Combined AKT and MEK Pathway Blockade in Pre-Clinical Models of Enzalutamide-Resistant Prostate Cancer

Despite recent improvements in patient outcomes using newer androgen receptor (AR) pathway inhibitors, treatment resistance in castrate resistant prostate cancer (CRPC) continues to remain a clinical problem. Co-targeting alternate resistance pathways are of significant interest to treat CRPC and delay the onset of resistance. Both the AKT and MEK signaling pathways become activated as prostate cancer develops resistance to AR-targeted therapies. This pre-clinical study explores co-targeting these pathways in AR-positive prostate cancer models. Using various in vitro models of prostate cancer disease states including androgen dependent (LNCaP), CRPC (V16D and 22RV1) and ENZ-resistant prostate cancer (MR49C and MR49F), we evaluate the relevance of targeting both AKT and MEK pathways. Our data reveal that AKT inhibition induces apoptosis and inhibits cell growth in PTEN null cell lines independently of their sensitivity to hormone therapy; however, AKT inhibition had no effect on the PTEN positive 22RV1 cell line. Interestingly, we found that MEK inhibition had greater effect on 22RV1 cells compared to LNCaP, V16D or ENZ-resistant cells MR49C and MR49F cells. In vitro, combination AKT and MEK blockade had evidence of synergy observed in some cell lines and assays, but this was not consistent across all results. In vivo, the combination of AKT and MEK inhibition resulted in more consistent tumor growth inhibition of MR49F xenografts and longer disease specific survival compared to AKT inhibitor monotherapy. As in our in vitro study, 22RV1 xenografts were more resistant to AKT inhibition while they were more sensitive to MEK inhibition. Our results suggest that targeting AKT and MEK in combination may be a valuable strategy in prostate cancer when both pathways are activated and further support the importance of characterizing the dominant oncogenic pathway in each patient’s tumor in order to select optimal therapy.


Introduction
Medical or surgical castration remains the first line of systemic therapy for metastatic prostate cancer (PCa) since its discovery over 70 years ago [1]. Unfortunately, cure remains elusive following castration and patients inevitably progress to develop castrate resistant prostate cancer

Cell proliferation assays
Cell viability was assessed in 96-well culture plates using the WST-1 reagent and/or crystal violet assay, as described previously [26].

Cell cycle analysis
Cell cycle analysis with propidium iodide staining was performed as previously described [29]. Relative DNA content was analyzed by FACS Canto II flow cytometer using the cyflogic v1.2.1 software (www.cyflogic.com) for analysis.

Caspase-3 activity assay
Caspase-3 activity was assessed using the Caspase 3 Assay kit, with the acetyl Asp-Glu-Val-Asp 7-amido-4-methylcoumarin (Ac-DEVD-AMC) fluorometric substrate (Enzo Scientific). Thirty micrograms of whole cell lysate was incubated with caspase-3 substrate AC-DEVD-AMC at 37.5°C for 2.5h and caspase-3 activity was quantified with a fluorometer with excitation set at 365nm and emission 460nm. Fold change differences from control were calculated following subtraction of readings from blank wells without lysate.

Quantitative reverse transcription (RT)-PCR
RNA extraction and RT-PCR were performed as previously described [30]. Real time monitoring of PCR amplification of cDNA was performed using the following primer pairs and probes: AR (Hs00171172_m1), PSA (Hs00426859_g1), and GAPDH (Hs03929097_g1) (Applied Biosystems, Foster City, CA) on the ABI PRISM 7900 HT Sequence Detection System (Applied Biosystems) using TaqMan Gene Expression Master Mix (Applied Biosystems). Target gene expression was normalized to GAPDH levels in respective samples as an internal control. both flanks. Nine mice per arm were randomized to vehicle (0.1% methylcellulose), AZD5363 100mg/kg BID, PD0325901 5mg/kg OD or the combination. ENZ 10mg/kg was administered prior to inoculation and continued until tumors reached 200mm 3 . Tumor measurements were measured biweekly and tumor volume calculated using the formula l x w x d x 0.5236. Drugs were administered as an oral gavage 5 days on, 2 off. PSA levels were measured weekly using automated enzymatic immunoassay (Cobas, Montreal, Quebec, Canada). Mice were sacrificed if the total tumor burden was >2000mm 3 or had >20% loss of body weight. For the 22RV1 xenografts, 2x10 6 cells were inoculated into the left flank of nude castrate mice. Eight mice per arm were randomized to vehicle, AZD5363 100mg/kg BID, selumitinib (AZD6244; ARRY-142886) 25mg/kg BID, and the combination. The 22RV1 xenografts were sacrificed when tumor volume was >1500mm 3 . All mice were monitored regularly for their clinical condition under the supervision of a veterinarian at the University of British Columbia. At sacrifice, all mice were deeply anesthetized with isoflurane prior to being euthanized with CO2. Tumors collected at sacrifice were divided into parts and snap frozen in liquid nitrogen or fixed in formalin. All animal procedures were performed according to Canadian Council on Animal Care guidelines and with approval of the Animal Care Committee of the University of British Columbia (protocol # A12-0210).

