HOXC6 Promotes Metastasis of MSI-H CRC via Interacting with M2 Macrophages

Background: HOXC6 was the most signicantly upregulated gene in right-sided colon cancer (RCC) compared to left-sided colon cancer (LCC) according to our previous study. It is known that RCC has a higher immunogenicity than LCC because much higher prevalence of MSI-H samples, however, the role of HOXC6 acted in MSI-H tumors remains poorly understood. Methods: Expression datasets from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus databases were used to analyze the differential expression and prognostic role of HOXC6 in CRC. ELISA and qRT-PCR were performed to examine the association of HOXC6 and CCL2. Immunohistochemistry and immunouorescence were performed to evaluate the correlation of HOXC6 and M2 macrophage inltration. CCK8 and Transwell were used to evaluate the proliferation and migration of tumor cells in vitro. Results: HOXC6 was overexpressed in MSI-H CRC and associated with poor prognosis. Upregulation of CCL2 by HOXC6 could attract more M2 macrophage inltration. IL6 secreted by M2 macrophages could induce epithelial-mesenchymal transition (EMT) of tumor cells by upregulating HOXC6. Overexpression of HOXC6 promoted the migration and invasion of CRC cells in vitro. Finally, inhibition of IL6/JAK pathway using ruxolitinib downregulated HOXC6 and suppressed invasion of HCT116 cells. Conclusion: Our study revealed that overexpression of HOXC6 attracted more M2 macrophage inltration and the positive crosstalk between M2 macrophages and HOXC6-highly expressed tumor led to EMT and enhanced the migration ability of CRC, which offered a promising therapic target for treatment of MSI-H CRC.


Introduction
Colorectal cancer (CRC) is one of the most common diagnosed malignant tumors, with the third morbidity and mortality among all malignant tumors in the United States [1]. Liver metastasis is observed in more than half CRC patients and less than 20% of newly diagnosed metastatic CRC patients survive more than ve years [2]. This situation highlights the pressing need to better understanding of metastatic development and better controlling of tumor progression in CRC.
In recent years, a bunch of studies have revealed that right-sided colon cancer (RCC) has a higher immunogenicity than left-sided colon cancer (LCC) because much higher microsatellite instability-high (MSI-H) frequency and better response to anti-PD-1/PD-L1 immunotherapies [3][4][5]. Though generally MSI-H confers CRC patients a favorable outcome, stage IV MSI-H CRC patients have a worse prognosis than microsatellite stable (MSS) patients (11.1 months vs. 22.1 months, P = 0.017) [6], with the mechanism remains poorly understood. We previous found MLH1 inactivation upregulates HOXC6 moderately, suggesting that other than intratumor per se, stromal or immune cells within tumor microenvironment (TME) may also has a role in regulating tumoral HOXC6 expression.
In this study, we con rmed that HOXC6 was overexpressed in MSI-H CRC and associated with a worse prognosis. Furthermore, we discovered a novel association between HOXC6 overexpressed CRC cells and M2 macrophages using bioinformatic exploration coupled with experimental validation. Upregulation of CCL2 by HOXC6 in tumor cells attracted more M2 in ltration within TME. Additionally, the regulatory axis IL6/HOXC6 in CRC cells contributing to EMT induced by M2 macrophages was identi ed and inhibition of HOXC6 signaling cascade by targeting IL6/JAK pathway suppressed EMT, which helped to better understand the crosstalk between CRC and tumor associated macrophages (TAM) and offered a good candidate target in CRC treatment.

