new insights on μ/δ selectivity of opioid peptides: conformational analysis of deltorphin...

New Insights on p/6 Selectivity of Opioid Peptides: Conformational Analysis of Deltorphin Analogues

1. TANCREDI,' P. A. TEMUSSI,' D. PICONE,2 P. AMODEO,' R. TOMATIS,3 S. SALVADOR1,3 M. MARASTON1,3 V. SANTAGADA,' and G . BALBONI3

' Istituto Chirnica M.I.B. del CNR, Via Toiano 6, 80072 Arc0 Felice, 'Dipartirnento di Chimica, University 01 Naples, via Mrrrocannone 4, 801 34 Napoli, and 3Dipartimento di Scienze Farmaceutiche, University of Ferrara, via Scandiana 21, Ferrara, Italy

SYNOPSIS

The message domain of dermorphin (Tyr-D-Ala-Phe) , a natural p-opioid heptapeptide, has long been considered the main cause of the high p selectivity of this peptide and of its analogues. The recent discovery, in the skin of Phyllomedusa sauvagei (i.e., the same natural source of dermorphin) and of Phyllomedusa bicolor of deltorphins, challenges this belief. Deltorphins, in fact, are three heptapeptides characterized by a message domain typical of p-selective peptides, but endowed of an extremely high 6 selectivity, the highest of all natural opioid peptides.

A conformational analysis of dermorphin and deltorphins, based on nmr studies in DMSO and cryoprotective mixtures and internal energy calculations, showed that the enormous differences in receptor selectivity can be interpreted on the basis of receptor models for p and 6 opioids that recognize the same @-turn in the N-terminal part, but discriminate for the conformation and polarity of the C-terminal part.

Here we present the synthesis, biological activity, and conformational analysis in solution of three deltorphin analogues with very similar constitution, but with different net charge, different location of negative residues, or even without negative residues, which confirm these hypotheses and show that His4 can play a specific structural role.

INTRODUCTION

Deltorphins are natural opioid peptides'-3 of the family of dermorphin, a potent p-selective peptide isolated from frog skin.4 They share with dermorphin the same or a very similar tripeptide message sequence ( Tyr-D-Ala-Phe for dermorphin and deltorphins I and 11, and Tyr-D-Met-Phe for deltor- ph in) , but contrary to dermorphin, are very 6 selective. These peptides may provide a better insight into the requirements for specific 6 or p activity (and selectivity, of course) not only with respect to en- kephalins ( which have lower activity, poor selectivity, and high conformational flexibility) and enkephalin analogues, but also with respect to alkaloid opiates, a t least for the fact that they are much bigger molecules: morphine analogues have molecular

Biopolymers, Vol. 31, 751- 760 (1991) (0 1991 . lohn Wiley & Sons, Inc. CCC 0006-35~5/91/0607~1- 10$04.00

weights on the order of 300-400 amu, whereas deltorphins have molecular weights of the order of 700- 1000 amu.

The presence of one acidic (negatively charged) residue in each of the sequences of natural deltorphins hints a t a possible role of membrane preselection, as proposed by Schwyzer.5 According to this model a positive net charge favors p selectivity since the p receptor is probably located on the negatively charged membrane surface. Conversely, a decrease of the net charge of the peptide might he expected to favor 6 selectivity. Even a superficial survey of literature data on opioid peptides, however, shows that peptides with net charge f l are not uniformly p selective, and those with zero net charge are not predominantly 6 selective.6 Besides, the change in selectivity from dermorphin to deltorphins is so large (i.e., four orders of magnitude, see Table I ) that it is inconsistent with all known examples attributed to changes of net charge. The causes of selectivity

75 1

752 TANCREDI ET AL.

must then be sought not only in changes of net charge, but also in specific receptor requirements that must be met by conformational and constitutional features of the message and address domains of the peptides. We have recently shown7 that Tyr- D-Ala-Phe, in a p-turn conformation, is the optimal message domain for both 6- or p-opioid peptides. It is the purpose of this work to investigate the relative importance of net charge and of the conformation of the two domains (i.e., message and address) in three deltorphin analogues, designed with strategic changes of net charge and charge location: Tyr-D- Met-Phe-His-Leu-Met-Asp-OH (net charge -1; henceforth called DTOH) , Tyr-D-Met-Phe-His- Leu-Met-Asn-OH (net charge 0; henceforth called DTNOH ) , and Tyr-D-Met-Phe-His-Leu-Met-Asn- NH2 (net charge +l; henceforth called DTN) . Table I summarizes the activity and selectivity data of dermorphin, deltorphins, and three deltorphin analogues. It is easy to see that the changes in activity and selectivity are surprisingly large, with respect to relatively minor constitutional changes; therefore it is necessary to perform a preliminary conformational analysis before attempting any stmcture-activity correlation. The conformational features of these peptides in solution have been studied by a combination of two-dimensional (2D) nmr and internal energy calculations.

