The Structure of the Organic Hydroperoxide Resistance Protein from Deinococcus radiodurans DO CONFORMATIONAL CHANGES FACILITATE RECYCLING OF THE REDOX DISULFIDE?*

The three-dimensional structure of the organic hydroperoxide resistance protein (OHRP) from Deinococcus radiodurans as determined using single crystal x-ray diffraction techniques is reported. Comparison of the structure with that obtained for OHRP from Pseudomonas aeruginosa reveals that the polypeptide chain of OHRPs can adopt two significantly different conformations (“in” and “out”) in the region of the active site disulfide moiety. It is postulated that the closed configuration is consistent with efficient catalysis of the reduction of organic hydroperoxides, whereas the open form is required for enzyme recycling. Comparison of the structures of OHRP and that of the osmotically induced protein C (OsmC) from Mycoplasma pneumoniae shows that OHRPs and OsmCs are structurally homologous, perhaps indicating related functions for the two families of proteins. The

The three-dimensional structure of the organic hydroperoxide resistance protein (OHRP) from Deinococcus radiodurans as determined using single crystal xray diffraction techniques is reported. Comparison of the structure with that obtained for OHRP from Pseudomonas aeruginosa reveals that the polypeptide chain of OHRPs can adopt two significantly different conformations ("in" and "out") in the region of the active site disulfide moiety. It is postulated that the closed configuration is consistent with efficient catalysis of the reduction of organic hydroperoxides, whereas the open form is required for enzyme recycling. Comparison of the structures of OHRP and that of the osmotically induced protein C (OsmC) from Mycoplasma pneumoniae shows that OHRPs and OsmCs are structurally homologous, perhaps indicating related functions for the two families of proteins.
The non-pathogenic, red-pigmented, extremophile Grampositive bacterium Deinococcus radiodurans (DEIRA) 1 is well known for its extreme resistance to ionizing and ultraviolet radiation, desiccation, and the presence of reactive oxygen species (1)(2)(3). The peculiar characteristics of DEIRA have made it the subject of intensive investigation, with DEIRA strain R1 being among the first organisms for which the complete genome sequence was made available (4). The extraordinary capability of DEIRA to repair double-stranded DNA breaks, which is perhaps linked to a unique packaging of DNA in cells of DEIRA, most likely lies at the heart of the bacterium's ability to recover from exposure to very high doses of radiation (1,5,6). However, its genome also contains myriad genes that code for proteins that defend cells against the reactive oxygen species that occur under conditions of oxidative stress, and it may be possible that this prevention of DNA damage caused by reactive oxygen species may also play a role in the extremophile nature of the bacterium.
Reactive oxygen species include the hydroxyl radical (OH ⅐ ), the superoxide radical (O 2 . ), hydrogen peroxide (H 2 O 2 ), and organic hydroperoxides (OHPs). Peroxiredoxins (Prxs) are thioldependent peroxidases, widespread in nature, that carry out the cellular detoxification of the latter two species (7)(8)(9)(10)(11)(12)(13). This family of enzymes includes the alkyl hydroperoxide reductase C22 (AhpC), recently shown to be the primary scavenger of endogenous H 2 O 2 in Escherichia coli. These enzymes accomplish detoxification by catalyzing the reduction of hydroperoxides to their corresponding alcohols using redox-active cysteines. In the mechanism for this reaction, one redox-active cysteine, called the peroxidatic cysteine (C P ), attacks the substrate and is converted to a cysteine-sulfenic acid. This acid is then attacked by the so-called resolving cysteine (C R ), resulting in the formation of a disulfide bond that must be re-reduced to reproduce the active enzyme Recently, a gene family that complements the Prx system has been discovered (14). This gene has been named the organic hydroperoxide resistance gene (Ohr), because its deletion in Xanthomonas campestris pv. phaseoli sensitizes this organism to organic hydroperoxides (mainly t-butylhydroperoxide and cumene