Activated tumor-infiltrating CD4+ regulatory T cells restrain antitumor immunity in patients with primary or metastatic liver cancer


  • Potential conflict of interest: Nothing to report.

  • Supported by the Erasmus MC University Medical Center (fellowship award to D. S.). D. S. was a clinical research trainee of the Netherlands Organisation of Scientific Research. Erasmus Trustfonds financial support was granted to A. P.-G.


The mechanisms that enable liver cancer to escape elimination by the immune system remain unclear, but their elucidation may provide novel therapeutic interventions. We investigated the influence of tumor-infiltrating regulatory T cells on tumor-specific T cell responses in patients with liver cancer, using ex vivo isolated cells from individuals with hepatocellular carcinoma (HCC) or liver metastases from colorectal cancer (LM-CRC). Here we report that in both HCC and LM-CRC, CD4+CD25+Foxp3+ regulatory T cells (Tregs) accumulate in the tumor milieu and are potent suppressors of autologous tumor-specific T cell responses. Especially in LM-CRC, where Treg accumulation is more prominent, there is good evidence for local proliferation of Tregs at the cancer site. We show that tumor Tregs up-regulate the expression of glucocorticoid-induced tumor necrosis factor receptor (GITR) compared with Tregs in tumor-free liver tissue and blood. Importantly, treatment with soluble GITR ligand (GITRL) induces a decrease in the suppression mediated by the activated tumor-infiltrating Tregs and restores the proliferative capacity and cytokine production of CD4+CD25− T cells. Conclusion: Our results show that tumor-associated Tregs are critical for immune evasion in liver cancer, and we propose that GITRL constitutes a rational treatment for this disease. (HEPATOLOGY 2013)

The two most common types of cancer affecting the liver are hepatocellular carcinoma (HCC) and liver metastases from colorectal cancer (LM-CRC).1, 2 The current therapeutic options for both malignancies are limited to liver surgery and local (ablative) therapy. At the time of diagnosis, the majority of HCC patients are not candidates for curative treatment, and in patients with LM-CRC there is a high rate of recurrence after treatment.1, 2 Consequently, there is a pressing need for novel therapeutic strategies. Immunotherapy is attractive because of the exquisite specificity of the immune response and may thus avoid many of the side effects associated with currently available clinical options.3 However, immunological tolerance of the liver or immunoregulatory mechanisms present in the tumor microenvironment, such as those described by us and others in breast,4-6 ovarian,7 and renal8 cancer, may contribute to tumor outgrowth and limit immunotherapeutic strategies by suppressing the local antitumor response.

There is accumulating evidence that CD4+FoxP3+ regulatory T cells (Tregs) hamper the development of effective tumor immunity in individuals with cancer.9, 10 Treg numbers are increased in blood and primary tumors of colorectal cancer (CRC) patients,11, 12 and circulating Tregs have been shown to exert tumor-specific suppression,13 suggesting a potential role in modulating tumor immunity. Nevertheless, it is not known whether the intratumoral presence of these cells has an impact on the tumor-specific T cell response. Furthermore, despite the common metastasis of CRC to the liver, which is the leading cause of CRC-related morbidity and mortality,2 there are no reports about the role for Tregs in hepatic CRC metastasis. Because growing evidence suggests that Tregs may be involved in the regulation of hepatic immune responses and in the unique immune environment in the liver which seems to favor tolerance,14, 15 these cells may also prevent the activation of tumor-specific immune responses in the liver, thereby favoring tumor development. Immunohistochemical studies have reported the presence of FoxP3+ T cells in HCC and their correlation with clinical prognosis.10, 16 However, few studies have analyzed Treg function in HCC patients,17-19 and they all used material from patients chronically infected with hepatitis B and C virus (HBV and HCV), both of which have been shown to induce intrahepatic accumulation of virus-specific Tregs in the absence of cancer,20-22 and so the potential role of Tregs in suppressing HCC-specific immune responses remains unclear.

