ABSTRACT
Manganese is a micronutrient required for activities of several important enzymes under conditions of oxidative stress and iron starvation. In Escherichia coli, the manganese homeostasis network primarily constitutes a manganese importer (MntH) and an exporter (MntP), which are regulated by the MntR dual regulator. In this study, we find that deletion of E. coli hflX, which encodes a ribosome-associated GTPase with unknown function, renders extreme manganese sensitivity characterized by arrested cell growth, filamentation, lower rate of replication, and DNA damage. We demonstrate that perturbation by manganese induces unprecedented influx of manganese in ΔhflX cells compared to that in the wild-type E. coli strain. Interestingly, our study indicates that the imbalance in manganese homeostasis in the ΔhflX strain is independent of the MntR regulon. Moreover, the influx of manganese leads to a simultaneous influx of zinc and inhibition of iron import in ΔhflX cells. In order to review a possible link of HflX with the λ phage life cycle, we performed a lysis-lysogeny assay to show that the Mn-perturbed ΔhflX strain reduces the frequency of lysogenization of the phage. This observation raises the possibility that the induced zinc influx in the manganese-perturbed ΔhflX strain stimulates the activity of the zinc-metalloprotease HflB, the key determinant of the lysis-lysogeny switch. Finally, we propose that manganese-mediated autophosphorylation of HflX plays a central role in manganese, zinc, and iron homeostasis in E. coli cells.
INTRODUCTION
Only a handful of enzymes require manganese as the sole cofactor for their activities (1). In the living world, Mn acts either as a Lewis acid, a property similar to magnesium (Mg), zinc (Zn), and calcium (Ca), or as an oxidation cofactor, resembling iron (Fe) or copper (2). In bacteria, Mn is involved in diverse cellular pathways like photosynthesis, gluconeogenesis, glycolysis, signal transduction, stringent response, sporulation, and pathogenesis (3). It has been proposed that the survival strategies of a few microorganisms, such as Borrelia burgdorferi, may have an obligate requirement for Mn (3, 4). Therefore, the need for Mn has evolved in parallel with the need for other metals in biological systems.
In recent times, Mn has gained great attention for its role in oxidative stress and Fe starvation (5–8). Oxidative stress stimulates the influx of Mn, leading to activation of Mn-dependent superoxide dismutase (SodA) to combat the pool of cytoplasmic superoxide radicals (5). Furthermore, Mn is also able to replace Fe from at least four nonredox Fe enzymes, viz. ribulose-5-phosphate-3-epimerase, peptide deformylase, threonine dehydrogenase, and cytosine deaminase (6, 7). Such a replacement of the Fe atom by nonoxidizable Mn restores the functions of these enzymes when Escherichia coli is under oxidative stress (6, 7). An alternative Mn-dependent ribonucleotide reductase (NrdEF) ensures chromosome replication when the Fe-dependent ribonucleotide reductase (NrdAB) is nonfunctional under Fe starvation in E. coli (8). However, excess Mn is hazardous to all life forms, since it has ability to inactivate enzymes or proteins by replacing Fe, Mg, or other metal cofactors (5, 9, 10).
The current picture of the Mn homeostasis network in standard laboratory strains of E. coli constitutes a Mn-dependent transcriptional regulator, MntR, which upregulates Mn efflux pump, MntP, and downregulates the Mn importer MntH (10, 11). However, some pathogenic E. coli strains have additional Mn importer-encoding genes, viz. sitABCD, which are also regulated by MntR (12, 13). A bifunctional RNA, which acts either as an mRNA encoding a small 42-amino-acid peptide (MntS) or as a small regulatory RNA (rybA), has been reported to be involved in Mn homeostasis (10, 14). It has been proposed that MntS exerts its role in Mn homeostasis in E. coli by acting as an Mn chaperone (10).
In this study, we demonstrate that a dispensable E. coli gene, hflX, is strongly linked with the process of Mn homeostasis. hflX was originally discovered as a part of genes belonging to the high-frequency lysogenization locus A (hflA) in the E. coli genome and thus was initially proposed to be involved in controlling the λ phage life cycle (15). However, the biological function of HflX remained “enigmatic,” as one of our earlier studies rejected such a role for HflX under standard laboratory conditions (16). Instead, we and others have shown that HflX is a P-loop GTP/ATPase that interacts with the 50S subunit of the ribosome in E. coli and other bacteria (16–19). In the present work, we show that ΔhflX E. coli is unable to regulate the influx of environmental Mn. Moreover, the Mn influx is found to perturb Zn and Fe homeostasis, altering the gene expression profiles of the regulatory networks of these metals in the ΔhflX strain. As a consequence, ΔhflX cells exhibit a number of abnormalities to Mn stress: growth retardation, cell filamentation, a lower rate of DNA replication, and DNA damage. Thus, HflX is a novel regulator of Mn homeostasis in E. coli.
MATERIALS AND METHODS
Strains, phages, and proteins.The strains C600, BW25113, JW2388 (ΔmntH::Kan), JW5830 (ΔmntP::Kan), JW0801 (ΔmntR::Kan), JW0669 (Δfur::Kan), JW1847 (ΔznuC::Kan), JW5831 (ΔznuA::Kan), and JW5165 (ΔhflD::Kan) were obtained from the Coli Genetic Stock Center (CGSC). The E. coli BW25113 ΔhflX strain was generated as described previously (16). The strain 10973 (BW28357-recA-gfp) was a gift from Max. E. Gottesman, Columbia University Medical Center, New York, NY. The P1 phage transduction method was used to construct ΔhflX ΔmntH, ΔhflX ΔmntP, ΔhflX ΔmntR, ΔhflX Δfur, ΔhflX ΔznuC, ΔhflX ΔznuA, BW25113-recA-gfp and ΔhflX-recA-gfp strains. The λNK1324 phage was obtained from Abhijit Sardesai, Center for DNA Fingerprinting and Diagnostics, India. The phage λc+cam105 was a generous gift from Sankar Adhya, NIH. The purification of 6×His-tagged HflX was performed as described previously (16).
