Enrichment and characterization of human-associated mucin-degrading microbial consortia by sequential passage

Abstract Mucin is a glycoprotein secreted throughout the mammalian gastrointestinal tract that can support endogenous microorganisms in the absence of complex polysaccharides. While several mucin-degrading bacteria have been identified, the interindividual differences in microbial communities capable of metabolizing this complex polymer are not well described. To determine whether community assembly on mucin is deterministic across individuals or whether taxonomically distinct but functionally similar mucin-degrading communities are selected across fecal inocula, we used a 10-day in vitro sequential batch culture fermentation from three human donors with mucin as the sole carbon source. For each donor, 16S rRNA gene amplicon sequencing was used to characterize microbial community succession, and the short-chain fatty acid profile was determined from the final community. All three communities reached a steady-state by day 7 in which the community composition stabilized. Taxonomic comparisons amongst communities revealed that one of the final communities had Desulfovibrio, another had Akkermansia, and all three shared other members, such as Bacteroides. Metabolic output differences were most notable for one of the donor’s communities, with significantly less production of acetate and propionate than the other two communities. These findings demonstrate the feasibility of developing stable mucin-degrading communities with shared and unique taxa. Furthermore, the mechanisms and efficiencies of mucin degradation across individuals are important for understanding how this community-level process impacts human health.


Introduction
Everyone has a personalized intestinal micr obiome, whic h assists in normal digestion that is cultivated through their lifetime and tightly linked to both health and disease .T he composition of each individual's gut microbiome is highly dependent on nutrient intake (Mohammadkhah et al. 2018 ).Although individual microorganisms can be associated with some disease states, other diseases are associated with microbial community features, examples include organisms that are efficient at dietary energy extraction associated with obesity (Liu et al. 2021 ) and organisms that contain sulfur metabolic machinery associated with colorectal cancer (Wolf et al. 2022 ).These micr oor ganisms may affect host health through the production of metabolites that interact with local and distal tissues (P ar ada Venegas et al. 2019 ).For example, the production of short-chain fatty acids (SCFA) by fermentative gut bacteria has been implicated in protection of the epithelium, imm une r egulation, and epithelial cell pr olifer ation and differ entiation (Bilotta and Cong 2019 ).The mechanisms that maintain and bolster these functions of the microbiome are poorly understood but of gr eat inter est because of their potential to modulate health.
Micr oor ganisms within the intestinal tract require basic resources including carbon and energy for growth and survival, which the y deri ve primarily from the host diet (Leeming et al. 2019 ).Dietary fibers, as one example, compose a broad array of pol ysacc harides exclusiv el y metabolized by diverse members of the gut microbiota.Recent evidence indicates that dietary fibers with more complex chemical structures necessitate a diverse array of enzymes for digestion often produced by specialist microorganisms (Lindemann 2020 ).These specialists enable a myriad of micr obial inter actions as they liberate smaller sugar subunits and produce fermentation products that are both used by other microbiota members and by host cells (Koropatkin et al. 2012 ).Comparativ el y simple fibers, suc h as fructooligosacc harides and type II r esistant starc h, r equir e fe wer enzymes for degradation, making these pol ysacc harides mor e widel y accessible to man y gut or ganisms (Cantu-Jungles et al. 2021 ).Man y individual micr oor ganisms within complex microbial communities also require exogenous forms of vitamins and amino acids due to underlying auxotrophies (Yu et al. 2022 ).While in the nativ e gastr ointestinal tr act, these essential nutrients may be derived from the host diet; in the cultivation of fecal microbial consortia, supplementation drastically affects final community outcomes (Zengler andZaramela 2018 , Yao et al. 2020 ).Auxotrophy for amino acids in vitro can be alle viated thr ough addition of free amino acids , peptides , or protein supplements.Ho w e v er, the ability of an organism to scavenge amino acids v aries; onl y some or ganisms pr oduce exopeptidases capable of breaking down complex proteins into smaller fr a gments, suc h as dipeptides or free amino acids, whic h ar e then transported into the cell (Zhang et al. 2022 ).T herefore , variability in host diets leads to a difference in dietary fibers and amino acid sources consumed, ultimately conferring advantages to specific sets of micr oor ganisms ca pable of br eaking the discr ete bond types and helping to shape the uniqueness of each microbiome.
Despite this individuality in diet, healthy humans secrete m ucins, a famil y of gl ycopr oteins decor ated with O-linked carbohydr ates in v arious configur ations pr oduced by specific epithelial cells in the intestinal, r espir atory, and genitourinary tr acts.Although primarily known for their role as a barrier against invasion, the high pol ysacc haride content of m ucins can potentiall y serve as a rich and reliable nutrient source for microorganisms (Tailford et al. 2015 ).Mucin structure is complex, fundamentally containing a serine-thr eonine-pr oline-ric h pr otein bac kbone, an N -acetylgalactosamine (GalNAc) O-linked to this protein, an array of sugars linked to the GalNAc, and a final terminal unit (Corfield 2015 ).The breadth in structures lies in the sugar array, which N -acetylglucosamine (GlcNAc).Furthermore, these structures can be capped by a sulfate group, fucose, or Nacetylneuraminic acid (sialic acid; NeuAc).Given this diversity of carbohydr ate c hemical linka ges and a limited number of amino acids, only a subset of gut microorganisms possess the functional capacity to hydrolyze and utilize mucins for growth (Glover et al. 2022 ).
Some intestinal micr oor ganisms, suc h as the nutritional generalist Bacteroides thetaiotaomicron , can switch between fermentation of dietary pol ysacc harides and m ucin gl ycans (Ravc hee v et al. 2013 ).During fermentation of either dietary fibers or mucin, byproducts , including the SCFA acetate , ar e pr oduced (Maho w ald et al. 2009 ).These metabolites stimulate mucin production by goblet cells in the gut, resulting in an intact mucus barrier between human and bacterial cells along the intestines (Willemsen et al. 2003, Adamberg et al. 2018 ).In fact, ther e a ppears to be a direct connection between the abundance of dietary fiber in the diet and colonic mucin thickness (Brownlee et al. 2003, Hedemann et al. 2009, Earle et al. 2015, Desai et al. 2016 ), likely mediated by the microbiota.
A smaller subset of organisms in the gastrointestinal tract, like the m ucin-degr ading specialist Akkermansia muciniphila , appear to depend on direct mucin fermentation for growth (Derrien et al. 2004 ).Ther e ar e man y carbohydr ate-activ e enzymes (CAZymes) responsible for metabolizing complex mucin glycans, where each enzyme (e.g.glycoside hydrolase-GH) is responsible for hydrolysis of a specific gl ycosidic bond, liber ating specific sugar moieties.Although it is possible that Akkermansia , a specialist that retains a full suite of m ucin-degr ading C AZymes , is capable of filling this m ucin-degr ading nic he and outcompeting other or ganisms for this resource, the incidence rate is only 50%-80% of the human population (Geerlings et al. 2018 ).Given that much of the established abundances of micr oor ganisms in the human gut are confined by available technology, including in vivo sampling, DNA extraction methods, primer choice, and detection limits (Brandt andAlbertsen 2018 , Abellan-Schneyder et al. 2021 ), Akkermansia may in fact be present in all gut communities.Alternatively, it is possible that gut micr obial comm unities divide the metabolic labor in which partial hydrolysis of mucin releases mono-or disaccharides that can then be fermented by other organisms (Bunesova et al. 2018 ).In this second scenario, the activities of secondary consumer microbiota would be supported by the intestinal mucin and further stim ulate m ucin pr oduction, ther eby limiting pathogen invasion.To determine whether mucin metabolism would be mo-nopolized by a single member or support a complex community, we sought to c har acterize the indigenous microbial communities capable of growth on mucin across multiple individuals.
Here, we use sequential transfer in media with mucin as the sole carbon source to investigate the diversity, composition, and taxonom y of nati v e gut micr obial consortia fr om thr ee human donors.We expected that each consortium would r eac h a steady state and, while some taxa may be shar ed acr oss donors, eac h community would be distinct.Additionally, within each donor, we expected replicate lineages to be similar due to strong, nutritional selection imposed by the m ucin gl ycopr otein.Ho w e v er, the amino acid sources provided will influence ov er all comm unity composition within each individual.By using a top-down appr oac h, we identify a core and unique set of micr oor ganisms involved in mucin metabolism with distinct SCFA production profiles.

