The Tn5 insertions in RT3-27 or WR1-14 also cause copper sensitivity
Acid-sensitive mutants of S. meliloti display pleiotropic phenotypic defects, including sensitivities to certain heavy metals (Glenn et al., 1999). The sensitivities of WR1-14 and RT3-27 to heavy metals were therefore compared with their respective wild types. Addition of CuSO4 to TY plates (at pH 7.0) at concentrations above 0.4 mM prevented growth of RT3-27 or WR1-14, whereas WSM419 and WSM710 could tolerate 1 and 2 mM, respectively. The metal sensitivity is restricted to copper since the mutants displayed the same tolerances on tryptone-yeast media (TY) plates as their respective wild types to AgNO3 (25 μM), CdCl2 (200 μM), CoCl2 (500 μM), HgCl2 (10 μM), Ni(NO3)2 (50 μM) or ZnSO4 (500 μM).
The relatively low concentration of copper (2 μM) present in the minimal medium was sufficient to terminate growth of the actP-defective mutants only at low pH; neither R. leguminosarum WR1-14 nor S. meliloti RT3-27 could grow at pH 5.5 or pH 5.7, respectively, unless copper was omitted from a minimal medium prepared with high-purity reagents. The latter finding is a contrast to the observations that acid tolerance cannot be restored to actA, actS or actR mutants by removing heavy metals (Glenn et al., 1999). These data indicate that the disruption of genes in RT3-27 and WR1-14 prevents the expression of a copper-export system. Here, we present an important insight into metal toxicity as the availability of a heavy metal may also change in response to pH-dependent binding.
Sequence and analysis
Full sequences for 5.1 and 3.2 kb DNA regions surrounding the Tn5 insertions in the mutants WR1-14 and RT3-27, respectively, showed that the disrupted genes code for cation-transporting ATPases; they have therefore been designated actP (acid-tolerance P-type ATPase). The Tn5 insertions are located 103 and 225 bp from the 3′ end of the actP open reading frame (ORF) in R. leguminosarum and S. meliloti respectively; the sequences immediately surrounding the insertions are not similar. These Tn5 insertions, which are in the C-terminal tenth of the ActP protein in both cases, apparently result in complete inactivation of the protein, possibly suggesting that (if the altered protein is actually produced) the C-terminal domain is functionally or structurally important. Several other genes were identified: hmrR occurred downstream to actP in both organisms while an additional two partial ORFs (treY and ecfR) were located in the R. leguminosarum strain sequence (Fig. 1A).
ActP is a P-type ATPase that belongs to the CPx family. The sizes of the ActP proteins deduced from R. leguminosarum and S. meliloti DNA sequences are 88 and 86 kDa respectively. The pI of S. meliloti ActP is 6.42 whereas that for R. leguminosarum is 5.72. Both proteins are very similar over their entire lengths: the two ActP proteins show 49% identity (and 65% similarity) over 664 amino acids.
A BLASTP search of the databases using the R. leguminosarum and S. meliloti ActP proteins revealed considerable similarity to putative cation-transporting P-type ATPases from numerous organisms. The matches to the R. leguminosarum ActP protein include a putative silver ion transporter (SilP; AF067954) from Salmonella typhimurium (53% identity over 808 amino acids), a putative copper ion transporter (PacS; D16747) from Synechococcus spp. (49% identity over 656 amino acids) and a heavy metal ion transporter (U42410) from Proteus mirabilis (47% identity over 671 amino acids). The matches to the S. meliloti ActP protein include the heavy metal ion-transporting P-type ATPase (U42410) from P. mirabilis (66% identity over 806 amino acids), a heavy metal ion-transporting ATPase (Z99121) from Bacillus subtilis (45% identity over 809 amino acids) and the copper ion-transporting ATPase (PacS; D16747) from Synechococcus spp. (47% identity over 756 amino acids). These proteins have been placed into the CPx subfamily of the P-type ATPases that recognize and transport heavy metals and are distinguished from other ATPases by the presence of the CP(C/H/S) motif, the P-residue generally located 43 amino acids upstream to an E1–E2 ATPase phosphorylation signature site (DKTGTLT) (Solioz and Vulpe, 1996). Both rhizobial ActP proteins contain the CPC motif, 41 amino acids distal to the phosphorylation site. This motif is located in the region considered to form an ion channel; members that transport a similar ion share more conserved residues in this region. Hydrophilicity profiles of both rhizobial ActP proteins calculated from the Kyte–Doolittle algorithm show a typical plot for CPx proteins.
