Proc. Natl. Acad. Sci. USA

Vol. 74, No. 3, pp. 811-815, March 1977

Chemistry

Biochemical catalysis involving coenzyme B-12: A rational stepwise mechanistic interpretation of vicinal interchange rearrangements [electrocyclic cleavage of cobalt corrins/oxidative addition to d8-Co(I)/reductive elimination to form d8Co(I)/cobalt-carbene complex/1,2-carbenoid rearrangement]

E. J. COREY*, N. JOHN COOPER*, AND M. L. H. GREENt *

Department of Chemistry, Harvard University, Cambridge, Massachusetts 02138; and t Department of Inorganic Chemistry, University of Oxford, Oxford

OXI 3QR, England Contributed by E. J. Corey, December 13, 1976

A mechanism is proposed for the catalytic acABSTRACT tion of coenzyme B-12 which is consistent with current knowledge of organometallic reactions and with the experimental data now available from biochemical studies. A key feature of the proposal is an electrocyclic cleavage of the coenzyme that reduces cobalt and also leads to a 1,19-seco-corrin. The seco-corrin serves as a tridentate ligand about Co(I). This arrangement permits the metal to take part in the kinds of organometallic reactions that are ideal for coenzyme B-12 catalysis, including oxidative addition and its reverse, reductive elimination. It is further proposed that the rearrangement steps involve cobaltcarbene complexes.

The history of the biochemistry of coenzyme B-12 has been replete with surprise since the discovery by Lenhart and Hodgkin (1) of a direct linkage between the 5' carbon of 5'deoxyadenosine and cobalt as shown in Fig. 1. Over the past two decades a wide variety of enzymic reactions involving coenzyme B-12 have been discovered (2-5), most of which are without precedent in terms of the reactions of organic chemistry. These corrinoid-dependent enzymic reactions include nine molecular rearrangements which can be formulated generally as a stereospecific interchange of two groups (H* and X) on adjacent carbons (vicinal interchange): H X X H* 1 y I- 2 I2 y Y H Among the specific examples of such a rearrangement are the conversions of (a) glutamate to ,B-methylaspartate, (b) methylmalonyl-CoA to succinyl-CoA, (c) 1,2-propanediol to propionaldehyde, and (d) 2-aminoethanol to acetaldehyde (2). Although the vicinal interchange process has been extensively studied, especially for the case of (c) and the enzyme 1,2-diol dehydrase (propanediol dehydratase; 1,2-propanediol hydrolyase, EC 4.2.1.28) (2, 3), the pattern of experimental observations has been enigmatic and no clear cut interpretation has emerged. The data obtained for 1,2-diol dehydrase may be summarized as follows. (1) Both R and S 1,2-propanediol are converted to propionaldehyde. (2) The hydroxyl group at C-2 migrates to C-1 to form a gem-diol structure from which one of the diastereotopic hydroxyls is lost stereospecifically. (3) The deuterium atom in (1R,2R)-1,2-propanediol-1-d migrates to C-2 during conversion to propionaldehyde. The rate of this process is one-twelfth that for undeuterated substrate

