Purification and characterization of a chimeric enzyme from Haemophilus influenzae Rd that exhibits glutathione-dependent peroxidase activity

While belonging to the same family of antioxidant enzymes, members of the peroxiredoxins do not necessarily employ one and the same method for their reduction. Most representatives become reduced with the aid of thioredoxin, whereas some members use AhpF, tryparedoxin, or cyclophilin A. Recent research on a new peroxiredoxin isoform (type C) from Populus trichocarpa has shown that these particular types may also use glutaredoxin instead of thioredoxin. This finding is supported by the occurrence of chimeric proteins composed of a peroxiredoxin and glutaredoxin region. A gene encoding such a fusion protein is enclosed in the Haemophilus influenzae Rd genome. We expressed the H. influenzae protein, denoted here as PGdx, in Escherichia coli and purified the recombinant enzyme. In vitro assays demonstrate that PGdx, in the presence of dithiothreitol or glutathione, is able to protect supercoiled DNA against the metal ion-catalyzed oxidation-system. Enzymatic assays did, indeed, characterize PGdx as a peroxidase, requiring the glutathione redox cycle for the reduction of hydrogen peroxide (k(cat)/K(m) 5.01 x 10(6) s(-1) m(-1)) as well as the small organic hydroperoxide tert-butylhydroperoxide (k(cat)/K(m) 5.67 x 10(4) s(-1) m(-1)). Enzymatic activity as function of the glutathione concentration deviated from normal Michaelis-Menten kinetics, giving a sigmoidal pattern with an apparent Hill coefficient of 2.9. Besides the formation of a disulfide-linked PGdx dimer, it was also shown by mass spectrometric analysis that cysteine 49, which is equivalent to the active site cysteine of the peroxiredoxins, undergoes glutathionylation during purification under nonreducing conditions. Based on these results, we propose a model for the catalytic mechanism.


Introduction
Aerobic organisms intrinsically encounter reactive oxygen species (ROS) 1 , such as hydrogen peroxide (H 2 O 2 ), the superoxide anion radical (O 2 -^) and the hydroxyl radical (OH^), during some stage of the four-electron reduction of O 2 to water, or following exposure to environmental factors (1,2). The unrestrained accumulation of these species gives rise to oxidative stress and can lead to cell damage, mutations or even death. One issue of the ROS detoxification concerns the decomposition of hydroperoxides. Most pro-and eukaryotic cells rely on the action of hemecontaining enzymes called catalases, which disproportionate H 2 O 2 to water and O 2 . Some eukaryotic cells may also use glutathione peroxidases to remove H 2 O 2 as well as organic and lipid hydroperoxides. In addition, research over the past decades has led to the characterization of a new family of peroxidases, collectively called 'peroxiredoxins' (Prxs) (3). They decompose organic hydroperoxides and H 2 O 2 by means of thiol-containing electron donors such as thioredoxin (Trx), AhpF, cyclophilin A, tryparedoxin, or, as recently reported, also the redox protein glutaredoxin (Grx) (4)(5)(6)(7)(8).
Haemophilus influenzae is an important, opportunistic, gram-negative human pathogen. The bacterium resides in the upper respiratory tract of humans where it generally grows aerobically, although facultative anaerobic growth is also possible (9). Besides oxidative stress from its aerobic respiratory metabolism, or as a result of the high O 2 tension at the nasopharynx, H.
influenzae may also be exposed to high levels of oxidants produced by the host's immune system, which uses the destructive power of ROS to eliminate bacterial infections (10).
Moreover, experimental data indicate that H. influenzae has to deal with H 2 O 2 secreted by peroxidogenic Streptococci (11). While the existence of a H 2 O 2 -inducible catalase (HktE) has been described in H. influenzae Rd, the enzyme seems to be redundant (12,13). As yet, no other 4 antioxidant enzyme has been identified that acts against hydroperoxides, making the ways in which the bacterium deals with hydroperoxide stress an interesting topic for future research.
Previously, we described a glutathione amide-dependent peroxidase from the phototrophic purple sulfur bacterium Chromatium gracile, capable of reducing both H 2 O 2 and tertbutylhydroperoxide (t-BOOH) at comparable high rates (14). By means of a BLAST-search using its deduced amino acid sequence, we were able to identify several homologs in different bacterial species, including one encoded by an open reading frame (HI0572) enclosed in the H.
influenzae Rd genome. The comparisons revealed the fusion of an N-terminal Prx region to a Cterminal Grx region, a unique feature typical for this novel family of homologs. This structure suggests that a thioltransferase reaction by the Grx moiety may be involved in the reduction of the Prx moiety (8,15,16). Grxs are small, ubiquitous thioltransferases that are specifically designed to use glutathione (GSH) for their reduction (17,18). They catalyze the reduction of protein disulfide groups and GSH-containing mixed disulfide groups either via a dithiol or monothiol mechanism (18).
In this paper, we expand our knowledge of the chimeric enzyme from H.  Final reaction volumes were 500 µL each. The decrease in NADPH absorbance was continuously monitored at 340 nm.  Another distinguishable fact is the apparent ability of type C Prxs to use Grx instead of Trx as electron donor (8). This especially merits to be mentioned since the C-terminal region of PGdx shares strong homology (52% identity, 69% similar) with Grx3 from E. coli. This Grxhomologous domain contains cysteine residues 180 and 183, arranged in a characteristic CPFC disulfide motif, and is coupled to the Prx region via a Gln rich stretch starting with a Pro.
Depicted in Figure 1 are the sequences of E. coli Grx3 and P. trichocarpa Grx. The latter reduces the poplar type C Prx, which is also included in the figure.
BLAST-searches of the H. influenzae PGdx against the Microbial Genomes database revealed numerous as yet uncharacterized homologs, the majority of them in microorganisms implicated in human disease (Fig.1). Similarity extends over the entire sequence, especially in the Prx region, and ranges from 60 to 95% identity and from 75% to 100% similarity.  (Table I, Fig.3).

