The immunostimulatory RNA RN7SL1 enables CAR-T cells to enhance autonomous and endogenous immune function

Abstract

Poor tumor infiltration, development of exhaustion, and antigen insufficiency are common mechanisms that limit chimeric antigen receptor (CAR)-T cell efficacy. Delivery of pattern recognition receptor agonists is one strategy to improve immune function; however, targeting these agonists to immune cells is challenging, and off-target signaling in cancer cells can be detrimental. Here, we engineer CAR-T cells to deliver RN7SL1, an endogenous RNA that activates RIG-I/MDA5 signaling. RN7SL1 promotes expansion and effector-memory differentiation of CAR-T cells. Moreover, RN7SL1 is deployed in extracellular vesicles and selectively transferred to immune cells. Unlike other RNA agonists, transferred RN7SL1 restricts myeloid-derived suppressor cell (MDSC) development, decreases TGFB in myeloid cells, and fosters dendritic cell (DC) subsets with costimulatory features. Consequently, endogenous effector-memory and tumor-specific T cells also expand, allowing rejection of solid tumors with CAR antigen loss. Supported by improved endogenous immunity, CAR-T cells can now co-deploy peptide antigens with RN7SL1 to enhance efficacy, even when heterogenous CAR antigen tumors lack adequate neoantigens.

Keywords: CAR-T cells, RN7SL1, 7SL, pattern recognition receptors, RIG-I, MDA5, interferon, exosomes, extracellular vesicles.

Figure 1.. Sensing of RN7SL1 by immune cells rather than cancer cells favors anti-tumor immune responses.

A. MHC-I expression on B16-F10 with or without RIG-I and/or MDA5 knockout after transfection with RN7SL1 (7SL), Scr RNA (Scr), poly I:C, or cellular RNA. Expression on Cas9 control cells (Cas) and after liposome-only transfection are also shown. B-C. Survival of mice bearing B16-F10 tumors after intratumoral injection of the indicated RNA on days 5, 8, and 11 combined with treatment with CD4 and CD8 T cell depleting antibodies (n=10 per group) (B) or treatment with anti-CTLA4 + anti-PD1 (n=15–16 per group) (C). D-E. GZMB and Ki67 expression in intratumoral CD3⁺ T cells (D) or DC infiltration (E) from mice with B16-F10 tumors injected with the indicated RNA. Tumors were harvested at day 15. F. Survival of mice with control or RIG-I knockout B16-F10 tumors after the indicated treatment (n=6–22 per group). G. Model for how PRR signaling affects different cellular compartments within the TME. P-values for survival are by log-rank test. For comparison between two groups, a two-sided T-test or Wilcoxon test is used for parametric or non-parametric data, respectively. See also Figure S1.

Figure 2.. Human CAR-T cells engineered to express RN7SL1 enhance solid tumor control

A. Construct design for M5BBz (M5) and 19BBz CARs with and without co-expression of RN7SL1 (7SL) or Scr RNA. EF1a promoter is utilized for human CAR vectors and 5’ LTR for mouse. B. Effector-memory differentiation pre-infusion defined by CCR7^-/CD45RO⁺ for the indicated CAR-T cells with or without RN7SL1 expression. Colors are independent donors. C. Tumor growth and survival from NSG mice bearing ASPC-1 tumors treated with untransduced (UTD), M5BBz, or M5BBz-7SL (7SL) CAR-T cells. N=15 mice/group, 3 independent donors. D. Cell populations in ASPC-1 tumors overlaid with expression for the indicated genes or an IFN-I ISG metagene (red is high expression and blue is low). E. Distribution of IFN-I ISG metagene values for cells in the indicated populations shown in (D). Labeled in red text are differences in the median values (maroon line) between M5BBz-7SL and M5BBz CAR-T cell treated groups. The p-value shows the significance of this difference when comparing CAR-T and cancer cell populations. F. M5BBz CAR-T cell subsets from (D) along with enrichment of gene sets for T cell exhaustion (Gen.Tex), memory precursor (MP), effector/central memory (EM_CM), or terminal effector (TE). Heatmap for select genes is also shown (right). G-H. Density plot (G) and quantification (H) of frequencies for indicated CAR-T cell subsets. I-J. Frequency of CAR-T cells post-infusion in ASPC-1 tumors (I) or peripheral blood (J) as measured by flow cytometry. K-L. CCR7^-/CD45RO⁺ effector memory differentiation of CAR-T cells in peripheral blood (K) or CAR-T cell exhaustion measured using TIM3 (L). N=5–15 mice/group, 3 independent donors. P-values for survival are by log-rank test. For comparison between two groups, a two-sided T-test or Wilcoxon test is used for parametric or non-parametric data, respectively. Mixed effect model is used for tumor growth analysis. See also Figure S2.

