The Multi-Targeted Kinase Inhibitor Sunitinib Induces Apoptosis in Colon Cancer Cells via PUMA

Constitutive activation of pro-survival kinases has become a promising target of small molecules with an increasing interest in developing multi-targeted agents. The mechanisms underlying the responsiveness to most agents targeting cancer specific survival pathways are still poorly understood but critical for their clinical application. In this study, we found that sunitinib, a small molecule inhibitor of multiple tyrosine kinases including VEGFRs and PDGFRs induces apoptosis and inhibits cell growth in colon cancer cells in cell culture and xenograft models via the BH3-only protein PUMA. Sunitinib treatment induced PUMA transcription via the AKT/FoxO3a axis. PUMA, BH3 mimetics, or 5-Flurourical sensitized colon cancer cells to sunitinib-induced apoptosis. Furthermore, PUMA was induced by sunitinib treatment in xenograft tumors, and deficiency in PUMA significantly suppressed the anti-tumor effects of sunitinib. Our study suggests that PUMA-mediated apoptosis is important for the therapeutic responses to sunitinib, and activation of the mitochondrial pathway by BH3 mimetics or PUMA manipulation may be useful for improving the antitumor activity of sunitinib. Modulation of PUMA and selective Bcl-2 family members might be potential biomarkers for predicting sunitinib responses.


Introduction
Colorectal cancer (CRC) is the third leading cause of cancerrelated death in the US and the incidence is on the rise in developing countries [1]. Even with the combination of improved chemotherapy and radiation in past decades, the 5 year survival of CRC patients with advanced disease remains unacceptably low. Aberrant activation of various kinase pathways is common in most solid tumors, which can lead to increased proliferation, survival, angiogenesis or invasion [2,3]. In recent years, considerable hope has been placed on agents developed to target oncogenic kinases, whose use in combination with chemotherapy or radiation might improve the survival and outcome of CRC patients [4]. The targeted approach is expected to ultimately deliver safer and more effective cancer therapeutics [5]. One major challenge in the clinical use of these agents is the prevalence of intrinsic and acquired resistance, whose underlying mechanisms remain largely unknown and a subject of intense investigation [4,5].
Sunitinib (also known as SU11248) was developed as a multitargeted receptor tyrosine kinase (RTK) inhibitor, and approved by the FDA in 2006 for the treatment of renal cell carcinoma (RCC) and imatinib resistant gastrointestinal stromal tumor (GIST) [6,7]. Ongoing clinical trials are being conducted to evaluate its efficacy in other tumor types including metastatic colon cancer [7,8] (http://clinicaltrials.gov/). Sunitinib inhibits a variety of receptor tyrosine kinases (RTKs) that are either mutated or activated in cancer. These include receptors for platelet-derived growth factor (PDGF-R a and b) and vascular endothelial growth factor receptors (VEGFR1, 2 and 3), as well as KIT (CD117), RET, CSF-1R, and flt3 [6,7]. Sunitinib has been recommended as a second-line therapy in GISTs that developed resistance to imatinib due to secondary mutations in c-KIT. Inhibition of angiogenesis, immune modulation and induction of apoptosis has been suggested to mediate the anti-tumor effects of sunitinib [7]. The mechanisms underlying the cell autonomous effect of sunitinib such as cell killing is not well-understood.
Mitochondria-mediated apoptosis plays an important role in the antitumor activities of a wide variety of conventional chemotherapeutic agents as well as targeted therapies [9,10]. The Bcl-2 family of proteins are the central regulators of mitochondriamediated apoptosis, which is engaged by the selective activation or induction of the proximal BH3-only members in response to distinct as well as overlapping signals [11,12]. The BH3-only protein PUMA plays an essential role in p53-dependent andindependent apoptosis in human cancer cells and mice [13], and activates the mitochondrial pathway via the Bcl-2 family member Bax/Bak following neutralizing all members of antiapoptotic Bcl-2 like molecules [14,15,16,17]. DNA damage induced by gammairradiation or commonly used chemotherapeutic agents such as 5fluorouracil (5-FU), adriamycin and etoposide, induce p53dependent induction of PUMA and apoptosis [15,18]. Nongenotoxic stresses such as growth factor deprivation, endoplasmic reticulum poisons and a number of kinase inhibitors induce PUMA through a number of other transcription factors including p73, NF-kB and FoxO3a [13,19,20,21,22,23].
In the current study, we demonstrated that sunitinib induces PUMA expression independent of p53 in colon cancer cells. The induction of PUMA was mediated by the transcription factor FoxO3a upon inhibition of AKT. PUMA deficiency led to resistance to sunitinib-induced apoptosis in cells as well as in xenografts. Our study provides a molecular mechanism of apoptosis induced by this non-selective kinase inhibitor in colon cancer cells, and has important implications for biomarker discovery and potential strategies to overcome resistance.
The release of cytochorme c and Smac was detected in the cytosol following subcelluar fractionation as described [20,26]. In brief, cells were treated in T75 flasks for indicated times and subject to differential centrifugation to obtain cytoplasmic and mitochondrial fractions. Concentrations of cytosolic fractions obtained were normalized using a protein assay dye reagent (Bio-Rad, Hercules, CA). The fractions were mixed with equal volumes of 2x Laemmli sample buffer and subjected to Western blotting analysis.

