Estrogen Receptor Beta Influences the Inflammatory p65 Cistrome in Colon Cancer Cells

Inflammation is a primary component of both initiation and promotion of colorectal cancer (CRC). Cytokines secreted by macrophages, including tumor necrosis factor alpha (TNFα), activates the pro-survival transcription factor complex NFκB. The precise mechanism of NFκB in CRC is not well studied, but we recently reported the genome-wide transcriptional impact of TNFα in two CRC cell lines. Further, estrogen signaling influences inflammation in a complex manner and suppresses CRC development. CRC protective effects of estrogen have been shown to be mediated by estrogen receptor beta (ERβ, ESR2), which also impacts inflammatory signaling of the colon. However, whether ERβ impacts the chromatin interaction (cistrome) of the main NFκB subunit p65 (RELA) is not known. We used p65 chromatin immunoprecipitation followed by sequencing (ChIP-Seq) in two different CRC cell lines, HT29 and SW480, with and without expression of ERβ. We here present the p65 colon cistrome of these two CRC cell lines. We identify that RELA and AP1 motifs are predominant in both cell lines, and additionally describe both common and cell line-specific p65 binding sites and correlate these to transcriptional changes related to inflammation, migration, apoptosis and circadian rhythm. Further, we determine that ERβ opposes a major fraction of p65 chromatin binding in HT29 cells, but enhances p65 binding in SW480 cells, thereby impacting the p65 cistrome differently in the two cell lines. However, the biological functions of the regulated genes appear to have similar roles in both cell lines. To our knowledge, this is the first time the p65 CRC cistrome is compared between different cell lines and the first time an influence by ERβ on the p65 cistrome is investigated. Our work provides a mechanistic foundation for a better understanding of how estrogen influences inflammatory signaling through NFκB in CRC cells.


INTRODUCTION
Colorectal cancer (CRC) accounts for the third highest number of cancer deaths among both women and men in the Western world (1). Benign adenomatous polyps evolve into carcinomas over 10-15 years, and screening is performed in many countries. While COX-2 inhibitors (e.g. celecoxib) or aspirin reduce inflammation and effectively prevent adenomatous polyp formation and CRC, adverse effects exclude their general usage (2)(3)(4). Better preventive treatments are required, and a deeper understanding of exactly how inflammation impacts CRC is needed.
During inflammation, cytokines such as tumor necrosis factor alpha (TNFa) are released by macrophages. TNFa activates the nuclear factor kappa-light-chain-enhancer of activated B cells (NFkB) transcription factor complex, which in turn induces several oncogenes and signaling pathways involved in tumor initiation and progression (5)(6)(7). Constitutive activation of NFkB has been observed in nearly 70% of CRC cell lines and 40% of clinical CRC specimens (8)(9)(10)(11). NFkB is a homo-or heterodimer comprised of two to five of subunits p65 (RelA/RELA), RelB (RELB), cRel (RELC), p105/p50 (NFkB1), or p100/p52 (NFkB2), that appears in multiple forms (12). The N-terminal Rel domain is present in all subunits and binds to a specific DNA sequence known as the kB site. Together with the recruitment of cofactors, this induces or suppresses expression of target genes. The various NFkB dimers differ in binding affinity and activation. p65 and cRel are the most potent transcriptional activators (13), and p65 together with p50 constitutes the most common NFkB heterodimer (14). While inflammation is critical in development of CRC, the function of NFkB complex in CRC is, however, poorly studied. Only one study describes the genome-wide chromatin binding of p65 in colon cancer, in cell line SW480 in the context of its interaction with p53 (TP53) mutants (15).
The hormone estrogen has been shown to reduce CRC incidence (16)(17)(18)(19)(20). Estrogen mainly acts through three receptors, of which estrogen receptor beta (ERb, ESR2) is present in epithelial colon and rectal cells (21,22). We have recently shown that intestinal epithelial ERb in vivo protects from the epithelial damage caused by TNFa and prevents tumor formation (22). Also, when reintroduced into CRC cell lines, ERb has antiproliferative and tumor-suppressive activity (22,23). ERb is a ligand-activated nuclear receptor which binds to genomic ERE (estrogen response elements). Its homologue, ERa (ESR1) is upregulated in breast cancer, where it promotes cell proliferation and interacts with p65 (24,25). ERb expression, in contrast, decreases during CRC development, and ERa is not expressed in the colon epithelial cells, nor tumors (26). Our hypothesis is that ERb in the normal colon opposes NFkBmediated inflammatory signaling and that this is an essential part of its tumor protective mechanism.
A crosstalk between the related ERa and NFkB has been extensively studied in breast cancer, albeit with some contradicting findings. A few studies report that ERa represses NFkB activity (27,28), whereas other reports that ERa, in the same cell lines (MCF7, T47D, ZR-75), enhances NFkB activity (24,25). Specifically, TNFa in ERa-positive MCF-7 breast cancer cells was shown to profoundly modify the ERa enhancer-binding landscape in an NFkB-dependent manner (29). Based on the homology between the two ERs DNAbinding domains, along with previous findings that ERb regulates NFkB key targets and reduces inflammatory signaling in colon, we speculated that ERb may impact the colon p65 cistrome. In the present study, we used p65 chromatin immunoprecipitation (ChIP) followed by sequencing (ChIP-Seq) to test this hypothesis, and to detail the p65 landscape in CRC cell lines.

