Crypt-Villus Scaffold Architecture for Bioengineering Functional Human Intestinal Epithelium

Abstract

Crypt-villus architecture in the small intestine is crucial for the structural integrity of the intestinal epithelium and maintenance of gut homeostasis. We utilized three-dimensional (3D) printing and inverse molding techniques to form three-dimensional (3D) spongy scaffold systems that resemble the intestinal crypt-villus microarchitecture. The scaffolds consist of silk fibroin protein with curved lumens with rows of protruding villi with invaginating crypts to generate the architecture. Intestinal cell (Caco-2, HT29-MTX) attachment and growth, as well as long-term culture support were demonstrated with cell polarization and tissue barrier properties compared to two-dimensional (2D) Transwell culture controls. Further, physiologically relevant oxygen gradients were generated in the 3D system. The various advantages of this system may be ascribed to the more physiologically relevant 3D environment, offering a system for the exploration of disease pathogenesis, host–microbiome interactions, and therapeutic discovery.

Graphical Abstract

1. INTRODUCTION

The small intestine is the major portion of the digestive tract responsible for the absorption of nutrients and drugs, as well as the primary target of numerous disease-causing toxins and pathogens that enter the body. As an organ that must provide absorptive and protective functions, along with withstanding mechanical abrasion, pH fluctuations, and colonization by billions of bacteria, the small intestine has evolved with a distinct crypt-villus structure, a critical characteristic of intestinal epithelial morphogenesis.¹ The small intestinal mucosa features thousands of finger-like protrusions, known as villi, which range from about 0.2 to 1 mm in length and 100 to 200 μm in diameter, varying from tip to base. Each villus is surrounded by small invaginations, known as crypts, which range from 100 to 200 μm in length and 50 to 150 μm in diameter.^2–5 Several epithelial cell types span the crypt-villus axis in vivo, including goblet cells, enteroendocrine cells, Paneth cells, and enterocytes. The most abundant intestinal epithelial cell lineage are the enterocytes, which are responsible for nutrient absorption. To achieve this function, fully differentiated enterocytes feature a brush border with two prominent morphological characteristics: uniform microvilli and highly ordered packing. The uniformity in dimensions and packing of the microvilli enables an increase in the inner surface area of the small intestine to a point that microvilli are estimated to amplify the surface area of the small intestine 9- to 16-fold.⁶ This increase in surface area improves nutrient absorption and digestive secretions and, consequently, any reduction in the surface area may lead to diseases such as inflammatory bowel disease, short bowel syndrome, and Hirschsprung disease.⁷

The distinct crypt/villi architecture of the intestine is also key to sustain intestinal homeostasis and renewal. The crypt is the basic architectural and functional unit of the highly proliferative stem cell population. The crypt-villus axis supports the rapid and continuous self-renewal, cell commitment, and cell differentiation of the intestinal epithelium such that proliferative stem cells residing within the crypts give rise to differentiated epithelial cells that migrate up the crypt-villus axis to eventually cover the villi projections.⁴ This continuous epithelial renewal allows for dying cells to shed from the tip of the villus while new, differentiated cells migrate upward and restock the diminished epithelial barrier under normal homeostatic conditions. Abnormal crypt-villus structures are linked to intestinal dysfunctions and disorders including celiac disease (CD), inflammatory bowel disease (IBD), and intestinal cancer.⁷

However, the detailed role of crypt-villus microenvironment in gastrointestinal disease development remains poorly understood, partially due to the lack of in vitro human intestinal epithelial models that can closely reproduce the human intestinal epithelium structure and phenotype. Animal models are commonly used for studies that pertain to causes and treatments for intestinal ailments. However, these animal models are not ideal because of cost and often lack correlation to human physiological responses, limiting extrapolations and predictions to patients.⁸ Biomimetic models of human tissues can provide a rapid and less expensive alternative platform to study human physiology. Therefore, the development of physiologically relevant human intestinal tissue structures as alternatives to animal models for drug development and/or disease modeling has gained interest.¹ Due to the importance of crypt-villus tissue architecture in the small intestine, an important goal is to generate crypt-villus architectures in such in vitro intestinal models.

One recent advance in the field that has led to major progress is the development of 3D organoid cultures. Organoids originate from primary source tissue and can self-organize into spheres with self-renewing crypts and proliferative cells that are able to differentiate into the major cell types within the intestinal epithelium.⁹ While organoids provide improvement with regard to physiological relevance and cell heterogeneity, they have limitations. In standard Matrigel droplet culture, organoids do not have an accessible lumen, making manipulation difficult and leading to necrosis of the core, thereby confounding utility. Further, they are not amenable to the inclusion of signaling gradients or mechanical forces, which play key roles in epithelial cell development and differentiation.¹⁰ To overcome some of these issues, 3D organoids have been digested into 2D monolayers, which can be seeded onto thin coatings of extracellular matrix (ECM) material in 96-well or Transwell formats.¹¹ Once digested and seeded into 2D monolayers, organoid-derived epithelial cells can be accessed easily from the luminal or basal sides, supporting studies involving drug absorption, nutrient absorption, microbe interactions, or the inclusion of relevant cell types such as stromal or immune cells. Epithelial monolayers therefore provide an easy and applicable model to investigate these intricate interactions. Despite their advantages, epithelial monolayers lack 3D architecture, dynamic mechanical forces, and biochemical gradients but are all critical within the native intestine.¹ It is therefore pertinent to use 3D systems to study the human intestine to ensure complete representation of cell–cell and cell–matrix interactions.¹²

