Novel Therapeutic Approach Targeting CXCR3 to Treat Immunotherapy Myocarditis

Abstract

Background: Immune checkpoint inhibitors (ICIs) are successful in treating many cancers but may cause immune-related adverse events. ICI-mediated myocarditis has a high fatality rate with severe cardiovascular consequences. Targeted therapies for ICI myocarditis are currently limited.

Methods: We used a genetic mouse model of PD1 deletion (MRL/Pdcd1^-/-) along with a novel drug-treated ICI myocarditis mouse model to recapitulate the disease phenotype. We performed single-cell RNA-sequencing, single-cell T-cell receptor sequencing, and cellular indexing of transcriptomes and epitopes on immune cells isolated from MRL and MRL/Pdcd1^-/- mice at serial time points. We assessed the impact of macrophage deletion in MRL/Pdcd1^-/- mice, then inhibited CXCR3 (C-X-C motif chemokine receptor 3) in ICI-treated mice to assess the therapeutic effect on myocarditis phenotype. Furthermore, we delineated the functional and mechanistic effects of CXCR3 blockade on T-cell and macrophage interactions. We then correlated the results in human single-cell multiomics data from blood and heart biopsy data from patients with ICI myocarditis.

Results: Single-cell multiomics demonstrated expansion of CXCL (C-X-C motif chemokine ligand) 9/10+CCR2+ macrophages and CXCR3hi (C-X-C motif chemokine receptor 3 high-expressing) CD8+ (cluster of differentiation) effector T lymphocytes in the hearts of MRL/Pdcd1^-/- mice correlating with onset of myocarditis development. Both depletion of CXCL9/10+CCR2+ (C-C motif chemokine receptor) macrophages and CXCR3 blockade, respectively, led to decreased CXCR3hi CD8+ T-cell infiltration into the heart and significantly improved survival. Transwell migration assays demonstrated that the selective blockade of CXCR3 and its ligand, CXCL10, reduced CXCR3+CD8+ T-cell migration toward macrophages, implicating this interaction in T-cell cardiotropism toward cardiac macrophages. Furthermore, cardiomyocyte apoptosis was induced by CXCR3hi CD8+ T cells. Cardiac biopsies from patients with confirmed ICI myocarditis demonstrated infiltrating CXCR3+ T cells and CXCL9+/CXCL10+ macrophages. Both mouse cardiac immune cells and patient peripheral blood immune cells revealed expanded TCRs (T-cell receptors) correlating with CXCR3hi CD8+ T cells in ICI myocarditis samples.

Conclusions: These findings bring forth the CXCR3-CXCL9/10 axis as an attractive therapeutic target for ICI myocarditis treatment, and more broadly as a druggable pathway in cardiac inflammation.

Keywords: T-lymphocytes; chemokines; immune checkpoint inhibitors; macrophages; myocarditis.

Figure 1.

