The Phosphohistidine Phosphatase SixA Targets a Phosphotransferase System

One common means to regulate protein activity is through phosphorylation. Protein phosphatases exist to reverse this process, returning the protein to the unphosphorylated form. The vast majority of protein phosphatases that have been identified target phosphoserine, phosphotheronine, and phosphotyrosine. A widely conserved phosphohistidine phosphatase was identified in Escherichia coli 20 years ago but remains relatively understudied. The present work shows that this phosphatase modulates the nitrogen-related phosphotransferase system, a pathway that is regulated by nitrogen and carbon metabolism and affects diverse aspects of bacterial physiology. Until now, there was no known mechanism for removing phosphoryl groups from this pathway.

The name for this superfamily is derived from the conserved histidine residue that is essential for catalysis and does not reflect the substrate specificity of the superfamily's members (4). In fact, the substrate preferences for this family of phosphatases are diverse, ranging from small phosphorylated metabolites to large phosphoproteins containing phosphorylated tyrosine, serine, threonine, or histidine residues. As the smallest member to have been crystalized, SixA's structure is representative of the minimal core fold of the superfamily's catalytic domain.
As of this writing, all E. coli isolates with fully sequenced genomes contain sixA, making it a member of the E. coli core genome. In addition, homologs of sixA are widespread among proteobacteria, actinobacteria, and cyanobacteria (5). Thus, SixA likely plays an important role in the physiology of these organisms. Previous work suggested that E. coli SixA dephosphorylates the histidine-containing phosphotransfer (HPt) domain of ArcB (6,7), a histidine kinase that, together with its partner response regulator ArcA, regulates the transition to anaerobiosis. Since those initial studies, however, no additional research has been published exploring the function of SixA.
Here, we reexamine the role of SixA in E. coli. We do not observe an effect of SixA on ArcB/ArcA-regulated transcription, and we find that SixA-null strains have a growth defect that is independent of ArcB. Suppressor screens and analysis of various mutants suggest that the growth defect is due to misregulation of the nitrogen-related phosphotransferase system (PTS Ntr ). We further show that the phosphorylation state of EIIA Ntr , one of the protein components of the PTS Ntr , is affected by SixA, and we propose a model in which SixA dephosphorylates the PTS Ntr protein NPr.

RESULTS
The absence of SixA causes a growth defect in minimal medium. We noticed that a sixA deletion in Escherichia coli MG1655 grows poorly in glycerol minimal medium. In aerobic cultures at 37°C, the wild-type and ΔsixA strains reached similar optical densities (OD) at stationary phase, but the ΔsixA strain grew more slowly, with an exponential-phase-doubling time twice that of the wild-type strain (Fig. 1A). In contrast, when grown in LB Miller medium, the two strains showed no difference in growth (Fig. 1B).
To facilitate growth comparisons between strains, we used an endpoint assay in which OD was measured after 12 h, a time point that gave a large difference in OD between the wild-type and ΔsixA strains (Fig. 1A). With this assay, we verified that reintroducing the sixA gene on a plasmid restores wild-type growth (Fig. 1C). Moreover, a plasmid expressing a catalytically inactive SixA mutant, SixA(H8A) (4,6), failed to complement the sixA deletion (Fig. 1C). Taken together, these results demonstrate that the growth defect is due to the absence of SixA and is likely due to the absence of SixA phosphatase activity.
We also found that the ΔsixA strain grows more slowly than the wild-type strain in minimal medium with acetate, maltose, or glucose as a carbon source. These observations for the ΔsixA mutant are consistent with the results of a chemical genomics screen that followed growth of E. coli mutants under various conditions (8). Data from that study also indicate that procaine has a more pronounced inhibitory effect on the ΔsixA mutant than on the wild-type strain, a result that we confirmed and also took advantage of for some genetic screens described below (see Materials and Methods).
