The O₂ gradient cassette was composed of a basal fluid reservoir, a customized hanging basket (similar to a Transwell insert) placed into the basal reservoir, and an O₂-impermeable machined plug for insertion into the luminal portion of the hanging basket (Supplementary Figure S1). A conventional 12-well plate was used for the basal reservoirs. The design and fabrication of the hanging basket was similar to that described previously (Gunasekara et al., 2018; Kim et al., 2019). Briefly, the hanging basket was fabricated from polycarbonate using a CNC milling machine. Next, a porous polyester (PET) membrane (0.4 µm pores, Sterlitech Cat #1300016) was attached to the bottom of the hanging basket using double-sided medical grade tape (3M, Cat #1504XL). Five minutes of plasma treatment (Harrick Plasma, Cat #PDC-32G) improved adhesion of subsequently added collagen to the basket. The basket with 200 µL neutralized collagen (rat tail, type I, Corning, Cat #354236) solution (1 mg/mL) (Kim et al., 2019; Kim et al., 2022) was incubated for 1 h at 37°C and then submerged in 3 mL crosslinking solution added to the basal reservoir for 30 min at 25°C. The crosslinking solution was 0.6 M 1-Ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC): 0.15 M N-hydroxysuccinimide (NHS): 0.1 M 2-(N-morpholino) ethanesulfonic acid (MES) buffer (pH 5.0) in water, (1:1:1 (vol). Later, the basket with partially crosslinked collagen matrix (1.1 cm² of surface area and 2.2 mm of thickness (Gunasekara et al., 2018)) was incubated in deionized water for >18 h at 25°C to leach out residual crosslinking reagent followed by sterilization in 70% ethanol and was then air-dried in a biosafety cabinet. The basket with collagen matrix could be stored in PBS buffer at 4°C for 2–3 weeks before adding the human primary colon cells. The cells were cultured aerobically to form a confluent monolayer on the collagen/PET membrane scaffold submerged in a well of the 12-well plate (see below). For O₂ gradient, an O₂-impermeable plug fabricated from polycarbonate was threaded into the luminal portion of the basket. A soft Viton^® fluoroelastomer O-ring (McMaster Car, Cat #1284N114) was used with the plug to form an O₂-impermeable seal. A port in the plug was used for insertion of an O₂ probe, and this port was sealed with ethylene propylene diene monomer (EPDM) rubber caps (McMaster-Carr, Cat #6448K117) when an O₂ sensor was not in place. The O₂ probe in the luminal compartment was a needle O₂ probe (PreSens, NTH-PSt7/Microx4, Germany). The tip of the probe was submerged into the luminal media to approximately 2 mm above the epithelial monolayer when O₂ was measured.

Primary human colonic epithelial stem cells from the transverse colon of a cadaveric donor (male, 23 years old, RRID: CVCL_ZR41 (https://web.expasy.org/cellosaurus/CVCL_ZR41)) were isolated, expanded, and maintained on a collagen scaffold in a 6-well plate in a stem cell media (SM, Supplementary Table S1) as described previously (Wang et al., 2017; Kim et al., 2019). Cells were used between passage 7 and 12. For each experiment, the human colonic stem cells were sub-cultured from a single well of the maintenance culture into 6 hanging baskets with collagen matrix prepared as described above. The human colonic cells were expanded on the collagen matrix in the hanging basket in expansion medium (EM, Supplementary Table S1 0.5 mL placed in the luminal compartment and 3 mL placed in the basal compartment) for 6 days with media exchange every other day (Kim et al., 2019; Kim et al., 2022). After a confluent monolayer was formed on day 7, both luminal and basal media were replaced by differentiation medium without antibiotics (DM-Ab, Supplementary Table S1 0.5 mL placed in the luminal compartment and 3 mL placed in the basal compartment). On day 8, the luminal medium was replaced with 0.5 mL 10% peptone-yeast-fructose media (PYF, Supplementary Table S2) in 1× PBS.

