Abstract
Acquired resistance to molecular targeted therapy represents a major challenge for the effective treatment of cancer. Hyperactivation of the PI3K/AKT pathway is frequently observed in virtually all human malignancies, and numerous PI3K and AKT inhibitors are currently under clinical evaluation. However, mechanisms of acquired resistance to AKT inhibitors have yet to be described. Here, we use a breast cancer preclinical model to identify resistance mechanisms to a small molecule allosteric AKT inhibitor, MK2206. Using a step-wise and chronic high-dose exposure, breast cancer cell lines harboring oncogenic PI3K resistant to MK2206 were established. Using this model, we reveal that AKT3 expression is markedly upregulated in AKT inhibitor–resistant cells. Induction of AKT3 is regulated epigenetically by the bromodomain and extra terminal domain proteins. Importantly, knockdown of AKT3, but not AKT1 or AKT2, in resistant cells restores sensitivity to MK2206. AKT inhibitor–resistant cells also display an epithelial to mesenchymal transition phenotype as assessed by alterations in the levels of E-Cadherin, N-Cadherin, and vimentin, as well as enhanced invasiveness of tumor spheroids. Notably, the invasive morphology of resistant spheroids is diminished upon AKT3 depletion. We also show that resistance to MK2206 is reversible because upon drug removal resistant cells regain sensitivity to AKT inhibition, accompanied by reexpression of epithelial markers and reduction of AKT3 expression, implying that epigenetic reprogramming contributes to acquisition of resistance. These findings provide a rationale for developing therapeutics targeting AKT3 to circumvent acquired resistance in breast cancer. Mol Cancer Ther; 15(8); 1964–74. ©2016 AACR.
Introduction
Breast cancer is the most frequently diagnosed malignancy in women. Genetic and epigenetic deregulation of the phosphoinositide 3-kinase (PI3K)/AKT signaling pathway is highly prevalent in breast cancer (1). Mutations in PIK3CA, the catalytic subunit of the p110α subunit of PI3K, loss of PTEN, and amplification of HER2 are common genomic abnormalities in tumor cells leading to hyperactivation of PI3K/AKT signaling and subsequent phenotypes associated with malignancy (2, 3). Based on this knowledge, a number of small molecule inhibitors targeting various components of the PI3K/AKT pathway are in clinical development and evaluation. For example, both allosteric and ATP-competitive AKT inhibitors (MK2206: allosteric; GDC0068 and GSK690693: ATP-competitive) are being assessed in clinical trials for various aggressive cancers as monotherapy or as combination strategies (4, 5). In cell-free assays using purified recombinant AKT, these pan-AKT inhibitors inhibit all three AKT isoforms with nanomolar potencies. Preliminary evidence of clinical activity is observed with combination of MK2206 and trastuzumab in patients with HER2-positive solid tumors in a phase I clinical trial (6). In a separate phase Ib study, the combination of the PI3K inhibitor BKM120 and trastuzumab was evaluated in patients with HER2-positive advanced/metastatic breast cancer resistant to trastuzumab-based therapy, and initial evidence of therapeutic efficacy has been presented (7). However, the experience from other molecular targeted therapies suggests that the clinical benefits of these PI3K/AKT inhibitors are likely to be limited by the development of acquired resistance in patients.
There are a few documented mechanisms of resistance to targeted therapy for inhibitors that target PI3K or AKT (8). Of these, overexpression and/or amplification of Myc, Notch, RSK3/4, HER3, and PI3K itself have been proposed to confer resistance to certain PI3K inhibitors in the context of breast cancer (9–14). In addition, RNA sequencing analysis revealed an induction of the receptor tyrosine kinase (RTK) AXL in head and neck squamous cell carcinomas (SCC) upon adaptation to the PI3Kα inhibitor BLY719. AXL dimerizes with EGFR and results in the activation of mTOR and resistance. Importantly, overexpression of AXL is observed in SCC tumors of patients treated with BLY719 (15). With respect to AKT inhibitors, upregulation and activation of RTKs have been implicated in acquisition of resistance to targeted therapy. For example, acute treatment of breast cancer cells with an allosteric AKT inhibitor induces marked upregulation of insulin receptor, IGF1R and HER3, via a FOXO-dependent manner (16). The same study also showed that phosphorylation of multiple RTKs is enhanced upon AKT inhibition, by relieving mTORC1-mediated feedback inhibition. HER3 expression has also shown to be induced in triple-negative breast cancer (TNBC) acutely treated with the catalytic AKT inhibitor GDC0068 (17). Consistent with preclinical studies, compensatory feedback activation of HER3 and ERK has been observed in tumor biopsies from patients treated with GDC0068 (18).
