A Mutant P53-Dependent Embryonic Stem Cell Gene Signature Is Associated with Augmented Tumorigenesis of Stem Cells

Author Manuscript Published OnlineFirst on August 28, 2018; DOI: 10.1158/0008-5472.CAN-18-0805 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

A mutant p53-dependent embryonic stem cell gene signature is associated with augmented tumorigenesis of stem cells

Gabriela Koifman1, Yoav Shetzer1, Shay Eizenberger1, Hilla Solomon1, Ron Rotkopf2, Alina Molchadsky1, Giuseppe Lonetto1, Naomi Goldfinger1, and Varda Rotter1,*.

1Dept. of Molecular Cell Biology, 2 Bioinformatic unit, Life Sciences Core Facilities, the Weizmann Institute of Science, Rehovot 7610001, Israel.

Running title: An oncogenic mutant p53-dependent embryonic gene signature

*Corresponding author

Varda Rotter, Ph.D.

Department of Molecular Cell Biology

Weizmann Institute of Science

Rehovot 76100, Israel

TEL: 972-8-9344501

FAX: 972-8-9465265

E-mail: [email protected]

The authors declare no potential conflicts of interest

Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 2018 American Association for Cancer Research. Author Manuscript Published OnlineFirst on August 28, 2018; DOI: 10.1158/0008-5472.CAN-18-0805 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Abstract

Mutations in the tumor suppressor p53 are the most frequent alterations in human cancer. These mutations include p53-inactivating mutations as well as oncogenic gain-of- function (GOF) mutations that endow p53 with capabilities to promote tumor progression. A primary challenge in cancer therapy is targeting stemness features and cancer stem cells (CSC) that account for tumor initiation, metastasis, and cancer relapse. Here we show that in vitro cultivation of tumors derived from mutant p53 murine bone marrow (BM) mesenchymal stem cells (MSC) gives rise to aggressive tumor lines (TL). These MSC-TL exhibited CSC features as displayed by their augmented oncogenicity and high expression of CSC markers. Comparative analyses between MSC-TL with their parental mutant p53 MSC allowed for identification of the molecular events underlying their tumorigenic properties, including an embryonic stem cell (ESC) gene signature specifically expressed in MSC-TL. Knockout of mutant p53 led to a reduction in tumor development and tumorigenic cell frequency, which was accompanied by reduced expression of CSC markers and the ESC MSC-TL signature. In human cancer, MSC-TL ESC signature-derived genes correlated with poor patient survival and were highly expressed in human tumors harboring p53 hotspot mutations. These data indicate that the ESC gene signature-derived genes may serve as new stemness-based prognostic biomarkers as well as novel cancer therapeutic targets.

Keywords: Mesenchymal stem cells, Cancer stem cells, Mutant p53, Embryonic stem cell signature, Gain Of Function.

Introduction Tumor development is a multistage process that is accompanied by gradual gain of intra- tumoral heterogeneity [1, 2]. One of the theories accounting for tumor heterogeneity is the existence of a rare sub-population of cells known as the CSCs that reside within tumors and harbor a unique ability to regenerate a tumor and capacities to metastasize and to resist chemotherapy [2]. In order to study the nature of these rare CSCs, several methods for their isolation were described, mainly by identification of specific CSC surface markers [3]. Although previous studies identified different CSC markers in various tumors, novel and more broadly expressed markers, that can be useful for early diagnostic and even for cancer therapy, are still urgently needed.

Mutations in the p53 gene are the most frequent alterations in human tumors [4, 5] that correlate with undifferentiated high-grade tumors [6-8]. Most of the p53 mutations are missense mutations that produce full-length mutant p53 proteins, which not only lack the p53 tumor-suppressor activity, but also gain new oncogenic functions that acquire the cell with enhance malignant properties [4, 5]. Ample reports indicate a possible mutant p53 GOF activity in the acquisition of de-differentiation and stemness phenotype that might lead to tumorigenesis. For example, mutant p53 GOF activity was shown to enhance the reprogramming efficiency of mouse embryonic fibroblasts (MEFs) and augment the tumorigenesis capacity of the reprogrammed cells [9]. Moreover, mutant p53 accumulation specifically in neuronal progenitors led to gliomagenesis, indicating improper maturation of neural stem cells by mutant p53 [10]. Notably, BM-MSCs harboring p53 mutation undergo malignant transformation that induces sarcomagenesis [11]. All in all, these studies imply that mutant p53 might have a GOF activity in stem cell transformation and de-differentiation but the molecular profile underlying these processes is still poorly understood.

Here we show that in-vitro cultivation of mutant p53 BM-MSCs derived tumors, led to the establishment of aggressive MSC tumor lines (TLs). These MSC-TLs exhibited an enhanced tumorigenic capacity compared with their parental cells, mutant p53 MSCs, as shown by their ability to form rapidly growing tumors by the injection of as few as 1×102 cells. By transcriptome profiling, we found that MSC-TLs express elevated levels of CSC

markers and a unique gene signature consisting of ESC genes. These results suggest that MSC-TLs may represent a mesenchymal CSC-like population that reside within mutant p53 MSCs derived tumors. Knocking-out mutant p53 significantly reduced tumor development and tumorigenic cell frequency and also resulted in a reduction in the expression of CSC markers and the identified ESC MSC-TLs signature. Importantly, analyses of human tumor datasets showed that the ESC MSC-TLs signature derived genes correlated with poor patient survival and were highly expressed in various human cancers harboring p53-hotspot missense mutations. Altogether, our data suggest a novel role of mutant p53 in expanding mesenchymal CSC-like cells that display expression of a unique mutant p53-dependent gene signature comprising of ESC genes. The fact that this mesenchymal CSC-like signature contains embryonic genes, which are not tissue specific, might suggests that these signature derived genes may serve as prognostic markers and potential cancer stemness therapeutic targets of mutant p53-dependent tumors and tumors at large.

Materials and Methods

Cells

Bone marrow MSCs were isolated from p53 WT or p53 Mut (R172H) mice and characterized as previously described [11]. The different MSC isolates were grown in MSC specific medium (MesenCultTM MSC Basal Medium (Mouse), STEMCELL Technologies) supplemented with 20% murine MesenCultTM MSC stimulatory supplement (Mouse) (STEMCELL Technologies), 60 mg/ml penicillin, 100 mg/ml streptomycin and 50 mg/ml kanamycin. p53 Mut MSC-TLs were established as follows; tumors were extracted and mechanically disaggregated by utilizing a cell strainer (542070, Greiner Bio-One). The isolated MSCs as mentioned previously [11], were examined for their authentication by the examination of their ability to differentiate to the mesodermal lineages adipocytes and osteocytes and also by verifying that the cells do not express hematopoietic markers. Tumor derived cells were culture on agar coated plates in DMEM medium supplemented with 15% (vol/vol) FCS, 1 mM sodium pyruvate, 2 mM L-glutamine, 0.1 mM nonessential amino acids, 0.1 mM β-mercaptoethanol, 1,000 units/mL leukemia inhibitory factor (ESG1107; [BD] at a ratio of 1:1). Cells were

incubated at 37˚ in a humidified atmosphere of 5% CO2. All cells were checked for mycoplasma contamination and in all the analyses, we verify that the cells did not exceeded 20 in-vitro passages.

Mice

C57BL/6 harboring p53 WT or p53 Mut (R172H) [12], kindly provided by Professor Guillermina Lozano (MD Anderson Cancer Center, Houston), Athymic Nude-Foxn1nu and NOD.CB17-Prkdcscid/NCrHsd (ENVIGO). Animal protocols were approved by the Institutional Animal Care and Use Committee of the Weizmann Institute of Science. CFU-fs Assay

Freshly bone marrow nucleated cells derived from two p53 WT or two p53 Mut (R172H) mice were plated at cell density of 5×106 in 10 cm BD falcon plates (BD). The cells were grown in MSC specific medium as described above and re-fed once a week without further treatment. After 14 days, colonies of triplicates were stained with Giemsa and counted. Cell proliferation assay

Cells were counted and plated (25,000 cells per well for 12 well plate or 140,000 cells for 6cm well plate). Each day cells in duplicate wells or plates were tripsinized and stained with Trypan blue, unstained cells were counted by using hemocytometer and light microscopy. The proliferation plots present results relative to day 1. Each curve represents p53 WT MSC or p53 Mut pMSC isolate or derived p53 Mut MSC-TL accordingly.