Immunohistochemistry
A tissue microarray of all MR49F xenograft tumors was constructed using a manual tissue microarrayer (Beecher Instruments, Inc., Sun Prairie, WI). Immunohistochemical staining was performed as previously reported [31]. All comparisons of staining intensities were done at 20X magnification on triplicate samples by a pathologist blinded to treatment assignment.

Statistical analysis
All results are expressed as the mean ±SEM, with one-way ANOVA used to detect significant differences between multiple treatments. When ANOVA showed significant differences (p<0.05), Tukey's HSD post-hoc test was used to compare means. Tumor growth velocity was calculated using linear regression. Kaplan Meier survival analysis compared cancer specific survival (CSS) and overall survival, with the log-rank test used to compare groups. CSS was defined as time from treatment start until total tumor volume exceeded 2000mm 3 and overall survival as the time from treatment start until sacrifice. The combination index (CI) was calculated using the Calcusyn software (Biosoft, Cambridge, UK). A CI <1 indicates synergy, a CI >1 indicates antagonistic interactions.

Effect of targeting MEK and AKT pathways on AR signaling pathway
To investigate the relevance of AKT and MEK pathways in PCa, we targeted respective pathways using AZD5363 (AKT inhibitor) and PD0325901 (MEK inhibitor). We evaluated their effects alone and in combination in models of different stages of prostate cancer including androgen sensitive cells (LNCaP), castrate resistant (V16D, 22RV1) and ENZ-resistant cells (MR49C and MR49F) (Fig 1). Blocking AKT with AZD5363 was evaluated by its effect on AKT downstream effector p6SK and not on AKT phosphorylation itself because of the nature of AZD5363. Basically, AZD5363 induces increased AKT phosphorylation which is inactive, a phenomenon occurs with many ATP competitive, catalytic inhibitors of AKT, and is due to the protein being held in a hyperphosphorylated but catalytically inactive form as a consequence of compound binding as has been established previously. AZD5363 was found to induce a decrease in S6K phosphorylation in all cell lines; this effect was more pronounced in the PTEN null cells MR49C, MR49F, LNCaP and V16D cells compared to PTEN positive 22RV1 cells (Fig 1). Similar effects were observed using other MEK inhibitor (UO126) and AKT inhibitor (LY294006) or in combination with Ay (Fig 1, S2 Fig). In 22RV1 cells, PD325901 completely abrogates ERK phosphorylation while AZD5363 had no effect on AKT downstream effector S6K phosphorylation which was only with combination of AZD5363 and PD0325901 (Fig 1). However, no additional decrease in pERK levels was observed with the combination compared to MEK inhibitor monotherapy. Moreover, we observed that AKT inhibition increased both AR and PSA at protein and RNA levels in LNCaP and its derivatives (Fig 1). Although AZD5363 increased AR and PSA levels in all cell lines, the effect of PD0325901 on AR and PSA levels alone or in combination with AZD5363 varied between cell lines (Fig 1). In contrast to LNCaP-based cell lines, PSA was further increased by the combination of AZD5363 and PD0325901 in 22RV1 cells. Notably, these results mirrored similar data with the previously tested successful combination of AZD5363 and ENZ where the changes in AR expression also differed between 22RV1 and LNCaP-based cell lines (S2 Fig) [32]. Taken together, we conclude that AKT and MEK cooperate to regulate the AR pathway.