ChIP-Sequencing
Libraries of the ChIP DNA were prepared and sequenced by the National Genomic Infrastructure (NGI) for Bioinformatics and Expression Analysis (BEA). DNA libraries were prepared using the NEB Next Ultra II DNA Library Prep Kit for Illumina (p/n NEB #E7645) and quality confirmed using TapeStation (DNA D1000 ScreenTape, Agilent). Libraries were loaded (1.8 pM end concentration of 1%) and sequenced (75 cycles, single read) using NextSeq 550 (Illumina).

ChIP-Seq Data Analysis
Spliced Transcripts Alignment to a Reference (STAR) was used to map unique ChIP-Seq reads to the human reference genome assembly hg38 (GRCh38) with the alignIntronMax flag set to 1. Peak calling was performed using Hypergeometric Optimization of Motif Enrichment (HOMER) over input with a four-fold enrichment as cutoff and applying a false discovery rate (FDR) less than 0.001. Peaks which overlapped within 200 bp and were present in at least two out of three biological replicates were used for downstream analysis. Raw tag counts were normalized using R and binding pattern differences were identified with edgeR package. To cluster and visualize the different peaks, Complex heatmap from R was used. Promoter regions were defined as -1kb to +100bp from TSS and genomic distribution of binding sites were identified by HOMER. Gene functional annotation was performed using Database for Annotation Visualization and Integrated Discovery (DAVID), with P <0.05 considered as significant.

Data Availability
The p65 ChIP-Seq data is deposited in the Gene Expression Omnibus (GEO) repository (GSE160856), TNFa bead array gene expression data was published previously (available at GSE65979), and SW480ERb input and HT29ERb input controls (GSE149979).

RESULTS
First, to understand the role of p65 transcriptional activity and oncogenic functions in CRC, we aimed to characterize its genome-wide binding in human CRC cells and correlate this to the TNFa-mediated transcriptional impact in the same cells. We used two well-characterized human colorectal adenocarcinoma cell lines: HT29 from a female primary tumor and SW480 from a male Dukes' type B primary tumor.

