Activation of the Proton Transfer Pathway in Catalysis by Iron Superoxide Dismutase*

Catalysis by Escherichia coliand Porphyromonas gingivalis iron superoxide dismutase was activated by addition of primary amines, as measured by pulse radiolysis and stopped-flow spectrophotometry. This activation was saturable for most amines investigated, and a free energy plot of the apparent second-order rate constant of activation was linear as a function of the pK a of the amine, indicating activation by proton transfer. Amines provide an alternate rather than the only pathway for proton transfer, and catalysis was appreciable in the absence of amines. Solvent hydrogen isotope effects were near unity for amine activation, which is consistent with rate-contributing proton transfer if the pK a of the proton acceptor on the enzyme is not in the region of the pK a values of the amines studied, from 7.8 to 10.6. The activation of catalysis by these amines was uncompetitive with respect to superoxide, interpreted as proton transfer in a ternary complex of amine with the enzyme-bound peroxide dianion.

Iron superoxide dismutase (Fe-SOD), found in prokaryotes, and Mn-SOD, found in mitochondria as well as in prokaryotes, are highly similar in sequence (5) and in crystal structure (6,7). Despite this high degree of structural homology, substitution of one metal for the other generally results in an inactive enzyme (8,9). A subclass of SOD exists that is able to utilize either manganese or iron as catalytic metal ion; these are termed cambialistic SODs and are of significant interest from the perspective of metal specificity as well as catalytic mechanism. A well studied cambialistic SOD is that from Porphyromonas gingivalis, the crystal structure of which is reported for the manganese-and iron-containing forms (10). Although they all share common features of their catalytic pathways, the copper/ zinc-and nickel-containing SODs have structures unrelated to the manganese-and iron-containing SODs (2,11).
The crystal structures of the iron-and manganese-containing superoxide dismutases generally show a trigonal bipyramidal geometry about the oxidized metal with three histidines, one aspartic acid, and one solvent molecule as ligands (6,7). Major features of the active site appear very similar in each enzyme, including the highly conserved hydrogen-bonded network connecting the metal-coordinated solvent molecule and residues Gln-69 2 (in Fe-SOD) or Gln-143 (in Mn-SOD), Tyr-34, His-30, and Tyr-166 from the adjacent subunit. Although this active site Gln emanates from a different backbone position in Fe-SOD and Mn-SOD, the side chain orientation is very similar in both. This Gln has a significant role in the modulation of the redox potential of the metal (8,(12)(13)(14). Catalytic measurements on Escherichia coli Fe-SOD by stopped-flow spectrophotometry (15) and on human Mn-SOD by pulse radiolysis (16) show efficient enzymes with a maximal catalytic turnover near 25-40 ms Ϫ1 and with k cat /K m near diffusion control. Catalysis by Mn-SOD exhibits a strong and reversible product inhibition that occurs within milliseconds in measurements by stoppedflow and pulse radiolysis (16,17). The identity of this inhibited form is suggested to be the side-on peroxo-complex of Mn(III)SOD (17), but structural data are not yet available to confirm this suggestion. Such a product inhibition is not observed with Fe-SOD (15), although Fe-SOD is subject to irreversible inhibition by the Fenton reactions of H 2 O 2 (3).
With maximal catalytic turnovers k cat in excess of 10 4 s Ϫ1 , both Fe-SOD and Mn-SOD appear determined in rate by proton transfer steps that must occur at a rate at least as rapid as k cat . The maximum velocity of catalysis by E. coli Fe-SOD has an appreciable solvent hydrogen isotope effect (SHIE), near 3 (18), and moreover, it is enhanced by millimolar concentrations of general acid catalysts such as primary amines (15), features that strongly suggest a rate-contributing proton transfer in the catalytic pathway. This involves either one or two proton transfers from solvent to the enzyme depending on whether product dissociates as HO 2 Ϫ or as H 2 O 2 , which has not been determined. Moreover, if two proton transfers are involved, it is not known at which stage of catalysis, Reaction 1 or 2, these proton transfers occur. Three ionizable groups that could possibly be * This work was supported by Grant GM 54903 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Box 100267, University of Florida, Gainesville, FL 32610-0267. involved in such proton transfers are Tyr-34, His-30, and the metal-bound water. These are all participants in the hydrogen bond network of the active site. Replacement of Tyr-34, His-30, as well as other residues within the hydrogen-bonded network of human Mn-SOD results in k cat values decreased at least an order of magnitude, suggesting a role for the hydrogen bond network in support of the proton transfers (19). In addition, the dissociation of the product-inhibited complex of human Mn-SOD is associated with a SHIE near 2, suggesting a role for proton transfer (20).
