Th17 cells inhibit CD8⁺ T cell migration by systematically downregulating CXCR3 expression via IL-17A/STAT3 in advanced-stage colorectal cancer patients

A total of 125 specimens (tumor and peripheral blood (PB) samples) were freshly obtained from the Department of Anorectal Surgery of the First Affiliated Hospital of Zhengzhou University (Zhengzhou, China). Healthy donors (HD, n = 50) were enrolled from the same hospital’s physical examinations center. Paraffin-embedded tissue samples from CRC patients (n = 75) diagnosed between 2011 and 2013 were obtained from the Pathology Department. All patients did not receive any therapeutic intervention such as chemo- or radiotherapy. All CRC patients were diagnosed histologically. Age- and sex-matched controls were selected and patients were staged according to the UICC-TNM classification. Early-stage patients included patients with stage I and II. Advanced-stage patients included patients with stages III and IV. The clinical data of the patients are shown in Table 1. Samples used in this study were approved by the Ethics Committee of the First Hospital of Zhengzhou University (approval number: Science-2010-LW-1213) and informed consent was obtained from each patient with available follow-up information.

Mononuclear cells from tumors and PB were isolated and stained for 30 min at 4 °C using saturating concentrations of the following antibodies: anti-human CD3, CD8, CD4, CD56, CXCR3, and 7-AAD antibodies and anti-mouse CD3, CD8, and CXCR3 antibodies (BioLegend, San Diego, CA). Intracellular staining of granzyme B, perforin, IFN-γ, and IL-10 was performed as previously described [32]. Cells were then analyzed by flow cytometry (BD FACSCanto II, BD Biosciences, Franklin Lakes, NJ).

Specimens from CRC patients and mice were formalin fixed, sectioned, and embedded into paraffin for immunohistochemistry. Immunofluorescence was performed with frozen samples and freshly fixed cells. The following antibodies were used: anti-CD8 (1:500), anti-CXCL10 (1:300), anti-CD4 (1:300), anti-IL-17A (1:300), and anti-CXCR3 (1:300; all Abcam, Cambridge, UK). Immunohistochemistry and immunofluorescence were performed as described elsewhere [33, 34]. For immunohistochemistry, three fields of view per sample were imaged. The intensity of immunostaining was considered when analyzing the data. The percentage scoring of immunoreactive cells was as follows: 0 (< 10%), 1 (10–40%), 2 (40–70%), and 3 (> 70%). Staining intensity was visually scored and stratified as follows: 0 (negative), 1 (yellowish), 2 (light brown), and 3 (dark brown). Immunoreactivity scores (IRS) were obtained by multiplying the two items to a total score and ranged from 0 to 9. Protein expression levels were further analyzed by classifying IRS values as low (based on an IRS value ≤ 5) and as high (based on an IRS value > 5).

Isolation of CD8⁺ T cells from HD samples was performed with CD8 magnetic beads (Miltenyi Biotech, Bergisch Gladbach, Germany) according to manufacturer’s instructions. Freshly isolated CD8⁺ T cells, anti-CXCR3 (10 ng/mL, Abcam), or Stattic (10 nM; S7024; Selleck Chem, Houston, TX)-treated CD8⁺ T cells as well as peripheral blood mononuclear cells (PBMCs) from HDs were seeded on the upper chambers of a 5.0-μm pore Transwell. Supernatants of tumor tissue, anti-CXCL10 (100 ng/mL, Abcam), or recombinant human (rh) CXCL10 (Peprotech, Rocky Hill, NJ) were added to the lower chambers. CD8⁺ T cells in the upper chambers were pretreated with rhIL-17A (20 ng/mL), Stattic (10 nM), and/or the supernatants of enriched Th17 cells (5 × 10⁶) with or without anti-IL-17A (10 ng/mL) for 48 h in vitro. Migrated cells in the lower chambers were then counted and analyzed by flow cytometry after 12 h.

Total protein was extracted after lysing cells in RIPA lysis buffer supplemented with protease inhibitor cocktail (P8340; Sigma-Aldrich, St. Louis, MO) and phosphatase inhibitor cocktail 2 (P5726, Sigma-Aldrich) as previously described [35]. The following primary antibodies were used: rabbit anti-phosphor-STAT3 (1:1000; Cell Signaling Technology, Danvers, MA), mouse anti-CXCR3 (1:1000, Cell Signaling Technology), and mouse anti-b-actin as control (1:3000, Cell Signaling Technology). Primary antibodies were detected with goat polyclonal rabbit or mouse IgG antibodies (1:1000, Cell Signaling Technology).

