Novel Mechanistic Targets of Forkhead Box Q1 Transcription Factor in Human Breast Cancer Cells

bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Novel mechanistic targets of Forkhead box Q1 transcription factor in human breast cancer cells

Su-Hyeong Kim*,1, Eun-Ryeong Hahm*,1, Krishna B. Singh*, and Shivendra V. Singh*,¶,2

From the *Department of Pharmacology & Chemical Biology, and ¶UPMC Hillman Cancer Center, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

Running Title: Role of FoxQ1 in breast cancer

2To whom correspondence should be addressed: 2.32A Hillman Cancer Center Research Pavilion, UPMC Hillman Cancer Center, 5117 Centre Avenue, Pittsburgh, PA 15213. Phone: 412-623-3263; Fax: 412-623-7828; E-mail: [email protected]

Keywords: Breast Cancer, FoxQ1, Cell Cycle, Interleukin-1α, Interleukin-8

The transcription factor forkhead box Q1 analysis. FoxQ1 overexpression resulted (FoxQ1), which is overexpressed in in downregulation of genes associated different solid tumors, has emerged as a with cell cycle checkpoints, M phase, and key player in the pathogenesis of breast cellular response to stress/external stimuli cancer by regulating epithelial- as evidenced from the Reactome pathway mesenchymal transition, maintenance of analysis. Consequently, FoxQ1 cancer-stem like cells, and metastasis. overexpression resulted in S, G2M and However, the mechanism underlying mitotic arrest in basal-like SUM159 and oncogenic function of FoxQ1 is still not HMLE cells, but not in luminal-type fully understood. In this study, we MCF-7 cells. There were differences in compared the RNA-seq data from FoxQ1 expression of cell cycle-associated overexpressing SUM159 cells with that of proteins between FoxQ1 overexpressing empty vector-transfected control (EV) SUM159 and MCF-7 cells. Finally, we cells to identify novel mechanistic targets show for the first time that FoxQ1 is a of this transcription factor. Consistent direct transcriptional regulator of with published results in basal-like interleukin (IL)-1α, IL-8, and vascular subtype, immunohistochemistry revealed endothelial growth factor in breast cancer upregulation of FoxQ1 protein in cells. Chromatin immunoprecipitation luminal-type human breast cancer tissue revealed FoxQ1 occupancy at the microarrays when compared to normal promoters of IL-1α, IL-8, and VEGF. In mammary tissues. Many previously conclusion, the present study reports reported transcriptional targets of FoxQ1 novel mechanistic targets of FoxQ1 in (e.g., E-cadherin, N-cadherin, fibronectin human breast cancer cells. 1, etc.) were verified from the RNA-Seq challenging aspect in clinical management Introduction of breast cancer relates to molecular Breast cancer remains an alarming heterogeneity of the disease that is health concern for women worldwide characterized by distinct gene expression reflected by more than 40,000 deaths each signatures and overexpression of driver year in the United States alone (1). A oncogenic proteins (2-5). Majority of the

1 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

invasive mammary ductal carcinomas are immortalized by hTERT and SV40 large T broadly grouped into four subtypes, antigen (HMLER) also promoted stem-like including luminal A type [estrogen receptor phenotype characterized by increased positive (ER+3), progesterone receptor mammosphere multiplicity (8). Importantly, positive (PR+), and human epidermal suppression of FoxQ1 expression in MDA- growth factor receptor 2 positive (HER2+)], MB-231 cells resulted in increased active luminal B type (ER+/PR+/HER2-), HER2- caspase-3 level in response to treatment with enriched, and basal-like (2,3). Nearly 75% several chemotherapy drugs, including 5- of basal-like breast tumors are triple- fluorouracil, taxol, and camptothecin (8). negative due to lack of ER, PR, and HER2 Transcriptional inactivation of E-cadherin expression (6). Further characterization of by FoxQ1 overexpression was also the disease subtype-independent oncogenic demonstrated in this study (8). dependencies in breast cancer is necessary to Subsequently, platelet-derived growth factor identify novel druggable targets to broaden receptor α/β were identified as additional therapeutic options for different subtypes of downstream targets of FoxQ1 in promotion breast cancer. of breast cancer stem cell-like phenotype as well as chemoresistance (9). We have also The transcription factor forkhead box shown previously that FoxQ1 promotes Q1 (FoxQ1) has recently emerged as a key breast cancer stem-like phenotype in a player in the pathogenesis of breast cancer luminal-type (MCF-7) and a basal-like (7-9). Zhang et al. (7) were the first to (SUM159) cell line by causing direct demonstrate a role for FoxQ1 in epithelial- transcriptional repression of the tumor mesenchymal transition (EMT) in breast suppressor Dachshund homolog 1 (DACH1) cancer metastasis using a cross-species gene (10). expression profiling strategy. Forced expression of FoxQ1 in a human mammary Published studies thus far clearly epithelial cell line (HMLE) and EpRas cells indicate that FoxQ1 regulates expression of increased their ability to migrate and invade multiple cancer-relevant genes to promote in vitro, and these effects were reversible by cell invasion/migration, stem-like knockdown of this protein in 4T1 mouse phenotype, and chemotherapy resistance in mammary carcinoma cells (7). Moreover, vitro and metastasis in vivo in breast cancer FoxQ1 overexpressing EpRas cells exhibited cell lines, but the possibility of additional increased propensity for pulmonary functionally important targets of this metastasis in vivo (7). Promotion of the transcription factor can’t be fully ignored. EMT phenotype by FoxQ1 overexpression To address this question, we compared was due to direct repression of E-cadherin RNA-Seq data from FoxQ1 overexpressing protein expression (7). In another study, SUM159 (hereafter abbreviated as FoxQ1 FoxQ1 expression was shown to correlate cells) cells with that of empty vector with high-grade basal-like breast cancers transfected control cells (hereafter and associated with poor clinical outcomes abbreviated as EV cells) to identify its (8). RNA interference of FoxQ1 in a highly additional mechanistic targets. The present invasive basal-like human breast cancer cell study reveals novel downstream targets of line (MDA-MB-231) attenuated EMT FoxQ1 in breast cancer including cell cycle phenotype and decreased its invasion ability control, interleukin (IL)-1α, IL-8, and (8). Ectopic expression of FoxQ1 in a vascular endothelial growth factor (VEGF). human mammary epithelial cell line

2 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Results Volcano plot in Fig. 1D depicts distribution of differentially expressed genes between FoxQ1 protein is overexpressed in luminal- FoxQ1 overexpressing SUM159 cells and type human breast cancers compared to EV cells with a threshold cut-off of 1.3-fold normal mammary tissues change in expression and at an adjusted p We have shown previously that the value of < 0.05 (Fig. 1D). The heatmaps of level of FoxQ1 protein and mRNA is three replicates of each group exhibiting relatively higher in basal-like human breast highly consistent transcriptional changes are cancer cell lines (e.g., MDA-MB-231) in shown in Fig. 1E. The Venn diagram shown comparison with a normal mammary in Fig. 1F shows unique and overlapping epithelial cell line (MCF-10A) or luminal- gene expression between FoxQ1 type cells (e.g., MCF-7 and MDA-MB-361) overexpressing SUM159 cells and EV cells. (10). Moreover, basal-like human breast cancers exhibited higher level of FoxQ1 Kyoto encyclopedia of genes and genomes protein when compared to normal mammary (KEGG) pathway analysis tissues (10). Initially, we compared the The top 5 pathways with upregulated expression of FoxQ1 gene expression in genes from the KEGG analysis included white versus black breast cancer patients and pathways in cancer (84 genes), mitogen- between ductal and lobular carcinomas in activated protein kinase signaling (56 breast cancer TCGA dataset. The expression genes), regulation of actin cytoskeleton (49 of FoxQ1 was marginally but significantly genes), focal adhesion (50 genes), and axon higher in the tumors of black women when guidance (37 genes) (Fig. 2A). The top 5 compared to white women (Fig. 1A). On the pathways with downregulated genes from other hand, the expression of FoxQ1 was the KEGG pathway analysis were cell cycle comparable between ductal and lobular (41 genes), oocyte meiosis (38 genes), breast cancers (Fig. 1A). In this study, we progesterone-mediated oocyte maturation also examined the expression of FoxQ1 (29 genes), lysine degradation (18 genes), protein in tissue microarrays consisting of and protein processing in endoplasmic luminal-type breast cancers and normal reticulum (46 genes) (Fig. 2B). mammary tissues. Fig. 1B shows immunohistochemical staining for FoxQ1 in The gene ontology (GO) and Reactome representative normal mammary tissues and pathway analyses luminal-type breast cancers. The H-score for All three ontologies including FoxQ1 expression was higher by 1.6-fold in cellular component, molecular function, and luminal-type breast cancers when compared biological processes are included in the GO to normal mammary tissues (Fig. 1C). These enrichment analysis. The GO enrichment results indicated overexpression of the analysis revealed that FoxQ1 overexpression FoxQ1 protein in luminal-type human breast caused upregulation of genes associated cancers. with positive regulation of locomotion, axon development, positive regulation of cellular FoxQ1-regulated transcriptome in component movement, cell SUM159 cells substrate/adherens junction, focal adhesion, Next, we performed RNA-Seq positive regulation of cell analysis using SUM159 cells to identify motility/migrations and a few other additional targets of FoxQ1. The mapping pathways (Fig. 2C). The downregulated results are summarized in Table 1. The genes following FoxQ1 overexpression in

