Learn more: PMC Disclaimer | PMC Copyright Notice
Structural basis for the allosteric inhibitory mechanism of human kidney-type glutaminase (KGA) and its regulation by Raf-Mek-Erk signaling in cancer cell metabolism
Associated Data
Abstract
Besides thriving on altered glucose metabolism, cancer cells undergo glutaminolysis to meet their energy demands. As the first enzyme in catalyzing glutaminolysis, human kidney-type glutaminase isoform (KGA) is becoming an attractive target for small molecules such as BPTES [bis-2-(5 phenylacetamido-1, 2, 4-thiadiazol-2-yl) ethyl sulfide], although the regulatory mechanism of KGA remains unknown. On the basis of crystal structures, we reveal that BPTES binds to an allosteric pocket at the dimer interface of KGA, triggering a dramatic conformational change of the key loop (Glu312-Pro329) near the catalytic site and rendering it inactive. The binding mode of BPTES on the hydrophobic pocket explains its specificity to KGA. Interestingly, KGA activity in cells is stimulated by EGF, and KGA associates with all three kinase components of the Raf-1/Mek2/Erk signaling module. However, the enhanced activity is abrogated by kinase-dead, dominant negative mutants of Raf-1 (Raf-1-K375M) and Mek2 (Mek2-K101A), protein phosphatase PP2A, and Mek-inhibitor U0126, indicative of phosphorylation-dependent regulation. Furthermore, treating cells that coexpressed Mek2-K101A and KGA with suboptimal level of BPTES leads to synergistic inhibition on cell proliferation. Consequently, mutating the crucial hydrophobic residues at this key loop abrogates KGA activity and cell proliferation, despite the binding of constitutive active Mek2-S222/226D. These studies therefore offer insights into (i) allosteric inhibition of KGA by BPTES, revealing the dynamic nature of KGA's active and inhibitory sites, and (ii) cross-talk and regulation of KGA activities by EGF-mediated Raf-Mek-Erk signaling. These findings will help in the design of better inhibitors and strategies for the treatment of cancers addicted with glutamine metabolism.
The Warburg effect in cancer biology describes the tendency of cancer cells to take up more glucose than most normal cells, despite the availability of oxygen (1, 2). In addition to altered glucose metabolism, glutaminolysis (catabolism of glutamine to ATP and lactate) is another hallmark of cancer cells (2, 3). In glutaminolysis, mitochondrial glutaminase catalyzes the conversion of glutamine to glutamate (4), which is further catabolized in the Krebs cycle for the production of ATP, nucleotides, certain amino acids, lipids, and glutathione (2, 5).
Humans express two glutaminase isoforms: KGA (kidney-type) and LGA (liver-type) from two closely related genes (6). Although KGA is important for promoting growth, nothing is known about the precise mechanism of its activation or inhibition and how its functions are regulated under physiological or pathophysiological conditions. Inhibition of rat KGA activity by antisense mRNA results in decreased growth and tumorigenicity of Ehrlich ascites tumor cells (7), reduced level of glutathione, and induced apoptosis (8), whereas Myc, an oncogenic transcription factor, stimulates KGA expression and glutamine metabolism (5). Interestingly, direct suppression of miR23a and miR23b (9) or activation of TGF-β (10) enhances KGA expression. Similarly, Rho GTPase that controls cytoskeleton and cell division also up-regulates KGA expression in an NF-κB–dependent manner (11). In addition, KGA is a substrate for the ubiquitin ligase anaphase-promoting complex/cyclosome (APC/C)-Cdh1, linking glutaminolysis to cell cycle progression (12). In comparison, function and regulation of LGA is not well studied, although it was recently shown to be linked to p53 pathway (13, 14). Although intense efforts are being made to develop a specific KGA inhibitor such as BPTES [bis-2-(5-phenylacetamido-1, 2, 4-thiadiazol-2-yl) ethyl sulfide] (15), its mechanism of inhibition and selectivity is not yet understood. Equally important is to understand how KGA function is regulated in normal and cancer cells so that a better treatment strategy can be considered.
