CXCL11 Correlates with Anti-tumor Immune Microenvironment and Improved Prognosis in Colon Cancer

Background: CXCL11 has been considered to be responsible for tumor development, but the specic effect of CXCL11 in colon cancer was still obscure. Therefore, the prognostic value and immunological regulation effect of CXCL11 in colon cancer were evaluated in this study. Methods: Three independent datasets were used for mRNA-related analysis: one dataset from the Cancer Genome Atlas (TCGA, n=451) and two single-cell RNA sequencing (scRNA-seq) datasets from Gene Expression Omnibus (GEO): GSE146771 and GSE132465. In addition, the patient cohort (the Yijishan Hospital cohort, YJSHC, n=108) was utilized for cell inltration-related analysis, accordingly. Both CXCL11 mRNA expression and CXCL11 + (CXCL11-producing) cells were assessed in colon cancer, whose effect on prognosis and immunological regulation was also studied. Results: High CXCL11 expression were associated with better prognosis in colon cancer, which was still signicant even if clinicopathological factors were adjusted. Furthermore, CXCL11 positively correlated with anti-tumor cells inltration, such as CD8 + T cells and natural killer cells. Meanwhile, CXCL11 correlated positively with several genes associated with DC, NK and T recruitment (cid:0) and a gene set of cytotoxic genes. Notably, CXCL11 correlated positively with several immune checkpoint related genes including of PD-L1. Conclusions: CXCL11 to could improve approach that expression of by genetic and (programmed death protein-1) antibody However, it

cytokines and cytokine receptors participating in in ammatory diseases and tumor development have been illustrated, but less is known about the speci c effect of CXCL11 in colon cancer. CXCL11 (or I-TAC, IFN-inducible T cell a-chemoattractant) belongs to the chemokines, which are a series of low-molecular-weight cytokines and can be categorized into 4 families (i.e. CXC, CC, C, and CX3C chemokines) according to the number and spacing of the amino-terminal cysteines [14,15]. CXC chemokines including of CXCL11 play a role in intercellular communication by mediating the activation and migration of speci c populations of leukocytes [16]. CXCL11 can be stimulated by interferon in leukocytes. However, CXCL11 is not only expressed in leukocytes, but also in broblast, endothelial cells and tumor cells [17][18][19]. Therefore, CXCL11 could make complex and unique effects by binding to CXCR3 and CXCR7 of all kinds of cells. On the one hand, CXCL11 may promote tumor development and progression in some situations [20][21][22]. On the other hand, it may inhibit tumor growth by enhancing antitumor immunity and inhibiting angiogenesis [23]. Although several gene expression signatures including of CXC chemokines have been set as the potential prognostic biomarkers in some studies [24][25][26], the role of CXCL11 in the tumor immune microenvironment and response to therapeutic treatment in colon cancer was still obscure. In order to provide additional information to help clinicians select more appropriate drugs and improve outcome in patients with colon cancer, more attention should be attracted to the prognostic value and immunological regulation effect of CXCL11 in colon cancer.
In this study, the prognostic value of CXCL11 mRNA expression and CXCL11 + cells in ltration in TCGA and YJSHC was investigated. Furthermore, the effect of CXCL11 on tumor immune microenvironment was evaluated. Currently, our work is the rst research to analyze the clinical value of CXCL11 mRNA and CXCL11 + cells, as well as the correlation between CXCL11 and tumor immune microenvironment comprehensively in colon cancer.

