Depletion of Conventional Type-1 Dendritic Cells in Established Tumors Suppresses Immunotherapy Efficacy

Cross-priming is a key function that controls the induction of antigen-specific CTLs. The ability to uptake cellular remains and process the engulfed antigens for MHC class I presentation was discovered to be mainly performed by a minority subset of dendritic cells (DC; ref. 1). Such myeloid DCs are known as conventional-type 1 dendritic cells (cDC1; ref. 2). They were first identified in the spleen of mice as the CD11c⁺ DEC205⁺ CD8α⁺ cells whose involvement in CD8 T-cell responses was shown (3–5).

Detailed study of the transcripts and proteins expressed by DCs better profiled this cross-presenting subset that is characterized by the surface coexpression of CD11c, XCR1, and DNGR-1 (2). In mice, two subsets of cDC1 cells have been identified depending on expression of the CD103 integrin, which defines the CD103⁺ migratory cDC1s and a CD103⁻ resident subtype constitutively located in secondary lymphoid organs (6).

Much of the research pertaining to cDC1 function comes from the fact that this subset of DC needs the transcription factors BATF3 and IRF8 for their ontogeny in the bone marrow (7). Hence, Batf3 knockout mice are selectively devoid of cDC1 cells (8, 9). Interestingly, Batf3^−/− mice are severely deficient when it comes to mounting antiviral CTL responses as well as rejecting immunogenic spontaneously regressing tumor cell lines (8). Similarly, other mouse strains lacking cDC1 cells (e.g., XCR-1 DTA mice) show impaired antitumor immune responses (10). The main mechanisms carried out by cDC1 cells are MHC class I presentation of exogenous antigens (cross-presentation) and responsiveness to viral double-stranded RNA via endosomal TLR3 (11), giving rise to high levels of IL12 production (12). The function of these DCs is critically upregulated by type-I IFNs (13, 14) and activated helper CD4 T cells expressing CD40 L (13). The exact molecular mechanism that redirects endosomal engulfed material to MHC I for antigen presentation is as yet incompletely understood, although there is genetic evidence for the involvement of proteins related to vesicular trafficking such as WDFY4 (15) and Sec22b (16).

Two groups simultaneously reported that tumor immunotherapy with anti–PD-1 and anti-CD137 mAbs was ineffective in cDC1-deficient Batf3^−/− mice (17, 18). The mechanisms could be traced to dysfunctional cross-priming of surrogate tumor antigens (18). Interestingly, adoptive T-cell therapy is also less efficient in Batf3 knockout mice showing a complete lack of antigen spreading that was needed to prevent tumor relapse (19–21). IL12 production by cDC1 cells has also been shown to play a key role in PD-1 blockade antitumor efficacy (22).

In humans, cDC1 cells can be identified because they share with their mouse counterparts the expression of XCR1 and DNGR-1 (23–25). In peripheral blood, they are best identified by the expression of CD141 (BDCA3; ref. 26). Numbers of cDC1 cells in mice and humans are greatly increased by treatment with a soluble form of Flt3L, which promotes their ontogeny in the bone marrow (27–29).

The involvement of cDC1 cells in cancer immunology and immunotherapy was underscored by observations according to which tumors failing to produce the CCL4 or CCL5 chemokines that attract cDC1 cells behave as poorly immunogenic (30, 31). In human tumors, a transcriptional cDC1 signature and the expression of Batf3 are strongly associated with the levels of CD8⁺ T-cell infiltration (32–34). Reportedly, migratory cDC1s take up tumor antigen in mouse models (35) and may transfer it to secondary lymphoid organ-resident cDC1 cells (36). cDC1 cells seem to have a close functional interaction with natural killer (NK) cells (32, 37), which is counter regulated by prostaglandins (38).

One important question that remains unanswered is whether cDC1 cells are required once immunotherapy has been instigated to mediate and sustain the beneficial effect of treatment. With the help of XCR1-DTR-Venus transgenic mice (39), we have demonstrated that diphtheria toxin (DT)–mediated depletion of cDC1 cells upon treatment hampers response to various immunotherapy agents. Moreover, microscopy imaging indicates that cDC1 cells actively interact with tumor antigen-specific T cells in the tumor microenvironment and are involved in tumor T-cell infiltration.

