Proximity-dependent labeling identifies dendritic cells that drive the tumor-specific CD4+ T cell response

Abstract

Dendritic cells (DCs) are uniquely capable of transporting tumor antigens to tumor-draining lymph nodes (tdLNs) and interact with effector T cells in the tumor microenvironment (TME) itself, mediating both natural antitumor immunity and the response to checkpoint blockade immunotherapy. Using LIPSTIC (Labeling Immune Partnerships by SorTagging Intercellular Contacts)-based single-cell transcriptomics, we identified individual DCs capable of presenting antigen to CD4⁺ T cells in both the tdLN and TME. Our findings revealed that DCs with similar hyperactivated transcriptional phenotypes interact with helper T cells both in tumors and in the tdLN and that checkpoint blockade drugs enhance these interactions. These findings show that a relatively small fraction of DCs is responsible for most of the antigen presentation in the tdLN and TME to both CD4⁺ and CD8⁺ tumor-specific T cells and that classical checkpoint blockade enhances CD40-driven DC activation at both sites.

Figure 1.. Using LIPSTIC to identify tumor antigen-presenting DCs.

(A) Schematic of the LIPSTIC system. T cells expressing S. aureus transpeptidase sortase A (SrtA) fused to CD40L transfer an injectable biotinylated peptide substrate (biotin-LPETG) onto five N-terminal glycines engineered into the extracellular domain of CD40 (G₅-CD40) on interacting DCs. (B) Experimental setup for panels (C-G). In (C, D, F, G) bilateral tumors were injected. (C) Percentage of labeled DCs in mice bearing B16^OTII or parental B16 tumors (left) and quantification of data (right) (n = 5, 4 mice per group). (D) Gating on resident and migratory DCs in tdLN (left), percentage of labeled migratory and resident DCs in tdLN (center), and quantification of data (n = 5 mice) (right). (E) Percentage of labeled DCs in mice treated with isotype control antibody vs. anti-CD40L antibody (left) and quantification of data for (n = 6 mice per group ) (right). (F) Contour plots (left, center) showing percentage of cDC1 and cDC2 among biotin⁺ and biotin⁻ DCs and quantification of data for (right) ( n = 5 mice). (G) Percentage of GFP⁺ and biotin⁺ DCs in tdLN (left) and quantification of GFP⁺ DCs among biotin⁺ and biotin⁻ DCs in tdLN ( n = 5 mice). (C, D, F, G each symbol represents one LN), E each symbol represents one mouse. (H) Proliferation of OT-II and OT-I T cells in vitro after 96 h of co-culture with DCs derived from mice carrying B16^OT-II or B16^OVA tumors, respectively (left) and quantification of T cells per well at the end of the culture period (right). For (H) tdLN were pooled from (n = 6-10 mice) each dot represents one culture well. P-values are for unpaired t test, except (F), where paired t test was used.

Figure 2.. Biotin⁺ DCs represent a transcriptionally distinct DCs state.

(A) t-distributed stochastic neighbor embedding (t-SNE) scatter plot showing clustering of DCs sorted from tdLN or steady-state iLN. Cells are pooled from 2 steady-state and 2 tumor-bearing mice from 2 independent experiments. (B) Distribution of steady-state iLN, biotin⁻ tdLN and biotin⁺ tdLN DCs in the same plot. Dotted lines indicate the approximate location of the cDC1/cDC2 boundary (left) and of biotin⁺ DCs (right). (C) The proportion of cells in each transcriptional cluster among steady-state, biotin⁻ and biotin⁺ DCs. (D) Expression of genes significantly upregulated in both biotin⁺ cDC1 and cDC2 compared to biotin⁻ cDC1 and cDC2. (E) Violin plots show the most upregulated gene signatures in both biotin⁺ cDC1 and cDC2 compared to biotin⁻ cDC1 and cDC2. (F) t-SNE contour density plots showing the distribution of biotin⁻ (left) and biotin⁺ DCs at 10 (center) or 15 (right) d.p.i. in tdLN of B16^OT-II bearing mice. Dotted circles indicate the approximate location of Cluster 0 (Fig. 2A) (G) Distribution of transcriptional clusters from Fig. 2A among biotin⁺ and biotin⁻ DCs from 10 and 15 d.p.i. Data for (F-G) are from the experiment described in Fig. 2A. (H) Contour plots show the percentage of biotin⁺ DC in tdLN of mice bearing B16^OT-II tumors at 10 (early) or 15 (late) d.p.i. (I) Quantification of data as in (H) (n = 10 mice per group from 4 independent experiments, each dot represents one mouse). (J-K) as in (H-I) but in mice bearing MC-38^OT-II tumors (n = 7 mice per group, with bilateral tumors; each dot represents one tdLN). Experiments in (H-K) used the Cd40lg^SrtAv1 allele. (L) t-SNE scatter plot showing clustering (left) and expression of a LIPSTIC+ gene signature (center) among tdLN myeloid cells under control conditions or upon depletion of Treg cells. Right, expression of the LIPSTIC+ signature by cluster. (M) Percentage of DCs from control vs. Treg-depleted conditions in Cluster 6 vs. in all cells. (N) Expression of the LIPSTIC+ signature in DCs from control or Treg depleted mice in Cluster 6 among total cDC1 and cDC2. Data in (L-N) are from [13]. Pearson’s chi-squared test was used for (C,G and M) Wilcoxon signed-rank test was used for (D, E and N), unpaired t-test was used for (I, K)

Figure 3.. Il27 promotes an antitumor CD4⁺ T cell response and drives antitumor immunity.

