydrogen-bonding energy). The standard SYBYL force field16 was employed. Once a reasonable starting geometry was obtained, the ring structures could be subjected to a systematic search for allowed conformations. The second step of the procedure consisted of a systematic conformational grid search through all conformational space, using the software program SEARCH.l7 This search program systematically checks for unfavorable vdw contacts around the nonbonded atoms by scanning all possible torsionai angles around all the rotatable bonds. The two criteria for an allowed conformation were the lack of unfavorable vdw interactions and the requirement for ring closure. In order to allow for thermal vibration, a vdw scaling factor of 0.80 was used for nonbonded atoms. A vdw scaling factor of 0.70 was used for 1,4-interactions1' and the vdw factor used for hydrogen-bonding interactions was 0.40. All amide bonds were held trans and planar. One of the bonds was chosen as the ring-closure bond and the remaining bonds were surveyed over a 30" grid over all space. An allowed conformation
91
was obtained if in a structure without unfavorable vdw contacts the ring closure bond could form within 0.25 of the length and within 15 degrees of the torsion angle of a normal carbon-carbon bond. In the case of the cystine-bridged analogue H-Tyr-D-Cys-Phe-Cys-NH,,the disulfide bond was
w A.
chosen as nng closure bond and the ring closure tolerance was 0.4 For each of the ring structures examined in this study, the search program was allowed to scan between lo9 and 10" possible combinations and usually required between 20 and 50 h central processing unit (CPU) time. For the bare ring structure of H-Tyr-u- Om-Phe(NMe)-Asp-NH
I J .
[Phe(NMe): N"-methylphenylalanine), only four solutions were found. For the other eight compounds, the ring search resulted in between 10 and 23 solutions. Each of the allowed ring structures was then minimized using the MINIM program, allowing all the atoms to relax, including those contained in the amide moieties. The solutions for each ring structure were ranked in order of increasing energy, and only those structures within 2 kcal/mol of the minimum energy found were retained for further study. In the third step the exocyclic Tyr' residue and the Phe" or Phg" (Phg: phenylglycine) side chain were linked to each low-energy ring conformer obtained in step 2. The exocyclic amide bonds were held trans and planar, and the remaining six or seven rotatable bonds outside the ring structure were surveyed over a 30" grid in order to determine allowed configurations for the Tyr' residue and Phe3 (Phg') side chain. In each case the program scanned approximately lo6 possible conformations and between 3000 and 10,000 solutions were obtained. The energies of these conformers were calculated, and the resulting solutions were grouped into two to ten low-energy families for each ring structure searched. The fourth step involved picking the lowest energy conformer in each low-energy conformational family and subjecting this structure to extensive energy minimization as described above. The obtained conformers could then be ranked in order of increasing energy. This entire procedure required between 100 and 200 h of CPU time for each compound. RESULTS AND DISCUSSION
The compounds studied and their p- and 6-opioid receptor affinities are listed in Table I. Opioid-
92
WILKES AND SCHILLER
Table I p- and 6-Opioid Receptor Affinities of Cyclic Dermorphin Analogues (taken from Ref. 9)
H-Tyr-D-Om-Phe-Asp-NH,
I
I
2200 k 65
10.4 & 3.7
213
J
> 12.4
I1
H-Tyr-L-Om-Phe-Asp-NH,
4830 & 610
> 60,000
I11
H-Tyr-D-Om-D-Phe-Asp-NH,
3040 5 470
> 380,000
IV
H-Tyr-D-Om-Phe-D-Asp-NH,
21.7 f 3.2
422 5 17
V
H-Tyr-D-Om-Phe(NMe)-Asp-NH,
3570 f 500
28,100 f 2500
7.87
VI
H-Tyr-D-Om-Phg-Asp-NH,
445 f 92
3570 f 370
8.02
VII
H-Tyr-D-Asp-Phe-Om-NH,
9.55 & 2.52
1320
150
138
24.8 i 1.1
4170 f 430
168
11.0 f 0.3
373 k 63
u
-
u
> 125
19.4
I
VIII
H-Tyr-D-Asp-Phe-A,bu-NH, I
IX
H-Tyr-D-Cys-Phe-Cys-NH, "Mean of three determinations
33.9
I
I
SEM.
