Oncogenic PIK3CA mutations reprogram glutamine metabolism in colorectal cancer

Hao, Yujun; Samuels, Yardena; Li, Qingling; Krokowski, Dawid; Guan, Bo-Jhih; Wang, Chao; Jin, Zhicheng; Dong, Bohan; Cao, Bo; Feng, Xiujing; Xiang, Min; Xu, Claire; Fink, Stephen; Meropol, Neal J.; Xu, Yan; Conlon, Ronald A.; Markowitz, Sanford; Kinzler, Kenneth W.; Velculescu, Victor E.; Brunengraber, Henri; Willis, Joseph E.; LaFramboise, Thomas; Hatzoglou, Maria; Zhang, Guo-Fang; Vogelstein, Bert; Wang, Zhenghe

doi:10.1038/ncomms11971

Download PDF

Article
Open access
Published: 20 June 2016

Oncogenic PIK3CA mutations reprogram glutamine metabolism in colorectal cancer

Yujun Hao^1,2,
Yardena Samuels^3,4^na1,
Qingling Li⁵^na1,
Dawid Krokowski⁶^na1,
Bo-Jhih Guan⁶^na1,
Chao Wang^1,2^na1^nAff15,
Zhicheng Jin⁵,
Bohan Dong^1,7,
Bo Cao^1,8,
Xiujing Feng^1,2,
Min Xiang^1,9,
Claire Xu¹⁰,
Stephen Fink^2,11,
Neal J. Meropol^2,11,
Yan Xu¹²,
Ronald A. Conlon^1,2,
Sanford Markowitz^2,11,
Kenneth W. Kinzler³,
Victor E. Velculescu³,
Henri Brunengraber⁵,
Joseph E. Willis¹³,
Thomas LaFramboise^1,2,
Maria Hatzoglou⁶,
Guo-Fang Zhang⁵,
Bert Vogelstein³ &
…
Zhenghe Wang^1,2

Nature Communications volume 7, Article number: 11971 (2016) Cite this article

16k Accesses
161 Citations
121 Altmetric
Metrics details

Subjects

Abstract

Cancer cells often require glutamine for growth, thereby distinguishing them from most normal cells. Here we show that PIK3CA mutations reprogram glutamine metabolism by upregulating glutamate pyruvate transaminase 2 (GPT2) in colorectal cancer (CRC) cells, making them more dependent on glutamine. Compared with isogenic wild-type (WT) cells, PIK3CA mutant CRCs convert substantially more glutamine to α-ketoglutarate to replenish the tricarboxylic acid cycle and generate ATP. Mutant p110α upregulates GPT2 gene expression through an AKT-independent, PDK1–RSK2–ATF4 signalling axis. Moreover, aminooxyacetate, which inhibits the enzymatic activity of aminotransferases including GPT2, suppresses xenograft tumour growth of CRCs with PIK3CA mutations, but not with WT PIK3CA. Together, these data establish oncogenic PIK3CA mutations as a cause of glutamine dependency in CRCs and suggest that targeting glutamine metabolism may be an effective approach to treat CRC patients harbouring PIK3CA mutations.

A CRISPRi/a screening platform to study cellular nutrient transport in diverse microenvironments

Article Open access 11 April 2024

Targeting p53 pathways: mechanisms, structures, and advances in therapy

Article Open access 01 March 2023

Cancer-associated fibroblast-derived acetate promotes pancreatic cancer development by altering polyamine metabolism via the ACSS2–SP1–SAT1 axis

Article Open access 01 March 2024

Introduction

Cancer cells are distinguished from most normal cells by metabolic reprogramming, including phenomena termed the Warburg effect and glutamine dependency^1,2. Normally, glucose is converted to acetyl-CoA, which enters the tricarboxylic acid (TCA) cycle. Cancer cells, however, convert glucose to lactate even in the presence of oxygen (Warburg effect) and utilize glutamine to replenish the TCA cycle³. To enter the TCA cycle, glutamine is first deaminated by glutaminases (GLSs) to glutamate⁴. Glutamate is then converted to α-ketoglutarate (α-KG), which is a substrate in the TCA cycle. Three groups of enzymes can convert glutamate to α-KG: (1) glutamate pyruvate transaminases (GPTs); (2) glutamate oxaloacetate transaminases (GOTs); and (3) glutamate dehydrogenases (GLUDs)⁴. The metabolic products of glutamine are utilized both to produce ATP and to synthesize macromolecules in the promotion of tumour growth⁴. Although glutamine is a non-essential amino acid, it has long been recognized that glutamine is a required supplement for culturing cancer cells.

Many oncogenes and tumour suppressors impact glutamine metabolism⁴. Myc overexpression affects cellular glutamine levels by inducing the transcription of GLS1 and the glutamine transporter SLC1A5 (a.k.a. ASCT2)^5,6. In contrast, SLC1A5 expression is repressed by the Rb tumour suppressor⁷, whereas GLS2 was identified as a transcriptional target of p53 (ref. 8). In addition, it has been shown that p53 represses the expression of malic enzymes ME1 and ME2, thereby regulating glutamine-dependent NADPH production⁹. A recent study showed that loss of tumour suppressor von hippel-lindau tumor suppressor (VHL) renders renal cell carcinomas sensitive to glutamine deprivation through hypoxia induced factor (HIF)-induced metabolic reprogramming¹⁰. Moreover, K-ras upregulates the aminotransferase GOT1 (ref. 11). Though all of these mechanisms impact the production or degradation of glutamine or its metabolites, the mechanisms by which many cancer cells become dependent on glutamine are still unknown or actively debated.

PIK3CA encodes the catalytic subunit of phosphatidylinositol 3-kinase α (PI3Kα), which plays a key role in regulating cell proliferation, survival and motility¹². PIK3α consists of a catalytic subunit p110α, and one of several regulatory subunits (a major one being p85α)¹³. On growth factor stimulation, p85 is recruited to phosphorylated receptor protein kinases and adaptor proteins, thereby activating PI3Kα. Activated PI3Kα converts phosphatidylinositol-4,5-biophosphate (PIP2) to phosphatidylinositol-3,4,5-triphosphate (PIP3). The second message PIP3 then activates PDK1 and AKT signalling. PIK3CA is mutated in a wide variety of human cancers including ∼30% of colorectal cancers (CRCs)¹⁴. Recent large-scale sequencing of human cancer genomes reveals that PIK3CA is the most frequently mutated oncogene in human cancer¹⁵. However, the fact that PIK3CA mutations can reprogram cancer metabolism, as demonstrated herein, was previously unknown.

We report that PIK3CA mutations render CRCs more sensitive to glutamine deprivation by upregulation of GPT2, an enzyme involved in glutamine metabolism. We further demonstrate that mutant p110α increases GPT2 gene expression through an AKT-independent signalling pathway. Moreover, we show that aminooxyacetate (AOA), a compound that inhibits enzymatic activity of aminotransferases, suppresses xenograft tumour growth of CRCs with PIK3CA mutations, but not with wild-type (WT) PIK3CA. These results suggest that reprogramming glutamine metabolism is crucial for the oncogenic function of PIK3CA mutations and that targeting glutamine metabolism may be an effective approach to treating CRC patients harbouring tumour mutations of this gene.

Results

PIK3CA mutations render CRC cells dependent on glutamine

Most PIK3CA mutations are clustered in two hotspots, with H1047R in the kinase domain and E545K in the helical domain the most common mutations¹⁶. We set out to determine whether PIK3CA mutations reprogram cell metabolism in CRCs. The CRC cell line HCT116 harbours a heterozygous H1047R mutation, whereas DLD1 CRC cells have a heterozygous E545K mutation (Fig. 1a). We exploited isogenic derivatives of these cell lines in which either the WT or mutant allele of PIK3CA is knocked out (Fig. 1a)¹⁷. The clones in which the mutant allele had been disrupted (and the WT allele was intact) were called ‘WT’ (Fig. 1a), whereas the clones in which only the WT allele had been disrupted (and the mutant allele was intact) were called ‘mutant’ (Mut, Fig. 1a)¹⁷. As reported previously¹⁷, the parental cells and their derived knockout clones grew at similar rate under normal conditions in the presence of both glucose and glutamine (Fig. 1b). However, in medium without glutamine, both parental cells and the two independently derived PIK3CA mutant clones died faster than did two independently derived PIK3CA WT clones (Fig. 1b). This relative sensitivity to glutamine deprivation dependent on PIK3CA status was observed in both HCT116 and DLD1 (Fig. 1b). Consistently, glutamine deprivation induced more apoptotic cells in the mutant clones than in the WT clones as assayed by percentages of sub-G1 cells and amounts of cleaved PARP (Fig. 1c,d). In contrast, differential sensitivity to deprivation of glucose was not found (Fig. 1b and Supplementary Fig. 1a). Surprisingly, these cell lines still grew equally well, although slowly, in the absence of glucose (Fig. 1b). Because recent studies showed that lung cancer cells can become independent of glucose by upregulation of mitochondrial phosphoenolpyruvate carboxykinase (PCK2), which generates phosphoenolpyruvate from glutamine to generate ribose for nucleic acid synthesis^18,19, we set out to determine whether PCK2 was upregulated in the HCT116 and DLD1 cell lines when deprived of glucose. As shown in Supplementary Fig. 1b, PCK2 was indeed upregulated in both cell lines.