migrates to C-2 during conversion to propionaldehyde and the overall deuterium kinetic isotope effect kH/kD = 1. (5) The hydroxyl at C-2 is replaced by hydrogen from C-1 with inversion of configuration at C-2. (6) During the catalytic process in the related transformation of 1-aminopropan-2-ol to propanal, 5'-deoxyadenosine (enzyme-bound, not free) is formed reversibly from the 5'-deoxyadenosyl ligand of the coenzyme. (7) Using coenzyme in which the C-5' position of the 5'deoxyadenosyl ligand is tritiated and unlabeled 1,2-propanediol, the enzymic reaction produces tritiated propionaldehyde. (8) Reaction of diol dehydrase, coenzyme, ethylene glycol (unlabeled), and tritiated 1,2-propanediol produces acetaldehyde that is tritiated. (9) No heavy isotope is incorporated into the aldehyde produced when tritiated water is used as solvent for the enzymic rearrangement of unlabeled diol. Although there have been a number of suggestions regarding the types of processes that might be involved in the catalysis of vicinal interchange rearrangements by the various enzymeB-12 coenzyme systems (3-9), most, if not all, are incompatible with existing data. The only mechanistic proposal (involving highly reactive free radicals) that seems to have attracted any acceptance whatsoever neglects the question of how vicinal migration occurs and is difficult to reconcile with chemical experience (vide infra). Derivation of a mechanistic hypothesis for the vicinal interchange reaction Chemical Background. It seems reasonable to start with two points that seem solidly based: (i) by some process the bond between the 5'-deoxyadenosyl ligand and cobalt is replaced by a bond between the 5'-deoxyadenosyl group and a hydrogen that was originally bonded to a carbon in the substrate, and (ii) the carbon of the substrate that gives up the hydrogen becomes bonded to cobalt. Given these two requirements, one process from the body of knowledge within organo-transition-metal chemistry stands out as preeminently suited to provide an effective mechanism: the oxidative addition-reductive elimination reaction (10-12). In its most common form the process can be formulated as 1 i=± 2, wherein 1 is a square planar complex with the metal M possessing the d8 electronic configuration. More specifically, in terms of the interaction between substrate (R2CH2CH2R3) and B-12 coenzyme, this model takes the form 3 + 4 z 5 ± 6 + 7, with

w~ I--,x

(kH/kD = 12).

Y

z

(4) The hydrogen atom at C-1 in (lR,2S)-1,2-propanediol-1-d

1

Abbreviation: CoA, coenzyme A.

A WNI|,B + A- B :;=

x 2

811

Chemistry: Corey et al.

812

Proc. Nati. Acad. Sci. USA 74 (1977)

facile oxidative addition would occur for higher valence states of cobalt or structures having a coordination number greater than four, despite the obvious capability of the enzyme to hold a substrate in close proximity to the metal. Granted that a substrate of -general type 4 could be converted by the oxidative addition-reductive elimination sequence to the complex 6, the question that arises next is whether an intermediate such as 6 is capable of undergoing the vicinal interchange rearrangement. Recent evidence suggests that such rearrangements are not only possible, but also relatively facile for the pathway indicated by 6 -- 15.

LCH2R3 LZ

L

6 H aR2 __ L Co.I L"' L'CH2R3% L 12

O1-kNH2

H N~J10JIITi2 >

,

%

R2

H

/ R3

lo

I

13

0

0 -H

H \

CH20H

R3

:4--

FIG. 1. Structure of coenzyme B-12.

H2 14

RICH2 being the 5'-deoxyadenosyl group. If L represents a

neutral two-electron ligand

LCo

CH2R, L

L

R2(CH2)2R3

Z

R3CH

R

CH2R3 -=± L >'o- ' CH\

L L, R6

+

R1CH3 7

6

(e.g., =N-), the cobalt in 3 and 6 would be uncharged and of formal valence state (I) in the d8 configuration. Both intermolecular (8 -. 9) (13) and intramolecular analogs (10 -_ 11) (14) of these oxidative additions are known. In the case of cobalt complexes

(CH3)A H

(CH3)2

(CH3)2

P

(CH)2+ H2 H2 CHC3) P P(CH3) p(CH3)2 I (CH3)2 (CH3)2P 9 8

CH2

(CH3)APX FeZP(CH3)3 (CH3)3P\ F jP(CH3)2 (CH3)3P

10

P(CH3)3

(CH3)3P

Li NCH2R,

L~co

+

N

16

5

P

WR2

2 CH3CH

LI--jH',

4

4

3

15 R,CH3

CH2R, +

*

NH P(CH3)3 11

the square planar, Co(I), d8 type structure is clearly the optimum arrangement for oxidative addition. It seems unlikely that

