Distinct TP63 Isoform-Driven Transcriptional Signatures Predict Tumor Progression and Clinical Outcomes Hussein A

Home , TP63

Published OnlineFirst November 27, 2017; DOI: 10.1158/0008-5472.CAN-17-1803 Cancer Tumor Biology and Immunology Research

Distinct TP63 Isoform-Driven Transcriptional Signatures Predict Tumor Progression and Clinical Outcomes Hussein A. Abbas1,2, Ngoc Hoang Bao Bui1,2,3, Kimal Rajapakshe4, Justin Wong3,5, Preethi Gunaratne6, Kenneth Y. Tsai2,7,8, Cristian Coarfa4, and Elsa R. Flores1,2,8

Abstract

TP63 is required to maintain stem cell pluripotency and sup- ing with TAp63 to modulate key biological pathways, principally presses the metastatic potential of cancer cells through multiple cell-cycle regulation, extracellular matrix remodeling, epithelial- mechanisms. These functions are differentially regulated by indi- to-mesenchymal transition, and the enrichment of pluripotent vidual isoforms, necessitating a deeper understanding of how the stem cells. Importantly, these TAp63 and DNp63 signatures prog- distinct transcriptional programs controlled by these isoforms nosticated progression and survival, even within specific stages, in affect cancer progression and outcomes. In this study, we con- bladder and renal carcinomas as well as low-grade gliomas. These ducted a pan-cancer analysis of The Cancer Genome Atlas to data describe a novel approach for understanding transcriptional identify transcriptional networks regulated by TAp63 and DNp63 activities of TP63 isoforms across a large number of cancer types, using transcriptomes derived from epidermal cells of TAp63 / potentially enabling identification of patient subsets most likely and DNp63 / mice. Analysis of 17 cancer developmental and 27 to benefit from therapies predicated on manipulating specific cancer progression signatures revealed a consistent tumor sup- TP63 isoforms. pressive pattern for TAp63. In contrast, we identified pleiotropic Significance: Transcriptomic analyses of patient samples and roles for DNp63 in tumor development and found that its regu- murine knockout models highlight the prognostic role of several lation of Lef1 was crucial for its oncogenic role. DNp63 performed critical mechanisms of tumor suppression that are regulated by a distinctive role as suppressor of tumor progression by cooperat- TP63. Cancer Res; 78(2); 1–12. 2017 AACR.

Introduction perplexing due to their variable expression in tissues, their complex interaction with each other and with TP53, and the lack of TP53, TP63, and TP73 family of genes transcriptionally regulate adequate antibodies to distinguish between their isoforms (3). genomic programs to induce apoptosis, cell-cycle arrest, senes- TP63 is of particular interest due to its high expression in cence, metabolic reprogramming, and stem cell maintenance epithelial tissues (11, 12), its clinical use as a diagnostic marker (1–8). Manipulating specific isoforms of Tp63 and Tp73 is an (13–15), and its role in stem cell maintenance, tumor suppres- effective approach to induce regression in Tp53-deficient and sion, and metastasis (3, 6, 8, 12, 16). Further, its two isoforms, mutant tumors (9). Although TP63 and TP73 roles in cancer have TAp63 and DNp63, exert various modulatory roles in stem cell been described (4, 5, 7, 9, 10), the transcriptional roles of TP63 maintenance, cancer metastasis, and miRNA biogenesis (6–9). and TP73 in tumor development and progression have been Yet, the broad tissue specific roles of the genes regulated by TAp63 and DNp63 in cancer development and progression remain to be deciphered and are important for the proper interpretation of its 1 fi Department of Molecular Oncology, H. Lee Mof tt Cancer Center, Tampa, Florida. use as a clinical diagnostic marker. 2 fi Department of Cutaneous Oncology, H. Lee Mof tt Cancer Center, Tampa, fi Florida. 3Graduate School of Biomedical Sciences, The University of Texas MD In this study, we identi ed the connection between the activity of Anderson Cancer Center, Houston, Texas. 4Department of Molecular and Cellular TP63 isoforms in stem cell maintenance and in regulatory networks Biology, Baylor College of Medicine, Houston, Texas. 5Department of Genetics, that control cancer development and progression. By applying an The University of Texas MD Anderson Cancer Center, Texas. 6Department of integrative model for TAp63 and DNp63 signatures from mice into 7 Biology and Biochemistry, University of Houston, Houston, Texas. Department of The Cancer Genome Atlas (TCGA)derived signatures, wefound that fi 8 Pathology, H. Lee Mof tt Cancer Center, Tampa, Florida. Cancer Biology and TAp63 transcriptional activity is lost in 13 cancer development Evolution Program, H. Lee Moffitt Cancer Center, Tampa, Florida. signatures, indicative of a tumor suppressor pattern, whereas Note: Supplementary data for this article are available at Cancer Research DNp63 activity is pleiotropic in cancer development. TAp63 and Online (http://cancerres.aacrjournals.org/). DNp63 exerted tumor suppressor activity in four cancers via mod- H.A. Abbas and N.H.B. Bui contributed equally to this article. ulation of stem cell, extracellular matrix (ECM), epithelial to mes- Corresponding Authors: Elsa R. Flores, Moffitt Cancer Center, 12902 Magnolia enchymal transition (EMT), and cell-cycle pathways. Interestingly, Drive, Tampa, FL 33612. Phone: 813-745-1473; E-mail: Elsa.Flores@Moffitt.org; we found that the ability of DNp63 to transcriptionally regulate Lef1 and Cristian Coarfa, 1 Baylor Plaza, Baylor College of Medicine, BCM-Cullen dictates its tumor suppressor or oncogenic activity. These signatures Building, Room 450A, MS: BCM130, Houston, TX 77030. Phone: 713-798-7938; are not only predictive of survival across stages, but can stratify Fax: 713-798-2716; E-mail: [email protected] patients within the same stage into different survival groups. These doi: 10.1158/0008-5472.CAN-17-1803 findings can be exploited by further classifying patients into groups 2017 American Association for Cancer Research. that can benefit from TP63-isoform targeted therapy (9).

www.aacrjournals.org OF1

Abbas et al.

