Candida species-specific colonization in the healthy and impaired human gastrointestinal tract as simulated using the Mucosal Ileum-SHIME® model

Abstract Candida species primarily exist as harmless commensals in the gastrointestinal tract of warm-blooded animals. However, they can also cause life-threatening infections, which are often associated with gut microbial dysbiosis. Identifying the microbial actors that restrict Candida to commensalism remains a significant challenge. In vitro models could enable a mechanistic study of the interactions between Candida and simulated colon microbiomes. Therefore, this study aimed to elucidate the spatial and temporal colonization kinetics of specific Candida, including C. albicans, C. tropicalis, and C. parapsilosis, and their relative Nakaseomyces glabratus, by using an adapted SHIME® model, simulating the ileum, and proximal and distal colons. We monitored fungal and bacterial colonization kinetics under conditions of eubiosis (commensal lifestyle) and antibiotic-induced dysbiosis (pathogenic lifestyle). Our findings highlighted the variability in the colonization potential of Candida species across different intestinal regions. The ileum compartment proved to be the most favourable environment for C. albicans and C. parapsilosis under conditions of eubiosis. Antibiotic-induced dysbiosis resulted in resurgence of opportunistic Candida species, especially C. tropicalis and C. albicans. Future research should focus on identifying specific bacterial species influencing Candida colonization resistance and explore the long-term effects of antibiotics on the mycobiome and bacteriome.


Introduction
Between 2.2 and 3.8 million fungal species inhabit Earth, colonizing a diverse range of ecological niches, including various sites both on and within the human body (Underhill andIliev 2014 , Ha wks worth andLucking 2017 ).Among these fungi, sever al str ains can pose significant health risks to humans, r esulting in a global health burden (Bongomin et al. 2017 ).Appr oximatel y 25% of the world's population suffers from skin infections, and about 75% of women will experience at least one episode of vulvov a ginal candidiasis during their lifetime (Brown et al. 2012 ).Furthermore, fungal systemic infections contribute to mortality rates r anging fr om 10% to 47% (Brown et al. 2012 ).Recognizing this, the World Health Or ganization (WHO) r ecentl y issued its first list of top-priority fungal pathogens, with Candida species pr ominentl y featur ed (Br own et al. 2012 ).Notabl y, Candida albicans , r esponsible for 80% of human candidiasis cases (Silva et al. 2012 ), is among the WHO's four critical priority fungal species, while Candida tropicalis , Candida parapsilosis , and Nakaseomyces glabratus (formerly Candida glabrata ; Takashima and Sugita 2022 , which we considered a Candida for ease of writing) are listed within the seven high-priority ones.
Candida species primarily exist as harmless commensals in the gastrointestinal tract of warm-blooded animals, particularly birds and mammals (Odds 1984 ).Howe v er, the Candida genus can also comprise pathogens and pathobionts, which can cause life-threatening infections .T hese infections mainl y affect imm unocompromised individuals and are often associated with microbial dysbiosis (Koh et al. 2008, Zhai et al. 2020, Li et al. 2022 ).Hence, both the host immune system and the gut microbiome play strategic roles in preventing the commensal-to-pathogen transition in Candida species (d'Enfert et al. 2021 ).In particular, the gut microbiome has a pivotal role in preventing candidiasis, serving as the first line of defense against pathological Candida outgrowth through a phenomenon termed "colonization resistance" (Lawley and Walker 2013 ).This colonization resistance involves microbemicr obe inter actions, suc h as competition for nutrients, nic hes, and binding sites, as well as the release of antimicrobial agents (Dabard et al. 2001, Momose et al. 2008, Gong et al. 2010, Rea et al. 2010, Deriu et al. 2013, Wagner 2022, Xu et al. 2023 ).Ho w e v er, identifying the k e y micr obial actors that r estrict Candida to commensalism remains a significant challenge .T he same can be said for comprehending the impact of antibiotic-associated dysbiosis in candidiasis pathology.
Current models available to study Candida -gut bacteriome inter actions hav e dr awbac ks.On the one hand, mouse models, historicall y r eliant on antibiotic tr eatment to establish stable Candida colonization in the gut (Fan et al. 2015 ), are limited in accur atel y replicating Candida 's commensal behaviour since the microbiome should be impaired to obtain C. albicans colonization (Neville et al. 2015, Perez 2019 ).On the other hand, in vitro models used to study Candida -bacteria interactions in bioreactors face challenges in culti vating faecal-deri ved strict anaerobes alongside oxygendependent fungi (Auchtung et al. 2018 ).Addressing this issue may r equir e separ ate cultiv ation of strict anaerobes and Candida (Ricci et al. 2022 ), whic h pr ecludes inv estigating dir ect interkingdom interactions, lo w ering its relevance.
Pr omisingl y, P ayne and colleagues (Payne et al. 2003, Wynne et al. 2004 ) studied C. albicans within a faecal-derived microbiome bac kgr ound using a m ultista ge fermentation model two decades ago, yet no follow-up studies ensued.To date, only Maas and colleagues, and our research group, have explored bacteria-fungi interactions in faecal-derived microbiomes, using the TIM-2 and the SHIME ® models (with features reviewed in Van de Wiele et al. 2015 ), r espectiv el y (Maas et al. 2023a, b , Marsaux et al. 2023 ).
In particular, we explored the natural engraftment of fungi in both eubiosis (commensal lifestyle) and dysbiosis (pathogenic lifestyle) states, and we highlighted the critical role of certain bacterial species in pr e v enting fungal outgr owth (Marsaux et al. 2023 ).Ho w e v er, despite examining numer ous faecal samples, onl y one individual contained Candida species, which was eventually washed out from the system under eubiosis conditions.Similarly, C. albicans did not consistently grow in the TIM-2 model (Maas et al. 2023a , b ).Hence, there is a pressing need to design complex in vitro models, which do not rely on the presence of Candida in the faecal inocula that will enable a mechanistic study of the interactions between Candida and simulated colon microbiomes.
In the present study, we used an adapted SHIME ® model to explor e whether an y Candida species of interest can be studied alongside individual colonic micr obiomes, deriv ed fr om a human faecal sample.Specifically, we sought to decipher the ecological factors influencing Candida colonization by artificially inoculating the SHIME ® model with species that have been linked to pathogenic properties, including C. albicans , C. tropicalis , and C. parapsilosis , as well as their r elativ e N. glabratus .As a control to distinguish active colonization from transient passage upon artificial inoculation, we also inoculated Sacc harom yces cerevisiae , a tr ansient micr obe of the gut micr obiome (Auc htung et al. 2018 ).To investigate the longitudinal colonization ability of the different Candida species along the simulated human gastrointestinal tract, we used the r ecentl y de v eloped Mucosal Ileum-SHIME ® model (Deyaert et al. 2022 ).This model allowed us to assess colonization dynamics in the simulated ileum, proximal colon, and distal colon, all in the presence of a mucosal compartment, which has been shown to be essential for maintaining crucial functional niches and the unique features of an individual's gut microbiome (Van den Abbeele et al. 2013 ).We monitored fungal and bacterial colonization kinetics in both eubiosis (commensal lifestyle) and upon antibiotic-induced disruption (pathogenic lifestyle).16S rRNA gene-targeted Illumina sequencing was conducted to e v aluate the bidirectional impact of Candida engraftment and antibiotic administrations on bacteriome div ersity.Finall y, to account for substantial interindividual variability across the population, we initiated the experiment starting from distinct faecal inocula obtained from three healthy individuals.Overall, this study serves as a proof of concept for using the Mucosal Ileum-SHIME ® model to investigate Candida -bacteria interactions and sets the stage for discovering factors contributing to Candida outgrowth and commensal-to-pathogen transition.