IL-17A concentrations in HD and CRC patient sera were measured using ELISA (R&D Systems Inc., Minneapolis, MN) according to manufacturer’s instructions.

Total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA), after which cDNA was reverse transcribed following manufacturer’s instructions. qRT-PCR was then performed using SYBR Green, and GAPDH was used as an internal control. Relative expression levels were determined using the 2^−ΔΔCt method. Primers used are listed in Supplementary Table 1.

Cytokine levels in the sera of CRC patients and supernatants from tumor tissues and paired normal tissues were analyzed using a Multi-Analyte Flow Assay Kit (BioLegend) that includes 13 human cytokines according to manufacturer’s instructions.

CD8⁺ T cells sorted from PBMCs of HDs were treated with 20 ng/mL rhIL-17A. Nuclear proteins were isolated with a Nuclear Extraction Kit (SK-0001; Signosis, Santa Clara, CA) and analyzed using a 96-well plate Transcription Factor (TF) Activation Profiling Array (FA1002; Signosis) according to manufacturer’s instructions. Quantification of each transcriptional factor was normalized as the fold change of TFIID activity.

CXCR3 upstream signaling was screened using radar tools of the Gene-Cloud of Biotechnology Information (GCBI) database (www.​gcbi.​com.​cn).

Human CD8 magnetic beads (Miltenyi Biotec) were used for isolating CD8⁺ T cells from PBMCs according to manufacturer’s instructions. CXCR3⁺CD8⁺ cells and CXCR3^-CD8⁺ cells were sorted from PBMCs using the MoFlo XDP cytometer (Beckman Coulter, Brea, CA). The positive rate of cells after purification was more than 90%.

Th17 cell culture and enrichment were carried out as described previously [36]. In brief, CD4⁺ memory T cells were isolated from freshly obtained PBMC using Memory CD4⁺ T Cell Isolation Kit (Miltenyi Biotech). The CD4⁺ memory T cells were treated with anti-CD3/CD28/CD2 beads for 7 d in the presence of recombinant IL-6 (40 ng/mL), IL-1b (10 ng/mL), IL-23 (50 ng/mL), neutralizing anti-IL-4 (0.5 mg/mL), and anti-IFN-g (5 mg/mL). Then cultured Th17 cells were enriched using an IL-17 Secretion Assay-Cell Enrichment and Detection Kit (130-094-542, Miltenyi Biotech). Cultured Th17 cells were further stimulated with cytostim (20 μL/mL) for 6 h at 37 °C. Then, the cells were labeled with IL-17 catch reagent, followed by incubation in closed tubes for 45 min at 37 °C under slow continuous rotation. The cells were then labeled with IL-17 detection antibody (PE) and anti-PE microbeads, followed by cell collection using magnetic separation with MS columns according to manufacturer’s instructions.

To generate a xenograft mouse model, 6-week-old female BALB/c nude mice were purchased from Vital River Laboratory Animal Technology Co. Ltd (Beijing, China) and randomly divided into groups. Luc-GFP-HCT116 cells (3 × 10⁶) were injected subcutaneously. 1 × 10⁶ CD8⁺ T cells were co-cultured with rhIL-17A (20 ng/mL) and/or Stattic (10 nM) for 48 h in vitro. Seven days after tumor cell implantation, pretreated CD8⁺ T cells were injected intravenously. After 48 h, CD8⁺ T cell infiltration in tumor tissues was examined via immunohistochemistry. To further evaluate the tumor growth, on day 0, CD8⁺ T cells, Th17 cells, or CD8⁺ T cells plus Th17 cells were injected intravenously. Next, 3.75 mg/kg Stattic was administrated to mice intraperitoneally every day for 1 week [37]. Tumorigenicity was evaluated using an in vivo imaging system (IVIS Lumina Series III; PerkinElmer, Waltham, MA) every 6 days. Mice were then sacrificed, after which tumors were harvested and blood collected from the tail vein for further analysis. All animal studies were approved by the Institutional Animal Care and Use committee of the First Affiliated Hospital of Zhengzhou University.

Data are expressed as the mean ± standard deviation. Differences among groups were compared using Student’s t test, chi-square test, and one-way ANOVA. OS curves were plotted according to the Kaplan-Meier method. Correlation between two variables was analyzed by Spearman’s rank-order correlation. Statistical analyses were performed using the GraphPad Prism 5 software (GraphPad Software Inc., La Jolla, CA). P values < 0.05 were considered statistically significant.

