Letters in Peptide Science, 4 (1997) 359–364

KLUWER/ESCOM 359
© 1997 Kluwer Academic Publishers. Printed in the Netherlands.

LIPS 200

Conformational studies of a benzodiazepine-like peptide in SDS

micelles by circular dichroism, 1 H NMR
and molecular dynamics simulation

Marc Lecouveya, Céline Frochota , Laurent Miclob, Piotr Orlewskia, Michel Marrauda ,
Jean-Luc Gaillardb , Manh Thong Cunga and Régis Vanderessea,∗
a LCPM, CNRS-URA 494, ENSIC-INPL, BP 451, F-54001 Nancy Cedex, France
b LBSA, Unité associée à l’INRA, Université Henri Poincaré, BP 239, F-54506 Vandœuvre lès Nancy Cedex, France

Received 14 March 1997

Accepted 23 May 1997

Keywords: Benzodiazepines; Circular dichroism; GABAA receptor; Micellar medium; Molecular dynamics; 2D NMR

SUMMARY

The conformation of a benzodiazepine-like decapeptide corresponding to the YLGYLEQLLR fragment of a casein

has been examined in a sodium dodecyl sulfate micellar medium using circular dichroism, two-dimensional 1 H NMR
spectroscopy and restrained molecular dynamics simulation. The decapeptide adopts an amphipathic 310 -helicoid structure
in which the E6 · · ·R10 ionic bridge stabilizes the C-terminus.

INTRODUCTION fragment is probably critical for the benzodiazepine-

like activity, and we have investigated its structure
Many authors have evidenced that various en- in a sodium dodecylsulfate (SDS) micellar medium
zymatic hydrolysates of milk proteins contain bio- by circular dichroism (CD), energy minimisation
logically active peptides [1–5], and some of these (EM) and restrained molecular dynamics (MD) using
peptides have been shown to be active in vivo [6]. nuclear Overhauser effects (NOEs) as geometrical
Recently, a tryptic casein hydrolysate has been found constraints.
to decrease pentylenetetrazole-induced convulsions
in rats, to present benzodiazepine-like activity in MATERIALS AND METHODS
a behavioral test revealing anxiolytic properties of
drugs [7], and to have an appreciable affinity to Synthesis and binding tests
the benzodiazepine site of the GABAA receptor [8]. The decapeptide casein fragment was syn-
Only the YLGYLEQLLR fragment of this casein
thesized on /4-(oxymethyl)-phenyl-acetamidomethyl
displays that activity.
(PAM) resin (0.56 mmol/g substitution, Neosys-
New agonists without the secondary effects as-
tem, Strasbourg, France), using Boc solid-phase
sociated with long-term benzodiazepine treatment
peptide synthesis protocols on a Dupont (Wilm-
should be of greatest pharmacological interest. In
ington, DE, U.S.A.) Coupler 250 peptide synthe-
this context, the conformation of the decapeptide
sizer. The peptide chain was assembled by sequential
∗ To whom correspondence should be addressed. coupling of the Boc-amino acids (3 equiv) in the
360

Fig. 1. Evolution of the CD spectrum for the casein fragment as a function of SDS concentration in water. The peptide concentration is
7.9 × 10−5 M in phosphate buffer (pH = 5.9).

presence of 2-(1H-benzotriazole-1-yl)-1,1,3,3-tetra- sorption laser desorption ionisation-time of flight

