Characterizing the mucin-degrading capacity of the human gut microbiota

Mucin-degrading microbes are known to harbor glycosyl hydrolases (GHs) which cleave specific glycan linkages. Although several microbial species have been identified as mucin degraders, there are likely many other members of the healthy gut community with the capacity to degrade mucins. The aim of the present study was to systematically examine the CAZyme mucin-degrading profiles of the human gut microbiota. Within the Verrucomicrobia phylum, all Akkermansia glycaniphila and muciniphila genomes harbored multiple gene copies of mucin-degrading GHs. The only representative of the Lentisphaerae phylum, Victivallales, harbored a GH profile that closely mirrored Akkermansia. In the Actinobacteria phylum, we found several Actinomadura, Actinomyces, Bifidobacterium, Streptacidiphilus and Streptomyces species with mucin-degrading GHs. Within the Bacteroidetes phylum, Alistipes, Alloprevotella, Bacteroides, Fermenitomonas Parabacteroides, Prevotella and Phocaeicola species had mucin degrading GHs. Firmicutes contained Abiotrophia, Blautia, Enterococcus, Paenibacillus, Ruminococcus, Streptococcus, and Viridibacillus species with mucin-degrading GHs. Interestingly, far fewer mucin-degrading GHs were observed in the Proteobacteria phylum and were found in Klebsiella, Mixta, Serratia and Enterobacter species. We confirmed the mucin-degrading capability of 23 representative gut microbes using a chemically defined media lacking glucose supplemented with porcine intestinal mucus. These data greatly expand our knowledge of microbial-mediated mucin degradation within the human gut microbiota.

To access mucin glycans, intestinal microbes must possess mucin-degrading glycosyl hydrolases ( Fig. 2A) 2 . Released mucin glycan oligosaccharides can then be used to support the growth of bacteria. Given the prominence of Akkermansia as a mucin-degrading genus, we first analyzed the genomes of human gut microbes A. glycaniphila and A. muciniphila (Fig. 2B). The one available genome of A. glycaniphila contained a least one gene copy of GH33 (sialidase), GH16 (endo-acting O-glycanase), GH29 (fucosidase), GH95 (fucosidase), GH20 (galactosidase), GH2 (galactosidase), GH35 (galactosidase), and GH84 (N-acetyl-glucosaminidases). Similarly, all the A. muciniphila genomes contained a least one gene copy of GH33, GH16, GH29, GH95, GH20, GH2, GH35 and GH84, as well as GH89, indicating that A. muciniphila can cleave sialic acid, fucose, galactose, and N-acetylglucosamine. Closer examination of the Akkermansia genomes revealed that the one genome of A. glycaniphila had six gene copes of GH33 and all 95 of the A. muciniphila genomes contained 2-4 genes copies Scientific Reports | (2022) 12:8456 | https://doi.org/10.1038/s41598-022-11819-z www.nature.com/scientificreports/ of GH33 (Fig. 2C), indicating that Akkermansia spp. have the capacity to remove sialic acid and initiate mucindegradation. The GHs with the largest gene copy range (6-13 gene copies) was GH20, a family containing β-Nacetyl-glucosaminidases (Table 1). No Akkermansia genomes contained GH42, 98, 101, 129 or 85, suggesting that Akkermansia is unable to degrade N-acetyl-galactosamine. To confirm the capacity of A. muciniphila to degrade intestinal mucus, we grew A. muciniphila ATCC BAA-835 is a chemically defined media ZMB1 lacking glucose, containing 100 mM glucose or containing 1 mg/mL porcine intestinal mucus (Fig. 2D). As expected, A. muciniphila had limited growth in ZMB1 with or without glucose but exhibited robust growth in media with porcine intestinal mucus. These findings complement our genome analysis of A. muciniphila ATCC BAA-835 (the BAA-835 genome analysis is found in the second column from the right in Fig. 2C). Additionally, various A. muciniphila strains were also examined to showcase the diversity of GH profiles across the genus, which supports the ability of this species to degrade mucins.  www.nature.com/scientificreports/ Next, we examined Victavallales bacterium in the Lentisphaerae phylum (Fig. 2E). Genome analysis revealed a similar GH profile to Akkermansia, with genes for GH33, GH16, GH29, GH95, GH20, GH2, GH35, and GH89, suggesting that Victavallales bacterium could enzymatically cleave sialic acid, fucose, galactose, and N-acetylglucosamine. Interestingly, Victavallales also harbored gene copies of GH42 and GH129, GHs not found in Akkermansia. The presence of GH129 indicates that Victavallales bacterium can release N-acetyl-galactosamine, a glycan which Akkermansia is not able to cleave. Victavallales bacterium possessed 4 genes copies of GH33 and 19 gene copies of GH2, which contains β-galactosidases. Although little information is available for Victavallales bacterium, the genome analysis reveals that Victavallales bacterium could degrade mucins.
