|Home | About | Journals | Submit | Contact Us | Français|
The arsenic resistance gene cluster of Microbacterium sp. A33 contains a novel pair of genes (arsTX) encoding a thioredoxin system that are cotranscribed with an unusual arsRC2 fusion gene, ACR3, and arsC1 in an operon divergent from arsC3. The whole ars gene cluster is required to complement an Escherichia coli ars mutant. ArsRC2 negatively regulates the expression of the pentacistronic operon. ArsC1 and ArsC3 are related to thioredoxin-dependent arsenate reductases; however, ArsC3 lacks the two distal catalytic cysteine residues of this class of enzymes.
Arsenic is widely dispersed in the environment and occurs primarily in two oxidation states, arsenate [As(V)] and arsenite [As(III)], and both are toxic to the majority of living organisms. The frequent abundance of arsenic in all environmental compartments has guided the evolution of detoxification systems in almost all microorganisms. Of these, the arsenic resistance system (ars) appears to be widely distributed among prokaryotes. It involves an arsenate reductase (ArsC), an arsenite efflux pump (ArsB or ACR3), and a transcriptional repressor (ArsR) (32), encoded by a set of genes that display large variations in their number and genomic organization. The early identified ars system of Escherichia coli plasmid R773 (41) has two additional components, ArsA, which acts as the catalytic subunit of the ArsAB arsenite extrusion pump (33), and ArsD, a metallochaperone protein that transfers As(III) to ArsA (18). In addition to these well-studied ars components, a variety of ars clusters contain additional genes whose functions in arsenic resistance have not been clearly established in many cases (31).
Members of the Microbacterium lineage of actinobacteria that can tolerate various metals, including nickel, chromium, and uranium (1, 16, 25), have been isolated from metal-rich environments. New examples of arsenic-resistant isolates of Microbacterium are continuously being reported (1, 2, 8, 10, 12, 21). In each case, however, the tolerance mechanism was not investigated, probably due to the lack of efficient genetic systems in this genus. Among actinobacteria, only Streptomyces sp. FR-008 (40) and Corynebacterium glutamicum ATCC 13032 (28) have been subjected to molecular characterization of determinants of defense against arsenic. In the former, the linear plasmid pHZ227 carries an arsenic resistance gene cluster with two novel genes, the arsO and arsT genes, which encode a putative flavin-binding monooxygenase and a putative thioredoxin reductase, respectively (40). The latter strain was recently shown to possess two members of a new class of arsenate reductases (Cg_ArsC1 and Cg_ArsC2) (30) and a transcriptional repressor (Cg_ArsR1) with a metalloid binding site unrelated to other previously characterized members of the ArsR/SmtB metalloregulatory proteins (29).
The present study focuses on Microbacterium sp. strain A33, a soil isolate previously shown to tolerate high concentrations of arsenite and arsenate (2). Here, we report on the isolation and functional characterization of an original set of arsenic resistance genes from this strain. Distinctive attributes of the ars cluster include the placement of two genes encoding a thioredoxin (Trx) system within the cluster, one gene encoding a putative Trx-coupled arsenate reductase lacking two catalytic cysteines, and a fusion gene whose product is a multidomain protein consisting of a transcription factor and an arsenate reductase.
Microbacterium sp. strain A33 was assayed for tolerance toward arsenic oxyanions in TRIS low-phosphate (LP) medium (27). Overnight cultures were diluted 1:100 in fresh LP medium supplemented with increasing concentrations of arsenate [As(V)] or arsenite [As(III)] and incubated at 28°C for 24 h before measurement of absorbance at 600 nm (A600). The concentrations of As(V) and As(III) that caused 50% growth inhibition (IC50s) were 400 mM and 28 mM, respectively. Cells were still able to grow (5% of maximum growth; A600 = 0.15) at 800 mM As(V) and 50 mM As(III), even without preexposure to lower concentrations of the oxyanions (Fig. (Fig.1).1). Similar and higher levels of tolerance to arsenite have been reported only for a few microorganisms, including Ochrobactrum tritici (6), Ferroplasma acidarmanus (5), an arsenite oxidizing bacterium identified as Microbacterium lacticum (24), and, recently, several Microbacterium isolates retrieved from deep-sea sediments (10). To our knowledge, however, the arsenate resistance level exhibited by Microbacterium sp. A33 appears to be the highest described so far.