hydroperoxide) but not to hydrogen peroxide or superoxides. The organic hydroperoxide resistance proteins (OHRPs) encoded by Ohr genes are generally 140 amino acids in length. They have high sequence homology with other members of the family (Fig. 1), are biologically active as homodimers (15,16), and contain two absolutely conserved cysteines in the polypeptide chain. The conserved cysteine occurring at around position 60 in the amino acid sequence is found in a very strictly conserved ACF motif at the end of a highly conserved stretch of ϳ16 amino acids. The conserved cysteine near amino acid position 121 is usually found in a VCPY motif embedded in a highly conserved chain of 12 residues (see Fig. 1). The environments of the two strictly conserved cysteine residues suggests that OHRPs function as a new type of 2Cys-Prx protein, and strong evidence, both biochemical (15,16) and structural (15), has emerged to support this hypothesis. From the available evidence, it is also clear that the two conserved cysteines form an intramolecular redox-active disulfide, with C P found toward the beginning of the amino acid sequence, and that the reaction catalyzed by OHRPs is 2RSH ϩ ROOH 3 RSSR ϩ ROH ϩH 2 O. The amino acid sequences of OHRPs also contain two absolutely conserved arginine residues near positions 15 and 128, and both biochemical and structural evidence (15) suggests that the former of these plays a crucial role in the enzymatic mechanism of OHRPs. The exact nature of the reducing equivalent required for recycling of the disulfide is not yet clear. However, the peroxidase activity of OHRP from X. fastidiosa appears to be supported only in the presence of dithiol-containing reducing agents such as dithiothreitol (DTT) or dihydroli-* 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.
The  (44) was based on the crystal structure of OHRP Dr superimposed with a ClustalW (45) alignment of the amino sequences of members of the OHRP and OsmC family members. The sequences are divided into three blocks: 1) OHRP family members; 2) main members of the OsmC family; and 3) a distant relative of the OsmC family for which the structure has been described recently (43)  poic acid (DHLA). Indeed, biochemical evidence suggests that OHRPs may actually be DHLA peroxidases (16).
Ohr genes are related to those of the OsmC family, with sequence identities of ϳ30% (17). OsmC family proteins also contain two absolutely conserved cysteines (near positions 55 and 120 in the polypeptide chain), and the environments of these cysteines (particularly the N-terminal one) ( Fig. 1) are very similar to those found in OHRPs. It might thus be expected that OHRPs and OsmCs have similar structures and functions. However, it is unclear as to whether this really is the case, given the results of two recent contradictory reports. The first of these notes the increased sensitivity of OsmC mutants of E. coli to t-butylhydroperoxide, suggesting that there are similar structures and functions for the two proteins (18). The second (17) reports that similar mutants of Pseudomonas aeruginosa show increased sensitivity only to osmotic stress (high salt concentration) and ethanol, and it suggests different functions for OHRPs and OsmCs.
The gene DR1857 encodes a 14.5-KDa (139 amino acids) protein (OHRP Dr ) that has been annotated as a member of the OHRP family ( Fig. 1) and predicted to be highly expressed by DEIRA (19). As part of a structural genomics project focusing on gene products that may be important in contributing to the extremophile nature of DEIRA strain R1, we now report the three-dimensional structure of OHRP Dr as determined using single-crystal x-ray diffraction techniques. The structure, when compared with that recently obtained for OHRP from P. aeruginosa (OHRP Pa ), allows a description of conformational differ-ences between the OHRPs in their redox-active site reduced and oxidized forms. This provides an explanation of how the enzyme might be recycled once neutralization of organic hydroperoxides is achieved. We also compare the structure of OHRP Dr to that recently reported for OsmC from Mycoplasma pneumoniae (OSMC Mp ) and comment on the structural and possible functional similarities of the two protein families.

