A dimeric chlorite dismutase exhibits O2-generating activity and acts as a chlorite antioxidant in Klebsiella pneumoniae MGH 78578.

Chlorite dismutases (Clds) convert chlorite to O2 and Cl(-), stabilizing heme in the presence of strong oxidants and forming the O═O bond with high efficiency. The enzyme from the pathogen Klebsiella pneumoniae (KpCld) represents a subfamily of Clds that share most of their active site structure with efficient O2-producing Clds, even though they have a truncated monomeric structure, exist as a dimer rather than a pentamer, and come from Gram-negative bacteria without a known need to degrade chlorite. We hypothesized that KpCld, like others in its subfamily, should be able to make O2 and may serve an in vivo antioxidant function. Here, it is demonstrated that it degrades chlorite with limited turnovers relative to the respiratory Clds, in part because of the loss of hypochlorous acid from the active site and destruction of the heme. The observation of hypochlorous acid, the expected leaving group accompanying transfer of an oxygen atom to the ferric heme, is consistent with the more open, solvent-exposed heme environment predicted by spectroscopic measurements and inferred from the crystal structures of related proteins. KpCld is more susceptible to oxidative degradation under turnover conditions than the well-characterized Clds associated with perchlorate respiration. However, wild-type K. pneumoniae has a significant growth advantage in the presence of chlorate relative to a Δcld knockout strain, specifically under nitrate-respiring conditions. This suggests that a physiological function of KpCld may be detoxification of endogenously produced chlorite.

O2 bond-forming processes are rare in biology. At present, two examples are well-described. First, the water-splitting reaction catalyzed by photosystem II (PSII) drives the photosynthetic fixation of CO2 into carbohydrates.[1] Second, a much more unusual reaction found in perchlorate (ClO4–)-respiring bacteria is used to detoxify chlorite (ClO2–), the end product of the perchlorate respiratory pathway. ClO2– is rapidly converted to O2 and Cl– via an enzyme known as chlorite dismutase (Cld).[2] In sharp contrast with the complexity of PSII,[1] Clds are soluble enzymes catalyzing an O–O bond-forming reaction with heme b as the sole cofactor.

Clds from perchlorate-respiring bacteria catalyze this reaction with great efficiency. The enzyme from Dechloromonas aromatica (DaCld), which we have characterized extensively and use here as an example of a highly adapted respiratory Cld, turns over >20000 equiv of chlorite per heme at rates nearing the diffusion limit before it is irreversibly inactivated.[3−7] The proposed mechanism for O2 formation (Scheme 1) involves initial transfer of an oxygen atom from ClO2– to the ferric heme to make an FeIV=O porphyrin (Por) π-cation radical (Compound I). The resulting hypochlorite (OCl–) leaving group is stabilized in its nucleophilic, anionic state by a positively charged arginine residue in a sterically confined pocket above the heme plane.[8−11] The “trapped” hypochlorite is then poised for nucleophilic attack on the electron-deficient ferryl oxygen atom to generate a putative ferric peroxyhypochlorite (OOCl–) intermediate that decomposes to yield Cl– and O2.

Scheme 1

The requirement for the Cld-catalyzed reaction by perchlorate-respiring bacteria is absolute; without it, ClO2– rapidly accumulates to toxic levels.[2] However, Clds are found in almost all bacterial phyla and in many archaea, the overwhelming majority of which are not involved in perchlorate respiration.[12] These Clds subdivide according to sequence into groups that appear to be functionally distinct from their respiration-associated counterparts and from each other (Figure 1), in spite of their high level of sequence similarity.[13] Investigation of the Cld from Staphylococcus aureus, a Gram-positive bacterium, provided a striking example. In its heme-bound form, it is completely inactive in the conversion of ClO2– to Cl– and O2.[14] Genetic and biochemical evidence instead suggests that the protein is essential for a terminal step in heme biosynthesis,[14,15] a biological function that appears to be common to the Clds from Gram-positive bacteria and potentially other species.[13,15]

Figure 1

Phylogenetic tree illustrating the major subdivisions of the Cld protein family discussed in the text. The respiratory Clds, coming mostly from Proteobacteria with some exceptions caused by lateral gene transfer, form the first group (red). Members of a second group of dimeric Clds (green), including KpCld, come from non-perchlorate-respiring species but retain active site features critical for ClO2–/O2 conversion. Members of a third broad group of Clds (blue) catalyze a terminal step in heme biosynthesis in Gram-positive bacteria and potentially some Archaea. These have consequently been renamed HemQs.[11] Sequence accession numbers are given in parentheses, with PDB entries used to indicate where structures are available. The phylogenetic tree was generated using MEGA6.[75] The bootstrap consensus tree inferred from 1000 replicates is taken to represent the evolutionary history of the taxa analyzed. Branches corresponding to partitions reproduced in <70% of bootstrap replicates are collapsed. Initial trees for the heuristic search were obtained by applying the Neighbor-Joining method to a matrix of pairwise distances estimated using a JTT model. A discrete Gamma distribution was used to model evolutionary rate differences among sites [five categories (+G, parameter = 2.4895)]; 5% alignment gaps, missing data, and ambiguous bases were allowed at any position.[76] The resulting tree is consistent with those previously reported by others.[11,13,77]

The Cld from Klebsiella pneumoniae strain MGH 78578 (KpCld) represents another subfamily that may have yet a third biochemical function. These enzymes are found in non-perchlorate-respiring bacteria from Gram-negative phyla (Figure 1).[12,13] As a consequence, they are not expected to be involved in either perchlorate respiration or heme biosynthesis. Unlike the respiration-associated Clds, they are homodimers rather than homopentamers, leading to a very different monomer–monomer interface and resulting structural context for their bound hemes (Figure 2A–C). Locally, however, they possess very similar active site residues and share structural attributes essential for efficient ClO2–/O2 conversion. In particular, they share the functionally important distal arginine above the ferric heme’s open coordination position (Figure 2D).[8,11] Consistent with chemical expectations, O2 production has been demonstrated for at least one protein from this subgroup, the Cld from Nitrobacter winogradskyi [NwCld (structure shown in Figure 2A,D)];[11] however, the dimeric enzymes have presumably not had the selection pressure to evolve to the same level of efficiency as Clds that are essential for metabolizing perchlorate.

Figure 2

Clds from perchlorate-respiring and nonrespiring Proteobacteria. (A) KpCld, like the NwCld structure shown (PDB entry 3QPI),[7] has a truncated monomer relative to the respiratory Clds. Both form functional homodimers. Protein monomers are rendered as cartoons in different colors. Hemes are rendered as sticks. (B) Respiration-associated DaCld (PBD entry 3Q08) forms a functional homopentamer, shown looking down the C5 axis.[4] (C) A monomer–monomer interface from the structure in panel B suggests differences in heme accessibility in the pentameric and dimeric Clds. (D) Active site environments for DaCld (cyan carbons) and NwCld (magenta carbons) are superimposed with several conserved residues around the heme labeled (DaCld numbering). The two active sites are highly similar; two notable exceptions are the orientation of the distal Arg side chain and the identity of the residue at the position of W227, which is a conserved glutamate in NwCld and other dimeric Clds.