In Bifidobacterium (Fig. 3C), we found that all 11 genomes of B. bifidum had 1-3 gene copies of GH33 and all genomes had GH29, GH95, GH20, GH2, GH42, GH101, GH129, GH89 and GH84. Additionally, 8 of the 11 B. bifidum genomes had one gene copy of GH16. These GH profiles are consistent with previous studies which identify B. bifidum as a mucin degrading microbe since it can remove all mucin glycans 20,22 . Within the 44 B. breve genomes, we found that 41 genomes had one gene copy of GH33 and the majority of strains had GH95, GH20, GH2, GH42, and GH129, covering all mucin glycan structures (Fig. 3C). B. longum had much more variability in terms of mucin-degrading GHs (Fig. 3D). Only 11 of the 54 genomes contained GH33, the majority of which belonged to the B. longum subspecies infantis subgroup. Variable presence for GH29, GH95, GH20, GH2, GH42, GH101, GH129 and GH85 was identified, with genomes harboring 0-5 gene copies. In contrast, B. angulatum only possessed 2 mucin-associated GHs: GH2 and GH42, suggesting that this species is likely unable to extensively degrade mucins. These data indicate that mucin degradation is species dependent in Bifidobacteria.
To confirm our genome findings, we also examined the growth of key Bifidobacteria in ZMB1 with or without glucose or intestinal mucus (Fig. 3G). Our genome analysis revealed that B. dentium ATCC 27678 and B. angulatum ATCC 27535 did not possess GH33 and had only 2-3 mucin-associated GHs, while B. longum and B. bifidum had several gene copies of GH33 and other mucin-degrading GHs. In our growth analysis, we did not detect growth above the ZMB1 media baseline when intestinal mucus was added, indicating that these species cannot degrade intestinal mucus to use as a carbon source. In contrast, B. longum subsp. infantis ATCC 15697, B. longum ATCC 55813, and B. bifidum ATCC 29521 had enhanced growth when mucus was present, indicating that these strains can degrade mucins.
Compared to the mucin-degrading microbes identified in other phyla, we observed far fewer mucin-degrading GHs in the Proteobacteria phylum, with only 3-4 GHs families found in Klebsiella, Mixta and Enterobacter spp. (Fig. 5A). All 46 of the Klebsiella aerogenes genomes had one gene copy of GH33 (sialidase) and 1-2 gene copies of GH2 (galactosidase). Ten of the 46 K. aerogenes genomes also had expression of GH42 (galactosidase) and 41 of the genomes had 1-2 gene copies of GH20 (galactosidase), suggesting the ability of these strains to remove galactose residues (Fig. 5B). Similarly, all 23 undefined Klebsiella spp. genomes had 1-3 gene copies of GH2, but only 13 of the 23 genomes had GH33, 14 genomes had GH42 genes and 3 of the genomes had GH20 (Fig. 5C). No other mucin-degrading GHs genes were observed. Of the four Mixta spp., which includes M. calida and M. intestinalis, we found that all three genomes had 1-2 gene copies of GH33, GH20 and GH2, but no other mucin-related GHs were identified (Fig. 5D). We observed large variation in the 8 Serratia fonticola genomes. Only one of the genomes had GH33, 6 genomes had GH16, 7 genomes had GH20 and all 8 genomes had GH2 gene copies. In the Enterobacter genera, only 15 of the 73 E. cloacae genomes had one gene copy of GH33, although most of the strains had GH20 and GH2 gene copies (Fig. 5E). Similarly, only 5 of the 36 undefined Enterobacter spp. had GH33, while almost all the strains had GH20 and GH2 (Fig. 5F). Growth analysis of E. coli Nissle 1917 in ZMB1 with or without mucus, which was not one of the E. coli with GH33 expression in our genome analysis, confirmed the inability of this species to use mucus as the sole carbon source (Fig. 5G). These data suggest that commensal Proteobacteria are far less adept at degrading mucin than their gut microbiota counterparts.
