Microbiome Mucin degrader Akkermansia muciniphila accelerates intestinal epithelial regeneration and repairs damaged intestinal mucosa

Background: Mucin-degrading bacteria are densely populated in the intestinal epithelium; however, their interaction with intestinal stem cells (ISCs) and their progeny has not been elucidated. To determine whether mucin-degrading bacteria play a role in gut homeostasis, mice were treated with Akkermansia muciniphila , a specialized species that degrades mucin. A total of 32 fecal samples were obtained from healthy volunteers and A. muciniphila was isolated from 11 samples. Mechanism of A. muciniphila was observed in vivo and in vitro , and studied using organoids, histology, metagenomics, and whole genome sequencing. Results: We found that administration of A. muciniphila for 4 weeks accelerated the proliferation of Lgr5 + ISCs and promoted the differentiation of Paneth cells and goblet cells in the small intestine (SI). The levels of acetic and propionic acids were higher in the cecal contents of A. muciniphila -treated mice than in PBS-treated mice. SI organoids treated with cecal content supernatant obtained from A. muciniphila -treated mice were larger and could be diminished by treatment with G protein-coupled receptor (Gpr)41/43 antagonists. Pre-treatment of mice with A. muciniphila reduced gut damage caused by radiation and methotrexate. A novel isotype of A. muciniphila strain was isolated from heathy human feces that possessed improved functions for intestinal epithelial regeneration. Conclusions: These findings suggest that mucin-degrading bacteria (such as A. muciniphila ) may play a crucial role in promoting ISC-mediated epithelial development and contribute to intestinal homeostasis maintenance. ISCs and Paneth cells from Lgr5-GFP crypt cell suspensions TrypLE for 10 at The dissociated cells were stained with the Live/Dead Cell Stain Fisher and anti-CD24 monoclonal Fisher Cell sorting was performed using a FACS Aria Ⅲ cell sorter. ISCs were sorted as Lgr5-GFP hi and Paneth cells were sorted as Lgr5-GFP - CD24 hi , respectively.


7
ISCs and Paneth cells from Lgr5-GFP mice, crypt cell suspensions were dissociated using TrypLE Express (Thermo Fisher Scientific) for 10 min at 37°C. The dissociated cells were stained with the Live/Dead Cell Stain kit (Thermo Fisher Scientific) and anti-CD24 monoclonal antibody (Thermo Fisher Scientific). Cell sorting was performed using a FACS AriaⅢ cell sorter. ISCs were sorted as Lgr5-GFP hi and Paneth cells were sorted as Lgr5-GFP -CD24 hi , respectively.

Organoid culture
For construction of organoids, 200-500 crypts per well were suspended in Matrigel (Corning) as described [35]. Complete ENR medium (all components from Thermo Fisher Scientific unless noted) comprised of advanced DMEM/F12 (Gibco), antibiotic-antimycotic (×100), 1 mM N-acetyl cysteine (Sigma-Aldrich), B27 supplement, N2 supplement, EGF, Noggin (R&D Systems), R-spondin-1-conditioned medium, and Y-27632 (Sigma). Y-27632 were added to the ENR medium for the first 48-72 hr of culture only and then removed during the medium change. The ENR medium was replaced every 2-3 days. Isolated ISCs and Paneth cells were co-cultured in ENR medium supplemented with Jagged-1 (1 µM, Anaspec). Wnt-C59 (50 µM, Abcam) was used as a porcupine (PORCN) inhibitor. The surface areas of SI organoids were measured microscopically by taking several random non-overlapping photos of organoids in a well using an inverted microscope (Carl Zeiss). Each photo was analyzed using ImageJ software (NIH) and the Zen image program (Carl Zeiss). Organoid perimeters for area measurements were defined manually using automated ImageJ software. 8

Histology
Ileum tissues were removed, opened longitudinally, and formed into Swiss rolls. The tissue was then fixed in 4% paraformaldehyde (PFA) and embedded in paraffin. Tissue sections were stained with hematoxylin-eosin (H&E) or periodic acid-Schiff (PAS).

