Phenotyping of Fecal Microbiota of Winnie, a Rodent Model of Spontaneous Chronic Colitis, Reveals Specific Metabolic, Genotoxic, and Pro-inflammatory Properties

The human gut hosts a complex and dynamically variable microbial community of trillions of microbial cells, which is shaped by host genetics and environmental factors [1]. Over the past 20 years, the number of studies investigating the association between onsets of inflammatory bowel diseases (IBD), comprising ulcerative colitis (UC) and Crohn’s disease (CD), and intestinal microbiota has undergone exponential growth, supported by the availability and large accessibility of ribosomal RNA gene-based metabarcoding and high-throughput metagenomics techniques [2]. While these studies have clearly demonstrated that IBD is associated with dramatic changes in the overall structure of the gut microbiota, specific changes in the gut microbiome and associated inflammatory and metabolic effects on the host in CD and UC are not well-defined [3]. An extensive unbiased meta-analysis of the gut microbiome data from five different IBD patient cohorts from five different countries has recently provided evidence that, while at higher phylogenetic levels gut microbiota community between healthy individuals and IBD patients is highly conserved, at or below the order level in the taxonomic rank, significant disease-specific alterations can be found. In addition, differential enrichment of microbial metabolic pathways in CD and UC, which included enriched pathways related to amino acid and glycan biosynthesis and metabolism, as well as other metabolic pathways, was reported [4].

One of the main obstacles to our understanding of how the gut microbiota affects the health of the host is our scarce information on evolutionary dynamics of the human gut microbiota even in healthy subjects, although its ecological and taxonomic properties have been extensively studied. Recent studies unraveled that the intestinal microbial populations are characterized by extensive longitudinal dynamics, and that, within-host adaptive evolution, these can affect resilience and long-term persistence [5, 6]. These studies open unanswered questions on the role of host-specific factors (such as diet, medication use, and host genetics), and inter-microbial interactions (metabolic interactions, spatial organization, communication signals) in microbial population dynamics. Also, they underlie the need of large longitudinal studies with all relevant information (diet, health status, medication use, weight gain or loss, travel, socioeconomic factors, and host genetics), and approaches for functional and strain-level profiling of the intestinal microbial populations [7]. Moreover, in vitro and in vivo model study systems, such as 3D cell cultures or mouse models, are required to test experimentally, under controlled conditions, working hypotheses mostly deriving from observational studies.

Although more than 60 experimental animal models are currently available for the study of IBD, including chemically induced disease models, congenic mice, models based on immune cell transfer, most of these models induce acute colitis and only a few chronic intestinal inflammations [8]. Winnie, a mouse (C57BL/6 background) carrying a missense mutation in the MUC2 mucin gene, is a valuable model for IBD [9, 10]. In humans, expression of MUC2 is reduced or depleted in ileal mucosa close to the ulcer margins in CD [11], and MUC2 production is also reduced in active UC [12]. MUC2 mucin, which is the main component of the intestinal mucus being present in the small and large intestine and forming the mucus skeleton, is the best characterized secreted mucin of the gastrointestinal tract [13]. In Winnie mice, the intestinal mucus is not firmly compacted in a tight inner layer, becoming permissive for host-microbial interaction [14]. Due to the primary epithelial defect, mice develop chronic intestinal inflammation, with signs of mucosal damage including thickening of muscle and mucosal layers, goblet cell loss, increased intestinal permeability, increased CD45-immunoreactivity in the distal colon, greatly enhanced susceptibility to luminal inflammation-inducing toxins, alteration of innervation in the distal colon, and symptoms of diarrhea, ulcerations, and rectal bleeding similar to those in IBD [14]. Signs and symptoms in Winnie mice have multiple similarities with those observed in patients with UC. Mice develop mild and progressive spontaneous inflammation in the distal colon from 6 weeks of age and a severe colitis starts to be detected after 16 weeks of age [15].

The missense mutation in the MUC2 mucin gene promotes dysbiosis in Winnie mice [15]. A recent study compared the gut microbiota of homozygous mutant mice (Winnie^−/−) with their wild-type littermates demonstrating that, although weaned from the same heterozygote mothers (Winnie^+/−) and caged separately according to genotype, dysbiosis in Winnie mice was established at 4 weeks of age, before histologic evidence of gut inflammatory changes [15]. Major hallmarks of Winnie mice microbiota, as compared with wild-type mice, were greater abundance of Bacteroidetes and Verrucomicrobia, lower abundance of Deferribacteres and Proteobacteria, and, at the species level, a greater abundance of Akkermansia muciniphila in 4-week-old mice [15].

