(2012) 31, 1794–1803 & 2012 Macmillan Publishers Limited All rights reserved 0950-9232/12 www.nature.com/onc ORIGINAL ARTICLE Pim1 is required to maintain tumorigenicity in -expressing cells

J Wang1, PD Anderson2, W Luo3, D Gius1,4,5, M Roh2 and SA Abdulkadir1,2

1Department of Cancer Biology, Vanderbilt University Medical Center, Nashville, TN, USA; 2Department of Pathology, Vanderbilt University Medical Center, Nashville, TN, USA; 3Department of Cell and Developmental Biology, Vanderbilt University Medical Center, Nashville, TN, USA; 4Department of Pediatrics, Vanderbilt University Medical Center, Nashville, TN, USA and 5Department of Radiation Oncology, Vanderbilt University Medical Center, Nashville, TN, USA

PIM1 kinase and MYC are commonly co-expressed in ment in mouse models (Verbeek et al.,1991;Wanget al., human prostate cancer and synergize to induce rapidly 2010). Whereas mice with homozygous deficiency in Myc progressing prostate cancer in mouse models. Deficiency of uniformly exhibit widespread developmental defects the Pim kinase is well tolerated in vivo, suggesting and embryonic lethality (Davis et al., 1993), mice with that PIM1 inhibition might offer an attractive therapeutic homozygous deficiency in all three Pim are viable, modality for prostate cancer, particularly for MYC- fertile and exhibit only minor phenotypic abnormalities expressing tumors. Here we examine the molecular (Mikkers et al., 2004). These observations show that consequences of Pim1 and MYC overexpression in the inhibition of PIM kinase activity is well tolerated in vivo, prostate as well as the effects of depleting Pim1 in prostate and suggest that drugs targeting PIM kinases might be carcinoma cells with high levels of MYC. Overexpression of better tolerated than those targeting MYC. Pim1 in the mouse prostate induces several pro-tumorigenic Using a prostate regeneration system, we have recently genetic programs including genes and Myc- shown that Pim1 potently interacts with MYC to regulated genes before the induction of any discernible accelerate prostate cancer progression, although Pim1 pathology. Pim1 depletion by RNA interference in mouse alone did not lead to significant pathology (Wang et al., and human prostate cancer cells decreased cellular pro- 2010). MYC/Pim1 expression in this model led to the liferation, survival, Erk signaling and tumorigenicity even development of neuroendocrine prostate cancer, appar- when MYC levels were not significantly altered. These ently by differentiation via an adenocarcinoma inter- results indicate that PIM1 may be necessary to maintain mediate. The mechanisms underlying tumorigenicity, and further support efforts aimed at between Pim1 and Myc in prostate tumorigenesis in vivo developing PIM1 inhibitors for prostate cancer therapy. have not been clearly elucidated. Studies have suggested Oncogene (2012) 31, 1794–1803; doi:10.1038/onc.2011.371; that Pim1 may enhance Myc tumorigenesis by increasing published online 22 August 2011 Myc protein stability or transcriptional activity (Chen et al., 2005; Zippo et al., 2007; Zhang et al., 2008; Kim Keywords: Pim1 kinase; MYC; expression; Erk; et al., 2010). We have reported that Pim1 overexpression tumorigenicity in the prostate in vivo produced minimal morphological alterationsby6weeks(Wanget al., 2010). These results prompted us to test whether we could discern molecular alterations induced by Pim1 overexpression in the absence Introduction of concurrent overexpression of Myc. In this report, we used profiling to investigate the molecular The PIM kinase family consists of three constitutively pathways modulated by expression of Pim1 and MYC active serine/threonine kinases, PIM1, PIM2 and PIM3, alone or together in the mouse prostate. We also knocked which have been implicated in diverse human malignan- down Pim1 expression using RNA interference to test cies including leukemias and lymphomas, as well whether Pim1 is necessary to maintain the malignancy of as pancreatic, hepatic, oral and prostate cancers MYC-expressing prostate cancer cells. Our results show (Shah et al., 2008; Magnuson et al., 2010). PIM1 kinase that Pim1 knock-down resulted in decreased proliferation, is an attractive molecular target for cancer pharma- cell survival, Erk signaling and tumorigenicity even when cotherapy because it enhances the transforming potential MYC levels were not significantly altered. of . For instance, Pim1 accelerates Myc- initiated lymphomagenesis and prostate cancer develop- Results Correspondence: Dr SA Abdulkadir, Department of Pathology, Vanderbilt University Medical Center, B33321A MCN, Nashville, Gene expression programs regulated by Pim1 and TN 37232, USA. E-mail: [email protected] c-MYC in prostate cells in vivo Received 22 February 2011; revised and accepted 19 July 2011; published To characterize the effects of Pim1 and MYC on online 22 August 2011 prostate tumorigenesis, we used a tissue recombination Pim1 in prostate cancer J Wang et al 1795 model in which lentiviral vectors overexpressing MYC, model of adipocyte differentiation of 3T3L1 cells Pim1, or MYC and Pim1 were used to infect normal induced by insulin, dexamethasone treatment (Burton mouse prostate cells. Cells were implanted under the et al., 2004). Many of the genes in this cluster are also renal capsule of SCID mice hosts, and regenerated related to the cell cycle as differentiation is associated prostate tissue grafts were harvested after 6 weeks. We with exit from the cell cycle. Another notable gene set have previously shown that at this time point, Pim1 identified by Gene Set