Delineating the spectrum of resistance mechanisms is critical for the development of strategies to prevent or treat resistant tumors. Because mechanisms of acquired resistance to AKT inhibitors have not been elucidated, in this study we used a breast tumor line harboring an activating mutation in PI3K and generated cells resistant to a selective AKT inhibitor, MK2206, by chronic adaptation. We show that these cells display upregulation of AKT3 protein and function, accompanied by a phenotypic switch in the epithelial to mesenchymal transition (EMT). Depletion of AKT3 in resistant cells restores sensitivity to MK2206, highlighting AKT3 as a candidate for conferring resistance to molecular targeted therapy in breast cancer.
Materials and Methods
Cell culture
T47D, HEK293T, and MDA-MB-468 cells were obtained from ATCC and maintained in Dulbecco's modified Eagle medium (DMEM; Cellgro) supplemented with 10% tet system-approved fetal bovine serum (FBS; Clontech). All cell lines obtained from the cell banks listed above are tested for authentication using short tandem repeat (STR) profiling and passaged for fewer than 6 months, and routinely assayed for mycoplasma contamination.
AKT inhibitor–resistant line generation
The MK2206-resistant cell line T47D MK0.2-5 was generated by gradual dose escalation of the AKT inhibitor MK2206 from 0.2 μmol/L to 5 μmol/L for a period of 2 months, then the cells were maintained in 5 μmol/L MK2206. T47D MK5-resistant line was generated by chronic exposure of cells to 5 μmol/L MK2206 for 2 months. T47D GDC0.2-5–resistant line was generated by culturing cells in increasing concentration of the AKT inhibitor GDC0068 from 0.2 μmol/L to 5 μmol/L for 2 months. T47D parental cells were maintained in growth medium containing dimethyl sulfoxide (DMSO).
Fresh AKT inhibitor or DMSO was replaced every 3 to 4 days. Cells were considered resistant when they could be cultured routinely in growth medium containing 5 μmol/L MK2206 or GDC0068. MK2206 and GDC0068 have been described previously (4, 19) and were obtained from Selleck Chemical.
3D cultures
3D cultures were prepared as previously described (20). Briefly, chamber slides were coated with growth factor-reduced Matrigel (BD Biosciences) and allowed to solidify for 30 minutes. Cells (1 × 104) in assay medium were seeded on coated chamber slides. Assay medium contained DMEM supplemented with 10% FBS and 2% Matrigel. The assay medium was replaced every 4 days. Doxycycline (dox, 100 ng/mL) was added every 2 or 3 days.
Antibodies
All primary antibodies in this study except p85 were obtained from Cell Signaling Technology. Anti-p85 polyclonal antibody was generated in-house and has been described (21). Horseradish peroxidase–conjugated anti-mouse and anti-rabbit immunoglobulin G (IgG) antibody were purchased from Chemicon.
RNA interference
For dox-inducible shRNA-mediated knockdown of AKT isoforms, a tet-on shRNA/pLKO system was used. The hairpin sequences targeting AKT1, AKT2, and AKT3 have been validated, and the construction of tet-on AKT isoform shRNA/pLKO vectors has been described previously (22). Akt1, sense, 5′-CCGGGAGTTTGAGTACCTGAAGCTGCTCGAGCAGCTTCAGGTACTCAAACTCTTTTTG-3′; Akt2, sense, 5′-CCGGGCGTGGTGAATACATCAAGACCTCGAGGTCTTGATGTATTCACCACGCTTTTTG-3′; Akt3, sense, 5′- CCGGCTGCCTTGGACTATCTACATTCTCGAGAATGTAGATAGTCCAAGGCAGTTTTTG-3′. To produce lentiviral supernatants, 293T cells were cotransfected with control or shRNA-containing tet-on pLKO vectors, VSVG and psPAX2 for 48 hours. Cells stably expressing dox-inducible shRNA were cultured in medium containing puromycin (2 μg/ml). Gene knockdown was induced by incubating cells with 100 ng/mL dox for 48 to 72 hours.
Plasmids
For dox-inducible overexpression of AKT3, cells were infected with HA-AKT3/pTRIPZ lentiviral vector. Construction of HA-AKT3/pTRIPZ has been described previously (23).
Cell viability assays
Cells were seeded 24 hours before inhibitor treatment into 96-well plates at density of 5,000 cells per well in 100 μL medium. Cell viability was measured 48 hours after inhibitor treatment using the WST-1 assay (Clontech) according to the manufacturer's protocol.
In vitro kinase assays
Akt3 was immunoprecipitated from cell extracts and incubated with 100 ng of GSK3β peptide in the presence of 250 μmol/L cold ATP in a kinase buffer for 1 hour at 30°C. The kinase reaction was stopped by the addition of SDS-PAGE loading buffer, and the samples were assayed by immunoblotting.