Soluble mucin preparation
Pr epar ation of soluble mucin was based on Kirmiz et al. ( 2020 ) and added to a final concentration of 0.5%.Briefly, soluble porcine gastric mucin type III (Sigma-Aldrich, USA) w as prepared b y autoclaving a 5% (w/v) solution in 0.01 M phosphate buffer (6.5 mM KH 2 PO 4 and 3.5 mM K 2 HPO 4 ), dialyzing with a 12-14-kDa membr ane (Spectr a/Por 4; Spectrum Labor atories, Ranc ho Dominguez, C A) o vernight in 10 dialysis volumes of 0.01 M phosphate buffer, centrifuging for 30 min at 30 000 rcf, and filtering through a 0.45μm syringe filter follo w ed b y a 0.2-μm syringe filter (Whatman GE Healthcar e Life Sciences, Chica go, IL).Sterile soluble m ucin was then added to the fermentation medium at a final concentration of 0.5%.

Mixing of mucin solution and fermentation medium
Briefly, concentrated fermentation medium (2X) was aerobically pr epar ed and diluted to a working concentration with soluble mucin and sterile water.Prior to transfer to Balch tubes, buffermucin mix was stored loosely covered in the anaerobic chamber (85% N 2 , 10% CO 2 , and 5% H 2 ) for 48 h to outgas.Each tube r eceiv ed 5 ml fermentation medium except the initial thr ee tubes which each received 4 ml.Fermentation medium containing mucin was aseptically distributed into Balch tubes in a 10% CO 2 , 5% H 2 , and 85% N 2 atmospher e befor e sealing with butyl rubber stoppers.After pr epar ation, tubes wer e stor ed at 37 • C prior to inoculation to detect any contamination.Immediately before inoculation, 0.05 ml ATCC vitamin mix (final 1% v/v, manufacturer's n umber: ATCC MD-VS, Hampton, NH) was ase ptically ad ded to each tube.An additional three tubes were maintained for the full duration of the experiment at 37 • C without inoculation as negativ e contr ols.

Inoculum prepar a tion
Appr oximatel y 4 ± 1.5 g of fecal material was collected from three healthy donors r epr esenting two female and one male donor ranging from 20 to 45 years of age.Donors had not r eceiv ed an y antibiotic treatment within 3 months and wer e gener all y in good health.The protocol for collecting human feces was a ppr ov ed by the Institutional Re vie w Board of California State Univ ersity, Northridge (#1516-146-f).Donors were asked to collect two Hershey's kiss size samples ( ∼4.2 g each) into separate preweighed conical vials using sterile wooden spatulas.Samples were maintained on blue ice during transfer following best practices for retaining viable organisms (Owens et al. 2013 ) and r eceiv ed within 3 hours of e v acuation.Fecal material was diluted 1:4 w/v with 1X fermentation medium containing mucin as described abo ve .Tubes w ere v ortexed for 20 s, returned to ice for up to 5 min, and vortexed again for 20 s .T he fecal slurry was then poured over 4-ply sterile cheesecloth, and the flow through was collected with a syringe as pr e viously described (Yao et al. 2020, Romero Marcia et al. 2021, Tuncil et al. 2017, 2018 ).To r epr esent day 0 (the initial inoculum), 500 μl of this fecal slurry was centrifuged at 10 000 × g for 5 min and the cell pellet stored at −20 • C until thawing for DNA extraction.Each tube containing 4 ml of fermentation medium then r eceiv ed 1 ml of fecal slurry totaling a 1:20 fecal dilution.The initial inoculation tubes r eceiv ed an additional 50 μl of 0.25 g/ml cysteine to reduce the initial inoculum for a final total cysteine concentration of 2.4 mM.Tubes were placed in a static 37 • C incubator inside of an anaerobic Coy chamber (10% CO 2 , 5% H 2 , and 85% N 2 atmosphere) for 23-25 h before sample collection and transfer as described below.