The aim of this study was to identify an immunosuppressive role for tumor-infiltrating Tregs that can be targeted to improve the efficiency of immunotherapeutic efforts intended to raise an effective tumor-specific T cell response in patients with liver cancer. Using ex vivo isolated cells of patients undergoing surgery for LM-CRC and HCC (no HBV/HCV), we show that Tregs accumulate in the tumor milieu. These tumor-associated Tregs are activated, express high levels of glucocorticoid-induced tumor necrosis factor receptor (GITR) and the inducible T cell costimulator (ICOS), and they are more potent suppressors of tumor-specific CD4+ T cell responses than circulating Tregs. Importantly, treatment with soluble GITR ligand (GITRL) decreases the suppression mediated by tumor-infiltrating Tregs derived from both groups of patients.


CFSE, carboxyfluorescein diacetate succinimidyl ester; CMV, cytomegalovirus; CRC, colorectal cancer; CTLA-4, cytotoxic T lymphocyte-associated antigen 4; DC, dendritic cell; ELISA, enzyme-linked immunosorbent assay; GITR, glucocorticoid-induced tumor necrosis factor receptor; GITRL, GITR ligand; GM-CSF, granulocyte-macrophage colony-stimulating factor; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; ICOS, inducible T cell costimulator; LM-CRC, liver metastases from colorectal cancer; mDC, myeloid dendritic cell; MNC, mononuclear cell; NK, natural killer; NKT, natural killer T; NL, normal liver; TL, tumor lysate; Treg, regulatory T cell; PB, peripheral blood; PBMC, peripheral blood mononuclear cell; TFL, tumor-free liver; TIL, tumor-infiltrating lymphocyte; TNF-α, tumor necrosis factor-α.

Patients and Methods


A total of 64 individuals who were eligible for surgical resection of HCC (n = 21) or LM-CRC (n = 43) were enrolled in the study between September 2009 and October 2011. Paired fresh liver tumor and tumor-free liver (TFL) tissue at the maximum distance from the tumor were used for isolating tumor-infiltrating lymphocytes (TILs) and intrahepatic lymphocytes. In addition, peripheral blood (PB) was collected. All patients were negative for antibodies against human immunodeficiency virus, HBV, and HCV, and in none of the patients was the tumor treated with chemotherapy or radiation prior to resection. There was no comorbidity that required immunomodulatory drugs (e.g., steroids). The clinical characteristics of the patients are summarized in Table 1. Most subjects included in the study were Caucasian. The study was approved by the local ethics committee, and all patients in the study gave informed consent before tissue donation.

Table 1. Patient Characteristics
 HCC (n = 21*)LM-CRC (n = 43)
  • Abbreviations: INR, international normalized ratio; TNM, tumor-node-metastasis.

  • *

    Etiology of liver disease: no known liver disease (n = 11), hemochromatosis (n = 4), nonalcoholic steatohepatitis (n = 3), alcohol-related liver disease (n = 2), porphyria (n = 1).

  • **

    Mean ± SEM.

Sex (men/women)13/827/16
Age (years)**64.6 ± 3.065.1 ± 1.7
Race (Caucasian/Asian/African)20/1/042/0/1
Alanine aminotransferase (U/L)**93.9 ± 28.633.2 ± 6.2
Bilirubin (μmol/L)**21.1 ± 6.68.2 ± 0.7
Prothrombin time (INR)**1.1 ± 0.021.0 ± 0.0
Liver fibrosis (metavir score) F0-F1/F2/F3-F4-cirrhosis11/4/643/0/0
Stage of disease (TNM)Stage I, n = 8; stage II, n = 13Stage IVa, n = 39; stage IVb, n = 4

Cell Preparation.

Peripheral blood mononuclear cells (PBMCs) were isolated via Ficoll density gradient centrifugation. Single cell suspensions from TFL and tumor were obtained via tissue digestion. Briefly, fresh tissue was cut into small pieces and digested with 0.5 mg/mL of collagenase (Sigma-Aldrich, St. Louis, MO) and 0.1 mg/mL of DNase I (Roche, Indianapolis, IN) for 30 minutes at 37°C. Cell suspensions were filtered through cell strainers and mononuclear cells (MNCs) were obtained by Ficoll density gradient centrifugation. Viability was determined by trypan blue exclusion.