Mice were housed at the animal facility of the CIMA Universidad de Navarra (Pamplona, Spain). XCR1-DTR-VENUS (Xcr1^{tm2(HBEGF/Venus)Ksho}) mice were obtained from the RIKEN Bioresource Research Center and have been described elsewhere (39). Mice were bred homozygous and crossed with C57BL/6 mice (Envigo) to obtain F1 heterozygous XCR1-DTR-Venus mice. hCD2RFP (C57BL/6 hCD2-DsRed) transgenic mice were a kind gift from Mark Coles from University of York (York, United Kingdom; ref. 40) and were bred to OT-I mice (Jackson, C57BL/6-Tg(TcraTcrb)1100Mjb/J, RRID:IMSR_JAX:003831) to obtain F1 hCD2RFPxOT-I mice. Batf3^−/− mice (B6.129S(C)-Batf3^tm1Kmm/J, RRID:IMSR_JAX:013755) were kindly provided by Kenneth M. Murphy (Washington University, St. Louis, MO; ref. 8). OT-I and CD45.1 (Jackson) were bred to homocygosity to allow in vivo cell tracking. All animal procedures were approved by the Institutional Animal Experimentation Ethics Committee and the regional government of Navarra (protocol 036-18, 089-18, and 087-21). Ages of mice included in experiments ranged from 8 to 12 weeks.

B16.OVA mouse melanoma cells and MC38 mouse colon carcinoma cell lines were kindly gifted by Dr. Lieping Chen (Yale University, New Haven, CT) and Dr. Karl E Hellström (University of Washington, Seattle, WA), respectively. E0771 tumor cells were obtained from ATCC. B16.OVAiRFP670 tumor cells were generated by lipofectamine 2000 (Thermo Fisher Scientific) transfection of B16.OVA cells with piRFP670-N1 (a gift from Vladislav Verkhusha; Addgene plasmid #45457; http://n2t.net/addgene:45457; RRID:Addgene_45457; ref. 41), an expression plasmid and selected on the basis of FACS in a MoFloAstrios EQ cell sorter (Beckman Coulter). Cultures were tested for Mycoplasma monthly using the Mycoalert kit (Promega). These cell lines were authenticated by Idexx Radil (Case 6592-2012). Cell lines were subjected to two or three passages before using in the experiments. Activated OT-I cells for therapy and imaging experiments were prepared as described previously (20). Briefly, total splenocytes of OT-I or OT-I x hCD2^RFP mice were cultured at 10⁶ cells/mL with 100 ng/mL of SIINFEKL peptide and 30 U/mL of hIL2 (Proleukin, Novartis) for 48 hours, then the medium was removed and cells were cultured at the same cell density for an additional 48 hours with 30 U/mL hIL2 in the absence of SIINFEKL peptide. All cells were grown in RPMI1640 media supplemented with GlutaMAX (Gibco), 10% heat-inactivated FBS, 50 μmol/L of 2-mercaptoethanol, 100 U/mL of penicillin, and 100 μg/mL of streptomycin at 37°C with 5% CO₂ (complete media).

A total of 5 × 10⁵ MC38, B16.OVA, or E0771 tumor cells were subcutaneously injected into the right flank of XCR1-DTR-Venus or Batf3^−/− mice. Mice were treated at the indicated time points with DT (Sigma) with a single intraperitoneal dose of 1 μg/mL plus two additional doses of 0.5 μg/mL every 3 days to maintain full cDC1 depletion (when indicated in the figures). Tumors were measured every 3 days with a caliper and the volume was calculated (length × width²/2). In addition, mice were monitored for survival and euthanized when any tumor size reached a diameter of 15 mm or mice showed signs of distress. In some experiments, C57BL/6 mice deficient for BATF3 were used. Mice were treated on the indicated days with 200 μg of anti–PD-1 (clone RPM1-14, Bio X cell, RRID:AB_1107747), anti-CD137 (clone 3H3, Bio X cell, RRID:AB_2687721), anti-CTLA4 (9D9, Bio X cell, RRID:AB_10949609), and Rat IgG control (Bio X cell, RRID:AB_2687813) when used in single treatment or with 100 μg/mL of each antibody when used in combination with other immunostimulatory antibodies. For adoptive transfer experiments, mice were intravenously injected with 2 × 10⁶ activated OT-I cells that were isolated and cultured as explained above.

