Mutation of ptsP encoding EINtr of the PTSNtr system in Rhizobium leguminosarum strain Rlv3841 caused a pleiotropic phenotype as observed with many bacteria. The mutant formed dry colonies and grew poorly on organic nitrogen or dicarboxylates. Most strikingly the ptsP mutant had low activity of a broad range of ATP-dependent ABC transporters. This lack of activation, which occurred post-translationally, may explain many of the pleiotropic effects. In contrast proton-coupled transport systems were not inhibited in a ptsP mutant. Regulation by PtsP also involves two copies of ptsN that code for EIIANtr, resulting in a phosphorylation cascade. As in Escherichia coli, the Rlv3841 PTSNtr system also regulates K+ homeostasis by transcriptional activation of the high-affinity ATP-dependent K+ transporter KdpABC. This involves direct interaction of a two-component sensor regulator pair KdpDE with unphosphorylated EIIANtr. Critically, ptsP mutants, which cannot phosphorylate PtsN1 or PtsN2, had a fully activated KdpABC transporter. This is the opposite pattern from that observed with ABC transporters which apparently require phosphorylation of PtsN. These results suggest that ATP-dependent transport might be regulated via PTSNtr responding to the cellular energy charge. ABC transport may be inactivated at low energy charge, conserving ATP for essential processes including K+ homeostasis.
Nitrogen fixation by Rhizobium–legume symbioses provides a significant proportion of the available nitrogen in the biosphere, making it agronomically and ecologically important. N2 is reduced to ammonium by specialized bacterial cells (bacteroids) inside legume nodules. Ammonium is the primary stable product of nitrogen fixation which bacteroids secrete to the plant where it is assimilated into glutamate by the combined activities of glutamine synthetase and glutamate oxoglutarate amidotransferase (GS/GOGAT) (Prell and Poole, 2006). However, nutrient exchange between the plant cytosol and bacteroid is more complex with amino acid transport being essential for fixation in pea nodules (Lodwig et al., 2003; Prell et al., 2009). This was revealed by mutation of the two ATP-binding cassette (ABC) broad specificity amino acid uptake systems (Aap and Bra) of Rhizobium leguminosarum causing plants to become severely nitrogen starved, even though bacteroids retain the ability to reduce 15N2 to ammonium (Lodwig et al., 2003). The reason is that plants must provide branched chain amino acids to allow full development and persistence of bacteroids (Prell et al., 2009; 2010), because bacteroids have greatly reduced synthesis of branched chain amino acids and become auxotrophic during symbiosis. Mutation of both Aap and Bra prevents bacteroids from receiving branched chain amino acids from the plant, causing a reduction in cell size, chromosome endoreduplication and early senescence.
Transport by Aap and Bra is post-transcriptionally downregulated when the intracellular glutamine to glutamate ratio is high (Mulley et al., 2011). This effect was independent of the classical Ntr-dependent transcriptional regulation of numerous nitrogen regulated operons. The alteration in the ratio of glutamine to glutamate was caused by mutation of gltB which encodes glutamate oxoglutarate amidotransferase (GOGAT). While the regulatory effect of mutation of gltB was clearly post-transcriptional second site suppressor mutations in hfq, glnB and ntrC increased transcription of aap and bra resulting in increased amino acid transport. However, while the post-transcriptional downregulation of aap and bra is correlated with a high glutamine to glutamate ratio this does not prove that these solutes are sensed directly. The regulation could be mediated by ATP or ADP, but the sensing mechanism is unknown and any suggestions remain speculative.
In Bradyrhizobium japonicum mutation of a phosphotransferase component, PtsP (EINtr), lowered the rate of transport by the ABC-type oligopeptide transport system (King and O'Brian, 2001). A similar effect was found when the adjacent gene encoding aspartokinase (ask or lysC) was mutated. These two proteins physically interact, suggesting they act together to alter oligopeptide transport. PtsP (EINtr) is part of the so-called PTSNtr system in numerous bacteria and transfers a phosphoryl group to Npr (PtsO) which in turn phosphorylates EIIANtr (PtsN) (Begley and Jacobson, 1994). The ptsN and ptsO genes are immediately downstream of rpoN in many bacteria although this is not the case in sequenced rhizobia (Barabote and Saier, 2005). PTSNtr was so named because ptsN and ptsO are co-transcribed with the global nitrogen regulator rpoN and initial growth experiments of ptsN mutants indicated deficient use of poor nitrogen sources (Powell et al., 1995). The PTSNtr system is related to the PTS transporters, which in many bacteria transfer phosphate from phosphoenolpyruvate (PEP) to incoming sugars. Whereas the PTS functions mainly in sugar uptake, PTSNtr lacks EIIB/EIIC components needed for solute translocation, indicating it has a purely regulatory function. PTSNtr mutant phenotypes have been linked to nitrogen metabolism in Escherichia coli and Klebsiella pneumoniae, to polyhydroxyalkanoate accumulation in Azotobacter vinelandii and Pseudomonas putida, to virulence in Legionella pneumophila and Pseudomonas aeroginosa and most recently to potassium homeostasis in E. coli (Deutscher et al., 2006; Pfluger-Grau and Gorke, 2010). Unphosphorylated PtsN binds to the Trk potassium transporter in E. coli and inhibits its uptake of potassium (Lee et al., 2007a). This leads to ptsN mutants becoming isoleucine pseudo-auxotrophs in high-potassium medium, because their over-accumulation of potassium inhibits acetohydroxyacid synthase I (AHAS I), the committed step in branched chain amino acid synthesis. However, this model has been challenged by a recent study demonstrating that ptsN phenotypes related to isoleucine pseudo-auxotrophy only occurs in the ilvG backgrounds (AHAS II) used in previous studies (Reaves and Rabinowitz, 2011). Unphosphorylated PtsN also activates transcription of the high-affinity potassium transporter (encoded by kdpABC) by binding to the two-component sensor KdpD (Luttmann et al., 2009). Thus while the role of PTSNtr in regulating branched chain amino acid synthesis has been challenged PTSNtr does have a central role in potassium homeostasis.