Enrichment Analysis analysis grafts were morphologically indistinguishable from consists of genes upregulated in doxorubicin-resistant control grafts, consisting of benign prostatic glands gastric cancer cells. Therefore, the gene sets identified as (Wang et al., 2010). MYC grafts, on the other hand, enriched in Pim1, MYC or MYC/Pim1 grafts consist of contained multiple foci of high-grade PIN (HGPIN) genes involved in cell cycle progression, differentiation lesions, (Wang et al., 2010). However, MYC/Pim1 grafts and drug resistance. The differential expression had the most severe pathology, consisting of poorly of several selected genes identified by this analysis was differentiated cancer with evidence of neuroendocrine confirmed by quantitative real-time PCR (Supplemen- differentiation. These observations indicate that Pim1 tary Figure S1B). Although observing these gene may be dependent on collaborating mutations in genes expression changes in MYC- or MYC/Pim1-expressign such as in Myc before it can manifest its tumorigenic neoplastic prostate grafts is not surprising, it is effects. unexpected that grafts expressing Pim1 alone also show To gain insight into the oncogenic functions of Pim1 such gene expression changes, considering that Pim1 in prostate cells in vivo, we compared the gene grafts at 6 weeks show no discernible morphological expression profiles of these regenerated mouse prostate abnormalities (Wang et al., 2010). Thus, Pim1 may grafts overexpressing human MYC, mouse Pim1, both perturb the expression of cancer-relevant genes before Pim1 and MYC or control lentivirus (Wang et al., 2010) the appearance of any discernible histological abnorm- using Affymetrix microarray analysis. By employing alities. Significance Analysis of Microarrays (SAM) analysis, The enrichment of these gene set modules in the we were able to identify 4172 probes (corresponding to prostate-regenerated grafts is not due to the presence of 2770 unique genes) that were differentially expressed a common set of genes in all modules, as leading edge between control, Pim1, MYC or MYC/Pim1 grafts analysis showed little overlap among them (Figure 1b; (Supplementary Table S1). To identify subgroups of Supplementary Figure S2). The ‘leading edge’ in a gene genes with similar expression profiles, we performed set represents the individual genes that are primarily hierarchical clustering (Supplementary Figure S1) for responsible for the observed association. Examination the genes deemed significant in the microarray analysis. of the ‘leading edge’ could provide important insights The most extreme gene expression changes were into the biological character of the gene set. The observed in three MYC/Pim1 grafts and revealed several similarity between modules enriched in Pim1-expressing distinct clusters of genes with altered expression. Gene grafts to those in MYC-expressing grafts (Figures 1a expression differences among samples are a probably a and c) is consistent with reports in the literature reflection of the stages of tumorigenesis in each of the that Pim1 enhances MYC transcriptional activity (Chen grafts. For example, one MYC/Pim1 array clustered et al., 2005; Zippo et al., 2007; Zhang et al., 2008; Kim with the MYC arrays (Supplementary Figure S1), et al., 2010). Thus, our gene expression analysis supports suggesting that RNA isolated from this sample might the hypothesis that overexpression of Pim1 stimulates have come from tissue consisting of mostly HGPIN. the transcriptional activity of endogenous Myc present Although single analysis of differentially expressed in the cells, resulting in the activation of gene programs genes can provide insight into the underlying biological regulated by Myc. differences between samples, it can miss modest but coordinated changes in groups of genes (such as those belonging to the same cellular pathway) that are Establishment of MYC/Pim1 tumor cell lines functionally significant in the tissue or cell type being To further investigate the relationship between Myc and studied. Gene Set Enrichment Analysis is an approach Pim1, we tested whether Pim1 has a role in maintaining that identifies pathway-level changes using groups the tumorigenic potential of Myc. To examine this of genes or ‘gene sets’ that are significantly enriched or possibility, we created a cell line (MPT: MYC and Pim1 depleted—as a group—in a pair-wise comparison of overexpressing Tumor cells) from MYC/Pim1 prostate experimental vs control samples (Mootha et al., 2003; tissue recombinant grafts. As we have shown, the Subramanian et al., 2005). We performed Gene Set MYC/Pim1 grafts formed aggressive prostate tumors Enrichment Analysis comparing the control grafts to (Figure 2a; Wang et al., 2010). MPT cells expressed each of the Pim1, MYC and MYC/Pim1 groups (a total MYC and Pim1 protein (Figure 2b), yellow fluorescent three comparisons). Remarkably, all three groups protein as a marker indicating their derivation from a showed significant enrichment for several gene sets, MYC/Pim1 tumor (Figure 2a), as well as the neuroen- including cell cycle genes and Myc target genes (Figure 1; docrine cell marker neuron-specific enolase (Figure 2b), Supplementary Table S2). Other gene sets significantly consistent with neuroendocrine differentiation in the enriched in Pim1, MYC and MYC/Pim1 grafts original MYC/Pim1 tumors (Wang et al., 2010). To include genes involved in differentiation. For example, study the effects of Pim1 expression, these cells were IDX/TSA cluster 3 refers to a gene set upregulated in a subsequently infected with shRNAmir against mouse

