Decoy-resistant IL-18 reshapes the tumor microenvironment and enhances rejection by anti-CTLA-4 in renal cell carcinoma

Abstract

The cytokine IL-18 has immunostimulatory effects but is negatively regulated by a secreted binding protein, IL-18BP, that limits IL-18's anticancer efficacy. A decoy-resistant form of IL-18 (DR-18) that avoids sequestration by IL-18BP while maintaining its immunostimulatory potential has recently been developed. Here, we investigated the therapeutic potential of DR-18 in renal cell carcinoma (RCC). Using pantumor transcriptomic data, we found that clear cell RCC had among the highest expression of IL-18 receptor subunits and IL18BP of tumor types in the database. In samples from patients with RCC treated with immune checkpoint inhibitors, IL-18BP protein expression increased in the tumor microenvironment and in circulation within plasma in nonresponding patients, and it decreased in the majority of responding patients. We used immunocompetent RCC murine models to assess the efficacy of DR-18 in combination with single- and dual-agent anti-PD-1 and anti-CTLA-4. In contrast to preclinical models of other tumor types, in RCC models, DR-18 enhanced the activity of anti-CTLA-4 but not anti-PD-1 treatment. This activity correlated with intratumoral enrichment and clonal expansion of effector CD8+ T cells, decreased Treg levels, and enrichment of proinflammatory antitumor myeloid cell populations. Our findings support further clinical investigation of the combination of DR-18 and anti-CTLA-4 in RCC.

Keywords: Cancer; Cytokines; Immunology; Immunotherapy; Oncology.

Figure 1. IL18BP and IL18R1 are expressed at high levels in ccRCC and elevated IL18BP is associated with cytokine and T cell activation and worse survival.

(A–C) IL18R1 and IL18BP expression from TCGA PanCancer Atlas for all tumors (ccRCC indicated with red asterisk) (A), RCC histologic subtypes (B), and for ccRCC (C), by stage. (D) Kaplan-Meier survival curves based on IL18BP expression in ccRCC, dichotomized by median expression. (E) Volcano plot of transcripts enriched with high versus low IL18BP expression in ccRCC (log₂ fold-change thresholds of 1 and –1; P value threshold of 1 × 10^–6). (F) The top gene sets from enrichment analysis of transcripts enriched with high IL18BP expression. For B and C, statistical testing was performed using Kruskal-Wallis test with Dunn’s correction for multiple comparisons. **P < 0.01; ***P < 0.001; ****P < 0.0001.

Figure 2. IL-18BP protein levels increase after immunotherapy in nonresponding patients with RCC.

(A) Kaplan-Meier curves of overall survival of patients with RCC after ICIs by IL-18BP protein expression, dichotomized by median qIF levels. (B) IL-18BP protein levels assessed by qIF in the same RCC patient cohort as A, before and after ICIs, in ICI responders/nonresponders. (C) Circulating plasma levels of IL-18BP, as assessed by ELISA, from patient-matched samples before and after ipi + nivo treatment in a different RCC patient cohort from A and B. (D) Circulating plasma levels of IL-18BP from patient-matched samples before- and during ipi + nivo treatment, separated by treatment response. (E) The ratio of post/pretreatment IL-18BP plasma levels by treatment response. (F) The directional change of IL-18BP plasma levels after treatment by response. (G) Kaplan-Meier curves of progression-free survival (PFS) after ipi + nivo by directional change in circulating IL-18BP levels after treatment, in the same RCC cohort as in C–F. Statistical testing was performed using Mann-Whitney U test (B and E), Wilcoxon matched-pairs signed rank test (C and D), and Fisher’s exact test (F). Due to small samples sizes, formal statistical testing was not conducted on G, and the analysis should be viewed as hypothesis generating. *P < 0.05; **P < 0.01.

Figure 3. DR-18 combined with anti–CTLA-4 extends survival in murine RCC models.

(A) WT immunocompetent balb/c mice were s.c. engrafted with 0.5 × 10⁶ Renca or 1.0 × 10⁶ RAG cells. Starting on day 7–10, mice were treated twice weekly with phosphate buffered saline (PBS), DR-18 (s.c.), and/or ICIs (anti–PD-1/anti–CTLA-4) i.p. Five treatments were given. Red triangles indicate timing of administration of depleting/neutralizing antibodies. (B–E) Kaplan-Meier survival curves and mean tumor growth curves of mice engrafted with Renca (B and C) and RAG (select treatment groups shown) (D and E) cells. Data are shown as mean ± SEM (C and E). (F) Survival of mice engrafted with Renca tumors and treated with control PBS or DR-18 + anti–CTLA-4, either alone (PBS depletion) or with depleting/neutralizing antibodies. Depleting/neutralizing antibodies were given 24 hours prior to treatment and twice weekly thereafter. NK cells were depleted using anti-Asialo GM1. Renca data were combined from 3 independent experiments; RAG data were combined from 2 independent experiments. For Kaplan-Meier curves, statistical testing was performed using the log-rank test with Bonferroni correction in comparison with control-treated mice. *P < 0.05; **P < 0.01; ****P < 0.0001

Figure 4. DR-18 + anti–CTLA-4 potently induces inflammatory cytokines/chemokines.