**Figure 1: *PIK3CA* mutation renders CRC cells more sensitive to glutamine deprivation.**

To determine if metabolic reprogramming by PIK3CA mutation was generalizable, we tested glutamine sensitivity in two CRC cell lines with PIK3CA mutations (RKO (containing a PIK3CA H1047R mutation) and HT29 (containing a PIK3CA P449T mutation)) and two CRC cell lines with WT PIK3CA (SW480 and LOVO). As shown in Fig. 1e, glutamine deprivation induced significantly more apoptotic cells in the PIK3CA mutant cell lines than in the WT PIK3CA cell lines.

Since KRAS increases glutamine metabolism in pancreatic cancers¹¹, we set out to determine if KRAS mutations affects glutamine dependency in CRC cells. HCT116 and DLD1 cells have heterozygous oncogenic KRAS mutations. Using isogenic HCT116 and DLD1 clones, with either WT or mutant KRAS knockout²⁰, we showed that the cells did not have differential sensitivity to glutamine dependent on KRAS genotype (Supplementary Fig. 1c,d). Taken together, the data suggest that PIK3CA mutations may be responsible for CRC cell dependence on glutamine for optimal growth.

PIK3CA mutations upregulates GPT2 expression

To determine whether PIK3CA mutations regulate enzymes involved in glutamine metabolism, we performed serial analysis of gene expression (SAGE)²¹ on the isogenic cell lines described above (Supplementary Data 1). Interestingly, the expression levels of mitochondrial GPT2, which converts glutamate to α-KG, were upregulated in both HCT116 and DLD1 PIK3CA mutant clones compared with the WT clones. This observation was confirmed by both reverse transcription–PCR (RT–PCR) and western blot analyses of the clones (Fig. 2a,b and Supplementary Fig. 2a). None of the other enzymes primarily involved in glutamine metabolism, including GLS, GPT1, GOT1, GOT2, GLUD or the glutamine transporter SLC1A5, exhibited differential protein levels among the PIK3CA mutant and WT clones (Fig. 2b). Moreover, both GPT2 RNA and protein levels were higher in the four different PIK3CA mutant lines (HCT116, DLD1, RKO and HT29) than in the two PIK3CA WT cell lines (SW480 and LOVO) (Fig. 2c,d and Supplementary Fig. 2b). In agreement with the notion that KRAS mutations do not play a role in glutamine dependency in CRC, none of the enzymes showed differential expression levels in the KRAS WT and mutant clones (Supplementary Fig. 2c).

**Figure 2: *PIK3CA* mutation upregulates GPT2 which renders CRC dependent on glutamine.**

We next tested if GPT2 expression is upregulated in CRC tumour specimens with PIK3CA mutations. We measured GPT2 RNA levels by quantitative RT–PCR (qRT–PCR) in 20 human CRC patient tumours, which we had previously sequenced the exomes (10 tumours with PIK3CA mutations and 10 tumours without mutations in the PIK3CA pathway)²². As shown in Fig. 2e, expression levels of GPT2 were significantly higher in the tumours with PIK3CA mutation than in the tumours with WT PIK3CA. To further extend this observation, we mined the TCGA data set of CRCs with RNA-seq data. As shown in Supplementary Fig. 2d, CRCs with PIK3CA mutation express significantly higher levels of GPT2 than CRCs with WT PIK3CA. Moreover, for those tumours with matched normal tissues, GPT2 expression is upregulated in nearly all of them (Supplementary Fig. 2e).

To determine whether the upregulation of GPT2 makes PIK3CA mutant cells dependent on glutamine, we knocked down GPT2 in the HCT116 PIK3CA mutant clone using two independent shRNAs (Fig. 2f). Because independently derived PIK3CA mutant and WT clones exhibited near-identical phenotypes, we chose one PIK3CA mutant clone (Mut 1) and one PIK3CA WT clone (WT 1) for in-depth analyses. Compared with cells with a control shRNA, the GPT2 knockdown cells grew more slowly under normal growth conditions (Fig. 2g), but they were less sensitive to glutamine deprivation as assayed by cell apoptosis (Fig. 2h). These results were further validated with GPT2 CRISPR/Cas9 knockouts in HCT116 cells (Supplementary Fig. 2f–i). Although GPT2 knockout cells grew slower than the parental cells, they were less sensitive to glutamine deprivation (Supplementary Fig. 2h,i). Moreover, reconstitution of the knockout cells by restoration of GPT2 expression made these cells grow faster and more sensitive to glutamine deprivation (Supplementary Fig. 2j–l). In contrast, knockdown of GPT2 in the HCT116 PIK3CA WT clone (WT 1) had no effect on their sensitivity to glutamine deprivation or proliferation under normal culture conditions (Supplementary Fig. 2m–o). Importantly, overexpression of GPT2 in the WT clone made it grow faster under normal culture conditions, and more sensitive to glutamine deprivation (Fig. 2i–k and Supplementary Fig. 2p). However, overexpression of an enzymatic inactive mutant GPT2 in the WT clone had no impact on either cell proliferation or sensitivity to glutamine deprivation (Fig. 2i–k). In aggregate, these data demonstrate that PIK3CA mutations render CRC cells more sensitive to glutamine withdrawal through upregulation of GPT2.

AOA inhibits the growth of PIK3CA-mutant CRCs in vivo

AOA inhibits the enzymatic activity of aminotransferases, including GPT2 (ref. 23). As shown in Supplementary Fig. 3a, both HCT116 and DLD1 PIK3CA mutant clones were more sensitive to AOA treatment than the WT clones in tissue culture. Moreover, AOA significantly inhibited the growth of HCT116 and DLD1 PIK3CA mutant clones when xenografted into nude mice (Fig. 3a and Supplementary Fig. 3b). In contrast, AOA had no effect on isogenic PIK3CA WT xenograft tumours (Fig. 3a and Supplementary Fig. 3b), although those tumours grew more slowly than their mutant counterparts in the absence of AOA (Fig. 3a and Supplementary Fig. 3b). Importantly, we further demonstrated that AOA inhibited growth of PIK3CA mutant xenograft tumours by targeting GPT2, because tumours established from the two GPT2 knockdown cell pools were insensitive to AOA treatment, although they grew much slower than HCT116 PIK3CA mutant tumours expressing a control shRNA (Fig. 3b). AOA treatment consistently reduced amounts of α-KG and alanine, but increased amounts of glutamate and pyruvate, in the HCT116 PIK3CA mutant tumours (Supplementary Fig. 3c).

**Figure 3: AOA inhibits xenograft tumour growth of *PIK3CA* mutant CRCs but not *PIK3CA* WT CRCs.**

To test whether our observations with the genetically engineered cell lines were generalizable, we expanded our xenograft study to a panel of CRC cell lines. AOA inhibited xenograft tumour growth of four PIK3CA mutant CRC cell lines (HCT116 (parental cells), DLD1 (parental cells), RKO and HT29; Fig. 3c). In contrast, AOA had no effect on xenograft tumour growth of two WT PIK3CA CRC cell lines (SW480 and LOVO; Fig. 3d). No weight loss was observed for the mice that were treated with AOA (Supplementary Fig. 3d), suggesting that the doses of AOA used in the experiments had minimal toxicity.

ATF4 regulates transcription of GPT2

Given that p110α is not a transcription factor, a key question raised by these data is how mutant p110α transduces the signals that activate GPT2 transcription. To address this question, we evaluated ATF4, as recent studies reported that ATF4 is involved in glutamine metabolism^6,24. Interestingly, ATF4 protein levels mirrored GPT2 protein levels in the PIK3CA mutant and WT clones (Fig. 4a). This correlation was maintained in xenograft tumours (Fig. 4b). Because ATF4 induces expression of pro-apoptotic BH3-only proteins PUMA and Noxa in neuroblastoma cells⁶, we also measured levels of these two proteins in the PIK3CA WT and mutant clones. However, PUMA and Noxa protein levels were not correlated with ATF4 levels (Fig. 4a), suggesting that the regulation of the two pro-apoptotic proteins by ATF4 may be cell-type specific or controlled by other factors.