In vitro sequential fecal fermentation
The sequential cultivation experiment was continued for 10 consecutive da ys , and each donor was cultured in triplicate without intermixing the r eplicates, r esulting in thr ee independent linea ges for each donor and amino acid source (Yao et al. 2020 ).Daily, immediately prior to inoculation, 50 μl of ATCC vitamins (final 1% v/v, manufacturer's number: ATCC MD-VS, Hampton, NH) was added to each 5 ml fermentation medium.Due to settling of micr oor ganisms fr om static incubation, tubes wer e gentl y inv erted 5x prior to sampling.Subsequently, as previously described (Yao et al. 2020 ), 50 μl of the incubated culture was tr ansferr ed to the corresponding tube containing 5 ml fermentation medium and incubated in the Coy chamber as indicated abo ve .
Daily, 500 μl of each incubated culture was collected and centrifuged at 10000 × g for 5 min.The cell pellet was stored at −20 • C until thawing for DNA extraction.On the final day, after centrifugation, the supernatant was tr ansferr ed to a separate microfuge tube and stored at −20 • C until thawing for SCFA detection as indicated below.

DN A extr action and sequencing
DN A w as extr acted fr om cell pellets collected abov e using the Zymo DNA Micr opr ep Kit #D4301 (Zymo Researc h Cor por ation, USA), with the following modifications.For the initial inoculum, 20% of the resuspended culture was processed to prevent overloading the kit.After resuspension and transfer to the Bead-Bashing lysis tubes, samples were placed on a horizontal vortex adapter at max speed for 10 min.DNA concentrations were read on a Qubit 2.0 fluorometer with the high-sensitivity kit, following manufacturer's protocol.A volume of 1 μl of each eluted sample was used in pol ymer ase c hain r eaction amplification of the variable region 4 of bacterial and archaeal 16S ribosomal RNA genes with barcoding primer set 515/806 based on the original Earth Microbiome Project protocol (Caporaso et al. 2011, Earth Micr obiome Pr oject 2024 ).Primers used wer e 515F (P ar ada et al. 2016 )-GTGYCAGCMGCCGCGGTAA and 806R-GGA CTA CHV GGGTWTCTAAT with barcoding on the 806R primer.PCR mixture conditions and thermal cycling steps are previously described (Herman et al. 2020 ).Triplicate PCR r eactions for eac h sample were combined, quantified with Quant-iT PicoGreen (Invitrogen) on a SpectraMax3 (Molecular Devices, USA), and pooled at equimolar concentrations before single-tube cleaning with the Invitrogen Gel and PCR Clean-up Kit (Invitrogen) and sequencing using an Illumina MiSeq (v2, 2 × 150 bp) instrument (Illumina, USA).

DNA sequence analysis
Sequences were analyzed following the QIIME-2 Atacama Desert pipeline (Qiime2Docs), briefly described her e.Sequences wer e denoised, der eplicated, and c himer as wer e r emov ed with Dada-2 (Callahan et al. 2016 ) with the following parameters: -p-trim-left 5 -p-trunc-len 150.Amplicon sequence variants (ASVs) were classified using sklearn and the silva 138.99 database (Quast et al. 2012 ).PCR blanks were used to remove contaminating ASVs following the decontamR pipeline (Davis et al. 2018 ) using a pr e v alencebased str ategy.Giv en that the experiment occurr ed with triplicate lineages, after combining the two datasets, ASVs that were present in less than two of the total samples were removed.Subsequentl y, samples with fe wer than 5000 r eads, r epr esenting the extr action blanks, wer e r emov ed.This r esulted in a total of 192 samples for downstream analysis, which included six samples rerun to compare differences across sequencing runs.QIIME-2 was used to calculate various α-diversity (total ASV counts and Shannon diversity index) and β-diversity (Bray-Curtis) metrics.

Determination of fatty acids
Final consortia cultures at day 10 were analyzed for SCFAs of acetate , propionate , butyrate , and branched-chain fatty acids (BC-FAs) of isobutyrate and isov aler ate.Samples wer e measur ed on a GC-FID system (GC-2030, Nexis, Shimadzu Cor por ation, K yoto , Ja pan).Cultur es wer e centrifuged (10 000 × g ) for 5 min to r emov e cell debris, mixed 4:1 with an internal standard mixture containing 4-methylvaleric acid, phosphoric acid, and copper sulfate pentahydr ate, and supernatants wer e tr ansferr ed to 2 ml scr e wthread autosampler vials.Volatile compounds were quantified on a GC-FID with a wax column (RESTEK Stabilwax-DA #11023, Bellefonte, PA).Helium was used as a carrier gas for the column at a flow of 1.41 ml/min.Samples were injected in a split mode (5 ul injection volume, 1:5 split ratio) under a linear velocity of 35 cm/s at 230 • C. The oven temperature was set as below: initial temperature at 40 • C for 2 min, following a ramp to 130 • C in 3 min, and then another ramp to 195 • C in 6.5 min, holding at 195 • C for 4 min.

pH and OD testing
On the final day of sampling, tubes were removed from the anaerobic chamber and opened aerobically.Cultures were transferred to a 15-ml conical vial and 100 μl of the culture was used to determine OD 600nm on a Biophotometer plus spectrophotometer (Eppendorf, German y).Cultur es with an OD 600nm > 1.0 were diluted with sterile water and r er ead.The pH was tested on the remaining volume using an accumet pH meter (Fisher Scientific, USA).