Identification of novel populations of CXCL9 (C-X-C motif chemokine ligand)/10+CCR2+ (C-C motif chemokine receptor) macrophages and CXCR3hi (C-X-C motif chemokine receptor 3 high-expressing) CD8+ (cluster of differentiation) effector T cells from heart and blood single-cell RNA-seq data from Murphy Roths large (MRL) and MRL/Pdcd1−/− mice at time points 1, 2, and 4 weeks. A, Workflow of time-lapse experiment. Peripheral blood mononuclear cell and hearts from MRL and MRL/Pdcd1^−/− mice are harvested at ages 1 week (MRL heart: n=6, MRL/Pdcd1^−/− heart: n=7, MRL blood: n=8, MRL/Pdcd1^−/− blood: n=6), 2 weeks (MRL heart: n=6, MRL/Pdcd1^−/− heart: n=6, MRL blood: n=6, MRL/Pdcd1^−/− blood: n=6), and 4 weeks (MRL heart: n=6, MRL/Pdcd1^−/− heart: n=6, MRL blood: n=6, MRL/Pdcd1^−/− blood: n=6). B, UMAP (uniform manifold approximation and projection) of immune cell populations at the CD45+. C, Feature plots of immune cell markers at the CD45+ level (Feature plots [UMAPs] with specific gene expression values represented by the color purple) are generated from scaled data using the Seurat FeaturePlot function. These are visual representations of the data where each feature [gene] is plotted on a scale that is standardized using Z scores, so the values are transformed to have a mean of 0 and an SD of 1, allowing for easier comparison between different features. The color bar on the right side of each feature plot then represents the Z score value for each data point, with different colors indicating high or low values relative to the overall distribution). D, UMAP shows changes in macrophage subpopulations. E, Quantification of CXCL9/10+CCR2+ macrophages shows a significant increase in the hearts of MRL/Pdcd1^−/− mice compared with MRL mice at 2 weeks (P=2.2×10⁻3 by Mann-Whitney U test) and 4 weeks (P=2.2×10⁻3 by Mann-Whitney U test). F, Quantification of cardiac resident macrophages shows a significant decrease in the hearts of MRL/Pdcd1^−/− mice compared with that of MRL mice at 2 weeks (P=4.3×10⁻3 by Mann-Whitney U test) and 4 weeks (P=2.2×10⁻3 by Mann-Whitney U test). G, UMAP shows changes in CD8+ T-cell subpopulations. H, Quantification of CXCR3hi CD8+ effector T cells shows a significant increase in the hearts of MRL/Pdcd1^−/− mice compared with that of MRL mice at 2 weeks (P=2.2×10⁻3 by Mann-Whitney U test) and 4 weeks (P=4.1×10⁻2 by Mann-Whitney U test). I, Quantification of all effector/effector memory CD8+ T cells shows a significant increase in the hearts of MRL/Pdcd1^−/− mice compared with that of MRL mice at 2 weeks (P=8.7×10⁻3 by Mann-Whitney U test) and 4 weeks (P=4.1×10⁻2 by Mann-Whitney U test). J, Feature plots indicate that CXCR3hi CD8+ effector T-cell populations strongly express cytotoxic markers. K, Violin plots of perforin and granzyme B show that expression is highest in CXCR3hi effector CD8+ and effector CD8+ subsets. L, UMAP at the CD8+ T-cell level shows expanded (≥3) cardiac CD8+ T-cell clones increase from 1 to 4 weeks of age. M, Quantification of clonally expanded (≥3) cardiac CXCR3hi effector CD8+ T cells shows a significant increase (P=1.5×10⁻2 by Mann-Whitney U test) in 4-week-old MRL/Pdcd1^−/− mice. CXCR3 indicates CXC chemokine receptor 3; and NK, natural killer cells. Biorender was used for A.

Figure 2.

