EP300 Knockdown Abrogates Cancer Stem Cell Phenotype, Tumor Growth and Metastasis in Triple Negative Breast Cancer

Background :Triple negative breast cancer (TNBC) is an aggressive breast cancer subtype with basal features, lacking the expression of receptors targeted successfully in other breast cancer subtypes. Treatment response to adjuvant and neoadjuvant chemotherapy is often short-lived and metastatic spread occurs at higher rates than other subtypes within the first five years after diagnosis. TNBCs exhibit stem cell features and are enriched for cancer stem cell (CSC) populations. E1A Binding Protein P300(EP300) is a large protein with multiple cellular functions, including as an effector in stem cell biology. Methods : We used a genetic knockdown (KD) model of EP300 in TNBC cell lines to investigate the effect on CSC phenotype, tumor growth and metastasis. Side population assay and tumorsphere suspension culture were used in vitro. Xenograft mouse models were used for in vivo studies. We performed in silico analysis of publicly available gene expression data sets to investigate CSC gene expression and molecular pathways as well as survival outcomes associated with EP300 expression in patients with TNBC and basal-like BC. Results : EP300 KD abolishedthe CSC phenotype by reducing ABCG2 expression, side population cells andtumorsphere formation capacity in vitro as well as tumor formation in a xenograft mouse model in vivo . Metastatic capacity was markedly reduced in EP300 KD cells in vivo, with no detection of circulating tumor cells.TCGA data analysis demonstrated that genes positively correlated with EP300 expression in TNBC and basal-like BC were associated with CSC biology. Survival analysis demonstrated that EP300 expression predicts poor recurrence free survival in TNBC and basal BC. Conclusion: We report a novel oncogenic role for EP300 in driving CSC phenotyperepresentinga potential target to address tumor initiation and metastatic spread in TNBC and basal-like BC. EP300 might serve as a prognostic marker and potential therapeutic target in TNBC. (both from Santa Cruz Biotechnologies). Secondary antibody incubation was performed for 60 min at room temperature. Secondary antibodies: goat anti-mouse IgG HRP conjugated; and goat anti-rabbit IgG HRP conjugated (both from Santa Cruz Biotechnologies). Subsequently the membranes were incubated with HRP substrate (GE Life Pittsburgh, PA) for 5 min. and chemiluminescence was recorded using a ChemiDoc MP gel imaging system (Bio-Rad). ImageJ software was used for quantification. Samples were prepared in technical duplicates. We demonstrated that knockdown of EP300 reduced the expression of ABCG2, eliminated side population cells and decreased tumorsphere formation potential in vitro . In vivo studies using stable EP300 knockdown cells showed a striking reduction in tumor initiation and tumor outgrowth. Knockdown cells also lacked the capacity for metastatic colonization potential and the formation of circulating tumor cell after tail vein injection. EP300 gene expression in large published breast cancer data sets positively corresponded with genes and molecular pathways associated with CSC function and predicted poor RFS in TNBC and basal-like BC. phenotype in TNBC and basal-like breast cancer. We show that knockdown of EP300 abrogates the CSC phenotype of TNBC cells and abolishes tumor growth and metastasis. Our results warrant further exploration of EP300 as a prognostic factor and potential therapeutic target in TNBC and basal-like BC.

Quantitative reverse transcription polymerase chain reaction (qRT-PCR) Total RNA extraction was performed using TRIzol reagent (Thermo Fisher Scientific) according to the manufacturer's protocol. First strand synthesis was performed using the qScript cDNA Supermix (Quantabio, Beverly, MA) and 1 µg of total RNA per reaction. The following reaction conditions were used on a T100 thermal cycler (Bio-Rad, Irvine, CA): 25 °C for 5 min, 42 °C for 30 min, 85 °C for 5 min. Samples were prepared as duplicates unless stated otherwise.
For gene expression quantification, SYBR Green RT-PCR master mix was used (Quantabio). Reaction were performed in 96-well plates (Bio-Rad), using 2 µl cDNA, 12.5 µl SYBR Green master mix, 1µL each forward and reverse primer probes and 8.5 µl nuclease free deionized water. The following reaction conditions were used on a MyiQ or CFX96 Real were then blocked in 5% skim milk for 60 min. Primary antibody incubation was performed in blocking solution over night at 4 °C. Primary antibodies: polyclonal rabbit anti-human p300 (N-15), monoclonal mouse anti-human ACTB (H-102) (both from Santa Cruz Biotechnologies). Secondary antibody incubation was performed for 60 min at room temperature. Secondary antibodies: goat anti-mouse IgG HRP conjugated; and goat anti-rabbit IgG HRP conjugated (both from Santa Cruz Biotechnologies).
Subsequently the membranes were incubated with HRP substrate (GE Healthcare Life Sciences, Pittsburgh, PA) for 5 min. and chemiluminescence was recorded using a ChemiDoc MP gel imaging system (Bio-Rad). ImageJ software was used for quantification. Samples were prepared in technical duplicates.