3 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

SUM159 cells were mostly associated with findings of Zhang et al. (7). Meng et al. (9) the GO pathway terms related to cell cycle showed downregulation of CST6, SEMA3A, regulation (GO terms: regulation of cell ADAM9, THBS1, FOXA1, and EDN1 but cycle phase transition, regulation of mitotic upregulation of PDGFRA, JAM3, and ZEB2 cell cycle phase transition, nuclear division, in FoxQ1 overexpressing human mammary nuclear chromosome segregation, epithelial cell line (HMLE) when compared microtubule, cell cycle G2/M phase to EV cells, and these changes were also transition, centromeric region, G2/M observed in the RNA-Seq data in the present transition of mitotic cell cycle, etc.) (Fig. study (Fig. 3B). We have shown previously 2D). that promotion of stem-like phenotype in The Reactome database contains FoxQ1 overexpressing SUM159 cell line is annotations for diverse set of molecular and associated with suppression of DACH1 and cell biological topics such as cell cycle, ZEB1 expression and upregulation of metabolism, signaling, transport, cell mRNA levels of MYC and TWIST2 (10) and motility, immune function, host-virus RNA-Seq results from the present study interaction, and neural function. Genes were consistent with these published associated with phospholipid metabolism, observations (Fig. 3C). At the same, some VEGF/VEGF receptor pathway, published changes in genes downstream of transforming growth factor β pathway were FoxQ1 (7-9,11) were not validated in the significantly upregulated in FoxQ1 RNA-Seq data from the present study, overexpressing SUM159 cells (Fig. 2E). including, PLD1, SNAI1, CTNNB1, JUP, Like KEGG and GO pathway analyses, most CD44, S100A4, DCN, PDGFRB, CADM3, downregulated genes from the Reactome COL1A1, COL6A1, DSG2, and TWIST1 analyses were associated with cell cycle (Fig. 4A-D). Two possibilities exist to regulation (Fig. 2F). Genes associated with explain the inconsistencies between cellular response to stress/external stimuli published data and RNA-Seq results. The were also significantly downregulated by RNA-Seq results were obtained using FoxQ1 overexpression (Fig. 2F). SUM159 cells and this cell line was not used in other published studies (8,9) except for Comparison of RNA-Seq data with our own published study (10). Secondly, the published literature possibility of regulatory functions of other Published studies have revealed transcription factors can’t be excluded. multiple downstream targets of FoxQ1 in breast cancer (7-12). For example, Zhang et Effect of FoxQ1 overexpression of cell al. (7) showed downregulation of CDH1 (E- cycle progression cadherin ) but upregulation of mesenchymal Flow histograms for cell cycle markers including CTNNA1 (catenin alpha distribution in EV and FoxQ1 1), CDH2 (N-cadherin), and FN1 overexpressing SUM159 cells after serum (fibronectin 1) in highly metastatic breast starvation and collection at different time cancer cell lines (e.g., MDA-MB-435, points after release into the complete MDA-MB-231, SUM149, etc.) in medium are shown in Fig. 5A. As expected, comparison with breast cancer cell lines serum starvation caused accumulation of with low metastatic capacity (e.g., MCF-7, G0/G1 phase fraction in both cell lines. The MDA-MB-361, BT20, etc.). As shown in FoxQ1 overexpressing SUM159 cells, but Fig. 3A, the RNA-seq results from the not MCF-7 cells, exhibited a significant present study were consistent with the decrease in fraction of G0/G1 phase cells that

4 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

was accompanied by S and G2/M phase cell overexpression with respect to cell cycle cycle arrest (Fig. 5B). These results progression. indicated S and G2/M phase cell cycle arrest upon FoxQ1 overexpression in SUM159 IL-1α, IL-8, and VEGF are novel cells but not in MCF-7 cells. downstream transcriptional targets of FoxQ1 Next, we performed western blotting The Reactome pathways analysis and flow cytometry to quantitate mitotic also showed downregulation of genes fraction in FoxQ1 overexpressing cells and associated with cellular response to stress corresponding EV cells. Opposite effects upon FoxQ1 overexpression in the SUM159 were observed in basal-like SUM159 cells cell line (Fig. 2F). Table 2 lists genes and luminal-type MCF-7 cells. FoxQ1 associated with cellular response to stress overexpression in SUM159 cells resulted in whose expression was altered upon FoxQ1 increased Ser10 phosphorylation of histone overexpression. We compared the H3, which is a marker of mitotic cells (Fig. correlations observed from the RNA-Seq 6A). In the MCF-7 cell line, Ser10 data (present study) with The Cancer phosphorylation of histone H3 was Genome Atlas (TCGA) dataset. Positive significantly lower in FoxQ1 overexpressing correlation was observed for several cells than in EV cells (Fig. 6A). These upregulated or downregulated genes from results were confirmed by flow cytometry the RNA-Seq data and the breast cancer (Fig. 6B,C). Because of the cell line-specific TCGA analysis (Table 2). On the other differences, we also performed similar hand, RNA-Seq data was not consistent with experiments in another basal-like cell line that of breast cancer TCGA for a few other (HMLE). The mitotic arrest by FoxQ1 genes (identified by negative correlation in overexpression was also observed in the Table 2). Because FoxQ1 has oncogenic HMLE cells similar to SUM159 (Fig. 6 function, we focused on IL-1α for further B,C). These results indicated different role investigation. RNA-seq data indicated a 23- for FoxQ1 in basal-like versus luminal-type fold increase in expression of IL-1α in breast cancer cells. FoxQ1 overexpressing SUM159 cells compared with EV cells (Fig. 8A). The Overexpression of FoxQ1 in the breast cancer TCGA analysis revealed a MCF-7 cell line resulted in suppression of significant positive association between protein levels of cyclin-dependent kinase 2 expression of IL-1α and that of FoxQ1 (Fig. (CDK2) and Cyclin D1, but upregulation of 8B). qRT-PCR confirmed overexpression of cell division cycle 25C (CDC25C), CDK4, IL-1α mRNA upon forced expression of and Cyclin B1 protein expression (Fig. 7). FoxQ1 in SUM159 cells (Fig. 8C). Secretion The protein levels of Cyclin A and CDK1 of IL-1α in the media was also significantly were not affected by FoxQ1 overexpression higher in the FoxQ1 overexpressing in the MCF-7 cell line (Fig. 7). The results SUM159 cells when compared to EV cells were strikingly different in the SUM159 cell (Fig. 8D). The promoter of IL-1α contains a line stably transfected with FoxQ1 that single FoxQ1 binding consensus sequence exhibited downregulation of CDC25C, of TGTTTA (9). Chromatin Cyclin B1, and CDK1 but upregulation of immunoprecipitation (ChIP) confirmed Cyclin D1, Cyclin A, and CDK4 (Fig. 7). recruitment of FoxQ1 at the promoter of IL- Collectively, these results indicated cell line- 1α in SUM159 cells (Fig. 8E). The specific responses upon FoxQ1 expression of IL-1α was very low in MCF-7