In a project looking for altered cell surface phenotypes of R. leguminosarum, we identified a mutant with dry colony morphology and subsequent analysis showed this was caused by mutation of ptsP. Given that PtsP regulates the opp ABC-type transporter in B. japonicum (King and O'Brian, 2001) and that Aap and Bra are subjected to post-transcriptional regulation (Mulley et al., 2011), we investigated the role of the PTSNtr system in R. leguminosarum bv. viciae Rlv3841. We show for the first time a role for PTSNtr in potassium homeostasis outside of enteric bacteria, but also demonstrate it has a far more wide reaching regulation of ABC transport systems. Possible roles of PTSNtr as a global regulator of ATP-dependent processes are considered.
Mutations of ptsP (EINtr) or both ptsN1 and ptsN2 (EIIANtr) cause similar changes to the growth and surface of Rlv3841
A Tn5 mutant library of R. leguminosarum was screened on TY plates for the appearance of dry colonies and A1049 formed small, flat and dry colonies, in contrast to the large, mucous colonies formed by the wild type. The DNA flanking the ends of the Tn5 from A1049 was cloned and end sequenced revealing that the transposon had inserted at bp 4 556 752 of the chromosome, 669 bp downstream of the start of ptsP (RL4283). The Tn5 in A1049 was transduced into Rlv3841 and transductants (PtsP107) were also dry demonstrating that the dry colony phenotype is tightly linked to the transposon insertion.
The R. leguminosarum 3841 genome (Young et al., 2006) contains 19 genes encoding proteins with homology to PTS components. These include PtsP (RL4283) with high similarity to published EINtr proteins (50% similar to E. coli PtsP) and two PtsN proteins (PtsN1: RL0425 and PtsN2: pRL110376), which encode proteins with high similarity to the EIIANtr component of the PTSNtr (58–60% similarity to E. coli PtsN). blast analysis did not reveal an Npr component, which in enteric bacteria transduces the signal between EINtr and EIIANtr, suggesting it is absent in Rlv3841 as is the case in all other sequenced rhizobia. However, there are components of sugar PTS systems including two putative Hpr proteins, RL0032 and RL2903. RL0032 is part of an operon RL0032–34 that also contains a single EIIAMan component and an Hpr kinase (HprK) that are probably orthologous to the characterized Man-PTS in Sinorhizobium meliloti (Pinedo et al., 2008; Pinedo and Gage, 2009). RL2903 appears to be part of an operon (RL2900–6) that may encode a dihydroxyacetone (DHA)-like PTS system. There are two further operons (RL1749–52 and pRL120386–7), which may encode functional DHA-PTS systems, as also reported by Barabote and Saier (2005) for B. japonicum USDA110, S. meliloti 1021 and Mesorhizobium loti MAFF303099. The Rlv3841 genome does not contain any PTS transport components (i.e. EIIB or EIIC) as already reported by Barabote and Saier (2005) for B. japonicum USDA110 and S. meliloti 1021. The gene adjacent to ptsP, RL4284, should be added to this PTS list because it encodes an aspartokinase (Ask), involved in PTS-dependent regulation of an oligopeptide permease in B. japonicum (King and O'Brian, 2001).
Since two genes in R. leguminosarum encode putative PtsN proteins that potentially interact with PtsP, both were mutated. ΩSpec and ΩTet cassettes were inserted in ptsN1 and ptsN2 but the resulting single mutants (LMB271 and RU4391 respectively) did not show a surface phenotype. However, the ptsN1 ptsN2 double mutant (LMB272) had a dry colony morphology, although slightly less pronounced than that of the ptsP mutant (PtsP107). The growth of PtsP107 and LMB272 was compared with wild-type Rlv3841 (Fig. 1). The generation time of PtsP107 and LMB272 increased when glutamate was used as a nitrogen source compared with ammonium. This is consistent with the suggested link of PTSNtr to nitrogen usage (Powell et al., 1995). The generation times of PtsP107 and LMB272 increased further with succinate as a carbon source compared with glucose. This is similar to the observation that mutation of ptsN and ptsP (named ptsA in the original paper) reduced growth on dicarboxylates in Rhizobium etli (Michiels et al., 1998). PtsP107 also grew on inositol as a sole carbon source on which it formed small dry colonies (data not shown). The growth deficiency on glutamate suggested a link to amino acid transport or metabolism.
