Journal of the National Cancer Institute Advance Access originally published online on July 24, 2007
JNCI Journal of the National Cancer Institute 2007 99(15):1188-1199; doi:10.1093/jnci/djm064
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© The Author 2007. Published by Oxford University Press.
ARTICLES |
Role of Tumor Endothelium in CD4+CD25+ Regulatory T Cell Infiltration of Human Pancreatic Carcinoma
Affiliations of authors: T cell Tumor Immunity group (DN, AB, PB) and Departments of Immunogenetics (ESP) and Cellular Immunology (VS), The German Cancer Research Center, Heidelberg, Germany; Department of Visceral Surgery, University Hospital of Heidelberg, Heidelberg, Germany (HSW, LG, DA, MK, MB, JW)
Correspondence to: Philipp Beckhove, MD, T cell Tumor Immunity group (D011), The German Cancer Research Center, INF280, 69120 Heidelberg, Germany (e-mail: p.beckhove{at}dkfz.de).
 |
ABSTRACT |
|---|
 Top Abstract Context and CaveatsPatients and MethodsResultsDiscussionFundingReferencesNotes |
|---|
Background: Regulatory T (Treg) cells have been detected in human carcinomas and may play a role in preventing the rejection of malignant cells.
Methods: We quantified Treg cells and the expression of the addressins and the respective ligands that attract them in blood and in human pancreatic tumors and adjacent nonmalignant tissues from 47 patients. The capacity of Treg cells to adhere to and transmigrate through autologous endothelial cells was tested in vitro using spheroid adhesion assays and in vivo using a xenotransplant NOD/SCID model and in the presence and absence of antibodies to addressins. All statistical tests were two-sided.
Results: More Treg cells infiltrated pancreatic carcinomas than adjacent nonmalignant pancreatic tissues (120 cells per mm2 versus 80 cells per mm2, difference = 40 cells per mm2, 95% confidence interval [CI] = 21.2 cells per mm2 to 52.1 cells per mm2; P<.001). In contrast to conventional CD4+ T cells, more blood-derived Treg cells adhered to (1.0% versus 5.2%, difference = 4.2%, 95% CI = 2.7% to 5.6%; P<.001) and transmigrated through (3332 cells versus 4976 cells, difference = 1644 cells, 95% CI = 708 cells to 2580 cells; P = .008) autologous tumor-derived endothelial cells in vitro and in vivo (458 cells versus 605 cells, difference = 147 cells, 95% CI = 50.8 to 237.2 cells; P = .04). Tumor-derived endothelial cells expressed higher levels of addressins—including mucosal adressin cell adhesion molecule-1 (MAdCAM-1), vascular cell adhesion molecule-1 (VCAM-1), CD62-E, and CD166—than endothelial cells from normal tissue. Experiments using antibodies to addressins showed that transmigration was mediated by interactions of addressins, including MAdCAM-1, VCAM-1, CD62-E, and CD166 with their respective ligands, 
7 integrin, CD62L, and CD166, which were expressed specifically on Treg cells.
Conclusions: Tumor-induced expression of addressins on the surface of endothelial cells allows a selective transmigration of Treg cells from peripheral blood to tumor tissues.
Prior knowledge Regulatory T cells may help tumor cells escape detection by the host's immune defense system. Study design Patient tumor and normal samples and in vitro and in vivo models were used to examine T-cell infiltration into pancreatic tumors and the mechanisms involved. Contribution More regulatory T cells were identified in pancreatic tumor samples than adjacent normal tissues; similar results were found in the in vitro and in vivo models. Addressins, proteins that target regulatory T cells, were highly expressed on the surface of the tumor-derived cells and were shown to be involved in the movement of regulatory T cells from the peripheral blood to tumor tissues. Implications Tumor-induced expression of addressins is a step in the infiltration of regulatory T cells into tumor tissue. Study limitations Other possible mechanisms, such as loss of conventional T cell homing receptors’ selective induction of addressins, were not specifically studied and cannot be ruled out.
|
Many cancer patients harbor tumor-specific lymphocytes. However, progressive tumor growth despite the infiltration of T cells demonstrates the failure of the immune system to efficiently combat malignant cells. Suppression of tumor-reactive helper (cluster of differentiation 4–positive, CD4+) and cytotoxic (CD8+) T cells by CD4+ suppressive regulatory T (Treg) cells may be a major tumor evasion strategy and an obstacle to successful tumor immunotherapy (1–3). Treg cells can be distinguished from other T-cell subsets by their constitutive expression of the interleukin (IL)-2 receptor alpha chain (CD25) and also by the expression of the transcription factor forkhead box P3 (FOXP3), a master regulator of Treg cell development.
Peripheral blood of patients with breast and pancreatic cancer (1), hepatocellular carcinoma (4), colorectal carcinoma (5,6), and lung cancer (6,7) has higher Treg cell counts than that of healthy individuals. Treg cells can also infiltrate the respective malignant tissues (8). Depletion of Treg cells before tumor challenge induces effective immunity in mice and spontaneous tumor rejection (9). Treg cells from solid tumors, like circulating Treg cells, secrete inhibitory cytokines (1) and exert T-cell suppressive activity (10). Treg cells can suppress the induction of the immune response in the draining lymph nodes and can block T-cell activity inside the target organ (11). Thus, inhibiting Treg cell infiltration into tumor tissues may be a promising strategy to improve T-cell–mediated tumor rejection.
Several chemokines have been described that may guide Treg cells to the inflamed organ or to the tumor. Chemokine receptors, such as C–C motif receptor (CCR)4 (12,13), CCR5 (14), and CCR8 (12,15) and homing receptors that are expressed on Treg cells can guide them to peripheral (16,17) or lymphoid (18) tissues. CCR4 may direct Treg cells to tumor tissues via the chemokine C–C motif ligand (CCL)22, which is expressed by tumor-infiltrating macrophages (10). CCR5 is selectively expressed by activated Treg cells and, in Listeria monocytogenes infection, can guide Treg cells to infected organs, such as the skin (19). CCR8 receptors can mediate Treg cell migration in response to skin or inflammatory tissue–derived CCL1 (12).
Although the infiltration of T cells into tissues is often regulated through interactions between chemokines and their receptors, it can also be regulated by interactions between T cells and endothelial cells. Specifically, T-cell infiltration is mediated by adressins that are expressed on the surface of endothelial cells, such as the selectins E- and P-selectin (CD62-E, CD62–P) and the immunoglobulin family members intercellular adhesion molecule (ICAM)-1, ICAM-2, mucosal adressin cell adhesion molecule-1 (MAdCAM-1), vascular cell adhesion molecule-1 (VCAM-1), and activated leukocyte cell adhesion molecule (ALCAM, CD166). The ligands for these addressins, including the adhesion molecules leukocyte function–associated-1 (LFA-1), CD24, and CD6, the selectin CD62-L, the selectin ligand P-selectin–binding glycoprotein-1 (PSGL1), and 1 and 7 integrins, are expressed in various combinations on the surface of different blood T-cell populations. The possible impact of the endothelial cell layer and particularly of these endothelial addressins and the impact of the expression of the respective T cell homing receptors on the migration of Treg cells into infected organs or tumor tissue have not yet been addressed. Using human pancreatic carcinomas as a model system, we analyzed whether Treg cell infiltration into tumor tissue is tumor selective and evaluated the role of addressins in the tumor endothelium in Treg cell recruitment.
|
Patients and Methods |
|---|
|
TopAbstractContext and CaveatsPatients and MethodsResultsDiscussionFundingReferencesNotes |
|---|
Tissue Samples
Pancreatic tissue and peripheral blood samples from 10 patients with histologically confirmed primary pancreatic carcinoma were collected during primary tumor resection (pancreatectomy). Nonmalignant pancreatic tissue from the same patients was also obtained during pancreatectomy and used as control tissue after pathologic exclusion of tumor infiltration. Peripheral blood samples were also obtained from 10 healthy donors. Tissue samples were either processed immediately or shock frozen in liquid nitrogen for immunohistology. Written informed consent was obtained from all participants, and the protocol was approved by the Ethical Committee of the University of Heidelberg.
Mice
Female 6- to 8-week-old immunodeficient NOD/SCID mice (total n = 45; Charles River Wiga GmbH, Sulzfeld, Germany) were housed under specific pathogen-free conditions at the animal facility of the German Cancer Research Center in accordance with institutional guidelines.
Isolation of T-cell Subsets
Mononuclear cells were isolated from peripheral blood samples by density gradient centrifugation using Ficoll (Biochrom, Berlin, Germany). CD3+ T cells were enriched by depleting contaminating cells using a T-cell–negative isolation kit (Dynal, Hamburg, Germany). CD4+CD25+ Treg cells and CD4+CD25– conventional T cells (Tcon) were then isolated using the magnetic bead–based CD4+CD25+ isolation kit (Dynal) that allows detachment of the beads from T cells through degradable DNA linkers. The purity of isolated cells was 85%–99%.
Isolation and Culture of Human Microvascular Endothelial Cells
Microvascular endothelial cells were isolated from samples of pancreatic tumors and from corresponding nonmalignant pancreatic tissues. Tissue samples were washed in phosphate-buffered saline (PBS; Invitrogen, Karlsruhe, Germany), dissected mechanically into small pieces (
3 mm2), and resuspended with endothelial cell basal medium (ECBM) + supplement (15% fetal calf serum, 2.5 ng/mL b-fibroblast growth factor, 10 µL/mL sodium-heparin, 1% penicillin/streptomycin; PromoCell, Heidelberg, Germany). The suspension was filtered through 40-µm cell strainers (Falcon BD, Heidelberg, Germany), and the resulting single cells were washed with PBS. Endothelial cells were magnetically isolated from this cell population using anti–CD31-Dynabeads (Dynal). Isolated endothelial cells were used immediately or transferred to gelatin-coated (2%) cell culture flasks (Biochrom) and cultured in supplemented ECBM until passage three. During endothelial cell culture, endothelial cells derived from primary pancreatic carcinomas were supplemented with 50 µg/mL of autologous tumor cell lysate to maintain the tumor endothelial phenotype during culture. Endothelial cells isolated from corresponding nonmalignant pancreatic tissue were supplemented with 50 µg/mL of autologous peripheral blood mononuclear cell (PBMC) lysate due to lack of control tissue lysates. Tumor tissue lysates were prepared by mechanical homogenization using an Ultra turrax T8 disperser (IKA, Werke Staufen, Germany). PBMC lysates were generated by five cycles of freezing in liquid nitrogen and thawing. Tumor and PBMC lysates were then centrifuged (20 minutes at 30000g) to remove cell debris and organelles. Protein concentrations of the lysate supernatants were determined by Bradford assay (BioRad, München, Germany). Endothelial cells were always washed carefully with PBS to remove traces of cell lysates before use in subsequent experiments.
Spheroid Adhesion Assay
Endothelial cells (2.0 x 104) that had been isolated from tumor or control pancreatic tissues and cultured in ECBM were cultured for 48 hours in spheroid medium (70% ECBM + supplement and 0.012 g/mL methylcellulose, PromoCell) together with either 50 µg/mL of autologous tumor lysate (tumor-derived endothelial cells) or autologous PBMC lysate (pancreas control tissue–derived endothelial cells), respectively, in nonadhesive polyethylene round-bottom 96-well plates (Biochrom, Berlin, Germany) to generate one adherent spheroid per well (10 wells per patient for control and tumor). Adherent spheroids were washed gently with PBS, and single control endothelium spheroids were removed and added to wells containing single tumor endothelium spheroids from the same patient. Then, 2.0 x 104 fluorescently labeled (carboxyfluorescein succimidyl ester, green) autologous Treg cells (Molecular Probes, Heidelberg, Germany) and fluorescently labeled (7-amino-4-chloromethylcoumarin, blue) Tcon cells (Molecular Probes) were added per well. After 4 hours, spheroids were washed with PBS, and numbers of tumor- and control spheroid–adherent Treg cells and Tcon cells were quantified by fluorescence microscopy. Tumor spheroids could be easily distinguished from control endothelium spheroids by their adherence to the well bottom. All experiments were performed in triplicate (n = 6 different experiments with six different donors).
In Vitro Transmigration Assay
Transwell membranes (3-µm pore size; 3420, Costar, Corning, NY) were coated for 30 minutes at 37 °C with fibronectin (5 µg/mL, Chemicon, Schwalbach, Germany). Isolated endothelial cells (3.0 x 104 per well) were added and cultured on the membranes together with 50 µg/mL of autologous tumor lysate (tumor-derived endothelial cells) or autologous PBMC lysate (pancreas control tissue–derived endothelial cells) in supplemented ECBM until they reached confluence (2 days). Endothelial cell layers were washed twice, and the medium in the lower chamber was supplemented with recombinant human stromal cell–derived factor 1 (100 ng/mL, CHM-262, TechnoGene Ltd, Rehovot, Israel) to establish a gradient for T-cell transmigration. Isolated autologous Treg or Tcon cell populations (30000 T cells per well) were then added to the upper chamber, and T cells that had transmigrated were quantified 24 hours later by using a Coulter counter (Beckmann Coulter GmbH, Krefeld, Germany). For blocking experiments, blood-derived T cells or endothelial cells were incubated before the cells were washed and used in the assay for 4 hours with 2–5 mg/mL of the following specific inhibitory monoclonal antibodies against cell adhesion molecules: goat anti-human CD62-E, ICAM-1, ICAM-2, MAdCAM-1, or CD166 or mouse anti-human CD6 (Santa Cruz Biotechnology, Heidelberg, Germany); rabbit anti-human CD62-P (US Biological, Hiddenhausen, Germany); goat anti-human VCAM-1 (R&D Systems, Wiesbaden-Nordenstadt, Germany); or mouse anti-human CD24 (Serotec, Oxford, U.K.) or with respective isotype antibodies (as specificity controls). All experiments were performed in triplicate (n = 4 independent experiments with four different donors).
In Vivo Transmigration Assays
In vivo transmigration chambers (3-µm pore) of 15 mm diameter were generated as follows: the bottoms of 0.5 mL safe-lock tubes (Eppendorf, Eppendorf, Germany) were excised, leaving approximately 5 mm of the upper part and the lid intact. Holes (8-mm diameter) were punched into the lid and closed firmly with 3-µm pore polycarbonate membranes (Corning, NY). Then, the bottoms were closed by silicon membranes (NeoLab, Heidelberg, Germany), and the chambers were sterilized in 70% ethanol for 24 hours. Endothelial cells (2 x 104–5 x 104 per well) from tumor or control pancreatic tissue were cultured in vitro on the micropore membrane until they reached confluence, as described above, and the chamber lids were closed. Immunodeficient NOD/Scid mice were anesthetized with Rompun (2%; Bayer, Leverkusen, Germany) and Ketanest (25 mg/mL; Parke-Davies, NY) in PBS in a ratio of 1:1:3 (vol/vol/vol). Two transmigration chambers coated with tumor-derived endothelial cells or control endothelial cells from the same patient were each implanted subcutaneously into the dorsal skin of anesthetized mice. The chambers were supplemented with either 50 mg/mL of autologous tumor lysate (chambers containing tumor-derived endothelial cells) or with an equal amount of autologous PBMC lysate (chambers containing control endothelial cells). After 7 days, when chamber membranes had become vascularized, 1.0 x 106 fluorochrome-labeled autologous T cells from the peripheral blood of the same patients were injected intravenously into each mouse. Mice were killed by CO2 asphyxiation 2–4 days later (this was enough time to allow T cells to infiltrate the chambers), after which time infiltrating cells were isolated from the chambers and transferred to glass slides by cytocentrifugation. Cells were then fixed in ice-cold acetone and stained with fluorescently labeled monoclonal mouse anti-human antibodies against CD3 (Invitrogen), CD4, CD8, and CD25 (Santa Cruz) (all diluted 1:100 in PBS), and stained cells were counted. To evaluate potential contributions of distinct homing receptors for T-cell transmigration through autologous endothelium into the transplanted chambers in vivo, T cells were cultured in the presence of 2–5 µg/mL of homing receptor–specific mouse monoclonal antibodies anti-CD166, anti-CD62L, and anti–7 integrin or respective isotype antibodies (as specificity control) for 4 hours before intravenous T-cell transfer. All analyses were performed in triplicate or duplicate mice (total n = 45) in 11 independent experiments with 11 different donors.
Proliferation Assays
Tcon cells from pancreatic cancer patients were incubated for 72 hours in 96-well plates (Nunc, Wiesbaden, Germany) that had been coated with mouse anti-human CD3 (0.5 µg/mL) and mouse anti-human CD28 (0.5 µg/mL) monoclonal antibodies alone or in coculture with autologous Treg cells (always 5.0 x 104 Tcon per well) at different ratios of Tcon:Treg (1:0, 4:1, 8:1, and 16:1). In addition, 5.0 x 104 Treg cells per well alone (0:1, negative control) and 1.0 x 105 Tcon per well alone (2:0, positive control) were cultured separately under the same conditions. After 72 hours at 37° C, [3H]thymidine at 1 µCi per well was added for an additional 16 hours of culture, and proliferation of T cells was measured by determining the amount of incorporated [3H] using a scintillation counter (Liquid Scintillation Counter [1450 MicroBeta] Perkin Elmer, Wellesley, MA), as described (20) in triplicate wells (n = 4 independent experiments with four different donors).
Flow Cytometry
Single-cell suspensions of tumor or control pancreatic tissue or PBMC (1.0 x 105–1.0 x 107 cells per well) were blocked with polyclonal human immunoglobulins (Endobulin, 2.5 mg/mL; Baxter Oncology, Frankfurt, Germany) and incubated with one of the following mouse anti-human monoclonal antibodies: anti–CD3-FITC, anti–CD3-PE, anti–CD4-FITC, anti–CD4-PE, anti–CD4-PECy5, anti–CD8-FITC, anti–CD8-PE, anti–CD31-FITC, anti–CD62L-PE, anti–PSGL1-PE, anti–CD25-FITC, or anti–LFA-1-FITC (all 1:20; BD Pharmingen, Heidelberg, Germany); mouse anti-human anti–1 integrin (1:100, Santa Cruz Biotechnology); or rabbit anti-human 7 integrin (1:100), goat anti-human CD62-E (1:100), goat anti-human ICAM-1 (1:100), goat anti-human ICAM-2, goat anti-human MAdCAM-1 (both 1:100), goat anti-human CD166 (1:100), mouse anti-human CD6 (1:100) (all Santa Cruz Biotechnology); rabbit anti-human CD62-P (1:200, US Biological); goat anti-human VCAM-1 (1:100, R&D Systems, Wiesbaden-Nordenstadt, Germany); or mouse anti-human CD24 (1:100, AbD Serotec, Oxford, U.K.) for 30 minutes on ice. Unconjugated antibodies were detected by respective chicken anti-mouse, chicken anti-goat, or chicken anti-rabbit AlexaFluor-594 secondary antibodies (1:400, Invitrogen). Dead cells, which were labeled with 1 µg/mL of propidium iodide (Abcam) immediately before flow cytometry, were excluded from analysis. Recordings were made from at least 1.0 x 105 cells on a FACS-Calibur flow cytometer (Becton Dickinson, Heidelberg, Germany) using FlowJo 4.3 software (TreeStar, San Carlos, CA).
Immunohistology
Pieces of freshly isolated tumor and control tissues were embedded in Tissue Tek embedding medium (Electron Microscopy Sciences, Hatfield, PA), snap frozen in liquid nitrogen, and stored at –80 °C until use. Cryosections (5 µm) were prepared from frozen tissue, fixed in ice-cold acetone, blocked with serum from the secondary antibody species (Santa Cruz Biotechnology), and incubated with one of the following primary antibodies: rabbit anti-human CD3 (SpringBioscience, Heidelberg, Germany), mouse anti-human CD4 (Santa Cruz Biotechnology), goat anti-human CD4 (Santa Cruz Biotechnology), mouse anti-human CD25 (DAKO, Hamburg, Germany) or rabbit anti-human CD25 (Santa Cruz Biotechnology), mouse anti-human CD31 (DAKO), rabbit anti-human CD62-P (US Biological), goat anti-human CD62-E (Santa Cruz Biotechnology), goat anti-human ICAM-1 (Santa Cruz Biotechnology), goat anti-human ICAM-2 (Santa Cruz Biotechnology), goat anti-human MAdCAM-1 (Santa Cruz Biotechnology), goat anti-human CD166 (Santa Cruz Biotechnology), goat anti–VCAM-1 (R&D Systems, Wiesbaden-Nordenstadt, Germany) (all diluted 1:100), or mouse anti-human FOXP3 (undiluted; kindly provided by Alison Banham, Oxford, U.K.). Antibody binding was detected with the appropriate secondary antibody: chicken-anti-mouse-AlexaFluor-594 (red; Invitrogen), goat-anti-mouse-AlexaFluor-350 (blue color; Invitrogen), chicken anti-goat AlexaFluor-594 (Invitrogen), or chicken anti-rabbit AlexaFluor-594 (Invitrogen) (all diluted 1:300). Slides were washed in PBS several times, and 4',6-diamidino-2-phenylindole (DAPI staining solution; Hoechst, Darmstadt, Germany) was added in a dilution of 1:4000 to detect nuclei. Tissue autofluorescence was blocked with CuSO4 solution (1–10 mM CuSO4 in 50 mM ammonium acetate buffer [pH 5.0]; Sigma, Deisenhofen, Germany), and slides were covered with glycerin–gelatine (Merck, Darmstadt, Germany). Slides were evaluated by counting labeled cells (for enumeration of total T cells, CD3+; Tcon, CD4+CD25–; and Treg, CD4+CD25+FOXP3+ cells per mm2) or by automatic determination of stained areas using AnalySIS software (Olympus Soft Imaging Solutions, Muenster, Germany). Quantitative analysis of slides was always based on a minimum of triplicate sections per sample (from n = 8–10 different donors).
Immunocytochemistry
T-cell subpopulations that were isolated from pancreatic tumor tissue or from corresponding normal pancreas tissue or re-isolated from transplanted migration chambers were cytocentrifuged onto glass slides, fixed with ice-cold acetone, and stained with one of following monoclonal antibodies: mouse anti-human CD3 (Invitrogen), rabbit anti-human CD3 (SpringBioscience), mouse anti-human CD4 (Santa Cruz Biotechnology), rabbit anti-human CD8 (Sant Cruz), mouse anti-human CD8 (Santa Cruz Biotechnology), mouse anti-human CD25 (DAKO), and rabbit anti-human CD25 (Santa Cruz Biotechnology) at dilutions of 1:100; and mouse anti-human IL-10 (Bender Med Systems GmbH, Vienna, Austria) or rabbit anti-human transforming growth factor-b1 (TGF-b1) (Santa Cruz Biotechnology) at dilutions of 1:50. Chicken anti-rabbit AlexaFluor-594 and chicken anti-mouse AlexaFluor-488 (both 1:400, Invitrogen) were used as secondary antibodies.
Statistical Analyses
P values were calculated by using two-sided Student's t test; P less than .05 was considered to be statistically significant. For comparison of corresponding tumor and normal tissues from the same patient, a paired t test was used instead. For certain analyses for which nonmalignant and malignant tissues were available from the same patient, we calculated the covariance (r) by its empirical mean. In all cases, r ranged between –.08 and +.11. We then used product–moment correlation to ensure that r was not statistically different from 0.
|
Results |
|---|
|
TopAbstractContext and CaveatsPatients and MethodsResultsDiscussionFundingReferencesNotes |
|---|
Treg Cell Infiltration in Pancreatic Tumor Tissue
We used immunohistology to evaluate absolute numbers of tissue-infiltrating total T cells and the Treg cell subpopulation of pancreatic carcinomas and nonmalignant pancreas tissue (Fig. 1, A–C). In addition, we used flow cytometry and fluorescence-activated cell sorting (FACS) to quantify the proportions of CD3+ CD4+ and CD8+ T cells among total CD3+ T cells and of the CD25+ subset among CD4+ T cells in peripheral blood and single-cell suspensions of pancreatic carcinomas and corresponding control pancreatic tissue (Fig. 1, B and C). More CD3+ T cells infiltrated pancreatic carcinoma tissue than nonmalignant pancreas tissue (means: 120 cells per mm2 versus 80 cells per mm2, difference = 40 cells per mm2, 95% confidence interval [CI] = 21.1 cells per mm2 to 52.1 cells per mm2; P<.001; Fig. 1, B). Interestingly, the increased infiltration of CD3+ T cells into tumor tissue was mainly ascribed to a tumor-selective CD4+ T-cell subset (tumor versus control, mean = 1.48% of total lymphocytes versus 0.61% of total lymphocytes, difference = 0.87%, 95% CI = 0.27% to 1.37%; P = .009), because proportions of tissue-infiltrating CD8+ T cells in pancreatic tumor tissues and control tissues were similar (Fig. 1, B). Proportions of CD25+ Treg cells among total CD4+ T cells in peripheral blood of pancreatic carcinoma patients were slightly but statistically significantly higher than in healthy donors (mean = 5.0% versus 2.1%, respectively; difference = 2.9%, 95% CI = 2.0 to 3.9; P<.001). In addition, the percentage of CD25+ Treg cells among total CD4+ T cells was higher in pancreatic carcinoma tissue than in paired nonmalignant pancreas tissue (mean = 39.2% versus 7.9%, respectively; difference = 31.3%, 95% CI = 14.6% to 33.4%; P = .02) (Fig. 1, C).
|
We evaluated the number of FOXP3+ Treg cells among CD4+CD25+ T cells using immunohistology. The percentage of CD4+CD25+ T cells expressing FOXP3 that had infiltrated pancreatic tumor tissues was much higher than that of pancreatic control tissues (mean = 74.6% versus 53.5%, difference = 21.1%, 95% CI = 9.1 to 33.1; P = .002; Fig. 1, C). The absolute numbers of FOXP3+ Treg cells were also statistically significantly higher in pancreatic carcinoma tissues than in corresponding control pancreas tissues (mean = 15.3 cells per mm2 versus 2.5 cells per mm2, difference = 12.8 cells per mm2, 95% CI = 7.6 to 17.9; P<.001; Fig. 1, C). CD4+CD25+ T effector cells (Teff), although clearly a minor component of this cell population, were also statistically significantly increased in tumor tissue compared with normal pancreatic tissue (mean = 4.9 cells per mm2 versus 2.6 cells per mm2, difference = 2.3 cells per mm2, 95% CI = 0.3 to 4.2; P = .025; data not shown).
To assess whether cells with a Treg cell phenotype also displayed functional Treg cell characteristics, we tested their capacity to suppress the proliferation of autologous cocultured CD4+CD25- Tcon cells after activation with polyclonal CD3 and CD28 antibodies. Blood-derived Treg cells efficiently suppressed Tcon cell proliferation in a dose-dependent manner (Fig. 2, A).
|
To evaluate the functional potential of tumor-infiltrating Treg cells, we measured the intracellular production of the Treg cell–associated cytokines, IL-10 and TGF-
1, by immunocytochemistry (Fig. 2, B). The majority of pancreatic carcinoma–derived Treg cells (TregTu) expressed IL-10 (55%) and TGF-1 (71%). In contrast, Tcon from tumor tissue (TconTu) as well as CD4+CD25+ Treg cells derived from nonmalignant pancreatic tissue (TregCon) contained statistically significantly lower proportions of IL-10–positive cells (TregTu versus TconTu, 55% versus 20%, difference = 35%, 95% CI = 25.5% to 45.1%; P<.001 and TregTu versus TregCon, 55% versus 35%, difference = 20%, 95% CI = 2.2% to 39.8%; P = .036) and TGF-1+ cells (TregTu versus TconTu, 71% versus 25%, difference = 46%, 95% CI = 37.9% to 57.5%; P<.001 and TregTu versus TregCon, 71% versus 33%, difference = 38%, 95% CI = 19.9% to 56.8%, P = .005) (Fig. 2, B). This result is consistent with the finding in Fig. 1 that the proportions of FOXP3+ Treg cells among CD4+CD25+ T cells were much higher in pancreatic carcinomas than in nonmalignant control pancreas tissues. Expression of Adhesion Molecules by Tumor-Derived Endothelial Cells and Treg Cells
Selective accumulation of Treg cells in tumor tissues could be due to an increased infiltration rate. To test this hypothesis, we characterized potential differences between the vasculature of pancreatic tumors and nonmalignant pancreas tissue (Fig. 3). We detected statistically significantly higher proportions of CD31+ (i.e., vascular endothelial) cells in tumor tissues than in nonmalignant tissues by flow cytometry (percentage of total cells, mean = 2.9% versus 0.9%; difference = 2.0%, 95% CI = 1.47 to 2.49, P = .005) and by immunohistology (percentage of stained area, mean = 2.3% versus 0.6%; difference = 1.7%, 95% CI = 1.14 to 1.94, P = .004) (Fig. 3, C). In addition, not only did the tumor tissue contain more CD31-positive cells but also CD31+ endothelial cells from such vessels displayed an increased expression of a variety of addressins, including E-selectin (CD62-E), ICAM-1 and -2, MAdCAM-1, VCAM-1, and CD166 (Fig. 3, D).
|
The differences in addressin expression observed between cultured endothelial cells that were isolated from tumor and nonmalignant pancreatic tissue were partly conserved for at least two to three passages (data not shown). However, the differences could be maintained stably by adding 50 µg/mL of protein from autologous pancreatic tumor cell lysate to the culture medium (Fig. 4, A and B), suggesting that factors derived from the tumor microenvironment account for the increased expression of endothelial addressins.
|
In peripheral blood, however, we did not find substantial differences in expression of addressins between Treg cells from healthy donors and pancreatic cancer patients (Supplementary Fig. 1, available online). Nor did we find differences in expression of these molecules on Treg cells and Tcon in the blood from pancreatic cancer patients (Fig. 4, C). In contrast, the expression of receptors that bind to the addressins and act as homing receptors on tumor-infiltrating Treg cells and Tcon differed greatly (Fig. 4, C). Treg cells isolated from tumor tissue showed enhanced expression of the
7 integrin CD62L and of CD166 and reduced expression of CD24 compared with Tcon. In contrast, 7 integrin and CD166 were increased in the Treg cells from tumor tissues compared with Treg cells from blood. Capacity of Treg Cells to Bind to and Transmigrate Through Tumor Endothelium In Vitro
To evaluate the capacity of peripheral blood–derived Treg cells to adhere to autologous tumor–derived endothelial cells, we established small endothelial cell spheroids of approximately 300-mm diameter that allow the adherence to and subsequent quantification of bound, fluorescently labeled T-cell populations (Fig. 5, A). Endothelial cells were isolated from nonmalignant or autologous pancreatic tumor tissue and used to generate spheroids. Treg and Tcon cells showed similar low binding capacity to endothelial cell spheroids that were derived from nonmalignant pancreatic tissue (Fig. 5, B). In contrast, Treg cells but not Tcon cells showed a strongly enhanced capacity to bind to tumor-derived endothelial cell spheroids (percent adherent cells, TconCon, mean = 1.2%, TconTu, mean = 1.0%, TregCon, mean = 0.93%, TregTu, mean = 5.2%; TregTu versus TconCon, difference = 3.98%, 95% CI = 3.04% to 4.93%; TregTu versus TconTu, difference = 4.2%, 95% CI = 2.7% to 5.6%; TregTu versus TregCon, difference = 4.25%, 95% CI = 3.12 to 5.38, all P<.001).
|
We next evaluated the capacity of separated blood-derived T-cell subsets to transmigrate in response to an SDF-1 (CXCL12) gradient through monolayers of autologous endothelial cells that were derived from pancreatic carcinomas or corresponding control pancreatic tissue. We detected a statistically significantly increased transmigration of both CD4+ and CD8+ T cells through tumor endothelium (Fig. 6, A). Transmigration of CD4+ T cells through tumor endothelium strongly exceeded that of CD8+ T cells (3609 cells versus 2009 cells, respectively; difference = 1600 cells, 95% CI = 1300 cells to 1901 cells, P<.001), whereas such a difference was not observed with regard to transmigration through normal endothelial cells (Fig. 6, A). In contrast, we observed a selective increase in the transmigration of Treg cells through autologous tumor endothelium (Treg versus Tcon, 4976 cells versus 3332 cells, difference = 1644 cells, 95% CI = 708 cells to 2580 cells, P = .008) (Fig. 6, B).
|
Role of Adhesion Molecules in Treg Cell Transmigration Through Tumor Endothelium
To determine which addressin–homing receptor interactions could account for the selectively increased transendothelial migration of patient-derived peripheral blood Treg cells, we used antibodies in the transmigration assay to block the function of a variety of adhesion molecules that were differentially expressed by tumor endothelial cells or by tumor-infiltrating Treg cells. Compared with using a mixture of respective isotype antibodies, specific blocking of these addressins had little effect on endothelial transmigration of Tcon through both tumor-derived and control endothelium (Fig. 6, C). In contrast, Treg cell transmigration through tumor-derived endothelium was strongly reduced by blocking CD62-E, MAdCAM-1, VCAM-1, or CD166, whereas blocking of CD62-P or ICAM-1 or -2 exerted less of an inhibitory effect on Treg cell transmigration (Fig. 6, C, upper panel). The inhibition of Treg cell transmigration through tumor endothelium was also statistically significantly higher than the inhibition of respective addressins on control endothelium (Fig. 6, C; compare top and bottom panels).
We next performed a similar in vitro assay with patient-derived peripheral blood T cells, in which we blocked those adressin ligands that showed differential increased expression on ex vivo isolated tumor-infiltrating Treg cells (Fig. 4, C). Antibodies against CD166 (homophilic binding to CD166), CD62L (binding to E-selectin, MAdCAM-1, and VCAM-1), and 7 integrin (binding to MAdCAM-1 and VCAM-1) all inhibited Treg cell transmigration through tumor endothelial cells statistically significantly more strongly than they inhibited transmigration through control endothelial cells (Fig. 6, D). Inhibition of Treg cell transmigration through tumor endothelium but not control endothelium was also statistically significantly enhanced compared with Tcon (Fig. 6, D).
Thus, the blocking of CD166, CD62E, VCAM-1, and MAdCAM-1, whose expression was selectively increased on tumor endothelium, as well as their ligands, whose expression was selectively increased on tumor-infiltrating Treg cells, led to a selective and strong reduction of Treg cell transmigration through tumor endothelium. Transmigration of Treg cells through control endothelium or transmigration of Tcon through tumor and control endothelium was only little affected by blocking antibodies.
Transendothelial Treg Cell Migration In Vivo
To evaluate whether the observed increase of Treg cell transmigration through tumor endothelium in vitro could also occur in vivo, we established a xenotransplant mouse model for human T-cell transmigration through human autologous endothelium (Fig. 7, A). CD3+ T cells predominantly infiltrated chambers containing tumor-derived endothelial cells (Fig. 7, B). Enhanced CD3+ accumulation was mainly based on infiltration by CD4+ T cells, thereby closely resembling tumor-selective CD4+ T-cell infiltration of primary pancreatic carcinomas in situ. Among CD4+ T cells, only the Treg cell subset showed a selective accumulation in chambers containing tumor endothelium (number of infiltrating CD4+ cells: Treg+ ECTu versus Treg + ECCon, 605 cells versus 382 cells, difference = 223 cells, 95% CI = 571 cells to 444 cells, P = .006; and Treg+ ECTu versus Tcon+ ECTu, 605 cells versus 458 cells, difference = 147 cells, 95% CI = 50.8 cells to 237.2 cells, P = .04) (Fig. 7, C). Thus, the increased migration of Treg cells through tumor endothelium in vitro was also observed in vivo. This finding supports and extends our hypothesis that the tumor endothelium promotes selective Treg cell accumulation in pancreatic cancer lesions.
|
We undertook a last set of experiments to evaluate whether T-cell transmigration through autologous endothelial cell layers in vivo could also be inhibited by specific blocking of T-cell homing receptors. We therefore treated in vitro separated Tcon or Treg cell subsets by inhibitory antibodies to the T-cell homing receptors CD166, CD62L, and
7 integrin and injected the T cells intravenously into NOD/SCID mice bearing xenotransplanted microchambers. We found that antibody blocking had less impact on transmigration of Tcon through either tumor endothelium or control endothelium than on the transmigration of Treg cells (especially through tumor endothelium) (Fig. 7, D). Blocking 7 integrin exerted the highest (50%) transmigration inhibition of Treg cells through tumor endothelium. |
Discussion |
|---|
|
TopAbstractContext and CaveatsPatients and MethodsResultsDiscussionFundingReferencesNotes |
|---|
We here demonstrate that Treg cells, but not CD4+ Tcon or CD8+ T cells, accumulate selectively in tumor tissue of pancreatic carcinoma but not in nonmalignant pancreas tissue of the same patients. These Treg cells seemed to be functionally active in situ because they secreted IL-10 and TGF-
1 after their isolation. The observed selective recruitment of Treg cells into tumor tissue was reflected by an increased Treg cell adhesion to and transmigration through tumor endothelium in vitro and in vivo. This transmigration was mediated by addressins, such as CD62-E, MAdCAM-1, VCAM-1, and CD166, which were overexpressed on tumor endothelium, and their ligands, CD166, CD62-L, and 7 integrin, which were overexpressed on tumor-infiltrating Treg cells. Therefore, our data demonstrate a previously unrecognized role of the tumor endothelium for the selective recruitment of Treg cells. We based our study on a careful analysis of primary human tissues and used in vitro and in vivo models to evaluate potential mechanisms underlying the observed situation in clinical samples. Because the experimental models chosen for our study do not exactly resemble physiologic conditions, our conclusion that tumor-driven changes of tumor endothelium play an important role in the selective recruitment of Treg cells into human pancreatic carcinomas remains unproven. This unavoidable limitation of an in vitro study with human material is to our point of view justified by its potential clinical relevance. However, an appropriate mouse model of pancreatic carcinoma should be established in the future to formally confirm our data.
We observed that the overall increased infiltration of tumor tissue by CD3+ TIL was due mainly to infiltration of Treg cells, suggesting that a selective mechanism for Treg cell accumulation is driven by tumor tissue. This assumption was substantiated by our finding of pronounced differences between the vasculature of malignant pancreatic tissue and of adjacent nonmalignant areas in the same patients. We detected an increased density of blood vessels in tumor tissue, together with increased expression of a broad variety of T-cell transmigration-relevant addressins (CD62-E, ICAM-1 and -2, MAdCAM-1, VCAM-1, or CD166). Changes in phenotype may be induced by the tumor microenvironment and may thereby provide a molecular basis for a selective recruitment of certain T-cell subsets (21–25). In addition, addressin expression on endothelial cells has been described to be modulated by tumor-derived factors, such as TGF-1 (26,27).
With respect to T-cell homing receptor expression, we detected no statistically significant differences between peripheral blood–derived Treg cells and Tcon cells that could explain their differential migratory capacities. Nevertheless, we cannot exclude the possibility that subsets among Treg cells, e.g., resting naive versus recently activated memory Treg cells (25,28), express distinct patterns of homing receptors and are thereby recruited selectively to tumor tissue. Such patterns of homing receptors might be induced on Treg cells when they encounter specific antigens and may distinguish them from Tcon cells. The majority of Treg cells express the lymph node homing receptor CD62L as well as the skin homing receptor CLA (13,29). Differential expression of additional homing receptors, such as 7 integrin, which promoted Treg cell infiltration into colonic mucosa in a mouse model (30), or specific patterns of expression of their ligands on tumor endothelium, as suggested in our study, may add a previously insufficiently recognized contribution to the observed predominant homing of Treg cells to human tumor tissue.
Interestingly, we detected increased expression of these homing receptors that could interact with several of the tumor-induced endothelial adhesion molecules on tumor-infiltrating Treg cells but not on tumor-infiltrating Tcon cells. Tumor-infiltrating Treg cells differentially overexpressed 7 integrin, CD166, and CD62L. These are all ligands of endothelial adhesion molecules that were overexpressed on tumor endothelium. Adhesive interactions have been suggested to play a role in Treg cell recruitment (10). Although Treg and Tcon cells expressed the P-selectin ligand PSGL1 at similarly high proportions, tumor-infiltrating Treg cells expressed lower levels of CD24, a ligand for P-selectin, than Tcon cells. CD24 ligation reduces T-cell migration along SDF-1 gradients (31). SDF-1 expression is a common feature of human pancreatic carcinomas (31). This may be one reason for the observed lower infiltration rate of Tcon compared with Treg cells, which interact with P-selectin mainly via promigratory PSGL1 (31).
The observed phenotypic differences between tumor-infiltrating Treg cells and Tcon cells may be explained by a selective, gradual loss of the respective homing receptors on Tcon cells during their persistence in tumor tissue or by a Treg cell–selective induction of homing receptors after their entry into tumor tissue, e.g., due to tumor-derived factors. Although our study cannot rule out the latter two possibilities, the data clearly demonstrated an impact of the identified addressins and their respective homing receptors on blood-derived T cells for the transmigration of Treg cells, but not Tcon cells, through tumor endothelial cells in vitro and in vivo. We therefore suggest that the expression pattern of homing receptors on Treg and Tcon cells at least partially influences their capacity to infiltrate pancreatic tumor tissue.
In this study, we compared Treg cells with nonactivated Tcon cells but not to recently activated effector T (Teff) cells that also express CD25. Isolation of tumor-specific CD25+ Teff cells for our experiments was not feasible. We cannot, therefore, exclude the possibility that activated CD25+ Teff cells migrate in response to the same signals as Treg cells. However, based on our analysis of the proportions of FOXP3+ cells among CD25+ T cells in target tissues, we can state that Treg cells accumulate to a much higher degree in tumor tissue than CD4+CD25+FOXP3– Teff cells (18 cells per mm2 tumor tissue versus 5 cells per mm2 tumor tissue). In fact, approximately 75% of CD25+CD4+ T cells in the tumor tissue expressed FOXP3. Thus, Treg cells, but not CD25+ Teff cells, are the major CD4 subset that infiltrates pancreatic tumor tissue.
The majority of peripheral blood CD25+CD4+ T cells in the patients expressed FOXP3 (75%, data not shown). Because the relative proportion of Treg cells to CD4+CD25+ Teff cells in the blood of pancreatic cancer patients was very similar to that of tumor-infiltrating cells (71%, data not shown), it is possible that both subsets possess a similar capacity to infiltrate tumor tissue. Naive Tcon cells infiltrated tumors to a much lower degree than Treg cells. In contrast, the capacity of Teff cell infiltration into corresponding control tissue was markedly increased compared with Treg cells. This difference can be deduced from the fact that approximately 50% of CD4+CD25+ T cells in control tissue were FOXP3– Teff cells, although their proportion in the blood was only 29%. Thus, in contrast to Teff cells, Treg cells were selectively increased in tumor as compared with normal tissue.
This conclusion is substantiated by our observation of an increased capacity of Treg cells to adhere to and transmigrate selectively through tumor-derived endothelial cells, compared with both non-Treg cells and control endothelial cells in vitro and in vivo. This capacity was mediated at least partly by interactions of distinct adhesion molecules because their blocking by antibodies selectively inhibited endothelial transmigration of Treg cells through tumor-derived endothelium.
Our findings demonstrate that Treg cells are selectively recruited to tumor tissues. Such recruitment involves selective interactions between Treg cells and the tumor vasculature mediated by adhesion molecules and respective ligands. Such tumor targeting by Treg cells may also contribute to the overall increased numbers of tumor-infiltrating Treg cells. Using blocking antibodies, we selectively reduced Treg cell transmigration through tumor endothelium. Transmigration of Treg cells through normal endothelial cell layers or transmigration of non-Treg cells through tumor endothelium was only slightly affected. These findings suggest that the development of new therapeutic strategies based on targeted endothelial blockade should be pursued. Anti-CD166 or a mixture of several ligand antibodies in suboptimal concentrations may enable Treg cell and T-cell access to normal endothelium but suppress the entrance of Treg cells to tumor tissue.
|
Funding |
|---|
|
TopAbstractContext and CaveatsPatients and MethodsResultsDiscussionFundingReferencesNotes |
|---|
Deutsche krebshilfe (German Cancer Aid; 102034) to E. Suri-Payer.
|
NOTES |
|---|
|
TopAbstractContext and CaveatsPatients and MethodsResultsDiscussionFundingReferencesNotes |
|---|
We thank Nina Oberle from the Department of Immunogenetics at the German Cancer Research Center for excellent technical advice and Alison Banham, Nuffield Department of Clinical Laboratory Sciences, University of Oxford, U.K., for kindly providing the anti-FOXP3 mAb 150D/E4.
The authors have no conflicting financial interests.
The sponsors had no role in the study design, data collection and analysis, interpretation of the results, or the preparation of the manuscript.
|
REFERENCES |
|---|
|
TopAbstractContext and CaveatsPatients and MethodsResultsDiscussionFundingReferencesNotes |
|---|
(1) Liyanage UK, Moore TT, Joo HG, Tanaka Y, Herrmann V, Doherty G, et al. Prevalence of regulatory T cells is increased in peripheral blood and tumor microenvironment of patients with pancreas or breast adenocarcinoma. J Immunol (2002) 169:2756–61.
(2) Shevach EM. CD4+ CD25+ suppressor T cells: more questions than answers. Nat Rev Immunol (2002) 2:389–400.[ISI][Medline]
(3) Zou W. Immunosuppressive networks in the tumour environment and their therapeutic relevance. Nat Rev Cancer (2005) 5:263–74.[CrossRef][ISI][Medline]
(4) Ormandy LA, Hillemann T, Wedemeyer H, Manns MP, Greten TF, Korangy F. Increased populations of regulatory T cells in peripheral blood of patients with hepatocellular carcinoma. Cancer Res (2005) 65:2457–64.
(5) Somasundaram R, Jacob L, Swoboda R, Caputo L, Song H, Basak S, et al. Inhibition of cytolytic T lymphocyte proliferation by autologous CD4+/CD25+ regulatory T cells in a colorectal carcinoma patient is mediated by transforming growth factor-beta. Cancer Res (2002) 62:5267–72.
(6) Wolf AM, Wolf D, Steurer M, Gastl G, Gunsilius E, Grubeck-Loebenstein B. Increase of regulatory T cells in the peripheral blood of cancer patients. Clin Cancer Res (2003) 9:606–12.
(7) Woo EY, Chu CS, Goletz TJ, Schlienger K, Yeh H, Coukos G, et al. Regulatory CD4(+)CD25(+) T cells in tumors from patients with early-stage non-small cell lung cancer and late-stage ovarian cancer. Cancer Res (2001) 61:4766–72.
(8) Unitt E, Rushbrook SM, Marshall A, Davies S, Gibbs P, Morris LS, et al. Compromised lymphocytes infiltrate hepatocellular carcinoma: the role of T-regulatory cells. Hepatology (2005) 41:722–30.[CrossRef][ISI][Medline]
(9) Yu P, Lee Y, Liu W, Krausz T, Chong A, Schreiber H, et al. Intratumor depletion of CD4+ cells unmasks tumor immunogenicity leading to the rejection of late-stage tumors. J Exp Med (2005) 201:779–91.
(10) Curiel TJ, Coukos G, Zou L, Alvarez X, Cheng P, Mottram P, et al. Specific recruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and predicts reduced survival. Nat Med (2004) 10:942–9.[CrossRef][ISI][Medline]
(11) Suri-Payer E, Fritzsching B. Regulatory T cells in experimental autoimmune disease. In: Springer Semin Immunopathol (2006) 28:3–16.[CrossRef][ISI][Medline]
(12) Iellem A, Mariani M, Lang R, Recalde H, Panina-Bordignon P, Sinigaglia F, et al. Unique chemotactic response profile and specific expression of chemokine receptors CCR4 and CCR8 by CD4(+)CD25(+) regulatory T cells. J Exp Med (2001) 194:847–53.
(13) Chen D, Bromberg JS. T regulatory cells and migration. Am J Transplant (2006) 6:1518–23.[CrossRef][ISI][Medline]
(14) Yurchenko E, Tritt M, Hay V, Shevach EM, Belkaid Y, Piccirillo CA. CCR5-dependent homing of naturally occurring CD4+ regulatory T cells to sites of Leishmania major infection favors pathogen persistence. J Exp Med (2006) 203:2451–60.
(15) Soler D, Chapman TR, Poisson LR, Wang L, Cote-Sierra J, Ryan M, et al. CCR8 expression identifies CD4 memory T cells enriched for FOXP3+ regulatory and Th2 effector lymphocytes. J Immunol (2006) 177:6940–51.
(16) Belkaid Y, Rouse BT. Natural regulatory T cells in infectious disease. Nat Immunol (2005) 6:353–60.[CrossRef][ISI][Medline]
(17) Huehn J, Siegmund K, Lehmann JC, Siewert C, Haubold U, Feuerer M, et al. Developmental stage, phenotype, and migration distinguish naive- and effector/memory-like CD4+ regulatory T cells. J Exp Med (2004) 199:303–13.
(18) Ermann J, Hoffmann P, Edinger M, Dutt S, Blankenberg FG, Higgins JP, et al. Only the CD62L+ subpopulation of CD4+CD25+ regulatory T cells protects from lethal acute GVHD. Blood (2005) 105:2220–6.
(19) Belkaid Y, Piccirillo CA, Mendez S, Shevach EM, Sacks DL. CD4+CD25+ regulatory T cells control Leishmania major persistence and immunity. Nature (2002) 420:502