Research article

Signal-transduction protein PII from Synechococcus elongatus PCC 7942 senses low adenylate energy charge in vitro

Oleksandra Fokina, Christina Herrmann, Karl Forchhammer


PII proteins belong to a family of highly conserved signal-transduction proteins that occurs widely in bacteria, archaea and plants. They respond to the central metabolites ATP, ADP and 2-OG (2-oxoglutarate), and control enzymes, transcription factors and transport proteins involved in nitrogen metabolism. In the present study, we examined the effect of ADP on in vitro PII-signalling properties for the cyanobacterium Synechococcus elongatus, a model for oxygenic phototrophic organisms. Different ADP/ATP ratios strongly affected the properties of PII signalling. Increasing ADP antagonized the binding of 2-OG and directly affected the interactions of PII with its target proteins. The resulting PII-signalling properties indicate that, in mixtures of ADP and ATP, PII trimers are occupied by mixtures of adenylate nucleotides. Binding and kinetic activation of NAGK (N-acetyl-L-glutamate kinase), the controlling enzyme of arginine biosynthesis, by PII was weakened by ADP, but relief from arginine inhibition remained unaffected. On the other hand, ADP enhanced the binding of PII to PipX, a co-activator of the transcription factor NtcA and, furthermore, antagonized the inhibitory effect of 2-OG on PII–PipX interaction. These results indicate that S. elongatus PII directly senses the adenylate energy charge, resulting in target-dependent differential modification of the PII-signalling properties.

  • cyanobacterium
  • energy charge
  • metabolic signal
  • nitrogen regulator
  • 2-oxoglutarate (2-OG)
  • PII protein


ATP in the cell provides energy for metabolic reactions, serves as a substrate for nucleotide synthesis and regulates cell metabolism as a signal molecule. The adenylate EC (energy charge) [(ATP+ADP)/(ATP+ADP+AMP)] is a measure of the energy available for metabolism [1]. Since in bacteria the concentration of AMP is constantly low, the adenylate EC depends mainly on the ATP/ADP ratio [2]. However, sensors of the adenylate EC have been poorly characterized. Previously, the PII signal-transduction proteins have been suggested to be involved in EC measurement. They respond to ATP, ADP and 2-OG (2-oxoglutarate) by binding these effectors in an interdependent manner [37], thereby transmitting metabolic information into structural states of the PII sensor protein [8,9]. PII, when fully occupied by either ATP or ADP, corresponds to non-physiological extremes of EC. To understand how PII responds to physiologically relevant changes in EC, experiments must be conducted with different ATP/ADP ratios that span physiological conditions.

PII signal-transduction proteins are widely distributed in bacteria, archaea and the chloroplasts of eukaryotes, where they regulate metabolic and regulatory enzymes, transcription factors and/or transport proteins [4,1012]. In addition to responding to the metabolites ATP, ADP and 2-OG, PII proteins can furthermore be subjected to signal-dependent reversible covalent modification, which occurs on the large surface-exposed T-loop [10,13]. PII proteins are compact homotrimers, with each subunit exposing three surface-exposed loops, termed the T-loop, B-loop and C-loop [4]. The large T-loop is highly flexible and adopts different conformations, taking part in the binding of effector molecules and being the dominant structure in protein–protein interactions. The B- and C-loops from opposite subunits face each other in the intersubunit cleft and take part in adenylate nucleotide binding [1416]. The PII trimer contains three adenylate nucleotide-binding sites, one in each intersubunit cleft, with ATP and ADP competing for the same site. In the presence of Mg2+-ATP, up to three 2-OG molecules can bind to the protein at the base of the T-loop in the immediate vicinity of the β- and γ-phosphate of ATP, which ligate 2-OG through a bridging Mg2+ ion [810]. Through this type of interaction, the binding of ATP and 2-OG at one effector-molecule-binding site is synergistic towards each other. However, the three effector-molecule-binding sites exhibit negative co-operativity towards each other [3,6,17,18]. The anti-co-operativity is mediated via intersubunit signalling [19]. The first structural insight into this process was obtained recently for sequential 2-OG binding [8]. Occupation of the first, high-affinity, site creates structural differences in the two neighbouring sites, with one site displaying a clear distortion in the conformation of the bound ATP molecule. The anti-co-operativity in the binding of the effector results in a subsensitive response to this stimulatory effector, which is ideal for accurate detection of a wide range of metabolite concentrations.

Adenylate nucleotide binding seems to be highly similar in all PII proteins from the three domains of life. In bacterial PII proteins, ADP generally does not support 2-OG binding [3,7], whereas in a plant (Arabidopsis thaliana) PII protein, 2-OG binding was supported both by ATP and ADP [20]. Effects of different ADP/ATP ratios have so far only been studied in vitro with Escherichia coli PII protein [3,18]. Increasing ADP levels act antagonistically to 2-OG in the UTase/UR (uridylyltransferase/uridylyl-removing enzyme)–PII–NRII (nitrogen regulator II)–NRI (nitrogen regulator I) signal-transduction cascade in vitro. Furthermore, ADP acts through PII as an activator of ATase (adenylyltransferase)-catalysed GS (glutamine synthetase) adenylylation in the ATase–GS monocycle. On the other hand, the uridylylation of PII by UTase is negatively influenced by ADP [3]. ADP increases the stability of the complex of the PII-family protein GlnK with the ammonium channel AmtB and antagonizes the effect of 2-OG, indicating the PII sensing of the adenylate EC [7]. In the photosynthetic bacterium Rhodospirillum rubrum, three PII homologues are involved in the regulation of the transcriptional activator NifA and the DRAT (dinitrogenase reductase ADP-ribosyltransferase)/DRAG (dinitrogenase reductase-activating glycohydrolase) system for the post-translational regulation of nitrogenase activity [2123]. The DRAT/DRAG system responds to the energy state, pointing towards a connection of PII with the energy status in the cell [22]. However, there is only a slight impact of low ATP levels alone on the PII regulation of both processes; supposedly, the ADP/ATP ratio is significant in this response [5].