C. scindens (ATCC 35704) was purchased from ATCC and propagated within an anaerobic chamber (Coy Laboratory Products). The bacteria were cultured in PYF medium (peptone 10 g/L, yeast extract 10 g/L, fructose 10 g/L, resazurin 1 mg/L, L-cysteine hydrochloride 0.5 g/L, CaCl₂·H2O 100 mg/L, MgSO₄·7H₂O 50 mg/L, K₂PO₄ 40 mg/L, KH₂PO₄ 40 mg/L, NaHCO₃ 0.4 g/L, NaCl 80 mg/L, hemin (5 mg/L) and vitamin K (1 mg/L)). C. scindens was plated on PYF agar (PYF with 15 g/L agar) and cultured anaerobically at 37°C using a BD GasPak EZ pouch system (BD, Cat # 260683). Two days before each experiment a fresh subculture was generated from a single colony for co-culture experiments by culturing in the PYF medium for 48 h.

S. aureus strain Newman was kindly provided by Dr. Michelle Reniere at the University of Washington and cultured under aerobic conditions in brain-heart-infusion media (BHI, Research Products International, Cat #B11000) at 37°C. S. aureus was plated on BHI agar (BHI with 15 g/L agar), and a single colony was used to prepare a bacterial sample after 18 h of culture in BHI broth before co-culture experiments.

After human colonic epithelial cells fully differentiated in the luminal compartment of the hanging baskets, the luminal medium was substituted with a deoxygenated medium consisting of 10% PYF in 1× PBS buffer. This co-culture medium was enriched with 200 µM cholic acid for bile acid metabolism. Simultaneously, the basal medium was refreshed with fresh DM-Ab media. For C. scindens co-culture, 5×10³ colony-forming unit (CFU) in total were added to the luminal media similar to that used previously for other Clostridioides (Kim et al., 2019). The sealed co-culture or control cultures without bacteria were incubated at 37°C for 48 h. Due to its rapid growth, a minimal number of S. aureus (5 × 10² CFU in total) was added to specified samples after 24 h of co-culture with C. scindens. The samples with added S. aureus were then incubated at 37°C for an additional 24 h. The deoxygenated luminal media was 10% PYF in 1 x PBS buffer for both C. scindens and S. aureus. All inoculations were performed in an anaerobic chamber.

To assess the viability of colonic epithelial cells cultured in the O₂ gradient cassette, DM-Ab medium (3 mL) supplemented with 10 μg/mL Hoechst 33342 (Thermo Fisher, Cat #H1399) was added to the basal compartment and then the cells were incubated at 37°C for 1 h. After washing with PBS, 2 μg/mL propidium iodide (PI, Thermo Fisher, Cat # P3566) in PBS buffer (3 mL) was added to the basal compartment, and then incubated at 37°C for 20 min. Hoechst 33342 was used to label DNA in all epithelial cells, and PI was used to label the DNA of dead epithelial cells. To measure apoptosis, a dead cell apoptosis kit with Annexin V (Thermo Fisher, Cat #V13242) was used to assess the presence of phosphatidylserine on the plasma membrane of cells. The working solution of the kit contained 1 μg/mL Annexin V, 1 μg/mL PI, 1.25 mM HEPES, 7 mM NaCl, 0.05 mM EDTA, and 0.005% bovine serum albumin (BSA) in 1× binding buffer (10 mM HEPES, 140 mM NaCl, and 2.5 mM CaCl₂). 500 μL of the working solution was used for luminal staining at 25°C for 15 min, then the reaction was terminated by mixing with 2 mL binding buffer and kept on ice before imaging.

Prior to microscopic imaging, each collagen scaffold with attached cells was carefully cut from the hanging basket, inverted, and placed onto a #1 coverslip. A confocal fluorescence microscope (Olympus, Fluoview 3,000, Japan) was used to measure the fluorescence of the epithelial cells. The entire epithelial monolayer (approximately 1.1 cm² surface area) was imaged as a 4 × 4 array with a 4× (0.16 N.A.) objective. The following fluorescence excitation/emission wavelengths were used- Hoechst 33342: Ex 405 nm, Em 430–470 nm; PI: Ex 561 nm, Em 610–670 nm; Annexin V: Ex 488 nm, Em 530–575 nm. All 16 images of each channel were stitched together to create an image of the entire well.