5 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

cells when compared to SUM159 and thus FoxQ1 and VEGFA in MCF-7 cells, ChIP these experiments were not performed in the assay was not performed in this cell line. former cell line. These results indicated that IL-1α is a direct transcriptional target of Discussion FoxQ1 at least in the SUM159 cell line. FoxQ1, also known as HNF-3/fkh homolog-1 or HFH-1, belongs to the The RNA-seq data indicated a forkhead family of transcription factors and significantly higher level of IL-8 mRNA in its normal physiological functions include FoxQ1 overexpressing SUM159 cells regulation of hair differentiation and control compared to EV cells (Fig. 9A). A of mucin expression and granule content in significant positive association between stomach surface mucous cells (12-15). The expression of IL-8 and that of FoxQ1 was FoxQ1 gene is located at chromosome also observed in the breast cancer TCGA 6p25.3 that encodes a 403-amino acid dataset (Fig. 9B). Overexpression of IL-8 in protein (16,17). Structurally, the FoxQ1 is FoxQ1 overexpressing SUM159 cell was characterized by alanine- and glycine-rich confirmed by qRT-PCR (Fig. 9C). The level region, the forkhead box domain (Winged- of IL-8 in the media was also significantly helix or DNA-binding domain), and proline- higher in the FoxQ1 overexpressing rich region (17). The FoxQ1 is an SUM159 cells when compared to EV cells evolutionary conserved protein with 100% (Fig. 9D). Three FoxQ1 binding consensus amino acid sequence similarity in the DNA sequences were observed at the promoter of binding domain between human, mouse, and IL-8. ChIP assay confirmed recruitment of rat (17,18). Overexpression of protein or FoxQ1 at two of those sites of IL-8 promoter mRNA levels of FoxQ1 has been reported in (Fig. 9E). Similarly, FoxQ1 overexpressing different cancers, including gastric cancer, MCF-7 cells exhibited a significantly higher hepatocellular carcinoma, laryngeal level of IL-8 secretion in comparison with carcinoma, pancreatic cancer, colorectal EV cells (Fig. 10A). Furthermore, FoxQ1 cancer, and breast cancer (8,10,19-24). We was recruited at two sites of the IL-8 showed previously that the level of FoxQ1 promoter in MCF-7 cells (Fig. 10B). protein is higher by 18.4-fold in basal-like human breast cancers compared to normal The Reactome pathway analysis also mammary tissue (10). Because luminal-type revealed upregulation of genes associated disease accounts for majority of human with VEGF/VEGFR2 pathway in FoxQ1 breast cancers, in this study we analyzed the overexpressing SUM159 cells. The RNA- level of FoxQ1 protein in this subtype. Like Seq data revealed upregulation of VEGFA basal-like breast cancers, the expression of mRNA in FoxQ1 overexpressing SUM159 FoxQ1 protein was significantly higher in cells when compared to EV cells (Fig. 11A). the luminal-type human breast cancers in Analysis of the breast cancer TCGA dataset comparison with normal breast tissues. also revealed a positive correlation between These results suggest that oncogenic expression of FoxQ1 and that of VEGF (Fig. function of FoxQ1 may span across disease 11B). The VEGF mRNA level (Fig. 11C) subtypes in human breast cancers. and/or secretion of the protein (Fig. 11C) were increased upon FoxQ1 overexpression. In breast cancer, FoxQ1 is well- ChIP assay indicated recruitment of FoxQ1 known for its role in promotion of EMT, cell at all 3 sites of VEGFA (Fig. 11E). Because migration and invasion, self-renewal of of weak association between expression of breast cancer stem like cells, and metastasis

6 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

(7-10). Consistent with these published vivo was also significantly suppressed by results (7-10), the GO pathways analysis of DACH1 overexpression in MDA-MB-231 the RNA-Seq data also indicated cells (26). DACH1 was also shown to inhibit upregulation of genes associated with EMT in breast cancer cells in vitro and positive regulation of locomotion, positive repress metastatic potential of 4T1/Luc regulator of cellular component movement, murine breast carcinoma cells in vivo in positive regulation of cell motility, and syngeneic mice (28). Our own group was positive regulation of cell migration. At the first to implicate FoxQ1 in direct same time, the RNA-Seq data also suggests transcriptional repression of DACH1 (10). In additional functions of FoxQ1 based on GO one study, DACH1 was shown to repress IL- and Reactome pathway analyses, including 8 level (30). Collectively, these studies regulation of cell morphogenesis, regulation indicate that FoxQ1 is a druggable target in of actin cytoskeleton organization, peptidyl- breast and other cancers but a selective tyrosine dephosphorylation, and so forth. inhibitor of this transcription factor is yet to Further work is necessary to systematically be identified. test the role of FoxQ1 in these processes. We found that downregulation of Studies have revealed direct genes associated with cell cycle regulation regulation of multiple cancer-relevant genes was the most prominent pathway enrichment by FoxQ1 (7-10). FoxQ1 functions to by FoxQ1 overexpression from KEGG, GO, repress E-cadherin expression in breast and Reactome pathway analyses (Fig. 2). cancer cell lines (8). For example, the We also found alterations in expression of promoter activity of E-cadherin was many key proteins of the cell cycle repressed by overexpression of FoxQ1 in regulation by FoxQ1 overexpression, but HMLER cells but restored by knockdown of there are differences between SUM159 and FoxQ1 in MDA-MB-231 cells (8). Another MCF-7 cells. Consequently, FoxQ1 study confirmed these findings in 293FT and overexpression in SUM159 cells, but not in MCF-7 cells (7). The three regulatory E- MCF-7, results in S, G2M and mitotic arrest. boxes of the E-cadherin promoter were From the cell cycle distribution changes at shown to cooperate in its repression least in SUM159 cells, one would expect a mediated by FoxQ1 overexpression (7). pronounced effect of FoxQ1 on cell Additional direct targets of FoxQ1 in breast proliferation and cell survival but the cancer include transcription factors (Twist1 published in vitro data do not support this and Zeb2) and PDGFRα/β growth factor possibility at least in breast cancer (7,8). To receptors to confer resistance to the contrary, overexpression of FoxQ1 in chemotherapy drugs like doxorubicin, non-small cell lung cancer cell lines (A549 paclitaxel, and imatinib in vitro and in vivo and HCC827) results in increased (9). The DACH1 homolog gene was proliferation when compared to originally identified as a critical regulator of corresponding EV cells (31). Cell eye and limb development in Drosophila proliferation is decreased upon FoxQ1 (25). Studies have pointed to a tumor knockdown in a laryngeal carcinoma cell suppressor function of DACH1 in breast line (21). Because the full spectrum of the cancer (26-30). For example, overexpression FoxQ1-regulated transcriptome is still not of DACH1 inhibited breast cancer stem-like fully appreciated, further work may shed fraction in vitro and in vivo (27). Colony light as to why FoxQ1 overexpression is formation in vitro and xenograft growth in

7 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

unable to increase breast cancer cell binds to VEGF receptor to promote vascular proliferation. endothelial growth, migration, survival, and lymphangiogenesis (44,45). The present The present study, for the first time study shows that VEGFA is a direct demonstrates that IL-1α, IL-8, and VEGFA transcriptional target of FoxQ1. However, are direct transcriptional targets of FoxQ1. this regulation seems more pronounced in FoxQ1 is recruited at the promoters of IL- the SUM159 than in MCF-7 cells. Thus, it is 1α, IL-8, and VEGFA genes. Moreover, reasonable to conclude that VEGFA protein levels of IL-1α, IL-8, and/or VEGFA regulation by FoxQ1 is partly responsible are increased by FoxQ1 overexpression in for its pro-migratory and metastatic effects. SUM159 and/or MCF-7 cells. In breast cancer patients, IL-1α protein secretion is In conclusion, the present study correlated with malignant phenotype (32). reveals novel targets and functions of FoxQ1 IL-1α-derived from cancer cells was shown in breast cancer. The RNA-Seq data also to promote autocrine and paracrine reveals additional direct regulatory targets of induction of pro-metastatic genes in breast FoxQ1 at least in SUM159 cells that require cancer cells (33). Interestingly, IL-1α further verification. For example, the protein expression in breast cancer Reactome pathways analysis of the RNA- correlated with that of IL-8 (34), and both Seq data from SUM159 cells indicates these cytokines are direct targets of FoxQ1 upregulation VEGF/VEGFR2 pathway, (present study). Similar to the IL-1α, studies which plays an important role in endothelial have established a pro-tumorigenic function cell proliferation and capillary tube of IL-8 in breast cancer (35-40). For formation (42), suggesting that this example, the IL-8 expression in breast transcription factor can promote tumor cancer cells is proportional to their invasion angiogenesis. potential (37,38). The invasiveness of breast cancer cells is increased by overexpression Experimental Procedures of IL-8 as well as treatment with recombinant IL-8 (37,38). RNA interference Reagents and cell lines of IL-8 in MDA-MB-231 cells diminishes Cell culture reagents including fetal its ability to invade (39). Studies have also bovine serum, cell culture media, phosphate- indicated that IL-8 can promote stemness buffered saline (PBS), and antibiotic mixture and EMT in breast cancer cells (41). These were purchased from Life Technologies- results indicate that promotion of EMT and Thermo Fisher Scientific (Waltham, MA). stem-like phenotypes in breast and, possibly Antibodies against CDK2, CDK4, Cyclin A, other cancers by FoxQ1 overexpression is Cyclin D1, CDC25C, Cyclin B1, CDK1, partly mediated by direct regulation of the and FoxQ1 (for chromatin IL-1α and/or IL-8 expression. immunoprecipitation assay) were from Santa Cruz Biotechnology (Dallas, TX). An More than 40,000 American women antibody specific for detection of phospho- succumb to metastatic breast cancer every (Ser10) histone H3 was from Cell Signaling year (1). Increased tumor angiogenesis is Technology (Danvers, MA). Anti-β-Actin considered critical for tumor growth antibody was from Sigma-Aldrich (St. (42,43). The VEGFA, routinely known as Louis, MO). VEGF, is one of the major mediators of The MCF-7 cell line was purchased from tumor neo-angiogenesis (44). The VEGF the American Type Culture Collection and

