Research ArticleCancer

Nuclear Phospho-Akt Increase Predicts Synergy of PI3K Inhibition and Doxorubicin in Breast and Ovarian Cancer

See allHide authors and affiliations

Science Translational Medicine  08 Sep 2010:
Vol. 2, Issue 48, pp. 48ra66
DOI: 10.1126/scitranslmed.3000630


The phosphatidylinositol 3-kinase (PI3K)–Akt signaling pathway is frequently disrupted in cancer and implicated in multiple aspects of tumor growth and survival. In addition, increased activity of this pathway in cancer is associated with resistance to chemotherapeutic agents. Therefore, it has been hypothesized that PI3K inhibitors could help to overcome resistance to chemotherapies. We used preclinical cancer models to determine the effects of combining the DNA-damaging drug doxorubicin with GDC-0941, a class I PI3K inhibitor that is currently being tested in early-stage clinical trials. We found that PI3K inhibition significantly increased apoptosis and enhanced the antitumor effects of doxorubicin in a defined set of breast and ovarian cancer models. Doxorubicin treatment caused an increase in the amount of nuclear phospho-AktSer473 in cancer cells that rely on the PI3K pathway for survival. This increased phospho-AktSer473 response to doxorubicin correlates with the strength of GDC-0941’s effect to augment doxorubicin action. These studies predict that clinical use of combination therapies with GDC-0941 in addition to DNA-damaging agents will be effective in tumors that rely on the PI3K pathway for survival.


The phosphatase and tensin homolog deleted from chromosome 10 (PTEN)–phosphatidylinositol 3-kinase (PI3K) signaling pathway is pivotal to several cellular processes including cell cycle progression and survival (1, 2). Aberrant activation of this pathway has been widely implicated in many cancers, and increased activity of this pathway is often a hallmark of resistance to conventional treatment modalities (35). In addition, increased activation of the PI3K pathway is crucial to cancer cell survival under the stressful tumor environment of limited nutrients, oxygen, and reduced pH (6). For these reasons, it has been suggested that cancer therapeutics may have enhanced efficacy when used in combination with PI3K pathway inhibitors (79).

Therapies designed to inflict DNA damage on a proliferating cancer cell are some of the most commonly used oncology treatments. Doxorubicin is an anthracycline antibiotic used in the treatment of a wide range of cancers, including breast and ovarian. It interacts with DNA by intercalation and inhibits the progression of topoisomerase II (10). After breaking the DNA chain for replication, the topoisomerase II complex is stabilized by doxorubicin, preventing the double helix from being repaired and thus stopping the process of replication. As a consequence of these events, proliferating cells usually respond by arresting in the G2 phase of the cell cycle in an attempt to repair the DNA damage (11). These data imply that abrogation of the G2 checkpoint would significantly enhance the activity of cytotoxic agents such as doxorubicin.

Many of the functions of the PI3K pathway are mediated by the serine-threonine kinase Akt, which is activated by many cellular stimulators and stressors. Two well-studied phosphorylation sites exist on Akt at Thr308 and Ser473, and phosphorylation of either site results in increased Akt enzymatic activity. Recent studies have shown that DNA damage can induce phosphorylation of Akt at the Ser473 site by DNA-dependent protein kinase (DNA-PK) in response to DNA damage, thereby promoting survival and therapeutic resistance (1216). In these studies, we explored the use of a highly selective class I PI3K inhibitor, GDC-0941, in combination with the DNA-damaging agent doxorubicin. We discovered that, in most cell lines where combination synergy was observed, there was an increase in phospho-AktSer473 in response to the doxorubicin treatment alone. Synergy correlated well with the phospho-AktSer473 induction by doxorubicin, whereas it did not correlate well with other potential markers of PI3K inhibitor sensitivity. These results suggest that a selective PI3K inhibitor could be particularly effective as a potentiator of chemotherapy in the set of tumors that use the Akt pathway to survive.


GDC-0941 and doxorubicin combine synergistically in a subset of breast cancer and ovarian cancer lines

