CD4⁺ T CELLS ELICIT HOST IMMUNE RESPONSES TO MHC-II⁻ OVARIAN CANCER THROUGH CCL5 SECRETION AND CD40-MEDIATED LICENSING OF DENDRITIC CELLS

Abstract

T cell adoptive transfer strategies that have produced clinical remissions against specific tumors have so far produced disappointing results against ovarian cancer. Recent evidence suggests that adoptively transferred CD4⁺ T cells can trigger endogenous immune responses in particular ovarian cancer patients through unknown mechanisms. However, conflicting reports suggest that ovarian cancer-infiltrating CD4⁺ T cells are associated with negative outcomes. Here we elucidate the phenotypic attributes that enable polyclonal CD4⁺ T cells briefly primed against tumor antigens to induce therapeutically relevant endogenous anti-tumor immune responses. Our results unveil a therapeutic mechanism whereby tumor-primed CD4⁺ T cells transferred into ovarian cancer-bearing mice secrete high levels of CCL5, which recruits endogenous CCR5⁺ dendritic cells to tumor locations, and activate them through CD40-CD40L interactions. These newly matured DCs are then able to prime tumor-specific endogenous CD8⁺ T cells, which mediate long-term protection. Correspondingly, administration of tumor-primed CD4⁺ T cells significantly delayed progression of MHC-II⁻ ovarian cancers, similarly to CD8⁺ T cells only, and directly activated wildtype, but not CD40-deficient DCs recruited to the tumor microenvironment.

Our results unveil a CCL5-, CD40L-dependent mechanism of transferring immunity from exogenously activated CD4⁺ T cells to tumor-exposed host cells resulting in sustained antitumor effects. Our data provide a mechanistic rationale for incorporating tumor-reactive CD4⁺ T cells in adoptive cell transfer immunotherapies against ovarian cancer and underscore the importance of optimizing immunotherapeutic strategies for the specific microenvironment of individual tumors.

INTRODUCTION

T cells are the only leukocyte subset in the ovarian cancer microenvironment known to exert spontaneous immune pressure against tumor progression, such that the infiltration of tumor islets by CD3⁺ T cells was definitively associated with a better outcome in a large cohort of patients (1). Interestingly, subsequent studies confirming the protective role of T cells in ovarian cancer suggest that CD4⁺ T cells actually have an unfavorable effect on prognosis, restricting the beneficial mediators of antitumor immune pressure to CD8⁺ T cells (2).

CD8⁺ lymphocytes have the ability to recognize and directly kill tumor cells, and can be readily isolated from certain tumors, activated and expanded ex vivo for reinfusion into cancer patients to achieve varying degrees of responses (3–7). Thus CD8⁺ T cells have been the focus in generating potent adoptive immunotherapies (8). Yet, studies dating back more than two decades, indicate that in the absence of CD8⁺ T cells, CD4⁺ T cells directly act to eradicate both solid and haematologic malignancies (9, 10). Since then, several reports have confirmed the contributory role of CD4⁺ T cells in antitumor immunity, culminating with a report in 2008 by Hunder et al, of a single case of complete response attained by a patient receiving adoptive immunotherapy with CD4⁺ T cell clones specific to the NY-ESO-1 antigen (11), demonstrating the compelling prospects of CD4⁺ T cells in adoptive immunotherapy against other tumors. Lastly, successful adoptive T cell administration protocols that have induced impressive clinical responses against other tumors have utilized a mixture of CD4⁺ and CD8⁺ T cells (4).

The beneficial or detrimental effect of including expanded CD4⁺ T cells in adoptive immunotherapies specifically against ovarian cancer remains unclear. To our knowledge, evidence is limited to a recent clinical report describing a therapeutic endogenous boost in anti-tumor immunity in 2 out of 4 ovarian cancer patients treated with intraperitoneally infused peripheral blood Th1 cells expanded against the tumor antigen MUC1 (12). Since the expression of MHC-II on these tumors is unknown, the mechanisms mediating this potentially Th1-dependent therapeutic effect need to be clarified. Confirming the therapeutic potential of CD4⁺ T cells and defining their mechanisms of anti-tumor activity is necessary for designing expansion protocols that maximize the phenotypic attributes required for their efficacy.

Conventional thought has limited the effects of the CD4⁺ T cell population in antitumor immunity to a role of merely providing additional and beneficial, yet non-essential stimuli for supporting the maintenance of the anti-tumor CD8⁺ T cell population (13, 14). CD4⁺ helper T cells may achieve this goal through the production of cytokines like IL-2 and IFN-γ that both regulate the responses of antigen-presenting cells and function in the differentiation, expansion and maintenance of cytolytic CD8⁺ T cells (15, 16). CD4⁺ T cells also contribute to the maturation of antigen presenting cells through interactions between CD40 on the Antigen-Presenting Cell (APC) and its cognate ligand CD40L/CD154 on the T cell resulting in the licensing of the APC. The hallmarks for this licensing process are the upregulation of co-stimulatory molecules, subsequent secretion of IL-12 and corresponding ability of the licensed APC to prime naïve T cells against antigen (17–19).