The ⌬sixA mutant growth defect is independent of ArcB and ArcA. Since SixA was previously reported to be a phosphohistidine phosphatase for the histidine kinase ArcB (6, 7), we tested whether the growth defect of a ΔsixA strain depends on the ArcB/ArcA two-component system. To our surprise, deletion of sixA produced a growth defect in strains lacking either arcA or arcB (Fig. 1D), indicating that the slow growth does not depend on ArcB/ArcA activity.
To our knowledge, there is only one published report supporting a role for SixA as an ArcB phosphatase in vivo (7). We therefore wanted to verify that the absence of SixA affects ArcA-dependent expression under anaerobic respiratory conditions, as previously reported (7). We tested the effect of a sixA deletion on transcription from the icdA P 1 promoter, which is repressed by phosphorylated ArcA and is not known to be regulated by other transcription factors (9) (Fig. 1E). For anaerobic growth with nitrate as an electron acceptor, deletion of sixA had no effect on transcription of a P 1icdA -lacZ reporter (Fig. 1E). Importantly, ␤-galactosidase levels for anaerobic growth in the absence of nitrate were 2-fold lower than the corresponding levels with nitrate present (30 Ϯ 5 Miller units and 60 Ϯ 7 Miller units,respectively), indicating that the reporter is not fully repressed and is sensitive to changes in ArcA phosphorylation under conditions of nitrate respiration. Thus, for at least our strains and assay conditions, we have no evidence that SixA affects ArcB/ArcA signaling in vivo.
Suppressors of the ⌬sixA growth defect indicate a connection between SixA and the PTS Ntr . To identify genes that interact with sixA, we looked for mutants that suppress the slow-growth phenotype of a ΔsixA strain. From a pool of random transposon mutants, we identified insertions in ycgO, which encodes a protein sharing homology with cation/proton antiporters. We verified that a clean deletion of ycgO in Optical density at 600 nM (OD 600 ) was monitored for two independent cultures of each strain. Symbols represent the average OD 600 ; bars indicate ranges and are not visible where smaller than the symbol. (C) Cultures of the wild-type (MG1655) and ΔsixA (JES13) strains, transformed with empty vector pTrc99a, SixA-expressing plasmid pSixA, or SixA(H8A)-expressing plasmid pSixA(H8A) were grown in glycerol minimal medium with ampicillin for 12 h. Symbols represent the OD 600 for individual cultures, and the horizontal black lines indicate the average OD 600 from three biological replicates. (D) Strains were grown in glycerol minimal medium for 12 h. Symbols are as described for panel C. Strains are MG1655, JES13, JES47, JES48, JES49, and JES50. (E) ␤-Galactosidase activity was measured for strains containing a P 1icdA -lacZ reporter following anaerobic growth in minimal medium with 40 mM KNO 3 . Symbols represent the activity (Miller units) measured for individual cultures, error bars report standard deviations, and the blue bars indicate the average levels of activity from the three biological replicates. Strains are JES252, JES253, and JES254. As shown in the illustration to the right of the graph, SixA was previously proposed to dephosphorylate ArcB.
Modulation of PTS Ntr by the Phosphatase SixA ® the ΔsixA strain suppresses the slow-growth phenotype ( Fig. 2A) and also that reintroduction of ycgO on a plasmid in the ΔycgO ΔsixA strain restores a growth defect (Fig. 2B).
We also screened a plasmid library of E. coli genomic DNA to search for multicopy suppressors. As expected, the screen yielded plasmids containing sixA. In addition, we isolated three distinct plasmids that had only one intact gene in common: ptsN. Using a plasmid containing only the ptsN gene (modified to encode a 3ϫFLAG-tagged protein), we confirmed that ptsN overexpression suppresses the ΔsixA growth defect (Fig. 2C). The gene ptsN encodes EIIA Ntr , a protein of the nitrogen-related phosphotransferase system (PTS Ntr ) (reviewed in reference 10). The PTS Ntr is composed of three proteins, EI Ntr , NPr, and EIIA Ntr , and it is regulated by a series of successive phosphoryl group transfers onto histidine residues: phosphoenolpyruvate to EI Ntr , EI Ntr -P to NPr, and NPr-P to EIIA Ntr (Fig. 2D). Several studies have found that inactivation of ptsN slows growth in minimal glucose medium (11,12), and it was reported that deletion of ycgO suppresses this growth defect (13). Taken together, these observations suggest that the ΔptsN and ΔsixA growth defects may be due to the same mechanism.