Oncogene Pim1 in prostate cancer J Wang et al 1796

Figure 1 Gene expression profiling identifies pathways dysregulated in Pim1 and MYC-expressing prostate tissue recombinant grafts. (a) GSEA (Gene Set Enrichment Analysis) analysis was performed to identify ‘gene sets’ significantly enriched in Pim1, MYC and Pim1/MYC grafts relative to control prostate grafts. The false discovery rates (FDR) are indicated for each gene set. See also Supplementary Figure S1 and Supplementary Tables S1 and S2. (b) Limited overlap among leading edge genes across the top gene sets enriched in Pim1, MYC and Pim1/MYC prostate tissue recombinant grafts. The numbers indicate percent overlap of leading edge genes in each gene set. These results indicate that the five gene sets are distinct. (c) Overlap of leading edge genes in gene sets enriched by expression of Pim1, MYC or Pim1/MYC. The numbers indicate percent overlap of leading edge genes in each experimental group. These data indicate that Pim1, MYC and MYC/Pim1 regulate a similar set of genes within each gene set. Adip_Diff, Adipocyte differentiation; Dox_Resist_Gastric, Doxorubicin-resistant gastric carcinoma cells; IDX, Insulin/Dexamethasone/ Isobutylmethyl- xanthine; TR, tissue recombinant graft; TSA, Trichostatin A; Yu_MYC, MYC target genes. A full colour version of this figure is available at the Oncogene journal online.

Pim1 (named shPim1#1). To control for off-target in Pim2 or Pim3 (Supplementary Figure S4). Notably, effects, additional stable cell line was established Pim1 knock-down did not appreciably change the levels expressing short hairpin RNA (shRNA) targeting of MYC protein in MPT cells (Figure 2b). Pim1 (named shPim1#2). Western blot analysis showed We also established a second cell line, MPT2 from a that Pim1 protein levels were dramatically reduced by 4-week MYC/Pim1 graft (Figure 2c; Supplementary both Pim1 RNAs (Figure 2b). Furthermore, Pim1 Figure S3). Unlike the 6-week grafts reported above, knock-down did not result in a compensatory increase histological and immunohistochemical analysis revealed

Oncogene Pim1 in prostate cancer J Wang et al 1797

Figure 2 Knock-down of Pim1 expression in MYC/Pim1-overexpressing prostate tumor derived (MPT) cell lines. (a) Establishment of MYC/Pim1 Tumor (MPT) prostate cell line from a MYC/Pim1-expressing tumor graft. The right panel shows the image of established cell line from 6-week MYC/Pim1 overexpressing tumor. YFP (yellow fluorescent protein ) fluorescence can be seen under fluorescence microscope as the cells contain lentivirus construct bicistronically expressing YFP in addition to MYC or Pim1 (scale bar: 100 mm). H&E, hematoxylin-eosin. (b) Western blot analysis of parental MPT cells or MPT cells stably expressing shRNAmir against Pim1 (shPim1#1), non-silencing shRNAmir (shControl#1), shRNA against Pim1 (shPim1#2) or control (shControl#2). MPT cells express NSE (neuron- specific enolase) and MYC, the levels of which were not altered by Pim1 knock-down. Actin served as loading control. (c) MPT2 cells were derived from 4-week MYC/Pim1 overexpressing prostate grafts (see Supplementary Figure S2) and Pim1 expression stably knocked down using shPim1#2. MPT2 cells expressed AR in early passage, but not in late passage. NSE was also expressed in MPT2 cells. Tubulin served as loading control. (d) Growth curve were generated by counting shPim1#1 and shControl#1 MPT cells in duplicates. *Po0.05 (e)Cell proliferation was measured using the CellTiter 96 AQueous One Solution Cell Proliferation Assay (MTS, Promega). The absorbance is directly proportional to the number of living cells in culture. Results represent the mean±s.d. of quadruplicate wells after 3 days of incubation of 2000 cells. *Po0.05. (f) Analysis of by activated caspase-3 staining (red), with nuclei counterstained with DAPI (blue; original magnification: Â 400). (g) Quantification of apoptosis rate by counting active capase-3 positive cells. The results represent the mean±s.d. of two independent experiments. *Po0.05. A full colour version of this figure is available at the Oncogene journal online. that 4-week MYC/Pim1 grafts exhibited HGPIN with synaptophysin negative (Supplementary Figure S3). We features of microinvasion. These lesions were androgen stably knocked down Pim1 expression in MPT2 cells receptor (AR) positive and neuroendocrine cell marker using shPim1#2 (Figure 2c). We observed that MPT2