In response to this need, recent studies have generated 3D culture systems to model the human intestine with 3D architectures or biochemical gradients.^1,13 Recent advances in biomaterials and 3D printing technology have enabled the generation of such systems, which incorporate features such as crypt/villi architectures. In a recent study, Caco-2 and HT29-MTX cells were cultivated together on poly(lactic-co-glycolic acid) scaffolds with a porous architecture and villi. The epithelial cells exhibited differentiation along the villi along with response to biochemical microenvironments of the intestine using both apical and basolateral feeding.¹⁴ Although this result was promising, full crypt-villus axes remain to be demonstrated, as only villus architecture, and exogenous basolateral signaling to simulate crypts was utilized. In a separate study, micro-molded cross-linked collagen and organoid-derived intestinal epithelial cells were combined to reproduce mature villus and crypt-like architecture.⁴ Another recent study used a cross-linked hydrogel and stereolithography to recreate the spatially organized crypt/villi architecture using Caco-2 cells.¹⁰ These studies demonstrated the importance of 3D topology in promoting cell polarization and differentiation. However, due to constraints of the fabrication techniques and the materials used, these systems were unable to reproduce the curvature of the native intestine or to incorporate other relevant cell types.

New microfabrication methods, such as 3D printing, have enabled the production of customized 3D biomimetic scaffolds with a high degree of complexity and precision to support multiple cell types.¹⁵ Despite these advances there has yet to be an in vitro tissue model incorporating multiple cell types, crypt/villi architecture, and curvature to mimic the native intestine. In this regard, the choice of biomaterial and fabrication method are critical for recreating a 3D environment capable of supporting intestinal epithelial cells while maintaining the ability to incorporate other relevant cell types, such as microbiota, stromal cells, and immune cells.

The present study combines silk fibroin protein and collagen type I to generate a system that is mechanically stable, cytocompatible, and versatile in morphology and structure to support intestinal features. Silk is a widely used natural polymer with tunable mechanical properties in addition to biocompatibility and ease of modification for a diversity of material formats.¹⁶ For the present study, silk solution was frozen, lyophilized, and cross-linkedinked into sponges that provide the porous structure for the integration of stromal or immune cells with mechanical stability. Additionally, 3D printing enables the generation of reverse molds, consisting of poly(dimethylsiloxane) (PDMS), which can be utilized to form the silk crypt and villi architecture within curved luminal surface. These sponges were then filled with collagen type I solution prior to cell seeding to provide support for cell attachment, proliferation, and differentiation. The combination of the silk and collagen proteins, the sponge structure, and 3D printing uniquely enabled a co-culture system, lumen curvature, and crypt/villi features of the small intestine, supporting results representative of the human small intestine functions.

2. MATERIALS AND METHODS

2.1. Fabrication of 3D Silk Scaffolds with Crypt and Villi Architecture.

Initial designs for the scaffolds were generated using computer-aided design (CAD) software Solidworks as described previously.¹⁷ Briefly, model designs were prepared in Solidworks and subsequently printed on a Formlabs Form 2 resin 3D printer (Somerville, MA). The printed resin molds were then used to generate reverse molds in PDMS (Sylgard 184 PDMS, Dow Corning, Midland, MI) to create silk scaffolds. The PDMS molds in the plates were cured in a 60 °C incubator for 2 h. The resin molds were then removed carefully to avoid damaging the luminal surface, and the reverse molds were washed to prepare for use with silk.¹⁷

Silk solution was prepared as previously described.¹⁸ To generate a thin film on the surface of the PDMS reverse molds, 100 μL of 6–7 wt % viscous silk fibroin solution was pipetted into the PDMS reverse molds to cover the surface of the mold. The molds coated with silk solution were placed under vacuum at −25 in Hg for 15 min and subsequently centrifuged at 800g for 5 min to permit efficient silk filling of the surface patterns. Excess silk solution was then removed to leave just a thin film on the surface of the reverse molds. The films were dried in a fume hood for 48 h. Once dried, the reverse molds with films were placed into a water annealer with a vacuum pump set to −25 kPa for 2 h at 60 °C and then dried overnight and subsequently filled with 6 wt % viscous silk solution for silk sponge formation. The molds filled with silk solution were frozen overnight at −20 °C and then moved into a lyophilizer for freeze-drying. After that, the resulting dried spongy silk scaffolds were autoclaved to induce the formation of β-sheets. Following this, the scaffolds were submerged and soaked in distilled water for 24 h to help with removal from the PDMS molds and trimmed into a rectangle (luminal size: 4 mm diameter, 8 mm length) with a patterned, indented lumen in the center. The fabrication process resulted in a half-scaffold with a patterned lumen (crypt-villus structure units) and a porous bulk space.