Finally, we examined the Firmicutes phylum and found that Abiotrophia, Blautia, Enterococcus, Paenibacillus, Ruminococcus, Streptococcus, and Viridibacillus species harbored several mucin-degrading GHs (Fig. 6A-C). The one genome of Abiotrophia defectiva had one gene copy of GH33 and 1-2 gene copies of GH29, GH20, GH2, GH35, GH101, and GH85. Within Blautia, B. coccoides and B. hansenii had one gene copy of GH33 and genes for GH29, GH95, GH20, GH2, GH101, GH85 and GH84 (Fig. 6D). In contrast, B. obeum, B. producta and undefined Blautia spp. had no gene copies of GH33, but did have variable gene copies (0-20) of GH16, GH29, GH20, GH2, GH35, and GH42, indicating the ability to remove fucose and galactose. Among the Enterococcus strains, only 1 of the 4 E. casseliflavus strains, 1 of the 4 E. durans strains, 2 of the 4 E. gallinarum, and 1 of the 4 undefined Enterococcus spp. possessed GH33. Variable numbers of gene copies were observed in GH29, GH95, www.nature.com/scientificreports/ GH20, GH2, GH35, GH42 and GH85. In Paenibacillus (Fig. 6E), we observed that P. barcinonensis and P. lautus genomes had one gene copy of GH33 and both genomes harbored GH16, GH29, GH95, GH2, and GH35, while P. lautus also had gene copies for GH20 and GH85. Of the 29 genomes of undefined Paenibacillus spp., we found www.nature.com/scientificreports/ that only 4 strains had GH33, but the majority of strains had gene copies of GH16, GH29, GH29, GH95, GH20, GH2, GH35 and GH42, suggesting that Paenibacillus spp. can remove fucose and galactose. We observed that all three Ruminococcus gnavus genomes had one gene copy of GH33, while undefined Ruminococcus spp. and R. torques did not harbor GH33 (Fig. 6F). Most Ruminococcus strains possessed GH29, GH85, GH2 and GH42. Among the streptococci, we found that all 8 S. intermedius genomes contained GH33, GH29, GH20, GH2, GH35, and GH85 (Fig. 6G). We also observed that 6 of the 9 S. mitis spp. had GH33 and most strains had gene copies of GH29, GH95, GH20, GH2, GH35 and GH85. Only one genome was available for Viridibacillus spp. and this genome had GH33 and GH35. Pathogenic Clostridium spp., such C. perfringens, have previously been shown to degrade mucins 35 , but little information exists on mucin degradation by commensal Clostridium spp. Of the 14 C. butryicum genomes, we found that only one of the genomes harbored GH33 and none of the C. sporogenes or undefined Clostridium spp. possessed GH33 (Fig. 6H). Compared to other species, commensal Clostridium spp. had only a few mucinassociated GHs, including GH16, GH95, and GH42. These profiles suggest that commensal Clostridium spp. are unlikely to be involved in substantial mucin degradation. Based on our genome analysis, we predicted that commensal Clostridium spp. could not degrade intact mucus and use mucus to enhance growth. To address this question, we examined the growth of several Clostridium spp., including C. butryicum CB, C. symbiosum ATCC 14940, C. inoculum ATCC 14501, C. clostridiforme ATCC 25532, and C. sporogenes DSMZ 795 in media with www.nature.com/scientificreports/ or without mucus (Fig. 6I). Consistent with our analysis, none of the Clostridium spp. had enhanced growth with mucus. Our genome analysis indicated that Blautia coccoides possessed multiple GHs involved in mucin degradation and we predicted that this strain would be capable of using mucin glycans as the sole carbon source. Similar to our GH profile, we found that B. coccoides had statistically significant growth with mucus compared to media without mucus. Finally, we examined Lactobacillus, which according to our genome analysis only have 1-4 mucin-associated GHs and do not harbor GH33. We grew Lactobacillus gasseri ATCC 33323, L. johnsonii ATCC 33200, L. brevis ATCC 27305, and L. acidophilus ATCC 4796 in media with and without mucus and found that mucus did not enhance the growth of many Lactobacillus spp. (Fig. 6J). These data provide a comprehensive analysis of mucin-associated GH profiles within commensal gut microbes and highlight that only specific gut strains have mucin-degrading capacity.

