Despite the high incidence of oncogenic mutations in PIK3CA, the gene encoding the catalytic subunit of PI3K, PI3K inhibitors have yielded little clinical benefit for breast cancer patients. Recent epidemiologic studies have suggested a therapeutic benefit from aspirin intake in cancers harboring oncogenic PIK3CA. Here, we show that mutant PIK3CA-expressing breast cancer cells have greater sensitivity to aspirin-mediated growth suppression than their wild-type counterparts. Aspirin decreased viability and anchorage-independent growth of mutant PIK3CA breast cancer cells independently of its effects on COX-2 and NF-κB. We ascribed the effects of aspirin to AMP-activated protein kinase (AMPK) activation, mTORC1 inhibition, and autophagy induction. In vivo, oncogenic PIK3CA-driven mouse mammary tumors treated daily with aspirin resulted in decreased tumor growth kinetics, whereas combination therapy of aspirin and a PI3K inhibitor further attenuated tumor growth. Our study supports the evaluation of aspirin and PI3K pathway inhibitors as a combination therapy for targeting breast cancer. Cancer Res; 77(3); 790–801. ©2016 AACR.
The PI3K signaling pathway plays a critical role in cell growth, survival, motility, and metabolism (1). Deregulated PI3K signaling is observed in numerous human pathophysiologies, including cancer. In breast cancer, somatic mutations in genes that encode proteins that activate, terminate, or transduce PI3K signaling are highly prevalent. Specifically, somatic mutations in PIK3CA, the gene encoding the catalytic subunit p110α, occur with a frequency of approximately 40% across all breast cancer molecular subtypes (2). The two most frequent mutations comprise single amino acid substitutions in two hotspot regions, His1047Arg and Gln545Lys (2). Expression of either of these PIK3CA mutants leads to elevated PI3K activity, downstream AKT activation, oncogenic transformation of mammary epithelial cells, and formation of heterogeneous mammary tumors in vivo (3, 4). Similarly, the lipid phosphatase, PTEN, which terminates PI3K signaling, is one of the most frequently mutated tumor suppressors in human cancers. Mutation or loss of at least one copy of PTEN occurs in approximately 50% of breast cancer patients, leading to hyperactivation of PI3K/AKT signaling (5). In addition, amplification and mutation of AKT genes have been identified in breast cancer, albeit with lower frequencies (6).
Given the frequency with which the PI3K/PTEN/AKT pathway is mutated in breast cancer, numerous small-molecule inhibitors have been developed as targeted therapy and are under clinical evaluation. These include pan- and p110 isoform-specific inhibitors, compounds that inhibit both PI3K and the downstream effector mTOR, and also pan-AKT inhibitors. To date, most of these inhibitors have shown limited efficacy in clinical trials due to dose-limiting toxicities as well as the emergence of drug resistance. However, it is likely that use of combination therapies that target both PI3K/PTEN/AKT and other key survival pathways may result in better therapeutic responses.
Aspirin (acetylsalicylic acid) is one of the most widely used NSAIDs. Its medicinal use for the treatment of pain, fever, and inflammatory ailment dates back to the time of Hippocrates (7). Aspirin is also widely used as an antiplatelet drug for the prevention of heart attacks and strokes (8). Recently, results from a number of observational and randomized clinical trials have suggested that regular use of aspirin reduces the risk of development and/or progression of several cancers, including breast cancer (9, 10). Although the effect of aspirin on breast cancer incidence remains poorly understood, recent observations from the Nurses' Health Study indicate that aspirin use is associated with a reduced risk of breast cancer distant recurrence and death (11). Additional independent observational studies have shown that aspirin use is associated with a significant improvement in survival for patients with mutant PIK3CA colorectal cancer but not for those with wild-type PIK3CA tumors (12, 13). Despite these observations, the molecular basis underlying the benefit of aspirin use in mutant PIK3CA cancers remains undefined.
Here, we evaluate the efficacy of aspirin either as a single agent or in combination with PI3K inhibitors in PI3K-driven breast cancer. We also investigate the mechanism by which aspirin may elicit a therapeutic effect in this disease.