Western blot analysis Cells were lysed in TLB buffer (50 mM TrisHCl, 100 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with Protease Inhibitor Coctail (Sigma- Aldrich) for 15 minutes on ice, followed by centrifugation. Protein concentration was determine using BCA reagent (Thermo Fisher scientific), 30-100 μg of protein extracts were separated by SDS-gel electrophoresis, and transferred to a nitrocellulose membrane at semi-dry conditions. Membranes were blocked using 5% dry milk in PBST (0.05% Tween in PBS), and incubated with anti-p53 (1c12, Cell Signaling Technology), anti-p21 (sc-397, Santa Cruz Biotechnology), anti-β-Actin (sc-47778, Santa Cruz Biotechnology),

anti-Sox2 (ab97959, ABCAM), anti-Hmga1 (ab129153, ABCAM), anti-Slc16a1 (GTX54699, GeneTex) and anti-Etv5 (GTX5114394) followed by appropriate horseradish peroxidase-conjugated secondary antibodies and the signal was obtained by ECL western blotting detection reagent (Thermo Fisher Scientific) and ChemiDoc MP (Bio-rad, Hercules, CA, USA). In all the experiments, β-Actin or GAPDH were used as a loading control. Multispectral imaging flow-cytometry (IFC) analysis For all the experiments, cells were imaged using multispectral imaging flow cytometry (ImageStreamX mark II imaging flow-cytometer; Amnis Corp, Seattle, WA, Part of EMD Millipore). Data was analyzed using image analysis software (IDEAS 6.2; Amnis Corp). Images were compensated for fluorescent dye overlap by using single-stain controls. Cells were gated for single cells, using the area and aspect ratio features, and for focused cells, using the Gradient RMS feature, as previously described [13]. For p53 staining, the cells were fixed with a fixation buffer (80% Ethanol, 20% HBSS), washed and stained with a primary p53 antibody (Cell signaling, 1C12) (1h, 4oC, gentle agitation) and then washed once with PBS-/- and DAPI for DNA staining. Nuclear staining p53 was verified using the Similarity feature (Calculated as log transformed Pearson’s correlation coefficient between the DAPI and p53 staining). For cell cycle analysis, we utilized Phase-Flow™ BrdU Cell Proliferation Kit (Biolegend, CA 92121). Cells after staining with antibody against BrdU were analyzed by multispectral imaging flow cytometry and the percentage of cells in each cell cycle phase was determined according to level of staining intensity. For CD44 staining, cells were incubated with the primary antibody for to detect CD44 (eBioscience, Clone IM7, 17-0441-81) (1h, 4oC, gentle agitation), washed once with PBS-/- and Hoechst for DNA staining. Cells were imaged using the multispectral imaging flow cytometry Approximately 1x104 cells were collected from each sample and Cells were then gated for high intensity of CD44 staining and their percentage was calculated and compared between samples. To identify small and circular cells, the area (The number of microns squared in a mask) and circularity (the degree of the object’s deviation from a circle) were calculated (three isolates each in each group). The population of cells with low area and high circularity was gated and their percentage was

calculated and compared between samples. CellTiter-Glo Luminescent Cell Viability Assay Cells were counted and plated (2,000 cells per well for 96 well plate). Each day Cell Titer-Glo kit reagent was added directly to cells (in triplicates) that lead to cell lysis and generation of a luminescent signal. The signal is proportional to the amount of ATP which is considered to be directly proportional to the presence of metabolically active cells in the culture.

Allograft tumor formation assay

Cells were tripsinized, stained with Trypan blue (Sigma-Aldrich T8154) and live cells (unstained) were counted by using hemocytometer. Further, cells were resuspended with PBS containing 1% of fetal calf serum and injected subcutaneously into Athymic Nude- Foxn1nu or NOD.CB17-Prkdcscid/NCrHsd (ENVIGO) between 6-8 week of age. Bioluminescence in-vivo imaging Cells were infected with pBABE-hygro retroviral vector expressing luciferase, in order to monitor tumor growth. Immunodeficient mice were subcutaneous injected with 1×102 or 1×103 infected cells into the nape of the neck. After 1 week, from the inoculation, mice were anesthetized and injected with luciferin and monitored by IVIS2000 system. Exposure time was 30 seconds for 20 images that were taken for each tumor. We selected the highest peak of luminescence values and normalized it to adjacent background. Tumors were allowed to reach to maximal size of 1 cm2 and then mice were sacrificed. Immunohistochemistry staining 4 µm paraffin embedded tissue sections were deparaffinized, rehydrated and pretreated for antigen retrieval. For Sox2, CD44 and PCNA the antigen retrieval was done in citric acid (pH 6), for 10 minutes, by microwave. For ALDH1 and CD34 staining the antigen retrieval was done in ice cold Acetone for 7 minutes, then washed in PBS and additional 10 minute in citric acid (pH 6) by microwave. Sections were than blocked and incubated with primary antibodies for at least 24 hours as in 4˚c, as follows; PCNA (FL-261) (sc- 7907, Santa cruz biotechnology), ALDH1 (AF5896 R&D system), Sox2 (ab97959, ABCAM), CD44 (553131, BD pharmingen) and CD34 (ACL8927AP, Accurate). Sections were washed and incubated with appropriate secondary biotinylated IgG and

avidin-biotin complex (Elite-ABC kit, Vector lab, CA, USA) followed by DAB reaction (Sigma-Aldrich). CD34 antibody incubation followed by incubation with secondary Cy3 conjugated antibody (1:100, Jackson ImmunoResearch, West Grove, PA) for 30-60 min and with Dapi for nuclear staining. Quantification of the staining was done utilizing Image Pro Plus software. RNA sequencing and analysis Three independent p53 Mut MSCs isolates and two individual p53 Mut MSC-TLs derived from primary tumors of each MSC isolate, as well as additional four p53 Mut MSC-TLs derived from secondary tumors (overall ten p53 Mut MSC-TLs) were subjected to RNA sequencing analysis. Cells were plated and when they reach to 80% confluent, RNA was extracted using Direct-zolTM RNA MiniPrep (ZYMO RESEARCH). Libraries preparation were done using the INCPM-RNA-seq. Briefley, polyA fraction (mRNA) was purified from 500ng of total RNA following by fragmentation and generation of double stranded cDNA.Then, end repair, A base addition, adapter ligation and PCR amplification steps were performed. Libraries were evaluated by Qubit and TapeStation. Sequencing libraries were constructed with barcodes to allow multiplexing of 22 samples in two lanes. 18 million single-end 60-bp reads were sequenced per sample on Illumina HiSeq 2500 V4 instrument. Reads were trimmed using cutadapt (http://dx.doi.org/10.14806/ej.17.1.200) and mapped to mm10 genome (downloaded from igenomes) using TopHat (v2.0.10) (http://dx.doi.org/ 10.1186/gb-2013-14-4-r36). Counting over refseq genes proceeded using htseq-count (http://dx.doi.org/10.1093/bioinformatics/btu638) (intersection-strict mode). Differential expression analysis was performed using DESeq2 (http://dx.doi.org/10.1186/s13059-014-0550-8) with the betaPrior, cooks Cutoff and independent filtering parameters set to False. The data presented in this study have been deposited in NCBI’s Gene Expression Omnibus [14] and are accessible through GEO series accession number GSE118173 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE118173)