2.2. 2D Cell Source, Maintenance, and Co-Culture in 2D Transwell System.

Human Intestinal Myofibroblasts (InMyoFibs) (Lonza, Portsmouth, NH) were plated in the basal compartment of a 24-well Transwell culture system (pore size 0.4 μm; Costar Corp.) at a density of 1 × 10⁴ cells/cm². Caco-2 (ATCC; Rockville, MD) and HT29-MTX (Public Health England Culture Collection; Salisbury, Great Britain) intestinal epithelial cell lines were then plated at a ratio of 3:1, respectively, in the apical compartment of the Transwell system at a density of 5 × 10⁴ cells/cm². The seeded epithelial cells were incubated at 37 °C for 1 h to allow for attachment before being placed into the 24-well plates (the basal compartment of the Transwell system) containing the InMyoFibs to commence experiments. The Transwell co-culture was fed every 2–3 days with DMEM/SMGM medium (1:1, DMEM: DMEM supplemented with 10% FBS (fetal bovine serum) and 10 μg/mL human transferrin (Gibco, Gaithersburg, MD), SMGM: Smooth Muscle Cell Growth Medium (Lonza; Basel, Switzerland)) for 1-, 2-, 3-, and 4-week collection time points. For the entire study, Caco-2 and HT29-MTX cells from passage numbers 45–60 were used for seedings, and for InMyoFibs, passage numbers 4–7 were used. All cells tested negative for mycoplasma contamination.

2.3. Cell Seeding and Culture in 3D Silk Scaffolds.

Seeding strategies for half-scaffold constructs have been previously described.¹⁷ Briefly, the lumen with crypts and villi of the 3D half-scaffolds were seeded with Caco-2/HT29-MTX cells in a 3:1 ratio at a density of 1 × 10⁶ cells/mL, while the bulk space with interconnected pores was seeded with InMyoFibs in a collagen gel as previously described.^17,19 The seeded scaffolds were left in the incubator for 1 h to allow for cell attachment. During the incubation, about 250 μL of 1:1 DMEM/SMGM medium was added into the bottom of the well plates containing the scaffolds to keep them from drying out. Scaffolds were fed every other day with 1 mL of 1:1 DMEM/SMGM medium for weekly time point collections up to 8 weeks of culture.

2.4. Immunofluorescence and Confocal Imaging.

Immunofluorescent staining and imaging strategies for half-scaffold constructs have been previously reported.¹⁷ Briefly, seeded silk scaffolds with crypt and villi architecture were fixed and subsequently trimmed to expose the crypts and villi to the blocking buffers and antibodies. Following permeabilization and blocking, the scaffolds were incubated overnight (12–16 h) at 4 °C with anti-human ZO-1 conjugated with Alexa Fluor 594 Conjugate (Thermofisher, 1:100), or primary antibodies including anti-human-MUC-2 (Santa Cruz Biotech, 1:50) and anti-Sucrose isomaltase (abcam, 1:100). The following day, the primary antibodies were removed from the scaffolds. The scaffolds were then stained with Alexa Fluor 488 donkey anti-mouse secondary antibody at 1:250 (Invitrogen) and counterstained with dihydrochloride (DAPI, Invitrogen) for nuclei. The stained specimens were mounted glycerin, and the luminal surface of the scaffolds was observed and laser-scanned under a Leica SP8 confocal microscope with a filter set for DAPI (Ex/Em: 350/470 nm), Texas Red (Ex/Em: 540/605 nm) and GFP/FITC (Ex/Em: 488/514 nm). Leica confocal software (application suite X, 3.3.0.16799 and ImageJ) was used to create 3D rendered images.

2.5. Scanning Electron Microscopy (SEM).

Both seeded and unseeded silk scaffolds were cross-linked in 2.5% glutaraldehyde (GA) for 1 h followed by rinsing twice with PBS and deionized water at room temperature. After this, scaffolds were submerged and incubated in a graded series of ethanol (30, 50, 75, 95 and twice in 100%, molecular biology grade) for 1 h at each concentration followed by overnight incubation of 100% ethanol at −20 °C. The samples were eventually subjected to critical point drying using a liquid carbon dioxide dryer (AutoSamdri-815, Tousimis Research Corp.). Once dry, samples were sputter-coated with a 10 nm of Pt/Pd with a sputter coater (208HR, Cressington Scientific Instruments, Inc.) prior to SEM imaging (Zeiss UltraPlus SEM, Carl Zeiss SMT, Inc.).

2.6. Oxygen Measurements.

Oxygen measurement methods for half-scaffold constructs have been previously described. ¹⁷ Briefly, following the manufacturer’s manual, a Microx TX3 oxygen meter (PreSens Precision Sensing GmbH) equipped with a precise needle-type oxygen microsensor (NTH-PSt1-L5-TF-NS40/0.8-OIW) was used to record real-time oxygen concentrations of half-scaffolds flipped upward or downward in culture. Seeded scaffolds were cultured for 3 weeks prior to oxygen measurements for 48 h.

