Changes in the gut microbiota of NOD mice in response to an oral Salmonella-based vaccine against type 1 diabetes

We developed an oral Salmonella-based vaccine that prevents and reverses diabetes in non-obese diabetic (NOD) mice. Related to this, the gastrointestinal tract harbors a complex dynamic population of microorganisms, the gut microbiome, that influences host homeostasis and metabolism. Changes in the gut microbiome are associated with insulin dysfunction and type 1 diabetes (T1D). Oral administration of diabetic autoantigens as a vaccine can restore immune balance. However, it was not known if a Salmonella-based vaccine would impact the gut microbiome. We administered a Salmonella-based vaccine to prediabetic NOD mice. Changes in the gut microbiota and associated metabolome were assessed using next-generation sequencing and gas chromatography-mass spectrometry (GC-MS). The Salmonella-based vaccine did not cause significant changes in the gut microbiota composition immediately after vaccination although at 30 days post-vaccination changes were seen. Additionally, no changes were noted in the fecal mycobiome between vaccine- and control/vehicle-treated mice. Significant changes in metabolic pathways related to inflammation and proliferation were found after vaccine administration. The results from this study suggest that an oral Salmonella-based vaccine alters the gut microbiome and metabolome towards a more tolerant composition. These results support the use of orally administered Salmonella-based vaccines that induced tolerance after administration.


Introduction
Type 1 diabetes (T1D) is characterized by autoimmune-mediated destruction of insulinsecreting pancreatic β-cells. Preproinsulin (PPI) is likely one of the diabetes-initiating self-antigens in mice [1] and humans [2][3][4]. We developed an oral antigen-specific vaccine using live attenuated Salmonella [5][6][7]. The Salmonella-based delivery system comprised PPI fused with effector (sseF) and expressed and secreted under control of the SPI2 promoter of the

Animal experiments
Seven-week-old female NOD/ShiLtJ (NOD) mice (Jackson Laboratories, Bar Harbor, ME) were maintained under pathogen-free conditions and housed at the animal care facility at City of Hope National Medical Center. The study was approved by the Institutional Animal Care and Use Committee (IACUC# 18017). Eight-week-old mice were orally vaccinated with Salmonella containing plasmid for the expression of IL10 and TFGβ and Salmonella expressing autoantigen PPI in 200 ml of 5% sodium bicarbonate on days 0 and 7. Vaccinated animals were also treated with anti-CD3 mAb for five consecutive days (days -1 to 3). The control (vehicle) treatment was 200 ml of 5% sodium bicarbonate given orally. Stools were collected from vaccine-, and control-treated animals at pre, 3-, 7-, 14-, 30-day post-vaccination.

DNA extraction
DNA was extracted from feces using the QIAamp Fast DNA Stool Mini Kit (Qiagen, Hilden, Germany) [26]. The quality and purity of the isolated genomic DNA were confirmed by gel electrophoresis and quantitated with a Qubit2.0 fluorometer using the Qubit dsDNA HS Assay (Thermo Fisher Scientific, Waltham, MA). DNA samples were stored at -20˚C.