The p65 Cistrome of Colon Cancer Cells
After optimizing the protocol for antibody specificity and including a double cross-linking (DSG-formaldehyde) procedure to capture long-range interactions of p65, we analyzed the chromatin bound by p65 in triplicate experiments of each cell line, HT29 and SW480. The sequencing produced between 24M and 65M (80%) of high-quality mapped reads per sample (Table S1). We identified a total of 12,504 (HT29) and 5004 (SW480) significantly enriched p65 peaks compared to input ( Table S1). Out of these, 3151 and 1459 binding sites were found in HT29 and SW480 cells, respectively ( Figure 1A). Whereas more p65 chromatin-binding sites were detected in HT29 cells overall, comparison between the two cell lines revealed that 63% (919) of sites found in SW480 were also found in HT29 ( Figures 1A, B). We next used HOMER to determine DNA motifs of the identified peaks. As expected, we identified RELA as the top motif, followed by JUN-AP1, in both cell lines ( Figure 1C). This corroborates the specificity of the antibody and the protocol. Further, the transcription factors HNF4A and NFAT motifs were present in 3-5% of the HT29 binding sites, and FOXA1 and RUNX1 were relatively abundant (12% and 17%) in the SW480 p65 cistrome. Thus, we present the p65 cistrome of two different CRC cell lines and identify a shared common core, as well as cell-line specific differences.

p65-Bound Genes Are Involved in Migration and Circadian Clock
In order to decipher how the above identified p65 binding may impact gene expression, we analyzed where the binding sites were located in relation to known genes. We found a highly similar pattern in both cell lines, with about 38% of sites located within introns, 34% in intergenic chromatin regions, and 22% within the promoter area (-1kb to +100bp from the transcription start sites, TSS) of genes ( Figure 1D). The top-20 most enriched promoter sites in both cell lines include well-known p65 targets such as NFkB regulators NFKBIB, NFKBIZ, TNFAIP3 (35), BCL3 (36), and BIRC3 (37, 38), NFkB subunits NFkB2 (p52) and RELB, tumor suppressor p53, TNIP1, and CREB1 (Table S2). p65-binding sites unique to either cell line also included wellknown p65 target genes (HT29: BCAR3, BIRC7, DUSP16, PTGS2, and TNFAIP8; SW480: CDX2, CDH10, CLRN3, ESR1 (ERa), and KCNH3, Figures 2A, B). Pathway analysis of genes bound by p65 (-50kb to +2kb of TSS) revealed that genes with transcriptional regulatory functions (e.g. JUND), cell adhesion and migration (e.g. WNT5B, BCAR1, TGFB1, CXCL16), NFkB signaling, TNFa signaling, and apoptosis were enriched. This is in accordance with general NFkB functions (39,40). We also identified a novel pathway, not previously associated with p65, including circadian rhythm in both cell lines ( Figure 1E). Circadian rhythm genes bound by p65 included the central circadian regulator CLOCK, BMAL2, CREB1, and KLF10. In conclusion, we note highly concordant binding to cis-regulatory chromatin in proximity of genes within expected functions and further identify potential mechanism for p65 regulation of the colon circadian rhythm.

p65 Recruitment Correlates With TNFa-Regulated Gene Expression in CRC Cells
While binding of a transcription factor indicates a potential gene regulation of nearby or distant genes, all such bindings do not translate into actual gene regulation. To determine the effect p65 binding has on transcriptional regulation of corresponding genes in CRC cells, we linked the p65 ChIP-Seq data to our previously generated TNFa (2-h treatment) transcriptome data set of the same cells (22). We found coordinated p65 binding and short-term TNFa regulation of 274 genes in HT29 and 82 genes in SW480 cells ( Figure 3A, Table S3). Out of these, 59 genes (72% of those identified in SW480) were bound by p65 and regulated by TNFa in both cell lines (Table S3). All 59 core genes were upregulated by TNFa in both cell lines. Motif enrichment analysis of the p65-bound and TNFa-regulated core genes demonstrated significant enrichment for the p65 motif. Overall, the TNFa-regulated genes associated with p65-binding sites were mainly involved in the gene ontology functions of NFkB signaling, TNFa signaling, and inflammatory pathways. These included regulations and binding of known NFkB target genes (BCL3, CCL20, CXCL1, CXCL8, and NFKBIA, Fig 3B-C). Other known targets were differently bound and regulated in HT29 (BCL11B, DUSP16, KLF6, RELA) and SW480 (IL23A, PRRG1) cells (Table S3). In conclusion, this data clearly shows a strong transcriptional activity by p65 in both CRC cell lines, with p65-bound and regulated genes involved in critical CRC pathways, including apoptosis and cell migration.

