Diet-driven differential response of Akkermansia muciniphila modulates pathogen susceptibility

The erosion of the colonic mucus layer by a dietary fiber-deprived gut microbiota results in heightened susceptibility to an attaching and effacing pathogen, Citrobacter rodentium. Nevertheless, the questions of whether and how specific mucolytic bacteria aid in the increased pathogen susceptibility remain unexplored. Here, we leverage a functionally characterized, 14-member synthetic human microbiota in gnotobiotic mice to deduce which bacteria and functions are responsible for the pathogen susceptibility. Using strain dropouts of mucolytic bacteria from the community, we show that Akkermansia muciniphila renders the host more vulnerable to the mucosal pathogen during fiber deprivation. However, the presence of A. muciniphila reduces pathogen load on a fiber-sufficient diet, highlighting the context-dependent beneficial effects of this mucin specialist. The enhanced pathogen susceptibility is not owing to altered host immune or pathogen responses, but is driven by a combination of increased mucus penetrability and altered activities of A. muciniphila and other community members. Our study provides novel insights into the mechanisms of how discrete functional responses of the same mucolytic bacterium either resist or enhance enteric pathogen susceptibility.


Introduction
The intestinal mucus layer is a protective and lubricating barrier of mucin glycoproteins covering the intestinal epithelium that is an integral part of the mucosal immune system (Johansson & Hansson, 2016).
Among the many functions of the gut mucus layer is the protection against enteric pathogens (Martens et al, 2018).The colonic mucus layer is mainly composed of mucin-2 (MUC2) glycoproteins, which are dynamically secreted by goblet cells in a greatly condensed form to generate a nearly impenetrable, netlike structure covering the epithelium (Johansson et al, 2008).Some members of the gut microbiota possess enzymatic capabilities to degrade the complex mucin glycoproteins (Martens et al, 2018) and constantly feed on the outer edges of this structure, resulting in a progressively looser layer further in the lumen (Johansson et al, 2008).Thus, the mucus layer is an important nutrient source for the gut microbiota (Schroeder, 2019;Luis & Hansson, 2023) and a critical interface between the microbiota and immune system (Paone & Cani, 2020).
The observation that a dietary fiber-deprived gut microbiota deteriorates the colonic mucus barrier has been replicated across labs, among mice harboring either complex and simplified microbial communities (Desai et al, 2016;Schroeder et al, 2018;Riva et al, 2019;Neumann et al, 2021;Parrish et al, 2023).
Moreover, we have shown previously that a fiber-deprived gut microbiota enhances susceptibility to an attaching and effacing, rodent mucosal pathogen, Citrobacter rodentium (Neumann et al, 2021;Desai et al, 2016), which is an important model pathogen for human enteropathogenic and enterohemorrhagic Escherichia coli (Mullineaux-Sanders et al, 2019).It is striking that a typically self-limiting pathogen such as C. rodentium leads to a lethal colitis in fiber-deprived mice with a microbiota (Neumann et al, 2021;Desai et al, 2016).This lethal colitis was not observed in fiber-deprived, germ-free mice that also exhibit a thinner colonic mucus barrier and a similar invasion of the pathogen in the mucosa as susceptible mice (Desai et al, 2016).Thus, it is evident that the pathogen alone is incapable of driving the lethal phenotype and an interaction of the microbiota with the pathogen plays a critical role.

4
Nevertheless, the mechanisms connecting the excessive mucus degradation, increased pathogen susceptibility, and specific microbial members remain unknown.To elucidate which specific bacteria or functions are responsible for increasing pathogen susceptibility in the absence of fiber, here we leverage a community ecology approach by performing multiple strain dropout experiments using a functionallycharacterized 14-member synthetic human gut microbiota in gnotobiotic mice.We identify a commensal microbe, Akkermansia muciniphila, to play a vital role in modulating infection dynamics of C. rodentium in a diet-dependent manner.Our study provides valuable lessons to better understand the pathogenesis mechanisms of an attaching and effacing pathogen by incorporating a community ecology perspective.

Stain dropout approach provides functionally relevant communities to study enteric infection
We colonized age-matched, germ-free Swiss Webster mice with different synthetic microbiota (SM) communities and maintained these mice on a standard lab chow, which we label a fiber-rich (FR) diet (Fig. 1a).Based on our previous work (Desai et al, 2016), we sought to tease apart how and which mucindegrading bacteria are directly responsible for the increased susceptibility to C. rodentium during dietaryfiber deprivation.We designed four new SMs by dropping out different mucolytic bacteria from the 14SM community (Desai et al, 2016) (Fig. 1b).The 10SM contains none of 4 mucolytic bacteria present in the 14SM: Akkermansia muciniphila, Barnesiella intestinihominis, Bacteroides caccae and Bacteroides thetaiotaomicron (Fig. 1b).The 11SM contains all members of the 10SM plus A. muciniphila, a mucin specialist bacterium that can utilize mucin O-glycans as the sole carbon and nitrogen source (Derrien et al, 2004;Desai et al, 2016).The 12SM consists of the 10SM plus both B. caccae and B. thetaiotaomicron, which are mucin generalists, meaning they can grow on a number of polysaccharides, including dietary fibers and mucin O-glycans (Desai et al, 2016) (Fig. 1b).Finally, the 13SM contains both mucin generalists in addition to the mucin specialist B. intestinihominis; therefore, only A. muciniphila is excluded from the community.

