Clec9A⁺ Dendritic Cells Are Not Essential for Antitumor CD8⁺ T Cell Responses Induced by Poly I:C Immunotherapy

Immunotherapy is used to stimulate antitumor immune responses that are particularly weak or lacking and overcome tumor-associated immune suppression. Dendritic cells (DCs) are useful targets for activation through immunotherapy, because improved DC function can lead to increased CD4⁺ and CD8⁺ T cell priming, more effective antitumor immune responses (1), and delayed tumor growth (2).

DCs are highly heterogeneous, and lymphoid and nonlymphoid tissues harbor several DC subsets that differ in surface phenotype and Ag-presenting properties. DCs in tumors are similarly heterogeneous. Surprisingly, despite this heterogeneity, recent studies have shown that most DC subsets in tumors do not appear to participate in the priming of antitumor immune responses. Instead, a discrete population of CD103⁺ DCs is necessary for the transport of tumor Ag to the draining lymph node (dLN) (3) and the priming of tumor-specific CD8⁺ T cell responses in the steady state (4, 5). Importantly, this key role of CD103⁺ DCs also extends to human populations, with the number of CD103⁺ DCs within tumors correlating with better prognosis in certain cancers (5).

CD103⁺ DCs, together with the lymphoid tissue–resident CD8α⁺ DCs, are characterized by the ability to efficiently cross-present Ags, and especially cell-associated Ags, in the context of MHC class I molecules (6–8). CD103⁺ and CD8α⁺ DCs, collectively referred to as “DC1s,” also share several other characteristics: they require BATF3 for their optimal development in vivo (4, 9) and express the surface C-type lectin “DC NK lectin group receptor-1,” or Clec9A, which binds actin and promotes the uptake of dying cells (10, 11). DC1s also express XCR1 (6, 12), the receptor for the chemokine XCL1 secreted by recently activated NK cells and CD8⁺ T cells, which promotes correct DC localization within the lymph node (LN) and efficient DC interaction with NK and CD8⁺ T cells (13).

Although current evidence suggests that DC subsets other than DC1s do not play a role in cross-presentation of tumor Ags in the steady state, the role of these cells during immunotherapy is less well understood. In previous work, CD8α⁻ splenic DCs were shown to cross-present Ag if activated through Fcγ receptors (14). In addition, all DCs are highly responsive to stimulation via TLR ligands and pathogen recognition receptors, and CD11b⁺ DCs appropriately activated via TLR7 can cross-present long peptides for tumor immunotherapy (15). During inflammation, monocyte-derived DCs (moDCs) are also recruited to the tumor (16) and the tumor dLN (1, 17). moDCs can cross-present Ag (18); although their ability to prime CD8⁺ T cells in vivo is unclear, they can produce inflammatory cytokines and are required for successful tumor immunotherapy (16, 19).

In this article, we investigate the requirement for DC1s in local tumor immunotherapy with poly I:C or monosodium urate crystals and Mycobacterium smegmatis (MSU+M.smeg), which are thought to activate tumor-associated DCs. These treatments were found to be effective as single agents (17) or in combination with peptide vaccines (20). Using Clec9A–diphtheria toxin receptor (DTR) and BATF3-knockout (KO) mice, we show that DC1s are not required for adaptive immune responses and delayed tumor growth after poly I:C immunotherapy. DC1s and DC2s (comprising the BATF3-independent CD11b⁺ and CD11b^low DC subsets) could take up soluble and cell-associated Ag in vivo and induce proliferation of Ag-specific CD8⁺ T cells ex vivo. Thus, our results suggest that appropriate treatments can expand the range of DCs able to cross-present tumor Ag for effective immunotherapy.

All mice were bred at the Malaghan Institute of Medical Research animal facility and were matched for age and sex within experiments. C57BL/6J, CD11c-DTR, and BALB/cByJ mice were originally from The Jackson Laboratory (Bar Harbor, ME). CD45.1⁺ OT-I mice were generated by breeding OT-I mice to B6SJL-Ptprc^a mice. BATF3-KO and Clec9A-DTR mice (4, 21), on a C57BL/6J or BALB/cByJ background, respectively, were bred from pairs originally provided by Dr. K. Murphy (Washington University, St. Louis, MO) and Dr. Christiane Ruedl (Nanyang Technological University, Singapore).

