Molecular dynamics simulations of opioid peptide analogs containing multiple conformational restrictions BRIAN C. WILKES and PETER W. SCHILLER
Laboratory of Chemical Biology and Peptide Research, Clinical Research Institute of Montrealt. Montreal, Canada
Received 8 February, accepted for publication 12 June 1992
- -
Molecular dynamics simulations were performed on the potent and slightly p-receptor selective cyclic dermorphin analog H-Tyr-D-Orn-Phe-Glu-NH2 as well as on analogs containing a conformationally restricted phenylalanine derivative in place of Phe in the 3 position of the peptide sequence. Peptides studied included the potent and highly p-selective analogs H-Tyr-~-Orn-Aic-Glu-NH2(Aic = 2-aminoindan-2-carboxylic acid), H-Tyr-~-Orn-Atc-Glu-NH2(Atc = 2-aminotetralin-2-carboxylic acid) and H-Tyr-D-Orn-D-Atc-GIu-NH2,
u
L___!
and the weakly active analog H-Tyr-~-Orn-Tic-Glu-NH~ (Tic = tetrahydroisoquinoline-3-carboxylicacid). u Four different starting conformations were chosen for each peptide, and after equilibration each simulation was allowed to proceed for 100 picoseconds at 600°K. The 14-membered ring structures in the Phe-, Aic-, L- and D-Atc-containing analogs showed moderate structural flexibility, while the peptide ring in the Ticcontaining analog was more rigid. As theoretically predicted, the @ and i,h angles of the Aic-, L- and D-Atc-containing analogs were limited to values of either about + 50" or -50" during almost the entire period of the simulations. In the Tic-containing analog the $3 and $3 angles were 0" and 90", respectively, and did not change for the entire duration of the simulation. The side chains of the constrained amino acids showed limited movement, but transitions between the allowed conformations did occur on the time scale of the simulations. One interesting aspect of the 5-membered ring of the Aic side chain was that it underwent a greater number of conformational transitions than the 6-membered rings contained in Atc and Tic, but covered a smaller volume of conformational space. Thus, the relative flexibility of these constrained amino acids was Phe > Aic > L- and D-Atc = Tic, but the relative volume of conformational space visited by these same residues was Phe> L- and D-Atc = Tic > Aic. In all compounds studied the exocyclic Tyr' residue had the greatest flexibility and covered a large volume of conformational space. Key words: conformational flexibility; conformational restriction; cyclic dermorphin analogs; cyclic phenylalanine derivatives; molecular dynamics simulations; opioid receptor selectivity; theoretical conformational analysis
Dedicated to Professor Bruce Merrifield on the occasion of his 70th birthday The enkephalins and other naturally occurring opioid peptides are highly flexible molecules capable of assuming a number of folded and extended conformations of Abbreviations: All optically active amino acids are of the relatively low energy (1). For example, it has been demL-configuration unless otherwise noted. Symbols and abbreviations onstrated convincingly that in solution the enkephalins are in accord with the recommendations of the IUPAC-IUB Com- exist in a conformational equilibrium (2). The structural mission on Biochemical Nomenclature [ J . Bid. Chem. 264,668-673 flexibility of these peptides may explain their lack of (1989)l. high selectivity towards the different opioid receptor Other abbreviations: Azhu, a,y-diaminobutyric acid; Aic, types (p, b, K). In an effort to determine the conforma2-aminoindan-2-carboxylic acid; Atc, 2-aminotetralin-2-carboxylic tional requirements of the various receptor types and t o acid; Tic, tetrahydroisoquinoline-3-carboxylicacid; NMePhe, Numethylphenylalanine; Om, ornithine; Pen, penicillamine; fs, femto- develop analogs with improved receptor selectivity, various investigators attempted to reduce the conforsecond; ps, picosecond. mational flexibility of opioid peptides through the int Affiliated with the University of Montreal. 249
B.C. Wilkes and P.W. Schiller corporation of conformational constraints. Peptide cyclizations through amino acid side chain groups proved to be very successful in obtaining opioid peptide analogs with improved stability as well as high selectivity towards either the p or the 6 opioid receptors. For example, the compounds H-Tyr-cyclo[-~-AzbuGly-Phe-Leu-] (3) and H-Tyr-D-Om-Phe-Asp-NH2
METHODOLOGY
Molecular dynamics calculations were performed using the molecular modelling software SYBYL (Tripos Associates, St. Louis, MO) version 5.30 on a VAXstation 3500. Molecules were viewed with an Evans and Sutherland PS330 computer graphics display terminal and a Hewlett Packard HP7475 plotter was used for the (4) were found to be potent and selective for the preparation of the figures. The standard SYBYL force p-receptor, whereas the cyclic opioid peptide analogs field was employed for energy calculations (lo), and the H-Tyr-D-Pen-Gly-Phe-(Dor L)-Pen-OH (5) and H-Tyr- Verlet method was used for integrating the equations of I I D-Cys-Phe-D-Pen-OH ( 6 ) showed markedly improved motion (11). A distance-dependent dielectric constant w of 78 was used to simulate an aqueous environment. 6 selectivity. Recent molecular mechanics studies performed with Low energy conformations of each compound were the cyclic analogs H-Tyr-~-Orn-Phe-Asp-NH2(7), and generated by means of a molecular mechanics approach described elsewhere (7, 8, 12), and four random low H-Tyr-D-Pen-Gly-Phe-D-Pen-OH (8) have shown that, energy structures were chosen for each compound as I while these compounds do have preferred conforma- starting conformations. Each dynamics simulation was tions, they are still somewhat flexible. In particular, the allowed to equilibrate to 600 OK for 2 picoseconds (ps) exocyclic Tyr' residue and the Phe3 (or Phe4) side chain and the simulation was allowed to proceed for 100 ps in these compounds still enjoy significant orientational at 600°K. This temperature was chosen to allow samfreedom. Because the tyramine moiety of the Tyr' res- pling of a representative amount of available conforidue and the aromatic side chain of the Phe3 (or Phe4) mational space in a reasonable time frame. A timestep residue are critical determinants for the observed bio- of 1 fs was employed and structures were recorded for logical activities of opioid peptides, it is