p65 Cistrome Differs Between Colon and Breast Cancer Cell Lines
To explore the extent that p65 binding is conserved between colon and breast cancer, we compared our generated p65 cistrome (colon) with previously published p65 ChIP-Seq data of the ER-positive breast cancer cell line MCF7. We selected a data set that also used double crosslinking (41). A heatmap illustrating p65 chromatin-binding sites in MCF7, HT29 and SW480 cells are shown in Fig 4A. Only 22% (230 sites) of MCF7 p65 binding sites were present in either CRC cell line ( Figure  4B). Also, a markedly lower fraction of p65 sites were located by promoters in MCF7 cells (12% versus 21-22%, Figure 1D). The predominant motifs in MCF7 were NFkB (p65) itself, supporting accuracy of this data set, but were otherwise different (FOXA and AP2, Figure 4C) from colon (AP1, Figure 1C). The enriched biological functions for p65 sites specific for MCF7 also included apoptosis, transcription regulation, cell cycle, and circadian clock ( Figure 4D). Thus, our study shows a cell specificity of p65 binding, where it binds different motifs and regulates different genes in different tissues or cell lines, but the biological functions of the regulated genes appear to have similar roles in cancer cell lines.

ERb Diminishes p65 Chromatin Binding in HT29 Cells
Next, we aimed to study whether the mechanism whereby estrogen impacts inflammatory signaling in colon involves p65 chromatin binding. As we have previously found that ERb can attenuate pro-inflammatory cytokine IL6 signaling in CRC cell lines (23) and regulate several important NFkB target genes and TNFa signaling in vivo (22), we explored whether ERb impacts the p65 cistrome. We performed p65 ChIP-Seq in the same CRC cells, with and without (mock) expression of ERb. In HT29, we found that whereas 1721 sites remained bound by p65 in both conditions, 1430 p65 binding sites were no longer detected in presence of ERb. Further, a smaller fraction of 228 new binding sites were identified, only in presence of ERb ( Figure 5A). Using density plot, we noted that ERb reduced the overall p65 binding in all three replicates ( Figure 5B). We also analyzed this using a sliding window approach, with a window of 200 bp and calling for enriched regions between mock and ERb, and identified the same trend ( Figure 5C). Next, to investigate whether specific p65-binding motifs were affected by ERb, we performed de novo motif analysis. The predominant p65 motifs in HT29 remained both in the absence and presence of ERb (p65, AP-1, HNF4A, Figure 5D). Thus, the presence of ERb reduced p65 chromatin binding and affected its distribution (numerous sites disappeared) but did not affect the type of motifs bound. Corresponding pathway analysis indicated that ERb hindered p65 chromatin binding to genes with activities in cell adhesion, migration and circadian clock, while enabling binding by genes related to cell proliferation and Notch signaling ( Figure 5E).
We also compared the p65 ChIP-Seq data with corresponding TNFa gene expression data, with and without ERb ( Figure 5F). We identified that 162 of p65-bound genes were regulated by TNFa differently in presence of ERb. Among those, ERb also inhibited p65 binding by 51 genes and enhanced binding of 7 genes. A heatmap illustrates how ERb and resulting lack of p65 binding, affects TNFa-mediated regulation of these 51 genes ( Figure 5G). Notably, presence of ERb either inhibited TNFamediated response (50%), or enabled TNFa induction (31%) for the majority of these genes. These genes were mainly involved in functions such as negative regulation of transcription, negative regulation of cell proliferation, and chromatin remodeling. The seven genes where ERb enhanced p65 binding were also impacted in terms of TNFa-mediated gene expression ( Figure  5H). For example, ERb enabled p65 binding and TNFa upregulation of TSC22D1 and binding followed by downregulation of ZNF341, but blocked (presumably through p65 recruitment) TNFa-mediated upregulation of VEGFA, AGO2 and IGFL4. We thus identified that ERb reduces a sizeable fraction of p65 binding, modifying TNFa regulation in HT29 cells, especially for genes involved in e.g. cell proliferation and cell-cell adhesion.