A search for domains (using the programs SCANPROSITE and PROFILESCAN) in the ActP proteins has identified a heavy metal-associated (HMA) domain in the ActP protein of S. meliloti (IEGMTCASCVRRVEKAIAAVPGVASANVNL). This sequence matches the consensus found in other CPx-type ATPases involved with heavy metal transport [L, I, V or N]–X2–[L, I, V, M, F or A]–X–C–X–[S, T, A, G, C, D, N or H]–C–X3–[L, I, V, F or G]–X3–[L, I or V]–X9–11–[I, V or A]–X–[L, V, F, Y or S] (Bull and Cox, 1994; Lin and Culotta, 1995); residues underlined indicate the selections from the consensus domain sequence, while those not underlined are spacer residues. Such an HMA domain is absent from the R. leguminosarum ActP. The Escherichia coli HRA-1, HRA-2 and Enterococcus hirae CopB P-type ATPases also lack the HMA domain, but contain histidine- and methionine-rich leader sequences which Trenor et al. (1994) have suggested bind to heavy metals. The R. leguminosarum ActP sequence also contains a histidine-rich leader which may serve a similar function.
In addition, a prokaryotic lipoprotein lipid attachment site [D, E, R or K]–X6–[L, I, V, M, F, W, S, T, A or G]–X2–[L, I, V, M, F, Y, S, T, A, G, C or Q]–[A, G or S]–C– (C is the lipid attachment site; PROSITE database) found in membrane-associated proteins is present within the ActP amino acid sequences of R. leguminosarum (LAAVAVLIIAC) and S. meliloti (VNAVAVLIIAC), although the initial consensus residue is absent.
HmrR is a putative transcriptional regulator. Directly downstream from actP in R. leguminosarum is an ORF that encodes a protein of 14.7 kDa and a pI of 5.78, whereas in S. meliloti this protein is 16.3 kDa and has a pI of 9.53. The two proteins are 55% identical (and contain 74% similarity) over 129 amino acids. It is interesting that the exopoly-saccharide (EPS) biosynthetic regulator ExoR is also more acidic in R. leguminosarum (pI 5.5) than in its S. meliloti counterpart (pI 7.2) (Reeve et al., 1997). This ORF downstream from actP in both species encodes a protein with highest similarity to transcriptional regulators of the MerR family; these proteins contain the MerR family helix–turn–helix DNA-binding signature in the N-terminal region (Helmann et al., 1989) (GEASKVSGVSSKMIRYYEQIGLI and GEASERSGLPSKTIRYYEDIGLI in S. meliloti and R. leguminosarum respectively; residues underlined are invariant across the proteins YBBI, YBBI and PMTR from E. coli, Haemophilus influenzae and P. mirabilis respectively).
The greatest identity (47% over 126 amino acids for R. leguminosarum and 60% over 128 amino acids for S. meliloti) occurs in the merR family with the heavy metal-dependent transcriptional regulator PMTR of P. mirabilis (Noll et al., 1998). The gene directly downstream to actP in the rhizobia has therefore been named hmrR (heavy metal-response regulator).
Function of TreY and EcfR. Upstream of actP in R. leguminosarum is a partial ORF (treY) that encodes a protein 34% identical over 346 amino acids to a malto-oligosyl trehalose synthase from Brevibacterium helvolum (accession number AAB95368), Arthrobacter sp. (accession number BAA09667) and Rhizobium sp. (accession number BAA11186). Malto-oligosyl trehalose synthase catalyses the conversion of maltodextrins to malto-oligosyl trehaloses through a transglucosylation reaction (Maruta et al., 1996). The match with TreY occurs over the entire length of the partially sequenced R. leguminosarum protein.
Downstream from hmrR in R. leguminosarum is an ORF for a protein with strong similarity to sigma factors of the extracytoplasmic factor (ECF) subfamily. The start codon for this ORF (ecfR) in R. leguminosarum could be an ATG codon or an upstream GTG; both are potential initiation codons, but only the GTG codon is preceded by a ribosome binding site (RBS). EcfR shows the greatest identity to the sigma-E factors (sigma-24) of Mycobacterium tuberculosis (32% over 162 amino acids of SigH; CAB08314), Streptomyces coelicolor (32% over 161 amino acids of SigR; CAB09088), Mycobacterium smegmatis (31% over 163 amino acids of SigH; AAD41810) and Treponema pallidum (33% over 145 amino acids of RpoE; AAC65087). In addition, the BLASTP search revealed 36% and 34% identity to a stretch of 144 amino acids of SigE from M. smegmatis (AA45268) and M. avium (AAC45219).
