Research ArticleCancer Metabolism

Oncogenic PI3K promotes methionine dependency in breast cancer cells through the cystine-glutamate antiporter xCT

See allHide authors and affiliations

Sci. Signal.  19 Dec 2017:
Vol. 10, Issue 510, eaao6604
DOI: 10.1126/scisignal.aao6604

Understanding the causes of methionine dependency

Cells can use the same precursor to produce either methionine or cysteine, but can free up cysteine from cystine, an oxidized cysteine dimer with structural functions. Methionine dependency, or the inability to efficiently produce methionine, is a metabolic vulnerability that could be exploited to treat certain cancers. Examination of several breast cancer cell lines by Lien et al. revealed a correlation between methionine dependency, the presence of oncogenic mutations in PIK3CA, which encodes the lipid kinase PI3Kα, and decreased expression of SLC7A11, which encodes the cystine transporter xCT. Oncogenic PI3Kα mutants not only transcriptionally suppressed SLC7A11 expression but also inhibited the activity of xCT through AKT-mediated phosphorylation. These results highlight how mutations in the PI3K pathway result in the metabolic reprogramming of cancer cells.

Abstract

The precursor homocysteine is metabolized either through the methionine cycle to produce methionine or through the transsulfuration pathway to synthesize cysteine. Alternatively, cysteine can be obtained through uptake of its oxidized form, cystine. Many cancer cells exhibit methionine dependency such that their proliferation is impaired in growth media in which methionine is replaced by homocysteine. We showed that oncogenic PIK3CA and decreased expression of SLC7A11, a gene that encodes a cystine transporter also known as xCT, correlated with increased methionine dependency in breast cancer cells. Oncogenic PIK3CA was sufficient to confer methionine dependency to mammary epithelial cells, partly by decreasing cystine uptake through the transcriptional and posttranslational inhibition of xCT. Manipulation of xCT activity altered the proliferation of breast cancer cells in methionine-deficient, homocysteine-containing media, suggesting that it functionally contributed to methionine dependency. We propose that concurrent with decreased cystine uptake through xCT, PIK3CA mutant cells use homocysteine through the transsulfuration pathway to synthesize cysteine. Consequently, less homocysteine is available to produce methionine, contributing to methionine dependency. These results indicate that oncogenic PIK3CA alters methionine and cysteine utilization, partly by inhibiting xCT to contribute to the methionine dependency phenotype in breast cancer cells.

INTRODUCTION

In the past decade, there has been a resurgence of interest in elucidating how metabolism is altered in cancer cells, with the goal of identifying cancer-associated metabolic dependencies that can be exploited for cancer therapy (1). Metabolic differences between cancerous and normal cells often involve differential utilization of key “junction” metabolites. For example, one aspect of the Warburg effect is the preferential usage of glycolysis-derived pyruvate to generate lactate in cancer cells, whereas in normal cells, pyruvate is primarily directed toward the tricarboxylic acid cycle. From a therapeutic standpoint, differences in how cancer cells regulate the fate of key metabolites may potentially provide a means of targeting these metabolic junctions for treatment.

Homocysteine (Hcy) is a key junction metabolite that lies at the nexus of two pathways involved in methionine (Met) and cysteine (Cys) metabolism. High concentrations of Hcy are toxic to cells, and medical disorders known as hyperhomocysteinemia and homocystinuria are characterized by the accumulation of Hcy in the blood, leading to various symptoms, such as stroke, vascular diseases, and intellectual disabilities (2). Therefore, cells must metabolize Hcy primarily through two different pathways: The methionine cycle and the transsulfuration pathway (Fig. 1A). In the methionine cycle, Hcy is methylated to produce Met, an essential amino acid critical for cell growth and function. In addition to contributing to protein synthesis, Met is a precursor for the generation of S-adenosylmethionine, which, as the principal methyl group donor, is critical for various cellular methylation reactions (3). S-adenosylhomocysteine is generated in the process and subsequently converted into Hcy, which is then used to regenerate Met to complete the cycle. Alternatively, Hcy can be metabolized through the transsulfuration pathway to synthesize the amino acid Cys, which is involved in multiple cellular antioxidant systems, such as the synthesis of glutathione (4). Depending on cellular demand, Hcy can be directed either toward the methionine cycle to increase methylation potential or through the transsulfuration pathway to contribute to antioxidant metabolism.

Fig. 1 Proliferation of breast cancer cell lines in MetHcy+ media.

(A) Schematic of the methionine cycle and transsulfuration pathway. Met, methionine; SAM, S-adenosylmethionine; SAH, S-adenosylhomocysteine; Hcy, homocysteine; Ser, serine; Cys, cysteine; αKB, alpha-ketobutyrate; MAT, methionine adenosyltransferase; AHCY, adenosylhomocysteinase; MTR, 5-methyltetrahydrofolate-homocysteine methyltransferase; CBS, cystathionine-beta-synthase; CTH, cystathionine gamma-lyase; AU, arbitrary units. (B) Cell lines were screened for their growth in MetHcy+ media for 4 days, and the proliferation of cells was determined using the sulforhodamine B (SRB) assay (n = 3 independent replicates). (C) Proliferation data from (B) were fit to an exponential curve to calculate the growth rate of each cell line in MetHcy+ media. (D) Pearson correlation of the growth rates of the cell lines in MetHcy+ media with their doubling time in Met+Hcy media. All error bars represent SEM.

In the context of cancer, the Hcy junction has been implicated in a cancer-associated metabolic vulnerability known as “methionine dependency,” in which a majority of cancer cells cannot proliferate in growth media in which Met is replaced by its precursor Hcy (MetHcy+ media). By contrast, most normal, nontumorigenic cells, such as fibroblasts and epithelial cells, are not methionine-dependent and can proliferate in MetHcy+ media (57). This phenotype has been demonstrated for various malignant cell lines and for patient tumors grown in primary culture from multiple cancers, including breast, bladder, colon, glioma, kidney, melanoma, and prostate cancer (813). Given these observations, methionine restriction has been proposed as a strategy for treating cancer, a notion that is supported by several preclinical models. For example, in animal models of various cancers, including rhabdomyosarcoma, Yoshida’s sarcoma, hepatoma, and colorectal cancer, methionine-restricted diets inhibit tumor growth, prevent metastases, and extend survival (1417). Other studies suggest that the enzyme methioninase, which degrades methionine, may be used pharmacologically to systemically deplete methionine levels to exert antitumor effects (1821). Finally, preclinical and clinical studies have indicated that methionine restriction can act synergistically with chemotherapeutic agents to effectively treat tumors (2226).