Sta tistical anal ysis
For pH, OD 600nm , SCFA, and alpha diversity analysis, statistically significant differences were calculated by Tuk e y's honest significance multiple comparisons test using R 4.1.1(R Foundation for Statistical Computing, Vienna, Austria).Comparisons were considered significant if corrected P -values were less than .05.For beta diversity analysis, dissimilarity matrices generated in Qiime-2 were used to calculate statistically significant differences with ADONIS from the vegan package in R 4.1.1.Comparisons were considered significant if the P -value of the F statistic was less than .01.Symbol style for figures: nonsignificant (ns), 0.05 ( * ), 0.01 ( * * ), 0.001( * * * ), and < 0.0001( * * * * ).

Results
To better understand the ecology of mucin degradation as a comm unity le v el pr ocess by the human gut microbiota, we performed 10-day sequential fecal enric hment fr om thr ee human donors in triplicate, using purified porcine gastric mucin as the sole carbon and energy source.As a secondary variable and to determine the impact of amino acid source on micr obial comm unity composition, we used two amino acid sources during community development in parallel enrichments for each donor.For each day, cultur es wer e tr ansferr ed at a high (10 −2 ) dilution rate and cell pellets were collected to monitor community development using 16S rRNA gene amplicon sequencing.On the final day, the OD 600nm and pH of cultures was tested, and the spent culture supernatant was collected and used to determine SCFA production.

Amino acid source does not govern growth or metabolic outputs
In cultur es fr om all thr ee donors, r obust gr o wth w as evident b y changes in media turbidity after each day of incubation, regardless of amino acid treatment.On the final day of growth, all comm unities gr e w to OD 600nm > 1 and ther e wer e no significant differences in OD 600nm within a donor's communities on either amino acid source ( Supplemental Fig. S1 ).Howe v er, acr oss donors, comm unities fr om donors one and thr ee gr e w to significantl y higher OD 600nm ( ∼1.4-1.6)than donor two ( ∼1.1) suggesting difference in gr owth yield (Fig. 1 A).Similarl y, dr ops in pH wer e also gener all y gr eater for comm unities fr om donors one and thr ee, except for on the amino acid mixture where donor two and three communities both dropped ∼0.3 pH units (Fig. 1 B).Interestingly, communities from donor two had a significantly different drop in pH between the two amino acid sources, with the culture growing on mixed amino acids dropping to pH 6.1 and the culture grown on tryptone to 6.3 ( Supplemental Fig. S1 ).This potentially suggests ammonification of the medium fr om degr adation of the more abundant proteins in the community cultivated on tryptone.

SCFA output is unique to each community
SCFAs are main metabolic byproducts of gut microbial fermentation of complex carbohydrates .T herefore , to determine the metabolic output of each community during mucin fermentation, the SCFA profile was assayed on day 10 of sample collection (Fig. 2 ).Ov er all, micr obial consortia fr om donors one and thr ee produced the highest total concentration of SCFAs ( ∼20 mM, sum of acetate , butyrate , and pr opionate) wher eas the donor two comm unity pr oduced lo w er concentr ations (14 mM), irr espectiv e of amino acid source .T hese differences wer e lar gel y driv en by acetate and pr opionate, wher e cultur es fr om donors one and thr ee pr oduced mor e than donor tw o.Ho w e v er, differ ences of specific SCFAs generated by each donor's communities were dependent on amino acid source as discussed below.
When cultivated on tryptone, communities from donors one and thr ee pr oduced equal amounts of propionate (3-4 mM), whereas the donor two consortium pr oduced significantl y less (0.5 mM).In contr ast, ther e was no significant differ ence in the production of acetate (9-14 mM) or butyrate (2.5-3.5 mM) across the donor communities, when grown on this amino acid source.Although the same general trend of SCFA production was observed in cultures grown with the amino acid mix, the differences between communities was more pronounced.In these communities, donor three microbiota produced the greatest amounts of propionate (5 mM), follo w ed b y donor one microbiota (2.5 mM), with donor two's microbiota generating low concentrations (0.5 mM).Concentrations of acetate produced by both donor one and donor three communities were similar (13-15 mM), with donor two's producing lower concentrations (9 mM).The production of butyrate in amino acid mix-consuming communities mirr or ed those cultivated on tryptone, where there was no significant difference in the amount of butyrate produced (2.5-3.5 mM).Furthermor e, we detected tr ace amounts of the BCFAs , iso valerate, and isobutyrate ( Supplemental Fig. S2 ).These minor BCFAs reflected the patterns observed for SCFAs but had very low concentr ations.Ov er all, these data coupled with the OD 600nm results, suggest mucin fermentation by microorganisms from all three donors.

rRNA amplicon sequence quality and taxonomy
Across two MiSeq Illumina sequencing runs, a total of 13 120 305 sequences wer e obtained.Eac h run was independentl y denoised, der eplicated, and c himer as wer e r emov ed.Amplification blanks were used to remove contaminating sequences, where 66 ASVs from the first sequencing run and 9 ASVs from the second were r emov ed.After r emov al of sequences pr esent onl y in a single sample and extraction blanks, the final number of sequences analyzed constituted 92.9% of the original sequence total (12 184 363 sequences).To determine whether sequencing run bias would influence the results, six samples r epr esenting the initial inoculum and mid-to-end timepoints from both donors one and two were sequenced on both sequencing runs and community composition was compared ( Supplemental Table S2 ).The differences in β-div ersity (Br ay-Curtis dissimilarity) acr oss samples sequenced on the first and second run were not significantly different as determined by adonis analysis ( R 2 = 0.003, P = .95).

