Mechanism by which a recently discovered allostericinhibitor blocks glutamine metabolism intransformed cellsClint A. Stalneckera, Scott M. Ulrichb, Yunxing Lia, Sekar Ramachandrana, Mary Kate McBrayerc, Ralph J. DeBerardinisd,Richard A. Cerionea,e,1, and Jon W. Ericksona

Departments of aChemistry and Chemical Biology and eMolecular Medicine, Cornell University, Ithaca, NY 14853; bDepartment of Chemistry, Ithaca College,Ithaca, NY 14850; cCenter for Dementia Research, Nathan S. Kline Institute, Orangeburg, NY 10962; and dChildrens Medical Research Institute, University ofTexas Southwestern Medical Center, Dallas, TX 75390

Edited by Joseph Schlessinger, Yale University School of Medicine, New Haven, CT, and approved December 5, 2014 (received for review July 23, 2014)

The mitochondrial enzyme glutaminase C (GAC) catalyzes the hy-drolysis of glutamine to glutamate plus ammonia, a key step in themetabolism of glutamine by cancer cells. Recently, we discovereda class of allosteric inhibitors of GAC that inhibit cancer cell growthwithout affecting their normal cellular counterparts, with the leadcompound being the bromo-benzophenanthridinone 968. Here, wetake advantage of mouse embryonic fibroblasts transformed by on-cogenic Dbl, which hyperactivates Rho GTPases, together with 13C-labeled glutamine and stable-isotope tracing methods, to establishthat 968 selectively blocks the enhancement in glutaminolysis neces-sary for satisfying the glutamine addiction of cancer cells. We thendetermine how 968 inhibits the catalytic activity of GAC. First, wedeveloped a FRET assay to examine the effects of 968 on the abilityof GAC to undergo the dimer-to-tetramer transition necessary forenzyme activation. We next demonstrate how the fluorescence ofa reporter group attached to GAC provides a direct read-out of thebinding of 968 and related compounds to the enzyme. By combiningthese fluorescence assays with newly developed GAC mutants trap-ped in either the monomeric or dimeric state, we show that 968 hasthe highest affinity for monomeric GAC and that the dose-dependentbinding of 968 to GACmonomers directly matches its dose-dependentinhibition of enzyme activity and cellular transformation. Together,these findings highlight the requirement of tetramer formation asthe mechanism of GAC activation and shed new light on how a dis-tinct class of allosteric GAC inhibitors impacts the metabolic programof transformed cells.

glutaminase | glutaminolysis | FRET | benzophenanthridines | Rho GTPases

Recently, the mitochondrial enzyme glutaminase (GLS1) hasgained significant attention as a therapeutic target for cancer

(1–3). GLS1 catalyzes the hydrolysis of glutamine to glutamate,which is used in the citric acid cycle (TCA) of cancer cells un-dergoing an aberrant glycolytic flux (i.e., the Warburg effect) asa non–glucose-derived source for anaplerosis. The elevation inglutamine metabolism exhibited by many cancer cells (“glutamineaddiction”) is critical for sustaining their proliferative capacity, aswell as for other aspects of their transformed phenotypes (4–9).Work from our laboratory has shown that a specific GLS1 splicevariant, glutaminase C (GAC), plays an essential role in thetransformation of NIH 3T3 fibroblasts by Rho GTPases, as wellas in the proliferative and invasive activities of various cancercells (10, 11). Thus, given the importance of GAC expression andactivation for oncogenic transformation, the identification ofinhibitors that target this metabolic enzyme offers new oppor-tunities for the development of anticancer drugs.Because glutamine is necessary for a range of biochemical

reactions, including nucleotide and protein synthesis, glutamineanalogs like the GLS1 inhibitor diazo-O-norleucine (DON) (12,13) are not ideal candidates for cancer drugs (14). However,two classes of allosteric inhibitors of GAC have been identifiedand offer more promising options as lead compounds for the

development of cancer therapeutics. One of these consists ofanalogs of bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl)ethylsulfide (BPTES), a reversible GAC inhibitor that has beenextensively characterized (15, 16). X-ray crystal structures of theGAC–BPTES complex show that BPTES effectively traps GACas an inactive tetramer (16–18).A second, more recently identified, class of allosteric GAC

inhibitors that is highly specific for inhibiting cancer cell growthwhile having little effect on normal (nontransformed) cells isrepresented by the benzophenanthridinone 968 (11). However,until now, very little was known regarding how this class ofmolecules functions. Here, we establish that the novel GAC in-hibitor, 968, negatively impacts glutaminolysis in transformedcells, as well as determine how it influences GAC activity. First,we demonstrate through 13C-isotopic labeling studies that 968inhibits the elevation in glutamine metabolism that accompaniesRho GTPase-dependent transformation. We go on to determinehow 968 regulates GAC activity in vitro. By developing a FRETassay to read out the ability of GAC to undergo a dimer-to-tetramer transition necessary for enzyme activation (15, 19, 20),we show that, although the binding of BPTES induces a stablehigh-affinity GAC tetramer, 968 neither inhibits nor promotestetramer formation. However, we discovered that the binding of968 to the FRET donor-labeled GAC caused a significant dose-dependent quenching of the donor fluorescence, even in theabsence of a FRET acceptor, which directly correlated with the

