Abstract
The success of adoptive CTL therapy for cancer depends on interactions between tumor-infiltrating CTLs and cancer cells as well as other cells and molecules in the tumor microenvironment. Tumor dendritic cells (DCs) comprise several subsets: CD103+CD11b− DC1 and CD11b+CD64− DC2, which originate from circulating precursors of conventional DCs, and CD11b+CD64+ DC3, which arise from monocytes. It remains controversial which of these subset(s) promotes intratumor CTL proliferation, expansion, and function. To address this issue, we used the Zbtb46-DTR–transgenic mouse model to selectively deplete DC1 and DC2 from tumors and lymphoid tissues. Wild-type and Zbtb46-DTR bone marrow chimeras were inoculated with B16 melanoma cells that express OVA and were treated with OT-1 CTLs. We found that depletion of DCs derived from precursors of conventional DCs in Zbtb46-DTR bone marrow chimeras abolished CTL proliferation and expansion in tumor-draining lymph nodes. By contrast, intratumor CTL accumulation, proliferation, and IFN-γ expression were unaffected by their absence. We found that adoptive cell therapy increases the frequency of monocyte-derived tumor DC3, which possess the capacity to cross-present tumor Ags and induce CTL proliferation. Our findings support the specialized roles of different DC subsets in the regulation of antitumor CTL responses.
This article is featured in In This Issue, p.1107
Introduction
Recent progress in the ability to identify cancer neoantigens, generate tumor-specific CTLs, and engineer potent immune checkpoint inhibitors has revitalized the field of cancer immunotherapy (1–3). Response rates have increased significantly, with some patients experiencing durable cures. Despite these advances, however, clinical responses remain heterogeneous and unpredictable (4). Improving outcomes in more patients requires better understanding of the mechanisms that influence CTL efficacy.
The tumor microenvironment (TME) contains a variety of accessory cells, including dendritic cells (DCs), monocytes, macrophages, granulocytes, myeloid-derived suppressor cells (MDSC), and stromal cells that can positively or negatively influence CTL migration, proliferation, survival, cytokine secretion, and tumor-killing capacity (5). DCs have emerged as a critical population because of their ability to prime and regulate CTLs in lymphoid tissues and tumors. Three major subsets of CD11c+ MHC class II (MHC II)+ DCs have been identified in the TME based on the expression pattern of several cell surface markers: DC1 (CD103+CD11b−), DC2 (CD103−CD11b+CD64−), and monocyte-derived DCs (CD103−CD11b+CD64+), which we have termed DC3 (6, 7). DC1 and DC2 arise from circulating precursors of conventional DCs (pre-cDCs) (8), which selectively express the transcription factor Zbtb46 (9, 10). Tumor DC1 share a common ontogenetic relationship with lymphoid tissue CD8+ DCs, with both requiring transcription factors IRF8 and BATF3 for terminal differentiation, whereas DC2 depend on IRF4 (11). DC3 arise predominately from circulating monocytes and resemble inflammatory DCs in inflamed tissues (12–14).
Recent reports highlight the essential role of migrating, pre-cDC–derived tumor DCs, especially DC1, in cross-priming naive, tumor Ag–specific T cells in tumor-draining lymph nodes (TDLNs) (15, 16). Intratumor DCs also promote the expansion and function of tumor-infiltrating CTLs (17, 18); however, controversy persists over which DC subpopulation is involved. Several reports have emphasized the dominance of DC1 based on their robust Ag cross-presenting activity (as compared with DC2 and other myeloid populations) and their apparent importance in inhibiting tumor growth in mice treated with adoptive CTL therapy (18–20). By contrast, Ma et al. (21) reported that treating mice with anti-CD11b Abs eliminated the immune-mediated benefits of anthracycline chemotherapy, whereas loss of DC1 in tumors implanted into BATF3−/− mice did not attenuate this effect of chemotherapy. Although their studies supported the importance of monocyte-derived DCs, anti-CD11b Abs deplete both DC2 and DC3. We reported that decreasing the frequency of CD11c+ DCs in the CD11c-Cre mouse model reduced intratumor CTL proliferation (17); however, this experimental approach depletes all DCs. Thus, the relevance of pre-cDC–derived versus monocyte-derived tumor DCs to intratumor CTL responses in vivo has yet to be established. Further delineation of their roles will help guide strategies to enhance immunotherapy.