Materials And Methods
Bioinformatic analysis of TCGA and GEO datasets All Level 3 CRC RNASeqV2 mRNA expression pro les were obtained from TCGA (08/ 26,2017). The raw CEL les of GSE39582 (Affymetrix HG U133 Plus 2.0 arrays) were downloaded from Gene Expression Omnibus (GEO) and processed using the affy package of BioConductor [10]. Then, the MAS5 algorithm was used for background correction, normalization and summarization of single probes for all probe sets. The identi cation of differentially expressed genes (DEGs) between subgroups was performed as in our previous studies [7,8]. Signi cant DEGs were selected according to a false discovery rate (FDR) adjusted P-value < 0.05 and fold change > 2. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed using the clusterPro ler package from Bioconductor based on the DEGs identi ed between the HOXC6+ (> median) and HOXC6-(< median) groups [9]. Signi cantly enriched GO terms and pathways were selected according to a FDR-adjusted Pvalue < 0.01. Hallmark gene sets from the molecular signatures database (MSigDB) [10] were used to determine whether any signatures were enriched in speci c groups by gene set enrichment analysis (GSEA) [11]. Signi cantly enriched hallmarks were chosen according to a P-value < 0.05.
To quantify the relative amount of distinct lymphocytes in CRC, CIBERSORT [12] was used to calculate the proportions of 22 lymphocytes in tumor tissue. The permutation was set to ≥ 100, and quantile normalization of the input expression mixture was set to FALSE for TCGA RNAseq dataset. Samples with a P-value > 0.05 were excluded from further comparisons. Consensus tumor purity was re ned based on a previous systematic pan-cancer measurement of tumor purity [13].
Immuno uorescence (IF) staining and calculation of CD206-positive area A total of 53 stage II RCC samples collected from Zhejiang University Cancer Institute (ZUCI) were used for IF staining to evaluate the expression of CD206 (a marker of M2 macrophages). Written informed consent was obtained from all patients before enrolment. A primary antibody against CD206 (1:200, HUABIO, ET1702-04) was used for IF staining. ImageJ was used to calculate the CD206-positive area percentage. In brief, the image was separated based on the channels and displayed as 8-bit images.
Then, the threshold was adjusted. The areas of DAPI (nuclear location) and CD206 were calculated. The percentage of CD206-positive area was calculated as (CD206-positive area)/(DAPI-positive area).

Cell culture and coculture
The HCT116 and THP-1 cell lines were purchased from ATCC and cultured with RPMI 1640 medium (Gibco) containing 10% fetal bovine serum (FBS, BI Industry). The cells were incubated at 37℃ with 5% Cocultivation of macrophages and HCT116 cells was conducted with the noncontact coculture Transwell system (Corning, USA). Inserts containing 1.0 × 10 6 THP-1 cells or M2 macrophages were transferred to 6well plates previously seeded with HCT116 cells (2.5 × 10 5 cells per well) and cocultured in 1.5% FBScontaining medium for 72 h. HCT116, THP-1, or M2 macrophages were cultured in 1.5% FBS-containing medium as a negative control. After coculture, macrophages, HCT116 cells and culture medium were harvested for use. IL6 was purchased from PeproTech (Catalog Number: 200-06).
Transwell migration assays Cell migration was examined by Transwell assays without Matrigel. Approximately 10 4 cells were plated into the upper chamber with RPMI 1640 medium without FBS. RPMI 1640 medium supplemented with 20% FBS was added to the lower chamber. After 48 h of culture for HCT116 cells, the cells in the upper chamber were scarpered, and cells under the upper chamber were xed with 4% formalin and stained with crystal violet. The migrated cells were counted by light microscopy, and the mean cell number of three random visual elds at a magni cation of 200 × was recorded.
Cell proliferation Approximately 1 × 10 3 cells were plated into 96-well plates. Cell viability was measured at 1, 3, 5, and 7 days after plating. Cell Counting Kit-8 (CCK-8, Dojindo, Japan, CK04) was utilized for cell viability testing. Cell culture medium was used as a blank control. After 2-3 hours of incubation, an optimal density (OD) value of 450 nm was used to detect cell proliferation. Experiments were carried out in triplicate. Ruxolitinib were purchased from Selleck.
Protein extraction and western blotting RIPA buffer (Beyotime) with 1% protease inhibitor cocktail (Roche Applied Science) was used for total protein extraction. After quanti cation of protein concentration and boiling with protein loading buffer, 10% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) gels were used to separate proteins and polyvinylidene uoride (PVDF) membranes by electrophoresis were used for protein transition. After blocking with 5% nonfat milk, the PVDF membranes were incubated with primary antibodies followed by horseradish peroxidase (HRP)-linked secondary antibodies. Enhanced chemiluminescence (ECL) reagent was used to detect the protein bands. The primary antibodies used for western blotting were as follows: anti-HOXC6 (Santa Cruz, sc-376330), anti-EMT antibody kit (Cell Signaling Technology, #9782), anti-IL6 (Abcam, ab233551), and anti-β-tubulin (Huabio, Hangzhou, M1305-2). β-tubulin was used as a protein loading control.