Depletion of monocytes/macrophages with liposomal clodronate attenuates immune checkpoint inhibitor (ICI) myocarditis in Murphy Roths large (MRL)/Pdcd1−/− mice. A, Workflow schematic of monocyte/macrophage depletion. B, Myocarditis-free (<10%) survival curve of MRL/Pdcd1^−/− mice injected with either liposomal control (n=12) or liposomal clodronate (n=9). Mice injected with liposomal clodronate have a significantly higher probability of myocarditis-free survival (P=2.7×10⁻3 by Log-rank test). C, left, Hematoxylin and eosin (H&E) stain of hearts of MRL/Pdcd1^−/− mice injected with either liposomal control or liposomal clodronate. Right, H&E stains are colocalized with caspase 3 and troponin-T staining. Control hearts are shown at ×200 and ×400 magnifications to indicate double positives for caspase 3 and troponin-I in cardiomyocytes near immune cell infiltrates. Liposomal clodronate hearts are shown at ×200 and ×400 magnifications and are negative for caspase 3. D, Quantification of percentage of myocarditis in the hearts of MRL/Pdcd1^−/− mice injected with either liposomal control or liposomal clodronate. Left, Mice injected with liposomal clodronate have significantly less cardiac immune cell infiltration per area (P=1.5×10⁻3 by Mann-Whitney U test). Right, The hearts of mice injected with liposomal clodronate have significantly less apoptosis (P=2.2×10⁻3, Mann-Whitney U test). E, UMAP (uniform manifold approximation and projection) of major immune cell populations identified at the CD45+ level from single-cell RNA-sequencing of the hearts of MRL/Pdcd1^−/− mice injected with liposomal control (n=6) or liposomal clodronate (n=6). F, UMAP of macrophage immune cell subpopulations in the hearts of MRL/Pdcd1^−/− mice injected with liposomal control or liposomal clodronate. G, Quantification of CXCL (C-X-C motif chemokine ligand) 9+CXCL10+CCR2+ (C-C motif chemokine receptor) macrophages analyzed via single-cell RNA-sequencing (scRNAseq) shows a significant decrease in the hearts of mice injected with liposomal clodronate compared with control (P=4.3×10⁻3 by Mann-Whitney U test). H, Quantification of cardiac resident macrophages analyzed via scRNAseq shows a significant increase in the hearts of mice injected with liposomal clodronate compared with control (P=1.7×10⁻2 by Mann-Whitney U test). I, UMAP of CD8+ (cluster of differentiation) T-cell subpopulations in the hearts of MRL/Pdcd1^−/− mice injected with liposomal control or liposomal clodronate. J, Quantification of all CXCR3hi (C-X-C motif chemokine receptor 3 high-expressing) effector CD8+ T cells in the hearts of MRL/Pdcd1^−/− mice shows a significant decrease in that of those injected with liposomal clodronate compared with the control (P=4.1×10⁻2 by Mann-Whitney U test). K, Feature plots indicate that CXCR3hi CD8+ effector T-cell populations strongly express cytotoxic markers such as perforin and granzyme B. L, UMAP at the CD8+ T-cell level shows expanded (≥3) cardiac CD8+ T-cell clones in MRL/Pdcd1^−/− mice injected with either liposomal control or liposomal clodronate. Expanded TCRs (T-cell receptors) are primarily concentrated in the CXCR3hi effector CD8+ cluster. M, Quantification of clonally expanded (≥3) cardiac CXCR3hi effector CD8+ T cells shows that 91.935% of expanded CD8+ T-cell clones are CXCR3 (CXC chemokine receptor 3)-positive. Biorender was used for A. Nk indicates natural killer cells.

Figure 3.

