|Home | About | Journals | Submit | Contact Us | Français|
T cells play a crucial role in preventing the growth and spread of colorectal cancer (CRC). However, immunotherapies against CRC have only shown limited success, which may be due to lack of understanding about the effect of the local tumor microenvironment (TME) on T cell function. The goal of this study was to determine whether T cells in tumor tissue were functionally impaired compared to T cells in non-tumor bowel (NTB) tissue from the same patients. We showed that T cell populations are affected differently by the TME. In the tumor, T cells produced more IL-17 and less IL-2 per cell than their counterparts from NTB tissue. T cells from tumor tissue also had impaired proliferative ability compared to T cells in NTB tissue. This impairment was not related to the frequency of IL-2 producing T cells or regulatory T cells, but T cells from the TME had a higher co-expression of inhibitory receptors than T cells from NTB. Overall, our data indicate that T cells in tumor tissue are functionally altered by the CRC TME, which is likely due to cell intrinsic factors. The TME is therefore an important consideration in predicting the effect of immune modulatory therapies.
Colorectal cancer (CRC) is the third most commonly diagnosed cancer, and causes the fourth most cancer-related deaths, worldwide.1 The immune system, particularly T cells, plays an important role in preventing the spread and growth of CRC.2,3 The Immunoscore, a score based on the number of tumor infiltrating CD3+ and CD8+ T cells, has been shown to be more accurate for predicting patient prognosis than traditionally used staging methods.4
Despite the value of measuring the immune response for predicting patient prognosis, immunotherapies such as cancer vaccines, adoptive cell therapy and check point blockade have only shown modest success against CRC.5 The lack of success may be due to limited understanding of the level and types of immunosuppression that occur within CRC tumors and how the tumor microenvironment (TME) differs from the natural immunosuppressive environment of the gut. T cells in the gut are kept under tight control to prevent unwanted immune responses against microflora and food antigens.6 Tumors also modulate the immune system in ways that attenuates the effectiveness of T-cell-mediated immune responses.7
Immune suppression mechanisms in CRC tumor tissue have been studied previously8; however, there have been very few studies that have directly investigated how the function of infiltrating T cells in CRC tumors differs from that of T cells within the unaffected bowel. We and others have shown a higher frequency of IL-17-producing T cell populations, including MAIT cells and γδ T cells, in tumor versus adjacent bowel tissue.9-14 Differences in the frequency of regulatory T cells (Tregs) have also been observed,10,11 but the effect of the function of these T cell populations on other infiltrating T cells is not clear. We hypothesized that T cells in tumor tissue would have impaired cytokine production and proliferative ability due to the suppressive environment in CRC tumors.
We validate, in a large cohort of patients, the finding that T cell populations are present in tumor tissue at different frequencies than in non-tumor bowel (NTB) of the same patients, including a novel population of T cells co-expressing IL-17 and IL-2. Further, we show that T cells in tumor tissue produce different amounts of cytokines per cell than those from NTB. Finally, the T cells infiltrating the tumor have an impaired proliferative ability compared to T cells from NTB. Interestingly, this impairment is not related to the frequency of IL-2 producing T cells or Tregs, but to a higher expression of inhibitory receptors.
The frequencies of regulatory and inflammatory T cells are elevated, and IL-2 producing T cells are decreased, in CRC tumor tissue compared to NTB tissue from the same patients.
To examine the influence of the CRC TME on T cell infiltrate, we compared the phenotype of infiltrating T cells in NTB tissue to those from CRC tumor tissue from the same patients using flow cytometry. Regulatory (CD25hiFOXP3+), IFNγ producing (IFNγ+), inflammatory (IL-17+) and IL-2 producing (IL-2+) T cells were identified within CD3+CD4+ and CD3+CD8+ populations. The gating strategy for identifying the different T cell subsets is displayed in Fig. 1A. The frequencies of the different subsets were compared between tumor tissue and matched NTB from the same patient, which allowed confounding factors such as age, sex, diet and host genetics to be controlled.
