Zinc Toxicity and Iron-Sulfur Cluster Biogenesis in Escherichia coli

Zinc toxicity has been implicated in causing various human diseases. High concentrations of zinc can also inhibit bacterial cell growth. However, the underlying mechanism has not been fully understood. Here, we report that zinc overload in Escherichia coli cells inhibits iron-sulfur cluster biogenesis by targeting specific iron-sulfur cluster assembly proteins. Because iron-sulfur proteins are involved in diverse physiological processes, the zinc-mediated inhibition of iron-sulfur cluster biogenesis could be largely responsible for the zinc-mediated cytotoxicity. Our finding provides new insights on how intracellular zinc overload may inhibit cellular functions in bacteria.

intracellular total zinc accumulates to about 0.2 mM when cells are grown in LB medium (9). Radioactive 65 Zn-labeling studies (10) and proteomic analyses (11) have revealed a large number of putative zinc-binding proteins in E. coli cells. On the other hand, the addition of 2.5 mM ZnSO 4 to LB medium (12) or 0.35 mM ZnSO 4 to M9 minimum medium (13) completely inhibits E. coli cell growth. Because of the zincmediated inhibition of cell growth, zinc compounds have been developed as antibacterial agents and preservatives. Furthermore, host-mediated zinc toxicity to pathogenic bacteria has been extensively investigated (14)(15)(16). However, the molecular mechanism underlying the zinc-mediated cytotoxicity has not been fully understood.
Our previous studies have shown that topoisomerase I (17,18) and its homolog YrdD (19) are iron and zinc binding proteins, and excess zinc can easily compete for iron binding in the proteins in vivo (17,19). This suggests that zinc and iron may have similar binding sites in proteins.
In the past decade, several "zinc finger" proteins have been identified as iron-sulfur proteins. For example, the mitochondrial outer membrane protein mitoNEET (20) and the cleavage and polyadenylation specificity factor 30 (CPSF30) (21) have a zinc finger motif which hosts an iron-sulfur cluster. Since zinc and iron-sulfur cluster have similar ligand coordination in proteins, it has been proposed that zinc may compete for iron-sulfur center binding sites in proteins and disrupt iron-sulfur clusters in proteins (22)(23)(24). Since iron-sulfur proteins are involved in diverse physiological functions (25), excess zinc may affect multiple cellular functions by disrupting iron-sulfur clusters in proteins.
Iron-sulfur clusters are assembled by a group of dedicated proteins. In E. coli, there are two iron-sulfur cluster assembly systems encoded by the housekeeping iscSUA-hscBA-fdx-iscX gene cluster (26) and the inducible sufABCDSE gene cluster (22). Among the proteins encoded by iscSUA-hscBA-fdx-iscX, IscS is a cysteine desulfurase that provides sulfur for iron-sulfur cluster assembly (27). IscU is a scaffold protein that assembles iron-sulfur clusters (28) and transfers the transient clusters to target proteins (29,30). IscA was thought to be an alternative scaffold (31). However, unlike the scaffold IscU, IscA has strong iron binding activity, and the iron center in IscA can be transferred to IscU for iron-sulfur cluster assembly (32)(33)(34). Thus, IscA is proposed as an iron chaperone for iron-sulfur cluster biogenesis. HscB and HscA are heat shock cognate proteins, which assist the iron-sulfur cluster transfer from IscU to target protein (35). Ferredoxin (Fdx) is a [2Fe-2S] cluster protein and may provide electrons for the iron-sulfur cluster assembly process (36). IscX has also been proposed as an iron donor for iron-sulfur cluster biogenesis (37). However, IscX has low iron binding affinity and interacts with IscS (38). The deletion of IscX does not significantly affect iron-sulfur proteins in E. coli cells (39). Therefore, the specific function of IscX remains to be defined.
In this study, we find that zinc overload in E. coli cells inhibits iron-sulfur cluster biogenesis without affecting the preassembled clusters in proteins. Additional studies reveal that zinc has strong interaction with the iron-sulfur cluster assembly proteins IscU, IscA, and ferredoxin, leading to inhibition of iron-sulfur cluster biogenesis in E. coli cells.