In vivo CXCR3 (CXC chemokine receptor 3) blockade in a pharmacological mouse model for immune checkpoint inhibitor (ICI) myocarditis significantly decreases immune cell infiltration in the heart. A, Workflow schematic of in vivo CXCR3 blockade in Murphy Roths large (MRL) mice treated with anti–PD1 (programmed cell death-1) and anti-CTLA4 combination immunotherapy. MRL mice are treated with immunotherapy and either anti-CXCR3 (n=6) or an IgG control (n=6). Heart and blood are harvested after 6 doses of immunotherapy. All sample numbers represent biological replicates. B, Myocarditis-free (<10%) survival curve of immunotherapy-treated MRL mice injected with either anti-CXCR3 (n=12) or IgG control (n=12). Mice injected with anti-CXCR3 have a significantly higher probability of myocarditis-free survival (P=9.9×10⁻3 by log-rank test). C, left, Hematoxylin and eosin (H&E) staining of hearts of immunotherapy-treated MRL mice injected with either anti-CXCR3 or IgG control. Right, H&E stains are colocalized with caspase 3 and troponin-T staining. Anti-PD1/anti-CTLA4 (cytotoxic T-lymphocyte associated protein 4) combination immunotherapy-treated hearts are shown at ×200 and ×400 magnifications to indicate double positives for caspase 3 and troponin-T in cardiomyocytes near immune cell infiltrates. Anti-PD1/anti-CTLA4 and anti-CXCR3–treated hearts are shown at ×200 and ×400 magnifications and are negative for caspase 3. D, Quantification of percentage of myocarditis in the hearts of immunotherapy-treated MRL mice injected with either anti-CXCR3 (n=12) or IgG control (n=12). Left, Mice injected with anti-CXCR3 have significantly less immune cell infiltration per area (P=3.6×10⁻2 by Mann-Whitney U test). Right, The hearts of mice injected with anti-CXCR3 have significantly less apoptosis (P=2.2×10⁻3, Mann-Whitney U test). E, UMAP (uniform manifold approximation and projection) of major immune cell populations identified at the CD45+ (cluster of differentiation) level from single-cell RNA-sequencing of the hearts and blood of immunotherapy-treated MRL mice injected with either anti-CXCR3 (n=6) or IgG control (n=6) as well as the hearts of MRL mice treated with only IgG control (n=3). F, Feature plots of major CD45+ immune cell markers. G, UMAP of T-cell subpopulations at the CD8+ level. H, Quantification of CXCR3hi (C-X-C motif chemokine receptor 3 high-expressing) effector CD8+ T cells shows a significant decrease in the hearts of mice injected with anti-CXCR3 (P=4.6×10⁻2 by Mann-Whitney U test). I, Feature plots indicate that effector and effector memory CD8+ T-cell subsets express cytotoxic markers granzyme B and perforin, as well as chemotactic markers such as Ccl5. J, CD8+ T-cell UMAP shows expanded (≥3) cardiac CD8+ T-cell clones in immunotherapy-treated MRL mice injected with either anti-CXCR3 or IgG control. There is decreased expansion in mice treated with anti-CXCR3. Expanded TCRs (T-cell receptors) are primarily concentrated in the CXCR3hi effector CD8+ cluster. K, Pie chart of expanded (≥3) CD8+ T-cell clones in immunotherapy-treated MRL mice injected with either anti-CXCR3 or IgG control. L, Workflow schematic of in vivo anti-CXCR3 blockade after immunotherapy-induced myocarditis is determined via troponin-I ELISA. M, Quantification of plasma troponin-I levels before CXCR3/IgG treatment shows that mice treated with immunotherapy have significantly elevated troponin-I levels compared with untreated mice (P=1.0×10⁻3 by Mann-Whitney U test). N, Survival curve of mice treated with CXCR3 blockade (n=5) compared with IgG control (n=6) shows that after development of immunotherapy-induced myocarditis, CXCR3 blockade significantly increases probability of survival (P=3.1×10⁻2 by Log-rank test). O, Quantification of plasma troponin-I levels after anti-CXCR3 or IgG control treatment reveals that CXCR3 blockade significantly decreases troponin-I concentration (P=3.6×10⁻2 via Mann-Whitney U test). P, H&E staining of hearts of 6 doses of anti-CXCR3 or IgG-treated mice started after 6 doses of immunotherapy. Q, Quantification of percent myocarditis shows significantly decreased immune cell infiltration in the hearts of mice that underwent CXCR3 blockade treatment after myocarditis has already been determined (n=5) compared with IgG controls (n=6; 1.73×10⁻2 via Mann-Whitney U test). Biorender was used for A and L. IT indicates immunotherapy; and NK, natural killer cells.

Figure 4.

