Different Foci for the Regulation of the Activity of the KefB and KefC Glutathione-gated K+ Efflux Systems*

KefB and KefC are glutathione-gated K+ efflux systems in Escherichia coli, and the proteins exhibit strong similarity at the level of both primary sequence and domain organization. The proteins are maintained closed by glutathione and are activated by binding of adducts formed between glutathione and electrophiles. By construction of equivalent mutations in each protein, this study has analyzed the control over inactive state of the proteins. A UV-induced mutation in KefB, L75S, causes rapid spontaneous K+ efflux but has only a minor effect on K+ efflux via KefC. Similarly amino acid substitutions that cause increased spontaneous activity in KefC have only small effects in KefB. Exchange of an eight amino acid region from KefC (HALESDIE) with the equivalent sequence from KefB (HELETAID) has identified a role for a group of acidic residues in controlling KefC activity. The mutations HELETAID and L74S in KefC act synergistically, and the activity of the resultant protein resembles that of KefB. We conclude that, despite the high degree of sequence similarity, KefB and KefC exhibit different sensitivities to the same site-specific mutations.

The carboxyl-terminal domain contains a sequence highly similar to a Rossman fold (7,8). A number of mutations that cause increased spontaneous activity in KefC have been characterized and fall in two regions: a region (the "HALESDIE" sequence) predicted to lie at the cytoplasmic face of the membrane domain and residues within, and adjacent to, the Rossman fold of the carboxyl-terminal domain (7)(8)(9). One mutation at the latter site alters the glutathione regulation of the KefC protein (9), but the specific mechanism of activation by the other lesions is not known.
The two E. coli glutathione-gated K ϩ efflux systems can be differentiated by their activation by MG (4). Methylglyoxal only weakly activates KefC, whereas KefB achieves almost maximum activity with this electrophile. In this study we sought to characterize the structural gene for KefB to determine the relatedness to KefC. The two proteins are similar at the sequence and organizational levels. However, the creation of equivalent mutations at a number of positions in KefB and KefC shows that the residues controlling the activation of the two systems are different.

EXPERIMENTAL PROCEDURES
Reagents-All chemical reagents were purchased from Sigma or BDH and were of analytical grade where possible. Chemicals used for preparation of complex growth medium were supplied by Oxoid. Restriction enzymes and Taq DNA polymerase were supplied by Boehringer. Pfu polymerase was obtained from Stratagene. The Qiagen Plasmid Preparation Kits were obtained from Qiagen. All primers used in this study were purchased from Genosys Biotechnologies Inc. The Wizard PCR Preps DNA Purification System was obtained from Promega. The PRISM Ready Reaction DyeDeoxy Terminator Cycle Sequencing Kit was obtained from Applied Biosystems Ltd.
Bacterial Strains-Bacterial strains used in this study are all derivatives of E. coli K-12 (Table I). Strains MJF270 and MJF276 were previously thought to carry an internal deletion in the kefB gene, as they were isolated as suppressers of the kefB110 mutation. 2 However, sequence analysis during the current work revealed that the strain carries two mutations, L75S and D157N, that together inactivate KefB.
Growth and Cell Viability-The growth medium used throughout was K X , where X is the concentration of K ϩ (10). Strains were grown overnight at 37°C in K 10 minimal medium supplemented with 0.04% (w/v) glucose and 1 g⅐ml Ϫ1 thiamine. Ampicillin (25 g⅐ml Ϫ1 ) was included if the strain carried a plasmid. Aliquots of 3 ml were washed in K 1 buffer, suspended in 30 ml of K 1 minimal medium containing 0.2% (w/v) glucose and 1 g⅐ml Ϫ1 thiamine placed at 37°C and the OD 650 monitored over time. For analysis of cell viability the appropriate strains were grown as above and grown to early exponential phase (OD 650 ϭ 0.4) before diluting 10-fold into fresh prewarmed medium containing MG from a 540 mM stock solution. Cell viability was determined exactly as described previously (4).
Potassium Efflux Experiments and Determination of Cytoplasmic pH-Potassium efflux and cytoplasmic pH determinations were carried out as described previously (3,5,6) with cells grown at 37°C in K 120 minimal medium (10) supplemented with 0.2% (w/v) glucose and 1 g⅐ml Ϫ1 thiamine. For the assay cells were washed and suspended in K 0 * This work was supported by The Wellcome Trust Grant 040174 and by Research Leave Fellowship 046289 (to I. R. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. buffer, which lacks ammonium sulfate and MgSO 4 . To determine the intracellular K ϩ content of cells during growth, samples were incubated in K 120 minimal medium to an OD 650 of 0.8 -1.0 and 6 ϫ 1-ml samples were centrifuged through 200 l of bromodecane oil and the supernatant and oil removed. The intracellular potassium concentration was then determined as for potassium efflux (3).