For the analysis of cDC1 depletion of XCR1-DTR-Venus in tumors, draining lymph nodes and spleen, MC38-bearing mice were treated with a single dose of 1 μg of DT and were sacrificed at the indicated time points. For the analysis of CD8 T cells, mice were injected with 0.5 × 10⁶ B16.OVA cells. At day 6, mice were treated with 1 μg of DT, at day 7 mice were transferred with 2 × 10⁶ activated OT-I CD8⁺ T cells, and at days 9 and 12 mice were treated with additional 0.5 μg of DT. Mice were treated with anti PD-1 and anti CD137 antibodies at day 10 and tumors and tumor-draining lymph nodes (TDLN) were harvested at day 14 after tumor inoculation. In the MC38 model, DT was injected at days 6 and 9 and antibodies at days 7 and 10 after tumor cell injection. Tumors and TDLN from euthanized mice were harvested at day 11. Organs were collected and single-cell suspensions were stained with the following antibodies: CD8α (BV510,RRID:AB_2561389; APC, RRID:AB_312750; BUV395, RRID:AB_2732919), F4/80 (PeCy7, RRID:AB_893490; BV421, RRID:AB_10901171), TCRβ BV605 (RRID:AB_2687544), CXCR6 PeCy7 (RRID:AB_2721669), PD-1 PercpCy5.5 (RRID:AB_10550092), Tox PE (RRID:AB_10853657), Ki67 AF488 (RRID:AB_10900418), TCF7 AF647 (RRID:AB_2797631), CD11c (APC, RRID:AB_313778; BV510, RRID:AB_2562010), IA/IE (APC Fluor 780, RRID:AB_1548783; Percp-Cy5.5, RRID:AB_2191072), CD103 PE (RRID:AB_535948), CD11b (BUV395, RRID:AB_2740936; PeCy7, RRID:AB_312798),CD45.2 (PB, RRID:AB_492873; BUV496, RRID:AB_2870691) CD45.1 BV786 (RRID:AB_2740538; all from BioLegend or BD, except TCF1 that was from Cell Signaling Technology and Tox from BD; ref. 42). Immunostained cells were analyzed with a Cytoflex LX or XS (Beckmann Coulter) or a FACScanto flow cytometer (BD). Fluorescence minus one or biological comparison controls were used for cell analysis. FlowJo software (TreeStar) was used for data analysis.

A six-color multiplex immunofluorescence panel based on tyramide signal amplification was used for simultaneous detection of CD3 (T cells, RRID:AB_443425), CD8 (CTLs, RRID:AB_2756376), Foxp3 (regulatory T cells, RRID:AB_2797979), CD4 (Th cells, RRID:AB_2798898), CD11b (myeloid cells, RRID:AB_1216361), and Ki67 (proliferating cells, RRID:AB_443209) and diamidino-2-phenylindole on tumor sections from formalin-fixed paraffin-embedded (FFPE) samples as described previously (43). Briefly, 4-μm-thick sections obtained from FFPE tissue blocks were deparaffinized and rehydrated from ethanol to water. Antigen retrieval with citrate (pH6, PerkinElmer) or ethylenediaminetetraacetic acid (EDTA; pH9, Dako) target retrieval solution was performed at the beginning of each sequential round of antibody staining. Each round consisted of heat-induced antigen retrieval followed by protein blocking (Antibody Diluent/Block, Akoya Bioscience), incubation with primary antibody, anti-rabbit secondary antibody (Opal Polymer anti-rabbit horseradish peroxidase Kit, Perkin Elmer) finishing with Opal fluorophore incubation diluted in 1XPlus Amplification Diluent (Akoya Bioscience). The panel primary antibody description is provided in Supplementary Materials and Methods and sections were mounted with Faramount Aqueous Mounting Medium (Dako).

Whole tissue sections were scanned on a Vectra-Polaris Automated Quantitative Pathology Imaging System (Akoya Biosciences). Akoya Biosciences’ Inform software (V.2.4.8) was used to remove the autofluorescence as determined by an unstained slide and to perform the spectral unmixing of the images. To analyze the invasive margin, six rectangular regions were randomly selected for each tumor containing the tumor margins, analyzed and the mean infiltration denoted by each marker for each tumor section was plotted.

