Oncogene (2014) 33, 1385–1394 & 2014 Macmillan Publishers Limited All rights reserved 0950-9232/14 www.nature.com/onc

ORIGINAL ARTICLE p120 catenin is a key effector of a Ras-PKCe oncogenic signaling axis

SG Dann1, J Golas1, M Miranda1, C Shi1,JWu2, G Jin1, E Rosfjord1, E Upeslacis1 and A Klippel1

Within the family of protein kinase C (PKC) molecules, the novel isoform PRKCE (PKCe) acts as a bona fide oncogene in in vitro and in vivo models of tumorigenesis. Previous studies have reported expression of PKCe in breast, prostate and lung tumors above that of normal adjacent tissue. Data from the cancer genome atlas suggest increased copy number of PRKCE in triple negative breast cancer (TNBC). We find that overexpression of PKCe in a non-tumorigenic breast epithelial cell line is sufficient to overcome contact inhibition and results in the formation of cellular foci. Correspondingly, inhibition of PKCe in a TNBC cell model results in growth defects in two-dimensional (2D) and three-dimensional (3D) culture conditions and orthotopic xenografts. Using stable isotope labeling of amino acids in a cell culture phosphoproteomic approach, we find that CTNND1/p120ctn phosphorylation at serine 268 (P-S268) occurs in a strictly PKCe-dependent manner, and that loss of PKCe signaling in TNBC cells leads to reversal of mesenchymal morphology and signaling. In a model of Ras activation, inhibition of PKCe is sufficient to block mesenchymal cell morphology. Finally, treatment with a PKCe ATP mimetic inhibitor, PF-5263555, recapitulates genetic loss of function experiments impairing p120ctn phosphorylation as well as compromising TNBC cell growth in vitro and in vivo. We demonstrate PKCe as a tractable therapeutic target for TNBC, where p120ctn phosphorylation may serve as a readout for monitoring patient response.

Oncogene (2014) 33, 1385–1394; doi:10.1038/onc.2013.91; published online 1 April 2013 Keywords: protein kinase C; epithelial to mesenchymal transition; triple negative breast cancer; p120 catenin; drug discovery

INTRODUCTION Here, we report that PKCe expression in normal mammary ducts Protein kinase C (PKC) molecules belong to the AGC kinase is confined to myoepithelial cells, and this compartmentalization superfamily and are divided into one of three classes (conven- is lost in breast cancer tissue. Exogenous expression of PKCe is tional, novel or atypical) determined by dependence on the sufficient to overcome contact inhibition in the non-tumorigenic second messengers Ca2 þ and diacylgycerol for activation.1 breast epithelial cell line MCF10A. We also show that PKCe activity Although several PKC isoforms have been implicated in cancer, in triple negative breast cancer (TNBC) cells is tightly correlated only the novel isoform e is causative for tumor formation in in vitro with phosphorylation of p120ctn on S268. Furthermore, activation and in vivo cancer models.2 Expression of the novel isoform PRKCE of oncogenic Ras induces a cell morphology and signaling (PKCe) is elevated in breast, prostate and lung tumors compared response that is dependent on PKCe activity. We present data with normal adjacent tissue.3 Pioneering studies in immortalized validating PKCe as a small molecule inhibitor target for oncology, murine cells showed that overexpression of PKCe is sufficient to and propose that high p120ctn P-S268 may reflect PKCe activity in promote growth in soft agar and tumor formation in nude mouse cancer cells. xenografts.4,5 Further, transgenic overexpression of PKCe in the mouse prostate leads to invasive prostatic intraepithelial neo- plasia, while knockout of even a single prkce allele is sufficient RESULTS to impair tumorigenesis in a genetically engineered mouse model PKCe expression drives focal growth in non-tumorigenic breast of prostate cancer.6,7 epithelial cells Recent data suggest that PKCe is indispensable for wound In order to evaluate PKCe expression by immunohistochemistry closure, cytokinesis and migration; all of which point toward (IHC), we qualified a rabbit monoclonal antibody against normal involvement in epithelial to mesenchymal transition (EMT).8,9 PKCe tissues from prkce À / À and À / þ mice as well as TNBC cells has also been implicated in several steps required for the EMT harboring a doxycycline (Dox)-inducible short hairpin RNA (shRNA) migratory phenotype.10,11 EMT of epithelial cancers is linked to a targeting PKCe (Supplementary Figure 1a and b). Satisfied with the broad transcriptional network induced by a set of embryonic fact that immunoreactivity was dependent on PKCe expression, transcription factors.12 The activity of this transcriptional network we surveyed normal and malignant mammary tissue samples by is generally scored by loss of epithelial markers, such as IHC. Normal breast tissue stains positive for PKCe in the E-cadherin, and gain of mesenchymal-specific markers, such as vasculature and myoepithelial cell layer surrounding epithelial N-cadherin and vimentin.13 Although the transcriptional ducts (Figure 1a and Supplementary Figure 1c). By comparison, network of EMT has been extensively studied, post-translational breast tumor tissue samples present with intense stromal staining mechanisms that mediate the effects on cell morphology are not for PKCe and with tumor epithelia-specific PKCe expression in a well understood.14 subset of cases (Figure 1a and Supplementary Figure 1c).

1Pfizer Oncology Research Unit, Pearl River, NY, USA and 2Pfizer Structural Biology and Biophysics, Groton, CT, USA. Correspondence: Dr SG Dann, Pfizer Oncology Research Unit, 401 N Middletown road, Pearl River, NY 10965, USA. E-mail: stephen.dann@pfizer.com Received 16 August 2012; revised 24 January 2013; accepted 4 February 2013; published online 1 April 2013 Oncogenic Ras-PKCe signaling requires p120ctn SG Dann et al 1386

Figure 1. PKCe overexpression is associated with altered morphology of breast epithelia and disruption of tissue organization. (a) Representative images taken at  20 magnification of normal breast (upper panels) and cancerous breast (lower panels) matched tissue stained for PKCe and p120ctn expression. (b) Analysis of the cancer genome atlas data for invasive breast cancer comparing PRKCE copy number in TNBC versus all other biomarker status. Oncomine (Compendia Bioscience, Ann Arbor, MI, USA) was used for analysis and visualization. (c) Relative PKCe expression levels in a panel of breast cancer cell lines. Protein extracts of comparable cell numbers from the indicated tumor lines were analyzed for PKCe expression and activation state by immunoblotting. (d) Ten days Dox-inducible expression of PKCe in MCF10A cells results in foci formation photographed at  10 magnification. Western blots showing relative PKCe expression are shown next to each panel. Quantitation was performed by bright-field phase contrast microscopy as described in the Materials and methods section.