In this study, we aimed to answer the question of whether the intestinal environment of Winnie, genetically determined by MUC2 mutation, can select an intestinal microbial community characterized by specific pro-inflammatory properties and metabolic features, which could imply a direct involvement in the pathogenesis of the chronic intestinal inflammation. This question was addressed using a variety of in vitro approaches for gut microbiota functional characterization including cell cultures and co-culture models for evaluation of geno-cytotoxic and pro-inflammatory properties, and cell-based phenotypic testing for metabolic profiling of the intestinal microbial communities.

The global burden of IBD, including UC and CD, can be seen in the context of current epidemiological transition [19]. Collectively, these chronic relapsing disorders of the intestinal tract, which can result in a range of debilitating symptoms including abdominal pain, diarrhea, vomiting and rectal bleeding, reduced quality of life, and physical, emotional, social and family problems, affect more than 2 million Europeans and 1.5 million North Americans, and about 6–8 million people are estimated to be affected by IBD worldwide [20]. While in Western countries the incidence of IBD has stabilized, and prevalence is increasing as a result of improved survival, in newly industrialized countries in South America, Eastern Europe, Asia, and Africa, the incidence is rapidly increasing even though the prevalence is still low [21]. The rising incidence in newly industrialized countries parallels with westernization of diet and culture, specifically with an increased intake of processed food, refined sugars, dairy, and less plant-based fibers [22].

Although the exact etiopathological mechanism remains elusive, it is generally accepted that IBD is the result of an inappropriate immune response against environmental factors, including luminal and microbial antigens, in genetically predisposed patients [23]. In the past, a large number of genome-wide association studies has led to the identification of more than 200 polymorphic loci genetically associated with IBD, some of which are also associated with other immune-mediated disorders, most notably with ankylosing spondylitis and psoriasis, and a considerable overlap between susceptibility loci for IBD and mycobacterial infection [24]. Most risk variants are shared across diverse ethnical groups, apart from several major European risk variants, which are not present in eastern Asians, including IL23R, and NOD2 variants that, notably, affect the interaction with the gut bacteria [24]. Despite substantial investments, these genetic studies have not yet brought health benefits through the identification of either genetic risk factors of disease susceptibility or prognostic markers or tools for predicting the response to therapy [25]. However, many studies have highlighted the importance of genetic traits that affect inflammation and the interaction with the intestinal microbiota [25]. This has shifted attention to the gut microbiota as a major player in IBD, the element on which host genetics and the environmental cues actually converge [26].

In this framework, the present study aimed to evaluate if and how a host genetically determined intestinal environment, such as that of the Winnie mouse with the MUC2 mutation, could select an intestinal microbial community characterized by specific pro-inflammatory properties and metabolic features, which may be directly implicated in IBD pathogenesis. It should be noted that in vitro approaches were mostly used in this study, which, due to their reductionistic nature and limitations (including (i) treatment and analysis of the microbiota samples under aerobic conditions; (ii) defined cell infection systems; (iii) limited set of metabolic, geno-cytotoxic, and pro-inflammatory markers), place caveat in the final interpretation of biological relevance and translational relevance of findings. Nonetheless, these approaches have provided some new insights that will certainly need to be confirmed in more appropriate in vivo settings as discussed below.

We started with 16S rRNA gene metabarcoding analysis that revealed substantial differences in the structure of gut bacterial communities from 4- to 16-week-old WN stool samples compared to parental WT mice (Fig. 1). It is worth to note that, at the genus level and, even more so, at the species level, the greatest differences concerned the less represented taxa (Fig. 6A, B). This situation can be well exemplified by Alistipes. We found Alistipes onderdonkii slightly more abundant in WN, Alistipes senegalensis less abundant in WN, while both Alistipes obesi and Alistipes shahii were much more abundant in WN. The different abundance of Alistipes species in WN and WT samples could be linked to different functional properties of each species in the relationship with the host. Alistipes is a relatively new sub-branch genus of the Bacteroidetes phylum with contrasting roles in host–pathogen interaction [27]. Alistipes has been involved in the progression of liver disease by affecting bile acids and short-chain fatty acid metabolism, progression of cardiovascular disease and hypertension, modulation of gut inflammation and immune response in cancer, and mental health [27]. On one hand, there is some evidence that Alistipes may protect against liver fibrosis, colitis, and cardiovascular disease, as well as against cancer immunotherapy by modulating the tumor microenvironment. On the other hand, several studies suggest that Alistipes may be pathogenic in colorectal cancer and may be associated with mental signs of depression [28]. These different behaviors may be due to specific properties of each species, although it is also conceivable that different Alistipes species may have different behaviors with respect to host nutrition and health depending on the host and host health. Noteworthy, among the two species much more abundant in WN than in WT, Alistipes obesi is the only motile species within the genus Alistipes, while Alistipes shahii has a clearly defined pro-inflammatory activity by TLR4-priming/TNF production that may increase local inflammation but may also lead to beneficial immunomodulation in cancer treatment [27, 28].

