E. coli cells devoid of MdtM are sensitive to alkaline pH
To investigate a physiological role for MdtM in basic pH tolerance we characterised the growth of wild-type and ΔmdtM single-deletion mutant E. coli BW25113 cells under various alkaline pH conditions in both solid and liquid media (Figure 1). On LB-agar plates, both strains exhibited similar growth at pH values of 8.5 to 9.25 (Figure 1A). However, as the pH of the media increased beyond pH 9.25, the growth of ΔmdtM cells was inhibited compared to wild-type cells and only the latter exhibited colony formation at pH 9.5 and pH 9.75. No colonies formed at pH 10. The growth assays in liquid media corroborated the results of the solid media assays and highlighted the deleterious effect of the chromosomal mdtM deletion on alkalitolerance under the experimental conditions employed (Figure 1B). At pH 8.5, the wild-type cells grew slightly better than those of the single-deletion mutant. However, as the pH of the medium was increased the effect of the mdtM deletion became more pronounced; at pH 9.0 and pH 9.25 the wild-type cells grew relatively well whereas the growth of the deletion mutant was suppressed, and even at pH 9.5 and 9.75 the wild-type cells still grew, albeit to a low density. Strikingly, at the latter pH values, growth of the deletion mutant was completely arrested. Neither strain grew at pH 10. Together, these data suggest a role for MdtM in conferral of alkalitolerance to E. coli cells within a narrow pH window framed by pH 9 and pH 10.
E. coli cells expressing MdtM from multicopy plasmid display an alkalitolerant phenotype
To further test the role of MdtM in alkalitolerance, we assayed growth of ΔmdtM cells expressing MdtM from a multicopy plasmid under control of the non-native araBAD promoter under basic pH conditions in solid and liquid media. The energetic costs of overexpressing the transporter resulted in differences in the growth characteristics displayed by cells harbouring plasmidic MdtM compared to those harbouring plain vector alone (data not shown). To account for this, ΔmdtM cells that overproduced dysfunctional MdtM from the pD22A plasmid were used as a control [24]. As shown in Figure 2A, on solid medium at pH 8.5, cells that overexpressed the dysfunctional transporter grew as well as those that overproduced wild-type MdtM. However, as the pH of the medium became more alkaline, growth of cells that synthesised the D22A mutant was progressively inhibited until, at pH 9.5 and 9.75, only the cells that overproduced functional MdtM were capable of colony formation. Both strains failed to grow on solid medium buffered to pH 10. Again, the results of the assays performed on solid medium were corroborated by assays performed in liquid medium (Figure 2B). The latter confirmed that growth of ΔmdtM cells complemented with pD22A was completely arrested above pH 9.25 whereas cells complemented with plasmidic DNA that encoded wild-type MdtM still retained capacity for limited growth up to a pH of at least 9.75. Liquid medium buffered to pH 10 did not support growth of either strain.
To provide further confirmation of the contribution of overproduced MdtM to an alkalitolerant phenotype, and to nullify any contribution by chromosomally-encoded MdfA in the previously described experiments, a set of growth assays that used the ΔmdfA BW25113 strain complemented with pMdtM and the pD22A mutant were performed in liquid medium buffered to pH values of between 8.5 and 9.5 (see Additional file 1). As observed in the assays that utilised ΔmdtM cells transformed with pMdtM and pD22A, there was no difference in the growth characteristics of ΔmdfA transformants cultured at pH 8.5 (see Additional file 1; top left panel). However, as the pH of the growth medium was made more alkaline the ΔmdfA pD22A transformants again became increasingly inhibited until, at pH 9.5, their growth was essentially halted (see Additional file 1; bottom right panel). In contrast, ΔmdfA cells that overproduced plasmidic, wild-type MdtM grew at all the alkaline pH values tested, thus underlining the ability of overexpressed MdtM to compensate for loss of MdfA and thereby support an alkalitolerant phenotype of E. coli.
Finally, to ensure that the observed differences in the cell growth assays were not due simply to differences in the expression levels of the wild-type and D22A mutant transporter, Western blot analysis of dodecyl-β-D-maltopyranoside (DDM) detergent-solubilized cytoplasmic membranes from each strain grown at different pH values was performed (Figure 2C). The analysis confirmed that the wild-type and mutant transporter were not only correctly targeted to the inner membrane but also that each was overexpressed to similar levels irrespective of the pH of the growth medium. Collectively, these results demonstrate that MdtM can confer E. coli with tolerance to alkaline pH values up to 9.75, provided it is functionally expressed from a multicopy plasmid.
