Heterogeneity among the flavin-containing NADH peroxidases of group D streptococci. Analysis of the enzyme from Streptococcus faecalis ATCC 9790.

Polyclonal antisera prepared against the purified NADH peroxidase from Streptococcus faecalis ATCC 9790 (Enterococcus hirae) do not cross-react with the ATCC 11700 enzyme. Comparative tryptic maps of the two proteins indicate that the differences in primary structures extend beyond those localized to respective antigenic epitopes. Alignments of the NH2-terminal and active-site cysteinyl peptide sequences of the two streptococcal peroxidases reveal identities of 50 and 67% in the respective overlap regions. Dithionite titrations of the ATCC 9790 enzyme reveal a separation in potentials (E2 - E1) for the nonflavin and flavin redox centers of 39 mV, a value nearly 50 mV lower than that observed with the ATCC 11700 peroxidase. Despite these changes in redox behavior NADH titrations of the ATCC 9790 enzyme give rise to both EH2 and EH2.NADH species as previously observed. The enzyme turnover number with hydrogen peroxide is approximately 60% that of the ATCC 11700 peroxidase; the ATCC 9790 peroxidase is also inhibited during turnover with ethyl hydroperoxide. These findings suggest that the flavoprotein NADH peroxidases may exhibit greater diversity among the group D streptococci than previously observed with the widely distributed enzymes of the related flavoprotein disulfide reductase class.

Polyclonal antisera prepared against the purified NADH peroxidase from Streptococcus faecalie ATCC 9790 (Enterococcus hirue) do not cross-react with the ATCC 11700 enzyme. Comparative tryptic maps of the two proteins indicate that the differences in primary structures extend beyond those localized to respective antigenic epitopes. Alignments of the NHz-terminal and active-site cysteinyl peptide sequences of the two streptococcal peroxidases reveal identities of 50 and 67% in the respective overlap regions. Dithionite titrations of the ATCC 9790 enzyme reveal a separation in potentials (& -E1) for the nonflavin and flavin redox centers of 39 mV, a value nearly 50 mV lower than that observed with the ATCC 11700 peroxidase. Despite these changes in redox behavior NADH titrations of the ATCC 9790 enzyme give rise to both El& and El&.NADH species as previously observed. The enzyme turnover number with hydrogen peroxide is approximately 60% that of the ATCC 11700 peroxidase; the ATCC 9790 peroxidase is also inhibited during turnover with ethyl hydroperoxide. These findings suggest that the flavoprotein NADH peroxidases may exhibit greater diversity among the group D streptococci than previously observed with the widely distributed enzymes of the related flavoprotein disulfide reductase class.
NADH peroxidase (l-4) is a unique flavoprotein which appears to serve two functions in the Gram-positive streptococci. First, it eliminates potentially toxic hydrogen peroxide under aerobic growth conditions and represents the only as yet identified enzymatic defense available against HnOn-mediated oxidative stress. Second, the enzyme presents an additional mechanism for regeneration of the oxidized pyridine nucleotide essential to the strictly fermentative metabolism of this organism. It has previously been observed that NADH peroxidase activity is more frequently associated with the group D Streptococcus faecalis (5), although some strains of the oral pathogen Streptococcus mutans (6) possess this adaptive enzyme as well. The Gram-positive streptococci are thought to have diverged from the Gram-negative bacteria well over 1 billion years ago (7-g), prior to the accumulation of oxygen in the atmosphere. These relatively primitive facultative anaerobes resemble a number of obligate anaerobes, such as Ckxtridium and Peptococcus, which also lack hemecontaining proteins (including catalase and the cytochromes) and are thought to reflect very closely the metabolic capacities of the earliest bacterial species (9).
There are several striking parallels which can be drawn between the streptococcal peroxidase and the flavoprotein disulfide reductases (10) such as glutathione reductase. Their redox-active components are similar, in that each contains 1 mol of FAD and one redox-active cysteinyl derivative. In the disulfide reductases this derivative is a protein disulfide; in the peroxidase an unusual stabilized cysteine-sulfenic acid (Cys-SOH) has been identified as the nonflavin redox center (4). With the exception of thioredoxin reductase (ll), each of these flavoenzymes yields a spectroscopically distinct intermediate (two-electron reduced NADH peroxidase, EH2) on two-electron reduction; the charge-transfer absorbance at long wavelength is due to the interaction between a nascent cysteine thiolate and the oxidized flavin. We have recently shown (3) that the single cysteine residue in the peroxidase polypeptide follows a putative /~Lv@ super-secondary structural element; this allows alignment with the corresponding reeidues in the glutathione reductase NH*-terminal sequence. These enzymes also show the same stereospecificity of hydride transfer with their respective pyridine nucleotide substrates (12,13), and their kinetic mechanisms appear similar (12,14). Yet the sequence of the active-site cysteinyl peptide from the peroxidase (2) bears no clear relationship to the corresponding peptide which is so highly conserved in all but one of the disulfide reductases (15,16).
