Analytica Chimica Acta, 210 (1988) 151-162 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
151
ELECTROSTATIC FEATURES OF MOLECULAR RECOGNITION BY CYCLIC UREA MIMICS OF CHYMOTRYPSIN
CAROL A. VENANZI* and KRISHNAN NAMBOODIRI
Department of Chemical Engineering and Chemistry New Jersey Institute of Technology, Newark, NJ 07102 (U.S.A.)
(Received 17th August 1987)
SUMMARY
The molecular electrostatic-potential pattern was used to investigate the electrostatic features of molecular recognition by two cyclic urea mimics of the active site of cr-chymotrypsin. The structures of the mimics were obtained by molecular-mechanics evaluation of the conformational potential-energy surface of the molecules. Calculations were done by using two different atomic point-charge sets in order to assess the effect of charge on the electrostatic potential pattern. The molecules studied were: (1) a “full” mimic of chymotrypsin containing the hydroxyl, imidazole, and carboxylate anion functionalities typical of the active site of the enzyme, and (2) a “partial” mimic with only the hydroxyl and imidazole functional groups. Comparison of the molecular elec- trostatic-potential patterns of the two mimics in both charge sets showed that the largest differ- ences were due to the structural addition of the carboxylata anion, rather than any particular differences in the choice of atomic point charge. For the full mimic, the pattern was essentially dominated by the negative charge on the carboxylate. Small structural changes which optimized the orientation of the catalytic components had little effect on the electrostatic potential pattern of the molecule. In the absence of the anionic functionality, greater differences were noted in the electrostatic potential pattern of the partial mimic in the two charge sets. The choice of atomic point charge was seen to influence the hydrogen-bonding pattern of the hydroxyl and imidazole moieties, resulting in differences in the spatial orientation of the electrostatic potential minima. In general, both charge sets produced molecular electrostatic-potential patterns which indicated that long-range electrostatic interactions would direct the cationic end of the substrate into the electron-rich binding site. However, specific local features of the electrostatic potential pattern were found to depend on point-charge set through the influence of charge on the hydrogen-bonding pattern.
Understanding of the process of molecular recognition is essential to the design of potent biomimetic catalysts. Rebek [ 1 ] has pointed out that molec- ular recognition by interacting partners occurs through the alignment of com- plementary shapes as well as complementary functionalities. In an attempt to elucidate these interactions at the molecular level, the techniques of compu- tational chemistry and computer graphics have been applied to the study of a
0003-2670/88/$03.50 0 1988 Elsevier Science Publishers B.V.
152
series of artificial enzymes, cyclic urea compounds, designed by Cram and co- workers [ 2-41 to mimic the action of the hydrolytic enzyme, a-chymotrypsin. The series consists of a macrocyclic synthetic precursor which selectively binds cationic ligands [ 21, two “partial” mimics built from the macrocycle with the incremental addition of hydroxyl [ 31 and imidazole [ 41 functionalities, and a “full” mimic, containing the functional groups of the catalytic triad, which has been proposed [ 31, but not yet synthesized by Cram’s group at the time of this writing. The partial mimics have been shown [3, 41 to approach enzymatic velocities in acyl-transfer reactions.
Molecular-mechanics conformational analysis has been used to investigate the structural components of molecular recognition in this series of analogues [ 5-81. The degree of rigidity of the macrocyclic binding site during complex- ation was studied and the site was shown to be essentially “pre-organized” for binding [ 51. It was also demonstrated that the addition of the hydroxyl group to the binding site causes little change in the conformational properties of the macrocycle [ 5,6]. In the analysis of the full mimic, the ability of the catalytic triad to adopt an orientation similar to that of chymotrypsin was assessed [ 71. Two alternative molecular designs which appeared to give a more optimal ori- entation of the catalytic triad were suggested [ 71.
In addition, the electrostatic components of molecular recognition were in- vestigated by analysis of the electrostatic potential patterns of the artificial enzymes. It was shown [5] that the precursor presents a “bull’s eye” pattern of increasing negative value, which can direct the cationic end of the ligand into the electron-rich binding site. It was demonstrated that the partial mimic (with only a hydroxyl functionality, called Mimetic 1 in previous work [ 5-81) and its substrate present complementary electrostatic potential patterns [ 61. It was also shown that the full mimic displays a pattern qualitatively similar to that of the catalytic triad of chymotrypsin [ 71. This work was summarized recently [ES].