Typical for ECF sigma factors, the EcfR protein from R. leguminosarum is shorter than other classes of sigma factors and it contains a fis-type (Kostrewa et al., 1992) helix–turn–helix DNA-binding motif EEQREALHLVAIEDLYSQEAAQ (residues underlined are invariant).
These sigma-24 proteins transcribe genes whose products are required for responses to oxidative stress in S. coelicolor (Paget et al., 1998) and both heat shock and oxidative stress in M. tuberculosis (Fernandes et al., 1999). Interestingly, SigE in M. smegmatis has been shown to promote survival against acidic pH in addition to high temperature, detergents and oxidative stress (Wu et al., 1997).
The phenotypic defects are not caused by a polar effect on hmrR expression
Since the hmrR gene is located downstream to actP in both organisms and is transcribed in the same direction, it became prudent to investigate whether the acid and copper sensitivity was caused by a Tn5 polar effect on hmrR expression.
Complementation of RT3-27. Ten individual clones were isolated from a DNA library of S. meliloti WSM419 that could complement both the acid- and copper-sensitive phenotype of RT3-27. Digestion of each complementing plasmid with EcoRI released a 4.1 kb fragment that was cut by HindIII into 2.5 and 1.6 kb fragments. The pattern observed corresponds to that obtained from the rhizobial DNA region in the clone pRT3-27 (Fig. 1C). A 4.8 kb HindIII fragment (in pNK52; Fig. 1) restores copper tolerance to RT3-27 but a 0.9 kb deletion between the EcoRV sites (in bold in Fig. 1) prevented complementation revealing that the actP ORF does indeed extend into the EcoRV fragment, as predicted by sequence analysis.
Disruption of hmrR does not prevent complementation by the plasmid pNK77 while the generation of a frame shift mutation in actP (actP SphI site inactivated; Fig. 1C) destroyed the ability of pNK57 to complement.
Mobilization of pNK52 (containing S. meliloti actP and hmrR) into WR1-14 restores growth of this mutant on JMM plates buffered at pH 5.5 (containing 2 μM CuSO4) and also on TY plates at pH 7.0 containing 0.75 mM CuSO4. Despite the difference in the proposed metal-binding domains for the ActP proteins, it is evident that ActP possessing the HRA domain may be substituted with ActP containing a HMA domain to serve the same function in R. leguminosarum.
Complementation of WR1-14. The plasmid pWR880 was obtained by forcing spontaneous precise excision of Tn5 from actP by selecting for copper-tolerant colonies of WR1-14 that originally contained pWR873. Using this strategy, precise excision (identified from the number of copper-tolerant colonies) could be detected approximately once in every 107 cells. All the copper-tolerant colonies had lost Tn5 from the plasmid, indicating that multiple copies of the wild-type actP gene provided a growth advantage to the cells on copper-containing plates over those cells that lost Tn5 from the single copy actP lesion already in WR1-14. Since the hmrR gene may be co-transcribed with actP in R. leguminosarum, we examined the complementation pattern of plasmids derived from pWR880. The plasmids pWR901 and pWR903 contained only a functional actP gene and were still able to restore the wild-type phenotype to WR1-14.
These observations indicate that the phenotypic defects in both organisms result directly from the loss of the ActP protein and not from a polar effect on the expression of the hmrR gene.
Expression of actP
An actP–gusA fusion present in the plasmid pWR668 was transcriptionally induced by copper in R. leguminosarum grown in MJMM-S (see Experimental procedures) containing glutamate as the nitrogen source (Fig. 2A). Transcription of actP increased further at low pH, paralleling the observed increase in copper sensitivity of the mutants at low pH. The expression appears to be specific to copper since no such response was observed after cell exposure to silver or zinc (data not shown). The lack of induction in response to silver indicates that the considerable similarity of ActP to SilS does not extend to function, in line with the lack of any increased sensitivity to silver for both actP mutants.
Figure 2. Transcriptional activation of a plasmid-borne R. leguminosarum actP–gusA fusion (pWR668) in WSM711 in response to copper.
A. MJMM-S containing 3 mM glutamate as the nitrogen source at pH 7.0 (filled circles) and pH 5.5 (filled squares).
B. MJMM-S containing 10 mM ammonium chloride as the nitrogen source at pH 7.0 (filled circles) and pH 5.5 (filled squares).