Despite these promising results, one critical concern is that complete and prolonged systemic depletion of methionine may be toxic and even lethal in humans (14, 27, 28). One challenge in more effectively exploiting methionine dependency in cancer treatment is that the mechanisms underlying this phenotype remain elusive. A few studies have evaluated the role of certain metabolic enzymes involved in methionine metabolism, especially methionine synthase (MTR), which converts Hcy to Met, and methylthioadenosine phosphorylase (MTAP), which is involved in the methionine salvage pathway (2935). Although these enzymes have been associated with methionine dependency, the exact metabolic alterations that are causally responsible for conferring methionine dependency have not been determined. Here, we provide evidence that oncogenic PIK3CA is sufficient to confer methionine dependency to mammary epithelial cells, partly by decreasing cystine (Cys2) uptake through the transcriptional and posttranslational inhibition of xCT, a Cys2 transporter encoded by SLC7A11.

RESULTS

Methionine dependency varies across different breast cancer cell lines

To study methionine dependency in breast cancer, we screened 13 breast cancer cell lines for their ability to proliferate in MetHcy+ media (Fig. 1B). On the basis of their proliferative capacities, we segregated the cell lines into three groups: Group A cells proliferated; group B cells did not proliferate, at least over the 4 days of the experiment; and group C cells died when cultured in MetHcy+ media. To quantify the differences in methionine dependencies across the cell lines, we used these proliferation curves to calculate a MetHcy+ growth rate for each cell line, again observing a segregation into the three groups (Fig. 1C).

Differences in basal proliferation rates may potentially explain the observed range of methionine dependencies, because more proliferative cells may require more methionine for growth and therefore be more methionine-dependent. However, we did not find a significant correlation between the MetHcy+ growth rates of the cell lines and their doubling times in regular growth media (Fig. 1D). Therefore, the range of methionine dependencies across the cell lines was not likely to be simply due to differences in basal proliferative capacity.

Oncogenic PIK3CA is sufficient to confer methionine dependency to mammary epithelial cells

We noticed that many of the most methionine-dependent breast cancer cell lines harbor a PIK3CA genomic mutation (Table 1), particularly two of the three group C cell lines (MCF7 and T47D) and the most methionine-dependent group B cell line (SUM-159). Notably, PTEN genomic mutations also tended to cluster toward the group B and C cell lines with negative MetHcy+ growth rates (Fig. 1C and Table 1). However, drawing a correlation between PTEN protein abundance and methionine dependency will require a quantitative analysis of PTEN protein abundance in these cell lines. All group A cell lines were negative for PTEN abundance at the protein level (Table 1), weakening the association of PTEN mutational status with methionine dependency. Given that two of the three group C cell lines harbored a PIK3CA mutation, we focused on assessing whether oncogenic PIK3CA might contribute to the methionine dependency phenotype. To test this hypothesis, we evaluated whether expression of oncogenic PIK3CA mutations might render the nontumorigenic mammary epithelial MCF10A cell line sensitive to the MetHcy+ media. Wild-type (WT) PIK3CA, PIK3CA(E545K), or PIK3CA(H1047R) were stably expressed at similar abundances in MCF10A cells (fig. S1A). Although these cell lines proliferated well in normal media containing Met (fig. S1B), they behaved differently in MetHcy+ media (Fig. 2A). Empty vector–transfected MCF10A cells proliferated in MetHcy+ media, as might be expected for a relatively normal cell line. Expression of WT PIK3CA inhibited the growth of MCF10A cells, whereas cells expressing PIK3CA(E545K) or PIK3CA(H1047R) died when cultured in the MetHcy+ media. Consistently, the PIK3CA mutant cells underwent a significantly greater degree of cell death in the MetHcy+ media, as measured by a propidium iodide (PI) assay (Fig. 2B) and by immunoblotting for cleaved poly[ADP (adenosine 5′-diphosphate)-ribose] polymerase (PARP) (Fig. 2C). Together, these results indicate that oncogenic PIK3CA is sufficient to confer methionine dependency to mammary epithelial cells.

Table 1 PIK3CA and PTEN mutational status of breast cancer cell lines studied.

N.D., not determined; WT, wild-type.

View this table:
Fig. 2 Oncogenic PIK3CA confers methionine dependency to MCF10A cells.

(A) Cells were grown in MetHcy+ media for 4 days, and the proliferation of cells was determined using the SRB assay (n = 3 biologically independent replicates). (B) Cells were grown in Met+Hcy or MetHcy+ media for 2 days, and cell death was measured using a propidium iodide (PI)–based plate-reader assay (n = 3 biologically independent replicates). (C) Cells were grown in Met+Hcy or MetHcy+ media for 2 days and were then immunoblotted for the indicated proteins (data are representative of three independent experiments). PARP, poly(ADP-ribose) polymerase. All error bars represent SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 by a two-sided Student’s t test.

Oncogenic PIK3CA inhibits cystine uptake through xCT

We reasoned that oncogenic PIK3CA may rewire cellular metabolism to confer methionine dependency, perhaps by altering the expression of genes involved in methionine and cysteine metabolism. To identify these candidate genes, we used published RNA expression data (36) to correlate the expression of a selected set of methionine and cysteine metabolism genes (listed in the Materials and Methods) to the MetHcy+ growth rates of our panel of cell lines (data file S1). Within this analysis, two genes were significantly correlated: Increased methionine dependency was associated with high MAT1A (which encodes methionine adenosyltransferase 1A) expression and low SLC7A11 (which encodes solute carrier family 7 member 11) expression (Fig. 3A and fig. S2, A and B). Next, we asked whether these transcriptional changes were also observed in the methionine-dependent PIK3CA mutant MCF10A cells. SLC7A11 expression was significantly lower in the PIK3CA mutant cells, whereas there was no significant difference in MAT1A expression (Fig. 3B). Consistently, SLC7A11 expression trended toward lower expression in the methionine dependent group C breast cancer cell lines, relative to the group A cell lines (Fig. 3C). Immunoblotting for endogenous SLC7A11 in two cell lines each from groups A to C revealed that the group C cell lines MDA-MB-468 and MCF7 had lower SLC7A11 protein abundance compared to the group A cell lines HCC38 and SUM-149 (Fig. 3D). SLC7A11, also known as xCT, is a cystine-glutamate antiporter that exchanges intracellular glutamate to take up extracellular cystine (Cys2), which is the oxidized form of Cys (37). Consistent with reduced xCT expression, PIK3CA mutant cells exhibited a significantly decreased rate of Cys2 uptake (Fig. 3E).

Fig. 3 Decreased xCT expression and cystine uptake is associated with increased methionine dependency.