Consequently, analysis of the TNBC subgroup within the invasive (Figure 1d and Supplementary Figure 2). Focal growth was breast cancer genome atlas finds PRKCE in the top 2% for all copy quantified using phase contrast bright-field microscopy as per the number gains (Figure 1b). In agreement with this data set, we find Materials and methods (Supplementary Figure 2 and Figure 1d). that immortalized but non-tumorigenic MCF10A breast epithelial cells express low levels of PKCe compared with a panel of cell lines derived from breast malignancies. Moreover, the TNBC cell lines PKCe is required for the growth of triple negative MDA-MB-231 MDA-MB-231 and BT549 express constitutively phosphorylated, or breast cancer cells in vitro and in vivo primed, PKCe, indicating hyperactivation of the kinase in this To further explore the role of PKCe in a disease relevant model, we context (Figure 1c). engineered MDA-MB-231 cells to stably express Dox-inducible To test the hypothesis that PKCe may drive tumorigenesis in shRNAs targeting PKCe. MDA-MB-231 cells are derived from a breast epithelial cells, we engineered MCF10A cells to express TNBC with activating mutations in KRAS and BRAF. Induction of wild-type PKCe (WTe), the kinase-defective mutant (K437R) or an PKCe knockdown for an extended period (10 days) resulted in empty vector control upon exposure to Dox. Dox-mediated reduced cell density over time, while the uninduced cells and induction of WTe and K437R mutant expression in each stable those harboring the non-targeting (NT)-shRNA displayed no line is shown in Figure 1d. Plated in the absence of Dox, each obvious growth or survival defect (Supplementary Figure 3). The stable MCF10A variant reached confluency and became contact PKCe loss of function (LOF) phenotype in two-dimensional (2D) inhibited (Figure 1d) as did the empty vector control and K437R in growth conditions prompted us to explore more stringent three- the presence of Dox. Upon induction of WTe, however, cells dimensional (3D) culture conditions by growing the cells overcame contact inhibition and cellular foci developed through- embedded in an extracellular matrix (ECM). In this environment out the plating surface, a hallmark of transformation in culture MDA-MB-231 cells form invasive structures,15 and expression of

Oncogene (2014) 1385 – 1394 & 2014 Macmillan Publishers Limited Oncogenic Ras-PKCe signaling requires p120ctn SG Dann et al 1387

Figure 2. PKCe activity can be monitored through p120ctn S268 phosphorylation and is required for MDA-MB-231 malignant cell growth in vitro and in vivo.(a) Stably transduced MDA-MB-231 cells with Dox-inducible expression of two independent PKCe or non-targeting control (NT) shRNAs were analyzed for growth in ECM. Cells were grown in the absence or presence of Dox for 7 days and photographed at  10 magnification. Black scale bars on micrographs represent 100 mM. The corresponding knockdown of parallel samples and the effect on p120ctn P-S268 is shown in adjacent western blots. (b) MDA-MB-231 cells stably expressing the indicated PKCe variants or b-Gal control were embedded into ECM and photographed 72 h later at  10 magnification. Black scale bars on micrographs represent 100 mM.(c) Expression of PKCe shRNA-A or NT-shRNA was induced in mice harboring MDA-MB-231 xenotransplants of the cells shown in (a) by feeding Dox supplemented chow; a parallel cohort of xenotransplanted animals received regular chow. Each cohort consisted of 15 mice evaluated for tumor presence weekly from 14 to 56 days. The graph shows the percentage of mice with measurable tumors for each cohort over the course of 6 weeks. The inset indicates the expression level of PKCe in parallel cell samples that were treated with or without 250 ng/ml Dox in culture. (d) Stably transduced MDA-MB-231 cells expressing NT or PKCe shRNA-A were treated with Dox for 10 days to ensure phenotypic changes comparable to those observed in Supplementary Figure 3b. Parallel lysates to those analyzed by LC MS/MS (Supplementary Table 1) were evaluated by western blotting as indicated. (e) Effect of b-Gal control, PKCe WT and PKCe kinase-defective (K437R) overexpression in MDA-MB- 231 on p120ctn P-S268 as measured by western Blot. (f) Dox-inducible expression of WT PKCe, but not K437R in MCF10A cells results in elevated p120ctn phosphorylation of S268 as measured by western blot. (g) Immunocytochemical detection of p120ctn total protein and P-S268 in MCF10A cells with and without induction of WT PKCe replicates western data in (f). (h) Recombinant p120ctn immunopurified from HEK293 cells was incubated in the absence ( À ) and presence ( þ ) of recombinant PKCe isolated from insect cells in an in-vitro kinase assay conditions for 30 min with the reaction products assayed for P-S268 by western blot (Lane L is HEK293 cellular lysate) All experiments were confirmed in triplicate, representative data is shown. shRNAs targeting PKCe attenuated invasive growth in ECM Having established that MDA-MB-231 cell growth in vitro is (Figure 2a). In agreement with LOF studies described above, dependent on PKCe by gain of function and LOF studies, we next MDA-MB-231 cells engineered to overexpress the kinase-defective sought to test the effects of targeting PKCe in vivo by producing K437R mutant showed impaired growth in ECM compared with orthotopic xenograft tumors of the shRNA-engineered MDA-MB- the b-galactosidase-expressing control, indicating a dominant 231 cells. The PKCe-targeting hairpin A resulted in 50% tumor negative effect in this context (Figures 2b and e). On the other attrition of MDA-MB-231 cells in vivo in mice fed chow hand, overexpression of the WT form accelerated formation of supplemented with Dox compared with both the NT-shRNA and invasive structures (Figures 2b and e). the same cells implanted in mice on normal chow. Relative to the

& 2014 Macmillan Publishers Limited Oncogene (2014) 1385 – 1394 Oncogenic Ras-PKCe signaling requires p120ctn SG Dann et al 1388

Figure 3. PKCe is required for Ras-induced morphology and p120ctn S268 hyperphosphorylation. (a)PKCe knockdown in MDA-MB-231 through antisense (AS) oligonucleotide were evaluated for E-cadherin expression by western blot. (b) E-cadherin/p120ctn complexes were co-immunoprecipitated from MCF10A cells with and without induction of PKCe expression by doxycyline (Dox) and assayed by western blot as indicated. Corresponding western blot analysis of cell lysates are shown to the right. (c)PKCe knockdown was induced using two 12 independent AS molecules (PKCe-AS A and B) or mismatch control (NT-AS) oligonucleotides 24 h before induction of ER-RasV with 100 nM 4-OHT for 72 h. Cell morphology was documented by photography at  10 magnification. Black scale bars on micrographs represent 100 mM. (d)PKCe knockdown was induced as described above and cell-signaling responses were documented by western blot. Quantitation on the right is relative P-S268 levels with NT-AS and PKCe-AS A normalized to vimentin. (e) Induction of ER-RasV12 signaling in response to increasing 4-OHT concentrations. Extracts of engineered HeLa cells treated with 0, 10, 25, 50 and 100 nM 4-OHT were analyzed by western blotting using 12 the indicated antibodies. Quantitation on the right is relative PKCe levels after induction of ER-RasV with 100 nM 4-OHT for 72 h normalized to vimentin.

