The Cu(II) Reductase RclA Protects Escherichia coli against the Combination of Hypochlorous Acid and Intracellular Copper

During infection and inflammation, the innate immune system uses antimicrobial compounds to control bacterial populations. These include toxic metals, like copper, and reactive oxidants, including hypochlorous acid (HOCl). We have now found that RclA, a copper(II) reductase strongly induced by HOCl in proinflammatory Escherichia coli and found in many bacteria inhabiting epithelial surfaces, is required for bacteria to resist killing by the combination of intracellular copper and HOCl and plays an important role in colonization of an animal host. This finding indicates that copper redox chemistry plays a critical and previously underappreciated role in bacterial interactions with the innate immune system.

thetaiotaomicron), and probiotics (e.g., Lactobacillus reuteri) (36), suggesting that RclA's function may be broadly conserved and not specific to a single niche or type of host-microbe interaction. RclR, RclB, and RclC are much less widely conserved and are found only in certain species of proteobacteria, primarily members of the Enterobacteriaceae ( Fig. 1; see Data Set S1 in the supplemental material). RclB and RclC both contribute to HOCl resistance, but their mechanistic roles have yet to be determined (26). RclB is predicted to be located in the periplasm, and RclC is predicted to be an integral inner membrane protein.
Here, we have now determined that RclA is a thermostable, HOCl-resistant copper(II) reductase that is required for efficient colonization of an animal host and protects E. coli specifically against the combination of HOCl and intracellular copper, possibly by preventing the formation of highly reactive Cu(III). We also found that, surprisingly, extracellular copper effectively protects bacteria against killing by HOCl both in cell culture and in an animal colonization model. These findings reveal a previously unappreciated interaction between two key inflammatory antimicrobial compounds and a novel way in which a commensal bacterium responds to and resists the combinatorial stress caused by copper and HOCl.
(This article was submitted to an online preprint archive [37].)

RclA contributes to HOCl resistance and host colonization.
An rclA mutant of E. coli is more susceptible to HOCl-mediated killing than is the wild type (26). To expand on these results, we utilized here a growth curve-based method to measure sensitivity to sublethal HOCl stress by quantifying changes in the lag-phase extension (LPE) of cultures grown in the presence of HOCl. Using this method, which we have found to be considerably more reproducible than other techniques for assessing bacterial HOCl sensitivity, we observed significant increases in LPE for a ΔrclA mutant strain compared to the wild type grown in the presence of various concentrations of HOCl (see Fig. S1A, B, and C in the supplemental material). We observed similar trends when the assay was performed with stationary-and log-phase cells (see Fig. S1D, E, and G), so stationaryphase cultures were used for subsequent assays. Furthermore, we determined that there was no decrease in CFU after treatment with HOCl at these concentrations, indicating that LPE measures the recovery of cultures from nonlethal stress (see Fig. S1F). These results confirm previous results and further illustrate the importance of RclA in resisting HOCl-mediated oxidative stress in E. coli.
To directly test the role of RclA in interactions with an animal host, we examined the ability of E. coli to colonize the intestine of Drosophila melanogaster, where the presence of enterobacteria is known to stimulate antimicrobial HOCl production by the dual oxidase Duox (20,22). Since an E. coli K-12 strain did not efficiently colonize D. melanogaster (see Fig. S2A), we used the colonization-proficient E. coli strain Nissle 1917 (EcN) (38,39) in these experiments. The genome of EcN encodes homologs of all the known HOCl resistance genes found in E. coli K-12 (40,41). Unlike K-12, EcN forms robust biofilms (42)(43)(44)(45), preventing the accurate measurement of growth curves and LPE (Fig. S2B), but EcN was slightly more sensitive to killing by lethal doses of HOCl than was MG1655 ( Fig. S2C and D). EcN ΔrclA mutants had a significant defect in their ability to colonize NP1-GAL4 D. melanogaster flies compared to wild-type EcN at 3 and 8 h postinfection (hpi) (Fig. 2, circles). This shows that rclA is important for EcN tolerance of host responses during early colonization.
