Increased intratumoral regulatory T cells are related to intratumoral macrophages and poor prognosis in hepatocellular carcinoma patients

Authors

  • Jia Zhou,

    1. State Key Laboratory of Biocontrol, Sun Yat-Sen (Zhongshan) University, Guangzhou, People's Republic of China
    Search for more papers by this author
  • Tong Ding,

    1. State Key Laboratory of Biocontrol, Sun Yat-Sen (Zhongshan) University, Guangzhou, People's Republic of China
    Search for more papers by this author
  • Weidong Pan,

    1. Department of Hepatobiliary Surgery, the Third Affiliated Hospital, Sun Yat-Sen (Zhongshan) University, Guangzhou, People's Republic of China
    Search for more papers by this author
  • Ling-yan Zhu,

    1. State Key Laboratory of Biocontrol, Sun Yat-Sen (Zhongshan) University, Guangzhou, People's Republic of China
    Search for more papers by this author
  • Lian Li,

    Corresponding author
    1. State Key Laboratory of Biocontrol, Sun Yat-Sen (Zhongshan) University, Guangzhou, People's Republic of China
    • College of Life Science, Sun Yat-sen (Zhongshan) University, Guangzhou 510275, People's Republic of China
    Search for more papers by this author
  • Limin Zheng

    1. State Key Laboratory of Biocontrol, Sun Yat-Sen (Zhongshan) University, Guangzhou, People's Republic of China
    2. State Key Laboratory of Oncology in Southern China, Cancer Center, Sun Yat-Sen (Zhongshan) University, Guangzhou, People's Republic of China
    Search for more papers by this author

Abstract

Immunosuppression mediated by regulatory T cells (Tregs) is a key facilitator of tumor immune evasion, but the source of these Tregs and their contribution to human cancer progression remains unclear. This study investigated the properties of FoxP3+ Tregs, their prognostic value in patients with hepatocellular carcinoma (HCC) and the underlying mechanisms of FoxP3+ Treg intratumoral accumulation. In addition to an increased number of circulating FoxP3+ Tregs, the results also showed that FoxP3+ Tregs gathered in the tumor site, where they suppressed tissue-derived CD4+CD25 T-cell activation (p < 0.001), promoting disease progression and poor prognosis in HCC patients (< 0.01). The intratumoral prevalence of FoxP3+ Tregs was associated with a high density of macrophages (Mφ) (p < 0.001). Depletion of tissue Mφ thus attenuated the increase of liver FoxP3+ Treg frequency attributed to in vivo inoculation with hepatoma (p = 0.01), whereas Mφ exposed to tumor culture supernatants from hepatoma-derived cell lines increased FoxP3+ Treg frequency in vitro (p < 0.001). This increase was partially blocked by antiinterleukin-10 antibody (p < 0.01). In conclusion, tumor-associated Mφ may trigger a rise of the intratumoral FoxP3+ Treg population, which in turn may promote HCC progression. © 2009 UICC

Hepatocellular carcinoma (HCC) is the fifth most common cancer, with increasing incidence worldwide, and is characterized by high mortality, frequent postsurgical recurrence and extremely poor prognosis.1–4 Immunotherapy, including anticancer vaccination and adoptive transfer of tumor-specific cytotoxic T lymphocytes, has been thought to be a promising strategy for curing HCC, although its effects on tumor regression remain limited.5–7 The inefficacy of such therapies implies that the hepatoma has developed diverse strategies of escaping tumor-specific immunity.8

Recent studies reported increased numbers of regulatory T cells (Tregs) in the peripheral blood and tumor-infiltrating lymphocytes (TILs) in patients with ovarian or gastric and esophageal cancers, which impaired cell-mediated immunity and promoted disease progression.9–11 Experimental depletion of Tregs in mice with melanoma improved tumor clearance and enhanced the efficacy of immunotherapy.12, 13 Furthermore, depletion of CD4+CD25+ Tregs enhanced T-lymphocyte and NK-cell effector function in end stage cancer patients.14, 15 These data suggest that Tregs may impair cell-mediated immune responses to tumors. In fact, Tregs accumulate and promote HCC progression in the peripheral blood and in situ tumors.16–22 However, based on the published data, it is still unclear whether Treg frequency truly increases in the peripheral blood and whether Treg intratumoral accumulation predicts poor survival in HCC patients.17, 21 The precise underlying mechanism by which Tregs accumulate at the tumor site in HCC patients is also unknown.

Soluble factors derived from the tumor microenvironment, especially those secreted by tumor cells and antigen-presenting cells (APCs), could induce the conversion, expansion and migration of Tregs.23 We recently observed that after exposure to tumor culture supernatants (TSNs) from hepatoma-derived cell lines, tolerogenic dendritic cells (DCs) could selectively induce the expansion of allogeneic Tregs.24 Our previous studies also demonstrated that the tumor microenvironment could affect the development of macrophages (Mφ) and DCs, which in turn may cause these cells to trigger T-cell dysfunction.25, 26 Because Mφ represent a major population of APCs that infiltrate tumors,27, 28 it would be interesting to study whether these cells are involved in the Treg accumulation observed in cancer patients.