**Figure 4: ATF4 activates GPT2 transcription in *PIK3CA* mutant cells.**

As shown in Fig. 4c, overexpression of ATF4 in the HTC116 PIK3CA WT clone increased GPT2 protein levels in a dose-dependent manner. Conversely, knockdown of ATF4 in the HCT116 PIK3CA mutant clone by two different siRNAs reduced both mRNA and protein levels of GPT2 (Fig. 4d). In contrast, knockdown of ATF4 did not affect the expression of other enzymes involved in glutamine metabolism, including SLC1A5, GLS1, GOT1, GOT2 and GLUD1 (Supplementary Fig. 4a). Similar results were observed in both HCT116 and DLD1 PIK3CA mutant clones (Fig. 4d and Supplementary Fig. 4a,b). Importantly, as in the case with GPT2, knockdown of ATF4 rendered PIK3CA mutant cells less proliferative under normal culture conditions, but more resistant to glutamine deprivation (Fig. 4e,f). Conversely, overexpression of a proteasome-resistant ATF4 S219A mutant in HCT116 WT cells increased GPT2 protein levels and cell proliferation under normal culture conditions (Fig. 4g,h), but made them more sensitivity to glutamine deprivation (Fig. 4i). These results suggest that the mutant p110α–ATF4–GPT2 axis regulates glutamine metabolism.

To determine whether ATF4 activates GPT2 gene transcription directly, we cloned 1.5 kb of genomic DNA upstream of the transcription start site of the GPT2 gene into a luciferase reporter plasmid. As shown in Fig. 4j, knockdown of ATF4 in the HCT116 PIK3CA mutant clone reduced transcriptional activity of the GPT2 reporter. Examining the DNA sequence in the 1.5-kb genomic DNA fragment of GPT2, we found two motifs matching ATF4 consensus-binding sites (Fig. 4k). Mutating either site individually or both together significantly diminished ATF4-mediated transcriptional activity (Fig. 4l).

Mutant p110α stabilizes the ATF4 protein

We next sought to determine how ATF4 is differentially regulated in PIK3CA mutant and WT cells. We first showed that ATF4 mRNA levels were similar in isogenic PIK3CA mutant and WT cell clones (Supplementary Fig. 4c). Translation of ATF4 mRNA is known to be induced by stress through phospho-eIF2α (p-eIF2α)-dependent translation initiation of upstream open reading frames (uORF)²⁵. To investigate whether mutant p110α upregulates ATF4 protein levels through this mechanism, we examined p-eIF2α levels in the PIK3CA mutant and WT clones. As shown in Supplementary Fig. 4d, similar levels of p-eIF2α were observed in the PIK3CA mutant and WT clones under normal culture conditions. This result was in agreement with the uORF reporter assays indicating that the uORF initiation activity of ATF4 was similar in the HCT116 PIK3CA mutant and WT cells (Supplementary Fig. 4e). Moreover, polysome profiles of ATF4 mRNA were similar in HCT116 PIK3CA mutant and WT clones (Supplementary Fig. 4f). We also tested whether mutant p110α affected ATF4 protein stability by regulating its ubiquitination. As shown in Fig. 5a,b, the ubiquitination levels of both endogenous and overexpressed ATF4 were much higher in the HCT116 PIK3CA WT clone than in the mutant cells. Consistent with this result, treatment of HCT116 and DLD1 PIK3CA WT cells with a proteasome inhibitor MG132 resulted in increased protein levels of ATF4 and GPT2 (Supplementary Fig. 4g). MG132 treatment also increased apoptosis in these cells in the presence or absence of glutamine (Supplementary Fig. 4h)—presumably because of effects on other pathways—preventing further investigation of MG132 treatments.

**Figure 5: The p110α–PDK1–RSK2 signalling axis regulates ATF4 protein stability.**

As expected, exposure to a pan PI3K inhibitor (LY294002), a p110α-specific inhibitor (BYL-719), a PI3K/mTOR dual inhibitor (BEZ235) and a PDK1 inhibitor (GSK2334470), each reduced ATF4 protein levels (Fig. 5c,d). Surprisingly, neither an AKT inhibitor (GSK690693) nor an inhibitor (CHIR-99021) of GSK3β (a known downstream effector of AKT) had any effect on ATF4 protein levels (Fig. 5c). In accord with this, expression of a constitutively active form of AKT1 (myristoylated AKT1) in the PIK3CA WT clone did not affect ATF4 or GPT2 proteins levels (Supplementary Fig. 5a). As a control for these experiments, we found that myristoylated AKT1 did increase the phosphorylation level of AKT and FOXO1 (Supplementary Fig. 5a), a well-known AKT kinase target. Moreover, overexpression of myristoylated AKT1 did not make HCT116 PIK3CA WT cells more sensitive to glutamine deprivation (Supplementary Fig. 5b,c). Although mTOR was activated in PIK3CA mutant cells when deprived of glutamine (Supplementary Fig. 5d), our data suggest that the activation of mTOR does not determine glutamine sensitivity in these cells, because the overexpression of myristoylated AKT1 increased mTOR activity in the PIK3CA WT in the absence of glutamine (Supplementary Fig. 5a).

Interestingly, ATF4 is reported to be a substrate of RSK2 (ref. 26). Although it is not as well-known an effector of PI3K as AKT, PDK1 also regulates the RSK2 kinase²⁷. Indeed, both a pan RSK inhibitor (BI-D1870) and a more selective RSK2 inhibitor (FMK) reduced ATF4 protein levels in the HCT116 PIK3CA mutant clone (Fig. 5c). In contrast, the inhibitors of p110α, PDK1 and RSK2 had little or no effect on ATF4 protein level in the HCT116 PIK3CA WT clone (Supplementary Fig. 4i), suggesting that the activity of the p110α–PDK1–RSK2–ATF4 pathway is low in these cells. Together, our data suggest a mutant p110α-activated PDK1–RSK2 pathway that regulates ATF4 protein levels (Fig. 5e, indicated by the red arrows).

To confirm the results obtained with the inhibitors, we first overexpressed p110α E545K and H1047R mutant constructs in the HCT116 PIK3CA WT clone. We found that overexpression of mutant p110α increased both ATF4 and GPT2 protein levels (Fig. 5f). We then attempted to assess the generality of these observations and ascertain whether the lipid kinase activity of p110α is required for the mutant p110α signalling pathway to stabilize the ATF4 protein. For this purpose, we knocked in a D933A mutation that inactivates its lipid kinase activity on top of the E545K mutant allele into the DLD1 PIK3CA mutant clone (Supplementary Fig. 5e,f). Figure 5g shows that the protein levels of both ATF4 and GPT2 were reduced in the double-mutant clones. As expected, AKT phosphorylation levels were also reduced in the double-mutant cells (Supplementary Fig. 5g). Importantly, the kinase inactivation mutation made the DLD1 PIK3CA E545K mutant clone less sensitive to glutamine deprivation (Fig. 5h). Moreover, knockdown of either PDK1 or RSK2 in both HCT116 and DLD1 PIK3CA mutant clones reduced ATF4 and GPT2 protein levels (Fig. 5i,j).

Phosphorylation of ATF4 by RSK2 protects it from degradation

RSK2 is a serine/threonine kinase that phosphorylates ATF4 at the serine 245 residue (S245)²⁶. Knockdown of RSK2 in HCT116 PIK3CA mutant clone reduced levels of pS245 ATF4 (Fig. 6a). Therefore, we hypothesized that phosphorylation of ATF4 at S245 by the mutant p110α–PDK1–RSK2 signalling axis stabilizes ATF4. To test this hypothesis, we first examined ATF4 S245 phosphorylation in the HCT116 PIK3CA WT and mutant clones. As expected, levels of pS245 ATF4 were higher in the mutant clone than in the WT clone (Fig. 6b). Second, compared with the expression of a WT ATF4 construct, the expression of an unphosphorylatable ATF4 S245A mutant construct in the HCT116 PIK3CA mutant clone resulted in less ATF4 protein expression (Fig. 6c). In contrast, the ATF4 S219A mutant that abolishes the binding of ATF4 to β-TrCP1, an ubiquitin E3 ligase of ATF4 (ref. 24), generated more protein than the WT ATF4 (Fig. 6c). In agreement with our hypothesis that phosphorylation of ATF4 at the S245 residue stabilizes it, the ubiquitination levels of the ATF4 S245A mutant were higher than that of the WT protein (Fig. 6d). These data led us to postulate that phosphorylation of ATF4 S245 by RSK2 either reduces its binding affinity to an ubiquitin E3 ligase or recruits a deubiquitinase, thereby protecting ATF4 from degradation. To this end, we first tested the binding of WT and S245A mutant ATF4 to β-TrCP1. However, both WT and the mutant ATF4 bound to a similar amount of β-TrCP1 (Supplementary Fig. 6a). We then turned our attention to deubiquitinases and tested the binding of the WT ATF4 and S245A mutant to eight ubiquitin specific peptidases (USPs) (USP1, USP2, USP7, USP8, USP9X, USP10, USP14 and USP18). Among the USPs tested, only USP8 exhibited differential binding to WT versus mutant ATF4 (Fig. 6e and Supplementary Fig. 6b). In agreement with our hypothesis, the ATF4 S245A mutant bound to less USP8 than WT ATF4 (Fig. 6e). Knockdown of USP8 by two independent siRNAs in the HCT116 PIK3CA mutant clone resulted in reduced ATF4 protein levels (Fig. 6f) and increased ATF4 ubiquitination (Fig. 6g). However, knockdown of USP8 failed to make the PIK3CA mutant cells less sensitive to glutamine deprivation (Supplementary Fig. 6c), which might be due to additional roles of USP8 in regulating anti-apoptotic protein(s). Consistently, knockdown of USP8 significantly increased apoptosis of PIK3CA mutant cells even in the presence of glutamine (Supplementary Fig. 6c).