MATERIALS AND METHODS
Cloning, expression, and purification of wild-type OHRP Dr modified by the addition of an N-terminal hexahistidine tag was accomplished as described elsewhere (20). The selenomethionine (SeMet)-substituted protein, the structure of which is described here, was then produced using established protocols (21) and purified in a similar manner, with the exception that 10 mM ␤-mercaptoethanol was added to the buffers used in the procedure. The pure protein fraction used for crystallization was then dialyzed against a storage buffer consisting of 20 mM Tris-HCl, pH 7.2, 200 mM NaCl, and 5 mM DTT and then concentrated to 10 mg/ml and stored at Ϫ20°C. Electrospray mass spectrometry confirmed the complete substitution of Met by SeMet Crystals of the SeMet-substituted protein appeared after several days from hanging drops containing 2 l of the protein stock solution and 2 l of a reservoir solution comprising 31-35% polyethylene glycol 4000, 0.1 M Tris-HCl, pH 8.4, 0.2 M MgCl 2 , and 0 or 2% (v/v) glycerol. The drops were equilibrated at 20°C against 1 ml of the reservoir solution. Structure solution was carried out using the single-wavelength anomalous diffraction (SAD) technique (22). X-ray diffraction data from a single crystal of the SeMet protein (Table I) were collected at 100 K, and ϭ 0.933 Å on the beamline ID14-EH2 of the European Synchrotron Radiation Facility. The same contiguous 180°segment of data was measured twice using 0.5°oscillation images and, initially, integrated to the limit of the inscribed circle on the ADSC Q4 detector (d min ϭ 2.3 Å) using the program MOSFLM. The data were then scaled and merged using SCALA, and the structure factors and anomalous differences were obtained by using the program TRUNCATE. This data set was used for both selenium atom substructure determination and initial phase calculations. A second data set was then derived, this time integrating diffraction spots over the entire area of the detector only for the first 180°of data collected. This data set includes reflections to a resolution of 1.9 Å and was used in both the automatic model building and refinement procedures outlined below. The data scaling and merging procedure clearly showed the crystals to be monoclinic, with cell dimensions of a ϭ 45.41, b ϭ 56.60, c ϭ 49.46 Å, and ␤ ϭ 90.27°, and an examination of systematic absences in the intensity data indicated the space group to be P2 1 . A calculated Matthew's coefficient of 2.2 Å 3 Da Ϫ1 and a theoretical solvent content of ϳ42% are consistent with the asymmetric unit containing the biologically active homodimer. Initial determination of the positions of the selenium atoms in the crystal was then carried out with anomalous difference data using direct methods as implemented in the program SHELXD (23,24). The best solution resulted in a (E obs , E calc ) correlation coefficient of 27.1 for all of the reflections used in the search procedure and clearly indicated that the asymmetric unit contains the four selenium sites expected for the presence of a dimer in the asymmetric unit. Once it was clear that the selenium atom substructure could be found readily, the programs SHARP/AutoSharp (25) were used for phase determination to d min ϭ 2.3 Å. These initial phases were then extended to the 1.9-Å resolution of the second data set described above by using solvent flattening, histogram matching, and non-crystallographic symmetry averaging in the program DM. In the first round of the averaging procedure, the initial averaging matrices were found using the program GETAX as implemented in AutoSharp, and an averaging mask was determined automatically. The phases resulting from this procedure allowed the building of a substantial part of the structure automatically using Arp/Warp (26), and this structure was used as a basis for improving the noncrystallographic symmetry matrix and building a more suitable averaging mask. Using these, the phase extension procedure was repeated, and the subsequent run of Arp/Warp allowed the building of the main chain of 244 residues in nine separate chains, giving a connectivity index of 0.93. A more complete model was then built using the program O (27), and this model provided the basis for a refinement protocol using the program REFMAC5 interspersed with rounds of manual rebuilding, during which further residues were added to the model, and water and glycerol moieties were included. Statistics showing the results of this refinement protocol are presented in Table I. Unless referenced specif-ically, all of the programs that were used during the data collection, structure determination, and refinement procedures are from the CCP4 package (28).