Why bacteria like K. pneumoniae MGH 78578, a respiratory pathogen, possess a potential means for chlorite detoxification is not clear. K. pneumoniae and other non-perchlorate-respiring species generally do not encode perchlorate or chlorate reductase enzymes,[17] yet pathogenic strains of clinically pervasive[18] species, including K. pneumoniae, Acinetobacter baumanii, and Pseudomonas aeruginosa, all possess genes from this cld subtype, often on plasmids or as part of pathogenicity islands. These genes appear unlikely to guard against environmental chlorite, because of its chemical instability and the inability of negatively charged chlorite to passively enter cells.[19] However, an indirect antioxidant function against ClO3– or ClO4– is possible, if these can enter the cell and be reduced by endogenous enzymes. To test this hypothesis, and to enhance our understanding of the influence of the unusual monomer and oligomerization states on an otherwise similar heme environment, KpCld was characterized in parallel with the Δcld strain of K. pneumoniae MGH 78578. The results presented here report on both the biological function of this group of Clds and the biochemical properties that make them unique.

Experimental Procedures

Chemicals

Chemicals were purchased from Alfa Aesar, Sigma-Aldrich, or VWR. Sodium chlorite, 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS), guaiacol, and 2-chloro-5,5-dimethyl-1,3-cyclohexanedione (MCD) stock solutions were made fresh daily in buffers prepared from distilled and deionized water and their concentrations determined spectrophotometrically: ε262 = 160 M–1 cm–1 (NaClO2), ε415 = 3.6 × 104 M–1 cm–1 (ABTS), and ε290 = 20100 M–1 cm–1 (MCD).[20] Iodometric titration was routinely used to quantify working solutions of sodium chlorite or hypochlorite [diluted from a 10–14% (w/w) stock], as well as unreacted chlorite remaining at the end of experiments.[21] Each substrate was dissolved directly in the specified aqueous buffers with the exception of guaiacol, for which a 0.9 M stock was prepared in dimethyl sulfoxide (DMSO). Stocks of hemin were prepared in DMSO.

Generation of the Δcld Strain

Mutagenesis of K. pneumoniae MGH 78578 was conducted as described previously.[22] Briefly, a derivative of pKOBEG encoding apramycin resistance, pKOBEGApra, was used to facilitate lambda Red-based replacement of cld with a 1652 bp FRT site-flanked hygromycin resistance cassette (FRT-hph-FRT) amplified from pJTAG-hyg (Deenathayalaguptha and Rajakumar, unpublished data) using primers GmF and GmR.[23] Targeting flanking sequences TS1 (572 bp) and TS2 (527 bp) were amplified using primer pair P1 and P2 and primer pair P3 and P4. TS1, TS2, and FRT-hph-FRT were joined by splicing overlap extension PCR (SOE-PCR) to produce a 2702 bp amplicon (Figure S1 of the Supporting Information). Arabinose-induced electrocompetent K. pneumoniae MGH 78578/pKOBEGApra maintained at 30 °C was electroporated with 100 ng of the SOE-PCR product, and transformants were recovered on LB/200 μg/mL hygromycin agar at 37 °C after overnight culture. Several hygromycin resistant colonies were examined, and the Δcld mutant was validated by PCR analysis (Figure S1 of the Supporting Information). Bacteria and plasmids used for cld mutagenesis are listed in Table S1 of the Supporting Information.

Phenotypic Characterization of the Δcld Strain

Frozen (−80 °C) glycerol stocks of wild-type (WT) and Δcld strains of K. pneumoniae were revived by streaking on Luria broth (LB) agar and LB agar/hygromycin B (75 mg/mL) plates, respectively. Hygromycin B was used in all growth experiments to maintain the mutant free of contamination. Single colonies were inoculated into 5 mL of LB or LB with hygromycin B and grown to an OD600 of 0.4. Cells were harvested in microcentrifuge tubes and washed three times with M9 medium [0.2% (v/v) glycerol, 58 mM K2HPO4, 22 mM Na2HPO4, 85 mM NaCl, and 18.7 mM NH4Cl] supplemented with the following trace elements in 134 μM EDTA: 6.15 μM ZnO, 570 nM CuCl2·2H2O, 340 nM CoNO3·6H2O, and 1.6 μM H3BO3. Cells were resuspended to a final optical density at 600 nm (OD600) of 0.4 and used as a uniform inoculate (1:100) for all experiments.

For determination of minimal inhibitory concentrations of ClO– (X = 2–4), cells were inoculated into 10 mL of anaerobic, N2-purged medium in sealed glass crimp-topped bottles containing increasing concentrations of NaClO2, NaClO3, or KClO4 (0.5–200 mM) and 50 mM KNO3. The sealed bottles were placed in a 37 °C incubator with gentle shaking. The minimal inhibitory concentration (MIC) was defined as the lowest concentration of a reagent that would inhibit the visible growth of K. pneumoniae after incubation for 24 h.[24] No-nitrate controls were conducted in tandem under aerobic conditions. Finally, for generation of growth curves, cells were inoculated into aerobic media and their growth was monitored via periodic measurement of their OD600 over time.

Growth, Purification, and Characterization of Wild-Type (WT) and Mutant KpCld

DNA containing the full-length coding region of chlorite dismutase from K. pneumoniae MGH 78578 (GenBank accession number CP000650.1) was amplified via PCR with primers KpCldFor (5′-CGC CAT AAT ACA CGA TTA TTT ACG TTC GCT GG-3′) and KpCldRev (5′-TTT GGATCC CTA GGC CGG CTC ATG CA-3′) from a K. pneumoniae genomic DNA template (added cut sites for NdeI and BamHI at the 5′ and 3′ ends underlined, respectively). The product was subsequently cloned into the pET-15b (Merck/Novagen) expression vector for production of protein with an N-terminal His tag. The Y62F and W97F mutants of KpCld and the Y118F mutant of DaCld were generated from the WT via PCR-based single-codon substitution using a QuikChange kit. The W227F, W155F, and W156F mutants of DaCld were generated in a similar manner and were available from prior work.[25]

All KpClds were expressed in Escherichia coli Tuner (DE3) cells (Merck/Novagen) grown in Terrific Broth (TB) with ampicillin (100 μg/mL). Expression cultures were grown at 37 °C in a shaker incubator (250 rpm) to midlogarithmic phase (OD600 = 0.5). Isopropyl β-d-thiogalactopyranoside (IPTG, 1 mM) and δ-aminolevulinic acid (50 mg/L) were added to induce heme protein expression, and the temperature was lowered to 20 °C. After 16 h, cell pellets were collected by centrifugation and stored at −80 °C. Cells were thawed and resuspended [20 mM phosphate buffer, 500 mM NaCl, 20 mM imidazole, and 1 mM phenylmethanesulfonyl fluoride (pH 7.4)] and lysed by pulsed sonication on ice (7 min). The lysates were clarified by centrifugation and supernatants loaded onto a 20 mL HisTrap column. The protein was eluted using a 20 to 500 mM linear gradient of imidazole in resuspension buffer. Eluted proteins were screened by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and pure fractions buffer-exchanged into 0.1 M phosphate buffer (0.1 M, pH 6.8) using Amicon centrifuge concentrators (molecular weight cutoff of 10000). Pure protein was concentrated to 10 mg/mL (20% glycerol), frozen in liquid N2, and stored at −80 °C. All KpCld concentrations are given as heme-bound monomer, where [heme] and [protein] were determined by the pyridine hemochrome and Bradford assays,[26] respectively.