GDC-0941 is a potent small-molecule inhibitor of class I PI3K isoforms and displays excellent selectivity against mammalian target of rapamycin (mTOR), DNA-PK, and a panel of >228 kinases (17). The compound shows potent growth-inhibitory activity as a single agent in vitro and in human xenograft models (table S1) (17). We used a panel of 20 breast cancer and 12 ovarian cancer cell lines to investigate the therapeutic potential of combinations of the anthracycline doxorubicin and the selective PI3K inhibitor GDC-0941. Doxorubicin and GDC-0941 were tested as single agents in the cell line panel and assayed in a CellTiter-Glo cell viability assay to determine their half-maximal effective concentration (EC50). For combination studies, the drugs were combined at a constant ratio of four times their EC50 values and evaluated in nine serial 1:2 compound dilutions. The data were analyzed to calculate a combination index (CI), which indicates the degree of synergy (18). Plots of the cell viability effects of these two drugs in two representative cell lines are shown in Fig. 1A. In MCF7-neo/HER2, the combination of the two agents markedly reduced viability compared to either agent alone, whereas in HDQ-P1, the combination was no more effective than either agent alone. Fractional analysis in the MCF7-neo/HER2 line indicates that combination synergy occurs over a broad range of concentrations. The CI at the EC50 concentration was calculated for all 32 lines (table S1). A CI of <0.3 indicates strong synergy, a CI between 0.3 and 0.7 indicates synergy, a CI between 0.7 and 1.1 indicates weak synergy to additivity, and a CI of >1.1 indicates antagonism (19). The CI values indicated strong synergy of the two compounds in MCF7-neo/HER2 cells and very weak additivity in HDQ-P1 cells. Overall, combinations in the cell line panel resulted in moderate or strong synergy (CI <0.7) in 16 of the 32 breast and ovarian cancer cell lines (table S1).

Fig. 1

Cell viability and pathway effects of GDC-0941 and doxorubicin in breast cancer lines. (A) GDC-0941 combines well with doxorubicin (Dox) in MCF7-neo/HER2 cells. The effect on viability of GDC-0941 and doxorubicin as single agents is shown in the blue and red lines, respectively, in two breast cancer cell lines. The combination effect of the two drugs is shown in the black dotted lines. (B) Increased apoptosis with the GDC-0941 and doxorubicin combination. Immunoblots from 24-hour–treated cell samples showing protein concentrations for phospho-AktSer473 (pAktSer473), phospho-SGK3Thr320 (pSGK3Thr320), cyclin D1, cleaved PARP, and cleaved caspase 3. Loading was assessed with an antibody to β-actin. (C) Doxorubicin increases phospho-Akt in MCF7-neo/HER2 cells. Protein concentrations of phospho-AktSer473, phospho-AktThr308, and total Akt (tAkt) were determined by enzyme-linked immunosorbent assay (ELISA) after treatment with 2 μM doxorubicin. Data are shown as a phospho-Akt/total Akt increase relative to samples that did not receive doxorubicin. (D) Dose-dependent increase in phospho-AktSer473 response to doxorubicin after 24 hours. Data are plotted as a phospho-AktSer473/total Akt increase relative to samples that did not receive doxorubicin. Error bars indicate ±SEM.

The effect of the compounds on the downstream PI3K pathway markers phospho-AktSer473 and phospho-SGK3Thr320 was investigated with single-agent and combination drug treatments at their EC50 concentrations at a 24-hour time point (Fig. 1B). SGK3 has overlapping substrates with Akt but has a PX domain, not a PH domain (20). Some cancer cell lines exhibit low levels of phospho-Akt with high amounts of phospho-SGK3 (21), so both SGK and Akt were assessed to determine whether lack of synergy in a subset of lines was due to SGK activity in the absence of Akt activity. We observed a direct correlation between levels of phospho-Akt and phospho-SGK3, and GDC-0941 caused a decrease of these markers in both cell lines in the presence or absence of doxorubicin, as expected for an inhibitor of PI3K. We detected an increase in phospho-AktSer473 in response to doxorubicin alone in the MCF7-neo/HER2 cell line, but not in the HDQ-P1 cell line. An increase in Akt phosphorylation leading to a survival response has been described previously in normal and cancer cells (1215, 22). To determine the kinetics of Akt activation in MCF7-neo/HER2 cells in response to DNA damage, we treated the cells with doxorubicin and measured the phosphorylation status at the kinase domain AktThr308 site in addition to the C-terminal AktSer473 site. A significant increase in phosphorylation at both of these sites was detected by 1 hour, and the amount of phosphorylation increased over time (Fig. 1C). To determine the amount of DNA damage required to phosphorylate Akt, we titrated the concentration of doxorubicin applied to MCF7-neo/HER2 cells. We observed a significant increase in Akt phosphorylation at doses as low as 0.25 μM (Fig. 1D), which is below the EC50 for doxorubicin (0.5 μM) in this cell line.

To investigate the effects of drug combinations, we looked at the cell cycle and apoptotic markers—cyclin D1, cleaved poly(ADP-ribose) polymerase (PARP), and cleaved caspase 3—after 24 hours of treatment (Fig. 1B). Cyclin D1 is expressed in proliferating cells and helps control the progression of cells through the cell cycle. Reduction of cyclin D1 was observed in both cell lines with single-agent treatments. This is consistent with cell cycle profiles that we observed in flow cytometry assays indicating G1 and G2 arrests with GDC-0941 and doxorubicin treatments, respectively (fig. S1). With the combined presence of both agents, however, there was complete loss of the cyclin D1 marker in the MCF7-neo/HER2 cell line, but not in HDQ-P1, where no synergy was observed.