5
Initial colonization of the each SM community was verified by qPCR using primers specific to each of the 14 bacterial strains as previously described (Desai et al, 2016;Steimle et al, 2021).Mice were maintained for 14 days on the FR diet before approximately half of the mice in each experimental group were switched to a fiber-free (FF) diet (Fig. 1a); the mice were maintained on the respective FR and FF diets for forty days.Forty days after the diet switch, a subset of mice from each group was sacrificed to collect the pre-infection readouts (Fig. 1a).The relative abundances of different members of the SM were determined Even without the mucolytic bacteria, the SM form a fairly stable community (Fig. EV1, Dataset EV1), which is further emphasized by examining the fold-changes of the individual strains between diets just prior to infection (Fig. 1e).We observed similar changes between diets across different SM combinations.
Based on the microbial abundance data, it seems that of the two mucin-degrading bacteria, A. muciniphila and B. caccae are favored, or at least not adversely impacted, under FF conditions (Fig. 1e).In contrast, B. thetaiotaomicron and B. intestinihominis seem to be outcompeted by the other mucin-degrading bacteria under FF conditions, resulting in a relatively lower abundance under these conditions (Fig. 1e).
Among the non-mucin-degrading bacteria, Desulfovibrio piger, Clostridium symbiosum and Collinsella aerofaciens also thrived under FF conditions (Fig. 1e).Of these, C. aerofaciens has been associated with a number of inflammatory diseases, in particular rheumatoid arthritis (Chen et al, 2016a).It is also known to be associated with D. piger as it produces lactate, H 2 and formate which serve as substrates for D. piger (Rey et al, 2013) and contribute to luminal H 2 S, which can damage the mucosa at high concentrations (Blachier et al, 2021).
In order to assess the presence of any baseline inflammation, which may be induced by the diet shift, we measured pre-infection fecal lipocalin-2 (LCN-2) concentration, which is a marker of low-grade inflammation (Fig. 1f) (Chassaing et al, 2012).The 14SM-and, to a lesser extent, the 11SM groupshowed statistically meaningful increases in LCN-2 on the FF diet (Fig. 1f).Meanwhile, only the 14SM and 13SM groups showed a significant shortening of the colon on the FF diet, which is another gross indicator of inflammation (Fig. 1g).It is worth noting that the 10SM mice exhibited shorter colons under both diets, underscoring that mucin-degrading microbes can reduce inflammation and promote epithelial barrier integrity in the context of a fiber-sufficient diet; our in vivo results provide evidence for the earlier data from in vitro models that mucin-degrading bacteria improve markers of barrier integrity (Pan et al, 2022).While direct observation of all mice showed no overt signs of disease, these results suggest that dietary fiber deprivation has the potential to induce low-grade inflammation, which appears to be further dependent on the community composition.
Considering the vital role of short-chain fatty acids (SCFAs) in promoting intestinal homeostasis (van der Hee & Wells, 2021) and susceptibility to C. rodentium (Osbelt et al, 2020;An et al, 2021), we suspected that possible shifts in the microbial SCFA production might alter responses to C. rodentium after infection.Although no differences were observed for acetate between the two diets in all SM groups, butyrate concentrations were significantly different only for the 14SM dietary groups (Fig. 1h).Interestingly, among the two groups containing A. muciniphila (14SM and 11SM), concentrations of cecal propionate-a metabolite produced by A. muciniphila (Derrien et al, 2004)-were reduced in FF groups (Fig. 1h).This observation is in contrast to the results of the complete community lacking A. muciniphila (13SM).These results suggest that the metabolism of this mucin specialist is altered in FF mice compared to FR mice.Importantly, our strain dropout approach successfully provides functionally relevant stable communities to possibly identify causal mucin-degrading microbes that might aid the increased susceptibility to C. rodentium in fiber-deprived mice.

A. muciniphila is responsible for increased C. rodentium susceptibility during fiber deprivation
The remaining mice (after those sacrificed for pre-infection readouts) were infected with approximately 10 9 CFU of C. rodentium (Fig. 1a).The C. rodentium-infected mice were monitored for up to 10 days postinfection (DPI) before they were sacrificed to collect final readouts (Fig. 1a).Overall, we could reproduce the higher pathogen susceptibility phenotype previously described in the 14SM colonized mice after 40 days of feeding on an FF diet (Desai et al, 2016).Comparing C. rodentium loads between colonization states, within the same diet, we reached an interesting conclusion that A. muciniphila tends to confer a resistance to infection, but only under fiber-rich conditions and in the absence of other mucin degraders in the community, that is, in the 11SM condition (Fig. 2a,b).Yet when fed an FF diet, the 11SM-colonized mice showed increased C. rodentium levels, as observed in the 14SM group (Fig. 2c,d).Meanwhile, the 12SM and 10SM groups, which lacked A. muciniphila, showed a trend towards elevated pathogen loads on the FF diet, but which was non-significant in most cases (Fig. 2d), potentially reflecting a generalized impact of diet on the susceptibility.Interestingly, in the 13SM community, lacking only A. muciniphila, the FF diet seemed to be marginally protective with overall lower pathogen loads (AUC) among the 13SM FF mice compared to the 14SM and 11SM groups (adjusted p = 0.0033 and 0.049, respectively, multiple Kruskal-Wallis tests with Benjamini-Hochberg correction), but not compared to the 12SM FF and 10SM FF groups fed the same diet (Fig. 2c).
The maximum weight loss following infection and fecal LCN-2 at the end of the infection were mostly in alignment with the pathogen loads, with the 14SM and 11SM groups showing significant weight loss and increased LCN-2 levels on FF diet (Fig. 2e,f).Surprisingly, after infection, the 10SM-colonized mice exhibited increased weight loss and higher levels of LCN-2 during fiber deprivation (Fig. 2e,f).As this is not in line with their markers of inflammation prior to infection (Fig. 1f,g) or their diet-dependent differences in pathogen load (Fig. 2c,d), it suggests that a complete absence of mucin-degrading bacteria is also a precursor to increased pathogen-induced inflammation in the context of a fiber-deficient diet.The colon lengths after the infection, however, were reflective of the pre-infection status, with the 14SM, 13SM, and 12SM groups showing a significant shortening on the FF diet (Fig. 2g).Histological analysis of the cecal tissue corroborated the increased inflammation during fiber deprivation in the 14SM and, to a lesser extent, in the 10SM group; however, we also observed elevated disease scores by histology in the 13SM group on the FF diet, relative to the FR diet (Fig. 2h).Overall, our data indicate that, during fiberdeprivation, the increased susceptibility to C. rodentium in the presence of mucolytic bacteria is exacerbated by a single mucin specialist, A. muciniphila.Furthermore, the presence of all other mucindegrading bacteria seems to amplify the physiological impact of A. muciniphila, while the lack of A. muciniphila is sufficient to avoid the excessive vulnerability to C. rodentium observed on the FF diet.
As A. muciniphila abundance is indirectly modulated by the presence of dietary fiber, we supplemented the FF diet with 7.5% acetylated galactoglucomannan (AcGGM), a purified plant polysaccharide (Fig EV2a).Although AcGGM is degraded by Roseburia intestinalis, Faecalibacterium prausnitzii, and Eubacterium rectale (La Rosa et al, 2019), three butyrogenic commensals present in the SM communities, we did not detect an increase in the relative abundances of these bacteria, but rather saw an expansion of In line with our earlier study suggesting that purified fibers do not alleviate the excessive microbial mucin foraging (Desai et al, 2016), the current results corroborate the conclusion that purified prebiotic supplementation is insufficient to prevent the higher pathogen susceptibility observed among FF-fed mice.This experiment serves as an important cautionary tale: the functional activities as predicted from single strains grown in monoculture do not necessarily translate into the same overall effect in a more complex community, therefore it is crucial to test prebiotic supplements using complex communities or defined consortia in vivo in order to validate their intended effect.
We then sought to determine whether the FF-diet led increased C. rodentium load in 14SM mice could also be seen with shorten feeding timelines, as we expected that the time spent on the FF diet might influence the detrimental impact on the host.Thus, we tested whether 5 or 20 days on a fiber-deprived diet could elicit a similar pathogen load.Overall, C. rodentium loads of the shorter FF feeding groups (5 and 20 days) were comparable to a 40-day FF diet feeding period (Fig EV2eg).Minor differences in pathogen loads are most likely attributable to the low sampling number of the 5-and 20-day groups (n=4 and n=3, respectively).These results suggest that the local microbiota effects, including mucus-microbe interactions, rather than the detrimental impact on the host owing to longer feeding periods (like 40 days), might play decisive roles in the increased pathogen susceptibility.

