Purpose: The cancer stem cell hypothesis predicts that standard prostate cancer monotherapy eliminates bulk tumor cells but not a tumor-initiating cell population, eventually leading to relapse. Many studies have sought to determine the underlying differences between bulk tumor and cancer stem cells.
Experimental Design: Our previous data suggest that the PTEN/PI3K/AKT pathway is critical for the in vitro maintenance of CD133+/CD44+ prostate cancer progenitors and, consequently, that targeting PI3K signaling may be beneficial in treatment of prostate cancer.
Results: Here, we show that inhibition of PI3K activity by the dual PI3K/mTOR inhibitor NVP-BEZ235 leads to a decrease in the population of CD133+/CD44+ prostate cancer progenitor cells in vivo. Moreover, the combination of the PI3K/mTOR modulator NVP-BEZ235, which eliminates prostate cancer progenitor populations, and the chemotherapeutic drug Taxotere, which targets the bulk tumor, is significantly more effective in eradicating tumors in a prostate cancer xenograft model than monotherapy.
Conclusion: This combination treatment ultimately leads to the expansion of cancer progenitors with a PTEN E91D mutation, suggesting that the analysis of PTEN mutations could predict therapeutic response to the dual therapy. Clin Cancer Res; 16(23); 5692–702. ©2010 AACR.
Tumor relapse remains a challenging issue in the treatment of cancer. The cancer stem cell hypothesis—which postulates the existence of a drug-resistant, tumor-initiating cell population—offers an intriguing cellular model system to pursue strategies that may more effectively treat cancers. Our data suggest that the use of a combination of the PI3K inhibitor NVP-BEZ235 and the conventional chemotherapeutic drug Taxotere directed against cancer progenitors and the bulk tumor population, respectively, leads to significant tumor regression in a prostate cancer xenograft model relative to standard chemotherapy. Our identification of a resistant mutation in relapsed xenograft tumor samples also suggests a way to test clinical samples to more effectively design personalized treatment regimens.
Prostate cancer is the most frequently diagnosed cancer in men. Although conventional therapies, such as surgery, hormonal therapy, radiotherapy, and chemotherapy, are effective in the initial phase of treatment, many prostate cancers eventually progress to invasive and drug-resistant metastatic cancers upon relapse (1). The recurrence and resistance to therapy have been attributed to the existence of stem cells referred to as cancer stem cell–like cells, or tumor-initiating cells (2–7). Tumor stem cell–like cells represent a population of drug-resistant cells that can survive conventional treatment and cause a tumor relapse. This new concept of tumorigenesis suggests that therapies that also target tumor progenitors may lead to more effective cancer treatments.
We showed previously that PI3K/AKT/FOXO3a signaling plays a critical role in the maintenance and viability of CD133+/CD44+ prostate cancer progenitors. Depletion of FOXO3a and PTEN increased the tumor-initiating cell population, and inhibition of PI3K activity by small molecules led to growth inhibition of prostate cancer progenitors while sparing differentiated cells (8). This finding is consistent with previous studies showing an important role for the PTEN/PI3K/AKT pathway in the regulation of malignant progenitor cell populations in acute myeloid leukemia and breast and brain tumors (9–11). These results suggest that inhibition of the PI3K/AKT and mTOR pathways in prostate cancer progenitors together with conventional chemotherapy may provide improved therapeutic outcomes (12, 13). Furthermore, the current development of anti-PI3K inhibitors should facilitate the translation of this combination strategy into clinical practice (14–16). One such drug is NVP-BEZ235, a dual pan class I PI3K and mTOR inhibitor that shows high antitumor activity in a variety of human cancer models (16).
Herein, we show that treatment of prostate tumor xenografts in mice with a combination of a standard chemotherapeutic drug targeting bulk tumor cells and NVP-BEZ235 targeting CD133+/CD44+ tumor progenitors leads to near-complete tumor regression. In contrast, the use of cytotoxic drugs such as Taxotere or 5-fluorouracil (5-FU) alone results in a decrease in bulk tumor cells but leads to an overall increase in the relative size of the tumor-initiating cell population—the source of tumor relapse and resistance. Our data suggest that the use of combination therapy strategy directed against the tumor-initiating cells together with the bulk tumor is more effective in eradicating xenograft tumors than monotherapy and may provide an advantage in cancer treatment. Moreover, we found that acquired mutations that reactivate PI3K signaling may serve as a good predictor for response to PI3K targeted therapy in the clinic.