Chemical products
All c hemicals wer e obtained fr om Merc k (Darmstadt, German y) unless otherwise stated.

The in vitro Mucosal Ileum-SHIME ® technology
The experiments were performed through modification of the M-SHIME ® tec hnology (Pr oDigest and Ghent University, Ghent, Belgium) pr e viousl y described by Van den Abbeele et al. ( 2012 ).The proximal colon compartment was preceded with an ileum compartment in a similar a ppr oac h as was r ecentl y published by our r esearc h gr oup (Deyaert et al. 2022 ).The configuration of the Mucosal Ileum-SHIME ® model is summarized in Fig. 1 (A).
Briefly, the entire setup was autoclaved at 121 • C for 20 min before inoculation to ensure sterility.Reactors were continuously stirred and maintained at 37 • C. Anaerobic conditions were ensur ed by spar ging the ileal nutritional medium with nitrogen (N 2 ) for 15 min each time it was introduced into the stomach compartment three times a day .Additionally , the headspace of each reactor was flushed with N 2 daily.The proximal colon compartments r eceiv ed 150 ml of ileal content thr ee times a day after 1.5 h of fermentation in the ileum compartment, along with 50 ml of fiber solution.To simulate the Mucosal component of the human gastrointestinal tract, we introduced mucin-alginate beads in the ileum compartments and m ucin-cov er ed micr ocosms in the proximal and distal colon compartments .T he pr epar ation and replacement of these Mucosal components were performed as pr e viousl y described (Deyaert et al. 2022 ).In short, mucin-alginate beads, which can be sampled from the ileum compartments in a sterile manner, wer e pr epar ed by dripping a m ucin-alginate solution into a cross-linking solution containing CaCl 2 (Deyaert et al. 2022 ).In contr ast, m ucin-cov er ed micr ocosms (length = 7 mm, diameter = 9 mm, total surface area = 800 m 2 m −3 , AnoxKaldnes K1 carrier, AnoxKaldnes AB, Lund, Sweden) were coated by submerging them in a mucin-agar solution (Van den Abbeele et al. 2012 ).Finally, the ileum compartments were inoculated at day 0 with a consortium of 12 bacterial species comprising Streptococcus bovis , Streptococcus intermedius , Ligilactobacillus salivarius , Limosilactobacillus reuteri , Enterococcus faecalis , Enterococcus faecium , Veillonella parvula , Veillonella dispar , Blautia obeum , Faecalibacterium prausnitzii , Clostridium nexile , and Prevotella melaninogica (Deyaert et al. 2022 ).They were selected to simulate the function and composition of a healthy individual's ileal microbiome, including microbes performing partial conversion and deconjugation of bile salts, as well as primary fermenters metabolizing simple carbohydrates into e .g .lactate , that support the growth of secondary fermenters, producing e.g .short-chain fatty acid (SCFA) thr ough cr oss-feeding (Deyaert et al. 2022 ).In contr ast, the pr oximal and distal colon compartments were inoculated with the faecal sample of a healthy donor.Details regarding the growth conditions of the ileal bacterial consortium, as well as the pr epar ation of the faecal inoculum, have been pr e viousl y described (Deyaert et al. 2022 ).

Eubiosis study to elucidate the ability of five fungal species to colonize a healthy microbial community
The eubiosis study aimed to assess the colonization ability of five distinct fungal species within the gut microbiome under healthy conditions .T he four Candida species, namely C. albicans , C. tropicalis , N. glabratus , and C. parapsilosis , were selected as they are among the most life-threatening fungi.As a control, S. cerevisiae was selected as model organism of transient passage, as it does not natur all y colonize the gut micr obiome in vivo .Consequentl y, S. cerevisiae can be used to differentiate active from transient passage of the four Candida species and to confirm that active colonization is not the result of artificial inoculation.Additionally, the eubiosis study sought to e v aluate the impact of the five fungal species inoculation on the microbiome's function and composition.
Briefly, the Mucosal Ileum-SHIME ® reactors were initially inoculated either with an ileal bacterial consortium (ileum compartment) or the faecal sample from a single donor (proximal and distal colon compartments).The system r eceiv ed thr ee feeding cycles each da y.T he study began with a stabilization period (d0-d19), during which the microbial community in each colon compartment diversified and eventually established stability.Following the stabilization period, a baseline period (d19-d35) allo w ed to study the baseline microbiota composition and metabolic activity before initiating any intervention.Subsequently, 1 ml of an ov ernight cultur e of eac h fungal species was inoculated once daily for thr ee consecutiv e days (d35-d37) in the ileum, and pr oximal and distal colon compartments .T he microbial function and composition wer e continuousl y monitor ed for 3 weeks (d35-d56) without altering the physiological conditions applied to the bioreactors, thereby maintaining eubiosis conditions.
An y c hanges observ ed in the micr obiome between the baseline and eubiosis study periods were attributed to the introduction of the five fungal species.Additionally, the study assessed the ability of these fungal species to colonize and thrive in different intestinal regions under eubiosis conditions.To account for potential interindividual differences in microbial composition, the study involved three healthy adult participants .T he experimental design is summarized in Fig. 1 (B) and the experimental timeline in Fig. 1 (C).
T hroughout the experiments , samples for metabolic , microbial, and flow cytometry analysis were collected before the start of a new feeding cycle, hence during the stationary phase of microbial growth.

Dysbiosis study to elucidate the ability of five fungal species to colonize an impaired microbial community
The dysbiosis study aimed to assess the ability of five distinct fungal species to colonize the gut microbiome under dysbiosis conditions, whic h wer e induced by antibiotic tr eatment.
Briefly, following the initial period of eubiosis conditions (d35-d56), clindamycin was administered in the ileum compartments.A clindamycin hydr oc hloride stoc k solution was pr epar ed twice a week at 3.15 g l −1 in water , filter -sterilized (0.2 μm) into an autoclav ed Sc hott bottle, and spar ged with N 2 to ensur e anaer obicity.The solution was pumped automatically into the ileum compartments e v ery 8 h, for a dur ation of 14 days (d56-d70).Clindamycin r eac hed a final concentration of 113.4 mg l −1 in the ileum compartments in ca. 3 da ys , and 85.0 mg l −1 in the proximal and distal colon compartments, in ca.6 days.Recognizing that the five fungal species might face challenges in colonizing the microbiome under eubiosis conditions, they were reintroduced once daily for thr ee consecutiv e days (d63-d65) after 1 week of antibiotic administration.
The first week of antibiotic tr eatment, befor e the reinoculation of the fungal species, aimed to confirm fungal engraftment that may have occurred during eubiosis conditions.Opportunistic fungi that managed to thrive under eubiosis conditions might exploit the reduced bacterial function and diversity induced by antibiotics, resulting in higher cell concentrations.In contrast, the second week of the antibiotic treatment, following the reinoculation of the five fungal species, was used to determine whether any of these species, which might not have thrived under eubiosis conditions, could now colonize in the dysbiosed microbiome.Following the antibiotic cessation, the r ecov ery of the microbial ecosystem was observed for one and a half weeks (d70-d79).This phase aimed to assess the competitive ability of opportunistic fungi against a recovering colonic microbiome .T he experimental timeline is described in Fig. 1 (C).
T hroughout the experiments , samples for metabolic , microbial, and flow cytometry analysis were collected before the start of a new feeding cycle, hence during the stationary phase of microbial growth.

