Activation of WNT/β‐catenin signaling results in resistance to a dual PI3K/mTOR inhibitor in colorectal cancer cells harboring PIK3CA mutations

Abstract

PIK3CA is a frequently mutated gene in cancer, including about ~15 to 20% of colorectal cancers (CRC). PIK3CA mutations lead to activation of the PI3K/AKT/mTOR signaling pathway, which plays pivotal roles in tumorigenesis. Here, we investigated the mechanism of resistance of PIK3CA‐mutant CRC cell lines to gedatolisib, a dual PI3K/mTOR inhibitor. Out of a panel of 29 CRC cell lines, we identified 7 harboring one or more PIK3CA mutations; of these, 5 and 2 were found to be sensitive and resistant to gedatolisib, respectively. Both of the gedatolisib‐resistant cell lines expressed high levels of active glycogen synthase kinase 3‐beta (GSK3β) and harbored the same frameshift mutation (c.465_466insC; H155fs*) in TCF7, which encodes a positive transcriptional regulator of the WNT/β‐catenin signaling pathway. Inhibition of GSK3β activity in gedatolisib‐resistant cells by siRNA‐mediated knockdown or treatment with a GSK3β‐specific inhibitor effectively reduced the activity of molecules downstream of mTOR and also decreased signaling through the WNT/β‐catenin pathway. Notably, GSK3β inhibition rendered the resistant cell lines sensitive to gedatolisib cytotoxicity, both in vitro and in a mouse xenograft model. Taken together, these data demonstrate that aberrant regulation of WNT/β‐catenin signaling and active GSK3β induced by the TCF7 frameshift mutation cause resistance to the dual PI3K/mTOR inhibitor gedatolisib. Cotreatment with GSK3β inhibitors may be a strategy to overcome the resistance of PIK3CA‐ and TCF7‐mutant CRC to PI3K/mTOR‐targeted therapies.

Short abstract

What's new?

Mutations in the PI3K/mTOR and WNT/β‐catenin pathways are common in colorectal cancer. Here, in colorectal cancer cells, the authors show that the frameshift mutation H155fs* in transcription factor TCF7 serves a critical role in mediating resistance to the PI3K/mTOR dual inhibitor gedatolisib. TCF7 H155fs* maintained mTOR signaling by inducing the active form of glycogen synthase kinase 3‐beta (GSK3β) and aberrant WNT/β‐catenin signaling. The latter conferred resistance to gedatolisib. Co‐treatment with a GSK3β inhibitor increased sensitivity to gedatolisib, a combination that synergistically inhibited colorectal tumor growth in mice. The findings shed light on gedatolisib resistance mechanisms and potential resistance biomarkers.

Abbreviations

CRC

colorectal cancers

GSK3β

Glycogen synthase kinase 3 beta

mTOR

mammalian target of rapamycin

PI3K

The phosphatidylinositol 3‐kinase

Introduction

The phosphatidylinositol 3‐kinase (PI3K), AKT serine/threonine kinase, mammalian target of rapamycin (PI3K/AKT/mTOR) signaling pathway is aberrantly activated in many types of cancer.1 For example, activating PIK3CA gene mutations affecting the PI3Kα catalytic subunit, such as H1047R, P449T, E545K and R88Q are detected in about ~15–20% of colorectal cancers (CRC).2, 3 Therefore, components of the PI3K/mTOR pathway have been considered attractive drug targets.4, 5 Most PI3K/mTOR inhibitors, such as dactolisib (BEZ235),6 Voxtalisib (XL765),7 and gedatolisib,8 are generally effective against solid tumors harboring PIK3CA mutations,9 whereas KRAS‐mutant tumors tend to be resistant.10 These inhibitors are often used as targeted therapies, but their mechanisms of action are uncertain and their effects are often limited by other mutations or compensatory mechanisms.11, 12 For these reasons, combination therapies that target other pathways are being investigated in an effort to overcome tumor resistance to dual PI3K/mTOR inhibitors.13, 14

Glycogen synthase kinase 3 beta (GSK3β) is a critical component of a degradative multiprotein destruction complex that includes adenomatous polyposis coli (APC), casein kinase 1 (CKI), AXIN2, and the key transcriptional coactivator for the WNT signaling pathway, β‐catenin.15 This complex acts as a negative regulator of WNT/β‐catenin signaling, which is frequently elevated in cancer.16 In addition to this role, GSK3β functions as a link between the PI3K/mTOR and WNT/β‐catenin pathways through AMPK. Although GSK3β is well known as a tumor suppressor, recent work suggests that it also functions in promoting the proliferation of many types of cancer cells.17, 18 In particular, GSK3β is thought to play both positive and negative roles in WNT/β‐catenin signaling, but the precise mechanisms have not yet been established.19

TCF7 is a transcription factor that mediates and positively regulates WNT/β‐catenin signaling by inducing transcription of target genes such as CCND1, AXIN2, TCF7, LEF1, MET, and WNT3A.20, 21 About 1–2% of CRC harbor a mutation in TCF7 that results in loss of the β‐catenin binding domain, which reduces its transcriptional activity and weakens WNT/β‐catenin signaling.22 However, several TCF7 frameshift mutations have also been reported in cancer, and current work suggests that they can increase or decrease transcription of WNT target genes.23

In our study, we explored the mechanism of resistance of CRC cells to the dual PI3K/mTOR inhibitor gedatolisib, focusing on the relationship between PI3K/mTOR, GSK3β, and WNT/β‐catenin signaling. We demonstrate a pivotal role for TCF7 H155fs* in increasing GSK3β activity and WNT/β‐catenin signaling and reveal their contribution to gedatolisib resistance in PIK3CA‐ and TCF7‐mutant CRC.

Material and Methods

Reagents

Gedatolisib (also known as PKI‐587 or PF‐05212384) was kindly provided by Pfizer (New York, NY). CHIR‐99021 (HCl) and SB216763 were purchased from Selleck Chemicals (Houston, TX). LiCl was purchased from Sigma Aldrich (St Louis, MO). Stock solutions were prepared in dimethyl sulfoxide or distilled water and stored at −20 °C.