**Figure 6: Phosphorylation of ATF4 S245 by RSK2 enhances its binding to USP8 and protects ATF4 from ubiquitin-mediated degradation.**

TCA cycle metabolites are higher in PIK3CA mutant clones

As shown in Fig. 7a and Supplementary Fig. 7a, both the HCT116 and DLD1 mutant clones consumed more glutamine than their WT counter parts. Glutamine is converted to α-KG to replenish the TCA cycle (Fig. 7b). Because GPT2, an enzyme that converts glutamate to α-KG, is upregulated in PIK3CA mutant CRC cells, we measured glutamine flux in the paired isogenic lines by tracing [¹³C₅]-glutamine over time. As shown in Fig. 7c, the flux of glutamine-derived TCA cycle intermediates including α-KG, succinate, fumarate and citrate was significant higher in the HCT116 PIK3CA mutant clone than in the WT clone at least at the early time points. Similar results were observed with the isogenic clones derived from DLD1 (Supplementary Fig. 7b). Although the flux of glutamine-derived glutamate was higher in DLD1 PIK3CA mutant cells than in the WT cells, the flux of glutamine-derived glutamate was similar in the HCT116 isogenic pair (Fig. 7c and Supplementary Fig. 7b). Nevertheless, the ratios of α-KG/glutamate were higher in the PIK3CA mutant clones than in the WT counterparts (Supplementary Fig. 7c). Similarly, compared with the isogenic PIK3CA WT clones, the pools of alanine and ratios of alanine/pyruvate were higher in the mutant counterparts (Fig. 7d and Supplementary Fig. 7d,e). These results support the premise that upregulation of GPT2 by oncogenic PIK3CA mutations leads to conversion of more α-KG from glutamine to replenish the TCA cycle. Consistently, knockdown of GPT2 in HCT116 PIK3CA mutant cells reduced levels of α-KG and TCA intermediates (for example, fumarate) that were derived from glutamine (Supplementary Fig. 7f,g). Conversely, overexpression of WT GPT2, but not enzymatically inactive mutant GPT2, increased glutamine-derived α-KG and TCA intermediates (Supplementary Fig. 7h,i).

**Figure 7: Mutation of *PIK3CA* increases glutamine consumption and conversion to α-KG.**

A major product of the forward TCA cycle is NADH, which couples with oxidative phosphorylation to generate ATP. We therefore measured the amounts of ATP and NADH in the PIK3CA WT and mutant clones. In the presence of glutamine, the amounts of ATP and NADH were significantly higher in both the HCT116 and DLD1 mutant clones than in their WT counterparts (Fig. 7e,f and Supplementary Fig. 7j,k). The ratios of NADH/NAD were also significantly higher in the mutant clones than in the WT clones (Supplementary Fig. 7m). Although not statistically significant, the ATP/ADP ratios trended higher in the mutant clones (Supplementary Fig. 7l). However, under glutamine deprivation, the amount of ATP and the ATP/ADP ratio were significantly lower in the mutant clones than in the WT clones (Fig. 7e and Supplementary Fig. 7j,l), whereas the amounts of NADH and the ratios of NADH/NAD were similar in the mutant and WT clones (Fig. 7f and Supplementary Fig. 7k,m).

α-KG rescues survival of Gln-deprived PIK3CA mutant cells

The results described above suggested that the generation of α-KG from glutamine to replenish the TCA cycle was critical to the survival of the PIK3CA mutant cells. To test this suggestion, we deprived the HCT116 PIK3CA mutant cell of glutamine and then supplemented the cells with 4 mM dimethyl-α-KG, a cell permeable form of α-KG. When deprived of glutamine, <8% cells survived after 3 days (Fig. 7g). In contrast, the addition of dimethyl-α-KG increased cell survival to ∼60%.

Discussion

Our data demonstrate that oncogenic PIK3CA mutations reprogram glutamine metabolism through the upregulation of GPT2 in CRCs. Although a previous study by Weinberg and colleagues suggests that GPT2 is involved in oncogenic KRAS mutation-mediated glutamine metabolism in HCT116 cells, they did not investigate how K-ras regulates GPT2 (ref. 28). Importantly, using isogenic cell lines with either mutant allele or WT allele knockout, we clearly demonstrated that KRAS mutation does not make HCT116 and DLD1 CRC cells more dependent on glutamine. Moreover, both SW480 and LOVO CRC cell lines harbour oncogenic KRAS mutations, but the two cell lines were resistant to glutamine deprivation. Thus, our data suggest that mutant KRAS is not a key determinant of glutamine dependency in CRCs. Consistently, knock-in of mutant K-ras in mouse and a human cell line actually reduces or does not affect PI3K/AKT activities^29,30, although WT K-ras interacts and activates PI3K (ref. 31). In contrast, our data provide compelling evidence that oncogenic PIK3CA mutations in CRCs render them more sensitive to glutamine deprivation. Given that several studies show that PI3K/AKT signalling regulates glucose metabolism^32,33, it is surprising to observe that HCT116 and DLD1 PIK3CA mutant and WT knockout clones could grow without glucose. Our data could be interpreted that upregulation of PCK2 in HCT116 and DLD1 cells deprived of glucose attenuates glucose dependency of these cells.

Our data also show that CRC cells harbouring PIK3CA mutations, but not those cells with WT PIK3CA, are sensitive to growth inhibition by AOA, a compound that blocks the conversion of glutamate to α-KG. These findings constitute a proof of principle that targeting glutamine metabolism could be a useful approach to treating CRCs harbouring PIK3CA mutations. Since the first PIK3CA mutations in human cancers were discovered¹⁴, huge efforts have been made to develop small molecule inhibitors of p110α (ref. 34). However, because the catalytic domains of class I lipid kinases, p110α, p110β, p110γ and p110δ, are highly conserved, it is difficult to design small molecules that specifically inhibit the p110α isoform³⁵, and pan-PI3K inhibitors are highly toxic. Moreover, the p110α inhibitors developed to date do not discriminate between mutant and WT p110α proteins. A recent study showed that cancer cells quickly develop resistance to a p110α-specific inhibitor BYL719 (ref. 36). Our study suggests that targeting glutamine metabolism may provide a specific form of therapy for cancer patients harbouring PIK3CA mutations.

The findings reported here demonstrate that GPT2 is the key determinant of glutamine sensitivity in PIK3CA mutant CRC cells. GPT2 is an aminotransaminase that converts glutamate to α-KG, which is a TCA cycle intermediate. Metabolic profiling shows that amounts of α-KG are significantly higher in the PIK3CA mutant clones than in the WT clones and that the other TCA cycle intermediates are also higher in the mutant clones than in the WT clones. Moreover, α-KG largely rescues PIK3CA mutant cells from cell death caused by glutamine deprivation, suggesting that α-KG is a key metabolite required for PIK3CA mutant cell growth. Together, these data suggest that upregulation of GPT2 by PIK3CA mutations produces more α-KG from glutamine to replenish the TCA, thereby generating more ATP and intermediates for macromolecule synthesis to sustain rapid growth of PIK3CA mutant tumours. This is in agreement with our observation that both the ATP concentration and the ATP/ADP ratios were higher in the PIK3CA mutant cells than in the WT cells.

