Humoral immune responses against gut bacteria in dogs with inflammatory bowel disease

Abstract

Inflammatory bowel disease (IBD) in dogs is associated with clinical signs of intestinal dysfunction, as well as abnormal lymphocytic and myeloid cell infiltrates in the small and/or large intestine. Thus, in many respects IBD in dogs resembles IBD in humans. However, the factors that trigger intestinal inflammation in dogs with IBD are not well understood and have been variously attributed to immune responses against dietary antigens or intestinal antigens. Previous studies in humans with IBD have documented increased production of IgG and IgA antibodies specific to intestinal bacteria, and this abnormal immune response has been linked to disease pathogenesis. Therefore, we investigated the humoral immune response against gut bacteria in dogs with IBD, using flow cytometry to quantitate IgG and IgA binding. Studies were also done to investigate the source of these antibodies (locally produced versus systemic production) and whether greater antibody binding to bacteria is associated with increased inflammatory responses. We found that dogs with IBD had significantly higher percentages and overall amounts of IgG bound to their intestinal bacteria compared to healthy dogs. Similarly, significantly higher percentages of bacteria were IgA+ bacteria were also found in dogs with IBD. Serum antibody recognition of gut bacteria was not different between healthy dogs and dogs with IBD, suggesting that anti-bacterial antibodies were primarily produced locally in the gut rather than systemically. Importantly, bacteria in the Actinobacteria phylum and in particular the genus Collinsella had significantly greater levels of antibody binding in dogs with IBD. Based on these findings, we concluded that antibody binding to commensal gut bacteria was significantly increased in dogs with IBD, that particular phyla were preferential targets for gut antibodies, and that anti-bacterial antibody responses may play an important role in regulating gut inflammation.

Fig 1. IgG⁺ and IgA⁺ fecal bacteria in healthy dogs and dogs with IBD.

(A) The percentages of IgG⁺ bacteria are plotted in dogs with IBD versus healthy dogs. (B) The amount of IgG bound to each bacterium (MFI) is plotted for the two groups of animals. IgA binding percentages and total IgA binding to each bacterium are depicted in C and D, respectively. Data are plotted as Mean ± SD. Statistical differences were calculated using two-tailed unpaired t-test (A,B,C) or a Mann-Whitney U test (D).

Fig 2. Ig-binding fecal bacteria.

Immunofluorescence staining and imaging of fecal bacteria from a healthy dog (top row) and from a dog with IBD (bottom row). IgA bound to bacteria indicated as green, while IgG⁺ bacteria indicated as red. Bacteria with both bound antibodies show up as yellow images in merged figures. Scale bar indicates 10 μm.

Fig 3. Serum IgG recognition of E. coli isolated from healthy dogs and dogs with IBD.

Six separate fecal isolates of E. coli (3 from dogs with IBD and 3 from healthy dogs) were incubated with serum from dogs with IBD (n = 20) and healthy dogs (n = 9), and IgG binding to the surface of bacteria was quantitated using flow cytometry, as noted in Methods. Scatter plots depicting IgG⁺ bacteria percentages in healthy versus IBD dogs plotted. The percentages of IgG⁺ bacteria were not significantly different between the two groups of animal sera (P = 0.41). (†) Indicated E. coli isolates from normal dogs, while (‡) indicated E. coli isolates from dogs with IBD. Data were plotted as Mean ± SD. Statistical differences were calculated using one-way ANOVA.

Fig 4. Macrophage phagocytosis of fecal bacteria from dogs with IBD versus healthy dogs.

(A) Fecal bacteria (PI staining; red) from dogs with IBD and healthy dogs (n = 5 per group) were incubated with primary cultures of canine monocyte-derived macrophages and bacterial uptake was determined using flow cytometry, as described in Methods. Images were obtained using confocal microscopy, with PI stained bacteria visualized as red objects within cultured macrophages. DAPI staining (blue) demonstrates cell nuclei. Similar results were obtained in at least n = 3 repeated, independent studies. Box plot comparing the percentage of macrophages containing intracellular bacteria (B) and the relative number of bacteria per macrophage (C), when bacteria from dogs with IBD and healthy control dogs were compared. Statistical differences were calculated using unpaired t-tests (B) and by the Mann-Whitney test (C). Scale bar as indicated.

Fig 5. Cytokine production by activated macrophages.

Canine monocyte-derived macrophages were activated by incubation and phagocytosis of non-viable fecal bacteria obtained from dogs with IBD (n = 5) and from healthy normal dogs (n = 5), as described in Methods. TNF-α and IL-10 concentrations in media obtained from macrophage cultures 24 hours after bacterial inoculation were measured using commercial canine-specific ELISA. Box plots comparing cytokine concentrations between the 2 groups of fecal bacterial samples are depicted. Statistical differences were calculated using unpaired t-tests. The assays were repeated for 3 times, total of 3 different PBMC donors.

Fig 6. Microbiome analysis.

IgG^hi sorted fecal bacteria from (n = 10) dogs with IBD, and non-sorted bacteria (n = 10; paired fecal samples from dogs with IBD) and bacteria from healthy control animals (n = 10) were analyzed by 16S rRNA sequencing, as described in Methods. (A) Species abundance heat map at taxonomic level representing average differences, with 0 = no difference, -1 and 1 representing maximum differences. (†) Showing the top 10 taxa abundance. (B) Bar graph depicting the relative abundance of 5 major phyla comparing the IgG^hi sorted population with non-sorted bacteria, obtained from same dogs with IBD. A significantly increased abundance of Actinobacteria phyla was found in IgG^hi sorted population. (C) Bar graph showed relative abundance comparing between IgG^hi sorted and non-sorted bacteria for members of Actinobacteria phyla. The data were reported as Mean ± SD, and statistical comparisons were calculated using paired t-test (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001).

Fig 7. Receiver operator curves for bacterial IgG assay.

To quantify the diagnostic ability of the bacterial IgG assay to discriminate dogs with IBD (n = 20) from normal dogs (n = 9) based on percentage IgG-binding gut bacteria, ROC curve analysis was performed. Area under the curve (AUC) was reported as 0.92, SD 0.06, P < 0.0001.