Autophosphorylation assay.Autophosphorylation of HflX was assayed by incubating 5 μg protein with 10 μCi [γ32-P]GTP (specific activity, 4,500 Ci mmol−1) in 20 μl phosphorylation buffer (20 mM Tris-HCl, 200 mM NaCl, 5 mM MgCl2, 1 mM phenylmethylsulfonyl fluoride [PMSF], 1 mM dithiothreitol [DTT]) at 37°C for 40 min. Similar reactions were set up by replacing MgCl2 from the phosphorylation buffer with 5 mM MnCl2 or CaCl2. EDTA (10 mM) was applied to the Mn-mediated autophosphorylation reaction to check whether the divalent ion is required for autophosphorylation, as illustrated in Fig. 1. The reaction was quenched by addition of 1 mM GTP and 5× SDS-PAGE sample loading buffer. Samples were applied to 12.5% SDS-PAGE to resolve the protein bands and were blotted onto the Immobilon-P membrane (Millipore). The blots were stained with Ponceau S to visualize the bands and subjected to autoradiography. The relative band intensities were estimated by Scion Image software.
Manganese-dependent autophosphorylation of HflX. His-tagged HflX and untagged HflX were autophosphorylated with [γ-32P]GTP in the presence of different divalent metal ions or EDTA, separated by SDS-PAGE, and transferred to polyvinylidene difluoride (PVDF) membrane. An autoradiogram of the PVDF membranes shows Mg-, Mn-, and Ca-mediated autophosphorylation of both His-tagged HflX (A) and untagged HflX (B). Mn stimulates autophoshorylation better than Mg or Ca. EDTA suppresses autophosphorylation. The relative band intensities were quantified by Scion Image software, as indicated.
Growth conditions and growth curves.Bacterial strains were grown in LB broth at 37°C. The growth curves were generated using an automated BioscreenC growth analyzer (Oy Growth Curves Ab, Ltd.). Ten microliters of log-phase cultures of the specified strain was freshly diluted to 1 ml of LB. Different metal salts and other supplements were added to the diluted culture, as specified. Two hundred microliters of these diluted cultures was grown in wells of honeycomb multiwell plates at 37°C.
Confocal microscopy and fluorescence assay.E. coli cells grown in the absence or presence of Mn were collected, fixed with 4% paraformaldehyde solution, and stained with diamidino-2-phenylindole (DAPI). The morphology of cells spotted on a clean glass slide was visualized by a confocal microscope (LSM 510; Carl Zeiss). The BW25113-recA-gfp and ΔhflX-recA-gfp strains were grown in the absence or presence of Mn, and a small drop was directly mounted on a slide to look for green fluorescent protein (GFP) fluorescent foci under a confocal microscope. The rest of the cells were harvested and lysed by sonication. GFP fluorescence was quantified by a fluorimeter (λ excitation, 395 nm; λ emission, 509 nm). Protein concentrations of the lysates were determined by Bradford reagent (Bio-Rad), and fluorescence values were normalized against total protein concentrations.
[3H]thymidine incorporation experiment.Cells were grown in the absence or presence of 1 mM Mn, and 0.2 ml of culture was pulse-labeled for 2 min with 10 μl of [3H]thymidine (MP Biomedicals) (80 Ci/mmol, 1 mCi/ml) at different time points. Ice-cold trichloroacetic acid (TCA) (final concentration, 10%) was mixed with labeled samples and incubated on ice for 30 min. Samples were filtered on Whatman glass-microfiber filters (GF/C) (Sigma) with vacuum assembly and washed three times with 10 ml of ice-cold 5% TCA. Filters were dried and added to scintillation cocktail fluid, and the amount of radioactivity incorporated into nucleic acid was determined by a scintillation counter (PerkinElmer).
Tn10::Camr mutagenesis strategy.A λNK1324 phage-mediated Tn10::Camr transposon mutagenesis strategy was used to identify the suppressors for the ΔhflX phenotype (20). The C600 strain was used to enhance the phage titer. Upon transduction to ΔhflX E. coli (multiplicity of infection, 0.1), the phage delivers the transposon into the cell. The transposon is inserted randomly into the genome and gives rise to ΔhflX suppressor colonies on an LB agar plate supplemented with 50 μg/ml chloramphenicol and 1 mM MnCl2. Genomic DNA from 10 suppressor colonies was isolated using a Qiagen kit. A total of 500 ng of genomic DNA was digested with Sau3AI followed by ligation. Inverted PCR was done to identify the genomic locus as described previously (21). We mapped the suppressors in either znuA (8 colonies), znuC (1 colony), or fur (1 colony) genes.
AAS.The intracellular concentrations of Mn, Fe and Zn were determined by atomic absorption spectroscopy (AAS). LB medium (500 ml) was inoculated with 5 ml of overnight-grown cultures in the absence or presence of Mn, and cells were allowed to further grow for 3 h at 37°C. The cell pellets were harvested and washed three times with 100 ml normal saline containing 1 mM EDTA to remove extracellular divalent ions and once with 100 ml normal saline to remove EDTA. Cells were resuspended in 3 ml normal saline and lysed by sonication. The protein concentration in the lysate was determined using Bradford reagent (Bio-Rad). One milliliter of lysate was digested with 7% HNO3 for 6 h at 85°C with proper shaking. AAS was performed to estimate Mn, Fe, and Zn concentrations by using the graphite furnace technology of Analytik Jena 5EA equipment (Carl Zeiss Jena GmbH). Each time, 20 μl of sample was injected, followed by drying, pyrolysis, and atomization of liquid samples. Atomization of Fe, Zn, and Mn was done at 1,850, 1,100, and 1,600°C, respectively. The values of the peak area were compared with the standard curve, generated using standard salt solutions, to determine the ion concentration. Cellular ion concentrations were calculated by normalizing them to a total intracellular protein concentration of 300 mg/ml (5). All glassware and plasticware used in this study were rinsed with 1% HNO3 several times, finally washed with double-distilled water, and dried.
Real-time PCR.Bacterial mRNAs were isolated from the strains grown in the absence or presence of 1 mM Mn for 3 h at 37°C, using TRIzol reagent and the Qiagen bacterial RNA isolation kit. DNase I treatment was done to remove residual DNA contaminant, and the integrity of the mRNA was checked on a 1% agarose gel. The mRNA concentration was determined by a NanoDrop spectrophotometer (Thermo Scientific) as well as by a conventional spectrophotometer. RNA samples (250 ng) were used for each quantitative reverse transcription-PCR (qRT-PCR) to amplify 100- to 150-bp fragments of target genes by one-step qRT-PCR (Invitrogen kit). No-template reactions were included as negative controls. At least three independent experiments were conducted for each primer pair, and cycle threshold (CT) values were obtained. The fold change in expression (Mn treated versus untreated) was calculated by the ΔΔCT method. The values were normalized to the level of tufA mRNA, which is expressed constitutively.
