TIM-3 restrains anti-tumour immunity by regulating inflammasome activation

Abstract

T cell immunoglobulin and mucin-containing molecule 3 (TIM-3), first identified as a molecule expressed on interferon-γ producing T cells¹, is emerging as an important immune-checkpoint molecule, with therapeutic blockade of TIM-3 being investigated in multiple human malignancies. Expression of TIM-3 on CD8⁺ T cells in the tumour microenvironment is considered a cardinal sign of T cell dysfunction; however, TIM-3 is also expressed on several other types of immune cell, confounding interpretation of results following blockade using anti-TIM-3 monoclonal antibodies. Here, using conditional knockouts of TIM-3 together with single-cell RNA sequencing, we demonstrate the singular importance of TIM-3 on dendritic cells (DCs), whereby loss of TIM-3 on DCs–but not on CD4⁺ or CD8⁺ T cells–promotes strong anti-tumour immunity. Loss of TIM-3 prevented DCs from expressing a regulatory program and facilitated the maintenance of CD8⁺ effector and stem-like T cells. Conditional deletion of TIM-3 in DCs led to increased accumulation of reactive oxygen species resulting in NLRP3 inflammasome activation. Inhibition of inflammasome activation, or downstream effector cytokines interleukin-1β (IL-1β) and IL-18, completely abrogated the protective anti-tumour immunity observed with TIM-3 deletion in DCs. Together, our findings reveal an important role for TIM-3 in regulating DC function and underscore the potential of TIM-3 blockade in promoting anti-tumour immunity by regulating inflammasome activation.

TIM-3 has emerged as an important checkpoint molecule whose expression correlates with terminal differentiation and exhaustion in tumours^1,2 and chronic viral infection^3,4. A number of clinical trials are under way using blocking monoclonal antibodies directed against TIM-3, however the exact mechanisms underlying the anti-tumour activity of these antibodies are not well understood.

To address the specific function of TIM-3 in T cells, we generated conditional-knockout mice (Extended Data Fig. 1a), deleting TIM-3 in CD4⁺ and CD8⁺ T cells, CD8⁺ T cells or T regulatory (T_reg) cells. Deletion in both CD4⁺ and CD8⁺ T cells led to a modest reduction in tumour burden using immunogenic MC38 colon carcinoma cells expressing ovalbumin (MC38-OVA^dim) (Fig. 1a), whereas deletion in either CD8⁺ T cells (Fig. 1b) or T_reg cells (Fig. 1c) had no effect. TIM-3 is also expressed on natural killer (NK) cells⁵; however, specific deletion using Ncr1^cre had no effect on tumour growth (Fig. 1d). These observations, although unexpected, led us to consider that the role of TIM-3 in regulating anti-tumour immunity may be predominantly mediated through myeloid cells^6,7 and especially DCs, in which TIM-3 is constitutively expressed at high levels^8–10. Recently, the importance of TIM-3 expression on myeloid cells has become apparent from observations made in human disease, where germline mutations in HAVCR2, the gene that encodes TIM-3, are associated with hyperactivation of myeloid cells and elevated circulating levels of IL-18¹¹. Furthermore, the therapeutic efficacy of TIM-3 blocking antibodies in acute myeloid leukaemia (AML) and myelodysplastic syndrome^12–14, suggested that TIM-3 may be particularly important in regulating myeloid cell function.

Fig. 1 |. Deletion of TIM-3 on DCs leads to reduced tumour burden.

a-d, MC38-OVA^dim cells were implanted subcutaneously into Havcr2^fl/fl and Havcr2^fl/flCd4^cre (n = 4) (a), Havcr2^fl/flE8i^cre (n = 5) (b), Havcr2^fl/flFoxp3-ERT2^cre (n = 5) (c) and Havcr2^fl/flNcr1^cre (n = 4) (d) mice and tumour growth was measured over time. e, Day 14 MC38-OVA^dim tumours were explanted. CD45⁺ cells from tumour, draining lymph nodes (dLN) and non-draining lymph nodes (ndLN) were analysed by scRNA-seq. Top, uniform manifold approximation and projection (UMAP) clustering and Havcr2 expression (exp.) in each cluster. Middle, clustering of CD45⁺ cells from wild-type tumours with cell annotation and Havcr2 expression. Bottom, violin plot showing normalized expression of Havcr2 in each CD45⁺ tumour cluster. CMP, common myeloid progenitors; LN, lymph node; mac, macrophages; mono, monocytes; mono-mac, monocyte-derived macrophages; PMN, polymorphonuclear cells. f, g, Growth curve of MC38-OVA^dim cells subcutaneously implanted into Havcr2^fl/fl and Havcr2^fl/flCx3cr1^cre (n = 3) (f) and Havcr2^fl/flCd11c^cre (n = 13) (g) mice. h, KP1.9 cells were injected intravenously into Havcr2^fl/fl and Havcr2^fl/fl Cd11c^cre mice (n = 5). Tumour burden was assessed by histological analyses of explanted lung tissue collected 4 weeks after implantation. i, Wild-type mice were subcutaneously implanted with B16-OVA cells (n = 4). In vitro-derived DC1 cells were generated from Havcr2^fl/fl and Havcr2^fl/flCd11c^cre mice, cultured with soluble OVA and injected subcutaneously after three days of culture. PBS was injected into non-transfer controls. j, k, Tumour growth in Havcr2^fl/fl and Havcr2^fl/flZbtb46^cre mice subcutaneously implanted with MC38-OVA^dim (n = 4) (j) or MC38 (n = 5) (k) cells. l, Growth curve of MC38-OVA^dim cells subcutaneously implanted in Havcr2^fl/fl, Havcr2^fl/flCd4^cre (n = 4) and Havcr2^fl/flCd11c^cre (n = 8) mice. Results are shown from one experiment, representative of at least three independent experiments. Data are mean ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001; two-way ANOVA.

To fully probe the expression profile of TIM-3 we performed single-cell RNA sequencing analysis (scRNA-seq) of tumour-infiltrating CD45⁺ cells (tumour-infiltrating lymphocytes (TILs)), with CD45⁺ cells from the draining and non-draining lymph nodes of wild-type mice bearing MC38-OVA^dim tumours. In lymphoid tissues, Havcr2 expression was most prominent in a cluster of cells expressing canonical DC1 markers including Xcr1 and Clec9a (Extended Data Fig. 1b). However, in tumours, Havcr2 was expressed in multiple cell types including DCs (clusters 1, 5 and 12), CD8⁺ T cells (cluster 7), T_reg cells (cluster 8), monocytes (cluster 3) and macrophages (cluster 0) (Fig. 1e). Flow cytometry analysis of CD45⁺ leukocytes from MC38-OVA^dim verified these expression patterns, with TIM-3 being most highly expressed on DC1s and migratory DCs (migDCs) (Extended Data Fig. 1c, d), a DC population that shares properties of both DC1s and DC2s and has been proposed in several studies to be important in mediating robust anti-tumour immune responses^15,16. We also identified abundant expression of TIM-3 on intratumoural macrophages and monocytes. These data, together with clinical observations made regarding the role of TIM-3 in human disease^11,17, prompted us to systematically Investigate the effect of deletion of TIM-3 in myeloid cells on tumour burden. Despite efficient deletion of TIM-3 in monocytes, macrophages and DC2s (Extended Data Fig. 1e, f), we did not observe any effect on tumour growth using Lysm^cre (Lysm is also known as Lyz2) or Cx3cr1^cre mice (Fig. 1f, Extended Data Fig. 2a), excluding a role for TIM-3 on macrophages, monocytes and DC2s In these models. By contrast, using Cd11c^cre (Cd11c Is also known as Itgax) to delete TIM-3 In DCs (TIM-3 conditional knockout (Havcr2^cko)), we observed significantly reduced tumour growth (Fig. 1g, Extended Data Fig. 2b). We further Investigated the effect of TIM-3 deletion on DCs In a model of non-small-cell lung carcinoma, using KP1.9 lung adenocarcinoma cells¹⁸ and found that Havcr2^cko mice had significantly reduced tumour burden (Fig. 1h). We additionally performed DC-transfer experiments (Fig. 1i) demonstrating that intrinsic loss of TIM-3 on DCs Is sufficient to drive anti-tumour Immunity, with increased frequency of functional antigen-specific effector CD8⁺ T cells (Extended Data Fig. 2c–h). Although we had not observed deletion of TIM-3 on non-myeloid cells using Cd11c^cre, we additionally confirmed our results using a Zbtb46^cre mouse line that specifically targets conventional DCs (cDCs) (Fig. 1j). Using less immunogenic MC38 tumour (Fig. 1k, Extended Data Fig. 2i,j) and B16F10 melanoma (Extended Data Fig. 2k) cells, we again found that loss of TIM-3 on DCs led to significant inhibition of tumour growth. Finally, we compared deletion of TIM-3 on DCs and T cells In parallel, and found that deletion of TIM-3 In DC populations led to superior Inhibition of tumour growth (Fig. 1l; Extended Data Fig. 2l–o).