Compared to WT, WN samples were also more enriched in cellulosolytic genera including Ruminococcus, Eubacterium, and Lachnospiraceae incertae sedis (i.s.) (Fig. 1), consistently with Biolog EcoPlate data showing that D-Cellobiose was better utilized by WN microbes (Fig. 2). Ruminococcus, Eubacterium, and Lachnospiraceae incertae sedis (i.s.) were, respectively, 25-, 20-, and fivefold more abundant in WN than in WT samples (Fig. 6A). This situation is reflected at the level of species with a prevalence of Clostridium saccharolyticum, Eubacterium coprostanoligenes, and Eubacterium hallii in WN samples (Fig. 6B). The family of Lachnospiraceae and the genus Eubacterium include major butyrate-producers capable of utilizing complex carbohydrates such as D-Cellobiose for growth [29, 30]. Flintibacter butyricus, which can produce butyrate from carbohydrates, was also enriched in WN samples [31]. Among Eubacterium species, it may also be worth to note the ability of E. coprostanoligenes to carry out the conversion of cholesterol to the poorly absorbed sterol coprostanol [32]. It is rather surprising to find prevalence in WN mice of cellulolytic and butyrate-producing bacteria, generally associated with a diet based on vegetables rich in fibers and a healthy state of the intestinal microenvironment [33‐35]. Wide evidence emphasized that butyrate-producing bacteria establish positive metabolic interactions with the intestinal cells. In fact, butyrate is crucial to maintain a healthy gut as it fuels enterocytes, signals to the host and is a critical mediator of the colonic inflammatory response, and the number of butyrate-producing bacteria is significantly reduced in a variety of diseases, including IBD and diabetes [36].

Apparently, this finding may be in contrast with the geno-cytotoxic and pro-inflammatory potential of WN mouse microbes compared to those of the WT mouse (Fig. 6). Our results suggest that, regardless of the MUC2 mutation, the Winnie gut microbiota seems to be sufficient to elicit an inflammatory response by showing for the first time the genes involved in the aforementioned response in two different in vitro systems. This finding is consistent with a recent study showing that Winnie colitis was substantially reduced although not abolished under germ-free conditions [37]. Furthermore, intestinal organoids obtained from Winnie mice retain the chronic intestinal inflammatory features characteristic of the parental tissue [38]. Despite the differences with the results obtained with Caco-2 cells seeded on plastic wells, also in the co-culture system, WN stool sample upregulated in Caco-2 cells genes mostly related to DNA damage repair, as compared to WT stool sample. WN microbes also upregulated several genes that are associated with pathogen-associated molecular patterns (PAMPs) and inflammatory response (TLR3, TLR4, IFNG, NFKB1, and TNF), while they downregulated genes mostly linked to cell proliferation and differentiation (Fig. 5).

Of note, the intestinal immune system is extremely dynamic, adapting the immune response to mucosal specific factors able to educate the intestinal immune cells [39]. The content of the intestinal lumen, including microbial communities, food-derived antigens, and various metabolites, directly influences the function of innate and adaptive immune cells in a cascade of events finally modulating host immune homeostasis [40]. Intestinal dysfunctions, particularly those involving TLRs, IFNG, NFKB1, and TNF, have established pathogenic roles in several mouse models of colitis and in the onset of human IBD [41, 42]. Lipopolysaccharide (LPS) signaling though TLR4, which requires the accessory molecules CD14, PBP, and MD-2, plays a key role in the pathogenesis of IBD. Under physiological conditions, the expression of this receptor complex is generally low in intestinal cells, but its expression is significantly upregulated by various cell types in both nonactive and active IBD and in experimental mouse models of colitis [43‐46]. Upregulation of MD-2/TLR4 may also result from stimulation of ligands other than LPS, and, furthermore, interferon gamma (IFN-g) and tumor necrosis factor alpha (TNF-a), which play significant roles in triggering IBD, have been found to upregulate intestinal epithelial TLR4 expression in vitro [47, 48]. Importantly, there is evidence that in active IBD, variant alleles in TLR4 induce dysregulation, and “gain-of-function” mutations exhibit pro-inflammatory effects in response to physiological LPS concentrations. In particular, D299G polymorphism, which occurs at a frequency of about 6% in Caucasian populations, has been associated with CD and UC [49]. Thus, it has been suggested changes in the microbiota composition in the genetically susceptible host may result in aberrant TLR4 hyperresponsiveness of the intestinal mucosa in IBD patients [48]. Our data appear to be consistent with this hypothesis. Conversely, while there is strong evidence that TLR4 is significantly upregulated in both UC and CD, TLR3, which was upregulated by WN microbes, was shown to be significantly downregulated in intestinal epithelial cells in active CD but not in UC [50]. However, the combination of genetic variations in TLR3 and TLR7 has been reported to influence the severity of UC, and TLR3 and TLR4 agonists stimulate IFN-g production [51].