To investigate the role of host-produced RCS in the colonization defect of the rclA EcN mutant, we reduced the gut-specific expression of Duox in the flies using Duox-RNAi and repeated the colonization experiments. Both strains colonized significantly better at 3 hpi when Duox was knocked down in the flies (Fig. 2, "ϫ" symbols), which was expected because rclA is not the only gene that contributes to HOCl resistance in E. coli (19). Importantly, the colonization defect of ⌬rclA EcN at 3 hpi was abrogated in the DuoxIR flies, with CFU/fly not being significantly different from wild-type EcN colonizing flies that are able to express Duox in the gut (Fig. 2). We confirmed that HOCl production was reduced in the DuoxIR flies using the HOCl-sensing fluorescent probe R19-S (see Fig. S3A to C) (46). Taken together, these results show that rclA facilitates early colonization of an animal host and indicate that rclA relieves stress caused by host-produced oxidation in early stages of colonization. Duox activation and HOCl production are rapid host immune responses that occur at early stages (minutes to first few hours) of bacterial colonization of the gut (20). However, ROS and RCS production are not the only antimicrobial responses in Drosophila, which may explain why rclA is only required in early colonization. As the course of infection progresses, additional antimicrobial effectors, such as antimicrobial peptides regulated by NF-B signaling, become more abundant (27). EcN infection induced robust production of antimicrobial peptide production in our fly model (Fig. S3D, E, and F).
RclA is homologous to mercuric reductase and copper response genes are upregulated after HOCl stress in EcN. Although the fact that rclA protects E. coli from RCS was previously known (26), the mechanism by which it does so was not. Based on its homology to other flavin-dependent disulfide oxidoreductases (47), we hypothesized that RclA catalyzed the reduction of an unknown cellular component oxidized by RCS. RclA is homologous to mercuric reductase (MerA), an enzyme that reduces Hg(II) to Hg(0) (48). These sequences are particularly well conserved at the known active site of MerA, a CXXXXC motif (Fig. 3A). However, MerA has an extra N-terminal domain and two additional conserved cysteine pairs used in metal binding (48), while RclA only has one conserved cysteine pair (see Fig. S4). This indicates that if RclA does interact with metal(s), the interactions must be mediated through mechanisms different from those of MerA. While the present manuscript was in revision, Baek et al. published a crystal structure of RclA (49), and alignment of this structure with that of MerA (50) clearly illustrates the homology between these two proteins as well as the location of the MerA-specific domains (Fig. 3B).
HOCl-stressed E. coli K-12 downregulates genes encoding iron import systems (e.g., fepABCD and fhuACDF) and upregulates genes for zinc and copper resistance (e.g., copA, cueO, cusC, zntA, and zupT) (25), also suggesting metals may play some role in RCS resistance. The genome of EcN encodes the same complement of known copper resistance genes as is found in E. coli K-12 (40,41). Transcriptomic profiling of EcN after treatment with a sublethal dose of HOCl confirmed the regulation of metal stress response genes by HOCl, including the upregulation of several genes encoding proteins involved in response to copper toxicity ( Fig. 4; see also Data Set S2 in the supplemental material), despite the very small amounts of copper present in the media used in that experiment (9 nM) (51). These included members of the Cus and Cue export systems, which are factors appreciated for their role in preventing copper toxicity and importance for E. coli colonization within mammalian hosts (52)(53)(54)(55). The expression of the rcl operon is not regulated by any of the known Cu-sensing transcription factors of E. coli (56) and is not affected by changes in FIG 2 EcN lacking rclA colonizes D. melanogaster less effectively, and early colonization with ⌬rclA EcN is improved in the absence of Duox-mediated oxidation. Flies were fed either wild-type or ⌬rclA EcN (1 ϫ 10 11 CFU/ml), and bacterial loads were measured at the indicated times postinfection (n ϭ 4 to 5, Ϯ the SD). RCS-deficient Duox-RNAi flies were obtained from crosses of UAS-dDuox-RNAi with