Immunofluorescence staining
Ileum tissues were fixed with 4% PFA and dehydrated with 15% then 30% sucrose in PBS.

Organoid treatment with cecal contents
Cecal contents (100 mg) were diluted in 1 ml of serum-free DMEM/F12 (Gibco) medium and vortexed for 1 hr. The contents were centrifuged at 4,000 rpm for 10 min and supernatants were passed through a 0.22-m syringe filter (Pall Corp.) before cultivation. To address the effectiveness of cecal contents in promoting organogenesis, we used ENR media supplemented with cecal supernatant diluted to 0.01% in advanced DMEM/F12. For inhibition of Gpr41/43, Gpr43 antagonist (GLPG0974, 0.1 µM, Tocris) and Gpr41 antagonist (β-hydroxybutyrate, 3 mM, Sigma) were used.

Microbiome data analysis pipeline
Total DNA was extracted from feces using QIAamp DNA stool mini kits (Qiagen) in accordance with the manufacturer's instructions. For bacterial PCR amplification, primers targeting 341F and 805R were used. The amplified product was purified and sequenced by Chunlab (Seoul, South Korea) with an Illumina Miseq Sequencing system (Illumina).
Processing raw reads started with quality check and filtering of low-quality (<Q25) reads by Trimmnomatic ver. 0.32. After a quality control pass, paired-end sequence data were merged together using VSEARCH version 2.13.4 using default parameters. Non-specific amplicons that did not encode 16S rRNA were detected by nhmmer in the HMMER software package version 3.2.1. We used the EzBioCloud 16S rRNA database for taxonomic assignment by precise pairwise alignment [36]. After chimeric filtering, reads that were not identified to the species level (with <97% similarity) in the EzBioCloud database were compiled. OTUs with single reads (singletons) were omitted from further analysis. The alpha diversity (Shannon index) and beta diversity for the sample difference were estimated. A taxonomic cladogram was generated using LEfSe with a threshold of 2 on the logarithmic LDA score [37]. A relationship based on a Pearson correlation between gut microbiota and SCFAs was visualized using Calypso software [38].

Whole genome sequencing
The integrity of gDNA was tested by running an agarose gel electrophoresis. gDNA was quantified using the Quant-IT PicoGreen (Invitrogen). The sequencing libraries were then prepared according to the manufacturer's instructions (20 kb template preparation and the BluePippin TM Size-Selection System) using the PacBio DNA template Prep Kit 1.0. The libraries were quantified using Quant-IT PicoGreen and qualified using the high-sensitivity DNA chip (Agilent Technologies). Subsequently, the libraries were sequenced using PacBio P6C4 chemistry in 8-well-SMART Cell v3 in PacBio RSII. The genome of the AK32 strain was constructed de novo using PacBio sequencing data. Sequencing analysis was performed by Chunlab. PacBio sequencing data were assembled with PacBio SMRT analysis 2.3.0 using the HGAP2 protocol (Pacific Biosciences). Resulting contigs from PacBio sequencing data were circularized using Circlator 1.4.0 (Sanger Institute). The gene-finding and functional annotation pipeline of whole genome assembly was used from the EzbioCloud genome database. Protein coding sequences were predicted by Prodigal 2.6.2. Genes coding for tRNA were searched using tRNAscan-SE 1.3.1. The rRNA and other non-coding RNAs were searched by Rfam 12.0 database. Comparative whole genome analysis was studied by average nucleotide identity base BLAST (ANIb). The ANIb value was calculated by ANI calculator from Kostas lab (http://enve-omics.ce.gatech.edu/ani). Operons including pdh and mmd were predicted by using MicrobesOnline Operon Predictions tools [39]. We estimated the -10 and -35 regions in promoter sequences of these genes by use of the phiSITE database [40].

Real-time PCR
Total RNA from the SI and SI-derived organoids was extracted using the RNeasy mini kit (Qiagen) and cDNA was synthesized using Superscript Ⅱ reverse transcriptase and oligo dT primer (Thermo Fisher Scientific). Total RNA of A. muciniphila was extracted using Trizol (Thermo Fisher Scientific). The ReverTra Ace qPCR RT master mix with gDNA remover (Toyobo) was used to synthesis cDNA from bacterial RNA. cDNA was used as the template for real-time PCR performed using SYBR green chemistry (Thermo Fisher Scientific) on a was normalized to the EIC of the internal standard. The peak area ratio of each metabolite 13 was normalized to the internal standard using serum volume or tissue weight in a sample and then used for relative comparison.