The present work is focused on the partial mimic shown schematically in Fig. 1 (called Mimetic 5 in previous work [8] ) and the full mimic (called Mi- metic 2 [7, 81) shown in Fig. 2. Cram et al. [4] have demonstrated that in contrast to Mimetic 1, acylation of Mimetic 5 occurs at the imidazole group. The full mimic, Mimetic 2, has not been synthesized (see above), but contains all the elements of the catalytic triad. The details of an extensive conforma- tional analysis of both molecules will be presented in a forthcoming paper, in which the effect of choice of atomic point charge on molecular conformation will be assessed and the relationship of structure to function will be examined by analyzing the orientation effect of the carboxylate anion on the imidazole group. In the present work, the global energy minimum structures determined in the above conformational analysis of Mimetics 2 and 5 are used to focus on the effect of atomic point charge on the electrostatic features of molecular recognition, as displayed through the molecular electrostatic-potential pattern.
153
Fig. 1. Schematic diagram of Mimetic 5. The torsional angles (Y (C4&zs-C,26-C,) and S (CT6- C66-C5,&56) define the orientation of the hydroxyl and imidazole groups with respect to the macrocycle. Distance A (HI16-Nil) indicates the distance between the hydroxyl and imidazole groups.
Fig. 2. Schematic diagram of Mimetic 2. The torsional angles (Y and Gare as defined in Fig. 1. The torsional angle y (C,,-C&-C&-C,,) is incremented to optimize the distance A (HI,,-N,,) between the hydroxyl and imidazole groups. Distance B (H9-0& indicates the distance between the im- idazole and carboxylate groups.
METHODS
Conformational analysis The conformational analysis was done with version 2.0 of the AMBER (As-
sisted Model Building with Energy Refinement) molecular mechanics pro- gram [9]. The AMBER united atom force field was supplemented with parameters appropriate to the mimics [ 71. In the united atom approximation, only those hydrogens capable of hydrogen bonding (H9 and H116 in Fig. 2) are included in the calculation. The model building procedures used to con- struct the mimics from the constituent cyclic urea, methylanisole, 3-hydroxy- methylbiphenyl, 4-phenylimidazole, and 3-methoxy-6-methylbenzoate anion have been described [ ~$71. In addition to the STO-3G-level Mulliken atomic point charges used previously [ 71, a set of “potential-derived” [lo] atomic point charges was constructed for each of the molecular fragments. The united atom potential-derived charges were obtained by requiring that the charges reproduce the electrostatic potential calculated on the surface of each molec- ular fragment. Calculations were done in the STO-3G basic set by using the GAUSSIAN 80 UCSF program package [ 111.
154
The same set of initial conformations was used for the energy minimization of Mimetics 2 and 5. The conformations differ chiefly in the orientation of the hydroxyl and imidazole functionalities with respect to the macrocycle. The position of the hydroxyl group is determined by the dihedral angle 6 (shown as C76-Cs6-C56-C156 in Figs. 1 and 2). The position of the imidazole function- ality is determined by the sum of a and 6, where (x is the dihedral angle Cd9- Czs-Cl&& (Figs. 1 and 2). Thirty-six conformations were obtained by in- crementing a! in units of 60’ and then incrementing S in units of 60’ for each value of Ly.
Each of the 36 conformations was minimized in both the charge sets. The global energy-minimum structure in each charge set was chosen for calculation of the molecular electrostatic-potential pattern. In addition, in the potential- derived charge set, the dihedral angle y (C12-C5-Cs-Nil, Fig. 2) of the global energy-minimum conformation was further varied from 0 to 180” in units of 30” in order to optimize the hydrogen-bond interaction between H,,, and N,, of the catalytic triad (shown as distance A in Fig. 2). The structure with the optimal distance A was chosen for analysis of the electrostatic potential pattern.
Molecular electrostatic-potential maps By using the global energy conformation of Mimetic 2 in the potential-de-
rived charge set as the template, a standard orientation was chosen with the cyclic urea oxygen atoms situated in the x-y plane at z = 0 A. The four other structures were fitted to this orientation with the RMS (root mean square) fitting facility of AMBER. The molecular electrostatic potential was approx- imated as the coulombic interaction between a positive point charge and the static charge distribution of the molecule, modeled by the atomic point charges positioned at the nuclei. The potential patterns were calculated in the x-y plane at z = 2 A using the “Set Map” facility of Chem-X (Chemical Design, Oxford, England). This plane was chosen because it indicates a typical position for a bound substrate. The plane contains the z-coordinate of the nitrogen atom of the substrate in the x-ray structure of a model system, the cyclic urea macro- cycle complexed with a tertiary butylammonium ligand [2]; the x-ray struc- ture of this complex was kindly provided by K.N. Trueblood.