PARP is one of the main cleavage targets of caspase 3, and cleaved PARP serves as a marker of apoptotic cells (23). Although only minimal amounts of cleaved PARP and caspase 3 were detected with single-agent treatments, there was a substantial increase of these markers in the MCF7-neo/HER2 cell line with the combination. Thus, in the MCF7-neo/HER2 cell line, both cell cycle perturbations and apoptosis may be drivers of synergy.

GDC-0941 and doxorubicin combination allows for cell cycle advance and increased DNA damage

Cdc2 kinase regulates cell entry into mitosis. An important step in activating cdc2 during the progression into mitosis is dephosphorylation of Tyr15, which is carried out by the cdc25 phosphatase [reviewed in (24)]. Phosphorylation of cdc2 at Tyr15 was detected in MCF7-neo/HER2 cells with doxorubicin treatment as expected with a G2 block (Fig. 2A). This phosphorylation, however, was decreased with GDC-0941 alone and with combination treatments. Analyses of phospho–histone H3Ser10 levels indicated that more cells proceeded through mitosis in the MCF7-neo/HER2 cell line with the combination than untreated cells. Cyclin B amounts were slightly increased with doxorubicin treatment and decreased with GDC-0941 or the combination. As expected due to cell cycle arrests at G1 for GDC-0941 and G2 for doxorubicin treatments, phospho–histone H3Ser10 levels were reduced with these single-agent treatments. Phospho–histone H3Ser10 was increased by combination treatment, indicating that many of the cells were not arrested in the cell cycle but were actively proceeding through mitosis. We did not detect an increase in this marker with HDQ-P1 cells.

Fig. 2

Cell cycle and DNA damage response to GDC-0941 and doxorubicin combinations. (A) Cell cycle markers are changed with single-agent and combination treatments. Nuclear lysates from MCF7-neo/HER2 and HDQ-P1 cells were analyzed by Western blot after 4-hour treatments for phospho-AktSer473, Akt, phospho-cdc2Tyr15, cyclin B, and phospho–histone H3Ser10. Nuclear extracts were confirmed with antibodies to α-tubulin and lamin A/C. (B) Increased DNA damage with the GDC-0941 and doxorubicin combination. The extent of DNA damage after 24-hour single-agent and combination treatments was determined by comet assay. Red arrows indicate comet tails.

Cells progressing through mitosis with DNA damage frequently undergo the type of cell death triggered by mitotic catastrophe (25). To further investigate the consequences of increased mitosis in the presence of a cytotoxic drug, we assessed the extent of DNA damage with the drug treatments (Fig. 2B). Cancer cells were treated at EC50 doses for 24 hours as single agents and in combination. We observed a detectable increase in DNA damage (comet tails present) with the combination in MCF7-neo/HER2 cells, but not with single-agent conditions.

Akt has a large number of substrates (26). We investigated the consequences of increased phosphorylation of Akt in response to DNA damage in whole-cell, cytoplasmic, and nuclear extracts (figs. S2 and S3). The only Akt substrate that changed in the MCF7-neo/HER2 line and not HDQ-P1 was the Ser642 site on Wee1. Akt phosphorylates Wee1 at this site, and the consequences of this phosphorylation for Wee1 activity have been controversial (27). Some studies have shown that it promotes 14-3-3 binding and increases Wee1 activity (28, 29). It has been shown that the phosphorylation results in cytoplasmic localization of Wee1, thereby inhibiting its ability to phosphorylate cdc2 and block cell cycle advance through mitosis (30). In our studies, we did not observe changes in Wee1 protein concentrations in nuclear or cytoplasmic cell fractions with doxorubicin treatment (fig. S3). Treatment with GDC-0941 reduced phospho-Wee1Ser642 levels induced by doxorubicin.

Increase in AktSer473 phosphorylation correlates with GDC-0941–doxorubicin combination synergy

It was intriguing that a specific subset of breast and ovarian cancer lines showed strong combination synergy. To follow up on this discovery, we searched for biomarker predictors for the synergy between GDC-0941 and doxorubicin, assessing mutational status of the PIK3CA and p53 genes, PTEN protein loss, basal phosphorylation levels of signaling proteins within the PI3K pathway, and breast cancer subtypes. Each of those potential markers was then correlated with the degree of synergy across the cell line panel (fig. S4). There was no clear association of any of these markers with the combination activity.

