Cancer-associated fibroblasts induce antigen-specific deletion of CD8 + T Cells to protect tumour cells

Abstract

Tumours have developed strategies to interfere with most steps required for anti-tumour immune responses. Although many populations contribute to anti-tumour responses, tumour-infiltrating cytotoxic T cells dominate, hence, many suppressive strategies act to inhibit these. Tumour-associated T cells are frequently restricted to stromal zones rather than tumour islands, raising the possibility that the tumour microenvironment, where crosstalk between malignant and "normal" stromal cells exists, may be critical for T cell suppression. We provide evidence of direct interactions between stroma and T cells driving suppression, showing that cancer-associated fibroblasts (CAFs) sample, process and cross-present antigen, killing CD8⁺ T cells in an antigen-specific, antigen-dependent manner via PD-L2 and FASL. Inhibitory ligand expression is observed in CAFs from human tumours, and neutralisation of PD-L2 or FASL reactivates T cell cytotoxic capacity in vitro and in vivo. Thus, CAFs support T cell suppression within the tumour microenvironment by a mechanism dependent on immune checkpoint activation.

Fig. 1

Cancer associated fibroblasts engulf and proteolytically process cellular debris and antigen. a Relative permeability of low-molecular weight dextran (LMWD) and high-molecular weight dextran (HMWD) across monolayers of normal (NORM, white bars) or cancer-associated fibroblasts (CAF, black bars). b, c Apparent permeability across NORM and CAF monolayers of increasing HMWD (b) and LMWD concentrations (c). d Representative confocal micrograph including z-plane of HMWD (green) within CAFs (red, Cell Mask). Scale bar, 10 μm. e Representative confocal snapshot of CAF (red) interacting with dead and dying tumour cells (green). Twenty-four hours after debris engulfment showing CAF (red, f), and intracellular tumour cell debris (green, g). h Uptake of FITC-labelled OVA by fibroblasts and tumour cells. i Processing of DQ-OVA by fibroblasts and tumour cells. j Representative micrographs of endosomal compartments (red) showing early endosomes (EEA1) and lysosomes (LAMP1) at 15 and 120 min post ovalbumin pulse (green). Areas of co-localisation (white arrows) and ovalbumin not within labelled compartments (blue arrows) are indicated in each image. Scale bar: 10 μm. Data shown as mean ± SEM. a *P < 0.05 vs. NORM fibroblasts (two-tailed unpaired Student’s t-test). h, i ****P < 0.0001 (two-way ANOVA with Tukey post hoc analysis vs. CAFs). NS, not significant. Assays performed in duplicate from three (h) or two independent experiments (i)

Fig. 2

CAF cross-presentation of processed antigen protects tumour cells from T cell killing. a Quantification of MHC I SIINFEKL detected in FRCs, CAFs and normal fibroblasts following OVA pulse (white bars), and when pulsed in tumour conditioned media (grey bars). b MHC I SIINFEKL detection in the presence of ammonium chloride (grey bars), chloroquine (black bars) and vehicle control (white bars). c MHC I SIINFEKL detected on CAFs after co culture with B16.OVA tumour cells, or CAFs pulsed with soluble OVA. d T cell killing of control parental (B16.F10) and antigen bearing (B16.OVA) target tumour cells in the absence of fibroblasts. e, f B16 viability when introduced to OT-I T cells previously conditioned in the presence/absence of normal fibroblasts (e), CAFs (f) and presence/absence of antigen as indicated on graphs. g Lung tumour cell viability when introduced to OT-I T cells previously conditioned in the presence/absence of normal fibroblasts (yellow bars), CAFs (red bars) and presence/absence of antigen as indicated on graphs. Data shown as mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001, one-way ANOVA with Tukey post hoc analysis. NS, not significant. Assays performed in triplicate (a, c, g) or duplicate (b) from two experiments or (d-f) three experiments. Comparisons indicated by horizontal lines

Fig. 3

CAFs reduce CD8⁺ T cell viability in an antigen-specific death ligand dependent manner. a-c Quantification of T cell viability after conditioning in culture (a) or in the presence of normal fibroblasts (b), CAFs (c) and presence/absence of antigen and target tumour cells as indicated on graphs. d Representative micrographs of T cells (green) co-cultured with normal fibroblasts (red, f-actin) and CAF (red, f-actin) showing intact T cells (yellow arrows), T cells surrounded by actin bundles (blue arrows), and fragmented T cells (white arrows). Nuclei stained with DAPI (blue). Quantification of PD-1 expression (e-g) and FAS expression (h-j) on T cells following culture with normal fibroblasts (f, i), CAFs (g, j). k-m Quantification of cognate ligand expression by co-cultured fibroblasts by flow cytometry; FASL (k), PD-L2 (l) and PD-L1 (m) expression by NORM (yellow bars) and CAFs (red bars) with and without antigen and antigen-specific OT-I T cells as indicated. n B16.OVA tumour cell viability following culture with CAFs, antigen and OT-1 T cells following CAF-specific neutralisation of PD-L2 and/or FASL. Data shown as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA with Tukey’s or Dunnett’s post hoc analysis. NS, not significant. Comparisons indicated by horizontal lines. Scale bar 40 μm. Assays performed in or quadruplet from two experiments (a-c) or duplicate from three (e-m) or two (n) independent experiments

Fig. 4

CAFs process and present model antigen and kill antigen-restricted T cells via PD-L1/2 and FASL. a Schematic of in vivo experimental design. Following intratumoural injection of DQ-OVA, tumours and lymph nodes were collected for analysis. b DQ-OVA processing by CAFs, LECs or BECs as a percentage of live CD45⁻ cells. c MHCI-SIINFEKL expression by DQ-OVA-negative CAFs compared with DQ-OVA-positive CAFs. d In vivo PD-L1 (white bars), FASL (grey bars) and PD-L2 (black bars) expression by tumour cells (Tum) and CAFs (CAF). e Representative confocal micrograph of B16.OVA tumour for podoplanin (CAFs, red), PD-L2 (green) and FASL (white). Scale bar: 30 μm. f Schematic of in vivo blocking antibody experiments. g Day 9 tumour volumes after PD-L2 neutralisation. h MHCI-SIINFEKL pentamer-specific CD8⁺ T cell frequency within control and treated tumours. i Day 9 tumour volumes after FASL neutralisation. j MHCI-SIINFEKL pentamer-specific CD8⁺ T cell frequency within control and treated tumours. k Publically available microarrays analysed for PD-L2 in normal fibroblasts and CAFs from human lung (GSE22862), colon (GSE46824 and GSE1257), pancreatic (GSE21440) and breast cancers (GSE29270). l Representative image of tumour-stroma interface of stage III lung adenocarcinoma from tissue microarray; CAFs (podoplanin, red), PD-L2 (green), nuclei (DAPI, blue). Dashed line: tumour border. Demonstrable staining was not detected in normal lung tissues. Scale bar: 50 μm. b, d Data shown as mean ± SEM. *P < 0.05, ***P < 0.001 (one-way ANOVA with Tukey’s or Dunnett’s post hoc analysis). NS, not significant. b n = 18 tumours from three experiments. c Data shown as mean ± SEM. *P < 0.05. n = 12–15 tumours from two experiments. g–k Data shown as mean ± SEM. *P < 0.05, **p < 0.01. (two-tailed unpaired Student’s t-test). n = 4 and six tumours from two experiments. Symbols represent individual tumours (b, c, g–j) or humans sample (k)