Significance

The work described here was motivated by our previous dis-covery of a connection between Rho GTPase activation and theup-regulation of mitochondrial glutaminase C (GAC), which isresponsible for satisfying the glutamine addiction of cancer cells.This connectionwas originally established by our identification ofa lead compound, 968, for a new class of inhibitors of oncogenictransformation. Although GAC was identified as the putativetarget for 968, how it regulated GAC was poorly understood.Here we provide important insights into the actions of 968,through the development of novel assays for its direct binding toGAC and its effects on enzyme activity. These findings offer ex-citing new strategies for interfering with the metabolic reprog-ramming critical for malignant transformation.

Author contributions: C.A.S., S.R., R.J.D., R.A.C., and J.W.E. designed research; C.A.S., Y.L.,M.K.M., and J.W.E. performed research; C.A.S., S.M.U., and R.J.D. contributed new re-agents/analytic tools; C.A.S., S.R., R.J.D., R.A.C., and J.W.E. analyzed data; and C.A.S.,R.A.C., and J.W.E. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: rac1@cornell.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1414056112/-/DCSupplemental.

394–399 | PNAS | January 13, 2015 | vol. 112 | no. 2 www.pnas.org/cgi/doi/10.1073/pnas.1414056112

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020

ability of 968 to inhibit GAC activity. Using this novel 968-binding assay, in combination with recently developed GAColigomerization-defective mutants, we show that 968 preferen-tially interacts with the monomeric form of GAC. Moreover, thedose–response for the binding of 968 to monomeric GACmatches that for its inhibition of oncogenic transformation.These findings provide important insights into the mechanism bywhich 968 blocks GAC activation and glutamine metabolism, aswell as open the way toward developing novel therapeutics tar-geting glutamine-dependent cancer cells.

ResultsThe GAC Inhibitor 968 Blocks Glutaminolysis in Transformed Cells.Previous work by our laboratory identified a potential connectionbetween glutamine metabolism and Rho GTPase-dependent onco-genic transformation through the discovery of the small moleculeinhibitor 968 (11). We showed that 968 specifically inhibited thegrowthof transformed cells and various cancer cells by blockingGACactivation, although the detailed mechanism was unclear. Here, wehave setout tobetter understandhow968 functions by first examiningits effects on glutamine metabolism in a well-defined model systemfor oncogenic transformation, in which the stable expression of theDbl oncogene in mouse embryonic fibroblasts (MEFs) is controlledby the removal of doxycycline (Dox).Induction of oncogenic Dbl in MEFs results in marked

changes in cell morphology, as a result of cytoskeletal rearr-angements caused by the activation of Rho GTPases (Fig. 1A)(10, 21–23). These morphological changes accompany the abilityof oncogenic Dbl-expressing cells to overcome contact inhibitionto form foci. Consistent with our previous results (11), treatment

of MEFs expressing oncogenic Dbl with 968 blocked focus for-mation (Fig. 1B). We examined whether these effects were ac-companied by an inhibition of glutaminolysis. The induction ofoncogenic Dbl expression in MEFs increased glutaminolysis andglutamine-dependent anaplerosis, as monitored by 13C enrichmentin TCA cycle intermediates derived from [U-13C]-glutamine (Fig.1 C and D). Treatment of Dbl-expressing cells with 968 causedsignificant reductions in the glutamine-derived 13C isotopic en-richment within TCA cycle intermediates, but did not result in thedepletion of relative pool sizes of each metabolite, with the ex-ception of an almost twofold reduction in intracellular glutamate(Fig. S1A, histograms labeled glutamate, compare red and yellowvs. black, green, and blue). The reduction in intracellular gluta-mate along with the reduced 13C labeling would be an expectedresult if 968 were indeed inhibiting glutaminase activity. To furthercorroborate these observations, we also measured the absoluterates of secreted ammonia and glutamate (i.e., the products of theglutaminase reaction) in the growth medium and found that in fact968 potently inhibited the accumulation of the direct products ofglutaminolysis (Fig. S1B). A modest inhibition of glutaminolysis by968 was also observed in MEFs not expressing Dbl (see Fig. S1Cfor the 13C enrichment diagram, and the M+5 histograms for Dbl-OFF, ±968, in Fig. S1D, and the M+4 histograms in Fig. S1 E–G).However, previous studies by Wang et al. (11) have reported nosignificant decrease in cell proliferation in nontransformed NIH3T3 (Fig. 1 D and E) and human mammary epithelial cells(HMECs; Fig. 4 F and G), suggesting that glutamine metabolismis critical for supporting the transformed phenotypes accompa-nying oncogenic Dbl expression but not for the proliferative ca-pability of normal MEFs. Treatment with the less potent 968-analog, compound 27, caused a weaker inhibition of the 13Cenrichment of metabolites (Fig. 1D), consistent with its reducedpotency to inhibit enzymatic activity (see below).Oncogenic Dbl induction did not cause marked increases in

