Jan 25, 1974 - As well, Koenig and Brown (22) have questioned the proposal, being unable to ..... Anfinsen, C. B., Jr., Anson, M. L., Edsall, J. T. &. Richards ...
Proc. Nat. Acad. Sci. USA Vol. 71, No. 5, pp. 1686-1690, May 1974
The Activity-Related Ionization in Carbonic Anhydrase (metalloenzymes/enzyme mechanism/imidazole chelates)
DAVID W. APPLETON AND BIBUDHENDRA SARKAR The Division of Biochemistry, The Research Institute, The Hospital for Sick Children, Toronto; and the Department of Biochemistry, University of Toronto, Toronto, Ontario, Canada
Communicated by Jacob Bigeleisen, January 25, 1974 The catalytic activity of carbonic anhyABSTRACT drase (EC 22.214.171.124) is linked to the ionization of a group in close proximity to the essential zinc ion. Studies have been undertaken to delineate the ionizations germane to the active-site chelate system. Several imidazole ligand systems were studied in order to approach a representative chelate. The simplest involved the complexation of Zn(II) by imidazole and by N-methylimidazole. As well, two bidentate systems, Zn(II)-4,4'-bis-imidazoylmethane and Co(II)-cyclic-L-histidyl-L-histidine were investigated. It was found that in a species containing metal-bound water and imidazole coordinated by means of the pyridinium nitrogen, the most acidic group was the pyrrole N-H in the imidazole ring. By the use of N-methylimidazole, the pKa of a metal-bound water molecule in a triimidazole ligand field was found to be 9.1. Noting the preference for labilization of the pyrrole hydrogen, the catalytic features of carbonic anhydrase are reexamined assuming that the pK... is associated with the N-H ionization, and not with the ionization of metal-bound water.
In 1964, Kernohan established a pH dependence for bovine carbonic anhydrase (carbonate hydro-lyase; EC 126.96.36.199) characterized by a pKenz of 6.9 (1). Subsequently, kinetic investigations were extended to include the human B and C isoenzymes (2, 3). The high-activity human C enzyme was found to have a pH dependence similar to that described for the bovine enzyme. Further work has revealed that the human B enzyme has a lower turnover number coupled with a more alkaline pH dependence, having a pKenz value near 8.2 (4, 5). Although the pH dependence of the intrinsic activity has been known for quite some time, the nature of the ionizing group responsible for the sigmoidal behavior has been, and still is, the seat of much discussion. It has been well established that the ionizing group must be intimately associated with the essential metal ion cofactor (6). Changes in the covalent domain of the metal ion can be selectively monitored if the native zinc is replaced by cobalt (7). Visible spectral (8, 9) and related analyses (10, 11) of the substituted enzyme reveal a close parallelism between the development of intrinsic activity and profound changes in the ligand system. The interaction of 83C1 with the bovine enzyme indicates that enzymic inhibition is concurrent with the binding of one ion per enzyme molecule, directly to the metal ion. This anionic inhibition is intimately associated with the pH dependence,
as a plot of the apparent pKenz against [Cl-] yields the expected pKenz when extrapolated to zero chloride concentration (12). Assuming that the ionization is metal ion linked, there exist two possible interpretations. Either the dissociation originates in the preexisting ligand system, or the ligands are exchanged, inducing the ionization of the incoming ligand. Of the two, the latter choice appears less attractive, for the following reason. It has been found from x-ray crystallographic studies on the human C isoenzyme, crystallized at pH 8.5, that the zinc ion is tetrahedrally coordinated to three imidazole rings and most likely a solvent molecule (13). From magnetic moment (14), spectral (15), and magnetic circular dichroism results (16), this situation is thought to be maintained in solution. Taken most simply then, the x-ray study defines the ligand field of the active enzyme. Presumably, a ligand exchange leading to the active enzyme would already have occurred. But, Coleman (4), by cyanide titration, has found the ionizing group to have an intrinsic pKa > 11, a pKa difficult to attribute to the ionization of a protonated imidazole side chain. Thus, in light of the arguments that would be needed to rationalize a ligand substitution process, the alternative of an ionization in the preexisting metal ion-ligand system seems more straightforward. In the alternative hypothesis, the dissociable hydrogens are the histidyl pyrrole hydrogens* and the metal-bound water molecule protons. The pyrrole hydrogens have as yet not been discussed in connection with the ionization, presumably because of their intrinsic pK.; the pKa of neutral imidazole is 14.5 (17). The known versatility of coordinated hydroxide (18, 19) coupled with its compatibility with most of the available criteria has led many workers (6, 20, 21) to accept the water dissociation hypothesis. Unfortunately, due to the problems inherent in the experimental design, little direct evidence is available to substantiate or refute this hypothesis. As well, Koenig and Brown (22) have questioned the proposal, being unable to reconcile either the nuclear magnetic relaxation dispersion data or Ward's 35CI relaxation data with the ionization of a metal-bound water molecule. In an attempt to more fully correlate the observed parameters pertinent to this ionization, we have begun an investi-
Abbreviations: Complexes of the type MpHqAr are defined as follows: p, q, r are the respective stoichiometries of the metal ion M, the hydrogen ion H, and the uncharged ligand A. The subscript, -q, indicates the uptake by the complex of q additional equivalents of base, due either to the titration of pyrrole or complexed water protons.