EXPERIMENTAL

Materials

Deltorphin was synthesized as described in Ref. 6. The three analogues were synthesized by the con- ventional method in solution as illustrated in the scheme of Figure 1. It involves the preparation of the protected segments IV and VI. Segment IV was

obtained by stepwise elongation starting from a C- terminal residue similar to that reported for DT6 and incorporated using the Rudinger azide procedure. Segments VI were obtained in good yields via 2 + 1 coupling. Na-Fmoc-Leu-Met-OH was then coupled to the aspartic acid a,P-di-t-butyl ester, asparagine t-butyl ester, and asparagine amide, re- spectively, using the dicyclohexyl-carbodiimide! HOSu (Hydroxysuccinimide) condensation method followed by a deblocking step with 10% dimethyl- amine solution in dimethylformamide.

Condensation of compound IV with different de- blocked tripeptides VI, i.e., H-Leu-Met-Asp ( OBut) - OBut, H-Leu-Met-Asn-OBut, and H-Leu-Met-Asn- NH2, was obtained by means of the azide procedure.8

The analogues were purified by silica gel column chromatography or by high performance liquid chromatography ( HPLC; Waters Delta Prep 3000 Preparative Chromatography System with a Delta Pak C18-300 A 30 mm X 30 cm, 15 p spherical) using a linear gradient of 40-70% CHBCN in 0.1% aqueous trifluoroacetic acid in 30 min at a flow rate of 30 mL/min with uv detection at 220 nm.

Purity was assessed by thin layer chromatography in two solvent systems on silica gel and by analytical high performance liquid chromatography.

‘H-nmr spectra were obtained for each analogue, and found consistent with the sequence and structure of the peptides.

All peptides were evaluated for their activity in two different radioligand assay systems, essentially as described by Gillan et al.’

[ 3H] [D-Ala’, MePhe*, Gly-01‘1 enkephalin (DAGO; 0.8 nM, 45 Ci mmol-’) and [3H] [D-Ser’, Leu’] enkephalin (DSLET; 1.5 nM, 40 Ci mmol-’) , purchased from New England Nuclear (USA), were used as labeled ligands specific for p and 6 receptors in rat brain membrane preparations. Incubations were carried out a t 25°C for 40 min in

Table I Sequences, Net Charge, and Selectivity Data of Deltorphins and Deltorphin Analogues

Peptide Charge (IC5,),” (IC5o)ab S/PC

Dermorphin Y -a-F-G-Y -P-S-NH2 +1 3.2 294 0.01 Deltorphin N Y-m-F-H-L-M-N-NH, +1 (+2) 380 20 19.0 Deltorphin NOH Y-m-F-H-L-M-N-OH 0 (+ I ) 660 22 30.0 Deltorphin OH Y -m-F-H-L-M-D-OH -1 ( 0 ) 600 18 33.0 Deltorphin Y -m-F-H-L-M-D-NH, 0 (+1) 1200 9 133.0 Deltorphin I Y -a-F-D-V-V-G-NH, 0 910 1.5 607.0 Deltorphin I1 Y -a-F-E-V-V-G--NH, 0 3200 3.8 842.0

a IC5, (nM) in inhibiting the binding a t the p receptor (vs [3H] DAGO), mean (?SEM) of 3-6 experiments carried out in duplicate. IC50 (nM) in inhibiting the binding at the 6 receptor (vs [3H] DSLET), mean ( S E M ) of 3-6 experiments carried out in duplicate. Selectivity ratio at the 6 sites, expressed as the ratio (IC50),,/(IC50)6.

BOC

BOC

BOC,

BOC,

B O C .

H.

