PIM Kinase Inhibitor AZD1208 for MYC-Driven Prostate Cancer
We conducted four independent AZD1208 treatment trials using nude mice carrying prostate tissue recombinant grafts overexpressing c-MYC and Pim1 (Figure 1, A and B). Prostate grafts were generated by coinfecting primary mouse prostate epithelial cells with lentiviruses coexpressing c-MYC or Pim-1 with yellow fluorescent protein (YFP), combined with urogenital mesenchyme in collagen and implanted under the renal capsule. AZD1208 (30mg/kg or 45mg/kg) or vehicle was given daily by oral gavage for two or three weeks (Figure 1A). Mice were weighed daily (Supplementary Figure 1A, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online). Two mice in the AZD1208 group receiving 45mg/kg lost between 16% to 19% of body weight, suggesting that dose toxicity level was reached on this schedule. Mice treated with AZD1208 for three weeks at either 30 or 45mg/kg had statistically significantly smaller grafts (P < .003) (Figure 1C). For one set of mice treated with 45mg/kg AZD1208, the grafts were 54.3±39% (P < .001) smaller than controls.
(Enlarge Image)
Figure 1.
PIM kinase inhibitor AZD1208 treatment scheme and histological analysis of treated prostate tissue recombinant grafts. A) Scheme for AZD1208 treatment (four trials) of MYC/Pim1-expressing prostate recombination grafts showing the timing and duration of treatment (shaded bars). B) Numbers of mice and individual grafts (number in parentheses) enrolled in the treatment trials. C) Relative change between vehicle-treated and AZD1208-treated tissue recombination graft volumes at harvest. Results are shown as mean ± SD. *Student's t test P < .05. #Mann-Whitney rank sum test significance at < .05. All statistical tests were two-sided. D) Hematoxylin and eosin (H&E)–stained sections demonstrate more severe lesion (HGPIN) in the vehicle-treated specimens compared with the 30mg/kg AZD1208-treated specimen. Corresponding sections demonstrate yellow fluorescent protein (YFP) immunofluorescence (detected with anti-GFP antibody), indicating the c-MYC/Pim1–transduced cells. E) H&E–stained sections with corresponding YFP by immunohistochemistry (brown) detected with anti-GFP antibody similarly demonstrate worse pathology (HGPIN) in the vehicle-treated specimen compared with the 45mg/kg AZD1208-treated specimen. Scale bars = 0.1mm (short bar), 0.05mm (long bar), respectively, for (D) and (E). H&E = hematoxylin and eosin stain; GFP = green fluorescent protein.
Histological analysis demonstrated more high-grade PIN (HGPIN) lesions in the vehicle-treated graft sections than AZD1208-treated samples (Figure 1, D and E). We could identify prostate glands derived from transduced cells by YFP coexpressed by the bicistronic lentiviral vector along with either c-MYC or Pim1. In all vehicle-treated and drug-treated specimens examined, Pim1 and c-MYC colocalized with YFP (Figure 2). No glands could be identified that expressed only c-MYC without Pim1 or vice versa, making YFP a valid surrogate marker for c-MYC and Pim1 expression in this model.
(Enlarge Image)
Figure 2.
PIM kinase inhibitor AZD1208 effect on PIN, proliferation, and apoptosis in prostate grafts. A) Incidence of PIN foci is lower in the AZD1208-treated samples compared with vehicle-treated grafts (χP = 1.65x10). B) PIN density measured as number of PIN foci per section is decreased in the AZD1208-treated samples compared with vehicle-treated specimens (χP = 3.06x10). N = 88 sections were analyzed for (A) and (B). C) Smooth muscle actin (SMA) immunostaining demonstrates substantial thinning of the SMA-containing boundary around abnormal prostate gland lesions in both the vehicle-treated and AZD1208-treated specimens (arrows). D) BrdU-positive nuclei (arrows) are fewer in AZD1208-treated specimens than controls. E) Increased cleaved caspase-3–positive cells (arrows) in AZD1208-treated specimens than the vehicle-treated specimens. Yellow fluorescent protein (YFP) was used to guide analysis to c-MYC/Pim1-expressing areas in "D" and "E". F) The percentage of BrdU-positive nuclei is lower in the AZD1208-treated (n = 11) specimens than in the vehicle-treated samples (n = 8) (P = .0016). G) The percentage of cleaved caspase-3–positive cells is greater in the AZD1208-treated (n = 11) specimens than in the vehicle-treated specimens (n = 11) (P = .01 for 30mg/kg, P = .039 for 45mg/kg). Results shown as mean ± SD. *Student's t test P < .05. #Mann-Whitney rank sum test significance at < .05. Scale bars = 0.05mm.