Measurement of Initial Rates of Chlorite-Decomposing Activity in the Steady State

A luminescence-based probe was used to measure O2 evolution by KpCld during chlorite decomposition under steady state conditions. Samples included 12 nM WT KpCld and 0.05–2 mM chlorite in the following buffer solutions: 50 mM phosphate-citrate (pH <6), 100 mM phosphate (pH 6–8), or 100 mM glycine (pH >8). The probe was equilibrated in the buffer/chlorite solution for 5–10 min prior to initiation of the reaction by introduction of enzyme. Kinetic traces were recorded at 1 s intervals for 3–10 min.

Monitoring Heme Chromophore Loss Caused by NaClO2 or NaOCl

Samples (10 μM, 200 μL) of heme-containing KpCld were titrated in UV–visible cuvettes with a concentrated chlorite or hypochlorite stock (600 mM) added in 1 μL increments [all solutions in 0.1 M citrate-phosphate buffer (pH 6.6)]. Samples were allowed to come to equilibrium following each addition and corrected for dilution.

Chromophore loss was also monitored over time following the addition of varying amounts of NaOCl as a function of pH. Reactions were conducted in 0.1 M citrate-phosphate at pH 6–8. Samples (10 μM, 200 μL) of heme-containing KpCld in UV–visible cuvettes were manually mixed with a NaOCl stock, yielding final NaOCl concentrations of 0–1 mM (0–2000 equiv). Spectra were measured every 6 s after mixing. The absorbance at the Soret band maximum (409 nm) was plotted versus time and fit to a single-exponential equation to obtain a first-order rate constant (kobs). Values of kobs measured at a given pH were plotted versus NaOCl concentration to obtain second-order rate constants (k is the slope).

Measurement of Residual Activities and Turnover Numbers

Chlorite’s potency as a suicide substrate was assessed by the method of Silverman.[27] Briefly, 5 μM samples of KpCld were incubated for 1 h at room temperature with an increasing number of equivalents of chlorite up to 3 × 104, with or without 0.5 mM added guaiacol, MCD, or ABTS [0.1 M citrate-phosphate buffer (pH 6.6)]. Reaction mixtures were dialyzed against chlorite-free buffer for two 1 h cycles to remove any unreacted substrate. The remaining KpCld reactivity was measured using a Clark oxygen electrode by adding 5 μL of each dialysate to 1.5 mL of 2 mM chlorite and measuring the initial rate of O2 production. Rates were measured three times, corrected for dilution, and averaged. Residual activity was computed by referencing the activity remaining after incubation in a given chlorite concentration to the activity of the zero chlorite control. The turnover number, defined as the total number of molecules of chlorite catalyzed per KpCld heme before the catalyst is irreversibly inactivated, was obtained by extrapolating a plot of residual activity versus [ClO2–]/[KpCld] to the x-intercept.

Vibrational Characterization of KpCld

Resonance Raman (rR) spectra were obtained with 413.1 or 441.6 nm excitation from a Kr+ or HeCd laser, respectively, using the 135° backscattering geometry for collection of Raman-scattered light. The spectrometer was calibrated against Raman frequencies of toluene, dimethylformamide, acetone, and methylene bromide. Spectra were recorded at ambient temperature from samples in spinning 5 mm NMR tubes. UV–visible absorbance spectra were recorded from the rR samples before and after spectral acquisition to assess whether sample integrity had been compromised by exposure to the laser beam. The laser power at ferric and ferrous samples ranged from 5 to 10 mW; no spectral artifacts due to photoinduced chemistry were observed with these irradiation powers. Ferric KpCld samples for the rR pH dependence study were prepared in the following 100 mM buffers: potassium phosphate (pH 5.7–7.5), Tris-HCl (pH 8.6–9.7), and borate buffer (pH 10.1).

Electron Paramagnetic Resonance (EPR) Spectroscopy

The 9 GHz EPR spectra were recorded on a Bruker EleXsys E500 spectrometer equipped with a standard Bruker ER 4102 X-band resonator and a liquid helium cryostat (Oxford Instruments, ESR 900). The spectra for WT KpCld at pH 6.0 were recorded at 4 K, a microwave power of 1 mW, a modulation amplitude of 4 G, and a modulation frequency of 100 kHz. The spectra at pH 8.0 were recorded at 12 K, a modulation amplitude of 10 G, a microwave power of 0.5 mW, and a modulation frequency of 100 kHz. Spectra for the mutant KpClds were recorded at 4 K, a microwave power of 1 mW, a modulation amplitude of 4 G, and a modulation frequency of 100 kHz. To avoid significant changes in pH upon freezing, Tris-maleate buffer, which covers the pH range from 5.2 to 8.6 with minor changes (0.1 pH unit) upon freezing, was used. No effect on the ferric EPR signal was observed for the Tris-maleate buffer as compared to that with phosphate buffer when using the same pHs. Typically, EPR samples of ferric KpCld and DaCld [in 0.1 mM Tris-maleate (pH 6.0 or 8.0) and at 0.6 mM enzyme] were measured in 4 mm quartz tubes. The buffer exchange in the pH range of 5.0–8.0, conducted with Centricon microconcentrators (Amicon), could be reversibly obtained without enzyme precipitation, degradation, or iron release as judged by the electronic absorption and EPR spectra recorded after each buffer exchange. Freezing and thawing cycles at high enzyme concentrations used for the EPR characterization did not induce any enzyme precipitation or changes in the heme environment.

Results

Heme Environment in KpCld As Measured by UV–Vis and rR

KpCld has a reversible pH-dependent transition in UV–visible spectrophotometric titrations. Experiments in which the enzyme was titrated with base (Figure 3A) and acid (data not shown) yielded the same pKa. Fits of UV–vis data at 390 nm (Figure 3A, inset), 413 nm, and 574 nm (not shown) indicated a pKa of 8.3, the same as that reported for the Azospira oryzae Cld[28] and slightly lower than that for Clds from Ideonella dechloratans (8.5)[29] and D. aromatica (8.7).[5] Heme proteins with neutral proximal histidine ligands have pKa values [heme oxygenase-1 and -2 (7.6 and 8.5, respectively), SmFixL (9.6), and HmuO (9.0)] lower than the pKa values of those with a proximal histidinate [peroxidases (pKa values of 11–12)].[30−32] Consistent with a pKa in this range, the Fe(II)–His vibrational frequency measured by rR for KpCld suggests a charge-neutral proximal histidine (Figure S2 of the Supporting Information; discussed further below).