Formalin-fixed, paraffin-embedded sections (6 μm) from liver tissues were used for immunohistochemistry. Deparaffinized sections were boiled for 10 minutes in Tris (10 mM)/ethylene diamine tetraacetic acid (1 mM) (pH 9.0) buffer for antigen retrieval. The sections were labeled with 10 μg/mL of anti-FoxP3 antibody (clone 236A/E7; AbCAM, Cambridge, UK). Endogenous peroxidase blockage and the secondary reagent used to detect the primary antibody were from the EnVision+ System-HRP kit (Dako, Denmark). Tissue sections were counterstained with hematoxylin.

Flow Cytometry Analysis.

PBMCs and MNCs isolated from TFL or tumor were analyzed for expression of surface and intracellular markers using the following anti-human antibodies: anti-ICOS, anti-GITR, anti-Ki67, anti-CD25, anti-CTLA-4, anti-granzyme B, anti-Perforin, anti-CD8, anti-HLA-DR, anti-FoxP3, anti–TNF-α, anti-CD4, anti-CD3, anti-CD56, anti-CD45, and anti-CD25 (see Supporting Information for details). Cells were analyzed in a FACSCanto II system (BD Biosciences, San Diego, CA).

Antigen-Specific T Cell Proliferation.

Tumor lysates were generated from freshly dissected tumors by five cycles of freezing and thawing in phosphate-buffered saline, followed by filtration (0.2 μm), and normal liver (NL) lysates were made by the same method from TFLT tissue. Myeloid dendritic cells (mDCs) were isolated from PBMCs by positive selection (BDCA-1 dendritic cell isolation kit, Miltenyi Biotec, Germany). mDCs were cultured overnight with media or 10 μg/mL autologous tumor lysate (TL), NL, or cytomegalovirus (CMV) antigens (Microbix Biosystems, Mississauga, Ontario, Canada) in the presence of 10 ng/mL of granulocyte-macrophage colony-stimulating factor (GM-CSF) (Miltenyi Biotec) and 0.1 μg/mL of polyinosinic:polycytidylic acid (InvivoGen, San Diego, CA). CD4+CD25− cells were isolated from PBMCs or TILs that were kept overnight at 4°C in medium supplemented with 10% fetal bovine serum, by magnetic sorting (Miltenyi Biotec). CD4+CD25− T cells were labeled with 0.1 μM carboxyfluorescein diacetate succinimidyl ester (CFSE, Invitrogen) and cocultured with autologous mDCs, pulsed with media, TL, NL, or CMV, at a ratio of 1:10 for 5 days in round-bottom 96-well plates with at least 5 × 104 CD4+CD25− T cells. Proliferation was measured by dilution of CFSE and reported as a fold increase of specific T cell proliferation, which was calculated by dividing the percentage of proliferating cells in the TL, NL, or CMV condition by that in control condition (media-pulsed dendritic cells [DCs]).

Suppression Assay.

CD4+CD25+ Tregs were isolated by magnetic sorting (Supporting Information). The suppressive effect of Tregs was assessed by coculturing isolated Tregs from peripheral blood or tumor tissue with autologous CFSE-labeled CD4+CD25− T cells that were activated as described for the antigen-specific T cell proliferation assay. Tregs were added at different ratios and cocultured for 5 days. Tregs were labeled with CellTrace Violet (Invitrogen) to be excluded from proliferating T cells (Supporting Fig. 1D). The concentration of tumor necrosis factor-α (TNF-α) in the culture supernatants was determined using an enzyme-linked immunosorbent assay (ELISA) kit (e-biosciences, San Diego, CA) according to the manufacturer's instructions. Furthermore, after coculture, T cells were restimulated overnight with autologous monocyte-derived DCs pulsed with media, TL, or CMV in the presence of brefeldin A and monensin (BD Biosciences). Proliferation and cytokine production were analyzed via flow cytometry. Inhibition of T cell proliferation by Tregs was determined via comparison with culture conditions without Tregs and is reported as the percentage of suppression of T cell proliferation. In some experiments, soluble GITRL (Enzo, Life Sciences) was added to the cocultures. Monocyte-derived DCs were obtained by culturing monocytes with 10 ng/mL interleukin-4 and 50 ng/mL GM-CSF for 5 days, after which immature DCs were pulsed with TL or CMV as described for mDCs in previous paragraph on Antigen-Specific T Cell Proliferation.

Statistical Analysis.

