Reversal of Syn-Anti Preference for Carboxylic Acids along the

Reversal of Syn-Anti Preference for Carboxylic Acids along the Reaction Coordinate for Proton Transfer. Implications for Intramolecular Catalysis. Tho...
0 downloads 0 Views 608KB Size
13171

J. Phys. Chem. 1994, 98, 13171-13176

Reversal of Syn- Anti Preference for Carboxylic Acids along the Reaction Coordinate for Proton Transfer. Implications for Intramolecular Catalysis1 Thomas A. Montzka, S. Swaminathan,?and Raymond A. Firestone” Bristol-Myers Squibb Pharmaceutical Research Institute, 5 Research Parkway, P. 0. Box 5100, Wallingford, Connecticut 06492-7660 Received: October 12, 1993; In Final Form: March 29, 1994@

Carboxylate anions are more basic on the syn side because carboxylic acids are more stable with the proton syn to the carbonyl oxygen than anti. This led Gandour to propose that carboxylic anions should act as catalytic bases, as in enzymes, better on the syn side than the anti. However, numerous experimental tests have failed to reveal a strong preference. We resolve this problem by showing, on the basis of Linnett structures and MO calculations, that as a proton gradually departs from its firm syn position in the un-ionized COOH group, its preference for syn over anti diminishes rapidly, crossing over to a preference for anti before the midpoint of the reaction coordinate has been reached. Therefore in catalytic events, where the proton is partially transferred, the preference could favor either syn or anti and is small.

Introduction In 1981, Gandour published a seminal paper dealing with the question of syn vs anti orientation of the basic lone pair of electrons on carboxylate anions 1 that participate as bases in the catalytic event of enzymes.z Since the acidic hydrogen is more stable syn to the carbonyl than anti by 4.5 kcal/mol for formic a ~ i d and ~ - even ~ more for acetic7 and propionic acids,* it was reesoned that the syn lone pair must be more basic than the anti by that amount and that at equilibrium syn outnumbers anti by 103-104.9-18 Consequently, a carboxylate that acts as a base must do so much more powerfully at the syn lone pair than the anti. Depending on the magnitude of the Bransted coefficient, the catalytic power of syn orientation was estimated to be 10-1000 times greater than that of anti. In support of this idea, a survey of 10 relevant enzymes whose structures were then known revealed that catalytic carboxylates were oriented syn in nine and possibly all 10 cases. Hydrogen bonds between carboxyl and imidazole in small molecules19and in proteinsz0 strongly prefer the syn orientation.z1 This is significant because in the simplest intramolecular arrangements, with carboxylates acting as neighboring catalytic groups as in 2, only anti lone pairs are accessible.

O\

R+O

0-syn anti

O

1

2

L

E

1

El = Electrophile

Background: Experimental Studies The elegance of Gandour’s insight struck a chord with many people who then sought experimental confirmation. However, several varied attempts to date have surprisingly failed to provide much support, if any, for the greater efficacy of syn over anti lone pairs as bases or nucleophiles. An early paper reported an unusually large ApK, (3-4 units) between the first and second ionization constants of a syndisposed dicarboxylic acid relative to anti-disposed diacids, @

Current address: Gilead Sciences, Foster City, CA 94404. Abstract published in Advance ACS Abstracts, November 1, 1994.

0022-36541942098-13171$04.50/0

suggestive of an exceptionally strong H-bond to the syn lone pair of the carboxylate aniomZz However, there was no anti comparison with a diacid of closely similar structure, and an enforced syn-syn array of two carboxylate anions 3 is unusually rich in electrostatic energy, so that capture of a proton must be exceedingly facile on this ground alone.