Discussion
To survive in the ever-changing environment of the gastrointestinal tract, gut microbes must be adept at foraging for nutrient sources. One way microbes deal with the varying availability of dietary carbohydrates is to forage glycans in the host mucus layer 3 . Mucin glycans are degraded by the sequential action of multiple mucin-associated GHs 30 . Sialic acid residues that terminate mucin glycans have been proposed to limit glycan degradation, thereby www.nature.com/scientificreports/ protecting the mucin glycan structure. Mucin glycans are also commonly terminated by fucose residues. As a result, mucin degrading microbes commonly encode sialidases and fucosidases to remove the terminal glycan structures, which then allow access to the extended core structures. Freed monosaccharides can then be used by the mucin-degrading bacteria themselves or scavenged by other bacteria 2,7 . Our comparative genomic approach has revealed that many gut microbes found in healthy individuals possess GH33 and other mucin-degrading GHs, indicating that these microbes have the capacity for extensive mucin degradation. Consistent with other findings, we found that Akkermansia harbored 9 different GH families and A. muciniphila ATCC BAA-835 was able to grow in a chemically defined medium with porcine intestinal mucus as the sole carbon source. We also identified several commensal bacteria with mucin-associated GH profiles and These genomic data suggest that several gut microbes may be able to completely degrade intestinal mucin glycans. Two microbes we identified that appear to possess the ability for extensive mucin degradation are Victavallales bacterium and Fermentimonas caenicola. The family Victivallaceae has only two cultured representatives: Victivallis vadensis strain CelloT and the uncharacterized strain NML 080035 36 . These microbes are Gram-negative and anaerobic. There are also 16S rRNA gene sequences from uncultured Victivallaceae. Culturable V. vadensis can use galactose as a primary carbon source 36 . In our genome analysis, we found that Victivallales bacterium possessed GH33, GH16, GH29, GH95, GH20, GH2, GH35, GH42, GH89 and GH129. Several of these glycosyl hydrolases are galactosidases (GH2, GH20, GH35, and GH42). As a result, we predict that Victivallales bacterium may cleave mucin galactose residues to use as a carbon source. F. caenicola was isolated from the stool of a healthy Senegalese child as part of a study aiming at cultivating human gut microbes 37 . F. caenicola is Gramnegative, facultatively anaerobic bacillus. Beye et al. found using an API 50 CH strip that F. caenicola also grows with galactose 37 . In our genome analysis, we found that F. caenicola harbors several galactosidases (GH2, GH20, and GH42), suggesting that removal of galactose from glycan core structures could promote F. caenicola growth. Our analysis also identified that the F. caenicola genome had gene copies of GH29, which contains a fucosidase. Interestingly, Beye et al. found that F. caenicola was unable to grow with d-fucose or l-fucose. Other microbes, like B. bifidum, have been shown to cleave fucose and cross-feed other bacteria, like Eubacterium hallii, which cannot degrade complex glycans 38 . It is possible that fucose release by F. caenicola may promote the growth of other microbes. Although there are no studies examining growth of Victavallales bacterium or F. caenicola with other mucin-related sugars or intact mucins, based on the GH profile, we predict that these microbes could use intact mucins and potentially use other mucin glycan sugars as carbon sources.
