The Emergence of 2-Oxoglutarate as a Master Regulator Metabolite

SUMMARY

The metabolite 2-oxoglutarate (also known as α-ketoglutarate, 2-ketoglutaric acid, or oxoglutaric acid) lies at the intersection between the carbon and nitrogen metabolic pathways. This compound is a key intermediate of one of the most fundamental biochemical pathways in carbon metabolism, the tricarboxylic acid (TCA) cycle. In addition, 2-oxoglutarate also acts as the major carbon skeleton for nitrogen-assimilatory reactions. Experimental data support the conclusion that intracellular levels of 2-oxoglutarate fluctuate according to nitrogen and carbon availability. This review summarizes how nature has capitalized on the ability of 2-oxoglutarate to reflect cellular nutritional status through evolution of a variety of 2-oxoglutarate-sensing regulatory proteins. The number of metabolic pathways known to be regulated by 2-oxoglutarate levels has increased significantly in recent years. The signaling properties of 2-oxoglutarate are highlighted by the fact that this metabolite regulates the synthesis of the well-established master signaling molecule, cyclic AMP (cAMP), in Escherichia coli.

INTRODUCTION

In order to cope with fluctuations in the availability of nutrients, organisms have developed a plethora of signal transduction systems to sense the prevailing nutritional status and generate the appropriate metabolic response. The two major nutrients needed by living organisms are carbon and nitrogen. Despite extensive knowledge on how both carbon and nitrogen metabolic pathways are regulated, one of the most fundamental questions that remain to be answered is how organisms, particularly microbes, coordinate nitrogen and carbon metabolism in order to maximize nutrient utilization and cell growth. An ideal signaling metabolite to coordinate the regulation of nitrogen and carbon metabolism should lie at the intersection between these two metabolic pathways. It has long been recognized that the tricarboxylic acid (TCA) cycle intermediate 2-oxoglutarate (2-OG) could, in theory, fulfill such a regulatory role, as 2-OG is used as the major carbon skeleton in nitrogen-assimilatory reactions (1). However, only recently have experimental data confirmed that 2-OG acts as a master regulatory metabolite.

This review aims to summarize recent advances in understanding the biological role of 2-OG as both a metabolic intermediate and an important signaling molecule. We focus primarily on Escherichia coli because the physiology and systems biology of metabolic control have been studied intensively in this model organism. In the first part of this review, we describe the roles of 2-OG as a TCA cycle intermediate and as the carbon skeleton for nitrogen assimilation reactions. This is followed by a review of methods and challenges of estimating 2-OG levels and a description of the major 2-OG-sensing proteins in prokaryotes. We then review the novel interplay between the 2-OG and cyclic AMP (cAMP) levels in E. coli. Finally, we briefly summarize the emerging role of 2-OG as a master regulator in eukaryotes and discuss the major points and challenges in 2-OG signaling that remain to be tackled.

METABOLIC ROLES OF 2-OG IN ENERGY PRODUCTION AND NITROGEN ASSIMILATION

When E. coli uses glucose, fatty acids, or some amino acids as a carbon source under aerobic conditions, these molecules are oxidized to acetyl-coenzyme A (acetyl-CoA), which is fed into the TCA cycle (Fig. 1). The acetyl group from acetyl-CoA is oxidized to two CO₂ molecules, and the energy derived from these reactions is conserved either in reduced electron carriers or in ATP. The TCA cycle also carries out important anabolic functions; for instance, 2-oxoglutarate and oxaloacetate act as precursors for the synthesis of amino acids and nucleotides (see below). It is also notable that alternative TCA cycles operate in some prokaryotes. These alternative pathways are likely to occur in organisms that were previously thought to have incomplete TCA cycles due to the lack of 2-OG dehydrogenase (2–4).

FIG 1.

Schematic representation of the TCA cycle and ammonium-assimilatory reactions in E. coli. Note that the GDH and GOGAT reactions produce one and two glutamate molecules, respectively (denoted by “1×” and “2×”). (Adapted from data in reference 139.)

Ammonium is the preferred nitrogen source for E. coli under aerobic conditions, as it supports high growth rates; this is also the case for many other microorganisms (5). Ammonium is assimilated into glutamine and glutamate; these two amino acids act as intracellular nitrogen donors to produce other nitrogen-containing organic compounds through transamination and transamidation reactions (5, 6). The synthesis of glutamine in E. coli and other bacteria occurs through glutamine synthetase (GS); this enzyme produces glutamine from ammonium and glutamate, in a reaction driven by the hydrolysis of one ATP molecule per one NH₄⁺ ion assimilated (Fig. 1). In E. coli, the productive synthesis of glutamate from ammonium can occur by the following two different pathways: (i) combining the glutamine produced by GS with 2-OG, through the action of the enzyme glutamate synthase (GOGAT), resulting in the production of two glutamates (Fig. 1); and (ii) directly assimilating ammonium into 2-OG by the action of glutamate dehydrogenase (GDH), producing one glutamate (Fig. 1). The GDH pathway has been presumed to be used when ammonium is abundant, as its K_m for ammonium is over 1 mM (7), whereas GS has a lower K_m for ammonium (∼0.1 mM) (8). However, kinetic flux profiling experiments showed that even with high ammonium availability (10 mM), the GS-GOGAT pathway is the major ammonium-assimilatory route in E. coli (9). Another difference between the GS-GOGAT and GDH pathways is their respective K_m values for 2-OG, which are ∼0.24 mM for GOGAT (10) and ∼0.64 mM for GDH (7, 11). As discussed below, 2-OG levels can fluctuate to levels below 0.1 mM; therefore, the GS-GOGAT pathway may be predominant under 2-OG deprivation conditions. Note that in either the GDH or GS-GOGAT pathway, the carbon receptor of the nitrogen atom is the TCA cycle intermediate 2-OG (Fig. 1). Therefore, 2-OG is at the intersection between carbon and nitrogen metabolism, connecting the catabolic function of the TCA cycle with the anabolic function of the nitrogen assimilation reactions.

FLUCTUATIONS IN THE 2-OG POOL IN RESPONSE TO NITROGEN AVAILABILITY

The determination of absolute metabolite concentrations within living cells is challenging. Many metabolites, such as 2-OG, turn over very quickly and may leak or degrade during extraction; hence, the sampling process must be quick to avoid perturbations in metabolite levels and to maintain metabolite stability (12–14). Furthermore, conversion of measured metabolite levels to molar concentrations within the cell is complicated by a number of factors, e.g., the internal cell volume may fluctuate (15), and it is hard to predict how the measured values correlate to free metabolites, as a fraction of metabolites are likely to be bound to proteins. Despite these challenges, many studies have determined the levels of 2-OG in E. coli cell extracts by applying either coupled enzymatic assays (16–18), mass spectrometry (9, 19–21), or high-pressure liquid chromatography (HPLC) methods (13). More recently, engineered fluorescent protein probes, or biosensors, have been developed to allow monitoring of 2-OG levels in cell extracts or directly inside living cells (22–24).