Cloning of kefB-A 2.6-kb fragment encompassing the yheR and kefB genes was amplified by PCR from strain MJF277 using primers KefB3 and KefB4 (Table II), both of which had BamHI restriction sites incorporated at their 5Ј ends. The PCR products obtained were end-filled by treating with the Klenow enzyme, restricted with BamHI, ligated into similarly restricted plasmid pHG165 (11) to create plasmid pKefB, and transformed into strain JM109. Klenow treatment, restriction enzyme digestion, ligation, and transformation procedures were carried out following standard protocols (12).
DNA Sequencing-For DNA sequencing, the cloned yheR and kefB genes from plasmid pKefB and mutant plasmids were amplified in 450 Ϯ 50 base pairs overlapping fragments, using primers designed specifically to complement the available yheR and kefB gene sequences from the E. coli genome project (13). The PCR products obtained were cleaned using Promega PCR DNA clean-up kit. A cycle sequencing reaction with each one of the primer pair used for amplification was performed using Applied Biosystems sequencing premix. The products were cleaned by ethanol precipitation and run on the Applied Biosystems 373A sequencer before being analyzed using the Applied Biosystems "Sequence Editor" program.
Site-directed Mutagenesis-To create plasmid pKefB-2, which carries an aspartate residue at position 262 of the KefB protein in place of the wild-type alanine residue (Table I), site-directed mutagenesis was performed. Primer KefB8 (Table II) was designed such that it encompassed bases 774 -792 of the kefB gene (amino acids 258 -264 of the KefB protein) and contained one mismatch base T781A, which created A262D. Used in conjunction with primer KefB3, KefB8 amplified, by PCR, a product of 1.6 kb in length. This product was Klenow-treated, cleaned, and restricted (as above) at two internal restriction sites, ClaI and DraIII. The resulting 1.15-kb restricted product was ligated into similarly restricted plasmid pKefB to create plasmid pKefB-2. The A262D mutation was confirmed by DNA sequencing of the 1.15-kb insert (as above).
All other mutant plasmids were obtained using the following method, which is based on a technique developed by Stratagene. Parental or wild-type plasmid DNA was purified from a strain that methylates its DNA (JM109 was used for this purpose), and this was used as template for 18 rounds of PCR using the appropriate mutagenic primers (Table  II) and Pfu polymerase. Restriction with DpnI, an enzyme that restricts methylated DNA only, digests template DNA, while leaving amplified and, therefore, mutated DNA undigested. After transformation of the restricted PCR reactions into JM109, the majority of colonies obtained, therefore, should contain the desired mutant plasmid. Analysis of the putative mutants was by restriction enzyme digestion followed by DNA sequencing (see above).

RESULTS
Cloning of the kefB Gene from E. coli-The kefB locus at 75 min on the E. coli genetic map is required for K ϩ efflux elicited pKefB-1 a Restriction sites introduced (ϩ) or deleted (Ϫ) to identify introduction of mutation.  (14). The putative yheR-kefB region was amplified and cloned into plasmid pHG165 to create pKefB (see "Experimental Procedures") and transformed into strain MJF276 (KefB Ϫ KefC Ϫ ). The cloned fragment was sequenced and confirmed to carry the same sequence as that deposited in the data base (13). The transformants were analyzed for electrophile-elicited K ϩ efflux activity and for restoration of protection against MG. Strain MJF276/pKefB rapidly lost 25% of the cell K ϩ pool on suspension into K 0 buffer (first time point 40 s after suspension in K 0 ), and the pool declined to less than 50% of the control over a 25-min incubation (Fig. 1A). The K ϩ pool of MJF276/pKefB was equal to, or greater than, that of MJF276 prior to suspen- sion in K 0 (705 Ϯ 21 and 598 Ϯ 51 mol⅐g Ϫ1 dry cell mass, respectively). Addition of MG caused more than 85% of the K ϩ pool to be lost in the first 7 min of the incubation with the electrophile. The MG-elicited rate of efflux was considerably faster than that observed with strain MJF276 (KefB Ϫ KefC Ϫ ) and MJF274 (KefB ϩ KefC ϩ ), which carries a single chromosomal copy of the kefB gene (note that KefC makes little contribution to MG-elicited efflux) (Fig. 1B). It is notable that the initial rate of K ϩ loss after addition of MG is slower than the maximum activity, which was achieved approximately 3-5 min after addition of the electrophile. Activation by NEM, which reacts spontaneously with glutathione to form the activator N-ethylsuccinimido-S-glutathione elicited high rates of K ϩ efflux from MJF276/pKefB (Fig. 1C). The rate of K ϩ efflux declined steadily as the K ϩ pool declined. When compared with data for strains MJF274 (KefB ϩ KefC ϩ ) and MJF277 (KefB ϩ KefC Ϫ ), which possess single copies of KefB, these data suggest a 10 -12-fold increased expression of the KefB protein in MJF276/pKefB.