EINtr and EIIANtr mutants are altered in ATP-dependent ABC transport but not proton-coupled succinate transport
Because of the growth deficiency of PtsP107 on glutamate and the inability of a B. japonicum EINtr mutant to transport δ-aminolevulinic acid (ALA), PtsP107 was tested for transport of various compounds. Transport of α-amino isobutyric acid (AIB; a non-metabolizable amino acid, which is taken up with very similar kinetics to glutamate), γ-amino butyric acid (GABA), ALA, glucose and myo-inositol, was greatly reduced in the EINtr mutant PtsP107 compared with wild-type Rlv3841 (Fig. 2A). These substrates are all taken up by ABC transporters, AIB via the general amino acid permeases Aap and Bra (Hosie et al., 2002), GABA via Bra (Hosie et al., 2002), ALA via the dipeptide permease Dpp (Carter et al., 2002), glucose via at least three different ABC permeases (Karunakaran et al., 2009) and myo-inositol via the Int ABC permease (Fry et al., 2001). In marked contrast, DctA mediated succinate transport, which is proton-coupled (Reid and Poole, 1998), was unaltered (Fig. 2A).
Transport of AIB by the ABC systems Aap and Bra was also tested in the EIIANtr (PtsN) single and double mutants (Fig. 2B). AIB transport of the ptsN1 mutant LMB271 showed a significant drop (P < 0.005; n = 6), although the colony morphology of this mutant was unaffected, while the ptsN2 mutant RU4193 showed no reduction in AIB transport. However, the double ptsN1 ptsN2 mutant LMB272 was greatly reduced in AIB transport (Fig. 2B).
An ask (aspartokinase) gene (RL4284) is next to ptsP in R. leguminosarum as it is in B. japonicum, in which Ask and PtsP physically interact to activate oligopeptide transport (King and O'Brian, 2001). Therefore an ask mutation was introduced into Rlv3841 and the mutant RU4392 showed a small but significant decrease (P < 0.005; n = 6) in AIB transport. However, while this decrease in AIB transport suggests Ask and PtsP may also interact in R. leguminosarum the rate of decrease in AIB uptake was not as marked as the decrease in ALA uptake in the ask mutant of B. japonicumKing and O'Brian (2001).
The dry surface phenotype is not responsible for the transport deficiency
Since the secretion of surface exopolysaccharides (EPS) is dependent on ABC transport systems in rhizobia (Becker et al., 1995), the dry surface phenotype of pstP (EINtr) and the pstN1 and pstN2 (EIIANtr) double mutant may be due to reduced EPS export. To exclude the possibility that it was the altered surface phenotype that caused the reduction in ABC-dependent transport, an EPS (pssA) mutation was transduced into Rlv3841 and PtsP107, forming LMB310 and LMB311 respectively, both of which lack any EPS. AIB transport in LMB310 (pssA::Tn5Spec) was comparable to Rlv3841, while in LMB311 (ptsP::Tn5Nm; pssA::Tn5Spec) the transport was comparable to PtsP107 (Fig. 2C). This indicates that the surface phenotype has no direct influence on transport rates.
However, while the ptsP mutation greatly reduced ABC transport rates, they were not shut down completely (Fig. 2). This could explain why PtsP107 still formed functional nodules on pea plants in contrast to the pssA mutant that formed fix minus nodules (data not shown).
ABC transport is post-translationally regulated
To evaluate at which molecular level ABC transport is regulated by the PTSNtr system a triplicate two-colour microarray experiment (including 1 dye-swap) was performed. RNA was extracted from independent cultures of the wild-type control Rlv3841 and the ptsP::Tn5 mutant PtsP107 grown in AMS glucose NH4Cl. No genes were identified whose expression was significantly increased or decreased more than twofold in PtsP107 compared with wild-type Rlv3841 (Array Express E-MEXP-3381). We therefore conclude that PtsP positively regulates the activity of these ABC transport systems in R. leguminosarum post-transcriptionally.
To test if the translation of aapJ (encoding the solute-binding protein of Aap) was decreased, the level of expression of an aapJ::phoA translational fusion (pRU1829) was analysed; the resulting alkaline phosphatase activity in PtsP107 was 195 (SEM ± 28; n = 3) Miller units compared with 221 (± 48) Miller units in Rlv3841. This indicates there is not a significant decrease of the translation of this PhoA fusion. To assess protein levels more directly, AapJ-Flag and AapJQMP-Flag fusion proteins were expressed from their own promoters in the low-copy-number plasmid pJP2 (pRU1113 and pRU1138 respectively) in PtsP107 and Rlv3841. The expression of AapJ-Flag (pRU1113) did not affect transport levels in either strain, but as expected, pRU1138 (AapJQMP-Flag) increased transport levels (about two-fold) in both, Rlv3841 and PtsP107, leaving the ratio of transport between them unchanged (data not shown). Fluorescent antibody quantification of both AapJ-Flag and AapP-Flag revealed a small drop of approximately 20% in PtsP107 relative to Rlv3841. A similar drop was observed using a native AapJ antibody to quantify the AapJ protein in periplasmic protein extracts of Rlv3841 and PtsP107. Since a 20% decrease is considerably smaller than the ∼ 80% decrease in transport rates by Aap these data suggest that the predominant regulation of Aap by PTSNtr is post-translational.