Because 2-OG acts as the carbon skeleton for the nitrogen assimilation reactions by either the GS-GOGAT or GDH pathway (Fig. 1), it is expected that the level of this metabolite will fluctuate according to the availability of ammonium. Indeed, pioneer studies demonstrated that the 2-OG concentration increases upon ammonium limitation under steady-state growth conditions (16, 17). Follow-up studies analyzed the dynamics of the 2-OG levels upon rapid shifts in ammonium concentration. In one case, the 2-OG level decreased from ∼1.4 mM under nitrogen starvation conditions to ∼0.3 mM just 1 min after the addition of 200 μM NH₄Cl to the cell culture, returning to the initial 2-OG level after 15 min (18). In another study, the 2-OG level decreased from ∼12 mM under nitrogen starvation conditions to ∼0.6 mM 3 min after the addition of 10 mM NH₄Cl (9). Conversely, E. coli cultivated with a high ammonium level (10 mM NH₄Cl) exhibited an increase in intracellular 2-OG, from ∼0.5 mM to ∼2.5 mM, 15 s after ammonium was removed from the medium (13). These experiments with E. coli showed that there is an inverse correlation between ammonium availability and 2-OG accumulation. The same correlation applies to organisms such as archaea (25), cyanobacteria (26, 27), and the yeast Saccharomyces cerevisiae (19). Although, as mentioned above, precise quantitation of metabolite levels in living cells is challenging, these studies indicate that 2-OG is well suited to act as a metabolic signal of small and transient fluctuations in ammonium availability in a wide range of microbes. As pointed out below, the relationship between 2-OG levels and the nitrogen status has allowed the evolution of a diverse range of 2-OG sensor proteins that coordinate nitrogen and carbon metabolism in many prokaryotes.

It is worth mentioning that 87% of the 2-OG drained from the TCA cycle during ammonium assimilation by either GDH or GOGAT (Fig. 1) is estimated to be recycled by transamination (Fig. 1, red line) (5). The remaining 13% should be replenished by anaplerotic reactions (see below). Hence, the decrease in 2-OG that occurs when N-starved E. coli receives an ammonium shock should be transient, returning to initial levels after ammonium consumption, and indeed this is what is observed experimentally (18). Rapid reestablishment of 2-OG levels after ammonium consumption may be facilitated by increased transamination (Fig. 1, red line), particularly through the enzyme aspartate aminotransferase, with the equilibrium driven toward the forward reaction (net aspartate production) by low levels of 2-OG (9). Despite the partial replenishment of the 2-OG pool after transamination, productive de novo synthesis of 2-OG must occur to replenish the 2-OG that was consumed during nitrogen assimilation but was not recycled. Given the cyclic nature of the TCA cycle, in principle, the carbon skeleton of any of the TCA cycle intermediates can be transformed into 2-OG. Reactions that fill the TCA cycle with carbon are known as anaplerotic reactions. In E. coli, anaplerosis is accomplished by phosphoenolpyruvate (PEP) carboxylase, which converts PEP to the TCA cycle intermediate oxaloacetate. E. coli can also replenish TCA cycle intermediates by using the glyoxylate shunt. However, the activity of the glyoxylate shunt is regulated by carbon catabolic repression (CCR) (28), and also by reversible phosphorylation of isocitrate dehydrogenase (29).

Apart from the de novo synthesis of 2-OG, E. coli can take up 2-OG from the medium by using a constitutively expressed 2-OG/H⁺ symport permease system encoded by kgtP (30). Direct uptake from the medium has been confirmed using an in vivo 2-OG biosensor (22, 31). However, 2-OG seems to leak from E. coli cells under certain culture conditions (13). Interestingly, in Xanthomonas oryzae, the kgtP-encoded transporter plays an important role as a plant virulence effector, since it is secreted by the type III secretion system to the plant cell membrane, where it may be involved in 2-OG acquisition from host rice cells (32).

FLUCTUATIONS IN THE 2-OG POOL IN RESPONSE TO CARBON AVAILABILITY

There is increasing evidence that intracellular 2-OG levels are sensitive to variations in the carbon supply in E. coli. E. coli cells cultured in glycerol have an intracellular 2-OG concentration of ∼0.5 mM as determined by HPLC. When these cells were washed with medium without glycerol, the intracellular 2-OG level dropped to ∼0.3 mM within 15 s (13). In a separate study, the intracellular concentration of 2-OG (as measured by HPLC) was 0.35 mM for E. coli incubated in carbon-free M9 medium, and this increased to 2.6 mM 2-OG 30 min after the addition of 10 mM glucose to the medium (22). Reductions in intracellular 2-OG levels when E. coli is starved for carbon have also been detected by mass spectrometry (14, 19, 33). These data have been corroborated by studies using intracellular 2-OG biosensors. Zhang and colleagues developed an in vivo 2-OG sensor based on fluorescence resonance energy transfer (FRET) by inserting the 2-OG binding domain of the Azotobacter vinelandii NifA protein (see “The Nitrogenase Transcriptional Regulator Protein NifA”) between the FRET pair yellow fluorescent protein (YFP)-cyan fluorescent protein (CFP) (22, 31). This biosensor enabled monitoring of fluctuations in 2-OG levels in real time in response to variations in the carbon supply. These experiments showed that there is a quick and significant increase in the 2-OG level when glucose is added to carbon-starved cultures of E. coli (22, 31). In summary, evidence is building to support the notion that intracellular 2-OG levels may be a good indicator of the carbon supply in E. coli. This hypothesis is further supported by recent findings demonstrating that the 2-OG level coordinates multiple carbon-dedicated pathways, such as fatty acid production and carbohydrate uptake systems (see the following sections).

PROTEINS THAT ACT AS 2-OG SENSORS

Given the key role played by 2-OG in central carbon and nitrogen metabolism and the ability of this metabolite to reflect the balance between the nitrogen and carbon nutritional statuses, it is not surprising that organisms have evolved a variety of proteins that are able to sense 2-OG levels. The best-described proteins that act as 2-OG sensors in prokaryotes are reviewed in this section.

The P_II Protein Family

Probably the most ancient sensors of 2-OG are proteins belonging to the P_II family. The P_II regulatory protein was initially identified through classical biochemical studies of the regulation of glutamine synthetase in E. coli and was the first protein to be recognized as a 2-OG sensor (34). P_II proteins are well conserved and widely distributed in bacteria, in archaea, and in the chloroplasts of plants (35). Since this protein family was recently reviewed comprehensively (36–38), in this review we briefly describe the P_II proteins and then focus on the regulation of acetyl-CoA carboxylase (ACC) by P_II.

Typical P_II proteins are homotrimers that form a barrel-like structure; each subunit carries a long and solvent-exposed loop, called the T loop, which is vital for P_II protein function. The three clefts formed between the P_II subunits can be occupied competitively by either ATP or ADP, suggesting that P_II proteins can potentially act as energy sensors in vivo (39). Each P_II trimer can also bind up to three 2-OG molecules; the 2-OG binding sites are formed between P_II subunits and require the previous binding of MgATP (Fig. 2) (40). Binding of MgATP, ADP, or MgATP plus 2-OG alters the P_II structure and the ability of P_II to bind and modulate the activity of a variety of target proteins (37, 41). Some P_II proteins, like E. coli GlnB, show strong negative cooperativity for the occupation of the three 2-OG binding sites when MgATP is saturating. The first 2-OG molecule binds with a K_d (dissociation constant) in the low micromolar range, and consecutive occupations of the second and third 2-OG binding sites exhibit K_ds that are 1 to 2 orders of magnitude higher (42, 43). This behavior makes P_II an ideal biological sensor of 2-OG with the ability to respond to a wide range of 2-OG concentrations (44). However, it should be noted that cooperative binding of 2-OG to P_II proteins is not a universal feature, since in some cases calorimetric titrations fit only a “one-set-of-sites” binding model (41, 45), perhaps reflecting a physiological requirement to respond to a more restricted range of 2-OG concentrations.

FIG 2.

P_II protein structure. (A) P_II trimer. The binding sites for MgATP (magenta) and 2-OG (black) at the lateral clefts between each subunit are indicated. (B) Closer view of the 2-OG binding site. 2-OG interacts with the side chains of conserved lysine (K58) and glutamine (Q39) residues (green sticks) and also with MgATP. 2-OG binding affects the P_II T-loop structure, thereby altering the affinity between P_II and its binding protein targets. In proteobacteria, the structure of the T loop is also influenced by reversible uridylylation at the conserved tyrosine residue indicated (Y51). The figure was prepared using Pymol and PDB entry 3MHY.

