p63 is an alternative p53 repressor in melanoma that confers chemoresistance and a poor prognosis

Primers designed to detect all splice variants of TAp63 (Koga et al., 2003) and ΔNp63 (Yang et al., 1998) were used to determine cellular expression in the skin. Both TA and ΔN p63 isoforms were expressed in samples obtained from different body sites from a range of skin phototypes (Fig. 1 A). TAp63 and ΔNp63 were expressed in keratinocytes; the relative expression was in keeping with a previous study (Senoo et al., 2007). TAp63 but not ΔNp63 was also expressed in fibroblasts. RT-PCR analysis failed to detect expression of either isoform in primary melanocyte cultures, but both isoforms were expressed in a metastatic melanoma cell line (WM1158). Despite the apparent lack of TP63 expression in primary melanocytes, quantitative PCR (Q-PCR) demonstrated significant up-regulation of TP63 in 24/33 (73%) melanoma cell lines (Fig. 1 B). No correlation was observed between expression of either TAp63 and/or ΔNp63 with BRAF^V600E mutation or NRAS mutation status (P > 0.05, Fisher’s exact test). In general, up-regulation of the two isoforms of TP63 was mutually exclusive with simultaneous up-regulation occurring infrequently (18%; n = 6). Western blotting data confirmed differential expression of p63 protein in melanoma cell lines compared with primary melanocyte cultures (Fig. 1 C).

Analysis of whole genome microarray data in melanoma tissue samples by Scatolini et al. (2010) revealed up-regulation of TP63 at all stages of melanoma progression but no significant up-regulation of TP53 or TP73 (Fig. 1 D). Immunohistochemical expression of p63 was investigated in 40 primary melanoma tissue samples in a melanoma tissue microarray (mTMA; Biomax). Overall, 16/40 (40%) samples in the mTMA demonstrated strong labeling of p63 and costaining with HMB-45 (human melanoma black-45), confirming expression of p63 in melanoma cells; two examples are shown in Fig. 1 E. Statistical analysis of clinical data available for the melanoma tissue array samples demonstrated no significant demographic differences between p63-positive and -negative melanomas (Table S1).

Primary melanocyte cultures were treated with UVB radiation and various chemotherapeutic agents (cisplatin, etoposide, and doxorubicin) from 6 to 48 h (Fig. 2 A and not depicted). No stabilization or reactivation of p63 was observed in melanocyte cultures despite induction of DNA damage and induction of an apoptotic pathway. Upon DNA damage, up-regulation of both nuclear and cytoplasmic p63 was observed in 8/9 (88%) established melanoma cell lines, with stabilization occurring as early as 2 h and in some cell lines persisting for 48 h (not depicted). Stabilization of the p63 gene and protein was demonstrated in established melanoma cell lines treated with various DNA-damaging agents (Fig. 2, C and D). No induction of p63 was observed upon treatment with dacarbazine (not depicted). Treatment with novel BRAF inhibitors (PLX4032 and PLX4720) affected p63 expression in a cell line–specific manner with stabilization of p63 in some cases (Fig. 2, E and F).

Cytoplasmic stabilization of p63 in melanoma cell lines led to investigation of relocation of p63 to different subcellular compartments. MitoTracker Orange was used as a marker of mitochondria to investigate localization of extranuclear p63 upon DNA damage. In untreated melanoma cells, localization of p63 is largely confined to the nucleus (Fig. 2 E), but upon treatment with paclitaxel (as early as 3 h) stabilization of p63 was observed in the cytoplasm and, more specifically, in the mitochondria, as demonstrated by colocalization with MitoTracker Orange (Fig. 2 G). Similar effects were observed in melanoma cells treated with doxorubicin and etoposide (not depicted). Collectively, these data confirm that p63 is up-regulated in response to genotoxic stress and is localized to both nuclei and mitochondria.

Subcellular fractions of melanoma cell protein lysates were analyzed using Western blotting to enrich for proteins of interest within mitochondrial and nuclear fractions. Purity was confirmed by incubation with anti–Lamin A (nuclear), anti-mtHsp70 (mitochondrial heat shock protein 70; mitochondrial), and anti-GAPDH (cytosolic) antibodies. In untreated WM1158 cells, TAp63α was largely expressed in the nuclear fraction, whereas TAp63γ was largely expressed in the mitochondrial compartment, with up-regulation of each isoform in their respective compartments in response to chemotherapeutic agents (Fig. 2 H). Results showed predominant p53 stabilization in the nucleus, with less pronounced stabilization in mitochondria (Fig. 2 H). In support of this, immunofluorescence microscopy of the same cells confirmed nuclear stabilization upon DNA damage treatment, suggesting that in melanoma cells, p53 also displays nuclear stabilization in response to genotoxic stress with possible translocation to the mitochondria (not depicted). In A375M cells, ΔNp63α and ΔNp63β were stabilized in nuclear and mitochondrial compartments upon treatment (not depicted).

Immunogold localization of protein using transmission electron microscopy was used to determine the exact location of p63 within mitochondria. Treated A375M and WM1158 cells were pelleted, fixed, and processed for transmission electron microscopy. Mitochondria were identified by their characteristic ultrastructure and confirmed by the presence of immunogold particles secondary to anti-mtHsp70 antibody (Fig. 2, I and J). Upon treatment with paclitaxel, immunogold particles of p63 were demonstrated in the nucleus and the mitochondria of both cell lines using various anti-p63 antibodies (Fig. 2, I and J). No association of immunogold particles with other subcellular structures identified by their characteristic ultrastructure, e.g., the Golgi body, was observed (not depicted). These data suggest that upon genotoxic stress, p63 translocates between the nucleus, cytoplasm, and mitochondrial compartments to exert its function.

Subcellular fractionation enriches for p63 protein within each fraction but does not allow for assessment of changes in concentration of p63 in response to genotoxic stress. To quantify the relative concentration of p63 in different cellular compartments, a new method recently developed in our laboratory was used as an alternative to the fractionation technique (summarized in Materials and methods and Fig. 3 A; Leverrier et al., 2007).

A375M cells were treated with etoposide in a time-dependent manner to establish the kinetic profile of p63 expression using flow cytometry (Fig. 3 A). To quantify the degree of p63 stabilization upon treatment, the percentage of positive events was combined with the relative fluorescent intensity of the Cy5 signal. Expression of p63 was determined by comparing the expression of p63 with the IgG mouse isotype control for each treatment. The relative concentration of p63 in treated A375M cells (considered total p63) and in fractionated cells in the nuclei and mitochondria was calculated by comparing concentration of p63-Cy5 in each fraction for each treatment with the untreated sample (Fig. 3 A). Histograms demonstrate stabilization of total p63 to a maximum at 6 h with a reduction thereafter to 24 h (Fig. 3 A, i and ii). When cells were lysed and reanalyzed (Fig. 3 A, iii and iv), expression of nuclear p63 linearly increased to a maximum at 24 h, whereas the stabilization profile of mitochondrial p63 reflected that of total p63 (Fig. 3 A, iii and v).