The frequencies of both CD4+ regulatory and CD4+ inflammatory T cells were elevated in tumor tissue compared to matched NTB tissue (median frequency increase of 2.678% and 1.229%, p < 0.001 and p < 0.0001, respectively; Fig. 1B). The frequencies of CD4+ and CD8+ IL-2-producing T cells were decreased in tumor tissue (median frequency decrease of 4.3% and 4.95%, respectively, p < 0.05; Fig. 1B, C). No differences in frequency were seen in patients with different stages of disease. Together, these data confirmed our previous published finding that CRC tumor tissue has a distinct T cell infiltrate compared to matched NTB tissue,11 and further characterized the TME with a higher frequency of CD4+ regulatory and inflammatory T cells and a lower frequency of IL-2-producing T cells.
Observations from cancer vaccine trials in mice and humans have shown that polycytokine-producing T cells correlate better with immune protection that monocytokine-producing T cells.15 IL-2 and IFNγ co-producing T cells are particularly important in tumor protection and therefore have enhanced antitumor functionality.16 A Boolean gating strategy was used to analyze the difference in frequencies of multicytokine-producing T cells between tumor tissue and NTB tissue. This strategy compared all possible combinations of cytokine-producing T cells. There was a significantly lower frequency of CD4+ IFNγ+IL-2+IL-17− T cells in tumor tissue compared to matched NTB tissue (Fig. 2B, median frequency decrease of 4.23%). Conversely, there was a higher frequency of CD4+ IFNγ+IL-17+IL-2− dual producing T cells in tumor tissue compared to matched NTB tissue (Fig. 2D, median frequency increase of 1.76%). This T cell subset has been described previously as playing an important role mediating autoimmune disease.17 These data demonstrate distinct differences in frequencies of multi-functional T cells between tumor and NTB tissue.
To further investigate potential functional differences in T cells between NTB and tumor tissue, the difference in median fluorescent intensity (MFI) of FOXP3, IFNγ, IL-17 and IL-2 between T cells in NTB and tumor tissue was investigated. The MFI of FOXP3 expression in Tregs has been shown to relate to suppressive ability.18 MFI of cytokine producing cells can be used to compare the amount of cytokine produced by individual cells.
There was no significant difference in the MFI of FOXP3 in Tregs between matched tumor and NTB tissue samples (Fig. S2A). There was also no difference in the MFI of IFNγ+ cells between NTB and tumor tissue (Fig. S2B). These data imply that there are no functional differences in these T cell populations between NTB and tumor.
The MFI of IL-17+ T cells (Fig. 3A) was significantly higher and the MFI of IL-2+ T cells (Fig. 3C) was significantly lower in tumor tissue compared to NTB tissue (Fig. 3B, D). These data demonstrate that different T cell subsets can be functionally impaired or enhanced by the TME.
T cell proliferation is an important in vitro measurement of in vivo immune function. In CRC, a high frequency of proliferating T cells in the tumor has been linked with improved disease-free survival.19 We assessed the ability of T cells from NTB and tumor to proliferate in response to a polyclonal stimulus. T cells from tumor tissue and NTB tissue were isolated and stimulated with anti-CD3/CD28 beads for 3 d and the frequency of Ki67+ cells measured by flow cytometry. T cells from tumor tissue displayed impaired proliferative ability compared to T cells from NTB tissue (Fig. 4A), which demonstrates that T cells in tumor tissue are functionally impaired compared to T cells in NTB tissue.
It was hypothesized that the decreased proliferation could be due to the lower frequency of IL-2-producing T cells or the higher frequency of Tregs in tumor tissue compared to NTB tissue (Fig. 1B). However, after 72 h of stimulation there was no difference in the frequency (Fig. 4B) or MFI (Fig. 4C) of IL-2-producing T cells between the tumor and NTB tissue. Exclusion of Tregs by fluorescence-activated cell sorting (FACS) did not significantly alter the frequency of Ki67+ T cells from NTB or tumor tissue (Fig. 4D, E).
Overall, these data further demonstrate that tumor-infiltrating T cells are functionally impaired compared to T cells from matched NTB tissue.