* Imidazoles are amphoteric compounds. Due to the basicity of the "pyridine-type" nitrogen, imidazoles readily form salts with acids and complexes with most metal ions. Imidazoles with the "pyrrole-type" N-H intact are weak acids with pKa values on the order of 14.
Proc. Nat. Acad. Sci. USA 71
gation into the dissociations associated with mixed complexes of the type metal ion [Zn(II) or Co(II)I, imidazole, water. In doing so, we have hopefully provided a case for an evaluation of the possibility of pyrrole hydrogen ionization, as well as some evidence about the possible pH dependencies for the two relevant ionizations.
Activity-Related Ionization in Carbonic Anhydrase
_ 02 c
MATERIALS AND METHOD
Analytical grade imidazole was twice recrystallized from benzene. The N-methylimidazole obtained from the Aldrich Chemical Co. was distilled under reduced pressure and stored under dry nitrogen. The cycdic-ihistidyl-ihistidine was prepared according to the method of Fischer and Suzuki (23), and twice recrystallized under argon, from hot water. The 4,4'-bis-imidazoylmethane was isolated as the dinitrate salt by a synthesis modified from Drey and Fruton (24). Ligand stock solution concentrations were checked repeatedly by titration with standardized KOH. Stock solutions of zinc nitrate were prepared by dissolving analytical grade zinc oxide in redistilled nitric acid. The cobalt nitrate solutions were prepared from analytical grade Co(NO3)2-6H20. The metal stock solutions were standardized by EDTA titration with xylenol orange as indicator. All titrations were carried out at 250 4 0.10 under washed argon with 0.16 M KNO3 as background electrolyte. The electrode linearity was calibrated with four National Bureau of Standards primary pH standard solutions, pH values 1.650, 4.008, 7.413, and 9.180 (25). All systems excepting the Zn(II)-N-methylimidazole were studied by continuous titration, with the equipment and method described in ref. 26. To avoid the problem of metal hydroxide precipitation, the Zn(II)-N-methylimidazole system was studied by discontinuous titration, with a medial ligand metal ratio of 258/1. At each of 30 pH values between 3.898 and 11.005, a series of five separate potentiostatic titrations was performed by the sequential addition of 0.1 through 0.5 ml of a 60.00 mM Zn(II) stock solution to 50.0 ml of an 77.37 mM solution of the ligand. The titration results were analyzed with the aid of our programs, PLOT, GUESS, and LEASK (27), on a GE-440 computer. Infrared spectra were recorded on a Beckman IR-20A. Visible spectra were recorded at 250 on a Cary 15 recording spectrophotometer. The titration apparatus was connected to a sealed cuvette, of path length 5.0 cm, by a syringe-driven flow system. RESULTS
Titration, with aqueous base of a solution 1.0 mM with respect to Zn(II) and 10.0 mM with respect to imidazole, causes the precipitation, beginning about pH 7.5, of a flocculent white solid. X-ray diffraction studies have shown that the precipitate, zinc imidazolate, is an infinite polymer in which each metal ion is tetrahedrally coordinated to four ligand molecules, with each imidazole acting as a bidentate link between two metal ions (28). The loss of the pyrrole hydrogen has been verified by infrared studies that demonstrated the absence of N-H stretching bands in the solid (29). We have completely substantiated the above findings by stoichiometric analysis and by infrared studies on precipitate isolated below pH 8 from both H20 and D20 solutions. Titration studies of the zinc imidazole system have, to date, not taken into account the pyrrole ionization (30), although
FIG. 1. The partial species distribution for a solution of 1.15 mM Zn(II), 19.64 mM imidazole, and 0.16 M KNO, at 250. The species shown are: (1) MA; (2) MA2; (3) MH-1A2; (4) MH-2A2. Values on the ordinate have been multiplied by 10-2.