⌬sixA cells have lower intracellular potassium levels than wild-type cells. It was previously observed that cellular potassium content is lower in a ΔptsN strain and is restored by deleting ycgO (13) (Fig. 2D). We therefore reasoned that the ΔsixA mutation might have a similar effect on the cell. We measured cellular potassium levels of wild-type and mutant strains by ICP-MS (13). The potassium levels of ΔsixA and ΔptsN strains were 2-fold lower than the levels of the wild-type strain (Fig. 2E), and deletion of ycgO restored the potassium content of the ΔsixA strain to a level comparable to that of the wild-type strain (Fig. 2E). These observations suggest that the absence of SixA, similarly to the absence of EIIA Ntr , may cause aberrant YcgO activity and misregulation of cellular potassium content. Changes in SixA phosphatase activity alter EIIA Ntr phosphorylation levels. The results described above suggest a model in which the SixA phosphatase modulates phosphorylation of the PTS Ntr . To test this, we assayed the relative amounts of EIIA Ntr phosphorylation with native gel electrophoresis, which has previously been shown to resolve the phosphorylated and unphosphorylated forms of the protein (11,14). To detect EIIA Ntr and EIIA Ntr -P by Western blotting, we used strains that produce 3ϫFLAGtagged EIIA Ntr from the native ptsN locus (15).
Comparing sixA ϩ and ΔsixA strains transformed with an empty vector and grown to stationary phase in LB medium, we found that the levels of phosphorylated and unphosphorylated EIIA Ntr were similar in the sixA ϩ strain whereas EIIA Ntr was predominantly phosphorylated in the ΔsixA strain ( Fig. 3A, lanes 1 and 2). In addition, expression of wild-type SixA from a plasmid decreased EIIA Ntr -P levels and increased EIIA Ntr levels relative to the empty vector control (Fig. 3A, lanes 2 and 3). Finally, expression of the catalytic-site mutant SixA(H8A) did not change the relative abundances of EIIA Ntr and EIIA Ntr -P (Fig. 3A, lanes 2 and 4).
Curiously, in testing LB and minimal medium cultures without a plasmid, we did not see a convincing difference in the levels of EIIA Ntr -P between sixA ϩ and ΔsixA strains. However, we note that for these cultures, EIIA Ntr was predominantly phosphorylated in the sixA ϩ strain, making it difficult to detect a further increase of EIIA Ntr phosphorylation in the ΔsixA strain. At present, we do not understand why the cells with and without a plasmid showed different effects. Nevertheless, the results described above show that SixA affects PTS Ntr phosphorylation. In particular, the absence of SixA can lead to hyperphosphorylation of EIIA Ntr .