Reverse transcription and QRT-PCR

RNA was extracted using Direct-zolTM RNA MiniPrep (ZYMO RESEARCH). Reverse transcription and QRT-PCR were performed as previously described [11]. The sequences of the specific gene primers are listed in Supplementary table 1. Knockout of mutant p53 by CRISPR technology Cells were transfected with Px330 backbone expressing sgRNA for targeting murine p53 and Cas9 by GeneExpressoTM in vitro DNA transfection reagent (EG-1031, Excellgen). The Px330 p53 plasmid was a gift from Tyler Jacks (Addgene plasmid 59910). After 48 hours from the infection, we plated 0.5 cell per well in a 96 wee plate in order to generate single cell clones. After the generation of single cell clones, we analyzed all cloned by western analysis in order to detect p53 knockout clones. 2 clones out of 71 clones were KO for mutant p53. We verified by DNA sequencing the DNA editing by the guided RNA molecule. Obtaining ESC gene sets Three gene sets were collected form independent studies. First, a previous study from our lab performed a RNA sequencing analysis on ESCs upon differentiation induction by retinoic acid [15]. We re-analyze the raw data and genes that were significantly downregulated upon differentiation (padj<0.05, FC<2, max reads>10) were selected as the ESC genes list. Second study, performed bioChIP-Chip analysis [16] we selected shared target genes of Nanog, Sox2 and Oct4. A third dataset was extracted from a study that performed a time course microarray experiment on reprogramed MEFs into induce pluripotent stem cells (iPSCs) by infecting with retoviruses encoding for the four yamanaka reprogramming factors (Oct4, Sox2, klf4 and c-myc). We selected genes that had expression levels more twofold after 8 days of the reprogramming process compared to day o and excluded genes that were upregulated in the control (Infection with retrovirus encoding for GFP) [17]. Gene-set enrichment analysis First, for each comparison, we estimated the significant genes (padj<0.05, FC>2, max reads>10). Moreover, from each dataset, we estimated the number of genes that had at least 10 reads (max) in our RNA sequencing analysis and analyzed if those lists are enriched in the differential gene expression in every comparison. We utilized The GeneProf hypergeometric probability calculator

(https://www.geneprof.org/GeneProf/tools/hypergeometric.jsp) to estimate the enrichment p-value of the different gene sets. GSEA was also performed by using GSEA software (http://software.broadinstitute.org/gsea/index.jsp). The enrichment of specific datasets was considered significant when the nominal (NOM) p value and false discovery rate (FDR) q-value were less than 0.05.

TCGA data analysis Survival analysis. Data for the survival analyses was downloaded from TCGA via "Xena". We examined whether the combined expression of each possible gene pair from the human orthologues of the MSC-TLs ESC gene list, affected patient survival by Cox proportional-hazards model via the R package `survival`. We classified the expression level per gene per sample as "high" or "low", using median of each gene value as cutoff. Next, In order to distill the genes that can predict poor prognosis, we first examined the combined expression of each possible gene pair from the ESC gene list and selected the pairs which significantly predicted poor survival in patients in each cancer type. (FDR<0.05, Hazard ratio>2). Next, we selected the significant genes derived from the pairwise survival analyses and performed multigene survival analyses via backward stepwise selection and arrived at a final survival model based on gene signatures that predicted poor patient survival. Next, we selected the most significant genes derived from the latter gene signatures and examined whether the combine expression of the selected genes can predict poor patient survival (by Cox proportional-hazards model). Pan-Cancer expression analysis. Mutation and expression (RNA-Seq) data was downloaded from TCGA (The Cancer Genome Atlas) via the R package ‘CGDSR’. We used the subset “3_way_complete” which contains all tumor samples that have mRNA, CNA and sequencing data. Differential expression between groups was tested using a t- test on log-2 transformed normalized counts. The frequency of the gene expression was calculated and normalized to the overall gene expression by Chi-squared test. The results shown here are based upon data generated by TCGA research network: http://cancergenome.nih.gov.

Results

Mutant p53 enhanced the self-renewal capacity and oncogenicity of primary bone marrow MSCs In order to examine the role of mutant p53 in stem cells transformation, we utilized a system containing BM-MSCs isolates harboring WT p53 or p53 R172H mutation, which is analogous to the frequent and known GOF hotspot mutation, R175H in human tumors [11]. In order to corroborate whether mutant p53 MSCs isolates display loss of WT p53 activity, we examined the DNA damage response. As expected, upon Doxorubicin treatment, p53 wild-type (p53 WT) MSCs showed an increase in nuclear p53 protein levels (Figures 1A, 1B, supplementary Figure 1A) that was accompanied by the elevation of its target gene, p21 (Figures 1A, C). In agreement with previous studies [4] mutant p53 was already accumulated in basal levels, regardless of the DNA damage response and there was no detection of p21 expression in p53 mutant (p53 Mut) MSCs upon Doxorubicin treatment (Figures 1A-C). Consistently, p53 Mut MSCs showed a higher proliferation rate compared with p53 WT MSCs (Figure 1D). Therefore, no genuine p53 activity was detected in p53 Mut MSCs. As mentioned above, mutant p53 promote cancer development [4, 5]. We further examined the tumorigenic potential. In agreement with our previous study [11], while all p53 Mut MSC isolates were able to form aggressive sarcomas, p53 WT MSCs lacked this ability (Supplementary Figure 1B-D). Tumor initiating capacity is ascribed to uncontrolled self-renewal pathways and acquisition of CSCs properties [18]. Next, Colony-Forming Unit-fibroblastic (CFU-Fs) assay showed that p53 Mut mice exhibited higher number of BM stem cells compared with p53 WT mice (Figure 1E), indicating a higher self-renewal ability of p53 Mut BM derived stem cells. Overall, mutant p53 enhanced the proliferation, tumorigenic potential and the self-renewal of BM-MSCs. Mutant p53 MSC derived tumor lines exhibited augmented tumorigenic capacity compared with their parental cells Tumor intra-heterogeneity that has been accounted for tumor development, is attributed to the existence of multiple sub-clones within a tumor that have different tumorigenic abilities [1]. In order to propagate the tumorigenic cells within p53 Mut MSCs derived tumors, we cultivated these tumors in-vitro and established various MSC-TLs. p53 Mut MSC-TLs exhibited higher proliferation and viability capacity compared with their p53

Mut parental MSCs (pMSCs) (Figure 2A, Supplementary Figure 2A). Consistently, as expected, cell cycle analysis revealed lower percentage of cells in the G1 phase and higher percentage of cells in the S phase in p53 Mut MSC-TLs compared with their parental cells, p53 Mut pMSCs (Supplementary Figure 2B-C). Furthermore, we found that p53 Mut MSC-TLs derived cells are smaller in size and more circular compared with their parental cells (Figures 2B, Supplementary Figures 2D-F). Interestingly, Stem cells and iPSCs are characterized by high proliferation rate, small cell size and circular morphology [19, 20]. In order to explore the possibility that the MSC-TLs exhibit stem cell features, we examine CD44 expression, a common marker for CSCs in various cancer models [21, 22]. The percentage of CD44high subpopulation in the p53 Mut MSC- TLs was higher compared with their parental cells (Figures 2C). Next, an in-vivo serial dilution transplantation assay showed higher tumor initiation capacity and tumorigenic cell frequency in p53 Mut MSC-TLs compared with their parental cells (Figures 2D). Strikingly, as few as 1×102 cells from p53 Mut MSC-TLs were able to generate tumors within an average of 34.7 days. Whereas, subcutaneous injection of 3×106 p53 Mut pMSCs gave rise to tumors within 97 days (mean) (Figure 2E, Table 1). Consistently, in- vivo imaging enabled the detection of tumors, following subcutaneously injection of 1×102 or 1×103 cells derived from p53 Mut MSC-TL, already within 14 days. Injection of the same amount of pMSCs yielded no tumor development (Figures 2F, Supplementary Figure 2G), even following six months. Furthermore, tumors derived from MSC-TLs displayed peripheral nerves and blood vessels invasion, reflecting aggressive features that were not detected in the pMSC derived tumors (Figure 2G). MSC-TLs derived tumors also demonstrated higher expression of the proliferative marker, PCNA, and of another common CSC marker [23], aldehyde dehydrogenase isoform 1(ALDH1), compared with the pMSCs derived tumors (Figures 2H, Supplementary Figure 2H-I). Altogether, these results suggest that upon cultivation of mutant p53 pMSC derived tumors, we enriched for highly tumorigenic cell lines with CSC features.

