Role of Innate Oral Immunity and the Salivary Fluid in Inflammatory Bowel Disease

Abstract

Background & Aims

Oral and gut health are tightly connected through their microbiome and immunity, including in disease states. The oral adaptive immunity contributes to the severity of inflammatory bowel disease (IBD). However, the role of oral innate immunity, and more specifically the saliva, in gut microbiome and IBD is poorly understood.

Methods

We used 2 mouse models with reduced saliva, nonobese diabetic (NOD) and aquaporin 5 (Aqp5)^-/- mice, and recovery of salivation in the NOD mice by treatment with a cystic fibrosis transmembrane regulator corrector to examine the role of salivation in oral and gut microbiome, IBD, and survival.

Results

Analysis of the oral microbiome at various conditions revealed that the saliva has a minimal role in shaping the oral microbiome. However, salivation affected the composition of the gut microbiome. Moreover, the lack of saliva significantly delayed development of dextran sodium sulphate-induced colitis, but resulted in a later, age-dependent, rapidly developed weight loss and death. The dual roles of the saliva were caused by 2 immunomodulatory peptides secreted by salivary glands. Fractionation and mass spectroscopy analysis identified trefoil factor 2 (TFF2) as a protective component and the cytokine macrophage migration inhibitory factor (MIF) as the damaging component of the saliva. The effects of the salivary fluid, TFF2, and MIF were primarily due to control of the gut barrier, rather than the gut microbiome. Scavenging salivary TFF2 and MIF with antibodies resulted in exacerbating and protection, respectively, of IBD.

Conclusions

The oral innate immunity has a major role in shaping the gut microbiome through secretion of MIF and TFF2. Control of MIF and TFF2 can benefit the treatment of colitis.

Graphical abstract

Summary.

The role of the oral innate immunity in oral and gut microbiome in health and during inflammatory bowel disease is not well-understood. The current findings reveal that the oral innate immunity has a minimal role in the oral microbiome but participates in shaping the gut microbiome primarily by modulating the gut barrier, in particular at the early (protective) and late (damaging) stages of inflammatory bowel disease. Mass spectroscopy analysis identifies the cytokine macrophage migration inhibitory factor as the damaging component and the peptide trefoil factor 2 as a protective component.

What You Need to Know.

Background

The role of the oral innate immunity in oral and gut microbiome in health and during inflammatory bowel disease is not well understood.

Impact

The current findings reveal that the oral innate immunity has a minimal role in the oral microbiome but participates in shaping the gut microbiome primarily by modulating the gut barrier, in particular at the early (protective) and late (damaging) stages of inflammatory bowel disease.

Future Directions

Molecular mechanism by which the salivary MIF acts as a damaging component and the peptide FFF2 as a protective component in the gut and their clinical relevance.

Salivary gland secretion is critical in shaping adaptive and innate oral immunity.^1,2 Saliva is secreted into the oral cavity and is composed of water, proteins, and antimicrobial peptides (AMPs).^3,4 The AMPs are part of the oral innate immunity, protecting the oral cavity and the gut from pathogenic bacteria and viruses,⁵ and they participate in maintaining the homeostasis of the oral cavity microbiome.⁶ The oral cavity has a complex microbiome that hosts a large number of bacterial species, eukaryotes, archaea, and viruses.⁷ Humans secrete and swallow approximately 1.5 liters of saliva daily, which contains many components of the oral microbiome and salivary proteins.^8,9

There is an intimate relationship between the oral and gut through their microbiome, immunity, and diseases.^2,10 The best-established link is through inflammatory bowel diseases (IBDs),² with a strong correlation between IBD, salivary secretion, and oral cavity health.¹¹ IBDs include Crohn’s disease and ulcerative colitis, with different patterns of gut inflammation.¹² The etiology of IBD is complex and includes multiple causes, including genetic, environmental, microbial, and immune components.^2,13 Moreover, IBD is affected by and is manifested in multiple other tissues.¹⁴ In the case of the oral cavity, IBD both influences and is influenced by oral health.¹⁵ For example, the systemic increase in inflammatory mediators in IBD is linked to a similar increase in the gingival and salivary level of these mediators.^16,17 On the other hand, dysbiosis of the oral microbiome can contribute to the severity of IBD.^18,19

In addition to the microbiome, the oral link to IBD is through the adaptive and innate immunity. There is a strong link between IBD and periodontitis immunity.²⁰ The adaptive immunity link²⁰ is manifested by the migration of oral and gut leukocytes to other tissues.²¹ In humans, ectopic colonization of oral bacteria in the gut induces expansion of Th1 cells and inflammation that is more severe during gut dysbiosis.²² In mice, the induction of periodontitis leads to expansion of oral pathobionts that cause generation of oral Th17 cells. The oral Th17 cells migrate to the gut, where they are activated by translocated oral pathobionts and exacerbate IBD.²³ However, much less is known about innate immunity in the oral cavity and the role of saliva and its components. In a previous study, we reported a major role of the pancreatic juice AMPs in controlling the gut microbiome and the gut innate immunity in mice.²⁴ Indeed, pancreatic secretion has a dominant role in maintaining the gut microbiome in humans.²⁵ Considering the large amount of microbiota components and AMPs in the saliva that is swallowed daily, we examined the roles of the saliva and its component in the oral and gut immunity in the present study.

We used 2 mouse models exhibiting reduced salivary secretion to address the role of saliva in oral and gut immunity: the aquaporin 5 (Aqp5)^-/- mouse and the nonobese diabetic (NOD) mouse. Surprisingly, we were not able to see a clear role of the saliva and salivary secretion in the oral microbiome in both mouse models, even when salivation was restored in the NOD mouse by treatment with the cystic fibrosis transmembrane conductance regulator (CFTR) corrector C18 (a derivative of VX-890). However, the saliva clearly affected the gut microbiome. Further exploring the role of the saliva in regulating the gut microbiome revealed an intricate role in IBD. Impaired salivation in the Aqp5^-/- mouse resulted in delayed manifestation of IBD, but once damage started, it developed rapidly and was more catastrophic. Analysis of the salivary components indicated that the damage caused by the saliva was mainly caused by the fluid devoid of the microbiome. Fractionation followed by mass spectrometry analysis identified macrophage migration inhibitory factor (MIF) as the damaging and trefoil factor 2 (TFF2) as the protective component of the saliva fluid. MIF and TFF2 exacerbated and protected the IBD induced by dextran sodium sulphate (DSS). These findings reveal the important role of the salivary innate immunity in shaping the gut microbiome and suggest that reducing MIF and elevating TFF2 can be beneficial for ameliorating IBD.

Results

Role of the Saliva in the Gut Microbiome

The role of the innate oral immunity in IBD is not well-established or understood. The saliva is considered the first line of defense in the oral and gastrointestinal tract.²⁶ To examine the role of the saliva in the oral microbiome, we first focused on the Aqp5^-/- mouse. Figure 1A shows the lack of Aqp5 protein in the Aqp5^-/- mouse submandibular glands, and Figure 1B shows close to 85% reduction in salivation. Sequencing the microbiome (Figure 1C and deposited data) and partial least-squares discriminant analysis (PLS-DA) (Figure 1D) revealed that deletion of AQP5 accounts for 78% of the variance. However, analysis of alpha diversity (Figure 1E), species richness (Figure 1F), total number of species (Figure 1G), and Shannon index (Figure 1H) revealed no significant difference between the oral microbiomes of wild-type (WT) and Aqp5^-/- mice. Only the Simpson index (Figure 1I) suggested a small but higher diversity of the Aqp5^-/- mice oral microbiome. Analysis of phyla (Figure 1J) and genus (Figure 1K) suggested that variance is mediated by a reduction in Firmicutes, increased Proteobacteria, reduced Streptococcus, and increased Pasteurellaceae in the Aqp5^-/- mice oral microbiome.