To investigate the nature of the genetic determinants responsible for the extreme arsenic tolerance of Microbacterium sp. A33, a genomic library was constructed by ligation of partially digested Sau3A fragments (7 to 10 kb) in the BamHI site of plasmid pSRK21 (39) and was subsequently introduced into the arsenic-sensitive strains E. coli MA1 (Table (Table1)1) and Corynebacterium glutamicum ArsB1-B2 (28) by electroporation (11, 38). E. coli transformants were selected on LB plates supplemented with 4 mM As(V) or 2 mM As(III), while a concentration of 12 mM As(III) was used for C. glutamicum, which is intrinsically less sensitive to arsenic than E. coli (28). A total of 126 arsenic-resistant transformants (91 with E. coli and 35 with C. glutamicum) were obtained and further analyzed for their plasmid content. Restriction mapping revealed that they all had common regions. Furthermore, PCR amplification using the dacr1F and dacr1R primers (2) produced a single amplicon of 0.75 kb in each case, indicating that all clones possessed the targeted ACR3 gene. Low-stringency Southern hybridization experiments with Microbacterium sp. A33 total DNA using this PCR fragment as a probe confirmed that the strain contains only one copy of the ACR3 gene (data not shown).
The DNA region encompassing the ACR3 gene was sequenced from a subset of clones of the gene library conferring arsenate and arsenite resistance to both E. coli and C. glutamicum. Six putative genes were identified, five of which were designated arsC3, arsT, ACR3, arsRC2, and arsC1 on the basis of their homology to known ars genes, and the remaining gene was called arsX. The organization of this six-gene cluster is depicted in Fig. Fig.2A.2A. arsC3 is transcribed divergently from the five other genes which appear to form an operon (see the transcription analyses below). An extensive BLAST search against the NCBI nr database was performed for the six putative encoded proteins.
ArsC1 and ArsC3 were found to share sequence similarity to Trx-dependent arsenate reductases from high-G+C-content, Gram-positive bacterial groups, such as Streptomyces sp. FR-008 (40) (70% and 40% amino acid identity, respectively), and, to a lesser extent, to the archetypical ArsC protein from Staphylococcus aureus plasmid pI258 (37% and 28% amino acid identity, respectively). Both proteins possess the consensus N-terminal presumed CX5R oxyanion-binding motif of Trx-coupled arsenate reductases (22). However, only the ArsC1 protein has the two additional cysteines (Cys88 and Cys91) required for the catalytic activity of this family of arsenate reductases. In addition to arsC1 and arsC3, a third arsC-like sequence fused to an arsR-like sequence (arsRC2) that could encode a single polypeptide of 331 amino acids was identified. The C-terminal portion (approximately 140 amino acids) of the predicted ArsRC2 protein is related to Trx-linked arsenate reductases from actinomycetes, like the above-mentioned ArsC1 protein. The three putative ArsC proteins share rather low sequence similarity, the best level of identity (52%) being between ArsC1 and the C-terminal domain of ArsRC2 and the least (39%) between ArsC1 and ArsC3. The 120-amino-acid N-terminal portion of ArsRC2 shows clear homology to transcriptional regulators of the ArsR family and shares 53% sequence identity with its closest homologue, ArsR from Thermobifida fusca XY (YP_289615.1). This domain contains both a putative arsenite binding motif (ESCVCDL) nearly identical to that of E. coli R773 ArsR and an adjacent DNA binding helix-turn-helix motif (35). ACR3 is located immediately upstream of arsRC2. It encodes a putative arsenite transporter of the Acr3(1)p subfamily (2) showing significant homology with previously characterized ArsB1 from C. glutamicum (58.2% identity) and Acr3p from Streptomyces sp. FR-008 (66.3% identity) (40).