Transwell invasion assays
Transwell filters (8-μm pore size; Corning) were coated with 1.5 μg Matrigel (BD Biosciences). A total of 1 × 105 cells in serum-free medium containing 0.1% BSA were added to upper Transwell chambers in triplicate. Conditioned medium from NIH 3T3 cells was used as chemoattractant, and was added to the lower chambers. After 7 hours of incubation at 37°C, non-invaded cells on Transwell filters were removed. Cells that had invaded and migrated to the bottom of the filters were fixed and stained using the Hema-3 stain set (Fisher HealthCare Protocol).
Quantitative real-time RT-PCR
Total RNA was isolated with an RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol. Reverse transcription was performed using random hexamers and multiscribe reverse transcriptase (Applied Biosystems). Quantitative real-time PCR was performed using an ABI Prism 7700 sequence detector. AKT3 primer: sense, 5′-GAAGAGGAGAGAATGAATTGTAGTCCA-3′; anti-sense, 5′-AGTAGTTTCAAATAGTCAAAATCATTCATTG-3′ (24); IGF1R primer: sense, 5′-TTCAGCGCTGCTGATGTG-3′; anti-sense, 5′-GGCTCATGGTGATCTTCTCC-3′ (25). PCR reactions were carried out in triplicate. Quantification of mRNA expression was calculated by the dCT method with GAPDH as the reference gene.
Copy number analysis with quantitative real-time PCR
Genomic DNA was isolated with the QIAamp DNA Mini Kit (Qiagen) according to the manufacturer's protocol. Real-time PCR was performed using an ABI Prism 7700 sequence detector. AKT3 primer: sense, 5′-CTGGACATCACCAGTCCTAGCTC-3′; anti-sense, 5′-ACCCTTGGCTGGTCTGGG-3′ (26); CEP17 primer: sense, 5′-GCTGATGATCATAAAGCCACAGGTA-3′; anti-sense, 5′-TGGTGCTCAGGCAGTGC-3′ (27). PCR reactions were carried out in triplicate. Quantification of copy number was calculated by the dCT method with CEP17 as the reference gene.
Immunoblotting
Cells were washed with PBS at 4°C and lysed in RIPA buffer [1% NP-40, 0.5% deoxycholic acid (SDC), 0.1% SDS,150 mmol/L NaCl, 50 mmol/L Tris-HCl (pH 7.5), proteinase inhibitor cocktail, 50 nmol/L calyculin, 1 mmol/L sodium pyrophosphate, 20 mmol/L sodium fluoride] for 15 minutes at 4°C. Cell extracts were precleared by centrifugation at 13,000 rpm for 10 minutes at 4°C, and protein concentration was measured with the Bio-Rad protein assay reagent using a Beckman Coulter DU-800 machine. Lysates were then resolved on 10% acrylamide gels by SDS-PAGE and transferred electrophoretically to nitrocellulose membrane (BioRad) at 100 V for 60 minutes. The blots were blocked in TBST buffer (10 mmol/L Tris-HCl, pH 8, 150 mmol/L NaCl, 0.2% Tween 20) containing 5% (w/v) nonfat dry milk for 30 minutes, 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 horseradish peroxidase–conjugated secondary antibody for 1 hour at room temperature. Membranes were washed 3 times and developed using enhanced chemiluminescence substrate (Pierce).
Results
Establishment and characterization of AKT inhibitor–resistant breast tumor lines
In vitro modeling of acquired resistance has successfully identified a resistance mechanism observed in cancer patients treated with kinase inhibitors. To explore mechanisms that mediate resistance to AKT inhibitors, we set out to model resistance to an allosteric AKT inhibitor, MK2206, currently in phase II clinical trials, using the luminal breast tumor cell line T47D, which harbors an activating PIK3CA mutation (H1047R). This cell line has been shown to be sensitive to MK2206 (28). MK2206-resistant derivative T47D lines were cultured either in gradually increasing doses (starting at 0.2 μmol/L) or at a constant concentration of 5 μmol/L inhibitor, until pools of cells growing in the presence of 5 μmol/L drug were established. The resulting pools of cells were termed T47D R (MK0.2-5; step-wise fashion) and T47D R (MK5; chronic high-dose fashion), respectively. Both resistant lines show resistance to MK2206 in cell viability assays, evidenced by >10-fold shifts in IC50 to MK2206 compared with the DMSO-treated parental pools (Fig. 1A). The T47D R lines also show cross-resistance to the ATP-competitive AKT inhibitor GDC0068 (Fig. 1A). In dose response studies, T47D R lines are broadly resistant to the inhibitory effect of MK2206 and GDC0068 at multiple nodes of the PI3K/AKT pathway, including pAKT S473, pPRAS40 T246, pGSK3β S9, and p4EBP1 S65 (Fig. 1B).