Flow cytometry experiments confirmed stabilization of p63 in both nuclear but to a greater extent in the mitochondrial compartment in other fractionated melanoma cell lines, including WM1158 (metastatic melanoma) and SBCl2 (primary radial growth phase [RGP] melanoma) cells, upon treatment with cisplatin, etoposide, and paclitaxel compared with untreated samples (Fig. 3, B and C; and not depicted). The purity of the subcellular fractions was confirmed using confocal microscopy (Fig. 3, D and E).

Posttranslational modifications, e.g., phosphorylation, of p63 are reported to significantly alter protein levels (Osada et al., 1998; MacPartlin et al., 2005; Westfall et al., 2005; Suh et al., 2006). Exposure to paclitaxel resulted in up-regulation of total phosphorylated p63 (ser 160/162), which constituted a small fraction of total p63 stabilized, as demonstrated using the flow cytometry fractionation method (Fig. 3 F). Stabilization of phosphorylated p63 was maximal at 90 min and 3 h in the nuclear and mitochondrial fractions, respectively. These data show that in melanoma cells, upon DNA damage, p63 could be phosphorylated in the nucleus before partially relocating to the mitochondria.

Three oligonucleotide sequences were used to silence both TA and ΔN p63 in melanoma cells (Fig. 4 A) to determine the effect on chemosensitivity. The morphology of A375M and WM1158 cells depleted of p63 using three different shRNA-p63 clones was not dissimilar to those stably expressing shRNA-scramble (not depicted). Significant depletion of p63 using all three shRNA-p63 clones (1, 2, and 3) was demonstrated using Q-PCR and Western blotting in WM1158 (Fig. 4 B) and A375M cells (Fig. 4 E). No difference in TAp63 and ΔNp63 mRNA expression was observed between nontransfected melanoma cells and those stably expressing scramble sequences, confirming the shRNA-scramble had no effect on p63 gene expression (Fig. 4, B and E). Depletion of p63 using shRNA clones 1 and 2 in both WM1158 (Fig. 4 C) and A375M cells (Fig. 4 F) demonstrated significantly increased apoptosis upon treatment with etoposide, paclitaxel, and cisplatin, as demonstrated by the Annexin V assay. No effect was observed with dacarbazine, even at increasing doses (Fig. 4, C and F; and not depicted). Treatment with novel BRAF inhibitors (PLX4032 and PLX4720) significantly increased apoptosis of melanoma cells that were depleted of p63 (Fig. 4, I and J).

Upon treatment of WM1158 cells (sh-Ctrl) with etoposide or paclitaxel (16 h), induction of the mitochondrial apoptotic pathway was demonstrated through cytosolic release of cytochrome c and up-regulation of Apaf-1 (apoptotic protease-activating factor 1; Fig. 4 D). Western blotting of WM1158 cells treated with various chemotherapeutic agents demonstrated significant stabilization of total and phosphorylated p53 (ser-15), MDM2 (mouse double minute 2), and MDM4 in p63-depleted cells (Fig. 4 G). In addition, up-regulation of proapoptotic BCL2 proteins such as Bax, Bak, and Puma and down-regulation of antiapoptotic factors such as Bcl-xL, Mcl-1 (myeloid cell leukemia 1), and Bcl-2 were detected in sh-p63 cells exposed to cisplatin, etoposide, or paclitaxel (Fig. 4 G). Among the down-regulated antiapoptotic factors, Bcl-xL was consistently down-regulated by all the treatments in the absence of p63, whereas Mcl-1 expression was significantly decreased only by paclitaxel, which is reported to phosphorylate Bcl-2 (Yamamoto et al., 1999). A similar pattern of expression was also observed in A375M cells upon treatment with paclitaxel (Fig. 4 H), supporting a role for p53 and its downstream targets in the chemosensitivity conferred by depletion of p63. Treatment with BRAF inhibitors (PLX4032 and PLX4720) inhibited phosphorylation of BRAF downstream targets, e.g., phosphorylated ERK (extracellular signal-regulated kinase; P-ERK) and P-MEK (mitogen-activated protein/ERK), whereas expression levels increased in those cells depleted of p63 (Fig. 4 K). Moreover, p53 was phosphorylated (at ser-15) in p63-depleted cells upon treatment with BRAF inhibitors, together with partial reduction in MDM2 expression and significant reduction in levels of MDM4.

Induction of p53 downstream targets and the mitochondrial apoptotic pathway raise the possibility that p63 may substitute for or affect the apoptotic function of WT-p53 in melanoma. Flow cytometry fractionation experiments were used to delineate the relationship between p53 and p63, using melanoma cells depleted from p53 (si-p53 A375M cells). MDM2 expression wasn’t affected in these cells although the expression of MDM4 significantly increased by depletion of p53 (Fig. 5 A, top). When comparing these melanoma cells with the parental cell line expressing WT-p53 (A375M), upon treatment with paclitaxel or etoposide, significantly greater total (Fig. 5 A), nuclear (Fig. 5 B), and mitochondrial p63 (Fig. 5 C) were observed in the WT-p53 melanoma cell line (A375M), suggesting that translocation of p63 to the mitochondria is dependent on the presence of p53. This was also further confirmed in p53-null melanoma cells (UISO-Mel-6) expressing mainly TAp63 isoform (Fig. 1 B). Upon treatment with paclitaxel, stabilization of TAp63 was predominantly observed in the nuclei (not depicted). Flow cytometry analysis demonstrated up-regulation of total p63 protein levels upon treatment with limited stabilization in either nuclear or mitochondrial fractions (not depicted). Using the same flow cytometry quantification method, significantly greater stabilization of total and nuclear p53 was observed in A375M sh-p63 cells compared with A375M sh-scramble cells upon treatment with etoposide or paclitaxel (Fig. 5 D, left and middle). In contrast, depletion of p63 significantly decreased translocation of p53 to the mitochondria (Fig. 5 D, right). These data suggest that translocation of p53 to the mitochondria may require p63, but more importantly, p63 physiologically prevents p53 nuclear stabilization in melanoma cells. Western blotting of nuclear extracts from A375M cells depleted of p63 after treatment with paclitaxel confirmed significant up-regulation of p53, particularly the phosphorylated forms at ser-15 and ser-46 (the latter reported to link p53 with induction of apoptosis [Oda et al., 2000]), suggesting that up-regulation of p53 in the nucleus results in induction of the mitochondrial apoptotic pathway (Fig. 5 E). Moreover, immunoprecipitation with specific anti-p53 antibody in untreated A375M cells provides evidence of the endogenous interaction with p63 in melanoma cells (Fig. 5 F).