The previous crystal structures of microbial (Mglu) and Escherichia coli glutaminases show a conserved catalytic domain of KGA (16, 17). However, detailed structural information and regulation are not available for human glutaminases especially the KGA, and this has hindered our strategies to develop inhibitors. Here we report the crystal structure of the catalytic domain of human apo KGA and its complexes with substrate (l-glutamine), product (l-glutamate), BPTES, and its derived inhibitors. Further, Raf-Mek-Erk module is identified as the regulator of KGA activity. Although BPTES is not recognized in the active site, its binding confers a drastic conformational change of a key loop (Glu312-Pro329), which is essential in stabilizing the catalytic pocket. Significantly, EGF activates KGA activity, which can be abolished by the kinase-dead, dominant negative mutants of Mek2 (Mek2-K101A) or its upstream activator Raf-1 (Raf-1-K375M), which are the kinase components of the growth-promoting Raf-Mek2-Erk signaling node. Furthermore, coexpression of phosphatase PP2A and treatment with Mek-specific inhibitor or alkaline phosphatase all abolished enhanced KGA activity inside the cells and in vitro, indicating that stimulation of KGA is phosphorylation dependent. Our results therefore provide mechanistic insights into KGA inhibition by BPTES and its regulation by EGF-mediated Raf-Mek-Erk module in cell growth and possibly cancer manifestation.
Results
Structures of cKGA and Its Complexes with l-Glutamine and l-Glutamate.
The human KGA consists of 669 amino acids. We refer to Ile221-Leu533 as the catalytic domain of KGA (cKGA) (Fig. 1A). The crystal structures of the apo cKGA and in complex with l-glutamine or l-glutamate were determined (Table S1). The structure of cKGA has two domains with the active site located at the interface. Domain I comprises (Ile221-Pro281 and Cys424 -Leu533) of a five-stranded anti-parallel β-sheet (β2↓β1↑β5↓β4↑β3↓) surrounded by six α-helices and several loops. The domain II (Phe282-Thr423) mainly consists of seven α-helices. l-Glutamine/l-glutamate is bound in the active site cleft (Fig. 1B and Fig. S1B). Overall the active site is highly basic, and the bound ligand makes several hydrogen-bonding contacts to Gln285, Ser286, Asn335, Glu381, Asn388, Tyr414, Tyr466, and Val484 (Fig. 1C and Fig. S1C), and these residues are highly conserved among KGA homologs (Fig. S1D). Notably, the putative serine-lysine catalytic dyad (286-SCVK-289), corresponding to the SXXK motif of class D β-lactamase (17), is located in close proximity to the bound ligand. In the apo structure, two water molecules were located in the active site, one of them being displaced by glutamine in the substrate complex. The substrate side chain is within hydrogen-bonding distance (2.9 Å) to the active site Ser286. Other key residues involved in catalysis, such as Lys289, Tyr414, and Tyr466, are in the vicinity of the active site. Lys289 is within hydrogen-bonding distance to Ser286 (3.1 Å) and acts as a general base for the nucleophilic attack by accepting the proton from Ser286. Tyr466, which is close to Ser286 and in hydrogen-bonding contact (3.2 Å) with glutamine, is involved in proton transfer during catalysis. Moreover, the carbonyl oxygen of the glutamine is hydrogen-bonded with the main chain amino groups of Ser286 and Val484, forming the oxyanion hole. Thus, we propose that in addition to the putative catalytic dyad (Ser286 XX Lys289), Tyr466 could play an important role in the catalysis (Fig. 1C and Fig. S2).
Schematic view and structure of the cKGA-l-glutamine complex. (A) Human KGA domains and signature motifs (refer to Fig. S1A for details). (B) Structure of the of cKGA and bound substrate (l-glutamine) is shown as a cyan stick. (C) Fourier 2Fo-Fc electron density map (contoured at 1 σ) for l-glutamine, that makes hydrogen bonds with active site residues are shown.
Allosteric Binding Pocket for BPTES.