RESULTS AND DISCUSSION
Structure of OHRP Dr -The structure of OHRP Dr , as presented in our crystal form (Fig. 2), is very similar to that reported recently for OHRP Pa (15), with which OHRP Dr shares 51% amino acid sequence identity. As is the case with the structure of OHRP Dr , the asymmetric unit of OHRP Pa contains the biologically active homodimer. A superposition of the structures of OHRP Dr and OHRP Pa aligns them with a 1.14-Å root mean square deviation for 267 common C ␣ atoms in the dimer (Fig. 3). The structure of each OHRP Dr monomer comprises two distinct subdomains. The N-terminal domain is small (residues 1-35) and consists of ␤-strands ␤1-␤3 (see Fig. 1 for secondary structure assignment) folded into a ␤-sheet. In the larger (residues 47-139) C-terminal domain, helices ␣1 and ␣2 (the latter severely kinked) are stacked on a ␤-sheet formed by strands ␤4 -␤6. To form the functionally active homodimer, ␣1 from one monomer is inserted between the N-and C-terminal domains of the second subunit and vice versa. This results in an elliptically shaped ␤-␣-␣-␤ sandwich structure in which the N termini of both ␣1 helices and the C termini of the ␣2 helices are stacked between a pair of six-stranded, concave ␤-sheets. In both monomers, the sulfur atoms of the conserved cysteine residues are found adjacent to each other, with C P situated near the midpoint of ␣1 and the resolving cysteine at the kinking point of ␣2.
Comparison of the Structures of OHRP Dr and OHRP Pa -Despite the global similarities of the structures of OHRP Dr and OHRP Pa , they differ in two noteworthy respects. The first of these differences concerns the conformation of the disulfides in the active sites. In both OHRP Dr and OHRP Pa , C P and the resolving cysteine (Cys57 and Cys121, OHRP Dr numbering) are found in the same polypeptide chain, and the two redox-active disulfides are located on opposite sides of the dimer close to monomer-monomer interfaces. In the structure of OHRP Pa , the disulfide groups are clearly in the reduced (S ␥ H-H ␥ S) state. In OHRP Dr , the S ␥ 57-S ␥ 121 distance is 2.8 Å in both active sites in the dimer. These distances are thus intermediate between the 2.2 Å that is expected for a fully oxidized disulfide bond and the 3.7 Å that one might expect for a fully reduced disulfide.
Although a reducing agent (DTT) was added to the buffer in which the SeMet protein was stored, it is well documented that its efficacy drops off over time (see Refs. 29 and 30 for examples of this). Thus, because DTT was not present in the reservoir solution used in the crystallization procedure, it is very likely that the crystal used here for data collection originally contained, in the large majority, molecules of OHR-P Dr in its oxidized (S ␥ -S ␥ ) form. One explanation of the intermediate S-S distances observed is that they are the time-and space-averaged results of a mixture of oxidized and reduced disulfides. However, the electron density in the crystal (Fig. 4) suggests that there is only a single major conformation for the disulfides. Thus, a more plausible reason for S-S distances observed in the structure of OHRP Dr is that they are the result of the photoreduction of the S-S bond due to x-ray irradiation during data collection. This phenomenon is more commonly known as radiation damage (31)(32)(33)(34). The S-S distances that we observed in the structure of OHRP Dr are almost exactly that calculated theoretically for disulfide radical anions (RS-SR . ) (35), and the disulfide bonds, although weakened, are not broken. The structure of OHRP Dr that we report here is thus representative of that for an OHRP in its oxidized (i.e. disulfide bond-containing) form.
It should be noted, however, that the difference electron density (mF obs Ϫ DF calc , ␣ calc ) maps (Fig. 4), which were calculated at the end of the refinement procedure, show features close to the disulfide groups that suggest the presence of an additional low occupancy conformation for the disulfides. A positive peak in the difference density clearly shows this conformation to be a more open configuration of the redox disulfide and suggests that the crystal also contains a small percentage of molecules with an open disulfide bond.