CellProfiler v4.0 was used to identify and calculate the image area with fluorescence intensity above an empirically set threshold to quantify the area of the epithelial monolayer positive for each of the stains. The processing protocol was modified from a previously reported method (Hinman et al., 2021). The stitched image at each emission wavelength was flat-field corrected using a Gaussian-smoothed illumination correction function (1000-pixel filter size). The corrected image was converted to binary using 2-class Otsu threshold selection. Then the area of the culture above the fluorescence threshold was calculated. The normalized PI + area was calculated as PI + area/Hoechst 33342+ area and was used as a metric for dead cells. The normalized Annexin V+ area was calculated as Annexin + area/Hoechst 33342+ area and used as a metric for apoptotic cells.

To determine whether tight junctions among the epithelial cells were disrupted in the presence of the bacteria, methanol (4°C) was added as a fixative into the luminal compartment, and the sample was placed at −20°C for 1 h. The fixed cells were then washed in PBS and blocked with 3% BSA in PBS for 1 h at 25°C. Next, 200 µL ZO-1 primary antibody (2 μg/mL, Proteintech, Cat #21773-1-AP, RRID:AB_10733242) was added to the luminal compartment and incubated with the cells at 4°C for 18 h. The sample was then washed 3 times with IF wash buffer (0.1% BSA, 0.2% Triton-X 100, 0.05% Tween 20 in 1× PBS) and then incubated with 500 µL secondary antibody conjugated with Alexa Fluor 647 (1 μg/mL, Thermo Fisher Scientific, Cat #A27040, RRID:AB_2536101) containing 10 μg/mL Hoechst 33342 for 1 h at 25°C. The sample was then washed twice with PBS buffer. Prior to confocal microscopy, each collagen scaffold with attached cells was carefully cut from the hanging basket, inverted, and placed on a #1 coverslip. Fluorescence was then quantified using a 30× objective oil lens (1.05 N.A.). Fluorescence was measured using the following parameters- ZO-1: Ex 640 nm, Em 650–750 nm; Hoechst: Ex 405 nm, Em 430–470 nm. Five different locations (top, bottom, right, left, center) of each sample were imaged. The images were processed and analyzed within CellProfiler v4.0 following a previous protocol (Hinman et al., 2021).

The luminal contents (media + collagen scaffold) were collected and mixed in a tube with 3 mL 500 U/mL collagenase in PBS, followed by 30 min incubation at 25°C under anaerobic conditions. Next, the suspension was vortexed for 2 min, and then plated on PYF (for C. scindens) or BHI (for S. aureus) agar. The CFU on the agar plates after 24–48 h incubation anaerobically (for C. scindens) or aerobically (for S. aureus) was counted and used as a measure of the viability of the bacterial cells.

To visualize the bacterial cells while still present on the epithelial monolayer, 200 µL 1 μM SYTO 9 (Thermo Fisher, Cat #S34854) in PBS was added to the luminal compartment for 20 min. Epithelial cells were visualized by addition of Hoechst 33342 (10 μg/mL, PBS) followed by incubation for 1 h. The collagen scaffold with attached cells was carefully cut from the hanging basket, inverted, and placed on a #1 coverslip. Both bacterial and epithelial cells were imaged by confocal fluorescence microscopy using a 60× oil objective lens (1.3 N.A.) under the following parameters- Hoechst 33342: Ex 405 nm, Em 430–470 nm; SYTO9: Ex 488 nm, Em 500–520 nm.

The IL-8 concentration in the basal media was measured using an interleukin 8 (IL-8) ELISA kit (Thermo Fisher Scientific, Cat #88-8086-22) following the manufacturer’s protocol. The basal media were diluted at 1:10 (control) or 1:20 (co-culture) so that the measured signals fit in the range of the standard curve provided with the kit. The MCP-1 concentration in the basal media was measured using a monocyte chemoattractant protein-1 (MCP-1) ELISA kit (Thermo Fisher Scientific, Cat #88–7399-88) following the manufacturer’s protocol.

After 48 h of co-culture in the presence of bile acid, both the luminal contents (media + collagen matrix) and basal media were collected separately. The luminal contents were degraded by collagenase as described above and the supernatant collected after centrifugation (10 min, 3,000×G). The luminal supernatant and basal media were assayed for bile acids by liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis as described previously (Ginos et al., 2018; Navarro et al., 2020) at the University of Washington Mitochondria and Metabolism Center.