All data set distributions were analyzed for normality using a Shapiro-Wilk normality test. The differences between paired groups of data were analyzed according to their distribution via t test or Wilcoxon matched pairs test. Differences between different groups of patients were analyzed via t test or Mann-Whitney test using GraphPad Prism Software (version 5.0). P values less than 0.05 were considered statistically significant 1 (*P < 0.05; 2 **P < 0.01; ***P < 0.001).


CD4+ T Cells Are the Predominant Lymphocyte Population Infiltrating Liver Tumors.

To compare the composition of TILs to that of TFL and PB, freshly isolated lymphocytes from individuals with HCC without HBV/HCV infection or from patients with LM-CRC were analyzed via flow cytometry (Fig. 1A,B). In both patient groups, of which the majority were of Caucasian origin, TFL displayed similar percentages of lymphocytes as reported for healthy livers23, 24; with a high proportion of natural killer (NK) cells (HCC, 29.4 ± 10%; LM-CRC, 30.3 ± 16.3%), natural killer T (NKT) cells (HCC, 8.7 ± 3.8%; LM-CRC, 15.99 ± 9.04%) and CD8+ T cells (HCC, 49.3 ± 14.8%; LM-CRC, 46.9 ± 12% of CD3+CD56− T cells). When TFL was compared with the corresponding liver tumors, we observed decreased proportions of NK and NKT cells in the tumor (NK cells: HCC, 9.5 ± 9%, P = 0.002; LM-CRC, 8.5 ± 7.3%, P < 0.0001; NKT cells: HCC, 2.6 ± 1.6%, P < 0.0001; LM-CRC, 6.5 ± 5.3%, P < 0.0001). In contrast, T cells (CD3+CD56−) were significantly concentrated at the tumor site (HCC, TFL 30.1 ± 13.5% of total CD45+ cells versus 63.6 ± 21.7% in the tumor, P < 0.0001; LM-CRC, TFL 28.7 ± 11.1% versus 52.2 ± 20.2%, P < 0.0001). In line with previous reports,23, 24 in TFL most T cells were CD8+ T cells. However, at the tumor site, the main population of T cells expressed CD4 (HCC, 65.7 ± 17.2%; LM-CRC, 61.4 ± 13.9%). Similar results were observed when the absolute numbers of cells were analyzed (Supporting Fig. 2).

Figure 1.

Flow cytometric analysis of hepatic and circulating lymphocytes from patients with liver cancer. (A) Viable (aqua dye-negative) leukocytes were gated based on CD45 expression, and then NK, NKT, and T cells were defined based on the expression of CD3 and CD56. CD3+CD56− cells were further analyzed for CD4+ and CD8+ T cells. (B) Percentages of NK (CD3−CD56+), NKT (CD3+CD56+), and T (CD3+CD56−) cells within the CD45+ fraction and proportions of CD4+ and CD8+ T cells within CD3+CD56− T cells isolated from liver tumor tissue, TFL, and PB from patients with HCC or LM-CRC.

Impaired Tumor-Specific T Cell Functionality at the Tumor Site.

To characterize the functionality and specificity of the large population of tumor-infiltrating CD4+ T cells, a proliferation-based assay was used to measure tumor-specific responses of CD4+CD25− T cells stimulated with autologous DCs pulsed with self-TL as a source of tumor antigens (Fig. 2). When CD4+CD25− T cells from PB were compared with tumor-derived CD4+CD25− T cells there was a significantly decreased proliferation to TL using both HCC and LM-CRC-infiltrating T cells (Fig. 2A).

Figure 2.

TILs display a compromised functionality. (A) CFSE-labeled CD4+CD25− T cells from blood or tumor tissue were stimulated with autologous blood-derived mDCs pulsed with autologous TL and cultured for 5 days. TL-specific proliferation measured at day 5, is reported as fold increase of TL-specific T cell proliferation compared with the proliferation in the control condition. Left panels show a representative experiment, right panels pooled data. (B) Phycoerythrin-labeled antibodies against granzyme B and perforin were mixed and used for intracellular staining of CD8+ lymphocytes from PB, TFL, and tumor tissue of patients with HCC or LM-CRC. One representative analysis is shown together with statistical comparisons for several patients of each group. The level of expression is reported as mean fluorescence intensity (MFI).