In this report, we investigate the effect of DC1 and DC2 deficiency in an adoptive CTL immunotherapy model of melanoma. Contrary to a prevalent view, our findings indicate that the absence of pre-cDC–derived DCs in tumors does not impact significantly on intratumor CTL frequency and function. Immunotherapy led to the expansion of tumor DC3 that cross-present tumor-derived Ags to Ag-specific CTLs. Our findings support the specialized roles of different DC subsets in the regulation of antitumor CTL responses.
Materials and Methods
Mice
Female C57BL/6, Zbtb46-DTR, and OT-1 mice and C57BL/6.SJL (CD45.1) congenic mice were purchased from The Jackson Laboratory (Bar Harbor, ME) or Taconic Farms and bred in our animal facility. CD8+ TCR-transgenic (P14) mice specific for lymphocytic choriomeningitis virus gp33–41 were kindly provided by P. Ohashi. To generate Zbtb46-DTR chimeras, female C57BL/6 wild type (WT) mice were irradiated twice (5.5 Gy 1 h apart) and reconstituted with 1 × 106 Zbtb46-DTR or C57BL/6 WT (control) female bone marrow cells. Mice were maintained in pathogen-free conditions in accordance with institutional guidelines and used at 2–3 mo of age. The Animal Research Committee of University Health Network reviewed and approved the studies.
Tumor models
B16-F10 melanoma (B16) was purchased from American Type Culture Collection. B16-OVA was kindly provided by R.W. Dutton at the Trudeau Institute (22). To establish tumors, 0.5–1 × 106 B16 or B16-OVA tumor cells in 100 μl PBS were injected s.c. into the flank of Zbtb46-DTR→WT and control WT→WT bone marrow chimeras. Tumor dimensions were measured with calipers, and tumor size was calculated using the formula ⅔ length × width.
Cell isolation
Tumors, spleens, and TDLNs from B16 tumor-bearing animals were minced, digested with collagenase and DNase I for 0.5 h at 37°C, and incubated in PBS containing 2 mM EDTA and 5% FCS for 10 min at room temperature. Mononuclear cells were isolated by Lympholyte-M (CEDARLANE) or Nycodenz (Axis-Shield Diagnostics) density gradient centrifugation to remove dead cells and further enriched for CD11c+ cells by positive selection using MACS (Miltenyi Biotec) and CD11c+ immunomagnetic beads. Cells were stained with anti–I-Ab or anti–I-A/I-E (allophycocyanin–eFluor 780), anti-CD11c (FITC or allophycocyanin), and anti-lineage markers (anti-CD3, anti-CD19, anti-B220, anti–Gr-1, and anti-CD49b). Desired populations were sorted on a BD FACSAria II using FACSDiva acquisition and analysis software (BD Biosciences, San Jose, CA) or the Sony SH800 cell sorter (Sony Biotechnology, San Jose, CA). The purity of the cell populations used was routinely >95% based on reanalyzed samples. Isolated cells were cultured in 96-well plates in 200 μl RPMI 1640 supplemented with 10% FBS, 50 μM 2-ME, 1 mM sodium pyruvate, 10 mM nonessential amino acids, 50 U/ml penicillin, and 50 μg/ml streptomycin (complete medium).
Flow cytometry
Cell suspensions were preincubated with anti-CD16/32 to block Fc receptors, then washed and incubated with the indicated mAb conjugates for 30 min at 4°C in a final volume of 100 ml PBS containing 0.5% BSA and 2 mM EDTA. In all experiments, appropriate control isotype-matched mAbs were included to determine the level of background staining. For intracellular cytokine detection, surface Ab–labeled cells were fixed, permeabilized, and stained with anticytokine Abs according to the instructions from the BD Cytofix/Cytoperm kit (BD Biosciences). Anti-CD11c (clone HL3), I-Ab (KH84, 25-9-17), I-A/I-E (MF/114.15.2), CD3 (17A2), CD8 and CD19 (1D3), CD49b (pan-NK, DX5), Gr-1 (RB6-8C5), CD11b (M1/70), B220 (RA3-6B2), CD45 (30-F11), CD45.1 (A20), CD40 (3/23), CD80 (16-10A1), Ly-6C (AL21), IFN-γ (XMG1.2), CD69 (H1.2F3), CD64 (X5405/7.1), IL-12/IL-23p40 (C17.8), TNF (MP6-XT22), inducible NO synthase (iNOS; clone 6), and Thy1.1 (OX-7) were purchased from BD Pharmingen, BioLegend, and eBioscience. These Abs were unlabeled or conjugated to FITC, PE, allophycocyanin, PE-Cy7, allophycocyanin–Cy7, V450, PerCP-Cy5.5, PE-CF594, AmCyan, or biotin as indicated. Biotinylated Abs were revealed with FITC, PE, allophycocyanin, Texas Red, Cy5, or Cy7.