2.7. Quantitative RT-PCR.

Human intestinal epithelial cells were carefully detached from the scaffolds and Transwells using a mini cell scraper. A Qiagen Mini mRNA Extraction kit was used to isolate mRNA from the collected epithelial cells for traverse transcription. A High-Capacity cDNA Reverse Transcription Kit (Invitrogen) was utilized to deliver single-stranded cDNA from 2 μg of mRNA following the manufacturer’s instructions. Gene expression was conducted by qRT-PCR using 6 ng of cDNA from assorted samples for each well, using ready-to-use primer sets (SI, Villin, Muc-2, ZO-1, Occludin-1, and Claudin-4) purchased from realtimeprimers.com. Three independent biological replicates and three experimental replicates were performed for each gene tested with GAPDH mRNA as a housekeeping gene for normalization.

2.8. Statistical Analysis.

Data were presented as mean ± SEM (n = 3–5). A two-tailed t-test was used to compare the difference between the means of two groups, whereas analysis of variance (ANOVA) was adopted to analyze and compare means of multiple groups. p-values ≤ 0.05 were considered statistically significant.

3. RESULTS

3.1. Design and Production of 3D Scaffolds with Crypt and Villi Architecture.

Our group has previously demonstrated the utility of a 3D tubular silk protein sponge scaffold with interconnected pores that mimics the geometry of native intestine. These scaffolds possess a hollow lumen in the center to support human intestinal epithelial cells^18,20,21 and a bulk space capable of supporting other intestinal cells including subepithelial cells (InMyoFibs),^18,21,22 neural stem cells,²³ immune cells²² and smooth muscle cells.²⁴ Here, the goal was to generate a more user-friendly version of these tubular scaffolds (half-scaffolds) and to incorporate crypt and villi architectures. To achieve these goals, 3D printing was used to generate resin molds with crypts and villi that were designed by CAD software Solidworks (Figure 1A). To mimic the physiology of the native intestine, the crypts and villi were 0.8 mm in length, within range of the villi of the native intestine, 0.5 mm in width at the base, and 0.4 mm at the top or bottom of the villi and crypts, respectively (Figure 1B). The spacing between crypts and villi was 0.79 mm while spacing between each villus or crypt within a row was 1 mm (Figure 1C). The silk solution in the reverse molds was freeze-dried to obtain silk sponges with interconnected pores in the bulk and a smooth luminal surface with crypts and villi (luminal size: 4 mm diameter, 8 mm length). Macroscopy and SEM imaging showed organized rows of crypts and villi within the PDMS reverse molds (Figure 1D) and matching organized rows of crypts and villi with a smooth luminal surface within the silk scaffolds (Figures 1E and S1A,B). SEM imaging showed smooth membranes with 2–5 μm diameter pores across the scaffold luminal surface (Figure S1C); while SEM imaging of the bottom view of the scaffolds revealed a porous sponge in the bulk space (Figure S1D,E) with pore sizes ranging from 100 to 300 μm (Figure S1F).

Figure 1.

3D half-scaffold system with crypts and villi for intestinal tissue engineering was fabricated using 3D printing techniques and silk fibroin. (A) Schematics of the fabrication process for generating silk-based porous half-scaffolds with crypts and villi for 3D human intestine engineering. Silk-based scaffolds are constructed by (1) designing a 3D model using 3D CAD software, (2 and 3) 3D printing resin molds, (4) casting PDMS reverse molds by placing the resin molds into liquid PDMS, (5) curing and obtaining the PDMS molds with heat exposure, (6) pipetting 7% aqueous silk solution into PDMS molds, (7) vacuuming and centrifuging silk filled molds then removing excess silk to leave a thin silk film on PDMS mold, (8) drying silk film then water annealing to achieve a smooth surface prior to filling with 6% silk solution, and (9) freezing, lyophilizing, and inducing β-sheets via autoclave. (B) Transverse cross section of half-scaffold constructs showing dimensions of crypt/villi height and diameter at both the base and the tip. (C) Longitudinal cross section of half-scaffold constructs showing distances between villi and crypts (scale bar: 1 mm). (D) Macroscopic top view of PDMS mold for casting silk scaffolds. (E) Macroscopic top view of silk scaffolds with crypts and villi (scale bar: 1 mm).

One of the main challenges in using tubular scaffolds was the inability to detect potential defects in the lumen prior to the addition of epithelial cells. Smooth and defect-free lumens are required to ensure proper cell compartmentalization, thus the goal was to generate scaffolds without defects that could be easily detected prior to adding the cells. The open architecture of this half-scaffold system enabled this quality control check with the use of a magnifying glass.

