PipX interacts specifically with nitrogen regulators PII and NtcA: two-hybrid analyses
In a previous work, we constructed Y2H libraries from Synechococcus DNA and screened them with GAL4BD:PII as bait (Burillo et al., 2004). In addition to argB clones, the Sau3A1 libraries produced prey clones that were named PipX (PII Interaction Protein X). pipX was predicted to encode an 89 amino acid protein with a pI of 8.97 and no homology to characterized protein domains. Interestingly, sequences with high homology to pipX were found in all cyanobacterial genomes available and not outside this phylum. An alignment of Synechococcus PipX with orthologous proteins is shown in the Fig. S1.
To provide an independent assay for PipX–PII interactions, we used another in vivo approach, the bacterial adenylate cyclase two-hybrid (BACTH) system, which is based on the interaction-mediated reconstitution of a cyclic AMP signalling cascade in an Escherichia coli cya strain (Karimova et al., 1998). We generated fusion proteins of each of the two fragments (T25 and T18) of the catalytic domain of Bortedella pertussis adenylate cyclase to PII and PipX proteins from Synechococcus and PII (GlnB) protein from E. coli, and analysed their ability to complement the Cya– phenotype. Both pairs of fusion proteins with PipX and PII from Synechococcus complemented the cya strain, thus indicating that PII and PipX retain their affinity in the E. coli system (Fig. S2). This interaction was specific, because none of the PipX fusion proteins interacted with E. coli GlnB. As expected, PII proteins from both Synechococcus and E. coli interacted with themselves and to each other, indicating that PII proteins behave in the BACTH system very much like in the Y2H system, already shown to reflect relevant interactions of PII proteins (Burillo et al., 2004).
Yeast mating of two strains, each one containing one of the GAL4 domains fused to a set of proteins of interest, facilitates systematic interaction analyses of relatively large number of proteins in all possible combinations, including each protein with itself. Positive clones in these assays provide direct evidence that each of the two fusion proteins of the diploid is appropriately expressed in yeast, which is important to interpret negative results. With this in mind, we performed Y2H analysis with PipX, heterologous PII proteins and additional Synechococcus proteins from our laboratory collection. The idea was to provide additional evidence of the specificity of the PII–PipX interaction and to use available fusion proteins, constructed with different purposes, to explore additional connections of PipX with Synechococcus proteins. In all cases, expression of the three reporters (HIS3, ADE2 and lacZ) in Y187/PJ696 diploids containing pairs of fusion proteins was determined and classified according to the strength of the signals, exactly as described (Burillo et al., 2004). For simplicity, only proteins giving signals with Synechococcus PII and/or with PipX have been included in Table 1.
Table 1. Yeast two-hybrid interactions involving PII and PipX proteins from Synechococcus.
| ||Synechococcus||E. coli||A. thaliana|
| NAGK||++||–||+|| ||–||–||–||–|
Control Y2H assays with PII fusions from E. coli (GlnB and GlnK) and A. thaliana confirmed that the interaction between Synechococcus PII and PipX was specific. Given that PII proteins from plants and cyanobacteria share common features (see Burillo et al., 2004 for an extended discussion), PII proteins from A. thaliana provided a good control for specificity of binding. The inability of PipX to interact in the Y2H system with any of the heterologous PII proteins tested indicated that only cyanobacterial PII proteins contain determinants for interactions with PipX.
Interestingly, the two PipX constructs also interacted with one of the proteins assayed, the transcriptional regulator NtcA. Consistent with this, two-hybrid screening of Synechococcus libraries using NtcA as bait also identified PipX as prey (data not shown). Taken together, the two-hybrid analyses performed here indicate that PipX is not a ‘sticky’ protein and that contacts between PipX and the nitrogen regulators PII and NtcA are specific.
Y2H and BACTH assays both rendered negative results for self interaction of PipX, suggesting that PipX does not oligomerize. It is worth noting that these assays are more likely to produce false negative results when the interaction determinants are localized at the N-terminal part of the protein assayed. The lack of signals between NtcA constructs in the Y2H assay (Table 1) may thus be due to occlusion, by GAL4 domains, of important dimerization determinants from the N-terminus of NtcA.
