Western Diet Induces Paneth Cell Defects through Microbiome Alterations and Farnesoid X Receptor and Type I Interferon Activation

Summary

Intestinal Paneth cells modulate innate immunity and infection. In Crohn’s disease, genetic mutations together with environmental triggers can disable Paneth cell function. Here, we find that a Western diet (WD) similarly leads to Paneth cell dysfunction through mechanisms dependent on the microbiome and farnesoid X receptor (FXR) and type I interferon (IFN) signaling. Analysis of multiple human cohorts suggest obesity is associated with Paneth cell dysfunction. In mouse models, consumption of a WD for as little as four weeks led to Paneth cell dysfunction. WD consumption in conjunction with Clostridium spp. increased the secondary bile acid deoxycholic acid levels in the ileum, which in turn inhibited Paneth cell function. The process required excess signaling of both FXR and IFN within intestinal epithelial cells. Our findings provide a mechanistic link between poor diet and inhibition of gut innate immunity, and uncover an effect of FXR activation in gut inflammation.

Graphical Abstract

eTOC blurb:

Small intestinal Paneth cells are gatekeepers of gut innate immunity. Liu et. al. identified a link between consumption of high fat, high sugar diet (Western diet; WD) and Paneth cell dysfunction. The mechanisms involve WD-associated microbiota conversion of bile acids, which activate both FXR and type I interferon pathways.

Introduction

Paneth cells are critical for gut innate immunity by producing and secreting antimicrobial peptides (Bevins and Salzman, 2011; Cederlund et al., 2011; Ouellette, 2010; Vaishnava et al., 2008) and providing a niche for intestinal stem cells (Sato et al., 2011). We and others have shown that Paneth cell defects/dysfunction could be the origin of gut inflammation in models of inflammatory bowel disease (IBD) (Kaser et al., 2008; Liu et al., 2017; Wehkamp et al., 2005). Paneth cell dysfunction leads to mucosal dysbiosis both in patients with Crohn’s disease (CD; a major subtype of IBD) and in mouse models (Gulati et al., 2012; Liu et al., 2016; Salzman et al., 2010). In addition to IBD and infection, Paneth cell homeostasis is also critical in the pathogenesis of graft-versus-host disease (Zhao et al., 2018), metabolism dysregulation (Larsen et al., 2019), and cirrhosis (Teltschik et al., 2012). Thus, optimal Paneth cell function is a critical anti-inflammatory mechanism in the intestine, and its dysfunction could lead to increased susceptibility to infection and IBD (Adolph et al., 2013; Burger et al., 2018).

While Paneth cell function can be indirectly evaluated by response to pathogen infections in mouse models (Bel et al., 2017; Burger et al., 2018), measuring this parameter in humans and laboratory mice is most easily accomplished by quantifying the distribution pattern of cytoplasmic antimicrobial proteins in Paneth cells on well-oriented tissue sections (Cadwell et al., 2008; Cadwell et al., 2010; Khaloian et al., 2020; Liu et al., 2016; Liu et al., 2018; Liu et al., 2017; VanDussen et al., 2014; Wu et al., 2015). Humans and mice with an elevated number of abnormal Paneth cells by this metric (abnormal Paneth cell phenotype; defined as <80% of the total Paneth cells contain normal cytoplasmic granule morphology) have reduced Paneth cell numbers, repressed mRNA expression for genes that control cell metabolism (Liu et al., 2018; VanDussen et al., 2014), and more aggressive disease course in CD (Khaloian et al., 2020; Liu et al., 2018; Liu et al., 2017; VanDussen et al., 2014).

An abnormal Paneth cell phenotype can be driven by either a genetic mutation or gene-environment interactions (Cadwell et al., 2008; Cadwell et al., 2010; Kaser et al., 2008; Liu et al., 2013; Liu et al., 2018). While we previously demonstrated that murine norovirus or cigarette smoking could trigger Paneth cell dysfunction in a host with Atg16l1 deficiency (Cadwell, 2010; Liu et al., 2018), the role of persistent viral infection in CD pathogenesis remains elusive (Norman et al., 2015), and abnormal Paneth cell phenotypes can also be found in CD patients lacking ATG16L1 T300A risk alleles or a cigarette smoking history (Liu et al., 2018). Our study of a pediatric CD cohort (of which none were active smokers) showed that an abnormal Paneth cell phenotype is present in ~50% of CD subjects (Liu et al., 2016), consistent with a previous study showing diminished α-defensins in pediatric CD (Perminow et al., 2010). Accordingly, we hypothesized that additional environmental factors could also trigger Paneth cell dysfunction. Previous studies suggest micronutrient candidates such as zinc, of which deficiency could lead to Paneth cell defects (Bohane et al., 1977; Elmes and Jones, 1980). However, no robust data exists to establish such connections in CD patients. Identifying additional environmental factors that could trigger Paneth cell defects could help develop intervention strategies.

Obesity has become an epidemic over the past 3 decades, with current prevalence of overweight/obese at 39% among adults and 18% among children in the USA (Hales et al., 2018). Given the link between obesity and the prevalence of IBD (Piovani et al., 2020), we hypothesized that diet-induced obesity is a potent environmental trigger for Paneth cell dysfunction. Herein we show that overweight and obese subjects are more prone to developing Paneth cell defects. Mice consuming an obesogenic diet (aka “western diet [WD]”) with high fat, high sugar content) (Naja et al., 2015) also develop Paneth cell defects via the secondary bile acid deoxycholic acid (DCA). Of note, the experimental diet used in this study included fat content (40%) closer to the average US adults (~34%) (Shan et al., 2019) then other preclinical studies. We show that WD-associated Paneth cell defects are dependent on enhanced signaling of two pathways with the intestinal epithelium: farnesoid X receptor (FXR) and type I interferon (IFN). Both pathways are required, as inhibition of either FXR or type I IFN signaling prevents WD-induced Paneth cell defects. Specifically, our findings have implications for how FXR activation may elicit previously unknown undesirable effects in the context of gut infection and inflammation.

Results

Overweight and obese individuals are prone to Paneth cell defects

While we can mechanistically explain the basis of abnormal Paneth cells in certain cohorts (i.e. ATG16L1 CD susceptibility alleles and smoking), for others (such as children with CD), the causal links, especially the environmental components, have remained elusive. To address this issue, we used our bank of Paneth cell-phenotyped individuals (now at 930 individuals from multiple disease states) to search for additional features that correlated with humans having abnormal Paneth cell phenotypes. For one cohort, North American adult non-IBD subjects (demographics described in Table S1), we found a correlation between the percentages of abnormal Paneth cells and elevated body mass index (BMI; ≥25) (Fig. 1A–B), and the number of Paneth cells per crypt also correlated with elevated BMI (Fig. S1A), in concordance with previous studies (Hodin et al., 2011; Tomas et al., 2016). Paneth cells appeared to be the primary intestinal epithelial cell type affected by elevated BMI, as the abundance of differentiated epithelial lineages, villous height and crypt depth did not correlate with BMI (Fig. 1C, Fig. S1B–S1E). The correlation of BMI to abnormal Paneth cell morphology also occurred in CD cohorts; CD subjects with BMI ≥25 had fewer normal Paneth cells (Fig. S2A). Taken together, we hypothesized that obesity can cause Paneth cell dysfunction.

Figure 1. Obesity is associated with Paneth cell defects in humans and mice.