Cell lines and cell culture

Human CRC cell lines, including HCT‐116, WiDr, HT‐29, SNU‐C5, DLD‐1, HCT‐15, and LS174T, were obtained from the Korean Cell Line Bank (Seoul, Korea).24 Cells were cultured in RPMI 1640 media containing 10% heat‐inactivated fetal bovine serum (Gibco, Carlsbad, CA) and gentamicin (10 μg/mL) at 37 °C in a 5% CO₂ atmosphere.

Cell viability assay

The viability of cells was assessed using MTT assays (Sigma‐Aldrich, St Louis, MO). A total of 2 × 10³ cells were seed in 96‐well plates, incubated for 24 h, and treated for 72 h with indicated drugs at 37 °C. after treatment, MTT solution was added to each well and incubated for 4 h at 37 °C. The medium was then removed, and dimethyl sulfoxide (DMSO) was added and mixed thoroughly for 10 min at room temperature. Cell viability was determined by measuring absorbance at 540 nm using a VersaMax microplate reader (Molecular Devices, Sunnyvale, CA). The concentration of drug required to inhibit cell growth by 50% was determined via interpolation from dose–response curves using Calcusyn software (Biosoft, Ferguson, MO). Six replicate wells were utilized for each analysis, and at least three independent experiments were conducted. The data from replicate wells are presented at the mean number of the remaining cells with 95% confidence intervals.

Western blotting antibodies

Cultured cells were washed with phosphate buffered saline (PBS) and lysed with lysis buffer (50 mM Tris–HCl, pH 7.5 1% NP‐40, 0.1% sodium deoxycholate, 150 mM NaCl, 50 mM NaF, 1 mM sodium pyrophosphate, 1 mM EDTA, and protease/phosphatase inhibitors). Lysates were clearly by centrifugation at 13,000 rpm for 20 min. Protein concentrations were quantified with a Bicinchoninic Acid Protein Assay Reagent (Pierce, Rockford, IL), according to manufacturer's instructions. Samples containing equal quantities of total proteins were resolved in SDS‐polyacrylamide denaturing gel, transferred to nitrocellulose membranes. The membranes were incubated in blocking buffer containing 1% skim milk and 1% bovine serum albumin(BSA) for an hour at room temperature and probed overnight at 4 °C with primary antibodies. Antibodies against PI3K p110, phosphorylated (p)‐AKT (pS473), p‐AKT (pThr308), p‐TSC2 (pThr1462), p‐mTOR (pS2481), p‐p70S6K (pThr389), p‐p70S6K (pS371), p‐S6K (pS240/244), p‐4E‐BP1 (pS65), p‐LRP6 (pS1490), p‐GSK3β (pS9/21), p‐β‐catenin, p‐GS, AXIN2, AKT, TSC1, TSC2, mTOR, Raptor, p70S6K, S6K, 4E‐BP1, GSK3β, TCF7, caspase‐3 and caspase‐7 were purchased from Cell Signaling Technology (Beverly, MA). Antibodies to p‐GSK3β (pTyr216), β‐tubulin, β‐catenin, and active β‐catenin were obtained from Santa Cruz Biotechnology (Dallas, TX). Quantitation of band intensity on western blots was performed with ImageJ software (National Institutes of Health, Bethesda, MD).

Cell cycle analysis and annexin V staining

Cells were treated for 24 h with drugs, washed twice in PBS, fixed in 70% ethanol, and then stored at −20 °C until analysis. Before analysis, the cell suspensions were washed with PBS, digested for 15 min with RNase A (50 μg/mL) at 37 °C, and stained with propidium iodide (50 μg/mL). The cell DNA contents (10,000 cells per experimental group) were determined with a flow cytometer (FACSCaliber, Becton Dickinson Biosciences) equipped with a ModFit LT program (Verity Software House, Inc.). Cells were left untreated or treated for 48 h with two different concentrations of gedatolisib and si‐RNA mediated GSK3β(50 nM) or CHIR‐99021 (0.1 and 1 μM/L of gedatolisib and 10 and 20 μM/L of CHIR‐99021), at which point the cells were collected and stained with Annexin V‐Fluorescein IsoThioCynate (V–FITC). Apoptotic cell death was assessed by counting the numbers of cells that stained positive for AnnexinV–FITC and negative for propidium iodide using an Apoptosis Detection Kit 2 (BD Pharmingen), coupled with fluorescence‐activated cell sorting analysis.

Colony‐forming assay

Colony‐forming assays were performed in 6‐well plates. Cells were plated in triplicate at 4 × 10³ or 1 × 10⁴ cells/well and incubated for 2 days before addition of vehicle, gedatolisib, and/or CHIR‐99021 at the concentrations indicated in the figures. Seven days later, the medium was removed and the cells were stained with Coomassie Brilliant Blue. Cells were washed five times in phosphate‐buffered saline and once in distilled water and then air dried. Colonies were enumerated using a GelCount colony counter (Oxford Optronix, Abingdon, UK) according to the manufacturer's instructions.

Immunoprecipitation

For immunoprecipitation of mTOR–Raptor and TSC1‐TSC2 complexes, HCT‐116, HCT‐15, and LS174T cells were seeded in 100 mm × 20 mm dishes and incubated for 24 h. Vehicle or gedatolisib was added to a final concentration of 1 μM and the cells were incubated for an additional 12 h and then harvested. Whole cell extracts were prepared and immunoprecipitations were performed as previously described.25

Small interfering RNA‐mediated knockdown

Small interfering RNAs (siRNAs) against GSK3β and TCF7 were purchased from mbiotech (Seoul, Korea). Cells were transfected with siRNAs (50 nM) according to the manufacturer's instructions, as previously reported. Cell lysates were prepared at 48 h after transfection. siRNA sequences were: negative control: sense 5′‐CCUCGUGCCGUUCCAUCAGGUAGUU‐3′ and anti‐sense 5′‐CUACCUGAUGGAACGGCACGAGGUU‐3′; GSK3β: sense 5′‐CAGCUAUACAGACACUAAAUU‐3′ and anti‐sense 5′‐UUUAGUGUCUGUAUAGCUGUU‐3′; TCF7: sense 5′‐GACAACUACGGGAAGAAGAUU‐3′ and anti‐sense 5′‐UCUUCUUCCCGUAGUUGUCUU‐3′.

