|Home | About | Journals | Submit | Contact Us | Français|
The growth properties of a new panel of Bacillus subtilis tetL deletion strains and of a derivative set of strains in which tetL is restored to the chromosome support earlier indications that deletion of tetL results in a range of phenotypes that are unrelated to tetracycline resistance. These phenotypes were not reversed by restoration of a tetL gene to its native locus and were hypothesized to result from secondary mutations that arise when multifunctional tetL is deleted. Such genetic changes would temper the alkali sensitivity and Na+ sensitivity that accompany loss of the monovalent cation/proton activity of TetL. Microarray comparisons of the transcriptomes of wild-type B. subtilis, a tetL deletion strain, and its tetL-restored derivative showed that 37 up-regulated genes and 13 down-regulated genes in the deletion strain did not change back to wild-type expression patterns after tetL was returned to the chromosome. Up-regulation of the citM gene, which encodes a divalent metal ion-coupled citrate transporter, was shown to account for the Co2+-sensitive phenotype of tetL mutants. The changes in expression of citM and genes encoding other ion-coupled solute transporters appear to be adaptive to loss of TetL functions in alkali and Na+ tolerance, because they reduce Na+-coupled solute uptake and enhance solute uptake that is coupled to H+ entry.
The tetL gene of Bacillus subtilis, originally identified in a screen of chromosomal mutations that conferred elevated resistance to tetracycline (Tc) (55), was also identified in a screen of transposon insertion mutants that were sensitive to growth in the presence of alkaline pH and elevated Na+ (7). Studies on the functions of tetL have since implicated this gene in several key areas of cell physiology. It has been shown that TetL plays important roles in alkaline pH homeostasis, Na+ resistance, and acquisition of K+, in addition to its role in Tc resistance (8, 35, 36, 51). The basis for the multiple physiological roles are the multiple catalytic capacities of the TetL antiporter. TetL supports antibiotic resistance by an efflux mechanism in which a Tc-divalent metal complex is exchanged for a greater number of external protons, i.e., electrogenic [Tc-Me]+/H+ antiport (20, 21, 39); neither Tc nor the divalent cation can serve as an efflux substrate without the other (9, 20, 39, 56). The pH homeostasis and Na+ resistance functions of TetL depend upon its additional capacity for Na+ (K+)/H+ antiport (8) and K+ acquisition modes. TetL-mediated K+ uptake occurs when K+ substitutes for some of the external H+ coupling ions in a competitive manner that depends upon their relative concentrations in the medium and their electrochemical potentials (21, 29).
The studies that established the physiological roles for TetL included the construction of strains that were deleted for the tetL locus, the properties of which indicated that TetL is a major Na+ (K+)/H+ antiporter required for pH homeostasis (8). Deletion of tetL results in strains that show varied phenotypes. One type is exemplified by JC112, a strain that is defective for growth at both neutral and alkaline pH and in the presence of limiting K+ and that is also defective in alkaline pH homeostasis and Na+ extrusion (8). These properties are consistent with the role of TetL as a major Na+/H+ antiporter. JC112 also grows poorly in the presence of Co2+, a property that was suggested to relate to up-regulation of other membrane proteins in response to the absence of TetL (51). Another type of tetL deletion mutant, which is the more prevalent isolate, is represented by JC112C, a strain that shows a more moderate phenotype than JC112 with respect to the above-mentioned properties (51).
A selected group of genes, predicted to encode various types of antiporters, were screened for increased expression in JC112 and JC112C. Two antiporters that showed elevated expression levels were nhaC, a Na+/H+ antiporter gene, and czcD-czcO, a divalent metal resistance locus. The increase in czcD-czcO mRNA in Northern analyses was especially pronounced when Co2+ was added to the medium, and neither this increase nor the increase in basal expression was altered by the presence of a wild-type tetL gene in the amyE locus (51). These studies led to the hypothesis that the ability of B. subtilis to compensate for deletion of tetL requires secondary-site mutations. The current studies were undertaken to test this hypothesis and to better understand the response of B. subtilis to the loss of tetL. For this purpose, a mutant with a substitution of the tetL coding sequence was constructed, the tetL gene was restored at the native locus in several deletion mutants, and the transcriptomes of a tetL deletion strain and its derivative tetL-restored strain were assessed for changes in gene expression.