Immunohistochemistry of benign and malignant melanocytic lesions was undertaken to determine the effect of p63 reactivity on patient outcomes. Initially, paraffin-embedded samples of 31 benign intradermal nevi in 17 individuals (6 males and 11 females) were analyzed for p63 expression (using H129 anti-p63 antibody). Mean age at diagnosis was 50.1 yr (range 33.9–70.2 yr). Positive internal control labeling of epidermal keratinocytes with p63 was used for specificity of staining. Overall, positive labeling of p63 in neval melanocytes was demonstrated with low frequency (3/31; 3.2%).

Overall, 81 paraffin-embedded primary cutaneous melanoma tissue samples from 80 individuals were analyzed for expression of p63. Mean age at diagnosis for primary melanomas was 58.4 yr (range 20.7–87.1 yr), male/female ratio was 1:1.3, and median follow up was 4.0 yr (range 0.1–15.3 yr). The mean Breslow thickness was 3.1 mm (range 0.2–13.5 mm). Overall, 47/81 (58%) primary melanoma tissue samples demonstrated positive labeling for p63 (Fig. 6 A). Statistical analysis revealed no significant association between p63 expression of primary tumor with age at diagnosis, gender, or site of melanoma (Table S2). Analyses of histopathological prognostic factors demonstrated significant correlation with Breslow thickness (P = 0.03) but no other features (e.g., histological classification, growth phase, ulceration status, regression, and mitotic rate; Table S2).

For these primary tumors, the mean time to recurrence (n = 7) was 1.60 yr (range 0.04–6.02 yr). Univariate and multivariate analysis demonstrated a trend toward increased recurrence rates in p63-positive tumors but a nonsignificant association (P = 0.197, Wald test; not depicted). Mean time to first metastasis in this cohort (n = 30) was 1.44 yr (range 0.1–4.69 yr). Univariate and multivariate analysis demonstrated an upward trend toward increased metastatic rates in the p63-positive cohort but failed to reach significance (P = 0.105, Wald test; not depicted).

Overall mortality rate for the primary melanoma cohort was 33% (n = 26), and melanoma-specific mortality was 21% (n = 17). Median time to death was 2.6 yr (range 0.5–6.2 yr). Univariate and multivariate analysis using Cox regression analysis for primary tumors demonstrated p63 status to be a significant predictor of worse melanoma-specific outcomes in the univariate analysis (hazard ratio [HR] 3.10; P = 0.04, Cox proportional HR; Table 1 and Fig. 6 B).

Overall, 19 recurrent melanoma samples and 56 metastatic melanoma tumor samples from 49 individuals (22 males and 27 females) were analyzed. The median age at diagnosis was 60.6 yr (range 30.6–107.6 yr). The median follow up for this cohort was 1.52 yr (range 0.04–8.07 yr). The proportion of p63-positive tumors comprised 10/19 (53%) of recurrent tumors (Fig. 6 C) and 37/56 (66%) of metastatic tumors (Fig. 6 D). Statistical analysis revealed no significant association between p63 expression of metastatic tumor with age at diagnosis, gender, or site of metastasis (Table S3).

Overall mortality for recurrent and metastatic tumors was 45% (n = 22), and melanoma-specific mortality was 37% (n = 18). Median age of death was 62.4 yr (range 30.7–86.7 yr), with median time to death from first recurrence/metastases being 1.25 yr (range 0.08–6.43 yr). Univariate and multivariate analysis using Cox regression analysis for metastatic tumors demonstrated that p63 status was a significant predictor of death from melanoma (multivariate analysis HR 3.86 [1.31–11.39]; P = 0.01; Table 2 and Fig. 6 E).

TP63, the master initiator of epithelium stratification, is often dysregulated in cancer progression (Koster et al., 2006; Mills, 2006; Marchini et al., 2008). To date, the expression pattern of p63 isoforms has not been robustly investigated in the melanocyte lineage. Q-PCR analysis, which displays increased sensitivity for detecting low expression levels of isoforms, demonstrated infrequent expression of TP63 isoforms in primary human melanocyte cultures but significantly elevated expression in >70% of established melanoma cell lines. In contrast with previous immunohistochemistry studies, which showed little or no expression of p63 protein (Brinck et al., 2002; Di Como et al., 2002; Reis-Filho et al., 2003b; Dotto and Glusac, 2006; Bourne et al., 2008; Morgan et al., 2008; Sakiz et al., 2009; Kanner et al., 2010), our data provide the first evidence of marked up-regulation of both p63 mRNA and protein in melanoma cell lines.

In an attempt to characterize the functional role of p63 in melanoma, we demonstrate stabilization of endogenous TA and ΔN p63 isoforms in both nuclear and mitochondrial compartments of melanoma cells in response to genotoxic stress. Depletion of p63 by RNA interference (RNAi) revealed that the expression of TA and/or ΔN p63 isoforms confers resistance to chemotherapy in melanoma cell lines and provides evidence for an oncogenic role of p63 in this tumor. We have demonstrated through immunoprecipitation experiments that p53 and p63 can interact and have shown that stabilization of both p63 and p53 occurs in the nucleus and both are able to translocate to the mitochondria upon genotoxic stress. The mechanism for the latter process appears to be one of codependence, whereby depletion of one protein limits translocation of the other. We propose a mechanism that supports an interaction between p63 and p53 in melanoma cells, whereby depleting p63 results in activation of the p53 mitochondrial apoptotic pathway, rendering melanoma sensitive to both standard chemotherapies and novel BRAF inhibitors. Our data suggest a putative antiapoptotic role for p63 by repression of nuclear p53, which may be one explanation for the failure of p53 apoptotic function in melanoma.

To support these findings, we have provided the first evidence that a significant proportion of cutaneous melanoma tumors express p63 protein. These data demonstrate nuclear and cytoplasmic expression of p63 in 40–63% of melanoma tissue samples, depending on stage of disease progression. Analysis of the p63-positive population demonstrated a significant association with Breslow thickness of primary melanoma and a positive trend toward shorter time to recurrence and to metastases. Moreover, individuals with p63-postive primary and metastatic tumors had significantly worse disease-specific outcomes.

This study has demonstrated the endogenous expression of p63 in melanoma cell lines and melanoma tissue samples using a variety of methods and demonstrated a physiological role for this gene in melanoma. Although physical localization of p63 to the mitochondria has not been previously reported, here we have shown using a range of methods (immunofluorescence, electron microscopy, Western blot of subcellular fractionation, and FACS quantification analysis) that translocation of both TA and ΔN p63 to the mitochondria occurs relatively rapidly compared with a more gradual stabilization of the protein in the nucleus, suggesting involvement of p63 in the mitochondrial apoptotic pathway, while still retaining nuclear transcriptional activity. This is the first study to demonstrate p63 relocation to the mitochondria but also to quantify the stabilization in various subcellular compartments using a novel flow cytometry fractionation technique.

