Discriminating Susceptibility of Xanthine Oxidoreductase Family to Metals

To examine the effect of Cu²⁺ and Cd²⁺ on the activity of the bacterial superoxide dismutase (SOD), we obtained the soluble protein fractions from wild-type (WT) R. gelatinosus cells grown in medium supplemented or not with excess CuSO₄ or CdCl₂. These protein fractions were resolved by non-denaturing PAGE and investigated using an in-gel SOD activity assay. The principle of this assay is based on the ability of superoxide to interact with nitroblue tetrazolium (NBT), reducing the yellow tetrazolium to a blue-purple precipitate within the non-denaturing PAGE gel (29). In this assay, riboflavin is used as a source of free radicals (superoxide) in the presence of light and oxygen. When a SOD is present on the PAGE gel, it appears as an achromatic band, because the SOD converts the superoxide radical into hydrogen peroxide and molecular oxygen. The remaining areas of the gel become blue-purple due to the precipitation of NBT-formazan. As shown in Fig. 1A (left), the assay showed only one achromatic band corresponding to the Fe-SOD (SodB) of R. gelatinosus (30) under all three tested conditions. However, an additional band of high molecular weight (between 130 and 250 kDa) that displayed a strong dark-purple staining, in sharp contrast with the SOD, was also detected. The purple band was present in all soluble fractions including those originating from cells exposed to excess metal (Fig. 1A). This dark purple color demonstrates that this unknown protein or complex can actively reduce NBT to NBT-formazan and suggests that it can generate superoxide radicals. As a proof of the Cu⁺ and Cd²⁺ stress on R. gelatinosus cells, a Western blot (Fig. 1A, right) showed the induction of CopI (a periplasmic protein involved in Cu and Cd homeostasis) by these ions as previously reported (26, 31).

To identify the unknown protein or complex in the purple band, protein identification was performed by nano-liquid chromatography–tandem mass spectrometry (nanoLC-MS/MS) after in-gel trypsin digestion of excised bands. The identified peptide sequences suggested that this band corresponds to a XOR family complex annotated putative mononuclear molybdenum carbon monoxide (CO) dehydrogenase of R. gelatinosus. Indeed, analysis of the MS data (Fig. 1B) showed the presence of the three subunits of a putative CO dehydrogenase (with a predicted mass of 126 kDa, encoded by the coxSLM operon) (Fig. 1C). CoxS is the small subunit encompassing two 2Fe-2S clusters (molecular mass, 17 kDa), CoxL corresponds to the large subunit (Moco subunit) (molecular mass, 84 kDa), and CoxM corresponds to the medium (FAD-containing) subunit (molecular mass, 27 kDa). Interestingly, CoxI (molecular mass, 35 kDa) was also detected and shares similarities with the XdhC protein required for the Moco cofactor insertion and the proper folding of the XOR enzyme (32).

To unambiguously demonstrate that the NBT-reducing active band of R. gelatinosus corresponds to the complex encoded by the coxSLM operon as suggested by the MS analysis, we disrupted the operon by the insertion of a trimethoprim (Tp) resistance cassette within the coding region of coxL. Subsequently, we assayed for the presence of the dark purple active band within the soluble protein extracts. During the previously described in-gel SOD activity assay, riboflavin was used as a source of free radicals (Fig. 2A), which does not allow us to formally confirm that the putative carbon monoxide dehydrogenase complex itself produces superoxide and reduces NBT. We therefore omitted riboflavin from subsequent in-gel assays. As shown in Fig. 2B, whether wild-type cells were untreated or exposed to excess CuSO₄ or CdCl₂, the putative complex was present in the soluble fractions and able to reduce NBT to NBT-formazan in the absence of riboflavin. Under these experimental conditions, the PAGE gel remained unstained and only a faint achromatic band corresponding to SOD could be observed (Fig. 2B). These data demonstrate that R. gelatinosus expressed a high-molecular-weight complex able to reduce NBT. It was also concluded that its activity was not abolished by 100 μM Cu⁺ or Cd²⁺ in the wild-type background.