Another challenge presented by the incorporation of crypt and villi architectures was to ensure that the silk solution completely filled the small crypt invaginations and covered all villi protrusions. To address this challenge, vacuum and centrifugation were utilized following initial filling of the resin molds with silk solution. The negative pressure created by the vacuum helped to create a tightly packed silk solution that could hug the villi, while centrifugation helped to ensure the solution filled the crypts. Excess silk could then be carefully removed, leaving a film to cover the resin mold. A further goal was to achieve smooth luminal surfaces without defects; thus, we placed the resin molds with fully dried films into a water annealer to facilitate the formation of β-sheet crystals for the formation of flexible and water-insoluble films.²⁵

3.2. 3D Crypt and Villi Architecture for the Co-Culture of Intestinal Epithelial Cells and Myofibroblasts.

The first goal was to investigate the impacts of 3D architecture and curvature on epithelial cell attachment and proliferation. The 3D silk half-scaffold system with crypt and villi architectures was used in an in vitro tissue system with Caco2 and HT29-MTX in the lumen and myofibroblasts in the bulk space (Figure 2A–C). Within the 3D silk scaffolds, InMyoFibs were incorporated into the bulk space as a means of providing support for epithelial cells. To initiate the co-culture, a 3:1 ratio of Caco2:HT29-MTX cells was seeded into the lumen of the scaffolds (Figure 2C). The 3D tissues were then maintained in a 1:1 ratio of Caco2/HT29-MTX medium/ InMyoFib medium for up to 8 weeks. For comparison, Transwell inserts were cultured simultaneously with Caco2 and HT29-MTX cells seeded in the apical chamber and myofibroblasts seeded in the basal chamber (Figure 2B). The transwells were maintained with Caco2/HT29-MTX medium in the apical chamber and InMyoFib medium in the basal chamber for up to 4 weeks, as cells in the transwell system began to lose function and die off after 3 weeks. Caco2 cells were used for this study to mimic enterocytes of the human intestine while HT29-MTX cells mimicked Goblet cells. These cells were used at a ratio of 3:1 (Caco2/HT29-MTX) to most mimic the ratio of enterocytes and goblet cells, respectively, found in the human gut and based on previous success using this ratio.^19,26,27 InMyoFibs were used as a supporting cell in the bulk space based on previous work from our group and others, showing that subepithelial myofibroblasts are capable of fostering epithelial growth and barrier function.^18,28 While previous studies had difficulties incorporating multiple cell types into systems with crypt and villi architectures due to the lack of a subepithelial compartment, our unique design with a 3D silk sponge supported the facile incorporation of multiple cell types in the subepithelial compartment.^3,4 While the present study focused on myofibroblasts to promote epithelial differentiation, we have previously demonstrated the ability to incorporate monocyte-derived macrophages, neuronal cells, and human microbiome.^{18,20,22,23,29,30}

Figure 2.

Caco2 and HT-29 cell lines formed confluent and functional monolayer in 2D Transwells and 3D silk half-scaffolds. (A–C) Illustration of cell culture with Caco2 and HT-29 cell lines and primary myofibroblasts (A), cell seeding strategies for Transwells (B), and 3D half-scaffolds with crypts and villi (C). (D) Top-view photos of silk half-scaffolds seeded with Caco2/HT29 and myofibroblasts prior to fixing and staining (scale bars: 4 mm). (E) Representative confocal z-stack of DAPI staining (blue) on luminal surfaces of the scaffolds showing crypts and villi well covered with a cell monolayer, outlined by white dashed lines (scale bar: 200 μm).

To ensure epithelial cell attachment and proliferation, all scaffolds were inspected for smooth lumens without cracks or larger pores. Once inspected, the scaffolds could be used for seeding myofibroblasts in the bulk and Caco2/HT29-MTX cells in the lumen (Figure 2C). Seeded scaffolds with smooth lumens and intact crypts and villi could be visually observed for improved quality control. The seeded scaffolds were then fixed 1 week post seeding and stained with DAPI to visualize epithelial cell coverage. Crypts and villi were all well covered with an epithelial cell monolayer, indicating successful cell attachment and proliferation on the patterned lumens (Figure 2E).

To further confirm epithelial cell growth in the scaffolds, samples were fixed at week 3 post seeding and processed for SEM. SEM images confirmed that epithelial cells were successfully seeded in scaffolds with the microarchitectures and successfully proliferated to cover all crypt-villus surfaces (Figure 3A). The differentiated monolayers showed a microvilli brush border both on the villi (Figure 3B–D) and within the crypts (Figure 3E–G). This polarized epithelium is typical of mature enterocytes with absorptive functions,⁶ indicating that the 3D half-scaffolds with crypt/villi features provided a suitable 3D niche for intestinal epithelial cells to attach and differentiate. Although it is not characteristic of the native intestine to have enterocytes within the crypts, this preliminary study was meant to show the capability of the crypts and villi to support cell adhesion, proliferation, and differentiation and future studies can focus on ensuring stem cells inhabit the crypt areas.

Figure 3.

SEM images on the 3D scaffolds seeded with cells (week 3 post cell seeding) revealed a full coverage of epithelial cells on the scaffold luminal surface and the apical surface of the epithelial cells were covered with dense microvilli. (A) SEM images of the luminal surface of the seeded 3D scaffold with crypt/villus pattern (scale bar: 500 nm). (B–D) SEM images with higher magnifications of the villi in the seeded 3D scaffold lumen. (B) Scale bar: 100 μm, (C) scale bar: 20 μm, (D) scale bar: 2 μm. (E–G) SEM images with higher magnifications of the crypt in the seeded scaffold lumen. (E) Scale bar: 200 μm, (F) scale bar: 20 μm, (G) scale bar: 10 μm.