Materials and Methods
Anti-p110α (#4249), anti-phospho-Akt Ser473 (#4060), anti-phospho-Akt Thr308 (#2965), anti-Akt (#4691), anti-phospho-Pras40 Thr246 (#2997), anti-Pras40 (#2691), anti-phospho-GSK3β Ser9 (#9336), anti-GSK3β (#9315), anti-β-actin (#4970), anti-phospho-IKKα/β Ser176/180 (#2697), anti-phospho-IκBα Ser32/36 (#9246), anti-IκBα (#9247), anti-phospho NF-κB p65 Ser536 (#3033), anti-NF-κB p65 (#8242), anti-AMPKα (#2532), anti-phospho-AMPKα Thr172 (#2535), anti-ACC (#3676), anti-phospho-ACC Ser79 (#3661), anti-S6K (#2708), anti-phospho-S6K Thr389 (#9205), anti-S6 (#2217), anti-phospho-S6 Ser240/244 (#5364), anti-4EBP1 (#9452), anti-phosho-4EBP1 Ser65 (#9451), and anti-TSC2 (#3990) were purchased from Cell Signaling Technology. Laminin V (#Z0097) and Ki67 (#M7240) were purchased from Dako. Horseradish peroxidase–conjugated anti-rabbit and anti-mouse immunoglobulin antibodies were purchased from Chemicon.
The IKKβ ATP-competitive inhibitor, Compound A, was a generous gift from the Baldwin Lab (Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC) and manufactured by Bayer Pharmaceuticals. Celecoxib (#S1261) was purchased from Selleckchem. BKM120 (#A-1108) and BYL719 (#A-1214) were purchased from Active Biochem. A769662 10 mg (#ab120335) was purchased from Abcam. Aspirin (#A2093), sodium salicylate (#A5376), and Bafilomycin A (#B1793) were purchased from Sigma Aldrich. Aspirin/salicylate was prepared as described previously (14). Briefly, aspirin was dissolved in 1 mol/L Tris-HCl (pH 7.5) to a stock concentration of 1 mol/L and final pH of 7.2. An equivalent volume of Tris-HCl (pH 7.2) was used as vehicle control.
JP1520-HA-PIK3CA-GFP, JP1520-HA-PIK3CA-WT (Addgene plasmid # 14570), and JP1520-HA-PIK3CA-HA-H1047R (Addgene plasmid # 14572) were generous gifts from Joan Brugge (Harvard Medical School, Boston, MA). pBABE-puro mCherry-EGFP-LC3B was a gift from Jayanta Debnath (University of California, San Francisco, San Francisco, CA; Addgene plasmid # 22418).
Stable cell lines expressing COX-2 shRNA constructs were maintained in 2 μg/mL puromycin. COX-2 shRNA plasmids were a kind gift from the Polyak lab (Dana Farber Cancer Center, Boston, MA; ref.15). RNAi sequences of shRNA clones used in study are as follows:
COX-2 ShRNA#1 GCAGATGAAATACCAGTCTTT
COX-2 ShRNA#2 CCATTCTCCTTGAAAGGACTT
For siRNA-mediated knockdown of TSC2 and AMP-activated protein kinase (AMPK) α1/α2, SMARTpool: ON-TARGETplus human TSC2 siRNA (#L-003029-00-0005) and SMARTpool: ON-TARGETplus human PRKAA1/AMPKα1 siRNA (L-005027-00-0005) and PRKAA2/AMPKα2 (L-005361-00-0005) were purchased from Dharmacon. Cells were transfected with each respective siRNA or control ON-TARGETplus Non-targeting Control Pool siRNA (Dharmacon) using Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher Scientific) according to the manufacturer's protocol.