Effect of combination of MEK and AKT blockade on cell apoptosis and proliferation
To determine the biological activity of targeting AKT and MEK signaling pathways on cell apoptosis, we treated our panel of cell lines with AZD5363, PD0325901 or the combination. Our data showed that combination of targeting AKT and MEK induces apoptosis as shown by increased sub G0/G1 cell cycle population. This effect was greater than AZD5363 monotherapy in CRPC cells V16D (17% vs 10%, P = 0.009) and ENZ-resistant MR49C (29% vs 12%, p = 0.005) and MR49F (12% vs 3%, p = 0.006) cells. None of the treatments exhibited any significant changes in the sub G1/G0 fraction in 22RV1 cells (Fig 2). In contrast, in androgen-sensitive LNCaP cells, combination treatment therapy did not show further induction of sub G1/ G0 compared to AZD5363 monotherapy (17% vs 16%, p = 0.76). S-phase and G2/M fractions appeared to decrease to a similar amount in MR49C, MR49F and V16D cells with AZD5363 or the combination, with significant differences for both treatments from control(P<0.001). Moreover, AZD5363 and PD325901 combination treatment induced cleaved PARP in LNCaPbased cell lines compared to PTEN positive cells 22RV1 cells (Fig 2). While not statistically different, similar trends were observed with caspase 3 activity assays (Fig 2).
We next investigated if the effect observed in cell apoptosis can be translated to cell viability. Our data show that targeting AKT using AZD5363 affects cell viability in all cell lines tested (Fig 3) while targeting MEK using either PD0325901 (Fig 3) or UO126 (S3 Fig) had greater activity in 22RV1 cells compared to the other cell lines further confirming our data on cell cycle population and PARP cleavage. Although we observed trends for decreased viability and synergy with the combination compared to monotherapy in LNCaP cells and V16D cells, in MR49C and MR49F cells the combination did not appear synergistic (Fig 3). This was further confirmed using the combination of MEK inhibitor U0126 with the AKT inhibitor LY294006 in MR49C, MR49F cells (S3 Fig). Synergy was observed in 22RV1 cells with a combination index <1 (Fig 3, S3 Fig).
Targeting AKT and MEK pathways in combination inhibit tumor growth and improved cancer specific survival Since the activity of combination therapy showed differences related to PTEN expression, we investigated in vivo how targeting AKT and MEK will influence tumor growth in two possible clinical settings (PTEN positive and negative tumors). We used ENZ-resistant MR49F cells and CRPC 22RV1 cells to assess the in vivo efficacy of the combination of targeting AKT and MEK pathways.
The MR49F xenografts were sensitive to AZD5363, making assessment of additional benefit with MEK blockade challenging. Monotherapy treatment with PD0325901 demonstrated tumor growth inhibition and demonstrated a non-significant trend to lower serum PSA compared to vehicle (Fig 4). Combination treatment with AZD5363 and PD0325901 in the MR49F xenografts did not result in greater decreases in tumor volume compared to AZD5363 monotherapy, but the combination did have significantly improved cancer specific survival (CSS) (P<0.001) compared to vehicle-and PD0325901-treatment arms (Fig 4). Median CSS was 17 days for vehicle-treated mice, 25 days for PD0325901-treated mice, and was not reached for either AZD5363 alone or in combination with PD0325901. No mice in the combination arm were euthanized due to tumor size after 35 days of treatment. Serum PSA was significantly reduced in both the AZD5363 and the combination treatment arms compared to vehicle (Fig  4).
We next assessed the combination of AKT and MEK pathway inhibition using 22RV1 xenografts in castrated mice. AZD5363 was again used as an AKT inhibitor while Selumetinib was chosen as a MEK inhibitor which may be more clinically relevant as it is currently in clinical evaluation and was recently approved for treatment of melanoma. In contrast to the MR49F model, 22RV1 tumors were relatively insensitive to AZD5363, but did demonstrate tumor growth inhibition to both Selumetinib and the combination of Selumetinib plus AZD5363. With all 22RV1 tumors growing relatively robustly despite treatment, no significant differences were found between treatment arms, though the greatest tumor growth inhibition was seen with combination treatment (Fig 4).
Analysis of the individual mice tumor volumes in the MR49F model suggests that combined blockade may benefit a subset of mice (Fig 5). In the AZD5363 monotherapy arm, while most tumors responded well, in select cases, the tumor growth inhibition was minimal. However, in the combination therapy group, no outliers were noted (Fig 5). Notably, immunohistochemistry staining of the MR49F tumors demonstrated pERK staining in the mice which demonstrated resistance to AZD5363 (Fig 5). No significant pERK staining was seen in the mice who responded to the treatment (Fig 5) nor were differences in pS6 staining or Ki67 staining detected between combination treatment and the AZD5363 monotherapy arm (Fig 5).

