conformational analysis of a triproline-derived helical template and its peptide conjugates
TRANSCRIPT
Pergamon
0040-4039(95)00673-7
letrahedron Letters, Vol. 36, No. 22, pp. 3813-3816, 1995 Elsevier Science Ltd
Printed in Great Britain 0040-4039/95 $9.50+0.00
Conformational Analysis of a Triproline-Derived Helical Template and its Peptide Conjugates
D.S. Kemp* and Jeffrey H. Rothman
Room 18-582, Deparlment of Chemistry, Massachusetts Institute of Technology, Cambridge, MA 02139
Abstract : The stable conformations in solution of a tfiproline-derived macrocyle and its peptide conjugates in
solution are assigned as cct (nonnucleating) and ctt (310 helical) by an 1H NMR analysis.
The accompanying paper reports the synthesis of a macrocyclic triproline derivative, prepared as a
potential N-terminal nucleating site for et-helices in linked polypeptides. An X-ray analysis establishes that in
the crystal, the acid assumes a non-nucleating cct conformation. Here we present a conformational analysis
based on 1H NMR spectroscopic data in a variety of solvents that demonstrates the amide orientations of the
template in solution to be the cot state 1 (S-up rll V +125°) and the ctt state 2 (S-down rll ~t -10°).
Is',t~O 1 o
1 e e t u p r l l
o .
2 ett down l 1 l
The 500 MHz 1H NMR spectrum of the methyl ester of the macrocycle reveals two sets of resonances
(ratios of 9:1 in CDCI3, 4:1 in CD3CN, and 6:4 in D20), corresponding to states that equilibrate slowly on the
NMR time scale and that differ in c-t orientation at the amide bonds, t The major state shows a large chemical
shift difference between the CH2-S (AS = 3.69-2.95 = 0.74 in CDC13) and two strikingly deshielded
resonances: an ct-methine at 5.65 5 in CDC13 (normal, 4.5-4.6 5) and a ~t -methine of a 4-thiaproline at 4.45 5
in CDC13; (normal 3-3.5 5). These imply an unusual magnetic environment for the methine hydrogens and
suggest conformational homogeneity. Unique assignments of resonances were made by COSY analysis of
three deuterated derivatives in which ring 1, or ring 2, or both rings 1 and 2 were perdeuterated. As noted in
Figure 1, the deshielded resonance at 5.53 ppm and that at 4.48 ppm can easily be identified as the ~-Hs of
proline 1 and 2, respectively. Resonances associated with the minor conformation are also clearly evident.
3813
3814
Pro l ine -2 p e r d o u t e r a t e d A t ~
Prol ine-1 p e r d e u t e r a t e d
Pro l ines-1 & 2 p e r d e u t e r a t e d
, i J I I i i , i I i
5 . 5 5 . 0
I I
A I I i I I I
4 . 5 p p m
l . .
I '
Figure 1 Portion of the 500 MHz 1H NMR Spectrum of the Macrocycle Ester in CD3CN
The spectroscopic results are best interpreted after more detailed conformational analysis using
molecular modeling. 2 Structure 1 depicts the calculated cct conformation of lowest energy and corresponds
closely with the X-ray structure determined for the crystalline acid. The cct states can be distinguished from
all other conformations by their placement of the ix-proton of ring 1 within 2.5-2.6/~ of the sulfur, the 8-proton
of ring 3, and the carbonyl of the amide side chain of ring 3. Nonbonded distances for the cct conformation
pertinent to the subsequent NOE analysis are listed in Table 1.
Table 1 Energies and Characterist ic Non-Bonded Distances for the States of the Macrocyc le
Nonbonded Distance (]k) State Calc. Energy
(kcal/mol) 1-~ & S-CH 2 1-ct & 3-8
cct S-up rll 1 2.84 2.7 2.6 3.1 2.0
cct S-up lrl 3.26 3.9 2.6 3.1 2.0
ctt S-dwn 1112 - 1.36 2.5 4.0 3.8 2.3
ctt S-up lrl 2.40 2.4 4.0 2.4 3.7
ttt S-up rll 1.17 4.7 4.4 2.9 2.3
ttt S-dwn lrr 3.25 4.5 4.4 3.8 2.3
3-7 & S-CH 2 3-8 & S-CH 2
Stereodiagram 2 depicts the ctt conformation calculated as most stable (S-down). The most stable
substate of the alternative ctt S-up orientation is calculated to be 3.8 kcal/mole less stable. The most stable ttt
state lies within the energy range of the calculated stabilities of the cct and ctt states. Table 1 shows relative
stabilities and characteristic nonbonded distances for the most stable states of each major species.
