Mechanism of 2-oxoglutarate signaling by the Synechococcus elongatus P_II signal transduction protein

Abstract

P_II proteins control key processes of nitrogen metabolism in bacteria, archaea, and plants in response to the central metabolites ATP, ADP, and 2-oxoglutarate (2-OG), signaling cellular energy and carbon and nitrogen abundance. This metabolic information is integrated by P_II and transmitted to regulatory targets (key enzymes, transporters, and transcription factors), modulating their activity. In oxygenic phototrophs, the controlling enzyme of arginine synthesis, N-acetyl-glutamate kinase (NAGK), is a major P_II target, whose activity responds to 2-OG via P_II. Here we show structures of the Synechococcus elongatus P_II protein in complex with ATP, Mg²⁺, and 2-OG, which clarify how 2-OG affects P_II–NAGK interaction. P_II trimers with all three sites fully occupied were obtained as well as structures with one or two 2-OG molecules per P_II trimer. These structures identify the site of 2-OG located in the vicinity between the subunit clefts and the base of the T loop. The 2-OG is bound to a Mg²⁺ ion, which is coordinated by three phosphates of ATP, and by ionic interactions with the highly conserved residues K58 and Q39 together with B- and T-loop backbone interactions. These interactions impose a unique T-loop conformation that affects the interactions with the P_II target. Structures of P_II trimers with one or two bound 2-OG molecules reveal the basis for anticooperative 2-OG binding and shed light on the intersubunit signaling mechanism by which P_II senses effectors in a wide range of concentrations.

The P_II proteins constitute one of the largest and most widely distributed family of signal transduction proteins present in archaea, bacteria, and plants. They control key processes of nitrogen metabolism in response to central metabolites ATP, ADP, and 2-oxoglutarate (2-OG), signaling cellular energy and carbon and nitrogen abundance (1–4). These effectors bind to P_II in an interdependent manner (see below), thereby transmitting metabolic information into structural states of this sensor protein (3, 5). Furthermore, P_II proteins may be posttranslationally modified (1, 6). Depending on the signal input states, P_II proteins bind and thereby regulate the activity of key metabolic and regulatory enzymes, transcription factors, or transport proteins (1–3). In cyanobacteria and plants, the controlling enzyme of arginine biosynthesis, N-acetyl-L-glutamate kinase (NAGK), is a major P_II target (7–9). Moreover, P_II affects gene expression in cyanobacteria through binding to the transcriptional coactivator of NtcA, PipX (10). In plants, acetyl-CoA carboxylase was recently shown to be regulated by P_II, providing an additional link between carbon and nitrogen regulation (11). Although these P_II targets share no common structural element, interaction with P_II is inhibited by 2-OG.

P_II proteins are homotrimers of 12- to 13-kDa subunits, built of a double ferredoxin-like fold-containing core (βαβ-βαβ), with a characteristic and highly conserved 3D structure, as revealed from numerous crystal structures (3, 12). The trimeric P_II architecture resembles a flattened barrel with long and flexible T loops extending outward, from the flat side (see Fig. 1 and Fig. S1). These T loops can adopt multiple conformations and mediate the versatile protein–protein interactions (3). Each subunit further comprises two small loops (B and C loop) in the intersubunit clefts, facing each other from opposing subunits and taking part in a unique mode of ATP-ADP binding (13–15). ADP and ATP compete here for the same site. In the presence of Mg-ATP, up to three 2-OG molecules can bind per trimer (1) with ADP antagonizing 2-OG binding (16). Notably, Arabidopsis thaliana P_II is an exception, because it binds 2-OG also in the presence of ADP (5, 17). Another feature characteristic for many P_II proteins is also intriguing: The three ATP-binding sites and the three 2-OG–binding sites each exhibit negative cooperativity. Anticooperativity implies strong conformational coupling between the subunits, and this feature probably allows P_II to sense a wide range of metabolite concentrations (5, 16, 18, 19). In contrast to the ATP-ADP–binding site, the 2-OG–binding site is controversial (3). From the crystal structure of a P_II paralogue from Methanococcus jannaschii, GlnK1, one 2-OG molecule was shown to bind from outside to the distal side of the T loop in the presence of Mg-ATP (20). By contrast, a recently published structure of a P_II homologue form Azospirillum brasiliense in complex with Mg-ATP and 2-OG revealed the 2-OG–binding site close to the base of the T loop and near the ATP-binding site (21). However, neither of these two structures was proved by biochemical studies nor could they explain the anticooperative binding of 2-OG.

