Stimulation of the potassium sensor KdpD kinase activity by interaction with the phosphotransferase protein IIANtr in Escherichia coli

Authors

  • Denise Lüttmann,

    1. Georg-August-Universität Göttingen, Abteilung für Allgemeine Mikrobiologie, Institut für Mikrobiologie und Genetik, Grisebachstrasse 8, D-37077 Göttingen, Germany.
    Search for more papers by this author
  • Ralf Heermann,

    1. Ludwig-Maximilians-Universität München, Bereich Mikrobiologie, Großhaderner Strasse 2–4, D-82152 Planegg-Martinsried, Germany.
    2. Munich Center of integrated Protein Science CiPSM, München, Germany.
    Search for more papers by this author
  • Björn Zimmer,

    1. Georg-August-Universität Göttingen, Abteilung für Allgemeine Mikrobiologie, Institut für Mikrobiologie und Genetik, Grisebachstrasse 8, D-37077 Göttingen, Germany.
    Search for more papers by this author
  • Antje Hillmann,

    1. Georg-August-Universität Göttingen, Abteilung für Allgemeine Mikrobiologie, Institut für Mikrobiologie und Genetik, Grisebachstrasse 8, D-37077 Göttingen, Germany.
    Search for more papers by this author
  • Isabella Sofia Rampp,

    1. Ludwig-Maximilians-Universität München, Bereich Mikrobiologie, Großhaderner Strasse 2–4, D-82152 Planegg-Martinsried, Germany.
    Search for more papers by this author
  • Kirsten Jung,

    1. Ludwig-Maximilians-Universität München, Bereich Mikrobiologie, Großhaderner Strasse 2–4, D-82152 Planegg-Martinsried, Germany.
    2. Munich Center of integrated Protein Science CiPSM, München, Germany.
    Search for more papers by this author
  • Boris Görke

    Corresponding author
    1. Georg-August-Universität Göttingen, Abteilung für Allgemeine Mikrobiologie, Institut für Mikrobiologie und Genetik, Grisebachstrasse 8, D-37077 Göttingen, Germany.
      *E-mail bgoerke@gwdg.de; Tel. (+49) 551 393796; Fax (+49) 551 393808.
    Search for more papers by this author

*E-mail bgoerke@gwdg.de; Tel. (+49) 551 393796; Fax (+49) 551 393808.

Summary

Proteins EINtr, NPr and IIANtr form a phosphoryl group transfer chain (Ntr-PTS) working in parallel to the phosphoenolpyruvate:carbohydrate phosphotransferase system (transport-PTS) in Escherichia coli. Recently, it was shown that dephosphorylated IIANtr binds and inhibits TrkA, a low-affinity potassium transporter. Here we report that the Ntr-PTS also regulates expression of the high-affinity K+ transporter KdpFABC, which rescues K+ uptake at limiting K+ concentrations. Transcription initiation at the kdpFABC promoter is positively controlled by the two-component system KdpD/KdpE in response to K+ availability. We found that kdp promoter activity is stimulated by the dephosphorylated form of IIANtr. Two-hybrid data and biochemical analysis revealed that IIANtr interacts with sensor kinase KdpD and stimulates kinase activity, resulting in increased levels of phosphorylated response regulator KdpE. The data suggest that exclusively dephosphorylated IIANtr binds and activates KdpD. As there is cross-talk between the Ntr-PTS and the transport-PTS, carbon source utilization affects kdpFABC expression. Expression is enhanced, when cells utilize preferred carbohydrates like glucose, which results in preferential dephosphorylation of the transport-PTS and also of IIANtr. Taken together, the data show that the Ntr-PTS has an important role in maintaining K+ homeostasis and links K+ uptake to carbohydrate metabolism.

Introduction

Nutrient supply is of prime importance for all living organisms including bacteria. Heterotrophic bacteria such as Escherichia coli rely on the uptake and subsequent metabolism of organic carbon sources to acquire energy and to synthesize all macromolecules of the cell. Hence, all other processes in the cell such as uptake of other nutrients and microelements must be co-ordinated with and adapted to the flux through the central carbohydrate metabolic pathways. However, the regulatory mechanisms underlying this co-ordination are still largely unknown.

In many bacteria the phosphoenolpyruvate (PEP):carbohydrate phosphotransferase system (PTS) represents the primary carbohydrate transport system. Two general phosphotransferases, EI and HPr, transfer phosphoryl groups derived from PEP to the various sugar-specific Enzyme II transporters, which subsequently phosphorylate their substrates during uptake (Postma et al., 1993). In addition, the PTS triggers signal transduction in response to carbohydrate availability by modulating the activities of transporters, enzymes and gene regulators either by protein–protein interaction or by phosphorylation of target proteins (Deutscher et al., 2006). Notably, the PTS is the central processing unit in carbon catabolite repression in many bacteria (Deutscher, 2008; Görke and Stülke, 2008). In enteric bacteria such as E. coli, this regulation is carried out by the IIAGlc subunit of the glucose transporter. In its dephosphorylated form IIAGlc binds and inhibits transporters of less preferred carbon sources such as lactose and galactose, and thereby prevents induction of the corresponding catabolic systems (inducer exclusion). Phosphorylated IIAGlc activates adenylate cyclase that in turn catalyses the synthesis of cAMP, which is a prerequisite for the expression of secondary carbon utilization systems. The phosphorylation state of IIAGlc, and of the PTS in general, is determined by PTS transport activity and the flux through glycolysis. Utilization of less preferred carbon sources yields a high PEP/pyruvate ratio and thereby favours phosphorylation of IIAGlc (Hogema et al., 1998; Bettenbrock et al., 2007). In contrast, this ratio is low and IIAGlc is predominantly dephosphorylated, when a preferred carbohydrate is utilized.

Many bacteria encode paralogues of EI and HPr, which form phosphoryl-transfer chains working in parallel to the canonical PTS (subsequently designated as ‘transport-PTS’). One of them, the so-called Ntr-PTS, is conserved in many proteobacteria (Deutscher et al., 2006). In this system, EINtr (encoded by ptsP) transfers phosphoryl groups to NPr, which finally phosphorylates protein IIANtr (Rabus et al., 1999; Zimmer et al., 2008). IIANtr and NPr are encoded by genes ptsN and ptsO respectively, which are present in the rpoN operon encoding the global nitrogen regulator σ54 (Fig. S1A). Genes ptsN and ptsO are separated by a further gene, yhbJ, whose product controls two small RNAs, which regulate expression of the glmS gene in E. coli (Görke and Vogel, 2008). As there is no obvious phosphoryl group acceptor for IIANtr∼P, it has been speculated that this system has regulatory rather than transport functions.

The role of the Ntr-PTS has been investigated in some detail in Pseudomonaceae. In Pseudomonas putida and in Azotobacter vinelandii, formation of polyhydroxyalkanoates is controlled by the Ntr-PTS (Velazquez et al., 2007; Noguez et al., 2008). Moreover, in P. putida catabolite repression of the xylene and toluene degradation pathways is controlled by IIANtr (Cases et al., 1999; del Castillo and Ramos, 2007). The phosphorylation state of IIANtr appears to play a crucial role in both processes. However, how the mechanism IIANtr exerts its regulatory function in these two examples is unknown. In E. coli, a ptsN mutation generates a pleiotropic phenotype. A ptsN insertion mutant is growth inhibited on Krebs cycle intermediates when alanine or a nucleoside base is the single nitrogen source. The addition of excess NH4+ restores growth (Powell et al., 1995). However, nitrogen control by σ54 was not affected by the mutation leaving the proposed function of this system as a nitrogen (Ntr)-PTS obscure (Reizer et al., 1996). Recently, overexpression of ptsN was found to suppress a lethal σE mutation, suggesting that IIANtr lowers outer membrane stress by a yet unknown mechanism (Hayden and Ades, 2008).