In vitro properties of PipX and PipX–PII complexes: effect of ATP and 2-oxoglutarate on complex formation
Subsequent biochemical analysis of complex formation between PipX and PII and its response to the PII effector molecules was performed using purified Strep-tagged PII (PII-ST) and H6-PipX protein. To provide in vitro evidence for the monomeric structure of PipX, suggested by YTH and BACTH analysis, we performed fast protein liquid chromatography (FPLC) gel filtration analysis with purified H6-PipX protein (His-tagged PipX). H6-PipX eluted from a calibrated Superdex 200 column in a single peak near the total volume, indicating a size of less than 10 kDa, which is in agreement with the assumed monomeric structure of the protein (Fig. 1A and B). Chromatographic mobility of PipX in the presence of PII revealed a significant shift towards increased molecular size, which partially overlapped with the elution of PII (Fig. 1C). Moreover, the elution of PII chromatographed together with PipX was slightly shifted towards increased size, as compared with the chromatorgraphic profile of PII alone (Fig. 1A). The elution shift of PipX towards increased molecular size implies complex formation between PipX and PII, and the partial overlap is an indication of slow complex dissociation during chromatography.
Figure 1. Gel filtration analysis of PipX and PipX–PII complexes. A. Determination of the apparent molecular weight of free PipX and PII and of PipX–PII complexes. The Superdex 200 HR 10/30 gel filtration column was calibrated with the molecular mass standards from Bio-Rad: (1) Thyroglobulin, 670 kDa; (2) Bovine gamma-globulin, 158 kDa; (3) Chicken ovalbumin, 44 kDa; (4) Equine myoglobin, 17 kDa; (5) Vitamin B12, 1,35 kDa. The elution volume (Ve) is shown as a fraction of the void volume (Vo). Below, the fraction numbers are given, corresponding to fraction numbers shown in B and C. (a) Elution peak of free PipX cromatographed in absence of PII; (b) elution peak of PII chromatographed in absence of PipX; (c) elution peak of PII from a PII/PipX mixture. B and C. Coomassie-stained SDS-polyacrylamide gel of the indicated fractions collected after gel filtration of H6-PipX (B) and Silver stained SDS-polyacrilamide gel of the indicated fractions collected after gel filtration of the H6-PipX/PII-ST mixture (C).
Download figure to PowerPoint
Further investigations of PipX–PII complex formation were performed by surface plasmon resonance (SPR) analysis. Initial attempts to stably bind H6-PipX to the Ni-NTA sensor chip were not successful: following initial binding, the protein was gradually washed off from the surface. This precluded the conventional SPR analysis of PII binding to surface-immobilized receptors. As an alternative method, H6-PipX was preincubated with PII-ST and the mixture was injected to the sensor chip. Flow chamber (FC) 1 was used as a control and FC2 was loaded with Ni2+ and the response difference (FC2 − FC1) was recorded. Compared with injection of H6-PipX alone, a significantly higher increase in resonance difference was observed (inset in Fig. 2A). Control experiments confirmed that this increase was indeed caused by complex formation between H6-PipX and PII-ST: no binding of PII-ST was detectable to the Ni-NTA surface alone (not shown) and in addition, when the PII-like GlnK protein from Bacillus subtilis was used instead of Synechococcus PII, no increase in resonance difference could be observed (Fig. 2B). Therefore, the resonance signal increase could be attributed to specific H6-PipX-PII-ST complex formation. Titration of H6-PipX with increasing amounts of PII-ST resulted in a saturation of the effect (Fig. 2A). Saturation started at a molar ratio of one PII-trimer per PipX monomer, indicating a stoichiometry of the complex of 1:1. The same type of binding analyses for H6-PipX-PII-ST in the presence of ATP, ADP or 2-oxoglutarate revealed that complex formation was not affected by any of these molecules individually. However, when increasing ATP concentrations were added in the presence of 2-oxoglutarate (Fig. 3A) or vice versa, increasing 2-oxoglutarate in the presence of ATP (Fig. 3B), complex formation decreased in response to increasing effector molecule concentrations. This effect was specific for 2-oxoglutarate, because other organic acids like pyruvate or succinate did not significantly affect the binding (Fig. 3C). This strongly suggests that in vivo, PipX–PII complex formation is regulated by the cellular 2-oxoglutarate level, and complex formation would be favoured under conditions of nitrogen excess (low 2-oxoglutarate levels).