Luciferase Assays
Cells were transfected with PUMA reporters containing either WT or mutant FoxO3a binding sites [19], with the transfection control b-galactosidase reporter pCMVb (Promega), and treated with 12 mM sunitinib for 24 hours. Cell lysates were collected and luciferase activities were measured as previously described [23]. All reporter experiments were done in triplicate and repeated three times.

Real-time Reverse Transcription-PCR
Total RNA was isolated from cells using the Mini RNA Isolation II kit (Zymo Research, Irvine, CA) according to the manufacturer's protocol. Total RNA (1 mg) was used to generate cDNA using SuperScript II reverse transcriptase (Invitrogen). Real-time PCR was carried out for PUMA and GAPDH as described [22].

Apoptosis Assays
Adherent and floating cells were harvested, stained with Hoechst 33258 (Invitrogen), and analyzed for apoptosis by nuclear staining assay. A minimum of 300 cells were analyzed for each treatment [25,27]. For colony formation assays, equal number of cells were subjected to various treatments and plated into 12-well plates at different dilutions. Colonies were visualized by crystal violet staining 11 to 14 days after plating as previously described [25,28]. Each experiment was performed in triplicate and repeated at least twice.

Xenograft Tumors
All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Pittsburgh. HCT 116 WT and PUMA KO xenografts were established and measured as described [25]. In brief, 5-6 week old female athymic nude mice (Harlan, Indianapolis, IN) were inoculated with 5610 6 cells per site on both flanks. Tumors were allowed to establish for 7 days. The mice were oral gavaged for 10 consecutive days with 80 mg/kg/day sunitinib diluted in sodium citrate buffer (pH 4.7, vehicle), or vehicle [29]. The tumor volumes were measured in two dimensions using a vernier caliper. Mice were randomized into groups, such that the average tumor volume across the groups was the same prior to treatment. For all in vivo experiments, tumor volumes were measured every other day in 2 dimensions and volumes were determined in mm 3 using the formula l6b 2 60.52 (where l is the larger diameter and b is the smaller diameter of the tumor). Mice were injected i.p. 2 h before sacrifice with a single dose of bromodeoxyuridine (BrdU) at 150 mg/kg to label cells in S phase. BrdU was dissolved in PBS to a final concentration of 30 mg/mL. Histologic and immunofluorescence analysis for apoptosis and proliferation were performed on 5-mM frozen sections, as described [25]. Mice were euthanized 21 days after the treatment, or 24 hours after the third treatment (day 4). Tumors were dissected and snap frozen for western blotting or fixed in 10% formalin before paraffin embedding.