Download figure to PowerPoint
Transcription is enhanced at low pH presumably because greater protonation of glutamate at acid pH than at neutral pH results in a decrease in copper chel-ation and increased free copper concentrations. In a liquid minimal medium containing 3 mM glutamate and 10 μM copper, the free copper concentration has been calculated to change from 0.00021 to 0.052 μM with a pH shift from 7.0 to 5.5. For the same total concentration of copper in the medium, but with 10 mM NH4Cl replacing the glutamate, the range of free copper concentrations has been calculated to change from 0.47 to 8.7 μM for the same pH change (P. May, personal communication). In line with the theoretical calculations, cells were far more sensitive to added copper in a medium containing NH4Cl, where 10 μM copper prevented growth, than one containing glutamate, which allowed growth in the presence of 100 μM copper.
When glutamate was replaced by NH4Cl in the liquid minimal medium, expression of actP–gusA in R. leguminosarum (Fig. 2B) was still pH dependent. In these conditions, however, the fusion was activated in response to far lower concentrations of total copper (compare Fig. 2B and Fig. 2A).
HmrR transcriptionally regulates actP
To investigate the role of hmrR in the copper-responsive regulation of actP, we constructed mutant strains by exchanging the wild-type allele for hmrR:CAS-Gm in R. leguminosarum (Fig. 3). The targeted disruption of hmrR created strains that were mildly copper sensitive, the mutants showing no growth at pH 7.0 on TY plates containing 1.75 mM CuSO4 whereas the wild type could grow at 2 mM. The transcriptional activation of the actP–gusA fusion in response to copper was eliminated in the hmrR– background (Fig. 4). Interestingly, in this background the basal level of expression of the actP–gusA fusion increased at both neutral and acidic pH in the absence of copper ions. In contrast, the level of expression of a copA–lacZ fusion in an copR mutant of E. coli did not change the activity of β-galactosidase in cells cultured in non-inducing conditions, unlike the wild-type background (Outten et al., 2000; Petersen and Moller, 2000). Our research reveals that HmrR negatively regulates the transcription of actP in the absence of copper ions but positively in the presence of copper ions. The increase in transcription of actP in the HmrR mutant would explain why this mutant tolerates almost as much copper as the wild type. HmrR is therefore the second member of the MerR family that responds to copper ions and therefore does indeed fall into the CopR subgroup initially proposed by Petersen and Moller (2000).
Figure 3. Insertional inactivation of hmrR.
A. Transposon excision. Tn5 was excised from pWR873 as described in the Experimental procedures section to construct pWR880.
B. Cassette insertion and subcloning. A 1.3 kb blunted HindIII fragment containing the CAS-Gm cassette was subcloned from pWRTACT-G2 into the blunted SphI site of hmrR in two orientations. The NotI fragments containing the hmrR:CAS-Gm mutation were subcloned into pJQ2ΩS (filled arrows represent residual aacC1 gene) to form pWR1033 and pWR1037.
C. Marker exchange. The latter plasmids were mobilized into WSM710 and a double crossover forced by plating the mating mixture onto JMM minimal medium containing gentamicin and sucrose (5% w/v). HindIII-restricted genomic DNA from the five streptomycin and spectinomycin-sensitive mutants WR1880 (1), WR1881 (2), WR1882 (3), WR1884 (4), WR1885 (5) and WSM710 (6) was hybridized to DIG-labelled pWR114L–Hp/E to probe for hmrR (D) and DIG-labelled pJQ200SK to probe for the gentamicin resistance gene aacC1 (E). Disruption of hmrR with CAS-Gm should result in a 1.2 kb increase (evident in D and E, lanes 2–5) in size of the 4.0 kb wild-type HindIII fragment (C; and D, lane 6). The size of the 1.2 kb HindIII fragment (C; lower hybridizing band in D) should not be disrupted during the crossover process. The size of fragments of the molecular weight marker from the top are approximately 23, 9.4, 6.6, 5.1, 4.9, 4.4, 4.2, 3.5, 2.3, 2.0, 1.9, 1.5, 1.3, 0.9, 0.8 and 0.5 kb.
Download figure to PowerPoint
Figure 4. The actP gene is transcriptionally regulated by HmrR in R. leguminosarum in response to copper. β-Glucuronidase expression from the actP–gusA fusion located on the broad host range plasmid pWR668 was measured from the wild-type (PW711) and hmrR mutants (WR1880 and WR1885) in the absence (non-filled bars) and presence (filled bars) of 50 μM CuSO4.
Download figure to PowerPoint