PIN incidence was reduced in a dose-dependent manner in AZD1208-treated samples as determined by a blinded pathologist (Figure 2A). While 91% of 22 analyzed vehicle-treated grafts contained PIN foci, only 50% of 10 grafts and 25% of 12 grafts treated with AZD1208 at 30 and 45mg/kg contained PIN foci, respectively (P < .001) (Figure 2A). AZD1208 treatment also reduced PIN density in a dose-dependent manner (P < .001) (Figure 2B). Both vehicle-treated and drug-treated specimens had lesions with thinning of the smooth muscle actin (SMA) layer (Figure 2C). We next analyzed the YFP-positive gland regions for proliferation and apoptosis by immunostaining for BrdU and cleaved caspase-3, respectively. AZD1208-treated samples at 45mg/kg had a 46±14% (P = .016) lower BrdU index and a 326±170% (P = .039) higher apoptotic index than vehicle-treated controls (Figure 2, D-G). A statistically significant increase in apoptosis of 200±100% (P = .01) was observed in AZD1208-treated specimens at 30mg/kg (Figure 2G). These data are consistent with previous studies showing decreased tumorigenicity of prostate cancer cells after Pim1 knockdown by shRNA.
Pim1 can modulate apoptosis by phosphorylating Bad at serine-112. Using YFP as a guide to analyze lentivirus transduced glands, we found that AZD1208 treatment decreased phospho-Bad(S112) immunostaining compared with vehicle (Figure 3, A and B). To test whether Pim inhibition modulated the activity of MYC, we assessed the effect of AZD1208 on the expression of serine-62–phosphorylated MYC and on MYC-regulated targets hexokinase II, Nedd4l, and Prdx6. AZD1208 treatment decreased the expression of phospho-S62-c-MYC and Hk2, with variable effects on Nedd4L and Prdx6 (Supplementary Figure 3, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online).
(Enlarge Image)
Figure 3.
AZD1208 treatment effect on phospho-Bad staining in prostate tissue recombinant grafts. A) Vehicle-treated and (B) AZD1208-treated tissue recombinant grafts with adjacent sections stained by hematoxylin and eosin (H&E), yellow fluorescent protein (YFP) (anti-GFP antibody), and phospho-Bad-S112 antibody. Abnormal gland foci are indicated by arrows. Phospho-Bad(S112) expression is reduced within the abnormal gland foci in AZD1208-treated specimens compared with vehicle-treated samples, indicating that PIM kinase function is inhibited in the prostate tissue. Scale bars = 0.1mm (short bar) and 0.05mm (long bar). H&E = hematoxylin and eosin; GFP = green fluorescent protein.
In order to explore the effects of Pim inhibition on established MYC-driven prostate tumors and the effects of combination therapy on recurrence, we turned to Myc-CaP cells, established from the Hi-Myc transgenic model of prostate cancer. We first showed that AZD1208 inhibited growth of Myc-CaP colonies in vitro in a concentration-dependent manner (Figure 4, A and B). Next, we established allograft Myc-CaP tumors in nude mice, and after tumors were grown for two weeks (average tumor size = 207.4 ± 34.7mm) the mice were treated with AZD1208 once daily by oral gavage for three weeks. Drug treatment was well tolerated (Supplementary Figure 1B, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online) and resulted in statistically significant inhibition of tumor growth (Figure 4C). Histologically, AZD1208-treated grafts contained more voids, consistent with the presence of increased apoptotic cells (186±94%, P = .011) (Figure 4E) and reduction in BrdU labeling index (75±73%, P < .001) (Figure 4D). AZD1208 also suppressed the growth of prostate cancer xenografts DU145 and CWR22rv1 (Figure 4, F and G), comparable with the effects of the established cancer drug docetaxel, as well as the growth of bone metastasis–derived PC3 tumor xenografts (discussed below).
(Enlarge Image)
Figure 4.
AZD1208 effects on the growth and tumorigenicity of mouse Myc-CaP and human castrate-resistant DU145 and CWR22Rv1 prostate cancer tumors. A) Myc-CaP colonies are decreased by the AZD1208 in a concentration-dependent manner. B) Clonogenic assay with Myc-CaP cells (replicates n = 3 per concentration) demonstrates decreased surviving fraction for colonies reaching 1mm in diameter at AZD1208 concentrations of 300nM and higher (* P < .05). C) Treatment with AZD1208 statistically significantly decreases the growth of Myc-CaP allografts compared with vehicle-treated control (n = 10 per group, *P < .05). D-E) AZD1208-treated tumors show reduced BrdU-positive nuclei (P < .001) and increased cleaved caspase-3–positive cells (P = .011) compared with vehicle-treated control (n = 3 per group). F) Treatment with AZD1208 statistically significantly decreases the growth of human prostate cancer cell line DU145 in xenografts (n = 10 per group, *P < .01). G) AZD1208 treatment impairs tumorigenicity of human prostate cancer cell line CWR22Rv1 xenografts similar to the effects of docetaxel (n = 10 per group, *P < .05). Results shown are mean ± SD for (B-G). *Student's t test P < .05. #Mann-Whitney rank sum test significance at < .05. All statistical tests were two-sided.
To obtain insights into the mechanisms by which AZD1208 inhibits prostate tumors we performed RNA-seq on Myc-CaP cells treated with vehicle or 1 μM AZD1208 overnight (In Vitro Study–acute treatment) and Myc-CaP tumors in mice treated with vehicle or 45mg/kg AZD1208 for 21 days (In Vivo Study–chronic treatment). Using DAVID functional clustering of statistically significantly altered genes (Supplementary Table 1, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online), we found enrichment of translation processes genes (peptidases, ribosomes), growth arrest genes (cell cycle, chromosomal replication), and perturbed vesicular movement (altered myosin) in AZD1208 treated samples in vivo, suggesting that PIM inhibition impairs tumorigenicity by engaging multiple cancer hallmarks.