RESULTS
The molecular electrostatic potential pattern of Mimetic 2 is given in Fig. 3, while that of Mimetic 5 is found in Fig. 4. In each case, the molecule is displayed in the united atom approximation with only the imidazole (H,) and hydroxyl (HI16) hydrogens explicitly shown. The contour levels in units of kcal mol-’ are: - 100, -75, -50, - 25,0,25,50,75. The distances between the catalytic groups are compared to the active site of chymotrypsin [ 12 J in Table 1.
Figure 3a shows the global energy minimum structure obtained for Mimetic
155
Fig. 3. The molecular electrostatic-potential patterns of Mimetic 2 in the Mulliken charge set (A) and in the potential-derived charge sets (B, C). (A) The conformation shown is that of the global energy-minimum structure at 47.66 kcal mol-‘. (B) The conformation shown is that of the global energy-minimum structure at - 17.64 kcal mol-‘. (C) The conformation shown is that of the local energy minimum with optimum distance A (see Fig. 2) and energy - 13.24 kcal mol-‘. @ all cases, the molecule is orientated with the cyclic uzea oxygen atoms in the x-y plane at z=O A. The electrostatic potential is calculated in the z = 2 A plane. The contour levels are: - 100, - 75, - 50, - 25,0,25,50,75.
2 in the Mulliken charge set. This structure, which corresponds to a minimized energy of 47.86 kcal mol-‘, is characterized by a hydrogen bond between the imidazole hydrogen and the carboxylate oxygen ( H9-Od8 = 2.0 A, see Table 1) . There is no hydrogen bond between the hydroxyl hydrogen (Hi& and either of the neighboring cyclic urea oxygens ( Oa or OY1 ) or the methyl anisole ox- ygena. The closer of the two 0 (cyclic urea)-H (hydroxyl) distances is 3.4 A. In addition, the hydroxyl hydrogen is far from the imidazole nitrogen (Hi,,-
156
Fig. 4. The molecular electrostatic potential of Mimetic 5 in the Mulliken charge set (A) and in the potential-derived charge set (B). (A) The conformation shown is that of the global energy- minimum structure at 26.36 kcal mol-‘. (B) The conformation shown is that of the global energy- minimum structure at - 10.89 kcal mol-‘. The molecular orientation, map plane, and contour levels are the same as in Fig. 3.
N 11 = 8.4 A). The molecular electrostatic potential of Mimetic 2 in the Mulli- ken charge set (Fig. 3a) displays an overall negative pattern because of the presence of the carboxylate anion. The chief features are the negative region at - 100 kcal mol-’ around the carboxylate, and broader regions of - 75 and - 50 kcal mol-’ that extend beyond the perimeter of the macrocycle. The pat- tern is not a symmetrical “bull’s eye” pattern, but rather is shifted towards the carboxylate group.
Figure 3b shows the electrostatic-potential pattern of Mimetic 2 in the po- tential-derived charge set for the global energy-minimum conformation at -17.64 kcal mol-‘. This structure is characterized by a hydrogen bond be- tween the_hydroxyl hydrogen and the neighboring cyclic urea oxygen (Hl16- 0 71 = 1.9 A, see Table 1). In addition, there is a hydrogen bond between the imidazole and carboxylate groups (H9-04s= 1.8 A). The hydroxyl hydrogen and the imidazole nitrogen are not hydrogen-bonded (H116-Nll = 5.2 A). The electrostatic-potential pattern of Mimetic 2 in the potential-derived charge set (Fig. 3b) is very similar to that in the Mulliken charge set (Fig. 3a). Both figures show broad regions at - 75 and - 50 kcal mol-’ with the same general spatial orientation. In addition, the overall pattern is negative, with the most negative contour ( - 100 kcal mol-l ) appearing around the carboxylate anion. In general, the qualitative features of the molecular electrostatic potential of Mimetic 2 appear to be the same for the two charge sets.