We recently demonstrated that the adoptive transfer of briefly primed, tumor antigen-reactive, polyclonal T cells, coupled with the depletion of tumor-associated immunosuppressive CD11c⁺ dendritic cells (DCs), depending on the tumor model, induced the rejection or significantly delayed the progression of established ovarian cancer (20). CCL5 was required for these therapeutic benefits, although its source remained undetermined, as long-term protection was directly mediated by host (endogenous) immune cells. The studies presented here aimed to elucidate the suitability of including tumor antigen-reactive CD4⁺ T cells in adoptive transfer protocols against ovarian cancer. Based on our results, we have defined certain phenotypic attributes required for maximizing their observed therapeutic effectiveness. We demonstrate that CCL5, which recruits antigen-presenting cells to the tumor site, is primarily produced by adoptively transferred CD4⁺ T cells. CD4⁺ T cells are also necessary for activating these recruited APCs at tumor locations through CD40-CD40L interactions, and thereby stimulating therapeutically relevant host-derived antitumor CD8⁺ T cells. Therefore, CD4⁺ T cells may be critical for the efficacy of future adoptive T cell therapies against ovarian cancer. Our results provide a mechanistic rationale for including CCL5-secreting, CD40L-expressing CD4⁺ T cells as a beneficial and pertinent part of the antitumor armament of adoptive immunotherapy, particularly against lethal epithelial ovarian cancer.

MATERIALS AND METHODS

Mice and Tumor lines

C57BL/6 (B6), CD45.1 (B6-Ly5.2) and FVB mice were purchased from the National Cancer Institute (Frederick, MD). CCL5-deficient (#005090) CD40L-deficient (#002770) and CD40-deficient (#002928) mice on a C57BL/6 background were purchased from The Jackson Laboratory. Experiments were conducted in accordance with the Dartmouth Medical School guidelines. Lewis lung carcinoma, MT2, ID8-Defb29/Vegf-a, cultured T cells and DCs were maintained in RPMI medium containing 10% fetal bovine serum.

Tumor Inoculation

To develop ovarian tumors, mice were injected i.p. with 1.5 × 10⁶ ID8-Defb29/Vegf-a cells. The ID8 ovarian cancer model was originally derived from the spontaneous transformation of C57BL6 mouse surface epithelial cells after rounds of in vitro passaging (21). The ID8-Defb29/Vegf-a ovarian cancer model was formed from the stable transfection of the original ID8 cell-line with Defb29 and VEGF-a, resulting in accelerated tumor progression and ascites formation compare to the parental cell-line, which recapitulates the microenvironment of human solid ovarian carcinomas in a more aggressive and more faithful manner (22–26). Lewis lung carcinoma cells (American Type Culture Collection, Manassas, VA, USA) and MT2 ovarian cancer cells (a kind gift of S. Orsulic (27, 28)), were intraperitoneally injected (1.0 × 10⁶ in 0.2 ml of PBS). Treatments were always administered to mice bearing established tumors (at least 7 days). For protection experiments, mice were subcutaneously challenged with 15×10⁶ ID8-Defb29/Vegf-a in Matrigel.

Flow cytometry and cell sorting

Anti-mouse antibodies specific for CD45.1 (A20), CD45.2 (104), CD3 (145.2C11), CD4 (GK1.5), CD8 (Ly-2), CD11c (N418), MHCII (M5/114.15.2), Gr-1 (RB6-8C5), CD14 (rmC5-3), CD80 (B7-1), CD86 (GL1) CD40 (HM40-3), CD40L (MR-1), CD70 (FR70) and CCR5 (7A4) were all obtained from either eBioscience, BD Biosciences or BioLegend.

IL-12 and CCL5 ELISA

Sorted CD4⁺, CD8⁺, unsorted CD3⁺ T cells and CD40L^−/− T cells that had been activated with tumor-pulsed dendritic cells were incubated at a concentration of 1.0 × 10⁶ cells/ml in complete media overnight. 100 μls of supernatants obtained from these cultures were then plated on a 96-well microplate coated with the CCL5 capture antibody from the mouse CCL5 DUOSET ELISA development kit (R&D Systems, Minneapolis, MN). The ELISA was conducted as per manufacturer’s instructions. IL-12 (p70) ELISA was performed with 100 μls of the supernatants obtained from wild-type or CD40^−/− dendritic cells cultured with wild-type or CD40L^−/− T cells for 24 hours. IL-12 (p70) ELISA was performed according to manufacturer’s directions (eBioscience, San Diego, CA).

T Cell Expansion, Adoptive Immunotherapy and DC Depletion

Bone marrow derived DCs were produced (20), then pulsed overnight (10:1) with irradiated (10, 000 Rads) and UV-light exposed ID8-Defb29/Vegf-a. Approximately 2.5 × 10⁸ splenocytes were harvested from CD45.1⁺ mice and cultured at a concentration of 2×10⁶ cells/ml for 7 days with ID8-Defb29/Vegf-a - pulsed DCs and 10 U/ml recombinant human IL-2 (Peprotech). 6–8 week old mice were injected i.p. with 1.5 ×10⁶ ID8-Defb29/Vegf-a cells and treated on day 7 and 14 with i.p. adoptive transfer of 1.5 × 10⁶ in vitro activated splenic T cells. Lymphopenia was induced by sublethal TBI (300 Rads) of tumor-bearing mice on day 7, 5 hours prior to ACT. DCs were eliminated by i.p. administration of 10 μg/mouse of an anti-CD11c-immunotoxin (25) on the day prior to T cell transfer (day 6) and thrice weekly for two weeks thereafter.

Protection Experiments

CD45.1⁺ mice bearing ID8-Defb29/Vegf-a tumors were treated with wild-type, CD40L^−/− or CCL5^−/− T cells along with 3 weekly injections of an anti-CD11c cell depleting immunotoxin. On day 28 of tumor progression, when transferred T cells are no longer detectable (20), host CD3⁺ T cells were obtained from the spleens of treated mice, by FACS. CD45.2⁺ mice that would become the recipients of the sorted host (CD45.1⁺) T cells were irradiated (300γ) at least 5 hours prior to transfer. 3×10⁶ host T cells from the mentioned groups were transferred i.v. into irradiated recipient mice. Two days after transfer, mice were challenged with subcutaneous flank injections of 15×10⁶ ID8-Defb29/Vegf-a tumor cells. Cell sorting was performed with a FACSAria (BDBiosciences). CD3 expression was confirmed by flow cytometry on a FACSCanto (BD Biosciences). Tumor growth was measured over time and reported in units of volume.