Total RNA was isolated using the RNeasy Kit following the manufacturer's instructions (Qiagen). Reverse transcription was performed using QuantiTect Reverse Transcription Kit according to the manufacturer's instructions (Qiagen). qRT-PCR was performed using SYBR Green PCR Master Mix (Bio-Rad) and the ABI Prism 7900 sequence detector (Applied Biosystems). Quantification of COX-2 mRNA expression was calculated by the ΔΔCt method with GAPDH as reference. qRT-PCR primer sequences are listed below:
COX-2 forward 5′-TCAGCCATACAGCAAATCCTT-3′
COX-2 reverse 5′-GTGCACTGTGTTTGGAGTGG-3′
GAPDH forward 5′-GCAAATTCCATGGCACCGT-3′
GAPDH reverse 5′-TCGCCCCACTTGATTTTGGAGG-3′
Cell culture and immunoblotting
MCF10A cells were cultured in DMEM/Ham's F12 medium supplemented with 5% equine serum (Gibco BRL), 10 μg/mL insulin (Sigma-Aldrich), 500 ng/mL hydrocortisone (Sigma-Aldrich), 20 ng/mL EGF (R&D Systems), and 100 ng/mL cholera toxin (List Biological Laboratories). MCF10A cells expressing PIK3CA wild type (WT) and mutants were generated and grown as described previously (16, 17). Stable pools were generated by selection in 2 μg/mL puromycin. SUM159-PT cells were grown in Ham's F12 medium (CellGro) supplemented with 5% FBS (Cyclone), 5 μg/mL insulin (Sigma-Aldrich), and 1 μg/mL hydrocortisone (Sigma-Aldrich). Cells used for animal studies were cultured in HUMEC medium (Cell application, 815-500) with DMEM-F12 (CellGro, 10-090-CV; 1:1) and 5% FBS (Cyclone). MCF7, MDA-MB-468, and MDA-MB-231 cells were maintained in DMEM (CellGro) supplemented with 10% FBS (Cyclone). MCF10A cells were provided by Joan Brugge (Harvard Medical School, Boston, MA), and SUM159-PT cells were obtained from Kornelia Polyak (DFCI, Harvard Medical School). All other cells lines were purchased and authenticated from ATCC. All cell lines were obtained and passaged for fewer than 6 months and routinely tested for mycoplasma contamination in the years 2012 to 2016. For immunoblotting, cells were rinsed with PBS and lysed in RIPA buffer containing protease and phosphatase inhibitors. Lysates were resolved by SDS-PAGE and transferred by electrophoresis to nitrocellulose membrane (Bio-Rad), followed by immunoblotting.
MCF10A cells were grown in three-dimensional Matrigel cultures as described previously (16). Briefly, chambers slides were coated with growth factor–reduced Matrigel (BD Biosciences) and allowed to solidify for 30 minutes. Cells (3 × 103) suspended in assay media containing 2% Matrigel were overlayed on coated chamber slides. Assay medium contained DMEM/Ham's F12 (CellGro) supplemented with 2% equine serum (CellGro), 10 μg/mL insulin (Sigma-Aldrich), 500 ng/mL hydrocortisone (Sigma-Aldrich), 5 ng/mL EGF (R&D Systems), and 100 ng/mL cholera toxin (List Biological Laboratories). For aspirin studies, acini were allowed to grow for 4 days, followed by treatment with aspirin every 2 days. Cells were then fixed and stained with Ki67 and laminin V on day 12 as described previously (16). Phase-contrast images were acquired using the Nikon Eclipse Ti microscope. Fluorescent images were acquired using Zeiss LSM 510 Meta confocal microscope.
Colony formation in soft agar
For colony formation in soft agar, 5 × 104 cells were suspended in growth media containing aspirin or vehicle and 0.4% Noble agar. Cell suspension was plated on top of a solidified layer of 0.8% Noble agar also containing aspirin. Cells were fed with growth media containing aspirin every 4 days. After 15 days, colonies were stained with 1 mg/mL iodonitrotetrazolium chloride and quantified using MATLAB software (MathWorks).
Sulforhodamine B assay
Cell viability was assessed using sulforhodamine B (SRB) assay as described previously (18). Briefly, adherent cells were fixed with 12.5% (w/v) trichloroacetic acid for 1 hour at 4°C. Cells were then rinsed three times with water and stained with a solution of 0.5% (w/v) SRB in 1% acetic acid for at least 30 minutes at room temperature. Cells were then washed three times with 1% acetic acid and allowed to dry. SRB was dissolved in 10 mmol/L Tris (pH, 10.5). Absorbance of solubilized SRB was measured at 510 nm.
Female NSG (NOD.Cg-PrkdcscidIL2rgtm1Wjl/SzJ; stock no. 005557) mice ages 6 weeks, were purchased from The Jackson Laboratory. All experiments were conducted in accordance with regulations of the Children's Hospital Institutional Animal Care and Use Committee (protocol 12-11-2308R). Bilateral orthotopic injections of SUM159-PT breast cancer cells were performed by resuspending 2 × 106 cells 1:1 (v/v) in Matrigel (BD Biosciences) in serum-free DMEM/F12 (CellGro). Cells were injected into the fourth inguinal mammary fat pad. Five mice were used for each treatment group. Tumors were measured at least once a week using a vernier caliper, and tumor volume was calculated using the formula 0.5 (length × width2). Aspirin used for in vivo studies involving daily oral gavage administration was prepared as a suspension in 0.5% methyl cellulose (Sigma).