The only direct mechanistic insight comes from a recent study demonstrating that dephosphorylated IIANtr binds to TrkA, a subunit of the Trk K+ transporters and thereby inhibits K+ uptake in E. coli (Lee et al., 2007). In E. coli, the intracellular K+ concentration is kept at approximately 200 mM (Silver, 1996). Under normal conditions, the low-affinity systems Trk and Kup carry out K+ uptake (Silver, 1996). Upon K+ limitation or high osmolarity, K+ uptake is assured by the Kdp P-type ATPase, which is a high-affinity K+ transport system. Kdp consists of four subunits, which are encoded in the kdpFABC operon. KdpA provides the K+ binding site and the K+ translocation channel (Buurman et al., 1995). K+ uptake is driven by the energy-coupling subunit KdpB, which becomes transiently phosphorylated during the transport cycle (Haupt et al., 2005). KdpC and KdpF are required for assembly and stabilization of the transport complex (Gassel et al., 1998; 1999). Expression of the Kdp system is controlled by the two-component system (TCS) KdpD/KdpE, which activates kdpFABC transcription under conditions of K+ limitation or osmotic salt stress (Walderhaug et al., 1992). Upon signal perception, the membrane-bound sensor kinase KdpD autophosphorylates and transfers the phosphoryl-group to the response regulator KdpE (Nakashima et al., 1992; Jung et al., 1997). Phosphorylated KdpE binds to a sequence upstream of the kdpFABC promoter and stimulates transcription initiation (Sugiura et al., 1992). KdpD exhibits both kinase and phosphatase activity towards KdpE (Jung et al., 1997). K+ inhibits the kinase activity of KdpD, whereas increase of the ionic strength stimulates this activity (Jung et al., 2000). Recently, it has been shown that the universal stress protein UspC binds to KdpD and stabilizes the KdpD/KdpE∼P/DNA complex, which leads to enhanced transcription of the kdp operon (Heermann et al., 2009). This mechanism explains how kdpFABC expression can be activated under salt stress, when the intracellular K+ concentration is also far above the normal level.

Here, we report that the Ntr-PTS regulates expression of the kdpFABC operon. We found that dephosphorylated IIANtr enhances activity of the kdp promoter. This is achieved by direct binding of dephosphorylated IIANtr to the sensor kinase KdpD. This interaction strongly stimulates KdpD kinase activity resulting in increased levels of phosphorylated response regulator KdpE, which in turn increases kdpFABC expression. First evidence is provided that this mechanism adjusts the rate of K+ uptake to carbon source utilization.

Results

Overexpression of ptsN leads to accumulation of KdpFABC

During a study dealing with the requirements for the phosphorylation of the IIANtr protein in vivo, the Görke laboratory made the observation that a phosphorylated protein accumulated in the cells in response to overproduction of IIANtr. Various E. coli mutants and transformants, which constitutively overexpressed genes of the rpoN operon from plasmids, were labelled with H3[32P]O4. Phosphorylated proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and analysed by phospho-imaging (Fig. S1B). The chromosomally encoded IIANtr∼P is not detectable by this approach due to its low abundance. However, IIANtr∼P becomes visible, when ptsN was overexpressed from a plasmid (Fig. S1B, compare lanes 1–6 with lanes 7, 11, 12, 13, 17 and 18; Zimmer et al., 2008). Interestingly, in wild-type cells a phosphorylated protein migrating at the top of the gel appeared in response to ptsN overexpression, whereas overexpression of yhbJ and/or ptsO had no effect (Fig. S1B, compare lanes 13, 17, 18 with 14–16). The same result was obtained in a mutant lacking the chromosomally encoded genes ptsN, yhbJ and ptsO (Fig. S1B, lanes 7–12), demonstrating that yhbJ and ptsO are not required for the phosphorylation of IIANtr and for the appearance of the additional phosphorylated protein. Phosphorylation of IIANtr in the absence of ptsO is explained by cross-talk with the canonical PTS as reported recently (Zimmer et al., 2008).

The band containing the labelled protein (Fig. S1B, lane 7) was cut out from the gel and subjected to mass spectrometry. As a control, the corresponding gel piece of the sample of the untransformed Δ[ptsN-O] mutant was used (Fig. S1B, lane 5). This analysis identified peptides of several proteins. Among them KdpB, a subunit of the KdpFABC potassium ion transporter was identified exclusively in the sample containing the labelled protein band. In order to clarify whether KdpB∼P accumulates in response to IIANtr overproduction, a ΔkdpB ΔptsN double mutant was generated and transformed with a plasmid expressing ptsN under Para promoter control. In the wild-type strain, the presence of this plasmid led to accumulation of IIANtr∼P and of the additional phospho-protein, when arabinose was added as inducer for ptsN expression (Fig. 1A, compare lane 4 with lanes 1–3). However, in the ΔkdpB ΔptsN mutant the additional phosphorylated protein did not accumulate anymore in response to ptsN overproduction, although IIANtr∼P was readily observed (Fig. 1A, compare lanes 7 and 4). The introduction of a compatible plasmid, which likewise expresses kdpB under control of the Para promoter, restored accumulation of the high-molecular-weight phospho-protein in the presence of arabinose, confirming that this protein is KdpB∼P (Fig. 1A, compare lanes 8 and 9 with 6 and 7). To demonstrate unequivocally the identity of KdpB, a Western blot analysis was performed using a polyclonal αKdpFABC antiserum (Heermann et al., 2009; Fig. 1B). As KdpF and KdpA are extremely hydrophobic, only KdpB (molecular weight = 72.2 kDa) and KdpC (molecular weight = 20.3 kDa) are detectable using this antiserum. Indeed, plasmid-driven overexpression of ptsN in the wild-type strain resulted in the accumulation of KdpFABC (Fig. 1B, compare lane 4 with lanes 1–3). In the ΔkdpB ΔptsN double mutant, neither KdpB nor KdpC was visible upon overexpression of ptsN (Fig. 1B, lanes 5–7). In the complemented strain, which expresses kdpB from a plasmid, KdpB became detectable again (Fig. 1B, lanes 8 and 9). KdpC could not be detected in this case, presumably because the kdpB deletion prevents expression of the downstream located kdpC gene due to a polar effect. Taken together, the data indicate that overproduction of IIANtr stimulates synthesis of the KdpFABC potassium transporter.

Figure 1.

Overproduction of IIANtr induces accumulation of phosphorylated KdpB. Strains R1279 (wild-type) and Z123 (ΔptsN ΔkdpB) carrying plasmid pBGG98 (kdpB on pBAD33) and/or plasmid pBGG86 (ptsN on pBAD18) as indicated were grown in the absence or presence of arabinose as inducer for gene expression and subsequently analysed by in vivo protein phosphorylation (A) and Western blotting using an αKdpFABC antiserum (B).
A. The transformants were labelled with H3[32P]O4 and phosphorylated proteins were separated by SDS-PAGE and analysed by phospho-imaging. Only the relevant parts of the gel are shown.
B. Detection of KdpFABC in cell extracts separated by SDS-PAGE and visualized by Western blotting.

A transposon mutagenesis screen for mutants with constitutive kdpFABC expression identifies insertions in ptsP

Using a different approach, the Jung laboratory obtained results, which likewise indicated that expression of the kdpFABC operon is affected by the ‘Ntr-PTS’. In these experiments E. coli strain RH002 carrying a chromosomal kdpFA′-lacZ reporter gene fusion was subjected to transposon mutagenesis and screened for insertions leading to increased synthesis of β-galactosidase and therefore kdpFABC expression in minimal medium containing 5 mM K+, i.e. conditions under which kdpFABC expression is only weakly induced in this strain (Fig. S2). Two independent transposon mutants were isolated out of 9000 tested, in which lacZ expression was significantly elevated. Both insertions mapped in gene ptsP, which encodes EINtr, the cognate phosphotransferase that initiates the phosphorylation cascade leading to phosphorylation of IIANtr. Quantitative determination of β-galactosidase activities in the presence of different K+ concentrations showed that the Tn10 insertions increased kdpFA′-lacZ expression at all tested K+ concentrations (Fig. S2).

IIANtr stimulates kdpFABC expression

The data indicated that the ‘Ntr-PTS’ somehow affects the cellular concentration of the proteins encoded in the kdpFABC operon. As IIANtr triggers the output in this phosphorelay system, we investigated its role in more detail. First, we analysed the effects of deletion and overexpression of ptsN on the KdpFABC induction profile. To this end, the bacteria were grown in minimal media containing various K+ concentrations and cell extracts were subjected to Western analysis using the αKdpFABC antiserum. As expected from previous data, KdpFABC was detectable in the wild-type strain exclusively in the presence of low K+ concentrations ≤ 20 mM, and the KdpFABC level increased with decreasing K+ concentrations (Fig. 2A, wild-type). However, in the ΔptsN mutant subunits of the KdpFABC complex were only detectable when cells were grown in media containing 1 mM and 5 mM K+. In addition, their concentration appeared to be reduced in comparison with the wild-type strain grown under the same conditions (Fig. 2A). Overexpression of ptsN caused the opposite effect. Production of KdpFABC was drastically increased compared with the wild-type strain, and the complex was detectable under all tested conditions, although its concentration decreased with increasing K+ concentrations (Fig. 2A).

Figure 2.

IIANtr increases KdpFABC production in a TrkA-independent manner.
A. Detection of KdpFABC in strain R1279 (wild-type), Z79 (ΔptsN) and strain Z79 carrying plasmid pBGG86 (ptsN on pBAD18) by Western analysis. Cells were grown in minimal medium containing different K+ concentrations as indicated. To induce expression of ptsN from plasmid pBGG86, arabinose was added. Cell extracts were separated by SDS-PAGE and KdpFABC production was determined using αKdpFABC antiserum.
B. The same experiments as in A but carried out with strains Z166 (ΔtrkA) and Z167 (ΔtrkA ΔptsN).