3815
Analysis of both 1 and 2D NOE experiments in CD3CN demonstrated the following interactions
between pairs of resonances: S-CH 2 with 3- ~-CH, and 1-tx-CH with both S-CH 2 and 3 8--CH. This pattern
is uniquely consistent with the cct S-up rll structure also observed in the crystal. Conformational assignment
for the state corresponding to the minor resonances was facilitated by the observation of chemical exchange
cross peaks due to saturation transfer 3 in 1D NOE and 2D ROESY 4 spectra of the methyl ester and its peptide
derivatives. These tie together pairs of resonances corresponding to major and minor conformations and
permit unique assignments of resonances to protons for the minor conformation. The pattern of chemical shifts
for this species lacks most of the anomalies seen for the more abundant cct conformation, and must therefore
belong to an alternative conformation with different nonbonded interactions. For example, in CDCI 3 the three
o~-methine resonances all appear in the range of 4.7-4.8 8, and the S-CH y-methine of ring 3 appears at 3.44 8.
A large chemical shift difference of 0.79 8 = 3.90 - 3.11 is seen for the diastereotopic S-CH 2 resonances.
Spectra of peptide derivatives were also examined for which the acid or ester function of the
macrocycle is replaced by L-alanine peptide bonds CO-(NH-CHMe-CO)n -OH, n = 1,2,3. These also show
two sets of resonances corresponding to a pair of detectable, slowly equilibrating conformational manifolds,
with spectroscopic properties for the macrocycle nearly identical with those observed for the acid or ester. The
important exception is the relative state abundances, which now disfavor the cct species. NOE studies of the
peptide conjugates in CD3CN and a ROESY spectrum in D20 of the template resonances of the now major
conformation demonstrate significant interactions between the S-CH 2 and both the 3- 8-CH and 1-¢t-CH
resonances. As seen from the analysis of Table l, this pattern is inconsistent with the ttt states and only
consistent with cct or ctt states, 1 or 2. Since the former is excluded by prior assignment, the template in its
second conformation must corresponds to the ctt state, which can only nucleate a 310-helix. With the particular
potential functions and parameter sets used, molecular mechanics thus includes the pair of states detected by
NMR in a short list of low energy states, but does not predict the observed order of stabilities.
Helices nucleated by the ctt conformation of 2 must assume a 310 helical geometry at the template -
peptide interface. Calculations by Marshall and coworkers of the relative stabilities of 310 and co-helices for
short peptides imply that these conformations are of comparable stability, 5 and interconversion between them
is predicted to be very rapid. The N-termini of tx-helices in proteins often show 310 character, 6 and Karle has
found that the same peptide can form distinct crystals that exhibit either 310 or co-conformations. 7
Alanine conjugates of template 2 were examined by 1H NMR in several solvents. Spectra of alanine
conjugates longer than triAla were complicated by peak overlap, broadening, and evidence of aggregation.
They were not examined in detail. Previous studies of helical peptides conjugated with a reporting helical
template show a progressive stabilization of the template nucleating state over the non-nucleating as the
peptide chain length is increased or the solvent polarity is decreased. 8 In CD3CN the [cct]:[ctt] ratio increases
from 4:1, 1:2, 1:3, 1:5 for the series consisting of the methyl ester, and the mono, di, and trialanine conjugates
of 2. This change is consistent with a small but progressive stabilization of the ctt state with increase of
peptide length. The corresponding numbers in water are: 64:36, 61:39, 59:41, 60:40, implying no incremental
stabilization by the alanine residues of the ctt state relative to the cct state. With this series a progressive
shielding of the cct-state H-1 resonance is seen with increase in n. Moreover NOEs are observed between the
Ala-1 NH and the et-CH resonance of Pro-1. Both results imply that the cct state does in fact undergo a
length-dependent peptide-template interaction. If both cct and ctt states are stabilized, the absence of a length-
dependent state ratio need not imply the absence of length-dependent ctt-state stabilization.