Determination of λ phage lysogenic frequency.The lysogenic frequencies were determined as described in reference 16, with the following modifications as specified below. The WT and ΔhflX strains were grown in 4 ml tryptone broth at 30°C up to an optical density at 600 nm (OD600) of 0.2. Next, MnCl2 was added to a final concentration of 0.5 mM and allowed to grow until the OD600 reached 0.5 to 0.8. Cells were harvested and resuspended in 200 μl of 10 mM MgSO4 plus 100 μM MnCl2 and incubated at 30°C for 20 min to generate the nutrient-deprived condition, wherever required. Where mentioned, the Zn chelator tetrakis-(2-pyridylmethyl) ethylenediamine (TPEN; Sigma Chemicals) was added at 200 μM at this point, and its concentration was maintained throughout the experiment. The λc+cam105 phage was added at a multiplicity of infection of 0.1, and adsorption was allowed for 30 min at 4°C. The free phage was removed by centrifugation, and cells were washed with 10 mM MgSO4. The bacteria were resuspended in 1 ml LB containing 0.5 mM MnCl2 and incubated for 30 min at 37°C. The infected bacteria were spread over an LB agar-MnCl2 (100 μM) plate supplemented with or without 50 μg/ml chloramphenicol and incubated at 37°C overnight to determine the number of lysogens and infection centers.
RESULTS
GTP-mediated autophosphorylation of HflX using Mn as a cofactor.The GTPase and ATPase activities of HflX were demonstrated earlier (16, 18). In this study, we observed GTP-mediated phosphorylation of purified HflX. Incubation of purified 50-kDa N-terminal His-tagged HflX with radioactive [γ-32P]GTP and three different metal ions (Mg2+, Mn2+, and Ca2+) induced different degrees of phosphorylation, in the order Mn2+ > Mg2+ > Ca2+ (Fig. 1A). Generally, protein phosphorylation requires Mg2+ as a cofactor, but phosphorylation of HflX was found to be more efficient in the presence of Mn2+ (Fig. 1A). When phosphorylation was performed (in the presence of the above-mentioned divalent cations), after removal of the N-terminal histidine tag of the recombinant HflX (by proteolysis by thrombin), the phosphorylation patterns and efficiencies remained the same (Fig. 1B). Thus, the histidine tag had no influence on the phosphorylation of HflX. To check whether ATP can also phosphorylate HflX, we replaced [γ-32P]GTP with [γ-32P]ATP in the reaction mixture. No radioactive signal was observed in the autoradiogram, indicating that phosphorylation was GTP specific (data not shown). Since no other protein was involved in the phosphorylation assay, we conclude that the HflX phosphorylation is an autophosphorylation event.
Mn-sensitive phenotypes of ΔhflX E. coli.The ΔhflX strain of E. coli was generated earlier (16). Analysis of phenotypic alteration in the absence of a gene is a primary criterion to annotate the function of the gene. However, the functions of about 33% of the genes of E. coli are not known since the corresponding gene knockout constructs exhibit no distinct phenotypic changes under standard growth conditions in the laboratory (22). Since the ΔhflX strain of E. coli falls under this category, it was an uphill task to find a particular condition that will bring out a phenotypic alteration in the ΔhflX strain. The preference of Mn to Mg in the autophosphorylation of HflX (Fig. 1) prompted us to check if HflX was involved in the Mn stress response mechanism. Indeed, the ΔhflX strain was found to exhibit Mn-mediated hypersensitivity, characterized by a dramatic inhibition of cell growth in the presence of increasing concentrations of MnCl2 (Fig. 2A). The growth profile of the wild-type (WT) strain, however, remained unchanged under identical conditions (Fig. 2A). In parallel, we checked the effect of a series of di- and trivalent metal ions (zinc, nickel, iron, lead, cobalt, copper, and calcium) on the growth profiles of WT and ΔhflX E. coli. These ions exhibited no differential impact on the growth of ΔhflX E. coli (see Fig. S1 in the supplemental material). These observations clearly imply the specific susceptibility of the ΔhflX E. coli strain to Mn. We, therefore, checked for possible morphological changes of ΔhflX cells under Mn stress by examining these cells under a confocal microscope. The ΔhflX E. coli cells exhibited a pronounced filamentous morphology (Fig. 2B), suggesting inhibition of cell division under Mn stress.
Extreme manganese-sensitive phenotypes of ΔhflX E. coli. (A) Growth curves of the WT and ΔhflX strains were generated with increasing concentrations of MnCl2 to show that the ΔhflX strain, but not the WT, is extremely sensitive to Mn. Error bars represent propagation of the standard errors from three independent experiments (B) The untreated and Mn-treated cells were stained with DAPI and mounted on a slide for observation under a confocal microscope. Mn-perturbed ΔhflX cells exhibit a cell filamentation phenotype. The morphologies of WT (untreated and Mn-treated) and ΔhflX (untreated) cells of E. coli are unaffected. (C) The Mn-treated ΔhflX cells, which harbor recA-gfp in their genome, exhibit enhanced GFP fluorescence, suggesting a DNA damage response. WT (untreated and Mn-treated) and ΔhflX (untreated) cells, which also harbor recA-gfp, show the background level of GFP fluorescence. DIC, differential interference contrast. (D) The fluorescence intensities were quantified separately using a fluorimeter and normalized against total cellular protein content. The 1.6-fold upregulation of GFP fluorescence in the Mn-perturbed ΔhflX strain supports the DNA damage response in vivo. An asterisk denotes significant difference (P < 0.01, Student's t test). (E) The replication fork progression is highly affected in the Mn-perturbed ΔhflX strain. Incorporation of [3H]thymidine was monitored in the unperturbed or Mn-perturbed WT and ΔhflX cells, and the percentage of inhibition was calculated by taking ratio of incorporated radioactive counts in the Mn-treated versus untreated strains at different time intervals and plotted accordingly. The rate of replication was decreased in the Mn-perturbed ΔhflX strain. Error bars represent propagation of standard errors from three independent experiments.