receptor affinities had been determined by displacement of the radioligands [3H]H-Tyr-~-AlaGly-Phe(NMe)-Gly-ol (p-selective) and [3H]H-TyrDSer-Gly-Phe-Leu-Thr-OH (&selective) from rat brain membrane preparations, as de~cribed.~ In comparison with linear opioid peptides, the conformational flexibility of the cyclic dermorphin analogues studied is greatly reduced due to the considerable rigidity of the relatively small ring structures (11-13 membered) they contain. For each compound lo9 to 10" possible conformations of the bare ring structure were scanned, yet only a very limited number (no more than 23) of allowed conformations was obtained. In each case, the number of low-energy ring conformers within 2 kcal/mol of the lowest energy ring structure found was no more than four. For the ring structure in H-Tyr-DOrn-Phe(NMe)-Asp -NH, (V) I
only four allowed conformations were found and only two of these were of sufficiently low energy to warrant further study. This result indicates that the latter ring structure is somewhat more rigid than the other ring structures studied due to the additional conformational constraint produced by the N-methyl group. The ring structures of the compounds described in the present paper are considerably more rigid than those in some other cyclic
opioid peptide analogues containing larger rings that have been studied to date. For example, in the case of the &selective cyclic enkephalin analogue H-Tyr-D-Pen-Gly-Phe-~-Pen-OH~, sixteen conformations of the 1Cmembered ring within 2 kcal/mol of the lowest energy ring structure were obtained by using the same a p p r o a ~ h . ' ~ The gross topological features of the low-energy ring conformations of the cyclic parent peptide H-Tyr-D- Om-Phe-Asp-NH, (I) and of the eight
u
analogues (11-IX) studied were all quite similar. The structural characteristics of the low-energy ring conformers of parent peptide I have been previously discussed in detail.I4 As in the case of compound I, the low-energy ring structures of analogues 11-IX were all fairly symmetrical and relatively flat when viewed from the side. The amide bonds within the rings were in general more or less perpendicular to the ring plane and no strong (linear), transannular hydrogen bonds were observed in any case. When the conformational analysis was extended to include the Tyr' residue and the Phe3 side chain attached to the various low-energy ring structures, it was found that in all cases the latter moieties enjoy a considerable amount of orientational freedom around the exocylic rotatable bonds. As expected, this was particularly the case for the Tyr' moiety. However, only a relatively limited number
CONFORMATION-ACTIVITYRELATIONSHIPS
93
of low-energy families was observed for all compounds. Four of the analogues-
the two aromatic rings in their lowest energy conformation (Figure 1). In the lowest energy conformation of the compound H-TP-D- Cys-Phe-Cys-
-
NH, (IXa) the Tyr' moiety is oriented away from the rest of the molecule and does not engage in any intramolecular interactions. However, in the second lowest energy conformation of this compound (IXb) (0.2 kcal/mol higher in energy than the lowest one) a tilted stacking interaction between the two aromatic rings is again observed. A recent survey of interactions between aromatic rings in proteins based on crystal structure data revealed that this staggered stacked interaction with tilting of the two rings is the most common and, presumably, the energetically most favorable, while fully stacked interactions are almost never observed.2" Interestingly, the five cyclic dermorphin analogues (I, IV, VII, VIII, and IX) showing a tilted stacking arrangement of the two aromatic rings in their lowest energy conformations all display high affinity for the p-opioid receptor and weak but somewhat variable affinity for the 6 receptor.
H-Tyr-D-Orn-Phe-Asp-NH, (I),
u
H-Tyr-D- Om-Phe-D-Asp-NH, (IV), H-Tyr-D-AspPhe-Om-NH, (VII),
u
and H-Tyr-D-Asp-Phe-A,bu -NH, (VIII) I
-showed
a tilted stacking arrangement between
u
Ip'
'm
IT.
PT
Figure 1. Lowest-energy conformations of H-Tyr-D-Om-Phe-Asp -NH, (I),
u
P Figure 2. Lowest energy conformations of
H-Tyr-D-Om-Phe-D-Asp -NH, (IV) ,
H-TY-L- Om-Phe-Asp -NH, (11) , I
H-Tyr-D- Asp-Phe-Om -NH, (VII),
H-Tyr-D-Om-D-Phe-Asp -NH, (111) ,
I
H-Tyr-D-Asp-Phe-A,bu -NH, (VIII) ,
H-Tyr-D-Om-Phe(NMe)-Asp -NH, (V) ,
1
and
and H-Tyr-D-Cys-Phe-Cys -NH, ( IXa and IXb) .
H-Tyr-D-Om-Phg-Asp -NH, (VI) .
