Leukemic extracellular vesicles induce chimeric antigen receptor T cell dysfunction in chronic lymphocytic leukemia

Abstract

Chimeric antigen receptor (CAR) T cell therapy has yielded unprecedented outcomes in some patients with hematological malignancies; however, inhibition by the tumor microenvironment has prevented the broader success of CART cell therapy. We used chronic lymphocytic leukemia (CLL) as a model to investigate the interactions between the tumor microenvironment and CART cells. CLL is characterized by an immunosuppressive microenvironment, an abundance of systemic extracellular vesicles (EVs), and a relatively lower durable response rate to CART cell therapy. In this study, we characterized plasma EVs from untreated CLL patients and identified their leukemic cell origin. CLL-derived EVs were able to induce a state of CART cell dysfunction characterized by phenotypical, functional, and transcriptional changes of exhaustion. We demonstrate that, specifically, PD-L1⁺ CLL-derived EVs induce CART cell exhaustion. In conclusion, we identify an important mechanism of CART cell exhaustion induced by EVs from CLL patients.

Keywords: CART cell exhaustion; chimeric antigen receptor T cells; chronic lymphocytic leukemia; extracellular vesicles; microenvironment.

Graphical abstract

Figure 1

Identification of CLL-derived extracellular vesicles (EVs) in patients with CLL (A–E) Dot plots showing total particle number (A) and EV levels (B–E) measured by nanoscale flow cytometry in platelet-poor plasma isolated from normal individuals (n = 10) and CLL patients (n = 50). (B–E) A panel of fluorescent antibodies was used to enumerate levels of EVs for (B) CD45⁺, (C) CD19⁺, (D) CD5⁺CD19⁺, and (E) PD-L1⁺. Values represent number of EVs per microliter transformed in a logarithmic scale (Mann-Whitney test; error bars, SD). (F) Correlation analysis of levels of CLL-derived CD5⁺CD19⁺ EVs and PD-L1⁺ EVs in CLL patients. Pearson correlation coefficient was calculated with a two-tailed p value. (G) Western blot showing expression of three EV-enriched markers (TSG101, CD9, CD81) and PD-L1 in a panel of six EV lysates obtained from platelet-poor plasma of CLL patients. A second band at higher molecular weight was detected for PD-L1 that corresponds to a glycosylated form of the protein. (H) Relative intensity of gel bands for total PD-L1 (left panel) and glycosylated PD-L1 (right panel). Levels of PD-L1 were increased by 1.4-fold (minimum [min]-maximum [max], 1.17–1.77). The glycosylated form of PD-L1 was markedly increased in PD-L1^high patients with a 3.0-fold increase (min-max, 1.78–4.37) compared to PD-L1^low patients. (I) Correlation analysis of levels of total PD-L1 (left panel) and glycosylated PD-L1 (right panel) between western blot and nanoscale flow cytometric quantification methods. Pearson correlation coefficient was calculated with a two-tailed p value.

Figure 2

CLL-derived EVs induce a state of CART cell dysfunction (A and B) Inhibitory receptor expression on activated CART cells is upregulated by CLL-derived EVs. CART19 cells were co-cultured for 24 h with JeKo-1 cells with different concentrations of EVs (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, one-way ANOVA; error bars, SEM; three biological and two technical replicates, two experiments). (C and D) CART19 cell antigen-specific proliferation and killing of CD19⁺ JeKo-1 cells were decreased in the presence of CLL-derived EVs (blue triangles) compared to controls (orange squares). EVs/CART19 cells at a 100:1 ratio were co-cultured for 6 h and plated at a 5:1 effector-to-target ratio (E:T ratio) with JeKo-1 cells (∗∗∗∗p < 0.0001, one-way ANOVA; error bars, SEM; three biological and two technical replicates, three experiments). (E) CART19 cell antigen-specific proliferation was further decreased in the presence of CLL-derived EVs at 1,000:1 and 10,000:1 compared to 100:1. EVs/CART cells were co-cultured for 6 h and plated at a 5:1 E:T ratio with JeKo-1 cells (∗p < 0.05, ∗∗∗∗p < 0.0001, one-way ANOVA; error bars, SEM; three biological and two technical replicates, one experiment). (F) Treatment of JeKo-1 cell xenografts with CART19 cells alone (red squares) improved survival compared to CART19 cells co-cultured with CLL-derived EVs (purple triangles) or untransduced (UTD) T cells (blue circles). NOD-SCID-γ^−/− mice were engrafted with the CD19⁺ luciferase⁺ cell line JeKo-1 (1 × 10⁶ cells intravenous [i.v.] via tail vein injection), and engraftment was confirmed through bioluminescence imaging (total flux, photons [p]/s). Mice were randomized to treatment with (1) UTD T cells, (2) CART19 cells, and (3) CART19 cells co-cultured ex vivo with CLL-derived EVs for 6 h prior to injection. All T cells were washed prior to injection. A single low dose of CART19 cells (2.5 × 10⁵) was injected to induce relapse (∗p = 0.0198, log-rank test; five mice per group).

