A novel iron- and copper-binding protein in the Lyme disease spirochaete

Authors


  • The authors declare no competing financial interests.

For correspondence. E-mail xin.li@cvm.osu.edu; Tel. (614) 292 6524; Fax (614) 292 6473.

Summary

Iron and copper are transition metals that can be toxic to cells due to their abilities to react with peroxide to generate hydroxyl radical. Ferritins and metallothioneins are known to sequester intracellular iron and copper respectively. The Lyme disease pathogen Borrelia burgdorferi does not require iron, but its genome encodes a ferritin-like Dps (DNA-binding protein from starved bacteria) molecule, which has been shown to be important for the spirochaete's persistence in the tick and subsequent transmission to a new host. Here, we show that the carboxyl-terminal cysteine-rich (CCR) domain of this protein functions as a copper-binding metallothionein. This novel fusion between Dps and metallothionein is unique to and conserved in all Borrelia species. We term this molecule BicA for Borrelia iron- and copper-binding protein A. An isogenic mutant lacking BicA had significantly reduced levels of iron and copper and was more sensitive to iron and copper toxicity than its parental strain. Supplementation of the medium with iron or copper rendered the spirochaete more susceptible to peroxide killing. These data suggest that an important function of BicA is to detoxify excess iron and copper the spirochaete may encounter during its natural life cycle through a tick vector and a vertebrate host.

Introduction

The unique redox properties of iron (Fe) and copper (Cu) make them ideal cofactors for many enzymes, especially the oxidoreductases (Waldron et al., 2009). However, these metals can also be toxic to cells due to their abilities to react with peroxide (Crichton et al., 2002; Dupont et al., 2011). Therefore, not only is Fe and Cu homeostasis tightly controlled at the level of import and export, cellular Fe and Cu ions are also sequestered or bound by proteins, making free ions virtually undetectable (Rae et al., 1999). Ferritins and metallothioneins are the most representative proteins that sequester intracellular Fe and Cu respectively (Andrews, 2010; Capdevila and Atrian, 2011).

The DNA-binding protein from starved bacteria (Dps) is a dodecameric ferritin-like molecule found in diverse bacterial species (Almiron et al., 1992; Chiancone and Ceci, 2010; Haikarainen and Papageorgiou, 2010). While not all Dps proteins have a DNA-binding domain or bind DNA, as suggested by the name, Fe mineralization is an activity conserved in all Dps molecules that have been characterized to date (Chiancone and Ceci, 2010; Haikarainen and Papageorgiou, 2010).

Borrelia burgdorferi (Bb), the causative agent of Lyme disease, has evolved to be independent of Fe for growth (Posey and Gherardini, 2000). The genome of this bacterium, however, encodes a Dps homologue (Fraser et al., 1997) capable of binding Fe (Li et al., 2007; Codolo et al., 2010). Mutagenesis and complementation studies have shown that BicA is critically important for the spirochaete's persistence in the tick vector and its subsequent transmission to a new host (Li et al., 2007). The presence of this Fe-binding Dps molecule in a bacterium that does not require Fe for growth suggests that its main function is for detoxification not for storage.

A comparison with other members of the Dps family has revealed a unique carboxyl-terminal cysteine-rich (CCR) domain in the Dps homologues from all Borrelia species, including both the Lyme disease and the relapsing fever spirochaetes (Fig. S1). Here, we report that this CCR domain functions as a Cu-binding metallothionein. We term this molecule BicA for Borrelia iron and copper-binding protein A to highlight the uniqueness of this novel fusion between a ferritin-like Dps and a Cu thionein.

Results

BicA-deficiency results in reduced levels of iron and copper in Bb

To determine the effect of BicA on metal homeostasis, we compared the metal content of an infectious Bb B31 strain with that of an isogenic bicA mutant constructed previously (Li et al., 2007) using inductively coupled plasma-sector field mass spectrometry (ICP-SFMS), also known as high resolution (HR)-ICP-MS (Jakubowski et al., 2011a,b). Spirochaetes were grown at 33°C to early stationary phase in standard BSK-H complete media (Pollack et al., 1993). Cellular levels of the six most common transition metals found in metalloproteins (Dudev and Lim, 2008; Waldron et al., 2009) – manganese (Mn), Fe, cobalt (Co), nickel (Ni), Cu, and zinc (Zn) – were measured (Fig. 1A). Analysis of the wild-type (WT) B31 showed that the Mn, Fe, Cu and Zn levels are significantly higher than the Co and Ni levels (P < 0.001, one-way anova with Tukey's multiple comparison test). In comparison with the WT, the bicA mutant had similar levels of Mn, Zn, Co and Ni, but a 49% reduction in Fe and a 96% reduction in Cu (Fig. 1A).

Figure 1.