T47D AKT inhibitor–resistant cells show resistance to inhibition along the PI3K/AKT axis. A, T47D parental (black) and MK2206-resistant (green) cells were seeded to 96-well plates in the absence of MK2206 for 24 hours. Cells were then treated with MK2206 or GDC0068 for 48 hours. Cell viability was determined by WST assays and calculated relative to the untreated cells. Error bars, mean ± SEM, with n = 3. B, T47D parental and MK2206-resistant cells were seeded to plates in the absence of MK2206 for 48 hours. Cells were then treated with MK2206 or GDC0068 for 1 hour. Whole-cell lysates were subjected to immunoblotting for AKT signaling pathway components. p85 was immunoblotted as a loading control.
Upregulation of AKT3 as a novel mechanism of acquired resistance to AKT inhibitors
The three AKT isoforms have distinct functions in modulating phenotypes commonly associated with cancer. For example, whereas AKT1 is a breast cancer metastasis suppressor (29–31), AKT2 promotes invasion and metastasis of breast cancer in vitro and in vivo (32, 33). In addition, recent studies have also revealed a specific role of AKT3 in regulating the growth of TNBC (23). Resistant lines were profiled with immunoblot analysis to identify relative expression levels of AKT1, AKT2, and AKT3. Cells resistant to MK2206 markedly exhibit increased AKT3 expression levels when compared with parental cells (Fig. 2A). Conversely, expression of AKT1 and AKT2 is indistinguishable between parental and resistant cells. In addition, AKT3 expression is increased in a T47D R line established by culturing cells in gradually increasing doses of GDC0068 (R (GDC0.2-5); Fig. 2A), suggesting that AKT3 upregulation is not specific to a particular class of the AKT inhibitor. Importantly, upregulation of AKT3, but again neither AKT1 nor AKT2, is also observed in multiple resistant derivatives of the triple-negative cell line MDA-MB-468, including cells resistant to the ATP-competitive AKT inhibitor GSK690693 (Supplementary Fig. S1). These data indicate that upregulation of AKT3 is a general feature of acquired resistance to AKT inhibitors in breast cancer cell lines.
Upregulation of AKT3 in AKT inhibitor–resistant breast tumor cells. A, Western blot analysis of lysates from T47D parental and resistant cells. It has been shown that the ATP-competitive inhibitor GDC0068 locks AKT in a nonfunctional yet hyperphosphorylated state (19). The slower migrating protein bands of AKT1-3 in GDC0068-resistant cells represent the hyperphosphorylated form of AKTs. Protein levels were quantified with ImageJ from NIH software. The levels of protein are expressed as a ratio relative to the p85 protein in each sample (n = 3). B, mRNA levels of AKT3 and IGF1R in T47D parental and resistant cells were analyzed by quantitative real-time RT-PCR. The levels of mRNA are expressed as a ratio relative to the GAPDH mRNA (n = 3). C, AKT3 gene copy levels in MDA-MB-468 cells, T47D parental line (10 subclones), as well as resistant cells were analyzed by quantitative PCR. The copy number of AKT3 is expressed as a ratio relative to the CEP17 reference gene in each sample (n = 3). D, T47D parental and resistant cells were seeded to plates in the absence of MK2206 for 24 hours. Cells were then treated with JQ1 (0.3 μmol/L; Cayman Chemical) or iBET151 (1 μmol/L; Cayman Chemical) for 30 minutes, followed by MK2206 (1 μmol/L) for 48 hours. Whole-cell lysates were immunoblotted for the indicated antibodies. E, T47D parental and resistant cells were seeded to plates in the absence of MK2206 for 48 hours. AKT3 was immunoprecipitated from the cell lysate. In vitro kinase assay was then performed using GSK3β peptides as substrates. The kinase reaction was terminated, and samples were immunoblotted.
It has been shown that IGF1R levels are enhanced in breast cancer cells treated with AKT inhibitors (16). Consistent with this, we observe upregulation of IGF1R in the T47D R lines (Fig. 2A). Upregulation of both AKT3 and IGF1R protein expression in resistant cells correlates well with upregulation at the transcriptional level as assessed by real-time RT-PCR (Fig. 2B) and RNA-sequencing (RNA-seq; Supplementary Table S1). To examine if AKT3 is amplified in resistant cells, quantitative real-time PCR were performed on 10 subclones of T47D parental lines as well as the resistant lines. There is no significant difference of AKT3 copy number in resistant cells as compared with parental cells, whereas AKT3 amplification is found in the triple-negative MDA-MB-468 line (Fig. 2C). Because it has recently been demonstrated that members of the bromodomain and extra terminal domain (BET) family of proteins act epigenetically to regulate various components of the PI3K pathway, including IGF1R (34), we next evaluated the effects of BET proteins inhibition on the expression of AKT3 and IGF1R in T47D cells using two different small-molecule inhibitors, JQ1 and iBET. In T47D parental cells, AKT3 and IGF1R are induced after 48 hours of MK2206 treatment (Fig. 2D). The upregulation of AKT3 and IGF1R is inhibited when cells were pretreated with JQ1 or iBET. Similarly, both JQ1 and iBET lower the expression of AKT3 and IGF1R in the resistant cells. These data demonstrated that an epigenetic pathway regulates the induction of AKT3 and IGF1R in T47D R cells. To examine if the intrinsic kinase activity of AKT3 is changed in the resistant cells and if its activity can be inhibited by MK2206, we have performed in vitro kinase assays. MK2206 was removed from cells for 48 hours before harvesting of cell lysates. When an equal amount of Akt3 in the parental and resistant cells was immunoprecipitated, AKT3 expressed in the resistant cells exhibits lower ability to phosphorylate GSK3β peptides (Fig. 2E). These data suggest that AKT3 in the resistant cells is not hyperactive, and that it could be bound to and inhibited by MK2206, which has a long half-life (60–90 hours; ref. 6).
