Sequence and Analysis of a Plasmid-Encoded Mercury Resistance Operon from Mycobacterium marinum Identifies MerH, a New Mercuric Ion Transporter

ABSTRACT In this study, we report the DNA sequence and biological analysis of a mycobacterial mercury resistance operon encoding a novel Hg2+ transporter. MerH was found to transport mercuric ions in Escherichia coli via a pair of essential cysteine residues but only when coexpressed with the mercuric reductase.

Narrow-spectrum mercury resistance (i.e., resistance to mercury salts) requires transport of mercuric ions across the membrane by specific transporters, typically MerT and MerP (10), cytoplasmic reduction to elemental mercury Hg(0) by mercuric reductase (MR) and passive diffusion of the volatile metal out of the cell (2). In broad-spectrum mercury resistance, the organomercurial lyase MerB catalyzes the breakdown of carbonmercury bonds and, with the transport and reductase proteins, confers resistance to a wide range of organomercurials (12). Both mechanisms are subject to tight regulation by a mercuryresponsive transcriptional regulator, MerR (5).
Although mercury resistance has been extensively studied and characterized in different bacterial genera, there are few reports of mycobacterial mercury resistance genes, and no mer operons have been described in mycobacteria. Most studies consist of the isolation and identification of mercury-resistant mycobacterial strains in which the mercuric reductase gene has been detected by PCR and MR activity has been shown by volatilization of radioactive mercury (11,21). Some strains were also positive for amplification of the merB gene. Mercury resistance was also demonstrated in mycobacteria expressing mer genes from Tn501, and mycobacterial expression vectors have been developed with mercury resistance as the only selectable marker (3).
Mycobacterium marinum, which causes a tuberculosis-like disease in fish and skin infections in humans, is closely related to Mycobacterium tuberculosis and is a good model for mycobacterial infection studies (6). The genome of a clinical isolate of M. marinum (strain ATCC BAA-535) (14) carrying a plasmid encoding potential mercury resistance genes was recently published (22). Here, we report the DNA sequence and analysis of the first Mycobacterium mercury resistance operon. We characterized a previously unknown membrane protein (MerH) from M. marinum, which we found transports mercuric ions across the inner membrane of Escherichia coli via a pair of cysteine residues located in the first transmembrane (TM) region but only when coexpressed with MR. In Mycobacterium smegmatis, MerH and MR confer resistance to mercuric chloride, which may be a useful, naturally occurring, nonantibiotic selection marker in mycobacteria.
A 3.6-kb mer operon confers mercury resistance to M. marinum strain ATCC BAA-535. M. marinum strain ATCC BAA-535 (GenBank accession number NC_010612) contains a 23-kb plasmid (pMM23; GenBank accession number NC_010604) which harbors a 3.6-kb mer operon ( Fig. 1) (22) containing five genes likely to be involved in mercury resistance ( Table 1). The operon is predicted to contain genes encoding functions associated with regulation, transport, and reduction of mercuric ions and organomercurial breakdown, but the gene arrangement is unusual, with two of the genes being divergently transcribed from the remaining three (Fig. 1). These mer genes are surrounded by three other genes that may not participate directly in mercury resistance-a predicted glutathione reductase and two recombinases that may promote mobility of the genetic element.
M. marinum strain ATCC BAA-535 was found to be three to four times more resistant to mercuric chloride than was M. smegmatis mc 2 155 in two different growth media (see Table 5), and similar levels of mercury resistance were observed for M. marinum and E. coli expressing mer genes from Tn501 on LB agar, which suggests that mercury resistance genes located on the pMM23 plasmid from M. marinum are functional.
One gene from the mer operon in pMM23.11 has no significant similarity to any protein in the nonredundant database but was predicted to be a membrane protein with four putative transmembrane helices. We hypothesized that this gene, designated merH, was involved in mercuric ion transport (Fig. 2B).
