Conformational changes in the activation loop of mitochondrial glutaminase C: A direct fluorescence readout that distinguishes the binding of allosteric inhibitors from activators

The first step in glutamine catabolism is catalysis by the mitochondrial enzyme glutaminase, with a specific isoform, glutaminase C (GAC), being highly expressed in cancer cells. GAC activation requires the formation of homotetramers, promoted by anionic allosteric activators such as inorganic phosphate. This leads to the proper orientation of a flexible loop proximal to the dimer-dimer interface that is essential for catalysis (i.e. the “activation loop”). A major class of allosteric inhibitors of GAC, with the prototype being bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl)ethyl sulfide (BPTES) and the related molecule CB-839, binds to the activation loop and induces the formation of an inactive tetramer (two inhibitors bound per active tetramer). Here we describe a direct readout for monitoring the dynamics of the activation loop of GAC in response to these allosteric inhibitors, as well as allosteric activators, through the substitution of phenylalanine at position 327 with tryptophan (F327W). The tryptophan fluorescence of the GAC(F327W) mutant undergoes a marked quenching upon the binding of BPTES or CB-839, yielding titration profiles that make it possible to measure the binding affinities of these inhibitors for the enzyme. Allosteric activators like phosphate induce the opposite effect (i.e. fluorescence enhancement). These results describe direct readouts for the binding of the BPTES class of allosteric inhibitors as well as for inorganic phosphate and related activators of GAC, which should facilitate screening for additional modulators of this important metabolic enzyme.

The increased reliance on glutamine catabolism by proliferating cancer cells, termed glutamine addiction, has recently attracted significant attention as a route to developing new therapeutics that target this unique metabolic requirement of transformed cells (1)(2)(3)(4). The outcome of elevated gluta-mine metabolism leads to increases in glutamine-fueled anaplerosis, where glutamine is first deamidated by the mitochondrial enzyme glutaminase (GLS) 3 to produce stoichiometric amounts of glutamate and ammonia. Glutamate is then deamidated by glutamate dehydrogenase (GDH) or by one of two transaminases (GOT/GPT), producing ␣-ketoglutarate, which is incorporated into the TCA cycle. In this way, glutamine, being the most abundant amino acid in the blood, acts as a primary carbon and nitrogen source for highly proliferative cells. Increased glutamine metabolism is triggered by several signal transduction pathways, including those influenced by HIF1␣, Myc, and Rho GTPases, as well as by Ras/MAPK-and mechanistic target of rapamycin (mTOR)/Akt-signaling activities (5)(6)(7)(8)(9)(10)(11)(12)(13).
There are two forms of GLS: one that is commonly referred to as kidney-type glutaminase (KGA) and the other a C-terminal splice variant, designated GAC. Because GAC is often highly expressed in cancer cells, it has been described as a "gatekeeper" enzyme for the elevated glutamine metabolism exhibited by these cells ("glutamine addiction"). Thus, GAC, as well as the longer GLS isoform KGA, are attractive therapeutic targets (14,15).
Currently, there are three classes of inhibitors for KGA and GAC. One class is represented by the benzophenanthridines, and, specifically, compound 968, which we previously demonstrated to act as a non-competitive allosteric inhibitor of GAC by interfering with its ability to undergo normal monomermonomer interactions that lead to GAC dimers and, ultimately, to activated tetramers (9,10,16). A second class of KGA/GAC inhibitors consists of analogs of the substrate glutamine, such as diazo-O-norleucine (DON), which binds to the enzyme active site and covalently modifies the catalytic serine (Ser-291) (17). The third class of inhibitors, depicted in Fig. 1A, consists of a number of bisthiadiazole derivatives, the prototype being BPTES (11, 18 -21). Gross et al. (22) have described a BPTES derivative, CB-839, which is a more potent inhibitor than 2Ϫ ) (red) or sulfate (SO 4 2Ϫ ) (blue) (B) or 500 nM BPTES or CB-839 (C) rapidly increases the FRET signal, indicating the formation of GAC tetramers. The increase in FRET is not readily reversible by addition of a 10-fold excess of unlabeled GAC subunits compared with DMSO control (black), reflecting the formation of stable drug-bound tetramers. In contrast, tetramers bound to anionic activators are readily reversible. The results shown are representative of three experiments. D, schematic illustrating the transition from a GAC dimer to a tetramer, where the tetramer species is depicted to give rise to the FRET signal and as being the active species. 488-labeled (green) GAC is the FRET donor, and QSY9-labeled (purple) GAC is the FRET acceptor. The allosteric activators inorganic phosphate (HPO 4 2Ϫ ) and sulfate (SO 4 2Ϫ ) are depicted to promote the dimer-to-tetramer transition to emphasize their ability to activate GAC, whereas the inhibitors BPTES and CB-839 bind to the FRET pairs at the dimer-dimer interface, forming a stable BPTES or CB-839-bound inactive GAC tetramer (BPTES is shown as green spheres).
BPTES, and showed it to be effective against triple-negative breast cancer cells. CB-839 efficacy has been examined in vivo and subsequently tested in clinical trials (23).
The discovery of BPTES as an inhibitor of KGA/GAC activity was first reported by Robinson et al. (18). Elucidation of the binding site of BPTES, based on the X-ray crystal structure solved for the inhibitor bound to the KGA/GAC enzymes, revealed that its interactions with a flexible loop within the dimer-dimer interface of the tetrameric forms of these enzymes (i.e. the "activation loop"), accounted for its mode of inhibition. Indeed, mutations along this loop ( 316 KEPSGLRFNKLF 327 ) can markedly impact enzyme activity. The functional consequences of these mutations vary from inducing constitutive activation in the absence of phosphate (K325A), to shifting the dose response for phosphate (F322Y/F327S, K316A), to inhibiting the formation of higher-order oligomers (K316Q) (20,24,25). Therefore, probing the conformation of this loop and how small molecules affect its orientation should provide a more detailed understanding of the fundamental mechanisms underlying the activation of the GLS isoforms.
Thus, we set out to introduce tryptophan residues at specific sites along the activation loop as conformational probes for monitoring the interactions of this loop with the BPTES class of allosteric inhibitors. In this study, we show that substituting a tryptophan residue for a phenylalanine at position 327 provides a sensitive reporter group for reading out the direct binding of BPTES and its clinically relevant analog CB-839. Moreover, this tryptophan reporter group now provides a real-time assay for the direct binding of allosteric activators such as inorganic phosphate and offers strong evidence for the distinct conformational changes induced within the activation loop by activators versus inhibitors.