Because the mechanism of combination effects includes a significant increase in apoptosis, we then assessed cells treated with doxorubicin for their reliance on the PI3K pathway for survival. All 32 cancer cell lines were treated with 2 μM doxorubicin for 24 hours, and levels of phospho-AktSer473 and total Akt were determined. A range of responses was observed because some cancer lines showed increased phospho-AktSer473, some showed decreased phospho-AktSer473, and some did not change (Fig. 3A). The cell lines showing more than a 50% increase in phospho-AktSer473 were more resistant to doxorubicin (Fig. 3B). However, there was no difference in the GDC-0941 EC50 values between the group with a less than 50% and that with a more than 50% increase in phospho-AktSer473 (Fig. 3B). The cell lines showing more than 50% increase in phospho-AktSer473 upon doxorubicin treatment had a significantly lower CI in both tumor tissue types (breast mean CI, 0.35; ovarian mean CI, 0.23) than those that did not (breast mean CI, 0.82; ovarian mean CI, 0.87) (Fig. 3C). Notably, this demonstrates that a robust Akt phosphorylation response to doxorubicin treatment is predictive for strong synergism of GDC-0941 and doxorubicin.

Fig. 3

Doxorubicin induced changes in phospho-AktSer473 in breast and ovarian cell lines. (A) Breast and ovarian cancer cell lines were treated with 2 μM doxorubicin for 24 hours, and in replicate samples the amounts of phospho-AktSer473 and total Akt protein concentrations were determined by ELISA. (B) Increases in phospho-AktSer473 correlate with doxorubicin resistance. The EC50 for single-agent doxorubicin and GDC-0941 treatments in each cell line is plotted against the AktSer473 phosphorylation increase. (C) Increases in phospho-AktSer473 correlate with combination efficacy. The CI at the EC50 concentration for GDC-0941 and doxorubicin in breast and ovarian cancer lines is plotted against the AktSer473 phosphorylation increase. The CI is divided into a greater than or less than 1.5-fold increase in phospho-AktSer473/total Akt in response to doxorubicin. The horizontal line indicates the median CI in each group.

AktSer473 phosphorylation after doxorubicin treatment occurs in the nucleus and is mediated by DNA-PK

We next investigated the mechanism of Akt phosphorylation in response to doxorubicin. PI3K-related protein kinase (PIKK) family members such as ATM (ataxia-telangiectasia mutated), ATR (ATM- and Rad3-related), and DNA-PKcs (DNA-PK catalytic subunit) are activated in response to DNA damage and function to relay signals responsible for cell cycle arrest and DNA repair (31). Previous studies with human umbilical cord endothelial cells or mouse embryonic fibroblasts have shown that AktSer473 phosphorylation in response to DNA-damaging agents occurs in the nucleus and is mediated by DNA-PK (12). To determine whether the response we observed in tumor cell lines was by this mechanism, we isolated nuclear and cytoplasmic extracts following a course of doxorubicin treatment in MCF7-neo/HER2 cells. Within 1 hour, we detected a significant increase in phospho-AktSer473 in the nucleus, but not in the cytoplasm (Fig. 4A). The amount of nuclear phospho-AktSer473 increased significantly up to 24 hours. We also detected a smaller and delayed increase in cytoplasmic phospho-AktSer473 by 8 hours. In total, the fractionation study data indicated that the phospho-AktSer473 increase occurs predominantly in the nucleus.

Fig. 4

GDC-0941 and NU-7026 effects on the protein concentrations of phospho-AktSer473 after doxorubicin treatment in MCF7-neo/HER2 cells. (A) Increase in phospho-AktSer473 after 0.6 μM doxorubicin treatment occurs in the nucleus. Phospho-AktSer473 and total Akt protein concentrations were determined by ELISA in replicate samples. Data are shown as a phospho-AktSer473/total Akt increase relative to samples that did not receive doxorubicin. (B and C) GDC-0941 reduces phospho-Akt with doxorubicin treatments. MCF7-neo/HER2 cells were treated for 1 hour with doxorubicin (0.6 μM), GDC-0941 (0.5 μM), or NU-7026 (1 μM) as single agents or in combination, and lysates were fractionated. Relative amounts of phospho-AktSer473, phospho-AktThr308, and total Akt proteins were determined by ELISA in replicate samples. Data are shown as a phospho-Akt/total Akt increase relative to samples that did not receive treatments. (D) GDC-0941 reduces cytoplasmic phospho-AktSer473 induced by EGF. MCF7-neo/HER2 cells were treated for 1 hour with EGF (10 nM), GDC-0941 (0.5 μM), or NU-7026 (1 μM) as single agents or in combination, and lysates were fractionated. Protein concentrations for phospho-AktSer473 and total Akt were determined by ELISA in replicate samples. Data are shown as a phospho-AktSer473/total Akt increase relative to samples that did not receive treatments.