When the soluble fractions of the wild-type and ΔcoxL strains were compared on the same PAGE gel, the NBT-reducing band was present in the wild-type and absent in the ΔcoxL sample (Fig. 2C), thereby confirming that the NBT-reducing complex is encoded by the coxSLM operon. The CoxSLM complex will be here referred to as pCOD for “putative CO dehydrogenase,” since the function of the complex to oxidize CO or reduce CO₂ has not yet been demonstrated.

The in-gel NBT reductase activity of the pCOD was previously unaffected by the presence of CuSO₄ (Fig. 1A). Since the Cu⁺ detoxification (CopA/CopI) system (21, 31) in the WT had been induced (as shown by increased abundance of CopI on a Western blot [Fig. 1B]), the effect of Cu excess on pCOD activity in the WT background could be underestimated. Because the induced efflux ATPase CopA expels Cu⁺ outside the cytoplasm, it renders the analysis of excess Cu⁺ effects irrelevant in this background. We therefore took advantage of a mutant devoid of the Cu⁺-efflux ATPase CopA that accumulates Cu⁺ in the cytoplasm (21) to measure the effect of cytoplasmic excess Cu⁺ on pCOD activity. For that purpose, the WT and copA-minus mutant cells, which withstand the sublethal concentration of 100 μM CuSO₄ (21), were grown in medium supplemented with or without 100 μM CuSO₄, and the resulting soluble fractions of these cells were analyzed as done previously. To confirm the Cu⁺ stress, Fig. 3A shows the induction of CopI in response to Cu⁺ excess in WT and copA-minus strains. The in-gel activity assay confirmed that for the WT strain, Cu⁺ does not affect the activity of pCOD. The controls (WT) from Fig. 3 are from the same experiments as those shown in Fig. 1 and 2 and are thus being reused. In contrast, for the copA-minus mutant, addition of CuSO₄ to the medium resulted in complete loss of the NBT reductase activity, in contrast to the SOD (Fig. 3B and C). These results showed that excess cytoplasmic Cu⁺ can inhibit in vivo the activity of the pCOD in R. gelatinosus.

The results also raised the question whether Cu⁺ inhibits the assembly or the activity of the enzyme. Indeed, direct Cu⁺ inhibition of Moco synthesis (9, 25) or Fe-S cluster biogenesis (33, 34) has been reported previously. This inhibition of synthesis could be at the origin of the observed copper effect on pCOD. To answer this question, we analyzed in vitro the effect of CuSO₄ treatment on the NBT reductase activity of the pCOD complex in the soluble protein fraction. For that purpose, soluble proteins from WT cells were incubated in buffer supplemented with increasing concentrations of CuSO₄ (from 1.6 to 2,000 μM) for 30 min at room temperature. In this experiment, we used the SOD activity assay as a control to simultaneously compare the effects of excess CuSO₄ on pCOD and on SOD. Similar to the in vivo data, incubation of soluble proteins with increasing concentration of CuSO₄ led to a decreased activity of the pCOD complex in a concentration-dependent manner (Fig. 4). In contrast, excess Cu⁺ only marginally affected the activity of the SOD during the in vitro assay. Altogether, these data showed that Cu⁺ directly inhibits the activity of the assembled pCOD complex.