Figure 1.

Effect of lack of salivation on the oral and cecal microbiome. (A) Expression of Aqp5 protein in WT and Aqp5^-/- submandibular glands obtained from 5 mice was assayed by Western blot. (B) Pilocarpine-stimulated saliva secretion in WT and Aqp5^-/- mice. Here and in all panels, each dot represents 1 mouse. (C) Heatmaps of the S16 analysis of the oral microbiome. Full list in accession number: PRJNA1260208. (D–I) Analysis of the WT (black and blue) or Aqp5^-/- mice (red) oral microbiomes: similarities (D); observed OTUs (E); Chao1 index (F); Ace index (G); Shannon index (H), and Simpson index (I). (J) Analysis of phyla in the oral microbiome of WT and Aqp5^-/- mice. (K) Analysis of genus in the oral microbiome of WT and Aqp5^-/- mice. (L) Heatmaps of the S16 analysis of the cecal microbiome. Full list in accession number: PRJNA1260208. (M–Q) Analysis of the WT (black and blue) or Aqp5^-/- mice (red) cecal microbiomes: similarities (M), Chao1 index (N), Shannon index (O); Simpson index (P); observed OTUs (Q). (R) Analysis of phyla in the cecal microbiome of WT and Aqp5^-/- mice. (S) Analysis of genus in the cecal microbiome of WT and Aqp5^-/- mice. (T) Level of intestinal neutrophils in WT and Aqp5^-/- mice.

Examining the role of reduced saliva on the cecal microbiome (Figure 1L) showed that deletion of Aqp5 accounted for an impressive 72.0% of the variance (Figure 1M), indicating that Aqp5 expression is a dominant factor in shaping the cecal microbiota. Significant change in alpha diversity (Figure 1N), Shannon index (Figure 1O), Simpson index (Figure 1P), and community richness (Figure 1Q), further distinguished between the WT and Aqp5^-/- mice microbiomes. The phyla (Figure 1R) and genus analysis (Figure 2) revealed changes in commensal bacteria (Figure 2A–E) and an increased Muribaculum (Figure 2F), which is potentially associated with IBD,²⁷ and increased Tannerella forsythia (Figure 2G), an oral bacterium commonly associated with severe periodontitis.²⁸ Indeed, analysis of inflammatory mediators in colonic tissue and serum indicated systemic inflammation of the Aqp5^-/- mouse with increased colonic level of neutrophils (Figure 1T), increased colonic interleukin (IL)1β, IL17, and IL10 (Figure 2H–M), and increased serum granulocyte-macrophage colony-stimulating factor (GM-CSF), IL1β, IL17, tumor necrosis factor alpha (TNFα), IL10, and interferon gamma (INFγ) (Figure 2N–S).

Figure 2.

Changes in the cecal microbiome and inflammatory mediators in mice with reduced saliva due to deletion of Aqp5 and treated with DSS. (A–G) Analysis of cecal genus spices in the microbiome of WT and Aqp5^-/- mice. (H–S) Measurement of inflammatory mediators in the colon (H–M) and the serum (N–S) of WT and Aqp5^-/- mice treated with DSS for 6 days. Each dot represents 1 mouse.

The unexpected finding of minimal effect of salivation on the oral microbiome and significant impact on the cecal microbiome prompted us to test this finding in another mouse model. We selected the NOD mice because we have shown these mice to exhibit reduced saliva due to mislocalization of the Cl^- channel CFTR, and that salivation is restored by treating the mice with CFTR correctors.²⁹ Therefore, we can test the effect of reduced and restored salivation on the oral and cecal microbiome. We used 11- to 13-week-old NOD mice before the onset of diabetes but with development of Sjögren’s disease and decreased salivation. Figure 3A confirms reduced salivation stimulated by pilocarpine and its restoration by C18. Most notably, Figure 3B shows that restoring CFTR and ductal function restored exocytosis by acinar cells, indicating that improved ductal function leads to improved acinar cell function.

Figure 3.

The oral and cecal microbiome of NOD mice treated with the CFTR corrector C18. (A) Pilocarpine-stimulated salivary secretion was measured in WT and NOD mice treated with DMSO (vehicle) or C18 for 1, 7, or 14 days, as indicated. (B) Isoproterenol-stimulated amylase secretion in WT and NOD mice treated with vehicle or C18. (C–E) Analysis of cecal microbiome similarities (C), observed OTUs (D), and genus species (E). (F–I) Analysis of the cecal microbiome of WT (black and blue) or NOD mice (red and green) treated with C18 (blue and green): similarities (F); observed OTUs (G); Ace index (H); Chao1 index (I). (J) Analysis of genus level in the cecal microbiome of all groups.

Analysis of the NOD mice oral microbiome (Figure 4A) and PLS-DA analysis (Figure 3C) showed that C18 treatment altered the microbial profile only in NOD mice, shifting it towards the WT microbiome. However, restoring salivation was not sufficient to restore the diversity parameters of the NOD mice microbiome (Figures 3D and 4C–F). The contribution of oral phyla Bacteroidetes to NOD microbiome diversity was unaffected by C18 treatment (Figure 4G–H). Genus species analysis (Figures 3E and 5A–E) showed limited changes in the NOD mice oral microbiome, with treatment by C18 having mixed effects. Analysis of the NOD mice cecal microbiome showed clear separation between the WT and NOD microbiomes in all parameters analyzed, except for the Shannon index, that were equalized by the C18 treatments (Figures 3F–I, 4I and J). Phylum analysis of the cecal microbiome showed changes in Bacteroidetes and Firmicutes only after C18 treatments (Figure 4K and L), whereas genus species showed a reduction in several commensal species (Figure 5F–O) and 2 pathogenic species (Eisenbergiella tayi and LLKBg) (Figure 5P and Q) in the NOD microbiome, all of which were equalized to the WT by the C18 treatment. Overall, our findings indicated a limited role of saliva in the oral and a significant role in shaping the gut microbiomes.

Figure 4.

Effect of correcting CFTR expression with the C18 correctors on diversity and phylum of the oral and cecal microbiome of NOD mice. (A and B) analysis of the oral (A) and cecal (B) microbiome of NOD mice that were also treated with C18 by S16 sequencing. Five mice in each group. Full list in accession number: PRJNA1260208. (C–F) Diversity and richness analysis of the oral microbiome. (G and H) Phylum analysis of the oral microbiome. (I and J) diversity and richness analysis of the cecal microbiome. (K and L) Phylum analysis of the cecal microbiome.

Figure 5.

The oral and cecal genus microbiome spies in NOD mice treated with C18. Analysis of oral (A–E) and cecal microbiome (F–Q) genus species in WT and NOD mice treated with the vehicle DMSO (black and red) or with C18 (blue and green).