NP1-GAL4 (enterocyte-specific driver). Statistical analysis was performed using a two-way analysis of variance (ANOVA) with Tukey's multiple-comparison test (****, P Ͻ 0.0001; ***, P Ͻ 0.001; ns, not significant). Derke et al. media copper concentrations (57,58). Our results suggested that copper might play an important role during HOCl stress for EcN. The homology between RclA and MerA and the indication that copper and HOCl responses may be connected in E. coli led us to investigate the role of rclA in resisting HOCl stress under growth conditions containing different amounts of copper. Extracellular CuCl 2 protects both wild-type and ⌬rclA E. coli strains against HOCl. How the presence of copper influences bacterial sensitivity to RCS has not been investigated before this study. However, it is important to note that Cu chemically catalyzes the decomposition of HOCl to nontoxic O 2 and Cl - (59)(60)(61)(62). We first used growth curves in the presence of copper and HOCl to identify how combinations of HOCl and extracellular copper influenced the sensitivity of wild-type and ΔrclA mutant  E. coli ( Fig. 5A and B). As noted above, the ΔrclA mutant was more sensitive to HOCl in MOPS, but the sensitivity to HOCl of both strains was greatly increased when Cu was removed from the media, indicating that the low concentration of Cu present in morpholinepropanesulfonic acid (MOPS) medium (9 nM) (51) was enough to react with the added HOCl and change the sensitivity of our strains. Consistent with this, addition of 10 M CuCl 2 to HOCl-containing media greatly decreased sensitivity of both the wild type and the rclA-null mutant to sublethal HOCl stress ( Fig. 5A and B; see also Fig. S5A, B, and C in the supplemental material). That copper is uniformly protective for both strains makes it likely that extracellular copper had reacted with and detoxified the HOCl before cells were inoculated into the media. To confirm this result, we tested whether the addition of exogenous copper (0.5 mM CuCl 2 ) could protect E. coli against killing by a very high concentration of HOCl (1 mM). We found that treatment with 1 mM HOCl resulted in complete killing of both strains (Fig. 5C). Both strains survived several orders of magnitude better when 0.5 mM CuCl 2 was added to the media immediately before HOCl stress (Fig. 5C). In addition, colonization of flies by EcN was enhanced when their diet was supplemented with copper (see Fig. S5D), showing that copper can influence the microbiome in vivo, and consistent with the model that copper detoxifies HOCl in the gut. Taken together, these results illustrate that the presence of exogenous copper strongly protects E. coli against HOCl.
RclA protects E. coli against the combination of HOCl and intracellular copper. Next, we sought to investigate how intracellular copper affects the HOCl resistance of E. coli. To address this, we grew wild-type and ΔrclA mutant E. coli strains overnight in minimal media with or without copper before inoculating the strains into copper-free media to perform HOCl-stress growth curves. Growing overnight cultures in media lacking copper was expected to starve the cells for this metal, thereby reducing the concentration of intracellular copper in those cultures. A broad range of HOCl concentrations were assayed to account for quenching of the oxidant by media components.
Consistent with the results shown in Fig. 5, E. coli was more sensitive to inhibition by HOCl in media without copper. The ΔrclA mutant was more sensitive to HOCl than the wild type when grown overnight in copper-containing media, but this phenotype was lost when cells were starved for copper before stress (Fig. 6). The wild-type strain was also slightly more sensitive to some concentrations of HOCl in the presence of intracellular copper (see Fig. S6B and C), but this difference was much more subtle than in the ΔrclA mutant. These results suggest that the physiological role of RclA is to resist the stress resulting from the combination of HOCl and copper in the cytoplasm. They are not consistent, however, with a model where RclA uses Cu to detoxify RCS or other oxidants in the cytoplasm (49), since in that case we would expect the sensitivity of the wild type to decrease to match that of the ΔrclA mutant in the absence of Cu, the opposite of what we actually observed (Fig. 6B).