There were no substantial differences in tumour immune infiltrates between Havcr2^fl/fl and Havcr2^cko mice (Extended Data Fig. 3a–k), except for a significant increase In CD8⁺ T cells in Havcr2^cko mice (Extended Data Fig. 3l). We assessed the expression of cytokine, chemokine, co-inhibitory and co-stimulatory molecules on intratumoural DC1s and migDCs (Extended Data Fig. 3m) but found no substantial changes. Together, these results indicated that TIM-3 has a dominant role In regulating DC function to promote anti-tumour Immunity, prompting us to comprehensively Investigate differences between tumours in Havcr2^fl/fl and Havcr2^cko mice.

TIM-3-deficient DCs expand stem-like CD8⁺ T cells

Using scRNA-seq, we profiled TILs from Havcr2^fl/fl and Havcr2^cko mice bearing MC38-OVA^dim tumours, identifying 15 distinct cell types, Including four lymphoid clusters, a NK cell cluster, seven distinct myeloid clusters and three cDC clusters (Extended Data Fig. 4a–c). All clusters were represented In samples from both Havcr2^fl/fl and Havcr2^cko mice (Extended Data Fig. 4d), but in different proportions. In particular cluster 0 (containing monocyte-derived macrophages; P = 4.74 × 10⁻⁵), clusters 10 and 12 (containing activated B cells; P = 1.16 × 10⁻² and P = 4.16 × 10⁻³, respectively), and cluster 7 (containing CD8⁺ T cells; P = 1.3 × 10⁻⁴), were more highly represented in tumours from Havcr2^cko mice (Fig. 2a, Extended Data Fig. 4e).

Fig. 2 |. Expansion of stem-like memory-precursor CD8⁺ T cells in Havcr2^cko tumours.

MC38-OVA^dim cells were implanted subcutaneously into Havcr2^fl/fl and Havcr2^fl/fl Cd11c^cre (Havcr2^cko) mice and explanted on day 14. CD45⁺ cells were Isolated from tumour for analysis by scRNA-seq. a, Proportions of CD8⁺ T cells in scRNA-seq data. b, Transcripts from CD8⁺ T cells in tumours from Havcr2^fl/fl and Havcr2^cko mice were assessed for enrichment of signatures of early-activated, effector, memory and memory-precursor T cells. c, d, Flow cytometry analysis of TIM-3 and PD1 expression on CD8⁺ TILs from MC38-OVA^dim (n = 10) and B16-OVA tumours (n = 4 or 5) on day 14 after implantation (c) and quantification of CD8⁺ cells expressing TIM-3 and PD1 (d). e-h, Flow cytometry analysis of IL-7R, CD5 and CXCR5 expression in CD8⁺PD1⁺ TILs (e) and proportions of cells expressing IL-7R (f), CD5 (g) and CXCR5 (h). i, TCF1 expression among CD8⁺PD1⁺ TILs. FMO, fluorescence minus one; gMFI, geometric mean fluorescence intensity. j, Representative CX3CR1 expression among CD8⁺PD1⁺ TILs. k, Representative histograms (k) showing expression of T-bet (summarized in l) and Ki67 (summarized in m) among CD8⁺PD1⁺ TILs. Results shown are from one experiment, representative of at least three independent experiments. n = 4 or 5. Student’s two-tailed t-test.

Analysing CD8⁺ T cells (cluster 7), we observed that cells from tumours in Havcr2^cko mice expressed higher levels of gene signatures found in a spectrum of functional states, including early activated¹⁹ (P = 2.6 × 10⁻³), effector²⁰ (P = 2.3 × 10⁻²), memory²⁰ (P = 2.4 × 10⁻⁰³) and memory precursor²¹ (P = 9.9 × 10⁻⁵) (Fig. 2b). Memory precursor cells are ‘stem-like’ CD8⁺ T cells that have been identified in several human cancers^19,22,23 and have been shown to be essential for providing a proliferative burst that supplies the effector T cell pool in response to checkpoint immunotherapy^21,24,25. We hypothesized that the loss of TIM-3 on DCs could contribute to induction or maintenance of stem-like CD8⁺ T cells. Supporting our hypothesis, we observed a significant increase in the frequency and absolute number of PD-1 single-positive TILs in tumours from Havcr2^cko mice (Fig. 2c, d, Extended Data Fig. 5a, b). PD1⁺ TILs have been shown to contain both stem-like and effector-like CD8⁺ TILs²¹; we therefore assessed the expression of cell surface markers that distinguish these two populations^21,26–28. We found increased expression of IL-7R, CD5 and CXCR5 on PD-1⁺ CD8⁺ T cells infiltrating tumours in Havcr2^cko mice (Fig 2e–h), indicating increased numbers of stem-like CD8⁺ TILs. In line with these results, we observed a significant expansion of CXCR5⁺TIM-3⁻CD8⁺ T cells in tumours from Havcr2^cko mice (Extended Data Fig. 5c), together with TCF1 (Fig 2i) which is a hallmark of memory precursor T cells.

We further found increased expression of CX3CR1, T-bet and Ki67 (Fig. 2j–m, Extended Data Fig. 5d), indicating an increase in the number of cells entering the effector lineage. These results indicate that loss of TIM-3 expression on DCs promotes the maintenance of stem-like CD8⁺ T cells and effector T cells, which together promote protective immune responses in tumours from Havcr2^cko mice.

TIM-3 deficiency boosts antigen-specific immunity

Subclustering only the myeloid cells from our single-cell dataset, we identified ten transcriptionally distinct subsets (Fig. 3a, Extended Data Fig. 6a–c), which were represented in tumours from both Havcr2^fl/fl and Havcr2^cko mice, but in different proportions. In particular, cluster 0 (CCR5-expressing monocyte-derived macrophages), was enriched in tumours from Havcr2^cko mice, consistent with increased levels of RANTES in these tumours (data not shown). Conversely, there was a smaller proportion of Havcr2^cko cells in cluster 9 (neutrophils), cluster 6 (DC1s) and cluster 8 (migDCs,) (Fig. 3b, Extended Data Fig. 6d, e).