Boc -- OSu H -- One

Boc One Boc -- OSU H -- I

TFA V BOC --OSu H OMe Boc

Boc OMe Boc

OSu H OMe H

One Fmoc

NH-NH Fmoc

I1 NIOH

TFA HC1

Fmoc-CI . I11

IV NH-NH VI 2 2

2 2 DMA

H N3

TFA 2 - 4

NEW INSIGHTS ON p/6 SELECTIVITY OF OPIOID PEPTIDES 753

TYr D -Met Phe H i s Leu Net xu

OW

O M 4

OH

OH

OH H -

DCC, HONSu

~ ~~~~~~~~~~~~~~~~~~~~~~~~ ~ ~

Figure 1. Scheme for the synthesis of compounds 2 (Xaa = Asp), and 3 and 4 (Xaa = Asn). Y: functional group of amino acid Xaa (NH,, OH) , or protecting group of the COOH function of amino acid Xaa (OBu' for Asp and Asn); P: protecting group of the COOH side-chain function of amino acid Xaa (OBu' for Asp).

L Y

L Y

L Y

L Y

- Y

50 m M Tris HC1 buffer (pH 7.4). Dose-response curves were constructed utilizing 6-10 dose levels and analyzed by computer programs to estimate the ICs0 value, the dose that produces a 50% inhibition of binding of each labeled ligand.

Methods All nmr spectra were run a t 400 MHz on a Bruker AM-400 spectrometer equipped with an Aspect 3000 computer and a variable temperature unit. Samples for nmr measurements were 6 m M in neat 99.98% DMSO-d, (Aldrich, Milwaukee, WI). All chemical shifts are referred to internal tetramethylsylane. Phase-sensitive double quantum filtered correlated spectroscopy (DQF-COSY ) , lo nuclear Overhauser enhancement spectroscopy (NOESY) ,I1 and total correlated spectroscopy (TOCSY) '' spectra were run a t 300 K. NOESY spectra were acquired using a 300-ms mixing time. TOCSY spectra were acquired in the low-power transmitter mode; the 90" flip angle in this condition was 70 ps. The total mixing time was 45 ms.

The united l3 and all-atoml4 parametrizations of the AMBER force field, as implemented in the SY - BIL package, were used in a series of energy minimization ( E M ) calculations on D T and its ana-

logues. The computational procedure can be divided into the following steps: (1) different starting guesses are generated (see below) ; ( 2 ) a united atoms EM calculation is performed, using a quasi- Newton method, the Broyden-Fletcher-Goldfarb- Shanno (BFGS) stopping when the gradient norm is lo-' or less; and (3) nonpolar hydrogen atoms are added to the resulting structure, and an all-atom BFGS EM is run until a gradient norm of lop3 is reached.

The starting structures used in E M calculations were obtained by means of two procedures:

1. A set of structures consistent with nmr data (see Results and Discussion) was generated. Owing to the nonunivocal nature of the nmr data on the first (N-terminal) and last two ( C-terminal) residues, different conformations relative to these residues were examined. In addition, a complete search on the dihedral angles describing the rotation of the two aromatic rings of Tyr' and Phe3 of D T was performed.

2. Starting from the lowest energy minimum found in set 1, several simulated annealing runs were performed, according to the follow-

754 TANCREDI ET AL.

ing scheme: ( a ) A set of random atomic ve- locities is chosen for the all-atom starting structure from a Boltzmann distribution at 2000 K; ( b ) 5 ps of constant temperature molecular dynamics ( M D ) simulation is performed, with the following parameter settings: temperature ( T ) = 2000 K and time step (ts) = 0.2 fs; ( c ) 2 ps of MD with T = 1000 K and ts = 0.2 fs; ( d ) 1 ps of MD with T = 500 K and ts = 0.2 fs; ( e ) 1 ps of MD with T = 250 K and ts = 0.2 fs; ( f ) 1 ps of MD with T = 100 K and ts = 0.5 fs; (g ) 1 ps of MD with T = 50 K and ts = 0.5 fs; ( h ) 1 ps of MD with T = 10 K and ts = 0.5 fs; ( i ) the structures generated in run ( h ) are analyzed, chosing the lowest potential energy conformation; ( j ) the ( i ) structure is replaced in step ( a ) for a new annealing cycle; ( k ) the ( i ) structure is fully minimized (see above) using BFGS algorithm.

In the final analysis only those EM conformers that are 20 kcal/mol or less higher in energy than the absolute minimum of the set have been considered.

These final D T conformations have been used also as starting structures for D T analogues.