Materials and Methods epidermal keratinocytes. We then analyzed TP63 isoform transcriptomes across TCGA tumors. Because stem cell mainte- Bioinformatics analysis nance in epithelial tissues is intimately tied with tumorigenesis, The primary and processed data for 9,578 tumor and 730 our approach integrates pathways in normal stem cell biology normal samples were downloaded from TCGA website between and cancer. July 2014 and February 2015. We aligned all TCGA sequencing We derived tumor development signatures by comparing the data to Genome Reference Consortium Human Reference 38 expression profile of tumors to normal tissues. Tumor progression (hg38) in order to identify all genes and TP63 isoforms (a, b, signatures were derived by comparing expression profile of high g for TAp63 and DNp63). TAp63 signatures were derived from / stage/grade tumors to low stage/grade tumors (Supplementary epidermal cells P1 TAp63 and wild-type mice at embryonic day File S2). We characterized TP63 isoforms transcriptomes in cancer 18.5 (6). For DNp63 murine gene signature, we used previously development and progression by employing the GSEA algorithm published transcriptome profiles (8). For putative tumor suppres- (17) in 17 tumor development and 27 tumor progression signa- sor role of the TAp63 or DNp63, we identified genes that were tures. Identification of putative tumor suppressor and oncogene significantly changing and in the same direction in both TAp63 / / activities of TP63 isoforms in cancer development and progres- (or DNp63 ) and the corresponding tumor signature. For onco- sion was based on significance, directionality, and enrichment genic roles, an inverse approach was taken. Pan-cancer survival concordance. Specifically, when both upregulated and downre- signature (PCSS) was identified separately for each gene signature gulated cancer signature genes are significantly enriched in con- and for each role (e.g., tumor suppressor or oncogene) using cordant direction [positive normalized enrichment scores (NES) association of gene scores and survival that conforms with the for upregulated genes, and negative NES for downregulated predicted function. The highest-ranking genes were then com- / / genes] in TAp63 or DNp63 signatures, this suggests a con- bined to establish a robust PCSS. Z-scores were used for all activity cordance with the TAp63- or DNp63-deficient genotypes, thus a scoring then ranked by the highest and lowest 25th percentile potential tumor suppressor activity. Alternatively, discordant (quartiles). We utilized BP, KEGG, and Reactome compendiums / / enrichment in TAp63 or DNp63 signatures suggests a pre- for pathway analysis. All studies were approved by the Institu- dominant wild-type phenotype, and hence a potentially corre- tional Review Board. All patient studies were conducted in accor- sponding oncogenic activity. We then identified TP63 isoforms- dance with the U.S. Common Rule ethical guidelines. derived signatures to measure corresponding transcriptional activity among patients and across stages, and to assess for survival In vitro assays (Fig. 1A). MCF10CA1D.cl1 breast cancer cell line, human lung squamous TAp63 consistently exhibited a tumor suppressor pattern in cell carcinoma HCC95 cell line, and human kidney clear cell 13 cancer development signatures, whereas DNp63 exhibited a (KIRC) carcinoma Caki-1 cell line were cultured in corresponding pleiotropic role in tumorigenesis: oncogenic pattern in seven media. For Western blot analysis, blots were probed with anti- signatures and a tumor suppressive pattern in three cancer devel- DNp63 (619002; Biolegend), Lef1 (2230S; Cell Signaling Tech- opment signatures (q < 0.0001 for all; Fig. 1B and Supplementary nology), Myc-tag (9B11; Cell Signaling Technology), and Flag-tag Table S1). (A8592; Sigma) overnight at 4C. Actin (A5441; Sigma) was used as loading control. For real-time PCR analysis, total RNA was TAp63 and DNp63 signatures predict patient survival isolated using miRNeasy Mini Kit and cDNA synthesized using To enable pan-cancer clinical assessment applications SuperScript First-Strand Synthesis System (Invitrogen) and informed by TAp63 and DNp63 transcriptional activities, we followed by qRT-PCR ran in triplicates using Taqman Universal refined TAp63 tumor suppressive and DNp63 oncogenic signa- PCR Master Mix (Applied Biosystems). Taqman probes included: tures (Supplementary File 3.1–3.2) based on a survival-predictive human Lef1 (Hs01547250_m1) and human GAPDH model dubbed PCSS (PCSSTAp63 and PCSSDNp63, respectively). (Hs03929097_g1) as internal controls. C values were normalized t PCSSTAp63 activity significantly decreased across various stages I to to GAPDH. For siRNA assays, 40 nmol/L of siLef1 IV cancers including breast (BRCA), KIRC, and kidney papillary (SASI_Hs02_00349169 and SASI_Hs01_00151600; Sigma) was (KIRP), lung adenocarcinoma (LUAD), and lung squamous cell transfected into 0.5 106 cells on 6-cm dishes using Lipofecta- (LUSC) carcinoma (ANOVA <0.0001–0.032) exhibiting a tumor mine RNAi Max (Invitrogen). One microgram of plasmids was suppressor pattern (Fig. 1C). However, PCSSDNp63 activity transfected into 0.5 106 cells using X-tremeGENE HP DNA increased significantly from stage I to IV in BRCA and LUAD transfection reagent. pcDNA3.1-DYK-Lef1 and pcDNA3.1-DYK- (ANOVA <0.0001) patterning an oncogenic role (Fig. 1D). empty plasmids were obtained from Genscript. DNp63a ORF was Patients with increased activity of PCSSTAp63 and PCSSDNp63 had cloned into pcDNA3.1-Myc-empty plasmid. Cells were collected significantly better and worse overall survival, respectively, within 48 hours posttransfection. Cells were stained with Click-iT EdU stages and across all patients (P < 0.0001–0.038; Fig. 1E and F). microplate assay (Invitrogen) and proliferation quantification Corresponding enriched pathways are summarized in Supple- was done using Celigo and Incucyte. For CFU assays, colonies mentary Fig. S1A and S1B and File 3.3–3.4. were stained with 0.005% crystal violet for 1 hour. DNp63 activates an oncogenic program through transcriptional Results regulation of its downstream target, Lef1 Gene set enrichment analysis and signature activity reveal To explore the functional pleiotropic role of DNp63, we first pleiotropic roles of TAp63 and DNp63 in tumor development manipulated DNp63 expression in breast and LUAD cell lines, and progression where we found DNp63 to be oncogenic. Knockdown of DNp63 in We generated TAp63 and DNp63 RNA signatures from primary MCF10-CA1D breast cancer cells, where DNp63 is overexpressed / / TAp63 (Supplementary File S1) and DNp63 (8) mouse (Supplementary Fig. S2A), reduced proliferation and soft agar

OF2 Cancer Res; 78(2) January 15, 2018 Cancer Research

TP63 Isoforms Transcriptional Signatures in Tumorigenesis

BLCA KICH A B TAp63 DNp63 UCEC KIRC

TAp63 THCA DNp63 KIRP

READ –/– –/– TAp63 DNp63 TCGA LGG

Oncogenic pattern PRAD Tumor suppressor pattern Not significant LIHC Insufficient data

PAAD

Summary GSEA ESCA

BRCA

CHOL COAD

STAD LUAD Z-Score HNSC LUSC

Survival Pathway Progression

D CDPCSSTAp63 Activity in tumor development PCSS Np63 Activity in tumor development Activity level Activity level BRCA*

CHOL *ANOVA < 0.05 LUSC LIHC* *ANOVA < 0.05

COAD BLCA READ THCA* ESCA

HNSC LUAD* STAD

KIRC* COAD

KIRP* BRCA* LUAD*

LUSC UCEC Stage I Stage II Stage I Stage III Stage IV Stage II Stage III Stage IV D EFSurvival based on PCSSTAp63 activity Survival based on PCSS Np63 activity STAD UCEC READ PRAD THCA LUSC Pattern Pattern Tumor suppressor Tumor suppressor LUAD Not significant Oncogenic Insufficient data LUSC Not significant LIHC –Log (log rank P) Insufficient data 10 –Log (log rank P) KIRP 1 LUAD 10 2 1 KIRC 2 3 HNSC COAD 3 4 ESCA 4 COAD BRCA CHOL BLCA BRCA Stage I Stage I Stage II Stage II Stage III Stage IV Stage III Stage IV ll Stages All Stages A

Figure 1. TAp63 and DNp63 have tissue-dependent activity in tumor development. A, Working model for identiﬁcation of transcriptional networks enriched in TCGA tumors B, Summary of GSEA concordant ﬁndings across cancer development and cancer progression signatures. C, Heatmap of the activity of PCSSTAp63 across different stages of tumorigenesis. D, Heatmap of the activity of PCSSDNp63 across different stages of tumorigenesis. E, Dot plot of the survival analysis of different tumors with respect to PCSSTAp63 activity. F, Dot plot of the survival analysis of different tumors with respect to PCSSDNp63 activity.

www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 OF3

Abbas et al.

