Activated T regulatory (Treg) cells are potent suppressors that mediate immune tolerance. We investigated the relationship between activated Treg cells and the progression of human colon cancer. We designed a cross-sectional study of CD4+Foxp3+ T cells from peripheral blood, primary tumor and nontumor colon tissue of 42 patients with colon cancer and correlated the percentages of different subgroups of Treg cells with colon cancer stage. The phenotypes, cytokine-release patterns and suppression ability of these Treg cells were analyzed. We found that Treg cells increased significantly in both peripheral blood and cancer tissue. In addition, the Treg cells expressed significantly lower levels of CCR7, CD62L and CD45RA in comparison to normal volunteers. Further dividing Treg cells into subgroups based on Foxp3 and CD45RA expression revealed that both activated Treg cells (Foxp3hiCD45RA−) and nonsuppressive Treg cells (Foxp3loCD45RA−), but not resting Treg cells (Foxp3lowCD45RA+), increased in the peripheral blood and cancer tissue of patients with colon cancer. Only the activated Treg cells expressed significantly higher levels of tumor necrosis factor receptor 2 and cytotoxic T-cell antigen-4. Activated Treg cells, however, secreted significantly lower levels of effector cytokines (interleukin-2, tumor necrosis factor-α and interferon-γ) than did resting Treg cells and nonsuppressive cells upon ex vivo stimulation. Activated, but not resting, Treg cells in cancer tissue correlated with tumor metastases. In summary, we confirmed that activated Treg cells are a distinct subgroup with effector memory phenotype and fully functional regulatory activity against human colorectal cancer immunity.
CD4+ Foxp3+ Regulatory T cells (Treg cells) are essential to maintain self-tolerance and immune hemostasis.1, 2 They play important roles in a variety of human diseases including autoimmune disease, chronic infection and cancers.3–5 A large body of evidence suggests that Treg cells are one of the major players for tumor immune suppression and the main obstacle to successful tumor immunotherapy.3, 6–8 The prevalence of Treg cells significantly increases in the human peripheral blood and/or tumors of multiple types of cancers.9–12 These cells are recruited to tumor sites, where they suppress antitumor cytotoxic responses.3, 13–15 Animal models show that inhibiting Treg cells improves tumor immunity or enhances tumor vaccine therapeutic effects.16, 17 In human colon cancer, studies have revealed that CD4+CD25+Foxp3+ Treg cells are increased in peripheral blood mononuclear cells (PBMCs) and draining lymph nodes. These Treg cells were capable of suppression on antigen-specific CD4+ T cells. Surgical removal of colon cancer reduces the Treg cell population and restores antigen-specific CD4+ activity.18, 19 However, it is believed that nonspecific suppression of Treg cells might be dangerous, as it may produce an overwhelming autoimmune response.
Human CD4+Foxp3+ Treg cells are thought to be functionally heterogeneous. Human Foxp3+ T cells are not all suppressors. Allan et al. postulated that activated CD4+ T cells might express Foxp3 but without regulatory activity.20 CD4+Foxp3− T cells might transiently express lower levels of Foxp3 during activation or generation of pathogenic memory T cells.21 Furthermore, Foxp3+ Treg cells could be naïve or effector memory and have different functions. Effector memory Treg cells, which were discovered by Hamann and Shevach, express integrin CD103 surface molecules that guide the cells to an inflamed site on activation.22–24 This subgroup of Treg cells are activated Treg cells and accumulated in chronic inflammation,25, 26 graft-vs.-host disease27 and cancer in mice and humans.28, 29 Besides, chemokine/chemokine receptor interactions traffic leukocytes to nonlymphoid tissues and are a specific feature of activated Treg cells. Treg cells expressing a variety of chemokine receptors with tissue-trafficking ability are classified as distinct subtypes of activated Treg cells capable of executing immune-suppression functions.30, 31
Sakaguchi's group demonstrated that human CD4+Foxp3+ Treg cells can be grouped into three phenotypically and functionally distinct subsets: CD45RA+Foxp3lo resting Treg cells, CD45RA−Foxp3hi activated Treg cells and CD45RA−Foxp3 lo nonsuppressive T cells (or non-Treg cells).32 The former two subsets were suppressive in vitro and the last comprised cytokine-secreting nonsuppressive T cells. Activated Treg cells had the most potent suppressive activity and were increased in patients with sarcoidosis, a representative chronic inflammatory disease.32 Although they did not include human cancer patients in their analysis of Treg cell subsets, there is sufficient evidence to support the hypothesis that this subgroup of Treg cells should increase in human cancer.
