A SINGLE-CELL SURVEY OF GASTRIC MUSCULARIS MACROPHAGES IN DIABETIC GASTROPARESIS

ABSTRACT

Background: Gastric muscularis propria immune cells play an instrumental role in homeostasis and disease. A subset of these cells, muscularis macrophages (MMs) are involved in the pathobiology of diabetic gastroparesis (DG) but are poorly understood. This study aims to survey transcriptional and functional profiling of gastric MMs in DG and diabetes. Methods: Full-thickness gastric body biopsies were obtained from patients with DG and diabetic controls. CD45+ cells were isolated from dissociated muscle tissue using magnetic beads. 10xGenomics was used for scRNA-seq library prep and cells sequenced by Illumina HiSeq4000. Bioinformatic analyses was performed using Suite and Seurat. Myeloid cells were annotated through a pseudogating strategy that identifies cells by differential expression levels of HLA-DR, CD14, CD11b, and CD11c based on flow cytometry-based gating utilized in a recent analysis of human small intestinal MMs. Canonical signaling pathways were determined using Ingenuity Pathway Analysis (IPA). Results: A total of 21,740 high-quality single-cell transcriptomes were generated from 16 subjects (DG=6, age 32±8 yr, BMI 23.7±3.9, 48.2±40.1% 4 hr gastric retention, average GCSI score 3.7±0.5;Diabetic controls= 10, age 53±13 yr, BMI 42.2±5.7). Through annotating 8,693 myeloid cells (DG 1509, Controls 7184), we characterized 1,788 as MMs (CD45+HLA-DR+) and 448 as dendritic cells (CD14-CD11c+). Utilizing a priori markers for pseudogating, the MMs were divided into four populations (Figure 1) subset 1 (CD14+CD11c+HLA-DRint, 5.6%), subset 2 (CD14+CD11c+HLA-DRhi, 36.0%), subset 3 (CD14+CD11c-CD11b-, 41.8%), and subset 4 (CD14+CD11c-CD11b+, 16.6%). The overall proportions of cells in the 4 subsets were similar to a prior approach in small bowel using gating. The expected ratio of cells from DG/diabetic control was 21% based on imputed cells. Subsets 1 and 4 were significantly decreased in DG compared to controls with ratios 15% and 14% respectively while subsets 2 and 3 were unchanged (21% and 20%). On IPA, phagosome formation and immune cell trafficking represented canonical signaling pathways of subset 1 and coronavirus phagocytosis pathway and phagosome formation of subset 4. Canonical genes of subset 1 included S100A12, A8, A9, and CSTA and subset 4 as LYVE1, MAF, MRC1 (CD206), MS4A4, and A2M. Subset 4 also had the highest expression of neuron-related genes (NPTX2, BMP2) similar to that observed in the small intestine. Conclusions: Pseudogating based on the transcriptomic expression of gastric immune cells reveal MM clusters similar in gene expression and proportions to previously characterized MMs in human small bowel using gating. The reduction of MM clusters associated with anti-inflammatory, phagocytosis, and neuronal signaling in specialized MMs subsets may suggest candidate targets in the pathophysiology of DG. Supported by NIHDK074008. (Figure Presented) Figure 1. Single-Cell RNA-Seq Profiling of Human Gastric Muscularis Macrophages in DG and Diabetes. T-distributed Stochastic Neighbor Embedding (tSNE) plot of muscularis macrophages in DG and diabetic control subjects by their differential genes from MAST (FDR < 0.05), color-coded by Status. *Mf1 and Mf2 not visualized as distinct clusters due to inadequate separation of overall gene expression in cells distinguished by HLA-DRint (Mf1) and HLA-DRhi (Mf2)