STUDIES OF STRUCTURE AND MECHANISM
Irwin Fridovich Department of Biochemistry, Duke University Medical Center, Durham, North Carolina 27710 INTRODUCTION Superoxide dismutases are enzymes of defense which serve to protect respiring cells against a product of their own respiration. This is achieved by a catalytic scavenging of the superoxide radical which, surprisingly, is a rather commonplace intermediate of the reduction of oxygen. The reaction catalyzed is:
+ 0 - + 2H+
Enhancing the rate of this particular reaction may appear to be a peculiar undertaking for an enzyme, since this reaction is reasonably rapid, even in the absence of catalysis and since several transition metal cations serve as effective catalysts. Nevertheless superoxide dismutases are indispensable because they operate with ultimate efficiency and are present in abundance, whereas free transition metal cations are less effective and are not plentiful inside cells. It is a fact that all of the superoxide dismutase activity, which can be measured in crude extracts of such diverse materials as erythrocytes, mammalian liver or Escherichia coli, can be accounted for in terms of the superoxide dismutases which they contain. There is thus no significant superoxide dismutase activity in these extracts save that due to these enzymes. The evidence which establishes the biological importance of superoxide dismutases has been reviewed (1-9). We shall therefore now eschew discussion of such matters and shall concentrate on structure and mechanism. Living forms contain several distinct superoxide dismutases. These fall naturally into two families. Within each family the 530
K. T. Yasunodu et al. (eds.), Iron and Copper Proteins © Plenum Press, New York 1976
STRUCTURE AND MECHANISM
enzymes are very similar, whereas comparisons between families show no likeness save that of a common catalytic effect. These two families of superoxide dismutases probably evolved independently in response to the common selection pressUre imposed by the oxygenation of the biosphere. One of these groups comprises enzymes containing copper and zinc whereas the members of the other family contain iron or manganese. COPPER AND ZINC-CONTAINING SUPEROXIDE DISMUTASES The cytosols of eukaryotic cells contain a superoxide dismutase whose molecular weight is 32,000. It is a homodimer and contains one eu++ and one Zn++ per subunit. The properties of these enzymes has been remarkably resistant to evolutionary change. Thus the enzymes isolated from yeast, Neurospora crassa, spinach, chicken or cow (10-15) are hardly distinguishable and show only minor differences in amino acid composition. in the super-hyperfine detall$ of their esr spectra (16) and in their optical spectra in the ultraviolet (17). One indication of the similarities among these enzymes is their ability to be isolated from diverse sources by application of essentially the same purification procedure (10, 12,14,15). These enzymes were isolated and were characterized, as proteins, well before their catalytic activity was discovered. Thus the first to be described, in 1938, was a "hemocuprein" from ox blood (18). It and similar "cupreins" were isolated on the basis of copper content and were thought to function in copper storage or transport. The relatively recent realization (15) that these cupreins were enzymes of vital importance to respiring cells caused a resurgence of interest, which has culminated in a thorough knowledge of the structure of one of them. Thus the complete amino acid sequence of the superoxide dismutase from bovine erythrocytes has been reported (19-21) and structural analysis by X-ray diffraction has progressed to 3i resolution (22,23). The Richardsons are at this moment refining this structural analysis to 2i resolution (24). The most prominent structural feature of the subunit of this enzyme is a cylinder whose wall is made up of eight strands of the peptide chain arranged in an antiparallel S structure. The segments of the sequence which constitute the slats of this S barrel are 2-11, 13-23, 26-35, 38-47, 80-88, 91-100, 112-118 and 142-149. Two non-helical coils, involving residues 48-79 and 119-141, protrude from one side of the S barrel and together enclose and constitute the active site. Each subunit is stabilized by an intrachain disulfide bond between cys-55 and cys-144. There is a free thiol residue on each subunit, that of cys-6, which is remarkably unreactive in the native molecule (19). The enzyme is unusually stable and retains activity in the presence of 9.0 M urea or 4% sodium dodecyl sulfate (36). Removal of the metal prosthetic groups or reduction of the intrachain disulfide bond diminishes the
