The gastric mucus layer protects the epithelium from gastric acid and ingested pathogens. However, studies of human gastric mucus have been limited due to poor accessibility of native human mucus and the abundance of contaminants in these samples. Here, we explored the potential of human gastric organoids as models for mucus production. 3-D gastric organoids were analyzed by microscopy and particle tracking microrheology. To collect organoid-produced gastric mucus, termed bioengineered gastric mucus (BGM), organoids were cultured as monolayers at the air-liquid interface (ALI), and apically secreted mucus was harvested and analyzed by MUC5AC ELISA, proteomics, cryo-field emission scanning electron microscopy (CryoFE-SEM), bulk rheometry, and particle tracking microrheology.

Human Gastric Tissue and Mucus Samples

Human gastric tissue samples for the generation of organoid cultures were obtained with informed consent and IRB approval from patients undergoing endoscopy and biopsy at the Bozeman Health Deaconess Hospital (protocol 2023-48-FCR). Alternatively, de-identified whole stomachs collected post-mortem from transplant donors or surgical discard material from sleeve gastrectomies were provided by the National Disease Research Interchange (NDRI, RRID: SCR_000550; protocol DB062615-EX). Tissue samples from the NDRI were transported to Montana State University via overnight shipping in DMEM on ice and were used for both organoid establishment and mucus collection. Additional mucus samples were collected from stomach samples collected from sleeve gastrectomies performed at West Virginia University (IRB protocol 2406990595) after incubating the resected tissues in DMEM on ice overnight. To collect native mucus from the tissue samples, the mucus was gently scraped off the luminal tissue surface using a clean glass slide. To ensure that the least-contaminated and most representative portion of the native mucus sample was used, the clearest portion of the mucus was removed from the tube with forceps, leaving behind any contaminating blood or gastric juice. All mucus samples were frozen at -20 °C before use, which has been shown not to alter rheological properties. Sources for all mucus samples, i.e., organoid lines and native tissue samples, along with age, sex, and ethnicity of the donors, are listed in Table S1.

3D Organoid Culture

Human gastric organoid cultures were established and maintained as previously described. Briefly, gastric glands were isolated by collagenase digestion and seeded into Matrigel (Corning). Polymerized Matrigel was overlaid with organoid culture medium containing the supernatants of murine L cells that secrete the growth factors Wnt3a, noggin, and R-spondin 3 (L-WRN cells, ATCC, CRL-3276). The composition of the organoid culture medium is listed in Table S2. Organoid cultures were passaged weekly and were generally used up to passage 10. 

Particle tracking microrheology

For particle tracking microrheology in organoids, gastric organoids cultured on 35 mm MatTek glass bottom plates were microinjected with fluorescent microspheres. A 2 µL glass capillary was backfilled with sterile mineral oil and loaded onto a micromanipulator-controlled Nanoject (Drummond 3-000-204). The capillary was then filled with a 4.55 x 10⁶ particle/mL solution of 1 µm Fluoresbrite polystyrene microspheres (Polysciences; yellow/green: #17154, polychromatic red: #18660) in PBS. Negatively charged microspheres were used because they have a lower charge-mediated diffusion impairment in mucus and mucin gels compared to positively charged particles. 9.2 nL of the particle solution was injected into organoids that had a diameter of at least 300 µm (n=10). The organoids were then incubated at 37 °C, 5% CO₂ for 24 h prior to imaging to allow equilibration of the system. Organoids were imaged on a Leica SP5 confocal laser scanning microscope on a heated stage with an environmental control chamber (37 °C, 5% CO2; Life Imaging Services). A 20x (0.70 NA) objective and an additional 4x optical zoom were used for particle tracking, capturing an average of 15 ± 5 trackable particles in 30 s videos at 26 frames per s. For the long-term time-lapse imaging, ten organoids were imaged over 72 h at 5 min intervals. For particle tracking microrheology in native gastric mucus, PGM (15 mg/mL), or bioengineered, organoid-derived gastric mucus, the mucus sample and 1 µm Fluoresbrite polystyrene microspheres were pipetted into 0.12 mm spacers (SecureSeal™, Grace Bio-Labs, Bend, OR) that were mounted on glass slides and then sealed with glass coverslips. A Keyence BZ-X810 microscope with a 60x oil-immersion objective (1.4 NA) was used for imaging. For each sample, we recorded six 30 s videos at a frame rate of 29 fps and a frame size of 640 x 480 pixels at room temperature.