(A) In non-IBD patients, those with BMI≥25 (n=57) showed reduced percentage of normal Paneth cells compared to those with BMI<25 (n=34) (P=0.0129). Representative images of Paneth cells from patients with BMI<25 and BMI≥25 stained with HD5 immunofluorescence (green) are shown in (B). Scale bars: 10μm. (C) There was no difference in goblet cell density (P=0.6256) between the 2 groups. Compared to standard diet (SD)-fed mice (n=30), mice fed with Western diet (WD; n=41) for 8 weeks also resulted in (D) reduced percentage of normal Paneth cells (P<0.0001), and no significant changes in (E) goblet cell density/villus (P=0.1767), (F) neuroendocrine cells/villus (P=0.1193), or (G) crypt base proliferation (P=0.8982). (H) A time course study showed that 4 weeks of WD was sufficient to trigger Paneth cell defects (P<0.0001), whereas (I) WD only induced significant permeability change at 16 weeks (P=0.0022). (J) A 4-week washout period was sufficient to restore the percentage of normal Paneth cells (P=0.0198 compared to baseline). (H): n=3~10/group. (I): n=7/group. (J): baseline: n=5; 2 and 4wk: n=7; SD control: n=10. (A, C-G): Statistical analysis was performed by Mann-Whitney test. (H): Statistical analysis was performed by two-way ANOVA. (I, J): Statistical analysis was performed by Kruskal-Wallis test. *: P<0.05; **: P<0.01; ****: P<0.0001. Error bars represent standard deviations.

WD consumption results in Paneth cell defects in mice

We first tested if an abnormal Paneth cell phenotype could be induced by obesity generated by excess food consumption due to a genetic susceptibility. We found no significant changes in Paneth cell morphology or numbers per crypt in two well-characterized strains of genetically obese mice (ob/ob and db/db) fed with standard diet (SD) compared to their littermate controls (Fig. S3). In addition, CD patients harboring risk alleles in ATG16L1 T300A (Fig. S2B–C) and NOD2 (Fig. S2D–E) did not have increased probability of developing Paneth cell defects when BMI was above 25 compared to those without these risk alleles. These results suggest that host genetic susceptibility does not play a prominent role in Paneth cell defects associated with obese individuals.

We next tested if diet-induced obesity caused Paneth cell defects. Wild type (WT) C57BL/6 mice were fed with SD or WD for 8 weeks. We used a WD that contained 40% fat content, as this level is most similar to the diet of the average US adult (~34%) (Shan et al., 2019); this percentage of fat is lower than what was used in other published mouse studies (~60%) (Beyaz et al., 2016; Everard et al., 2019; Fu et al., 2019; He et al., 2019; Whitt et al., 2018). WD consumption in mice led to reduced numbers and percentages of morphologically normal Paneth cells (Fig. 1D, S1F, S1K, and S1L). There were no significant alterations in the abundance or distribution of other intestinal epithelial cell types (Fig. 1E–1G, Fig. S1G–S1J).

A critical functional consequence in mice with >6 months WD consumption is mucosal barrier dysfunction, which manifests as increased gut permeability (Genser et al., 2018; Guerville et al., 2017; Lam et al., 2012) and can lead to increased inflammation in the gastrointestinal tract (Li et al., 2008). To understand if Paneth cell defects may be a consequence of increased epithelial permeability, we determined the temporal relationship between WD-induced Paneth cell defects and permeability changes. A 4-week consumption of WD was sufficient to trigger Paneth cell defect (Fig. 1H), whereas increased gastrointestinal tract permeability required 16 weeks in our facility (Fig. 1I). Therefore, with WD consumption, the Paneth cell defects we observed are likely not secondary to barrier dysfunction.

Previous studies determining the effect of dietary changes on microbiome composition and metabolism suggested that certain community members and hepatic triglyceride metabolism are sensitive to dietary perturbation, and the resulting changes are irreversible after a washout period during which the diet was switched back to SD (Heden et al., 2014; Howe et al., 2016). Mice that were fed with WD for 8 weeks were switched back to SD for either 2 or 4 weeks. A 4-week washout was required for the Paneth cell defects to reverse (Fig 1J). Thus, Paneth cell defects associated with a WD consumption in mice are reversible.

Enhanced FXR and type I IFN signaling is associated with WD-induced Paneth cell defects

We used data from multiple OMICs approaches to determine the candidate pathways that mediate the WD-induced Paneth cell defects. First, we performed RNA-seq analysis on full thickness small intestine from mice fed with WD and SD. We used this approach because Paneth cells can receive signals from other cell types (Sehgal et al., 2018). We focused on metabolic pathways, as defective Paneth cells show profound alterations in metabolism (Adolph et al., 2013; Liu et al., 2018; Pentinmikko et al., 2019). Utilizing the Compbio analysis platform (Gehrig et al., 2019), we found that genes associated with Fxr activation were strongly associated with WD consumption (Fig. 2A, Fig. S4A). RT-qPCR of the well-established Fxr target gene, Fgf15 (Kim et al., 2007), showed highly enhanced expression in ileal samples from WD-fed mice than SD-fed mice (Fig. 2B). Furthermore, we found Fgf15 mRNA expression was increased in the crypt base compartment (enriched with Paneth cells (Fig. S4B) of WD fed mice. For a second approach, we performed bulk RNA-seq analysis on full-thickness ileal samples from non-IBD subjects. Subjects with BMI≥25 had elevated mRNA expression of genes associated with an activated FXR pathway (Fig. S4C). Collectively, these data show that activation of the FXR pathway was associated with Paneth cell defects in obese/overweight humans and a diet-induced obesity mouse model.

Figure 2. Activated FXR and type I IFN signaling in WD-associated Paneth cell defects in mice.

(A) A heat map showing upregulated Fxr-associated genes in full-thickness ileum from WD-fed mice and their level of contribution to specific themes in the data set. (B) RT-qPCR of Fxr target gene Fgf15 showed that WD treatment resulted in induction of Fgf15 in the ileum (n=5/group; P=0.0079). By mining publically-available dataset GSE74101 (FXR agonist PX20606 treatment in WT mice), we found that FXR activation reduced Paneth cell-associated (C) Defa6 (adjusted P=0.002) and (D) Reg3g (adjusted P=0.009) expression, without significantly altering the expression of (E) gob let cell-specific gene Muc2 (adjusted P=0.1143). PX20606 treatment also increased expression of (F) Irf7 (adjusted P=0.01) and (G) Oas1 (adjusted P=0.03). (H) Ileum from WD-fed mice showed enhanced expression of Oas1 (P=0.0079). (I) WD-fed mice showed higher serum type I IFN activity (P=0.0159). (C-G): Benjamini & Hochberg adjustment to the P value was applied (n=4/group). (B, H, I): Statistical analysis was performed by Mann-Whitney test (n=5/group). (B-H): Error bars represent standard deviations.

For a third approach, we mined a transcriptomic dataset (GSE 74101) from a published study where WT mice were administered vehicle or the FXR agonist PX20606 for 2 weeks (de Boer et al., 2017), and full-thickness small intestine tissue was collected for RNA-seq analysis. PX20606 significantly repressed the expression of Paneth cell-associated genes Defa6 (Fig. 2C) and Reg3g (which is also expressed in other intestinal epithelial cell types) (Fig. 2D), whereas expression of the goblet cell-specific gene, Muc2 was not altered (Fig. 2E), consistent with our findings with WD consumption. Therefore, excess FXR activation by a pharmacological approach specifically represses genes associated with Paneth cell function.

Previous studies by our group and others have shown that cytokines are important mediators of Paneth cell defects (Dannappel et al., 2014; Gunther et al., 2019; Liu et al., 2018). We queried the GSE 74101 dataset to determine if any signatures associated with cytokine activation were present. We found that PX20606 administration also led to the enhanced expression of the IFN-simulated genes Irf7 and Oas1 (Fig. 2F–G). We also found enhanced expression of IFN-stimulated gene Oas1 in the ileum of WD-fed mice (Fig. 2H). These results are consistent with elevated serum type I IFN bioactivity of WD-fed mice (Fig. 2I).