To confirm that proliferation of PB-derived CD4+CD25− T cells induced by TL was tumor-specific, we compared it with the proliferation induced by a lysate derived from TFL. We observed that DCs pulsed with TFL lysates induced a level of T cell proliferation that was comparable to that observed using a control with media-DCs only, which was significantly lower than the proliferation induced by TL (Supporting Fig. 3). These findings indicate that the observed responses to TL are tumor-specific.

Additionally, tumor-infiltrating CD8+ T cells had a considerably decreased expression of the cytolytic enzymes perforin and granzyme B in both groups of patients compared with CD8+ T cells from TFL and PB (Fig. 2B). Thus, T cells in HCC and metastatic CRC displayed impaired tumor-specific functionality, demonstrating the need for immunotherapeutic modulation to restore the local immunity against malignant cells.

Activated Tregs Are Sequestered at the Tumor Site.

Following the observation of increased numbers of CD4+ T cells at the tumor site, we asked whether these cells contained CD4+ regulatory T cells that may suppress local T cell responses. Therefore, we analyzed the frequencies and absolute numbers of CD3+CD4+CD25+FoxP3+ Tregs in the blood and liver of the patients. Tregs were present in all compartments analyzed, but significant accumulation was observed in the tumor areas compared with TFL and blood in both HCC and LM-CRC patients (Fig. 3A and Supporting Fig. 2). Tumor-infiltrating Tregs represented 7.2 ± 3% of total tumor CD4+ T cells in HCC and 9.8 ± 5% in LM-CRC. To analyze their distribution within the tissues, we performed immunohistochemistry and observed that FoxP3+ cells were very scarce in TFL (data not shown). In contrast, they were enriched within the tumors. In both LM-CRC and HCC tissue, the majority of FoxP3+ cells were localized within the tumor stroma (Fig. 4B,D). However, Tregs were also found among lymphocytic infiltrates in the tumor bed (Fig. 4A,C) and in the marginal zones separating the tumor from surrounding liver tissue (Fig. 4A).

Figure 3.

Tregs accumulate in liver tumors and display an activated phenotype. The presence of Tregs (CD3+CD4+CD25+FoxP3+) was analyzed by flow cytometry in liver tissue and peripheral blood from patients with liver cancer. (A) Representative analysis of one patient. PBMCs or liver MNCs were first gated on viable CD45+CD3+CD4+ cells, and then Tregs were defined by the expression of CD25 and FoxP3. The percentages of Tregs among total CD4+ T cells of each compartment evaluated are summarized in the corresponding graphs. (B) Phenotypic characterization of Tregs in HCC and LM-CRC patients was performed by measuring the surface expression levels of ICOS, GITR, HLA-DR, and CD25, and intracellular expression of CTLA-4. Histograms show representative analysis and dot plot graphs show a summary of the mean fluorescence intensity for each marker.

Figure 4.

Distribution of tumor-infiltrating Tregs in liver cancer. Immunohistochemistry analysis in paraffin-embedded tumor sections from patients with LM-CRC (A,B) or HCC (C,D) is shown. (A) Marginal zone of liver metastasis with lymphocytic infiltration containing FoxP3+ cells (arrowheads) and tumor-infiltrating Tregs (arrows). (B) FoxP3+ Tregs infiltrating the tumor stroma. (C) TILs containing Tregs (brown) in HCC. (D) Infiltration of the tumor bed and tumor stroma by FoxP3+ Tregs. Magnification ×100 (A,C,D) and ×200 (B).

Furthermore, we analyzed the expression of HLA-DR, CD25, ICOS, GITR, and cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) in blood, TFL, and tumor CD4+ CD25+FoxP3+ Tregs (Fig. 3B). Based on the expression of these markers, we concluded that intrahepatic Tregs displayed a more activated phenotype than circulating Tregs. Interestingly, tumor Tregs expressed significantly more ICOS and GITR than Tregs from TFL and blood. In addition, Tregs from LM-CRC expressed significantly higher levels of CTLA-4, HLA-DR, and CD25 than their counterparts from TFL, indicating a higher state of Treg activation at the tumor site. When comparing expression of these markers between both patient groups, we observed that expression of the activation markers HLA-DR and CD25 was higher on Tregs from LM-CRC than HCC tumors (HLA-DR LM-CRC, 3,398 ± 2,182 mean fluorescence intensity (MFI) versus HCC, 1,681 ± 1,064 MFI, P = 0.0337; and CD25 LM-CRC, 5,257 ± 3,110 MFI versus HCC, 2,022 ± 1,189 MFI, P = 0.0018), whereas expression of the other markers was similar. Thus, tumor-derived Tregs were characterized by the expression of high levels of GITR and ICOS and Tregs in LM-CRC displayed an even more activated phenotype than those in HCC.

