Mitesh Dongre

Top 30 markers of clusters in CAF subset of single cell data

Lists of genes used to calculate signature scores for cytolytic activity, T cell activation and M1 vs M2 macrophage polarization in single cell data analysis

Orthotopic transplantation of gal 4–KO cell lines identifies differences in CD4+ and CD8+ cell infiltration and function. A, Schematic of the orthotopic transplantation experiments using gal 4–KO and R26 control cell lines transplanted into immune competent C57BL6/J mice. B, Representative IHC images of gal 4 in R26 control and gal 4–KO tumors (top scale bar, 50 μm; bottom, scale bar, 20 μm). A total of 10 tumors were stained, 5 of each condition. C–E, Representative IHC images (left) and quantification (right) of CD4+ cells (P = 0.0032; C), CD8+ cells (P = <0.0001; D), and FOXP3+ cells (ns., no significant difference; E), in orthotopically transplanted tumors with gal 4 (R26 controls) and no gal 4 expression (Gal 4–KO); scale bar, 100 μm. Significance calculated using the Mann–Whitney test. Error bars show mean ± SEM. F, Representative contour plots illustrating gating strategies of intratumoral CD8+ T cells (left). Scatter plot with bar graph showing the fraction of CD8+ cells out of the total amount of CD45+ CD3+ cells in R26 control and gal 4–KO tumors (right; P = 0.0001). Significance calculated using the Mann–Whitney test. Error bars show mean±SEM. G Representative contour plots of intratumoral CD45+, CD3+, CD8+ cells (CD8+ T cells gated as in Fig. 3 F) expressing the exhaustion markers PD-1 (x-axis) and Tim-3 (y-axis) in R26 control and gal 4–KO tumors (left). Scatter plot with bar graph s (right) showing quantification of CD8+ T cells exhaustion (PD1+ and Tim-3 cells in Q2) in R26 control and gal 4–KO tumors (P = 0.046). Significance calculated using the Mann–Whitney test. Error bars show mean with SEM. **, P < 0.01; ***, P < 0.001.

Top 30 markers of clusters in Immune subset of single cell data

Galectin 4 is a cancer cell–derived marker of disease progression and prognosis in pancreatic cancer, with an association to reduced immune activity. A, Representative ISH images of cytokeratin 18 (Krt18, blue, top image) and fibroblast activation protein alpha (Fap, blue, the bottom image), together with gal 4 (Lgals4, in red), in mouse model of PDAC tumor tissue. 4 tumors analyzed in total; scale bar, 10 μm. B, Violin plot of single-cell RNA-seq data from Peng et al. (40) showing LGALS4 expression in cell types from human PDAC tumors. C, Western blot and quantification (semiquantitative) of gal 4 protein in normal murine ductal organoid (N, two different cell lines), PanIN organoid (P, two different cell lines), and tumor organoid (T, three different cell lines) cell lines. Vinculin was used as loading control. D, Kaplan–Meier survival curve adapted from the TCGA data set (26), with patients grouped into high (n = 26) and low (n = 49) gal 4 mRNA expression (groups defined in Materials and Methods) and relationship to survival in patients with PDAC with moderately differentiated tumors. *, Significance (P = 0.0254) calculated using the Mantel–Cox test. E, Z-score comparison of high (n = 26) and low (n = 49) gal 4-expressing tumors within the group of moderately differentiated PDAC tumors in the TCGA data set. Genes associated with cytolytic effect of immune cells analyzed in the left graph (P = 0.036), and CD8 activation scores analyzed in the right graph (P = 0.0253). Genes used for CD8 activation and cytolytic effect are listed in Supplementary Table S2. *, Indicates significance calculated using an unpaired t test. Error bar indicates mean ± SEM.

Raw read count, normalized expression values (TPM) and sample metadata for bulk RNA sequencing data