As tumor-infiltrating CD8⁺ T cells are indicators of an active host immune response against cancer [4], we quantified the infiltrating CD8⁺ T cells in tumor tissues of early- and advanced-stage CRC patients. CD8⁺ T cell density was found lower in advanced-stage tumor tissues compared with early-stage tumor tissues, and high expression of CD8 was associated with a favorable prognosis (Fig. 1a–c). Given that T cell infiltration of tumors is a multi-step process that is mediated, in part, by chemokine-chemokine receptor pathways [38], we examined the potential chemokines contributing to T cell infiltration and found that CXCL10 expression were significantly increased in advanced-stage tumor tissue compared with early-stage tumor tissues. Other chemokines exhibited no significant difference in their expression levels, which were also inconsistent with CD8⁺ T cell infiltration patterns (Fig. 1d, e). These findings were similar at the protein level by IHC (Fig. 1f). The staining results showed that CXCL10 was predominately produced from tumor cells rather than from stroma cells (Fig. 1f). We then utilized Transwell migration assays to test the role of CXCL10 in cell recruitment and found that CD8⁺ T cell migration increased significantly after CXCL10 treatment; in contrast, CD4⁺ T cells and NK cell migration remained unchanged (Fig. 1g). Next, CD8⁺ T cells were isolated from freshly obtained HD PBMCs using magnetically activated cell sorting (MASC) and the purity obtained was greater than 90% (data not shown). CD8⁺ T cell movement was also markedly enhanced along with increasing concentrations of rhCXCL10 (Fig. 1h). We further investigated the regulatory effects of CXCL10/CXCR3 on CD8⁺ T cell migration using supernatants from freshly obtained CRC tumor tissues. Supernatants derived from tumor tissues markedly facilitated CD8⁺ T cell migration, which was attenuated by anti-CXCL10 or anti-CXCR3 neutralizing antibodies. These findings indicate that CXCL10 attracts CD8⁺ T cell migration. This attraction was dependent on CXCR3, as a decreasing ratio of CXCR3⁺CD8⁺ cells to CXCR3^-CD8⁺ cells led to a significantly reduced chemotactic ability of CD8⁺ T cells in the presence of the same concentrations of rhCXCL10. When the ratio decreased to 1:8, increased concentrations of rhCXCL10 also did not affect the chemotactic ability of CD8⁺ T cells (Supplementary Figure 1). Thus, we theorized that CXCR3 expression on CD8⁺ T cells plays an important role in the degree of CD8⁺ T cell infiltration in tumor tissues of CRC patients.

Previous studies have shown that CXCR3, the only receptor for CXCL10, facilitates CD8⁺ T cell recruitment in inflammatory and malignant diseases [36, 39]; thus, we examined CXCR3 expression on CD8⁺ T cells in PB using flow cytometry. Compared with CD8⁺ T cells from HDs, CXCR3 expression was found decreased in CRC patients (Fig. 2a). Interestingly, the percentage of CXCR3⁺CD8⁺ T cells was also lower in advanced-stage CRC compared with early stage, which was consistent with CD8⁺ T cell levels (Fig. 2b); a similar result was observed in CRC patients with distant metastasis compared with those without (Fig. 2c). CXCR3 expression levels were also found lower in tumor tissues with advanced-stage compared with early-stage CRC (Fig. 2d). To assess the function of CXCR3⁺CD8⁺ T cells, we identified the intracellular factors and found that the percentages of IFN-g granzyme B, and perforin were significantly higher and IL-10 levels were decreased in the CXCR3⁺CD8⁺ T cell subset compared with the CXCR3^-CD8⁺ T cell subset (Fig. 2e). Taken together, these findings exhibited that the expression level of CXCR3 on CD8⁺ T cells was decreased in advanced-stage CRC PB.