Host immune and pathogen transcript readouts suggest differential response of A. muciniphila behind increased susceptibility
As A. muciniphila has been studied for its potential immunomodulatory properties (Luo et al, 2021;Ansaldo et al, 2019;Ashrafian et al, 2021;Xie et al, 2023;Zhang et al, 2023), we supposed that the relative expansion of A. muciniphila on the FF diet might have repercussions on the host immune response, thus explaining the increased C. rodentium susceptibility.As the initial expansion of C. rodentium is heavily affected by dietary-fiber deprivation, we chose to assess the immune cell populations at 3 DPI using fluorescence-activated cell sorting (FACS).Additionally, we have previously observed that by 4 DPI, the pathogen successfully enters into the intestinal tissue (Desai et al, 2016), therefore we surmised that 3 DPI would allow us to capture the early innate immune responses known to be driven by C. rodentium (Mullineaux-Sanders et al, 2019) before the complete invasion of the pathogen.We also used the same time point (3 DPI) to study both host colonic-and microbial transcriptomes.In order to facilitate comparison with our previous work (Desai et al, 2016), we continued using a 40-day FF feeding period for this and subsequent experiments in the current study.
Overall, different immune cell populations showed similar trends under fiber deprivation, but nearly all comparisons were statistically non-significant after correction for multiple comparisons (Fig EV3a,b).
Exceptionally, the RORγt-positive T helper cell population (CD4 + RORγt + ) was significantly decreased in the 14SM FF diet group compared to the FR group, while it did not show a significant difference in the 13SM group (Fig. 3a).As Th17 cells play an essential role in the clearance of C. rodentium (Atarashi et al, 2015), these data point to a lack of Th17 induction as a possible culprit behind the increased pathogen susceptibility observed among the 14SM FF diet group compared to the 14SM FR.However, comparing 14SM and 13SM under the same diet revealed no statistically significant shifts in proportions of the immune cell populations assessed here, prompting us to investigate other facets of the host-microbiome axis for explanations of the altered susceptibility.
To further investigate the role of A. muciniphila in modulating pathogen susceptibility, we analyzed both mouse colonic tissue (3 DPI) and C. rodentium transcripts (3 DPI, from microbial metatranscriptomics data) in both diet groups of the 13SM and 14SM mice.As with the FACS data, there was considerable variability in the host transcriptomic data, which is why no distinct clustering by group could be observed in the PCA plot (Fig. 3b).In contrast, C. rodentium transcripts showed clear clustering based on diet (Fig. 3c).Analysis of the host transcript counts (Dataset EV2) showed only a limited number of genes differentially transcribed between the four groups (Dataset EV3).Within the 13SM group, 19 host genes were upregulated in the FR group and 8 in the FF group (Fig. 3d, Dataset EV3).Meanwhile, in the 14SM group, 9 host genes were upregulated during the FR diet and 13 during the FF diet (Fig. 3d, Dataset EV3).Only two transcripts, C4b-a complement factor involved in innate immune response elevated in FF-fed mice-and Klf6-a key regulator of pathogenic myeloid cell activation increased in FR-fed micewere significantly altered in both the 13 and 14SM groups (Dataset EV3).Evaluating the effects of the presence of A. muciniphila in the same fiber-deprived conditions, revealed an enrichment of transcripts involved in immunoglobulin production (Igkv12-89, Igkv1-110) and zinc homeostasis (Slc39a4) in the 14SM FF group, which are both associated with infection response, however it is unclear how increased transcription of these genes might lead to increased susceptibility to infection.Overall, the host transcriptomic profiles offered limited explanation as to why the FF-fed 14SM mice were more susceptible to C. rodentium infection.
Lacking evidence for altered susceptibility in the host transcriptome data, we next examined whether changes in the transcriptomic profile of C. rodentium (Dataset EV4) might provide greater insight.We identified numerous diet-specific changes in C. rodentium gene expression, however the presence of A. muciniphila within the same FF diet background (13SM FF vs 14SM FF) had no effect on C. rodentium transcription (Fig. 3e, Dataset EV5).While there are some intriguing observations in the transcript data of C. rodentium, in particular the transcription of a number of phage-type genes in the 13SM FR group, no transcripts of genes encoded on the LEE pathogenicity island were altered in the C. rodentium transcriptome under any conditions (Dataset EV5).
In summary, our host immune and transcriptome as well as pathogen transcriptome readouts did not provide clear clues about how A. muciniphila could be altering susceptibility to C. rodentium in the context of FF diet.Nevertheless, these results still hold importance, because they clarify that the pathogen and the host respond the same way in 13SM FF and 14SM FF mice, despite resulting in strikingly different susceptibility to the pathogen.Thus, these results point to the fact that the altered responses of A. muciniphila itself and also the accordingly altered responses of other community members likely play a critical role in enhancing susceptibility to C. rodentium in FF-fed mice.