Patient samples
Three independent datasets were adopted in this study, which consisted of one dataset from TCGA (the Cancer Genome Atlas, n=451) and two scRNA-seq (single-cell RNA sequencing) datasets from GEO (Gene Expression Omnibus): GSE146771 and GSE132465. In addition, the patient cohort: YJSHC (the Yijishan Hospital cohort, n=108) was also involved. The characteristics of patients in TCGA were downloaded from http://www.cbioportal.org at 18 April 2020. The patients' statistical description of TCGA and YJSHC in this study was provided in supplementary

Immunohistochemistry (IHC) staining and evaluation
We collected Formalin-xed para n-embedded surgical specimens, which were used for tissue microarray construction and subsequent immunohistochemistry studies. Firstly, tissue microarray sections were rehydrated, which were treated with hydrogen peroxide and heat-mediated antigen retrieval by microwave with sodium citrate. Next, the slides were incubated with indicated antibodies which were obtained from Abcam (Cambridgeshire, UK) in a humidi ed box at 4℃ overnight. The instructions of four antibodies in this study were provided in supplementary Table S1. Following, color-reaction was realized by DAB substrate kit and nucleus counterstaining was carried out by hematoxylin. The level of protein expression was assessed based on the intensity of staining and the extent of staining at 200X under the microscope. For the extent of staining, the quantity score (0-4) denoted 0%, 1-10%, 11-50%, 51-80% and 81-100% of positive cells, respectively. The degree of staining intensity was divided into three grades, which were weak, moderate and strong staining. What should be noted was that the corresponding intensity scores were 1-3. The nal IHC score was calculated by multiplying the quantity and intensity scores.

Immune cells in ltration analysis
The relative abundance of the tumor in ltrated lymphocytes (TILs) in different CXCL11 mRNA expression status of colon cancer was calculated by CIBERSORT (https://cibersort.stanford.edu/) [27] and TISIDB (http://cis.hku.hk/TISIDB/index.php) [28]. For each cancer type, GSVA (gene set variation analysis) was adopted to infer the relative proportion of various TILs based on the immune-related gene expression pro le of 28 TIL types from Charoentong's study [29]. The correlation between CXCL11 and TILs was measured by Spearman's test. In this study, we also investigated the differential expression of CXCL11 between tumor and adjacent normal tissues across all TCGA tumors in TIMER (https://cistrome.shinyapps.io/timer/) [30].

Differential expression analysis
Differential expression analysis was conducted by R package "limma". Differentially expressed genes were up-regulated or down-regulated genes with fold change ≥2 and false discovery rate (FDR) p value <0.05 between CXCL11-high and CXCL11-low tumor samples.

Gene set enrichment analysis
The signi cantly enriched pathways in CXCL11-high tumor samples were identi ed by gene set enrichment analysis (GSEA) performed by "MSigDB" (Molecular Signatures Database, version 6.0).
Meanwhile, the enrichment score was assessed by the expression of a gene set and the "enriched" gene set represented the majority of a gene set had higher expression.

Single-cell transcriptome analysis
External single-cell transcriptome data of colon cancer patients were retrieved from Gene Expression Omnibus (GSE146771 and GSE132465). Single-cell data analysis was performed using scanpy v.1.4 (https://scanpy.readthedocs.io/en/stable/). Since single-cell technologies currently capture only a portion of the transcripts in any cell, we generated the "pseudobulks" as technical replicates by summing the raw UMI counts for each gene from the same patient.
Statistical analysis R and corresponding R packages were utilized for the statistical analysis and the Cutoff Finder platform (http://molpath.charite.de/cutoff) was utilized to automatically determine the cutoff points in this study [31]. Results were all displayed by mean ± SD. The two-tailed p value <0.05 was considered statistically signi cant. Student's t-test was adopted for the analysis of continuous variables. Spearman's correlation was made for the analysis of the correlation. Kaplan-Meier analysis was used for survival analysis and the log-rank test were used for the comparison of survival among multiple groups, respectively.

