Endogenous Glucocorticoid Signaling Regulates CD8+ T Cell Differentiation and Development of Dysfunction in the Tumor Microenvironment

Abstract

Identifying signals in the tumor microenvironment (TME) that shape CD8⁺ T cell phenotype can inform novel therapeutic approaches for cancer. Here, we identified a gradient of increasing glucocorticoid receptor (GR) expression and signaling from naïve to dysfunctional CD8⁺ tumor-infiltrating lymphocytes (TILs). Conditional deletion of the GR in CD8⁺ TILs improved effector differentiation, reduced expression of the transcription factor TCF-1, and inhibited the dysfunctional phenotype, culminating in tumor growth inhibition. GR signaling transactivated the expression of multiple checkpoint receptors and promoted the induction of dysfunction-associated genes upon T cell activation. In the TME, monocyte-macrophage lineage cells produced glucocorticoids and genetic ablation of steroidogenesis in these cells as well as localized pharmacologic inhibition of glucocorticoid biosynthesis improved tumor growth control. Active glucocorticoid signaling associated with failure to respond to checkpoint blockade in both preclinical models and melanoma patients. Thus, endogenous steroid hormone signaling in CD8⁺ TILs promotes dysfunction, with important implications for cancer immunotherapy.

Keywords: CD8(+) T cell; Nr3c1; TCF-1; cancer; dysfunction; exhaustion; glucocorticoid; immune checkpoint blockade; steroid; tumor-associated macrophages.

Figure 1:. A gradient of glucocorticoid receptor expression and signaling in CD8⁺ TILs

GR expression in TILs harvested from mice bearing MC38-Ova^dim colon carcinoma (tumor size 100–120 mm²) (A,B) or from human colon carcinoma (C). A) Representative histograms of GR expression and summary data of mean fluorescence intensity (MFI) in the indicated CD8⁺ TILs populations. (n=5) B) Representative histograms of GR expression and summary data of MFI in OVA-specific CD8⁺ TILs. (n=5) C) Representative histograms of GR expression and summary data of MFI in CD8⁺ TILs. Data are normalized to the expression level in Tim-3^-PD-1^- CD8⁺ TILs. (n=7) D) tSNE plot showing projection of a (I) GC signature, (II) naïve CD8⁺T cell signature, (III) CD8⁺ T cell dysfunction signature onto the single-cell RNA profiles of CD8⁺ TILs (Singer et al., 2016). The contour marks cells showing highest expression and the color scale indicates low (blue) to high (red) expressing cells. (IV) Each cell in the dataset was scored for the three normalized signatures: GC, Naïve, and Dysfunction. Cells were then sorted based on their expression of the glucocorticoid signature from low (blue) to high (red) (x-Axis). The y-axis indicates the naive (blue) and dysfunction (red) signature score for each of the sorted cells. Moving average (shaded area) and smoothing conditional means (solid line) was used to aid visualization. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. One-way ANOVA (Tukey’s multiple comparisons test) or unpaired Student’s t test. Mean ± SEM is shown.

Figure 2:. Glucocorticoid signaling promotes checkpoint receptor expression and dampens CD8⁺ T cell effector functions

Murine (A-B) or human (C) naïve CD8⁺ T cells were repeatedly activated (anti-CD3/28) in the presence or absence of GC (Dex). Data shown are representative of 3 independent experiments B) Representative flow cytometry data and summary plots of the frequency and MFI the indicated cytokines following polyclonal activation (n=5) B and C) Representative flow cytometry data and summary plots of the frequency and MFI of the indicated checkpoint receptors (n=5 for B), (n=6 for C) *p<0.05, **p <0.01, ***p < 0.001, ****p<0.0001, unpaired Student’s t test. Mean ± SEM is shown.