The Saliva in IBD

Considering the role of the saliva in intestinal innate immunity and the profound role of the gut microbiome in IBD,^13,18 we examined the impact of reduced salivation in IBD. We used the colitis mouse model of feeding mice with 3% DSS. In Figure 6A, mice were fed with DSS for 15 days, and WT mice showed the typical reduction in weight that continued for the first 10 days of DSS feeding. Unexpectedly, the Aqp5^-/- mice were partially protected and showed delayed weight loss by 6 days. Their weight loss curve paralleled that of the WT mice after 6 days (Figure 6A). To confirm this surprising finding and to test the effect of mouse age on the impact of reduced saliva, we used the same procedure with 8- and 13-week-old mice (Figure 6C and D). Weight loss in Aqp5^-/- mice was delayed by 2 days at 8 weeks of age but then progressed more rapidly and catastrophically, with 5 of 7 Aqp5^-/- mice dead by day 13 (Figure 6C). At 13 weeks of age, the Aqp5^-/- mice weight loss was delayed until day 7, and after day 10 progressed rapidly, with 3 of 6 mice dead by day 12 (Figure 6D). Because saliva had 2 age-dependent effects in IBD, the experiments below were with mice between 18 and 20 weeks of age to minimize the effect of age on the roles of saliva.

Figure 6.

Effect of lack of saliva on development of DSS-induced colitis. (A) WT (black and green) and Aqp5^-/- mice (blue and red) were fed with 3% DSS (green and red) for 15 days. Mice were 22 weeks old, and the numbers in parenthesis indicate the number of mice in each group. The P values are the difference between WT and Aqp5^-/- mice fed with DSS. (B) Time course of development of disease activity of the mice in (A). (C and D) WT and AQP5^-/- mice age 8 (C) and 13 (D) weeks were treated with 3% DSS added to drinking water. The number of mice in each group is indicated in parenthesis, and the results are mean ± SEM. The P values are relative to WT mice fed with DSS. (E) Intestinal permeability expressed as fold change relative to the permeability measured before DSS treatment started and was taken as 1 for each group of mice. (F) Example images and averages of intestinal length. (G–I) Example images (G) and the level of ZO1 (H) and Occludin (I) measured by immunofluorescence staining. (J–L) Example H&E and PAS images (J) and analysis of cell proliferation (K) and goblet cell density (L). (M and N) Serum (M) and intestinal MPO (N) levels as reporters of neutrophils activity. (O and P) WT and AQP5^-/- mice age 18 weeks (5 mice in each group) were treated with 3% DSS and were gavaged with 50 μL PBS (Black, blue, gray and green) or with 50 μL whole saliva collected from 3 WT mice by stimulation with pilocarpine. The mice were gavaged on days 0, 3, and 6. Mice weight (O) was measured in 2-day intervals, and intestinal permeability (P) was measured on days 1, 3, 6, and 12. The results are mean ± SEM.

Analysis indicated that the delay in weight loss in Aqp5^-/- mice fed DSS correlated with the disease activity index (DAI), determined as a combination of diarrhea, blood in stool, and body weight (Figure 6B), and was delayed in the Aqp5^-/- mice. Most notably, measurement of intestinal permeability revealed that, after 8 days of DSS treatment, the intestinal permeability of WT mice increased significantly, whereas that of the Aqp5^-/- mice remained intact (Figure 6E). By days 12 and 15 of DSS treatment, the intestinal permeability (Figure 6E) and intestinal length (Figure 6F), were similar in both lines. Analysis of the key intestinal permeability barrier proteins ZO1 and Occludin³⁰ showed a similar reduction in these proteins after 15 days of DSS treatment (Figure 6G–I), but a difference was observed after 8 days of treatment (see below). Similarly, intestinal length reduction (Figure 6F), histological damage (Figure 6J), increased intestinal proliferation, and reduced intestinal goblet cells (Figure 6K and L) were the same in WT and Aqp5^-/- mice after 15 days of treatment with DSS. We also analyzed colonic and serum inflammatory mediators after 15 days. Figure 2 shows that DSS treatment increased all mediators tested in both the colon (Figure 2H–M) and serum (Figure 2N–S) of WT and Aqp5^-/- mice, with only minor differences. Serum at day 6 of DSS treatment of the same mice (Figure 6M) showed increased myeloperoxidase (MPO) in the WT mice, but not in the Aqp5^-/- mice, and it was similar by day 10. Intestinal MPO was still higher in WT than in Aqp5^-/- mice even after 15 days of DSS treatment (Figure 6N).

The Saliva Components Have a Different Role in Intestinal Health

To directly test the role of saliva in IBD, we collected saliva from WT mice and administered it by gavage to WT and Aqp5^-/- mice. We collected saliva by stimulating the glands with the parasympathetic agonist pilocarpine to collect fluid with AMPs and the oral microbiome and minimize secretion of mucins and amylase. The results in Figure 6O show that gavage of the whole saliva to WT and Aqp5^-/- mice had no significant impact on the mice’s weight. Intestinal permeability was lower in Aqp5^-/- mice than in WT mice treated with DSS for 6 days, and gavage of the whole saliva increased intestinal permeability of WT but not Aqp5^-/- mice treated with DSS for 6 days (Figure 6P).

To further evaluate the role of the saliva, as illustrated in Figure 7A, we collected saliva from WT mice and separated it into a particulate fraction containing the microbiome (referred to below as microbiome) and a fluid component that was passed through a 0.2-μm filter to collect the fluid with the AMPs (referred to below as fluid). We gavaged WT and Aqp5^-/- mice with the 2 components separately. Mice were treated with DSS for 6 days and allowed to recover for 9 days (Figure 7). The microbiome fraction had no obvious effect on body weight in WT mice (Figure 7B and C). Interestingly, the fluid fraction derived from WT saliva significantly reduced the loss of body weight and disease activity in WT mice (Figure 7B and C). By contrast, Figure 7D shows that the microbiome fraction of WT saliva slightly accelerated the DSS-induced weight loss in Aqp5^-/- mice but reduced the extent of the weight loss, apparently by hastening weight recovery. The fluid fraction markedly accelerated weight loss (Figure 7D) and disease activity in the Aqp5^-/- mice (Figure 7E).

Figure 7.

Effect of saliva fluid and microbiome fraction on development of DSS-induced colitis. (A) Illustration of the protocol and procedures used to collect the salivary microbiome and fluid fractions. (B and D), WT (B) and Aqp5^-/- mice (D) were fed with water (black) or 3% DSS (blue, green and red) for 6 days and allowed to recover for 9 days. The mice were gavaged with PBS (black), the microbiome fraction in PBS (green), or the salivary fluid fraction (red) on days 1, 3, and 6, as indicated. Mice were 18 to 22 weeks old, and the numbers in parenthesis indicate the number of mice in each group. The P values are relative to mice fed with DSS. (C and E) Time course of development of disease activity of the mice in (B and D), respectively. (F) Intestinal permeability expressed as fold change measured on day 8. (G) Example images and averages of intestinal length measured on day 8. (H–J) Example immunofluorescence and bright field images (H) and analysis of ZO1 (I) and Occludin (J) measured on day 8. (K–M) Example H&E and PAS images (K) and analysis of cell proliferation (L) and goblet cell density (M). (N and O) Serum (N) and intestinal MPO (O) levels.

