ON THE MECHANISM OF CATALYSIS BY VITAMIN B12 BY JONATHAN D. BRODIE DEPARTMENT OF BIOCHEMISTRY, STATE UNIVERSITY OF NEW YORK AT BUFFALO Communicated by H. A. Barker, December 5, 1968 Abstract.-The requirements for alkylation and binding of nitrogenous ligands to cobalamins and cobinamides suggest that the central cobalt atom can exist either approximately in the plane of the corrin or axially displaced above this plane. Distortion of the normal octahedral complex by axial displacement results when the dimethylbenzimidazole moiety of the coenzyme is prevented from coordinating with the cobalt. A general hypothesis of B12 coenzyme catalysis that involves the potential of Co'II-Re and CoI-Re transitions is proposed on the basis of these considerations. Since the original isolation of deoxyadenosyl-b12, a number of unusual enzymatic transformations catalyzed by this coenzyme have been delineated.' In general, these reactions may be viewed as a double 1,2 shift in which one of the migrating groups is hydrogen.2 The central role of the carbon-cobalt bond in the catalytic function of B12 coenzymes has been revealed in the last few years in the laboratories of Abeles3 and Arigoni,4 who used the deoxyadenosyl-b12-dependent propanediol dehydrase system from A. aerogenes that catalyzes the reaction: OH I CH37---CH20H~z:CH3CH2--CH(OH)2i--CH,3CH2CHO. H The labeling data have clearly established the mechanism to be the abstraction of hydride by the 5' carbon of the coenzyme followed by hydroxyl migration and the addition of hydride from the coenzyme to the substrate. 4 The similar 1,2 shift catalyzed by methylmalonyl CoA mutase, COOH / CH3CH=HOOC-CH2-CH2COSCoA, COSCoA cannot be rationalized by an analogous hydride mechanism, since this would require attack of a primary carbonium ion on the carbonyl carbon of the thiolester. The hydrogen transfer mechanisms are probably disparate, although the incorporation of hydrogen from the substrate to the 5' carbon of the adenosyl moiety of the coenzyme appears to be identical to that observed in propanediol dehydrase.' The cleavage of the carbon-cobalt bond appears to be an intrinsic step in all known B,2-dependent reactions.2 Accordingly, we wish to report on the nature of this carbon-cobalt bond as delineated by ligand binding studies and the spatial relationship of the substituents of the d2 orbital to the corrinoid system. 461
462 BIOCHEMISTRY: J. D. BRODIE PRoc. N. A. S. Materials and Methods.-Vitamin B12 was purchased from Pierce Chemical Co. A series of alkyl halides and substituted amines were products of Distillation Products Industries. Aquocobinamide was prepared by the cerous ion-catalyzed hydrolysis of aquocobalamin (B12a) as previously described.6 Alkyl cobinamides were prepared by borohydride reduction of aquocobinamide to Co'-cobinamide followed by addition of the appropriate alkyl bromide or iodide in the dark, and purified by chromatography on Sephadex G-10 and CM-cellulose. Ligand binding studies were performed at room temperature (230) in the dark, and spectra were recorded with a Cary 14 recording spectrophotometer. Results.-Properties of secondary alkyl cobinamides: Cyclohexyl cobinamide and sec-butyl cobinamide gave essentially identical spectra. Figure 1 shows the spectrum of cyclohexyl cobinamide before and after photolysis. The photolysis 1.0 0.5- O.D.,, 2501 1 3501 4501 5501 650 FIG. l.-pectrum of cyclohexyl cobinamide. Solid line is cyclohexyl cobinamide (2.2 X 10-5 M) in water. Dashed line is aquocobinamide formed after exposure to light. Concentration determined after conversion to dicyanocobinamide, e368 = 30.4 X 103. product is identical to authentic aquocobinamide. There are remarkable similarities between the secondary alkyl cobinamide spectrum and that of Co",-cobinamide. In contrast to the latter, however, these alkyl products are not susceptible to air oxidation and are unaffected by the presence of mercaptans, but the alkylated compounds are extremely photolabile. Reaction of methyl cobinamide and cyclohexyl cobinamide with cyanide: Figure 2 shows the spectrum of methyl cobinamide before and after the addition of excess KCN. The spectrum of the methylcyanocobinamide reverts to the original methyl cobinamide spectrum upon acidification and phenol extraction, which shows that cyanide has added to the d6 position only. In contrast, Figure 3 shows that treatment of cyclohexylcobinamide under the same conditions gives dicyanocobinamide on standing, an indication that the carbon-cobalt bond of the secondary alkyl cobinamide is labilized in the presence of cyanide. Binding of amines to methyl and cyclohexyl cobinamide: The spectra of the simple alkyl-cobinamides are unaffected by ph. However, a nitrogenous ligand
VOL. 62, 1969 BIOCHEMISTRY: J. D. BRODIE 463 O.D., A 250 350 Ir 450 I I 450 550 650 FIG. 2.-Spectrum of methyl cobinamide and methyl cyanocobinamide. (A) Methyl cobinamide (1.4 X 10' M) in water. (B) Spectrum in 0.98 M KCN. O.D. Iftf t hw FIG. 3.-Spectrum of cyclohexyl cobinamide in presence of KCN. Cyclohexyl cobinamide (2.4 X 10- M), e3o5 = 43.3 X 103, in 0.96 M KCN. Allowed to stand in dark at room temperature. in the d6 position profoundly alters the spectrum to yield a spectrum very close to that of methyl cobalamin. Titration of methyl cobinamide with NH40H and pyridine gives the results shown in Figures 4 and 5. It can be seen that the nature of the nitrogenous ligand has very minor effects on the spectra, with differences noted only in the region near 340 m. MIethylamine and ethylamine complexes form at concentrations intermediate between ammonia and pyridine. Piperidine, although an excellent nucleophile, is a very poor ligand-former, presumably due to steric hindrance. We have not been able to detect interaction of any amine with cyclohexyl cobinamide, even upon raising the temperature to 60.