Animal experiments
Eight to ten weeks old female NOD scid gamma (NSG) mice (NOD-scid IL2Rgamma null , The Jackson For tail vein injections, the mice were placed under a heating lamp for 2 min to dilate blood vessels. After subsequent immobilization in a rodent holder (Kent Scientific, Torrington, CT) 1 × 10 6 cells in 100 µl culture medium were injected per mouse using hypodermic syringes.
In vivo imaging: Tumor growth and spread after tail vein injection was monitored intra-vitally once per week for a total of 5 weeks. Luciferin (Promega, Madison, WI) was prepared at a final concentration of 300 mg/ml. After induction of anesthesia as described above, 300 µl of luciferin was injected intraperitoneally per mouse and allowed to distribute through the body of the animal for 15 min before imaging. Imaging was performed using an IVIS Spectrum pre-clinical in vivo imaging system (PerkinElmer, Waltham, MA) under continued anesthesia. Images were taken after 10 s, 1 min and 3 min each for supine and prone position. Bioluminescence was quantified using the Living Image 4.2 software (PerkinElmer).

Gene association and survival analysis
For gene association studies publicly available data sets from the The Cancer Genome Atlas (TCGA) breast cancer cohort were used and analyzed via Ingenuity pathway (IPA) tool and cBioPortal [21][22][23].
The Kaplan-Meier (KM) plotter web interface was used to assesses the effect of EP300 gene expression on survival in breast cancer patients [24]. KM plotter is a meta-analysis-based biomarker assessment tool using a manually curated gene expression database downloaded from Gene Expression Omnibus (GEO), European Genome-phenome Archive (EGA) and The Cancer Genome Atlas (TCGA). The auto selected cut-off was used to define high vs. low expression, which computes the best threshold between the upper and lower quartiles based on sample selection. Recurrence free survival (RFS) was reported at 60-month survival.

Statistical analysis
GraphPad PRISM (GraphPad Software Inc., La Jolla, CA) was used for statistical analysis of all experiments. Non-normal distribution was assumed, and appropriate non-parametric statistical tests were used; for experiments with single variables and groups of two Mann-Whitney test was used; for three or more groups, non-parametric one-way ANOVA (Kruskal-Wallis and Dunn's multiple comparison) was used. For two variables and groups of two or more two-way ANOVA was used.
Statistical hypothesis testing (Sidak test) was used to correct for multiple comparison. Data for repeats were presented as means and standard deviation. A two-tailed statistical significance level Tumorsphere formation capacity is reduced after EP300 KD in TNBC cells MDA-MB-231 EP300KD clones showed significantly reduced tumorsphere formation capacity in vitro (scramble 78 ± 5, KD clone 1 10 ± 2 n.s., KD clone 2 2 ± 3 p = 0.02, n = 3 each) ( Fig. 2A). Transient siRNA-mediated KD of EP300 in two additional TNBC cell lines (BT20 and Hs578T) also showed reduction of tumorsphere formation, albeit to a lesser degree than stable KD (BT20 KD 43 ± 6 vs. scramble 77 ± 14, p = 0.018; Hs578T KD 47 ± 6 vs. scramble 72 ± 5, p = 0.005; n = 3 each) (Fig. 2B). These results demonstrated that anchorage independent sphere formation as a characteristic of cell with CSC phenotype is reduced by EP300 knockdown to a varying degree in TNBC cell lines. EP300 correlated genes show CSC related transcription in TNBC and basal-like BC The TCGA breast cancer cohort was filtered for TNBC (n = 82) or basal-like breast cancers (n = 81), which based on gene expression showed an overlap of n = 33 cases between both sample groups (Fig. 5A). Gene expression demonstrated that n = 3037 and n = 509 genes in TNBC and basal-like BC, respectively, correlated positively with EP300 expression (FDR < 0.05), with n = 298 genes correlating in both subgroups (Fig. 5B and supplementary table S1). The top associated pathways and biological functions in Ingenuity Pathway Analysis (IPA) (Qiagen, Hilden, Germany) are shown in Fig. 5B. While genes correlating with EP300 in basal breast cancer were associated with stem cell biology and cancer metastasis, EP300 correlated genes in TNBC were associated with growth factor signaling pathways (Fig. 5C). All three sets of genes positively correlated with EP300 (TNBC, basallike BC and the overlap between both) were linked to PTEN and Protein kinase A (PKA) signaling (Fig. 5C).