Another interesting characteristic of P_II is its potential to integrate the 2-OG signal with energy sensing. Given that MgATP and ADP bind to P_II competitively and that 2-OG binding requires preoccupancy with MgATP, an increased apparent affinity for 2-OG is observed at high ATP/ADP ratios (43, 46, 47). This interplay between the ATP/ADP ratio and 2-OG binding is further complicated by the fact that some P_II members apparently function as an ATPase whose activity is regulated by 2-OG levels (48).

The P_II protein structure can be affected further by reversible covalent modification at a conserved tyrosine residue located at the top of each of its three T loops (Fig. 2). In proteobacteria, P_II proteins are found in uridylylated forms under nitrogen-limiting conditions and are rapidly deuridylylated under conditions of nitrogen excess (37, 49). In most cases studied so far, the nitrogen signaling molecule glutamine controls P_II uridylylation through allosteric regulation of the bifunctional uridylyl transferase/uridylyl-removing enzyme GlnD (50). Covalent modification by GlnD is modulated by binding of the ligands ATP, ADP, and 2-OG to P_II (43, 47, 51, 52).

Regulation of acetyl-CoA carboxylase by P_II and 2-OG.

Since P_II is able to sense important metabolites, it is not surprising that P_II proteins control the activity of a vast range of target proteins (37, 53). In E. coli, the archetypical function of P_II is the regulation of nitrogen assimilation. It fulfils this function by controlling the activity of the ammonium transporter AmtB (54) and the expression and activity of the glutamine synthetase enzyme (50, 55). More recently, it was also observed that E. coli GlnB can modulate the activity of an enzyme dedicated to the carbon metabolism, i.e., acetyl-CoA carboxylase (ACC) (56).

ACC catalyzes the first and committed step in fatty acid biosynthesis, the carboxylation of acetyl-CoA to generate malonyl-CoA. ACC is divided into the following three functional modules: the biotin carboxylase (BC), the carboxyltransferase (CT), and the biotin carboxyl carrier protein (BCCP) (57). BC catalyzes the first half-reaction, the carboxylation of a biotin group attached to BCCP, using bicarbonate and ATP as substrates. The second half-reaction is catalyzed by CT, which transfers the carboxyl from carboxyl-biotin to acetyl-CoA to generate malonyl-CoA (57, 58). Typical ACC enzymes can be divided into two classes. Those found in most eukaryotes have all three functional modules within a single polypeptide. On the other hand, in most prokaryotes and in the chloroplasts of most plants, ACC is formed by combining different polypeptides, corresponding to the BC, CT, and BCCP functional modules (58).

Searches for novel targets of P_II in the Arabidopsis thaliana chloroplast and in the bacterium Azospirillum brasilense both led to the identification of BCCP as a binding partner of P_II (59, 60). This interaction was further observed in E. coli by use of purified proteins. The detection of the BCCP-P_II interaction in distantly related organisms supports the hypothesis that BCCP is an ancient binding partner of P_II (60). The functional role of the BCCP-P_II interaction is also conserved. Measurements of ACC activity using A. thaliana chloroplast extracts or using purified E. coli ACC came to the same conclusion: P_II binding to BCCP leads to ACC inhibition by reducing the ACC turnover rate (Fig. 3). The BCCP-P_II interaction and ACC inhibition are relieved by 2-OG in a dose-dependent manner (56). Hence, P_II acts as a 2-oxoglutarate-sensitive dissociable regulatory subunit of ACC (Fig. 3). In vivo studies of plants have corroborated these observations. A knockout mutation in the gene encoding P_II increased the accumulation of fatty acids in A. thaliana seeds (61). These studies support a model in which 2-OG levels act as a conserved switch to regulate ACC activity, and thus fatty acid biosynthesis, in bacteria and plants.

FIG 3.

GlnB acts as a 2-OG-sensitive dissociable regulatory subunit of acetyl-CoA carboxylase (ACC). (A) With low 2-OG levels, GlnB is bound to 3 MgATP molecules (magenta) and interacts with the BCCP component of ACC, turning down the enzyme activity. (B) When the 2-OG concentration increases, the three 2-OG binding sites of GlnB are occupied (dark blue circles), resulting in a GlnB T-loop structure that abrogates interaction with and inhibition of ACC.

The Nitrogenase Transcriptional Regulator Protein NifA

P_II proteins are involved in the regulation of nitrogen fixation in both archaea and bacteria, affecting the transcription of nitrogen fixation genes (nif genes) and/or nitrogenase activity (36, 53, 62). In proteobacteria, P_II influences nif gene expression by controlling the activity of a transcriptional activator named NifA in response to the level of fixed nitrogen. The N status can thus be conveyed to NifA via the uridylylation state of P_II proteins and by the capacity of these signal transduction proteins to sense 2-OG levels as described above. The potential for P_II proteins to sense the adenylate energy charge (ATP/ADP ratio) may also be important in this context, as nitrogen fixation is a very energy-intensive process.

In some diazotrophs, the activity of NifA is regulated by stoichiometric interaction with a partner regulatory protein, NifL, which senses the redox status and inhibits NifA when the oxygen concentration is unfavorable for nitrogen fixation (63–65). The P_II protein GlnK modulates the interaction between NifL and NifA in response to the N status (66). In the model diazotroph Azotobacter vinelandii, deuridylylated GlnK interacts with NifL under conditions with high glutamine levels, resulting in the formation of a GlnK-NifL-NifA ternary complex in which NifA is inactivated under conditions of N excess (45, 67). In contrast, with low glutamine levels, uridylylated GlnK does not interact with NifL, consequently allowing NifA activity under N-limiting conditions (Fig. 4). Unlike the E. coli P_II proteins, A. vinelandii GlnK does not exhibit strong negative cooperativity with respect to the binding of 2-OG. Surprisingly, the interaction between nonmodified GlnK and NifL is maintained even at a high 2-OG concentration (2 mM) in vitro, suggesting that this target interaction is regulated primarily by the uridylylation state of GlnK (45).

FIG 4.

Involvement of 2-OG and glutamine in the regulation of NifA activity by NifL and GlnK. (A) At low-2-OG concentrations, NifL forms a complex with NifA and NifA activity is inhibited. (B) At high concentrations, 2-OG (dark blue circles) binds to NifA to prevent inhibition by NifL. At low glutamine concentrations, GlnK is uridylylated and unable to interact with NifL. At high glutamine concentrations, GlnK is deuridylylated, and the nonmodified form of GlnK interacts with NifL, resulting in the formation of a ternary GlnK-NifL-NifA complex. This complex is formed even when GlnK is saturated with MgATP (magenta circles) and 2-OG (dark blue circles).

Most NifA proteins contain an N-terminal GAF domain, which apparently regulates the catalytic activity and/or the DNA binding functions of the activator. In the case of A. vinelandii NifA, the GAF domain regulates the interaction with NifL in response to the concentration of 2-OG, which binds directly to the GAF domain, with a dissociation constant of ∼60 μM (68). When NifL is absent, binding of 2-OG to the GAF domain of the isolated NifA protein does not influence its activity. However, both the oxidized and reduced forms of NifL inhibit NifA activity in vitro in the absence of 2-OG (68, 69) (Fig. 4). Further support for a physiological role for 2-OG comes from the observation that a substitution in the GAF domain which prevents 2-OG binding to NifA obviates escape from inhibition by NifL (70).