Materials and Methods
Cells, reagents, and animals
DU145, PC3, 22RV1, and LNCaP (prostate cancer cells) were obtained from ATCC and were cultured in Dulbecco's Minimal Essential Medium (DU145 cells), Ham's/F12 Medium (PC3 cells), or RPMI-1640 Medium (22RV1 and LNCaP cells), supplemented with 10% fetal calf serum, 300 μg/mL of glutamine, 100 IU/mL of penicillin, and 100 μg/mL of streptomycin. NOD.CB17-Prkdc (SCID) mice were obtained from Jackson Laboratories and maintained under standard conditions according to institutional guidelines. The antibodies used were anti-AKT (pan), phospho-AKT(Ser473), phospho-AKT(Thr308), PTEN, FOXO3a, phospho-FOXO3a(Ser253), Cell Signaling Technology; β-actin (mAb), Sigma; rabbit IgG, HRP-linked whole Ab, GE Healthcare; and mouse IgG, HRP-linked whole Ab, GE Healthcare. PI3K inhibitor NVP-BEZ235 was obtained from Novartis, Taxotere (docetaxel) was purchased from LC Laboratories, and 5-FU and Oxaliplatin were purchased from Sigma.
Flow cytometric analysis of CD44 and CD133 expression
For flow cytometry, cells were dissociated with Accutase (Innovative Cell Tech Inc.) and washed twice in staining solution containing Ca2+- and Mg2+-free PBS with 1 mmol/L of EDTA, 25 mmol/L of HEPES (pH 7.0; Invitrogen), 0.5% FBS. Cells were stained with conjugated anti-CD44 (mouse anti-human CD44-APC-H7; BD Pharmingen) and anti-CD133 (mouse anti-human CD133/2(293C3)-PE; Miltenyi Biotec) antibody (50 minutes at 4°C). Samples were analyzed on a BD LSR II flow cytometer (Beckton Dickinson Immunocytometry Systems). Cell debris and clumps were electronically gated. A minimum of 500,000 viable cell events were collected per sample. For sorting, 2 × 107 cells were processed for CD44 and CD133 staining along with appropriate negative controls, single color-positive controls, and isotype negative control (mouse IgG2b; Miltenyi Biotec). The CD133+/CD44+ and CD133−/CD44− populations were sorted on a BD FACS Diva cell sorter.
In vivo tumorigenicity experiments and mouse treatment
Xenograft tumors were established using DU145 cells within 4 passages of receipt from ATCC. For subcutaneous tumor development, 5 × 105 DU145 cells were injected subcutaneously into 5- to 8-week-old NOD.CB17-Prkdc (SCID) mice in 100 μL of BD matrigel (BD Biosciences). When the tumors reached a size of 300 mm3, mice were treated with Taxotere or the PI3K/mTOR modulator NVP-BEZ235, alone or in combination. The mice treated with vehicle were used as the control group. For each experiment, a stock solution of Taxotere was prepared by dissolving docetaxel in ethanol. Final drug dilution was obtained by adding polysorbate 80 and 5% glucose in water (5:5:90; v/v/v). A stock solution of NVP-BEZ235 was prepared by dissolving NVP-BEZ235 in 1-methyl-2-pyrrolidone. The final drug solution was obtained by adding polyethylene glycol 300 (1:9; v/v). The animals received Taxotere (10 mg/kg, intravenously) once per week and NVP-BEZ235 (12.5 mg/kg, perorally) once per day for 4 weeks. Ten mice were used in each experimental group. The tumors were measured every 4 to 5 days. The mice were observed for 3 to 5 months for the appearance and development of tumors.