RclA reduces copper(II) to copper(I). Based on the effect of copper starvation on the HOCl sensitivity of the ΔrclA mutant, the sequence homology between RclA and MerA, and the predicted oxidoreductase activity of RclA (47), we hypothesized that the substrate of RclA might be copper. The reaction between copper and HOCl is known to generate strong oxidizing intermediates, most likely highly reactive Cu(III) (59)(60)(61)(62)(63). HOCl is also capable of oxidizing other transition metals, including iron (64, 65) and manganese (66,67). We therefore measured the specific activity (SA) of purified RclA in the presence of a panel of biologically relevant metals. We also included mercury in the panel of metals because of RclA's homology to MerA, although it is unlikely to be physiologically relevant since we do not expect E. coli to encounter this metal in its environment under normal conditions. We note that the oxidized forms of many transition metals are insoluble in aqueous solution, which limited the set of substrates we could test with this experiment.  (51). In cells containing intracellular copper (A), the ΔrclA mutant has delayed growth relative to the wild type; there is no difference between the strains when the cells were starved for intracellular copper (B). One HOCl concentration is shown for simplicity (see Fig. S6 in the supplemental material for growth curves showing more HOCl concentrations and the statistical analysis comparing LPE values between the strains at each condition). (C) Average LPE values for growth curves shown in panels A and B and statistics comparing wild-type and ΔrclA strains under all conditions. Differences in average LPE values between the strains were analyzed using two-way ANOVA with Tukey's multiple-comparison test.
In the absence of any metal, RclA slowly oxidized NADH (0.0303 mole NAD ϩ min Ϫ1 mg Ϫ1 RclA), consistent with the background activity of other flavin-dependent oxidoreductases in the absence of their specific substrates (68,69). Three of the metals we tested significantly affected RclA SA, as measured by NADH oxidation. Copper and mercury both significantly increased the SA of RclA, whereas zinc caused a decrease in SA (Fig. 7A). As mentioned above, while the present manuscript was in revision, Baek et al. (49) reported crystal structures of E. coli RclA in the presence or absence of bound copper, along with testing a similar panel of metals as the substrates. The results from that study are consistent with ours, supporting our identification of RclA as Cu(II) reductase. Copper is a potent inhibitor of MerA activity (70), further emphasizing the distinct nature of these two enzymes. RclA oxidized NADPH at similar rates to NADH in the absence of metals, but there was no significant increase to SA when copper was added to the reactions. The addition of exogenous thiols, commonly added as ␤-mercaptoethanol (BME), is required for MerA activity (71). To determine whether exogenous thiols increase the reaction rate of RclA, we added 1 mM BME to the RclA reactions. BME rapidly reduced Cu(II) to Cu(I) in the absence of RclA but had no effect on the SA of NADH reduction by RclA with or without the addition of copper (see Fig. S7A and B).
Since RclA is an NADH oxidase, the results shown in Fig. 7A strongly suggested that this enzyme was concurrently reducing copper. Copper exists in four possible oxidation states, Cu(I), Cu(II), and the less common and highly reactive Cu(III) and Cu(IV) states at 37°C using the injector system of a Tecan Infinite M1000 plate reader. All reactions were carried out in 20 mM HEPES-100 mM NaCl (pH 7). NADH absorbance at 340 nm was measured each minute for 5 min. "Buffer only" denotes NAD(P)H oxidation in the presence of the indicated metals, and the "FAD" reactions were performed with 3 M FAD to control for any possible free cofactor that may contribute to metal reduction in the RclA positive reactions. Differences in SA in the presence of each metal were analyzed using a two-way ANOVA with Dunnett's multiple-comparison test using the no metal reaction as the control (****, P Ͻ 0.0001; **, P Ͻ 0.01). (B) Cu(I) accumulates after the RclA and NADH/copper (II) reaction, as measured by BCS/Cu(I)-complex absorption. Each reaction described in panel A was then stopped at 5 min with 10 l of a BCS (400 M final) and EDTA (1 mM final) solution using the injector system of the plate reader. The stopped reaction mixtures were incubated at 37°C for 5 min, with the absorbance of BCS/Cu(I) complex being measured at 483 nm each minute to ensure saturation of BCS. Differences in the amount of BCS/Cu(I) complex between the buffer-only and RclA reactions were analyzed using a two-way ANOVA with Dunnett's multiple-comparison test using the buffer-only sample as the control for each reaction start solution (****, P Ͻ 0.0001; ns, not significant).
