FOXM1c promotes oesophageal cancer metastasis by transcriptionally regulating IRF1 expression

Objectives We aimed to elucidate the role and molecular mechanisms of FOXM1 in regulating metastasis in oesophageal squamous cell carcinoma (ESCC) as well as its clinical implications. Materials and methods The expression levels of four isoforms of FOXM1 were analysed by real‐time PCR. Next, genetically modification using overexpression and RNAi systems and transwell were employed to examine FOXM1c function in invasion and migration. Dual luciferase and ChIP assays were performed to decipher the underlying mechanism for transcriptional regulation. The expression levels of FOXM1 and IRF1 were determined by immunohistochemistry staining in ESCC specimens. Results The FOXM1c was predominantly overexpressed in ESCC cell lines compared to the other FOXM1 isoforms. Ectopic expression of FOXM1c promoted invasion and migration of ESCC cells lines, whereas downregulation of FOXM1c inhibited these processes. Moreover, FOXM1c expression was positively correlated with IRF1 expression in ESCC cell lines and tumour specimens. IRF1 is, at least in part, responsible for FOXM1c‐mediated invasion and migration. Mechanistically, we identified IRF1 as a transcriptional target of FOXM1c and found a FOXM1c‐binding site in the IRF1 promoter region. Furthermore, high expression levels of both FOXM1c and IRF1 were positively associated with low survival rate and predicted a poor prognosis of oesophageal cancer patients. Conclusion FOXM1c promotes the metastasis by transcriptionally targeting IRF1 and may serve as a potential prognostic predictor for oesophageal cancer.


| INTRODUC TI ON
The incidence of oesophageal cancer has rapidly increased in the United States and western countries over the past decades. 1,2 In Asia, oesophageal cancer is also one of the most aggressive cancers, with a high prevalence. 3 As a major subtype of oesophageal cancer, oesophageal squamous cell carcinoma (ESCC) accounts for approximately 90% of oesophageal cancers and has been ranked as the fourth leading cause of cancer-related mortality in China. 4 Oesophageal cancer has a poor prognosis due to early metastasis and direct invasion, the 5-year overall survival rate of which is less than 20%. 5 The majorities of oesophageal cancer patients develop resistance to both chemo-and radiotherapy despite initial response. [6][7][8][9][10] Moreover, patients with resistance to these treatments are frequently diagnosed with metastasis. 11,12 Therefore, elucidation of the mechanisms involved in oesophageal cancer metastasis is urgently needed.
FOXM1 is strongly overexpressed in almost all types of human cancers and is highly associated with cancer progression, including that of oesophageal cancer. 13 In general, there are four distinct isoforms of FOXM1: FOXM1a, FOXM1b, FOXM1c and FOXM1d, due to the alternative splicing of its exons Va and VIIa. 13,14 FOXM1a, which retains both exons Va and VIIa, is predominantly located in the cytoplasm and is transcriptionally inactive, probably due to the disruption of its transactivation domain. 15 Both FOXM1b (without both Va and VIIa) and FOXM1c (with only exon Va), which are predominantly located in the nucleus, play similar but not identical biological roles and have different binding affinities and partners. 13 We recently identified FOXM1d as a novel FOXM1 isoform that has exon VIIa but lacks exon Va and is predominantly located in the cytoplasm; thus, it does not have direct transcription-regulating functions. 14 FOXM1 has widely been recognized as a proliferation-specific oncogenic transcription factor 13 that transcriptionally regulates a number of genes that are involved in the G2-M progression, such as Plk1, AuroraB, Cyclin B1, CDC25B, CENP-A and Survivin. 16,17 In addition, FOXM1 plays an essential role in the regulation of a wide spectrum of biological processes, such as inflammation, metabolism, angiogenesis, apoptosis and metastasis. [18][19][20][21] Overexpression of FOXM1 is highly associated with tumour cell survival, epithelial-to-mesenchymal transition (EMT), chemo-/radio-resistance and metastasis. 18,19 Downregulation of FOXM1 inhibits matrix metalloproteinases (MMPs), including MMP2 and MMP9, and inhibits nasopharyngeal carcinoma metastasis. 22,23 FOXM1 is associated with metastasis in colorectal cancer through induction of EMT. 24  In this study, we investigated the abundance of each FOXM1 isoform in oesophageal cancer cells and found that FOXM1c was the predominant isoform. FOXM1c modulates oesophageal cancer invasion and migration by regulating IRF1 transcription and subsequently MMP2/9 expression. We further observed that both FOXM1c and IRF1 were positively correlated with poor prognosis and low survival rate in oesophageal cancer patients. These findings suggest that FOXM1c and IRF1 may be potential diagnostic biomarkers and drug targets for oesophageal cancer.