Fiber-deprived A. muciniphila increases mucus penetrability and alters community activity
We have previously shown that during fiber deprivation, the 14SM community is characterized by a thinner mucus layer (Desai et al, 2016).In 40-day FF-fed, uninfected mice, we validated this phenotype in the current study and further show that there is a similar trend in the 13SM group, though the difference is more pronounced due to a generally higher mucus thickness among 13SM FR-fed mice (Fig. 4a).In addition to the thickness of the mucus layer, its integrity is important (Schroeder et al, 2018); thus, we used bacteria-sized beads (1 μm) to measure mucus penetrability (Fig. 4b).We observed a significant increase in the penetrability of the colonic mucus layer of the 14SM FF group compared to the FR group, but no difference between the 13SM groups (Fig. 4b), which suggests a potential mechanism by which A. muciniphila could increase the C. rodentium susceptibility under FF conditions.It bears noting that the presence of A. muciniphila strongly influences the gut mucus layer; accordingly, we decided to limit confounding factors by restricting our comparisons of the mucus penetrability to the FR and FF groups of a specific microbial composition (13SM or 14SM).These results suggest alterations in mucus integrity that can stem from changes in host mucin secretion or glycosylation or from altered degradation by the commensal microbiome.The importance of the mucus layer integrity compared to its thickness is further highlighted by our prior work in germ-free mice, which, despite possessing a thin mucus layer, exhibit a diet-independent resistance toward C. rodentium infection comparable to 14SM FR-fed mice (Desai et al, 2016).
To further investigate the role of A. muciniphila, we analyzed its microbial meta-transcriptome data in the infected mice (same data from which C. rodentium transcripts were analyzed in Fig. 3).A 12-fold higher proportion of transcripts mapped to A. muciniphila on the FF diet, compared to on the FR diet (Fig. 4c).
Focusing on mucin-targeting enzymes in particular, sialidase and β-N-acetylgalactosaminidase expression was significantly higher in FF mice compared to their FR counterparts (Fig. 4d).When comparing 14SM and 13SM FF-fed mice, however, the overall levels of transcripts mapping to mucin-targeting enzymes did not differ in the presence of A. muciniphila, as the niche was apparently taken up by mucin generalists such as B. caccae (Fig. 4d).Taking together, these results are important to support the conclusion that microbial over-foraging of the mucus layer alone is insufficient to explain the difference in C. rodentium susceptibility, as it is observed even in the group that shows resistance to infection (13SM FF).
Given that the only clear difference between the groups stemmed from the presence or absence of A. muciniphila, we returned to the metatranscriptome to identify significant changes in transcription within this taxon.Among the 97 differentially abundant transcripts mapping to A. muciniphila between FR and FF-fed mice, 21 are hypothetical proteins, that is, without a known function (Fig. 5a).Taking a more targeted approach, we noted that FF-fed mice exhibited an increase in transcripts that have been reported to have anti-inflammatory properties via induction of IL-10: UniRef90_B2UR41 or Amuc_1100 encoding a pili-like protein (Ottman et al, 2017) and UniRef90_A0A139TV36 encoding threonine--tRNA ligase (Kim et al, 2023) (Fig. 5b).While these are promising microbial-derived factors in the treatment of autoimmune diseases, it is possible that their immunosuppressive effects are deleterious in the context of enteropathogenic infection.
Finally, as elegantly demonstrated by Weiss et al. (Weiss et al, 2023), bacterial community dynamics are highly dependent on both the nutritional conditions and the presence or absence of other key consortium members.Overall, the total number of differentially expressed genes (DEGs) was lower between 13SM and 14SM mice fed the FF diet, than between the same SM fed the FR diet, highlighting the strong impact of diet on the activity of the overall community (Fig. 5c).Among FF-fed mice, Bacteroides spp., C. symbiosum, C. aerofaciens, and E. coli appeared most responsive to the absence of A. muciniphila, as detailed in Fig EV4 .A limitation of this analysis is that each locus in the genome is treated independently, even if they encode the same product.on the unstratified data and then visualized according to the stratified output (Fig. 5d).Here, we can see the altered transcriptional profiles in greater detail, particularly the contribution of transcripts from A. muciniphila, and note that many of the protein families correspond to functions to combat oxidative stress (PF00667, PF00724, PF10417), gluconeogenesis (PF00821, PF17297), or are involved in protein secretion (PF12951, PF00482) (Fig. 5d).
The increased susceptibility to C. rodentium infection observed in 14SM FF-fed mice cannot be attributed to either the diet or microbiome alone, but rather depends on both factors to produce a local environment that supports infection (Fig. 6).The host and pathogen both behave as expected-mounting an innate immune response against infection or expressing virulence factors to establish infection, respectively-regardless of the diet or SM.The decisive factor determining the outcome of this battle rests in the microbiome, which, unlike the host or pathogen, collectively alter their gene expression in a diet-and microbiome-dependent manner.Ultimately, in leveraging a strain dropout approach, we find that a single mucin-degrading bacterium, A. muciniphila, plays a pivotal role in modulating pathogen susceptibility in a diet-dependent manner (Fig. 6).