Figure 3

pH-dependent behavior of KpCld indicates a water-accessible heme. (A) UV–visible pH titration of KpCld. Spectra shown at pH 7.13, 7.38, 7.85, 8.1.4, 8.46, 8.78, 9.14, 9.61, 10.07, and 10.43. Arrows indicate the direction of the absorbance change with pH. The inset shows the absorbance at 390 nm as a function of pH; a least-squares fit to the data (●) gives a pKa of 8.3. (B) Speciation of ferric KpCld as observed in rR pH titration. Resonance Raman spectra were obtained with 413.1 nm excitation at the indicated pH values. (C) Soret-excited rR spectra of the oxygen isotopologs of alkaline KpCld [50 mM Ches (pH 9.8)]. The top three spectra were measured in the indicated solvent at identical acquisition times. The bottom trace is the difference spectrum generated by a 1:1 digital subtraction (H2O – H218O).

At pH 6.0, ferric KpCld exhibits a Soret band maximum at 405 nm, a broad α/β band envelope at 505 nm with a small shoulder at 540 nm, and a charge transfer (CT) band at 645 nm (Figure 3A). This spectrum is similar to that reported for the related dimeric NwCld (405, 506, 543, and 640 nm) and the pentameric Cld from CandidatusNitrospira defluvii at pH 7.0,[10,11] but different from Clds from D. aromatica (393, 506, and 648 nm, pH 7.0), A. oryzae (392–394 nm), and I. dechloratans (392, 509, and 648 nm, pH 7).[5,9,29] The latter three proteins, which come from perchlorate respirers, share similar five-coordinate, high-spin (5cHS) heme sites, pentameric oligomerization states, and hydrophobic, solvent-enclosed heme environments, to which their blue-shifted Soret bands have been attributed. In contrast, the visible spectrum of acidic KpCld corresponds to a mixture of coordination states; this is confirmed by its rR spectra (below).

The rR spectra (Figure 3B) of ferric KpCld in a mildly acidic solution have a broad HS ν3 envelope spanning 1482–1490 cm–1, consistent with a mixture of six-coordinate, high-spin (6cHS) and 5cHS heme. The temperature dependence of the HS ν3 envelope supports the presence of 5cHS and 6cHS hemes. When the temperature of ferric KpCld is decreased from 19 °C at pH 6.0, the 6cHS intensity of the ν3 envelope at 1484 cm–1 increases at the expense of the 5cHS component at 1491 cm–1. Below −14 °C, the envelope narrows to a single band centered at 1484 cm–1, indicating complete conversion of the HS heme to a 6c complex that is likely an aqua complex, whose formation is exothermic (Figure S3 of the Supporting Information). The acid–base behavior of this complex (see below) supports this assignment. These data indicate that a considerable fraction of the resting HS enzyme is coordinatively saturated at near-physiological temperatures. In contrast, the acidic form of DaCld contains only 5cHS heme.[5]

Another small ν3 band occurs at 1505 cm–1 at 19 °C and pH ∼6 (Figure 3B). The intensity of this band increases in concert with that of a new 6cHS band at 1478 cm–1 as the pH is increased. Bands in the 1505 cm–1 frequency range can be consistent with 6cLS species or 5cQS (5c quantum mechanical spin admixture of S = 3/2 and S = 5/2 ferric heme). The appearance of UV–visible bands near 572 nm (typical for 6cLS species) and 646 nm (CT) at pH 7.0 argue for assignment of the ν3 band to a 6cLS species. At pH 10, ferric KpCld has a Soret maximum at 413 nm, α/β bands at 572 and 540 nm, and small absorbance bands near 490 and 603 nm. The 572 and 540 nm bands are consistent with a 6cLS complex, while the bands at 490 and 603 nm are suggestive of a 6cHS species (Figure 3A). These features are very similar to those reported for alkaline ferric DaCld.[5] The mixture of spin states suggested by the UV–visible spectrum is confirmed by the appearance of two ν3 bands at 1478 and 1505 cm–1 in the rR spectrum of alkaline KpCld. Growth of these bands is accompanied by the appearance of a shoulder at 1640 cm–1, which can be attributed to ν10 for 6cLS heme, and a 2 cm–1 upshift in the frequency of ν4 (not shown), both indicating growth in the population of 6cLS heme. The formation of heme hydroxides is often characterized by parallel growth in the intensity of bands arising from 6cHS and 6cLS hemes, which indicate its presence as a thermal spin state equilibrium. Similar mixtures attributed to such equilibria have been reported for a number of heme proteins, including the alkaline forms of DaCld, myoglobin, hemoglobin, Mycobacterium tuberculosis HbN, and SmFixL.[5,35−38]

To determine whether the sixth ligand in KpCld is hydroxide at pH 9.8, the νFe–OH modes were identified by isotopic substitution (Figure 3C). Bands at 509 and 442 cm–1 were assigned to the Fe–OH stretching modes for the LS and HS heme hydroxides, respectively. The LS νFe–OH band shifts to 494 cm–1 in D2O and to 488 cm–1 in H218O. In the case of the HS heme, no deuterium shift is detected. In H218O, the νFe– band shifts by 22 cm–1 to 420 cm–1. Thus, the pKa of 8.3 corresponds to the formation of the KpCld–OH complex. Consistent with our previous report about the DaCld–OH complex,[5] the νFe–OH bands of both HS and LS KpCld–OH complexes fall at the low end of the Fe–OH stretching frequency range for heme hydroxides.

Finally, the UV–visible spectrum of ferrous KpCld (pH 7.0) generated by reduction of the ferric protein with a 12-fold redox excess of sodium dithionite exhibited a Soret band maximum at 433 nm, typical of 5cHS ferrous heme, and α/β bands at 586 and 555 nm, respectively (Figure S2 of the Supporting Information). The 413.1 nm excited Fe(II) KpCld spectrum at pH 7.0 has a ν4 at 1356 cm–1, typical of ferrous heme, and a ν3 at 1472 cm–1, typical of 5cHS ferrous heme. The 441.6 nm excited Fe(II) KpCld spectrum revealed an Fe–His stretching band at 229 cm–1, 7 cm–1 higher than that observed for DaCld (222 cm–1 at pH 6.8),[5] consistent with the Fe–His bond being slightly stronger in KpCld assuming similar normal mode compositions in the two enzymes.