However, Fig. 3b shows additional regions of - 100 kcal mol-’ around the benzyl alcohol and phenylimidazole fragments. These contours are due to the
157
TABLE 1
Distances” of functional groups in catalytic triad
Structure Hydroxyl-cyclic urea Hydroxyl-imidazole Imidasole-carboxylate
Hi,&-& Hi,,-% HwN,, H&&s
Mimetic 2 3.4 4.7 3.4 2.0 Mb, 47 86”
Mimetic 2 pi’, - 17.64
5.7 1.9 5.2 1.8
Mimetic 2 P, - 13.24
5.9 3.6 2.0 1.8
Hydroxyl-cyclic urea Imidazole-cyclic urea
Hi,,-0, HUt+&* Hs-% Hs-071
Mimetic 5 5.4 6.1 4.6 3.4 M, 26.36
Mimetic 5 P, - 10.89
2.0 5.0 1.9 4.9
0 (Hydroxyl ) -N ( imidazole ) N (Imidazole) -0 (carboxylate)
Cbymotrypsind 4.0 4.2
“Distances given in HngstrSm units (A). bAtomic point-charge sets are indicated by M (Mulliken) or P (potential-derived). %nergy in kcal mol- . 1 dThese distances are from the 1.67-A resolution structure of the chymotrypsin dimer by R.A. Blevins and A. Tulinsky (Brookhaven Protein Data Bank (cf. [ 121) ).
larger negative values of the potential-derived charge set, as well as to the different spatial orientation of the phenylimidazole group in the two figures. The charges on the atoms of the benzyl alcohol and phenylimidazole fragments are given in Table 2. The atoms are labeled according to the schematic diagram in Fig. 2. The value of the z-coordinate of each of the atoms is given in the table. The electrostatic potential contour maps were calculated in the z = 2 A plane. For both charge sets, Table 2 shows that this plane cuts through the benzyl alcohol fragment, passing between atoms CY6 and Cs6 on one side of the phenyl ring, and atoms C 146 and C,,, on the other, while passing close to HII+ However, the charge of atom C&, in particular, is much more negative in the potential-derived set ( -0.570) than in the Mulliken set ( -0.009). This ac- counts for the additional contour at - 100 kcal mol-’ in the region of the ben- zyl alcohol fragment in the potential-derived charge set (Fig. 3b ).
Comparison of parts a and b of Fig. 3 shows that the phenylimidazole frag- ment has a slightly different spatial orientation in the two charge sets. Table 2 shows that, in the potential-derived charge set calculation, the z= 2 A plane
159
between the hydroxyl and imidazole groups (distance A, Fig. 2) obtained by rotation of the phenylimidazole torsional angle. Table 1 shows that this struc- ture is characterized by hydrogen bonds between the hydroxyl and imidazole groups ( Hw-N~ 1 - -2.0 A), as well as between the imidazole and carboxylate groups (H.&L = 1.8 A). Comparison of parts b and c in Fig. 3 shows that the patterns are very similar. The chief difference is the appearance of a region a,t - 100 kcal mol-’ extending from OT1 to He Table 2 indicates that the z=2 A plane cuts through the imidazole group, passing near several atoms with neg- ative partial charges (NB at z= 1.4 A, -0.569; Nll at z=2.9 A, -0.512). Be- cause the structure in Fig. 3c shows that N,, has moved closer towards the hydroxyl group, this accounts for the extension of the - 100 kcal mol-’ con- tour into this region. In general, patterns b and c (Fig. 3) are qualitatively similar, indicating that, within a particular charge set, small structural varia- tions have little effect on the overall molecular electrostatic-potential pattern.
The molecular electrostatic-potential pattern of Mimetic 5 in the Mulliken charge set is shown in Fig. 4A. The structure shown is that of the global energy minimum conformation at an energy of 26.36 kcal mol-‘. The structure shows no hydrogen-bonding pattern; neither the hydroxyl nor the imidazole groups are hydrogen-bonded to the cyclic urea oxygens (see Table 1) . In comparison to the molecular electrostatic potential pattern of Mimetic 2 in the same charge set (Fig. 3A), that of Mimetic 5 is less negative because of the absence of the carboxylate anion. The most negative contours (at - 25 and - 50 kcal mol-’ ) are associated with the positions of the hydroxyl and imidazole groups. Table 2 shows that the z= 2 A plane cuts through the phenyl and imidazole groups, passingnearC$ (z=l.gA) and& (z=2.6A) ofthephenylringandc, (2=2.0 A) and C, (z= 2.0 A) of the imidazole. The appearance of a contour at -50 kcal mol-’ near Nll is due to the fact that N,, (at z = 1.5 A; with charge - 0.267) is close to the plane of the map. The contour at -25 kcal mol-’ around the region of the hydroxyl group is due to the negative partial charge ( - 0.322) on the oxygen (O,,). Table 2 shows that in the absence of any hydrogen-bonding interactions with the neighboring cyclic urea oxygens, the hydroxyl group has rotated away from the macrocycle and, therefore, from the plane of the map. In the Mulliken charge set calculation of Mimetic 5, Olos is located at z= 4.7 A and H,,, at z = 4.9 A.