The B. subtilis parent strain was BD99, which is hisA1 thr-5 trpC2. The preparation and transformation of B. subtilis competent cell cultures were as described previously (13). Transformed strains were selected on solid Luria-Bertani medium with 80 mM K+ replacing Na+ (LBK) (18), using 4 μg/ml chloramphenicol (Cm), 100 μg/ml spectinomycin (Sp), 3 μg/ml phleomycin (Pm), or 5 μg/ml erythromycin (Em). Escherichia coli strains DH5α (19) and XL1Blue (Stratagene) were the hosts for plasmid constructions.
B. subtilis strains were grown at 30°C. TTM medium was used for assays of Tc resistance to minimize competitive inhibition by the high potassium ion content of other media (29). TTM consists of 100 mM Tris-HCl buffer (pH 7.0 or 8.3); 1 mM potassium phosphate; 0.01% MgSO4; 0.2% (NH4)2SO4; 50 μg (each) of l-threonine, l-histidine, and l-tryptophan per ml; 0.1% yeast extract; and 50 mM Tris malate. The medium was adjusted to pH 7.0 for assays of Tc resistance. For TKM medium, potassium malate replaced the Tris malate. Pm resistance was assayed in LBK medium at pH 7.5. For all other experiments, including isolation of RNA for microarray analyses, cells were grown in Spiz-KM medium, consisting of 44 mM KH2PO4, 80 mM K2HPO4, 15 mM (NH4)2SO4, 3.4 mM sodium citrate, 50 mM potassium malate, 0.1 mM MgSO4, 0.01% yeast extract, and 50 μg/ml each of histidine, threonine, and tryptophan. Where specified, potassium citrate replaced sodium citrate in this medium. The pH of the medium is indicated for individual experiments.
To replace the tetL leader peptide coding sequence (CDS) and tetL CDS with a Cm resistance gene (cat) CDS, plasmid pYW112 was used. The complete cat CDS from plasmid pC194 (26) was amplified by PCR as an NdeI-DraI fragment; the NdeI restriction site (CATATG) contained the ATG start codon, and the DraI restriction site (TTTAAA) contained the TAA stop codon. Two other PCR fragments were prepared; one was a 1,140-bp fragment containing sequences upstream of tetL on a BstBI-NdeI fragment, and the other was a 570-bp fragment containing sequences downstream of tetL on a DraI-HindIII fragment. The three PCR fragments were ligated into the vector plasmid pYH56 (10), between ClaI and HindIII sites, giving plasmid pYW112. Plasmid pYW112 was used to transform BD99, with selection for Cm resistance. Recombination by double crossover resulted in a clean replacement of the leader peptide CDS and TetL CDS with a cat CDS, whose expression is controlled by the tetL promoter and transcription terminator and the tetL leader peptide translational signals. This and all other plasmid constructs were confirmed by DNA sequencing.
To restore tetL in the deletion mutants, tetL-deleted strains were transformed with genomic DNA from BG384, a B. subtilis thr-5 trpC2 strain that has a Sp resistance cassette inserted in the MunI site upstream of tetL, as shown in Fig. Fig.11 (1).