necessary to analysis every 50 fs. Each dynamics trajectory was anobtain detailed information on the relative orientation alyzed for selected distances between functional groups of these functional groups in order to gain insight into and all torsional angles were monitored. Conformathe distinct bioactive conformations of these peptides tions were sampled along the trajectory and minimized, as well as combined to show the accessible conformaat the p and the 6 receptor. We recently prepared a series of conformationally tional space of the various residues in each analog. restricted phenylalanine derivatives and incorporated RESULTS AND DISCUSSION them in place of Phe3 in the cyclic opioid peptide analog H-Tyr-~-Orn-Phe-Glu-NH2which contains a 14In each dynamics simulation, the various functional membered ring structure (9). Because H-Tyr-D- groups underwent multiple conformational transitions Om-Phe-Glu-NH2 is very potent but has only slight and appeared to cover a fairly large amount of available u preference for p receptors over 6 receptors, it represents conformational space. In the case of the parent peptide an ideal parent peptide to study the effects of additional H-Tyr-~-Orn-Phe-Glu-NH2,examination of 20 conconformational restrictions at the 3-position residue on formations sampled at 5 ps intervals along one trajecthe conformational behavior and on opioid receptor tory demonstrated that this cyclic peptide is structurally affinity and selectivity. We report here molecular dy- still quite flexible (Fig. 2). In agreement with the results namics simulations of the essentially non-selective par- of molecular mechanics studies of related peptides (7, ent peptide H-Tyr-D-Om-Phe-Glu-NHz, of the three 12), the Tyr' residue showed greatest structural flexjpotent and very p-selective analogs H-Tyr-D- bility due to relatively unhindered rotation around the Om-Aic-Glu-NH2, H-Tyr-~-Orn-Atc-Glu-NH2 and four exocyclic rotatable bonds at the N-terminus. The u u Phe3 side chain was also very mobile during the simH-Tyr-~-Orn-~-Atc-Glu-NH2, and of the weakly ac- ulations, behaving somewhat like a spinning top. Be-
-
-
-
-
u
tive analog, H-Tyr-D-Om-Tic-Glu-NH2 (9) (Fig. 1). I
1 H-Tyr-~-Orn-Phe-Glu-NH2 U
2 H-Tyr-~-Orn-Aic-Glu-NH2 U
3 H-Tyr-D-Om-Atc-Glu-NH2 U
4 H-Tyr-~-Orn-Tic-Glu-NH~ U
FIGURE 1 Structures of cyclic dermorphin tetrapeptide andogs 1-4. 250
cause the high dielectric constant of water was employed, no strong hydrogen bonds were observed in these simulations. Replacement of Phe3 with the various conformationally restricted aromatic amino acids produced conformational constraints both in the side chain and in the peptide backbone at the 3-position. Backbone torsional angles (@,$) for the Aic- and Atc-containing peptides were limited to values around $3 = - 50", $3 = - 50" and $3 = + 50", $3 = + 50" during most of the time period of the simulations (Figs. 3 and 4), in agreement
Molecular dynamics of opioid peptides
0
50
100
50
0
2
1
d)
t 1 8 0 F
0
50
100
0
50 TIME (PSI
TIME (PS)
3
-
4
FIGURE 2 Twenty snapshots of H-Tyr-~-Om-Phe-Glu-NH2(l),H-Tyr-D-Om-
-
I00
TIME (PS)
TIME (PS)
100
FIGURE 4 $3 angle of analogs 1 (a), 2 (b), 3 (c) and 4 (d), monitored along the dynamics trajectories.
L
analog are presented in the Figures. In the Ticcontaining analog the @ and $3 angles were found to be 0 " and 90 O , respectively, and did not change for the gen atoms were omitted. entire duration of the simulations. This observation reflects the structural rigidity of the peptide backbone at the 3-position in this peptide as we11 as the fact that the bond defining the @ angle is part of two fused ring structures. Due to the five- or six-membered ring structure present in the conformationally constrained amino acids Aic and Atc, the range of accessible side chain torsional angles (XI, 2 2 ) is severely limited in these res. . ., . . . , idues (Figs. 5 and 6). While in the case of phenylalanine b 50 100 0 50 100 TIME (PS) TIME (PSI the x 1 angle can assume three different values [g+ (t 60°), g- (-60°), and t (180")l and preferred values for x2 are 90", the side chain torsional angles of the conformationally constrained amino acids can assume only two values. Thus, the side chain torsional angles of Aic are limited to values of 1 1 = - 80°, x 2 = -20" and X I = -160", x 2 = + 2 0 ° . For Atc the -90 !f two side chain conformations are characterized by the torsion angles x1 = 180" (t), x 2 = + 25 and X I = - 60" -1804-LrA-. (g - ), x 2 = - 25 O, whereas for D-Atc the corresponding 0 50 100 0 50 100 TIME (PS) TIME (PSI values are X I = 180" (t), x 2 = -25" and X I = +60° (g+ ), x 2 = + 25". In the case of L-Tic, the accessible FIGURE 3 values are = t 60" (g+), x 2 = - 25" and 311 = - 60" @ angle of analogs 1 (a), 2 (b), 3 (c) and 4 (d), monitored along the (g- ), x 2 = + 25 ". In the course of the molecular dydynamics trajectories. namics simulations, the conformationally restricted amino acid residues in all these compounds spent about with results obtained from studies with Ca-methyl equal time in each of the two accessible side chain amino acids (13). In all the dynamics simulations, the conformational states. While the substitution of the cyclic phenyIalanine behavior of the L - A ~ cand ~ ~-Atc~-containing peptides was similar and, therefore, only the results for the L - A ~ c ~analogs severely reduced the side chain- and backbone Aic-Glu-NHZ (2), H-Tyr-~-Orn-Atc-Glu-NH2 (3), and H-Tyr-D-
I
Om-Tic-Glu-NHt (4), taken along the dynamics trajectories. Hydro-
-1801
--
'1
1
-
,
i
O
25 1
B.C. Wilkes and P.W. Schiller volume of conformational space. Thus, the rank order of the relative mobility of these various residues was Phe>Aic>L- and D-Atc= Tic, but the rank order of the relative volume of conformational space available to these sameresidues was Phe > L- and D-Atc = Tic > Aic. The fact that the Aic-containing peptide as well as the L- and D-Atc-containing analogs are potent and selective p receptor agonists suggests that the positioning of the Aic residue in H-Tyr-~-Om-Aic-Glu-NH2may be I_) nearly optimal for the interaction with the p-receptor. In Fig. 2, 20 snapshots of H-Tyr-~-Om-Aic-Glu-NH2 1_1
-w
2
+go o
W z 4
z c3 -90
a -90
-180
-180
0
50 TIME (PSI
100
0
50 TIME (PS)
100
FIGURE 5 x: angle of analogs 1 (a), 2 (b), 3 (c) and 4 (d), monitored along the dynamics trajectories.