In this study, we investigated the distribution, prognostic significance and phenotypic and functional characteristics of forkhead/winged helix scurfy (FoxP3)+ Tregs within the tumors of HCC patients. We also estimated the correlation between intratumoral FoxP3+ Treg and Mφ densities and tried to elucidate the mechanisms by which Mφ enlarge the FoxP3+ Treg population in tumors.

Abbreviations:

APCs, antigen-presenting cells; DCs, dentritic cells; HCC, hepatocellular carcinoma; Mφ, macrophages; NILs, nontumor-infiltrating lymphocytes; PBMCs, peripheral blood mononuclear cells; SNs, supernatants; TILs, tumor-infiltrating lymphocytes; Tregs, regulatory T cells; TSNs, tumor culture supernatants.

Material and methods

Patients and specimens

A total of 213 patients with HCC pathologically confirmed at the Cancer Center of Sun Yat-Sen University were enrolled in this study. None of the patients received anticancer therapy before sampling. Concurrence of autoimmune disease, human immunodeficiency virus (HIV) and syphilis was excluded for all enrolled individuals. Paraffin-embedded, formalin-fixed liver sections were obtained from 8 healthy donors undergoing liver transplantation and 131 HCC patients (Group 1) who underwent surgical resection between 1999 and 2002 (Bank of Tumor Resource, Cancer Center of Sun Yat-Sen University). Of these 131 patients, 85 patients who underwent curative resection with follow-up were further enrolled in survival analysis. Overall survival (OS) was defined as the interval between the dates of surgery and death. Disease-free survival (DFS) was defined as the interval between the dates of surgery and recurrence or the last follow-up. Blood samples from 64 patients (Group 2) and paired fresh tumor and nontumor (at least 3 cm distant from the tumor site) tissues from 18 patients (Group 3) undergoing surgical resection were used for isolating peripheral blood mononuclear cells (PBMCs), tumor-infiltrating lymphocytes (TILs) and nontumor-infiltrating lymphocytes (NILs). Clinical stages of tumor progression were determined according to the tumor-nodes-metastasis (TNM) classification system of the International Union Against Cancer (Edition 6). The clinical characteristics of all patients were summarized in Table I. Control blood samples were obtained from 48 healthy donors at The First Affiliated Hospital of Sun Yat-sen University, all of whom were negative for antibodies to Hepatitis C, Hepatitis B, HIV and syphilis. All samples were coded anonymously according to local ethical guidelines, as stipulated by the Declaration of Helsinki, with written informed consent and a protocol approved by the Review Board of the Cancer Center.

Table I. Characteristics of HCC Patients
VariablesGroup1
123
  • HBsAg, hepatitis B surface antigen.

  • 1

    Group 1: Paraffin-embedded formalin-fixed section samples; Group 2: peripheral blood samples; Group 3: fresh tissue samples.

  • 2

    Mean ± SD.

  • 3

    The biggest diameter of single tumor or multiple tumor.

  • 4

    According to the TNM classification system of international Union Against Cancer (edition 6).

No. of patients1316418
Age (years)249 ± 1253 ± 1153 ± 11
Gender (male/female)113/1858/613/5
HBsAg (+/−)118/1357/715/3
Cirrhosis (+/−)110/2129/3512/6
Child-Pugh score (A/B)124/753/1113/5
AFP, ng/ml (≤25/>25)37/9417/475/13
Tumor size3, cm (≤5/>5)37/9428/366/12
Tumor multiplicity (solitary/multiple)108/2342/2212/6
Vascular invasion (+/−)10/12124/407/11
TNM stage4 (I+II/III+IV)94/3736/2810/8

Immunohistochemistry

Paraffin-embedded, formalin-fixed 5-μm-thick tissue sections were processed for immunohistochemical staining with mouse anti-human CD68 antibodies (Abs; DakoCytomation, Glostrup, Denmark) and mouse anti-human FoxP3 Abs (Abcam, Cambridge, UK) as described.24 The density of CD68+ or FoxP3+ cells was evaluated quantitatively by the mean count of 5 representative fields at 400 × magnification (0.146 mm2/field) by 2 independent observers who were blinded to the clinical outcome. A Leica DM IRB inverted research microscope was used (Leica Microsystems, Wetzlar, Germany).

Cell preparation

PBMCs were isolated by Ficoll density gradient centrifugation.24, 25 NILs and TILs were isolated as previously described with some modifications.29 Fresh tissues were cut into small pieces and digested at 37°C for at least 20 min in RPMI 1640 supplemented with 0.05% collagenase Type IV (Sigma-Aldrich, St. Louis, MO), 0.002% DNase I (Roche, Basel, Switzerland) and 20% fetal calf serum (FCS, HyClone Laboratories, Logan, UT). Dissociated cells were then filtered through a 150-μm mesh and mononuclear cells (MNCs) were obtained by Ficoll density gradient centrifugation.

