Fig 2: TCGA dataset analyses of basal-like breast cancer patients reveal a highly significant correlation between the expression of TNFa, IL-1ß and IFN? and the expression of PD-L1 and PD-L2. The TCGA breast cancer dataset was used to identify correlations in gene expression patterns in basal-like patients. n = 141. (A) Associations of TNFa (A1), IL-1ß (A2) and IFN? (A3) with PD-L1. (B) Associations of TNFa (B1), IL-1ß (B2) and IFN? (B3) with PD-L2. Correlation coefficients and p values are demonstrated in the Figure. Statistical analyses were performed as described in Section 2.

Fig 3: The cell-autonomous activities of PD-L1 shorten the lag period of tumor appearance in vivo and depend on the integrity of the S283 residue of PD-L1. (A) Description of the experimental procedure taken in order to determine the cell-autonomous roles of PD-L1 and its S283 residue in regulating tumor growth and metastasis. Tumor cells were administered to the mammary fat pad of female nude mice, deficient in T cell activities. This part of the Figure demonstrates the time points in which the mice of each group were sacrificed and analyzed for tumor volumes and metastasis formation. All cells were infected to express mCherry; tumor growth was determined by caliper every 3–4 days and was validated at different time points also by intravital imaging of the mammary region. Two independent experiments were performed, with the following mouse groups: (1) Group 1 = KO-PD-L1-MDA cells, with a total of n = 10 mice in the two experiments together. To follow up on the regulations of the Ethics Committee of Animal Use, the experiments of this group of mice were terminated at day 56 after tumor cell inoculation. Of note, as indicated in Figure 5(A1), in these cells the expression of endogenous PD-L1 was knocked out and the cells were infected with the control vector of PD-L1; thus, these cells do not express PD-L1 at all. (2) Group 2 = WT-PD-L1-MDA cells (generated as indicated in Figure 5(A1)), with a total of n = 9 mice in the two experiments together; the experiments of this group were terminated at day 49. (3) Group 3 = S283A-PD-L1-MDA cells (generated as indicated in Figure 5(A1)), with a total of n = 9 mice in the two experiments together, followed up until day 77. Follow up of five of the nine mice was continued until day 111. (B) Kinetics of tumor cell appearance. p values are demonstrated in the Figure.

Fig 4: Cytokine production and degranulation by PD-L1–CAR T cells following stimulation with breast cancer cells. (A) Functional and cytokine release assays of PD-L1–CAR T cells targeted against MCF-7 (PD-L1low/-) or MDA-MB-231 (PD-L1+) cancer cell lines. Degranulation assay, assessed by CD107a staining (left panel), TNFa release (middle panel), and IFN? release (right panel) were measured after 4 hours of coincubation of target and effector cells at the E:T ratio of 2:1. The experiment was repeated in duplicates three times. Bars represent the mean value±SD. The normality was checked using the Shapiro-Wilk test. The p values derived from unpaired t-test: ****p<0.0001. (B) Functional and cytokine release assays of PD-L1–CAR T cells targeted against MCF-7 pLVX (PD-L1low/-) and MCF-7 PD-L1 and MDA-MB-231 sgNTC (PD-L1+) or MDA-MB-231 sgPD-L1 (PD-L1-) cancer cell lines were assessed by flow cytometry. Degranulation assay, assessed by CD107a staining (left panel), and IFN? release (right panel) were measured after 4 hours of coincubation of target and effector cells at the E:T ratio of 2:1. The experiment was repeated in duplicate three times (for IFN? release by MCF-7 for two times). Bars represent the mean value±SD The p values derived from unpaired t-test or Mann-Whitney test depending on data distribution (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001). (C) The potential of killing tumor cells by control (pSEW-GFP) and PD-L1–CAR T cells was measured by impedance analysis for MDA-MB-231 cells. Cancer cell lines were left to adhere and form a monolayer on the E plates for 24 hours. The next day, PD-L1–CAR T cells or control (pSEW-GFP) T cells were added to the monolayers for 12 hours at the indicated E:T ratios. Representative mean impedance curves from two wells were shown. The experiment was repeated in duplicates two times. (D) The potential of killing tumor cells by control (pSEW-GFP) and PD-L1–CAR T cells was measured by impedance analysis for MCF7 cells. Cancer cell lines were left to adhere and form a monolayer on the E plates for 24 hours. The next day, PD-L1–CAR T cells or control (pSEW-GFP) T cells were added to the monolayers for 12 hours at the indicated E:T ratios. Representative mean impedance curves from two wells were shown. The experiment was repeated in duplicates two times. CAR, chimeric antigen receptor, PD-L1 - programmed death-ligand 1, TNFa - tumor necrosis factor a, IFN? - interferon ?, GFP - green fluorescent protein, SSC - side scatter

Fig 5: Workflow for generation of UM-SCC-49-GeCKO pool, cell sorting, and sequencing of final populations. 1. UM-SCC-49 was transduced with a lentiviral pooled sgRNA library (GeCKO v2). 2. Live cells were selected based on PDL1 expression using a PE-conjugated antibody directed against PDL1. PE fluorescence distribution was plotted on a histogram. The fractions of cells with the highest and lowest PE fluorescence (10% each) were expanded separately in culture and re-sorted to enhance respective phenotypes. 3. DNA was harvested from final populations for next-generation sequencing of sgRNA barcodes.