Our murine colon cancer models and others have shown that Treg cells with the effector memory phenotype, the CD103+ Treg cells, accumulate in tumor sites and are responsible for the loss of immunosurveillance.28 The presence of tissue-homing molecules on tumor-infiltrating Treg cells of various types of human cancers were thought to be activated Treg cells that mediated tumor immunosuppression.31 In this article, we will demonstrate that CD4+Foxp3+ Treg cells in human colon cancer predominantly possess the CD45RA− phenotype. Activated Treg cells accumulate in tumor-infiltrating lymphocytes (TIL) and express suppressive phenotypes. The proportion of activated Treg cells correlates with tumor stage. Our finding confirmed that a specific subtype of effector memory-like Treg cells, like those discovered in the murine model, exists in human cancer. This subgroup of Treg cells could be a target for improving tumor immunotherapy.
Materials and Methods
Patients and blood/tissue samples
From May 2010 to November 2011, 42 patients with different stages of colorectal cancer (CRC) who received surgical treatment at Chang Gung Memorial Hospital, Linkou Medical Centre, were enrolled. During the same period, 19 healthy volunteers were also enrolled. PBMCs were isolated by the Ficoll/Paque™ PLUS density gradient centrifugation method. Thirty-one fresh tumor samples and nontumor tissues were collected during surgery and chopped into small pieces using a razor blade in roswell park memorial institute medium 1640 medium. The tumor tissues were removed from the main part of the tumor after gross pathological examination. Nontumor parts of the colon specimens were obtained from the distal ends of the surgical margin, which was proven free of cancer involvement by microscopic examination. These tissues were mixed with collagenase-D (0.1%; Gibco) in hank's buffered salt solution for 30 min at 37°C and then filtered through 70-μm nylon mesh. Single cell suspensions were separated with Ficoll (GE Healthcare), and leukocytes were recovered from the interphase.
Medical information was retrieved from the charts of CRC patients. Collected data included demographics, tumor pathology, stage, completed blood counts, biochemistries and level of carcinoembryonic antigen.
All patients and healthy donors provided written informed consent before enrollment. The study protocol conformed to the ethical guidelines of the 2008 Declaration of Helsinki and was approved by the Institution Review Board of Chang Gung Memorial Hospital.
Freshly obtained human lymphocytes were stained with anti-CD4 (-PerCP-Cy 5.5 from BD Bioscience or −APC from R&D Systems), anti-CD25 (-PE or −PECy5 from BD Bioscience) and anti-hCD45RA (-PE-Cy7 from BD Bioscience or −FITC Beckman Coulter). Intracellular detection of Foxp3 with anti-hFoxp3 (PE [BD biosciences]) was performed on fixed and permeabilized cells with the Foxp3 staining buffer set (e-Bioscience), according to the manufacturer's instructions. The following antibodies and reagents were used: CD62L (PE), CCR7 (PE), CD45RA (FITC or PE), cytotoxic T-cell antigen-4 (CTLA-4; APC), interleukin-2 (IL-2; FITC), tumor necrosis factor-α (TNF-α; FITC) and interferon-γ (IFN-γ; FITC or APC) obtained from BD and interleukin-17 (IL-17; FITC) mAbs from eBioscience.