The cloned kefB gene provided full protection against MG. When incubated with 0.4 mM MG growth of E. coli cells was inhibited but strain MJF274 (KefB ϩ KefC ϩ ) and MJF276/ pKefB recovered and subsequent growth occurred at the same rate. Strain MJF276, which lacks functional KefB and KefC systems, also recovered but at a much slower rate (data not shown). When exposed to higher concentrations of MG cell death ensued and the degree of survival was greater in MJF276/pKefB than in MJF274 (Fig. 2A). The enhanced protection afforded by the higher activity of KefB in strain MJF276/pKefB correlated with the rate and magnitude of the lowering of cytoplasmic pH (pH i ). Thus, on addition of MG, the cytoplasmic pH of MJF276/pKefB fell rapidly to a level lower than in either MJF276 or MJF274 (Fig. 2B), and these observations are consistent with our previously published model (5,6).
Mutations Affecting the Regulation of KefB Activity-UVinduced chromosomal kefB mutants, which exhibit a rapid K ϩ leak, have been isolated previously (1,10). Five independent mutants, MJF110, MJF111, MJF113, MJF115, and MJF117, were analyzed by PCR amplification of gene fragments from the mutant kefB genes. In each case the same single amino acid change was observed, L75S. This residue is strongly conserved in members of the KefB/C family for which the gene sequence is available (Fig. 3). The amino acid change causes rapid spontaneous K ϩ efflux via the chromosomally encoded KefB system (Fig. 4A). The addition of either MG or NEM did not greatly amplify the rate of K ϩ efflux, which may indicate that the L75S mutation causes the protein to achieve almost maximum activity.
We have shown previously that the wild-type kefC gene can suppress mutations that cause partial spontaneous activation of KefC. The rapid K ϩ leak seen in strain MJF276/pKefB was not observed in strain Frag5/pKefB; there was no immediate loss of K ϩ on suspension in K 0 medium, and the cells retained a similar K ϩ pool to Frag5 throughout the incubation (Fig. 4B). Since Frag5 is the isogenic parent of MJF111, this enabled the potential suppression of the KefBL75S mutant by the wild type gene to be analyzed (Fig. 4B). Potassium loss was consistently observed to be slower from MJF111/pKefB than from MJF111. However, the effect was small relative to the suppression seen previously with the cloned kefC gene and kefC missense mutants (9,15). Introduction of pkC11, which carries the kefC gene in the same plasmid vector as pKefB, did not alter the rate or extent of K ϩ loss. These data suggest that the small effect seen with pKefB is specific and is not due to a general change in membrane organization consequent upon the higher level of expression of the KefB system in MJF111/pKefB. Thus, the L75S mutation has a profound effect on the regulation of the activity of the KefB system and is dominant over the wild-type allele.
Regulatory Mutations in KefB and KefC-We have described previously the effects of a number of mutations that increase the spontaneous activity of the KefC system (9). Since KefB and KefC display significant similarity of sequence and organization, we sought to determine whether each was affected by mutations that affect the spontaneous activity of the other, i.e. do they share common control points.
A262D-We have established previously that the KefC mutation D264A causes high rates of spontaneous K ϩ efflux (9). The KefB protein carries an alanine at the equivalent position (A262) as the wild type sequence (Fig. 3). Since pKefB causes a spontaneous leak we generated the mutation A262D, predicting that it would reduce the leak, but strain MJF276/pKefB-2, which carries the A262D mutation, exhibited only a slightly reduced rate of spontaneous K ϩ efflux (data not shown). The initial rate of MG-induced K ϩ efflux was significantly inhibited and there was a slight reduction in NEM-elicited efflux (data not shown). Therefore, it is clear that this residue plays a less significant role than D264 in KefC.
