Gut dysbiosis modulates the immune response to factor VIII in murine hemophilia A

Abstract

The development of neutralizing FVIII antibodies is the most serious complication of hemophilia A treatment. The currently known patient- and treatment-related risk factors for inhibitor development do not accurately predict this adverse event in all patients. The composition of the gut microbiota has been shown to influence immune-mediated diseases at distant anatomical sites (eg, lungs, brain, and joints). We demonstrate that a disrupted gut microbiota can be created in a mouse model of hemophilia A using a broad-spectrum antibiotic. Under controlled conditions, this sustained dysbiosis was associated with an increase in splenic B cells and the development of higher titer, FVIII-specific immunoglobulin G antibodies after FVIII challenge. Splenic and mesenteric lymph node cytokines, T cells, and dendritic cells were unaffected before administration of FVIII. However, the immune transcriptome of both aforementioned secondary lymphoid organs was significantly modified. Short-chain fatty acids (SCFAs), which are immunomodulatory microbial metabolites, were depleted in cecal contents of the dysbiotic mice. Furthermore, supplementation of the drinking water with butyrate, the most immunologically active SCFA, successfully achieved attenuation of the FVIII immune response. Collectively, data from this exploratory study suggest that the composition of the gut microbiota alters the FVIII immune response via the action of specific microbial metabolites on the immune cell transcriptome and that oral supplementation with butyrate effectively reduces the FVIII immune response.

Graphical abstract

Figure 1.

Study overview and effects of ampicillin on the gut microbiome. (A) Study overview. Mice aged 3 weeks were weaned and then immediately commenced treatment with ampicillin, administered every 12 hours by gastric gavage for 7 days (n = 15) or not (n = 14). After a 2-week rest period, mice remaining in the study were challenged with 50 ng IV-administered FVIII. The mice were euthanized and samples collected (blue circles) for analyses at 6 weeks of age, immediately before the FVIII challenge began and at the study end point (10 weeks of age). (B) Differences in the gut microbiota between ampicillin-treated and control mice derived from a heat map of calculated z-scores. Bacterial 16s rRNA was sequenced, and samples were clustered based on study time point (6 weeks vs 10 weeks) and treatment (ampicillin vs control), using Pearson’s distance and Ward’s clustering methods. (C) Principal coordinate analysis based on weighted UniFrac distances of each microbiome sample. (D) α-Diversity was measured by using Faith’s phylogenetic diversity (PD) metric (means compared by unpaired, 2-tailed Student t test). Standard deviations are shown. (E) Relative abundance of bacterial phyla (left) and class (right). Each vertical bar corresponds to an individual sample. Sample size for all microbiome analyses remain consistent (n = 4-5). ****P < .0001.

Figure 2.

Dysbiosis enhances the immune response to FVIII. (A) Plasma was assessed at the study end point (10 weeks of age) for FVIII-specific IgG and was quantified by the highest titer or dilution at which the ELISA was positive. (B) The magnitude of inhibitory activity was assessed by Bethesda assay. (C) FVIII-specific IgG was quantified based on IgG subclass using a mouse reference serum standard curve. The means and standard deviations were compared using Mann-Whitney U test for total FVIII-specific IgG and an unpaired, 2-tailed Student t test for all other analyses (n = 10). **P < .01; ***P < .001; ****P < .0001.

Figure 3.

Dysbiosis affects the total number of B cells in spleen cells but not in MLNs. Cell enumeration by flow cytometry was performed on whole-cell isolates from dissected spleen and MLNs of control and ampicillin-treated mice at 6 weeks of age, before FVIII infusion (n = 9 for each cohort). B220⁺ B cells (A) and CD4⁺ T cells (B) as a proportion of total cells. Other cells analyzed include: CD25⁺FoxP3⁺ (C) and CD25⁻Foxp3⁺ (D) T cells represented as a proportion of total CD4⁺ cells, and CD11c⁺ DCs (E) as a proportion of total immune cells, and CD103⁺ cells (F) as a proportion of total CD11c⁺ DCs. The means and standard deviations within each immunological organ were compared by using an unpaired, 2-tailed Student t test. *P < .05.

Figure 4.

Dysbiosis does not affect homeostatic splenic cytokine levels before FVIII challenge. Spleens were isolated before FVIII challenge in ampicillin-treated (n = 3) and control (n = 4) mice. The protein content of splenic lysate was normalized and then analyzed on an 18-plex cytokine array (A) in addition to a TGF-β panel (B). The means and standard deviations in the cohorts were compared with the unpaired, 2-tailed Student t test. Multiple test corrections were made by the Holm-Šidák method.

Figure 5.

Dysbiosis alters the mRNA transcript profile in both spleen and MLNs. RNA was isolated from whole-cell isolates of the spleen and MLNs of ampicillin-treated compared to control mice at 6 weeks of age, before FVIII infusion (n = 3 in each). RNA was analyzed using a Nanostring murine immunology panel of 547 genes. (A) Differences in gene expression within the spleen (left) and MLNs (right) as a result of ampicillin-treatment is visualized on a volcano plot. An arbitrary fold-change cutoff of ±1.25 and an α = 0.05 was applied to focus on potentially significant genes, and those that met criteria are highlighted in red. Significance was calculated using the Welch t test for unequal variances. (B) Differences in gene expression between both cohorts are represented through their respective z-scores on a heat map. Clustering was completed by using the Pearson squared distance metric of z-scores with Ward’s clustering method.

Figure 6.

Dysbiosis results in reduced cecal SCFAs and supplementation with butyrate attenuates anti-FVIII immune response. (A) Levels of the cecal SCFAs acetate, butyrate, and propionate were quantified by NMR at 6 weeks of age, before FVIII infusion, and compared between ampicillin-treated (n = 5) and control (n = 4) mice. (B) In a separate experiment, mice were maintained in regular housing conditions and supplemented with 100 mM butyrate in the drinking water starting from 3 weeks of age (n = 7). They were then challenged with FVIII as previoulsy described. Plasma was analyzed for FVIII-specific IgG and compared with plasma of mice receiving only water (n = 10). The means and standard deviations of SCFA levels and FVIII-specific IgG were compared by using the unpaired, 2-tailed Student t test and the Mann-Whitney U test, respectively. **P < .01; ***P < .001; ****P < .0001.