Library preparation and sequencing
The amplicon library was cleaned and barcoded, followed by emulsion PCR using the Ion Torrent next-generation sequencing data analysis workflow (Thermo Fisher Scientific). Equal volumes of bacterial 16S rRNA and fungal ITS amplicons were pooled, cleaned with AMPure XP beads (Beckman Coulter, Brea, CA) to remove unused primers, and then exposed to endrepair enzyme for 20 minutes at room temperature. After an additional AMPure cleanup, ligation was performed at 25˚C for 30 minutes using Ion Torrent P1 and a unique barcoded "A" adaptor per pooled sample. After AMPure removal of residual adaptors, samples were concentrated to one quarter volume for 1 hour using a vacuum (Labconco, Kansas City, MO) under heat. All separate barcoded samples were then pooled in equal amounts (10 μL) and sizeselected for the anticipated 16S and ITS range (200 to 800 bp) using Pippin Prep (Sage Bioscience, Beverly, MA). The library was amplified for 7 cycles and quantitated on a StepOne qPCR instrument (Thermo Fisher Scientific) ahead of proper dilution to 300 pM going into the Ion-Sphere templating reaction on the Ion Chef (Thermo Fisher Scientific). Library sequencing was completed on an Ion Torrent S5 sequencer (Thermo Fisher Scientific), and barcode-sorted samples were analyzed in our custom pipeline based on GreengenesV13_8 and Unite database V7.2, designed for the taxonomic classification of 16S rRNA and ITS sequences, respectively. Sequencing reads were clustered into operational taxonomic units (3% distance), described by community metrics and taxonomically classified within the Qiime bioinformatics pipeline (ver. 1.9)

Metabolomics profiling
Metabolomic profiling was performed as published [27]. Feces (20 mg) was mixed with methanol: chloroform (1:1) and vortexed for 15 minutes. Samples were centrifuged at 10,000 RPM for 10 minutes and the filtrates were collected and dried using a rotatory evaporator (Buchi, Germany). The metabolite extracts were derivatized by adding 50 μL of N-trimethylsilyl-Nmethyl trifluoroacetamide and trimethylchlorosilane (MSTFA + 1% TMS) mixture followed by incubation at 50˚C for 30 minutes prior to gas chromatography-mass spectrometry (GC-MS) analysis as described [28]. GC-MS measurements were carried out using an Agilent Model 7683 Autosampler, 6890 Gas Chromatograph and 5975 Inert Mass Selective Detector in the Electron Impact (EI) mode. EI energy was set to 70 eV. Separation was carried out on an Agilent HP5-MS column with dimensions 30 m x 250 μm x 0.25 μm. Ultra-High Purity Grade He (Airgas) was used as the carrier gas with the flow set to 0.8 mL/minute in constant flow mode. The initial oven temperature was set to 45˚C for 1 minute followed by a 30˚C/minute ramp to a final temperature of 300˚C, which was maintained for 3 minutes. A 3.2-minute solvent delay was used. The injector temperature was set at 220˚C. The MSD was set to scan the 40-1050 m/z range. Data collection and analysis were performed using MSD Enhanced Chemstation software (Agilent). Product spectra were identified by comparison of the measured fragmentation patterns to those found in the NIST 08 Mass Spectral Library. Differential metabolite expression was calculated as the fold-change from day 0 (before vaccination) by employing Ingenuity Pathway Analysis (IPA, Qiagen, version 84978992) core analysis with metabolomics analysis by expression Log Ratio using default settings and no cutoffs. Assessment between timepoints was created using the comparison analysis tool in IPA. Analyzed pathway data was exported into CSV (comma-separated values) files and graphed in Prism.

Statistical analysis
A custom pipeline based on the Greengenes V13_8 database designed for the taxonomic classification of 16S rRNA sequences was employed. Downstream data analysis was performed using Qiime Platform (ver. 1.9) [29]. The Dunnett's multiple comparisons test and Dunn's multiple comparisons test (non-parametric) were used to analyze the differences in the percent abundance of gut microbiome within the same cohorts before and after vaccination (paired for longitudinal analysis). Bonferroni's and Mann-Whitney multiple comparisons tests were used to compare differences between vaccinated and vehicle-treated groups at different time points. Multiple paired t tests with adjusted p-values using the two-step Benjamini, Kreiger, and Yekutiele procedure were used to compare differences in post-vaccinated metabolite concentrations in animals on day 3, 7, 14, and 30 with pre-vaccinated levels. For all tests, a p < 0.05 was considered significant. Statistics were performed using Graphpad Prism software.