Notably, Gene Set Enrichment Analysis revealed striking downregulation of multiple MYC target gene sets, indicating reduced MYC activity in AZD1208-treated cells both in vitro and in vivo (Table 1). PIM1 has been shown to increase MYC protein stability and MYC phosphorylation to enhance activity and to stimulate RNA polymerase II binding to increase the transcription of MYC target genes. Consistent with these earlier studies, our GSEA analyses found, using a global unbiased approach, strong evidence for downregulation of MYC activity by AZD1208 treatment.
We also found evidence of downregulation of p53 transcriptional activity by AZD1208 treatment in vitro and in vivo (Table 1). A link between Pim1 and p53 has been previously suggested by cell culture studies indicating that Pim1 overexpression leads to the stabilization of p53 protein. Pim inhibition can therefore be expected to suppress the p53 pathway by reducing p53 stability. Our results indicate that while AZD1208 treatment activates several anticancer mechanisms, such as reduced proliferation, enhanced apoptosis, and reduced MYC activity, it can also suppress the p53 pathway. One implication of this observation is that therapeutic approaches that counteract AZD1208-induced p53 pathway suppression might enhance the drug's efficacy.
Pim1 expression can be induced by stresses such as hypoxia and radiation in different cell systems as a cell survival mechanism. We reasoned that the efficacy of Pim inhibition might be increased in tumor cells exposed to radiation or hypoxia, which induce Pim1 as a prosurvival mechanism. Additionally, activation of the p53 pathway by these stressors may overcome the p53 pathway suppression observed upon AZD1208 treatment. To test these hypotheses, we subjected Myc-CaP cells to radiation and/or hypoxia. Both the 44 and 33kDa isoforms of Pim1 were increased by exposure to radiation+hypoxia (Figure 5A). Clonogenic assays using Myc-CaP cells treated with radiation and AZD1208 at 100nM indicate that AZD1208 functions as a radiation sensitizer (Figure 5B). Myc-CaP tumors subjected to radiation alone or radiation+hypoxia also showed higher Pim1 expression detected by immunohistochemistry, particularly in the nuclei (Figure 5C).
(Enlarge Image)
Figure 5.
AZD1208 effects on sensitivity of prostate cancer cells to radiation. A) Immunoblots demonstrate induction of Pim1 (44kD and 33 kD isoforms) by radiation and hypoxia in Myc-CaP cells. B) Clonogenic assay demonstrates that AZD1208 acts as a radiation sensitizer by decreasing the surviving fraction when compared with radiation with vehicle (replicates n = 4 per radiation dose). *Student's t test P < .05. All statistical tests were two-sided. C) Increased Pim1 expression in Myc-CaP tumors treated with radiation and hypoxia. Scale bar = 0.05mm.
Having demonstrated that the PIM inhibitor AZD1208 is effective at sensitizing Myc-CaP cells to radiation, we sought to determine if the sensitizing effects of AZD1208 are also manifested in CRPC in vivo. Myc-CaP tumors were allowed to grow in mice to an average size of 500mm. Castration of the mice resulted in tumor regression in all cases over six to nine days. Following a recovery period, the grafts regrew in the androgen-deprived mice. At this point, mice were treated with AZD1208 alone, radiation alone, or a combination of AZD1208 and radiation (Figure 6A). AZD1208 at 45mg/kg was given daily for three weeks. After five doses of AZD1208, three once-daily 2 Gy radiation treatments were delivered. As shown in Figure 6B, treatment with either AZD1208 or radiation alone slowed the growth of recurrent tumors, and combination treatment with AZD1208+radiation resulted in more sustained inhibition of tumor growth. The mean graft growth rates for each treatment group were determined for the last eight days of drug treatment, after completing radiation treatment. An exponential fit to each curve revealed the following growth rates: 11% per day for vehicle treatment, 8.4% per day for radiation treatment, 10% per day for AZD1208 treatment, and 5% per day for combined radiation and AZD1208 treatment (Figure 6C).
(Enlarge Image)
Figure 6.
AZD1208 and radiation treatment of castrate-resistant recurrent prostate cancer. A) Schematic of Myc-CaP allograft experiment and (B) growth curves of Myc-CaP grafts that were treated with androgen deprivation (surgical castration) and followed over time. Tumors regressed, then started regrowing, after which mice were treated daily with Pim kinase inhibitor AZD1208 at 45mg/kg or vehicle. Additionally, some mice received three once-daily treatments of 2 Gy of radiation (XRT). AZD1208-only and radiation-only treated tumors responded initially, but started growing back. AZD plus radiation-treated tumors show sustained inhibition of growth (n = 6–8 per group). Results shown as mean ± SD. *Student's t test P < .05. #Mann-Whitney rank sum test significance at < .05. All statistical tests were two-sided. C) Last 8 days of tumor volume measurements with daily growth percentage determined by the exponent coefficient for an exponential curve fit (R is the goodness of fit).