the presence of the precipitate is proof that the species (MH-2A2)n is forming. This factor would then introduce a heavy bias into ni calculations of the initial stepwise constants. As a consequence, studies were undertaken to reevaluate the system, both to better understand the stepwise complexation and to describe the pH dependence of the pyrrole deprotonation. In short, complete analysis proved. impossible. The amount of any species present in the system is obtained by a least-squares fit (Program LEASK) of the species curve shapes (Program GUESS) to the calculated bound metal function. However, it was found that five stepwise species, plus the pyrrole species, occur simultaneously within one pH unit. Given normal fitting errors and the lack of data beyond pH 7.5, there was no unique solution. In Fig. 1 are shown representative results. In curve 1, the species MA is rigorously fitted and log #lo, = 2.48 agrees well with published values (30). The species MA2, curve 2, is inordinately broad. Its proportion depends heavily on the amount of MA3 present in the system. But MA3 and MA4 cannot be calculated separately because the curve shapes are prematurely truncated due to the precipitation problem. The lower limit of MA2 has been included for interest, and the composite MA3 and MA4 has been omitted. Since none of the pyrrole species peak before precipitation, rigorous fittings in the context of this method is negated. However, information on the pH dependence can be obtained since the "pyrrole deprotonation envelope" can be calculated. We have fitted the envelope, assuming two species MH-1A2 and MH-2A2. Obviously the second species will not exist to any great extent, in nonpolynuclear form. However, all polymers of either species will have maxima at lower respective pH values, such that the fitted curves represent the most alkaline pH dependencies. In an effort to extend the analysis of the pyrrole ionization to a broader pH range, two bidentate ligands were subsequently studied. Drey and Fruton (31), reported that 4,4'bis-imidazoylmethane readily forms a bis complex, MA2, with zinc. At the time, they reported that further ionization of an unspecified nature did occur. Reanalysis of this system showed that the species MA2 declined with increasing pH concomitant with the dissociation of pyrrole hydrogens and subsequent precipitation. The first ionization fits as MH-1A2 and again because of precipitation, species leading from MH_-A2 to (MH-2A2). were fitted under MH-2A2. The pKa
Biochemistry: Appleton and Sarkar
Proc. Nat. Acad. Sci. USA 71
TABLE 1. Summary of the ionization constants METALION-ON ~ Ka SPECIES ION__ LIGAND (A) _IZATI
FIG. 2. The species distribution for a solution of 0.299 mM Zn(H), 77.0 mM N-methylimidazole, and 0.16 M KNOs at 250. The species shown are: (I) unbound metal; (2) MA; (8) MA2; (4) MA3; (5) MAX; (6) MA.5; (7) MA6; (8) MH_ IA4; (9) MH-rA4.
Of the pyrrole ionization giving rise to MH-jA2 is 7.90. This is to be compared with a value of 7.0-7.3 for MH-,A2 in the
Zn(HI)-imidazole system (Table 1). It has been reported that in cyclic-L-histidyl-L-histidine, the two imidazole side chains share, in close proximity, the space over the diketopiperazine ring (32). To exploit the obvious advantages such a bidentate ligand may possess, an analysis of the interaction with Co(HI) was carried out. Cyclic-L-histidyl-L-histidine forms a single complex with Co(II), MA (log #lo, = 2.68). Above pH 7, MA begins dissociating pyrrole hydrogens. The dissociation of a second pyrrole hydrogen from MA leads to polynucleation although no precipitate is seen until above pH 9. The species MH-2A presumably can remain in solution longer, due to the steric inhibition of precipitation posed by the bulky diketopiperazine ring. The pK. for the first pyrrole ionization is 8.50. Spectral fitting (33) using spectra recorded below pH 8.5 was used to calculate the individual species spectra; MA is octahedral, L'Max 19; MH-,A has an em.. = 120; while MH-2A has an em.. = 277. The species having dissociated pyrrole hydrogens are extremely sensitive to oxidation of the `