ABCA1D CA1D C CA1D D CA1D

+siC +siDNp63 +siC +siDNp63 EdU / DAPI m 200 mm 200 mm 500 m 500 mm D % EdU-positive cells +siDNp63 +si Np63

+siC Number of colonies/field +siC

EF G H HCC95 HCC95 HCC95 HCC95 +siC +siDNp63 +siC +siDNp63 EdU / DAPI 200 mm 200 mm 500 mm 500 mm

% EdU-positive cells D +siC +siDNp63 Number of colonies/field +siC +si Np63

IJ KL Caki-1 Caki-1 Caki-1 Caki-1 +EV +DNp63a +EV +DNp63a

m m EdU / NucRed 200 mm 200 mm 200 m 200 m

% EdU-positive cells +DNp63a +DNp63a +EV Number of colonies/field +EV

Figure 2. DNp63 plays a pleiotropic role as a tumor suppressor or oncogene. Quantification and fluorescence representative images of EdU (green) incorporation MCF10-CA1D (A and B), HCC95 (E and F), and Caki-1 (I and J) cell lines. C and D, G and H, K and L, Quantification and bright-field representative images of anchorage-independent colony formation of MCF10-CA1D (C and D), HCC95 (G and H), and Caki-1 (K and L) cell lines in soft agar assay (per 10 field). Data are mean SD, n ¼ 3. Asterisks indicate statistical significance. , P < 0.005 versus siRNA scramble control (siC) or empty vector (EV), two-tailed t test.

colony formation (Fig. 2A–D; Supplementary Fig. S2B). In and B), indicating that Lef1 is downstream of DNp63. We then HCC95 LUAD cells, where DNp63 is overexpressed (Supplemen- knocked down Lef1 in both MCF10-CA1D and HCC95 cell tary Fig. S2A), knockdown of DNp63 decreased soft agar colony lines (Fig. 3C), and measured soft agar colony growth. Similarly formation but not proliferation (Fig. 2E–H; Supplementary to the results where DNp63 was knocked down, the Lef1 knock- Fig. S2B). To assess tumor suppressor activities of DNp63, we down reduced MCF10-CA1D and HCC95 colony formation overexpressed DNp63a in Caki-1 human renal cancer cells (Sup- (Fig. 3D–G). In addition, overexpressing Lef1 in DNp63-knocked plementary Fig. S2B), where we found DNp63 as tumor suppres- down cells (Supplementary Fig. S3A) significantly increased sive (Fig. 1B) and whereas we did not observe significant differ- MCF10-CA1D and HCC95 colony formation (Fig. 3H–K), and ences in proliferation (Fig. 2I–J), the number of colonies MCF10-CA1D cell proliferation (Supplementary Fig. S3B and decreased by two-fold compared with control (Fig. 2K–L). These S3C), but not HCC95 proliferation (Supplementary Fig. S3D and data further support that DNp63 plays a pleiotropic role depend- S3E). Together, these data demonstrated that DNp63 functions as ing on tissue of origin. an oncogene via regulating Lef1. Importantly, the DNp63-over- To identify the pleiotropic mechanism exerted by DNp63 in expressing Caki-1 cells had increased Lef1 mRNA and protein tumorigenesis, we overlapped previously published human expression (Supplementary Fig. S4A and S4B), which further DNp63 Chip-Seq data (18) with our TCGA-derived signatures, confirmed that DNp63 regulates Lef1 expression. To assess wheth- which revealed enrichment for EMT genes. Next, we conducted er the tumor suppressive role of DNp63 functions through Lef1, hypergeometric analysis to identify transcription factors that we overexpressed Lef1 in Caki-1 cells, and noted a nonstatistically regulate these EMT genes and overlap DNp63 keratinocyte sig- significant decrease in soft agar colony formation (Supplementary natures. We identified Lef1 targets (INHBA, USP2, COL5A1, Fig. S4C and S4E). When Lef1 was knocked down in Caki-1 cells BMP1, PMEPA1) to be highly significant (P ¼ 0.0009). To assess where we overexpressed DNp63 (Supplementary Fig. S4F), no whether the oncogenic role of DNp63 functions through Lef1,we significant difference was noted with respect to cell proliferation assessed Lef1 mRNA expression in MCF10-CA1D and HCC95 cell and colony formation (Supplementary Fig. S4G–S4J), suggesting lines after DNp63 knockdown. Lef1 expression decreased (Fig. 3A the involvement of other factors in addition to Lef1 in the tumor

OF4 Cancer Res; 78(2) January 15, 2018 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2017 American Association for Cancer Research. www.aacrjournals.org xrsigete i rsi or siC either expressing e1i ontemof downstream is Lef1 3. Figure otaa sa pr10 (per assay agar soft control. loading a as used xrsini C1-ADadHC5clsepesn ihrsCo siD or siC either expressing cells HCC95 and MCF10-CA1D in expression HI DE A FG JK Downloaded from

Number of colonies/field Number of colonies/field Number of colonies/field Number of colonies/field Relative mRNA expression 0.5 1.0 1.5 0.0 10 15 20 10 15 10 15 10 15 20 0 5 0 5 0 5 0 5 +siC +siC +siC D +siC +siC Published OnlineFirstNovember27,2017;DOI:10.1158/0008-5472.CAN-17-1803 p3addcae h ucinof function the dictates and Np63 fi D D l) aaaemean are Data eld). CA1D p3 seik niaesaitclsigni statistical indicate Asterisks Np63. – +siDNp63 K, Quanti sLf- +siLef1-2 +siLef1-1 sLf- +siLef1-2 +siLef1-1 +siDNp63 +siDNp63 cancerres.aacrjournals.org CA1D CA1D HCC95 HCC95 Lef1 fi +siC ainadbright- and cation +Lef1 +siDNp63 +siDNp63 +Lef1 HCC95 +siDNp63 SD, n ¼ fi .Atrssidct ttsia signi statistical indicate Asterisks 3. D l ersnaieiae facoaeidpnetcln omto fidctdcl ie in lines cell indicated of formation colony anchorage-independent of images representative eld p3a uo upesro oncogene. or suppressor tumor a as Np63 (higher exposure) Cancer Research. fi BC cance. p3( Np63 Actin Lef1 Lef1 +siC +siC +siC +siC on September 26, 2021. © 2017American Association for B , ri C1-ADadHC5clstetdwt i,sLf-,o ie12( siLef1-2 or siLef1-1, siC, with treated cells HCC95 and MCF10-CA1D in or ) P < .0 esssC two-tailed siC, versus 0.005 200 mm2 200 mm2 500 mm5 500 mm5 +siC CA1D +siDNp63 P3Ioom rncitoa intrsi Tumorigenesis in Signatures Transcriptional Isoforms TP63 fi