| Quantitative real-time polymerase chain reaction (qRT-PCR)
Total RNA was extracted from oesophageal cancer cells using TRIzol reagent (Life Technology). Next, cDNA was obtained from 2 μg of total RNA using a reverse transcription kit (TaKaRa, Tokyo, Japan). qRT-PCR analyses of the expression of the FOXM1a, FOXM1b, FOXM1c, FOXM1d and IRF1 genes were performed on an ABI Prism 7900 System with SYBR Premix Ex Taq II (TaKaRa). The primers were designed against the region that locates exclusively in each isoforms. And the specificity and amplification efficiency were verified previously. 14 All primers are listed in Table S1. The data were analysed using QuantStudio TM Real-Time PCR software, and the relative expression was analysed using the 2 −ΔΔCt method. Three separate experiments were performed.

| Construction of reporter plasmids
A 958 bp sequence from −820 to +138 bp of IRF1 (NM_002198.2) relative to the transcriptional start site was subcloned into the KpnI and XhoI sites of the pGL3-basic vector (Promega), using the following primers: forward-KpnI: 5'-CGGGTACCCGACCTTGAAAACTACTCAGC-3' and reverse-XhoI: vectors were also established based on potential FOXM1 binding sites predicted in the website https://www.genomatix.de/#. All reporter plasmid constructs were verified by sequencing.

| Dual luciferase assay
For analysis of the effect of FOXM1c on regulating IRF1 transcription, we employed dual luciferase reporter assays as described previously. 29 In brief, 293T cells were transiently transfected with the different pGL3-IRF1 plasmids together with FOXM1c-expressing or FOXM1c shRNA plasmid. The IRF1 promoter activity was normalized via co-transfection with a Renilla luciferase reporter gene. The luciferase activity was quantified using a dual luciferase assay kit (Promega) 48 hours after transfection.

| Chromatin immunoprecipitation (ChIP) assay
ChIP assays were used to identify the physical binding of FOXM1c to the IRF1 promoter. Oesophageal cancer cells (2 × 10 6 ) were prepared for ChIP assays using a ChIP assay kit (Merck Millipore, Billerica, MA) according to the manufacturer's protocol. The resulting immunoprecipitated DNA specimens were analysed using three ChIP primers to amplify three regions of the IRF1 promoter; the PCR products were 198, 220 and 194 bp. The ChIP primers are as follows: 5'-GATTTCCCCTGGTCCAGCA-3' (forward) and 5'-GAATCTCCCGACTGGCAGC-3' (reverse). The PCR products were resolved electrophoretically on a 2% agarose gel and visualized using ethidium bromide staining.

| Statistical analysis
Statistical evaluation was conducted with SPSS 22.0 (SPSS Inc., Chicago, IL). The chi-square test was used to analyse the relationship between clinicopathological parameters and the expression of CD59.
The 5-year overall survival (OS) and disease-free survival (DFS) were calculated by the Kaplan-Meier method, and differences in variables were compared using log-rank tests. The significance of the in vitro and in vivo data was determined using Student's t test (two-tailed).
All data are shown as the mean ± SD Experiments were repeated at least three times. P values less than 0.05 were considered significant.

| FOXM1c was the predominant isoform in ESCC cells
Although FOXM1 was highly expressed in oesophageal cancer and correlated with poor prognosis, 25 isoform with specific primers. 14 We observed that the FOXM1c isoform was uniformly expressed at a much higher level than the other three isoforms in all four cell lines ( Figure 1B and Figure  and FOXM1c. 32 Considering that FOXM1a was negligibly expressed in ESCC ( Figure 1B) and in pancreatic and colorectal cancers as described previously, 14 found that the pattern of FOXM1c expression in four ESCC cells was consistent with FOXM1 expressed at similar levels ( Figure 1C).
Therefore, we concluded that FOXM1c was the predominant isoform in ESCC cells.

| Genetic alteration of FOXM1c expression level affected ESCC cell metastasis
Given the evidence that FOXM1 coincides with metastasis of breast cancer, pancreatic cancer and prostate cancer, 28

| IRF1 mediated FOXM1c-induced cell migration and invasion via MMP2/9
To identify the downstream targets that are potentially regulated by FOXM1c and simultaneously contribute to oesophageal cancer metastasis, we next tested 15 previously reported genes that are regulated by FOXM1 and highly associated with cancer metastasis ( Figure   S4). 23,[36][37][38][39][40] To identify which gene is responsible for FOXM1c regula- In addition, based on the GEO public database analysis, we found that IRF1 expression in oesophageal cancer was significantly higher than that in paired normal tissues, indicating a positive correlation between IRF1 and oesophageal cancer progression (GSE23400) ( Figure 3E).
Therefore, we chose IRF1 for further investigation.
Furthermore, we silenced IRF1 with shRNA to identify the subsequent alterations of MMP2/9 and the effects on cell migration and invasion. The results demonstrated that both MMP2 and MMP9 were substantially downregulated in Eca109 and KYSE180 cells ( Figure 3F,G). Accordingly, the cell migration and invasion capacities were strongly suppressed ( Figure 3H,J, the quantitative results shown in Figure 3I,K). Knockdown of IRF1 with specific siRNAs also impaired the wound healing abilities ( Figure S3C,D). Therefore, these findings suggest that FOXM1c promoted oesophageal cancer metastasis, at least in part, by regulating the IRF1-MMP2/9 signalling axis.