Intracellular staining for IL-2, TNF-α, IL-17 and IFN-γ were performed as follows. Lymphocytes were freshly isolated and activated with phorbol myristate acetate/ionomycin for 5 and 1 hr with Golgi stop. Cells were stained for cell surface markers and then washed, fixed and permeabilized with the Foxp3 staining buffer set (e-Bioscience) for intracellular cytokine staining. BD fluorescence-activated cell sorting (FACS) Caliber was used to determine fluorescence intensity and Flowjo cytometry analysis software (Tree Star) was used for data analysis.
In vitro suppression assay of Treg cells
In suppression assay, 1 × 104 responder cells (CD25−CD45RA+CD4+ T cells) were labeled with 1 μM carboxyfluorescein diacetate succinimidyl ester and cocultured with unlabeled CD25loCD45RA+CD4+T cells, CD25hiCD45RA−CD4+T cells or CD25loCD45RA−CD4+ T cells assessed for their suppressive activity and CD3 and CD28 beads were used for stimulation. After cocultured for 86 hr, the cells were harvested and analyzed by FACS Caliber. median fluorescence intensities (MFI) were represented the suppressive ability of the each group of Treg cells.
Differences between groups were assessed using the Mann–Whitney U-test, t-test, paired t-test or Kruskal–Wallis test. The correlation between Treg cell population and tumor stage was determined by one-way analysis of variance. p Values were considered significant at p < 0.05 (*p < 0.05; **p < 0.01; ***p < 0.005). Statistical analyses were performed in Graph Pad Prism 5.0 software.
Increased percentage of CD4+Foxp3+ T cells with effector/memory phenotype in patients with CRC
Forty-two patients with colon cancer and 19 healthy volunteers were prospectively enrolled. Patient demographics are summarized in Table 1. There were 27 male and 15 female patients, with a mean age of 65 years. All patients were diagnosed with colorectal adenocarcinoma and had received surgical resections. None of the patients were associated with inflammatory bowel disease. Peripheral blood and colon tissues were obtained for analysis. As shown in Figures 1a and 1b, the percentages of CD4+Foxp3+Treg cells were increased in the PBMC of CRC patients (CRC vs. normal: 7 ± 3.2% vs. 1.97 ± 0.58%, p < 0.005). The percentage of Treg cells also increased in when compared to the nontumorous colon tissue in the same patients (TIL vs. nontumor tissue: 10.39 ± 4.97% vs. 4.7 ± 0.9%; p < 0.005).
Table 1. Patient demographics and tumor characteristics (N = 42)
In a murine colon cancer model, Treg cells in TIL predominantly exhibited the effector/memory phenotype. This subtype of Treg cells was CCR7−, CD62Llo and CD103+.28 Therefore, we analyzed the surface markers representing the effector/memory phenotype in human T cells. CCR7, CD62L and CD45RA expression decreased in the PBMC of patients with CRC (Fig. 1c). In patients with CRC, CCR7 expression further decreased in the TILs (median CRC PBMC 17.8 ± 9.9% vs. TIL 6.2 ± 3.5; p = 0.06). Similar phenomena were found for CD62Llo (median CRC PBMC 741.3 ± 181.8 vs. TIL 138.3 ± 29.6 MFI; p < 0.0001) and CD45RA− (median CRC PBMC 17.9 ± 3.4 vs. TIL 2.3 ± 1.8%; p < 0.0001) (Fig.1d). These findings were consistent with the murine model in which effector/memory Treg cells accumulated in TILs and PBMC, suggesting this may be an important feature of human tumor immune escape.
