Colon cancer cells secreted CXCL11 via RBP‐Jκ to facilitated tumour‐associated macrophage‐induced cancer metastasis

Abstract Metastasis is the main cause of colon cancer‐related deaths. RBP‐Jκ is involved in colon cancer development, but its function in colon cancer metastasis is still unclear. Tumour‐associated macrophages are the main cell components in tumour microenvironments. Here, we aimed to determine the function of RBP‐Jκ in colon cancer metastasis and its underlying mechanisms for modulating interactions between colon cancer cell and tumour‐associated macrophages. Through bioinformation analysis, we found that RBP‐Jκ was overexpressed in colon cancer tissues and associated with advanced colon cancer phenotypes, macrophage infiltration and shorter survival overall as confirmed by our patients’ data. And our patients’ data show that RBP‐Jκ expression and tumour‐associated macrophages infiltration are associated with colon cancer metastasis and are independent prognostic factors for colon cancer patients. Tumour‐associated macrophages induced colon cancer cell migration, invasion and epithelial‐mesenchymal transition through secreting TGF‐β1. Colon cancer cells with high RBP‐Jκ expression induced the expression of TGF‐β1 in tumour‐associated macrophages by secreting CXCL11. Our research revealed that colon cancer cells secreted CXCL11 via overexpression of RBP‐Jκ to enhance the expression of TGF‐β1 in tumour‐associated macrophages to further promote metastasis of colon cancer cells.

RBP-Jκ. By targeting Notch pathway with secretase inhibitor, the production of NICD is prevented and the RBP-Jκ-dependent transcriptional activity is regulated. Recently, RBP-Jκ has been shown to be overexpressed in various human cancers, such as osteosarcoma, 7 glioblastoma 8 and squamous cell skin carcinoma, 9 indicating that it may be an oncogene or act as a tumour promoter. But secretase inhibitor is not like knocking down RBP-Jκ which can inhibit the proliferation of cancer cells to the greatest extent. This shows that the function of promoting tumour cell proliferation of RBP-Jκ is not completely dependent on Notch signal. 10 Although a recent study reported that RBP-Jκ knockdown suppressed oral cancer cell EMT, 11 the exact function of RBP-Jκ in colon cancer metastasis remains unclear.
Metastasis is a complicated process that not only involves cancer cells but also the tumour microenvironment (TME). 12 Monocytes enter tumour lesions and polarize to various subtypes of macrophage as different stimulants in the TME. Macrophage polarize into the M1 type by interferon-gamma and lipopolysaccharide or into the M2 type (also known as TAMs) by IL-4 and IL-13. 13 TAMs (tumour-associated macrophages) play a key role in tumour development; however, the function of TAMs in colon cancer metastasis remains unclear.
Therefore, in this study, we explored the clinical significance of the expression of RBP-Jκ and the infiltration of TAMs, and the revealed the underlining mechanisms of RBP-Jκ and TAMs in colon cancer metastasis.

| Immunohistochemical staining
Formalin-fixed and paraffin-embedded tissue samples of human and mouse xenograft tissues were used for IHC based on a previous study. 15 The antibodies used were shown in Table S1. The stained sections were semiquantitatively scored based on a previous study. 16

| Cell culture and transfection
The human colon adenocarcinoma cell lines RKO  and SW480-NC (negative control of SW480-R cells). Furthermore, tumour-associated macrophages (TAMs) were plated in 6-or 24well plates for transfection with TGF-β1 small interfering RNA (GenePharma Co. Ltd., Shanghai, China) or a plasmid carrying TGF-β1 cDNA (GenePharma) using Turbo-Fect TM (Thermo Fisher Scientific, Waltham, MA, USA). The RKO-shR cells in 6-well plates were also transfected with a plasmid carrying C-X-C motif chemokine 11 (CXCL11) cDNA (GenePharma) using Turbo-Fect TM (Thermo Fisher Scientific), and the SW480-R cells in 6-well plates were transfected with CXCL11 siRNA (GenePharma) using Turbo-Fect TM (Thermo Fisher Scientific). mRNA expression was determined after 24 h by real-time PCR (qPCR). Total cell protein was extracted for Western blot analysis after 48 h.