We used several assays to determine the physiological changes responsible for the effects of the fluid and microbiome fractions. The changes in body weight correlated with the change in intestinal permeability (Figure 7F). No major difference in intestinal length was found, except for further reduction caused by the salivary fluid in the Aqp5^-/- mice intestine (Figure 7G). The change in intestinal permeability correlated with a reduction in the intestinal barrier proteins ZO-1 and Occludin (Figure 7H–J); with DSS-challenged mice, the fluid fraction partially restored the levels of both proteins in WT mice but reduced them in Aqp5^-/- mice. Histologically, the density of goblet cells was higher in Aqp5^-/- mice compared with WT mice, which was reduced by DSS treatment and was reduced further by the fluid fraction in Aqp5^-/- mice (Figure 7K and M). The fluid fraction also restored villus cell proliferation in DSS-treated WT intestines (Figure 7L). Finally, intestinal (Figure 7N) and serum MPO (Figure 7O) and various inflammatory mediators (Figure 8) showed that most mediators were significantly lower in the intestine and serum of the Aqp5^-/- mice relative to those measured in WT mice, although no consistent effects of the microbiome and fluid fractions were obvious due to high variability.

Figure 8.

Effect of the separated WT microbiome and salivary fluid on the serum and colonic inflammatory mediators in WT and Aqp5^-/- mice. Inflammatory mediators were measured at days 8 and 15 of WT (open columns) and of Aqp5^-/- mice (filled columns) treated with DSS (blue) and gavaged with WT microbiome fraction (green) of fluid fraction (red).

Considering the effect of the salivary fluid and microbiome fractions on intestinal integrity, we asked whether they affected the gut microbiome. As expected, the WT and Aqp5^-/- mice microbiomes differed significantly and were altered by the DSS treatment (Figure 9A–C), which also reduced the observed operational taxonomic units (OTUs) (Figure 9D) and Chao index (Figure 9E) but not the Shannon diversity index (Figure 9F) in the 2 mouse lines. Nevertheless, gavage of the microbiome and fluid fractions had a marginal effect on PLS-DA (Figure 9C), although they restored the observed OTUs (Figure 9D) and Chao1 index (Figure 9E) parameters in WT, but not Aqp5^-/- mice. Examining the phyla (Figure 9G) and genus (Figure 9H) showed multiple changes caused by the DSS treatment. However, as shown in the representative examples of Figure 9I–N, no clear pattern emerged when examining the effects of the microbiome and the fluid fractions. Therefore, with changes in intestinal permeability and not with the mice microbiomes, the foremost effect of the saliva is on maintaining the integrity of the mouse mucosal barrier.

Figure 9.

Effect of the WT microbiome and salivary fluid on the microbiome of WT and Aqp5^-/- mice. (A and B) Heatmaps of the WT and Aqp5^-/- mice microbiomes treated with DSS and the microbiome or fluid fractions. Full list in accession number: PRJNA1260208. (C–F) Analysis of the cecal microbiome of WT (open symbols and open columns) and of Aqp5^-/- mice (closed symbols and filled columns) treated with DSS (red, green and blue) and gavaged with WT microbiome (green) of fluid fraction (red). (G and H) Analysis of Phyla and Genus level of the various mice. (I–N) Analysis of selective cecal genus spices in the microbiome of WT and Aqp5^-/- mice treated with DSS and gavaged with the microbiome of fluid fraction obtained from WT mice, as indicated.

The Impact of the Native Microbiome

The marginal effect of the saliva on the mice microbiome (Figure 9) raised the question of whether the resident microbiome had any role in the effects of the saliva. To address this, mice were treated with a broad-spectrum antibiotic cocktail for 1 week prior to the start of DSS treatment. Figure 10A–C show that in WT mice, in the absence of the gastrointestinal (GI) microbiome, DSS was more toxic, causing significant death starting at day 8 of DSS treatment. The salivary fluid reduced weight loss, disease activity, and mortality. The salivary microbiome (without the salivary fluid) had a minimal effect on weight loss or disease activity caused by DSS but significantly hastened mortality that started at day 4 of DSS treatment (Figure 10C). Remarkably, depletion of the Aqp5^-/- mice resident GI microbiome markedly reduced the toxicity of DSS treatment, minimizing weight loss, disease activity, and mortality and prevented the effect of the salivary microbiome and fluid fractions in the Aqp5^-/- mice (Figure 10D–F). Notably, again, the impact of DSS and of the saliva components correlated very well with intestinal permeability and barrier function (Figure 10G).

Figure 10.

Effect of saliva fluid and microbiome fractions on development of DSS-induced colitis in the absence of the resident microbiome. (A and D) WT (A) and Aqp5^-/- mice (D) were treated with broad spectrum antibiotics for 1 week to clear the resident microbiome prior to the start of the experiment. The mice were fed with water (black) or 3% DSS (blue, green and red) for 6 days and allowed to recover for 6 days. The mice were gavaged with PBS (black), the microbiome fraction in PBS (green), or the salivary fluid fraction (red) on days 1, 3, and 6, as indicated. Mice were 18 to 22 weeks old, and the numbers in parenthesis indicate the number of mice in each group. The P values are relative to mice fed with DSS. (B and E) Time course of development of disease activity of the mice in (A and D), respectively. (C and F) Survival curves of WT (C) and Aqp5^-/- mice (F). P values between the indicated groups are shown in the Figures. (G) Intestinal permeability expressed as fold change measured on day 8.

Identifying Peptides Mediating the Beneficial and Toxic Effects of the Salivary Fluid

For an unbiased identification of saliva fluid peptides and proteins affecting the mice, we used mass spectrometry (MS) to analyze the protein in the saliva collected, as in Figure 7A. The salivary fluid was fractionated by filtration to collect proteins of molecular weight (MW) of 100 kDa, 50 kDa, 30 kDa, and 10 kDa (Figure 11A). These fractions were tested in WT (Figure 11B) and Aqp5^-/- mice (Figure 11C) treated with DSS. In WT mice, only the 50-kDa fraction significantly affected weight loss by preventing the recovery from DSS treatment. In Aqp5^-/- mice, the 100-kDa fraction slightly increased weight recovery, whereas the 30-kDa fraction slightly increased the rate of weight loss. The 50-kDa fraction greatly increased both the rate and extent of weight loss and noticeably slowed the weight recovery from DSS treatment. The proteins in each of the fractions were analyzed by MS, and their abundance in the fractions is shown in Figure 11D. The analysis showed that 23 proteins were found only in the 50-kDa fraction, and they are listed in Figure 11E. Of the 23 proteins, 2 stood out as involved in IBD, MIF^31,32 and TFF2.^33,34 MIF is a cytokine with immunosuppressive and proinflammatory functions³⁵ that has a role in tissue repair.³⁶ TFF2 is a member of the trefoil factor family (TFF) peptides that are secreted by exocrine cells and interact with mucins in epithelial surfaces.³⁷

Figure 11.

Effect of saliva fluid fractions, MIF, and TFF2 on development of DSS-induced colitis. (A) Coomassie-stained proteins in whole saliva fluid, (1) >100-kDa fraction, (2) >50-kDa fraction, (3) >30-kDa fraction, and (4) <30-kDa fraction. (B and C), WT (B) and Aqp5^-/- mice (C) were fed with water (black) or 3% DSS (blue, green, red and dark yellow) for 6 days and allowed to recover for 9 days. The mice were gavaged with the indicated fractions of the salivary fluid on days 1, 3, and 6, as indicated. Mice were 18 to 22 weeks old, and the numbers in parenthesis indicate the number of mice in each group. The P values are relative to mice fed with DSS. The results with DSS shown in dotted lines were taken from Figure 7 and are shown to better illustrate the effect of the fractions. (D) Venn chart of the proteins in each fraction. (E) A list of the proteins in the 50-KDa fraction. MIF and TFF2 are highlighted in red. (F and G) Protocol as in (B and C), except that the mice were gavaged with 50 μL of 0.01 μg/μL MIF (green) or 0.01 μg/μL TFF2 (red). (H and I) Example images (H) and averages of intestinal length (I) measured on day 8. (J–L) Example H&E images (J) and analysis of cell proliferation (K) and goblet cell density (L). (M) Intestinal permeability expressed as fold change measured on day 8. (N–P) Example of immunofluorescence and bright field images (N) and analysis of ZO1 (O) and Occludin (P) measured on day 8. (Q and R) Intestinal (Q) and serum (R) MPO levels.