For drug combination studies, preparation of aspirin for in vivo administration was performed as described previously (19). For these experiments, mice were injected intraperitoneally with aspirin (Sigma, A2093-100G) at 100 mg/kg. BYL719 (10mg/kg) was prepared in 0.5% methyl cellulose (Sigma) and administered daily by oral gavage. A similar volume:weight ratio of vehicle (DMSO in PBS or 0.5% methyl cellulose) was administered to control animals. Tumor volumes were compared using Student t test with reported P values with correction for multiple hypothesis testing using FDR method.
In addition, to compare tumor volume of the different treatment groups at a single time point, we performed a tumor growth rate analysis based on the longitudinal tumor growth measurements. For this purpose, we performed a linear mixed effects model analysis (20) using R lme4 package (21). Because of the study design, we considered as fixed effects in the model, treatment type and the time points at which the tumor volume was measured. To account for possible random effects due to individual mouse characteristics and side of inoculation and pretreatment tumor volume, we also included these three variables as random effects in the model. Visual inspection of residuals plots did not reveal any obvious deviations from homoscedasticity or normality.
Aspirin selectively inhibits growth of mutant PIK3CA breast cancer cells
To determine whether aspirin has any effect on mutant PIK3CA breast cancer cells, we first examined the effects of aspirin on the immortalized nontumorigenic breast epithelial cell line MCF10A stably expressing the oncogenic hotspot PIK3CA mutations H1047R or E545K. We examined cell growth in three-dimensional Matrigel culture as a surrogate measure of in vivo tumorigenicity. In contrast to wild-type PIK3CA cells, mutant PIK3CA H1047R- and E545K-expressing cell populations both formed large multi-acinar structures (Fig. 1A) and displayed irregular deposition of the basement membrane marker laminin V (Fig. 1B), features of malignant transformation commonly observed with oncogenes such as HER2 and KRAS (16, 22). Aspirin treatment of both oncogenic mutant PIK3CA populations resulted in reversion to spheroid-like acinar structures and a localization of laminin V to the periphery of the acini, similar to that of untransformed parental MCF10A cells (Fig. 1B). The aspirin concentrations (1–5 mmol/L) used were consistent with measurements of aspirin in the serum of patients treated for chronic inflammatory diseases (23–25). Aspirin treatment also resulted in a concomitant decrease in the marker Ki67, and an overall reduction in average spheroid size, supporting an antiproliferative effect due to aspirin treatment (Fig. 1B).
A decrease in cell viability and stellate-shaped acinar formation was also observed in aspirin-treated SUM159-PT breast cancer cells that harbor an endogenous oncogenic PIK3CA mutation (H1047L; Fig. 1C). Aspirin also robustly inhibited anchorage-independent growth of the mutant PIK3CA breast cancer cell lines SUM159-PT (H1047L) and MCF-7 (E545K) in soft agar (Fig. 1D). Moreover, daily administration of aspirin resulted in a statistically significant decrease in tumor volume and growth kinetics in SUM159-PT orthotopic mouse xenografts (Fig. 1E).
Aspirin suppresses growth of mutant PIK3CA breast cancer cells in an NF-κB and COX-2–independent manner
To explore the mechanistic basis for the growth-inhibitory effects of aspirin in oncogenic PIK3CA breast cancer cells, we tested the contribution of COX-2 and IKKβ/NF-κB, two well-characterized targets of aspirin (25–29). As revealed by RT-PCR, MCF10A cells expressing PIK3CA H1047R displayed elevated levels of COX-2 mRNA (Supplementary Fig. S1A). Similar induction of COX-2 mRNA was observed in MDA-MB-231 breast cancer cells expressing PIK3CA H1047R (Supplementary Fig. S1A). Expression of mutant PIK3CA also led to elevated COX-2 protein (Supplementary Fig. S1B). Consistent with this observation, increased levels of PGE2α were observed in media from MCF10A cells expressing mutant PIK3CA (Supplementary Fig. S1C). Oncogenic PIK3CA also increased NF-κB activation as measured by phosphorylation of IKKα/β, IκB, and p65 (Supplementary Fig. S1B), in agreement with previous studies (17). This observation coincides with the role for NF-κB as a transcriptional regulator of PTGS2 (30–32).