The remaining genes (arsT and arsX) are located proximally in the arsT-arsX-ACR3-arsRC2-arsC1 cluster. They were found to overlap by a 4-bp sequence involving fusion of the termination codon of arsT with initiation codon of arsX (ATGA), suggesting that they are translationally linked. An identical 4-bp overlap was also observed between arsX and ACR3. ArsT is homologous to thioredoxin reductases and mostly relates to those from actinobacterial groups such as Brevibacterium linens BL2 (65.7% identity to ZP_00377913) or the recently identified ArsT protein from Streptomyces sp. FR-008 (58.9% identity). ArsX is a member of the thioredoxin family. As with the ArsT protein, the closest homolog of ArsX is from B. linens BL2 (72% identity to ZP_00377914). However, BLAST searches revealed that these two putative thioredoxins are more distantly related to other actinobacterial thioredoxins than those from alphaproteobacteria, which showed the best hits (data not shown). To our knowledge, this is the first description of a bacterial ars cluster harboring genes for a putative thioredoxin system.
A BamHI-ScaI DNA fragment carrying the whole ars cluster was cloned in the pSRK21 vector to yield plasmid pF1. Deletion mutagenesis was used to elucidate the function of the ars genes that contributed to arsenic tolerance. For this purpose, plasmid pF1 was digested with restriction endonucleases as indicated in Fig. Fig.2B2B and Table Table1.1. The constructs, as well as pF1 and empty pSRK21 control plasmids, were inserted into E. coli AW3110, a strain lacking the chromosomal arsRBC operon (9), and the transformants were assayed for arsenic resistance in LP medium. Introduction of pF1 in cells of AW3110 led to a restoration of both arsenate and arsenite resistances (Fig. (Fig.2B).2B). Compared to what was observed for pF1, deletion of arsC1 to yield plasmid pF1ΔC1 completely abolished resistance to As(V) but not to As(III). Removal of arsC3 (pF1ΔC3) resulted in a similar phenotype; however, only partial loss of As(V) resistance was observed. These results support the hypothesis that ArsC1 and ArsC3 function as arsenate reductases, which are known to be required for As(V) detoxification. ArsC1 appears to be the major contributor of arsenate resistance in E. coli. Cells carrying plasmid pF1ΔC1RC2 (lacking arsC1 and arsRC2) exhibited the same levels of resistance to As(V) and As(III) as those containing pF1ΔC1; therefore, no clear function could be inferred from that construct for ArsRC2. Extending the deletion to include ACR3 (pF1ΔC1RC2ACR3) resulted in a total loss of tolerance toward As(III) and As(V). This is reminiscent of what is observed in most bacterial ars systems when arsenite transporter genes are inactivated, i.e., hypersensitivity to arsenite and arsenate due to the inability of cells to extrude As(III) subsequent to As(V) reduction by ArsC. Cells transformed with pF1ΔT (lacking arsT) partly lost resistance to arsenate, in agreement with the predicted function of ArsT as a thioredoxin reductase and its possible involvement in the electron transfer to arsenate reductase. However, the deletion also led to a decreased level of arsenite resistance, which may be attributed to a polar effect on downstream genes. To test this possibility, Cys137 and Cys140 residues within the highly conserved CXXC motif that characterizes the active site of thioredoxin-related proteins were replaced with serine residues by using a megaprimer-based mutagenesis method (37). The first PCR product obtained with the proC3/C137-140Santi primer pair (see the supplemental material) was used as a megaprimer, together with primer TR, in a second round of PCR. The resulting product was digested with Kpn2I to give a mutated fragment, which was inserted into the Kpn2I site of pF1ΔT. E. coli harboring this arsT-mutated plasmid (pF1mT) displayed the same reduction in arsenate resistance level as with pF1ΔT, while tolerance to arsenite remained unaffected (Fig. (Fig.2B).2B). This result provides substantial evidence that ArsT contributes to arsenate tolerance via a thioredoxin reductase activity that depends on the presence of the CXXC redox motif. Heterologous expression of the various ars constructs in the arsenic-sensitive strain C. glutamicum ArsB1-B2 produced the same phenotypes as those observed in AW3110, albeit with a less pronounced effect (Fig. (Fig.2B2B).