V427A-Mutations in the Rossman fold of KefC (R416S and V427A) result in a similar phenotype to that seen with the L75S mutation in KefB (9). Val 427 is conserved in the KefC family of proteins (Fig. 3) and the E. coli KefC mutant V427A exhibits rapid K ϩ efflux when present in single or low copy number (9). Strain MJF276/pkC11-4 (V427A), which is a multicopy plasmid based on pkC11 (Table I), failed to grow even in K 120 medium suggesting that the K ϩ leak is too severe to allow growth. In contrast, only a small increase in spontaneous K ϩ leak was observed when the equivalent V428A mutation was introduced into KefB ( Fig. 5; cf. pKefB and pKefB-4). Rates of MG-elicited efflux were rapid but showed no significant difference between pKefB and pKefB-4 (V428A) (data not shown).
The HALESDIE Sequence-Located between two highly con-served regions of KefC is a variable sequence HALESDIE that contains three acidic residues in all four known sequences (Fig.  3). Two UV-induced mutations in this region in E. coli KefC, D264A and E262K, enhance spontaneous K ϩ efflux (9). The KefB protein also has three acidic residues in the equivalent sequence (HELETAID), but also carries an alanine residue at position 262, echoing the D264A mutation in KefC. Therefore, we determined whether it was the presence of three acidic residues or their location at specific positions that controlled the activity of the KefC system. We exchanged the equivalent regions from KefB and KefC, namely the HELETAID and HALESDIE motifs, respectively, and measured the spontaneous and electrophile-induced rates of K ϩ efflux (Figs. 5 and 6, A and B). Replacement of the KefB HELETAID with KefC HALESDIE in plasmid pKefB-3 (Table I) had only a small effect on spontaneous efflux, enhancing the initial rate approximately 2-fold (Fig. 5). The mutation did not significantly affect the rate of electrophile-elicited efflux, which was faster than the spontaneous rate of K ϩ loss (data not shown). Combinations of the HALESDIE motif and V428A in KefB (pKefB-5) also led to higher spontaneous rates of K ϩ efflux, but the double change did not emulate the severity of the combination in KefC. Electrophile-elicited efflux was not significantly affected in the KefB mutant (data not shown). In contrast, in KefC, replacement of the HALESDIE sequence with HELETAID (plasmid pkC11-3) significantly enhanced the spontaneous K ϩ loss (Fig.  6A). This multiple change creates in KefC the D264A mutation but leaves three acidic amino acids in the motif. As a control, an equivalent plasmid pkC11-2 (KefC D264A) was created. Strain MJF276/pkC11-2 failed to grow in K 120 medium, suggesting that the K ϩ leak overwhelms the uptake capacity of the strain. In contrast, MJF276/pSM19 (D264A), which has reduced expression of KefC due to a deletion 5Ј to the structural gene, was able to grow normally in K 120 medium (9). Strain MJF276/ pkC11-3 (HELETAID), which recreates the D264A mutation but in a different context to pkC11-2 (KefC D264A) (Table I), grew normally in K 120 medium and exhibited only a moderate K ϩ leak. These data suggest that the D264A mutation in plasmid pkC11-3 (HELETAID) is partially compensated by the presence of the three acidic residues in the motif. L75S-The KefB L75S mutants exhibited rapid spontaneous efflux (Fig. 4). The importance of this residue in regulating KefC activity was therefore investigated. Strain MJF276/ pkC11-1 (L74S) exhibited spontaneous K ϩ efflux, such that there was a rapid initial loss of K ϩ (approximately 16% of the K ϩ pool) followed by a slower loss of 45% of the K ϩ pool over 25 min (Fig. 6A). However, given that the mutation in the kefC gene is carried on a multicopy plasmid, which leads to an approximately 20-fold increase in KefC protein (4), this rate of K ϩ loss is slow compared with the rate of spontaneous K ϩ efflux observed from the chromosomal KefBL75S mutant (MJF111). This observation applies to both spontaneous (Fig.  6A) and MG-elicited (Fig. 6B) efflux. Transformants carrying pKefB-1 (L75S), a construct equivalent to pkC11-1 (KefC L74S), grew poorly in K 120 medium and could not be assayed for K ϩ efflux. Therefore, the L75S mutation has a much greater effect on the activity of KefB than on KefC.