To restore tetL to its original chromosomal site without introducing an antibiotic cassette near the tetL locus, an intermediate plasmid, named pYW124, was constructed. A 4.5-kilobase chromosomal HindIII fragment containing the tetL gene (7) was amplified in two parts, with the internal primers providing a BamHI site by an A-to-T change 180 bp downstream of the tetL stop codon. The two fragments were cloned into pGEM-9Zf(+) (Promega), yielding plasmid pYW124. The HindIII insert was then transferred to the temperature-sensitive plasmid pG+host4 (37), to give plasmid pYW124G. pYW124G was used to transform BD99, JC112, or JC210, with selection for Em resistance at 30°C. Colonies with integration of pYW124G were selected on Em plates at 42°C. Integration at the tetL locus was confirmed by PCR. Colonies with integrated pYW124G were grown at 30°C without selection and were spread for isolated colonies. Colonies that had undergone a double crossover to restore tetL were screened by Em sensitivity and, in the case of JC112, Tc resistance, and the desired construct was confirmed by BamHI digestion of the PCR-amplified tetL locus. Southern blot analysis was done to confirm that the tetL locus was restored and that there was no amplification of the tetL gene.
For citM knockout strain construction, plasmid pYW119 was used. This plasmid contained the following three PCR amplification products: a 510-bp BamHI-XbaI fragment that contained sequences from upstream of the citM gene to bp 45 of the citM CDS, an 850-bp XbaI-KpnI fragment that contained the Pm resistance gene of plasmid pUB110 (45), and a 570-bp KpnI-EcoRI fragment that contained the last 60 bp of the citM CDS followed by downstream sequence. The three PCR products were ligated together between the BamHI and EcoRI sites of pGEM3-Zf(+).
B. subtilis strains used in the study are listed in Table Table11.
RNA isolation and Northern blot analysis were done exactly as described previously (11), except that the cells were grown in Spiz-KM at pH 7.0. Uniformly labeled, antisense riboprobes were synthesized by T7 RNA polymerase transcription, in the presence of [α-32P]UTP, of gel-purified DNA fragments that were generated by PCR amplification of the various coding sequences. The upstream primer in the PCRs contained the 17-nucleotide T7 RNA polymerase promoter sequence at its 5′ end. Probing of 5S RNA for quantitation of Northern blots was done using a 5′-end-labeled oligonucleotide.
The microarray experiments were done as described previously (11). Briefly, total RNA was extracted by hot-phenol extraction, using a buffer containing 50 mM sodium acetate and 1 mM EDTA (pH 5.1). cDNA was synthesized in the presence of cyanine 3-dCTP (Cy3) and cyanine 5-dCTP (Cy5) dyes (Perkin-Elmer Life Sciences). The Cy3-cDNA (from the reference strain BD99 or JC112) and Cy5-cDNA (from JC112 or JC112-R) probes were concentrated and mixed. The mixture was hybridized to B. subtilis microarray slides (Eurogentec) overnight at 42°C in an Amersham Lucidea Slidepro hybridization station. The microarray slides contained whole open reading frame sequences for 4,096 of the 4,106 B. subtilis genes, and each B. subtilis gene was represented twice on the microarray. After washing, the slides were scanned at excitation wavelengths of 550 nm (Cy3) and 640 nm (Cy5) in a ScanArray ExpressHT scanner. Fluorescence was measured at 570 nm (Cy3) and 670 nm (Cy5). The fluorescence intensity of microarray spots was quantified and analyzed using the ScanArray Express program. A normalized Cy5/Cy3 mean fluorescence ratio was used to quantify changes in gene expression. Genes with a Cy5/Cy3 mean fluorescence ratio of >3 were considered significantly up-regulated, while genes with a mean fluorescence ratio of <0.33 were considered significantly down-regulated.
