The transcription factor HIF-1α in NKp46+ ILCs limits chronic intestinal inflammation and fibrosis

Abstract

Innate lymphoid cells (ILCs) are critical for intestinal adaptation to microenvironmental challenges, and the gut mucosa is characterized by low oxygen. Adaptation to low oxygen is mediated by hypoxia-inducible transcription factors (HIFs), and the HIF-1α subunit shapes an ILC phenotype upon acute colitis that contributes to intestinal damage. However, the impact of HIF signaling in NKp46⁺ ILCs in the context of repetitive mucosal damage and chronic inflammation, as it typically occurs during inflammatory bowel disease, is unknown. In chronic colitis, mice lacking the HIF-1α isoform in NKp46+ ILCs show a decrease in NKp46⁺ ILC1s but a concomitant rise in neutrophils and Ly6C^high macrophages. Single-nucleus RNA sequencing suggests enhanced interaction of mesenchymal cells with other cell compartments in the colon of HIF-1α KO mice and a loss of mucus-producing enterocytes and intestinal stem cells. This was, furthermore, associated with increased bone morphogenetic pathway-integrin signaling, expansion of fibroblast subsets, and intestinal fibrosis. In summary, this suggests that HIF-1α-mediated ILC1 activation, although detrimental upon acute colitis, protects against excessive inflammation and fibrosis during chronic intestinal damage.

Figure 1.. Deletion of HIF-1α in NKp46+ cells leads to a proinflammatory phenotype in the proximal colon upon chronic dextran sodium sulfate exposure.

(A) Disease activity index of HIF-1α KO compared with WT mice. Mice were treated with 2% dextran sodium sulfate in drinking water four times for 3 d with the recovery phase of 4 d; n WT = 5, n KO = 4; data are mean values ± SD; *P < 0.05, two-tailed t test. (B) Representative images of H&E-stained proximal colon and analysis of the histological score of inflammation. Pooled data of three experiments; n WT = 17, n KO = 8; data are mean values ± SD; *P < 0.05, two-tailed t test. (C) Absolute number of CD45⁺ cells and abundance of different innate lymphoid cell (ILC) subsets in % of alive CD45⁺ cells at the endpoint. ILC1s defined as alive CD45⁺, Lin negative (CD11c, CD19, Ly6G, Ter119, TCR-β, TCR-γ/δ), CD127+, RORγt negative, NKp46+, T-bet+, CD49a+, NK1.1+. NK cells as alive CD45⁺, Lin negative, CD127 negative, NKp46+, Eomes+, NK1.1+. ILC3 as alive CD45⁺, Lin negative, CD127+, RORγt+, NKp46+. Pooled data of two independent experiments; n WT ≥ 10, n KO ≥ 7; data are mean values ± SD; *P < 0.05, two-tailed t test. (D) Fold change of the IFN-γ mRNA expression of whole colon of WT and HIF-1α KO mice. Data were normalized to 16s. n WT = 3, n KO = 3. Data are mean values ± SD; *P < 0.05, two-tailed t test. Two outliers have been excluded using the ROUT method (Q = 1%). (E) Flow cytometric analysis of alive CD45⁺, neutrophils, macrophages, eosinophils, CD4⁺ and CD8⁺ T cells, and M1 (iNOS+) and M2 (CD206+) macrophages at the endpoint. Pooled data of two experiments; n KO ≥ 3, n WT ≥ 4; data are mean values ± SD; *P < 0.05, two-tailed t test. (F) (Left) Representative flow cytometric images of Ly6C+ gated on macrophages (alive CD45⁺, CD11b+, F4/80+) in HIF-1α WT and KO in NKp46+ cells and results (right) of % low/intermediate/high Ly6C. Pooled data of two experiments; n KO = 10, n WT = 9; data are mean values ± SD; *P < 0.05, two-tailed t test.

Figure S1.. Representative gating strategy for innate lymphoid cells (ILCs) and myeloid cells.

