Elevated and localized GM-CSF expression in inflamed Crohn’s Disease ileums with stable M-CSF and G-CSF levels.

A. Ileal Crohn’s disease tissue sections (inflamed and uninvolved regions) were profiled using 10X Xenium. Transcript density maps for M-CSF, GM-CSF, and G-CSF are color-coded: yellow (high), purple (intermediate), and black (none). Scale bar, 2000 µm. B. Bar graph quantifying the percentage of cells expressing each CSF per Xenium slide from 10 patients (10 inflamed and 10 uninvolved slides). C. Bar graph compares CSF expressions between inflamed and uninvolved regions (N = 10, **, P < 0.01). D. Heat map of CSF expression per cell cluster (yellow: high, green: medium, purple/black: low/none). E. Images of CSF with markers: TPSAB1 (mast cells), IL7R (T/ILC), and CD34 (endothelial cells). Scale bar, 20 μm. F. Bar graphs of ligand-receptor interactions in inflamed vs. uninvolved ileal sections. The significant Spot to N spot sample ratio was calculated as the number of ligand-receptor interactions in the same grid (Spot) relative to the total spots in the full mucosa area (10 uninvolved slides and 8 inflamed slides). **, P < 0.01.

Human GM-CSF prevents injury in zebrafish DSS model of intestinal injury through Csf2rb signaling.

A. Diagram of strategies on zebrafish treatments and data collections. B. Bar graph of intestinal length relative to total body length for WT (blue, N = 12-19/condition) and csf2rb-/-(yellow, N = 10-19/condition) larval zebrafish treated with DSS and GM-CSF. *, P < 0.05, **, P < 0.01. C. Bar graph of neutral red staining in the zebrafish intestine relative to genotype-matched controls for WT (blue, N = 28-70/condition) and csf2rb-/- (yellow, N = 25-68/condition) larval zebrafish treated with DSS and GM-CSF. ****, P < 0.0001. D. Quantification of RNA foci of RNAscope images for the proinflammatory cytokines tnfα, il1b, and osm in larval zebrafish intestines treated with DSS and human GM-CSF (N = 5, six images per larval gut). **, P < 0.01; ****, P < 0.0001. E. RNAscope detection of csf2rb transcripts (green) in Tg(mpeg1:mCherry) larval zebrafish, in which macrophages are labeled with mCherry (red). Representative csf2rb-positive mpeg1:mCherry-positive macrophages in the retina, intestine, and midbody are highlighted by arrows and magnified in the boxed insets. Additional csf2rb-positive cells lacking mpeg1:mCherry signal are also present, indicating that csf2rb expression is not restricted to macrophages. Scale bar, 100 µm. F. Immunostaining of phospho-STAT5 (red) and nuclei (blue, DAPI) with wildtype larval zebrafish treated with DSS and GM-CSF. Scale bar: 50 µm. G. Quantification of relative intensity of phospho-STAT5 per larval zebrafish gut. N = 10 per treatment conditions. **, P < 0.01; ***, P>0.001; ****, P < 0.0001.

Single-cell transcriptomic profiling reveals conserved ILCs and marker signatures in the larval zebrafish intestine.

A. Diagram of strategies on isolation and single-cell analysis of innate lymphoid cells (N = 200 larval guts/condition). B. UMAP of lck-positive ILCs from larva zebrafish intestine. C. Dot plot generated from the scRNA-seq dataset of FACS-isolated intestinal lck:GFP+ lymphocytes from larval zebrafish, showing expression of conserved ILC3 markers (rorc and il23r), previously reported zebrafish ILC markers (il7r, nitr4a, nitr5, nitr9, and ccl20a.3), and the proliferation marker mki67 across the identified ILC clusters. D. Top differentially expressed genes for zebrafish proliferative ILCs (top bar, green), ILC1s (top bar, blue), and ILC3s (top bar, orange) in no treatment control (bottom bar, red), GM-CSF (bottom bar, blue), DSS (bottom bar, yellow) and DSS GM-CSF cotreatment (bottom bar, green). The yellow rectangle highlights the induction of putative ILC1 gene module by DSS treatment in proliferative ILCs and ILC3s, while the green rectangle indicates the reversal of the putative ILC1 gene module upon cotreatment with human GM-CSF. E. Heatmap of conserved ILC1 (top panel) and ILC3 (bottom panel) marker genes in zebrafish intestine, human ileum, and human rectum. F. Venn diagram of conserved transcriptional regulators predicted by IPA for ILC1s (top panel) and ILC3s (bottom panel) across human ileum and rectum, and larva zebrafish intestine.

GM-CSF represses ILC1 activity and boosts the tissue repair function of ILC3s in intestinal inflammation.