The MYC transgene in Myc-CaP cells is driven by the androgen-regulated PB4 promoter. We therefore sought to determine the dynamic pattern of MYC and AR expression in Myc-CaP tumors before castration, in regressed tumors postcastration, and in recurrent tumors postcastration. In control uncastrated mice treated with vehicle or AZD1208, we observed robust nuclear MYC expression (Supplementary Figure 4A, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online). Residual regressed tumors postcastration showed a heterogeneous pattern of MYC, while recurrent tumors regardless of whether they also received AZD1208 or radiation treatment expressed nuclear MYC in a homogenous pattern (Supplementary Figure 4A http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1). AR expression was nuclear and homogenous in intact mice with or without AZD1208 treatment (Supplementary Figure 4B, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online). In regressed tumors, AR expression was heterogeneous and present in both the nuclei and cytoplasm. AR expression becomes more homogeneous in recurrent tumors postcastration with primarily cytoplasmic AR expression in AZD1208-treated tumors postcastration (Supplementary Figure 4B, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online). These results suggest the presence of tumor cells that retain MYC and AR expression in regressed tumors postcastration. These cells may be the origin of recurrent tumors.
The fact that postcastration recurrent Myc-CaP tumors subjected to AZD1208 treatment were able to regrow after a lag period prompted us to examine possible mechanisms of this resistance. To this end, we established a panel of six cell lines, each from independent recurrent tumors, three from vehicle-treated (CxVehicle) and three from AZD1208-treated (CxAZD) surgically castrated animals. Analysis of in vitro growth rates showed no consistent differences between the cells derived from vehicle-treated and AZD1208-treated tumors (data not shown). We next used Affymetrix Mouse Gene 2.0 ST Arrays to profile gene expression changes in the cell lines. Using GSEA, we identified several pathways statistically significantly enriched in CxAZD cell lines compared with CxVehicle cells (Figure 7, A and B). Remarkably, the top enriched gene set is p53_DN.V2_UP, representing genes upregulated in HEK293 cells upon P53 knockdown by RNAi. These results suggest that CxAZD cells have lower p53 pathway activity, consistent with our RNA-seq profiling studies of Myc-CaP cells and tumors (Table 1). Another gene set enriched is JAK2_DN.V1_UP, consisting of genes upregulated in human erythroleukemia cells after knockdown of JAK2 by RNAi. PIM kinases have been linked to the Jak/Stat pathway. However, Jak/Stat pathway activity was unchanged in CxAZD cells as measured by p-Stat3 levels (Figure 7C). We confirmed reduced p53 protein levels in CxAZD cell lines by immunoblotting (Figure 7C), and intact/postcastration AZD1208-treated tumors expressed reduced levels of p53 protein by immunohistochemistry (Figure 7D). On the other hand, AZD1208-treated tumors exposed to radiation expressed high levels of p53. Thus, reduced p53 expression may represent a mechanism of resistance to AZD1208 by tumors that can be overcome by radiation treatment, which induces p53. This may provide one possible (but not the only) basis for the cooperativity we observed between AZD1208 treatment and radiation in inhibiting Myc-CaP prostate tumor growth. Indeed, treatment of the p53-null human PC3 prostate cancer cell line xenografts with AZD1208 and/or radiation showed that the combination treatment was still more effective in suppressing tumor growth than either single treatment (Supplementary Figure 5, A and B http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1).
(Enlarge Image)
Figure 7.
GSEA analysis of gene expression changes in AZD1208-resistant castration-resistant prostate cancer (PC) cells. A) The list of the top 10 gene sets enriched in tumor cell lines established from recurrent, castrated, AZD1208-treated mice (CxAZD) relative to cell lines established from castrated, recurrent, vehicle-treated recurrent tumors (CxVehicle). B) Enrichment analysis plots showing enrichment of genes upregulated in P53 knockdown cells (left panel) and JAK2 knockdown cells (right panel) in CxAZD tumor cell lines. C) Immunoblots showing reduced p53 protein levels in AZD1208-treated castrated Myc-CaP allograft cell lines, but unchanged p-Stat3 and Stat3 levels. D) Reduced p53 expression in AZD1208-treated Myc-CaP tumors by immunohistochemical staining both before and after castration (upper four panels), while tumors exposed to radiation treatment (XRT) or radiation plus AZD1208 (XRT/AZD) expressed high levels of p53 (bottom two panels). Scale bar = 0.05mm E) Gene expression profile of AZD1208-resistant recurrent tumor cells is associated with human prostate cancer recurrence. Cross-species Gene Set Enrichment Analysis analysis shows association of gene expression signature from AZD1208-treated recurrent tumors with biochemical recurrence in prostate cancer patients from.
Finally, we sought to determine if genes dysregulated in postcastration AZD1208-resistant recurrent tumors are associated with clinical outcome in human prostate cancer. Using cross-species GSEA analysis, we found that genes upregulated in AZD1208-resistant tumors are statistically significantly enriched in the prostate tumors of patients who experienced biochemical recurrence (Figure 7E). Conversely, genes downregulated in AZD1208-resistant tumors are statistically significantly enriched in patients without recurrence.