NT shRNA, the PKCe-targeting hairpin A also resulted in MDA-MB- Phosphosignal reduction of a-catenin, b-catenin, d-catenin, 231 xenograft attrition (23%) even without Dox-administration OB-cadherin and Desmoplakin under PKCe knockdown conditions albeit to a lower extent than with Dox (Figure 2c). We hypothesize suggested that PKCe has an especially important role in regulating that this is due to leakiness of inducible shRNA expression adherens junction. The phosphorylated d-catenin (herein referred resulting in B50% lower PKCe levels in the absence of Dox as to as p120ctn) peptide identified in the MS analysis corresponds reflected by western blot analysis (Figure 2c inset). to S268. Knockdown of PKCe by two independent shRNAs (A and B) results in a reduction in the phosphorylation of S268 as compared with the non-silencing control (Figure 2a). p120ctn phosphorylation at S268 in TNBC is tightly linked to PKCe To further validate the effect of PKCe on p120ctn P-S268, we activity evaluated the MDA-MB-231 cells stably overexpressing WTe and Although the genetic gain of function and LOF studies in MCF10A K437R, and found P-S268 increased in only the WTe-expressing and MDA-MB-231 support an oncogenic role for PKCe, it is difficult cells (Figure 2e). Induction of WTe in MCF10A cells, but not the to attribute phenotypical alterations to specific pathway modula- K437R or vector control, also results in elevated P-S268 p120ctn tion without a discrete signaling readout. Therefore, we used (Figure 2f). Likewise, P-S268 p120ctn induction in the MCF10A MDA-MB-231 cells stably expressing inducible PKCe or NT shRNAs model could be visualized by immunocytochemistry upon to perform a SILAC (stable isotope labeling by amino acids in cell expression of PKCe (Figure 2g). To explore a possible direct culture) phosphoproteomic screen and identify possible biomar- relationship between PKCe and p120ctn, we performed standard kers of PKCe activity (Figure 2d). Liquid chromatography–tandem in-vitro protein kinase assays using full-length p120ctn immuno- mass spectrometry (LC–MS/MS) was performed on SILAC-labeled purified from mammalian cells as substrate. Indeed, incubation of lysates and phospho-enriched peptides with changes greater than full-length p120ctn with purified PKCe in vitro resulted in direct two-fold in PKCe knockdown are listed in Supplementary Table 1. phosphorylation of p120ctn at S268 (Figure 2h). We were unable

Oncogene (2014) 1385 – 1394 & 2014 Macmillan Publishers Limited Oncogenic Ras-PKCe signaling requires p120ctn SG Dann et al 1389 to detect P-S268 signal by IHC analysis of the tissue microarray, version of the enzyme primes the hydrophobic motif in the however, IHC of total p120ctn in normal and malignant breast catalytic domain for phosphorylation by its upstream-activating tissues was successful (Figure 1a). In normal tissue, p120ctn kinase. Applying this approach, PF-5263555 exhibited an IC50 of expression is segregated from PKCe and confined to the B250 nM in cells as reflected by its ability to induce the membrane of epithelial cells of the mammary duct. In the tumor phosphorylation of S729 in the K437R PKCe (Figure 4c). As predic- cases, however, we find that p120 staining is reduced at the ted by Cameron et al.18 overexpressed WT PKCe is constitutively epithelial membrane with granular intracellular staining accom- phosphorylated at this site and no further phosphosignal is panying PKCe expression (Figure 1a). observed upon addition of PF-5263555 (Figure 4c). As p120ctn phosphorylation state is tightly regulated by PKCe The in vitro IC50 results for a spectrum of PKC enzymes have been activity, we sought to determine whether p120ctn functions as a published previously, highlighting the selectivity of PF-5263555 for crucial mediator of PKCe-driven growth effects. MDA-MB-231 cells PKCe and PKCy.17 As PKCy cannot be detected by western blotting were stably engineered to express inducible p120ctn shRNA in HeLa ER-RasV12 or MDA-MB-231 cells, the selectivity for PKCy and molecules (Supplementary Figure 2b). Although no defect in e over PKCd and Z indicates that PKCe is the major, if not sole, growth was observed in 2D cell culture conditions, induction of target for PF-5263555 in these cellular models. p120ctn-specific shRNA molecules A and B showed distinct inhibitory effects on MDA-MB-231 invasive behavior and growth PF-5263555 normalizes Ras-driven morphology, PKCe-driven focal when embedded in ECM (Supplementary Figure 2b), thereby growth and PKCe-p120ctn signaling phenocopying the results obtained with shRNA targeting PKCe. Given the activity of PF-5263555 against PKCe, we attempted pharmacological recapitulation of PKCe knockdown in in-vitro and Oncogenic H-Ras activation results in a PKCe-dependent in-vivo models of RasV12-mediated EMT and breast cancer growth. mesenchymal phenotype Similar to the results attained using antisense knockdown of PKCe, Given the modulation of adherens junction protein’s phosphor- treatment of RasV12 HeLa cells with PF-5263555 resulted in a ylation state observed in the SILAC experiment and the known reversion of the cellular morphology promoted by RasV12 role of p120ctn in stabilizing E-cadherin, we