Knockdown of EP300 abolishes tumor formation and causes G2/M blockade in vivo
Using the overlap gene signature (n = 298 genes) to analyze associated biological function via cBioPortal showed that the top associated pathways were TP53/BRCA, Notch and Wnt signaling ( Fig. 5D and supplementary fig. S4).
The expression of affected genes in these pathways (Notch -CREBBP, SPEN; Wnt -GSK3B) showed strong positive correlation with EP300 expression on a per sample basis (Fig. 5E). The graphs for Fig. 5D and E were created and directly exported form cBioPortal [23]. These results demonstrated that EP300 expression correlated with CSC related genes and pathways (Notch, Wnt, BRCA/TP53) suggesting at complex role of EP300 in CSC biology.

EP300 is prognostic in TNBC and basal-like BC
The association between EP300 expression and RFS was assessed in 6234 breast cancer patients, of which 198 TNBC cases were available for analysis. High EP300 gene expression was statistically significantly correlated with worse RFS in TNBC and basal-like BC patients with lymph node metastasis or high-grade tumors (G3) (Fig. 6).
Although not all results reached statistical significance, there was a clear trend towards increasing hazard ratios (HR) associated with higher histological grade (G3) and lymph node (LN) metastasis, with the highest HR in TNBC G3 and positive LN metastasis (Fig. 6). Comparing all PAM50 subtypes to basal-like BC had no significant effect on recurrence free survival. Of note, while higher EP300 RNA expression was associated with a better prognosis in BC in general (Supplementary fig. S5A), high EP300 protein expression demonstrated worse RFS (HR 3.32, 1. 53-7.19) in all BC patients (Supplementary fig. S5B). The results demonstrated that high EP300 RNA expression might function as prognostic marker for poor RFS in TNBC/basal-like breast cancer.