Purified monocytes were isolated as described before24 and cultured with or without 15% TSNs for 7 days to obtain Mφ. CD3+CD25 T cells were isolated from PBMCs using a Pan T Cell Isolation Kit II followed by using anti-CD25 magnetic beads in the MACS column purification systems according to the manufacturer's instructions (Miltenyi Biotec, Bergisch Gladbach, Germany). Enriched cells were >90% pure. To isolate CD4+CD25high and CD4+CD25 T cells from fresh tissues, MNCs from TILs and NILs were sorted with anti-CD25 and anti-CD4 using the FACSVantage SE cell sorting system (Becton Dickinson, Immunocytometry Systems, San Jose, CA). The resultant cell purity was about 90%.

Flow cytometric analysis

Fluorochrome-conjugated monoclonal Abs (mAbs) against CD3, CD4, CD25, CD152 (also known as cytotoxic T lymphocyte-associated antigen-4, or CTLA-4), CD45RO, CD62L or relevant control mAbs (BD Pharmingen, San Diego, CA), were used. For intracellular staining, cells were fixed and permeabilized with IntraPre Reagent (Beckman Coulter, Fullerton, CA). An APC or PE anti-Foxp3 Staining Set (eBioscience, San Diego, CA) was used for FoxP3 analysis according to the manufacturer's instructions. Data were acquired and analyzed using FACSVantage SE with CellQuest software (Becton Dickinson).

Confocal microscopic analysis

Frozen HCC tissue sections were stained with mouse anti-human FoxP3 Abs (Abcam, Cambridge, UK) and rabbit anti-human CD68 Abs (Sata CruZ Biotechnology, Santa Cruz, CA) followed by labeling with Alexa Fluor 555-conjugated donkey anti-mouse IgG and Alexa Fluor 488-conjugated donkey anti-rabbit IgG (Molecular Probes). Positive cells were assessed using a laser scanning confocal microscope (TCS-SP2, Leica, Jena, Germany) by sequential scanning at wavelengths of 543 nm and 488 nm. The data were analyzed using ImagePro Plus software.

Cell lines and preparation of TSNs

Human hepatoma (HepG2, Huh7) and a leukemia (THP1) cell line were obtained from the American Type Culture Collection. All cells were tested for mycoplasma and lipopolysaccharide (LPS) contamination. TSNs were prepared as described previously.25

In vitro immunosuppressive assay and mixed lymphocyte reaction

For the immunosuppressive assay, 5 × 104 purified CD4+CD25 T cells were cultured with varying numbers of autologous CD4+CD25high cells from TILs or NILs in the presence of 2.5 μg/ml soluble anti-CD3 mAbs plus 5 μg/ml anti-CD28 mAbs (eBioscience). After 4 days, the cells were pulsed with bromodeoxyuridine (BrdU) and assessed for BrdU incorporation 4 hr later using a Cell Proliferation ELISA, BrdU (colorimetric) kit (Roche, Mannheim, Germany), according to the manufacturer's instructions.

For the mixed lymphocyte reaction (MLR), 1 × 105 autologous Mφ were cocultured with 2 × 105 CD14+-depleted PBMCs in the presence of 0.5 μg/ml anti-CD3 mAbs or with 1 × 105 purified CD3+CD25 T cells in the presence of 2 μg/ml anti-CD3 mAbs plus 1 μg/ml anti-CD28 mAbs, with or without 5 μg/ml of blocking mAbs against IgG1 or interleukin-10 (IL-10; R & D Systems, Abingdon, UK), for 4 days.

Cytokine production

The concentration of interferon-γ (IFN-γ) or IL-10 in the culture supernatants was determined using enzyme-linked immunosorbent assay (ELISA) kits (eBioscience) according to the manufacturer's instructions.

In vivo depletion of tissue Mφ

The handling of mice and experimental procedures were conducted in accordance with the animal guidelines of Sun Yat-Sen University under an approved protocol. C57BL/6 mice (females, 5–6 weeks old) were obtained from the Experimental Animal Center of Sun Yat-sen University (Guangzhou, PRC). Mouse hepatoma Hepa1-6 cells provided by the Shanghai Cell Bank of the Chinese Academy of Sciences (Shanghai, PRC) were injected subcutaneously to develop a tumor mass in C57BL/6 mice. Ten days later, the tumor mass was cut into small pieces and inoculated into the livers of C57BL/6 mice for 5 days. Thereafter, liver Mφ were depleted in vivo after intraperitoneal injections with gadolinium chloride (GdCl3, 10 mg/kg; Sigma-Aldrich) every 5 days for 20 days.30, 31 The liver lymphocytes were isolated using a Ficoll density gradient as described29 and analyzed for FoxP3+ Tregs by FACS.