Microbiota analysis
Fifty fecal samples were collected from vaccine-(n = 25) and control (vehicle)-treated NOD mice. The total number of bacteriome and mycobiome reads was 3.7 million. Annotated sequences were used to study the microbial community using the phylotype approach. Sequences were assembled in operational taxonomic units (OTUs) at the lowest annotation level (genus). The relative abundance for each OTU was calculated. The resulting abundance matrix was used to calculate the community alfa diversity indexes. Relative abundances of bacterial and fungal phyla, genera, and species are represented in S1 and S2 Figs.

Effect of oral Salmonella-based combination therapy on gut bacteriome
To answer the question if the oral bacterial vaccine alters the gut bacterial community composition, the difference between vaccine-and vehicle-treated groups was assessed before and after vaccination on days 3, 7, 14, and 30 post-vaccination. The dynamics of the murine gut microbiota was studied at the levels of phylum and genus.
The murine gut microbiota was dominated by two bacterial phyla Firmicutes and Bacteroidetes that accounted for over~95% of sequence reads at all time points (S3A Fig). The remaining diversity was attributed to Verrucomicrobia, Tenericutes, Actinobacteria, and Proteobacteria that accounted for~4% of the remaining microbiota abundance. The phyla with abundance less than 1% were combined and designated "other" (S3A Fig). Longitudinal analysis of the fecal bacterial composition was conducted. At the phylum level, a significant decrease in Firmicutes on days 14 and 30 (p = 0.005) (Fig 1B) was observed for vaccine-treated mice, while no change in the Firmicutes was found in vehicle-treated mice (Fig 1A). Bacteroidetes was decreased on day 3 and significantly increased on day 30 (p = 0.009) (Fig 1B) in vaccinetreated mice, but not in vehicle-treated mice (Fig 1A). Actinobacteria was significantly increased in vaccine-treated mice on days 3, 7 and 30 (p = 0.01, p = 0.02, and p = 0.02, respectively) (Fig 1B), while a significant increase in Actinobacteria on day 30 (p = 0.028) was observed in vehicle-treated mice (Fig 1A). No change was detected in fecal Protobacteria in vaccine-treated mice (Fig 1B), while in vehicle-treated mice a significant decrease in Protobacteria (p = 0.011) was noted (Fig 1A).
The genera levels of microbial dynamics are displayed in S3B Fig. Ten genera with abundance >1% are presented. The genera with abundance <1% were combined and designated "other" (S3A Fig). There genera dominated, Lactobacillus, Oscillospira, and Akkermansia accounting for~75%. In longitudinal analysis, Lactobacillus was significantly increased on day 3 (p = 0.043) in vaccine-treated mice (Fig 2B) and significantly increased on days 7 (p = 0.007) and 30 (p = 0.034) in vehicle-treated mice (Fig 2A). The genus Oscillospira was significantly decreased on day 14 (p = 0.012) (Fig 2B) in vaccine-treated animals and little changed in vehicle-treated mice (Fig 2A). The genus Ruminococcus was significantly decreased on days 3 (p = 0.04) and 30 (p = 0.007) (Fig 2B) but decreased without significance in vehicletreated mice (Fig 2A). No change in the abundance of Akkermansia in any of the mice was observed (Fig 2).
At the phylum level, changes over the 5 time points for the 6 most abundant fecal bacteria in vaccine-treated mice compared with the vehicle-treated mice are illustrated in Fig 3. The relative abundance of the phylum Bacteroidetes was significantly increased in the gut of vaccine-treated mice on day 30 (p = 0.024) and on day 7 (p = 0.03) for Proteobacteria compared with vehicle-treated mice (Fig 3). In contrast, the abundances in the bacterial phyla Firmicutes and Actinobacteria in vaccine-treated mice were significantly decreased on day 30 (p = 0.023, and p = 0.008) compared with vehicle-treated mice (Fig 3). No changes were observed in the relative abundance of the bacterial phyla Verrucomicrobia and Tenericutes between vaccineand vehicle-treated groups at any time points (Fig 3).