Tumor-Infiltrating Tregs Are Functionally Superior to Circulating Tregs, and LM-CRC Derived Tregs Suppress Better than Tregs from HCC.

To analyze whether the enhanced activation status of tumor-derived Tregs was reflected in superior suppressive functionality compared with their counterparts from PB, cocultures of Tregs from blood or tumor with CD4+CD25− T cells from PB stimulated with PHA, TL, or CMV were performed. PHA-stimulated CD4+CD25− T cells responded with a robust proliferation, and both circulating and tumor-derived Tregs inhibited T cell proliferation in a dose-dependent manner (Supporting Fig. 4). However, tumor Tregs showed stronger inhibitory capacity than circulating Tregs. To compare inhibition of tumor-specific T cell responses, autologous peripheral CD4+CD25− T cells were stimulated with DCs pulsed with TL and cocultured with Tregs (Fig. 5). Both HCC- and LM-CRC–derived Tregs were more potent suppressors of tumor-specific T cell proliferation than circulating Tregs, but only LM-CRC–derived and not HCC-derived Tregs convincingly inhibited the cytokine production of responder T cells better than circulating Tregs. However, CMV-specific T cell proliferation and cytokine production in both HCC and LM-CRC are more potently suppressed by tumor-derived Tregs than blood-derived Tregs.

Figure 5.

Tumor Tregs are more potent suppressors of antigen-specific CD4+ T cell responses than circulating Tregs. At least 5 × 104 CFSE-labeled CD4+CD25− T cells from peripheral blood were stimulated with autologous mDCs pulsed with TL or CMV for 5 days in the absence or presence of autologous Tregs isolated from blood or tumor tissue. The Treg/CD4+CD25− T cell ratio was 1/5. Proliferation and cytokine production (TNF-α) were measured after restimulation with autologous Mo-DC pulsed with TL or CMV. (A) Representative analysis of one patient with LM-CRC. (B) Statistical analysis of the suppressive activity of Tregs from several HCC or LM-CRC patients exerted on CD4+CD25− T cell proliferation and TNF-α production as measured by flow cytometry and (C) using enzyme-linked immunosorbent assay in the culture supernatants obtained before restimulation (mean ± SEM).

Whereas no significant differences were found between HCC patients with and without cirrhosis with regard to tumor-infiltrating CD4+ T cells, including Tregs and their functionality (Supporting Fig. 5), phenotypic and functional differences were noted comparing tumor-infiltrating Tregs from patients with HCC and LM-CRC. As seen in Fig. 3, the frequency of Tregs at the tumor site is significantly higher in LM-CRC versus HCC patients (P = 0.04). In addition, the expression of both HLA-DR and CD25 were significantly higher in Tregs from LM-CRC than in HCC, suggesting a higher activation state in the former disease. This is reflected in the superior capacity of LM-CRC–derived Tregs to inhibit both TL- and CMV-specific T cell proliferation (percentages of suppression, respectively; TL: HCC, 42 ± 12% versus LM-CRC, 68 ± 21%, P = 0.0381; CMV: HCC, 39 ± 21% versus LM-CRC, 65 ± 24%, P = 0.0305). These differences were not observed using Tregs from PB. Altogether, these data indicate that tumor-infiltrating Tregs are highly activated and are potent suppressors of the tumor-specific and non–tumor-specific effector CD4+ T cell responses. Furthermore, LM-CRC–derived Tregs appear to be better suppressors of T cell responses than Tregs in HCC.

In LM-CRC, Local Proliferation of Tregs Contributes to Their Intratumoral Accumulation.