-+00

0

j

6 3

Subsequent studies revealed that syn carboxylate exceeds anti by only 0.3-0.7 pK units in increasing the pKa of imidazoli~m;~~ both as a nucleophile and as a general base, imidazole is activated by carboxylate with no syn-anti p r e f e r e n ~ e ; ~ ~ ~ ~ ~ ~ ketone enolization catalyzed by intramolecularsyn carboxylates exhibits effective molarities (EMS) of only 0.5,7.0, and 17 M25 (cf. 56 M for a simple case with an anti carboxylatez6);and there is no advantage for syn over anti carboxylate in intramolecular SN2 to form lactones.z7 Thus to date, experimental studies have provided scant support for Gandour’s hypothesis. We propose that this situation may stem from focusing attention only on the stereochemistry of approach to the carboxylate anion. During an enzyme’s catalytic cycle, a carboxylate anion reversibly becomes a carboxylic acid. Could the orientation of the acid rather than the anion be the important factor? This question has not figured in previous discussions, probably because it follows from the greater stability of the syn carboxylic acid that the anti form must have the greater acidity, by the same factor. Therefore, since the high population of syn times its low acidity must exactly equal the low population of anti times its high acidity, the catalytic power of both conformers of the un-ionized carboxylic acid might appear to be the same. Syn vs Anti along the Reaction Coordinate. However, an implicit though unstated assumption in Gandour’s argument is that the proton wants to be syn, not only when bound to the carboxylate but also at all intermediate stages on the reaction coordinate (RC) for the ionization of the carboxylic acid, Figure 1. Thus, not only does the proton preferentially occupy the syn position in the un-ionized carboxylic acid by a substantial factor but it also preferentially attacks the syn position of the 0 1994 American Chemical Society

13172 J. Phys. Chem., Vol. 98, No. SO, 1994

Montzka et al.

I

I

RCOO‘ + H+

RCOOH

---- Apparent thermodynamically favored positionof proton Apparent acidity of proton

Figure 1. Syn vs anti position and acidity of proton along the RC according to Gandour (implied). We define the RC as the degree to which the proton has been transferred from formic acid to formate anion.

carboxylate anion and presumably prefers syn at all intermediate positions on the RC as well. It follows from this that an antisituated proton is more acidic than a syn proton by the same factor at all positions on the RC. However, this picture is based on only one firmly established data point, (I), indicated on Figure 1. At the extreme left, one can confidently place this point at a high sydanti ratio. At the extreme right there is no position of the proton at all. All intermediate points are assumed. Yet the intermediate points are what govern the experimental situation, because during the catalytic act, whether the catalyst is a carboxylic acid or its anion, the proton must be in some state of partial transfer at the transition state (TS). Therefore, we must look into these points more closely. In this paper we look at the reaction coordinate both in terms of qualitative electronic structure analysis (Linnett analysis) and by large basis set ab initio MO calculations on two model systems. The MO calculations were carried out with the Hondo 728$29 program using Dunnings split valence polarization basis sets (DZP).30 All the calculations were performed on an IBM 3090 mainframe computer. The model systems that we considered are the formic acid-formaldehyde and formic acidformate anion. Since the model systems used for this analysis involved partially and fully charged molecules, the electrostatic contributions to the reaction coordinate were minimized by using the extended conformation of the sydanti formic acid and acceptor complex.

RCOO-

R COOH

+

H+

---- position of proton acidity of proton

Figure 2. Syn vs anti position and acidity of proton along the RC

according to Linnett structures. possible with Linnett structures because double bonds are viewed as two-membered rings.ls According to this analysis, systems X=C(Y)-A-B ought generally to exist preferentially in the conformer having B eclipsed by X rather than by Y, and this has long been known to be the case.13 A 3 kcal/mol synanti energy difference is not out of line with experimental values. The earliest conceivable event along the RC for which a Linnett structure different from 4 can be written is 5, in which there is an unsymmetrical H-bond with the proton still firmly bonded to the carboxylate, which is polarized by the approaching base. The most interesting feature of 5 is that the two spinsets no longer coincide as they did in 4. One of the spin-sets in 5 (the 0’s) retains the possibility, by rotation, of existing in either the staggered or eclipsed conformation, but the other spinset (the x’s) is now fixed into the same position for both syn and anti, so that it no longer distinguishes between them. For this reason, the energy by which the proton prefers syn over anti is cut in h a p s (point ( 2 ) , Figure 2).

K

0

Linnett Analysis A Linnett a n a l y ~ i s of ~ ~the . ~ RC ~ produces some startling results. Looking first at the un-ionized carboxylic acid 4, we see that the array of spin-sets differs between syn and anti only at the valence shell of the hydroxyl oxygen atom. Q

.R

6

a

ta

0= 0--

.