glucose-fueled anaplerosis, as measured by 13C enrichment in cit-rate, when using [U-13C]-glucose as a tracer (see the M+2 histo-grams in Fig. S2A), demonstrating that a highly specific stimulationof glutamine metabolism accompanies Rho GTPase-dependenttransformation. However, 968 was observed to inhibit glucose la-beling of citrate isotopologues (see M+2 histogram in Fig. S2B).This presumably was due to the inhibition of glutamate flux by 968.Overall, these results show how 968 attenuates cellular glutaminemetabolism and restores a normal growth phenotype in cellsexpressing oncogenic Dbl, thus highlighting the role of glutamineas a critical source for anaplerosis during cellular transformation.

Examining the Effects of 968 on the Dimer-to-Tetramer Transition ofGAC. The transition of GAC from a dimer to a tetramer has beensuggested to be essential for enzyme activity (15, 19, 20). A well-established allosteric inhibitor of GAC, BPTES, has been shownto stabilize an inactive tetrameric state of the enzyme (15). Thus,we examined whether 968 acted in a similar manner by de-veloping a real-time read out for the GAC dimer-to-tetramertransition. Fig. 2A depicts the proposed FRET assay, whereoligomer formation is monitored between two populations ofpurified recombinant GAC, labeled with either the highly fluo-rescent AlexaFluor 488 (donor) probe or with the nonfluorescentQSY9 (acceptor) chromophore. A major advantage of usingFRET as a direct read out for GAC tetramer formation is theability to monitor oligomer formation at the low concentrationsof GAC commonly used for assaying its enzymatic activity.The addition of QSY9-GAC to 488-GAC yielded a dose-

dependent quenching of the donor 488 emission due to FRET,which was reversible on the addition of unlabeled GAC, dem-onstrating that GAC tetramer formation is a dynamic process(Fig. 2B). The dose-dependent binding isotherm obtainedfrom the QSY9-GAC titration profile directly correlated withthe basal activation of GAC that occurs at increasing protein

A

B

C

D

Fig. 1. Dbl-induced transformation and increased glutaminolysis are inhibitedby 968. (A) Fluorescent staining before (+Dox) and after (−Dox) a 24-h inductionof Dbl-inducible MEFs with anti-actin (red) and anti-HA (green) antibodies. (B)Expression of Dbl confers the ability of MEFs to form foci, which is blocked bytreatment with 10 μM 968. (C) Diagram showing 13C enrichment from [U-13C]-glutamine into TCA cycle intermediates, where GAC activation downstream fromDbl is highlighted. 13C-carbons are shown as dark-filled circles and 12C-carbons aslight-filled circles. (D) Glutamine-derived metabolites (glutamate M+5, fumarateM+4, malateM+4, citrateM+4) were normalized to 13C enrichment observed forMEFs not expressing Dbl. Comparisons were made between treatment with 968,its less potent analog 27 (see Fig. 3D for molecular structures), and untreatedcells. Bars represent the mean (±SD) of triplicate determinations. P values weredetermined by the Student t test (*P < 0.05, **P < 0.005).

