Change in Populations of Macrophages Promotes Development of Delayed Gastric Emptying in Mice

Abstract

Background & aims: Muscularis propria macrophages lie close to cells that regulate gastrointestinal motor function, including interstitial cells of Cajal (ICC) and myenteric neurons. In animal models of diabetic gastroparesis, development of delayed gastric emptying has been associated with loss of macrophages that express cytoprotective markers and reduced networks of ICC. Mice with long-term diabetes and normal gastric emptying have macrophages that express anti-inflammatory markers and have normal gastric ICC. Mice homozygous for the osteopetrosis spontaneous mutation in the colony-stimulating factor 1 gene (Csf1op/op) do not have macrophages; when they are given streptozotocin to induce diabetes, they do not develop delayed gastric emptying. We investigated whether population of the gastric muscularis propria of diabetic Csf1op/op mice with macrophages is necessary to change gastric emptying, ICC, and myenteric neurons and investigated the macrophage-derived factors that determine whether diabetic mice do or do not develop delayed gastric emptying.

Methods: Wild-type and Csf1op/op mice were given streptozotocin to induce diabetes. Some Csf1op/op mice were given daily intraperitoneal injections of CSF1 for 7 weeks; gastric tissues were collected and cellular distributions were analyzed by immunohistochemistry. CD45⁺, CD11b⁺, F4/80⁺ macrophages were dissociated from gastric muscularis propria, isolated by flow cytometry and analyzed by quantitative real-time polymerase chain reaction. Cultured gastric muscularis propria from Csf1op/op mice was exposed to medium that was conditioned by culture with bone marrow-derived macrophages from wild-type mice.

Results: Gastric muscularis propria from Csf1op/op mice given CSF1 contained macrophages; 11 of 15 diabetic mice given CSF1 developed delayed gastric emptying and had damaged ICC. In non-diabetic Csf1op/op mice, administration of CSF1 reduced numbers of gastric myenteric neurons but did not affect the proportion of nitrergic neurons or ICC. In diabetic Csf1op/op mice given CSF1 that developed delayed gastric emptying, the proportion of nitrergic neurons was the same as in non-diabetic wild-type controls. Medium conditioned by macrophages previously exposed to oxidative injury caused damage to ICC in cultured gastric muscularis propria from Csf1op/op mice; neutralizing antibodies against IL6R or TNF prevented this damage to ICC. CD45⁺, CD11b⁺, and F4/80⁺ macrophages isolated from diabetic wild-type mice with delayed gastric emptying expressed higher levels of messenger RNAs encoding inflammatory markers (IL6 and inducible nitric oxide synthase) and lower levels of messenger RNAs encoding markers of anti-inflammatory cells (heme oxygenase 1, arginase 1, and FIZZ1) than macrophages isolated from diabetic mice with normal gastric emptying.

Conclusions: In studies of Csf1op/op and wild-type mice with diabetes, we found delayed gastric emptying to be associated with increased production of inflammatory factors, and reduced production of anti-inflammatory factors, by macrophages, leading to loss of ICC.

Keywords: Cytokine; Gastroparesis; ICC; Immune Response.

Figure 1. CSF1 treatment populates the gastric muscularis propria of Csflo^p/op mice with macrophages

(A) Experimental design. Gastric emptying was determined weekly (arrows). Diabetes was induced from 12 weeks old. CSF1 (2.5μg) was given daily. Groups were as follows: CSF1-treated dbCsf1^op/op with normal (NGE) and delayed gastric emptying (DGE) , age-matched, untreated diabetic and non- diabetic Csfl^op/op (op/op, ), controls (WT ) and CSF1-treated, non-diabetic Csfl^op/op , diabetic WT with DGE (dbWT DGE, ) or NGE (dbWT NGE, ). (B) Areas studied for immunohistochemistry. (C) F4/80⁺ macrophages in the groups of mice. Scale bar: 50μm. (D) F4/80⁺ macrophages in the gastric circular muscle (lower panels) and myenteric region (upper panels) in CSF1-treated, Csfl^op/op mice. Scale bar: 25μm. (E) Quantification of macrophages in n=27 fields, N=3 mice. Data are means±SEM. (lway-ANOVA, P<0.05, Significant differences indicated for Tukey post-test comparisons by a vs non- db Csfl^op/op, b vs dbCsfl^op/op, c vs CSF1- treated non diabetic Csfl^op/op, d vs CSF1-treated diabetic Csfl^op/op, e vs WT). (F) Scatter plots of events from N=3 dissociations from gastric muscularis propria of Csfl^op/op and (G) WT mice (WT).