methyluronium tetrafluoroborate (TBTU, 3 equiv) mass spectrometry (MALDI-TOF MS) was carried
and diisopropylethylamine (DIEA, 9 equiv) in out on a Bruker (Rheinstetten, Germany) Protein
dimethylformamide/dichloromethane (1/3) mixture. TOF spectrometer (M + 1 = 1268.48 (calculated),
Standard cleavage using trimethylsilyl trifluoro- 1268.27 (found)). The IC50 value in binding as-
methane-sulfonate (TMSOTf) and trifluoroacetic says to the GABAA receptor in competition with
acid (TFA) with thioanisole and 1,2-ethanedithiol [3 H]flunitrazepam is 5.9 × 10−4 M.
as scavengers afforded the crude peptide, which
was desalted on Sephadex G-25 and eluted with
CD and NMR spectroscopy
the quaternary solvent BuOH/pyridine/AcOH/H2O
(15:10:3:12). Reversed-phase HPLC was performed CD spectra were recorded at 25 ◦ C on a Jobin
on a Merck (Darmstadt, Germany) L-6200 chro- Yvon (Lonjumeau, France) CD6 spectrophotometer
matograph coupled to a Jasco (Tokyo, Japan) 875 using a 1 mm path-length quartz cell. The decapep-
UV detector using a 1×25 cm Lichrocart packed tide (7.9 × 10−5 M) was investigated in phosphate
with Lichrospherr WP300 RP-18 (5 µm) using a buffer (pH 5.9) with increasing SDS concentration
linear gradient of A: 0.1% TFA in water and B: up to 0.2 M. CD spectra are reported in ellipticity
0.1% TFA in acetonitrile, from 69% A to 39% A units per mole of peptide residue ([θ]R in deg cm2
over 40 min (4 mL min−1 , 214 nm). Matrix ad- dmol−1 ).
 361

NMR experiments were performed at 25 ◦ C

on the casein fragment (5 mM) in H2 O/D2 O
(95:5) containing perdeuterated SDS-d25 (250 mM),
using a Bruker DRX-400 spectrometer and 3-
(trimethylsilyl)propionic-2,2,3,3-d4 acid, sodium
salt (TMPS) as an internal standard. The 2D NOESY
experiments were recorded with four mixing times
(100, 200, 350 and 500 ms) in order to deter-
mine the mixing time interval where no spin diffu-
sion occurs [9]. The WATERGATE sequence was
used to suppress the water resonance and retain
the exchangeable proton signals [10]. The inter-
proton distances were calculated from the ratios of
the cross-relaxation NOEs [9] and from the Y4 -
Cβ H2 interproton distance as a reference. A set of Fig. 2. Variation of the CD ellipticity at 220 nm as a function of
39 backbone–backbone and 10 backbone–side chain SDS concentration in water.
constraints with interproton distances equal to the
RESULTS AND DISCUSSION
NOE distances ±0.5 Å were used for restrained MD
calculations.
The dependence of the CD spectrum for the
casein fragment on SDS concentration (Fig. 1) is
Molecular modeling
typical of a transition from a random conformation
On the basis of the CD analysis and NOESY re-
in pure water (negative band at 198 nm) [11] to an
sults (vide infra), the decapeptide casein fragment
was built in the 310 and α-helix form, and submitted ordered form presenting two negative bands at 208
to EM and MD calculations using the DISCOVER and 220 nm. The ellipticity at 220 nm exhibits a
(Biosym, San Diego, CA, U.S.A.) program, and rapid variation with SDS concentrations below the
applying force constants of 12 and 4 kcal mol−1 critical micellar concentration (about 8 × 10−3 M),
Å−2 to the backbone–backbone and backbone–side and reaches a plateau above the SDS critical micel-
chain constraints, respectively. Both calculations lar concentration (Fig. 2). Under these conditions,
converged toward two quite similar conformers. In the relative intensities of the two minima at 208 and
vacuo MD simulations of 0.1 ps each, using a six- 220 nm are in favor of a 310 -helix rather than of
step temperature elevation at 50, 100, 150, 200, an α-helix [12], although a certain percentage of the
250 and 300 K, were run and followed by a 25 ps latter cannot be excluded. Similarly, the appearance
qrestrained MD run under temperature control. Af- of NOE connectivities on addition of SDS in water
ter thermal equilibration, a 320 ps restrained MD (Fig. 3) reflects the peptide–micelle interaction and
with weak coupling (0.1 ps) to the thermal bath was the restricted flexibility of the peptide in this micel-
performed. The last 30 ps period of the MD simula- lar medium. Due to proton resonance broadening in
tion, which exhibited no substantial conformational the presence of SDS, the vicinal coupling constants
changes, was used for conformational averaging and could not be determined.
analysis. A 3r dielectric constant for Coulombic in- The temperature coefficients (Table 1) reveal that
teractions and a double cutoff at 12.5 and 14 Å for the C-terminal NHs are less solvent-exposed than
all nonbonded interactions were applied. the N-terminal ones, and support the hypothesis of
a helicoid structure. The R10 -Ne H proton and one
R10 -Nη H guanidinium proton are also solvent pro-
tected and most probably involved in a salt bridge
with E6 -Cδ O2 − and/or R10 -CO2 − [13]. The other
three R10 -guanidinium protons are in rapid exchange
362
TABLE 1
DIMENSIONS (Å, ◦ ) AND OCCURRENCE (%) OF THE HYDROGEN BONDS IN THE DECAPEPTIDE CASEIN FRAGMENT
DURING THE LAST 30 PS OF RESTRAINED MD SIMULATIONS