conformational mobility at the 3-position residue in these compounds, transitions between the various allowed conformations were nonetheless observed in the course of the performed molecular dynamics simulations. A transition in the @ angle of Atc or Aic from + 50" to -50" was often (but not always) followed by a corresponding change in the $3 angle from + 50" to -50" within a few ps (Figs. 3 and 4). When a conformational transition in the side chain of any of the conformationally constrained amino acid residues occurred, both the X I and the xz angle changed simultaneously (Figs. 5 and 6 ) . However, transitions between the two allowed side chain conformational states occurred independently from backbone conformational transitions in all the compounds studied. Interestingly, the molecular dynamics simulations showed that in thesecyclic opioid peptides the 3-position residues containing a six-membered ring (L- and D-Atc and L-Tic) underwent fewer side chain conformational transitions than the Aic3 residue, which contains a 5-membered ring, or than the unrestricted Phe3 residue (Figs. 5 and 6 ) . In order for the constrained amino acids to undergo a successful side chain conformational transition, the torsional angles in the five- or sixmembered rings of these amino acids must change in a concerted manner. One interesting aspect of the obtained results was that, while the five-membered ring contained in the Aic residue of the Aic3-analog was shown to undergo a greater number of conformational transitions than the six membered rings of the Atc3- and Tic3 residues in these cyclic analogs, the Aic residue covered a smaller amount of torsional space than the Atc- and Tic residues and, therefore, covered a smaller 252
taken at 5 ps intervals along the dynamics trajectory are superimposed. This figure clearly demonstrates the limited conformational space available to the Aic side chain. The low activity observed with the Tic-containing peptide, H-Tyr-D-Om-Tic-Glu-NH2, may not be due to the conformational .restriction imposed by the Tic residue per se, but may be due to the fact that the sixmembered ring of Tic is formed through the amine nitrogen of this residue. A related compound containing an Nz-methyl-Phe residue, H-Tyr-DOm-NMePhe-Glu-NHz, was also found to have weak w activity (9), suggesting that the loss of activity shown by both the Tic- and the NMePhe-containing peptide may be due to the unavailability of the amide proton in the 3-position of the peptide sequence for the formation of an important hydrogen bond with a receptor moiety. An alternative explanation is that the steric bulk around the a-nitrogen in these compounds may just not be tolerated at the receptor.
-
0
50
100
50 TIME ( P S )
0
TIME (PS)
1 -804 1 -
0
50 TIME (PS)
100
-1804~, 0
.
,
,
,,
.
50 TIME (PSI
100
,
,
I
100
FIGURE 6 1; angle of analogs 1 (a), 2 (b), 3 (c) and 4 (d), monitored along the dynamics trajectories.
Molecular dynamics of opioid peptides Substitution of the cyclic phenylalanine analogs for Phe3 in the parent peptide in general caused primarily a local conformational restriction of the peptide backbone around the 3-position residue but had a lesser effect on the conformational behavior of the rest of the peptide ring structure, with the exception of the Ticcontaining analog. Monitoring of the $ angle of the Glu4residue along the dynamics trajectory (Fig. 7) provided an indication of the relative structural flexibility of the 14-membered peptide ring structure in these compounds. Several transitions in the x: angle were observed for the Phe-, Aic- and Atc-analogs, but not for the Tic-analog. In the case of the Phe-, Aic- and Atcanalogs, transitions between the various side chain conformers of the D-0m2 and Glu4 residues were usually observed several times during each simulation and in many cases several adjacent torsional angles underwent simultaneous transitions. Several 180' amide bond rotations (but not cis-trans isomerizations) were observed in a few of the dynamics simulations, but this event was rare (data not shown), and did not occur during any of the simulations of the Tic-containing peptide. In fact, in two of the four dynamics simulations performed with H-Tyr-~-Om-Tic-Glu-NH2(Figs. 3d, u 4d and 7d), the peptide was held rigid with the 14membered ring vibrating around a single conformation. In all compounds studied the exocyclic tyrosine residue displayed high conformational flexibility, as indicated by the frequent fluctuations in the $1 angle (Fig. 8) and in the x: and xi angles (data not shown) observed dong the dynamics trajectory. Because of this orientational freedom of the Tyrl residue (see also Fig. 2) the intramolecular distance between the two aromatic rings
d
W
-90 -180
50
TIME (PSI
. . .