Fig. 1.

Structural basis of P_II regulation by 2-oxoglutarate. (A–C) Side views of different S. elongatus P_II structures in surface representation. The location of the ATP-binding site is indicated by a yellow dashed circle; the red boxes highlight the structure part discussed detailed in D. (A) Ligand-free P_II modeled together with ATP (placed according to the superimposed structure). The subunits are color coded in dark blue, light orange, and warm pink. The extended C termini and T loops of the proteins chains are marked CT and T-loop, respectively. (B) P_II/ATP/2-OG () complex structure; subunits color coded in orange, marine blue, and red. The structure adopts a compact conformation because of the back folding of the C terminus (CT) onto the core domain structure and the T loop observed in a partially disordered conformation. (C) Structure of P_II in the P_II–NAGK complex (PDB ID code: 2V5H) modeled together with ATP. Subunits are color coded in magenta, deep olive, and salmon. (D) Superposition of the ligand-free P_II and structure as ribbon models in the color codes derived from A and B. Secondary structure elements are marked with β1–β5 and α1–α2 and ATP and 2-OG are shown in stick representation, whereas the Mg ion is marked in green. The sites of two mutations, R9L and K58M, introduced to prove the binding mode of 2-OG are marked with green dots. Remarkable conformational transitions are marked with dashed lines: movement of the C terminus, a shift of helix 1, and displacement of the T-loop base by 2-OG while the tip becomes disordered. (E) Superposition of and P_II–NAGK structures as ribbon models in the color codes derived from B and C. Structural details are indicated as in D.

The structures of complexes of P_II with its regulatory target NAGK from Synechococcus elongatus and A. thaliana are highly similar (22, 23), and the mode of interaction and regulation is apparently conserved in cyanobacteria and plants (24). The P_II–NAGK complex involves one hexameric (trimer of dimers) NAGK toroid sandwiched between two P_II trimers with the threefold axis aligned (23). Each P_II subunit engages two contact surfaces in NAGK binding: A smaller surface involves the B loop and a larger is formed by the T loop, which adopts a tightly bent conformation that fits into the interdomain crevice of NAGK. Binding of P_II enhances the catalytic activity of NAGK and alleviates feedback inhibition by arginine. To bind NAGK, free P_II has to contract its extended T loop. Recently, a two-step process of P_II–NAGK binding was proposed on the basis of the properties of a newly identified S. elongatus P_II variant (I86N), which mimics the P_II conformation in the NAGK-bound state (18): First, a salt bridge between P_II-E85 and NAGK-R233 forms, triggering the extended T loop to fold into the tightly bent conformation that fits into NAGK.

Detailed structural information together with the well-studied biochemical features and highly sensitive complex formation assays render the P_II–NAGK system ideal to study the transduction of metabolic signals into protein action. Until now, it was unclear how the 2-OG signal, perceived by P_II, results in inhibition of P_II–NAGK complex formation. To answer this question, we solved structures of P_II with ATP, Mg²⁺, and 2-OG, revealing the mechanism by which the 2-OG signal is perceived by P_II and controls receptor interactions.

Results

Structure of P_II in Complex with ATP and 2-OG.

Crystals of recombinant P_II protein from S. elongatus were obtained in the presence of excess 2-OG/Mg-ATP () and at substoichiometric amounts of 2-OG () (Table S1). All structures were solved by molecular replacement using the free S. elongatus P_II as a model [Protein Data Bank (PDB) ID code 1QY7]. The crystal diffracting to 2.2 Å contains two identical trimers in the asymmetric unit (with an rmsd of 0.2 Å² for all C atoms superposed). Each trimer contains three Mg²⁺, ATP, and 2-OG molecules. In Fig. 1, this structure is compared to the structures of ligand-free P_II (14) and P_II in complex with NAGK (23). In the ligand-free structure, the T loop, and the C terminus adopt highly extended conformations away from the ATP-binding site (see Fig. 1 A and D). Here, P_II offers an open binding cavity volume of approximately 1,000 Å³ for the ligands. Upon ATP and 2-OG binding, a significant conformational change in the T loop and the C terminus occurs: The C-terminal β-sheet of the free form moves toward the ATP-binding site by up to 20 Å (for the terminal residue I112) to change into a small helix (see Fig. 1D). The T loop contracts and the tip (from residues 44–50) becomes disordered. Part of the flexible T loop moves toward the ATP-binding site to form a kink-like structure (residues 36–41), thereby forming the scaffold for proper ATP and 2-OG assembly. As a result, the ATP/2-OG–binding cavity is completely enclosed by this unique T-loop conformation and the movement of the C terminus (see Fig. 1B, yellow circle). In the P_II structure complexed with NAGK, the T loop is clamped in a different conformation (see Fig. 1 C and E). Here, access to the ligand-binding cavity is slightly reduced compared to the ATP-free form because of a bend in the T-loop conformation, thereby reducing the cavity volume to about 600 Å³.