Secondary bile acid, deoxycholic acid, mediates WD-FXR-associated Paneth cell defects

The transcriptome analysis above suggests that excess FXR activation secondary to WD consumption causes Paneth cell defects. One consequence of defective Paneth cells is susceptibility to infections (Holly and Smith, 2018; Lassen et al., 2014; Ouellette, 2006; Wehkamp and Stange, 2006). Susceptibility to Salmonella typhimurium infection has been particularly shown to correlate with Paneth cell function (Bel et al., 2017; Holly and Smith, 2018; Lassen et al., 2014; Lu et al., 2020; Salzman et al., 2003). Therefore, we treated Salmonella-infected WT mice with vehicle or the FXR agonist GW4064. All mice received GW4064 upon infection, with two groups additionally getting pretreatment for 4 weeks or at the same time of Salmonella infection. Mice receiving the pretreatment of GW4064 showed increased mortality as compared to those treated with GW4064 only after Salmonella infection or with vehicle control (no treatment with GW4064) (Fig. 3A). This correlated with increased bacterial titers in the spleen and intestine of GW4064-pretreated mice (Fig. S4D, E). The detrimental effects on mortality with GW4064 pretreatment were reduced in Salmonella-infected Fxr^−/− mice (Fig. 3B). These results show that excess FXR activation can lead to increased mortality with Salmonella infection, further suggesting the Paneth function may be a target of excess FXR signaling.

Figure 3. WD-DCA-mediated FXR activation induces Paneth cell defects.

(A) WT mice pretreated with FXR agonist GW4064 showed worse mortality after Salmonella Typhimurium infection compared to non-GW4064 treated or treated with GW4064 at the onset of infection (n=10/group; P=0.0035). (B) GW4064 pretreatment increased the mortality of Fxr^+/+ mice after Salmonella Typhimurium infection but not in Fxr^−/− mice (n=5/group; P=0.0084). Distal ileum from WD-fed mice had higher levels of (C) DCA (P<0.0001) and (D) lithocholic acid (LCA; P=0.0021). n=10/group. (E) Reduction of DCA from WD-fed mice by bile acid sequestrant cholestyramine prevented Paneth cell defects (P<0.0001). Total n: no cholestyramine: n=6, with cholestyramine: n=14. (F) Administration of DCA but not LCA in SD-fed mice recapitulated WD-induced Paneth cell defects (P<0.0001). Total n: SD, WD, DCA: n=15/group, LCA: n=10. (G) Fxr^−/− mice and littermates (Fxr^+/+) were treated with SD, WD, SD+DCA, or SD+GW4064, and analyzed for Paneth cell morphology. While WD, DCA, and GW4064 all induced Paneth cell defects in the Fxr^+/+ mice (P=0.0005), the Fxr^−/− mice did not develop this phenotype (P=0.1257). n: 4–14/group. (H) WT, WD-fed mice treated with FXR antagonist guggelsterone did not develop Paneth cell defects (P=0.0286). n=4/group. (A, B): Kaplan-Meier curve analysis was performed with Logrank test. (C-H): Statistical analysis between 2 groups was performed by Mann-Whitney test. Statistical analysis between more than 2 groups was performed by Kruskal-Wallis test. *: P<0.05; **: P<0.01; ***: P<0.001. ****: P<0.0001. Error bars represent standard deviations.

The best characterized mechanism by which WD activates FXR pathway is through the induction of bile acids (Dermadi et al., 2017; Jena et al., 2018). Primary bile acids, the end product of cholesterol catabolism, are produced in the liver, and converted to secondary bile acids in the terminal ileum/cecum by Clostridium spp. and subsequently reabsorbed in the ileum as part of the enterohepatic circulation (Chiang, 2013). We measured the concentrations of primary bile acids cholic acid (CA), chenodeoxycholic acid (CDCA), muricholic acid (MCA) and secondary bile acids deoxycholic acid (DCA) and lithocholic acid (LCA) in the small intestine by mass spectrometry. These bile acids were analyzed because they are present in both humans and mice in significant quantity with the exception of MCA, which is predominantly present in mice (Chiang, 2013). Mice fed with WD had significantly increased DCA (13-fold) and LCA (7.6-fold) levels in the ileum (Fig. 3C–D), without changes to CA, CDCA, or MCA levels (Fig. S5). Therefore, WD exposure was associated with elevated secondary bile acids in the distal/terminal ileum.

To test if elevated WD-associated secondary bile acids caused Paneth cell defects, we performed the following experiments: (1) concomitant lumenal treatment with bile acid sequestrant cholestyramine to lower bile acid levels and (2) administration of secondary bile acids DCA and LCA in mice fed with SD. Co-administration of cholestyramine prevented Paneth cell defects in WD-fed mice (Fig. 3E), and reciprocally, DCA administration induced Paneth cell defects in SD-fed mice (Fig. 3F). The latter effect was specific for DCA, as LCA did not trigger Paneth cell defects. Therefore, increased levels of ileal DCA contribute to the WD-mediated Paneth cell defects in mice.

We tested the causality between WD-associated Fxr activation and Paneth cell defects in Fxr^−/− mice and littermates. These mice were treated with SD, WD, SD+DCA, or SD+FXR agonist GW4064. As expected, Paneth cell defects were observed with WD, DCA, or GW4064 fed/treated WT littermates but not in Fxr^−/− mice (Fig. 3G). WD-fed, WT mice additionally treated with FXR antagonist guggelsterone did not develop defective Paneth cells (Fig. 3H). These data further support the hypothesis that elevated activity of the FXR pathway in WD fed mice induces Paneth cell defects.

Excess FXR activation directly affects Paneth cells in the intestinal epithelium

FXR is active in a wide variety of cell types (Jiang et al., 2015a; Shih et al., 2001; Trabelsi et al., 2015). We first tested if DCA-mediated FXR activation in the intestinal epithelium was required for Paneth cell dysfunction. We determined the effect of WD/DCA on intestinal epithelium using intestinal epithelium-specific (Villin-Cre) Fxr conditional knockout mice (Fxr^ΔIEC mice). In these experiments, Paneth cell defects were induced by WD, DCA, and GW4064 in the littermate control Fxr^fl/fl mice but not in similarly treated Fxr^ΔIEC mice (Fig. 4A). Similar results were obtained in vitro, by treating Fxr^+/+ and Fxr^−/− ileal organoids with DCA or GW4064. Excess FXR stimulation caused a reduction in Paneth cells in Fxr^+/+ but not Fxr^−/− organoids (Fig. 4B and Fig. S6A–E) without significantly affecting organoid sizes (Fig. S6F). Of note, the reduction of Paneth cells in vitro occurred using a DCA concentration comparable with the DCA concentration in the ileum of WD-fed mice (~45 μM; Fig. 3).

Figure 4. Fxr signaling in Paneth cells is essential for Western diet (WD)-mediated Paneth cell defects.

(A) WD, DCA, and GW4064 triggered Paneth cell defects in Fxr^fl/fl mice, but the effects were abrogated in the Fxr^ΔIEC mice (P=0.0005). Total n=5~13/group. (B) Ileal organoids derived from Fxr^−/− and Fxr^+/+ mice showed differential responses to DCA and GW4064. Whereas DCA and GW4064 both reduced the percentages of Paneth cells/organoid in the Fxr^+/+ organoids, such an effect was abrogated in the Fxr^−/− organoids (P=0.0036). (C) WD-mediated Paneth cell defects were abrogated in the Fxr^ΔPC mice (P=0.0006). Total n=7/group. (D) Ileal organoids derived from the Fxr^ΔPC mice did not show reduced percentages of Paneth cells/organoid when treated with DCA or GW4064. (B, D): Results were from 3 independent experiments, with each experiment containing 20 organoids/group. (A-D) Statistical analysis between 2 groups was performed by Mann-Whitney test. Statistical analysis between more than 2 groups was performed by Kruskal-Wallis test. *: P<0.05; **: P<0.01; ***: P<0.001; ****: P<0.0001. Error bars represent standard deviations.