Oncogene Pim1 in prostate cancer J Wang et al 1798 cells expressed neuron-specific enolase and initially Pim1 knock-down impairs prostate tumor cell expressed AR (Figure 2c). However, AR expression tumorigenicity was lost with increasing passage in culture (Figure 2c). We used a focus-formation assay to test the effects of Further characterization of MPT and MPT2 cells Pim1 knock-down on prostate cancer cell tumorigenicity showed that Pim1 knock-down substantially reduced in vitro. The results showed that control MPT and cellular proliferation compared with cells expressing MPT2 cells lost contact inhibition and formed foci, control shRNA (Figures 2d and e). PIM1 has been which is typical of cancer cells. However, Pim1 knock- implicated in antiapoptotic functions via interactions down in both cell lines sharply reduced focus-forming with several molecules including BAD (Aho et al., 2004) ability (Figures 3a and b). To examine in vivo tumor- and ASK1 (Gu et al., 2009). Accordingly, Pim1 knock- igenicity, we injected either Pim1 knock-down down increased the rate of apoptosis as determined (shPim1#1) or control (shControl#1) MPT cells by activated caspase-3 staining in both MPT and MPT2 subcutaneously into athymic nude mice and monitored cells in low-serum conditions (Figures 2f and g). tumor growth over 12 weeks. Although no tumors

Figure 3 Pim1 expression is required to maintain tumorigenicity of MYC/Pim1 tumor cells. Stable knock-down of Pim1 by shRNA reduces colony formation in MPT (a) and MPT2 (b) cells. (c, d) Pim1 knock-down (shPim1#1) abrogates tumorigenicity of MPT cells when injected subcutaneously into nude mice. (e) Representative H&E images of graft sections. Control group showed high-grade tumor consistent with the original MYC/Pim1 tissue recombinant graft tumor whereas Pim1 knock-down group showed fat and blood vessel cells with absence of tumor cells. Scale bar: 100 mm. Insets: higher-magnification images. (f) Immunohistochemical staining showed that sections from control grafts were synaptophysin (SYP) positive and AR negative. Scale bar: 100 mm. Pim1 knock-down grafts could not be analyzed because no tumor cells were found by H&E staining. Insets: AR positive control. A full colour version of this figure is available at the Oncogene journal online.

Oncogene Pim1 in prostate cancer J Wang et al 1799 developed in the Pim1 knock-down group (N ¼ 10), initiate the development of invasive prostate cancer, 60% of mice in the control group (N ¼ 10) formed large continued Pim1 expression is required to maintain tumors (Figures 3c and d). In histology, the control prostate cancer cell tumorigenicity. grafts consisted of sheets of tumor cells expressing the neuroendocrine differentiation marker synaptophysin, Pim1 knock-down impairs ERK signaling pathway but not AR, recapitulating the features of original activation MYC/Pim1 tumor from which the cell line was We explored signaling pathways that may be dysregu- established (Figures 2a, and 3e and f). lated in Pim1 knock-down cells, including the PI3- We extended these findings beyond the two MPT cell kinase (phosphatidylinositol-3-kinase)/Akt and MAPK lines, by examining the effects of PIM1 knock-down in (mitogen-activated protein kinase)/Erk (extracellular DU145, a human prostate cancer cell line with appreci- signal-regulated kinase) signaling pathways. Although able endogenous PIM1 and MYC expression. To reduce we found no consistent changes in Akt signaling PIM1 expression in DU145, we used shPim1#2 (not shown), western blot analysis showed that ERK1/ that targets both human and mouse PIM1 mRNA. We 2 phosphorylation was consistently reduced by Pim1 verified that PIM1 was substantially knocked down in knock-down in MPT, MPT2 and DU145 cells compared DU145 cells (Figure 4a). Furthermore, unlike MPT with corresponding control cells (Figures 5a–d). We next cells, endogenous MYC levels were reduced by PIM1 examined whether Pim1 overexpression is sufficient to knock-down in DU145 cells (Figure 4a). Neuron- induce active ERK in prostate carcinoma cells. We specific enolase was also expressed by DU145 cells, found that stable overexpression of Pim1 in LNCaP and consistent with previous results (Leiblich et al., 2007), PC3 prostate cancer cells (Roh et al., 2003; Kim et al., and its expression was not altered by PIM1 knock-down 2010) did not alter ERK phosphorylation (Figures 5e (Figure 4a). PIM1 reduction impaired the proliferation and f). However, examination of Pim1-expressing tissue and survival of DU145 cells (Figures 4b and c). recombinant prostate grafts showed evidence of en- In addition, DU145 cell tumorigenicity as determined hanced Erk1/2 phosphorylation (Figure 5g). As these by growth in soft agar was reduced following PIM1 Pim1-expressing grafts showed no evidence of prostate depletion (Figures 4d and e). These results, taken neoplasia (Wang et al., 2010), the results indicate that in together with our previously published results on the prostate tissue in vivo, Pim1 overexpression is associated effects of Pim1 overexpression on prostate cell transfor- with enhanced Erk1/2 activity. Overall, these results mation (Wang et al., 2010), indicate that although indicate that PIM1 may be necessary for maintaining Pim1 overexpression alone may not be sufficient to ERK signaling in prostate cells.

Figure 4 Pim1 knock-down in DU145 cells reduces proliferation, survival and tumorigenicity (a) Western blot analyses for PIM1, MYC and NSE (neuron-specific enolase) in DU145 cells with stable PIM1 knock-down by shRNA. Tubulin served as a loading control. (b) Cell proliferation was measured by MTS assay. The results are mean±s.d. of quadruplicate wells after 3 days of incubation of 4000 cells. *Po0.05. (c) Quantitation of apoptosis rate by counting active capase-3 positive cells after DU145 cells were treated with serum-free medium overnight. The results are mean±s.d. of two independent experiments. *Po0.05. (d) Representative images of soft agar assay show that PIM1 knock-down in DU145 cells decreased the size and number of colonies (original magnification: Â 4). (e) Quantitation of colonies from soft agar assay of Pim1 knock-down and control DU145 cells. The results are mean±s.d. of triplicates. *Po0.05.