Generation of effector OT-1 and P14 effector T cells
OT-1 C57BL/6.SJL (CD45.1) transgenic mice were injected i.p. and s.c. with 1 × 109 viral particles of the adenovirus vector encoding OVA. Effector CD8+ T cells were isolated 3 d later from spleens and lymph nodes by negative selection using a mixture of biotin-conjugated CD4, CD19, CD49b, and CD11b Abs and streptavidin magnetic beads (Miltenyi Biotec). P14 mice (Thy1.1+) received one i.v. injection of 5 μg of gp33 in HBSS, as described previously (23). Three days later, lymphocytes were isolated from spleens and s.c. and mesenteric lymph nodes.
Diphtheria toxin treatment
Zbtb46-DTR→WT and control WT→WT chimeras received 100 ng/g body weight of diphtheria toxin (DT) (Sigma-Aldrich, Oakville, ON) i.p. initially, then 20 ng/g i.p. every other day.
Antitumor CTL responses in vivo
Effector CD45.1 OT-1 cells (5–10 × 106) were labeled with CFSE and injected i.v. into mice bearing B16-OVA tumors 0.5–1 cm in diameter; in every experiment, the same number of OT-1 T cells was injected into the mice. The tumors were recovered 3 or 6 d later to assess the frequency and proliferation in tumors and lymphoid tissues by flow cytometry. Intracellular IFN-γ production in OT-1 CTLs was assessed after restimulation with SIINFEKL peptide in vitro.
Analysis of CTL proliferation ex vivo
WT and Zbtb46-DTR bone marrow chimeras bearing B16-OVA tumors 0.5–1 cm in diameter were injected i.v. with 5–10 × 106 CD45.1 OT-1 CTLs or Thy1.1+ P14 CTLs and treated with DT on days 1, 3, and 5. Six days after CTL transfer, tumor-infiltrating CD45+ cells were sorted into CD11c+ MHC II+ DC and non-DC populations. The tumor-infiltrating non-DCs were combined, labeled with CFSE, and incubated with or without sorted tumor DCs (1–2 × 104) from Zbtb46-DTR or WT bone marrow chimeras for 3 d. Because P14 CTLs were absent from tumors, they were sorted from lymph nodes from the same tumor-bearing mice using anti-Thy1.1 Ab and added to the sorted non-DC population to constitute a frequency of ∼2%. In some experiments, sorted tumor DCs were pulsed with exogenous OVA protein (100 μg/ml) in the presence of GM-CSF (10 ng/ml) and LPS (1 μg/ml) for 4 h, then washed three times before incubation with tumor-infiltrating non-DCs.
Statistical analysis
Results are expressed as mean ± SEM. Data were analyzed by Student t test and two-way ANOVA. The analyses were performed with the GraphPad Prism statistical program. The p values <0.05 were considered significant.
Results
Depletion of pre-cDC–derived DCs (DC1 and DC2) in tumors of Zbtb46-DTR mice
We made use of the Zbtb46-DTR transgenic mouse model to deplete pre-cDC–derived DCs. Because prolonged DT treatment in these mice causes systemic toxicity (9), we created Zbtb46-DTR→WT and control WT→WT bone marrow chimeras in lethally irradiated mice for the following studies in this report. We injected B16 melanoma cells s.c. into Zbtb46-DTR and control bone marrow chimeras 8–10 wk after bone marrow reconstitution and gave them DT (100 ng/g body weight) when the tumors reached ∼5 mm in diameter. Spleens, TDLNs, and tumors were removed 1 d later for flow cytometric analysis. We gated on the CD45+ lineage (CD3, B220, CD49b, Ly-6G)neg tumor-infiltrating cell population. Consistent with previous reports (9), DT rapidly depleted DC1 and DC2 and spared DC3 in spleen, TDLN, and tumors in Zbtb46-DTR bone marrow chimeras, but not in control chimeras, resulting in a striking contraction of the DC compartment (Fig. 1A). As expected, DT depleted bone marrow pre-cDC in Zbtb46-DTR bone marrow chimeras, the precursors for tissue DC1 and DC2 (Fig. 1B), without affecting the frequency of bone marrow promonocytes (CD117+CD115+Ly-6Clo) and monocytes (CD117−CD115+Ly-6Chi) (Fig. 1C). These results confirm that DT treatment in Zbtb46-DTR bone marrow chimeras selectively depletes pre-cDC–derived DCs in lymphoid tissues and tumors.