tetL deletion mutants JC112 and JC112C were constructed previously by transforming wild-type strain BD99 with a plasmid that had a Cm resistance cassette replacing sequences from the ClaI site located 272 bp upstream of the tetL CDS to an NdeI site located 159 bp downstream of the tetL CDS (Fig. (Fig.1).1). Thus, there was some concern that the phenotypes observed with JC112 and JC112C may be due, at least in part, to effects of the deleted tetL locus on expression of neighboring genes. We observed previously that there is a significant amount of readthrough past the tetL transcription terminator (8). Therefore, a plasmid that had a precise substitution of only the tetL leader peptide and CDS was constructed, using the cat CDS from the same element that was used to knock out tetL previously (Fig. (Fig.1).1). Control experiments demonstrated that the same Cm cassette in the amyE locus did not produce detectable phenotypes other than Cm resistance (strain BD99-05; data not shown). The plasmid was used to transform BD99. Twenty-seven Cm-resistant, Tc-sensitive clones were obtained by selection on LBK plates, as opposed to the more defined Spiz-KM used in the isolation of the original set of tetL deletion mutants. Genomic DNA from the new clones was analyzed by PCR to confirm the presence of the coding sequence substitution and by Southern blot analysis to eliminate clones that might have amplifications of the tetL locus, as has been observed previously by us and others (1, 28). The clones were grouped into categories based on 15-hour growth in 0.2 μg/ml Tc in semidefined TKM medium (see Materials and Methods). Nine clones were chosen for further analysis, and these were grouped into categories based on 15-hour growth in 1.4 M NaCl in TKM medium (pH 7.0) and 8-hour growth in 400 μM CoCl2 in Spiz-KM medium (data not shown). Of these, three clones were chosen (JC203, JC205, and JC206) for the following analyses: MICs for growth in Tc in TTM medium, growth in the presence of inhibitory levels of NaCl in TKM medium at pH 7.0 and 8.3, and growth in Spiz-KM medium in the presence of 300 μM CoCl2 (Table (Table2).2). The three clones showed a range of phenotypes, similar to the range found previously when the greater tetL locus was deleted (51). For example, JC205 was the most sensitive to alkali conditions but the least sensitive to the presence of Co2+. JC206, on the other hand, was the least sensitive to alkali but the most sensitive to Na+. None of the three tetL deletion mutants had the same phenotypes as the original JC112 and JC112C. These results demonstrated that the absence of TetL protein alone was responsible for defects in growth at high pH and elevated monovalent and divalent cation concentrations. We postulate that loss of tetL results in selective pressure that leads to second-site mutations that allow better growth of tetL-deleted strains.
If deletion of tetL was indeed accompanied by second-site mutations, it was predicted that restoration of tetL to a deleted strain would not necessarily complement all of the mutant phenotypes. To examine more precisely the effect of restoring tetL, derivatives of the three clones mentioned above, as well as of JC112 and JC112C, were constructed in which the deleted tetL locus was restored to wild type. The tetL-deleted strains were transformed with chromosomal DNA from a strain that had the wild-type tetL locus and a Sp resistance gene inserted at a MunI site located 1,804 bp upstream of the tetL CDS (Fig. (Fig.1).1). This site is in the coding sequence of a gene named yyaM, which is annotated in the SubtiList database as having an unknown function and being similar to other B. subtilis proteins of unknown function; a control wild-type strain with this insertion (JC210; Table Table2)2) was phenotypically similar to the wild type with respect to Tc resistance, Na+ sensitivity, and Co2+ sensitivity. The presence of wild-type tetL in the strains transformed to Sp resistance was confirmed by PCR analysis. Revertants were designated with the addition of an “R” to the strain number.
The data in Table Table22 show that the presence of the wild-type tetL restored Tc resistance, as previously observed (51). On the other hand, wild-type growth yields under other conditions were not restored and, in some cases, adding back the tetL gene actually resulted in lower levels of growth, e.g., growth of JC206R in the presence of elevated sodium, and growth of JC203R in the presence of Co2+. For strains JC112 and JC112C, there were no significant effects of tetL restoration on growth in the various conditions.