RNA isolation and qRT-PCR
Total RNA was extracted from HCT116 cells using Trizol following a standard protocol. The Takara PrimeScript TM RT Master Mix Kit (Takara, RR036Q) was used for reverse transcription. The iTaq Universal SYBR Green Supermix (BioRad) and Applied Biosystems 7500 Fast Real-Time PCR System were applied for qRT-PCR. GAPDH was used as the loading control. Experiments were carried out in triplicate. The results were calculated as follows: ΔCT = CT Experimental/NC -CT GAPDH ,ΔΔCT = ΔCT Experimental/NC -ΔCT NC, fold change = 2 −ΔΔCT . The primers used for qRT-PCR are as follows.

Statistical analysis
All statistical analyses and graphical representations were performed in the R programming language (× 64, version 3.5.1), IBM SPSS Statistics 22 and GraphPad Prism 7 unless otherwise speci ed.

HOXC6 was overexpressed in MSI-H CRC and led to poor prognosis
We previously revealed that HOXC6 was upregulated in RCC compared to LCC and normal colon tissues [7]. Then, we further explored the expression pattern of HOXC6 in MSI-H and MSS samples. Results showed that HOXC6 had the highest expression level in RCC MSI-H than in RCC MSS and LCC samples, suggesting that a close association of HOXC6 and MSI-H (P < 0.001, Fig. 1A).
Next, we assessed the prognostic value of HOXC6 based on TCGA and GSE39582 samples and found that high expression of HOXC6 was signi cantly associated with poor clinical outcome in MSI-H CRC (log rank P = 0.02, Fig. 1B).
HOXC6 recruited M2 macrophage in ltration via upregulating cytokine CCL2 To explore whether HOXC6 plays a role in the interaction between tumor cells and tumor microenvironment in CRC, we investigated in detail the functional phenotype under HOXC6 regulation. In this context, we compared the overall transcriptome variation between the HOXC6+ (mean RSEM: 133) and HOXC6-groups (mean RSEM: 12). There were 1,226 DEGs identi ed between the HOXC6 + and HOXC6-groups, most of which (1,001, 82%) were potentially upregulated by HOXC6 ( Fig. 2A). For example, cytokines such as CCL2, CCL5, and IL6 had a much higher expression level in the HOXC6 + group ( Fig. 2A). Furthermore, GO enrichment dissection revealed that these upregulated DEGs mainly participated in leukocyte migration (P < 0.001, FDR adjusted), extracellular matrix organization (P < 0.001), and leukocyte chemotaxis (P < 0.001, Fig. 2B). KEGG pathway enrichment also showed that cytokine − cytokine receptor interaction (P < 0.001), osteoclast differentiation (P < 0.001) and chemokine signaling pathway (P < 0.001, Fig. 2C) were the top enriched pathways, indicating that HOXC6 was positively linked to TME reshu ing. Thus, tumor purity was compared between the HOXC6 + and HOXC6-groups. Not surprisingly, tumor purity was signi cantly decreased in the HOXC6 + group (P = 8.6E-5, Wilcoxon test, Fig. 2D). To further determine the cell types accounting for the lower tumor purity in the HOXC6 + group, CIBERSORT was used to estimate the abundance of diverse cell types in CRC. We found that leukocytes were signi cantly increased in the HOXC6 + group (P = 3.1E-5, Wilcoxon test, Fig. 2D), and this difference was mainly due to M2 macrophage in ltration (P = 0.02, Wilcoxon test, Fig. 2D). Therefore, we speculated that M2 macrophages could be recruited into the tumor region by CCL2 under high levels of HOXC6.
To test this hypothesis, we experimentally investigated the relationship between HOXC6 and CCL2/M2 macrophage in ltration. First, we selected HCT116 cell line for HOXC6 overexpressed and knockdown cell line construction (Fig. S1A-C) because HCT116 belonging to one of the dMMR cell lines. At the mRNA level, CCL2 was elevated by 1.8-fold (P < 0.05, t-test) in the HCT116 HOXC6-OE group and decreased 41% (P < 0.05, t-test) in the HCT116 siHOXC6 group compared to the NC group (Fig. 3A). At the protein level, the CCL2 concentration in cell culture supernatants was elevated from 7.9 to 10.2 pg/ml (P < 0.01, t-test) by overexpressing HOXC6 in HCT116 cells (Fig. 3B). Furthermore, immuno uorescence was performed in 53 stage II RCC samples collected from the ZUCI cohort to evaluate M2 macrophage in ltration by the CD206-positive area (Fig. 3C). Through combined evaluation of HOXC6 and CD206, we con rmed a signi cantly positive correlation between M2 macrophage in ltration and HOXC6 expression (P = 0.0002, r = 0.48, Fig. 3D). In conclusion, these results revealed that upregulation of HOXC6 could recruit more macrophages by upregulating CCL2.
Overexpression of HOXC6 was associated with IL6/JAK pathway and EMT Next, we explored the detailed molecular mechanism in the crosstalk between HOXC6 overexpressed tumors and M2 macrophages. GSEA was performed to identify prominent signatures associated with high HOXC6 expression. Interestingly, IL6/JAK/STAT3 signaling (NES = 1.78, P = 0.014, Fig. 4A) and EMT (NES = 1.71, P = 0.038, Fig. 4B) were characteristic of HOXC6 overexpression. Considering that HOXC6 was related to M2 macrophage in ltration by CCL2 and that cancer cell EMT could be induced by IL6 secreted from TAMs [14], we hypothesized that M2 macrophages induce EMT in tumor cells by regulating the IL6/HOXC6 axis.