The enhanced geno-cytotoxic and pro-inflammatory potential of the Winnie gut microbiota may be related to an enhanced capability to establish physical (and metabolic) interactions with intestinal cells consistently with the results of bacteria-Caco-2 cell adhesion assay (Fig. 4). This may be the consequence of a process of microbial selection triggered by the alteration of the intestinal micro-environment caused by the missense mutation in the MUC2 gene. MUC2 mucin is present in Winnie but is not firmly compacted in a tight inner layer [10]. It is reasonable to assume that the lower compactness of intestinal mucin in the Winnie’s gut can favor the contact of microbes with the epithelial surface and, therefore, locally favor the selection of groups of microorganisms that interact closely with the intestinal cells. This pathogenetic hypothesis, which is focused on the role of two major players, i.e., the intestinal mucin metabolism and the local microbiota, in inducing or enhancing the inflammatory state through a self-sustaining process, could be tested in vivo by fecal microbiota transplant experiments (FMT) with germ-free or antibiotic-treated mice [52‐54]. If the biological relevance of our findings were confirmed with FMT experiments in mice, it would open up the possibility of extending these findings to human IBD.

There is strong evidence that implicates mucin biosynthesis and turnover in IBD pathogenesis. In IBD cases, structural changes in the glycoprotein nucleus of mucin, sulfation, and sialylation of mucin oligosaccharide residues and quantitative changes in mucin production are commonly observed [55]. These changes, which lead to a diminished functionality of the mucous barrier, have been associated to various genetic mutations or polymorphisms, but they may also be induced by exposure to environmental cues [56‐58]. Dynamic changes in mucin glycosylation pattern may be induced by the local microbiota, immune factors, and dietary components (fat, fiber, prebiotics, protein, and food additives) [59]. For instance, there is evidence that dietary fiber appears to improve intestinal barrier function in pigs through increased mucin production capacity (i.e., goblet cell numbers, MUC2 gene expression) and secretion (i.e., fecal mucin output) [60]. Mucin synthesis and mucin degradation are in equilibrium in healthy individuals. There is accumulating evidence that in UC patients, there is a decrease in the number of goblet cells and subsequently a decrease in mucin secretion, and an increase in the relative number of immature goblet cells that produce unstable hypo-glycosylated and/or hypo-sulfated mucin [55]. This results in infiltration of the luminal microbiota into the inner mucus layer, and increased contact with intestinal cells leading to inflammation. Our results with the Winnie mouse model, if confirmed by FMT experiments in mice, would allow us to add another piece to this picture: an increased contact between microbes and the epithelial surface (as a consequence of less compact and partially de-structured MUC2 mucin) may, in turn, favor the proliferation in the gut of bacteria whose lifestyle is determined by intimate interactions with the intestinal cells. The pro-inflammatory properties of these microbes may amplify and self-sustain the inflammatory process.

Were the biological relevance of our findings confirmed in vivo, it would also open up the possibility of using the proposed in vitro approaches for gut microbiota phenotyping in the clinical practice. Indeed, the gut microbiome exhibits high taxonomic compositional variation across individuals and extensive temporal dynamics [61], although gene composition and functional capabilities tend to be conserved. Indeed, there the gut microbiota is characterized by high degree of functional redundancy since phylogenetically unrelated taxa may carry out similar functions [62]. Moreover, remarkable functional variations exist even within the same species [63]. These limitations make it difficult to draw firm conclusions from the structural analysis of the microbiome using conventional rRNA-based methods, and highlight the need to develop new strategies and approaches, which can be suitably transferred to clinical practice. High-throughput metagenome sequencing, metatrascriptomics, single-cell sequencing, proteomics, and metabolomics are all very powerful technologies, but difficult to apply in clinical routine. Better applicability might have several methods that aim to comprehensively assess the phenotype of a microbial community such as those proposed in this study.