Figure 3.. CAR-T cells deploying RN7SL1 in extracellular vesicles preferentially deliver RNA to endogenous immune cells and improve anti-tumor response.

A. Detection of RN7SL1 (7SL) or Scr RNA (Scr) from extracellular vesicles (EVs) secreted by mouse 19BBz CAR-T cells as assayed by qRT-PCR. B. Ectopic expression of human CD19 on mouse B16 melanoma cells (B16-h19). C. Transfer of RNA from CD45.1⁺ CAR-T cells to CD45.2^– cancer cells or CD45.2⁺ immune cells in B16-h19 tumors (top). Prior to RNA labeling with SytoRNA, CAR-T cells were treated with DMSO or a neutral sphingomyelinase inhibitor (NSMi) to block EV secretion and then transferred to tumor-bearing mice. RNA transfer was assessed 24 hours later. D. Presence of 7SL or Scr from sorted intratumoral CD45.2⁺ immune cells positive for SytoRNA. E. Relative transfer of CAR-T cell RNA to either CD45.2⁺ immune cells or CD45.2^– cancer cells quantified by flow cytometry of individual tumors. Fold change from equal transfer is shown. Positive and negative values indicate greater transfer to immune cells or cancer cells, respectively. F-G. Survival of mice with B16-h19 tumors treated with 19BBz, 19BBz-7SL (7SL), 19BBz-Scr (Scr), or untransduced (UTD) CAR-T cells on days 5 and 12 (n=5–10 per group) (F) or combined with anti-CTLA4 on days 8, 11, and 14 (n=14–20 per group) (G). No treatment (No Tx) group is also shown (n=5). H. Frequency of transferred CD45.1⁺ CAR-T cells or endogenous CD45.2⁺ immune cells in B16-h19 tumors at day 15 after treatment. Frequencies are relative to live cells. I. RNA transfer from human 19BBz-7SL CAR-T cells to donor-matched PBMCs following stimulation with the indicated beads. J. Survival of mice bearing B16 tumors with or without human CD19 expression following treatment with indicated CAR-T cells (n=7–8 per group). P-values for survival are by log-rank test. For comparison between two groups, a two-sided T-test or Wilcoxon test is used for parametric or non-parametric data, respectively. See also Figure S3.

Figure 4.. Deployment of RN7SL1 by CAR-T cells restricts suppressive features of myeloid cells and promotes co-stimulatory features of dendritic cells.

