A Myeloid Lineage Signifying Anti-tumor Necrosis Factor Resistance in Crohn's Disease

Abstract

Background & aims: Although anti-tumor necrosis factor (TNF) therapy has improved Crohn's disease (CD) management, the development of a refractory phenotype having resistance to the drug is not uncommon. The mechanisms behind this anti-TNF nonresponse are unknown but are likely multifactorial. Here, we examined myeloid cells expressing signal regulatory protein α (SIRPα) for their potential role in refractory CD.

Methods: Response to anti-TNF was defined as having reached endoscopic and histological healing, whereas nonresponders did not. Isolated cells from peripheral blood and mucosal biopsies were analyzed by high dimensional flow cytometry, single-cell and bulk RNA sequencing, and Luminex. Ileal organoids were also challenged with secretomes from stimulated SIRPα⁺ cells.

Results: Among the CD phenotypes, patients with anti-TNF refractory CD had the highest levels of CD33⁺HLA-DR⁺CD11c⁺SIRPα⁺ cells in their intestinal mucosa, but the levels in peripheral blood were unchanged. SIRPα⁺ cells from the gut displayed a higher proinflammatory transcriptome, with increased levels of interleukin (IL)-6, TNFα, p40, and IL-1β expression. When isolated and stimulated in vitro with flagellin, these SIRPα⁺ cells showed a more pro-inflammatory transcriptome during CD47 ligation than with an IgG control. Moreover, the secretomes of flagellin/CD47-stimulated SIRPα⁺ cells from patients with refractory CD increased cell death and promoted gene expression associated with Rho GTPase and innate immune responses in epithelial cells, while downregulating their gene expression involved in RNA, lipid metabolism, and adaptive response signaling.

Conclusions: Increased levels of myeloid lineage expressing CD33⁺HLA-DR⁺CD11c⁺SIRPα⁺ cells in the intestinal mucosa negatively impact epithelial cell function, possibly explaining one mechanism for anti-TNF resistance. The abundance of mucosal SIRPα⁺ cells should be further explored as a biomarker and therapeutic target.

Keywords: Anti-TNF; CD47; Inflammatory Bowel Disease; Refractory Crohn’s Disease; SIRPα.

Graphical abstract

Figure 1

Quality control and annotation of blood and intestinal single-cell data. (A) Schematic showing the quality control steps and downstream processing for blood and gut samples. (B) UMAP embedding shows the cellular landscape colored by cell type annotations. (C) Dot plot shows the expression of marker genes of the cell types in the single cell UMAP above.

Figure 2

SIRPA-expressing myeloid cells exhibit distinct pro-inflammatory profiles in blood and intestinal mucosa. (A) UMAP of integrated single-cell RNA sequencing data of immune cell populations derived from peripheral blood and intestinal samples (125K cells) of individuals with CD and non-CD controls. (B) UMAP plot colored by cellular compartments. (C) UMAP density plot depicting SIRPA expression levels across immune populations. (D) Dot plot highlighting myeloid cells with the highest SIRPA expression. (E) Bar plot quantification showing that SIRPA-expressing myeloid cells represent less than 25% of the total myeloid compartment in both blood and intestinal tissues. (F) Pseudo bulk PCA analysis of SIRPA-expressing myeloid cells from blood, rectum, and ileum. (G) UMAP visualization of SIRPA-expressing myeloid subpopulations and their distribution across blood and intestine (ileum and rectum). (H) Bar plot showing significantly enriched pathways in SIRPA-expressing myeloid cells from blood and intestinal tissues, relative to their myeloid counterparts with undetectable levels of SIRPA expression. (I) Dot plot displaying the expression of key proinflammatory genes across SIRPA-expressing myeloid subpopulations in blood and intestine.

Figure 3

PC loading genes and SIRPA-expressing sub-cluster annotations in blood and intestinal single-cell data. (A) Bar plot shows the immune cell proportions across the blood and mucosal samples. (B) PC loading genes associated with PC1 and PC2 shown in bar plots. (C) Feature plot showing the expression of HLA-DRA and CD33 expression levels in the SIRPA-expressing myeloid cells. (D) Dot plot shows the expression of gene markers used for SIRPA-expressing myeloid sub cluster annotations.