RESULTS
Zinc overload selectively inactivates iron-sulfur enzymes in E. coli cells. In wild-type E. coli cells, the "free" intracellular zinc concentration is in the femtomolar range (9). Zinc homeostasis in E. coli cells is regulated primarily through a network of zinc influx and efflux pumps. The major zinc efflux system ZntA, a P-type ATPase transporter, is upregulated by the transcription factor ZntR when intracellular zinc concentration is high (40). The deletion of ZntA results in an E. coli strain that is hypersensitive to zinc (41). To explore the effect of intracellular zinc overload on iron-sulfur proteins in E. coli cells, we have constructed an E. coli mutant in which both the zinc efflux pump ZntA and the transcription factor ZntR were deleted. Table 1 shows that the deletion of ZntA and ZntR resulted in accumulation of intracellular zinc in E. coli cells grown in LB medium supplemented with 200 M ZnSO 4 under aerobic growth conditions. ZnSO 4 at 200 M was chosen, as it inhibited cell growth of the E. coli zntA zntR double mutant in LB medium by about 50% and did not significantly affect the cell growth of wild-type E. coli (see Fig. S1 in the supplemental material).
To investigate the effect of zinc overload on iron-sulfur proteins in E. coli, we first utilized fumarases. There are three fumarases in E. coli, fumarase A and fumarase B, which require a [4Fe-4S] cluster for their catalytic activity (42); and fumarase C, which has no iron-sulfur clusters (43). Each fumarase was expressed in the E. coli zntA zntR mutant cells grown in LB medium supplemented with or without 200 M ZnSO 4 under aerobic conditions. Figure 1 shows that the addition of ZnSO 4 (200 M) to LB medium  largely eliminated the iron-sulfur cluster content ( Fig. 1A and B) and the enzyme activity ( Fig. 1D) of fumarases A and B in the E. coli zntA zntR mutant cells. On the other hand, the same zinc treatment did not affect the enzyme activity of fumarase C in E. coli zntA zntR mutant cells ( Fig. 1C and D). The results suggested that zinc overload in E. coli cells selectively inhibits iron-sulfur cluster-containing fumarases A and B without inhibiting fumarase C, which does not have iron-sulfur clusters.

Effect of zinc overload on other iron-sulfur proteins in E. coli cells.
To further explore the effects of zinc overload on iron-sulfur proteins, we used biotin synthase, which contains a [2Fe-2S] cluster and a [4Fe-4S] cluster (44). Biotin synthase (BioB) converts dethiobiotin into biotin by inserting a sulfur atom between C-6 and C-9 of dethiobiotin in an S-adenosylmethionine (SAM)-dependent reaction (45). In the experiments, recombinant BioB was expressed in the E. coli zntA zntR mutant cells grown in LB medium supplemented with or without 200 M ZnSO 4 under aerobic conditions. Figure 2A shows that the addition of ZnSO 4 (200 M) also decreased the iron-sulfur cluster contents of recombinant biotin synthase in the E. coli zntA zntR mutant cells.
We then analyzed the activity of the endogenous NADH dehydrogenase I, which contains multiple iron-sulfur clusters (46). Using deamino-NADH as a specific substrate for NADH dehydrogenase I (47), we found that the deamino-NADH oxidation rate of the E. coli zntA zntR mutant cells progressively decreased when the ZnSO 4 concentration in LB medium was gradually increased from 0 to 400 M (Fig. 2B). The cell growth of the E. coli zntA zntR mutant was also inhibited as the ZnSO 4 concentration in LB medium was increased (Fig. 2B). Since the NADH dehydrogenase I contains multiple iron-sulfur clusters (48), the correlation between the decrease in the deamino-NADH oxidation rate and inhibition of cell growth by zinc in LB medium suggested that zinc toxicity could be closely associated with the inhibition of iron-sulfur proteins in E. coli cells.