For intravital microscopy of B16.OVA tumors engrafted in the liver, the spleens of XCR1-DTR-Venus mice were surgically exposed under isoflurane anesthesia and 10⁶ B16.OVA.iRFP670 cells were injected in 50 μL of PBS. Then, spleens were surgically removed, and mice were allowed to recover. Four days later, mice were intravenously injected with activated 5 × 10⁶ OT-IxhCD2^RFP CTLs. Mice were intraperitoneally injected with anti–PD-1 plus anti-CD137 or Rat IgG control mAbs at day 7 or 9 after tumor cell injection and DT was given 24 hours before such Ab treatments as indicated in the figures. At days 9 and 11 after tumor cell injection, intravital confocal microscopy of the liver was performed as described previously (44, 45). Mice were injected with 5 μg of CD31-AF647 antibody (RRID:AB_2161030). Briefly, the liver was surgically exposed and isolated from respiratory movements by separating the rib cage by pulling with a thread and keeping the liver tissue adhered to the coverslip with the help of wet pieces of paper. The temperature of the mice was maintained with a rectal probe connected to a heating blanket (Kemp) and mice were kept asleep under Isoflurane anesthesia (2%). Mice were placed on a custom-built stage and imaged with an LSM880 inverted microscope (Zeiss) equipped with a 25X water immersion objective (NA, 0.8). Imaging sessions took from 2 to 4 hours per mice and time-lapse acquisitions lasted from 30 minutes to 2 hours with frames taken every 2 minutes. Several tumor foci were imaged per session and mice. Time-lapse videos were analyzed using the IMARIS (Bitplane) software. Both Venus-positive (Venus⁺) and red fluorescent protein–positive (RFP⁺) cells were segmented using the spots tool and tracked manually overtime. Individual cell motility parameters were generated from such tracks. Speed (track length/time), chemotactic index or straightness (track displacement length/track length), and motility index (track displacement length²/4*track duration) were calculated. Interactions of T cells with Venus⁺ cells were quantified selecting time points in which Venus⁺ cells were closer than 12 μm from an RFP⁺ CD8⁺ T cell.

Time-lapse videos were generated in IMARIS software and edited with Final Cut Pro (Apple software).

Data were processed using GraphPad Prism 6.0. Means and SEM are presented as averages and error bars unless otherwise indicated in the figure legends. All experimental repetitions and numbers of specimens and mice are indicated in the figure legends. For the statistical comparison of tumor growth, individual AUC were calculated for each mouse and compared using ANOVA tests, which were used to analyze statistical differences between independent groups unless otherwise indicated. P values are shown for any relevant statistical difference in the figures.

The data generated in this study are available upon request from the corresponding authors.

To investigate the need for the presence of cDC1 cells during immunotherapy, we made use of XCR1-DTR-Venus transgenic mice (39). cDC1 cells can be identified upon flow cytometry analysis of CD11c⁺ CD11b⁻ cells because of selective expression of the Venus fluorescent protein. We engrafted MC38-derived tumors under the skin of XCR1-DTR-Venus mice and intraperitoneally injected DT (Fig. 1A).

Twenty-four hours after DT treatment, splenic CD11c⁺ CD11b⁻ CD8α⁺ cells almost disappeared, a phenomenon that was also observed in terms of the disappearance of MHC-II⁺ Venus⁺ cells (Fig. 1B–D; Supplementary Fig. S1A). TDLNs experienced a similar outcome following DT treatment (Fig. 1B–D; Supplementary Fig. S1A). In excised tumors, all Venus⁺ DCs and most CD103⁺ CD11c⁺ MHCII⁺ F4/80⁻ DCs were absent as a result of DT treatment (Fig. 1B–D; Supplementary Fig. S1A). In contrast, cDC2s were respected upon DT treatment (Supplementary Fig. S1B). Depletion of cDC1 cells following a single DT injection lasted for at least 5 days (Supplementary Fig. S1C). cDC1 cell numbers remained low for 5 days and eventually recovered when mice were analyzed at day 7 after DT treatment (Supplementary Fig. S1D).

We used the MC38 tumor model to characterize the need for cDC1 cells during treatment with efficacious immunotherapies. MC38-derived subcutaneous tumors partially respond to anti-CD137, or to anti–PD-1 mAbs and intensely respond to a combination of anti-CD137 plus anti–PD-1 mAbs (18). Experiments of DT depletion and immunostimulatory mAb treatments were performed in MC38 tumor-bearing XCR1-DTR-Venus transgenic mice and in Batf3 knockout mice (Fig. 2A). DT treatment started at day 4 after tumor inoculation, while the first immunostimulatory mAb was given at day 5. At day 4, the expansion of anti-gp70 tumor-reactive CD8 T cells were readily detectable in such tumor-bearing mice (Supplementary Fig. S2A). The partial efficacy of treatment of anti–PD-1 or anti-CD137 mAbs was lost in the groups of mice pretreated with DT (Fig. 2B). The synergistic immunotherapy combination of anti–PD-1 plus anti-CD137 mAbs was very efficacious, eradicating 14 of 20 tumors, and again such efficacy was completely lost in DT-treated mice and in cDC1 constitutively deficient Batf3^−/− mice (Fig. 2C). In sentry mice from the experiments in Fig. 2B and C, cDC1 cells were verified as having been efficiently depleted (Supplementary Fig. S2B). Of note, DT treatment did not impair the efficacy of immunotherapy in wild-type (WT) mice (Supplementary Fig. S2C).