Discussion
Dietary fiber consumption is well known to confer a myriad of health benefits (Sonnenburg & Sonnenburg, 2014;Makki et al, 2018;Gill et al, 2021).Moreover, fiber consumption is indispensable to avoid a functional shift of the gut microbiota toward mucin degradation, which erodes the gut mucus layer and increases pathogen susceptibility (Desai et al, 2016;Neumann et al, 2021).By manipulating the presence of functionally relevant members of the microbiota, our work offers insights into the multifactorial and ecological nature of susceptibility to enteropathogenic infection, with the key finding that A. muciniphila increases susceptibility to infection in concert with other members of the microbiome, but only in the context of a fiber-deficient diet (Fig. 2, Fig. 6).Importantly, adding A. muciniphila as the sole mucindegrader to the synthetic microbiota appeared to be protective to C. rodentium colonization in the context of a fiber-sufficient diet.However, we also show that resistance to infection can be achieved on a fiberdeprived diet through removal of A. muciniphila, which contributes to excessive erosion of the gut mucus layer (Fig. 4b) and alters the activity of other gut community members (Fig. 4c,d).Our study highlights the power of a strain dropout approach within a defined consortium to elucidate possible mechanistic links in the microbiota-pathogen-host axis.
Intriguingly, a fiber-deprived microbiota-led reduced mucus barrier alone is insufficient to heighten susceptibility to an enteric pathogen, but the presence of a microbial biomarker species-that is a species whose presence or activity can be used to predict susceptibility to disease or disease course-is essential.In this study, A. muciniphila can be considered as a biomarker species since its presence or absence under fiber-deprived dietary conditions determines the course of C. rodentium infection.A fiber-deprived community harboring A. muciniphila exhibits increased penetrability of the mucus barrier, thereby likely increasing the influx of microbial antigens that might aid in the heightened susceptibility.
Furthermore, the beneficial aspects of mucolytic bacteria are also evident, as we noticed increased markers of pathogen susceptibility in the absence of all four mucolytic bacteria, possibly owing to immature mucus barrier function in the complete absence of mucolytic bacteria.
As A. muciniphila is regarded as a potential probiotic bacterium (Zhai et al, 2019;Ashrafian et al, 2021;Cani et al, 2022;Daniel et al, 2023), the effects of this bacterium on the activity of other members of the gut microbial community must be considered when designing probiotic therapies using this microbe.The opposing and intriguing observation that A. muciniphila can reduce the pathogen load in the presence of a fiber-sufficient diet and can exacerbate pathogen susceptibility in the absence of fiber, underscores the nuanced and widely overlooked concept that, depending on the dietary context, this bacterium can be either beneficial or detrimental.This observation echoes views of some other researchers that A. muciniphila's impact on health need to be rationally considered (Cirstea et al, 2018;Luo et al, 2022).Our microbial transcriptome analyses highlight that A. muciniphila behaves differently on two diets with many differentially transcribed genes encoding hypothetical proteins.A key question that remains unanswered is which of these differentially expressed gene products play an important role in altering the pathogen susceptibility, which relies in part on increasing knowledge of the structure and function of these hypothetical proteins.
A related aspect for future studies is to understand the connection between increased mucus penetrability and differentially transcribed genes of A. muciniphila.One possible explanation is that the increased expression of pili-like proteins (Ottman et al, 2017) and threonine--tRNA ligase (Kim et al, 2023) by A. muciniphila under fiber-deprived conditions might tune the immune system in a way that increases vulnerability to the pathogen.In this case of the aforementioned components, IL-10 induction might represent a deleterious immune response that ultimately contributes to the exacerbated pathogen susceptibility.Differences in the microbial compositions of laboratory mice can impact susceptibility to C. rodentium infection through altered production of SCFAs (Osbelt et al, 2020).Since we observed reduced propionate-a SCFA produced by A. muciniphila (Derrien et al, 2004)-in A. muciniphila-containing FFfed mice, it is possible that the altered metabolism of fiber-deprived A. muciniphila contributes to increased pathogen susceptibility.Altogether, the molecular mechanisms through which fiber-deprived A. Using such mutants might also help to understand which specific responses of A. muciniphila majorly impact the broader community dynamics, especially considering that the degradation of the gut mucus by mucolytic bacteria releases glycan residues that can be metabolized by either the mucolytic bacteria themselves or other members of the gut microbiota (Belzer et al, 2017;Crost et al, 2018;Raimondi et al, 2021).Changes in which mucosal sugars are released the presence or absence of A. muciniphila under fiber-deprived conditions could also play a role in the susceptibility, as it was recently shown that sialic acid-a sugar present in mucin glycoproteins-plays a vital role in aiding transition of C. rodentium from lumen to mucosa (Liang et al, 2023).Another recent study showed that dietary restriction of L-serine enhances mucin degradation by A. muciniphila and promotes encroachment of adherent-invasive Escherichia coli (AIEC) to the epithelial niche, where the pathogen acquires host L-serine from the epithelium to proliferate (Sugihara et al, 2022).Future studies are needed to determine whether a possible metabolic interaction between A. muciniphila and C. rodentium exists.
In line with a previous study that showed how high-fat diet impacts susceptibility to C. rodentium (An et al, 2021), our study highlights how microbial ecology and the presence or absence of certain key taxa impact disease susceptibility by altered dietary habits.The role of diet in this dynamic is particularly relevant given that dietary fiber intake in many industrialized countries remains below the recommended intake for adults of 25 g/day (Scientific Opinion on Dietary Reference Values for carbohydrates and dietary fibre, 2016).It is likely that A. muciniphila is just one of the potential biomarkers that can be used to predict susceptibility to enteric pathogen infection.Identifying additional such biomarker species and understanding how their role in disease susceptibility can be modified through the diet could prove to be an important tool to help lower the burden of human foodborne enteropathogenic infections-a looming challenge in the face of the dual threats of altered food supply systems and antimicrobial resistance (Willett et al, 2019).

Animal diets
In both animal facilities, the fiber-rich (FR) diet was an autoclaved rodent chow (LabDiet #5013, St. Louis, MO, USA).For experiments in the United States, the fiber-free (FF) diet was manufactured and irradiated by Envigo (TD.140343,Indianapolis, IN, USA) and is based on the Harlan.TD08810 diet (Envigo, Indianapolis, IN, USA) (Desai et al, 2016).The fiber-free diet used in the Luxembourg facility was manufactured and irradiated by SAFE Diets (Augy, France) based on the TD.140343 diet formulation used in the United States.The FF diet lacks dietary fiber, which has been compensated by an increase in glucose.The FF diet also contains crystalline cellulose, a polysaccharide that cannot be degraded by any member of the SM.The AcGGM diet was a modified version of the TD.140343 diet, manufactured and irradiated by Envigo (Indianapolis, IN, USA), which was supplemented with 7.5% of the dietary fiber acetylated galactoglucomannan (AcGGM).