Structural Features of the KpCld Heme Environment, Including Second-Sphere Interactions As Revealed by 9 GHz EPR Spectroscopy

EPR spectroscopy is a sensitive probe of the extended hydrogen bonding network of heme active sites, including structural water molecules as well as second-sphere (and beyond) amino acid residues.[39,40] In the absence of a crystal structure for KpCld and to expand the structural view of the heme site provided by rR, the ferric EPR spectra of wild-type KpCld and key variants in its heme extended environment were examined as a function of pH. These were compared to their DaCld counterparts, for which the WT crystal structure has been determined.[8]

Figure 4 shows the 9 GHz EPR spectra of wild-type KpCld recorded at cryogenic temperatures (4 and 12 K). The ferric EPR spectrum of KpCld remained invariant in the pH range of 5.5–7.9 (top, black trace) but showed a dramatic change at pH 8.0 (top, gray trace). At pH 6.0 (Figure 4, top, black trace), the EPR spectrum of KpCld showed an axial signal, with observed effective g values of g⊥ = 5.92 and g∥ = 1.99, consistent with a ferric high-spin species. At pH 8.0 (Figure 4, top, gray trace), the KpCld EPR spectrum showed a distinct LS ferric species, with effective g values of gC = 2.54, gC = 2.19, and gC = 1.87. The almost complete conversion of the ferric EPR signal from high-spin to low-spin at pH 8.0, observed in frozen solutions, differs from the more gradual effect observed in the rR experiments in solution (Figure 3B) and most possibly reflects the preferential configurations of the heme environment locked in the frozen samples. The ferric LS EPR spectrum overall is consistent with a nitrogenous amino acid side chain not directly coordinated to the heme iron, in the vicinity of the Fe–OH bond revealed by the rR characterization described in the previous section. The relatively smaller contribution of another LS signal with a large gmax of 3.10 (the other two components being very weak and too broad to be detected) consistent, in this case, with a nitrogenous ligand on the heme distal side[8] was also observed. The identity of this ligand is not known. It is possible that, even after thorough dialysis, some imidazole used in the purification of the protein remained in the active site. Alternatively, repositioning of the distal arginine, with a deprotonated guanidinium group replacing water molecules close to the iron [for example, in the structural water molecules shown in the NwCld structure (Figure 2)], could be envisioned if considering the crystal structures of the N. defluvii Cld with thiocyanate bound in the sixth coordination position of the heme.[10]

Figure 4

9 GHz EPR spectra of ferric Clds from K. pneumonia (top) and D. aromatica (bottom) as a function of pH. A spin switch from high-spin ferric (black trace) to low-spin ferric (gray trace) is observed at pH ≥8.0. The conversion of the high-spin ferric signal was estimated to be 95% in DaCld and 85% in KpCld. The spectra at pH 6.0 were recorded at 4 K, a microwave power of 1 mW, a modulation amplitude of 4 G, and a modulation frequency of 100 kHz. The spectra at pH 8.0 were recorded at 12 K, a modulation amplitude of 10 G, a microwave power of 0.5 mW, and a modulation frequency of 100 kHz.

The EPR spectra of DaCld differed substantially from those of KpCld. Specifically, the spectrum at pH 6.0 showed the contribution of two ferric HS EPR signals (Figure 4, labeled with subscripts A and B), with effective g values of (gA = 6.67, gA = 5.20, gA = 1.96) and (gB = 6.25, gB = 5.24, gB = 1.99). Two distinct LS ferric species (Figure 4, subscripts C and D), with effective g values of (gC = 2.56, gC ≈ 2.18, gC = 1.87) and (gD = 2.64, gB ≈ 2.18, gB = 1.82), were consistently observed at pH 8.0. These observations of two HS and two LS EPR signals are consistent with our earlier reports that closed (more active) and open (less active) DaCld conformers are present between pH 5.6 and 9.1.[16]

To improve our understanding of the lack of pH-induced changes in the HS EPR spectra as well as the difference in ferric heme signals of the various Clds, structurally conservative mutations at highly conserved Trp and Tyr residues within hydrogen bonding distance of the heme propionates were constructed. Specifically, the crystal structures of the Da and Nw Clds show that a highly conserved tyrosine (Tyr118, Figure 2D; Tyr62 in KpCld) appears to make hydrogen bonding contacts to both heme propionates via its phenol oxygen and amide nitrogen. The indole N atom of a conserved tryptophan (Trp155 of DaCld or Trp97 of KpCld) lies within hydrogen bonding distance of one heme propionate. A second strictly conserved Trp (Trp156 of DaCld or Trp98 of KpCld) is sterically close to the same propionate but oriented such that the indole nitrogen cannot form a hydrogen bond to it. Accordingly, we anticipated that substitutions of these conserved residues could considerably affect the orientation of the propionate(s) and/or heme planarity, resulting in measurable changes in the ferric EPR spectra.

Figure 5 shows dramatic changes observed in the ferric EPR spectra of KpCld and DaCld upon mutation of these active site residues. The axial HS EPR signal of WT KpCld (Figure 5, top, dotted black trace) fully converted to a rhombically distorted spectrum in the Y62F mutant (Figure 5, top, magenta trace), with the contribution of two species [effective g values of (6.46, 5.39, 1.98) and (6.57, 5.54, 1.99)]. The W97F mutation in KpCld induced a partial conversion of the wild-type axial signal to one of the rhombically distorted HS forms of the Y62F variant, with effective g values of (6.57, 5.54, 1.99), with a contribution of the 6cLS form with a large gmax of 3.10 (Figure 5, top, purple trace). The same LS high-gmax form was observed in WT KpCld, but only at basic pH (Figure 4, top, gray trace). In the Y118F mutant of DaCld, positionally equivalent to Y62F in KpCld, the EPR component with the largest rhombic distortion disappears (Figure 5, bottom, magenta trace). The ferric EPR spectrum of the Y118F DaCld variant then becomes very similar to the previously reported rhombic EPR spectra of the Clds from A. oryzae and I. dechloratans,[9,29] and also similar to that of Y62F KpCld. Accordingly, these results show that breaking the hydrogen bonds to the propionates allows more flexibility of the heme.

Figure 5

9 GHz EPR spectra of selected mutations on the heme environment of KpCld (top) and DaCld (bottom) at pH 6.0. Both spectra of the wild-type Clds (black dotted traces) are shown for comparison. Figure 2D shows the crystallographic structure of the extended heme environment of NwCld (PDB entry 3QPI), in which the amino acid residues at positions equivalent to those mutated in KpCld (Y62F and W97F, at the same positions as Tyr118 and Trp155) are shown. The other Trp mutated in DaCld (Trp227) is not conserved in NwCld or KpCld. Spectra were recorded at 4 K, a microwave power of 1 mW, a modulation amplitude of 4 G, and a modulation frequency of 100 kHz.

To explore these differences further, we investigated the DaCld W227F mutant. This residue is conserved among the Clds from known perchlorate respirers but not in the dimeric Clds (Figures 1 and 2) and could represent a key differentiation between the two groups. Trp227 is not connected to the heme by hydrogen bonds yet is sterically close to those other residues having hydrogen bonding interactions to the heme. Interestingly, two significant changes were observed in the ferric EPR spectrum of W227F DaCld relative to WT: the disappearance of the component with the largest rhombic distortion, as in DaCld Y118F, and an increase in the contribution of the axial EPR signal (Figure 5, bottom, violet trace). Moreover, the ratio of the axial and rhombically distorted signals of the W227F DaCld EPR spectrum became pH-dependent, as in the case of Burkholderia pseudomallei KatG,[41] fully converting to the axial signal at pH 5.0 (Figure 5, bottom, green trace). Hence, the resulting EPR spectrum for DaCld W227F at pH 5.0 is the same as that of wild-type KpCld (Figure 5, top, dotted black trace). Notably, the mutant protein could be isolated in a stably heme-bound state at low pH if concentrated and frozen immediately following purification (see Experimental Procedures).[25] These results reinforce the idea that second-sphere coordination influences the electronic structure of the heme iron. They also suggest that the conserved residue Trp227 is indeed important for distinguishing the heme environments of DaCld and KpCld even though it is not directly connected to the heme via hydrogen bonding.