Figure 3:. Glucocorticoid signaling regulates effector differentiation in CD8⁺ TILs

A) MC38-Ova^dim was implanted into WT (E8i-Cre^-Nr3c1^fl/fl) and E8i-Cre⁺Nr3c1^fl/fl mice (n=8–9). Mean tumor growth is shown, ***p< 0.001, linear mixed model. Data are representative of 3 independent experiments. B-G) TILs were harvested from mice bearing MC38-Ova^dim at early (size 40–60 mm²) and intermediate (size 120–150 mm²) stages of tumor progression as determined by the growth observed in WT controls. B) Representative flow cytometry data and summary plots of the frequency of OVA-specific CD8⁺ TILs at early (n=7) and intermediate (n=4) stages. (C-E) TILs were activated with OVA_257–264 followed by intracellular staining. C) Representative flow cytometry data and summary plots of the frequency of the indicated cytokines in CD8⁺ TILs at early (n=7) and intermediate (n=9–10) stages. Data are pooled from 2 independent experiments for the intermediate stage. D) Representative flow cytometry data and summary plot of frequency of CD107a⁺ GzmB⁺ CD8⁺ TILs at early (n=7) and intermediate (n=6) stages. E) Representative flow cytometry data and summary plot of frequency of IL10-producingCD8⁺ TILs at early (n=7) and intermediate (n=5) stages. F) Representative flow cytometry data and summary plot of frequency of TCF-1⁺ cells within Ova-specific CD8⁺ TILs at early (n=7) and intermediate (n=4) stages. G) Representative flow cytometry data and summary plot of frequency of checkpoint receptor expressing CD8⁺ TILs at early (n=7) and intermediate (n=6–7) stages. NS, not significant, **p<0.01, ***p< 0.001, unpaired Student’s t-test. Mean ± SEM are shown. H) Experimental design: congenically marked WT (blue) and E8i-Cre⁺Nr3c1^fl/fl (red) CD8⁺ T cells were transferred to Rag^−/− recipients along with WT CD4⁺ T cells (green). MC38-Ova^dim was implanted 2 days post T cell transfer. TILs were harvested at the intermediate stage of tumor growth and analyzed (n=6). NS, not significant, *p< 0.05, **p<0.01, ***p<0.001, ****p<0.0001, unpaired Student’s t-test or paired Student’s t-test (H). Mean ± SEM are shown.

Figure 4:. Glucocorticoid signaling transactivates checkpoint receptor and IL-10 expression and induces T cell dysfunction genes

(A-E) Luciferase activity in 293T cells transfected with pGL4.23 or pGL4.10 luciferase reporters for the loci of the indicated checkpoint receptors or IL10 together with either empty vector (control) or vector encoding Nr3c1. Cells were treated with GC (Dex) after 24h. Firefly luciferase activity was measured 48 h after transfection and is presented relative to constitutive Renilla luciferase activity. NS, not significant, ****p<0.0001, two-way ANOVA (Tukey’s multiple comparisons test). Data are mean ± SEM and are representative of 2 independent experiments. F) Volcano plot showing the overlap of genes suppressed by GC (Dex) with genes expressed in Tim-3^-PD-1⁻ CD8⁺ TILs (p=1.4.0×10⁻²⁶) and genes induced by GC (Dex) with Tim-3⁺PD-1⁺ CD8⁺ TILs (p=9.4×10⁻⁵²) (Mean-rank Gene Set Test).

Figure 5:. Intra-tumoral production of glucocorticoid affects tumor progression

A) Pregnenolone levels in the indicated tissues were quantified by ELISA (n=5). B) qPCR analysis of Cyp11a1 mRNA expression in the indicated cells. Data are pooled from 2 independent experiments (n= 5–6). C) MC38-Ova^dim was implanted in LysMCre⁻ Cyp11a1^fl/fl and LysMCre⁺Cyp11a1^fl/fl mice (n=5). Mean tumor growth is shown, ***p<0.001, linear mixed model. Data are representative of 2 independent experiments. D) Analysis of CD8⁺TILs at early stage of tumor development (tumor size 40–60 mm²) (n=5) E) Corticosterone levels were quantified by ELISA (n=5). F) Lin⁻CD45⁺CD24⁻ monocyte-macrophage lineage cells were isolated from MC38-Ova^dim tumors and cultured in the presence or absence of Metyrapone. At 24hrs corticosterone levels were quantified by ELISA (n=5). G) MC38-Ova^dim was implanted in WT mice (n=5). Metyrapone or vehicle control was administered intra-tumorally on Days 5,6,7 and 9 post-tumor implantation. Mean tumor growth is shown ***p< 0.001, linear mixed model. Data are representative of 2 independent experiments. H) MC38-Ova^dim was implanted in WT mice (n=5). Metyrapone or vehicle control was administered intra-tumorally on Days 5 and 6 post-tumor implantation. 24hrs later, TILs were harvested (tumor size 55–65 mm² in both groups) and analyzed by flow cytometry. Summary plots show the frequency of the indicated populations. ND, not detected, NS, not significant *p<0.05 **p<0.01, ***p<0.001, ****p<0.0001, unpaired Student’s t-test or One-way ANOVA (Tukey’s multiple comparisons test). Data are mean ± SEM.