Stalnecker et al. PNAS | January 13, 2015 | vol. 112 | no. 2 | 395

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020

concentrations (i.e., due to tetramer formation through massaction), yielding an apparent KD of 164 nM (±20 nM) for tet-ramer formation (Fig. 2C), supporting the contention that theGAC tetramer is the minimal unit for enzymatic activity.We then compared the effects of 968, vs. BPTES, on the GAC

dimer-to-tetramer transition. Consistent with previous findingsthat BPTES stabilizes GAC as an inactive tetramer (16), wefound that it caused an immediate quenching of 488-GACfluorescence emission when added to a mixture of 488-GAC andQSY9-GAC (Fig. 2D), i.e., due to the ability of BPTES to pro-mote the formation of 488-GAC:QSY9-GAC (donor:acceptor)tetramers. These stable GAC:BPTES tetrameric complexes wereless susceptible to reversal by the addition of unlabeled GAC(Fig. 2D, compare the green trace for the addition of unlabeledGAC in the presence of the vehicle control DMSO vs. the blacktrace that represents the addition of unlabeled GAC in thepresence of 5 μM BPTES). Interestingly, 968 elicited a markedlydifferent response, causing a significant change in the fluores-cence emission of 488-GAC, followed by a partial fluorescencerecovery on the addition of excess unlabeled GAC (Fig. 2E). Therecovery of 488 fluorescence when adding excess unlabeled GACwas due to the elimination of FRET between 488-GAC and

QSY9-GAC, following the formation of mixed tetramers be-tween 488-GAC or QSY9-GAC and unlabeled GAC. Thus, 968does not appear to interfere with GAC tetramer formation.However, the inability to achieve a full recovery of the fluorescenceemission suggested that 968 binding was directly affecting 488-GACdonor fluorescence emission. Indeed, we found that 968 causeda dose-dependent quenching of 488-GAC emission (in the absenceof the FRET acceptor QSY9-GAC) that matched the 968-mediatedinhibition of GAC activity (Fig. 2F and Fig. S3). Taken together,these findings show that 968 does not mimic the actions of BPTESby trapping GAC in an inactive tetrameric state but instead regu-lates GAC activity through a distinct allosteric mechanism.

Real-Time Monitoring of 968 Binding to GAC and Its Inhibition ofEnzyme Activity. We developed a real-time enzyme activity assayto simultaneously monitor both the binding of 968 to GAC and itseffects on enzyme activity. The enzymatic activity of 488-GAC isassayedbymonitoringNADHproduction (i.e., fluorescenceemissionat 460 nm) that accompanies the conversion of glutamate (theproductof theGAC-catalyzedreaction) toα-ketoglutarate, catalyzedby glutamate dehydrogenase. Fig. 3A depicts the coupling of thesetwo fluorescence assays and Fig. 3B shows the results of an

A B

C D

E F

Fig. 2. Real-time fluorescence assay detecting GAC tetramer formation. (A) Schematic depiction of the FRET assay. (B) 25 nM 488-GAC (donor) fluorescence isquenched on addition of QSY9-GAC (acceptor) in a dose-dependent manner and reversed with the addition of a 10-fold excess of unlabeled GAC. (C) FRETresulting from the titration of 25 nM 488-GAC with increasing amounts of QSY9-GAC (○) overlaid with concentration-dependent in vitro activation of GAC(●). FRET data were fit to a quadratic binding isotherm. Points represent the mean ± SD of three independent experiments. (D) Increasing amounts of BPTESwere added to 25 nM 488-GAC and 25 nM QSY9-GAC to examine the effects of the inhibitor on GAC tetramer formation. A 10-fold excess of unlabeled GACwas added to attempt to reverse tetramer formation. (E) 968 induces a dose-dependent quenching of 488-GAC fluorescence that is distinct from thequenching induced by the addition of QSY9-GAC. (F) Fluorescence quenching on addition of different concentrations of 968 to 10 nM 488-GAC in the absenceof QSY9-GAC.

396 | www.pnas.org/cgi/doi/10.1073/pnas.1414056112 Stalnecker et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020

experiment simultaneously monitoring the direct binding of 968 toGAC (green solid line) and its inhibition of enzyme activity (bluesolid line; the blue dashed line represents the control enzymeactivity treated with the solvent vehicle DMSO). Unlike 968,BPTES does not directly affect 488-GAC fluorescence, underconditions where it strongly inhibits GAC activity (Fig. 3B, seethe green and blue dotted traces, respectively).We then adapted these assays to a 96-well plate format and

showed that 968 exhibited an overlapping dose-dependent in-hibition of both 488-GAC and unlabeled GAC activity (Fig. 3C,black closed and open circles, respectively), as well as an over-lapping dose–response for its direct binding to 488-GAC (Fig.3C, green closed circles). We tested the robustness of these high-throughput binding and enzymatic assays by examining a groupof newly synthesized 968 derivatives (compounds SU-1, SU-2,SU-7, SU-8, and SU-14 in Fig. 3D), together with molecules 031,27, and 742 that were previously characterized and shown to beGAC inhibitors (11, 24). A direct correlation exists between theability of different 968 analogs to bind to GAC and to inhibit its

enzymatic activity (Fig. 3E). The results of these analyses, andparticularly the finding that substituting the napthyl group of 968with a quinoline moiety (e.g., compound 27) markedly affectedboth binding and inhibitory activity, suggests that hydrophobicityat this position is required for maximal efficacy.Previous studies of the 968-mediated inhibition of recombinant