cance. +siC HCC95

+siDNp63 +siDNp63 +siDNp63 +siLef1-1 +siLef1-1 , HCC95 HCC95 A, P CA1D CA1D R-C for qRT-PCR < .0 esssCo si or siC versus 0.005 t Actin test. Lef1 0 0 0 0 0 B 0 0 0

mm2 mm2 mm5 mm5 and Lef1 acrRs 82 aur 5 2018 15, January 78(2) Res; Cancer +siC C, nMF0C1 n C9 cells HCC95 and MCF10-CA1D in etr ltaayi o Lef1 for analysis blot Western +siLef1-1 CA1D +siDNp63 +Lef1 +siDNp63 +Lef1 D +siLef1-2 p3 two-tailed Np63, +siLef1-2 +siLef1-2

+siC

+siLef1-1 HCC95 +siLef1-2 0 0 0 0 C 0 0 0 0 .Atnwas Actin ).

t m m m m test. OF5 Published OnlineFirst November 27, 2017; DOI: 10.1158/0008-5472.CAN-17-1803

Abbas et al.

suppressor roles of DNp63. Hence, DNp63 regulation of Lef1 TP63 isoform signatures. We identified multiple clusters (C) dictates its oncogenic roles in different tissues while the tumor corresponding to TAp63- and DNp63-derived signatures suppressor pathways of DNp63 are still not identified. (Fig. 6A and B). The corresponding activity and survival pattern was significantly different across the clusters identified in all TAp63- and DNp63-regulated transcriptional activities are lost cancers (Fig. 6C and D). Specifically, patients with the highest in higher stages and can predict survival of BLCA, KIRC, KIRP, TAp63 and DNp63 activity had significantly better survival than and LGG the other clusters (Fig. 6E and F). These findings provide strong GSEA analysis revealed a suppressive role for both TAp63 and evidence of distinct TAp63 and DNp63 transcriptome profiles that DNp63 in the progression of four tumors: bladder cancer (BLCA), identify distinct groups of patients within cancers and can predict KIRC, KIRP, and low-grade glioma (LGG; Fig. 4A). To explore the their survival. This is of major significance given tumor hetero- role of TP63 isoforms in tumor progression, we first identified geneity and the difficulty of identifying patients with different TAp63- and DNp63-regulated genes that were enriched in cancer prognosis within the same stage or group. progression signatures and the corresponding isoform signature, hereafter dubbed TAp63BLCA, DNp63BLCA, TAp63KIRC, DNp63KIRC, Cell-cycle activity, pluripotency potential, EMT, and ECM TAp63KIRP, DNp63KIRP, TAp63LGG, and DNp63LGG (Supplemen- remodeling in high-stage tumors modulated by TAp63 and tary File 4.1–4.2 and Supplementary Fig. S5A and S5B). DNp63 signatures We found that high-grade/stage KIRC, KIRP, and LGG tumors Cell cycle, sister chromatid/chromosome, and organelle regu- had significantly lower activities of their corresponding TAp63- latory pathways were the most highly enriched pathways in driven signatures compared to low-grade/stage tumors (P < TAp63 signatures (Supplementary Fig. S7A and Supplementary 0.0001; Fig. 4B). KIRC (P < 0.0001), KIRP (P ¼ 0.023), and LGG File S4.3). For DNp63-derived signatures, there were two genes (P < 0.0001) patients with higher corresponding TAp63 activity (CTHRC1 and COL5A1 both are major ECM constituents) and 14 had significantly better survival than patients with lower corre- pathways that were common among DNp63BLCA, DNp63KIRC, sponding TAp63 activity (Fig. 4C). Similarly, high-grade/stage DNp63KIRP, and DNp63LGG (Supplementary File S4.4). The high- BLCA, KIRC, KIRP, and LGG tumors had significantly lower est enrichments in DNp63 signatures were developmental and activities of their corresponding DNp63-driven signatures com- ECM remodeling pathways (Supplementary Fig. S7B), which were pared to low-grade/stage tumors (P < 0.0001; Fig. 4D), and KIRC of significant interest as these were indicative of stem cell plur- (P < 0.0001) and LGG (P < 0.0001) patients with higher corre- ipotency, EMT, and metastatic mechanisms. Because TAp63 and sponding DNp63 activities had better survival (Fig. 4e). These DNp63 were both acting as suppressors of tumor progression in data suggest that higher grade/stage tumors lose TAp63 and the same cancers, we investigated whether these roles are exerted DNp63 activities when progressing. Association between activity via common or distinct transcriptome networks. Interestingly, and survival was also evident within the same grade/stage of there were few overlapping genes regulated by corresponding tumors (Fig. 4F and G). This is of particular importance as it TAp63 and DNp63 signatures in KIRC, BLCA, LGG, and KIRP suggests that there is significant heterogeneity in the activity of progression (Supplementary Fig. S7C–S7F). DNp63 within high-grade/stage of KIRC, KIRP, and LGG that may CTHRC1 and COL5A1 are DNp63-regulated genes common warrant different therapy. TAp63BLCA consisted of one gene only to BLCA, KIRC, KIRP, and LGG progression signatures. In (COL8A1) and was not assessed. addition to the ECM remodeling role of CTHRC1 and COL5A1, We then used independent cohorts to validate our findings. COL5A1 is also highly expressed in known stem cell signatures. With respect to TAp63, high-grade (19) and stage (20) KIRC, high- Hence, we investigated whether the DNp63 activity of these two stage KIRP (21), and high-grade LGG (22) tumors had signifi- genes, hereafter dubbed DNp63CTHRC1/COL5A1, could predict cantly lower corresponding TAp63-signature activities (Fig. 5A; staging and survival. In addition to the ECM remodeling role Supplementary Fig. S6A). Also, KIRP patients (21) with higher of CTHRC1 and COL5A1, COL5A1 is also highly expressed in TAp63KIRP activity had significantly better survival than KIRP known stem cell signatures (24). We found significantly lower patients with lower TAp63KIRP activity (Fig. 5B). Similarly, DNp63CTHRC1/COL5A1 activity in the higher grade/stage of BLCA, high-stage BLCA (23), high-grade (19) and high-stage (20) KIRC, KIRC, KIRP, and LGG (P < 0.0001 for all; Supplementary high-stage KIRP (21), and high-grade LGG (22) patients had Fig. S8A). Further, KIRC (P ¼ 0.0009), KIRP (0.0086), and significantly lower corresponding DNp63-signature activities LGG (P ¼ 0.0241) patients with higher activity of (Fig. 5C; Supplementary Fig. S6B). These findings also translated DNp63CTHRC1/COL5A1 had significantly better survival than into significantly better survival for BLCA and KIRP patients with patients with lower activity (Supplementary Fig. S8B). We then higher corresponding DNp63-signature activities (Fig. 5D). These assessed whether single gene expression profile of COL5A1 and findings indicate that TAp63- and DNp63-derived signatures can CTHRC1 correlates with survival in BLCA, KIRC, KIRP, and predict survival across cancers and within specific stages in dif- LGG tumors. Interestingly, patients with higher expression of ferent cohorts and can identify patients with better survival COL5A1 or CTHRC1 almost consistently had worst prognosis among advanced tumors. Further, the large number of genes and (P < 0.05; Supplementary Fig. S9A and S9B). These findings pathways identified can be exploited in the clinic to identify show that a DNp63-regulated signature driven by two genes therapies that could target these downstream effectors and can that are known to modulate ECM and stem cell pluripotency explain the heterogeneity in responding to certain treatments. can predict progression and survival of patients. We then investigated whether previously published stem cell Patient prognosis correlates with expression profiles of TAp63- signatures were enriched in high-grade/stage tumors and / and DNp63-regulated signatures DNp63 keratinocytes. GSEA analyses identified a significant We conducted unsupervised hierarchal clustering on BLCA, enrichment of cancer invasive and primary stem cells (25, 26), KIRC, KIRP, and LGG patients with respect to corresponding mouse and human embryonic stem cell signatures (24, 27), and