As a control, cholic or deoxycholic acid was added to S. aureus cultured in 96-well plates. Diluted S. aureus (1/10⁶) was added to wells containing 200 µL 10% PYF (in 1× PBS) medium supplemented by 25–200 µM cholic acid or deoxycholic acid and incubated anaerobically. After 24 h at 37°C, the optical density (O.D.) at 600 nm of the medium in each well was measured.

The luminal contents (media + collagen scaffold) were collected and mixed with collagenase as described above. The supernatant was collected after centrifugation. The concentration of S. aureus enterotoxin in the supernatant was measured using a sandwich enzyme immunoassay kit (RIDASCREEN^® SET Total, R-Biopharm AG, German) following the manufacturer’s protocol. The O.D. of each sample was measured and an O.D. of 0.15 greater than that of the negative control value was set as a threshold for a positive sample. The concentration of enterotoxin was represented as relative O.D., i.e., normalized to the positive control mean (set equal to 100%).

Origin 9 was used for statistical analyses of the data. All p values were calculated by ordinary one-way ANOVA. A significance level of 0.05 was used, and any p-value below this threshold was considered statistically significant. (*: p < 0.05, **: p < 0.01, ***: p < 0.001). In this manuscript, we define a “technical replicate” as being separate experiments on different days but using the cells of a single donor.

C. scindens is a bacterium often found in the human colon (White et al., 1980; MORRIS et al., 1985). C. scindens is remarkable for its 7-dehydroxylation activity that transforms primary bile acids (e.g., cholic acid) to secondary bile acids (deoxycholic acid and lithocholic acid). Secondary bile acids are reported to inhibit the growth of intestinal pathogens, such as S. aureus, supporting a role for C. scindens as a probiotic species enhancing resistance to gastrointestinal infection (Buffie et al., 2015). S. aureus is a common cause of gastrointestinal illness after consumption of S. aureus-contaminated food (CDC, 2018). The toxicity of S. aureus is derived from enterotoxins that increase the permeability of the epithelial barrier and initiate epithelial cell death (Moretó and Pérez-Bosque, 2009; Šuligoj et al., 2020). The overall goal of this study was to co-culture these two bacterial species with primary human gastrointestinal epithelial cells to demonstrate the utility of multi-species gut-on-chip models for intestinal microbiota research and the study of the interactions between the microbiome and human host.

Clostridial species such as C. scindens tolerate O₂ concentrations up to 2%–3% (Morvan et al., 2021; Chia, 2023), but higher concentrations are lethal. These obligate anaerobes are able to survive in the environment of the luminal-facing differentiated colonic epithelium that exists in physiologic hypoxia of <1–3% O₂. In contrast, the O₂ concentration increases to 10% near the colonic epithelial cells at the crypt base, i.e., deeper in the submucosa, where the stem-cell niche produces proliferative cells to replenish the epithelial layer (Keeley and Mann, 2019; Schwerdtfeger et al., 2019). To achieve these distinct O₂ conditions, a modified cell culture cassette for O₂-gradient formation was previously demonstrated for co-culture of epithelial cells and anaerobes (Kim et al., 2019; Kim et al., 2022). This cassette, based on a hanging basket, possessed luminal and basal reservoirs so that both sides of the colonic epithelial cell culture were accessible. To create an O₂ gradient across the tissue surface, the side walls of the luminal compartment were impervious to O₂ and an O₂ impermeable plug was inserted into the luminal reservoir to prevent atmospheric O₂ ingress. In contrast, the basal reservoir was open to the environment. Under these conditions, a confluent monolayer of epithelial cells has been shown to consume O₂ at a sufficient rate to decrease the luminal O₂ to <1% in 3.75 h (Kim et al., 2019). In the current experiments, epithelial stem cells were plated on the luminal side of the hanging basket’s porous membrane in the cassette and cultured in a medium rich in growth factors (expansion medium, EM) without the luminal plug so that a high O₂ tension was maintained (Figures 1A,B). After forming a confluent monolayer, the cells were fully differentiated by removal of growth factors (differentiation medium without antibiotics, DM -Ab) producing a high-resistance epithelial monolayer (TEER = 507 ± 126 Ω·cm², n = 3, Figure 1C). The luminal DM -Ab was then replaced with a minimal nutrient medium (PYF) suitable for bacterial culture (Figure 1D). The cassette was then transferred into an anaerobic chamber (Figure 1E), the plug was installed, and the assembled cassette was placed in a standard tissue-culture incubator (Figure 1F). Under these conditions, the maximal O₂ concentration in the luminal compartment was 0.6% ± 0.4% after plug installation and decreased to a steady state value of 0.4% ± 0.1% during culture due to continued O₂ consumption by the epithelial cells (Supplementary Figure S2). C. scindens was subsequently inoculated onto the intestinal epithelium at 10³ colony-forming unit (CFU)/ml (1,300 ± 500 cells in total). The CFU increased to 3×10⁴ ± 0.95 × 10⁴ CFU/mL (approximately 15,500 ± 5,000 in total, doubling time 13 h) after 48 h (p = 0.045, Supplementary Figure S3A), suggesting that C. scindens not only tolerated the luminal microenvironment but was able to replicate.