(A) Expression of Il27 and Ebi3 in DCs. t-SNE plot as in Fig. 2A. (B) Schematic for IL-27 (p28) blocking in mice bearing B16^OT-II tumors. (C) Expression of CXCR3⁺ (left) or TNFα and IFNγ (right) among OT-II T cells in tdLN of isotype or anti-p28-treated mice. Graphs summarize data from 2 independent experiments with (n = 13 mice per group.) (D) Percentage of OT-II cells among all CD4⁺ T cells (left) and expression of TNFα and IFNγ among OT-II T cells (right) in the tumors of isotype or anti-p28-treated mice. Graphs summarize data from 2 independent experiments with (n = 12 and 18 mice per group). (E) Schematic for IL-27 (p28) blocking in mice bearing B16^mOVA tumors. (F) Tumor growth curves and tumor weights for isotype control and anti-p28-treated groups (n = 9, 10 mice per group). (G) Tumor growth curves and tumor weights for Il27^f/f (control) and Il27^f/f Itgax-Cre groups. Data are from 4 independent experiments with n = 9 (control) and n = 13 (Il27^f/f Itgax-Cre). P-values are for unpaired t test.

Figure 4.. Using LIPSTIC to identify antigen-presenting APCs inside the tumor.

(A) Experimental setup for panels (B-D). (B) Percentage of labeled APCs in (Cd40^G5/+Cd40lg^SrtAv2) mice bearing B16^OTII tumors (left) and quantification of data (right) (n = 10 mice, 4 independent experiments) Cd40^G5/G5 mice were used as a baseline for biotin. (C) Mean fluorescent intensity (MFI) normalized to the MFI of biotin-negative DC for the indicated molecules. Biotin+(red) and biotin-(blue) DCs within TME (for CD40, CD86, CD80 n = 14 from 4 independent experiments, CD200, PD-L1 n = 3). (D) Experimental setup as in (A) except anti-MHC-II or isotype control were injected 2h prior to substrate injection intratumorally (200ug). (Left) Contour plots show percentage of labeled DCs, (right) quantification of labeled DCs (n = 5, from 2 independent experiments). (E) Proliferation of OT-II and OT-I T cells in vitro after 96 h of coculture with DCs derived from mice carrying B16^OT-II or B16^mOVA tumors, respectively (left) and quantification of T cells per well at the end of the culture period (right). For (E), tumors were pooled from (n = 5-9 mice, two independent experiments). Each dot represents one culture well. P-values are for one-way ANOVA(B), paired t-test (C), and unpaired t-test (D).

Figure 5.. Single-cell interaction-based transcriptomics of APCs within TME.

(A) Uniform Manifold Approximation and Projection (UMAP) plot showing manually annotated clustering of MHC-II⁺ myeloid cells sorted from the TME at 15 days post-tumor inoculation. See Fig. S5B for unsupervised clustering. (B) Distribution of LIPSTIC signal (anti-biotin DNA hashtag labeling) in the same cells, in log-normalized counts. (C) Expression of the tdLN_LIPSTIC⁺ signature within biotin⁺ and biotin⁻ cells in the indicated populations. P-value is for Wilcoxon signed-rank test. (D) Scatter plot showing the expression of CD40 mRegDC gene signatures for the indicated populations of cells (left), quantification of indicated populations in the “mRegDC x CD40 diagonal” and “mRegDC^hi CD40^hi” clusters. (E) Pseudotime trajectory showing the transition from cDC1 to mRegDC1 (left) and biotin acquisition and expression of the mRegDC program along that trajectory (right). (F) as in (E) but for cDC2. (G) Cartoon depicting the tumor photoconversion experiment. (H) Percentage of photoconverted migratory and resident DCs in the tdLN at 10 (early, top) and 15 (late, bottom) d.p.i with (n = 6 mice per group; representative plots are shown). (I) UMAP showing correspondence between tumor APCs and photoconverted DCs arriving at the tdLN, using the MapQuery function of Seurat.

Figure 6.. Checkpoint blockade amplifies CD40-CD40L interaction axis in the tdLN and TME.

(A) Experimental setup for B and C. (B) Contour plot and quantification of biotin⁺ DCs in the tdLN in the indicated groups. (n = 7 for iso and n = 8 for CTLA-4, 2 independent experiments) (C) Contour plot and quantification of CD11b⁺ and CD11b⁻ DCs among biotin⁺ cells. (D) Experimental setup for E and F. (E) Contour plot and quantification of biotin⁺ DCs in the TME in the indicated groups. (n = 5 for isotype control and n = 7 for anti-CTLA-4, 2 independent experiments) (F) Contour plot and quantification of biotin⁺ Ly6C⁺ APCs in the TME in the indicated groups. P-values were calculated using an unpaired t-test. (G) UMAP plot showing the distribution of LIPSTIC signal (anti-biotin DNA hashtag labeling, in log-normalized counts) in anti-CTLA-4 vs isotype control-treated groups (n = 4 for control and n = 3 for anti-CTLA-4). (H) Violin plots showing anti-biotin DNA hashtag labeling (LIPSTIC signal) in the indicated populations in anti-CTLA-4 and isotype control groups. Perrcentage of biotin+ cells are given for each violiln. (D) Violin plots showing expression of the tdLN_LIPSTIC⁺ signature within anti-CTLA-4 and isotype control-treated groups within the indicated populations. For (H-I) P values were calculated using Wilcoxon signed-rank test. (J) Volcano plots showing differentially expressed genes between biotin⁺ APCs from isotype control and anti-CTLA-4-treated mice. Statistically significant genes are colored in red.