Fungal inoculum prepar a tion and inoculation
Candida albicans strain SC5314 (patient with disseminated candidiasis) (Odds et al. 2004 ) was provided by the r esearc h gr oup of Dr Salomé LeibundGut-Landmann (University of Zürich, Zürich, Switzerland).Candida tropicalis strain IHEM6204 (human sputum from a patient with disseminated aspergillosis and lung cancer), N. glabratus strain IHEM19218 (human blood from a patient with candidemia and prostate cancer), C. parapsilosis strain IHEM17737 (human faeces), and S. cerevisiae strain IHEM17798 (human faeces from a patient with fungemia and stubborn anemia) were purc hased fr om the BCCM/IHEM fungi collection (Sciensano, Brussels, Belgium).Upon r eceiv al, eac h fungal species identity was confirmed by Sanger sequencing.
Each fungal species was initially cultured separately on Sabour aud dextr ose a gar (Av antor™) for 48 h under anaer obic conditions, using an Anaer oGen™ ba g, at r oom temper atur e. Subsequently, a single colony was selected to inoculate 20 ml Sabour aud dextr ose br oth (2% w/v), and this culture was further incubated under anaerobic conditions within an anaerobic chamber at 37 • C, for 48 h to pr epar e the first inoculum (to be used on day 35 or 63).Ho w e v er, after 24 h of incubation, 400 μl of eac h ov ernight cultur e was used to inoculate distinct 20 ml fresh Sabour aud dextr ose br oth (2% w/v), and this was also incubated anaer obicall y for 48 h (to be used on day 36 or 64).This process was repeated once more to prepare the third inoculum (to be used on day 37 or 65).Ther efor e, onl y fr esh cultur es wer e used to pr epare the different inocula.
To pr epar e those inocula, equal volume of the cultur es fr om each species were pooled in a 100 ml-clear glass vial (reference A05521, Nov olab, Geraar dsbergen, Belgium) closed with a butyl stopper (r efer ence 7395, Rubber B.V., Loosdr ec ht, The Netherlands) and a crimp cap (reference 548-3317, Avantor™) that had been preflushed with N 2 to ensure the preservation of an anaerobic environment.After homogenizing the mixture, 5 ml was withdrawn using a syringe mounted with a needle, and this inoculum was used to inoculate each intestine compartment once a day (on days 35-37 and days 63-65) after the feeding cycle was completed.The headspace of the ileum, proximal colonic, and distal colon compartments was flushed with N 2 immediately after fungal inoculation.

Faecal sample collection and donor description
The three healthy adults had no antibiotic nor pre-or probiotic dietary supplements intake in the pr e vious 3 months, and no constipation.All faecal samples wer e immediatel y tr ansferr ed to a recipient containing an AnaeroGen™ bag to limit the samples' exposure to oxygen, and transferred to the lab for further use .T he proximal and distal colon compartments were inoculated within maximum 13 h postcollection of the faecal samples by the donors.

Metabolic analysis
To assess the impact of the fungal inoculation under eubiosis and dysbiosis conditions on the metabolic activity of the gut microbiome , lactate , ethanol, SCFA (acetate , propionate , and butyrate), and br anc hed short-c hain fatty acid (BSCFA, isobutyr ate , iso valerate, and isoca pr oate) wer e quantified at all sampling time points.
Lactate was quantified on cell-free fermentation samples (centrifuged for 5 min at 7690 × g ) using a commercially available enzymatic assay kit (R-Biopharm, Darmstadt, Germany) according to the manufacturer's instructions.
Ethanol, SCF A, and BSCF A were quantified using a GC-2030 gas c hr omatogr a ph (Shimadzu, Hertogenbosc h, Netherlands).It was equipped with a GC SH-polarD capillary column (30 mm × 0.32 mm ID-BP 21 × 0.25 μm, Shimadzu), a flame-ionization detector, and a split injector.Nitrogen (90.3 ml min −1 ) was used as a carrier gas; nitrogen (24 ml min −1 ), hydrogen (32 ml min −1 ), and air (200 ml min −1 ) were used as make-up gas .T he injector and detector temper atur es wer e set at 200 • C and 240 • C, respectiv el y.The column temper atur e pr ofile was set from 40 • C to 220 • C with a constant temper atur e incr ease of 10 • C min −1 after whic h temper atur e was held for 5 min at 220 • C. Fermentation samples collected during the Mucosal Ileum-SHIME ® experiment were diluted 1:1 with a mixture composed of acetonitrile, 0.5% (v/v) formic acid, 0.008% (v/v) 2-methyl hexanoic acid (internal standard), and a saturating amount of sodium chloride; mixed for 15 min using a HulaMixer™ Sample Mixer (ThermoFisher Scientific, MA, USA); centrifuged (15 080 × g , 15 min); and run with the a ppr opriate external standards.Injection was performed in split mode with a split ratio of 40; the injected volume was 1 μl.
All anal ysis wer e performed in simplicate.Due to limited concentration of BSCFA (Supplementary Raw data), only lactate, ethanol and SCFA data are reported in the manuscript.

Li v e bacterial cell count quantification by flow cytometry
Live bacterial cells were quantified via flow cytometry to e v aluate the impact of the five fungal species inoculation on bacterial survival in eubiosis and dysbiosis.Samples were 1:1-diluted in cryoprotectant and stored at -80 • C until analysis as previously described (Hoefman et al. 2012 ).Upon staining with 0.01 mM SYTO24 (ThermoFisher Scientific) and 0.0 mM propidium iodide (Ther-moFisher Scientific) at room temperature for 15 min in the dark, samples were analysed on a BD Facsverse (BD, Franklin Lakes, NJ, USA) using the high flo wrate setting.Flo w c ytometry data w ere analysed using FlowJo version 10.5.0.(BD).

Microbial community analysis by qPCR
Total bacterial and species-specific concentrations were determined by quantitative polymerase chain reaction (qPCR) on all time points of the experiment, on the three faecal inocula, and the different fungal overnight cultures.Briefly, DN A w as isolated as pr e viousl y described (Boon et al. 2003 ) with minor modifications (Duysburgh et al. 2019 ) from 1 ml luminal samples or 0.25 g of faecal matter.Subsequently, qPCR was performed using a QuantStudio 5 Real-Time PCR system (Applied Biosystems, Foster City, CA, USA).Each sample was run in technical triplicate.
The standards used were synthesized gBlock Gene Fragments (Integr ated DNA Tec hnologies, Inc., Coralville, IA, USA) of 1000 bp designed by aligning the primer pairs on the r efer ence genome of the species of inter est, whic h was obtained from NCBI (Bethesda, MD, USA).For the total bacteria standard, Esc heric hia coli str ain 97-3250 was used.The complete sequence of each standard can be found in the Supplementary Raw data.
To better a ppr eciate the differ ences in concentr ation between eac h fungal species, qPCR v alues wer e corr ected using flo w c ytometry data.To do so, four to six pure overnight cultures of each fungal species were analysed both by qPCR (triplicate of analysis) and b y flo w c ytometry (simplicate of analysis).T he a v er a ge quotient of the gene copy number and the active fluorescent unit (afu) count was used as a correction factor of the qPCR data obtained for each fungal species present in a mixed community (faecal or luminal samples).Ther efor e, the qPCR data for each fungal species are expressed in log 10 (afu ml −1 ).

