Dermal vascular endothelial cells are important targets for cGVHD by expressing DBY/HLA class II complexes. (A) PLAs to visualize protein–protein interactions between DBY and HLA-DR in cGVHD. Colocalization of DBY and HLA-DR induced PLA signals (red). Nuclei were stained with DAPI (blue). Bar plots indicate the PLA signals per vessel from the skin of female patients with cGVHD (n = 9) and patients with cGVHD with F-to-M HCT (n = 12). Data are presented as the mean ± SD. P values were calculated using unpaired 2-tailed Student t test. ∗P < .05. PLA signals of patients who underwent F-to-M HCT vs female patients. (B) Representative immunofluorescence images of CD31 (red) as an endothelial marker and α-smooth muscle actin (αSMA; green) as a myofibroblast marker in skin biopsy samples from healthy donors (n = 2) and patients with cGVHD (n = 7). (C) Analysis of CDC using anti-DBY/HLA class II complex antibodies. Purified total IgG from cGVHD plasma, including anti-DBY/HLA class II complex antibodies, were mixed with 293T cells transfected with mock (light green line), DBY (purple line), HLA-DRB1∗09:01 (dark green line), HLA-DRB1∗12:01 (orange line), HLA-DRB1∗15:02 (pink line), DBY and HLA-DRB1∗09:01 (light blue line), DBY and HLA-DRB1∗12:01 (dark blue line), or DBY and HLA-DRB1∗15:02 (red line) followed by incubation with complement. Dead cells were assessed using flow cytometry. Data are presented as the mean ± SD of 3 independent experiments. Mock: pSI vector. Scale bars: 20 μm (A) and 50 μm (B). Original magnification, ×400 (A) and ×200 (B). DAPI, 4',6-diamidino-2-phenylindole.