CXCR3 expression was reported to be regulated by several inflammatory factors, including IL-10, IL-4, and IL-15 [40‐42]. To identify which factor(s) participate in the downregulation of CXCR3 on CD8⁺ T cells during CRC malignancy, we tested 13 inflammatory cytokines using a multiplex array in the following three groups: sera of CRC patients vs. HDs, sera of advanced-stage patients vs. early-stage patients, and supernatants of tumor tissues vs. paired normal tissues (Fig. 3a). We identified IL-17A, IL-17F, and IFN-g as significantly upregulated cytokines in the former of all the three groups (Fig. 3b). To further test which factors play important roles in regulating CXCR3 expression, we evaluated CXCR3 expression levels on CD8⁺ T cells after treatment with three recombinant proteins of IL-17A, IL-17F, and IFN-g and found that IL-17A markedly downregulated CXCR3 expression, whereas IL-17F and IFN-g did not lead to marked downregulation (Fig. 3c). Therefore, we decided to focus on the effects of IL-17A on CXCR3 expression regulation. Via immunofluorescence, we found that the fluorescence intensity and positive rate of CXCR3 expression can be inhibited by IL-17A (Fig. 3d). The results of ELISA further verified that IL-17A expression in sera was higher of CRC patients than of HDs and it was also higher in advanced-stage patients than in early-stage patients (Fig. 3e, f). Moreover, correlation analysis showed that IL-17A expression levels were negatively associated with the percentage of the CXCR3⁺CD8⁺T cells in PB (Fig. 3g). To verify whether IL-17A influenced the expression of CXCR3 ligands, we stimulated HCT116 cells with rhIL-17A. The results showed that CXCR3 ligands (CXCL9,10,11) were not changed significantly (Supplementary Figure 2). These results indicate that IL-17A plays a major role in downregulating CXCR3 expression on CD8⁺ T cells in advanced-stage CRC.

To explore the mechanisms underlying IL-17A regulation of CXCR3 expression on CD8⁺ T cells, we performed a multiplex screening assay for the DNA-binding activity of 96 transcriptional factors in IL-17A-treated CD8⁺ T cells sorted from the PBMCs of HDs. IL-17A treatment activated several transcriptional factors, including Pit, STAT3, and HoxA-5, as evidenced by a > 4-fold increase in their activity (Fig. 4a). To further examine which transcriptional factors are involved in regulating CXCR3, we used gene radar prediction from the GCBI database to search for upstream molecules of CXCR3, and 96 transcriptional factors were found (Fig. 4b). Based on the above results, STAT3 was identified as the key transcriptional factor of IL-17A regulating CXCR3 on CD8⁺ cells (Fig. 4c). To test this, we assessed CXCR3 levels on CD8⁺ T cells after treatment with Stattic, a STAT3 inhibitor. After incubating PBMCs with Stattic for 24 h and 48 h, the percentage of CXCR3⁺CD8⁺ T cells increased (Fig. 4d). Similar results were also obtained with purified CD8⁺ T cells sorted from PBMCs of HDs (Fig. 4e). Next, we evaluated whether inhibition of STAT3 signaling can reverse the IL-17A-induced inhibition of CXCR3 expression and found that this inhibitory effect was attenuated in CD8⁺ T cells treated with Stattic (Fig. 4f). In addition, western blot analysis showed that STAT3 phosphorylation increased, while CXCR3 levels decreased in a concentration-dependent manner after treatment with rhIL-17A (Fig. 4g). Consistent with previous results, the effect of rhIL-17A was markedly attenuated upon treatment with Stattic (Fig. 4h). Moreover, the percentage of P-STAT3⁺CD8⁺ T cells in total CD8⁺ T cells was found increased in PB of CRC patients vs. HDs and advanced-stage patients vs. early-stage patients (Supplementary Figure 3). Overall, our findings indicate that IL-17A inhibits CXCR3 expression through STAT3 signaling.

We further co-cultured sorted CD8⁺ T cells with sera collected from advanced-stage CRC patients and found that the percentage of CXCR3⁺CD8⁺ T cells was reduced; this percentage was upregulated after treatment with anti-IL-17A or Stattic (Fig. 4i). These results support the notion that CXCR3 expression on CD8⁺ T cells is inhibited by IL-17A and STAT3 signaling in the PB of advanced-stage CRC patients. Finally, Transwell migration assays demonstrated that IL-17A suppressed CD8⁺ T cell recruitment, which was rescued by the addition of Stattic (Fig. 4 j). Taken together, our findings indicate that STAT3 signaling is critical for the IL-17A-mediated regulation of CXCR3 expression.

We performed immunofluorescence assays to investigate which cells secrete IL-17A and found that IL-17A was predominately produced from CD4⁺ T cells rather than from CD8⁺ T cells (Supplementary Figure 4). Thus, we examined the expression of Th17 cells in PB and found increased levels in CRC patients compared with HDs (Fig. 5a) that were especially higher in advanced-stage than early-stage patients (Fig. 5b). We next induced and enriched Th17 cells in vitro and the purity was more than 70% (Supplementary Figure 5). Using multiplex arrays, we identified that IL-17A was the most secreted cytokine among the 13 inflammatory cytokines (Fig. 5c). Furthermore, most enriched CD4⁺ T cells were found co-localized with IL-17A via immunofluorescence (Fig. 5d). In addition, CD8⁺ T cells treated with conditional medium collected from Th17 cells exhibited a decreased percentage of CXCR3⁺CD8⁺ T cells, which was rescued after adding anti-IL-17A or Stattic (Fig. 5e). The same trend was found for the migratory ability of CD8⁺ T cells (Fig. 5f). These findings indicate that Th17 cells inhibit the recruitment of CD8⁺ T cells via IL-17A/STAT3 signaling.