Figure 2. SPR analysis of PII–PipX complexes. A. Titration of H6-PipX (150 nM) with increasing amounts of PII-ST (from 25 nM to 450 nM PII trimers). The increase of resonance difference (difference in resonance units between Ni-loaded flow chamber 2 and control flow chamber 1) by addition of PII to 150 nM H6-PipX (compared with binding of H6-PipX in the absence of PII) is plotted against the molar ratio of PII (trimer) to PipX (monomer). Standard deviations from three independent experiments are indicated. A representative Biacore experiment is shown as an inset. The graph shows the difference in response units (response difference, RD) between Ni-loaded FC2 – control FC1. The first 60 s correspond to the injection phase (binding of proteins to the surface) followed by buffer flow over the chip (dissociation phase). The response difference obtained immediately after the injection phase is taken as a measure of specific binding. The bottom dotted line corresponds to a control binding experiment of H6-PipX in the absence of PII. The direction of the arrow indicates increasing amounts of PII added. B. Specificity of the PipX–PII interaction. H6-PipX (500 nM) was preincubated with 500 nM PII-ST (PII), GlnK-ST (GlnK from B. subtilis) or no protein (dotted line) and injected to the sensor chip. The response difference between Ni-loaded FC2 and control FC1 is shown.
Download figure to PowerPoint
Figure 3. SPR analysis of the effect of ATP, organic acids and S49D substitution on PipX–PII complex formation. The sensorgrams show the response difference (RD) between Ni-loaded flow chamber and control chamber. The dashed (top) and dotted (bottom) lines correspond, respectively, to controls without effector molecules or without PII-ST protein. A. Effect of various ATP concentrations on the interaction of PII-ST (500 nM) with H6-PipX (500 nM). The direction of the arrow indicates increasing ATP concentrations (0, 0.1, 0.5, 1 and 2 mM) in the presence of 1 mM 2-oxoglutarate. B. Effect of different concentrations of 2-oxoglutarate (0, 0.1, 0.5, 1 and 2 mM) in the presence of 1 mM ATP. C. Effect of 1 mM of the indicated organic acids. D. Interaction of PipX–PIIS49D in the presence of 1 mM 2-OG and 1 mM ATP (+ effectors) or in its absence. Details in B–D are as in A.
Download figure to PowerPoint
Effect of mutations at Ser49 on PipX–PII complexes
Most of the PII residues known to be essential for interactions with receptors map to the T-loop, a flexible region of PII proteins (Jaggi et al., 1996; Jiang et al., 1997; Martinez-Argudo and Contreras, 2002; Burillo et al., 2004; Heinrich et al., 2004). In Synechococcus and other cyanobacterial PII proteins, Ser49 is phosphorylated in response to nitrogen-limiting conditions (Forchhammer and Tandeau de Marsac, 1995; Hisbergues et al., 1999; Lee et al., 2000). PII mutants thought to mimic the phosphorylated (PIIS49D and PIIS49E) and non-phosphorylated (PIIS49A) forms of PII were unable to interact with NAGK but showed no impaired interaction with PipX (Burillo et al., 2004). This suggested that residue S49 of PII is critical for interaction with NAGK but not for interaction with PipX and that phosphorylation is probably not involved in the control of PipX binding to PII.
To biochemically analyse the impact of mutations at Ser49 on PII–PipX complexes, we compared PII-ST and PIIS49D-ST proteins by SPR analysis. As shown in Fig. 3D, no significant differences were observed between the two PII variants in binding to H6-PipX and in their sensitivity to ATP and 2-oxoglutarate. Therefore, the interaction between PipX and PIIS49D is also regulated by 2-oxoglutarate and ATP. If the PIIS49D proteins can mimic the phosphorylated state of PII, then our Y2H and SPR data indicate that interactions with PipX would not depend on the phosphorylation status of PII.
Interactions of cyanobacterial PII proteins with NAGK (Maheswaran et al., 2004) and PamA (Osanai et al., 2005), a protein not conserved within cyanobacteria, are also controlled by 2-oxoglutarate levels. The ability of PIIS49A and PIIS49E proteins to bind effectors in vitro (Lee et al., 2000) and the similar behaviour of PII and PIIS49A, PIIS49D and PIIS49E constructs on interactions with PipX (Fig. 3D and Burillo et al., 2004) further support the importance of 2-oxoglutarate for PII functions.