The second major difference between the crystal structures of OHRP Pa and OHRP Dr is the fact that the loop containing an arginine residue (Arg-18 in the sequence of OHRP Pa and Arg-15 in that of OHRP Dr ) conserved in the sequences of all bacterial OHRPs adopts quite different conformations. In the crystal structure of OHRP Pa (blue in Fig. 3), the loop from subunit B is closed in and points toward the subunit A active site disulfide and vice versa. The side chain of the arginine forms an inter-subunit salt bridge with a glutamic acid residue conserved in the sequences of both OHRPs and OsmCs (Glu-50 in OHRP Pa ) and approaches to within 3.4 Å of the S␥ of C P . Lesniak and colleagues (15) have reported that an R18Q mutant of OHRP Pa possesses a substantially compromised ability to metabolize hydroperoxides. The mechanism that they therefore propose for peroxide reduction by OHRPs (15) is essentially that of other 2Cys-peroxiredoxins (7), as the conformation of the arginine-containing loop in the structure of OHRP Pa places this conserved residue in an ideal position to stabilize the ionized peroxidatic cysteine that attacks the peroxide. This mechanism also neatly explains why OHRPs are biologically active as homodimers.
Based on sequence alignments (Fig. 1), it is highly probable that Arg-15 in OHRP Dr plays the same crucial role in the stabilization of C P as does Arg-18 in OHRP Pa . However, in both subunits in the crystal structure of OHRP Dr , the loop containing Arg-15 has flipped away from the catalytic disulfide (Fig.   FIG. 3. Comparison of  3). Additionally, because the electron density for the side chain of the arginine residue is no longer visible (this residue has been truncated at the C ␤ atom in both monomers of OHRP Dr ), it must therefore be directed toward the solvent region in the crystals. It is thus clear that, in OHRPs, this loop can adopt two significantly different conformations ("in" and "out"). In contrast to the structure of OHRP Pa , our crystal contains molecules of OHRP that are not fully reduced. It thus seems reasonable to suggest that the in conformation of the loop occurs when the active site disulfide is open (i.e. reduced) and that the out conformation occurs when it is closed (i.e. oxidized).
Wood and colleagues noted (36) that the 2Cys-Prxs they investigated also have two stable conformations that appear to be correlated with the catalytic cycle. Those conformations that contain a reduced disulfide are found with what is termed a "fully folded" conformation, whereas for those conformations in which a disulfide bond formation is observed, there is a local unfolding of the protein molecule close to the active site. The in and out conformations observed for OHRPs would appear to be analogous to these conformations, and it is probable that, in vivo, the arginine-containing loop in OHRPs acts a lid that closes when the redox disulfide is in its reduced, active form and opens when the disulfide is oxidized.
In the fully folded (i.e. reduced) conformation for the 2Cys-Prxs examined by Wood and colleagues (36), the two cysteine sulfur atoms that comprise the redox active disulfide are Ͼ10 Å apart, regardless of whether disulfide bond formation is inter or intramolecular. The local unfolding observed for these enzymes thus appears to be required to allow disulfide bond formation. Comparison of the structures of OHRP Pa and OHR-P Dr reveals that, similarly as with thioredoxin-like proteins (30,(37)(38)(39), only minor structural changes around the disulfide active sites would be required for disulfide bond formation. Thus, an interchange between the two different conformations observed for OHRPs is not needed to allow disulfide bond formation to occur. It therefore seems reasonable to suggest that the flipping of the arginine loop, which is observed in the structures of OHRPs, is necessary either for involvement in signaling processes similar to those attributed to other 2Cys-Prxs (40 -42) or for allowing the recycling of the redox disulfide that completes the catalytic cycle. Support for the latter idea comes from the crystal structure of tryparedoxin from Trypanosoma brucei (38). Here, the side chain of a tryptophan residue forming a lid over the active site is flipped out, and it is proposed that the resulting larger active site cleft might aid the interaction of the redox components in the trypanothione peroxidase pathway.