Results
AZD1208 Inhibition of c-MYC/Pim1-Prostate Tissue Recombinant Grafts
We conducted four independent AZD1208 treatment trials using nude mice carrying prostate tissue recombinant grafts overexpressing c-MYC and Pim1 (Figure 1, A and B). Prostate grafts were generated by coinfecting primary mouse prostate epithelial cells with lentiviruses coexpressing c-MYC or Pim-1 with yellow fluorescent protein (YFP), combined with urogenital mesenchyme in collagen and implanted under the renal capsule. AZD1208 (30mg/kg or 45mg/kg) or vehicle was given daily by oral gavage for two or three weeks (Figure 1A). Mice were weighed daily (Supplementary Figure 1A, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online). Two mice in the AZD1208 group receiving 45mg/kg lost between 16% to 19% of body weight, suggesting that dose toxicity level was reached on this schedule. Mice treated with AZD1208 for three weeks at either 30 or 45mg/kg had statistically significantly smaller grafts (P < .003) (Figure 1C). For one set of mice treated with 45mg/kg AZD1208, the grafts were 54.3±39% (P < .001) smaller than controls.
(Enlarge Image)
Figure 1.
PIM kinase inhibitor AZD1208 treatment scheme and histological analysis of treated prostate tissue recombinant grafts. A) Scheme for AZD1208 treatment (four trials) of MYC/Pim1-expressing prostate recombination grafts showing the timing and duration of treatment (shaded bars). B) Numbers of mice and individual grafts (number in parentheses) enrolled in the treatment trials. C) Relative change between vehicle-treated and AZD1208-treated tissue recombination graft volumes at harvest. Results are shown as mean ± SD. *Student's t test P < .05. #Mann-Whitney rank sum test significance at < .05. All statistical tests were two-sided. D) Hematoxylin and eosin (H&E)–stained sections demonstrate more severe lesion (HGPIN) in the vehicle-treated specimens compared with the 30mg/kg AZD1208-treated specimen. Corresponding sections demonstrate yellow fluorescent protein (YFP) immunofluorescence (detected with anti-GFP antibody), indicating the c-MYC/Pim1–transduced cells. E) H&E–stained sections with corresponding YFP by immunohistochemistry (brown) detected with anti-GFP antibody similarly demonstrate worse pathology (HGPIN) in the vehicle-treated specimen compared with the 45mg/kg AZD1208-treated specimen. Scale bars = 0.1mm (short bar), 0.05mm (long bar), respectively, for (D) and (E). H&E = hematoxylin and eosin stain; GFP = green fluorescent protein.
Histological analysis demonstrated more high-grade PIN (HGPIN) lesions in the vehicle-treated graft sections than AZD1208-treated samples (Figure 1, D and E). We could identify prostate glands derived from transduced cells by YFP coexpressed by the bicistronic lentiviral vector along with either c-MYC or Pim1. In all vehicle-treated and drug-treated specimens examined, Pim1 and c-MYC colocalized with YFP (Figure 2). No glands could be identified that expressed only c-MYC without Pim1 or vice versa, making YFP a valid surrogate marker for c-MYC and Pim1 expression in this model.
(Enlarge Image)
Figure 2.
PIM kinase inhibitor AZD1208 effect on PIN, proliferation, and apoptosis in prostate grafts. A) Incidence of PIN foci is lower in the AZD1208-treated samples compared with vehicle-treated grafts (χP = 1.65x10). B) PIN density measured as number of PIN foci per section is decreased in the AZD1208-treated samples compared with vehicle-treated specimens (χP = 3.06x10). N = 88 sections were analyzed for (A) and (B). C) Smooth muscle actin (SMA) immunostaining demonstrates substantial thinning of the SMA-containing boundary around abnormal prostate gland lesions in both the vehicle-treated and AZD1208-treated specimens (arrows). D) BrdU-positive nuclei (arrows) are fewer in AZD1208-treated specimens than controls. E) Increased cleaved caspase-3–positive cells (arrows) in AZD1208-treated specimens than the vehicle-treated specimens. Yellow fluorescent protein (YFP) was used to guide analysis to c-MYC/Pim1-expressing areas in "D" and "E". F) The percentage of BrdU-positive nuclei is lower in the AZD1208-treated (n = 11) specimens than in the vehicle-treated samples (n = 8) (P = .0016). G) The percentage of cleaved caspase-3–positive cells is greater in the AZD1208-treated (n = 11) specimens than in the vehicle-treated specimens (n = 11) (P = .01 for 30mg/kg, P = .039 for 45mg/kg). Results shown as mean ± SD. *Student's t test P < .05. #Mann-Whitney rank sum test significance at < .05. Scale bars = 0.05mm.
PIN incidence was reduced in a dose-dependent manner in AZD1208-treated samples as determined by a blinded pathologist (Figure 2A). While 91% of 22 analyzed vehicle-treated grafts contained PIN foci, only 50% of 10 grafts and 25% of 12 grafts treated with AZD1208 at 30 and 45mg/kg contained PIN foci, respectively (P < .001) (Figure 2A). AZD1208 treatment also reduced PIN density in a dose-dependent manner (P < .001) (Figure 2B). Both vehicle-treated and drug-treated specimens had lesions with thinning of the smooth muscle actin (SMA) layer (Figure 2C). We next analyzed the YFP-positive gland regions for proliferation and apoptosis by immunostaining for BrdU and cleaved caspase-3, respectively. AZD1208-treated samples at 45mg/kg had a 46±14% (P = .016) lower BrdU index and a 326±170% (P = .039) higher apoptotic index than vehicle-treated controls (Figure 2, D-G). A statistically significant increase in apoptosis of 200±100% (P = .01) was observed in AZD1208-treated specimens at 30mg/kg (Figure 2G). These data are consistent with previous studies showing decreased tumorigenicity of prostate cancer cells after Pim1 knockdown by shRNA.