Using the small-molecule DNA-PK inhibitor NU-7026, we next investigated whether the increases in phospho-AktSer473 were due to DNA-PKcs activity. This molecule is selective and potent for DNA-PK and does not show activity against the other PIKK family kinases activated in response to DNA damage, namely, ATM or ATR (32). NU-7026 treatment for 1 hour had no effect on phospho-AktSer473 or phospho-AktThr308 in cytoplasmic extracts (Fig. 4, B and C). NU-7026 treatment was able, however, to strongly inhibit the phospho-AktSer473 increase in nuclear extracts from doxorubicin-treated cells, indicating that DNA-PKcs phosphorylates Akt at Ser473 in response to DNA damage. Inhibition of Akt phosphorylation at the Thr308 site was not observed with NU-7026 treatment (Fig. 4C). We also wanted to determine the influence of 1-hour treatment of GDC-0941 on phospho-AktSer473 and phospho-AktThr308 in the cytoplasm and nucleus in response to DNA damage. In both fractions, GDC-0941 treatment decreased the amount of Akt phosphorylation at both sites (Fig. 4, B and C). The Akt phosphorylation decrease in the cytoplasmic fraction was equivalent in cells treated with GDC-0941 regardless of doxorubicin treatment. However, in the nuclear fraction, GDC-0941 could more robustly decrease phospho-AktSer473 in the absence of doxorubicin. Thus, GDC-0941 decreased the relative amounts of phospho-AktSer473 but could not fully block Akt phosphorylation by DNA-PK in response to doxorubicin.

To further investigate the effect of NU-7026 and GDC-0941 on AktSer473 phosphorylation in a growth factor–driven manner, we stimulated the cells with epidermal growth factor (EGF). As expected, EGF increased phospho-AktSer473 levels in the cytoplasm and not in the nucleus (Fig. 4D). Appropriately, NU-7026 had no effect as a single agent or in combination with EGF in the cytoplasmic or nuclear fractions. However, GDC-0941 treatment decreased the amounts of phospho-AktSer473 in both fractions and was able to effectively block the increase in phospho-AktSer473 in response to EGF stimulation.

Doxorubicin increases AktSer473 phosphorylation and enhances antitumor responses in combination with GDC-0941 in a breast cancer xenograft model

Using the human MCF7-neo/HER2 breast cancer xenograft model, we further investigated combinations of GDC-0941 and doxorubicin. After implantation of cells into mice, tumors were allowed to reach a mean tumor volume of 250 to 350 mm3 before dosing was initiated. For these experiments, doxorubicin was administered intraperitoneally twice weekly for 3 weeks at doses of 0.5, 1.0, 2.0, and 4.0 mg/kg. Maximum efficacy was achieved at 4 mg/kg. Phospho-AktSer473 and total Akt were measured in xenograft tumors sampled at 1 and 4 hours after the last administered dose of doxorubicin. The phospho-AktSer473/total Akt levels increased as early as 1 hour and sustained at 4 hours (Fig. 5A). Total Akt levels did not change with doxorubicin treatment in vivo. Analogous to our in vitro studies, GDC-0941 decreased phospho-AktSer473 levels in the presence or absence of doxorubicin in xenograft tumors (Fig. 5B). GDC-0941 was more effective in reducing phospho-AktSer473 in the absence of doxorubicin, suggesting that GDC-0941 reduced basal phospho-AktSer473 but was unable to inhibit DNA-PK phosphorylation of AktSer473. Levels of phospho-AktSer473 and total Akt were also measured in murine platelets 1 hour after the last administered doxorubicin dose (4 mg/kg) (Fig. 5C). We did not detect a significant difference between the vehicle- and the doxorubicin-treated platelet samples.

Fig. 5

Efficacy of GDC-0941 and doxorubicin combinations in the MCF7-neo/HER2 xenograft model. (A) Phospho-AktSer473 increases are detected in vivo. Tumor-bearing mice were dosed intraperitoneally with vehicle (0.9% saline) or doxorubicin at the concentrations indicated. Phospho-AktSer473 and total Akt protein concentrations were determined by ELISA in xenograft tumors at 1 and 4 hours after dosing. (B) GDC-0941 decreases phospho-AktSer473 in vivo. Tumor-bearing mice were dosed orally with GDC-0941 (100 mg/kg) or intraperitoneally with doxorubicin (4 mg/kg). Phospho-AktSer473 and total Akt protein concentrations were assessed by ELISA in xenograft tumors 1 hour after dosing. (C) Detection of phospho-AktSer473 in platelets. Platelets were purified from murine blood at 1 hour after doxorubicin dosing, and phospho-AktSer473 and total Akt protein concentrations were determined by ELISA. (D) Significant in vivo efficacy with the GDC-0941 and doxorubicin combination. Tumor-bearing mice were dosed daily with GDC-0941 orally (PO) and vehicle (MCT) and biweekly intraperitoneally (IP) with doxorubicin for 21 days. (E) Mean body weights from efficacy study.