Solvation effects may be relevant in solutions of medium-high polarity solvents, such as the media employed in the present study; in order to account for them three different approaches were used in preliminary calculations. In particular, the following procedures were tested on a single conformation of each considered molecule: ( 1 ) a distance-dependent dielectric constant t = r ; ( 2 ) a fixed t = 10 value; and ( 3 ) a damped distance depending t = 10 r . Apart from a general increase in the average hydrogen bond lengths, and consequently, a decrease of their stability, the only noteworthy effect of increasing the value of t is represented by the substantial weak- ening of the ionic interaction between the Asp carboxyl group and the N-terminal NH: . However, the folding patterns here proposed for each molecule are essentially insensitive to t variations, whereas side chains and terminal regions are affected to a larger extent. Consequently all final calculations and the corresponding conformations reported under Con- formational Energy Calculations were obtained with t = r .

RESULTS AND DISCUSSION

NMR Study

All peptides of Table I were studied a t 400 MHz in DMSO-d, a t several tempcratures. Residue types

and sequential assignments were performed by means of standard 2D techniques: DQF-COSY, lo

NOESY, l 1 and TOCSY." Chemical shifts of the la- bile protons and of Met y protons are summarized in Table 11.

In all peptides the side-chain protons of the second residue (particularly the y protons) are shifted to unusually high field. This finding is consistent with similar observations on deltorphin itself, 'J' dermorphin, 2o and its analogues, 7,21 and seems to be a general feature of opioid peptides containing the Tyr-D-Xaa-Phe message domain. I t can be interpreted as a strong indication that the side chain of the second residue is sandwiched between the aromatic rings of Tyr and Phe3, and suggests that the backbone conformation of the message domain is characterized by a type 11' P-turn whose central residues are D-Met and Phe: this kind of arrangement had been previously proposed, both for p and 6 ag- onists, on the basis of a comparison of the conformational behavior of the N-tetrapeptide fragment of dermorphin ( Tyr-D-Ala-Phe-Gly-NH,) with the structures of many nonpeptidic opiates and, in particular, with the structure of methyl fentanyl?I2' The chemical shifts of the other nonexchangeable protons are not outstanding; however, they differ from typical "random coil" values reported in the literature, a further indication that there is a t least a fraction of ordered conformations in equilibrium with extended forms. Most of the NH temperature coefficients are close to -5 ppb/K, a value typical of exposed protons, but the distribution around -5 ppb/K is not uniform, as can be expected from a modulation resulting from complex conformational equilibria.

Table I1 Chemical Shifts (6) of NH and of Protons of Deltorphins in DMSO"

DT DTN DTNOH DTOH

Tyr' Met' Met' (yCH2) Phe" His4 Leu5 Met6 Met' (yCH,) Asp' Asn'

8.15 8.47 8.46 1.80 1.84 8.41 8.41 8.52 8.53 8.11 8.13 8.28 8.26 2.40 2.44 8.02

8.01

8.49 1.83 8.42 8.53 8.15 8.19 2.44

8.11

8.47 1.83 8.43 8.47 8.09 8.24 2.44 8.16

The 6's are in parts per million from internal tetramethyl- silane.

NEW INSIGHTS ON p / 6 SELECTIVITY OF OPIOID PEPTIDES 755

These indications can be put on a more quanti- tative basis by measuring NOEs. This task is usually arduous for medium-sized peptides a t high field since their correlation time brings about the condition W ~ T : = 1, which makes all NOEs close to zero." We have been able to show, in the case of enkephal in~, '~ that this condition is not simply due to the molecular weight of these compounds, but rather to the unfavorable coupling of rotational and internal motions. These motions can be partly decoupled by a combination of low temperature and viscosity, assured for instance by the use of cryoprotective mix- t u r e ~ , ' ~ or in favorable cases, i.e., when the peptides have a fairly high intrinsic tendency to assume folded conformations, 25 also by the viscosity of DMSO that, a t room temperature, is about twice as large than that of water. In the case of deltorphin a considerable improvement in the NOESY spectrum was in fact obtained by running the spectrum in a DMSO /water cryomixture a t low t empera t~ re . ' ~ The three analogues, however, do give good NOESY spectra even in neat DMSO-d, a t room temperature. Table I11 summarizes the relevant cross peaks of all four deltorphin peptides: it can be seen that, for the analogues, the number of effects in DMSO-d, is comparable to that of DT in the cryomixture, in-

Table I11 NOESY Spectra of DT, DTN, DTNOH, and DTOH in DMSO"

Relevant NOE Cross Peaks in the

DT DTN DTNOH DTOH

DMSO Cryo DMSO DMSO DMSO

+ + + + + + + + +

+ +

+ + + + + + + + + + + + + + + +

+ + + + + + + + + + + + + + + +

+

+ + + + + + + + + + +

+ + + + + + + + + + + + + + +

'I For DT, data relative to a DMSO/water cryoprotective mix- ture are also reported.

dicating that their tendency to form folded structures is even stronger than that of DT. Accordingly, we decided to limit the comparative conformational analysis to a study in DMSO-4,.