(A) Spearman correlation of the growth rates of the panel of breast cancer cell lines in MetHcy+ media from Fig. 1C with their expression of a selected set of methionine and cysteine metabolism genes (see Materials and Methods for full list). Values above the red dashed line represent P < 0.05. (B) SLC7A11 and MAT1A mRNA abundances were measured by quantitative real-time polymerase chain reaction (qRT-PCR) and are expressed as fold changes relative to MCF10A empty vector–transfected (EV) cells (n = 3 biologically independent replicates). (C) SLC7A11 mRNA abundance was measured by qRT-PCR (n = 3 biologically independent replicates). (D) Cells were grown in Met+Hcy media and immunoblotted for the indicated proteins (data are representative of two independent experiments). (E) [14C]-cystine uptake was measured over 5 min and is expressed as fold changes relative to MCF10A EV cells (n = 5 biologically independent replicates). All error bars represent SEM. **P < 0.01 by a two-sided Student’s t test.

In addition to transcriptional regulation, we also noted that multiple mass spectrometry–based studies have identified xCT to be phosphorylated on Ser26, and the amino acid sequence surrounding this phospho-site is similar to the AKT substrate motif RxRxxS/T, with an Arg at the −3 position (Fig. 4A; www.phosphosite.org). To determine whether xCT Ser26 was phosphorylated downstream of phosphoinositide 3-kinase (PI3K), we stably expressed both WT and the S26A mutant of a C-terminally hemagglutinin (HA)-FLAG–tagged xCT in MCF10A WT or PIK3CA(H1047R) cells. We then serum-starved these cells, and by immunoprecipitating xCT with an HA antibody and immunoblotting with an antibody recognizing the p-RxxS/T phospho-motif, we found that xCT was constitutively phosphorylated in the PIK3CA(H1047R) cells. The S26A mutation abolished phosphorylation, indicating that xCT was phosphorylated at Ser26 downstream of the PI3K pathway (Fig. 4B). We also stimulated serum-starved WT MCF10A cells expressing xCT-HA-FLAG with insulin to activate PI3K. xCT phosphorylation was stimulated by insulin within 5 to 10 min and was attenuated by 30 to 60 min (Fig. 4C). The kinetics of xCT phosphorylation was similar to that of a canonical AKT substrate, PRAS40 (Fig. 4C), suggesting that AKT may be the kinase responsible for xCT phosphorylation. Phosphorylation of xCT in MCF10A PIK3CA(H1047R) mutant cells was abolished by the PI3K inhibitor GDC-0941, the AKT inhibitors GDC-0068 and MK2206, and the mechanistic target of rapamycin (mTOR) catalytic inhibitor Torin1. By contrast, the mTORC1/S6K inhibitor rapamycin did not affect xCT phosphorylation (Fig. 4D). Moreover, in MCF10A cells, we detected phosphorylation of endogenous xCT upon insulin stimulation, which was inhibited by both PI3K and AKT inhibitors (fig. S3). These results suggest that xCT is phosphorylated either by AKT or a kinase downstream of AKT that is not mTORC1. To further confirm that AKT was the responsible kinase, we conducted in vitro kinase assays with immunoprecipitated xCT and recombinant AKT1 (Fig. 4E). In vitro assays indicated that AKT1 phosphorylated xCT, but not xCT(S26A), demonstrating that AKT can directly phosphorylate xCT at Ser26 in vitro. Finally, to determine the functional consequence of xCT Ser26 phosphorylation, we measured glutamate secretion to assay xCT activity. We found that MCF10A cells expressing xCT(S26A) secreted more glutamate than cells expressing WT xCT (Fig. 4F), suggesting that phosphorylation of xCT at Ser26 inhibits its cystine transport activity. Together, the results above support a model in which oncogenic PIK3CA inhibits xCT activity through two mechanisms: (i) transcriptional inhibition of SLC7A11 expression and (ii) AKT-mediated phosphorylation of xCT at Ser26.

Fig. 4 xCT Ser26 is phosphorylated upon PI3K pathway activation.

(A) Comparison of the consensus AKT substrate motif with the amino acid sequence surrounding xCT Ser26. (B) MCF10A PIK3CA(+/+) and PIK3CA(H1047R/+) knock-in cells expressing EV, xCT–hemagglutinin (HA)-FLAG, or xCT(S26A)–HA-FLAG were serum-starved. HA immunoprecipitates from cell lysates were immunoblotted for the indicated proteins (data are representative of three independent experiments). (C) Serum-starved MCF10A cells expressing EV or xCT–HA-FLAG were treated with 100 nM insulin for the indicated times. HA immunoprecipitates from cell lysates were immunoblotted for the indicated proteins (data are representative of three independent experiments). (D) Serum-starved MCF10A PIK3CA(+/+) and PIK3CA(H1047R/+) knock-in cells were treated with vehicle, 1 μM GDC-0941, 1 μM GDC-0068, 1 μM MK2206, 1 μM Torin 1, or 100 nM rapamycin for 15 min. HA immunoprecipitates from cell lysates were immunoblotted for the indicated proteins (data are representative of three independent experiments). (E) xCT–HA-FLAG or xCT(S26A)–HA-FLAG were immunoprecipitated from human embryonic kidney–293T cells treated with 1 μM MK2206 and incubated with active glutathione S-transferase (GST)–AKT1 for an in vitro kinase assay. Lysates were immunoblotted for the indicated proteins (data are representative of three independent experiments). (F) Release of glutamate into the media by the indicated cell lines was measured over 24 hours (n = 3 biologically independent replicates). IP, immunoprecipitation; IB, immunoblot; WCL, whole cell lysate; p, phospho. All error bars represent SEM. *P < 0.05 by a two-sided Student’s t test.

xCT functionally contributes to the methionine dependency phenotype

The above results indicated that oncogenic PIK3CA decreased Cys2 uptake through both the transcriptional and posttranslational inhibition of xCT, and this metabolic phenotype was correlated with increased methionine dependency. On the basis of this model, we next explored the potential role of the inhibition of xCT and Cys2 uptake in contributing to the methionine dependency induced by oncogenic PIK3CA.

We hypothesized that xCT overexpression may reverse methionine dependency because PIK3CA mutant MCF10A cells expressed lower amounts of xCT. We found that MCF10A PIK3CA(H1047R) cells overexpressing xCT proliferated in both normal media containing Met and in MetHcy+ media (Fig. 5A and fig. S4A). Similarly, xCT overexpression restored the viability of these PIK3CA mutant cells in MetHcy+ media, as measured with a PI-based cell death assay (Fig. 5B). Together, these results indicate that at least in the context of MCF10A cells, xCT is a mediator of the methionine dependency induced by oncogenic PIK3CA, because its expression was sufficient to rescue the methionine dependency of the PIK3CA mutant cells.