Recent reports suggest that the interactions of PII proteins with receptors may involve regions outside the T-loops and rather larger surfaces of PII proteins (Zhang et al., 2004; Zhu et al., 2006). In this context, the small size of PipX, the inferred stoichiometry of one PipX monomer per PII trimer (Fig. 2A) and the insignificant impact of substitutions at Ser49 on PipX–PII complexes (Fig. 3D and Burillo et al., 2004), suggest that the main determinants for interactions are at the main core of the PII trimer, rather than at the T-loops.
In vitro properties of PipX–NtcA complexes: effect of ATP and 2-oxoglutarate on complex formation
To obtain in vitro evidence of the interaction between NtcA and PipX, complex formation was analysed between purified proteins using SPR analysis. H6-NtcA could be stably bound to the Ni-NTA sensor chip and PipX-ST was used as analyte. Injection of PipX-ST led to a H6-NtcA dependent and specific increase in difference resonance units (H6-NtcA covered FC2 – control FC1) during the injection phase, followed by a rapid dissociation of the complex. Interestingly, the effect of 2-oxoglutarate on NtcA–PipX complex formation is opposite to that observed for the PipX–PII complex: 2-oxoglutarate significantly stimulates NtcA–PipX complex formation (Fig. 4A). In contrast to PipX–PII binding, ATP was not required for the 2-oxoglutarate effect (Fig. 4B). Further confirmation of the 2-oxoglutarate requirement for PipX–NtcA complex formation was obtained by pull-down experiments. Co-elution of PipX-ST and H6-NtcA from Strep-Tactin columns could only be achieved when 2-oxoglutarate was included in all buffers used for affinity chromatography (Fig. 4C).
Figure 4. SPR analysis of PipX–NtcA complex formation. A and B. BIAcore sensorgrams showing the response difference (FC2 FC1) from binding of PipX-ST to H6-NtcA immobilized surface. (A) 2-Oxoglutarate-dependence of PipX binding to NtcA. The direction of the arrow indicates increasing 2-oxoglutarate concentrations (0, 0.5, 2 and 4 mM 2-oxoglutarate) in the presence of 4 mM ATP. (B) Effect of ATP on the PipX-ST H6-NtcA interaction. Increasing concentrations of ATP (0, 0.5, 2 and 4 mM) were added in the presence of 4 mM 2-oxoglutarate. The dotted bottom line shows PipX binding in the absence of effector molecules. C. Co-elution of PipX-ST and H6-NtcA from a gravity flow Strep-Tactin Superflow mini column and separation of samples in 15% SDS-PAGE gels stained with silver nitrate. Lane F, flow-through; lane W, wash fractions I–II; lane E, elution fractions I-IV. D. As in C, except that 2-oxoglutarate was present in all buffers at 10 mM. A protein molecular weight marker (Fermentas) is shown at the left panel.
Download figure to PowerPoint
The basic properties of PipX–PII and PipX–NtcA complexes indicate that binding of PipX to these regulators is inversely affected by 2-oxoglutarate. In particular, the effectors 2-oxoglutarate and ATP impaired the binding of PipX to PII, while the binding of PipX to NtcA could only be observed in the presence of 2-oxoglutarate. The dependence of 2-oxoglutarate for alternative binding of PipX to NtcA and PII is very significant, given the importance of 2-oxoglutarate as a signal of the carbon/nitrogen balance in cyanobacteria. Concentrations disrupting or stimulating, respectively, PipX–PII and PipX–NtcA complexes (Figs 3 and 4) were similar to those reported to affect PII–NAGK and PII–PamA complexes (Maheswaran et al., 2004; Osanai et al., 2005) or to maximally activate transcription initiation by NtcA (Tanigawa et al., 2002) and therefore the effects of 2-oxoglutarate we observe on PipX complexes are likely to be of physiological significance, strongly suggesting that PipX plays a role in the context of global nitrogen regulation. The estimated 2-oxoglutarate concentrations in cyanobacterial cells are in the 0.06–0.44 mM range (Muro-Pastor et al., 2001; Tanigawa et al., 2002), somehow lower than the concentrations used in most in vitro studies of PII and NtcA proteins.