The crystal structure of cKGA: BPTES complex shows that BPTES occupies a previously unsuspected allosteric pocket, in the solvent-exposed region at the dimer interface of cKGA, located ≈18 Å away from the active site serine (Ser286) (Fig. 2A). The chemical structure of BPTES has an internal symmetry, with two exactly equivalent parts including a thiadiazole, amide, and a phenyl group (Fig. S3A), and it equally interacts with each monomer. The thiadiazole group and the aliphatic linker are well buried in a hydrophobic cluster that consists of Leu321, Phe322, Leu323, and Tyr394 from both monomers, which forms the allosteric pocket (Fig. 2 B–E). The side chain of Phe322 is found at the bottom of the allosteric pocket. The phenyl-acetamido moiety of BPTES is partially exposed on the loop (Asn324-Glu325), where it interacts with Phe318, Asn324, and the aliphatic part of the Glu325 side chain. On the basis of our observations we synthesized a series of BPTES-derived inhibitors (compounds 2–5) (Fig. S3 A–F and SI Results) and solved their cocrystal structure of compounds 2–4. Similar to BPTES, compounds 2–4 all resides within the hydrophobic cluster of the allosteric pocket (Fig. S3 C–F).The side chain of Tyr394, backbone amide of the Phe322 and Leu323, makes hydrogen-bonding contacts to inhibitors. Moreover, the residues Leu321 and Phe322 flipped out ≈180° to enhance the hydrophobic interactions with the inhibitors. Thus, the conformational changes are required to form the allosteric pocket. Further kinetic studies showed that compounds 2–5 inhibit cKGA to a lesser extent than BPTES (Fig. S3B), and we propose that the binding specificity of BPTES is dictated by the thiadiazole, amide, and the hydrophobic linker regions, whereas the potency is primarily determined by the presence of phenyl rings at both ends. Thus, BPTES was subsequently used to further delineate the structural, functional, and regulation aspects of KGA.
Structure of cKGA: BPTES complex and the allosteric binding mode of BPTES. (A) Structure of cKGA dimer and BPTES is shown as a cyan stick. (B) A close-up view of the interactions of BPTES in the cKGA allosteric inhibitor binding pocket. (C) Electron density map (2Fo \x{2013} Fc map, contoured at 1.0σ) for BPTES is shown. (D) A close-up view of the BPTES binding pocket on the surface exposed region of the loop Glu312-Pro329 at the dimer interface. (E) Perpendicular view of dimer interface formed by the sulphate ion, hydrogen bonding, salt bridge, and hydrophobic interactions between residues from each monomer. (F) Conformational changes on cKGA induced by binding of the BPTES. For clarity only half of the BPTES is shown. Structure superposition of monomeric BPTES complex (magenta) and apo cKGA (green), showing conformational changes of key residues on the loop Glu312-Pro329. The BPTES binding site is located ∼18 Å away from the active site (Ser286).
Allosteric Binding of BPTES Triggers Major Conformational Change in the Key Loop Near the Active Site.
The overall structure of these inhibitor complexes superimposes well with apo cKGA. However, a major conformational change at the Glu312 to Pro329 loop was observed in the BPTES complex (Fig. 2F). The most conformational changes of the backbone atoms that moved away from the active site region are found at the center of the loop (Leu316-Lys320). The backbone of the residues Phe318 and Asn319 is moved ≈9 Å and ≈7 Å, respectively, compared with the apo structure, whereas the side chain of these residues moved ≈14 Å and ≈12 Å, respectively. This loop rearrangement in turn brings Phe318 closer to the phenyl group of the inhibitor and forms the inhibitor binding pocket, whereas in the apo structure the same loop region (Leu316-Lys320) was found to be adjacent to the active site and forms a closed conformation of the active site. Specifically, in apo structure Phe318 makes hydrophobic interactions with Tyr466, and side chain of the Asn319 makes hydrogen-bonding contact with backbone of the Asn335 (≈2.8 Å). Notably, the residues Tyr466 and Asn335 are involved in binding to l-glutamine and catalysis. These observations suggest that binding of BPTES induces conformational changes of the key residues of the loop (Glu312-Pro329) to stabilize an open and inactive conformation of the catalytic site. Fig. S4 A and B show the electrostatic surface potential of the open and closed conformation of active site. Besides, we have determined the cKGA–glutamate–BPTES structure and revealed similar conformational changes, suggesting that BPTES can stabilize an inactive glutamate-bound form of the enzyme (Fig. S4C). This result is consistent with previous kinetics studies showing that BPTES, which is an uncompetitive inhibitor, can inhibit both the enzyme–substrate or enzyme–product complex (15).
Binding of BPTES Stabilizes the Inactive Tetramers of cKGA.