Similar conclusions were drawn from a set of strains in which tetL was restored without introducing an antibiotic resistance cassette in the tetL locus. A marked copy of the tetL gene containing a single-base-pair change located 180 bp downstream of the tetL CDS, which created a BamHI site, was introduced into the chromosome of JC112 (see Materials and Methods). This version of the tetL-restored JC112, which, unlike JC112-R, did not contain an inserted Sp resistance cassette, also did not show restoration of the tetL deletion phenotypes (data not shown).
The remaining studies focused on JC112, the best characterized of the tetL deletion mutants.
Microarray experiments were performed to assess global changes in gene expression in JC112 and the effect of restoring tetL in JC112-R. mRNA levels in these two strains were compared to those in BD99. The data in Tables Tables33 and and44 are averages from duplicate samples in two independent experiments (i.e., averages of four measurements). Shown are changes that were more than threefold, which was considered significant. Five genes, which encoded products that could be related to the alkali, Na+, or Co2+ sensitivity of the deletion strain, were selected for Northern blot analysis, to confirm the microarray findings (Fig. (Fig.22).
The data for 37 genes that had a positive fold change (i.e., overexpression) (Table (Table3)3) can be summarized as follows. Genes in the rocG and rocD operons, required for arginine metabolism (15), were overexpressed in JC112. Northern blot analysis of the rocD operon, using a rocD riboprobe, gave results similar to those with the microarray (Fig. (Fig.2).2). The Northern blot analysis further showed that the increased level of rocD mRNA was abolished upon deletion of rocR, the positive regulator of the roc operons (6), from JC112 (strain BG520) (Fig. (Fig.2).2). We postulated that increased arginine uptake and degradation to glutamate would provide a means of acidifying the cytoplasm and/or the medium in the absence of the major proton uptake activity provided by tetL. Additionally, the up-regulated arginine (ornithine) transporters RocE and RocC probably import the amino acid together with H+ rather than with Na+, since they are related to H+-coupled LysP from E. coli rather than to families of Na+-coupled amino acid transporters (www.membranetransport.org).
Genes in three operons involved in carbohydrate metabolism—the aco, rbs, and yvd operons—were overexpressed in JC112. How this relates to deletion of tetL was not clear, and this was not addressed in the current study.
Overexpressed monocistronic genes, or cases where only one gene in an operon was overexpressed, are listed in the top part of Table Table3.3. As could be expected, several stress protein-encoding genes were included, as well as other genes involved in carbohydrate metabolism. Of particular interest was the fivefold increase in citM gene expression, since citM encodes a divalent cation/citrate symporter (31, 34, 52). The increase in citM mRNA was confirmed by Northern blot analysis (Fig. (Fig.22).
In the comparison of changes in gene expression between JC112 and JC112-R, 33 out of 37 genes had similar levels of expression (we arbitrarily chose a less than twofold change between JC112 and JC112-R as being similar). The restoration of wild-type tetL resulted in virtually none of the 37 genes being expressed at the wild-type level (the one case of yvdF is likely unreliable, as other genes in the same operon showed no change according to our arbitrary definition and the significance of a threefold change was borderline). In fact, the only notable change between JC112 and JC112-R was the approximately threefold higher expression of three genes (gsiB, ydaG, and ykzA) encoding stress proteins.
The data for 26 genes that had a negative fold change (i.e., underexpression) (Table (Table4)4) can be summarized as follows.
Half of the underexpressed genes in JC112 were in operons of the PBSX prophage, and these were generally reverted to wild-type levels in JC112-R. Since none of the overexpressed genes had returned to wild-type levels in JC112-R, this result was important in demonstrating that restoration of tetL could indeed reestablish altered gene expression in some cases.
The remaining half of the down-regulated genes included the maeN gene, which encodes a Na+/malate symporter (53); maeN was expressed at very low levels in JC112 and JC112-R relative to the wild type. This was confirmed by Northern blot analysis, which also showed that the rocR status of JC112 did not affect the level of maeN expression (Fig. (Fig.2).2). The rationale for a mutation resulting in such a change seemed straightforward: import of Na+ would need to be avoided in the absence of TetL, the major contributor to alkaline pH homeostasis and a contributor to Na+ resistance. Since the uptake of malate by maeN is coupled to uptake of Na+, the cell would benefit from reducing this mode of using malate as a carbon source.