Metadata and single cell annotation for all 7677 cells in single cell dataset

High levels of galectin 4 in pancreatic tumors and in the circulation of patients with pancreatic cancer. A, Volcano plot illustrating fold-change in extracellular expression of proteins detected from quantitative mass spectrometry performed on purified ECM from human PDAC tumors and normal pancreatic tissue (data from; ref. 22). Detected members of the galectin family of proteins in red. Arrow denotes galectin 4. Dotted horizontal line indicate significance level P = 0.05 (n = 5 tumors, n = 2 normal pancreatic tissue). B, Mass spectrometry Z scores for galectin 4 (data from; ref. 22) in mouse and human tissue from normal pancreas, PanIN, chronic pancreatitis, and tumor samples (subcategorized by early- and late-stage for mouse samples). Error bars indicate mean ± SEM of Z-scores. C, Representative IHC staining of gal 4 in murine and human normal pancreatic tissue, human pancreatitis, murine and human PanIN precursor lesions, KPC-derived murine cancer and well-differentiated human PDAC, and human liver and lymph node metastases. Arrows denote normal pancreatic ducts. 5 samples from each tissue condition were stained; scale bar, 50 μm. D, PDAC tumors counterstained with HTX (top left) and labeled with UCNPs (bottom left) on the same section. Semiquantitative analysis of gal 4 signal intensity in areas proximal to or distal from the epithelial cells (right). The UCNP emission was quantified in areas proximal to epithelial cells (blue dotted lines represent the area where the luminescent signal was quantified) or distal from tumors (red dotted lines represent the area where the luminescent signal was quantified) in 5 tumors; scale bar, 20 μm. E, ELISAs to detect serum galectin 4 protein in patients with PDAC (red) and aged-matched controls (blue). Significance calculated using the Mann–Whitney test (P = 0.0012). Error bars show mean ± SEM. **, P < 0.01.

Ex vivo models of PDAC reveal immunosuppressive effects of galectin 4 through CD3δ. A, Schematic depicting two ex vivo model systems using splenocytes isolated from immunocompetent C57BL6/J mice added to KPC-derived 2D cells (Monolayer model) or tumor organoids/tumor spheroids (Matrigel dome and Transwell models). Splenocytes were isolated from aged-matched C57BL6/J mice. B, Representative confocal microscope image of a KPC cell–derived spheroid (GPF, green) surrounded by T cells (cell tracker deep red stain, red) after 48 hours of incubation; scale bar, 10 μm. C, Representative brightfield images of gal 4–KD and scramble-hairpin control organoids at 24, 96, and 120 hours in cocultures with immune cells (scale bar, 100 μm). The experiment was repeated 3 times. D, Representative contour plot showing gating strategy for 7AAD+ epithelial cells from dissociated Matrigel domes in scramble-hairpin controls and gal 4–KD cell lines in the tumor organoid ex vivo model. Quantifications from this experiment are presented in E–G. E–G, Scatter plot with bar graph showing flow cytometry quantification of apoptosis in organoid cells from dissociated Matrigel domes with added splenocytes (E), without splenocytes (organoids only; F), organoids with splenocytes in Transwell at 120 hours timepoint (G). Scramble-hairpin controls (n = 6) and gal 4–KD (n = 7); ns., not significant. H, Representative contour plots showing gating strategy for 7AAD+CD8+ T cells from the monolayer model using R26 control and gal 4–KO cell lines. Quantifications from this experiment are presented in I. I, Scatter plot with bar graph showing flow cytometry quantification of 7AAD+CD8+ T cells added to R26 control (n = 4) and gal 4–KO (n = 8) in the monolayer model (R26 normalized to 1) at 24 hours in culture. Significance (**, P = 0.004) calculated using the Mann–Whitney test. J, SPR binding graph showing binding between human recombinant gal 4 (hgal 4) and human recombinant CD3ε/δ and CD3ε/γ heterodimer chains. The experiment was repeated 3 times. K, Line graph of SPR data for carbohydrate inhibition of binding between human recombinant gal 4 and human recombinant CD3ε/δ heterodimer treated with either galactose or fructose. The experiment was repeated once for fructose and twice for galactose. L, Bar graph of SPR data for binding between human recombinant gal 4 and human recombinant CD3ε/δ heterodimer following treatment with a glycosidase targeting N-linked glycosylations. The experiment was repeated 2 times. M, Representative contour plot showing gating strategy for AnnexinV+CD8+ T cells from monolayer model used in the gal 4-blocking assay. Quantifications from this experiment are presented in N–O. N, Scatter plot with bar graph showing flow cytometry quantification of AnnexinV+CD8+ T cells added to R26 control (n = 4) and gal 4–KO (n = 6) supplemented with BSA in the monolayer model (R26 normalized to 1). Significance (P = 0.019) calculated using the Mann–Whitney test. O, Scatter plot with bar graph showing flow cytometry quantification of AnnexinV+ CD8+ T cells added to R26 control (n = 3) and gal 4–KO (n = 7) supplemented with 60 μg/mL soluble CD3 chains in the 2D cell-based ex vivo model (R26 normalized to 1). Non-significant, ns.; *, P = 0.5167 calculated using the Mann–Whitney test.