KpCld Exhibits Chlorite-Decomposing Activity

KpCld is a competent catalyst of chlorite decomposition; values of kcat and kcat/KM are maximal near pH 5.0, at (1.9 ± 0.2) × 103 s–1 and (2.5 ± 0.4) × 106 M–1 s–1, respectively (20 °C) (Figure S4 of the Supporting Information). These values are approximately 10-fold lower than the corresponding parameters for DaCld measured at its pH 5.2 optimum (4 °C): kcat = (2.0 ± 0.6) × 104 s–1, and kcat/KM = (3.2 ± 0.4) × 107 M–1 s–1.[5] A pH optimum between 5 and 6 was recently measured for the kcat for the N. defluvii Cld.[10,42]

Steady state pH–rate profiles for KpCld (Figure S4 of the Supporting Information) and DaCld[5] are broadly similar, with each protein possessing a more active acidic form and less active alkaline form. Turning points in the plots of log kcat versus pH were fit at pH 6.5 and 8.7 (DaCld).[5] A transition from a highly active low-pH form to a less active alkaline form is apparent near pH 7 for KpCld. Turning points in log–log plots of kinetic constants are associated with pKa values. Though not explicitly assigned here for KpCld, the lower-pH turning point in DaCld was previously assigned to the distal arginine, first modeled as an explicit deprotonation and later associated with movement of the distal pocket Arg between less reactive “out” (alkaline) and more reactive “in” (acidic) conformations (Figure 2). The latter model is supported by both spectroscopic evidence and reactivity data with H2O2.[5,6,16]

Chlorite Acts as a Potent Suicide Substrate

Titration of KpCld with chlorite (Figure 6A) demonstrated near-complete elimination of the heme’s Soret band following exposure to roughly (6.0 ± 0.3) × 103 equiv of the oxidant. This value is intermediate between DaCld’s (2.0 × 104) and SaCld’s (≤5).[5,14] To relate heme destruction quantitatively to the loss of catalytic activity, KpCld’s turnover number was measured (Figure 6B). A line fit to a plot of residual activity versus chlorite equivalents intercepts the x-axis at 5.3 × 103. The same data could also be fit to an exponential decay curve, consistent with some protection of the enzyme from degradation by the accumulation of high product concentrations.[43] The exponential curve predicts heme degradation after exposure to even fewer equivalents of chlorite, that is, the number of equivalents extrapolated from where the initial, linear phase intercepts the x-axis. The number extrapolated from a straight line fit to all of the data is therefore likely an upper limit for turnovers.

Figure 6

Chlorite acts as a suicide substrate and is rescued by H atom donors and chlorination substrates. (A) ClO2– was added in 300 equiv increments and the heme chromophore observed to diminish to baseline after roughly 6 × 103 equiv. The inset shows the Soret band absorbance plotted vs added chlorite equivalents. (B) Residual enzymatic activity following incubation with increasing equivalents of chlorite. Linear extrapolation to the x-axis yields the turnover number: no additives (■, 5.3 × 103), with added ABTS (●, —, 1.5 × 104), and with added MCD (○, −––, 1.5 × 104).

The measured turnover number is similar to but slightly less than the number of equivalents required to completely eliminate the Soret band (Figure 6A, inset), suggesting that loss of activity is correlated with destruction of the heme and also some degree of protein damage. Consistent with these results, a detailed study of chlorite-mediated damage to the pentameric Cld from N. defluvii showed that chlorite effected damage in a number of ways, including oxidation of methionine residues, chlorination of aromatic side chains, and heme lysis.[44]

H Atom Donors and Chlorination Traps Increase KpCld’s Turnover Number

The turnover number was remeasured in the presence of excess guaiacol and ABTS, which can act as sacrificial hydrogen atom donors toward highly reactive oxidants. These could include ferryl heme species, (H)OCl produced as intermediates in the O2-generating reaction, or chlorine dioxide (ClO2) from the one-electron oxidation of chlorite by a ferryl heme. In the presence of guaiacol, the turnover number for DaCld increased by approximately 10-fold.[3] Here, in guaiacol or ABTS, the turnover number increased more modestly, to 1.5 × 104 (∼2-fold larger). The same trend was observed in the presence of MCD, a trap used to detect chlorinating agents (HOCl, ClO2, or OCl–) in enzymatic reactions.[45,46] These results indicated that a species capable of reacting with guaiacol, ABTS, and MCD was responsible for chlorite-mediated inactivation of KpCld.

(H)OCl Is Produced by KpCld during Steady State Turnover with Chlorite

To directly detect the species generated during turnover, the reaction between ClO2– (3 mM) and KpCld (0.1 μM) was monitored via UV–vis under steady state conditions. Unreacted chlorite was quantified by iodometric titration. ClO2– decomposition was incomplete: 36 ± 5% of the initially present chlorite had degraded by the end of the experiment, indicating (6.7 ± 0.3) × 103 turnovers per heme; representative results are shown, and turnover numbers are an average of three experiments (Figure 7A). Chlorite decomposition moreover occurred with no observable accumulation of ClO2 (ε262 = 160 M–1 cm–1). When 50 μM MCD was added to the same reaction mixture, conversion of the MCD to DCD could be readily detected via the loss of the MCD chromophore (Figure 7B). The reaction with ClO2– also proceeded much closer to completion: 73 ± 5% consumed, or (1.4 ± 0.1) × 104 turnovers per heme.

Figure 7

Chlorite decomposition is enhanced by the presence of (H)OCl trapping agents. (A) An inhibitory concentration of ClO2– (3 mM) was added to a catalytic amount (0.1 μM) of KpCld [0.1 M citrate-phosphate (pH 6.6)]. Spectra were measured every 30 s (gray lines) until the reaction had reached completion (black lines; t = 0 and 30 min); 36% of the initially present ClO2– decomposed without the appearance of ClO2. (B) In the presence of 50 μM MCD and the same amount of ClO2–, the turnover number more than doubles. MCD’s chromophore disappears, consistent with conversion to the chlorinated product DCD. (C) Colorless DMSO, used as a HOCl-specific trap, enhances ClO2– turnover to a similar extent.