Discovery of the mechanism(s) that cause the functional impairment of tumor-infiltrating T cells could provide information about how CRC tumors suppress immune responses in patients. A cardinal feature of T cell exhaustion is the expression of inhibitory receptors including CTLA-4, LAG-3, PD-1 and Tim-3.
There was a significantly higher frequency of T cells expressing PD-1 in tumor tissue compared with those from matched NTB tissue (Fig. 5A). There was no difference in the frequency of cells expressing other inhibitory receptors. There was also no significant difference in the level of expression of any of the inhibitory receptors as measured by MFI between tumor and matched NTB (Fig. 5B). However, a decrease in MFI of CTLA-4+ T cells in tumor compared with NTB was close to significance (p = 0.0542, Fig. 5B).
Boolean gating was used to quantify T cells expressing more than one receptor, as it has been shown that T cells expressing multiple inhibitory receptors are more exhausted.20 We found the following significant differences between the tissue types: (i) an increase in the frequency of LAG-3+PD-1+Tim-3−CTLA-4− T cells and (ii) a decrease in frequency of T cells expressing no inhibitory receptors, in tumor compared to matched NTB (Fig. 5C). These data showed that a higher frequency of T cells from the tumor have an exhausted phenotype compared to T cells from matched non-tumor bowel.
T cells play a crucial role in preventing the growth and spread of CRC.21 The microenvironment of a tumor can result in functionally impaired T cells, which prevent tumor destruction.22 In the current study, we have demonstrated functional impairments in the cytokine production and proliferation of T cells infiltrating tumor in CRC patients and shown that this functional impairment may be related to T cell exhaustion.
T cells in the peripheral blood, tumor-draining lymph nodes and tumor tissue of CRC patients have been shown to have an exhausted phenotype,20,23,24 including a higher frequency of PD-1+ T cells in tumor compared to adjacent uninvolved bowel.10 However, analysis of multiple inhibitory receptors on T cells from tumor compared to T cells from NTB has not previously been published. We showed a higher frequency of PD-1+LAG-3+ double positive T cells in tumor compared to NTB, although these data should be replicated in a larger cohort of patients. LAG-3 binds to MHC Class II, a molecule shown to be expressed on CRC cells.25 Given that blockade of LAG-3 increases T cell cytokine production and proliferation,26 it may be possible that the higher expression of this molecule influences both the cytokine phenotype and limited proliferative capacity of the T cells as described here. T cells expressing PD-1 are susceptible to negative regulation by tumors (as well as Tregs) via PD-L1 – a molecule expressed on CRC tumors.27 The co-expression of LAG-3 may determine the nature of the regulation, that is, via inhibition of proliferation and cytokine production. Our data support this idea – there was no difference in the frequency of LAG-3+ T cells in tumor compared to NTB, but a difference was seen in T cells co-expressing both molecules.
Inhibitory receptors that are expressed at high frequency in NTB tissue may play an important role in maintaining immune homeostasis in the gut, while inhibitory receptors expressed on immune cells in tumor tissue may prevent immune eradication of the tumor. Therefore, comparison of T cells from NTB and tumor tissue may have identified inhibitory receptors that inhibit the antitumor immunity but contribute less to peripheral tolerance in the gut. This may help to explain why anti-CTLA-4 therapy may not be effective in patients with CRC, whereas PD-1 blockade may be effective, as reported in a recent clinical trial.26,27 The expression of PD-L1 on tumor cells as well as other infiltrating immune cells such as Tregs and macrophages may also determine the effect of T cells expressing inhibitory receptors on cancer growth.28 PD-1 blockade is more effective in people with microsatellite instability (MSI)-high CRC.29 MSI data were unavailable for our cohort, so comparisons with T cell expression of inhibitory receptors in MSI-high or MSI-low patients could not be performed. However, with methodology established to measure proliferation of T cells ex vivo from NTB and tumor tissue, potential therapies can be tested on clinical samples easily and safely. In particular a combination of LAG-3 and PD-1 blockade would be an intriguing combination.26
To our knowledge, this is the first time that T cells co-producing IFNγ and IL-2 have been investigated in CRC tumors, although co-expression of other cytokines have been studied.10,11 We showed these IFNγ+IL-2+ T cells to be at a lower frequency in CRC tumors than matched NTB tissue. The dual production of IFNγ and IL-2 in T cells has previously been described as an important anticancer T cell subset.16,30 These cells correlate strongly with immune protection against tumors in preclinical tumor models and human vaccine studies.16,30 Dual IFNγ and IL-2-producing T cells produce more IFNγ on a per cell basis and are more likely to become memory cells than single IFNγ producing T cells.16,31,32 These attributes are beneficial to T cells that destroy tumor cells. In mice, dual IFNγ and IL-2-producing T cells require a greater amount of T cell receptor (TCR stimulation than mono IFNγ producing T cells.33 TCR specificity varies widely between T cells in NTB tissue and CRC tumor tissue34; potentially T cells in NTB tissue could bind with higher affinity to their target antigen and therefore are more likely to be dual IFNγ and IL-2 producers than T cells located in the tumor.