TUNEL and Immunostaining
Histological analysis was performed by hematoxylin and eosin staining. Terminal deoxyribonucleotidyl transferase -mediated dUTP nick end labeling (TUNEL) staining on frozen sections was done with recombinant terminal transferase (Roche) and uUTP-Alexa 594 (Invitrogen) according to the manufactures' instructions, and counterstained with 49,6-diamidino-2-phenylindole (DAPI) with minor modification made for paraffin samples [25,30,31]. Apoptotic cells were counted under a fluorescence microscope in randomly chosen fields, and the apoptosis index was calculated as a percentage of TUNEL-positive cells in at least 1,000 scored cells.
BrdU incorporation was visualized with anti-BrdU-Alexa 594 antibody (Invitrogen) and nuclei were visualized with DAPI. The proliferation index was calculated by counting cells under fluorescence microscope in several randomly chosen fields. The BrdU index was calculated as a percentage of BrdU-positive cells in at least 1,000 scored cells. Phospho-AKT, phospho-FoxO3a and active Caspase-3 immunohistochemistry was performed as described [19].

Statistical Analysis
Statistical analysis was carried out using GraphPad Prism IV software. All P-values were calculated by the student's t-test, and P,0.05 was considered significant. Means 6 one standard deviation (SD) were displayed in figures where applicable.

PUMA is Induced by Sunitinib in Colon Cancer Cells
The expression of PUMA is low in most unstressed cells, and is induced by various genotoxic and non-genotoxic stresses [23]. To determine a potential role of PUMA in sunitinib-induced response, we analyzed the levels of PUMA, p21 and p53 before and after treatment in five colon cancer cell lines. Three of these lines, HCT 116, RKO and SW 48 contain WT p53 while two lines HT 29 and SW 480 contain mutant p53. PUMA was induced by sunitinib in all five cell lines (Fig. 1A). PUMA induction was dosedependent and occurred in both HCT 116 WT and p53 knockout (KO) cells (Fig. 1B). Basal levels of PUMA were lower in p53 knockout cells compared to WT cells as previously reported (Fig. 1B) [15]. p53 and p21 were also induced by sunitinib treatment. PUMA mRNA was rapidly induced by sunitinib, preceding protein induction ( Fig. 1C and 1D). These data suggest that PUMA is transcriptionally induced by sunitinib independent of p53 in colon cancer cells.

PUMA Mediates Sunitinib-induced Apoptosis
To examine a potential role of PUMA in sunitinib-induced apoptosis, we compared the responses of HCT 116 cells with isogenic PUMA knockout (KO) cells [15]. PUMA KO cells were highly resistant to sunibinib-induced apoptosis ( Fig. 2A). As expected, activation of caspase-3 and -9, or cytosolic release of cytochrome c or Smac was impaired in PUMA KO cells, but not in p53 KO cells (Fig. 2B). In long-term colony formation assays, PUMA KO cells generated over 4 times more colonies than WT cells (Fig. 2C). PUMA was also significantly induced by sunitinib in p53 mutant DLD 1 cells while PUMA deficiency significantly blocked sunitinib-induced apoptosis and caspase activation in these cells (Fig. 2D). We also examined the levels of several BH3only proteins and antiapoptotic Bcl-2 family members after sunitinib treatment. No consistent change was observed in most of them, except for a rapid Mcl-1 downregulation and a delayed induction of Bim after 24 hours, in HCT 116 cells (Fig. S1A). However, little or no Mcl-1 downregulation or Bim induction was observed in DLD1 cells (Fig. S1B). Interestingly, the levels of several Bcl-2 family members increased in PUMA KO cells compared to WT cells, perhaps reflecting their degradation by proteases activated during the treatment and apoptosis (Fig. S1B). These results suggest that PUMA plays a key role in the apoptotic responses to sunitinib in colon cancer cells.