The Y2H, BACTH and SPR approaches used here detected interactions of PipX with Synechococcus PII but not with control PII proteins from other organisms (A. thaliana, E. coli or B. subtilis), indicating that PII determinants for interactions with PipX are not conserved in any of the heterologous proteins tested. Consistent with the idea of coevolution of proteins working on the same pathways (Pellegrini et al., 1999), NtcA and PipX show identical phylogenetic profiles, that is, both proteins are encoded only by cyanobacterial genomes. While the exclusive and constant presence of pipX and ntcA in all cyanobacterial genomes suggest their adaptative importance, the specific interactions of PipX with nitrogen regulators PII and NtcA strongly suggest that PipX is also involved in nitrogen signal transduction.
To study the role of PipX in transcriptional regulation, NtcA activity was analysed in null (PipX-) and constitutively expressed (PipXc) mutants of pipX containing promoter derivatives glnB::luxAB and glnN::luxAB. These promoters are strictly dependent of NtcA but exhibit different induction profiles (Aldehni et al., 2003; Aldehni and Forchhammer, 2006). Each of the two promoter fusions carries a single NtcA dependent promoter, which in the case of the glnN promoter contains an ‘imperfect’ NtcA-binding site that determines a rather weak and slow induction. Reporter expression was determined by bioluminescence measurements from cultures grown to mid-exponential phase in either the presence of ammonium or nitrate or after cultures were shifted from ammonium-containing to nitrogen-depleted medium.
As shown in Fig. 5, the nitrogen starvation response was severely impaired in PipX- and enhanced in PipXc strains, indicating that PipX plays a positive regulatory role in transcription of the two NtcA dependent promoters analysed. It is worth noting that despite the drastic effect of pipX inactivation on promoter activity, PipX- cultures subjected to nitrogen starvation showed very low but still significant expression of both promoters. Taking into account that in the NtcA- strain there is no residual activation of glnB::luxAB and glnN::luxAB promoters (Aldehni et al., 2003; Aldehni and Forchhammer, 2006), the results indicate that PipX is required for physiological regulation of NtcA promoters, but not essential for transcriptional activation.
Figure 5. Effect of genetic background on glnB::luxAB and glnN::luxAB expression. The time course induction of light emission (mV) of cultures grown with ammonium and shifted to medium without added nitrogen over a time period of 48 h is shown for WT, PipX- and PipXc strains. In each case, the data correspond to a representative time course out of four independent experiments yielding similar results. Wild type, circles; PipX-, squares; PipXc, triangles. Values in the inset tables correspond to the mean values from four independent experiments (with standard deviation) of the same three strains grown with ammonium or nitrate to mid-exponential phase. Values in the ammonium column correspond to time 0 in the graph. A. glnB::luxAB. B. glnN::luxAB.
Download figure to PowerPoint
Interestingly, overexpression of PipX did not increase promoter activity when the cells were cultured with ammonium. In nitrate grown cultures of PipXc, some induction was observed with glnB but not with the lower affinity glnN promoter (insets in Fig. 5). The finding that overexpression of PipX or NtcA (Luque et al., 2004) cannot override the requirement of ammonium deprivation emphasizes the importance of the 2-oxoglutarate signal in transcriptional regulation of the NtcA regulon.
In summary, under conditions of nitrogen deficiency, corresponding to high intracellular levels of 2-oxoglutarate, PipX functions as an activator of NtcA-dependent promoters. Given the 2-oxoglutarate requirement for both NtcA–PipX complex formation (Fig. 4) and in vitro transcriptional activation by NtcA (Tanigawa et al., 2002), it is tempting to propose that the PipX–NtcA complex can activate transcription of NtcA-dependent promoters.