bound cobalt. To investigate the possible hydrolysis of zinc-bound water in an imidazole ligand field, a study was carried out on the Zn(II)-N-methylimidazole system. Although the pyrrole hydrogen has been replaced-by a methyl group, the basicity of the pyridinium nitrogen is not greatly changed. The pK. of imidazole at 25° in 0.16 M KNO3 is 7.125 4- 0.014, while of the N-methylimidazole is 7.209 0.008. 1 the pK4 Therefore, N-methylimidazole should, be representative of imidazole but will be freespeie trb complications associated with the
pyrrole hydrogen. Analysis of the Zn(hi)-N-methylimidazole system provided an unexpected wealth of information (Fig. 2). The stepwise species MA through MA4 all maximize within a range of 1.2 pH units. Then follows a break of 1.2 pH units before the apvalueo by a scant 0.24 pH units. pearance of MA5 anedwih, Accompanying this unusual pH distribution are "roller coaster" stability constants (Table 2). The log K values follow a repeating pattern of low, high through all six stepwise species, MA4 showing the highest stability. The uniqueness of this
pattern at first caused some concern. However, the further addition of data points did not alter the low, high pattern. As well, the pattern is maintained both in regions of lower and extremely high experimental accuracy, and is maintained in MA5 and MA6, which are clearly separated from the preceding species. This finding is not without precedent, since several zinc systems have been analyzed that do not follow the normal pattern, log K1 > log K2> log K3 > log K4 (34-36). Returning to the question of hydrolysis, it is clearly evident that this system extensively, hydrolyzes metal-bound water at higher pH. The progenitor is MA4, which loses a zincbound water proton with a pKa of 9.12. At higher pH a second bound hydroxide is formed. It should be stressed that there is no visual, pH drift, or curve-fitting evidence for polymerization or for metal hydroxide formation in this system, as far as it has been studied (pH 11.5). It is of interest to add that the curve shape for the species MH-jA3 has a pH dependence very similar to that described for MH- 1A4. DISCUSSION
The intrinsic pK. values of water and of neutral imidazole fall within an order of magnitude. Enhancement of the respective proton labilities will be expected if the water molecule or the imidazole pyridinium nitrogen is bound to the zinc. The question then is one of the relative transmittability of the metal ion influence to the two ionizing centers. In the systems studied, it would appear that the transmittability is accentuated in the case of the pyrrole ionization. That is, given a system in which both ionizations can occur, the ionization of a pyrrole hydrogen will take precedence over the ionization of metal bound water. This is clearly evident since the ionization of the bound water can only be studied when the pyrrole hydrogen has been replaced by a methyl group. The degree to which the pyrrole ionization is favored varies, depending on the nature of the complex. The most acidic pyrrole hydrogen is found in the Zn(II)-imidazole system (Table 1). Since the species maximum of MH-jA2 occurs following the onset of precipitation, the pKa can be obtained only after extrapolation of the curve shape to a maximum. Consequently, the pK. is approximated as lying between 7.0 and 7.3. The other two chelates studied in the context of pyrrole ionization both have progressively less acidic hydrogens, giving rise to a variation in pKa of about 7 through 8.5.
Proc. Nat. Acad. Sci. USA 71
Activity-Related Ionization in Carbonic Anhydrase
TABLE 2. Stability constants for the Zn(HI)-N-methylimidazole 8y8tem* p
1 1 1 1 1 1 1
0 0 0 0 0 -1 --2
2 3 4 5 6 4 4
4.924 6.600 9.214 10.005 11.047 0.157 -10.615
0 K1 = 2.380 K2 = 2.544
Ks= 1.676 K4 = -2. 614 K5=0.791
* Log stability constants (log .pqr) of complex species MpHqAr [M = Zn(H), A = N-methylimidazole] in 0.16 M KNOs at 250. The log K. values are given as well, where log K. = log 61On log j#1O(n-1).