The ability of ArsRC2 to play a role in arsenate resistance in E. coli was tested with E. coli JW3470, a strain lacking arsC (3). The complete arsRC2 coding region was amplified from plasmid pF1 by using the ExpRC2F/ExpRC2R primer pair, digested with HindIII and XbaI, and ligated into plasmid pBAD18 (15) under the control of the PBAD promoter to create pBAD-arsRC2. Cells of JW3470 harboring pBAD-arsRC2 or pBAD18 exhibited the same low level of arsenate resistance whether or not they were cultured under inducing conditions, i.e., with added 0.2% arabinose. Considering that the fused gene product is actually produced in the pBAD-arsRC2 construct (see below), the inability to complement the arsC mutant suggests that ArsRC2 has no significant arsenate reductase activity in E. coli.
Reporter gene studies using arsT-gfp and arsC3-gfp fusions were performed to study the expression of the divergent ars gene cluster in E. coli. A fragment encompassing the arsT-arsC3 intergenic region and 66 and 102 bp downstream of the initiation codons of arsT and arsC3, respectively, was PCR amplified using primers ProT and ProC3. The PCR product was digested with EcoRI and ligated in both orientations into the promoter-probe vector pPROBE-NT (23) to generate plasmids pT-GFP and pC3-GFP (Table (Table1).1). The orientation of the insert in recombinant clones was established by PCR with primers ProT and ProC3 used in concert with vector-specific primer PGFP. These constructs were electroporated into E. coli DH10B, together with the compatible plasmid pBAD-arsRC2, which provided the putative regulator in trans. For induction assays, cultures of transformants were exposed or not exposed to arsenate and/or 0.2% arabinose before the normalized fluorescence level (relative level of green fluorescent protein [GFP]/optical density at 600 nm [OD600]) was measured with a FLX Xenius spectrofluorimeter (SAFAS, Monaco). In the absence of arabinose, the arsT-gfp fusion (pT-GFP) produced a high level of fluorescence independent of the concentration of added arsenate (Fig. (Fig.3).3). Expression was nearly abolished in the presence of arabinose alone and further increased in a dose-dependent manner upon arsenate addition. Arsenite and antimonite were found to elicit equivalent responses (data not shown). In contrast, low levels of fluorescence were measured for the arsC3-gfp fusion regardless of the presence or absence of arabinose or arsenate (Fig. (Fig.3).3). From these findings, we concluded that ArsRC2 serves as a transcriptional repressor of arsT but not arsC3 and that As(V), As(III), and Sb(III) are inducers. A 17-bp imperfect inverted repeat (TTGTATCGATAAGTGTC-N6-GACACATGTCGATTCAA) that may constitute an ArsRC2 binding site was found in front of arsT, overlapping the ATG initiation codon. Another inverted repeat sequence (TCCGGCGGGC-N5-GCCCGCCGGA) was present 15 nucleotides upstream of the putative GTG initiation codon of arsC3, suggesting the existence of a specific transcription regulator for this gene in Microbacterium sp. A33. Neither of these two dyad sequences resembles the ArsR binding sites identified so far in E. coli, Synechocystis, and C. glutamicum (20, 29, 35).
As mentioned above, the organization of the ars gene region suggested that arsT-arsX-ACR3-arsRC2-arsC1 might form an operon divergently transcribed from arsC3. Reverse transcription-PCR (RT-PCR) experiments were performed to test this possibility. Total RNA was isolated from 5-ml samples of Microbacterium sp. A33 cultures at the mid-exponential phase, grown for 4 h in LP medium containing 5 mM As(V) or no arsenic. RNA was extracted using Nucleospin RNAII (Macherey-Nagel Duren, Germany) with DNase treatment according to the kit's manual. First-strand cDNA was synthesized from 2 μg RNA with Moloney murine leukemia virus (MMLV) reverse transcriptase (Invitrogen) and random hexamers in accordance with the recommendations of the supplier. Synthesized cDNA was used as the template in subsequent PCR amplification reactions with primers designed to span the ars gene region or targeting the 16S rRNA gene. RT-PCR products of the expected sizes were obtained for primer pairs covering the region from arsT to arsC1, showing that these five genes form a single transcriptional unit (Fig. (Fig.4).4). An internal 80-bp arsC3 product was also obtained, indicating that this gene is indeed expressed in Microbacterium sp. A33.