For OHRPs, disulfide bond recycling is most likely carried out by dithiol-containing reducing agents, with DHLA being a good candidate for performing this role (16). The in conformation of the arginine loop results in the formation of a coneshaped pocket that would appear to allow the funneling of an alkylperoxide substrate toward the redox disulfide (Fig. 5a). If OHRPs permanently maintained this conformation, then, once the reaction products have diffused away, a dithiol-containing reducing agent could perhaps attack the newly formed disulfide bond via this channel. In the crystal structure of OHRP Pa in which the disulfide is in its reduced form, this cleft contains a molecule of DTT. One of the sulfur atoms of DDT approaches to ϳ3 Å from the S␥ atom of the peroxidatic cysteine and, at first sight, this seems to explain how a dithiol might get close enough to the disulfide to allow its re-reduction. However, when the disulfide is fully oxidized, the much shorter S-S distance means that C P is likely to retreat further into the cleft, and access to it by a dithiol become more difficult. A more open conformation around the active site would therefore seem to be essential to allow re-reduction of an oxidized disulfide to occur more readily.
The out conformation of the arginine loop, which is seen for the majority of OHRP molecules in the structure of OHRP Dr , would appear to allow this greater access. Movement of the arginine loop creates a large, highly hydrophobic pocket, the surface and shape of which is eminently suited to the binding of molecules such as DHLA. In fact, manual modeling of DHLA into this newly created pocket is a very straightforward procedure (Fig. 5b). The results of this procedure show that, by . This may allow molecules such as DHLA, which is modeled in stick conformation in panel b, to approach close to redox disulfide in its oxidized state and, thus, effect its re-reduction. In panel a the DTT molecules found in the active site cavity of OHRP Pa are shown in stick representation. For both panels, any glycerol or water molecules moiety found in the active site cavity were not included in the surface calculations. occupying the space vacated by the side chain of the flipped-out conserved arginine, thiol groups of molecules such as DHLA can approach close to redox disulfide in its oxidized state and thus effect its re-reduction. This provides further evidence that the out conformation for the arginine loop seen in the structure of OHRP Dr may be required for the recycling of OHRPs in general.
Comparison of the Structures of OHRPs and OsmCs-Recently, the crystal structure of a protein annotated as an osmotic stress-induced protein, OsmC from Mycoplasma pneumoniae, has been reported (43). Although OSMC Mp is one of the more distant relatives of the OsmC family (Fig. 1), this allows a comparison of the structures of proteins from the related Ohr and OsmC gene families that may shed more light on the function of the latter.
