Journal: bioRxiv

Article Title: Actionable spatial prostanoid barriers constrain BiTE-driven adoptive T cell immunity in intact human tumors

doi: 10.64898/2026.03.26.713601

Figure Lengend Snippet: A, Schematic of the tiered benchmarking assay comparing anti-MSLN CAR-T cells and anti-MSLN BiTE T cells in OVCAR-8 2D monolayers and matrix-free 3D spheroids after 24 hrs coculture at a 1:1 effector-to-target (E:T) ratio. B, Representative flow cytometry plots and quantification of CD8 + CD137 + cells among CD8 + T cells after coculture of MSLN-CAR T cells alone (T cell only), with 2D OVCAR-8 cells, or with 3D OVCAR-8 spheroids. C, Representative flow cytometry plots and quantification of CD8 + CD137 + cells among CD8+ T cells after coculture of MSLN-BiTE T cells alone (T cell only), with 2D OVCAR-8 cells, or with 3D OVCAR-8 spheroids. D, Supernatant concentrations of granzyme B, TNFα, and IFNγ after 24-hour coculture of MSLN-CAR (top) or MSLN-BiTE (bottom) T cells with 2D OVCAR-8 cells or 3D spheroids. E, Representative confocal immunofluorescence images of embedded 3D spheroids stained for nuclei (DAPI), CD8, and MSLN at baseline and at 5 and 24 hrs after exposure to MSLN-CAR or MSLN-BiTE T cells. Scale bars, 100 μm. F, Quantification of intratumoral CD8 density in 3D spheroids across conditions. Statistical analysis using one-way ANOVA (+Kruskal-Wallis) G, Quantification of intratumoral CD8 density in 3D spheroids across conditions. H, Interpatient heterogeneity of MSLN expression in PDTs assessed by flow cytometry, schematic of Matrigel-free PDT coculture with MSLN-BiTE T cells at a 1:1 E:T ratio, and correlation between epithelial target abundance (%CK + MSLN + of CK + cells) and CD8 activation (FC of CD8 + CD137 + over UT control). I , Multiparametric functional profiling of PDT responses to MSLN-BiTE T cells. Upper part, CD8 activation (%CD8⁺CD137⁺) under untreated versus +MSLN-BiTE conditions. Below, heatmap summarizing changes in secreted mediators (log2 fold-change over UT control) following co-culture across the indicated tumoroid lines. One-way repeated-measures ANOVA (+Tukey) was performed in GraphPad Prism v10.2 (n=3-4 technical replicates per patients). Exact P values and comparisons are indicated in the panels. In B and C , representative plots are shown and quantification reflects 3 to 4 technical replicates per condition. In D , n = 3 technical replicates per condition. In F , n = 12 to 22 regions of interest per condition. Exact P values and comparisons are indicated in the panels. Figure created with BioRender.

Article Snippet: All image stacks were imported into Imaris software v10.0 (RRID:SCR_007370) and analyzed using surface detection and surface creation tools to generate three dimensional reconstructions of regions of interest and to quantify exogenous T cell infiltration within tumor compartments.

Techniques: Flow Cytometry, Immunofluorescence, Staining, Expressing, Activation Assay, Control, Functional Assay, Co-Culture Assay

Journal: bioRxiv

Article Title: Actionable spatial prostanoid barriers constrain BiTE-driven adoptive T cell immunity in intact human tumors

doi: 10.64898/2026.03.26.713601

Figure Lengend Snippet: A, Schematic of the NSCLC patient-derived explant (PDE) assay. Fresh or cryopreserved resected tumor tissue from 13 patients was dissected into homogeneous 1 to 2 mm³ fragments and cultured for 24 hours under untreated, untransduced (UT), or MSLN-BiTE conditions. B, FC in the frequency of activated CD8⁺CD137⁺ T cells (out of total CD8⁺) across conditions (P values shown). C, Correlation between epithelial target abundance, quantified as the percentage of EpCAM + MSLN + tumor cells among EpCAM + cells, and ex vivo T-cell activation, quantified as the frequency of CD8 + CD137 + cells among CD8 + T cells. D, FC of granzyme B relative to the UT control condition. E-J, FC of soluble mediators relative to the UT control condition: IFNγ ( E ), TNFα ( F ), GM-CSF ( G ), CCL5 ( H ), CXCL9 ( I ), and CXCL10 ( J ). symbols are color-coded by lesion/patient ID; bars indicate mean ± SEM. Statistical testing was performed on lesion-level values using one-way repeated-measures ANOVA with appropriate multiple-comparisons correction (as indicated by the plotted comparisons). Exact P values are displayed on the graphs (GraphPad Prism v10.1.2). K, Pairwise correlation analyses linking key components of the inflammatory and chemokine response program, including CCL5 versus IFNγ, CCL5 versus CXCL9, CXCL9 versus GM-CSF, and CXCL10 versus GM-CSF. Red and black points denote untreated and MSLN-BiTE conditions, respectively. In B-J , points represent individual patients color-coded; bars indicate mean ± SEM. Exact P values are shown in the panels. Statistical testing was performed at patient-level values using one-way repeated-measures ANOVA with multiple-comparisons correction, as indicated by the plotted comparisons. In K , each point represents one explant fragment.

Article Snippet: All image stacks were imported into Imaris software v10.0 (RRID:SCR_007370) and analyzed using surface detection and surface creation tools to generate three dimensional reconstructions of regions of interest and to quantify exogenous T cell infiltration within tumor compartments.