resistance of the enzyme towards denaturing stresses (19,37). The subunits are joined by non-covalent interactions and the large area of contact helps explain the strength of this association. The Cu++ and Zn++, at the active sites, are in close proximity, as was predicted from physico-chemical studies (25-28). Indeed the Cu++ and Zn++ are joined by a common ligand, which is the imidazolate ring of histidine 61. The Cu++ is relatively exposed to the solvent whereas the zinc is more buried within the structure. The Cu++ appears to be in contact with a freely exchangeable molecule of water and this contact point is probably the site of direct interaction with the substrate, 02- (29). In addition to the bridgin~ imidazolate of histidine 61, the other groups liganded to the Cu+ are the imidazole rings of histidines 44, 46, and 118; while the other ligands of the Zn++ are the imidazole rings of histidines 78 and 69 and the carboxyl group of aspartate 81 (22). Analysis of the epr spectrum of the enzyme and studies with group specific reagents had indicated that nitrogen atoms of imidazole rings formed the ligand field of the Cu++, before the X-ray structure was in hand (25,30-33). The eu++ and the Zn++ can be reversibly removed with concomitant loss and regain of catalytic acti vi ty. Indeed acti vi ty can be largely restored to the apoenzyme by Cu++ alone (15,16,34). Attempts to substitute other metals for Cu++, with retention of activity, have been uniformly unsuccessful; whereas the Zn++ may be replaced by Co++, Hg++ or Cd++, without gross modification of activity. It is clear that catalytic activity is centered at the eu++ whereas the Zn++ plays a secondary role. It is certainly the case that the Zn++ contributes to the stability of the enzyme. Replacement of Zn++ by Hg++ gave an active enzyme whose stability actually exceeded that of the native enzyme (36). The catalytic mechanism of the bovine erythrocyte enzyme has been probed by pulse radiolysis. This technique allows the rapid introduction of 02- into buffered solutions of the enzyme, under conditions which allow monitoring either the rate of decay of the 02- or the degree of bleaching of the enzyme. The enzyme has been foun~ to react with 02- at a rate of approximately 2.2 x 109 M-l sec-. This rate was unresponsive to changes of pH in the range 5.5-9.5. Since the pKa for H0 2 • is 4.8 we must conclude that 02-' rather than H0 2 ·, is the substrate. If the enzyme was fully in the cupric form when first exposed to 02- then a partial bleaching of the enzyme occurred during catalysis. If the enzyme was fully reduced prior to exposure to 02-' then a partial oxidation of the enzyme, with augmentation of its 680 nm absorption occurred during the catalytic cycle. This behavior indicates that the copper was alternately reduced and reoxidized during the catalytic cycle (38-43) as follows:
STRUCTURE AND MECHANISM
+ E-Cu + 02
One group reported the rate constants for reactions I and II to be 1.2 x 109 M-l sec-l and 2.2 x 109 M-l sec-I, respectively (40), while anQthef grou£ (43), in substantial agreement, found 2.4 x 10~ M- sec- for both reactions. They furthermore concluded that the reaction was truly diffusion limited, since the energy of activation was only 4.6 Kcal/mole and since the rate constant decreased with increasing viscosity of the solvent, without changes in the energy of activation. Given the enormous turnover rate of this enzyme, it is not surprising that the 4reaction showed no signs of saturation by substrate up to 2.4 x 10- M 02-; which was the maximum concentration of 02- achievable. If the rate constants for reactions I and II are in fact equal then the steady-state bleaching of the enzyme during its catalytic cycle should be 50%. The bleaching actually observed was closer to 25% and on this basis it has been proposed (43) that the enzyme exhibits ~alf of the sites' reactivity. Since the active sites are on opposite sides of the molecule, approximately 36 ~ apart, the interactions needed to explain 'half of the sites' behavior are difficult to envision in this case and one is tempted to seek other explanations for the less than expected bleaching during catalysis. Enzymes facilitate the forward and reverse aspects of any given reaction to the same degree. The superoxide dismutase should therefore catalyze the oxidation of H2 0 2 by oxygen. It was possible to demonstrate this effect by using tetranitromethane as a very efficient scavenger of 02-. Superoxide dismutase was able to catalyze an oxygen-dependent reduction of tetranitromethane by H2 02 (44). The reactions which explain this are:
E-Cu+ + 02
Reactions III and IV are simply the reverse of reactions I and II. In view of reaction III it is not surprising that H202 has been seen to reduce the eu++ and thus to bleach the visible color of the enzyme (30,40,45-48). Less expected was the irreversible inactivation of the enzyme by H2 02 (45-49). It has been reported that one histidine residue per subunit is destroyed during this activation (47). The enzyme acts as a peroxidase towards a variety of compounds which, in being peroxidized, protect the enzyme against inactivation
by H202 (50), Inactivation of the enzyme was accompanied by chemiluminescence. The peroxidation of imidazole or of xanthine, by the superoxide dismutase, was also accompanied by luminescence. In addition, it was observed that oxygen protected the enzyme against inacti vation by H2 02 • The explanation offered for these observations (50) envisions the generation of a potent oxidant as a consequence of the interaction of the cuprous enzyme with H202' This oxidant, which remains bound to the copper, could then attack an adjacent imidazole and thus inactivate the enzyme; or it could oxidize various small molecules, in which case we would observe peroxidase action rather than inactivation. The luminescence comes from electronically-excited products during their return to their ground states. The scheme proposed (50) was: Enz-Cu++ + H202 Enz-Cu+ + H202 Enz-Cu++-OH + ImH
°2- + 2H+
Enz-Cu++ + 1m' + H2O
Reaction III is the reduction of the enzyme by H2 02 already discussed. Reaction V is analagous to the reaction of Fe++ with H202, as in Fenton's reagent, which generates OH·. In this case the OH· is considered to remain bound to the copper. If the bound oxidant attacked an adjacent imidazole ring (ImH), which is part of the enzyme, then the active site would be destroyed as in reaction VI. If, on the other hand, the bound oxidant attacked exogenous imidazole or xanthine or anyone of a number of other compounds, we would observe peroxidase action rather than inactivation. The ability of oxygen to protect the enzyme against destruction by H20 2 can also be explained. Thus if the cuprous enzyme reacts with oxygen, as in reaction IV, then the oxygen would be in competition with H202' That is, reaction IV competes with reaction V for the cuprous enzyme and thus diminishes the rate of generation of bound oxidant and hence diminishes the rate of inactivation of the enzyme. The inactivation of the enzyme by one of its reaction products, H20 2 , is very interesting in that it aids studies of the intimate meclianism. Is this inactivation of any consequence in vivo? Since the concentration of H2 02 needed in order to see reasonable rates of inactivation is approximately 10 ~ and since the highest concentrations of H2 02 reached in vivo are less than this by several orders of magnitude (51) we would answer this question in the negative. Furthermore the inactivation by H2 02 is most rapid at elevated pH and is really rather sluggish close to neutrality (50). Finally, the compounds which protect the enzyme against this inactivation, such as glutathione, are abundant within cells. The level of superoxide dismutase activity within human erythrocytes does not diminish during the life of these cells, whereas the levels of numerous less stable enzymes, do decrease with time (52). It
STRUCTURE AND MECHANISM
is apparent then that inactivation of superoxide dismutase by H202 is not a biological problem. The dismutation of superoxide radicals to H202 plus 02 requires two protons. It is surprising therefore that the enzymatic dismutation is independent of pH over the range 5.5-9.5 (38-43) and furthermore shows no deuterium isotope effect at pH 10 (50). One way to explain this is to suppose that only one proton is transferred during the catalyzed dismutation, such that the product is H0 2- rather than H20 2 and further to provide for some means of facilitating this proton transfer. The imidazolate of histidine 61, which bridges the Cu++ and Zn++ in the resting enzyme, could function to conduct protons as follows (50): I
[-Zn++-Im--Cu++] I I
[-Zn I I
[-Zn I H0 2-
+ ° 2-
I ++ [-Zn -ImH
'6J.++] + °2I
H02 - + [-Zn
6u+] I "
,/ ++] -Im--- Cu I