To test if DCA acted directly on Paneth cells, we generated a Paneth cell-conditional knockout of FXR (α-defensin-4-IRES-Cre; herein termed Fxr^ΔPC) (Liu et al., 2018) and showed that the Fxr^ΔPC mice were protected from WD-driven Paneth cell defects (Fig. 4C). Similarly, ileal organoids from Fxr^ΔPC mice did not show a reduction in Paneth cells with treated with either DCA or GW4064 (Fig. 4D). These data are consistent with a direct effect of DCA on Paneth cells through FXR.

The role of microbiota in WD-mediated Paneth cell defects

Dietary modulation of gut microbiome composition is well-established (Bisanz et al., 2019), and certain host phenotypes (including obesity) can be horizontally transmissible via microbiota transfer (Turnbaugh et al., 2009; Turnbaugh et al., 2006). WD-fed mice in our facility showed an increase in the abundance of Firmicutes and a relative reduction in abundance of Bacteroides, consistent with previous studies (Fig. S7A, 7B) (Ley et al., 2005; Turnbaugh et al., 2006). We next tested if WD-induced Paneth cell defects could be transmitted between mice. We first performed co-housing in a specific pathogen-free (SPF) facility where WT, microbiota-recipient mice were pretreated with broad-spectrum antibiotics to create a niche for donor microbiota. Microbial donors were mice fed for 4 weeks with WD or SD. The donors were replaced every other day to ensure fresh WD microbiota were provided (Fig. S7C), as microbial community composition changes rapidly as a function of diet (Turnbaugh et al., 2008). At the end of the co-housing period, the WD microbiota recipients showed no difference in the percentages of normal Paneth cells or Paneth cell density (Fig. S7D–E). In a second experiment, we performed microbiota transfer using cecal contents from SPF-housed WT mice fed either SD or WD into germ-free recipient mice, which were maintained on SD throughout the experiment. PBS-gavaged germ-free mice had lower percentages of normal Paneth cells (Fig. S7F), consistent with a prior study (Kernbauer et al., 2014). However, we observed no significant difference in Paneth cell phenotypes between mice gavaged with cecal contents from SD- and WD-fed SPF mice (Fig. S7F). Therefore, WD-associated Paneth cell defects were not induced by microbiome transfer alone.

The experiments above suggest that the primary role of the microbiome in WD-mediated Paneth cell defects is to facilitate the conversion of primary to secondary bile acid. Clostridium spp. harboring a BaiCD operon, are over-represented in mice fed with WD (Jena et al., 2018), and can mediate the conversion of primary to secondary bile acids (Jain et al., 2018). We found that the ileal mucosal microbiome in WD-fed, SPF-housed mice were indeed enriched with BaiCD expression (Fig. 5A). We next treated WD-fed, SPF-housed mice with or without vancomycin to deplete gram-positive bacteria, including Clostridium spp. Vancomycin-treated mice showed a reduction in BaiCD gene detection in the microbiome (Fig. 5B). As expected, vancomycin-treated mice were protected from WD-mediated Paneth cell defects (Fig. 5C). Furthermore, we gavaged WD-fed, germ-free mice with either PBS or Clostridium scindens (a well-characterized Clostridium sp. containing the BaiCD operon whose gene products convert primary to secondary bile acids) (Jain et al., 2018). Mice gavaged with C. scindens showed increased DCA levels in the ileum (Fig. S7G) and reduced percentages of normal Paneth cells (Fig. 5D). Thus, C. scindens promotes WD-induced Paneth cell defects. To determine if diet alone was sufficient to elicit Paneth cell defects in the absence of the microbiome, we also fed germ-free mice with either SD or WD for 4 weeks, without fecal transplantation. As expected, mice in each of these groups did not show increased abnormal Paneth cells (Fig. 5E).

Figure 5. The role of microbiota in WD-mediated Paneth cell defects.

(A) Mucosal microbiome from WD-fed mice was enriched with BaiCD operon (n=8/group; P=0.0281). (B) The amount of BaiCD operon in the microbiota was diminished after vancomycin treatment (n=5/group; P=0.0079). (C) Vancomycin treatment prevented WD-mediated Paneth cell defects in SPF housed mice (n=10/group; P<0.0001). (D) Germ-free mice (GF) fed with WD and gavaged with C. scindens showed Paneth cell defects compared to those gavaged with PBS (P=0.0003). Total n: PBS: 7, C. scindens: 8. (E) GF mice fed with SD or WD without cecal content transfer did not develop Paneth cell defects (n=10/group; P=0.4309). Statistical analysis was performed by Mann-Whitney test. *: P<0.05; **: P<0.01; ***: P<0.001; ****: P<0.0001. Error bars represent standard deviations.

WD-FXR signaling activates Type I IFN and mediates Paneth cell defects

A second target of DCA-FXR signaling is type I IFNs that also have important roles in inhibition of Paneth cell function (Liu et al., 2013). FXR activation can lead to production and activation of type I IFN in the context of viral infection (Honke et al., 2017; Winkler et al., 2020). As noted above, we found evidence of elevated type I IFN activity in WD-fed mice (Fig. 2I) consistent with a previous study showing a direct correlation between systemic IFN-α and triglyceride levels (Tominaga et al., 2010). SD-fed mice administered with the FXR activator, GW4064 also showed increased serum type I IFN activity (Fig. 6A), as well as increased expression of Irf1/9 in the crypt base compartment (Fig. 6B, C). To test if elevated levels of type I IFN activity played a role in Paneth cell defects, we treated WD-fed mice with either a blocking monoclonal antibody to the type I IFN receptor (anti-Ifnar1, MAR1–5A3) or an isotype control antibody (GIR208). Remarkably, anti-Ifnar1 administration prevented WD-induced Paneth cell defects (Fig. 6D). Consistent with this result, Ifnar1^−/− mice were also protected from WD-induced Paneth cell defects as compared to littermate genetic controls (Fig. 6E). Finally, Stat1^ΔIECmice were protected from WD-induced Paneth cell defects (Fig. 6F), even though they maintained high levels of Fgf15 expression in the ileum (Fig. 6G), suggesting that type I IFN signaling in the intestinal epithelium was also critical in mediating Paneth cell defects.

Figure 6. Type I IFN mediates WD-associated Paneth cell defects.

(A) Administration of GW4064 in WT, SD-fed mice showed induction of type I IFN (Ctl: n=5; GW4064: n=10; P=0.0193). This was associated with enhanced expression of type I IFN associated genes (B) Irf1 and (C) Irf9 in the crypt base compartment. (D) Treatment with anti-Ifnar1 (MAR1–5A3) but not isotype control antibody (GIR208) prevented WD-induced Paneth cell defects (n=5/group; P=0.0079). (E) Ifnar^−/− mice were protected from WD-associated Paneth cell defects (Ifnar^+/+: n=13, Ifnar^−/−: n=14; P<0.0001). (F) Stat1^ΔIEC mice fed with WD did not develop Paneth cell defects (P=0.0238). Total n: Stat1^fl/fl: n=3; Stat1^ΔIEC: n=6. (G) Stat1^ΔIEC mice fed with WD maintained high Fgf15 expression compared to the Stat1^fl/fl mice (P=0.1). (H) Fxr^ΔIEC mice fed with WD did not show abrogation of type I IFN induction (P=0.6523). Total n: Fxr^fl/fl: n=8; Fxr^ΔIEC: n=6. (I) GW-4064-treated mice did not show increased lamina propria F4/80+ cells (P=0.6560). (J) WD-fed WT mice treated with clodronate were protected from Paneth cell defects (P=0.0031). Total n: Ct: n=10; clodronate: n=6. (K) DCA treatment induced type I IFN activity in macrophages in vitro (P<0.0001). n=4/group. (L) Fxr^ΔMye mice fed with WD showed abrogation of type I IFN induction (P=0.0193). Total n: Fxr^fl/fl: n=3; Fxr^ΔMye: n=4. (M) Fxr^ΔMye mice were protected from WD-mediated Paneth cell defects (P=0.0033). Total n: Fxr^fl/fl: n=15; Fxr^ΔMye: n=16. Statistical analysis for all panel was performed by Mann-Whitney test. *: P<0.05; **: P<0.01, ****: P<0.0001. Error bars represent standard deviations.