Fe and Cu are present in Bb and their homeostasis is affected by a deficiency in BicA or BmtA. Cellular levels of Mn, Fe, Co, Ni, Cu and Zn in (A) an infectious B31 strain and its isogenic bicA mutant and (B) an infectious 297 strain and its isogenic bmtA mutant were determined by ICP-SFMS. Spirochaetes were grown at 33°C to early stationary phase in the BSK-H media. Data represent mean (+ s.e.m.) of three independent samples. P-values are derived from a two-way anova followed by Bonferroni post-test: ***P < 0.001; ns, P > 0.05.

Fe and Cu have never been reported as major transition metals in Bb. Given the complexity of the BSK-H medium, which contains bovine serum albumin, rabbit serum, peptone, and yeast extract, the Fe and Cu detected here could be due to non-specific association of serum proteins with the cell surface. Therefore, we further determined whether the presence of Fe and Cu is dependent on Borrelia metal transporter A (BmtA) by comparing the metal content of an infectious 297 strain with that of an isogenic bmtA mutant (Ouyang et al., 2009). Like B31, the WT 297 strain also had much higher levels of Mn, Fe, Cu and Zn than Co and Ni (P < 0.001, one-way anova with Tukey's multiple comparison tests) (Fig. 1B). The bmtA mutant, in comparison with its parental 297 strain, had significantly reduced levels of Mn (99%) and Cu (90%), significantly increased levels of Zn (1.5-fold) and Fe (2.8-fold), and similarly low levels of Co and Ni (Fig. 1B). Notably, these data independently confirmed a previous finding by Ouyang and colleagues that the bmtA mutant had a reduced level of Mn but an increased level of Zn (Ouyang et al., 2009). These data further showed that Cu and Fe, which were not reported in the previous study, were readily detectable in the WT 297 strain, and that the bmtA mutant had a reduction in the Cu level but an increase in the Fe level (Fig. 1B). The altered Fe and Cu homeostasis due to an absence of BmtA, a metal transporter, suggests that these metals may be internalized rather than being non-specifically associated with the cell surface.

Mn is the most concentrated transition metal in the spirochaete

We next determined the concentration of these transition metals in the BSK-H medium used to cultivate the spirochaete (Fig. 2A). Of the four major transition metals found in Bb, Fe was detected at the highest level (26 μM) in the medium, which is followed by Zn at 6 μM, Cu at 2 μM and Mn at 0.1 μM. To calculate the relative fold of concentration increase in the cell, we converted the number of atoms per cell to μM based on the assumption of an average volume of 10−9 μl for a spirochaete (a spiral cylinder of 0.25 μm in diameter and 20 μm in length). Interestingly, the fold of concentration increase in the cell follows the exact reverse order of the abundance of these metals in the medium, with Mn being the most concentrated in the cell and followed in order by Cu, Zn and Fe (Fig. 2B). The significant increase in Mn concentration (∼ 3300-fold) and the modest increase in Fe concentration (∼ 10-fold) are consistent with a more prominent role for Mn than Fe in this bacterium. Instead of Fe, the superoxide dismutase and the peptide deformylase in Bb have been shown to use Mn and Zn respectively (Nguyen et al., 2007; Troxell et al., 2012).

Figure 2.

Mn is the most concentrated transition metal in Bb.

A. Metal concentrations in standard BSK-H complete medium used for cultivation of Bb.

B. Fold of increase in metal concentration in the cell as compared with that in the medium. Calculation of cellular metal concentration was based on an estimated volume of 10−9 μl for each spirochaete.

BicA functions as a Cu(I)-binding metallothionein

Given the ferritin nature of all Dps molecules, it was not unexpected that the bicA mutant had a reduced level of Fe. However, the bicA mutant had a much more significant reduction in Cu than in Fe (96% vs 49%). Therefore, we investigated whether the CCR domain of BicA (Fig. 3A) could function as a Cu-binding metallothionein.

Figure 3.

The CCR domain of BicA functions as a Cu-binding metallothionein.

A. A schematic illustration of BicA showing the primary sequence of the CCR domain as well as the five residues constituting the ferroxidase centre, which was revealed in a crystal structure of this protein (Codolo et al., 2010).

B. A TEM image of purified recombinant BicA. Bar, 20 nm.

C. Fluorescence emission scan (left) and emission maximum at 600 nm (right) of the Cu(I) : BicA complex at molar ratios of 2, 4, 6, 8, 10, 12, 16, 18, 20, 30, and 40. Data are representative of three independent experiments.

D. Coomassie blue staining of the proteins (left) and Prussian blue staining of Fe (right) following native polyacrylamide gel electrophoresis (PAGE) of the Fe-loaded proteins.

E. Fluorescence emission scan of WT and mutant BicA at 5, 15, 25, 35, 45, 55, and 65 min after the addition of Cu.

F. The Cu(I)-binding capacities of the GST fusions of WT and mutant BicA. Data represent mean (+ s.e.m.) of two independent samples.