Site‐directed mutagenesis

Plasmid pcDNA3.1 encoding TCF7 was purchased from Addgene (Cambridge, MA), and pOTB7 encoding GSK3β was purchased from KHGB (Korea Human Gene Bank, Seoul, Korea). The TCF7 H155 frameshift and GSK3β S9A mutations were introduced using a QuikChange Site‐Directed Mutagenesis kit (Stratagene, La Jolla, CA) with full‐length TCF7 and GSK3β as the templates. The primers to introduce the TCF7 H155fs mutation were: forward 5′‐TTGGGGGACACCGGTGGGGGGGCTG‐3′ and reverse 5′‐CAGCCCCCCCACCGGTGTCCCCCAA‐3′, and the primers to introduce the GSK3β S9A mutation were: forward 5′‐TTGGGGGACACCGGTGGGGGGGCTG‐3′ and reverse 5′‐CAGCCCCCCCACCGGTGTCCCCCAA‐3′. Mutagenesis reactions were performed according to the manufacturer's instructions. The mutated sequences were confirmed by Sanger sequencing (Macrogen, Seoul, Korea).

BrdU incorporation and cell growth analysis

HCT‐15 and LS174T cell lines were transfected with plasmids encoding GSK3β wild‐type and GSK3β‐S9A mutant proteins using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). BrdU staining was performed as previously described.18

Establishment of patients‐derived organoids and organoid culture

Human Colon cancer tissues were obtained from Seoul National University Hospital (Seoul, Korea). The study protocol was approved by the Institute Review Board(IRB) of Seoul National University Hospital with informed consent (1608–054‐784). Tumor cell isolation and organoid culture were processed as previously described.26

Organoid electroporation and 3D drug response assay

Organoids were transfected with TCF7 H155fs mutation using NEPA21 Electroporation (Nepagene, Japan) as previously described.27 After 24 h, drug was treated and cell viability was assessed using Cell‐titer Glo 3D (Promega, US) after 6 days. Organoids images were obtained using ImageXpress Nano system (Molecular Devices, US).

Mouse xenograft model

All animal experiments were approved by the Institute Laboratory Animal Resources and Use Committee of Seoul National University. Four‐week‐old female BALB/c nude 20 mice were purchased from Central Lab Animal Inc. (Seoul, Korea) and weighed 18–20 g on arrival. Mice were allowed to acclimatize for 1 week before experiments. HCT‐15 cells (1 × 10⁷/mouse) were injected subcutaneously into the right flank of the mouse. When the mice had developed a tumor volume of 100–200 mm³, they were randomized into the treatment groups (n = 5 per group) and treatment with vehicle, gedatolisib, LiCl, or gedatolisib plus LiCl was initiated. Mice treated with LiCl were also given additional NaCl to prevent electrolyte imbalance, as previously reported. 28 Gedatolisib (10 mg/kg) and LiCl (80 mg/kg) were injected intravenously and intraperitoneally, respectively, twice a week for 3 weeks. The tumors were then measured with calipers three times per week, and the tumor volume in mm³ was calculated in accordance with the after formula: {(width)² × (height))/2}. After the final treatment on day 21, all of the mice were euthanized.

Immunohistochemistry (IHC) staining

Xenograft tumor specimens were harvested at the end of an efficacy study and fixed for 24 h in 10% buffered formalin. After embedding in paraffin and sectioning, general tissue morphology was visualized via H&E staining. Ki67 procedure has previously been described.29.

Statistical analysis

All experiments were conducted in duplicate or triplicate, with at least two biological replicates. Data are expressed as the mean ± standard deviation (s.d.). Statistical significance was calculated using unpaired Student's t‐test, except for the differences between tumor sizes in xenograft‐bearing mice, which was analyzed by ANOVA. p < 0.05 was considered statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Results

PIK3CA‐mutant CRC cell lines with elevated phosphorylated mTOR and reduced phosphorylated TSC2 expression are sensitive to gedatolisib

To explore the effects of the dual PI3K/mTOR inhibitor gedatolisib in CRC cells, we first determined the concentration of gedatolisib that reduced cell viability by 50% (IC₅₀) in a panel of 29 CRC cell lines (Fig. 1 a). The cells were then divided into two groups, designated sensitive or resistant to gedatolisib, based on IC₅₀ values of less than or greater than 0.5 μM, respectively. Seven of the 29 CRC cells harbored at least one PIK3CA mutation (Supporting Information Table S1); of these cells, 5 were gedatolisib sensitive (HCT‐116, WiDr, HT‐29, SNU‐C5 and DLD‐1; IC₅₀ < 0.5 μM) and 2 were resistant (HCT15 and LS174T; IC₅₀ > 7 μM). Next, we investigated the basal expression of additional PI3K/AKT/mTOR pathway molecules and observed that the levels of p‐mTOR, which phosphorylated ser2481 ensues from an auto‐phosphorylation mechanism and is found in both mTORC1 and 2 complexes,30 were higher and the levels of p‐TSC2 were lower in the gedatolisib‐sensitive than the ‐resistant cell lines (Fig. 1 b). In addition, since all seven cell lines have a PIK3CA hotspot mutation, it can be concluded that the PI3K/AKT/mTOR pathway is activated, the levels of p‐AKT Thr308, readout of PI3K activation, and the levels of p‐AKT ser473, readout of mTORC2 activation, are both high.

Figure 1.