The PI3K/AKT/FOXO3a pathway is highly activated in prostate tumor–initiating cells in vivo
Recent evidence showed that the PTEN/PI3K/AKT pathway is critical for stem cell maintenance but that aberrant activation of PI3K signaling contributes to transformation of quiescent normal stem cells to cancer stem cells in the hematopoietic system, brain, intestine, mammary gland, and prostate (9–12, 17–20). Cell populations expressing the surface markers CD133 and CD44 have been identified as putative stem cell populations in the prostate gland. We and others have confirmed that CD133+/CD44+ cells from established prostate cancer cell lines are self-renewing, differentiate into heterogeneous tumors, and have a strong tumorigenic potential in vivo (6, 8, 21–24; Supplementary Fig. S1). Moreover, we recently showed that inhibition of PI3K signaling and activation of FOXO3a-dependent transcription results in a decrease in prostate progenitor survival in vitro by targeting prostate cancer progenitors in androgen refractory prostate cancers (8). In addition, we found that androgen-sensitive prostate cancer cells grown under sphere-forming conditions also showed an activation of PI3K signaling and repression of FOXO3a-dependent gene expression as compared with the cells grown under monolayer conditions (Supplementary Fig. S2).
We have extended these earlier studies to assess the role of PI3K activation in vivo by characterizing the degree of colocalization of CD133 (a marker for prostate cancer progenitors) and phospho-FOXO3A (an indicator of PI3K pathway activation) in tumor samples. Indeed, in 3 of 3 primary human prostate tumor samples, 95.5% ± 4.8% of CD133+ cells have a high level of FOXO3a phosphorylation compared with 59.5% ± 8.6% of phospho-FOXO3a+ cells in CD133− cells (Fig. 1A and C–E). Similarly, in mouse xenografts of the DU145 prostate cancer cell line, ∼93% ± 0.9% of CD133+ cells highly express phosphorylated FOXO3a compared with CD133− cells (Fig. 1B). This finding is consistent with our previous data showing that prostate cancer cells in which FOXO3a expression is knocked down possess high tumorigenic potential in vivo. These results suggest that FOXO3a phosphorylation may be important for the maintenance of cancer progenitors in vivo and targeting PI3K signaling could provide a means to specifically eliminate prostate cancer stem cell-like cells.
Combinatorial treatment of DU145 and PC3 human prostate carcinoma xenograft tumors using Taxotere and the dual PI3K/mTOR inhibitor NVP-BEZ235
To demonstrate that the combination of PI3K pathway inhibition and conventional chemotherapy targets both tumor-initiating and bulk tumor cancer cell populations, we tested the effect of the PI3K/mTOR modulator NVP-BEZ235 as well as conventional chemotherapeutic drugs on the survival of CD133+/CD44+ prostate cancer progenitor cells and CD133−/CD44− differentiated cell populations in the prostate cancer cell lines DU145 and PC3. NVP-BEZ235 is a potent PI3K/mTOR inhibitor that significantly reduces AKT phosphorylation and increases FOXO3a nuclear localization and transcriptional activity in vitro (Supplementary Fig. S3A–C). PC3 and DU145 cells were treated with 20 nmol/L of NVP-BEZ235 (IC50 = 20 nmol/L), 5 nmol/L of Taxotere (IC50 = 10 nmol/L), 5 to 20 μg/mL of fluorouracil (IC50 = 20 μg/mL), or 5 μmol/L of Oxaliplatin (IC50 = 10 μmol/L) for 3 days in serum-free epithelial growth medium (progenitor medium conditions). We observed a significant decrease in the CD133+/CD44+ population for both PC3 and DU145 cells with NVP-BEZ235 treatment (a 2.8- and 2.2-fold decrease, respectively; Fig. 2A), whereas this population was not changed or even increased in response to conventional therapy. In contrast, the CD133−/CD44− population was sensitive to conventional chemotherapy (a 1.7- to 7.9-fold decrease) but was relatively unresponsive to NVP-BEZ235 treatment (≤1.3-fold decrease). In addition, PC3 and DU145 cells that were treated with the combination of NVP-BEZ235 and any of the chemotherapeutic drugs had a 2-fold or greater decrease in CD133+/CD44+ progenitor cell populations; a 2-fold or greater decrease in the CD133−/CD44− cell population with the combination of NVP-BEZ235 and 5-FU or Oxaliplatin; and a 1.5-2.0-fold decrease in the CD133−/CD44− population with the combination of NVP-BEZ235 and Taxotere (Fig. 2A). These in vitro findings demonstrate that the combination of NVP-BEZ235, which preferentially reduces prostate cancer progenitor populations, together with chemotherapeutic drugs that target the bulk tumor can be more effective than monotherapy.