hypothesize that PKCe activation (Figure 4d). Treatment of the Hela ER-RasV12 cells with activity could contribute to cadherin remodeling. Indeed, knock- PF-5263555 also resulted in loss of P-S268 signal (Figure 4e). down of PKCe was sufficient to rescue E-cadherin protein levels in To interrogate possible downstream effects of p120ctn P-S268, MDA-MB-231 and BT549 TNBC cells (Figure 3a and Supplementary we generated S268A and S268E mutants mimicking the depho- Figure 4a). Further upon induction of PKCe expression in MCF10A sophorylated and phosphorylated sites, respectively. As expected, cells, we find that less E-cadherin is associated with p120ctn as inhibition by PF-5263555 reduced P-S268 in both the over- assayed by immunoprecipitation (Figure 3b). expressed and endogenous p120ctn; however, the phospho- HeLa cells engineered to stably express a 4-hydroxytamoxifen specific antibody failed to recognize both the S268A (non- (4-OHT) responsive form of oncogenic H-ras, ER-RasV12, undergo a phoshpo) and the S268E (phospho-mimetic) mutants (Figure 4f mesenchymal morphology shift when exposed to 4-OHT (16; and Supplementary Figure 4b). We found that PF-5263555 Figure 3c). PKCe knockdown in this context reverted the RasV12- reduced b-catenin levels, whereas overexpression of the S268E induced cell morphology (Figure 3c). Correspondingly, a change in mutant protein resulted in higher basal b-catenin levels, largely HeLa cell morphology similar to that induced by RasV12 was refractory to PKCe inhibition (Figure 4f). Correspondingly, in ER- observed by expression of a constitutively active variant of PKCe RasV12 HeLa cells, b-catenin levels were reduced upon antisense (Supplementary Figure 5). Further, induction of RasV12 signaling in knockdown of PKCe in the absence and presence of Ras activation HeLa cells resulted in a significant increase in P-S268 that was (Figure 3d). Finally, by blocking global translation with cyclohex- sensitive to knockdown of PKCe by antisense oligonucleotides amide and assaying b-catenin decay over time, we find that (Figure 3d). In this model, and to a lesser extent in the MDA-MB- overexpression of p120ctn in HeLa cells elevates b-catenin 231 model, we observe a reduction in total p120ctn upon PKCe expression and reduces decay compared with vector control knockdown, perhaps due to impaired stability of p120ctn upon (Supplementary Figure 4c). Both S268A and S268E p120ctn dephosphorylation (Figures 2a and 3d). Of particular note, we mutants also stabilized total b-catenin with the S268E mutant observed an increase in total PKCe levels in the presence of having a more pronounced effect at time 0 (Supplementary activated Ras, (Figure 3d) which we quantified in a 4-OHT dose Figure 4c). Thus, it is likely that elevated p120ctn expression escalation (Figure 3e). 4-OHT dose-responsive phosphorylation of contributes to overall stability of b-catenin, and in this context canonical Ras pathway targets Akt (S473) and MAPK (Y-202/242) phosphorylation at S268 is a positive effector. were also recorded in this model. In summary, these data suggest Given that focal growth observed in MCF10A cells can be driven that, aside from its inherent oncogenic activity, PKCe is by PKCe expression, treatment with PF-5263555 should be upregulated by oncogenic Ras activation and functions as a sufficient to revert cells to contact inhibition. Indeed, PF- mediator of Ras-dependent p120ctn phosphorylation. 5263555 treatment of MCF10A cells expressing PKCe results in loss of foci formation (Figure 4g). When the downstream signaling PF-5263555, a PKCy-targeted inhibitor, is also potent against PKCe was interrogated in these cells, we also observed reduced p120ctn P-S268 with addition of PF-5263555 (Figure 4h). PF-5263555 is an orally available 5-aryl/heteroaryl-3-pyridinecar- bonitrile previously developed as an inflammation therapeutic targeting PRKCQ (PKCy) (Figure 4a;17). We took advantage of the PF-5263555 inhibition of PKCe recapitulates the genetic LOF reported cross-selectivity for PKCe in an IMAP biochemical assay phenotype in cells and inhibits tumor growth in vivo and revisited this aspect in the more sensitive UltraLance assay In agreement with previous experiments, PF-5263555 treatment of 17 format. The Km for ATP of PKCe by UltraLance was 4.4 mM with MDA-MB-231 cells stably expressing WTe significantly attenuated PF-5263555 showing classical ATP-competitive kinetics with an p120ctn P-S268 signal (Figure 5a). Treatment of both, BT549 and IC50 (50% inhibition of enzyme activity) of 1.2 nM obtained at 5 mM MDA-MB-231 cells with PF-5263555 resulted in a rescue of ATP concentration (Figure 4b). E-cadherin levels, substantiating the role for PKCe in maintaining We established a cell-based assay for testing the effectiveness an EMT-dependent phenotype in TNBC cells (Figure 5b and of PF-5263555 and structurally related derivatives on PKCe based Supplementary Figure 4d). PF-5263555 also produced a dose- on a methodology described by Cameron et al.18 and Ohno et al.19 dependent change in MDA-MB-231 morphology characterized Here, binding of ATP-competitive inhibitors by the kinase dead by a flattened epithelial-like morphology (Figure 5c). Likewise,