Comparison of the effects of allosteric activators and BPTES class inhibitors on GAC
Previous studies by our laboratory and others have demonstrated the requirement for both KGA and GAC to form homotetramers to become activated (10, 26 -28). Using a FRET assay, we were able to monitor the effects that allosteric activators and inhibitors (Fig. 1A) have on GAC tetramer formation in real time. We observed an increase in the degree of FRET by the addition of the allosteric activators inorganic phosphate (HPO 4 2Ϫ ) and sulfate (SO 4 2Ϫ ) (Fig. 1B). The phosphate/sulfatebound GAC tetramers were found to be readily reversible upon addition of excess unlabeled GAC subunits, which compete with the GAC FRET pairs, resulting in a decreased FRET signal. Similarly, addition of the allosteric inhibitors BPTES and CB-839 to GAC labeled with FRET pairs increased the FRET readout, reflecting the rapid formation of tetramers upon the binding of these inhibitors (Fig. 1C). However, in the presence of either BPTES or CB-839, GAC tetramers were resistant to dissociation following the addition of excess unlabeled GAC subunits, demonstrating the formation of stable inhibitorbound tetramers. Under the same conditions, in the absence of either BPTES or CB-839, the addition of excess unlabeled GAC resulted in the ready exchange of dimeric subunits and the effective dilution of donor-acceptor FRET pairs (Fig. 1C, compare black (DMSO) versus blue (CB-839) and red (BPTES) curves). Notably, CB-839-bound GAC tetramers labeled with FRET pairs were more resistant to dissociation compared with their BPTES-bound GAC counterparts (Fig. 1C, compare blue (CB-839) versus red (BPTES)). GAC tetramer formation, which underlies the FRET changes and the stabilization provided by anions that stimulate glutaminase activity (i.e. SO 4 2Ϫ and HPO 4 2Ϫ ), or by inhibitors that enhance the formation and stability of GAC tetramers (i.e. BPTES and CB-839), are summarized schematically in Fig. 1D.
We also compared the ability of BPTES and CB-839 to inhibit recombinant GAC activity by monitoring the NADH fluorescence generated as an outcome of the coupled reactions used to assay the enzyme (10). Fig. 2A shows a direct comparison of the inhibition by BPTES and CB-839 following initiation of the reaction by the addition of glutamine but prior to the addition of the activator HPO 4 2Ϫ . Consistent with the increased stability of the GAC tetramer conferred by CB-839 compared with BPTES in GAC FRET assays (Fig. 1C, compare red versus blue traces), tetramers that formed in the presence of CB-839 were more resistant to subsequent activation by inorganic phosphate ( Fig. 2A, compare red (BPTES) and blue (CB-839) traces following addition of unlabeled GAC). The relative phosphate-stimulated activity is shown in Fig. 2B, where an initial rate analysis reveals a stronger inhibition by CB-839 compared with the same concentration of BPTES upon stimulation by an excess of allosteric activator.
Given the inhibitory properties of the BPTES class of molecules toward GAC and the established link between the degree of cellular glutaminase activity and glutamine dependence in cancer cell lines (29), we were interested in establishing an assay to probe the direct binding of these inhibitors to the enzyme and the underlying mechanisms by which they influence catalytic activity. Of particular interest was to better understand how they interact with the activation loop within the dimerdimer interface of GAC, where BPTES has been shown to bind based on X-ray crystallographic studies, and how such interactions inhibit the enzyme. Thus, we set out to engineer a fluorescence reporter group within the flexible activation loop to monitor its real-time movements in response to allosteric inhibitors and to compare such effects to those elicited by allosteric activators.