-180 50 TIME (PS)
100
50
0
100
TIME (PSI
loo
0
.
I
,
50
.
'
50 (PS)
100
0
TIME
50 TIME (PS)
100
FIGURE 8 $1 angle of analogs 1 (a), 2 (b), 3 (c) and 4 (d), monitored along the dynamics trajectories.
observed in these computer simulations for both the active analog! and for the inactive one covered a wide range (3-17 A). CONCLUSION
-
The observation that in two of the loops molecular dynamics runs of H-Tyr-~-Orn-Tic-Glu-NH2no conformational transitions of the 14-membered ring structure were observed, whereas in the two other runs such transitions did occur, clearly suggests that both relatively long simulations and the use of different starting conformations are necessary to adequately sample the conformational space available to these peptides. In addition, the dynamics simulations may have to be run at high temperatures in order to overcome transitional barriers. For example, in a recent molecular dynamics study (14) performed with the cyclic enkephalin apalog H-Tyr-D-Pen-Gly-Phe-D-Pen-OH at 3 16OK only one starting conformation was used and, therefore, only limited sampling of the conformational space around the starting conformation may have been achieved. More extensive molecular dynamics studies at 1000°K (15) found additional conformations. Through incorporation of additional conformational constraints at the 3-position we obtained a series of opioid peptide analogs which in comparison with the cyclic parent peptide H-Tyr-D-Om-Phe-Glu-NHz were L-----' considerably more rigid but still displayed some limited structural flexibility. Replacement of Phe3 with the various cyclic phenylalanine derivatives produced analogs showing greatly reduced mobility of the peptide back-
-
o
z 0
0
0
I
100
TIME (PS)
FIGURE 7 $ angle of analogs 1 (a), 2 (b), 3 ( c ) and 4 (d), monitored dong the dynamics trajectories.
253
B.C. Wilkes and P.W. Schiller bone around the 3-position residue in all cases. The rest of the 14-membered peptide ring structure in the Aic3and Atc3-analogs also showed reduced structural mobility in comparison with the Phe3-parent peptide, but the most drastic rigidification of the entire peptide ring structure was observed with the Tic3-analog which in two of the four molecular dynamics simulations showed no conformational transitions at all for that part of the molecule. As expected, the side chains of the conformationally restricted 3-position residues showed greatly reduced structural flexibility, insofar as transitions between only two side chain conformational states were observed. A particularly interesting result of the simulations was the observation that the Aic3 residue underwent more frequent conformational transitions than the other cyclic Phe residues in these compounds, but at the same time had the least amount of torsional space available. The exocyclic tyrosine residue displayed high conformational mobility in all compounds studied. Obviously, additional conformational constraints will have to be introduced at the Tyr' residue in order to obtain more definitive insight into the receptor-bound conformation of this class of cyclic opioid peptides.
2. Fishman, A.J., Riemen, M.W. & Cowburn, D. (1978) FEES Lett. 94, 236-240 3. Schiller, P.W. & DiMaio, I. (1982) Nature 297, 74-76 4. Schiller, P.W., Nguyen, T.M.-D. & Maziak, L.A. (1985)J. Med. Chem. 28, 1766-1771 5. Mosberg, H.I., Hurst, R., Hruby, V.J., Gee, K., Yammura, H.I., Galligan, J.J. & Burks, T.F. (1983) Proc. Nafl. Acad. Sci. USA 80, 5871-5874 6. Mosberg, H.I., Omnaas, J.R., Medzihradsky, F. & Smith, G.B. (1988) Life Sci. 43, 1013-1020 7. Wilkes, B.C. & Schiller, P.W. (1987) Biopolymers 26,1431-1444 8. Wilkes, B.C. & Schiller, P.W. (1991) J . Cornput.-Aided Mol. Design 5, 293-302 9. Schiller, P.W., Weltrowska, G., Nguyen, T.M.-D., Lemieux, C., Chung, N.N., Marsden, B J . & Wilkes, B.C. (1991)J. Med. Chem. 34, 3125-3132 10. Clark, M., Cramer 111, R.D. & Opdenbosch, N.V. (1989) J . Comp. Chem. 10, 982-1012 11. Verlet, L. (1967) Phys. Rev. 159, 98-103 12. Wilkes, B.C. & Schiller, P.W. (1990) Biopo!ymers 29, 89-95 13. Marshall, G.R. & Bosshard, H.E. (1972) Circulation Res. (Supp. II) 30 and 31, 11143-11150 14. Smith, P.E., Dang, L.X. & Pettitt, B.M. (1991)J. Am. Chem. SOC. 113, 67-73 15. Pettitt, B.M., Matsunaga, T., Al-Obeidi, F., Jehrig, C., Hruby, V.J. & Karplus, M. (1991) Biophys. J. 60, 1540-1544
ACKNOWLEDGMENTS This work was supported by grants from the Medical Research Council of Canada (grant MT-10131) and the National Institute on Drug Abuse (grant DA-04443). We thank Johanne Duhaime and Franqois Boucher for the excellent upkeep of the computer facility and Helen Zalatan for typing the manuscript.