GAC activity showed that 968 was much more effective when itwas added before glutamine and inorganic phosphate (the latterbeing an allosteric activator that stimulates GAC tetramer for-mation and GAC activity) compared with when it was added afterthe addition of phosphate (24). Therefore, we examined whetherthe ability of 968 to bind to GAC was compromised under con-ditions where the enzyme was pretreated with inorganic phos-phate and assumed an activated tetrameric state. In fact, wefound that 968 was capable of binding to a tetrameric GACspecies comprised of 488-labeled GAC and QSY9-labeled GACdimers, as read out by the quenching of 488 fluorescence emission(Fig. S4 A and B). Moreover, 968 was able to bind to GAC thathad been preincubated with inorganic phosphate (Fig. S4C,closed vs. open circles). However, under these conditions, 968 wasmuch less effective at inhibiting enzyme activity (Fig. S4D, closedvs. open circles). Thus, phosphate induces an activated state thatis less sensitive to 968 inhibition, even though 968 is able to bindto phosphate-activated GAC. In contrast, when GAC was pre-incubated with 968 before adding phosphate, the enzyme activitywas strongly inhibited and directly correlated with the binding of968 (Fig. S4 C and D, open circles).

968 Preferentially Binds to the Monomeric State of GAC. Dockinganalyses using the X-ray structure of the GAC tetramer, togetherwith mutagenesis studies, suggested that 968 binds in a cove be-tween the monomer-monomer interface (24). To examine theability of 968 to bind to different oligomeric states of GAC, we tookadvantage of recently solved X-ray structures of GAC (16) to designmutants trapped as either monomers or dimers. Fig. 4A depicts theBPTES-binding sites within the GAC tetramer interface andthe proposed 968-binding pocket at the C-terminal region of themonomer-monomer interface. Residue contacts that were mutatedto create constitutive monomeric and dimeric GAC mutants arehighlighted at the GAC-tetramer helical interface (Lower Inset), aswell as at the GAC dimer interface (Upper Inset). When a pointmutation was incorporated at the tetramer interface of mouse GAC(D391K), tetramer formation was disrupted with the resulting GACmutant being trapped in a dimeric state, as determined by multi-angle light scattering (MALS) (Fig. 4B, blue trace). Introducingpoint mutations at the dimer interface of mouse GAC (K316E andR459E), within the background of the dimeric GAC mutant,resulted in a monomeric GAC (D391K, K316E, and R459E) spe-cies (Fig. 4B, green trace). As expected, the monomeric and dimericGAC mutants showed neither a concentration-dependent basalenzymatic activity nor phosphate-stimulated activity (Fig. S5A andB). Although the addition of WT QSY9-labeled GAC to WT 488-labeled GAC resulted in the expected FRET due to tetramer for-mation (Fig. 4C, red trace), the addition of either the QSY9-labeledGAC (D391K) dimer or the GAC (D391K, K316E, and R459E)monomer to WT 488-labeled GAC failed to result in a significantquenching of the 488 donor fluorescence (Fig. 4C, blue and greentraces, respectively). The addition of the dimeric QSY9-GAC(D391K) to WT 488-GAC did induce a minor quenching of the488-GAC emission; however, this was most likely due to the for-mation of mixed donor and acceptor labeled dimers, which resultfrom a relatively minor exchange of the monomeric GAC units.We found that 968 was capable of binding to WT 488-GAC, as

well as to both the dimeric GAC (D391K) and the monomericGAC (D391K, K316E, and R459E), with the monomeric GAChaving the highest affinity for 968 (Fig. 4D). These results sug-gested that 968 should be most effective at inhibiting WT GACat relatively low enzyme concentrations, i.e., where equilibrium