Role of overexpression of EINtr and EIIANtr on ABC-dependent transport
To verify that the phenotype caused by the ptsP (EINtr) mutation was not a polar effect on downstream genes, ptsP was cloned (forming pLMB151) and expressed from the lacZ promoter. Plasmid pLMB151 rescued the rough colony morphology (not shown) and the AIB transport deficiency of PtsP107 (Fig. 3) and in Rlv3841 significantly (P < 0.001; n = 6) increased AIB transport compared with the wild type.
Introduction into LMB272 (ptsN1/N2) of plasmids carrying ptsN1 or ptsN2 restored both the colony surface (data not shown) and AIB transport (Fig. 3), confirming the mutations in these genes are not polar. A site-directed mutant of ptsN1 (ptsN1-H66A) was generated in which the conserved histidine 66 is changed to alanine. This position is homologous to the H73 phosphorylation site of EIIANtr of E. coli previously reported by Luttmann et al. (2009). A plasmid carrying the mutant ptsN1-H66A allele was almost as effective as a plasmid carrying the WT allele for complementation of AIB transport by LMB272 (ptsN1/N2) (Fig. 3). This was rather surprising and indicates that either unphosphorylated PtsN is capable of activation of ABC-dependent transport when overexpressed or there may be other phosphorylation sites. The simplest interpretation is that PtsN∼P is a high-affinity activator of ABC transport, hence the requirement for PtsP, while PtsN is a low-affinity activator that requires overexpression for activation of transport. However, because some components of the phosphorylation cascade of the Rlv3841 PTSNtr are not known many different models can be constructed.
Non-phosphorylated EIIANtr interacts with KdpD and activates the KdpABC potassium transporter
In E. coli the PTSNtr system has been linked to potassium homeostasis (Lee et al., 2007a; Luttmann et al., 2009). The E. coli EIIANtr protein encoded by ptsN post-translationally regulates TrkA, a component of the low-affinity Trk potassium transport system (Lee et al., 2007a). Unphosphorylated PtsN forms a tight complex with TrkA and inhibits K+ uptake by this system. Additionally unphosphorylated EIIANtr transcriptionally activates the high-affinity potassium transporter KdpABC via the two-component transcriptional regulatory system KdpDE (Luttmann et al., 2009). KdpABC is an ATP-dependent transporter that does not belong to the ABC family. KdpABC is transcriptionally elevated by KdpDE under low intracellular K+. It is the membrane bound sensor KdpD that detects and transmits this signal by phosphorylation to the receiver domain of KdpE, which in turn activates transcription of kdpABC. Rlv3841 does not contain a homologue of Trk but does contain a full-length KdpABCDE transport system, encoded by pRL110381-77. This directly precedes the second copy of EIIANtr (ptsN2: pRL110376), suggesting a functional relationship. Since in E. coli the unphosphorylated form of EIIANtr activates KdpABC at low potassium, mutation of ptsP (EINtr) in Rlv3841 may lead to unphosphorylated EIIANtr, which could constitutively activate KdpABC. Conversely, a ptsN (EIIANtr) mutant may be unable to activate potassium transport resulting in reduced K+ uptake when extracellular K+ levels are limiting. Therefore, LMB272 (ptsN1/N2), which cannot make EIIANtr, was tested on varying potassium concentrations and found to grow only if greater than 10 µM KCl was present in minimal medium, whereas Rlv3841 grew well down to 1 µM KCl (Fig. 4). A KdpA mutant (RU4193) behaved identically to LMB272, requiring at least 10 µM KCl for growth (Fig. 4). The single ptsN mutants (LMB271 and RU4193) showed only slightly increased generation times in potassium limiting medium (data not shown). Thus PtsN1/N2 are required to activate high-affinity K+ transport. The ptsP mutant PtsP107 grew as well as Rlv3841 down to 1 µM KCl (Fig. 4). Since PtsP is not required for Kdp activation this indicates that the EIIANtr protein in its unphosphorylated form activates the high-affinity transporter KdpABC to allow growth in potassium-limiting medium.
As predicted from E. coli and the results above, the KdpABC operon should be constitutively transcribed in a ptsP (EINtr) mutant, because of its inability to phosphorylate the EIIANtr proteins. Additionally kdpABC expression should be repressed or stay unchanged in an EIIANtr double mutant because of the absence of EIIANtr mediated activation of expression. These predictions were tested using quantitative reverse transcriptase PCR to measure the expression of kdpA and kdpB. In PtsP107 compared with the wild-type Rlv3841 expression of kdpA and kdpB, was increased by 4.9-fold (P = 0.016) and 7.1-fold (P = 0.048) respectively in cells grown in AMS glucose/NH4Cl medium with the normal, non-limiting potassium levels (1 mM). In contrast, expression of kdpA and kdpB in LMB272 (ptsN1/N2) relative to Rlv3841 were 0.55-fold (P = 0.55) and 0.28-fold (P = 0.29), respectively, and therefore not significantly different. These expression data confirm the growth results in potassium-limiting medium and support the existence of a phospho-relay system between the R. leguminosarum EINtr and EIIANtr proteins.
EIIANtr and KdpD physically interact
Bacterial two-hybrid (BTH) analysis was used to investigate interactions between the PTSNtr components themselves and between the EIIANtr components and the KdpDE regulatory elements. A strong interaction between KdpD and KdpE was observed, about equivalent to that seen with the positive control (the sensor kinase/response regulator pair NtrB and NtrC) (Fig. 5). No evidence for an interaction between EINtr and EIIANtr was observed, but such an interaction could require another protein equivalent to Hpr/Npr neither of which has been identified. No interaction was detected between EINtr or EIIANtr and Ask (data not shown), in spite of the demonstrated interaction between Ask and EINtr in B. japonicum (King and O'Brian, 2001).