3.3. 3D Crypt and Villi Architecture Supported Differentiation of Intestinal Epithelial Cells.

Caco-2 cells have been shown to spontaneously differentiate over time. If cultured in DMEM for roughly 3 weeks, Caco-2 cells develop characteristic intestinal enterocyte properties including the formation of tight junctions and cellular polarization.^12,31,32 HT29-MTX cells, on the other hand, will differentiate into a secretory lineage of cells similar to intestinal Goblet cells.³³ Intestinal myofibroblasts have been previously identified as a significant source of differentiation and growth, stem cell, and chemotactic factors when placed in the subepithelial space of in vitro intestinal models.³⁴ We therefore sought to investigate the ability of our 3D scaffold with crypts and villi to support the growth and differentiation of Caco-2 and HT29-MTX cells in the luminal space along with InMyoFibs in the bulk space. To this end, samples were harvested and processed for immunostaining after 3 weeks in culture. Immunofluorescence of these tissues showed that epithelial cells could differentiate into heterogeneous populations of cells expressing differentiation markers typical of human intestinal epithelium,³⁵ including Zonula occludens (ZO)-1 (tight junctions, Figure 4A,B), mucin 2 (Muc-2) (Goblet cells, Figure 4C,D), and sucrase isomaltase (SI) (enterocytes, Figure 4E,F). Confocal z-stack reconstructions showed confluent monolayers of differentiated intestinal epithelia in the crypts and villi within the lumens, further confirming that the crypt and villi architecture supported cell adhesion, proliferation, and differentiation.

Figure 4.

Intestinal epithelial cells (Caco2 and HT29-MTX) formed a confluent and functional monolayer in 3D silk half-scaffolds. Schematic of Caco2/HT29-MTX monolayer and primary myofibroblasts. (A–F) Representative confocal z-stack of the immunostained luminal surface of the scaffold (scale bar: 200 μm). Confocal immunofluorescence images of the epithelia for ZO-1 (A, B), Muc-2 (C, D), and SI (E, F). Scale bars: 20 μm. (A, C, E) 3D reconstruction of z-stack of fluorescence microscope images of the scaffolds. (B, D, F) Confocal maximum projection fluorescence microscope images of the scaffolds with higher magnification on the luminal surface.

3.4. 3D Half-Scaffolds with Crypts and Villi Increased Culture Time of Epithelial Cells and Promoted Differentiation Compared to 2D Transwells.

Growing evidence has indicated that physical cues of cellular environment, including substrate geometry, impact various cell behaviors such as adhesion, proliferation, differentiation, migration, and permeability.³⁶ For comparison of physiological relevance to the native small intestine, the 3D scaffolds with crypt/villi were compared with 2D Transwells both of which are seeded with a mixture of Caco2 and HT29-MTX and human intestinal myofibroblasts. In this context, epithelial cell differentiation and function within the crypt- and villi-laden scaffolds were characterized in comparison to 2D Tranwells. To quantify gene expression of various markers of intestinal differentiation in both 2D and 3D cultures over time (Figure 5A–F), mRNA of tissue samples from different time points was isolated and processed for RT-PCR. Consistent with previous studies using Caco2 and HT29-MTX to represent the intestinal epithelium, gene expression for all markers was relatively low at week 1 and increased in both systems through week 3.¹² After week 3, gene expression of all markers evaluated decreased in the Transwell systems, but either increased or remained stable in the 3D scaffold system through week 4. No data was collected for Transwells for weeks 5 through 8; however, by week 8, scaffolds held equal or higher gene expression levels than their respective levels at week 4 in the Transwells. Taken together, these data indicate that the 3D architecture of the scaffolds increased stable cell culture time and promoted increased differentiation of intestinal epithelial cell lines. These data align with previous studies which have demonstrated sustained cell culture in 3D structures compared to Transwells.¹⁴ In previous studies, it was found that monolayers on Transwells may begin to be disrupted after around 4 weeks in culture due to cell overgrowth and multilayer formation,¹⁴ a finding observed in the present study. It is hypothesized that in the 3D environment, which better replicates the in vivo environment, sustained gene expression is supported due to the presence of physiological and topological cues that are different than in 2D, such as tight junction formation (barrier integrity), mucus secretion, and digestive enzyme production.³⁷

Figure 5.

3D architecture and multicellular co-culture system significantly improved the overall differentiation and function of epithelial cells. (A–F) Gene expression level of intestinal epithelial biomarkers, including SI (A), Villin (B), and Muc-2 (C), and cell–cell junction-related genes, including ZO-1 (D), Occludin-1 (E), and Claudin-4 (F), evaluated by quantitative reverse transcription-polymerase chain reaction (qRT-PCR). Data are presented as mean ± SEM (n = 3, ***p < 0.001, *p < 0.05).