In cyanobacteria, PII is involved in nitrate utilization [24,25], regulation of gene expression by sequestering the co-activator PipX of the general transcription factor NtcA [2628], and in arginine biosynthesis by regulating the controlling enzyme NAGK [NAG (N-acetyl-L-glutamate) kinase] [29,30]. The binding of PII from the cyanobacterium Synechococcus elongatus enhances the activity of NAGK and relieves it from feedback inhibition by arginine [31]. In higher plants, PII also regulates NAGK [32] and, moreover, A. thaliana PII binds and inhibits a key enzyme of fatty acid metabolism, acetyl-CoA carboxylase [33], representing another link between PII and carbon metabolism. The structure of the PII–NAGK complex, NAGK regulation by PII and its response towards 2-OG are highly conserved between higher plants and cyanobacteria [3436]. In the presence of Mg2+-ATP, 2-OG strongly inhibits PII–NAGK complex formation. Whereas Mg2+-ATP alone has only a slight effect on complex formation in cyanobacteria [6,31], it favours the binding of PII to NAGK in A. thaliana [34]. ADP accelerates the dissociation of the cyanobacterial proteins but has no major effect on the A. thaliana proteins [36].

The S. elongatus PII protein is subjected to phosphorylation and dephosphorylation in response to cellular 2-OG and ATP levels [37]. In vitro, high concentrations of 2-OG in the presence of ATP cause phosphorylation of Ser49, an exposed residue at the apex of the T-loop [38]. The PII kinase has not yet been identified. The phosphatase of phosphorylated PII, PphA from Synechocystis PCC 6803, readily dephosphorylates PII in the absence of effector molecules. This reaction is partially inhibited by ATP or ADP, but is strongly inhibited by Mg2+-ATP-2-OG [39]. Combining different mixtures of ADP/ATP and 2-OG revealed that the PII dephosphorylation reaction responded with highly sensitivity towards the 2-OG levels; however, changing the ATP/ADP ratio had little effect [40], since both adenylate nucleotides inhibit the dephosphorylation reaction. In contrast, preliminary results indicate that ATP and ADP antagonistically influence the binding of PII with its targets NAGK or PipX [28]. This could indicate that the interaction of PII with its downstream signalling targets is indeed responsive to the ATP/ADP level, as was reported for the interaction of the PII protein from E. coli with its targets NtrB and ATase [3].

In the present study, we investigated the ability of S. elongatus PII to act as an EC sensor in vitro, and found that, indeed, it has this ability. Interestingly, the two targets NAGK and PipX respond differently to changing ADP/ATP ratios, with PipX being more sensitive towards low ADP levels. Furthermore, we observed that the effects of EC were mediated both indirectly, by changing the 2-OG-binding properties of PII, and directly, by the alteration of PII-output activities upon the binding of mixtures of nucleotides to PII.


Construction of heterotrimeric PII protein

Heterotrimeric PII proteins, consisting of the Strep-tagged and non-tagged subunits, were constructed as follows: two glnB genes from S. elongatus, one carrying a Strep-tag fusion and another without a Strep-tag, were cloned into the pETDuet-1 vector (Merck). First, native glnB was amplified using primers containing AatII and AvrII restriction sites: PIInfwd2 (5′-ATGCGACGTCGATAACGAGGGCAAAA-3′) and PIInrev2 (5′-ATGCCCTAGGGTAAACGGCAGACAAA-3′). The PCR product was restricted with AatII and AvrII and cloned into MCS2 (multiple cloning site 2) of pETDuet-1 vector. The resulting plasmid was restricted with NcoI and HindIII for the insertion of Strep-tag-fused glnB into MCS1. The amplification of the second gene was performed using following primers: PIIsfor (5′-ATGCCCATGGTTACCACTCCCTATCAGT-3′) and PIIsrev (5′-ATGCAAGCTTCGCAGTAGCGGTAAAC-3′). The clones were checked by sequencing with the primers pET-Upstream (5′-ATGCGTCCGGCGTAGA-3′) for MCS1 and T7-Terminator (5′-GCTAGTTATTGCTCAGCGG-3′) for MCS2.

Overexpression and purification of recombinant PII, NAGK, PipX and AGPR (N-acetyl-γ-glutamyl-5-phosphate reductase)

The glnB gene from S. elongatus, cloned into the Strep-tag fusion vector pASK-IBA3 (IBA), was overexpressed in E. coli RB9060 [41] and purified using affinity chromatography as described previously [30]. The His6-tagged recombinant NAGK and His6-tagged PipX from S. elongatus and AGPR from E. coli were overexpressed in E. coli strain BL21(DE3) [42] and purified as reported previously [26,30,43]. Heterotrimeric PII was overexpressed in E. coli strain BL21(DE3) and purified as described previously [30] with the following modification: protein elution was performed with a gradient of 30 μM, 200 μM and 1 mM desthiobiotin and analysed by SDS/PAGE. PII trimers with one/two Strep-tags were localized in 200 μM desthiobiotin fractions.

SPR (surface plasmon resonance) detection

SPR experiments were performed using a BIAcore® X biosensor system (Biacore) at 25°C in HBS-Mg buffer (10 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM MgCl2 and 0.005% Nonidet P40), as described previously [31]. The purified His6–NAGK was immobilized on the Ni2+-loaded NTA (nitrilotriacetate) sensor chip to FC2 (flow cell 2) in a volume of 50 μl at a concentration of 30 nM (hexamer) to receive a binding signal of approximately 3000 RU (resonance units), which corresponds to a surface concentration change of 3 ng/mm2. To determine the influence of different ATP/ADP ratios on the association and dissociation of the PII–NAGK complex, a solution of 100 nM PII was injected over the sensor-chip-immobilized His6–NAGK surface in the presence of 2 mM ATP with 0, 1, 2, 3 or 4 mM ADP, as well as in the presence of 2 mM ADP alone. PII was afterwards eluted by an injection of the same proportion of the metabolites. Furthermore, 100 nM PII was bound to the immobilized NAGK with or without 1 mM ADP and eluted by an injection of 1 mM ADP. To load fresh proteins on to the NTA sensor chip, bound proteins were first removed by injection of 25 μl of 0.4 M EDTA, pH 7.5; subsequently, the chip could be loaded again with 5 mM Ni2SO4 solution and His6–NAGK as described earlier.