u
94
WILKES AND SCHILLER
On the other hand, the cyclic analogues displaying weak affinity for p receptors (compounds 11, 111, V, and V1) were unable to adopt this tilted stacking interaction of the two aromatic rings in their low-energy conformations (Figure 2). In the case of analogues I1 and 111, this is due to the fact that configurational inversion a t the 2- or 3-position of the peptide sequence places the side chains of Tyr' and Phe3 on opposite sides of the peptide ring structure, thereby preventing an intramolecular interaction between the two aromatic rings. Examination of the low-energy conformations obtained for the weakly active analogue H-Tyr-D-Om-Phe(NMe)-Asp-NH, (V) I
revealed that aromatic ring stacking is not possible because of steric interference by the bulky Nmethyl group introduced in the 3-position of the peptide backbone. An almost fully stacked parallel arrangement of the two aromatic rings is observed in the lowest energy conformer of the weak y agonist H-Tyr-D-Om-Phg-Asp-NH, (VI), in contrast to the tilted stacking interaction seen in the lowest energy conformations of analogues with high preceptor affinity. Taken together, these results suggest that a tilted stacking interaction between the Tyr' and Phe" aromatic rings may represent an important structural requirement for high preceptor affinity of the cyclic dermorphin analogues examined in this study. In agreement with the present findings, the results of a molecular dynamics study recently performed with the analogues H-Tyr-D-Om-Phe-Asp-NH, (I) i 1
and H-Tyr-D-Asp-Phe-Orn-NH, (VII) id also indicated a relatively close proximity of the two aromatic rings in these two compounds." A conformational comparison of the cyclic dermorphin analogues with the p-selective opioid peptide morphiceptin (H-Tyr-Pro-Phe-Pro-NH,) is of interest because the latter peptide also contains a Phe3 residue in the 3-position and is also structurally somewhat constrained due to the presence of the two proline residues. The results of a re-
cently performed conformational analysis of morphiceptin and its analogue PL017 [H-Tyr-Prophe(NMe)-~-Pro-NH,],using essentially the same methodoIogica1 approach as described in the present paper, indicated that the lowest energy conformers of both these peptides are also characterized by a close proximity of the Tyr' and Phe3 aromatic rings.22 However, in a recently proposed candidate structure for the bioactive conformation of morphiceptin, the intramolecular distance between these same two rings is considerably larger.23 It should be realized that the relative spatial disposition of the Tyr' and Phe3 side chains could change upon binding to the receptor. In order to still better define the bioactive conformation a t the p receptor, conformational studies on active, cyclic dermorphin analogues with conformationally restricted side chains in the 1- and 3-position will have to be performed. This work was supported by operating grants form the Medical Research Council of Canada (MT-5655and MA10131) and the National Institute on Drug Abuse (DA04443). We thank Mamdouh Mikhaic, Pierre Chinier, and Nicole Fiori for the excellent upkeep of the VAX 11/750. Atomic coordinates for the compounds studied will be made available upon request.
REFERENCES 1. Schiller, P. W. (1984) in The Pepttdes: Analysis, Synthesis, Biology, Vol. 6, Udenfriend, S. & Meienhofer, J., Eds., Academic Press, Orlando, FL, pp. 2 19- 268. 2. Schiller, P. W. & DiMaio, J. (1982) Nature (London) 297, 74-76. 3. DiMaio, J. & Schiller, P. W. (1980) Proc. Natl. Acad. Scl. USA 77,7162-7166. 4. DiMaio, J., Nguyen, T. M.-D., Lemieux, C. & Schiller, P. W. (1982) J . Med. Chem. 25, 1432-1438. 5. Berman, J. M., Goodman, M., Nguyen, T. M.-I). & Schiller, P. W. (1983) Biochem. Biophys. Res. Commun. 115, 864-870. 6. Richman, S. J., Goodman, M., Nguyen, T. M.-D. & Schiller, P. W. (1985) Znt. J . Peptide Protein Res. 25, 648-662. 7. Mosberg, H. F., Hurst, R., Hruby, V. J., Gee, K., Yamamura, H. I., Galligan, J. J. & Burks, T. F. (1983) €'roc. Natl. Acad. Sci. 80,5871-5874. 8. Schiller, P. W., Nguyen, T. M.-D., Lemieux, C. & Maziak, L. A. (1985) J. Med. Chem. 28, 1766-1771. 9. Schiller, P. W., Nguyen, T. M.-D., Maziak, L. A., Wilkes, B. C. & Lemieux, C. (1987) J . Med. Chem. 30,2094-2099. 10. SchiIler, P. W. & Wilkes, B. C. (1988) in Recent Progress in the Chemistry and Biology of Opiolcl
CONFORMATION-ACTIVITYRELATIONSHIPS Peptrdes, (NIDA Research Monograph 87), Rapaka, R. S. & Dhawan, B. N., Eds., U.S. Government Printing Office, Washington, DC,pp. 60-73. 11. Hall, D. & Pavitt, N . (1984) Biopolymers 23, 1441- 1455. 12. Mammi, N. J., Hassan, M. & Goodman, M. (1985) J . Am. C h m . Soc. 107,4008-4013. 13. Maigret, B., Fournib-Zaluski, M.-C., Roques, B. P. & Premilat, S. (1986) Mol. Pharmacol. 29,314-320. 14. Wilkes, B. C. & Schiller, P. W. (1987) Bwpolymers 26, 1431-1444. 15. Smith, G. M. & Veber, D. F. (1986) Bwchem. Biophys. Res. Commun. 134,907-914. 16. Vinter, J. G., Davis, A. & Saunders, M. R. (1987) J . Cornput.-AidedMol. Design 1, 31-51. 17. Kalman, B. L. (1982) Technical Memorandum, No. 49, Department of Computer Science, Washington University, St. Louis, MO.