After maeN, the gene that was most highly repressed in JC112 as well as JC112-R was the nrgA gene, which encodes an ammonium transporter (12). Northern blot analysis of nrgA mRNA also showed down-regulation (Fig. (Fig.2),2), although not as dramatic as that suggested by the microarray data. We postulated that the down-regulation of nrgA in JC112 might be linked to roc operon overexpression, since roc-mediated arginine utilization would generate sufficient internal NH4+ to eliminate the requirement for additional NH4+ import via nrgA. The observed down-regulation of gltA in JC112 and JC112-R was also expected in the context of overexpressed roc operons, for two reasons: first, conversion of arginine to glutamate via the roc pathway would obviate the need for glutamate synthesis by gltAB, and second, gltA expression is positively regulated by gltC, and expression of gltC is in turn negatively regulated by a rocG-dependent pathway (3). Thus, the 10-fold increase in rocG expression in JC112 would result in repression of gltC, which is needed for up-regulation of gltA. To test these hypotheses with respect to nrgA and gltA down-regulation in JC112, Northern blot analysis of nrgA and gltA expression was performed on BD99, JC112, and their derivative rocR-deleted strains, BG519 and BG520. As shown in Fig. Fig.2,2, Northern blot analysis of gltA expression in the roc-deleted strains showed that gltA expression was restored as expected almost to the wild-type level in BG520, the JC112 strain that had rocR deleted. However, the results did not support the idea that elevated roc operon expression is the cause of decreased nrgA mRNA, since nrgA expression was down even in the rocR-deleted strains (Fig. (Fig.22).
(Note that the tetL gene itself is not included in Table Table4,4, although it is present in the wild type and absent in JC112, because the microarray slides we used do not contain the complete set of B. subtilis genes. About 10 open reading frames are not represented, including tetL.)
Additional data (Table (Table5)5) include genes that are unchanged in JC112 but that have higher or lower levels of expression in JC112-R. These were not investigated further.
As another test of the effect of tetL deletion and restoration, measurements of the pH of the growth medium were made during growth of BD99, JC112, and JC112-R. During aerobic growth of bacteria on LB or semidefined media containing amino acids and nonfermentative organic acids, alkalinization of the medium occurs during the last half of the growth curve (46, 48, 54). The data in Fig. Fig.33 demonstrate that growth of JC112 led to a smaller change in pH than growth of BD99. Restoration of the tetL gene in JC112-R resulted in only a marginal recovery of this property. Deletion of rocR increased the alkalinization observed with either the wild type or JC112-R (data not shown), consistent with indications from E. coli studies that different modes of amino acid catabolism are adaptive to different challenges of external pH (4, 5, 16, 38).
Previous studies failed to clarify the basis for the extreme Co2+ sensitivity of JC112 (51). The microarray experiments revealed up-regulation in JC112 of citM, which is capable of taking up citrate in complex with a number of divalent cations, including toxic divalent cations such as Co2+ (31, 33). We hypothesized that up-regulation of citM in JC112 could facilitate use of citrate to offset the reduced use of malate caused by the down-regulation of maeN, since the Spiz-KM medium used routinely in growth experiments contained both malate and citrate. This relationship was shown explicitly by growing BD99, JC112, and JC112-R in medium containing malate plus citrate, malate alone, or citrate alone (Fig. (Fig.4A).4A). Growth of JC112 or JC112-R was still observed on malate alone, but the yield was much lower than on malate plus citrate, which was not the case with the wild-type strain. This presumably reflects the repression of maeN. Also, in contrast to the case for the wild type, the growth yields of JC112 and JC112-R on citrate alone were higher than those on malate alone, which correlated with increased expression and use of citM.