Figure 4B shows the molecular electrostatic potential of Mimetic 5 in the potential-derived charge set. The structure is that of the global energy mini- mum at - 10.89 kcal mol- ‘. In contrast to the structure in the Mulliken charge set (Fig. 4A), the potential-derived charge set shows hydrogen bonding both between the hydroxyl group and the neighboring cyclic urea oxygen (Hlle- 0 ge6:
= 2.0 A, see Table 1) as well as between the imidazole and cyclic urea oxy- ( H9-Oe3 = 1.9 A). As in the Mulliken charge set, the position of the nega-
tive contours ( - 25 and - 50 kcal mol-l) is determined by the orientation of the phenylimidazole and hydroxyl moieties. Table 2 shows that the z= 2 A
160
plane passes through$he imidazole group, close to C, (z= 1.9 A), N8 (z= 1.8 A), and C,, (z = 2.4 A). The appearance of the contour at - 25 kcal mol-l extending from Ns past Nll is due to the presence of Ns at z= 1.8 A with a charge of - 0.569 and N,, at z= 3.0 A with a charge of -0.512. Comparison of patterns A and B in Fig. 4 shows a similar negative region around the imidazole group in the Mulliken charge set (Fig. 4A). However, because of lack of hy- drogen bonding between the imidazole group and Oe3 in the Mulliken charge set (Fig. 4A), the imidazole and hence its related negative contour are found in different regions of space than in the potential-derived charge set calcula- tion (Fig. 4B).
Figure 4B also shows a broad region of - 25 kcal mol- ’ enclosing two smaller regions of - 50 kcal mol-‘, which extends from the benzyl alcohol group to the benzene ring of the phenylimidazole group. This effect is due chiefly to the large negative partial charges on Cs ( - 0.535, z = 3.0 A) of the benzene ring and C& ( - 0.570, z = 2.7 A) of the benzyl alcohol fragment. This contour is not seen in the Mulliken charge set calculation (Fig. 4A) because, although the atoms have a similar position with re!pect to the z=2 A plane, the Mulliken charges are nearly zero: CIZ (2=3.8 A, 0.009) and C& (z=2.9 A, -0.009). Pattern A also differs from pattern B (Fig. 4) in the prtsence of a zero-level contour around Hl16. Table 2 shows that HJl6 (at z = 2.0 A, 0.268) is very close to the z= 2 A plane, while Olos (at z = 3.0 A, - 0.436) is further away. There- fore, the positive partial charge on HII6 balances the negative partial charge on O,,, resulting in a zero-level contour. This is in contrast to the Mulliken pattern, which shows a negative contour in this region because H1i6 is rotated away from the plane.
Comparison of the electrostatic potential contour maps of Mimetics 2 and 5 shows that conformational analysis with the two different charge sets may result in different hydrogen-bonding arrangements. This, in turn, can lead to qualitatively different electrostatic potential patterns, especially in the case of the uncharged species, Mimetic 5.
DISCUSSION
Comparison of the molecular electrostatic-potential patterns of the two mimics in both charge sets shows that the largest differences are due to the structural addition of the carboxylate anion, rather than to any particular dif- ferences in the choice of atomic point charges. For example, there are larger differences between the pattern of Mimetics 2 and 5 in the same charge set (Figs. 3A and 4A, respectively) than between those of Mimetic 2 in the two different charge sets (Fig. 3A and B). The dominant feature of the electrostatic potential pattern of Mimetic 2 in both charge sets is the broad region of neg- ative contours at - 100 and - 75 kcal mol-’ extending outward from the car- boxylate group. In comparison to this feature, the differences in the
161
electrostatic-potential pattern caused by the alternative hydrogen-bonding patterns of the hydroxyl group in the two charge sets (see Fig. 3) are minor.