Videos were analyzed in MATLAB v7.9.0 using the “polyparticletracker” particle tracking routine, which finds the center of intensity for each particle using a polynomial Gaussian fit. Drift was corrected, if necessary, using a 3D Drift Correction plugin for ImageJ before analyzing the videos. Parameters within the particle tracking software were adjusted to track only single particles. Using the trajectory data for each particle, the time-dependent mean squared displacement (MSD) was calculated:

in which N = total number of particles to be averaged, xⁱ (0) = initial particle reference position, and xⁱ (t) = particle position at time t.

An exponential model was fit to each tracked MSD profile, and the power constant α was recorded:

in which α = exponential slope of the MSD profile. In the instance α ≈ 1.00, linear regression was conducted to determine the average slope of MSD(t) and subsequently used to calculate the diffusion coefficient using the generalized Stokes-Einstein equation (GSE). G^~(s) was calculated by taking the unilateral Laplace transform of MSD(t) for each particle using the GSE. Allowing s = i𝜔, we obtained frequency-dependent storage (elastic) (G’) and loss (viscous) (G’’) moduli.

 Monolayer Culture

Transwell permeable supports with PET membranes (surface area: 0.33 cm²; pore diameter: 0.4 µm; Corning, 3470) were coated with 15 µg/cm² rat tail collagen I (Corning, 354236) for 1 h at room temperature (23, 24). After aspirating the remaining collagen, the basolateral chambers were filled with 600 µL organoid culture medium and incubated for at least one h at 37°C, 5% CO₂. Human gastric organoids (cultured as described above) were expanded, harvested, and trypsinized for 10 min in a 37°C water bath (24). The resulting organoid fragments were mechanically dissociated as previously described, and pipetted through a 70 µm cell strainer to eliminate cell clusters (42). The filtrate was centrifuged at 400 x g for 5 mi,n and the resulting cell pellet was resuspended in organoid culture medium. A minimum of 2.× 10 ⁵⁰5^ cells were seeded onto each insert, agitated gently for 5 min, and incubated undisturbed at 37°C, 5% CO² for at least 24 h. Every other day, media was replenished,d and transepithelial electrical resistance (TEER) was measured using an EndOhm Chamber attachment to an EVOM 2 epithelial voltohmeter (World Precision Instruments). Once a transepithelial electrical resistance (TEER) of at least 200 Ω*cm² was reached, apical medium was removed to begin culture at the air-liquid interface (ALI)–a process termed “airlifting”. Cells were then left undisturbed until there was visible production of at least 50 µL apical mucus for collection. Our gastric epithelial cells typically form confluent monolayers within one week, and mucus production was observed within one week of airlifting.

Harvesting and Storage of Bioengineered Gastric Mucus

Bioengineered mucus (BGM) was harvested off the apical epithelium at either biweekly or weekly intervals using either forceps or a wide-bore pipette and transferred to a microcentrifuge tube. The wet weight of the mucus harvested from each well was recorded. Samples were then either used immediately for experiments, frozen at -20°C, or frozen at -80°C for long-term storage. For most experiments, BGM and native mucus (NM) samples were used fresh to preserve their natural state as much as possible, or were frozen at -20°C for transport and storage before analysis. Samples for which dry weight was measured were lyophilized, as described below. Porcine gastric mucin (PGM), prepared as previously described, was rehydrated at a concentration of 10-15 mg/mL for experiments.

 Proteomics

Proteomic analysis of BGM, M, and PGM was performed at the GlycoMIP core facility at Virginia Tech. Cysteine disulfide bonds were reduced using 4.5 mM dithiothreitol (DTT) and incubated at 37˚C for one h. Free sulfhydryl groups were alkylated with 10 mM iodoacetamide (IAA) at room temperature for 30 min in the dark, then unreacted IAA was quenched with 10 mM DTT. Protein was precipitated by the addition of o-phosphoric acid to 1.2% (v/v) and 1 mL methanol, followed by overnight incubation at -80˚C. Precipitated protein was loaded onto a micro S-Trap (Protifi) by centrifugation at room temperature for 1 min at 1,000 x g and washed extensively with methanol. Samples were digested using 2 µ. Pierce Trypsin Protease, MS grade (ThermoFisher Scientific) in 50 mM triethylammonium bicarbonate (pH 8.5) at 37˚C overnight. Peptides were recovered by sequential washings of the S-Trap with 25 µL solvent A (0.1% formic acid), then 25 µL 50:50 solvent A: solvent B (80% acetonitrile with 0.1% formic acid), and finally 25 µL solvent B. Excess acetonitrile was removed by vacuum centrifugation and peptide concentrations were determined using a nano-UV/Vis spectrometer (DeNovix) to measure the absorbance at 215 nm. The processed samples were analyzed by LC/MS on a Bruker MALDI-2 Mass Spectrometer (timsTOF Pro/flex) equipped with a Vanquish Neo UPLC unit. The gradient used LC/MS-grade water with 0.1% formic acid as solvent A and 80% acetonitrile with 0.1% formic acid as solvent B. The column used was a μPACTM HPLC column (ThermoFisher Scientific, #COL-NANO200G1B), and the flow rate was 8 μL/min. The run was 95 min long. The starting percentage of solvent B was 2%. The final percentage of solvent B was 98%. The mass spectrometer was operated in Parallel Accumulation Serial Fragmentation (PASEF) mode as described in the Supplementary Data 1.