Expression, Purification and Physical
When the non-reduced monomeric form was analyzed by ESI-MS we observed a peak with a mass of 26915.1-Da that is 305-Da higher than expected (Table I, Fig.3). This prompted us to investigate the possibility whether the enzyme was modified by glutathionylation during the overexpression and subsequent isolation from E. coli. Addition of GSH, indeed, did reduce the mass to that of the fully reduced monomer (Table I, Fig.3), a phenomenon that was not observed 12 when ascorbate was added (not shown). In order to determine the location of the GSH-moiety in the peptide chain, we performed a tryptic digest and analyzed the digest mixture of the reduced and non-reduced condition by mass spectrometry (spectra not shown). The mass spectrum of the reduced PGdx digest mixture contained a peak at 2316.2-Da, corresponding to the monoisotopic mass of the unmodified peptide T35-56R. Under non-reducing conditions, this peak was absent and replaced by a peak at 2621.2-Da, which agrees with the monoisotopic mass of the modified peptide T35-56R. To confirm this observation and the location of the modification, the peptide was subjected to collision-induced fragmentation mass spectrometry. As shown in Figure 4, the MS/MS spectrum was found to be consistent with the sequence of the peptide. This figure also shows an increment in mass of 305-Da after Cys49 (note the shift of the y"-ions), which points to the GSH-molecule being linked to the Cys49 residue.
PGdx Protects Supercoiled DNA from Oxidative Damage -Supercoiled DNA is prone to nicking when exposed to oxidative radicals such as those generated by the MCO system. Therefore, PGdx was tested for its ability to protect supercoiled DNA from degradation induced by the MCO system in the presence of either DTT, GSH or ascorbate (Fig.5). Absence of PGdx resulted in open coiled or nicked DNA, while addition of 10 mM EDTA completely inhibited degradation. PGdx, in combination with DTT or GSH, was successful in protecting the DNA.
When DTT or GSH was replaced by ascorbate as electron donor, the enzyme was unable to protect the DNA at a concentration that was sufficient to provide full protection against degradation when a thiol was present.