Posttranslational modification analysis of p63 revealed that phosphorylation occurs within 3 h of exposure to genotoxic agents, predominantly in the nucleus but also in the mitochondria. This may affect stability of the protein and/or its transactivation abilities. Phosphorylation of p63 in keratinocytes has no effect on subcellular localization (Westfall et al., 2005), whereas it has been shown recently that phosphorylation of p53 may have a role in targeting the protein to the mitochondria (Nemajerova et al., 2005; Mancini et al., 2009). Based on these data, we speculate that, analogous to p53, phosphorylation of p63 may contribute to the stability of the protein in the nucleus and may also assist in targeting the protein to the mitochondria. Further work is required to characterize targeting of p63 to the mitochondria.

p63 has been linked to chemosensitivity in several tumors: in head and neck squamous cell carcinoma, ΔNp63 was a key determinant of therapeutic response (Rocco et al., 2006), and expression correlated with clinical response to cisplatin (Zangen et al., 2005); in breast cancer, p63 expression positively correlated with chemosensitivity to cisplatin (Rocca et al., 2008); and in hepatoma cells, transfection of TAp63α sensitized cell lines to chemotherapy and transient depletion led to chemoresistance (Gressner et al., 2005). Studies have reported p63 to be a biomarker for poor prognosis and cancer progression in breast cancer and follicular cell lymphoma, respectively (Ribeiro-Silva et al., 2003; Fukushima et al., 2006). Although p63 has been reported to primarily be a nuclear protein (el-Deiry et al., 1995; Dellavalle et al., 2001; Di Como et al., 2002), aberrant cytoplasmic expression is reported in lung and prostate cancers (Narahashi et al., 2006; Dhillon et al., 2009) and was associated with increased prostate cancer–specific mortality, reduced apoptosis, and higher proliferative activity, suggesting an oncogenic role in prostate cancer progression and survival (Dhillon et al., 2009). Our data support these findings in melanoma. The heterogeneity of this cancer raises the possibility that such a finding could provide a novel therapeutic approach aimed at counteracting p63 expression in melanoma cells and consequently sensitizing melanoma to standard chemotherapeutic agents.

Our data propose a physiological interaction between p53 and p63, providing an explanation for the failure of p53-mediated apoptosis in melanoma, in particular in response to chemotherapeutic agents. Although both TA and ΔNp63 isoforms can act as transcription factors when homodimerized, the ΔN isoforms can heterooligomerize with the TAp63 isoforms (as well as with other p53 family members) and modify their activity in vitro (Hibi et al., 2000; Choi et al., 2002; Serber et al., 2002; Chan et al., 2004). This could be a result of heterooligomer formation or co-translocation by molecular chaperones recognized to transport p53, e.g., hsp90 or hsp70 (Walerych et al., 2004, 2009; Whitesell and Lindquist, 2005).

Endogenous p63 prevents nuclear p53 stabilization after a stress response. This effect could be caused by either inhibition of p53 entry into the nucleus or, more likely, an inhibitory effect on stability of p53 through induction of ubiquitin ligases, e.g., MDM2, which is often overexpressed in melanoma and recognized as one of the main causes of p53 inactivity in melanoma cells (Polsky et al., 2002; Muthusamy et al., 2006; Ji et al., 2012). Reactivating p53 to restore its tumor-suppressive capacity has received great attention, with emphasis on reduction of mdm2 to achieve this (Smalley et al., 2007; Gomez-Monterrey et al., 2010; Michaelis et al., 2011; Verhaegen et al., 2012). Recently, elevated expression of MDM4 was demonstrated to be a key determinant of impaired p53 function in melanoma (Gembarska et al., 2012). Although previous in vitro data indicate that both MDM2 and MDM4 could not repress transactivation of p63 isoforms (Little and Jochemsen, 2001), our results suggest otherwise in melanoma. Our data propose a role for p63 in modulating expression of p53, MDM2, and MDM4 through depletion of MDM2 (possibly through degradation) and MDM4, thus increasing p53-mediated chemosensitivity in melanoma. We have demonstrated that depletion of p63 significantly increases nuclear expression of p53 and mdm2 in conjunction with total MDM4 protein levels. Although we cannot definitively confirm whether p63 directly represses MDM2 and MDM4 in addition to p53, or whether this is an indirect effect caused by their feedback loop with p53 (Li et al., 2010), these data support a role for p63 to disrupt the p53–MDM2 and p53–MDM4 axis in melanoma with consequent reactivation of chemosensitivity. We propose that expression of p63 in melanoma might be considered when designing novel targets aimed at disrupting the p53–MDM2 and/or p53–MDM4 interactions.

In summary, we have demonstrated the first evidence of a biological role for p63 in melanoma cells supporting preferential up-regulation of TP63 over p53 mutations, whereby p63 acts to negatively regulate p53 function. In response to DNA damage, p63 is stabilized in the nuclei and translocates to the mitochondria where it acts to repress apoptosis through direct and indirect inhibitory function on proapoptotic effectors in the p53 apoptotic pathway. These data unexpectedly implicate both TA and ΔN p63 isoforms in mediating chemoresistance in melanoma and impact novel therapeutic strategies used to treat melanoma.

33 established melanoma cell lines and 8 human primary melanocytes cultures were profiled (Tables S4 and S5). Human melanoma cell lines WM35, WM1575, WM1552C, WM793, WM115, WM278, WM1158, WM9, WM852, WM983B, and WM239A were donated by M. Herlyn (Wistar Institute, Philadelphia, PA). SBCl2 RGP-like cells were a gift of B.C. Giovanella (Stehlin Foundation for Cancer Research, St. Joseph Hospital, Houston, TX). LM3, LM4, LM5, LM32, LM34, and LM37 were provided by M. Rodolfo (Fondazione IRCCS Istituto Nazionale Tumori, Milan, Italy). VMM39, Mel 224, Mel 505, Mel 501, SK-MEL 5, and SK-MEL 24 were provided by T. Crook (University of Dundee, Dundee, Scotland, UK). DX3, DX3LT5.1, and A375M were donated by J.F. Marshall (Queen Mary University of London [QMUL], London, England, UK). C8161 cells were a gift from M.J.C. Hendrix (Children’s Memorial Research Center, Chicago, IL). SK-MEL 31, HBL, and MALME-3M were provided by J. Vachtenheim (University Hospital Prague, Prague, Czech Republic) and by G.E. Ghanem (University of Brussels, Brussels, Belgium). p53-null MM-AN cells were a gift from B.A. Gilchrest (Boston University School of Medicine, Boston, MA), whereas UISO-mel-6 were provided by T.K. Das Gupta (University of Illinois at Chicago, Chicago, IL). Detailed methods for cell culture and conditioned media are provided in Table S5. HEK293 cells were cultured in DMEM supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin.