The ability of 2-OG to prevent the interaction between NifL and NifA is completely overridden when either the PAS domain of NifL becomes oxidized or GlnK is deuridylylated. Hence, 2-OG is effective as an allosteric regulator only under reducing, N-limiting conditions. This may explain why the GlnK-NifL interaction is relatively insensitive to 2-OG, as this will permit maintenance of the inhibitory GlnK-NifL-NifA ternary complex even at high 2-OG concentrations (45). Overall, the function of 2-OG in this regulatory system is to antagonize inhibition by NifL under conditions that are favorable for nitrogen fixation (Fig. 4). This would potentially enable sensing of the carbon status by NifA in order to ensure that the high carbon (energy) demand for nitrogen fixation can be met. However, it is also possible that 2-OG sensing by NifA provides a backup mechanism for sensing the nitrogen status.

In diazotrophic bacteria that do not contain NifL, the response to the fixed nitrogen status is mediated by direct interaction of P_II proteins with the GAF domain of NifA. In some organisms, for example, A. brasilense, Herbaspirillum seropedicae, and Rhodospirillum rubrum, the uridylylated form of the P_II protein is apparently required to prevent intramolecular repression of NifA activity by the GAF domain under nitrogen-limiting conditions (71–73). In contrast, in Rhodobacter capsulatus and Azorhizobium caulinodans, P_II proteins are required to inactivate NifA under conditions of nitrogen excess (74, 75). However, the role of 2-OG in these interactions is not well understood.

The NtcA Transcriptional Regulator in Cyanobacteria

In cyanobacteria, some of the functions of the P_II protein in regulating nitrogen assimilation have been taken over by NtcA, a global transcriptional regulator belonging to the Crp-Fnr family (76–78) that senses the N status. As noted above, in cyanobacteria, high 2-OG levels are indicative of nitrogen deficiency (26). 2-OG binds to NtcA and acts as an allosteric effector by increasing the affinity of NtcA for specific DNA binding targets, resulting in transcriptional activation in vitro (79–81). Canonical NtcA binding sites are palindromic and, in promoters subject to activation, are frequently centered at position ∼−41.5 with respect to the transcriptional start site (77, 82), similar to class II promoters activated by Crp and Fnr (83). However, unlike Crp, NtcA can also act as a repressor.

Crystal structures of NtcA in both the apo and 2-OG-bound forms reveal, as expected, that NtcA has a fold similar to that of the cAMP-binding protein Crp and that the NtcA subunits form a tight dimer in the asymmetric unit (84, 85). Both proteins contain an effector binding domain (EBD) adopting a β-barrel motif connected to a DNA binding domain via a long α-helix (the C-helix). The positioning of the ligand in the EBD is similar in both proteins, and they even share some binding residues (84, 85). However, the modes of allosteric activation by the ligands appear to be different. In the case of Crp, the binding of cAMP to the EBD causes an extension of the C-helices, resulting in rotation and appropriate positioning of the DNA recognition helices to enable stable DNA binding (86, 87). In contrast, in NtcA signal transmission, 2-OG binding to the EBD results in a relatively small change in the conformation of the bridging C-helices, forming a tightly coiled coil, which shortens the distance between the two recognition helices in the DBD to adopt a more favorable DNA binding conformation (Fig. 5). This may explain why binding of 2-OG to NtcA results in a relatively small enhancement in the affinity for DNA, compared with a large change in affinity brought about by the binding of cAMP to Crp (85, 88). Structural characterization of the 2-OG binding site in NtcA has been valuable for designing chemical derivatives of 2-OG that mimic its signaling function in vivo (89).

FIG 5.

Role of 2-OG in partner switching by PipX and coactivation of NtcA. (A) At low 2-OG concentrations, PipX binds to P_II and NtcA is primarily in the apo form (not bound to 2-OG). Binding of ADP to P_II (green circles) also favors formation of the P_II-PipX complex. (B) At high concentrations of 2-OG, P_II is saturated with 2-OG (dark blue circles) and ATP (magenta circles), disfavoring its interaction with PipX. Conversely, the binding of 2-OG to NtcA favors its interaction with PipX, resulting in an enhancement of transcriptional activation. Structures of protein complexes and unbound proteins are shown beneath the schematic. PipX is shown as a surface structure, and 2-OG is shown as blue and red spheres. Binding of 2-OG to NtcA results in realignment of the recognition helices to increase its affinity for DNA. Binding of 2-OG to P_II retracts the extended T loop, inhibiting complex formation with PipX. The PDB codes for the structures are as follows: 2XG8 (for P_II-PipX; note that this complex consists of trimeric P_II bound to a trimer of PipX), 3LA7 (apo-NtcA), 2XUL (P_II with high 2-OG level), and 2XKO (PipX-NtcA, which comprises a dimer of PipX bound to a dimer of NtcA).

The rather simplistic model in which the binding of 2-OG to NtcA influences its affinity for DNA and, consequently, transcriptional activation has been complicated by the discovery of the protein PipX as the binding partner of NtcA. PipX acts as a coactivator of NtcA and links P_II-mediated signaling with NtcA-regulated gene expression via a partner-swapping mechanism (90). PipX was originally identified as a P_II-interacting protein (91) but was subsequently shown to interact with NtcA, in addition to P_II, in a 2-OG-dependent manner (90). Whereas the PipX-NtcA complex is stabilized at high 2-OG levels, the interaction between PipX and P_II is impaired by the presence of 2-OG and ATP (Fig. 5). This leads to a model in which PipX is bound to P_II under conditions of nitrogen sufficiency (low 2-OG) and coactivates NtcA under nitrogen-deficient conditions (high 2-OG) (90) (Fig. 5). Intriguingly, the 2-OG-controlled partner swapping of PipX between NtcA and P_II is modulated by subtle fluctuations in ADP levels, which favor formation of the P_II-PipX complex at high ADP concentrations. This results in a turning down of the 2-OG signal by ADP so that P_II can efficiently compete with NtcA for PipX and deactivate this transcription factor in the presence of 2-OG (92). This highly sophisticated partner-switching mechanism results in fine-tuning of NtcA-dependent gene expression (77, 93). The crystal structure of the PipX-NtcA complex contains two monomers of PipX interacting with each subunit of the 2-OG-bound NtcA dimer. Domain interactions in the activated form of NtcA are stabilized by the binding of PipX, which suggests that PipX drives the equilibrium toward the active conformation of NtcA (84, 94). In agreement with this, biochemical studies revealed that PipX increases the affinity of NtcA for promoters and the effective affinity of NtcA for 2-OG (88). Modeling of the PipX-NtcA complex on DNA suggests that PipX may also enhance NtcA-dependent transcription through interaction with the C-terminal domain of the α-subunit of RNA polymerase (84).

The Archaeal Transcriptional Repressor NrpR

In the archaeon Methanococcus maripaludis, genes required for nitrogen assimilation under nitrogen-deficient conditions are subject to regulation by the transcription factor NrpR. This regulatory protein represses the expression of the nitrogen fixation (nif) and glutamine synthetase (glnA) genes by binding to operator sequences when fixed nitrogen is abundant (95). The affinity of NrpR for the operator sites is lowered under nitrogen deficiency conditions due to increased 2-OG levels (25, 96). Homologues of M. maripaludis NrpR are widely distributed in the Euryarchaeota and also found in a few bacterial species. However, the domain architecture of these NrpR homologues is not conserved, and they can be classified into three different groups (97). In M. maripaludis, NrpR has an N-terminal helix-turn-helix (HTH) DNA binding domain fused to two consecutive NrpR domains (Fig. 6). Other organisms, such as Archaeoglobus fulgidus, encode an NrpR consisting of an N-terminal HTH linked to a single NrpR-like domain. A third class is found in organisms, such as Methanosarcina mazei, which encode two different NrpR polypeptides, one similar to the second group (HTH fused to a single NrpR domain; NrpR1) and a second polypeptide carrying a single NrpR domain without a DNA binding motif, named NrpR2 (97).