3

The rearrangement 6-.- 12 parallels the steps 17-. 18 (in which Cp is cyclopentadienyl (15) and (19 -- 20) (16), which provide the most reasonable basis for understanding the observed chemistry. Further, many complexes of methylenes (or "carbenes") with various transition metals are now known (16-18) and, in consequence, the accessibility of structure 12 as a reactive

CP2W+-CH3

+CP2W

17 a.,

18

CH2

: ± Cp2WX P

+

CH2PR3

[(CH3)3CCH2[C5Ta)30CH2]4TaO-IH ""'SCH-C(CH3)3 19 20

(CH3)3CCH3 + [(CH3)3CCH2]3Ta=CHC(CH3)3 intermediate appears assured. The vicinal carbon-to-carbon rearrangement 13-. 14 finds analogy in the transformation of the carbene complex 21'to the product 22 (19). Of course, 1,2-rearrangements of carbenes themselves (of type 23 -. 24) are among OCH2R (CO)5W C > (CO)5Wc |

I-,'CHR

21 R = H or n-C3H8 or CH-CH2

HC6H5

22

Proc. Natl. Acad. Sci. USA 74 (1977)

Chemistry: Corey et al. CH2R l

813

mCsH3 CH3

2

CH

26*

26

25 L

HOCH CH2OH

z L?..Co 1-'L2

H~ICos

H'_

2

-

-

HOCH2

CH20H CH-OH

HOCH ROH3 + L-o

01L

CH2. R 27

CH(O91)2 UN

RCH3 26*+ CH3CHO a~-

411'

28

HO\°s

L

H

____3HO"0I--,

L ILL

30

31

HO

HO I

H

-

L

29

25 RCH2: 5' -deoxyodeosyl FIG. 2. Proposed sequence for catalysis by coenzyme B-12 of the rearrangement of ethylene glycol to acetaldehyde. Electrocyclic cleavage of coenzyme B-12 to a 1,19-seco-Co(I)-corrin 26* of d8 electronic configuration; oxidative addition of substrate ethylene glycol to 26* forming 27; reductive elimination of 5'-deoxyadenosine from 27; hydrogen shift in 28, hydroxyl rearrangement in 29, and reductive elimination to 31; oxidative addition of 5'-deoxyadenosine to 31 and reductive elimination of acetaldehyde with regeneration of the catalytic species 26*.

the most facile and general reactions of the free species. Such a rearrangement within a transition metal-carbene complex can be expected to be especially favorable for weakly held carbene ligands (20): From the intermediate 15 all that is required to C-C

23 24

complete the scheme corresponding to a vicinal interchange is a sequence of reductive elimination and oxidative addition steps that provide the overall rearrangement product 16 and the original catalytic species 3.

Application to Coenzyme B-12. The relevance of the mechanistic scheme derived above for catalysis by a cobalt complex of the vicinal interchange reaction (3 + 4-several steps

3 +

16)

to

the

specific

case

of coenzyme B-12 is

not

immediately obvious. The coenzyme appears to contain too many ligands in coordination with cobalt and also to be in too

high an oxidation state [formally Co(III)]. The lower axial ligand (Fig. 1), 5,6-dimethylbenzimidazole, is known to dissociate from the metal very easily, but such a cleavage would still leave one too many ligands about cobalt. The recognition that one of the coordinating corrin nitrogens must also be lost raises questions with regard to the corrin ring system itself because the intact structure could not serve as a tridentate ligand. Still another problem connected with the maintenance of the intact corrin ring during the catalytic process arises from the fact that the planar nature of the four ligands of the corrin structure allows two ligands only at opposite axial positions where they are unlikely to interact with one another as required. Such considerations led to the mechanistic sequence illustrated by 25-31 (Fig. 2), wherein the coenzyme is depicted in an abbreviated way that omits many corrin substituents for the sake of convenience and clarity. The key step in this scheme is the very first (25 -* 26). By an electrocyclic process the bond between carbons 1 and 19, which directly joins the A and D rings, is cleaved as indicated. Nucleophilic attack by the acetamide group at C-2 (an excellent and well positioned neighboring group in this case) allows