Fig. 5 xCT and cystine uptake functionally contribute to the methionine dependency phenotype.

(A) Cells were grown in MetHcy+ medium for 4 days, and the proliferation of cells was determined using the SRB assay (n = 3 biologically independent replicates). (B) Cells were grown in Met+Hcy or MetHcy+ media for 2 days, and cell death was measured using a PI-based plate-reader assay (n = 3 biologically independent replicates). (C and D) Cells were grown in (C) MetHcy+, Cys2Hcy+, or Met Cys2Hcy+ media or (D) MetHcy+ media, with or without 1 mM sulfasalazine (SSA), for 4 days, and the proliferation of cells was determined using the SRB assay (n = 3 to 4 biologically independent replicates). (E) Cells were grown in MetHcy+ media for 6 days, and the proliferation of cells was determined using the SRB assay (n = 3 biologically independent replicates). All error bars represent SEM. *P < 0.05 by a two-sided Student’s t test. #P < 0.05 by a paired t test.

We also assessed whether manipulation of xCT expression and activity affects the methionine dependency of our panel of cell lines (Fig. 1B). Because group A cells, which proliferated in MetHcy+ media, expressed higher amounts of xCT (Fig. 3, A and C), then inhibition of Cys2 uptake in these cells should confer the methionine dependency phenotype. We found that removal of Cys2 from the growth media to inhibit Cys2 uptake led to proliferation of each of the group A cell lines in either MetHcy+ media or Cys2Hcy+ media. However, if both Met and Cys2 were removed (MetCys2Hcy+ media), then the absence of Cys2 conferred methionine dependency to these cells (Fig. 5C). Alternatively, inhibition of xCT with sulfasalazine (SSA) (38) consistently impaired the ability of group A cells to grow in the MetHcy+ media (Fig. 5D). Short hairpin RNA–mediated depletion of xCT also impaired growth in MetHcy+ media, especially for the highly MetHcy+ resistant HCC70 cell line (fig. S4, B to D). The proliferation of SUM149 cells in MetHcy+ media was only modestly impaired by xCT depletion (fig. S4, B to D), consistent with the observation that of the three group A cell lines, growth of SUM149 cells in MetHcy+ media is least affected by Cys2 deprivation or SSA (Fig. 5, C and D). Together, these results suggest that Cys2 uptake is required for the group A cell lines to regulate Hcy utilization in a manner that permits proliferation in the MetHcy+ media.

Conversely, the methionine-dependent group C cell lines expressed lower amounts of xCT, and we hypothesized that xCT overexpression may reverse their methionine dependency. Although the rescue was not as robust as in MCF10A cells (Fig. 5, A and B), xCT expression did modestly improve the survival of the group C cells in MetHcy+ media, particularly for T47D cells, and to a lesser extent, for MCF7 cells (although this rescue was reproducible, it was small and not statistically significant) (Fig. 5E). This result suggests that although xCT may be required for the proliferation of group A cells in MetHcy+ media, xCT expression was only partially sufficient in reversing the methionine dependency of two of the three group C cells. Therefore, although xCT may partially contribute to methionine dependency in the context of these two cell lines, it is likely that other factors are also involved. Overall, these analyses in the group A and C cell lines suggest that xCT is at least one of the mediators of the methionine dependency phenotype because manipulation of its activity or expression modulated the degree to which cells can grow in MetHcy+ media.

Oncogenic PIK3CA alters methionine utilization through the transsulfuration pathway

Finally, we asked whether oncogenic PIK3CA alters methionine metabolism and the ability of cells to use Hcy to either regenerate Met through the methionine cycle or synthesize Cys through the transsulfuration pathway. We hypothesized that because oncogenic PIK3CA decreased Cys2 uptake in MCF10A cells (Fig. 3E), PIK3CA mutant MCF10A cells may compensate by synthesizing Cys from Hcy through the transsulfuration pathway. Consistently, we observed increased steady-state pools of the transsulfuration pathway intermediate cystathionine in MCF10A PIK3CA(H1047R) cells relative to control cells (Fig. 6A). We next labeled cells with [U-13C5]-methionine, which results in either (i) [M+4] cystathionine if the methionine carbon enters the transsulfuration pathway or (ii) [M+4] methionine if the methionine carbon cycles through the methionine cycle once (Fig. 6B). After 1 hour of labeling, a time point at which steady-state labeling had not been reached, we found greater incorporation of the 13C label into [M+4] cystathionine in the PIK3CA mutant MCF10A cells relative to empty vector–transfected cells (Fig. 6C), which together with increased steady-state cystathionine pool sizes (Fig. 6A), suggests an increased flux through the transsulfuration pathway. We then provided these cells with [U-13C5]-methionine for 24 hours to reach steady-state labeling in the presence or absence of the PI3K inhibitor GDC-0941. Most of the detected cystathionine was labeled at steady-state in both the empty vector–transfected and PIK3CA(H1047R) cells, and the fractional contribution of [U-13C5]-methionine to [M+4] cystathionine was inhibited by GDC-0941, indicating that transsulfuration pathway activity depended on PI3K signaling (Fig. 6D). Furthermore, concurrent with a decrease in [M+4] cystathionine induced by GDC-0941, PI3K inhibition also increased labeling in [M+4] methionine, suggesting that activation of PI3K directs the utilization of Met through Hcy toward the transsulfuration pathway and away from remethylation through the methionine cycle to regenerate Met (Fig. 6E). Notably, inhibition of [M+4] cystathionine by GDC-0941 was more robust in the empty vector–transfected cells relative to the PIK3CA(H1047R)-expressing cells (Fig. 6, D and E), which reflected the stronger inhibition of PI3K signaling by GDC-0941 in the empty vector–transfected cells under these conditions (fig. S5). This result therefore further suggests a correlation between PI3K activity and conversion of Hcy to cystathionine. Together, these results are consistent with a model in which cells with oncogenic PIK3CA cannot proliferate in MetHcy+ media because the supplemented Hcy is used for transsulfuration and is not efficiently remethylated to produce Met.

Fig. 6 Oncogenic PIK3CA alters methionine utilization through the transsulfuration pathway.