FIG 6.

Effect of 2-OG on NrpR structure and DNA binding activity. The crystal structure of the NRD2 domain of NrpR (PDB code 4NEX; rainbow ribbon) is shown, fitted into the electron microscopy electron density maps of full-length M. maripaludis NrpR (green mesh) in both the apo form (EMD2221; left) and the 2-OG-bound form (EMD2222; right). In the apo state, the two HTH DNA binding domains of NrpR (not shown) are presumed to be positioned in the correct orientation and distance to enable binding of NrpR to two adjacent major grooves (highlighted with yellow dotted squares) in the operator sequence (modeled as B-form DNA and shown above the protein density). The conformational change induced by the binding of 2-OG converts NrpR from a U-shaped molecule to a trapezium-shaped particle in which the two DNA binding domains are too far apart to recognize the operator sequence, and consequently, transcription of nitrogen assimilation genes is derepressed.

Genome-wide expression analysis of an M. mazei NrpR1 knockout strain showed elevated transcription of 27 genes under nitrogen sufficiency conditions, including nifH, glnA1, and glnK1 (98). The M. mazei NrpR1 and NrpR2 proteins physically interact and bind as a complex to the nifH and glnK1 promoters in the presence of low 2-OG levels (99). The NrpR2 polypeptide also interacts with the general transcriptional factors (TBP and TFB) in the presence of low 2-OG levels. Hence, the NrpR1-NrpR2 complex binds to operator sequences when the 2-OG concentration is low, through both the NrpR1 DNA-binding HTH motif and interactions between NrpR2 and the general transcriptional factors; this allows repression of nitrogen assimilation genes under nitrogen-sufficient conditions (99).

The structure of one of the NrpR-like domains of the M. maripaludis NrpR protein has been solved. This domain forms a dimer in which each monomer has a cleft predicted to bind 2-OG, as suggested by molecular docking analysis (100). The putative NrpR 2-OG binding site is conserved among other 2-OG binding proteins. It was observed that Asn, Tyr, Ser, and Arg residues are commonly found in 2-OG binding sites (100). A model structure of the full-length NrpR protein of M. maripaludis was fitted into the electron microscopy electron density maps obtained in the presence or absence of 2-OG. The data suggest that 2-OG induces a major conformational change in NrpR which separates the two HTH motifs in the NrpR dimer, with the consequence that the HTH motifs are then too far away to allow stable interaction with the operator sites on DNA (100) (Fig. 6).

The Carbohydrate PTS

The carbohydrate phosphotransferase system (PTS) is present in a number of bacterial species and is responsible for the uptake of hexoses (PTS sugars), including glucose, mannose, and glucosamine; this system can also transport bi- and trisaccharides. The PTS also plays a number of regulatory roles that are thought to be the primary function of this system (101). Since the PTS has been studied extensively and comprehensively reviewed elsewhere (102, 103), in this review we only briefly describe the overall features of the PTS, focusing primarily on its regulation by 2-OG in E. coli.

The PTS acts as a phosphorylation cascade coupling the transport of carbohydrates to their phosphorylation. The basic components of the PTS are similar in all species studied (102). In E. coli, the phosphorelay begins in the cytosol with the EI protein, which catalyzes its autophosphorylation using the glycolysis intermediate phosphoenolpyruvate (PEP) as a phosphoryl donor (104). The EI phosphoryl group is then transferred to another cytosolic protein, HPr. The cytosolic EI and HPr proteins act as common phosphoryl donors to a group of membrane-associated transporters, the EII components, which are specific for different carbohydrates (105). Typical EII components consist of an integral membrane domain, which facilitates sugar transport, together with associated cytoplasmic domains that can be carried by separate polypeptides (102). For instance, the glucose-specific EII complex in E. coli consists of the cytosolic EIIA^Glc component (encoded by crr) and the integral membrane component EIICB^Glc (encoded by ptsG) (105). The phosphoryl group is transferred from HPr to EIIA^Glc and then relayed to EIICB^Glc, where it is then transferred to the incoming glucose (Fig. 7).

FIG 7.

Regulation of AC activity by 2-OG levels. (A) When glucose is unavailable, phosphorylated EIIA (EIIA-P) results in accumulated activated AC, cAMP production, and the use of alternative carbon sources. Under these conditions, 2-OG levels are low, allowing EI autophosphorylation and active phosphorelay to EIIA. (B) In the presence of glucose, EIIA is mainly dephosphorylated (denoted by the pale P signal) due to phosphotransfer to the incoming glucose. High glucose increases 2-OG levels, thereby inhibiting EI autophosphorylation and phosphorelay to EIIA. The accumulation of unphosphorylated EIIA inhibits AC activity and cAMP production. AC activity is further inhibited by 2-OG, by an unknown mechanism. (C) Nitrogen levels may also affect cAMP production. In a situation where E. coli is using glucose and a poor nitrogen source and is suddenly presented with a good nitrogen source (i.e., ammonium shock conditions), a sudden increase in the assimilation of ammonia into amino acids (denoted by a large red arrow) will drain carbon skeletons (and 2-OG), augmenting the demand for carbon and energy for protein biosynthesis. This reduction in 2-OG levels is expected to activate the PTS phosphorelay, thereby augmenting the concentration of EIIA-P and partially activating AC. cAMP will activate carbon catabolic pathways, thereby fulfilling the increased carbon and energy demand.

Components of the PTS, particularly EIIA, also act as signaling modules controlling the activity of a number of downstream targets. For example, when glucose is available, its transport results in net dephosphorylation of the PTS proteins (Fig. 7B). Unphosphorylated EIIA^Glc directly binds to a number of transporters for nonpreferred carbon sources (non-PTS sugars), inhibiting their activity and thereby ensuring that the genes required for their catabolism are not induced, in a mechanism known as inducer exclusion (102, 103). Unphosphorylated EIIBC^Glc also sequesters the transcriptional repressor Mlc to the cell membrane, affecting the expression of genes involved in carbohydrate metabolism. On the other hand, when glucose is limiting, the PTS proteins are mainly observed to be phosphorylated. The phosphorylated form of EIIA^Glc activates adenylate cyclase (AC), resulting in the accumulation of cAMP, which binds to the catabolic repressor protein (CRP), affecting the expression of a vast array of genes, in a regulatory mechanism known as carbon catabolite repression (CCR).

Even though this model for regulation of AC is widely accepted, other unknown factors are believed to participate in the control of AC activity, as supported by in vitro studies (106; for a review, see reference 107). The presence of additional regulatory factors is further supported by the observation that non-PTS carbon sources and low nitrogen availability tend to reduce cAMP production. As noted below, recent data support a model in which 2-OG is a key player in controlling AC activity, and this may explain, at least in part, some of the “unidentified signaling elements.”

Regulation of carbohydrate PTS phosphorylation by 2-OG.

Doucette et al. observed that glucose and d-mannitol uptake in E. coli is inhibited by nitrogen limitation. Using metabolomic approaches together with mutant analysis, they concluded that the inhibition of PTS carbohydrate uptake was caused by the accumulation of 2-OG that occurs under nitrogen limitation conditions (108). In vitro analysis with purified proteins demonstrated that the inhibition of carbohydrate uptake was caused by the inhibition of EI autophosphorylation in the presence of 2-OG. These authors noted that 2-OG directly influences EI activity in a noncompetitive fashion, suggesting that PEP and 2-OG have different binding sites on EI (108). Direct binding of 2-OG to the C-terminal region of E. coli EI has been confirmed by nuclear magnetic resonance (NMR), with a reported K_d of ∼2.2 mM (109). However, in contrast to the data reported by Doucette et al., the NMR analysis indicated that PEP and 2-OG compete for the same binding site on EI. Indeed, molecular docking models indicate that 2-OG can occupy the PEP binding site on the EI C-terminal domain (109, 110).