CD11c-DTR bone marrow (BM) chimeras were generated as described (17) by irradiating (2 × 550 rad) C57BL/6J hosts, followed by i.v. transfer of 1 × 10⁷ CD11c-DTR BM cells. Chimeras were rested for ≥8 wk before experimental use. All experimental protocols were approved by the Victoria University of Wellington Animal Ethics Committee.

The 4T1 murine mammary carcinoma cell line was purchased from the American Type Culture Collection (ATCC; Manassas, VA; ATCC number CRL-2539) and maintained in ATCC-formulated RPMI 1640 supplemented with 10% FCS. Extended in vitro passaging was avoided. Before tumor challenge, cells were washed three times in PBS, and mice were injected with 1 × 10⁴–3 × 10⁴ tumor cells orthotopically into the mammary fat pad. The B16-F1 (ATCC) and B16.OVA (22) murine melanoma lines were maintained in complete IMDM, as described (23). Before tumor challenge, cells were washed three times in PBS, and mice were injected s.c. with 1 × 10⁵ tumor cells into the flank. Tumor size and survival were calculated as described (23).

Mice were treated with 50 μg of poly I:C (low m.w., 0.2–1 kb; InvivoGen) or with 4 × 10⁶ CFU M. smegatis (mc²155) combined with 250 μg of monosodium urate crystals containing <0.01 endotoxin units/10 mg, prepared as described (24). Treatments were injected in a total volume of 100 μl of PBS into the area directly adjacent to the tumors. PBS (Invitrogen) was used as a vehicle control. Treatments involved four injections, began when the tumor was palpable (∼days 7–9), and were repeated every second day.

CD8⁺ T cells were depleted by i.p. injection of 200 μg of purified 2.43 Ab (Bio X Cell), administered on days −1 and 0 with respect to tumor inoculation. Depletion was assessed 7 d later in the blood. For Clec9A⁺ DC depletion, 20 ng of diphtheria toxin (DT) from Corynebacterium diphtheriae (Sigma) per gram of body weight was delivered by i.p. injection every 3–4 d after the tumor was palpable, for a maximum of five treatments. PBS (Invitrogen) was used as a vehicle control. For CD11c-DTR BM chimeras, DT was given (15 ng/g of body weight) 18 h before each immunotherapy treatment.

LNs, spleens, or tumors were digested using DNase I and Liberase TL (Roche). Single-cell suspensions were resuspended in FACS buffer (PBS with 10 mM EDTA, 2% FBS, and 0.01% NaN₃), and Fc receptors were blocked with anti-mouse CD16/32 (2.4G2) before staining with Abs specific for the following markers: CD8 (2.43; prepared in-house; CD11b (M1/70), CD8 (53-6.7), CD103 (M290), Ly6G (1A8; all from BD Biosciences); MHC class II (M5/144.15.2), CD11c (N418), Ly6C (HK1.4), XCR1 (ZET; all from BioLegend); Thy1.1 (30-F11; eBioscience); and Ly6B (7/4; AbD Serotec). Dead cells were excluded on the basis of positive DAPI staining. Acquisition was performed on a BD LSR II SORP or a BD LSRFortessa SORP (both from Becton Dickinson), and data were analyzed using FlowJo version 9.9 (TreeStar).

C57BL/6J mice were injected s.c. with 50 μg of poly I:C or PBS on day 1 and with 50 μg of Alexa Fluor 647–labeled OVA (AF-OVA) on day 2, with or without poly I:C. dLNs were collected on day 3 to analyze AF-OVA uptake by flow cytometry. Alternatively, C57BL/6J mice were injected s.c. with 5 × 10⁶ irradiated CellTracker Orange–labeled B16.OVA (CTO-B16.OVA) cells, with or without poly I:C. dLNs were collected 1 or 2 d later and analyzed for CTO-B16.OVA uptake by flow cytometry. PBS was used as a control.