Like all other Dps homologues (Chiancone and Ceci, 2010; Haikarainen and Papageorgiou, 2010), recombinant BicA exists as dodecamers (Li et al., 2007; Codolo et al., 2010), which can be seen as ∼ 10 nm diameter particles under a transmission electron microscope (TEM) (Fig. 3B). We utilized a characteristic fluorescent signal emitted by Cu(I)-metallothionein complex to detect Cu(I)-binding by BicA (Lerch, 1980; Beltramini and Lerch, 1981). Gold and colleagues showed that in the presence of a reducing agent such as DTT or ascorbate, Cu(II) could be used as a Cu(I) donor (Gold et al., 2008). Calderone and colleagues also succeeded in incorporating Cu(I) into a synthetic apopeptide of yeast Cu thionein by supplying Cu(II) in the presence of excessive β-mercaptoethanol (Calderone et al., 2005). Therefore, we established a Cu(I)-specific fluorescence assay that can be performed under the atmospheric air, bypassing the need for an anaerobic chamber (see Experimental procedures for details). Upon incubation with increasing molar ratios of Cu(I), recombinant BicA protein yielded fluorescent signals characteristic of Cu(I)-thiolate core formation (Beltramini and Lerch, 1981; Gold et al., 2008), which reached a plateau at the ratio of 10 atoms of Cu per monomer (Fig. 3C).

To demonstrate that it is the CCR domain and particularly the cysteines of the CCR domain that are responsible for Cu(I)-binding, we constructed mutants that either lacked the entire CCR domain (ΔCCR) or had cysteine-to-alanine replacement in all 6 cysteines of the CCR domain (CAR6). These mutations did not compromise the dodecameric structure of the molecule, which was evident in their migration pattern on a native polyacrylamide gel (Fig. 3D) and confirmed by TEM and chromatographic analyses (data not shown). The structural integrity of the ΔCCR and the CAR6 mutants was further demonstrated in their ability to mineralize Fe (Fig. 3D). The Cu(I)-specific fluorescent signal, however, was not detected in either the ΔCCR or the CAR6 mutants (Fig. 3E). Addition of six molar excess of cysteine to the ΔCCR mutant reaction failed to restore the fluorescent signal (Fig. S2). Therefore, it is the cysteine residues in the context of the CCR domain that are responsible for the Cu(I)-specific fluorescent signal.

Next, we determined whether Cu(I)-binding by BicA is independent of the Dps domain. We targeted residues D65 and E69 for mutagenesis because they have been shown in a recent crystal structure of BicA (Protein Data Bank: 2PYB) to directly co-ordinate Fe ions in the ferroxidase centre (Codolo et al., 2010) and because these residues are conserved in the ferroxidase centre of all Dps molecules that have been structurally characterized (Chiancone and Ceci, 2010). The D65N&E69Q mutant failed to mineralize Fe (Fig. 3D), but was still capable of binding Cu (Fig. 3E). The D65N&E69Q&ΔCCR mutant, as expected, did not mineralize Fe or bind Cu (Fig. 3D and E). These data are consistent with a previous study showing that mutations targeting the ferroxidase centre of a Dps molecule abolished its ability to mineralize Fe (Pulliainen et al., 2005). In conclusion, Cu(I)-binding by the CCR domain and Fe mineralization by the Dps domain are two independent activities of BicA.

Finally, we quantified the Cu(I)-binding capacity of the WT and the mutant BicA by ICP-SFMS. Since the glutathione S transferase (GST) fusion of BicA also formed functional dodecamers that bind Fe and Cu (Fig. S3), we utilized the GST-tag to affinity-capture protein-bound Cu. As purified, the GST fusions of WT and mutant BicA contained a low level of Fe but not any significant level of Cu (Fig. S4). After incubation with Cu(I), the GST fusions of the WT and the D65N&E69Q mutant captured as many as seven atoms of Cu per monomer, whereas the GST fusions with the ΔCCR mutation captured only one atom of Cu per monomer (Figs 3F and S4). These data indicate that a single CCR domain is capable of binding approximately six atoms of Cu, and a dodecameric BicA complex is capable of binding 72 atoms of Cu.

Collectively, these biochemical analyses demonstrate that the CCR domain of BicA indeed functions as a Cu(I)-binding metallothionein, explaining the reduced level of Cu in the bicA mutant. Therefore, BicA represents a novel fusion between a Cu thionein and a ferritin-like Dps molecule.

BicA protects the spirochaete against Fe and Cu toxicity

We next examined whether BicA has a role in protecting the spirochaete against Fe or Cu toxicity. The WT and the bicA mutant spirochaetes were grown at 33°C in standard BSK-H complete media (Pollack et al., 1993) either not supplemented or supplemented with increasing concentrations of Fe or Cu, and the number of viable spirochaetes in each culture was determined by the limiting dilution method and the percentage of growth in Fe or Cu supplemented media was normalized by growth in media without supplementation (see Experimental procedures for details). As shown in Fig. 4, the bicA mutant is significantly more sensitive to Fe and Cu than its parental WT strain. Although the difference between these two strains was only evident within a narrow range of Fe or Cu concentration, the magnitude of the difference was quite significant – the mutant was > 3 orders more sensitive than the WT. The reduced Fe and Cu levels in the bicA mutant (Fig. 1A) suggest that the spirochaete may have other mechanisms to alleviate metal toxicity, for example, by controlling metal influx and efflux. Therefore, the contribution of BicA sequestration of excess Fe or Cu to alleviating the toxicity of these metals would only be evident under conditions where the spirochaete's ability to maintain metal homeostasis through import and export is being severely challenged but the spirochaete is not yet killed.