Aberrant expression and prognostic value of CXCL11 in colon cancer
As shown in Figure 1a, the Kaplan-Meier survival curve revealed that patients with high CXCL11 mRNA expression had better prognosis (p=0.0053) in TCGA. Next, TISIDB was utilized to examine the association between CXCL11 and clinical outcome across all TCGA tumors and we found that patients with high expression of CXCL11 mRNA had good prognosis especially in COAD (colon cancer) ( Figure   1b). Furthermore, we explored the differential expression of CXCL11 between tumor and adjacent normal tissues across all TCGA tumors in TIMER and found that CXCL11 was signi cantly up-regulated in tumor than adjacent normal tissues across several tumors ( Figure 1c). In addition, we also validated the up- Association of CXCL11 with tumor immune microenvironment Firstly, the differential tumor immune microenvironment in different groups of CXCL11 mRNA expression in TCGA was investigated by CIBERSORT. High CXCL11 mRNA expression group had higher proportion of anti-tumor immune cells, such as: CD8 + T cells (CD8T, p<0.001), activated natural killer cells (NKa, p<0.05), as shown in Figure 3a. On the contrary, high CXCL11 mRNA expression group had lower proportion of pro-tumor immune cells, such as: M0 macrophages (M0, p<0.05), resting natural killer cells (NKr, p<0.001), monocytes (Mono, p<0.05).
Next, TISIDB [28] was employed to validate which kinds of intratumoral lymphocytes might be regulated by CXCL11 across all TCGA tumors (supplementary Figure S1a). Surprisingly, we found that CXCL11 positively correlates with activated CD8 + T cells (Act CD8, r=0.55, p<0.001; Figure 3b Genetic alteration and enriched biological process in the high and low CXCL11 metagene expression groups Firstly, we performed GSEA and identi ed that biological pathways enriched in high-CXCL11 tumor samples were immune-activated process, which consisted of chemokine signaling pathway, cytokinecytokine receptor interaction, antigen processing and presentation, T cell receptor signaling pathway natural killer cell mediated cytotoxicity, intestinal immune network for IgA production, interferon alpha response interferon gamma response , as shown in Figure 3e-3l. We then analyzed the genetic alteration and found that differentially expressed genes in tumors revealed several immune-activated genes in high CXCL11 mRNA expression group. The volcano plot was shown in Figure 4a. As shown in Figure 4b  Cytotoxic genes [32][33][34] including IFNG, GZMA, GZMB, GZMK, GZMM and PRF1 were positively correlated with CXCL11 mRNA expression (Spearman's q=0.49, 0.55, 0.22, 0.37, 0.21 and 0.37; p<0.001, p <0.001, p<0.001, p=0.045, and p=0.004; respectively), as shown in Figure 4d. Interestingly, CXCL11 positively correlated with several immunosuppressive molecules [34] including of PDCD1,PD-L1, CTLA4 (Spearman's q=0.35, 0.56, and 0.44; p<0.001, p<0.001 and p<0.001; respectively), as shown in Figure 4e. Surprisingly, we also found the up-regulated expression of CXCL11 in tumor compared to adjacent normal tissue (Figure 4f and 4h Validating the association of CXCL11 + cells with TILs and PD-L1 Further analysis in YJSHC was conducted to validate the ndings in TCGA. As was expected, tumor with high abundance of CXCL11 + cells in ltration tended to have high abundance of intratumoral CD8 + T cells and CD56 + NK cells in ltration. In addition, high abundance of CXCL11 + cells in ltration was associated with high abundance of intratumoral PD-L1 + cells. Accordingly, the expression of PD-L1, CD8A, CD56 was higher in CXCL11-high group than CXCL11-low group (Figure 5b-5d) and the positive correlation between the expression of CXCL11 and PD-L1, CD8A, CD56 (r=0.62, p<0.001; r=0.34, p<0.001; r=0.32, p<0.001; respectively; Figure 5e-5g) was signi cant within YJSHC.