No significance changes occurred in the 6 most abundant fecal bacterial compositions at the genus level over any time points except a significant increase in the genus Anaeroplasma on day 7 (p = 0.008) and Coprococcus on day 14 (p = 0.03) in vaccine-compared with vehicletreated mice (Fig 4). There is a relation between Salmonella and phylum Protobacteria. Thus, a closer look into this was done. A significant increase in the phylum Protobacteria (relative abundance 0.5%) on day 7 (p = 0.029) occurred in vaccine-treated mice compared to vehicle-treated mice (Fig 5). The class Gammaproteobacteria (~0.3%) significantly increased on day 7 in vaccine-compared to vehicle-treated mice (p = 0.019) (Fig 5). Analyzing the class Gammaproteobacteria including families Enterobacteriaceae (0.3%) and Pseudomonadaceae (0.1%) revealed a relative abundance of fecal bacteria from Enterobacteriaceae with a significant increase noted in vaccinecompared to vehicle-treated on day 7 (p = 0.022) (Fig 5). Also, a significant increase in the relative abundance of the bacterial family Pseudomonadaceae was noted in vaccine-compared to vehicle-treated mice on days 7 (p = 0.008) and 30 (p = 0.008) (Fig 5).

Effect of oral Salmonella-based combination therapy on the gut mycobiome
The mouse gut mycobiome is dominated by the fungal phylum Ascomycota that accounted for over 99% of sequence reads (S4A  In summary, vaccination had no significant effect on gut microbiota composition immediately after initiation of treatment. The effects were noted at later times and were characterized by the increased abundance of the phylum Bacteroidetes and decreased abundance in Firmicutes and Actinobacteria.
Saturated fatty acids (SFA) including myristic acid on days 14 and 30 and palmitic acid at day 7 were significantly decreased in feces following vaccination. On the other hand, stearic acid was significantly increased on day 30 in feces following vaccination while other SFA such as hexadecanoic acid ethyl ester (palmitic acid), pentadecanoic acid, and heptadecanoic acid (margaric acid) varied but without significant changes (Fig 6).
Unsaturated fatty acids (UFA) such as 10-undecynoic acid (at days 7 and 14) and oleic acid (on days 3 and 30) were significantly decreased following vaccination while α-linolenic acid was detected only on days 3 and 30 following vaccination (Fig 6). Interestingly, azelaic acid as The amount represents the relative percentage of a metabolite in relation to total metabolites detected in the fecal pellet for 5 mice before and on days 3, 7, 14, and 30 post-vaccination ± SD. The statistical significance was calculated using multiple paired t test with Benjamini, Kreiger, and Yekutiele corrected significance level indicated (p value < 0.05). Red p value means significant increase and blue p value means significant decrease in comparison to the pre-treated group.
Other acids such as 3-(3-hydroxyphenyl) propanoic acid (3-hydroxycinnamic acid) was significantly increased in the feces at day 3 following vaccination and then returned to normal levels on day 30. Other metabolites such as phloretic acid (3-(4-hydroxyphenyl) propanoic acid) and 4-hydroxybenzeneacetic acid were not changed in the feces after vaccination while ethanolamine was significantly decreased on days 7 and 30 after vaccination. Eicosane, a saturated hydrocarbon, was significantly increased after vaccination while heptadecane was significantly decreased (Fig 6).
Sugar metabolites including levoglucosan (on day 30), and glucopyranose, and methyl alpha-D-glucofuranoside were significantly decreased in the feces following the vaccination, while D-ribose, D-(+)-talofuranose, xylose, beta-D-(+)-xylopyranose and N-acetyl-D-galactosamine did not change in the feces after vaccination. L-rhamnose was significantly increased on day 7 in the feces after vaccination (Fig 6).