The activation of E. coli Fe-SOD by general acid catalysts provides an opportunity to examine in closer detail the proton transfer steps in the overall catalytic pathway. Such activation by exogenous proton donors has been informative in elucidating proton transfer pathways in other protein systems such as carbonic anhydrase, giving information on the properties of the proton transfer and of the proton acceptor on the enzyme (21). We used stopped-flow and pulse radiolysis techniques to show that activation of Fe-SOD by primary amines is an example of nonessential activation; that is, it occurs via proton transfer as an alternate rather than an exclusive pathway for proton transfer. Moreover, the mechanism of this activation is uncompetitive with respect to substrate superoxide, limiting the steps in the pathway where this activation could occur. The SHIE for proton transfer by exogenous amines and the linear free energy plot of the activation of catalysis of Fe-SOD establish limits for the pK a of the proton acceptor group on the enzyme.

MATERIALS AND METHODS
Enzymes-Purification of E. coli Fe-SOD was carried out as described in Slykhouse and Fee (22) with the following modifications: Sod A Ϫ /Sod B Ϫ E. coli containing a construct with E. coli Fe-SOD (gift of Dr. Anne-Francis Miller) was grown to an A 600 of 0.4 before induction with isopropyl-b-D-thiogalactopyranoside. Cells were harvested by centrifugation and lysed. Purification proceeded as described, omitting the hydroxylapatite column. Enzyme purity was monitored at each step by SDS-polyacrylamide gels, which upon the elution from the final column showed only one intense band. Enzyme concentration was determined spectroscopically, as was metal occupancy, by measuring the A 350 , which corresponds to the Fe 3ϩ with ⑀ ϭ 1850 M Ϫ1 cm Ϫ1 (22). The total enzyme concentration was taken as the total metal-occupied enzyme. P. gingivalis Fe-SOD was graciously supplied by Dr. Fumiyuki Yamakura. Metal occupancy was determined by flame atomic absorption spectroscopy. For P. gingivalis Fe-SOD, iron occupancy was 59%, and manganese occupancy was 0.6%.
Stopped-flow Spectrophotometry-Activation of catalysis by E. coli Fe-SOD was measured by the stopped-flow technique of McClune and Fee (23) in which a solution of KO 2 in Me 2 SO is diluted into an aqueous solution containing enzyme and buffers. KO 2 (Aldrich, spectrophotometric grade) was dissolved in the dimethyl sulfoxide with an approximately equimolar amount of 18-crown-6 ether to increase solubility. A dual-drive sequential mixing stopped-flow spectrophotometer (Applied Photophysics, SX.18MV) was used to measure the change in the absorption of superoxide at 250 nm (⑀ ϭ 2000 M Ϫ1 cm Ϫ1 (24)). To enhance the efficiency of mixing, we used a sequence of two dilutions. In the first, superoxide in Me 2 SO from a 250-l syringe was mixed with an aqueous solution containing 2 mM Caps and 1 mM EDTA at pH 11 from a 2.5-ml syringe, a 10:1 dilution of the Me 2 SO solution. After a 500-ms delay, this solution was rapidly diluted again in a 1:1 ratio with a solution containing enzyme, 200 mM buffer (sodium borate, Taps, or Ches), 1.0 mM EDTA, and primary amines as activators. Average dead time was 4 ms for the mixing of enzyme and substrate. Final Me 2 SO concentration was 4.5% by volume. No enhancement of catalytic activity of E. coli Fe-SOD was detected when borate or Taps buffer were increased in concentration from 5 to 180 mM; ionic strength was maintained with 0.1 M Na 2 SO 4 . Concentrations of borate buffer above 0.2 M depressed the activity of Fe-SOD by at most 40%, an effect noted previously (25). D 2 O (99.9 atom %, Isotec, Miamisburg, OH) was used to measure solvent hydrogen isotope effects; D 2 O was filtered through charcoal and glassdistilled. All experiments were carried out at 25°C at either a 2-or 10-mm path length. The average of four to eight reaction traces of the first 5-10% of the reaction was used to determine initial rates. The uncatalyzed rates were subtracted.