To study when cDC1 cells were needed for immunotherapy efficacy, we performed DT-mediated depletion at different time points. Mice that recovered from preengraftment transient depletion attained similar efficacy upon anti-CD137 plus anti–PD-1 treatment (Fig. 2D). However, tumor growth prior to immunotherapy onset was hastened in predepleted groups, suggesting an immunotherapy-independent cDC1-mediated control of tumor progression at least in this tumor model (Fig. 2D). When DT treatment was performed 1 day after the first immunotherapeutic mAb injections (Fig. 2E), anti–PD-1 and anti-CD137 efficacy were reduced upon cDC1 depletion but the efficacy of the combination of both antibodies was retained irrespective of cDC1 cell depletion at that timepoint.

In the clinic, anti–PD-1 plus anti-CTLA4 (nivolumab plus ipilimumab) constitutes the approved treatment of choice in a number of malignancies (46–48). Figure 3A shows that treatment with a mouse surrogate of such a combination lost most of its efficacy upon DT treatment against the transplanted E0771 breast cancer model. Interestingly, when testing cDC1 depletion in anti–PD-1– and anti-CTLA4–based immunotherapies, cDC1 cells were only required for the efficacy of PD-1 blockade but not for the efficacy of anti-CTLA4 mAbs (Fig. 3B).

Adoptive T-cell therapy is also gaining clinical application (49, 50). To model it, we used B16.OVA engrafted tumors and adoptive transfer of activated OT-I TCR-transgenic T cells recognizing ovalbumin (OVA). Results in Fig. 3C indicate that the control of tumors by adoptively transferred OT-I cells was partially lost upon cDC1 depletion. These results are consistent with previous observations in Batf3^−/− tumor-bearing mice treated with adoptive T-cell therapy (19), but highlight the relevance of cDC1s specifically during the administration of T cells.

In all previous experiments, we used homozygous XCR1-DTR-Venus knock-in mice, that happen to have a disrupted XCR1 locus due to the fact that depletions were more efficient in such DTR homozygous mice. However, this created a doubt as to the potential importance of XCR1 for the efficacy of immunotherapy. This was ruled out as shown in Supplementary Fig. S3, because therapies with anti-CD137+ anti–PD-1 (Supplementary Fig. S3A), anti–PD-1+anti-CTLA4 (Supplementary Fig. S3B) and OT-I adoptive T-cell transfer (Supplementary Fig. S3C) were as efficacious in homozygous as in heterozygous transgenic mice, which have a functional XCR1 gene in the latter case.

Focusing on the efficacy of anti–PD-1+anti-CD137 antibodies, we used multiplex tissue immunofluorescence to study the lymphocyte infiltrates of MC38 tumors under therapy with and without cDC1 depletion.

As can be seen in Fig. 4A at two different magnifications, DT depletion caused a dramatic reduction of T cells infiltrating the tumor on day 6 after treatment onset, when mice had received two doses of immunostimulatory mAbs. Interestingly, upon cDC1 depletion, T cells remained on the rim of the invasive tumor margins without penetrating to the center of the engrafted tumor nodules (Fig. 4A). Quantitative analysis and statistical comparisons of the images are provided in Fig. 4B and C, showing that the observed effects regarding T-cell density were less prominent at the tumor margins.