The in vivo colonic lumen acts as a host to C. scindens with co-existence of and mutual benefit for both the mammalian and bacterial cells. C. scindens ferments fiber and metabolizes bile acids to support both the health of the microbial community as well as that of the epithelial cells (Wlodarska et al., 2017; Kang et al., 2019). In turn the epithelial cells form a tight barrier to prevent the intrusion of bacteria into the tissue and collaborate with the immune system to repel invading microbes (Chen et al., 2021). To determine whether the epithelial cells might be able to co-exist with the obligate anaerobes in the absence of an immune system, C. scindens was cultured on a confluent, differentiated epithelial cell layer. The impact on epithelial cell viability and monolayer barrier function was then assessed. When stained with propidium iodide (PI), no significant difference in epithelial cell death was observed in the presence and absence of C. scindens (9.8% ± 7.5% PI + area normalized to total Hoechst 33342+ cell area for the control vs. 10.8% ± 4.2% after 48 h of co-culture with C. scindens, p = 0.64, Figure 2A; Supplementary Figure S4A). Programmed cell death was assessed using an annexin V conjugate assay to identify phosphatidylserine in the outer membrane leaflet. No significant difference in the binding of annexin V to the epithelial cells was observed in the absence and presence of C. scindens (9.5% ± 4.0% annexin V+ area normalized to total Hoechst 33342+ cell area for the control vs. 6.1% ± 1.6% after 48 h of co-culture with C. scindens, p = 0.24, Figure 2B; Supplementary Figure S4B). Barrier integrity was measured by assaying the TEER after 48 h of culture with and without C. scindens. In both instances the TEER decreased over time but was not significantly different (control 168 ± 18 vs. 150 ± 37 Ω·cm² with C. scindens, p = 0.45, Supplementary Figure S5A). Since TEER represents a global measurement of barrier function, the integrity of cell-cell junctions was also assessed by fluorescent staining for ZO-1, which is a constituent of tight junctions located on the cytoplasmic membrane surface (McNeil et al., 2006). When visualized by microscopy, cells with and without C. scindens co-cultured for 48 h possessed similar ZO-1 staining patterns (Figure 2C). When the area, circularity, and perimeter of the cells was measured, no significant differences were identified for the epithelial cells with and without C. scindens (area: 77 ± 19 μm² without vs. 70 ± 15 μm² with, p = 0.3; circularity: 0.78 ± 0.02 without vs. 0.79 ± 0.02 with, p = 0.2; perimeter: 34 ± 5 µm without vs. 32 ± 4 µm with, p = 0.3, N = 3 technical replicates, Supplementary Figure S6). These data suggest that although the number of C. scindens expanded 10-fold, the epithelial cells did not die or lose barrier integrity during the 48 h of co-culture. However, extremely high numbers of inoculated C. scindens (∼5×10⁶ cells) did result in an increase in epithelial cell death at 48 h of co-culture (Supplementary Figure S7).