ERb Enhances p65 Chromatin Binding in SW480 Cells
Similarly, we explored the impact by ERb on p65 chromatin binding in SW480 cells. Contrary to HT29, few binding sites were decreased upon introduction of ERb and nearly all (1433 sites) remained. However, a high number (5796) of new p65 binding sites appeared in presence of ERb ( Figure 6A). This enhanced p65 binding was also evident for all replicates in the density plot ( Figure 6B). Pathway analysis showed that p65 binding sites dependent on ERb were located by genes involved primarily in transcription regulation, GTPase activity, apoptotic process, protein phosphorylation, cell migration, and MAPK cascade ( Figure 6C, right panel). The p65 and AP-1 motifs remained highly enriched in SW480 ERb, but we note that RUNX2 motifs were more common when ERb was present ( Figure 6D). RUNX2 is also upregulated by ERb (23).
In terms of gene expression, ERb modulated transcription of 110 TNFa-regulated genes that also had p65 binding sites in SW480ERb cells ( Figure 6E). About half of these (53 genes) had p65 bound by regulatory chromatin only in presence of ERb. The corresponding TNFa-mediated gene regulation is illustrated in a heatmap ( Figure 6F). Interestingly, ERb (presumably by recruiting p65 to the chromatin) inhibited the TNFa activation of the majority of these genes (56%, or 30 genes). Another 16 TNFa-response genes required ERb and resulting p65 binding for their induction ( Figure 6F), including DUSP5 [which regulates inflammatory gene expression of TNFa (42)], nuclear receptor NR2F6, and KLF9. Gene ontology enrichment reveals that genes within cell proliferation and cell migration were enriched among the p65 regulations modified by ERb in SW480 cells. Despite the finding that ERb enhanced p65 binding in SW480 cells, the resulting transcription of these genes was mostly inhibited. ERb also modulated expression of p65-bound TNFa-regulated genes, without impacting p65 chromatin binding (57 genes). Among these, expression was attenuated in most (24), and enhanced in some (14). Genes suppressed by ERb included BCL3, BIRC3, CCL20, NFkB2 and RELB ( Figure 6G), all of which are associated with poor prognosis in CRC (43)(44)(45)(46)(47).
Thus, ERb clearly modulates p65 binding and TNFa response also in SW480 cells. ERb appeared to enhance p65 binding, but still repress TNFa transcriptional activity.

