Structural and functional features of the Escherichia coli hydroperoxide resistance protein OsmC

Abstract

The osmotically inducible protein OsmC, like its better-characterized homolog, the organic hydroperoxide protein Ohr, is involved in defense against oxidative stress caused by exposure to organic hydroperoxides. The crystal structure of Escherichia coli OsmC reported here reveals that the protein is a tightly folded domain-swapped dimer with two active sites located at the monomer interface on opposite sides of the molecule. We demonstrate that OsmC preferentially metabolizes organic hydroperoxides over inorganic hydrogen peroxide. On the basis of structural and enzymatic similarities, we propose that the OsmC catalytic mechanism is analogous to that of the Ohr proteins and of the structurally unrelated peroxiredoxins, directly using highly reactive cysteine thiol groups to elicit hydroperoxide reduction.

Pathogenic microorganisms have developed complex strategies to detoxify and repair damage caused by reactive oxygen species. These include organic hydroperoxides produced during bacterial aerobic respiration, as well as by the host immune system cells as a defense mechanism against the infectious bacteria. Organic hydroperoxides are highly toxic and are capable of modifying and inactivating a variety of cellular macromolecules (Niki 1992; Mongkolsuk et al. 1998). Because of its inherently high reactivity, hydroperoxide metabolism to less noxious byproducts is necessary for microbial survival and proliferation in the infected host, and proteins involved in hydroperoxide protection play important roles in host–pathogen interactions (Mongkolsuk et al. 1998).

The Ohr/OsmC proteins were identified as a family of bacterial proteins involved in the detoxification of organic hydroperoxides (Mongkolsuk et al. 1998; Atichartpongkul et al. 2001). They share no sequence homology to other prokaryotic or eukaryotic proteins. The sequence identity within each of the Ohr and OsmC subfamilies is between 40%–70%, and ~20% between the two subfamilies (Fig. 1A ▶). Their more conserved carboxy-terminal region contains two invariant cysteine residues that play a critical role in hydroperoxide detoxification (Lesniak et al. 2002).

Figure 1.

(A) Structure-based sequence alignment of E. coli OsmC and P. aeruginosa Ohr. Identical amino acids within the OsmC and Ohr subfamilies are marked in red; conservative substitutions, in orange; and the semiconserved residues, in yellow. The positions of the catalytically active cysteines are indicated with circles, enzymatically equivalent arginines with diamonds, and the identical residues in the active sites with stars. (B) Topology diagram of the OsmC monomer. β-Strands (S1–S6) are in blue; the 3₁₀-helix (H1) and α helices (H2–H4) are in light purple.

We recently determined the crystal structure of Pseudomonas aeruginosa Ohr and demonstrated that in the presence of the reducing agent dithiothreitol (DTT), Ohr displays hydroperoxidase activity toward both inorganic and organic hydroperoxides (Lesniak et al. 2002). The Ohr catalytic mechanism is similar to that of the structurally unrelated peroxiredoxins, where both protein families directly use highly reactive cysteine thiol groups to elicit hydroperoxide reduction (Lesniak et al. 2002; Cussiol et al. 2003). Escherichia coli lacking OsmC is sensitive to exposure to hydrogen peroxide and tert-butylhydroperoxide, suggesting that like its homolog Ohr, OsmC is involved in defense against oxidative stress (Conter et al. 2001). Here, we describe the first crystal structure of an OsmC subfamily member, analyze its activity toward both organic and inorganic peroxide substrates, and discuss the implications of the structural data for the substrate specificity of this class of peroxidases.

Results and Discussion

The OsmC structure

OsmC is a domain-swapped dimer, composed of two monomers arranged in a head-to-tail orientation (Fig. 2A ▶). The two active sites, each containing two conserved cysteines Cys59 and Cys125 (marked • in Fig. 1A ▶) are located in the region of the dimer interface. The major contacts between the two monomers are due to the interactions of two long helices (H2) at the center of the hydrophobic core and two large antiparallel β-sheets wrapped around the central helices. The OsmC monomer is composed of two domains (Figs. 1B ▶ and 2B ▶). The larger carboxy-terminal subdomain contains an antiparallel β-sheet composed of strands S4, S5, and S6 and three α-helices (H2, H3, H4), whereas the smaller amino-terminal subdomain consists only of a three-stranded antiparallel β-sheet (strands S1, S2, and S3) and one 3₁₀ helix (H1).