Figure 4.

BicA protects the spirochaete against Fe and Cu toxicity. The WT (solid circles) and the bicA mutant (open circles) spirochaetes were cultured in the BSK-H media either not supplemented or supplemented with (A) Fe or (B) Cu at the indicated concentrations. The number of viable spirochaetes in each culture was determined by the limiting dilution assay as described in Experimental procedures. The percentage of growth was normalized against growth in media not supplemented with Fe or Cu. Data represent mean (+ s.e.m.) of three independent measurements. P-values are derived from a two-way anova followed by Bonferroni post-test, and only P-values of statistical significance are indicated: *P < 0.05, ***P < 0.001.

Fe and Cu supplementation in media renders the spirochaete more sensitive to peroxide stress

Fe and Cu toxicity is often attributed to their ability to react with peroxide and generate hydroxyl radicals (Crichton et al., 2002; Dupont et al., 2011). Here, we examined the effect of Fe and Cu supplementation in media as well as the effect of BicA-deficiency on spirochetal resistance to peroxide stress. The WT and the bicA mutant spirochaetes were grown either in standard BSK-H complete media or in media supplemented with 20 μM Fe or 100 μM Cu, and then tested for their ability to survive treatment with 0.25, 0.5, 1, 2, and 4 mM of hydrogen peroxide (H2O2).

After being cultivated in media supplemented with either Fe or Cu, the WT spirochaete became significantly more sensitive to H2O2 (Fig. 5A). In comparison, media supplementation with Fe or Cu did not have as a significant effect on the bicA mutant (Fig. 5B). Fe supplementation appeared to render the mutant more resistant to H2O2, but the difference was only statistically significant at 2 mM of H2O2. Cu supplementation appeared to render the mutant more sensitive, but the difference was only statistically significant at 4 mM of H2O2.

Figure 5.

The spirochaete becomes more sensitive to H2O2 after growing in media supplemented with Fe or Cu. The WT (solid symbols) and the bicA mutant (open symbols) were grown to early stationary phase in the BSK-H media either not supplemented (circles) or supplemented with 20 μM of (NH4)2Fe(SO4)2 (triangles) or 100 μM of CuSO4 (diamonds). Spirochaetes were harvested and assayed for resistance to H2O2 as described in the Experimental procedures. Data represent mean (+ s.e.m.) of two independent cultures. P-values are derived from a two-way anova followed by Bonferroni post-test, and only P-values of statistical significance are indicated: *P < 0.05, **P < 0.01, ***P < 0.001. P-values in panels A–B are for comparisons between spirochaetes grown in media supplemented with Fe or Cu and spirochaetes grown in media not supplemented. P-values in panels C–E are for comparisons between the WT and the bicA mutant.

Consistent with our previous report (Li et al., 2007), the loss of BicA had no adverse effect on the spirochetal resistance to H2O2 (Fig. 5C). To the contrary, when cultivated in media supplemented with Fe or Cu, the bicA mutant was actually more resistant to H2O2 than the WT (Fig. 5E and F). These data appeared to be inconsistent with the established role of Dps in protecting bacteria against oxidative stress (Chiancone and Ceci, 2010; Haikarainen and Papageorgiou, 2010), which could be attributed to the Cu(I)-binding activity of BicA.

Spirochetal resistance to peroxide stress correlates inversely with cellular Cu level

We next determined the effect of media supplementation with Fe or Cu on the levels of transition metals in the spirochaete. For the WT, Fe supplementation in the media resulted in a significant increase of Fe in the spirochaete, but no significant change in other metals (Fig. 6). When cultivated in media supplemented with 20 μM Fe, the bicA mutant did not accumulate more Fe, but instead had significantly higher levels of Mn and Co. This could be due to misplacement of Fe in (or Fe poisoning of) cellular enzymes that normally utilize Mn or Co, which drives further import of these metals. Notably, these data, again, showed that the bicA mutant had significantly reduced levels of Fe than the WT whether the spirochaetes were grown in the standard media or in media supplemented with Fe (Fig. 6B), suggesting that in the absence of BicA, the spirochaete could not accumulate excess Fe.

Figure 6.

Metal homeostasis in the spirochaete is affected by media supplementation with Fe or Cu. The WT (solid bars) and the bicA mutant (open bars) were grown to early stationary phase in the BSK-H media either not supplemented or supplemented with 20 μM Fe or 100 μM Cu. Spirochaetes were harvested and the cellular levels of (A) Mn, (B) Fe, (C) Co, (D) Ni, (E) Cu and (F) Zn were determined by ICP-SFMS as described in Experimental procedures. Data represent mean (+ s.e.m.) of two independent cultures. P-values are derived from a two-way anova followed by Bonferroni post-test, and only P-values of statistical significance are indicated: *P < 0.05, **P < 0.01, ***P < 0.001.