& 2014 Macmillan Publishers Limited Oncogene (2014) 1385 – 1394 Oncogenic Ras-PKCe signaling requires p120ctn SG Dann et al 1390

Figure 4. PF-5263555 is an ATP-competitive inhibitor with activity against PKCe in biochemical and cell-based assays. (a) Chemical structure of PF-5263555. (b) Dose response of PKCe inhibition with PF-5263555 at 5 mM ATP in a Lance Ultra in-vitro kinase assay format (IC50 B1.2 nM). (c) Western blot analysis of MDA-MB-231 cells overexpressing recombinant kinase-deficient mutant (K437R) or WT PKCe protein. Total PKCe and P-S729 protein levels were detected after a 24-h dose escalation of PF-5263555 (0, 67.5, 125, 250, 500 and 1000 nM). (d) Effect of 24-h 12 500 nM PF-5263555 treatment on induced (72 h, 100 nM 4-OHT) and uninduced HeLa ER-RasV cells in 2D cell culture (  10 magnification). 12 Black scale bars on micrographs represent 100 mM (e) HeLa ER-RasV cells were induced for 72 h (100 nM 4-OHT) and treated with vehicle or 500 nM PF-5263555 for 48 h. Western blotting analysis of comparable amounts of cell extracts are shown. Quantitation on the right indicates relative P-S268 levels with vehicle and PF-5263555 normalized to vimentin. (f) HeLa cells transiently expressing the indicated V5-tagged p120ctn proteins were treated with 500 nM PF-5263555 for 24 h, and extracts were subjected to western blotting. Recombinant (upper) and endogenous (lower) S268 phosphorylated p120ctn forms are indicated by arrowheads on the left. (g) Dox-inducible expression of PKCe in MCF10A cells results in foci formation, which is rescued by treatment with PF-5263555 photographed at  10 magnification. Black scale bars on micrographs represent 100 mM. Quantitation was performed by bright-field phase contrast microscopy as described in the Materials and methods section. (h) Dox-inducible expression of PKCe in MCF10A cells and treated with 500nM PF-5263555 analyzed by western blot with the antibodies indicated.

Oncogene (2014) 1385 – 1394 & 2014 Macmillan Publishers Limited Oncogenic Ras-PKCe signaling requires p120ctn SG Dann et al 1391

Figure 5. PF-5263555 attenuates TNBC growth in vitro and in vivo, and recapitulates effects of PKCe genetic LOF. (a) The effect of stable, constitutive overexpression of PKCe (WT), K437R (KD) PKCe or b-Gal control in MDA-MB-231 treated with or without 500 nM PF-5263555 was analyzed by western blot of cell extracts. Quantitation on the right indicates relative P-S268 levels with vehicle and PF-5263555 normalized to vimentin (b) Cell extracts of MDA-MB-231 treated with PF-5263555 for 48 h were subjected to western blotting as indicated. (c) MDA-MB-231 cells treated for 24 h with vehicle or PF-5263555 (250, 500, 1000 nM) grown under standard culture conditions (2D plastic) or embedded in ECM for 7 days (3D Matrigel) and photographed at  10 magnification. Black scale bars on micrographs represent 100 mm. (d) Mice with MDA-MB- 231 orthotopic xenografts were dosed orally (5 days on, 2 days off cycles) with 100 mg/kg PF-5263555 or vehicle (10 mice for each cohort; P-value determined by analysis of variance); 4/10 mice treated with PF-5263555 and 0/10 mice in the vehicle control had tumors regress to undetectable size.