| RNA isolation and qPCR
Total RNA was isolated from cells using Fast200 (Tiangen, Beijing, China) and was then reverse transcribed into cDNA using Prime Script TM RT Master Mix (Takara, Shiga, Japan) according to the manufacturer's protocols. The relative expression of RBP-Jκ, TGF-β1, CXCL11 and GAPDH mRNA was measured by qPCR with SYBR® Premix Ex Taq™ II (Perfect Real Time, Takara); the specific primer sequences are shown in Table S2. Their relative levels were quantified using the 2 (−ΔΔCT) method against the GAPDH level in each cell line.
The experiments were performed in triplicate and were repeated three times.

| Protein extraction and western blot analysis
Western blot analysis was performed according to a previous study. 15 The antibodies used were shown in Table S1. The experiments were repeated three times.

| TAMs induction and flow cytometry
To induce TAMs, Thp-1 cells were plated in 6-or 24-well plates at a density of 3 × 10 6 /ml and were treated with 100 ng/ml phorbol 12-myristate 13-acetate (PMA) (Sigma) in the dark for 24 h and the cells were further treated with 20 ng/ml interleukin 4 (IL-4) (Sigma) and 20 ng/ml IL-13 (Sigma) for 36 h. 13,17 The expression of CD68 and CD163 was used to identify type 2 TAMs. Next, TAMs were digested with an ethylene diamine tetra acetic acid-free pancreatic enzyme solution and were collected and washed twice with phosphate-buffered saline (PBS), incubated with FcBlock (564219, BD Biosciences, Franklin Lakes, NJ, USA) at room temperature for 10 min and incubated with CD68 monoclonal antibody (11-0689, Thermo Fisher Scientific) and CD163 monoclonal antibody (12-1639, Thermo Fisher Scientific) in the dark for 30 min at 37°C. The cells were then measured using a flow cytometer and analysed with FlowJo software (BD Biosciences).

| Tumour cell migration and invasion assays
Tumour cell migration and invasion ability were assessed using 24-well Transwell chambers with a pore size of 8 µm (Corning). For the invasion assay, the Transwell filters were precoated with 50 µl of 200 mg/ml Matrigel (356234, BD Biosciences). In brief, 1 × 10 5 colon cancer cells (for the migration assay) or 3 × 10 5 colon cancer cells (for the invasion assay) in 200 µl of FBS-free medium were plated in the upper chamber, and the serum-supplemented medium or TAMs were used as attractants in the lower chamber. Then, the cells were incubated at 37°C for 24 h, and cells migrating or invading the underside of the filters were fixed in 100% methanol, stained with a 0.5% crystal violet solution and counted under a microscope (Olympus Corporation, Tokyo, Japan). The means of the triplicate assays were used for statistical analysis.

| Wound healing assay
Cells were seeded in 6-well plates. Then, a wound was created across the entire well using a sterile pipette tip. After washing with PBS, the cells were further cultured in a serum-free medium for an additional 48 h, and then, wound closure was measured after capturing photographic images from six randomly selected microscopic fields. The means of the triplicate assays were used for statistical analysis.

| Enzyme-linked immune sorbent assay
Cell culture supernatants were harvested, and the concentrations of TGF-β1 and CXCL11 were determined using TGF-β1 and CXCL11 ELISA kits (R&D Systems, Minneapolis, MN, USA) according to the manufacturer's protocols. The assays were conducted in triplicate and were repeated three times.

| Animal experiments
The animal experiments in this study were approved by the Institutional Animal Care and Use Committee of Xi'an Jiaotong University. Four-week-old male BALB/c nude mice were purchased from the Animal Center of Xi'an Jiaotong University and housed in a specific pathogen-free facility with free access to autoclaved food and water. The mice were randomized into 12 groups (n = 6) and were implanted subcutaneously in the back with the following: 1 ×

| RNA sequencing
Total RNA prepared from RKO-NC or RKO-shR cells (see qPCR section) was purified and quantified for construction of RNA sequencing libraries. Then, the libraries were sequenced on the Illumina HiSeq platform; reads containing adapter or poly-N or of low quality were removed, and high-quality and clean reads were further analysed (i.e. they were Q20 [>96%], Q30 [>92%], with an uncertainty rate [<0.01%] for the clean data). The mapped reads were obtained using Tophat2 after alignment to the reference of the human genome. The number of mapped clean reads for each unigene was counted and then normalized to reads/kb/million reads to calculate the unigene expression levels.