It can be noted that in deltorphin and its analogues there are similar, diagnostically critical, NH- NH effects: H4-L5, L5-M6, and M6-D7 in DT; H4- L5 and L5-M6 in DTOH; H4-L5 in DTNOH; and H4-L5, L5-M6, and M6-D7 in DTN, which suggest that folded conformations characterized by a succession of @-turns. Moreover, the effects are probably even more numerous, since the close prox- imity of the diagonal peaks makes it impossible to identify other significant NH-NH NOEs, in particular for the N-terminal region. It must be noted, however, that the presence of a type 11' 8-turn in the N-terminal message domain, suggested by chemical shift data (vide supra), is substantiated by other NOEs, in particular B1-N2, F2,-/3' and YA,-N' as reported in Table 111.

Although it is very difficult to derive a detailed unique conformation from the nmr data alone, it is possible to interpret NOESY data in general terms to say that deltorphin and its analogues show a marked tendency to form folded conformations similar to those of deltorphins I and I1 in the DMSO /water cryomixture.6 A common backbone structure, however, although consistent with the high activities of all three analogues, cannot explain the changes in selectivity since the constitutional differences are small and contradictory with respect to the accepted role of net charges.5

The 6 selectivity decreases by a factor of 4 when going from DT to DTOH, i.e., with the only apparent change of the decrease of net charge from 0 to -1; moreover, the persistence of a fairly high 6 selectivity is assured entirely by the high value of 6 activity since the p activity increases from an ICs0 of 1200 to an ICs0 of 600, in complete contradiction of the possible membrane preselection mechanism based on net ~ h a r g e . ~ On the other hand, DTNOH, which differs from deltorphin for the position of the carboxyl group in the last residue but has the same net charge, shows a decrease of 6 selectivity by a factor of 4.4; a surprisingly large factor even from the point of view of constitution, since DT and DTNOH can be considered as position isomers. Even more sur- prising is the retention of considerable 6 selectivity in DTN, a peptide that would be expected to be p selective since it bears a positive net charge and has a constitution very similar to that of DT but also to some p-selective dermorphin analogues.26

Since the overall backbone structures have been shown, on the basis of the nmr data, to be rather

756 TANCREDI ET AL.

similar, it is necessary to resort to internal energy calculations to find possible local differences among the three peptides.

Conformational Energy Calculations

Nuclear magnetic resonance data can be interpreted in terms of several consecutive @turns, provided we allow for a certain variation of the “canonical” cp, rc/ angles. That is, it is possible to build a unique folded conformation characterized by a type 11’ @-turn involving residues 1-4, followed by two consecutive type I P-turns involving residues 2-5 and residues 3-6, provided the cp, $ angles of the “junction” between the type 11’ ,f3-turn and the initial type I ,& turn are a compromise between the best values for the i + 2 residue of the 4 + 1 turn and the best values for the i + 1 values of the 5 + 2 turn.

A model characterized by the following internal rotation angles-D-Met’ ( c p = +60; rc/ = -120); Phe3 (cp = -70; 1c/ = -15); H is4 (cp = -70; rc/ = -15); Leu5 (cp = -90; rc/ = 0) -was used as input of internal energy calculations based on the united13 and all atoms14 parametrization of the AMBER force field (see Experimental). In order to account for solvation effects, three different approaches were used in preliminary calculations: ( 1 ) a distance-dependent dielectric constant t = r ; ( 2 ) a fixed t = 10 value; and ( 3 ) a damped distance depending t = 10 r . Final calculations were performed with a dielectric constant whose actual value is a function of interatomic distance ( see Experimental). Alternative starting models were built by combining the type 11’ p turn of the first four residues with only one of the other p turns consistent with the NOESY data in the C- terminal part, and by simulated annealing runs, according to the procedure described in Experimental section.