The Effects of MIF and TFF2 in the IBD Mouse Model

Considering the physiological roles of MIF and TFF2 and their presence exclusively in the 50-kDa fraction, we tested their effect in the DSS model. Figure 11F and G shows that the application of recombinant MIF by gavage fully reproduces the effect of the fluid fraction in both WT and Aqp5^-/- mice. MIF inhibited the recovery of body weight in WT mice and accelerated the reduction of body weight in Aqp5^-/- mice. Interestingly, TFF2 was protective in both mouse lines, reducing the weight loss in WT mice (Figure 11F) and largely preventing the weight loss in Aqp5^-/- mice (Figure 11G). Accordingly, when assayed at 8 days of DSS treatment, MIF had no further effect on intestinal length in WT mice but reduced the intestinal length in Aqp5^-/- mice, whereas TFF2 prevented the reduction in intestinal length in both mouse lines (Figure 11H and I). Figure 11J and K shows that MIF increased cell proliferation, as reported by villus height, whereas TFF2 had no effect or reduced cell proliferation. MIF treatment increased tissue damage, which was estimated from goblet cell density in WT and Aqp5^-/- mice (Figure 11J and L). The effects of MIF and TFF2 correlated very well with their impact on intestinal permeability and the level of barrier proteins. Figure 11M–P showa that MIF caused a prominent increase in intestinal permeability and reduced the levels of ZO-1 and Occludin, whereas TFF2 reduced intestinal permeability in WT mice and increased the levels of ZO-1 and Occludin in WT and Aqp5^-/- mice treated with DSS. Finally, both MIF and TFF2 reduced the intestinal and serum neutrophil marker MPO in WT mice and had no major effect on the MPO, which is minimally affected by DSS in Aqp5^-/- mice (Figure 11Q–R).

To establish a direct role of salivary MIF and TFF2 in IBD, we determined the presence of the peptides in saliva and the effect of salivary fluid from which MIF and TFF2 were partially depleted (Figure 12). Using a titration curve with recombinant MIF showed that the salivary fluid contained about 27.2 pg/mL MIF, and it was reduced to 22.5 pg/mL in saliva collected from DSS-treated mice, and preincubation of the fluid with anti-MIF antibodies reduced the MIF to 17.0 pg/mL (Figure 12A). The TFF2 titration curve indicated that salivary fluid contained about 111.8 pg/mL TFF2, which was unchanged by DSS treatment and was reduced to 79.6 pg/mL by treatment with anti-TFF2 antibodies (Figure 12B). Despite the partial depletion of the salivary fluid MIF and TFF2, the antibody-treated salivary fluid showed reduced MIF and TFF2 effects (Figure 12C–F). Thus, although depletion of MIF had a minimal impact, depletion of TFF2 eliminated the protective effect of the salivary fluid in WT mice (compare Figure 12C and D and Figure 7B and C). In Aqp5^-/- mice, depletion of MIF converted a toxic effect of the salivary fluid to a protective effect, and depletion of TFF2 hastened weight loss and delayed recovery (compare Figure 12E and F and Figure 7D and E). These effects translated to reduced disease parameters by depletion of salivary MIF in both or one of the mice lines that included improved colon length (Figure 12G); reduced proliferation; increased goblet cells (Figure 12H and I); reduced colonic and serum MPO as indicative of inflammation (Figure 12J and K) and intestinal Occludin level, although not permeability and ZO-1 (Figure 12L–N). On the other hand, depletion of TFF2 worsened most of the same disease parameters (Figure 12G–N). The findings in Figure 12 further support the notion that salivary MIF and TFF2 mediate the harmful and protective effects of the salivary fluid, respectively.

Figure 12.

Scavenging saliva fluid MIF and TFF2 ameliorates effects of saliva fluid on development of DSS-induced colitis. (A and B) Standard curves (left) and the level of MIF (A) and TFF2 (B) in saliva from control (black) and DSS-treated mice (red) and saliva treated with the respective antibodies (green). (C and E) WT (C) and Aqp5^-/- mice (E) were fed with water (black) or 3% DSS (blue, green and red) for 6 days and allowed to recover for 6 days. The mice were gavaged with the salivary fluid treated with anti-MIF (green) and anti-TFF2 antibodies (red) on days 1, 3, and 6, as indicated. Mice were 18 to 22 weeks old, and the numbers in parenthesis indicate the number of mice in each group. The P values are relative to mice fed with DSS. The results with DSS shown in dotted lines were taken from Figure 7 and are shown to better illustrate the effect of MIF and TFF2 depletion. (D and F) Time course of development of disease activity of the mice in (C and E), respectively. (G) Intestinal length measured on day 8. (H and I) Analysis of cell proliferation (left) and goblet cell density (right). (J and K) Intestinal (J) and serum (K) MPO levels. (L) Intestinal permeability expressed as fold change measured on day 8. (M and N) Analysis of ZO1 (M) and Occludin (N) measured on day 8.

Discussion

The overall findings of the current studies allowed several conclusions. It is widely assumed that saliva plays a significant role in controlling the oral microbiome both in healthy and disease states.^6,7,38 Moreover, impaired saliva secretion is associated with increased incidence of oral infection.³⁹ However, in the 2 mouse lines with reduced salivary secretion used in the present studies, changes in the oral microbiome were minimal. Moreover, restoring salivary secretion in NOD mice failed to affect the oral microbiome. The simplest explanation for the lack of effect of the reduced saliva on the oral microbiome is that the residual saliva is sufficient to maintain an intact oral microbiome. The reduced saliva in oral diseases such as Sjögren’s disease and radiation therapy that do affect the oral microbiome^40,41 can be secondary to the inflammation and tissue damage typical of these conditions, which have a more significant impact on the oral microbiome than the reduced saliva alone, as studied here.

Nonetheless, the microbiome of mice with reduced salivation does have a role in IBD. It was surprising to find out that the weight loss caused by DSS was strongly blunted by depletion of the resident Aqp5^-/- mice microbiome. DSS is believed to cause colitis by hyperosmotic stress and injury to colonic epithelial cells and disruption of the epithelial barrier.⁴² It is expected that such physical tissue injury should be independent of biological functions. Therefore, the most likely explanation for the reduced toxicity of DSS in Aqp5^-/- mice is a tighter and improved intestinal barrier in the Aqp5^-/- mice. Indeed, overall, we compared intestinal permeability in 35 WT and 35 Aqp5^-/- mice in the same experiments (Figure 6, Figure 7 and Figure 10, Figure 11, Figure 12), and in these experiments, the intestinal permeability of Aqp5^-/- mice was not or only minimally reduced by the DSS treatment, indicating a superior intestinal barrier integrity. It is possible that depletion of the toxic Aqp5^-/- mice resident GI microbiome resulted in an even tighter intestinal barrier, sufficient to reduce the harmful effects of DSS alone and in combination with the salivary fluid fraction. The improved barrier function may also improve food intake, which is reduced in DSS-induced colitis.⁴³ The lack of effect of the salivary fluid fraction in these mice can result from the tighter intestinal barrier, although it may also indicate that the fluid fraction needs to interact with the resident Aqp5^-/- mice microbiome to exert toxicity.