| Statistical analysis
The data are expressed as the mean ±standard deviation for each case number. Differences between groups were analysed using Student's t test, and differences among multiple variable groups were analysed using one way ANOVA analysis. The associations between the percentage of positive staining and clinical parameters were assessed using Pearson's χ 2 test. Moreover, overall survival (OS), which was defined as the time from tumour diagnosis to patient death or the last follow-up, was used to generate Kaplan-Meier curves, was statistically analysed using the log-rank test and was further assessed using univariate and multivariate analyses. All statistical analyses were conducted using SPSS 22.0 software (International Business Machines Corporation, Armonk, NY, USA) with two-side analysis, and a p value <0.05 was considered statistically significant for levels of *p < 0.05, **p < 0.01 and ***p < 0.001.

| RBP-Jκ overexpression was associated with macrophage infiltration in colon cancer tissues
GEPIA was used to assess the expression of RBP-Jκ in colon cancer tissues. We found that RBP-Jκ expression in colon cancer tissues (n = 275) was significantly higher than that of normal colon tissue (n = 41, p<0.01; Figure S1A). Survival analysis showed that high expression of RBP-Jκ indicated shorter OS (p = 0.048) and DFS (p = 0.060; Figure S1B-S1C). Then, we used Oncomine to further confirm this result (p < 0.001; Figure S1D-S1G). TCGA data showed that RBP-Jκ was overexpressed in colon cancer tissues (31 pairs, p = 0.047; 478 tumour tissues vs 31 para-carcinoma tissues, p = 0.046; Figure S1H-S1I). Next, we collected clinical parameters of 247 patients from TCGA data set. Correlation analysis showed that high expression of RBP-Jκ was associated with depth of tumour invasion (p < 0.001) and distance metastasis (p = 0.022, Table S3). The above analysis results implicated that RBP-Jκ was overexpressed in colon cancer tissues and was associated with metastasis.
Finally, we used TIMER to analyse the relevance between immune cell infiltration and RBP-Jκ in colon cancer. Results showed that the macrophage infiltration was associated with overall survival of colon cancer patients. That was patients with high macrophage infiltration had a shorter overall survival than patients with low macrophage infiltration (p = 0.035; Figure S1J). RBP-Jκ was positively correlated with macrophage infiltration (r = 0.486, p = 2.72e-25; Figure S1K). Further analysis revealed that RBP-Jκ was positively correlated with CD163, which is the marker of M2 type macrophage (r = 0.47, p = 1.02 × 10 −23 ; Figure S1L).

| RBP-Jκ and CD163 were independent prognostic predictors of colon cancer patients
To verify the results above, we compared the expression of RBP-Jκ and CD163 in colon cancer and normal tissues from 201 patients ( Figure 1A). Our IHC data showed that high RBP-Jκ expression was observed in 124 colon cancer cases (61.69%) compared to 41 cases (20.40%) of para-tumour tissues (p<0.001; Table 1). High  Table 1). Moreover, RBP-Jκ expression exhibited a significant positive association with CD163 (r = 0.562, p < 0.001, Table S4).
Moreover, patients with high tumour RBP-Jκ expression had significantly shorter OS (p < 0.001; Figure 1B) and patients with high CD163 expression in colon cancer had a significantly shorter OS (p < 0.001; Figure 1C). Furthermore, combining RBP-Jκ expression and CD163 expression, stratification analysis showed that patients with high RBP-Jκ expression and high CD163 expression had the shortest OS (p < 0.001, Figure 1D, 1E).

| TAMs promoted colon cancer cell metastasis by secreting TGF-β1
As high CD163 expression was associated with cancer metastasis, we assessed the effects of CD68 + CD163 + TAMs on colon cancer cells. We induced Thp-1 into TAMs and the CD68 and CD163 double-positive rate of TAMs was 87.85% ±7.19% ( Figure S2A).
In addition, we also assessed the effect of TAMs on the induction of colon cancer cell EMT in the nude mouse xenograft tumour ( Figure 5A) by using RKO or SW480 cells with or without macrophages. CD163 immunostaining of the mesenchyme in the cocultured group showed higher expression than that of the RKO or SW480-alone groups, whereas E-cadherin immunostaining in the cocultured group was lower than the RKO or SW480-alone groups.
The expression of N-cadherin was higher in the cocultured group than that in the RKO or SW480-alone groups ( Figure 5B, 5C). Lung metastasis models ( Figure 5D)  cells formed smaller and fewer metastasis nodules in the lungs than that with macrophages ( Figure 5E).

