Longitudinal Single-Cell Transcriptomics Reveals a Role for Serpina3n-Mediated Resolution of Inflammation in a Mouse Colitis Model

Abstract

Background & aims: Proper resolution of inflammation is essential to maintaining homeostasis, which is important as a dysregulated inflammatory response has adverse consequences, even being regarded as a hallmark of cancer. However, our picture of dynamic changes during inflammation remains far from comprehensive.

Methods: Here we used single-cell transcriptomics to elucidate changes in distinct cell types and their interactions in a mouse model of chemically induced colitis.

Results: Our analysis highlights the stromal cell population of the colon functions as a hub with dynamically changing roles over time. Importantly, we found that Serpina3n, a serine protease inhibitor, is specifically expressed in stromal cell clusters as inflammation resolves, interacting with a potential target, elastase. Indeed, genetic ablation of the Serpina3n gene delays resolution of induced inflammation. Furthermore, systemic Serpina3n administration promoted the resolution of inflammation, ameliorating colitis symptoms.

Conclusions: This study provides a comprehensive, single-cell understanding of cell-cell interactions during colorectal inflammation and reveals a potential therapeutic target that leverages inflammation resolution.

Keywords: Colitis; Serpina3n; Single Cell RNA-Sequencing; Stromal Cell.

Graphical abstract

Figure 1

Single-cell RNA-seq time course of induced colon inflammation. (A) Scheme illustrating our time-course analysis in the murine DSS-induced colitis model. Body weight was recorded every day. Mean ± standard error of the mean is indicated. P values were calculated with two-way analysis of variance (ANOVA) (n = 3 mice per group). ∗P < .05, ∗∗P < .01, ∗∗∗P < .001. (B and C) Colonic length was measured at each time point (days 0, 3, 6, 9, 12, and 15). P values were calculated with one-way ANOVA (n = 3 mice per group). ∗P < .05 (B). Histologic assessment of colon samples (days 0, 3, 6, 9, 12, and 15). The representative section was stained with H&E or alcian blue. Scale bar: 500 μm (C). (D) Data visualization using a UMAP plot. Annotated cell types are listed. (E) Marker genes for each cluster are illustrated. Dot size represents percentage of cells expressing the particular gene within the cluster. Intensity of dot color shows mean expression level. (F) Percentage of cell proportion in each cluster over time.

Figure 2

Marker genes of cell clusters. (A) Marker genes expressed on cell clusters were illustrated in the UMAP plot. (B) UMAP plot showing the stem/progenitor marker genes within epithelial: Abs & sec cell cluster. (C) Cell proportions in clusters are shown in the bar graph; y-axis: cell count, bottom: time point.

Figure 3

Phenotypic changes during inflammation. (A) Inflammation score was calculated on the basis of expression levels of inflammatory genes expressing in the MNP cluster. P values were calculated using one-way ANOVA Bonferroni post-test. ∗P < .01, ∗∗P < .001. (B) Analysis of pseudotime changes in cell clusters. (C) Difference between maximum and minimum pseudotime within cell clusters. (D) Number of DEGs in each cluster is indicated. ND, no detection of DEGs.

Figure 4

IBD risk genes expressed in multiple cell clusters. (A) 79 IBD-related genes that are expressed at one or more time points in at least one of the cell types. (B) Heatmap representing expression levels of IBD risk genes. Maximum gene expression levels across time points are shown. (C) Number of cell types expressing the given gene. (D) Heatmap based on K-means clustering showing expression patterns of IBD risk genes. (E) Expression pattern of Ifng and Ahr across the time course of inflammation.

Figure 5

Hub function of stromal cells during inflammation. (A) Heatmap showing the number of ligands in senders and receptors in receivers across time points. Right: cell type; bottom: time point. Analysis conducted using NicheNet. (B) Heatmap showing the number of ligand-receptor connections between cells. (C) Total number of ligands in senders.

Figure 6

Cell-cell communication during DSS-induced inflammation. Circos plots showing the ligand-receptor links between each cell cluster. Blunt end of the ribbon represents the ligand in senders, and arrowhead points to the receptor in receivers. The ligand-receptor number is indicated by the size and color of the ribbon.