To understand the role of oligomerization in KGA function, dimers and tetramers of cKGA were generated using the symmetry-related monomers (Fig. 2 A–E and Fig. S4 D and E). The dimer interface in the cKGA: BPTES complex is formed by residues from the helix Asp386-Lys398 of both monomers and involves hydrogen bonding, salt bridges, and hydrophobic interactions (Phe389, Ala390, Tyr393, and Tyr394), besides two sulfate ions located in the interface (Fig. 2E). The dimers are further stabilized by binding of BPTES, where it binds to loop residues (Glu312-Pro329) and Tyr394 from both monomers (Fig. 2 D and E). Similarly, residues from Lys311-Asn319 loop and Arg454, His461, Gln471, and Asn529-Leu533 are involved in the interface with neighboring monomers to form the tetramer in the BPTES complex. The interactions among the monomers are relatively weaker in the apo tetramer than in the BPTES complex (Fig. S4 D and E). We infer that the binding of BPTES promotes the formation of a stable but inactive tetramer.
BPTES Induces Allosteric Conformational Changes That Destabilize Catalytic Function of KGA.
Subsequently we examined how BPTES binding could inhibit KGA activity by comparing the structures of the cKGA: BPTES inhibitor complex with the apo cKGA structure. Because the loop Glu312-Pro329 is located near the active site, we hypothesize that BPTES binding induces conformational changes of the key residues of the loop (Glu312-Pro329) to stabilize an open and inactive conformation of the catalytic site. To test this hypothesis, wild-type KGA and structure-guided KGA mutants Phe318Ala, Leu321Ala, Phe322Ala, and Leu323Ala, as well as a Tyr394Ala variant from the dimerization helix, were each expressed in human embryonic kidney epithelial cells 293T at equal levels, and homeostasis levels of glutamate inside the cells were determined. Fig. 3A shows that 293T cells overexpressing KGA produced higher level of glutamate compared with the vector control cells. Most significantly, all of these mutants, except Phe322Ala, greatly diminished the KGA activity.
Mutations at allosteric loop and BPTES binding pocket abrogate KGA activity and BPTES sensitivity. (A) 293T cells were transfected with vector control or plasmids expressing wild-type, single, or multiple point mutants of HA-tagged KGA for 24 h before cell lysates were prepared for glutaminase assays. For clarity, the dotted line was included to indicate the basal level. Equal expression levels of the wild-type and mutants KGA were verified with Western blots. Each value represents the mean ± SD of three independent experiments. (B) Mutational analyses of cKGA residues in the BPTES binding pocket. BPTES sensitivity for the wild-type and cKGA mutants indicated were measured and their IC50 values calculated. Each value represents the mean ± SD of three independent experiments performed in duplicate.
Next, we examined whether these residues are specifically involved in stabilizing the BPTES–KGA interactions. Unlike all of the previous mutants that have Ala substitutions to knock out their direct contribution to the actual enzymatic activities, a set of recombinant cKGA mutants (Phe318Tyr, Phe322Ser, Phe318Tyr/Phe322Ser, and Tyr394Ile) were instead generated to test their BPTES sensitivity. In particular, Phe318Tyr/Phe322Ser double mutant was used to mimic the corresponding residues in the liver form of glutaminase, LGA. Results show that all these mutants still retain the same KGA activity as the wild-type control (Fig. S7A). However, the three mutants Phe322Ser or Phe318Tyr/Phe322Ser and Tyr394Ile showed significantly reduced sensitivity to BPTES (1,140-, 970-, and 910-fold, respectively) (Fig. 3B). In contrast, Phe318Tyr, which retains the aromatic ring and is still active, remains highly sensitive to BPTES. In summary, consistent with our structural analysis, all of the key residues in the loop Glu312 to Pro329 and Tyr394 are essential for conferring KGA activity, and at least Phe322 and Tyr394 are involved in stabilizing the BPTES–KGA interaction. Any conformational change upon BPTES binding would severely affect the stability of the catalytic core of KGA, hence affecting its activity.
Raf-Mek-Erk Signaling Module Regulates KGA Activity.