The raw data were analyzed using Byonic Software (Protein Metrics). Proteins identified as “reverse sequences” and those flagged as “common contaminant proteins”, such as porcine albumin, were excluded from the data set post-processing. Protein isoforms were treated as separate entries. Proteins were categorized based on the human protein atlas (79) as (i) mucins; (ii) gastric proteins, i.e., secreted proteins from the GI tract or proteins known to have a specific function in the stomach; (iii) cellular proteins, i.e., cytoplasmic, nuclear, or membrane proteins with no specific known function in the stomach; (iv) contaminants, i.e., proteins derived from blood/serum or extracellular matrix; or (v) unknown proteins. The full datasets and the specifications for Byonic analyses have been deposited at PRIDE (PXD072060) via the ProteomeXchange.         

Cryo-Field Emission Scanning Electron Microscopy

Cryo-Field Emission Scanning Electron Microscopy (cryoFE-SEM) was performed at the Imaging and Chemical Analysis Laboratory (ICAL, RRID: SCR_026325) at Montana State University. NM from three tissue donors and freshly harvested BGM samples from three different organoid lines were analyzed. Approximately 50 µL of mucus was sandwiched between two gold-coated (Emitech K575X) silicon wafers, flash-frozen in liquid nitrogen, and manually fractured very quickly along scribed lines on the wafers using a flathead screwdriver before imaging on a Zeiss SUPRA 55VP Field Emission Scanning Electron Microscope. Images were acquired at -140°C with secondary electrons in high vacuum (~1^-6 mbar) at accelerating voltages of 0.8-1 kV, working distances of 4.8-6.7 mm, and with 10 µm and 30 µm apertures.

Image analysis was performed in FIJI using a Segment Anything Model (SAMJ; Fig. S1). Images were first converted to grayscale and automatically enhanced for contrast and brightness. Structures of interest (intact honeycombs or pores) were first manually localized using bounding boxes drawn with the rectangle tool, after which SAMJ automatically delineated boundaries. Identified ROIs were measured and exported as binary masks. To distinguish honeycomb cavities from pores, ROI measurements were gated based on FIJI circularity and solidity descriptors, applying thresholds derived from receiver operating characteristic (ROC) analysis of manually annotated ground-truth structures (Supp. Fig. 2A). In FIJI, a circularity is 4π(area/perimeter²) (with 1.0 indicating a perfect circle), while solidity is the ratio of area to convex hull area (with lower values indicating more concave or irregular boundaries).

 pH measurements and acid titration

  pH of the mucus samples was measured using pH paper (Hydrion, MicroEssential laboratory, #165/1-12), or a GIDIGI food digital electronic pH tester (JiNan, #B0FXWJVY7K). Buffer capacity of the mucus samples was determined by titration with 0.1 – 3 N HCl and was calculated as β = Δn / ΔpH, where Δn is the number of nmoles of HCl added per µL of mucus (equivalent to moles/L), and ΔpH is the measured change in pH. For particle tracking microrheology, polystyrene microspheres were added before adjusting the pH of the samples, and then samples were left to equilibrate for 30 min before imaging. The maximum amount of acid added to any sample undergoing particle tracking microrheology was 5.5% v/v of the final volume.

Consideration of biological variables, rigor and reproducibility, and statistical analysis

Tissue samples were obtained from donors of any sex or ethnicity and within an age range of 17-58 (Table S1). Each experiment was repeated three or more times with different organoid lines or native mucus samples. All data were analyzed using GraphPad Prism version 10.6.1 (San Diego, CA, USA). Data are presented as the mean ± SD. Student’st-testss and one-way ANOVA with Tukey’s or Dunnett’s multiple comparisons tests were used for normally distributed data. The Kolmogorov-Smirnov test or the Kruskal-Wallis test with Dunn’s multiple comparison test was used for data without Gaussian distribution to assess statistical significance. Data were tested for normal distribution using the Shapiro-Wilk test. P ≤ 0.05 was considered to indicate a statistically significant difference. The Robust Regression and Outlier Removal (ROUT) coefficient method with a Q value of 1% was used to remove outliers.