95
18. Motoc, I. & Marshall, G. R. (1985) Chem. Phys. Lett. 116,415-419. 19. Wilkes, B. C. & Schiller, P. W. (1989) in Proceedings of the 11th American Peptide Symposium, Rivier, J & Marshall, G. R., Eds., ESCOM Science Publishers, Leiden, The Netherlands, in press. 20. Singh, J. & Thomton, J. M. (1985) FEBS Lett. 191, 1-6. 21. Mierke, D. F., Schiller, P. W. & Goodman, M. (1989) Bwpolyners, in press. 22. Wilkes, B. C. & Schiller, P. W., unpublished results. 23. h e w , G., Keys, C., Luke, B., Polgar, W. & Toll, L. (1986) Mol. Pharmcol. 29,546-553.
Received March 28, 1989 Accepted June 9, 1989
SYNOPSIS
A theoretical conformational analysis (molecular mechanics study) of nine cyclic tetrapeptides, structurally related to the highly p-receptor-selective dermorphin analogue H-Tyr-D-Om-Phe-Asp -NH,, was performed. These compounds display considerable
u
diversity in their preceptor affinity and selectivity. A systematic search and subsequent energy minimization in absence of the exocycIic Tyr' residue and Phe3 side chain revealed the constrained nature of the 11-13-membered ring structures contained in these analogues. No more than four low-energy conformers (within 2 kcal/mol of the lowest energy conformation) were found in each case. After attachment of the Tyr' moiety and Phe:' side chain to the "bare" low-energy ring structures, a systematic search and energy minimization of these exocyclic moieties resulted in a limited number of low-energy confomationd families for dl compounds. Five analogues with high preceptor affinityH-Tyr-D- Om-Phe-Asp -NH, ,H-Tyr-D-Om-Phe-D-Asp -NH, ,
u
U
H-Tyr-D-Asp-Phe-Om -NH,, H-Tyr-D-Asp-Phe-A,bu-NH, (A, bu: a, y-diaminobutyric I
acid) and H-Tyr-D-Cys-Phe-Cys-NH,
u
-all showed a tilted stacking interaction between the Tyr' and Phe3 aromatic rings in the lowest or second lowest energy conformation found. The same kind of stacking was not possible in low-energy conformers of the four analogues with poor affinity for the p-receptor
[ H-Tyr-L-Om-PheAsp-NH, , u H-Tyr-D- Om-D-Phe-Asp-NH, ,
H-Tyr-D-Om-Phe(NMe)-Asp-NH, [Phe(NMe): Nu-methylphenylalanine],and I
I
H-Tyr-D-Om-Phg-Asp -NH, (Phg: phenylglycine) } . J
It is concluded that a tilted stacking arrangement of the two aromatic rings may represent a structural requirement for high preceptor affinity of the examined cyclic dermorphin analogues. T: 1990 John Wiley & Sons, Inc. CCC C~S-3525/90/010089-07 $04.00 Biopolymers, Vol. 29,89-95 (1990)
Abbreviations: Symbols and abbreviations are in accord with recommendations of the IUPAC-IUB Commi&on on Bio. chemical Nomenclature. (1984) Biochern. J . 219, 345-373.
INTRODUCTION
Since the discovery of the enkephalins [H-Tyr-GlyGly-Phe-Mewor Leu)-OH] numerous conformational studies On these pentapeptides have been performed with the goal of elucidating their bioac89
90
WILKES AND SCHILLEH
tive conformation(s) (for a review, see Ref. 1).These conformational studies made use of a variety of different methods, including theoretical conformational analysis, x-ray diffraction analysis, and various spectroscopic techniques, and led to the overall conclusion that the natural enkephalins can assume a number of both folded and extended conformations of comparatively low energy. Due to this structural flexibility, conclusive information about the bioactive (receptor-bound) conformation cannot be obtained from conformational analysis of the enkephalins and of other linear opioid peptides. The situation is further complicated by the demonstrated existence of a t least three different types of opioid receptors ( p , 3, and K), each of which has distinct structural and conformational requirements.2 The enkephalins interact preferentially with &receptors but also bind to p receptors, albeit with somewhat lower affinity. The lack of receptor selectivity of these linear pentapeptides is apparently due to their conformational flexibility, which permits structural adaptation to the receptor topography of both the 3 and the p receptor. In recent years, several investigators have prepared conformationally constrained enkephalin analogues in an attempt to improve the receptor selectivity of these peptides. Conformational restriction through various peptide cyclizations (side chain to end group or side chain to side chain) has been particularly successful. The first active cyclic enkephalin analogues were obtained through substitution of a D-diamino acid (Daa) in the 2-position and cyclization between the Daa side-chain amino group and the C-terminal carboxyl function.',' The resulting compounds, H-Tyr-cyclo[-D-Daa-Gly-Phe-Leu-] [Daa = a, pdiaminopropionic acid(A2pr), a ,y-diaminobutyric acid(A,bu), Om, or Lys], not only were potent and enzyme resistant, but also showed considerable preference for p receptors over 3 receptors. Several partial retro-inverso analogues of H-Tyr-cyclo[-DA,bu-Gly-Phe-Leu-] were also found to be very potent, extremely stable, and quite p selective."' An example of a side chain to side chain cyclized enkephalin analogue is [D- Pen2, D(or ~ ) - P e n ~ ] e n -