Next, we wanted to determine whether IIANtr affects the kdp operon at the transcriptional level. For this purpose a transcriptional kdpFA′-lacZ fusion encompassing the kdp promoter and its regulatory region was placed on a low copy plasmid. This plasmid was introduced into the wild-type strain, the ΔptsN mutant and the ptsN overexpressing strain. Cells were grown in minimal media containing different K+ concentrations and the β-galactosidase activities were determined. Indeed, expression of the kdpFA′-lacZ fusion was significantly reduced in the ΔptsN mutant, whereas it was much higher expressed upon overproduction of IIANtr (Table 1). Taken together, these data show that the IIANtr-mediated upregulation of KdpFABC production is caused by higher kdpFABC transcript levels.

Table 1.  Regulation of kdpFABC expression by IIANtr.
K+ (mM)β-Galactosidase activity (Miller units)
Wild type
(R1279)
ΔptsN
(Z79)
ΔptsN + ptsN on pBAD18
(Z79/pBGG86)a
ΔtrkA
(Z166)
ΔtrkA, ΔptsN
(Z167)
ΔtrkA, ΔptsN+ptsN on pBAD18 (Z167/pBGG86)a
  • a. 

    For the induction of ptsN expression arabinose was added.

  • β-Galactosidase activities of strains carrying a transcriptional kdpFA′-lacZ fusion on plasmid pBGG158. Strain and plasmid designations are in parentheses.

0.12848 ± 8242048 ± 8211 431 ± 1 3212851 ± 8392141 ± 339791 ± 1717
11594 ± 227766 ± 2010 491 ± 8141470 ± 1061234 ± 538978 ± 1515
5229 ± 11081 ± 15 599 ± 906880 ± 67369 ± 1286801 ± 960
1042 ± 1923 ± 25 489 ± 1178629 ± 19110 ± 465228 ± 638
2014 ± 612 ± 13 055 ± 1110323 ± 6069 ± 164868 ± 822
4011 ± 510 ± 13 214 ± 894151 ± 8726 ± 184344 ± 993
808 ± 410 ± 13 352 ± 31113 ± 319 ± 42196 ± 286
1158 ± 38 ± 32 961 ± 2148 ± 113 ± 72213 ± 267

Regulation of kdpFABC expression by IIANtr is independent from IIANtr-mediated inhibition of the Trk potassium transporter

It has recently been shown that IIANtr forms a complex with the low-affinity TrkA potassium transporter and thereby inhibits the intracellular accumulation of K+ (Lee et al., 2007). Therefore, it appeared conceivable that the modulation of kdpFABC expression by IIANtr was the indirect consequence of the IIANtr-mediated inhibition of the Trk K+ transporter. The absence of IIANtr could increase K+ uptake by Trk and thereby inhibit induction of kdpFABC expression, whereas its overexpression might limit Trk-mediated K+ transport, which would lead to induction of kdpFABC expression even at high K+ concentrations. To investigate this possibility, we checked the effects of ptsN deletion and overproduction on kdpFABC expression in a ΔtrkA mutant. To this end, we analysed KdpFABC production as well as expression of the kdpFA′-lacZ fusion at different K+ concentrations (Fig. 2B and Table 1). As expected from previous data (Laimins et al., 1981), the trkA mutation shifted induction of the kdpFABC operon towards higher K+ concentrations (compare lanes 1–7 in Fig. 2A and B; compare wild-type and ΔtrkA in Table 1). Notably, the ΔptsN mutation decreased expression of kdpFABC as well as KdpFABC production also in the trkA mutant background (compare ΔtrkA and ΔtrkA ΔptsN in Fig. 2B and in Table 1 respectively). Moreover, plasmid-driven overexpression of ptsN drastically enhanced kdpFABC expression and KdpFABC production also in the ΔtrkA mutant (Fig. 2B and Table 1). In conclusion, stimulation of kdpFABC expression by IIANtr is independent from its inhibitory effect on the Trk transporter.

The dephosphorylated form of IIANtr upregulates kdpFABC expression

It has been shown that EINtr and NPr control, at least in part, the phosphorylation state of IIANtr (Rabus et al., 1999; Zimmer et al., 2008). The finding that a Tn10 insertion in ptsP enhanced expression of a kdpFA′-lacZ fusion raised the possibility that the phosphotransferases EINtr and NPr affect kdpFABC expression by controlling the phosphorylation state of their cognate phosphoryl group acceptor IIANtr. To explore the roles of EINtr and NPr in more detail, we determined expression of the kdpFA′-lacZ fusion in ΔptsO as well as ΔptsP mutants in the presence of different K+ concentrations (Fig. 3). Indeed, expression of kdpFABC was significantly enhanced in both mutants when compared with the wild type. As phosphorylation of IIANtr is diminished in ptsP as well as ptsO mutants (Zimmer et al., 2008), these results suggest that non-phosphorylated IIANtr may stimulate kdpFABC expression. In addition, a Δ[ptsN-O] mutant lacking both ptsO and ptsN was tested. Expression of the kdpFA′-lacZ fusion in this strain was comparable to the ptsN single mutant (Fig. 3), verifying that the stimulatory effect of the ptsO mutation on kdp expression relies on ptsN.

Figure 3.

Roles of EINtr, NPr and IIANtr in regulation of kdpFABC expression. β-Galactosidase activities of strains R1279 (wild-type), Z79 (ΔptsN), R2404 (ΔptsP), Z80 (ΔptsO) and R2413 (Δ[ptsN-ptsO]) carrying a transcriptional kdpFA′-lacZ reporter gene fusion on plasmid pBGG158. Cells were grown in minimal medium containing different K+ concentrations as indicated.

To further corroborate the hypothesis that non-phosphorylated IIANtr stimulates expression of the kdp operon, we used mutant ptsN alleles, which encode IIANtr variants with amino acid exchanges of the His73 phosphorylation site. In vivo, phosphorylation of these mutant IIANtr proteins is prevented (Zimmer et al., 2008). Plasmids carrying these mutant ptsN alleles under Para control were introduced into the ΔptsN mutant. With these strains, two types of analyses were carried out. On the one hand, we analysed whether expression of the mutant ptsN alleles still results in accumulation of phosphorylated KdpB (Fig. 4A). In parallel, we checked the KdpFABC amount at different K+ concentrations by Western blotting (Fig. 4B). Both the expression of the ptsN-H73A and that of the ptsN-H73D allele resulted in accumulation of KdpB∼P to much higher concentrations than those observed upon overexpression of wild-type ptsN (Fig. 4A, compare lanes 5–8 with lanes 3 and 4). This result was confirmed at the protein level. Expression of the ptsN-H73A and ptsN-H73D alleles caused strong accumulation of KdpFABC at all K+ concentrations, and its concentration was always higher in comparison with the strain expressing wild-type ptsN (Fig. 4B).

Figure 4.

Non-phosphorylatable mutant IIANtr proteins strongly enhance KdpFABC production.
A. In vivo protein phosphorylation of strain Z79 (ΔptsN) encoding different IIANtr variants on plasmids. Strain Z79 carried either plasmid pBGG86 (wt-ptsN; lanes 3, 4), pBGG93 (ptsN-H73A; lanes 5, 6), pBGG92 (ptsN-H73D; lanes 7, 8) or the empty expression vector pBAD18 as control (lanes 1, 2). Arabinose as inducer for ptsN expression was added when indicated. Phosphorylated proteins were separated by SDS-PAGE and analysed by phospho-imaging.
B. Detection of KdpFABC by Western analysis in strain Z79 carrying the plasmids described in A. Cells were grown in minimal medium containing different K+ concentrations as indicated. Arabinose was present for the induction of ptsN expression. Cell extracts were separated by SDS-PAGE and KdpFABC production was visualized by Western blotting using αKdpFABC antiserum. Only the relevant parts of the gels are shown.

To confirm that the IIANtr derivatives exert their effects at the level of kdpFABC expression, the plasmid carrying the kdpFA′-lacZ reporter fusion was introduced into the strains used in the experiments above. These cells harbouring two respective plasmids were grown in the presence of different K+ concentrations and the β-galactosidase activities were determined (Table S1). Indeed, expression of the ptsN-H73A and ptsN-H73D alleles resulted in higher β-galactosidase activities in comparison with wild-type ptsN. In particular, expression of the reporter fusion was not anymore repressed by high K+ concentrations in the presence of these mutant ptsN alleles. In sum, these results demonstrate that the dephosphorylated form of IIANtr stimulates expression of the kdpFABC operon.