In reaction VII reduction of the Cu++ by 02- is accompanied by release of the imidazolate bridge from its attachment to the copper. The imidazolate, which remains attached to Zn++, is a strong base and it protonates as in reaction VIII. When 02- then interacts with the reduced enzyme it simultaneously acquires an electron from the Cu+ and a proton from the imidazole of histidine 61. Th~s 02is thus converted to H0 2- while.the bridging of Cu++ and Zn+ is reestablished, as in reaction IX. Finally HOi becomes protonated in free solution as in reaction X. Cyanide and azide both ligate to the active site of this enzyme (25) yet have disparate effects. Thus cyanide is an effective inhibitor, whereas azide is not (30,38,50,53). It has been suggested (25) that azide binds to the Zn++ while cyanide binds to ~~e Cu++. It is possible that both cyanide and azide bind to the Cu but that azide merely displaces one of the non-bridging ligands whereas cyanide binds to that site on the copper which is ordinarily available for interaction with 02-. Binding of N3 - in this way would merely replace one nitrogenous ligand by another and should have relatively minor effects on the epr spectrum and on the activity. Cyanide, in contrast, binds by its carbon end (54) and should thus grossly change the epr spectrum, as it does (25). Specific inhibitors of superoxide dismutase could be very useful both in advancing our knowledge of its catalytic mechanism and in more firmly establishing its biological functions. Unfortunately the list of effective inhibitors is short and does not include compounds expected to exhibit specificity. Thus, as
already stated, cyanide inhibits. Since it is only the copper-zinc enzymes which are cyanide sensitive, whereas the mangani and the ferri superoxide dismutases are not, cyanide has been useful in distinguishing these families of enzymes (14,55). Several copperchelating agents, such as diethyldithiocarbamate, xanthogenate and diphenylthiocarbazone have recently been reported to inhibit the copper-zinc superoxide dismutase (56). The generality that copper-zinc superoxide dismutases are characteristic of the cytosol of eukaryotes has been challenged. Thus a superoxide dismutase bearing one atom of copper and two of zinc per molecule has been reported in a symbiotic luminescent bacterium (57). The ultraviolet absorption spectrum of this enzyme differed strikingly from that seen with the copper-zinc enzymes from eukaryotes. Its place in the evolution of superoxide dismutases will remain uncertain until some knowledge of its amino acid sequence is available. The manganese and the iron-containing superoxide dismutases and their evolutionary relationships will be discussed by Dr. Joe McCord. REFERENCES Accounts Chem. Res.
Fridovich, I. (1972)
McCord, J. M., Beauchamp, C. 0., Goscin, S., Misra, H. P., and Fridovich, I. (1973) in Proc. 2nd Int. Symp. Oxidases and Related Redox Systems, Memphis, 1971 (King, T. E., Mason, H. S., and Morrison, M., eds) pp. 51-76, University Park Press, Baltimore.
Fridovich, I. (1974)
Fridovich, I. (1974) in Molecular Mechanisms of Oxygen Activation (Hayaishi, 0., ed.) pp. 453-477, Academic Press, New York.
Sawada, Y., and Yamazaki, I. 19, 527-536.
Halliwell, B. (1974)
Bors, W., Saran, M., Lengfe1der, E., Stottl, R., and Michel, C. (1974) Curro Top. Radiat. Res. Q. ~, 247-309.
Fridovich, I. (1975)
Bouanchaud, D. (1975)
Advan. Enzymol. 41, 35-97.
(1974) Tampakushktsu Kakusan Koso
New Phytology 73, 1075-1086.
Ann. Rev. Biochem. 44, 147-159. Recherche~,
STRUCTURE AND MECHANISM
J. BioI. Chem. 247,
Misra, H. P., and Fridovich, I. (1972) 3410-3414.
Rapp, U., Adams, W. C., and Miller, R. W. (1973) Biochem. 51, 158-171.
Goscin, S. A., and Fridovich, I. (1972) Acta 289, 276-283.
Asada, K., Urano, M., and Takehashi, M. (1973) Europ. J. Biochem. 36, 257-266.
Weisiger, R. A., and Fridovich, I. (1973) 248, 3582-3592.
McCord, J. M., and Fridovich, I. (1969) 6049-6055.
Beem, K., Rich, W. E., and Rajagopalan, K. V. (1974) Chem. 249, 7298-7305.
Calabrese, L., Frederi ci, G., Bannister, W. H., Bannister, J. V., Roti1io, G., and Finazzi-Agro, A. (1975) Europ. J. Biochem. 56, 305-309.
Mann, T., and Kei1in, D. (1938) B126, 303.
Abernethy, J. L., Steinman, H. M., and Hill, R. L. (1974) J. BioI. Chem. 249, 7339-7347.
Evans, H. J., Steinman, H. M., and Hill, R. L. (1974) Chem. 249, 7315-7325.
Steinman, H. M., Naik, V. R., Abernethy, J. L., and Hill, R. L. (1974) J. BioI. Chem. 249, 7326-7338.
Richardson, J. S., Thomas, K. A., Rubin, B. H., and Richardson, D. C. (1975) Proc. Nat. Acad. Sci. U.S.A. 72, 1349-1353.
Richardson, J. S., Thomas, K. A., and Richardson, D. C. (1975) Biochem. Biophys. Res. Commun. 63, 986-992.