Several additional conclusions are offered by the overall findings. First is the role of the saliva in the gut microbiome in the absence of disease, which has not been examined extensively before. Such a relationship was found in the present studies, in which reduced salivation affected all parameters of the cecal microbiome, relatedness, richness, and diversity. Interestingly, these parameters were corrected by restoring salivation in the 11- to 13-week-old NOD mice (Figure 3F–I) before the onset of diabetes and inflammation in these mice.^44,45 These findings suggest that the innate oral immunity contributes to the homeostasis of the gut microbiome even in the absence of disease. In a previous study, we showed that the pancreatic innate immunity affects the homeostasis of the gut microbiome primarily by secreting antimicrobials.²⁴ Oral secretion of antimicrobials appears to have similar roles.

The mutual relationship between the oral cavity and the gut microbiome in IBD is well-established.^10,20,23 IBD is associated with periodontal diseases, and periodontitis exacerbates IBD.^2,46, 47, 48 The adaptive oral immunity has a significant role in IBD, responding to oral bacteria present in the gut.²³ However, it is not clear whether and how the saliva participates in the pathogenesis of IBD. The present studies describe a new relationship between oral and gut health, with implications for IBD. The saliva appears to have different effects in different stages of IBD. In the first stage of the disease, the lack of saliva substantially delayed development of IBD, whereas at the second stage, IBD developed more rapidly in the continued presence of the insult (DSS feeding), or the recovery from IBD was delayed. The protective and noxious effects of the saliva were clearly associated with improved and impaired intestinal barrier, respectively, that correlated very closely with the expression of the barrier proteins examined, ZO1 and Occludin. It should be noted that IBD is a multifaceted disease involving malfunction of several biological pathways, including major aspects of the immune system, epithelial defense, and transcriptional regulation. Indeed, there are numerous IBD models, including other chemically induced, cell-transfer, congenial mutant, and genetically engineered IBD models.⁴⁹ Whether and how the saliva impacts these forms of IBD is not known at present, but our findings suggest that this should be examined to fully appreciate the role of the saliva in IBD.

The beneficial and noxious effects of the saliva were mainly in the fluid fraction, with minimal contribution of the oral microbiome to the effects of the saliva on IBD (Figure 7). We interpret these findings to suggest that the salivary antimicrobials and innate oral immunity have a major role in IBD in the absence of oral cavity disease and inflammation. This is different from the effect of the oral cavity on IBD during periodontitis, when the oral microbiome and adaptive oral immunity are the major contributors to IBD.^2,23 MS analysis identified 2 of the peptides that mediate the effects of the saliva, TFF2 and MIF. Beneficial and curative roles of TFF2 in IBD have been reported in several studies. Lack of TFF2 caused a more severe IBD in mice, and secretion of TFF2 is essential for maintaining the intestinal barrier.³³ Supplementation of TFF2 reduced tissue injury and lymphocyte recruitment in mouse models of IBD.^34,50 Although the major source of TFF2 is the gastric mucosa,^37,51 salivary glands secrete TFF peptides, with TFF3 as the major peptide over TFF1 and TFF2.⁵² Analysis of the active fraction in the mouse salivary fluid detected only TFF2, and the effects of supplementing the mice with recombinant TFF2 and partially depleting the salivary fluid of TFF2 were consistent with TFF2 as the major protective factor in the saliva.

The toxic component of the salivary fluid appears to be mediated by the pleiotropic cytokine MIF. In the gut, deletion of MIF damages the intestinal barrier by increasing intestinal permeability.⁵³ CD74 is the MIF receptor⁵⁴; it is increased in patients with IBD⁵⁵ and promotes mucosal healing.⁵⁶ However, genome-wide association studies (GWAS) and polymorphism analysis suggested that increased MIF is associated with Crohn’s disease and ulcerative colitis,³¹ raising the possibility that MIF may have adverse roles in disease states. Moreover, transgenic mice overexpressing MIF in the colon are susceptible to colitis.⁵⁷ Effects of MIF are mediated by binding to the MIF receptor CD74. The level of CD74 is very low in the mouse intestine and is markedly increased in IBD.⁵⁶ Therefore, it is possible that expression of intestinal CD74 is delayed in the Aqp5^-/- mice. MIF has both extrinsic and intrinsic effects mediated by different signaling pathways that can be beneficial, promoting cell and tissue renewal, or harmful, promoting inflammatory mediators and inflammation, causing apoptosis, necrosis, and cell death.³⁶ Salivary MIF level responds to the induction of IBD and is reduced within 8 days of treatment with DSS (Figure 12A), perhaps to limit its toxicity.

The overall findings with TFF2 and MIF suggest that control of their level in IBD may be of clinical relevance. TFF2 was used in a small clinical trial to treat SARS-CoV-2 infection⁵⁸ with a potential for improved patient recovery. MIF inhibitors are promising in the treatment of several cancers. BAX69 was reasonably tolerated in phase I advanced solid tumor trials,⁵⁹ and the MIF inhibitor MN-166 was used in several clinical trials.^60,61 The TFF2 and MIF activity modifiers can be of use in other inflammatory diseases in association with IBD, such as Sjögren’s disease and pancreatitis, in which the level of MIF is increased.^62,63 In addition, controlling TFF2 and MIF can be relevant to subgroups of patients with Sjögren’s disease,⁶⁴ treated with radiation therapy,⁶⁵ and treated with immune checkpoint inhibitor,⁶⁶ which severely reduce salivation.

Methods

Collection of Saliva and Cecal Microbiome and Fluid and Microbiome Fractions

All animal protocols have been reviewed and approved by the National Institutes of Health, National Institute of Dental and Craniofacial Research (NIDCR) animal use committee (protocol ASP 22-1090) and comply strictly with all NIH guidelines. C57BL/6, and NOD mice (Jackson Laboratory), and AQP5^-/- mice⁶⁷ were maintained in sterile control environment at the NIDCR mouse facility. Mice of both sexes were used in these experiments, ages 8 to 22 weeks. Mice were anesthetized with intraperitoneal injection of ketamine (60 mg/kg body weight [BW]), and xylazine (8 mg/kg BW) and pilocarpine (0.5 mg/kg BW) were injected at the neck to stimulate saliva secretion. Saliva was collected from 10 to 15 WT BL6 mice for 20 minutes. To obtain saliva fluid and microbiome fractions, saliva was centrifuged for 10 minutes at 12,000 rpm. The pellet was redissolved in phosphate-buffered saline (PBS) as the microbiome fraction. The supernatant was collected and passed through 0.22-uM filters to remove any leftover bacteria/fungus from the supernatant and is referred to as saliva fluid fraction. For the collection of cecal content, surgical instruments were sterilized by autoclaving, and procedures were conducted within a laminar flow hood. The GI tract was extracted by grasping the colon with forceps, severing it with scissors, and carefully moving it onto a surgical table set up with clean towels. Fecal material from the cecum was then gently extracted into a 2-mL cryovial and immediately flash-frozen in liquid nitrogen. These samples were stored at −80°C until further analysis.