Examination of the BPTES binding site based on X-ray crystal structures
With respect to their ability to drive GAC tetramer formation, both inhibitors and activators result in similar FRET increases and yet elicit opposite outcomes regarding their effects on enzyme activity. To examine this at a structural level, we compared the available X-ray crystal structures for GAC and KGA. Currently, there are 16 high-resolution X-ray crystal structures for these GLS isoforms. It is important to note that KGA and GAC differ only in their C-terminal sequences, which are unresolved in all X-ray crystal structures. Upon careful analysis, only six of these structures provide sufficient electron density to resolve unambiguously the atomic positions of some of the residues within the activation loop (i.e. residues 320 GLRFNKL 326 ) (11,20,21). Of those structures with a resolved activation loop, four are co-crystal structures with BPTES (PDB codes 3VOZ, 3VP1, 3UO9, and 4JKT), one is a co-crystal structure with the active site inhibitor DON (PDB code 4O7D), and the other is without either BPTES or DON but includes two inorganic anions of sulfate per KGA monomer (PDB code 3VOY). Interestingly, in this KGA structure (PDB code 3VOY), one sulfate is bound proximal to the activation loop through interactions formed directly with tyrosine 399 and lysine 403 and represents the likely binding site for allosteric activators, although this has not been directly demonstrated.
We noted significant differences when comparing the two KGA/GAC structures that lack bound BPTES but have a sufficiently resolved activation loop (PDB codes 4O7D and 3VOY) with all BPTES-bound structures. To illustrate this, Fig. 3A presents a space-filling model of a DON-bound GAC tetramer (PDB code 4O7D), which has been aligned with the BPTES/ SO 4 2Ϫ -bound GAC structure (3VOZ), to show the binding sites for BPTES and sulfate. The marked changes in the orientation of the activation loop are illustrated in Fig. 3B, with the loop in the DON-bound GAC complex (cyan) being significantly displaced relative to its position in the BPTES-bound GAC structure (magenta). The displacement of the activation loop represents the extremes of the available conformational states observed in the KGA/GAC X-ray crystal structures to date, where the loop assumes a variety of conformations. We reasoned that it might be possible to observe this conformational transition in real time by replacing a wild-type residue with the amino acid tryptophan, i.e. to use it as a fluorescent reporter group. With this in mind, three residues were selected for tryptophan substitution; specifically, Phe-323, Phe-327, and Tyr-399. Both Phe-323 and Phe-327 are significantly displaced in the BPTES-bound structures (Fig. 3B, the Phe-327 movement is indicated by the red arrow), and Tyr-399 is directly adjacent to the thiadiazole ring of BPTES. These residues have been shown previously to be critical for the activation of the KGA/GAC enzymes as well as for the inhibition of their activity by BPTES (24,25).