To investigate whether local proliferation of Tregs is involved in their accumulation at the tumor site, we measured the expression of the proliferation marker Ki67 in freshly isolated cells. In HCC patients, Ki67 expression in Tregs from PB, TFL, and tumor was similar. In contrast, in LM-CRC patients, elevated proportions of Tregs expressing Ki67 are observed in tumor tissue (Fig. 6). Ki67 expression in LM-CRC–derived Tregs was also significantly higher than in those isolated from HCC (P = 0.0428). Thus, in metastatic CRC, liver cancer local proliferation of tumor-infiltrating Tregs may contribute to the accumulation of these cells in the tumor bed.

Figure 6.

Differential proliferative activity distinguishes tumor Tregs from HCC and LM-CRC. Intracellular expression of Ki-67 was measured in fresh ex vivo obtained cell suspensions from tumor tissues via flow cytometry. CD4+CD25+FoxP3+ Tregs were gated from total viable cells, and actively dividing cells were identified by expression of Ki67 and reported as percentage of positive cells of total Tregs. Representative dot plots from HCC and LM-CRC patients are shown. Graphs summarize percentages of Ki67 expressing cells in blood, TFL, and tumor from HCC and LM-CRC.

GITR Ligation Prevents Hyporesponsiveness of Effector T Cells Coincubated with Tumor-Derived Tregs.

Engagement of GITR can reverse the suppressive effect of Tregs on effector T cells, either by a direct stimulating effect on effector T cells or indirectly by interfering with the suppression induced by Tregs.25, 26 Because tumor-derived Tregs displayed a more prominent expression of GITR compared with Tregs isolated from TFL or from blood, we hypothesized that soluble GITRL would prevent effector T cell hyporesponsiveness provoked by tumor-derived Tregs. Treatment with a dose of 10 μg/mL soluble GITRL was unable to increase proliferation or TNF-α production by CD4+CD25− T cells activated with autologous DCs pulsed with CMV in the absence of Tregs (Fig. 7A,B). In contrast, this concentration of soluble GITRL significantly reduced Treg-mediated inhibition of CMV-specific CD4+CD25− T cell proliferation and cytokine production (Fig. 7B) in six out of seven patients tested (41.3 ± 19% suppression of T cell proliferation and 50.6 ± 34.5% suppression of TNF-α production in the presence of tumor Tregs versus 25.8 ± 21% suppression of T cell proliferation and 33.5 ± 30.5% suppression of TNF-α production when tumor Tregs and GITRL were present in the coculture; P = 0.011 and 0.0313, respectively [n = 7]). The effect on cytokine production was corroborated via ELISA of the culture supernatants (Fig. 7C). A higher concentration of soluble GITRL (20 μg/mL) was found to also stimulate the proliferation of responder CD4+CD25− T cells in the absence of Tregs (Fig. 7A). Thus, treatment with soluble GITRL alleviated the suppression mediated by highly activated tumor infiltrating Tregs, and it may therefore be considered as an adjuvant to immunotherapeutic interventions aimed at stimulating efficient antitumor T cell activity at the tumor site.

Figure 7.

GITR engagement partially abrogates suppression mediated by tumor-infiltrating Tregs. At least 5 × 104 CFSE-labeled CD4+CD25− T cells isolated from peripheral blood were stimulated with autologous CMV-pulsed mDCs for 5 days in the absence or presence of soluble GITRL. (A) Representative data of three experiments using different doses of GITRL, showing that a higher concentration induces proliferation of CD4+CD25− responder T cells in the absence of Tregs (upper plots), and that when a dose of 10 μg/mL is used, the observed effects are predominantly mediated by inhibition of suppression by Tregs rather than by direct induction of proliferation on the responder cells. (B) Autologous tumor-derived Tregs from HCC patients (black circles) or LM-CRC (white circles) were added in the absence or presence of 10 μg/mL of soluble GITRL. The Treg/CD4+CD25− T cell ratio was 1:5. T cell proliferation and TNF-α production was measured via flow cytometry after restimulation with autologous Mo-DC pulsed with CMV, and (C) TNF-α production was analyzed via ELISA in the culture supernatants before restimulation (n = 7; mean ± SEM).


Normal liver contains significant numbers of lymphocytes, with high frequencies of CD8+ T cells, NK cells, and NKT cells.23, 24 All of these cell types are potential effectors of tumor growth control. Interestingly, we found that their frequencies were decreased in the tumor bed in both HCC and LM-CRC patients. Furthermore, tumor-derived CD4+CD25− T cells displayed significantly decreased tumor-specific proliferative capacity compared with those from PB, and tumor-infiltrating CD8+ T cells exhibited a reduced expression of cytotolytic effector molecules, confirming similar findings in HCC in viral hepatitis patients.17, 27 These dramatic changes in the composition and function of lymphocytes within the tumors suggest that an immunopermissive environment is essential for liver cancer development.