Knockdown of AKT3 in resistant cells restores sensitivity to the AKT inhibitor MK2206
To investigate the role of AKT3 in determining sensitivity to MK2206, AKT3 was expressed using a tetracycline-on (dox)-inducible system. Whereas T47D cells are sensitive to MK2206 (IC50: 0.17 μmol/L), ectopic expression of AKT3 results in a 16-fold increase in IC50 (2.71 μmol/L, P < 0.0001; Fig. 3A). By contrast, AKT3 depletion in T47D cells using shRNA leads to a 3.5-fold decrease in IC50 (0.06 μmol/L vs. 0.21 μmol/L in control cells; P < 0.0001; Fig. 3B). Dox itself has no effect on drug sensitivity because the IC50 of tumor cells containing empty vectors in the absence of dox is indistinguishable from dox-treated cells (Supplementary Fig. S2a). To further determine the specific role of AKT3 in inhibitor sensitivity, dox-inducible shRNA to deplete AKT1, AKT2, or AKT3 in parental and T47D R lines was used. Upon dox administration, AKT isoforms are depleted specifically and quantitatively (Fig. 3C). We have demonstrated a specific role for AKT3 in regulating TNBC spheroid growth using a three-dimensional (3D) cell culture system (23). In the luminal T47D model, depletion of AKT3 in parental cells results in a small, but statistically significant reduction in spheroid size (23% reduction; Fig. 3C). Conversely, depletion of AKT1 or AKT2 has minimal effect on spheroid growth. In resistant lines, depletion of AKT3 potently inhibits spheroid growth relative to parental cells (47% reduction; Fig. 3C), consistent with the notion that proliferation of resistant cells is driven by increased expression of AKT3. By contrast, the effect of AKT1 or AKT2 depletion on spheroid growth is much more modest. A 21% reduction in spheroid size is observed in AKT1-depleted cells, whereas there is no significant difference in spheroid size when AKT2 is depleted. We next assessed the contribution of AKT isoforms in determining sensitivity to MK2206. In parental lines, whereas AKT1 or AKT2 depletion has no effect on sensitivity to MK2206, loss of AKT3 results in a 4.8-fold decrease in IC50 (Fig. 3D). Importantly, depletion of AKT3 restores sensitivity of T47D R (MK5) to MK2206 to levels equivalent to parental cells (Fig. 3D). In contrast, knockdown of AKT1 or AKT2 in the resistant cells by using shRNAs targeting two distinct regions of the AKT1 and AKT2 transcripts did not affect MK2206 sensitivity (Fig. 3D; Supplementary Fig. S2b). Similarly, AKT3 depletion in another T47D R line (MK0.2-5) sensitizes cells to MK2206 in a dose-dependent manner (Supplementary Fig. S3). We also examined if the activation of AKT3 in resistant cells is mediated by IGF1R. Treatment of resistant cells with an IGF1R inhibitor, AEW541, greatly reduces the phosphorylation of AKT and PRAS40, as well as cell viability (Fig. 3E), suggesting that the IGF1R pathway drives AKT3 activation in T47D R cells. Taken together, these findings demonstrate that AKT3 upregulation in breast cancer cells confers resistance to the AKT inhibitor MK2206.