The F327W mutation is sensitive to inhibitor binding
Upon individual substitution of these three residues to tryptophan, we initially analyzed their fluorescence properties (excitation, 285 nm; emission, 310 -390 nm) before and after the addition of BPTES. We found that BPTES induced a marked quenching in the tryptophan fluorescence of the GAC(F327W) mutant (Fig. 4A, compare black with red curves), whereas it did not change the fluorescence emission of WT GAC (Fig. 4B) or the F323W or Y399W mutants (data not shown). We then determined the molecular size of WT GAC and the individual tryptophan mutants in the absence and presence of BPTES using multiangle light scattering downstream of size exclusion chromatography (SEC-MALS). Significant shifts in the molecular weight distribution for the F327W mutant and WT GAC were observed following their preincubation with BPTES ( Fig.  4, C and D; solid lines represent elution of GAC, and broken lines represent the calculated molecular weight). The comparable increases in the mean molecular weights of WT GAC and the F327W mutant following preincubation with BPTES further demonstrate that the F327W mutant is able to bind to and form complexes with the inhibitor molecule in a manner similar to the wild-type enzyme. Additionally, this mutant exhibited V max and K M values for phosphate-stimulated activity that closely matched those of WT GAC (Fig. 4E), further demonstrating the functional equivalence of GAC(F327W) with the wild-type protein.
Based on the available crystallographic data, the ability of the F327W mutation to read out BPTES binding is likely due to the interactions of the inhibitor with the peptide backbone of the substituted tryptophan and not the outcome of direct contact with the indole side chain (Fig. 5A). Consistent with this, the orientation of the native phenylalanine in the BPTES-bound GAC structures indicates that the phenylalanine side chain does not interact with other residues. Alternatively, DeLaBarre et al. (20) described the electrostatic interaction of BPTES with the peptide backbone of residues Leu-326 and Phe-327. These interactions are formed between the peptide backbone and the nitrogen atoms within the thiadiazole ring of BPTES (Fig. 5A, dotted lines). Others have reported the observed electrostatic interaction between BPTES and the peptide backbone of these two residues but involving the sulfur atom of the thiadiazole ring of BPTES rather than its nitrogen moieties (Fig. 5B) (11,21). The requirement of BPTES analogs to possess a thiadiazole ring has been shown to be necessary to maintain potency (18,30,31). The interactions depicted in these two bound BPTES orientations suggest that the molecular basis for the observed changes in GAC(F327W) tryptophan emission, upon activation loop repositioning, results from the interaction of the loop with the thiadiazole group present in both BPTES and CB-839.

Changes in F327W fluorescence provide a direct binding readout for BPTES and CB-839
Derivatives of BPTES have proven to be effective inhibitors in vivo, with compound CB-839 being the most potent to date with a reported IC 50 value of ϳ30 nM (22). However, unlike BPTES, CB-839 achieved its maximal potency only when preincubated with GAC prior to assaying enzyme activity, suggesting that it has a slower rate of association with the enzyme (i.e. a slower on rate (22)). Similarly, CB-839 appeared to dissociate more slowly from GAC compared with BPTES (i.e. a slower off rate (22)). By monitoring the tryptophan fluorescence of the F327W mutant, we were able to directly compare the kinetics and binding efficiencies of these two inhibitors. As shown in Fig. 6, A and B, the addition of BPTES or CB-839 to the F327W mutant resulted in a dose-dependent quenching of tryptophan fluorescence. Each inhibitor reached equilibrium within 10 min, with CB-839 displaying slower binding kinetics than BPTES. The binding isotherms for the two inhibitors yielded K D values of 34 Ϯ 5 nM for CB-839 and 70 Ϯ 5 nM for BPTES (Fig. 6C). Thus, the binding affinities of these two inhibitors for GAC are more similar than the effects they show on the initial rates of phosphate-stimulated enzyme activity (Fig. 2, A and B) or when comparing their IC 50 values (i.e. 23 Ϯ 1 nM for CB-839 versus 108 Ϯ 17 nM for BPTES; data not shown). As will described under "Discussion," the differences observed between the direct binding and enzyme activity assays may reflect the differences in the on and off rates for the two inhibitors.

BPTES requires tetramer formation to bind to GAC
To demonstrate the requirement of BPTES to bind to the tetrameric form of GAC, we took advantage of a mutation along the dimer-dimer interface, namely D391K, that was shown previously to block the ability of GAC to form tetramers (10). Aspartate 391 is critical for GAC tetramer formation, where it forms a salt bridge with Lys-401 (both residues reside at the interface between two GAC dimers). Because of the twofold axis of symmetry of the GAC tetramer, the single change of aspartate 391 to lysine, and hence the introduction of a chargecharge repulsion, result in the disruption of four salt bridges per GAC tetramer that is sufficient to prohibit tetramer formation (10). To investigate whether BPTES binds to the activation loop of an inactive GAC dimer, we introduced the D391K mutation into the GAC(F327W) background and found that this single substitution completely ablated the ability of BPTES to induce a change in tryptophan fluorescence (Fig. 7A, compare the red and black curves). We further examined the GAC(F327W,D391K) double mutant using SEC-MALS to con-  firm that it existed in a dimeric state both with and without pre-incubation of BPTES. We found that the GAC double mutant indeed exhibited a molecular weight distribution consistent with a dimer, in the presence and absence of BPTES, thus further illustrating the requirement of a GAC tetramer to form for BPTES to bind and induce an inhibitory state (Fig. 7B, solid lines represent elution of GAC, and broken lines represent the measured molecular weight).