Next, we translated these findings to an in vivo model by treating DU145 hormone-refractory prostate carcinoma xenograft tumors with NVP-BEZ235 and the cytotoxic antimicrotubule agent Taxotere. Taxotere is the only chemotherapy drug currently approved for the treatment of advanced prostate cancer, and thus its application in our model is more clinically relevant than the use of either 5-FU or Oxaliplatin. DU145 cells were injected subcutaneously into NOD/SCID mice, which have been shown to be a relevant xenograft model to study prostate cancer (25, 26). When the tumors had grown to a size of 300 mm3, the mice were treated with Taxotere (10 mg/kg, injected once-weekly intravenously), NVP-BEZ235 (12.5 mg/kg given every day perorally), or the combination for 5 weeks (Fig. 2B). Combination therapy led to the inhibition of tumor growth in the first 3 weeks and to reduction of tumor size beginning in the fourth week of treatment. After the treatment period, the xenograft tumors treated with either drug alone showed an approximate 2-fold decrease in tumor growth rate compared with the control group. The combination treatment, on the other hand, led to a 7-fold decrease in tumor growth relative to the control (Fig. 2B).
Flow cytometric analysis revealed a more than 2-fold attenuation of the CD133+/CD44+ population in the xenograft tumors treated with NVP-BEZ235, alone or in combination with Taxotere (Fig. 2C). In contrast, the CD133+/CD44+ population was not attenuated, or even increased, in the Taxotere monotherapy cohort. Consistent with these results, we also observed a decrease in the fraction of CD133+/CD44+ and CD133+/phospho-FOXO3a populations in NVP-BEZ235–treated tumors by immunostaining (Figs. 3 and 4A). Western blot analysis showed a significant decrease in FOXO3a phosphorylation in the tumors treated with NVP-BEZ235 alone or in combination with Taxotere (Fig. 4B). In addition, the CD133+/CD44+ cell population from mice treated with the combination of 5 nmol/L of NVP-BEZ235 and 1 nmol/L of Taxotere for 3 weeks continue to maintain stem cell–like biological properties, possess multipotency, and have high spherogenic and tumorigenic potential (Supplementary Fig. S4). Notably, treatment with a lower dose of NVP-BEZ235 (6 mg/kg given every day perorally) only moderately reduced tumorigenicity but caused tumor shrinkage in mouse xenografts in combination with conventional chemotherapy, suggesting that the tumor-initiating population could be very sensitive to PI3K inhibition (Supplementary Fig. S5). Collectively, these studies identify the PI3K signaling pathway as an important component in prostate tumor progenitor maintenance and the use of combination therapy directed against prostate cancer progenitor cells and bulk tumor cells is more effective in xenograft tumors than conventional chemotherapy alone.
Inactivation of the tumor suppressor PTEN in prostate cancer progenitor cells could be an underlying mechanism in tumor regrowth
Relapse often occurs after conventional treatment of prostate cancer in the clinic even after multiple cycles of chemotherapy (27). Likewise, tumor growth is temporarily delayed in rodent models of prostate cancer after chemotherapy, but reported protocols are not sufficiently efficacious to eradicate tumors (28, 29). As expected, tumors treated with Taxotere alone quickly resumed growth after Taxotere withdrawal (Fig. 2B). While Taxotere treatment reduced tumor size by 45% relatively to vehicle-treated tumors during the first 45 days of treatment, tumor size increased by 112% over the next 40 days after Taxotere withdrawal. Unexpectedly, we also observed that the tumors treated with both Taxotere and NVP-BEZ235 showed slow regrowth during the second week after the treatment was terminated (Fig. 2B). Tumor volume increased by 104% after Taxotere/BEZ235 treatment was stopped, although the endpoint tumors were still a modest 263 mm3 on average compared with nearly 1,000 mm3 for Taxotere alone. Interestingly, we observed a selective 3-fold or greater increase in the CD133+/CD44+ population as compared with the tumors analyzed after 5 weeks of treatment (Supplementary Fig. S6A). One possibility is that mutations in the PI3K/AKT signaling components underlie cancer stem cell escape and tumor regrowth.