As copper is biologically and redox active, we questioned whether a redox-nonactive toxic metal such as cadmium could inhibit the activity of the pCOD complex of R. gelatinosus. To address this question, we used the mutant devoid of the Cd²⁺-efflux ATPase CadA to assess the effect of cytoplasmic accumulation of Cd²⁺ on the pCOD activity. While WT R. gelatinosus can tolerate up to 3 mM CdCl₂ in its growth medium, the ΔcadA mutant is limited to low concentrations of up to 200 μM (26). In subsequent assays, the WT and ΔcadA mutant cells were grown in the presence of 100 μM CdCl₂. The activity of pCOD in the soluble fractions of these cells was then analyzed by non-denaturing PAGE as described above. The in-gel activity confirmed that for the WT strain, Cd²⁺ does not affect the activity of pCOD, very likely because Cd²⁺ is expelled outside the cytoplasm. To confirm the Cd²⁺ stress, Fig. 5A shows the induction of CopI in response to Cd²⁺ excess in a ΔcadA strain as previously reported (26). In contrast, for the ΔcadA mutant, addition of 100 μM CdCl₂ to the medium resulted in a drastic loss of the NBT reductase activity (Fig. 5B and C). These results clearly showed that Cd²⁺ can inhibit the activity of the R. gelatinosus pCOD complex in vivo. Similarly, to show that Cd²⁺ affects the assembled pCOD complex, we analyzed the in vitro effect of CdCl₂ on the NBT reductase activity of the pCOD complex and on the SOD activity in enriched soluble protein fractions. In this experiment, soluble proteins from WT cells were incubated in buffers supplemented with increasing concentrations of CdCl₂ (from 0 to 2,000 μM) at room temperature for 30 min. Similar to the in vivo data, incubation of soluble proteins with increasing concentrations of CdCl₂ led to a decline in the activity of the pCOD complex (Fig. 5D), while cadmium did not affect the activity of the SOD, similarly to copper excess. We should note, however, that the effect of Cd²⁺ in vitro was not as strong as the effect of Cu⁺ (Fig. 4). Nevertheless, taken together the data strongly suggested that Cd²⁺ directly inhibits the activity of the assembled pCOD complex.

In an independent study, we sought to identify metal stress-responsive and/or target proteins in R. gelatinosus, through a quantitative differential shotgun proteomics (35) analysis of cells grown either with 100 μM CuSO₄ or 100 μM CdCl₂ or under FeSO₄-limited conditions (unpublished data). Here, we extracted the specific peptide signals of the different subunits of the pCOD complex and compared the abundances of these proteins in WT, copA-minus, and ΔcadA cells grown in the presence or absence of metals. The abundance of the three subunits decreased significantly by as much as 17-fold for CoxL, 4-fold for CoxS, and 13-fold for CoxM in the Cu⁺-efflux ATPase mutant copA-minus strain, when exposed to 100 μM CuSO₄, compared to their abundance in unexposed control cells (Table 1). Similar fold decreases in the abundance of CoxLMS proteins were also observed in the ΔcadA mutant in the presence of 100 μM CdCl₂, compared to the unexposed control cells. Conversely, but as expected, proteins required for metal homeostasis, such as CopA, CopI, CopR, and CadA, were strongly increased in abundance in stressed cells compared to the control (Table 1). These data confirmed that excess Cu⁺ and Cd²⁺ affected the amount of pCOD complex subunit, in agreement with the decrease or loss of its activity in cells facing excess Cu⁺ or Cd²⁺.

Non-denaturing PAGE preserves the native state of protein complexes, enabling subsequent identification using activity assays. To identify the substrate of the newly identified pCOD complex, we tested different molecules for their ability to significantly increase the activity of the pCOD-containing polyacrylamide band (Fig. 6A). Different aldehydes (formaldehyde, acetaldehyde, n-octyl-aldehyde, and butyraldehyde) did not change the activity of the pCOD complex. In CO₂-enriched growth medium, the activity of the pCOD was slightly increased, suggesting that the pCOD could have a CO/CO₂ dehydrogenase/reductase activity. Yet, further experiments are required to confirm this observation. Interestingly, the addition of xanthine to the native PAGE gel revealed a second active band of approximately 250 kDa. These data strongly suggested that this new complex corresponds to the xanthine dehydrogenase (XDH) of R. gelatinosus. An operon encompassing three genes (xdhABC) that could encode the three subunits of the xanthine dehydrogenase in R. gelatinosus was therefore cloned from the genome (Fig. 6B). To identify the activity of the complex encoded by these genes, an ΔxdhB deletion mutant was generated. Soluble fractions from the WT and the ΔxdhB mutant were assayed for the presence of the active band in the presence of xanthine as the substrate. As shown in Fig. 6C, the pCOD band was revealed in both samples, while the band corresponding to the XDH was absent in the ΔxdhB deletion mutant, suggesting that the 250 kDa complex corresponds to the XDH complex. According to the theoretical molecular weight of the three subunits (XdhB, 82 kDa; XdhA, 54 kDa; and XdhC, 27 kDa), the XDH complex would be a homodimer on the native gel, which has been observed for other bacterial and mammalian XDH enzymes.