DC maturation assays

CD40L^−/− or wild-type T cells were expanded with DCs pulsed with irradiated tumor cells and IL-2 as described above (T cell Expansion), for 7 days. CD45.2⁺CD11c⁺MHCII⁺ cells were obtained from the peritoneum of 7-day old ID8-Defb29/Vegf-a tumor-bearing CD40^−/− and wild-type mice by cell sorting, then co-cultured with expanded T cells at a ratio of 1 DC: 2 T cells. At least 10⁵ DCs were cultured per condition. The expression of the maturation markers CD70, CD80, CD86 and MHCII were then determined by flow cytometry at the indicated timepoints.

Enzyme-linked Immunospot Assay (ELISPOT)

T cells from wild-type or CD40L-deficient T cells were activated for 7 days against tumor antigen as described above (T cell Expansion). Wild-type or CD40-deficient DCs sorted from tumor-bearing mice were co-cultured for 24 hours with activated wild-type or CD40L-deficient T cells then admixed with tumor-infiltrating CD8⁺ T cells for 48hours in ELISPOT analyses. Flat-bottomed, 96-well nitrocellulose-lined plates (Millipore, Bedford, MA) were coated with IFN-γ mAb (AN-18; eBioscience, San Diego, CA) or Granzyme-B mAb (R&D Systems, Minneapolis, MN) and incubated overnight at 4°C. After washing with Washing Buffer (0.05% Tween 20 in PBS), plates were blocked with RPMI supplemented with 10% FBS for 2 h at RT or blocking buffer (R&D). BMDCs preciously primed for 24h with T cells as indicated, were cultured with sorted 10⁵ CD8⁺ T cells from tumor-bearing mice at a 10 T cell: DC ratio for 48 h in 10% FBS RPMI medium. After incubation, the plates were washed with washing buffer, and after addition of biotinylated secondary IFN-γ mAb (R4-6A2; eBioscience) or Granzyme-B mAb(R&D) were left to incubate for 24h at 4°C. Plates were washed and developed with Avidin-horseradish peroxidase (eBioscience) or Streptavidin-AP (R&D) for 2 h at room temperature. After washing, fresh substrate (3-amino-9-ethyl carbazole, Sigma, St. Louis, MO) or BCIP/NBT (R&D) was added and the plates incubated for approximately 45 minutes.

Enumeration of adoptively transferred and host cells

At indicated timepoints, peritoneal lavages were performed on mice. Upon RBC lysis, cells were enumerated by trypan blue exclusion then analyzed by flow cytometry for expression of CD3, CD4, CD11c, MHCII, CD80 and congenic markers – CD45.2 (host) and CD45.1 (transferred). Absolute cell numbers were determined by multiplying the percentage of the relevant population (as determined by flow cytometry) by the total cell count per organ.

Statistical Analysis

Tumor survival curves were compared using Log Rank Test (Graphpad, Prism). Statistical differences between numbers of spots in ELISPOT assays, proportions of cells in flow cytometry, total number of cells or tumor volumes were determined by Mann Whitney test (Graphpad, Prism). Probability (P) values less than 0.05 were considered statistically significant.

RESULTS

CCL5 is required within the transferred T cells for their ability to trigger sustained host immunity

We previously demonstrated that a mixed population of naïve T cells shortly primed against tumor antigens was able to induce antitumor effects leading to delayed progression of tumors in preclinical models of established ovarian cancers. The transfer of these tumor-specific T cells was found to evoke host (endogenous) anti-tumor responses, which were paramount for the observed therapeutic effects and involved the upregulation of CCL5 (20). To demonstrate that CCL5 secreted by transferred lymphocytes (and not by the host (endogenous) immune cells activated by treatment) is required for inducing therapeutic immunity against our aggressive intraperitoneal ID8-Defb29/Vegf-a ovarian cancer model (20, 22, 24, 25), we first administered equal numbers of CCL5-deficient or wild-type tumor-primed total T cells into mice growing established tumors. Restricting the requirement of CCL5 to the transferred T cell compartment, tumor-primed CCL5^−/− lymphocytes administered with or without depleting immunosuppressive DCs with an anti-CD11c immunotoxin (20), induced no significant survival increase, compared to tumor DC depletion alone (Figure 1A – B). In contrast, equal numbers of identically primed wild-type T cells with similar CD4:CD8 ratios of 2:3, significantly increased the survival of tumor-bearing mice, and their therapeutic benefits synergized with the concurrent elimination of tumor-associated immunosuppressive DCs, supporting our previous report (20) (Figure 1A – B). The therapeutic effects elicited by the administration of wild-type tumor-reactive T cells were associated with a stronger accumulation of activated (CD80⁺) DCs expressing CCR5 (Supplemental Figure 1A) at tumor (peritoneal) locations, compared to the transfer of CCL5^−/− T cells (Figure 1C). In contrast, the absence of CCL5 within the host population had no effect on the therapeutic benefits elicited by the adoptive transfer of wild-type tumor-reactive T cells in conjunction with the elimination of CD11c⁺ regulatory DCs (Figure 1D). Thus, ID8-Defb29/Vegf-a tumors progressed similarly in untreated CCL5^−/− mice as in wild-type mice, while treatments induced comparable survival increases (Figure 1D). Taken together, these data indicate that CCL5 is required within the transferred T cell compartment but not within the host compartment to obtain the therapeutic benefits of ACT in established preclinical ovarian cancer.