& 2014 Macmillan Publishers Limited Oncogene (2014) 1385 – 1394 Oncogenic Ras-PKCe signaling requires p120ctn SG Dann et al 1392 MDA-MB-231 cells in 3D culture lose their ability to invade into the to be p120ctn dependent.29 Suprisingly, however, oncogenic ECM when treated with increasing doses of PF-5263555, which H-Ras-mediated anchorage-independent growth was refractory to may be representative of a more differentiated phenotype p120ctn knockdown in MDCK cells.29 This may be partly explained resulting in ductal structures (Figure 5c). by the work of Soto et al.30 who describe a tumor-suppressive Finally, mice harboring MDA-MB-231 orthotopic xenografts function for p120ctn when E-cadherin is high (as in the case of treated with 100 mg/kg PF-5263555 daily via oral gavage showed MDCK cells), but oncogenic when E-cadherin is low as in the case inhibition of tumor growth compared with vehicle-treated of MDA-MB-231 breast cancer cells. Similarly, Shibata et al.31 controls (Figure 5d) in agreement with the results of the shRNA- describe an oncogenic role for p120ctn in E-cadherin low breast induced inhibition of PKCe expression in MDA-MB-231 cells cancer cell lines and cytoplasmic localization of p120ctn when (Figure 2d). Of particular note, treatment of mice with the PF- E-cadherin is absent. Mislocalization of p120ctn from cell junctions 5263555 inhibitor resulted in tumor regression in 4/10 mice to cytoplasm has been reported in colon cancer cells undergoing reminiscent of the inducible shRNA effects in this model. Upon EMT.32 Not surprisingly, then loss of E-cadherin in breast cancer analyses of MDA-MB-231 xenograft tumor lysates, we find reduced patients correlates well with cytoplasmic localization of p120ctn, total p120ctn in three of the five tumors remaining after dosing and both of these factors independently predict poor outcome with PF-5263555 as compared with vehicle control (Supple- with respect to patient survival and disease progression.33 mentary Figure 4e). Although different Ras isoforms (that is, H-, K- and N-) are not interchangeable in regards to cellular phenotype, our results suggest that activated Ras can mediate changes in cell DISCUSSION morphology that are dependent on PKCe signaling. Integrin Various PKC family members have been previously implicated in signaling through focal adhesion kinase has been reported to be driving the malignant state of cancer cells.20 Early efforts in essential to Ras-driven transformation, and Ras signaling is oncology drug discovery to target PKC have largely focused on frequently associated with cadherin switching in EMT progres- the cancer cell autonomous effects of PKCa (ISIS 3125, PKC412, sion.34,35 Although Ras mutations are not common in breast UNC101) or the pro-angiogenic effects of PKCbII (Enzastaurin).1 To cancers, there are multiple paths to activated Ras where PKCe may date, targeting PKCs has been met with limited clinical success be a downstream effector, such as loss of the Ras-GAP NF1 or the due to poor compound properties or pleiotropic pharmacological miRNA let-7.36,37 In light of the data presented here, correlation of effects, leading to dose-limiting toxicity before reaching efficacy. PKCe expression in breast cancer patients with expression and Based on our results, we propose the specific targeting of the localization of adherens junction proteins, including p120ctn, may novel isoform PKCe, as this kinase has oncogenic properties when be useful in generating patient stratification hypotheses. Our data expressed at elevated levels in cell culture and in transgenic further indicate that PKCe is an important mediator of Ras signal mouse models. Recently, mesenchymal stem cell sub-populations transduction and malignant growth in TNBC cells, and represents isolated from an immortalized mammary epithelial cell line were a unique opportunity for the development of targeted cancer described as enriched for PKC pathway activation.21 Further, by therapies. forcing the bulk population of mammary epithelial cells to undergo EMT, strong activation of PKC pathways was obser- ved.21 Other groups have related EMT specifically to PKCe in MATERIALS AND METHODS human cancer through mir205 activity and more recently through Antisense-mediated knockdown a c-MET/mir103-signaling axis in lung cancer.22,23 Third-generation genebloc antisense molecule gapmers38 were synthe- Previous studies by the Ivaska and Parker labs have described sized by Biospring (Frankfurt, Germany). Each antisense molecule used PKCe involvement in integrin recycling and zonula occludens was controlled by a corresponding 4 nucleotide control oligonucleotide. Transient knockdown experiments, and subsequent signaling or pheno- stability, providing a mechanistic explanation for the observed 38 effects of PKCe on cell invasion.10,11 Profiling studies of tumor typic analysis were carried out as described previously. Briefly, GAPmer oligonucleotides targeting PRKCE were transfected with Atufect transfec- samples from breast cancer patients and cell lines have revealed a V12 24 tion reagent (Silence Therapeutics, Berlin, Germany). HeLa ER-Ras cells stem cell-like gene signature associated with EMT. Of particular were seeded 300 000 cells/10-cm plate, 24 h before transfection. relevance is the finding that breast cancer samples with stem cell- Oligonucleotide and transfection reagent were complexed according to like characteristics show upregulated b-catenin and OB-cadherin the manufacturer’s instructions and delivered to cells at a final concen- mRNA, both of which were also identified as being modified in our tration of 30 nM and 1.2 mg/ml, respectively. Antisense oligonucleotide SILAC analysis in a PKCe-dependent manner. sequences used in this study are found in Supplementary Table 2. To the best of our knowledge, no previous reports have linked deregulation of the armadillo catenin members in response to loss 3D cell culture assays and/or inhibition of PKCe. In the phosphoproteomic screen Experiments involving MDA-MB-231 cells used an ECM-embedded presented here we see a reduction in phosphorylated peptides protocol as follows. MDA-MB-231 cells engineered with inducible shRNA of a-catenin, b-catenin, Desmoplakin, and p120ctn. Although constructs were induced with 250 ng/ml Dox for 48 h before embedding in p120ctn S879 was reported to become phosphorylated upon 300 ml of ice-cold 80% liquid ECM (BD Biosciences) 20% full media. A total treatment with PDGF in a PKCa-dependent manner, the PKC of 100 000 cells were resuspended in ECM. This ECM cell suspension was isoforms responsible for other p120ctn-specific phosphorylation allowed to harden for 2 h, and 1 ml of media supplemented with vehicle, events have remained largely unknown.25,26 This may explain the Dox or PF-5263555 was added. conflicting results from Xia et al.,27 where P-S268 is downregulated upon high PKC activity. Previous studies have reported opposing MDA-MB-231 orthotopic xenografts and PF-5263555 dosing effects on growth and proliferation mediated by PKCd and PKCe.4 MDA-MB-231 stable cells (1 Â 106) expressing Dox-inducible PKCe or NT Thus, it is possible that activation of the tumor-suppressive PKCd shRNAs cultured in the absence of Dox were injected orthotopically into in the context outlined by Xia et al.27 has a dominant effect over the right abdominal mammary fat pad of female nude mice. Groups of 15 PKCe activity, resulting in a net dephosphorylation of S268. mice received mammary fat pad inoculation of pooled PKCe or NT shRNAs, expressing MDA-MB-231 cells. Dox was administered through food at Tyrosine phosphorylation of p120ctn has previously been 28 625 mg/kg. All animals were completely dissected and photographically reported in response to activation of oncogenic Src. Although documented. this phosphorylation does not alter its binding to E-cadherin, it For compound efficacy experiments, 1 Â 106 MDA-MB-231 cells were does result in a reduction in cell adhesion. Further, constitutive injected orthotopically into the right abdominal mammary fat pad of Src- and Rac-mediated anchorage-independent growth was found female nude mice. When tumors reached an average tumor volume of