It has been reported that compounds containing copper or zinc are potential inhibitors of xanthine XOR/XDH activity in vitro using mouse liver tissue homogenate (36). However, direct evidence showing an inhibitory effect of metals on XDH activity in vivo was still lacking. We therefore tested the effect of excess Cu⁺ or Cd²⁺ on the activity of the cytosolic XDH in vivo by growing WT, copA-minus, or ΔcadA mutant R. gelatinosus cells with or without excess Cu or Cd. The soluble fractions were analyzed on non-denaturing PAGE gels for XDH NBT reductase activity. In the WT background, XDH activity was unaffected by the addition of Cu or Cd (Fig. 7A), similarly to pCOD activity. In contrast, in the strains that accumulate Cu⁺ or Cd²⁺, the activities of both pCOD and XDH were strongly inhibited by both ions (Fig. 7B). These data clearly showed that Cu⁺ and Cd²⁺ inhibit the activity of different enzymes in the bacterial XDH/XOR family. We also sought to verify whether mammalian XDH activity could be compromised by exposure to Cu⁺ or Cd²⁺. In mammalian cells, xanthine oxidase (XO) and xanthine dehydrogenase (XDH) are interconvertible forms of the same enzyme, known as xanthine oxidoreductase (XOR) (28, 37). The xanthine oxidoreductase activity of bovine milk XOR was tested either in liquid or by non-denaturing PAGE in the presence of xanthine with increasing concentrations of CuSO₄ or CdCl₂. During the in-liquid reactions, as expected the untreated samples showed the formation of a dark blue color formed due to the interaction of superoxide ions with NBT (Fig. 8A). Low concentrations of CuSO₄ in the reaction buffer generated a light blue color, demonstrating a decrease in the enzyme activity. No color was developed in the samples exposed to 250 or 500 μM CuSO₄. Similarly, on the non-denaturing PAGE gel, CuSO₄ affected the activity of the enzymes (XO and XDH) in a concentration-dependent manner (Fig. 8B). We also noticed the presence of protein aggregates (high-molecular-weight bands) in the presence of metals. In both assays, copper effectively inhibited the XOR activity, but the mammalian XOR was not affected or only very slightly affected by Cd²⁺ under these assay conditions, in contrast to the bacterial XDH. Altogether, our data thoroughly support the inhibition of pCOD and XDH by excess Cu or Cd, thus illustrating the harmful effect of metals on the XOR family enzymes.

R. gelatinosus S1 (47) was grown overnight at 30°C in malate medium under photosynthetic condition (light and microaerobiosis, in filled tubes with residual oxygen in the medium). E. coli was grown overnight at 37°C in LB medium. Antibiotics were added at the indicated concentrations when appropriate: kanamycin (50 μg/mL) and trimethoprim (50 μg/mL). Bacterial strains and plasmids used in this study are listed in Table 2. Standard methods were performed according to the work of Sambrook et al. (48) unless indicated otherwise. To inactivate coxL, a 2.2-kb fragment was amplified using the primers coxLF (5′-CTACCAGAACATCGTCAAGG-3′) and coxLR (5′-CATCGTGTAGTCCATGTAGC-3′) and cloned into the PCR cloning vector pGEM-T to give pGcoxL. The coxL gene was then inactivated by deletion of a 279-bp BglII fragment and the insertion of the 0.7-kb Tp cassette (49) at the BglII site within the coxL coding sequence. The resulting recombinant plasmid was designated pGcoxL::Tp. This plasmid was used to transform the WT strain. To disrupt xdhB, a 1,711-bp fragment was amplified using the primers xdhBF (5′-CAACTGGCCGAGATGTGGCG-3′) and xdhBR (5′-CCAGTAGCAGTTGTCGAAGT-3′) and cloned into the PCR cloning vector pGEM-T to give pGxdhB. The xdhB gene was then inactivated by insertion of the 0.7-kb Tp cassette at the MscI site within the xdhB coding sequence. The resulting recombinant plasmid was designated pGxdhB::Tp. This plasmid was used to transform the WT strain. Transformants were selected on malate plates supplemented with the appropriate antibiotics (trimethoprim or kanamycin) under photosynthesis conditions. Following transformant selection, template genomic DNA was prepared from the ampicillin-sensitive transformants and confirmation of the antibiotic resistance marker’s presence at the desired coxL or xdhB locus was performed by PCR.