These studies imply that the status of EI phosphorylation, and thus of all subsequent proteins in the PTS cascade, is controlled by the 2-OG/PEP ratio. Hence, 2-OG has the potential to influence not only PTS carbohydrate transport but also the PTS signaling functions. These include inducer exclusion, regulation of AC activity and consequent cAMP accumulation, and CCR (Fig. 7). Indeed, pioneer studies showed that when 2-OG is added to E. coli cultivated in the presence of glycerol, there is a decrease in cAMP synthesis (111). A similar effect was observed by Rabinowitz and coworkers, who noted a reduction in the rate of accumulation of cAMP when a membrane-permeating 2-OG analog (dimethyl-ketoglutarate) was added to E. coli using glycerol as a carbon source (108).

Strikingly, however, 2-OG also appears to participate in an additional signaling loop, in which 2-OG (and other α-keto acids, such as oxaloacetate) influences AC activity in the absence of a functional PTS. By utilizing E. coli strains in which elevated transport of 2-OG was achieved through upregulation of the kgtP transporter, Hwa and coworkers demonstrated that the addition of 2-OG causes transient repression of lacZ expression and reduces the level of cAMP (112). This corresponds to the classical phenomenon of CCR mediated by CRP-cAMP. Since the transient decrease in cAMP production mediated by 2-OG was completely independent of the PTS^Glu system, it appears that this represents a novel signaling loop providing hierarchical control of CCR, in which 2-OG either directly regulates AC or interacts with a 2-OG-responsive regulatory partner of AC (112) (Fig. 7B).

Overall, these data now establish that 2-OG is a master regulator that not only influences the activities of the PTS but also controls CCR via a PTS-independent route (112, 113). 2-OG therefore plays a pivotal role in coordinating carbon and nitrogen metabolism in this system, since, as mentioned above, the level of 2-OG is inversely correlated with nitrogen availability. As a consequence, the quality of the nitrogen source influences the accumulation of cAMP (e.g., see Fig. 4D in the work of Doucette et al. [108]) (Fig. 7C).

The Regulatory PTS (PTS^Ntr)

Apart from the canonical sugar-related PTS, some bacteria encode another system, named PTS^Ntr, which is not involved in carbohydrate uptake and apparently is dedicated to regulatory functions (101, 114, 115). In E. coli, the PTS^Ntr comprises EI^Ntr, encoded by ptsP; NPr, the product of ptsO; and EIIA^Ntr, encoded by ptsN. These proteins are paralogs of the sugar PTS components EI, HPr, and EII, respectively. EIIA^Ntr can be phosphorylated in vitro by adding PEP, EI^Ntr, and NPr (114). Hence, the phosphorelay cascade in PTS^Ntr is analogous to that in the canonical sugar PTS (Fig. 8). In fact, there is evidence for cross talk between the canonical sugar PTS and the PTS^Ntr (114, 116). However, notable differences between the two systems include the following: first, the presence of a regulatory N-terminal GAF domain in E. coli EI^Ntr (similar to that found in NifA), and second, the absence of an identified final acceptor of the phosphoryl group from EIIA^Ntr. These findings support the hypothesis that PTS^Ntr has a regulatory function.

FIG 8.

Hypothetical relationship between nitrogen status, phosphorylation of EIIA^Ntr, and nitrogen metabolism. Ammonium limitation increases the intracellular 2-OG/glutamine ratio, enhancing EI^Ntr phosphorylation and phosphorelay to EIIA^Ntr, which accumulates in the phosphorylated form. P-EIIA^Ntr does not inhibit Trk activity, resulting in an accumulation of intracellular K⁺ and thereby favoring the binding of σ^S to RNA polymerase and the expression of genes involved in the general starvation response. This scenario is likely to occur under conditions of high extracellular K⁺, where Trk is active and Kdp expression is repressed.

The genes encoding the PTS^Ntr components, ptsN and ptsO, were first identified in Klebsiella pneumoniae and in E. coli, adjacent to the rpoN gene, which encodes the alternative RNA polymerase sigma factor σ⁵⁴, required for transcription of nitrogen assimilation genes (114, 117–119). Similar genomic contexts are found in other proteobacteria (102), suggesting that PTS^Ntr could be involved in the regulation of nitrogen metabolism by controlling σ⁵⁴ activity (117). However, the effect of a ptsN knockout on the activity of σ⁵⁴-dependent promoters is at most modest (114, 117, 120). Furthermore, deletion of genes encoding the PTS^Ntr caused no growth phenotype or defect in nitrogen utilization. Some of the phenotypes attributed to ptsN mutations in early studies were traced to an ilvG frameshift mutation present in some E. coli strains (114, 121–123).

Regulation of PTS^Ntr phosphorylation by 2-OG.

Recent studies by Seok and colleagues support a link between PTS^Ntr and nitrogen regulation. Using an in vitro-reconstituted system, they demonstrated that 2-OG and glutamine (signals of the carbon and nitrogen statuses) reciprocally regulate the phosphorylation state of PTS^Ntr. Glutamine inhibits EI^Ntr autophosphorylation, whereas 2-OG stimulates it, which is consistent with the accumulation of unphosphorylated EIIA^Ntr under ammonium-sufficient conditions in vivo (124). The binding sites for glutamine and 2-OG were mapped to the GAF domain of EI^Ntr (124). However, glutamine, not 2-OG, was found to bind the GAF domain of EI^Ntr from Sinorhizobium meliloti (125). Hence, glutamine sensing might be a primary function of EI^Ntr. Although these studies help to explain how nitrogen signals control the PTS^Ntr phosphorelay, a direct connection between the PTS^Ntr phosphorylation status and nitrogen regulation remains to be determined. A variety of different cellular processes are controlled directly or indirectly by PTS^Ntr in various organisms (103, 115), but none of these are directly related to nitrogen metabolic pathways.

In E. coli, the best-characterized function of PTS^Ntr involves regulation of the expression and activity of K⁺ transporters (126, 127), a function also conserved in Rhizobium leguminosarum (128, 129). In E. coli, unphosphorylated EIIA^Ntr has two effects on K⁺ accumulation: first, it binds to the low-affinity K⁺ transporter Trk and inhibits the accumulation of high intracellular K⁺ levels (126), and second, it activates transcription of the high-affinity K⁺ transporter KdpFABC through direct interaction with the potassium sensor kinase KdpD (127) (Fig. 8). Trk and Kdp are believed to be the major routes of K⁺ uptake in E. coli, with K_m values for K⁺ of 1.5 mM and 10 μM, respectively (130, 131). It was speculated that PTS^Ntr would function to regulate intracellular K⁺ according to the carbon supply. When a good carbon source is available, EIIA^Ntr would be dephosphorylated (due to cross talk with the canonical sugar PTS), resulting in increased Kdp expression with a consequent accumulation of K⁺. Intracellular K⁺ would, in turn, act as an allosteric activator of metabolic enzymes (127).

Although the studies described above point to a possible role for 2-OG in communicating the carbon status to the PTS^Ntr, the observation that glutamine binds to the GAF domain of EI^Ntr suggests that this system is responsive to the nitrogen status (124, 125). In the following paragraph, we discuss a potential hypothesis in which the PTS^Ntr may signal nitrogen availability via regulation of potassium transport.