A. Differentiation trajectory and cell states for myeloid cells and DCs from B16-h19 tumors treated with CAR-T cells alone (top). Pseudotime values are color-coded. The identity of each state was determined by enrichment of gene sets (bottom) for monocytic (Mono), macrophage (Mac), or DC or monocytic DC (MonoDC) cell types. Also included is a gene set for MDSCs (Gen.MDSC). B. Distribution of myeloid cells in each cell state after treatment with the indicated CAR-T cells. Shown are relative densities overlaid on the differentiation trajectory plot. C. Enrichment scores for type I IFN-stimulated genes (ISGs) for each myeloid state and after treatment with the indicated CAR-T cells. Black boxes in the heatmap outline values with p < 0.05. D. Average expression of MDSC genes for each myeloid state and the frequency of myeloid cells in each of the two states most enriched in MDSC genes. E. Expression of Tgfb1 for each myeloid state and after CAR-T cell therapy. Black boxes in the heatmap outline values with p < 0.05. F. DC states from CD45.2⁺ intratumoral immune cells from (A) were reclustered and shown on a UMAP plot (top). Clusters were assigned based on enrichment of gene sets for the indicated DC subtype (bottom). G. Distribution of DC subtypes shown by relative densities overlaid on the UMAP plot. H. Enrichment scores for ISGs for each DC subtype and after CAR-T cell therapy. Black boxes in the heatmap outline values with p < 0.05. I-J. Frequency of pDC-like cells (I) and percentage of DCs expressing the pDC marker CD209a by flow cytometry after CAR-T cell therapy (J). K-L. Enrichment scores for co-stimulatory (left) and negative immunoregulatory (right) genes for each DC subtype (K), and ratio of DCs expressing CD86 or PDL1 by flow cytometry (L) after CAR-T cell therapy. Black boxes in the heatmap outline values with p < 0.05. For comparison between two groups, a two-sided T-test or Wilcoxon test is used for parametric or non-parametric data, respectively. See also Figure S4.

Figure 5.. CAR-T cell delivery of RN7SL1 results in greater DC stimulation and expansion of endogenous effector-memory-like CD8 T cells.

A. Differentiation trajectory and cell states for CD45.2⁺ non-naïve CD8 T cells from B16-h19 tumors treated with CAR-T cells alone (top). The identity of each state was determined by enrichment of gene sets (bottom) for CD8 T cell exhaustion (Core.Tex) followed by gene sets for exhausted subsets (transitory, progenitor 2: Prog 2, intermediate: Int Tex, and terminal: Term Tex) or for non-exhausted subsets (memory precursor: MP, effector-memory: EM, terminal-effector/effector-memory: TE_EM). B. Distribution of endogenous CD8 T cells in each cell state after treatment with indicated CAR-T cells. Relative densities are overlaid on the trajectory plot. C. Expression of select marker genes associated with exhausted and non-exhausted CD8 T cells. D. Frequency of CD8 T cells in each state after CAR-T cell therapy. E-F. Percent of CD8 T cells expressing ITGB7 (E) or TOX (F) by flow cytometry. G. Expression of the indicated markers on OT-I T cells after stimulation of BMDCs from wildtype or MAVS KO mice with RN7SL (7SL) or Scr RNA (Scr). Fold change is relative to control liposomes (Cont). P-values for effect of 7SL and whether this effect is influenced by BMDC genotype (7SL:Genotype) are shown in the right-hand margin. H. Increase in activation markers on OT-I T cells after BMDCs were stimulated with EV RNA from 19BBz or 19BBz-7SL (7SL) CAR-T cells. I. Increase in type I IFN genes by qRT-PCR relative to liposome control (Cont) from BMDCs stimulated with 7SL. J. Increase in OT-I T cells expressing GZMB after BMDC stimulation with the indicated RNA as described in (G) with or without anti-IFNAR antibody. Shown are P-values for effect of 7SL and whether this effect is influenced by anti-IFNAR treatment (7SL:Treatment). For comparison between two groups, a two-sided T-test or Wilcoxon test is used for parametric or non-parametric data, respectively. Multivariable linear regression is used to determine significance of 7SL and interaction effects. See also Figure S5.

Figure 6.. Endogenous T cells, dendritic cells, and activation of host RIG-I are required for RN7SL1 to enhance CAR-T cell response.