In the absence of the anionic functionality (Mimetic 5, Fig. 4), the differ- ences in the molecular electrostatic-potential pattern caused by the magni- tudes of the point charges, as well as to the hydrogen-bonding patterns, become more obvious. For example, the contour at -50 kcal mol-’ found near the benzyl alcohol fragment only in the potential-derived charge set (Fig. 4B) was shown to be the result of the substantially more negative charges on the C,, and Cs6 atoms in that charge set. Because of the different hydrogen-bonding interactions of the imidazole functionality in the two charge sets, however, the negative contour associated with the imidazole group is found in different re- gions of space in pattern A and B of Fig. 4.
In general, both charge sets produce molecular electrostatic-potential pat- terns with contour levels of the same magnitude. The patterns of both Mimet- its 2 and 5 give a qualitative picture of the electron-rich binding site produced by the lone-pair electrons of the cyclic urea and methyl anisole oxygens of the macrocycle. In this sense, the electrostatic potential of both molecules is com- plementary to that of the cationic ammonium group of the L-alanyl p-nitro- phenyl ester perchlorate substrate. Long-range electrostatic interactions would, therefore, direct the ligand into the electron-rich binding site.
However, the placement of the minima in the electrostatic-potential pattern is dependent on the choice of atomic point charge. This feature varies with local charge differences and, more significantly, with hydrogen-bonding inter- actions. In general, the Mulliken charge set underestimates the hydrogen- bonding interactions. This is because the AMBER force field [ 131 used in this calculation contains parameters which have been “fine-tuned” by Weiner et al. to reproduce hydrogen-bonding interactions in protein and nucleic acid model systems calculated in a related potential-derived charge set. This indi- cates that care must be taken to use a charge set compatible with the molecular mechanics force field. Although both charge sets predict the same gross elec- trostatic features of molecular recognition, the sets differ significantly in the specific, localized features of the electrostatic-potential pattern. Further anal- ysis of the effect of atomic point charge on molecular conformation, as seen through a detailed study of the potential energy surface of Mimetics 2 and 5, will be the subject of a forthcoming paper.
This work was supported by grants to C.A.V. from the National Science Foundation (CPE-8404363 ), the donors of the Petroleum Research Fund of the American Chemical Society, the New Yersey Commission on Science and Technology, and the Ciba-Geigy, Schering-Plough, Hoffmann-LaRoche, and Squibb Corporations. C.A.V. thanks New Jersey Institute of Technology for a generous grant of computer time.
162
REFERENCES
1 J. Rebek, Jr., in J. Liebman and A. Greenberg (Eds.), Molecular Structure and Energetics, Vol. 10, VCH, New York, in press.
2 D. J. Cram, I. B. Dicker, C. B. Knobler and K. N. Trueblood, J. Am. Chem. Sot., 104 (1982) 6828.
3 D. J. Cram and H. E. Katz, J. Am. Chem. Sot., 105 (1983) 135. 4 D. J. Cram, P. Y.-S. Lam and S. P. Ho, J. Am. Chem. Sot., 108 (1986) 839. 5 C. A. Venanzi and J. D. Bunce, Int. J. Quantum Chem., 12 (1986) 69. 6 C. A. Venanzi and J. D. Bunce, Ann. N.Y. Acad. Sci., 4’71 (1986) 318. 7 C. A. Venanzi and J. D. Bunce, Enzyme, 36 (1986) 79. 8 C. A. Venanzi, in J. Liebman and A. Greenberg (I!%.), Molecular Structure and Energetics,
Vol. 10, VCH, New York, in press. 9 P. K. Weiner and P. A. Kollman, J. Comput. Chem., 2 (1981) 765.
10 U. C. Singh and P. A. KoIIman, J. Comput. Chem., 5 (1984) 129. 11 U. C. Singh and P. KolIman, QCPE Bull., 2 (1982) 17. Gaussian80 UCSF is available from
the Quantum Chemistry Program Exchange, Indiana University. 12 F. C. Bernstein, T. F. Koetzle, G. J. B. Williams,E. F. Meyer, Jr., M. D. Brice, J. R. Rodgers,
0. Kennard and T. Shimanouchi, J. Mol. Biol., 112 (1977) 535. 13 S. J. Weiner, P. A. Kollman, D. A. Case, U. C. Singh, C. Ghio, G. Alagona, S. Profeta, Jr. and
P. Weiner, J. Am. Chem. Sot., 106 (1984) 765.