Transcriptome analysis revealed that the expression of many genes is affected by a ptsN deletion, with the most dramatic effect being upregulation of the σ^S regulon (132). It was concluded that K⁺ regulates the selectivity of RNA polymerase toward σ⁷⁰ or σ^S such that, with high K⁺ levels, RNA polymerase preferentially binds σ^S (132). Induction of the σ^S (RpoS) regulon by E. coli enables acclimatization to nutrient starvation, including nitrogen deprivation (133–136). The signaling mechanism for σ^S regulon induction under nitrogen starvation conditions remains elusive (136), although RpoS activity is indirectly regulated by the Ntr regulatory system (137). Therefore, it is tempting to speculate that the PTS^Ntr may signal nitrogen starvation to activate the σ^S starvation response. With a low ammonium supply (high 2-OG and low glutamine levels), phosphorylated EIIA^Ntr would accumulate, avoiding inhibition of Trk activity and resulting in increased intracellular K⁺ levels, thereby favoring an interaction between σ^S and RNA polymerase (Fig. 8). Such a scenario is likely to occur when extracellular K⁺ levels are relatively high, when Trk is active and Kdp expression is repressed (Fig. 8). The activity of RpoS is also affected by EIIA^Ntr in A. vinelandii. The accumulation of unphosphorylated EIIA^Ntr reduced the expression of the RpoS-dependent gene arpR (138). Although the connection between RpoS activity and EIIA^Ntr has not been established in A. vinelandii, these data fit with the model suggested in Fig. 8. Accumulation of unphosphorylated EIIA^Ntr would downregulate Trk activity, reducing the intracellular K⁺ level and, thereby, RpoS activity.

Other, Less-Characterized Prokaryotic 2-OG Sensor Proteins

The Bacillus subtilis transcriptional regulator GltC.

The expression of the B. subtilis gltAB genes, encoding the GOGAT enzyme (Fig. 1), is regulated by two transcription factors, GltC and TnrA (139). GltC is a member of the LysR family, whose DNA binding activity is typically affected by interaction with small molecules. It was observed that GltC activates gltAB transcription, which is enhanced in the presence of 2-OG and reduced in the presence of glutamate. These data support a model in which GltC acts as a sensor of the 2-OG/glutamate ratio (140).

The cyanobacterial transcriptional regulator CcmR.

Efficient carbon fixation through RubisCO in phototrophic cyanobacteria relies on the ability to concentrate inorganic carbon (C_i) near the RubisCO active site. High-affinity C_i transporter systems are induced upon C_i limitation. In Synechocystis sp., a LysR transcription factor, namely, CcmR, acts a repressor of some C_i-concentrating mechanisms (141). The DNA binding activity of CcmR is enhanced in the presence of 2-OG, NADP⁺, 2-phosphoglycolate, or ribulose bisphosphate (142).

The two-component 2-OG signaling system KguS/KguR in uropathogenic E. coli.

In a search for novel two-component signaling systems specifically present in uropathogenic E. coli, Cai and coworkers identified a novel regulatory pair, KguS/KguR, which is presumably responsive to 2-OG (143). Expression of the KguS/KguR pair is induced upon oxygen limitation, and as long as 2-OG is present, this activates the transcription of genes located on a pathogenicity island which are involved in the utilization of extracellular 2-OG as a carbon source under anaerobic conditions (143).

2-OG SIGNALING IN EUKARYOTES

As mentioned previously, nitrogen starvation leads to increased 2-OG levels in the unicellular eukaryote S. cerevisiae (19). Increasing evidence supports the view that 2-OG also acts as a conserved starvation signaling molecule in Metazoa (144). When the nematode Caenorhabditis elegans experiences starvation, 2-OG levels increase (144), and similar increases in 2-OG levels were observed in humans subjected to prolonged or acute exercise (145, 146). In animals, either starvation or exercise is expected to activate amino acid catabolism in order to provide carbon skeletons for anaplerosis, which could explain the increased 2-OG levels observed under these circumstances (144).

The addition of 2-OG to the culture medium increased the life span of C. elegans (144). In order to identify the sensor protein responsible for this effect, the drug affinity responsive target stability (DARTS) technique, which relies on increased resistance to proteolysis when a protein interacts with a small molecule (147), was used to detect 2-OG binding proteins. This analysis identified the β-subunit of the mitochondrial ATP synthase as the 2-OG binding target in both human cell lines and C. elegans. The binding of 2-OG to ATP synthase reduced the activity of the enzyme and mitochondrial oxygen consumption (144). These data support the view that 2-OG acts as an important signaling molecule in Metazoa, and this proposition is further supported by the identification of 2-OG as a ligand for a human G-coupled receptor, named either GPR99 or OXGR1 (148). This receptor is mostly expressed in the kidney and may regulate acid-base balance (149).

The importance of 2-OG as a signaling molecule has similarly long been recognized in algae and plants (150, 151), supported by the presence of chloroplast-located P_II signal transduction proteins in these organisms. Indeed, both fatty acid and arginine biosynthesis pathways are regulated by 2-OG in A. thaliana, by controlling the interaction between P_II and the acetyl-CoA carboxylase and the N-acetyl-glutamate kinase (NAGK) enzymes, respectively (59, 152, 153). Surprisingly, however, a recent study revealed that the structural features of P_II proteins from members of the Brassicaceae, such as A. thaliana, are atypical compared with those found in most of the plant kingdom. In contrast to A. thaliana, green plant and algal P_II proteins contain a plant-specific C-terminal extension that forms a novel loop in the structure, termed the Q loop, which confers a low-affinity glutamine binding site (154). In vitro studies with Chlamydomonas reinhardtii P_II demonstrated that it binds NAGK in a glutamine-dependent manner, in addition to the 2-OG-regulated interaction. A recombinant form of A. thaliana P_II containing a functional Q loop from C. reinhardtii exhibited glutamine dependency in the NAGK interaction, confirming that the plant-specific C-terminal extension mediates the glutamine response. These observations suggest that glutamine sensing is a conserved feature of P_II signaling in most plant chloroplasts (154).

Fluctuations in 2-OG levels may affect the activity of enzymes that use 2-OG as the substrate. Enzymes belonging to the family of 2-OG-dependent dioxygenases (2-OGDO) are particularly worthy of mention in this context, since they are widespread in nature and have diverse metabolic functions, including regulation of secondary metabolism in plants (151) and the catalysis of both DNA and histone demethylation in animals (155). There is increasing evidence that 2-OG levels can influence the epigenetic landscape through regulation of chromatin structure and histone modifications, thus providing a link between cellular metabolism and cell proliferation and differentiation (156).

CONCLUDING REMARKS

The coordination of microbial metabolic pathways is mediated by fluctuations in the concentrations of key metabolites, which enable allosteric transfer of this information to sensor proteins, thereby generating the appropriated metabolic response (157). Examples of metabolites that act as key players in microbial signal transduction are glutamine, cAMP, and ppGpp. Each of these metabolites is dedicated to the transfer of information upon the availability of a specific class of nutrient, namely, nitrogen, carbon, and amino acids, respectively (157). The ability of these metabolites to cross-regulate unrelated metabolic pathways seems to be relatively minor. For example, in addition to its huge effect on carbon metabolism (102), cAMP influences nitrogen metabolism by regulating glnA and glnHPQ expression in E. coli via CRP (158, 159).

