Intratumoral immune triads are required for adoptive T cell therapy-mediated elimination of solid tumors

Abstract

Tumor-reactive CD8 T cells found in cancer patients are frequently dysfunctional, unable to halt tumor growth. Adoptive T cell transfer (ACT), the administration of large numbers of in vitro-generated cytolytic tumor-reactive CD8 T cells, is an important cancer immune therapy being pursued. However, a limitation of ACT is that transferred CD8 T cells often rapidly lose effector function, and despite exciting results in certain malignancies, few ACT clinical trials have shown responses in solid tumors. Here, we developed preclinical cancer mouse models to investigate if and how tumor-specific CD4 T cells can be enlisted to overcome CD8 T cell dysfunction in the setting of ACT. In situ confocal microscopy of color-coded cancer cells, tumor-specific CD8 and CD4 T cells, and antigen presenting cells (APC), combined with functional studies, revealed that the spatial positioning and interactions of CD8 and CD4 T cells, but not their numbers, dictates ACT efficacy and anti-tumor responses. We uncover a new role of antigen-specific CD4 T cells in addition to the known requirement for CD4 T cells during priming/activation of naïve CD8 T cells. CD4 T cells must co-engage with CD8 T cells and APC cross-presenting CD8- and CD4-tumor antigens during the effector phase, forming a three-cell-cluster (triad), to license CD8 T cell cytotoxicity and mediate cancer cell elimination. Triad formation transcriptionally and epigenetically reprogram CD8 T cells, prevent T cell dysfunction/exhaustion, and ultimately lead to the elimination of large established tumors and confer long-term protection from recurrence. When intratumoral triad formation was disrupted, adoptively transferred CD8 T cells could not be reprogrammed, and tumors progressed despite equal numbers of tumor-infiltrating CD8 and CD4 T cells. Strikingly, the formation of CD4 T cell::CD8 T cell::APC triads in tumors of patients with lung cancers treated with immune checkpoint blockade was associated with clinical responses, but not CD4::APC dyads or overall numbers of CD8 or CD4 T cells, demonstrating the importance of triads in non-ACT settings in humans. Our work uncovers intratumoral triads as a key requirement for anti-tumor immunity and a new role for CD4 T cells in CD8 T cell cytotoxicity and cancer cell eradication.

Figure 1 |. Tumor-specific CD4 T cells prevent and reverse CD8 T cell dysfunction/exhaustion within solid tumors and mediate tumor elimination.

a. Scheme: tumor models, adoptively transferred effector T cells, and experimental schemes. b. B16 OVA-GP_61–80 (B16-OG) tumor growth (right) and Kaplan–Meier survival curve (left) of tumor-bearing B6 WT mice (CD45.2; Thy1.2) receiving effector TCR_OTI CD8 T cells alone (CD45.1) (black; TCR_OT1) or together with TCR_SMARTA CD4 T cells (Thy1.1) (red; TCR_OT1^(+CD4)) (ACT = adoptive T cell transfer). Data is representative of 5 independent experiments (n=5 mice/group). Values are mean ± SEM. Significance is calculated by multiple t test. Kaplan–Meier curve; **p=0.00021; Mantel–Cox test. c. MCA205 OVA-GP_61–80 (MCA-OG) tumor outgrowth and survival in B6 mice treated as described in b; **p=0.0003; Mantel–Cox test. Data is representative of 2 independent experiments (n=5–6 mice/group). d. TCR_OTI (% of total of CD8+ T cells) within progressing B16-OG tumors 8–9 days post transfer +/− TCR_SMARTA CD4 T cells. Data pooled from 2 independent experiments (n=8 mice/group). Each symbol represents an individual mouse. e. IFNγ and TNFα production of TCR_OTI isolated from B16-OG tumors 8–9 days post transfer +/− TCR_SMARTA CD4 T cells. Cytokine production was assessed after 4-hr peptide stimulation ex vivo. Data show 2 pooled independent experiments (n=5–7). f. Inhibitory receptor expression, and g. TOX expression of B16-OG tumor-infiltrating TCR_OTI isolated 8–9 days post transfer +/− TCR_SMARTA. Graphs depict relative MFI normalized to naive TCR_OTI; two pooled independent experiments (n=5–7mice/group). h. Mice with B16-OG tumors received effector TCR_OTI CD8 T cells 14 days post tumor transplantation; 9 days later, TCR_SMARTA CD4 T cells were adoptively transferred (red); B16-OG tumor growth in mice receiving only TCR_OT1 are shown in black. Data is representative of 2 independent experiments (n=8 mice/group). Values are mean ± SEM. Significance is calculated by multiple t test.