PtsN1 appeared to interact strongly with KdpD (Fig. 5; black bars) and surprisingly a similar interaction could be observed with both the C-terminal part of KdpD (KdpDc) and the N-terminal part (KdpDn). This interaction was not greatly increased when PtsN1 lacking its phosphorylation site (H66A) was used in contrast to the observations of Luttmann et al. (2009). The apparent interaction between KdpD and PtsN2 was less strong but still significant (Fig. 5; white bars). However, PtsN1 and AapP, the ATP binding cassette component of the Aap transport complex which might be a target for interaction, failed to interact (data not shown).
The intracellular K+ level is unaltered in a PtsP mutant
The data above show that the PTSNtr system regulates both high-affinity K+ transport via Kdp and a wide range of ABC transport systems via an unknown post-translational mechanism. Since K+ is a global regulator of bacterial gene expression to stresses, such as osmolarity, the intracellular K+ concentration might regulate the activity of ABC transporters. If this were the case then PTSNtr could indirectly control ABC transporters via the intracellular K+ level. Therefore, the intracellular concentration of K+ was measured in Rlv3841 and the PtsP107 (ptsP) mutant, which exhibits constitutive kdpABC expression. There was no change in the intracellular K+ concentration in Rlv3841 compared with PtsP107 [400 (SEM ± 12; n = 3) and 391 (± 10: n = 3) mM K+ respectively], even though there are large decreases in the transport rates of ABC systems (Fig. 2). This shows that ABC transporters are not regulated directly by the intracellular K+ level and that there are parallel control circuits from PTS to Kdp and ABC transporters.
From this study it is clear that PTSNtr is essential for activation of a broad range of ABC-type transport systems (Fig. 6), while not altering transport by proton-coupled systems such as the dicarboxylate transport system. The effect was shown by microarray, PhoA fusion and Western blotting to act at the post-translational level. The ABC superfamily is one of the largest classes of transporters capable of either uptake or export and is found in all organisms from archaea to bacteria and higher eukaryotes (Higgins, 1992; Davidson and Chen, 2004; Lee et al., 2007b). ABC-type systems are particularly abundant in rhizobia with 269 ABC genes in R. leguminosarum compared with 67 in E. coli (Mauchline et al., 2006). ABC exporters have a common minimum structure consisting of four domains, two of which are hydrophobic integral membrane domains and two cytoplasmically located but membrane-associated ATP-binding cassettes (Davidson and Chen, 2004). In addition ABC uptake systems present in archaea and bacteria have a periplasmically located solute-binding protein capable of delivering solute to the membrane transport complex (Lee et al., 2007b). ABC systems are also involved in protein export (Type I secretion systems in bacteria), DNA compaction and DNA repair (Hopfner et al., 2000; Schmitt et al., 2003; Nasmyth and Haering, 2005). Mechanistically ATP hydrolysis drives solute translocation by all of the diverse transport systems. Therefore regulation of ABC transporters by PTSNtr would have a profound impact on many aspects of cellular physiology leading to pleiotropic effects for ptsntr mutants. This is seen in this study in a dry surface morphology as well as poor utilization of organic nitrogen sources because of severe reduction in amino acid transport by Aap and Bra (Figs 1, 2 and 6).
PtsP is needed for the post-translational activation of amino acid transport suggesting that phosphorylation is required (Fig. 6). Overexpression of PtsP on a plasmid in Rlv3841 increased the level of transport by Aap and Bra confirming the activity of PtsP results in activation. By analogy to the E. coli EINtr system and carbohydrate PTS systems, PtsP in R. leguminosarum can be predicted to alter the phosphorylation state of PtsN1 and PtsN2, perhaps via a not yet identified Hpr/Npr homologue. There was a strong surface phenotype and loss of ABC transport activity only when both ptsN1 and ptsN2 were mutated suggesting they can functionally interchange. However, even a single ptsN1 mutant reduced amino acid transport by Aap and Bra while a single ptsN2 mutant was without effect, suggesting PtsN1 has a greater effect on Aap and Bra than PtsN2 (Fig 2B). One of the simplest interpretations of these data is that phosphorylated PtsN activates ABC transporters although in the absence of direct physical evidence of the phosphorylation pathway and intermediates it is possible that additional factors are involved. An alternative model is that un-phosphorylated PtsN inhibits ABC transporters. However, if PtsN1/N2 really were inhibitors of ABC transport then LMB272 (pstN1/ptsN2) should have fully activated ABC transport which is the opposite of what occurs. Furthermore, overexpression of PtsN1 or PtsN2 as well as a PtsN1 version with a site-directed mutation in the likely phosphorylation site complemented ptsN1/N2 mutations. It should be noted that overexpression of ptsN was facilitated using a pBBR-based plasmid that has a high copy number. Overall, these data strongly suggest that PtsN∼P is a potent activator, while unphosphorylated PtsN is a weak activator that requires strong overexpression for activation of ABC transporters. It was also observed that the phenotype of LMB272 (ptsN1/N2) was slightly weaker than that of PtsP107 (ptsP), suggesting there may be weak cross-talk with other PTS EII proteins (Fig. 2B). There is no obvious Npr