IIANtr stimulates kdpFABC promoter activity

The next most important question was how to determine the molecular mechanism by which IIANtr enhances kdpFABC expression. The data obtained with the kdpFA′-lacZ reporter fusion indicated that IIANtr increases kdpFABC expression. However, so far a post-transcriptional regulation could not be ruled out. A post-transcriptional mechanism, e.g. by base-pairing of a small RNA with the kdp leader region, appeared imaginable because ptsN maps next to yhbJ, whose product regulates two small RNAs in the cell (Görke and Vogel, 2008). To discriminate between a transcriptional and a post-transcriptional mechanism, we replaced the kdpFABC promoter region in the kdpFA′-lacZ fusion by the constitutive Ptac promoter leaving the untranslated leader region unaffected. This construct was introduced into the wild-type strain and the ΔptsN mutant. Cells were grown at different K+ concentrations and the β-galactosidase activities were determined. The obtained data show that regulation of the Ptac-kdpFA′-lacZ fusion was completely lost (Table 2). Neither the K+ concentration nor the ΔptsN deletion had any effect on kdpFA′-lacZ expression. To corroborate these data we also tested the effect of the hfq-1 mutation on KdpFABC production. Hfq is involved in many post-transcriptional regulatory events and required for many small RNAs to act. However, the hfq-1 mutation had no effect on KdpFABC production at any K+ concentration. It did not affect KdpFABC concentrations in either a ptsN+ or a ΔptsN mutant background and it had also no effect in a IIANtr-overproducing strain (data not shown). Taken together, the data exclude that IIANtr acts post-transcriptionally or on transcription elongation. Hence, transcription initiation at the kdp promoter must be controlled by IIANtr.

Table 2.  Replacement of the kdp- with the tac-promoter results in loss of regulation by IIANtr.
K+ (mM)β-Galactosidase activity (Miller units)
Wild type (R1279)ΔptsN (Z79)
  1. β-Galactosidase activities of strains carrying a transcriptional Ptac-kdpFA′-lacZ fusion on plasmid pBGG180. Strain designations are in parentheses.

0.17997 ± 333310 217 ± 2 833
17143 ± 402511 098 ± 3 255
56893 ± 27889 162 ± 2 252
106421 ± 24228 671 ± 2 803
207930 ± 30659 109 ± 2 015
407709 ± 41279 502 ± 325
806965 ± 38597 641 ± 962
1158142 ± 35328 708 ± 2 092

IIANtr interacts with sensor kinase KdpD

IIANtr lacks any DNA binding domain and therefore direct binding to the kdpFABC promoter region appeared unlikely. Moreover, IIANtr inhibits the Trk K+ transporter by direct interaction with TrkA. Similarly, the IIANtr paralogue IIAGlc exerts its regulatory functions by binding to protein partners. Therefore, we tested whether IIANtr interacts with the histidine kinase KdpD or the response regulator KdpE, which control expression of the kdp operon in response to K+. To this end, we used a bacterial two-hybrid system. In this system the proteins of interest are fused to the T25 and T18 fragments of Bordetella pertussis adenylate cylase and the fusion proteins are expressed in an E. coli strain lacking its cognate adenylate cyclase. Interaction of the hybrid proteins restores adenylate cyclase activity and allows expression of cAMP-CRP-dependent reporter genes such as lacZ. Hence, protein–protein interaction can be detected in β-galactosidase assays. Therefore, bacteria harbouring the two-hybrid constructs were grown in Luria–Bertani (LB) and samples were harvested in the exponential growth phase as well as during transition to stationary phase and the β-galactosidase activities were subsequently determined. First of all, we tested already known interactions. Indeed, the assays revealed dimerization of the KdpD sensor kinase as well as interaction between KdpD and KdpE as compared with the background controls and the positive control (Fig. 5, compare columns 5 and 6 with 1–4). Intriguingly, the data demonstrate that IIANtr interacts with KdpD, whereas no interaction was detectable between IIANtr and KdpE (Fig. 5, columns 9 and 7). Next, the non-phosphorylatable IIANtr-H73A mutant was tested. Again, there was no interaction detectable between this mutant protein and KdpE (Fig. 5, columns 8). However, the data revealed a strong interaction between IIANtr-H73A and KdpD (Fig. 5, columns 10). These results are in perfect agreement with our results demonstrating a much stronger enhancement of kdpFABC expression by IIANtr-H73A than by wild-type IIANtr (Fig. 4 and Table S1).

Figure 5.

IIANtr interacts with sensor kinase KdpD. Protein–protein interaction was analysed using a bacterial two-hybrid system. For this purpose, the proteins of interest were fused to the N- or C-terminal ends of the T25 and T18 fragments of B. pertussis adenylate cyclase as indicated in the figure. Strain BTH101 was co-transformed with plasmids encoding the various T25- and T18-hybrid proteins respectively. In columns 2 and 3, strain BTH101 carried plasmids pKT25 + pUT18 and pKNT25 + pUT18C respectively. These cells expressed the unfused T25- and T18-fragments and served as negative controls, in addition to the untransformed strain (columns 1). Cells, which produced T25- and T18-fragments fused to the leucine zipper of transcription factor GCN4, were used as positive controls (columns 4). Cells were grown in LB supplemented with 0.5 mM IPTG for the induction of expression of the fusion genes and harvested at OD600 of 0.5 and 1.3 respectively. Subsequently, the β-galactosidase activities were determined, which reflect the strength of interaction of the fusion proteins.

The data do not exclude the possibility that IIANtr is generally able to interact with members of the protein family of sensor kinases. Therefore, we tested a possible interaction of IIANtr with sensor kinase AtoS, which regulates the expression of genes involved in short-chain fatty acid catabolism in E. coli (Kyriakidis and Tiligada, 2009). Similar to KdpD, AtoS carries its cytoplasmic transmitter domain at the C-terminus. As can bee seen from the data, no interaction was detectable between AtoS and IIANtr or the IIANtr-H73A mutant (Fig. 5, columns 11 and 12). Hence, interaction of IIANtr appears to be restricted to KdpD and does not apply to sensor kinases in general.

IIANtr increases the phosphorylation of KdpD and KdpE in vitro

Our data demonstrated that IIANtr enhances kdpFABC transcription and interacts with the sensor kinase KdpD, which controls together with the response regulator KdpE activity of the corresponding promoter. Therefore, the most likely explanation was that IIANtr modulates KdpD/KdpE signalling, resulting in increased levels of phosphorylated response regulator KdpE. To prove this hypothesis, we tested the effect of purified IIANtr protein on the in vitro reconstructed KdpD/KdpE phosphorylation cascade. Recombinant IIANtr provided with an N-terminal deca-His-tag was overproduced and purified to homogeneity (Fig. S3A and B). Functionality of 10His-IIANtr was verified in vivo (Fig. S3D), indicating that the 10His-Tag has no influence on IIANtr activity. Next, proteoliposomes containing purified KdpD-6His were prepared and mixed with purified 10His-KdpE in a molar ratio of these proteins of 1:4. Additionally, the reaction mixture contained DNA fragments comprising the KdpE-binding site. Phosphorylation was started by adding a mixture of radioactive ATP and ADP, which reflects the ratio found in vivo (Ohwada and Sagisaka, 1987). As it is known that phosphorylation of the in vitro reconstructed KdpD/KdpE signalling cascade is inhibited by K+ (Heermann et al., 2009), the influence of IIANtr on the phosphorylation cascade was tested at different K+ concentrations (Fig. 6A). Intriguingly, the presence of purified IIANtr stimulated the phosphorylation cascade at all tested K+ concentrations, resulting in a stable phosphorylation of KdpE. As a result, IIANtr increased the phosphorylation level of KdpD/KdpE twofold (no K+), threefold (50 mM K+) and 10-fold (250 mM K+) respectively. To see if the phosphorylation state of IIANtr is important for triggering phosphorylation of KdpD/KdpE, we also tested the purified non-phosphorylatable IIANtr-H73A variant (Fig. S3C). As it can be seen in Fig. 6B, addition of IIANtr-H73A enhanced the phosphorylation level of KdpD/KdpE 10-fold in the presence of 250 mM K+. These data show that non-phosphorylated IIANtr stimulates phosphorylation of KdpD/KdpE.

Figure 6.

IIANtr (A) and IIANtr-H73A (B) stimulate phosphorylation of KdpD and KdpE in vitro. Purified KdpD-6His in proteoliposomes was mixed with 10His-KdpE and DNA fragments comprising the KdpE-binding site in buffer containing the indicated K+ concentrations. When indicated, purified 10His-IIANtr or 10His-IIANtr-H73A, respectively, was added. The phosphorylation cascade was started at time 0 by adding a mixture of [γ32P]-ATP and ADP in a molar ratio of 8:1. At the indicated times, the reactions were stopped, the samples were subjected to SDS-PAGE, and phosphorylated KdpD and KdpE were detected by phospho-imaging and quantified. The autoradiographs are representative of three independently performed experiments.