Richardson, J. S., and Richardson, D. C., personal communication.
Fee, J. A., and Gaber, B. P. (1972) J. BioI. Chem. 247,60-65.
Fee, J. A. (1973)
J. BioI. Chem. 248, 4229-4234.
Fee, J. A. (1973)
Biochim. Biophys. Acta 295, 107-116.
J. BioI. Chem. J. BioI. Chem. 244,
Proc. Roy. Soc. (London)
Rotilio, G., Calabrese, L., Mondovi, B., and Blumberg, W. E. (1974) J. Biol. Chem. 249, 3157-3160.
Gaber, B. P., Brown, R. D., Koenig, S. H., and Fee, J. A. (1972) Biochim. Biophys. Acta 271, 1-5.
Rotilio, G., Morpurgo, L., Giovagnoli, C., Calabrese, L., and Mondovi, B. (1972) Biochemistry 11, 2187-2192.
Forman, H. J., Evans, H. J., Hill, R. L., and Fridovich, I. (1973) Biochemistry 12, 823-827.
Stokes, A. M., Hill, H. A., Bannister, W. H., and Bannister, J. V. (1973) FEBS Lett. 32, 119-123.
Haffner, P. H.,. and Coleman, J. E. (1973) J. BioI. Chem. 248, 6626-6629.
Fee, J. A., and Briggs, R. G. (1975) 400, 439-450.
Rotilio, G., Calabrese, L., and Coleman, J. E. (1973) Biol. Chem. 248, 3855-3859.
Forman, H. J., and Fridovich, I. (1973) 2645-2649.
Beauchamp, C. 0., and Fridovich, I. (1973) Biochim. Biophys. Acta 317, 50-64.
Rotilio, G., Br~, R. C., and Fielden, E. M. (1972) Biochim. Biophys. Acta 268, 605-609.
Klug, D., Rabani, J., and Fridovich, I. (1972) J. BioI. Chem. 247, 4839-4842.
Klug-Roth, D., Fridovich, I., and Rabani, J. (1973) J. Am. Chem. Soc. 95, 2786-2790.
Fielden, E. M., Roberts, P. B., Br~, R. C., and Rotilio, G. (1973) Biochem. Soc. Transactions !, 52-53.
Bannister, J. V., Bannister, W. H., Bray, R. C., Fielden, E. M., Roberts, P. B., and Rotilio, G. (1973) FEBS Lett. 32, 303-306.
Fielden, E. M., Roberts, P. B., Bray, R. C., Lowe, D. J., Mautner, G. N., Rotilio, G., and Calabrese, L. (1974) Biochem. J. 139, 49-60.
Hodgson, E. K., and Fridovich, I. (1973) Biochem. Biophys.
Biochim. Biophys. Acta
J. BioI. Chem. 248,
STRUCTURE AND MECHANISM
Res. Commun. 54, 270-274. 45.
Symonyan, M. A., and Na1bandyan, R. M. (1972) FEBS Lett. 28, 22-24.
Roti1io, G., Morpurgo, L., Calabrese, L., and Mondovi, B. (1973) Biochim. Biophys. Acta 302, 229-235.
Bray, R. C., Cockle, S. H., Fielden, E. M., Roberts, P. B., Roti1io, G., and Calabrese, L. (1974) Biochem. J. 139, 43-48.
Roti1io, G., Calabrese, L., Bossa, F., Barra, D., Finazzi-Agro, A., and Mondovi, B. (1972) Biochemistry 11,2182-2187.
Beauchamp, C. 0., and Fridovich, I. (1973) Biochim. Biophys. Acta 317, 50-64.
Hodgson, E. K., and Fridovich, I. (1975) Biochemistry, in press.
Oshino, N., Jamieson, D., Sugano, T., and Chance, B. (1975) Biochem. J. 146,67-77.
McCord, J. M., personal communication.
Tyler, D. D.
Haffner, P. H., and Coleman, J. E. (1973) J. Bio1. Chem. 248, 6626-6629.
Beauchamp, C. O. (1973)
Hirata, F., and Hayaishi, O. (1975) J. Bio1. Chem. 250, 5960-5966.
Fuget, K., and Michelson, A. M. (1974) Biochem. Biophys. Res. Commun. 58, 830-838.
(1975) Biochem. J. 147, 493-504.
Ph.D. dissertation, Duke University.