in R. leguminosarum, which in E. coli is orthologous to Hpr of carbohydrate PTS systems and is responsible for transferring a phosphoryl group from EI to EIIA. While, there are two candidate Hpr proteins, encoded by RL0032 and RL2903, they have high identity to carbohydrate PTS components and the S. meliloti orthologue of RL0032 has been characterized as an Hpr (Pinedo et al., 2008). Another possible intermediate is Aspartokinase (Ask) the gene for which is next to ptsP. Mutation of ask in B. japonicum caused a large decrease in oligopeptide transport by Opp (King and O'Brian, 2001) but an ask mutant of R. leguminosarum only lowered transport by Aap and Bra by ∼ 20%. While the weak phenotype of RU4392 (ask) may just reflect redundancy with other proteins having similar activity, inspection of the Rlv3841 genome does not reveal other proteins with high identity to Ask. This is curious because Ask is normally the first committed step in the synthesis of the aspartate family of amino acids including lysine, threonine and methionine. However, mutation of ask did not result in amino acid auxotrophy in R. leguminosarum or in B. japonicum (King and O'Brian, 2001), suggesting these amino acids can be made in the ask mutants. However, PtsP is a very large and complex protein of 755 amino acids with four domains including an N-terminal GAF domain, and may be capable of binding multiple ligands as well as Ask. Our data suggest that Ask is unlikely to have an obligate role in signalling by PTSNtr but may alter or extend the signalling range of EINtr. Our failure to detect an interaction between Ask and EINtr may simply reflect the limitations of bacterial two-hybrid analysis or their interaction may be weaker under the conditions used for Rlv3841 compared with B. japonicum (King and O'Brian, 2001).
We also showed that as in E. coli PTSNtr directly regulates K+ accumulation by the EIIANtr-dependent activation of KdpD, leading to enhanced transcription of KdpABC. The activation is dependent upon unphosphorylated EIIANtr, explaining why PtsP107 is unaltered in the concentration of K+ required for growth and is fully activated for KdpABC transcription. However, constitutively increased transcription of Kdp did not result in increased intracellular K+ concentrations. Additionally we demonstrated a direct physical interaction of the EIIANtr proteins with KdpD in bacterial two-hybrid assays. We were not able to clone a full-length kdpD into the bacterial two-hybrid assay vectors and decided to split the protein into its N-terminal and C-terminal ends excluding the membrane anchor. Interestingly, both ends of the KdpD protein interacted with PtsN1 and PtsN2. The C-terminal end of KdpD contains its phosphorylation site and autophosphorylation domain and would be the logical partner for a protein–protein interaction (see also discussion by Luttmann et al., 2009). However, if the N- and C-termini of KdpD fold close together then EIIANtr may make contact with both regions.
The KdpABC system is a K+-dependent ATPase, where KdpA is the K+ channel, while KdpB is transiently phosphorylated by ATP during the transport cycle (Buurman et al., 1995; Haupt et al., 2006). Thus the PTSNtr system in R. leguminosarum regulates both ABC transporters and the Kdp K+ transporter. However, they are regulated in the opposite direction such that ABC transporters are primarily upregulated by phosphorylated EIIANtr while KdpABC is upregulated by unphosphorylated EIIANtr. This is a striking difference that suggests PTSNtr is regulating different ATP-dependent transport systems in response to ATP levels or perhaps more broadly to the adenylate charge in the cell. Physiologically this makes sense because ABC transporters are extremely numerous and at low cellular ATP levels, presumably reflected in a low level of EIIANtr∼P, it may be crucial to reduce their activity to conserve ATP. This effect occurs post-translationally so it will rapidly reduce ATP consumption by ABC transporters. However, under the same conditions of low intracellular ATP the transcription of kdpABC is elevated which should lead to elevated levels of KdpABC. Given that K+ homeostasis is essential for survival of bacteria, under conditions of reduced ATP the level of KdpABC may need to be enhanced to maintain K+ homeostasis. Effectively, under ATP limitation the PTSNtr system would switch ATP utilization away from ABC transporters towards an essential cellular process such as K+ transport by Kdp. Many of the pleiotropic effects of ptsP/ptsN mutations would be explained by reduction in activity of ABC transporters since they control the import and export of so many small molecules, proteins and cell surface precursors. However, if PTSNtr really is a sensor of adenylate charge then many of the effects on metabolism such as a reduction in the production of storage compounds (Pflüger and de Lorenzo, 2007; Velazquez et al., 2007), the synthesis of which would directly or indirectly consume ATP, may also be directly regulated by EIIANtr. This would also explain the dependence of energy intensive nitrogen fixation in K. pneumonia and R. etli on EIIANtr (Merrick and Coppard, 1989; Michiels et al., 1998). However, since it has not been demonstrated that PTSNtr senses adenylate charge this idea is speculative although the proven phosphorylation of PtsP by PEP or ATP (King and O'Brian, 2001; Deutscher et al., 2006) indicates this is not an unreasonable suggestion. If PTSNtr really is a PTSATP or perhaps PTSAdenylate then numerous facets of its complex pleitropic effects fall into place.