Bacterial community analysis by 16S rRNA gene-targeted sequencing
Bacterial community composition was assessed on a selection of time points.Samples were sent out to LGC Genomics GmbH (Berlin, Germany) for Illumina sequencing of the 16S rRNA gene amplicons of the V3-V4 region.T he 341F (5 -CCTACGGGNGGCWGC AG-3 ) and 785R (5 -GA CTA CHV GGGTATCTAAKCC-3 ) primers were used (Herlemann et al. 2011 ).The amplification by PCR, sequencing of the amplicons, and pr epar ation of the Illumina libr aries wer e performed as pr e viousl y described (Marsaux et al. 2023 ).

Da ta anal ysis
In the case of total bacterial, species-specific and cell count concentr ations, data wer e log-tr ansformed because of their lognormal distribution.
The impact of the fungal inoculation on the change in total bacterial, species-specific, cell count, and metabolic concentrations was assessed using a Kruskal-Wallis test follo w ed b y Dunn's multiple comparison test.During the eubiosis study, the outcomes wer e compar ed for c hange between the baseline period (av er a ge of d19-d35), the week 1 of the eubiosis period (d36-d42), and the weeks 2 and 3 of the eubiosis period (d44-d56).During the dysbiosis study, the outcomes were compared for change between the week 3 of the eubiosis period (d51-d56), the first week of the dysbiosis period (d58-d63), the second week of the dysbiosis period (d64-d70), and the r ecov ery period (d71-d79).Statistical analysis was performed using Gr a phP ad Prism v ersion 9.5.1 (733) for Windows (Gr a phP ad Softwar e, San Diego, CA, USA).
Read assembly and cleanup was lar gel y deriv ed fr om the MiSeq standard operating procedure described by the Schloss lab (Schloss andWestcott 2011 , Kozich et al. 2013 ).In brief, the Mothur software package (v.1.44.3) was used to assemble reads into contigs, perform alignment-based quality filtering (alignment to the Mothur-reconstructed SILV A SEED alignment, v .138),remov e c himer as (vsearc h v2.13.3), assign taxonom y using a naïve Bayesian classifier (Wang et al. 2007 ) and SILVA NR v138, and cluster contigs into OTUs at 97% sequence similarity.Only sequences having at least 5 counts in one sample and assigned to the bacteria kingdom were retained.For each OTU, re presentati ve sequences wer e pic ked as the most abundant sequence within that OTU.
Data at genus le v el was processed after a ppl ying a total sum normalization.To account for variations in total bacterial cell counts among samples, the r elativ e abundances of bacterial genera was corrected with 16S rRNA qPCR data, enabling the gener ation of quantitativ e micr obiome pr ofiles.Additionall y, the gut bacteriome α-diversity was assessed by using Shannon's and Simpson's indexes (calculated using phyloseq, v1.44.0;McMurdie and Holmes 2013 ), while the β-diversity was visualized through discriminant analysis of principal components (DAPC), principal component analysis (PCA), and redundancy analysis (RDA) based on the r elativ e abundances of bacterial OTU.Specifically, the DAPC was conducted by hier arc hicall y clustering Euclidean distances between samples using W ard' s minimum variance method.The DAPC plots were constructed with two discriminants and 80% of r etained v ariance in the principal components using adegenet (v2.1.10)(J ombart 2008 , J ombart et al. 2010 ).The RDA was performed with the taxonomic r elativ e abundances of bacterial OTU as a response variable and the treatment as explanatory variable using type 2 scaling.The RDA model and its statistical significance were calculated in vegan (v2.6-4).DAPC and RDA plots were visualized using ggplot2 (v3.4.3).PCA were performed on normalized data of the 16S rRNA gene-targeted sequencing using ClustVis ( https:// biit.cs.ut.ee/ clustvis/ , 19 July 2023, date last accessed) with parameters as previously described (Van den Abbeele et al. 2021 ).

Quantification of the five fungal species in the donor faecal samples and the fungal inocula used in the eubiosis and dysbiosis studies
The concentrations of C. albicans , C. parapsilosis , C. tropicalis , N. glabratus , and S. cerevisiae were determined by qPCR in the three faecal samples and the fungal inocula used in the eubiosis and dysbiosis studies.
For the three faecal samples, all five fungal species displayed le v els below the limit of quantification, as indicated by all qPCR analyses (Fig. 2 A-E).
The three fungal inocula used in the eubiosis study (d35-d37) contained a significantly higher quantity of C. parapsilosis cells compared to C. tropicalis and S. cerevisiae (Fig. 2 G).Conv ersel y, during the d ysbiosis stud y (d63-d65) the inocula contained a significantl y higher concentr ation of S. cerevisiae and N. glabratus cells than C. albicans and C. parapsilosis (Fig. 2 G).In addition, the inocula used in the eubiosis study contained significantly more C. albicans , C. parapsilosis , and S. cerevisiae cells compared to the inocula used in the d ysbiosis stud y (Fig. 2 A-E).T his discrepancy ma y lead to species-specific differences within the SHIME ® reactors between the different study periods .Regardless , all fungal species were present in high enough concentrations in the inocula to obtain v alues abov e the limit of quantification in the r espectiv e SHIME ® reactors upon inoculation.

Cell concentrations of the five fungal species and total bacteria in eubiosis conditions
The concentrations of C. albicans , C. tropicalis , N. glabr atus , C. par apsilosis , S. cerevisiae , and total bacteria were determined in the SHIME ® using qPCR (Fig. 3 A-G), and the live bacteria concentr ations wer e quantified b y flo w c ytometry (Fig. 3 H) at all sampling time points during the eubiosis study.Detailed statistical analysis for each fungal species is provided in Fig. S1 (A-E).Similar r esults wer e observ ed during the thr ee experiments despite the fact that different donors were used to inoculate the colon compartments.
As expected from the low initial le v els of the five fungi present in the r espectiv e faecal inocula, none of the intestinal regions in the SHIME ® contained le v els of the fungal species abov e the limit of quantification during the baseline period (Fig. 3 A-E).Subsequently, upon inoculation of the fungal strains, all fungal species concentr ations significantl y incr eased, particularl y in the ileum compartments.In the second-and thir d-w eek postinoculation, concentrations of C. tropicalis , N. glabratus , and S. cerevisiae significantl y decr eased, r eac hing le v els below the limit of quantification.Conv ersel y, C. albicans and C. parapsilosis maintained relativ el y high cell counts, e v entuall y r eac hing similar concentr ations in all intestinal compartments.
Importantl y, the engr aftment success is independent of inoculation density: N. glabratus did not persist in the bioreactors de-spite the fact its inoculum had, on av er a ge, one of the largest concentration of cells (Fig. 2 G).
Lastly, the total bacterial concentration gradually and significantl y decr eased in all intestinal r egions between the baseline period and the second-and thir d-w eeks postinoculation (Fig. 3 G, Fig. S1 F).For live bacteria specifically, it was only in the ileum compartment that there was a significant decrease, as quantified by flo w c ytometry (Fig. 3 H, Fig. S1 G).