Figure 7

Expression and function of Serpina3n in stromal cells during inflammation. (A) Heatmap showing the expression patterns of DEGs (across time points) in stromal cells. (B) Expression level of genes in the K4 cluster. (C) Expression levels of Serpina3n across time points. (D) UMAP plot showing expression of Serpina3n in the stromal cell cluster (C5) at day 9 in the DSS-induced colitis model. (E) mRNA expression of Serpina3n in colon was analyzed by real-time qPCR; quantitative result is normalized with actb. P values were calculated with one-way ANOVA (n = 3 mice for each time point). ∗P < .001. (F) qPCR/scRNA-seq trend comparison was shown in line plot. (G) Immunostaining of Serpina3n in the colon of DSS-induced inflammation across the time course. Expression levels were quantified using IMARIS software (BitPlane). Scale bar: 100 μm. P values were calculated with one-way ANOVA (n = 3 mice for each time point). ∗P < .01. (H) Western blot of Serpina3n in colon. Histone H3 is an internal control. Bottom: time point. (I and J) Immunoprecipitation of Serpina3n and blotting with Serpina3n (F) or elastase (G). Input: whole cell lysate.

Figure 8

Serpina3n is involved in the resolution of colon inflammation. (A) Expression of Serpina3n in the colon was confirmed using immunostaining. Colon samples were collected from wild-type (Serpina3n +/+) or Serpina3n knockout (Serpina3n -/-) mice at day 9 after DSS induction. (B) Body weight changes in wild-type (Serpina3n +/+) or Serpina3n knockout (Serpina3n -/-). P values were calculated with two-way ANOVA (n = 6 mice per group from 2 independent experiment). ∗P < .05, ∗∗P < .01, ∗∗∗P < .001. Mean ± standard error of the mean is indicated. (C) Colonic length was measured at day 15. P values were calculated using one-way ANOVA (n = 6 mice per group from 2 independent experiments). ∗P < .05. (D) Histologic assessment of colon samples (day 15); representative sections were stained with H&E; lower image is a magnified region from the upper image. Scale bar: 500 μm.

Figure 9

Serpina3n treatment ameliorates IBD symptoms. (A) Body weight changes relative to control (black circle). DSS-induced mice were intravenously treated with PBS (red circle) or Serpina3n (green circle). P values were calculated using two-way ANOVA (n = 6 mice per group from 2 independent experiments). ∗P < .05, ∗∗P < .01, ∗∗∗P < .001. (B and C) Colonic length was measured at day 9 of DSS-induced colitis model. P values were calculated using one-way ANOVA (n = 6 mice per group from 2 independent experiments). ∗P < .05, ∗∗P < .01 (B). Histologic appearance of colon samples (day 9); the lower image is a magnified region of the upper image. Scale bar: 500 μm (C). (D and E) UMAP plot showing the cell clusters in all treatment groups (D). Shown is the proportion of cells (left y-axis) for each cell subset (right y-axis); treatment group is shown on the x-axis (E). (F) Quantification of S100A8-expressing granulocytes and CD19-expressing B cells in control-treated colons (black bar), DSS-induced mice treated with PBS (red bar) or Serpina3n (green bar). P values were calculated using one-way ANOVA. ∗P < .01. (G) Number of DEGs in each cell cluster is shown in the bar graph. ND, no detection of DEGs. (H) Inflammation score was calculated on the basis of inflammatory genes expressed in the MNP cluster. P values were calculated using a one-way ANOVA Bonferroni post-test. ∗∗P < .001. (I) Gene ontology analysis shows biological pathways enriched in the stromal cell cluster. Bottom: P value. (J) Cell count analysis showing the S100a8 and S100a9 expressing cells within granulocyte cluster in all treatment groups. P values were calculated by one-way ANOVA Bonferroni post-test. ∗∗∗P < 2e^-16.

Figure 10

Administration of Serpina3n suppresses cell-cell communication in colitis. (A) NicheNet plot showing the ligand-receptor link between the sender and receiver for DSS-induced mice treated with PBS (left) or Serpina3n (right). (B and C) Heatmap showing the number of ligands and receptors in cell clusters under DSS-induced group treated with PBS or Serpina3n. (D) Circos plot showing ligand-receptor links between clusters in the DSS-induced group treated with PBS or Serpina3n.