Glycerol monostearate (2-monostearin) and glyceryl palmitate (1-monopalmitin) were significantly decreased on day 7 following vaccination and returned to normal on day 14, and 1,3-propanediol was significantly decreased on day 14. Glycerol and glyceric acid were increased after vaccination and become significantly increased on day 30 (Fig 6).
Fecal metabolites such as succinic acid (butanedioic acid) and oxalic acid were not changed following vaccination (Fig 6). Urea was detected in the feces of the animal on day 3 then sharply increased on day 7 followed by a decrease on day 14 and then a final increase on day 30 post-vaccination while uracil (a nucleic acid) was significantly decreased following vaccination (Fig 6).
To understand the significance of the metabolomic changes, we utilized Ingenuity Pathway Analysis (IPA). Metabolite data ( Table 1) were entered as fold change in relation to data collected from day 0 (pre-vaccine). The predicted disease and functions pathways identified within molecular functions showed an overall decrease in the ATP production, as seen in the pathways 'Release of ATP', 'Concentration of ATP', and 'Synthesis of ATP' (Fig 7A). Upregulated pathways involved cell death (Apoptosis, Necrosis) on days 3, 14, and 30 post-vaccines but these decreased on day 7 (S6A Fig). A potential activation of immune cell functions (Endocytosis by eukaryotic cells, Engulfment of cells) was noticed with an increased activity at all days post vaccination. Other pathways showed a trending decrease in 'Release of reactive oxygen species' and 'Generation of superoxide' or a trending increase in 'Synthesis of carbohydrate', 'Concentration of cholesterol', 'Synthesis of D-glucose', and 'Biosynthesis of hydrogen peroxide' (S6A Fig). Further analysis of immune-focused diseases and functions was performed (Fig 7B). The 'Activation of leukocytes' was increased on days 3, 7, and 14 post-vaccines but decreased on day 30. Other pathways, such as 'Inflammatory response', 'Activation of phagocytes', 'Activation of macrophages', 'Activation of myeloid cells' and 'Proliferation of CD4+ T-lymphocytes were increased on all days except day 7. Finally, in the first 14 days, 'Leukopoiesis' (formation of white blood cells) and 'Inflammation of organ' were decreased. This suggests a vaccine-associated activation of the immune system (S6B Fig). The metabolite data were processed to determine if the changes of global regulators could be predicted. Using IPA Upstream Regulator Analysis, we identified proteins that are predicted to be activated or inhibited (Fig 7C). Of note, the anti-inflammatory cytokine IL-37 [30] is predicted to be activated days 3, 7, and 30 post-vaccination. Also, different regulators of mitochondria and metabolism were predicted to be upregulated on day 7 including NAD(P) transhydrogenase (NNT: mitochondria redox) [31], lipid-sensing peroxisome proliferator-activated receptor delta (PPARD) [32], and long-chain fatty acid transport carnitine palmitoyltransferase I (CPT1B) [33]. While both NNT and PPARD were downregulated on days 3 and 30. Other transcriptomic co-factors and regulators, such as ARNT (aryl hydrocarbon receptor nuclear translocator) [34], MYC [35] and PPARGC1 (Peroxisome proliferator-activated receptor gamma coactivator 1) [36], were predicted to be upregulated on days 3, 14 and 30 post-vaccination. Uncoupling protein 1 (UCP1) [37] growth factor and fatty acid binding protein 4 (FABP4) [38] were upregulated on days 3 and 30 but downregulated on days 7 and 14 postvaccination. Matrix metalloproteinase 11 (MMP11) [39], an inflammatory mediator, and stearoyl-CoA desaturase (SCD) [40], an enzyme that participates in the synthesis of UFA from SFA, were downregulated on days 3 and 30 post-vaccination (S6C Fig). Overall, this analysis indicated a significant change in metabolism, inflammation and proliferation following vaccine administration.