Final community composition and diversity is dependent on initial inoculum
Six m ucin-degr ading comm unities wer e established fr om thr ee human donors to test whether mucin deterministically selects for specific community compositions both across and within individual human donors.To determine community composition and div ersity thr ough time, standard ecological div ersity metrics wer e calculated based on profiling of the V4 region of the 16S rRNA gene sequence.As expected, initial fecal communities were compositionall y unique fr om one another although we could not statistically test this since only a single sample was collected from  eac h donor.Subsequentl y, since we de v eloped triplicate linea ges for each amino acid sour ce, w e could statistically test for differ ences acr oss and within donors thr ough time.Acr oss donors, eac h comm unity was compositionall y distinct thr oughout de v elopment regardless of nitrogen source ( Supplemental Table S3 ).Within each donor, microbial community development follo w ed the same trajectory across the replicate lineages regardless of amino acid source (Fig. 3 ), suggesting that community composition is gov erned mor e by mucin than by amino acid source under these conditions .T hese results also show that community assembly within a donor is replicable under high dilution pr essur e with mucin.
To determine whether communities were persistent throughout time, we first calculated the Bray-Curtis dissimilarity and compar ed differ ences between consecutiv e days within each donor ( Supplemental Fig. S3 ).Given that there were no significant differ ences within eac h donor's comm unities between amino acid sour ces, w e then grouped amino acid sources for each day to identify community plateau for each donor ( Supplemental Table S3 ).
Comparing the communities at each day to the previous day indicates that the differences between communities after day 6 or 7 was not significant ( Supplemental Table S4 ), suggesting communities had stabilized by this point.
From an alpha diversity perspective, similarities and differences within and across donor comm unities wer e observ ed depending on diversity metric ( Supplemental Table S5 ).For example, by a ppr oximatel y day 5, ric hness (ASV count) of each community plateaued (Fig. 4 , Supplemental Table S6 ).In contrast, the Shannon diversity metric, which accounts for both richness and e v enness, suggests that donor one communities leveled off earlier than donors two and three (Fig. 4 , Supplemental Table S7 ).Inter estingl y, the amino acid source affected community richness for Donor 1 on days 2 and 4, and comm unity e v enness for Donor 2 on days 4, 6, 7, and 9, but not for the Donor 3 community ( Supplemental Tables S5 , S8, and S9 ).Ov er all, these r esults demonstrate that although the timing differs across donors, all six communities reached a steady-state.
Each donor started and ended with a different final number of ASVs (Fig. 4 ), where donors one and two started with the gr eatest div ersity at 171 ± 8 ASVs and 164 ± 6 ASVs, respectiv el y.These comm unities plateaued to communities with 63 ± 7 (donor one) and 48 ± 4 (donor two) ASVs.Inter estingl y, donor three had much lo w er initial diversity, starting at about half the initial number of ASVs (96 ± 5) compared to donors one and two, and r eac hed a corr espondingl y lo w er number of 31 ± 2 ASVs.Ho w e v er, the % loss was a ppr oximatel y equal across donor sam-F igure 3. Principal coor dinates of anal ysis plots of Br ay-Curtis dissimilarity between comm unities for eac h donor ov er 10 days with supplementation of either amino acids or tryptone.Donors are represented by fill color, days by hue of the dominant color, amino acid supplementation b y bor der color, and lineage by shape.Principle coordinates 1 and 2 (A) and 1 and 3 (B) indicate that the community from each donor follows a distinct trajectory.
ples, wher e eac h final comm unity r epr esented ∼30% of the initial community (Donor 1, 35.5%; Donor 2, 29.7%; and Donor 3, 31.8%Supplemental Fig. S4 ).Furthermore, the point at which the loss in ASVs plateaued differed among the comm unities; comm unities belonging to donors one and two took 4 and 5 da ys , r espectiv el y, to r eac h this plateau, whereas donor three communities took 3 days ( Supplemental Table S10 ).Taken together, these data suggest that mucin can sustain diverse fermenting consortia over sequential dilutions in a manner related to the initial community structure and diversity.
In addition to shared features of the three communities, each also had unique members.In donor two communities, a bloom of Desulfobacterota a ppear ed at day 3 and remained abundant throughout the experiment.On the other hand, donor three displayed a bloom of Akkermansia that had the highest abundances on days 4 and 5, but whose population declined until the termination of the experiment on day 10.To identify less-obvious associations of individual taxa with their donor inoculum, we compared the taxa of the day 10 microbial communities using linear discriminant analysis effect size (LEfSe, Supplemental Fig. S6 , Supplemental Table S12 ).Inter estingl y, some of the or ganisms that wer e shar ed acr oss the thr ee donors wer e discriminants for a specific donor due to the differential abundance patterns.At the genus le v el, in addition to Desulfovibrio, other discriminating taxa included Faecalitalea , Merdibacter , and Erysipelatoclostridium for the donor two consortium ( Supplemental Fig. S6 ).Akkermansia was a discriminant of the donor three community, as were Ruminococcus and Oscillospiraceae .Although donor one did not have an y ob vious associations with specific taxa in the visualizations, the FCS020_group and Howardella (in the Lachnospiraceae family), Prevotella , and Megamonas were all discriminants of this donor.Inter estingl y, at the ASV le v el, eac h donor had multiple discriminating ASVs belonging to the genus Bacteroides and family Lachnospiraceae ( Supplemental Table S12 ).