In light of these data, we investigated whether the growth-inhibitory effects of aspirin on mutant PIK3CA breast cancer cells is attributable to NF-κB and/or COX-2 activity. However, we observed that treatment of mutant PIK3CA MCF10A cells or SUM159-PT breast cancer cells with Compound A (an IKKβ ATP-competitive inhibitor; ref. 33), or celecoxib (a COX-2–specific inhibitor), had no effect on cell growth (Fig. 2A). This finding is consistent with previous work showing that attenuation of canonical NF-κB signaling in MCF10A cells expressing mutant PIK3CA E545K or H1047R with either the IKKβ ATP-competitive inhibitor BAY-65-1942, or by ectopic expression of an IκBα superrepressor does not affect monolayer growth or colony formation in soft agar (17). Furthermore, shRNA-mediated suppression of COX-2 in SUM159-PT (Fig. 2B) cells did not affect cell proliferation (Fig. 2C). Collectively, these findings suggested that the inhibitory effects of aspirin on proliferation of breast cancer cells expressing oncogenic PIK3CA are unlikely to be mediated exclusively by induction of NF-κB or COX-2.
Aspirin activates AMPK and decreases mTORC1 signaling in mutant PIK3CA cells
Recently, aspirin was shown to target intracellular energy homeostasis and metabolism through direct activation of AMPK and subsequent inhibition of mTOR signaling (34, 35). AMPK and mTORC1 are major sensors of nutrient and growth factor signaling that regulate cell proliferation and size (36, 37). Thus, we investigated the contribution of AMPK and mTORC1 signaling to the growth defects induced by aspirin in mutant PIK3CA cells.
Expression of mutant PIK3CA E545K and H1047R in MCF10A cells resulted in attenuation of AMPK activity relative to cells expressing GFP vector control, or wild-type PIK3CA, as measured by phosphorylation of AMPK (pT172) and its substrate acetyl-CoA carboxylase (ACC; pS79). This was concomitant with increased mTORC1 signaling, as analyzed by phosphorylation of S6K1 (pT389), 4EBP1 (pS65), and S6 (pS240/244), all of which were blocked by the PIK3CA-specific inhibitor BYL719 (Supplementary Fig. S2A). As expected, oncogenic PIK3CA also increased phosphorylation of AKT (pS473) and downstream substrates PRAS40 (pT246) and GSK-3β (pS9) relative to PIK3CA wild-type cells (Supplementary Fig. S2A). In summary, these data demonstrate that mutant PIK3CA antagonizes AMPK signaling but positively regulates mTORC1 activity.
Interestingly, in breast cancer cells harboring oncogenic PIK3CA, aspirin treatment as well as its metabolite salicylate, led to activation of AMPK coincident with inhibition of mTORC1 activity, in a dose-dependent manner (Fig. 3A; Supplementary Fig. S2B). This aspirin-induced attenuation of mTORC1 activity resulted in reactivation of AKT signaling (Fig. 3A), similar to that observed with chronic rapamycin treatment (38, 39). Importantly, inhibition of COX-2 and IKKβ/NF-κB did not significantly attenuate mTORC1 activity (Supplementary Fig. S2B). Therefore, aspirin-induced inhibition of mTORC1 occurred independently of COX-2 and IKKβ/NF-κB signaling. Interestingly, both celecoxib and Compound A modestly increased AMPK phosphorylation (pT172) relative to vehicle; however, this did not translate into inhibition of mTORC1 signaling (Supplementary Fig. S2B). The mechanism that accounts for this increase remains unclear. Importantly, treatment with the AMPK activator A769662 led to a dose-dependent decease in mTORC1 signaling as indicated by the decrease in phosphorylation of S6K1 (pT389), S6 (pS240/244), and 4EBP1 (pS65; Fig. 3B) relative to vehicle-treated cells, as well as a dose-dependent decrease in cell viability (Fig. 3C). Interestingly, knockdown of AMPKα1/α2 partially rescued this attenuation of mTORC1 signaling upon treatment with aspirin (Supplementary Fig. S2C). In support of a role for mTORC1 inhibition in growth suppression, we observed that the mTORC1 inhibitor rapamycin also decreased the growth of SUM159-PT cells (Fig. 3D). Furthermore, siRNA depletion of tuberous sclerosis complex 2 (TSC2), which results in an increase in phosphorylation of S6K1 (pT389), S6 (pS240/244), and 4EBP1 (pS65; Fig. 3E), partially rescued the suppression of mTORC1 signaling induced upon aspirin treatment (Fig. 3E). Taken together, these mechanistic studies demonstrate that the growth-inhibitory effect of aspirin can be ascribed, at least in part, to enhanced activation of AMPK and inhibition of mTORC1 signaling. They also provide a rationale for the increased sensitivity of mutant PIK3CA cells to aspirin treatment, given the importance of AMPK/mTORC1 signaling in these cells.