The overexpression of citM immediately provided an explanation for the extreme Co2+ sensitivity of JC112. The increased use of citM would, as a by-product, also result in increased import of Co2+ in a Co2+-citrate symport reaction when Co2+ is added to the medium. Thus, we predicted that deletion of the citM gene in JC112 would restore Co2+ resistance. The citM genes of the wild-type and JC112 strains were substituted with a Pm resistance cassette. The data in Fig. Fig.4B4B show that the predicted link between citM and Co2+ sensitivity was correct. The growth yield of JC112 in the presence of 200 μM Co2+ was negligible, but it was substantially restored when the citM gene was deleted.
The new tetL mutants constructed in this study differed from the earlier tetL deletion strains represented by JC112 and JC112C in having a smaller deletion and being selected on LBK medium instead of Spiz-KM medium (7). Both sets of mutants exhibited the same phenotypes in a range of magnitudes. These phenotypes consistently included Co2+ sensitivity (Table (Table2),2), a phenotype that has no obvious relationship to the catalytic or physiological functions of TetL. The results support the conclusion that deletion of the tetL gene itself is responsible for the multiple phenotypes. Although in both sets of tetL deletion strains the small tetL leader sequence that is upstream of the tetL CDS is also deleted, that sequence is apparently involved solely in translational regulation of tetL (49). The current experiments further show that return of tetL to its original chromosomal position does not reverse the Na+ and alkali sensitivity or the Co2+ sensitivity of the tetL deletion strains. The microarray analyses of JC112, JC112-R, and wild-type B. subtilis strongly support the explicit hypothesis that secondary mutations arise upon tetL deletion, ameliorating the physiological consequences of the mutation. Presumably, the significant genomic changes and consequent patterns of gene expression in the mutants account for the failure of tetL reintroduction into the chromosome to restore a wild-type phenotype. Yet, we did not experience difficulty in isolating tetL deletion mutants or discern a low frequency of their selection in side-by-side experiments in which the amyE gene was deleted (J. Jin and D. H. Bechhofer, unpublished data). The mechanism whereby mutations arise at high frequency under stress conditions is the focus of investigation in other laboratories (17, 25, 44) and is not addressed in our study.
The microarray analyses and the correlated physiological experiments also indicated specific ways in which changes in expression of ion-coupled membrane transporters would be adaptive in JC112. The activities of these transporters are summarized diagrammatically in Fig. Fig.5A.5A. Loss of functional TetL simultaneously reduces Tc and Na+ efflux capacity and both the H+ and K+ accumulation capacity, resulting in the alkali and Na+ sensitivity. Without compensation, the cumulative results of these defects would probably be dire, since their individual effects would act synergistically in causing stress on the cells, in two ways. First, Na+ cytotoxicity increases with increasing cytoplasmic pH, so a deficit in alkaline pH homeostasis makes otherwise subinhibitory Na+ concentrations inhibitory, and elevated cytoplasmic Na+ reduces the upper pH limit for growth (41, 42). Second, the cytotoxicity of Na+ is more closely related to the Na+/K+ ratio than to the concentration of Na+ itself (23). Although the basis for reduced expression of nrgAB in JC112 and JC112-R was not clarified, reduced NrgA function would be adaptive, assuming this protein functions analogously to its E. coli homologue, AmtB, as an NH3 gas channel (30, 47, 57). The residual pH homeostasis of JC112 that is provided by other B. subtilis Na+ (K+)/H+ antiporters, such as NhaC and Mrp (27, 51; see below), would be adversely affected by NH3 because it would consume cytoplasmic H+ accumulated by these antiporters as it comes into equilibrium with its protonated NH4+ form in the relatively acidified cytoplasm (30, 43). Reduced malate uptake by MaeN would be adaptive because it lowers