The antibodies used were GAPDH and APAF1 from Abcam; anti-p63 4A4 from Neomarkers (for Western blot) and H-129/H137 from Santa Cruz Biotechnology, Inc. (for immunofluorescence); anti-BAK from EMD Millipore; anti–cytochrome c from BD; anti–Bcl-XL (H-62), anti–MDM-2 (C-18/N-20), anti-Bcl2 (C-8), and anti-p53 (DO-1) from Santa Cruz Biotechnology, Inc.; anti–phospho-p53 (ser15), anti–phospho-p53 (ser46), anti–Mcl-1, and anti-Puma from Cell Signaling Technology; and anti-Hdmx/MDM4 from Bethyl Laboratories, Inc.

p63 plasmids were donated by G. Melino (University of Leicester, Leicester, England, UK). HEK293 cells were transfected with Lipofectamine 2000 reagent (Invitrogen) at a 1:2 ml/mg ratio with DNA using 5 mg plasmid DNA. Cells were treated with UV-B as indicated in the figure legends. Cisplatin, etoposide, doxorubicin, dacarbazine, and paclitaxel were used as indicated in the figure legends.

Cells were washed in PBS and then lysed using lysis buffer (1 M Tris, 2.5 M NaCl, 10% glycerol, 0.5 M glycerophosphate, 1% Tween 20, 0.5% NP-40, and 1× EDTA-free Complete Protease Inhibitor tablet [Roche]) for 15 min on ice. Extracts were separated on SDS 10 or 12% polyacrylamide gels and transferred to a nitrocellulose transfer membrane (Whatman). The blots were incubated with the specific antibodies and developed according to the manufacturer’s instructions (ECL plus; GE Healthcare).

This was adapted from published methods (Mihara and Moll, 2003; Arnoult, 2008). Three 10-cm dishes of cells at a confluency of 70–80% were required to obtain sufficient levels of mitochondrial protein. All steps were performed on ice or at 4°C. Each plate was washed once with 1 ml ice-cold PBS/1 mM EDTA before scraping the cells into a 15-ml falcon tube. Cells were pelleted at 750 g for 5 min and then washed in 3 ml ice-cold PBS. The cells were pelleted again for 5 min at 750 g. The pellet was then resuspended in 400 µl of cold mitochondrial isolation buffer (MIB) with protease inhibitor cocktail (PIC; 1:20). Cells were transferred to an ice-cold Dounce homogenizer (Wheaton) and homogenized with a minimum of 150 strokes, while monitoring under the microscope. Trypan blue staining was used to confirm cell membrane disruption. The solution was transferred to an Eppendorf and centrifuged for 5 min at 800 g to isolate the nuclear fraction. The pellet was used for nuclear isolation. The supernatant was centrifuged for 30 min at 10,000 g where the resulting pellet contained the mitochondrial fraction. The supernatant containing cytoplasmic proteins was added to the Eppendorf containing the nuclear fraction that was to be used in the nuclear isolation protocol. The mitochondrial pellet was washed in 500 µl of 1× MS buffer (+PIC) and centrifuged again at 10,000 g for 15 min. The mitochondrial fraction was resuspended in 50 µl MIB with 1% Triton-X 100. Protein concentration was determined and assayed using SDS-PAGE.

The nuclear pellet and cytoplasmic supernatant from the mitochondrial extract method were resuspended and centrifuged at 900 g for 5 min at 4°C. The supernatant was discarded. The pellet was resuspended in 300 µl nuclear isolation (NI) lysis buffer supplemented with PIC and incubated on ice for 15 min. This was centrifuged at 3,500 rpm for 10 min at 4°C. The supernatant was transferred to a new Eppendorf-labeled cytoplasmic fraction. The pellet was resuspended again in 300 µl ice-cold NI lysis buffer (+PIC) and centrifuged at 3,500 rpm for 10 min at 4°C, and supernatant was transferred to the cytoplasmic fraction Eppendorf. The nuclear pellet was washed in 1 ml ice-cold Tris-EDTA twice and centrifuged at 3,500 rpm for 10 min at 4°C between each wash. The supernatant was discarded each time. The pellet was resuspended in 300 µl ice-cold TGN lysis buffer (+PIC) and incubated on ice for a total of 30 min, vortexing every 10 min. The sample was centrifuged again at 15,000 rpm for 30 min at 4°C. The protein concentration of the supernatant containing nuclear proteins was determined and assayed using SDS-PAGE.

Cells were lysed in NP-40 lysis buffer and precleared with protein G beads for 1 h at 4°C. The protein concentration was determined, and then 1 mg of the extract was incubated either with p53 or control IgG antibody prebound to protein G beads for 4 h or overnight at 4°C. The beads were washed twice in NP-40 lysis buffer and twice in 100 mM NaCl, 1 mM EDTA, and 10 mM Tris, pH 8. The immunoprecipitation beads were mixed with 5× Laemmli buffer and loaded onto an SDS–polyacrylamide gel. The gels were transferred (wet) to nitrocellulose membranes, and the resulting blots were incubated with specific antibodies and developed according to the manufacturer’s instructions (ECL plus).

RNA was extracted from keratinocytes, fibroblasts, melanocytes, and melanoma cells by using the RNeasy kit (QIAGEN). A total of 500 ng RNA was used for using the SuperScript III RT kit (Invitrogen).The primers used for the PCR were as follows: ΔNp63 forward, 5′-GGAAAACAATGCCCAGACTC-3′; and reverse 5′-GAAGGACACGTCGAAACTGTG-3′; TAp63 forward, 5′-GGTGCGACAAACAAGATTGAG-3′; and reverse, 5′-GAAGGACACGTCGAAACTGTG-3′; TP53 forward, 5′-GTCACTGCCATGGAGGAGCCGCA-3′; and reverse, 5′-GACGCACACCTATTGCAAGCAAGGGTTC-3′; GAPDH forward, 5′-CTCCTCCACCTTTGACGCTG-3′; and reverse, 5′-CCACCCTGTTGCTGTAGCCA-3′; and GUS forward, 5′-AAACGATTGCAGGGTTTCAC-3′; and reverse, 5′-CTCTCGTCGGTGACTGTTCA-3′.