Gedatolisib sensitivity and resistance of CRC cell lines harboring PIK3CA mutations. (a) A total of 29 CRC cell lines were treated with vehicle or gedatolisib at the indicated concentrations for 72 h, and IC₅₀ values were determined using the MTT viability assay. Untreated cells were considered 100% viable. (b) Western blot analysis of basal levels of the total or phosphorylated PI3K/mTOR pathway‐associated proteins in CRC cell lines. Data are representative of three independent experiments. β‐Tubulin was probed as a loading control in all western blots. (c) Colony‐forming assay of CRC cell lines treated with vehicle or gedatolisib at 0.1 or 1 μM for 10 days. (d) Annexin V‐FITC staining and western blotting of CRC cell lines treated with vehicle or gedatolisib at 1 μM for 72 h. (e) Human organoids, O28T(PIK3CA wild‐type) and O29T(PIK3CA R88Q), were treated gedatolisib at 0.01 or 0.1 or 1 μM for 6 days measured cell viability using the cell titler glo. (f) CRC cell lines were treated with vehicle or gedatolisib at 1 μM for 24 h, and cell lysates were subjected to western blotting with antibodies specific for total or phosphorylated proteins as indicated. [Color figure can be viewed at wileyonlinelibrary.com]

Furthermore, we evaluated the cell lines for their colony‐forming ability. As expected, gedatolisib treatment markedly decreased both the number and the size of colonies formed by gedatolisib‐sensitive cell lines, but had no effect on the resistant cell lines (Fig. 1 c). To confirm these findings, we examined the cell cycle status by flow cytometry of cells, stained with the DNA‐intercalating dye propidium iodide or by western blot analysis of cell cycle regulatory proteins. In the sensitive cell lines, gedatolisib dose‐dependently increased the proportion of cells in sub‐G1 phase. However, the resistant cell lines showed no significant change in sub‐G1 phase (data not shown). So, we evaluated apoptosis induction by Annexin V‐FITC staining and the number of Annexin V stained cells were increased in five sensitive cell lines, showing increased expression of active caspase‐3 and active caspase‐7, but had no effect on the resistant cell lines (Fig. 1 d). In addition, we used patient‐derived colorectal cancer organoids to determine whether they show similar tendencies to cell line. O29T organoids with PIK3CA R88Q mutation were more sensitive compared to O28T with PIK3CA wild‐type. Next, phosphorylation of several components of the PI3K/mTOR signaling pathway, indicative of their activation, was also inhibited by gedatolisib treatment in the sensitive cell lines, whereas it was either maintained or upregulated in the resistant cell lines (Fig. 1 f). These results showed that gedatolisib effectively suppressed activation of the PI3K/mTOR pathway, reduced cell growth and proliferation, and induced apoptosis in sensitive PIK3CA‐mutant CRC cell lines and this sensitivity was demonstrated also in human organoid samples.

Gedatolisib‐resistant cell lines exhibit sustained mTOR signaling which is dependent on GSK3β

HCT‐15 and LS174T cell lines are resistant to gedatolisib despite carrying activating PIK3CA mutations. Moreover, gedatolisib treatment increased the levels of p‐mTOR in these cells (Fig. 1 f), suggesting that a molecule upstream of mTOR might be involved in the mechanism of gedatolisib resistance. Tuberous sclerosis proteins 1 and 2 (TSC1/TSC2) are upstream inhibitors of mTOR and function as integrators between mTOR and other signaling pathways. AKT negatively regulates the formation of TSC1‐2 complexes by phosphorylating TSC2.31 Thus, based on our results and previous data,31 we hypothesized that TSC1‐2 complex levels would be higher in gedatolisib‐treated than ‐untreated sensitive cell lines, due to the inhibition of AKT activity by gedatolisib. Conversely, TSC1‐2 complex formation was expected to be unaffected by gedatolisib in the resistant cell lines. To test this hypothesis, we examined TSC1‐2 complex formation by immunoprecipitating TSC2 and western blotting to detect associated TSC1. As expected, increased amounts of TSC1 were complexed to TSC2 after gedatolisib treatment in the sensitive cell line HCT‐116, and, accordingly, the levels of p‐p70S6K, p‐S6K, and p‐4E‐BP1 were decreased (Fig. 2 a). However, formation of TSC1‐2 complexes was relatively unaffected by gedatolisib treatment in the resistant cell lines, and, in fact, was slightly decreased.

Figure 2.

GSK3β sustains mTOR signaling in gedatolisib‐resistant cell lines. (a) HCT‐116 (sensitive) and HCT‐15 and LS174T (resistant) cells were treated with vehicle or gedatolisib at 1 μM for 12 h. TSC2 and Raptor were immunoprecipitated from cell lysates and subjected to western blotting with antibodies to TSC1 or TSC2 and mTOR or Raptor (upper rows). 30% of input cell lysate was blotted with antibodies to total or phosphorylated forms of the indicated proteins (lower rows). (b) Western blot analysis and phosphorylated and total GSK3β protein expression in the indicated CRC lines. The graph shows the ratio of p‐GSK3β(ser9)/p‐GSK3β(tyr216) to total GSK3β expression levels. (c) HCT‐15 and LS174T cells were transfected with control or GSK3β‐targeting siRNA for 48 h and then treated with vehicle or gedatolisib 1 μM for 12 h. mTOR was immunoprecipitated from cell lysates and subjected to western blotting with antibodies to Raptor or mTOR (Upper rows). 30% of input lysate was blotted with antibodies to total or phosphorylated forms of the indicated proteins. β‐Tubulin was probed as a loading control in all western blots. Data are representative of three independent experiments.

Because mTOR–Raptor complexes have also been reported to be affected by GSK3β,32 we examined mTOR–Raptor complexes formation by immunoprecipitating mTOR and western blotting to detect associated Raptor. Immunoprecipitation of Raptor and western blotting for mTOR indicated that the formation of mTOR–Raptor complexes was decreased by gedatolisib treatment in sensitive cell lines due to the reduction in the activity of mTOR signaling, whereas complex formation was unchanged in the resistant cell lines (Fig. 2 a).