Allosteric activators enhance GAC(F327W) fluorescence
Previous studies of the activation of the KGA/GAC enzymes revealed that their enzymatic activity was markedly increased  upon addition of various inorganic anionic ions (18,26,27,32). The most potent of these activators, inorganic phosphate, has been the most studied, having been shown to affect the oligomeric state of both the KGA and GAC isoforms. Additionally, recent high-resolution X-ray crystal structures have revealed potential binding sites for these anions (Fig. 3, A and B) and suggest that they exert their effects on enzymatic activity by changing the conformation of the activation loop (11).
Here we examined the effects of these anionic activators on the tryptophan fluorescence of the GAC(F327W) mutant. We first tested anions that have been reported to activate glutaminase enzymes; namely, inorganic phosphate and sulfate. Unlike the changes observed for BPTES and CB-839, addition of these anions to GAC(F327W) resulted in marked enhancement of tryptophan fluorescence (Fig. 8A), thus providing evidence that they directly affect the conformation of the activation loop. Interestingly, we found that the dose-dependent increases of fluorescence enhancement correlated well with the ability of a given anion to activate GAC. More specifically, sulfate (SO 4 2Ϫ ) was ϳ50% as effective as phosphate (HPO 4 2Ϫ ), which matched their abilities to enhance the tryptophan fluorescence of the F327W mutant (i.e. 15% versus 25%, respectively). Titrations with these anions yielded binding curves that were in good agreement with their ability to activate WT GAC (Fig. 8B, compare closed black and red circles with blue diamonds and triangles), suggesting that they act through a similar mechanism by affecting the conformation of the activation loop.
We next investigated the effects of these ions on the molecular size distribution of GAC as read out by SEC-MALS. Previous studies have described the formation of polytetrameric complexes that result from phosphate binding to high concentrations of enzyme (i.e. Ͼ1 mg/ml), where GAC tetramers bind end-on-end to form polymers. This led to the suggestion that these polytetrameric complexes might be necessary for stimulating enzyme activity (25,26). Interestingly, we found that sulfate, which has not been described previously to have an effect on the formation of GAC oligomers, stimulated the formation of a mixture of 16-mers and 8-mers but not the larger oligomers that are induced by phosphate (Fig. 8C, compare the dashed and dotted blue lines representing molecular weight for eluted species). These results, together with the GAC(F327W) fluorescence readout, suggest that the orientation of the activation loop results in an active conformation of the enzyme, which can then facilitate the formation of polytetrameric oligomers. They are also reflective of the differences in the ability of sulfate versus inorganic phosphate to stimulate GAC enzymatic activity and to enhance F327W fluorescence.

Inorganic phosphate binds to the tetrameric state of GAC
The mechanism by which inorganic phosphate activates GAC activity is still unclear. In fact, it has been suggested that phosphate binds within the active site following catalysis, where it creates an electrostatic repulsion with the product to increase its off rate, thereby effectively increasing catalytic turnover (21). The second proposed mechanism is that phosphate interacts with the activation loop at the dimer-dimer interface, much like allosteric inhibitors such as BPTES, and enables the loop to access an active conformation to stimulate catalysis (11). The enhancement of the tryptophan fluorescence of the F327W mutant observed here is consistent with the latter mechanism. Furthermore, we found that the fluorescence enhancement induced by phosphate addition was dependent on GAC concentration and, therefore, the proportion of tetramers in solution. When this enhancement was plotted with respect to the concentration of GAC, it was in good agreement with FRET measurements that reflect the binding isotherm representing the transition of GAC dimers to tetramers and the concentration-dependent activation of the WT enzyme (Fig. 8D, compare the black triangles on the left axis with the blue and green circles on the right axis). Taken together, these results suggest that phosphate binding promotes catalysis by interacting with the GAC tetramer, thereby inducing a confor- mational change within the activation loop, which, in turn, promotes enzymatic turnover.