(A) Gating for ILCs with ILC1 defined as alive CD45⁺, Lin negative (CD11c, CD19, Ly6G, Ter119, TCR-β, TCR-γ/δ), CD127+, RORγt negative, NKp46+, T-bet+, CD49a+, NK1.1+. NK cells as alive CD45⁺, Lin negative, CD127 negative, NKp46+, Eomes+, NK1.1+. ILC3 as alive CD45⁺, Lin negative, CD127+, RORγt+, NKp46+. NK cells as alive CD45⁺, Lin negative, CD127 negative, NKp46+, Eomes+, NK1.1+. ILC3 as alive CD45⁺, Lin negative, CD127+, RORγt+, NKp46+. (B) Gating for myeloid and lymphoid cells among alive CD45⁺ with neutrophils (Ly6G+, CD11b+), eosinophils (CD11b+, Siglec-F+), macrophages (F4/80+, CD11b+), CD4⁺ T cells (TCR-β+, CD4⁺), and CD8⁺ T cells (TCR-β+, CD8⁺).

Figure S2.. Marker genes reveal different cell types in the colon with the increased abundance of immune cells in the colon of mice lacking HIF-1α in NKp46+ cells.

(A) Dot plot of top 10 marker genes (colored) of the corresponding cell types. (B) UMAP plot of the immune cell compartment with clustering of the different immune cell subtypes.

Figure 2.. Single-nucleus RNA sequencing reveals a proinflammatory phenotype of HIF-1α KO mice with increased overall signaling.

(A) UMAP plots and corresponding cell types of the whole colon tissue at the endpoint of WT and HIF-1α KO in NKp46+ cells. A total of 9,668 nuclei were analyzed. (B) Left: distribution of immune cells, mucus-producing enterocytes, and intestinal stem cells. Middle: representative images of immunofluorescent LGR5 staining (green) and DAPI (blue) on colon sections. Data were analyzed and calculated (cells per mm²). Right: representative images of immunofluorescent MUC2 staining (green) and DAPI (blue) on colon sections. Pooled data of two experiments; n KO ≥ 5, n WT = 5; data are mean values ± SD; *P < 0.05, two-tailed t test. (C) CellChat analysis of all cell types in the colon. Red arrows show increased, and blue arrows show decreased, signaling in the colon of HIF-1α KO. (D) Number of interactions from and toward mesenchymal cells. (E) Bubble plots of corresponding ligands–receptors from mesenchymal cells toward the mucus-producing enterocytes, intestinal stem cells, and the mesenchymal cell compartment of WT and HIF-1α KO mice. Pooled samples of n = 3 animals per group at the endpoint.

Figure S3.. Increased profibrotic signature in single-nucleus RNA sequencing in mesenchymal cells in the colon of HIF-1α KO in NKp46+ cells.

Violin plots with profibrotic genes and myofibroblast markers in mesenchymal cells. Pooled samples of n = 3 animals per group at the endpoint.

Figure 3.. Single-nucleus RNA sequencing reveals an increased frequency of profibrotic (myo)fibroblast subsets and histological increased collagen deposition in the lamina muscularis in mice colon lacking HIF-1α in NKp46+ cells.

(A) UMAP plots of mesenchymal cells zoom in with different fibroblast and smooth muscle cell subsets. (B) Dot plot of profibrotic gene and myofibroblast marker expression for each subset. (C) Top: the abundance of fibroblasts and smooth muscle cells in % of mesenchymal cells of HIF-1α KO (right) and WT (left). Bottom: the relative abundance of the different subsets of fibroblasts and smooth muscle cells. (D) GO terms and associated genes for subset Fibro 2. (E) RNA velocity analysis of mesenchymal cell comparison between HIF-1α KO (right) and WT (left). (F) Representative images and analysis of PDGFR-a+ area in the submucosa of WT and HIF-1α KO colon sections. (G) Analysis of average fibrotic thickness of WT and HIF-1α KO colon sections. Pooled data of two experiments; n KO = 5, n WT = 5; data are mean values ± SD; *P < 0.05, two-tailed t test. (H) Left: representative images of Sirius Red/Fast Green–stained lamina muscularis sections and enhanced collagen signal. Right: analysis of % Sirius Red–positive area in the lamina muscularis of the colon of HIF-1α KO and WT mice at the endpoint. Pooled data of three experiments; n WT ≥ 11, n KO ≥ 7; data are mean values ± SD; *P < 0.05 and ***P < 0.001, two-tailed t test.