Impact of the fungal inoculation on the metabolic activity of the gut microbiome in eubiosis conditions
The metabolic activity of the ileal micr obiome r emained lar gel y stable between the baseline period and the eubiosis period following fungal inoculations, with the exception of lactate and acetate (Fig. 4 A-E, Fig. S2 A-E).Lactate v alues str ongl y incr eased in the ileum compartment 1 but r emained exceptionall y low ( < 0.06 mM).Acetate and propionate concentrations gradually declined in all ileum compartments, and values were significantly lo w er during the second-and thir d-w eeks postinoculation compared to the baseline period.Despite these changes, the acetate:pr opionate:butyr ate r atio r emained stable thr oughout the experiment (Fig. 4 F-H).
In contrast, in each colon region, all metabolic markers significantl y decr eased between the baseline period and the period following the fungal inoculations, except for lactate, which r emained unc hanged (Fig. 4 A-E, Fig. S2 A-E).Despite a significant reduction in acetate , propionate , and butyrate , the acetate:pr opionate:butyr ate r atio r emained stable thr oughout the experiments, regardless the donor that was used (Fig. 4 F-H).Furthermor e, ethanol concentr ations also significantl y decreased.Detailed statistics for each metabolite are provided in Fig. S2 (A-E).

Impact of the fungal inoculation on the gut bacteriome composition and diversity in eubiosis conditions
The composition of the gut bacteriome remained stable in all intestinal regions between the baseline period and the period following fungal inoculation in eubiosis conditions (Fig. 5 A-K, Table S1 ).In the ileum compartments, Streptococcus , Veillonella , and Enterococcus were dominant throughout the eubiosis experiment (Fig. 5 A-C).Similarl y, the r atios between the dominant phyla r emained stable in the proximal, and, albeit to a lesser extent, in the distal colon compartments ( Table S1 ).At genus le v el, tempor al changes in relative abundance were limited.Supporting these observ ations, the α-div ersity indexes of the gut bacteriome remained stable throughout all time points in the ileum compartments, while limited temporal variations were observed in the colonic regions (Fig. 5 D-F).Both the Shannon's and the Simpson's indexes were similar on days 28 and 56 in most colon compartments.Furthermor e, the anal ysis of the gut bacteriome β-diversity sho w ed that samples tended to cluster by donor and intestinal regions, maintaining donor-specific features (Fig. 5 G-K).Notably, samples collected on days 28 and 56 sho w ed close clustering in all colon compartments.
Ov er all, these data demonstr ated that distinct micr obial comm unities wer e established in eac h intestinal compartment, and that the interindividual differ ences wer e maintained between donors .Hence , our observations suggest that the inoculation of the fiv e differ ent fungal species into a wide range of donor-and r egion-specific micr obiota had no impact on the ov er all bacterial composition.

Cell concentrations of the five fungal species and total bacteria in impaired conditions
The antibiotic treatment significantly impacted total bacterial concentrations in the ileum compartments but had a minor effect on the colon compartments, regardless of the donor used during the experiment (Fig. 3 G-H, Fig. S1 F-G).Upon cessation of an-tibiotic administration, bacterial concentrations significantly decreased in the proximal colon compartments, remained stable in the distal colon compartments, but significantly increased in the ileum compartments.Similar patterns were observed by flow cytometry, with the antibiotic treatment mostly impacting the ileum compartments (Fig. 3 H).
With respect to the fungal species, only C. tropicalis concentrations sho w ed significant increases during the first w eek of antibiotic treatment in the ileum compartments, while N. glabratus and S. cerevisiae remained below the limit of quantification (Fig. 3 A-F, ) (H) in the ileum, and the proximal and distal colon compartments inoculated at day 0 with the faecal samples from donor 1, 2, or 3.The fungal species values were corrected with flow cytometry data obtained on their r espectiv e pur e ov ernight cultur e, hence r esulting in cell counts instead of gene copy counts .T he uncorrected a v er a ge v alues acr oss the thr ee donors for each intestinal region and each species is also provided (F).Samples were collected during the baseline period (d19-d35), during the eubiosis study (d36-d56) following the fungal inoculations for three consecutive days (d35-d37), during the first week of clindamycin treatment (d56-d63; AB = antibiotic), during the second week of clindamycin treatment (d63-d70; AB + F = antibiotic with fungal reinoculation), which started with the reinoculation of the fungal species for three consecutive days (d63-d65), and during the recovery period (d70-d79; Reco = reco very).T he fungal (re-) inoculations are indicated by an horizontal line .T he limit of quantification ( = LOQ) is indicated by an horizontal dotted line.
In the colon compartments, the fungal concentrations were donor specific.Candida parapsilosis , C. tropicalis , and N. glabratus concentr ations significantl y incr eased acr oss all thr ee pr oximal colon compartments, although N. glabratus remained be-low the limit of quantification in donors 1 and 2, and C. tropicalis in donor 3. Candida albicans concentrations also increased in all three proximal colon compartments, albeit not significantl y.Sacc harom yces cerevisiae le v els r emained stable and below the limit of quantification.In contrast, no changes were observed in the distal colon compartment from donor 1.In the one fr om donor 2, concentr ations of C. albicans , C. parapsilosis , C. trop-  During the r ecov ery period, C. albicans , N. glabratus , and C. tropicalis concentrations remained stable across all intestinal regions.Candida parapsilosis significantly decreased in the ileum compartments, decreased in the proximal compartments, and remained stable in the distal colon compartments.Sacc harom yces cerevisiae concentr ations decr eased significantly in all intestine compartments, reaching values below the limit of quantification.Importantly, C. parapsilosis was the only fungal species that reached higher levels in the proximal and distal colon compartments compared to the ileum compartments by the end of the experiment for all three donors.
Finall y, the engr aftment success is independent of inoculation density: C. albicans persisted in the bioreactors despite the fact its inocula had, on av er a ge, the lo w est amount of cells.In contrast, N. glabratus and S. cerevisiae inocula contained significantl y mor e cells and did not persist (Fig. 2 G).Detailed growth statistics for each fungal species and total bacteria are provided in Fig. S1 (A-F).

Metabolic activity of the gut microbiome in impaired conditions
The metabolic activity of the ileal microbiome was profoundly affected by the antibiotic treatment (Fig. 4 A-E, Fig. S2 A-E).Acetate and propionate, but not butyrate, concentrations significantl y decr eased in the ileum compartments after 1 week of antibiotic treatment.These concentrations remained stable for the remainder of the experiment, except in the ileum compartment 2 in which the concentrations returned to levels comparable to the ones before treatment.These results suggest that the fungal reinoculation had limited impact on the metabolic activity of the ileum micr obiome, compar ed to the antibiotic treatment.This initial reduction influenced the acetate:pr opionate:butyr ate ratio, shifting it by the end of the experiment to w ar ds increased acetate and butyrate in the ileum compartments 1 and 3, and acetate only in the ileum compartment 2 (Fig. 4 F-H).The reduced production of SCFAs resulted in an accumulation of lactate due to reduced cross-feeding activity.Indeed, lactate concentr ations significantl y incr eased at the start of the antibiotic treatment, and although they decreased by the end of the experiment, they r emained ele v ated.Ethanol concentr ations onl y significantl y increased once the fungal species were reintroduced and returned to the original le v el upon discontinuation of the antibiotic administration.
The metabolic activity of the colon microbiome was significantly impacted by the antibiotic treatment (Fig. 4 A-E, Fig. S2 A-E).In the proximal colon compartments, butyrate and acetate significantl y decr eased after 1 week of antibiotic treatment.Although pr opionate concentr ations decr eased, this r eduction was not statistically significant.Similar, though nonsignificant, decreases in acetate and butyrate, but not propionate, were observed in the distal colon compartments .T he SCFAs le v el r emained lar gel y similar in all colon compartments during the second week of antibiotic treatment, except propionate, which significantly increased in the distal colon compartments.After ceasing antibiotic administr ation, le v els in acetate , propionate , and butyrate changes were donor dependent.Acetate , propionate , and butyr ate concentr ations increased in both colon compartments from donor 1.For donor 2, acetate r eincr eased in both colon compartments, while pr opionate onl y r eincr eased in the pr oximal colon compartment.Finall y, onl y butyr ate r eincr eased in both colon compartments fr om donor 3. Consequentl y, the acetate:pr opionate:butyr ate r atio str ongl y shifted in the colon compartments from donors 2 and 3 but not donor 1 (Fig. 4 F-H).These changes did not result in lactate accum ulation; instead, lactate onl y significantl y incr eased during the first week of antibiotic administration but returned to compar able le v els by the end of the experiment.Additionally, while ethanol concentr ation significantl y incr eased upon the r eintr oduction of the fungal species, it returned to comparable levels after antibiotic cessation.