Because GSK3β is a well‐known regulator of TSC1‐2 and mTOR signaling,33, 34 we examined its involvement in gedatolisib sensitivity in more detail. Interestingly, the gedatolisib‐resistant cell lines expressed much lower levels of the inactive, phosphorylated form of GSK3β(ser9) and higher levels of the active, phosphorylated form of GSK3β(tyr216) compared to the sensitive cell lines (Fig. 2 b), suggesting that GSK3β is more active in the resistant cells. Previous studies have shown that low p‐GSK3β levels and high GSK3β activity can promote the proliferation of human ovarian and pancreatic cancer cells.25, 35 To determine whether this is also observed in the gedatolisib‐resistant cell lines, we transfected HCT‐15 and LS174T cells with GSK3β‐S9A, a constitutively active form of GSK3β, and examined cell proliferation (5‐bromo‐2′‐deoxyuridine [BrdU] incorporation) and colony formation. Indeed, GSK3β‐S9A expression significantly increased BrdU incorporation and colony formation by the gedatolisib‐resistant cells (Supporting Information Fig. S2). This confirms that the constitutive activation of GSK3β does not affect cell proliferation in the sensitive cell line. These data suggested that active form of GSK3β might enhance the proliferation in gedatolisib‐resistant cell lines.

However, transfection of the resistant cells with GSK3β‐targeting siRNA led to a reduction in mTOR–Raptor complex formation upon treatment with gedatolisib, which was accompanied by a reduction in p‐mTOR, p‐P70S6K, p‐S6K, and p‐4E‐BP1 levels (Fig. 2 c). These data indicate that the gedatolisib‐induced reduction in AKT activity increases the formation of TSC1‐2 complexes, which induce separation of mTOR–Raptor complexes and subsequently decrease mTOR signaling. Furthermore, siRNA‐mediated GSK3β knockdown (KD) promoted formation of TSC1‐2 complexes in gedatolisib‐treated resistant cell lines (Supporting Information Fig. S3).

Furthermore, it appears that active form of GSK3β is critically involved in the mechanism of resistance to PI3K/mTOR inhibition through its ability to sustain mTOR signaling.

Increased WNT signaling induced by TCF7 H155fs* mutation confers gedatolisib resistance

The gedatolisib‐resistant cell lines HCT‐15 and LS174T both carry a frameshift mutation in TCF7 (c.465_466insC; Supporting Information Fig. S1), a critical transcription factor in the WNT/β‐catenin pathway. The frameshift results in loss of a corepressor binding domain and increases TCF7 transcriptional activity. Therefore, we hypothesized that the mutant TCF7 protein may increase the levels of active GSK3β and the WNT/β‐catenin signaling pathway. Consistent with this, basal levels of p‐LRP6, AXIN2 and active β‐catenin, critical components in WNT/β‐catenin signaling, were much higher in gedatolisib‐resistant HCT‐15 and LS174T cells than in the sensitive cell lines (Fig. 3 a). To investigate if resistant cell lines have WNT/β‐catenin signaling activation, we analyzed WNT/β‐catenin activity in 7 cell lines using TOP‐flash reporter assay and also checked FOP‐flash activity as a negative control. The basal activity of WNT/β‐catenin was enhanced in resistant cell lines, HCT‐15 and LS174T (Fig. 3 a). Therefore, we thought that the TCF7 mutation was responsible for the resistance of the resistant cell line to gedatolisib. In the resistant cells, siRNA‐mediated KD of TCF7 reduced the levels of p‐LRP6, AXIN2, active β‐catenin and active p‐GSK3β‐tyr216 and specifically increased the levels of inactive p‐GSK3β‐ser9 (Supporting Information Fig. S4). In addition, KD of TCF7 in combination with gedatolisib treatment inhibited both mTOR and WNT/β‐catenin signaling in the resistant cells (Fig. 3 b). These results support the possibility that the TCF7 H155fs* mutation is responsible for the elevated the levels of active GSK3β and WNT/β‐catenin pathway signaling. In addition, we identified that in the presence of AES protein, which functions as transcription repressor,23 TCF7 frameshift mutation increased transcription activity in WNT/β‐catenin signaling compared to Mock(empty vector) or TCF7 wild‐type, using luciferase assay (Supporting Information Fig. S5). Also, TCF7 transcription activity was measured in seven cell lines, and it was confirmed that the activity was further increased in resistant cell lines (Supporting Information Fig. S5). To confirm these results, we examined the effect of TCF7 H155fs* expression on the gedatolisib‐sensitive cell lines HCT‐116 and HT‐29. Interestingly, expression of TCF7 H155fs* rendered the cells more resistant to gedatolisib (>1 μM IC₅₀) compared to the parental cell lines harboring Mock (empty vector) (Fig. 3 c). In addition, the levels of p‐AKT, p‐mTOR, p‐S6K, p‐P70S6K, p‐4E‐BP1, p‐LRP6, AXIN2 and active β‐catenin in these cells was either unaffected or upregulated by gedatolisib treatment (Fig. 3 d). p‐GSK3β also showed an increase in its active form through low phosphorylation expression under TCF7 H155fs* mutation. In other words, these results showed the relationship between GSK3β activation and TCF7 mutation. Similar results were obtained with a second gedatolisib‐sensitive cell line, WiDr (Supporting Information Fig. S6) and patient‐derived colorectal cancer organoids (Fig. 3 e).

Figure 3.