Figure 1. CCL5 is required within transferred T cells for effective adoptive immunotherapy.

A. Survival of ID8-Defb29/Vegf-a-tumor bearing mice treated with wild-type (wt) T cells, CCL5^−/− T cells or left untreated. B. Survival of ID8-Defb29/Vegf-a-tumor bearing mice treated with wild-type (wt) T cells or CCL5^−/− T cells combined with depletion of immunosuppressive DCs with an anti-CD11c immunotoxin (IT). C. Quantification of MHCII⁺CD80⁺CD11c⁺ cells in the peritoneum of mice treated with wild-type (wt) T cells, CCL5^−/− T cells or left untreated. Mice were depleted of immunosuppressive dendritic cells with an anti-CD11c immunotoxin prior to T cell transfer. D. Survival of ID8-Defb29/Vegf-a challenged wild-type or CCL5-deficient mice (host) treated with wild-type (wt) anti-tumor T cells or left untreated (PBS). (n=6 mice per group in three independent experiments). (* means P<0.05; **means P<0.01).

We and others have demonstrated that the long-term antitumor immunity elicited by transferred (exogenous) tumor-reactive T cells is ultimately elicited in ovarian cancer by T cells of host (endogenous) origin (12, 20). We therefore investigated whether CCL5 specifically produced by adoptively transferred T cells was required for the mobilization of endogenous memory T cells that provide sustained protection. For that purpose, we depleted immunosuppressive DCs from mice bearing established intraperitoneal tumors and treated them with congenically labeled (CD45.1⁺) T cells, briefly primed against ID8-Defb29/Vegf-A-tumor, from either wild-type or CCL5-deficient mice. After 28 days of tumor-progression, splenic T cells of host origin (CD45.2⁺) were obtained from both groups, and from mice that did not receive T cell transfer but were treated with PBS only, and transferred (3×10⁶/mouse) into irradiated naive mice, which were challenged 48 hours later with flank ID8-Defb29/Vegf-A tumors (Experimental Design, Supplemental Figure 1B). As shown in Figure 2 and Supplemental Figure 1C, host-derived T cells transferred from tumor-bearing mice previously treated with (exogenous) wild-type tumor-reactive T cells elicited significant protection against ID8-Defb29/Vegf-A tumors, compared to endogenous T cells transferred from mice treated with either (exogenous) CCL5-deficient T cells, or PBS (P<0.01 for both comparisons, Mann-Whitney). Taken together, these results suggest that CCL5 specifically secreted by adoptively transferred T cells mediates the recruitment of activated (CCR5⁺) DCs to the tumor microenvironment (20), and demonstrate that CCL5 is essential for inducing persistent changes in host (endogenous) T cells that mediate sustained anti-tumor protection.

Figure 2. Adoptive (T) Cell Therapy (ACT) induces sustained host immunity that is protective against secondary tumor challenges and is dependent upon the presence of CCL5 in transferred T cells.

Mice bearing i.p. ID8-Defb29/Vegf-a tumors were depleted of immunosuppressive dendritic cells with an anti-CD11c immunotoxin (IT) then treated with expanded T cells from wild-type or CCL5-deficient mice on days 7 and 14 of tumor progression. After 28 days of tumor progression host CD3⁺ were sorted from treated mice and transferred into naïve mice that received s.c. ID8-Defb29/Vegf-a tumors 2 days later. Tumor growth of s.c. ID8-Defb29/Vegf-a tumors in mice receiving host CD3⁺ T cells from mice treated with wildtype T cells, CCL5^−/− T cells or left untreated. (n=4 mice per group in three independent experiments).

Independent adoptive transfer of CD4⁺ and CD8⁺ T cells elicit comparable therapeutic effects in hosts bearing MHC-II⁻ tumors, but are inferior to mixed T cells

To dissect the mechanisms whereby individual T cell subsets contribute to the observed therapeutic effect, CD45.1⁺ T cells were primed as previously described (20), then CD4⁺ and CD8⁺ T cells were separated by FACS sorting and transferred individually into established ovarian cancer-bearing mice, depleted of tumor-associated immunosuppressive DCs through the intraperitoneal administration of an anti-CD11c immunotoxin (20). As expected, CD8⁺ T cells were independently able to significantly enhance the survival of mice bearing aggressive orthotopic ID8-Defb29/Vegf-a tumors by 30.1 % (Figure 3A). Surprisingly, identical numbers of CD4⁺ T cells alone induced comparable therapeutic benefits in multiple experiments (25.3% increase in survival, Figure 3A), although the transfer of a combination of CD4⁺ and CD8⁺ T cells (CD3) had the greatest impact on survival (53% increase in survival; Figure 3A). Corresponding effects were observed in the absence of regulatory DC depletion although, supporting our previous findings (20, 25), the therapeutic benefit elicited by adoptively transferred tumor-primed T cells alone was significantly weaker (Supplemental Figure 1D).

Figure 3. CD4⁺ T cells are pertinent for the effects of adoptive T cell therapy in enhancing the survival of mice bearing ovarian tumors.