Intratumor CTLs proliferate and expand in the absence of tumor DC1 and DC2
To investigate the role of pre-cDC–derived DCs in regulating CTL responses, we inoculated Zbtb46-DTR and control bone marrow chimeras with B16 tumors that express OVA (B16-OVA). We transferred CFSE-labeled OT-1 CTLs, which recognize the OVA-derived SIINFEKL peptide bound to H-2 Kb, 10 d after tumor inoculation and recovered lymphoid tissues and tumors 3 and 6 d later to assess their frequency and proliferation, as determined by CFSE dilution. Preliminary studies indicated that durable depletion of tumor DCs required higher doses of DT than those used in earlier reports to deplete lymphoid tissue DCs. We started DT treatment 1 d prior to CTL transfer at a dose of 100 ng/g body weight and continued with a dose of 20 ng/g body weight on days 1, 3, and 5 after CTL transfer.
In both WT and Zbtb46-DTR bone marrow chimeras, the spleen contained small numbers of OT-1 CTLs with low rates of proliferation at days 3 and 6, consistent with limited OVA Ag presentation in this organ (Fig. 2A, 2B). In TDLNs, we detected no difference in the frequency or rate of proliferation of OT-1 CTLs at day 3 in WT and Zbtb46-DTR bone marrow chimeras; however, at day 6, their frequency and proliferation rate increased significantly in control but not in Zbtb46-DTR bone marrow chimeras. These findings support the current view that tumor Ag presentation in TDLNs relies on pre-cDC–derived DCs (15, 16).
OT-1 CTLs constituted ∼1% of CD45+ cells in tumors at day 3 in WT and Zbtb46-DTR bone marrow chimeras (Fig. 2A). Their frequency and absolute number per gram of tumor (data not shown) increased significantly at day 6 in both control and Zbtb46-DTR bone marrow chimeras. In both groups of mice, ∼40% of tumor OT-1 CTLs had divided at day 3, and this frequency increased further at day 6 (Fig. 2B). In Zbtb46-DTR bone marrow chimeras, the rate of OT-1 CTL division in tumors was 4-fold higher than in TDLNs (as compared with 2.5-fold higher in WT mice), consistent with in situ proliferation being a key determinant of intratumor CTL frequency in this model. We found that tumor OT-1 CTLs recovered from both groups of mice 6 d after cell transfer expressed similar levels of intracellular IFN-γ after incubating them with SIINFEKL peptides in vitro (Fig. 2C). We also found no difference in the mean tumor weight in the two groups of mice 6 d after cell transfer (tumors were recovered and analyzed at the same time after tumor inoculation in all mice) (Fig. 2D). Thus, there was no difference in tumor OT-1 CTL frequency and IFN-γ expression in WT and Zbtb46-DTR bone marrow chimeras.
Immunotherapy expands the frequency of tumor DCs
Collectively, our results suggested that pre-cDC–derived tumor DC1 and DC2 are dispensable for intratumor OT-1 CTL expansion and function. These results surprised us because previous studies in our laboratory (J. Diao and M.S. Cattral, unpublished observations) and other laboratories have shown that the Ag-presenting efficacy of CD103+ DCs excels over monocyte-derived DCs (15, 18). Further, our initial studies (Fig. 1) indicated that the overall frequency of DCs in tumors and lymphoid tissues had contracted markedly after a single dose of DT despite the persistence of DC3.