side chain bridging between appropriate residues substituted in positions 2 (D-configuration) and 4 has recently been Several cyclic lactam tetrapeptide amides of this type with small ring structures (12 or 13 membered) and a cystinebridged tetrapeptide amide (11-membered ring) showed high preference for p receptors over 6 receptors. Among these compounds, the highest preceptor selectivity was displayed by the analogue H-TJT-D-Om-Phe-Asp-NH,, the most selec1_1
tive cyclic opioid peptide analogue with p-agonist properties reported to date. Several studies on the conformational behavior of various cyclic opioid peptide analogues have been carried out (for a review, see Ref. lo). The major focus has been on the moderately p-selective prototype compound H-Tyr-cyclo[-D-A,bu-GlyPhe-Leu-]."-" These studies have not yet led to a consensus concerning a possible unique bioactive conformation of this analogue a t the p receptor. The most important finding of these endeavors was the realization that the 14-membered ring structure in H-Tyr-cycle[-~-A,bu-Gly-Phe-Leu-] still retains some flexibility, and that various intramolecular hydrogen bonds are constantly formed, broken, and reformed again, as shown most conclusively in a molecular dynamics study.I2 We recently performed a theoretical conformational analysis of the highly p-selective cyclic dermorphin analogue H-TV-D-Om-Phe-Asp-NH 2.14
-
The lowest energy conformer found in this molecular mechanics study was characterized by a lack of linear transannular hydrogen bonds and by a tilted stacking interaction of the two aromatic rings. In the present study we extend this type of analysis to several cyclic dermorphin analogues structurally related to H-Tyr-D-Om-Phe-Asp-NH,, and show-
u
ing great diversity with respect to their affinity for p- and 8-opioid receptors. We report a common low-energy conformation for analogues with high preceptor affinity, and propose that this conformation may be of relevance as the bioactive conformation a t the p receptor.
J
kephalin ('Pen: penicillamine), a compound displaying greatly improved 3-receptor selectivity? Whereas the enkephalins contain a phenylalanine residue in the 4-position of the peptide sequence, the opioid peptides dermorphin and pcasomorphin (morphiceptin) have that same residue in the 3-position. A series of cyclic dermorphin-(l-4) tetrapeptide analogues resulting from side chain to
METHODOLOGY
All calculations were performed using the molecular modeling software SYBYL (Tripos Associates, St. Louis, MO) on a VAX 11/750 mainframe using VMS version 4.2. Molecules were viewed with an Evans and Sutherland PS 300 computer graphics
CONFORMATION-ACTIVITYRELATIONSHIPS
display terminal. A Hewlett Packard H P 7475 plotter was used for the preparation of the figures. A comprehensive theoretical analysis of the parent compound H-TP-D- Om-Phe-Asp-NH, has been
u
reported e1~ewhere.l~The methodological approach taken in the latter study was used with some minor modifications to determine the lowenergy conformations of the eight analogues examined in the present study. The first step in ths stepwise procedure was the construction of the "bare" ring structure for each compound. Only atoms directly attached to the ring, with associated hydrogen atoms, were included in this part of the analysis. This approach has been used by other in~estigators'~ and was taken to permit the identification of the greatest possible number of solutions in the conformational search to be undertaken in the second step of the ana!ysis procedure. Each "crude" ring structure was minimized prior to the conformational search using the energy minimization program MINIM. The steepest descent approach was used and the potential energies were calculated from
3; =
WstrEstr
+
+
YorEtor
WangEmg
+
WvdwEvdw
where the Ws represent weight constants and the E s represent the energy terms for the bondstretching (str) energy, angle-bending (ang) energy, torsional (tor) energy, and van der Waal's (vdw) contact energy (including hydrogen-bonding energy). The standard SYBYL force field16 was employed. Once a reasonable starting geometry was obtained, the ring structures could be subjected to a systematic search for allowed conformations. The second step of the procedure consisted of a systematic conformational grid search through all conformational space, using the software program SEARCH.l7 This search program systematically checks for unfavorable vdw contacts around the nonbonded atoms by scanning all possible torsionai angles around all the rotatable bonds. The two criteria for an allowed conformation were the lack of unfavorable vdw interactions and the requirement for ring closure. In order to allow for thermal vibration, a vdw scaling factor of 0.80 was used for nonbonded atoms. A vdw scaling factor of 0.70 was used for 1,4-interactions1' and the vdw factor used for hydrogen-bonding interactions was 0.40. All amide bonds were held trans and planar. One of the bonds was chosen as the ring-closure bond and the remaining bonds were surveyed over a 30" grid over all space. An allowed conformation
91
was obtained if in a structure without unfavorable vdw contacts the ring closure bond could form within 0.25 of the length and within 15 degrees of the torsion angle of a normal carbon-carbon bond. In the case of the cystine-bridged analogue H-Tyr-D-Cys-Phe-Cys-NH,,the disulfide bond was
w A.