REFERENCES 1. Schiller, P.W. (1984) in The Peptides: Analysis, Synthesis, Biology (Udenfriend, S . & Meienhofer, J., eds.), pp. 219-268, Academic Press, Orlando, FA
254
Address: Dr. Peter W. Schiller Dr. Brian C. Wilkes Laboratory of Chemical Biology and Peptide Research Clinical Research Institute of Montreal 110 Pine Avenue West Montreal, Que. Canada H2W 1R7
Laboratory of Chemical Biology and Peptide Research, Clinical Research Institute of Montrealt. Montreal, Canada
Received 8 February, accepted for publication 12 June 1992
- -
Molecular dynamics simulations were performed on the potent and slightly p-receptor selective cyclic dermorphin analog H-Tyr-D-Orn-Phe-Glu-NH2 as well as on analogs containing a conformationally restricted phenylalanine derivative in place of Phe in the 3 position of the peptide sequence. Peptides studied included the potent and highly p-selective analogs H-Tyr-~-Orn-Aic-Glu-NH2(Aic = 2-aminoindan-2-carboxylic acid), H-Tyr-~-Orn-Atc-Glu-NH2(Atc = 2-aminotetralin-2-carboxylic acid) and H-Tyr-D-Orn-D-Atc-GIu-NH2,
u
L___!
and the weakly active analog H-Tyr-~-Orn-Tic-Glu-NH~ (Tic = tetrahydroisoquinoline-3-carboxylicacid). u Four different starting conformations were chosen for each peptide, and after equilibration each simulation was allowed to proceed for 100 picoseconds at 600°K. The 14-membered ring structures in the Phe-, Aic-, L- and D-Atc-containing analogs showed moderate structural flexibility, while the peptide ring in the Ticcontaining analog was more rigid. As theoretically predicted, the @ and i,h angles of the Aic-, L- and D-Atc-containing analogs were limited to values of either about + 50" or -50" during almost the entire period of the simulations. In the Tic-containing analog the $3 and $3 angles were 0" and 90", respectively, and did not change for the entire duration of the simulation. The side chains of the constrained amino acids showed limited movement, but transitions between the allowed conformations did occur on the time scale of the simulations. One interesting aspect of the 5-membered ring of the Aic side chain was that it underwent a greater number of conformational transitions than the 6-membered rings contained in Atc and Tic, but covered a smaller volume of conformational space. Thus, the relative flexibility of these constrained amino acids was Phe > Aic > L- and D-Atc = Tic, but the relative volume of conformational space visited by these same residues was Phe> L- and D-Atc = Tic > Aic. In all compounds studied the exocyclic Tyr' residue had the greatest flexibility and covered a large volume of conformational space. Key words: conformational flexibility; conformational restriction; cyclic dermorphin analogs; cyclic phenylalanine derivatives; molecular dynamics simulations; opioid receptor selectivity; theoretical conformational analysis
Dedicated to Professor Bruce Merrifield on the occasion of his 70th birthday The enkephalins and other naturally occurring opioid peptides are highly flexible molecules capable of assuming a number of folded and extended conformations of Abbreviations: All optically active amino acids are of the relatively low energy (1). For example, it has been demL-configuration unless otherwise noted. Symbols and abbreviations onstrated convincingly that in solution the enkephalins are in accord with the recommendations of the IUPAC-IUB Com- exist in a conformational equilibrium (2). The structural mission on Biochemical Nomenclature [ J . Bid. Chem. 264,668-673 flexibility of these peptides may explain their lack of (1989)l. high selectivity towards the different opioid receptor Other abbreviations: Azhu, a,y-diaminobutyric acid; Aic, types (p, b, K). In an effort to determine the conforma2-aminoindan-2-carboxylic acid; Atc, 2-aminotetralin-2-carboxylic tional requirements of the various receptor types and t o acid; Tic, tetrahydroisoquinoline-3-carboxylicacid; NMePhe, Numethylphenylalanine; Om, ornithine; Pen, penicillamine; fs, femto- develop analogs with improved receptor selectivity, various investigators attempted to reduce the conforsecond; ps, picosecond. mational flexibility of opioid peptides through the int Affiliated with the University of Montreal. 249
B.C. Wilkes and P.W. Schiller corporation of conformational constraints. Peptide cyclizations through amino acid side chain groups proved to be very successful in obtaining opioid peptide analogs with improved stability as well as high selectivity towards either the p or the 6 opioid receptors. For example, the compounds H-Tyr-cyclo[-~-AzbuGly-Phe-Leu-] (3) and H-Tyr-D-Om-Phe-Asp-NH2
METHODOLOGY
Molecular dynamics calculations were performed using the molecular modelling software SYBYL (Tripos Associates, St. Louis, MO) version 5.30 on a VAXstation 3500. Molecules were viewed with an Evans and Sutherland PS330 computer graphics display terminal and a Hewlett Packard HP7475 plotter was used for the (4) were found to be potent and selective for the preparation of the figures. The standard SYBYL force p-receptor, whereas the cyclic opioid peptide analogs field was employed for energy calculations (lo), and the H-Tyr-D-Pen-Gly-Phe-(Dor L)-Pen-OH (5) and H-Tyr- Verlet method was used for integrating the equations of I I D-Cys-Phe-D-Pen-OH ( 6 ) showed markedly improved motion (11). A distance-dependent dielectric constant w of 78 was used to simulate an aqueous environment. 6 selectivity. Recent molecular mechanics studies performed with Low energy conformations of each compound were the cyclic analogs H-Tyr-~-Orn-Phe-Asp-NH2(7), and generated by means of a molecular mechanics approach described elsewhere (7, 8, 12), and four random low H-Tyr-D-Pen-Gly-Phe-D-Pen-OH (8) have shown that, energy structures were chosen for each compound as I while these compounds do have preferred conforma- starting conformations. Each dynamics simulation was tions, they are still somewhat flexible. In particular, the allowed to equilibrate to 600 OK for 2 picoseconds (ps) exocyclic Tyr' residue and the Phe3 (or Phe4) side chain and the simulation was allowed to proceed for 100 ps in these compounds still enjoy significant orientational at 600°K. This temperature was chosen to allow samfreedom. Because the tyramine moiety of the Tyr' res- pling of a representative amount of available conforidue and the aromatic side chain of the Phe3 (or Phe4) mational space in a reasonable time frame. A timestep residue are critical determinants for the observed