To determine whether possible genetic alterations reactivate the PI3K/AKT pathway in the cancer progenitor population, we analyzed mutations in the coding sequences of the candidate genes, PIK3CA, MTOR, and PTEN . Mutations in these genes are some of the most frequent genetic aberrations in human prostate cancers and may determine sensitivity to PI3K/mTOR inhibitors (30, 31). Exon sequencing was done for these genes in CD133+/CD44+ and CD133−/CD44− cell populations purified from tumors treated with combination therapy or vehicle. Tumors were treated with drug combination for 5 weeks and samples were collected 4 weeks after treatment was terminated to obtain sufficient cells for DNA sequencing. CD133+/CD44+ and CD133−/CD44− cells were isolated from 3 pooled xenograft tumors by FACS and exon DNA was amplified and sequenced. Two independent groups of animals were treated, followed by exon sequencing in duplicate. When compared with the CD133+/CD44+ population from vehicle-treated tumors or to bulk cells from tumors treated with the drug combination, exon sequencing of PIK3CA and MTOR genes in the Taxotere/NVP-BEZ235–treated tumors did not reveal any mutations. However, exon sequencing of PTEN in the CD133+/CD44+ population from the Taxotere/NVP-BEZ235–treated tumors revealed a missense E91D mutation within the active site pocket of the phosphatase domain (Fig. 5A). Notably, this mutation does not occur in bulk tumor cells treated with the drug combination (Supplementary Fig. S6B). Our results are in agreement with recently published data showing that stem cells are more susceptible to tumorigenic mutations than differentiated cell population (32, 33). The identified PTEN mutation could affect differentiation of prostate progenitors in vivo and may explain the selective increase in the CD133+/CD44+ population in xenograft tumors treated with drug combination after termination of treatment (Supplementary Fig. S6A).
PTEN is a lipid/protein phosphatase that negatively regulates the PI3K/AKT pathway by dephosphorylating phosphatidylinositol (3,4,5)-triphosphate and is one of the most commonly mutated tumor suppressor genes in human cancers (31). PTEN levels/activity is a key determinant in prostate cancer progression (34). Previously, we showed that PTEN loss has important roles in tumor initiation potential and maintenance of stem cell–like cell populations in prostate cancers (8). The PTEN D92A mutant has diminished phosphatase activity (30) and was used as a positive control. To study the functional consequence of the E91D mutation, we stably transfected PTEN+ DU145 and PTEN− PC3 and LNCaP prostate cancer cells with vectors encoding PTEN-WT, PTEN-E91D, or PTEN-D92A and assessed their clonogenic and spherogenic potential. PTEN-WT, PTEN-D92A, and PTEN-E91D were expressed at similar levels as determined by Western blotting (Fig. 5D). PTEN-E91D- and PTEN-D92A–transfected cells showed an increase in sphere and colony formation over PTEN-WT–transfected cells (up to 6- and 9-fold increase, respectively; Fig. 5B). Moreover, PTEN-E91D- and PTEN-D92A–expressing cells possess higher levels of AKT phosphorylation than PTEN-WT–transfected DU145, LNCaP, and PC3 cells (Fig. 5D). We further determined PTEN lipid phosphatase activity by measuring the conversion from PIP3 to PIP2 in cell lysates of 293T cells transfected with vectors encoding PTEN-WT, PTEN-E91D, or PTEN-D92A. The PTEN-E91D and PTEN-D92A mutant proteins showed a 2- and 3-fold decrease, respectively, in phosphatase activity compared with PTEN-WT (Fig. 5E).