The in vivo increase in phospho-Akt upon doxorubicin treatment indicated that MCF7-neo/HER2 xenograft tumors rely on the pathway for a survival response to DNA damage and that greater antitumor activity may be achieved by combining doxorubicin and GDC-0941. To test this hypothesis, after tumors were allowed to reach a size of 250 to 350 mm3, we gave doxorubicin (4 mg/kg) twice weekly intraperitoneally and GDC-0941 orally at 100 mg/kg daily for 21 days. On the days in which the animals received both drugs, doxorubicin was administered 1 hour before GDC-0941. Plasma concentrations of the compounds were assessed in the animals, and no changes in drug concentrations were observed due to coadministration of the compounds. The combination of doxorubicin with GDC-0941 caused significant tumor growth inhibition that was greater than what was observed with either agent alone (Fig. 5D). The combination of GDC-0941 and doxorubicin inhibited tumor growth significantly compared to single-agent GDC-0941 or doxorubicin. The doses tested for GDC-0941, doxorubicin, and the combination of both agents were well tolerated, and body weights were not significantly different among the treated groups and vehicle controls (Fig. 5E).


Historically, it has been challenging to predict the successes and failures of drug combinations in cancer patients and to understand the mechanisms of response and resistance to chemotherapeutic agents. Here, we used a panel of breast and ovarian cancer cell lines to investigate the in vitro efficacy of a specific PI3K inhibitor in combination with the DNA-damaging agent doxorubicin. We discovered that a doxorubicin resistance–survival pathway relying on Akt phosphorylation becomes activated in a subset of cancer lines after DNA damage occurs. Notably, we also discovered a correlation between cancer lines that benefit from the drug combination and those in which an increase in Akt phosphorylation is detected after doxorubicin treatment.

In most cell lines examined, the doxorubicin and PI3K inhibitor combination was either additive or synergistic. This level of synergy of drug combinations has been observed in several studies with the PI3K inhibitors LY294002 or wortmannin (3336). Previous results with those PI3K inhibitors, however, may not be solely due to PI3K inhibition because both inhibitors exhibit substantial off-target activity (37, 38). In addition, activation of the PI3K pathway has been shown to be a critical player in the resistance of cancers to therapeutics (13, 16, 39). Therefore, it was not surprising that blockade of the PI3K pathway in the presence of doxorubicin resulted in increased efficacy. It was unexpected, however, that synergy would be observed in many of the lines regardless of GDC-0941 potency, PI3K pathway activation status, or p53 mutational status.

We discovered that, for lines demonstrating synergistic combination, there was a coordinate increase in AktSer473 phosphorylation in response to doxorubicin treatment. The cancer lines that showed this response were also more resistant to single-agent doxorubicin. It is unlikely that cell lines with this response are inherently more dependent on the PI3K pathway for growth and survival because those lines exhibited a range of GDC-0941 EC50s and a wide range of basal phospho-Akt levels for both responders and nonresponders (table S1 and fig. S4). Despite the phospho-AktSer473 increase observed with the cytotoxic agent alone, cotreatment with GDC-0941 reduced phospho-AktSer473 levels, thus decreasing the common Akt phosphorylation response that many tumor cells engage to survive.

Phosphorylation of Akt in response to DNA-damaging agents has been described previously in breast cancer lines in response to high concentrations of chemotherapeutics, trastuzumab, or tamoxifen (13). In those studies, high concentrations of LY294002 alone or in combination with chemotherapeutics resulted in increased apoptosis and induction of Akt phosphorylation. PI3K inhibitors have been reported to induce a cell cycle block (40), and in the cell line panel, we observed reduced cyclin D1 within 24 hours of GDC-0941 treatment (Fig. 1B). We also detected at 24 hours a strong increase in cleaved PARP, a marker of apoptosis. Our combination studies measure cell viability after 96 hours; therefore, the effects are likely the result of both G1 cell cycle arrest and apoptosis.