Derke et al. (72). The copper salt used in our RclA SA determinations was CuCl 2 , suggesting that RclA was reducing this Cu(II) species to Cu(I). This was initially surprising to us, since Cu(I) is often thought of as a toxic species that causes oxidative stress (54,73). We therefore first sought to validate that RclA was in fact reducing Cu(II) to Cu(I) while oxidizing NADH to NAD ϩ . We measured Cu(I) accumulation in RclA reactions directly using the Cu(I)-specific chelator bathocuproinedisulfonic acid (BCS) (74). NADH spontaneously reduces Cu(II) (75) at rates too slow to impact the measurements made here, but BCS increases the rate of this nonenzymatic copper reduction by shifting the equilibrium of the reaction toward Cu(I) (76). Stopping RclA reactions with a mixture of BCS and EDTA, to chelate any remaining Cu(II), allowed us to observe RclA-dependent Cu(I) accumulation (see Fig. S7C and D). We observed a significant increase in BCS/Cu(I) complex formation only in reaction mixtures containing RclA, NADH, and Cu(II) and not in reaction mixtures lacking any single component ( Fig. 7B; see also Fig. S7C and D in the supplemental material). Furthermore, we validated that the copper reductase activity of RclA was maintained when the reactions were performed in an anaerobic chamber (see Fig. S7E) and that FAD alone did not catalyze Cu(II) reduction (see Fig. S7A and B). Taken together, our results show that RclA has Cu(II) reductase activity and directly demonstrate that RclA generates Cu(I) as a product. This is consistent with the findings of Baek et al. (49), who also used site-directed mutagenesis to clarify the roles of the active site cysteines of RclA in Cu(II) reductase activity and binding sensitivity.
RclA is thermostable and resistant to denaturation by HOCl and urea. We hypothesized that the copper reductase activity of RclA was likely to be relatively stable under denaturing conditions because it must remain active during exposure to HOCl stress, which is known to cause extensive protein misfolding and aggregation in vivo (19,27,28,(32)(33)(34). To test this hypothesis, we first measured RclA activity after treatment with protein denaturing agents (HOCl and urea) in vitro. HOCl treatment (with 0-, 5-, 10-, and 20-fold molar ratios of HOCl to RclA) was performed on ice for 30 min, and urea treatment (0, 2, 4, and 6 M) was carried out at room temperature for 24 h. RclA retained full copper reductase activity at all HOCl levels tested, indicating that it is highly resistant to treatment with HOCl (Fig. 8A). By comparison, the NADH oxidase activity of lactate dehydrogenase was significantly decreased after treatment with a 5-fold excess of HOCl (Fig. 8B). RclA also retained a remarkable 35.8% of full activity after being equilibrated in 6 M urea (Fig. 8C). Finally, we used circular dichroism (CD) spectroscopy to measure the melting temperature (T m ) of RclA, which was 65°C ( Fig. 8D; see also Fig. S7F), indicating that RclA is thermostable relative to the rest of the E. coli proteome, which has an average T m of 55°C (standard deviation [SD] ϭ 5.4°C) (60,77).