OF6 Cancer Res; 78(2) January 15, 2018 Cancer Research

TP63 Isoforms Transcriptional Signatures in Tumorigenesis

B C DE TAp63 DNp63

High ΔNp63BLCA Activity (53) Low ΔNp63BLCA Activity (56)

BLCA Activity BLCA BLCA

ΔNp63 P = 0.087 P < 0.0001

High ΔNp63KIRC Activity (86) High TAp63 KIRC Activity (86) Low ΔNp63KIRC Activity (87) Low TAp63 KIRC Activity (87) Activity Activity

KIRC KIRC KIRC P < 0.0001 P < 0.0001 ΔNp63 TAp63 P < 0.0001 P < 0.0001

High ΔNp63KIRP Activity (60) KIRP High TAp63 Activity (60) Low ΔNp63KIRP Activity (61) Low TAp63KIRP Activity (61) Activity KIRP Activity KIRP KIRP KIRP KIRP

P = 0.023 P ΔNp63 = 0.08 TAp63 P < 0.0001 P < 0.0001

High TAp63LGG Activity (114) High ΔNp63LGG Activity (113) Low TAp63LGG Activity (115) Low ΔNp63LGG Activity (113) Activity LGG Activity LGG LGG

P P < 0.0001 ΔNp63 < 0.0001 TAp63 P < 0.0001 P < 0.0001

F KIRC (Stage IV) KIRP (Stage III/IV) LGG (G3) High TAp63KIRC Activity (20) High TAp63LGG Activity (59) KIRC Low TAp63 Activity (21) KIRP Low TAp63LGG Activity (60) High TAp63 Activity (15) P P = 0.009 Low TAp63KIRP Activity (15) < 0.0001 P = 0.0153 TAp63

G KIRC (Stage IV) LGG (G2) LGG (G3) High ΔNp63KIRC Activity (20) High ΔNp63LGG Activity (58) KIRC LGG Low ΔNp63 Activity (21) LGG Low ΔNp63 Activity (60) P High ΔNp63 Activity (54) = 0.0002 Low ΔNp63LGG Activity (53) P < 0.0001 P DNp63 = 0.0203

Figure 4. Tumor suppressor activity of TAp63 and DNp63 in cancer progression. A, NES of upregulated and downregulated cancer progression signatures / in TAp63 and DNp63 / signatures. B, Activity score of TAp63KIRC, TAp63KIRP, and TAp63LGG in corresponding cancer patients at high- and low-grade/stages. C, Kaplan–Meier survival of KIRC, KIRP, and LGG patients with respect to the corresponding TAp63 signature activity. D, Activity score of DNp63BLCA, DNp63KIRC, DNp63KIRP,andDNp63LGG in corresponding cancer patients at high- and low-grade/stages. E, Kaplan–Meier survival of BLCA, KIRC, KIRP, and LGG patients with respect to the corresponding DNp63 signature activity. F, Kaplan–Meier survival within stage IV KIRC, stage III/IV KIRP, and grade (G) 3 LGG with respect to TAp63KIRC, TAp63KIRP, and TAp63LGG activities. G, Kaplan–Meier survival within stage IV KIRC, and G2 and G3 LGG patients with respect to DNp63KIRC and DNp63LGG activities, respectively. Number of cases in each survival group is listed between parenthesis where indicated.

www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 OF7

Abbas et al.

A B CD TAp63 DNp63 High ΔNp63BLCA Activity (41) Low ΔNp63BLCA Activity (42) Activity BLCA BLCA

ΔNp63 P = 0.0019 P = 0.0011 GEO#: GSE13507 Activity Activity KIRC KIRC KIRC TAp63 P < 0.0001 ΔNp63 P < 0.0001 GEO#: GSE40435

High ΔNp63KIRP Activity (6) KIRP High TAp63 Activity (5) Low ΔNp63KIRP Activity (8) Low TAp63KIRP Activity (8) Activity Activity KIRP KIRP KIRP

ΔNp63 P = 0.016 TAp63 GEO#: GSE2748 P = 0.0002 P = 0.02 P < 0.0001 Activity Activity LGG LGG LGG TAp63 ΔNp63 GEO#: GSE35158 P = 0.0054 P = 0.0022

Figure 5. Validation of TAp63 and DNp63 in independent cohorts. A, Validation of corresponding TAp63 activities in independent cohorts of KIRC, KIRP, and LGG. B, Survival of KIRP in independent cohorts with respect to TAp63KIRP activity.C,Validation of corresponding DNp63 activities in independent cohorts of BLCA, KIRC, KIRP, and LGG. D, Survival of BLCA and KIRP in independent cohorts with respect to DNp63BLCA and DNp63KIRP activities, respectively. Number of cases in each group is listed between parenthesis where indicated.

Nanog, Oct4, and Sox2 (NOS) targets (27) in high-grade/stage one stem cell pluripotency gene COL5A1 is regulated by DNp63, / tumors and DNp63 epidermal cells signatures (Fig. 7A). Sim- common to all of DNp63BLCA, DNp63KIRP, DNp63KIRC,and ilarly, there was significant enrichment of primary tumor and DNp63LGG signatures, and is also a major ECM remodeling metastatic ECM signatures (28) in high-stage/grade and factor; we showed it to be a major predictor of staging / DNp63 epidermal cell signatures (Fig. 7B). We found a signif- and survival when combined with CTHRC1 (Supplementary / icant enrichment of EMT in DNp63 keratinocytes and progres- Fig. S8A-B). To note, the role of DNp63 in pluripotency in the sion signatures of BLCA, KIRC, and KIRP (Fig. 7c), which is different tumors is exerted via a large network of genes as evident interesting given the role of TP63 isoforms in stem cell from the little overlap among the stem cell pluripotency genes of maintenance. DNp63BLCA, DNp63KIRP, DNp63KIRC,andDNp63LGG. The sup- To identify specific DNp63-regulated genes that promote stem pressive activity of DNp63 in tumor progression is consistent cell pluripotency in cancer progression, we compared with previous reports indicating that DNp63 serves to promote DNp63BLCA, DNp63KIRP, DNp63KIRC,andDNp63LGG signatures differentiation of epithelial tissues and that loss of DNp63 leads to the known stem cell signatures that were enriched in our to pluripotency (8, 29). The progression of these tumors entails progressioncancermodelshowninFig.7A(24–27). We found loss of TAp63 and DNp63 activities, which leads to increased 11, 17, 43, and 23 DNp63-regulated stem cell pluripotency stem cell pluripotency, ECM remodeling, and EMT via DNp63, genes to be upregulated in DNp63BLCA, DNp63KIRP, DNp63KIRC, and disruption of normal cell-cycle regulation by TAp63, which and DNp63LGG and at least one previously published stem cell ultimately leads to progression and metastasis (Fig. 7E). These signature, respectively (Fig. 7D). We also identified overlapping findings identify patients within different stages that could genes between the different signatures (Fig. 7D). Interestingly, benefit from treatments that specifically target these pathways.