(A) Steady-state total pool sizes of cystathionine measured by liquid chromatography–tandem mass spectrometry (LC-MS/MS) in MCF10A cells growing in standard culture media (n = 3 biologically independent replicates). (B) Schematic of [U-13C5]-methionine labeling into the methionine cycle ([M+4] methionine) or the transsulfuration pathway ([M+4] cystathionine), which are detected by LC-MS/MS. (C) Total ion counts of [M+4] cystathionine in MCF10A cells labeled with [U-13C5]-methionine for 1 hour (n = 3 biologically independent replicates). (D) Fractional labeling of [M+4] cystathionine in MCF10A cells labeled with [U-13C5]-methionine for 24 hours and treated with or without 1 μM GDC-0941 (n = 3 biologically independent replicates). (E) Total ion counts of [M+4] cystathionine and [M+4] methionine in MCF10A cells labeled with [U-13C5]-methionine for 24 hours and treated with or without 1 μM GDC-0941 (n = 3 biologically independent replicates). All error bars represent SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 by a two-sided Student’s t test.

DISCUSSION

Here, we demonstrated that in breast cancer cell lines, oncogenic PIK3CA was associated with methionine dependency, a cancer-associated metabolic dependency in which cells cannot proliferate in MetHcy+ media. In mammary epithelial MCF10A cells, oncogenic PIK3CA was sufficient to confer methionine dependency through a PI3K-induced decrease in Cys2 uptake mediated by both transcriptional and posttranslational inhibition of the Cys2 transporter xCT. xCT at least partially mediated the methionine dependency phenotype because manipulation of xCT activity or expression modulated the methionine dependency of breast cancer cells. Together, our data provide evidence to link oncogenic PIK3CA and xCT-mediated Cys2 uptake with methionine dependency, a mechanistically poorly understood cancer-associated metabolic dependency.

The mechanism by which oncogenic PIK3CA decreases xCT expression remains an open question. We have previously reported that expression of an oncogenic AKT mutation in MCF10A cells also decreases xCT expression (39). The transcription factor Nrf1 has been reported to act as a repressor of xCT expression (40), and mTORC1 is an activator of Nrf1 (41). Because mTORC1 signals downstream of PI3K, one potential hypothesis is that oncogenic PIK3CA may decrease xCT expression through the mTORC1-mediated activation of Nrf1.

Notably, posttranslational regulation of xCT downstream of PI3K resulting in phosphorylation of Ser26 and inhibition of Cys2 uptake has been previously reported (42), and as such, is consistent with the results presented here. Gu et al. identified mTORC2, but not AKT, as the kinase responsible for phosphorylation of xCT at Ser26 in cells. However, consistent with our data, they also showed that AKT can potentially phosphorylate xCT with in vitro kinase assays (42). The motif surrounding Ser26 in xCT conforms only partially to both the mTOR (43, 44) and the AKT consensus motifs (45, 46), and the responsible kinase in vivo may therefore conceivably depend on distinct cellular contexts. In our system, xCT phosphorylation was robustly inhibited by the AKT inhibitors GDC-0068 and MK2206 but not the mTORC1 inhibitor rapamycin, suggesting that xCT Ser26 may be phosphorylated by either AKT or a downstream kinase that is not mTORC1. However, an alternative explanation that could reconcile our results with those of Gu et al. (42) is the positive feedback loop from AKT to mTORC2 (47). If this feedback loop is functional in the cells used in our study, then AKT inhibitors would impair activation of mTORC2 by AKT, leaving open the possibility that mTORC2 could be the responsible kinase for xCT phosphorylation, at least under these conditions. Regardless, because both AKT and mTOR catalytic inhibitors impair xCT phosphorylation, our data indicate that both mTORC2 and AKT are equally required for xCT phosphorylation.

xCT-mediated Cys2 uptake and the transsulfuration pathway constitute two main mechanisms by which a cell can obtain Cys. Our [U-13C5]-methionine tracing data suggested that concurrent with decreased Cys2 uptake through xCT, PIK3CA mutant MCF10A cells used Met-derived Hcy for synthesizing Cys through the transsulfuration pathway, as opposed to remethylating Hcy to regenerate Met. These results are consistent with a model in which the ability of a cell to take up exogenous sources of Cys through xCT may influence whether that cell synthesizes Cys intracellularly through Hcy transsulfuration. In turn, the degree to which a cell uses the transsulfuration pathway for Cys may then affect the regulation of the Hcy junction by modulating how efficiently Hcy is remethylated through the methionine cycle to produce Met. Therefore, cells with more transsulfuration pathway activity would be more methionine-dependent, because less Hcy flux is devoted toward the methionine cycle to generate sufficient Met to continue proliferating in MetHcy+ media.

In support of this model, group A cells, including WT MCF10A cells, expressed higher amounts of xCT and may therefore use the transsulfuration pathway less to obtain Cys. As a result, in MetHcy+ media, Hcy can be devoted toward Met production through the methionine cycle to restore sufficient Met levels for proliferation. Conversely, in group C cells, including PIK3CA mutant MCF10A cells, Cys2 uptake was decreased through inhibition of xCT expression and activity by oncogenic PI3K. These cells therefore used the transsulfuration pathway to a greater extent to obtain Cys, resulting in less Hcy-to-Met conversion, impaired proliferation, and cell death in MetHcy+ media. Notably, MDA-MB-468 cells, a group C cell line, exhibit increased transsulfuration pathway activity in MetHcy+ media, which is consistent with our hypothesis (48). Finally, manipulation of xCT activity or expression may modulate the methionine dependency phenotype by altering the dependence of cells on transsulfuration-derived Cys, thereby influencing the relative diversion of Hcy into the methionine cycle compared to the transsulfuration pathway.

Although xCT contributes to methionine dependency, it is unlikely to be the only functional mediator of this phenotype. In the context of the group C breast cancer cell lines, xCT overexpression was only partially sufficient in reversing their methionine dependency. Moreover, it is important to note that although our data provides evidence for oncogenic PIK3CA mutations promoting methionine dependency, the correlation between methionine dependency and alterations within various nodes of the PI3K pathway, in general, was not as strong. For example, in our analysis of our panel of breast cancer cell lines, although genomic mutations in both PIK3CA and PTEN clustered toward cell lines with negative MetHcy+ growth rates, all of the group A cell lines are also negative for PTEN protein abundance. On one hand, this observation may reflect the idea that genetic alterations at different nodes of the PI3K pathway may not function equivalently, as has been demonstrated for other phenotypes such as drug sensitivity and organoid growth (4951), and may provide a rationale for testing how PTEN alterations contribute to methionine dependency in an isogenic system. However, this observation also points toward the likelihood that other genes may more strongly modulate the capacity to proliferate in MetHcy+ media. Other methods, such as focused single-guide RNA or complementary DNA overexpression screens targeting genes involved in Met and Cys metabolism, may be used to identify other factors that modify the methionine dependency phenotype.