A signaling metabolite that is well suited to fulfill a broader signaling function and to coordinate both carbon and nitrogen metabolism is 2-OG. This compound is a key intermediate of the TCA cycle and is used as a carbon skeleton for nitrogen assimilation (Fig. 1). As such, 2-OG reflects the carbon and nitrogen balance (1, 157), a signaling characteristic which seems to be conserved throughout nature. Not surprisingly, the list of regulatory proteins and metabolic pathways controlled by 2-OG has expanded exponentially and now includes nitrogen- and carbon-dedicated metabolic pathways (Fig. 9). In prokaryotes, 2-OG regulates several transcriptional factors and P_II protein activity (Fig. 9). The recent inclusion of PTS^Glu, PTS^Ntr, and adenylate cyclase as targets of 2-OG has extended the regulatory repertoire of this molecule to an unprecedented level, raising 2-OG above cAMP in the signal transduction hierarchy (Fig. 9).

FIG 9.

Schematic representation of proteins and metabolic pathways whose activity is affected by 2-OG levels. The 2-OG level is influenced by the supply of carbon and nitrogen. In turn, intracellular fluctuations in 2-OG are perceived by various sensory proteins, including P_II, the EI component, both PTS^Glu and PTS^Ntr, and a range of transcriptional factors. In E. coli, 2-OG regulates adenylate cyclase (AC) through the action of PTS^Glu and also through an as yet unidentified mechanism (question mark). In Metazoa, 2-OG levels directly affect the activity of ATP synthase.

Despite the emergence of 2-OG as a master signaling metabolite, a number of important questions remain to be answered. Although 2-OG is central to the interface between carbon and nitrogen metabolism, it is not always employed as the sole integrator of these signals. Indeed, under some conditions, this could be a dangerous physiological strategy, since the level of 2-OG might not always be representative of the carbon/nitrogen ratio. For example, under conditions of carbon starvation, a low concentration of 2-OG might not necessarily be indicative of nitrogen excess if the cell is also suffering from nitrogen starvation. The P_II signal transduction proteins provide fail-safe integration of the carbon and nitrogen statuses through directly sensing the 2-OG concentration and also by sensing glutamine, either directly through the Q linker or indirectly via posttranslational modifications. Another well-suited module for integrating the carbon and nitrogen statuses is the E. coli PTS^Ntr, as its phosphorelay cascade is modulated by both 2-OG and glutamine (124). However, despite several reported functions of PTS^Ntr in different model bacteria, a direct connection between PTS^Ntr and regulation of a specific metabolic pathway remains elusive. Moreover, S. meliloti PTS^Ntr binds only glutamine, suggesting that in this case the primary function is to sense the nitrogen status.

Although 2-OG appears to play an overarching role in carbon sensing through modulating the level of cAMP, the question remains as to why there is apparent functional redundancy in the control circuitry. Since 2-OG can apparently bypass the PTS^Glu to regulate adenylyl cyclase activity, why is there a need for a PTS-dependent route for controlling AC in response to 2-OG (Fig. 9)? The PTS-dependent route is dependent on relatively high concentrations of 2-OG, as the K_d for binding to the C-terminal domain of EI is ∼2 mM. It is possible that the primary function of this route is feedback inhibition of glucose transport and that the effect on AC activity is secondary. This is one case where 2-OG clearly acts to integrate signals of carbon and nitrogen regulation when, for example, a sudden addition of excess nitrogen results in the depletion of carbon skeletons, which is homeostatically compensated by increased glycolytic flux (see Fig. 7C in reference 108). In contrast, the PTS-independent route of inhibition of AC activity by 2-OG (112) explains why non-PTS carbon sources can influence CCR. However, a mechanistic understanding of this newly discovered role for 2-OG is lacking. We still do not know, for example, which protein is the direct target for 2-OG in this case and the affinity of its interaction with this target, let alone the mechanism whereby adenylyl cyclase activity is regulated.

Even more enigmatic is the physiological role of 2-OG (and glutamine) in controlling PTS^Ntr activity in E. coli. While this could maintain the linkage between nitrogen availability and σ^S activity via accumulation of intracellular potassium (Fig. 8), other functions are also plausible. One potential hypothesis is that PTS^Ntr controls nonspecific uptake of ammonium by the potassium transporters in response to nitrogen availability. Ammonium (NH₄⁺) is the preferred nitrogen source for E. coli, and NH₄⁺ and K⁺ have similar charges, ionic radii, and hydration shells (160). Hence, K⁺ and NH₄⁺ transporters from various sources do allow significant amounts of these similar ions to permeate (160–165). In Saccharomyces cerevisiae, ammonium becomes toxic when cells are cultured with low K⁺ levels, and such toxicity is apparently caused by ammonium uptake via potassium transporters (163). When E. coli is cultured in media containing low potassium levels (which favors Kdp activity), the presence of high ammonium concentrations increases the rate of oxygen and glucose consumption by a wild-type strain but not an isogenic kdp null strain (161). These results were interpreted as nonspecific uptake of NH₄⁺ by the Kdp transporter resulting in a transient increase in the intracellular ammonium concentration, followed by leakage of NH₃ out of the cells via passive diffusion across the cell membrane (161). This would result in a Gibbs energy-wasting futile cycle, explaining the increase in cellular energy needs in the wild-type strain (166). Although ammonium toxicity has not been documented for E. coli, tight control of ammonium uptake is vital to restrict the energetic cost of futile cycling (167). Consequently, the activity of the ammonium transporter AmtB is tightly regulated in response to the external ammonium concentration, through the formation of a protein complex with the dedicated P_II-like protein GlnK (54). Since nonspecific ammonium uptake by K⁺ transporters will result in unrestricted futile cycling, we favor the hypothesis that the PTS^Ntr may also participate in the control of nonspecific ammonium transport, particularly through the low-affinity K⁺ transporter Trk, which is likely to exhibit enhanced transport of noncognate ions. According to this scenario, nonphosphorylated EIIA^Ntr would accumulate under conditions of ammonium excess, inhibiting Trk activity and enhancing the expression of Kdp, which is expected to be more efficient than Trk in discriminating between K⁺ and NH₄⁺. However, there is little evidence at present to support this hypothesis.

Given the signaling properties of 2-OG, it is highly likely that other regulatory proteins responsive to 2-OG remain to be described. The deployment of the DARTS technique to successfully identify ATP synthase as a 2-OG binding target in eukaryotes (144) provides an elegant example of how this field is likely to move ahead in the future. There is already strong evidence to support the notion that 2-OG acts as a master regulatory metabolite in eukaryotes, a view already anticipated by the presence of P_II proteins in algae and plants. The recent development of 2-OG-specific affinity columns (168) may help to identify novel 2-OG sensors and extend the number of metabolic pathways in which this exciting regulatory metabolite plays a role.

Biographies

Luciano F. Huergo studied biology for his undergraduate degree at the Universidade Federal do Paraná, Brazil, and obtained his Ph.D. in biochemistry at the Universidade Federal do Paraná in 2006. He worked as a visiting scientist at John Innes Centre, United Kingdom (2005), and at the Institute for Glycomics at Griffith University, Australia (2013). From 2008 to 2015, he was a professor at the Universidade Federal do Paraná in Curitiba, Brazil. In 2015, he moved to the Setor Litoral of Universidade Federal do Paraná in Matinhos, Brazil, where he is currently establishing a new microbiology and molecular biology lab. The major focus of his current research is the biology of P_II proteins and the regulation of bacterial metabolism.

Ray Dixon is Research Group Leader at the John Innes Centre in Norwich, United Kingdom. He studied microbiology for his undergraduate degree at the University of Reading and obtained a D.Phil. in microbial genetics at the University of Sussex in 1973. He remained at the Unit of Nitrogen Fixation, University of Sussex, to lead a research group focusing on the molecular biology of nitrogen fixation before moving to his current position in Norwich in 1995. The major aim of his current research is to understand how nitrogen fixation is regulated in response to the oxygen, carbon, and fixed nitrogen status.