Experimental design
Germ-free (GF), 6-to 8-week-old, age-matched, Swiss Webster mice were housed in iso-cages with maximum five animals of the same sex per cage.Light cycles consisted of 12 h of light and sterile water, with diets provided ad libitum.The experiments were performed in two different facilities, one facility in Luxembourg and the other in the United States.Results were reproduced independent of the location of the facility.Researchers were blinded to the treatment group for the histological disease scoring, SCFA quantification, and FACS analysis.Prior to and 14 days following the initial gavage, all mice were maintained on the FR diet.Two weeks after the gavage, half of the mice per SM group were switched the FF diet.Mice were maintained on their respective diets for up to 40 days and fecal samples were collected as represented in Fig. 1a.One group of mice was kept on the FF diet for 5 days (n=4) or 20 days (n=3), respectively, to determine the timeline to develop a susceptible phenotype (Fig EV2e).Aside from the shortened feeding period, these mice were treated identically to the mice under the 40-day feeding regimen.At the end of the feeding period, mice were either euthanized for pre-infection readouts or infected with ~10 9 CFUs of C. rodentium as previously described (Desai et al, 2016).Mice were observed daily following infection.Mice dedicated for FACS analysis were euthanized 3 days post-infection (DPI), while the remaining mice were observed for up to 10 DPI before euthanasia.In the US, mice were euthanized by CO 2 asphyxiation followed by cervical dislocation, while mice in Luxembourg were euthanized directly by cervical dislocation.Colons were excised and either processed for bead penetration assay measurements or stored in Methacarn fixative for histological assessment.For FACS analysis in a subset of mice, colons were excised and processed, as described in the section "Immune cell profiling of colonic lamina propria".For RNA extraction, mesenteric lymph nodes and 2 mm sections of colonic tissue were stored in 1 ml RNAlater for up to 2 weeks, then the RNAlater was removed and the tissue stored at −80°C until subsequent processing.Cecal contents were flash-frozen prior to storage at −80°C for RNA extraction or SCFA measurements.Mice fed the AcGGM diet followed the same protocol as the mice fed the FF diet.

Cultivation and administration of the SM
All SM-constituent strains were cultured and intragastrically gavaged into germ-free mice, as previously described (Steimle et al, 2021;Desai et al, 2016).For the experiment in the US, 14SM bacteria were cultured individually in modified yeast extract, casitone and fatty acid (YCFA) medium (Roseburia intestinalis, Faecalibacterium prausnitzii and Marvinbryantia formatexigens), tryptone yeast extract glucose (TYG) medium (Collinsella aerofaciens), modified Baar's medium for sulfate reducers (Desulfovibrio piger) or custom chopped meat broth (Bacteroides caccae, Bacteroides thetaiotaomicron, Bacteroides ovatus, Bacteroides uniformis, Barnesiella intestinihominis, Eubacterium rectale, Clostridium symbiosum, Escherichia coli and Akkermansia muciniphila) (Desai et al, 2016).For the experiments in Luxembourg, all 14SM bacterial cultures were grown individually in a further modified yeast extract, casitone, and fatty acid medium (mYCFA) (Steimle et al, 2021).Mice were orally gavaged with 0.2 ml of one of the five synthetic human gut microbiota (SM) combinations on three consecutive days.Mice the bacteria mixtures were freshly prepared with approximately equal proportions of each strain, adjusted based on the OD 600 (individual cultures ranged from OD 600 of 0.5-1.0).Selected strains were not included in the gavage mix depending on the specific SM (Fig. 1b).

Colon length and mucus thickness measurements
Colon length was measured by taking pictures of the colons in their histology cassettes, followed by length measurements with ImageJ using the cassette size as reference.Measurement of the colonic mucus layer thickness was performed as previously described (Desai et al, 2016), with cross-sectional images analysed in BacSpace (Earle et al, 2015).

Histological disease scoring
Histological disease scoring was performed by Prof. Kathryn A. Eaton from the University of Michigan Medical School according to a modified version of the protocol by Meira et al (Meira et al, 2008).Briefly, scores were assigned based on inflammation, epithelial damage, hyperplasia/dysplasia and submucosal edema.

Ex vivo mucus layer penetrability assessment
Gut mucus layer penetrability was measured using fluorescent beads, as previously described (Schroeder et al, 2018).Briefly, the colons were removed from euthanized mice and gently flushed using oxygenated ice-cold Krebs buffer (116 mM NaCl, 1.3 mM CaCl 2 x 2H 2 O, 3.6 mM KCl, 1.4 mM KH 2 PO 4 , 23 mM NaHCO 3 , and 1.2 mM MgSO 4 x 7H 2 O).The muscle layer was removed by blunt microdissection while keeping colon tissue suspended in oxygenated, ice-cold Krebs buffer.Then, the distal mucosa was mounted in a perfusion chamber.The apical chamber contained oxygenated, ice-cold Krebs-mannitol buffer (Krebs buffer with 10 mM mannitol, 5.7 mM sodium pyruvate and 5.1 mM sodium-L-glutamate), while the basolateral chamber contained oxygenated, ice-cold Krebs-glucose buffer (Krebs buffer with 10mM Glucose, 5.7 mM sodium pyruvate and 5.1 mM sodium-L-glutamate) with 0.6 µg/ml SYTO-9.After a 10 min incubation in the dark, at room temperature, FluoSpheres™ carboxylate beads (1 µm, red 580/605, Invitrogen) were added apically and allowed to sediment on the tissue for 5 min.Next, the apical chamber was gently flushed with Krebs-mannitol buffer to remove excess beads and the tissue was incubated for an additional 10 min before visualizing with a microscope.For each tissue, 4-7 confocal images were taken as XY stacks with 5 µm intervals, from the epithelium to the beads.Mucus penetrability was calculated according to the distance of each bead to the epithelium.

Nucleic acid extraction
Bacterial DNA was extracted from fecal pellets by phenol-chloroform extraction, as previously described (Steimle et al, 2021).Total RNA was extracted from the host mesenteric lymph nodes, colonic tissue, or cecal contents at 3 DPI, as previously described (Grant et al, 2023;Parrish et al, 2022).RNA integrity was determined using an Agilent 2100 Bioanalyzer system and the RNA was stored at -80°C until further analysis.

