Transforming Growth Factor β Inhibits the Antigen-Presenting Functions and Antitumor Activity of Dendritic Cell Vaccines

INTRODUCTION

In recent years, DCs 4 have become popular candidates in cancer vaccine development because of their crucial role in inducing T-cell responses. Upon antigen uptake, DCs residing in peripheral tissues internalize and process antigen and migrate to secondary lymphoid organs where they stimulate naïve T lymphocytes in the context of class I and class II MHC antigens (1) . The effectiveness of DCs as antigen-presenting cells provides the rationale for their use as cancer vaccines with the objective of inducing durable antitumor immune responses. In numerous rodent models, vaccines consisting of tumor antigen-pulsed DCs are effective in inducing CTL responses and providing protection against subsequent tumor challenge (2, 3, 4, 5) . In contrast, DC vaccines have been less effective in abrogating established tumors in mice (2, 3, 4, 5) and human cancer patients (6) . The insensitivity of established tumors to DC therapy is likely because of factors present within the tumor microenvironment that are inimical to the optimal induction of an antitumor immune response (7) . Examination of circulating and tumor-infiltrating DCs in tumor-bearing animals and in cancer patients has revealed that DCs are functionally impaired in their ability to induce T-cell responses (8, 9, 10, 11) . These deficits have been associated with down-regulation of MHC and costimulatory molecules (10 , 11) and tumor-induced apoptosis of DCs (12) .

One of the factors produced within the tumor microenvironment that might interfere with DC functions is TGF-β. TGF-β is a pleiotropic cytokine produced by cancer cells of different histological types (13, 14, 15, 16) . Among the plethora of immunosuppressive effects of TGF-β (17) is the capacity to interfere with several DC functions. These include down-regulation of cell surface MHC antigens, costimulatory molecules, chemokine receptors, as well as impairment of in vitro chemotaxis (18, 19, 20) .

Although the in vitro effects of TGF-β on DCs are relatively well known, the impact of TGF-β on the ability of DCs to migrate to secondary lymphoid organs and induce specific antitumor T-cell responses in vivo remains to be determined. In this study, we demonstrate that TGF-β inhibits DC migration to DLN and diminishes their capacity to stimulate IFN-γ secretion by tumor-sensitized T lymphocytes. Most importantly, we show that the combined use of antisense TGF-β gene transfer plus TGF-β-neutralizing antibody increases the efficacy of DC vaccines in treating established TGF-β-secreting 4T1 mammary tumors.

MATERIALS AND METHODS

Animals.

Six-week-old female BALB/c and C57BL/6 (B6) mice were purchased from The Harlan Laboratory (Indianapolis, IN). Six-week-old female BALB/c-TgN (DO11.10) 10 Loh mice were purchased from The Jackson Laboratory (Bar Harbor, ME). All mice were housed at the University of Arizona Animal Facilities in accordance with the Principles of Animal Care (NIH publication no. 85-23, revised 1985).

Tumors.

4T1 murine mammary tumor cells were kindly provided by Dr. Fred Miller (Michigan Cancer Foundation, Detroit, MI) and maintained as described previously (21) .

TGF-β-neutralizing Antibody (2G7).

The 2G7 mouse IgG1 mAb was generated after immunization of BALB/c mice with recombinant human TGF-β₁. 2G7 neutralized the growth inhibitory activity of TGF-β₁, TGF-β₂, and TGF-β₃ on Mv1Lu epithelial cells (22) .

Generation of DCs and TGF-β Treatment.

Bone marrow cells were harvested from flushed marrow cavities of femurs and tibiae under aseptic conditions and cultured with 100 units/ml granulocyte macrophage colony-stimulating factor and 100 units/ml interleukin 4 (Peprotech, Rocky Hill, NJ) at 10⁶ cells/ml in complete media (RPMI 1640 containing 10% heat-inactivated FBS, 0.1 mm nonessential amino acids, 1 μm sodium pyruvate, 2 mm l-glutamine, 100 μg/ml streptomycin, 100 units/ml penicillin, 0.5 μg/ml fungizone, and 5 × 10⁻⁵ m 2-mercaptoethanol). Cytokines were replenished on day 4. On day 6 of culture, DCs were collected and cultured at 10⁶ cells/ml with granulocyte macrophage colony-stimulating factor and interleukin 4 with or without the addition of 10 ng/ml of recombinant human TGF-β₁ (R&D Systems, Minneapolis, MN) for 6 days. DCs were matured with 200 units/ml of TNF-α (Peprotech) for 48 h.

FACS Analysis.