Discussion
Despite the success of castration and newer potent AR pathway inhibitors such as enzalutamide, treatment resistance occurs in all cases. Persistent AR signaling is usually clinically evident by a rising PSA. The persistence of AR-signaling may be driven by several mechanisms of resistance including AR splice variants, intra-tumoral production of androgens and activation of alternate pathways which support AR signalling, including the AKT and MEK pathways [33]. Combination therapy in PCa holds potential for increasing patient survival by blocking these resistance pathways. Early clinical experience in PCa patients suggests that targeting AKT alone is not an effective strategy. Clinical trials of AKT inhibitor monotherapy in CRPC have demonstrated limited success with this approach [34,35]. Prior in vitro work suggests that PI3K/AKT inhibition can  Combination Targeting of AKT and MEK Pathways in Prostate Cancer result in up-regulation of the Raf/MEK/ERK pathway [36], and our results in the MR49F xenograft model also suggests this may occur in some cases. The well-recognized crosstalk between these pathways provides a rationale for combining both of these inhibitors [19][20][21][22][37][38][39]. Clinical experience with dual targeting of MEK and AKT pathways is limited; early results in advanced malignancies suggests that dual targeting of both pathways improves oncologic efficacy at the cost of greater toxicity [40].
Our study highlights the differences which can occur between AR positive PCa models and especially the relevance of PTEN status when targeting AKT and MEK pathways. Interestingly, targeting the AKT pathway increased PSA transcription in the surviving cells and this effect, noted previously by others [41], may be important to consider in the design of future clinical trials with these targeted agents since PSA is commonly used as a marker of progression. In the LNCaP-derived ENZ-resistant models, MEK inhibition added a modest benefit compared to AKT inhibition; interestingly the synergy appeared greater in vivo than in vitro. In 22RV1 cells, MEK inhibition alone had significant impact on cell proliferation and signaling, but the combination of MEK and AKT inhibition demonstrated minimal additional benefit in vitro. Thus, while improvements in anticancer activity were noted by some metrics with the combination in each model, each model appeared relative addicted to AKT and MEK, respectively, thus limiting the benefit of combination therapy.
Combined targeting of the MEK and AKT pathway has been investigated in several other cancers in pre-clinical models, with results supporting the use of the combination [19][20][21][22][37][38][39]. In a pre-clinical prostate cancer in vivo study, PD0325901 was assessed in combination with the mTOR inhibitor rapamycin [14]. While we did not observe any significant decrease in Ki67 with the combination, this may be related to collecting all of our tumors at the end of study at which point the xenograft tumors were treatment-resistant. Recently, Park et al, showed the combination of selumitinib and the pan-PI3K inhibitor GSK2126458 improved tumor growth inhibition compared to either monotherapy in AR-negative DU145 and PC3 xenografts [42]. Overall, their results in AR-negative models are comparable to our results. Taken together, our data suggest that the efficacy of targeting AKT and MEK is independent of the disease state of prostate cancer and that the efficacy correlates with the activation of these pathways in the tumor.
Our study is limited by several aspects of our prostate cancer models. LNCaP cells appear to be highly dependent on the AKT pathway. Nonetheless, as AR positive, PSA-producing cell lines with an activated PI3K/AKT pathway, they do resemble a common CPRC phenotype encountered [43]. Moreover, our ENZ-resistant models demonstrate several phenotypes consistent with clinical ENZ-resistant disease, such as persistent AR nuclear localization, AR F876L mutation, and upregulation of steroiodogenic enzymes [26,32,44]. 22RV1 cells harbor androgen splice variants, which predict a poorer response to AR-pathway inhibitors [45]. The presence of the enzalutamide-agonistic F876L mutation and the presence of dominant AR splice variants limited our ability to test for the efficacy of combined AKT and MEK inhibition together with AR inhibition in advanced prostate cancer models. Further, while our in vitro studies were not performed in androgen deficient media, it is unclear whether this would result in significantly different results, particularly as CRPC tumors can produce their own androgens through de novo synthesis pathways [46]. Accordingly, our results in castrate conditions in vivo were not substantially different.
In conclusion, our results in ENZ-resistant CRPC cell models suggest that combination treatment with AKT and MEK inhibition may be a rational combination in a subset of resistant prostate cancer cases needing further investigation. Our results also suggest the importance of targeting inhibitors to tumor pathways which are known to be activated; in well-selected patients monotherapy may be as effective as combination therapy. Overall, this emphasizes the need for rational, biomarker-driven selection of patients as targeted therapeutics are evaluated in clinical trials.