Oncogene Pim1 in prostate cancer J Wang et al 1800

Figure 5 Pim1 is required to maintain Erk signaling in prostate carcinoma cells. (a–c) Western blots showed that Pim1 knock-down in MPT and MPT2 cells by two different sets of shPim1 constructs decreased phosphorylation of Erk1/2 compared with control shRNA. (d) PIM1 knock-down in DU145 cells decreases ERK1/2 phosphorylation of Erk1/2 (e, f) Western blots showed that Pim1 overexpression did not increase ERK phosphorylation in LNCaP and PC3 cells compared with vector control cells (Neo). (g) Western blot analysis of 6-week tissue recombinant grafts generated from cells expressing control or Pim1 lentivirus. Note increased phosphorylation of Erk1/2 in Pim1 graft compared with control.

Discussion correlates its kinase activity (Qian et al., 2005). To understand the function of Pim1 on Myc/Pim1 coopera- In this report, we investigated the molecular changes tion in prostate tumorigenicity, we knocked down Pim1 induced by expression of Pim1 and MYC in the mouse expression in MYC/Pim1 mouse tumor-derived cell lines prostate in vivo as well as the functional consequences of and a human prostate cancer cell line DU145. We found depleting Pim1 in prostate tumor cells. We found that that, while Pim1 failed to initiate development of Pim1 expression engages multiple gene programs invasive prostate adenocarcinoma, depletion of Pim1 important for tumorigenesis even in the absence of any expression resulted in decreased growth rate, survival discernible morphological abnormalities. Several gene and tumorigenic potential, indicating that Pim1 is modules were disrupted by Pim1 overexpression, in- required for maintenance of the tumorigenic phenotype. cluding modules involving cell cycle genes, genes altered In MPT cells, Pim1 knock-down reversed their tumori- during differentiation, MYC targets and genes upregu- genic phenotype even when MYC level was not lated in doxorubicin-resistant cancer cells. Dysregula- decreased, which suggested that Pim1 and Myc synergism tion of MYC target genes in Pim1-expressing prostate in prostate cancer depend on other mechanisms besides grafts is consistent with published reports showing Pim1 enhancing Myc activity and stability. Unlike in DU145 enhances MYC protein stability and transcriptional cells, Pim1 knock-down in MPT and MPT2 cells did not activity (Chen et al., 2005; Zippo et al., 2007; Zhang decrease Myc levels perhaps due to the fact that these cells et al., 2008; Kim et al., 2010). It has been shown that express exogenous MYC. It is important to point out that hypoxia can induce Pim1, which contributes to solid MPT cells are derived from MYC/Pim1 neuroendocrine tumor formation (Chen et al., 2009a). The gene prostate cancer without AR expression, so it may not programs we have shown to be regulated by Pim1 may reflect the situation in most prostate tumors, which are also be dysregulated when Pim1 is induced in hypoxic AR positive. However, we found similar results in MPT2 tissue and thereby sensitize the tissue to tumorigenesis cells, which are derived from AR positive HGPIN/cancer due to expression of oncogenes such as MYC. lesions. In addition, data from multiple human prostate Although Pim1 is weakly oncogenic, it synergizes cancer cell lines consistently showed that depletion potently with Myc to induce invasive prostate cancer of PIM1 attenuates tumor cell proliferation and/or with features of neuroendocrine differentiation (Wang tumorigenicity. Our results indicate that PIM1 shRNA et al., 2010). The basis for this synergism between Pim1 impairs proliferation and tumorigenicity of DU145 cells, and Myc is not well understood. Pim1 is a constitutively consistent with other reports in 22rv1(Morishita et al., active serine/threonine kinase, so its protein level 2008) and PC3 cells (Hu et al., 2009).

