High-level chromate resistance in Arthrobactersp. strain FB24 requires previously uncharacterized accessory genes

Identification of a chromate resistance determinant (CRD) in Arthrobactersp. strain FB24

Arthrobacter sp. strain FB24 genome analysis deduced a 450 amino acid (aa) sequence Arth_4248 with similarity to chromate ion transporters. Phylogenetic analysis of the sequence with 512 other characterized and putative ChrA sequences (see Figure 1 and Additional files 1 and 2) suggests that it forms a new branch in the CHR superfamily [22] that is composed of Actinobacteria. This group likely has unique evolutionary features since the majority (70%) of ChrA ortholog sequences used in the comparison is from Proteobacteria yet it formed its own branch. In fact, most of the clades are composed of specific phyla/classes of biota (Additional file 1).

The genome neighborhood of Arth_4248 consists of a 10.6-kb region of five putative chromate resistance genes and three proximal genes of unknown function located on a 96-kb plasmid (Figure 2). Of five genes similar to ones associated with Cr(VI) resistance in other organisms, two encode ChrA efflux protein orthologs (Arth_4248 and 4251) and three are similar to different regions of a putative regulatory protein, ChrB (Arth_4249, 4253 and 4254). The remaining three genes (Arth_4247, 4252 and 4255) have not been previously shown to be associated with chromate resistance. The region between Arth_4251 and Arth_4249 is an approximate 1.3 kb region of low complexity. Currently, there is no strong indication of functional genes within this region.

The chromate resistance determinant in Arthrobacter sp. strain FB24 has a similar genetic arrangement to that found in chromate-resistant Arthrobacter sp. CHR15, but is markedly different than in the two well-studied Proteobacteria, P. aeruginosa and C. metallidurans (Figure 2). More recently, a transposable element conferring chromate resistance in Ochrobactrum tritic was found to have a similar genetic makeup to the chr1 determinant in C. metallidurans [17], while a chromate resistance operon containing chrA, chrB and chrC was found in Shewanella sp. strain ANA-3 [16]. Additional genes involved in chromate resistance in C. metallidurans, such as the superoxide dismutase gene chrC, chrI and rpoH [21] are not present within the CRD of strain FB24. This could point to functional and regulatory differences in chromate resistance between these distantly related taxa. Thus, we were led to investigate Arth_4247, 4252 and 4255, as well as previously characterized chrA and chrB sequences. Due to the potential involvement of Arth_4247, 4252 and 4255 in chromate resistance, we have named these genes chrL, chrK and chrJ, respectively (Figures 2 and 3).

Sequence analysis of the CRD

Arth_4248, the putative 450 amino acid (aa) chromate ion transporter, is most similar to ChrA from Rhodococcus sp. RHA1 (79%). The protein is predicted to have 12 transmembrane helices and two CHR domains defined by a conserved GGX₁₂VX₄WX₁₆PGPX₁₀GX₇G motif, placing it within the LCHR family of chromate ion transporters [22, 24]. However, there is little sequence similarity (35% similarity across 106 of 225 amino acids) between the amino and carboxy halves of Arth_4248; hence, it does not appear to have arisen by direct tandem duplication of the same CHR domain-containing open reading frame. Because of the topological diversity of the CHR superfamily proteins [22, 25] and the observed preponderance of conserved residues in the N-terminal half of the P. aeruginosa ChrA protein [26], it is expected that the amino and carboxy termini may carry out different functional roles in chromate efflux. Alignment of the amino acid sequence of Arth_4248 with that of P. aeruginosa ChrA indicated that residues which resulted in Cr(VI) sensitivity following mutation in P. aeruginosa [26] are also conserved in Arth_4248.

The other chrA ortholog Arth_4251 is predicted to be 137-aa protein with sequence similarity to known ChrA transporters. The protein sequence aligns to the N-terminus of C. metallidurans ChrA1 with 71% similarity across 49 amino acids. In comparison, Arth_4251 is only 52% similar to the N-terminus of Arth_4248 across 44 amino acids. The CHR domain in Arth_4251 contains the GGX₁₂VX₄W motif, but lacks the PGPX₁₀GX₇G motif. In addition, no definitive transmembrane helices were predicted for Arth_4251. Small (<200 aa) proteins containing a single CHR domain have been recognized as a separate group of proteins within the CHR family. Recently, Bacillus subtilis SCHR orthologs ywrA and ywrB were shown to confer chromate resistance in E. coli; however, both genes were required for the resistant phenotype [27]. Genes encoding SCHR proteins are usually present as pairs within a genome [22]. In FB24, though, there does not appear to be a partner SCHR gene for Arth_4251 and the aa sequence is more closely related to CHR domains from LCHR proteins than to those of the SCHR family [23].