Discussion
Mucins provide an endogenous nutritional resource in the human gut upon which additional nutrients from the hosts diet are overlaid in vivo .As suc h, m ucins ar e one of the most consistent nutritional r esources av ailable to gut micr oor ganisms.Although man y individual m ucin-degr ading bacteria hav e been identified (Tailfor d et al. 2015 ), w e w anted to determine the selectivity of mucin across distinct donor communities and begin to understand the fundamental ecology of m ucin degr adation by gut microbiota.Giv en that eac h human hosts a taxonomically distinct gut microbiota, we used fecal samples from three unrelated donors to select for m ucin-degr ading comm unities ov er 10-days of sequential passage and assess community composition and metabolic output using 16S rRNA gene sequencing and SCFA quantification, respectiv el y.Although m ucin is a gl ycopr otein, the ability of an or ganism or community to access and metabolize the relatively simple protein backbone likely varies .T herefore , to determine whether the provision of amino acids or oligopeptides would differ entiall y impact community structure, replicate communities were developed on either tryptone or a proteinogenic amino acid mixture.At the conclusion of the experiment, we obtained steady-state mucindegr ading comm unities with distinct compositions across donors and minor compositional differences within a donor cultivated with either amino acid source.While similarities in final community composition did exist, notable taxonomic differences across donors included the presence of the mucin-degrading specialist Akkermansia in the donor three community and the presence of the sulfate-reducing Desulfovibrio in the donor two community.Metabolic output differences were most apparent for donor two communities, as they produced significantly less acetate and Figure 4. Bacterial diversity at each passage for all three donors (Donor 1, D1; Donor 2, D2; and Donor 3, D3) when supplemented with amino acids provided as an equimolar mix ("AminoAcids") or peptides ("Tryptone").Species richness as measured by the number of ASVs (A) and evenness as measured by Shannon entropy (B) plateaued for all three communities.Mean, first, and third quartiles are represented with fill color indicating amino acid source.Statistically significant differences at day 10 are calculated by Tuk e y's multiple comparisons test.Symbol style: nonsignificant (ns), 0.05 ( * ), 0.01 ( * * ), 0.001( * * * ), and < 0.0001( * * * * ).
propionate than the other two communities .T hese findings reinforce the concept that the gut microbiota is taxonomically distinct, yet retains many microbial members that are capable of growth on complex carbohydrates such as mucin.Although all thr ee comm unities had r obust gr owth on m ucin, the taxonomic differ ences r esult in distinct metabolic outputs, suggesting that the mechanisms and efficiencies of m ucin degr adation ar e important for understanding how this comm unity-le v el pr ocess impacts human health across individuals.
Notable taxonomic similarities across developed communities include the pr e v alence of Bacteroides in all thr ee final comm unities despite not a single ASV from this genus being shared across donors.Although some members of the Bacteroides have known ca pacity for m ucin degr adation, suc h as B. fragilis and B. thetaiotaomicron , most are not specialized on this carbon source and instead target dietary fibers (Salyers et al. 1977, Pudlo et al. 2015 ).Ho w e v er, the br eadth of GHs harbor ed by these or ganisms enables subsistence on mucin in the absence of dietary fibers (Glover et al. 2022, Sonnenburg et al. 2005 ) or during strong competition by other community members (Sonnenburg et al. 2006 ).Within the Bacteroides populations we note two interesting observations across donors; one, that each donor's microbiota contained an ASV assigned as B. vulgatus that had an initial bloom and then disa ppear ed fr om the comm unities; and two, ther e wer e differ ent dominant Bacteroides ASVs by day 5 that were maintained through the remainder of the experiment.B. vulgatus is not a known mucin degrader and likely requires other members of the community to liber ate sugars fr om complex substr ates; for example, cocultur es with B. ovatus on inulin have demonstrated growth dependency of B .vulgatus on B .ov atus , likel y mediated thr ough extr acellular GHs (Rakoff-Nahoum et al. 2016 ).The other Bacteroides that dominated donor communities one and two after day 5 could not be classified be yond gen us with the SILVA database.Ho w e v er, manuall y using NCBI BLAST to assign and confirm taxonomy of these sequences ( Supplemental Table S11 ), we found that they closely matched B. caccae (community one), B. thetaiotaomicron (community two), and confirmed the sequence of B. fragilis (community three).All three of these dominant Bacteroides are known to degrade mucin (Desai et al. 2016 ), suggesting that each donor has a unique Bacteroides pr ofile ca pable of m ucin consumption that would enable surviv al in the absence of dietary fiber.It would be interesting to identify the full spectrum of mucin degrading consortia, specifically with respect to known mucin degraders, within a larger study population or across the same individual to determine whether these micr oor ganisms ar e r etained thr oughout the life of the host.
After 10 consecutive days of growth on mucin, each donor community maintained ∼30% of its initial diversity, demonstrating that a complex gl ycopr otein like m ucin can sustain r elativ el y div erse micr obial comm unities.While some members of these comm unities ar e certainl y primary m ucin degr aders, man y of the taxa observ ed ar e not.Instead, div ersity is at least partially maintained by generalist bacteria feeding on mucin sugars liberated by primary degraders and by bacteria cross-feeding on fermentation end pr oducts (Kor opatkin et al. 2012 ).For example, str ains of B. thetaiotaomicron are able to liberate sialic acid from mucins but unable to import and catabolize it for carbon and energy (Marcobal et al. 2011 ).Similarl y, B. bifidum r eleases m ucin monosacc harides, such as fucose, that support other members of the gut microbiota (Bunesov a et al. 2018 ).Faecalibacterium , whic h is a genus of Firmicutes observed in all final donor communities here, is one example of an organism able to use mucin sugars (i.e.GlcNAc) but not intact mucin oligosaccharides (Lopez-Siles et al. 2012 ).In addition to the liberated monosaccharides, other bacteria could be sustained by fermentation end products.For example, acetate produced by A. muciniphila during mucin fermentation can support the growth of the butyrogenic Eubacterium hallii in vitro (Belzer et al. 2017 ).How pr e v alent this type of cross-feeding across the three donor communities is unknown but could play a significant role in the final amount of SCFAs observed here.Although it is difficult to do more than hypothesize about these metabolic interactions based on 16S rRNA gene sequencing alone, these types of syntr ophic inter actions ar e important driv ers of ov er all micr obiome function and gut health.Future work should look more directly at these interactions using a multiomics approach of these communities and also coculture experiments.