Impact of the antibiotic treatment on the gut bacteriome composition and diversity
The gut bacteriome was substantially impacted by the antibiotic treatment in all intestinal regions (Fig. 6 A-K).In the ileum compartments, the microbiome shifted from a predominance of Strep-tococcus to a dominance of Enterococcus by the end of the antibiotic treatment (Fig. 6 A-C).As a result, the Shannon's and Simpson's indexes decreased either by the end of the antibiotic period (ileum compartment 2) or at the beginning of the r ecov ery period (ileum compartments 1 and 3) (Fig. 6 D-F).By the end of the experiment, comparable index values were observed compared to the pr eantibiotic tr eatment le v els in the ileum compartments 1 and 2, while the indexes remained low in the ileum compartment 3. Despite these variations, two distinct clusters of samples were observed, with samples collected on days 77 and 79, after cessation of the antibiotic treatment, on one hand, and those from day 56, before initiation of the antibiotic treatment, on the other (Fig. 6 G), demonstrating the consistent and long-lasting effect of clindamycin on the ileal microbiome.
The ratios between the dominant phyla also shifted in the colon compartments, resulting in the dominance of Bacteroides , at genus le v el, r egardless of the donor used (Fig. 6 A-C).The colon bacteriome α-diversity was more impacted in the proximal colon compartments compared to the distal ones (Fig. 6 D-F).While both the Shannon's and Simpson's indexes decreased in the colon compartments from donor 1, to then return to their initial levels by the end of the experiment, they steadily increased in the proximal colon compartments from donors 2 and 3. Despite these variations, all the samples collected during the antibiotic treatment and r ecov ery cluster ed closel y together, distinct fr om the samples collected before the initiation of antibiotic treatment (Fig. 6 H-K).This was associated with a loss of donor-specific features, as observed on the RDA plot which only account for 9% of explanatory variations on the first two dimensions (RDA1 and RDA2) of the ordination space (Fig. 6 K).