Three open reading frames (ORFs) designated Arth_4253, Arth_4254 and Arth_4249 in the putative CRD region share sequence similarity to the C. metallidurans ChrB proteins. Arth_4253, which encodes a 171 aa protein, aligns with the N-terminal portion of the C. metallidurans ChrB1 protein [GenPept: YP_582012] (42% similarity across 133 aa). Arth_4254 is a predicted 143 aa protein that exhibits 53% similarity across 132 aa of the C-terminal portion of the C. metallidurans ChrB1 protein. Together, Arth_4253 and Arth_4254 appear to encode the complete sequence for a full-length ChrB gene, but the gene sequences overlap by 4 nucleotides and a potential Shine-Dalgarno sequence is present upstream of the predicted start codon of Arth_4254. Repeated sequencing of this region did not reveal any potential sequencing errors that could explain this observation. RT-PCR analysis revealed that Arth_4253 and Arth_4254 can form a dicistronic mRNA (operon structure analysis provided in Additional file 3). Arth_4249 contains 430 nucleotides, but does not yield any hits to known genes at the nucleotide level. A BLASTx search of the translated nucleotide sequence versus the protein database shows that the predicted amino acid sequence is 76% similar to Arth_4254 across 77 aa.

Arth_4252 encodes a 344 aa protein containing a 40-residue YVTN family beta-propeller repeat and a WD40 repeat domain (with 81% sequence similarity to ORF18 in Arthrobacter sp. strain CHR15) with an N-terminal signal sequence. The function of Arth_4252 is presently unknown, but other proteins within the WD40 repeat domain family are associated with the regulation of signal transduction and sensing membrane stress [28, 29]. Arth_4252 also shares 62% sequence similarity to Rmet_6194, which is located approximately 4 kb downstream of the C. metallidurans chrA1 gene, Rmet_6202. However, a functional role for Rmet_6194 in chromate resistance in this organism has not been established. Orthologs of Arth_4252 were also found in close proximity to chrA genes in Arthrobacter sp. strain CHR15 and several species of Burkholderia as revealed by a gene ortholog neighborhood search in the Integrated Microbial Genomes database http://img.jgi.doe.gov.

Arth_4247 has an expected protein sequence of 337 aa with a putative overlapping signal sequence and transmembrane helix at the N-terminus, which suggests that it is a membrane-anchored protein. The protein sequence shares 75% aa similarity with lipoproteins of the LppY/LpqO family, which were first described in Mycobacterium tuberculosis but have not been functionally characterized. Other mycobacterial lipoproteins have been shown to perform such diverse roles as binding solutes in ABC transporter complexes, sensing environmental stressors and participating in signal transduction mechanisms [30]. M. tuberculosis, like strain FB24, is a high GC% Gram positive bacterium of the order Actinomycetales. The role of lipoproteins in the response to Cr(VI) has not been established in other organisms. Other lipoproteins have been shown to participate in the response to divalent metals such as copper and lead [31, 32]. In the case of copper, NlpE stimulated the CpxAR envelope stress response pathway in copper-exposed E. coli cells [32]; however, it is not known if an analogous function exists for LppY/LpqO proteins. As in the case for Arth_4252, orthologs of Arth_4247 are also present near chrA orthologs in Arthrobacter sp. strain CHR15 (81% similarity to ORF 27) and C. metallidurans (52% similarity to Rmet_6195).

Arth_4255 encodes a putative malate:quinone oxidoreductase of 517 aa with 77% similarity to Arthrobacter aurescens TC1 Mqo. This class of proteins generally functions in energy production, but the biochemical role of Arth_4255 in the context of Cr(VI) resistance is not known. In Agrobacterium tumefaciens, insertional inactivation of an operon specifying NADH:quinone oxidoreductases similar to malate:quinone oxidoreductases (MrpA, MrpC and MrpD) resulted in the loss of arsenite oxidation. The phenotype was recovered via complementation with the intact Mrp operon [33]. In other bacteria, NADH-dependent oxidoreductases have been shown to reduce Cr(VI) [34]; however, there is no conclusive evidence of Cr(VI) reduction in FB24, and it is unlikely that Arth_4255 is a Cr(VI) reductase.