Discussion
Prior studies of the structure-function relationships of glutaminases have revealed their essential role in glutamine metabolism within mammalian cells, where their activity is governed by their self-association to form tetramers (28,(32)(33)(34)(35). However, the mechanism by which tetramer formation stimulates enzymatic activity is not fully understood, presenting a significant challenge for rational drug design targeting this family of enzymes. The fact that a class of small molecule inhibitors, namely the bisthiadiazole derivatives, appeared to stabilize an inactive, tetrameric state of the KGA/GAC enzymes was an important discovery and further highlights the interest in understanding the mechanisms regulating glutaminase activity. Therefore, we set out to develop a spectroscopic readout for the flexible activation loop of GAC, where BPTES binds, with the aim of obtaining mechanistic insights into how allosteric activators and inhibitors impact the loop and thereby exert their regulatory functions.
Recently, there has been a concerted effort to find small molecule inhibitors of the glutaminase enzymes using medicinal chemistry approaches (22,31,36,37). Here we first compared the differences between the allosteric activators inorganic phosphate and sulfate and the two most commonly used allosteric inhibitors, CB-839 and BPTES, in fluorescence assays that directly monitor GAC tetramer formation. We found that both allosteric activators and inhibitors induced the formation of GAC tetramers; however, inhibitors were distinguished from activators by their ability to better stabilize the tetrameric state. CB-839 appears to act similarly to BPTES, so that the addition of either drug enhances tetramer formation. However, the CB-839-bound GAC tetramers labeled with FRET pairs were much less able to undergo exchange with excess unlabeled GAC subunits compared with BPTES-bound tetramers. Overall, these results agree with previous observations that CB-839 acts through a similar mechanism as BPTES, by binding at the dimer-dimer interface to promote the formation of an inactive tetramer, but with CB-839 having a significantly slower off rate compared with BPTES.
To address how allosteric activators and inhibitors both enhance GAC tetramer formation but exert opposing effects on enzyme activity, we examined the available X-ray crystal structures of the KGA/GAC enzymes and noticed differences between the conformations of the activation loop. We reasoned that, by comparing the activation loop in the BPTES-bound GAC structure versus that for free GAC, we might be able to identify potential residues that, when substituted with the fluorescent amino acid tryptophan, would be sensitive to the dynamics of the loop. Indeed, we found that the selective substitution of a single amino acid within the activation loop (specifically, changing phenylalanine 327 to a tryptophan residue) yields an enzyme that retains normal catalytic activity while providing a useful readout for the binding of allosteric inhibitors and activators. The native phenylalanine at position 327 has been shown previously to be an important residue for the activation of the KGA/GAC (GLS) isoforms (20). Additionally, Phe-327 is one of only two residues that differ between the liver-type (GLS2) and kidney-type (GLS) glutaminase enzymes in the activation loop, where the residue corresponding to phenylalanine at position 323 on the GLS enzymes is a tyrosine in GLS2, and phenylalanine 327 in GLS corresponds to a serine in GLS2 (11,20). GLS and GLS2 have been reported to have different catalytic properties (27), and it is of interest in this study that, of the various substitutions examined in the activation loop of GAC, only the F327W mutant provided a useful readout for both allosteric activators and inhibitors.
CB-839 has been suggested to be a more potent inhibitor of glutaminase activity compared with BPTES, with reported IC 50 values of 30 nM for CB-839 and between 80 nM and 3 M for BPTES (11, 18 -20, 22, 31, 38). However, the direct binding of these inhibitors to KGA or GAC has never been compared, which we now have been able to do by taking advantage of the quenching of the intrinsic tryptophan fluorescence of the GAC(F327W) mutant induced by these compounds. Based on their titration profiles, we determined that the two inhibitors have comparable dissociation constants (30 nM and 70 nM for CB-839 and BPTES, respectively (Fig. 6C)). The differences between the range of values reported for the IC 50 of BPTES from cell studies versus the K d value we have determined from direct binding assays may, at least in part, reflect a reduced efficiency in the ability of BPTES to reach its cellular target GLS because of the potential of this drug to accumulate in membranes. However, analysis of the real-time fluorescence data demonstrates that the rates of association and dissociation of CB-839 versus BPTES from the enzyme are different, with CB-839 showing slower binding kinetics (Fig. 6, A and B). This indicates a longer residence time for CB-839 that may contribute to some of its enhanced ability to inhibit phosphate-stimulated enzyme activity.
The fluorescence assays described here for GAC tetramer formation, together with those for inhibitor binding and enzyme catalysis, allow for a more detailed analysis of how different classes of small molecule inhibitors interact with glutaminase to alter the activity of this important metabolic enzyme. Fig. 9 depicts a model in which BPTES-like inhibitors are incorporated into GAC tetramers in a manner that positions the activation loop in an orientation that prohibits catalysis and is reflected in the quenching of intrinsic GAC(F327W) fluorescence. Stabilization of the tetrameric form of GAC is similarly achieved by addition of anionic activators, but, conversely, the binding of phosphate or sulfate increases intrinsic GAC(F327W) fluorescence in proportion to the observed enzyme activity. Our model proposes that the attendant increases in activity and fluorescence result from the stabilization of the activation loop in a conformation that reconfigures the glutamine binding site for catalysis. Although the model as drawn suggests that the binding of activators and inhibitors is mutually exclusive, we cannot rule out the possibility that both reside simultaneously on GAC, competing for loop positioning and the corresponding fluorescence/activity changes on the same GAC monomer.
The real-time fluorescence assays we developed have provided an important and additional benefit; namely, the ability to directly monitor the interactions of allosteric anionic activators with GAC. Allosteric activators, such as inorganic phosphate, have been shown to promote changes in the oligomeric state of KGA/GAC enzymes that accompany their activation. However, although KGA/GAC oligomer formation has been consistently observed, the binding site for these anionic activators and the mechanistic outcome of oligomer formation are still not Figure 9. Schematic of the dimeric and tetrameric species of GAC as influenced by allosteric activators and BPTES class inhibitors. Inactive GAC dimers associate to form active GAC tetramers in the absence of allosteric activators or inhibitors (basal enzyme activity). Anionic activators, such as inorganic phosphate, bind to the GAC tetramer at the activation loop to induce an active conformation, resulting in activated tetramers with the potential of forming higher-order oligomers (not shown). BPTES class inhibitors bind to the activation loop in GAC tetramers, locking the activation loop in an inactive state, forming highly stable drug-bound tetramers.
well understood. Recent studies have shown that mutations along the activation loop constitutively activate KGA/GAC so that anionic activators elicit no additional stimulatory effects (20,24,25,39), suggesting that they exert their actions through the activation loop. Our findings show that allosteric activators like inorganic phosphate induce an enhancement of the tryptophan fluorescence of the GAC(F327W) mutant, directly correlating with their ability to stimulate GAC activity. These results strongly support the idea that allosteric anionic activators bind at the dimer-dimer interface, changing the conformation and environment of the loop to varying degrees to promote activated GAC tetramers (Fig. 9). Our findings further suggest that the activation loop influences the global structure of the enzyme, which can lead to the formation of polytetrameric GAC complexes (25,26).
In conclusion, our results highlight the important roles the activation loop of GAC plays in regulating enzyme activity. The results presented here show that BPTES and CB-839 interact with the activation loop in a similar manner and demonstrate the utility of this novel F327W mutant and accompanying fluorescence assays for screening the effects bisthiadaizoles exert on the GLS enzymes. Similarly, we show that allosteric anionic activators directly impact the conformational dynamics of the activation loop, pointing to the potential usefulness of the F327W mutant in screening for additional small molecules capable of activating these enzymes.