C57BL/6J or BATF3-KO mice were primed with 25 μg of poly I:C on day 0. The following day, mice received 250 μg of OVA with 25 μg of poly I:C. dLNs were harvested on day 2, and preparations were pre-enriched for DCs using a Dynabeads Mouse DC Enrichment Kit. The positive fraction was stained and sorted using a BD Influx cell sorter. CD8⁺ T cells were purified from CD45.1⁺ OT-I mice using a Dynabeads Untouched Mouse CD8 Cells Kit. Cells were stained with 100 nM CFSE. OT-I cells were plated at 1 × 10⁵ cells per well with titrated DC subsets (5,000–20,000). Cells were harvested 2 d after plating and assessed for OT-I proliferation by flow cytometry analysis.

Statistical analyses were performed with GraphPad Prism 5. Tumor growth curves were analyzed by two-way ANOVA with the Bonferroni correction. Survival curves were analyzed by the Mantel–Cox test. Multiple groups were compared using the Kruskal–Wallis test with the Dunn posttest. Differences of p < 0.05 were deemed statistically significant.

Immunotherapy with poly I:C or MSU+M.smeg is successful at delaying primary tumor growth in a model of orthotopic murine 4T1 mammary carcinoma in BALB/cByJ mice (17). To investigate the contribution of CD8⁺ T cells to antitumor activity in vivo, we used Ab treatment to deplete CD8⁺ T cells prior to tumor engraftment and throughout tumor growth. In the absence of CD8⁺ T cells, immunotherapy with poly I:C or MSU+M.smeg was no longer able to prolong survival (Fig. 1) and was unable to delay 4T1 tumor growth (Supplemental Fig. 1). Therefore, CD8⁺ T cells are required for the efficacy of poly I:C and MSU+M.smeg immunotherapy.

We used Clec9A-DTR mice that can be selectively depleted of cross-presenting DC1 populations by DT treatment (21) to assess the requirement for these DCs during immunotherapy with poly I:C and MSU+M.smeg. Mice were treated with DT every 3–4 d throughout immunotherapy, which successfully depleted the large majority of CD8α⁺ and CD103⁺ DCs from dLN and spleen (Fig. 2A, Supplemental Fig. 2A). DC1 depletion removed the ability of MSU+M.smeg immunotherapy to delay the growth of 4T1 orthotopic tumors, suggesting that this DC population is necessary for antitumor responses induced by this treatment (Fig. 2B). In contrast, DC1 depletion had no detectable effect on tumor growth after poly I:C immunotherapy (Fig. 2B), suggesting that this treatment does not require specialized cross-presenting DCs.

To investigate whether the nonessential role of DC1s during immunotherapy with poly I:C could extend to other mouse strains and tumor models, we used BATF3-KO mice, which lack migratory CD103⁺ DCs but retain some CD8α⁺ LN-resident DCs (Supplemental Fig. 2B) and are highly susceptible to tumor growth (4). Similar to the observation in Clec9A-DTR mice, MSU+M.smeg treatment could no longer delay primary tumor growth in BATF3-KO mice. In contrast, poly I:C immunotherapy remained fully effective (Fig. 2C).

To assess whether any DC subsets were required for successful poly I:C immunotherapy, CD11c-DTR BM chimeras were challenged with B16 tumors and treated with DT starting 18 h before each immunotherapy treatment to deplete all CD11c⁺ cells (Supplemental Fig. 3). Treatment with DT completely abrogated the impact of poly I:C immunotherapy on primary tumor size and led to decreased survival (Fig. 2D). This requirement for DCs is similar to what was observed for MSU+M.smeg immunotherapy (19).

Therefore, CD8⁺ T cells and CD11c⁺ cells are required for the efficacy of tumor immunotherapy with poly I:C, whereas the cross-presenting DC1s are dispensable. Taken together, these results suggest that a DC population other than DC1s is involved in the presentation of tumor Ag in poly I:C–treated mice.