Loss of plasmid DNA from strain FB24 results in metal sensitivity and increased intracellular chromium accumulation

A chromate-sensitive mutant (D11) was obtained after successive culturing of FB24 for 90 generations in the absence of chromate. Loss of plasmid DNA was assessed by Southern hybridization using a 10.6-kb probe for the CRD, and the results were validated by a PCR screen using gene-specific primers (data not shown). Strain D11 was hypersensitive to low levels (0.5 mM), whereas the wild type grew prolifically on 0.1X nutrient agar (NA) plates amended with 5 mM chromate. Strain D11 was also very sensitive to lead, zinc and cadmium. Jerke et al (2008) had shown that FB24 contained 3 plasmids, each with genes that confer resistance to lead, zinc and cadmium [35]. Whereas FB24 attained maximal cell densities in 200 μM lead, zinc and cadmium in mXBM, growth of strain D11 was strongly inhibited by 10 μM lead, 50 μM zinc and 1 μM cadmium (data not shown).

Total intracellular chromium content was measured in chromate-exposed cells of FB24 and D11 to determine if the loss of chromate resistance in strain D11 correlated with increased intracellular accumulation of chromium. There was a significant difference (p = 0.015) in chromium content between strain D11 (2.8 × 10^-7 mol mg protein^-1) and FB24 (9.2 × 10^-8 mol mg protein^-1). Chromium was undetectable in FB24 and D11 cells that were not exposed to chromate. Similar decreases in chromium accumulation were found between chromate-resistant and -sensitive strains of P. aeruginosa and C. metallidurans which contain ChrA efflux pumps [15, 36]. The comparable change in chromium accumulation between resistant and sensitive strains of Arthrobacter sp. FB24 and the two organisms in which the operation of an efflux pump has been biochemically demonstrated supports the hypothesis that plasmid-encoded Cr(VI) resistance in strain FB24 results from the function of a ChrA-like efflux pump.

Complementation of strain D11 with CRD

To localize the essential determinants for chromate resistance within the CRD, a series of plasmids were designed and tested for their capacity to confer chromate resistance in the chromate-sensitive strain D11 (Figure 3). Only D11 transformed with pKH12 (the complete 10.6 kb region) was able to grow comparably to FB24 on 0.1X (NA) plates containing 5 mM chromate (Figure 3). The other transformants, in which regions of the CRD were deleted, were able to grow only at lower levels of chromate (0.5 to 2 mM). In particular, chrA produced a resistance level of 0.5 mM Cr(VI) regardless of the presence of chrB-Nterm and chrB-Cterm.

Expression of chromate resistance genes in strain FB24 under chromate stress

Potential regulatory element within the CRD

ChrB has been proposed to function as an activator of the chromate resistance determinants in C. metallidurans [21]. A bioinformatics analysis using protein function prediction software [37] suggested possible DNA-binding and kinase activities for ChrB-Cterm and ChrB-Nterm, respectively. In addition, proteins containing WD40 repeats, such as Arth_4252, have been associated with signal transduction and regulatory mechanisms [29, 38]. To determine if chrK, chrB-Nterm and chrB-Cterm influence expression of chrA, strain D11 bearing plasmids pKH22 and pKH32 was grown in the presence and absence of chromate, and qRT-PCR was used to quantify chrA expression under these conditions. Expression of chrA was induced to higher levels by chromate in strain D11 bearing pKH22 than when the putative regulatory genes were absent (pKH32) (Figure 4). This difference is not likely to be attributable to differences in plasmid copy number provided that chrA expression in both strains without chromate was similar.

Bacterial strains and growth conditions

Induction of Cr(VI) resistance genes was assessed in Arthrobacter sp. strain FB24 cells by culturing in 150 ml NB to early mid-log phase (OD₆₀₀, 0.3) at 30°C with shaking at 200 rpm. Cells were harvested by centrifugation, washed once with 0.2X NB and suspended in 15 ml 0.2X NB. Chromate (K₂CrO₄) was added to final concentrations of 0, 0.005, 0.025, 0.05, 0.1, 5, 20 or 100 mM. To test for specificity of induction, additional cultures were incubated in the presence of 0, 0.5, 5 and 50 μM PbNO₃ in mXBM; 0, 0.5, 5 and 50 mM Na₂HAsO₄·7 H₂O in 0.2X NB; and 0, 0.5, 5, 50 mM hydrogen peroxide (H₂O₂) in 0.2X NB. Cells were incubated for 2.5 hours at 30°C with agitation. Induction experiments with Cr(VI)-sensitive strain D11 transformed with pKH22, pKH23 and pKH24 were carried out in the same manner with the following exceptions: kanamycin was added to a concentration of 30 μg ml^-1 and chromate was added to one culture at a concentration of 0.025 mM.