.R

Q’

00

‘0

H 4 syn

Q

--@-H

@

@

4

anti

Syn 4 is clearly lower in energy than anti 4 because in syn 4 the spin-sets along the C-0 bond are staggered, whereas those in anti 4 are eclipsed. The energy difference can be approximated by using ethane as a in which the staggered conformer is 3 kcaymol more stable than the eclipsed.34 The staggered-vs-eclipsed analysis of the C - 0 bond in 4 becomes

0 5 anti

This conclusion is not obvious from an inspection of ordinary Lewis structures and leads to a departure from the (assumed) corollary of Gandour’s hypothesis (vide supra) that the catalytic power of the syn and anti conformers of an intact carboxylic acid ought not to differ. For a TS at point (2) (see Figure 2 ) , a syn carboxylic acid should have a distinct advantage, because there has been a fundamental change from point (I), the unperturbed carboxylic acid. At point (I), the product of population (which favors syn) times the acidity (which favors anti) exactly cancels, but at point (2) it does not cancel. The approach of a base toward RCOOH is fixed by the intramolecular geometry to be either syn or anti. If syn, it will find a properly aligned proton in 299% of the encounters which, however, is initially a very poor acidic catalyst. In contrast, an anti-positioned base, while finding a properly positioned proton

Syn- Anti Preference for Carboxylic Acids

J. Phys. Chem., Vol. 98, No. 50, 1994 13173

only rarely, will enjoy very superior catalytic power during those rare encounters. When the reaction has proceeded along the RC to point (2), Figure 2, the position of the proton, whether syn or anti, is fixed by the base, so that the numerical advantage of syn over anti is preserved. Now, however, the superiority in acidity of anti over syn no longer corresponds to the superiority in numbers of syn over anti, but to only half of it (on an energy scale). Consequently, there is a net gain in relative catalytic power of syn over anti carboxylic acids. If 3 k c d m o l is taken as the difference in stability between syn and anti, the gain in catalytic power of a carboxylic acid toward a syn- relative to an anti-positioned base becomes about 1.5 kcaY mol, for a factor of about 10 in those cases where the TS comes early in the RC. The factor grows with AE(syn-anti). Symmetrical H-Bond. A surprising result is obtained when the TS occurs at the midpoint of the RC. In this case the H-bond is symmetrical, as in 6.36 The array of spin-sets is the same as in 5, but with two important differences: (1) 6 syn corresponds to 5 anti, and vice versa; (2) the proton in 6 is no longer placed on a close pair as in 5, but rather on a non-close-paired electron. As in 5, one spin-set has lost the syn-anti preference and one retains it, but now the advantage in stability lies with the anti proton rather than the syn (point (3), Figure 2), a reversal from their relative positions earlier on the RC (points (1) and (2), Figure 2).37-39This conclusion is counter to that implied in Gandour's proposal. Note that the disagreement exists only in the middle of the RC, not at the ends.

0 6 syn

@

TABLE 1: Formic Acid-Formaldehyde H-Bond 0-0 distance, A

conformation

7 syn

-302.710 -302.718 -302.700 -302.712

W

2.904"

7 anti

E, hartrees

W

2.874"

116 297 815 757

AE, k c d m o l 0 0 5.84 3.48

" H-bond E minimum geometry optimized. (2), and (3), Figure 2, support the Linnett analysis in every important way. In accord with previous work3-6,9-12 we calculate syn formic acid to be more stable than anti by 5.84 kcaymol using HONDO 7 (DZP basis set). Formic Acid-Formaldehyde. The formic acid-formaldehyde system, 7, was studied as representative of a highly asymmetrical H-bond only a short distance along the RC, point (2) on Figure 2. This system has the advantage that both partners are uncharged, with charge separation still minimal at this early stage of proton transfer. Therefore, electrostaticeffects are unimportant. The results are presented in Table 1. The extended conformation of each conformer was fully optimized using HONDO 7 (DZP). The optimal 0-0 distances were 2.904 and 2.874 8, for syn and anti conformers, respectively, with the syn more stable than the anti by 3.48 kcal/mol. This is about half the AE for syn vs anti formic acid itself (point (l), Figures 1 and 2). This halving of AE at an early stage of proton transfer is precisely as predicted by the Linnett structures 4 and 5, corresponding to points (1) and (2), respectively (Figure 2), in which two spin-sets favor syn over anti in 4, but only one spinset does this in 5. It is noteworthy that to whatever extent charge transfer might play a role in electrostatically favoring one of the forms it would favor syn, not anti, so that if the small electrostatic contribution were subtracted from the AE(antisyn), it would be even less than 3.48 kcal/mol.