Oncogene Pim1 in prostate cancer J Wang et al 1801 We observed reduced Erk phosphorylation in Pim1 by Vanderbilt Functional Genomics shared resource (Open knock-down cells. It has been shown that Pim1-depleted Biosystems, Huntsville, AL, USA). Lentiviral pLKO.1 shRNA and Pim1 inhibitor-treated bone marrow cells have targeting human, mouse, rat Pim1 (RHS3979–9631245) and impaired Erk phosphorylation (Grundler et al., 2009). empty vector were obtained from Open Biosystems. Lentiviral The Raf–MEK–ERK/MAPK pathway is subject to shRNAmir or shRNA vectors with D8.9 and VSVG (provided many levels of regulation. Pim1 may have common by Dr David Baltimore, Caltech, Pasadena, CA, USA) were transfected into 293FT cells (Invitrogen, Carlsbad, CA, USA) targets that affect the MAPK pathway. In Em-Myc mice, to obtain viral supernatants as described (Wang et al., 2010). Tpl2 can compensate for the absence of Pim1 and Pim2 MPT, MPT2 or DU145 cells were infected by virus in the (Berns et al., 1999). Tpl2 is also known as Map3k8, and presence of 8 mg/ml polybrene. In all, 10 or 5 mg/ml Puromycin has been reported to be involved in Mek/Erk1/2 was added to select stably transduced cells after 2 days of activation (Jager et al., 2010). It is still unclear whether infection. Pim1 directly or indirectly affects Erk phosphorylation. In the primary mitogen-regulated pathway, extracellular Transcriptional profiling analysis stimuli activate Raf-1, which phosphorylates and Total RNA was isolated with TRIzol (Invitrogen) and activates MAPK kinase (Mek), leading to the phos- RNAeasy kit from Qiagen (Germantown, MD, USA). RNA phorylation and activation of the extracellular signal quality was checked on an Agilent Bioanalyzer. All samples regulated kinases (Erk; Kolch 2000). Activated Erk/ used for microarray analysis have high quality score (RIN Mapk translocates to the nucleus and activates gene 47). RNA (1 mg) was reverse transcribed with T7–oligo(dT) expression by phosphorylation of a series of substrates primer and labeled with biotin using Affymetrix One Cycle including MYC. It remains to be determined whether Target Labeling kit (Santa Clara, CA, USA) following Pim1 affects RafÀ1, or Mek phosphorylation, manufacturer’s protocol. Three to four replicates of each or interacts with other proteins that are involved in this group were prepared, labeled, and hybridized to Affymetrix Mouse Gene 1.0 ST v1.r4 arrays and scanned on Affymetrix pathway. GeneChip scanner 3000. Data were collected using Affymetrix Several studies suggest that targeting Pim1 could GCOS software and pre-processed in Expression Console provide a promising strategy in anticancer therapy. (version 1.1; Affymetrix) using the RMA-Sketch method. Importantly, Pim1 depletion has only subtle effects in Normalized data were imported to Microsoft Excel for SAM normal cells. Pim1-deficient mice are ostensibly normal, (Tusher et al., 2001). Multiclass SAM was performed healthy and fertile (Laird et al., 1993). Pim1, Pim2 and (delta ¼ 0.43, false discovery rates o0.0152) without array Pim3 compound knockout mice are viable and fertile, centering. Genes deemed significant in the multiclass SAM but show a reduction in body size, suggesting Pim were ordered based on their similarity in expression patterns kinases might act as sensitizers for growth factor with hierarchical clustering in R. Distance was calculated using signaling pathways (Mikkers et al., 2004). Pim kinase the Euclidean model, and clusters were generated using the complete linkage method. Gene Set Enrichment Analysis inhibition using SGI-1776 in prostate cancer cells or v2.0.6 (Subramanian et al., 2005) was used to test for CLL cells (B-cell chronic lymphocytic leukemia) led enrichment in collections C2–C5 of the Molecular Signatures to a concentration-dependent induction of apoptosis Database (MSigDB version 2.5). Comparisons were made in (Mumenthaler et al., 2009; Chen et al., 2009b). A PIM1- three ways: control vs Pim1 overexpressing, control vs MYC specific monoclonal antibody has also shown efficacy in overexpressing and control vs MYC/Pim1 overexpressing. inhibiting the tumorigenicity of PIM1-expressing tumors in mouse models (Hu et al., 2009). In conjunc- Western blot analysis tion with these studies, our results further support that Cells were washed with PBS, RIPA buffer with protease Pim1 represents an efficient target for prostate cancer inhibitors was added, and cells were harvested with cell therapy. scraper, briefly sonicated and spun down. For analysis of tissues, prostate grafts from control and Pim1-expressing tissue recombinants grown for 6 weeks (Wang et al., 2010) were used Materials and methods to make lysates. Lysate was run on SDS–polyacrylamide gel electrophoresis, transferred to PDVF membranes. Membranes Cell lines and constructs were blotted with the following antibodies: c–MYC, Pim1, AR DU145 human prostate cancer cells (ATCC) were cultured in and Actin (Santa Cruz Biotechnology, Santa Cruz, CA, USA); RPMI medium with 10% fetal bovine serum (FBS). MPT p44/42 MAPK (Erk1/2), phospho-p44/42 MAPK (Thr202/ and MPT2 mouse prostate cancer cells were established Tyr204; Cell Signaling, Danvers, MA, USA); neuron-specific from MYC/Pim1 prostate tissue recombinant tumors. The enolase (Neomarker); b-actin (Santa Cruz Biotechnology); generation of MYC/Pim1 graft tumors has been described b-tubulin (Sigma, St Louis, MO, USA). (Wang et al., 2010). Briefly, MYC/Pim1 tumors were minced and digested by collagenase, then plated on collagen coat plate Proliferation assay with DMEM/F12/10% FBS medium. After reaching conflu- MPT Cells were seeded at 20 000 cells per well in a six-well ence, the cells were split into regular dish. plate. Viable cell number was determined by hemacytometer counts of trypan blue, excluding cells for 4 days. For MTS Lentivirus preparation, transduction and stable (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2- knock-down of Pim1 (4-sulfophenyl)-2H-tetrazolium) assay, 2000 MPTs cells or Lentiviral pGIPZ shRNAmir against mouse Pim1 4000 DU145 cells were plated in quadruplicate per well in (V2LMM_46214) and the sequence-scrambled, non-silencing- 96-well plates. Three days later, CellTiter 96 Aqueous One GIPZ lentiviral shRNAmir control (RHS4346), were provided Solution (Promega, Madison, WI, USA) was added to each