Embryonic gene signature expressed in mutant p53 MSC-TLs

To identify gene networks that underlie the accentuated oncogenic activity of p53 Mut MSC- TLs, we subjected three independent parental p53 Mut MSCs isolates, two individual p53

Mut MSC-TLs derived from primary tumors of each MSC isolate, as well as additional four p53 Mut MSC-TLs derived from secondary tumors, which exhibited similar tumorigenic capacity as the primary MSC-TLs (Supplementary figure 3A) to transcriptome profiling. While the principal component analysis (PCA) and Hierarchical clustering revealed transcriptional variances between individual p53 Mut MSC-TLs, the largest transcriptome variation was observed between p53 Mut pMSCs and their derived MSC-TLs (Supplementary Figure 3B-C). This finding indicates a tumorigenic transcriptional core that is shared between all p53 Mut MSC-TLs. The aggressive nature of p53 Mut MSC-TLs led us to examine whether they express a CSC signature. We utilized a define murine CSC signature derived from head and neck Squamous cell carcinoma [24]. We found that this CSC signature was enriched (p-value=2.2×10-16) and the CSC signature derived genes were more frequent in the upregulated gene fraction of p53 Mut MSC-TLs compared with their parental cells (Chi-squared test p-value=0.0003) (Figure 3A, supplementary table 2). Next, a Gene Set Enrichment Analysis (GSEA) showed that p53 Mut MSC-TLs signature was enriched for stemness and undifferentiated cancer gene sets (Figure 3B). Epithelial to Mesenchymal Transition (EMT) was shown to be associated with acquisition of stemness and CSC phenotype [25]. we further found that an EMT gene signature is enriched in p53 Mut MSC-TLs compared with their parental cells, p53 Mut pMSCs (Figure 3B). Analyzing the genes that were upregulated in p53 Mut MSC-TLs compared with their parental cells, by Ingenuity Pathway Analysis (IPA) tool [26] linked them to biological functions including cell movement, proliferation and survival. Whereas, this analysis showed downregulation of genes associated with organismal death and growth failure (Figure 3C, Supplementary table 3). The activated canonical pathways in the p53 Mut MSC-TLs, according to IPA, unveiled cancer associated pathways including Paxilin, Gα12/13, PAK, Integrin and CDK5 signaling (Supplementary table 3). Interestingly, IPA also highlighted the activation of stem cell upstream regulators in the p53 Mut MSC-TLs that are involved in different developmental pathways, including Wnt, Hedgehog and Notch pathways (Figure 3D, Supplementary table 3). Furthermore, the activated upstream regulators also included oncogenic pluripotency regulators, Myc and Klf4, as well as embryonic stem cells regulators including EZH2 and BRG1. On the contrary, upstream regulators that were

inhibited in the p53 Mut MSC-TLs comprised of tumor-suppressors including CDKN1B, TSC2, STK11, mir-141-3p and mir-34a-5p (Figure 3D, Supplementary table 3). Notably, mir-34-5p was shown to negatively regulate the Wnt signaling pathway [27, 28]. Furthermore, IPA analysis, identified, as the highest scored network in p53 Mut MSC- TLs, a gene network associated with cellular, embryonic and organismal development (Supplementary Figure 3D, Supplementary table 3). To further examine these observations, we obtained murine ESC signature datasets from three independent published studies (Supplementary table 2). A Hypergeometric probability test was aimed to examine whether the differential genes between p53 Mut MSC-TLs and p53 Mut pMSCs, contained higher number of genes than randomly expected, from a particular ESC dataset, indicating ESC dataset enrichment. To that end, we obtained a dataset from a previous comparative analysis of p53 WT murine ESCs before and after retinoic acid induced differentiation [15]. We focused on genes that were downregulated during ESCs differentiation, since these genes, most likely, represent the ESC expression signature. This gene set was enriched in p53 Mut MSC-TLs and included 129 overexpressed genes as compared with the parental MSCs (p-value=0.02). A second gene set was extracted from a published study identifying target genes of embryonic transcription factors [16]. Shared target gene list of the ESC transcriptional factors; Nanog, Sox2 and Oct4, yielded a significant enrichment score in the up-regulated gene fraction of p53 Mut MSC-TLs compared with their parental cells (p-value=0.005). Finally, we analyzed a geneset comprising of genes that were upregulated upon induction of iPSCs [17]. This list contained genes that showed, at least, 2-fold increase in their expression following eight days of pluripotency induction. Consistently, the reprogramming geneset contained 39 overexpressed genes in the p53 Mut MSC-TLs and yielded a significant enrichment score (p-value=1.23×10-7). However, enrichment analysis of the down-regulated genes in the p53 Mut MSC-TLs compared with their parental cells did not yielded a significant p-value in any ESC dataset. Overall, we detected 184 ESC genes derived from the three datasets that were upregulated in the p53 Mut MSC-TLs (Figure 3E, Supplementary table 2). The ESC transcription factor, Sox2, known to be essential for maintaining pluripotency of ESCs, was shared by all the ESC datasets and was highly expressed in the p53 Mut MSC-TLs as compared with their parental cells (Figure 3F-G). The

expression of additional ESC signature derived genes were validated (Supplementary Figure 3E). Interestingly, ESC gene signature derived genes that were downregulated in p53 Mut MSC-TLs were associated with promoting apoptosis and attenuating cell proliferation (Supplementary figure 3F). Overall, p53 Mut MSC-TLs highly expressed cancer and CSC associated genes and an ESC gene signature.

Mutant p53 knock-out reduced tumor initiating capacity of mutant p53 MSC-TL sub-clones.

The gain of oncogenic activities by mutant p53 that promote tumor developmet is a well- accepted notion [4, 5]. To determine whether the enhanced tumorigenic capacity of MSC-TLs is mutant p53 dependent, we knocked-out mutant p53 in a MSC-TL (Supplementary Figure 4A). Knock-out (KO) of mutant p53 resulted in a significant reduction in tumor initiating capacity, tumor development and tumorigenic cell frequency (Figures 4A-D). Tumors derived from p53 KO MSC-TL sub-clones were significantly smaller compared with tumors derived from p53 Mut MSC-TL sub-clones, exemplified by tumor weight and size (Figure 4A-B). Furthermore, tumor latency was longer upon mutant p53 KO (Figure 4C). Notably, injection of as low as 1×102 showed significant reduction in the incidence of tumors upon KO of mutant p53 (Figure 4D). Furthermore, injection of 10 cells, yielded tumors only upon injection of mutant p53 expressing MSC-TL sub-clones (Figure 4C-D), indicating a mutant p53 GOF activity in expanding tumorigenic cell population. Tumors derived from mutant p53 KO MSC-TL sub-clones displayed wider area of tumor necrosis, as well as a reduction in tumor vascularization, as determined by CD34 staining, compared with tumors derived from mutant p53 MSC-TL sub-clones (Figure 4E, Supplementary Figure 4B). This phenomenon is in line with the notion that mutant p53 exhibits a GOF in angiogenesis [4, 29]. Furthermore, upon mutant p53 KO there was a significant reduction in the expression of the two CSC markers, CD44 and ALDH1, that were shown to be highly expressed in p53 Mut MSC-TLS compared to their parental cells (Figures 4F-G). Altogether, these results suggest that mutant p53 enhances tumor growth, angiogenesis, survival, expression of CSC markers and most importantly expansion of tumor initiating cells within the MSC-TL.

Mutant p53 knock-out led to a reduction in the expression of the ESC gene signature expression in MSC-TL sub-clones.