Generation of chromate-sensitive FB24 derivative

The lead- and chromate-sensitive mutant, D11, was generated from the resistant wild-type strain FB24 by growing cells in LB without chromate. Cultures were transferred daily by diluting cells 1:1000 into fresh media. Transfers were maintained for approximately 90 generations at 30°C with shaking at 200 rpm and then screened for cells sensitive to 75 μM lead on mXBM agar plates. Lead-sensitive colonies were then tested for Cr(VI) sensitivity on 0.1X nutrient agar (NA) plates supplemented with 0.5, 1, 2 and 5 mM K₂CrO₄. Loss of plasmid DNA in strain D11 was assessed by Southern hybridization and rep-PCR. Loss of the CRD genes was confirmed by PCR using gene-specific primers.

Chromate resistance determination

Starter cultures were grown overnight at 30°C in 0.2X NB with appropriate selection. Cultures for minimal inhibitory concentration (MIC) determination were diluted 1:1000 in 3 ml of 0.1X NB for chromate cultures or mXBM plus glucose for divalent cationic metals in borosilicate glass tubes and maintained at 30°C with shaking at 200 rpm. The OD₆₀₀ was measured daily for a period of 3 days until growth stabilized. Divalent cationic metals used in MIC experiments were added as lead nitrate (Pb(NO₃)₂, zinc chloride (ZnCl₂), or cadmium sulfate (CdSO₄) at concentrations ranging from 0 to 200 μM. Cultures were prepared in triplicate for each growth or MIC experiment. D11 transformants were screened for chromate resistance by streaking single colonies onto 0.1X nutrient agar plates containing 0, 0.5, 1, 2, or 5 mM chromate.

Sequence analysis of putative chromate efflux gene

The genome sequence is available in the GenBank database under accession numbers NC_008537 to NC_008539 and NC_008541. The genome was sequenced by the Department of Energy Joint Genome Initiative (DOE-JGI) and can be accessed at http://genome.jgi-psf.org/finished_microbes/art_f/art_f.home.html. The annotated sequence at this site was used for locating the CRD and construction of primer sequences. Multiple sequence alignment of Arth_4248 (ChrA) with other described and putative members of the CHR family of chromate efflux proteins [24] was performed using the ClustalX program and default setting for Gonnet series for protein weight matrix [51] and bootstrap Neighbor Joining tree options with 1000 resamplings. Output trees were visualized in Fig Tree http://tree.bio.ed.ac.uk/software/figtree/. Sequences were retrieved from the UniProt database [52] by conducting a search for ChrA sequences according to Diaz-Perez et al [22]. Some additional eukaryotic sequences were found by conducting a similar search of the GenBank database [53]. All short ChrA (SCHR) sequences (<350 amino acids) were excluded from the alignment. A total of 513 sequences (Additional files 1 and 2) were retrieved and aligned. Transmembrane helices were predicted using the TMHMM 2.0 server [54].

Cloning of chromate resistance determinant

A series of constructs were created to test expression of chromate resistance in strain D11. All constructs, except for pKH62 and pKH72, were prepared by subcloning into pBluescript SK+ (Stragene, La Jolla, CA) prior to cloning into pART2 [55]. Recombinant plasmid DNA was transformed into strain D11 by electroporation as described elsewhere [56]. Ampicillin was used for selection at a concentration of 100 μg ml^-1 for pBluescript-derived transformants, and kanamycin was used at a concentration of 40 μg ml^-1 for pART2-derived transformants. Plasmids were submitted to the Purdue University Core Genomics Center for validation of insert sequences.