Statistical analysis

All data were summarized as mean ± SE unless otherwise indicated in the text. Statistical analyses were performed with SPSS 13.0 (SPSS, Chicago, IL) using 2-tailed Student's t test. Cumulative survival times were calculated by the Kaplan–Meier method and different survival functions between groups were analyzed by the log-rank test. A multivariate Cox proportional hazards regression model was used to identify independent prognostic factors. To test the association between various variables, the χ2 test was used. Before χ2 testing, unless otherwise noted, the median value was used as a cutoff to dichotomize the series. p values < 0.05 were considered statistically significant.

Results

Intratumoral prevalence of FoxP3+ Tregs could predict poor survival and recurrence in HCC patients

To observe FoxP3+ Tregs in situ, we performed immunohistochemical staining of FoxP3 in paraffin-embedded tissues from HCC patients (Group 1, n = 131). Notably, FoxP3+ Tregs accumulated in the intratumoral region (2.68 ± 0.48 cells/field; range, 0–17.8 cells/field) but were rarely observed in the peritumor and nontumor regions (range, 0–1 cells/field) (Fig. 1a). Moreover, FoxP3+ Tregs were undetectable in liver tissues from healthy donors (n = 8).

Figure 1.

Prognostic significance and intratumoral prevalence of FoxP3+ Tregs in HCC patients. (a) Representative images of immunohistochemical staining of FoxP3+ lymphocytes in the nontumor, peritumor and intratumoral regions. Positive cells were stained brown, as indicated by arrows. The median count of intratumoral FoxP3+ cells was 1.05 cells/field. (b) The cumulative overall survival (OS) and disease-free survival (DFS) of HCC patients were estimated using the Kaplan–Meier method and compared by the log-rank test. Patients with high-FoxP3+ Treg levels (dashed lines) had significantly poorer survival compared with individuals with low FoxP3+ Treg density (solid lines). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

To address whether the intratumoral accumulation of FoxP3+ Tregs affects survival, 85 HCC patients from Group 1, who had undergone curative resection with follow-up evaluation, were divided into 2 groups based on the median value of FoxP3+ lymphocytes/field during immunohistochemical analysis (median, 1.075 cells/field). As illustrated in Figure 1b, there was a significant inverse correlation between intratumoral FoxP3+ Treg density and both OS (p = 0.003) and DFS (p = 0.008). Patients with a higher density (n = 36) of intratumoral FoxP3+ Tregs had significantly shorter OS and DFS than patients evidencing a lower density (n = 49) (OS median, 26 vs. 46 months; DFS median, 8 vs. 19 months). During multivariate analysis, intratumoral FoxP3+ Treg density emerged as an independent prognostic factor of OS (Hazard ratio, HR, 2.315; p = 0.011) and DFS (HR, 1.944; p = 0.012) (Table II). These results suggested that the intratumoral prevalence of FoxP3+ Tregs may serve as an independent predictor of poor HCC prognosis.

Table II. Univariate and Multivariate Analysis of Factors Associated with Survival and Recurrence
 VariablesOverall survivalRelapse free survival
UnivariateMultivariate1UnivariateMultivariate1
HRpHRpHRpHRp
  • Univariate and multivariate analysis; Cox proportional hazards regression model.

  • HR, Hazard Ratio; NA, not adopted.

  • 1

    Multivariate variables were adopted by univariate (p < 0.05).

Age, years> 48vs. ≤ 481.0550.859 NA1.0000.999 NA
GenderMalevs. Female1.2090.690 NA1.1820.678 NA
HBsAgPositivevs. Negative3.2950.100 NA2.2890.109 NA
CirrhosisPresentvs. Absent1.6530.200 NA1.8160.067 NA
Child-Pugh scoreAvs. B1.1780.872 NA2.0380.485 NA
AFP (ng/ml)> 25vs. ≤ 252.0890.0482.0830.0671.5170.127 NA
Tumor size (cm)> 5vs. ≤ 51.1190.719 NA1.3300.288 NA
Tumor multiplicityMultiplevs. Solitary5.2530.0016.3500.0028.679<0.0017.6970.001
Vascular invasionPresentvs. Absent4.4000.0060.8200.7868.678<0.0014.9580.007
TNM stageIII+IVvs. l+II2.3770.0081.2700.5502.0520.0141.1770.638
FoxP3Highvs. Low2.4020.0042.3150.0111.9160.0121.9440.012

Phenotypic and functional characteristics of FoxP3+ Tregs in HCC patients

To further characterize the properties of FoxP3+ Tregs in HCC patients, CD4+CD25+FoxP3+ T cells derived from PBMCs (normal control, NC, n = 48; HCC patients, Group 2, n = 64), as well as from NILs and corresponding TILs (Group 3, n = 18), were analyzed by FACS. We observed that the percentage of circulating FoxP3+ Tregs among the CD4+ T cell population significantly increased in HCC patients compared with NC (mean ± SE, 5.7 ± 0.3% vs. 4.3 ± 0.2%, p < 0.001) (Figs. 2a and 2b). Additionally, FoxP3+ Treg frequency were significantly evaluated in TILs than corresponding NILs (14.2 ± 1.5% vs. 5.3 ± 0.5%, paired t test, p < 0.001) (Figs. 2a and 2b).