Next, we examined whether the reduction in tumorigenicity following mutant p53 KO is associated with a decrease in the ESC gene signature expression, described above. Therefore, we subjected mutant p53 MSC-TL and the corresponding mutant p53 KO MSC-TL sub-clones to a transcriptome profiling by RNA sequencing (Supplementary table 4). IPA analysis of the differential genes upon KO of mutant p53 in the MSC-TL revealed increase of biological functions such as organismal death, morbidity and mortality and decrease of cell migration and proliferation (Figure 5A, Supplementary table 5). Interestingly, we identified inactivation of several cancer related pathways upon mutant p53 KO. These included Integrin, Paxilin, Gα12/13, PAK and CDK5 signaling pathways that were previously shown to be activated in the MSC-TLs compared to their parental cells. Additional cancer related pathways that were inhibited upon mutant p53 KO included RhoA, CXCR4, PDGF, ILK and VEGF signaling pathways. Mutant p53 abolishment also led to the downregulation of genes associated with mouse ESC pluripotency pathway (Supplementary table 5). Finally, we detected downregulation of two reprogramming regulators Myc and Klf4 following mutant p53 depletion (Figure 5B). Furthermore, we found that the p53 Mut MSC-TLs ESC signature was significantly downregulated upon mutant p53 abolishment (p-value=5×10-8) (Figure 5C, Supplementary table 4). Notably, the ESC genes, shown to be downregulated upon retinoic acid differentiation induction [15], yielded a further enrichment in the downregulated genes upon mutant p53 knock-out (P-value=5.9×10-9) (Figure 5D, Supplementary table 4). The Polycomb Repressive Complex 2 (PRC2) regulates the pluripotency of ESCs by repressing developmental associated genes [30]. we found that a gene set of PRC2 repressed target genes [31] was enriched and more frequent in the upregulated gene fraction following mutant p53 KO (Enrichment p-value=1×10-15, Chi-squared test p- value=0.02) (Figure 5E-F, Supplementary table 4). This might suggest that in p53 Mut MSC-TL sub-clones PRC2 is active, thus contributing to the ESC signature expression phenotype. Moreover, the expression of Sox2, previously shown to be elevated in the different MSC-TLs compared with their parental cells, was downregulated following

mutant p53 KO (Figures 5G-H). We validated the downregulation of additional ESC signature derived genes following mutant p53 KO (Figure 5G-H, Supplementary Figure 5A). By utilizing BindDB tool [32], we showed that the proximal promoters of mutant p53 MSC-TL dependent genes are enriched for chromatin transcriptional activation marks in ESCs (Supplementary figure 5B). These findings suggest that mutant p53 induces the expression of ESC gene signature in MSC-TLs. The MSC-TLs ESC signature derived genes are associated with poor patient survival and correlate with human tumors harboring p53 missense mutations. To assess the relevance of the murine MSC-TL ESC signature in human cancer patients, we utilized datasets provided by The Cancer Genome Atlas (TCGA). We focused on the human orthologues of ESC gene signature, that was detected in the MSC-TLs, and additional mutant dependent ESC genes. We identified unique gene signatures, whose combined expression predicted poor patient survival in several cancer types (Figure 6A, Supplementary Figure 6, Supplementary table 6, detailed in material and methods section). Of note, a single gene survival analysis yielded lower hazard ratios compared with the combined expression of genes, indicating the importance of analyzing the combined expression of genes (Supplementary table 6). Furthermore, the combined expression of four transcription factors, derived from the MSC-TLs ESC gene signature, E2F2, HMGA1, SOX2 and ZIC2 predicted poor patient survival in various cancer types (Supplementary Figure 6, Supplementary table 6). To explore a correlation between p53-hotspot mutations and the expression of the ESC gene signature we performed a Pan-cancer gene expression analysis according to TCGA datasets. Datasets derived from different cancer types that includes tumor samples with a characterized p53 status were studied (Supplementary table 7). As previously described, missense mutations in the p53 gene are the most common mutations found in human cancers and usually occur in six hotspot amino acids in the DNA binding domain of the protein. These mutations produce full-length mutant p53 proteins that were shown to exhibit oncogenic GOF [4, 5]. Similar to the IARC p53 database, these six hotspot mutations were highly frequent in the TCGA database (Supplementary Figure 7). Of note, we compared between tumor samples harboring one of the p53-hotspot missense mutations and the entire tumor samples that harbor other p53 mutations or WT p53. We

assumed that tumors expressing WT p53 probably entailed inactivation of p53 associated pathways, due to mutations in other downstream genes [33]. The frequency of highly expressed MSC-TLs ESC genes in tumors harboring hotspot missense p53 mutations was significantly higher compared with the rest of tumor samples (Figure 6B, Supplementary table 8) (Chi-squared test p-value=0.01). By analyzing a previously described human ESC signature [30], we found a higher proportion of human ESC derived genes that were highly expressed in tumors harboring one of the six p53-hotspot missense mutations compared with the rest of tumor samples (Chi-squared test p-value=0.004) (Figure 6C and Supplementary table 8). Importantly, SOX2 was upregulated in p53-hotspot missense mutations tumor samples. Box plot of SOX2 and additional ESC genes are presented (Figure 6D, Supplementary Figure 8). These analyses indicated that like in our mouse model, expression of mutant p53 proteins in human tumors correlated with the expression of the presently described p53 Mut MSC-TLs associated ESC gene signature and human ESC signature. Discussion p53 mutations are the most common alterations in human tumors that lead to the gain of oncogenic traits, which promote cancer development [4, 5]. CSCs were suggested to be an important milestone underlying oncogenicity. The intrinsic stemness potential of adult stem cells, together with their long life span, enabling mutations accumulation, make them suitable candidates to be the cell of origin of CSCs. p53 was shown maintaining the normal equilibrium between self-renewal and differentiation of stem cells [21, 34]. Various studies support the notion that loss of WT p53 in stem cells leads to tumor initiation [21]. Here, we demonstrate that BM-MSCs derived from p53 mutant mice exhibit an augmented sarcomagenesis and higher self-renewal capacity. This is in agreement with previous studies showing that p53 preserves the genomic integrity of MSCs and prevents their malignant transformation [11, 35]. Cultured MSCs, similarly to cancer cells, exhibit intra-population heterogeneity with variable proliferation and differentiation capacities and display a repertoire of surface markers [36]. Accordingly, the low frequency of tumorigenic cells seen within the parental p53 Mut MSCs may account for intra-population heterogeneity with a restricted number of tumorigenic cells within the population (Figure 2D). Our data shows that by culturing mutant p53 MSC

derived tumors, we enriched for aggressive TLs. These MSC-TLs exhibited CSC expression markers and an augmented tumorigenic capacity, by the injections of as few as 1×102. Notably, under the same conditions, no tumors were evident upon the injection of their parental p53 Mut MSCs. Transcriptome analysis revealed that the augmented tumorigenic capacity of the established p53 Mut MSC-TLs was associated with the expression of CSC gene signature and a unique gene signature consisting of ESC genes. ESC genes that were downregulated in p53 Mut MSC-TLs were linked with positive regulation of apoptosis and cell proliferation attenuation. These results indicate that the aggressive p53 Mut MSC-TLs adopt only a part of the ESC signature, that favors tumorigenesis, but suppress the part of the ESC signature that negatively regulates the tumorigenic processes. MSC-TLs derived from identical parental MSCs were clustered together. This transcriptional clustering and the shared tumorigenic phenotype by all p53 Mut MSC-TLs in conjunction with intra-population heterogeneity in MSCs [36] might indicate the existence of a common origin within the mutant p53 MSC population rather than a stochastic event. Overall, these results suggest that in-vitro cultivation of the mutant p53 MSC derived tumors indeed selected for a CSC-like population that reside within MSC derived tumors. This led to the identification of a mesenchymal CSC gene signature. p53 inactivation in human breast tumors were shown to correlate with expression of an ESC signature [37]. Furthermore, WT p53 was shown to negatively regulates the expression of ESC transcription factors [21]. These findings suggest that the ESC signature expression in tumors lacking WT p53 activity may result from the absence of a p53 mediated repression of ESC factors expression.

As mentioned above, p53-hotspot missense mutations are the most common alteration in human tumors that produce mutant p53 proteins that exhibit oncogenic traits that are beyond the abrogation of the normal tumor-suppressor activity [4, 5]. We and others have shown a mutant p53 GOF activity in promoting EMT [38, 39]. Notably, the EMT was shown to exert stem cell characteristics [25]. Furthermore, mutant p53 was shown to induce de-differentiation of a human osteosarcoma cell line [40]. Additionally, mutant p53 was shown to facilitate the reprogramming process and to generate iPSCs with tumorigenic ability [9]. Recently, we reported that mutant p53 upregulate the expression

of CSC markers in colon cell lines [41]. These observations might indicate a mutant p53 GOF activity in enhancing cancer plasticity and stemness phenotype, thus promoting the malignant process. However, the molecular profiling of endowing cells with stemness phenotype by mutant p53 protein remains to be elucidated.