To provide an individual assessment of expression of the six ars genes, we performed real-time RT-PCR experiments with the same cDNA as that described above. Transcriptional induction by As(III) and Sb(III) was not tested, since these two metalloids were shown (see above) to act as inducers of the operon comprising arsT to arsC1. Primers were designed using Primer3 software (http://frodo.wi.mit.edu/primer3/input.htm) to yield PCR products ~100 bp in length. Amplifications were carried out with a Bio-Rad iCycler iQ with 50-μl reaction mixtures containing 2 μl cDNA, 500 nM forward and reverse primers, and iQ SYBR green Supermix (Bio-Rad) according to the manufacturer's instructions and conducted under the following conditions: 4 min at 95°C, followed by 40 cycles of 15 s at 95°C and 45 s at 58°C. Each reaction was run in duplicate, and amplification efficiencies for each primer pair were determined from standard curves generated with serial dilutions of cDNA. Because the PCR efficiencies for all genes were close to each other (95 to 115%), relative mRNA expression levels were normalized for input RNA against the level of 16S rRNA gene transcripts and quantified by use of the comparative cycle threshold (2ΔΔCt) method (19). The exposure of Microbacterium sp. A33 cells to arsenate (5 mM) resulted in a 14-fold increase in arsC3 expression compared to the level for the control. The expression levels of both arsT and arsX were highly enhanced (98- and 60-fold, respectively) by arsenate treatment, while the induction levels of the downstream ACR3, arsRC2, and arsC1 genes were less pronounced (7.7-, 7.3-, and 5.3-fold, respectively). Similar polar expression effects have been reported for the ars operons from Synechocystis sp. PCC 6803 (20) and Desulfovibrio desulfuricans G20 (17) but not for the two ars operons from C. glutamicum (28).
The facts that arsT and arsX occupy the two first positions in the ars operon and that they show the most-dramatic transcriptional induction by arsenic are indicative of their importance in the response of Microbacterium A33 to arsenic stress. The presence of a putative thioredoxin reductase in an arsenic resistance gene cluster has been reported only for the linear plasmids pHZ227 and pREL1 in Streptomyces sp. FR-008 and Rhodococcus erythropolis PR4, respectively (34, 40). We examined the occurrence of similar gene association in annotated bacterial genome sequences available in the nonredundant databases in the Integrated Microbial Genomes (IMG) system (Joint Genome Institute, Walnut Grove, CA [http://img.jgi.doe.gov/cgi-bin/pub/main.cgi]). In addition to the two above-mentioned cases, arsT homologs were found in the vicinity of ars genes in Streptomyces coelicolor A3(2) (NP_630905), plasmid pBD2 from Rhodococcus erythropolis BD2 (NP_898753), Arthrobacter sp. FB24 (YP_829715), Arthrobacter aurescens TC1 (YP_949317) and its plasmid TC2 (YP_950223), and Brevibacterium linens BL2 (ZP_00377913) as well as in the two gammaproteobacterial strains Stenotrophomonas maltophilia R551-3 (ZP_01642200) and Acinetobacter baumannii AYE (YP_001715387). For the latter strain, there is some evidence that the ars genes are located in a genomic resistance island presumably acquired by horizontal gene transfer (13). The only case of collocation of thioredoxin carrying arsX homolog and ars genes was found in the genome of B. linens BL2 (ZP_00377914). Interestingly, all of these clusters were found to contain at least one arsC-like gene encoding a thioredoxin-linked arsenate reductase in the same orientation as arsT or arsX homologs (Fig. (Fig.5).5). Assuming that operons are composed of genes that cooperate in the same biological process, we hypothesize that coexpression of a thioredoxin system and arsenate reductase would enhance the efficiency of arsenate reduction and/or contribute to a better protection against arsenic-induced oxidative stress. Nevertheless, the evolutionary significance of the predominant occurrence of such an unusual genetic arrangement in actinobacteria awaits further clarification.