The final analysis of the conformer set so obtained shows that only three minimum energy conformations (labeled I, I1 and 11) are sizably more stable than other structures. The main geometrical parameters associated with these structures are reported in Table IV. Structure I is slightly higher in energy but resembles the structure suggested by nmr data, being characterized by backbone angles consistent with a set of p-turns (with the partial ex- ception of $4 and q5), whereas structures I1 and 111, characterized by a series of y-turns, reflect optimal adaptation to local low-energy situations.

It is important to point out that each conformer ought to be regarded as a set of accessible conformations, since rotation of backbone terminal regions and of selected side chains gives rise to several quasi- isoenergetic structures (see below). However, the flexibilities of the three conformers are fairly different in some relevant features, strongly affecting the comparison with experimental NOEs and the compatibility with the changes in chemical constitution explored in the three analogues (vide infra) .

In fact, while no EM structure can be ruled out on the basis of intrabackbone NOE distances, the relative orientations of Tyr’, D-Met’, and Phe3 side chains, and the limited mobility allowed to aromatic residues (especially Phe ) in both annealing-derived structures (I1 and I11 of Table IV), are not able to account simultaneously for the characteristic high- field shift of the side-chain peaks of D-Met’ protons and for NOEs of Tyr’, D-Met’, and Phe3 side chains.

A description of the EM-refined structure and its comparison with NOE data is therefore performed exclusively on structure I of Table IV, discussing only the main discrepancies found in conformers I1 and 111.

The results of this conformational analysis can be summarized as follows: The starting structure of

Table IV Resulting from EM Calculations

Relevant Internal Coordinates of the Main Molecular Models of DT,

I I1 I11

Residue cp 1c/ x1 x2 x3 cp 1c/ x1 xz x3 P 1c/ x1 x2 x3

Tyr’ - 150 180 -95 - -53 -174 14 - -57 -56 97 met’ 50 -101 62 177 180 77 84 -169 -65 76 -69 66 65 -70 180 Phe3 -52 -48 176 -132 -140 170 -175 60 -66 69 -54 -55 His4 -90 108 -57 169 -56 -49 -177 -11 -177 -42 179 -12 Leu‘ -64 -77 180 60 -82 52 -55 180 -91 -167 -67 -59 Met‘ -47 -45 -175 177 180 66 -69 -161 -69 -64 -77 56 -61 179 180 Asp’ -41 - 171 -49 -149 - -160 -87 59 - -171 -93


DT refined to a minimum energy conformation (see Figure 2 ) characterized by a Clo comprising Y '-m2- F3-H4, a C7 comprising F3-H4-L5, a CI3 comprising H4-L5-M6-D7, and the terminal NH2 and a CI7 whose end groups are the side chains of H 4 and D 7 .

The most prominent calculated hydrogen bonds, i.e., characterized by H-0 distances of the order of 0.2 nm or less, are Y ' CO-H4 NH, F3 CO-L5 NH, H 4 CO-D7 NH2, L5 CO-D7 NH2, D7 COO--imidazol NH, D7 COO--Y1 NH;, H 4 CO-Y1 NH;.

Comparison with the experimental NOEs shows that all prominent cross peaks do have a counterpart in short backbone distances. The relative arrangement of the aromatic rings of Tyr' and Phe3 with respect to the side chain of D-Met2 accounts for the high-field shift of the side-chain resonances of D- Met ', but it is quite likely that the position of the side chains of Tyr and Phe cannot be reproduced exactly by a single, static energy calculation. All side chains have, in general, more conformational free-

dom than backbone protons, but those of Tyr and Phe are particularly flexible, being rather external with respect to the backbone structures. In order to ascertain the degree of conformational flexibility of these side chains, we calculated the energy profiles corresponding to full rotations of the Tyr and Phe side chains as a function of their x angles. That is, a full rigid geometry scan, with 30" resolution, has been performed on x1 and x2 angles of both residues, followed by a complete geometry relaxation in the main relative minima of the map. The results of these calculations show that the positions of the Phe and Tyr side chains corresponding to the three clas- sical staggered conformations ( X I ) differ by less than 500 cal/mol, i.e., small enough values to justify si- multaneous observation of NOEs from different side-chain-side-chain interactions, also in view of the fact that the interconversion barriers are rather small, i.e., even if calculated by keeping the rest of the molecule fixed, they are of the order of 5 kcal/ mol.

Figure 2. minimization of the starting model suggested by the nmr data.