IT means anti-CD11c immunotoxin, CD4 represents mice receiving CD4⁺ T cell transfer, CD8 represents mice receiving CD8⁺ T cell transfer and CD3 represents mice receiving a mixed population (3 CD8⁺: 2 CD4⁺) of CD3⁺ T cells. A. Mice bearing 7-day-old ID8-Defb29/Vegf-a tumors were treated on day 7 and 14 of tumor progression with either a mixed population of CD3⁺ T cells or pure populations of CD4⁺ or CD8⁺ T cells combined with the depletion of CD11c⁺ cells. Mice were monitored for survival. B. Secretion of CCL5 as determined by ELISA, for a mixed population of CD3⁺ T cells and individual pure cultures of CD4⁺ and CD8⁺ T cells immediately after expansion. C. Total numbers of host CD80⁺MHCII⁺CD11c⁺ cells accumulated at the tumor site 3 days after treatment. D. Survival of mice bearing ID8-Defb29/Vegf-a treated with wildtype or CCL-5^−/− deficient CD4+ T cells along with depletion of CD11c⁺ cells. E. Total numbers of host CD80⁺MHCII⁺CD11c⁺ cells accumulated at the tumor site 3 days after treatment in D. (n=6 mice per group in four independent experiments). (* means P<0.05; **means P<0.01).

As ID8-Defb29/Vegf-a tumor cells do not express MHC-II in culture or in vivo (Supplemental Figure 1E), they are not directly cleared by CD4⁺ T cells. Interestingly, we found that the individual transfer of CD4⁺ T cells resulted in 32% more CCR5⁺CD11c⁺ DCs at the tumor (peritoneal) site, compared to the individual transfer of CD8⁺ T cells (Supplemental Figure 1F). This was not due to in situ differentiation of immature progenitors into DCs, because >95% of tumor-associated leukocytes in both groups were either: 1) bona fide CD3⁺ T cells; 2) already differentiated CD11c⁺ DCs; or 3) committed F4/80⁺ macrophages at the time of adoptive transfer, supporting the concept that these were newly recruited DCs. Correspondingly, adoptively transferred CD4⁺ T cells were found to produce 4 times more CCL5 than the same number of CD8⁺ T cells (Figure 3B), suggesting that transferred (exogenous) CD4⁺ T cells represent the main source of this chemokine. In addition, there was a significantly higher proportion of CD80⁺MHCII⁺CD11c⁺ DCs in mice receiving adoptively transferred CD4⁺ vs. CD8⁺ T cells (Figure 3C).

CCL5 specifically produced by transferred tumor-reactive CD4⁺ T cells is required for the elicitation of therapeutic immunity in multiple models of peritoneal carcinomatosis

To confirm that the effects specifically triggered by CD4 T cells were mediated at least in part through their superior production of CCL5, negatively immunopurified CCL5-deficient and wild-type CD4 T cell splenocytes were simultaneously activated against tumor antigen and then transferred into wild-type mice growing established ID8-Defb29/Vegf-a ovarian tumors. Supporting the importance of CCL5 for the therapeutic effectiveness of adoptively transferred lymphocytes, CD4⁺ T cells that were deficient in CCL5 were impaired in their ability to delay tumor progression in comparison to wild-type T cells (Figure 3D). Correspondingly, fewer total DCs and thus fewer mature (MHCII⁺CD80⁺) DCs were present at the tumor site upon adoptive transfer of CCL5^−/− CD4⁺ T cells compared to wild-type CD4⁺ T cells (Figure 3E). These data suggest that CD4⁺ T cells not only induce the recruitment of greater numbers of host-derived DCs, but also play a role in their activation at tumor locations, and that CCL5 within the CD4 compartment is necessary for these effects.

Although the ID8-Defb29/Vegf-a model recapitulates the massive recruitment of CD11c⁺DEC205⁺CD8a⁺ leukocytes found in most solid human ovarian carcinoma specimens better than any other orthotopic model (22, 23, 25), we next aimed to rule out the unlikely event that the transduction of this aggressive cell line could obscure any homeostatic host events. Supporting the general applicability of our findings, mice with advanced Lewis lung peritoneal carcinomatosis treated with T cells expanded to tumor antigen survived significantly longer than untreated mice (Figure 4A). The transferred T cells secreted levels of CCL5 comparable to T cells activated against ID8-Defb29/Vegf-a tumors (Figure 4B), which was required for their therapeutic effect, as transferred CD4⁺ T cells deficient in CCL5 were unable to induce the same delay in tumor progression as wild-type CD4⁺ T cells (Figure 4C), suggesting a similar mechanism maybe playing a role. Furthermore, FVB mice challenged with the MT2 strain of ovarian carcinoma also presented increased numbers of DCs expressing higher levels of CD80 and CD70 in mice treated with tumor-activated T cells (that also secrete CCL5 Figure 4B) than their untreated counterparts (Figure 4D–E). Therefore, the induction of host immunity to peritoneal carcinomatosis through adoptive T cell therapy may be independent of genetic and haplotype differences, and relevant to the natural diversity observed within patients with peritoneal tumors of different types.

Figure 4. CCL5 specifically produced by transferred tumor-reactive CD4⁺ T cells is required for the elicitation of therapeutic immunity in multiple models of peritoneal carcinomatosis.

A. Survival of advanced Lewis lung carcinoma-bearing mice treated with wild-type (wt) T cells (day 7) and regulatory CD11c⁺ cell depletion from tumor locations (day 6, day 8 and day 10), or left untreated. B. Quantification of CCL5 produced by C57BL/6 T cells directly after their in vitro expansion against ID8-Defb29/Vegf-a (BL/6 v ID8) tumor cells or Lewis lung carcinoma cells (BL/6 v LLC) and FVB T cells against MT2 tumor cells (FVB v MT2). C. Survival of advanced Lewis lung carcinoma -bearing mice treated with wildtype (wt) CD4⁺ T cells or CCL5^−/− CD4⁺ T cells combined with depletion of immunosuppressive DCs with an anti-CD11c immunotoxin (IT). D. CD80⁺CD11c⁺ and E. CD70⁺CD11c⁺ cells in the peritoneum of MT2-tumor bearing mice treated with wild-type (wt) T cells (day 7) and DC elimination (day 6, 8, 10, 13 and 15) or left untreated. (n=6 mice per group in three independent experiments). (* means P<0.05; **means P<0.01).