To explore the differential impact of MSU+M.smeg versus poly I:C immunotherapy on tumor-associated APC populations, we examined serum cytokines after local immunotherapy of orthotopic 4T1 tumors. We found that, 2 h after treatment, serum CCL2 (also known as MCP1) was significantly elevated in poly I:C–treated mice compared with controls (Fig. 3A), whereas CCL2 levels in MSU+M.smeg-treated mice remained low. The high CCL2 levels in poly I:C–treated mice were reflected in more CD11b⁺Ly6B⁺Ly6C⁺ cells (Fig. 3B) (moDCs in this article) in the tumor dLN, whereas the number of Ly6C⁺ cells in the tumor was not significantly increased (Fig. 3C). In contrast, in MSU+M.smeg-treated mice, the number of moDCs was markedly elevated in the tumor and only modestly increased in the dLN (Fig. 3C). When Clec9A-DTR mice bearing 4T1 tumors were treated with DT, DC1 depletion did not impair the poly I:C–induced accumulation of moDCs in the dLN (Fig. 3D).

Thus, treatment with poly I:C induces the accumulation of a population of CD11b⁺Ly6B⁺Ly6C⁺ moDCs in the dLN, and this accumulation does not require the presence of DC1s.

Poly I:C treatment, at doses higher than those used in this study, has been reported to reduce the accumulation of myeloid-derived suppressor cells (MDSCs) in spleens and tumors of mice bearing 4T1 mammary carcinomas (25), thus facilitating antitumor immune responses. Therefore, we examined the accumulation of CD11b⁺Ly6C⁺ and CD11b⁺Ly6G⁺ MDSCs in mice bearing 4T1 tumors, using a previously described strategy (26) but gating out CD11c⁺ cells to exclude moDCs (Fig. 4A, Supplemental Fig. 4). Compared with naive mice, tumor-bearing mice had greatly increased numbers of splenic Ly6C⁺ and Ly6G⁺ MDSCs when they were sacrificed on day 21 after tumor injection (Fig. 4B). Treatment with poly I:C or MSU+M.smeg did not significantly affect MDSC numbers compared with PBS-treated controls, although a modest decrease was detectable in poly I:C–treated mice 7 d following the end of immunotherapy (Fig. 4C). This reduction might reflect the decreased tumor size in poly I:C–treated mice at the time of sacrifice (Fig. 4C).

The number of Ly6C⁺ and Ly6G⁺ MDSCs was also assessed at an earlier time, 18 h after the fourth poly I:C or MSU+M.smeg treatment, when the impact of immunotherapy on MDSCs might be more marked and less affected by variation in tumor size. MDSC numbers in spleens and tumors were similar in all groups, regardless of immunotherapy treatment (Fig. 4D). Thus, at the doses used in this study, poly I:C and MSU+M.smeg treatments do not appear to affect the numbers of Ly6C⁺ and Ly6G⁺ MDSCs in mice bearing 4T1 tumors.

To compare the capacity of different DC subsets to take up Ag in LNs, we used AF-OVA. C57BL/6J mice were inoculated s.c. with AF-OVA alone or AF-OVA + poly I:C, and dLNs were harvested the following day to examine AF-OVA uptake. Treatment with poly I:C increased the total numbers of AF647⁺ DCs in the dLN, as well as the total population counts (Fig. 5A). In mice treated with AF-OVA only, the AF647 label was detected in each of the dLN DC subsets in comparable proportions (15–30% of each subset). In mice treated with poly I:C, AF-OVA uptake by each DC subset was increased compared with mice not treated with poly I:C. The highest number of AF647⁺ cells was in the CD11b-expressing DC2 subset.

Uptake of cell-associated Ag was assessed by injecting mice with irradiated B16.OVA tumor cells labeled with CTO. CTO-B16.OVA cells were injected, with or without poly I:C, 2 or 1 d prior to dLN harvest. All DC subsets appeared capable of some CTO-B16.OVA uptake. Again, coadministration of poly I:C induced a substantial increase in LN cellularity, which was especially marked on day 2. In addition, poly I:C treatment had an impact on the proportions of DCs taking up CTO-B16.OVA. CD11b⁺ DCs represented almost 70% of the DCs taking up CTO-B16.OVA in untreated mice but only ∼50% in mice treated with CTO-B16.OVA + poly I:C. In contrast, the proportion of moDCs was increased in poly I:C–treated mice at both time points, whereas the proportion of CD8α⁺ DCs was increased on day 1 (Fig. 5B). When the number of CTO⁺ DCs in each subset and the number of total DCs were compared, a small percentage of DCs in each subset were capable of CTO-B16.OVA uptake. Taken together, these results show that all DC subsets are capable of soluble and cell-associated Ag uptake. Interestingly, in each case, the largest number of DCs taking up Ag expressed the DC2 marker CD11b.