Fig. 3 |. TIM-3 deficiency promotes DC functionality and enhances antigen-specific anti-tumour immunity.

a, Annotated UMAP plot showing scRNA-seq analysis of myeloid cells from tumours in Havcr2^fl/fl and Havcr2^fl/fl Cd11c^cre (Havcr2^cko) mice. b, Heat map showing expression of selected genes, cluster as identified in a is indicated on the right, the bar on the left indicates Havcr2^fl/fl (grey) or Havcr2^cko (red). c, UMAP showing enrichment of mreg DC signature in cluster 8 (migDCs). d, Empirical cumulative distribution function (ECDF) plot of enrichment score for mregDC signature in Havcr2^fl/fl (grey) or Havcr2^cko (red) migDCs, (P = 9.4 × 10⁻⁴). e, ECDF plot of MHC class I presentation signature (P = 1.01 × 10⁻³). f, Dot plot of selected genes from MHC class I presentation signature. g, Frequency of OVA-specific CD8⁺ T cells from tumours in Havcr2^fl/fl and Havcr2^cko mice at day 14. h-l, MC38-OVA^dim cells were implanted subcutaneously into Havcr2^fl/fl and Havcr2^cko mice. Tumours were collected on day 14 and CD8⁺ TILs were analysed by flow cytometry for expression of IFN-γ and TNF (h), IL-2 (i), granzyme B (j), CD107α (k) and perforin (l). Results shown are representative of at least three independent experiments; n = 4 (h), n = 9 per group (other panels). Student’s two-tailed t-test.

The migDC population has been shown to acquire a regulatory program in the tumour microenvironment (resulting in mature DCs enriched in immunoregulatory molecules (mregDCs)), restraining the immunogenic potential of these cells¹⁶. Scoring this signature on our myeloid cells indeed highlighted cluster 8, migDCs (Fig. 3c) with a lower enrichment score in Havcr2^cko mice (Fig. 3d; P = 9.4 × 10⁻⁰⁴), concomitant with reduced expression of immunoregulatory molecules enriched in mregDCs¹⁶ including IL-4R, CD200, CD83 and OX-40 (Extended Data Fig. 7a). As a key cytokine driving the mregDC program, we investigated the effect of IL-4 blockade in Havcr2^cko tumour-bearing mice (Extended Data Fig. 7b). Treatment with anti-IL-4 led to a profound delay in tumour growth in Havcr2^fl/fl mice but had no effect in Havcr2^cko mice. This suggests that deficiency of TIM-3 might curtail acquisition of the mregDC program, enabling migDCs to attain a full immunogenic program.

In the context of tumour, a critical function of DCs^15,29,30, is to promote and educate antigen-specific CD8⁺ T cells for durable anti-tumour immunity. Scoring the myeloid clusters for a major histocompatibility complex (MHC) class I presentation signature, we found significant enrichment in Havcr2^cko cluster 8 migDCs (Fig. 3e; P = 1.01 × 10⁻³), in which several genes related to MHC class I antigen presentation, including Wdfy4, Cybb, Tapbpl and H2-M2 were upregulated (Fig. 3f). To investigate the role of TIM-3 in cross-presentation we performed in vitro assays using particulate antigen. OT-I T cells cultured with dead-cell-associated or bead-coupled OVA had a higher rate of proliferation and activation (CD44⁺CTV^lo) when cultured with Havcr2^cko DC1s (Extended Data Fig. 7c). In line with this result, we found an increased frequency of antigen-specific CD8⁺ T cells in tumours in Havcr2^cko mice (Fig. 3g), producing increased levels of effector cytokines interferon-γ (IFN-γ), TNF and IL-2 (Fig. 3h, i) with greater cytotoxic potential, assessed by increased expression of granzyme B, CD107α and perforin (Fig. 3j–l). This resulted in increased lytic activity of CD8⁺ T cells in vivo (Extended Data Fig. 7d).

Loss of TIM-3 increases inflammasome activation

To identify potential cell-cell interactions between CD8⁺ T cells (expanded and activated in Havcr2^cko mice) and migDCs (with enhanced antigen-presentation) in Havcr2^cko tumours, we used a method recently developed³¹ as part of the Waddington OT package. Although multiple ligand-receptor gene pairs scored similarly between tumours in Havcr2^fl/fl and Havcr2^cko mice, (for example, Il12b-Il12rb2), several putative ligand-receptor gene pairs had a higher interaction score in tumours from Havcr2^cko mice, notably Il18-Il18r1 and Il18-Il18rap (Fig. 4a). In line with this finding, expression of Il18rap was significantly higher in CD8⁺ T cells from tumours in Havcr2^cko mice (Extended Data Fig. 7e, f). Although Il1r1 transcript was undetectable in our dataset, we scored CD8⁺ T cells from tumours in Havcr2^fl/fl and Havcr2^cko mice for enrichment of Il18 and Il1 gene signatures, finding that both were enriched in CD8⁺ T cells from tumours in Havcr2^cko mice (Fig. 4b, c; P = 2.8 × 10⁻⁴ and 7.02 × 10⁻⁴, respectively). Cytokine analysis of homogenates of tumours from Havcr2^fl/fl and Havcr2^cko mice also showed higher levels of IL-1β and IL-18 (Extended Data Fig. 7g).

Fig. 4 |. Loss of TIM-3 on DCs promotes inflammasome activation.

a, Unidirectional analysis of the fraction of DCs expressing a ligand and the fraction of T cells expressing the cognate receptor; ligand (migDCs)-receptor (CD8) interaction from tumours in Havcr2^fl/fl (left) and Havcr2^cko (right) mice. b, c, ECDF plots showing enrichment of IL-18 (b) and IL-1 (c) signatures in Havcr2^fl/fl (grey) and Havcr2^cko (red) CD8⁺ TILs. d, e, ECDF plot of enrichment score for inflammasome signature (d; P = 1.5 × 10⁻³) and dot plot for selected genes from inflammasome signature (e) in Havcr2^fl/fl (grey) and Havcr2^cko (red) migratory DCs. f, g, ECDF plot of enrichment score for oxidative stress signature (f) and dot plot of selected genes from oxidative stress signature (g) in Havcr2^fl/fl (grey) and Havcr2^cko (red) migratory DCs. h, i, MC38-OVA^dim cells implanted subcutaneously in Havcr2^fl/fl and Havcr2^cko mice treated with N-acetylcysteine (NAC) (n = 4 or 5). h, Tumour size. i, Tumour supernatant was collected, and levels of IL-1β were determined by ELISA. Data are mean ± s.e.m. Student’s two-tailed t-test. j-l, Tumour growth curve of MC38-OVA^dim cells implanted subcutaneously in Havcr2^fl/fl and Havcr2^cko mice treated with vehicle control or caspase 1 inhibitor (Casp1) (j; n = 4), MCC950 (k; n = 5) or anti-IL-1β and anti-IL-18 (l; n = 4 or 5). m-q, B16-OVA cells were implanted subcutaneously in Havcr2^fl/fl and Havcr2^cko mice treated with anti-IL-1β and anti-IL-18 (n = 5). Flow cytometry analysis of CD8⁺ TILs (m) from Havcr2^fl/fl and Havcr2^cko mice bearing B16-OVA tumours and treated with isotype control or anti-IL-1β and anti-IL-18. Expression of TCF1+ (n), Ki67 (o), T-bet (p) and IFN-γ (q). Results shown are from one experiment, representative of at least two independent experiments; n = 4 or 5 per group; h, j-l, two-way ANOVA; i, m-q, one-way ANOVA.