As previously seen using SEM, a polarized epithelial layer with a microvilli brush border was formed, verified by the gene expression of Sucrase Isomaltase (SI) (Figure 5A). Villin, an actin-binding protein localized to the intestinal epithelial brush border, was also highly expressed in the crypt-villi scaffolds (Figure 5B) as well as Muc-2 (Figure 5C), an indicator of Goblet cell function. At week 4, Muc-2, Villin, and SI were all expressed at significantly higher levels in the scaffolds than in the Transwells, indicating that such biomaterial scaffolds enhance epithelial cell differentiation and function (Figure 5A–C).

Tight junctions (TJs) significantly contribute to the function of intestinal barrier and the regulation of epithelial paracellular transport.³⁸ Claudin, occludin, and zonula occluden protein families are the major players with respect to TJ-associated proteins.³⁹ To evaluate whether the 3D scaffolds with crypts and villi impacted epithelial barrier function compared to 2D Transwells, various tight junction proteins, including ZO-1, Occludin-1, and Claudin-4, were analyzed (Figure 5D–F) for gene expression level. Following 3 weeks of culture and, therefore, spontaneous differentiation of the epithelial cells, significantly higher ZO-1 gene expression (week 4, p < 0.001, n = 3), Occludin-1 (week 4, p < 0.001, n = 3), and Claudin-4 (week 4, p < 0.001, n = 3) was observed in the 3D scaffolds (Figure 5D–F). This trend indicated that scaffold architecture improved tight junction formation in intestinal epithelial cells compared to 2D platforms such as Transwells.

3.5. 3D Half-Scaffolds with Crypts and Villi Were Capable of Emulating Physiological Oxygen Gradients in the Human Intestine.

There are two major oxygen gradients existing in the healthy human intestinal tract: one running from the stomach to the distal colon in a proximal-to-distal manner and one running from the intestinal submucosal wall to the mucus layer in radial manner.⁴⁰ Here, the 3D scaffolds with crypt and villi architecture were manipulated to attain similar oxygen gradients by placing the scaffolds facing upward such that the lumen was exposed to the environment (Figure 6A) or facing down (Figure 6B) such that the lumen was facing to the bottom of the culture wells filled with culture media. The oxygen measurements were conducted 3 weeks post cell seeding using an oxygen probe, as confluent intestinal monolayers are established in the scaffolds at this point. Scaffolds in cultures with the lumen facing up demonstrated microaerobic conditions (pO₂ ~5 to 7%) that stabilized approximately half an hour after culture initiation (Figure 6A). When scaffold lumens were placed facing down, microaerobic conditions (pO₂ ~1 to 2%) were attained (Figure 6B). These spontaneous microaerobic conditions were maintained for up to 48 h in both upward-facing and downward-facing culture conditions (Figure 6A,B). We did not continue oxygen measurements past this point due to the inherent difficulty in changing culture media with the probe in place; however, as the oxygen levels were stable from 2 to 48 h, we predict they would continue to be stable for the duration of cell culture.

Figure 6.

3D bioengineered intestinal tissues generated low luminal oxygen levels. (A) Scaffolds cultured in an upward position generated microaerobic conditions in the lumen (pO₂ between 5 and 7%), stable for up to 48 h. (B) Scaffolds cultured in a downward position generated anaerobic conditions (pO₂ ~1 to 2%) stable for up to 48 h.

4. DISCUSSION

4.1. Silk as a Biomaterial for Tissue Engineering.

The selection of suitable biomaterials for scaffold fabrication is a key to success in tissue engineering.¹⁰ In the present study, silk protein was chosen as the scaffolding material due to its biocompatibility and mechanical properties.¹⁶ The silk has a propensity to self-assemble into hydrophobic β-sheet structures, resulting in strong and resilient materials that are cell-friendly.^16,17 We have previously established scaffold designs using silk protein to generate sponges with 3D compartmentalized structures that mimic the biologically relevant geometry of the native intestine. The first design featured a tubular design with a central hollow lumen allowed intestinal epithelial cells to grow in vitro in a 3D architecture mimicking that the native human intestine^18,20,21 and further supported other intestinal cells including subepithelial cells (InMyoFibs),^18,21,22 neural stem cells,²³ immune cells,²² and smooth muscle cells.²⁴ The second design featured a half-scaffold with an open lumen, analogous to a longitudinal section of the intestine, but walled off on either side to provide a half-cylinder-shaped lumen.⁴¹ This design still supports the culture of epithelial cells in the lumen and other supporting cells in the bulk space compared to 2D transwells, while also enabling control of the luminal surface architecture and direct imaging of this region. In the present study, we took advantage of this ability to control the luminal surface architecture using 3D printing technology and mold casting to incorporate a crypt and villi architectures into the lumen of these half-scaffolds to reproduce more complex 3D intestinal models (Figure 1A).

4.2. 3D Printing Technology to Control Scaffold Surface Pattern and Curvature.