Figure 4

Representative gating strategy of the SIRPα+ cells in the blood and intestinal tissue. (A) Representative gating strategy for isolating CD33⁺HLA-DR⁺CD11c⁺SIRPα⁺ cells from intestinal tissue using flow cytometry. (B) Representative gating strategy for isolating CD33⁺HLA-DR⁺CD11c⁺SIRPα⁺ cells from peripheral blood samples.

Figure 5

SIRPα⁺ cells are more abundant in the gut during refractory CD. (A) Representative contour plots of CD33⁺HLA-DR⁺CD11c⁺SIRPA⁺ cells from each phenotypic group after immunophenotyping by flow cytometry. (B) Quantified levels of SIRPA⁺ myeloid cells from (A) for each group. (C) Representative contour plots showing IL-6, TNFα, p40, IL-1β, and IFN-γ expression by flow cytometry in CD33⁺HLA-DR⁺CD11c⁺SIRPA⁺ cells from ileal mucosa for each group. (D) Corresponding bar graphs for (C) display ANOVA analysis quantifying cytokine expression levels across the disease phenotypes. ∗∗ P < .01, ∗P < .05.

Figure 6

Cytokine expression in rectal SIRPα+ phagocytic cells in flow cytometry analysis data. (A) Representative contour plots showing the levels of IL-6, TNFα, p40, IL-1β, and IFN-γ expression by flow cytometry in CD33⁺HLA-DR⁺CD11c⁺SIRPα⁺ cells from rectal mucosa for each group. (B) Corresponding bar graphs for (A) displaying ANOVA analysis comparing cytokine expression levels across the disease phenotypes.

Figure 7

Refractory CD has higher levels of SIRPA-expressing inflammatory macrophages. (A–F) Validation cohort analysis of ileum single-cell transcriptomes from CTRL, TN-CD, REF-CD and Es-CD patients (130K cells). (A) UMAP showing cellular subtype identities of the ileum mucosa. (B) Feature plot with density gradient showing SIRPA expression. (C) Dot plot illustrating SIRPA expression levels across cell types. (D) Relative abundance of SIRPA-expressing macrophages (resident vs inflammatory) populations across disease phenotypes. (E) Relative abundance of SIRPA-expressing inflammatory macrophages between TN-CD and REF-C. Wilcox test, P < .05. (F) Volcano plot of DE between TN-CD and REF-CD SIRPA-expressing macrophages. (G) Comparative analysis of cellular communication networks among SIRPA-expressing macrophages in TN-CD and REF-CD.

Figure 8

Ileal epithelial cells communication to SIRPA+ resident and inflammatory macrophages in TN-CD and REF-CD. Cell-cell communication plots showing the receptor-ligand pairs from epithelial cells to SIRPA-expressing inflammatory and SIRPA-expressing resident macrophages across TN-CD and REF-CD.

Figure 9

Representative gating strategy of sorting SIRPα+ cells and Live organoids. (A) Representative gating strategy for sorting SIRPα⁺ phagocytes from ileal and rectal tissue of patients with REF-CD. (B) Representative gating strategy for identifying live organoid cells in flow cytometry analysis.

Figure 10

SIRPα⁺ signaling enhances pro-inflammatory processes and negatively affects the intestinal epithelium. (A–C) Isolated SIRPα⁺ cells from the gut were stimulated in vitro under standard cell culture conditions for 48 hours with FLG, along with either CD47-Fc or with IgG1-Fc as a control. (A) Up and downregulated pathways identified in bulk RNA sequencing of stimulated SIRPα⁺ cells. (B) Top DEGs between CD47-Fc and IgG1-Fc treatments. (C) Bar graphs of multiplex Luminex data of culture supernatants from stimulated SIRPα⁺ cells. Statistical significance assessed by Welch’s t-test, ∗P < .05. (D and E) Intestinal organoids were grown in conditioned media from stimulated SIRPα⁺ cells (diluted 1:1 with Intesticult media) for 48 hours before bulk RNA sequencing analysis of the relevant pathways being up- or downregulated (D). (E) Flow cytometry analysis of organoids (trypsinized to single cells) after propidium iodine staining showing the relative percentage of dead cells for each treatment (Anova test, ∗P < .05 and ∗∗P < .01).