Oncogene (2014) 1385 – 1394 & 2014 Macmillan Publishers Limited Oncogenic Ras-PKCe signaling requires p120ctn SG Dann et al 1393 200 mm3, they were randomized into groups and received either culture flasks with the appropriate selection (G418 at 500 mg/ml, Blasticidin PF-5263555 or vehicle (methocel tween) dose via oral gavage once daily at 10 mg/ml or Zeocin at 100 mg/ml) until resistant cell colonies formed, and with weekend drug holidays for 49 days. antibiotic induced cell death ceased. Several hundred resistant colonies were pooled and expanded to form the stable cell lines employed herein. SILAC phosphoproteomic LC–MS/MS Immunoblotting and immunoprecipitations Equal amounts of light and heavy arginine/lysine isotope samples were combined and subjected to proteolytic digestion with Lys-C, trypsin and Standard western blotting techniques were used for all immunblotting Glu-C. Phosphopeptides were enriched by strong-cation exchange experiments described. Cell lysates were generated from 2D and 3D fractionation followed by the titanium dioxide chromatography, as growth conditions using RIPA buffer containing 40 mM HEPES, pH 7.5, described previously.39 Peptide identification and relative quantification 120 mM NaCl, 1 mM EDTA, 5% glycerol, 1% NP40, 1% Na deoxycholate, 0.1% were performed on a reversed-phase Magic C18 nanocolumn coupled with SDS, protease inhibitor cocktail (Sigma-Aldrich, St Louis, MO, USA) and an LTQ-Orbitrap XL mass spectrometer (Thermo Fisher, Waltham, MA, phosphatase inhibitor cocktails I and II (Sigma-Aldrich). Antibodies used in USA). MS/MS raw spectra were searched against the human IPI database this study are found in Supplementary Table 3. using Biowork 3.3.1 (Thermo Fisher). The relative ratio of the intensity of Immunoprecipitations were performed using antibodies indicated in the heavy versus the light peptides was used to express the degree of Supplementary Table 3. In brief, cells were lysed in CHAPS-based lysis phosphorylation change of a given protein. buffer containing 40 mM HEPES pH 7.5, 120 mM NaCl, 1 mM EDTA, 5% glycerol, 1% CHAPS, protease inhibitor cocktail (Sigma-Aldrich), and phosphatase inhibitor cocktails I and II (Sigma-Aldrich). Upon removal of In vitro and cell-based PKC assays cell debris by centrifugation, lysates were precleared with protein G The inhibitory potency of compounds against PKCe was determined in the GammaBind beads (GE Healthcare, Buckinghamshire, UK) with rotation for LANCE Ultra kinase assay format. The assay was carried out in a white 1 h at 4 1C. Protein A beads were removed and mouse monoclonal anti- 384-well plate (Matrix Technologies, Hudson, NH, USA) by incubation of p120ctn antibody was added to a final concentration of 1/250 mg protein PKCe (5 pM, Invitrogen, Grand Island, NY, USA) with inhibitor, Ulight-labeled with rotating incubation overnight at 4 1C. p120ctn immunocomplexes peptide substrate (50 nM CRFARKGSLRQKNV, Perkin Elmer) and ATP (5 mM) were captured with protein G GammaBind beads by constant rotation for 1 in assay buffer (20 mM HEPES, pH 7.2, 2.5 mM MgCl2,2mM Cys, 0.005% 4 h at 4 C, recovered by centrifugation and then washed six times with Brij35). The reaction was stopped with EDTA (20 mM) after 60 min ice-cold CHAPS lysis buffer. p120ctn immunocomplexes were eluted with incubation at room temperature. The formation of phosphopeptide standard SDS gel-loading buffer by boiling for 5 min and analyzed by product was determined by Europium-labeled anti-phospho PKC (A25S) standard western blotting techniques. antibody (1 nM, Perkin Elmer, Waltham, MA, USA). The TR-FRET signals at 615 and 665 nm wavelengths upon excitation at 340 nm were measured Microscopy, immunocytochemistry and IHC using the Envision fluorescence plate reader from Perkin Elmer. The signal Micrographs of live cells grown in 2D and 3D cell culture were taken using ratio of 665/615 (Ulight/Europium) was used in all data analyses. The IC50 phase contrast settings on a Zeiss Axioscope (Carl Zeiss AG, Oberkochen, value was obtained by fitting the data to a dose-response inhibition equa- Germany) at the indicated magnifications. Focal growth was quantified tion using GraphPad Prism (GraphPad Software, Inc., La Jolla, CA, USA). from images recorded at identical focal plane, brightness and exposure in The cell-based assay of KD PKCe P-S729 was performed in engineered phase contrast conditions at  2.5 magnification. Microscopy images were MDA-MB-231 cells described above. Cells were treated with PF-5263555 in analyzed using the Adobe Photoshop CS5 (Adobe Systems Inc., San Jose, a dose response for 24 h, lysed and analyzed by western blot as described CA, USA) extended version measurements tool, where integrated density below. Rationale and full description of this approach can be found in 18 of the micrograph is calculated from the product of the image area (in Cameron et al. pixels) and the mean gray value (measurement of image brightness). Immunoflourescence was carried out on cells grown on collagen-coated Molecular cloning, viral preparations and stable cell-line coverslips. Briefly, cells were fixed with 4% paraformaldehyde in engineering Phosphate-buffered saline (PBS) and permeabilized with 0.5% Triton V12 X-100 PBS. Samples were washed and blocked with 3% BSA in PBS, and The 4-OHT-inducible oncogenic form of H-Ras, ER-Ras , has been then incubated for 1 h with primary antibody. Cells were washed (PBS) previously described.16 PKCe and p120ctn expression plasmids suitable before a 30 min incubation with Alexaflour488 (Invitrogen)-conjugated for lentiviral recombination were created using Invitrogen’s ViraPower secondary antibody (Invitrogen) and then washed again. Cells were b b Lentiviral Expression Systems ( -galactosidase, -Gal, control supplied mounted on slides with Gold antifade mounting media with 4’,6-diamidino by Invitrogen). Briefly, PRKCE and CTNND1 open-reading frames were -2-phenylindole and phalloidin stains. Micrographs of immunofluorescent cloned into pENTR-D followed by transformation into TOP10 competent labels were taken on a Zeiss confocal microscope (Carl Zeiss AG, Escherichia coli (Invitrogen). Positive clones were sequenced and Oberkochen, Germany) at the indicated magnification. recombined into pLenti4-TO/V5 or pLenti6.3-TO/V5 through homologous IHC was performed on formalin-fixed paraffin-embedded tissue sections recombination followed by transformation of competent STBL3 E. coli using standard protocols and developed through 3,3-Diaminobenzidine (Invitrogen). Ampicillin/Zeocin- or Ampicillin/Blasticidin-resistant colonies detection and horseradish peroxidase conjugated secondary antibody. were screened by PCR and prepared as endofree plasmid (Qiagen, Tissue microarrays with both normal and tumor tissue were procured from Germantown, MD, USA). Site-directed mutagenesis was performed on US Biomax (BR1921). Slides were conterstained using standard hematox- e pENTR-D-PKC to create kinase-inactive (K437R) and constitutively active ylin staining protocol. (A159E) forms as well as pENTR-D-p120ctn to create S268A and S268E mutants. All shRNA lentiviral plasmids were constructed using annealed oligonucleotides using standard cloning methods into the pLentiCiti 7SK Statistical methods vector.40 Primer sequences used for cloning steps including shRNA Western blots and phase contrast microscopy were quantified by using the sequences are found in Supplementary Table 2. Adobe Photoshop CS5 extended version measurements tool based on Pseudotyped retroviral particles were created by transfecting the packaging overall brightness. For western blots, arbitrary units of brightness cell line GP2-293 with pLIXN-ER-RasV12 (5 mg) and pVSVG (7 mg) using normalized to a loading control were compared between experimental Effectene transfection reagent (Qiagen) in the presence of 2.5 mM chloroquine. conditions by two-tailed Student’s t-test assuming unequal variance. For The next day media was aspirated and replaced with full media without phase contrast images arbitrary units of the reciprocal function from chloroquine, and retroviral particles were harvested after 3 days. Pseudotyped integrated density values were compared between experimental condi- lentiviral particles (harboring shRNA or PKCe, b-Gal and p120ctn expression tions by two-tailed Student’s t-test assuming unequal variance. In-vivo cassettes) were obtained through similar methodology except that the tumor growth was measured weekly using standard caliper measurements packaging cell line used is 293-FT, and 7 mg of the virapower packaging mix and statistically analyzed using repeated measures analysis of variance. (Invitrogen) was co-transfected with 5 mg of lentiviral plasmid DNA.40 Stable HeLa and MDA-MB-231 cell lines were created by exposing cells at 50% confluency in a six-well plate to pseudotyped retroviral or lentiviral CONFLICT OF INTEREST particles at a multiplicity of infection of one or less for a 24-h period in the All the authors are full time employees of Pfizer, Inc., with no relationships that they presence of 8 mg/ml polybrene. After 24 h media was replaced with full believe could be construed as resulting in an actual, potential, or perceived conflict of media, cells were grown to confluence and then passaged into T-175 interest with regard to the manuscript submitted for review.