The observation that p53 knock-out in p53 Mut MSC-TL led to a reduction of tumorigenesis and CSC markers expression indicates a role of mutant p53 in expanding the CSC population. As expected, we found downregulation of cancer associated genes following mutant p53 KO. Interestingly, the ESC signature, which was identified in the p53 Mut MSC-TLs, was also downregulated following mutant p53 abrogation. This ample change in ESC gene signature expression, regulated by the presence of mutant p53 protein, rather than loss of WT p53 function, is novel. The significant differences in transcriptome profile between p53 Mut MSC-TL and p53 Mut KO MSC-TL sub-clones implies an involvement of a broader transcriptional regulation, such as an epigenetic regulation. Indeed, we showed that promoters of the mutant p53 MSC-TL dependent genes are enriched for chromatin transcriptional activation marks in ESCs. Furthermore, we found that Polycomb Repressive Complex 2 (PRC2) target genes are downregulated in p53 Mut MSC-TL sub-clones. This may indicate that, as in ESC cells, PRC2 is active in p53 Mut MSC-TL sub-clones and accordingly can negatively regulate the expression of different developmental associated genes by histone methyltransferase activity. Altogether, these analyses may imply that one possible mechanism in which mutant p53 regulates the expression of the ESC derive genes is by regulating the epigenetic mark state. Furthermore, the BindDB tool allowed us also to identify transcription factors that bind to the promoters of the mutant p53 MSC-TL dependent genes, thus suggesting that they may be involved in mechanisms in which mutant p53 regulates the ESC gene signature (supplementary figure 5B). In all, we suggest a novel mutant p53 GOF activity in upregulating the expression of a mesenchymal CSC-like signature that comprised of ESC genes. Core developmental transcription programs were shown to be shared between human and mouse species [42]. Therefore, it was of interest to examine the human relevance of the p53 Mut MSC-TLs ESC gene signature. We identified specific gene signatures, derived from the human orthologues of p53 Mut MSC-TLs ESC gene signature, that predicted

poor patient survival in different cancer types. When performing Pan-cancer expression analyses, we found that the human orthologues of the identified MSC-TLs ESC signature were preferentially expressed in human cancers harboring p53 hot-spot missense mutations. Furthermore, analyzing a previously defined human ESC signature, that was shown to correlate with high grade breast tumors with poor clinical outcome [30], showed similar preferential expression. These observations might explain the association between p53 mutations, and high grade tumors with poor prognosis [6-8, 43, 44].

In conclusion, our study demonstrates that mutant p53 GOF activity is associated with the enrichment of mesenchymal CSC-like population along with upregulation of a unique signature containing ESC genes. This ESC gene signature was also relevant to patient outcome and to human tumors harboring p53-hotspot missense mutations. Notably, carcinomas occasionally undergo epithelial to mesenchymal transition, a process that was also associated with acquisition of stemness traits [25]. Thus, our data might suggest that the specific genes in the ESC signature, identified in our MSC model, can be regarded as a broader signature of CSCs and tumor progression towards advanced disease. Gene members of this newly defined signature may therefore serve as prognostic biomarkers and new therapeutic targets for targeting cancer stemness and CSCs in tumors, thereby preventing cancer recurrence.

Acknowledgments

We thank Dr. Tali Shalit and Dr. Gil Hornung for RNA sequencing analysis. We also thank Dr. Raya Eilam and Dr. Ori Brenner for their help with immunohistochemistry and pathological analyses, respectively. In addition, we thank Dr. Ziv Porat for his help with Multispectral imaging flow-cytometry (IFC) data analysis. We would like to acknowledge TCGA research network for providing TCGA datasets. The work by V.R. was supported by a Center of Excellence Grant from the Israel Science Foundation and a Center of Excellence Grant from the Flight Attendant Medical Research Institute. V.R. is the incumbent of the Norman and Helen Asher Professorial Chair of Cancer Research at The Weizmann Institute.

References

1. Marjanovic ND, Weinberg RA, Chaffer CL: Cell plasticity and heterogeneity in cancer. Clin Chem 2013, 59(1):168-179. 2. Visvader JE, Lindeman GJ: Cancer stem cells in solid tumours: accumulating evidence and unresolved questions. Nat Rev Cancer 2008, 8(10):755-768. 3. Medema JP: Cancer stem cells: the challenges ahead. Nat Cell Biol 2013, 15(4):338-344. 4. Brosh R, Rotter V: When mutants gain new powers: news from the mutant p53 field. Nat Rev Cancer 2009, 9(10):701-713. 5. Muller PA, Vousden KH: Mutant p53 in cancer: new functions and therapeutic opportunities. Cancer Cell 2014, 25(3):304-317. 6. Junttila MR, Karnezis AN, Garcia D, Madriles F, Kortlever RM, Rostker F, Brown Swigart L, Pham DM, Seo Y, Evan GI et al: Selective activation of p53- mediated tumour suppression in high-grade tumours. Nature 2010, 468(7323):567-571. 7. Ahmed AA, Etemadmoghadam D, Temple J, Lynch AG, Riad M, Sharma R, Stewart C, Fereday S, Caldas C, Defazio A et al: Driver mutations in TP53 are ubiquitous in high grade serous carcinoma of the ovary. J Pathol 2010, 221(1):49-56. 8. Fujimoto K, Yamada Y, Okajima E, Kakizoe T, Sasaki H, Sugimura T, Terada M: Frequent association of p53 gene mutation in invasive bladder cancer. Cancer Res 1992, 52(6):1393-1398. 9. Sarig R, Rivlin N, Brosh R, Bornstein C, Kamer I, Ezra O, Molchadsky A, Goldfinger N, Brenner O, Rotter V: Mutant p53 facilitates somatic cell reprogramming and augments the malignant potential of reprogrammed cells. J Exp Med 2010, 207(10):2127-2140. 10. Wang Y, Yang J, Zheng H, Tomasek GJ, Zhang P, McKeever PE, Lee EY, Zhu Y: Expression of mutant p53 proteins implicates a lineage relationship between neural stem cells and malignant astrocytic glioma in a murine model. Cancer Cell 2009, 15(6):514-526. 11. Shetzer Y, Kagan S, Koifman G, Sarig R, Kogan-Sakin I, Charni M, Kaufman T, Zapatka M, Molchadsky A, Rivlin N et al: The onset of p53 loss of heterozygosity is differentially induced in various stem cell types and may involve the loss of either allele. Cell Death Differ 2014, 21(9):1419-1431. 12. Lang GA, Iwakuma T, Suh YA, Liu G, Rao VA, Parant JM, Valentin-Vega YA, Terzian T, Caldwell LC, Strong LC et al: Gain of function of a p53 hot spot mutation in a mouse model of Li-Fraumeni syndrome. Cell 2004, 119(6):861- 872. 13. George TC, Fanning SL, Fitzgerald-Bocarsly P, Medeiros RB, Highfill S, Shimizu Y, Hall BE, Frost K, Basiji D, Ortyn WE et al: Quantitative measurement of nuclear translocation events using similarity analysis of multispectral cellular images obtained in flow. J Immunol Methods 2006, 311(1-2):117-129. 14. Edgar R, Domrachev M, Lash AE: Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic acids research 2002, 30(1):207-210.