SixA modulates EIIA Ntr phosphorylation through NPr. In our working model for the ΔsixA growth defect, SixA-null strains have increased phosphorylation of one or more PTS Ntr proteins. This in turn causes the putative potassium-proton antiporter YcgO to become more active, and the ΔsixA cells suffer from the consequences of YcgO overactivity (Fig. 2D) (13). Consistent with this model, we found that the ΔsixA and ΔptsN strains have similar growth defects (Fig. 3B). The double deletion strain grew Optical density at 600 nM (OD 600 ) was measured for two independent cultures of each strain. Symbols represent the average OD 600 ; bars indicate ranges and are not visible where smaller than the symbol. Samples for this growth curve were collected at the same time as the samples for the growth curve in Fig. 3B, so the data for the wild-type and ΔsixA strains are identical between figures. Strains are MG1655, JES13, JES30, and JES31. (B) Growth curves were measured for strains transformed with empty vector pMG91 or YcgO-expressing plasmid pYcgO in glycerol minimal medium with chloramphenicol. Symbols are as described for panel A. Strains are MG1655, JES13, and JES31. (C) Cultures of the wild-type (MG1655) and ΔsixA (JES13) strains, transformed with empty vector pTrc99a or ptsN-3ϫFLAG-expressing plasmid pEIIA Ntr , were grown in glycerol minimal medium with ampicillin for 13 h. Symbols represent the OD 600 for individual cultures, and the horizontal black lines indicate the average OD 600 from three biological replicates. (D) EIIA Ntr is part of the nitrogen-related phosphotransferase system (PTS Ntr ). Phosphoryl groups entering the system originate from phosphoenolpyruvate and are passed by successive phosphotransfers between the three PTS Ntr proteins. According to a previously proposed model (13), unphosphorylated EIIA Ntr inhibits YcgO, decreasing potassium efflux. (E) Exponential-phase cultures, grown in glycerol minimal medium, were assayed for potassium by inductively coupled plasma mass spectrometry (ICP-MS). Symbols represent the cellular potassium content for individual cultures, error bars report standard deviations, and the purple bars indicate the average cellular potassium content from all biological replicates. Strains are MG1655, JES208, JES13, and JES31. The average cellular potassium content for the wild-type strain is 243 nmol K ϩ /OD 600 . slightly more slowly than strains with either deletion alone, which accounts for the OD of the double deletion being slightly lower than the ODs of the single deletions.
According to a model proposed previously (13), unphosphorylated EIIA Ntr inhibits YcgO, and slow growth results from loss of this inhibition. We reasoned that the absence of EI Ntr or NPr, the first two components of the PTS Ntr (Fig. 2D), would cause EIIA Ntr to be fully unphosphorylated. In this case, eliminating the SixA phosphatase would have no effect on growth. Indeed, deletion of sixA had no effect on the growth of an NPr-null strain, as anticipated (Fig. 3C). However, we found that deletion of sixA still produces a growth defect in a strain lacking EI Ntr .
We interpret this unexpected result in the context of our model in which the growth defect arises from increased phosphorylation of the PTS Ntr . The fact that deleting sixA affected growth of an EI Ntr -null strain indicates that there is another source of phosphoryl groups for the PTS Ntr , which we hypothesize is the sugar phosphotransferase system (PTS sugar ) (Fig. 3D). The PTS sugar is a pathway homologous to the PTS Ntr that facilitates carbohydrate uptake into the cell by phosphorylating transported carbohydrates (reviewed in references 16 and 17). Previous work has shown that EIIA Ntr can be phosphorylated by PTS sugar components (11,(18)(19)(20)(21). This cross talk might be expected to occur during growth on non-PTS carbohydrates such as glycerol, which is characterized by high phosphorylation levels of PTS sugar proteins, but not for growth on a PTS sugar such as glucose, for which PTS sugar proteins are less phosphorylated (21,22). We therefore hypothesized that growth on glucose would limit cross talk between the PTS sugar and PTS Ntr . With this additional source of phosphorylation eliminated, we expected our original reasoning to apply, i.e., that deleting sixA should have no effect on growth in strains lacking either EI Ntr or NPr. In agreement with this hypothesis, deleting sixA had no effect on the growth of either EI Ntr -null or NPr-null strains in glucose minimal medium (Fig. 3E). These observations support our model suggesting that the ΔsixA slow-growth phenotype requires phosphoryl group transfer through the PTS Ntr . Furthermore, the results presented in Fig. 3C and E imply that SixA modulates PTS Ntr through NPr, since in the absence of NPr, the deletion of sixA has no effect.
To test whether SixA modulation of EIIA Ntr phosphorylation depends on NPr, we overexpressed SixA in the presence or absence of NPr and monitored levels of EIIA Ntr and EIIA Ntr -P. We observed significant EIIA Ntr phosphorylation in the absence of NPr for cells growing on LB (Fig. 3F, lanes 1 and 3), which is consistent with cross talk to EIIA Ntr . Furthermore, we found that SixA expression had no observable effect on EIIA Ntr -P in an NPr-null strain (Fig. 3F, lanes 3 and 4), supporting a model in which SixA acts on NPr.