To distinguish definitively between HOCl or ClO2 as the MCD-reactive species, DMSO was assayed for its ability to enhance turnover. DMSO has previously been used as a highly selective trapping agent for HOCl in aqueous ClO2/ClO2– mixtures.[47] Neutral HOCl is the reactive species (HOCl ⇆ OCl– + H+; pKa = 7.5), serving as an electrophilic oxidant via transfer of OH+ to a nucleophilic acceptor substrate (here, the lone pair of electrons on the sulfoxide).[47] When the KpCld/chlorite reaction shown in Figure 7A was monitored in the presence of 50 μM DMSO (Figure 7C), the reaction proceeded rapidly and nearly to completion [86 ± 7%; (1.6 ± 0.2) × 104 turnovers per heme]. This strongly suggested that HOCl and not ClO2 is released from KpCld during turnover with ClO2–. The results in panels A and B of Figure 6 are moreover consistent with the measured turnover number for chlorite and expected increases in turnover afforded by MCD (Figure 6), and they support the conclusion that MCD’s protective function is due to its reaction with HOCl.

Hypochlorous Acid Avidly Degrades KpCld’s Heme

To quantify the influence of chlorite-derived HOCl on heme degradation, the effect of exogenously added sodium hypochlorite on the heme spectrum was measured titrimetrically at pH 6.6, where it is expected to equilibrate rapidly to form HOCl. The spectrum diminished to baseline following addition of 800 ± 20 equiv of NaOCl (Figure S5 of the Supporting Information), versus (6.0 ± 3) × 103 equiv of ClO2– (Figure 3A). If HOCl alone were responsible for catalytic inactivation, this would suggest that approximately one in eight turnovers would lead to HOCl rather than O2 during catalysis of ClO2–. This number is likely an upper estimate, because the Fe(IV)=O Por•+ generated concomitantly with OCl– (Scheme 1) would also likely lead to protein damage in the absence of a reductant.

The measurements examining heme and catalytic stability above (Figures 6 and 7) were conducted at a single pH (6.6). However, both the enzyme and the HOCl suicide reactant are known to undergo pH-dependent transitions. Second-order rate constants for the heme/NaOCl reaction (k) were therefore measured 1.5 pH units below and 0.5 pH unit above the HOCl/OCl– pKa of 7.5 to assess which is the more likely reactive form. Heme decomposition was 1 order of magnitude faster in the presence of HOCl (pH 6; k = 0.13 M–1 s–1) than in the presence of OCl– (pH 8; k = 0.010 M–1 s–1) (Figure S6 of the Supporting Information). However, the spin state of the heme also clearly undergoes changes from HS to LS within this pH range (Figures 3 and 4). Changes in the heme electronic state could therefore also contribute to these differences in reaction rate.

Measurement of MICs

K. pneumoniae exhibits substantial ClO4– and ClO3– tolerance under aerobic conditions, withstanding concentrations of ≥100 mM without significant growth defects. It is much more strongly affected by ClO2–. No growth was observed even at 20 mM NaClO2, the lowest concentration tested. The mechanism of toxicity of ClO2– administered in the extracellular environment of bacteria has not been described. Lipid epoxidation, protein unfolding, and amino acid side chain modifications have all been observed in the presence of (H)OCl and could occur in the presence of ClO2–.[48] However, the ClO2–/HClO2 pKa (1.8) is significantly lower than that for OCl–/HOCl.[49,50] Hence, chlorite and not hypochlorous acid is expected to predominate under biological conditions, where its negative charge should bar its passive entry into cells. Consistent with an expected extracytoplasmic mechanism of toxicity, ClO2– is equally toxic to cells possessing KpCld in the cytoplasm and those in which it is absent.

We hypothesized that ClO3– could be more toxic to K. pneumoniae that actively reducing NO3– is as a respiratory substrate. Consistent with that hypothesis, WT K. pneumoniae appears to become sensitized specifically to ClO3– (not ClO4– or ClO2–) under nitrate-respiring conditions (anaerobic, 50 mM KNO3–), exhibiting a MIC near 70 mM (Table 1). This shifts to 60 mM for the Δcld mutant. These results suggest that K. pneumoniae metabolizes ClO3– under these conditions and that KpCld affords some protection against ClO2– produced endogenously from ClO3–. By contrast, ClO4– has no effect (up to 80 mM) on either the WT or Δcld mutant strain of K. pneumoniae, under either aerobic or nitrate-respiring conditions. These results collectively suggest that ClO4– is not incorporated and/or metabolized by K. pneumoniae.

Table 1

Determination of ClO3– MICs for WT and Δcld K. pneumoniae MGH under Aerobic and Nitrate-Respiring Conditionsa

	WT		Δcld
[ClO3–] (mM)	aerobic	anaerobic	aerobic	anaerobic
30	++	++	++	++
40	++	++	++	+
50	++	++	++	+
60	++	+	++	–
70	++	+	++	–
80	++	–	++	–
100	++	–	++	–

All cultures were grown in the presence of 50 mM added KNO3. Legend: ++, robust growth; +, minor growth; −, no observable growth.

Discussion

Chlorite dismutases are heme-binding proteins catalyzing steps in biological processes as divergent as perchlorate respiration and heme biosynthesis. Genetic and biological evidence has pointed toward a third, functionally distinct subgroup of the Cld family (Figure 2), members of which are found in diverse Gram-negative bacteria, including many obligate pathogens.[11,17] These bacteria are not known to respire (per)chlorate, nor do they lack any of the canonical genes required for making heme.[57] Their Cld proteins nonetheless share an almost identical active site with the respiration-associated Clds, though the overall monomer structure and oligomerization state differ. The biological role and distinct molecular features of this subgroup have been examined here, through biochemical investigation of KpCld and phenotypic characterization of the corresponding gene knockout in a pathogenic strain of K. pneumoniae (MGH 78578).

The Δcld strain of K. pneumoniae had neither a growth defect nor any obvious sign of impaired heme metabolism. This contrasts sharply with the Δcld strain of S. aureus, a slow-growing small colony variant with global deficiencies in its cellular heme levels.[14] We therefore investigated the hypothesis that the cld acts as an antioxidant toward chlorite or a related chlorine oxoanion. It was discovered that WT K. pneumoniae gains a noticeable growth advantage over its Δcld counterpart in the presence of ClO3–. This advantage is observed specifically under anaerobic, nitrate-respiring conditions (Table 1), even though K. pneumoniae does not have a dissimilatory reductase for either ClO4– or ClO3–, nor have these anions been shown to support anaerobic growth. It has been suggested that ClO3– might be taken up and metabolized by widespread bacterial nitrate-associated pathways.[56,58] Catalytic reduction of chlorate could be catalyzed by nitrate reductases because of the favorable reduction potential and kinetic lability of ClO3– relative to those of ClO4–; indeed, some nitrate reductases have been shown to accept chlorate as a substrate in vitro.[59,60] The product ClO2–, trapped inside the cell because of its charge, would be expected to quickly reach toxic levels if not enzymatically removed.[2] Hence, the Cld acts as an antioxidant against the ClO2– generated endogenously from ClO3–.