Preparation of recombinant GAC
GAC was expressed and purified as described previously (10). Briefly, a plasmid encoding the mouse kidney-type glutaminase isoform 2 (GAC, NP_001106854.1, residues 73-603) was cloned into a pET28a vector containing an N-terminal histidine (His) tag and thrombin cleavage site. The expressed protein was purified using Co 2ϩ affinity beads (Clontech), followed by anion exchange (GE Healthcare) and gel filtration chromatography. Purified GAC was stored in a high salt-containing buffer (20 mM Tris-HCl (pH 8.5), 500 mM NaCl, and 1 mM NaN 3 ) at Ϫ80°C following snap-freezing in liquid N 2 for longterm use.

Protein labeling with FRET pairs
Labeling recombinant GAC with small molecule probes was performed by methods described previously (10). Briefly, 1.5 mg of GAC was first exchanged into 50 mM HEPES (pH 7.2) and 100 mM NaCl (labeling buffer) using a PD10 desalting column (GE Healthcare). The protein was then incubated with either 50 M (5-fold excess over enzyme) Alexa Fluor 488 succinimidyl ester or QSY9 succinimidyl ester (Molecular Probes) for 1 h at 4°C. The labeling reaction was quenched with 150 mM Tris-HCl (pH 8.5), and unreacted probe was separated from labeled enzyme using a PD10 desalting column, eluting labeled GAC into the high salt-containing buffer.

MALS analysis
Purified GAC and GAC mutants were subjected to MALS as described previously by Møller et al. (26). Briefly, 50-l samples of 5 mg/ml GAC with or without 30-min preincubation with 100 M BPTES were injected onto a BioSep-SEC-S4000 size exclusion column (Wyatt Technology) coupled to a static 18-angle light scattering detector (DAWN HELEOS-II) and a refractive index detector (OptiLab T-rEX, Wyatt Technology) at 23°C. The size exclusion column was equilibrated with 20 mM Tris-HCl (pH 8.5) and 200 mM NaCl and, when appropriate, 50 mM K 2 HPO 4 or K 2 SO 4 . The flow rate was kept at 1 ml/min. Root mean square radius and mass distribution (polydispersity) were analyzed using ASTRA software, with monomeric BSA (Sigma) serving to normalize the light-scattering signal.

Glutaminase assays
Activity assays used to evaluate the activity of GAC mutants and inhibition by BPTES and CB-839 followed a two-step protocol adapted from Robinson et al. (18). For example, when assaying phosphate-or sulfate-stimulated activity, 20 l of 20 mM glutamine, 50 mM Tris acetate (pH 8.5), and 0.1 mM EDTA, containing the indicated concentrations of K 2 HPO 4 or K 2 SO 4 , were added to a UV-transparent Costar 96-well plate (Corning). For inhibition assays, drugs dissolved in DMSO were added to 20 l of 20 mM glutamine, 50 mM Tris acetate (pH 8.5), 0.1 mM EDTA, and 50 mM K 2 HPO 4 . To initiate the reaction, 5 l of a solution of GAC prepared in 20 mM Tris-HCl (pH 8.5), 100 mM NaCl, and 1 mM NaN 3 , to give a final enzyme concentration of 50 nM in the assay, or a GAC concentration of 100 nM for inhibition assays, was added to the glutamine solution and incubated at 23°C for 2 min before the reaction was quenched using 2.5 l of 3 M HCl. The second step was initiated by addition of 200 l of 12 units/l GDH, 2 mM NAD ϩ , 100 mM hydrazine (Sigma), and 100 mM Tris-HCl (pH 9.2) on top of the first quenched reaction and incubated for 45 min at 23°C before reading NADH absorbance. Glutamate produced by the first reaction was determined from the amount of NADH generated in the second reaction by using the extinction coefficient for NADH (6220 M Ϫ1 cm Ϫ1 ).

Fluorescence measurements
Fluorescence experiments were performed with a Varian Cary Eclipse fluorimeter in counting mode, using 1 ml samples with continuous stirring at 20°C in 50 mM Tris acetate (pH 8.5) and 0.1 mM EDTA. For FRET assays, 75 nM QSY9-labeled GAC was added to 25 nM 488-labeled GAC while monitoring fluorescence emission at 520 nm with excitation at 490 nm. After equilibration for 10 min, 100 mM K 2 HPO 4 or K 2 SO 4 , or either 500 nM BPTES or CB-839, was added and equilibrated for 5 min. Finally, 1 M unlabeled WT GAC was added to the assay incubation. For tryptophan emission scans, the excitation and emission wavelengths were 285 and 310 -390 nm, respectively. For kinetic experiments, the excitation and emission wavelengths were 285 and 340 nm, respectively. For drug titrations, BPTES and CB-839 at the indicated concentrations were added to 100 nM GAC(F327W) to give less than 0.5% (v/v) DMSO. Points for the dose-dependent quenching by inhibitors were taken from the equilibrated binding curves and fit to a bimolecular ligand binding equation after correcting for fractional saturation using the relation of free and unbound enzyme/drug complex (i.e. f bound ϩ f free ϭ 1, where f bound ϭ F/F max ). Similarly, 1 M BPTES was added to 100 nM GAC(D391K-F327W). For allosteric activator titrations, the indicated concentrations of K 2 HPO 4 or K 2 SO 4 were added to 500 nM GAC(F327W).
Author contributions-C. A. S., J. W. E., and R. A. C. conceived and designed the study. C. A. S. performed the experiments. C. A. S., J. W. E., and R. A. C. wrote and edited the manuscript.