Zinc overload inhibits iron-sulfur cluster biogenesis in E. coli cells. It was proposed that zinc may directly attack iron-sulfur clusters in proteins to produce the apo form in cells (22). On the other hand, zinc may block iron-sulfur cluster biogenesis, thus producing apo-form inactive proteins in cells. To delineate the two possibilities, we added ZnSO 4 (200 M) to the E. coli zntA zntR mutant cells in LB medium before and after recombinant iron-sulfur protein was expressed. The relative dehydrogenase I activity and relative cell growth were defined as the percentage of the E. coli zntA zntR double-mutant cells in LB medium with ZnSO 4 over that without ZnSO 4 . The relative cell growth inhibition rate was calculated by 100% minus the relative growth rate of the E. coli zntA zntR double-mutant cells. The relative dehydrogenase I activity (closed circles) and relative cell growth inhibition rate (closed squares) were plotted as a function of the ZnSO 4 concentration in LB medium. The 100% cell growth represented the cell density (OD at 600 nm) of ϳ3.0 after 5 h at 37°C in LB medium with aeration. The results were the most representative of three independent experiments.
In the experiment, recombinant dihydroxy-acid dehydratase (IlvD), which contains a [4Fe-4S] cluster for its enzyme activity in the branched-chain amino acid biosynthesis pathway (49), was expressed in the E. coli zntA zntR mutant cells grown in LB medium with or without ZnSO 4 (200 M). IlvD was then purified from the cells. Figure 3A shows that the addition of ZnSO 4 (200 M) to LB medium before recombinant IlvD was expressed largely eliminated iron-sulfur clusters in the protein. However, the addition of zinc to LB medium after IlvD was expressed did not significantly affect iron-sulfur clusters in the protein. The enzyme activity measurements further showed that zinc blocked iron-sulfur cluster assembly without affecting the preassembled iron-sulfur clusters in IlvD in the E. coli cells (Fig. 3B).
The recombinant endonuclease III (Nth), a DNA repair enzyme which hosts a stable [4Fe-4S] cluster (50), was also investigated. Figure 3C shows that while the addition of NADH dehydrogenase I activity was measured by monitoring the oxidation of deamino-NADH at 340 nm (extinction coefficient, 6.22 mM Ϫ1 cm Ϫ1 ) at room temperature. The unit of NADH dehydrogenase I enzyme activity referred to the reduction of substrate (micromolar) per minute per OD at 600 nm. The relative activity of control sample was considered to be 100%, and the relative activity of other samples was obtained by dividing the control activity. The results represent average Ϯ standard deviation from three independent experiments. ZnSO 4 (200 M) to LB medium before endonuclease III was expressed in the E. coli zntA zntR mutant cells largely prevented iron-sulfur cluster assembly in the protein, the addition of ZnSO 4 (200 M) to LB medium after endonuclease III was expressed did not significantly affect the preassembled iron-sulfur cluster in endonuclease III.
To further explore the inhibition of zinc overload on endogenous iron-sulfur cluster biogenesis, we measured the activity of the native NADH dehydrogenase I in the E. coli cells. Figure 3D shows that while the addition of ZnSO 4 (200 M) to LB medium followed by 5 h of growth of the E. coli zntA zntR mutant cells significantly decreased the enzyme activity, the addition of ZnSO 4 (200 M) to LB medium after 5 h of cell growth did not affect the enzyme activity of the NADH dehydrogenase I in the cells.
In E. coli, in addition to the housekeeping iscSUA-hscBA-fdx-iscX iron-sulfur gene cluster assembly system, there is another stress-inducible sufABCDSE system. Since increased expression of the gene cluster sufABCDSE is an indication of the iron-sulfur cluster assembly deficiency in E. coli cells (51,52), we also analyzed the expression of the sufA operon in the E. coli zntA zntR mutant cells in response to ZnSO 4 in LB medium and found that expression of the sufA operon was indeed induced by ZnSO 4 treatment (Fig. S2B), suggesting that zinc has a general inhibitory effect on iron-sulfur cluster biogenesis in E. coli cells.
Taken together, the results suggested that ZnSO 4 (200 M) inhibits iron-sulfur cluster biogenesis without affecting the preassembled iron-sulfur clusters in proteins in E. coli cells.