Oncogene Pim1 in prostate cancer J Wang et al 1802 well and results were read at 490 nm in a plate reader according both the flanks of 8-week-old male athymic nude mice (BALB/ to the manufacturer’s instruction. c strain; Harlan Sprague Dawley, Indianapolis, IN, USA). Grafts were measured weekly as they were visible. All mice Active caspase-3 staining were killed by 12 weeks after injection. Animal care and Cells grown were plated on glass coverslips. Cells were fixed in experiments were carried out according to the protocols 4% paraformaldehyde for 10 min followed by permeabiliza- approved by the Institutional Animal Care and Use Commit- tion in 1% Triton X-100 for 10 min. After washing in PBS, tees at Vanderbilt University. Grafts were paraffin-embedded, cells were blocked in 10% goat serum, stained with activated sectioned, stained with H&E and analyzed by light micro- caspase-3 (Cell signal) at 1:500. Coverslips were mounted on scopy. A tumor was defined as a palpable mass that contained slides using Vectashield mounting medium (Vector Labora- carcinoma cells upon histological examination. Immunostain- tories, Burlingame, CA, USA). An average of 1000 cells were ing was performed as described (Abdulkadir et al., 2000, 2001) counted per coverslip. using anti-synaptophysin (BD Biosciences, Franklin Lakes, NJ, USA), anti-AR (Santa Cruz Biotechnology). Colony formation assay In all, 5000 of MPT cells were plated in 10 cm dishes. Triplicate experiments were performed for each cell line. The medium Conflict of interest was changed in every 3–4 days. After 10 days, the cells were fixed and stained with crystal violet. The authors declare no conflict of interest. The funders had no role in study design, data collection and analysis, decision Soft agar assay to publish, or preparation of the manuscript. A 2 ml of lower layer of 0.6% agar in RPMI-10% FBS was placed into each well of six-well plate. After agar solidified, 2 ml of 0.3% top agarose in RPMI-10% FBS containing 10 000 cells was added to each well. The cells were fed in every Acknowledgements 3–4 days with RPMI-10% FBS. The plates were incubated at 37 1C, in 5% CO2 incubator for 2 weeks. The colonies those We thank Drs Robert Matusik, Simon Hayward, Vito are larger than 100 mm were counted. Quaranta, Xiuping Yu and Fritz Parl for helpful discussions. Micrographs were taken from Cell Imaging Core at Vanderbilt In vivo tumorigenicity assay University Medical Center. This work was supported by A total of 1 Â 105 control MPT cells or Pim1 knock-down cells Grant RO1CA123484 from the NCI to SAA. while DG is were mixed with 15 ml of Matrigel (Becton Dickinson Labware, supported by R01CA152601 from the NCI and BC093803 Franklin Lakes, NJ, USA) and injected subcutaneously in from the DOD.

References

Abdulkadir SA, Carbone JM, Naughton CK, Humphrey PA, Grundler R, Brault L, Gasser C, Bullock AN, Dechow T, Woetzel S Catalona WJ, Milbrandt J. (2001). Frequent and early loss of et al. (2009). Dissection of PIM serine/threonine kinases in the EGR1 corepressor NAB2 in human prostate carcinoma. FLT3-ITD-induced leukemogenesis reveals PIM1 as regulator of Hum Pathol 32: 935–939. CXCL12-CXCR4-mediated homing and migration. J Exp Med 206: Abdulkadir SA, Carvalhal GF, Kaleem Z, Kisiel W, Humphrey PA, 1957–1970. Catalona WJ et al. (2000). Tissue factor expression and angiogenesis Gu JJ, Wang Z, Reeves R, Magnuson NS. (2009). PIM1 phosphor- in human prostate carcinoma. Hum Pathol 31: 443–447. ylates and negatively regulates ASK1-mediated apoptosis. Oncogene Aho TL, Sandholm J, Peltola KJ, Mankonen HP, Lilly M, 28: 4261–4271. Koskinen PJ. (2004). Pim-1 kinase promotes inactivation of the Hu XF, Li J, Vandervalk S, Wang Z, Magnuson NS, Xing PX. (2009). pro-apoptotic Bad protein by phosphorylating it on the Ser112 PIM-1-specific mAb suppresses human and mouse tumor growth gatekeeper site. FEBS Lett 571: 43–49. by decreasing PIM-1 levels, reducing Akt phosphorylation, and Berns A, Mikkers H, Krimpenfort P, Allen J, Scheijen B, Jonkers J. activating apoptosis. J Clin Invest 119: 362–375. (1999). Identification and characterization of collaborating Jager J, Gremeaux T, Gonzalez T, Bonnafous S, Debard C, Laville M oncogenes in compound mutant mice. Cancer Res 59: et al. (2010). Tpl2 kinase is upregulated in adipose tissue in 1773s–1777s. obesity and may mediate -1beta and tumor necrosis Burton GR, Nagarajan R, Peterson CA, McGehee Jr RE. (2004). factor-{alpha} effects on extracellular signal-regulated kinase Microarray analysis of differentiation-specific gene expression activation and lipolysis. Diabetes 59: 61–70. during 3T3-L1 adipogenesis. Gene 329: 167–185. Kim J, Roh M, Abdulkadir SA. (2010). Pim1 promotes human Chen J, Kobayashi M, Darmanin S, Qiao Y, Gully C, Zhao R et al. prostate cancer cell tumorigenicity and c-MYC transcriptional (2009a). Hypoxia-mediated up-regulation of Pim-1 contributes to activity. BMC Cancer 10: 248. solid tumor formation. Am J Pathol 175: 400–411. Kolch W. (2000). Meaningful relationships: the regulation of the Chen LS, Redkar S, Bearss D, Wierda WG, Gandhi V. (2009b). Ras/Raf/MEK/ERK pathway by protein interactions. Biochem J Pim kinase inhibitor, SGI-1776, induces apoptosis in chronic 351(Part 2): 289–305. lymphocytic leukemia cells. Blood 114: 4150–4157. Laird PW, van der Lugt NM, Clarke A, Domen J, Linders K, Chen WW, Chan DC, Donald C, Lilly MB, Kraft AS. (2005). McWhir J et al. (1993). In vivo analysis of Pim-1 deficiency. Nucleic Pim family kinases enhance tumor growth of prostate cancer cells. Acids Res 21: 4750–4755. Mol Cancer Res 3: 443–451. Leiblich A, Cross SS, Catto JW, Pesce G, Hamdy FC, Rehman I. Davis AC, Wims M, Spotts GD, Hann SR, Bradley A. (1993). A null (2007). Human prostate cancer cells express neuroendocrine cell c-myc mutation causes lethality before 10.5 days of gestation in markers PGP 9.5 and chromogranin A. Prostate 67: 1761–1769. homozygotes and reduced fertility in heterozygous female mice. Magnuson NS, Wang Z, Ding G, Reeves R. (2010). Why target PIM1 Genes Dev 7: 671–682. for cancer diagnosis and treatment? Future Oncol 6: 1461–1478.