The CD4+Foxp3hi CD45RA− T cells were increased in PBMCs and preferentially accumulated in TILs of human colon cancer
Miyara et al. highlighted an important functional delineation of CD4+Foxp3+T cells based on the expression of CD45A and Foxp3. Their classification described three subtypes of CD4+Foxp3+T cells including resting Treg cells, activated Treg cells and nonsuppressive T cells. In chronic inflammatory disease, the activated Treg cells but not resting Treg cells were increased in PBMCs.32 As the tumor environment is considered an inflamed site33 and the Treg cells in CRC patients were predominantly devoid of CD45RA expression, we were prompted to investigate whether the activated Treg cell subgroup increased in patients with CRC. We found the percentages of resting Treg cells were similar in the peripheral blood of normal volunteers and CRC patients, as well as in TILs and nontumor tissue of CRC patients. Nonsuppressive T cells were higher in PBMCs and TILs of CRC patients than in normal sources in healthy donors (CRC PBMC vs. normal PBMC: 3.6 ± 2.1% vs. 1.4 ± 0.4%; p = 0.0002; TILs vs. nontumor tissue: 4.36 ± 2.4% vs. 1.85 ± 0.3%; p = 0.106), but were similar between PBMCs and TILs in CRC patients (CRC PBMC vs. CRC TIL: 3.6 ± 2.1% vs. 4.36 ± 2.4%; p = 0.057). The most striking difference was in the activated Treg cell population. It appears that TILs from CRC patients had the highest percentage of activated Treg cells (5.3 ± 2.95%) vs. PBMCs from CRC patients (2.2 ± 1.3%; p < 0.0001). The percentage of activated Treg cells from healthy volunteers and normal colon tissues was much lower (p < 0.001) (Fig. 2b). We then sorted Treg cells from PBMC of patients with colon cancer, according to the definition of Miyara et al. Activated Treg cells suppressed CD4+CD25−CD45RA+ responder T cell in vitro, while the rest of the Treg cell populations were significantly low (Fig. 2c). In summary, activated Treg cells increased in PBMC, the TIL of CRC patients and exerted a suppressive effect, as we observed in the murine colon cancer model.
Activated Treg cells exhibited higher expression of CTLA4 and TNFRII in CRC patients and did not secrete effector cytokines
We investigated the phenotypes of the different subgroups of CD4+Foxp3+T cells in patients with CRC. Tumor necrosis factor receptor type 2 (TNFRII) is an important cytokine receptor expressed by highly suppressive Treg cells in human and mouse.34 TNFRII was enriched in activated Treg cells in comparison to other subtypes of Treg cells in PBMC (Fig. 3a). Another major regulatory player is the CTLA-4 surface molecule.35 Upregulation of CTLA-4 is associated with increased Treg cell number and activity in inflammatory bowel disease models.36 CTLA-4 blockade could boost antitumor immune responses and is used in clinical therapy.37 We therefore examined CTLA-4 expression in patients with CRC and found that it was also significantly increased only in the activated Treg cell population (Fig. 3a).
We also studied the functional cytokine pattern in the CD4+Foxp3+ T-cell subgroups in PBMCs of patients with colon cancer after ex vivo stimulation. Activated Treg cells secreted very little IL-2, IFN-γ or TNF-α (p < 0.05). IL-17 was expressed in few Treg cells and was similar between subgroups (Fig. 3b). This feature suggests that activated Treg cells are suppressors. In contrast, CD4+Foxp3loCD45RA− nonsuppressive T cells secreted IL-2, IFN-γ and TNF-α and an insignificant amount of IL-17. This subgroup of Foxp3+ T cells, however, was suggested by Miyara et al. to be effector T cells.32 Thus, activated Treg cells were the most important subgroup of CD4+Foxp3+ T cells that expressed a tumor immune suppression phenotype in CRC patients.