PII–PipX complex formation on the NTA chip was measured as described previously [26]. The His6–PipX (500 nM) was pre-incubated with homotrimeric Strep–PII (100 nM monomer) in the absence and in the presence of effectors and injected on to the Ni2+-loaded NTA chip. As a control, His6–PipX was bound to the chip in the absence of PII. The response difference between binding of His6–PipX alone and in the presence of PII at t=197 s after the start of the injection phase was taken as a measure of protein binding.

To assay the binding of PipX to immobilized PII, a CM5 sensor chip was treated with an amino-coupling kit (Biacore) to bind Strep-Tactin protein (50 μg/μl) in a 50 μl volume on the surface [28]. Thereafter, the purified heterotrimeric or homotrimeric Strep-tagged PII (40 ng/μl) was immobilized on the chip surface in FC2 in a 20 μl volume. PipX (21 ng/μl) was injected in a volume of 20 μl in the presence of the following effectors: 2 mM ADP, 2 mM ATP and 1 mM 2-OG. To remove PII from the surface, the chip was washed with 5 mM desthiobiotin and regenerated by injecting HABA [2-(4′-hydroxyazobenzene)benzoic acid] buffer (IBA).

ITC (isothermal titration calorimetry)

ITC experiments were performed using a VP-ITC microcalorimeter (MicroCal) in buffer comprising 10 mM Hepes/NaOH, pH 7.4, 50 mM KCl, 50 mM NaCl and 1 mM MgCl2 at 20°C.

Isotherms of 2-OG binding to PII (33 μM trimer concentration) were determined in the presence of various ATP/ADP ratios (1:0.25 mM, 1:0.5 mM and 1:1 mM, as well as only 1 mM ATP or 1 mM ADP). For one measurement, 5 μl of 2 mM 2-OG was injected 35 times (4.2–293.7 μM) to the measuring cell containing PII protein (cell volume=1.4285 ml) with stirring at 350 rev./min. The binding isotherms were calculated from the recorded data and fitted to one-site and three-sites binding models using MicroCal Origin software as indicated.

Direct coupled NAGK-activity assay

The specific activity of NAGK from S. elongatus was assayed by coupling NAGK-dependent NAG phosphorylation to AGPR-catalysed reduction of NAG-phosphate with NADPH as a reductant and recording the change in absorbance at 340 nm [43]. The reaction buffer consisted of 50 mM potassium phosphate, pH 7, 50 mM KCl, 20 mM MgCl2, 0.2 mM NADPH and 0.5 mM dithiothreitol. Each reaction contained 10 μg of AGPR, 50 mM NAG and 2.4 μg of PII with 6 μg of NAGK or 1.2 μg of PII with 3 μg of NAGK. The metabolites ATP, ADP, arginine and 2-OG were varied depending on the experiment being performed. The reaction was started by the addition of NAGK. Phosphorylation of one molecule of NAG leads to oxidation of one molecule of NADPH, which was followed by the linear decrease in absorbance at 340 nm, recorded in a volume of 1 ml over a period of 10 min with a SPECORD® 200 photometer (Analytik Jena). One unit of NAGK catalyses the conversion of 1 mmol of NAG per min. The reaction velocity was calculated with a molar absorption coefficient of NADH of ϵ340=6178 l·mol−1·cm−1 from the slope of the change in absorbance against time.

PII–PipX in vitro cross-linking

PII–PipX interaction was analysed using glutardialdehyde cross-linking. PII (0.1 μg/μl) was pre-incubated with PipX (0.2 μg/μl) in the absence or in the presence of the effectors ATP, ADP and 2-OG in 20 μl of buffer (10 mM potassium phosphate, pH 7.4, 100 mM NaCl and 2 mM MgCl2) at 4°C. After 5 min, 0.1% (w/w) glutardialdehyde was added and the samples were incubated for 5 min at 25°C. The cross-linking reaction was stopped by the addition of 100 mM Tris/HCl, pH 7.4. The cross-link products were analysed by SDS/PAGE (12.5% gel) followed by immunoblot analysis with a PII-specific antibody as described previously [44].


Different ATP/ADP ratios affect the activation of NAGK by PII and its relief from arginine inhibition

The effect of different ADP/ATP ratios on the activation of NAGK by PII and on the PII-mediated relief from arginine inhibition could not be determined in the assay used previously [36], in which ATP consumption was coupled to NADH oxidation. However, by assaying the activity of NAGK in a reaction, where the phosphorylation of NAG is linked to the subsequent reduction of NAG-phosphate by AGPR with NADPH as a reductant [43], it is possible to determine the activity of NAGK under almost physiological conditions and at variable ATP/ADP levels. Increasing ATP concentrations from 0.5 mM to 4 mM enhanced the activity of NAGK both in the presence and absence of PII (Figure 1), which was expected from the Km value of NAGK for ATP (without PII, Km=0.6 mM; in the presence of PII, Km=1.1 mM) [36]. Addition of ADP monotonically reduced the activity of PII-complexed NAGK at any fixed ATP concentration. The higher the ADP concentration, the greater the decrease in activity of NAGK. At a low constant ATP concentration, the relative decrease in NAGK activity by ADP addition was more pronounced than at a high constant ATP concentration (a 6.4-fold decrease from 0 to 4 mM ADP at a fixed 0.5 mM ATP compared with a 2.6-fold decrease from 0 to 4 mM ADP at a fixed 4 mM ATP). At the highest ADP/ATP ratio (4 mM ADP with 0.5 mM ATP), NAGK activity in the presence of PII was as low as NAGK activity in the absence of PII (Figure 1). In the absence of PII, NAGK responded only weakly to different ATP/ADP ratios (Figure 1B), with a 1.8-fold reduction of activity comparing 0 and 4 mM ADP at any fixed ATP concentration. This indicates that the response of NAGK activity in presence of PII towards different ADP/ATP ratios operates through the adenylate-binding properties of PII.