Beyond sugar liber ation, primary m ucin degr aders could also sustain other functional groups of bacteria, including sulfate reducers, as suggested by a bloom of Desulfovibrio sp . in the donor two comm unity.Man y m ucin oligosacc harides ar e ca pped with sulfate groups that must be removed using sulfatase enzymes before accessing the underlying sugar moieties (Katoh et al. 2017, Luis et al. 2021 ).Although we do not know the functional capabilities of the strain observed here, other Desulfovibrio spp.from the human gut do not produce sulfatase enzymes capable of this acti vity (Re y et al. 2013 ).This suggests that other members of the comm unity m ust liber ate sulfate to sustain the Desulfovibrio bloom observed in the donor 2 community.For example, several species belonging to the Bacteroides are capable of removing sulfate gr oups fr om m ucin (Luis et al. 2021, Ulmer et al. 2014 ).We propose that the Desulfovibrio in community two reduces liberated sulfate into hydrogen sulfide, as has been pr e viousl y demonstrated in coculture experiments (Rey et al. 2013 ).Although we did not measure sulfate accumulation in this experiment, we expect a buildup of sulfate in the donor 1 and 3 communities because of the absence on known sulfate r educers.Wher e r esidual sulfate in the colon is likely excreted and is unlikely to affect the host (Florin et al. 1991 ), high le v els of hydr ogen sulfide hav e str ong links to colorectal cancer (Wolf et al. 2022 ).Our study points to the importance of future work aimed at understanding the full metabolic profile of mucin-degrading communities.
Although ther e wer e similar tr ends in comm unity composition across all three donors, of particular interest was the community containing the known mucin-degrading specialist, Akkermansia .Akkermansia ar e widel y r egarded as beneficial members of the gut microbiota and are known to specialize on mucin glycoproteins because of auxotrophies for GlcNAc and threonine, whic h ar e both pr e v alent in m ucin (v an der Ark et al. 2018 ).An initial bloom of Akkermansia was observed to r epr esent ∼20% of the community by day 5, which then dwindled by da y 10.T his community also contained an ASV assigned as B. fragilis , another well-established m ucin degr ader that incr eased in abundance as Akkermansia decr eased, suggestiv e of competition (Roberton andStanley 1982 , Huang et al. 2011 ).This a ppar ent competition between Akkermansia and B. fragilis may be strain-or conditionde pendent.Recently, the gen us Akkermansia has been classified into four species-le v el phylogr oups (AmI-AmIV) based on phylogenomic analyses (Guo et al. 2017, Kirmiz et al. 2020 ).These genotypic differences are recapitulated in phenotypes, for example, wher e specific phylogr oups of Akkermansia synthesize vitamin B 12, potentially outcompeting other organisms in a vitamin B 12deficient environment (Kirmiz et al. 2020 ).Other differences may be dir ectl y r elated to m ucin metabolism, wher e phylogr oups AmII, AmIII, and AmIV retain a larger number of GHs related to human milk oligosacc haride metabolism, whic h has structur al similarity with mucin (Luna et al. 2022, Becken et al. 2021 ).Although the genomic patterns differ across Akkermansia , these differences are unidentifiable within the V4 region of the 16S rRNA gene studied her e.Furthermor e, str ain-le v el div ersity anal yses within B. fragilis hav e identified r epeated m utations acr oss individuals in pol ysaccharide utilizing loci, suggesting continued within-person evolution, which may be directed by competition with other members of the micr obiota, suc h as other members of the Bacteroides or, in the case of mucin, Akkermansia (Zhao et al. 2019 ).
Each donor community was developed in triplicate while maintaining individual lineages with little variability in taxonomy, div ersity, and composition acr oss r eplicates thr oughout the experiment.We inter pr et these observ ations to suggest that mucin as the sole carbon source is a strong selective force driving deterministic community assembly, though the stabilized community selected depended gr eatl y upon initial inoculum.These findings are in line with other sequential cultivation experiments in which diverse and replicable communities have been developed on mucins (Lou et al. 2023 ) and arabinoxylans (Yao et al. 2023 ) under high dilution pr essur e and short incubation times, though unlike m ucins, comm unity structur es on ar abinoxylans conv er ge acr oss donors (Yao et al. 2023 ).The complexity in glycosidic bonds, which sustain greater species diversity (Midani and David 2023 ), are likely to drive this selection and limit the ability for any one organism to dominate.Although the dilution rate and sampling times may select for fast growing organisms (Adamberg and Adamberg 2018 ), these same organisms are expected to thrive in native conditions, under which the average gut transit time is 24 h (Asnicar et al. 2021, Nandhra et al. 2020 ).In our experiments, the mucin solution is first dialyzed to remov e small oligosacc harides ( < 12 kDa) and then filtered ( < 0.2 uM), resulting in large, soluble glycoproteins that are likely processed extr acellularl y befor e tr ansport.This m ucin pr epr ocessing step is likely effective, as simple sugars, when present in these types of cultivation experiments, can result in communities that are dominated by one or two species (Lou et al. 2023, Yao et al. 2020, Wu et al. 2023 ).While the composition of the porcine gastric mucin used in these experiments is unknown, a pr e vious study found similarities and differences in both protein and carbohydrate composition compared to human mucins (Miner-Williams et al. 2009 ).Porcine gastric mucin tends to have a higher percenta ge of carbohydr ates b y w eight while human mucins are more pr otein ric h.In terms of whic h sugars ar e pr esent, porcine gastric mucin tends to have more GlcNAc , galactose , and fucose as compared to human colonic mucins.Despite these differences, others have noted that mucin and mucin glycans originating from both porcine and human sources result in the ability to support similar div erse comm unities (Wu et al. 2023 ).Additionally, similar filtration and dialysis steps on labeled porcine mucin result in large glycoproteins that are metabolized by mucin degrading specialists (Davey et al. 2023 ).T herefore , it will be interesting to determine the inflection point of mucin complexity at whic h the tr ansition fr om single-or ganism cultur e to m ultispecies consortium occurs and if mucin source influences community de v elopment.
In addition to nutrient availability, the absence of specific nutrients in these systems may be influential on selection and community assembl y.Auxotr ophies , both for vitamins and amino acids , ar e highl y pr e v alent acr oss man y micr obial comm unities (Yu et al. 2022 ).In similar sequential batc h-cultur e cultiv ation experiments, the relief of these auxotrophies has a demonstrable effect on final community composition (Yao et al. 2020 ).In the experiments here, common vitamin auxotrophies were alleviated through addition of the ATCC vitamin supplement.Ho w ever, in hand with amino acid auxotrophies, the abilities of an organism to scavenge amino acids into its cells vary; further, some organisms produce exopeptidases capable of breaking down complex proteins into smaller fragments, as dipeptides or free amino acids, whic h ar e then tr ansported into the cell (Zhang et al. 2022 ).Although the mucin backbone does provide a potential source of amino acids, not all bacteria produce peptidases able to cleave the m ucin bac kbone .T her efor e, to amelior ate an y amino acid auxotr ophy without pr o viding additional carbon sources , we provided a low concentration of free amino acids or tryptone in the fermentation medium and assessed whether the complexity of amino acids influenced the final microbial community structure.The lack of ASVs discriminated by LEfSe analysis comparing the final comm unities cultiv ated with the two amino acid sources is in line with observations of bacterial growth in minimal amino acid-free media (Price et al. 2018 ) and cross-feeding interactions in complex systems (Du et al. 2022 ).These dynamic interactions by other members of the microbial community may be speciesor strain-specific (Ashniev et al. 2022 ), which argues for a comprehensiv e anal ysis of the genomic ca pacity of these or ganisms using metagenomics and assembly of species' genomes to fully determine whether organisms with auxotrophies were maintained in these experiments.