8 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

authenticated by us in 2015 and 2017. The for RNA-Seq analysis. Prior to RNA-Seq SUM159 cell line was purchased from analysis, overexpression of FoxQ1 was Asterand (Detroit, MI) and authenticated by confirmed as done in our previous study us in 2015 and 2017. Details of stable (10). Cells (5×105 cells/6-cm dish) were transfection of MCF-7 and SUM159 cells harvested by trypsinization and used for with pCMV6 empty vector and the same RNA isolation and RNA-Seq analysis. The vector encoding FoxQ1 and their culture RNA-Seq analysis was performed by conditions have been described by us Novogene (Sacramento, CA). Other details previously (10). The HMLE cells stably of RNA-Seq analysis were essentially same transfected with FoxQ1 and EV cells were as described by us previously (46). Analysis generously provided by Dr. Guojun Wu was performed with a combination of (Karmanos Cancer Institute, Departments of software programs including STAR, HTseq, Oncology and Pathology, Wayne State Cufflink, and wrapped scripts. Tophat University, Detroit, MI) and maintained as program was used for alignments and recommended by the provider (7). DESeq2 was used for analysis of differential gene expressions. RNA-Seq data presented Immunohistochemistry for FoxQ1 protein in this study have been submitted to the expression in tissue microarrays Gene Expression Omnibus of NCBI Expression of FoxQ1 protein in (GSE151059). tissue microarrays of human luminal-type breast cancers (US Biomax, Rockville, MD; Flow cytometry for analysis of cell cycle catalog # BR1508) and normal human distribution and mitotic fraction mammary tissues (US Biomax, Rockville, EV cells and FoxQ1 overexpressing MD; catalog # BRN801a) was determined SUM159 (3 × 105 cells/dish) and MCF-7 by immunohistochemistry essentially as cells (6 × 105 cells/dish) were plated in 6-cm described by us previously (10). At least culture dishes in triplicate. After overnight three randomly-selected and non- incubation, the cells were rinsed with PBS overlapping fields on each core of the tissue and changed to serum-free medium for 16 microarray were examined using nuclear hours to synchronize cells in G0/G1 phase. algorithm of the Aperio ImageScope The cells were then released into serum- software (Leica Biosystems, Buffalo Grove, containing complete medium to resume cell IL). Data is expressed as H-score that is cycle. The cells were harvested with based on intensity (0, 1+, 2+, and 3+) % trypsinization at specified time points positivity (0-100%) and calculated using the followed by fixation in 70% ethanol following formula: (% of negative cells × 0) overnight at 4°C. Fixed cells were washed + (% of 1+ cells ×1) + (% of 2+ cells ×2) + with PBS, stained with propidium iodide (50 (% of 3+ cells ×3). Some specimens for µg/mL), and RNase A (80 µg/mL) for 30 normal breast tissue (n=10) and luminal- minutes and then analyzed using a BD type breast cancer (n=2) were not available AccuriTM C6 flow cytometer (BD for analysis due to poor staining and/or less Biosciences). For quantitation of the mitotic than optimal sectioning on the tissue fraction, cells were plated (SUM159 and microarray (cracked or folded). HMLE cells - 3 × 105 cells/6-cm dish and MCF-7 cells - 6 × 105 cells/6-cm dish), RNA-seq analysis incubated overnight, and then replaced with FoxQ1 overexpressing SUM159 fresh medium and incubated for 24 hours. cells and EV cells (n=3 for each) were used The cells were collected by trypsinization

9 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

and fixed in 70% ethanol at 4 °C for 2 hours. Reverse; 5’- Subsequently, the cells were permeabilized GGTGTCGCTGTTGAAGTCAGAG-3’. with 0.25% Triton X-100 for 15 minutes, and incubated with Alexa Fluor 488- ChIP assay conjugated phospho-(Ser10) histone H3 The ChIP assay was performed antibody for 1 hour followed by staining according to the manufacturers’ protocol with propidium iodide (50 µg/mL) and (Magnetic Chip kit, Pierce, Rockford, IL) RNase A (80 µg/mL) for 30 minutes at room using normal mouse IgG and FoxQ1 temperature. Stained cells were analyzed antibodies. Other details of ChIP assay have using BD AccuriTM C6 flow cytometer. been described by us previously (10). Putative FoxQ1 binding sites at the IL-8, IL- Western blotting 1α, and VEGFA promoters were amplified Details of western blotting have been (60°C, 1 min, 40 cycles) with the following described by us previously (47). The blots region-specific primers: for IL-8 site #1, 5′- were stripped and re-probed with β-Actin ATGCACTGTGTTCCGTATGC -3′ antibody for normalization. Immunoreactive (forward) and 5′- bands were detected by the enhanced GCTTTGCTAGTACAGGACAGG-3′ chemiluminescence method. Densitometric (reverse); site #3, 5′- quantitation was done using UN-SCAN-IT TGCTTTCTTCTTCTGATAGACCA-3′ software (Silk Scientific, Orem, UT). (forward) and 5′- TGTTAACAGAGTGAAGGGGCA-3′ (reverse); for IL-1α, 5’- Quantitative real-time polymerase chain TTCTTTGGTGAACTGAGGCA-3’ reaction (qRT-PCR) (forward) and 5’- qRT-PCR was performed as GTGAGCTGCTATGGAGATGC-3’ described by us previously (46) and relative (reverse); for VEGFA site #1, 5’- gene expression was calculated using the AAGGTGAGGCCCTCCAAG-3’ (forward) method of Livak and Schmittgen (48). PCR and 5’-ACCTAGCAGATTGGGGGAAG- was performed using 2× SYBR green qPCR 3’(reverse); site #2, 5’- kit (Thermo Fisher Scientific) with 95ºC (15 GGAGGACAGTTGGCTTATGG- sec), 60ºC (30 sec), and 72ºC (20 sec) for 40 3’(forward) and 5’- cycles. Primers for IL-8 and IL-1α were GCCAACAGACCTGAAAGAGC- purchased from GeneCopoeia (Rockville, 3’(reverse); site #3, 5’- MD). VEGFA was amplified (95ºC-15 sec, TCCAGATGGCACATTGTCAG- 60ºC-1 min for 40 cycles) with the 3’(forward) and 5’- following primers and Glyceraldehyde 3- TCTGGCTAAAGAGGGAATGG- phosphate dehydrogenase (GAPDH) was 3’(reverse). Fold enrichment was used as a normalization control. The primers normalized to the input. used were as follows: for VEGFA Forward:5’- Analysis of breast cancer TCGA data ATCTTCAAGCCATCCTGTGTG-3’; RNA-Seq data from TCGA database Reverse; 5’- for breast cancer (n = 1,097) was analyzed CAAGGCCCACAGGGATTTTC-3’ and for using the University of California Santa GAPDH Forward: 5’- Cruz Xena Browser GGACCTGACCTGCCGTCTAGAA-3’; (https://xena.ucsc.edu/public/). The correlation coefficient and statistical significance was determined by Pearson test.

10 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Measurement of IL-8, IL-1α, and VEGFA levels Cells were plated in 10-cm dishes at a density of 6 × 105 (SUM159) or 1 × 106 (MCF-7) cells per dish. The secretion of IL-1α, IL-8, and VEGFA in medium was determined using kits from Quantikine R & D System ELISA kits (Minneapolis, MN) according to the instructions provided by the supplier. The value of IL-1α in EV cells was below detection limit, and therefore a value of 1 was assigned for comparison with FoxQ1 overexpressing SUM159 cells.

Statistical analysis Statistical tests were performed using GraphPad Prism (version 7.02). Student’s t test was performed for statistical comparisons between two groups.

11 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Footnotes Funding: This study was supported by the USPHS grant RO1 CA219180 and RO1 CA142604 awarded by the National Cancer Institute. This research used the Flow Cytometry Facility and the Tissue and Research Pathology Facility supported in part by Cancer Center Support Grant from the National Cancer Institute (P30 CA047904). 1Both authors contributed equally to this work.

3Abbreviations: CDK, cyclin-dependent kinase; CDC25C, cell division cycle 25C; DACH1, Dachshund homolog 1; EMT, epithelial-mesenchymal transition; ER, estrogen receptor; EV, empty vector transfected control; GO, gene ontology; HER2, human epidermal growth factor receptor 2; IL, interleukin; KEGG, Kyoto Encyclopedia of Genes and Genomes; PBS, phosphate-buffered saline; PR, progesterone receptor; qRT-PCR, quantitative real-time polymerase chain reaction; TCGA, The Cancer Genome Atlas; VEGF, vascular endothelial growth factor.

Conflict of Interest: None of the authors has any conflict of interest.