ERb Modulates p65 Signaling in Colon Cells
From the above findings, we conclude that ERb impacted p65 binding in both CRC cell lines, partly in different ways but with similar outcome in terms of TNFa-mediated gene regulation. Here, we compare p65 chromatin binding (ChIP-Seq data) between the two cell lines, with and without ERb (Figure 7). A heatmap illustrates the reduced p65 binding upon ERb expression in HT29, and the enhancement noted in SW480 ( Figure 7A). A Venn diagram comparing p65 binding in the four conditions (HT29 and SW480 with and without ERb, Figure 7B) shows that a large fraction of p65 sites (727 sites) are bound in all conditions, and 77 sites are enhanced by ERb in both HT29 and SW480. ERb appears to enable p65 binding 56 kb downstream of TSS of PROX1 in both cell lines ( Figure 7D). PROX1 is a transcription factor highly upregulated in colon cancers and previously shown to be regulated by ERb posttranscriptionally through miR-205 (21). Overall, 15 genes were bound by p65 and regulated by TNFa, both in absence and presence of ERb, in both cell lines ( Table S4). Out of these, the TNFa response of 12 genes was modulated in opposite direction by ERb in HT29 and SW480 cells, which include important target genes such as BIRC3, CXCL1, CXCL8 and PDGFB ( Figure 7C). We further identified 596 p65 binding sites opposed by presence of ERb in HT29 but, in contrast, bound only in presence of ERb in SW480 cell line ( Figure 7B). This core set of genes includes the well-known p65 interacting protein AP-1, NFIB, and circadian clock genes (CLOCK, CXCL10, RUNX1, TP53I11, NFIB, BMAL2/ARNTL2, Figure  7D). Altogether, these results indicate that in addition to conserved patterns there are also considerable cell specific differences in p65 binding between HT29 and SW480 cells, and that ERb impacts the p65 cistrome and TNFa response in both cell lines.