Figure 1 Effect of ADP on NAGK activity in the AGPR-coupled assay

Assays were performed in the presence of ATP at a concentration of 4 mM (dashed line), 2 mM (continuous line), 1 mM (dotted line) and 0.5 mM (dotted and dashed line) as indicated. AGPR-coupled NAGK assays were performed as described in the Experimental section. NAGK activity was plotted against the respective analyte concentrations and the data points were fitted to a hyperbolic curve. Results shown are means±S.D. for three independent measurements. (A) Enzyme activity with PII protein. (B) Effect on NAGK without PII.

In the AGPR-coupled assay, NAGK was more sensitive to arginine inhibition than in the previously reported PK (pyruvate kinase)/LDH (lactate dehydrogenase)-coupled assay [31,36], which keeps the ATP level constant at 10 mM. Using the AGPR-coupled assay, 10 μM arginine inhibited the free enzyme, but PII relieved NAGK from arginine inhibition (Figure 2A), as described previously for the colorimetric assay and the PK/LDH-coupled assay [31,36]. The inhibitory effect of ADP on NAGK in the presence of PII was tested at different arginine concentrations (20 μM, 40 μM and 60 μM), which are completely inhibitory for free NAGK, but are not or only moderately inhibitory for PII-complexed NAGK. In these experiments, ATP was fixed at 2 mM. Arginine inhibition was efficiently relieved by PII in the presence of ATP alone; increasing concentrations of ADP increasingly diminished this effect, but did not eliminate it. If ADP completely inhibited complex formation between PII and NAGK, full inhibition of NAGK would be expected at high ADP concentrations in the presence of arginine. The lack of full inhibition implies that ADP did not fully prevent PII–NAGK interaction in the presence of 2 mM ATP.

Figure 2 Arginine inhibition and antagonistic effect of 2-OG on NAGK activation by PII in the presence of arginine

AGPR-coupled NAGK assays were performed as described in the Experimental section. NAGK activity was plotted against the respective analyte concentrations and the data points were fitted to a hyperbolic curve. Results shown are means±S.D. for three independent measurements. (A) ADP enhances inhibition of the NAGK activity by arginine in the presence of 2 mM ATP. NAGK activity in the absence of PII and ADP (thick continuous line); and in the presence of PII without ADP (dashed line), with 0.5 mM ADP (dotted line), with 1 mM ADP (thin continuous line), with 2 mM ADP (dotted-dashed line) and with 4 mM ADP (dotted-dotted-dashed line). (B) Effect of 2-OG on NAGK induction by PII in the presence of 2 mM ATP and 30 μM arginine without ADP (continuous line) and with 2 mM ADP (dashed line).

2-OG is a key effector molecule in PII-mediated signal transduction; micromolar amounts of 2-OG in the presence of ATP negatively affected PII–NAGK complex formation [31]. Previous studies using the PK/LDH-coupled assay demonstrated the inhibitory effect of 2-OG on PII–NAGK interaction by antagonizing the protective effect of PII on NAGK activity in the presence of 50 μM arginine [36]. In a similar experimental setting using the AGPR-coupled assay, NAGK activity decreased 10-fold when the 2-OG concentration was increased from 0 to 250 μM in the presence of PII, 30 μM arginine and 2 mM ATP (Figure 2B). The apparent IC50 of 2-OG was estimated to be 78 μM, a similar result to that obtained in our previous study [6]. To reveal how ADP affects the response towards 2-OG, the same experiment was performed in the presence of 2 mM ATP together with 2 mM ADP (Figure 2B). The activity in the absence of 2-OG was 3-fold lower than in the control in the absence of ADP, and it decreased 5-fold when 2-OG was titrated up to 250 μM. The apparent IC50 for 2-OG in the presence of ADP was approximately 145 μM, showing that 2-OG also inhibited PII–NAGK complex formation in the presence of ADP, although moderately less efficiently.