Donor Temperature Acceptor Hydrogen N-H· · ·O Occurrence

coefficient bond Distance Angle (%)
(10−3 ppm/K) type (Å) (◦ )

L2 -NH –5.1
G3 -NH –6.4
Y4 -NH –4.4
L5 -NH –5.8
E6 -NH –1.4 L2 -CO i+4 → i 2.78 132 68
Q7 -NH –4.0 Y4 -CO i+3 → i 2.80 137 90
L8 -NH –2.1 L5 -CO i+3 → i 3.18 125 67
L9 -NH –2.9 E6 -CO i+3 → i 2.85 134 74
R10 -NH –3.3 E6 -CO i+4 → i 2.63 170 90
Q7 -NH2 –4.1
Q7 -NH2 –4.2
R10 -Ne H –1.8 R10 -CO2 − 2.82 144 98
R10 -Nη H –1.6 E6 -Cδ O2 − 2.95 114 75

The time-averaged structure obtained from the

150 conformers in the last 30 ps MD calculations and
subsequent EM is depicted in Fig. 4. Only two of the
backbone–backbone and four of the backbone-side
chain interproton distances are out of the margins
by less than 0.45 Å, and the largest violations are
observed for the latter. Such differences are com-
monly observed for oligopeptides which are known
to adopt rather flexible structures, and especially in
the orientation of their side chains [14]. The short
N· · ·O distances (Table 1) show three consecutive
i+3 → i interactions, typical of a 310 -helix encom-
Fig. 3. 1 H/1 H NOE connectivities for the casein fragment in SDS passing the Y4 -L-E-Q-L-L9 sequence, and two i+4
aqueous solution. Strong, medium and weak intensity of the NOE → i interactions, typical of an α-turn, involving the
cross peaks is indicated by the thickness of the lines.
N-terminal L2 -G-Y-L-E6 and C-terminal E6 -Q-L-L-
R10 sequences.
with water and not visible at room temperature. The The C-terminus of the molecule exhibits the R10 -
successive medium αN(i,i+2) cross peaks argue for N H to R10 -CO2 − and R10 -Nη H to E6 -Cδ O2 − ionic
e

the occurrence of a 310 -helix spanning the Y4 -L- interactions, in agreement with the small tempera-
E-Q-L-L9 sequence, in agreement with the small ture coefficients pointed out for these guanidinium
resonances. An ionic interaction between arginine
temperature coefficients for the last five C-terminal
guanidinium and glutamic carboxylate is actually
residues (Table 1). The weak αN(i,i+4) E6 -Cα H/R10 - known to be an important helix-promoting factor
NH NOE cross peak suggests that the R10 C-terminal [17]. One notes the amphiphilic character of the re-
residue could participate in an α-turn. sulting structure (Fig. 4), placing the Y1 , L2 , Y4 ,
 363

Fig. 4. Stereoview of the time-averaged structure from the last 30 ps of restrained MD simulation for the decapeptide casein fragment.