Under inflammatory conditions, Ly-6Chi monocytes are recruited to sites of injury where they can differentiate into inflammatory monocyte-derived DCs (24). Given that immunotherapy can augment inflammation in the TME (25), we hypothesized that adoptive T cell therapy in Zbtb46-DTR bone marrow chimeras increased the frequency of DC3, mitigating the loss of DC1 and DC2. We therefore reanalyzed DC populations in tumors, spleens, and TDLNs of WT and Zbtb46-DTR bone marrow chimeras treated with adoptive CTLs. We confirmed that DC1 and DC2 were virtually absent in lymphoid tissues and tumors in Zbtb46-DTR bone marrow chimeras at day 3 and 6 after CTL therapy (Fig. 3). The residual Lin−CD11c+MHC II+CD11b+ cells were CD135 (Flt3) negative, further confirming their absence.
We compared the frequency of DCs in mice 6 d after CTL therapy with those that did not receive CTLs (Fig. 3B). CTL therapy increased the frequency of tumor DCs 3-fold in both WT and Zbtb46-DTR bone marrow chimeras; expansion of both tumor DC2 and DC3 accounted for this increase in WT mice, whereas the frequency of DC1 did not change appreciably. The overall frequency of tumor DCs in the Zbtb46-DTR bone marrow chimeras remained similar to WT controls because of the marked expansion of DC3. Adoptive T cell therapy had little impact on the frequencies of spleen and TDLN DCs in Zbtb46-DTR and WT bone marrow chimeras.
In addition to DC3, monocytes produce macrophages and MDSC in the TME (26, 27), raising the possibility these populations could also be affected in Zbtb46-DTR bone marrow chimeras. We found that the frequencies of tumor monocytes/macrophages (Lin−CD11c−MHC II−CD11b+), MDSC (CD11b+Gr-1+Ly-6G−), and granulocytes (Gr-1+Ly-6G+) were similar in Zbtb46-DTR bone marrow chimeras and controls 6 d after CTL transfer (Supplemental Fig. 1). A report noted an association between the frequency of DC3 and M1-like MHC IIhi tumor-associated macrophages in various implantable tumors (14); however, we found no correlation between the frequency of tumor DC3 and this tumor-associated macrophage population in WT and Zbtb46-DTR bone marrow chimeras (data not shown).
Tumor DC3 stimulate CTL proliferation
The increased frequency of tumor DC3 in Zbtb46-DTR bone marrow chimeras suggested that they could compensate for the loss of DC1 and DC2. Many reports indicate, however, that monocyte-derived DCs stimulate T cells poorly despite their prodigious capacity to uptake and process Ags (18, 28). Flow cytometric analysis of freshly isolated tumor DCs from Zbtb46-DTR and WT bone marrow chimeras revealed that they express similar levels of intracellular iNOS and TNF. Tumor DCs from both groups of mice expressed low levels of IL-12, but after LPS stimulation the levels were higher in DCs from WT bone marrow chimeras (Fig. 4A).
To further understand the functional relevance of DC3, we designed an experiment to test their ability to stimulate CTLs under conditions that mimicked the TME (Fig. 4B). We injected WT and Zbtb46-DTR bone marrow chimeras bearing B16-OVA tumors with unlabeled OT-1 CTLs and treated them with DT on days 1, 3, and 5, which depleted DC1 and DC2 in Zbtb46-DTR bone marrow chimeras but not in WT bone marrow chimeras. Six days after OT-1 CTL transfer, we sorted tumor-infiltrating CD45+ cells into two populations: 1) CD11c+MHC II+ DCs and 2) non-DCs (OT-1 CTLs constitute ∼5–10% of these cells). The tumor-infiltrating non-DCs from Zbtb46-DTR and WT bone marrow chimeras were combined to control for potential differences in these cell populations, labeled with CFSE, and incubated with or without sorted- tumor DCs from either Zbtb46-DTR or WT bone marrow chimeras. In the absence of tumor DCs, OT-1 CTLs exhibited low levels of proliferation, which did not increase significantly with the addition of a large amount of exogenous OVA to the culture medium, indicating that the non-DC population had limited intrinsic Ag-presenting capacity (Fig. 4C). By contrast, the addition of tumor DCs from either WT or Zbtb46-DTR mice stimulated CTL proliferation; notably, these DCs had acquired OVA directly from the TME prior to tumor recovery. Addition of exogenous OVA to these cultures had no effect on proliferation. Collectively, these data indicate that DC3 possess the capacity to cross-present Ags and stimulate proliferation of Ag-specific CTLs.