Inhibitory effect of ADP on 2-OG binding of PII in the presence of ATP

The 2-OG-binding site in PII is created by a Mg2+ ion, which is co-ordinated by the γ-phosphate of a bound ATP molecule and amino acid residues of one monomer of the PII trimer itself. Binding of a single 2-OG molecule to PII leads to a strong conformational change in the T-loop extending from the subunit, which ligates 2-OG, and, moreover, to subtle changes in the other two binding sites as a result of negative co-operativity [8]. 2-OG binding in the presence of ATP was previously measured by ITC and the raw data could be fitted to a three-sequential-sites binding model, that revealed the anti-co-operative occupation of the three binding sites [6]. Since ADP and ATP compete for the same binding site, an ADP molecule bound to one monomer of PII might influence the affinity for 2-OG of the other two monomers because of the PII intersubunit communication. Therefore next we studied the effect of ADP on the binding of 2-OG to PII. The raw data were fitted according to a three-sequential-sites binding model, but were also fitted using a one-site binding model, because in a mixture of ATP and ADP the number of available 2-OG-binding sites in PII cannot be reliably predicted. Data fitting according to a model with three sequential binding sites could only be achieved for measurements obtained in an excess of ATP over ADP (Table 1). At a molar ratio of 1:1 ADP/ATP, the binding isotherm was only consistent with the one-site model. Fitting to a one-binding-site model reveals a combined Kd value of all binding sites and a mean stoichiometry of bound ligands. As shown in Figure 3(A), 2-OG exhibited high affinity towards PII in the presence of 1 mM ATP. The combined Kd value of 39 μM for all three binding sites was calculated from the binding isotherms of three independent experiments. On the other hand, in the presence of Mg2+-ADP, titration with 2-OG did not yield any calorimetric signals (Figure 3B), obviously because ADP is not able to create the 2-OG-binding site. To reveal how ADP affects the binding of 2-OG to PII in presence of ATP, 2-OG was titrated to various mixtures of ADP/ATP in presence of PII. The addition of 0.25 mM ADP to 1 mM ATP (ADP/ATP ratio of 1:4) had a negative effect on 2-OG binding (Figure 3C), with the apparent Kd value increasing to 78 μM and the N (binding stoichiometry) value decreasing from 1.73 to 1.35 (Table 1). Increasing concentrations of ADP led to a further increase in the Kd value for 2-OG binding and to a decrease in the binding stoichiometry (Figures 3D and 3E and Table 1). In the presence of ATP alone, data fitting retrieved a mean stoichiometry of 1.73 2-OG molecules per PII trimer. Since each PII trimer has three binding sites for 2-OG, data fitting according to a one-binding-site model underestimates the actual number of binding sites, which could be due to the anti-co-operativity between the sites. The N value in the presence of 1 mM ATP and 1 mM ADP (1:1 ratio, N=0.75) therefore indicates that one of the three binding sites can be efficiently occupied by 2-OG under these conditions. Furthermore, ADP increases the Kd value of the remaining 2-OG binding site to approximately 183 μM.

View this table:
Table 1 2-OG binding to PII in the presence of various ADP/ATP ratios

Values correspond to the means±S.E.M. for three experiments. The raw data were fitted using one-site and three-sites binding models for a PII trimer. For comparison, data for 2-OG binding in the presence of 1 mM ATP, fitted according to the three-sites binding model, are given in parentheses. NF, non-fittable.

Figure 3 ITC measurement of 2-OG binding to PII protein

Upper panels: raw data of the heat effect during the titration of 33 μM PII solution (trimer concentration) with 2 mM 2-OG. Lower panels: binding isotherm and best-fit curve according to the one-site binding model. (A) 2-OG binding (titration 4.2–293.7 μM) in the presence of 1 mM ATP. (B) 2-OG binding in the presence of 1 mM ADP. (CE) 2-OG binding in the presence of 1 mM ATP and 0.25 mM (C), 0.5 mM (D) and 1 mM (E) ADP.

Different ATP/ADP ratios alter the association and dissociation of the PII–NAGK complex

As shown previously, ADP negatively affects PII–NAGK interaction by increasing the dissociation of the complex [6]. To reveal in more detail how ADP affects PII–NAGK association, PII was first loaded on to the NAGK surface in the absence of effectors and was than eluted by injection of 1 mM ADP (Figure 4A, left-hand side). Subsequently, the same amount of PII was injected on to the NAGK-sensor surface in the presence of 1 mM ADP, which again resulted in rapid complex formation. At the end of injection, after an initial decrease in RU, dissociation ceased. A further injection of 1 mM ADP again resulted in rapid dissociation of bound PII (Figure 4A, right-hand side). Comparing the association curves shows that, in the presence of ADP, the curve rapidly reaches a plateau, which is at approximately one-third of the maximal level obtained in the absence of ADP. The steep initial increase in RU and rapid equilibration indicates that ADP enhances both the association and dissociation kinetics of PII–NAGK complex formation. The fact that PII-ADP did not fully dissociate from NAGK after the end of the injection, but could again be rapidly dissociated after a further injection of ADP, can be explained as follows. At the end of the PII-ADP injection, when the sensor ship is rinsed with ADP-free running buffer, ADP from PII dissociates faster than PII-ADP from NAGK, leaving ADPfree PII, which can stay bound to NAGK.

Figure 4 SPR analysis of the ADP/ATP ratio influence on the association and dissociation of the PII–NAGK complex

NAGK was bound to FC2 of a Ni2+-loaded NTA sensor chip (see the Experimental section), and FC1 was used as a background control. The response difference (ΔRU) between FC1 and FC2 is shown. (A) Effect of 1 mM ADP on the dissociation of the wild-type PII–NAGK complex following association without effectors (left-hand side) and in the presence of 1 mM ADP (right-hand side) (B) Association and dissociation of the PII–NAGK complex under the influence of different ADP/ATP ratios: 2 mM ATP (continuous line), 2 mM ATP+1 mM ADP (dotted line), 2 mM ATP+2 mM ADP (short-dashed line), 2 mM ATP+4 mM ADP (long-dashed line) and 2 mM ADP (dotted and dashed line).

The influence of various ADP/ATP ratios on PII–NAGK interaction was investigated by injecting PII on to a sensor chip with immobilized NAGK in the presence of different concentrations of ADP and ATP, and subsequently injecting the same buffer in the absence of PII, to determine the ATP/ADP-dependent dissociation of the complex (Figure 4B). In the presence of ATP, rapid association of PII–NAGK was observed, and injection of ATP to the complex did not accelerate the dissociation. Addition of 1 mM ADP to 2 mM ATP had only a minor effect on complex formation (association was slightly faster and the maximal level was slightly lower), but drastically accelerated complex dissociation (Figure 4B). Higher concentrations of ADP (2 and 4 mM) at fixed 2 mM ATP lowered the steady-state binding equilibrium during the first injection phase and increased the dissociation velocity to reach the velocity caused by ADP alone. Remarkably, in the excess of ADP over ATP, there was clearly more PII–NAGK complex formation than with ADP alone, although complex dissociation was similarly fast in both cases.