DISCUSSION
Targets of SixA phosphatase activity. Only one target of the SixA phosphohistidine phosphatase has been reported to date: E. coli ArcB. The interaction between these proteins was discovered through a screen for inhibitors of OmpR phosphorylation by the ArcB histidine phosphotransfer (HPt) domain (6). Further support was provided by in vitro data showing that SixA dephosphorylates ArcB and by in vivo results indicating that deleting sixA affects expression of an ArcA-dependent sdh transcriptional reporter under conditions of anaerobic respiration (6,7). However, our experiments with a different ArcA-dependent reporter growing anaerobically on nitrate did not show any evidence that SixA modulates ArcA phosphorylation (Fig. 1E), and another study failed to see an effect of SixA on ArcA-regulated transcription under conditions of microaerobic growth (23). Thus, while the in vitro results reported in references 6 and 7 indicate that SixA can function as a phosphohistidine phosphatase, the role of SixA in modulating ArcB/ArcA signaling in vivo is unclear and requires further investigation.
The genetic and biochemical evidence presented here reveals a different-ArcBindependent-function for SixA: modulation of PTS Ntr phosphorylation. As with other phosphotransferase systems, the activity of the PTS Ntr depends on the phosphorylation states of its protein components. For example, many pathways are reportedly mediated by the unphosphorylated form of EIIA Ntr (12, 13, 24-29; reviewed in reference 17). The phosphorylation states of the PTS Ntr proteins depend on a balance between the rate of phosphoryl transfer into the system from phosphoenolpyruvate and the rate of dephosphorylation. However, the mechanism by which PTS Ntr proteins are dephosphorylated has not been established. Whereas the PTS sugar transfers phosphoryl groups to incoming sugars, providing a phosphoryl group sink for this system (Fig. 3D), no analogous sink for the PTS Ntr has been reported. A recently proposed model for a global stress response pathway in Acinetobacter baumannii, a bacterium whose PTS Ntr lacks an EIIA Ntr protein, suggests a dephosphorylation pathway for NPr via phosphotransfer to a serine on another protein, GigB, which is subsequently dephosphorylated by a phosphoserine phosphatase (30). However, E. coli and many other bacteria that have a PTS Ntr do not have GigB.
One possible path for PTS Ntr dephosphorylation is through cross talk with the PTS sugar (11,(18)(19)(20)(21) (Fig. 3C), although the extent of this cross talk in wild-type bacteria is unknown. It is also possible that the PTS Ntr can be dephosphorylated by spontaneous hydrolysis, due to the intrinsic instability of phosphohistidines of the phosphorylated PTS Ntr proteins (31). Our work reveals a different mechanism that is mediated by the SixA phosphatase. On the basis of our observation that NPr is required for SixA to decrease EIIA Ntr phosphorylation (Fig. 3F), we propose that SixA removes phosphoryl groups from the PTS Ntr via NPr. Furthermore, since phosphohistidine phosphatase activity has been demonstrated in vitro for SixA, we hypothesize that SixA directly dephosphorylates NPr-P.
We were led to the observation that SixA modulates the PTS Ntr through our discovery of a growth defect for the ΔsixA mutant and through our suppressor analysis, which pointed to a connection with the growth defect associated with the ΔptsN mutant. We note, however, that we do not yet understand the mechanism underlying this growth defect or its suppression by ΔycgO.