Statistics
Statistical analyses were performed with Prism software (GraphPad, La Jolla, CA). For pairwise and two independent group comparisons, a two-tailed t-test was used. Data are presented as mean ± SEM. The values p < 0.05, p < 0.01, and p < 0.001 were considered as statistically significant.

Results
Oral administration of A. muciniphila may promote epithelial differentiation in the small intestine.
To address whether mucin-degrading bacteria regulate IEC differentiation, A. muciniphila and Spdef1 that regulate differentiation of secretory lineage cells [42][43][44] in A. muciniphilatreated mice compared with controls ( Figure 1H). These results suggest that A. muciniphila may promote differentiation of secretory lineage cells in the SI. 15 Oral administration of A. muciniphila may accelerate ISC proliferation.
Because secretory subtypes of IECs are derived from Lgr5 + ISCs, we next investigated whether A. muciniphila modulates the proliferation of Lgr5 + ISCs. Lgr5-GFP mice administered A. muciniphila had more GFP-expressing Lgr5 + ISCs in the SI crypt than PBStreated mice (Figure 2A and 2B). Organoids derived from the SI crypt of A. muciniphilatreated mice were larger than those of PBS-treated mice ( Figure 2C and 2D). Furthermore, Lgr5 expression was upregulated in the SI and SI organoids from A. muciniphila-treated mice compared with controls ( Figure 2E). RNA ISH analysis indicated increased production of Olfm4 + ISCs, which were confirmed in the SI crypt of A. muciniphila-treated mice ( Figure   2F). In addition, protein and mRNA levels of Muc2 and Lyz1 were upregulated in the SI organoids of A. muciniphila-treated mice compared with PBS-treated mice ( Figure 2G and 2H). These results imply that A. muciniphila may play a critical role in accelerating the proliferation of ISCs.
Oral administration of A. muciniphila enhances ISC proliferation by Wnt signaling.
Because Wnt signaling is involved in maintaining ISC stemness in the SI crypt, we next investigated whether A. muciniphila treatment regulates the Wnt pathway. Mice given A.
muciniphila orally had increased expression of Wnt3, Axin2, and Ctnnb1 in their SI tissues ( Figure 3A). The upregulated expression of Wnt3 and Axin2 in the SI crypt of A. muciniphilatreated mice was further confirmed by RNA ISH ( Figure 3B). In addition, Wnt3 protein levels were higher in the SI crypt and SI organoids of A. muciniphila-treated mice than in PBS-treated control mice ( Figure 3C and S1B). Of note, β-catenin protein levels were upregulated in the nuclei of the SI crypt of A. muciniphila-treated mice compared with PBS-treated mice ( Figure 3D) indicating increased translocation from the cytoplasm. Because the Wnt/β-catenin pathway activates RAS-ERK signaling that in turn promotes stemness [45], we next examined ERK phosphorylation (pERK) in the SI crypt. As expected, oral administration with A. muciniphila increased pERK expression in the SI crypt compared with controls ( Figure S1C). To address whether A. muciniphila activates Wnt3 signaling in the SI, Lgr5-GFP hi ISCs isolated from naive mice were co-cultured with Paneth cells isolated from A.
muciniphila-treated or PBS-treated mice. Interestingly, SI organoids grew significantly more in co-cultures isolated from A. muciniphila-treated mice than in PBS-treated mice ( Figure 3E and 3F). This effect was diminished when the porcupine inhibitor (Wnt-C59) was added to the co-cultures ( Figure 3E and 3F). Together, these results demonstrate that A. muciniphila promotes the secretion of Wnt3 from Paneth cells that support ISC proliferation in the SI.