& 2014 Macmillan Publishers Limited Oncogene (2014) 1385 – 1394 Oncogenic Ras-PKCe signaling requires p120ctn SG Dann et al 1394 ACKNOWLEDGEMENTS 21 Scheel C, Eaton EN, Li SH, Chaffer CL, Reinhardt F, Kah KJ et al. Paracrine and The authors would like to thank Drs Robert T Abraham and Valeria Fantin for critical autocrine signals induce and maintain mesenchymal and stem cell states in the review of the manuscript. breast. Cell 2011; 145: 926–940. 22 Garofalo M, Romano G, Di Leva G, Nuovo G, Jeon YJ, Ngankeu A et al. EGFR and MET receptor tyrosine kinase-altered microRNA expression induces tumorigenesis and gefitinib resistance in lung cancers. Nat Med 2012; 18: 74–82. REFERENCES 23 Gandellini P, Folini M, Longoni N, Pennati M, Binda M, Colecchia M et al. miR-205 1 Mackay HJ, Twelves CJ. Targeting the protein kinase C family: are we there yet? Exerts tumor-suppressive functions in human prostate through down-regulation Nat Rev Cancer 2007; 7: 554–562. of protein kinase Cepsilon. Cancer Res 2009; 69: 2287–2295. 2 Griner EM, Kazanietz MG. Protein kinase C and other diacylglycerol effectors in 24 Hennessy BT, Gonzalez-Angulo AM, Stemke-Hale K, Gilcrease MZ, Krishnamurthy cancer. Nat Rev Cancer 2007; 7: 281–294. S, Lee JS et al. Characterization of a naturally occurring breast cancer subset 3 Gorin MA, Pan Q. Protein kinase C epsilon: an oncogene and emerging tumor enriched in epithelial-to-mesenchymal transition and stem cell characteristics. biomarker. Mol Cancer 2009; 8:9. Cancer Res 2009; 69: 4116–4124. 4 Mischak H, Goodnight JA, Kolch W, Martiny-Baron G, Schaechtle C, Kazanietz MG 25 Brown MV, Burnett PE, Denning MF, Reynolds AB. PDGF receptor activation et al. Overexpression of protein kinase C-delta and -epsilon in NIH 3T3 cells induces p120-catenin phosphorylation at serine 879 via a PKCalpha-dependent induces opposite effects on growth, morphology, anchorage dependence, and pathway. Exp Cell Res 2009; 315: 39–49. tumorigenicity. J Biol Chem 1993; 268: 6090–6096. 26 Reynolds AB, Roczniak-Ferguson A. Emerging roles for p120-catenin in cell 5 Cacace AM, Guadagno SN, Krauss RS, Fabbro D, Weinstein IB. The epsilon isoform adhesion and cancer. Oncogene 2004; 23: 7947–7956. of protein kinase C is an oncogene when overexpressed in rat fibroblasts. 27 Xia X, Carnahan RH, Vaughan MH, Wildenberg GA, Reynolds AB. p120 serine and Oncogene 1993; 8: 2095–2104. threonine phosphorylation is controlled by multiple ligand-receptor pathways but 6 Hafeez BB, Zhong W, Weichert J, Dreckschmidt NE, Jamal MS, Verma AK. Genetic not cadherin ligation. Exp Cell Res 2006; 312: 3336–3348. ablation of PKC epsilon inhibits prostate cancer development and metastasis in 28 Ozawa M, Ohkubo T. Tyrosine phosphorylation of p120(ctn) in v-Src transfected L transgenic mouse model of prostate adenocarcinoma. Cancer Res 2011; 71: 2318–2327. cells depends on its association with E-cadherin and reduces adhesion activity. 7 Benavides F, Blando J, Perez CJ, Garg R, Conti CJ, Digiovanni J et al. Transgenic J Cell Sci 2001; 114(Pt 3): 503–512. overexpression of PKCepsilon in the mouse prostate induces preneoplastic 29 Dohn MR, Brown MV, Reynolds AB. An essential role for p120-catenin in Src- lesions. Cell Cycle 2011; 10: 268–277. and Rac1-mediated anchorage-independent cell growth. J Cell Biol 2009; 184: 8 Leask A, Shi-Wen X, Khan K, Chen Y, Holmes A, Eastwood M et al. Loss of protein 437–450. kinase Cepsilon results in impaired cutaneous wound closure and myofibroblast 30 Soto E, Yanagisawa M, Marlow LA, Copland JA, Perez EA, Anastasiadis PZ. p120 function. J Cell Sci 2008; 121(Pt 20): 3459–3467. catenin induces opposing effects on tumor cell growth depending on E-cadherin 9 Saurin AT, Durgan J, Cameron AJ, Faisal A, Marber MS, Parker PJ. The regulated expression. J Cell Biol 2008; 183: 737–749. assembly of a PKCepsilon complex controls the completion of cytokinesis. Nat Cell 31 Shibata T, Kokubu A, Sekine S, Kanai Y, Hirohashi S. Cytoplasmic p120ctn regu- Biol 2008; 10: 891–901. lates the invasive phenotypes of E-cadherin-deficient breast cancer. Am J Pathol 10 Ivaska J, Vuoriluoto K, Huovinen T, Izawa I, Inagaki M, Parker PJ. PKCepsilon- 2004; 164: 2269–2278. mediated phosphorylation of vimentin controls integrin recycling and motility. 32 Bellovin DI, Bates RC, Muzikansky A, Rimm DL, Mercurio AM. Altered localization EMBO J 2005; 24: 3834–3845. of p120 catenin during epithelial to mesenchymal transition of colon carcinoma is 11 Tuomi S, Mai A, Nevo J, Laine JO, Vilkki V, Ohman TJ et al. PKCepsilon regulation of prognostic for aggressive disease. Cancer Res 2005; 65: 10938–10945. an alpha5 integrin-ZO-1 complex controls lamellae formation in migrating cancer 33 Paredes J, Correia AL, Ribeiro AS, Milanezi F, Cameselle-Teijeiro J, Schmitt FC. cells. Sci Signal 2009; 2: ra32. Breast carcinomas that co-express E- and P-cadherin are associated with p120- 12 Polyak K, Weinberg RA. Transitions between epithelial and mesenchymal states: catenin cytoplasmic localisation and poor patient survival. J Clin Pathol 2008; 61: acquisition of malignant and stem cell traits. Nat Rev Cancer 2009; 9: 265–273. 856–862. 13 Zeisberg M, Neilson EG. Biomarkers for epithelial-mesenchymal transitions. J Clin 34 Doehn U, Hauge C, Frank SR, Jensen CJ, Duda K, Nielsen JV et al. RSK is a principal Invest 2009; 119: 1429–1437. effector of the RAS-ERK pathway for eliciting a coordinate promotile/invasive 14 Singh A, Greninger P, Rhodes D, Koopman L, Violette S, Bardeesy N et al. A gene gene program and phenotype in epithelial cells. Mol Cell 2009; 35: 511–522. expression signature associated with ‘K-Ras addiction’ reveals regulators of EMT 35 Pylayeva Y, Gillen KM, Gerald W, Beggs HE, Reichardt LF, Giancotti FG. and tumor cell survival. Cancer Cell 2009; 15: 489–500. Ras- and PI3K-dependent breast tumorigenesis in mice and humans requires focal 15 Kenny PA, Lee GY, Myers CA, Neve RM, Semeiks JR, Spellman PT et al. The adhesion kinase signaling. J Clin Invest 2009; 119: 252–266. morphologies of breast cancer cell lines in three-dimensional assays correlate 36 Craig DW, O’Shaughnessy JA, Kiefer JA, Aldrich J, Sinari S, Moses TM et al. with their profiles of gene expression. Mol Oncol 2007; 1: 84–96. Genome and transcriptome sequencing in prospective refractory metastatic triple 16 Leenders F, Mopert K, Schmiedeknecht A, Santel A, Czauderna F, Aleku M et al. negative breast cancer uncovers therapeutic vulnerabilities. Mol Cancer Ther 2013; PKN3 is required for malignant prostate cell growth downstream of activated 12: 104–116. PI 3-kinase. EMBO J. 2004; 23: 3303–3313. 37 Yu F, Yao H, Zhu P, Zhang X, Pan Q, Gong C et al. let-7 regulates self renewal and 17 Prashad AS, Wang D, Subrath J, Wu B, Lin M, Zhang MY et al. C-5 substituted tumorigenicity of breast cancer cells. Cell 2007; 131: 1109–1123. heteroaryl-3-pyridinecarbonitriles as PKCtheta inhibitors: part II. Bioorg Med Chem 38 Sternberger M, Schmiedeknecht A, Kretschmer A, Gebhardt F, Leenders F, Lett 2009; 19: 5799–5802. Czauderna F et al. GeneBlocs are powerful tools to study and delineate signal 18 Cameron AJ, Escribano C, Saurin AT, Kostelecky B, Parker PJ. PKC maturation is transduction processes that regulate cell growth and transformation. Antisense promoted by nucleotide pocket occupation independently of intrinsic kinase Nucleic Acid Drug Dev 2002; 12: 131–143. activity. Nat Struct Mol Biol 2009; 16: 624–630. 39 Wu J, Shakey Q, Liu W, Schuller A, Follettie MT. Global profiling of 19 Okuzumi T, Fiedler D, Zhang C, Gray DC, Aizenstein B, Hoffman R et al. Inhibitor phosphopeptides by titania affinity enrichment. J Proteome Res 2007; 6: hijacking of Akt activation. Nat Chem Biol 2009; 5: 484–493. 4684–4689. 20 Weinstein IB. The origins of human cancer: molecular mechanisms of carcino- 40 Czauderna F, Santel A, Hinz M, Fechtner M, Durieux B, Fisch G et al. Inducible genesis and their implications for cancer prevention and treatment—twenty- shRNA expression for application in a prostate cancer mouse model. Nucleic Acids seventh G.H.A. Clowes memorial award lecture. Cancer Res 1988; 48: 4135–4143. Res 2003; 31: e127.

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

Oncogene (2014) 1385 – 1394 & 2014 Macmillan Publishers Limited