OF8 Cancer Res; 78(2) January 15, 2018 Cancer Research

TP63 Isoforms Transcriptional Signatures in Tumorigenesis

A KIRC KIRP LGG C-1 C-2 C-3 C-1 C-2 C-1 C-2

TAp63

B BLCA KIRC KIRP LGG C-1 C-2 C-1 C-2 C-1 C-2 C-3 C-2 C-1 C-3

DNp63

C KIRC KIRP LGG 20 10 4 ANOVA = 0.022 P < 0.0001 P < 0.0001 2 10 0 0 0 Activity Activity –10 Activity TAp63 –2 –10 LGG KIRC –4 KIRP –20 P-value –20 –6 C1-C2 =0.49 –30 –8 C1-C3 =0.01 –30 TAp63 TAp63 TAp63 –10 C2-C3 =0.01 –40 –40 C-1 C-2 C-3 C-1 C-2 C-1 C-2 D BLCA KIRC KIRP LGG 20 20 P < 0.0001 P < 0.0001 50 ANOVA = 0.022 25 ANOVA = 0.022 0 10 0 0 0 –50 –25 –20 Activity Activity –10 –100 Activity DNp63 Activity –50 –40 –20 –150 KIRP KIRP LGG –75 BLCA BLCA KIRC –30 –200 –60 P-value –100 P-value –40 p63 –250 C1-C2 < 0.0001 C1-C2 <0.0001 –80 –50 C1-C3 < 0.0001 –125 C1-C3 <0.0001

D N –300 D Np63 D Np63 D Np63 C2-C3 = 0.8 C2-C3 <0.0001 –100 –60 –350 –150 C-1 C-2 C-1 C-2 C-1 C-2 C-3 C-1 C-2 C-3

E KIRC KIRP LGG C-1 (73) 1.0 1.0 1.0 C-1 (162) C-1 (326) C-2 (205) C-2 (130) C-3 (67) 0.8 C-2 (83) 0.8 0.8 P = 0.01 P = 0.0076 P = 0.007 0.6 0.6 0.6 TAp63 0.4 0.4 Log rank P-value 0.4 C1-C2 =0.767 0.2 0.2 0.2 C1-C3 =0.027

C2-C3 =0.009 probability Survival Survival probability Survival 0.0 Survival probability Survival 0.0 0.0 0204060 80 100 120 0 50 100 150 200 0 50 100 150 200 Time (months) Time (months) Time (months) F BLCA KIRC KIRP LGG )912( 1-C )912( C-1 (158) 1.0 C-1 (303) 1.0 C-1 (118) 1.0 1.0 C-2 (32) C-2 (79) C-2 (122) C-3 (55) 0.8 C-2 (102) 0.8 P < 0.0001 0.8 0.8 C-3 (71) P = 0.259 P = 0.009 P < 0.0001 0.6 Log rank P-value 0.6 0.6 0.6 C1-C2 = 0.03 DNp63 Log rank P-value 0.4 0.4 0.4 0.4 C1-C3 < 0.0001 C1-C2 = 0.0011 C2-C3 < 0.0001 0.2 0.2 0.2 C1-C3 = 0.042 0.2 C2-C3 = 0.59 Survival probability Survival Survival probability Survival Survival probability Survival 0.0

Survival probability Survival 0.0 0.0 0.0 0204060 80 100 120 0204060 80 100 120 0 50 100 150 200 0 50 100 150 200 Time (months) Time (months) Time (months) Time (months)

Figure 6. Expression profiles of TAp63- and DNp63-regulated signatures identify different groups of BLCA, KIRC, KIRP, and LGG patients with different prognosis. A and B, Unsupervised hierarchal clustering of BLCA, KIRC, KIRP, and LGG patients based on expression profile of corresponding TAp63 (A)andDNp63 (B) signatures. C and D, Activity score of TAp63KIRC, TAp63KIRP, and TAp63LGG (C) and DNp63BLCA, DNp63KIRC, DNp63KIRP, and DNp63LGG (D) in corresponding cancer patients of different clusters. E and F, Kaplan–Meier survival of different clusters of BLCA, KIRC, KIRP, and LGG patients with respect to corresponding TAp63 (E) and DNp63 (F) expression profile. Number of cases in each survival group is listed between parenthesis where indicated. www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 OF9

Abbas et al.

ABStem cell signatures ECM Signatures

9.0 5.0

8.0 4.0 7.0 3.0 6.0 5.0 2.0 4.0 1.0 3.0 Mouse ESC-like ECM Primary Core ESC 0.0 2.0 Human ESC-like ECM Metastatic ESC1 ECM (Primary + Metastatic) 1.0 ESC2 BLCA EPHB2HI(ISC) KIRC KIRP 0.0 + - LGG KC CD44 CD24 (IGS) Normalized enrichment score –/– NOS Targets

Normalized enrichment score BLCA DNp63 KIRC KIRP LGG KC –/–

DNp63 C D DNp63BLCA pluripotency genes DNp63KIRC pluripotency genes Downregulated Upregulated KIRP 6 –/– DNp63 pluripotency genes in EMT in EMT DNp63 KC DNp63LGG pluripotency genes 4 BLCA KIRC 2 KIRP

–2

–4

–6

–8

Normalized enrichment score –10

Extracellular matrix Tumor progression

remodelling Tumor progression Tumor

DNp63 Tumor progression Tumor

TAp63 Pluripotent stem cells

Figure 7. Enrichment of pluripotency, ECM and EMT in DNp63-driven progression signatures, and cooperation between TAp63 and DNp63 to suppress progression. / A–C, NES of BLCA, KIRC, KIRP, and LGG progression and DNp3 signatures in known stem cells and pluripotency (A), ECM (B), and EMT (C) signatures. D, ClueGO diagram representing enrichment for pluripotency genes in DNp63-driven signatures of BLCA, KIRC, KIRP, and LGG. E, Suggested model of cooperativity between TAp63 and DNp63 to suppress tumor progression via modulation of EMT, ECM, pluripotency, and cell-cycle activity.