Mn stress affects DNA metabolism in the ΔhflX strain.Filamentation of E. coli is often observed as a result of bacteria responding to inhibition of replication and DNA damage (23). Inhibition in the progression of the replication fork causes fork regression, which ultimately can lead to DNA double-strand breakage (24). Conversely, DNA damage can also lead to inhibition of replication fork movement (25, 26). In particular, Mn-dependent infidelity of DNA synthesis by DNA polymerases can increase the rate of mutagenesis (27, 28), which may lead to DNA damage, SOS response, and inhibition of the replication fork. We examined the DNA damage phenomenon by introducing a recA-gfp reporter into the ΔhflX E. coli cells. This reporter system is useful for monitoring DNA damage in vivo, as SOS response upregulates RecA-GFP, which forms distinct green fluorescent foci at the damage site of DNA (29). As expected, strong GFP fluorescent foci in the filamentous ΔhflX cells were visualized under a confocal microscope (Fig. 2C). We also quantified average fluorescence intensities (normalized against total protein content) of the Mn-treated or untreated cells. The fluorescence intensity of Mn-treated ΔhflX cells increased 1.6-fold compared to that of the untreated WT cells (Fig. 2D). The untreated ΔhflX cells also exhibited marginal enhancement of fluorescence compared to the WT cells, while no change in the fluorescence intensity was observed for Mn-treated WT cells (Fig. 2D). In another experiment, the rate of [3H]thymidine incorporation was measured in the DNA of E. coli strains. As shown in Fig. 2E, the rate of [3H]thymidine incorporation in the ΔhflX strain was drastically reduced, while WT cells showed an unaltered rate of incorporation of [3H]thymidine in the presence of Mn. This experiment confirms that the progression of the replication fork is severely affected under Mn stress in the ΔhflX strain. Together, these results suggest that cell filamentation is the result of an impaired progression of the replication fork and a DNA double-strand break in the Mn-perturbed ΔhflX cells.
HflX maintains Mn homeostasis by an MntR-independent mechanism.Mn-mediated hypersensitivity of ΔhflX E. coli appeared to be the result of an excessive accumulation of Mn in the cell. Initially we assumed that HflX maintains Mn homeostasis in E. coli by inhibiting the MntH importer, stimulating the MntP exporter system, or both. Thus, the absence of HflX might result in the accumulation of excessive Mn. Since MntH is the sole importer of Mn in the laboratory strain of E. coli known to date (10), we expected a total recovery of Mn-sensitive growth upon deletion of mntH from the ΔhflX strain. Surprisingly, comparison between the growth curves of the ΔhflX and the ΔhflX ΔmntH strains revealed that the absence of the MntH importer in the latter could only partially recover the Mn-mediated growth inhibition observed in the former (Fig. 3A). In addition, the ΔhflX ΔmntH strain was more sensitive to Mn than the ΔmntH strain (Fig. 3A). These two observations suggest an alternative route of Mn import, which is activated in the absence of hflX in E. coli. It is noteworthy that a low-affinity alternative route for Mn transport in E. coli has also been proposed previously (5). It will be interesting to check if this low-affinity route of Mn import works as a high-affinity importer in the absence of HflX. Next, the growth profiles of ΔmntP and ΔhflX ΔmntP strains, grown in the absence or presence of Mn, were compared with that of the ΔhflX strain. As expected, the ΔhflX ΔmntP strain exhibited a greater growth defect than the ΔhflX and ΔmntP strains under Mn stress, suggesting a synergistic inhibition of growth by Mn (Fig. 3B). If HflX stimulated MntP function, one would expect a greater Mn sensitivity of the ΔmntP strain than the Mn sensitivity of the ΔhflX strain. Quite the opposite occurred: Mn stress exerted a lesser growth defect in the ΔmntP strain than the growth defect of the ΔhflX strain (Fig. 3B). Therefore, this result suggests that HflX cannot be a modulator of MntP efflux function. Since the transcription regulator MntR upregulates MntP and downregulates MntH in order to maintain Mn homeostasis (10, 11), we wondered whether stimulation of MntR activity by HflX is responsible for this Mn homeostasis. If this were the case, ΔmntR cells would be more sensitive to Mn than the ΔhflX cells. Interestingly, we observed that the ΔhflX strain, but not the ΔmntR strain, was highly susceptible to Mn (Fig. 3C). Therefore, we conclude that HflX is not a part of the mntR regulon but maintains Mn homeostasis in the laboratory strain of E. coli via an unknown mechanism.
HflX-mediated Mn homeostasis is not linked to the MntR regulon (A) The ΔhflX ΔmntH strain only partially rescues the growth defect that was observed in the ΔhflX strain under 1.5 mM Mn. This observation suggests that apart from the MntH importer, there is another alternative route of Mn import activated in the absence of HflX. (B and C) Comparison between the growth curves of the ΔhflX and ΔmntP strains or ΔhflX and ΔmntR strains, shows that ΔhflX is more sensitive to Mn than the ΔmntP or ΔmntR strain. That the ΔhflX ΔmntP and ΔhflX ΔmntR double mutants were found to be more sensitive than single mutants, suggests a synergistic effect of inactivation of two different Mn homeostasis pathways. Therefore, HflX cannot be a stimulator of either MntP or MntR. Error bars represent the standard errors calculated from the three independent experiments.
Mn influx elevates the cellular Zn level and reduces the Fe level.Spontaneous suppressors that overcome the growth defect and cell filamentation (ΔhflX phenotype) under Mn stress originated at a high rate (>10−6) on an LB agar plate containing 1 mM Mn (data not shown). In order to determine whether evolution of these suppressors was the result of the inactivation of the proposed alternative route(s) of Mn import in the ΔhflX strain, we took advantage of a λNK1324 phage-mediated transposon mutagenesis strategy (20). The Tn10::Camr transposon harbored in the phage DNA was allowed to insert randomly into the genome to disrupt certain genes of ΔhflX E. coli, and, thereby, a number of suppressor colonies that overcame the ΔhflX phenotype were isolated (Fig. 4A). We picked a few of the suppressors and identified the disruptions at znuC or znuA (genes that encode the subunits of the Zn ABC importer) or at fur (a gene that encodes the ferric uptake regulator that represses Fe import) (30–32). Mapping of these suppressors elucidates that stimulated Zn influx and Fe starvation are the major causes of the ΔhflX phenotype. We next added Zn or Fe to the growth medium in order to check whether these two ions aggravate or rescue the existing growth defect, respectively, of the ΔhflX cells under Mn stress. As expected, Zn reduced the growth rate further, while Fe triggered substantial rescue of the growth defect of Mn-perturbed ΔhflX cells (Fig. 4B and C). In another experiment, desferrioxamine (DFO), a cell-penetrable Fe chelator, was added to the growth medium. DFO was found to magnify Mn toxicity (Fig. 4D), which reinforces our claim that Fe gets deprived in the Mn-stressed ΔhflX cells.