IIANtr modulates kinase rather than phosphatase activity of KdpD

So far, the experiments revealed that IIANtr enhances phosphorylation of KdpD and KdpE (Fig. 6). This could result either from a stimulation of the kinase activity or alternatively from the inhibition of the phosphatase activity of KdpD. To discriminate between these possibilities, we tested the effect of IIANtr on these activities of KdpD in vitro in the presence of 250 mM K+. To test kinase activity, KdpD was first phosphorylated in the absence or presence of IIANtr for 3.5 min and subsequently non-phosphorylated KdpE protein was added to the reaction mixture. Indeed, presence of IIANtr strongly enhanced autophosphorylation of KdpD (i.e. 10-fold, when compared at t = 2 min) and the subsequent phospho-transfer to KdpE (Fig. 7A). To test the effect of IIANtr on KdpD phosphatase activity, phosphorylated KdpE protein was isolated and subsequently inverted membrane vesicles containing KdpD were added in the absence or presence of IIANtr. Previous control experiments have shown that dephosphorylation of KdpE∼P depends on the presence of KdpD in these assays (Jung and Altendorf, 1998a). The data show that IIANtr had only a minor effect on the dephosphorylation of KdpE, regardless of the tested K+ concentration (Fig. 7B). Quantification revealed that the KdpE∼P signal was twofold enhanced in the presence of IIANtr. In conclusion, interaction with IIANtr predominantly increases the kinase activity of KdpD.

Figure 7.

IIANtr increases the kinase activity of KdpD (A) but has only a minor effect on KdpD phosphatase activity (B).
A. Inverted membrane vesicles containing KdpD-6His were incubated with [γ32P]-ATP in the presence of 250 mM K+ with or without 10His-IIANtr. After 3.5 min, equimolar concentrations of 10His-KdpE were added. Aliquots were removed at the indicated times (before and after addition of KdpE) and the reaction was stopped. Proteins were tested in equimolar concentrations.
B. 10His-KdpE∼32P was incubated in the presence or absence of 250 mM K+ with inverted membrane vesicles containing KdpD and 10His-IIANtr when indicated (equimolar concentrations to KdpE). At the indicated times, aliquots were removed and the reaction were stopped. All samples in A and B were subjected to SDS-PAGE and phosphorylated proteins were detected by phospho-imaging and quantified. The autoradiographs are representative for three independently performed experiments.

The phosphorylation state of IIANtr links KdpFABC-mediated K+ uptake to carbohydrate utilization

We have previously shown that there is cross-talk between the ‘Ntr-PTS’ and the canonical transport-PTS in vivo (Zimmer et al., 2008). Therefore, the phosphorylation state of IIANtr is determined in part also by the transport-PTS. If this is the case, the phosphorylation state of the transport-PTS should affect expression of kdpFABC and thus K+ uptake. The phosphorylation state of the canonical PTS is determined by the carbon source that is taken up and utilized. The utilization of preferred carbon sources allowing for high growth rates leads to preferential dephosphorylation of the PTS proteins, whereas these proteins are predominantly phosphorylated when cells metabolize less preferred carbohydrates (Deutscher et al., 2006; Bettenbrock et al., 2007). In order to determine whether the nature of the utilized carbon source affects expression of the kdpFABC operon, we integrated the kdpFA′-lacZ fusion into the chromosomes of the wild-type strain and of the ΔptsN mutant. β-Galactosidase activities exhibited by these strains when grown at different K+ concentrations verified that the kdpFA′-lacZ fusion is considerably lower expressed in the ΔptsN mutant (Fig. S4), reflecting well the data obtained before with the plasmid-borne reporter system (Fig. 3, Table 1). Next, we grew the strains in minimal media, which were supplemented with different carbohydrates but contained a fixed K+ concentration of 5 mM. The data in Fig. 8 show that expression of the kdpFA′-lacZ fusion indeed varied with the carbon source. In the wild-type strain the kdpFA′-lacZ fusion was significantly higher expressed, when the cells grew on carbohydrates like glucose, mannitol or GlcNAc in comparison with substrates such as galactose, GlcN, glycerol and mannose. This correlated well with the previously determined phosphorylation state of IIAGlc (Bettenbrock et al., 2007), which reflects phosphorylation of the canonical phosphotransferases EI and HPr (Deutscher et al., 2006). Upon growth on glucose, mannitol and GlcNAc, IIAGlc is considerably less phosphorylated (5–17% IIAGlc∼P) than on the other tested substrates (47–64% IIAGlc∼P). No significant differences in kdpFA′-lacZ expression were detectable in the ΔptsN mutant, suggesting that IIANtr is required for the carbon source-dependent differences in kdp operon expression. To verify that the differences of kdpFA′-lacZ expression in the presence of the different carbohydrates are indeed determined by the phosphorylation states of the sugar phosphotransferases EI and HPr, we determined the influence of different carbon sources on kdpFABC expression in a ΔptsP mutant lacking EINtr (Fig. 8). The data show that kdpFABC was higher expressed in this mutant in comparison with the wild type, supporting the data obtained with the plasmid-based reporter system (Fig. 3). Notably, kdpFABC expression in the ΔptsP mutant showed a similar carbon source-dependent pattern as observed with the wild-type strain.

Figure 8.

Carbon source utilization modulates kdpFABC expression in an IIANtr-dependent manner. β-Galactosidase activities of strains Z267 (wild type), Z268 (ΔptsN) and Z269 (ΔptsP) carrying a transcriptional kdpFA′-lacZ reporter gene fusion ectopically integrated into the chromosome. Cells were grown in minimal medium containing 5 mM K+ and supplemented with the carbon source as indicated. The fractions of phosphorylated IIAGlc protein, as previously determined by Bettenbrock et al. (2007), are shown for comparison.

As the differences in kdpFA′-lacZ expression in response to the different carbon sources were only two- to three-fold, we decided to confirm these data using a different reporter system. For this purpose, we used E. coli strain HAK006, which carries a chromosomal kdpFA′-lacZ reporter gene fusion (Nakashima et al., 1993). This strain lacks the kdpFABC operon as well as kdpD, but carries a functional kdpE gene. Transformation of this strain with plasmid pBD carrying kdpD under Para control leads to restoration of the KdpD/KdpE signalling cascade (Heermann et al., 2003). This strain was grown on different carbohydrates and kdpFA′-lacZ expression was monitored. The results correlated well with the data obtained before. Expression of kdpFABC was two- to three-fold higher when cells utilized glucose, mannitol or GlcNAc in comparison with the other substrates (Fig. S5). Deletion of ptsN in strain HAK006 carrying the kdpD-containing pBD plasmid led to very low β-galactosidase activities, verifying the positive effect of IIANtr on the kdpFABC operon in this second strain background also. In sum, the data suggest that IIANtr links kdpFABC expression to the phosphorylation state of the transport-PTS (see model in Fig. 9).

Figure 9.

Model for the carbon source-dependent regulation of kdpFABC transcription by IIANtr. Dephosphorylated IIANtr binds to sensor kinase KdpD and stimulates its kinase activity. This increases phosphorylation of response regulator KdpE, which enhances transcription initiation at the kdpFABC promoter and thereby synthesis of the high-affinity K+ transporter. As a result, uptake of K+ is enhanced. The phosphorylation state of IIANtr is determined, at least partially, also by the phosphorylation state of the transport-PTS, because there is cross-talk between both PTS systems in vivo (Zimmer et al., 2008). Hence, utilization of a preferred carbon source, such as glucose, leads to preferential dephosphorylation of the transport-PTS and thereby increases the amount of dephosphorylated IIANtr, which stimulates expression of the kdpFABC operon. In addition, phosphorylation of IIANtr might be influenced by an unknown signal (marked with ?) perceived by the GAF domain that could modulate autophosphorylation of EINtr.

Discussion

The regulatory function of the so called nitrogen-PTS remained enigmatic for a long time. Originally, a role of this system in nitrogen regulation was assumed (Powell et al., 1995), basically because the genes for IIANtr, NPr and σ54 colocalize in the same operon. The putative role of the Ntr-PTS as a system devoted to nitrogen regulation became questionable by a recent study demonstrating that IIANtr controls activity of the low-affinity K+ transporter TrkA by direct binding (Lee et al., 2007). Here, we show that the Ntr-PTS controls expression of a further K+ uptake system, the high-affinity K+ transporter KdpFABC. Therefore, evidence accumulates that the Ntr-PTS is a more global system that connects carbohydrate metabolism and adjustment of K+ uptake.