Figure 2.

(A) Structure of the OsmC dimer. One monomer is in green; the other is in blue. The active site cysteines are shown in red. (B) Structure of the OsmC monomer highlighting the amino-terminal subdomain (red) and the carboxy-terminal subdomain (green). The locations of Cys59 (blue circle), Cys125 (purple circle), and Arg39 (orange circle) are also marked. (C) Superposition of the structures (main chain trace) of the OsmC (green) and Ohr (yellow) with marked secondary structural elements. (D) Molecular surface rendering of the OsmC and Ohr. The attacking Cys59 in OsmC and Cys60 in Ohr (shown in red) lie at the bottom of the active site pockets partially surrounded by hydrophobic residues (green).

The OsmC dimer has two active sites located on opposite sides of the enzyme, with each active site containing the conserved Cys59 and Cys125 contributed by the same monomer. Cys59 is surrounded by Pro126, Val127, Leu 66, and Arg39 side chains, whereas Cys125 is located next to His56, Phe60, Pro126, Ser128, Val127, Ala121, and Ala63.

Structural comparison between OsmC and Ohr

Overall, both the tertiary and quaternary structures of Ohr and OsmC are very similar. Superposition of the two molecules results in a root-mean-square deviation (rmsd) of 1.47 Å for 105 α carbon positions (Fig. 2C ▶). The main structural differences between the two proteins involve rearrangements of secondary structural elements away from the active sites. The seven amino-terminal amino acids in Ohr form an extended coil that wraps around the other monomer, whereas in OsmC they are structured into a longer, more extended β-strand S1. The second difference between the two structures involves β-strands S1 and S2, which are longer in OsmC than in Ohr. A third difference is that helix H3 (OsmC amino acids 111–123) is positioned further away from the enzyme’s core region in OsmC than in Ohr (displacements in this region ranging from 3.45 Å to 2.78 Å). Fourth, the loop region (OsmC amino acids 131–134) linking helix H4 and β-strand S6 in the proteins is substantially more tightly folded and compact in OsmC. Finally, the strand S6 of OsmC is longer than the one found in Ohr.

The active site residues (indicated with ♦ in Fig. 1A ▶) of both enzymes are conserved and similarly positioned, including Pro48 (Pro49 in Ohr), Phe60 (Phe61 in Ohr), Glu49 (Glu50 in Ohr), and Ser128 (Ser127 in Ohr). One exception is the orientation and position of the side chain of Arg39, which in OsmC hydrogen bonds with Cys59 (S) (3.2 Å), but does not form a salt bridge with Glu49. In contrast, in Ohr, the functionally equivalent Arg18 both hydrogen bonds with the corresponding Cys60 (3.4 Å), and forms a salt bridge with Glu50. The Arg18–Cys60 interaction has been shown, using site-directed mutagenesis, to be important for optimal catalytic activity of Ohr, presumably by lowering the pK_a of the attacking Cys60 thiol (Lesniak et al. 2002). It is interesting that the proteins from the OsmC subfamily lack Arg18, which is fully conserved in the Ohr subfamily (marked with ♦ in Fig. 1A ▶), but have created the same spatial arrangement of catalytic residues by using Arg39, which is in turn completely conserved in the OsmC subfamily. Position 18 (Arg in Ohr; Gly in OsmC) is located in the beginning of strand S2, whereas position 39 (Gly in Ohr; Arg in OsmC) is at the end of the nearby helix H1, therefore the side chain of an arginine at any of these two locations would easily fall within the hydrogen-binding distance to the attacking Cys59 and Cys60 in OsmC and Ohr, respectively.