Modulators of PTS Ntr activation. PTS Ntr activation is influenced by both nitrogen and carbon metabolism. In particular, EI Ntr is inhibited by glutamine and may be stimulated by ␣-ketoglutarate (32)(33)(34). In addition, EIIA Ntr phosphorylation is influenced by PTS sugars through cross talk from the PTS sugar in at least some genetic backgrounds (19)(20)(21). Such cross talk could account for our observation that the severity of the ΔsixA growth defect depends on the carbon source. Non-PTS carbon substrates such as glycerol, maltose, and acetate produce pronounced ΔsixA growth defects that are easily observable as small colonies on solid media. Growth on the PTS sugar glucose, however, yields a subtle growth defect that we could detect only with optical density measurements. These phenotypes suggest a model in which, during growth of E. coli ΔsixA on non-PTS substrates, increased phosphorylation of the PTS sugar increases phosphorylation of the PTS Ntr through cross talk, thereby exacerbating the ΔsixA slow-growth phenotype. In the context of this model, SixA is important for limiting cross talk from the PTS sugar in wild-type E. coli. However, it is also possible that the dependence of the ΔsixA slow-growth phenotype on carbon source reflects differences in phosphoenolpyruvate levels.
SixA may also provide an additional point of control for the PTS Ntr , since factors affecting SixA expression or activity would modulate PTS Ntr activation. For example, SixA protein levels may increase in response to cell envelope stress since sixA is a member of the E regulon (35)(36)(37)(38). Several examples of interactions between the PTS Ntr and the cell envelope have been identified (39)(40)(41). Thus, increased dephosphorylation of the PTS Ntr by SixA could play an important role in the cell envelope stress response.
Concluding remarks. SixA is a well-conserved protein found in proteobacteria, actinobacteria, and cyanobacteria (5). Gene expression and proteome analyses have implicated SixA in biofilms of E. coli and Acidithiobacillus ferrooxidans (42)(43)(44)(45), in nutrient limitation-induced phenotypes of Legionella and Salmonella (46)(47)(48), and macrophage infection with Salmonella (38). The roles of SixA in these contexts were interpreted to involve oxygen limitation, ArcB, or other two component systems but may instead involve SixA regulation of phosphotransferase systems. Does SixA function as a PTS Ntr modulator in other bacterial species? Are there other targets of SixA? Our understanding of the roles of phosphatases in modulating histidine phosphorylation is in its early stages in bacterial and eukaryotic systems alike. Tables 1, 2, and 3, respectively. Growth conditions. Rich medium consisted of LB Miller medium (49) (containing [per liter] 10 g tryptone, 5 g yeast extract, and 10 g NaCl). Defined medium consisted of minimal A medium (49)  ΔsixA   SO 4 , and 0.5 g Na citrate dihydrate) with 0.2% of the indicated carbon source and 1 mM MgSO 4 added after autoclaving. Minimal medium plates contained 1.5% agar. Bacteria were grown at 37°C, except during some steps of strain construction, when cultures were grown at 30°C to maintain temperature-sensitive plasmids. The final concentrations of the antibiotics ampicillin, kanamycin, and chloramphenicol were 50 to 100, 25 to 50, and 12 to 25 g/ml, respectively. Plasmids containing sixA or ptsN were used without IPTG (isopropyl-␤-Dthiogalactopyranoside) induction and thus expressed the corresponding proteins from the basal activity of the trc promoter. For anaerobic growth, minimal medium containing 0.2% glucose and 0.1% Casamino Acids was inoculated with a single colony and transferred to an anaerobic chamber, and the culture was grown overnight to stationary phase. The medium for the following day's cultures, which was the same as that used for the overnight cultures but also contained 40 mM KNO 3 , was also put in the anaerobic chamber at that time. The overnight cultures were used to inoculate the medium containing KNO 3 (1:500 dilution of the inoculum) and were grown for 7 to 8 doublings anaerobically.

Strains, plasmids, and primers used in this study are listed in
Strain construction. Phage transduction was performed with P1 vir as described in reference 49. FLP recombination target (FRT)-flanked kanamycin resistance cassettes were excised from the genome with FLP recombinase expressed from pCP20 (50). Gene deletions from the Keio collection (51) were confirmed by PCR.