IL-1β and IL-18 are both tightly regulated cytokines that are released as mature proteins in a process mediated by inflammasomes^32,33. Investigation of genes that were differentially expressed in Havcr2^fl/fl and Havcr2^cko mice, we found significant enrichment for an inflammasome gene signature in Havcr2^cko migDCs (Fig. 4d; P = 1.5 × 10⁻³) including adaptor molecules involved upstream (Aim2 and Nlrp3) and downstream (Casp1, Casp4 and Il18) of inflammasome activation (Fig. 4e). Furthermore, agonists of NLRP3, AIM2 and NLRC4 substantially increased levels of IL-1β in supernatants from Havcr2^cko bone-marrow-derived DC1s, whereas levels of TNF were only moderately increased (Extended Data Fig. 8a–c).

We hypothesized that endogenous damage-associated molecular patterns, including reactive oxygen species, oxidized lipids and oxidized DNA^34–37, might be driving inflammasome activation in tumour-infiltrating Havcr2^cko DCs. Indeed, Havcr2^cko migDCs scored more highly on a gene signature associated with oxidative stress (Fig. 4f; P = 1.9 × 10⁻⁶), with upregulation of the key enzymes SOD1, SOD2, BAG3 and PRDX5^38,39 (Fig. 4g). We investigated the reactive oxygen species activity of DCs from MC38-OVA^dim tumours, and found significantly increased activity in Havcr2^cko DC1s and migDCs (Extended Data Fig. 8d). Furthermore, treatment with antioxidant N-acetylcysteine completely reversed the protective inhibition of tumour growth in Havcr2^cko mice (Fig. 4h). Mitigation of oxidative stress in tumours in Havcr2^cko mice also completely abrogated tumour IL-1β levels (Fig. 4i).

Finally, we used three different approaches to target the inflammasome pathway in vivo: (1) a caspase 1 inhibitor to interfere with pro-IL1β and pro-IL-18 cleavage (Fig. 4j); (2) MCC950 to perturb oligomerisation of the inflammasome complex protein ASC⁴⁰ (Fig. 4k); and (3) anti-IL-1β and anti-IL-18 blocking antibodies (anti-IL-1β/IL-18) to inhibit the downstream effector cytokines (Fig. 4l). Each of these approaches abolished the protective anti-tumour immunity in Havcr2^cko, with anti-IL-1β/IL-18 having the most potent effect (Fig. 4l). Furthermore, flow cytometry analysis of the treated tumours from Havcr2^cko mice showed complete reversal of the CD8⁺ phenotype (Fig. 4m–q, Extended Data Fig. 8e–k). Together with recent findings, these data highlight the role that innate cytokines such as IL-1 and IL-18 can have in curtailing an exhaustion trajectory, sustaining effector T cell responses for anti-tumour immunity. It is noteworthy that TIM-3 has also been shown to negatively regulate signalling through the IL-1 and IL-18 receptors by interacting with TIR8⁴¹. How this interaction may function in DCs remains unknown, but could potentially implicate an autocrine feed-forward loop where loss of TIM-3 on DCs leads to more IL-1β production and also heightened responses to IL-1β, together promoting immunogenic DCs.

Previous preclinical studies have observed high expression of TIM-3 on intratumoural DCs and concluded that the therapeutic efficacy of anti-TIM-3 was probably mediated by DC1s⁹. However, TIM-3 is expressed by multiple cell types, confounding the interpretation of these studies. We implanted B16-OVA tumour cells in Havcr2^fl/fl, Havcr2^fl/fl Cd4^cre and Havcr2^fl/fl Zbtb46^cre mice and treated the mice therapeutically with anti-TIM-3 and anti-PDL1. Havcr2^fl/fl recipients showed a modest reduction in tumour burden with either anti-TIM-3 or anti-PDL1, with a further reduction using combined blockade (Extended Data Fig. 8l). Whereas the additive effect of anti-TIM-3 and anti-PDL1 was also observed in Havcr2^fl/fl Cd4^cre tumour-bearing mice, (Extended Data Fig. 8m), it was completely lost in the Havcr2^fl/fl Zbtb46^cre mice (Extended Data Fig. 8n), suggesting that anti-TIM-3 works primarily on DCs to induce anti-tumour immunity. Of note, TIM-3 has been identified as a marker for leukaemic stem cells in AML¹², with particularly high expression in patients with refractory AML¹³. Therefore, TIM-3 has become an attractive therapeutic target, and emerging data indicate that TIM-3 blockade in AML and myelodysplastic syndrome induces 50–60% response rates¹⁴, suggesting TIM-3 blockade as a therapeutic strategy for myeloid tumours where other checkpoint inhibitors have shown limited success^42–44.

Together, we show that loss of TIM-3 prevents migDCs from acquiring a regulatory program, facilitating the maintenance of the CD8⁺ effector T cell pool driven by IL-1β and IL-18. These results build on growing interest in stimulating inflammasome pathways for cancer treatment⁴⁵, and underscore the value of TIM-3 blockade in liberating inflammasome activity to potentiate anti-tumour immunity.

Methods

No statistical methods were used to predetermine sample size. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment.

Mice

Six-to eight-week-old C57BL/6, Cd11c^cre, Zbtb46^cre, Cd4^cre, E8i^cre, Lysm^cre, Cx3cr1^cre and Foxp3-ERT2^cre mice were purchased from the Jackson Laboratory. Havcr2^fl/fl mice were generated as described in supplementary materials. TIM-3 conditional knockout mice were generated by crossing to the above cre lines. For knockin/knockout alleles (Lysm^cre) the appropriate controls were used; that is, Havcr2^fl/+ × Lysm^cre+/−, for all other cre lines fl/fl mice were used as controls. For experiments with Foxp3-ERT2^cre mice, mice were orally gavaged with 8 mg tamoxifen 3 days before tumour implantation and every 3 days thereafter for the duration of the experiments. Havcr2^+/+ Foxp3-ERT2^cre were used as an additional wild-type control but results were comparable to Havcr2^fl/fl and were therefore not included. Deletion efficiency was determined by flow cytometry (not shown). Animal experiments were done in accordance with the guidelines of the institutional Animal Care and Use Committee (IACUC) at Brigham and Women’s Hospital and Harvard Medical School. All animals were euthanized before reaching humane endpoint, with tumour growth no greater than 2 cm in any one direction or of a total of 400 mm² overall. Mice of both sexes were used throughout the study; sex-matched and age-matched (8–12 weeks) controls were used in individual experiments.

Gene-targeted deletion of mouse Havcr2 locus

To generate TIM-3 floxed mice on the C57BL/6 background, targeting vectors containing genomic fragments of the Havcr2 gene were constructed by using C57BL/6 BAC clones. Linearized targeting vector was transfected into B6 embryonic stem (ES) cells. Homologous recombinants were identified by Southern blot analysis and implanted into foster B6 albino mothers. Chimeric mice were bred to C57BL/6 mice, and the F₁ generation was screened for germline transmission. The Neo gene was removed by breeding F₁ mice with a strain of actin promoter-driven flipase transgenic mice (Jackson Laboratory) (Extended Data Fig. 1).