Techniques: Derivative Assay, Cell Culture, Ex Vivo, Activation Assay, Control

Journal: bioRxiv

Article Title: Actionable spatial prostanoid barriers constrain BiTE-driven adoptive T cell immunity in intact human tumors

doi: 10.64898/2026.03.26.713601

Figure Lengend Snippet: A, Basal PGE 2 quantified in NSCLC PDE supernatants (pg/mL), shown as a heatmap illustrating inter-lesion variability. B, Correlation between basal PGE 2 and MSLN-BiTE-induced CD8+ T cell activation in NSCLC PDEs, expressed as FC in CD8 + CD137 + frequencies relative to UT CTRL (Holm-corrected P value shown). C, Correlation between basal PGE 2 and MSLN-BiTE-induced granzyme B release in NSCLC PDEs, expressed as log2 FC relative to UT CTRL per patient (r and P value shown). D, COX blockade rescue experiment in two NSCLC cases (Lung 31 and Lung 33): PDEs were pretreated with ketorolac (COXi, 1 µM) for 1 h prior to co-culture with MSLN-BiTE T cells, and CD8 + CD137 + activation together with granzyme B, CXCL9, and CXCL10 were quantified as FC over UT CTRL. E, Pharmacodynamic verification of prostanoid pathway inhibition in the HGSOC validation cohort (n = 11), showing PGE 2 levels at baseline and after COXi (ketorolac, 1 µM) or EP2/EP4 inhibition (EP2/4i, 1 µM). F, CD8 + T cell activation in HGSOC PDE co-cultures, reported as %CD8 + CD137 + among CD8 + cells across conditions (condition matrix shown). G-H, FLAG-based discrimination of engineered versus bystander/resident CD8 + T cells in HGSOC PDEs, showing frequencies of CD8 + CD137 + FLAG + exogenous cells ( G ) and CD8 + CD137 + FLAG - bystander/ endogenous cells ( H ). (I) IFNγ secretion in HGSOC PDE co-cultures, expressed as FC over UT CTRL across conditions. ( J ) Cytokine and chemokine coordination across HGSOC PDEs visualized as Pearson correlation matrices for untreated, MSLN-BiTE, MSLN-BiTE + COXi, and MSLN-BiTE + EP2/4i conditions (circle size indicates correlation magnitude; color indicates direction). K, Myeloid polarization in HGSOC PDEs quantified as %CD68 + CD80 + among CD68 + macrophages across conditions. ( L-N ) TNFα ( L ), GM-CSF ( M ), and IL-10 ( N ) secretion in HGSOC PDE co-cultures, expressed as FC over UT CTRL. Boxplots show median and interquartile range with whiskers as displayed; each dot represents one patient. For HGSOC panels ( E-N ), statistics were performed on patient-level values using one-way repeated-measures ANOVA with multiple-comparisons correction as indicated (GraphPad Prism v10.1.2). Figure generated with BioRender.

Article Snippet: All image stacks were imported into Imaris software v10.0 (RRID:SCR_007370) and analyzed using surface detection and surface creation tools to generate three dimensional reconstructions of regions of interest and to quantify exogenous T cell infiltration within tumor compartments.

Techniques: Activation Assay, Co-Culture Assay, Inhibition, Biomarker Discovery, Generated

Journal: bioRxiv

Article Title: Actionable spatial prostanoid barriers constrain BiTE-driven adoptive T cell immunity in intact human tumors

doi: 10.64898/2026.03.26.713601

Figure Lengend Snippet: A , Schematic of the volumetric two-photon imaging pipeline used to quantify intratumoral localization of fluorescently labeled exogenous T cells in PDEs after 24 h co-culture (UT T cells, MSLN-BiTE, MSLN-BiTE +COXi, or MSLN-BiTE +EP2/EP4i), followed by fixation, pan-cytokeratin staining to define tumor epithelium, DAPI counterstaining, tissue clearing, and two-photon acquisition. B, Representative 3D reconstructions showing nuclei (blue), cytokeratin-defined tumor epithelium (magenta), and infiltrating CD8⁺ T cells (green), with corresponding tumor and stroma masks generated by cytokeratin-based segmentation. C, Boxplots quantifying exogenous CD8⁺ T cell density within the CK⁺ tumor compartment and CK⁻ stromal compartment. D, Normalized to imaged surface area, with representative segmentation overlays. E, Representative H&E and IF images of DNA damage after 24 h co-culture, showing TUNEL (green), DAPI (blue), cytokeratin (red), and merged channels across conditions. F-G , Violin plots showing the fraction of TUNEL⁺ cells among CK⁺DAPI⁺ epithelial cells and among CK⁻DAPI⁺ non-epithelial cells. H-I, Violin plots quantifying cleaved-caspase-3 (CC3) positivity within CK⁺MSLN⁺ epithelial cells and within CK⁻ cells. Each dot represents an individual PDE fragment. Statistical analyses were performed using one-way ANOVA (GraphPad Prism v10.1.2); exact P values are indicated on the plots. Figure generated with BioRender. Error bars indicated ± SEM.

Article Snippet: All image stacks were imported into Imaris software v10.0 (RRID:SCR_007370) and analyzed using surface detection and surface creation tools to generate three dimensional reconstructions of regions of interest and to quantify exogenous T cell infiltration within tumor compartments.

Techniques: Imaging, Labeling, Co-Culture Assay, Staining, Generated, TUNEL Assay