All antibodies used were purchased from Caltag Laboratories (Burlingame, CA) unless otherwise noted. For analysis of DCs, samples were stained with PE-conjugated anti-CD11c (BD PharMingen, San Diego, CA), FITC-conjugated anti-I-A^d (BD PharMingen), PE-conjugated anti-B7.1 (CD80), FITC-conjugated anti-B7.2 (CD86), or PE-conjugated anti-CD40. T cells were stained with PE-conjugated anti-CD3, FITC-conjugated anti-B220, PE-conjugated anti-CD8, or FITC-conjugated anti-CD4. Cells were analyzed using a FACStar^PLUS flow cytometer (Becton Dickinson Immunocytometry Systems, San Jose, CA).

Induction of Allogeneic Mixed Lymphocyte Reactions.

Spleen cells from B6 mice were harvested and enriched for CD3-positive cells using a T-cell enrichment column (R&D Systems). Cells were 80–95% CD3 positive as determined by FACS analysis. Varying numbers of DCs were incubated with 2 × 10⁵ T lymphocytes for 5 days in 96-well tissue culture plates (Sarstedt, Newton, NC) with the addition of 1 μCi of [³H]thymidine (Perkin-Elmer Life Sciences, Boston, MA) for the final 18 h of culture.

OVA Peptide Presentation Assay.

Spleen cells from DO11.10 OVA T cell receptor transgenic mice were enriched for CD3 positive cells as described above. Varying numbers of DCs were incubated with 2 × 10⁵ T lymphocytes in the presence of 1 μm of OVA peptide (ISQAVHAAHAEINEAGR; United Biochemical Research, Seattle, WA) in 96-well tissue culture plates for 5 days with the addition of 1 μCi of [³H]thymidine for the final 18 h of culture.

Endocytosis and Phagocytosis Assays.

Endocytosis and phagocytosis assays were performed using modifications of previously described procedures (23) . Endocytosis was measured by incubating 2 × 10⁵ DC with 400 μg of FITC-conjugated dextran beads, 40,000 MW (Molecular Probes, Eugene, OR) for 30 min at 4°C or 37°C. Phagocytosis was measured by incubating DCs with FITC-conjugated Escherichia coli (Molecular Probes) at a ratio of 100 E. coli particles to 1 DC for 60 min at 4°C or 37°C. After incubation, cells were washed extensively with PBS containing 0.5% bovine albumin and 0.1% sodium azide and analyzed by flow cytometry.

Stimulation of Tumor-sensitized T Lymphocytes.

Bone marrow-derived DCs were pulsed with 4T1 tumor cell lysate at a ratio of 3 tumor cell equivalents/DC for 24 h in the presence or absence of 10 ng/ml TGF-β₁. DCs were then matured with 20 ng/ml TNF-α in the presence or absence of 10 ng/ml TGF-β₁. Splenic T lymphocytes were purified from mice bearing 14 day 4T1 tumors as indicated above. Ten thousand DCs were incubated with 2 × 10⁵ splenic T lymphocytes in 96-well tissue culture plates for 5 days with the addition of 1 μCi of [³H]thymidine for the final 18 h of culture. One million tumor DLN cells from mice bearing 14 day 4T1 tumors were incubated with 2.5 × 10⁵ DCs for 48 h, and IFN-γ production was evaluated by ELISA (R&D Systems).

In Vivo Migration Assay.

Bone marrow-derived DCs were matured with 200 units/ml TNF-α with or without 10 ng/ml TGF-β₁ for 48 h. DCs were labeled with 10 μm PKH-67L, green fluorescent dye (Sigma, St. Louis, MO) as described previously (20) . Naive mice received s.c. injections in the right flank with 5–8 × 10⁶ DCs. Forty-eight h after injection, mice were sacrificed, and inguinal lymph nodes were harvested and disaggregated. Lymph node cells were centrifuged (Shandon, Pittsburgh, PA) onto glass slides at 700 rpm for 4 min. The slides were fixed with 4% paraformaldehyde and stained with a propidium iodide/RNase solution (Phoenix Flow Systems, San Diego, CA). Slides were analyzed using a laser scanning cytometer (CompuCyte, Cambridge, MA). Detection and contouring of cells was keyed by propidium iodide signal, whereas DCs were identified by their green fluorescence signal; 35,000 propidium iodide events were analyzed from each treatment group/experiment. The identity of each DC detected during scanning was confirmed visually by direct microscopic observation using the instrument’s relocation function.

In Vitro Chemotaxis Assay.