Mn-perturbed ΔhflX in E. coli affects Mn, Zn, and Fe homeostasis. (A) Growth profile of the suppressors (Sup 1 to Sup 5) originated by λNK1342 phage-mediated mutagenesis. Mid-log-phase cultures of the strains were serially diluted and spotted onto an LB agar plate supplemented with 1 mM Mn. The suppressors exhibit normal growth comparable to that of the WT strain. The ΔhflX strain shows extreme Mn sensitivity. (B) Indicated cells were grown in the absence or presence of 0.5 mM MnCl2. Addition of 0.5 mM ZnCl2 elevates the growth defect of Mn-stressed ΔhflX E. coli, suggesting that Zn is toxic under Mn stress. (C) Similarly, E. coli strains were grown in the absence or presence of 1 mM Mn. Supplementation of increasing concentrations of FeCl3 in the growth medium rescues the growth inhibition of Mn-stressed ΔhflX E. coli. (D) Chelation of intracellular Fe by 0.25 mM DFO magnifies the growth defect of Mn-perturbed ΔhflX E. coli. At least three independent experiments were performed to represent standard error bars. (E) Intracellular concentrations of different metal ions in the WT and ΔhflX strains (untreated or Mn treated) were determined by AAS analysis. The bar diagram suggests that Mn influx is positively correlated with simultaneous Zn influx while inversely correlated with Fe influx in the Mn-treated ΔhflX strain. The numbers on top of each bar indicate the μM concentrations of corresponding ions. *, significant difference (P < 0.01, Student's t test) compared with the corresponding Mn levels in the WT (both unperturbed and Mn perturbed) and ΔhflX cells; #, significant difference (P < 0.01, Student's t test) compared with the corresponding Zn levels in the WT (both unperturbed and Mn perturbed) and ΔhflX cells; ◆, significant differences (P < 0.01, Student's t test) compared with the corresponding Fe levels in the WT (both unperturbed and Mn perturbed) and ΔhflX cells. (F) Under the nutrient-deprived condition, Mn-perturbed ΔhflX cells exhibited low-frequency lysogenization of λ phage. Addition of TPEN, a Zn chelator, induces the lysogenic frequency in the Mn-perturbed ΔhflX cells. These observations lead to speculation that increased Zn transport in the Mn-treated ΔhflX strain elevates the protease activity of HflB. * and # denote the significant differences (P < 0.01, Student's t test) compared with the unperturbed and Mn-perturbed ΔhflX strain, respectively. (G) The frequency of lysogenization in the ΔznuA strain, a zinc transporter-defective mutant, was recorded to be significantly higher in comparison to the WT, which suggests that import of Zn has a role to regulate lysis and lysogeny. However, the frequency of lysogenization is considerably lower than in the ΔhflD strain, a mutant that shows a high-frequency lysogenization phenotype. * and ◆, significant differences (P < 0.01, Student's t test) compared with the ΔhflD and WT strains, respectively.
To determine the cellular levels of Zn and Fe under Mn stress, we carried out atomic absorption spectroscopy (AAS). A 7-fold accumulation of Mn in the ΔhflX cells (663 μM versus 96 μM in WT cells) confirmed the existence of an uncontrolled influx of supplemental Mn (Fig. 4E). However, WT and ΔhflX cells without Mn supplement showed levels of 21 and 39 μM Mn, respectively (Fig. 4E). The cellular accumulation of Zn (460 μM versus 187 μM in the WT) and reduction of Fe (360 μM versus 1.3 mM in WT) (Fig. 4E) directly authenticates our claim of Fe starvation and Zn accumulation in the Mn-stressed ΔhflX cells (Fig. 4B, C, and D). These data also show the intricate connection among the Mn, Fe, and Zn homeostasis networks.
Although the current set of observations suggests that Mn influx caused an elevation of the level of Zn and reduction of Fe in the Mn-perturbed ΔhflX strain, it is difficult to predict a possible alternative channel for Mn import. Since 9 out of 10 suppressors in our study were found to evolve by disruption of znuA or znuC, we speculate that the high-affinity Zn transporter, ZnuABC, may transport both zinc and manganese in the absence of HflX. This issue will be discussed in detail later.
Zn-enriched ΔhflX cells exhibit a low frequency of lysogenization of bacteriophage λ.Earlier we were unsuccessful to connect hflX with the decision making process of the λ phage lysis-lysogeny life cycle (16), because we lacked the information that the cellular level of Zn increases in the Mn-perturbed ΔhflX strain. The latter has been established in this study (Fig. 4E). Thus, we assumed that the elevated Zn level may activate HflB, an essential Zn metalloprotease that primarily determines the phage life cycle by degrading λCII protein (33) in the Mn-perturbed ΔhflX strain. To investigate this possibility, we estimated the frequency of lysogenization of bacteriophage λ in the Mn-perturbed E. coli cells. WT and ΔhflX strains were grown in the absence or presence of Mn and subjected to nutrient deprivation to enhance frequency of lysogenization. In this way, the lysogenization frequencies in both WT and ΔhflX cells were recorded as ∼15% in the absence of Mn (Fig. 4F). As expected, a striking reduction in the lysogenization frequency (from ∼15% to ∼5%) was observed in the Mn-perturbed ΔhflX strain (Fig. 4F). Addition of the Zn chelator TPEN enhanced the frequency of lysogenization (∼5% to ∼12%) in the Mn-perturbed ΔhflX strain (Fig. 4F). A marginal reduction or enhancement of the lysogenization frequency was also observed in the Mn-perturbed WT strain in the absence or presence of TPEN, respectively (Fig. 4F). To rule out the counterargument that stimulation of HflB protease activity could be the result of hflB overexpression, we performed quantitative-RT-PCR (qRT-PCR) to show that the level of hflB mRNA in the Mn-perturbed ΔhflX strain remained unaltered (see Table S1 in the supplemental material).