o

6 anti

7 syn

With a TS that is centered on the RC, the syn conformer is preferred on both counts, population and acidity. Its advantage in population is retained by the H-bond, and yet it holds an absolute advantage in acidity as well, since it now has become the less stable conformer. This conclusion is opposed to the premise2J9,21-25*27 that the syn side of the carboxylate anion is always the more basic one. When the proton moves further away from the carboxylate, however, the picture fades because the carboxylate's electrons become freer to assume their optimum positions for the carboxylate anion, and at the extreme right side of the RC ther is no syn-anti preference at all. That's why the Rebek and Zimmerman experiments found none. It is at the carboxylic acid rather than the carboxylate anion side of the RC that the search for Grandour's syn > anti effect should take place.

Formic Acid-Formate Ion. The formic acid-formate ion system, 8, was followed down the RC for complete proton transfer, for four extended conformations: S-A (syn formic acid-anti formate), A-A, S - S , and A-S. Formate ion was chosen as the base for simplicity. Data are presented for each conformation in Table 2 and Figure 3.

a

8 S-A

A-A

H

Molecular Orbital Calculations The predictions made on the basis of Linnett structures were then compared with results obtained from rigorous molecular orbital calculations. In order to correlate our results with Linnett analysis, we identified the RC as the movement of the hydrogen atom from one molecule to another. By optimizing the geometry at each point, we have identified the TS along this RC. The energetics from the MO calculations for points (l),

7 anti

a

A-s

H

8 S-S

On Figure 3 point (1) on the RC represents formic acid iformate anion at infinite distance. Point (2) is the first energy

13174 J. Phys. Chem., Vol. 98, No. 50, 1994

Montzka et al.

TABLE 2: Proton Transfer Formic Acid-Formate Anion (See Figure 3) conformation

point on RC

S-A

(1) (2) TS (1) (2)

A- A

-377.052 -377.090 -377.090 -377.042 -377.096 -377.094 -377.042 -377.097 -377.090 -377.052 -377.091 -377.089

TS A-S

(1) (2) TS (1) (2) TS

s-s

Ea, kcaVmol

AE, k c d m o l

E, hartrees" 0

241 817 802 941 857 188 941 528 802 241 407 586

0-H,Ab 0.950 1.053 1.100 0.945 1.005 1.185 0.945 0.997 1.100 0.950 1.024 1.191

3.79 0.01 5.84

0 1.68 5.84

0 4.22

0 3.84 1.14

0-0,

A

00

2.442 2.398 m

2.502 2.370 m

2.537 2.398 00

2.502 2.382

Energies were calculated with full geometry optimization except when defining the reaction coordinate. In the unsymmetrical H transfer the

H-0 distance was fixed over the reaction coordinate range, and in the symmetrical cases, symmetry operations were used. For the formic acid molecule losing the proton.

-

Energy S-A Energy A-A

* Transition States

(2)

(3)

Reaction Coordinate -377.04

1

-377.05

-377.06

P a3

* 3

-377.07

c

w

Energy S-S Energy A-S

Transition States

-377.00

-377.09

-377.10

!,

1

(1)

(2)

1 (3)

(4)

Reaction Coordinate

Figure 3. Proton transfer from formic acid to formate anion. Data are from Table 2. minimum, i.e. an almost symmetrical H-bond with the proton still somewhat closer to its original formate partner. Transition states are marked with the usual symbol *. Point (3) is the second energy minimum, and point (4) is formate anion formic acid at infinite distance, after complete proton transfer. All points (except TSs) are lined up vertically, so that the abscissa depicts progress along the RC rather than absolute distance. Points ( 2 ) , *, and (3) are close in structure and, as a group, correspond to point (3) in Figure 2. The points at the left, of course, simply tell us that syn formic