The increased percentage of activated Treg cells correlated with tumor metastatic ability
The clinical impact of increased Treg cells on tumor stages was examined. In PBMCs, the mean percentage of CD4+Foxp3+ T cells was significantly higher in patients with lymph node and/or distant metastases than in those without metastases (p < 0.05) and healthy volunteers (p < 0.005). In TIL, the mean percentage of Treg cells in tumors with metastases tended to be higher; the difference was not significant, but was still significantly higher than that in nontumor tissue (p < 0.05) (Fig. 4a). We also questioned whether activated Treg cells correlated with tumor stage. The mean percentage of activated Treg cells in PBMC and TIL was higher in patients with CRC and even higher in patients with metastases (p < 0.05) (tumor without metastasis vs. tumor with metastasis; PBMC: 1.8 ± 1.2 vs. 2.5 ± 1.3; TIL: 3.8 ± 1.09 vs. 6.4 ± 3.4; p < 0.0001). Resting Treg cells did not differ between groups; while nonsuppressive Treg cells increased in PBMC of patients colon cancer (p < 0.0001) and TIL (p < 0.001), however, there was no difference between tumor with or without metastases (Fig. 4b). In conclusion, our data suggest an association between the accumulation of activated Treg cells at the tumor site in CRC patients and disease progression. These results imply that activated Treg cells are the major subgroup of CD4+Foxp3+ T cells responsible for the suppression of antitumor immunity.
This study highlights an important immunological feature of human tumor immunology. We carefully delineated the subgroups of CD4+ Treg cells in patients with CRC and found that CD4+ Treg cells were increased in CRC patients. Most of them were CD45RA−, suggesting an activated status, particular at the tumor site. The subgroup of Foxp3hi Treg cells increased at tumor sites and expressed high levels of TNFRII and CTLA-4 with decreased cytokine expression. This phenotypic expression profile suggests that this subgroup of Treg cells are fully immune suppressive. We also validated the finding from Sakaguchi's group,32 that the classification of CD4+Foxp3+ Treg cell subgroups is applicable in human cancer. Although this was a cross-sectional study aimed at the discovery of specific active subgroups of CD4+Foxp3+ Treg cells in human cancer, it provided invaluable evidence to support the notion that activated Treg cells in TIL (CD4+Foxp3hiCD45RA− T cells) express the effector memory phenotype and are responsible for tumor immunity escape, as we discovered in the murine colon model.28
Increased percentages of Treg cells in PBMCs have been reported in patients with almost every kind of human cancer.10, 11 Curiel et al. has reported increased Treg in PBMCs and malignant ascites associated with ovarian cancers.38 These cells express Foxp3 and inhibit tumor-associated antigen-specific CD8+ T cells ex vivo. Increases in the Treg population also predict poor prognosis in human cancers.15, 39 Increase in the number of Treg cells in patient's PMBCs after IL-2 administration has been associated with treatment failure.40 With the availability of Foxp3 immunohistochemical staining, accumulating evidence from frozen and paraffin-embedded specimens suggest the presence of Foxp3+ T cells in the tumor environment are associated with prognosis. The emerging data from human tumor studies, however, are controversial. Salama et al. suggested that better survival was associated with a high density of intratumor Foxp3+ Treg cells in CRC.41 Correale et al. reported a similar trend in patients with colon cancer receiving chemoimmunotherapy.42 Others have reported that increased intratumor Foxp3+ Tregs are associated with poor prognosis in gastric cancer,39 nonsmall cell lung cancer,43 thyroid cancer44 and even in colon cancers. For example, Sinicrope et al. suggested that a low CD3+/Foxp3+ ratio predicted a poor outcome.45 Our study and others may yield conflicting results because of the inconsistent subclassification of Treg cells.
Although Foxp3 is the master gene for CD4+ Treg-cell development and function, Foxp3 can also be transiently expressed in T-cell antigen receptor-activated human non-Treg cells.20 Recent data indicate that some nonlymphoid cells might express lower levels of Foxp3.46 Zhou et al. further suggested that activated CD4+ T cells transiently expressing Foxp3 might differentiate into memory T effector cells.21 Therefore, use of Foxp3 as the sole marker of Treg cells might lead to discordant conclusions. Further subgrouping to identify active Treg cells and their suppression mechanism in specific diseases is a newly active research field.