AKT3 expression determines sensitivity of breast cancer cells to the AKT inhibitor. A, T47D cells infected with tet-on AKT3/pTRIPZ lentiviral vector were treated with dox for 3 days. Cell viability was assessed by WST assays. Data, mean ± SEM, n = 2. Cell lysates were analyzed by immunoblotting. B, T47D cells expressing tet-on AKT3 shRNA were treated with dox for 3 days. Cell viability was calculated relative to the untreated cells. Data, mean ± SEM; n = 3. Knockdown of AKT3 was confirmed by Western blot analysis. C, T47D parental and resistant cells containing tet-on AKT1 (#1), AKT2 (#1), or AKT3 shRNA were grown in 3D culture for 5 to 8 days in the presence or absence of dox. Spheroid size was quantified in pixel area using ImageJ and depicted in the bar graph. Error bars, mean ± SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001 (Student t test, n ≥ 40). Knockdown of AKT isoforms was confirmed by treating cells with dox for 72 hours, followed by immunoblotting. D, T47D parental and resistant cells containing tet-on AKT1 (#1), AKT2 (#1), or AKT3 shRNA were treated with dox for 3 days, followed by preforming WST cell viability assays. Data, mean ± SEM; n = 3. E, T47D-resistant cells were seeded to plates in the absence of MK2206 for 48 hours. Cells were then treated with AEW541 (1 μmol/L) for 1 hour, followed by MK2206 (0.1 μmol/L) for 1 hour. Whole-cell lysates were analyzed by immunoblotting. To assess cell viability, T47D-resistant cells were seeded to 96-well plates in the absence of MK2206 for 24 hours. Cells were then treated with AEW541 (1 μmol/L) and/or MK2206 (0.15 μmol/L) for 48 hours, followed by WST assays. Data, mean ± SEM. *, P < 0.05; **, P < 0.01; ns, not significant (Student t test, n = 3).
Cellular reprogramming and reversibility of MK2206-resistant T47D cells
Next, to determine if genetic or epigenetic mechanisms are responsible for the establishment of AKT inhibitor resistance, we examined if the resistance phenotype is reversible upon drug discontinuation. MK2206 was removed from resistant lines for 3 weeks prior to analysis. Compared with cells continuously treated with MK2206, resistant cells that had drug removed partially reacquire AKT inhibitor sensitivity (Fig. 4A). Furthermore, when the cells of drug removal were rechallenged with MK2206, they regain resistance rapidly after 2 weeks of drug exposure (Fig. 4B). The AKT signaling profile of MK2206-withdrawn cells is comparable with parental cells (Fig. 4C). Whereas phosphorylation of PRAS40 and 4EBP1 is largely refractory to the inhibitory effects of MK2206 in T47D R lines, upon drug removal for 9 days, MK2206 inhibits PRAS40 and 4EBP1 phosphorylation to levels similar to those seen in parental cells. The reversibility of both signaling and resistance phenotypes is indicative of an epigenetic mechanism.
Induction of EMT in AKT inhibitor–resistant cells is associated with upregulation of AKT3. A, to assess the reversibility of the effects of chronic AKT inhibition, a fraction of T47D-resistant cells was split from the MK2206 culture and maintained in DMSO for 3 weeks. Cells were then treated with MK2206 (2.5 μmol/L) for 48 hours, followed by WST cell viability assays. Data, mean ± SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant (Student t test, n = 3). B, to assess the ability of cells to regain resistance after drug removal, a fraction of cells that have been cultured in the absence of drug for 2 weeks were rechallenged with MK2206 for 1 or 2 weeks, followed by WST assays. wk, weeks. C, T47D parental and resistant cells, as well as resistant cells that have been cultured in the absence of MK2206 for 9 days were seeded to plates in the absence of MK2206 for 48 hours. Cells were then treated with MK2206 (0.1 or 1 μmol/L) for 1 hour. Whole-cell lysates were immunoblotted for the indicated antibodies.
Because increasing evidence indicates an important role for the EMT in drug resistance (35), and there is clear evidence for epigenetic reprogramming in epithelial–mesenchymal plasticity (36), we next investigated if the resistant cells show hallmarks of EMT. Expression of the epithelial gene E-cadherin is decreased whereas the mesenchymal markers N-cadherin and vimentin are increased in T47D R lines, relative to parental cells (Fig. 4C). In Boyden chamber invasion assays, T47D R lines also exhibit enhanced invasiveness (Fig. 5A), consistent with the acquisition of EMT. In addition, expression of E-cadherin is inversely correlated with the expression levels of AKT3, whereby E-cadherin expression is decreased and AKT3 expression is increased in T47D R lines, again compared with parental cells. Conversely, upregulation and downregulation of E-cadherin and AKT3, respectively, is observed upon drug removal in resistant cells (Fig. 4C). These findings indicate that acquisition of AKT inhibitor resistance in T47D tumor cells is accompanied by an EMT program that is associated with upregulation of AKT3.
AKT3 regulates invasiveness of AKT inhibitor–resistant breast tumor cells. A, T47D parental and resistant cells were subjected to a Transwell invasion assay. Relative invasion (y axis) = ratio of the number of invaded cells in test versus control. *, P < 0.05 (Student t test, n = 3). B, T47D parental and resistant cells expressing tet-on AKT isoform shRNA were grown in 3D culture for 8 to 15 days in the presence or absence of dox. Morphology of spheroids is shown in the representative phase-contrast images.