AOA potently inhibits xenograft tumour growth of CRCs with PIK3CA mutations, despite its low potency to inhibit growth of CRC cells in culture. A similar discrepancy was also observed with a breast cancer cell line²³. We postulate that tumour cells are more dependent on glutamine in the in vivo microenvironment than the in vitro tissue culture conditions. Although AOA is a pan-aminotransferase inhibitor, our data clearly demonstrate that the tumour inhibitory effect of AOA on PIK3CA mutant xenografts is indeed on-target to GPT2, because GPT2 knockdown cells grow much slower in nude nice and these tumours are insensitive to AOA treatment. Consistently, AOA treatment reduces amounts of α-KG and alanine, but increases amount of pyruvate and glutamate, in PIK3CA mutant xenograft tumours. Currently, only pan-aminotransferase inhibitors are available. It is conceivable that a GPT2-specific inhibitor could be more potent and less toxic. Given that GPTs are also involved in gluconeogenesis in liver and kidney, GPT inhibitors may cause hypoglycaemia. However, a study by Lindblom et al. shows that GPT1, but not GPT2, is predominantly expressed in liver and kidney³⁷. Thus, GPT2-specific inhibitors could still have a favourable therapeutic index.

It is generally believed that AKTs are the key mediators of the oncogenic signalling of PI3Ks (ref. 38). This study, however, revealed a novel p110α–PDK1–RSK2–ATF4–GPT2 pathway that regulates glutamine metabolism. We demonstrated that blocking this pathway inhibits PIK3CA mutant tumour growth in vitro and in vivo, suggesting that this novel signalling pathway also plays a critical role in tumorigenesis driven by PIK3CA mutations. While our previous studies showed that AKT signalling drives metastasis of PIK3CA mutant tumours³⁹, here we propose the p110α–PDK1–RSK2–ATF4–GPT2 signaling axis reprogrammes glutamine metabolism that is required to generate energy and metabolites to sustain rapid growth in PIK3CA mutant tumours.

In summary, we have discovered a novel mechanism through which mutant p110α perturbs the regulation of glutamine and makes cancer cells more dependent on adequate glutamine levels. We demonstrated that mutant p110α activates RSK2 kinase through PDK1. Activated RSK2 then phosphorylates ATF4 at the serine residue 245, which in turn recruits the deubiquitinase USP8 and protects ATF4 from ubiquitin-mediated degradation (Supplementary Fig. 8). This pathway can be targeted by various means, which may provide potential selective therapies for the treatment of the large number of patients whose tumours harbour PIK3CA mutations. Future experiments may extend our study to a large panel of CRC cell lines, patient-derived xenografts and transgenic PIK3CA mutant mouse models.

Methods

Cell culture

CRC cell lines HCT116, DLD1, RKO, HT29, SW480 and LOVO were purchased from American Type Culture Collection. Isogenic PIK3CA mutant and WT and KRAS mutant, and WT HCT116 and DLD1 cell lines were generated previously^17,20. All cell lines were maintained in McCoy’s 5A medium (Fisher Scientific, Catalogue No. SH30200) containing 10% fetal bovine serum (FBS, Fisher Scientific, Catalogue No. SH30910). Tissue cultures were routinely assayed for mycoplasma to ensure that they were mycoplasma-free. For glutamine deprivation, cells were cultured in glutamine- and pyruvate-free DMEM with 4.5 g l⁻¹ glucose (Fisher Scientific, Catalogue No. SH30081) with 10% dialysed FBS (dFBS; Invitrogen, Catalogue No. 26400). For glucose deprivation, cells were cultured in glucose-free DMEM (Invitrogen, Catalogue No. A14430) with 10% of dFBS.

Reagents

L-Glutamine, L-glutamine-¹³C₅, D-glucose, dimethyl α-KG and AOA were purchased from Sigma. Inhibitors LY294002 (pan-PI3K inhibitor), BYL-719 (p110α-specific inhibitor), BEZ235 (PI3K/mTOR dual-specificity inhibitor), GSK2334470 (PDK1 inhibitor), GSK690693 (AKT inhibitor), CHIR-99021 (GSK3β inhibitor), rapamycin (mTOR inhibitor), BI-D1870 (pan-RSK inhibitor), FMK (RSK2 inhibitor) and MG132 (proteasome inhibitor) were purchased from Selleck Chemicals. siRNAs for ATF4 (SI03019345 and SI04236337) and PDK1 (SI00301140 and SI00301154) were purchased from Qiagen. siRNAs for RSK2 (J-003026-10 and J-003026-12) were purchased from Dharmacon. siRNAs for USP8 (SR306014A and SR306014B) were purchased from Origene. shRNAs for GPT2 (TRCN0000035025 and TRCN0000035026) were purchased from Sigma. Antibodies used in this study are listed in Supplementary Table 1.

Plasmid construction

A GPT2 cDNA clone was purchased from Open Biosystems. The GPT2 ORF was subcloned into the pCMV-3Tag1A vector with HindIII and SalI. Then the FLAG-GPT2 sequence was PCRed out and subcloned into pCDNA3.1zeo with KpnI and XbaI. An enzymatically inactive form of GPT2 (K341H) was generated with the Quick-change Site-Directed Mutagenesis kit (Agilent Technologies). ATF4 and β-TrCP plasmids were obtained from Addgene. A FLAG-tagged ATF4 expression vector was constructed by subcloning ATF4 ORF into the pCMV-3Tag1A vector with BamHI and XhoI. ATF4 S219A and S245A mutations were made with the Quick-change Site-Directed Mutagenesis kit (Agilent Technologies). Then the pCMV-ATF4 S219A construct was digested with NotI and XhoI, and the ATF4 S219A fragment was subcloned into pCDNA3.1zeo. To make the construct expressing sgRNA to knock out GPT2, the CRISPR/Cas9 vector pX330 was digested with BbsI and ligated with annealed oligonucleotides. All primers used in this study are listed in Supplementary Table 2.

Plasmid transfection

Plasmids were transfected into cells with Lipofectamine 3000 (Invitrogen) according to the manufacturer’s instructions. For transient expression, cells were lysed 72 h after transfection. For stable expression (FLAG-GPT2 expression), cells were selected with 0.1 mg ml⁻¹ Zeocin (Invitrogen) for 7 days.

Serial analysis of gene expression

SAGE libraries were generated from 5 μg of total RNA, with reagents from the NlaIII Digital Gene Expression-Tag Profiling preparation kit (Illumina, San Diego, CA) according to the manufacturer’s instructions. This protocol results in a SAGE library wherein each tag is surrounded by adaptors suited for sequencing on the Illumina Genome Analyzer platform (Illumina). Each library was loaded on one channel of an Illumina flow cell, and 17 cycles of sequencing were performed. Sequence tags with Phred-equivalent score >20 for every base were used for analysis.

RT–PCR analysis

RNAs were extracted using RNeasy mini kits (Qiagen) according to the manufacturer’s instructions. A unit of 1 μg total RNA was used for reverse transcription using the Superscript First-Strand kit (Invitrogen) and cDNAs were used as templates for PCR. Primers are listed in Supplementary Table 2.

Real-time PCR analysis

RNAs from frozen tumours were extracted as described above. The Taqman assay system was used for qRT–PCR using GPT2 probes (Hs00370287, Applied Biosystems) with IQ super mix (Catalogue No. 170-8860, Bio-Rad). Expression levels of GPT2 in each tumour was normalized to that of β-2-microglobulin. The PIK3CA mutation status of the human CRC specimens is listed in Supplementary Table 3.

Mining TCGA data

For TCGA colon adenocarcinoma and rectal adenocarcinoma patients, RNA-seq files in the *rsem.genes.normalized_results format were downloaded from the TCGA website for all tumours and, where available, matched normal tissue samples. This file format provides normalized fragments per kilobase of exon per million reads mapped (FPKM) values, using the RSEM software⁴⁰, for each gene. In addition, the somatic mutation.maf files were downloaded from the same site for each tumour type. For the PIK3CA mutant group, tumours with the following characteristics were excluded: (1) nonsense mutations that generate stop codons; (2) unknown mutations that were not confirmed; and (3) possible tumour–normal mismatches that had multiple mutations in PIK3CA.

Gene targeting

GPT2 knockout clones were generated with CRISPR/Cas 9. A pX330 plasmid expressing sgRNA that targets exon 4 of GPT2 genomic sequences was transfected into HCT116 cells, and single clones were selected. Knockout clones were verified by sequencing and western blot analysis.

The PIK3CA D933A targeting vector was constructed with the USER system, and targeted cells were generated with rAAV as described previously^41,42. Briefly, vector arms were created by PCR from genomic DNA using HiFi Taq (Invitrogen) and validated by sequencing before viral production and infection. Stable G418-resistent clones were then selected for PCR screening as reported⁴². Targeted clones were genotyped by RT–PCR and sequencing. Primers for constructing the targeting vector and screening for knock-in clones are listed in Supplementary Table 3.