Our phenotypic profiling also shows more pronounced ability of WN microbes, as compared to WT microbes, to induce the expression of a number of DNA damage repair related genes in Caco-2 cells (Fig. 5). Thus, it would be interesting to understand if the bacteria that are numerically enriched in the WN sample have the capacity to produce genotoxic substances. It is well known that patients with IBD, including CD and UC, have an increased risk of developing colorectal cancer, and have, in their gut, higher levels of bacteria producing colibactin, a genotoxic peptide-polyketide hybrid molecule produced primarily by some E. coli strains [64]. This group of strains may also encode cyclomodulins, cycle inhibiting factors (CIF) capable of blocking mitosis independently of DNA damage and inducing apoptosis in exposed epithelial cell lines [65, 66]. Moreover, in colorectal cancer cases driven by chronic inflammation, mutations in TP53 gene are found to be an early event [67]. In addition to colibactin-positive and CIF-positive E. coli, other bacteria with potential genotoxic capacity have been associated with IBD, including several strains of Bacteroides fragilis producing metalloproteinase enterotoxin (BFT), and Enterococcus faecalis producing genotoxic peroxide [68]. Genotoxicity of BFT is the result of cross-talk with intestinal cells; it has been associated to the ability of BFT to induce an intense Th-17-mediated inflammatory response [69], but also to induce cleavage of E-cadherin, leading to increased paracellular permeability and the activation of β-Catenin and a subsequent increase in cell proliferation [70]. Other by- or end-products of microbial primary metabolism, such as H₂S produced by sulfate-oxidizing bacteria, and phenol, indoles and 4-cresol produced by bacteria that ferment aromatic amino acids, could also contribute to genotoxicity [68]. Finally, we must not forget the ability of some bacteria to produce secondary metabolites with powerful genotoxic capabilities, such as, for example, the enediynes which are produced by certain groups of actinomycetes, and which are used in the clinic as anticancer agents [71]. In this regard, it may be interesting to note greater abundance of gene clusters encoding secondary metabolites in the bacterial species that were enriched in WN compared to WT mice (Tables 1 and 2). Genome analysis of reference strains belonging to these species (Table S1) by antiSMASH revealed a substantial enrichment for cluster encoding arylpolyenes, linear azol(in)-e-containing peptides, non-ribosomal peptides, thiopeptides, ranthipeptides and RiPP recognition element (RRE) for peptide maturation in bacteria coming from Winnie mouse (Tables 1 and 2). Therefore, if this picture is correct, manipulation of the gut microbiota could represent an important prophylactic or therapeutic strategy in controlling IBD progression and colorectal cancer.

In this study, we have pursued two aims: (i) to propose a polyphasic approach for functional characterization of gut microbiota by using in vitro approaches, and (ii) to investigate the role of intestinal microbiota in the pathophysiology of IBD in a Winnie mouse model harboring MUC2 mutation. Firstly, we have characterized the structure of prokaryotic communities in WT and WN fecal microbiota by 16S rRNA gene metabarcoding. Then, the prokaryotic structural profiles were compared with the microbial metabolic profiles of WT and WN stool samples, outlined by Biolog EcoPlates phenotyping. Results emerging from structural analysis were consistent with metabolic data, demonstrating that Biolog EcoPlate technology is a reliable method to explore the metabolic potential of microbiota in animal fecal samples. Afterwards, in order to investigate the ability of fecal bacteria to interact with intestinal cell, a microbial adhesion assay was performed, using Caco-2 cell lines. Two different approaches were used to detect and quantify the number of bacteria adhering to Caco-2 cells, and both revealed that Winnie fecal microbiota had greater ability to interact with intestinal cells than WT. The increased capability of Winnie fecal microbiota to establish physical and metabolic interactions with intestinal cells compared to WT microbiota was accompanied by enhanced genotoxic, cytotoxic and pro-inflammatory potentials, as determined in two different in vitro cellular systems that were used to screen the mRNA expression pattern of 43 marker genes, by customized qPCR arrays. In conclusion, we both provide (i) a reliable method for phenotyping fecal microbial communities and (ii) in vitro evidence that the Winnie mouse fecal microbiota, selected by host genetic alterations, may elicits specific pro-inflammatory, genotoxic, and cytotoxic responses in the intestinal tract, suggesting a direct role of intestinal microbial community in IBD pathogenesis.