To further examine the role of IL-17A/STAT3 signaling on CD8⁺ T cell migration in vivo, we utilized a xenograft mouse model of CRC. Tumor-bearing mice were intravenously injected with purified CD8⁺ T cells, which were pretreated with rhIL-17A or Stattic for 48 h in vitro (Fig. 6a). After 48 h, CD8⁺ T cell infiltration in tumor tissues was examined via immunohistochemistry. IL-17A suppressed the infiltration of CD8⁺ T cells in tumor tissues, which was attenuated by Stattic (Fig. 6b, c). To further examine whether blockade of STAT3 signaling can decrease tumor progression by rescuing the Th17-induced inhibition of CD8⁺ T cell infiltration in tumor tissues, Stattic was administered by intraperitoneal injection to Luc-GFP-HCT116-tumor bearing mice (Fig. 6d). CD8⁺ T cell infusion suppressed tumor growth; this effect was significantly diminished when CD8⁺ T cells were injected with Th17 cells (Fig. 6e, f). Consistently, tumor growth was significantly suppressed after Stattic was injected (Fig. 6e, f). Notably, we found that Th17 cells decreased CXCR3 expression levels on CD8⁺ T cells in PB, which was rescued by Stattic (Fig. 6g); CXCR3⁺CD8⁺ T cell infiltration in tumor tissues was consistent with these results (Fig. 6h). Moreover, flow cytometry and immunofluorescence indicated that CD8⁺ T cells were expressed at lower levels in the co-injection group of CD8⁺ T cells and Th17 cells and elevated in the Stattic group (Fig. 6i, j). These findings indicate that Th17 cells promote CRC tumor progression by partly blocking CXCR3⁺CD8⁺ T cell homing to tumor tissues via STAT3 signaling. Thus, STAT3 signaling may serve as a potential therapeutic target for CRC treatment.

To investigate the clinical significance of Th17 cells and CXCR3⁺CD8⁺ cells in CRC patients, tumor tissues were collected. The results showed that expression levels of Th17 cells were significantly higher in advanced-stage than in early-stage tumor tissues, whereas the levels of CXCR3⁺CD8⁺ cells were lower (Fig. 7a). Previous studies have shown that CXCR3 is expressed on both tumor cells and stromal cells [43, 44]; therefore, we analyzed the distribution of CXCR3-positive cells in CRC tissues via immunofluorescence and found that cells highly expressing CXCR3 were predominantly located in stromal cells rather than in tumor cells (Supplementary Figure 6A, B). In the stroma, cells highly expressing CXCR3 were strongly co-localized with CD8, indicating that CXCR3 is highly expressed on CD8⁺ T cells and may be associated with the antitumor immune response (Supplementary Figure 6C). Given that IL-17A-positive cells were mainly derived from the CD4⁺ cell population, Th17 cells may be the primary source of IL-17A in tumor tissues (Supplementary Figure 6D). The expression levels of Th17 cells were negatively associated with the levels of CXCR3⁺CD8⁺ cells using correlation analysis (Fig. 7b). In the CRC tissue sections, a positive correlation between CXCR3 and CD8 was observed (Fig. 7c), while a negative correlation between IL-17A with CXCR3 and CD8 was identified (Fig. 7c). According to the IRS and positive rate, CRC patients with high levels of IL-17A in tumor tissues had a poor prognosis. In patients with higher CXCR3 expression, OS was significantly improved (Fig. 7d). According to the expression levels of CD8, CXCR3, and IL-17A, we divided all patients into different groups and performed survival analysis, which indicated that the following four groups have better prognosis: high CXCR3 and CD8 expression group, high CD8 and low IL-17A expression group, high CXCR3 and low IL-17A expression group, and high CD8 and CXCR3 and low IL-17A expression group (Fig. 7e). These findings further support the correlation between IL-17A, CXCR3, and CD8 in CRC tumor tissues.

Overall, our findings support a model wherein Th17 cells inhibit the expression of CXCR3 on CD8⁺ T cells in blood circulation by secreting IL-17A, which activates STAT3 signaling and downregulates CD8⁺ T cell infiltration in tumor tissues; these effects can be attenuated via Stattic (Fig. 7f).