Both GDC-0941 and the DNA-PK inhibitor NU-7026 can decrease nuclear phospho-Akt, but there are likely different mechanisms by which these inhibitors hamper Akt phosphorylation in response to DNA damage. NU-7026 inhibits DNA-PK activity directly, which blocks subsequent AktSer473 phosphorylation after doxorubicin treatment. GDC-0941 does not directly bind to or inhibit DNA-PK, but instead works by reducing basal phospho-AktThr308 and phospho-AktSer473 levels (41). The two activation sites of Akt influence each other; thus, phosphorylation at one site can alter the phosphorylation status of the other (42, 43). Thr308 phosphorylation confers greater Akt enzyme activity than Ser473; thus, PI3K-PDK1 (phosphoinositide-dependent kinase 1) inhibition stops Thr308 phosphorylation and is more important for reducing the function of Akt (44). With a PI3K inhibitor diminishing phospho-Akt Thr308, DNA-PK can still phosphorylate Akt at the Ser473 site upon DNA damage but Akt activity will be below a threshold necessary for Akt function.

Our experiments with the MCF7-neo/HER2 breast cancer xenograft model suggest that combinations of drugs made up of PI3K inhibitors and doxorubicin can enhance antitumor responses in vivo. The preclinical model results demonstrated that it is feasible to sample in vivo tumors and observe transient Akt phosphorylation after DNA damage with doxorubicin from tumors in vivo. Akt phosphorylation in murine platelets after doxorubicin treatment did not change in our studies, but other surrogate tissues can be explored to ensure that toxicity is not also increased as a result of chemopotentiation by PI3K inhibitors. In the clinic, it would be beneficial to determine whether increased phospho-AktSer473 in tumors is confirmed as a metric in the assessment of resistance to DNA-damaging agents. Together, these results encourage further investigations for the clinical utility of PI3K inhibitors in combination with DNA-damaging agents.

Materials and Methods

Cell culture

Human breast and ovarian cell lines were obtained from the American Type Culture Collection (ATCC) or German Resource Centre for Biological Material. Cell lines were cultured in Dulbecco’s modified Eagle’s medium or RPMI 1640 supplemented with 10% fetal bovine serum, penicillin (100 U/ml), and streptomycin (100 μg/ml) at 37°C under 5% CO2. MCF7-neo/HER2 is an in vivo–selected tumor cell line developed at Genentech and derived from the parental MCF7 human breast cancer cell line (ATCC).


GDC-0941 was obtained from Genentech. The DNA-PK inhibitor NU-7026 and doxorubicin were from Sigma. For in vivo studies, doxorubicin was purchased as the drug Doxil from St. Mary’s Pharmacy. Antibodies to phospho-AktSer473, Akt, phospho-SGK3Thr320, cyclin D1, cyclin B, phospho-Wee1Ser642, phospho-cdc2Tyr15, phospho-Chk1Ser280, phospho-cdc25CSer216, cdc25C, phospho–histone H3Ser10, α-tubulin, and lamin A/C were obtained from Cell Signaling. Antibodies to PTEN and β-actin were obtained from BD Pharmingen and Sigma, respectively.

Mutation detection

Mutational screening was done by direct sequencing (Polymorphic DNA Technologies) of exons and adjacent intronic junctions. Primers for polymerase chain reaction (PCR) and sequencing were designed with the Primer3 program ( and purchased from Integrated DNA Technologies. Nested PCR products were treated with ExoSAP (USB) per the manufacturer’s instructions. PCR products were sequenced with BigDye Terminator Mix (Applied Biosystems) on an ABI3730xl (Applied Biosystems). Trace files were analyzed with Agent (Paracel), Sequencher (GeneCodes), and Mutation Surveyor 3.0 (SoftGenetics).

Cell viability and combination assays

Cells were seeded in 384-well plates for 16 hours. Diluted compounds were added to quadruplicate wells in 384-well cell plates. After 4 days of incubation, relative numbers of viable cells were measured by luminescence with CellTiter-Glo (Promega) and read on a Wallac Multilabel Reader (Perkin-Elmer). EC50 values were calculated with Prism 4.0 (GraphPad). Drugs in combination assays were dosed starting at four times their EC50 concentrations. The CI was assessed by the method of Chou (19) with CalcuSyn software (Biosoft).

Comet assay

Cells were treated with an EC50 concentration of GDC-0941, doxorubicin, or both for 24 hours. The comet assay kit was obtained from Trevigen and performed according to the manufacturer’s instructions.

Protein assays

Cells were treated with an EC50 concentration of GDC-0941, doxorubicin, or both. After treatment, cells were lysed in cell extraction buffer from Biosource supplemented with protease inhibitors (Roche), 1 mM phenylmethylsulfonyl fluoride (PMSF), and Phosphatase Inhibitor Cocktails 1 and 2 (Sigma). Protein concentration was determined with the BCA Protein Assay kit (Pierce).