15. Rivlin N, Katz S, Doody M, Sheffer M, Horesh S, Molchadsky A, Koifman G, Shetzer Y, Goldfinger N, Rotter V et al: Rescue of embryonic stem cells from cellular transformation by proteomic stabilization of mutant p53 and conversion into WT conformation. Proc Natl Acad Sci U S A 2014, 111(19):7006-7011. 16. Kim J, Chu J, Shen X, Wang J, Orkin SH: An extended transcriptional network for pluripotency of embryonic stem cells. Cell 2008, 132(6):1049-1061. 17. Koga M, Matsuda M, Kawamura T, Sogo T, Shigeno A, Nishida E, Ebisuya M: Foxd1 is a mediator and indicator of the cell reprogramming process. Nat Commun 2014, 5:3197. 18. Pardal R, Molofsky AV, He S, Morrison SJ: Stem cell self-renewal and cancer cell proliferation are regulated by common networks that balance the activation of proto-oncogenes and tumor suppressors. Cold Spring Harb Symp Quant Biol 2005, 70:177-185. 19. Smith ZD, Sindhu C, Meissner A: Molecular features of cellular reprogramming and development. Nat Rev Mol Cell Biol 2016, 17(3):139-154. 20. Courtot AM, Magniez A, Oudrhiri N, Feraud O, Bacci J, Gobbo E, Proust S, Turhan AG, Bennaceur-Griscelli A: Morphological analysis of human induced pluripotent stem cells during induced differentiation and reverse programming. Biores Open Access 2014, 3(5):206-216. 21. Rivlin N, Koifman G, Rotter V: p53 orchestrates between normal differentiation and cancer. Semin Cancer Biol 2015, 32:10-17. 22. Zoller M: CD44: can a cancer-initiating cell profit from an abundantly expressed molecule? Nat Rev Cancer 2011, 11(4):254-267. 23. Ma I, Allan AL: The role of human aldehyde dehydrogenase in normal and cancer stem cells. Stem Cell Rev 2011, 7(2):292-306. 24. Chen D, Wu M, Li Y, Chang I, Yuan Q, Ekimyan-Salvo M, Deng P, Yu B, Yu Y, Dong J et al: Targeting BMI1(+) Cancer Stem Cells Overcomes Chemoresistance and Inhibits Metastases in Squamous Cell Carcinoma. Cell stem cell 2017, 20(5):621-634.e626. 25. Mani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, Zhou AY, Brooks M, Reinhard F, Zhang CC, Shipitsin M et al: The epithelial-mesenchymal transition generates cells with properties of stem cells. Cell 2008, 133(4):704- 715. 26. Kramer A, Green J, Pollard J, Jr., Tugendreich S: Causal analysis approaches in Ingenuity Pathway Analysis. Bioinformatics 2014, 30(4):523-530. 27. Kim NH, Kim HS, Kim NG, Lee I, Choi HS, Li XY, Kang SE, Cha SY, Ryu JK, Na JM et al: p53 and miRNA-34 are Suppressors of Canonical Wnt Signaling. Sci Signal, 4(197):ra71. 28. Hashimi ST, Fulcher JA, Chang MH, Gov L, Wang S, Lee B: MicroRNA profiling identifies miR-34a and miR-21 and their target genes JAG1 and WNT1 in the coordinate regulation of dendritic cell differentiation. Blood 2009, 114(2):404-414. 29. Bossi G, Marampon F, Maor-Aloni R, Zani B, Rotter V, Oren M, Strano S, Blandino G, Sacchi A: Conditional RNA interference in vivo to study mutant

p53 oncogenic gain of function on tumor malignancy. Cell Cycle 2008, 7(12):1870-1879. 30. Ben-Porath I, Thomson MW, Carey VJ, Ge R, Bell GW, Regev A, Weinberg RA: An embryonic stem cell-like gene expression signature in poorly differentiated aggressive human tumors. Nat Genet 2008, 40(5):499-507. 31. Ku M, Koche RP, Rheinbay E, Mendenhall EM, Endoh M, Mikkelsen TS, Presser A, Nusbaum C, Xie X, Chi AS et al: Genomewide analysis of PRC1 and PRC2 occupancy identifies two classes of bivalent domains. PLoS Genet 2008, 4(10):e1000242. 32. Livyatan I, Aaronson Y, Gokhman D, Ashkenazi R, Meshorer E: BindDB: An Integrated Database and Webtool Platform for "Reverse-ChIP" Epigenomic Analysis. Cell stem cell 2015, 17(6):647-648. 33. Joerger AC, Fersht AR: The p53 Pathway: Origins, Inactivation in Cancer, and Emerging Therapeutic Approaches. Annu Rev Biochem 2016, 85:375-404. 34. Solozobova V, Blattner C: p53 in stem cells. World journal of biological chemistry 2011, 2(9):202-214. 35. Rodriguez R, Rubio R, Menendez P: Modeling sarcomagenesis using multipotent mesenchymal stem cells. Cell Res 2012, 22(1):62-77. 36. Pevsner-Fischer M, Levin S, Zipori D: The origins of mesenchymal stromal cell heterogeneity. Stem Cell Rev 2011, 7(3):560-568. 37. Mizuno H, Spike BT, Wahl GM, Levine AJ: Inactivation of p53 in breast cancers correlates with stem cell transcriptional signatures. Proc Natl Acad Sci U S A 2010, 107(52):22745-22750. 38. Kogan-Sakin I, Tabach Y, Buganim Y, Molchadsky A, Solomon H, Madar S, Kamer I, Stambolsky P, Shelly A, Goldfinger N et al: Mutant p53(R175H) upregulates Twist1 expression and promotes epithelial-mesenchymal transition in immortalized prostate cells. Cell Death Differ 2011, 18(2):271- 281. 39. Dong P, Karaayvaz M, Jia N, Kaneuchi M, Hamada J, Watari H, Sudo S, Ju J, Sakuragi N: Mutant p53 gain-of-function induces epithelial-mesenchymal transition through modulation of the miR-130b-ZEB1 axis. Oncogene 2013, 32(27):3286-3295. 40. Di Fiore R, Marcatti M, Drago-Ferrante R, D'Anneo A, Giuliano M, Carlisi D, De Blasio A, Querques F, Pastore L, Tesoriere G et al: Mutant p53 gain of function can be at the root of dedifferentiation of human osteosarcoma MG63 cells into 3AB-OS cancer stem cells. Bone 2014, 60:198-212. 41. Solomon H, Dinowitz N, Pateras IS, Cooks T, Shetzer Y, Molchadsky A, Charni M, Rabani S, Koifman G, Tarcic O et al: Mutant p53 gain of function underlies high expression levels of colorectal cancer stem cells markers. Oncogene 2018. 42. Xue Z, Huang K, Cai C, Cai L, Jiang CY, Feng Y, Liu Z, Zeng Q, Cheng L, Sun YE et al: Genetic programs in human and mouse early embryos revealed by single-cell RNA sequencing. Nature 2013, 500(7464):593-597. 43. Andersson J, Larsson L, Klaar S, Holmberg L, Nilsson J, Inganas M, Carlsson G, Ohd J, Rudenstam CM, Gustavsson B et al: Worse survival for TP53 (p53)-

mutated breast cancer patients receiving adjuvant CMF. Ann Oncol 2005, 16(5):743-748. 44. Murakami I, Hiyama K, Ishioka S, Yamakido M, Kasagi F, Yokosaki Y: p53 gene mutations are associated with shortened survival in patients with advanced non-small cell lung cancer: an analysis of medically managed patients. Clin Cancer Res 2000, 6(2):526-530.

Table 1 Mutant p53 MSC derived tumor lines exhibited augmented tumorigenic capacity compared with their parental cells.

Cell type Number of Tumor/ Tumor take Days to injected cells mice (%) tumor detection (Mean) p53 Mut MSCs (1) 102 0/4 0 - p53 Mut MSCs (1) 103 0/4 0 - p53 Mut MSCs (1) 105 0/4 0 - p53 Mut MSC-TL (1) 102 3/4 75 33 p53 Mut MSC-TL (1) 103 3/4 75 28.3 p53 Mut MSC-TL (1) 105 4/4 100 21.7 p53 Mut MSCs (2) 102 0/4 0 - p53 Mut MSCs (2) 103 0/4 0 - p53 Mut MSCs (2) 105 0/4 0 - p53 Mut MSCs (2) 3x106 4/5 80 73.5 p53 Mut MSC-TL (2) 102 4/9 44.4 42.2 p53 Mut MSC-TL (2) 103 9/9 100 29.4 p53 Mut MSC-TL (2) 105 9/9 100 18.7 p53 Mut MSCs (3) 102 0/4 0 - p53 Mut MSCs (3) 103 0/4 0 - p53 Mut MSCs (3) 105 1/4 25 144 p53 Mut MSCs (3) 3x106 9/9 100 120.6 p53 Mut MSC-TL (3) 102 9/11 81.8 29 p53 Mut MSC-TL (3) 103 9/14 64.2 23.1 p53 Mut MSC-TL (3) 105 5/5 100 15.2