814

Proc. Natl. Acad. Sci. USA 74 (1977)

Chemistry: Corey et al.

transfer of positive charge from cobalt to an iminium grouping with consequent reduction of formal valence from Co(III) to Co(I). Driving force for this transformation could be expected to come from the substrate-carrying enzyme and could take the form of (i) abstraction of the benzimidazole ligand, thereby destabilizing Co(III); (ii) electrostatic interactions that favor the change in charge distribution; and (iii) hydrogen bonding and other interactions that produce a torque on the corrin ring favoring rotation of ring A out of the corrin plane and assumption of the geometry shown in 26*. This structure includes a square planar arrangement of four ligands about Co(I), as required for the ideal model (3) and substantial rotation about the bond between carbons 4 and 5. It should be noted that the planar coordination of all four nitrogens of rings A, B, C, and D to cobalt in 26 would also be disfavored by the steric repulsion between proximate substituents on rings A and D that would result. Modifications of 26 other than to 26* are also possible, and are not now excluded; for example, a structure in which the nitrogen of ring A remains coordinated to cobalt and that of ring D moves away by rotation about the bond between carbons 14 and 15. In fact a variety of alternatives to 26 (such as 32) are also possible and not unreasonable. The common denominator of all these alternatives H2N+

H NCO

A

H3C

3 0 CH-

H B ...CH3

CH3

32 would be the electrocyclic cleavage of the corrin ring system at the C(1)-C(19) bond and the transfer of charge with conversion of Co(III) to Co(I). One satisfying feature of the C(1) -C(19) cleavage tactic is the simplicity with which nature's construction of the corrin system and the unique direct linkage of rings A and D can be viewed; such direct linkage is both nonaccidental and essential. The remainder of the process of catalysis for the conversion of ethylene glycol to acetaldehyde as illustrated by formulas 27 to 31 closely follows the general formulation of the preceding section. The rearrangement 29 -- 30 can be regarded as a migration of hydroxyl with the electron pair of the C-OH bond to an electrophilic adjacent carbon and simultaneous bonding of cobalt (through its vacant coordination site) to the methylene carbon in a backside displacement of OH. The carbon ligand in alkoxycarbene-transition metal complexes is known to be electrophilic (17, 18, 21). Reductive ("alpha") elimination of 30 leads to the gem-diol complex 31, with Co(I) formal valence and d8 electronic configuration. Oxidative addition of 5'deoxyadenosine to 31 followed by a-elimination would form acetaldehyde (or its hydrate) with regeneration of 26*, which could continue the cycle or revert to the corrin system of the B-12 coenzyme. The mechanism outlined in Fig. 2 accords with all the experimental data for the enzymic reaction that converts 1,2-diols to aldehydes (items 1-9 in the introduction) with just one proviso. Since the isotope effect for the transfer of hydrogen from 5'-tritiated 5'-deoxyadenosyl ligand to the aldehyde is known to be large, there would have to be a correspondingly