When the L74S and HELETAID mutations were combined in KefC (plasmid pkC11-5) spontaneous K ϩ efflux was so rapid that at the first time point (approximately 40 s) the cells were completely depleted of K ϩ (Fig. 6A). These cells grew poorly and even in K 120 medium achieved a rate that was only 78% of that of MJF276/pkC11 ( ϭ 0.6 h Ϫ1 and 0.47 h Ϫ1 , for MJF276/ pkC11 and MJF276/pkC11-5 (HELETAID ϩ L74S), respectively. Thus, L74S acted synergistically with the HELETAID mutation. These data are consistent with the effect of the L75S mutation on KefB, which naturally possesses the HELETAID motif, and suggest that these two regions are critical to maintenance of the closed state of KefB. DISCUSSION These studies were undertaken to ascertain whether the amino acid residues critical to the regulation of two homologous K ϩ efflux systems were the same. KefB and KefC are 601 and 620 amino acid proteins, respectively, and are 42% identical and 70% similar in their sequences. The linker regions (amino acids 380 -400 in KefB) are quite diverse and the major points of sequence deviation lie in the extreme carboxyl-terminal region. In view of their overall similarity, it was reasonable to expect that they might possess common regions responsible for the regulation of their activity. KefC is maintained in an inactive state even when present on a multi-copy plasmid, except in the presence of an activating electrophile (7). We have documented previously a number of KefC mutations that increase the spontaneous K ϩ efflux via this protein (9). The mutations substantially increased the rate of K ϩ loss from cells such that they could not grow in media low in K ϩ (K 0.2 ) (1,9,15). The mutations with the greatest effect on activity clustered to two sequences, the Rossman fold and HALESDIE, suggesting that these might be significant controlling regions in the protein.
However, this study suggests that KefB and KefC have evolved different critical residues and that sequence conservation alone is not a guide to the identification of important sequences.
The HALESDIE region is different in KefB and KefC despite strong conservation in the flanking sequences (Fig. 3). Both proteins, and the KefX proteins of H. influenzae and M. xanthus, contain three acidic residues in this sequence, but it is noteworthy that their positions are not conserved. This study aimed to analyze the relative importance of position and sequence. Cells overexpressing the KefC D264A mutation in the HALESDIE context exhibit a much more profound growth defect in K 120 medium than those where the mutation is surrounded by HELETAID, which retains the three acidic residues. The rate of spontaneous K ϩ loss in MJF276/pkC11-3 (KefC HELETAID) is similar to that observed previously in MJF276/pSM26 (9), which carries the kefC D264A mutation but expresses the KefC protein at an approximately 20-fold lower level. Consistent with this observation, MJF276/pkC11-2 (KefC D264A), which has high level expression of the KefC system, cannot grow in K 120 medium. These data suggest that the context of the D264A mutation is a significant determinant of its impact on KefC activity and is consistent with the hypothesis that the number of acidic residues in the HALESDIE region of KefC is more important than their absolute position.
Five independent UV-induced mutations causing fast spontaneous K ϩ leak via KefB were found to be L75S. The importance of Leu 75 is consistent with the observations on the L74S/ HELETAID double mutant of KefC. Combination of L74S and HELETAID in KefC resulted in spontaneous efflux characteristics, resembling those of KefB (L75S). The KefB mutation is more severe than the change in KefC, since strain MJF276/ pKefB-1 (L75S) could not grow in K 120 medium, whereas MJF276/pkC11-5 (KefC L74S/HELETAID) grew, albeit with a reduced growth rate. The combination of the two mutations had a synergistic effect on spontaneous K ϩ loss via KefC (Fig.  6A). These data strongly suggest a possible interaction between FIG. 6. The effect of HELETAID on KefC regulation and activation. K ϩ efflux was measured after suspension of the cells in K 0 buffer. The electrophile was added (arrow) after 3-min incubation of the cells in K 0 buffer. A, spontaneous; B, 3 mM MG-elicited potassium efflux. Symbols: MJF276 (KefB Ϫ KefC Ϫ ) transformed with: pkC11 (wild-type KefC) (f), pkC11-3 (KefC HELETAID) (Ⅺ), pkC11-1 (KefC L74S) (E), pkC11-5 (KefC HELETAID/L74S) (‚), and pKefB (KefB ϩ ) (q). All pkC11 constructs carry the upstream transcription and translation sequences, the yabF gene, and the kefC gene. the region surrounding L75S and the HELETAID motif that leads to the maintenance of the protein in the closed state.