RNA sequencing and analysis
The RNA sequencing library was prepared using an Illumina Stranded Total RNA Prep with Ribo-Zero Plus kit (San Diego, CA, USA) and sequenced using a NovaSeq 6000 SP Reagent Kit v1.5 in 2 × 75 bp configuration on an Illumina NovaSeq 6000 system (San Diego, CA, USA) at the LuxGen Platform (Dudelange, Luxembourg).Raw data files were cleaned using kneaddata (https://github.com/biobakery/kneaddata)including adapter removal, discarding reads less than half the expected length, and removal of reads mapping to contaminant databases (ribosomal RNA and mouse genome, in the case where host transcripts were not the target).Colonic tissue transcripts were mapped to the Mus musculus genome, while cecal content transcripts were mapped individually to each of the 14SM strain genomes (Appendix Table 1) using Salmon (Patro et al, 2017).Host transcripts that were not found at least once on average across all samples were filtered out; for metatranscriptomic analyses, a cutoff of 2.5 cpm in at least 50% of the sample libraries was used, as described by Chen et al (Chen et al, 2016b).Differential expression analysis was carried out using DESeq2 1.30.1 with correction for multiple comparisons using the Benjamini-Hochberg method (Love et al, 2014).For metatranscriptomic analysis in HUMANn3, the forward and reverse fastq files were concatenated and then analysed by aligning to a custom taxonomic profile of the 14SM members and C. rodentium, followed by remapping of UniRef90 identifiers to Pfam identifiers (Beghini et al, 2021).

Bacterial relative abundance quantification
Initial colonization of the SM was verified using phylotype-specific qPCR primers (Steimle et al, 2021).
Analysis of the final relative abundances was performed following 16S rRNA gene sequencing of the V4 region using primers described by Kozich et al. (Kozich et al, 2013).Amplicon libraries were prepared using the Quick-16S™ NGS Library Prep Kit (BaseClear, Leiden, NL) and run on an Illumina MiSeq (San Diego, CA, USA) using the MiSeq Reagent Kit v2 (500-cycles) at the Integrated BioBank of Luxembourg (IBBL, Dudelange, Luxembourg).Raw, demultiplexed sequences were processed and analyzed using mothur version 1.41.3 (Schloss et al, 2009).Reads were assigned taxonomy using the k-Nearest Neighbor method based on a custom database containing the sequences of the 14SM bacteria and C. rodentium.

C. rodentium culturing and enumeration
C. rodentium was cultivated aerobically in LB medium at 37°C while shaking and CFUs were enumerated using LB-agar plates containing kanamycin, as previously described (Desai et al, 2016).

Intestinal short-chain fatty acid analysis
Thirty to 100 mg of flash-frozen cecal content collected from uninfected mice was used for the fatty acid analysis by gas chromatography-mass spectrometry (GC-MS), as previously described (Greenhalgh et al, 2019;Wolter, 2021).

Immune cell profiling of colonic lamina propria
Colons from infected mice were open longitudinally and gently washed with ice-cold 1X PBS before being placed in Hanks' balanced salt solution (HBSS) with phenol red and without calcium and magnesium (Lonza, Basel, Switzerland) on ice.Once all tissues were collected, immune cells from the colonic lamina propria were extracted using the Lamina Propria Dissociation Kit for mice and gentleMACS dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany) following the manufacturer's instructions.Approximately 1.5 × 10 6 cells per animal were stained and acquired using a NovoCyte Quanteon flow cytometer (ACEA Biosciences Inc., San Diego, CA, USA), as previously described (Grant et al, 2023).The gating strategy can be found in Fig EV3b.

Statistical analyses
Unless otherwise stated, statistical analyses were performed using Prism 8.1.1.(GraphPad Software, Inc., San Diego, CA, USA).Outlier removal was performed using the ROUT method with coefficient Q = 1% (Motulsky & Brown, 2006).Data were assessed for normality using the Kolmogorov-Smirnov test.For normally distributed data, multiple unpaired two-tailed t-tests were used with Welch's correction (no assumption of equal standard deviations), otherwise, multiple Mann-Whitney tests were used.In all cases, p-values were adjusted for pre-selected comparisons (between diets within same SM) using the Benjamini-Hochberg method and exact values are displayed on the plots if less than 0.01, otherwise, the comparisons are indicated as not significant (ns).The specific test and the number of animals used for each experiment is detailed in the figure legends.for FF diet with 20-day feeding period, and 10 for FF diet with 5-day feeding period, n=1-4 mice per sampling point.g Fecal C. rodentium load of 14SM mice fed the FF diet for 5, 20 or 40 days before infection.5-day feeding n=4, 20-day feeding n=3, 40-day feeding n=12.Some days it was impossible to recover fecal pellets from certain mice.Error bars represent SEM; unpaired two-tailed t-test between 5day feeding and 40-day feeding (top significance labels) groups or between 20-day feeding and 40-day feeding (bottom significance labels) groups.

Extended
throughout the experimental timeline by 16S rRNA gene sequencing (Fig EV1), with the averages for each group prior to the infection depicted in Fig. 1c.Across different SM combinations, two consistent shifts could be observed in the FF group, relative to the FR group: 1) an expansion of the mucolytic bacteria A. muciniphila and/or B. caccae (when present in the SM); and 2) a reduction of fiber-degrading bacteria such as Eubacterium rectale and Bacteroides ovatus (Fig. 1c, Fig EV1).These microbial changes resulted in an overall increased relative abundance of mucolytic bacteria in place of fiberdegrading bacteria during the FF diet (Fig. 1d).

B
. uniformis (Fig EV2a).Although we aimed to rescue the vulnerable 14SM phenotype by supporting the growth and metabolic activity of the aforementioned butyrogenic commensals, the AcGGM supplemented group showed similar C. rodentium loads (Fig EV2b), LCN-2 levels (Fig EV2c) and colon lengths (Fig EV2d) as the 14SM FF group.
For example, transcripts corresponding to TonB-dependent receptors are simultaneously differentially expressed in 13SM FF and 14SM FF because they corresponded to different loci in the B. ovatus genome (Fig EV4).Therefore, in order to calculate differential expression according to the unique gene product, we used HUMANn3 to map the reads to a custom database of the 14SM plus C. rodentium and regrouped the unstratified output to the Pfam database (Mistry et al, 2021) for consistent nomenclature (Dataset EV6).Again, differentially expressed gene products between the non-susceptible and susceptible (14SM FF) phenotypes were identified based muciniphila alters susceptibility to C. rodentium could involve multiple independent contributors.The next step to precisely decipher contribution of individual molecular mechanisms, of how A. muciniphila modulates disease susceptibility in a diet-dependent manner, includes the implementation of genetic systems, for example by using transposon mutants, as introduced by Davey et al. (Davey et al, 2023).
Animal experiments in the United States were approved by the University of Michigan Institutional Animal Care & Use Committee.Animal experiments in Luxembourg were approved by both the University of Luxembourg Animal Experimentation Ethics Committee and by the Luxembourgish Ministry of Agriculture, Viticulture, and Rural Development (authorization no.LUPA 2019/52) and carried out according to the "Règlement Grand-Ducal du 11 janvier 2013 relatif à la protection des animaux utilisés à des fins scientifiques", according to Directive 2010/63/EU on the protection of animals used for scientific purposes.