These and other observations have raised questions about the evolution of the streptococcal peroxidase gene. Interest in this regard has been heightened with the recent observation (17,18) that the streptococcal NADH oxidase, a flavoenzyme capable of reducing O2 -+ 2Hz0, is related to the NADH peroxidase. The development of these two gene products thus appears to parallel, to a certain extent, the evolutionary history of lipoamide dehydrogenase and glutathione reductase (8,15,19). Since the streptococci appear to lack the high levels of glutathione found in Escherichiu coli (8) the NADH peroxidase may have arisen in an even more urgent response to the accumulation of oxygen in the environment. The genes encoding lipoamide dehydrogenase and glutathione reductase appear to have evolved relatively slowly (19); comparisons (20) of the mercuric reductase genes from the Gram-positive Bacillus sp. and from the Tn501 transposon found in the disulfide reductase as well. The immunological cross-reactivity observed (21) for all mercuric reductases from Gramnegative bacteria (with only one exception (22)) lends further support to this conclusion.
In order to begin to trace the evolutionary history of the

Enzyme
Puri+ztion-The results of our purification scheme for the ATCC 9790 NADH peroxidase are given in Table I. Approximately 7 mg of pure enzyme were obtained in an overall 15% yield from 140 g (wet weight) of S. fuecalis. There are important differences between this protocol and that previously established for the ATCC 11700 enzyme (1). At 55% ammonium sulfate, over 60% of the ATCC 9790 peroxidase is precipitated, whereas the S. faecalis 1OCl enzyme remains in the supernatant under these conditions. In order to provide better resolution by reversed-phase ammonium sulfate chromatography, we chose a phenyl-Sepharose matrix and eluted the enzyme with a gradient of 38 to 0% ammonium sulfate. The ion-exchange chromatography step was also modified slightly to accommodate differences in behavior observed with the ATCC 9790 peroxidase. SDSpolyacrylamide gel electrophoresis analysis of the purified enzyme ( Fig. 1) gives an apparent subunit molecular weight of 50,000 and shows that there are no intersubunit disulfides. The specific activity of the purified ATCC 9790 NADH peroxidase is -60% that of the S. fueculis lOC1 enzyme as assayed under identical conditions. The spectral properties of the ATCC 9790 enzyme are very similar to those of the previously purified flavoprotein peroxidase (1). Absorbance ratios are 7.5, 1.03, and 1.0 at 280, 378, and 450 nm, respectively.
In order to better assess similarities in overall covalent structure between the two peroxidases the enzymes were subjected to parallel trypsin digestions without prior modification. The tryptic maps in Fig. 2  A 10% acrylamide gel was run with samples pretreated by heating for 5 min in sample buffer containing 2% SDS with (lanes 2-8) or without (Inn? 9) 5% 2-mercaptoethanol. The gel was electrophoresed at 25 mA in the presence of 0.1% SDS. Lanes 2-7, pooled samples from Steps I-VI of the purification; lanes 8 and 9, purified enzyme pretreated in the presence and absence, respectively, of 2mercaptoethanol. Lanes 1 and IO contain molecular weight standards, as indicated.
common. The unmodified cysteinyl peptide of the S. fuecalis lOC1 peroxidase has been shown to elute at -100 min in this HPLC system, based on experiments with the ""S-labeled enzyme (4).
The differences in overall structure between the two peroxidases were reinforced by the results of the Western analysis shown in Fig. 3. Polyclonal rabbit antisera to the purified ATCC 9790 enzyme detect the protein with good sensitivity (below 2 ng) in both crude extracts and in purified form. In addition, these affinity-purified antisera also react strongly with a commercially available NADH peroxidase preparation. However, they do not recognize the purified ATCC 11700 peroxidase when present at even lo-fold higher levels. In separate experiments not shown we have also determined that polyclonal rabbit antisera against the S. fuecalis lOC1 enzyme react very poorly, if at all, with the ATCC 9790 protein. We conclude that antigenic epitopes are not strongly conserved between the two enzymes.
Active-site Peptide Isolation and Sequence--5,5'-Dithiobis(2-nitrobenzoate) titration of the purified ATCC 9790 peroxidase under anaerobic conditions in the presence of 4 M guanidine HCl demonstrated the absence of free sulfhydryls. Anaerobic reduction of the native enzyme with 1 eq/FAD of NADH, followed by denaturation with 4 M guanidine HCl and 5,5'-dithiobis(2-nitrobenzoate) titration, resulted in a value of 0.9 thiols/FAD in the two-electron reduced enzyme. These data are consistent with the presence of a single redox- with TPCK-treated trypsin, and the resulting peptide mixture was resolved by HPLC (Fig. 4). When fractions were collected manually over the range 63-80 min two radiolabeled peaks were identified.