464 BIOCHEMISTRY: J. D. BRODIE PROC. N. A. S. 1.0-0.5-0.0. 250 350 450 550 650 FIG. 4.-Titration of methyl cobinamide with NH40H. Methyl cobinamide (1.7 X 10- M) was titrated with NH40H to a final concentration of 15.5 M base. Ligand binding is associated with a decrease in absorbance at 305 and 462 my and an increase at 262, 370, and 530 mjs. The ligand concentration for 50% change in absorbance at the indicated wavelengths is 4.7 M. 1.0 05Ak 250 350 450 550 65( ml FIG. 5.-Titration of methyl cobinamide with pyridine. Methyl cobinamide (1.7 X 10' M) was titrated with pyridine to a final concentration of 0.66 M base. The ligand concentration for 50% change in absorbance at 305, 340, 460, and 515 my was determined to be 0.11 M. Synthesis of cyclohexyl "cobalamin:" The preceding results indicate that the absence of a nitrogen base at da (cobinamide) enables the introduction of a bulky alkyl substituent at the e1 position. However, borohydride reduction of aquo-
VOL. 62, 1969 BIOCHEMISTRY: J. D. BRODIE 465 cobalamin followed by reaction with cyclohexyl or other secondary alkyl halides failed to yield alkylated products. A variety of platinum, palladium, rhodium, and cobalt catalysts, while effecting partial reduction of Co"'-cobalamin, did not enable alkylation even at ph 1.0. By analogy with the properties of the cobinamides, we had expected that protonation of the dimethyl benzimidazole moiety would enable alkylation. Accordingly, a different route of synthesis was attempted (as shown in Fig. 6) that involved reversible blocking of the cobalt as a key step, followed by quaternization of the dimethylbenzimidazole to prevent reassociation. The wellknown displacement of the da ligand by cyanide7 enabled alkylation of N7 of the benzimidazole with methyl iodide in refluxing aqueous methanol in the presence of potassium carbonate. Acidification to ph 2 followed by phenol extraction gave N-methyl monocyano-"cobalamin," with absorption maxima at 351, 495, and 525 mjy. Reduction with borohydride enabled the smooth alkylation by cyclohexyl iodide to form cyclohexyl-"cobalamin"' that had the spectral characteristics of cyclohexyl cobinamide. CN CN CN Co Co0 CH3I Co C-h QCN (CN N N-CH3 N CN INYCo NaBH CL N-CH3 0 r C N-CH3 FIG. 6.-Synthesis of cyclohexyl-"cobalamin." Experimental details are given in the text. The product is cyclohexyl-3,5,7-trimethylbenzimidazolyl cobamide. A control experiment showed that the phosphate ester linking the nucleoside to the C ring of the corrin was not hydrolyzed by the quaternizing conditions. Discussion.-A fundamental relationship between the polarizability of the carbon-cobalt bond and the steric environment of the cobalt in the B12 coenzymes is established by the chemical and spectroscopic studies reported in this paper. This relationship offers a means of reconciling the apparent differences in mechanism of two similar deoxyadenosyl-b,2-dependent reactions that involve hydrogen transfer to the 5' carbon of the adenosyl moiety, and a means of including the methyl transfer in methionine biosynthesis in a generalized mechanism of B12 coenzymes. The differences in chemical reactivity of cobinamide (I) when compared with cobalamin (II) were not entirely unexpected. The trans effect is clearly evi-