A

B D

C E

Fig. 3. Development of real-time 968 binding and inhibition assays. (A)Schematic model of real-time 968 binding and inhibition assays. Monitoring488-GAC fluorescence quenching serves as a read-out for 968 binding, andenzymatic activity is monitored through the generation of NADH fluores-cence on addition of 20 mM glutamine and 50 mM phosphate to an assayincubation containing labeled GAC together with 10 units of glutamatedehydrogenase (GDH) and 2 mM NAD+. (B) Fluorescence of 10 nM 488-GAC(520-nm emission, green curves) was monitored on addition of 20 μM 968 (–),10 μM BPTES (•••), or DMSO (—) at the indicated time. Simultaneously,NADH fluorescence (460-nm emission, blue curves) was monitored followingthe addition of 20 mM glutamine and 50 mM phosphate at 120 s. (C) Real-time 968 binding and inhibition assays adapted to a 96-well plate formatshow overlapping inhibition and fluorescence quenching profiles for 10 nM488-GAC and 10 nM WT unlabeled GAC. Data points are the average ± SD ofthree independent experiments. The solid line shows the semilog plot of thebinding isotherm with KD = 3 μM. (D) Structures of 968 and 968-like analogsused in real-time binding and inhibition assays. (E) Plotted IC50 (±SD) valuesfrom inhibition data and measured KD (±SD) values from fluorescencequenching data for a representative group of 968 analogs (depicted in D).The compounds a–i correspond to the letter designations shown in D. Valuesobtained from inhibition data and quenching data were fit to a ligandbinding equation for a biomolecular interaction. The line represents a linearregression fit with the following values (R2 = 0.92, slope = 1.10).

Stalnecker et al. PNAS | January 13, 2015 | vol. 112 | no. 2 | 397

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020

conditions favor GAC initially existing as a monomer. Fig. 4Eshows that when the concentration of GAC was decreased from50 to 5 nM, 968 was able to inhibit GAC activity with greaterpotency. Furthermore, the 968-mediated inhibition of GAC ac-tivity at these low enzyme concentrations correlated well with itsinhibition of oncogenic transformation (Fig. 4E).

DiscussionPrevious work from our laboratory aimed at identifying inhib-itors that specifically block Rho GTPase-dependent trans-formation led to the discovery of the benzophenanthridinone 968(11). Unexpectedly, the protein target for 968 appeared to be a

specific splice variant (GAC) of a family of enzymes collectivelycalled glutaminase that catalyzes the hydrolysis of glutamine toglutamate with the production of ammonia. This highlighted apreviously unappreciated connection between the roles of RhoGTPases in driving oncogenic transformation and the regulationof glutamine metabolism. Given the striking specificity that 968exhibited in its ability to inhibit transformed cells and cancer cells,with little or no effect on their normal cellular counterparts, it wasof interest to better understand how 968 functions.We took advantage of an inducible expression system for on-

cogenic Dbl that allowed us to temporally control the expressionof this upstream activator of Rho GTPases in a well-defined

A

B C

D E

Fig. 4. Examination of 968 binding to monomeric and dimeric GAC mutants. (A) Crystal structure of the GAC tetramer (human isoform) in complex with bothBPTES and glutamate (PDB 3UO9), with the proposed 968-binding pocket indicated by the arrow pointing toward the C-terminal monomer-monomer in-terface. Insets highlight critical monomer-monomer (Upper) and dimer-dimer (Lower) contacts, with the corresponding human and mouse GAC isoformresidue numbering. (B) Multiangle light scattering profiles of WT GAC (red), D391K-GAC (blue), and K316E-D391K-R459E-GAC (green), 250 μg (each), wherethe solid line represents the elution of each species by monitoring refractive index (R.I.), and the broken line designates the calculated molecular weight forthe species eluted at that time. Reference lines for the molecular weights of the monomeric, dimeric, and tetrameric forms of the enzyme are included at 58,116, and 232 kDa, respectively. (C) FRET assays on addition of 200 nM WT QSY9-labeled GAC (red), the dimeric QSY9-GAC (D391K) (blue), and monomericQSY9-GAC (K316E, D391K, R459E) (green) to 20 nM WT 488-labeled GAC. (D) 968 binding monitored by its quenching of the fluorescence of WT 488-labeledGAC, dimeric 488-GAC (D391K), and the monomeric GAC (K316E, D391K, and R459E) (10 nM total monomer in each sample). Data points represent the mean(±SD) of three independent experiments and were fit as in Fig. 3C. (E) In vitro inhibition curves of 50 nM (●) and 5 nM WT GAC (○) preincubated withincreasing concentrations of 968. Data points represent the mean (±SD) of three independent experiments and were fit to a logistic four parameter curve.Overlaid is the dose-dependent inhibition by 968 of Dbl-induced focus formation (triangles).

