T-cell PD-1 engagement by tumoral PD-L1 is widely recognized as one of the main immunosuppressive mechanisms driving cytotoxic T-cell exhaustion. However, it is unclear how tumoral PD-L1 modulates immune evasion in non-T cell PD1+ immune populations, such as tumor-associated macrophages (TAMs). 
To interrogate this, we generated a PD-L1 KO TNBC-like tumor model in the murine EO771 cell line using CRISPR/Cas9 editing, allowing us to profile the immune infiltrates of the tumoral microenvironment (TME) in vivo during tumoral progression. 
Using flow cytometry (FC) to characterize the immune infiltrates of early vs late-stage WT tumors, we found a late-stage decrease in F480+ CD206+ populations, suggesting that M2 TAM polarization is inhibited during tumor development. Furthermore, analyzing PD-1 expression of immune infiltrates we observed an increase in PD1+ M2 TAMs at late-stage, suggesting that advanced tumors are more responsive to PD-L1. 
In addition, examining PD-L1 KO vs WT tumors at early and late-stage we found that tumoral PD-L1 inhibits M2 TAM polarization exclusively at late stage. Using bone marrow-derived macrophages in tumoral cells & conditioned media co-culture experiments, we found that M2 TAM inhibition is a direct effect that involves tumor cell-to-macrophage contact.
Interestingly, M2 TAMs from late-stage WT tumors showed increased MHCII+ expression, suggesting an improvement in antigen-presentation potential. Moreover, using FC to analyze GFP+ tumor cell phagocytosis, we found that tumor progression triggers phagocytosis exclusively in M2 TAMs. Comparing PD-L1 KO vs WT tumors, we observed that tumoral PD-L1 inhibits TAMs phagocytosis both in vivo and in vitro.
Altogether, these results suggest that M2 TAMs acquire anti-tumoral features during tumor progression and that tumoral PD-L1 dependent inhibition of M2 polarization plays a critical role in TAM immunosuppression during late stages of TNBC progression.