It has been recently reported that during chronic lymphocytic choriomeningitis (LCMV) infection cDC1 are able to support TCF1⁺ CD8⁺ precursor cells (51). The absence of cDC1 cells during the use of PD-1 blockade in the viral model induces a loss of CD8⁺ T cells, with precursor features (TCF1⁺) enforcing their excessive proliferation and leading faster to exhaustion. To test the functional effects of cDC1 absence during immunotherapy in the cancer model setting, we implanted B16.OVA tumors in XCR-1 DTR mice and adoptively transferred activated OVA-specific CD8⁺ CD45.1⁺ OT-I T cells to such mice to henceforth be able to trace antigen-specific CD8 T-cell responses (Fig. 5A). cDC1 depletion did not have any impact on the proportion of transferred OT-I T cells recovered from the tumor draining lymph nodes at the evaluated timepoint (5 days after starting cDC1 depletion, 4 days after immunotherapy onset). However, the recovered OT-I CD8⁺ T cells showed markedly less TCF1 expression, particularly in the absence of cDC1 during immunotherapy and showed upregulated CXCR6 expression as recently described in the LCMV chronic infection model (Fig. 5B; ref. 51). Immunotherapy enhanced the overall number of proliferating OT-I cells, but proliferating cells were TCF1^low, mainly when cDC1 cells were absent. This was in contrast with the nondepleted mice in which proliferating cells were mainly TCF1^hi (Fig. 5C). We also analyzed the endogenous CD8⁺ PD1⁺ cells in the TDLNs (Fig. 5D) and observed very similar results regarding TCF1 expression reduction (Fig. 5E) and higher ki67-denoted proliferation in such TCF1^low CD8⁺ T cell (Fig. 5F). In the tumor compartment, we confirmed the impaired infiltration of transferred CD8⁺ T cells (Fig. 5G) as a result of cDC1 depletion. Moreover, we observed increased expression of the exhaustion marker TOX2 in the absence of cDC1 cells during immunotherapy in OT-I CD8⁺ tumor-infiltrating lymphocytes (Fig. 5H). However, in this intratumoral location, the effects of cDC1 depletion on TCF1 and proliferation were not as evident as in the TDLNs (Fig. 5,I and J). In the endogenous CD8⁺ tumor-infiltrating lymphocytes less tumor infiltration, less exhaustion marker expression levels and similar TCF1 levels were also observed (Fig. 5,K–M).

To test whether these changes in CD8 T cells were observable in another tumor model, we analyzed endogenous CD8⁺ T cells infiltrating MC38 tumors and from TDLNs in the same experimental conditions. The effects observed in the TDLN of such mice (Supplementary Figs. S4A–S4C) were similar to the ones observed in B16.OVA TDLNs, but even more clear, with almost complete loss of TCF1 expression upon cDC1 depletion. In MC38 tumors, infiltrating CD8⁺ T cells showed similar changes following cDC1 depletion, that included upregulation of exhaustion markers, loss of TCF1 expression and enhanced proliferation of the TCF1^low population (Supplementary Figs. S4D–S4F).

To further understand the effects of cDC1 to sustain already activated T cells, we transferred T cells from MC38-bearing mice previously cured with immunotherapy (anti–PD-1+anti-CD137 treated) into Batf3^−/− recipient mice having no cDC1 cells or control WT mice. Transferred T cells were able to more efficaciously control tumors in WT mice than in cDC1-deficient Batf3^−/− mice (Supplementary Fig. S4G).

The selective expression of Venus in XCR1⁺ cDC1 cells in XCR1-DTR-Venus mice offered an opportunity for in vivo time-lapse confocal microscopy to trace their presence and behavior. We chose a model in which mice bear metastatic melanoma lesions in the liver (Fig. 6A) to allow tumor visualization. In this setting, tumor cells are injected into the spleen to disseminate through the portal vein 10 days before imaging. Of note, spleens are removed to avoid unwanted tumor cell engraftment in this organ. Multiple tumor nodules appeared in the liver when a B16.OVA melanoma variant transfected to express the fluorescent protein iRFP670 was intrasplenically injected. Hence, fluorescent tumor cells expressing the surrogate antigen OVA can be visualized. Moreover, XCR1-DTR-Venus mice were adoptively transferred with preactivated TCR-transgenic OT-1 T cells that are cotransgenic for RFP to simultaneously permit visualization and tracking of anti-OVA tumor-specific CD8⁺ T cells (Fig. 6B). The liver central lobe can be surgically exposed and microscopy imaging can be performed for up to 4 or 5 hours. Figure 6B and Supplementary Movie S1 show that cDC1 cells (bright green), OT-I cells (red), and B16.OVA melanoma cells (blue) can be covisualized overtime in this experimental setting. OT-I cells were activated before adoptive transfer to bypass possible effects of cDC1 cells on priming in these experiments.