The 2-OG–binding site of P_II is formed by an ATP-chelated Mg²⁺ ion and residues from one side of the intersubunit cleft (Fig. 2). The ATP molecule bound in the intersubunit cleft is ligated mainly by arginine residues (R38 from monomer A; R101 and R103 from monomer B) together with K90 from monomer A and a small number of hydrogen bonds. ATP fixed by these residues forms the scaffold for Mg²⁺-mediated binding of 2-OG. The Mg²⁺ ion has an almost perfect hexagonal coordination sphere of oxygen atoms, three of which are contributed by oxo groups of the α-, β-, and γ-phosphate of ATP. Two additional ligating atoms are accounted for by the O2 and O5 of 2-OG. The last coordination position is contributed by the OE1 atom of residue Q39. Notably, the 2-OG–binding sphere mainly comprises residues from the T loop. Backbone atoms from residues Q39, K40, and G41 (all contacting O1 or O2 of 2-OG) together with the Q39 side-chain ligate the O5 atom and form one area of interactions. A second interaction region is composed of the backbone nitrogen of G87 (forming a second H bridge to O5) and a strong salt bridge of K58 toward the O3 and O4 atoms (distance of 2.7 Å). Furthermore, residue R9 approaches the O4 to 3.5 Å. All these residues with the exception of K40 are highly conserved in P_II proteins (4) (Fig. S1).

Fig. 2.

Stereo image of the 2-OG–binding site. Residues involved in binding of 2-OG (atoms are marked with small numbers) and the hydrophilic portion of ATP are numbered according to the sequence. Cofactors as well as side- and main-chain atoms are marked in stick representation; Mg²⁺ is marked as a green sphere. Colors of residue numbers (orange and blue) correspond to those of the respective subunits. Residues conserved in standard alignments are boxed. Dashed blue lines represent bonds between residues and ATP and black lines indicate bonds for the ligation of 2-OG, whereas green lines mark the hexagonal coordination of the Mg²⁺ ion.

To validate that 2-OG in the crystal occupies the true binding site, P_II variants were constructed, in which residues K58 and R9, whose side chains according to the structure specifically interact with 2-OG, were replaced by similar-sized uncharged residues (K58M and R9L). Binding of 2-OG in the presence of Mg-ATP was determined by isothermal calorimetry (ITC). Indeed, the P_II K58M variant was completely unable to bind 2-OG, although ATP could still be bound, showing that the loss of 2-OG binding is a specific effect of the K58 replacement. In further agreement with the structural prediction, the P_II R9L variant was strongly impaired in 2-OG binding (Table 1). Moreover, both P_II variants were impaired in NAGK binding, confirming that K58 is indeed pivotal for folding the T loop in the tightly bent structure. The R9 side chain is near the contact surface to NAGK and appears to stabilize the B-loop–T-loop interface (23) (Fig. S2).

Table 1.

Effector molecule binding to P_II variants R9L and K58M

	Kd1, μM	Kd2, μM	Kd3, μM
2-OG (+1 mM ATP)
R9L	441 ± 40	123 ± 7	509 ± 116
K58M	ND	ND	ND
(WT)	(5.1 ± 4.0)	(11.1 ± 1.8)	(106.7 ± 14.8)
ATP
R9L	NM	NM	NM
K58M	10 ± 5	262 ± 136	31 ± 15
(WT)	(4.0 ± 0.1)	(12.5 ± 0.9)	(47.4 ± 21.9)

Values correspond to the mean of two independent experiments ± SEM. The raw data were fitted by using a three-site binding model for a P_II trimer. For comparison, the original data for WT P_II protein are given in parentheses. ND, not detectable; NM, not measured.

Structural Basis of Sequential/Anticooperative 2-OG Binding.