Enzymatic activity of OsmC toward peroxide substrates

Recombinant OsmC was tested for peroxidase activity toward both inorganic (H₂O₂) and organic hydroperoxide (CHP) using a colorimetric FOX assay, as previously described (Lesniak et al. 2002). The results presented in Fig. 3A–D ▶ show that H₂O₂ and CHP can both serve as OsmC substrates. Nevertheless, the enzyme’s ability to turn these compounds over differs substantially. H₂O₂ is a relatively poor OsmC substrate: 10 μM of OsmC can remove ~13 μM of H₂O₂ in a minute, whereas only 3 μM of OsmC can metabolize >2000 μM CHP during the same time period. In addition, it takes >700 sec for 1 μM OsmC to process ~20 μM H₂O₂, but only 1–2 sec for the equivalent amount of CHP. These results are similar to the ones previously reported by us for Ohr, and are consistent with an apparent K_m value for H₂O₂ in the 200–800 μM range, and an approximately two-order of magnitude lower K_m for CHP, demonstrating that both proteins evolved to detoxify organic peroxide substrates and not H₂O₂.

Figure 3.

FOX enzymatic assays document that OsmC metabolizes (A, B) cumene hydroperoxide (CHP) and (C, D) H₂O₂ (HP) in a protein-concentration and time-dependent manner in the presence of DTT. Each point represents a mean measurement value from three separate experiments and the bars represent standard errors.

Biological implications

The overall structural similarities between Ohr and OsmC, and in particular the structural conservation of the active sites, including the location of the activating arginine, imply that the two proteins share a common enzymatic mechanism. Therefore, we propose that the OsmC protein subfamily metabolizes organic hydroperoxides by directly using cysteine thiol groups as the catalytic centers. The overall enzymatic scheme can be represented as follows:

1. 
2. 
3. 

Briefly, the activated Cys59 reacts with peroxide (ROOH) and is oxidized to a cysteine sulfenic acid (Cys59–SOH) intermediate, whereas the peroxide is reduced to alcohol (ROH). Second, the Cys59–SOH condenses with the Cys125 thiol located in the immediate vicinity, leading to the formation of an intramolecular disulfide bond and release of water. Finally, the oxidized OsmC is regenerated back to its enzymatically active, reduced state using an unidentified endogenous reductant R(SH)₂.

Cussiol et al. (2003) have recently provided evidence that purified Xylella fastidiosa Ohr can be reduced back to its enzymatically active state by dihydrolipoic acid (DHLA), which would suggest that the protein is a dihydrolipoic acid peroxidase. This would also probably be true for the E. coli homolog. E. coli genome encodes enzymes necessary for the synthesis of DHLA: dihydrolipoamide succinyltransferase (GenBank ID: AAG55051), dihydrolipoamide dehydrogenase (GenBank ID: AAG54420.1), and lipoate synthase (GenBank ID: AAG54962.1).

Some bacteria express both Ohr and OsmC proteins, despite the fact that they are structurally and functionally similar. There are two plausible, and not mutually exclusive, explanations for this. First, it is possible that each protein resides in a distinct subcellular location, which may be beneficial, if, for example, one is primarily responsible for detoxification of exogenous peroxides produced by the host immune system, whereas the other inactivates the peroxide byproducts of bacterial metabolism. Second, it is possible that Ohr and OsmC have evolved to target different subsets of substrates. The OsmC structure suggests that the second possibility is probably true. Although both OsmC and Ohr proteins show a strong preference for the metabolism of organic hydroperoxides (tert-butyl and CHP) over an inorganic H₂O₂, the active site of OsmC and the surrounding surface region shows different distribution of hydrophobic residues from that of Ohr (Fig. 2D ▶). In addition, the overall three-dimensional shape of the active site cavities differs between the two enzymes. Whereas the active site of Ohr is relatively wide, the one in OsmC is more elongated and narrow (Fig. 2D ▶). Thus, OsmC and Ohr may be responsible for the detoxification of structurally different, albeit still hydrophobic peroxide substrates. Understanding of the precise biological role of Ohr and OsmC would require identification of the endogenous substrates for each enzyme, and careful characterization of the kinetic parameters for each substrate. Finally, unlike other peroxidases, both the Ohr and OsmC hydroperoxide reductase subfamilies are present only in bacteria (most of which are pathogenic to plants and humans), and therefore, could potentially represent viable therapeutic drug targets.

The atomic coordinates and structure factors have been deposited in the Protein Data Bank under the code 1QWI.