Tumour induction

The C57BL/6-derived colon carcinoma MC38 and MC38-OVA^dim cells were maintained at 37 °C with 5% CO₂ in DMEM supplemented with 10% heat-inactivated fetal calf serum (FCS), penicillin and streptomycin and 1% sodium pyruvate. The C57BL/6-derived melanoma cell lines B16- and B16-OVA-transfected clones were maintained at 37 °C with 5% CO₂ in RPMI medium supplemented with 10% heat-inactivated fetal calf serum, penicillin and streptomycin. All cell lines used in this study tested negative for murine pathogens. MC38-OVA^dim (0.5 × 10⁶), B16F10 or B16-OVA (0.25 × 10⁶) and MC38 (1.0 × 10⁶) cells were injected subcutaneously in 100 μl PBS on the flank. Tumour size was determined by the formula L × W, where L is length and W is width. In some experiments, mice were treated with anti-IL-1β (B122) and anti-IL-18 (YIGIF74-1G7) or appropriate control hamster immunoglobulin and rat IgG2a (all at 8 mg kg⁻¹) intraperitoneally on days 3, 6 and 9 after tumour implantations. For immune-checkpoint experiments anti-TIM-3 was administered at a dose of 8 mg kg⁻¹ and anti-PDL1 at a dose of 2 mg kg⁻¹. Isotype controls were used at the same concentration in control mice. Neutralizing anti-IL-4 was used at 1.25 mg kg⁻¹, anti-IL-12 at 25 mg kg⁻¹. For MCC950 experiments, mice were treated with MCC950 (R&D Systems) at a dose of 8 mg kg⁻¹ or PBS as control intraperitoneally on days 0, 2, 4, 6, 8 and 10 until the end of the experiment. For caspase 1 inhibitor (Invivogen) experiments, mice were orally gavaged with 8 mg kg⁻¹ inhibitor or control (water) at on days 0, 2, 4, 6, 8 and 10 until the end of the experiment. For N-acetylcysteine experiments mice were given N-acetylcysteine daily in drinking water at a concentration of 40 mM.

Lung tumour

Lung tumour experiments were conducted using orthotopic syngeneic KP1.9 lung adenocarcinoma cells. These cells are derived from lung tumours of C57BL/6 KP mice harbouring Kras and Trp53 mutations⁴⁶. Tumour cells were injected intravenously (2.5 × 10⁵ cells in 100 μl PBS) to develop orthotopic tumours. Animals reproducibly show macroscopic lung tumour nodules at 3–4 weeks post injection. Evaluation of lung tumour burden was assessed by histological analyses based on haematoxylin and eosin (H&E) staining of explanted lung tissue collected 4 weeks post implantation. Tumour area was calculated as a percentage of area occupied by the tumour as a part of total lung tissue.

Preparation of cell suspensions

Single-cell suspension was obtained after tumour digestion with 20 μg ml⁻¹ Liberase (Roche) and 20 ng ml⁻¹ DNase I (Roche) at 37 °C for 30 min. Haematopoietic cells were enriched by density gradient centrifugation with 30% Percoll (GE Healthcare Life Sciences) for 20 min at 1,800 rpm (no brake) followed by passing through a 40-μm cell strainer. DCs were obtained from LNs after 20 μg ml⁻¹ Liberase (Roche) and 20 ng ml⁻¹ DNase I (Roche) at 37 °C for 30 min.

Flow cytometry

Single cell suspensions were incubated with anti-CD16/32 (BioXcell) for 15 min at 4 °C, followed by staining with fluorescently conjugated antibodies for 30–45 min at 4 °C.

The following cell populations were identified based on cell marker expression: neutrophils (CD45⁺CD11b⁺Ly6C⁻Ly6G⁺), monocytes (CD45⁺ CD11b⁺Ly-6G⁻Ly-6C^high), CD11b⁺ macrophage (CD45⁺CD11b⁺Ly6G⁻Ly6C⁻F4/80⁺), T cells (CD45⁺CD3⁺CD4⁺ or CD8⁺), B cells (CD45⁺B220⁺CD19⁺), NK cells (CD45⁺CD3⁻NK1.1⁺), DC1 (CD45⁺CD3⁻CD19⁻NK1.1⁻, MHCII⁺CD11c⁺ CD64⁻CD11b⁻CD103⁺ (tumour) XCR1⁺ (lymphoid tissue)), DC2 (CD45⁺ CD3⁻CD19⁻NK1.1⁻MHCII⁺CD11c⁺CD64⁻CD11b⁺CD103/XCR1⁻), migDC DC1 (CD45⁺CD3⁻CD19⁻NK1.1⁻MHCII⁺CD11c⁺CD64⁻CD11b⁺CD103⁺).

The following antibodies (Biolegend and BD) were used: CD45 (30-F11), CD4 (RM4-5), CD8a (53-6.7), B220 (RA3-6B2), NK1.1 (PK136), CD11b (M1/70), CD11c (N418), CD103 (P84), XCR1 (ZET), CD64 (X54-5/7.1), CD86 (GL-1), PDL1 (10F.9G2), CD27 (LG.3A10), IL-7Ra (A7R34), CXCR5 (L138D7), CX3CR1 (SA011F11), CD5 (53-7.3), CD24 (M1/69), PDL2 (122), Lag3 (C9B7W), TIGIT (GIGD7), TIM-3 (5D12), Ly6C (HK1.4), Ly6G (1A8), TCRb (H57-597), F4/80 (BM8), Ki67 (16A8), T-bet (4B10) and TCF1 (C63D9).

Re-stimulation and intracellular cytokine staining

For intra-cellular cytokine (ICC) staining cells were stimulated with phorbol-12-myristate 13-acetate (PMA) (50 ng ml⁻¹, Sigma-Aldrich) and ionomycin (1 μg ml⁻¹, Sigma-Aldrich) in the presence of Golgi Plug (BD Biosciences) and Golgi Stop (BD Biosciences) for 4 h before cell surface and ICC staining. Antibody detecting CD107α (1D4B) was added to the cells during cell culture incubation. Following fixation and permeabilization, staining with antibodies against the following was performed with the following antibodies purchased from Biolegend or BD Biosciences: IL-2 (JES6-5H4), TNF (MP6-XT22), IFN-γ (XMG-1.2), granzyme B (GB11), IL-10 (JES5-16E3), Perforin (eBioOMAK-D), IL-12p40 (C17.8), IL-27p28 (MM27-7B1), LAP (TW7-20B9), CXCL9 (MIG-2F5.5). Antigen specific T cells were determined by H-2Kb/OVA_257-264 dextramer staining following the manufacturer’s protocol (Immudex). All data were collected on a BD Fortessa or Symphony (BD Biosciences) and analysed with FlowJo 10.4.2 software (TreeStar).

In vitro bone-marrow-derived DC cultures and DC transfer

Bone marrow cells were isolated by flushing femurs, tibias and humeri with PBS, supplemented with 0.5% BSA, 2 mM EDTA, and 1% penicillin/streptomycin (P/S). Bone marrow cells were strained through a 70-μm filter and centrifuged before resuspension in 1 × Ack lysis buffer (Gibco) for 5 min on ice. For generation of DC2 cells were plated in RPMI medium with 10% FCS, 1% L-glutamine, 1% sodium pyruvate, 1% MEM non-essential amino acids, 1% P/S, 55 μM 2-mercaptoethanol 10 ng ml⁻¹ GMCSF (Peprotech) and 5 ng ml⁻¹ IL-4 (R&D). DC1 cells were plated as above but with 200 ng ml⁻¹ recombinant Flt-3 ligand (Biolegend). DCs were analysed on days 7-9. For transfer experiments, cells were sorted for expression of XCR1/CD103 (DC1) or Sirpα (DC2). On day 3 after tumour implantation DCs were transferred (0.5 × 10⁶ per mouse) in 50 μl PBS subcutaneously adjacent to the tumour. For sham control mice PBS alone was injected.

