Fig 2: Activity of CAR T cells expressing VHH variants.(A) Flow cytometric histograms showing the cell surface expression of JZQ-B4 VHH CAR variants in primary CAR T cells. (B) Surface expression of human MSLN in tumor cell lines. NCI-H226/KO represents NCI-H226 with human MSLN knockout using a CRISPR/Cas9 system. MSTO-211H/hMSLN is MSTO-211H cell line that was transduced for human MSLN overexpression. (C) Bioluminescence-based cytotoxicity assay using NT cells as control. CAR T cell samples were normalized to contain the same percentage of CAR-expressing cells (50% CAR+) using donor-matched NT cells. Target cells were cocultured with T cells at an effector-to-target ratio of 2.5:1, and the percentages of target cell lysis were normalized to target cell–only group at 24 hours after coculture. Data represent mean ± SD of 4 technical replicates from 1 independent experiment.

Fig 3: Alanine scanning of JZQ-B4 VHH using a yeast surface display system.(A) The schematic of the cell surface display of the VHH variants (created with BioRender.com). Different VHH variants were subcloned into the vector within NheI-BamHI restriction sites. (B) The schematic illustrating the topology of the fusion proteins on the yeast cell surface (created with BioRender.com). Aga2p is disulfide bonded to Aga1p, forming α-agglutinin. The HA tag at the N-terminus was used for the detection of displayed fusion proteins while Alexa Fluor 647–labeled human MSLN was used to measure the affinity of VHHs. (C) The bar graph shows relative binding affinities of VHH variants with alanine substitutions in CDRs, normalized to the level of the parental JZQ-B4 (n = 1 culture per variant). (D) Dot plots are shown for yeast cells expressing the parental JZQ-B4 and CDR3 VHH variants after staining with anti-HA antibody and Alexa Fluor 647–labeled human MSLN.