Since ionization of the pyrrole hydrogen more or less precludes the ionization of bound water in the same system, it was necessary to study N-methylimidazole to characterize the ability of an imidazole chelate system to labilize the hydrogens on a bound water molecule. The pK. for the ionization of a water molecule in a zinc complex containing four bound imidazoles was found to be 9.12 (Table 1). This value is sufficiently higher than, say, the value of 7.90 obtained with the Zn(II)-4,4'-bis-imidazoylmethane to explain the presence of only the pyrrole ionization in systems which can ionize either the N-H or the water. It could be argued that a reduction of the ligand field to three imidazoles may labilize still further the water hydrolysis. To rule this out, the species matrix was solved including MH-1A3. It was found that MH-1A3 and MH-1A4 have almost identical pK. values, suggesting that the ionization of metal-bound water is not overly sensitive to changes in the neutral ligand field. Other features of the Zn(II)-N-methylimidazole system may also prove to be pertinent to a comprehensive understanding of these ionizations. From the species distribution, considering the high stability of MA4, one is led to believe that most of MA4 is tetrahedral, with a small amount aquated and octahedral. This small amount of octahedral MA4 would then be the template for the formation of the four higher pH species. Possibly, the relatively higher stability of MA2 reflects the initial change from octahedral to tetrahedral zinc. In any case, the tetrahedral-like environment presented by the enzyme does coincide with the preferred geometry of zinc, at least in an imidazole ligand field. In this conjunction it is perhaps worthwhile to point out that the enzyme, in the protonated state, may have one or more of the pyrrole hydrogens hydrogen-bonded to adjacent acceptor groups on the protein backbone (13). Subsequent deprotonation would alleviate this point of attachment, allowing the chelate to move towards a more preferred conformation. This would adequately explain the magnitude of the spectral change noted on deprotonation of the cobalt enzyme. Considering that the hydration of CO2 involves the addition of no less than a hydroxyl ion to the substrate, the formation of a metal-bound hydroxyl ion represents the most compelling feature of the ionizing water hypothesis. A mechanism dependent on the ionization of a pyrrole hydrogen must develop the nucleophile indirectly. Such an indirect formation could occur if the imidazolate anion accepted a water proton
FIG. 3. A mechanism for the catalysis of C02 hydration by a zinc-bound anionic imidazole.
locally forming an OH- ion. In Fig. 3 is shown a mechanism that illustrates how such OH- ion could be used. This mechanism pertains to the human B enzyme, since Khalifah (3) has shown that imidazole is a competitive inhibitor of the B enzyme, binding directly to the zinc ion (40). This inhibitor is unique, since it does not flood the system with a negative charge. This would allow the imidazole anion to exist in the absence of metal bound water. If the C02 hydrogen-bonded to the water, the pyrrole ionization may be perturbed slightly, raising the pK,, causing the nitrogen to bind a proton and form the nucleophile. Recently, Koeftig et al. (41) have presented further evidence supporting the formation of H2CO3 as product. In this context, the above mechanism is more easily rationalized since the metal-bound water would then completely donate the proton, temporarily "inhibiting" the pyrrole ionization. Although such a mechanism is less straightforward, it does have intriguing possibilities. Unless one anionic pyrrole nitrogen is specifically stabilized, the net dissociation may extend over the three rings. This "tautomeric set" could facilitate the use of much of the upper chelate surface for catalysis. To carry the model studies to a logical conclusion, we are presently synthesizing a triimidazole chelate. This will hopefully enable us to more fully characterize the tendencies inherent in non enzymatic systems. Perhaps this work, in conjunction with further purposeful work on the enzymet, will establish the true nature of the ionizing group. This work was supported by search Council of Canada.
grant from The Medical Re-
t Proton magnetic resonance studies on the human B enzyme have shown resonances with chemical shifts characteristic for histidine C2 protons, which were classed as nontitrating or anomalous (37, 38). Studies in this laboratory on Zn(II)-4,4'bis-imidazolylmethane solutions indicate that the binding of Zn(II) shifts the free base C2 proton resonances downfield by more than 0.6 ppm. As well, Pugmire and Grant (39) have shown that the change in chemical shift of the C2 proton accompanying the deprotonation of the free base is only 16% the magnitude of that seen for the cationic imidazole deprotonation. Therefore, the downfield resonances in the enzyme spectra must be considered suspect, and warrant a close reexamination as to subtle titration behavior.