398 | www.pnas.org/cgi/doi/10.1073/pnas.1414056112 Stalnecker et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020

manner. Using this system, we were able to establish a directcorrelation between the ability of 968 to prevent a key outcome ofDbl-induced transformation, namely focus formation, and tospecifically inhibit glutaminolysis. Thus, the inhibitory actions of968 on oncogenic transformation appear to be a direct outcomeof its ability to interfere with glutamine metabolism.We then set out to understand how 968 inhibits the activity of

a key enzyme in glutamine metabolism, GAC. Because BPTES,a well characterized allosteric inhibitor of GAC, has been shownto bind and stabilize an inactive tetrameric form of the enzyme,one possibility was that 968 had a similar effect. Previous studiesused analytical ultracentrifugation, gel filtration, and electron mi-croscopy to investigate the oligomeric transitions of GAC; how-ever, these analyses were performed at GAC concentrations abovethe KD for tetramer formation reported here (15, 19, 20, 25, 26).Thus, we made use of a real-time FRET assay for monitoring GACtetramer formation. The highly sensitive FRET assay enabled us todirectly monitor GAC tetramer formation and show that it corre-lates with enzyme activation, as well as to compare the effects of968 and BPTES on the dimer-to-tetramer transition. We foundthat unlike BPTES, 968 does not stabilize an inactive tetramericstate of GAC. However, during the course of these FRET ex-periments, we discovered that the binding of 968 to GAC resultedin a quenching of the reporter group fluorescence, thus providingus with a direct spectroscopic read out for the ability of this in-hibitor and various analogs to bind to the enzyme.By taking advantage of a direct binding assay for 968, together

with the recent development of GAC mutants that exist asmonomers or dimers, we discovered that 968 has a marked pr-eference for binding to the monomeric form of the enzyme.Although 968 is able to bind, albeit more weakly, to a GAC di-mer, as well as to a GAC tetramer that has been activated by theallosteric regulator inorganic phosphate, it is unable to inhibitthe activity of the activated enzyme tetramer. Therefore, 968preferentially binds to an inactive, monomeric state of GAC andprevents it from undergoing activating conformational changes,whereas, if GAC reaches an activated state before 968 binding,then 968 is unable to inhibit enzyme activity.These findings highlight the distinction between the two classes of

allosteric GAC inhibitors for which BPTES and 968 are the pro-totypes. BPTES is able to bind and inhibit activated GAC, whereas968 binds preferentially to and stabilizes an inactive state of the

enzyme. In addition, these results shed light on the reason forprevious discrepancies when comparing the 968 dose dependenciesfor the inhibition of recombinant GAC activity vs. oncogenictransformation (24). Specifically, in those earlier experiments, theconcentrations of recombinant GAC routinely being assayed rep-resented a mixture of dimers and tetramers. Consequently, the IC50values for 968 reflected its weaker binding to these oligomeric GACspecies. Indeed, when the binding of 968 to GAC, together with itsability to inhibit enzyme activity, is assayed at GAC concentrationswhere it initially exists predominantly as a monomer, the dose–response profiles for these binding assays match the dose-dependent inhibition of transformation in cell culture.In conclusion, we show that 968 functions as a highly specific

inhibitor of oncogenic transformation by blocking a key step inglutamine metabolism necessary for sustaining the transformedstate. We demonstrate that 968 is capable of directly binding toGAC, a key enzyme responsible for elevated glutamine metabolismin transformed cells and cancer cells, and that 968 preferentiallybinds to a monomeric, inactive state of the enzyme. Although anX-ray crystal structure of 968 bound to GAC has not yet beenachieved, these findings provide rationale as to why this has been sochallenging, given that crystallization trials have been routinelyperformed at GAC concentrations where it exists as a tetramer, i.e.,the least favorable species for binding 968 (13, 16–18). Our ability togenerate monomeric GACmutants now provides new opportunitiesfor achieving such a structure. Moreover, the availability of a directbinding read out adapted for plate reader assays offers excitingpossibilities for the identification of 968-like allosteric inhibitors thatcould yield new therapeutic strategies against cancer.

MethodsRecombinant GAC, prepared as described in ref. 11, was labeled with 50 μMof the fluorescent probes and resolved from unreacted probe using a PD10desalting column. Additional details of experimental methods, including cellculture, transformation assays, metabolic tracing experiments, GAC prepa-ration, fluorescence labeling and analyses, enzymatic activity and FRETassays, and synthesis of 968 analogs, are presented in SI Methods.

ACKNOWLEDGMENTS. We thank Cindy Westmiller for excellent secretarialassistance. This study was supported by grants from the National Institutes ofHealth (to R.A.C.) and Award T32GM008500 from the National Institute ofGeneral Medical Sciences (to C.A.S.).

1. DeBerardinis RJ, Cheng T (2010) Q’s next: The diverse functions of glutamine in me-tabolism, cell biology and cancer. Oncogene 29(3):313–324.

2. Wise DR, Thompson CB (2010) Glutamine addiction: A new therapeutic target incancer. Trends Biochem Sci 35(8):427–433.

3. Hensley CT, Wasti AT, DeBerardinis RJ (2013) Glutamine and cancer: Cell biology,physiology, and clinical opportunities. J Clin Invest 123(9):3678–3684.