Pulse Radiolysis-Pulse radiolysis experiments were carried out at Brookhaven National Laboratory using the 2 MeV van de Graaff accelerator. Dosimetry was determined using the KSCN dosimeter as described by Cabelli et al. (26). All UV spectra were recorded at 25°C on a Cary 210 spectrophotometer with a path length of either 2 or 6.1 cm. Solutions contained enzyme, 30 mM formate, and 0.5 M ethanol (as hydroxyl radical scavengers), 50 M EDTA, and 2 mM Taps at pH 8.2. Superoxide radicals formed in an aqueous air saturated solution via the mechanisms described by Schwarz (27). Decay of superoxide was observed spectrophotometrically at 260 nm. Initial rates were calculated from the first 10% of the progress curve and were subsequently used to determine the steady-state parameters through a least-squares fitting process (Enzfitter; Biosoft).

RESULTS
We observed the saturable activation by exogenous amines of catalysis by E. coli Fe-SOD at steady state. Results determined by stopped-flow spectrophotometry for ethanolamine are shown in Fig. 1 (plotted using the concentration of ethanolammonium ion) in which the concentration of superoxide was maintained at the near saturating level of 780 M. K m O _ ⅐ 2 for superoxide was determined to be 100 Ϯ 20 M under the conditions of Fig. 1; this and other steady-state constants for catalysis by E. coli Fe-SOD are given in Table I and are in agreement with those reported earlier (15). Fig. 1 demonstrates that there is considerable catalytic activity in the absence of ethanolamine.
Bull and Fee (15), in their study of activation of E. coli Fe-SOD, did not observe saturation even at amine concentrations near 100 mM. In our hands, this activation was saturable for the activating amines of Table II,  conditions of Fig. 2; that is, the data at different ethanolamine concentrations came to a common intercept on the ordinate.
Thus, the activation of superoxide dismutase by these exogenous amines was found to be an example of nonessential, uncompetitive activation (28). This behavior is consistent with the nonessential activator mechanism shown in Scheme 1 and described by Equation 1.
Here BH ϩ is the exogenous proton donor, and ␤ is the factor by which k cat is enhanced by activator. At the saturating concentration of O 2 . , the result shown in Equation 2 occurs, v ͓E͔ ϭ k cat obs ϭ k cat shows that a plot of 1/(k cat obs Ϫ k cat ) versus 1/[BH ϩ ] should be linear with an intercept that gives the maximal velocity attributed to the activation k cat (␤ Ϫ 1) and a slope that gives the (reciprocal of the) apparent second-order rate constant due to the activating amines, (␤ Ϫ 1)k cat /␤K m B . Such a plot is shown in the inset to Fig. 1 Table II along with ␤ and (k cat /K m ) donor values. Weak acids that are not primary amines such as morpholine and 1,2-dimethylimidazole did not enhance catalysis under the conditions of Fig. 1. Activation by primary amines was observed in catalysis with Fe-SOD from P. gingivalis (Table I).