We assessed whether different DC subsets can cross-present OVA for Ag-specific CD8⁺ T cell proliferation in vitro. DCs were flow sorted from the dLNs of C57BL/6J mice treated with OVA and poly I:C in vivo and tested for the ability to induce division of OVA-specific OT-I T cells in vitro without further Ag supplementation. Three LN DC populations were compared: XCR1⁺ DCs, which include the cross-presenting CD8α⁺ and CD103⁺ DC subsets, as well as a functional CD103⁻XCR1⁺ migratory DC subset (12), XCR1⁻ DCs, and Ly6C⁺ moDCs (Fig. 6A). XCR1⁺ DCs supported the highest total cell recovery (including T cells and DCs) at the end of culture. Coculture with XCR1⁻ DCs also resulted in good cell recovery; however, 2-fold more XCR1⁻ DCs were required to induce a response comparable to that of XCR1⁺ DCs (Fig. 6B). moDCs were the least effective, with low cell recovery. Flow cytometry analysis of the cells recovered from culture revealed that a high percentage of OT-I T cells cocultured with XCR1⁺ DCs expressed high levels of CD44 and CD69, which is consistent with their recognition of cognate Ag and consequent activation (Fig. 6C). In addition, up to 30% of these cells had undergone cell division (Fig. 6D). Coculture with XCR1⁻ DCs also induced OT-I T cell activation and division, but these were lower compared with XCR1⁺ cultures. Again, the lowest response was observed in cocultures of OT-I T cells and moDCs.

These data suggest that XCR1⁺ and XCR1⁻ DCs can cross-present Ag ex vivo to induce the division of Ag-specific OT-I T cells. Although DC1s were the most effective stimulators on a per-cell basis, XCR1⁻ DC2s were also effective, and their numbers in dLNs were three to four times higher than DC1s (Fig. 5).

In this article, we investigate the contribution of DC1s to antitumor immune responses during immunotherapy. We report that DC1s are required for delayed tumor growth after local immunotherapy with MSU+M.smeg. In contrast, poly I:C immunotherapy did not require DC1s to be effective. Treatment with poly I:C improved the ability of multiple DC subsets to take up soluble and cell-associated Ag in vivo; these DCs could present Ag to specific CD8⁺ T cells in vitro. These results suggest that, when exposed to the appropriate stimuli, multiple DC populations are capable of cross-presenting Ag and priming antitumor immune responses.

Data in the literature indicate that DC1s are essential to antitumor immune responses: BATF3-KO mice that are defective in the CD103⁺ DC population are unable to control tumor growth (4), and the expansion of CD103⁺ DCs by FLT3 treatment can enhance antitumor immune responses (3). CD103⁺ DCs are found, albeit in very low numbers, in tumors and can migrate to LNs to prime antitumor immune responses (3, 5). These properties are consistent with the reported ability of CD103⁺ DCs to take up dying cells via receptors such as Clec9A (10, 11), and their role in antiviral immune responses. Although clearly essential, CD103⁺ DCs in tumors are also rare; any treatment to improve the cross-presenting ability of other more abundant DC subsets may have substantial benefits for immunotherapy. In this article, we show that poly I:C treatment appears to have some of these properties and was able to promote antitumor immune responses in the absence of cross-presenting DC1s. This was observed in BATF3-KO mice, which lack CD103⁺ migratory DCs but maintain a normal CD8α⁺ LN-resident DC population, as well as in the combined absence of all DC1s, as in DT-treated Clec9A-DTR mice (21); this ruled out the possibility that “leakiness” of BATF3 inactivation in inflammatory conditions might allow the development of some CD103⁺ DCs (27) or that LN-resident CD8α⁺ DCs might be contributing to cross-presentation of tumor Ags when CD103⁺ DCs are absent.