Na+ or K+ cations are required for MdtM-mediated alkaline pH tolerance
Inward active transport of protons by antiporters involved in alkaline pH homeostasis in bacteria is usually driven by outward co-transport of monovalent cations such as Na+ or K+[1]. Therefore, we characterised the requirement of Na+ or K+ for MdtM-mediated alkalitolerance by performing growth experiments with E. coli BW25113 ΔmdtM cells complemented with pMdtM in salt-free liquid medium supplemented with different concentrations (ranging from 20 mM to 86 mM) of NaCl or KCl at different pH values. Cells grown at neutral pH did not exhibit any Na+ or K+-dependence (Figure 3A and B, top panels). However, as pH of the medium increased, cell growth showed distinct NaCl or KCl concentration dependence, suggesting that the presence of Na+ or K+ ions is required for MdtM-mediated basic pH tolerance (Figure 3). Notably, at alkaline pH, cells grown in the presence of the higher concentrations of K+ (Figure 3B) achieved higher optical density than those grown in the presence of the corresponding concentrations of Na+ (Figure 3A). The stronger growth of cells observed in the presence of K+ in the external medium probably reflects the activity of the chromosomally encoded ChaA K+/H+ antiporter [12].
Cells were unable to grow in liquid medium in which choline chloride (Figure 4A) or sucrose (Figure 4B) replaced the chloride salt of sodium or potassium, thereby negating a role for either chloride ions or osmotic pressure in MdtM-mediated alkalitolerance. Further evidence of a dependence upon Na+ or K+, but not Cl-, for alkalitolerance came from growth experiments performed in medium containing either sodium gluconate (Figure 4C) or potassium gluconate (Figure 4D); both these compounds supported the growth of MdtM-expressing cells at pH 9.5 and did so in a concentration-dependent manner that reflected the results of the growth experiments performed in liquid medium containing NaCl or KCl (Figure 3). As observed in the experiments that tested the effects of added NaCl and KCl on cell growth at alkaline pH values, cells grown at pH 9.5 in the presence of added K+ gluconate achieved higher optical densities at all the concentrations tested than those cultured in medium that contained Na+ gluconate.
A further indication that the observed alkalitolerance was mediated by MdtM-catalysed monovalent metal cation transport came whole cell transport assays that used fluorescence spectroscopy measurements of the effects of increasing concentrations of NaCl on the EtBr efflux activity of pMdtM transformants of E. coli UTL2 cells (Figure 5). In the absence of NaCl, addition of 0.5% (w/v) glucose to energize the cells resulted in a steady decrease in the fluorescence intensity as EtBr was actively extruded against its concentration gradient (Figure 5, trace A). Dissipation of the proton electrochemical gradient by addition of the ionophore carbonyl cyanide 3-chlorophenylhydrazone (CCCP) caused the fluorescence signal to rise again, indicating disruption of EtBr efflux. In contrast to the results obtained from MdtM-expressing cells, the fluorescence of control cells that expressed the dysfunctional MdtM D22A mutant decreased more slowly and by a much smaller amount over the timescale of the assay (Figure 5, trace E). In this control the residual EtBr efflux is likely due to the activity of chromosomally encoded transporters that recognise EtBr as a substrate. As expected, the addition of 100 mM NaCl to control cells harbouring pD22A had no noticeable effect on the shape or magnitude of the trace (data not shown). In contrast, addition of Na+ cations to UTL2 cells transformed with pMdtM clearly inhibited EtBr efflux (Figure 5, traces B, C and D). Moreover, this inhibition was titratable; addition of increasing concentrations of Na+ resulted in an increasing inhibition of EtBr efflux. Addition of choline chloride had no measurable effect on EtBr efflux (data not shown), thereby establishing that the inhibition of EtBr efflux by NaCl was due solely to Na+ ions. Together, the results of the whole cell transport assays suggest that EtBr and Na+ utilise the same binding site and/or translocation pathway in MdtM. Indeed, in the closely related MdtM homolog MdfA, the multidrug and Na+ cation translocation pathways overlap [9].
MdtM catalyses K+/H+ and Na+/H+ exchange activities
The growth assay and whole cell EtBr efflux data implied that MdtM-catalysed K+/H+ and Na+/H+ antiport activities underpinned alkalitolerance. To examine if MdtM mediated the exchange of K+ and Na+ for protons, we measured the changes in luminal pH of inverted membrane vesicles generated from antiporter-deficient TO114 cells [26] that overexpressed wild-type MdtM by monitoring the fluorescence dequenching of acridine orange upon addition of Na+ gluconate or K+ gluconate to the transport assay buffer at the indicated alkaline pH values (Figure 6). Inverted vesicles prepared from TO114 cells that overproduced dysfunctional MdtM D22A mutant were used as controls.