We found that a likely source of the type I IFN enhanced by FXR/WD diet was intestinal myeloid cells. Fxr^ΔIEC mice showed similar levels of serum type I IFN activity as controls (Fig. 6H). We did not identify differences in the amount of F4/80+ cells infiltrating the lamina propria of GW4064-treated mice (Fig. 6I), suggesting the effect of FXR activation was not due to changes in macrophage cell numbers, but rather the function of these cells. WT mice fed with WD and treated with intraperitoneal clodronate to deplete intestinal myeloid cells (Weisser et al., 2012) were protected from WD-associated Paneth cell defects (Fig. 6J). Consistent with this result, DCA induced type I IFN bioactivity in macrophages in vitro (Fig. 6K). Finally, mice with Fxr conditionally knocked out in myeloid cells (Fxr^ΔMye) when fed with WD, failed to show elevated serum type I IFN bioactivity compared to littermates (Fig. 6L), and these mice were protected from WD-mediated Paneth cell defects (Fig. 6M). These results support a role for elevated Type I IFNs in myeloid cells that was necessary for Paneth cell defects in vivo.

Discussion

We previously showed that Paneth cell defects can be induced in the context of a gene-environment interaction (Liu et al., 2018). In this current study, we utilized our extensive collection of cohorts with Paneth cell phenotypes and observed a significant subset of overweight/obese human subjects had Paneth cell defects. Our studies in mice revealed that the obesity-associated Paneth cell dysfunction was not related to genetics but instead was likely related to WD consumption. Mechanistically, the defective Paneth cells driven by WD consumption was due to increased secondary bile acid (specifically DCA) in the ileum that were converted from primary bile acids by BaiCD-expressing Clostridium spp. This process was dependent on FXR signaling in two cell types: activation of FXR in Paneth cells and in myeloid cells; the latter augmented type I IFN production in response to FXR activation. Myeloid cells -derived type I IFNs in turn acted on the intestinal epithelium to produce abnormal Paneth cells. Both of these enhanced signaling activities are required to produce abnormal Paneth cells. (Fig. 7).

Figure 7. Proposed model of how WD triggers Paneth cell defects.

Our current study shows that potent environmental factors (such as WD consumption) could induce Paneth cell defects without host genetic susceptibility. WD induces FXR signaling that directly target Paneth cells, and also induces type I IFN production in myeloid cells such as macrophages. Both FXR and type I IFN signaling are required to trigger Paneth cell defects.

Paneth cell dysfunction indicates defects in metabolism that are driven by genetics and/or the environment. In this regard, mouse and human Paneth cells have shown remarkable concordance in several studies (Adolph et al., 2013; Cadwell et al., 2008; Cadwell et al., 2010; Jackson et al., 2020; Kaser et al., 2008; Liu et al., 2017; Lu et al., 2020), enabling reverse and forward translational approaches. Here we show that WD, which has been suspected to induce and exacerbate IBD (Ananthakrishnan, 2015; Ng et al., 2015), has a profound effect on Paneth cell function, which could lead to increased susceptibility and/or severity of IBD (Liu et al., 2018; Liu et al., 2017; VanDussen et al., 2014). This observation is of particular relevance, as the incidence of IBD is on the rise in historically low incidence areas which have now transitioned to a more western lifestyle (Garduno-Diaz and Khokhar, 2013; Ng et al., 2018; Siddiquee et al., 2015). While previous studies have shown that WD consumption affects nutrient absorption, metabolism, and cell cycle (Ait-Omar et al., 2011; de Wit et al., 2008; Newberry et al., 2009), how WD impacts low-abundance but functionally important cell types such as Paneth cells is relatively unknown. A key finding of our study is that Paneth cell dysfunction can be induced after 8 weeks of WD consumption, which in turn can lead to more aggressive disease in IBD. While we cannot exclude that intestinal stem cells or their niche is affected by bile acids, we did not observe alteration in lineage allocation in other secretory lineages and stem/progenitor proliferation was not altered. Prolonged WD consumption period may result in additional changes secondary to its effect on the stem cell niche, as suggested by a recent study (Beyaz et al., 2016). Nonetheless, our data suggest that, prior to permanent changes in intestinal stem cells, WD consumption is capable of inducing Paneth cell defects. Moreover, our study also suggests the potential reversibility of Paneth cell defect after cessation of short-term WD consumption. This has important implications in the management of IBD.

Using complementary models, we showed that WD has a direct effect on the intestinal epithelium, and that the nuclear receptor FXR is a central target. FXR signaling regulates metabolism in many cell types (Ryan et al., 2014; Sun et al., 2018), and appears to have this role in dysfunctional Paneth cells known to have altered metabolism (Cadwell et al., 2008; Liu et al., 2018). FXR widely affects many organs and may have different physiological effects in specific locations (Gonzalez et al., 2016; Jiang et al., 2015a; Rao et al., 2016; Trabelsi et al., 2015). In the gut, FXR has been implicated in improving microbiota composition, barrier function (Sorribas et al., 2019), and stem cell proliferation (Fu et al., 2019). Strategies of using FXR agonists are being developed for several diseases, including metabolic syndrome, insulin resistance, obesity, and nonalcoholic steatohepatitis (Fang et al., 2015; Jiang et al., 2015b; Mudaliar et al., 2013; Neuschwander-Tetri et al., 2015). Our study provides molecular and functional insight into how FXR activation may have unexpected and undesirable effects on gut innate immunity.

Our study also suggests that Paneth cell defects associated with WD consumption are mediated by commensal Clostridium spp. by affecting primary to secondary bile acid conversion, as suggested by a previous study (Jena et al., 2018). Because Paneth cell defects can trigger mucosal dysbiosis in CD patients (Liu et al., 2016), these data indicate that expansion of Clostridium spp., as part of WD-mediated dysbiosis, further affects bile acid-FXR-IFN signaling.

In addition to directly activating type I IFN by FXR under virus infection (Honke et al., 2017), previous studies have also shown that IFN regulatory transcription factor Irf7 can be recruited to the Fxr promotor (Renga et al., 2013) and that an Fxr deficiency results in repressed expression of type I IFN (Zheng et al., 2017). Our findings agree with previous reports showing that type I IFN signaling has a critical role in diet-induced obesity (Hannibal et al., 2017). Furthermore, Irgm knockout mice (with constitutively high levels of type I IFN) also possess Paneth cell defects (Liu et al., 2013; Sun et al., 2015). Studies using intestinal epithelium-specific Ifnar1-deficient mice showed that type I IFN signaling is critical in Paneth cell homeostasis (Tschurtschenthaler et al., 2014). While our in vitro data suggest that macrophages may be a source of type I IFN, it is possible that additional cell types targeted by the LysM-Cre (i.e., myeloid cells) are capable of producing type I IFN in this context. In addition, our study does not exclude the possibility that type I IFN activity can be induced by FXR-independent mechanisms. Together, with our data, an emerging theme is that dietary modulation of the FXR-type I IFN axis affects gut innate immunity and homeostasis. The involvement of type I IFN in modulating Paneth cell defects also raises the possibility that other disease states by which type I IFN is activated, such as chronic viral infections (Norman et al., 2015), could impact CD pathogenesis. In addition, our finding that WD-fed Stat1^ΔIEC mice showed an even higher level of Fgf15 expression compared to WD-fed Stat1^f/f mice suggests that while FXR activation is upstream of type I IFN activation, it is possible that a negative feedback loop is involved. Here, type I IFN could modulate the expression of FXR pathway-associated genes.