As can be seen from Fig. 6 OSMC Mp is homodimeric and has a tertiary structure very similar to that of OHRPs. A superposition of the structures of OHRP Dr and OSMC Mp results in a root mean square deviation in positions of 1.7 Å for 185 common C ␣ atoms within a 3.8-Å cutoff. The conserved cysteine residues (see Fig. 1) in OsmC gene products superimpose almost exactly with those found in the sequences of OHRPs. For the main members of the OsmC family, the amino acid sequence environment in which the cysteine residues are found indicates that they form a redox active site rather than a structural disulfide pair. This site, coupled with the very close similarities between the OHRP and OsmC structures compared here, suggests that, like OHRPs, the main members of the OsmC family also function as 2Cys-Prxs with the N-terminal cysteine playing the peroxidatic role. Thus, structural evidence supports the observations of Conter and colleagues (18) that OsmC-deficient E. coli is hypersensitive to t-butylhydroperoxide. However, the catalytically important argi-nine (Arg-15 in OHRP Dr ) that is absolutely conserved in the amino acid sequences of OHRPs is not conserved in the sequences of OsmCs (Fig. 1). Thus, if OsmCs do act as 2Cys-Prxs, they must use a different residue to help stabilize the ionized peroxidatic cysteine formed during the catalytic cycle. Examination of the amino acid sequences of OSMCs ( Fig. 1) reveals two possibilities. The first is a conserved arginine found at around position 39 in the amino acid sequences of the main members of the OsmC family. Assuming that the main members of the OsmC family exhibit the same type of the structural homology with OHRPs as does OSMC Mp , then, in the structures of OsmCs, this residue will be found on the loop joining the ␤-strand S3 and the helix (H1) containing the peroxidatic cysteine. In the structures of both OHRP Pa and OHRP Dr , this loop from one monomer helps to shape the active site of the second monomer in the homodimer. Only a small change in the conformation of the loop would be needed to allow an arginine residue to be in the correct position to interact with both the peroxidatic cysteine and the glutamic acid conserved on the sequences of both OHRPs and OsmCs. A more attractive possibility, however, is that any ionized peroxidatic cysteine formed during the catalytic cycle of OSMCs is stabilized by a residue that occurs in the same structural motif as is the case for OHRPs and that such an amino acid exists in the sequence of OSMCs. A lysine residue (Lys-17 in the sequence of OSMC from Deinococcus radiodurans) appears to be highly conserved throughout the main members of the OSMC family. Based on the apparent structural homology of OHRPs and OSMCs, this lysine residue, in the structure of the latter, is likely to be found in the OSMC equivalent of the OHRP arginine loop described above and would thus be an excellent candidate to be the residue in OSMCs that stabilizes any ionized peroxidatic cysteine.
Although it is clear that the main members of the OsmC family act as peroxidases, it is worth noting that the amino acid sequence of OSMC Mp , (see Fig. 1) contains very little similarity with those of other OsmCs or OHRPs. In fact, the annotation of OSMC Mp as an OsmC family member appears to be based solely on the disposition of the cysteine residues. Thus, how this particular enzyme might catalyze peroxide reduction or even whether it acts a peroxidase remains unclear.
Summary-We have reported here the crystal structure of the organic hydroperoxide resistance protein from D. radiodurans. Although the results of our analysis show a global three-dimensional structure very similar to that recently reported for OHRP from P. aeruginosa, they also clearly indicate conformational differences between the enzyme with its redox disulfide in its reduced, active form and its oxidized, inactive form. The conformation in the latter form results in the formation of a large cleft close to the active site disulfide that appears to facilitate the recycling of the redox active site disulfide by dithiol-containing reducing agents. A comparison of the structures of OHRPs and those of a distant OsmC family member show that the OHRP and OsmC families of proteins are highly homologous structurally and provide further evidence that the latter are, like OHRPs, also 2Cys-peroxidases. Sequence alignments of OHRP and OsmC family members, coupled with the high structural homology of the two systems, suggests that one of two conserved, positively charged amino acids in the sequences of the latter family may stabilize the ionized peroxidatic cysteine that is an essential part of the mechanism by which alkyl hydroperoxides are metabolized by peroxidases. While this manuscript was under review, the crystal structure of OSMC from E. coli was reported (47) along with biochemical evidence that OSMCs do act as 2Cys-Prxs. As expected, the overall structure of this E. coli OSMC is very similar to those of OHRP Dr , OHRP Pa , and OSMC Mp . Based on the spatial arrangement of amino acid residues close to the catalytic redox center, Arg-39 was tentatively identified as the residue involved in the stabilization of C P . However, no biochemical evidence to support this hypothesis is currently available, and we await the results of such studies with interest. The report on the structure and function of E. coli OSMC provides yet more evidence that OSMCs can act as 2Cys-peroxidases, and the initial finding (17) that the expression of OsmC gene products is induced under conditions of osmotic stress but not in the presence of oxidants is, thus, all the more intriguing.