Affinity of PipX to PII in the presence of ADP

Three PipX monomers can bind to one PII trimer in the absence of 2-OG [28]. However, at elevated 2-OG levels in the presence of ATP, PII–PipX complex formation is impaired and free PipX can now bind to and activate the transcription factor NtcA [28]. These in vitro properties are correlated in vivo to nitrogen-limiting conditions. In the present study, we used SPR spectroscopy and glutardialdehyde cross-linking to determine the influence of ADP on 2-OG-modulated PII–PipX complex formation.

In a first set of experiments, the His6–PipX was pre-incubated with homotrimeric Strep-tagged PII protein in the absence or in the presence of effectors and injected on to the Ni2+-loaded NTA chip. To ensure that all PII trimers are fully occupied with PipX and enough free PipX is present, a 5:1 PipX/PII monomer ratio was used. In a control binding experiment, His6–PipX was loaded on to the chip in the absence of PII. PipX monomers alone showed relatively low binding to the chip due to the single short His6-tag and therefore reached the equilibrium quickly and dissociated from the surface after the end of the injection (see Supplementary Figure S1 at Effectors had no influence on the interaction of His6–PipX with the Ni2+-loaded NTA chip surface. Addition of PII to PipX increased the binding signal in the sensorgram 2.5-fold and decreased the dissociation rate, which can be explained by enhanced binding of the PII–PipX complexes, containing three His6-tags that strongly bind to the surface, compared with monomeric PipX. PII in the absence of PipX did not bind to the chip surface. Thus the increase in RU induced by the addition of PII to PipX can be used to quantify PII–PipX complex formation. Addition of 1 mM ADP to PII–PipX had a strong positive effect on the binding signal, probably due to the stabilization of the PII–PipX interaction, allowing more protein complexes to bind to the chip (Supplementary Figure S1). The experiment was repeated in the presence of different ATP/ADP ratios and 2-OG. The increase in RU (ΔRU) from the start to the end of the injection indicates the amount of protein binding to the chip surface. The ΔRU values (at t=197 s) for the binding of PipX without PII was subtracted from the ΔRU values obtained in the presence of PII and different effector molecules, and the results are shown in Figure 5(A). In the absence of effectors, the PII–PipX complex binding reached more than 800ΔRU. The amount of PII–PipX complexes was enhanced by ADP and, less efficiently, by ATP; intermediate binding levels were obtained in a mixture of 1 mM ATP and ADP. Figure 5(A) shows that 2-OG in the presence of ATP is a strong inhibitor of PII–PipX complex formation: the binding levels were almost the same as for free PipX. ADP acted antagonistically to 2-OG: even small amounts of ADP (0.05 mM) could significantly repress the 2-OG signal in the presence of 1 mM ATP in vitro. The amount of PII–PipX complex in the presence of 1 mM ADP, 1 mM ATP and 1 mM 2-OG was even higher than in the absence of any effectors, showing that 1 mM ADP can almost completely erase the inhibitory effect of 2-OG.

Figure 5 Effect of ATP, ADP and 2-OG on PII–PipX complex formation

(A) His6–PipX (500 nM) was pre-incubated with homotrimeric Strep–PII (100 nM) in the absence or presence of effectors, as indicated, and injected on to the Ni2+-loaded NTA chip for SPR detection. As a control, His6–PipX was injected under the same conditions, but in the absence of PII. The response signal at t=197 s after the start of the injection phase was taken as a measure of protein binding. The binding of His6–PipX alone was subtracted from the results measured in the presence of PII. Results shown are means±S.D. for three independent measurements. (B) PII–PipX binding assay using glutardialdehyde cross-linking in vitro. PII was pre-incubated with PipX in the presence and absence of effectors, as indicated, and was subseqently cross-linked using glutardialdehyde. The cross-link products were analysed by SDS/PAGE followed by immunoblot analysis with a PII-specific antibody. (C) Effect of ADP, ATP and 2-OG on PipX-binding to immobilized heterotrimeric Strep-tagged PII protein. PII (3.2 μM) was bound to FC2 of a CM5 chip and FC1 was used as a background control. The response difference (ΔRU) between FC1 and FC2 is shown. PipX (2 μM) was injected in the absence of effector molecules (continuous line), in the presence of 2 mM ADP (long-dashed line), 2 mM ATP+1 mM 2-OG (medium-dashed line), 2 mM ADP+2 mM ATP+1 mM 2-OG (short-dashed line) and 4 mM ADP+2 mM ATP+1 mM 2-OG (dotted line).

This result was confirmed in a PII–PipX-binding assay, where proteins were pre-incubated under different conditions and cross-linked using glutardialdehyde. Subsequently, the products were visualized by SDS/PAGE and immunoblot analysis with PII-specific antibody. As shown in Figure 5(B), in the absence of PipX, the cross-links of PII trimers were prevalent and the dimeric form was only slightly visible; PII monomers were not present. Although no other bands were detected in the presence of PipX alone, with 1 mM ADP one additional band was observed, with higher molecular mass than the PII trimer, consistent with a PII–PipX cross-link product that could be formed when ADP stabilized the complex. ATP had no such effect, but in the presence of both 1 mM ATP and 1 mM ADP the characteristic band was also observed. In comparison with the SPR experiments, higher concentrations of ADP were needed to neutralize the negative effect of 2-OG.