Peptide T-l eluted at -69 min and accounted for about one-third (5500 dpm) of the label; peptide T-2 eluted at 72 min and contained 64% (9600 dpm) of the combined radioactivity.
Tryptic peptide T-2 was purified by rechromatography and, on automated Edman degradation, gave the sequence MSCGMELYLEDQVTDVNDV(R/K).
The presence of the phenylthiohydantoin-carboxamidomethylcysteine residue in cycle 3 was confirmed by counting all 20 cycles for radioactivity.
In order to identify the secondary radiolabeled peptide (T-l) eluting at -69 min in the primary chromatogram, the alkylation with iodo[l-'4C]acetamide was repeated with a modified protocol (see "Experimental Procedures"). Although we were not able to separate the radiolabeled peak from one major contaminating peptide, sequence and radiolabel analyses indicate that tryptic peptide T-l results from a secondary cleavage of the T-2 peptide after the NHn-terminal Met residue. In addition, tryptic peptide T-2 was also isolated as a result of this second alkylation experiment and its sequence confirmed that observed previously.
No attempt was made to analyze any of the smaller radioactive peaks identified in the chromatogram given in Fig. 4. Reductiue Titrations-Dithionite titration of the NADH peroxidase requires 2 eq/FAD of reductant, as shown in Fig.  5, and proceeds through a characteristic two-electron reduced intermediate (EH2) with charge-transfer absorbance at 540 nm. Comparison of experimental and theoretical curves for this titration, however, indicates that only 69% of the enzyme appears as the EH, species on half-reduction.
The corresponding formation constant K can be calculated from the relationship: The value of 20 determined for this enzyme is considerably lower than that for the ATCC 11700 enzyme (K = 780 under identical conditions (3)) and indicates a significant increase in the extent of disproportionation of EH, to oxidized (E) and fully reduced (EH,) enzyme at equilibrium.
From the formation constant one can estimate that the separation in potentials (Ep -E,; where Ez is the midpoint potential for redox couple E/EHn and El is the midpoint potential for redox couple EH2/EH4) between the nonflavin redox center and the flavin is 39 mV. The redox potentials of the two centers are considerably closer in this enzyme than in the S. fuecalis lOC1 enzyme, where Ez -E, is ~86 mV (3). The enhanced disproportionation in the present case also explains why the absorbance at 450 nm for the EHp intermediate is only 87% of that observed with the ATCC 11700 enzyme at half-reduction; this difference is attributable to the 15% of the total enzyme present as EH, at equilibrium.
The altered redox behavior of the ATCC 9790 peroxidase is somewhat similar to that of the S. fuecalis lOC1 enzyme in the presence of low concentrations of urea (3)  The EH, spectra of the dithionite-and NADH-reduced S. fuecalis lOC1 peroxidase are virtually identical, although oxidized pyridine nucleotide appears to remain tightly bound in the latter case (1, 3). With the ATCC 9790 peroxidase, however, the A460 and ASdO values of the NADH-titrated enzyme (Fig. 6) are -20% higher than those of dithionite-generated EH2. These observations provide further support for a specific, high-affinity interaction between the two-electron reduced peroxidase and NAD'. We conclude that the presence of NAD' leads to preferential stabilization of EH2: The subsequent addition of a second eq/FAD of NADH under anaerobic conditions (Fig. 6) yields an EH,.NADH complex similar to that previously observed (1). Recent studies with the ATCC 11700 enzyme indicate that the higher-extinction charge-transfer band of EH,.NADH is primarily due to an NADH+FAD interaction (4); the very low Kd (~10~* M) for this complex suggests a probable role in catalysis, as described recently (4). Inhibition by Ethyl Hydroperoxide-As discussed previously, the proposed mechanism for the streptococcal NADH peroxidase, involving a reducible cysteine-sulfenic acid (4), is very similar to that proposed for glutathione peroxidase, which is thought to shuttle between selenolate (Cys-Se-) and selenenic acid (Cys-SeOH) redox states in catalysis (28). The latter enzyme is also known to catalyze the reductions of alkyl hydroperoxides at rates comparable to those of hydrogen peroxide reduction (28). There are, however, major diff:rences in active-site structures for the two enzymes. The 2-A x-ray structure of the selenoenzyme shows the essential Cys-Seresidue to be readily accessible from solvent channels (28). Chemical modification studies of the streptococcal peroxidase, on the other hand, indicate a very restricted access to solvent for the reduced active-site cysteine (3). In order to extend the analysis of the flavoprotein peroxidase with respect to its specificity for peroxide substrates, we investigated ethyl hydroperoxide as a substrate with the ATCC 9790 enzyme. Whereas initial rates of NADH oxidation by the streptococcal peroxidase under standard assay conditions (1.3 mM H202, pH 5.4) are linear for several minutes (l), substitution of ethyl hydroperoxide at the same concentration resulted in nonlinear progress curves as followed at 340 nm. Fig. 7 (12); the concentration of NADH remaining under the conditions of the assay at complete inhibition is on the order of 0.1 mM. As shown in Fig. 7, the rate of inhibition is essentially first-order and corresponds to an apparent rate of 0.07 min-' at 5.3 mM ethyl hydroperoxide.