Bacterial growth and media
The bacterial strains, plasmids and primers used in this study are detailed in Table 1. Rhizobium strains were grown at 28°C in either Tryptone Yeast extract (TY) (Beringer, 1974) or acid minimal salts medium (AMS) (Poole et al., 1994) with 10 mM d-glucose or 10 mM di-sodium succinate as carbon sources and 10 mM sodium glutamate or 10 mM NH4Cl as nitrogen sources. In AMS cultures with defined potassium levels K2HPO4 was replaced by Na2HPO4 and KCl added to appropriate concentrations. Antibiotics were used at the following concentrations (µg ml−1): streptomycin (Str), 500; neomycin (Nm), 80; tetracycline (Tet), 5; gentamicin (Gm), 20 and spectinomycin (Spec), 100.
Doubling times were determined in 100 µl of cultures in 96-well microtitre plates in triplicate shaking at 200 r.p.m. in a PowerWave 340 (BioTek, USA) plate reader at 28°C over a maximum time of 20 h with the OD595 determined every 30 min.
Mutant and plasmid construction
Strain Rlv3841 was mutagenized with Tn5 by conjugation with E. coli S17-1 containing pSUP202-1::Tn5 as previously described (Walshaw et al., 1997). Strain PtsP107 (ptsP::Tn5) was isolated by using the general transducing phage RL38 to lyse strain A1049 (ptsP::Tn5) and transduce the Km/Nm marked ptsP::Tn5 mutation back into Rlv3841 as previously described (Buchanan-Wollaston, 1979).
The ptsN1 (RL0425) mutant LMB271 was isolated as follows. The ptsN1 gene was amplified with flanking DNA using primers p1700/p1701. The resulting PCR product was cloned into pCR2.1 following the manufacturer's instructions and a BoxI restriction site in ptsN1 was used to insert a SmaI-digested ΩSpec cassette (Fellay et al., 1987). From that construct an ApaI/XbaI fragment was cloned into pJQ200SK to form pRU2225. Strain LMB271 (ptsN1::ΩSpec) was generated in the Rlv3841 wild-type background by selecting for recombination using the sac mutagenesis strategy as previously described (Kumar et al., 2005). Strain LMB271 was checked by PCR using primers pOTforward/p1702 and pOTforward/p1703, respectively, to confirm from both sides that the correct insertion was present.
The ptsN2 (pRL110376) mutant RU4391 was generated as follows. The ptsN2 gene was amplified with flanking DNA using primers p1704 and p1705. The resulting PCR product was cloned into pCR4 following the manufacturer's instructions and an EcoRV restriction site in ptsN2 was used to insert a SmaI-digested ΩTet cassette (Fellay et al., 1987). An ApaI/XbaI fragment was cloned from this into pJQ200SK to form pRU2227. Strain RU4391 (ptsN2::ΩTet) was then generated using the same sac mutagenesis strategy as above and the insertion PCR mapped with primers p1706 and p1707.
The ptsN1::ΩSpec ptsN2::ΩTet double mutant LMB272 was generated using pRU2227 (ptsN2::ΩTet) with LMB271 (ptsN1::ΩSpec) as a recipient.
The ask (RL4284) mutant RU4392 was generated as follows. The ask gene was amplified with flanking DNA using the primers p1708 and pr1709. The resulting PCR product was cloned into pCR4 following the manufacturer's instructions and an EcoRV restriction site in ask was used to insert a SmaI-digested ΩTet cassette (Fellay et al., 1987). From this an ApaI/XbaI fragment was cloned into pJQ200SK to form pRU2234. Strain RU4392 (ask::ΩTet) was generated in the Rlv3841 wild-type background using the same sac mutagenesis strategy as above and the insertion PCR mapped with primers p1710 and p1711.
The kdpA (pRL110381) mutant RU4393 was generated as follow. The kdpA gene was amplified with flanking DNA using the primers p1716 and p1717. The resulting PCR product was cloned into pCR2.1 following the manufacturer's instructions and an NruI restriction site in kdpA was used to insert a SmaI-digested ΩTet cassette. From this a PstI/XbaI fragment was cloned into pJQ200SK to form pRU2226. Strain RU4393 (kdpA::ΩTet) was then isolated using the sac mutagenesis strategy as above.
Strain LMB310 (pssA::TnSpec) and LMB311 (ptsP::Tn5; pssA::TnSpec) were made by using the general transducing phage RL38 to lyse strain A1204 (pssA::TnSpec) and transduce the Spectinomycin marked pssA::TnSpec mutation back into Rlv3841 to generate LMB310 and into PtsP107 (ptsP::Tn5) to generate LMB311 as described previously (Buchanan-Wollaston, 1979).
A TnphoA translational fusion to AapJ was isolated as follows. Plasmid pRU1134 containing the aap region was transformed into E. coli MC1061. Lambda TnPhoA was used to mutagenize MC1061 as previously described (Simon et al., 1989). Plasmids were recovered from this strain and transformed into E. coli DH5α with selection for kanamycin (TnphoA) and tetracycline (plasmid marker). One plasmid containing an in-frame TnphoA insertion was selected and sequenced using primer IS50R to confirm insertion at bp171 of aapJ.