M2 macrophages induced EMT of cancer cells by secreting IL6 and regulating the IL6/HOXC6 axis
To test our hypothesis, coculture assay was performed to examine the crosstalk between HCT116 cells and M2 macrophages. M2 macrophages were sequentially induced by PMA, IL4 and IL13 from progenitor THP-1 cells in vitro (Fig. 4C). CD163 and CD206, two M2 macrophage surface markers, were signi cantly upregulated in the induced M2 macrophages compared to THP-1cells (Fig. 4D). Then, the optimal coculture time for M2 macrophages/IL6 and tumor cells was determined to be 72 h/48 h according to the expression levels of HOXC6 ( Fig. 4E and F). After 72 h coculture, the expression of IL6 in M2 macrophages was upregulated three-fold compared to that in M2 macrophages alone (P < 0.01, t-test, Fig. 4G). Furthermore, both the mRNA and protein levels of IL6 in M2 macrophages were signi cantly downregulated when coculture with ltv-shHOXC6 cells and upregulated by HOXC6-OE cells compared to shCtrl cells in the HCT116 cell background ( Fig. 4H and I).
Next, the in uence of M2 macrophages on tumor cells was also investigated. EMT of HCT116 cells was induced by downregulating ZO-1 and E-cadherin and upregulating snail, when coculturing with M2 macrophages, which was similar to coculture with human recombinant IL6 (Fig. 5A). In addition, HOXC6 could be upregulated when coculturing with M2 macrophages or IL6 only (Fig. 5A). When IL6 was knocked down in M2 macrophages, EMT was reversed and HOXC6 was downregulated as well (Fig. 5A).
In addition, in spite of coculturing with M2 macrophages, MET was accomplished by knockdown of HOXC6 in HCT116 cells (Fig. 5B), indicating that HOXC6 played a critical role in metastasis. To test our hypothesis, we performed Transwell assay to evaluate whether HOXC6 could enhance the migration ability of tumor cells by inducing EMT. The results showed that the migration ability was greatly enhanced under high expression of HOXC6 ( Fig. 5C and D), however, the proliferation ability of tumor cells remained unchanged ( Fig. S2A and B). In summary, increased secretion of IL6 from M2 macrophages is the results of high HOXC6 expression in tumor cells and in turn M2 macrophages could induce EMT of tumor cells through the IL6/HOXC6 axis.