12 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

References

1. Siegel, R.L., Miller, K.D., and Jemal, A. (2019) Cancer statistics, 2019. CA Cancer J Clin 69, 7-34. 2. Perou, C.M., Sørlie, T., Eisen, M.B., van de Rijn, M., Jeffrey, S.S., Rees, C.A., Pollack, J.R., Ross, D.T., Johnsen, H., Akslen, L.A., Fluge, O., Pergamenschikov, A., Williams, C., Zhu, S.X., Lønning, P.E., Børresen-Dale, A.L., Brown, P.O., and Botstein, D. (2000) Molecular portraits of human breast tumours. Nature 406, 747-752. 3. Sørlie, T., Perou, C.M., Tibshirani, R., Aas, T., Geisler, S., Johnsen, H., Hastie, T., Eisen, M.B., van de Rijn, M., Jeffrey, S.S., Thorsen, T., Quist, H., Matese, J.C., Brown, P.O., Botstein, D., Lønning, P.E., and Børresen-Dale, A.L. (2001) Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci USA 98, 10869-10874. 4. Vargo-Gogola, T., and Rosen, J.M. (2007) Modelling breast cancer: one size does not fit all. Nat Rev Cancer 7, 659-672. 5. Sørlie, T. (2007) Molecular classification of breast tumors: toward improved diagnostics and treatments. Methods Mol Biol 360, 91-114. 6. Saha, P., and Nanda, R. (2016) Concepts and targets in triple-negative breast cancer: recent results and clinical implications. Ther Adv Med Oncol 8, 351-359. 7. Zhang, H., Meng, F., Liu, G., Zhang, B., Zhu, J., Wu, F., Ethier, S.P., Miller, F., and Wu G. (2011) Forkhead transcription factor foxq1 promotes epithelial-mesenchymal transition and breast cancer metastasis. Cancer Res 71, 1292-1301. 8. Qiao, Y., Jiang, X., Lee, S.T., Karuturi, R.K., Hooi, S.C., and Yu, Q. (2011) FOXQ1 regulates epithelial-mesenchymal transition in human cancers. Cancer Res 71, 3076-3086. 9. Meng, F., Speyer, C.L., Zhang, B., Zhao, Y., Chen, W., Gorski, D.H., Miller, F.R., and Wu G. (2015) PDGFRα and β play critical roles in mediating Foxq1-driven breast cancer stemness and chemoresistance. Cancer Res 75, 584-593. 10. Kim, S.H., Kaschula, C.H., Priedigkeit, N., Lee, A.V., and Singh, S.V. (2016) Forkhead Box Q1 Is a Novel Target of Breast Cancer Stem Cell Inhibition by Diallyl Trisulfide. J Biol Chem 291, 13495-13508. 11. Hardy, K., Wu, F., Tu, W., Zafar, A., Boulding, T., McCuaig, R., Sutton, C.R., Theodoratos, A., and Rao, S. (2016) Identification of chromatin accessibility domains in human breast cancer stem cells. Nucleus 7, 50-67. 12. Carlsson, P., and Mahlapuu, M. (2002) Forkhead transcription factors: key players in development and metabolism. Dev Biol 250, 1-23. 13. Hoggatt, A.M., Kriegel, A.M., Smith, A.F., and Herring, B.P. (2000) Hepatocyte nuclear factor-3 homologue 1 (HFH-1) represses transcription of smooth muscle-specific genes. J Biol Chem 275, 31162-311670. 14. Martinez-Ceballos, E., Chambon, P., and Gudas, L.J. (2005) Differences in gene expression between wild type and Hoxa1 knockout embryonic stem cells after retinoic acid treatment or leukemia inhibitory factor (LIF) removal. J Biol Chem 280, 16484-16498. 15. Potter, C.S., Peterson, R.L., Barth, J.L., Pruett, N.D., Jacobs, D.F., Kern, M.J., Argraves, W.S., Sundberg, J.P., and Awgulewitsch, A. (2006) Evidence that the satin hair mutant gene

13 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Foxq1 is among multiple and functionally diverse regulatory targets for Hoxc13 during hair follicle differentiation. J Biol Chem 281, 29245-29255. Epub 2006 Jul 10. 16. Katoh, M., and Katoh, M. (2004) Human FOX gene family (Review). Int J Oncol 25, 1495- 1500. Review. 17. Li, Y., Zhang, Y., Yao, Z., Li, S., Yin, Z., and Xu, M. (2016) Forkhead box Q1: A key player in the pathogenesis of tumors (Review). Int J Oncol 49, 51-58. 18. Abba, M., Patil, N., Rasheed, K., Nelson, L.D., Mudduluru, G., Leupold, J.H., and Allgayer, H. (2013) Unraveling the role of FOXQ1 in colorectal cancer metastasis. Mol Cancer Res 11, 1017-1028. 19. Bieller, A., Pasche, B., Frank, S., Gläser, B., Kunz, J., Witt, K., and Zoll, B. (2001) Isolation and characterization of the human forkhead gene FOXQ1. DNA Cell Biol 20, 555-561. 20. Liang, S.H., Yan, X.Z., Wang, B.L., Jin, H.F., Yao, L.P., Li, Y.N., Chen, M., Nie, Y.Z., Wang, X., Guo, X.G., Wu, K.C., Ding, J., and Fan, D.M. (2013) Increased expression of FOXQ1 is a prognostic marker for patients with gastric cancer. Tumour Biol 34, 2605- 2609. 21. Zhang, J., Li, W., Dai, S., Tai, X., Jia, J., and Guo, X. (2015) FOXQ1 is overexpressed in laryngeal carcinoma and affects cell growth, cell cycle progression and cell invasion. Oncol Lett 10, 2499-2504. 22. Wang, W., He, S., Ji, J., Huang, J., Zhang, S., and Zhang, Y. (2013) The prognostic significance of FOXQ1 oncogene overexpression in human hepatocellular carcinoma. Pathol Res Pract 209, 353-358. 23. Cao, D., Hustinx, S.R., Sui, G., Bala, P., Sato, N., Martin, S., Maitra, A., Murphy, K.M., Cameron, J.L., Yeo, C.J., Kern, S.E., Goggins, M., Pandey, A., and Hruban, R.H. (2004) Identification of novel highly expressed genes in pancreatic ductal adenocarcinomas through a bioinformatics analysis of expressed sequence tags. Cancer Biol Ther 3, 1081-1089. 24. Kaneda, H., Arao, T., Tanaka, K., Tamura, D., Aomatsu, K., Kudo, K., Sakai, K., De Velasco, M.A., Matsumoto, K., Fujita, Y., Yamada, Y., Tsurutani, J., Okamoto, I., Nakagawa, K., and Nishio, K. (2010) FOXQ1 is overexpressed in colorectal cancer and enhances tumorigenicity and tumor growth. Cancer Res 70, 2053-2063. 25. Mardon, G., Solomon, N.M., and Rubin, G.M. (1994) dachshund encodes a nuclear protein required for normal eye and leg development in Drosophila. Development 120, 3473-3486. 26. Wu, K., Li, A., Rao, M., Liu, M., Dailey, V., Yang, Y., Di Vizio, D., Wang, C., Lisanti, M.P., Sauter, G., Russell, R.G., Cvekl, A., and Pestell, R.G. (2006) DACH1 is a cell fate determination factor that inhibits cyclin D1 and breast tumor growth. Mol Cell Biol 26, 7116- 7129. 27. Wu, K., Jiao, X., Li, Z., Katiyar, S., Casimiro, M.C., Yang, W., Zhang, Q., Willmarth, N.E., Chepelev, I., Crosariol, M., Wei, Z., Hu, J., Zhao, K., and Pestell, R.G. (2011) Cell fate determination factor Dachshund reprograms breast cancer stem cell function. J Biol Chem 286, 2132-2142. 28. Zhao, F., Wang, M., Li, S., Bai, X., Bi, H., Liu, Y., Ao, X., Jia, Z., and Wu, H. (2015) DACH1 inhibits SNAI1-mediated epithelial-mesenchymal transition and represses breast carcinoma metastasis. Oncogenesis 4, e143. 29. Wu, K., Chen, K., Wang, C., Jiao, X., Wang, L., Zhou, J., Wang, J., Li, Z., Addya, S., Sorensen, P.H., Lisanti, M.P., Quong, A., Ertel, A., and Pestell, R.G. (2014) Cell fate factor