Here, using tissue from predominantly Caucasian patients, we show that functional Tregs accumulated in the tumors of patients with HCC without viral hepatitis infection. In fact, more than half of HCC patients had no known prior liver disease, and the histology of 15 of 21 HCC patients showed no or mostly mild fibrosis in the surrounding liver tissue (Metavir score F1-F2). These data suggest that the HCC microenvironment by itself can induce the presence of high numbers of functional Tregs that can locally suppress the tumor-specific T cell response. Compared with HCC, in tissue from LM-CRC we found an even higher frequency of tumor-infiltrating Tregs, which also displayed a more activated phenotype and superior inhibition of CD4+ T cell responses to tumor and nontumor antigens. Our data suggest that local proliferation of tumor-infiltrating Tregs may contribute to the higher frequencies observed in LM-CRC tumors. A recent study in HCC in viral hepatitis patients suggested the possible migration of Tregs mediated by CCL-20 produced at the tumor site,28 and even though this mechanism is still elusive, in LM-CRC the expression of CCL20 appears to be higher than in HCC,29 which may also contribute to the increased numbers of Tregs observed in LM-CRC. Together, these findings support potential tumor-specific rather than organ-specific Treg recruitment and activation in primary and secondary liver cancers.

The high frequencies of CD4+CD25+FoxP3+ Tregs in the tumors and their profound suppression of T cell responses as observed in this study strongly support the possibility that in vivo at the tumor site, these Tregs may inhibit local antitumor immunity, which might promote tumor survival and may also interfere with immunotherapeutic efforts to induce efficient antitumor immunity. However, in addition to CD4+FoxP3+ Tregs, other suppressive populations may be present in the CD4+ T cell fractions that contribute to their hyporesponsiveness to tumor antigens.

Tumor-infiltrating Tregs in both types of liver tumors are characterized by significantly higher expression levels of GITR and ICOS compared with Tregs from TFL and blood. These molecules are regulators of their suppressive function30, 31 and can be targeted for immunotherapeutic intervention. Several reports suggest that signaling through GITR interferes with Treg-effector T cell interaction, either by abrogating Treg-suppressive function26, 31, 32 or by conferring effector T cells resistant to Treg-mediated suppression.33, 34 Furthermore, GITR is up-regulated on activated conventional (FoxP3−) T cells, and GITR ligation may enhance effector T cell proliferation.25 Here we show that soluble GITRL partially prevents hyporesponsiveness of effector T cells coincubated with Tregs derived from both types of liver tumors. In our experiments, GITRL mediates its effect either by inhibition of Treg-mediated suppression or in combination with stimulation of responder T cell proliferation, depending on the concentration used (10 versus 20 μg/mL). The lower concentration was used in our assays to allow easy interpretation of GITRL-induced effects on Treg suppression, without interference by its T cell stimulatory capacity. However, both the effect of GITRL on Tregs and T cells support the use of GITRL as a possible treatment in cancer, because it could simultaneously abolish the suppression mediated by Tregs and booster tumor-specific T cell responses. Although more research is required to further understand the mechanism, the data suggest that manipulation of the GITR pathway holds promise as immunotherapeutic intervention in patients with HCC and LM-CRC. Potentially, it may serve as an adjuvant to immunotherapeutic interventions aimed at stimulating efficient effector T cell antitumor activity.

In conclusion, our data demonstrate that in both primary and secondary liver cancer, the tumor-specific T cell response is compromised. These tumors contain high numbers of activated Tregs, and these cells suppress tumor-specific T cell activity. GITR ligation is able to prevent hyporesponsiveness of effector T cells when coincubated with tumor-derived Tregs, and GITR may therefore be a target for immunotherapeutic intervention.


We thank the surgeons and pathologists at Erasmus MC for providing and assisting with tissue handling, Andrea Woltman and Andre Boonstra for helpful discussion, and Ernesto Vargas-Mendez and Gertine van Oord for technical assistance.