| RBP-Jκ promoted colon cancer cell metastasis through inducing TAMs secret TGF-β1
Taken together, our data show that RBP-Jκ and TAMs participate in promoting colon cancer metastasis. We then detected the rela-  Figure 6D, Figure S4H).  Figure 7B, 7C). Similarly, data from the lung metastasis model ( Figure 7D) also confirmed these findings, which showed that metastasis nodules in the RKO-shR plus M2-T or SW480-R plus M2-NC1 groups were larger and more prevalent than those in the RKO-shR plus M2-NC2 or SW480-R plus M2-siT groups ( Figure 7E).
These data suggest that RBP-Jκ expression in colon cancer cells promotes tumour cell metastasis via TAM expression of TGF-β1.

| Colon cancer cell with overexpressed RBP-Jκ inducted TAMs to express TGF-β1 by secretion of CXCL11
We demonstrated that colon cancer cells with RBP-Jκ overexpression induced TAMs to express TGF-β1, but the underlying molecular events still remained to be determined. Therefore, we performed an and IGFBP1, insulin-like growth factor binding protein 1). Then, we used TIMER to analyse the relevance between macrophage infiltration and the 7 genes. Results showed that CP, A2 M, CXCL11 and IGFL2 were positively related to macrophage infiltration, while FGB and FGA were negatively related to macrophage infiltration, and IGFBP1 had little to do with macrophage infiltration ( Figure 8C).

The sequencing analysis was confirmed by real-time PCR and
Western blot analyses which showed that the CXCL11 mRNA and protein levels were lower in RKO-shR and SW480-NC cells than those in RKO-NC and SW480-R cells, respectively ( Figure 8F, 8G, Figure S4J).
Moreover, the RKO-shR or SW480-NC-cultured supernatant concentration of CXCL11 was also significantly lower than that of RKO-NC  Figure 8H).
We cocultured TAMs with RKO-NC, RKO-shR-NC, RKO-shR-C, SW480-R, SW480-R-NC or SW480-R-siC cells for Western blot analysis of TGF-β1. We found that RKO-shR-C cells or SW480-R and SW480-R-NC cells stimulated TAMs to express TGF-β1 ( Figure 8K, Figure S4L). These data indicate that RBP-Jκ induces TAMs to express TGF-β1 by increasing colon cancer cell secretion of CXCL11. Using an RNA sequencing assay, we found that RBP-Jκ was able to upregulate the expression of CXCL11 in colon cancer cells. CXCL11 is a member of the CXC chemokine family. 28 CXCL11 is the ligand with the highest affinity to CXCR3, followed by CXCL9 and CXCL10. 29 As an ELR (Glu-Leu-Arg)-negative CXC chemokine, CXCL11 can generally attenuate angiogenesis and thus have an antitumour effect. 30 In this study, we established a xenograft tumour model for tail vein To our knowledge, our study being conducted at present is the first one that investigates the association between RBP-Jκ and CXCL11 expression in colon cancer, as well as high RBP-Jκ and  Figure S3).

ACK N OWLED G EM ENTS
We thank TCGA data set for providing data and all the patients participated in our study.

CO N FLI C T O F I NTE R E S T S TATE M E NT
All the authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available at gepia2.cancer-pku.cn/, www.cancer.gov, www.oncom ine.org, https://cistr ome.shiny apps.io/timer/. The CXCL11 concentrations in the cell culture supernatants showed that silencing or overexpression of RBP-Jκ decreased or increased CXCL11 levels, respectively. (I) Western blot analysis. The CXCL11 protein was overexpressed or silenced in RKO-shR and SW480-R cells, respectively, after transfection of CXCL11 cDNA or siRNA. (J) Real-time PCR analysis. CXCL11 mRNA was overexpressed or silenced in RKO-shR and SW480-R cells, respectively, after transfection of CXCL11 cDNA or siRNA. (K) Western blot analysis. TGF-β1 expression in TAMs was up-or downregulated after coculturing with RKO-shR-C or SW480-R-siC cells, respectively. **p < 0.01 and ***p < 0.001. NC, negative control