Discussion
An oral antigen-specific vaccine using live attenuated Salmonella prevented and reversed T1D in NOD mice [5][6][7]. To better understand possible effects of the vaccine, changes in the composition of the gut microbiota were determined at different time points following vaccination. This is important because the microbiota can protect the host against microbial pathogens [41] while effects of Salmonella infection on the gut microbiota may lead to inflammation that favors the growth of pathogens [42,43]. And while the Salmonella strain employed in the vaccine was attenuated, it still retained some infectious capacity.
Changes in the gut microbiota were implicated in altering diabetes susceptibility in NOD mice [44,45]. Related to this, changes in the gut microbiome were seen between healthy and diabetic individuals mimicking to an extent data from NOD mice [46]. In this study, we compared the gut microbiome in vaccine-and vehicle-treated prediabetic mice. We found no changes in the composition of gut microbiota on the phylum level after vaccine administration except for a decrease in Firmicutes, and Actinobacteria on day 30 and increase in Bacteroidetes on day 30. Furthermore, no changes in the composition of bacteria genera between vaccineand vehicle-treated mice were noted at any time points except an increase of Anaeroplasma and Corprococcus on days 7 and 14, respectively. Our vaccine combination caused an increase in the gut phylum Proteobacteria, the class Gammaproteobacteria, and the families Pseudomonadaceae, and Enterobacteriaceae. Salmonella genus belongs to Enterobacteriaceae (family), Gammaproteobacteria (class), and Proteobacteria (phylum) [12]. This could be important for the changes seen after vaccination. Overall, the data indicate that administration of the Salmonella-based vaccine altered the balance of the gut microbiota in prediabetic NOD mice. However, the modest changes noted before and after treatment within the same treatment-group (either vaccine group or vehicle group) were likely due to other factors such as type of food, handling, temperature, and age. The slight increase in family Pseudomonadacea may be attributed to an imbalanced gastrointestinal environment due to temporary changes in the composition of the gut microbiome [47].
Studying the changes in the composition of gut microbiota at short (7-14 days post vaccination) and longer times (30 days post-vaccination) reflects the expected lifecycle of the vaccine. Short term, there was an increase in only a few taxa by the vaccine, suggesting a successful infection and proliferation of the vaccine within the GALT. At day 30, the composition of microbiota mostly returned to its pre-vaccinated state except for an increase in the Bacteroidetes phylum and decrease in the Firmicutes and Actinobacteria. Narratively, this aligns with our data on the approximate 3-week clearance time of the vaccine. This suggests that the microbiota may have assumed a more tolerant state. When the vaccine successfully infiltrated the GALT, Treg abundance was increased throughout several lymphatic tissues especially on day 30 post-vaccination and this was associated with less beta cell injury [6,7]. Of some possible relevance, gut Bacteroidetes were associated with increased Tregs [22]. Based on our data, we postulate that the microbiota responded to the vaccine and participated in immune regulation and induction of tolerance by increasing Tregs and regulatory cytokines and decreasing inflammatory cytokines [6, 7]. The intestinal epithelium plays an important role as a filter that translocates water, nutrients, and bio-reactive compounds from the intestine to the circulation [48]. Inflammation can increase the intestinal permeability to permit antigens to induce autoimmune response and promote the development of diabetes [49,50].