RclA influences copper homeostasis, but HOCl stress does not lead to copper export in wild-type E. coli. Since the copper exporters of E. coli (copA and cusCFBA) are upregulated by HOCl treatment (Fig. 4; see also Data Set S2 in the supplemental material) (25) and only transport Cu(I) (54,55,78,79), one possible model for how RclA protects against HOCl is that RclA might facilitate the rapid export of cytoplasmic copper, allowing it to react with and eliminate HOCl outside the cell. To test whether the Cu(II) reductase activity of RclA is important for exporting copper during HOCl stress, we measured intracellular copper concentrations in E. coli MG1655 before and after HOCl stress with ICP mass spectrometry (see Fig. S8A). The copper content of the wild type did not change upon HOCl stress, indicating that copper export is not dramatically upregulated under these conditions. We did find that the ΔrclA mutant contained, on average, more intracellular copper before HOCl stress than did the wild type but that both strains contained similar amounts of copper after HOCl stress. This suggested that RclA has a role in copper homeostasis under nonstress conditions but that any copper export stimulated by HOCl was RclA independent and, in fact, only occurred in the absence of RclA. To attempt to further probe the effect of intracellular copper on HOCl survival, we constructed mutants lacking copA, which is reported to result in increased intracellular copper (55), but unexpectedly found that the copA rclA double mutant had a substantial growth defect in copper-free media, even in the absence of HOCl (see Fig. S8B and C). We do not yet know the explanation for this intriguing result, since there are no known essential copper-containing proteins in E. coli (80), although perhaps the most likely candidate under our growth conditions is the Cu-containing cytochrome bo 3 ubiquinol oxidase CyoB, which is involved in aerobic respiration at high O 2 concentrations (81,82). Our results suggest that there is considerable complexity in the interactions between Cu homeostasis and RclA under different growth conditions, and future work in our laboratory is focused on exploring these interactions in more detail. However, our current results clearly indicate that Cu is important to understanding bacterial HOCl sensitivity and that the Cu(II) reductase RclA is involved in modulating that process.

DISCUSSION
The antimicrobial function of copper in host-microbe interactions is well established (15,17,52,83), although the exact mechanism(s) by which copper kills bacteria remain incompletely known (84,85). In the present study, we identified a new way in which copper toxicity contributes to host-bacterium interactions via its reactions with RCS. We identified RclA as a highly stable Cu(II) reductase (Fig. 7). This is consistent with the simultaneous report by Baek et al. (49), who also found that both Cu(II) and Hg(II) increase the rate of NADH oxidation by RclA. Importantly, we have now shown that RclA is required in vivo for resisting killing by the combination of HOCl and intracellular copper in E. coli (Fig. 6). In the absence of rclA, E. coli had a significant defect in initial colonization that was partially eliminated when production of HOCl by the host was reduced (Fig. 2). The amount of copper in bacterial cells is low (15,16), but how much is unbound by protein and its redox state under different conditions are unknown (15). Given the broad conservation of RclA among host-associated microbes, we propose that there is likely to be a common and previously unsuspected role for copper redox reactions in interactions between bacteria and the innate immune system.
Copper accumulates in host tissues during inflammation (86,87), as do RCS (88, 89). Our discovery that even very low concentrations of extracellular copper can protect  Fig. S7F in the supplemental material) at each temperature used to determine the T m of RclA (65°C). Differences in SA (n ϭ 6, Ϯ the SD) after treatment were analyzed using two-way ANOVA with Sidak's multiple-comparison test for HOCl treatment (A and B) and Dunnett's test using the buffer-only reaction as the control for the urea-treated samples (C) (****, P Ͻ 0.0001; *, P Ͻ 0.05; ns, not significant).

Derke et al.
® bacteria against RCS both in vitro and in vivo adds a new and important facet to understanding copper's role in innate immunity. Since a large proportion of host tissue damage during inflammation is due to HOCl (90,91), the presence of copper in inflamed tissues may play an important role not only in killing bacteria but potentially also in protecting host cells, although this hypothesis will require further testing. Our results also show that media copper concentrations are a key variable in experiments testing the sensitivity of cells to HOCl and that care must be taken to account for media copper content and use metal-free culture vessels in such experiments.