bio- of 1 fs was employed and structures were recorded for logical activities of opioid peptides, it is necessary to analysis every 50 fs. Each dynamics trajectory was anobtain detailed information on the relative orientation alyzed for selected distances between functional groups of these functional groups in order to gain insight into and all torsional angles were monitored. Conformathe distinct bioactive conformations of these peptides tions were sampled along the trajectory and minimized, as well as combined to show the accessible conformaat the p and the 6 receptor. We recently prepared a series of conformationally tional space of the various residues in each analog. restricted phenylalanine derivatives and incorporated RESULTS AND DISCUSSION them in place of Phe3 in the cyclic opioid peptide analog H-Tyr-~-Orn-Phe-Glu-NH2which contains a 14In each dynamics simulation, the various functional membered ring structure (9). Because H-Tyr-D- groups underwent multiple conformational transitions Om-Phe-Glu-NH2 is very potent but has only slight and appeared to cover a fairly large amount of available u preference for p receptors over 6 receptors, it represents conformational space. In the case of the parent peptide an ideal parent peptide to study the effects of additional H-Tyr-~-Orn-Phe-Glu-NH2,examination of 20 conconformational restrictions at the 3-position residue on formations sampled at 5 ps intervals along one trajecthe conformational behavior and on opioid receptor tory demonstrated that this cyclic peptide is structurally affinity and selectivity. We report here molecular dy- still quite flexible (Fig. 2). In agreement with the results namics simulations of the essentially non-selective par- of molecular mechanics studies of related peptides (7, ent peptide H-Tyr-D-Om-Phe-Glu-NHz, of the three 12), the Tyr' residue showed greatest structural flexjpotent and very p-selective analogs H-Tyr-D- bility due to relatively unhindered rotation around the Om-Aic-Glu-NH2, H-Tyr-~-Orn-Atc-Glu-NH2 and four exocyclic rotatable bonds at the N-terminus. The u u Phe3 side chain was also very mobile during the simH-Tyr-~-Orn-~-Atc-Glu-NH2, and of the weakly ac- ulations, behaving somewhat like a spinning top. Be-
-
-
-
-
u
tive analog, H-Tyr-D-Om-Tic-Glu-NH2 (9) (Fig. 1). I
1 H-Tyr-~-Orn-Phe-Glu-NH2 U
2 H-Tyr-~-Orn-Aic-Glu-NH2 U
3 H-Tyr-D-Om-Atc-Glu-NH2 U
4 H-Tyr-~-Orn-Tic-Glu-NH~ U
FIGURE 1 Structures of cyclic dermorphin tetrapeptide andogs 1-4. 250
cause the high dielectric constant of water was employed, no strong hydrogen bonds were observed in these simulations. Replacement of Phe3 with the various conformationally restricted aromatic amino acids produced conformational constraints both in the side chain and in the peptide backbone at the 3-position. Backbone torsional angles (@,$) for the Aic- and Atc-containing peptides were limited to values around $3 = - 50", $3 = - 50" and $3 = + 50", $3 = + 50" during most of the time period of the simulations (Figs. 3 and 4), in agreement
Molecular dynamics of opioid peptides
0
50
100
50
0
2
1
d)
t 1 8 0 F
0
50
100
0
50 TIME (PSI
TIME (PS)
3
-
4
FIGURE 2 Twenty snapshots of H-Tyr-~-Om-Phe-Glu-NH2(l),H-Tyr-D-Om-
-
I00
TIME (PS)
TIME (PS)
100
FIGURE 4 $3 angle of analogs 1 (a), 2 (b), 3 (c) and 4 (d), monitored along the dynamics trajectories.
L
analog are presented in the Figures. In the Ticcontaining analog the @ and $3 angles were found to be 0 " and 90 O , respectively, and did not change for the gen atoms were omitted. entire duration of the simulations. This observation reflects the structural rigidity of the peptide backbone at the 3-position in this peptide as we11 as the fact that the bond defining the @ angle is part of two fused ring structures. Due to the five- or six-membered ring structure present in the conformationally constrained amino acids Aic and Atc, the range of accessible side chain torsional angles (XI, 2 2 ) is severely limited in these res. . ., . . . , idues (Figs. 5 and 6). While in the case of phenylalanine b 50 100 0 50 100 TIME (PS) TIME (PSI the x 1 angle can assume three different values [g+ (t 60°), g- (-60°), and t (180")l and preferred values for x2 are 90", the side chain torsional angles of the conformationally constrained amino acids can assume only two values. Thus, the side chain torsional angles of Aic are limited to values of 1 1 = - 80°, x 2 = -20" and X I = -160", x 2 = + 2 0 ° . For Atc the -90 !f two side chain conformations are characterized by the torsion angles x1 = 180" (t), x 2 = + 25 and X I = - 60" -1804-LrA-. (g - ), x 2 = - 25 O, whereas for D-Atc the corresponding 0 50 100 0 50 100 TIME (PS) TIME (PSI values are X I = 180" (t), x 2 = -25" and X I = +60° (g+ ), x 2 = + 25". In the case of L-Tic, the accessible FIGURE 3 values are = t 60" (g+), x 2 = - 25" and 311 = - 60" @ angle of analogs 1 (a), 2 (b), 3 (c) and 4 (d), monitored along the (g- ), x 2 = + 25 ". In the course of the molecular dydynamics trajectories. namics simulations, the conformationally restricted amino acid residues in all these compounds spent about with results obtained from studies with Ca-methyl equal time in each of the two accessible side chain amino acids (13). In all the dynamics simulations, the conformational states. While the substitution of the cyclic phenyIalanine behavior of the L - A ~ cand ~ ~-Atc~-containing peptides was similar and, therefore, only the results for the L - A ~ c ~analogs severely reduced the side chain- and backbone Aic-Glu-NHZ (2), H-Tyr-~-Orn-Atc-Glu-NH2 (3), and H-Tyr-D-
I
Om-Tic-Glu-NHt (4), taken along the dynamics trajectories. Hydro-
-1801
--
'1
1
-
,
i
O
25 1
B.C. Wilkes and P.W. Schiller volume of conformational space. Thus, the rank order of the relative mobility of these various residues was Phe>Aic>L- and D-Atc= Tic, but the rank order of the relative volume of conformational space available to these sameresidues was Phe > L- and D-Atc = Tic > Aic. The fact that the Aic-containing peptide as well as the L- and D-Atc-containing analogs are potent and selective p receptor agonists suggests that the positioning of the Aic residue in H-Tyr-~-Om-Aic-Glu-NH2may be I_) nearly optimal for the interaction with the p-receptor. In Fig. 2, 20 snapshots of H-Tyr-~-Om-Aic-Glu-NH2 1_1
-w
2
+go o
W z 4
z c3 -90
a -90
-180
-180
0
50 TIME (PSI
100
0
50 TIME (PS)
100
FIGURE 5 x: angle of analogs 1 (a), 2 (b), 3 (c) and 4 (d), monitored along the dynamics trajectories.