Paneth cells are important mediators for IBD and other disease processes (Wehkamp and Stange, 2020). Our study identifies specific signaling pathways by which WD consumption, a common environmental factor for these diseases, can trigger defective Paneth cells. These results have implications for the pathogenesis of IBD and for the potential utility for testing and monitoring the function of Paneth cells in IBD patients. Future experiments dissecting the different components of WD (e.g., fat and sugar) will provide additional insight into the pathogenesis of WD-associated Paneth cell defects.

STAR methods:

RESOURCE AVAILABILITY

Lead contact

Further information for resources and reagents should be directed to the Lead Contact, Thaddeus S. Stappenbeck (stappet@ccf.org).

Materials availability

The intestinal stem cells isolated for this study will be made available upon request and provided with completed Materials Transfer Agreement. The human material is part of routine clinical care and is under legal requirement to be maintained in the Pathology Departments of our institutions, and as a result will not be available.

Data and code availability

The full-thickness ileum transcriptomic dataset has been deposited to ArrayExpress, with accession number: E-MTAB-8332. The microbiome dataset has been deposited to ArrayExpress, with accession number: E-MTAB-8334.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Human subjects

The study was approved by the Institutional Review Boards of Washington University School of Medicine (St. Louis) and Cedars-Sinai Medical Center (Los Angeles). CD or non-IBD subjects who underwent ileocolectomy between 2006–2013 at Washington University, St. Louis, or Cedars-Sinai Medical Center, Los Angeles, were previously described (Liu et al., 2018; Liu et al., 2017; VanDussen et al., 2014). De-identified tissue samples from ileal resection margins that were free of acute inflammation were used for Paneth cell phenotype analysis.

The following information was retrieved from the medical record: sex, body weight, height at time of operation (for body mass index [BMI] calculation). The genotypes of the patients were obtained using immunochip (VanDussen et al., 2014) or through Taqman genotyping assay (ThermoFisher) with genomic DNA extracted from formalin-fixed, paraffin-embedded tissue based on the vendor’s instructions (Liu et al., 2018; Liu et al., 2017). The average age, gender and their influences on the results are included in Table S1.

Mouse treatments

The animal studies were approved by the ethical committee at Washington University School of Medicine. Mice were either obtained from Jackson Laboratory or bred and maintained at Washington University School of Medicine. Mice were maintained under specific pathogen free (SPF) conditions unless otherwise specified. Only 4–6 week-old mice (post-weaning) in C57/BL6 background were used. Genotype information is included in Key Resources Table. Mice were exposed to the following diet: SD (Lab Supply Labdiet 5053; with calories provided by 62% carbohydrates, 13% fat, and 25% protein), WD (Research Diets D09100310; with calories provided by 40% carbohydrates, 40% fat, and 20% protein), and WD with 2% cholestyramine admixed (Research Diets D18030505; with calories provided by 40% carbohydrates, 40% fat, and 20% protein). Mice were exposed to different diets for the time indicated (4–24 weeks) and then sacrificed for tissue collection. For antibiotic treatment experiments, mice received vancomycin for the entire duration of diet exposure (Jain et al., 2018). For bile acid administration, sodium deoxycholate (0.2%; Sigma-Aldrich #30970) was added to the drinking water for 4 weeks. Lithocholic acid (20 mg/kg; Sigma-Aldrich #L6250) was administered via intraperitoneal (i.p.) route daily for 4 weeks. GW4064 (Tocris #2473) was administered i.p. at 10 mg/kg/day. Guggulsterone (Tocris #3570) was administered i.p. at 30 mg/kg/day. For type I IFN blockade, mouse anti-mouse monoclonal antibody specific for Ifnar1 (MAR1–5A3; Leinco) or isotype control antibody GIR208 (Leinco) was administered by using a loading dose of 2.5 mg/mouse, followed by a weekly dose of 0.5 mg/mouse (Pinto et al., 2011) per vendor instructions. To ensure that the bile acid levels did not fluctuate per fast/feed cycle change, all the mice were sacrificed in the hours between 10am – 3pm.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Lysozyme C Antibody (C-19)	Santa Cruz	Cat.# sc-27958
Cleave caspase-3 Antibody	Cell signaling	Cat.# 9661
Defensin alpha 5 Antibody (8C8)	Novus	Cat.# NB110–60002
Mucin 2 Antibody (H-300)	Santa Cruz	Cat.# sc-15334
F4/80 Antibody	Abcam	Cat.# 6640
Lab Vision™ Ki-67, Rabbit Monoclonal Antibody	ThermoFisher	Cat.# RM9106
Donkey anti-Rabbit IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor 488	ThermoFisher	Cat. # A-21206
Donkey anti-Goat IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 594	ThermoFisher	Cat. # A-11058
Donkey anti-Mouse IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor 488	ThermoFisher	Cat. # A-21202
Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Biotin-XX	ThermoFisher	Cat. # B-2770
Chromogranin A antibody	DAKO	Cat. # A0430
QIAamp DNA Stool Mini Kit	QIAGEN	Cat. # 51604
Quant-iT™ RiboGreen® RNA Assay Kit	ThermoFisher	Cat. # R11490
iScript™ Reverse Transcription Supermix	Bio-Rad	Cat. #1708840
TB Green Advantage qPCR Premix	Clontech	Cat. #639676
RNeasy FFPE Kit	QIAGEN	Cat. # 73504
RNeasy Mini Kit	QIAGEN	Cat. # 74104
Bacterial and Virus Strains
Clostridium scindens	ATCC	Cat.# 35704
Salmonella enterica serovar Typhimurium	ATCC	Cat.# 14028
Biological Samples
Human ileum tissue blocks	This study	N/A
Chemicals, Peptides, and Recombinant Proteins
GW4064	Tocris	Cat. # 2473
Sodium deoxycholate	Sigma-Aldrich	Cat. # 30970
Lithocholic acid	Sigma-Aldrich	Cat. # L6250
Z-Guggulsterone	Tocris	Cat. # 3570
MAR1-5A3	Leinco	Cat. # I-401
GIR208	Leinco	Cat. # I-443
Vancomycin hydrochloride	Sigma-Aldrich	Cat. # V2002
Fluorescein isothiocyanate-dextran	Sigma-Aldrich	Cat. # FD4
Clodrosome,Clodronate liposome	Encapsula Nano Sciences	Cat. # CLD-8909
Encapsome, Control liposome	Encapsula Nano Sciences	Cat. # CLD-8910
Dulbacco’s modified Eagle’s medium (DMEM)	Life Technologies	Cat. # 11885–084
Horse Serum	Sigma-Aldrich	Cat. # H-1207
Fetal Bovine Serum	Sigma-Aldrich	Cat. # F-2442
Penicillin-Streptomycin	Sigma-Aldrich	Cat. # P-4333
HEPES	ThermoFisher	Cat. # 15630080
L-glutamine	ThermoFisher	Cat. # 25030081
Sodium Pyruvate	ThermoFisher	Cat. # 11360070
Recombinant Mouse M-CSF (carrier-free)	Biolegend	Cat. # 576402
Critical Commercial Assays
Custom TaqMan™ SNP Genotyping Assay	ThermoFisher	Cat. # 4332072
Deposited Data
GSE 74101	de Boer et al., 2017	N/A
E-MTAB-8332	This study	N/A
E-MTAB-8334	This study	N/A
Experimental Models: Cell Lines
Small intestinal organoids from wild type mice	This study	N/A
Small intestinal organoids from Fxr−/− mice	This study	N/A
Experimental Models: Organisms/Strains
SPF mice (C57BL/6J)	Jackson Laboratory	Cat.# 000664
Fxr−/−	Jackson Laboratory	Cat.# 007214
Stat1−/−	Jackson Laboratory	Cat.# 012606
ob/ob	Jackson Laboratory	Cat.# 000632
db/db	Jackson Laboratory	Cat.# 000697
Villin-Cre	Jackson Laboratory	Cat. # 004586
LysM-Cre	Jackson Laboratory	Cat. # 004781
Fxrfl/fl	Gonzalez lab	N/A
α-defensin-4-IRES-Cre	Dempsey lab	N/A
Oligonucleotides
BaiCD Forward: GGWTTCAGCCCRCAGATGTTCTTTG	Wells et al., 2003	N/A
BaiCD Reverse: GAATTCCGGGTTCATGAACATTCTKCKAAG	Wells et al., 2003	N/A
Fgf15 Forward: AGTACCTGTACTCCGCTGGT	NM_008003.2	N/A
Fgf15 Reverse: CAGCCCGTATATCTTGCCGT	NM_008003.2	N/A
Oas1 Forward: CGCACTGGTACCAACTGTGT	Steed et al., 2017	N/A
Oas1 Reverse: CTCCCATACTCCCAGGCATA	Steed et al., 2017	N/A
Recombinant DNA
Software and Algorithms
Graphpad Prism	Graphpad Software, Inc.	Version 9.0.0
Olympus cellSens Dimension	Olympus	Version 1.17
SAS	SAS Institute	Version 9.4
Compbio	Washington University	N/A
Other
Standard diet	Lab Supply	Cat. # 5053
Western diet	Research Diets	Cat. # D09100310
Western diet (with 2% cholestyramine)	Research Diets	Cat. # D18030505