AKT3 depletion promotes epithelial phenotype in MK2206-resistant breast tumor spheroids
We next explored whether AKT3 regulates epithelial characteristics of AKT inhibitor–resistant cells. T47D R (MK5) and parental cells containing tet-on AKT3 shRNA were cultured in 3D. In agreement with published findings (37), T47D cells form normal round 3D spheroids (Fig. 5B). Depletion of AKT3 has minimal effect on spheroid morphology. Consistent with changes in EMT markers, MK2206-resistant spheroids display invasive morphogenesis, and AKT3 depletion reverses the resistant spheroid phenotype to organized round structures (Fig. 5B).
Discussion
Despite the initial response of tumors to targeted therapeutic agents, most patients relapse and develop resistance, leading to limited clinical benefit. Various resistance mechanisms have been identified for inhibitors targeting the EGFR, BRAF, and MAPK pathways (38, 39). By contrast, relatively few mechanisms have been identified and implicated in resistance to PI3K pathway inhibitors (8–14). Importantly, despite over 80 current clinical trials for various AKT inhibitors, resistance mechanisms in cells and patients treated chronically with these compounds have yet to be identified. In the present study, we set out to identify and delineate resistance mechanisms to an AKT inhibitor, MK2206, using the T47D breast cancer cell line that harbors a PIK3CA mutation. We show that AKT3, but not AKT1 or AKT2, is upregulated at the mRNA and protein level in resistant cells. Functional studies show that acquisition of resistance is specifically due to the increased expression of AKT3, as evidenced by the restoration of sensitivity to MK2206 in resistant cells upon AKT3 depletion. To our knowledge, this is the first report of a chronic resistance model for an AKT inhibitor in cancer.
In our breast tumor model of resistance, we observe AKT3 upregulation not only with the allosteric AKT inhibitor MK2206, but also with the ATP-competitive inhibitors GDC0068 and GSK690693. Whereas MK2206 has ∼5-fold lower IC50 toward recombinant AKT1 and AKT2 than AKT3 (AKT1: 8 nmol/L; AKT2: 12 nmol/L; AKT3: 65 nmol/L; ref. 4), the ATP-competitive inhibitors have similar potency toward all three AKT isoforms (GDC0068: AKT1: 5 nmol/L; AKT2: 18 nmol/L; AKT3: 8 nmol/L, GSK690693: AKT1: 2 nmol/L; AKT2: 13 nmol/L; AKT3: 9 nmol/L; refs. 40, 41). In addition to the luminal subtype of breast cancer, AKT3 is also overexpressed in resistant cells of the triple-negative subtype. In this context, a specific role of AKT3 in regulating TNBC growth has been demonstrated in vitro and in vivo, and the cell cycle inhibitor p27 appears to be critical for the ability of AKT3 in modulating proliferation (23). The underlying molecular mechanism(s) for AKT3-mediated drug resistance in breast cancer is yet to be defined. In human glioblastoma, AKT3 is highly expressed and its expression is significantly correlated with DNA repair genes (42). Moreover, in a mouse model of glioma, AKT3 overexpression enhances DNA repair pathways and confers resistance of tumor cells to radiation and temozolomide. It will be interesting to determine if DNA repair is involved in conferring resistance to AKT inhibitors. It has also been reported that in response to targeted agents, cancer cells develop resistance by upregulating both the levels of the targeted kinase and increasing intrinsic kinase activity (43). To examine this, we performed in vitro kinase assays, and our data indicated that AKT3 expressing in the MK2206-resistant cells is not hyperactive and could still be inhibited by MK2206. Although upregulation of AKT3 as a resistance mechanism has yet to be verified in matched patient biopsies of pre- and posttreatment of AKT inhibitors, copy number alteration analysis of TNBC clinical samples shows that AKT3 is amplified in ∼15% of chemotherapy-resistant tumors (44). In addition, in a systematic functional screen performed in breast cancer cell lines, AKT3 is one of the genes shown to support proliferation and survival of tumor cells upon PI3K inhibition (13). In a separate study using HER2+ mammary tumor cells from Balb-neuT mice, AKT3 depletion upregulates estrogen receptor alpha and sensitizes tumor cells to the estrogen receptor modulator tamoxifen (45). In addition to breast cancer and glioblastoma, AKT3 has also been implicated in resistance of other aggressive tumors. In metastatic melanoma, AKT3 plays a critical role in mediating resistance to an inhibitor targeting mutant BRAF (46). BRAF inhibitor–induced upregulation of BH3-only proapoptotic proteins Bim-EL and Bmf is attenuated by ectopic expression of AKT3. Our current work focuses primarily on the T47D breast tumor line, which has a PIK3CA mutation, whether the isoform-specific function of AKT3 plays a critical role in acquired resistance in other breast tumor lines, contexts, and tumor types awaits further studies.