Figures legends

Figure 1. Mutant p53 enhanced the oncogenicity of primary bone marrow MSCs. Doxorubicin treatment (0.5µg/ml) for 6 hours. (A) western blot analysis of p53 and p21. Representative experiment of three experiments. (B) p53 expression by ImageStreamX. Images of representative cells (upper panel). Representative plot of p53 intensity (Lower panel) (C) p21 expression by QRT-PCR. Results are presented as mean±SD of two experiments. (D) Proliferation plot. *p<0.05 by repeated-measures ANOVA. Representative experiment of two experiments. (E) CFU-Fs from the BM of p53 WT and p53 Mut mice (upper panel). The results are displayed as means±SEM (lower panel). *p<0.05 by two-tailed Student’s t-test. Representative experiment of two experiments. Figure 2. Mutant p53 MSC derived tumor lines exhibited augmented tumorigenic capacity compared with their parental cells. (A) Proliferation plot. Representative experiment of two experiments. **p < 0.01 by repeated-measures ANOVA. (B) 1×104 -2×104 cells were imaged by ImageStreamX. (Left panel) The percentage of the sub-population with low area and high circularity was determined. Results are presented as mean±SEM. *p<0.05 by paired two-tailed Student’s t-test. (Right panel) Representative plot of circularity versus area. The red gate contains the small and circular sub-population. (C) CD44 expression was determined by ImageStreamX. The results are from two experiments and presented as mean±SEM (Upper panel). *p <0.05 by paired one-tailed Student’s t-test. Representative image of CD44HIGH and CD44LOW expression levels. BF=bright field (Lower panel). (D) Three p53 Mut MSC isolate and their derived MSC-TL were subcutaneously injected into Athymic Nude-Foxn1nu mice. Tumor take percentage are presented as mean ± SEM. ***p<0.001 by proportion test. (E) Distribution of tumors latency is presented as mean±SEM. Survival analysis showed higher hazard-ratio in the p53 Mut MSC-TLs group compared to p53 Mut pMSCs group (115-fold increase in tumor development, ***p<0.001). (F) Tumorigenic cell frequency (http://bioinf.wehi.edu.au/software/elda/). CI, confidence interval. (G) Bioluminescence monitoring of growing tumors in mice subcutaneously injected with p53 Mut pMSC isolate and a derived MSC-TL. Results are shown as means ± SEM. ****p<0.0001 by repeated-measures ANOVA (Representative images in Supplementary Figure 2G) (H) H&E staining of MSC-TL derived tumor indicating (a)

Invasion of a peripheral nerve. Arrowheads identify the outline of the infiltrated nerve. (b) Blood vessel invasion (vein). Arrowheads identify the severely damaged vein wall. The lumen is filled with neoplastic cells. (A-artery). (c) Invasive growth. The neoplastic cells form finger-like extensions (arrows) which infiltrate the surrounding skeletal muscle. (d) High mitotic rate (arrows mark cells in mitoses) (I) PCNA and ALDH1 expression in p53 Mut MSCs and MSC-TLs derived tumors by immunohistochemistry staining. (Quantification and statistic in Supplementary Figures 2H-I). Figure 3. Embryonic gene signature expressed in mutant p53 MSC-TLs. RNA sequencing of p53 Mut pMSCs and p53 Mut MSC-TLs (A) Heat map of CSC gene signature derived genes. (B) GSEA enrichment plots (C) Diseases or Functions that are increase or decrease in the p53 Mut MSC-TLs versus p53 Mut MSCs according to IPA. (D) Activated or inactivated upstream regulators in p53 Mut MSC-TLs versus p53 Mut MSCs according to IPA. (E) Heat maps representing the three ESC gene datasets derived genes. (F) Sox2 expression by QRT-PCR. **p<0.01 by two-tailed Student’s t-test (G) immunohistochemistry staining of Sox2 in MSCs and MSC-TLs derived tumors. Four tumors (two tumors per slide) in each group were stained and the average of the percentage of Sox2 positive cells of, at least, 13 random fields in each slide was estimated. The results are presented as mean±SEM. ***p<0.001 by two-tailed Student’s t-test. Representative images are presented.

Figure 4. Mutant p53 knock-out reduced the tumor initiating capacity of mutant p53 MSC-TL sub-clones. p53 Mut and p53 KO MSC-TLs sub-clones were subcutaneously injected into Athymic Nude-Foxn1nu mice (See the blot of mutant p53 KO in Supplementary Figure 4) (A) 14 days after injection of 1×105 cells, tumors were removed. Tumors weight are presented as mean±SEM. ***p<0.001 by two-tailed Student’s t-test (B) 1×104 cells from each group (n=3) were subcutaneously injected for tumor growth observation. The results are shown as mean±SEM. *p<0.05 by log-transformed rate t-test. (C) Tumors latency is presented as mean±SEM. Survival analysis showed higher hazard-ratio in the p53 Mut MSC-TLs group compared to p53 KO group (5.2-fold increase in tumor development, **p<0.01). (D) Percentage of tumor take are presented as mean±SEM. Number of mice in each cell amount engraftment; (n=10,1×101), (n=17,1×102), (n=7,1×103) in each group. **p<0.01

by proportion test. (E) Tumorigenic cell frequency (http://bioinf.wehi.edu.au/software/elda/). CI, confidence interval. (F) Representative H&E staining of p53 KO (a+b) and p53 Mut (c+d) MSC-TLs sub-clones derived tumors. a. Low magnification of a p53 KO MSC-TL sub-clone derived tumor with extensive necrosis, seen as pink-stained areas (asterisks). The boxed area is shown in b. c. Low magnification of a p53 Mut MSC-TLs sub-clone derived tumor composed of viable tissue. The boxed area is shown in d. (G) Immunohistochemistry analysis of ALDH1 and CD44. ALDH1 analysis contained sections from three tumors in each group and the CD44 analysis contained sections from four tumors in each group. The staining area in, at least, 10 random fields in each slide was estimated. The results are presented as mean±SEM. ***p<0.001 by two-tailed Student’s unpaired t-test. Representative images are presented.

Figure 5. Mutant p53 knock-out led to a reduction in the expression of the ESC gene signature expression in MSC-TL sub-clones.

Transcriptional analysis of p53 Mut and p53 KO MSC-TL sub-clones by RNA sequencing. (A) Diseases or functions that are increase or decrease following KO of p53 Mut according to IPA. (B) Myc and Klf4 normalized counts. (C) Heat map of MSC-TLs ESC gene signature derived genes following KO of p53 Mut. (D) The volcano plot presents differential genes (marked as black) and ESCs derived genes that were downregulated (marked as blue) or upregulated (marked as red) upon KO of mutant p53 in MSC-TLs sub-clones. (E) Heat map of PRC2 target genes following KO of p53 Mut. (F) The volcano plot presents differential genes (marked as black) and PRC2 target genes that were downregulated (marked as blue) and upregulated (marked as red) upon KO of mutant p53 in MSC-TLs sub-clones. (G) QRT-PCR analysis of ESC genes expression. (H) Western blot analysis of p53 and ESC derived genes.

Figure 6. The MSC-TLs ESC signature derived genes are associated with poor patient survival and correlate with human tumors harboring p53 missense mutations. (A) Kaplan-Meier graph of the indicated datasets showing significant relationship between genes expression levels and patient survival. (B) Volcano plot of the identified

MSC-TLs ESC gene signature and mutant p53 dependent ESC genes in the p53-hotspot mutations TCGA samples versus other p53 statuses TCGA samples (Chi-squared test p- value=0.01). (C) Volcano plot of a previously described human ESC signature in the p53- hotspot mutations tumor samples versus other p53 statuses samples in TCGA datasets (Chi-squared test p-value=0.004). (D) Box plots of RNA expression in TCGA samples harboring one of the p53-hotspot mutations versus other p53 statuses. Statistical analysis was determined by t-test on log2 transformed normalized counts (p-value<0.05, - 1.5>FC>1.5) FDR corrected p-value is indicated on the top of the box plot.

A mutant p53-dependent embryonic stem cell gene signature is associated with augmented tumorigenesis of stem cells

Gabriela Koifman, Yoav shetzer, Shay Eizenberger, et al.

Cancer Res Published OnlineFirst August 28, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-18-0805

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2018/08/28/0008-5472.CAN-18-0805.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet been Manuscript edited.

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/2018/08/28/0008-5472.CAN-18-0805. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.