+

acid (+formate) is more stable than anti (+formate) by 5.84 kcal/mol. Each graph converges at point (4) on the right because the energy difference fades as the proton departs from its parent formic acid. The most important feature of Figure 3 is the energy crossover that occurs between the beginning of the reaction, point (l), and the well-established, almost symmetrical H-bonds about halfway down the RC, point (2). This crossover confirms the changeover that was predicted by the Linnett analysis, from syn > anti for formic acid to anti > syn for the midpoint of the RC

Syn-Anti Preference for Carboxylic Acids -

".&

J. Phys. Chem., Vol. 98, No. 50, 1994 13175 0

6-

S-A

A-A

s-s

A-S 0' 6 6Figure 4. Electrostatic relationships of formic acid-formate anion arrays at the first minimum of the RC.

(point (3), Figure 2). The result is the same whether the base, formate ion, is positioned syn or anti. Another important prediction is also c o n f i i e d by these data. The more stable form of the acid must be the less acidic and vice versa. Therefore, Ea's for proton transfer for each of the four conformations were calculated. The Ea's are indeed much lower for S-S and S-A, which have become the higher-energy conformations after the crossover, than for their lower-energy counterparts A-S and A-A.40 If syn formic acid becomes higher in energy than anti as the midpoint of the RC is approached, not for reasons of internal bonding as we propose, but rather because electrostatic repulsion of the approaching base (formate ion) is greater37 (see Figure 4), then further approach of the base would be more difficult toward syn formic acid than toward anti, resulting in a higher, not lower, Ea for proton transfer from the syn acid. This is one argument for the position that electrostatic considerations, while not negligible, are not dominant in this situation. Another argument is to consider the relative energies of S-A/ A-A vs S-S/A-S at the first minimum (Figure 4). When the base is oriented anti, one of its two negatively charged oxygen atoms is further from the scene than when it is oriented syn. Switching from an anti- to a syn-oriented base ought then, on purely electrostatic grounds, to increase even further the stability of anti acid vis-&vis syn. In other words, the difference between the energies of S-A and A-A would be smaller than that of S-S and A-S at the first minimum. However, this is scarcely observed. The energetic advantage of anti formic acid vs syn at the first minimum is 0.00604 hartrees with anti-oriented base and 0.00612 hartree with syn-oriented base. The effect of base orientation is indeed in the right direction for an electrostatic effect, but negligible in magnitude (0.00008 hartree = 0.05 k c d mol). Therefore, we can confidently say that the changeover in relative energies of syn and anti formic acid that occurs between points (1) and (3), Figure 2, is not caused primarily by electrostatic effects.

Conclusions Linnett theory predicts a rapid diminution in syn =- anti stability (and thus in anti > syn acidity) of carboxylic acids as the proton is gradually removed by a base, with a crossover occurring before the midpoint of the RC. High basis set molecular orbital calculations support these predictions for formic acid with formaldehyde and formic acid anion and say furthermore that, after the crossover, the syn form becomes kinetically as well as thermodynamically the more acidic one. Thus, Gandour's postulate that carboxylic acid anions should be more basic on their syn side than anti, although correct for complete protonation of the anion to the un-ionized carboxylic acid, does not hold for all intermediate points on the RC. The fact that enzymes preferentially orient catalytic carboxylates syn could then result, depending on the position of the catalyzed TS along the RC, not only from carboxylate anion catalysis in some cases but also in others from catalysis by the carboxylic acid form.