To construct JES270, a chloramphenicol-resistant and kanamycin-sensitive ΔptsO strain, a PCR fragment containing FRT-cat-FRT was amplified from pKD3 with primers psyn-u1 and oriR6kseqprim1. The resulting DNA segment was then electroporated into strain JES185/pEL8, and cells were selected on 12 g/ml chloramphenicol to exchange FRT-kan-FRT with FRT-cat-FRT. The resulting strain was confirmed to be kanamycin and ampicillin sensitive, indicating loss of both the kanamycin resistance gene and plasmid pEL8.
Plasmid construction. The insertions in all engineered plasmids were verified to be correct by DNA sequencing.
To construct sixA expression plasmid pSixA, primers NcoI-sixA-F and HindIII-sixA-R were used to amplify the sixA gene from MG1655 genomic DNA. The resulting DNA segment was digested with NcoI and HindIII and cloned into pTrc99a, which was digested with the same enzymes. A plasmid expressing the H8A sixA mutant was constructed by site-directed mutagenesis as follows: plasmid pSixA was amplified with primers sixA-dead-F and sixA-deadCRIM-R2, the ends were phosphorylated with T4 polynucleotide kinase, and the resulting DNA was circularized by blunt-end ligation with T4 DNA ligase, generating pSixA(H8A).
To construct the ycgO expression plasmid pYcgO, the ycgO gene and roughly 500 bp of upstream DNA were amplified from MG1655 genomic DNA with primers EcoRI-PcvrA-F and HindIII-cvrA-R. The resulting PCR product was digested with EcoRI and HindIII and ligated into EcoRI-and HindIII-digested pMG91, a derivative of the single-copy plasmid pSMART.
The plasmid for FLAG-tagged ptsN expression was constructed as follows. The ptsN-3ϫFLAG sequence from Z501 genomic DNA was amplified with primers EcoRI-ptsN-F and HindIII-ptsN-3ϫFLAG-R2. The resulting PCR product was cloned into EcoRI-and HindIII-digested pEB52, a derivative of pTrc99a, generating pEIIA Ntr .
Growth curves. LB Miller medium was inoculated with a single colony and grown to stationary phase with aeration in a roller drum. The overnight culture was then used to inoculate 50 ml of medium with a 1:1,000 dilution, and the flasks were aerated on a shaker. For each time point, three aliquots were removed and the optical density at 600 nm (OD 600 ) of each was measured. For determination of the wild-type strain's growth curve in glycerol minimal medium (Fig. 1A), the exponential-phase and stationary-phase measurements were obtained from two separate experiments. The data for the growth curves presented in Fig. 2A and Fig. 3B were collected in the same experiment, so the curves for MG1655 and JES13 are identical.
Optical density endpoint assay. Our initial experiments with endpoint assays revealed that cells in some cultures were aggregating, which was manifested by the flocculation and sedimentation of standing cultures and was also visible by microscopy. We suspect that spontaneous activation of antigen 43, which is subject to random phase variation and causes autoaggregation when expressed (52, 53), was GACACAGGAACACTTAACGGC PCR-amplify FRT-cat-FRT pBR322-seq-For CCTGCTCGCTTCGCTACTTG Sequence insertions from plasmid library screen pBR322-seq-Rev GCGATATAGGCGCCAGCAAC Sequence insertions from plasmid library screen responsible, since deletion of flu eliminated this flocculation phenotype. The flocculating cultures were not used for endpoint assays because they produced lower optical densities than nonflocculating cultures. LB Miller cultures were inoculated with single colonies and grown overnight to stationary phase with aeration in a roller drum. These cultures were then left standing on the benchtop at room temperature for at least 1 h to monitor for flocculation of the culture. Those tubes that did not show flocculation were then used to inoculate 2-ml cultures of minimal medium (1:1,000 dilution), which were grown with aeration in a roller drum. Incubation times were chosen to catch wild-type cultures in exponential phase.
␤-Galactosidase assays. ␤-Galactosidase assays were performed essentially as described previously (49). To ensure that protein synthesis was stopped at the appropriate times, 100 g/ml streptomycin was added to cultures after they were removed from the anaerobic chamber and placed on ice.