Pim1 can modulate apoptosis by phosphorylating Bad at serine-112. Using YFP as a guide to analyze lentivirus transduced glands, we found that AZD1208 treatment decreased phospho-Bad(S112) immunostaining compared with vehicle (Figure 3, A and B). To test whether Pim inhibition modulated the activity of MYC, we assessed the effect of AZD1208 on the expression of serine-62–phosphorylated MYC and on MYC-regulated targets hexokinase II, Nedd4l, and Prdx6. AZD1208 treatment decreased the expression of phospho-S62-c-MYC and Hk2, with variable effects on Nedd4L and Prdx6 (Supplementary Figure 3, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online).
(Enlarge Image)
Figure 3.
AZD1208 treatment effect on phospho-Bad staining in prostate tissue recombinant grafts. A) Vehicle-treated and (B) AZD1208-treated tissue recombinant grafts with adjacent sections stained by hematoxylin and eosin (H&E), yellow fluorescent protein (YFP) (anti-GFP antibody), and phospho-Bad-S112 antibody. Abnormal gland foci are indicated by arrows. Phospho-Bad(S112) expression is reduced within the abnormal gland foci in AZD1208-treated specimens compared with vehicle-treated samples, indicating that PIM kinase function is inhibited in the prostate tissue. Scale bars = 0.1mm (short bar) and 0.05mm (long bar). H&E = hematoxylin and eosin; GFP = green fluorescent protein.
AZD1208 Treatment Inhibition of Established c-MYC-Driven Prostate Cancer
In order to explore the effects of Pim inhibition on established MYC-driven prostate tumors and the effects of combination therapy on recurrence, we turned to Myc-CaP cells, established from the Hi-Myc transgenic model of prostate cancer. We first showed that AZD1208 inhibited growth of Myc-CaP colonies in vitro in a concentration-dependent manner (Figure 4, A and B). Next, we established allograft Myc-CaP tumors in nude mice, and after tumors were grown for two weeks (average tumor size = 207.4 ± 34.7mm) the mice were treated with AZD1208 once daily by oral gavage for three weeks. Drug treatment was well tolerated (Supplementary Figure 1B, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online) and resulted in statistically significant inhibition of tumor growth (Figure 4C). Histologically, AZD1208-treated grafts contained more voids, consistent with the presence of increased apoptotic cells (186±94%, P = .011) (Figure 4E) and reduction in BrdU labeling index (75±73%, P < .001) (Figure 4D). AZD1208 also suppressed the growth of prostate cancer xenografts DU145 and CWR22rv1 (Figure 4, F and G), comparable with the effects of the established cancer drug docetaxel, as well as the growth of bone metastasis–derived PC3 tumor xenografts (discussed below).
(Enlarge Image)
Figure 4.
AZD1208 effects on the growth and tumorigenicity of mouse Myc-CaP and human castrate-resistant DU145 and CWR22Rv1 prostate cancer tumors. A) Myc-CaP colonies are decreased by the AZD1208 in a concentration-dependent manner. B) Clonogenic assay with Myc-CaP cells (replicates n = 3 per concentration) demonstrates decreased surviving fraction for colonies reaching 1mm in diameter at AZD1208 concentrations of 300nM and higher (* P < .05). C) Treatment with AZD1208 statistically significantly decreases the growth of Myc-CaP allografts compared with vehicle-treated control (n = 10 per group, *P < .05). D-E) AZD1208-treated tumors show reduced BrdU-positive nuclei (P < .001) and increased cleaved caspase-3–positive cells (P = .011) compared with vehicle-treated control (n = 3 per group). F) Treatment with AZD1208 statistically significantly decreases the growth of human prostate cancer cell line DU145 in xenografts (n = 10 per group, *P < .01). G) AZD1208 treatment impairs tumorigenicity of human prostate cancer cell line CWR22Rv1 xenografts similar to the effects of docetaxel (n = 10 per group, *P < .05). Results shown are mean ± SD for (B-G). *Student's t test P < .05. #Mann-Whitney rank sum test significance at < .05. All statistical tests were two-sided.
Effect of AZD1208 on Gene Expression in Prostate Tumor Cells
To obtain insights into the mechanisms by which AZD1208 inhibits prostate tumors we performed RNA-seq on Myc-CaP cells treated with vehicle or 1 μM AZD1208 overnight (In Vitro Study–acute treatment) and Myc-CaP tumors in mice treated with vehicle or 45mg/kg AZD1208 for 21 days (In Vivo Study–chronic treatment). Using DAVID functional clustering of statistically significantly altered genes (Supplementary Table 1, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online), we found enrichment of translation processes genes (peptidases, ribosomes), growth arrest genes (cell cycle, chromosomal replication), and perturbed vesicular movement (altered myosin) in AZD1208 treated samples in vivo, suggesting that PIM inhibition impairs tumorigenicity by engaging multiple cancer hallmarks.