The WNT/β‐catenin pathway is hyperactivated in CRC cell lines rendered resistant to gedatolisib by the TCF7 H155 frameshift mutation. (a) Western blotting of basal levels of total and phosphorylated/active forms of the indicated WNT/β‐catenin pathway proteins. TOP/FOP flash luciferase assay was conducted in 7 cell lines under the Wnt3a stimulation. (b) Western blot analysis of total and phosphorylated/active forms of the indicated proteins in HCT‐15 and LS174T cells transfected with control or GSK3β‐targeting siRNAs and treated with vehicle or gedatolisib 1 μM for 24 h. (c) Viability of HCT‐116 and HT‐29 cells after transfection with Mock(empty vector) or TCF7 H155fs* (TCF7 MT) followed by treatment for 72 h with vehicle or 0.1 μM gedatolisib. The percentage of viable cells is shown relative to the untreated controls (*p < 0.05, ***p < 0.001). (d) HCT‐116 and HT‐29 parental and TCF7 MT cell lines were treated with vehicle or 0.1 μM gedatolisib for 24 h, and cell lysates were subjected to western blotting of total and phosphorylated/active forms of the indicated proteins. β‐Tubulin was probed as a loading control in all western blots. Data are representative of three independent experiments. (e) Viability of O29T organoids after transfection with Mock (empty vector) or TCF7 H155fs* (TCF7 MT) followed by treatment for 6 days with vehicle or 0.01 μM or 0.1 μM gedatolisib using cell tilter glo. The percentage of viable cells is shown relative to the untreated controls. [Color figure can be viewed at wileyonlinelibrary.com]

Collectively, these results indicate that the TCF7 H155fs* mutation decrease the levels of inactive GSK3B suggesting an increase of the its active form. In addition, they also demonstrate that TCF7 H155fs* mutation leads WNT/β‐catenin signaling and induction of gedatolisib resistance.

Inhibition of GSK3β confers sensitivity to gedatolisib

Because GSK3β cross‐regulates both the PI3K/mTOR and WNT/β‐catenin signaling pathways, we hypothesized that inhibition of GSK3β might attenuate gedatolisib resistance caused by the mutant TCF7‐induced increase in WNT/β‐catenin signaling. To test this, we cotreated the resistant cell lines with gedatolisib and si‐GSK3β. Indeed, the combination treatment of si‐GSK3β and gedatolisib inhibited not only colony formation ability but also cell viability of HCT‐15 and LS174T cells (Fig. 4 a). The combination treatment induced apoptosis, as shown by Annexin V‐FITC staining and increased expression of active caspase‐3 and active caspase‐7 (Fig. 4 b). Furthermore, it decreased the levels of p‐mTOR, p‐P70S6K, p‐S6K, and p‐4E‐BP1 (Fig. 4 c) compared to the effects of either agent alone. Interestingly, the levels of p‐LRP6, AXIN2, and active β‐catenin were also decreased by the combination treatment (Fig. 4 c). Similarly, treatment of the resistant cells with gedatolisib plus CHIR‐99021(CHIR), a selective GSK3β inhibitor, significantly inhibited cell viability and colony‐forming ability (Figs. 4 d and 4 e), induced apoptosis showing increased active caspase‐3/7 (Fig. 4 f). The combination treatment of gedatolisib and CHIR‐99021 decreased the levels of p‐mTOR, p‐P70S6K, p‐S6K, p‐4E‐BP1, p‐LRP6, p‐GS, AXIN2, and active β‐catenin (Fig. 4 g). Similar results were obtained when the resistant cell lines were treated with gedatolisib plus SB216763(SB) or LiCl, two additional GSK3β inhibitors (Supporting Information Figs. S7 and S8). In contrast, treatment of the gedatolisib‐sensitive cell lines WiDr and HT‐29 with gedatolisib and CHIR induced mTOR and WNT signaling by increasing the levels of the phosphorylated and total S6 and 4E‐BP1 and AXIN2 and active β‐catenin (Supporting Information Fig. S9). Taken together, these results suggest that inhibition of GSK3β restores the sensitivity of HCT‐15 and LS174T cells to PI3K and mTOR inhibition.

Figure 4.

PI3K/mTOR and GSK3β inhibitors have synergistic effects on gedatolisib‐resistant CRC cell lines. (a) Upper: Colony‐forming assay in HCT‐15 and LS174T cells transfected with control or GSK3β‐targeting siRNAs and treated with vehicle or gedatolisib 1 μM for 10 days. Lower: Viability of HCT‐15 and LS174T cells transfected with control or GSK3β‐targeting siRNAs and treated with vehicle or gedatolisib 1 μM for 48 h. (***p < 0.001) (b) Annexin V‐FITC staining and western blotting of HCT‐15 and LS174T cells transfected with control or GSK3β‐targeting siRNAs and treated with vehicle or gedatolisib 1 μM for 48 h. (c) Western blot analysis of total and phosphorylated/active forms of the indicated proteins in HCT‐15 and LS174T cells transfected with control or GSK3β‐targeting siRNAs and treated with vehicle or gedatolisib 1 μM for 24 h. (d) Viability of HCT‐15 and LS174T cells at 72 h after treatment with the indicated combinations of vehicle, gedatolisib (1 μM), and/or the GSK3β inhibitor CHIR‐99021(10 μM and/or 20 μM) (***p < 0.001). (e) Colony‐forming assay of HCT‐15 and LS174T cells treated with the indicated concentrations of vehicle, gedatolisib, and/or CHIR for 10 days (***p < 0.001) (f) Annexin V‐FITC staining of HCT‐15 and LS174T cells treated with indicated combinations of vehicle, gedatolisib (1 μM), and/or the GSK3β inhibitor CHIR‐99021(10 μM and/or 20 μM) for 48 h. (g) Western blot analysis of basal levels of total and phosphorylated/active forms of the indicated proteins in HCT‐15 and LS174T cells. β‐Tubulin was probed as a loading control in all western blots. Data are representative of three independent experiments.

Gedatolisib and the GSK3β inhibitor LiCl synergistically inhibit the growth of CRC tumors in a xenograft mouse model

To determine whether GSK3β inhibition restores sensitivity to gedatolisib, we examined xenograft tumor growth. HCT‐15 cells were injected subcutaneously into BALB/c nude mouse, and groups of mice were treated for 3 weeks with vehicle, gedatolisib alone, LiCl alone, or both gedatolisib and LiCl. Mice treated with LiCl were administered additional NaCl to prevent electrolyte imbalance, as previously reported.28 While monotherapy with gedatolisib or LiCl alone had no effect on (or marginally increased) the tumor size compared to vehicle treatment, the combination of gedatolisib plus LiCl markedly decreased tumor growth (Fig. 5 a). In addition, We also analyzed the effects of the treatments on PI3K/mTOR and WNT/β‐catenin pathway activation in tumor tissue excised at the end of the experiment (day 19). In Figure 5 b, ki67 staining, which marks cell proliferation, decreased in combination therapy treated tumor tissue compared to control tumor tissue. We found that the levels of p‐GS, p‐AKT, p‐S6K, p‐4E‐BP1, p‐LRP6, AXIN2, active β‐catenin, and p‐GSK3β(tyr216) were decreased and level of p‐GSK3β(ser9) was increased by treatment with both gedatolisib and LiCl, but not with the single agents (Fig. 5 c). Taken together, these results indicate that treatment of CRC cells with both gedatolisib and LiCl inhibits oncogenic signaling through PI3K/AKT/mTOR and WNT signaling pathways, resulting in apoptosis.