Cu supplementation resulted in a higher level of Cu in the bicA mutant but not in the WT, suggesting that the Cu level in the WT may have already reached saturation (Fig. 6E). Supplementation of media with Cu affected the levels of other metals in the spirochaete: the bicA mutant had a decrease in Zn (Fig. 6F); the WT had an increase in Mn (Fig. 6A) but a reduction in Fe (Fig. 6B). Interestingly, it has been suggested that an increased Mn/Fe ratio could be an indicator for bacterial adaptation to peroxide stress (Faulkner and Helmann, 2011). Our data showed that the WT spirochaetes cultivated in Cu-supplemented media were indeed the most susceptible to peroxide stress (Fig. 5).

These data clearly indicate that the homeostasis of one metal is intrinsically linked to the homeostasis of other metals in the cell. To systematically sort out the relationship between cellular metal content and peroxide resistance, we performed a matrix correlation analysis of the cellular levels of various transition metals and the spirochaete's survival rates at various concentrations of H2O2 (Table S1). This analysis revealed a strong negative correlation between spirochetal resistance to peroxide stress and the cellular level of Cu (Fig. 7A). Due to a negative correlation between the cellular levels of Cu and Ni (Table S1), there was also a positive correlation between spirochetal resistance to peroxide stress and the cellular level of Ni (Fig. 7B).

Figure 7.

Spirochetal resistance to peroxide stress correlates with cellular levels of Cu and Ni. Matrix correlation analysis indicates (A) a strong inverse correlation between the spirochetal resistance to H2O2 and the cellular Cu level, and (B) a strong positive correlation between the spirochetal resistance to H2O2 and the cellular Ni level. The Spearman correlation efficient and P-value for all correlations are presented in Table S1.

In summary, our data show that Fe and Cu are present in the Lyme disease spirochaete. Fe and Cu homeostasis in the spirochaete is altered by a deficiency in BmtA, a metal transporter, and by a deficiency in BicA, which is a novel fusion between the ferritin-like Dps molecule and a Cu thionein. BicA protects the spirochaete against Fe and Cu toxicity, but Cu accumulation by BicA renders the spirochaete more susceptible to peroxide killing.

Discussion

Although eukaryotic metallothioneins have been studied extensively since the discovery of horse cadmium-binding protein in 1957 (Margoshes and Vallee, 1957; Lerch, 1980; Vasak and Meloni, 2011), bacterial metallothioneins are rare (Blindauer, 2011), with only two types characterized to date: the Zn-binding bacterial metallothionein (BmtA) family that was first discovered in a cyanobacterium (Olafson et al., 1988; Huckle et al., 1993) and later found in other bacteria (Blindauer et al., 2002) and the Cu(I)-binding MymT from mycobacteria (Gold et al., 2008). The CCR domain of BicA is a small (19-a.a.), cysteine-rich (∼ 30%) peptide that binds Cu(I), which fits the broad definition of metallothionein (Coyle et al., 2002; Capdevila and Atrian, 2011). Other than being cysteine-rich, the CCR domain does not share significant sequence homology with either BmtA (the metallothionein not the transporter) or MymT. This discovery offers further support for Robinson's prediction that bacterial metallothioneins may be more widespread than previously thought but have gone unnoticed due to limitations in annotation (Robinson, 2008).

Cu(I)-binding metallothioneins are notoriously recalcitrant to crystallization, and to date, only one group has succeeded in solving the crystal structure of a yeast Cu thionein using Cu(I)-loaded synthetic peptide (Calderone et al., 2005). This crystal structure and other solution structures (Peterson et al., 1996; Cobine et al., 2004) of Cu(I)-binding metallothioneins have shown that the protein backbone has a random coil structure and lacks defined structure such as α-helix or β-strand. This may explain why the CCR domain is not visible in a crystal structure of BicA (Codolo et al., 2010). Likely, Cu(I) ligand may be needed to stabilize the protein structure (Calderone et al., 2005). Nevertheless, the visible structure of BicA clearly shows the C-terminal tails protruding outward from the ferritin-like cage (Fig. S5). This structural model indicates a spatial separation of Fe mineralization inside the cage and Cu-binding outside of the cage, which agrees with the biochemical data showing that these two activities of BicA are largely independent of each other (Fig. 3).