Crystallization of P_II protein in the presence of low 2-OG amounts yielded P_II structures with differing 2-OG content. The structure resolved at a resolution of 1.95 Å contains three P_II trimers in the asymmetric unit; one trimer contained three ATP, one Mg²⁺, and one 2-OG (), the second three ATP, two Mg²⁺, and two 2-OG (), and the third three each ATP, Mg²⁺, and 2-OG () (Fig. 3 and Fig. S3). Additional details of structural parameters are given in SI Text (Tables S1 and S2). Binding of 2-OG does not significantly render the B-factor distribution of the three monomers significantly, unless the mobile elements (C terminus and T loop) contributing to binding are involved (Fig. S4). A superimposition of the three structurally similar trimers (Fig. 3 A and B) reveals the structure identity of the S1 site, which is occupied by 2-OG in all three trimers, and the divergence in the S2 and S3 sites, respectively (Fig. 3B). The ligands are bound in S1 identical in topology to the mode described for the structure (Figs. 2 and 3B). The structure reveals that binding of the first 2-OG molecule to P_II (S1 site) generates unequal ligand-binding sites in the adjacent monomers, and, remarkably, sites not occupied by 2-OG also lack the Mg²⁺ ion. The conformational differences in the nonoccupied binding sites provide a structure-based explanation for the anticooperativity observed in biochemical experiments: After occupation of the first site, the K_d for the second site increases slightly, but after occupation of the second site, the K_d for the third site increases strongly (about 20-fold higher than the K_d for the first site; see Table 1). In , the ATP molecule attached to the S2 site exhibits a significantly altered conformation of the phosphate moiety (Fig. 3 B–D); furthermore, the T-loop basis is displaced and the C terminus is ordered similar to the S1 site (Fig. 3 C and D). The S2 site in resembles the S3 site of the structure, which, according to the sequential 2-OG–binding mode, corresponds to the lowest affinity site (for detailed comparison of the binding sites, see Fig. S3). The S3 site of exhibits further changes, visible most significantly in the T loop, the C terminus, and a small distortion in the β4-strand. Together these changes can lead to the strongly altered affinity of 2-OG toward the stereochemically differing S2 and S3 sites.

Fig. 3.

Anticooperativity of 2-OG–binding sites. (A) Top view of the structure as a ribbon plot and superposition of the (in blue), (in green), and (in orange) structures. The three ATP/2-OG–binding sites are marked by dashed circles and numbered (S1, S2, and S3). The picture on the right represents the cofactors bound in the individual sites (highlighted in yellow) with three ATP and 2-OG molecules in S1, three ATP and two 2-OG molecules in S2, and three ATP and one 2-OG molecules located in S3, respectively. The clockwise consecutive 120° binding into S1 → S2 → S3 sites is shown by an arrow. (B) Zoom in (side view) of the three binding sites after superposition of the molecules. The content of the individual binding sites is marked below the picture. T and B loops are marked with T and B, respectively, for clarity. (C) Binding sites S1, S2, and S3 of the structure. In the S1 site, ATP, 2-OG, and Mg²⁺ (green sphere) are bound, whereas S2 and S3 contain only ATP and no Mg²⁺. Significant changes in the C terminus and the T loop in site S2 are marked with numbered circles. (D) Superposition of effector molecules bound to sites S1, S2, and S3 in the structure. The ATP molecule observed in the S2 site is significantly distorted relative to that in S1 and S3.

Effect of 2-OG on the Dissociation of the P_II–NAGK Complex.