Cells were disrupted by sonication in 50 mM Tris-HCl buffer (pH 7.6) containing 1 mM phenylmethylsulfonyl fluoride. Unbroken cells and membranes were removed by ultracentrifugation (200,000 × g, 90 min, 4°C) to collect the soluble fraction. Protein concentration was estimated using the bicinchoninic acid assay (Sigma), with bovine serum albumin as the standard. For soluble protein metal treatment, required concentrations of metal solution were mixed with a total protein concentration of 20 mg/mL at room temperature (RT) for 30 min.

Forty micrograms of soluble proteins was separated on a 12% non-denaturing polyacrylamide gel and stained for SOD activity as described in the work of Weydert and Cullen (29), with minor modifications. Nitroblue tetrazolium (NBT) (2 mg/mL) was added first for a 15-min incubation in the dark at RT followed by the addition of TEMED (N,N,N′,N′-tetramethylethylenediamine) (0.85%) and riboflavin-5-phosphate (56 μM) for 15 min in the light. The gel was washed twice in double-distilled water (ddH₂O) and then left in ddH₂O at RT on a light box until SOD-positive staining appeared. For NBT reductase activity, the gel was incubated in 50 mM Tris-HCl buffer (pH 7.6) supplemented only with NBT (2 mg/mL) for 15 to 30 min in the dark at RT.

Peptide mixtures generated by standard in-gel digestion with trypsin (Gold; Promega) were SpeedVac treated for 10 min and then analyzed with a quadrupole time of flight (QTOF) Premier mass spectrometer coupled to an Acquity nano-ultraperformance liquid chromatograph equipped with a trapping and an analytical column as described in reference 31.

Bacterial pellets were diluted in 1× LDS (Thermo) buffer at a ratio of 7.5 μL/mg. They were heated for 10 min at 99°C, sonicated for 10 min, and then subjected to bead-beating lysis as previously described (50). The extracted proteins (25 μL) were subjected to denaturing electrophoresis onto a NuPAGE 4 to 12% Bis-Tris gel (Thermo) for 4 min. For each sample, a single gel band containing the whole cellular proteome was excised and processed as described previously (51) for in-gel trypsin proteolysis. Peptides were analyzed using a Q-Exactive HF tandem mass spectrometer (Thermo) coupled to an Ultimate 3000 nanoRSLC nanoLC system (Thermo) in data-dependent acquisition mode as previously described in reference 52.

XO from bovine milk was purchased from Sigma-Aldrich Co. (reference no. X1875; ≥0.4 units/mg protein units/mL) and diluted in 50 mM Tris-HCl buffer (pH 7.6). For the assay, 100 μL of XOR (0.01 U/mL) in the same buffer was mixed or not with increasing CuSO₄ concentrations. After 30 min of incubation at room temperature, 50 μL of buffer containing NBT (2 mg/mL) and xanthine (2 mmol/L) was added to the XOR samples to start the XOR reaction.

Equal amounts of soluble proteins (20 μg) were separated on SDS-PAGE gels and transferred onto a Hybond ECL polyvinylidene difluoride membrane (GE Healthcare). The membrane was then probed with His-Probe-HRP (horseradish peroxidase; from Pierce) according to the manufacturer’s instructions. His-Probe-HRP allows detection of CopI and SodB, two naturally histidine-rich proteins in the R. gelatinosus soluble fraction (30). Positive bands were detected using a chemiluminescent HRP substrate according to the method of Haan and Behrmann (53). Image capture was performed with a ChemiDoc camera system (Bio-Rad).

The bands in the in-gel assay were quantified using the ImageJ program. The relative amount was calculated based on the signal obtained in the untreated sample. Statistical significance was calculated using an ordinary one-way analysis of variance (ANOVA)/GraphPad Prism (ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

The mass spectrometry proteomics data have been deposited in the Figshare repository with the data set identifier https://doi.org/10.6084/m9.figshare.23615265.