14 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

DACH1 represses YB-1-mediated oncogenic transcription and translation. Cancer Res 74, 829-839. 30. Wu, K., Katiyar, S., Li, A., Liu, M., Ju, X., Popov, V.M., Jiao, X., Lisanti, M.P., Casola, A., and Pestell, R.G. (2008) Dachshund inhibits oncogene-induced breast cancer cellular migration and invasion through suppression of interleukin-8. Proc Natl Acad Sci USA 105, 6924-6929. 31. Feng, J., Xu, L., Ni, S., Gu, J., Zhu, H., Wang, H., Zhang, S., Zhang, W., and Huang, J. (2014) Involvement of FoxQ1 in NSCLC through regulating EMT and increasing chemosensitivity. Oncotarget 5, 9689-9702. 32. Singer, C.F., Kronsteiner, N., Hudelist, G., Marton, E., Walter, I., Kubista, M., Czerwenka, K., Schreiber, M., Seifert, M., Kubista, E. (2003) Interleukin 1 system and sex steroid receptor expression in human breast cancer: interleukin 1α protein secretion is correlated with malignant phenotype. Clin Cancer Res 9, 4877-4883. 33. Nozaki, S., Sledge, G.W. Jr., and Nakshatri, H. (2000) Cancer cell-derived interleukin 1alpha contributes to autocrine and paracrine induction of pro-metastatic genes in breast cancer. Biochem Biophys Res Commun 275, 60-62. 34. Kurtzman, S.H., Anderson, K.H., Wang, Y., Miller, L.J., Renna, M., Stankus, M., Lindquist, R.R., Barrows, G., and Kreutzer, D.L. (1999) Cytokines in human breast cancer: IL-1α and IL-1β expression. Oncol Rep 6, 65-70. 35. Singh, J.K., Simões, B.M., Howell, S.J., Farnie, G., and Clarke, R.B. (2013) Recent advances reveal IL-8 signaling as a potential key to targeting breast cancer stem cells. Breast Cancer Res 15, 210. 36. Todorović-Raković, N., and Milovanović, J. (2013) Interleukin-8 in breast cancer progression. J Interferon Cytokine Res 33, 563-570. 37. Freund, A., Chauveau, C., Brouillet, J.P., Lucas, A., Lacroix, M., Licznar, A., Vignon, F., and Lazennec, G. (2003) IL-8 expression and its possible relationship with estrogen-receptor- negative status of breast cancer cells. Oncogene 22, 256-265. 38. Lin, Y., Huang, R., Chen, L., Li, S., Shi, Q., Jordan, C., and Huang, R.P. (2004) Identification of interleukin-8 as estrogen receptor-regulated factor involved in breast cancer invasion and angiogenesis by protein arrays. Int J Cancer 109, 507-515. 39. Yao, C., Lin, Y., Chua, M.S., Ye, C.S., Bi, J., Li, W., Zhu, Y.F., and Wang, S.M. (2007) Interleukin-8 modulates growth and invasiveness of estrogen receptor-negative breast cancer cells. Int J Cancer 121, 1949-1957. 40. Clark-Lewis, I., Schumacher, C., Baggiolini, M., Moser, B. (1991) Structure-activity relationships of interleukin-8 determined using chemically synthesized analogs. Critical role of NH2-terminal residues and evidence for uncoupling of neutrophil chemotaxis, exocytosis, and receptor binding activities. J Biol Chem 266, 23128-23134. 41. Fernando, R.I., Castillo, M.D., Litzinger, M., Hamilton, D.H., and Palena, C. (2011) IL-8 signaling plays a critical role in the epithelial-mesenchymal transition of human carcinoma cells. Cancer Res 71, 5296-5306. 42. Bielenberg, D.R, and Zetter, B.R. (2015) The contribution of angiogenesis to the process of metastasis. Cancer J 21, 267-273.

15 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

43. Folkman, J. (2002) Role of angiogenesis in tumor growth and metastasis. Semin Oncol 29 (suppl. 16), 15-18. 44. Ferrara, N (2002). VEGF and the quest for tumour angiogenesis factors. Nat Rev Cancer 2, 795-803. 45. Hicklin, D.J., and Ellis, L.M. (2005) Role of the vascular endothelial growth factor pathway in tumor growth and angiogenesis. J Clin Oncol 23, 1011-1027. 46. Kim, S.H., Hahm, E.R., Singh, K.B., Shiva, S., Stewart-Ornstein, J., and Singh, S.V. (2020) RNA-seq reveals novel mechanistic targets of withaferin A in prostate cancer cells. Carcinogenesis pii: bgaa009. doi: 10.1093/carcin/bgaa009. 47. Xiao, D., Srivastava, S.K., Lew, K.L., Zeng, Y., Hershberger, P., Johnson, C.S., Trump, D.L., and Singh, S.V. (2003) Allyl isothiocyanate, a constituent of cruciferous vegetables, inhibits proliferation of human prostate cancer cells by causing G2/M arrest and inducing apoptosis. Carcinogenesis 24, 891-897. 48. Livak, K.J., and Schmittgen, T.D. (2001) Analysis of relative gene expression data using -ΔΔC real-time quantitative PCR and the 2 T Method. Methods 25, 402-408.

16 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure legends

Figure 1. FoxQ1 expression was significantly higher in luminal-type human breast cancer specimens compared to normal human mammary tissues. A, expression of FoxQ1 in mammary tumors of black and white women, and in ductal and lobular bacrcinomas. B, representative immunohistochemical images (200× magnification; scale bar = 100 μm) for FoxQ1 expression in normal human mammary tissues and luminal-type human breast cancer tissues. C, quantification of FoxQ1 expression in normal human mammary tissues and luminal- type human breast cancer tissues. Results shown are mean ± S.D. *P < 0.05 by two-sided Student's t-test. D, volcano plot for differential gene expression. E, heatmap of the differentially expressed genes in empty vector transfected cells (EV) and FoxQ1 overexpressing SUM159 cells. F, venn diagram showing unique and overlapping genes between EV and FoxQ1 overexpressing SUM159 cells.

Figure 2. Functional analysis of the differential expressed genes by FoxQ1 overexpression in SUM159 cells. A-B, scatter plots for Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses showing upregulation (A) or downregulation (B) by FoxQ1 overexpression. C- D, scatter plots for gene ontology (GO) enrichment analyses showing upregulation (C) or downregulation (D) by FoxQ1 overexpression. E-F, scatter plots for Reactome pathway analyses showing upregulation (E) or downregulation (F) by FoxQ1 overexpression. padj, adjusted p- value.

Figure 3. Consistent data from RNA-Seq analysis (present study) and published literature (7,9,10) on known targets of FoxQ1. A, RNA-seq analysis of genes associated with epithelial- mesenchymal transition (7). CDH1, E-cadherin; CTNNA1, catenin alpha 1; CDH2, N-cadherin; FN1, fibronectin 1. B, RNA-seq analysis of genes involved in breast cancer stemness and chemoresistance (9). CST6, cystatin 6; SEMA3A, semaphorin 3A; PDGFRA, platelet derived growth factor receptor alpha; JAM3, junctional adhesion molecule 3; ADAM9, ADAM metallopeptidase domain 9; THBS1, thrombospondin 1; ZEB2, zinc finger E-box binding homeobox 2; FOXA1, forkhead box A1; EDN1, endothelin 1. C, RNA-seq analysis of genes related to breast cancer stemness (10). DACH1, dachshund homolog 1; ZEB1, zinc finger E-box binding homeobox 1; TWIST2, twist basic helix-loop-helix transcription factor 2. Results shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test.

Figure 4. Inconsistent data between RNA-Seq analysis (present study) and published literature (7-9,11). A-D, RNA-seq analysis of genes not consistent with the published literatures (7,9,11). Results shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test. PLD1, phospholipase D1; SNAI1, snail family transcriptional repressor 1; CTNNB1, catenin beta 1; S100A4, S100 calcium binding protein A4; DCN, Decorin; PDGFRB, platelet derived growth factor receptor beta; CADM3, cell adhesion molecule 3; COL1A1, collagen type I alpha 1; COL6A1, collagen Type VI alpha 1; DSG2, desmoglein 2; TWIST1, twist basic helix-loop-helix transcription factor 1.

Figure 5. FoxQ1 overexpression resulted in cell cycle arrest at S and/or G2/M phases in SUM159 cells, but not in MCF-7 cells. A, representative flow histograms showing serum-

17 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

starved synchronization at G0/G1 phase and resumption of cell cycle after release of cells in complete medium. B, percentage of G0/G1, S, and G2/M population. Results shown are mean ± SD (n = 3). *P < 0.05 by two-sided Student's t test. Experiments were done at least three times with similar results. Figure 6. FoxQ1 overexpression resulted in mitotic arrest. A, Western blot analysis for phospho-(Ser10) histone H3 protein using lysates from (EV) cells or FoxQ1 overexpressing SUM159, MCF-7, and HMLE cells. The numbers on top of the bands represent changes in protein level compared to corresponding EV cells. B, representative flow histograms showing mitotic fraction in EV cells or FoxQ1 overexpressing SUM159, MCF-7, and HMLE cells. C, quantitation of mitotic fraction. Data shown are mean ± SD (n = 3-6). *P < 0.05 by two-sided Student's t-test. Experiments were done twice with comparable results. Figure 7. Overexpression of FoxQ1 modulated expression of proteins associated with cell cycle regulation. Representative immunoblots for CDK2, CDC25C, Cyclin D1, Cyclin A, CDK4, Cyclin B1, and CDK1 using lysate from EV or FoxQ1 overexpressing SUM159 or MCF- 7 cells. Numbers above the bands are relative expression level of each protein compared to EV cells. The -Actin blot is the same for some proteins. Experiment were done at least twice with comparable results. CDK2, cyclin dependent kinase 2; CDC25C, cell division cycle 25C; CDK4, cyclin dependent kinase 4; CDK1, cyclin dependent kinase 1. Figure 8. FoxQ1 is a novel regulator of IL-1α. A, RNA-seq analysis of IL-1α gene expression in EV cells FoxQ1 overexpressing SUM159 cells. Data shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test. B, positive correlation between FoxQ1 and IL-1α expression in breast tumor TCGA dataset (n = 1097). Pearson's test was used to determine statistical significance of the correlation. C, quantification of IL-1α mRNA expression in EV and FoxQ1 overexpressing SUM159 cells. Data shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test. D, quantification of IL-1α secretion in the media of EV or FoxQ1 overexpressing SUM159 cells. Data shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test. E, identification of a putative FoxQ1 binding site in the promoter region of IL-1α. The bar graph shows recruitment of FoxQ1 at the IL-1α promoter by ChIP analysis. The results shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test. Experiments were done twice with similar results. Figure 9. IL-8 is a novel downstream target of FoxQ1 in SUM159 cells. A, RNA-seq analysis for IL-8 gene expression in EV cells and FoxQ1 overexpressing SUM159 cells. Data shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test. B, positive correlation between FoxQ1 and IL-8 expression in breast tumor TCGA dataset (n = 1097). Pearson's test was used to determine statistical significance of the correlation. C, quantification of IL-8 mRNA expression by qRT-PCR in EV and FoxQ1 overexpressing SUM159 cells. Data shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test. D, quantification of IL-8 secretion in the media of EV and FoxQ1 overexpressing SUM159 cells. Data shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test. E, identification of putative FoxQ1 binding sites at the promoter region of IL-8. The bar graph shows recruitment of FoxQ1 at the IL-8 promoter by ChIP analysis. The results shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test. Experiments were repeated with similar results.