large isotope effect for the oxidative addition of 5'-deoxyadenosine (enzyme bound) to 31, or competition between ethylene glycol and RCH3 for 31. The data on the kinetic isotope effect using deuterated 1,2-propanediol would seem to imply that step 26* --27 or 31 -- 26* (or both) would be rate limiting for the overall process. Certain aspects of the new mechanism proposed herein are subject to experimental test. The question of whether the corrin system might be accessible synthetically by a process such as 26 -- 25 is also of considerable interest. The extension of this mechanism to the carboxylic rearrangements (e.g., methylaspartate - glutamate) is straightforward, but one especially interesting point emerges. The cobalt-complexed carbene which would participate in the carboxyl or COSCoA migration (analogous to 29-. 30) is expected to be nucleophilic (22) and so it can be anticipated that the replacement of carboxyl by cobalt and then later by hydrogen would occur with retention of configuration. In fact, such a stereochemical course has been demonstrated experimentally (23). Finally, some comment is required on the previously proposed "free radical" mechanism (3), which depends on an initial homolytic cleavage of the bond between cobalt and the 5'deoxyadenosyl ligand and subsequent hydrogen atom transfer from substrate to the 5'-deoxyadenosyl radical. The following points would seem to support a skeptical position. (i) It is not clear that the extremely reactive 5'-deoxyadenosyl radical could be prevented from attacking nearby groups (including its own adenine ring) indiscriminately, or constrained so strongly in space so as to attack only the pro-S hydrogen of (2R)-1,2-propanediol or the pro-R hydrogen of (2S)-1,2-propanediol. (ii) The free radical mechanism provides no clarification or rational description of the vicinal interchange itself. (iii) It is far from clear that the radical mechanism is consistent with the various isotope effects that have been observed. (iv) There is no convincing direct evidence for the intermediacy of carbon radicals. This work was assisted financially by a grant from the United States National Science Foundation. 1. Lenhart, P. G. & Hodgkin, D. C. (1961) Nature 192,937-938. 2. Barker, H. A. (1972) Annu. Rev. Biochem. 41, 55-90. 3. Abeles, R. H. & Dolphin, D. (1976) Acc. Chem. Res. 9, 114120. 4. Brown, D. G. (1973) Prog. Inorg. Chem. 18, 177-286. 5. Stadtman, T. C. (1971) Science 171, 859-867. 6. Schrauzer, G. N. & Sibert, J. W. (1970) J. Am. Chem. Soc. 92, 1022-1030. 7. Whitlock, H. W., Jr. (1963) J. Am. Chem. Soc. 85,2343-2344. 8. Dowd, P., Shapiro, M. & Kang, K. (1975) J. Am. Chem. Soc. 97, 4754-4757. 9. Breslow, R. & Khenna, P. L. (1976) J. Am. Chem. Soc. 98, 1297-1299, 6765. 10. Collman, J. P. & Roper, W. R. (1968) Adv. Organometal. Chem.

7,54-94. 11. 12. 13. 14.

Collman, J. P. (1968) Acc. Chem. Res. 1, 136-143. Parshall, G. W. (1975) Acc. Chem. Res. 8, 113-117. Chatt, J. & Butter, S. A. (1967) Chem. Commun. 501-502. Rathke, J. W. & Mutterties, E. L. (1975) J. Am. Chem. Soc. 97,

3272-3273. 15. Cooper, N. J. & Green, M. L. H. (1974) Chem. Commun. 761-762. 16. Schrock, R. R. (1974) J. Am. Chem. Soc. 96,6796-6797. 17. Fischer, E. 0. (1972) Pure Appl. Chem. 30,353-372. 18. Fischer, E. 0. (1974) Angew. Chem. 86,651-682 (Nobel Lec-

ture).

Proc. Nati. Acad. Sci. USA 74 (1977)

Chemistry: Corey et al. 19. Fischer, E. 0. & Held, W. (1976) J. Organometal. Chem. 112, C59-C62. 20. Labinger, J. A. & Schwarz, J. (1975) J. Am. Chem. Soc. 97, 1596-1597.

815

21. Cardin, D. J., Cetinkaya, B., Doyle, M. J. & Lappert, M. F. (1973) I

C mhem. Soc. Rev. 2, 99-144.

22. Schrock, R. R. (1975) J. Am. Chem. Soc. 97,6577-6578. 23. Retey, J. & Zagalak, B. (1973) Angew. Chem. 85, 721-722.

Biochemical catalysis involving coenzyme B-12: a rational stepwise mechanistic interpretation of vicinal interchange rearrangements.

Proc. Natl. Acad. Sci. USA Vol. 74, No. 3, pp. 811-815, March 1977 Chemistry Biochemical catalysis involving coenzyme B-12: A rational stepwise mec...
700KB Sizes 0 Downloads 0 Views