A. Zebrafish ILC proportions from scRNA-seq dataset. B. RNAscope images showing TBX21 probe staining in WT and csf2rb-/- zebrafish intestines treated with DSS and GM-CSF. Scale bar, 50 µm. C. Quantification of RNA foci for ILC markers tbx21 (ILC1) and rorc (ILC3) (N = 5 zebrafish, six images per gut). *, P < 0.05; **, P < 0.01; ****, P < 0.0001. D. Expression of the putative ILC1-like transcriptional module in proliferative ILCs, putative ILC3-like cells, and putative ILC1-like cells under control, GM-CSF, DSS, and DSS + GM-CSF conditions (red: control, blue: GM-CSF, yellow: DSS, green: DSS + GM-CSF). Module scores were calculated using gene in figure 3E. ****, P < 0.0001; *, P < 0.05. E. RNAscope images with il22 probe staining of zebrafish intestines treated with DSS and GM-CSF. Scale bar: 100 µm. F. Quantification of il22 RNA foci (N = 6-7, one image per gut). ***, P < 0.001. G. RNAscope images with ifng1-1 probe staining of zebrafish intestines treated with DSS and GM-CSF. Scale bar, 100 µm. H. Quantification of ifng1-1 RNA foci (N=6-7, one image per gut). **, P < 0.01; *, P < 0.05.

Plasticity of intestinal ILCs.

A. Unspliced (red) to spliced transcript (blue) ratio of intestinal ILC and enterocytes in control (top left panel), GM-CSF (bottom left panel), DSS (top right panel) and DSS with GM-CSF co-treated (bottom right panel) larval zebrafish determined by single-cell velocity analysis. B. Unspliced to spliced transcript ratios of proliferative ILC marker gene, hmgn2 in control (top row, blue punctation), GM-CSF (second row, purple punctation), DSS (third row, red punctation) and DSS with GM-CSF co-treated (bottom row, pink punctation) larval zebrafish (N = 15/data point). C. Proportion of unspliced relative to spliced transcripts for the indicated gene module. The y-axis provides the quantitative measure of unspliced transcript abundance, while the color scale visualizes the corresponding transcriptional state, with higher relative levels of unspliced transcripts indicating a more transcriptionally active state. *, P < 0.05. D. Quantification of relative ILC1 and ILC3 cell proportion in inflamed and uninvolved mucosa of human ileal tissue. *, P < 0.05.

Spatial activation of phopho-STAT5 signaling pathway surrounding proinflammatory macrophage aggregates.

A. H&E image of inflamed CD section with non-ulcerated mucosa (white), ulcer exudate (pink) and granulation (yellow). Scale bar, 200 μm. B. Xenium images of inflamed CD section with non-ulcerated mucosa (white), ulcer exudate (pink), and granulation tissue (yellow). Green cells represent the epithelial compartment and are distinct from the CD14-positive macrophage aggregates within the ulcer exudate. Scale bar, 500 μm. C-E. Quantification of ulcerative epithelial disruptions identified by loss or disruption of the epithelial barrier in pathologist-classified inflamed and uninvolved tissue sections. Ulcer numbers, macrophage adjacency, and macrophage density quantifications. * P < 0.05, **** P < 0.0001. N = 4 patients with paired inflamed and uninvolved sections. F. GM-CSF and cytokine (IL1B, TNF, OSM) density map. Scale bar, 5000 μm. G. Expression map of GM-CSF (red), IL1B (light blue), OSM (yellow), and TNF (green) showing IL1B and OSM enrichment in macrophage aggregates at ulcer exudate. Scale bar, 250 μm. H. Bar graphs of cytokine and CD14 transcript density in non-ulcered mucosa, ulcer exudate, and granulation tissue. *, P < 0.05. N = 4 patients with paired inflamed and uninvolved sections. I. Schematic of M-CSF, GM-CSF and G-CSF signaling through pERK, pSTAT5, pSTAT3/pERK, respectively. J. Bar graph showing the Pearson correlation coefficient from averaging five macrophage aggregate regions per inflamed ileal section. K. Representative images of inflamed ileal tissue with co-immunostaining for DAPI, CD14, GM-CSF, p-STAT5, p-STAT3 and p-ERK1/2 (top panel). Scale bar, 5000 μm. Zoomed-in regions of macrophage aggregates show DAPI, CD14, GM-CSF, p-STAT5, p-STAT3 and p-ERK1/2 (bottom panel). Scale bar, 250 μm.

GM-CSF in ulceration modulates ILC state and proinflammatory cytokines.

A. Model: M-CSF, GM-CSF and G-CSF in homeostasis and Crohn’s disease inflammation.