Plasmid pKH11 was generated by amplifying a 10.6 kb fragment bearing bases 72880 to 83464 of pFB24-104 using the TripleMaster PCR system (Eppendorf North America, Inc., Westbury, NY) according to the manufacturer's specifications and primers C42/F and C42/R. The PCR product was digested with HindIII and XbaI and ligated into pBluescript SK+ to give pKH11. Plasmid pKH21 contains a 7.3 kb insert bearing bases 74642 to 81771 from FB24-104; the insert was isolated by digesting pAOWA10128 (obtained from DOE-JGI) with XbaI and HindIII. The remaining constructs (Table 3) were generated by restriction digestion of either pKH11 or pKH21 using standard cloning procedures [50].

Expression analysis by quantitative reverse transcriptase PCR (qRT-PCR)

Primer sequences for qRT-PCR are listed in Table 4. Total RNA was extracted from Arthrobacter cell pellets using the FastRNA PRO Blue Kit (MP Biomedical, Solon, OH) and treated with Turbo DNA-Free DNAse (Ambion, Austin, TX) to remove contaminating DNA. RNA concentrations were quantified by measuring the A₂₆₀ on a Smart Spec 3000 spectrophotometer (Bio-Rad, Hercules, CA). cDNA was synthesized from 100 ng total RNA using ImProm II reverse transcriptase (RT) (Promega, Madison, WI) following the manufacturer's reaction conditions. PCR was performed using the following conditions: 98°C for 5 min, followed by 30 cycles of 94°C for 30 s, 56-58°C (depending on the primer pair) for 30 s, 72°C for 1 min, with a final extension step at 72°C for 10 min.

For real-time PCR, 1 μl of the reverse transcription reaction mixtures prepared as described above was used as the template. The PCR mixture contained 1 U of HotMaster Taq (Eppendorf North America, Inc., Westbury, NY), 1× HotMaster Taq PCR buffer with 25 mM MgCl₂, 1% bovine serum albumin, 0.2 mM each of dNTPs, 0.25 mM each of a forward and reverse primer, SYBR Green (1:30,000; Molecular Probes, Eugene, OR) and 10 nM FITC (Sigma, St. Louis, MO) in a final volume of 25 μl. Reactions were carried out using a Bio-Rad MyIQ single-color real time PCR detection system, and data were analyzed using the MyIQ Optical System software version 2.0. Transcript copy numbers were calculated from a standard curve using known concentrations of pKH11. Plots of transcript copy number ng^-1 total RNA versus chromate concentration were prepared in GraphPad Prism 4 (GraphPad Software, San Diego, CA). The mean and standard errors were determined from 6 qRT-PCR reactions per chromate treatment (3 independent cultures × 2 reactions per culture). Significant differences among chromate treatments for each gene were determined by generating least square means in PROC GLIMMIX with the LS MEANS option in SAS version 9.1. Multiple comparisons were adjusted using Tukey's test. To normalize the variance of the model residuals, a negative binomial distribution was used for each set of gene expression data.

Chromium content in chromate-exposed cells

Arthrobacter strains FB24 and D11 were grown to mid-log phase (OD₆₀₀, ~0.2) in 50 ml 0.2X NB at which time four replicate cultures were amended with 2 mM chromate (final concentration). One culture per strain was incubated without chromate. All cultures were incubated for an additional 2 h. Aliquots of 40 ml of cells were harvested by centrifugation and washed 4 times with ddH₂O. Cell pellets were solubilized in concentrated nitric acid (cHNO₃) and heated at 95°C for 2 h. Samples were adjusted to a final concentration of 2% HNO₃ with double distilled water and analyzed for total chromium content at the Purdue University Mass Spectrometry Center. The ⁵²Cr inductively coupled argon plasma mass spectrometry (ICPMS) results were obtained using an ELEMENT-2 (ThermoFinnigan, Bremen, Germany) mass spectrometer in the medium resolution mode. The samples were introduced into the plasma using an Aridus desolvating system with a T1H nebulizer (Cetac Technologies, Omaha NE), which is used to enhance sensitivity and reduce oxide and hydride interferences. The argon sweep gas and nitrogen of the Aridus is adjusted for maximum peak height and stability using ⁷Li, ¹¹⁵In and ²³⁸Upeaks obtained from a multi-element standard (1 ng/ml, Merck & Co.). Chromium concentration was normalized per mg protein. Total soluble cell protein concentration was determined using the Lowry method [57] after collecting cells by centrifugation and extracting protein with 1N NaOH at 100°C. Student's t-test was used to determine statistically significant differences in the average chromium content between strains D11 and FB24 at the 95% confidence level.