Aspirin sensitizes mutant PIK3CA breast cancer cells to PI3K inhibitors
Given that aspirin treatment resulted in reactivation of AKT signaling, we hypothesized that aspirin, by modulating PI3K/mTORC1 signaling, might sensitize or prime cells for PI3K inhibition. Therefore, we investigated the ability of aspirin to attenuate the growth of PIK3CA-mutant breast cancer cell lines in combination with PI3K inhibitors. Suboptimal levels of aspirin were used to reveal any enhanced effects due to the combination of aspirin with various PI3K inhibitors. In contrast to GFP- or PIK3CA wild-type expressing cells, MCF10A cells expressing oncogenic PIK3CA (H1047R) displayed a significant decrease in the number of viable cells following treatment of aspirin in combination with BYL719, relative to either drug alone (Fig. 4A; Supplementary Fig. S3). Similarly, in SUM159-PT cells (PIK3CA H1047L), enhanced growth inhibition was observed upon cotreatment with aspirin and BYL719, compared with each drug alone (Fig. 4B).
In contrast, the combination of aspirin and BYL719 showed little additive effect in MDA-MB-468 cells, a PTEN-mutant breast cancer line (Fig. 4C). However, cotreatment of aspirin and BKM120, a pan–class I PI3K inhibitor, showed an additive effect (Fig. 4C) in these cells, consistent with the notion that PTEN-deficient tumors signal primarily through p110β (PIK3CB) and would therefore be insensitive to p110α-specific inhibitors (40). Importantly, MDA-MB-231 cells that do not harbor PI3K pathway mutations were insensitive to the combination of aspirin with either BYL719 or BKM120 (Fig. 4D).
Cotreatment of aspirin and PI3K inhibitors leads to enhanced activation of AMPK, inhibition of mTORC1, and induction of autophagy
Next, we investigated the molecular basis for growth suppression of PI3K pathway mutant breast cancer cells in response to dual treatment of aspirin and PI3K inhibitors. In terms of signaling mechanisms, the combination of aspirin with either PI3K inhibitors resulted in increased phosphorylation of AMPK (pT172) and ACC (pS79) along with decreased phosphorylation of S6K1 (pT389), S6 (pS240/244), and 4EBP1 (pS65) relative to either drug alone (Fig. 5A). Similar to aspirin, cotreatment of SUM159-PT cells with rapamycin and both PI3K inhibitors (BYL719 and BKM120) also led to a significant decrease in cell viability relative to either drug alone (Fig. 5B). This observation also held true for the AMPK activator A769662 (Fig. 5C).
AMPK and mTORC1 are critical regulators of autophagy, a catabolic process that facilitates organelle and protein recycling and/or degradation to maintain cellular homeostasis (41). Activation of autophagy can be tumor suppressive as demonstrated by loss-of-function studies of autophagy genes, such as Beclin1 (42). Early stages of autophagy involve recruitment and processing of cytosolic light chain LC3-I to membrane-associated LC3-II, whereas the latter stages involve maturation and fusion of the autophagosome to the lysosome (41). Given that cotreatment of aspirin and BYL719 led to activation of AMPK and suppression of mTORC1 signaling and that aspirin results in an increase in autophagy (Fig. 5D), we investigated the effect of this drug combination on autophagy. To assess activation of autophagy, we employed a mCherry-GFP-LC3B reporter system (43, 44). Upon activation of autophagy, cytosolic mCherry-GFP-LC3B is first recruited to autophagosomes and can be visualized as distinct puncta. Following this, autophagosomes undergo fusion with the lysosome to form the autolysosome. Because of the acidic environment of the autolysosomes, positive pH-insensitive mCherry fluorescence is mainly observed, while pH-sensitive GFP signal is diminished. Thus, formation of mCherry-LC3B–positive puncta can be used as a surrogate marker for activation of autophagy. We observed that treatment of SUM159-PT cells with a combination of aspirin and BYL719 resulted in an increase in LC3I/II total protein (Fig. 5D) as well as an increase in the number of fluorescently labeled mCherry-LC3B puncta, relative to each drug alone (Fig. 5E and F). Together, these observations demonstrate that aspirin in combination with BYL719 results in activation of autophagy.