Materials and methods

Protein preparation, data collection, structure determination, refinement, and enzymatic assays

OsmC from E. coli strain O157:H7 EDL399 was cloned, expressed, and purified as described for P. aeruginosa Ohr (Lesniak et al. 2002). Purified OsmC was concentrated to 20 mg/mL in a buffer containing 5 mM Tris at pH 8.0 and 5 mM DTT and was crystallized at 4°C using the hanging drop vapor diffusion method against a reservoir containing 0.2 M ammonium sulfate, 30% PEG 5K MME and 0.1 M MES at pH 6.0. The crystals were flash-frozen in a cryoprotectant buffer containing 20% glycerol (v/v). Diffraction data were collected at the Brookhaven National Synchrotron Light Source beamline X9A. Attempts to determine the structure using molecular replacement and Ohr as a model were unsuccessful. The structure of the selenomethionine-derivatized OsmC was therefore solved using the multiwavelength anomalous dispersion (MAD) method (Hendrickson 1991) to 2.0 Å resolution (Table 1). Oscillation images were integrated, scaled, and merged using Denzo and Scalepack (Otwinowski and Minor 1997). Solve (Terwilliger and Berendzen 1999) was used to locate the eight selenium atoms (two per monomer) as well as to calculate the initial phases. Model contained four OsmC monomers in the asymmetric unit. Solvent flattening and noncrystallographic symmetry averaging was performed with the program Resolve (Terwilliger 2000) to improve the experimental electron-density map. The automatic chain tracing procedure of wARP (Perrakis et al. 1999) was used to build 508 of 568 OsmC residues, whereas the program O (Jones et al. 1991) was used to complete the tracing and sequence assignment. The final model was refined against a native data set collected to 1.8 Å through a conventional least squares algorithm with CNS (Brunger et al. 1998). The final model contains 142 amino acids of 143 (the amino-terminal Met residue is post-translationally removed in BL21 cells) and is refined to an R-factor of 22%, with restrained temperature factors. Stereochemical analysis using Procheck of the CCP4 Package (CCP4, 1994) revealed main chain and side chain parameters better than or within the typical range of values for protein structures determined at corresponding resolutions. None of the OsmC residues fell in the disallowed region of the Ramachandran plot. Molecular graphic figures were created with MolScript (Kraulis 1991), Raster3D (Merritt and Bacon 1997), and Grasp and POV4GRASP (Nicholls et al. 1991). Metabolism of peroxides by OsmC was measured using the ferrous oxidation in xylenol (FOX) method as previously described for the Ohr protein (Lesniak et al. 2002).

Table 1.

Summary of crystallographic analysis

	Infection	Peak	Remote	Native
Wavelength (Å)	0.9795	0.9793	0.9770	0.9790
Resolution (Å)	2.0	2.0	2.0	1.8
Completeness (%)	99.7	97.5	99.8	N/A
	(98.7)c	(94.1)	(99.7)
Anomalous
Completeness (%)	97.7	97.5	97.1	99.0
	(94.6)c	(94.1)	(94.9)	(98.3)
Rmergea	3.9%	3.8%	4.0%	4.8%
I/σI	17.3	17.4	14.9	11.0
Redundancy (fold)	2.1
F.O.M. (SOLVE)	0.42
F.O.M. (RESOLVE)	0.56
Space group	P21
Cell dimensions (Å)	a = 65.75	b = 64.23	c = 67.59
	α = 90.00	β = 92.62	γ = 90.00
Refinement
Resolution (Å)	500–1.8
Reflections
Working	93,520
Test	4810
Nonhydrogen atoms	4656
Number of waters	504
Rcrys/Rfree	22%/24%
R.m.s. deviationsb
Bonds (Å)	1.36
Angles (°)	0.006

^a R_merge = ∑|I − 〈I〉|/∑I, where I = observed intensity, 〈I〉 = average intensity obtained from multiple observations of symmetry related reflections.

^b r.m.s. deviations in bond lengths and angles are the respective r.m.s. deviations from ideal values.

^c ( ) Indicates highest resolution shell: 2.09–2.00 Å for SeMet; 1.89–1.80 for native.