The structure of the P_II Mg-ATP/2OG complex suggests that 2-OG prevents interaction of P_II with NAGK by hindrance of the T loop folding into the tightly bent conformation needed for NAGK binding: The NAGK-bound P_II structure involves a salt bridge between K58 and E44 (18, 23), but because K58 is an important ligand for 2-OG, formation of this salt bridge is prevented. Furthermore, binding of 2-OG introduces a significant bend in the backbone of residues 38–43 (Fig. 1E) and together with the side chain of residue 42 induces a new T-loop conformation, which is incompatible with NAGK binding. When the P_II–NAGK complex has already been formed, is 2-OG still able to bind to P_II and antagonize the P_II–NAGK complex? Because this issue has not yet been investigated, the dissociation of the P_II–NAGK complex by 2-OG was studied. First, complex dissociation was directly recorded by surface plasmon resonance (SPR) spectroscopy (Fig. 4A). The P_II–NAGK complex was formed on the sensor chip, and, subsequently, 2-OG was injected (Fig. 4A, arrow) to dissociate the complex. No dissociation was observed in the presence of Mg-ATP alone; with 0.5 mM 2-OG, the complex decayed slowly with a rate of 1.8 × 10^-3 s^-1. With 1 mM 2-OG the decay rate increased to 9.0 × 10^-3 s^-1 and at 3 mM 2-OG to 28.6 × 10^-3 s^-1. By comparison, association of the complex was inhibited by much lower 2-OG concentrations, with an IC₅₀ of approximately 130 μM (18). In the second assay, the catalytic activity of NAGK/P_II in the presence of 50 μM arginine as an indicator of the degree of complex formation (24) was assayed (Fig. 4B). When 2-OG was added after formation of the P_II–NAGK complex, the inhibitory 2-OG concentration had an IC₅₀ of 0.9 mM. By contrast, addition of 2-OG to P_II prior to the addition of NAGK inhibited the activity with an IC₅₀ for 2-OG of approximately 120 μM (Fig. 4B, Inset) (18). Thus, 2-OG is able to dissociate the P_II–NAGK complex; however, one order of magnitude higher 2-OG concentrations are required to achieve dissociation compared to those required to inhibit association.

Fig. 4.

The 2-OG effect on P_II–NAGK complex dissociation. (A) SPR analysis of 2-OG-induced dissociation of NAGK–P_II complex in the presence of 1 mM ATP. The response difference (ΔRU) between flow cells FC2 and FC1 (control) is shown. After binding of 100 nM P_II to NAGK in FC2, a mixture of 1 mM ATP, 1 mM MgCl₂, and 2-oxoglutarate [at a concentration of 0 (solid line), 0.5 (dotted line), 1 (dashed line), and 3 mM (dot-dashed line), as indicated] was injected at the point indicated by an arrow. (B) The effect of 2-OG on NAGK activity in the presence of 50 μM arginine. Increasing 2-OG concentrations were added to reaction mixtures after the formation of the P_II–NAGK complex, and NAGK activity was determined as detailed in Materials and Methods. (Inset) Effect of 2-OG on NAGK activity in the presence of P_II, when 2-OG was preincubated with P_II.

Discussion

The structures presented here explain the known features of P_II-mediated 2-OG signaling: A Mg²⁺ ion, chelated by the phosphates of ATP, ligates carboxylate oxygens of 2-OG, and, therefore, Mg-ATP binding is the prerequisite for 2-OG binding to P_II. Binding of 2-OG to A. thaliana P_II in the presence of Mg-ADP could involve additional residues, possibly from its prolonged C-terminal segment, which contacts the effector molecule-binding site (22). The fact that all residues revealed here to be involved in 2-OG binding are highly conserved among P_II proteins (see also Fig. S1) strongly suggests that the reported mode of 2-OG binding could apply to all P_II proteins. The only contradiction is the previously reported structure of the P_II family member GlnK1 from M. jannaschii, where 2-OG bound from outside to the apex of a bent T loop (20). Because no biochemical evidence to support this binding mode was provided, it remains to be clarified whether this 2-OG binding mode is a peculiarity of the archaeal P_II protein or whether this mode of binding resulted from special crystallization conditions. In contrast, a recently described structure of the P_II homologue GlnZ from the proteobacterium A. brasiliense in complex with Mg-ATP and 2-OG revealed a mode of 2-OG binding, which is highly similar to the 2-OG binding described here, in particular the involvement of the Mg²⁺ ion and the highly conserved key residues Q39 and K58 (21). This one and our P_II structures perfectly agree with the properties of S. elongatus P_II variants bearing mutations in residues R9 and K58 described in this work and with the previously described I86N variant, displaying a closed 2-OG–binding site (18). Furthermore, they agree with previously reported properties of other P_II mutant variants. For instance, a Q39 mutation was shown to strongly impair 2-OG binding, whereas a deletion of the apical T-loop residues did not prevent 2-OG binding to Escherichia coli P_II (25). Furthermore, a K58 substitution abolished 2-OG signaling in Rhodospirillum rubrum P_II (26). Moreover, the actual structure reveals how precisely the carboxylate oxygens of 2-OG are probed by Mg²⁺ coordination and by interactions with protein backbone and side-chain atoms, explaining the high selectivity of P_II proteins for 2-OG (1, 16, 19). All together, these evidences strongly imply that the mode of 2-OG binding described here can be generalized for P_II proteins.