The Mechanism of PUMA Induction by Sunitinib
The PI3K/AKT pathway is a common effector downstream of multiple kinases targeted by sunitinib. We therefore examined AKT activation in a time course experiment following sunitinib treatment, and found AKT de-phosphorylation within minutes (Fig. 3A). FoxO3a is a transcription factor and well-established target of AKT, and its phosphorylation by AKT leads to inactivation and nuclear exclusion. Sunitinib treatment led to a rapid de-phosphorylation of FoxO3a, yet had no obvious effect on total FoxO3a levels (Fig. 3A). It is noted that changes in FoxO3a and AKT phosphorylation, or PUMA mRNA are dynamic and transient; while induction of PUMA protein is persistent and uniform across different cell lines (Figs. 1 and 3A). Exogenous expression of active AKT suppressed PUMA induction by sunitinib (Fig. 3B). The induction of PUMA was significantly inhibited by FoxO3a knockdown by either transient expression of siRNA or stable expression of shRNA in both WT and p53 KO HCT 116 cells (Fig. 3C and 3D). p53 KO cells were used to reduce p53-depedent PUMA induction by transfection.
We next determined whether FoxO3a directly activates PUMA transcription by chromatin Immunoprecipitation (ChIP) assay. Two FoxO3a binding sites are located in the first intron of PUMA [19]. The recruitment of FoxO3a to the PUMA promoter containing these sites increased as early as 8 hours following sunitinib treatment (Fig. 4A). Using reporter assays, we found that mutations in the FoxO3a binding sites significantly reduced PUMA reporter activity following sunitinib treatment (Fig. 4B). Furthermore, FoxO3a stable knockdown cells were found to be resistant to sunitinib-induced apoptosis (Fig. 4C). Mcl-1 levels were restored by FoxO3a siRNA in sunitinib-treated cells (Figs. 1D and 3C, S1), suggesting its degradation could be an additional mechanism of sunitinib-induced apoptosis in HCT 116 cells. Sunitinib-induced apoptosis occurred on other CRC lines including HT29, and was suppressed by overexpression of Mcl-1 or Bcl-2 (Figs. 4D and S2). These data indicate that FoxO3a regulates PUMA induction and the mitochondrial pathway in sunitinib-induced apoptosis.

BH3 Mimetics or Elevated PUMA Levels Sensitize Colon Cancer Cells to Sunitinib
The above observations predict high levels of PUMA, BH3only proteins, or small molecule BH3 mimetics sensitize cancer cells to sunitinib-induced apoptosis. Several BH3 mimetics including HA14-1, gossypol and ABT-737, and PUMA adenovirus (Ad-PUMA [15]), were able to sensitize HCT 116 cells to sunitinib. These agents alone induced little or limited apoptosis (Fig. 5A). Interestingly, BH3 mimetics induced significant apoptosis and caspase activation in PUMA KO cells when combined with sunitinib ( Fig. 5A and 5B). DNA damaging agents such as 5-FU induces PUMA and Noxa in p53 WT cells [14,32,33,34]. 5-FU also synergized with sunitinib to induce apoptosis in HCT 116 cells (Fig. 5C), which is associated with enhanced PUMA induction (Fig. S3). This synergy was attenuated but not blocked in PUMA KO cells (Fig. 5C) perhaps due to modulations of other Bcl-2 family members via both p53-dependent and independent mechanisms. These data demonstrate that elevated levels of PUMA or BH3 mimetics can enhance apoptotic responses to sunitinib, even in apoptosisresistant cells.

PUMA Mediates the Therapeutic Responses to Sunitinib in Xenograft Models
To assess whether PUMA modulates therapeutic responses in vivo, WT and PUMA KO HCT 116 cells were injected subcutaneously into the flanks of BALB/c (nu/nu) nude mice to establish xenografts. Compared to the vehicle, sunitinib treatment resulted in 62% and 38% growth inhibition in HCT 116 WT and PUMA KO tumors, respectively (Fig. 6A). Differences between PUMA genotypes were statistically significant in the sunitinib arm (P,0.01), but not in the vehicle arm regarding efficiency or growth rate in tumor establishment (Fig. 6A).
PUMA was significantly induced in WT tumors from sunitinibtreated mice (Fig. 6B). Modulations of Mcl-1, phosphorylated FoxO3a and AKT were also observed ( Fig. 6B and 6C). Analyzing tumor sections from mice one day after the third sunitinib administration (day 4), we found a significantly lower rate of apoptosis and higher rate of cell proliferation in PUMA KO tumors when compared to WT tumors (Figs. 6D and S4A). Active caspase-3 staining confirmed significantly reduced apoptosis in PUMA KO tumors (,80%), compared to WT tumors treated identically (Fig. S4B). These results demonstrate that the therapeutic response to sunitinib in vivo is mediated by PUMAdependent apoptosis.