We demonstrate that IIANtr positively regulates activity of the kdpFABC promoter. This regulation still works in a trkA mutant background and is therefore independent from the Trk transporters. Bacterial two-hybrid assays demonstrate that IIANtr interacts with KdpD, the sensor kinase of the cognate TCS that controls transcription initiation at the kdpFABC promoter. In vitro experiments using KdpD reconstituted in proteoliposomes and purified response regulator KdpE showed that phosphorylation of both proteins is strongly enhanced in the presence of IIANtr. Further analysis revealed that IIANtr predominantly stimulates kinase activity, whereas phosphatase activity of KdpD is only slightly affected. The data suggest that IIANtr enhances autophosphorylation activity of KdpD (Fig. 9). To our knowledge, this is the first report that a PTS protein regulates activity of a sensor kinase by direct interaction.

Several lines of evidence indicate that the dephosphorylated form of IIANtr regulates kdpFABC expression. First, in mutants lacking EINtr or NPr, kdpFABC transcription is considerably increased. In both mutants, phosphorylation of IIANtr is reduced (Zimmer et al., 2008). Second, non-phosphorylatable IIANtr mutants carrying Ala- or Asp-exchanges of the active site His73 increase kdpFABC expression to much higher levels than the wild-type protein does. Third, the IIANtr-H73A mutant protein interacts with KdpD and stimulates phosphorylation of the TCS in vitro much stronger than the wild-type protein. Interestingly, the Ala- and Asp- exchanges of His73 in IIANtr yielded the same effects. Thus, in contrast to other proteins of the PTS, such as HPr from Bacillus subtilis (Deutscher et al., 1994), phosphorylation of His73 in IIANtr cannot be mimicked by a negatively charged Asp residue. In comparison with IIA proteins of the transport-PTS, the active centre of IIANtr shows pronounced structural differences (van Montfort and Dijkstra, 1998), which might explain this result.

It is noteworthy that both processes, inhibition of TrkA at high external K+ concentrations and stimulation of kdpFABC at limiting K+ concentrations, are carried out by the dephosphorylated form of IIANtr. Hence, apparently dephosphorylated IIANtr helps to avoid deleterious conditions of K+ surplus as well as depletion in the cell, suggesting that IIANtr is important for K+ homeostasis. It has been shown for E. coli as well as for P. putida, that cross-phosphorylation occurs between the transport-PTS and the Ntr-PTS (Pflüger and de Lorenzo, 2008; Zimmer et al., 2008). Therefore, the phosphorylation state of IIANtr is determined by the activities of both systems. Consequently, as shown here, expression of the kdpFABC operon is linked to the available carbon source. Utilization of a preferred carbohydrate, such as glucose, triggers the preferential dephosphorylation of the transport-PTS, and also of IIANtr. As a result, the kdpFABC operon is higher expressed as compared with less preferred carbon sources (Fig. 9). This implies that cells take up more K+, when a rich carbon source is utilized. What is the biological meaning of the higher K+ requisition in this case? Potassium is the major monovalent intracellular cation in E. coli and all other organisms (Silver, 1996). It is required for maintenance of intracellular osmolarity, cell turgor and for pH homeostasis. Intriguingly, many enzymes require K+ for activity [reviewed in Di Cera, 2006; Page and Di Cera, 2006]. In bacteria, activity of S-adenosylmethionine synthase and of chaperone GroEL requires K+. For other bacterial enzymes, such as ribokinase, tryptophanase and tyrosinase, K+ is an activating allosteric effector. It is tempting to speculate that under conditions of supply with energy-rich carbon sources, cells require more K+ to meet the requirements of higher cellular enzymatic activities. The connector between these two processes seems to be IIANtr. The phosphorylation state of IIANtr modulates induction of the high-affinity KdpFABC transporter as well as the activity of the low-affinity transporter Trk. In addition to cross-phosphorylation of IIANtr, there could be an additional signal which is perceived by EINtr. EINtr contains an N-terminal GAF domain. GAF domains were shown to bind signal molecules such as nitric oxide or cyclic nucleotides (Hurley, 2003; D'Autreaux et al., 2008). It remains open whether the GAF domain of EINtr plays a regulatory role in response to a yet unidentified signal.

The two-hybrid analysis detected interaction of IIANtr with KdpD. KdpD consists of N- and C-terminal cytoplasmic domains, which are connected by four transmembrane helices (Zimmann et al., 1995). The C-terminal domain contains the transmitter domain of sensor kinases including the site of autophosphorylation His673. In addition, adjacent to transmembrane helix IV, a putative K+ binding site might be present in this domain (Rothenbücher et al., 2006; Zimmann et al., 2007). It is important to note that K+ has an inhibitory effect on the kinase activity of KdpD (Jung et al., 2000; Heermann et al., 2009). The N-terminal cytoplasmic domain contains a regulatory ATP binding site and is important for a regulated phosphatase activity of KdpD (Heermann et al., 2000). Moreover, this domain stabilizes binding of KdpE to its target site on the DNA and contains the UspC interaction surface (Heermann et al., 2003; 2009). In this respect, it is tempting to speculate that IIANtr binds to the C-terminal domain of KdpD and may either block the presumptive K+ binding site or stimulate the autophosphorylation activity.

IIANtr can be assigned to an emerging class of proteins, termed TCS connectors, which regulate the activity of TCSs (Mitrophanov and Groisman, 2008). In Bacillus subtilis, the phosphorelay regulating initiation of sporulation is subject to extensive regulation by TCS connectors. Several proteins can inhibit autophosphorylation of the cognate sensor kinase KinA, whereas others stimulate dephosphorylation of the response regulator Spo0A. The PmrD protein from Salmonella enterica binds to response regulator PmrA and protects it from dephosphorylation. In contrast, activation of a histidine kinase by a TCS connector appears to be less frequent. Recently, a small inner membrane protein B1500 was described in E. coli that binds and regulates the sensor kinase PhoQ resulting in increased phosphorylation of the response regulator PhoP. Expression of B1500 is in turn controlled by the EvgS/EvgA TCS, which functionally connects both TCSs (Eguchi et al., 2007). However, it is unknown whether B1500 stimulates kinase activity or inhibits phosphatase activity of PhoQ. Therefore, it is demonstrated here for the first time that activity of a sensor kinase can be stimulated by a TCS connector, in this case IIANtr. Moreover, the activity of the TCS connector IIANtr is itself regulated by phosphorylation, which is an unprecedented case.

Experimental procedures

Growth conditions, strains and plasmids

Bacteria were routinely grown in LB broth under agitation (200 r.p.m.) at 37°C. Where appropriate, media were supplemented with antibiotics (ampicillin: 100 μg ml−1; chloramphenicol: 15 μg ml−1; kanamycin: 30 μg ml−1; tetracycline: 12.5 μg ml−1). For the induction of pBAD-driven gene expression 0.2% (w/v) arabinose was added to the culture. For monitoring expression of the kdpFABC operon, bacteria were grown in phosphate-buffered minimal medium that allows variation of the K+ concentration without affecting osmolarity (Epstein and Kim, 1971; Sardesai and Gowrishankar, 2001): K115 medium contains 115 mM K+ and consists of: 46 mM K2HPO4, 23 mM KH2PO4, 8 mM (NH4)2SO4, 0.4 mM MgSO4, 6 μm FeSO4, 1 mM Na3 citrate, 0.2% (w/v) glucose, 1 mg l−1 thiamine/HCl, 20 μg ml−1l-proline and 0.5% (w/v) casamino acids. For the experiments in Fig. 8 and Fig. S5, glucose was replaced by the indicated carbon source. In K0 medium the potassium salts are replaced by equimolar concentrations of the respective sodium salts. Intermediate K+ concentrations were obtained by mixing suitable proportions of the K115 and K0 media. The strains and plasmids used in this study are described in Table 3. Oligonucleotides are listed in Table S2 (see Supporting information). The ΔkdpB and ΔtrkA gene deletions were constructed according to a standard procedure (Datsenko and Wanner, 2000) using the oligonucleotide pairs BG302/BG303 and BG416/BG417 respectively. In the resulting strains Z119 and Z166 genes kdpB and trkA were each replaced by a chloramphenicol resistance (cat) cassette. Subsequently, the cat cassette in strain Z119 was removed with help of plasmid pCP20 as described, resulting in strain Z123. Integration of the kdpFA′-lacZ-fusion into the λ attachment site (attB) of the chromosome was achieved as described (Diederich et al., 1992). The ΔtrkA::cat and ΔptsN::cat mutations were moved by bacteriophage T4GT7 transduction (Wilson et al., 1979). Constructed strains were verified by PCR using appropriate primers. For the construction of plasmid pBGG98 carrying kdpB under Para promoter control, kdpB was amplified by PCR using primers BG311 and BG312 and the DNA fragment was inserted between the SacI and XbaI sites on plasmid pBAD33. For the construction of the transcriptional kdpFA′-lacZ fusion on plasmid pBGG158, the kdp promoter region was amplified by PCR using primers BG389 and BG390 and used to replace the SalI-XbaI fragment in the lacZ reporter plasmid pKES15. In the otherwise isogenic construct pBGG180, the kdpFABC promoter was replaced by the Ptac promoter. For its construction the SalI-EcoRI fragment in pBGG158 was replaced by a PCR fragment that was obtained using primers BG151 and BG152 and plasmid pFDX3453 (Görke and Rak, 1999) as template. Plasmid pBGG190 carrying 10His-ptsN under tacOP control was constructed in two steps. First, gene yhbJ was amplified by PCR using primers BG396 and BG397. This amplification added 10 histidine codons to the 5′ end of yhbJ and introduced a NheI site between the His-codons and the yhbJ start codon. The PCR fragment was subsequently inserted between the NdeI- and PstI-sites on plasmid pKES170 resulting in plasmid pBGG162. Next, ptsN was amplified using oligonucleotides BG462 and BG463 and plasmid pBGG86 as template and the obtained DNA fragment was used to replace the NheI-PstI fragment in plasmid pBGG162. Plasmid pBGG211, which expresses 10His-ptsN-H73A under tacOP control, was constructed in the same way, with the difference that plasmid pBGG93 was used as template for PCR. For the generation of the bacterial two-hybrid vectors the following genes were cloned between the PstI and XbaI sites of plasmid pKT25 using the primers given in parentheses for amplification: ptsN (BG548/BG545), ptsN-H73A (BG548/BG545), kdpE (BG549/BG547). These constructions resulted in plasmids pBGG261, pBGG262 and pBGG263 respectively. Similarly, kdpE (primers BG546 and BG547), kdpD (primer BG550/BG551) and atoS (primers BG592 and BG593) were amplified by PCR and subsequently cloned between the PstI and XbaI sites of plasmid pUT18C resulting in plasmids pBGG260, pBGG264 and pBGG333 respectively. Moreover, kdpD was cloned between the PstI and XbaI sites of plasmid pKNT25 following amplification using the primers BG550 and BG551.