For co-housing experiments, 5 WT mice that had been and continued to be exposed to WD were designated as microbial donors. Additional WT mice exposed to SD designated as microbial recipients were exposed to an antibiotic cocktail of vancomycin, neomycin, ampicillin, and metronidazole for 2 weeks (Moon et al., 2015), followed by co-housing with microbial donors. The microbial donors and recipients were co-housed and all exposed to standard diet. A different microbial donor was used every other day to ensure the donor continue to provide microbes that are consistent with WD exposure. Co-housing lasted 2 weeks.

For germ-free mice experiment, mice were fed with irradiated SD or WD as indicated. Cecal contents from SD- or WD-fed, SPF-housed mice were used for fecal microbiota transplant every 2 weeks. For gavage of Clostridium scindens (ATCC #35704), mice were orally administered PBS or 7×10⁸ bacteria per mouse every 2 weeks as described previously (Jain et al., 2018).

Salmonella infection model

Salmonella enterica serovar Typhimurium (ATCC # 14028) was grown overnight at 37°C in LB broth under mild aeration. Bacteria were washed twice in cold PBS and resuspended at 10⁸ CFU/50 μL. Water and food were withdrawn 4 hours prior to treatment with 20 mg streptomycin in 75 μL; immediately following streptomycin treatment, water and food were again provided. Twenty hours later, water and food were again removed for 4 hours, after which mice were infected with 10⁸ CFU S. Typhimurium per oral gavage. Water was resumed immediately and food was provided 2 hours post-infection.

METHOD DETAILS

Immunofluorescence and immunohistochemistry

For immunofluorescence, unstained slides were deparaffinized, followed by antigen retrieval using Trilogy (Cell Marque #920P-09; Rocklin, CA). The slides were blocked with 1% bovine serum albumin and 0.1% Triton X-100 in phosphate buffer saline for 1 hour, before applying primary antibody overnight. After overnight incubation, secondary antibody was applied for 45 minutes, and slides were subsequently washed and stained with Hoechst (1:20,000; Sigma-Aldrich #33258). The following primary antibodies were used: lysozyme (1:100; Santa Cruz# sc-27958), defensin 5 (1:2000; Novus #NB110–60002), chromogranin A (1:500; Dako # A0430), and MUC-2 (1:200; Santa Cruz# sc-15334). Secondary antibodies used include donkey anti-rabbit, donkey anti-goat, and donkey anti-mouse antibodies (1:500; Life Technologies). For Ki-67 immunohistochemistry, unstained slides were deparaffinized, followed by antigen retrieval using Diva Decloaker buffer (Biocare Medical). The slides were then quenched with 10% H₂O₂ in methanol and blocked as described above. The rabbit anti-human Ki-67 antibody was then applied (1:400; Lab Vision) overnight, followed by goat anti-rabbit biotinylated secondary antibody (1:200; Thermo Fisher Scientific). Signal detection was performed by using Vectastain Elite ABC kit (Vector) and DAB (Vector). The stains were reviewed using an Olympus BX53 microscope equipped with an Olympus DP73 camera and cellSens Dimension software for adjusting brightness and contrast and image cropping.

Paneth cell phenotype analysis

Lysozyme and defensin 5 immunofluorescence were interpreted as described previously (Cadwell et al., 2008; Cadwell et al., 2010; Liu et al., 2014; Liu et al., 2016; VanDussen et al., 2014). For both human and mouse samples, each Paneth cell was classified into normal or one of the 5 abnormal categories, including: disordered (abnormal distribution and size of the granules), diminished (≤10 granules), diffuse (smear of lysozyme or defensin within the cytoplasm with no recognizable granules), excluded (majority of the granules do not contain stainable material), and enlarged (rare, megagranules) (Liu et al., 2014; VanDussen et al., 2014). The last two categories were only observed in human samples.

BMDM stimulation

BMDMs were isolated as described previously (Steed et al., 2017). Briefly, cells from femurs and tibia of mice were isolated and plated in sterile dishes containing high glucose DMEM (Life Technologies) with 5% horse serum (Sigma-Aldrich), 10% fetal bovine serum (Sigma-Aldrich), Penicillin-Streptomycin (Sigma-Aldrich), 10mM HEPES (Corning), 1% L-glutamine, 1% sodium pyruvate and Macrophage Colony-Stimulating Factor (Biolegend) at 10 ng/ml for 7 days prior to experimental procedure.

For stimulation experiments, BMDMs were collected in DMEM by cell scraper and reseeded in 6 well plates containing DMEM supplemented with 10% fetal bovine serum, Penicillin-Streptomycin (Sigma-Aldrich), 10mM HEPES (Corning) and allowed to adhere. After 24 hours, cells were washed with sterile PBS, fresh media was added and cells were stimulated with DCA (10μM) or vehicle (PBS). 24 hours post-stimulation, supernatants were collected and assayed for type I IFN serum activity.

Laser capture microdissection

The distal ileum of the mice was prepared and fixed in methacarn as previously reported (Cadwell et al., 2010). The crypt base epithelial cells enriched for Paneth cells were captured and RNA isolated by previously described techniques (Cadwell et al., 2010; VanDussen et al., 2014).

Intestinal permeability assay

Food and water were removed from the mouse cage for 4 hours prior to performing permeability assays. FITC-dextran (average molecular weight 4,000; Sigma-Aldrich #FD4) solution (100 mg/ml) was gavaged into mice at 44 mg/100g. After 3 hours, serum was collected, and FITC-dextran concentration analyzed by fluorescence.

Clodronate treatment

Clodronate or control liposomes (40 μl from 23mg/ml stock; Encapsula Nano Sciences) were mixed with 160μL of molecular-grade PBS. Mice were then injected with clodronate or control liposomes intraperitoneally twice per week for 8 weeks.

In vitro studies

Small intestinal stem cells were isolated as described previously (VanDussen et al., 2015). Ileal crypts were cultured in Matrigel (BD Biosciences) in the presence of 50% L-WRN conditioned medium (contains WNT3A, R-SPONDIN 3 and NOGGIN) (Miyoshi et al., 2012; Miyoshi and Stappenbeck, 2013). Compared to other methods (Sato et al., 2009), this protocol allows for rapid expansion of intestinal epithelial spheroids enriched with stem cells. The spheroids were plated out at Day 0. From Day 1 to Day 7, the culture cells were cultured in ENR medium (Madison et al., 2015) which contains EGF, NOGGIN, and R-SPONDIN 1, to allow for Paneth cell differentiation. DCA and GW4064 were added to the ENR medium on Day 7. The organoids were harvested at Day 14 and processed for immunofluorescence.