In vitro cross-presentation assays

Splenic DC1 s were cultured with Cell Trace Violet (CTV) labelled OT-I cells together with (1) HLA mismatched apoptotic splenocytes osmotically loaded with ovalbumin, or (2) latex beads passively coupled with OVA. Non-OVA loaded dead cells or non-OVA coupled beads served as control, respectively. After 72 h co-culture the proliferation of CD8⁺ T cells was assessed by flowcytometry where expression of CD44 together with dilution of CTV dye was indicative of activation and proliferation.

In vivo killing assay

On day 12 of tumour growth mice were injected intravenously with splenocytes differentially labelled with CFSE or CTV (Invitrogen) and loaded with 100 nM OVA_257-264 or MOG_37-46 (Irrelevant Antigen). Cells were injected at 50:50 ratio into MC38-OVA^dim bearing Havcr2^fl/fl or Haver2^cko mice respectively and collected 24 h after injection. Percentage cytotoxicity calculated as 100 − ([CTV]/[CTV] + [CFSE]).

ROS activity assay

Single cell suspensions were generated as described and cells were resuspended in HBSS containing 0.1% BSA followed by FACS antibody surface marker staining for 30 min on ice. Cells were then washed and resuspended in PBS-EGG buffer (1 mM EDTA, 0.05% gelatine, 0.09% glucose) and 5.0 μM DHR123 probe (Thermo Fisher Scientific) was added for 30 min at 37 °C. The reaction was stopped by placing the cells on ice and washing with PBS-EGG buffer. Cells were resuspended in PBS containing 0.1% BSA and activated rhodamine 123 signal (activated DHR 123) was analysed in the FITC channel on a BD Symphony within 30 min. Hydrogen peroxide was added to some samples to serve as a positive control.

Inflammasome activation

Bone-marrow-derived DC1s were generated as described above. XCR1⁺ cells were sorted after 10 days of differentiation and seeded at a density of 0.5 × 10⁶. Sorted cells were either unstimulated or primed with LPS for 3 h followed by the addition of oxidised phospholipids (ox-PAPC) (100 μg ml⁻¹; Invivogen), pdA:dT (1 μg ml⁻¹; Invivogen), flagellin (1 μg ml⁻¹; Invivogen), Clostridium difficile (1 μg ml⁻¹; List Biologicals) or nigericin (10 mM; Invivogen). Following overnight cultures supernatants were collected and ELISA was performed to detect IL-1β and TNF (non-inflammasome regulated control).

Statistical analysis

Graphs were compiled and statistical analyses were performed with Prism software (GraphPad). Statistical significance was evaluated with the two-tailed unpaired t -test when comparing two groups and two-way ANOVA when comparing more than two groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; NS, not significant.

Multiplexing samples by cell-hashing

After flow sorting, cells from each sample (tumour or dLN or ndLN from an individual mouse) was resuspended in PBS containing 2% FCS and stained with oligo tagged TotalSeq antibodies directed against CD45 for 30 min on ice. Cells were washed and pooled accordingly, centrifuged at 1,200 rcf for 5 min at 4 °C and resuspended in PBS + 2% FCS. For each 10X channel, 6 samples were combined: Havcr2^fl/fl 1-Tumour, dLN, ndLN; KO1-Tumour, dLN, ndLN; Havcr2^fl/fl 2-Tumour, dLN, ndLN; KO2- Tumour, dLN, ndLN.

scRNA-seq library preparation

Droplet-based 3′ end massively parallel single-cell RNA sequencing (scRNA-seq) was performed by encapsulating sorted live CD45⁺ tumour-infiltrating cells into droplets and libraries were prepared using Chromium Single Cell 5′ Reagent Kits v2 according to the manufacturer’s protocol (10X Genomics). The generated scRNA-seq libraries were sequenced using an Illumina HiSeq2500 to an average depth of 60.4 million paired end reads per sample.

scRNA-seq data analysis

We conducted all the pre-processing steps: generating, demultiplexing FASTQ files from the raw sequencing reads (bclfastq, v2.20), aligning to UCSC mm10 mouse transcriptome (cellranger, v3.0.2, 10X Genomics), demultiplexing cell-hashing data, and generating of raw gene count matrices using Cumulus (v0.8.0), a cloud based data analysis framework⁴⁷. The pipelines were run in Terra Cloud Platform⁴⁷ (http://app.terra.bio).

Downstream analysis was performed using the R software package Seurat⁴⁸ (v2.3.4; http://satijalab.org/seurat/). We removed low-quality cells with less than 500 genes detected and removed genes expressed in fewer than 50 cells. The filtered raw gene expression matrix was cell-normalized over total number of counts, multiplied by 10,000 and log transformed. Principal component analysis (PCA) was performed on a submatrix of the top 1,000 most variable genes (computed using FindVariableGenes from Seurat), and the number of top principal components for further analysis was determined by the elbow method, keeping 12 and 15 principal components for analysis of tumour only or all cells (plus dLN, ndLN). We clustered cells used a shared nearest neighbour (SNN) modularity optimization-based clustering algorithm as implemented in a function FindClusters from the Seurat package. We computed cluster-specific differentially expressed genes using function FindAllMarkers from Seurat with default test parameters. Gene expression values, gene signature scores, and clustering results were visualized by embedding cells in a UMAP⁴⁹ of the same dimensionally-reduced space as in clustering of cells. Annotating cell clusters was performed by assessing known cell-type markers and comparison with public myeloid tumour datasets^16,50,51.

Creating gene signatures

Curated gene signatures were constructed from various databases of gene signatures as summarized in Supplementary Table 1. IL-1 signature was determined from differential genes between IL-1b-treated and vehicle CD8⁺ T cells⁵². The expression matrix was created from data downloaded from GEO database, accession number GSE127234. We used edgeR⁵³ to compute the differentially expressed genes, and the IL-1 signature was created from top genes upregulated in IL-1b-treated cells (log(fold change) > 3, false discovery rate < 0.1).

Given a set of genes, we computed signature score activity levels using the AddModuleScore function in Seurat with the parameters defining the control gene-sets specified by n.bin = 25 and ctrl.size = 100. The difference in gene signature score activity levels between Havcr2^fl/fl and Havcr2^cko was examined using Wilcoxon rank-sum test and visualized using violin or ECDF plots.

Determining potential receptor-ligand Interactions

To characterize potential paracrine signalling between migDCs and CD8⁺ T cells, we used the method proposed by Schiebinger et al.³¹. In particular, we defined a list of ligands from the following GO terms: cytokine activity (GO:0005125), growth factor activity (GO:0008083), and hormone activity (GO:0005179). The set of receptors was defined by the GO term receptor activity (GO:0004872). We determined the set of potential ligand-receptor pairs using a curated database of mouse protein-protein interactions⁵⁴. We quantified the potential ligand-receptor interactions by an interaction score defined as the product of the fraction of cells in the cell cluster expressing ligand and the fraction of cells in the cell cluster expressing the cognate receptor. To remove ubiquitously expressed ligands and receptors, we used 10,000 permutations to generate empirical null distribution of interaction scores by randomly shuffling the cell labels. Then we standardized each ligand-receptor interaction score by computing the distance between the interaction score and the mean interaction score in units of standard deviations from the permuted data as described³¹.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.

Data availability

Data have been uploaded to NCBI Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/) under data repository accession number GSE151914. Source data are provided with this paper.