Schematic model of the preferred conformation of DT, obtained from an energy

758 TANCREDI E T AL.

This kind of conformational analysis of the Tyr and Phe side chains showed that the x1 angle of Tyr can only assume two of the staggered values ( +60° and 180") while the third one (-60') is prohibited by highly unfavorable interactions of the aromatic ring with the D-Met side chain.

From this point of view, the conformational behavior of I1 and I11 EM minima differs significantly from that of I, as the rotation of Phe3 side chain is strongly limited by its interactions with backbone and the relative arrangement of D-Met and Phe side

chains prevents D-Met PCHZ protons-Phe aromatic H distances to be less than 5.1 A.

The whole conformational search hitherto dis- cussed for D T was not repeated for its analogues; instead, we simply used, as starting structures, the main conformations found for DT. On these structures a limited search, involving only those residue differing from DT, has been performed. For all analogues, the EM minimum derived from conformer I of D T resulted in by far the most stable structure, with conformational energies of 8-40 kcal/mol lower

Figure 3. Comparison of the schematic models of the preferred conformations of DT (blank balls) and DTOH (full balls), obtained from an energy minimization of the starting models suggested by the nmr data. It is possible to note that substitution of the terminal NH2 with an oxygen favors a double salt bridge with the N-terminal NH;.


with respect to the corresponding structures derived from conformers I1 and I11 of DT.

The final conformers found for these analogues differ only slightly from the original structures; the main results and the comparison among these molecules can be summarized as follows:

All peptides are characterized by the same interaction between the side chains of His4 and of the last residue, a structural feature that is responsible oft he formation of a large hydrophobic surface above the plane of the 0-turn of the message domain. From this point of view, it is interesting to note that His4, in spite of the different charge, plays in DT and its analogues a structural role analogous to those of the acidic residues in the other two deltorphins.'

The most prominent differences in the case of the DT analogues were found, as expected, only in the ('-terminal part of the sequences. DTN is characterized by the most flexible terminal structure. Formation of the CI7 ring is still possible but it is no longer stabilized by the salt bridge between D7 COO Y ' NH;; overall it was not possible to identify a preferred unique structure for the last three residues, also because the absence of the salt bridge makes these residues more exposed to solvent than in DT. In DTNOH the C17 ring is stabilized by a salt bridge between terminal ( a ) D' COO - and Y ' N H t , resulting in a stability slightly lower than in 11'1'.

The most interesting structure is that of DTOH: as can be appreciated in the comparison with the structure of DT (Figure 3 ) , the CI7 ring is stabilized both t q a salt bridge between terminal ( a ) D7 COO- and \- * NH [, and by a salt bridge between terminal ( 8 ) D' ('00- and Y NH;; this feature also allows an additional Clo comprising H4-L5-Mfi-D7 that substitutes the C,, comprising H4-L5-M6-D7 and the terminal NH2 found in DT. The structure of DTOH appears globally more stable than that of DT, albeit characterized by analogous hydrophilic, hydrophobic surfaws. Thus, the main consequence of substituting the terminal amide with a carboxylate seems a sta- bilizat ion of the hydrophobic surface (essential for 6 selectivity6) that compensates the increased hy- drophilicity conferred by removal of the amide terminal

CON CLUS 10 N

The comparative study of deltorphin and of its analogues shows that the Tyr-D-Xaa-Phe message domain and its proposed type 11' 0-turn conformation

are indeed compatible not only with p agonism but equally well with 6 agoni~m.~," Membrane selection does not play a major role in p / 6 selectivity; it can only favor p specificity for peptides with high, positive net charge, but it is not efficacious for peptides with net charges of +1 or lower.

The main causes of p / 6 selectivity are conformational; in particular, they can be traced to the conformation of the C-terminal part. All peptides examined are characterized by similar C-terminal (folded) conformations, which generate a highly hydrophobic surface above the mean molecular plane of the N-terminal 0-turn; formation and stabiliza- tion of these conformations, in the case of DT, DTOH and DTNOH is favored also by electrostatic interactions involving one or two carboxyl groups of Asp( Asn) and by an unusually strong hydrogen bond formed by the imidazol NH of His4 with the terminal CO. The structural role of His4 in deltorphin and its analogues is particularly clear in the case of DTN, in which this hydrogen bond appears to be the main stabilizing force of the C-terminal conformation.

This work was performed as part of the national program Progetto finalizzato Chimica Fine e Secondaria I1 of CNR. The financial support of the CNR is gratefully acknowl- edged.