The control vesicles (Figure 6; grey traces) exhibited negligible Na+/H+ or K+/H+ activities at pH values of 9.0 to 9.75. This was expected because the TO114 cells from which the inverted vesicles were generated are devoid of the major antiporters NhaA, NhaB and ChaA that function primarily in monovalent metal cation/H+ exchange at alkaline pH [12, 26]. However, at pH 8.5 the controls exhibited some degree of exchange activity; this activity was more pronounced upon addition of K+ ions and resulted in ~30% dequenching of the initial lactate-induced fluorescence quench (Figure 6B, top panel). It is conceivable that this dequenching was due to the activity of other, chromosomally-encoded antiporters that operate in the same pH range and that have a greater affinity for K+ than Na+ ions. In all control experiments, addition of 100 μM CCCP at the time indicated resulted in dissipation of the ΔpH, as revealed by an instantaneous dequenching of the fluorescence signal. This confirmed that the inverted vesicles had maintained integrity over the lifetime of the assay.
In contrast to the controls, addition of Na+ or K+ to inverted vesicles containing recombinant wild-type MdtM resulted in a rapid and significant dequenching of the lactate-induced, acridine orange steady state fluorescence at all the alkaline pH values tested (Figure 6; black traces), thus indicating that MdtM was responsible for catalysing both Na+/H+ and K+/H+ exchange reactions. The magnitude of the dequenching at each pH value, however, varied depending upon the pH and the metal cation added; in the case of added Na+ the most pronounced dequenching was observed at pH 9.25 (Figure 6A; black traces) whereas the maximal K+-induced dequenching occurred at pH 9.0 (Figure 6B; black traces). As observed from the assays performed on control vesicles, the addition of CCCP to the reaction mixtures resulted in a further dequenching of the fluorescence signal, confirming that the MdtM-containing inverted vesicles had also maintained integrity for the lifetime of the assay.
pH profiles of MdtM-catalysed K+/H+ and Na+/H+ exchange activities
Measurements of the acridine orange fluorescence dequenching enabled a plot of the K+/H+ and Na+/H+ exchange activities (expressed as the percentage dequenching of the lactate-induced fluorescence quenching) as a function of pH to be constructed, and this revealed a clear pH-dependence for both (Figure 7A). At pH ≤6.5, no transport of the probed K+ and Na+ cations was detected, providing further evidence that MdtM does not operate as a monovalent metal cation/H+ antiporter at acidic pH. However, as the pH increased and became more alkaline, a significant exchange activity was recorded. From no detectable activity at pH 6.5, the activity of MdtM in the presence of 100 mM Na+ or K+ increased to ~20% dequenching at the pH range of 7.0 to 8.0 (Figure 7A). Between pH 8.0 and 9.75, the pH profiles for both exchange activities were essentially bell-shaped, with the activity optimum for MdtM-catalysed K+/H+ antiport at pH 9.0 and that of Na+/H+ antiport at pH 9.25. The activity of MdtM at each pH optimum was similar, attaining a mean corrected fluorescence dequenching of ~ 80%.
Apparent affinity of MdtM for transported Na+ and K+ is low
To permit a crude assessment of the affinity of MdtM for the transported metal cations, a series of dose–response experiments, covering substrate ranges of 5 mM - 125 mM Na+ and K+ (Figures 7B & C), were performed on inverted vesicles at the pH optimum of each substrate using the acridine orange fluorescence quenching /dequenching assay as described in the Methods section. Although it was not possible to access actual K
m values using these assays, they did permit the concentrations of Na+ and K+ required for the half-maximal response to be estimated and the results implied that MdtM has low apparent affinity for monovalent metal cations, with [Na+]1/2 of 38±6 mM (Figure 7B) and [K+]1/2 of 32±7 mM (Figure 7C).
MdtM also catalyses Rb+/H+ and Li+/H+ antiport but not Ca2+/H+ exchange
Bacterial Na+/H+ and K+/H+ antiporters that function in alkaline pH homeostasis can often also transport cations of other metals such as rubidium, lithium and calcium [12, 27–29]. Therefore, the capacity of inverted vesicles of TO114 cells transformed with pMdtM to support the exchange of Rb+, Li+ and Ca2+ for protons was examined at pH 9.0 using the acridine orange fluorescence quenching/dequenching assay. Not unexpectedly, the addition of 40 mM Rb2SO4 to the inverted vesicles containing wild-type MdtM resulted in ~35% dequenching of the lactate-induced fluorescence quench, indicating that MdtM was capable of catalysing the exchange of the potassium analogue Rb+ for protons (Figure 8A; black trace). A similar magnitude of dequenching was observed when 40 mM Li2SO4 was added to inverted vesicles (Figure 8B; black trace), confirming that Li+/H+ exchange is also catalysed by MdtM. In contrast, the addition of 40 mM CaSO4 to inverted vesicles did not elicit a dequenching of the respiration induced fluorescence quenching (Figure 8C; black trace), demonstrating that Ca2+ is not a substrate for MdtM under the conditions tested; this assay also confirmed that the activity observed upon addition of the sulphate salts of Rb+ and Li+ to the inverted vesicles was not due to any MdtM-mediated exchange of sulphate anions for protons. Additionally, magnesium sulphate or choline chloride at final concentrations of 40 mM also failed to dequench the fluorescence (data not shown). Control assays conducted with inverted vesicles that contained the dysfunctional MdtM D22A mutant did not exhibit any fluorescence dequenching in response to the addition of any of the cations tested (Figure 8; grey traces), thereby providing further robust evidence that the dequenching observed upon the addition of Rb+ and Li+ to vesicles generated from TO114 cells transformed with pMdtM was due to a process mediated by the functionally expressed recombinant transporter.