Note also that the transsulfuration pathway is restricted to certain normal tissues in an organism, such as the liver, pancreas, and kidney (52). However, this pathway is also active in various cancer cells, including breast cancer cells (48, 5355). Moreover, the transsulfuration pathway can serve as a source of Cys for glutathione synthesis, and it is possible that the regulation of this pathway by oncogenic PIK3CA may support the PI3K-mediated stimulation of glutathione synthesis (39). It would also be interesting to explore whether a subset of cancer cells with transsulfuration pathway activity depend on this pathway to maintain their cellular Cys pools for growth and survival. If so, the transsulfuration pathway could be a potential therapeutic target for this subset of cancer cells.

In summary, we have identified a junction in metabolism that appears to be regulated differently in a subset of breast cancer cells: The metabolism of Hcy in either the methionine cycle to produce Met or in the transsulfuration pathway to synthesize Cys. The utilization of Hcy through these two pathways, which is partly altered by oncogenic PIK3CA and the degree to which cells depend on Cys2 uptake through xCT, then contributes to the methionine dependency phenotype. This work provides another example of the role of the PI3K pathway in mediating metabolic reprogramming in various cancers (56). In addition, xCT alters the utilization of other nutrients by cancer cells. For example, xCT overexpression results in higher glutamate export, rendering cancer cells more sensitive to glucose starvation (57, 58). Increased xCT-dependent cystine uptake and glutamate release also leads to dependence by some cancer cell lines on glutamine anaplerosis for proliferation, which results in greater sensitivity to glutaminase inhibitors (59, 60). By showing that PI3K-mediated regulation of xCT also influences methionine and cysteine metabolism in breast cancer cells, our work adds to the body of evidence describing how xCT alters nutrient utilization by cancer cells to modulate their dependency on different nutrients for survival.

MATERIALS AND METHODS

Cell lines

Cell lines were obtained from the American Type Culture Collection and authenticated using short tandem repeat profiling. MCF10A cells with a knock-in PIK3CA(H1047R) mutation were obtained from Horizon Discovery. No cell lines used in this study were found in the database of commonly misidentified cell lines, which is maintained by the International Cell Line Authentication Committee and National Center for Biotechnology Information BioSample. MCF10A cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM)/Ham’s F12 (CellGro) supplemented with 5% equine serum (CellGro), insulin (10 μg/ml) (Life Technologies), hydrocortisone (500 ng/ml) (Sigma-Aldrich), epidermal growth factor (EGF; 20 ng/ml) (R&D Systems), and cholera toxin (100 ng/ml) (Sigma-Aldrich). MDA-MB-231, MDA-MB-468, and MCF7 cells were maintained in DMEM (CellGro) supplemented with 10% fetal bovine serum (FBS) (Gemini). T47D and BT549 cells were maintained in RPMI 1640 (CellGro) supplemented with 10% FBS (Gemini) and insulin (10 μg/ml) (Life Technologies). HCC38, HCC70, HCC1143, HCC1806, and ZR-75-1 cells were maintained in RPMI 1640 (CellGro) supplemented with 10% FBS (Gemini). SKBR3 cells were maintained in McCoy’s 5A (CellGro) supplemented with 10% FBS (Gemini). SUM149 and SUM159 cells were maintained in Ham’s F12 (CellGro) supplemented with insulin (5 μg/ml) (Life Technologies), hydrocortisone (1 μg/ml) (Sigma-Aldrich), and 5% FBS (Gemini). Cells were passaged for no more than 6 months and routinely assayed for mycoplasma contamination.

Met+Hcy and MetHcy+ media

DMEM lacking glutamine, pyruvate, methionine, and cystine was obtained from Gibco and was supplemented as necessary with 2 mM glutamine (Gibco), 0.2 mM l-methionine (Sigma-Aldrich), 0.2 mM l-cystine (Sigma-Aldrich), 0.1 mM l-homocystine (Sigma-Aldrich), and 10% dialyzed FBS (CellGro). For MCF10A cells, the medium was also supplemented with 5% dialyzed FBS (CellGro), insulin (10 μg/ml) (Life Technologies), hydrocortisone (500 ng/ml) (Sigma-Aldrich), EGF (20 ng/ml) (R&D Systems), and cholera toxin (100 ng/ml) (Sigma-Aldrich).

Growth factors and inhibitors

For insulin stimulation, cells were serum-starved for 20 to 24 hours and stimulated with 100 nM insulin (Life Technologies). Inhibitors were used as follows: 1 μM GDC-0941 (LC Laboratories), 1 μM GDC-0068 (Active Biochem), 1 μM MK2206 (Selleck Chemicals), 100 nM rapamycin (Cayman Chemicals), 1 μM Torin 1 (Tocris), and 1 mM SSA (Sigma-Aldrich).

Antibodies

All antibodies were purchased from Cell Signaling Technology: p-AKT Ser473 (#4060; 1:1000), p-AKT Thr308 (#2965; 1:1000), AKT (#4691; 1:1000), p-PRAS40 Thr246 (#2997; 1:1000), PRAS40 (#2691; 1:1000), PARP (#9542; 1:1000), p-S6 Ser240/244 (#2215; 1:1000), p-RxxS/T (#9614; 1:1000); HA-tag (#2367; 1:1000); xCT/SLC7A11 (#12691; 1:1000), β-Actin (#4970; 1:5000), and conformation specific mouse anti-rabbit immunoglobulin G horseradish peroxidase (HRP) conjugate (#5127; 1:3000).

Plasmids

pJP1520-HA-GFP, pJP1520-HA-PIK3CA WT, pJP1520-HA-PIK3CA(E545K), and pJP1520-HA-PIK3CA(H1047R) were gifts from A.S. Baldwin (61). pENTR223-SLC7A11 was obtained from the Dana-Farber/Harvard Cancer Center DNA Resource Core, and SLC7A11 was subcloned into the pHAGE-C-HA-FLAG lentiviral expression vector (W. Harper, Harvard Medical School). The SLC7A11 S26A mutation was generated by site-directed mutagenesis (Qiagen). pLKO.1 vector was used as empty vector control.