Figure 2 |. Tumor-specific CD4 T cells transcriptionally and epigenetically reprogram tumor-specific CD8 T cells and prevent terminal differentiation/exhaustion.

a. MA plot of RNA-seq data showing the relationship between average expression and expression changes of TCR_OT1 and TCR_OT1^(+CD4) TIL.Statistically significantly DEGs (false discovery rate (FDR) < 0.05) are shown in red and blue, with select genes highlighted for reference. b. Heat map of RNA-seq expression (normalized counts after variance stabilizing transformation, centered and scaled by row for DEGs) (FDR < 0.05) in TCR_OT1 and TCR_OT1^(+CD4) TIL. c. Selected GO terms enriched for genes up-regulated in TCR_OT1 (blue) and TCR_OT1^(+CD4) (red) TIL. d. Chromatin accessibility (ATAC-seq); (left) heatmap of log2-transformed normalized read counts transformed with variance stabilization per for regions with differential chromatin accessibility; (right) each row represents one peak (differentially accessible between TCR_OT1 and TCR_OT1^(+CD4) TIL; FDR < 0.05) displayed over a 2-kb window centered on the peak summit; regions were clustered with k-means clustering. Genes associated with the two major clusters are highlighted. e. ATAC-seq signal profiles across the Tox, Pdcd1, Lag3, Tcf7, and Lef1 loci. Peaks significantly lost or gained are highlighted in red or blue, respectively. f. Top 10 most-significantly enriched transcription factor motifs in peaks with increased accessibility in TCR_OT1^(+CD4) TIL (red) or TCR_OT1 TIL (blue). g. Enrichment of gene sets in TCR_OT1 and TCR_OT1 (^+CD4), respectively, described for human tumor infiltrating (TIL) CD8 T cell subsets (CD69- CD39) stem-like CD8 T cells/TIL (responders) or (CD69+ CD39+) terminally differentiated CD8 T cells/TIL (non-responders) from metastatic melanoma patients receiving ex vivo expanded TIL for ACT (S. Krishna et al, Science 2020). TCR_OT1^(+CD4) are enriched in genes observed in CD69- CD39- stem-like T cells/TIL from responders in contrast to TCR_OT1 which are positively enriched for genes in CD69+ CD39+ terminally differentiated CD8 T cells/TIL from non-responders. NES, normalized enrichment score.

Figure 3 |. Tumor elimination requires tumor antigen/epitope linkage and unique spatial orientation of tumor-specific CD8 T cells, CD4 T cells and CD11c+ dendritic cells (DC) within tumors.