conformational mobility at the 3-position residue in these compounds, transitions between the various allowed conformations were nonetheless observed in the course of the performed molecular dynamics simulations. A transition in the @ angle of Atc or Aic from + 50" to -50" was often (but not always) followed by a corresponding change in the $3 angle from + 50" to -50" within a few ps (Figs. 3 and 4). When a conformational transition in the side chain of any of the conformationally constrained amino acid residues occurred, both the X I and the xz angle changed simultaneously (Figs. 5 and 6 ) . However, transitions between the two allowed side chain conformational states occurred independently from backbone conformational transitions in all the compounds studied. Interestingly, the molecular dynamics simulations showed that in thesecyclic opioid peptides the 3-position residues containing a six-membered ring (L- and D-Atc and L-Tic) underwent fewer side chain conformational transitions than the Aic3 residue, which contains a 5-membered ring, or than the unrestricted Phe3 residue (Figs. 5 and 6 ) . In order for the constrained amino acids to undergo a successful side chain conformational transition, the torsional angles in the five- or sixmembered rings of these amino acids must change in a concerted manner. One interesting aspect of the obtained results was that, while the five-membered ring contained in the Aic residue of the Aic3-analog was shown to undergo a greater number of conformational transitions than the six membered rings of the Atc3- and Tic3 residues in these cyclic analogs, the Aic residue covered a smaller amount of torsional space than the Atc- and Tic residues and, therefore, covered a smaller 252
taken at 5 ps intervals along the dynamics trajectory are superimposed. This figure clearly demonstrates the limited conformational space available to the Aic side chain. The low activity observed with the Tic-containing peptide, H-Tyr-D-Om-Tic-Glu-NH2, may not be due to the conformational .restriction imposed by the Tic residue per se, but may be due to the fact that the sixmembered ring of Tic is formed through the amine nitrogen of this residue. A related compound containing an Nz-methyl-Phe residue, H-Tyr-DOm-NMePhe-Glu-NHz, was also found to have weak w activity (9), suggesting that the loss of activity shown by both the Tic- and the NMePhe-containing peptide may be due to the unavailability of the amide proton in the 3-position of the peptide sequence for the formation of an important hydrogen bond with a receptor moiety. An alternative explanation is that the steric bulk around the a-nitrogen in these compounds may just not be tolerated at the receptor.
-
0
50
100
50 TIME ( P S )
0
TIME (PS)
1 -804 1 -
0
50 TIME (PS)
100
-1804~, 0
.
,
,
,,
.
50 TIME (PSI
100
,
,
I
100
FIGURE 6 1; angle of analogs 1 (a), 2 (b), 3 (c) and 4 (d), monitored along the dynamics trajectories.
Molecular dynamics of opioid peptides Substitution of the cyclic phenylalanine analogs for Phe3 in the parent peptide in general caused primarily a local conformational restriction of the peptide backbone around the 3-position residue but had a lesser effect on the conformational behavior of the rest of the peptide ring structure, with the exception of the Ticcontaining analog. Monitoring of the $ angle of the Glu4residue along the dynamics trajectory (Fig. 7) provided an indication of the relative structural flexibility of the 14-membered peptide ring structure in these compounds. Several transitions in the x: angle were observed for the Phe-, Aic- and Atc-analogs, but not for the Tic-analog. In the case of the Phe-, Aic- and Atcanalogs, transitions between the various side chain conformers of the D-0m2 and Glu4 residues were usually observed several times during each simulation and in many cases several adjacent torsional angles underwent simultaneous transitions. Several 180' amide bond rotations (but not cis-trans isomerizations) were observed in a few of the dynamics simulations, but this event was rare (data not shown), and did not occur during any of the simulations of the Tic-containing peptide. In fact, in two of the four dynamics simulations performed with H-Tyr-~-Om-Tic-Glu-NH2(Figs. 3d, u 4d and 7d), the peptide was held rigid with the 14membered ring vibrating around a single conformation. In all compounds studied the exocyclic tyrosine residue displayed high conformational flexibility, as indicated by the frequent fluctuations in the $1 angle (Fig. 8) and in the x: and xi angles (data not shown) observed dong the dynamics trajectory. Because of this orientational freedom of the Tyrl residue (see also Fig. 2) the intramolecular distance between the two aromatic rings
d
W
-90 -180
50
TIME (PSI
. . .