IL-8 and MCP-1 are chemokines produced by epithelial cells in response to bacterial products and are often used as markers of an epithelial cell inflammatory response (Deshmane et al., 2009; La Pine and Hill, 2011). The concentrations of IL-8 and MCP-1 in the basal media were measured in the absence and presence of C. scindens co-culture after 48 h. The concentration of IL-8 was not significantly different in the two culture conditions (687 ± 149 pg/mL without vs. 789 ± 199 pg/mL with C. scindens, p = 0.38, Supplementary Figure S8A). Likewise, the concentration of MCP-1 was not significantly different (9.8 ± 2.1 pg/mL without vs. 7.4 ± 1.8 pg/mL with C. scindens, p = 0.21, Supplementary Figure S8B). The concentration of IL-8 secreted into the basal reservoir was similar to that in previous studies (300–600 pg/mL) (Wang et al., 2018b; Kim et al., 2022). Additionally, the concentration of MCP-1 was near that reported for colonic epithelial cell cultures (6.9–27 pg/mL) (Futagami et al., 2003; Fujita et al., 2010). Under these conditions, the co-culture of C. scindens did not alter the inflammatory response of the colonic cells.

Bile acids flow from the liver into the small intestine via the bile duct. The major portion of bile acids (∼80%) are actively transported back to liver in the terminal ileum. However, ∼15% of bile acids transit through the digestive tract to reach the colon and are then metabolized by bacteria into secondary bile acids that diffuse passively across the colonic epithelium to return to the liver (Di Ciaula et al., 2017). C. scindens is known to convert primary bile acids such as cholic acid into a variety of secondary bile acids (e.g., deoxycholic acid and lithocholic acid) in the colon (White et al., 1980; Ridlon et al., 2014; Kang et al., 2019). To assess bile acid conversion by C. scindens, cholic acid (200 µM) was added in the luminal reservoir in the presence of epithelial cells with or without C. scindens and the concentrations of cholic acid and its metabolites were measured by mass spectrometry (Figure 3). In the absence of C. scindens, only cholic acid (99.37% of total luminal bile acids) was found in the luminal reservoir. In these cultures, cholic acid was also detected in the basal compartment (98.06% of total basal bile acids). In contrast, the luminal reservoir of epithelial cells co-cultured with C. scindens demonstrated a range of secondary bile acids (>99% conversion) with the predominant being apocholic acid (58.33% of total luminal bile acids), which is a byproduct of the dehydration of cholic acid (Shi et al., 2023). Other prominent metabolites identified in the luminal compartment included 3α-hydroxy-12 ketolithocholic acid (21.96% of total luminal bile acids) and deoxycholic acid (11.84% of total luminal bile acids). These two metabolites are formed via a dehydroxylation pathway of C. scindens and have been observed previously in vitro and in vivo (Marion et al., 2019; Funabashi et al., 2020). The basal compartment possessed a distinct pattern of bile acids. The predominant secondary bile acid was deoxycholic acid (50.49% of total basal bile acids) in the presence of C. scindens, but untransformed cholic acid was also present (41.61% of total basal bile acids) as were other products. The presence of the bile acids in the basal compartment was most likely due to passive diffusion as occurs in the colon in vivo (Mekhjian et al., 1979). These data suggest that molecular interactions between C. scindens-epithelial cells, such as bile acid metabolism were present in this in vitro system.

S. aureus is a facultative anaerobe normally found on human skin and within the nasal cavities but can be an opportunistic pathogen of the colon (O'Gara, 2017). S. aureus is able to produce a range of toxins including enterotoxins (SEs), hemolysins, and leukotoxins (Kong et al., 2016) so that colonization of the colon by S. aureus can lead to severe nausea, vomiting, and diarrhea (CDC, 2018). To examine the cytotoxicity of S. aureus to colonic epithelial cells, we inoculated S. aureus onto the epithelial monolayer at 900 ± 370 CFU/mL (450 ± 190 cells in total). After 24 h of co-culture in the anaerobic cassette, S. aureus numbers increased to 1.4×10⁸ ± 6 × 10⁷ CFU/mL (7.0×10⁷ ± 2.8×10⁷ total cells), which was significantly higher than the initial inoculum (p = 0.02, Figure 4A; Supplementary Figure S3B) and represents a doubling time of ∼83 min. Growth of S. aureus on the epithelial cell culture was substantially more robust than that of C. scindens (Moretó and Pérez-Bosque, 2009; Martens et al., 2021). As the epithelial cells died as a result of co-culture with S. aureus, many nutrients may have become available to benefit the growth of S. aureus. In these co-cultures, S. aureus grew both in a dispersed manner as well as in clusters or sheets of bacteria fully covering the monolayer, suggesting the formation of biofilms.