18 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 10. FoxQ1 regulated IL-8 expression in MCF-7 cells. A, quantification of IL-8 secretion in the media of EV and FoxQ1 overexpressing MCF-7 cells. Data shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t test. B, ChIP assay for FoxQ1 recruitment at the promoter region of IL-8. The results shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t-test. Experiments were done twice with similar results. Figure 11. FoxQ1 regulates VEGFA expression in SUM159 cells. A, RNA-seq analysis of VEGFA gene expression difference between EV and FoxQ1 overexpressing SUM159 cells. Data shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t test. B, TCGA dataset showed the correlation between FoxQ1 and VEGFA in breast tumor (n = 1097). Pearson's correlation coefficient was used for the analysis. C, quantification of VEGFA mRNA expression by quantitative real-time PCR in EV or FoxQ1 overexpressing SUM159 and MCF-7 cells. Data shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t test. Experiments were done twice with comparable results. D, quantification of VEGFA secretion in EV or FoxQ1 overexpressing SUM159 and MCF-7 cells after 16 hours of serum-starvation. Data shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t test. Experiments were done twice with comparable results. E, identification of putative FoxQ1 binding sites in the promoter region of VEGFA. The bar graph shows recruitment of FoxQ1 in the VEGFA promoter by ChIP analysis. The results shown are mean ± S.D. (n = 3). *P < 0.05 by two-sided Student's t test. Experiments were done twice with similar results.

19 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 1

A D

] Breast cancer TCGA • Upregulated genes (n=3247) )

1 12 • Downregulated genes (n=3391)

+ t n 300

n *

c _

e 8

z x

) i

l 200

1 m

padj r

4 (

x n

(

F g Log

o 100 -

L 0 [ Black White

(n=150) (n=598)

] )

1 12 1.3

n n

o threshold i

u -4 0 4

o s

c Log2(Fold change) _

e 8

d p

e E

1 m

r 4

(

g o

L 0 [ Ductal Lobular (n=784) (n=203)

B Normal Luminal FoxQ1 expression FoxQ1

C EV FoxQ1

n 90 o

i * F

s FoxQ1 EV

)

e r

r 60

- 1

H 30

( Q

x 627 11263 467 o

F 0 Normal Luminal (n=60) (n=148) bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 2

A Upregulation B Downregulation KEGG enrichment KEGG enrichment

0.02 0.04 0.06 0.08 0.10 0.01 0.02 0.03 0.04 GeneRatio GeneRatio C Upregulation D Downregulation GO enrichment GO enrichment

positive regulation of cellular component movement

0.02 0.03 0.04 0.02 0.03 0.04 GeneRatio GeneRatio E Upregulation F Downregulation Reactome enrichment Reactome enrichment

0.01 0.02 0.03 0.02 0.03 0.04 0.05 0.06 0.07 GeneRatio GeneRatio bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 3

A )

0.020 n 120 40 200

)

r i

M *

o s

* e

i h

s *

P e

0.015 30 s 150

)

e a

6 80

(

0 x

0.010 o 20 100

(

( ( (

A s

2 r

0.005 10 N 50

e C 0.000 0 C 0 0 EV FoxQ1 EV FoxQ1 EV FoxQ1 EV FoxQ1

B n

1.0 n 2.0 15 15

n i

n *

s o

o *

i s

0.8 s

s e

e 1.5

)

) e e 10

p 10

M x

0.6 x

e x

x 1.0

P e

e *

F F

0.4 F

( ( (

(

3 6

R 5 5

T F

0.5 M

M G

0.2 A

D S 0.0 * 0.0 P 0 0 EV FoxQ1 EV FoxQ1 EV FoxQ1 EV FoxQ1

40 600 15 1.5

s s

30 s *

)

r r

* 400 e 10 1.0

e e

20 e

P P

P *

( (

(

( 2

5 A 0.5

S 200

X B

A 10

T F A 0 0 * 0 0.0 EV FoxQ1 EV FoxQ1 EV FoxQ1 EV FoxQ1

o i

s 30

)

p K

x 20 P

e *

( 1

N 10

D E 0 EV FoxQ1 C

0.4 15 250 25

o i n *

i *

i s

i 200 20

s s s *

0.3 s

)

r r

e 10

M p

r 150 15

x e

0.2 x

F F

F 100 10

( ( ( (

1 H

5 C

Y I C 0.1

* E 50 5

W D 0.0 0 0 T 0 EV FoxQ1 EV FoxQ1 EV FoxQ1 EV FoxQ1 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 4

1.5 4

s s

s 3

)

) e

e 1.0

* r

K x

x 2 *

( ( 1

0.5 I D

A 1

P S 0.0 0 EV FoxQ1 EV FoxQ1

B C 5

n 200 60 160

)

i n

M *

* n

P e

150 P 120

)

t F

40 e

(

- x

100 o 80

P P

 i

(

s 4

N r

50 P 40

J e C 0 0 0 EV FoxQ1 EV FoxQ1 EV FoxQ1 D

40 6 n 4 8

s s

30 e 3 6

)

r r

4 p

e *

e e

20 x 2 4

( (

(

( R

A 2

M F 0 *

10 C

1 D 2

D S * C 0 0 P 0 0 * EV FoxQ1 EV FoxQ1 EV FoxQ1 EV FoxQ1

200 8 50 15

s s

40 s

150 6 s *

)

r r

e 10

M M

M 30

e x

100 4 e *

F F

F 20

( (

A T

A 5

G S

50 2 I

L L

* S 10

O W

* D

T C 0 C 0 0 0 EV FoxQ1 EV FoxQ1 EV FoxQ1 EV FoxQ1 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 5 A SUM159 Control Serum starvation 2 h 4 h 6 h 12 h EV

2N 4N 2N 4N 2N 4N 2N 4N 2N 4N 2N 4N FoxQ1

2N 4N 2N 4N 2N 4N 2N 4N 2N 4N 2N 4N B

SUM159 MCF-7 )

80 EV FoxQ1 ) 80 EV FoxQ1

(

e s

60 s 60 a

* a

h p

* p

* 1 * 1

G 40 40

n i

20 i 20

e C 0 C 0 2 h 4 h 6 h 12 h 2 h 4 h 6 h 12 h

30 EV FoxQ1 40 EV FoxQ1 )

* )

(

e 30

e s

20 s

h p

* p *

* 20 *

n i

10 i

s l

l 10

C C 0 0 2 h 4 h 6 h 12 h 2 h 4 h 6 h 12 h

30 EV FoxQ1 ) 20 EV FoxQ1

( ,

e s

s 15

a h

20 * h *

p p

* *

M M

/ 10

e C 0 C 0 2 h 4 h 6 h 12 h 2 h 4 h 6 h 12 h bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 6

A SUM159 MCF-7 HMLE 1 5.8 kDa 1 <0.1 kDa 1 >10 kDa Phospho-(Ser10) 17 17 17 histone H3 12 12 12 52 52 52 β-Actin 38 38 38 1 2 1 2 1 2 1:EV, 2:FoxQ1 1:EV, 2:FoxQ1 1:EV, 2:FoxQ1

B EV FoxQ1 C 2

) SUM159

% (

l e

c 1

t i H3 M 0 EV FoxQ1 EV FoxQ1

histone 2

) MCF-7

(

s l

l * e Ser10) Ser10) (

c 1

t i

M 0 Phospho EV FoxQ1 EV FoxQ1 3

) HMLE

% (

s l

l 2

c i

t 1

t i

M 0 Propidium iodide (DNA content) EV FoxQ1 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 7

SUM159 MCF-7 1 0.8 kDa 1 0.6 kDa 38 38 CDK2 CDK2 31 31 β-Actin 52 β-Actin 52 38 38 1 0.2 1 2.4 76 76 CDC25C 52 CDC25C 52 52 β-Actin 38 1 2.7 1 0.5 Cyclin D1 38 Cyclin D1 38 31 31 β-Actin 52 38 1 2.1 1 0.9 76 76 Cyclin A 52 Cyclin A 52 52 β-Actin 52 β-Actin 38 38 1 1.7 1 1.7 38 38 CDK4 CDK4 31 31 52 β-Actin 38 1 0.1 1 7.4 76 76 Cyclin B1 52 Cyclin B1 52 52 52 β-Actin β-Actin 38 38 1 0.1 1 1.0 38 38 CDK1 31 CDK1 31 52 52 β-Actin β-Actin 38 38 1 2 1 2 1:EV, 2:FoxQ1 1:EV, 2:FoxQ1 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 8