Immunoblotting and immunoprecipitation

Cells were lysed in RIPA buffer (10 mM Tris (pH 7.4), 150 mM NaCl, 5 mM EDTA (pH 8.0), 0.1% SDS, 1% Triton-X100, 1 mM dithiothreitol, 1 mM phenylmethyl sulphonyl fluoride and complete Protease Inhibitor Cocktail tablet (Roche); supplemented with phosphatase inhibitors (1 mM Na₃VO₄, 50 mM NaF, 1 mM β-glycerophosphate and 20 mM sodium pyrophosphate)). Lysates were cleared by centrifugation at 14,000 r.p.m. for 10 min, and protein concentration in supernatants was determined by the BCA protein assay kit (Pierce). Equal amounts of total protein were used for immunoblotting. For immunoprecipitation, cells were lysed as described above. Cleared cell lysates were incubated with antibody for 1 h, and then protein A and/or protein G for 1 h. Protein A/G beads were washed with lysis buffer three times, and then boiled with SDS-loading buffer followed by immunoblotting. Original images of western blots are shown in Supplementary Fig. 9.

Luciferase reporter assays

The 1.5-kb promoter of GPT2 was subcloned into the pGL3 vector (Promega) to obtain a pGL3-GPT2 promoter-LUC plasmid. pGL3-GPT2 promoter-LUC was co-transfect with pCMV-ATF4 and an internal control β-galactosidase expressing pCH110 plasmid (abbreviated as pCH110 β-gal) or Renilla luciferase expressing pRL (Promega). Forty-eight hours after transfection, cells were collected for Luciferase assay according to the manufacturer’s instructions (Promega). Luminescence was measured with an EnVision 2103 Multilabel Plate Reader (PerkinElmer). β-galactosidase activity was measured with a β-Gal assay kit (Invitrogen). Relative luciferase activity was expressed as the ratio of luminescence normalized to the internal control. Two putative ATF4-binding sites in the GPT2 promoter were predicted by TFseach v1.3. Mutations in the GPT2 promoter were generated with the Quickchange kit (Agilent Technologies) and primers in Supplementary Table 3. The uORFATF4 plasmid was a kind gift from Dr Ron Wek at Indiana University.

Cell proliferation assays

Cells were plated in 96-well plates at 2,000 cells per well, 24-well plates at 1 × 10⁴ cells per well and 6-well plates at 2 × 10⁵ cells per well in complete DMEM (20 mM glucose, 2 mM glutamine, 10% dFBS; Invitrogen). After 24 h, cells were washed with PBS, and changed to either glutamine-free DMEM (with 20 mM glucose) or glucose-free DMEM (with 2 mM glutamine) containing 10% dFBS. Cells (including floating cells in medium) were collected and counted by Trypan-Blue exclusive assay. Cell numbers in 96-well plates were measured by Cell Counting Kit 8 (Dojindo) according to the manufacturer’s instructions.

Flow cytometry

Cells were fixed with methanol and then incubated at 37 °C for 30 min in 5% normal goat serum diluted in PBS. Propidium iodide solution was used to stain cells at 4 °C for 1 h. Cells were analyzed on an Epics XL flow cytometer. WinMDI2.9 software was used for data analysis. Cell debris and aggregates were excluded by propidium iodide gating. Percentages of sub-G1, G1, S and G2/M populations were determined by histograms generated by WinDI2.9.

Gene silencing

Plasmids expressing shRNAs were transfected into cells. Two days post transfection, cells were selected with 1 μg ml⁻¹ puromycin for 7 days. Puromycin-resistant cells were pooled, amplified and analysed. For genes silenced by siRNAs, siRNAs were transfected into cells with Lipofectamine 3000. Three days post transfection, cells were collected for further analyses.

Ubiquitination assays

Cells were pretreated with 5 μM of MG132 for 6 h as previously described⁴³, and cell lysates were immunoprecipitated with antibodies against either ATF4 or FLAG. Beads were washed three times with washing buffer (10 mM Tris (pH 7.4), 1 M NaCl, 1 mM EDTA (pH 8.0), 1% NP-40 and 0.2% SDS). The immunocomplexes were resolved in SDS–PAGE gels for western blot analyses.

Polysome profile analysis

Three tumours (∼250 mm³ size) of each genotype were snap-frozen in liquid nitrogen, pulverized and then lysed in 1 ml of lysis buffer (50 mM HEPES-KOH (pH 7.4), 5 mM MgCl₂, 250 mM KCl, 2% Triton X-100, 8.5% sucrose, 100 μg ml⁻¹ cycloheximide, 1 mM dithiothreitol, 200 U ml⁻¹ RNase inhibitor (RNaseOUT, Invitrogen), EDTA-free protease inhibitor (Roche Applied Science) and 10 mM ribonucleoside vanadyl complex (New England Biolabs)), kept on ice for 20 min and then passed 15 times through a 23-gauge needle. Lysates were spun at 14,000 r.p.m. for 15 min, and supernatants were collected. Approximately 10–15 absorbance units (260 nm) of lysates were layered over cold 10–50% sucrose gradients in buffer (50 mM HEPES-KOH (pH 7.4), 5 mM MgCl₂ and 250 mM KCl). Gradients were centrifuged at 17,000 r.p.m. in a Beckman SW28 rotor for 15 h at 4 °C. After centrifugation, 12 fractions (1.2 ml per fraction) were collected. RNA from each fraction was isolated using TRIzol LS reagent (Invitrogen), and an equal volume from each fraction was used for cDNA synthesis. The relative quantities of specific mRNAs were measured by qRT–PCR.

Inhibitor treatments

To assay ATF4 protein stability, HCT116 PIK3CA mutant or WT cells were cultured in complete McCoy5A medium and treated with vehicle or inhibitors for 12 h, then lysed for western blot analyses. For ubiquitination assay by immunoprecipitation, cells were treated with 5 μM MG132 for 6 h before they were lysed. For long-term treatment with MG132, cells were exposed to 0.2 μM MG132 for 48 h. Cells were lysed for western blot analysis, or fixed for flow cytometry.

Xenograft studies

Animal experiments were approved by the Case Western Reserve University Institutional Animal Care and Use Committee. As described in ref. 44, 3 million cells were injected subcutaneously into the flanks of 4- to 6-week-old female athymic nude mice. Mice were randomly assigned into treatment groups (5 mice per group). When average tumour volume reached about 100 mm³, mice were injected intraperitoneally with vehicle or with 5 mg kg⁻¹ or 10 mg kg⁻¹ of AOA every day. Tumour volumes were measured with an electronic caliper and calculated as length × width²/2. Body weights were measured at the beginning and at the end of treatment. These experiments were not blinded.

Glutamine consumption

One million cells were plated in complete medium in a T25 flask. When the cells were ∼70% confluent, they were washed twice with PBS, and changed into medium containing 2 mM of [¹³C₅]L-glutamine (C_Gln0) and cultured for 24 h. Medium was collected and frozen at −80 °C. To measure the glutamine concentration remaining in the medium (C_Gln__end), 50 μl of medium was mixed with 50 μl of 2 mM [¹²C]L-glutamine as internal standard and 500 μl of pre-chilled (−80 °C) methanol was added. The supernatants were dried with nitrogen gas. TBDMCS (tert-butyldimethylchlorosilane, REGIS Technologies):acetonitrile (2:1) was used for derivatization of metabolites at 60 °C for 1 h. Samples were injected into GC-MS (Agilent Technologies) for analysis. Glutamine concentration in medium was determined by normalization to the internal standard. Glutamine (Gln) consumption rate=(C_Gln0−C_Gln__end) × medium volume/cell numbers/incubation time.

Metabolic assays and stable isotope tracing

A million cells were plated in each T25 flask. When cells reached ∼70% confluency, they were washed with PBS twice and changed to medium containing 2 mM of [¹³C₅-]glutamine. Cells were collected with 1 ml pre-chilled (−80 °C) methanol at 0.5, 1, 2 and 4 h. A unit of 5 μM of heptadecanoic acid, 2.5 μM of [3,3,4,5,5,5-²H6]4-hydroxypentanoate and 2.5 μM of [2,2,3,3,4,4,5,5,6,6,7,7,7-²H13] heptanoate were added as internal standards. Metabolites were extracted by homogenization and sonication on ice. Cell debris was removed by centrifugation at 14,000 r.p.m. for15 min at 4 °C. The supernatant was dried with nitrogen gas. TBDMCS:acetonitrile (2:1) was used for derivatization of metabolites at 60 °C for 1 h. Samples (1 μl) were injected into GC-MS (Agilent Technologies) for metabolite profiling. The direct [¹³C₅-]glutamine metabolites are ¹³C₅-glutamate (M5 glutamate), ¹³C₅-α-KG (M5 α-KG), ¹³C₄-succinate (M4 succinate), ¹³C₄-fumarate (M4 fumarate) and ¹³C₄-citrate (M4 citrate). For metabolite flux analyses, the total pool of each metabolite was considered as 100%. Metabolite fluxes were presented as percentage of direct glutamine-derived metabolites in the total pool. For analysis of metabolite levels, the amounts of labelled glutamate, α-KG and TCA intermediates at 2 h were measured and normalized to internal standards. Unlabelled pyruvate and alanine were measured and normalized to D13 heptanoate.