Q-PCR reactions were set up in 96-well plates using Brilliant II SYBR Green QPCR Master Mix (Agilent Technologies). This Master Mix includes a SureStart Taq DNA polymerase with hot start capability and contains MgCl₂ at a concentration of 2.5 mM in the 1× solution. A passive reference dye (ROX) was added to the mix to compensate for non-PCR–related variations in fluorescence. 25-µl reactions were set up. Experimental reactions were performed in triplicate, and duplicate no-template controls were also run. Once sample cDNA was added, the 96-well plate was briefly centrifuged to remove bubbles and ensure mixing. Data were collected by running a three-step cycling protocol using the AB7500 Fast Real-time PCR System (Applied Biosystems). The temperature cycler was set to detect and report fluorescence during the annealing and extension step of each cycle. Formation of nonspecific products was checked using gel analysis. TP63 gene expression levels in samples were standardized to the housekeeping gene, β-glucuronidase (GUS). GUS was chosen because of its stable expression in melanoma cell lines and because it was a medium-expressed gene that was appropriate because of the medium-low expression levels detected for ΔNp63 in RT-PCR experiments. TP63/GUS expression levels in samples were compared with the mean of gene expression levels in the five primary melanocyte cultures to determine the extent of up-regulation of p63 in melanoma cell lines considering a threefold increase in gene expression as a stringent cut-off for significant up-regulation.

RNA was extracted from melanoma cells by using the RNeasy kit, and cDNA was obtained by RT using the SuperScript III RT kit (Invitrogen) according to the manufacturer’s instructions. B-RAF (exon 15 for mutations at codon 600) and N-RAS (exons 1–2 for mutations at codons 12, 13, and 61) genes were amplified by PCR with BIOTAQ DNA Polymerase (Bioline) and sequenced with different primers pairs: B-RAF forward, 5′-GCACAGGGCATGGATTACTT-3′; and reverse, 5′-ATTAATCTCTTCATGGCTTT-3′; and N-RAS forward, 5′-CGCCGACTGATTACGTAGC-3′; and reverse, 5′-TCGCCTGTCCTCATGTATTG-3′.

PCR products were sequenced using the BigDye Terminator version 3.1 Cycle Sequencing kit (Applied Biosystems). 0.3 µl PCR product was added to a reaction containing 1 µl BigDye terminator mix (version 3.1; Applied Biosystems), 1 µl of 10 µM primer, and 3 µl Better Buffer (Microzone) made up to 11.5 µl with dH₂O. The following PCR cycle for Sanger sequencing was performed: 95°C for 10 s (25 cycles), followed by 58°C for 15 s (25 cycles) and 60°C for 4 min (25 cycles). Sequencing reactions were transferred to plates and precipitated with 2.5 µl of 125 mM EDTA and 30 µl ice-cold 100% ethanol, left on ice for 10 min, and centrifuged at 4,400 g for 20 min. Plates were inverted and centrifuged at 135 g for 10 s to remove the supernatant. The pellet was washed with 135 µl of 70% ethanol and centrifuged for 5 min, and the supernatant was removed as before. Samples were dried in a heating block at 95°C for 10 s. Before being loaded on the sequencer, samples were resuspended in 10 µl HiDi formamide, transferred to an ABI plate (both Applied Biosystems), and denatured at 95°C for 3 min. Sequencing reactions were run on a PRISM 3130xl Genetic Analyzer (Applied Biosystems). Samples were viewed and analyzed using Chromas software (Technelysium Pty Ltd).

Total RNA was extracted from 8 primary RGP malignant melanomas, 15 primary vertical growth phase malignant melanomas, and 5 melanoma metastases and processed as described in Scatolini et al. (2010). Commercial RNA (BDTM Human Universal Reference Total RNA; Takara Bio Inc.) was used as baseline in comparative hybridization experiments on oligonucleotide glass arrays (Human Whole Genome 44K Oligo Microarray Array; Agilent Technologies) using a dye-swap duplication scheme. Single channel intensities were then processed and combined with the Resolver SE System (Rosetta Biosoftware). Mean expression intensities and SEs for the TP63, TP53, and TP73 genes in the three sample classes were retrieved, as shown in Fig. 1 D.

Cultured cells were plated onto glass coverslips (confluency of 150,000 cells/cm²) in 12-well culture plates and allowed to attach overnight at 37°C. Media was discarded and replaced with media containing MitoTracker Orange (dilution 1:10,000). Cells were incubated with MitoTracker Orange for 30–40 min before washing twice in PBS followed by treatment with either UVB or pharmacological drug. Cells were washed in PBS for 5 min at 5–24 h after treatment, fixed in 1 ml of 4% formaldehyde/PBS for 10 min at room temperature, washed twice in fresh PBS for 5 min, and stored at 4°C in PBS. PBS was removed from the cells, and 1 ml of 0.1% Triton X-100 in PBS was added for 3 min at room temperature to allow permeabilization of the cell membranes. Cells were washed twice with PBS for 10 min to remove residual detergent. To avoid nonspecific reaction with the secondary antibody, cells were incubated with 500 µl of 5% goat serum/PBS for 30 min. After removal of the serum, cells were incubated with primary antibody diluted in 5% goat serum/PBS overnight at 4°C. Cells were washed three times for 10 min in PBS and incubated with secondary antibody for 1 h at room temperature in the dark. Cells were washed again three times for 10 min in PBS and incubated with 500 µg/ml DAPI (Invitrogen) for 10 min, followed by two further washes in PBS for 10 min at room temperature. Coverslips were mounted onto a glass slide using Vectashield mounting medium (Vector Laboratories) to prevent photobleaching over time. Mounted slides were sealed with clear colorless nail varnish and stored at 4°C protected from light.

Immunohistochemical expression of p63 was investigated in an mTMA (Biomax US ME481) comprising a panel of 8 normal skin cores and 40 melanoma cores, each of 1.5-mm diameter and 5 µM. The slide was initially baked at 60°C for 2 h. Tissue sections were subjected to deparaffinization and rehydration using a xylene and ethanol (EtOH) series: (a) xylene: ×2 at 5 min each, (b) 100% EtOH: twice at 5 min each, (c) 90% EtOH: once at 3 min, (d) 70% EtOH: once at 3 min, (e) 50% EtOH: once at 3 min, (f) distilled water: once at 5 min, and (g) PBS: once at 5 min. The slide was incubated in citrate buffer, pH 6, placed in a microwave at 300 W for 5 min for three cycles, and subsequently cooled for 15 min. It was returned to the microwave for 5 min at 300 W, cooled to room temperature, and incubated in 5% goat serum/PBS for 2 h at room temperature. The slide was placed in a humid chamber and incubated with the primary antibody (1:50 dilution of H137 and H129 anti-p63 antibodies) overnight at 4°C. It was washed in PBS buffer (three times for 10 min) and incubated with the secondary antibodies at room temperature in 5% goat serum/PBS for 1 h. It was washed in PBS (two times for 10 min) followed by 10 min in DAPI in 500 µg/ml PBS and washed (two times for 10 min) in PBS to remove excess DAPI. The slide was mounted using Vectashield to prevent bleaching of fluorophores. HMB-45 was used and is a highly specific marker routinely used in the diagnosis of primary and metastatic melanoma cells to confirm melanoma cells (Colombari et al., 1988; Baisden et al., 2000; Yaziji and Gown, 2003).