Discussion
In this study, we found that patients with high abundance of intratumoral CXCL11 + cells in ltration had better survival and CXCL11 was found to have independent prognostic value in colon cancer. Considering that CXCL11-producing cells were a variety of heterogeneous cells [17] and the intercellular communication mediated by CXCL11 was complicated, the discoveries make sense [26,35].
Furthermore, we found that CXCL11 positively correlated with CCL4, CCL5, CXCL9, CXCL10 in colon cancer, which were associated with DC, NK and T recruitment and may played an important role in inhibiting tumor growth and improving prognosis [36,37]. Consequently, we found that tumor with high CXCL11 mRNA expression and CXCL11 + cells had high abundance of anti-tumor CD8 + T cells, CD56 + NK cells in ltration [38,39], which was validated in TCGA and YJSHC. These data indicated that CXCL11 in colon cancer could function by mediating the in ltration of other immunocytes. Accordingly, the positive correlation between CXCL11 and cytotoxic genes (IFNG, GZMA, GZMB, GZMK and GZMM) [32][33][34] was discovered, which indicated that CXCL11 could enhance the cytotoxic function and contribute to the immunostimulatory microenvironment. However, it requires further investigation to discover the molecular mechanism underlining the correlation between CXCL11 and tumor immune microenvironment.
It is well known that ICIs (immune checkpoint inhibitors) therapy has been widely adopted in a variety of cancers [40,41]. Several studies have showed that tumor immune signatures could provide more information for clinicians to make appropriate choice of immunotherapy treatment [42][43][44]. In this study, we found that CXCL11 positively correlated with PD-L1 in TCGA and YJSHC, which was also veri ed in colon cancer single-cell RNA-seq dataset GSE146771 and GSE132465. The discovery was consistent with the initial study that PD-L1 blockade combined with an oncolytic vaccinia virus expressing CXCL11 in murine tumor models signi cantly shrinked tumor burden and improved prognosis [45]. Several studies have showed that massive intratumoral CD8 + T cells in ltration was able to alleviate the resistance to ICB (immune checkpoint blockade) in some way and intratumoral CD8 + T cells in ltration had the predictive value of therapeutic response to ICB treatment [41]. Nevertheless, prospective investigations need to be conducted to validate whether CXCL11 can predict the response to ICIs treatment in colon cancer. It's promising that increased expression of CXCL11 contributes to the anti-tumor immune microenvironment and sensitize patients who are resistant to anti-PD-1blockade treatment.
Somehow, our study also has some limitations. On the one hand, our analysis can only re ect certain aspects of tumor status and need to be veri ed in additional patient cohorts. On the other hand, in vitro and in vivo studies should be performed to discover the latent molecular mechanism.

Conclusions
In conclusion, CXCL11 was identi ed as a prognostic biomarker in patients with colon cancer in this study. This was the very rst study which revealed the clinical value and the correlation between CXCL11 and anti-tumor immune microenvironment in colon cancer. In addition, inducible expression of CXCL11 in colon cancer may improve the therapeutic response of patients receiving anti-PD-1 antibody treatment. No matter how, our ndings can provide some novel insights to assist clinicians in choosing appropriate drugs for their patients and improve the long-term outcome of patients with colon cancer.      The in ltration of various TILs and enriched biological pathways between high and low CXCL11 groups. The in ltration of various TILs identi ed by CIBERSORT between high and low CXCL11 groups in TCGA Association of CXCL11 with immune-related genes. The volcano plot showed the differentially expressed genes between high and low CXCL11 groups (a). Correlation between CXCL11 and indicated transcripts of DC and T cell-recruiting chemokines (b and c). Correlation between CXCL11 and cytotoxic molecules (d), immunosuppressive genes (e). The differential expression of CXCL11 in N (normal tissue), P (paracarcinoma tissue) and C (cancer tissue) of single-cell RNA-seq dataset GSE146771 (g) and GSE132465 (i). Spearman's correlation between CXCL11 and PD-L1 in single-cell RNA-seq dataset GSE146771 (f) and GSE132465 (h). Each dot represents one patient in GSE146771 and GSE132465.  Veri cation of the TILs between different groups of CXCL11+ cell in ltration by immunohistochemistry (IHC). Comparison of the in ltation of PD-L1+, CD8A+, CD56+ cells between high and low intratumoral CXCL11+ cells groups in YJSHC (a-d