Figure 1 :
Figure legendsFigure1: Species-dropout approach enables distinct functionally characterized synthetic communities in vivo a Experimental timeline.Age-matched germ-free Swiss Webster mice were gavaged with one of the different synthetic gut microbiota (SM) on three consecutive days.These mice were maintained for 14 days on the fiber-rich (FR) diet, after which approximately half of the mice were switched to a fiber-free (FF) diet.The mice were maintained for 40 days on their respective diets before they were infected with Citrobacter rodentium.After infection, mice were closely observed for up to ten days.b Tablerepresentingfive different synthetic microbiota compositions (10SM, 11SM, 12SM, 13SM and 14SM).Bacteria which are present in a given SM are represented by a plus (+) sign.c Relative abundance of the gut bacteria as determined by 16S rRNA gene sequencing of stools of uninfected mice.d Relative proportion of fiber-and

Figure 2 :
Figure 2: A. muciniphila drives increased C. rodentium susceptibility during fiber-deprivation a Fecal C. rodentium load of gnotobiotic Swiss Webster mice from 1-10 days post-infection (DPI) among mice fed a standard, fiber-rich (FR) diet to compare between SM compositions.Log-transformed CFU/g were analyzed using a mixed effects model with matching on mouse ID, Geiger-Greenhouse correction, and p-value adjustment using the Benjamini-Hochberg method with 14SM as the control for multiple comparisons.Number of datapoints are indicated on each graph; 6 outliers were excluded.Error bars represent SEM.ns, non-significant, *p<0.05;**p<0.01.Asterisk color represents the group exhibiting a significant difference compared to 14SM.b,c Area under the curve (AUC) for fecal C. rodentium loads across the 10 days post-infection for (b) 11SM and 10SM FR mice and (c) each SM and diet combination.Unpaired Mann-Whitney test with p-value adjustment using the Benjamini-Hochberg method between (b) SM or (c) diets.Missing values were imputed by calculating the geometric mean of adjacent time points or assuming the value of the adjacent timepoint if the missing value occurred at the start of the series.In

Figure 3 :
Figure 3: Exploration of host immune response and pathogen transcriptome to tease apart factors behind increased pathogen susceptibility a Select immune cell populations as a percent of the parent population among mice at 3 DPI, determined by fluorescence-activated cell sorting (FACS).See Fig EV3b for the gating strategy.Population percentages were analyzed by multiple Mann-Whitney tests with p-value adjustment using the Benjamini-Hochberg method.Error bars show IQR.b PCA plot of the mouse colonic tissue transcripts at 3 DPI separated by experimental group.Not all gene names are visible due to text overlaps; see Dataset EV2 for full results.c PCA plot of the C. rodentium transcripts at 3 DPI separated by experimental group.d Volcano plot of murine host gene transcription in colonic tissues at 3 DPI.e Volcano plot of C. rodentium

Figure 4 :a
Figure 4: Dietary fiber deprivation-induced erosion of the gut mucus layer only partially explains increased C. rodentium susceptibility.a Histologically assessed mucus thickness of uninfected 13SM and 14SM mice on the FR and FF diet based on Muc2 antibody staining.Error bars represent SEM; unpaired two-tailed t-test b Ex vivo mucus layer penetrability assessment on uninfected mice.AUC of the normalized bead penetrability was determined for each mouse.Error bars represent SEM; unpaired two-tailed t-test.ns, non-significant; *p<0.05.c Mapping rates of cecal RNA reads to the 14SM and C. rodentium genomes, averaged by diet and SM group.d Normalised counts (per million reads) of transcripts mapping to genes for mucintargeting enzymes according to UniRef90 names using HUMANn3.Error bars represent SD; one-way ANOVA with p-value adjustment using the Benjamini-Hochberg method.

Figure 5 :
Figure 5: Cecal metatranscriptomes highlight altered community activities and persistent functional knowledge gaps a Volcano plot of Akkermansia muciniphila transcription in cecal contents at 3 DPI in 14SM FR (C.rodentium resistant) and 14SM FF (C.rodentium susceptible) mice, mapped using Salmon.Significance based on adjusted p-value <0.05 using the Benjamini-Hochberg method in DESeq2.b Fold-change plot for known immunomodulatory gene products transcribed by Akkermansia muciniphila based on UniRef90 identifiers in HUMANn3.Left: Pili-like membrane protein Amuc_1100.Right: threonine--tRNA ligase.Copies per million reads attributed to each transcript, p-values calculated using Welch's t-test.c Barplot indicating the number of differentially expressed genes using DESeq2 in the indicated group-wise comparison for each member of the 14SM community plus C. rodentium.Transcripts mapped using Salmon, with significance based on adjusted p-value <0.05 using the Benjamini-Hochberg method in DESeq2.d Left: Heatmap of the top 25 (lowest adjusted p-value, all <0.05) differentially expressed genes according to Pfam identifier, regrouped from UniRef IDs in HUMANn3, for all C. rodentium resistant

Figure 6 :a
Figure 6: Impaired mucosal integrity and altered community dynamics underlie observed microbiota and diet-dependent susceptibility to Citrobacter rodentium infection.Visual illustration of the impact of mucin-degrading bacterium A. muciniphila in mice fed a fiber-deprived diet on C. rodentium susceptibility.While host immune response and pathogen activities are unchanged in these conditions, mucosal barrier integrity and altered transcriptional profiles suggest a combination of these factors contributes to the observed disease phenotype.14SM, 14-member synthetic microbiota; 13SM, 13-member synthetic microbiota (lacking Akkermansia muciniphila); R; resistant to C. rodentium infection; S, susceptible to C. rodentium infection.Created using BioRender.com.

Data Figure 3 :a
Host immune cell populations remain largely unchanged upon diet switch.Non-significantly altered immune cell population as a percent of the parent population among mice at 3 DPI, as determined by fluorescence-activated cell sorting (FACS).Population percentages were analyzed