Because KGA supports cell growth and proliferation, we first validated that treatment of cells with BPTES indeed inhibits KGA activity and cell proliferation (Fig. S5 A–D and SI Results). Next, as cells respond to various physiological stimuli to regulate their metabolism, with many of the metabolic enzymes being the primary targets of modulation (18), we examined whether KGA activity can be regulated by physiological stimuli, in particular EGF, which is important for cell growth and proliferation. Cells overexpressing KGA were made quiescent and then stimulated with EGF for various time points. Fig. 4A shows that the basal KGA activity remained unchanged 30 min after EGF stimulation, but the activity was substantially enhanced after 1 h and then gradually returned to the basal level after 4 h. Because EGF activates the Raf-Mek-Erk signaling module (19), treatment of cells with Mek-specific inhibitor U0126 could block the enhanced KGA activity with parallel inhibition of Erk phosphorylation (Fig. 4A). Interestingly, such Mek-induced KGA activity is specific to EGF and lysophosphatidic acid (LPA) but not with other growth factors, such as PDGF, TGF-β, and basic FGF (bFGF), despite activation of Mek-Erk by bFGF (Fig. S6A). We next investigated whether Raf-Mek-Erk activated by EGF could indeed directly regulate the KGA activity. Cells were transfected with KGA with wild-type, the “dominant negative and kinase-dead,” or the constitutive active mutants of Raf-1 and Mek2 and the levels of glutamate determined. Strikingly, KGA activity was completely inhibited by both the dominant negative mutants of Raf-1 (K375M) and Mek2 (K101A), whereas both wild-type and constitutive active mutants of Raf-1 (Raf-1-Y340D) and Mek2 (Mek2-S222, 226D) did not lead to any further increase in the glutamate production (Fig. 4B). Moreover, expression of these mutants did not affect the expression levels of either ectopically expressed or the endogenous level of KGA (Fig. 4C and Fig. S6B), indicating that any changes in the KGA activity was not due to their protein stability but was due to some posttranslational modifications of KGA. To examine whether such regulation is directly associated with the Raf-Mek-Erk complex, overexpressd KGA was immunoprecipitated from the cells, and the presence of Raf-1, Mek2, or Erk1/2 (endogenous or overexpressed) was examined. The results show that KGA could interact equally well with the wild-type or mutant forms of Raf-1 and Mek2 (Fig. 4C). Importantly, endogenous Raf-1 or Erk1/2, including the phosphorylated Erk1/2 (Fig. 4 C and D), could be detected in the KGA complex. Taken together, these results indicate that the activity of KGA is directly regulated by Raf-Mek-Erk downstream of EGF receptor. To further show that Mek2-enhanced KGA activity requires both the kinase activity of Mek2 and the core residues for KGA catalysis, wild-type or triple mutant (Leu321Ala/Phe322Ala/Leu323Ala) of KGA was coexpressed with dominant negative Mek2-KA or the constitutive active Mek2-SD and their KGA activities measured. The result shows that the presence of Mek2-KA blocks KGA activity, whereas the triple mutant still remains inert even in the presence of the constitutively active Mek2 (Fig. 4E), and despite Mek2 binding to the KGA triple mutant (Fig. S7B). Consequently, expressing triple mutant did not support cell proliferation as well as the wild-type control (Fig. S7C).
EGFR-Raf-Mek-Erk signaling stimulates KGA activity. (A) 293T cells expressing HA-tagged KGA were starved for 24 h and then stimulated with EGF (100 ng/mL) for the times indicated. Cells were lysed and assayed for their glutaminase activities. The expression levels of KGA and the Erk activation profile (as indicated by levels of phosphorylated Erk) were verified by Western blot analyses. (B) Cells expressing Flag-tagged KGA with or without the HA-tagged wild-type, dominant negative mutants (Raf-1-K375M; Mek2-K101A) or constitutive active mutants (Raf-1-Y340D; Mek2-S222, 226D) were lysed and assayed for glutaminase activity. (C) Same batch of cell lysates prepared for the glutaminase assay in B were subjected to immunoprecipitation (IP) with anti-Flag M2 beads. Bound proteins and their expression in whole-cell lysates (WCL) were analyzed with Western blot. (D) Cells expressing Flag-tagged KGA were lysed for immunoprecipitation using anti-Flag M2 beads and analyzed for the presence of endogenous Raf-1 or Erk1/2 by Western blot analyses. Arrow denotes band for Raf-1. (E) 293T cells were transfected with vector control or plasmids expressing wild-type KGA or the KGA triple mutant (L321A/F322A/L323A), in the absence or presence of Mek2-K101A or Mek2-S222,226D for 24 h before lysates were prepared for glutaminase assays. Expression levels of these proteins were verified by Western blot analyses. All values are mean ± SD of three independent experiments, each with multiple replicates. Data sharing different letters are statistically significant at P values as indicated, tested by ANOVA or t test.