Discussion
In the eubiosis study, we investigated the colonization potential of C. albicans , C. tropicalis , N. glabratus , and C. parapsilosis within the Mucosal Ileum-SHIME ® model, alongside S. cerevisiae , a transient microbe of the human gastrointestinal tract (Auchtung et al. 2018 ).Despite their global pr e v alence (d'Enfert et al. 2021, Delavy et al. 2022 ), none of those fungal species were detected in the faecal samples used for inoculation.Even if they would be present (at le v el below the limit of detection), their nic he occupancy will have a negligible impact compared to that of the highly inoculated fungi.
Upon inoculation, the five fungi exhibited a remarkably similar behaviour across the different donors' SHIME ®, although colonized by distinct colonic micr obiomes.Notabl y, S. cerevisiae did not persist in the model, serving as a control to confirm that artificial inoculation with highl y concentr ated species does not enable tr ansient micr obes to engr aft the micr obiome.In addition, C. albicans and C. parapsilosis colonized all intestine compartments whereas C. tropicalis and N. glabratus concentrations remained below the limit of quantification.This indicated that Candida could successfully colonize the gut microbiome under the applied eubiosis physiological conditions, yet in a species-(or strain-) dependent manner.
The absence of growth of N. glabratus and C. tropicalis may be attributed to intrinsic colonization capabilities or potential interspecies competition in the compartments.Notably, C. albicans (23.7%) and C. parapsilosis (10.7%) were the most common Candida species found in the faeces of healthy volunteers in the Human Micr obiome Pr oject, compar ed to C. tropicalis (6.5%) and N. glabratus (0.5%) (Nash et al. 2017 ).This suggests that some species may be better adapted to thrive in the human gastrointestinal tract  (Sandai et al. 2012 ).The adequate simulation of a diverse range of sugars in the ileal nutritional medium and fibers in the fiber solution used in our Mucosal Ileum-SHIME ® model may ther efor e specificall y stim ulate micr obial species having a particular carbohydrate consumption flexibility, in casu C. albicans.
It is, ho w e v er, note worthy that the cell concentrations used for inoculating each fungal species were not uniform.These five fungal species were introduced into an established micr obial comm unity, whic h natur all y offers colonization r esistance (Lawley and Walker 2013 ).As a result, we reason that the ability of a fungal species to persist in any intestinal compartment does not depend on the inocula concentration but on inherent characteristics like adaptation to physiolog ical, ecolog ical, and envir onmental str essors within the existing microbial network and the SHIME ® model itself.Yet, the inoculum may affect interactions between the inoculated fungi, particularly if they compete for limited ecological nic hes.Consequentl y, the lac k of normalization in cell numbers r epr esents a limitation in our study when inter pr eting inter-Candida competition.
Evaluating the region-specific longitudinal colonization of C. albicans and C. parapsilosis is complex, given that both the proximal and distal colon compartments r eceiv ed luminal suspension from their preceding compartments three times daily.Yet, by the end of the eubiosis period, C. albicans and C. parapsilosis concentr ations wer e higher in both colon compartments of donor 1 and 2, r espectiv el y, as compar ed to their preceding ileum, inferring activ e gr owth.This higher le v el may stem fr om the longer r etention time in the colon compartments.To further substantiate our observations, we estimated the specific growth rates of each fungal species in each intestine compartment.We direct the reader to the Supplementary Specific gr owth r ate for the complete ov ervie w of the methods used and results obtained while only summarizing here our main findings.Higher specific gr owth r ates of C. albicans and C. parapsilosis in the ileum compartment compared to the colon ones were observed, suggesting that the lo w er bacterial diversity, and thus colonization resistance, and the presence of more abundant and diverse nutrients favoured their growth.Nonetheless, both fungal species demonstrated active growth in the colon compartments.Further r esearc h is, ho w e v er, r equir ed to confirm the growth capacity of each fungal species in separate SHIME ® colonic regions.
The production of lactate and the ratio between the three SC-FAs remained constant throughout the eubiosis study.This observation suggests a degree of stability in microbiome activity and diversity.Acetate and lactate are known to be converted to pr opionate and butyr ate by v arious colonic micr obes (Martin-Gallausiaux et al. 2021 ).Hence, the absence of alterations in metabolic profiles suggests a consistent cross-feeding pattern implying that the microbiome composition and its associated metabolic activity did not dr asticall y c hange .T his stability was further confirmed by the observation of a consistent microbiome composition at phylum and genus le v els.Mor eov er, the bacteriome α-diversity remained unaffected, and β-diversity displayed no significant clustering on PCA, D APC , or RD A throughout the eubiosis period.Similar findings wer e observ ed in mice in absence of antibiotic tr eatment (Gutierr ez et al. 2020, McDonough et al. 2021 ).The fact that artificial inoculation of Candida species in the Mucosal Ileum-SHIME ® did not disturb the gut microbiome diversity is particularly relevant considering we aimed to simulate Candida commensal lifestyle in our in vitro model, which should not be associated with bacterial disruptions under eubiosis conditions.
The commensal-to-pathogen transition of Candida is often associated with bacterial dysbiosis and Candida outgrowth (Zhai et al. 2020 ).Ho w e v er, the specific bacterial species and their metabolic functions involved in this transition remain poorly understood.Ther efor e, our second objectiv e was to induce bacterial dysbiosis using clindamycin and study the commensal-topathogen transition of Candida .Clindamycin is a lincosamide antibiotic known for its broad-spectrum activity against aerobic Gr am-positiv e cocci and a r ange of anaer obic Gr am-positiv e and Gr am-negativ e bacteria (Klainer 1987 ).It has pr e viousl y been shown to induce outgrowth of the mycobiome in vitro (Marsaux et al. 2023 ), and of C. albicans specifically in mice (Fan et al. 2015 ).
While the total bacterial concentration statistically and consistentl y decr eased in the three ileum compartments during the 2 weeks of antibiotic administration, limited impact was observed in the simulated colon microbiomes .T he distinction between the ileal and colonic microbiomes lies in bacterial div ersity, whic h was substantially lo w er in the ileum.Consequently, the ileal microbiome may exhibit limited resilience and resistance to antibiotic treatment, as was previously observed in the SHIME ® comparing babies and adults (Marsaux et al. 2023 ).Despite the limited impact on the colon total bacterial concentrations, bacterial activity, especiall y butyr ate and acetate, was significantl y r educed acr oss all intestinal regions.As a result, a substantial shift occurred to w ar ds higher propionate production compared to butyrate, which did not reverse upon discontinuation of the antibiotic treatment.
The 16S rRNA gene-targeted sequencing analysis further supported a strong shift at community composition levels due to antibiotic tr eatment.For instance, ther e was a dr astic shift fr om Bifidobacterium and Streptococcus to Bacteroides (supporting the incr eased pr opionate pr oduction), Enterococcus , and Lac hnoclostridium dominance in the colon microbiome of donor 1.These genera r emained lar gel y stable, with only Faecalibacterium showing substantial r ecov ery once the antibiotic administr ation ceased, likel y contributing to the r eincr eased butyr ate pr oduction.Additionall y, bacteriome α-div ersity decr eased during the antibiotic tr eatment but r ecov er ed by the end of the experiment, except in the distal colon, where it remained lo w er.The β-diversity analysis revealed significant clustering on PCA, D APC , or RD A, with samples before antibiotic treatment separated from all other samples.Importantly, it is likely that a more long-lasting shift happened in the SHIME ® model compared to in vivo , because there is no r eintr oduction of gut-viable micr obes fr om the envir onment.Nonetheless, similar findings have been reported in humans treated with ciprofloxacin (Dethlefsen and Relman 2011 ), whereby the microbiome e v entuall y r eac hed a stable but distinct composition 2 months after the treatment, without pathological symptoms associated.
Due to the changes in bacteriome activity and composition, C. albicans , C. parapsilosis , and C. tropicalis , but not N. glabratus and S. cerevisiae concentr ations, significantl y incr eased during the first week of antibiotic tr eatment, befor e their r eintr oduction.This observation is particularly significant for C. tropicalis , as it had fallen below the limit of quantification under eubiosis conditions.Ho w e v er, its r esur gence suggests that it was still present in the compartments, albeit at le v els too low to be quantifiable.Similar findings in both baby and adult SHIME ® studies were previously reported (Marsaux et al. 2023 ), indicating that opportunistic fungi may be unquantifiable under eubiosis conditions using molecular-based methods .T his implies that the pr e v alence of some Candida species may be underestimated in healthy individuals.
Following their r eintr oduction into the bior eactors, the concentrations of all fungal species increased during the second week of antibiotic treatment.Ho w ever, S. cerevisiae levels rapidly declined, confirming its inability to grow.By contrast, N. glabratus concentrations, unquantifiable until its reinoculation, remained stable in the ileum compartments, whereas it varied among donors in the colon compartments .T he ability of N. glabratus to colonize under conditions of dysbiosis but not eubiosis might be the consequence of a reduced bacteria α-diversity, which provides a reduced colonization resistance to w ar ds opportunistic fungi, as recently suggested (Marsaux et al. 2023 ), and/or opens up an ecological niche.Ho w e v er, the estimated specific growth rates (Supplementary Specific growth rate) sho w ed that C. tropicalis and N. glabratus did not gr ow in an y colon compartments, wher eas C. parapsilosis did in all the sim ulated pr o ximal colons.Ad ditionall y, the specific gr owth rates of C. albicans in the sim ulated pr oximal colons seemed to inv ersel y corr elate with the incr eased concentr ations in the simulated ilea, leading to cell decay during the second week of antibiotic treatment.
Our findings suggest that the source of Candida ultimately causing infections might be species-dependent.While N. glabratus might originate from a contaminated environment and is only able to colonize vulnerable indi viduals, lik e other nosocomial pathogens , C. albicans infections ma y arise from the endogenous microbiome itself, as suggested by other reports (Gabaldon andFairhead 2019 , Zhai et al. 2020 ).Ho w e v er, as all fungi were inoculated together, it must be noted that we cannot exclude interspecies competition in the different gut compartments, resulting in these different patterns of colonization.
The ability to persist in a r ecov ering micr obiome, upon cessation of antibiotic treatment, exhibited species-, intestinal region-, and donor-dependent variations.Concentrations of C. albicans , and to a lesser extent C. tropicalis , remained stable across all intestine compartments, whereas N. glabratus concentrations varied between donors.Candida parapsilosis concentrations remained r elativ el y high and stable in the colon compartments compared to the simulated ileum, confirming its adaptation.These novel observ ations pr ompt questions about the long-term effects of antibiotics, a well-documented phenomenon for bacteria but less understood in the context of the mycobiome (Dethlefsen andRelman 2011 , Seelbinder et al. 2020 ).Seelbinder et al. ( 2020 ) demonstrated that while the bacteriome recovered about 30 days postantibiotic treatment, the mycobiome was shifted from mutualism to competition in humans.Compared to their study, our results underscor e the v ariability in colonic micr obiome r ecov ery among indi viduals, indicating different susce ptibilities to long-lasting opportunistic emergence.
Finall y, ethanol pr oduction mirr or ed the kinetics of S. cerevisiae and N. glabratus , with a significant increase occurring only during the second week of antibiotic tr eatment, e v entuall y r eturning to pr etr eatment le v els by the end of the experiment.Notabl y, S. cerevisiae is known for its ability to perform alcoholic fermentation from glucose or fructose to ethanol (Maicas 2020 ).The contribution of the Candida species to produce ethanol under limited oxygen conditions remains to be determined.Importantly, the role of ethanol produced by the microbiome, both in general (Meijnikman et al. 2022 ) and by fungi in particular (Demir et al. 2022 ), in nonalcoholic fatty liver disease has been proposed.In our study, we observe a correlation between antibiotic use, colonic fungi, and ethanol production, providing a foundation for further r esearc h.