a. B16-OG tumor outgrowth in CD11c-DTR/GFP bone marrow (BM) chimeras (scheme, top; DTR→WT or WT→WT) treated with diphtheria toxin (DT). In vitro activated TCR_OTI and TCR_SMARTA were adoptively transferred into lymphodepleted tumor-bearing BM chimeras. 5 days post ACT, mice were treated with DT. Representative of 2 independent experiments (n=3 mice/group). Values are mean ± SEM. Significance is calculated by multiple t test. b. (Top) Experimental scheme of tumor models A and B: 2.5×10⁶ B16-OG cancer cells (B16 OG; model A) or 1.25×10⁶ B16-OVA (B16-O) mixed with 1.25×10⁶ B16-GP_61–80 cancer cells (B16 O+G; model B) were transplanted into B6 WT mice. (Bottom), (left) Tumor outgrowth of B16-OG or B16 O+G tumors after TCR_OTI and TCR_SMARTA ACT. Representative of 2 independent experiments (n=7 mice/cohort). Data are shown as mean ± SEM. Significance is calculated by multiple t test. (Right) Kaplan–Meier curve; **p=0.0002; Mantel–Cox test. c. Percentage of TCR_OT1^(+CD4) (out of total CD8⁺ TIL) 9 days post ACT. d. Percentage of TCR_SMARTA (out of total CD4⁺ TIL) 9 days post ACT. Data represent 2 pooled, independent experiments (n=8 mice/tumor model). Each symbol represents an individual mouse. e. IFNγ, TNFα, CD107, Granzyme B production of TCR_OT1^(+CD4) isolated from B16-OG or B16 O+G tumors, or f. isolated from tumor-draining lymph nodes of B16-OG or B16 O+G tumor-bearing hosts. Cytotoxic molecules and cytokine production assessed after 4-hr peptide stimulation ex vivo. Representative of 2 independent experiments (n=3 mice/tumor). Data are shown as mean ± SEM. *p<0.05, unpaired two-tailed Student’s t test. NS, not significant. g. Mosaic, clonal growth of B16 OVA-EGFP mixed with B16 GP_61–80-Cerulean tumor cells (B16 O+G) in B6 WT mice. Shown are confocal microscopy sections of tumors with B16 OVA (green) and B16 GP (red) distinct tumor regions. h. Proposed model: Triad formation (three-cell-type clusters; CD8 T cells::CD4 T cells:: APC) form in B16 OG tumors (Model A) where CD8- and CD4-tumor antigens/epitopes are linked and co-presented on the same APC within tumors; tumor-specific CD8 and CD4 T cells engage on same APC; CD4 T cells reprogram CD8 T cells. Model B: B16 O+G; triads cannot form due to CD8- and CD4-tumor antigens being presented on distinct APC.

Figure 4 |. Intratumoral immune triads (three-cell-types clusters; CD8 T cell::CD4 T cell::APC) are required for CD8 T cell reprogramming and tumor elimination.

a. Color-coded mouse models to determine intratumoral immune triad formation (Models A and B (see Fig. 3)). B16 OG (Model A) or B16 O+G (Model B) tumors were established in CD11c-YFP mice (yellow); effector TCR_OTI-RFP (red) and TCR_SMARTA-EGFP T cells (green) were adoptively transferred into tumor-bearing hosts. Confocal microscopy analysis of frozen tumor tissue sections. Arrows indicate triads. b. Numbers of triads per field of view (FOV), and c. (left) Fold increase of triads normalized to total numbers of CD11c⁺YFP⁺ cells/FOV (right). c. Quantification of fold increase of numbers of CD4 T cell-DC dyads normalized to total number of infiltrating CD11c⁺YFP⁺ cells/FOV. Each symbol represents an individual frozen tumor section (n=3 mice/group/model). Data are shown as mean ± SEM. *** P <0.001, unpaired two-tailed Student’s t test. (e.-g). Increased triads in patients with Malignant Pleural Mesothelioma (MPM) treated with checkpoint immunotherapy is associated with pathologic responses. e. Treatment regimen and methodology used to determine triads (CD8 T cell::CD4 T cell::APC) and dyads (CD4::APC). Pipeline of co-localization detection by imaging mass cytometry (IMC; see Methods for more details). Briefly, FFPE tumor tissues were stained with 35 target-specific antibodies. Automated cluster detection estimated cluster boundaries by expanding the perimeter of nuclei, identified by Cell ID Intercalator-iridium (191Ir). IMC images were quantified through FIJI, and protein expression data extracted through mean intensity multiparametric measurements performed on individual clusters. Acquired cluster data were normalized with CytoNorm tools, and normalized cytometric data transferred into additional Spanning-tree Progression Analysis of Density-normalized Events (SPADE) to generate automated clustering algorithm and applied cytometric analysis in FlowJo. f. Representative multiplexed mass cytometry images of triads and dyads. g. Fold change of triads and dyads of pre- and post-immune checkpoint therapy (Tx) density (numbers/mm²) in responders (R) and non-responders (NR); *p=0.02; n.s. p=0.34 (not significant). h. Proposed model of TRIAD-associated cancer elimination.