Characterization of the merH gene product in E. coli. When expressed in E. coli in place of MerT within a minimal mercury resistance operon, MerH could be recovered in the membrane protein fraction, indicating that it is inserted in the membrane (data not shown). Five fragments of merH were PCR amplified using primers listed in Table 2 and cloned using NcoI/BstBII in pYZ-TBL (4), making ␤-lactamase fusions at five different locations in MerH. (Fig. 2B). ␤-Lactamase activity of the resulting hybrid proteins was measured in E. coli C43 cells (cul-tured, washed, and diluted in 0.85% saline solution before plating onto L agar containing increasing concentrations of carbenicillin). ␤-Lactamase confers high-level resistance to ampicillin or carbenicillin only when located in the periplasmic compartment (9). As indicated by the carbenicillin MICs in Fig. 2B, ␤-lactamase protein fusions at residues alanine 54, leucine 67, and serine 115 of MerH resulted in a carbenicillinresistant phenotype for E. coli cells. This suggests that loops between TM regions 1 to 2 and 3 to 4 of MerH are located in the periplasm. In contrast, fusions at residues histidine 83 and valine 99 led to inactive ␤-lactamase and a carbenicillin-sensitive phenotype in E. coli. These data are compatible with the SOSUI model for MerH topology, which predicts both Nterminal and C-terminal regions to be located in the cytoplasm.
MerH was expressed in E. coli by replacing merT with merH within a cloned minimal Tn501 mer operon containing the merR, merT, merP, and merA genes in plasmid pBRmerT (Table 3). When merT was replaced by merH in plasmid pBRmerH, the mercuric chloride MIC for E. coli cells decreased from 111 to 81 M but remained much higher than that for control cells expressing no mercuric ion transporters (MIC of 15 M). This result suggests that MerH acts as a mercuric ion transporter in E. coli.
In vivo 203 Hg volatilization assays were performed as described elsewhere (17) in E. coli TG2 strains expressing either MerT, MerC, MerF, or MerH as the sole mercuric ion transporter within a minimal Tn501 mer operon (Fig. 3A). Mercury volatilization observed in the absence of a mercuric ion transporter was probably the result of cell permeabilization and was subtracted from all samples. MerT conferred the highest rate of mercury volatilization in E. coli, while MerH conferred intermediate rates, between the volatilization rates achieved with and without MerT, that were equivalent to the mercury volatilization rates achieved with MerF and MerC, mercuric ion transport proteins from Tn5053 and Tn21, respectively (8,24). These data show that MerH is able to import mercuric ions into the E. coli cytoplasm for reduction by MR.
The following five combinations of cysteine-to-serine mutations were created in MerH using a two-step PCR method: C38S, C39S, C169S, C38S/C39S, and C38S/C39S/C169S (20) ( Table 2). These were confirmed by sequencing both DNA strands. Mutants were cloned in place of merT in the pBRmerT plasmid and expressed from the P merT promoter in E. coli TG2 upon induction with 0.4 M HgCl 2 . Expression and membrane localization of the cysteine mutants were confirmed by 35 S labeling of plasmid-encoded proteins (data not shown).
The effects of the cysteine mutations on mercuric ion resistance and transport were tested by HgCl 2 MICs (Table 4) and 203 Hg 2ϩ volatilization assays (Fig. 3B) using E. coli TG2 cells expressing the different MerH cysteine mutants. The cysteine residues in MerH did not play an equal role in mercury resistance. Cysteine 169 did not participate in mercuric ion transport, since its mutation to serine did not decrease mercury resistance or mercuric ion transport activity compared to wildtype MerH. Mutation of cysteine 38 decreased mercury resistance by approximately 25%, whereas mutation of cysteine 39 completely abolished resistance. Both were predicted to be in the first TM region of MerH. Cells expressing the MerH double (C38S/C39S) or triple (C38S/C39S/C169S) mutants were as  sensitive to mercuric chloride as were cells expressing no mercuric ion transport proteins (Table 4). Similarly, mutation of cysteine 38 significantly decreased Hg volatilization, and mutation of cysteine 39 completely abolished volatilization (Fig. 3B). As expected, the double and triple cysteine mutants of MerH were not able to transport mercuric ions.