Saliva Fluid Protein Fractionation

Saliva samples were fractionated using Amicon Ultra 0.5-mL ultracentrifugation filters (Millipore, UFC5100, UFC5050, UFC5030, UFC5010) with sequential MW cutoffs of 100, 50, 30, and 10 kDa. The filtration process was performed progressively from higher to lower MW cutoffs. After passing through each filter, part of the flow through was collected and stored, and the remaining sample was subjected to the next lower MW filter. The fractions collected were designated as ≤100 (100–50 kDa), ≤50 kDa (50–30 kDa), ≤30 kDa (30–10 kDa), and ≤10 kDa (10 and lower kDa).

Trypsin Digestion and Protein Identification by MS

Aliquots of salivary protein fractions were reduced and alkylated with 1 mM Tris (2-carboxyethyl) phosphine (TCEP)/ 5 mM chloroacetic acid (CAA), and incubated for 30 minutes at 50°C, and the reduced/alkylated sample was transferred to 10-kDa filters. The samples were washed twice with 200 μM tetraethylammonium bicarbonate buffer (TEAB) by centrifuge at 900 g for 1 minute, and 2 μg of MS grade gold trypsin (Promega, Cat no. V5280) were added. The samples were digested overnight at 37°C. The digests were washed with 50 μL of 0.1% trifluoroacetic acid (TFA) by centrifuge at 900 g for 1 minute and passed through C18 columns for peptide cleanup. The eluted tryptic peptides were dried under vacuum.

Liquid Chromatography-Tandem MS Analysis

Nano liquid chromatography-tandem MS (LC-MS/MS) analysis of tryptic peptides was carried out with a Thermo Scientific Fusion Lumos tribrid mass spectrometer interfaced to a UltiMate3000 RSLCnano HPLC system. For each analysis, ∼1 μg of the tryptic digest was loaded and desalted in an Acclaim PepMap 100 trapping column (75 μm × 2 cm) at 5 μL/min for 4 minutes. Peptides were then eluted into an AuroraXT column (75 μm × 250 mm; Ionoptiks) and chromatographically separated using a binary solvent system consisting of A: 0.1% formic acid and 2.5% acetonitrile, and B: 0.1% formic acid and 80% acetonitrile at a flow rate of 300 nL/min. A gradient was run from 1% B to 42% B over 90 minutes, followed by a 5-minute wash step with 80% B and 10-minute equilibration at 1% B before the next sample was injected. Precursor masses were detected in the Orbitrap at R = 120,000 (m/z 200). Fragment masses were detected in the Orbitrap at R = 30,000 (m/z 200). Data-dependent MS-MS was carried with top speed setting; cycle time was 2 seconds with dynamic exclusion of 20 seconds.

Data Analysis

Protein identification was carried out using Proteome Discoverer software package (v 3.1, Thermo Scientific). Raw data was searched against a mouse proteome database from Uniprot along with a common contaminant protein database with Sequest HT search engine. C alkylation was set as fixed modification, and NQ deamidation, M oxidation, were set as variable modifications.

Recombinant Peptide and Application by Gavage

Two colitis-associated recombinant peptides were obtained, MIF (Abcam, Cat no. AB219128) and TFF2 (ProteinTech Cat no. 315-30), and were diluted at 500 ng/100 μL in PBS. WT and AQP5^-/- mice were gavaged with MIF and TFF2 (0.5 μg/mice/gavage) on the indicated days during colitis induction. Because gastric mucosa synthetizes and secretes both MIF^68,69 and TFF2,⁵¹ thus they are likely to survive the gastric environment. MIF or TFF2 protein level in saliva collected from WT mice before and after DSS treatment was measured by commercial sandwich enzyme-linked immunosorbent assay (ELISA) kits (MyBioSource, MIF Cat no. MBS164186, TFF2 MBS763147) according to the kit protocol. The depletion of MIF or TFF2 protein from the saliva fluid was by incubation of the fluid for overnight at 4°C with anti-MIF (Cell Signaling, Cat no.88186S) or anti-TFF2 (ProteinTech, Cat no. 13681-1-AP) neutralizing antibodies at a dilution of 5 μg/mL for both peptides. The extent of depletion was assessed with ELISA assays.

Induction of Colitis and Sample Collection

DSS was obtained from MP Biomedicals (Cat no 160110). For 15 days of DSS treatment, following a 1-week acclimation period, both WT and Aqp5^-/- mice were randomly divided into a control group, receiving PBS in drinking water, and a DSS group, receiving 3% DSS in drinking water. For 6 days of DSS treatment, following a 1-week acclimation period, both WT and Aqp5^-/- mice were randomly divided into: a control group receiving PBS by gavage; 3% DSS receiving PBS by gavage; 3% DSS receiving salivary microbiome fraction by gavage; and 3% DSS receiving salivary fluid fraction by gavage. Mice had free access to a standard mouse chow and either tap water (control group) or 3% DSS in tap water. PBS, salivary microbiome, or salivary fluid in a volume of 50 μL were administered intragastrically by gavage.

Throughout the experiment, BW, presence of blood in stool, and degree of diarrhea were monitored to assess the DAI. Mice were euthanized on day 8 or at the end of the experiments, and colonic weight and length were measured. Colonic tissue, serum, cecal contents, and fecal samples were stored at −80°C for further analysis. Additionally, a 0.5-cm section of the colon was fixed in 4% paraformaldehyde for histological analysis. Fresh cecal content from euthanized control and DSS-treated groups was aseptically collected on day 8, around the peak of the disease and at the end of the experiments.

In Vivo Intestinal Permeability

At the indicated days, mice were gavaged with 50 uL of 150 μg/mL fluorescein isothiocyante (FITC)-dextran 4KDa (Sigma-Aldrich, Cat. FD4), and blood was collected 3 hours after gavage. About 0.3 mL of blood was collected by retro-orbital bleeding procedure. A heparinized hematocrit capillary tube was inserted into the medial canthus of anesthetized mouse eye socket and gently rotated to puncture the orbital sinus to collect blood via capillary action. Serum was separated from whole blood by centrifugation at 5000 g for 10 minutes. FITC-dextran fluorescence in serum was measured at 520 nm using Flexi station spectrofluorometer (Molecular Device, BMG LABTECH; FLUOstar Omega; Serial # 415-1214).

Histological Analysis

Tissues were collected, fixed, and stored at −80^oC until processing. Subsequently, tissues from all treatment groups were processed at the same time, with the same antibody stokes and dilution and by identical procedures so that results can be compared. Paraffin sections of colon tissue were stained with hematoxylin and eosin (H&E). In brief, sections were deparaffinized and incubated with Mayer’s hematoxylin for 2 minutes and washed in tap water for 8 minutes, followed by eosin staining for 4 minutes. The sections were dehydrated with alcohol-xylene and were mounted in paramount mounting media (Permount Mounting Medium, Cat no: 17986, Electron Microscopy Science). Images were captured with a S60 NanoZoomer Digital Scanner microscope (Hamamatsu), and histological scores were assessed based on the crypt damage, the extent of inflammatory infiltration, and the state of edema, as previously described.⁷⁰ Goblet cells were analyzed by Alcian blue/periodic acid-Schiff (AB-PAS) staining and counting the number of colonic goblet cells per villus. In brief, sections were successively incubated with acetic acid (3%) for 2 minutes, Alcian blue (pH 2.5) for 15 minutes, and washed in tap water for 2 minutes. The sections were submerged in 1% periodic acid solution for 5 minutes, washed twice in distilled water, dipped in Schiff’s solution (pH 9.8) for 15 minutes, and finally washed in tap water for 2 minutes. Nuclei were counter-stained (hematoxylin), and the sections were imaged.