The GSH/GR/NADPH System Provides Electrons for PGdx-catalyzed Hydroperoxide Reduction
-We set up a reconstitution assay by which we demonstrated that PGdx can use the GSH/GR/NADPH-system and that the reduction of hydroperoxides depends on the presence of each component. The high activity observed after addition of GR (Fig.6.A) is due to the accumulation of its substrate, GSSG. In contrast, no discernible peroxidase activity was observed when GSH and GR were replaced by Trx and TR. For both H 2 O 2 (not shown) and t-BOOH For a kinetic analysis with GSH we chose t-BOOH as substrate, since its spontaneous reaction at physiological pH with GSH is less pronounced compared to H 2 O 2 . In order to avoid extensive background activity with GSH we also used a pH of 7.1 instead of pH 7.8, the established pH optimum for PGdx (not shown). Measurements revealed a sigmoidal substrate-velocity curve ( Fig.7.B). By fitting our data into the Hill equation we obtained an apparent Hill coefficient (n app ) of 2.9, indicating a phenomenon of strong cooperativity. K m,app and k cat /K m,app were 3.11 mM and 3.01 x 10 3 s -1 .M -1 , respectively. V max,app was 20.98 µmol/min/mg PGdx. The insets in Figure 7 by guest on March  Rouhier and colleagues previously described the Grx-dependent reduction of a poplar phloem Prx (8,15). They suggested a mechanism where the sulfenic acid of the oxidized Prx becomes reduced by formation of a disulfide linkage with the N-terminal cysteine residue of the CysXXCys motif from Grx. This disulfide bond then becomes reduced either through the monoor the dithiol mechanism characteristic for Grx activity (Fig.8). On the basis of our results, it appears that Cys49 has an affinity for GSH. Therefore, we propose another possible reaction mechanism for PGdx, in which the reduction of hydroperoxides is accompanied by the formation of a glutathionylated Prx-cysteine. The GSH-mixed disulfide is subsequently reduced by the action of the C-terminal Grx region, following a monothiol pathway. The mechanism can be summarized as follows, and is schematically given in Figure 8, Reaction 3 describes the dethiolation of the Prx region by Cys180 of the Grx region. In Reaction 4, PGdx becomes regenerated by GSH, forming GSSG, which in turn will be reduced by GR in Reaction 5. Glutathionylation of Prxs has already been mentioned in numerous cases (26)(27)(28) where it functions as a regulatory mechanism in which the Prx gets inactivated and protected against further oxidation of its active site cysteine into the more stable forms of sulfinic (Cys-

SO 2 H) or sulfonic acid (Cys-SO 3 H).
Other examples of such protection and regulation are already known to occur in protein tyrosine phosphatases where reactivation takes place with either GSH or Grxs (29)(30)(31). Dethiolation via Grxs follows a monothiol mechanism, requiring only one Cys residue of the redox active disulfide motif. Besides the inability of monocysteinic mutants of Trx to follow a monothiol pathway (32), Grxs are also 10 times more effective, on a molar basis, than Trxs in reducing GSH-mixed protein disulfides (31,33). In addition, the Cterminal Grx region of PGdx shares strong homology with Grx3 from E. coli, which has a higher activity as reductant of glutathionylated proteins than E. coli Grx1 and 2 (34). Hence, from an evolutionary point of view, the mechanism we propose provides a possible explanation for the fact that Grx-instead of Trx-homologs constitute these chimeric enzymes.
18 structures formed by some Prxs (36,37). Yet, no sigmoidal kinetics were mentioned in these cases. So, although the situation of a cooperative active oligomeric PGdx complex is possible, in which binding of GSH induces an increase in peroxidase activity of the other associated enzymes, we are not convinced this is indeed the case. Rather, we believe that other factors are responsible for the sigmoidal features of the velocity curve. Hence, the Hill coefficient obtained does not relate to the number of cooperative interacting sites, but is the intrinsic result of the sigmoidal velocity curve. Substrate depletion seems unlikely because the decrease in activity sets in when GSH-concentrations are still high and, moreover, the lowest GSH-concentration used was still 12,000 times that of the total enzyme concentration.
Kinetic measurements were performed with a non-reduced PGdx sample, thus containing both the glutathionylated monomer and the homodimer. Although sigmoidicity remained with a protein sample purified under reducing conditions (not shown), we cannot exclude the possibility that, in the presence of low GSH concentrations, formation of dimeric PGdx through oxidation prevails. Since participation in our proposed catalytic mechanism requires PGdx to remain monomeric, the homodimer formed needs to be reduced. Hence, low GSH concentrations may lead to a lower availability of active monomeric species and, therefore, to lower activity.
Given the proposed reaction scheme (Fig.8), the unproductive side reaction in the monothiol mechanism of a Grx may also be considered to be responsible for the sigmoidicity. When in H. influenzae ever characterized, it is also, to our knowledge, the first prokaryotic peroxidase effectively using GSH for its reduction, albeit in a manner different from eukaryotic GSH peroxidases. Besides a more in depth investigation of the catalytic role behind each of its three cysteine residues and their roles in catalysis, regulatory as well as physiological studies are currently undertaken to gain more insight into its in vivo functions. Studies with the separate regions of the PGdx enzyme will provide new insights into the catalytic mechanism and the interaction between its two regions. pBSK, plasmid BlueScript.