Oral administration of A. muciniphila promotes SCFA secretion and ISC-mediated epithelial development.
Because microbiota-derived metabolites are a key factor in gut homeostasis, we next investigated whether metabolites produced from A. muciniphila treatment effected ISCmediated epithelial development. To address this question, the cecal contents from PBS-or A.
muciniphila-treated mice were isolated and then applied to SI organoids from naive B6 mice.
SI organoids treated with A. muciniphila-treated mouse cecum were significantly larger than those treated with PBS-treated cecum ( Figure 4A). Furthermore, the mRNA levels of Lgr5, Lyz1, Muc2, and Wnt-related genes (Wnt3, Axin2, Ctnnb1) increased in the presence of cecal contents obtained from A. muciniphila-treated mice compared with PBS-treated mice ( Figure   4B). To identify which metabolites are associated with A. muciniphila-mediated epithelial development, the levels of SCFAs were examined in cecal contents. As predicted, higher levels of SCFAs, including acetic, propionic, and butyric acids, were present in the cecal contents from A. muciniphila-treated mice than from PBS-treated mice ( Figure 4C). Of these, acetic and propionic acids, but not butyric acid, were highly associated with increased SI organoid growth ( Figure S2A and S2B). Treatment with the Gpr41/43 antagonist reduced SI organoid growth ( Figure 4D), suggesting that A. muciniphila-derived SCFA metabolites play an important role in ISC-mediated epithelial development.
Oral administration of A. muciniphila alters gut microbiota composition and SCFA production.
We next addressed whether oral administration of A. muciniphila altered the composition of gut microbiota and, as predicted, found that it did ( Figure S3A). At the phylum level, the gut microbiota from A. muciniphila-treated mice showed an increased proportion of phyla Bacteroidetes and Proteobacteria and decreased numbers of phyla Firmicutes, compared with PBS-treated mice ( Figure S3B). Further, linear discriminant analysis (LDA) with LEfSe confirmed that several bacteria genera were prominently changed after A. muciniphila treatment ( Figure 5A and 5B). The genera Muribaculum, Alistipes, Akkermansia, Helicobacter, and Desulfovibrio showed an upper 2 LDA score after A. muciniphila treatment compared with control mice ( Figure 5B). Furthermore, the Shannon index was significantly increased in A. muciniphila-treated mice compared with PBS-treated mice, indicating alteration of the bacterial community structure ( Figure 5C). Unifrac-based PCoA analysis demonstrated that the two groups were clustered separately ( Figure 5D). Interestingly, a positive correlation was observed between the Akkermansia-induced population and presence of SCFA metabolites (acetic, propionic, and butyric acids) ( Figure 5E). In summary, these results suggest that A. muciniphila treatment promotes ISC-mediated epithelial development by altering the gut microbiota composition that in turn activates SCFA secretion.
Oral administration of A. muciniphila may repairs radiation and chemotherapy gut damage.
Since A. muciniphila promotes ISC-mediated epithelial development, we next investigated whether A. muciniphila plays a role in preventing gut damage. Our previous study [22] showed that radiation (R; 10 Gy) and methotrexate (M; MTX) cause severe damage to mouse SI tissues. In this study, we assessed PBS-treated mice (PBS+R+M) ( Figure 6A and 6B) and mice treated with A. muciniphila for 4 weeks prior to radiation and MTX treatment (A. muciniphila+R+M). The treated group had less severe damage ( Figure 6B). In addition, more Lgr5 + ISCs were maintained in the SI crypt of A. muciniphila+R+M mice compared with the PBS+R+M mice ( Figure 6C). As predicted, A. muciniphila+R+M mice lost less weight than the PBS+R+M mice ( Figure 6D). The organoid size and number derived from the SI of A. muciniphila+R+M mice were significantly increased in comparison with PBS+R+M-derived SI organoids, indicating that pre-treatment with A. muciniphila reduced damage and may play a protective role in the gut ( Figure 6E and 6F). These results suggest that the symbiotic actions of A. muciniphila may promote gut repair following damage provoked by cancer therapy.