To further explore potential effects of these mutations on cancer progenitors, we measured the CD133+/CD44+ population in DU145 cell lines stably transfected with vectors encoding PTEN-WT, PTEN-E91D, or PTEN-D92A by flow cytometric analysis. In contrast to PTEN-WT–transfected cells, cells expressing phosphatase defective PTEN-E91D and PTEN-D92A proteins showed an increase in CD133+/CD44+ cancer progenitors and possessed higher tumorigenic properties in vivo (Fig. 5F; Supplementary Fig. S7). In addition, PTEN-E91D mutation confers DU145 cells resistance to NVP-BEZ235 treatment alone and in combination with Taxotere under sphere-forming conditions (Fig. 5G). Collectively, these results suggest that the E91D mutation in PTEN could be one of the molecular mechanisms driving subsequent tumor regrowth in the animal experiments. This observation is consistent with our previously reported data that showed that PTEN loss leads to increased cancer progenitors with high clonogenic and tumorigenic potential.
It has been shown recently that continuous treatment with conventional chemotherapy often fails to eradicate tumors because of the accumulation of drug-resistant cancer progenitors, or cancer stem cells (35–37). Thus, targeting both differentiated and stem cell tumor populations is predicted to have greater efficacy in tumor reduction and prevention of relapse. We and others have identified the PI3K/AKT pathway as a critical mediator of maintenance of stemness in prostate cancer cells (8, 18). In this study, we show that the combination of PI3K inhibitor NVP-BEZ235 and the chemotherapeutic drug Taxotere, targeting cancer progenitors and the bulk tumor population, respectively, leads to significant tumor regression relative to standard chemotherapy in a xenograft model.
The use of drug combinations for the treatment of cancer originally focused on the use of cytotoxic molecules with independent modes of action in analogy with combinatorial antibiotic use, without regard to cellular heterogeneity in tumors (38). Recently, the cancer stem cell hypothesis has given impetus to the pairing of chemotherapeutic agents based on target cell populations rather than pathways within an individual cell. In one example, metformin, a drug that was found to selectively kill breast cancer stem cells, was dosed with doxorubicin and the combination virtually eliminated xenograft breast tumors, whereas doxorubicin alone decreased tumor volume 2-fold and metformin alone had little effect (39). Another study showed that a combination of a DR5 agonistic monoclonal antibody, which was found to target pancreatic cancer stem cells, with the chemotherapeutic drug gemcitabine is more efficacious in the elimination xenograft pancreatic cancers than monotherapy (40). Our results in prostate xenografts reflected a similar pattern—while treatment with either Taxotere or NVP-BEZ235 alone reduced tumor growth by 2-fold, the addition of NVP-BEZ235 to Taxotere led to a 7-fold decrease in tumor size compared with vehicle. Moreover, NVP-BEZ235 significantly reduced the tumor-initiating cell population present in the tumors whether used alone or in combination with Taxotere, in agreement with our observation that the state of PI3K pathway activation correlates with stem cell marker expression and tumorigenicity. Taken together, these studies suggest that combinatorial strategies based on bulk tumor reduction and cancer stem cell–specific pathway inhibition offer a promising treatment modality.
The PI3K/AKT pathway has been intensively studied as a major contributor to tumorigenesis and tumor progression in several types of cancers, including prostate (41). Several nodes, including mTOR, within the pathway can be targeted for cancer therapy. The mTOR inhibitor CCI-779 alone or in combination with Taxotere causes a tumor growth delay in xenograft prostate cancer models (13). Although our studies show that PI3K inhibition is sufficient to induce differentiation of prostate cancer progenitors, the relationship between mTOR inhibition and stemness in prostate cancer stem cells is not clear. However, the CCI-779/Taxotere combination study suggests that the inhibition of bulk tumor cell proliferation by NVP-BEZ235 may also contribute to the synergistic tumor growth reduction observed in this study. Nevertheless, our data suggest that prostate cancer progenitor cells are preferentially sensitive to PI3K inhibition compared with differentiated cell populations.
We observed more than 2-fold decrease in the CD133+/CD44+ population for both PC3 and DU145 cells with NVP-BEZ235 treatment in vitro, whereas the CD133−/CD44− population was relatively unresponsive to NVP-BEZ235 treatment showing 1.3-fold decrease or less. This result agrees with published observations that tumor-initiating populations can be targeted even at low doses of PI3K inhibitors that do not significantly reduce bulk tumor cells (8, 20). These results are also consistent with published data showing preferential sensitivity of breast and brain cancer progenitors to PI3K and AKT inhibition (20, 42).