For p63 knockdown, A375M and WM1158 cell lines were transfected with a SMARTpool of three siRNAs targeting TP63 (ID 217143, ID 4893, and ID 217144; Applied Biosystems). Transfection was performed according to the manufacturer’s protocol and optimized for a 6-well plate. Cells were plated at 50% confluency subjected to transfection the following day using HiPerFect (QIAGEN) transfection reagent and 60 nM final concentration of each siRNA. Transfection media was replaced with complete RPMI-1640 media after 24 h. p63 protein expression was analyzed by Western blot at 48 h after transfection. Cells incubated with the transfection reagent only (control) as well as cells transfected with a pool of nontargeting siRNAs (siCONTROL nontargeting siRNA pool) were used as negative controls. For stable infection with shRNA, the pSUPERIOR.retro.puro vector constructs each containing one of three annealed oligonucleotide sequences targeting p63 (clones 1, 2, and 3) were transfected into the packaging cell line Phoenix cells before introducing the retrovirus into the melanoma cell lines. Infected cells were selected using 0.9–1.25 µg/ml puromycin to establish a stable cell line for shRNA expression that was transcribed in cells from a DNA template as a single-stranded RNA molecule. Effects in p63 mRNA were subsequently analyzed. For p53 knockdown, A375M cells were transfected with the stealth RNAi siRNA for p53 (ID 000546; Invitrogen; Fenouille et al., 2011). Transfection was performed according to the manufacturer’s instructions with the Lipofectamine RNAiMAX transfection reagent (Invitrogen).

Translocation of p63 and p53 between subcellular compartments was quantified by assessing p63 or p53 fluorescence intensity using a novel flow cytometry technique (Leverrier et al., 2007). The advantages of this method include increased sensitivity, requirement for fewer cells, marginal spillover between cellular compartments, and reproducibility, which allows quantification of relative protein concentration within a cellular compartment. Moreover, data from quantification experiments were validated using confocal microscopy to confirm purity of the fractions (Fig. 3 D). Three different fluorescent stains were used: MitoTracker Orange (CMTMRos; Molecular Probes) for the mitochondria, Hoechst (Bisbenzimide Hoechst 33342; Sigma-Aldrich) for the nucleus, and secondary antibodies conjugated to Cy5 (Cy5-conjugated AffiniPure F(Ab)₂ fragment goat anti–mouse IgG; Jackson ImmunoResearch Laboratories, Inc.) for p63 or p53 labeling. Live melanoma cells were incubated with MitoTracker Orange to fluorescently label mitochondria, and cells were labeled with anti-p63 or IgG mouse isotype primary antibody and Cy5-conjugated anti–mouse secondary antibody followed by Hoechst (to label the nuclei). The anti-p63 antibody used (4A4) detected all isoforms of p63. Intact A375M cells, characterized by their double positivity to Hoechst and MitoTracker Orange labeling, were analyzed. Cells were disrupted using a Dounce homogenizer, and the resulting homogenates were reanalyzed by flow cytometry. For each cell line, preliminary control experiments included MitoTracker Orange or Hoechst fluorescently labeled cells only, confirming location of the fractions in the homogenized sample. Hoechst-positive and MitoTracker Orange–negative populations were defined as free nuclei; Hoechst-negative and MitoTracker–positive populations were defined as free mitochondria ,and the final region gated comprised intact cells. After gating for whole cell, mitochondrial, and nuclear populations, fluorescence intensity of p63-Cy5 was analyzed using the FlowJo software (Tree Star). To establish purity of the nuclear and mitochondrial fractions, homogenized cells were sorted according to Hoechst and MitoTracker Orange labeling using the FACSAria cell sorter (BD). Sorted fractions were reanalyzed to confirm purity of each fraction (Fig. 3 D). Sorted mitochondrial fractions were centrifuged (13,000 g for 30 min at 4°C), and the supernatant was discarded. Mitochondrial isolates fluorescently labeled with MitoTracker Orange were mounted and visualized using confocal microscopy. The nuclear fraction was prepared in the same way except centrifugation was performed at 4,500 rpm for 15 min at 4°C before mounting isolates onto a slide. Imaging of the intact cells, nuclei, and mitochondria by confocal microscopy validated the analysis gates used to separate fractions comprising of pure nuclei and mitochondria and not fragments of cells. Further verification of this technique included detection of p63 protein by visualizing p63-Cy5 in mitochondrial and nuclear fractions using confocal microscopy after fluorescence-activated cell sorting (FACSAria flow cytometer; BD; Fig. 3 E). An LSRII (BD) fitted with 488-, 405-, 350–360-, and 633-nm lasers with FACS Diva software version 4.1.2 was used to acquire 30,000 whole cells or 1 million fractionated events, using the Argon 488-nm laser line and a 575/26-nm band pass filter to detect MitoTracker Orange; the UV 350–360 laser and a 440/40-nm band pass filter were used to detect Hoechst 33342; the Red HeNe 633-nm line and 660/20-nm band pass filter were used to detect Cy5. No compensation was required because of the use of separate laser lines to detect fluorophores, which did not display any spectral cross talk.

The Annexin V assay was used to detect apoptotic cells. Cells were treated with chemotherapeutic agents for 16–24 h, with or without prior DNA transfection or shRNA infection. Cells were trypsinized, and all cells (living and dead) were pelleted. Cells were resuspended in 400 µl of 1× Binding Buffer (BD; 10× binding buffer, 0.1 M Hepes, pH 7.4, 1.4 M NaCl, and 25 mM CaCl₂). 5 µl Annexin V–FITC (BD) was added and incubated in the dark at room temperature for 15 min. Samples were analyzed on the flow cytometer (LSRII) within an hour. Viability dye (5 µg/ml propidium iodide [PI]) was added just before reading. The following controls were used to set up compensation and quadrants: unstained cells, cells stained with Annexin V–FITC only (no PI), and cells stained with PI only (no Annexin V–FITC). Apoptotic cell populations included Annexin V–FITC positive, PI negative and Annexin V–FITC positive, and PI positive (right top and bottom quadrants of scatter plot). Results were analyzed using the FACSDiva software (BD) and expressed as mean ± SEM values of three independent experiments performed in triplicate.