Bone marrow-derived DCs were matured with TNF-α in the presence or absence of TGF-β₁ for 48 h as indicated above. An in vitro chemotaxis assay was performed as described previously (24) . SLC (Peprotech) and MIP-3β (R&D Systems) were diluted with serum-free media to a final volume of 600 μl of 100 ng/ml chemokine and added to 24-well tissue culture plates (Corning Costar, Cambridge, MA). Transwell culture inserts (Corning Costar) with 6.5-mm diameter and 5.0-μm pore-size were inserted into each well, and DCs (4 × 10⁵ cells/each well) were added to the top chamber in serum-free media at a final volume of 100 μl. After the plates were incubated at 37°C in 5% CO₂ for 4 h, the cells in the bottom chamber were recovered, the migrating cells were counted, and an aliquot was stained with anti-CD11c mAbs to be analyzed by FACS. Controls included wells with chemokine in both the top and bottom chambers. The number of migrated cells was determined by subtracting the number of migrated cells in control wells from the number of migrated cells in experimental wells.

Treatment of Established Tumors.

Six-week-old BALB/c mice were orthotopically injected with 10⁴ 4T1-N or 4T1-asT tumor cells into the mammary gland. DCs were pulsed with 4T1 tumor cell lysate at a ratio of 3 tumor cell equivalents/dendritic cell for 24 h. After pulsing, DCs were matured with 200 units/ml TNF-α for 48 h. Mice were injected i.t. with 1.5 × 10⁶ tumor cell lysate-pulsed, matured DCs in 50 μl of PBS on day 15 when tumors were palpable. Vaccination was repeated on days 20 and 25. Two h before each vaccination, mice received i.p. injections of 300 μg of 2G7 mAb. In combination with DCs or alone, mice received 100 μg of 2G7 mAb i.t. Primary tumors were measured as reported previously (21) . Mice exhibiting complete tumor regression were challenged with 10-fold more 4T1 tumor cells (10⁵) and monitored for tumor growth.

Statistical Analysis.

For all analyses, student t tests were performed using Prism software (GraphPad, San Diego, CA). Ps of <0.05 were considered to indicate significant differences between data sets.

RESULTS

Antigen Uptake and Presentation by DCs.

Before evaluating the effect of TGF-β in vivo, we evaluated its effects on several DC functions in vitro. Treatment of DCs with TGF-β caused a significant (P < 0.05) decrease in the levels of B7.1 and CD40 (Fig. 1A) ⇓ , as well as in the percentage of cells expressing CD40 (Fig. 1B) ⇓ . Next, we evaluated the ability of DCs to endocytose dextran particles and phagocytose E. coli particles after TGF-β treatment. TGF-β exposure inhibited the endoocytic and phagocytic capacities of DCs by 58% and 87%, respectively (Fig. 1C), a pattern that was observed in all three experiments conducted. To evaluate the effect of TGF-β exposure on antigen presentation, mixed lymphocyte reactions were performed. Stimulation of allogeneic T lymphocytes (Fig. 1D) ⇓ and presentation of OVA peptide (Fig. 1E) ⇓ by DCs was significantly (P < 0.05) suppressed after TGF-β exposure. These results demonstrate that TGF-β inhibits the uptake and presentation of antigens by DCs.

Induction of Tumor-sensitized T Lymphocytes.

Observing that TGF-β inhibits the antigen uptake and presentation capacities of DCs, we evaluated its effect on the stimulation of tumor-sensitized T lymphocytes by tumor cell lysate-pulsed immature and mature DCs. TGF-β-treated mature DCs were significantly less effective (P = 0.0002) than untreated mature DCs at stimulating the proliferation of tumor-sensitized T lymphocytes (Fig. 2A) ⇓ . The ability of immature and mature DCs to stimulate IFN-γ production by tumor-sensitized lymphocytes was also significantly inhibited (P < 0.0001, P = 0.0003) by 60 and 22%, respectively, after TGF-β exposure. Taken together, these data demonstrate that TGF-β exposure reduces the ability of DCs to stimulate antitumor immune responses.

In Vivo Migration and In Vitro Chemotaxis.

Because migration of DCs to secondary lymphoid organs is necessary for T-cell priming (25) , we assessed the effect of TGF-β treatment on the ability of mature DC to migrate to DLNs. For this purpose, TGF-β-treated DCs were labeled with PKH-67 and injected s.c. into the hind flank of mice. Forty-eight h later, draining inguinal lymph nodes were harvested, and infiltrating DCs were enumerated by scanning laser cytometry. TGF-β treatment resulted in a decrease in migration of mature DCs to the DLNs (Table 1) ⇓ . This was a significant (P < 0.05) trend observed over four independent experiments. To determine the mechanism responsible for decreased migration of mature DCs, in vitro chemotaxis toward SLC and MIP-3β was evaluated. DCs migrating to the bottom chambers were recovered and analyzed by flow cytometry for CD11c expression. No significant difference in CD11c expression was observed between untreated and TGF-β-treated DCs (data not shown). TGF-β treatment caused reduced migration of DC toward both SLC and MIP-3β, however, only migration toward MIP-3β was significantly decreased (P < 0.05; Fig. 3 ⇓ ).