Tumor relapse and the associated increase in cancer progenitor cells can be attributed to genetic aberrations that can occur in a subpopulation of cancer progenitors making them resistant to PI3K inhibition. Clonal expansion of tumor cells with advantageous mutations leading to aggressive progression of disease is a clinical phenomenon underlying drug resistance and tumor regrowth in cancer patients (43, 44). To understand the genetic mechanism behind tumor regrowth, we analyzed CD133+/CD44+ and CD133−/CD44− subpopulations within the relapsed tumors from combination therapy for mutations of key genes in the PI3K pathway. We found a novel missense mutation E91D in the phosphatase domain of the tumor suppressor PTEN. Mutation of E91 in PTEN has been reported in glioblastoma and metastatic prostate cancer (45, 46) and leads to a loss in phosphatase activity. Notably, the PTEN mutation was found selectively in the CD133+/CD44+ cell population in tumors treated with combination therapy, suggesting that PTEN inactivation in CD133+/CD44+ cells could be a mechanism for preventing cell differentiation and maintaining a cancer progenitor pool. Overexpression of PTEN E91D led to an increase in AKT activation and cancer progenitor proliferation compared with wild-type PTEN. Moreover, the PTEN-E91D mutation confers resistance in prostaspheres to NVP-BEZ235 treatment alone or in combination with Taxotere, suggesting that acquired mutations in tumor suppressor PTEN can serve as a mechanism of drug resistance in prostate cancer progenitor cells. A number of clinical studies suggest that mutations of PTEN occur in 60% to 80% of patients with advanced prostate cancer and are associated with the development of resistance to androgen deprivation and chemotherapy (47).
These observations suggest a model of drug resistance in prostate cancer progenitors after treatment with the dual PI3K/mTOR inhibitor NVP-BEZ235 in vivo (Fig. 6). NVP-BEZ235 preferentially targets prostate cancer progenitors while still affecting growth of the bulk tumor. Drug-resistant mutations develop in the progenitor population. After cessation of drug treatment, remaining bulk tumor and progenitor cells regrow; the increased level of PI3K/AKT activation in drug resistant cells favors self-renewal compared with asymmetric division, which, in turn, favors the spread of the drug-resistant mutation in the progenitor population. These mutated progenitors could form the nucleus of a drug-resistant tumor upon relapse. We assume that this mutation could be subsequently detectable throughout the bulk tumor upon more prolonged regrowth in the presence of combination treatment; at the time of tumor harvest for sequencing, the tumors were only 260 mm3, which was 17% of NVP-BEZ235–treated tumors, so drug-resistant progenitors had little time to repopulate the bulk tumor.
In summary, our results suggest that combination of therapies that target cancer stem cells with conventional chemotherapy can be a useful strategy for cancer treatment. Moreover, screening for patients with PTEN mutations could serve as a good predictor for response to PI3K-targeted therapy in the clinic and underscores the utility of profiling individual tumors.
Disclosure of Potential Conflicts of Interest
J. Elliott, R.J. Salamone, J.R. Walker, J. Watson, M. Sauveur-Michel and C.Y. Cho are employees of Novartis.
We thank Vasyl Lukiyanchuk for his support with sequence analysis, Dr. Serge Batalov for his help with exon sequencing, Christopher Trussel for his technical assistance with flow cytometry, and Dr. Yong Jia and Christopher Quinn for their help with PTEN phosphatase assays. We thank Drs. Costas Lyssiotis, Heiko Wurdak, and Brooke Grandinetti for critical reading of the manuscript.
Grant Support: This study was supported by Susan G. Komen for the Cure Grant PDF0707903 A. Dubrovska the Novartis Research Foundation, and the Skaggs Institute for Chemical Biology.
Current address of V.A. Reddy: Oncology – Sanofi-Aventis US, 270 Albany St, Cambridge, MA 02139.
Current address of C. Garcia-Echeverria: Oncology – Sanofi-Aventis Fr, 13 quai Jules Guesde, Vitry-Sur-Seine, 94403 Cedex, France.
Current address of S. Kim: Amgen Inc., 1120 Veterans Blvd., South San Francisco, CA 94080.
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
- Received June 17, 2010.
- Revision received September 24, 2010.
- Accepted October 5, 2010.