DISCUSSION
Activation of NFkB in the intestinal epithelia can lead to colitisinduced CRC (48). Our study attempts to understand the molecular mechanisms behind the CRC promoting role of inflammation, by studying the cistrome of p65 and investigating how this is impacted by ERb. We characterize the p65 genome-wide chromatin binding in two different CRC cell lines, and specify similarities and differences. We find that presence of ERb impacts p65 binding and corresponding TNFa-mediated transcription. Our work emphasizes that NFkB binds primarily through RELA and JUN-AP1 motifs in cells with a colon origin. Genes nearest to p65-bound chromatin were involved in inflammatory response, cell proliferation, cell migration, and, interestingly, circadian clock (BMAL2, CLOCK). A previous study has shown that dysregulation of circadian rhythm increases the risk for colorectal cancer (49). Indeed, CLOCK gene mutations have been identified in 53% of CRC with microsatellite instability (MSI) (50). Moreover, another study identified that mutations in the CLOCK1 gene increased the risk of developing CRC (51). In breast cancer cells, several studies identified a link between circadian genes and NFkB signaling pathway (52,53). In colon cancer, two studies have shown that REV-ERB-a through NFkB modulates circadian clock and reduced DSS-induced colitis (54,55). In our recent studies, we have showed that ERb can modulate the impact of TNFa-NFkB activity in CRC cell lines and in vivo using the AOM-DSS mouse model (22). We have also demonstrated that intestinal ERb regulates the expression of the circadian clock gene Bmal1 (Arntl1) in colon of HFD-fed mice (56). Here, in addition to identifying p65-binding sites, we demonstrate that the activation of the TNFa-NFkB axis impacts the expression of circadian genes. Moreover, we show that ERb interferes with the general p65 chromatin binding, including the circadian genes CLOCK and BMAL2 ( Figure 7C). Taken all this into account, our interpretation is that p65 modulates circadian genes in the colon in the pro-inflammatory pro-tumorigenic condition, and that ERb can change this and thereby oppose the inflammatory condition that drives development of colon cancer. Interestingly, p65 chromatin binding appears relatively distinct between the two CRC cell lines, in support with the fine-tuned cell-specific manner whereby NFkB controls transcriptional regulation. The cell lines are indeed different in several respects. While both are derived from primary colon adenocarcinomas, the HT29 cell line is derived from a likely premenopausal (44-year-old) woman, whereas SW480 originates from a 50-year-old man (57,58). We have also reported sex differences in the non-tumor and tumor transcriptome of CRC patients, which impacted biomarker discovery (59). The different female-male origin of the cell lines used here, may indeed impact the different regulation of p65 cistrome or its modulation by ERb. However, further studies are needed to clarify this. Moreover, HT29 cells are CIMP (CpG island methylator phenotype) positive, and SW480 cells are CIMP negative. Aberrant methylation of the CpG islands has been shown to impact chromatin binding and accessibility to transcription factors (60,61). Their mutational profile also differs, with HT29 having mutations in BRAF (V600E), PIK3CA (P449T), and p53 (R273H), and SW480 in KRAS (G12V) and p53 (double mutant alleles R273H and P309S, however still retaining functionality of many p53-associated pathways) (62,63). These proteins are important transcriptional regulators that can also influence the binding of transcription factors (64,65). In parental CRC cell lines HT29 and SW480, neither ERa nor ERb is expressed, while MCF7 cell line expresses ERa, which has been shown to interact with p65 (24). These factors may all modulate the p65 cistrome.
While the p65 binding pattern was similar between the two CRC cell lines, the p65 cistrome of breast cancer cell line MCF7 was more distinctly different. One of the well-known interactionpartner of p65 is the p53 protein (65). Recently, it was shown that mutant p53 enhances NFkB activity in mice, leading to chronic inflammation and associated CRC (65). Another study demonstrated that p53 mutants directly interact with NFkB in SW480 cells (15). MCF7 cell line has wild type p53, whereas both HT29 and SW480 cells express the R273H p53 mutant protein, which inhibits DNA binding (66,67). Hence it is possible that the p53 status impacts p65 cistrome in these cell lines, and further studies are needed to explore this hypothesis.
We have previously shown that TNFa triggers a transcriptional response in both CRC cell lines, and that ERb modulates this (22). Here, we correlate the transcriptional response with p65 chromatin binding sites, and how ERb modifies the p65 cistrome. To be noted, in order to optimize experiments, different treatments times and concentrations of TNFa were used in ChIP-Seq (30 ng/ml, 30 min) and for transcriptional analysis (10 ng/ml, 2h). Further, the transduced ERb is expressed at higher levels in SW480 compared to HT29 (1.8 times more), and previous data suggests that TNFa may increase transactivation of ERb (22). These factors may all influence the kinetics of the mechanisms described here, but are not expected to have a major influence on the mechanism per se. Our findings offer mechanistic underpinnings of how inflammation modulates specific signaling pathways and how ERb can attenuate cytokine-induced carcinogenic response in CRC cells.
The strength of this study includes the genome-wide approach to decrypt these interactions, which together with the validated high-quality ChIP-Seq data generates unbiased and reliable data. The significance of these findings is reinforced by the use of two different CRC cell lines and the comparison with transcriptional impact, as well as comparisons between our results and published data generated from cells of other origin. A shortcoming includes the use of exogenous expression of ERb in the CRC cell lines. However, cell lines mostly lack endogenous expression of ERb (21,22). Further, it may be preferable to use non-tumor colonic cell lines, as one aim of the study was to investigate how p53 can prevent CRC through its impact on NFkB signaling. However, a key interest was also to decipher the oncogenic NFkB signaling in CRC, and along with the lack of suitable non-tumor cell lines at hand, and our access to highly characterized cell lines with exogenous expression of ERb, this is the balance we chose. Further, the difference of antibodies used between our study and the breast cancer tissue study (41), may contribute to the differences found.
In conclusion, we provide a mechanistic foundation for a better understanding of how estrogen influences inflammatory signaling through NFkB in CRC cells.

DATA AVAILABILITY STATEMENT
The p65 ChIP-Seq data is deposited in the Gene Expression Omnibus (GEO) repository (GSE160856), TNFa bead array gene expression data was published previously (available at GSE65979), and SW480ERb input and HT29ERb input controls (GSE149979).

AUTHOR CONTRIBUTIONS
CW contributed to conceptualization. Methodology performed by RI, LH, and AA. Validation done by RI and LH. Formal analysis was done by RI and LH. Curation, RI. RI and CW wrote the first original draft and all authors commented or edited. Visualization, RI. Supervision, CW and AA. Project administration, CW. All authors contributed to the article and approved the submitted version.