Although we focused on microbes with GH profiles indicative of more complete mucin-degradation, it is well known that microbes can act in concert to break down glycan structures. In pioneering studies in the 1980s, Hoskins et al. examined fecal bacteria grown in mucin-based medium and found that 1% of the microbiota was able to use mucin as a carbon source, including the genera Bifidobacterium and Ruminococcus 39,40 . Recent in silico analysis, which is not reliant on culturing techniques, has demonstrated that up to 86% of the human gut microbiota encode genes for cleavage of mucin glycans 23 . We also found that 62% of all microbes and 83% of human gut microbes in the CAZy database encode genes for mucin-degradation. These studies, as well as our own, have found that only specific bacterial species have a sufficient repertoire of enzymes to disassemble complex mucin glycans and that the complete degradation of mucin often requires the action of several bacteria. Our analysis reveals that many bacteria possess multiple gene copies of GHs targeting internal glycans. These findings suggest that mucolytic bacteria with GH33 may initiate glycan break-down and then the less-specialized bacteria with internal glycan GHs can participate in degradation.
Based on our studies, we believe that the core GH-ome for mucin degradation includes GH33, 29, 95, and 20/35. More extensive degradation of internal glycans incorporates GH84/85/89 and 101. Our genome analysis suggests that mucin degrading microbes possess > 4 mucin-associated GHs. Additionally, microbes that extensively degrade mucin, like A. muciniphila, B. bifidum, and B. thetaiotaomicron, possess > 9 mucin-associated GHs. Despite the fact that many commensal microbes are not capable of extensive mucin degradation, the GH profile of bacteria such as Clostridium indicates that they could contribute to degradation when paired with another bacteria. For example, if A. muciniphila removes sialic acid, several Clostridium species could remove fucose with GH95. After A. muciniphila removes N-acetyl-glucosamine, almost all Clostridium could remove galactose with GH2 or GH42. These cross-feeding events likely occur in vivo and contribute to the health of the mucus layer. Future studies using mucus cross-feeding will likely shed light into the complex interplay because mucin-degrading microbes.
Since mucin glycan degradation disrupts the first protection of the host mucus layer, host glycan foraging by mucolytic bacteria is commonly considered an initial stage in pathogenesis. While this notion likely only holds true for excessive mucin degradation, many consider mucin-glycan break-down to be a normal process and a key strategy for mucus-associated microbes. Given the continuous turnover of the epithelial cells and mucus in the human gastrointestinal tract, mucin degradation by commensal gut microbes is not likely to contribute to barrier dysfunction. Additionally, the capacity to degrade mucin is particularly important for early colonizers of the gut. Infants are commonly colonized with mucin degrading Bifidobacterium bifidum, B. longum subsp. infantis, and B. breve [41][42][43][44] . One study in Sweden identified the establishment of mucin-degrading bacteria during the first months of life 44 . We speculate that mucin glycan degradation gives colonizing microbes an advantage after the termination of breast milk and allows them to exert their beneficial influence on gut homeostasis.
One potential limitation of this work is that it is difficult to predict the exact specificity of a CAZyme based on family membership 45 . However, substrate categories can be broadly inferred within the CAZy families, even if the precise specificity of each protein in the family is challenging to predict 3 . An advantage of this type of genome www.nature.com/scientificreports/ analysis is that it does not require complex culturing, which can be challenging since many intestinal microbes are not able to grow in classical laboratory media. Genomic strategies have now been widely applied and are bringing new information about the diversity and function of human gut microbiota. Our genomic characterization has shed light on commensal gut species with mucin-degrading properties. We believe this work enhances the foundation for examining mucin-degradation within the human intestine.