Both HOCl and copper can cause oxidative stress in bacteria and Cu(I) is generally considered more toxic than Cu(II) (15,19,52,83,84,92), so we were initially surprised that a Cu(II) reductase protected E. coli against HOCl. Copper reacts with the ROS hydrogen peroxide (H 2 O 2 ) to form highly reactive hydroxyl radicals in vitro (17,53,73,85), but there is also strong evidence that oxidation is not the major cause of Cu toxicity in E. coli (85,93). CusRS and CueR are exceptionally sensitive to changes in Cu concentrations in the periplasm and cytoplasm, respectively (94,95). CueR, for example, has zeptomolar Cu binding affinity (95). The upregulation of the CusRS and CueR regulons under HOCl stress (Fig. 4) indicates that free Cu is increasing in both the cytoplasm and the periplasm, which could plausibly result from the oxidation of cysteine and histidine residues in Cu-binding proteins by HOCl (96,97). Redox proteomics of RCS-stressed E. coli (98,99) have not identified oxidized Cu-binding proteins, but the methods used in those studies to date are limited to detection of the most common proteins in the cell. Detection of oxidation in less abundant proteins will require more specialized methods (30).
How the presence of copper influences bacterial sensitivity to RCS has not been investigated before this study, but the chemistry of reactions between HOCl and copper is complicated and different from that of reactions between ROS and Cu. HOCl can oxidize Cu(II) to highly reactive Cu(III) (59)(60)(61)(62)(63), and both Cu(I) and Cu(II) are known to catalyze the breakdown of HOCl (59)(60)(61)(62). At nearly neutral pH, similar to that in the large intestine or bacterial cytoplasm, Cu(I) accelerates the rate of decomposition of HOCl to O 2 and chloride ions by as much as 10 8 -fold (60). One possibility to explain the protective effect of RclA is that it might facilitate an HOCl-degrading Cu(I)/Cu(II) redox cycle in the cytoplasm. Alternatively, Baek et al. (49) note the increased O 2 consumption by RclA in the presence of Cu(II), and propose that RclA protects against oxidative stress by lowering O 2 levels. If either of these were the case, however, RclA would require copper to drive HOCl resistance and the ΔrclA mutant would become more sensitive to HOCl in the absence of copper, the opposite of what we actually observed (Fig. 6). We therefore propose that RclA-catalyzed reduction of Cu(II) to Cu(I) may act to limit the production of Cu(III) in the cytoplasm (Fig. 9). Uncontrolled production of Cu(III) could greatly potentiate the ability of HOCl to kill bacterial cells Alternatively, RclA may also ensure that intracellular copper remains in the Cu(I) state, where it can be bound by Cu chaperones like CopA(Z) (100). All of the known proteins involved in Cu homeostasis and export in the E. coli cytoplasm are specific to Cu(I) (52,54,55). Although our data (see Fig. S8A in the supplemental material) indicate that Cu export is not detectably upregulated in HOCl-stressed wild-type cells, we cannot rule out a role for RclA in FIG 9 Proposed model for RclA activity in reducing the toxicity of HOCl. Oxidation of Cu by HOCl can result in the production of the highly-reactive and unstable Cu(III) (72,110). Limiting the amount of cytoplasmic Cu(II) available would prevent the accumulation of Cu(III) during HOCl stress, thereby reducing the toxicity. Converting intracellular Cu(II) to Cu(I) may also facilitate Cu export via CopA(Z) mediated export systems (55).
RclA Protects against HOCl and Intracellular Copper ® maintaining Cu homeostasis under nonstress conditions, especially given the unexpected phenotype of a copA rclA double mutant (see Fig. S8B and C). This is an active area of research in our lab, and we are currently exploring how RclA and the various Cu homeostasis mechanisms of E. coli interact both under nonstress conditions and in the presence of HOCl.
The rate at which RclA oxidized NADH in the presence of copper in vitro was slow (approximately 4.4 min Ϫ1 ) (Fig. 7) (49), suggesting that we have not yet identified optimal reaction conditions for this enzyme. However, expression of rclA is rapidly induced Ͼ100-fold after sublethal doses of HOCl in E. coli ( Fig. 4; see Data Set S2 in the supplemental material) (26), which could compensate in vivo for the low rate of NADH turnover we observed in vitro.