Extended Data

Extended Data Fig. 1 |. Generation of conditional knockout mice for TIM-3.

a, Strategy for generating the Havcr2-targeting vector to target the Havcr2 allele. Blue boxes represent exons (E). The 5′ external probes for Southern Blot are indicated by thick red line. Targeted events were identified by Southern blot analysis of Afl2-digested genomic ES cell DNAs with the 5′ flanking probe as shown in A. b, MC38-OVA^dim (0.5 × 10⁶ cells) were subcutaneously implanted into Havcr2^fl/fl and Havcr2^fl/fl Cd11c^Cre mice. On D14 dLNs were explanted followed by cell sorting for sc-RNA-seq of CD45+ cells. UMAPs of canonical cDC1 markers Xcr1, Clec9a and Flt3, (bottom) UMAP of Havcr2 expression among clusters found in dLN, violin plot from scRNA-seq displaying normalized expression of Havcr2 in each cluster. c, d, WT mice were implanted with MC38 cells (1.0 × 10⁶). On D21 tumours were explanted followed by flow cytometric analysis of TIM-3 (gMFI) on tumour infiltrating immune cells (n = 3-4). e, MC38-OVA^dim (0.5 × 10⁶ cells) tumour cells were subcutaneously implanted into Havcr2^fl/fl, Havcr2^fl/fl LysM^Cre (n = 3), Havcr2^fl/flCx3cr1^Cre (n = 3) and Havcr2^fl/flZbtb46^Cre (n = 4) animals. Representative Flow cytometric analysis of TIM-3 expression on DC1, DC2, migDCs, macrophages and monocytes. DCs were gated as in Ext Fig. 2: CD45⁺, CD3⁻CD19⁻NK1.1⁻, ClassII⁺CD11c⁺, Ly6c⁻CD64⁻ and DC1: CD103⁺CD11b⁻, DC2: CD11b⁺ CD103⁻ migDCs CD11b⁺CD103⁺. Macrophages: CD45⁺, CD3⁻CD19⁻NK1.1⁻, ClassII^lo Ly6c^loCD64⁺F480⁺CX3CR1⁺, monocytes ClassII^lo Ly6c^hiCD64^loLy6g⁻. (right) Percentage expression and gMFI of TIM-3. f, MC38-OVA^dim (0.5 × 10⁶ cells) were subcutaneously implanted into Havcr2^fl/fl (n = 5), Havcr2^fl/flCd4^Cre (n = 3), Havcr2^fl/flCd11c^Cre (n = 5) and Havcr2^fl/fl Zbtb46^Cre (n = 4) mice. On D14 tumours were explanted followed by flow cytometric analysis of TIM-3 expression on CD4 TILs, CD8 TILs and tumour infiltrating DC1 from Havcr2^fl/fl, Havcr2^fl/fl × CD4^Cre, Havcr2^fl/fl × CD11c^Cre and Havcr2^fl/fl × Zbtb46^Cre. The results shown are from one experiment, representative of at least 3 independent experiments. ***P < 0.001; ****P < 0.0001 (One-Way ANOVA). Data shown (f) as mean ± s.e.m. *P < 0.05; ****P < 0.0001 (Student Two-Tailed t-test).

Extended Data Fig. 2 |. Deletion of TIM-3 in cDC using Zbtb46 recapitulates findings using CD11c cre.

a, Tumour growth curve of MC38-OVA^dim subcutaneously implanted into Havcr2^fl/fl and Havcr2^fl/flLysM^Cre (n = 6). b, Tumour growth curve MC38-OVA^dim OVA subcutaneously implanted Havcr2^fl/+ × CD11c^Cre and Havcr2^fl/fl × CD11c^Cre (n = 5). c, Tumour of Havcr2^fl/fl mice implanted with B16-OVA. On D3 XCR1+ BMDC1 were sorted, pulsed with OVA and injected into tumour bearing mice. d, Flow-cytometric analysis of frequency (n = 9), and absolute number (n = 4) of OVA specific CD8⁺ T cells from tumours injected with Havcr2^fl/fl or Havcr2^cko DC1. e-h, Flow-cytometric analysis of OVA specific CD8⁺ T cells from tumours injected with Havcr2^fl/fl or Havcr2^cko DC1 (n = 4). i, MC38-OVA^dim (0.5 × 10⁶ cells) tumour cells were subcutaneously implanted into Havcr2^fl/fl and Havcr2^fl/fl × Zbtb46^Cre animals (n = 4-5). Flow cytometric analysis (d14) of TIM-3 expression on tumour infiltrating DC1, DC2, migDCs and pDC from Havcr2^fl/fl and Havcr2^fl/fl × Zbtb46^Cre. j, Tumour weight and total CD45+ cells of MC38 subcutaneously implanted Havcr2^fl/fl and Havcr2^fl/fl × Zbtb46^Cre (n = 5). k, Tumour growth curve of B16 subcutaneously implanted Havcr2^fl/fl and Havcr2^fl/fl × Zbtb46^Cre (n = 9). l, m, Tumour growth curve of B16F10 melanoma (l) and B16-OVA (m) subcutaneously implanted Havcr2^fl/fl, Havcr2^fl/fl × CD4^Cre and Havcr2^fl/fl × CD11c^Cre cre in parallel (n = 4-5). n, Weights of tumours from (Fig. 1l). o, B16-OVA subcutaneously implanted Havcr2^fl/fl, Havcr2^fl/fl × CD4^Cre and Havcr2^fl/fl × Zbtb46^Cre in parallel (n = 4). The results shown are from one experiment, representative of at least three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (Two-Way ANOVA). Data shown (d-h) as mean ± s.e.m. *P < 0.05; **P < 0.01 (Student Two-Tailed t-test).

Extended Data Fig. 3 |. Deficiency of TIM-3 on DC leads to increased numbers of tumour infiltrating CD8⁺ T cells.

MC38-OVA^dim (0.5 × 10⁶ cells) tumour cells were subcutaneously implanted into Havcr2^fl/fl and isolated at D14. a, Gating strategy and phenotype of intratumoral myeloid cells. b-l, Flow cytometric quantification of immune cells in tumours from Havcr2^fl/fl and Havcr2^fl/fl × CD11c^Cre mice at d14 harvest. m, Flow cytometric analysis of DC1 and Mig DC from Havcr2^fl/fl and Havcr2^cko tumours at d14 harvest following in vitro stimulation for 4 h in the presence of Brefeldin A and Monensin. Data shown (l) as mean ± s.e.m. *P < 0.05 (Student Two-Tailed t-test) n = 5-9/group.

Extended Data Fig. 4 |. scRNA-seq of Havcr2^fl/fl and Havcr2^cko total CD45⁺ cells.

a, UMAP scRNA-seq plot of annotated total cells from Havcr2^fl/fl and Havcr2^fl/fl × CD11c^Cre (Havcr2^cko) tumours. b, UMAP scRNA-seq plots showing select marker gene expression. c, Heat map from scRNA-seq displaying normalized expression of select genes in each cluster. d, UMAP scRNA-seq plot showing distribution of Havcr2^fl/fl (blue) and Havcr2^cko (orange) cells. e, Bar graph showing frequency of Havcr2^fl/fl (blue) and Havcr2^cko (orange) cells in each cluster.

Extended Data Fig. 5 |. Expansion of CD8⁺ PD1⁺ cells in Havcr2^cko tumours.