4;66 BIOCHEMISTRY: J. D. BRODIE PRoc. N. A. S. denced by the effect of the d6 substituent on the susceptibility to alkylation. In aquocobinamide, where d6 is H20 or hydroxide, a hindered secondary alkyl halide (cyclohexyl) can easily be attacked by the Co' species. On the other hand, d5 d5 0 Co d6 kd I II alkylation of aquocobalamin, in which the d6 position is coordinated with nitrogen-7 of the dimethylbenzimidazole, could not be effected. Although it might be argued that the secondary alkyl substituent is destabilized by induction, it is probable that in the case of the cobalamins, the trans inductive effects are minimal when compared with steric factors.8 The experimental results support this view. Thus, the Col cobinamide can be alkylated by a bulky substituent because of the absence of a strong ligand in the d6 position, which allows the cobalt to remain above the plane of the corrin system, which hinders the binding of nitrogenous ligands. Conversely, the presence of a strong ligand (dimethylbenzimidazole) in the Co'-cobalamin series tends to force the cobalt into the plane of the corrin, which precludes the formation of a carbon-cobalt bond with a bulky alkyl group. This is consistent with the alkaline cyanide decomposition of cyclohexyl cobinamide (which may be visualized as the addition of a strong ligand to the d6 position that tends to force the cobalt into the plane of the corrin, with concomitant labilization of the carbon-cobalt bond caused by steric hindrance) followed by rapid addition of cyanide to give the observed dicyanocobinamide, with no evidence of a stable monocyano cyclohexyl derivative.' More direct support for this view may be found by examination of the synthesis of the cyclohexyl-"cobalamin" that is formed only after quaternization of the nitrogenous ligand. An understanding of the enzymatic behavior of the B,2 coenzymes may be provided by the preceding considerations in which a hydride transfer mechanism (operative for propanediol dehydrase) would be facilitated by a carbon-cobalt cleavage into a Co1-R9 species that would be stabilized by the absence of an electron donor in the d6 position. This would suggest that in carbonium ion transfers, as in methionine biosynthesis'0 or hydride abstractions (cf. ref. 33 cited in ref. 2), the enzyme binds the coenzyme in a manner in which the dimethylbenzimidazole is not coordinated with the cobalt. On the other hand, a proton abstraction reaction, which may be used to explain the isomerization of methylmalonyl CoA to succinyl CoA, would be visualized as involving a heterolytic carbon-cobalt cleavage to form a Co,,'-Re pair in which the dimethylbenzimidazole coordinated to the cobalt would stabilize the intermediate CoIII species. Thus, it appears likely that the key transition in all coenzyme-bn2 catalyzed enzymatic reactions may be effected merely by altering the distance of the cobalt-nitrogen bond, which in turn changes the symmetry of the cobalt and the polarization of the carbon-cobalt bond. We shall test this hypothesis by circular
VOL. 62, 1969 BIOCHEMISTRY: J. D. BRODIE 467 dichroism and magnetic resonance studies with model compounds and appropriate enzymatic experiments. The author gratefully acknowledges Mr. Melvin Spadford for his excellent technical assistance, Drs. Curtis Hare, R. G. Wilkens, and R. M. McLean for helpful discussions, and Dr. W. B. Elliott for the use of his Cary 14 spectrophotometer (provided by USPHS grant no. GM-06241). * This work was supported by grant no. AM-10479 from the National Institutes of Health. lbaker, H. A., H. Weissbach, and R. D. Smyth, these PROCEEDINGS, 44, 1093 (1958). 2 Hogenkamp, H. P. C., Ann. Rev. Biochem., 37, 225 (1968). 3 Frey, P. A., M. K. Essenberg, and R. H. Abeles, J. Biol. Chem., 242, 5369 (1967). 4 R6tey, J., A. Umani-Ronchi, J. Seibl, and D. Arigoni, Experientia, 22, 502 (1966). 5 Cardinale, G. J., and R. H. Abeles, Biochim. Biophys. Ada, 132, 517 (1967). 6 Friedrich, W., and K. Bernhauer, Chem. Ber., 89, 2507 (1956). 7 Ibid., p. 2030. 8 Schrauzer, G. N., and R. J. Windgassen, J. Am. Chem. Soc., 89, 1999 (1967). 9 Hogenkamp, H. P. C., J. E. Rush, and C. A. Swenson, J. Biol. Chem., 240, 3641 (1965). 10 The experimental data obtained with methionine synthetase are essentially considered in this light by Hogenkamp.2