It has been shown that 2-OG controls P_II target interactions that involve the T loop, with X-ray structural information available for the S. elongatus and A. thaliana P_II–NAGK complex (22, 23), the S. elongatus P_II–PipX complex (27), and the E. coli GlnK–AmtB complex (28, 29). The mechanism of 2-OG-mediated P_II-target control was clarified here with the cognate P_II–NAGK complex. When P_II binds 2-OG, the base of the T loop (R38-G41) wraps around this metabolite, thereby adopting a unique retracted conformation. Furthermore, residues K58 and R9, which are involved in folding the T loop into the tightly bent conformation that fits into the NAGK crevice, perform ionic and H-bond interactions with the 2-OG γ-carboxylate oxygens, preventing formation of this fold. The IC₅₀ for 2-OG to inhibit P_II–NAGK association (120–130 μM, depending on the method) matches the dissociation constant of the third, low-affinity 2-OG–binding site (approximately 110 μM). This correlation implies that all three sites in P_II have to be occupied by 2-OG in order to inhibit NAGK binding. Consequently, P_II partially loaded with 2-OG should be able to bind NAGK, whereby 2-OG should be displaced from P_II. The driving force squeezing out 2-OG could be provided by the encounter complex between P_II and NAGK, which, according to a recent analysis, could be formed by an ionic interaction of the B-loop residue E85 of P_II with R233 of NAGK (18).

The present study also revealed how 2-OG dissociates the P_II–NAGK complex. As shown in Fig. 1C, 2-OG can access its binding site from the P_II periphery, which is not shielded by NAGK in the complex. However, approximately 10-fold higher concentrations of 2-OG are required to dissociate the P_II–NAGK complex than to inhibit its association. The difference could be explained by the 2-OG–binding site being closed in the P_II–NAGK complex by the tightly bent T loop. The 2-OG should unlock this compact conformation to gain access to its binding site, and this process probably requires much higher concentrations than binding to the open sites, which are accessible when P_II is not attached to NAGK.

The structure of the second cyanobacterial P_II target complex, P_II–PipX, has been determined recently (27). It reveals three PipX molecules bound on the flat bottom surface of the P_II body (orientation of Fig. 1), trapped between vertically extended T loops whose tip residues grasp the PipX monomers. Notably, this extended T-loop conformation is incompatible with the T-loop fold imposed by Mg-ATP-2-OG binding (see structure overlay in Fig. S5). Binding of 2-OG to the P_II–PipX complex will retract the extended T loop, releasing the bound PipX molecules, which explains the biochemical data, showing that binding of PipX to P_II is antagonized by Mg-ATP/2-OG (10). A similar antagonistic mechanism of Mg-ATP/2-OG can be assumed for the complex of the P_II family member GlnK with the ammonium transport channel AmtB, as deduced from the complex structure of the E. coli proteins (28, 29). In complex with AmtB, the T loop is in a vertically extended structure, resembling the T loop of S. elongatus P_II in complex with PipX. In the AmtB complex, GlnK residue Q39 interacts with K58 and ADP is bound to the adenylate-binding pocket (28, 29). Given that the binding mode of Mg-ATP/2-OG to GlnK is identical as outlined above, the resulting T-loop conformation will be incompatible with formation of the GlnK–AmtB complex (21). Studies with other E. coli P_II receptors such as NtrB imply that 2-OG does not always inhibit complex formation, but it may affect receptor activity at a postbinding step (16). In this case, it is conceivable that receptor binding occurs apart from the T loop (like the B-loop interaction of P_II with NAGK) and the conformational changes of the T loop imposed by 2-OG binding to P_II are transduced into conformational changes in the receptor, thereby altering its activity.

P_II proteins are highly sophisticated devices for measuring the concentration of central metabolites ATP, ADP, and 2-OG in an interdependent manner. This study reveals the mechanisms underlying this process. Binding of one, two, or three 2-OG molecules generates, via intersubunit communication, distinct structural states of P_II. Intermolecular signaling is based on the highly conserved trimeric architecture of the P_II proteins. The β2-strands, which directly connect the three binding sites, could play an important role. Binding of 2-OG to one site affects the two neighboring sites asymmetrically, generating the anticooperativity that allows metabolite sensing in a wide concentration range. Moreover, the free site in clockwise orientation displays a characteristic T-loop structure. P_II receptors perceive the signal via intimate T-loop interactions, which affect binding or influence the receptor at a postbinding stage (16, 18). This mode of signal transduction by P_II is unique, and the complexity of interactions explains the remarkably high conservation of P_II proteins.