We next investigated whether activation of autophagy was responsible for the decrease in cell viability observed upon cotreatment with aspirin and BYL719. To address this, we treated SUM159-PT cell with the autophagy inhibitor, Bafilomycin A. This autophagy inhibitor partially rescued the decrease in cell viability induced by cotreatment of aspirin and BYL719 (Fig. 5G).
Aspirin augments tumor growth suppression induced by BYL719
Given the pleiotropic effects of aspirin, we investigated the effects of aspirin in combination with BYL719 in SUM159-PT mouse xenografts. Tumor cells were implanted into NSG mice and allowed to grow for a period of 22 days, when average tumor volume attained approximately 120 to 160 mm3. Mice were then divided into four cohorts with comparable tumor size across all cohorts and treated with vehicle control, aspirin alone, BYL719 alone, or aspirin + BYL719 combination (Fig. 6A). Tumor-bearing mice were treated daily with BYL719 by oral gavage, while aspirin was administered by intraperitoneal injection, as indicated (Fig. 6A). At the end of the experiment, tumors of mice treated with BYL719 alone showed a significant lower volume compared with vehicle control (P = 0.03184). The volume reduction was even more pronounced in mice treated with BYL719 in combination with aspirin (BYL719 + aspirin vs. control: FDR P = 0.00224; Fig. 6B). No statistically significant changes were observed in tumor volume between mice treated with BYL719 alone versus the combination of BYL719 with aspirin (FDR P = not significant) at this single time point (day 33). However, based on our mixed effects model of tumor growth and, independently of the time point selected, we found a significant decrease in the tumor growth rate of approximately 0.21% (P = 0.02) with the addition of aspirin to BYL719. The combination treatment group also showed a decrease of 44% (P < 0.0001) of tumor growth rate relative to the vehicle control group. Notably, aspirin alone did not result in a reduction in tumor growth as previously observed in Fig. 1E. This is likely due to the lower frequency of aspirin administration, which was initially employed to minimize any toxicity or negative side effects from daily treatment of the above drug combination. This observation reinforces epidemiologic findings that highlight the importance of duration and dosage of aspirin used to obtain a chemotherapeutic benefit (11).
Although aspirin is primarily administered as an antipyretic and analgesic, numerous observational and randomized clinical trials have suggested a potential chemotherapeutic use in cancer patients (9, 10, 45). Two independent studies have demonstrated that aspirin use is associated with increased survival in colorectal cancer patients with mutant PIK3CA but not among those with the wild-type gene (12, 13). Despite these observations, the molecular basis for this phenomenon remains poorly understood. By contrast, there are few well-characterized studies on a potential therapeutic benefit from aspirin intake in breast cancer. However, somatic PIK3CA mutations are highly prevalent in breast cancer, and aspirin use is associated with a decrease in breast cancer mortality and distant recurrence (11).
In this study, we investigated whether adjuvant aspirin treatment could improve the efficacy of specific PIK3CA and PI3K inhibitors currently under clinical trial evaluation for the treatment of breast cancer. To date, many of these inhibitors display limited clinical efficacy. Our data indicate that aspirin decreases the growth and viability of mutant PIK3CA breast cancer cells both in vitro and in vivo. Cells expressing mutant PIK3CA are more sensitive to aspirin compared with cells expressing wild-type PIK3CA in vitro. Notably, relatively lower concentrations of aspirin robustly blocked the growth of PIK3CA-mutant cells grown in soft agar, suggesting that under conditions of stress, PIK3CA-mutant cells may be more sensitized to aspirin treatment.