Such an antioxidant function, even in the absence of perchlorate or chlorate respiration, is consistent with both the catalytic properties of KpCld and the potent toxicity of ClO2–. Like superoxide dismutase or catalase, antioxidant enzymes against reduced oxygen species, KpCld is relatively fast [kcat = (1.9 ± 0.2) × 103 s–1; kcat/KM = (2.5 ± 0.4) × 106 M–1 s–1 (25 °C, pH 5)]. This suggests that it may effectively outcompete reactions between ClO2– and reactive intra- or extracellular components. Potential targets of ClO2–-mediated damage are not as well characterized as those of HOCl;[48] however, ClO2– is known to act as an oxygen atom donor toward heme Fe(III) and reactive double bonds, as well as a source of highly reactive ClO2 and HOCl.[44,45,61] These properties support its industrial use as a microbicide and bleach.[62]

KpCld’s reactivity toward ClO2–, while rapid, is limited by the instability of the heme cofactor in the presence of its strongly oxidizing substrate. The heme spectrum and catalytic activity are completely eliminated by exposure to roughly 6000 equiv of ClO2– (Figure 6). The turnover number is enhanced by the inclusion of sacrifical reductants or chlorination traps in the reaction mixture, including DMSO, a reagent that combines rapidly and specifically with HOCl.[47] This strongly suggests that OCl–, generated along with FeIV=O Por•+ from the heterolytic cleavage of the O–ClO– bond (Scheme 1), is partially responsible for loss of the heme and its associated catalytic activity. This is consistent with recently reported results for a respiratory-type Cld from N. defluvii (discussed further below).[44]

The results also indirectly support the mechanism for O2 generation proposed in Scheme 1, wherein the OCl– leaving group recombines with the Fe(IV)=O Por•+ to produce O2 and Cl–. Kinetic sluggishness in the recombination would create the opportunity for release of OCl–. Thus, the role of the arginine likely involves both steering the OCl– leaving group and maintaining it in its more nucleophilic, anionic form, thereby mechanistically favoring the O–O bond forming reaction, as illustrated in Scheme 1. Evidence in support of this model is found in the sensitivity of Clds lacking the distal Arg to oxidative conditions. For example, the Cld from S. aureus, which contains a glutamine at the homologous position, undergoes complete beaching of the heme following exposure to only 5 equiv of ClO2–.[14] By the same token, substitution of the arginine in the N. defluvii Cld with a neutral residue strongly sensitizes its heme toward oxidant-mediated degradation.[44]

However, while HOCl is readily observed during chlorite turnover with KpCld, no MCD-trappable HOCl could be detected for DaCld under conditions similar to those used here (Figure S7).[3]DaCld is at the same time significantly less prone to catalytic inactivation (20000 turnovers vs 6000) and (H)OCl loss (Figure 6).[3,4] Hence, in spite of their shared O2-evolving activity and conserved active site residues, including the distal Arg, the heme environment in dimeric KpCld and its pentameric DaCld counterpart must be distinct in some ways.

The best insight into the nature of these distinctions comes from spectroscopic analysis of DaCld and KpCld, which provides a nuanced view of the heme environment. First, it is clear from both EPR and rR that KpCld and DaCld are different. The occurrence of a single axial ferric high-spin signal in the EPR spectrum of KpCld at pH <8.0 strongly contrasts with the case for DaCld, which exhibits two rhombically distorted ferric EPR signals. In fact, among the few Clds for which the ferric EPR spectra have been reported so far, only NwCld shows the same axial signal as KpCld.[41] While the rR spectra of DaCld and KpCld at pH 6 indicate high-spin heme, the enzymes differ in coordination state with DaCld being 5c and KpCld having a significant population of 6c heme. Second, both EPR and rR spectra are consistent with water as the sixth ligand in KpCld at neutral pH. An increased level of access of water to the active site pocket in KpCld relative to DaCld is suggested by the crystal structures of these two enzymes (Figure 8).

Figure 8

Surface rendering of NwCld and DaCld in their expected native oligomerization states, showing differences in the entryway leading to the heme: (A) DaCld pentamer (PDB entry 3Q08) and (B) NwCld dimer (PDB entry 3QPI). The monomers are rendered in different colors and the hemes in the green monomers shown as black sticks. The tunnel leading from the surface to the heme is more open in NwCld than in DaCld. Figures generated by PyMol (www.pymol.com).

Functionally, a more open catalytic pocket could explain the postulated lower fidelity in the recombination of Fe(IV)=O Por•+ with OCl–, leading to a lower turnover number and a higher frequency of hypochlorous acid escape in KpCld (Figure 8). The Cld from N. defluvii shares spectral features with KpCld that are consistent with a more solvated distal pocket.[11] It likewise produces readily trappable HOCl during turnover with ClO2–,[44] and the EPR spectrum of its bound heme becomes undetectable after exposure to roughly 3000 equiv of ClO2–. Third, EPR characterization of mutants positioned within hydrogen bonding distance to the heme propionates (Y118F in DaCld and Y62F KpCld) or within the second coordination sphere of the heme (W227 and W156 in DaCld and Trp97 in KpCld) clearly indicates that both direct and indirect interactions influence the electronic structure of the heme. The fact that some of these (W227) are conserved only in the respiratory Clds suggests that they could contribute to the unique functional characteristics of those enzymes. Finally, the absence of second-sphere nitrogenous ligands that could explain the spin switch observed in frozen solution at pH 8 [85% conversion of the high-spin to low-spin ferric EPR signal (see Figure 4)] strongly argues for a repositioning of the guanidinium group of the distal side Arg in concert with formation of the Cld–OH complex and/or changes in hydrogen bonding interactions to the heme propionates. Effects of these kinds were reported for cytochrome c peroxidase, in relation to the formation of the ferryl–oxo intermediate.[5656]

The relatively less robust hemes in KpCld and potentially other proteins in this subfamily are consistent with a biological role in providing intermittent protection against environmental ClO3–. The same may perhaps be true of the N. defluvii Cld that, though pentameric in structure like DaCld and the other respiratory Clds (Figure 1, red), is extraordinary for being found within a species that is both non-perchlorate- or non-chlorate-respiring and phylogenetically distinct from any known respirers.[12] The gene’s presence in this organism appears to be the result of lateral transfer; hence, as in KpCld, it may provide protection against environmental chlorate that is metabolized via nitrate-reducing pathways.

Of course, a chlorate-directed antioxidant role for KpCld and related enzymes (Figure 1, green) depends on the availability of chlorate in the natural environment. Perchlorate is used as an oxidant and propellant for rockets and fireworks, while chlorate is a common constituent of herbicides.[63] Because of their pervasive industrial use and the lack of a known geochemical mechanism for their formation, perchlorate and chlorate have long been thought to be largely unnatural in origin. However, the discovery of widespread perchlorate in pristine ground waters, ice cores, and undisturbed deserts prompted deeper investigation of its natural sources.[64−66] Potentially ancient atmospheric mechanisms for the generation of both perchlorate and chlorate are now favored on Earth and on Mars, where abundant perchlorate has been discovered.[67−72] Hence, Clds of the KpCld subtype may presently play an unaccounted-for role in Earth’s chlorine cycle and could have done so for some time. It is likewise possible that they could serve a similar role against other halogen oxide species (e.g., iodate) that are especially abundant in nonterrestrial environments such as the open ocean.[73−75]