Figure 2.

Phenotypic and functional characteristics of FoxP3+ Tregs in HCC patients. Tregs (CD4+CD25+FoxP3+ T cells) were gated from CD3+CD4+ T cells by FACS analysis. (a) Representative plots and (b) statistical analysis showed that the frequency of FoxP3+ Tregs was higher in HCC patients, especially among TILs. (a,c) The mean fluorescence intensity (MFI) of FoxP3 in Tregs from the 4 groups studied (n = 13 for each group). The percentages represent the frequency of FoxP3+ Tregs, and the numerical values represent the MFI of FoxP3 in Tregs in (a). Data in (b) are expressed in box plots, in which the horizontal lines illustrate the 25th, 50th and 75th percentiles. (d) Representative CTLA-4, CD62L and CD45RO expression profiles in Tregs from the 4 groups studied. Purified CD4+CD25high T cells (Tregs) from TILs or NILs, (e) which were mostly FoxP3+ cells (low panel, thick lines), (f) could similarly inhibit the proliferation and IFN-γ production of autologous CD4+CD25 T cells (T) in a dose-dependent manner (n = 3 for each group). Proliferation was measured by BrdU incorporation and IFN-γ secretion was detected by ELISA. #Paired t test. *p < 0.05; **p < 0.001, compared with controls (Tregs:T = 0:1). Data in c and f represent the mean ± SE.

The mean fluorescence intensity (MFI) of FoxP3 in FoxP3+ Tregs derived from TILs was also significantly higher than that from corresponding NILs (paired t test, p = 0.002) or PBMCs (p = 0.017) (Figs. 2a and 2c). Moreover, most FoxP3+ Tregs among NILs and TILs exhibited an activated memory phenotype (CD45RO+CD62L), with the lowest CD62L expression level in FoxP3+ Tregs from TILs (Fig. 2d). We also observed variable CTLA-4 expression by FoxP3+ Tregs from different subsets (TILs > NILs > PBMCs) (Fig. 2d).

We also found that CD4+CD25high T cells (Tregs) from TILs or NILs, which mostly expressed FoxP3 and proliferated poorly, similarly inhibited the proliferation and IFN-γ secretion of autologous CD4+CD25 T cells (T) from the same region in a dose-dependent manner (n = 3, p < 0.001) (Figs. 2e and 2f). These data indicated that FoxP3+ Tregs with immunosuppressive properties were concentrated in the TILs of HCC patients.

Intratumoral prevalence of FoxP3+ Tregs correlated with high intratumoral Mφ density in HCC patients

The earlier results suggested that the tumor microenvironment may contribute to the increased presence of FoxP3+ Tregs at the tumor site. To test this hypothesis, we assessed the association between clinical characteristics and increased intratumoral FoxP3+ Treg levels using χ2 test. This analysis revealed that the intratumoral prevalence of FoxP3+ Tregs was correlated with the presence of cirrhosis (p = 0.010) and later TNM stages (p = 0.029) (Table III), indicating that the accumulation of these cells is associated with disease progression.

Table III. Association of Intratumoral FoxP3+ Tregs and Other Variables
VariablesIntratumoral FoxP3+ cellsp
≤ Median1> Median
  • One hundred thirty one HCC patients were enrolled.

  • 1

    Median level of intratumoral FoxP3+ cells was 1.05 cells/field.

  • 2

    Two groups were divided by the median value. The number follow “≤” or “>” represented the median value. χ2 test or Fisher exact test.

Age (year)2≤ 4831360.335
> 483529 
GenderMale58550.588
Female810 
HBsAgNegative670.748
Positive6058 
CirrhoslPresent50600.010
Absent165 
Child-Pugh scoreA64600.425
B25 
AFP (ng/ml)≤2522150.192
>254450 
Tumor size (cm)≤519180.889
>54747 
Tumor multiplicitySolitary56520.466
Multiple1013 
Vascular invasionPresent280.095
Absent6457 
TNM stageI+II53410.029
III+IV1324 
Numbar of CD68+ cells in situ tumor2≤8342270.011
>832438 

APCs are crucial for initiating and maintaining T-cell immunity and Mφ markedly outnumber other APCs in tumor tissues.27, 28, 32 Intratumoral Mφ counts were divided along the median value of CD68+ cells/field (median, 83 cells/field; range, 15–367 cells/field), and the association of the 2 resultant groups with increased intratumoral FoxP3+ Treg levels was assessed by χ2 test. The results showed that there was a positive correlation between the 2 cell types (Table III). Moreover, patients in Group 1 (n = 131) with an intratumoral Mφ density above the median usually had more intratumoral FoxP3+ Tregs (4.04 ± 0.63 cells/field, n = 69) than that with Mφ counts below the median (1.56 ± 0.24 cells/field, n = 62, p < 0.001) (Figs. 3a and 3b). In addition, confocal microscopic analysis revealed that intratumoral FoxP3+ Tregs were usually adjacent to CD68+ Mφ (Fig. 3c), suggesting an interaction between FoxP3+ Tregs and Mφ at the tumor site. Furthermore, these results implied that intratumoral Mφ may be involved in elevating intratumoral FoxP3+ Treg levels in HCC patients.