The PII–PipX interaction was also studied using C-terminal Strep-tagged PII protein immobilized on a Strep-Tactin II-coated sensor chip surface, where PipX was injected as an analyte. Immobilized PII protein was able to build a complex with PipX, but the interaction was insensitive to the effector molecules ATP and 2-OG (see Supplementary Figure S2 at We hypothesized that, by fixing the C-terminus of the PII subunits on the chip surface, the conformational change of the C-terminus that occurs upon binding of ATP and 2-OG (a movement of the C-terminus towards the ATP molecule) [8] was prevented and, therefore, the PII–PipX complex was not affected by these effectors. To resolve this issue, a heterotrimeric Strep-tagged PII protein was constructed by co-expressing Strep-tagged and untagged PII subunits und purifying heterotrimers containing only one or two Strep-tagged subunits. If the abovementioned assumption was true, the untagged subunits should be able to respond to the effectors. Heterotrimeric PII was fixed on to the Strep-Tactin-coated chip, and PipX could bind to this surface as efficiently as to the homotrimeric Strep-tagged PII protein (Figure 5C). In contrast with the Strep-tagged PII homotrimer, the interaction was indeed weakened by 1 mM 2-OG in the presence of 2 mM Mg2+-ATP, although not completely, because of the remaining Strep-tagged PII subunits. Injection of PipX in the presence of 2 mM ADP on to the immobilized heterotrimeric PII resulted in an enhanced velocity of complex association compared with binding in the absence of effector molecules. Furthermore, ADP was able to antagonize the negative effect of 2-OG on PipX–PII binding: when ADP was added to a mixture of ATP and 2-OG, more PipX could bind to the PII surface than in the absence of ADP.


An important feature of cyanobacterial PII signal transduction is the sensing of the balance of carbon and nitrogen status through binding of the central metabolite 2-OG. The results of the present study indicate how PII could also act as a sensor of adenylate EC in cyanobacteria. Binding of 2-OG to the ATP-ligated PII protein is the basis of PII-mediated signal transduction. ADP was shown to occupy the ATP site of PII proteins [3,14,15]. Therefore ADP binding prevents the co-ordination of the bridging Mg2+ ion by the γ-phosphate of ATP, which is essential for 2-OG binding [8]. In agreement, no binding of 2-OG to ADP-occupied S. elongatus PII protein occurs. When ATP and ADP are present simultaneously, as is the case in living cells, there will be competition for the three adenylate-binding sites of the trimeric PII protein. Moreover, it has to be considered that the three binding sites are interacting in an anti-co-operative manner. In contrast with the E. coli system, where ADP binds better than ATP [3], the affinity of each of the three binding sites of the cyanobacterial PII towards ADP (Kd values of 10, 19 and 133 μM) is approximately 2–3-fold lower than towards ATP (Kd values of 4, 12 and 47 μM) [6]. So, when PII is exposed to a mixture of ATP and ADP, a mixed occupation of the adenylate-binding sites is expected with a concomitant decreased capability of binding 2-OG. In agreement, the number of 2-OG-binding sites per PII trimer was reduced by the addition of ADP. As long as ATP stayed in excess over ADP, the 2-OG-binding process could be fitted according to a three-sequential-sites binding model, but the average number of 2-OG-binding sites in the mixed population of PII trimers decreased. This indicates that part of the PII population did not exhibit three 2-OG-binding sites any more. This part of the population probably has at least one ADP molecule bound. Furthermore, the affinity for 2-OG binding decreases. At a 1:1 molar ratio of ADP/ATP, the dominating PII population should consist of PII trimers with a mixed occupation of ATP and ADP. Considering the 2–3-fold higher affinity for ATP, the PII species PII-ATP2ADP1 should clearly prevail over the species PII-ATP1ADP2. Under these conditions, on average, only one 2-OG molecule can apparently bind to PII. The binding process can no longer be fitted according to a three-sequential-sites binding model. The affinity of the remaining 2-OG-binding site seems to be as low as that of the third, low-affinity, 2-OG-binding site of ATP-ligated PII (Table 1). This implies that binding of ADP to one site in PII has a strong negative effect on the remaining 2-OG-binding sites, so that probably only one site can be efficiently occupied. The binding of 2-OG to the lowest-affinity site appears to be crucial for inhibition of PII–NAGK interaction, since the IC50 of 2-OG at a 1:1 ratio of ADP/ATP (145 μM) for inhibiting PII–NAGK interaction is near the Kd measured for 2-OG binding at the same ADP/ATP concentrations (183 μM). This mechanism allows the PII–NAGK complex to stay sensitive towards 2-OG under low-energy conditions.

The ADP-ligated PII protein appears to be able to bind to NAGK, but the binding is clearly different to that of ATP-ligated PII, as revealed by SPR spectroscopy. In particular, the dissociation rate of the PII–NAGK complex was greatly increased by ADP. The increased dissociation rate results in a lower steady-state binding level of PII-ADP to NAGK. PII species occupied by a mixture of ATP and ADP display intermediate binding properties. At 1:2 and 2:2 mM mixtures of ADP/ATP, the PII-ATP2ADP1 species should prevail. These conditions already lead to enhanced complex dissociation. In mixtures of 3:2 or 4:2 mM ADP/ATP, more and more of the PII-ATP1ADP2 species should appear; however, fully ADP-ligated PII species are very unlikely. In agreement, the binding curves under these two conditions were very similar to each other and distinct from that of fully ADP-ligated PII with NAGK.

We have previously suggested a two-step model for the binding of PII to NAGK [6]. Following the formation of a transient-encounter complex involving B-loop residues of PII, the T-loop folds into a compact conformation, which tightly associates with NAGK. In ADP-ligated PII protein, the T-loop may be impaired in folding into the perfect NAGK-binding conformation, explaining the fast dissociation of the complex. Measuring NAGK activity in mixtures of ATP, ADP and PII suggests that PII protein occupied by ATP/ADP mixtures is unable to enhance the catalytic activity of NAGK. On the other hand, NAGK, which forms a complex with ATP/ADP-occupied PII, is protected from arginine inhibition as efficiently as by purely ATP-occupied PII. These properties suggest that protection from arginine inhibition and activation of catalytic activity of NAGK by PII operates by different mechanisms: the loose binding of ATP/ADP-ligated PII to NAGK is sufficient to relieve arginine inhibition, but is insufficient to rearrange the catalytic centre of NAGK, resulting in enhanced activity. This suggestion is in agreement with two different contact surfaces of PII with NAGK [35]. The rearrangement of the catalytic centre of NAGK by PII has been shown to require a tight hydrogen-bonding and ion-pair network involving the distal part of the T-loop of PII and the N-domain of NAGK, tightening the catalytic centre of NAGK [35]. The inability of ATP/ADP-ligated PII to enhance NAGK activity is thus in agreement with the lax binding of this PII species, probably mediated by an imperfect fit of the T-loop to the corresponding NAGK-recognition site. In contrast, the relief from arginine inhibition appears to be mediated by the C-terminus of NAGK [36] interacting with the body of the PII protein. In essence, NAGK complexed to partially ADP-ligated PII is not catalytically induced, but is still relieved from feedback inhibition by arginine. In such a state, PII-complexed NAGK would not support elevated arginine synthesis for efficient nitrogen storage, but could quickly respond to increasing energy levels.