Table 3. E. coli strains and plasmids used in this study.
NameGenotype or relevant structuresReference or construction
  1. ori, origin of replication; MCS, multiple cloning site; RBS, ribosomal binding site.

Strains
 BL21(DE3/pLysS)FompT rB mB dcm gal (DE3) pLysS (CmR) tonAStudier and Moffatt (1986)
 BTH101FcyaA-99 araD139 galE15 galK16 rpsL1 hsdR2 μrA1 μrB1Karimova and Ladant (2005)
 DH5αgyrA96 recA1 relA1 endA1 thi-1 hsdR17 glnV44 deoR Δ(lacZYA-argF)U169
[φ80dΔ(lacZ)M15]
Woodcock et al. (1989)
 HAK006CSH26 ΔkdpABCD::neo λattB::kdpFA′-lacZΔ(lac-pro) ara thiNakashima et al. (1993)
 JM109recA1 endA1 gyrA96 thi hsdR17 supE44 lrelA1 (lac-proAB)/F′traD36
proA+B+lacIqlacZDM15
Yanisch-Perron et al. (1985)
 R1279CSH50 Δ(pho-bgl)201 Δ(lac-pro) ara thiGörke and Rak (1999)
 R1653as R1279, but Δ[ptsH ptsI crr]::neoGörke and Rak (1999)
 R2404as R1279, but ΔptsPKalamorz et al. (2007)
 R2409as R1279, but ΔptsPΔ[ptsH ptsI crr]::neoKalamorz et al. (2007)
 R2413as R1279, but Δ[ptsN, yhbJ, ptsO]Kalamorz et al. (2007)
 R2415as R1279, but Δ[ptsN, yhbJ, ptsO]Δ[ptsH ptsI crr]::neoKalamorz et al. (2007)
 RH002as HAK006, but kdp+Heermann et al. (2003)
 RH30as RH002, but ptsP::Tn10This work
 TKR2000ΔkdpFABCDE trkA405 trkD1 atp706Kollmann and Altendorf (1993)
 Xl1-BluerecA1 endA1 gyrA96 thi1 hsdR17 glnV44 relA1 lac[F′proAB lacIq
Δ(lacZ)M15 Tn10 (Tetr)]
Bullock et al. (1987)
 Z77as R1279, but ΔptsN::catZimmer et al. (2008)
 Z78as R1279, but ΔptsO::catZimmer et al. (2008)
 Z79as R1279, but ΔptsNZimmer et al. (2008)
 Z80as R1279, but ΔptsOZimmer et al. (2008)
 Z99as R1279, but ΔptsO::catΔ[ptsH ptsI crr]::neoZimmer et al. (2008)
 Z101as R1279, but ΔptsOΔ[ptsH ptsI crr]::neoZimmer et al. (2008)
 Z119as R1279, but ΔptsNΔkdpB::catPCR BG302/BG303 → Z79; this work
 Z123as R1279, but ΔptsNΔkdpBZ119 cured from cat; this work
 Z166as R1279, but ΔtrkA::catT4GT7 (Z167) → R1279; this work
 Z167as R1279, but ΔptsNΔtrkA::catPCR BG416/BG417 → Z79; this work
 Z185as HAK006, but ΔptsN::catT4GT7 (Z77) → HAK006; this work
 Z186as HAK006, but ΔptsNZ185 cured from cat; this work
 Z267as R1279, but λattB::[aadA, kdpFA′-lacZ]pBGG158 → R1279; this work
 Z268as R1279, but ΔptsN λattB::[aadA, kdpFA′-lacZ]pBGG158 → Z79; this work
 Z269as R1279, but ΔptsP λattB::[aadA, kdpFA′-lacZ]pBGG158 → R2404; this work
Plasmids
 pBAD18ori pMB1, Para, MCS, blaGuzman et al. (1995)
 pBAD33ori p15A, Para, MCS, catGuzman et al. (1995)
 pBDkdpD under PAra-control in pBAD18Jung and Altendorf (1998a)
 pBGG86ptsN under PAra-control in pBAD18Zimmer et al. (2008)
 pBGG92as pBGG86, but ptsN with a His73Asp exchange (CAT(r)GAT)Zimmer et al. (2008)
 pBGG93as pBGG86, but ptsN with a His73Ala exchange (CAT(r)GCT)Zimmer et al. (2008)
 pBGG98kdpB under PAra-control in pBAD33This work
 pBGG158kdpFA′-lacZ, neo, attP, aadA, ori p15AThis work
 pBGG180as pBGG158, but kdpFABC promoter replaced by PtacThis work
 pBGG19010His-ptsN under Ptac-control in pKES170This work
 pBGG21110His-ptsN-H73A under Ptac-control in pKES170This work
 pBGG260kdpE in pUT18CThis work
 pBGG261ptsN in pKT25This work
 pBGG262ptsN with a His73Ala exchange (CAT(r)GCT) in pKT25This work
 pBGG263kdpE in pKT25This work
 pBGG264kdpD in pUT18CThis work
 pBGG267kdpD in pKNT25This work
 pBGG333atoS in pUT18CThis work
 pBluescript SKII+ori pUC, ori f1, Plac::MCS, blaStratagene (La Jolla, CA, USA)
 pEE10His-kdpE under PT7-control in pET16B, blaHeermann et al. (2003)
 pFDX4291Operator-less Ptac, sacB-RBS, pSC101-ori, catKalamorz et al. (2007)
 pFDX4292ptsO under Ptac-control in pFDX4291Kalamorz et al. (2007)
 pFDX4294ptsN under Ptac-control in pFDX4291Kalamorz et al. (2007)
 pFDX4296ptsN, yhbJ, ptsO under Ptac-control in pFDX4291Kalamorz et al. (2007)
 pFDX4320yhbJ, ptsO under Ptac-control in pFDX4291Kalamorz et al. (2007)
 pFDX4322ptsO and ptsN under Ptac-control in pFDX4291Kalamorz et al. (2007)
 pFDX4324yhbJ under Ptac-control in pFDX4291Kalamorz et al. (2007)
 pKES15bgl′-lacZ, neo, attP, aadA, ori p15ANagarajavel et al. (2007)
 pKES170lacIq, Ptac, T7gene10-RBS, NdeI, PstI, rrnBT1/T2, bla, ori pMB1Karin Schnetz
 pKJ2–6HiskdpD-6His under Ptac-control in pKK223-3, blaJung et al. (1997)
 pKNT25ori p15A, Plac::MCS, cyaA 1–224, neoKarimova and Ladant (2005)
 pKT25ori p15A, Plac::cyaA 1–224, MCS, neoKarimova et al. (2001)
 pKT25-zipori p15A, Plac::cyaA 1–224φGCN4-zip, neoKarimova and Ladant (2005)
 pNK2883ats, Tn10 in pBR322, tet, blaKleckner et al. (1991)
 pUT18CColEI-ori, Plac::cyaA 225–399, MCS, blaKarimova et al. (2001)
 pUT18C-zipColEI-ori, Plac::cyaA 225–399φGCN4-zip, blaKarimova et al. (2001)

In vivo protein phosphorylation

Labelling of phosphorylated proteins in vivo with H3[32P]O4 was performed as described recently (Zimmer et al., 2008).