Interestingly, when cells expressing both KGA and Mek2-K101A were treated with subthreshold levels of BPTES, there was a synergistic reduction in cell proliferation (Fig. S6C and SI Results). Lastly, to determine whether regulation of KGA by Raf-Mek-Erk depends on its phosphorylation status, cells were transfected with KGA with or without the protein phosphatase PP2A and assayed for the KGA activity. PP2A is a ubiquitous and conserved serine/threonine phosphatase with broad substrate specificity. The results indicate that KGA activity was reduced down to the basal level in the presence of PP2A (Fig. 5A). Coimmunoprecipitation study also revealed that KGA interacts with PP2A (Fig. 5B), suggesting a negative feedback regulation by this protein phosphatase. Furthermore, treatment of immunoprecipitated and purified KGA with calf-intestine alkaline phosphatase (CIAP) almost completely abolished the KGA activity in vitro (Fig. S6D). Taken together, these results indicate that KGA activity is regulated by Raf-Mek2, and KGA activation by EGF could be part of the EGF-stimulated Raf-Mek-Erk signaling program in controlling cell growth and proliferation (Fig. 5C).
KGA activity is regulated by phosphorylation. (A) Lysates were prepared from cells expressing Flag-tagged KGA in the presence or absence of myc-tagged catalytic subunit of the protein phosphatase PP2A and assayed for the glutaminase activity. Values are means ± SD of three independent experiments. Data sharing different letters are statistically significant at P < 0.02, as tested by ANOVA. (B) Separate aliquots from the same batch of cell lysates prepared for the glutaminase assay in A were subjected to immunoprecipitation (IP). Bound myc-PP2A and their expression levels in whole-cell lysates (WCL) were analyzed by Western blots. (C) Schematic model depicting the synergistic cross-talk between KGA-mediated glutaminolysis and EGF-activated Raf-Mek-Erk signaling. Exogenous glutamine can be transported across the membrane and converted to glutamate by glutaminase (KGA), thus feeding the metabolite to the ATP-producing tricarboxylic acid (TCA) cycle. This process can be stimulated by EGF receptor-mediated Raf-Mek-Erk signaling via their phosphorylation-dependent pathway, as evidenced by the inhibition of KGA activity by the kinase-dead and dominant negative mutants of Raf-1 (Raf-1-K375M) and Mek2 (Mek2-K101A), protein phosphatase PP2A, and Mek-specific inhibitor U0126. Consequently, inhibiting KGA with BPTES and blocking Raf-Mek pathway with Mek2-K101A provide a synergistic inhibition on cell proliferation. Refer to the text for more details.
Discussion
Small-molecule inhibitors that target glutaminase activity in cancer cells are under development. Earlier efforts targeting glutaminase using glutamine analogs have been unsuccessful owing to their toxicities (2). BPTES has attracted much attention as a selective, nontoxic inhibitor of KGA (15), and preclinical testing of BPTES toward human cancers has just begun (20). BPTES selectively suppresses the growth of glioma cells (21) and inhibits the growth of lymphoma tumor growth in animal model studies (22). Wang et al. (11) reported a small molecule that targets glutaminase activity and oncogenic transformation. Despite extensive studies, nothing is known about the structural and molecular basis for KGA inhibitory mechanisms and how their function is regulated during normal and cancer cell metabolism. Such limited information impedes our effort in producing better generations of inhibitors for better treatment regimens.
Comparison of the complex structures with apo cKGA structure, which has well-defined electron density for the key loop, we provide the atomic view of an allosteric binding pocket for BPTES and elucidate the inhibitory mechanism of KGA by BPTES. The key residues of the loop (Glu312-Pro329) undergo major conformational changes upon binding of BPTES. In addition, structure-based mutagenesis studies suggest that this loop is essential for stabilizing the active site. Therefore, by binding in an allosteric pocket, BPTES inhibits the enzymatic activity of KGA through (i) triggering a major conformational change on the key residues that would normally be involved in stabilizing the active sites and regulating its enzymatic activity; and (ii) forming a stable inactive tetrameric KGA form. Our findings are further supported by two very recent reports on KGA isoform (GAC) (23, 24), although these studies lack full details owing to limitation of their electron density maps. BPTES is specific to KGA but not to LGA (15). Sequence comparison of KGA with LGA (Fig. S8A) reveals two unique residues on KGA, Phe318 and Phe322, which upon mutation to LGA counterparts, become resistant to BPTES. Thus, our study provides the molecular basis of BPTES specificity.
It was recently reported that KGA is up-regulated by Myc and Rho and is subjected to ubiquitination (9, 11, 12). However, little is known regarding how the glutaminolytic pathway is functionally linked to growth-promoting signaling pathway(s). Many metabolic enzymes are thought to serve as housekeeping enzymes that control fluxes of metabolites to sustain rather than to primarily regulate cell growth. Here we show that a high level of KGA activity can lead to the increased cell proliferation that is inhibited by BPTES. Most significantly, we show that KGA activity is activated by EGF via the Raf-Mek-Erk signaling module because inhibition of both kinases by their dominant negative mutants or treatment with a specific Mek inhibitor completely abrogates the KGA activity. Consistent with the regulation being phosphorylation-dependent, coexpressing KGA with protein phosphatase PP2A or treatment of purified KGA with alkaline phosphatase all block the elevated KGA activity. When key residues on the loop that stabilizes the catalytic core of KGA are mutated, such stimulation is lost. These results indicate that Raf-Mek-Erk signaling is linked to reprogramming of growth metabolism.
We show that the combined inhibitory effect of Mek2 and BPTES on KGA and cell proliferation could offer an exciting regime for multidrug therapy in cancers. Understanding how KGA itself is activated by the Raf-Mek signaling will also provide an alternative approach to further examine whether deregulation of KGA and its hyperactivation can be linked to this pathway. This could be the basis for abnormal cell growth in cancers and possibly other metabolic and neuronal disorders involving glutamate or ammonia as a main metabolite, such as hyperinsulinism/ hyperammonemia/hepatic encephalopathy and neurotransmission (25, 26). In this regard, we show that KGA forms a complex with Raf-1, Mek2, and Erk1/2, as well as the protein phosphatase PP2A, which can be anchored on a common scaffold protein or present as different subcomplexes. Whether this activation would involve KGA directly as a substrate of Raf-1, Mek2, or/and Erk or whether other immediate substrate(s) of these kinases exist to act as a coregulator for KGA remains to be further investigated. This is of particular interest because we show that bFGF, despite activating Mek/Erk, does not activate the KGA activity. This result highlights tight regulation of these three-tier kinases that could be spatially coordinated through different scaffolds or by other yet-unknown coregulators. In addition, the domain II of cKGA is homologous to several DEATH domain-containing proteins (Fig. S8B), which are known to regulate many signaling pathways, such as NF-κB and apoptosis (27).
In summary, the current structural, biochemical, molecular, and cell-based studies provide detailed insights into allosteric inhibition of human KGA by BPTES, the regulation of KGA activity and cell proliferation by EGF receptor-mediated Raf/Mek/Erk signaling module, and phosphorylation-dependent regime in cancer cell metabolism. This could pave the way for developing more specific therapeutic inhibitors toward KGA and Mek2-linked pathways and may offer a more effective strategy to tackle glutamine-addicted cancers with greater efficacy.
Materials and Methods
Protein production, enzymatic assays, and crystallographic studies of the cKGA construct is described in detail in the SI Text. Plasmid purification, transfection, cell proliferation assay, and immuno precipitation experiments were performed as previously described (28, 29). Details are provided in SI Text.
Acknowledgments
We thank Dr. Norman P. Curthoys (Colorado State University) for KGA plasmid; and MAX-lab (Lund, Sweden), Brookhaven National Laboratory, and National Synchrotron Radiation Research Center (NSRRC) (Taiwan) for synchrotron beamlines. This work was supported by Biomedical Research Council of Singapore Grants R154000461305 (to J.S.) and 07/1/21/19/506 (to B.C.L.), and partially supported by the Mechanobiology Institute of Singapore (to B.C.L.). H.S. and T.K. were supported by the Structural Genomics Consortium, a registered charity (1097737) that receives funds from Sweden, the United Kingdom, and Canada (http://www.thesgc.org/about/sponsors.php/). K.T. is a graduate scholar supported by the National University of Singapore.
Footnotes
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: Crystallography, atomic coordinates, and structure factors reported in this paper have been deposited in the Protein Data Bank, www.pdb.org (PDB ID codes 3VOY, 3VP0, 3CZD, 3VOZ, 3VP1, 3VP2, 3VP3, and 3VP4).
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1116573109/-/DCSupplemental.