PipX is another known binding partner of PII protein in S. elongatus. It switches between binding to PII or NtcA, depending on the 2-OG concentration. Low adenylate EC (increased ADP/ATP ratio) enhances the association of the PII–PipX complex in vitro, thereby antagonizing the signal of nitrogen limitation (elevated 2-OG levels). PII–PipX complex formation is highly ADP-sensitive; ADP promotes complex formation and protects it from dissociation by 2-OG in vitro. Therefore increasing ADP levels should decrease the activation of NtcA-dependent genes, which depends on PipX–NtcA interaction, due to efficient competition by PII for PipX. Remarkably, the negative effect of 2-OG on PII–PipX interaction was absent when Strep-tagged PII was fixed with its C-terminus to the sensor surface, but could be partially restored using heterotrimeric Strep-tagged PII, which consists partially of untagged monomers with a free C-terminus. Although PII–PipX binding itself does not involve conformational changes of the C-terminus of PII, the movement of the C-terminus imposed by 2-OG binding is essential for the response to the effector molecules and incorporating metabolic signals [8,9]. Taken together, the present study shows how the S. elongatus PII signal-transduction protein is capable of responding in a fine-tuned manner to the change of the EC in vitro. Through PII-dependent adenylate energy signalling, increasing ADP levels should diminish the NtcA-dependent activation of genes required for nitrogen assimilation under nitrogen-limiting conditions (high 2-OG levels). On the other hand, under nitrogen-excess conditions (low 2-OG levels), increased ADP levels, via PII signalling, should diminish the activation of NAGK, thereby reducing the flux into the arginine biosynthesis pathway. In both cases, PII-dependent signalling of increased ADP levels should dampen energetically expensive anabolic reactions. Comparing the response of PipX with NAGK to EC signalling by PII, it appears that PipX is more sensitive to low ADP levels.

The ability to bind 2-OG and adenylate nucleotides is conserved among PII proteins in all three domains of life. Several studies have indicated the involvement of PII in sensing the energy status in various bacteria. It was suggested that PII proteins in the photosynthetic proteobacterium R. rubrum respond to the cellular EC [22,45]. Mutants of R. rubrum with an impaired purine-synthesis pathway were created and then provided with an exogenous adenine source to test the effect of different cellular ATP levels on PII signal transduction. However, depleted ATP levels had little effect on PII-mediated regulation of NifA and nitrogenase activity. It was suggested that the ADP/ATP ratio provides the actual signal for PII protein, but the direct effect of ADP/ATP ratios has not yet been shown [5]. Jiang and Ninfa [3,18] showed for the first time, with reconstituted signal-transduction systems using the E. coli PII protein and its receptors NtrB, ATase and UTase/UR, that ADP affected almost all signalling properties of E. coli PII by antagonizing 2-OG-mediated responses. The metabolite sensing involves intersubunit signalling in the PII trimer itself with co-operation of the multiple effector binding sites [19]. Furthermore, ADP antagonized the inhibitory effect of 2-OG on the binding of the PII paralogue GlnK to the ammonium transport protein AmtB in E. coli [7]. The present study extends these insights: in the case of the cyanobacterial system, ADP does not always antagonize the 2-OG signal, but differentially affects the interaction of PII with its targets. ADP modulates PII signalling to the receptor NAGK primarily at low 2-OG levels without antagonizing the effect of 2-OG, whereas it antagonizes the inhibitory effect of 2-OG for PII–PipX interaction. There is still discussion on how much unbound ATP is available in the cell at any given moment [46], but it appears that the ratio of ADP to ATP, instead of the absolute concentration of ATP, affects PII signal transduction. Further studies using in vivo systems could shed light on the physiological impact of this remarkable and complex signalling system.


Oleksandra Fokina performed experiments, analysed the results and wrote the paper. Christina Herrmann performed experiments. Karl Forchhammer designed the experiments, analysed the results, wrote the paper and reviewed/edited the paper prior to submission.


This work was supported by the Deutsche Forschungsmeinschaft (DFG) [grant number Fo195/9].


We thank Christopher Schuster and Sebastian Kindermann (Interfakultäres Institut für Mikrobiologie und Infektionmedizin, Eberhard-Karls-Universität Tübingen, Tübingen, Germany) for their help in creating the heterotrimeric Strep-tagged PII protein.

Abbreviations: AGPR, N-acetyl-γ-glutamyl-5-phosphate reductase; ATase, adenylyltransferase; DRAG, dinitrogenase reductase-activating glycohydrolase; DRAT, dinitrogenase reductase ADP-ribosyltransferase; EC, energy charge; FC, flow cell; GS, glutamine synthetase; ITC, isothermal titration calorimetry; LDH, lactate dehydrogenase; MCS, multiple cloning site, N, binding stoichiometry; NAG, N-acetyl-L-glutamate; NAGK, N-acetyl-L-glutamate kinase; NTA, nitrilotriacetate; 2-OG, 2-oxoglutarate; PK, pyruvate kinase; RU, resonance units; SPR, surface plasmon resonance; UR, uridylyl-removing enzyme; UTase, uridylyltransferase


View Abstract