CD40 signaling is required for the antitumor effects elicited by adoptively transferred T cells

As the expanded (CD4⁺) T cells express CD40L (Figure 5A), we hypothesized that they could activate newly recruited host DCs through CD40 signaling. Supporting this proposition, transferred T cells that were deficient in CD40L recruited similar numbers of leukocytes to the tumor site, but this population contained only half as many mature host CD80⁺MHCII⁺CD11c⁺ DCs as in wild-type treated mice (Figure 5B). Correspondingly, tumor-bearing mice depleted of tumor-associated regulatory DCs survived for significantly shorter periods upon adoptive transfer of CD40L-deficient tumor-reactive T cells than wild-type T cells (Figure 5C). In agreement with our previous findings, similar effects were observed without eliminating immunosuppressive DCs (20, 25), although the therapeutic responses observed after the individual administration of tumor-primed T cells were not as dramatic (data not shown). Importantly, this reduced antitumor effect was not caused by impaired secretion of the chemokine CCL5, as CD40L^−/− T cells secreted as much CCL5 as wild-type T cells (Figure 5D) and show identical CD4:CD8 ratios (Figure 5E) and activation markers (Figure 5F – G) after expansion, suggesting that the inability of the (CD4⁺) T cells to signal through CD40 on newly recruited DCs prevented their maturation in situ.

Figure 5. CD40-CD40L interactions are required for the therapeutic effects and induction of host immunity elicited by ACT.

A. Expression of CD40L on T cells primed against tumor antigens for 7 days. Shaded gray histogram represents Isotype control; unshaded histogram represents CD40L staining. B. Quantification of MHCII⁺CD80⁺CD11c⁺ cells in the peritoneum of mice treated with wild-type (wt) T cells, or CD40L^−/− T cells. C. Survival of ID8-Defb29/Vegf-a tumor-bearing mice treated with wild-type (wt) T cells, CD40L^−/− T cells, combined with the elimination of regulatory DCs (IT), or left untreated. D. Concentration of CCL5 secreted by wild-type (wt) or CD40L^−/− T cells upon 7-day expansion, as determined by ELISA. (n=6 mice per group in four independent experiments). E. T cells expanded from wild-type (wt) or CD40L-deficient T cells show comparable proportions of CD8 vs CD4 T cells by flow cytometry. F. Activation status of T cells expanded from wild-type (wt) or CD40L-deficient T cells. G. IFN-γ ELISPOT analysis of wild-type (wt) or CD40L-deficient T cells primed against ID8-Defb29/Vegf-a tumor antigens. H. Mice bearing i.p. ID8-Defb29/Vegf-a tumors were depleted of immunosuppressive dendritic cells with an anti-CD11c immunotoxin (IT) then treated with expanded T cells from wild-type or CD40L-deficient mice on days 7 and 14 of tumor progression. After 28 days of tumor progression host CD3⁺ were sorted from treated mice and transferred into naïve mice that received s.c. ID8-Defb29/Vegf-a tumors 2 days later. Mice receiving host CD3⁺ T cells from mice treated with wild-type T cells or CD40L^−/− T cells were challenged with s.c. ID8-Defb29/Vegf-a tumors. Tumor growth was monitored over time. (n=3/4 mice per group in three independent experiments) (* means P<0.05; **means P<0.01).

Additionally, host-derived (endogenous) T cells obtained from regulatory DC depleted, CD40L-deficient T cell treated mice, failed to protect mice challenged with flank ID8-Defb29/Vegf-a tumors (Figures 5H and Supplemental Figure 2). Taken together, these data suggest that transferred T cells recruit APCs to the tumor site through the secretion of CCL5 (predominantly secreted by the CD4⁺ T cell subset). In the presence of tumor antigen (presumably released by transferred CD8⁺ T cells (20)), adoptively transferred CD4⁺ T cells signal through CD40 on the newly recruited host DCs to trigger their maturation. The mature DCs then activate tumor-specific host-derived (endogenous) T cells to control tumor progression.

Tumor-primed T cells activate tumor-associated dendritic cells

To confirm that the activating function of adoptively transferred T cells is indeed mediated by CD40 stimulation on host DCs recruited to tumor locations, we briefly primed T cells against tumor antigen and cultured them with DCs sorted from wild-type or CD40-deficient mice bearing ovarian tumors. As expected, wild-type tumor-derived dendritic cells significantly upregulated expression of the costimulatory molecules CD70, CD86 and CD80, as well as MHC II, after culture with wild-type T cells, compared to unstimulated dendritic cells (Figure 6). In addition, these dendritic cells also secreted significantly higher levels of IL-12 than their unstimulated counterparts (Figure 7A). In contrast, DCs sorted from the peritoneal cavity of tumor-bearing CD40-deficient mice cultured with wild-type T cells primed in an identical manner expressed significantly reduced levels of CD70, CD86, CD80 and MHC II, (Figure 6), and also secreted significantly less IL-12 than wild-type DCs (Figure 7A). Further supporting the requirement of interactions between CD40 (on host DCs) and CD40L (on adoptively transferred T cells) for the maturation of newly recruited tumor DCs, CD40L-deficient T cells briefly primed to tumor antigens were incapable of inducing the maturation and secretion of IL-12 by cultured tumor-derived dendritic cells to the extent observed by identically primed wild-type T cells (Figures 6 & 7A). Interestingly, differences in DC activation were not as strong when wild-type DCs were cultured with CD40L^−/− T cells as seen when DCs lacking CD40 were cultured with wild-type T cells, suggesting that while these ID8-Defb29/Vegf-a tumor-primed T cells act to mature dendritic cells through CD40 signaling, there probably exists other mechanisms through which they additionally act to mature DCs. Nevertheless, together, these data confirm that adoptively transferred (exogenous), briefly primed (CD4⁺) T cells induce the maturation of tumor-associated DCs of host (endogenous) origin, suggesting that these DCs are licensed in situ at tumor locations to induce the expansion and activation of host-derived (endogenous) tumor-specific T cells.