Another unique feature of the Microbacterium sp. A33 ars genes lies in arsC3. We showed that this gene, together with arsC1, could rescue the arsenate sensitivity phenotype of E. coli AW3110 (Fig. (Fig.2B)2B) and that its expression is inducible by arsenate in Microbacterium, suggesting that it encodes a functional arsenate reductase. However, ArsC3 lacks two of the three catalytic cysteine residues of Trx-dependent arsenate reductases and does not exhibit recognizable sequence similarity to ArsC from E. coli plasmid R773, the representative of the second major class of glutathione-linked arsenate reductases (22). It seems unlikely that ArsC3 belongs to this class, since glutathione is generally absent from the actinomycetes, which instead produce mycothiol (26). Whether ArsC3 is a member of the recently identified third class of mycothiol-dependent arsenate reductases (30) is also improbable. Indeed, ArsC3 exhibits only 32% amino acid identity with the two representative enzymes of this novel class, Cg_ArsC1 and Cg_ArsC2 from C. glutamicum, and seems to be functional in E. coli, which rules out its dependence on mycothiol. This raises the question of the precise function of the ArsC3 protein and the way it contributes to arsenate resistance.
The arsRC2 gene also constitutes an unusual feature of the ars cluster. Fusion of the ArsR and ArsC domains has been described only for Leptospirillum ferriphilum (36) and could therefore be considered a rare event. However, a CDART (14) search identified 73 proteins with the same domain architecture, essentially from alphaproteobacteria (37), high-G+C-content Gram-positive bacteria (23), and green nonsulfur bacteria (7) and 13 proteins with an inverse ArsCR architecture, exclusively from gammaproteobacteria. Such a wide distribution suggests that these fusions are nonrandom events and presumably confer a selective advantage to their host (42). Although most ars operons are inducible by arsenate and arsenite, the ArsR repressor has been shown to control the expression in response to As(III), so that As(V) has to be reduced in vivo to induce the ars system. We therefore speculate that the physical coupling of functional ArsC and ArsR domains in a fusion protein might alleviate the problem of diffusion of As(III) to the inducer binding site and increase the efficiency of transcription in response to As(V). In this situation, the ArsC domain would primarily serve to provide As(III) to the repressor rather than to the ACR3 efflux protein, which would not necessarily require a high level of arsenate reductase activity. This may explain both the inability of arsRC2 to complement an arsC mutation in E. coli JW3470 and the presence of the additional arsC1 gene, which appears to play the major role in the detoxication of arsenate.
In summary, the ars system of Microbacterium sp. A33 exhibits remarkable features, including (i) the presence of genes required to provide electrons for the reduction of arsenate in a single operon rather than being dispersed in the genome, (ii) an ArsC3 arsenate reductase which contributes to arsenate resistance while lacking typical catalytic cysteine residues of related Trx-dependent enzymes, and (iii) an unusual ArsRC2 fusion protein which acts as an arsenic-dependent transcriptional repressor. Because the ars cluster has been found to confer arsenic tolerance to heterologous hosts, it is most probably functional in its natural host. However, its actual contribution to the exceptional arsenic tolerance of Microbacterium sp. A33 still remains to be determined. The development of effective genetic systems for Microbacterium would greatly facilitate addressing these questions.
The sequence of the ars gene cluster from Microbacterium sp. A33 has been deposited in the EMBL database under accession number AM283030.
This work was supported by GISFI (Groupement d'Intérêt Scientifique sur les Friches Industrielles) and ZAM (Zone Atelier Moselle). A. Achour-Rokbani gratefully acknowledges the support of the Ministère de l'Enseignement Supérieur, de la Recherche Scientifique et de la Technologie Tunisien and of Région Lorraine (France).
We thank Francis Arhin and Mandeep Dadhwal for critical reading of the manuscript.
Published ahead of print on 4 December 2009.
†Supplemental material for this article may be found at http://aem.asm.org/.