The impact of S. aureus on the epithelial cells was assessed by measuring epithelial cell death and monolayer barrier integrity. Compared to control cultures, the viability of colonic epithelial cells in the presence of S. aureus was significantly decreased after 24 h of co-culture (9.8% ± 7.5% PI + area normalized to total Hoechst 33342+ cell area for the control vs. 68% ± 16% after co-culture with S. aureus, p = 0.0007, Figure 5E). Visually, a majority of the epithelial cells appeared dead in the cultures (Figure 4B). This is consistent with the measured TEER of 70 ± 12 Ω·cm² (p < 0.001 relative to control, Supplementary Figure S5B), indicating a complete loss of barrier function in the presence of S. aureus after 24 h. To examine cell-cell tight junctions, the epithelial monolayers were stained for ZO-1, which showed a loss of cell boundaries and an appearance of large aggregates of ZO-1 in the colonic cells. This finding indicated that the intercellular tight junctions were disrupted in these cultures (Figure 4C). Further confirmation was obtained by quantification of the epithelial cells’ morphology with the area, perimeter, and circularity all significantly different compared to that of colonic cells cultured in the absence of S. aureus (area: 77 ± 19 μm² without vs. 161 ± 43 μm² with, p < 0.001; circularity: 0.78 ± 0.02 without vs. 0.69 ± 0.04 with, p < 0.001; perimeter: 34 ± 5 µm without vs. 54 ± 7 µm with, p < 0.001, N = 3 technical replicates, Supplementary Figure S6). The area and perimeter of the epithelial cells increased in the presence of S. aureus while the circularity decreased, in agreement with the observed distorted cell boundaries on the micrographs. Consistent with these results, the IL-8 concentration in the basal compartment of the epithelial cells was significantly increased after 24 h of S. aureus co-culture compared to that for the epithelial cells alone (531 ± 196 pg/mL vs. 1,382 ± 329 pg/mL, p = 0.02, Figure 5F). It is noteworthy that S. aureus was inoculated onto the epithelial monolayer at a 2-to-3-fold lower CFU than C. scindens. The shorter S. aureus doubling time and potentially toxin production under these conditions enabled S. aureus to expand in number rapidly, leading to a cellular inflammatory response, cell death, and destruction of the monolayer barrier integrity.

As a commensal strain, C. scindens exerts a protective effect on the colon by converting 7α-dehydroxylating primary bile acids to form secondary bile acids which then inhibit pathogenic bacterial growth (Greathouse et al., 2015; Kang et al., 2019). To explore this defensive behavior of C. scindens, epithelial monolayers were colonized with C. scindens by co-culture for 24 h. S. aureus was then inoculated onto the C. scindens/epithelial cell co-culture (Figure 5A). After an additional 24 h of incubation, epithelial cell death, cell morphology, and secreted cytokines were measured. Pre-culture of the epithelial cells with C. scindens significantly enhanced the viability of the epithelial cells after addition of S. aureus (p = 0.02, Figures 5B,E). Remarkably, epithelial cell death in the presence of both C. scindens and S. aureus was not significantly different from epithelial cells in the absence of bacteria (p = 0.18, Figure 5E). Furthermore when ZO-1 immunofluorescence was measured, the presence of pre-cultured C. scindens also reduced the alterations in cell morphology in the co-culture with S. aureus compared to culture in the absence of C. scindens (area: 161 ± 43 μm² with S. aureus only vs. 116 ± 30 μm² with both, p = 0.01; circularity: 0.69 ± 0.04 with S. aureus only vs. 0.74 ± 0.03 with both, p = 0.003; perimeter: 54 ± 7 µm with S. aureus only vs. 43 ± 7 µm with both, p = 0.008, N = 3 technical replicates, Supplementary Figures S6, S10). The presence of C. scindens, however, did not significantly alter the amount of IL-8 secreted in the presence of S. aureus alone (p = 0.8, Figure 5F) suggesting that the epithelium was still inflamed by the presence of the S. aureus, but protected from death by the co-culture with C. scindens. The measured S. aureus CFU in the absence or presence of C. scindens was not significantly different (7.0×10⁷ ± 2.8×10⁷ without vs. 6.3×10⁷ ± 2.4×10⁷ with C. scindens, p = 0.83, Supplementary Figure S3B), therefore repression of S. aureus growth was not an explanation for the increased epithelial cell survival, which was affirmed when the two bacteria were co-cultured in the absence of an epithelium (Kang et al., 2019).