MdtM-catalysed K+/H+ and Na+/H+ antiport is electrogenic
Generally, cation/proton antiporters involved in alkaline pH homeostasis are required to mediate an electrogenic antiport that is energized by the transmembrane electrical potential, Δψ [5]. Therefore, to probe whether MdtM catalyses electrogenic antiport, inverted vesicles were generated from TO114 cells transformed with pMdtM and assayed for electrogenicity in a chloride-free and potassium-free buffer using the Δψ–sensitive fluorophore Oxonol V. Inverted vesicles produced from TO114 cells transformed with pD22A were used as a negative control. In all the assays, energization of the vesicles by lactate resulted in a rapid quench of Oxonol V fluorescence indicating the generation of respiratory Δψ (Figure 9). To ensure the suitability of the experimental conditions for detection of electrogenic antiport, a positive control (Figure 9F) was performed using inverted vesicles produced from E. coli BW25113 cells that contained a full complement of electrogenic antiporters. This control experiment was performed at pH 8.5 to specifically enable detection of NhaA-catalysed, electrogenic Na+/H+ exchange [30]. Addition of Na+ to these vesicles caused a rapid partial dequenching of the Oxonol V fluorescence, indicating electrogenic antiport. Addition of the protonophore CCCP at the time indicated resulted in dissipation of the respiratory Δψ.
Addition of Na+ (Figure 9A) or K+ (Figure 9C) to inverted vesicles produced from TO114 cells that overexpressed wild-type recombinant MdtM resulted in a partial depolarization of Δψ, whereas addition of the same metal cations to negative control vesicles containing dysfunctional MdtM resulted in no detectable depolarization (Figures 9B and 9D). In each case, addition of the protonophore CCCP at the times indicated resulted in dissipation of Δψ. In another control experiment, addition of the ionophore nigericin to TO114/pMdtM vesicles pre-incubated in the presence of 50 mM K+ gluconate resulted in a small increase in the magnitude of Δψ due to conversion of ΔpH to Δψ by the electroneutral K+/H+ exchange activity of nigericin (Figure 9E). Addition of valinomycin to the same vesicles at the time indicated completely dissipated Δψ.
Together, these qualitative data clearly indicate that MdtM-catalysed Na+/H+ and K+/H+ antiport at alkaline pH is electrogenic, with > 1 H+ exchanged per Na+ or K+; quantitative determination of the stoichiometry of antiport will require reconstitution of purified MdtM into proteoliposomes as was done for NhaA [31], and this is the subject of ongoing work by our laboratory.
In alkaline environments, MdtM functions to maintain a cytoplasmic pH that is acidic relative to external pH
Taken together, all the previous data strongly support the idea that MdtM contributes to cytoplasmic pH homeostasis under conditions of alkaline stress. Therefore, to demonstrate directly a role for MdtM in this process, in vivo measurements of the intracellular pH of E. coli BW25113 ΔmdtM transformed with pMdtM or pD22A at different external alkaline pH values between pH 7.5 and pH 9.5 were performed in the presence of NaCl using fluorescence measurements of the free acid of the pH-sensitive probe 2,7-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl ester (BCECF-AM). Calibration of our system resulted in a reasonably linear correlation between intracellular pH and the 490 nm/440 nm fluorescence ratio over a range of pH values from 7.5 to 9.5 (Figure 10A) thereby making internal cellular pH measurements over this range amenable. The intracellular pH of cells that overexpressed wild-type MdtM from a multicopy plasmid remained relatively constant (at between pH 7.5 and 8.0) over the range of external alkaline pH values tested (Figure 10B; filled symbols). In contrast, cells expressing the dysfunctional D22A mutant of the transporter were unable to maintain a stable cytoplasmic pH, acidic relative to the outside; as the external pH increased there was a concomitant alkalinisation of the cell cytoplasm (Figure 10B; empty symbols). These results uphold our contention that MdtM contributes to alkaline pH homeostasis in E. coli.