Figure 5.

Co‐treatment with PI3K/mTOR and GSK3β inhibitors decreases CRC tumor growth in a xenograft mouse model. (a) Groups of 4‐week‐old female BALB/c nude mice were subcutaneously injected with HCT‐15 cells (1 × 10⁷/mouse) and administered vehicle, gedatolisib (10 mg/kg), and/or LiCl (80 mg/kg) twice weekly for 3 weeks. Tumor size was measured every other day. Data are the mean ± s.d. of 5 mice per group. Representative pictures of HCT‐15 tumors excised from mice on day 19 of the experiment shown in (a). (b) Representative images of IHC staining of tumor tissues of mice with control and gedatolisib plus LiCl groups. H&E staining (upper) and Ki‐67 staining (lower). (c) Western blot analysis of the total and phosphorylated/active forms of the indicated proteins extracted from HCT‐15 tumors excised on day 19. β‐Tubulin was probed as a loading control.

Discussion

In our study, we identified a novel mechanism of resistance to the dual PI3K and mTOR inhibitor gedatolisib in CRC. We confirmed that most cell lines carrying PIK3CA mutations are sensitive to gedatolisib and demonstrated that increased WNT/β‐catenin signaling induces resistance to gedatolisib in the presence of mutant PIK3CA. Our data demonstrate that the TCF7 H155fs* mutation reduces levels of inactive GSK3B suggesting its possible activation and, as a result, elevated WNT/β‐catenin signaling. This particular mechanism of resistance to a dual PI3K and mTOR inhibitor has not been previously been described in CRC. There are evidences to date, suggesting that resistance of CRC to dual PI3K/mTOR ‐targeted therapy can occur via several routes; dysregulation of the KRAS‐RAF–MEK pathway,10, 36 activation of ERBB2 and ERBB3 through FOXO3a,37 c‐Myc induction by the Notch pathway,38 and/or activation of the IGF1R/AKT pathway.39 The present study is the first to suggest a WNT/β‐catenin signaling‐related mechanism of gedatolisib resistance in PIK3CA‐mutant CRC.

It has been well characterized that GSK3β has multifunctional roles in cell death pathways, cancer‐associated signaling pathways, cell proliferation and growth. It is notable that phosphorylation of GSK3β by AKT is considered as inactive form. Through this, GSK3β acts as a negative regulator by inducing up‐regulation of WNT/β‐catenin signaling.34, 40 On the other hand, as a positive regulator, GSK3β modulates mTOR activity by regulating formation of TSC1‐2 complexes,33 which, in turn, influence mTOR/Raptor complex formation and the activity of downstream effectors such as ribosomal protein S6 kinase (S6K).32, 41, 42 Our study identified that mTOR/Raptor complex was affected by GSK3β and it also contribute to drug response through decreased the levels of active mTOR (Fig. 2 c). Furthermore, active GSK3β was noted to be significantly higher in tumor cells than normal cells in 63.6% of CRC patients.28 These data suggested that active GSK3β is associated with cancer progression and poor response to anticancer therapy.43 Thus, GSK3β is emerging as a potential therapeutic target. When targeted, it results in induction of cell death through the after mechanisms; regulation of MDM2‐p53 pathway,44 JNK‐dependent mechanisms25, MYCN mRNA levels45 and disruption of centrosome regulation in cancer cells.42 It is known that GSK3β mediates LRP6 phosphorylation, which activates WNT/β‐catenin signaling.46 However, a study showing the oncogenic role of GSK3β in direct relationship with the WNT/β‐catenin pathway, remains to be elucidated. In our study, we showed active form of GSK3β in gedatolisib‐resistant cell lines harboring TCF7 frameshift mutation and it resulted in increased WNT/β‐catenin signaling. These results conflict with previous studies stating that inhibition of GSK3β induces WNT/β‐catenin signaling. Therefore, it seems likely that GSK3β may have dual activity, acting as both negative and positive regulator of WNT/β‐catenin signaling.

Among the components of WNT/β‐catenin signaling, TCF/LEF family of transcription factors are essential molecules.21 Both TCF and LEF have β‐catenin binding domains, which allow binding of β‐catenin and initiate transcription process of WNT/β‐catenin signaling. Thus, lacking of β‐catenin binding domain results in suppressed activity of WNT/β‐catenin signaling.47 High expression of TCF/LEF and their truncated form that lacks a β‐catenin binding domain were also found in many cancer types. In addition to these mutations, mutations in other TCF/LEF domains are also found. For example, a variant of TCF4 is highly expressed in CRC, and TCF4 is mutated in about 40% of CRC patients with microsatellite instability and patients with leukemia.48 Recent study suggested that TCF7 is also upregulated in CRC.49, 50 Another TCF7 frameshift mutation, c. 460delC, which also confers high transcriptional activity, has been reported to increase the expression of cyclin D1, a target of WNT/β‐catenin signaling, although the underlying mechanism was not clear.23 We found novel TCF7 H155 frameshift mutation (c.465_466insC) which results in a truncated molecule that lacks a binding domain for the transcriptional repressor protein AES/Groucho.51 Therefore, AES/Groucho protein could not bind to the binding domain to suppress transcription of target genes in WNT/β‐catenin signaling. Thus, TCF7 H155fs* has enhanced transcriptional activity, resulting in increased WNT/β‐catenin signaling activity. We demonstrated that TCF7 H155 frameshift mutation remarkably showed high transcription activity in either the absence or presence of AES/Groucho protein (Supporting Information Fig. 5). Although the TCF7 H155 frameshift mutation has been reported only rarely (<1%) in CRC, our findings may be useful in guiding the current and future use of combination therapies for an n‐of‐1 trial.