/ / Discussion were generated from TAp63 and DNp63 epidermal cells In this study, we conducted a pan-cancer analysis using TCGA exhibiting profound phenotypes in epithelial stem cell mainte- data and informed by unique transcriptome proﬁles of TP63 nance and renewal, and thus, allowed for an assessment of p63 isoforms, TAp63 and DNp63. These unique transcriptome proﬁles isoform-driven stem cell biology in cancer development and

OF10 Cancer Res; 78(2) January 15, 2018 Cancer Research

TP63 Isoforms Transcriptional Signatures in Tumorigenesis

progression (Fig. 1A; refs. 6, 8, 12, 29). TP63 alterations have been pluripotency, ECM, and EMT markers that can confer resistance previously implicated in many cancers (30–33), and TP63 is to treatment. frequently used as a diagnostic marker (34, 35). Our mechanistic understanding and the clinical significance of the underlying Disclosure of Potential Conflicts of Interest mechanisms of TP63 isoforms in tumor suppression and onco- No potential conflicts of interest were disclosed. genesis are still poorly defined. Here, we found that although TAp63 activity follows a suppressive pattern in tumor develop- Authors' Contributions ment and progression, the activity of DNp63 was pleiotropic in fi Conception and design: H.A. Abbas, N.H.B. Bui, C. Coarfa, E.R. Flores cancer development. Importantly, we identi ed Lef1 as a major Development of methodology: H.A. Abbas, J. Wong, P. Gunaratne, C. Coarfa, downstream target of DNp63 that mediates the switch between E.R. Flores the tumor suppressive and oncogenic activities of DNp63 in a Acquisition of data (provided animals, acquired and managed patients, tissue-specific manner. The DNp63-Lef1 axis seems to be crucial provided facilities, etc.): H.A. Abbas, N.H.B. Bui, K. Rajapakshe, P. Gunaratne, for its role as an oncogene whereas the underlying factors that C. Coarfa, E.R. Flores Analysis and interpretation of data (e.g., statistical analysis, biostatistics, dictate its role as a tumor suppressor gene remain to be elucidated. computational analysis): H.A. Abbas, N.H.B. Bui, K. Rajapakshe, Importantly, we found no evidence that TAp63 regulates Lef1 K. Rajapakshe, J. Wong, P. Gunaratne, K.Y. Tsai, C. Coarfa, E.R. Flores neither in our genomic analysis nor via chromatin immunopre- Writing, review, and/or revision of the manuscript: H.A. Abbas, N.H.B. Bui, cipitation (ChIP) assay (Napoli and Flores, unpublished data). K. Rajapakshe, K. Rajapakshe, J. Wong, K.Y. Tsai, C. Coarfa, E.R. Flores This study constitutes the largest analysis from TCGA plat- Administrative, technical, or material support (i.e., reporting or organizing form in terms of number of cancers analyzed and in light of data, constructing databases): H.A. Abbas, C. Coarfa, E.R. Flores Study supervision: C. Coarfa, E.R. Flores important epidermal stem cell and cancer regulators, TAp63 and DNp63, and their transcriptional networks. We unveiled Acknowledgments TAp63- and DNp63-driven signatures that can predict tumor E.R. Flores was supported by a grant from NCI (R01CA160394 and development, progression, and patient survival. Although R35CA197452), CPRIT (RP140271 and RP150094), NCI-Cancer Center Core DNp63 was previously thought to act primarily as an oncogene, Grant (CA-16672; University of Texas M.D. Anderson Cancer Center), the DNp63 has more recently been found to be lost during tumor Hildegardo E. and Olga M. Flores Foundation, and the Mel Klein Foundation. progression and proposed to be able to suppress metastasis E.R. Flores is an NCI Outstanding Investigator, a Moffitt Distinguished Scholar, a (36, 37). Our results demonstrate that DNp63 plays a dual role scholar of the Leukemia and Lymphoma Society, the Rita Allen Foundation, and the V Foundation for Cancer Research. C. Coarfa was partially supported by an as a tumor suppressor or oncogene in cancer development. Its Alkek Center for Molecular Discovery Pilot grant, and by the Cancer Prevention regulation of Lef1 is crucial for its oncogenic roles. In addition, & Research Institute grants [RP120092 (Texas Proteomics & Metabolomics Core DNp63 acts as a suppressor of tumor progression. Using inte- Facility), RP170295, RP170005]. K.Y. Tsai was supported by a grant from NCI grative bioinformatics analyses, we identified TAp63-driven (R01CA194617 and R01CA194062). H.A. Abbas was supported by an Odyssey and DNp63-driven transcriptional networks that can predict Fellowship at University of Texas MD Anderson Cancer Center. survival and tumor progression within and across cancer stages. fi fi The costs of publication of this article were defrayed in part by the payment of Our ndings identify patient populations that may bene t page charges. This article must therefore be hereby marked advertisement in from targeting TAp63 and DNp63 pathways in order to improve accordance with 18 U.S.C. Section 1734 solely to indicate this fact. their survival. Further, the TAp63 and DNp63 signatures establish an expression profile that can identify different clusters Received June 21, 2017; revised October 25, 2017; accepted November 14, of patients with differential potential to express stem cell 2017; published OnlineFirst November 27, 2017.

References 1. Chang CJ, Chao CH, Xia W, Yang JY, Xiong Y, Li CW, et al. p53 regulates 9. Venkatanarayan A, Raulji P, Norton W, Chakravarti D, Coarfa C, Su X, et al. epithelial-mesenchymal transition and stem cell properties through mod- IAPP-driven metabolic reprogramming induces regression of p53-deficient ulating miRNAs. Nat Cell Biol 2011;13:317–23. tumours in vivo. Nature 2015;517:626–30. 2. Lane DP. Cancer. p53, guardian of the genome. Nature 1992;358:15–6. 10. Tomasini R, Tsuchihara K, Wilhelm M, Fujitani M, Rufini A, Cheung CC, 3. Su X, Chakravarti D, Flores ER. p63 steps into the limelight: crucial roles in et al. TAp73 knockout shows genomic instability with infertility and tumor the suppression of tumorigenesis and metastasis. Nat Rev Cancer 2013; suppressor functions. Genes Develop 2008;22:2677–91. 13:136–43. 11. Mills AA, Zheng B, Wang XJ, Vogel H, Roop DR, Bradley A. p63 is a p53 4. Flores ER, Tsai KY, Crowley D, Sengupta S, Yang A, McKeon F, et al. p63 and homologue required for limb and epidermal morphogenesis. Nature 1999; p73 are required for p53-dependent apoptosis in response to DNA damage. 398:708–13. Nature 2002;416:560–4. 12. Yang A, Schweitzer R, Sun D, Kaghad M, Walker N, Bronson RT, et al. p63 is 5. Flores ER, Sengupta S, Miller JB, Newman JJ, Bronson R, Crowley D, et al. essential for regenerative proliferation in limb, craniofacial and epithelial Tumor predisposition in mice mutant for p63 and p73: evidence for development. Nature 1999;398:714–8. broader tumor suppressor functions for the p53 family. Cancer Cell 13. Abd Raboh NM, Hakim SA. Diagnostic role of DOG1 and p63 immuno- 2005;7:363–73. histochemistry in salivary gland carcinomas. Int J Clin Exp Pathol 6. Su X, Paris M, Gi YJ, Tsai KY, Cho MS, Lin YL, et al. TAp63 prevents 2015;8:9214–22. premature aging by promoting adult stem cell maintenance. Cell Stem Cell 14. Kiselyov A, Bunimovich-Mendrazitsky S, Startsev V. Key signaling path- 2009;5:64–75. ways in the muscle-invasive bladder carcinoma: clinical markers for disease 7. Su X, Chakravarti D, Cho MS, Liu L, Gi YJ, Lin YL, et al. TAp63 suppresses modeling and optimized treatment. Int J Cancer 2016;138:2562–9. metastasis through coordinate regulation of Dicer and miRNAs. Nature 15. Ma Y, Fan M, Dai L, Kang X, Liu Y, Sun Y, et al. Expression of p63 and CK5/6 2010;467:986–90. in early-stage lung squamous cell carcinoma is not only an early diagnostic 8. Chakravarti D, Su X, Cho MS, Bui NH, Coarfa C, Venkatanarayan indicator but also correlates with a good prognosis. Thorac Cancer 2015; A, et al. Induced multipotency in adult keratinocytes through 6:288–95. down-regulation of DeltaNp63 or DGCR8. Proc Nat Acad Sci U S A 16. Senoo M, Pinto F, Crum CP, McKeon F. p63 Is essential for the proliferative 2014;111:E572–81. potential of stem cells in stratified epithelia. Cell 2007;129:523–36.