Cellular metal homeostasis is intrinsically linked to oxidative stress (Faulkner and Helmann, 2011). Dps has been shown to protect many bacteria against oxidative stress due to its ability to sequester Fe (Chiancone and Ceci, 2010; Haikarainen and Papageorgiou, 2010). We had previously reported (Li et al., 2007) and showed here again that the bicA mutant was not more sensitive to oxidative stress than the WT. To the contrary, after being cultured in media supplemented with Fe or Cu, the bicA mutant was more resistant than the WT to peroxide stress. The newly discovered Cu thionein function of BicA offers an explanation for such a departure from other Dps homologues. Unlike Dps storage of Fe, during which Fe(II) is oxidized to Fe(III), Cu ions bound by metallothioneins remain in the Cu(I) state. Considering the two main mechanisms underlying Cu toxicity – attacking metal binding sites of enzymes that normally bind other metals and reacting with peroxide to generate hydroxyl radicals (Dupont et al., 2011) – the physical sequestration of Cu(I) by BicA may prevent the former but not the latter. Indeed, the data presented here showed a strong inverse correlation between spirochetal resistance to peroxide and the cellular level of Cu: the higher the Cu level is in the spirochaete, the more sensitive the spirochaete becomes to peroxide stress.

When cultivated in standard BSK-H complete medium, the bicA mutant had a 49% reduction in Fe and 96% reduction in Cu as compared with the WT B31 (Fig. 1A), which suggests that in the absence of BicA sequestration of Fe and Cu, the spirochaete may be able to either reduce import or enhance export of these metals. The model organism Escherichia coli has three systems for Cu detoxification: the periplasmic multicopper oxidase CueO, the inner membrane Cu-translocating ATPase CopA, and the periplasm-spanning Cus efflux system (Rensing and Grass, 2003). The Bb genome encodes a multicopper oxidase (bb0467), which, however, shares little homology with CueO, and its function in Bb has not been investigated. Although Bb does not have a CopA homologue, an efflux system homologous to Cus has been identified and shown to be not only important for spirochetal resistance to antibiotics but also required for infection in mice (Bunikis et al., 2008). Whether this efflux system plays any role in Fe and/or Cu detoxification remains unclear.

The presence of BicA and its function in Fe and Cu detoxification in the Lyme disease spirochaete do not necessarily imply utilization of these metals by this bacterium. In fact, this bacterium has been widely recognized as one of only two organisms that can live without Fe (Archibald, 1983; Posey and Gherardini, 2000). Even though the Bb genome encodes a SufS-like cysteine desulfurase (bb0084) and an IscU-like scaffold protein (bb0085), which are known to be involved in biogenesis of Fe-S proteins in other organisms (Lill, 2009), and homologues of Fe-S proteins such as CoxS (bb0555) and CoxM (bb0556) (Dobbek et al., 1999), the functions of these genes have not been established in this bacterium. Since there is no known Cu enzyme in the cytosol (Rensing and Grass, 2003), BicA sequestration of Cu in the cytosol is also more likely for detoxification than for storage.

How does Fe or Cu enter the spirochaete? The metal transporter BmtA is thought to be involved in transporting Mn (Ouyang et al., 2009), the transition metal that is the most concentrated in the spirochaete relative to the medium. Here, we showed that the bmtA mutant also had a reduced level of Cu, suggesting that Cu may gain entry into the cell through BmtA. The loss of BmtA resulted in an increase in both Zn and Fe levels in Bb, again suggesting that Fe could gain entry into the cell through the yet-to-be identified Zn transporter, which may be upregulated to compensate the loss of BmtA. Interestingly, with a remarkable 99% reduction in Mn, the major transition metal in the spirochaete, the bmtA mutant still grows in the BSK-H medium, albeit at a slower rate (Ouyang et al., 2009), suggesting that Mn may not be essential to the spirochaete. However, the complete loss of the bmtA mutant in its ability to infect mice or colonize ticks (Ouyang et al., 2009) suggests that the metal requirement for Bb growth in vivo is far more stringent than the metal requirement for Bb growth in vitro.

What are the niches in the spirochaete's life cycle through a tick and a mouse that could potentially pose Fe or Cu toxicity to the bacterium? A large body of literature suggests that Fe-limitation is a mechanism of innate host defence against bacterial infection (Nairz et al., 2010; Skaar, 2010), and it is thought that the Lyme disease spirochaete has evolved to be independent of Fe to evade the Fe-limiting environment in the host (Posey and Gherardini, 2000). The data presented here, however, indicate that in a bacterium that has no recognizable Fe acquisition system, keeping Fe out of the cell could also be a potential challenge. The influx of blood meal during tick engorgement could expose the spirochaetes to excesses of Fe and Cu in the blood. Construction of Bb mutants lacking either the Fe-binding or the Cu-binding activity of BicA and characterization of their phenotypes in an experimental tick–mouse infectious cycle could help further determine whether these activities are indeed responsible for the physiological role of BicA in Bb.

In this study, we also showed that spirochaetes cultivated in media supplemented with Cu became extremely sensitive to peroxide stress. Interestingly, a growing body of evidence suggests that at the host–pathogen interface, host innate immune defence also utilizes Cu to kill bacteria (Dupont et al., 2011; Festa and Thiele, 2012). It warrants further investigation whether patients with Lyme disease have an elevated level of Cu in the serum and whether dietary supplementation of Cu can improve the efficacy of antibiotic therapy for Lyme arthritis.