The full ptsP open reading frame was cloned for complementation and overexpression. Primers pr137b and pr0138b were used to amplify ptsP. The resulting fragment was directly cloned as a HindIII/XbaI fragment into pRK415 forming plasmid pLMB151. The full ptsN1 and ptsN2 ORFs were amplified using primers pr0139b/pr0140b and pr0141b/pr0142b and cloned into pBBRMCS5 following the above strategy forming pLMB159 and pLMB161 respectively. The ptsN1-H66A derivative pLMB160 was generated using site-directed mutagenesis.
The kanamycin resistance casette in pssA::Tn5 (strain A1073) was replaced with a spectinomycin resistance casette by conjugating in pJQ15-6 and selecting for marker exchange by sucrose selection (Quandt et al. 2004). The resulting strain was named A1204.
Rhizobium leguminosarum uptake assays were performed with 25 µM (4.625 kBq of 14C) solute (Hosie et al., 2002), using cultures grown in AMS with 10 mM glucose and 10 mM NH4Cl to an OD600 of ∼ 0.4.
Microarrays and quantitative RT-PCR analysis
Microarrays and qRT-PCR analysis were performed as described previously (Karunakaran et al., 2009). Rlv3841 and PtsP107 cultures were grown in triplicate to mid-log phase in AMS glucose/NH4Cl cultures. Primers for kdpA and kdpB qRT-PCR reactions are detailed in Table 1 (pr0686–89). Mdh (p1310/11) was used as a calibrator gene and results were analysed using REST 2005 BetaV1.9.12 (Bustin, 2000).
A triplicate two-colour microarray experiment (including 1 dye-swap) was performed using RNA extracted from three independent cultures of wild-type R. leguminosarum bv. viciae 3841 (control condition) and the ptsP::Tn5 mutant (PtsP107) grown overnight in AMS supplemented with 10 mM glucose and 10 mM NH4Cl. Array results were deposited in ArrayExpress (accession E-MEXP-3381).
AapJ with a Flag tail at the 3′ end was PCR-amplified using primers p357 and p400 (including Flag epitope Table 1) from R. leguminosarum strain A34 genomic DNA. The resulting PCR product was cloned into pET2.1 (Invitrogen, USA) and further cloned as an XhoI/BamHI fragment into pJP2 forming pRU1113. AapP with a Flag tail was PCR-amplified using primers p357 and p399 (introducing the Flag epitope together with a His-tag; Table 1) from A34 genomic DNA. The resulting PCR product was cloned into pCR2.1 (Fermentas, USA) and further cloned as an XhoI/HindIII fragment into pJP2 forming pRU1138. Vectors, pRU1113 and pRU1138 were conjugated into Rlv3841 forming RU4420 and RU4421 and conjugated into PtsP107 forming LMB221 and LMB222 respectively. Transport assays were conducted on these strains to investigate the influence of the Flag-tagged proteins on transport rates.
Cultures were grown in AMS glucose/NH4Cl medium to an OD600 of ∼ 0.4. Cells were harvested by centrifugation and broken in a FastPrep FP120 ribolyser (Bio101, Thermo). Cell debris was removed by centrifugation, protein levels determined using Bradford reagent and equal amounts of protein (∼ 10 µg) i.e., 10 micrograms were separated on 12% polyacrylamide gels and transferred to Nitrocellulose membrane. Western blots were performed following standard procedures (Sambrook and Russell, 2001). Flag-tagged proteins were detected using a monoclonal ANTI-FLAG® M2-Cy3™ antibody (Sigma-Aldrich, USA). A polyclonal antibody against AapJ was derived from rabbit serum raised against purified AapJ (BioGenes, Germany) and detected with a secondary Cy3-labelled anti-rabbit antibody (Sigma-Aldrich, USA). The hybridization was quantified on a Fuji Phosphoimager FLA-7000 (Fujifilm, Europe) using the MultiGauge2.0 software.
Bacterial two-hybrid analysis
Interaction of proteins was assayed using the BACTH system based on reconstitution of adenylate cyclase activity as described previously (Karimova et al., 1998). Briefly, E. coli strain BTH101 was co-transformed with pUT18C derived plasmids and derivatives of plasmid pKNT25. The proteins tested were cloned into the BamHI/KpnI restriction sites of pUT18C and pKNT25 using primers detailed in Table 1. In case of kdpD the C-terminal end (kdpDc; residues 514–902) of the coding region following the membrane anchor was cloned, as well as the N-terminal end (kdpDn; residues 1–387). Transformants were selected using kanamycin (50 µg ml−1) and ampicillin (100 µg ml−1). Cells were grown at 30°C for 6–7 h in LB containing the appropriate antibiotics and 0.5 mM IPTG. These pre-cultures (50 µl) were used to inoculate 7 ml of the same medium and grown for 16h at 30°C. The β-galactosidase activities were determined according to the procedure described previously (Miller, 1972).
Cells were grown in AMS glucose/NH4Cl to an OD600 of ∼ 0.4 and harvested on 0.4 µm nitrocellulose filters. The cells were then quickly washed with K+ free AMS containing 50 mM NaCl (Lee et al., 2007a) and washed down into 5 ml of Millipore water. An OD600 read was taken as a reference point. Then cells were boiled, cell debris removed by centrifugation and K+ content was quantified using ICP-MS (Thermo scientific). Intracellular concentrations of K+ were determined from the known intracellular volume of Rlv3841 being 1.45 ml per g dry weight (Dilworth and Glenn, 1982).