4. Cheng T, et al. (2011) Pyruvate carboxylase is required for glutamine-independentgrowth of tumor cells. Proc Natl Acad Sci USA 108(21):8674–8679.

5. Le A, et al. (2012) Glucose-independent glutamine metabolism via TCA cycling forproliferation and survival in B cells. Cell Metab 15(1):110–121.

6. Seltzer MJ, et al. (2010) Inhibition of glutaminase preferentially slows growth ofglioma cells with mutant IDH1. Cancer Res 70(22):8981–8987.

7. Metallo CM, et al. (2012) Reductive glutamine metabolism by IDH1 mediates lipo-genesis under hypoxia. Nature 481(7381):380–384.

8. Son J, et al. (2013) Glutamine supports pancreatic cancer growth through a KRAS-regulated metabolic pathway. Nature 496(7443):101–105.

9. van den Heuvel AP, Jing J, Wooster RF, Bachman KE (2012) Analysis of glutaminedependency in non-small cell lung cancer: GLS1 splice variant GAC is essential forcancer cell growth. Cancer Biol Ther 13(12):1185–1194.

10. Lin R, Cerione RA, Manor D (1999) Specific contributions of the small GTPases Rho,Rac, and Cdc42 to Dbl transformation. J Biol Chem 274(33):23633–23641.

11. Wang JB, et al. (2010) Targeting mitochondrial glutaminase activity inhibits onco-genic transformation. Cancer Cell 18(3):207–219.

12. Shapiro RA, Clark VM, Curthoys NP (1979) Inactivation of rat renal phosphate-dependent glutaminase with 6-diazo-5-oxo-L-norleucine. Evidence for interaction atthe glutamine binding site. J Biol Chem 254(8):2835–2838.

13. Brown G, et al. (2008) Functional and structural characterization of four glutaminasesfrom Escherichia coli and Bacillus subtilis. Biochemistry 47(21):5724–5735.

14. Ahluwalia GS, Grem JL, Hao Z, Cooney DA (1990) Metabolism and action of aminoacid analog anti-cancer agents. Pharmacol Ther 46(2):243–271.

15. Robinson MM, et al. (2007) Novel mechanism of inhibition of rat kidney-type gluta-

minase by bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl)ethyl sulfide (BPTES). Bio-

chem J 406(3):407–414.16. DeLaBarre B, et al. (2011) Full-length human glutaminase in complex with an allo-

steric inhibitor. Biochemistry 50(50):10764–10770.17. Thangavelu K, et al. (2012) 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. Proc Natl Acad Sci USA 109(20):7705–7710.18. Cassago A, et al. (2012) Mitochondrial localization and structure-based phosphate

activation mechanism of Glutaminase C with implications for cancer metabolism. Proc

Natl Acad Sci USA 109(4):1092–1097.19. Godfrey S, Kuhlenschmidt T, Curthoys P (1977) Correlation between activation and dimer

formation of rat renal phosphate-dependent glutaminase. J Biol Chem 252(6):1927–1931.20. Kenny J, et al. (2003) Bacterial expression, purification, and characterization of rat

kidney-type mitochondrial glutaminase. Protein Expr Purif 31(1):140–148.21. Rossman KL, Der CJ, Sondek J (2005) GEF means go: Turning on RHO GTPases with

guanine nucleotide-exchange factors. Nat Rev Mol Cell Biol 6(2):167–180.22. Etienne-Manneville S, Hall A (2002) Rho GTPases in cell biology. Nature 420(6916):629–635.23. Olivo C, et al. (2000) Distinct involvement of cdc42 and RhoA GTPases in actin orga-

nization and cell shape in untransformed and Dbl oncogene transformed NIH3T3

cells. Oncogene 19(11):1428–1436.24. Katt WP, Ramachandran S, Erickson JW, Cerione RA (2012) Dibenzophenanthridines as

inhibitors of glutaminase C and cancer cell proliferation.Mol Cancer Ther 11(6):1269–1278.25. Ferreira AP, et al. (2013) Active glutaminase C self-assembles into a supratetrameric olig-

omer that can be disrupted by an allosteric inhibitor. J Biol Chem 288(39):28009–28020.26. Møller M, et al. (2013) Small angle X-ray scattering studies of mitochondrial gluta-

minase C reveal extended flexible regions, and link oligomeric state with enzyme

activity. PLoS ONE 8(9):e74783.

Stalnecker et al. PNAS | January 13, 2015 | vol. 112 | no. 2 | 399

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

23, 2

020