IscU, IscA, and ferredoxin are the major zinc targets among the housekeeping iron-sulfur cluster assembly machinery. The proteins encoded by iscSUA-hscBA-fdx-iscX represent the housekeeping iron-sulfur cluster biogenesis machinery in E. coli cells. If zinc inhibits iron-sulfur cluster biogenesis, it is possible that zinc may directly interact with iron-sulfur cluster assembly proteins. To test this idea, we expressed each protein encoded by the gene cluster iscSUA-hscBA-fdx-iscX in the E. coli zntA zntR mutant cells grown in LB medium supplemented or not with ZnSO 4 (200 M). Purified proteins were then subjected to the UV-visible absorption measurements and metal content analyses. Figure 4 shows that the addition of ZnSO 4 (200 M) to LB medium had no effect on the UV-visible absorption spectra of IscS (Fig. 4A), IscU (Fig. 4B), HscB (Fig. 4D), HscA (Fig. 4E), and IscX (Fig. 4G) expressed in the E. coli zntA zntR mutant cells. On the other hand, the addition of ZnSO 4 (200 M) to LB medium significantly decreased the iron binding peak at 315 nm of IscA (Fig. 4C) and the iron-sulfur cluster binding peaks at 415 nm and 459 nm of ferredoxin (Fig. 4F) expressed in the E. coli zntA zntR mutant cells, suggesting that zinc overload may block the iron binding in IscA and the iron-sulfur cluster binding in ferredoxin. The zinc content measurements in purified proteins (Fig. 4H) showed that IscU, IscA, and ferredoxin proteins contained 0.85 Ϯ 0.16, 0.94 Ϯ 0.04, and 1.69 Ϯ 0.12 zinc atoms per protein monomer (n ϭ 3), respectively. The stoichiometry of zinc binding in IscU is consistent with previous studies showing that each IscU monomer contains one zinc atom (24). On the other hand, other iron-sulfur cluster assembly proteins had only very little or no zinc binding (Fig. 4H). Thus, IscA, IscU, and ferredoxin are the major targets of zinc overload in the E. coli zntA zntR mutant cells.
The conserved cysteine residues in IscA, IscU, and ferredoxin are required for their zinc binding activity. To explore the zinc binding sites of IscU, IscA, and ferredoxin, we constructed an IscU mutant (IscU-3M) in which three cysteine residues (Cys-37, Cys-63, and Cys-106) were replaced with serine, an IscA mutant (IscA-3M) in which three cysteine residues (Cys-35, Cys-99, and Cys-101) were replaced with serine (48), and a ferredoxin mutant (Fdx-4M) in which four cysteine residues for binding the [2Fe-2S] cluster (Cys 42, Cys 48, Cys 51, and Cys 87) were replaced with serine. Wild-type IscU, IscA, and ferredoxin and their mutants (IscU-3M, IscA-3M, and ferredoxin-4M, respectively) were then expressed in the E. coli zntA zntR mutant cells grown in LB medium supplemented with increasing concentrations of ZnSO 4 (0 to 400 M). Each protein was then purified from the E. coli cells. Figure 5 shows that zinc binding in IscU, IscA, and ferredoxin was gradually increased in the E. coli zntA zntR mutant cells as the concentration of ZnSO 4 in LB medium was increased. In contrast, the mutant proteins (IscU-3M, IscA-3M, and Fdx-4M) expressed in the E. coli zntA zntR mutant cells had very little or no zinc binding even after 400 M ZnSO 4 was added to LB medium. Thus, IscU, IscA, and ferredoxin have specific zinc binding activity, and the conserved cysteine residues in IscU, IscA, and ferredoxin are essential for their zinc binding activity.
Zinc overload emulates the phenotype of an E. coli mutant with the deletion of IscU, IscA, and ferredoxin. If IscU, IscA, and ferredoxin are the major targets of zinc overload in the E. coli cells, the deletion of these genes would emulate the effects of zinc overload on iron-sulfur cluster biogenesis. To test this idea, we deleted the genes encoding IscU, IscA, and ferredoxin to produce an E. coli iscU iscA fdx mutant. Figure 6A shows that the deletion of IscU, IscA, and ferredoxin decreased the iron-sulfur cluster assembly in endonuclease III, which was similar to the inhibition of the iron-sulfur cluster assembly in endonuclease III in the E. coli zntA zntR mutant cells grown in LB medium supplemented with 200 M ZnSO 4 (Fig. 6B). We also measured the cell growth of the E. coli mutant with the deletion of IscA, IscU, and ferredoxin and found that those deletions resulted in slow growth (Fig. 6C), which was also similar to that of the E. coli zntA zntR mutant cells grown in LB medium supplemented with 200 M ZnSO 4 (Fig.  6D). Thus, zinc overload in E. coli cells appears to emulate the phenotype of the E. coli mutant cells with deletion of the iron-sulfur cluster assembly proteins IscU, IscA, and ferredoxin.