Assays of ATP/ADP and NADH/NAD

The amounts of ATP and the ATP/ADP ratios were measured with an ADP/ATP ratio assay kit (Abcam) according to the manufacturer’s instructions. The amounts of ATP were determined by an ATP standard curve, and normalized to the protein concentrations. The amounts of NADH and NADH/NAD ratios were measured with an NAD/NADH colorimetric kit (BioVision). The NADH concentrations were determined by a NADH standard curve, and normalized to the protein concentrations.

Statistical analyses

We applied the t-test to compare the means between two groups, assuming unequal variances. For xenograft growth, analysis of variance was performed for repeated measurements to test whether there was an overall difference in the tumour size.

Data availability

The TCGA data referenced during the study are available in a public repository from the TCGA website. The authors declare that all the other data supporting the findings of this study are available within the article and its Supplementary Information files and from the corresponding author on reasonable request.

Additional information

How to cite this article: Hao, Y. et al. Oncogenic PIK3CA mutations reprogram glutamine metabolism in colorectal cancer. Nat. Commun. 7:11971 doi: 10.1038/ncomms11971 (2016).

References

Dang, C. V. Links between metabolism and cancer. Genes Dev. 26, 877–890 (2012).
Article CAS Google Scholar
Vander Heiden, M. G., Cantley, L. C. & Thompson, C. B. Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 324, 1029–1033 (2009).
Article CAS ADS Google Scholar
Abraham, R. T. & Eng, C. H. A metabolic (re-)balancing act. Mol. Cell 38, 481–482 (2010).
Article CAS Google Scholar
Hensley, C. T., Wasti, A. T. & DeBerardinis, R. J. Glutamine and cancer: cell biology, physiology, and clinical opportunities. J. Clin. Invest. 123, 3678–3684 (2013).
Article CAS Google Scholar
Gao, P. et al. c-Myc suppression of miR-23a/b enhances mitochondrial glutaminase expression and glutamine metabolism. Nature 458, 762–765 (2009).
Article CAS ADS Google Scholar
Qing, G. et al. ATF4 regulates MYC-mediated neuroblastoma cell death upon glutamine deprivation. Cancer Cell 22, 631–644 (2012).
Article CAS Google Scholar
Reynolds, M. R. et al. Control of glutamine metabolism by the tumor suppressor Rb. Oncogene 33, 556–566 (2014).
Article CAS Google Scholar
Suzuki, S. et al. Phosphate-activated glutaminase (GLS2), a p53-inducible regulator of glutamine metabolism and reactive oxygen species. Proc. Natl. Acad. Sci. USA 107, 7461–7466 (2010).
Article CAS ADS Google Scholar
Jiang, P., Du, W., Mancuso, A., Wellen, K. E. & Yang, X. Reciprocal regulation of p53 and malic enzymes modulates metabolism and senescence. Nature 493, 689–693 (2013).
Article CAS ADS Google Scholar
Gameiro, P. A. et al. In vivo HIF-mediated reductive carboxylation is regulated by citrate levels and sensitizes VHL-deficient cells to glutamine deprivation. Cell Metab. 17, 372–385 (2013).
Article CAS Google Scholar
Son, J. et al. Glutamine supports pancreatic cancer growth through a KRAS-regulated metabolic pathway. Nature 496, 101–105 (2013).
Article CAS ADS Google Scholar
Cantley, L. C. The phosphoinositide 3-kinase pathway. Science 296, 1655–1657 (2002).
Article CAS ADS Google Scholar
Hao, Y., Zhao, S. & Wang, Z. Targeting the protein-protein interaction between IRS1 and mutant p110alpha for cancer therapy. Toxicol. Pathol. 42, 140–147 (2014).
Article CAS Google Scholar
Samuels, Y. et al. High frequency of mutations of the PIK3CA gene in human cancers. Science 304, 554 (2004).
Article CAS Google Scholar
Lawrence, M. S. et al. Discovery and saturation analysis of cancer genes across 21 tumour types. Nature 505, 495–501 (2014).
Article CAS ADS Google Scholar
Zhao, L. & Vogt, P. K. Class I PI3K in oncogenic cellular transformation. Oncogene 27, 5486–5496 (2008).
Article CAS Google Scholar
Samuels, Y. et al. Mutant PIK3CA promotes cell growth and invasion of human cancer cells. Cancer Cell 7, 561–573 (2005).
Article CAS Google Scholar
Leithner, K. et al. PCK2 activation mediates an adaptive response to glucose depletion in lung cancer. Oncogene 34, 1044–1050 (2015).
Article CAS Google Scholar
Vincent, E. E. et al. Mitochondrial phosphoenolpyruvate carboxykinase regulates metabolic adaptation and enables glucose-independent tumor growth. Mol. Cell 60, 195–207 (2015).
Article CAS Google Scholar
Yun, J. et al. Glucose deprivation contributes to the development of KRAS pathway mutations in tumor cells. Science 325, 1555–1559 (2009).
Article CAS ADS Google Scholar
Velculescu, V. E., Zhang, L., Vogelstein, B. & Kinzler, K. W. Serial analysis of gene expression. Science 270, 484–487 (1995).
Article CAS ADS Google Scholar
Wood, L. D. et al. The genomic landscapes of human breast and colorectal cancers. Science 318, 1108–1113 (2007).
Article CAS ADS Google Scholar
Thornburg, J. M. et al. Targeting aspartate aminotransferase in breast cancer. Breast Cancer Res. 10, R84 (2008).
Article Google Scholar
Csibi, A. et al. The mTORC1 pathway stimulates glutamine metabolism and cell proliferation by repressing SIRT4. Cell 153, 840–854 (2013).
Article CAS Google Scholar
Guan, B. J. et al. Translational control during endoplasmic reticulum stress beyond phosphorylation of the translation initiation factor eIF2alpha. J. Biol. Chem. 289, 12593–12611 (2014).
Article CAS Google Scholar
Yang, X. et al. ATF4 is a substrate of RSK2 and an essential regulator of osteoblast biology; implication for Coffin-Lowry Syndrome. Cell 117, 387–398 (2004).
Article CAS Google Scholar
Frodin, M. et al. A phosphoserine/threonine-binding pocket in AGC kinases and PDK1 mediates activation by hydrophobic motif phosphorylation. The EMBO J. 21, 5396–5407 (2002).
Article Google Scholar
Weinberg, F. et al. Mitochondrial metabolism and ROS generation are essential for Kras-mediated tumorigenicity. Proc. Natl Acad. Sci. USA 107, 8788–8793 (2010).
Article CAS ADS Google Scholar
Tuveson, D. A. et al. Endogenous oncogenic K-ras(G12D) stimulates proliferation and widespread neoplastic and developmental defects. Cancer Cell 5, 375–387 (2004).
Article CAS Google Scholar
Arena, S. et al. Knock-in of oncogenic Kras does not transform mouse somatic cells but triggers a transcriptional response that classifies human cancers. Cancer Res. 67, 8468–8476 (2007).
Article CAS Google Scholar
Castellano, E. & Downward, J. RAS Interaction with PI3K: more than just another effector pathway. Genes Cancer 2, 261–274 (2011).
Article CAS Google Scholar
Taniguchi, C. M. et al. Divergent regulation of hepatic glucose and lipid metabolism by phosphoinositide 3-kinase via Akt and PKClambda/zeta. Cell Metab. 3, 343–353 (2006).
Article CAS Google Scholar
Hu, H. et al. Phosphoinositide 3-kinase regulates glycolysis through mobilization of aldolase from the actin cytoskeleton. Cell 164, 433–446 (2016).
Article CAS Google Scholar
Engelman, J. A. Targeting PI3K signalling in cancer: opportunities, challenges and limitations. Nat. Rev. Cancer 9, 550–562 (2009).
Article CAS Google Scholar
Vadas, O., Burke, J. E., Zhang, X., Berndt, A. & Williams, R. L. Structural basis for activation and inhibition of class I phosphoinositide 3-kinases. Sci. Signal. 4, re2 (2011).
Article Google Scholar
Costa, C. et al. Measurement of PIP Levels reveals an unexpected role for p110bemta in early adaptive responses to p110alpha-specific inhibitors in luminal breast cancer. Cancer Cell 27, 97–108 (2014).
Article Google Scholar
Lindblom, P. et al. Isoforms of alanine aminotransferases in human tissues and serum—differential tissue expression using novel antibodies. Arch. Biochem. Biophys. 466, 66–77 (2007).
Article CAS Google Scholar
Klempner, S. J., Myers, A. P. & Cantley, L. C. What a tangled web we weave: emerging resistance mechanisms to inhibition of the phosphoinositide 3-kinase pathway. Cancer Discov. 3, 1345–1354 (2013).
Article CAS Google Scholar
Ericson, K. et al. Genetic inactivation of AKT1, AKT2, and PDPK1 in human colorectal cancer cells clarifies their roles in tumor growth regulation. Proc. Natl. Acad. Sci. USA 107, 2598–2603 (2010).
Article CAS ADS Google Scholar
Li, B. & Dewey, C. N. RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinformatics 12, 323 (2011).
Article CAS Google Scholar
Zhang, X. et al. Epitope tagging of endogenous proteins for genome-wide ChIP-chip studies. Nat. Methods 5, 163–165 (2008).
Article CAS Google Scholar
Hao, Y. et al. Gain of interaction with IRS1 by p110alpha-helical domain mutants is crucial for their oncogenic functions. Cancer Cell 23, 583–593 (2013).
Article CAS Google Scholar
Du, Z. et al. DNMT1 stability is regulated by proteins coordinating deubiquitination and acetylation-driven ubiquitination. Sci. Signal. 3, ra80 (2010).
Article Google Scholar
Zhao, Y. et al. Identification and functional characterization of paxillin as a target of protein tyrosine phosphatase receptor T. Proc. Natl. Acad. Sci. USA 107, 2592–2597 (2010).
Article CAS ADS Google Scholar