In conclusion, our study demonstrates that PIK3CA‐mutant CRC cells develop resistance to gedatolisib through activation of GSK3β and WNT/β‐catenin signaling resulting from an activating TCF7 mutation. We suggest that, although there is little clinical data on this particular TCF7 mutation, it could be used as a biomarker in clinical trials to identify patients likely to be resistant to dual PI3K/mTOR inhibitors.

Supporting information

Supplementary Table 1. Mutational status of the CRC cell lines used in our study. Genomic DNA was extracted from cell lines using a QIAamp DNA Mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Mutations were identified by whole‐exome sequencing (Celemics, Seoul, Korea) and validated using Sanger sequencing (Macrogen, Seoul, Korea).

Figure S1. Schematic showing the wild‐type and truncated H155fs* mutant forms of TCF7.

Figure S2. Fluorescence and light microscopy of HCT‐15 and LS174T cells transfected with plasmids encoding EGFP alone (control) or EGFP‐fused to constitutively active GSK3β‐S9A. Cell growth was assessed using BrdU incorporation (red), nuclei were stained with 4′,6‐diamidino‐2‐phenylindole (DAPI, blue), and EGFP is shown in green. To test colony formation ability, transfected cells were seeded in six well‐plate for 10 days.

Figure S3. HCT‐15 and LS174T cells were transfected with control or GSK3β‐targeting siRNA for 48 h and then treated with vehicle or gedatolisib 1 μM for 12 h. TSC2 was immunoprecipitated from cell lysates and subjected to western blotting with antibodies to TSC1 or TSC2 (Upper rows). 30% of input lysate was blotted with antibodies to total or phosphorylated forms of the indicated proteins. β‐Tubulin was probed as a loading control in all western blots. Data are representative of three independent experiments.

Figure S4. HCT‐15 and LS174T cells were transfected with control or TCF7‐targeting siRNA, and cell lysates were subjected to western blotting of total and phosphorylated/active forms of the indicated proteins.

Figure S5. Left; The TCF7 frameshift mutation increases transcriptional activity of the WNT/β‐catenin signaling pathway in the presence of AES protein under Wnt3a stimulation. HEK293 cells were co‐transfected with pGL3‐basic‐Luc (0.4 μg) and pCMV‐β‐Gal (0.1 μg; to normalize transfection efficiencies) and various combinations of plasmids encoding wild‐type TCF7, TCF7 H155fs mutant TCF7, (0.2 μg/sample), and/or AES (0.2 μg/sample) using Lipofectamine 2000. Cells were lysed and luciferase activity was evaluated using a TR717 microplate luminometer (Applied Biosystems, Foster City, CA), in accordance with the manufacturer's instructions. HEK293T cells were transfected with Monk, TCF7 wild‐type, mutant TCF7 H155fs*, or AES. Cells were lysed, and the relative luciferase activity (normalized to β‐galactosidase activity) was evaluated. Western blot shows the level of AES in transfected cells. Right; 7 cell lines co‐transfected with pGL3‐basic‐Luc (0.4 μg) and pCMV‐β‐Gal (0.1 μg; to normalize transfection efficiencies) and various combinations of plasmids encoding wild‐type TCF7, TCF7 H155fs mutant TCF7, (0.2 μg/sample), and/or AES (0.2 μg/sample) using Lipofectamine 2000. The experimental method is the same as mentioned above.

Figure S6. The TCF7 H155fs* mutation induces resistance to a dual PI3K/mTOR inhibitor. WiDr cells were transfected with Mock(empty vector) or TCF7 H155fs* and treated with vehicle or gedatolisib 0.1 μM for 72 h (viability) or 24 h (western blots). The percentage cell viability is shown relative to untreated controls. Whole cell lysates were analyzed by western blotting with antibodies specific for the phosphorylated/active forms of the indicated proteins. β‐Tubulin was probed as a loading control.

Figure S7. Inhibitors of PI3K/mTOR (gedatolisib) and GSK3β (SB216763, SB) show synergistic effects in gedatolisib‐resistant CRC cell lines. Upper panels: HCT‐15 and LS174T cells were treated with the indicated combinations of vehicle, gedatolisib 0.1 μM, and SB 20 or 40 μM, and cell viability was measured after 72 h. Middle panel: Colony‐forming assay of HCT‐15 and LS174T cells treated for 10 days with vehicle, gedatolisib, and SB as described for the upper panel. Lower panel: Western blot analysis of HCT‐15 and LS174T cells treated with vehicle or gedatolisib 0.1 μM in the presence or absence of 40 μM SB for 72 h. Blots were analyzed with antibodies specific for total and phosphorylated/active forms of the indicated proteins. β‐Tubulin was probed as a loading control.

Figure S8. Inhibitors of PI3K/mTOR (gedatolisib) and GSK3β (LiCl) show synergistic effects in gedatolisib‐resistant CRC cell lines. The experiment was performed as described for Figure S6 except that HCT‐15 and LS174T cells were treated with the indicated combinations of vehicle, gedatolisib 0.1 μM, and LiCl 1 or 2 mM.

Figure S9. mTOR and WNT/β‐catenin signaling pathways are activated in gedatolisib‐sensitive CRC cell lines treated with a combination of the PI3K/mTOR inhibitor gedatolisib and the GSK3β inhibitor CHIR‐99021(CHIR). Western blot analysis of WiDr and HT‐29 cells treated with the indicated combinations of vehicle, gedatolisib 0.1 μM and CHIR 20 μM for 72 h. Blots were analyzed with antibodies specific for total and phosphorylated/active forms of the indicated proteins. β‐Tubulin was probed as a loading control.