Figure 6:. Glucocorticoid signaling in CD8⁺ T cells affects responses to immunotherapy

A) Correlation of Cyp11a1 mRNA expression with survival in patients with colon adenocarcinoma (COAD) and stomach adenocarcinoma (STAD) using TIMER. B) MC38-Ova^dim was implanted into WT (E8i-Cre⁻Nr3c1^fl/fl) and E8i-Cre⁺Nr3c1^fl/fl mice (n=7–8). Anti-PD1 was administered i.p on Days 5,8 and 11. Mean tumor growth is shown. NS, not significant, ****p< 0.0001, linear mixed model. C) MC38 was implanted into WT mice. On Day 7 post-tumor implantation, mice were treated with GC (Dex) or anti-PD1+anti-CTLA-4 or both. Antibody was administered bi-weekly for a total of 5 treatments (n=6–10). GC was administered for 10 consecutive days. NS, not significant, *p<0.05, ***p<0.001, linear mixed model. D) tSNE plot of single-cell TILs data from melanoma patients treated with anti-PD-1, anti-CTLA-4, or anti-CTLA-4 + anti-PD-1(Sade-Feldman et al., 2018). I) CD8 expression, II) CD4 expression, III) pre- (orange) versus post- (purple) treatment samples, IV) Responder (red) versus non-responder (blue), V) Projection of CD8⁺ TILs dysfunction signature, VI) Projection of the GC signature. VII) Box plots show the GC signature score in responder versus non-responders in pre- (p=3.246×10⁻¹³) and post- (p <2.2×10⁻¹⁶⁾ treatment samples (Welch Two Sample t-test). The lower and upper hinges correspond to the first and third quartiles. The upper and lower whiskers extend from the hinge to the largest and smallest value no further than 1.5 times the distance between the first and third quartiles, respectively. Data beyond the end of the whiskers are outlying points and are not plotted individually.

Figure 7:. Glucocorticoid and IL-27 signaling co-operate to regulate CD8⁺ T cell phenotype in the TME

A-C) Naïve CD8⁺ T cells were cultured in vitro with anti CD3/28 and GC (dexamethasone), IL-27, or GC+IL-27. Cells were harvested on Day 9 and gene expression analyzed by RNA sequencing. A) Principle component analysis (PCA) of Ctrl, GC, IL-27, and GC+IL-27 treated CD8⁺ T cells. The percentage of explained variance for each principal component is indicated. B) Mean delta Euclidean distance between the GC, IL-27, or GC+IL-27-treated groups to thecontrol group, adjusted p-values were calculated using one-way ANOVA (p=9.89×10⁻⁰⁹), followed by Tukey’s multiple comparisons test, *p<0.05, ****p<0.001. C) Heatmap of DE genes between Ctrl and GC + IL-27 treatment. Tick marks indicateselected genes associated with CD8⁺ T cell dysfunction. D) CD8⁺ T cells from either WT (E8i-Cre^-Nr3c1^fl/fl), E8i-Cre⁺Nr3c1^fl/fl, WSX1^−/− or and E8i-Cre⁺Nr3c1^fl/fl WSX1^−/− (DKO) mice and CD4⁺ T cells from WT mice were transferred to Rag^−/−mice (n=5–6/group), MC38-Ova^dim cells were implanted two days post T cell transfer. Mean tumor growth is shown. *p<0.05, ***p< 0.001, linear mixed model. Data are representative of 2 independent experiments. E) qRT-PCR analysis of IL-27 (p28 and Ebi3) mRNA expression in the indicated cells. Dataare pooled from 2 independent experiments. **p<0.01, ***p<0.001, ****p<0.0001. One-way ANOVA (Tukey’s multiple comparisons test). Data are mean ± SEM.