(A) Schematic of treatment and sample collection time points for cytokine/chemokine profiling and scRNA/TCR-Seq in the Renca model. (B) Heatmap of the natural logarithm of circulating cytokine/chemokine levels in mice for the indicated treatments and time points (n = 3 mice/group, with the same mice collected at each time point), with unsupervised hierarchical clustering on the y axis. Data were generated using 31-plex Mouse Cytokine/Chemokine Array from Eve Technologies (MD31). (C) Volcano plots of the same data as in B, comparing circulating cytokine/chemokine levels with DR-18 + anti–CTLA-4 treatment (Combo) to PBS (log₂ fold-change thresholds of 0.5 and –0.5; P value threshold of 0.05; cytokine/chemokine changes with FDR < 0.05 highlighted as indicated). (D) Absolute levels of the indicated cytokines/chemokines at each time point for each treatment. Statistical testing performed using 2-way ANOVA with Tukey’s multiple comparisons test comparing all conditions within a given time point; only significant comparisons are shown. Tx, treatment; hr, hours. *P < 0.05; ***P < 0.001; ****P < 0.0001

Figure 5. DR-18 alters immune subset composition in Renca tumors, including enrichment and clonal expansion of CD8⁺ effector T cells.

(A) Uniform Manifold Approximation and Projection (UMAP) dimensionality reduction plot of clustering and annotation of all cell populations isolated from Renca tumors treated for 3 cycles with PBS, DR-18, anti–CTLA-4, or DR-18 + anti–CTLA-4 (Combo) (n = 3 mice/group, pooled) based on scRNA-Seq analysis. Annotations were performed using SingleR. (B) Quantification of the proportion of each cell population from A within each of the treatment groups, showing enrichment of granulocytes with DR-18 treatment and CD8⁺ and CD4⁺ T cells with DR-18 + anti–CTLA-4. For select cell populations (boxed), the percentages within each treatment group are shown. (C) Neighborhood group plot from Milo analysis of T cell subsets from scRNA-Seq data. (D) Differential abundance fold changes of the neighborhood groups in C, comparing the Combo treatment with control, showing enrichment and deenrichment of certain groups. (E) Heatmap of the top differentially expressed genes between neighborhood group #7, enriched with DR-18 + anti–CTLA-4 treatment and with high expression levels of markers of T cell activation, cytolytic activity, and exhaustion, versus neighborhood group #4, deenriched with combination treatment. (F) Relative proportion of the top 20 clonotypes out of the total for each treatment group based on TCR analysis. (G) Clonotype proportions by size category based on TCR analysis, showing clonal expansion with DR-18 + anti–CTLA-4 (Combo). Statistical testing performed using Fisher’s exact test comparing control with all other treatment conditions, with only significant comparisons shown F, and χ² test comparing DR-18 + anti–CTLA-4 (Combo) to all other conditions. *P < 0.05; ****P < 0.0001

Figure 6. DR-18 + anti–CTLA-4 leads to intratumoral expansion of proinflammatory myeloid populations.

(A and B) UMAP plots of all macrophages/monocytes identified by scRNA-Seq analysis with overlaid treatment groups (A) and annotated clusters (B). Annotation was performed based on the phenotypic groups and markers described in Ma et al. (18). (C) Quantification of the proportion of each macrophage/monocyte subtype from B within each of the treatment groups, showing relative enrichment of proinflammatory and loss of protumorigenic subtypes. For select cell populations (boxed), the percentages within each treatment group are shown. (D) UMAP plot of all granulocytes identified by scRNA-Seq analysis with overlaid treatment groups. (E) Volcano plot of differential gene expression between granulocytes from tumors treated with combination DR-18 + anti–CTLA-4 (Combo) versus all other treatment groups (Other) (log₂ fold-change thresholds of 0.5 and –0.5; P value-adjusted threshold of 1 × 10^–6). (F) The top gene sets from enrichment analysis of genes enriched in granulocytes from Combo-treated tumors. (G and H) UMAP plot of all neutrophils from scRNA-Seq analysis with overlaid neutrophil subtype classification based on Zilionis et al. (25) (G), with quantification of the relative proportion of each subtype by treatment group (H). For select cell populations (boxed), the percentages within each treatment group are shown. (I) UMAP plots of neutrophils showing trajectory analysis using Slingshot from the given starting point, with overlaid treatment groups (left) and neutrophil subtypes (right), as in G.