A. muciniphila AK32 from healthy human feces is superior to BAA-835 for ISCmediated epithelial development. 19 We next investigated whether an A. muciniphila strain isolated from healthy human feces promotes ISC-mediated epithelial development compared with the common strain (ATCC BAA-835 T ). By use of selective media and species-specific PCR analysis, we obtained 11 different A. muciniphila strains. To evaluate the effect of A. muciniphila on ISC-mediated epithelial development, SI-derived organoids were cultured with culture supernatant from one of the strains. Only treatment with the AK32 strain significantly increased organoid size ( Figure 7A). To address whether increased ISC-mediated epithelial development by the AK32 strain is dependent on SCFAs, the Gpr41/43 antagonist was applied to the cultures. Treatment with the Gpr41/43 antagonist significantly reversed the AK32-mediated effect on SI-derived organoid size ( Figure 7B). As anticipated, treatment with the AK32 increased production of acetic acids and propionic acids compared with the A. muciniphila type strain ( Figure S4A).
To examine how the AK32 strain increased SCFA secretion, the expression levels of two important enzymes, pyruvate dehydrogenase E1 component (Pdh) and Na + translocating methylmalonyl-CoA/oxaloacetate decarboxylase (Mmd) were examined ( Figure S4C). The mRNA expression levels of pdh and mmd from the AK32 strain were higher than those of BAA-835 type strain ( Figure 7C). Next, we used whole genome sequencing to analyze genetic characteristics of strain AK32, including the pdh and mmd coding genes. Figure S5 shows a complete genome map of AK32-based Clusters of Orthologous Groups (COG). The genomic characteristics (genome size and number of coding sequences) in the AK32 strain differed from those of type strain BAA-835 (Table S1). As the mRNA expression differences might be attributed to regulation of transcription by promoter sequences, we assessed the operons, including pdh and mmd. Interestingly, strains AK32 and BAA-835 had the same promoter sequences and amino acid sequences of pdh but those of mmd differed ( Figures S6   and S7). To further examine the in vivo function of strain AK32, mice were treated with either 20 the AK32 strain or BAA-835 for 4 weeks. Of note, the SI organoids from AK32-treated mice were significantly larger those from mice treated with the BAA-835 type strain ( Figure 7D).
Administration of the AK32 strain increased SI crypt height and the number of mucinproducing goblet cells compared with the SI of mice treated with BAA-835 ( Figure 7E).
Administration of strain AK32 resulted in increased mRNA expression of Lgr5, Lyz1, Muc2, Wnt3, Axin2, and Ctnnb1 in the SI compared with findings in mice treated with strain BAA-835 ( Figure 7F). Furthermore, higher levels of acetic and propionic acids were detected in the cecal contents from AK32-treated mice than in mice treated with BAA-835 ( Figure S4B).
Thus, we concluded that the newly identified A. muciniphila AK32 strain was superior to A.
muciniphila BAA-835 in terms of ISC-mediated epithelial development.