chosen as nng closure bond and the ring closure tolerance was 0.4 For each of the ring structures examined in this study, the search program was allowed to scan between lo9 and 10" possible combinations and usually required between 20 and 50 h central processing unit (CPU) time. For the bare ring structure of H-Tyr-u- Om-Phe(NMe)-Asp-NH
I J .
[Phe(NMe): N"-methylphenylalanine), only four solutions were found. For the other eight compounds, the ring search resulted in between 10 and 23 solutions. Each of the allowed ring structures was then minimized using the MINIM program, allowing all the atoms to relax, including those contained in the amide moieties. The solutions for each ring structure were ranked in order of increasing energy, and only those structures within 2 kcal/mol of the minimum energy found were retained for further study. In the third step the exocyclic Tyr' residue and the Phe" or Phg" (Phg: phenylglycine) side chain were linked to each low-energy ring conformer obtained in step 2. The exocyclic amide bonds were held trans and planar, and the remaining six or seven rotatable bonds outside the ring structure were surveyed over a 30" grid in order to determine allowed configurations for the Tyr' residue and Phe3 (Phg') side chain. In each case the program scanned approximately lo6 possible conformations and between 3000 and 10,000 solutions were obtained. The energies of these conformers were calculated, and the resulting solutions were grouped into two to ten low-energy families for each ring structure searched. The fourth step involved picking the lowest energy conformer in each low-energy conformational family and subjecting this structure to extensive energy minimization as described above. The obtained conformers could then be ranked in order of increasing energy. This entire procedure required between 100 and 200 h of CPU time for each compound. RESULTS AND DISCUSSION
The compounds studied and their p- and 6-opioid receptor affinities are listed in Table I. Opioid-
92
WILKES AND SCHILLER
Table I p- and 6-Opioid Receptor Affinities of Cyclic Dermorphin Analogues (taken from Ref. 9)
H-Tyr-D-Om-Phe-Asp-NH,
I
I
2200 k 65
10.4 & 3.7
213
J
> 12.4
I1
H-Tyr-L-Om-Phe-Asp-NH,
4830 & 610
> 60,000
I11
H-Tyr-D-Om-D-Phe-Asp-NH,
3040 5 470
> 380,000
IV
H-Tyr-D-Om-Phe-D-Asp-NH,
21.7 f 3.2
422 5 17
V
H-Tyr-D-Om-Phe(NMe)-Asp-NH,
3570 f 500
28,100 f 2500
7.87
VI
H-Tyr-D-Om-Phg-Asp-NH,
445 f 92
3570 f 370
8.02
VII
H-Tyr-D-Asp-Phe-Om-NH,
9.55 & 2.52
1320
150
138
24.8 i 1.1
4170 f 430
168
11.0 f 0.3
373 k 63
u
-
u
> 125
19.4
I
VIII
H-Tyr-D-Asp-Phe-A,bu-NH, I
IX
H-Tyr-D-Cys-Phe-Cys-NH, "Mean of three determinations
33.9
I
I
SEM.
receptor affinities had been determined by displacement of the radioligands [3H]H-Tyr-~-AlaGly-Phe(NMe)-Gly-ol (p-selective) and [3H]H-TyrDSer-Gly-Phe-Leu-Thr-OH (&selective) from rat brain membrane preparations, as de~cribed.~ In comparison with linear opioid peptides, the conformational flexibility of the cyclic dermorphin analogues studied is greatly reduced due to the considerable rigidity of the relatively small ring structures (11-13 membered) they contain. For each compound lo9 to 10" possible conformations of the bare ring structure were scanned, yet only a very limited number (no more than 23) of allowed conformations was obtained. In each case, the number of low-energy ring conformers within 2 kcal/mol of the lowest energy ring structure found was no more than four. For the ring structure in H-Tyr-DOrn-Phe(NMe)-Asp -NH, (V) I
only four allowed conformations were found and only two of these were of sufficiently low energy to warrant further study. This result indicates that the latter ring structure is somewhat more rigid than the other ring structures studied due to the additional conformational constraint produced by the N-methyl group. The ring structures of the compounds described in the present paper are considerably more rigid than those in some other cyclic
opioid peptide analogues containing larger rings that have been studied to date. For example, in the case of the &selective cyclic enkephalin analogue H-Tyr-D-Pen-Gly-Phe-~-Pen-OH~, sixteen conformations of the 1Cmembered ring within 2 kcal/mol of the lowest energy ring structure were obtained by using the same a p p r o a ~ h . ' ~ The gross topological features of the low-energy ring conformations of the cyclic parent peptide H-Tyr-D- Om-Phe-Asp-NH, (I) and of the eight
u
analogues (11-IX) studied were all quite similar. The structural characteristics of the low-energy ring conformers of parent peptide I have been previously discussed in detail.I4 As in the case of compound I, the low-energy ring structures of analogues 11-IX were all fairly symmetrical and relatively flat when viewed from the side. The amide bonds within the rings were in general more or less perpendicular to the ring plane and no strong (linear), transannular hydrogen bonds were observed in any case. When the conformational analysis was extended to include the Tyr' residue and the Phe3 side chain attached to the various low-energy ring structures, it was found that in all cases the latter moieties enjoy a considerable amount of orientational freedom around the exocylic rotatable bonds. As expected, this was particularly the case for the Tyr' moiety. However, only a relatively limited number