Focusing on cDC1 cells (Fig. 6C), we were able to trace their behavior and their interactions with OT-I cells. As can be seen in Fig. 6C and Supplementary Movie S2, some cDC1 cells actively moved (termed migrating cells in Fig. 6C) and interacted with T cells (cell #2 in Fig. 6C). First, when these experiments were performed in mice receiving anti–PD-1 plus anti-CD137 mAbs, given 48 hours earlier, more cDC1 cells could be found in the metastatic foci (Fig. 6D). We observed a nonstatistically significant upregulation of CCL5 mRNA within MC38 tumors treated with immunotherapy but no major changes in other DC chemoattracting chemokines as XCL1 or CCL4 were found (Supplementary Fig. S5A) Monitoring cDC1 behavior overtime, we detected that a fraction of cDC1 cells established more durable interactions with OT-I cells in mice receiving immunotherapy as compared with cDC1 cells in mice receiving IgG control antibodies (Fig. 6E). However, the cDC1 cells also tended to move at higher speeds (Fig. 6F) in a direction that seemed purposeful, because the directionality of migration was also significantly higher under immunotherapy with the immunostimulatory antibodies (Fig. 6G). When such experiments were performed in the absence of OT-I (Supplementary Fig. S5B), we observed a similar effect regarding enhanced speed of cDC1 cells upon immunotherapy treatment but without increased directional movement, therefore implying that adoptively transferred OT-I cells may underlie enhanced DC activation in these experiments.

Next, we focused on T-cell behavior comparing conditions in which mice had been depleted of cDC1 cells 72 hours earlier with DT and/or given immunotherapy with anti–PD-1 plus anti-CD137 mAbs. Most importantly, under conditions of immunotherapy, OT-I cells readily infiltrated the tumor, but such an increase was impaired by DT-mediated depletion of cDC1 cells (Fig. 7A and B; Supplementary Movie S3).

We analyzed the behavior of T cells under immunotherapy with and without cDC1 depletion. Previous results using anti-CD137 treatment had shown that OT-I cells upon such immunotherapy stayed focused on targets and showed less overall motility (52). In keeping with these previous findings, OT-I cells under anti–PD-1 plus anti-CD137 immunotherapy showed reduced speed and motility (distance away from the initial location; Fig. 7C and D) and were more focused on the tumor target cells (Supplementary Movie S3). Notably, this focusing effect was lost when cDC1 cells were depleted by DT. Interestingly, CD8⁺ T cells showed no differences in speed or motility in the surrounding nontumor liver tissue (Fig. 7C and D).

Overall, this intravital microscopy experimentation indicates that cDC1 cells are not only important to mediate infiltration of T cells recognizing tumor antigens, but also are able to change the functional T-cell behavior in the tumor tissue microenvironment.

The requirement for cDC1 cells to be present and functional immediately before immunotherapy treatments with immunostimulatory mAbs is clearly demonstrated here. We took advantage of XCR1-DTR-Venus transgenic mice, which had been used before to reveal the role of cDC1 cells in the cross-priming of listeria antigens as well as in LCMV chronic infection (39, 51). DT depletions in the tumor microenvironment and TDLNs were almost complete and lasted for at least 5 days. Nonetheless, our DT dosing scheme was more frequent to ensure complete cDC1 absence. These results are compatible with the lack of efficacy of various immunotherapies in Batf3^−/− mice (17, 18, 53) and make the point that the presence of cDC1 cells at the onset of treatment is a must for immunotherapy to be efficacious. Our experiments do not yet address whether migratory (CD103⁺) or secondary lymphoid organ-resident (CD103⁻) cDC1 cells are those that are necessary and it may well be the case that both dendritic cell subsets play key roles (36).

Our previous experiments in Batf3^−/− mice had shown that these cells were critically needed for antigen-specific CD8 T-cell priming in tumor-bearing mice (18). Importantly, these cDC1 cells are also key to prime against antigen material derived from cytotoxicity as mediated by NK, CTL, and chimeric antigen receptor T cells (21, 54). Therefore, a virtuous circle is postulated according to which the cytotoxic effects of immunotherapy provide more raw material for tumor antigen cross presentation by cDC1 cells. The common feature is that cDC1 cells are continuously needed for immunotherapy agents such as anti–PD-1, anti-CD137, intratumoral poly I:C (53), radioimmunotherapy (55), and adoptive T-cell therapy (19–21). These facts speak of the key importance of the functions mediated by cDC1 cells in oncoimmunology approaches, either with approved or with promising agents under clinical trial development (56).

In keeping with previous observations (30), we observe that cDC1 cells are critical to allow intratumoral T-cell infiltration. Our multiplex immunofluorescence images and time-lapse confocal intravital imaging are conclusive in that regard. The underlying mechanisms remain under investigation and probably involve chemokine loops under the IL12-IFNγ axis as reported previously (30, 31). Observations following OT-I CD8⁺ T-cell transfer are also in agreement with this conclusion. Furthermore, cDC1 seem to be necessary to avoid the expression of exhaustion markers and maintain the markers related to stemness and self-renewal in such TCR transgenic OT-I cells (51).