Download references

Acknowledgements

We thank Dr Anthony Wynshaw-Boris for critical reading of the manuscript. This work is supported by NIH grants R01CA196643, R21CA160060, R21CA181859, R01CA127590, P50CA150964 and P30 CA043703 (to Z.W.) and R37-DK060596 and R01-DK053307 (to M.H.).

Author information

Chao Wang
Present address: Present address: Division of Cellular & Molecular Research, National Cancer Center Singapore, Singapore 169610, Singapore.,
Yardena Samuels, Qingling Li, Dawid Krokowski, Bo-Jhih Guan and Chao Wang: These authors contributed equally to this work.

Authors and Affiliations

Department of Genetics and Genome Sciences, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, 44106, Ohio, USA
Yujun Hao, Chao Wang, Bohan Dong, Bo Cao, Xiujing Feng, Min Xiang, Ronald A. Conlon, Thomas LaFramboise & Zhenghe Wang
Case Comprehensive Cancer Center, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, 44106, Ohio, USA
Yujun Hao, Chao Wang, Xiujing Feng, Stephen Fink, Neal J. Meropol, Ronald A. Conlon, Sanford Markowitz, Thomas LaFramboise & Zhenghe Wang
Ludwig Center and Howard Hughes Medical Institute, Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, 21231, Maryland, USA
Yardena Samuels, Kenneth W. Kinzler, Victor E. Velculescu & Bert Vogelstein
Department of Molecular Cell Biology, The Weizmann Institute of Science, Rehovot, 76100, Israel
Yardena Samuels
Department of Nutrition, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, 44106, Ohio, USA
Qingling Li, Zhicheng Jin, Henri Brunengraber & Guo-Fang Zhang
Department of Pharmacology, School of Medicine, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, 44106, Ohio, USA
Dawid Krokowski, Bo-Jhih Guan & Maria Hatzoglou
Department of Biochemistry and Molecular Biology, Wannan Medical College, Wuhu, 241000, China
Bohan Dong
Department of Pharmacognosy, School of Pharmacy, Third Military Medical University, Chongqing, 400038, China
Bo Cao
Department of Pharmacy, Suzhou Health College, Suzhou, 215009, Jiangsu, China
Min Xiang
Hathaway Brown School, 19600 North Park Boulevard, Shaker Heights, 44122, Ohio, USA
Claire Xu
Department of Medicine, University Hospitals Case Medical Center, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, 44106, Ohio, USA
Stephen Fink, Neal J. Meropol & Sanford Markowitz
Department of Chemistry, Cleveland State University, 2121 Euclid Avenue, Cleveland, 44115, Ohio, USA
Yan Xu
Department of Pathology, University Hospitals Case Medical Center, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, 44106, Ohio, USA
Joseph E. Willis

Authors

Yujun Hao
View author publications
You can also search for this author in PubMed Google Scholar
Yardena Samuels
View author publications
You can also search for this author in PubMed Google Scholar
Qingling Li
View author publications
You can also search for this author in PubMed Google Scholar
Dawid Krokowski
View author publications
You can also search for this author in PubMed Google Scholar
Bo-Jhih Guan
View author publications
You can also search for this author in PubMed Google Scholar
Chao Wang
View author publications
You can also search for this author in PubMed Google Scholar
Zhicheng Jin
View author publications
You can also search for this author in PubMed Google Scholar
Bohan Dong
View author publications
You can also search for this author in PubMed Google Scholar
Bo Cao
View author publications
You can also search for this author in PubMed Google Scholar
Xiujing Feng
View author publications
You can also search for this author in PubMed Google Scholar
Min Xiang
View author publications
You can also search for this author in PubMed Google Scholar
Claire Xu
View author publications
You can also search for this author in PubMed Google Scholar
Stephen Fink
View author publications
You can also search for this author in PubMed Google Scholar
Neal J. Meropol
View author publications
You can also search for this author in PubMed Google Scholar
Yan Xu
View author publications
You can also search for this author in PubMed Google Scholar
Ronald A. Conlon
View author publications
You can also search for this author in PubMed Google Scholar
Sanford Markowitz
View author publications
You can also search for this author in PubMed Google Scholar
Kenneth W. Kinzler
View author publications
You can also search for this author in PubMed Google Scholar
Victor E. Velculescu
View author publications
You can also search for this author in PubMed Google Scholar
Henri Brunengraber
View author publications
You can also search for this author in PubMed Google Scholar
Joseph E. Willis
View author publications
You can also search for this author in PubMed Google Scholar
Thomas LaFramboise
View author publications
You can also search for this author in PubMed Google Scholar
Maria Hatzoglou
View author publications
You can also search for this author in PubMed Google Scholar
Guo-Fang Zhang
View author publications
You can also search for this author in PubMed Google Scholar
Bert Vogelstein
View author publications
You can also search for this author in PubMed Google Scholar
Zhenghe Wang
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

Z.W., Y.H., B.V., G.-F.Z., M.H., V.E.V., K.W.K., H.B., Y.S., C.W., S.M., R.C., X.Y. and N.J.M. designed the experiments; Y.H., Y.S., Q.L., D.K., B.-J.G., C.W., J.E.W., Z.J., B.D., B.C., X.F., M.X. and C.X. performed the experiments; Y.H., T.L. and Z.W. analysed the data; Z.W., B.V., Y.H., R.C., M.H., K.W.K. and S.M. wrote the manuscript.

Corresponding author

Correspondence to Zhenghe Wang.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Supplementary information

Supplementary Information

Supplementary Figures 1-9 and Supplementary Tables 1-3 (PDF 2657 kb)

Supplementary Data 1

Gene expression analysis of isogenic PIK3CA WT and mutant only cells. (XLSX 65 kb)

Rights and permissions

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Reprints and permissions

About this article

Cite this article

Hao, Y., Samuels, Y., Li, Q. et al. Oncogenic PIK3CA mutations reprogram glutamine metabolism in colorectal cancer. Nat Commun 7, 11971 (2016). https://doi.org/10.1038/ncomms11971

Download citation

Received: 11 January 2016
Accepted: 17 May 2016
Published: 20 June 2016
DOI: https://doi.org/10.1038/ncomms11971

This article is cited by

Glutamine addiction in tumor cell: oncogene regulation and clinical treatment
- Xian Li
- Xueqiang Peng
- Hangyu Li
Cell Communication and Signaling (2024)
SIRT5-mediated ME2 desuccinylation promotes cancer growth by enhancing mitochondrial respiration
- Peng Teng
- Kaisa Cui
- Zhaohui Huang
Cell Death & Differentiation (2024)
Metabolic reprogramming in colorectal cancer: regulatory networks and therapy
- Jieping Zhang
- Shaomin Zou
- Lekun Fang
Cell & Bioscience (2023)
Metastasis organotropism in colorectal cancer: advancing toward innovative therapies
- Kai He
- Zhihan Wang
- Hai-Ning Chen
Journal of Translational Medicine (2023)
Phosphorylation at tyrosine 317 and 508 are crucial for PIK3CA/p110α to promote CRC tumorigenesis
- Ting Wang
- Longci Sun
- Yujun Hao
Cell & Bioscience (2023)

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.