MPO and Cytokine Assay

The MPO level in the cleared serum and colonic tissue lysates was measured using Abcam MPO assay kit (cat no. ab273334) according to the kit protocol. Cytokines were measured with a customized mouse multiplex cytokine assay kit (R &D System, Cat no. LXSAMSM). The frozen colonic tissues were thawed in ice and dissolved in a protease inhibitor cocktail (Cat no.1803670A, TaKaRa) containing 1× RIPA buffer at 4°C, homogenized, and pulse sonicated (10 seconds, 3–4 times). After 20 minutes on ice, the lysates were centrifuged (10,000 g, 10 minutes at 4°C), and the supernatants were collected for further analysis.

Immunofluorescence Staining

Colonic tissues were snap-frozen in OCT (Tissue-Tek, Cat no. 4583) and sectioned. Sections were fixed with methanol at −20°C for 10 minutes and permeabilized and blocked with 1% Saponin and 0.5% Triton-X in 5% normal donkey serum in PBS for 1 hour. Antibodies for ZO-1 (Cat no. SAB1306492, Sigma) and Occludin (Cat no. SAB4200593, Sigma) were incubated at a 1:100 dilution in blocking media overnight at 4°C. After 5 washes with 0.1% Saponin and 0.05% Triton-X in PBS (washing buffer), the slices were incubated with Alexa Fluor 488 conjugated Donkey anti-rabbit IgG (Jackson Immuno-Research, Cat AB_2313584) for 60 minutes and washed 5 times with washing buffer. The sections were mounted using Fluoromount-G Mounting Medium with 4′,6-diamidino-2-phenylindole (DAPI; Themo Scientific, Cat no: 00-4959-52), and imaged with a 60× oil immersion objective on an Olympus Fluoview Confocal microscope.

Western Blot

Western blot was performed as described previously.²⁹ In brief, the submandibular glands were harvested and finely minced. Membranes were isolated by homogenization in a buffer containing 250 mM sucrose, 10 mM HEPES, pH 7.4, and protease inhibitor cocktail (Roche). The homogenate was centrifuged at 500 g for 10 minutes at 4°C, and the supernatant was collected and centrifuged at 20,000 g for 20 minutes at 4°C. The pellet was extracted with RIPA lysis buffer for 30 minutes at 0°C, cleared by centrifugation at 20,000 g for 20 minutes at 4°C, and the supernatants were separated by sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Aqp5 was detected by 1:100 dilution of anti-Aqp5 antibodies (Alomone Labs, AQP-005Cat).

DNA Extraction for Microbiome Analysis

DNA extraction was performed using QIAamp DNA microbiome kit (Qiagen Cat no. 51604) according to the manufacturer’s instructions. Briefly, samples were mixed with Buffer AHL, incubated for 30 minutes at room temperature with rotation, and centrifuged for 10 minutes at 10,000 g. The supernatant was discarded, and the pellet was resuspended with RDD buffer and benzonase, followed by stirring at 600 rpm at 37°C for 30 minutes. Proteinase K was added to each sample and incubated for a further 30 minutes at 56°C. Samples were transferred to pathogen lysis tubes, ATL buffer was added, and the tubes were centrifuged for 1 minute at 10,000 g. The supernatants were transferred into new microcentrifuge tubes, and proteinase K was added and incubated at 56°C for 30 minutes while stirring at 600 rpm. APL2 buffer was added, content was mixed and incubated at 70°C for 10 minutes, and ethanol was added to the mixture. The mixtures were applied to the QIAamp UCP mini spin columns and centrifuged at 6000 g for 1 minute to discard the flow-through. The columns were washed twice with AW1 and AW2 buffers, and the DNA was eluted from the membrane with AVE buffer.

16S rDNA Amplification

Polymerase chain reaction (PCR) was performed to assess the rRNA operon copy number per bacterial genome. The PCR was carried out in a total volume of 25 μL containing the following: 2 μL (900nM) forward primer (5'-TCCTACGGGAGGCAGCAGT-3'), 2 μL reverse primer (5'-GGACTACCA GGGTATCTAATCTT-3'), 12 μL 2X Hot StarTaq master mix (Qiagen, Cat no./ID. 203443), 7 μL dH2O, and 2 μL (200 μg) of genomic DNA. Amplification of the 16S was performed at 50°C for 2 minutes, and 10 minutes at 95°C, followed by 40 cycles of 95°C for 15 seconds, 60°C for 60 seconds, and held at 37°C. The quantity and concentration of the PCR product were analyzed using 2% agarose gel electrophoresis.

16S rDNA Sequencing

The 16S rRNA coding sequence was used to identify the bacterial community by amplifying the V3-V4 hypervariable region. The following procedures were used.

First PCR Amplification (Gene-specific Amplification)

In sterile thin-walled tubes, the following was mixed in a final volume of 25 μL: template DNA 10.5 μL (100 ng); 12.5 μL (0.2 μM) or forward and reverse primers, respectively, 5'-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGTGCCAGCMGCCGCGGTAA-3′ and GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGGACTACHVGGGTWTCTAAT-3′. The reaction cycles were 1 cycle at 98°C for 30 seconds, 10 cycles of 98°C for 10 seconds + 60°C for 30 seconds/1 degree decrease/cycle + 72°C for 30 seconds + 98°C for 30 seconds + 50°C for 30 seconds, 7 cycles of 72°C for 30 seconds + 72°C for 7 minutes, and finally samples were kept at 4°C. Seventeen cycles of amplification were carried out. PCR products were purified using Ampure XP beads (Beckman coulter Cat# A63881) at a ratio of 1:1. The final product was eluted with 50 μL EB buffer (Qiagen, Cat#19086).

Second PCR Amplification (Illumina Bar Code Amplification)

In sterile thin-walled tubes, the mixtures below in a final volume of 25 μL were used. DNA (Purified PCR product of the first) in 8.5 μL, 2× Phusion High Fidelity in 12.5 μL, and 2.5 μM Index 1 (N7xx) and 2.5 μM index 2 (S5XX), each in 2 μL. The PCR amplification cycles were 98°C for 30 seconds 1 cycle, 98°C for 10 seconds + 55°C for 30 seconds 8 cycles and 72°C for 30 seconds + 72°C for 7 minutes for 8 cycles, and then samples were kept at 4°C.

Nine cycles of amplification were carried out, and PCR products were purified using Ampure XP beads (Beckman coulter; Cat# A63881) at a ratio of 1:1. The final product was eluted with 50 μL EB buffer (Qiagen, Cat#19086). The concentration of the library was determined by using the KAPA Library quantification kit (Kapa Biosystems; cat#KK4873,) on Quantstudio 6 Flex (Thermofisher). The same concentration of samples was combined in a sterile microcentrifuge tube. The pooled library was quantified with Qubit, and the pooled library was sequenced using the Miseq machine with a 10% phix spike in.

Analysis of Cecal and Oral Microbiota

Sequence analysis was processed by Quantitative Insights into Microbial Ecology (QIIME, Version 2, https://qiime2.org) analysis pipeline for taxonomic classification and alpha and beta diversity. Read count data were exported as BIOM tables, analyzed, and visualized using the “Phyloseq” package (Version 1.14) and distributed as part of the Bioconductor (version 2.30 https://bioconductor.org/packages/release/bioc/html/phyloseq.htm) repository for the R statistical programming language (Version 3.2.2 https://www.r-project.org).

Statistical Analysis

Data were analyzed using SPSS (Version 24) (IBM Corp) and GraphPad Prism (Version 14).