Supplementary Material Available: Molecular orbital calculation data for determining reaction coordinates in Figures 3 (2 pages). Ordering information is given on any current masthead page. References and Notes (1) Application of the Linnett Electronic Theory to Organic Chemistry, Part 8. Part 7: Firestone, R. A. J. Org. Chem. 1980, 45, 3604. (2) Gandour, R. D. Bioorg. Chem. 1981, 10, 169. (3) Peterson, M. R.; Csizmadia, I. G. J . Am. Chem. SOC. 1979, 101, 1076. (4) Miyazawa, T.; Pitzer, K. S . J. Chem. Phys. 1959, 30, 1076. (5) Lide, D. R., Jr. Trans. Am. Crystallogr. Assoc. 1966, 2, 106. (6) Mariner, T.; Bleakney, W. Phys. Rev. 1947, 72, 792. (7) Meyer, R.; Ha, T.-K.; Gunthard, H. H. Chem. Phys. 1975, 9, 393. (8) Allinger, N. L.; Chang, S . H. M.Tetrahedron 1977, 33, 1561. (9) Other publications on this topic give syn-anti differences of the same order of magnitude. Experimental AE = 4.1 kcdmol (see ref 10). Calculated AE = 5.93, 6.98 for formic and acetic acids, respectively (see ref 11). Calculated AE = 5.8 (SCF) or 5.4 (MP2) for formic acid (see ref 12). (10) Hocking, W. H.; Winnewisser, G. Z. Naturforsch. 1976,31A, 422, 438, 995. Winnewisser, G.Z. Naturforsch. 1976, 31A, 1113. Hocking, W. H.; Winnewisser, G. Z. Naturforsch. 1977, 32A, 1108. (1 1) Marcoccia, J. F.; Csizmadia, I. G.; Peterson, M. R.; Pokier, R. A. G a u . Chim. Ital. 1990, 120, 77. (12) Basch, H.; Stevens, W. J. J . Am. Chem. SOC. 1991, 113, 95. (13) The preference for syn over anti carboxylic acid is part of a larger pattern of conformational preferences in which sp3 carbon and divalent oxygen bonded to C=C or C-0 adopt preferred conformations in which C-H, C-CH3, C-Cl, 0-H, and O-CH3 are eclipsed by the vinyl or carbonyl group (see refs 14 and 15). A possible explanation for this apparent anomaly is that double bonds can be viewed as two-membered rings (see refs 16- 18), which would then make the preferred conformations staggered ones and hence energy minima on the rotational reaction coordinate (see ref 14). (14) Conformational Analysis; Eliel, E. L., Allinger, N. L., Angyal, S. J., Momson, G.A., Eds.; Interscience: New York, 1965; p 19. (15) Karabatsos, G.J.; Fenoglio, D. J. J. Am. Chem. SOC.1%9,91,1124. (16) Mulliken, R. S. Tetrahedron 1959, 6, 68. (17) Pople, J. A. Q. Rev., Chem. SOC. 1957, 11, 273. (18) Linnett, J. W. J. Am. Chem. SOC. 1961, 83, 2643; The Electronic Structure of Molecules; Methuen: London, 1964. (19) Nabulsi, N. A. R.; Fronczek, F. R.; Gandour, R. D. Biochemistry 1990, 29, 2199. (20) Ippolito, J. A.; Alexander, R. S.; Christianson, D. W. J . Mol. Biol. 1990, 215, 457. (21) Gandour, R. D.; Nabulsi, N. A. R.; Fronczek, R. R. J . Am. Chem. SOC. 1990, 112, 7816. (22) Rebek, J., Jr.; Duff, R. J.; Gordon, W. E.; Panis, K.J . Am. Chem. SOC. 1986, 108, 6068. (23) (a) Zimmerman, S. C.; Cramer, K. D. J. Am. Chem. SOC. 1988, 110, 5906. (b) Huff, J. B.; Askew, B.; Duff, R. J.; Rebek, J., Jr. J. Am. Chem. SOC.1988,110,5908. (c) Zimmerman, S . C.; Korthals, J. S.; Cramer, K. D. Tetrahedron 1991, 47, 2649. (24) Cramer, K.D.; Zimmerman, S. C. J. Am. Chem. SOC. 1990, 112, 3680.