To verify that OT-1 CTL proliferation in this assay was Ag specific, we added P14 CTLs to the sorted non-DC tumor-infiltrating population, labeled them with CFSE, and incubated them with or without sorted B16-OVA tumor-infiltrating CD11c+MHC II+ DCs from Zbtb46-DTR and WT bone marrow chimeras, as described in Fig. 4B. Neither population of tumor DCs stimulated proliferation of P14 CTLs, whereas P14 CTLs isolated from the spleen of tumor-bearing mice proliferated in the presence of gp33–41 peptide, but not control peptide (Supplemental Fig. 2).
The TME is a well-recognized cause of DC dysfunction (29), which can be overcome, at least partly, by stimulating them with TLR agonists (15, 30). To further assess their potential for Ag presentation, we pulsed freshly isolated tumor DCs from WT and Zbtb46-DTR bone marrow chimeras with OVA and LPS before culturing them with the non-DC tumor-infiltrating cell population, as described in Fig. 4B. Under these conditions, tumor DCs from Zbtb46-DTR and WT bone marrow chimeras induced similar levels of OT-1 T cell proliferation and expansion (Fig. 5A, 5B).
Discussion
Recent studies in mouse tumor models have revealed several roles of pre-cDC–derived tumor CD103+ DC1 that contribute to the generation of antitumor immune responses: transportation of tumor Ags to TDLNs, presentation of tumor Ags to cognate T cells in TDLNs, and spreading of tumor Ags to other resident DC subpopulations in TDLNs (15, 16). Several reports also highlight their importance in presenting tumor Ags to intratumor CTLs and in attracting CTLs to the TME (18–20). We found that depletion of pre-cDC–derived DCs in Zbtb46-DTR bone marrow chimeras abolished CTL proliferation and expansion in TDLNs, confirming their dominance in presenting tumor Ags in lymphoid tissues. Unexpectedly, intratumor CTL accumulation and IFN-γ expression were unaffected by their absence. We show that adoptive CTL therapy increases the frequency of monocyte-derived tumor DC3, which possess the capacity to cross-present tumor Ags and support CTL proliferation.
Although monocyte-derived DCs constitute a high proportion of the total DC population in many tumors, distinguishing their function(s) from pre-cDC–derived DCs has proven to be difficult because of overlapping cell surface markers. The Zbtb46-DTR mouse model enabled us to overcome this hurdle; with continuous DT therapy, we could maintain selective depletion of DC1 and DC2 in both lymphoid and tumor tissues. Three days after CTL transfer, the distribution and frequency of CTLs in lymphoid and tumor tissues was similar in both Zbtb46-DTR and WT bone marrow chimeras, indicating that their initial recruitment to these sites occurs independently of DC1 and DC2. Six days after transfer, we detected CTL proliferation in the TDLNs in WT but not in Zbtb46-DTR bone marrow chimeras. Although a previous report indicated that monocyte-derived DCs can present tumor Ags to tumor-specific CD8+ T cells in TDLNs under inflammatory conditions (31), this mechanism seemed inefficient in Zbtb46-DTR bone marrow chimeras. Thus, proliferation and expansion of intratumor CTLs in Zbtb46-DTR bone marrow chimeras reflect stimulation within the tumor.
CTL stimulation in tumors requires interactions between their TCRs and antigenic peptides presented on MHC class I molecules. Live-cell microscopy of tumors shows that as CTLs travel through the TME, they engage tumor-infiltrating DCs, other myeloid cells, stromal cells, and tumor cells, all of which could potentially stimulate CTLs (18, 32–34). Our studies of freshly isolated tumor-infiltrating cells indicate that Ag cross-presentation activity resides mostly in DCs. Tumor DCs from WT and Zbtb46-DTR bone marrow chimeras induced similar levels of CTL proliferation in vitro, which was augmented by maturing DCs with LPS. Although isolated DC1 cross-presents Ags more effectively to T cells than the other DC subsets on a per-cell basis, their impact in this assay is likely limited by their low frequency.