Another assay was conducted under the nutrient-enriched condition to compare the lysogenization frequency in the ΔznuA strain, a Zn transporter-defective mutant, with the lysogenization frequency in the WT and ΔhflD strains. Unlike the nutrient-deprived condition, the nutrient-enriched condition allows a very low frequency of lysogenization in the WT strain (16). On the other hand, the ΔhflD mutant naturally exhibits a very high frequency of lysogenization of bacteriophage λ (33). The ΔznuA strain exhibited a frequency of lysogenization (11%) that was substantially higher than that in the WT (4 to 5%), although not very high compared to that exhibited by the ΔhflD strain (20%) (Fig. 4G). Nevertheless, the enhanced frequency of lysogenization in the ΔznuA strain compared to that in the WT strain indicates that the limited influx of the Zn in the former could promote the choice for lysogenization by bacteriophage λ (Fig. 4G). Considering all of these arguments, we postulate that the increased level of the intracellular Zn in the Mn-perturbed ΔhflX strain (Fig. 4E) boosts λCII degradation by enhancing HflB protease activity.
Altered transcription profile of metal regulatory networks in the Mn-perturbed ΔhflX strain.Intracellular metal ions are known to regulate gene expression by activating transcription regulators in order to maintain their physiological levels. Mn and Fe, for example, activate the MntR (MntR-Mn2+ complex) and Fur (forms the Fur-Fe2+complex) transcription regulators, respectively (10, 11, 14, 31). Similarly, Zn activates the Zur or ZntR (Zur-Zn2+ or ZntR-Zn2+ complex) regulatory networks (30, 34). However, alternative mechanisms of regulation, viz. Mn-mediated activation of Fur, Fe-mediated activation of MntR, or Mn-mediated inactivation of Zur, have also been proposed (10, 30, 34, 35, 36). In our present study, accumulation of Mn in the Mn-perturbed ΔhflX strain increases Zn and decreases Fe levels, suggesting that an influx of Mn is capable of affecting both Fe and Zn regulatory networks (Fig. 4E). To examine these possibilities, we performed qRT-PCR to check the expression profiles of the genes that are involved in the metal regulatory networks, mentioned above. As expected, a huge downregulation of rybA/mntS, mntH, and dps and upregulation of mntP were observed in the Mn-stressed ΔhflX cells (Fig. 5), indicating that an elevated level of intracellular Mn activated the MntR regulator. On the other hand, a repression of Fe transport genes (fepA and feoB) and of enterobactin biosynthetic genes (entA and entB), which are under the control of the Fur regulon (1), supports our hypothesis that elevated levels of Mn can activate Fur in the Mn-stressed ΔhflX cells (Fig. 5). This effect of Mn, causing repression of the genes involved in Fe transport and enterobactin biosynthesis, explains why Mn-stressed ΔhflX cells suffer from Fe starvation. Although Fur represses and activates sodA and sodB, respectively (37), quite opposite expression profiles of these two genes were observed under Mn stress (Fig. 5). These discrepancies may arise from the presence of complex transcription regulation by other multiple regulators that additionally regulate sodA and sodB expression as represented in the EcoCyc database (1). On the other hand, Mn-mediated inactivation of the Zn regulatory network was observed earlier (30, 34). An especially high level of Mn alters the pattern of the Zur regulator footprint on the DNA element, suggesting a possible inactivation of Zur by Mn (34). Therefore, we examined whether an elevated level of Mn in the ΔhflX cells facilitates accumulation of Zn by upregulating and downregulating Zn importer and exporter genes, respectively, in the Zur and ZntR regulatory networks. Surprisingly, a slight downregulation of the high-affinity Zn importer genes (znuA and znuC) and upregulation of the Zn exporter gene (zntA) were observed (Fig. 5; see Table S1 in the supplemental material). To explain this result, we argue that the increasing level of Zn, which is the natural cofactor for Zur and ZntR regulators, in the Mn-perturbed ΔhflX cells may compete with the increasing level of Mn in order to bind the regulators. Thus, Zn-mediated reactivation of the regulators nullifies the deactivation effect of Mn in the ΔhflX cells. It is also notable that downregulation of zinT (3-fold), coding for a Zn binding protein, upregulation of fieF (4.5-fold), coding for a Zn/Fe exporter, and upregulation of zraP (3.5-fold), coding for a periplasmic Zn binding chaperone (Fig. 5; see Table S1), reflected the presence of high level of intracellular Zn in the Mn-perturbed ΔhflX cells (Fig. 4E).
Gene expression profile of metal regulatory networks. We performed qRT-PCR to measure the up- or downregulation of genes of the Mn, Fe, and Zn regulatory networks in the Mn-perturbed ΔhflX strain. The y axis indicates folds of up- or downregulation of the mRNA levels in the Mn-perturbed ΔhflX strain compared to the ΔhflX strain. *, significant difference (P < 0.01, Student's t test) with at least 2-fold up- or downregulation of the genes in the Mn-perturbed ΔhflX cells compared with expression in the unperturbed ΔhflX cells.
Activated Fe import or inactivated Zn import reduces Mn import in the suppressors.We knocked out fur, znuC, or znuA (Δfur, ΔznuC, or ΔznuA, respectively) from the ΔhflX strain by P1 transduction to generate clean suppressors for Mn-sensitive phenotypes. The comparative growth curves of the ΔhflX strain and the suppressors (ΔhflX Δfur, ΔhflX ΔznuC, or ΔhflX ΔznuA) are shown in Fig. 6A. The growth profile of the Mn-perturbed ΔhflX Δfur strain is quite similar to the growth profile of the ΔhflX strain (Fig. 6A). However, the ΔhflX ΔznuC and ΔhflX ΔznuA strains represented a prolonged lag phase, which signifies that the major loss of active Zn import is also not very healthy for the normal growth of E. coli (Fig. 6A). Hence, an optimum level of Zn in the cell seems to be crucial for the maximum growth of E. coli. This observation and the result illustrated in Fig. 4C indicate that upregulation of Fe import probably ensures proper metallation of Fe-apoproteins in order to allow healthy growth of E. coli under Mn stress in the absence of HflX. On the other hand, the Mn-resistant phenotypes of the ΔhflX ΔznuC and ΔhflX ΔznuA suppressors (Fig. 6A) apparently reflect that excess Zn import in the Mn-perturbed ΔhflX strain elevates Zn toxicity (Fig. 4B). We asked whether evolution of such suppressors was also meant to overcome the accumulation of Mn in the cell. To check this, the cellular Mn levels in the Mn-perturbed suppressors were estimated. A moderate level of Mn accumulation in the suppressors (42 μM in ΔhflX Δfur cells, 60 to 70 μM in ΔhflX ΔznuA cells, and 50 to 60 μM in ΔhflX ΔznuC cells) compared to that in the Mn-perturbed ΔhflX cells (663 μM) suggests that reduction in Mn accumulation in the suppressors is also important in combating Mn stress (Fig. 6B). Therefore, a link between the increased Fe transport and the decreased Zn transport with the reduced Mn transport in the suppressors once again elucidates that homeostases among the metal trio are intricately interconnected. We performed qRT-PCR to check the expression profile of the metal transport genes. Analysis of these data reveals a metal-mediated complex regulatory pattern by activation or inactivation of Fur, MntR, Zur, and ZntR regulatory networks, as described in the supplemental material (see Fig. S2).