13176 J. Phys. Chem., Vol. 98, No. 50, 1994 (25) Tadayoni, B. M.; Huff, J.; Rebek, J., Jr. J . Am. Chem. SOC.1991, 113, 2247. Tadayoni, B. M.; Parris, K.; Rebek, J., Jr. J . Am. Chem. SOC. 1989, 111, 4503. (26) Harper, D. C.; Bender, M. L. J . Am. Chem. SOC.1965, 87, 5625. (27) Tadayoni, B. M.; Rebek, J., Jr. Bioorg. Med. Chem. 1991, I, 13. (28) Dupui, M.; Watts, J. D.; Villar, H. 0.; Hurst, G. F. B. Compuf. Phys. Commun. 1989, 53, 415. (29) Hondo 7 was obtained courtesy of IBM Corp. (30) (a) Dunning, T. H., Jr.; Hay, P. J. In Modem Theoretical Chemistry; Schaefer, H. F., HI,Ed.; Plenum: New York, 1977; Vol. 4. (b) Dunning, T. H., Jr. J. Chem. Phys. 1970,53,2823. (c) It has been shown in hydrogenbonding systems that the interaction energy calculated from minimal basis set SCF suffers from superposition errors, which diminish when one uses split valence basis sets. In this instance, we have used a split valence basis set along with polarization functions to minimize superposition errors further. One example of hydrogen-bonding energy in the case of water dimers as a function of basis sets: Matsuoka, 0.;Clementi, E.; Yoshimine, M. J . Chem. Phys. 1976, 64, 1351 and references cited therein. (31) Firestone, R. A. J. Org. Chem. 1969, 34, 2621. (32) In Linnett’s structural theory (see refs 18 and 31), for atoms with eight valence electrons, the electrons, as nonpaired spatial orbitals (npso’s), are arranged in two spin-sets, each with four electrons of like spin depicted as x’s and 0’s. Each spin-set is a tetrahedron, centered on the nucleus, whose shape is determined by both electrostatic repulsion and the Pauli principle, more rigid therefore for first- than second-row elements. The two spin-sets can be placed independently, and although the Pauli principle does not apply between them, they prefer not to coincide, owing to electrostatic repulsion. A single covalent bond enforces coincidence of the tetrahedra at one comer, with the other six npso’s not coinciding, but in an atom with two or more single bonds, or a double bond, the two tetrahedra must coincide. Covalent bonds of one, two, three, four, five, or six electrons are possible. In odd-electron bonds the shared npso’s do not coincide. Spinsets of fewer than four electrons are possible, but do not appear in this paper.

Montzka et al. (33) Ethane’s barrier to rotation arises from C-H bond repulsions: Sovers, 0. J.; Kem, C. W.; Pitzer, R. M.; Karplus, M. J . Chem. Phys. 1968, 49,2592. Jorgensen, W. L.; Allen, L. C . J . Am. Chem. SOC.1971,93,567. (34) Weiss, S.; Leroi, G. E. J. Chem. Phys. 1968, 48, 962. (35) Structure 5 can also be viewed as a polarized resonance form of 4, which is normally a minor contributor to the resonance hybrid, but which is now stabilized by the approach of the base. (36) A reviewer has pointed out that 6 syn and anti, both planar, are not the only conceivable Linnett shapes, since the H-bond could also be attached to the spin-set of unpaired o’s, which are not in the plane of the RCOO group. However, by the time 6 has been reached along the RC, the position of the base has already been fixed as either syn or anti, and this in tum fixes the proton’s position in one or the other planar configuration. (37) There are other examples in the literature in which an anti preference is calculated for a partially transferred proton, but on grounds quite different from ours. Biformate ion is predicted to favor anti-anti over syn-syn (see ref 38), not for reasons related to the Linnett structural analysis but instead because formate-formate electrostatic repulsion is lower in antianti, in which the second oxygen atom in each formate (the one not bonded to H), with its partial negative charge, is pointed away from rather than toward the other formate. The most stable form was actually predicted to be syn-anti, with the proton closer to the anti-formate, from considerations partially of CH-O bonding and partially electrostatic. In proton transfer between formate and acetaldehyde anions (at a-carbon), an intermediate point was calculated to favor anti-formate by 2.5 kcaYmol in the gas phase, but once again principally for electrostatic reasons, since solvation of the enolate reversed the prediction toward syn-formate with a 5 kcal advantage (see ref 39). (38) Emsley, J.; Hoyte, 0.P. A.; Overill, R. E. J . Am. Chem. SOC.1978, 100, 3303. (39) Li, Y.; Houk, K. N. J . Am. Chem. SOC.1989, 111, 4505. (40) This is not merely a truism, because examples are known in which the more energy-rich compound of a pair reacts more slowly, yet by the same mechanism (see ref 1).