Intravital imaging of tumor metastases nested in the liver permitted time-lapse imaging of the tumors and the surrounding nontumor liver parenchyma. Three facts became clear in these intravital microsocopy experiments: (i) cDC1 cells change their behavior under immunotherapy conditions within the tumor microenvironment; (ii) CD8⁺ T cells recognizing tumor antigens are attracted and stimulated by cDC1 cells in the baseline tumor microenvironment and more intensely once immunotherapy has been instigated; and (iii) cDC1 cells are required for optimal dynamics and function of CD8⁺ T cells within the tumor microenvironment during immunotherapy. The quantitative parameters of motility, number and duration of cellular interactions in the liver nested tumors are indirect measurements that estimate profound functional changes, even if the molecular mechanisms in action are probably multiple and remain to be fully elucidated. Complex bidirectional functional consequences of cDC1 interactions with cognate T cells are likely to ultimately underlie the efficacy of immunotherapy treatments.

Given the absolute requirement for the presence of cDC1 cells for immunotherapy of established tumors to be efficacious, strategies that enhance their presence and activity in tumors make a great deal of sense in clinical translation. For instance, sFLT3L (28, 29) or local delivery of chemokines attracting cDC1s (42) are strong partner candidates to enhance efficacy of various immunotherapy approaches.

I. Melero reports grants and personal fees from BMS, AstraZeneca, Genmab, Roche-Genentech, Alligator, Bioncotech; personal fees from F-Star, pharmamar, F-Star, Numab, Biolinerx, Dompe, Boston Therapeutics, Third Rock, Servier, and Pierre Fabre outside the submitted work; in addition, I. Melero has a patent for Bioncotech issued and a patent for Genmab pending. No disclosures were reported by the other authors.

A. Teijeira: Conceptualization, resources, data curation, formal analysis, supervision, investigation, methodology, writing–original draft, writing–review and editing. S. Garasa: Data curation, formal analysis, validation, investigation, methodology. C. Luri-Rey: Investigation, methodology, writing–review and editing. C. de Andrea: Data curation, formal analysis, validation, methodology. M. Gato: Formal analysis, investigation, methodology, writing–review and editing. C. Molina: Investigation, methodology, writing–review and editing. T. Kaisho: Resources, writing–review and editing. A. Cirella: Investigation, methodology, writing–review and editing. A. Azpilikueta: Data curation, investigation, methodology, writing–review and editing. S.K. Wculek: Resources, methodology, writing–review and editing. J. Egea: Investigation, methodology, writing–review and editing. I. Olivera: Data curation, investigation, methodology, writing–review and editing. I. Rodriguez: Investigation, writing–review and editing. A. Rouzaut: Resources, formal analysis, writing–review and editing. V. Verkhusha: Resources, writing–review and editing. K. Valencia: Validation, investigation, writing–review and editing. D. Sancho: Conceptualization, resources, writing–review and editing. P. Berraondo: Resources, data curation, methodology, writing–review and editing. I. Melero: Conceptualization, resources, supervision, funding acquisition, writing–original draft, project administration, writing–review and editing.

This work was supported by Spanish Ministry of Economy and Competitiveness and Spanish Ministry of Research [MINECO SAF2014-52361-R and SAF 2017-83267-C2-1R and PID2020-112892RB-100, PID2020-113174-RA-100 (AEI/FEDER,UE), financed by MCIN/AEI/10.13039/501100011033], Cancer Research Institute under the CRI-CLIP, Asociación Española Contra el Cancer (AECC) Foundation under Grant GCB15152947MELE, Joint Translational Call for Proposals 2015 (JTC 2015) TRANSCAN-2 (code: TRS-2016-00000371), projects PI14/01686, PI13/00207, PI16/00668, PI19/01128, funded by Instituto de Salud Carlos III and co-funded by European Union (ERDF, “A way to make Europe”), European Commission within the Horizon 2020 Programme (PROCROP - 635122), Gobierno de Navarra Proyecto LINTERNA Ref: 0011-1411, Mark Foundation, Fundación BBVA and Fundación Olga Torres. A. Teijeira is supported by the Ramon y Cajal program from the Spanish Ministry of Science (RYC2019-026406-I financiada por MCIN/AEI /10.13039/501100011033 y por El FSE invierte en tu futuro).

Esther Guirado and Dr. Belen Palencia are acknowledged for project managing, Dr. Diego Alignani for excellent flow cytometry assistance, the light microscopy unit and Carlos Ortiz de Solórzano for their support in microscopy experiments, and Dr. Paul Miller for English editing.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.