Experimental procedures

Bacterial strains and media

The bicA (bb0690) mutant of an infectious B31 strain and the bmtA (bb0219) mutant of an infectious 297 strain were generated in previous studies (Li et al., 2007; Ouyang et al., 2009). Unless otherwise noted, the spirochaetes were cultured at 33°C in standard BSK-H complete media (Sigma-Aldrich), which contained 6% rabbit serum (Pollack et al., 1993). The bicA and the bmtA mutants are resistant to kanamycin due to the deletion and insertion mutations at the specific loci, but the mutations are stable without the antibiotic selection (Li et al., 2007; Ouyang et al., 2009). To avoid the Cu chelating effect or any other unknown side-effect of kanamycin (Szczepanik et al., 2004), the mutants were cultivated in the same medium as the WT strains, without any antibiotic selection.

Inductively coupled plasma-sector field mass spectrometry

Spirochaetes were grown to early stationary phase, collected by centrifugation, washed three times with sterile Dulbecco's phosphate-buffered saline (PBS, Invitrogen) and once with the UltraPure water (Invitrogen), and then digested with acid. Bacterial pellets (2–3 × 109 spirochaetes) and protein samples (10–15 μg in a 100 μl volume) were digested with 100 μl of concentrated nitric acid (16 N, double distilled; GFS Chemicals) at 100°C for 15 min. The final volume of each sample was adjusted to 1 ml with the addition of Milli-Q water. As an internal standard, indium (In) was added to each sample to a final concentration of 10 ppb. Metal analyses were performed using a ThermoFinnigan Element 2 ICP-SFMS instrument at the Trace Element Research Laboratory at the Ohio State University. A certified standard (ICP-MS-68B-A from High-Purity Standards) containing all the elements to be analysed (Mn, Fe, Co, Ni, Cu, Zn and In) was diluted in 10% nitric acid to make 1, 10, 50, and 200 ppb solutions, which were used to generate linear calibration curves. Samples were introduced into the instrument using a perfluoroalkoxy (PFA) sample introduction system (Elemental Scientific) at an uptake rate of 100 μl min−1, and all elements were measured at medium resolving power (R = m/Δm = ∼ 4000) as recommended for detection of the transition metals included in our analysis (Jakubowski et al., 2011a,b).

Expression and purification of WT and mutant BicA

The WT BicA was expressed in and purified from E. coli TOP10 (Invitrogen) using the GST fusion system (expression vector pGEX-6P-2; GE Healthcare Life Sciences) as described previously (Li et al., 2007). The expression vectors for the mutants were derived from the WT plasmid by PCR mutagenesis, and nucleotide sequencing was performed to verify that only the intended mutations were introduced during PCR and cloning. Upon IPTG induction, the GST-BicA fusion protein was found in both the insoluble and the soluble fractions of the French Pressed lysate. All protein preparations used in this study were purified from the soluble fraction on Glutathione 4B Sepharose resin (GE Healthcare Life Sciences). On-column digestion using the PreScission Protease (GE Healthcare Life Sciences) released BicA from the GST tag, which was further purified on a Superdex 200 10/300 GL column (GE Healthcare Life Sciences) to remove minor contaminating proteins and to change the buffer to 50 mM MOPS containing 1 mM DTT, pH 7.0 (the MOPS/DTT buffer). A representative chromatogram and a representative sodium dodecyl sulphate-polyacryalmide gel are shown in Fig. S6.

Transmission electron microscopy

A drop (10 μl) of purified recombinant BicA (500 μg ml−1) was placed on a 200-mesh Cu TEM grid with carbon support film and negatively stained with 1% uranyl acetate. The specimens were characterized using a FEI Tecnai G2 F20 cryo-S-TEM equipped with a Gatan Ultrascan 1000 CCD camera at an accelerating voltage of 200 kV.

Cu(I)-thiolate fluorescence assay

Purified recombinant BicA (12 μM in the MOPS/DTT buffer) was incubated at room temperature with 24, 48, 72, 96, 120, 240, 360, or 480 μM of CuSO4 in 100 μl reactions set up in a 96-well plate (Microfluor 1 White, Thermo Scientific). Of note, 1 mM DTT was included in all reactions to keep Cu in the cuprous form (Gold et al., 2008). When DTT was omitted, fluorescent signal was not detected in any of the reactions. For each Cu concentration tested, a mock reaction containing the specific concentration of Cu and 1 mM DTT but no protein was used for background subtraction. Fluorescent signals were detected using a SpectroMax M2 instrument (Molecular Devices) with an excitation wavelength of 280 nm and an emission scan from 500 nm to 750 nm at 5, 30 and 60 min after the addition of Cu. The result shown in Fig. 3C was obtained at 60 min after the addition of Cu, and was representative of three independent experiments using three different preparations of purified WT BicA. The Cu(I) luminescent assay of the WT and mutant proteins shown in Fig. 3E was carried out at a ratio of 20 atoms of Cu per monomer, and the signals were measured at 5, 15, 25, 35, 45, 55 and 65 min after the addition of Cu.