Treatment of Established Tumors with DCs and Neutralizing TGF-β Antibody.

The impairment of critical DC functions, including antigen uptake and antigen presentation, tumor-specific T lymphocyte stimulation, and in vivo migration by TGF-β, suggested that inhibition of TGF-β production in tumor-bearing animals may improve the efficacy of DC vaccines. To test this possibility, TGF-β production was suppressed by transfer of an antisense TGF-β transgene into 4T1 cells (4T1-asT) as described previously (21) . Expression of the transgene resulted in >90% inhibition of TGF-β production (0.083 ± 0.003 ng/ml in 4T1-asT compared with 1.244 ± 0.188 ng/ml in mock-transduced cells (4T1-N). Treatment of 4T1-asT tumors with 2G7 mAb alone or DCs plus 2G7 mAb significantly inhibited (P < 0.05) tumor growth compared with 4T1-asT tumors treated with DCs alone (Fig. 4) ⇓ . Furthermore, complete tumor regression occurred in 40% (two of five) of 4T1-asT-tumor-bearing mice that were treated with DCs plus 2G7 mAb (Fig. 4) ⇓ . Similarly, mock-transduced (4T1-N) tumors responded significantly better (P < 0.05) to treatment with DCs plus 2G7 mAb as compared with treatment with DCs alone; however, tumor growth inhibition in this group was inferior to that observed in animals bearing 4T1-asT tumors. (Fig. 4) ⇓ . Tumor growth was not affected by treatment with isotype control (IgG) antibody (data not shown). To determine whether the mice that exhibited complete tumor regression had developed long-term immunity, they were rechallenged with parental 4T1 tumor cells. These mice failed to develop tumors (zero of two); however, all control naive mice challenged with tumor cells developed tumors (three of three). These data suggest that neutralization of TGF-β in mice bearing TGF-β-secreting tumors enhances the effectiveness of DC vaccines in treating established tumors.

DISCUSSION

In this study, we evaluated the impact of TGF-β on in vivo migration and immunostimulatory activities of DCs. Our results demonstrate that TGF-β treatment diminishes the ability of DC to migrate to secondary lymphoid organs and to induce T-cell responses. Most importantly, suppression of tumor-derived TGF-β by antisense TGF-β gene transfer plus neutralization of secreted TGF-β with anti-TGF-β mAb significantly improved the antitumor activity of intratumorally injected tumor cell lysate-pulsed DCs. Previous studies have demonstrated that suppression of TGF-β by antisense gene transfer (13 , 21 , 26) or abrogation of TGF-β using neutralizing antibody (27 , 28) increases tumor immunogenicity, leading to tumor growth inhibition or rejection. To our knowledge, this is the first study to evaluate the impact of both approaches simultaneously on DCs in controlling established tumors. The finding that the antitumor effect was most evident when antisense TGF-β-expressing tumors were treated with DCs plus anti-TGF-β mAb directly implicates tumor-derived TGF-β in tumor progression. In our study, TGF-β within the tumor milieu could be promoting tumor growth by interfering with the antigen-presenting and effector functions of DCs at the tumor site, as well as preventing the emigration of injected DCs to DLNs to activate naïve tumor-specific T lymphocytes. The former possibility is supported by the recent findings by Kirk et al. (29) who suggested that migration of i.t. injected DCs to DLNs is not required for the induction of an antitumor response. Using SLC gene-modified DCs, they demonstrated comparable tumor growth inhibition in normal mice and lymphotoxin a^−/− mice lacking peripheral lymph nodes (29) . The latter possibility is supported by the decreased chemotactic response of TGF-β-treated DCs to MIP-3β and SLC observed in our study. These chemokines produced in the lymph nodes recruit DCs via their interaction with the chemokine receptor, CCR7 (20 , 30) . A possible explanation is that CCR7 gene expression is inhibited in DC by TGF-β treatment as has been reported by others (19 , 20) .

As with previously published studies (29 , 31 , 32) , only a minute fraction (<1%) of s.c. injected DCs in our study migrated to DLNs. It is yet to be determined if these lymph node-infiltrating DCs represent a unique subpopulation capable of singularly stimulating the antitumor response or require the participation of endogenous DCs to achieve this goal. In the setting of DC vaccination to prevent or treat inaccessible micrometastases, it is desirable that adoptively transferred DCs are able to migrate to secondary lymphoid organs to stimulate naïve T lymphocytes. Thus this LN-infiltrating, ex vivo manipulated DC population needs to be more actively studied.

In summary, this study demonstrates the potential usefulness of a combined therapeutic approach to eliminate immunosuppressive tumor-derived factors to improve the effectiveness of DC-based vaccines.