A model for nitrogen control in cyanobacteria: 2-oxoglutarate, PipX and partner swapping
2-Oxoglutarate, the signal of carbon/nitrogen status, plays a key role on regulation of both PII and NtcA proteins, although in the case of NtcA, no direct binding of 2-oxoglutarate has yet been shown. Evidence of regulatory connections between NtcA and PII include the NtcA regulation of PII at the transcriptional and post-translational levels (Lee et al., 1999), the stimulatory role of PII on NtcA activity in nitrogen starved cells (Aldehni et al., 2003; Paz-Yepes et al., 2003) and its inhibitory role when cells are grown with nitrate (Aldehni et al., 2003; Aldehni and Forchhammer, 2006). Although NtcA and 2-oxoglutarate can activate transcription of a target promoter in vitro (Tanigawa et al., 2002), a classical inducer-activator complex does not seem enough to explain regulation of NtcA activity in vivo. Importantly, PipX is clearly involved in the induction of NtcA genes by nitrogen starvation (Fig. 5). To reconcile all these observations, we propose a regulatory model in which the PipX–NtcA complex is part of the mechanism ensuring 2-oxoglutarate dependency for transcriptional activation by NtcA. The function of PipX–PII complexes would be to assist regulation of NtcA–PipX interactions.
According to the model schematically represented in Fig. 6, when cells are grown with ammonium the 2-oxoglutarate concentration favours binding of PipX to PII. In the absence of 2-oxoglutarate, PipX would not bind to NtcA, that would then be inactive. On the contrary, high levels of 2-oxoglutarate favours (ternary) complex formation between PipX and NtcA, bringing about transcriptional activation of target genes. At intermediate levels of 2-OG, PipX could form complexes with either of the two protein partners. The inhibitory effect of PII on NtcA activated genes when cells are grown with nitrate would reflect competition for PipX binding.
Figure 6. Schematic model of the proposed mode of action of PipX. Under low 2-oxoglutarate levels (i.e. ammonium sufficiency) NtcA (white square pairs) is inactive and PipX (black octagons) is bound to PII (trimeric white spheres). In nitrate-grown cultures, part of the PII pool is phosphorylated (black dots) and the moderated levels of intracellular 2-oxoglutarate allow PipX swapping between PII and NtcA, leading to NtcA activation (grey square pairs). In these conditions different conformations of NtcA (active/inactive) and of PII (phosphorylated/non-phosphorylated) coexist. High 2-oxoglutarate levels (i.e. nitrogen starvation) further increases PipX binding to NtcA, which is now fully activated.
Download figure to PowerPoint
As a working hypothesis, we propose that the function of PipX–PII complexes is to fine tune regulation of NtcA–PipX interactions. Indirect support for this idea has been obtained while trying to find evidences of a role of PipX in modulating PII function. For instance, an up to sixfold excess of PipX over PII trimers did not affect PII stimulation of NAGK activity (data not shown), indicating that PipX is probably not involved in regulation of the PII dependent NAGK activity. Moreover, PipX does not seem to modulate PII accumulation or phosphorylation (data not shown) and the PII–PipX interactions are not affected by substitutions at the site of phosphorylation (Fig. 3D; Burillo et al., 2004). The basis of the stimulatory role of PII on NtcA-activated genes under nitrogen deficiency is still unknown. One possibility, currently under investigation, could be that PII deficient mutants accumulate less PipX.
Reported examples of bacterial proteins functioning as coactivators of DNA-binding activators are scarce and they include the phage T4 protein AsiA, which functions in concert with the MotA activator to change promoter specificity by ‘σ appropriation’ (Browning and Busby, 2004; Hinton et al., 2005), the IpgC chaperone (Mavris et al., 2002) and the dihydroxyacetone kinase subunit DhaL (Bachler et al., 2005). A remarkable feature of the NtcA system is that PipX appears to signal the carbon to nitrogen balance, just as 2-oxoglutarate does. This apparent redundancy may nevertheless be important in vivo to avoid non-appropriated induction of the very large NtcA regulon and/or to efficiently amplify the nitrogen starvation signal.
This work places a previously unknown protein, PipX, at the centre of the nitrogen signalling network of cyanobacteria, providing a new element to understand previously reported genetic connections between glnB and ntcA. We have shown that (i) the PipX protein can form complexes with the key nitrogen regulators PII and NtcA, (ii) the 2-oxoglutarate concentration differentially affects PipX–PII and PipX–NtcA complex formation, (iii) the pipX gene encodes a positive regulator of the NtcA-dependent promoters glnB and glnN, and (iv) the regulatory activity of PipX is also dependent of the 2-oxoglutarate signal. The ability of PII to bind to PipX under non-inducing conditions and the sensitivity of PII–PipX complexes to 2-oxogulatarate may also play a role in the regulation of NtcA activity by PipX.