The Tokuyasu method was used for immunogold localization of p63 (Tokuyasu, 1980). Cells were seeded at 70% confluency in two 100-mm plates and treated with 2 µM paclitaxel for 6 h. 8% formaldehyde + 0.2% glutaraldehyde in 100 mM Pipes buffer, pH 7.2, was added to the media (1:1), and plates were refrigerated (1 h). Cells were scraped into an Eppendorf tube and centrifuged (1,500 rpm for 5 min). The supernatant was discarded, and cells were resuspended in Pipes buffer for transport. The pellet was washed twice in Pipes buffer, and supernatant was discarded, ensuring removal of fixative. The pellet was resuspended in 10% warm gelatin and centrifuged (6,000 rpm for 3 min). To solidify, the pellet of cells in gelatin was incubated at 4°C (30 min). Gelatin containing the cells was cut into small cubes ∼1 mm³, which were incubated in 2.3 M sucrose at 4°C overnight for cryoprotection. Cubes were mounted on specimen pins, excess sucrose was blotted away, and the cubes were cryofixed by plunging into liquid nitrogen. Cryosections, 90-nm thick, were cut from these blocks with glass knives at −100°C using an MTXL ultra-microtome with a CRX cryo-adaptation (RMC). Groups of sections were picked up on a droplet of 2.3 M sucrose and transferred onto Pioloform-coated nickel grids. Grids were floated onto standard buffer consisting of PBS + 1% BSA + 0.1% sodium azide, pH 7.4. Grids were incubated (three times for 5 min) on PBS–0.05 M glycine to remove remaining aldehyde groups persisting from the fixation process before being washed in incubation buffer (PBS + 0.1% BSA-c [Aurion] + 0.1% azide, pH 7.4) and transferred onto droplets of primary antibody (anti-p63 antibodies H129, 4A4, and H137 [Santa Cruz Biotechnology, Inc.] and anti-mtHsp70 [Grp75; Abcam]) diluted in incubation buffer (1:50) for overnight incubation at 4°C. The primary antibody was removed by passing over six droplets of incubation buffer solution, 5 min per droplet, before incubation for 1 h at room temperature with the secondary antibody, diluted in incubation buffer (1:100). Secondary antibody was removed by passing grids over three droplets of PBS (5 min), and the reaction was fixed by exposing grids to PBS–2% glutaraldehyde. Grids were washed over distilled water (three times for 5 min). Sections were subsequently embedded in 2% methylcellulose/3% uranyl acetate solution in a 9:1 ratio. Sections were examined, and micrographs were obtained using a T12 transmission electron microscope (FEI) at an accelerating voltage of 120 kV.

Clinicopathologically characterized, formalin-fixed paraffin-embedded material representing benign melanocytic nevi and all stages of melanoma development (primary melanoma, recurrent melanoma, and metastases) were analyzed for expression of p63 (Ethical approval number 07/QO604/23 by the East London & The City HA Local Research Ethics committee 2, UK). The effect of storage of tissue samples on p63 reactivity was notable; with increased storage times, a progressive and significant decrease in intensity of p63 staining was observed, with evidence of significant decline occurring as early as 2 wk of storage (Hameed and Humphrey, 2005; Burford et al., 2009). We also observed this effect and therefore, for optimal detection of p63 in tissue samples, staining was undertaken within 2 wk of sectioning of tumor samples.

Melanoma tissue samples representing different stages of disease progression were selected from the archive of paraffin-embedded tissue samples at Barts and the London National Health Service Trust. Clinical data were collected from patients attending the skin cancer multidisciplinary clinic and through accessing electronic medical records. Pathology reports confirmed the diagnosis of melanoma in all cases, and slides were subsequently reviewed by a dermatopathologist (R. Cerio). Immunostaining was undertaken with a Vector RTU system (red). Initial optimization was performed using anti-p63 antibodies H129 and 4A4 (dilution 1:100). Vector VIP (SIC-4600) was applied for 20 min, and a green counterstain was used to distinguish the red stain clearly. As both anti-p63 antibodies demonstrated similar p63 reactivity, the H129 antibody (routinely used for the analysis of clinical samples and for clinical trials by the Barts Pathology Service) was used for all tumor samples.

Positive reactivity to p63 in basal keratinocytes in the epidermis or ductal/glandular epithelia in the dermis of skin samples was used as a positive internal control. Negative controls comprised omission of the primary antibody. Positive reactivity of p63 in melanoma tumor samples was confirmed when melanoma cells showed staining of nucleus and/or cytoplasm. Nevi and melanoma sections labeled with p63 were reviewed by three dermatologists (R.N. Matin, C.A. Harwood, and S. Law Pak Chong) and one dermatopathologist (R. Cerio) blinded to clinical outcomes. Given the heterogeneity displayed by melanoma cells, only tumors demonstrating >50% positive reactivity of melanoma cells were included in the statistical analysis.

Statistical analysis was undertaken in collaboration with D. Mesher (Centre for Cancer Prevention, QMUL). Clinicopathological variables were tabulated and compared with p63 status. A power calculation was performed using melanoma-specific death as the primary end point. For 80% power, 200 melanomas from individuals with follow up over 5 yr needed to be analyzed to detect a 5% significant difference in outcomes between p63-positive and -negative tumors. For survival (overall and melanoma-specific analysis), patients were censored at date of death or date of last follow up. For recurrence and metastases, patients were censored at date of recurrence/metastases, date of death, or date of last follow up. Analysis of end points was performed by Cox proportional hazards model. Where non-independent observations were included in Cox regression analysis (i.e., multiple tumors from the same patient), robust SEs were used to adjust for this using the cluster option in Stata. Variables were examined univariately, and subsequently a multivariate model was constructed using backward stepwise selection method. P-values were presented comparing the multivariate model with and without p63 status included. All p-values were two-sided, and all statistical analyses were performed using Stata 10.0.

Table S1 provides statistical analysis of demographic data for p63-expressing primary melanoma tissue samples (mTMA). Tables S2 and S3 show the clinicopathological features of primary and metastatic melanoma archival tissue samples, respectively. Table S4 summarizes the mutational status of p53, BRAF, and NRAS in melanoma cell lines used in the study. Table S5 shows the culture media and supplements for the melanocytes and established melanoma cell lines used in the study.

We are grateful to Dr. Gary Warnes, Flow Cytometry Core Facility Manager, for technical advice, to Dr. Claire Scott for the BRAF and NRAS mutation sequencing of the cell lines, and to the core Pathology laboratory for the technical support with the tissue sample immunostaining at the Blizard Institute. We also thank Dr. Alice Warley (Centre for Ultrastructural Imaging CUI at King’s College London, London, England, UK) for technical advice with transmission electron microscopy methods.

Funding support was provided by the Medical Research Council, Cancer Research UK, and Barts and The London Charity.

The authors declare no competing financial interests.