Stimulation of Fe import or inhibition of Zn import in the suppressors also inhibits Mn import. (A) The growth patterns of the different suppressors (ΔhflX Δfur, ΔhflX ΔznuC, and ΔhflX ΔznuA) exhibit a significant suppression of growth retardation compared to that in the Mn-perturbed ΔhflX strain. (B) The AAS result compares cellular Mn concentrations in the unperturbed and Mn-perturbed E. coli strains. *, significant difference (P < 0.01, Student's t test) between the marked groups.
DISCUSSION
A robust link between HflX and Mn homeostasis is discovered.Despite the fact that E. coli has been studied extensively as a model organism, the functions of about one-third of its genes are still unknown (22). In this work, we unravel the function of one such nonessential gene, hflX, which was discovered almost 3 decades ago by Banuett and Herskowitz (15). We demonstrate that an uncontrolled influx of Mn leads to catastrophic consequences for the DNA metabolism and normal cellular growth of the ΔhflX strain (Fig. 2). As a result, the SOS response and filamentous phenotype of the Mn-perturbed ΔhflX cells are observed (Fig. 2). No other divalent metal ions exert such a deleterious impact on the growth of the ΔhflX strain (see Fig. S1 in the supplemental material). Thus, HflX is involved in the process of Mn homeostasis in E. coli. So far, MntR, an Mn binding dual regulator activated by increasing concentrations of intracellular Mn, is known to regulate Mn homeostasis in E. coli (10, 11). Activated MntR downregulates the MntH importer and upregulates the MntP exporter in order to maintain the cellular level of Mn (10, 11). Our results demonstrate that the HflX-mediated Mn homeostasis is unrelated to the function of MntH-, MntP-, or MntR-dependent homeostasis of Mn (Fig. 3). Hence, HflX-mediated Mn homeostasis is considered a novel pathway completely independent of the MntR miniregulon. Furthermore, the growth of the ΔhflX strain is much more retarded than that of the ΔmntR or ΔmntP strains in the presence of 0.5 mM Mn (Fig. 3). This observation suggests that HflX-mediated Mn homeostasis is more robust than MntR-mediated Mn homeostasis.
Autophosphorylation of HflX could be a starting point of Mn homeostasis signaling.GTP-dependent autophosphorylation of HflX using Mn as a primary cofactor guided us to demonstrate the relationship between HflX and Mn homeostasis. On the other hand, the GTPase and ATPase activities of HflX (16, 18), which requires Mg as the sole cofactor, are strongly inhibited by Mn (data not shown). Therefore, we speculate that Mn-mediated autophosphorylation of HflX (Fig. 1) is most likely the starting point of a signal transduction cascade that ultimately inhibits toxic Mn import. Molecular details of the autophosphorylation of HflX are under investigation. In addition, HflX is a ribosome-associated protein in E. coli (16, 18). The biological implication of this ribosome-HflX interaction has remained unexplained to date. Most recently, it has been shown that apart from the canonical nucleotide binding C-terminal domain (G-domain), the N-terminal glycine-rich domain of HflX also exhibits a noncanonical nucleotide binding property (38). Interestingly, these two nucleotide binding domains of HflX influence each other's NTPase property (38). Therefore, we believe that HflX is a multifunctional protein that not only plays a role in metal homeostasis but also modulates ribosome functions or its biogenesis under certain conditions.
HflX restores multiple metal homeostasis networks under Mn stress.We identified two different types of suppressors, which can either derepress Fe import (ΔhflX Δfur) or inhibit Zn import (ΔhflX ΔznuC and ΔhflX ΔznuA). Consistently, supplemental Fe in the medium rescues the growth defect, whereas supplemental Zn or the iron chelator DFO affects the growth further (Fig. 4B, C, and D). Therefore, our results indicate that the suppressors may have evolved to overcome Fe starvation or excess Zn accumulation in the Mn-perturbed ΔhflX strain (Fig. 4A, B, C, and D). To support this assumption, we show that the cellular level of Zn is increased and the Fe level decreased due to uncontrolled influx of Mn in the ΔhflX strain (Fig. 4E). Intracellular Mn activates MntR, and Fe activates Fur to regulate homeostasis of the respective metal ions (10, 11, 14). Our qRT-PCR analysis indicates that apart from activating MntR, a high level of cellular Mn also activates the Fur regulon (Fig. 5). As a consequence, import of cellular Fe is hindered in the Mn-perturbed ΔhflX strain (Fig. 4E). Conversely, the presence of constitutive Fe import in the ΔhflX Δfur suppressor that reduces the cellular level of Mn (Fig. 6B) suggests that Fe also plays a role in the Mn homeostasis. Similar Mn-mediated activation of Fur- and Fe-mediated activation of MntR was also observed earlier (10, 35, 36). Our data also suggest that the MntR-Fe2+ complex is able to suppress mntH expression (see Fig. S2D in the supplemental material). However, the MntR-Fe2+ complex, unlike MntR-Mn2+, fails to activate mntP expression (see Fig. S2E).
In contrast to such an inverse relationship between intracellular levels of Mn and Fe, the relationship between Mn and Zn appears to be direct. Our experiments reveal that Mn import in the Mn-perturbed ΔhflX cells leads to Zn accumulation (Fig. 4E). On the other hand, inactivation of Zn import in the ΔhflX ΔznuC and ΔhflX ΔznuA suppressors also inhibits Mn accumulation (Fig. 6B). Since Mn-mediated derepression of the Zur regulator to upregulate Zn-transport genes was proposed earlier (30, 34), we initially assumed that an upregulation and downregulation of the Zn import and export genes, respectively, could explain the direct relationship between cellular Mn and Zn transport. However, a very minor change has been recorded in the expression profiles of Zn transport genes in the Mn-perturbed ΔhflX strain (Fig. 5). This discrepancy has been addressed with the argument that an increasing concentration of Zn (Fig. 4E) can compete with Mn for binding to the transcription regulators (Zur and ZntR). Such a competition may quench the Mn-mediated derepression of Zn transport genes. Consistent with this, the ΔhflX ΔznuC suppressor, a Zn-transport-defective mutant, exhibits 5-fold upregulation of znuA and 2-fold downregulation of zntA, under Mn stress (see Fig. S2F and G in the supplemental material).