Our findings demonstrate that the cell growth–inhibitory effects of aspirin on mutant PIK3CA cells are unlikely to be mediated exclusively by induction of NF-κB or COX-2. Instead, the aspirin-induced growth suppression phenotype can be ascribed to activation of AMPK and inhibition of mTORC1 signaling. Although COX-2 expression strongly associates with the benefit of aspirin intake in colorectal cancer, epidemiologic studies have shown that improved breast cancer survival associated with aspirin use is independent of COX-2 status (46). This is consistent with mechanistic studies, which demonstrated that the antitumor effects of aspirin are COX-2 independent and are mediated by other pathways, including Wnt/β-catenin signaling (47, 48).
Although our data point to COX-2 and NF-κB–independent mechanism(s) in the cell-autonomous effects of aspirin in PIK3CA-mutant breast cancer, we cannot completely rule out the contribution of these two aspirin targets in PI3K-driven breast cancer. In the context of the tumor microenvironment, attenuation of COX-2 activity and NF-κB signaling blocks cancer growth and metastasis via several mechanisms that target the immune system, platelet activation/angiogenesis, and other components of the tumor milieu (49–51). This notion is consistent with the finding that mutant PIK3CA activates IKKβ/NF-κB, which in turn plays a critical role in the regulation of inflammatory genes, including the PTGS2 gene, which encodes for COX-2. Therefore, the multiple direct targets of aspirin, including COX-2, NF-κB, as well as AMPK, likely explain the sensitivity of mutant PIK3CA cancers to this drug.
We also demonstrate that aspirin in combination with PI3K inhibitors results in enhanced growth suppression of PIK3CA/PTEN–mutant breast cancer cells. This growth suppression phenotype is due to enhanced AMPK activation and inhibition of mTORC1 signaling. We further show that mutant PIK3CA inhibits AMPK activity but activates mTORC1 signaling, most likely due to mutant PIK3CA–induced activation of AKT. AKT phosphorylation of AMPKα1 at pS487 has been shown to inhibit AMPKα1 activation and phosphorylation at pT172 (52). In the case of mTORC1, AKT activation promotes mTORC1 signaling through several mechanisms, including phosphorylation of TSC2 and PRAS40 (53, 54). Although activation of AMPK has been shown to inhibit mTORC1 signaling, future studies aimed at deciphering whether aspirin-induced inhibition of mTORC1 signaling is AMPK dependent are warranted.
Aspirin is a readily available and inexpensive drug already approved by regulatory agencies worldwide and with many potential chemotherapeutic properties. In contrast to other drugs being tested for cancer treatment, it has fewer severe side effects and due to its pleiotropic targets, may be effective at targeting both the primary tumor and distant metastases. Taken together, our findings provide a mechanistic rationale for the use of aspirin in combination with PI3K inhibitors for the treatment of breast cancers that show PI3K pathway dependency. It also highlights the importance of assessing whether PIK3CA mutation status would be a reliable biomarker for identifying breast cancer patients who are most likely to benefit from adjuvant aspirin therapy. Our study reinforces the need in performing clinical trials to evaluate the potential chemotherapeutic effects of aspirin in breast cancer patients, and whether any beneficial outcome outweighs side effects due to chronic aspirin intake, and ultimately how this ancient drug compares with other chemotherapeutic agents currently under clinical evaluation.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: W.S. Henry, A. Toker
Development of methodology: W.S. Henry, A. Toker
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Laszewski, T. Tsang, S.S. McAllister
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W.S. Henry, F. Beca, A.H. Beck, A. Toker
Writing, review, and/or revision of the manuscript: W.S. Henry, F. Beca, A.H. Beck, S.S. McAllister, A. Toker
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Toker
Study supervision: S.S. McAllister, A. Toker
We thank Drs. John Rinn, Carl Novina, Lewis Cantley, Michelle Holmes, and members of the Toker laboratory for productive discussions, Joan Brugge, Kornelia Polyak, Albert Baldwin, and Brendan Manning for providing reagents, and Amey Barakat for technical assistance.
This study was supported in part by grants from the NIH (CA177910 to A. Toker), DOD/CDMRP/BCRP Era of Hope Scholar Award (S.S. McAllister), and a Howard Hughes Medical Institute International Student predoctoral fellowship (W.S. Henry).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.