To explore genes that may be involved in the regulation of AKT3 and IGF1R, as well as to examine global changes of gene expression, we analyzed the transcriptome of T47D parental and MK2206-resistant cells using RNA-seq. We identified 525 and 402 protein-coding genes that are upregulated (log of change > 0.5) or downregulated (log of change < 0.5), respectively, in MK2206-resistant lines (Supplementary Table S1). In addition to AKT3 and IGF1R, a few other genes in the PI3K/AKT pathway are found to be upregulated in the resistant lines, including HER3, insulin receptor substrate 2 (IRS2), and discoidin domain receptor 1 (DDR1). Agreeing with previous studies that implicated a role for HER3 in AKT inhibitor resistance (16–18), HER3 expression is increased in our resistant lines. An upregulation of IRS2, a cytoplasmic adaptor protein that mediates the activation of PI3K (47), suggests a feedforward loop activating AKT signaling in the resistant cells. In addition, DDR1, which has been shown to interact with and positively regulate IGF1R expression, is upregulated. This collagen receptor tyrosine kinase has also been demonstrated to promote EMT and cancer progression (48). Two other upregulated genes that warrant further studies are the transcription factor FOXD3, where its consensus binding sequence is found on the AKT3 promoter, as well as BCL2, an antiapoptotic protein that drives cell survival.
The reversibility of drug resistance in the context of increased AKT3 expression suggests that epigenetic alterations are responsible. Indeed, induction of AKT3 in MK2206-resistant cells is regulated epigenetically by BET proteins. Our RNA-seq data also showed that one of the BET proteins, BRD1, is upregulated in the MK2206-resistant lines (Supplementary Table S1). It would be interesting to examine if BRD1 is the major protein mediating the epigenetic regulation of AKT3. The EMT phenotype is associated with both intrinsic and acquired resistance to various kinase inhibitors (49, 50), and is subject to epigenetic regulation. In our resistant lines, we observe changes in the expression of multiple EMT-associated genes, accompanied by altered AKT3 expression. Consistent with this, expression of AKT3 is highly correlated with EMT activators such as ZEB1 in clinical breast tumors (Spearman's correlation: 0.61; cBioportal.org; refs. 51, 52). Our resistant lines also show mesenchymal-associated characteristics, including the enhancement of invasiveness. We observe that MK2206-resistant spheroids are more invasive than spheroids from parental cells. Notably, AKT3 depletion reduces invasiveness of cells in resistant tumor spheroids. It has been shown that AKT1 and AKT2 have opposing roles in breast cancer cell invasion and metastasis (53). By contrast, the role of AKT3 in this phenotype has not been examined in detail, although the present findings are consistent with previous studies showing that unlike AKT1 depletion, silencing AKT3 does not result in an invasive morphology of MCF10A spheroids (23). It will also be interesting to examine if inhibition of AKT3 reduces metastasis of resistant cells in an in vivo setting.
These findings have important clinical implications because AKT inhibitors in clinical development target all three isoforms, and an AKT3-selective inhibitor has yet to be developed. Given the undesired metastatic phenotype of breast cancer cells upon inhibition of AKT1, and the critical role of AKT2 in regulating glucose homeostasis, combined with studies highlighting a role for AKT3 in TNBC growth, these findings advocate for treating TNBC and other breast tumors overexpressing AKT3 with an AKT3-selective inhibitor to curb toxicities (23). Our data also reveal an isoform-selective role of AKT3 in an in vitro resistance model. Based on the concept that it is preferable to prevent the emergence of resistance, rather than treating resistance once it develops, if AKT3, but not AKT1 or AKT2, induction is observed in patients that develop resistance, these findings provide a rationale for the development of potent AKT3-selective small molecule inhibitors for treating breast tumors in which AKT3 is a driver for growth.
Disclosure of Potential Conflicts of Interest
Y.R. Chin reports receiving a commercial research grant from Pfizer CTI. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: Y.R. Chin
Development of methodology: Y.R. Chin
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Stottrup, T. Tsang, Y.R. Chin
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Stottrup, T. Tsang, Y.R. Chin
Writing, review, and/or revision of the manuscript: Y.R. Chin
Study supervision: Y.R. Chin
Grant Support
This study was supported in part by a grant from the NIH National Cancer Institute (Y.R. Chin and C. Stottrup; R00CA157945), a grant from the V Foundation for Cancer Research (Y.R. Chin), and a sponsored research grant from Pfizer CTI (Y.R. Chin and T. Tsang).
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.
Acknowledgments
The thank Samuel Klempner for advice in generating drug-resistant tumor lines, Ruslan Sadreyev and Fei Ji from the Nextgen Sequencing Core of Massachusetts General Hospital for their assistance in analyzing the RNA sequencing data, members of the Chin laboratory for discussions, and members of Alex Toker laboratory for advice.
Footnotes
Note: Supplementary data for this article are available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/).
- Received September 16, 2015.
- Revision received May 24, 2016.
- Accepted May 27, 2016.
- ©2016 American Association for Cancer Research.
References
Volume 15, Issue 8