Blocking of IL6/HOXC6 axis is a promising strategy for MSI-H CRC treatment
In view of TAMs induce EMT via the IL6/HOXC6 axis, we next assessed whether blockade of HOXC6oriented signaling pathway would be a feasible therapeutic regimen for CRC treatment. Given HOXC6 is not an approved druggable target so far, we explored whether inhibiting the IL6/JAK pathway was a potential alternative therapeutic schedule for HOXC6-overexpressed MSI-H CRC. Interestingly, ruxolitinib, a selective JAK1/2 inhibitor, inhibited tumor cell EMT by upregulating E-cadherin and downregulating snail and HOXC6 in a dose-dependent manner (Fig. 6A). Moreover, HOXC6 was downregulated by ruxolitinib despite the presence of IL6 (Fig. 6B). Furthermore, IL6 could not upregulate HOXC6 and snail nor downregulate E-cadherin when ruxolitinib existence. These results indicated that HOXC6 could be indirectly druggable by ruxolitinib.

Discussion
In this study, HOXC6 was con rmed to be overexpressed in MSI-H CRC and associated with poor prognosis, which is in line with previous studies in other cancer types [15][16][17]. As a heterogeneous ecosystem, tumors are composed of various components, such as diverse stromal cells and lymphocytes [18]. Deconvolution of 22 lymphocytes in tumor tissue using CIBERSORT revealed a signi cant enrichment of M2 macrophages in HOXC6 high-expression groups in CRC. TAMs are the main component of in ltrated immune cells in TME [19] and have been shown to play an important role in tumorigenesis [20], angiogenesis [21,22], metastasis [23,24], and chemoresistance [25][26][27][28][29][30]. Here, we rst reported the reciprocal interaction of HOXC6 between CRC and TAMs. We found that cytokine CCL2, a known strong monocyte-attracting cytokine that recruits macrophages to the tumor region, could be upregulated by HOXC6. Furthermore, IL6 was upregulated in macrophages by coculture with HOXC6-OE CRC cells and vice versa. IL6 secreted by macrophages was associated with EMT of tumor cells, which was consistent with a previous study [14]. Additionally, IL6 secreted by TAM has various effect on tumor cells, such as promotes expansion of human hepatocellular carcinoma stem cells [31] and cancer immune evasion [32].
In summary, we rst found that high expression of HOXC6 could attract more macrophages into the tumor area by promoting CCL2 secretion, which in turn resulted in upregulation of HOXC6 in tumor cells via IL6 secreted from macrophages (Fig. 6C).
With regard to the downstream regulatory cascade of HOXC6 participating in CRC progression, we found high expression of HOXC6 was signi cantly associated with IL6/JAK/STAT3 pathway and subsequent EMT of CRC cells, and targeting JAK not only inhibited EMT but also downregulated HOXC6, which indicated that the potential IL6/JAK/HOXC6 axis present in CRC. In view of there is no HOXC6-derived drug currently, targeting IL6/IL6R or JAK is a feasible strategy to block the positive crosstalk for HOXC6 overexpressed or high TAM in ltrated CRC patients, such as sarilumab (IL6R inhibitor) approved by the

Conclusion
In conclusion, the most important nding in this study is high expression of HOXC6 attracts more M2 macrophage in ltration via upregulating CCL2 and M2 macrophages induced EMT of cancer cells by regulating IL6/HOXC6 axis. In addition, block IL6/JAK pathway accompanied by downregulating HOXC6 and inhibiting EMT, which provided an alternative therapy strategy for indirectly targeting HOXC6.
Consequently, HOXC6 could serve as a biomarker for CRC metastatic potential predicting and a drug target but warrant further investigation.

Consent for publication
All authors read and are consent for the publication of this manuscript.

Availability of data and materials
The datasets used during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.