We also found higher frequencies of IL-17+ inflammatory T cells and more IL-17 secreted per inflammatory T cell in tumor tissue compared to NTB tissue. While higher frequencies of IL-17-producing cells have been reported in colorectal tumors,9-12,14 to our knowledge this is the first time that enhanced cytokine production per cell by inflammatory T cells infiltrating the CRC TME has been reported. The TME has high levels of cytokines that boost inflammatory T cell differentiation and function such as IL-6.35,36 IL-17 can promote tumor cell proliferation, survival and metastasis, and therefore the tumor is likely enhancing the function of T cells that promote tumor growth.37 As well as promoting tumor growth, high levels of IL-17 can also inhibit immune function by reducing the expression of antitumor cytokines,38 which likely makes enhanced inflammatory T cells function to be beneficial to CRC tumors. In support, T cells expressing inflammatory genes (IL17A, RORC) have been linked with poor patient prognosis in CRC.39
We showed that T cells infiltrating tumor tissue have impaired proliferation compared to T cells recovered from NTB tissue. In cancer patients, peripheral blood mononuclear cells (PBMCs) with higher proliferative proficiency have been linked with longer survival.40 Specifically, in CRC, a high frequency of proliferating T cells in tumors has been linked with good prognosis.19 Impaired proliferation of tumor-infiltrating T cells has also been linked to the level of IL-15 in CRC tumors.19 IL-15 is an important cytokine for T cell proliferation that is mainly produced by monocytes and dendritic cells.41 In our study, impaired proliferation of tumor-infiltrating T cells was observed, once removed from the main sources of IL-15. This finding demonstrates that IL-15 may not be solely responsible for the impaired proliferation of tumor-infiltrating T cells. The data from this study showed that impaired proliferation of T cells from tumor tissue was likely due to intrinsic factors such as T cell exhaustion, and not extrinsic factors such as suppressive T cell subsets.
Discovery of how T cells are affected by the TME is critical for the understanding the impact of the immune response on cancer outcome and potential targets for immunotherapy. This study describes an exhausted T cell phenotype and functional impairment of tumor-infiltrating T cells in CRC, and may provide insights that are pertinent to developing immunotherapies for this disease.
Patients undergoing elective surgery for CRC at Dunedin Hospital were invited to participate. The study was approved by the Lower South Regional Ethics Committee (LRS/10/11/054 and LRS 11/04/017). All patients were over 18 y old and gave written informed consent before inclusion in the study in accordance with the Treaty of Helsinki. Resected specimens were dissected by a pathologist, and fresh samples were obtained from the tumor and NTB tissue. Samples from patients with metastatic disease were taken from the primary tumor only for analysis. The NTB samples were obtained from macroscopically normal appearing colon more than 5 cm from the tumor. The samples were transported directly to the laboratory for immediate processing. Patients were entered into a prospectively maintained database with detailed clinicopathological information and clinical follow-up (Table S1).