While RclA itself is widely conserved, the rclABCR locus as a whole is restricted to certain enteric proteobacteria, including E. coli, Salmonella, Citrobacter, Raoultella, Serratia, and Shigella. These genera are notable for their close association with gut inflammation and the ability of pathogenic strains to bloom to very high levels in the gut in disease states (2, 3, 7-10, 101, 102). We hypothesize that the ability to survive increased levels of antimicrobial compounds (including RCS) in the inflamed gut is important for the ability of enterobacteria to exploit this niche, and our in vivo results with the ΔrclA mutant generally support this idea (Fig. 2). Many noninflammatory commensal bacteria do encode rclA homologs ( Fig. 1; see Data Set S1 in the supplemental material), including members of the Bacteroidetes, Clostridiaceae, and Lactobacillaceae, where their physiological roles are unknown. Expression of the rclA homolog of the probiotic Lactobacillus reuteri is induced modestly by HOCl, but an L. reuteri rclA mutant is not sensitive to HOCl stress (36). It is unclear, however, whether this is because RclA has a different physiological function in L. reuteri (possibly related to Cu homeostasis) or because RclA requires either strong induction or the presence of RclB and RclC to protect against HOCl under laboratory growth conditions. We do not currently know the physiological roles of RclB, which is a small predicted periplasmic protein, or RclC, which is a predicted inner membrane protein, although deletion of either of these genes results in increased HOCl sensitivity in E. coli (26). We hypothesize that they may form a complex with RclA in vivo and enhance its copper-dependent protective activity and are currently pursuing experiments to test this idea.
Protein expression and purification. Expression of twin-strep-tagged RclA was done in MJG0586 containing pRCLA11 (MJG1338) in M9 minimal media (106) containing 2 g liter Ϫ1 glucose and 100 g ml Ϫ1 ampicillin. Overnight cultures of MJG1338 were diluted 1:100 and grown to an A 600 of 0.4 at 37°C with shaking. When the A 600 reached 0.4, expression was induced with IPTG (isopropyl-␤-D-thiogalactopyranoside; 1 mM final concentration) and allowed to continue for 12 to 18 h at 20°C. Purification of recombinant RclA was achieved to high purity using a 1-ml StrepTrap HP column (GE, 28-9136-30 AC) according to the manufacturer's instructions. Purified protein was subsequently saturated with FAD cofactor by incubating the protein preparation with a 10-fold molar excess of FAD at room temperature for 45 min. Excess FAD and elution buffer were dialyzed away with three exchanges of 1 liter of RclA storage buffer ( Copper(I) quantification. Cu(I) accumulation after the course of the RclA reaction was measured by using bathocuproinedisulfonic acid disodium salt (BCS; Sigma-Aldrich, B1125). NADH oxidation reactions were carried out as in the previous section but were started with either NADH (200 M final), CuCl 2 (200 M final), NADH and CuCl 2 (both at 200 M final), or reaction buffer as a negative control to observe levels of background copper in the reagents used. Each reaction was stopped at 5 min with a BCS (400 M final) and EDTA (1 mM final) solution using the injector system of a Tecan Infinite M1000 plate reader. The stopped reaction mixtures were incubated at 37°C, with the absorbance of BCS/Cu(I) complex being measured at 483 nm every minute for 5 min to ensure complete saturation of the BCS. The amount of the BCS/Cu(I) complex after 5 min was determined using the molar extinction coefficient 13,000 M Ϫ1 cm Ϫ1 (74).
NADH oxidase activity after treatment with urea and HOCl. The SA of RclA after being treated with denaturing agents was measured by determining the SA as described above after incubation of RclA with either urea or HOCl. Melting temperature determination. CD spectra were obtained on a Jasco J815 circular dichroism spectrometer. CD spectra were collected on purified recombinant RclA exchanged into 20 mM HEPES and 100 mM NaCl (pH 7.5). Room temperature CD spectra in the range of 260 to 190 nm were obtained in 0.1-mm demountable quartz cells. Thermal CD data between 30 and 90°C were obtained in standard 1.0-mm quartz cells. All data were collected with a 1.0-nm step size, an 8-s averaging time per point, and a 2-nm bandwidth. Data were baseline corrected against the appropriate buffer solution and smoothed with Jasco software.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.