Figure 3.

Association between intratumoral FoxP3+ Tregs and Mφ in HCC patients (n = 131, Group 1). (a) Representative images and (b) statistical analysis of FoxP3+ Treg density in adjacent sections showed that patients with an intratumoral Mφ density above the median (CD68+ high) tended to have more intratumoral FoxP3+ Tregs than those patients below the median (CD68+ low). The median level of intratumoral Mφ density was 83 cells/field. Positive cells were stained brown, as indicated by arrows. (c) Representative pictures of FoxP3+ Tregs (red) and CD68+ Mφ (green) in the intratumoral region (n = 3).

In vivo depletion of tissue Mφ decreased the frequency of FoxP3+ Tregs at the tumor site

To confirm whether Mφ are involved in increasing the prevalence of FoxP3+ Tregs at the tumor site, we depleted liver Mφ in vivo as described under Material and Methods section. As expected, the FoxP3+ Treg frequency among liver CD4+ cells was augmented in hepatoma-bearing mice (mean ± SE, 9.25 ± 1.65% in tumor vs. 5.4 ± 0.85% in control, n = 3, p = 0.03) (Figs. 4a and 4b). This increase was partially attenuated by the depletion of tissue Mφ in the tumor (9.25 ± 1.65% in untreated tumor vs. 6.27 ± 0.78% in GdCl3-treated tumor, n = 3, paired t test, p = 0.01) (Figs. 4a and 4b). In contrast, treatment with GdCl3 did not affect the proportion of FoxP3+ Tregs in the liver tissues of control mice (Figs. 4a and 4b). These data indicated that Mφ were involved in the accumulation of FoxP3+ Tregs at the tumor site.

Figure 4.

Depletion of tissue Mφ could reverse the intratumoral increase in FoxP3+ Tregs in vivo. Hepatoma (Hepal-6) was inoculated into the liver of C57BL/6 mice. Five days later, the mice were left untreated or injected intraperitoneally with GdCl3 (10 mg/kg) every 5 days until day 20. Thereafter, the liver lymphocytes were isolated and analyzed by FACS. Representative plots (a) and statistical analysis (b) of the Treg frequency among CD4+ cells in the livers of hepatoma-bearing mice (tumor), normal mice (control), untreated mice (solid bar) or mice treated with GdCl3 (blank bar) are shown (n = 3 for each group). *p < 0.05. #Paired t test, p < 0.05. Data in (b) represent the mean ± SE.

Mφ exposed to TSNs from hepatoma-derived cell lines increased the population of FoxP3+ Tregs

The results described earlier suggested that intratumoral Mφ increase the population of FoxP3+ Tregs in HCC. Therefore, we investigated the effect of TSNs-exposed Mφ on FoxP3+ Treg frequency using an MLR assay, as described under Material and Methods section. Briefly, CD14+ monocytes were cultured for 7 days with medium alone (Mφ) or with TSNs from hepatoma-derived cell lines (HepG2-Mφ and Huh7-Mφ) or a leukemia cell line (THP1-Mφ), followed by coculture with autologous responder cells. As shown in Figure 5a and Supplementary Table 1, when cocultured with CD14+-depleted PBMCs during stimulation, HepG2-Mφ increased the FoxP3+ Treg frequency among CD4+ T cells compared with either THP1-Mφ (mean ± SE, 4.38 ± 0.24% vs. 2.54 ± 0.26%, n = 5, p < 0.001) or Mφ (4.38 ± 0.24% vs. 1.98 ± 0.25%, n = 5, p < 0.001) (Fig. 5a). More IL-10 secretion were detected in cocluture with HepG2-Mφ as well (HepG2-Mφ, 78.7 ± 4.1 pg/ml vs. THP1-Mφ, 12.8 ± 0.6 pg/ml or vs. Mφ, 9.1 ± 0.4 pg/ml, p < 0.001) (Fig. 5a). When purified CD3+CD25 T cells were used as responder cells, similar results were found for FoxP3+ Treg frequency (HepG2-Mφ, 10.9 ± 0.47% or Huh7-Mφ, 9.53 ± 0.37% vs. THP1-Mφ, 3.13 ± 0.59% or vs. Mφ, 2.37 ± 0.19%, n = 6, p < 0.001) (Fig. 5b and Supplementary Table 1) and IL-10 secretion (HepG2-Mφ, 199 ± 29 pg/ml or Huh7-Mφ, 173 ± 33 pg/ml vs. THP1-Mφ, 37 ± 1 pg/ml or vs. Mφ, 13 ± 1 pg/ml, p < 0.001) (Fig. 5b). It should be noted that coculture of HepG2-Mφ with purified CD3+CD25 T cells significantly increased IL-10 production in the supernatants, although HepG2-Mφ alone could release substantial amounts of IL-10 (199 ± 29 pg/ml vs. 70 ± 3 pg/ml, p < 0.001, Fig. 5b). Furthermore, the augmented FoxP3+ Treg population consisted predominantly of CD4+CD25+FoxP3+ T cells with a typical Treg phenotype (Fig. 5c). TSNs from hepatoma-derived cell lines alone were unable to increase the FoxP3+ Treg frequency and IL-10 secretion (Supplementary Fig. 1). These data indicated that Mφ exposed to TSNs from hepatoma-derived cell lines could increase the FoxP3+ Treg population.

Figure 5.

Mφ exposed to TSNs from hepatoma-derived cell lines augmented the FoxP3+ Treg population. Human monocytes were cultured for 7 days with medium alone (Mφ) or with TSNs from a leukemia cell line (THP1-Mφ) or hepatoma cell lines (HepG2-Mφ, Huh7-Mφ). Thereafter, Mφ or TSNs-exposed Mφ were cocultured with autologous responder cells and stimulated with anti-CD3/CD28 for 4 days. Statistical analysis showed that Mφ exposed to TSNs from hepatoma-derived cell lines (HepG2-Mφ or Huh7-Mφ) increased the FoxP3+ Treg frequency among CD4+ T cells more and secreted more IL-10 than either Mφ or THP1-Mφ when cocultured with autologous CD14+-depleted PBMCs (a) (n = 5) or CD3+CD25 T cells (b) (n = 6). FoxP3+ Treg frequency and IL-10 secretion when cultured HepG2-Mφ, Mφ or CD3+CD25 T cells alone upon anti-CD3/CD28 stimulation were also shown (b). Representative histograms (c) revealed that CD4+FoxP3+ T cells exhibited a regulatory phenotype after coculture. **p < 0.001, compared with responder cells cocultured with Mφ or THP1-Mφ. Data in (a) and (b) represent the mean ± SE.

Mφ exposed to TSNs from hepatoma-derived cell lines increased the FoxP3+ Treg population, partially via IL-10.

The earlier results, together with our previous data showing that TSNs-exposed Mφ and in situ tumors may secrete IL-10,25 imply that Mφ may affect the FoxP3+ Treg population via this cytokine. In support of this hypothesis, blocking mAbs against IL-10 partially attenuated the typical increase in FoxP3+ Treg frequency resulting from coculture of CD3+CD25 cells with autologous HepG2-Mφ upon anti-CD3/CD28 stimulation (anti-IL-10, 5.7 ± 0.9% vs. medium, 10.9 ± 1.1%, n = 3, p = 0.002 or vs. anti-IgG1, 10.8 ± 0.4%, n = 3, p = 0.006) but had little effect on coculture with Mφ (anti-IL-10, 2.2 ± 0.2% vs. medium, 2.7 ± 0.1% or vs. anti-IgG1, 2.6 ± 0.1%, n = 3, p > 0.05) (Fig. 6). In addition, supernatants (SNs) from HepG2-Mφ alone could only cause a marginal increase in the FoxP3+ Treg frequency (Supplementary Fig. 2), which could be due to the facts that IL-10 production was significantly increased in supernatants from coculture of HepG2-Mφ with purified CD3+CD25 T cells than that from HepG2-Mφ alone (Fig. 5b). Therefore, these results indicated that IL-10, attributed to Mφ exposed to TSNs from hepatoma-derived cell lines, is involved in amplifying the FoxP3+ Treg population.

Figure 6.

Mφ exposed to TSNs from hepatoma-derived cell lines could increase the FoxP3+ Treg population, partially through IL-10. CD3+CD25 T cells were cocultured for 4 days with autologous HepG2-Mφ or Mφ, with or without blocking mAbs against IL-10 or IgG1 (5 μg/ml), and in the presence of 2 μg/ml anti-CD3 mAbs plus 1 μg/ml anti-CD28 mAbs. Statistical analysis of the FoxP3+ Treg frequency among CD4+ T cells is shown. *p < 0.05, compared with medium (solid bar) or anti-IgG1 (blank bar). Data represent the mean ± SE from 3 experiments.

Discussion

Treg-mediated immunosuppression is a crucial strategy of tumor immune evasion and a main obstacle for successful cancer immunotherapy.23 However, the source of Tregs and their contribution to tumor progression in human cancers are still unclear. This study demonstrated that FoxP3+ Tregs with immunosuppressive properties were concentrated within HCC tumors and that the intratumoral prevalence of FoxP3+ Tregs was associated with disease progression and poor prognosis. We also found that the elevated intratumoral FoxP3+ Treg population was correlated with high-intratumoral Mφ density in HCC patients. Depletion of liver Mφ thus decreased the frequency of liver FoxP3+ Tregs in hepatoma-bearing mice. On the basis of further studies, we concluded that Mφ exposed to TSNs from hepatoma-derived cell lines augmented the FoxP3+ Treg population, partially via IL-10.

FoxP3 is a master regulator of Tregs and its expression in activated human CD4+CD25+ T cells is correlated with immunosuppressive function.33, 34 It has been suggested that once an activated T cell expresses FoxP3, pathways similar to those of Tregs are triggered, causing Treg-like activity.35, 36 In this study, we observed an increase in circulating and tumor-infiltrating FoxP3+ Tregs in HCC patients and a high occurrence of intratumoral FoxP3+ Tregs. This intratumoral prevalence of FoxP3+ Tregs was associated with lower OS and DFS and served as an independent predictor of poor HCC prognosis. The CD4+CD25high (FoxP3+) T cells derived from both TILs and NILs inhibited the activation of autologous CD4+CD25 T cells from the liver tissue of HCC patients, which is in agreement with previous reports demonstrating the suppressive effects of CD4+CD25+ Tregs on circulating T cells in HCC.16–19 In this study, we found that the levels of the immunosuppressive molecules CTLA-4 and FoxP333 were higher in Tregs derived from TILs than those from corresponding NILs or PBMCs. However, CD4+CD25high T cells from NILs or TILs were similar with regard to their immunosuppressive effects. These data suggested that the prevalence, rather than the superior immunosuppressive activity, of Tregs in the tumor microenvironment results in impairment of tumor-specific cell-mediated immunity in HCC, which is supported by a previous report on ovarian carcinoma.10

Tregs may expand or be induced from CD4+CD25 T cells at the tumor site mediated by tumor-associated DCs.23, 24, 37 Murine lamina propria Mφ may help to generate FoxP3+ Tregs in vitro.38 However, the correlation between intratumoral Mφ and FoxP3+ Tregs in situ in HCC patients and the mechanisms underlying this correlation are still undefined. In this study, we found that high-Mφ density was correlated with increased FoxP3+ Treg density in the intratumoral region of HCC patients. Moreover, depletion of liver Mφ partially attenuated this increase in FoxP3+ Treg frequency in the livers of hepatoma-bearing mice. We also found that Mφ exposed to TSNs from hepatoma-derived cell lines could significantly augment the FoxP3+ Treg population when cocultured with responder cells. However, in the absence of Mφ, TSNs alone could not increase the percentage of FoxP3+ Tregs upon anti-CD3/CD28 stimulation. These data indicated that HCC-associated immunosuppressive Mφ can increase the intratumoral FoxP3+ Treg population.

We have recently shown that the tumor microenvironment, including hepatoma, dynamically educates monocytes, resulting in the development of immunosuppressive HLA-DRlowIL-10high Mφ.25 This study provided evidence that IL-10 released from these immunosuppressive Mφ is involved in increasing the intratumoral prevalence of FoxP3+ Tregs. It should be noted that a blocking mAb against IL-10 only partially attenuated the HepG2-Mφ-mediated increase in FoxP3+ Treg frequency, indicating that the effect of HepG2-Mφ on FoxP3+ Treg frequency may involve additional factors or mechanisms. Indeed, the supernatants from HepG2-Mφ alone, which contained significantly less IL-10 than the coculture of HepG2-Mφ with CD3+CD25 cells, had only a marginal effect on FoxP3+ Treg induction. These data imply that interactions between HepG2-Mφ and CD3+CD25 T cells, including cell-cell contact, are crucial for FoxP3+ Treg induction by immunosuppressive tumor Mφ. Although the precise underlying mechanisms are still elusive, it has been recently reported that Tregs can trigger the IL-10 production by APCs, which in turn stimulate APC B7-H4 expression and render APCs immunosuppressive via increase the Treg suppressive capacity.39 Therefore, characterization of the factors responsible for the induction of FoxP3+ Tregs by suppressive Mφ may provide new avenues for development of novel immune-based therapies to enhance antitumor immunity in HCC.

HCC is typified by poor clinical prognosis and limited therapeutic options.1–3 Certain immunotherapies, which have demonstrated potent antitumor effects in vitro and in animal models, may have only limited effects on clinical HCC.5–8 This may be due to the documented immunosuppressive hepatic microenvironment in HCC. Hoechst et al.40 have recently characterized a novel population of myeloid derived suppressive cells, (CD14+HLA-DR−/low) which is also increased in blood and tumor of HCC patients. Interestingly, these CD14+HLA-DR−/low cells could promote the generation of Foxp3+ Treg population after their coculture with autologous T cells.40 Consistent with these findings, we have previously observed that tumor microenvironments, including hepatoma, can drive blood monocytes to develop into tolerogenic semimature DCs or suppressive Mφ, which exhibited an IL-12lowIL-10high phenotype with retained CD14 and reduced HLA-DR expression.25, 26 Most likely, the suppressive Mφ in this study belong to the population of CD14+HLA-DR−/low myeloid derived suppressive cells described by Hoechst et al. Therefore, a combination of Treg depletion and reversion of the abnormalities of myeloid-derived cells may be a novel, effective immunotherapy for reducing disease recurrence and prolonging survival in HCC patients.

Acknowledgements

The authors thank Ms. Jing Xu and Ms. Qiyi Zhao at Sun Yat-Sen University for their excellent technical assistance.

Ancillary