Fresh tissue samples were weighed and then washed with phosphate-buffered saline (PBS; 0.8% NaCl; ThermoFisher Scientific, 0.114% Na2HPO4; VWR International; 0.02% KH2PO4, Merck; 0.02% KCl, VWR International and distilled water). Tissue was manually dissociated and suspended in 5 mL complete RPMI (supplemented with 100 U/mL Penicillin, 100 μg/mL Streptomycin, 55 μM 2-mercaptoethanol, all from Gibco; Invitrogen) containing 0.5 mg/mL of collagenase (Sigma-Aldrich). Half of the sample was then used to measure ex vivo phenotype by stimulation with 10 ng/mL phorbol 12-myristate 13-acetate (PMA) and 500 ng/mL ionomycin (both from Sigma-Aldrich) to detect cytokine production. A 4 h incubation period was used for digestion at 37°C for both stimulated and ex vivo fractions. Brefeldin A (BFA, Sigma-Aldrich) 1 μg/mL was added after 2 h.11
Lymphocytes from single cell suspensions obtained from NTB and tumor tissue were isolated over a discontinuous Ficoll gradient, as described previously.42 For ex vivo studies, live T cells were isolated from NTB and tumor via FACS sorting based on their expression of CD3 and the exclusion of live/dead (L/D) dye. Tregs were isolated from NTB and tumor tissue by sorting for live CD3+CD25hi T cells and live CD3+CD25low/- cells, respectively. The Treg population was verified by FOXP3 expression (Fig. S1A, B).
T cells were plated in a 96-well plate (Falcon) at 1 × 105 cells per mL in RPMI-HS (complete RPMI supplemented with 100 μg/mL Penicillin, 100 μg/mL Streptomycin, 55 μM 2-mercaptoethanol and 10% human serum (Life Technologies)) for 3 d at 37°C. Cells were stimulated with anti-CD3/CD28 beads (Thermo Fisher Scientific) at a 1:2 bead to cell ratio. Proliferation was then measured by detection of Ki67 expression using flow cytometry.
Samples were resuspended in 1 mL PBS with 0.5 µL per sample of L/D Fixable Red Dead Stain (Invitrogen) for 30 min on ice in darkness. Cells were washed and resuspended in 100 μL of FACS buffer ((PBS, 0.01% sodium azide (VWR International), 0.5% fetal calf serum (FCS; PAA Strasse), 0.075% EDTA; VWR International)) containing the appropriate antibodies (listed in Table S2) against surface markers. After 30 min on ice, the samples were washed and fixed in PBS containing 1% paraformaldehyde (PFA; Sigma-Aldrich). Cells were washed and resuspended with 1 mL permeabilization buffer (FACS buffer + 0.5% saponin (EMD Biosciences)) for 1 h, then washed and incubated for 60 min in permeabilization buffer containing the appropriate intracellular antibodies (listed in Table S2). Samples were then washed three times in permeabilization buffer, and resuspended in 250 μL FACS buffer for acquisition. All antibodies were titrated on PBMCs before use. Fluorescence minus one controls were used as gating controls. All flow cytometry data acquisition was performed on either an LSR-FORTESSA (Becton Dickinson) or FACSAria IIu (Becton Dickinson). Data was acquired using FACSDiva software (Becton Dickinson) and analyzed using FlowJo software (version 9.7.6, Tree Star, Ashland, OR, USA).
T cell subsets were sorted from NTB and tumor specimens using a FACSAria II flow cytometer (Becton Dickinson). Sterility could not be maintained during cell sorting so cells were sorted into 15 mL Falcon tubes containing 1 mL complete RPMI media with 20 ug/mL gentamicin + 2 ug/mL fungizone (both from Invitrogen).
Wilcoxon-matched pairs signed rank test was used to test for significant differences between non-parametric paired data. Statistical analyses of multiple variables between matched pairs were carried out with two-way ANOVA with Holm-Šídák post-hoc analysis to correct for multiple comparisons.
No potential conflicts of interest were disclosed.
We wish to thank all of the patients who have contributed to this study. We thank Kirsten Ward-Hartstonge, Shirley Shen and Sam Norton for critical review of the manuscript.
This study was supported by the Cancer Society of New Zealand, the Genesis Oncology Trust and the Health Research Council of New Zealand. ET was supported by a University of Otago PhD scholarship.
Michael A. Black http://orcid.org/0000-0003-1174-6054