CONFORMATION-ACTIVITYRELATIONSHIPS
93
of low-energy families was observed for all compounds. Four of the analogues-
the two aromatic rings in their lowest energy conformation (Figure 1). In the lowest energy conformation of the compound H-TP-D- Cys-Phe-Cys-
-
NH, (IXa) the Tyr' moiety is oriented away from the rest of the molecule and does not engage in any intramolecular interactions. However, in the second lowest energy conformation of this compound (IXb) (0.2 kcal/mol higher in energy than the lowest one) a tilted stacking interaction between the two aromatic rings is again observed. A recent survey of interactions between aromatic rings in proteins based on crystal structure data revealed that this staggered stacked interaction with tilting of the two rings is the most common and, presumably, the energetically most favorable, while fully stacked interactions are almost never observed.2" Interestingly, the five cyclic dermorphin analogues (I, IV, VII, VIII, and IX) showing a tilted stacking arrangement of the two aromatic rings in their lowest energy conformations all display high affinity for the p-opioid receptor and weak but somewhat variable affinity for the 6 receptor.
H-Tyr-D-Orn-Phe-Asp-NH, (I),
u
H-Tyr-D- Om-Phe-D-Asp-NH, (IV), H-Tyr-D-AspPhe-Om-NH, (VII),
u
and H-Tyr-D-Asp-Phe-A,bu -NH, (VIII) I
-showed
a tilted stacking arrangement between
u
Ip'
'm
IT.
PT
Figure 1. Lowest-energy conformations of H-Tyr-D-Om-Phe-Asp -NH, (I),
u
P Figure 2. Lowest energy conformations of
H-Tyr-D-Om-Phe-D-Asp -NH, (IV) ,
H-TY-L- Om-Phe-Asp -NH, (11) , I
H-Tyr-D- Asp-Phe-Om -NH, (VII),
H-Tyr-D-Om-D-Phe-Asp -NH, (111) ,
I
H-Tyr-D-Asp-Phe-A,bu -NH, (VIII) ,
H-Tyr-D-Om-Phe(NMe)-Asp -NH, (V) ,
1
and
and H-Tyr-D-Cys-Phe-Cys -NH, ( IXa and IXb) .
H-Tyr-D-Om-Phg-Asp -NH, (VI) .
u
94
WILKES AND SCHILLER
On the other hand, the cyclic analogues displaying weak affinity for p receptors (compounds 11, 111, V, and V1) were unable to adopt this tilted stacking interaction of the two aromatic rings in their low-energy conformations (Figure 2). In the case of analogues I1 and 111, this is due to the fact that configurational inversion a t the 2- or 3-position of the peptide sequence places the side chains of Tyr' and Phe3 on opposite sides of the peptide ring structure, thereby preventing an intramolecular interaction between the two aromatic rings. Examination of the low-energy conformations obtained for the weakly active analogue H-Tyr-D-Om-Phe(NMe)-Asp-NH, (V) I
revealed that aromatic ring stacking is not possible because of steric interference by the bulky Nmethyl group introduced in the 3-position of the peptide backbone. An almost fully stacked parallel arrangement of the two aromatic rings is observed in the lowest energy conformer of the weak y agonist H-Tyr-D-Om-Phg-Asp-NH, (VI), in contrast to the tilted stacking interaction seen in the lowest energy conformations of analogues with high preceptor affinity. Taken together, these results suggest that a tilted stacking interaction between the Tyr' and Phe" aromatic rings may represent an important structural requirement for high preceptor affinity of the cyclic dermorphin analogues examined in this study. In agreement with the present findings, the results of a molecular dynamics study recently performed with the analogues H-Tyr-D-Om-Phe-Asp-NH, (I) i 1
and H-Tyr-D-Asp-Phe-Orn-NH, (VII) id also indicated a relatively close proximity of the two aromatic rings in these two compounds." A conformational comparison of the cyclic dermorphin analogues with the p-selective opioid peptide morphiceptin (H-Tyr-Pro-Phe-Pro-NH,) is of interest because the latter peptide also contains a Phe3 residue in the 3-position and is also structurally somewhat constrained due to the presence of the two proline residues. The results of a re-
cently performed conformational analysis of morphiceptin and its analogue PL017 [H-Tyr-Prophe(NMe)-~-Pro-NH,],using essentially the same methodoIogica1 approach as described in the present paper, indicated that the lowest energy conformers of both these peptides are also characterized by a close proximity of the Tyr' and Phe3 aromatic rings.22 However, in a recently proposed candidate structure for the bioactive conformation of morphiceptin, the intramolecular distance between these same two rings is considerably larger.23 It should be realized that the relative spatial disposition of the Tyr' and Phe3 side chains could change upon binding to the receptor. In order to still better define the bioactive conformation a t the p receptor, conformational studies on active, cyclic dermorphin analogues with conformationally restricted side chains in the 1- and 3-position will have to be performed. This work was supported by operating grants form the Medical Research Council of Canada (MT-5655and MA10131) and the National Institute on Drug Abuse (DA04443). We thank Mamdouh Mikhaic, Pierre Chinier, and Nicole Fiori for the excellent upkeep of the VAX 11/750. Atomic coordinates for the compounds studied will be made available upon request.
