Aberrant PD-L1 Expression in CLL As a Result of Adaptive Immune Resistance Mediated By Tumor-Secreted Circulating miRNA Binding to Toll-like Receptor 7

Background: Chronic lymphocytic leukaemia (CLL) is a model cancer to study immune evasion via PD-L1/PD-1 signalling: aberrantly expressed PD-L1 on CLL and PD-1 on CD8 T cells are key mediators of poor anti-tumor immune responses, which are a major hallmark of CLL. Several preclinical studies suggest that aberrant PD-L1 expression is a result of adaptive immune resistance and is induced during immune responses within the tumor microenvironment. It has recently been proposed that specific circulating microRNAs (miRNAs) shed by malignant cells participate in the complex crosstalk between cancer cells and microenvironment, and that they activate immune cells via Toll-Like Receptor (TLR) 7 due to their structural similarity to its natural ligand. In addition, TLR signaling has been demonstrated to result in the upregulation of costimulatory molecules such as CD80, CD86 and PD-L1/2. In CLL, aberrant circulating miRNA and TLR expression patterns have been well characterized. Therefore, we hypothesized that aberrant PD-L1 expression in CLL is a result of continuous TLR7 signaling mediated by circulating miRNAs. Our aims were to demonstrate that (1) specific circulating miRNAs induce PD-L1 expression and have functional consequences, (2) miRNA/PD-L1 associations are mediated via TLR7, and (3) miR/TLR/PD-L1 interactions are subject to the dynamics of CLL development.

Methods and Materials: Mononuclear cells from spleen cell suspensions from 3 month old TCL1 transgenic, wild-type (WT) or TLR7-/- mice (total n=13) were treated ex vivo for 18hours with liposomal formulations of synthetic scrambled miRNA or miRNAs -16 (negative control), -21, -29, -150 and -155, which are reported to have an effect on immune cells and to be released by CLL cells. Specific TLR7 and TLR9 agonists were included for comparison. Primary human CLL cells and healthy B cells were treated with specific TLR2/6, TLR7 and TLR9 agonists. In adoptive transfer experiments, young WT mice (n=15) were injected with 4x10⁷CLL cells from TCL1 transgenic mice. Mice were sacrificed at days 3, 6, 9, 12 and 15, and spleen cells were treated ex vivo as above. Changes in surface PD-L1, CD69 and CD86 expression on DAPI-negative CD19+ B cells/CLL cells were determined by flow cytometry. Supernatant cytokines were screened by multiplex ELISA.

Results: PD-L1 surface expression on spleen B cells from both WT and non-leukemic TCL1 mice was strongly induced by miRs -21 and -29, and moderately by -150 and -155, but not by miR-16 negative control. The degree of PD-L1 upregulation by miRs -21 and -29 was comparable to the effect of direct TLR7 and TLR9 binding by specific agonists. Similar patterns were seen for CD69 and CD86 expression. Across treatment conditions, PD-L1 expression was highly correlated with the expression of CD69 (r .7777, p<.0001) and CD86 (r .7516, p<.0001). This observation strongly suggests that PD-L1 expression after TLR engagement is a marker of activation/costimulation, and therefore a physiological adaptive immune response to TLR binding in healthy B cells. Functionally, miR treatments resulted in increased IL-6, IL-10 and TNFα, with miR-29 having the strongest effect. PD-L1, CD69 and CD86 could also be induced by TLR engagement in healthy B cells, but not in CLL patient cells, where PD-L1 was confirmed to be already aberrantly expressed. To elucidate when in the course of CLL development PD-L1 expression becomes aberrant and if it ceases to be inducible by miR treatment, we sacrificed adoptively transferred mice every 3 days to simulate tumor development. With increasing CLL the magnitude of the fold-change of PD-L1 expression following miR treatment decreased substantially, and the baseline expression of miR-untreated B cells increased consistently until day 15. Interestingly, although the PD-L1 response was substantially decreased with tumor load, it was not completely abrogated, even on day 15 when mice had a median CLL load of 71%. Importantly, miR treatment did not result in increased PD-L1, CD69 or CD86 expression in B cells from TLR7-/- mice, indicating that the miR/PD-L1 interactions are indeed mediated by TLR7.

Conclusions: Our findings support that PD-L1 expression on B cells can be induced by specific miRNAs known to be produced by CLL cells, and that this effect is mediated via TLR7. Therefore, aberrant PD-L1 expression on CLL is likely to be a result of adaptive immune resistance mediated by tumor cell-produced circulating miRNAs.