Figure 2. Distribution of macrophages in the gastric myenteric regions of CSF1 treated Csf1^op/op mice

(A) MHCII⁺ macrophages (green) and NRP-1⁺ structures (red). Scale bar: 50μm. (B) Quantification of proximity of macrophages to blood vessels and (C) proximity of macrophages to neuronal structures. Data are means±SEM from n=27 fields, N=3 mice. (1way-ANOVA, P<0.05, significant differences indicated for Tukey post-test by a vs CSF1-treated non-db Csf1^op/op, b vs CSF1-treated db Csf1^op/op, c vs WT).

Figure 3. Physiological data from CSF1 treated Csf1^op/op mice

(A) T_½ values. Normal range is shown as horizontal lines. Whiskers are medians with IQRs. Each point represents one animal. Red dots indicate values above normal range. (B) Blood glucose. Red dots indicate values for mice with DGE in (A). (Whiskers are means±SEM, P<0.05, 1way-ANOVA, for (A) and (B), significant differences indicated by a vs non-db Csf1^op/op, b vs CSF1-treated non-db Csf1^op/op). Effect of CSF1 on T_1/2 in non-diabetic (C) and diabetic Csf1^op/op mice (D) (ttest,**P<0.01). (E) Serum malondialdehyde levels. (Whiskers are medians with IQRs and circles represent individual animals, P<0.05, significant differences by Dunn's Multiple comparison by a vs dbCsf1^op/op and b vs CSF1-treated dbCsf1^op/op NGE).

Figure 4. ICC networks are damaged in diabetic Csf1^op/op mice with delayed gastric emptying

(A) Macrophages (green) and ICC (red) in a CSF1-treated non-diabetic Csf1^op/op mouse. En face and cross sectional views. Scale bar: 50μm (B) ICC in the different groups of Csf1^op/op mice. Scale bar: 50μm. Scores for ICC networks: (C) density, (D) connectivity and (E) integrity. Whiskers are means±SEM from N=4 mice. (1way-ANOVA, P<0.05, significant differences indicated for Tukey's posttest by a vs dbCsf1^op/op, b vs CSF1-treated non-db Csf1^op/op NGE).

Figure 5. Quantification of Neurons in CSF1 treated Csf1^op/op mice

(A) Images of HuC/D⁺ and NOS1⁺ neurons. Scale bar: 50μm. (B) Quantification of HuC/D⁺ neurons, (C) NOS1⁺ neurons and (D) proportions of NOS1⁺ neurons (n=22 fields, N=4 mice). Points represent individual fields, Bars and whiskers indicate means±SEM. (1way-ANOVA, P<0.05, significant differences indicated for Tukey's Multiple comparison test by a vs WT, b vs Csf1^op/op).

Figure 6. IL6Rα/TNFα neutralizing antibodies preserve ICC in organotypic culture

(A) ICC in fresh and cultured muscularis propria. (B) Bone marrow-derived macrophages treated with/without TBH (100μM) and/or hyperglycemia (5g/L). (C) RNA expression levels following treatment (Data are means±SD, lway-ANOVA, P<0.05, significant differences indicated for Tukey's posttest). (D) ICC after exposure to conditioned medium, with/without antibodies to IL6Ra (1 μg/mL) and TNFα (0.4μg/mL). (E) Quantification of ICC (n=12 fields, N=3 mice. lway-ANOVA, P<0.05, Significant differences indicated for Kruskal-Wallis with Dunn's Multiple comparison test by * vs TBH+IL6Ra/TNFa neutralizing abs. Scale bars in A, 50μm, B, 10μm, D, 40μm.

Figure 7. Characterization of gastric MPMs from diabetic mice with delayed and normal gastric emptying

(A) T_½ values. Whiskers are medians with IQRs. Each point represents one animal. (P<0.001, 1way-ANOVA, significant differences indicated for post-tests by * vs WT and db WT with NGE). (B) Blood glucose and (C) Serum malondialdehyde. Red dots indicate mice with DGE. Whiskers are means±SEM. (P<0.001, 1way-ANOVA, significant differences indicated for post-tests by * vs WT and db WT with NGE). Scatter plots of events dissociated and sorted from diabetic WT with NGE (D) and DGE (E), N=4. (F) RNA levels in CD45⁺CD11b⁺F4/80⁺ cells from db WT with NGE compared to mice with DGE (*P<0.05, ttest).