The Treg cell classification proposed by Miyara et al. has been proven relevant to immune suppression. Miyara et al. classified CD4+CD45RA−Foxp3lo T cells as non-Treg for their ability to secrete cytokines and elevated IL-17 expression. In their disease model, (sarcoidosis and systemic lupus erythematous), the percentage of activated and nonsuppressive Treg cells were reversed. In our observations of human cancer, nonsuppressive Treg cells increased in parallel with activated Treg cells and were enriched in CRC patient PBMCs. We have found that this subgroup of Treg cells did not exhibit suppressive function in vitro. We hypothesized that the CD45RA−Foxp3lo nonsuppressive Treg cells could be a heterogeneous population in human cancer. Many of them might be non-Treg and could differentiate into memory effector CD4+ T cells as others have proposed.21 The increased percentage of this subgroup could be the result of persistent antigen exposure in a tumor environment, like CD8+ T cells in human and murine models. In human renal cell carcinoma, Atting et al. found simultaneous infiltration of both polyfunctional effector and suppressor CD4+T cells.47 Further study of the function and differentiation of this distinct subgroup of CD4+ T cells in human cancer will be intriguing. In our opinion, investigators should carefully characterize Foxp3+ T-cell function and identify the active subgroup of Foxp3+ Treg cells rather than the whole CD4+Foxp3+ T-cell population in future human studies.
Our findings suggested that using Foxp3 and CD45RA to define a subgroup of CD4+ Treg cells in human CRC was appropriate and consistent with the findings of the Sakaguchi group. This marker is good for identifying human effector/memory Treg cells that fit the same population identified in the murine model. We found that the activated Treg subgroup is the principle Treg population in human tumor immune suppression. Furthermore, activated Treg cells in TIL were CD62Llo and expressed high levels of TNFRII and CTLA-4, but very few of them secreted effector cytokines, a feature of Treg cells. Chen et al. discovered that human and mouse effector Treg cells preferentially express TNFRII, which mediates the activating effect of TNF on Treg cells.34 This receptor is critical in the activation, proliferative expansion, and survival of Treg cells. They proposed that TNFRII is a key marker for functional CD4+Foxp3+ Treg cells.48 CTLA-4 is also a major effector marker for controlling Foxp3+ Treg cell function.49 Thus, the expression phenotype of activated Treg cells and the increased population of activated Treg cells in TIL suggest that activated Treg cell are the most potent suppressive CD4+ T cells in human colon cancer. Treg cells in human colon cancer might express IL17, and Treg might suppress Th17 in colon tumors.50 Our study showed that none of the Treg cell subsets expresses a significant level of IL-17. As our patients were patients with colon cancer devoid of inflammatory bowel disease, it is possible that the function of activated Treg cells in this context is distinct from its function in ulcerative colitis associated colon carcinoma. The diverse features of these cells in different tumor microenvironment will be the topic of future study.
Activated Treg cells were enriched in TIL and increased in advanced tumors. This finding was compatible with what we found in the murine model.28 Activated Treg cells accumulated at the tumor site to execute suppressing tumor immunity are central to tumor immune tolerance. An increase in the activated Treg cell population suggests loss of antitumor immunity. Although we cannot use this population to predict prognosis now, its correlation with tumor stage remains a valuable parameter for clinical utility. A longitudinal follow-up for this cohort of patients to validate the prognostic value of this subset of Treg cells is warranted. Our study also found that the total CD4+Foxp3+ T-cell population in PBMCs significantly correlated with tumor stage, but not in TIL. We feel, however, this definition for cancer-specific Treg cells should be used with caution due to the heterogeneity of the human Treg cell population, especially as the role of nonsuppressive Treg cells remains unclear. Misuse of this population may create contradictory results as described above. Finally, as the activated Treg cell population is the major suppressor in tumor immunity, targeting this subpopulation rather than whole Treg cell population may be a future method of tumor immunotherapy.
In summary, we confirmed that CD4+Foxp3hiCD45RA− is a distinct subgroup of Treg cells with an effector memory phenotype and fully functional regulatory activity against human CRC immunity. This cell population increases with advanced stage, implying that these Tregs are central to tumor immune suppression. Selective suppression of this subgroup of Treg cells would be an attractive strategy for future tumor immunotherapy.