Although cross-presenting DC1s are superior at Ag presentation in multiple tumor models (3–5), other DC subsets have been implicated in cross-presentation. Studies have reported that CD8α⁺, as well as CD8α⁻, DCs could cross-prime cell-associated Ag (28). To identify the DC subset that might be responsible for cross-presenting tumor Ags in mice depleted of DC1s and treated with poly I:C, we examined the uptake of soluble or cell-associated Ags by DCs in vivo. As reported by other investigators (7), we observed that soluble and cell-associated Ags could be taken up by multiple DC subsets, although there were differences between the uptake of soluble protein and cellular material. Remarkably, in both cases, the majority of Ag was taken up by a population of CD11b⁺ DCs, which constituted ∼40% of the Ag⁺ DCs in LNs and was 4-fold more abundant than Ag⁺ CD103⁺ or CD8α⁺ DCs. In vitro proliferation assays using flow-sorted DCs from immunized mice showed that this population of DCs, which does not express the XCR1 surface marker, was able to induce T cell division. Although XCR1⁻ DCs were less efficient stimulators than XCR1⁺ DCs on a per-cell basis, this relative deficiency might well be compensated for by their higher numbers in the LN. Interestingly, moDCs could only induce minimal Ag-specific T cell proliferation. Although these cells have been shown to cross-present soluble protein in vivo in some models (29) and were necessary for antitumor immune responses after chemotherapy (16) or immunotherapy with MSU+M.smeg (19), their role in the priming of CD8⁺ T cell responses remains unclear.

The mechanism by which poly I:C treatment enables the generation of antitumor immune responses in the absence of DC1 subsets is yet to be defined. MDSCs, which are abundant in the spleens and tumors of mice carrying 4T1 tumors, were not decreased by poly I:C immunotherapy at the dose used in our experiments. Although DC1s were not required, we show in this article that CD11c⁺ DCs and CD8⁺ T cells were necessary for the antitumor effect of poly I:C. Indeed, poly I:C acts at many levels in the immune response. Through the induction of IFN-I and IL-12, it supports NK cell activation (30), which is necessary for the optimal development of Th1 and CD8⁺ T cell responses, including the antitumor immune responses in this study (17). Poly I:C and IFN-I also promote the proliferation of CD8⁺ T cells in vivo (31, 32) and support CD8⁺ T cell memory formation (33). Importantly, poly I:C treatment also supports cross-presentation by DCs, through the induction of IFN-I (34) and through the direct stimulation of TLR3-expressing DC subsets (35); however, it is important to note that endosomal TLR3 is not the only receptor for poly I:C. The cytosolic receptors melanoma differentiation-associated protein 5 (MDA5) and retinoic acid–inducible gene I protein (RIG-I), via the adaptor protein mitochondrial antiviral signaling, can initiate signaling pathways that are similar to those of TLR3, including activation of IRF3 and NF-κB, with consequent IFN-I secretion and production of proinflammatory cytokines (36). Interestingly, unlike TLR3, MDA5 and RIG-I are broadly expressed on DC subsets, especially the CD4⁺ LN-resident and CD11b⁺ migratory DC2 subsets (37). Thus, poly I:C may promote antitumor immune responses by activating TLR3⁻ DC subsets via the cytoplasmic receptors RIG-I and/or MDA5.

These results suggest that, in the absence of DC1s, poly I:C treatment enables other DC populations to cross-present Ag and is sufficient for inducing effective antitumor immune responses. Thus, appropriate immunotherapies may expand the range of DC subsets that are able to prime antitumor immune responses to improve the impact of immunotherapies.

We thank Dr. Ken Murphy for the BATF3-KO mice and all colleagues at the Malaghan Institute of Medical Research for advice and discussion. The expert animal husbandry of the staff of the Malaghan Institute animal facility and the flow cytometry support of the Hugh Green Cytometry Core facility staff are also gratefully acknowledged.

The authors have no financial conflicts of interest.