Conclusions
In this study, we investigated the spatial and temporal colonization and growth dynamics of Candida species in a multicompartmental semidynamic gastrointestinal system.Our r esearc h r evealed intriguing insights into how these fungal species interact with the intestinal environment and provided critical information about their behaviour under eubiosis and dysbiosis conditions.First, our findings highlighted the variability in the colonization potential of Candida species acr oss differ ent intestinal r egions .T he ileum compartment pr ov ed to be the most favourable environment for C. albicans and C. parapsilosis under conditions of eubiosis.Second, antibiotic-induced dysbiosis had a significant impact on both the fungal (mycobiome) and bacterial (bacteriome) communities in the gastrointestinal system.This disruption resulted in shifts in bacterial diversity and metabolite production.Notably, opportunistic Candida species, especially C. tropicalis and C. albicans , experienced a r esur gence during antibiotic-induced dysbiosis .T hese species were unquantifiable in some conditions under eubiosis conditions using molecular-based a ppr oac hes, suggesting that the pr e v alence of Candida presence may be underestimated in healthy individuals.Nakaseomyces glabratus gr e w under conditions of dysbiosis, but not eubiosis .Hence , our study also shed light on the source of Candida infections, which may vary depending on the species involv ed.Nakaseom yces glabratus , which colonized dysbiosed microbiome only, might originate from external environments and pose a risk to vulnerable individuals, while C. albicans infections could arise from the endogenous microbiome.Interspecies competition and microbiome diversity were identified as potential factors influencing colonization patterns.Understanding the factors driving fungal colonization and the consequence of microbial disruption is crucial for elucidating the dynamics of commensal and pathogenic fungal communities in the gut.Future research should focus on identifying specific bacterial species and their associated microbial functions influencing colonization resistance and explore the long-term effects of antibiotics on the mycobiome and bacteriome.

Institutional review board
The study was conducted in accordance with the Declaration of Helsinki and a ppr ov ed by the Ethics Committee of the University Hospital Ghent (r efer ence number B670201836585).

Informed consent
Informed consent of donors was obtained after providing them with detailed information about the project and the use of the samples.

Figure 2 .
Figure 2. Candida albicans (A), C. tropicalis (B), C. parapsilosis (C), N. glabratus (D), S. cerevisiae (E), and total bacteria concentrations (F) in the faecal samples from donor 1, 2, or 3, and in the fungal inocula used during the eubiosis stud y at days 35-37 and the d ysbiosis stud y at days 63-65 (G).Bo x plots are box and whiskers showing minimum and maximum values.Significant differences were assessed using Mann-Whitney tests for comparisons between the fungal inocula used in the two study periods, Kruskal-Wallis tests with Dunn's multiple comparisons for differences between fungal species within inocula, and Kruskal-Wallis tests with Dunn's multiple comparisons for total bacteria concentrations between the three faecal samples.Significant differences are marked with asterisks ( * P < .05,* * P < .01,* * * P < .001,and * * * * P < .0001).

Figure 3 .
Figure 3. Candida albicans (A), C. tropicalis (B), C. parapsilosis (C), N. glabratus (D), S. cerevisiae (E), and total bacteria concentrations (G) measured by qPCR ( n = 3) (A) and total live bacteria concentrations measured b y flo w c ytometry ( n = 1) (H) in the ileum, and the proximal and distal colon compartments inoculated at day 0 with the faecal samples from donor 1, 2, or 3.The fungal species values were corrected with flow cytometry data obtained on their r espectiv e pur e ov ernight cultur e, hence r esulting in cell counts instead of gene copy counts .T he uncorrected a v er a ge v alues acr oss the thr ee donors for each intestinal region and each species is also provided (F).Samples were collected during the baseline period (d19-d35), during the eubiosis study (d36-d56) following the fungal inoculations for three consecutive days (d35-d37), during the first week of clindamycin treatment (d56-d63; AB = antibiotic), during the second week of clindamycin treatment (d63-d70; AB + F = antibiotic with fungal reinoculation), which started with the reinoculation of the fungal species for three consecutive days (d63-d65), and during the recovery period (d70-d79; Reco = reco very).T he fungal (re-) inoculations are indicated by an horizontal line .T he limit of quantification ( = LOQ) is indicated by an horizontal dotted line.

Figure 5 .
Figure 5. Quantitativ e micr obiome pr ofiling fr om the ileum inoculated with a bacterial consortium, fr om the pr oximal colonic, and distal colonic compartments inoculated with faecal inoculum of donor 1 (A), 2 (B), or 3 (C), and their resulting bacterial α-diversity indexes (D-F) and β-diversity (G-K), from samples collected during the baseline and eubiosis periods.(A-F) The dotted lines separate the baseline period from the eubiosis one, which starts with the inoculation of the fungal species for thr ee consecutiv e days (d35-d37).(A-C) Each bar represents the total bacteria concentration in the sample, whereas the colours represent the proportion of each bacterial genera.Gut bacteriome β-diversity based on operational taxonomic unit (OTU) r elativ e abundances and anal ysed thr ough separ ated PCA for all ileum (G), pr oximal (H), and distal (I) colon compartments, and through DAPC (J) and RDA (K) for all colon compartments only.PC = proximal colon; DC = distal colon; DAPC = discriminant analysis of principal component; LD = linear discriminant; RDA = redundancy analysis; PCA = principal component analysis; and PC ( x -and y -axis of PCA plot) = principal component.parapsilosis .On the other hand, C. parapsilosis , C. albicans , and C. tropicalis concentr ations significantl y incr eased in the distal colon compartments, while S. cerevisiae and N. glabratus concentrations remained below the limit of quantification, except for donor 2. Candida albicans le v els especiall y incr eased in the distal colon compartment of donor 2. During the r ecov ery period, C. albicans , N. glabratus , and C. tropicalis concentrations remained stable across all intestinal regions.Candida parapsilosis significantly decreased in the

Figure 6 .
Figure 6.Quantitativ e micr obiome pr ofiling fr om the ileum inoculated with a bacterial consortium, fr om the pr oximal colonic, and distal colonic compartments inoculated with faecal inoculum of donor 1 (A), 2 (B), or 3 (C), and their resulting bacterial α-diversity indexes (D-F) and β-diversity (G-K), from samples collected during the antibiotic treatment and the recovery periods.(A-F) The left dotted lines separate the end of the eubiosis period from the antibiotic treatment during which the five fungal species were reintroduced in the bioreactors (d63-d65), and the right dotted lines the end of the antibiotic treatment and the start of the recovery period.(A-C) Each bar represents the total bacteria concentration in the sample, whereas the colours r epr esent the proportion of each bacterial genera.Gut bacteriome β-diversity based on OTU relative abundances and analysed through separated PCA for all ileum (G), proximal (H), and distal (I) colon compartments, and through DAPC (J) and RDA (K) for all colon compartments only.PC = proximal colon; DC = distal colon; DAPC = discriminant analysis of principal component; LD = linear discriminant; RDA = redundancy analysis; PCA = principal component analysis; and PC ( x -and y -axis of PCA plot) = principal component.