DISCUSSION
In this study, we report that zinc overload in the E. coli zntA zntR mutant cells inhibits iron-sulfur cluster biogenesis without affecting the preassembled iron-sulfur clusters in proteins or proteins without iron-sulfur clusters. Additional studies show that among the housekeeping iron-sulfur cluster assembly proteins encoded by gene cluster iscSUA-hscBA-fdx-iscX in E. coli, IscU, IscA, and ferredoxin have strong zinc binding activity in E. coli cells, and the conserved cysteine residues in the proteins are essential for their zinc binding activity. The deletion of IscU, IscA, and ferredoxin in E. coli cells appears to emulate the zinc overload-mediated inhibition of iron-sulfur cluster biogen- esis and the slow-growth phenotype of the E. coli zntA zntR mutant cells. The results suggest that zinc overload inhibits iron-sulfur cluster biogenesis by specifically targeting IscU, IscA, and ferredoxin in cells.
Iron-sulfur clusters in proteins are generally vulnerable to reactive oxygen species (51) and nitric oxide (53). Recently, it has also been shown that iron-sulfur clusters in dehydratases are sensitive to copper (54), zinc, and silver (22). Here, we find that iron-sulfur proteins, including fumarases A and B, endonuclease III, biotin synthase, and endogenous NADH dehydrogenase I, are readily inactivated by zinc overload in E. coli cells. On the other hand, the preassembled iron-sulfur clusters in proteins are not disrupted by zinc overload in the cells. Thus, zinc overload appears to target iron-sulfur cluster biogenesis instead of disrupting iron-sulfur clusters in proteins in cells. This notion of zinc toxicity is analogous to that of copper toxicity, which also inhibits iron-sulfur cluster biogenesis without affecting preassembled iron-sulfur clusters in cells (48,55).
One finding of this study is that zinc can effectively inhibit the enzyme activity of NADH dehydrogenase I in the E. coli zntA zntR double-mutant cells. This is partially in agreement with the previous observation that zinc inhibits the respiratory chain by both respiratory oxidases (56,57) and NADH dehydrogenase (57) in E. coli. However, we also find that the activity of preexisted NADH dehydrogenase I is not affected significantly by zinc in E. coli zntA zntR double-mutant cells. The apparent contradiction with the previous report that zinc inhibits NADH dehydratase directly in vitro (57) is likely due to the different experimental conditions. A possible consideration is that the activity of NADH dehydrogenase is sensitive to zinc by determination in vitro. In our experiment, prior to the determination of NADH dehydratase I activity, most of the residual zinc after treatment was removed by washing cells twice with buffer. When we measured the activity of NADH dehydratase I, there was little or no zinc in the reaction solution. The activity of the NADH dehydratase I should be unaffected as long as the complex and the bound iron-sulfur clusters are completely assembled and not destroyed by zinc treated in living cells. Altogether, zinc may not only inhibit the activity of NADH oxidase and quinol oxidases directly but also decrease NADH dehydratase I activity by blocking the iron-sulfur cluster assembly in the enzyme complex in E. coli cells.