Biochemistry: Appleton and Sarkar
1. Kernohan, J. C. (1964) Biochim. Biophys. Acta 81, 346356. 2. Gibbons, B. H. & Edsall, J. T. (1964) J. Biol. Chem. 239, 2539-2544. 3. Khalifah, R. G. (1971) J. Biol. Chem. 246, 2561-2573. 4. Coleman, J. E. (1967) J. Biol. Chem. 242, 5212-5219. 5. Fabry, M. E., Koenig, S. H. & Schillinger, W. E. (1970) J. Biol. Chem. 245, 4256-4262. 6. Lindskog, S., Henderson, L. E., Kannan, K. K., Liljas, A., Nyman, P. 0. & Strandberg, B. (1971) in The Enzymes, ed. Boyer, P. D. (Academic Press, New York), Vol. 5, pp. 587-665. 7. Lindskog, S. (1970) Struct. Bonding (Berlin), 8, 153-196. 8. Lindskog, S. (1963) J. Biol. Chem. 238,945-951. 9. Lindskog, S. & Nyman, P. 0. (1964) Biochim. Biophys. Acta 85, 462-474. 10. Coleman, J. E. (1965) Biochemistry 4, 2644-2655. 11. Lindskog, S. (1966) Biochim. Biophys. Acta 122, 534537. 12. Ward, R. L. (1969) Biochemistry 8, 1879-1883. 13. Liljas, A., Kannan, K. K., Bergst6n, P.-C., Waara, I., Fridborg, K., Strandberg, B., Carlbom, U., JRrup, L., Lovgren, S. & Petef, M. (1972) Nature New Biol. 235, 131137. 14. Lindskog, S. & Ehrenberg, A. (1967) J. Mol. Biol. 24, 133137. 15. Rosenberg, R. C., Root, C. A., Wang, R., Cerdonio, M. & Gray, H. B. (1973) Proc. Nat. Acad. Sci. USA 70, 161-163. 16. Coleman, J. E. & Coleman, R. V. (1972) J. Biol. Chem. 247, 4718-4728. 17. Walba, H. & Isensee, R. W. (1956) J. Chem. Soc. 21, 702704. 18. Chaffee, E., Dasgupta, T. P. & Harris, G. M. (1973) J. Amer. Chem. Soc. 95, 4169-4173. 19. Breslow, E. (1966) in The Biochemistry of Copper, eds. Peisach, J., Aisen, P. & Blumberg, W. E. (Academic Press, New York), pp. 149-157. 20. Prince, R. H. & Woolley, P. R. (1972) Angew. Chem. Int. Ed. Engl. 11, 408-417. 21. Lindskog, S. & Coleman, J. E. (1973) Proc. Nat. Acad. Sci. USA 70, 2505-2508.
Proc. Nat. Acad. Sci. USA 71
22. Koenig, S. H. & Brown, R. D., III (1972) Proc. Nat. Acad. Sci. USA 69, 2422-2425. 23. Fischer, E. & Suzuki, U. (1905) Chem. Ber. 38, 4173-4196. 24. Drey, C. N. C. & Fruton, J. S. (1965) Biochemistry 4, 1-5. 25. Bates, R. G. (1964) in Determination of pH, Theory and Practice (John Wiley and Sons, Inc., New York), pp. 123130. 26. Kruck, T. P. A. & Sarkar, B. (1973) Can. J. Chem. 51, 35493554. 27. Sarkar, B. & Kruck, T. P. A. (1973) Can. J. Chem. 51, 35413548. 28. Freeman, H. C. (1967) in Advances in Protein Chemistry, eds. Anfinsen, C. B., Jr., Anson, M. L., Edsall, J. T. & Richards, F. M. (Academic Press, New York), Vol. 22, pp. 290-294. 29. Cordes de N. D., M. & Walter, J. L. (1968) Spectrochim. Acta Part A 24, 237-252. 30. Martel, A. E. & Sill6n, L. G. (1964) in Stability Constants of Metal-Ion Complexes (special publication No. 17, the Chemical Society, London). 31. Drey, C. N. C. & Fruton, J. S. (1965) Biochemistry 4, 12581263. 32. Ziauddin, Kopple, K. D. & Bush, C. A. (1972) Tetrahedron Lett. 483-486. 33. Kruck, T. P. A. & Sarkar, B. (1973) Can. J. Chem. 51, 3563-3571. 34. Mironov, V. E., Kul'ba, F. Ya., Rutkovskii, Yu. I. & Ignatenko, E. I. (1966) Zh. Neorg. Khim. 11, 955-958. 35. Hershenson, H. M., Brooks, R. H. & Murphy, M. E. (1957) J. Amer. Chem Soc. 79, 2046-2048. 36. Nyman, C. J. (1953) J. Amer. Chem. Soc. 75, 3575-3576. 37. King, R. W. & Roberts, G. C. K. (1971) Biochemistry 10, 558-565. 38. Cohen, J. S., Yim, C. T., Kandel, M., Gornall, A. G., Kandel, S. I. & Freedman, M. H. (1972) Biochemistry 11, 327-334. 39. Pugmire, R. J. & Grant, D. M. (1968) J. Amer. Chem. Soc. 90, 4232-4238. 40. See footnote 3, in ref. 3, p. 2566. 41. Koenig, S. H., Brown, R. D., Needham, T. E. & Matwiyoff, N. A. (1973) Biochem. Biophys. Res. Commun. 53, 624630.