www.aacrjournals.org Cancer Res; 78(2) January 15, 2018 OF11

Abbas et al.

17. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, 27. Ben-Porath I, Thomson MW, Carey VJ, Ge R, Bell GW, Regev A, et al. An et al. Gene set enrichment analysis: a knowledge-based approach for embryonic stem cell-like gene expression signature in poorly differentiated interpreting genome-wide expression profiles. Proc Nat Acad Sci U S A aggressive human tumors. Nat Genet 2008;40:499–507. 2005;102:15545–50. 28. Naba A, Clauser KR, Whittaker CA, Carr SA, Tanabe KK, Hynes RO. 18. McDade SS, Henry AE, Pivato GP, Kozarewa I, Mitsopoulos C, Fenwick K, Extracellular matrix signatures of human primary metastatic colon cancers et al. Genome-wide analysis of p63 binding sites identifies AP-2 factors as and their metastases to liver. BMC Cancer 2014;14:518. doi: 10.1186/ co-regulators of epidermal differentiation. Nucleic Acids Res 2012; 1471-2407-14-518. 40:7190–206. 29. Chakrabarti R, Wei Y, Hwang J, Hang X, Andres Blanco M, Choudhury A, 19. Wozniak MB, Le Calvez-Kelm F, Abedi-Ardekani B, Byrnes G, Durand G, et al. DeltaNp63 promotes stem cell activity in mammary gland develop- Carreira C, et al. Integrative genome-wide gene expression profiling of clear ment and basal-like breast cancer by enhancing Fzd7 expression and Wnt cell renal cell carcinoma in Czech Republic and in the United States. PloS signalling. Nat Cell Biol 2014;16:1004–15, 1–13. One 2013;8:e57886. 30. Hoadley KA, Yau C, Wolf DM, Cherniack AD, Tamborero D, Ng S, et al. 20. Pena-Llopis S, Vega-Rubin-de-Celis S, Liao A, Leng N, Pavia-Jimenez A, Multiplatform analysis of 12 cancer types reveals molecular classification Wang S, et al. BAP1 loss defines a new class of renal cell carcinoma. Nat Gen within and across tissues of origin. Cell 2014;158:929–44. 2012;44:751–9. 31. Cancer Genome Atlas Research N. Comprehensive genomic characteriza- 21. Yang XJ, Tan MH, Kim HL, Ditlev JA, Betten MW, Png CE, et al. A molecular tion of squamous cell lung cancers. Nature 2012;489:519–25. classificationofpapillaryrenal cellcarcinoma.CancerRes 2005;65:5628–37. 32. Cancer Genome Atlas N. Comprehensive molecular characterization of 22. Gorovets D, Kannan K, Shen R, Kastenhuber ER, Islamdoust N, Campos C, human colon and rectal cancer. Nature 2012;487:330–7. et al. IDH mutation and neuroglial developmental features define clinically 33. Cancer Genome Atlas N. Comprehensive genomic characterization of head distinct subclasses of lower grade diffuse astrocytic glioma. Clin Cancer Res and neck squamous cell carcinomas. Nature 2015;517:576–82. 2012;18:2490–501. 34. Orzol P, Holcakova J, Nekulova M, Nenutil R, Vojtesek B, Coates PJ. The 23. Kim WJ, Kim EJ, Kim SK, Kim YJ, Ha YS, Jeong P, et al. Predictive value of diverse oncogenic and tumour suppressor roles of p63 and p73 in cancer: a progression-related gene classifier in primary non-muscle invasive bladder review by cancer site. Histol Histopathol 2015;30:503–21. cancer. Mol Cancer 2010;9:3. doi: 10.1186/1476-4598-9-3. 35.SethiI,RomanoRA,GluckC,SmalleyK,VojtesekB,BuckMJ,etal.A 24. Wong DJ, Liu H, Ridky TW, Cassarino D, Segal E, Chang HY. Module map global analysis of the complex landscape of isoforms and regulatory of stem cell genes guides creation of epithelial cancer stem cells. Cell Stem networks of p63 in human cells and tissues. BMC Genomics 2015; Cell 2008;2:333–44. 16:584. doi: 10.1186/s12864-015-1793-9. 25. Merlos-Suarez A, Barriga FM, Jung P, Iglesias M, Cespedes MV, Rossell D, 36. Adorno M, Cordenonsi M, Montagner M, Dupont S, Wong C, Hann B, et al. et al. The intestinal stem cell signature identifies colorectal cancer stem cells A Mutant-p53/Smad complex opposes p63 to empower TGFbeta-induced and predicts disease relapse. Cell Stem Cell 2011;8:511–24. metastasis. Cell 2009;137:87–98. 26. Liu R, Wang X, Chen GY, Dalerba P, Gurney A, Hoey T, et al. The prognostic 37. Barbieri CE, Tang LJ, Brown KA, Pietenpol JA. Loss of p63 leads to increased role of a gene signature from tumorigenic breast-cancer cells. N Engl J Med cell migration and up-regulation of genes involved in invasion and 2007;356:217–26. metastasis. Cancer Res 2006;66:7589–97.

OF12 Cancer Res; 78(2) January 15, 2018 Cancer Research

Distinct TP63 Isoform-Driven Transcriptional Signatures Predict Tumor Progression and Clinical Outcomes

Hussein A. Abbas, Ngoc Hoang Bao Bui, Kimal Rajapakshe, et al.

Cancer Res Published OnlineFirst November 27, 2017.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-17-1803

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2017/11/23/0008-5472.CAN-17-1803.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2018/01/08/0008-5472.CAN-17-1803. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.