Materials and Methods

Full protocols are available in SI Materials and Methods.

Overexpression and Purification of Recombinant P_II and NAGK.

The R9L and K58M variants were created with artificial glnB genes carrying the respective mutations and cloned into the Strep-tag fusion vector pASK-IBA3 (IBA) after restriction with BsaI as described previously (7). Overexpression of wild-type and mutant S. elongatus glnB in E. coli RB9060 (30) and purification of recombinant P_II proteins with a C-terminal-fused Strep-tag II peptide were performed according to Heinrich et al. (7). His₆-tagged recombinant NAGK from S. elongatus was overexpressed in E. coli strain BL21(DE3) (31) and purified as reported previously (8).

SPR Detection.

SPR experiments were performed by using a BIAcore X biosensor system (GE Healthcare) at 25 °C in Hepes-buffered saline-Mg buffer as described previously (8). In order to analyze the effect of 2-OG on the dissociation of the P_II–NAGK complex, 100 nM P_II was first bound to immobilized His₆–NAGK in flow cell 2 (FC2) (ascending curves). Subsequently, 50 μL buffer containing 1 mM ATP and different concentrations of 2-OG was injected (start of injection indicated by the arrow). Binding and dissociation of P_II to NAGK was recorded as the response signal difference (ΔRU) of FC2-FC1; FC1, reference cell without His₆–NAGK.

ITC.

ITC experiments were performed on a VP-ITC microcalorimeter (MicroCal, LCC) in ITC buffer containing 10 mM Hepes-NaOH, pH 7.4, 50 mM KCl, 50 mM NaCl, and 1 mM MgCl₂ at 20 °C as described previously (18). For determination of ATP- and 2-OG–binding isotherms for P_II variants R9L and K58M, solutions with different protein concentration were titrated with 1 mM ATP or 4 mM 2-OG (in the presence of 1 mM ATP). The binding isotherms were calculated from received data and fitted to a three-site binding model using the MicroCal ORIGIN software (Northampton) as indicated.

Coupled NAGK Activity Assay.

Activity of NAGK was assayed by a coupled assay (32) with modifications as described previously (24), in the buffer consisting of 50 mM imidazole, pH 7.5, 50 mM KCl, 20 mM MgCl₂, 0.4 mM NADH, 1 mM phosphoenolpyruvate, 10 mM ATP, 0.5 mM DTT, 11 U lactate dehydrogenase, 15 U pyruvate kinase, 50 μM arginine, 1.2 μg P_II, and 3 μg NAGK. The mixture was preincubated for 3 min to allow P_II–NAGK complex formation. Then the reaction was started by the addition of 50 mM NAG and 2-OG (to determine the effect of increasing 2-OG concentrations on disruption of P_II–NAGK complex in the presence of NAGK-inhibiting concentrations of arginine). Then, 20 s after addition of substrate, the change in absorbance at 340 nm was recorded for 10 min. Linear kinetics were observed over that period of time.

Crystallization of Recombinant S. elongatus P_II Protein.

Crystallization was performed with the sitting-drop technique by mixing 400 nL of the protein solution with equal amounts of the reservoir solution by using the honeybee robot (Genomic Solutions Ltd). Drops were incubated at 20 °C, and pictures were recorded by the RockImager system (Formulatrix). The crystallization buffer was composed of 10 mM Tris (pH 7.4), 0.5 mM EDTA, 100 mM NaCl, 1% glycerol, and 2 mM ATP-Mg, and also 2 mM 2-OG was added; crystals appeared in a precipitant condition containing PEG 4000. Glycerol was used as the cryoprotectant, and the crystals were flash-frozen in liquid nitrogen. Diffraction data were collected at the Swiss Light Source. Diffraction images were recorded on a MarCCD camera 225 (Marresearch), and images were processed by using the XDS/XSCALE software (33). The structure was solved by molecular replacement using the program Molrep (34). Rebuilding of the structure and structure refinement was performed by using the programs Coot and Refmac (35, 36). The quality of the structure was analyzed by the Procheck program (37). Figures were generated by using PyMOL (www.pymol.org).