Figure 3

EVs from CLL patients induce phenotypical, functional, and transcriptomic changes of exhaustion in T cells (A) CLL-derived EVs do not express E-cadherin. E-cadherin was measured on EVs derived from normal donor (ND) and CLL patients by nanoscale flow cytometry compared to measurements of CD19 on CLL-derived EVs (∗p < 0.05, one-way ANOVA; error bars, SEM; three to five biological replicates, two technical replicates, one experiment). (B) CLL-derived EVs decrease E-cadherin CART cell antigen-specific proliferation. EVs/CART cells at a ratio of 100:1 were co-cultured for 6 h and plated at an E:T ratio of 5:1 with the E-cadherin⁺ breast cancer cell line MCF-7 (∗∗∗∗p < 0.0001, one-way ANOVA; error bars, SEM; three biological replicates, two technical replicates). The absolute number of live T cells significantly decreased when E-cadherin CART cells were co-cultured with MCF-7 cells in the presence of CLL-derived EVs (blue triangles) compared to E-cadherin CART cells co-cultured with MCF-7 alone (orange squares). The UTD negative control (pink circles) shows background proliferation. (C and D) CART19 cell transcriptome is modulated by CLL-derived EVs. CART19 cells were co-cultured with irradiated JeKo-1 cells for 24 h at a ratio of 10:1, 1:1, or 0:1 EVs/CART19 cells and then isolated by magnetic sorting (three biological replicates, adjusted p value < 0.05). Gene expression with 10:1 EVs/CART19 cells (green columns) and 1:1 EVs/CART19 cells (blue columns) compared to CART19 cells alone (salmon column). EVs increase the expression of AP-1 (FOS-JUN) and YY1. (E) Principal component analysis of CART19 cell RNA-sequencing samples. Similar gene expression patterns were noted between both 1:1 EV/CART19 cell (blue circles) and 10:1 EVs/CART19 cells (red circles). (F) Ingenuity Pathway Analysis predicts increased activation of the AP-1 pathway (FOS-JUN, orange) in CART19 cells co-cultured with CLL-derived EVs. (G) Gene set enrichment analysis for significantly upregulated genes shows enrichment for pathways associated with CD4 (p = 0.037) and CD8 (p = 0.0033) T cell signaling as well as AP-1 transcription factors (p = 0.0445) (red bars, p < 0.05).

Figure 4

CART cell dysfunction is facilitated by PD-L1⁺ CLL-derived EVs (A and B) CART19 cells alone (red squares) control tumor burden better compared to CART19 cells co-cultured ex vivo with PD-L1^high CLL-derived EVs (purple triangles) (∗∗p = 0.0088, two-way ANOVA; error bars, SEM; five mice per group). NOD-SCID-γ^−/− mice engrafted with the CD19⁺luciferase⁺ cell line JeKo-1 Luc-ZsGreen (1 × 10⁶ cells i.v. via tail vein injection) and engraftment confirmed through bioluminescent imaging (total flux, p/s). Mice were then randomized for treatment with (1) UTD T cells, (2) CART19 cells, (3) CART19 cells co-cultured ex vivo with PD-L1^high CLL-derived EVs for 6 h prior to injection, or (4) CART19 cells co-cultured ex vivo with PD-L1^low CLL-derived EVs for 6 h prior to injection. A single low dose of CART19 cells (2.5 × 10⁵) was injected to induce relapse. Mice treated with UTD T cells (blue squares) had continued progression of disease. Mice treated with CART19 cells that were pre-cultured with PD-L1^low CLL-EVs had a non-statistically significant impairment of anti-tumor activity (green triangles). Mice treated with CART19 cells that were pre-cultured with PD-L1^high CLL-EVs had significant impairment of anti-tumor activity. (C) CLL-derived PD-L1^high EV impairment of CART19 cells is not significantly reversed by PD-L1 blockade. CART19 cells were co-cultured for 6 h with and without PD-L1^high CLL-derived EVs (100:1 EV/CART cell ratio) and with and without anti-PD-L1 antibody. CD19⁺ JeKo-1 cells were added at an E:T ratio of 5:1. CART19 cell antigen-specific proliferation was significantly impaired in the presence of PD-L1^high CLL-derived EVs (p < 0.01, two-way ANOVA). This inhibited CART19 cell antigen-specific proliferation did not improve following a co-culture with anti-PD-L1 antibody (n = 11 biological replicates, two technical replicates, four experiments).