CXCL9 (C-X-C motif chemokine ligand) and CXCL10 mediate CD8+ (cluster of differentiation) T-cell migration toward macrophages in a CXCR3 (CXC chemokine receptor 3)-dependent manner. A, Workflow schematic of in vitro transwell assay of CD8+CXCR3+ T-cell and macrophage crosstalk. CD8+CXCR3+ T cells are isolated from the pooled hearts of anti–PD1 (programmed cell death-1)/anti-CTLA4 (cytotoxic T-lymphocyte associated protein 4)–treated Murphy Roths large (MRL) mice (n=6, biological replicates) and plated in the top insert of a transwell. Cardiac macrophages from pooled of anti-PD1/anti-CTLA4–treated MRL mice (n=6, biological replicates) are plated in the bottom well. Migration of CXCR3+CD8+ T cells is assessed via flow cytometry of the bottom well. B, Flow gating of the bottom well of in vitro transwell assay of cardiac CD8+CXCR3+ T-cell chemotaxis toward cardiac macrophages. C, Histogram and quantification of CXCR3+ fluorescence in the bottom well in control, IgG-treated, and anti-CXCR3– treated groups shows significantly decreased migration of CD8+CXCR3+ T cells treated with CXCR3 blockade (P=2.6×10⁻2, unpaired t test). D, Violin plot of Itga4 expression in cardiac CD8+ T cells from 4-week-old MRL and MRL/Pdcd1^−/− mice. E, Quantification revealed a significant increase (1-way ANOVA) in CD49d expression in CD8+ T cells via CXCL9/10 treatment when compared with control. CXCR3 blockade significantly decreased (1-way ANOVA) CD49d expression that had been increased by CXCL9/10 and returned expression to baseline (n=6). F, Workflow schematic of cardiomyocyte and CXCR3+CD8+ T-cell coculture apoptosis assay (n=6). G, Quantification of cardiomyocyte apoptosis shows that cardiomyocytes cocultured with CD8+CXCR3+ T cells have significantly increased apoptosis compared with cardiomyocytes in single culture (P=8.7×10⁻3, unpaired t test). Biorender was used for A and F. CM indicates cardiomyocyte; FACS, fluorescence activated cell sorting; FSC-A, forward scatter area; FSC-H, forward scatter height; Itga, integrin subunit alpha; ns, non-significant; and SSC-A, side scatter area.

Figure 5.

Heart and blood samples from patients with immune checkpoint inhibitor (ICI) myocarditis reveal CXCR3hi (C-X-C motif chemokine receptor 3 high-expressing) CD8+ (cluster of differentiation) T cells and CXCL9/10+ (C-X-C motif chemokine ligand) macrophage populations. A, Single-cell sequencing data of human samples included peripheral blood mononuclear cells from healthy controls not on ICIs (n=6), group A of patients treated with ICIs but having no immune-related adverse event (irAEs; n=8), group B of patients treated with ICIs and having nonmyocarditis irAEs (n=8), and group C of patients that were treated with ICIs and had ICI myocarditis (n=8). All samples represent biological replicates. B, Feature plot of CXCR3 (CXC chemokine receptor 3) expression on a UMAP (uniform manifold approximation and projection) of annotated CD8+ T-cell subpopulations shows overlap with immune cells previously labeled as Temra CD8+. C, Single-cell TCR (T-cell receptors)-sequencing of clonally expanded (≥3) CD8+ T-cell clones in the different patient groups. D, left, In total, 86% of clonally expanded CD8+ T cells express CXCR3. Right, Quantification of the CXCR3+CD8+:CD8+ T-cell ratio in nonexpanded vs expanded CD8+ T cells clones shows significantly more CXCR3+CD8+ T cells in expanded clones (P=1.2×10⁻4, Mann-Whitney U test). G, Hematoxylin and eosin (H&E) and immunoperoxidase staining for CXCR3, CXCL9, and CXCL10 in the hearts of 3 patients with confirmed ICI myocarditis. The lymphocytes show immunoreactivity for CXCR3 and the macrophages stained show CXCL9 and CXCL10 expression. Biorender was used for A.

Figure 6.

Diagram of the pathogenesis of immune checkpoint inhibitor (ICI) myocarditis. CCR indicates C-C motif chemokine receptor; CD, cluster of differentiation; CTLA4, cytotoxic T-lymphocyte associated protein 4; CXCL, C-X-C motif chemokine ligand; ICI, immune checkpoint inhibitor; IFN, interferon; IL, interleukin; PD1, programmed cell death-1; and VLA-4, very late activation antigen-4. Biorender was used to create this figure.