-180 50 TIME (PS)
100
50
0
100
TIME (PSI
loo
0
.
I
,
50
.
'
50 (PS)
100
0
TIME
50 TIME (PS)
100
FIGURE 8 $1 angle of analogs 1 (a), 2 (b), 3 (c) and 4 (d), monitored along the dynamics trajectories.
observed in these computer simulations for both the active analog! and for the inactive one covered a wide range (3-17 A). CONCLUSION
-
The observation that in two of the loops molecular dynamics runs of H-Tyr-~-Orn-Tic-Glu-NH2no conformational transitions of the 14-membered ring structure were observed, whereas in the two other runs such transitions did occur, clearly suggests that both relatively long simulations and the use of different starting conformations are necessary to adequately sample the conformational space available to these peptides. In addition, the dynamics simulations may have to be run at high temperatures in order to overcome transitional barriers. For example, in a recent molecular dynamics study (14) performed with the cyclic enkephalin apalog H-Tyr-D-Pen-Gly-Phe-D-Pen-OH at 3 16OK only one starting conformation was used and, therefore, only limited sampling of the conformational space around the starting conformation may have been achieved. More extensive molecular dynamics studies at 1000°K (15) found additional conformations. Through incorporation of additional conformational constraints at the 3-position we obtained a series of opioid peptide analogs which in comparison with the cyclic parent peptide H-Tyr-D-Om-Phe-Glu-NHz were L-----' considerably more rigid but still displayed some limited structural flexibility. Replacement of Phe3 with the various cyclic phenylalanine derivatives produced analogs showing greatly reduced mobility of the peptide back-
-
o
z 0
0
0
I
100
TIME (PS)
FIGURE 7 $ angle of analogs 1 (a), 2 (b), 3 ( c ) and 4 (d), monitored dong the dynamics trajectories.
253
B.C. Wilkes and P.W. Schiller bone around the 3-position residue in all cases. The rest of the 14-membered peptide ring structure in the Aic3and Atc3-analogs also showed reduced structural mobility in comparison with the Phe3-parent peptide, but the most drastic rigidification of the entire peptide ring structure was observed with the Tic3-analog which in two of the four molecular dynamics simulations showed no conformational transitions at all for that part of the molecule. As expected, the side chains of the conformationally restricted 3-position residues showed greatly reduced structural flexibility, insofar as transitions between only two side chain conformational states were observed. A particularly interesting result of the simulations was the observation that the Aic3 residue underwent more frequent conformational transitions than the other cyclic Phe residues in these compounds, but at the same time had the least amount of torsional space available. The exocyclic tyrosine residue displayed high conformational mobility in all compounds studied. Obviously, additional conformational constraints will have to be introduced at the Tyr' residue in order to obtain more definitive insight into the receptor-bound conformation of this class of cyclic opioid peptides.
2. Fishman, A.J., Riemen, M.W. & Cowburn, D. (1978) FEES Lett. 94, 236-240 3. Schiller, P.W. & DiMaio, I. (1982) Nature 297, 74-76 4. Schiller, P.W., Nguyen, T.M.-D. & Maziak, L.A. (1985)J. Med. Chem. 28, 1766-1771 5. Mosberg, H.I., Hurst, R., Hruby, V.J., Gee, K., Yammura, H.I., Galligan, J.J. & Burks, T.F. (1983) Proc. Nafl. Acad. Sci. USA 80, 5871-5874 6. Mosberg, H.I., Omnaas, J.R., Medzihradsky, F. & Smith, G.B. (1988) Life Sci. 43, 1013-1020 7. Wilkes, B.C. & Schiller, P.W. (1987) Biopolymers 26,1431-1444 8. Wilkes, B.C. & Schiller, P.W. (1991) J . Cornput.-Aided Mol. Design 5, 293-302 9. Schiller, P.W., Weltrowska, G., Nguyen, T.M.-D., Lemieux, C., Chung, N.N., Marsden, B J . & Wilkes, B.C. (1991)J. Med. Chem. 34, 3125-3132 10. Clark, M., Cramer 111, R.D. & Opdenbosch, N.V. (1989) J . Comp. Chem. 10, 982-1012 11. Verlet, L. (1967) Phys. Rev. 159, 98-103 12. Wilkes, B.C. & Schiller, P.W. (1990) Biopo!ymers 29, 89-95 13. Marshall, G.R. & Bosshard, H.E. (1972) Circulation Res. (Supp. II) 30 and 31, 11143-11150 14. Smith, P.E., Dang, L.X. & Pettitt, B.M. (1991)J. Am. Chem. SOC. 113, 67-73 15. Pettitt, B.M., Matsunaga, T., Al-Obeidi, F., Jehrig, C., Hruby, V.J. & Karplus, M. (1991) Biophys. J. 60, 1540-1544
ACKNOWLEDGMENTS This work was supported by grants from the Medical Research Council of Canada (grant MT-10131) and the National Institute on Drug Abuse (grant DA-04443). We thank Johanne Duhaime and Franqois Boucher for the excellent upkeep of the computer facility and Helen Zalatan for typing the manuscript.
REFERENCES 1. Schiller, P.W. (1984) in The Peptides: Analysis, Synthesis, Biology (Udenfriend, S . & Meienhofer, J., eds.), pp. 219-268, Academic Press, Orlando, FA
254
Address: Dr. Peter W. Schiller Dr. Brian C. Wilkes Laboratory of Chemical Biology and Peptide Research Clinical Research Institute of Montreal 110 Pine Avenue West Montreal, Que. Canada H2W 1R7