A RNA-seq B TCGA-BRCA (n=1097)

30 ]

) 12 r = 0.2779, P < 0.0001

+ )

* t

c P

_ 8

F ( 20 d



m r

e 4

( x

10 2

 o

L [

1 0 L I 0 4 8 12 0 FoxQ1 EV FoxQ1 [Log2(normalized_count+1)] C D

25 15 n

SUM159 ) SUM159

i s

* t

e s

20 t n

e *

r r

p 10

15 h

g N

10 e

p R

( 5

m o  5 f

 1

I L I 0 0 EV FoxQ1 EV FoxQ1 E 3′ IL1α -1640 TGTTTA -1635 IL1α 5′ promoter 15

t SUM159

e m

h *

c 10

l o

f 5



L I 0 IgG FoxQ1 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 9

A B RNA-seq TCGA-BRCA (n=1097)

600 ] ) 15 r = 0.2812, P < 0.0001

* 1

)

c P

_ 10

F (

400 d

m s

r 5

( p

200 2

8 [

- 0 L I 0 4 8 12 0 FoxQ1 EV FoxQ1 [Log2(normalized_count+1)] C 8 D 2.0

n SUM159 SUM159

)

n i

s *

s t

e 6 1.5 *

e g

4 1.0

(

m 2 0.5

I L I 0 0.0 EV FoxQ1 EV FoxQ1 E 3′ Site #1 -1758 TGTTTG -1753 Site #3 IL-8 5′ IL-8 promoter -1937 ATGTTT -1932 Site #2 -600 TGTTTG -595

10 SUM159 15 SUM159

t t n

n * e

e 8 Site #1 Site #3 m

m * h

h 10

c c

i i r

r 6

n n

e e

d 4

l l o

o 5

f f

8 8 -

- 2

L L

I I 0 0 IgG FoxQ1 IgG FoxQ1 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 10 A

200 MCF-7

)

i e

t 150

o r

p *

g 100

p (

L I 0 EV FoxQ1

B 3′ Site #1 Site #3 IL-8 5′ -1758 TGTTTG -1753 IL-8 promoter -1937 ATGTTT -1932 Site #2 -600 TGTTTG -595

4 MCF-7 4 MCF-7

t t n

n * * e

e Site #1 Site #3 m

m 3 3

h h

c c

i i

r r n

n 2 2

e e

d d

l l

o o f

f 1 1

8 8

- -

L L

Figure 11

A B

20 RNA-seq TCGA-BRCA (n=1097)

) ]

M * 20

) 1

K r= 0.0672, P=0.026

+ P

15 t

( u

z i

10 G 10

n 5

( A

5 2

L [ E 0

V 0 5 10 15 0 EV FoxQ1 FoxQ1 [Log2(normalized_count+1)]

C D n

6 ) 15

r r

p *

x 4 * 10

R (

2 5

A *

G V E 0 0 V EV FoxQ1 EV FoxQ1 EV FoxQ1 EV FoxQ1 SUM159 MCF-7 SUM159 MCF-7 E 3′ Site #1 -2179 TGTTTA -2174 Site #3 VEGFA 5′ VEGFA

promoter -2929 AGTTTG -2924 Site #2 -680 TGTTTG -675

t t

12 t 15 15

n n

SUM159 n SUM159 SUM159

m m

Site #1 m Site #2 Site #3 *

h h

9 h

r r

r 10 10

n n

* n *

f f

f 5 5

A A

3 A

V V 0 V 0 0 IgG FoxQ1 IgG FoxQ1 IgG FoxQ1 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Table 1. The summary of mapping result

Sample name EV-SUM159 FoxQ1-SUM159 Mean SD Mean SD Total reads 65198263 4386610 72027681 10214885 Total mapped 58914439 3655237 64895915 8672753 (90.38%) (0.51%) (90.17%) (0.82%) Multiple mapped 1147142 81213 1224173 154648 (1.76%) (0.01%) (1.70%)* (0.03%)* Uniquely mapped 57767297 3574643 63671741 8518408 (88.62%) (0.52%) (88.46%) (0.80%) Reads map to '+' 28883648 1787322 31835871 4259204 (44.31%) (0.26%) (44.23%) (0.40%) Reads map to '-' 28883648 1787322 31835871 4259204 (44.31%) (0.26%) (44.23%) (0.40%) Non-splice reads 32843918 2036952 36497716 5119388 (50.39%) (0.30%) (50.68%) (0.56%) Splice reads 24923378 1546835 27174025 3464790 (38.24%) (0.33%) (37.79%) (1.14%)

*, Statistically significant analyzed by unpaired t test (P < 0.05).

Table 2. Correlation between FoxQ1 and genes associated with “Cellular response to stress” category from TCGA-BrCa (n=1097) and RNA-seq results. Student t test was used for statistical analysis. TCGA-BrCa analysis RNA-Seq data Correlation Genes fold change Pearson r P value P value (FoxQ1 vs. EV) Positive TUBB6 0.4657 < 0.0001 + 1.59 0.0032 HMGA2 0.3699 < 0.0001 + 1.56 0.0541 CDK6 0.3684 < 0.0001 + 1.29 0.0978 MAP4K4 0.3093 < 0.0001 + 1.65 0.0001 SOD2 0.2869 < 0.0001 + 2.18 < 0.0001 IL8 (=CXCL8) 0.2812 < 0.0001 + 2.42 0.0001 IL1A 0.2779 < 0.0001 + 23.04 0.0002 ETS2 0.2657 < 0.0001 + 1.53 0.0047 C11orf73 0.2568 < 0.0001 + 1.37 0.0512 (=HIKESHI) PHC2 0.2453 < 0.0001 + 1.49 0.0001 IGFBP7 0.2410 < 0.0001 + 4.04 0.0015 ETS1 0.2288 < 0.0001 + 1.23 0.0341 HMGA1 0.2181 < 0.0001 + 1.22 0.0233 JUN 0.1691 < 0.0001 + 1.14 0.0265 TFDP2 0.1550 < 0.0001 + 1.74 0.0007 AR -0.2917 < 0.0001  9.71 < 0.0001 MAP2K4 -0.1899 < 0.0001  1.56 0.0003 BMI1 -0.1888 < 0.0001  1.24 0.0048 DYNLL2 -0.1810 < 0.0001  1.32 0.0017 HSPA1L -0.1789 < 0.0001  3.45 0.0019 ATP7A -0.1738 < 0.0001  1.24 0.0500 CBX4 -0.1715 < 0.0001  2.02 < 0.0001 DYNC1LI1 -0.1714 < 0.0001  1.23 0.0025 MAPK3 -0.1645 < 0.0001  1.79 0.0042 UBN1 -0.1638 < 0.0001  1.14 0.0049 GSR -0.1506 < 0.0001  1.39 0.0095 Negative BAG2 0.3692 < 0.0001  1.21 0.0004 CA9 0.3279 < 0.0001  11.51 0.0119 WTIP 0.3160 < 0.0001  2.46 0.0017 CBX2 0.2512 < 0.0001  3.05 < 0.0001 CCNE1 0.2497 < 0.0001  1.31 0.0304 PSMB9 0.2361 < 0.0001  1.34 0.0600 RPS6KA3 0.2278 < 0.0001  1.77 0.0004 PSMB2 0.2273 < 0.0001  1.25 0.0431 EGLN3 0.2229 < 0.0001  2.25 0.0010 MAPK7 0.2111 < 0.0001  1.79 < 0.0001 TUBA1A 0.2091 < 0.0001  1.88 0.0015 E2F3 0.2000 < 0.0001  1.32 0.0109 UBE2D1 0.1940 < 0.0001  1.29 0.0022 PSMB10 0.1877 < 0.0001  1.79 0.0020 DNAJC2 0.1778 < 0.0001  1.27 0.0011 HIF1A 0.1683 < 0.0001  1.38 0.0018 PSMC1 0.1678 < 0.0001  1.49 0.0086 IL6 0.1650 < 0.0001  2.55 < 0.0001 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.26.117176; this version posted May 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

TUBB2A 0.1633 < 0.0001  1.39 0.0089 UBE2C 0.1511 < 0.0001  1.32 0.0047 ERO1L (=ERO1A) 0.1509 < 0.0001  1.21 0.0711

DCTN4 -0.2692 < 0.0001 + 1.15 0.0035 MDM2 -0.2195 < 0.0001 + 1.26 0.0478 HIST4H4 -0.2000 < 0.0001 + 3.34 0.0083 H2AFJ -0.1580 < 0.0001 + 2.30 0.0119 NUP210 -0.1561 < 0.0001 + 14.72 < 0.0001

H2AFV -0.1501 < 0.0001 + 1.16 0.0022