The effects of high zinc content on wild-type E. coli cells have been extensively investigated by transcriptomics (13), proteomics (58) and metalloproteomics (59) approaches. In response to the elevated intracellular zinc content, the cells will express multiple proteins, including ZraP, a putative zinc storage protein (59), a major zinc efflux system, ZntA, a P-type ATPase transporter, and a transcription factor, ZntR (40), among others. When ZntA and ZntR are deleted, the E. coli mutant cells accumulate intracellular zinc content, resulting in zinc overload. Because zinc and the iron-sulfur cluster have similar ligand coordination in proteins, it is conceivable that zinc overloading in cells may compete with iron or iron-sulfur cluster binding in the proteins and inhibit iron-sulfur cluster biogenesis in cells. With the exception of IscS, the major iron-sulfur cluster assembly proteins have either iron or iron-sulfur cluster binding sites. Here, we found that zinc overload in the E. coli zntA zntR mutant cells inhibits iron-sulfur cluster biogenesis by specifically binding to IscU, IscA, and ferredoxin. The zinc binding in the iron-sulfur cluster assembly protein IscU has previously been reported (60), which is consistent with our results. It may be envisioned that zinc overload forces IscU to bind zinc, which would prevent IscU from assembling iron-sulfur clusters. Zinc binding in IscA has not been previously reported. In the crystal structure, Cys-99 and Cys-101 of IscA are not visible, likely because of their flexible structure (61,62). Nevertheless, it has been postulated that IscA may form a cysteine pocket with Cys-99 and Cys-101, which are responsible for binding iron (63) and facilitate the binding of other transition metal ions, such as copper or zinc. Our previous study showed that excess copper in E. coli cells does lead to copper binding in IscA under aerobic (48) and anaerobic (55) growth conditions and inhibits iron-sulfur cluster assembly. Here, we found that zinc overload in E. coli cells results in zinc binding in IscA and blocks iron-sulfur cluster biogenesis. Similarly, ferredoxin hosts a [2Fe-2S] cluster via four cysteine residues (Cys-42, Cys-48, Cys-51, and Cys-87) (64), and zinc overload in E. coli cells leads to zinc binding in ferredoxin. Since mutations of the conserved cysteine residues in IscU, IscA, and ferredoxin almost abolish the zinc binding activity of the proteins, these residues are critical for both iron/iron-sulfur clusters and zinc binding in the proteins. We propose that zinc overload in cells results in zinc binding in IscU, IscA, and ferredoxin and inhibits iron-sulfur cluster biogenesis in E. coli cells.
Since iron-sulfur proteins are involved in diverse physiological processes ranging from energy metabolism to DNA repair and replication (65), the inhibition of iron-sulfur cluster biogenesis by zinc will have a broad impact on diverse cellular functions. It should be pointed out that iron-sulfur proteins are also the targets of cobalt (66) and copper (48,54,55,67,68) toxicity. Thus, iron-sulfur cluster biogenesis could be the primary target of heavy-metal toxicity in cells.

MATERIALS AND METHODS
Gene knockout in E. coli cells. ZntA and ZntR, two major proteins regulating intracellular zinc homeostasis, were deleted from wild-type E. coli (MC4100) following procedures described previously (69). The constructed E. coli zntA zntR mutant cells grow normally in LB medium but become hypersensitive to zinc in the medium as reported by Binet and Poole (41). Genes encoding IscU, IscA, and ferredoxin were also deleted from wild-type E. coli cells (MC4100). The gene deletion was confirmed by PCR. All primers for the gene deletion and confirmation were synthesized by TaKaRa Co. (Dalian, China).
Protein expression and purification. Genes encoding fumarases A, B, and C from E. coli were amplified using PCR and cloned to an expression plasmid, pBAD, as described previously (55). The plasmids expressing E. coli IscS, IscU, IscA, HscB, HscA, ferredoxin, IscX, dihydroxy-acid dehydratase (IlvD), endonuclease III (Nth), and biotin synthetase (BioB) were previously prepared. Each plasmid was introduced into the E. coli zntA zntR mutant cells. Cells containing the expression plasmid were grown to an optical density at 600 nm (OD 600 ) of 0.6. ZnSO 4 was then added to Luria-Bertani (LB) medium 10 min before recombinant protein was induced with 0.02% arabinose at 37°C for 4 h with aeration. Cells were harvested and washed twice with protein purification buffer (NaCl [500 mM], Tris [20 mM, pH 8.0]). Proteins were purified as described previously (55). The purity of purified protein was judged from SDS-PAGE, followed by the Coomassie blue staining. The concentration of purified protein was determined from the absorption peak at 280 nm using the published extinction coefficients.