In summary
Identification of individual m ucin-degr ading consortia necessitated the use of in vitro models, in which mucin was the sole carbon source .For example , in in vivo studies in both animals and humans, the nutritional role of mucins is obscured by dietary conditions .T her efor e, the batc h-model system used in this r esearc h provides essential advantages to identify the ecological networks of m ucin degr adation.Ho w e v er, v ariability in m ucin structur e and composition along the length of the intestinal tract (Robbe et al. 2004 ), secretion of antimicrobial peptides (Bevins 2007 ), and spatial organization of individual microbes (Mark Welch et al. 2017 ) are remaining aspects of mucin degradation that are hostr ele v ant and difficult to replicate in vitro .De v elopment of these communities in vitro using a top-down a ppr oac h allows the examination of micr obe-micr obe inter actions while accounting for natur all y occurring str ain heter ogeneity amongst core bacterial gr oups, an adv anta ge not afforded by bottom-up a ppr oac hes that r el y on assembling communities using type strains originating fr om differ ent sources.

Figur e 1 .
Figur e 1. Final (da y 10) optical density and pH measur ements of micr obial consortia fr om thr ee human donors established with m ucin as the sole carbon source supplemented with different amino acid sources.Each community grew to different final optical densities (A) and produced different le v els of acid (B) after 10 days of incubation.Av er a ges of three lineages from each final donor community ((Donor 1, D1; Donor 2, D2; and Donor 3, D3) are displayed with error bars representing standard deviation as calculated in R. Statistically significant differences are calculated by Tuk e y's multiple comparisons test with P < .05.Symbol style: nonsignificant (ns), 0.05 ( * ), 0.01 ( * * ), 0.001( * * * ), and < 0.0001( * * * * ).

Figure 5 .
Figure 5. Relative abundance of bacterial taxa for each donor over 10 days with supplementation of either amino acids or tryptone.Av er a ged abundance of three replicate lineages for each donor indicates mucin sustains diverse fermenting consortia over sequential dilutions that is dependent on the initial community.Day 0 ( n = 1) is the initial fecal inoculum prior to incubation, whereas each subsequent day reflects the average of thr ee linea ges after incubation.Eac h shade r epr esents a distinct ASV classified down to the le v el of genus, with the color r epr esenting a phylum: Bacter oidota (pur ple), Desulfobacter ota (y ello w), Firmicutes (green), Proteobacteria (grey), and Verrucomicrobiota (red).Numbers following genus names r epr esent differ ent ASVs .T he top 40 ASVs across all donors (Donor 1, D1; Donor 2, D2; and Donor 3, D3) are shown, wher e r emaining low-abundance ASVs are grouped in Other (light grey).