Values of the apparent second-order rate constant for activation (k cat /K m ) donor are plotted versus the solution value of the pK a of the activating amine in Fig. 3. Even the largest value of (k cat /K m ) donor (4.7 M Ϫ1 s Ϫ1 for glycine methyl ester; Table II) is not close to diffusion control, in contrast to k cat /K m O _ ⅐ 2 , which is closer to diffusion control at 250 M Ϫ1 s Ϫ1 for E. coli Fe-SOD (Table I). The free energy plot of Fig. 3 is fit to a straight line with a slope of 0.50 Ϯ 0.07. ␤ and (␤ Ϫ 1)k cat values did not display a linear correlation on a free energy plot. Solvent hydrogen isotope effects D [(k cat /K m ) donor ] 4 were all within error unity with the exception of ethanolamine, which has a large standard deviation (Table II). We have no data to indicate the pK a of the amines when bound to enzyme in the complex that leads to activation of catalysis; in the absence of such data, we have used the solution values of pK a of amines in Fig. 3. This may account in part for the scatter of points in this figure; additional scatter could be attributed to structural variations in the general acid catalysts. The binding of these amines to E. coli Fe-SOD is weak as suggested by the K m B values and the concentrations of amines required to approach saturation; the K m B values range from about 4 to 50 mM for the amines of Fig.  3. Perhaps such weak binding would indicate little change in pK a of the amine upon binding. This was the case with activating derivatives of imidazole and pyridine during catalysis by carbonic anhydrase (29). DISCUSSION General Acid Catalysis and Fe-SOD-A role for exogenous proton donors and acceptors in catalysis is well studied in enzymatic reactions that consume or generate protons, such as the hydration/dehydration of CO 2 /HCO 3 Ϫ catalyzed by carbonic anhydrase (29,30) and the activation of a site-specific mutant of aspartate aminotransferase (31). However, there is an interesting difference between activation of Fe-SOD and these examples. In carbonic anhydrase, the steady-state turnover in the catalyzed dehydration of HCO 3 Ϫ approaches zero as the concentration of exogenous proton donor is reduced to concentrations below 1 mM (32,33). For many isozymes of carbonic anhydrase, His-64 acts as a proton shuttle to transfer protons from buffers in solution to the zinc-bound hydroxide at the active site (34). After each dehydration step that releases CO 2 , the zinc-bound hydroxide must be reprotonated from solution to complete the catalytic cycle. Water itself is not sufficient as a proton donor to the zinc-bound hydroxide or to His-64 to carry out this process at a rate exceeding about 10 3 s Ϫ1 (35). As a result, in the absence of exogenous buffers, catalysis is nearly abolished.
In catalysis by Fe-SOD, the steady-state velocity does not approach zero in the absence of exogenous amine but maintains an appreciable value at 2 ms Ϫ1 (Fig. 1). This fact excludes a catalytic mechanism in which a residue of the enzyme itself acts as proton shuttle and the function of the exogenous amine is to rapidly protonate this shuttle residue. This mechanism is excluded because it requires a rapid reprotonation rate of the shuttle residue, and catalysis would decrease greatly in the absence of exogenous proton donor, which is not observed (Fig.  1). These observations suggest that solvent itself is the predominant mechanism that provides the proton in catalysis by Fe-SOD. This was also suggested on the basis of kinetic and structural data, for example the pH independence of k cat , the SHIE of 3 on k cat (15,18), and the crystal structures of E. coli Fe-SOD and azide-inhibited complexes (7). Since the metalbound water molecule will have its pK a depressed, this is a likely source of rapid proton transfer in the active site. When the iron of Fe-SOD is in the oxidized state, the solvent ligand has pK a near 5, whereas in the reduced state, this ligand has pK a from 9 to 11 (36). Further analysis of the activation data,  Fig. 1 (pH  8.3, 25°C); data for P. gingivalis FeSOD are determined by stoppedflow at pH 9.0, 100 mM TAPS with other conditions as described in Fig.  1. Standard errors are typically 15% or less. discussed below, suggests that amines as exogenous proton donors provide an alternate pathway to the proton transfer in which the metal-bound water is the proton donor and the bound peroxide anion is the proton acceptor. We observed activation by amines of P. gingivalis Fe-SOD with kinetic constants similar to those for E. coli Fe-SOD (Table I), indicating that this activation is not a specific property of the E. coli enzyme but is probably general among many Fe-SODs.
Free Energy Plot for General Acid Catalysis-Significant features of the free energy plot or Brønsted plot for activation by exogenous amines (Fig. 3) yield further information on the catalysis. The logarithm of the rate constant (k cat /K m ) donor describing the activation of Fe-SOD catalysis by amines was a linear function of the pK a of activating amines (Fig. 3); this confirms that activation occurs by proton transfer. We did not observe any significant correlation in k cat obs or ␤, which may mean that the component of k cat due to activation does not have a significant contribution from proton transfer steps or that we lack accuracy on this parameter, which is difficult to measure, requiring large concentrations of both superoxide and amine. The slope of a free energy plot for a proton transfer reaction provides evidence on the position of the transition state with respect to reactants and products (37). The slope in Fig. 3, which is the Brønsted coefficient ␣, is 0.50 Ϯ 0.07, indicating a position midway between reactants and products. In a free energy plot with extensive curvature, it is possible to estimate the pK a of the donor group on the enzyme (35,38). However, with the free energy plot of Fig. 3, it is not possible to estimate the pK a of the proton acceptor group on Fe-SOD.