In this study, the metabolic profiling of the vaccine-treated animals was determined and identified. Of the 60 identified metabolites, almost a third was associated with diabetes. The rest of the common metabolites were associated with digestion and were expected based on the diet of the mice. For example, amino acids (AA) including glycine and alanine are associated with glucose metabolism through promoting insulin secretion [51,52]. Furthermore, elevated levels of valine, isoleucine, leucine, tyrosine, and phenylalanine are correlated with insulin intolerance and secretion suppression [53,54]. Throughout, the levels of most of AAs showed high variance from day to day. However, by day 30 most of the AAs returned to pre-vaccination levels. Of note, multiple predicted Upstream Regulators were up or down regulated based on the combination of changes in amino acid concentrations (Fig 7C). For example, CPT1B was predicted to be upregulated because of the increase in leucine and valine, and the decrease in alanine, glycine, and uracil (S6C Fig). When MMP11 is active, it could up-regulate amino acids such as alanine, glycine, isoleucine, tyrosine, phenylalanine, and ethanolamine. Also, due to the combined effects of alanine and glycine, in association with butyric acid and succinic acid, the anti-inflammatory protein IL37 is predicted to be activated on all days observed except day 14 (S6C Fig). Lactate is produced by gut bacteria under anaerobic conditions and is metabolized in the colon by lactate-utilizing bacteria to produce beneficial short-chain fatty acids (SCFAs) including butyrate and propionate. Lactate metabolites play an important role in the stability of the gut microbiome [55]. SCFA metabolites generated by gut microbiota exert systemic antiinflammatory effects through involvement in the production of immunoglobulin A and immunosuppressive cytokines [56]. These SCFAs may have an effect on metabolic and immune response, and inflammatory disease [57,58]. For example, changes in the 'Biosynthesis of hydrogen peroxidase are due, in part, to the changes in butyric and succinic acids (S6A Fig). SCFAs such as acetate and propionate can influence insulin sensitivity and glucose tolerance via glycemic-mediated response. Propionate can decrease the fasting blood glucose, reduce gluconeogenesis, promote the utilization of glucose, and elevate glucagon-like peptide [59].
Butyrate maintains the integrity of the gut epithelium by inducing mucin synthesis [57]. NOD mice had lower abundance of SCFAs, especially butyrate, when compared to non-diabetic mice [60]. Significant elevations in 3-hydroxybutyric acid (β-hydroxybutyric acid, BHBA) is an indicator of diabetic ketoacidosis in T1D [61,62]. In this study, our vaccine was associated with a decrease in BHBA (Table 1). However, in general, the Salmonella-based vaccine increased the levels of SCFAs in the feces of NOD mice.
An increase in glutamate and a decrease in TCA metabolites were observed in diabetic NOD mice [72]. Our vaccinated mice showed a significant decrease in L-5-oxoproline (pyroglutamic acid) ( Table 1). Vaccination was not associated with a change in the TCA metabolite, oxalic acid. Monoacylglycerol increased the glucagon-like peptide-1 (GLP-1) and gastric inhibitory polypeptide (GIP) levels following administration to the small intestine [73]. Our vaccine did not show effect on the abundance of monoacylglycerol ( Table 1). Steroid metabolites in the gut of NOD mice were changed as a result of Salmonella-induced inflammation [74]. Steroids (cholesterol, coprostanol, stigmasterol, and campesterol) and eicosanoid can impact wound healing, sugar metabolism, and immune system regulation [74]. In fact, changes in cholesterol influenced multiple Disease and Functions pathways such as 'Concentration of ATP', 'Biosynthesis of amide', 'Synthesis of lipid' (S6A Fig), and 'Activation of leukocytes', 'Activation of phagocytes', and 'Inflammation of organ' (S6B Fig). Many of the metabolites assessed in the feces of vaccine-treated mice were altered over time. However, analysis was not obtained on samples from unvaccinated mice. Hence, it is unknown if the changes in the metabolome were secondary to the vaccine alone.
In summary, we demonstrated that an oral Salmonella vaccine did not adversely alter gut microbiota diversity and proportions. At 30 days post-vaccination, the results suggested a transition in microbiota towards a more tolerant composition as the Tregs increased in mucosal tissues. Metabolic analysis indicated that after vaccination changes in metabolism, inflammation, and proliferation occurred and this was associated with immune activation and increased regulatory and decreased inflammatory cytokines. These results support the oral administration of Salmonella vaccines as carriers for autoantigens and immunomodulators against T1D. Further studies to understand the mechanistic role of the microbiota in immune function and to relate the composition of the microbiota to oral vaccines are required.