Figure 6. T cells expanded to tumor antigens induce maturation of tumor-associated DCs.

Expression profile of MHCII, CD80, CD70 and CD86 on wild-type or CD40^−/− dendritic cells obtained from tumor-bearing mice and subsequently cultured with tumor-primed wild-type or CD40L^−/− T cells as specified. (n=4 per group in three independent experiments). (* means P<0.05; **means P<0.01).

Figure 7. Wild-type T cells interact with CD40 on wild-type tumor associated DCs to induce their production of IL-12 and license them to prime antitumor CD8⁺ T cells.

T cells from wild-type or CD40L-deficient T cells were activated for 7 days against tumor antigen. Wild-type or CD40-deficient DCs sorted from mice bearing ID8-Defb29/Vegf-a tumors for 7 days were co-cultured for 24 hours with activated wild-type or CD40L-deficient T cells and the supernatants thus obtained were tested in ELISA for IL-12 production. A. IL-12 secretion by tumor-associated wild-type (wt) or CD40^−/− DCs that were cultured with tumor-primed wild-type (wt) or CD40L^−/− T cells. B. T cells from wild-type (wt) or CD40L-deficient T cells were activated for 7 days against tumor antigen. Wild-type (wt) or CD40-deficient DCs sorted from tumor-bearing mice were co-cultured for 24 hours with activated wild-type (wt) or CD40L-deficient T cells then admixed with tumor-infiltrating CD8⁺ T cells for 48hours in ELISPOT analyses. C. ELISPOT analyses of the number of CD8⁺ T cells secreting IFN-γ after culture with dendritic cells primed as indicated. C. ELISPOT analyses of the number of CD8⁺ T cells secreting Granzyme-B after culture with dendritic cells primed as indicated. (n=4 per group in four independent experiments). (* means P<0.05; **means P<0.01).

Naïve T cells briefly primed to tumor antigens transform tumor-derived immunosuppressive dendritic cells into effective tumor antigen-presenting cells through CD40-CD40L interactions

We next sought to confirm the functional, and thus, potential therapeutic relevance of our findings that expanded T cells license tumor-associated DCs to activate endogenous antitumor cytotoxic T lymphocytes (CTLs). For that purpose, we FACS-sorted CD11c⁺MHC-II⁺ DCs cells from the peritoneal cavity of untreated mice growing established intraperitoneal ovarian cancer (Experimental Design Supplemental Figure 3). As we reported, these cells are strongly immunosuppressive (26). When these tumor-derived DCs were incubated for 24 h. with “adoptive transfer-ready” wild-type T cells from healthy mice (20), they were able to induce the secretion of both cytolytic Granzyme-B and IFN-γ by high numbers of subsequently added CD3⁺CD8⁺ CTLs sorted from tumor (peritoneal) locations in ELISPOT analysis (Figures 7B–C). In contrast, when tumor-derived CTLs were added to cultures in which tumor-associated DCs had been incubated with CD40L-deficient T cells primed in an identical manner, the number of lymphocytes producing Granzyme-B and IFN-γ was significantly decreased (Figures 7B–C). These results indicate that adoptively transferred (CD4⁺) T cells transform tumor-associated immunosuppressive DCs into effective APCs, and, as no additional antigenic stimulus was added, the data confirm the notion that tumor-associated DCs spontaneously engulf tumor materials (26).

To reciprocally confirm that the capacity of licensed tumor DCs to activate (rather than inhibit, as they spontaneously do (20)) tumor-infiltrating CTLs was dependent on CD40 signaling, we challenged CD40-deficient mice with intraperitoneal ovarian tumors and first confirmed that their ascitic fluid contained the same numbers and proportions of microenvironmental cells as wild-type mice (data not shown). When tumor-associated (CD40-deficient) DCs sorted from these mice were incubated with either wild-type or CD40L-deficient T cells de novo primed against tumor antigens, the number of tumor-derived CTLs producing Granzyme-B or IFN-γ was again significantly decreased compared to wild-type DCs (Figures 7B–C). Together, these data indicate that adoptively transferred polyclonal T cells briefly primed de novo against tumor antigens can license host-derived tumor-associated DCs through CD40 to become mature DCs capable of activating host (endogenous) tumor-associated T cells to induce protection against subsequent challenge with the same tumor.

DISCUSSION

CD4⁺ T cells have been associated with a negative outcome in ovarian cancer patients (2) and the therapeutic potential of adoptively transferring tumor-reactive Th1 cells to ovarian cancer patients demands further investigation (12).

We previously reported that the transfer of a mixed population of polyclonal T cells briefly primed de novo against tumor antigens promotes the regression of established ovarian cancer, or significantly delays the progression of a more aggressive model of ovarian cancer, when coupled with the depletion of tumor-associated immunosuppressive CD11c⁺ DCs (20). In this current study, we demonstrate that CD4⁺ T cells are necessary for maximum effectiveness. We further elucidated the mechanisms of the contributions of such CD4⁺ T cells as being two-fold, initially through the recruitment of host immune cells to the tumor site via CCL5 secretion, and more importantly, by directly interacting with and licensing host dendritic cells to become capable of activating potent anti-tumor CD8⁺ T cells.