This lack of curvature is in contrast to the very curved free energy plots for bimolecular proton transfer between nitrogen and oxygen acids and bases occurring non-enzymatically in solution (39) and is also in contrast to the very curved plots and low intrinsic kinetic barrier observed for general acid catalysis in site-specific mutants of carbonic anhydrase (40). The lack of curvature in Fig. 3, although over a narrow range of pK a values, indicates that the proton transfer to the Fe-SOD from exogenous amines has a large intrinsic kinetic barrier when  1 and 2) and solvent hydrogen isotope effects for activation of FeSOD catalysis by various amines. The entries are the exogenous proton donors as numbered in Fig. 3 with conditions as described in Fig. 1  compared with that for the non-enzymatic transfer of protons between electronegative atoms, which is near 2 kcal/mol (39).
Previous reports on proton transfer between electronegative atoms describe increased values of the SHIE in a narrow range of pK a values for the proton donors (2-6 pK a units with a maximal value of the SHIE at pK a(donor) Ϫ pK a(acceptor) near zero) with SHIE near unity outside this range. This observation was made both for non-enzymatic proton transfer between electronegative atoms (41,42) and for the enzymatic case of carbonic anhydrase (40,43). An explanation for a maximum in isotope effect near ⌬pK a value of zero in proton transfer reactions is based on a comparison of symmetric and asymmetric transition states by Westheimer (44). The SHIE on the rate constant (k cat /K m ) donor is unity within error for activation of Fe-SOD by exogenous amines (Table II). That we do not observe a region of increased SHIE for rate constant (k cat /K m ) donor (Table II) indicates that the proton-donating amines we used do not have pK a values close to that of the proton acceptor in Fe-SOD. This suggests, for example, that Tyr-34 (pK a 8.5 for E. coli Fe(II)SOD; Ref. 45) is not the proton acceptor, and perhaps neither is the aqueous ligand of Fe(II)SOD if it has a pK a less than 11. These conclusions are consistent with the pH-independent nature of k cat for E. coli Fe-SOD in the pH range of 7-11 (15), since k cat for the overall catalysis is limited by the proton transfer step. The data are also consistent with proton transfer to the peroxide dianion bound to iron, which is anticipated to have very high pK a .
Uncompetitive Activation-The data for activation of E. coli Fe-SOD by glycine (Fig. 2) and ethanolamine (data not shown) indicate an uncompetitive activation with respect to superoxide. The steady-state rate equation derived from the catalytic pathway of Reactions 1 and 2 is complex and contains terms with substrate concentration squared. However, we can draw some conclusions from the uncompetitive nature of the activation of catalysis by exogenous amines. The most straightforward interpretation based on known features of the pathway is that the activators donate a proton within a ternary complex of enzyme, peroxide, and amine. 5 We assume that rate-limiting proton transfer by amines assists product release, as shown in Scheme 1. Scheme 1 reflects uncompetitive activation if ␤ in Equations 1-3 is greater than unity (inhibition if ␤ is less than unity). Scheme 1 is based on the 5-6-5 mechanism of Lah et al. (7) in which substrate superoxide binding in the active site expands the coordination shell of the metal to octahedral. Scheme 1 does not show the coordination sites of three histidines and one aspartate. Thus, our data are consistent with the amines providing an alternate pathway to proton donation by the iron-bound water. The data do not indicate where the amine binds for this proton transfer; the active site channel of Fe-SOD is narrow (7). (k cat /K m ) donor values are near 1 M Ϫ1 s Ϫ1 and far from diffusion control. These exogenous donors could bind at some distance from the metal and transfer protons through a hydrogen-bonded chain to the iron-bound peroxide anion.