REFERENCES 1. Schiller, P. W. (1984) in The Pepttdes: Analysis, Synthesis, Biology, Vol. 6, Udenfriend, S. & Meienhofer, J., Eds., Academic Press, Orlando, FL, pp. 2 19- 268. 2. Schiller, P. W. & DiMaio, J. (1982) Nature (London) 297, 74-76. 3. DiMaio, J. & Schiller, P. W. (1980) Proc. Natl. Acad. Scl. USA 77,7162-7166. 4. DiMaio, J., Nguyen, T. M.-D., Lemieux, C. & Schiller, P. W. (1982) J . Med. Chem. 25, 1432-1438. 5. Berman, J. M., Goodman, M., Nguyen, T. M.-I). & Schiller, P. W. (1983) Biochem. Biophys. Res. Commun. 115, 864-870. 6. Richman, S. J., Goodman, M., Nguyen, T. M.-D. & Schiller, P. W. (1985) Znt. J . Peptide Protein Res. 25, 648-662. 7. Mosberg, H. F., Hurst, R., Hruby, V. J., Gee, K., Yamamura, H. I., Galligan, J. J. & Burks, T. F. (1983) €'roc. Natl. Acad. Sci. 80,5871-5874. 8. Schiller, P. W., Nguyen, T. M.-D., Lemieux, C. & Maziak, L. A. (1985) J. Med. Chem. 28, 1766-1771. 9. Schiller, P. W., Nguyen, T. M.-D., Maziak, L. A., Wilkes, B. C. & Lemieux, C. (1987) J . Med. Chem. 30,2094-2099. 10. SchiIler, P. W. & Wilkes, B. C. (1988) in Recent Progress in the Chemistry and Biology of Opiolcl
CONFORMATION-ACTIVITYRELATIONSHIPS Peptrdes, (NIDA Research Monograph 87), Rapaka, R. S. & Dhawan, B. N., Eds., U.S. Government Printing Office, Washington, DC,pp. 60-73. 11. Hall, D. & Pavitt, N . (1984) Biopolymers 23, 1441- 1455. 12. Mammi, N. J., Hassan, M. & Goodman, M. (1985) J . Am. C h m . Soc. 107,4008-4013. 13. Maigret, B., Fournib-Zaluski, M.-C., Roques, B. P. & Premilat, S. (1986) Mol. Pharmacol. 29,314-320. 14. Wilkes, B. C. & Schiller, P. W. (1987) Bwpolymers 26, 1431-1444. 15. Smith, G. M. & Veber, D. F. (1986) Bwchem. Biophys. Res. Commun. 134,907-914. 16. Vinter, J. G., Davis, A. & Saunders, M. R. (1987) J . Cornput.-AidedMol. Design 1, 31-51. 17. Kalman, B. L. (1982) Technical Memorandum, No. 49, Department of Computer Science, Washington University, St. Louis, MO.
95
18. Motoc, I. & Marshall, G. R. (1985) Chem. Phys. Lett. 116,415-419. 19. Wilkes, B. C. & Schiller, P. W. (1989) in Proceedings of the 11th American Peptide Symposium, Rivier, J & Marshall, G. R., Eds., ESCOM Science Publishers, Leiden, The Netherlands, in press. 20. Singh, J. & Thomton, J. M. (1985) FEBS Lett. 191, 1-6. 21. Mierke, D. F., Schiller, P. W. & Goodman, M. (1989) Bwpolyners, in press. 22. Wilkes, B. C. & Schiller, P. W., unpublished results. 23. h e w , G., Keys, C., Luke, B., Polgar, W. & Toll, L. (1986) Mol. Pharmcol. 29,546-553.
Received March 28, 1989 Accepted June 9, 1989