Discussion
Our findings shed new light on the role of EP300 in the cancer stem cell phenotype, tumor initiation and metastasis in TNBC. We demonstrated that knockdown of EP300 reduced the expression of ABCG2, eliminated side population cells and decreased tumorsphere formation potential in vitro. In vivo studies using stable EP300 knockdown cells showed a striking reduction in tumor initiation and tumor outgrowth. Knockdown cells also lacked the capacity for metastatic colonization potential and the formation of circulating tumor cell after tail vein injection. EP300 gene expression in large published breast cancer data sets positively corresponded with genes and molecular pathways associated with CSC function and predicted poor RFS in TNBC and basal-like BC.
Collectively these data demonstrate a novel oncogenic role for EP300 in TNBC and basal-like breast cancer biology.
ATP Binding Cassette Subfamily G Member 2 (ABCG2), also known as the breast cancer resistance protein (BCRP) is involved in the trafficking of molecules across cellular membranes and has been demonstrated to play a pivotal role in drug resistance to chemotherapeutics used in the treatment of TNBC (e.g. anthracyclines and topoisomerase inhibitors) [25]. ABCG2 is a pivotal marker of side population cells, which have been shown to be enriched for cells with CSC characteristics in breast cancer [26,27]. Our study showed a reduction in ABCG2 expression after EP300 knockdown and a strong abrogating effect on the side population phenotype in vitro as well as tumor initiation and lung colonization in vivo. Patrawala et al. showed that while the side population was enriched for tumorigenic CSC-like cells, ABCG2 levels did not affect their tumorigenic potential [26]. We showed that EP300 KD in Hs578T cells did not reduce the expression of ABCG2 but had a significant effect on tumorsphere formation. Other CSC markers have been proposed in breast cancer, such as the EPCAM + CD44 + CD24 − phenotype [11], SOX2 [28], and ALDH1 [29]. It would be of interest to investigate the effect of EP300 KD on these cell populations as distinct CSC phenotypes seem to represent unique rather than common cell populations [30].
Further investigation into the underlying mechanisms by which EP300 knockdown inhibits tumor growth in vivo is warranted. As stromal cells play a pivotal role in tumor development [31], perhaps microenvironmental factors such as paracrine signaling and physical cues from stromal cells affect tumor growth after EP300 knockdown [32]. Xenograft co-culture of MDA-MB-231 EP300KD with MDA-MB-231 WT did not stimulate outgrowth of KD cells in vivo, yet these same knockdown cells exhibited regular growth behavior in vitro before and after xenograft culture. Factors such as oxygen and glucose levels did not influence growth in vitro. While stroma cells can function as tumor suppressors [33], it has been shown that factors from CSCs can corrupt these cells to promote cancer growth [34]. Conceivably after EP300 KD and reduced CSC-like phenotype, the balance of factors in the cancer cell microenvironment may be tipped towards the suppression of tumor outgrowth. The varying degree of reduction in tumorsphere formation capacity after EP300 KD also points towards a role of physical factors in the growth behavior of these cells. The observed differences in tumorsphere formation capacity after EP300 KD between different TNBC cell lines could be due to technical differences in gene knock down (stable vs. transient) or have distinct biological reasons such as the claudin-low versus basal-like subtypes [35].
Since we used an immunocompromised (NSG) mouse model, immunological factors are unlikely to play a role in the absence of in vivo tumor formation through EP300 KD. Yet, as immunotherapies become increasingly important in the treatment of solid tumors, including BC [6], it will be of interest to investigate the effect of EP300 KD in TNBC cells in the context of a functional immune system [36]. For example, Xiong et al. showed that EP300 expression might be predictive of hyper progressive disease after PD-L1 blockade [37]. EP300 has been shown to increase migratory potential and invasiveness of cancer cells [38,39]. Reduction in lung colonization and distant metastasis as well as circulating tumor cells after tail vein injection suggested that EP300 KD TNBC cells potentially either cannot survive in circulation or are unable to seed tumors at distant sites.  [44], BRCA [45] and embryonic stem cells (ES) transcriptional networks. The genes/signaling pathways common to TNBC, basal-like BC and the overlap between both were PTEN and protein kinase A (PKA). Both pathways have previously been described as regulators and potential targets in CSC biology [46,47]. Other genes positively correlated with EP300 that overlap between TNBC and basal BC (GSK3B, CREBBP, SPEN) and associated with the top affected pathways (Wnt and Notch signaling) have been linked to CSC biology and metastasis in breast cancer [48][49][50].
Survival analysis in an in-silico cohort of breast cancer cases demonstrated worse RFS correlated with high EP300 expression in TNBC and basal-like BC patients with high grade (G3) or LN metastasized tumors. Although not statistically significant due to small patient numbers, there was a trend toward increased hazard ratios for patients with both high grade and LN metastasized tumors. Validation of these results in larger patient cohorts as well as in BC patients with distant metastasis is warranted.
Several studies have shown that EP300 has tumor suppressor function [16,17]. A specific mutation in EP300 (G211S) has been shown to be present in a subgroup of TNBC patients with very low overall mutational burden [51]. Interestingly, these patients had a lower risk of relapse and breast cancer-related mortality. Other groups demonstrated that EP300 may also function as an oncogene [18,19]. Collectively, these findings raise the possibility that context matters. Another famous tumor suppressor, TP53, has been shown to potentially have pro-tumorigenic effects via increased inflammation [52] or anti-apoptotic mechanisms [53]. Similarly, we demonstrate here that EP300 might context-dependently favor a CSC phenotype.
Naturally there are limitations to our study. The fact that only a single cell line was used for in vivo studies, as well as the use of immunocompromised mice limits the extrapolation of our results to a more generalizable situation found in individual BC patients. The effect of the immune system on tumor growth is one of the most exciting and promising area of cancer research and the effect of EP300 expression in this context should be explored. The role of the tumor microenvironment more generally seems to play a critical role in the findings described here and invites further detailed experimentation.

Conclusion
Taken together, the data presented here is to our knowledge the first study describing an oncogenic role of EP300 in CSC phenotype in TNBC and basal-like breast cancer. We show that knockdown of EP300 abrogates the CSC phenotype of TNBC cells and abolishes tumor growth and metastasis. Our results warrant further exploration of EP300 as a prognostic factor and potential therapeutic target in TNBC and basal-like BC.    or EP300 KD cells (same mice as shown in Figure 4D). (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001).

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.