Site-directed mutagenesis. The plasmid pBAD-IscA-3M was previously constructed. The IscU-3M (C37/63/106S) and ferredoxin-4M (C42/48/51/87S) mutants were constructed by the site-directed mutagenesis. Each of the conserved cysteine residues in the proteins was replaced with serine. PCR primers used in this work are shown in Table 2, and specific mutations were confirmed by direct sequencing.
Enzyme activity assays for fumarases, dihydroxy-acid dehydratase, NADH dehydrogenase I, and cysteine desulfurase. The activities of purified fumarases A, B, and C were measured by monitoring the reaction product (fumaric acid) in a reaction mixture containing 50 mM sodium phosphate (pH 7.4) and 50 mM substrate (malate) at 250 nm using an extinction coefficient of 1.48 cm Ϫ1 mM Ϫ1 (55). For dihydroxy-acid dehydratase, activity was measured using DL-2, 3-dihydroxy-isovalerate as the substrate. In the assay, 10 l purified IlvD was added to 390 l preincubated solution containing 50 mM Tris (pH 8.0) and 10 mM substrate. The reaction product (keto acids) was monitored at 240 nm using an extinction coefficient of 0.19 cm Ϫ1 mM Ϫ1 . NADH dehydrogenase I activity of E. coli cells was measured following procedures described previously, with some modifications (48). Briefly, the inverted membrane vesicles of E. coli cells were prepared by passing the cells through low-temperature ultrahigh-pressure continuous flow cell disrupter (JN-3000 Plus) once. Inverted membrane vesicles were added to the reaction solution containing 20 mM Tris (pH 8.0), NaCl (200 mM), deamino-NADH (100 M), sodium azide (400 mM), and plumbagin (400 M). The NADH dehydrogenase I activity was determined by measuring the oxidation of deamino-NADH at 340 nm ( ϭ 6.22 mM Ϫ1 cm Ϫ1 ) at room temperature. In this situation, deamino-NADH as a specific substrate for NADH dehydrogenase I provides electrons (47), sodium azide inhibits the terminal cytochrome oxidases (52), and plumbagin abstracts the electrons directly from the NADH dehydrogenase I (59). The cysteine desulfurase activity of E. coli IscS was measured by incubating IscS with dithiothreitol (2 mM) and L-cysteine (0.1 mM) at 37°C. The amount of sulfide produced by IscS in the solution was measured according to Siegel's method (70).
Metal content analyses. Total zinc content in protein samples was determined using the zinc indicator PAR [4-(2-pyridylazo)-resorcinol]. The iron content of protein samples was measured according to Fischer's method (71). Zinc and iron contents in protein samples were also analyzed by the inductively coupled plasma-emission spectrometry (ICP-MS). The results from the two methods were very similar to each other.
For total zinc content of E. coli, cells was also determined by ICP-MS. Particularly, the E. coli zntA zntR mutant and its parental wild-type strain MC4100 cells were grown in LB medium supplemented or not with 200 M ZnSO 4 at 37°C under aerobic conditions. Cells were harvested from 50 ml LB medium by centrifugation when the OD at 600 nm of the cells reached 0.6. The cell pellet was washed twice with 50 ml of 170 mM NaCl, 20 mM Tris-HCl (pH 8.0), and 2 mM EDTA, resuspended in 5 ml of 170 mM NaCl, and transferred to a microwave digestion vessel with 4 ml nitric acid (35% [vol/vol], Shanghai Yiqian Technology Co. Ltd.) added. The vessel was sealed and placed in the microwave chamber. The following steps were run: the temperature was ramped to 120°C, held for 1 min, ramped to 160°C, held for 6 min, ramped to 180°C, held for 20 min, and reduced to room temperature over 30 min. Once digestion was complete, the samples were diluted using deionized water. The zinc content of above-mentioned samples was determined by ICP-MS.

SUPPLEMENTAL MATERIAL
Supplemental material for this article may be found at https://doi.org/10.1128/AEM .01967-18.

ACKNOWLEDGMENTS
This work was supported by the National Natural Science Foundation of China (grants 81671124, 31629003, 31870775, and 81500440), Key Discipline of Zhejiang Province in