Immunoblotting and immunoprecipitation

Cells were washed with phosphate-buffered saline (PBS) at 4°C and lysed in radioimmunoprecipitation assay buffer ]1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 50 mM tris-HCl (pH 7.5), proteinase inhibitor cocktail, 50 nM calyculin, 1 mM sodium pyrophosphate, and 20 mM sodium fluoride] for 15 min at 4°C. Cell extracts were precleared by centrifugation at 14,000 rpm for 5 to 10 min at 4°C, and protein concentration was measured with the Bio-Rad DC protein assay. For HA-tag immunoprecipitations, lysates were incubated with 10 μl of HA beads (Thermo Fisher Scientific) for 3 hours. Beads were then washed four times with NETN buffer [20 mM tris-HCl (pH 8), 100 mM NaCl, 0.5% NP-40, and 1 mM EDTA] and one time with PBS before elution with SDS sample buffer. For endogenous xCT immunoprecipitations, lysates were collected as described above and incubated with anti-xCT primary antibody for 2 hours, followed by 2 more hours with 20 μl of protein A beads. Lysates were then resolved on acrylamide gels by SDS–polyacrylamide gel electrophoresis and transferred electrophoretically to nitrocellulose membrane (Bio-Rad) at 100 V for 90 min. The blots were blocked in tris-buffered saline (TBST) buffer [10 mM tris-HCl (pH 8), 150 mM NaCl, and 0.2% Tween 20] containing 5% (w/v) nonfat dry milk for 1 hour and then incubated with the specific primary antibody diluted in blocking buffer at 4°C overnight. Membranes were washed three times in TBST and incubated with HRP-conjugated secondary antibody for 1 hour at room temperature. For endogenous xCT immunoprecipitations, a conformation-specific secondary antibody conjugated to HRP (mouse anti-rabbit-HRP; Cell Signaling Technology, #5127) was used to facilitate the detection of endogenous xCT phosphorylation. Membranes were washed three times and developed using enhanced chemiluminescence substrate (EMD Millipore).

In vitro kinase assay

Human embryonic kidney–293T expressing empty vector, pHAGE-C-HA-FLAG xCT, or pHAGE-C-HA-FLAG xCT(S26A) were serum-starved for 20 hours and then treated with 1 μM MK2206 for 30 min before collection of lysates and HA-xCT immunoprecipitation, as described above. Immunoprecipitated samples were incubated at 30°C for 1 hour in the presence of kinase assay reaction buffer (Cell Signaling Technology, #9802), adenosine triphosphate (200 μM), and dithiothreitol (2mM), with or without 0.5 μg of active glutathione S-transferase (GST)–AKT1 (Sigma-Aldrich, #SRP5001). The reaction was stopped by adding 12 μl of 6× SDS loading buffer. Samples were resolved by Western blotting, as described above.

Proliferation assays

Cells were seeded into 24- or 96-well plates (Corning) at the following densities: MCF10A, 25,000 cells per well (24-well) and 3000 to 8000 cells per well (96-well); MDA-MB-468, 75,000 cells per well (24-well); MCF7, 75,000 cells per well (24-well); T47D, 50,000 cells per well (24-well); HCC38, 30,000 cells per well (24-well); HCC70, 75,000 cells per well (24-well); and SUM149, 50,000 cells per well (24-well). After 24 hours, the medium was changed to either Met+Hcy, MetHcy+, Cys2Hcy+, or MetCys2Hcy+ media every 2 days. Two hundred microliters of media were used per well on a 96-well plate, and 1 ml of media was used per well on a 24-well plate. Cell number was measured at the indicated time points using sulforhodamine B staining, as previously described (62).

PI viability assay

Cell viability was assayed with a PI-based plate-reader assay, as previously described (63). Briefly, cells in 96-well plates were treated with a final concentration of 30 μM PI for 60 min at 37°C. The initial fluorescence intensity was measured in a SpectraMax M5 (Molecular Devices) at 530-nm excitation/620-nm emission. Digitonin was then added to each well at a final concentration of 600 μM. After incubating for 30 min at 37°C, the final fluorescence intensity was measured. The fraction of dead cells was calculated by dividing the background-corrected initial fluorescence intensity by the final fluorescence intensity. Viability was calculated by (1 – fraction of dead cells).

Growth rate and gene expression correlation analysis

The proliferation data for each cell line in Met+Hcy or MetHcy+ medium were fit to an exponential curve to calculate growth rates and doubling times. For the correlation of MetHcy+ growth rates to the expression of Met and Cys metabolism genes, RNA expression data was obtained from (36) for the following genes: GOT2, MAT2A, ODC1, AHCYL1, AHCY, AMD1, MARS, SRM, DNMT1, SMS, MTRR, SAT1, MPST, MTR, CDO1, MTAP, MAT1A, CTH, BHMT, TRDMT1, MTHFR, GOT1, SHMT1, GNMT, AHCYL2, CBS, SHMT2, IL4I1, ADI1, ENOPH1, MAT2B, DNMT3A, APIP, MSRB2, MSRA, BHMT2, DNMT3L, DNMT3B, and SLC7A11. A Spearman rank correlation was then conducted between the MetHcy+ growth rates of the panel of cell lines with the expression of the listed genes.

Quantitative real-time polymerase chain reaction

Total RNA was isolated with the NucleoSpin RNA Plus (MACHEREY-NAGEL) according to the manufacturer’s protocol. Reverse transcription was performed using the TaqMan Reverse Transcription Kit (ThermoFisher Scientific). Quantitative real-time polymerase chain reaction (qRT-PCR) was performed using an ABI Prism 7700 sequence detector. SLC7A11 primer: sense, 5′-TGCTGGGCTGATTTATCTTCG-3′, antisense, 5′-GAAAGGGCAACCATGAAGAGG-3′; MAT1A primer: sense, 5′-ATGCTACCGACGAGACAGAG-3′, antisense, 5′-GATGACGATGGTGTGGATGC-3′; 18S primer: sense, 5′-GTAACCCGTTGAACCCCATT-3′, antisense, 5′-CCATCCAATCGGTAGTAGCG-3′. PCR was carried out in triplicate. Quantification of mRNA expression was calculated by the ΔCT method with 18S ribosomal RNA as the reference gene.

14C-Cystine uptake assay

Fresh medium was added to the cells 2 hours before the experiment. [14C]-Cystine (L-[3,3′-14C]-Cystine, 110 mCi/mmol; PerkinElmer) was then added to the media at 0.9 μM for 5 min at 37°C. Uptake was terminated by rinsing the cells two times with ice-cold PBS. Cells were then lysed with 400 μl of 0.2 M NaOH containing 1% SDS, and lysates were centrifuged at 13,000 rpm for 5 min at room temperature. Two hundred fifty microliters of the lysate was mixed with 5 ml of scintillation cocktail (Ultima Gold, PerkinElmer), and radioactivity was measured using a scintillation counter. The remaining lysate was used for protein determination by the Bio-Rad DC protein assay, and the scintillation count for each sample was normalized by protein concentration.