A-B. Survival of T cell-deficient Tcra^−/− (A) or cDC1-deficient Batf3^−/− (B) mice bearing B16-h19 tumors treated with indicated CAR-T cells plus anti-CTLA4 (aCTLA4) (n=7–9 per group). C. Percentage of endogenous CD8 T cells (left) or CAR-T cells (right) in wildtype (WT) or Batf3^−/− mice bearing B16-h19 tumors and treated with CAR-T cells. D-E. Frequency of CAR-T cells (D) or endogenous immune cells (E) per gram of tumor following CAR-T cell therapy with or without FTY720. Contribution of Ki67⁺ T cells is separately shown (E, right). F-H. Survival of Rig-I^−/− mice (n=5–9 per group) (F), Rig-I^−/− bone marrow chimera mice (n=6–7 per group) (G), or anti-IFNAR (aIFNAR) treated wildtype mice (n=9–11 per group) (H) bearing B16-h19 tumors and treated with CAR-T cells plus anti-CTLA4. Flow plots in (G) show representative bone marrow reconstitution from WT or Rig-I^−/− donor. I-K. Percentage of intratmoral DC1 dendritic cells (I), endogenous ITGB7⁺ effector-memory-like CD8⁺ T cells (J), or endogenous T cells expressing GZMB or Ki67 (K) after CAR-T cell therapy with or without anti-IFNAR. ITGB7⁺ T cell data from Fig. 5E is provided for reference. Shown are P-values for effect of 7SL and whether this effect is influenced by anti-IFNAR treatment (7SL:Treatment). P-values for survival are by log-rank test. For comparison between two groups, a two-sided T-test or Wilcoxon test is used for parametric or non-parametric data, respectively. Multivariable linear regression is used to determine significance of 7SL and interaction effects. See also Figure S6.

Figure 7.. CAR-T cell deployment of RN7SL1 with peptide antigen overcomes resistance of poorly immunogenic tumors with heterogeneous CAR antigen expression.

A. Percentage of TRP2⁺ CD8 T cells from B16-h19 tumors expressing GZMB or Ki67 after CAR-T cell therapy. Shown are P-values for effect of 7SL and whether this effect is influenced by anti-IFNAR treatment (7SL:Treatment). B. Survival after CAR-T cell therapy of mice bearing a heterogenous tumor comprised of a 1:1 mix of B16 cells with and without human CD19 (n=6–7 per group). C. Design of 19BBz CAR vector expressing Ova (Ova-19BBz) or Ova plus RN7SL1 (Ova-19–7SL) (top) and representative flow cytometry plots showing detection of SIINFEKL peptide on CAR⁺ and CAR^– T cells (bottom). D. Detection of SIINFEKL peptide on B16 cells loaded with indicated concentrations of EVs from Ova-19BBz CAR-T cells or 19BBz CAR-T cells (Cont) (flow plots, top). OT-I T cells were then added and activation was measured (flow plots, bottom) and quantitated (right) using GZMB and Ki67. E. SIINFEKL peptide transfer to cancer and immune cells measured in mixed CD19⁺ and CD19^– B16 tumors following treatment in vivo with 19BBz or Ova-19BBz (Ova) CAR-T cells. Representative flow cytometry plots of cancer cells are shown (left). F-G. Endogenous Ova-specific T cell expansion measured by tetramer and Ki67 expression (F), and frequency of Ova-specific and Ki67-positive CD8 T cells (G). H. Growth of heterogenous CD19⁺/CD19^– B16 tumors following treatment with indicated CAR-T cells. I. Growth of heterogenous CD19⁺/CD19^– KP mixed tumors following CAR-T cell therapy with anti-CTLA4 plus anti-PD1. 5×10⁵ OT-I T cells were transferred prior to tumor implantation. J. Model for how CAR-T cells delivering RN7SL1 +/− peptide antigen improve solid tumor response. P-values for survival are by log-rank test. Multivariable linear regression is used to determine significance of 7SL and interaction effects. Mixed effect model is used for tumor growth analysis. See also Figure S7.