Discussion
Emerging evidence suggests that induction of apoptosis is an important mechanism of a wide variety of anticancer agents [2,10,35]. Evasion of cell death is a hallmark of cancer and an important contributor to therapeutic resistance [2,3]. In addition to well-documented effects of sunitinib in inhibiting tumor angiogenesis [6,7,36,37], our work demonstrates that sunitinib exhibits a strong pro-apoptotic activity in colon cancer cells via PUMA induction through transcription factor FoxO3a, but not p53, NF-kB p65 or p53 homologue p73 and p63. Sunitinibinduced apoptosis is associated with the induction of Bim or down regulation of Mcl-l in some colon cancer cell lines we tested. Earlier work demonstrated the involvement of Bim and STAT3 during sunitinib-induced apoptosis in other cells types [38,39,40]. Together, these data suggest that induction of BH3-only proteins might be a common mechanism underlying sunitinib-induced cancer cell killing that might be affected by status of various kinases, and different BH3-only proteins might be important in different cells types.
A number of more selective VEGFR inhibitors were also found to induce PUMA and apoptosis in colon cancer cells (data not shown), supporting a non-angiogenic role of anti-VEGFR therapies. It will be important to determine whether sunitinibinduced apoptosis is mediated by PUMA or other BH3-only proteins in other solid tumors such as renal cancer and GISTs, and their potential role in the apoptotic responses to other VEGFR and PDGFR inhibitors. Reduced sensitivity to sunitinib was suggested to be linked to mutations in KIT or VEGFR/FDGFR or in other RTKs, as well as decreased expression of soluble VEGF receptors (sVEGFs) [6,7], which can suppress cell death or promote survival [2,5]. In addition, inhibition of tumor angiogenesis or other components in the microenvironment might indirectly activate cell death pathways [2]. The use of isogenic cell lines as we did here could be particularly useful in understanding drug targets and mechanisms [41].
Despite the excitement in the development of agents targeting oncogenic kinases, clinical data demonstrate that most of these agents are generally efficacious only in a minor fraction of patients [4,5]. One major change is to identify biomarkers to help patient selection and stratification. Our data showed that PUMA persists 24-48 hours after sunitinib treatment. In contrast, inhibition of AKT/FoxO3a is more transient and recovers in hours, likely reflecting secondary and survival attempts in tumor cells following activation of apoptotic signaling. These findings potentially explain why upstream signaling molecules are less suitable as biomarkers, and suggest modulation of the mitochondrial death pathway might be a more useful readout for the overall therapeutic activity of anticancer agents. It may be particularly relevant to explore the regulation of Bcl-2 family members as they are rarely found mutated in solid tumors [2,3], and can be functionally achieved by ''BH3 profiling'' as demonstrated recently in leukemia patients following chemotherapy [42,43]. Current biomarkers of sunitinib include plasma levels of VEGF, (soluble) sVEGFR2, sVEGFR3, and sKIT [7,8]. Interrogation of changes in PUMA and other Bcl-2 family members using biopsies collected before and after treatment would certainly be possible as well as informative. Therefore, a combination of markers monitoring tumor microenvironment, angiogenesis (non cell-autonomous), cell survival and death (cell-autonomous) might ultimately prove more useful. Current treatment modalities provide limited benefits to patients with advanced colorectal cancer [1], and targeted agents might bring new hopes. A better understanding of their mechanisms of action and identification of biomarkers are expected to help guide their further development, clinical testing and use. Our work suggests that PUMA induction predicts the apoptotic responses to sunitinib in colon cancer cells, and provides potential strategies for combination therapies. In preclinical models, small molecule BH3 mimetics synergize with a wide variety of anticancer agents, and generally have none-overlapping side effects with chemo-drugs or kinase inhibitors [7,44,45]. Therefore, it is possible that carefully designed combinations will provide more effective treatment and long-term management of cancer [2].
In summary, our study provides a novel anti-tumor mechanism of sunitinib. In line with recent findings on several BH3-only proteins in targeted therapies [13,19,20,46,47,48], we propose that induction, rather than the steady-state levels, of selective BH3-only proteins as potential biomarkers in both treatment naive and none-naïve patients. This is an important concept as many kinases inhibitors are used in heavily pretreated patients whose cancer might already have elevated levels of BH3-only proteins. It will be important to determine whether BH3 mimetics or certain chemodrugs can improve the efficacies of kinase inhibitors in preclinical models and clinical trials.