Nuclear and cytoplasmic extracts were made with the Qiagen Cell Fractionation kit supplemented with protease inhibitors, 1 mM PMSF, and Phosphatase Inhibitor Cocktails 1 and 2.

Levels of phospho-AktThr308, phospho-AktSer473, and total Akt were assessed with kits from Meso Scale Discovery. For Western blots, equal amounts of protein were separated by electrophoresis through NuPAGE bis-tris 4 to 12% gradient gels (Invitrogen), and proteins were transferred onto nitrocellulose pore membranes with the iBlot system and protocol from Invitrogen.

In vivo xenograft studies

MCF7-neo/HER2 cells resuspended in a 1:1 mixture of Hanks’ balanced salt solution and Matrigel Basement Membrane Matrix (no. 356237, BD Biosciences) were subcutaneously implanted into the mammary fat pad of each mouse. Before cell inoculation, 17β-estradiol (0.36 mg per pellet, 60-day release, no. SE-121), obtained from Innovative Research of America, was implanted into the dorsal shoulder blade area of nude mice. After implantation of cells into mice, tumors were monitored until they reached a mean tumor volume of 250 to 350 mm3 and distributed into groups of 10 animals each before dosing was initiated. Doxorubicin was dissolved in 0.9% saline. GDC-0941 was obtained as a solution and dissolved in 0.5% methylcellulose with 0.2% Tween 80 (MCT) vehicle. Female nude (nu/nu) mice that were 6 to 8 weeks old and weighed 20 to 30 g were obtained from Charles River Laboratories. For combination efficacy experiments, tumor-bearing mice were dosed daily for 21 days with 100 μl of vehicle (MCT) and GDC-0941 at the concentrations indicated. Doxorubicin was dosed biweekly at the concentrations indicated in the figure legends.

Tumor volume was measured in two dimensions (length and width) with Ultra Cal-IV calipers (model 54-10-111; Fred V. Fowler Company) and analyzed with Excel version 11.2 (Microsoft). Tumor volume (mm3) was calculated as follows: [longer measurement × (shorter measurement)2] × 0.5. Animal body weights were measured with an Adventurer Pro AV812 scale (Ohaus). Percent weight change was calculated as follows: [1 − (new weight/initial weight)] × 100. Tumor sizes were recorded twice weekly over the course of the study (21 days). Mouse body weights were also recorded twice weekly, and the mice were observed daily. Mice with tumor volumes of ≥2000 mm3 or with losses in body weight of ≥20% from their initial body weight were promptly euthanized per Institutional Animal Care and Use Committee guidelines. Mean tumor volume and SEM values (n = 10) were calculated with JMP statistical software version 5.1.2 at the end of treatment. Percent tumor inhibition was calculated as follows: 100 × (mean volume of tumors in vehicle-treated animals − mean volume of tumors in test article–treated animals)/mean volume of tumors in vehicle-treated animals.

For pharmacodynamic analysis, MCF7-neo/HER2 tumors were excised from animals and immediately snap-frozen in liquid nitrogen. Frozen tumors were weighed and lysed with a polypropylene pestle (Scienceware) in cell extract buffer (Biosource) supplemented with protease and phosphatase inhibitors as described previously. Platelets were purified according to standard procedures (45) and lysed immediately in cell extract buffer.


Significance (P values) was determined by Student’s t test.

Supplementary Material

Materials and Methods

Table S1. Summary of in vitro combination studies in breast and ovarian cancer cell lines.

Fig. S1. Flow cytometric analysis of cell cycle effects in MCF7-neo/HER2 and HDQ-P1 cell lines.

Fig. S2. Changes in Akt substrates with doxorubicin treatment.

Fig. S3. Nuclear and cytoplasmic changes in Akt substrates with doxorubicin treatment.

Fig. S4. Correlation of combination index with breast subtypes, PI3K pathway alterations, basal phospho-AktSer473 levels, and p53 status.


  • Citation: J. J. Wallin, J. Guan, W. W. Prior, K. A. Edgar, R. Kassees, D. Sampath, M. Belvin, L. S. Friedman, Nuclear phospho-Akt increase predicts synergy of PI3K inhibition and doxorubicin in breast and ovarian cancer. Sci. Transl. Med. 2, 48ra66 (2010).

References and Notes

  1. Acknowledgments: We thank Genentech chemists for PI3K compounds, L. Salphati for pharmacokinetic analysis, and S. Seshagiri for cell line sequence analysis. Author contributions: J.J.W., J.G., W.W.P., K.A.E., R.K., and D.S. performed and designed all experiments. J.J.W., M.B., D.S., and L.S.F. wrote the paper. Competing interests: All authors are employees of Genentech Inc.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article