Secondary bile acids are reported to diminish the negative impact of S. aureus by disruption of the bacterium’s membrane (Sannasiddappa et al., 2017), and by minimizing excessive inflammation through inhibition of NF-κB and NLRP3 pathways (Zhao et al., 2023). Cholic acid (200 µM) was added to the luminal compartment during C. scindens colonization to enable the formation of secondary bile metabolites. S. aureus was then added, and epithelial cell death and IL-8 secretion were measured after 24 h. Epithelial cell death was not significantly decreased by the addition of cholic acid to the C. scindens-S. aureus mixed culture (p = 0.13, Figure 5B,D,E). IL-8 production was also not significantly altered in the presence of cholic acid relative to that without cholic acid (p = 0.92, Figure 5F). Cholic acid added to epithelial monolayers with only S. aureus (no C. scindens) did not significantly alter epithelial cell viability (p = 0.88) or IL-8 production (p = 0.46) compared to S. aureus only co-cultures (Figure 5C,E,F). The measured S. aureus CFU in the absence or presence of C. scindens + cholic acid was not significantly different (7.0×10⁷ ± 2.8×10⁷ without vs. 6.6×10⁷ ± 3.8×10⁷ with C. scindens/cholic acid, p = 0.93, Supplementary Figure S3B), so repression of S. aureus growth was again not a likely explanation for the improved epithelial cell survival. These results depart from the findings of prior studies in which bile acids repressed the growth of S. aureus (Sannasiddappa et al., 2017; Kang et al., 2019); however, we note that epithelial cells were absent from those experiments. For this reason, we cultured S. aureus in the presence of cholic acid and deoxycholic acid but without epithelial cells. We observed a significantly decreased growth rate as measured by O.D. (Supplementary Figure S9). It is likely that the epithelial cells, C. scindens, or their metabolic products other than bile acids are acting in a complex manner to impact S. aureus growth.

S. aureus enterotoxins (SEs) are potent exotoxins that damage colonic epithelial cells by stimulating the secretion of cytokines and provoking inflammation (Aljasir and D’Amico, 2020; Cretenet et al., 2011). Enterotoxin B (SEB) also disrupts the integrity of epithelial barriers, initiating cell death (Martens et al., 2021). To determine whether C. scindens reduces the cytotoxicity of S. aureus by altering toxin formation, enterotoxin (SE) production by S. aureus was measured in the luminal reservoir. For simplicity, the selected assay measured total enterotoxin (SEA, SEB, SEC, and SED). Addition of C. scindens with or without cholic acid significantly diminished total SE production by S. aureus compared to cultures without C. scindens pre-culture (p = 0.007, C. scindens; p = 0.012 C. scindens + cholic acid) (Figure 6). Addition of cholic acid to the luminal reservoir of S. aureus/epithelial cell cultures in the absence of C. scindens did not significantly alter toxin production compared to the cultures without cholic acid (p = 0.5) suggesting that cholic acid itself did not modulate toxin production (Figure 6). These results indicate that the production of SE by S. aureus was suppressed when C. scindens was present, but this suppression was not reliant on the creation of secondary bile acids. Similar findings have been observed in co-cultures of S. aureus with other bacterial species in vitro (Zdenkova et al., 2016; Nogueira Viçosa et al., 2019; Rodríguez-Sánchez et al., 2022). However, it is noteworthy that this phenomenon was observed for the first time in S. aureus-C. scindens co-culture in the presence of colonic cells.