Inflamed tissues attract DC precursors by releasing a diverse array of molecules, including GM-CSF, chemokines, and ATP (21, 35, 36). Our study illustrates that adoptive cell therapy with CTLs increases the frequency of tumor DCs in both WT and Zbtb46 bone marrow chimeras, presumably by augmenting proinflammatory signals in the TME (25). In WT bone marrow chimeras, DC2 and DC3 increased, whereas the frequency of DC1 remained constant. Whether the apparent static response of DC1 is real—possibly due to a shift in the differentiation of pre-cDCs toward DC2—or whether preferential migration of DC1 to TDLNs masks an increase requires further investigation. We and others have detected a higher frequency of circulating monocytes in Zbtb46-DTR than in WT bone marrow chimeras with DT treatment (9), which may potentiate the expansion of tumor DC3 following adoptive cell therapy. A recent report suggested that tumor DC3 generation in the Lewis lung carcinoma model may depend on instructional cues from tumor DC1/DC2 (14); our findings indicate that tumor DC3 in B16 melanoma can develop and function independently of pre-cDC–derived DCs. The increased frequency of DC3 following adoptive CTL therapy in our study was not accompanied by a change in the frequency of tumor macrophages or MDSC. We speculate that cytokines released in the TME following adoptive cell therapy skews monocyte differentiation toward DC3 rather than to these other cell types. For example, IFN-γ released from CTLs and NK cells promotes development of DCs and their production of IL-12, a key driver of CTL cytotoxicity (25, 37). In some settings, however, IFN-γ may switch monocyte differentiation from DCs to macrophages (38).
Reports of other studies have shown that monocyte-derived tumor DCs suppress CTL proliferation (14), which was linked to their expression of iNOS and production of NO. In our study, ∼50% of DC3 express iNOS in both WT and Zbtb46-DTR bone marrow chimeras. As compared with DC1/DC2, DC3 showed a similar capacity to stimulate OT-1 CTLs, suggesting that iNOS expression per se may not reliably predict APC activity and CTL responses in vivo. Interestingly, Marigo et al. (25) reported that CTLs stimulate tumor DCs to produce NO, which inhibited tumor growth.
The identification of DC subsets in mice that specialize in tumor Ag cross-presentation has ignited interest in targeting these cells to improve antitumor immune responses (15, 18). This interest has been bolstered by analyses of The Cancer Genome Atlas showing that patient survival for many cancers correlates positively with the gene expression signature for human DC1 (18, 20, 39). Whether this correlation relates to better priming of tumor Ag–specific T cells in lymphoid tissues, the TME, or both has been difficult to discern. We believe the Zbtb46-DTR mouse provides an ideal model to evaluate CTL responses in the TME with limited or no contribution from TDLNs. Our results suggest that although pre-cDC–derived DCs are critical during the priming phase of antitumor immunity and secondary expansion of CTLs in lymphoid tissues, they appear to have a less vital role in the TME during the effector phase.
The ability of tumor DC3 to support CTL expansion in the TME after adoptive CTL therapy is consistent with studies in a variety of infection, vaccination, and autoimmunity models showing that monocyte-derived DCs play a key role in cross-priming T cell responses, particularly at the local tissue level (6, 40, 41). Indeed, monocyte-derived DCs generated with GM-CSF and IL-4 cross-present Ags to CD8+ T cells as effectively as spleen CD24+ cDC (42). In tumor models, myeloid cells influence the antitumor immune effects of chemotherapy, radiation therapy, immune checkpoint inhibitors, and adoptive cell therapy (21, 43, 44). The success of PD-1–based immunotherapies depends on CD28 costimulation by B7-expressing tumor APCs to promote T cell expansion and prevent T cell exhaustion (45, 46). Given the natural abundance of tumor DC3 in various human cancers (14), they provide an additional therapeutic target to boost antitumor immune responses.
Acknowledgements
We thank R. Gorczynski for reviewing the manuscript. We thank P. Ohashi for providing P14 mice and the gp33 peptide and A. Elford for assisting in the P14 mouse studies.
Footnotes
This work was supported by the Canadian Institutes for Health Research (CIHR) (130438 to M.S.C.), Astellas Inc., and the Toronto General Hospital Transplant Program. M.S.C. is the holder of a CIHR/Astellas Research Chair. M.T. is the recipient of a CIHR training award (121831).
The online version of this article contains supplemental material.
References
Disclosures
The authors have no financial conflicts of interest.