The observed rate increases to 0.15 min-' at 12.4

DISCUSSION
The analysis presented in this report suggests that considerable heterogeneity exists among the streptococcal NADH peroxidases. The comparative tryptic maps of the ATCC 9790 and ATCC 11700 enzymes indicate extensive differences between the respective polypeptides.
The immunochemical analysis, which shows essentially no cross-reactivity between the two peroxidases, demonstrates that antigenic epitopes are not conserved in the two proteins. In contrast, previous reports have shown that all mercuric reductases from Gramnegative sources (21), with the exception of the enzyme from Thiobacillus ferrooxidans (22), are immunochemically crossreactive. The strong cross-reaction observed between the ATCC 9790 NADH peroxidase and a commercially available preparation from S. faecalis ATCC 8043 is consistent with the results of recent molecular and chemotaxonomic analyses since these strains are now both designated as E. hirae (23, 29); this species does not include the ATCC 11700 strain which has been reclassified as Enterococcus faecalis (29).
The chromosomal merA gene of the Gram-positive Bacillus sp. (20) encodes a polypeptide which exhibits only 40% amino acid sequence identity when compared to the Gram-negative Tn501 (P. aeruginosa) mercuric reductase, and does not appear to cross-react with the anti-Tn501 mercuric reductase sera (22). Despite the heterogeneity observed among mercuric reductase sequences from Gram-negative and Gram-positive bacteria Laddaga et al. (30) have shown that the percentage identity in those amino acid sequences corresponding to active-site positions and to FAD and NADPH contacts is greater than 90%. The NHz-terminal sequence of the ATCC 11700 NADH peroxidase (3) contains a putative &YP super-secondary structural element which is likely to represent either an FAD or NADH binding site. Alignment with the corresponding segment of the ATCC 9790 enzyme (Table II) shows that this motif is conserved, although only 7 of 14 residues in the overlap region are identical. The NHB-terminal segments of the ATCC 11700 peroxidase and the NADH oxidase of this strain also show nearly 50% identity over a 15-residue overlap. The active-site cysteinyl peptide sequences given for the two peroxidases are identical in 8 of 12 positions within the indicated overlap; again, a similar comparison can be made for the active-site sequences of the ATCC 11700 NADH peroxidase and NADH oxidase. These limited sequence comparisons, confined to predicted FAD or NADH binding regions and to active-site positions, suggest that the two peroxidase genes have undergone considerable divergence relative to the genes encoding mercuric reductase (30) and glutathione reductase (31). With respect to the active-site peptides, however, different structural requirements and stringencies must be met in the latter two cases to accommodate the redox-active disulfides in the respective oxidized enzymes. Furthermore, the conformation about the disulfide in glutathione reductase is somewhat unique; the main chain in this region is subject to mechanical strain which could be of functional importance (32). As the peroxidases do not contain disulfides at their active centers, more flexibility in amino acid replacement may have been tolerated over time.
The results of this analysis also serve to further distinguish the streptococcal NADH peroxidases from the flavin-dependent alkyl hydroperoxide reductase recently purified from Salmonella typhimurium (33). The latter enzyme appears to exist as a heterodimer (apparent subunit molecular mass values of 57 and 22 kDa); the larger subunit contains 1 mol of FAD. The enzyme is specific for alkyl hydroperoxides; hydrogen peroxide appears to inactivate the alkyl hydroperoxide reductase during turnover with NADPH. The 22-kDa subunit is thought to contain one redox-active disulfide which represents the site of alkyl hydroperoxide reduction. Aside from the differences in quaternary structure and in the nature of the nonflavin redox centers we have also shown that ethyl hydroperoxide actually inhibits the streptococcal peroxidase in turnover. Furthermore, the NADH peroxidase is specific for its pyridine nucleotide substrate whereas the S. typhimurium enzyme uses either NADH or NADPH. In addition, NH2terminal sequence comparisons (Table II) show no clear relationship between the two flavin-dependent hydroperoxidases. These structural and mechanistic distinctions serve to further underscore the parallels which exist between the enteric streptococci and the enteric Gram-negative bacteria, in terms of their fIavin-linked evolutionary responses to oxygen accumulation.