Microbiota collection

Ileal mucosal microbiome was collected from the indicated groups, and microbial genomic DNA was isolated using the QIAamp DNA Stool Mini Kit (QIAGEN), followed by quantitative RT-PCR. The fecal microbiota was collected as described previously (Liu et al., 2018).

16s rRNA sequencing assay and data analysis

Fecal samples were processed for 16s rRNA sequencing as previously reported (Liu et al., 2018). Reads from the V4 region amplicon were used for downstream analyses. Analysis of alpha diversity, including Faith’s phylogenetic diversity (Faith and Baker, 2006) and Shannon diversity, was performed using alpha_diversity.py in QIIME on rarefied data. For microbiome studies, principal coordinate analysis was performed using ANOSIM with 999 permutations. Relative OTU abundance data were input into LEfSe to determine biomarkers with significant linear discriminant analysis effect size (Segata et al., 2011). The determination of sample size and data analysis for animal studies followed the general guideline of Festing and Altman (Festing and Altman, 2002). Based on the law of diminishing returns, Mead recommended that a degree of freedom (DF) of 10–20 associated with error term in an ANOVA will be adequate to estimate preliminary information (Mead, 1988).

Transcriptomic profiling and analysis

For mouse tissue, full-thickness small intestine tissue was preserved in RNAlater. RNA was isolated using Qiagen RNeasy Mini kit (#74104), according to the kit protocol. Total RNA was quantified using the Quant-iT^™ RiboGreen® RNA Assay Kit (ThermoFisher; #R11490) and normalized to 4ng/μl. For human tissue, unstained slides from ileal resection margin were used to isolate RNA using RNeasy FFPE Kit (QIAGEN; # 73504) as described (VanDussen et al., 2018). Transcriptomic profiling was performed at the Genome Technology Access Center (GTAC) at Washington University, which has performed microarray (Liu et al., 2018) and RNA-seq before. RNA library preparation was performed with RiboZero and sequencing performed on an Illumina HiSeq 2500. RNA-seq reads were aligned to the GRCm38.76 assembly from Ensembl with STAR version 2.0.4 (Dobin et al., 2013). Gene counts were derived from the number of uniquely aligned unambiguous reads by Subread:featureCount version 1.4.5. Transcript counts were produced by Sailfish version 0.6.3. Sequencing performance was assessed for the total number of aligned reads, total number of uniquely aligned reads, genes and transcripts detected, ribosomal fraction and known junction saturation and read distribution over known gene models with RSeQC version 2.3. The results of the RNA-seq were further analyzed using the CompBio software package (PERCAYAI Inc., www.percayai.com/compbio). CompBio performs a literature analysis to identify relevant biological processes and pathways represented by the differentially expressed entities (genes, proteins, miRNA’s, or metabolites). This is accomplished with an automated Biological Knowledge Generation Engine that extracts all abstracts from PubMed that reference entities of interest (or their synonyms), using contextual language processing and a biological language dictionary that is not restricted to fixed pathway and ontology knowledge bases. Conditional probability analysis calculated the statistical enrichment of biological concepts (processes/pathways) over those that occur by random sampling. Related concepts built from the list of differentially expressed entities were further clustered into higher-level themes (e.g., biological pathways/processes, cell types and structures, etc.).

For analysis of the effect of FXR agonist on intestinal signaling, we used a publically-available transcriptomic dataset (GSE 74101), based on a published study testing FXR agonist PX20606 on intestine (de Boer et al., 2017). Expression of selected genes were retrieved and compared. Adjusted P values used in the original analysis were used.

RNA isolation and quantitative RT-PCR

To confirm the expression of selected genes, total RNA was purified from samples using the RNeasy Mini Kit (Qiagen #74104). cDNA was synthesized using the iScript^™ Reverse Transcription Supermix (Bio-Rad #1708840). qRT-PCR was performed using TB Green Advantage qPCR Premix (Clontech #639676) with Applied Biosystems QuantStudio 3 and the primer sequences listed in STAR Methods. Expression levels of each target gene were normalized to Gapdh, which was similarly expressed in all samples, and analyzed using the ΔΔCT method.

Bile acid measurement

Tissues stored at −80°C were homogenized by sonication (Branson Sonifier 450, output 2.5, 50% duty cycle, 10– 20 sec) in 10 volumes of 50% MeOH in water. Homogenates were centrifuged (13,000 × g, 10 min, 4°C), and cleared supernatants were transferred to new tubes. One third volume of chloroform was added to the supernatant, mixed (vigorous shaking for 10 sec), and centrifuged (13,000 × g, 5 min, 4°C). The aqueous phase was transferred to a new tube and the chloroform addition was repeated and centrifuged again. After the second chloroform treatment, the clear aqueous solution (100 μl) was dried under vacuum at room temperature and stored in −80°C until analysis.

On the day of analysis, lyophilized samples were reconstituted with 50 μl of 50% MeOH in water, vortexed, and then centrifuged (13,000 × g, 5 min, 4°C). Clear supernatants were transferred to HPLC vials and 10 μl of each sample was used for the analysis. Bile acids were measured using LC-MS (Agilent 1290 LC and 6470 triple quadrupole). Samples and the series of deoxycholic acid and lithocholic acid dilutions (0 – 10 ng/ml in 50% MeOH in water, Sigma-Aldrich) were injected into HILIC HPLC column (Kinetex 2.6 μm HILIC, 150 × 2.1 mm) with mobile buffer (30% MeOH, 70% 5mM ammonium formate) and eluted by the MeOH gradient (0–2min 30% MeOH, 2–6min 30–50% MeOH, 6–8min 50–80% MeOH, 8–10min 80% MeOH, 10–15min 80–30% MeOH, and 15–25min 30% MeOH). Bile acids were detected using single ion monitoring mode (m/z = 391.3 for deoxycholic acid and m/z = 375.3 for lithocholic acid with fragmentation voltage 240V) and quantified using the standard curve.

Type I IFN bioactivity assay

Levels of biologically active type I IFN in serum were determined using an encephalomyocarditis virus (EMCV) L929 cell cytopathic effect bioassay as described previously (Sheehan et al., 2006; Sheehan et al., 2015). Briefly, serial dilutions of sera in DMEM containing 10% FBS were applied in duplicate to L929 cells (2 × 10⁴ cells/well) in 96-well plates. Following incubation for 14 hours at 37°C, cells were infected with EMCV diluted in DMEM containing 2% FBS at a MOI of 7. At 8 hours post-infection, IFN-mediated protection was assayed using CellTiter 96 aqueous cell proliferation assay as per the manufacturer’s protocol (Promega). IFN levels were calculated from an IFN-β standard curve. To ensure protection from EMCV-induced death was due to type I IFN activity, L929 cells also were pretreated with anti-Ifnar1 mAb (MAR-1–5A3, 50 ug/mL) prior to the addition of serum samples.

QUANTIFICATION AND STATISTICAL ANALYSIS

For analysis between two groups, a nonparametric t test was used. For analysis between more than two groups, such as comparing different bile acids or different genotype and diet exposure combinations, Kruskal-Wallis tests followed by Dunn’s tests between groups was performed. Nonparametric statistical tests were selected based on the data distribution. All tests were two-tailed and a P value of <0.05 was considered significant. Data were plotted and analyzed using GraphPad Prism (version 9.0) and SAS version 9.4 (SAS Institute, Cary, North Carolina). All data represent mean ± SD. All statistical details of the experiments can be found in the figure legends, including the statistical tests used and exact value of n. n represents exact number of animals for in vivo studies and exact numbers of biological repeats for in vitro studies.

Highlights:

• Diet-induced obesity results in Paneth cell dysfunction in humans and mice.

• Consumption of Western diet leads to Clostridium-mediated deoxycholic acid conversion.

• Deoxycholic acid activates both FXR and type I interferon (IFN) pathways.

• Both FXR and type I IFN pathways are essential in triggering Paneth cell defects.