Glutamate release rate

MCF10A cells were seeded at 250,000 cells per well on a six-well plate. On the following day, the cells were washed twice with PBS, and 2 ml of Met+Hcy media were placed in each well. One milliliter of media was immediately removed and centrifuged at 3000 rpm, and the supernatant was frozen for analysis. Cell number on the plate was also determined using the sulforhodamine B assay. One day later, medium was harvested and cell number was determined in the same manner. Glutamate concentrations in the media were measured by gas chromatography–mass spectrometry (GC-MS). Ten microliters of media was added to 10 μl of isotopically labeled internal standards of amino acids (Cambridge Isotope Laboratory, MSK-A2-1.2), and this mixture was extracted in 300 μl of ice-cold high-performance liquid chromatography (HPLC)–grade methanol (Sigma-Aldrich), vortexed for 10 min, and centrifuged at 15,000 rpm for 10 min. Two hundred eighty microliters of the supernatant was dried under nitrogen gas, and the remaining pellet was analyzed by GC-MS after methoxamine–tert-butyldimethylsilyl derivatization, as previously described (59). To calculate the glutamate release rate (picomoles per cell per day), changes in glutamate concentration were normalized to the integrated growth curve of each culture (number of cell·days during which glutamate was released), which was calculated by using the cell numbers at the initial day and day 1 counts to fit an exponential growth equation.

LC-MS/MS metabolomic profiling

Cells were maintained in the indicated growth media, and fresh medium was added 2 hours before the experiment. For metabolite extraction, medium was aspirated, and 80% (v/v) methanol at dry ice temperatures was added. Cells and the metabolite-containing supernatants were collected, and the insoluble material in lysates was centrifuged at 4000g for 5 min. The resulting supernatant was evaporated using a refrigerated SpeedVac. Samples were resuspended using 20-μl HPLC grade water for MS. Ten microliters was injected and analyzed using a 5500 QTRAP hybrid triple quadrupole mass spectrometer (AB/Sciex) coupled to a Prominence UFLC HPLC system (Shimadzu) via selected reaction monitoring (SRM) of a total of 254 endogenous water soluble metabolites for steady-state analyses of samples. Some metabolites were targeted in both positive and negative ion mode for a total of 285 SRM transitions using positive/negative switching. Electrospray ionization voltage was +4900 V in positive ion mode and −4500 V in negative ion mode. The dwell time was 4 ms per SRM transition, and the total cycle time was 1.89 s. About 9 to 12 data points were acquired per detected metabolite. Samples were delivered to the MS via normal-phase chromatography using a 4.6-mm inside diameter × 10 cm Amide XBridge hydrophilic interaction liquid chromatography (HILIC) column (Waters Corp.) at 300 μl/min. Gradients were run starting from 85% buffer B (HPLC grade acetonitrile) to 42% B from 0 to 5 min; 42 to 0% B from 5 to 16 min; 0% B was held from 16 to 24 min; 0 to 85% B from 24 to 25 min; and 85% B was held for 7 min to re-equilibrate the column. Buffer A was composed of 20 mM ammonium hydroxide/20 mM ammonium acetate (pH 9.0) in water/acetonitrile (95:5). Peak areas from the total ion current for each metabolite SRM transition were integrated using MultiQuant v2.0 software (AB/Sciex). For 13C-labeled experiments, SRMs were created for expected 13C incorporation in various forms for targeted liquid chromatography–tandem mass spectrometry (LC-MS/MS). Data analysis was performed in MatLab. Metabolite total ion counts represent the normalized (by cellular protein content), integrated total ion current from a single SRM transition.

Isotope labeling

For labeling in MCF10A cells, DMEM lacking glutamine, pyruvate, methionine, and cystine were obtained from Gibco and supplemented with 2 mM glutamine (Gibco), 0.2 mM l-cystine (Sigma-Aldrich), 0.2 mM [U-13C5]-l-methionine (Cambridge Isotope Labs), 5% dialyzed FBS (CellGro), insulin (10 μg/ml) (Life Technologies), hydrocortisone (500 ng/ml) (Sigma-Aldrich), EGF (20 ng/ml) (R&D Systems), and cholera toxin (100 ng/ml) (Sigma-Aldrich). Fresh unlabeled medium, with or without 1 μM GDC-0941, was added to the cells 2 hours before labeling. Labeled medium, with or without 1 μM GDC-0941, was then added to the cells, and cellular metabolites were extracted as described above at 1 and 24 hours.

Statistics and reproducibility

Sample sizes and reproducibility for each figure are denoted in the figure legends. Unless otherwise noted, all Western blots are representative of at least three biologically independent experiments. Statistical significance between conditions was assessed by two-tailed Student’s t tests. All error bars represent SEM, and significance between conditions is denoted as *P < 0.05, **P < 0.01, and ***P < 0.001.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/10/510/eaao6604/DC1

Fig. S1. Characterization of MCF10A cells stably expressing PIK3CA variants.

Fig. S2. SLC7A11 and MAT1A expression correlate with methionine dependency in breast cancer cell lines.

Fig. S3. Endogenous xCT phosphorylation is inhibited by PI3K and AKT inhibitors.

Fig. S4. xCT functionally contributes to the methionine dependency phenotype.

Fig. S5. Inhibition of AKT signaling by GDC-0941 during metabolic labeling with [U-13C5]-methionine.

Data File S1. Spearman rank correlation between methionine dependency and expression of methionine and cysteine metabolism genes.

REFERENCES AND NOTES

Acknowledgments: We thank J. Brugge (Harvard Medical School), B. Manning (Harvard School of Public Health), K. Cichowski (Brigham and Women’s Hospital), I. Harris (Harvard Medical School), C. Lyssiotis (University of Michigan Medical School), and members of the Toker (Beth Israel Deaconess Medical Center), Cantley (Weill Cornell Medical College), and Vander Heiden (Massachusetts Institute of Technology) labs for suggestions; M. Vander Heiden, M. Yuan (Beth Israel Deaconess Medical Center), and S. Breitkopf (Beth Israel Deaconess Medical Center) for technical assistance with mass spectrometry; R. Arora (Beth Israel Deaconess Medical Center) for evaluation of statistical analyses, and C. Dibble (Beth Israel Deaconess Medical Center) for critical reagents. Funding: Research support was derived in part from NIH [R01CA200671 (A.T.)] and the Ludwig Center at Harvard (A.T.). E.C.L. was a predoctoral fellow of the NSF graduate research fellowship program (NSF DGE1144152) and is a Damon Runyon Fellow supported by the Damon Runyon Cancer Research Foundation (DRG-2299-17). Author contributions: E.C.L. and A.T. designed the study and interpreted the results. E.C.L performed all experiments unless otherwise noted. L.G. performed the xCT phosphorylation studies. R.C.G. performed a subset of the proliferation and Western blot assays. J.M.A. assisted with the LC-MS/MS metabolomic studies and data analysis. E.C.L. and A.T. wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article