When in excess, Mn is toxic in E. coli as it may inactivate enzymes that use other divalent metal ions. Mn is known to replace Fe from the cellular proteins (6, 7, 39). The functions of many Fe-containing proteins (e.g., cytochromes, dehydrogenases, and iron-sulfur proteins), which are involved in diverse cellular processes, may also be affected due to replacement of Fe by a very high level of Mn in the ΔhflX cells under Mn stress. In addition, an inadequate supply of Fe in the Mn-perturbed ΔhflX cells can directly affect proper metallation and biogenesis of these Fe-containing proteins. Therefore, evolution of the ΔhflX Δfur suppressor or Fe supplementation (Fig. 4C) may play an important role in suppressing the ΔhflX phenotype by restoring the function of the Fe-containing proteins. On the other hand, the intracellular free Zn level is at femtomolar ranges (less than 1 free Zn ion per cell), while the protein-bound level of Zn is much higher (40) (Fig. 4E), implying that free Zn could be toxic. Indeed, excess intracellular Zn is reported to have a damaging impact on the bacterial cell envelope (41). Thus, HflX-mediated Mn homeostasis plays an important role in protecting Fe and Zn homeostasis in order to maintain healthy growth and propagation of E. coli cells under Mn stress.
An intricate functional link between the hflA locus and HflB activity.E. coli hflX is a part of a complexly regulated “superoperon” that encodes several important cellular proteins, which are activated under multiple stress conditions in order to take part in DNA repair, RNA metabolism, cell wall biogenesis, tRNA modulation, and proteolysis (42). The hflA locus, which constitutes the hflX, hflK, and hflC genes, is the most downstream segment of this superoperon. Although HflK and HflC are known to act in concert with Zn-dependent HflB metalloprotease to maintain the housekeeping function of the cell and to control the λ phage life cycle (33), the relationship between HflB and HflX was poorly understood. Earlier we failed to establish a connection of HflX with the HflB-HflKC protease complex (16). In the present study, we demonstrate that the intracellular Zn level can play a role in the lysis-lysogeny decision making process of bacteriophage λ (Fig. 4F and G). We propose that the high level of intracellular Zn accumulation in the Mn-perturbed ΔhflX cells may activate HflB, which leads to the low frequency of lysogenization of bacteriophage λ (Fig. 4F).
MntH-independent alternative route for Mn import is activated in the absence of HflX.Deletion of the gene coding for the sole Mn importer, mntH, can only partially overcome the Mn toxicity in the ΔhflX cells, and thus, an alternative route for Mn import, which is primarily activated in the absence of the HflX, is predicted (Fig. 3A). Strikingly, a large number of identified suppressors (9 out of 10) exhibit disruption in either znuA or znuC, which are defective for high-affinity Zn import, in E. coli. Apart from reducing Zn import, the ΔhflX ΔznuA and ΔhflX ΔznuC suppressors also reduce the cellular Mn level in order to rescue the cell from Mn sensitivity (Fig. 6B). These two observations provoked us to consider whether HflX regulates the activity of the high-affinity ZnuABC importer to block Mn transport (Fig. 7A) and whether the latter acts promiscuously in the absence of HflX to import both Mn and Zn (Fig. 7B). If so, then the ZnuABC transporter is the alternative channel for Mn import, which results in elevation of the level of Mn and Zn ions in the ΔhflX cells (Fig. 7B). We hypothesize that Mn-mediated phosphorylation of HflX could determine the specificity of ZnuABC for Zn in the Mn-enriched environment (Fig. 7). Particularly, ZnuA, a member of the solute binding proteins (SBPs), acquires Zn in the periplasm and initiates its transport (43). Although the biochemical analysis suggests little possibility of Mn binding to ZnuA (43), the overall structural and sequence similarity of the latter to other Mn-binding SBPs make it difficult to assign the exact mechanism of metal specificity determined in these proteins (44, 45). A structural reorganization of ZnuA may take place when perturbed by Mn. The function of HflX may restrict this structural modulation and restore the Zn specificity of the transporter under Mn perturbation. When the constitutive Fe transport in the ΔhflX Δfur suppressor possibly blocks ZnuABC-mediated Mn import by an unknown mechanism, the evolution of the ΔhflX ΔznuA and ΔhflX ΔznuC suppressors indicates the presence of a direct mechanism to block Mn as well as Zn transport (see Fig. S3 in the supplemental material). However, we cannot rule out the possibility of any other alternative mechanism of Mn accumulation in ΔhflX E. coli cells unless this hypothesis is proven convincingly.
Schematic to explain HflX-mediated Mn homeostasis (A) Model showing metal importers and their regulatory mechanisms in the Mn-perturbed WT strain. Fe transporters are shown in general. ZupT, an alternative Zn transporter that is expressed constitutively in E. coli, is also shown. Based on our experimental findings, the ZnuABC transporter is assumed to be a possible alternative channel for Mn import. Mn-mediated phosphorylation of HflX probably starts a signal transduction to block noncanonical influx of Mn via the ZnuABC transporter and thus determines the Zn specificity of this importer. (B) Due to the lack of HflX-mediated signaling and regulation, the Mn-perturbed ΔhflX strain exhibits an uncontrolled influx of Mn via the ZnuABC transporter. Mn influx also stimulates Zn influx by an unknown mechanism and represses Fe transport by activating Fur. Mn also heavily activates the MntR regulator to repress MntH-mediated import of Mn. A fraction of cellular Zur regulators is shown to be inactivated by Mn.
ACKNOWLEDGMENTS
D.D. is grateful to Pradip Sen and Beena Krishnan for reading the manuscript.
Funding for this work is supported by CSIR-IMTECH. D.D. is the recipient of a Ramanujan Fellowship funded by DST, India.
FOOTNOTES
- Received 31 March 2014.
- Accepted 26 April 2014.
- Accepted manuscript posted online 2 May 2014.
- Address correspondence to Dipak Dutta, dutta{at}imtech.res.in.
S.S. and V.K. contributed equally to this article.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.01717-14.
REFERENCES
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