Most adoptive immunotherapy regimens focus on enhancing the effect of CD8⁺ T cells, partly due to their ability to eradicate highly prevalent MHC-I⁺ tumor cells. Yet, most methods currently employing the individual transfer of CD8⁺ T cells have not been met with much success (29–31). This may be due to the fact that while these CD8⁺ T cells may initially induce strong tumorlytic effects, they eventually succumb to the paralytic tumor microenvironment and become tolerized and thus, incapable of eliciting sustained anti-tumor effects. Developing strategies that awaken and permit host immune cells to overcome such debilitating effects constitute the future for adoptive cell therapies. With this intention, and given the fact that the earlier the activation phenotype of the transferred T cells the better they fare, we aimed to develop a clinically translatable technique of activating T cells. A potential source of non-paralyzed T cells for activation and infusion may be aphaeresis samples from ovarian cancer patients. We recapitulate this stage wherein there are few antigen-experienced T cells by utilizing splenic T cells from healthy unchallenged mice, which we previously demonstrated are as effective as those from tumor-bearing mice at inducing enhanced survival of tumor bearing mice (20).

With the understanding that there are several immunosuppressive networks at play in the context of ovarian cancer, we initially sought to establish the contribution of individual T cell subsets to adoptive cell therapy with the hypothesis that CD4⁺ T cells would prove to be required for maximal therapeutic benefits by providing help for CD8⁺ T cells. In fact CD4⁺ T cells were found to be the major producers of CCL5, which was required for the accumulation of CCR5⁺ activated dendritic cells and antitumor T cells of host origin at the tumor site. This is consistent with the results of Dobrzanski et al, that attribute the therapeutic benefits triggered by adoptively transferred Th1 cells to subsequently activated CCR5⁺ T cells of host origin (12). CCL5 secretion is also pertinent for the long-term antitumor protective effects, as activated host T cells mediated protection against secondary tumor challenge only when CCL5 was produced by the transferred T cell compartment.

Importantly, the protective effects mediated by endogenous T cells re-activated upon adoptive T cell therapy against a secondary tumor challenge were also abrogated by the elimination of CD40 signaling in the transferred T cells. Thus, adoptively transferred wild-type, but not CD40L^−/− (CD4⁺) T cells, were capable of licensing dendritic cells newly recruited to tumor locations. DCs licensed by transferred CD4⁺ T cells were then capable of activating tumor-reactive CD8⁺ T cells of host origin. Endogenous CD8⁺ T cells primed in this fashion had a greater capacity for secreting the tumorlytic factors IFN-γ and granzyme-B and also displayed increased capacity for long-term persistence and anti-tumor immunity. This translated to their capacity to induce sustained host immune responses such that their infusion into naive mice, was protective against a secondary challenge with s.c. tumors of the same strain. Since T cells deficient in CD40L were unable to induce such effects, it was apparent that this mechanism of DC licensing, completely independent of the direct cytotoxic activity recently found in non-epithelial tumor models (32, 33), was absolutely necessary for the prolonged CTL effector functions and later memory/protective responses.

These studies have revealed that adoptively transferred CD4⁺ T cells license DCs to prime antitumor CD8⁺ T cells. We have previously demonstrated that within 3 days of adoptive T cell transfer and CD11c⁺ regulatory DC depletion we observe increased maturation of returning DCs, which remain in this state up to 7 days after T cell transfer (20). Upon administration of the CD11c⁺ immunotoxin, ~70% of CD11c⁺ regulatory DCs are eliminated within the peritoneum but there is a rapid turnover such that within 48 hours there is an almost complete repopulation (25). Furthermore, transferred T cells do not persist for long periods after T cell transfer. This implies that the T cell licensing occurs within a short span of T cell transfer as confirmed by our in vitro data in which activated T cells induce DCs to upregulate activation markers after 24 hours in culture. Importantly, the transmission of immunity from transferred T cells to the host population through licensing of host DCs, though rapid, is long-lasting, supported by the ability of host T cells to protect against secondary tumor challenge.

Traditionally, licensing of dendritic cells is considered to occur through the interactions with CD4⁺ T cells in the lymph node and to be followed with the subsequent priming of circulating CD8⁺ T cells there. However, reports present evidence for the in situ licensing of dendritic cells and their further in situ priming of resident CD8⁺ T cells (34, 35). As such, since CD4⁺ T cells, CD40 signaling and CCL5 were all required for successful ACT, we propose a mechanism whereby, CD4⁺ T cells may, upon administration in the peritoneum, secrete CCL5 initiating the recruitment of immune cells to the tumor site. CD8⁺ T cells administered concurrently, may begin killing tumor cells, while the CD4⁺ T cells primed against tumor antigens prior to transfer and from the repeated exposure to antigen produced by the CD8⁺ T cell mediated tumor cell death, act upon endogenous tumor-resident dendritic cells to license them to further prime host CD8⁺ T cells also present or recruited to the tumor site. These newly awakened host CD8⁺ T cells can now additionally function to fight against the tumor.

Collectively, the data presented here show that CD4⁺ T cells contribute positively and significantly to the beneficial effects of adoptive T cell therapy against ovarian cancer, and define the phenotypic attributes that can maximize their effectiveness. These results emphasize the need for consistently studying methods of achieving optimal effects in adoptive cell therapies and advance the field’s understanding of the mechanisms at play in the context of ovarian cancer.

Abbreviations

ACT

Adoptive Cell Therapy

APC

Antigen Presenting Cell

DC

Dendritic Cell

IT

Immunotoxin (Anti-CD11c Immunotoxin)

IP

Intraperitoneal