Notably, Gene Set Enrichment Analysis revealed striking downregulation of multiple MYC target gene sets, indicating reduced MYC activity in AZD1208-treated cells both in vitro and in vivo (Table 1). PIM1 has been shown to increase MYC protein stability and MYC phosphorylation to enhance activity and to stimulate RNA polymerase II binding to increase the transcription of MYC target genes. Consistent with these earlier studies, our GSEA analyses found, using a global unbiased approach, strong evidence for downregulation of MYC activity by AZD1208 treatment.
We also found evidence of downregulation of p53 transcriptional activity by AZD1208 treatment in vitro and in vivo (Table 1). A link between Pim1 and p53 has been previously suggested by cell culture studies indicating that Pim1 overexpression leads to the stabilization of p53 protein. Pim inhibition can therefore be expected to suppress the p53 pathway by reducing p53 stability. Our results indicate that while AZD1208 treatment activates several anticancer mechanisms, such as reduced proliferation, enhanced apoptosis, and reduced MYC activity, it can also suppress the p53 pathway. One implication of this observation is that therapeutic approaches that counteract AZD1208-induced p53 pathway suppression might enhance the drug's efficacy.
Effect of Radiation and Hypoxia on Pim1 Expression in Prostate Cancer Cells
Pim1 expression can be induced by stresses such as hypoxia and radiation in different cell systems as a cell survival mechanism. We reasoned that the efficacy of Pim inhibition might be increased in tumor cells exposed to radiation or hypoxia, which induce Pim1 as a prosurvival mechanism. Additionally, activation of the p53 pathway by these stressors may overcome the p53 pathway suppression observed upon AZD1208 treatment. To test these hypotheses, we subjected Myc-CaP cells to radiation and/or hypoxia. Both the 44 and 33kDa isoforms of Pim1 were increased by exposure to radiation+hypoxia (Figure 5A). Clonogenic assays using Myc-CaP cells treated with radiation and AZD1208 at 100nM indicate that AZD1208 functions as a radiation sensitizer (Figure 5B). Myc-CaP tumors subjected to radiation alone or radiation+hypoxia also showed higher Pim1 expression detected by immunohistochemistry, particularly in the nuclei (Figure 5C).
(Enlarge Image)
Figure 5.
AZD1208 effects on sensitivity of prostate cancer cells to radiation. A) Immunoblots demonstrate induction of Pim1 (44kD and 33 kD isoforms) by radiation and hypoxia in Myc-CaP cells. B) Clonogenic assay demonstrates that AZD1208 acts as a radiation sensitizer by decreasing the surviving fraction when compared with radiation with vehicle (replicates n = 4 per radiation dose). *Student's t test P < .05. All statistical tests were two-sided. C) Increased Pim1 expression in Myc-CaP tumors treated with radiation and hypoxia. Scale bar = 0.05mm.
Effect of PIM Inhibition and Radiation Treatment on CRPC
Having demonstrated that the PIM inhibitor AZD1208 is effective at sensitizing Myc-CaP cells to radiation, we sought to determine if the sensitizing effects of AZD1208 are also manifested in CRPC in vivo. Myc-CaP tumors were allowed to grow in mice to an average size of 500mm. Castration of the mice resulted in tumor regression in all cases over six to nine days. Following a recovery period, the grafts regrew in the androgen-deprived mice. At this point, mice were treated with AZD1208 alone, radiation alone, or a combination of AZD1208 and radiation (Figure 6A). AZD1208 at 45mg/kg was given daily for three weeks. After five doses of AZD1208, three once-daily 2 Gy radiation treatments were delivered. As shown in Figure 6B, treatment with either AZD1208 or radiation alone slowed the growth of recurrent tumors, and combination treatment with AZD1208+radiation resulted in more sustained inhibition of tumor growth. The mean graft growth rates for each treatment group were determined for the last eight days of drug treatment, after completing radiation treatment. An exponential fit to each curve revealed the following growth rates: 11% per day for vehicle treatment, 8.4% per day for radiation treatment, 10% per day for AZD1208 treatment, and 5% per day for combined radiation and AZD1208 treatment (Figure 6C).
(Enlarge Image)
Figure 6.
AZD1208 and radiation treatment of castrate-resistant recurrent prostate cancer. A) Schematic of Myc-CaP allograft experiment and (B) growth curves of Myc-CaP grafts that were treated with androgen deprivation (surgical castration) and followed over time. Tumors regressed, then started regrowing, after which mice were treated daily with Pim kinase inhibitor AZD1208 at 45mg/kg or vehicle. Additionally, some mice received three once-daily treatments of 2 Gy of radiation (XRT). AZD1208-only and radiation-only treated tumors responded initially, but started growing back. AZD plus radiation-treated tumors show sustained inhibition of growth (n = 6–8 per group). Results shown as mean ± SD. *Student's t test P < .05. #Mann-Whitney rank sum test significance at < .05. All statistical tests were two-sided. C) Last 8 days of tumor volume measurements with daily growth percentage determined by the exponent coefficient for an exponential curve fit (R is the goodness of fit).