Transposon mutagenesis

For transposon mutagenesis, E. coli RH002 harbouring a chromosomal kdpFA′-lacZ reporter gene fusion was transformed with plasmid pNK2883. The cells were plated on LB medium supplemented with 100 μg ml−1 carbenicillin and incubated at room temperature for 48 h allowing for transposition. Thereafter, all colonies were collected from the plates and inoculated in LB medium supplemented with 12.5 μg ml−1 tetracycline and incubated overnight at 37°C. Subsequently, cells were plated on K5 minimal medium (5 mM K+) supplemented with 0.05% (w/v) X-gal and tetracycline (12.5 μg ml−1). Blue colonies were isolated and the β-galactosidase activity was quantified (Miller, 1972). To make sure that the mutants contained only one copy of Tn10, P1 lysates were prepared (Leder et al., 1977) and transduced into strain RH002. Loss of plasmid pNK2883 was verified by testing loss of resistance to carbenicillin. To map the Tn10 insertions, genomic DNA of the mutants was isolated, digested with PstI, and the fragments were subsequently cloned into PstI-digested pBluescript SKII+. The DNA sequences flanking the transposons were determined by DNA sequencing with primers annealing to the Tn10 region.

Detection of KdpFABC production by Western blotting analysis

Escherichia coli cells were grown to an OD600 = 0.5. Cells were harvested, resuspended in SDS-sample buffer and 3.25 μg of total protein was separated by electrophoresis on a 12.5% SDS polyacrylamide gel. Thereafter, the proteins were transferred to a polyvinylidene difluoride membrane (Bio-Rad) by semidry blotting (60 min at 0,8 mA cm−2). KdpFABC was detected with a polyclonal rabbit antibody directed against KdpFABC (Heermann et al., 2009) diluted 1:15 000. The antibody was visualized using anti-rabbit IgG-AP secondary antibodies (Promega, USA) and the CDP* detection system (Roche Diagnostics, Germany).

β-Galactosidase assays (determination of kdpFABC expression in vivo)

Overnight cultures grown in Kx minimal medium containing the desired potassium concentration and the appropriate antibiotics were inoculated into the same medium to an OD600 of 0.1. Arabinose [0.2% (w/v)] was eventually added for the induction of PBAD-driven gene expression. If not otherwise indicated, the cultures were grown at 37°C to an OD600 of 0.5–0.7, and subsequently harvested. Determination of β-galactosidase activities was performed as described previously (Miller, 1972). Enzyme activities are expressed in Miller units. Enzyme assays were performed at least in triplicate from independent cultures.

Bacterial two-hybrid assay

Interaction of proteins was assayed using the BACTH system based on reconstitution of adenylate cyclase activity essentially as described (Karimova et al., 1998). Briefly, E. coli strain BTH101 was co-transformed with pUT18C-derived plasmids and derivatives of plasmids pKT25 or pKNT25. Transformants were selected using kanamycin (50 μg ml−1) and ampicillin (100 μg ml−1). Cells were grown overnight in LB containing the appropriate antibiotics and 0.5 mM IPTG. These precultures were used to inoculate 10 ml of the same medium to an OD600 = 0.1. The cultures were grown at 30°C and 1 ml samples were harvested when the cells reached an OD600 = 0.5 (exponential growth phase) and an OD600 = 1.3 (transition between exponential and stationary phase). The β-galactosidase activities were determined according to the procedure described previously (Miller, 1972).

Cell fractionation and preparation of inverted membrane vesicles

Membrane vesicles containing KdpD-6His were prepared from strain TKR2000 carrying plasmid pKJ2-6His, as described previously (Jung et al., 1997). Cytosolic fractions containing 10His-KdpE were isolated from E. coli BL21(DE3)/pLysS carrying plasmid pEE as described (Heermann et al., 2003).

Protein purification

Solubilization and purification of KdpD-6His as well as purification of 10His-KdpE was carried out by Ni2+-NTA affinity chromatography as described before (Jung et al., 1997; Heermann et al., 2003). KdpD-6His was reconstituted into E. coli lipids in a lipid : protein ratio of 20:1 as described (Jung et al., 1997). For the purification of 10His-IIANtr and 10His-IIANtr-H73A proteins, transformants of E. coli strain DH5α carrying plasmids pBGG190 and pBGG211, respectively, were grown in 1 l LB to an OD600 = 0.5 and gene expression was induced by the addition of 1 mM IPTG. Growth was continued for 60 min before the cultures were harvested and passed three times through a French pressure cell at 1000 p.s.i. The lysates were centrifuged at 40 000 r.p.m. for 45 min and the supernatants were loaded on pre-equilibrated Ni2+-NTA superflow columns (Qiagen). Proteins were eluted with an imidazole gradient and the 150 mM fractions containing the pure protein were dialysed against buffer (10 mM Tris/HCl pH 7.4, 200 mM NaCl, 25% (v/v) glycerol) and used for subsequent experiments.

Reconstitution of the KdpD/KdpE phosphorylation cascade in vitro

The KdpD/KdpE signal transduction cascade was completely reconstructed in vitro. Purified KdpD-6His was reconstituted in lipids and subsequently mixed with purified 10His-KdpE in a ratio of 1:4 μM and with 100 μM of a DNA fragment carrying the KdpE binding site. The phosphorylation reactions were carried out as recently described (Heermann et al., 2009) in a buffer containing 50 mM Tris/HCl, pH 7.5, 10% glycerol (v/v), 0.5 M NaCl, 2 mM dithiotreitol with the indicated KCl concentrations (0 mM, 50 mM, 250 mM) and 10His-IIANtr or 10His-H73A-IIANtr (in equimolar concentrations to KdpE) respectively. At different times, aliquots were removed, and the reactions were stopped by addition of SDS-sample buffer. Thereafter, samples were subjected to SDS-PAGE. Shortly before stopping SDS-PAGE, an [γ-32P]-ATP standard was loaded on the gels, which allowed for quantification of phosphorylated proteins. Gels were dried, and phosphorylated proteins were detected by phospho-imaging.

Phosphorylation and dephosphorylation assays

KdpD kinase and phosphatase activity assays were performed as described (Jung et al., 1997). Briefly inverted membrane vesicles (2 mg ml−1) containing ∼10% KdpD-6His were incubated at room temperature in phosphorylation buffer [50 mM Tris/HCl, pH 7.5, 10% glycerol (v/v), 0.5 M NaCl, 10 mM MgCl2, 2 mM DTT], and phosphorylation was started by addition of 20 μm[γ-32P]-ATP (2.38 Ci mmol−1). Aliquots were removed at different times and phosphorylation was stopped by addition of SDS-sample buffer. After incubation for 3.5 min, an equal volume of purified 10His-KdpE (0.2 mg ml−1) was added to the remaining reaction mixture and incubation was continued. Additional aliquots were removed at different times and reactions stopped by addition of SDS-sample buffer. The phosphorylation was carried out in the absence or presence of IIANtr in concentrations, which were equimolar to 4 μM KdpE. For dephosphorylation assays, inverted membrane vesicles (1 mg ml−1) containing KdpD, 20 mM MgCl2 and 20 μM ATPγS were added to 1 μM 10His-KdpE∼32P, which was obtained by a procedure described elsewhere (Jung and Altendorf, 1998a). The reactions were performed in the absence or presence of IIANtr (equimolar to KdpE). At different times, aliquots were removed and the reactions were stopped by addition of SDS-sample buffer. Phosphorylated proteins were detected and quantified as described above.

Acknowledgements

Research in the laboratory of B.G. was supported by Grants GO1355/2-1 and GO1355/4-1 of the Deutsche Forschungsgemeinschaft. B.G. thanks Jörg Stülke for support and lab space. Work of K.J. was financially supported by the Deutsche Forschungsgemeinschaft (Exc114/1), and the BMBF (SysMo, project KOSMOBAC). Sonja Kroll is thanked for excellent technical assistance. We are grateful to Oliver Valerius for mass spectrometry and to Karin Schnetz for plasmid pKES170.

Ancillary

Advertisement