MerH cannot transport Hg 2؉ across the cytoplasmic membrane of E. coli without MR. Cells expressing mercuric ion transporters in the absence of MR were expected to show a  hypersensitive phenotype to mercuric chloride, since they specifically import mercuric ions (21). Plasmids expressing different mercuric ion transporters (MerT, MerC, MerF, and MerH) without expression of MR were constructed and transformed into E. coli TG2 cells (Table 3) to test the HgCl 2 sensitivity of resulting strains (Fig. 3C). Expression of MerT or MerC in the absence of MR resulted in an HgCl 2hypersensitive phenotype in TG2. Their sensitivity to mercuric chloride increased by approximately fourfold compared to that of cells expressing no mercury resistance proteins or no transporters. In the case of MerF, a 1.5-fold increase in HgCl 2 sensitivity was observed. However, for MerH, this hypersensitive phenotype was not observed. Cells expressing MerH unexpectedly showed a slight increase in mercury resistance (1.2-fold increase) compared to cells expressing no mer genes. These data suggest that MerH transports mercuric ions across the inner membrane of E. coli cells only when MR is present in the cytoplasm. Apart from the predicted length and number of TM regions, the major difference between MerH and other mercuric ion transporters is that MerH lacks a second pair of cysteine residues predicted to be located in the cytoplasm (Fig. 2A). In MerT, MerC, and MerF, this pair of cysteine residues has been shown to be involved in mercury transport, since their mutation resulted in a reduced rate of mercury volatilization (13,24). These data suggest that MerH is not able to import mercuric ions to the E. coli cytoplasm because it lacks the cysteine pair in this compartment. In order to achieve the transport of Hg 2ϩ into the cytoplasm, MerH may require MR cysteine residues to act as acceptors for mercuric ions, or MR may be required to cause a productive conformational change in MerH. It is also possible that these cytoplasmic cysteines are required for Hg 2ϩ transport in E. coli and not in the natural host, M. marinum. merA and merH from M. marinum confer mercury resistance on M. smegmatis. The merH and merA genes were separately and jointly cloned in pVV16 (an E. coli-Mycobacterium shuttle vector) and constitutively expressed in M. smegmatis mc 2 155 (Table 3). Transformants were selected on 25 g ml Ϫ1 kanamycin and 50 g ml Ϫ1 hygromycin. After 4 days of incubation at 37°C, single colonies of mc 2 155 expressing different mer genes were observed under the microscope. Expression of merA slightly affected the mc 2 155 colony morphology but did not affect the growth rate. Expression of merA with merH resulted in a significant change in the morphology of the colonies (which were much smoother) but did not affect the growth rate. However, expression of merH from the constitutive hsp60 promoter of pVV16 was toxic to the cells. Ringshaped colonies appeared on L agar plates, and cells could not be grown in L broth. The effects of merA and merAH expression upon mercury resistance in M. smegmatis were determined by MIC assays. Levels of mercury resistance of mc 2 155 strains expressing merA or merA and merH from the hsp60 promoter in the pVV16 vector were determined on L agar with antibiotic selection or 7H11 agar without antibiotic selection pressure   Table 5). Expression of merA in M. smegmatis resulted in a fourfold increase in mercury resistance, while coexpression of merA and merH in resulted in a 10-fold increase. These data show that the merA gene product is sufficient to confer some mercury resistance on M. smegmatis but that expression of the putative Hg 2ϩ transport gene (merH) is required for maximal mercury resistance. On 7H11 agar plates, M. smegmatis also showed maximal mercury resistance upon coexpression of merA and merH but with much greater MICs, suggesting that the kanamycin selection of pVV16 strongly decreased mercury resistance. However, expression of merAH in mc 2 155 conferred sufficient mercury resistance (Ͼ30 M) to allow selection of recombinant M. smegmatis cells on HgCl 2 , indicating that these two genes can be used as a non-antibiotic-selectable marker in mycobacteria.
We are grateful to Delon Barfuss of Georgia State University, who provided us with 1 mCi of 203 HgCl 2 , and to Jon Hobman for discussion and advice. This work was supported by a studentship awarded to M.S. from the Darwin Trust of Edinburgh, United Kingdom. G.S.B. acknowledges support from James Bardrick in the form of a Personal Chair, from the Lister Institute as a former Jenner Research Fellow, from the Medical Research Council (United Kingdom), and from the Wellcome Trust.