L5 , L8 and L9 hydrophobic side chains on one side NMR and restrained MD simulations. The time-
of the molecule, and the three E6 , Q7 and R10 hy- averaged structure of the decapeptide is an am-
drophilic side chains on the opposite side. Such an phiphilic 310 -helix initiated and terminated by an
amphiphilic structure allows its interaction with the α-turn. The ionic interactions between the R10 -
apolar core of the micelles [15,16], which is a rough guanidinium and the R10 - and E6 -carboxylates illus-
approximation of the cellular membrane in which the trate the key role of the C-terminal arginine residue in
benzodiazepine receptor is anchored. the stabilization of the helix structure. The tyrosine
Although the affinity of the natural casein frag- phenol rings probably play a similar role to that of
ment to the GABAA receptor is much lower than the aromatic rings in the benzodiazepines.
that of the benzodiazepines [8], their competi-
tive recognition by the same receptor suggests ACKNOWLEDGEMENTS
that they should share common structural prop-
erties. Active benzodiazepines contain two aro- A doctoral fellowship from the CNRS and the
matic rings, but only the benzo ring seems to Région Lorraine is warmly acknowledged by C.F.
be required for a good recognition by the re- L.M. and J.L.G. thank the Prospérité Fermière (Ar-
ceptor, as for the strong antagonist 8-fluoro- ras, France) for financial support.
5,6-dihydro-5-methyl-6-oxo-4 H-imidazo[1,5a][1,4]
benzodiazepine-3-carboxylic acid ethyl ester (Flu- REFERENCES
mazenil) [18]. This observation suggests that one
1 Brantl, V. and Teschemacher, H., Naunyn-Schmiedeberg’s
of the phenol rings could play the same role as the Arch. Pharmacol., 306 (1979) 301.
benzo ring in benzodiazepines. Experiments aiming 2 Yoshikawa, M., Tani, F., Ashikaga, T., Yoshimura, T. and
at the determination of the essential residues will be Chiba, H., Agric. Biol. Chem., 50 (1986) 2951.
3 Maruyama, S. and Suzuki, H., Agric. Biol. Chem., 46 (1982)
reported elsewhere. 1393.
4 Parker, F., Migliore-Samour, D., Floc’h, F., Zerial, A.,
Werner, G.H., Jollès, J., Casaretto, M., Zahn, H. and Jollès,
CONCLUSIONS P., Eur. J. Biochem., 145 (1984) 677.
5 Jollès, P., Levy-Toledano, S., Fiat, A.-M., Soria, C.,
The conformation of the decapeptide casein frag- Gillessen, D., Thomaidis, A., Dunn, F.W. and Caen, J.P., Eur.
J. Biochem., 158 (1986) 379.
ment (YLGYLEQLLR) has been determined in 6 Yamamoto, N., Akino, A. and Takano, T., J. Dairy Sci., 77
the SDS–water micellar medium by combining 2D (1994) 917.
364

7 Pellow, S., Chopin, P., File, S.E. and Briley, M., J. Neurosci. 13 Tsikaris, V., Tzovaras, D., Sakarellos-Daitsiokis, M.,
Methods, 14 (1985) 149. Sakarellos, C., Orlewski, P., Cung, M.T. and Marraud, M.,
8 Miclo, L., Perrin, E., Driou, A., Iung, C., Boudier, J.-F. and Biopolymers, 38 (1996) 673.
Linden, G., European Patent No. 95-02-697.7-2105, 1995. 14 Geyer, A., Müller, G. and Kessler, H., J. Am. Chem. Soc.,
9 Wüthrich, K. (Ed.) NMR in Biological Research: Peptides 116 (1994) 7735.
and Proteins, North-Holland, Amsterdam, The Netherlands, 15 Jung, H., Windhaber, R., Palm, D. and Schnackerz, K.D.,
1976. Biochemistry, 35 (1996) 6399.
10 Piotto, M., Saudek, V. and Sklenár, V., J. Biomol. NMR, 2 16 Bairaktari, E., Mierke, D.F., Mammi, S. and Peggion, E.,
(1992) 661. Biochemistry, 29 (1990) 10090.
11 Fasman, G.D. (Ed.) Circular Dichroism and the Conforma- 17 Huyghues-Despointes, B.M.P., Scholtz, J.M. and Baldwin,
tional Analysis of Biomolecules, Plenum Press, New York, R.L., Protein Sci., 2 (1993) 80.
NY, U.S.A., 1996. 18 File, S.E., Dingemanse, J., Friedman, H.L. and Greenblatt,
12 Manning, M.C. and Woody, R.W., Biopolymers, 31 (1991) D.J., Psychopharmacology, 89 (1986) 113.
569.