MC38-OVA^dim (0.5 × 10⁶ cells) were subcutaneously implanted into Havcr2^fl/fl and Havcr2^fl/fl × CD11c^Cre (Havcr2^cko) mice and harvested on D14. a, Frequency (n = 9-10) and absolute numbers (n = 4-5) of CD8+ PD1+ TILs from Havcr2^fl/fl and Havcr2^cko tumours. b, Analysis of expression of PD1 versus TIM-3, Lag3 and TIGIT in CD8⁺ TILs (n = 4-5). c, Flow cytometry (d14 harvest) of CD8 TILs from Havcr2^fl/fl and Havcr2^cko for expression of TIM-3 and CXCR5 (n = 4-5). d, Flow cytometry of CD8⁺ PD1⁺ TILs from Havcr2^fl/fl and Havcr2^fl/fl × Zbtb46^Cre for expression of IL-7R, SLAMF6 CX3CR1, IFNγ, IL-2, TCF1, Ki67 and T-bet (bottom right) Representative histograms of data in d. The results shown are from one experiment, representative of at least three independent experiments, n = 4-5 group. *P < 0.05; **P < 0.01; (Student Two-Tailed t-test).

Extended Data Fig. 6 |. Identification of tumour infiltrating myeloid cells in Havcr2^fl/fl and Havcr2^cko tumours.

a, UMAP scRNA-seq plot of annotated total myeloid cells from Havcr2^fl/fl and Havcr2^fl/fl × CD11c^Cre (Havcr2^cko) tumours. b, UMAP scRNA-seq plots showing select marker gene expression. c, Heat map from scRNA-seq displaying normalized expression of select genes in each cluster. d, UMAP scRNA-seq plot showing distribution of Havcr2^fl/fl (blue) and Havcr2^cko (orange) cells. e, Bar graph showing frequency of Havcr2^fl/fl (blue) and Havcr2^cko (orange) cells in each cluster.

Extended Data Fig. 7 |. Decreased expression of mregDC markers in TIM-3-deficient migDCs.

a, MC38-OVA^dim (0.5 × 10⁶ cells) tumour cells were subcutaneously implanted into Havcr2^fl/fl and Havcr2^fl/fl × CD11c^Cre (Havcr2^cko) animals and Flow cytometric analysis of DC populations was performed on D14 to assess expression of described mregDC markers including CD200, CD83, 1L4R and OX40. The results shown are from one experiment, n = 5 per group. *P < 0.05; **P < 0.01; ***P < 0.001; (One-Way ANOVA). b, Tumour growth curves of MC38-OVA^dim (0.5 × 10⁶ cells) subcutaneously implanted into Havcr2^fl/fl and Havcr2^cko mice treated with either Isotype control, anti-IL-12 (500μg/mouse) or anti-IL-4 (25μg/mouse). Treatment was initiated on D3 and antibodies were delivered i.p. every 3 days until experiment cessation. The results shown are from one experiment, n = 4-5 per group. ***P < 0.001; ****P < 0.0001 (Two-Way ANOVA). c, Splenic DC were sorted from Havcr2^fl/fl and Havcr2^cko animals and cultured with dead HLA mismatched splenocytes osmotically loaded with 10mg/ml Ova together with CTV labelled naïve OTI cells. Representative plots of CD44⁺ CTV^lo T cells after 72-h co-culture. Mean ± s.e.m. of 3 individual mice. Alternatively, DC from Havcr2^fl/fl and Havcr2^cko animals, were cultured with beads passively adsorbed with Ova together with CTV labelled naïve OTI cells. Representative plots of CD44⁺CTV^lo T cells after 72-h co-culture. Mean ± s.e.m. of 3 individual mice, *P < 0.0001 (Student Two-Tailed t-test). d, CFSE or CTV labelled splenocytes were pulsed with OVA_257-264 or MOG_37-46 (Irrelevant Antigen) and injected at 50:50 ratio into MC38-OVA^dim bearing Havcr2^fl/fl or Havcr2^cko mice. Percentage cytotoxicity calculated as 100-(CTV/CTV+CFSE) (n = 5). e, Bar plot of data from Fig. 4a, demonstrating unidirectional analysis of the fraction of DCs expressing ligand X and the fraction of T cells expressing the cognate Receptor; Ligand (migDCs): Receptor (CD8) interaction from Havcr2^fl/fl (grey) and Havcr2^cko (red) tumours. f, UMAP showing expression of Il18r1 and Il18rap on cluster 7 (CD8⁺ T cells), with violin plots showing the differential expression of both receptor in of Havcr2^fl/fl (blue) and Havcr2^cko (orange) CD8⁺ T cells, Havcr2^fl/fl and Havcr2^cko (e) tumours were harvested and mechanically dissociated. Tissue supernatant was collected, and levels of cytokines were determined relative to mg protein per sample, n = 4/group, **P < 0.01; (Student Two-Tailed t-test).

Extended Data Fig. 8 |. Enhanced inflammasome activation in TIM-3-deficient^cko DC.

BMDC were differentiated in the presence of FLT3L for 10 days. a, Flow cytometric analysis assessing typical DC1 and DC2 markers. XCR1+ cells were sorted after 10 days of differentiation and seeded at a density of 0.25 × 10⁶. Sorted cells were either unstimulated or primed with LPS (1μg/ml) for 3 h followed by the addition of oxidised phospholipids (ox-PAPC) (100μg/ml), pdA: dT (1μg/ml), Flagellin (1μg/ml), C. difficile (1μg/ml), or Nigericin (10mM). b, c, Following overnight cultures supernatants were harvested and ELISA was performed to detect IL-1β (b) and TNF (c) (non-inflammasome regulated control). The results shown are from one experiment (n = 3 per group), representative of at least 3 individual experiments, *P < 0.05 **P < 0.01; ***P < 0.001 (Student Two Tailed t-test). d, MC38-OVA^dim was subcutaneously implanted into Havcr2^fl/fl and Havcr2^cko mice and on D14 mononuclear cells were isolated and incubated with DHR123 as a measure of ROS activity (n = 4). e, Tumour growth curve of MC38-OVA^dim subcutaneously implanted Havcr2^fl/fl and Havcr2^cko treated with control or anti-IL-1β and anti-IL-18 (n = 4). f, Weights of B16-OVA (0.25 × 10⁶ cells) subcutaneously implanted into Havcr2^fl/fl and Havcr2^fl/fl × Zbtb46^Cre and treated with either Isotype control (Hamster IgG and Rat IgG2a) or anti-IL-1β and anti-IL-18 (Hamster IgG and Rat IgG2a respectively), all at a dose of 8mg/kg. g-k, Flow cytometric analysis of CD8⁺ TILs harvested from MC38-OVA^dim tumours subcutaneously implanted into Havcr2^fl/fl and Havcr2^fl/fl × Zbtb46^Cre and treated with either Isotype control upon termination of experiment (d15). The results shown are from one experiment, representative of at least two independent experiments n = 4-5/group. ***P < 0.001; ****P < 0.0001 (Two-Way ANOVA). l-n, Havcr2^fl/fl (l), Havcr2^fl/fl × CD4^Cre (m) and and Havcr2^fl/fl × Zbtb46^Cre (n) mice were implanted with B16-OVA and monitored for development of a palpable tumour. On D6 when tumours reached ~30-50mm² mice were randomized and treated with either (i) Isotype controls (IgG2a and IgG2b), (ii) anti-TIM-3, (iii) anti PD-L1 and (iv) anti-TIM-3 + PD-L1. Anti-TIM-3 was administered at a dose of 200μg/mouse and anti-PDL1 at a dose of 50μg/mouse. All tumours were measured daily for the duration of the experiment. Antibody treatment was initiated on D6 and administered again on D9 and D12. Area under the curve (AUC) was calculated from graphs in (k-m). The results shown are from one experiment, n = 4 per group. **P < 0.01; ***P < 0.001; ****P < 0.0001 (Two-Way ANOVA). Area under curve data-**P < 0.01; ***P < 0.001; ****P < 0.0001 (One-Way ANOVA).