Oncogene Pim1 in prostate cancer J Wang et al 1803 Mikkers H, Nawijn M, Allen J, Brouwers C, Verhoeven E, Jonkers J Shah N, Pang B, Yeoh KG, Thorn S, Chen CS, Lilly MB et al. (2008). et al. (2004). Mice deficient for all PIM kinases display reduced Potential roles for the PIM1 kinase in human cancer—a molecular body size and impaired responses to hematopoietic growth factors. and therapeutic appraisal. Eur J Cancer 44: 2144–2151. Mol Cell Biol 24: 6104–6115. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Mootha VK, Lindgren CM, Eriksson KF, Subramanian A, Sihag S, Gillette MA et al. (2005). Gene set enrichment analysis: a Lehar J et al. (2003). PGC-1alpha-responsive genes involved knowledge-based approach for interpreting genome-wide expression in oxidative phosphorylation are coordinately downregulated in profiles. Proc Natl Acad Sci USA 102: 15545–15550. human diabetes. Nat Genet 34: 267–273. Tusher VG, Tibshirani R, Chu G. (2001). Significance analysis of Morishita D, Katayama R, Sekimizu K, Tsuruo T, Fujita N. (2008). microarrays applied to the ionizing radiation response. Proc Natl Pim kinases promote cell cycle progression by phosphorylating Acad Sci USA 98: 5116–5121. and down-regulating p27Kip1 at the transcriptional and posttran- Verbeek S, van Lohuizen M, van der Valk M, Domen J, Kraal G, scriptional levels. Cancer Res 68: 5076–5085. Berns A. (1991). Mice bearing the E mu-myc and E mu-pim-1 Mumenthaler SM, Ng PY, Hodge A, Bearss D, Berk G, Kanekal S transgenes develop pre-B-cell leukemia prenatally. Mol Cell Biol 11: et al. (2009). Pharmacologic inhibition of Pim kinases alters prostate 1176–1179. cancer cell growth and resensitizes chemoresistant cells to taxanes. Wang J, Kim J, Roh M, Franco OE, Hayward SW, Wills ML et al. Mol Cancer Ther 8: 2882–2893. (2010). Pim1 kinase synergizes with c-MYC to induce advanced Qian KC, Wang L, Hickey ER, Studts J, Barringer K, Peng C et al. prostate carcinoma. Oncogene 29: 2477–2487. (2005). Structural basis of constitutive activity and a unique Zhang Y, Wang Z, Li X, Magnuson NS. (2008). Pim kinase-dependent nucleotide binding mode of human Pim-1 kinase. JBiolChem280: inhibition of c-Myc degradation. Oncogene 27: 4809–4819. 6130–6137. Zippo A, De Robertis A, Serafini R, Oliviero S. (2007). PIM1- Roh M, Gary B, Song C, Said-Al-Naief N, Tousson A, Kraft A et al. dependent phosphorylation of histone H3 at serine 10 is required for (2003). Overexpression of the oncogenic kinase Pim-1 leads to MYC-dependent transcriptional activation and oncogenic transfor- genomic instability. Cancer Res 63: 8079–8084. mation. Nat Cell Biol 9: 932–944.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene