Tumor-infiltrating regulatory dendritic cells inhibit CD8⁺ T cell function via L-arginine metabolism

Abstract

Dendritic cells (DC) have a critical impact on the outcome of adaptive immune responses against growing tumors. While it is generally assumed that the presence of phenotypically mature DC should promote protective anti-tumor immunity, evidence to the contrary does exist. We describe here a novel mechanism by which tumor-infiltrating dendritic cells (TIDC) actively contribute to the suppression of protective CD8⁺ T cell based antitumor immunity. Using the BALB/NeuT model of spontaneously arising mammary carcinoma, we found that canonical MHC II⁺/ CD11b⁺/ CD11c^high TIDC act as regulatory DC to suppress CD8⁺ T cell function, resulting in diminished T cell-based antitumor immunity in vivo. Stimulation of naive T cells with regulatory TIDC resulted in an altered cell fate program characterized by minimal T cell expansion, impaired IFNγproduction, and anergy. Suppression by regulatory TIDC overcame stimulatory signals provided by standard DCs, occurred in the absence of cognate interactions with T cells, and was mediated primarily by arginase metabolism of L-arginine. Immunosuppressive TIDC were found in every murine tumor type examined, and were phenotypically distinct from tumor-infiltrating CD11c^int-low/CD11b⁺/ Gr 1⁺ myeloid derived suppressor cells (MDSC). Thus, within the tumor microenvironment, MHC II⁺ TIDC can function as potent suppressors of CD8⁺ T cell immunity.

Introduction

The development of protective T cell based antitumor immunity is critically dependent upon the normal function of DCs. This has led to a great interest in using DC-based therapies for the treatment of progressively growing tumors. However, clinical results to date have yielded minimal success (1), and the reasons for this remain unclear.

Many tumors, including human breast cancers, contain appreciable numbers of tumor-infiltrating DCs (TIDC) (2–4) but the stimulatory capacity of these DCs is often compromised [reviewed in (5–7)]. For example, tumor-derived cytokines such as VEGF can block DC maturation, leading to failed T cell priming in tumor-draining lymph nodes (6, 8, 9). However, blocked DC maturation is not the only obstacle to be overcome. In mice, DC maturation is characterized phenotypically by upregulation of surface markers such as MHC II, CD80 and CD86. Although it was once thought that phenotypically mature DCs exclusively promoted T cell activation, this paradigm is now known to be incorrect (10). For example, disruption of E-cadherin adhesion induces DCs to upregulate MHC II and costimulatory molecules, yet simultaneously programs them to acquire a tolerogenic function (11). In breast carcinomas, MHC II⁺ DCs can aid tumor progression by generating IL-13 producing CD4⁺ T cells (12). MHC II⁺ DCs in ovarian cancer patients have also been shown to express inhibitory molecules such as B7-H1 that can down modulate T cell effector responses via ligation of PD-1 (13). Clearly, a more thorough understanding of the ways in which DCs participate in immunosuppression is essential if T cell and DC-based tumor immunotherapies are to achieve greater clinical efficacy.

In non-tumor model systems, DCs have been generated in vitro that function as regulatory cells to suppress CD4⁺ T cell function even in the presence of stimulatory DCs (14–18). This ability to dominate over signals provided by standard DCs differentiates such regulatory DCs (regDCs) from simply immature or tolerogenic DCs. We hypothesized that regDCs would develop naturally during the outgrowth of tumors, and utilized the BALB/NeuT model of spontaneous murine mammary carcinoma to explore this possibility.

We found that NeuT tumors are infiltrated by fully committed, CD11b⁺ CD11c^high DCs that express MHC II, CD80 and CD86, yet which impair CD8⁺ T cell antitumor immunity in vivo and T cell function in vitro. Thus, tumors can induce matured, canonical TIDC to act as potent suppressors of T cell-based antitumor immunity. Our work adds to the growing body of evidence that indicates phenotypically “mature” MHC II⁺ TIDC may function primarily to impair protective T cell immunity, rather than promote it.

Materials and Methods

Mice

BALB/c NeuT mice, kindly provided by Jay Berzofsky (National Cancer Institute, Bethesda MD), and DUC18 TCR transgenic mice have been described previously (19–22). BALB/c mice were purchased from the NCI. All animal studies were performed in accordance with institutional guidelines at the Washington University School of Medicine.

Immunofluorescence

Tumors were suspended in OCT, frozen in 2-methylbutane cooled with liquid nitrogen, sectioned, fixed with acetone, and stored at −20°C until use. Prior to staining, sections were blocked with 5% normal goat serum, Fc receptor blocking 2.4G2 antibody (BD Biosciences), and 0.1% Tween 20. Staining was done with anti-CD11c (clone HL3, BD Biosciences), followed by anti-rat Cy3, and anti- I-A^d-FITC (BD Biosciences). Immunofluorescence was visualized under a fluorescent microscope and pictures taken with a Nikon DXM1200 digital camera. Final image processing was performed using Photoshop 7.0 software (Adobe).

DC isolation

DCs were harvested from NeuT mammary carcinomas, CMS5 fibrosarcomas, or 4T1 mamary carincomas, when tumors were >10 mm in diameter. Control spDCs were isolated from spleens of tumor-free BALB/c mice. Organs were dissociated as described (20). For enrichment of spDCs, cells were incubated with Miltenyi CD11c Microbeads and purified over two sequential Macs LS columns according to the manufacturer’s protocol (Miltenyi Biotec) to achieve >93% CD11c⁺ DCs. Unless otherwise noted, TIDC were sort-purified based on their co-expression of CD11c and CD11b, and exclusion of Gr-1 using a MoFlo high speed flow cytometer (Dako Cytomation). Fc receptors were blocked with 2.4G2 (BD Biosciences) and normal mouse serum prior to surface staining.

T cell proliferation assays

T cells were harvested from DUC18 transgenic mice, and purified over two sequential Macs LS columns using CD8 Microbeads. The percentage of DUC18 T cells was determined via flow cytometry based on co-expression of CD8 and Vβ8.3. Naive DUC18 T cells were cultured as described (20) at 5 × 10⁴ cells/well in flat bottom 96 well plates with indicated numbers of spDCs, TIDC, or both. Prior to plating, DCs were pulsed with 0.5 µM tERK peptide (22). On day three, wells were pulsed with ³H-thymidine for 18 hours, harvested, and counted on a beta counter. The following were used where indicated: 10mM L-Norvaline (Sigma), 100µM L-nil (Sigma), 1mM L-NNA (Sigma), 2mM L-arginine (Sigma). In some experiments, T cells were labeled with CFSE (20) and plated at 2 × 10⁶ DUC18 T cells/well with 2 × 10⁵ DCs in 6 well plates for 96 hr prior to harvest and analysis.

L-arginine assays

L-arg incorporation was performed by culturing 5 × 10⁵ DC/ml in a 200 µl volume in 96 well flat bottom plates in complete medium made with RPMI 1640 containing 100 µM L-arginine. 5µCi/well of ³H-L-arginine was added at time 0 and plates were harvested at the indicated times. Arginase activity in cell lysates was determined by measuring the production of L-ornithine as described (23) after culturing 2 × 10⁶ Miltenyi macs purified TIDC or spDC overnight with or without the addition of 100µg/ml rmIL-4 (Peprotech). NO production was evaluated by measuring the total concentration of nitrate/nitrite in culture supernatant after culturing cells overnight in the absence or presence of 250ng/ml IFNγ (kindly provided by Dr. Robert Schreiber, Washington University School of Medicine), using the Nitrate/Nitrite Colorimetric Assay Kit (Cayman Chemical).

Cell staining for flow cytometry

T cells were routinely stained with anti-CD8 APC (BioLegend), 7 amino-actinomycin D (Sigma), and Vβ8.3 PE or FITC (BD Biosciences) to allow gating on live, DUC18 T cells. CD3ζ staining was performed after addition of 500 µg/ml digitonin and anti-TCRzeta (Abcam). Intracellular cytokine staining was performed as described (24) using anti-IFNγ APC (BioLegend), anti-CD8 PE/Cy5 (BioLegend), and anti- Vβ8.3 FITC.

DCs were routinely stained with anti-CD11c biotin (clone HL3, BD Biosciences) plus streptavidin APC (Caltag Laboratories), 7AAD, anti-CD11b PE (clone M1/70, BioLegend), and one of the following: anti- Gr-1 (RB6,8C5; BD Biosciences), H-2K^d FITC (SF1-1.1, BD Biosciences), I-A^d FITC (39-10-8, BD Biosciences), CD40 FITC (3/23, BD Biosciences), CD80 FITC (16-10A1, BioLegend), CD86 FITC (GL-1, BioLegend), CD54 FITC (3E2, BD Biosciences).

Tumor studies

The CMS5 fibrosarcoma cell line (25) was cultured as described (20) and injected s.c at 3 × 10⁶ cells/ mouse in the right hind flank, with or without the addition of 1 × 10⁶ Macs-purified TIDC. For mice receiving T cells, 1 × 10⁶ naive DUC18 T cells were injected i.v. 4 days after tumor cell challenge. Recipient mice were either BALB/c controls, or NeuT mice at approximately 14–15 weeks of age, when spontaneous mammary tumors were just becoming palpable (26). The 4T1 mammary carcinoma cell line was cultured as described (27) and injected s.c. into mammary gland #4 at 7.5 × 10⁵ cells/mouse. For all tumors, areas (product of orthogonal diameters) were measured every other day for 40 days, or until tumors were ≥ 20mm in diameter. Progressively growing tumors were defined as those tumors that displayed a continuous increase in cell size during the course of the observation period.

Statistical analyses

Statistics were performed using a two-tailed Student’s t-test for unpaired observations, with correction for unequal variations as needed, via Graph Pad software.

Results

NeuT mammary tumors are heavily infiltrated by phenotypically mature DCs

To determine whether TIDC could actively suppress CD8⁺ T cell function, we utilized BALB/c NeuT mice (hereafter referred to as “NeuT”). Progressively growing NeuT mammary tumors contained numerous CD11c⁺ cells that infiltrated tumor beds beyond the CD31⁺ vasculature (Fig. 1A), and showed positive co-expression of I-A^d along with characteristic dendritic cell morphology.

Figure 1.

Canonical CD11c⁺/ MHC II⁺ DCs infiltrate NeuT tumors. A, NeuT tumors were stained for CD11c-Cy3 (red) and CD31 FITC or I-Ad FITC (green). Cells expressing both appear in yellow. One tumor, representative of 5 is shown. B, Spleens from tumor-free BALB/c mice and mammary tumors from NeuT mice were harvested and stained. The percentage of live, gated CD11c^high cells is shown, along with maturation marker expression on these cells. LPS-treated BALB/c mice received 70µg of Salmonella abortus LPS one day prior to harvest and staining of splenocytes. C, Gates for CD11c^high TIDC and CD11c^int-low MDSC are shown, along with surface expression of Gr 1. Specific antibodies = blue lines, red lines = isotype controls. The mean fold increase of Gr1 MFI over isotype MFI is shown at right for 6 individual mice from 3 experiments.

Flow cytometric analysis of NeuT tumors revealed a striking accumulation of CD11c^high/ CD11b⁺ cells with advancing tumor progression, such that they accounted for greater than 10% of the live cells present by the time NeuT mice had reached 20 weeks of age (Fig. 1B and Supplemental Fig. 1A). CD11c^high cells from NeuT tumors appeared to be conventional myeloid DCs, with low to negative co-expression of CD4, CD8, B220, CD45RB and the lineage markers CD3, CD19 and DX5 (Supplemental Fig. 1B and data not shown). Tumor-infiltrating CD11c^high cells expressed higher levels of I-A^d, CD80 and CD86 than did CD11c^high/ CD11b⁺ splenic DCs (spDCs) from tumor-free mice (Fig. 1B), yet displayed a phenotype that was distinct from that of spDC that were matured in vivo in response to LPS administration (Fig. 1B). Ex vivo stimulation of CD11c^high/CD11b⁺ cells with CpG 1826 led to increased expression of CD40 and CD86, suggesting that these cells could be matured further, and were not a terminally differentiated, “end point” population (Supplemental Fig. 1C). Collectively, these data illustrate that tumor-infiltrating CD11c^high/CD11b⁺ cells are canonical, committed DC, and we refer to them henceforth as tumor-infiltrating dendritic cells (TIDC).

To determine if NeuT TIDC were phenotypically distinct from previously described CD11b⁺/ Gr1⁺ myeloid derived suppressor cells (MDSC), we analyzed tumor- infiltrating cells from multiple mice for CD11c, CD11b and Gr 1 expression. MDSC are heterogeneous populations of suppressor cells defined phenotypically in mice by their co-expression of CD11b and Gr-1, and by their immature state in the presence of tumor-derived factors (6, 28–31). CD11c^high NeuT TIDC do not co-express Gr-1 (Fig. 1C), differentiating them from CD11b⁺/ Gr-1⁺ MDSCs that exist in the peripheral blood of NeuT mice (32) and which also infiltrated NeuT mammary tumors (Fig. 1C). NeuT TIDC expressed MHC II and CD86 (Fig. 1B), while tumor-infiltrating NeuT MDSCs did not (data not shown). A minor, third population of CD11c^high/ CD11b^high cells was identified that had bimodal expression of Gr 1 (Fig. 1C); as these cells could represent immature DC, they were excluded from further analysis.

NeuT TIDC promote tumor outgrowth in vivo

We next wanted to determine whether NeuT TIDC supported or inhibited CD8⁺ T cell-mediated tumor immunity. We began by examining TIDC function in an adoptive transfer model in vivo, and utilized the CMS5 fibrosarcoma model, in which small established tumors are rejected following transfer of naive, tumor antigen-specific CD8⁺ TCR transgenic DUC18 T cells (20, 22, 26). The kinetics of DUC18 T cell-mediated CMS5 rejection are well-characterized and highly reproducible, making this an ideal system with which to examine the in vivo effects of TIDC function.

CMS5 fibrosarcoma cells were either injected into recipient BALB/c mice alone, or as a 3:1 mixture with purified CD11c⁺ TIDC from NeuT tumors. Purified TIDC preparations contained less than 2% contaminating CD11b⁺/ Gr-1⁺ MDSC (Supplemental Fig. 2A). In the absence of tumor antigen-specific DUC18 T cells, both groups of subcutaneous CMS5 tumors grew progressively and with similar kinetics through day 16 post-challenge, at which time animals were sacrificed due to the presence of unacceptably large tumors in some mice (Fig. 2A).

Figure 2.

TIDC from NeuT mammary tumors promote tumor outgrowth in vivo. A, CMS5 fibrosarcoma cells were were injected s.c. alone, or together with NeuT TIDC on day 0 into BALB/c mice. Tumor outgrowth through day 16 is shown as means +/− SEM; n = 11 mice for CMS5 alone and n = 10 for CMS5+ TIDC. Open circles = CMS5 alone, closed diamonds = CMS5+TIDC. Data are cumulative from 3 independent experiments. Center panel: Individual tumor regression data for 4 mice receiving CMS5+ TIDC on day −4 with naive DUC18 T cells transferred i.v. on day 0. Right panel: Individual tumor regression data for 4 mice receiving CMS5+ TIDC on day −4, with naive DUC18 T cells transferred i.v. on day 0. B, Cumulative scatter plots of tumor sizes at day 40, from all individual mice used in multiple experiments. C, Progressive tumor outgrowth in NeuT recipient versus BALB/c recipient mice after DUC18 T cell transfer on day 0. For BALB CMS5 vs BALB CMS5+TIDC, p = .0202; for BALB CMS5 vs. NeuT CMS5, p = .1208; for BALB CMS5 vs. NeuT CMS5+TIDC, p= .0001; for NeuT CMS5 vs NeuT CMS5+TIDC, p= .0014 (all, two-tailed t test with Welch’s correction). For NeuT recipients only, the relative % of gated tumor-infiltrating DUC18 T cells making IFNγ at day 6 post-transfer is shown as means from 4 –5 individual mice.

The effects of TIDC on CD8⁺ T cells were examined by injecting CMS5 cells alone or with enriched TIDC as above, then adoptively transferring 1 × 10⁶ naive DUC18 T cells 4 days later. In mice that received CMS5 cells plus DUC18 T cells, a short period of tumor outgrowth occurred prior to tumor rejection (Fig, 2A). Only 7% of these mice experienced any tumor re-growth through day 40 (1/15 mice in 4 experiments) (Fig. 2B). A reversal of this trend was seen when TIDC were admixed with CMS5 cells, such that 100% of mice (15/15 mice in 3 experiments) ultimately experienced progressive tumor outgrowth (Fig. 2A and B ). Inhibition of antitumor immunity was specific to TIDC, as sort purified CD11c^high/ CD11b⁺ spDCs from tumor-free BALB/c mice failed to similarly impact late tumor outgrowth (Fig. 2B). Thus, it appeared that TIDC promoted eventual tumor outgrowth, but did not noticeably affect initial T cell priming and tumor regression.

We were surprised that co-injection of TIDC with CMS5 cells affected tumor outgrowth in such a delayed manner. To determine if TIDC could suppress initial T cell-mediated tumor regression if recipient mice already bore established tumors, we repeated the above experiments, using as recipients either tumor-free BALB/c mice (as above) or NeuT mice at approximately 14–15 weeks of age, when mammary tumors were just becoming palpable. At this age, relatively few myeloid DC are present endogenously within developing NeuT tumors (Suppl. Fig. 1A). Previous work in our laboratory had demonstrated that distal NeuT mammary tumors do not affect the ability of DUC18 T cells to reject small, transplanted CMS5 flank tumors (26). However, when CMS5 + TIDC were injected into NeuT mice and T cells were transferred 4 days later, nearly 40% of recipient mice showed uncontrolled CMS5 tumor outgrowth from day 2 onward, and this increased steadily to 85% by day 20 (Fig. 2C). Therefore, progressive CMS5 outgrowth occurred in NeuT mice despite the presence of tumor-antigen specific DUC18 T cells. This suggested that when transferred into mice already bearing distal mammary tumors, TIDC exerted a stronger suppressive effect in vivo.

To further examine the functional effects of TIDC on DUC18 T cells, tumor-infiltrating DUC18 T cells from NeuT mice were examined for their ability to produce IFNγupon restimulation ex vivo with antigen-pulsed spDCs. As shown in Figure 2C, the percentage of IFNγ⁺ DUC18 T cells from CMS5+ TIDC tumors was decreased 43%, relative to DUC18 T cells from CMS5 control tumors. Collectively, these results suggested that NeuT TIDC act primarily to inhibit CD8⁺ T cell-mediated anti-tumor immunity, and promote tumor outgrowth. However, because these experiments were unable to distinguish between TIDC suppression of T cell mediated antitumor immunity, and TIDC promotion of tumor outgrowth via enhanced angiogenesis or other non-immune-mediated mechanisms, we next turned to examining the direct effects of TIDC on CD8⁺ T cells in vitro.

NeuT TIDC anergize naive CD8⁺ T cells

Previous reports on in vitro generated regDCs described them as being poor stimulators of naive CD4⁺ T cells, inducing minimal T cell proliferation and IFNγ production (14, 15). To determine whether NeuT TIDC functioned similarly, TIDC were sort-purified, pulsed with tERK peptide, and used to stimulate naive CD8⁺ DUC18 T cells. This resulted in diminished T cell proliferation compared to controls (Fig. 3A). Decreased T cell proliferation in the presence of TIDC was not due to increased apoptosis (Fig. 3B) but appeared to result from fewer T cells progressing normally through the cell cycle (Supplemental Fig. 2B). Similar to what had been reported by Zhang et al. (14), TIDC-stimulated DUC18 T cells showed impaired upregulation of CD25 (Fig. 3C), suggesting an impaired ability of T cells to respond to IL-2. Diminished T cell proliferation could be due to a decrease in the quantity or quality of antigen being presented by TIDC, or a shortened duration of antigen presentation. Gating on CD11c⁺ cells from the T cell + DC co-cultures shown in Figure 3B revealed that a similar percentage of viable, Annexin V⁻ spDC and TIDC were present after 96 hours of culture (12.5% viable spDC versus 18.9% TIDC). Additionally, expression of CD69 on T cells was equivalent at 96 hours, indicating that the duration of antigen presentation and the quantity of antigen being presented were comparable between spDC and TIDC (Fig. 3C). In contrast, IFNγ production was dramatically impaired in TIDC-stimulated T cells (Fig. 3C). Thus, it appears that TIDC-stimulated T cells were receiving a qualitatively different priming signal than were spDC-stimulated T cells.

Figure 3.

TIDC stimulation of DUC18 T cells leads to impaired proliferation and anergy. A, Proliferation of naive DUC18 T cells cultured with tERK peptide-pulsed spDC or NeuT TIDC for 96 hours, at the indicated T cell: DC ratios. Data from 1 experiment, representative of more than 15 are shown. * = p < .001, versus spDC at each ratio. B, Annexin V staining on gated Thy1.1⁺ T cells after 96 hours of culture. C, Naive DUC18 T cells were CFSE labeled, then cultured at a 10:1 ratio with spDC or NeuT TIDC. Gated on live, CD8⁺ DUC18 T cells at 96 hours, with percentages of divided cells in specific quadrants indicated. Data represent more than 5 experiments. D, Following primary stimulation with TIDC, DUC18 T cells become unresponsive to secondary stimulation with spDCs as evidenced by a lack of IFNγ production at 48 hours.

The combined lack of sustained T cell proliferation and INFγ production, coupled with intact CD69 upregulation, was reminiscent of cells becoming programmed for anergy. To investigate this possibility, enriched naive CD8⁺ T cells from DUC18 mice were cultured with either tERK-pulsed spDCs or TIDC for four days, then purified and rested. Then both sets of T cells were re-stimulated with tERK-pulsed spDCs. T cells that had undergone primary stimulation with TIDC produced virtually no IFNγ upon secondary stimulation (Fig. 3D). Therefore, primary encounter of naive CD8⁺ T cells with TIDC initiates a cell fate program that leads to aborted proliferation, minimal cytokine production, and anergy.

NeuT TIDC act as regulatory cells to suppress T cell proliferation in the presence of stimulatory spDCs

A defining hallmark of in vitro generated regDCs is their ability to act in a dominant fashion to suppress naive CD4⁺ T cell proliferation in the presence of standard, stimulatory DCs. We therefore asked whether NeuT TIDC could actively suppress CD8⁺ T cell proliferation, as evidence of their regulatory nature. To this end, NeuT TIDC were sort-purified according to the gate shown in Figure 4A, and pulsed with tERK peptide. As before, T cell proliferation was robust in the presence of tERK-pulsed spDCs. When this 10 :1 T cell to spDC ratio was maintained, and increasing numbers of TIDC were titrated into the culture, T cell proliferation steadily dropped (Fig. 4A). Overall, the mean level of TIDC suppression was 43% when equal numbers of spDC and TIDC were present in co-cultures with T cells (Fig. 4B, n = 15 experiments). To determine if the decrease in T cell proliferation were due to active suppression by TIDC or passive interference with access to stimulatory DCs, we combined equal numbers of spDCs, one set pulsed with tERK and the other with an irrelevant but H2-K^d-binding Influenza peptide (22), rather than adding TIDC to spDCs. Minimal inhibition of T cell proliferation ensued (Fig. 4B). Therefore, NeuT TIDC were acting as regulatory TIDC to actively, not passively, impede T cell expansion.

Figure 4.

NeuT TIDC act as regDCs to dominantly suppress naive CD8⁺ T cell proliferation. A, The gate used to sort-purify NeuT TIDC is indicated (small upper left panel). Naive DUC18 T cells were incubated alone, or in the presence of tERK-pulsed NeuT TIDC, BALB/c spDCs, or both, at the indicated T cell: DC ratios. One assay, representative of 17, is shown. B, T cells, spDCs, and NeuT TIDC, were cultured at a 10:1:1 ratio. The mean percent inhibition of T cell proliferation, as compared to T cells plus spDCs alone, is shown +/− the SEM. For TIDC pulsed with tERK (black bars) n= 17 experiments, for spDCs pulsed with K^d FLU peptide (white bars) n= 5 experiments, for TIDC pulsed with K^d FLU peptide (grey bars) n= 8 experiments. C, T cells were cultured as in panel B, with either NeuT TIDC, or CMS5 TIDC, or 4T1 TIDC. The mean % inhibition of T cell proliferation is shown for results from 4–15 independent experiments.

We then asked if cognate interactions between DUC18 T cells and TIDC were required for suppression to occur, by using TIDC that were pulsed with the irrelevant but H2-K^d -binding Influenza peptide. As with tERK pulsed TIDC, the mean level of suppression exceeded 40% (Fig. 4B, n= 8 experiments). Therefore, inhibitory signals from TIDC dominate over positive signals provided by standard DCs, and TIDC presentation of cognate antigen is not required for suppression to occur.

We then asked whether suppressive TIDC were present in other murine tumors. Whereas small transplanted CMS5 fibrosarcomas can be rejected by DUC18 T cells (Fig. 3B), large CMS5 tumors can not (22, 33). Therefore, CD11c^high/ CD11b⁺ TIDC were sort-purified from large, non-rejectable CMS5 tumors and used in suppression assays, as were CD11c^high/ CD11b⁺ TIDC from primary 4T1 mammary carcinomas. While the mean level of suppression varied among the three tumor models, CD11c^high/ CD11b⁺ TIDC from all tumors were able to suppress T cell proliferation in the presence of stimulatory spDC from tumor-free mice (Fig. 4C). Therefore, regulatory TIDC are not unique to NeuT tumors.

TIDC regulatory function depends upon arginase metabolism of L-arginine

To determine the mechanism by which NeuT TIDC suppressed T cell proliferation, we began by evaluating well-known soluble inhibitors such as VEGF, indoleamine 2–3 dioxygenase (IDO) (34), and IL-10. Blockade of these factors with antibodies or pharmacologic agents only minimally impacted T cell proliferation (data not shown). We then examined the involvement of L-arginine metabolism in TIDC-mediated suppression. During a 24 hr culture, NeuT TIDC incorporated 5 times more L-arginine than did spDCs (Fig. 5A). After uptake into cells, L-arginine is metabolized through two main pathways by iNOS and arginase (29). Culture of TIDC with IL-4 led to increased Arginase I protein expression (Supplemental Fig. 2C) and Arginase activity (Fig. 5B). In contrast, NO production was not elevated in NeuT TIDC versus spDCs (Fig. 5B), even after an overnight stimulation with IFNγ that led to increased iNOS protein expression (Supplemental Fig. 2D). As further evidence for the role of L-arginine metabolism in TIDC-mediated suppression, we tested T cells for evidence of phosphorylated eIF2a, the initial sensor of cellular stress caused by amino acid starvation (35, 36), and found it to be increased (Fig. 5C). Additionally, culture of T cells with regulatory TIDC led to decreased CD3ζ chain expression, as had been reported in other models of L-arginine deprivation (37). However, T cells cultured with TIDC showed no evidence of tyrosine nitrosylation (data not shown) and DCFDA staining showed negligible production of reactive oxygen species by NeuT TIDC in response to co-culture with DUC18 effector T cells (Supplemental Fig. 2E). Collectively, these data suggested a role for arginase, but not iNOS, metabolism of L-arginine during TIDC suppression of T cell proliferation.

Figure 5.

NeuT TIDC use arginase to metabolize L-arginine. A, ³H-L-arginine incorporation of purified TIDC and spDCs was measured at 0, 0.5, 3, 18 and 24 hrs. The means +/− s.d. for triplicate wells are shown; error bars are too small to be seen. B, Arginase activity measured in lysates from spDCs, pooled from 12 mice, or TIDC, taken from 3 individual NeuT mice. Cells were cultured in medium alone or with IL-4 prior to harvest and lysis. TIDC versus spDC controls * indicates p < .001. NO production as measured by the concentration of total nitrite/ nitrate after culturing spDCs or TIDC for 24 hours, with or without IFNγ. Cumulative data from four experiments are shown. C, Increased expression of the phosphorylated form of eIF2a when T cells are cultured with TIDC. D, Downregulation of surface CD3ζ chain expression on DUC18 T cells stimulated with TIDC.

We then directly addressed whether L-arginine metabolism by arginase was required for NeuT TIDC inhibition of CD8⁺ T cell proliferation. Pharmacologic inhibitors were selected for their specific activity on either iNOS or arginase pathways, and titrations were performed to determine the optimal doses for use in our model (data not shown). The addition of 2mM L-arginine, to prevent its depletion from culture medium, resulted in an average 40% restoration in T cell proliferation (n= 10 experiments) (Fig. 6A). Use of either L-nil or L-NNA, both iNOS inhibitors, did not significantly increase T cell proliferation, supporting our functional data that showed no increase in NO production by TIDC relative to spDCs. Norvaline is a commonly used arginase inhibitor (38) and addition of this compound restored T cell proliferation to nearly 80% of that seen for T cells cultured with spDCs alone (mean = 79%, n = 6 experiments). Taken together, these data indicate that TIDC inhibit T cell proliferation primarily via arginase-mediated metabolism of L-arginine.

Figure 6.

NeuT TIDC use arginase metabolism of L-arginine to suppress CD8 T cell proliferation. A, The mean percent restoration in T cell proliferation, relative to T cells plus spDC controls, is shown for T cells cultured with spDC plus TIDC +/− L-arginine, L-nil, L-NNA, or norvaline. Data are expressed as the mean +/− SEM for 9–17 independent experiments. B, Propidium idodide cell cycle analysis on naive DUC18 T cells cultured as indicated for 96 hours. The percentages of both dying and replicating cells are given for each condition.

Studies in other models have shown that limited L-arginine availability can lead to T cell hyporesponsiveness and cell cycle arrest (39). DUC18 T cells cultured with spDCs in the absence of L-arginine showed deficient cell cycle progression, in comparison to controls cultured with L-arginine (Fig. 6B). A nearly identical cell cycle profile was observed when DUC18 T cells were cultured with TIDC and spDCs in the presence of L-arginine. The addition of norvaline completely reversed this trend and restored normal cell cycle progression. Therefore, arginase-expressing TIDC from NeuT mice caused T cells to undergo cell cycle arrest equivalent to that observed when T cells were cultured in the absence of L-arginine.

Discussion

Tumor derived immunosuppression is a major impediment to successful clinical applications of immunotherapy. Recently, intense research efforts have focused on identifying populations of immunosuppressive tumor-associated cells, and the intracellular pathways that direct their functions, with the goal of ultimately improving the clinical efficacy of immune-based antitumor therapies (5, 7, 40, 41).

Our results provide clear evidence that tumors can induce canonical, MHC II⁺ TIDC to acquire regulatory functions and support tumor outgrowth by suppressing CD8⁺ T cell function. Previous studies had shown that MHC II⁺ DCs contributed to tumor outgrowth by subverting CD4⁺ T cell differentiation towards IL-13 production (12) or by down-regulating effector T cell function through B7-H1 ligation of the inhibitory receptor PD-1 (13). We now show that MHC II⁺ TIDC can actively suppress CD8⁺ T cell priming and expansion via metabolism of L-arginine, and can counteract CD8⁺ T cell-mediated tumor rejection in vivo. Collectively, these data suggest that within the tumor microenvironment, MHC II⁺ DCs act primarily to inhibit protective anti-tumor immunity. Therefore, therapeutic strategies aimed at increasing the numbers of mature DCs within tumors may fail due to TIDC acquisition of regulatory functions.

Currently, much research into tumor-associated suppressor cell populations focuses on MDSCs. MDSCs are a heterogeneous population of immature myeloid cells that share a “lack or reduced expression of markers of mature myeloid cells, expression of both Gr-1 and CD11b molecules in mice, inability to differentiate into mature myeloid cells in the presence of tumor-derived factors, high levels of reactive oxygen species, and activation of arginase I and other molecules.” (30). According to this definition, and the results presented here, CD11c^high/ CD11b⁺/Gr-1⁻ regulatory TIDC constitute a distinct population of myeloid-lineage suppressor cells within solid tumor masses. Thus, it is possible that regulatory TIDC complement the suppressive actions of MDSC and tumor-associated macrophages in vivo, and that the relative contribution of each population is dictated by both local cell numbers and suppressive capacity on a per cell basis.

During our investigation into the functional capacity of regulatory TIDC, we found that the magnitude and timing of TIDC suppression were critically linked to the host environment. In tumor-free BALB/c mice, co-injection of CMS5 fibrosarcoma cells with regTIDC had no apparent effect on tumor antigen-specific T cell priming, and initial tumor regression occurred with normal kinetics (Fig. 3A). During the equilibrium phase of the anti-tumor immune response (42), co-injection of regTIDC 30–40 days earlier shifted the balance so that tumor outgrowth ensued. Inhibition of T cell effector/memory responses occurred in 100% of BALB/c mice receiving CMS5 cells plus regTIDC (Fig. 3B). One explanation for this is that, similar to what we observed in vitro, the quality of initial T cell priming was altered so that downstream effector and memory cells were defective in their abilities to respond to secondary or tertiary antigenic encounters. In contrast, when regTIDC were co-injected into a recipient in which spontaneous mammary tumors were already developing, suppression of anti-tumor immunity was apparent more rapidly. Thus, it appears that systemic changes caused by developing mammary carcinomas created an environment that was more conducive to regTIDC suppression, and in this situation, regTIDC had a pronounced effect on T cell priming. It is important to consider, however, that regTIDC could also be promoting tumor outgrowth independent of any effects on the antitumor immune response. For example, DC in other systems have been shown to promote tumor angiogenesis (43, 44) and it is possible that regTIDC are exerting similar effects in our model system.

L-arginine is a conditionally required amino acid, and as with indoleamine 2,3-dioxygenase (IDO) metabolism of tryptophan, elevated metabolism of L-arginine by suppressor cells can lead to diminished T cell function in vitro and in vivo (23, 29, 34, 45, 46). Limited L-arginine availability results in blocked cell cycle progression of in vitro stimulated T cells (39). We found that co-culture of CD8⁺ T cells with regTIDC also resulted in impaired cell cycle progression, even in the presence of stimulatory spDCs (Supplemental Fig. 2 and Fig. 6). However, it does not appear that decreased L-arginine availability is the sole reason for diminished T cell expansion in our model system. Providing an excess of L-arginine restored T cell proliferation by only 40%, whereas inclusion of the arginase inhibitor norvaline restored T cell proliferation to 79% of that seen in positive controls. From this, we conclude that a separate mechanism, downstream of arginase activity, is equally responsible for the induction of anergy and cell cycle arrest in DUC18 T cells cultured with regTIDC. The nature of this arginase-dependent mechanism remains to be determined.

Our study demonstrates a role for MHC II⁺ arginase-expressing TIDC in the suppression of anti-tumor immunity. We predict that a re-evaluation of “mature” MHC II⁺ TIDC in other systems will reveal the presence of DCs functioning in a regulatory capacity. Although we identified regulatory TIDC in every murine tumor model examined, the extent to which matured TIDC contribute to the impairment of anti-tumor immunity in cancer patients, remains to be seen. If present, this would suggest that strategies aimed at specifically eliminating TIDC with regulatory functions could enhance the efficacy of T cell-based immunotherapies.