| FOXM1c regulated IRF1 transcription
To further reveal the mechanism of IRF1 transcription regulated by FOXM1c, we employed dual luciferase reporter assays by cloning the IRF1 promoter region (−820 bp to +138 bp) into the pGL3 vector to generate the reporter pGL3 plasmid. Next, this plasmid was co-transfected into 293T cells with the pRL inter-control plasmid and different doses of the FOXM1c-expressing plasmid. The results showed that the transcriptional activity represented by the relative luciferase activity was gradually elevated by the FOXM1c plasmid in a dose-dependent manner ( Figure 4A). In contrast, there was a significant decrease in transcriptional activity in FOXM1c knockdown cells compared to control cells ( Figure 4B). These results indicate that Quantitative results in I and K. Data represent the mean ± SD; n = 3; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; and analysis with Student's t test (unpaired, two-tailed) FOXM1c may be involved in IRF1 transcription. We also noted that the transcriptional activity of the IRF1 promoter remained at a high level in FOXM1-insufficient cells, indicating that other transcription factors may also participate in IRF1 transcriptional regulation.
To further determine the FOXM1c binding sites in the IRF1 promoter region, we constructed a series of pGL3 plasmids containing 5' truncations of the IRF1 promoter with different lengths ( Figure 4C).
These plasmids were then co-transfected into 293T cells with the We next performed ChIP assays to further verify the physical binding of FOXM1c to the promoter region identified above. After 293T cells were transiently transfected with the Flag-FOXM1c-expressing plasmid, we found that an anti-Flag antibody but not isotype IgG could effectively capture the binding site of FOXM1c ( Figure 4E,F).
Therefore, we concluded that FOXM1c effectively regulated IRF1 transcription by directly binding to the specific promoter region.

| IRF1 was highly associated with FOXM1c and both were correlated with oesophageal cancer progression
To investigate the relationship between the expression levels of FOXM1c and IRF1 in clinical specimens, we collected 120 oesophageal cancer samples (Table 1), which were classified as high or low stage based on their TNM classifications. The expression levels of FOXM1c and IRF1 were determined by IHC with the specific antibodies. We found that IRF1 expression was highly associated with FOXM1c. The representative images are shown in Figure 5A, in which both FOXM1c and IRF1 were strongly expressed in the samples classified as high stage, while slightly expressed or undetectable in the low-stage samples ( Figure 5A). The statistical analysis showed a significant positive correlation between FOXM1c and IRF1 ( Figure 5B). Moreover, these  (Table 1). Together, these results indicate that both FOXM1c and IRF1 were independent prognostic indicators and might be potential drug targets for oesophageal cancer.

| D ISCUSS I ON
FOXM1 produces four isoforms due to alternative splicing. FOXM1a function has been poorly characterized due to its extremely low expression, FOXM1b and FXOM1c mainly regulate oncogene transcription in the nucleus, and FOXM1d promotes cancer EMT and progression via interactions with ROCKs in the cytoplasm. 13  transcription, while NF-κB, PEA-3 and AP-1 regulate MMP-9 transcription. [43][44][45] Here, we interestingly found that the transcription factor of IRF1 simultaneously regulated both MMP-2 and MMP-9 in oesophageal cancer cells, possibly at the transcriptional level that requires future investigation, thus affecting cancer metastasis.
As a transcription factor, FOXM1 regulates many genes that are involved in different stages of cancer, including initiation, progression and metastasis. 13,21,32 We screened a series of genes that are important for tumour metastasis and highly associated with FOXM1 expression. IRF1 was uniformly identified to be significantly downregulated by FOXM1c insufficiency in three oesophageal cancer cell lines. IRF1 is a transcription factor that regulates a number of IFN-inducible genes in response to viral infection or interferon stimulation. 46 The role of IRF1 in cancer progression is controversial depending on cancer types. 47,48 Our current study demonstrated that IRF1 was transcriptionally regulated by FOXM1c and was an important regulator for the oesophageal cancer cell invasion and migration via MMP2/9. We also found that there was a high correlation between FOXM1c and IRF1 in 120 oesophageal cancer specimens; more importantly, both FOXM1c and IRF1 were co-overexpressed in oesophageal cancer patients in the advanced stage and with poor prognosis. Therefore, FOXM1c and IRF1 may be potential independent biomarkers for prediction of oesophageal cancer prognosis.
In summary, we determined that FOXM1c was the predominant isoform among the four isoforms of FOXM1 in oesophageal cancer.
Moreover, we unveiled a novel mechanism of FOXM1c in regulating cancer invasion and migration, that is, FOXM1c transcriptionally regulated IRF1 by directly binding to its promoter region, and IRF1 further modulated MMP2/9 expression. Using clinical samples, we further demonstrated the close correlation between FOXM1c and IRF1 expression levels, and their expression levels were highly associated with oesophageal prognosis. These findings suggest the potential of FOXM1c and/or IRF1 as independent prognosis biomarkers or drug targets for oesophageal cancer.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.