Discussion
In this study, A. muciniphila was found to play a crucial role in ISC-mediated epithelial development by activation of the Wnt signaling pathway and repair of the damaged gut.
Treatment with A. muciniphila upregulated the expression of genes involved in the Wnt signaling pathway and increased production of SCFA metabolites, such as acetic and propionic acids, which in turn maintain stemness of the ISCs. A novel A. muciniphila strain was isolated from healthy human stools that promoted the expression of genes involved in acetic acid and propionic acid production, and therefore may have improved functionality for maintaining gut homeostasis.
The Lgr5 + Ki67cells located at crypt positions +4/+5 of the SI are destined to differentiate into secretory lineage cells [41]. Administration of A. muciniphila resulted in an increased density of Lgr5 + Ki67cells in the SI crypt, which led to an increased number of secretory lineage cells, such as goblet and Paneth cells. The most important pathway for IEC development is Wnt signaling [6,46]. In support of this theory, inhibition of the Wnt pathway was shown to reduce the number of Math1 + precursor cells, resulting in a depletion of secretory lineage cells [7]. Paneth cells play a critical role in maintaining the ISC niche and produce Wnt3 [5]. In this study, the co-culture of ISCs and Paneth cells demonstrated that Paneth cells from A. muciniphila-treated mice directly promoted ISC proliferation in a Wnt3dependent manner. Thus, we speculate that metabolites produced by A. muciniphila stimulate Paneth cells to secrete Wnt3, which then promotes the proliferation of Lgr5 + Ki67cells at SI crypt positions +4/+5.
A previous study reported that A. muciniphila uses mucin as an energy source by 22 converting it into acetic and propionic acids [47]. Paradoxically, others have found that A.
muciniphila promotes the generation of mucin-secreting goblet cells that were depleted by a high-fat diet [27,32,48]. In this study, our aim was to investigate how A. muciniphila activates mucin secretion. We propose that acetic and propionic acids produced by A.
muciniphila may be key factors for supporting the maturation of mucin-secreting goblet cells.
Another study reported that administration of A. muciniphila did not reconstitute the gut microbiome [31]. In contrast, we found that A. muciniphila altered the gut microbiota composition and structure, which may affect the pattern of metabolite secretion. Germ-free mice supplemented with A. muciniphila showed an exacerbated infection by Salmonella typhimurium, suggesting that an over-abundance of A. muciniphila and reduced microbiome diversity leads to a deleterious modification of the gut environment [49]. Also, an accumulation of antimicrobial peptide produced by Paneth cells may contribute to a change in gut microbiota [50]. We therefore hypothesize that treatment with A. muciniphila may alter the bacterial composition and SCFA production in the gut; for example, by altering the abundance of phyla Bacteroidetes, which produces acetic and propionic acids [51], or Alistipes and Rikenellaceae, which produce SCFAs [52,53]. Therefore, we conclude that A. muciniphila increases SCFA production directly via mucin degradation and indirectly by altering the microbiome composition.
Individual bacterial strains, even the same species, have strain-specific abilities.
Because lactic acid bacterial strains of the same species show different enzyme activity [54], we wondered if different A. muciniphila strains might have different effects on ISC-mediated epithelial development. We found that the newly identified A. muciniphila AK32 strain activated the expression of genes involved in SCFA production and increased the secretion of 23 acetic and propionic acids. As expected, promoter sequences of mmd in AK32 differed by type strain, but those of pdh were highly conserved. The detailed mechanism by which the expression occurs is yet to be determined. We did identify a strain-specific alteration of the Mmd amino acid sequence, which may influence the enzyme activity by changing propionic acid levels. Thus, we conclude that strain AK32 has a characteristic genome, unlike the A.
muciniphila type strain, that leads to increased SCFA production, especially acetic and propionic acids.
Treatment with A. muciniphila had a greater effect on the production of propionic acid than on acetic acid. Previously, it was suggested that A. muciniphila and propionic acid regulate the expression of genes associated with the host lipid metabolism and activate the epigenome [55]. Accumulating evidence suggests that propionic acid may modulate the host physiology in several ways. For example, propionic acid stimulates the release of peptide YY and glucagon-like peptide-1 in human colonic cells, and thereby reduces energy intake and weight gain [56]. Intriguingly, propionic acid stimulates Muc2 production by IECs by regulating the expression of the prostaglandins [57]. A recent study proposed that supplementation of propionic acid improves the Treg/Th17 imbalance in multiple sclerosis patients [58]. Taken together with our results, we conclude that propionic acid may play an important role in IEC homeostasis and the overall gut and therefore may modulate host physiology.
Several studies have reported an interaction between gut metabolites and IEC development [22,59]. Our prior study revealed that microbiota-derived lactate promotes IEC development [22]. By contrast, no significant changes in lactate were found in mice treated with A. muciniphila, suggesting that lactate is not a crucial metabolite involved in A. 24 muciniphila-mediated IEC development ( Figure S8a). A previous study reported that fatty acids such as palmitic acid, the main metabolite produced by gut microbiota, enhanced ISC proliferation [59]. Furthermore, our recent study demonstrated that dietary cellulose prevented gut inflammation by increasing the A. muciniphila population and modulating production of lipid metabolites [60]. Taken together, A. muciniphila treatment may increase the production of lipid metabolites, including myristic and palmitic acid, which influence IEC development ( Figure S8b). Further investigation is warranted to rule out this possibility. 25

Consent for publication
Not applicable.