REFERENCES

1. Richter, K., Egger, R. & Kreil, G. (1987) Science 238, 200-202.

2. Kreil, G., Barra, D., Simmaco, M., Erspamer-Falcon- ieri, G., Melchiorri, P., Negri, L., Severini, C. & Corsi, R. (1989) Eur. J. Pharmacol. 162, 123-128.

3. Erspamer, V., Erspamer-Falconieri, G., Melchiorri, P., Negri, L., Corsi, R., Severini, C., Barra, D., Simmaco, M. & Kreil, G. 3 (1989) Proc. Natl. Acad. Sci. U S A

4. Erspamer, V. & Melchiorri, P. (1980) Trcnds Pharm.

5. Schwyzer, R. (1986) Biochemistry 25, 6335-6342. 6. Balboni, G., Marastoni, M., Picone, D., Salvadori, S.,

Tancredi, T., Temussi, P. A. & Tomatis, R. (1990) Biochem. Biophys. Res. Commun. 169, 617-622.

7. Castiglione-Morelli, M. A., Lelj, F., Pastore, A,, Sal- vadori, S., Tancredi, T., Tomatis, R., Trivellone, E. & Temussi, P. A. (1987) J. Med. Chem. 30, 2067- 2073.

8. Salvadori, S., Sarto, G. P. & Tomatis, R. ( 1982) Int. J . Pept. Protein Res. 19, 536-541.

9. Gillan, M. G. C., Kosterlitz, H. W. & Paterson, S. ( 1980) Br. J. Pharmacol. 70, 481-488.

86, 5188-5192.

Sci. 1, 391-395.

760 TANCREDI ET AL.

10. Rance, M., Smensen, 0. W., Bodenhausen, G., Wag- ner, G., Ernst, R. R. & Wuthrich, K. (1983) Biochem. Biophys. Res. Commun. 117, 479-485.

11. Macura, S. & Ernst, R. R. (1979) Mol. Phys. 41,95- 101.

12. Bax, A. (1985) J. Magn. Reson. 65, 335-360. 13. Weiner, S. J., Kollman, P. A., Case, D. A,, Chandra

Singh, U., Ghio, C., Alagona, G., Profeta, S. & Weiner, P. (1984) J. Am. Chem. SOC. 106, 765-771.

14. Weiner, S. J., Kollman, P. A,, Nguyen, D. T. & Case, D. A. (1986) J. Comp. Chem. 7, 230-242.

15. Broyden, C. G. (1970) J. Inst. Math. Appl. 6, 76-82. 16. Fletcher, R. (1970) Comput. J . 13, 317-322. 17. Goldfarb, D. (1970) Math. Comput. 24, 23-28. 18. Shanno, D. F. (1970) Math. Comput. 24, 647-653. 19. Temussi, P. A., Picone, D., Tancredi, T., Tomatis, R.,

Salvadori, S., Marastoni, M. & Balboni, G. (1989)

20. Pastore, A., Temussi, P. A,, Tancredi, T., Salvadori, S. & Tomatis, R. (1984) Biopolymers 23, 2349-2360.

FEBS Lett. 247, 283-288.

21. Castiglione-Morelli, M. A., Salvadori, S., Tancredi, T., Temussi, P. A., Trivellone, E. & Tomatis, R. ( 1987) in Peptides 1986, Proceedings of the X I X European Peptide Symposium, Theodoropoulos, D., Ed., Walter de Gruyter, Berlin and New York, pp. 443-446.

22. Motta, A., Picone, D., Tancredi, T. & Temussi, P. A. ( 1987) J. Magn. Reson. 75, 364-370.

23. Motta, A., Picone, D., Tancredi, T. & Temussi, P. A. (1988) Tetrahedron 44,975-990.

24. Douzou, P. & Petsko, G. A. (1984) Adv. Protein Chem.

25. Temussi, P. A., Picone, D., Castiglione-Morelli, M. A., Motta, A. & Tancredi, T. (1989) Biopolymers

26. Sagan, S., Amiche, M., Delfour, A,, Camus, A., Mor, A. & Nicolas, P. (1989) Biochem. Biophys. Res. Com- mun. 163, 726-732.

36, 245-361.

28, 91-107.

Received August I , 1990 Accepted November 12, 1990

new insights on μ/δ selectivity of opioid peptides: conformational analysis of deltorphin...

Documents