Genetic screens. For transposon mutagenesis, electrocompetent ΔsixA JES13 was transformed with pRL27, which contains a hyperactive Tn5 transposase (54). After electroporation and recovery in SOC medium (containing [per liter] 20 g tryptone, 5 g yeast extract, 0.5 g NaCl, with 20 mM MgSO 4 and 0.4% glucose added after autoclaving) for over 1 h, the cells were incubated in SOC medium with kanamycin for 6 h to select for mutants. Aliquots were then plated on maltose minimal medium agar, and Ͼ16,000 colonies were screened to identify suppressor mutants. Genomic DNA sequences flanking the sites of transposon insertions were identified as described previously (55,56).
For the multicopy suppressor screen, JES13 was transformed with a plasmid library prepared from MG1655 genomic DNA fragments cloned into pBR322 (57) and was screened on minimal agar medium containing several different supplements: (i) glucose with 0.1% Casamino Acids and 15 to 30 mM procaine HCl, (ii) maltose, (iii) glycerol, and (iv) acetate. Over 11,000 colonies were screened to identify plasmids that suppress slow growth. Plasmids were isolated, retransformed into JES13, and tested on maltose minimal medium agar plates to confirm that the plasmid suppressed the slow-growth phenotype. Insert sequences were determined by sequencing with primers pBR322seq-For/pBR322-seq-Rev.
Potassium measurements. LB Miller overnight cultures inoculated from single colonies were used to inoculate 2-ml cultures of glycerol minimal medium, which were grown overnight in a roller drum to optical densities of approximately 0.2. Cultures were harvested by centrifugation through a layer of a 2:1 (vol/vol) mixture of dibutyl phthalate and dioctyl phthalate following the procedure described in reference 13. The cell pellet was suspended in 0.1 ml 1 N nitric acid and digested for 1 to 3 days. The samples were then diluted with 0.95 ml diluent (2% nitric acid, 0.5% HCl) and centrifuged at 13,000 ϫ g for 30 min. One milliliter of supernatant was removed and mixed with 2 ml more diluent. 39 K was measured in the samples and in standard reference controls using an Agilent 7900 inductively coupled plasma mass spectrometer (ICP-MS). The limit of quantification was determined to be 0.1 mg/liter. Potassium content was determined as nanomoles of potassium per OD 600 of the culture prior to harvesting.
EIIA Ntr phosphorylation assay. EIIA Ntr phosphorylation assays were performed similarly to those described previously (14,15,58). Native gels were prepared with 10% 37.5:1 acrylamide and 375 mM Tris HCl (pH 8.8) and cast using a Mini-Protean II system (Bio-Rad). Cultures were grown overnight in LB Miller medium containing ampicillin to stationary phase. Cell pellets from 0.5 ml or 1 ml of culture were suspended in 0.4 ml or 0.8 ml of loading dye (10% glycerol, 40 mM glycine, 5 mM Tris HCl, 0.005% bromophenol blue, pH 8.8) for the experiments whose results are presented in Fig. 3A or Fig. 3F, respectively. The cells were then sonicated on ice and centrifuged to pellet cell debris. Samples were run at 4°C with a running buffer consisting of 25 mM Tris, 192 mM glycine (pH 8.3) until the dye reached the end of the gel.

ACKNOWLEDGMENTS
We thank Manuela Roggiani and Annie Chen for preparing and sharing the genomic DNA plasmid library, Christopher Sikich for assistance in validating strains, Patricia Kiley and Boris Görke for providing strains, and Erica R. McKenzie for processing ICP-MS samples.
This work was supported by a grant from the National Institutes of Health (R01GM080279) and by a National Science Foundation Graduate Research Fellowship under grant no. DGE-1321851.
J.E.S. and M.G. both contributed to the design, execution, analysis, and interpretation of experiments and both also drafted, revised, and approved the manuscript.