MYC and AR Expression in Regressed Myc-CaP Tumors Postcastration
The MYC transgene in Myc-CaP cells is driven by the androgen-regulated PB4 promoter. We therefore sought to determine the dynamic pattern of MYC and AR expression in Myc-CaP tumors before castration, in regressed tumors postcastration, and in recurrent tumors postcastration. In control uncastrated mice treated with vehicle or AZD1208, we observed robust nuclear MYC expression (Supplementary Figure 4A, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online). Residual regressed tumors postcastration showed a heterogeneous pattern of MYC, while recurrent tumors regardless of whether they also received AZD1208 or radiation treatment expressed nuclear MYC in a homogenous pattern (Supplementary Figure 4A http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1). AR expression was nuclear and homogenous in intact mice with or without AZD1208 treatment (Supplementary Figure 4B, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online). In regressed tumors, AR expression was heterogeneous and present in both the nuclei and cytoplasm. AR expression becomes more homogeneous in recurrent tumors postcastration with primarily cytoplasmic AR expression in AZD1208-treated tumors postcastration (Supplementary Figure 4B, http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1 available online). These results suggest the presence of tumor cells that retain MYC and AR expression in regressed tumors postcastration. These cells may be the origin of recurrent tumors.
Gene Expression Analysis in Cells Derived From AZD1208-Resistant Tumors
The fact that postcastration recurrent Myc-CaP tumors subjected to AZD1208 treatment were able to regrow after a lag period prompted us to examine possible mechanisms of this resistance. To this end, we established a panel of six cell lines, each from independent recurrent tumors, three from vehicle-treated (CxVehicle) and three from AZD1208-treated (CxAZD) surgically castrated animals. Analysis of in vitro growth rates showed no consistent differences between the cells derived from vehicle-treated and AZD1208-treated tumors (data not shown). We next used Affymetrix Mouse Gene 2.0 ST Arrays to profile gene expression changes in the cell lines. Using GSEA, we identified several pathways statistically significantly enriched in CxAZD cell lines compared with CxVehicle cells (Figure 7, A and B). Remarkably, the top enriched gene set is p53_DN.V2_UP, representing genes upregulated in HEK293 cells upon P53 knockdown by RNAi. These results suggest that CxAZD cells have lower p53 pathway activity, consistent with our RNA-seq profiling studies of Myc-CaP cells and tumors (Table 1). Another gene set enriched is JAK2_DN.V1_UP, consisting of genes upregulated in human erythroleukemia cells after knockdown of JAK2 by RNAi. PIM kinases have been linked to the Jak/Stat pathway. However, Jak/Stat pathway activity was unchanged in CxAZD cells as measured by p-Stat3 levels (Figure 7C). We confirmed reduced p53 protein levels in CxAZD cell lines by immunoblotting (Figure 7C), and intact/postcastration AZD1208-treated tumors expressed reduced levels of p53 protein by immunohistochemistry (Figure 7D). On the other hand, AZD1208-treated tumors exposed to radiation expressed high levels of p53. Thus, reduced p53 expression may represent a mechanism of resistance to AZD1208 by tumors that can be overcome by radiation treatment, which induces p53. This may provide one possible (but not the only) basis for the cooperativity we observed between AZD1208 treatment and radiation in inhibiting Myc-CaP prostate tumor growth. Indeed, treatment of the p53-null human PC3 prostate cancer cell line xenografts with AZD1208 and/or radiation showed that the combination treatment was still more effective in suppressing tumor growth than either single treatment (Supplementary Figure 5, A and B http://jnci.oxfordjournals.org/content/107/2/dju407/suppl/DC1).
(Enlarge Image)
Figure 7.
GSEA analysis of gene expression changes in AZD1208-resistant castration-resistant prostate cancer (PC) cells. A) The list of the top 10 gene sets enriched in tumor cell lines established from recurrent, castrated, AZD1208-treated mice (CxAZD) relative to cell lines established from castrated, recurrent, vehicle-treated recurrent tumors (CxVehicle). B) Enrichment analysis plots showing enrichment of genes upregulated in P53 knockdown cells (left panel) and JAK2 knockdown cells (right panel) in CxAZD tumor cell lines. C) Immunoblots showing reduced p53 protein levels in AZD1208-treated castrated Myc-CaP allograft cell lines, but unchanged p-Stat3 and Stat3 levels. D) Reduced p53 expression in AZD1208-treated Myc-CaP tumors by immunohistochemical staining both before and after castration (upper four panels), while tumors exposed to radiation treatment (XRT) or radiation plus AZD1208 (XRT/AZD) expressed high levels of p53 (bottom two panels). Scale bar = 0.05mm E) Gene expression profile of AZD1208-resistant recurrent tumor cells is associated with human prostate cancer recurrence. Cross-species Gene Set Enrichment Analysis analysis shows association of gene expression signature from AZD1208-treated recurrent tumors with biochemical recurrence in prostate cancer patients from.
Finally, we sought to determine if genes dysregulated in postcastration AZD1208-resistant recurrent tumors are associated with clinical outcome in human prostate cancer. Using cross-species GSEA analysis, we found that genes upregulated in AZD1208-resistant tumors are statistically significantly enriched in the prostate tumors of patients who experienced biochemical recurrence (Figure 7E). Conversely, genes downregulated in AZD1208-resistant tumors are statistically significantly enriched in patients without recurrence.