entry of Na+. The two alternate malate uptake systems of B. subtilis that account for growth of JC112 on malate are MleN and CimH. They achieve Na+/H+ antiport and net H+ uptake, respectively, during malate uptake (32, 34, 53) (Fig. (Fig.5B).5B). Use of MleN would ease both the Na+ efflux and H+ accumulation deficits, and use of CimH would ease the H+ accumulation defect. Although the alternate malate transporters do not support yields of JC112 growth on malate alone that are comparable to wild-type levels (Fig. (Fig.4A),4A), the increased activity of CitM and the arginine uptake proteins RocC and RocE provide additional carbon sources without coupling to Na+. Transport of amino acids by RocC and RocE is probably coupled to H+ uptake, providing an added benefit. Increased acid production from metabolism of arginine, and perhaps other amino acids, could be an additional strategy of JC112 to compensate for its poor pH homeostasis capacity. Taken together, the increased CitM, RocC, and RocE function; decreased MaeN and NrgAB function; compensatory use of alternate malate uptake systems; and increased acid production from amino acid catabolism in JC112 could be critical adaptations to the major defect in pH homeostasis and the accompanying Na+ sensitivity that result from loss of tetL. One of those changes, the elevated expression of CitM, was shown in this study to explain the Co2+ phenotype of tetL deletion mutants, a conclusion strongly supported by the greater Co2+ resistance of JC112 when citM was deleted (Fig. (Fig.4B4B).
Elevated expression of czcD and nhaC was found earlier in JC112 and JC112C, although these increases were below the threshold for this study (51). CzcD is an antiporter that effluxes divalent metal ions, including Co2+, in exchange for H+ and K+ (22, 51), and NhaC is a Na+/H+ antiporter (51) (Fig. (Fig.5A).5A). The increase in czcD expression was pronounced only when Co2+ was included in the medium (51), and the levels of CzcD in the membrane would be predicted to similarly increase when excess Co2+ is taken up by the high CitM activity of JC112. The Co2+/H+-K+ antiport activity of CzcD would then concomitantly help reduce the levels of CitM-mediated Co2+ accumulation in the cytoplasm, while bringing H+ and K+ inward, mitigating the alkali sensitivity and diminished K+ acquisition capacity of tetL deletion strains (Fig. (Fig.5A).5A). Deletion of czcD was earlier found to exacerbate the K+ acquisition phenotypes of tetL deletion strains JC112 and JC112C (51). Increased Na+/H+ activity of the NhaC and Mrp antiporters probably facilitates residual cytoplasmic pH homeostasis in the absence of TetL (8, 27, 42).
Finally, this study did not identify master genes whose altered activity accounts for the pattern of altered gene expression observed in JC112 and JC112-R. The genes that are up- or down-regulated might constitute a regulon that is coordinately controlled by a combination of stresses, e.g., elevated cytoplasmic pH and/or Na+ (or the Na+/K+ ratio). However, there is no evidence to date supporting such coordinated regulation. Current information about the genes and operons whose expression is altered in JC112 and JC112-R indicates that they are regulated by a variety of global and specific regulators. For example, some of the affected genes in JC112 are a subset of the members of the SigL regulon (rocA, rocD, rocG, and acoA) (24), some are regulated by CcpA (e.g., rocG, the rbs operon, treA, treP, and gltAB) (2, 40, 50), and regulation of both citM and maeN involves individual two-component systems (34). Analysis of global gene expression in tetL deletion strains other than JC112 would show whether the cellular response to the lack of Tet(L) is a conserved one. This finding would suggest the existence of one or a limited number of controlling genes that can act to coordinate the cellular reaction to the loss of a major Na+ and pH regulator.
This work was supported by research grant GM52837 from the National Institute of General Medical Sciences.
We thank A. L. Sonenshein for informative discussions on amino acid metabolism and S. Wallenstein for assistance in statistical analysis of the data.