Prussian blue staining of iron bound to BicA

Purified WT and mutant BicA were incubated at room temperature for 30 min in 20 μl reactions containing 24 μM protein, 1 mM ammonium ferrous sulphate, and 50 mM MOPS, pH 7.0. Proteins were then subjected to native polyacrylamide gel electrophoresis on 4–20% gradient gels (Bio-Rad). The gels were stained either with Coomassie brilliant blue for detection of protein or with Prussian blue for detection of iron (Leong et al., 1992).

GST-tag-based affinity capture of protein-bound Cu

The GST fusions of WT and mutant BicA were each purified on 50 μl Glutathione 4B Sepharose resin (GE Healthcare Life Sciences) in Pierce spin columns (Thermo Scientific). The protein-bound resins were then incubated at room temperature with 500 μl of 400 μM CuSO4 (in the MOPS/DTT buffer) followed by 3 to 5 washes each with 500 μl of the MOPS/DTT buffer. After the washes, proteins were eluted with 150 μl of 10 mM glutathione (in the MOPS/DTT buffer). The protein concentration in each sample was determined by the Bradford method (Bio-Rad), and the metal concentration in each sample was quantified by ICP-SFMS. Direct binding of Cu by the Glutathione 4B Sepharose resin (w/o protein) under the experimental conditions described above was negligible but was nevertheless subtracted as background. We also used this method to determine whether BicA binds Fe and Mn by replacing CuSO4 with (NH4)2Fe(SO4)2 and MnSO4 respectively, at the incubation step (Fig. S4). The data showed that the WT and the ΔCCR mutant bound significantly more Fe than the D65N&E69Q and the D65N&E69Q&ΔCCR mutants. The low-levels of Mn bound to the proteins are likely to be non-specific, as similar levels were detected in all constructs.

Fe and Cu toxicity assay

The WT and the bicA mutant of Bb strain B31 (Li et al., 2007) were subcultured into 96-well plate in standard BSK-H complete media (Sigma-Aldrich) (Pollack et al., 1993) either not supplemented (controls) or supplemented with the indicated concentration of (NH4)2Fe(SO4)2 or CuSO4. The plate was incubated at 33°C in a 5% CO2 incubator until the controls reached early stationary phase. In the experiment shown in Fig. 4, the number of viable spirochaetes in each culture was determined by the limiting dilution method using three independent series of 10-fold dilutions. The bacterial growth in media supplemented with Fe or Cu was normalized against the bacterial growth in media without supplementation. In a pilot experiment done earlier, we compared the limiting dilution method with the method of using Petroff-Hausser counting chamber to determine the total number of spirochaetes (Fig. S7). Both methods showed the same trend that the bicA mutant was more susceptible than the WT to Fe and Cu toxicity. We later chose the limiting dilution method because of its ability to distinguish viable cells from dead cells and because it is capable of detecting as low as 50 spirochaetes per ml, much lower than the 5 × 104 spirochaetes per ml detection limit for the counting method.

Peroxide killing assay

The WT and the bicA mutant of Bb strain B31 (Li et al., 2007) were cultured at 33°C to early stationary phase in BSK-H complete media (Sigma-Aldrich) (Pollack et al., 1993) either not supplemented or supplemented with 20 μM of (NH4)2Fe(SO4)2 or 100 μM of CuSO4. Two replicate cultures were set up for each condition tested. The spirochaetes were collected by centrifugation, washed three times with sterile PBS (Invitrogen), diluted to 108 spirochaetes per ml in PBS or in PBS containing 0.25, 0.5, 1, 2, or 4 mM of H2O2, and then incubated at the ambient temperature for 1 h. The number of viable spirochaetes in each sample was determined by the limiting dilution method using three independent series of 10-fold dilutions. The percentages of survival were calculated by normalizing the values of peroxide-treated samples with the value of the control sample, which was treated with PBS only.

Statistical analysis

The software GraphPad Prism 5 (version 5.01) was used to for graphic and statistical analyses. The statistical test for each P-value is indicated in the text. All P-values reported are two-tailed. P-values < 0.05 are considered statistically significant.

Acknowledgements

We thank M.V. Norgard for providing strains, D. Acosta for technical assistance, K. Hayes-Ozello for editing the manuscript, and A. Bird, K. Boris-Lawrie, P. Boyaka, C. L. Brooks, M. Chan, E. Fikrig, L. Hu, H.L.T. Mobley, M. Oglesbee, U. Pal, Y. Rikihisa, and X.F. Yang for comments and suggestions. X.L. is the recipient of an Arthritis Investigator Award from the Arthritis Foundation. This study is supported in part by start-up funds to X.L. from the Public Health Preparedness for Infectious Diseases Program and the College of Veterinary Medicine of the Ohio State University.

Ancillary