A helical peptide is expected 8 to exhibit characteristic chemical shift values and temperature
3816
dependences, as well as NOE interactions diagnostic of local structure. All these are seen for the ctt states. For
the di and triAla conjugates of 2 in H20-D20, the average NH chemical shift is 8.11 ppm (SD 0.08) with a
temperature dependence of - 7.2 (0.6) x 10 -3 ppm/K for the cct state, and 7.72 (0.31) with a temperature
dependence of- 4.3 (2.8) x 10 -3 ppm/K for the ctt state. For the helical structure-breaking solvent DMSO-d 6,
the averages are 8.00 (0.1) with a T dependence of - 3.8 (0.7) x 10 -3 ppm/K for the cct state and 7.42 (0.29)
with a T dependence of - 2.3 (2.7) x 10 -3 ppm/K for the ctt state. (The large standard deviations seen for the
cct state reflect a substantial increase of chemical shift and T dependence with alanine rank.) ROESY-
determined NOE interactions of the cct states of the Ala conjugates in DMSO-d 6 or in HDO include no
intrapeptide interactions characteristic of structure. Two peptide-macrocyle intereactions are seen between the
NH of Ala-1 and the ct-CH of ring 1 as well as a 7-CH of ring 1 or 2. By contrast, the ctt states reveal helical
NN(i,i+l) interactions between successive alanine NH resonances. (A macrocycle-peptide cz[3(i, i+2)
interaction may be present but peak overlaps prevent explicit assignment.) A pepdde-macrocycle interaction
between the NH of Ala-1 and the SCH 2 confirms the S-down ctt state assignment and is expected for a 310
helical hydrogen bonding pattern for the first NH residue of the peptide. The consistently different NMR
structural parameters seen for these short peptides linked to cct and to ctt states of 2 imply that helix nucleation
occurs only with the latter state.
The aggregation observed with short peptide conjugates and the failure to observe large changes in
[ctt]/[cct] ratio with changes in peptide length or solvent render 2 much less useful for quantitation of helicity
than our more localized tricyclic helix-nucleating template. 8 However, the properties of 2 do provide a
valuable reference for comparison with the strikingly different helix initiation behavior that we have observed
with a close structural analog of 2 that is conformationally more constrained. These results and the
comparisons with the properties of 2 will be reported subsequently.
Acknowledgements Financial support from Pfizer Research and from the National Science Foundation,
Grant 9121702-CHE is gratefully acknowledged.
References
1. Grathwohl, C.; Wuethrich, K., Biopolymers, 1981, 20, 2623-2633.
2. Molecular modeling was carried out using the CHARMm QUANTA© 3.3 software of Molecular
Simulations Inc.. Structures were generated systematically as described in the accompanying manuscript
and subjected to energy minimization; random conformational search and dynamics algorithms were used
to check inclusiveness. NOESY and ROESY measurements were carried out as described previously. 8
3. Jeener, J.; Meier, B.H.; Bachmann, P.; Ernst, R.R., J.Chem. Phys., 1979, 71, 4546; Kalman, J.R.;
Williams, D.H., J.Am.Chem.Soc., 1980, 102,906.
4. Bothner-By, A.A.; Stephens, R.L.; Lee, J-M.; Warren, C.D.; Jeanloz, R.W., J. Am.CherrLSoc., 1984, 106, 811.
5. Hodgkin, E.E.; Clark, J.D.; Miller, K.R.; Marshall, G.R., Biopolymers, 1990, 30, 533-546.
6. Nemethy, G.; Phillips,D.C.; Leach, S.J.; Scheraga, H..A., Nature, 1967, 214, 363-365.
7. Karle, I.L.; Flippen-Anderson, J.L.; Sukumar, M.; Balaram, P., Int. J. Pept. Prot. Res., 1988, 31,567- 576.
8. Kemp, D.S.; Curran, T.P.; Boyd, LG.; Allen, T., J. Org. Chem. 1991, 56, 6683-6697; Kemp, D.S.; Boyd,
J.G.; Muendel, C.C., Nature, 1991, 352, 451-454.
(Received in USA 2 August 1994; revised 3 April 1995; accepted 6 April 1995)