Many recent studies in tissue engineering have employed 3D printing to achieve similar surface architectures. One study used bioprinting to create a 3D model of the intestine consisting of two printed cell layers: human intestinal and stromal fibroblast cell layers.⁴² Many key features of the native tissue were displayed in this model, including the expression of differentiation markers, cell polarization, and enhanced CYP450 activity. However, 3D crypt-villus architecture was not incorporated into this model. Other recent studies have gone a step further, using 3D printing to incorporate multiple cell types and a crypt/villi architecture. More recently, using collagen-based bioinks, two research groups have developed crypt and villi 3D geometry.^3,43 One group only used one bioink composed of Caco-2 cells to create an epithelial layer,³ while the other went a step further using two bioinks: one composed of Caco-2 cells for an epithelial layer and one composed of Human Umbilical Vein Endothelial cells (HUVECs) to create capillary-like structures underlying the Caco-2 epithelium.⁴³ In both cases, it was determined that crypt/villi architecture contributed to increased homogeneity, higher proliferation rates and improved expression of differentiation markers compared to systems lacking crypt-villus architectures. While these studies improved upon previous models of the human intestine, they did not incorporate luminal curvature as present in the native intestine. In our system, the length of the villi within our scaffold systems was within the range of the native intestine, while the other dimensions were similar but not the same as those in the native intestine. A higher-resolution 3D printer would allow higher resolution of the morphological features as supported by the Solidworks design. With higher-resolution printing, the present design could be tuned to more precisely mimic the architecture of different regions of the intestine or to mimic disease-specific topographies.^7,44 Additionally, the model incorporates crypt/villi architectures in conjunction with a curved luminal surface, thus increasing physiological relevance.

Luminal curvature is critical to the physical rearrangement, motility, and function of many cell types, including stromal cells and corneal, kidney, lung, and intestinal epithelial cells.^45–48 Further, mesoscale substrate curvature overrules nanoscale contact guidance in directing cell behavior⁴⁸ To mimic the varying degrees and types of natural curvature present in vivo, many strategies have been employed. Observed effects of micron-scale curvature utilized in these methods include the notion that there is a dynamic relationship between the cytoskeleton and the nucleus within the cell with the nucleus working as a mechanical sensor that causes cell migration within concave rather than convex curvatures.^49,50 Though migration is favored in concave areas, it appears that attachment and proliferation are preferred in convex regions.⁵¹ This effect may be cell-line-dependent, as it appears to be more pronounced in certain cell types, such as mesenchymal stem cells or fibroblasts, as opposed to macrophages or epithelial cells, which tend to exhibit more complex patterns with regard to substrate topography.⁴⁹ Further study of this phenomenon demonstrated that apical stress fibers tended to align in the radial direction and form chords across concave gaps to avoid bending, whereas basal stress fibers tend to bend in the convex direction.⁵⁰ This differential bending of the apical and basal stress fibers allows for improved apical–basal polarization on curved substrates as opposed to flat ones.⁴⁵ Within intestinal epithelial cells, substrate curvature promoted apical–basal polarization to support ZO-1 protein expression and thus enhanced the intestinal barrier function.^45,47 This is consistent with our findings where the 3D curved scaffolds supported higher expression of ZO-1 compared to 2D Transwell systems over an 8-week period (Figure 5D). Further, Caco-2/HT29-MTX cells seeded on our scaffolds showed increased expression of various intestinal epithelial markers over an 8- week period compared to those seeded in a Transwell system (Figure 5). For all markers tested, expression increased up to week 3 within the Transwell system but then decreased at week 4. Within the scaffolds with crypts and villi, expression of all markers was still increasing at week 4 and was then stably maintained through week 8 with only a slight decrease in expression over the 8-week period. This supports the idea that 3D topology strongly influences cell behavior and differentiation and highlights the importance of recreating the intricate architecture of the native intestine in vitro. ¹⁰

While similar studies have found curvature to increase the expression of ZO-1 and other tight junction (TJ) proteins, there are discrepancies between these studies on whether crypt and villi architecture increases or decreases tight junction formation. Multiple studies have noted that TJ expression decreased from the tip of the villi to the base of the crypts.^14,47 Based on this finding, it would follow that incorporating both crypts and villi would result in decreased TJ expression compared to 2D cultures. However, both these and other studies have noted that curvature influences cell polarization, which has a direct influence on ZO-1 expression and tight junction formation. As ZO-1 localizes to the apical side of epithelial cells, correct polarization is necessary to form a continuous, functional barrier.^14,45,47 In the context of our findings, it would appear that the increase in ZO-1 and TJ formation as a result of incorporating a curved surface eclipses the possible decrease resulting from incorporating the crypt and villi architecture.

4.3. Cell Source of Intestinal Epithelial Cells for Intestine Tissue Engineering.

In modeling the human intestine, it is common to use immortalized cell lines such as Caco-2, HT-29, or T84 as they are cheap and easy to use.^26,52–54 Co-culturing Caco-2 cells to mimic enterocytes with HT-29 cells to mimic goclet cells has been shown to better mimic small intestine nutrient absorption and mucus secretion compared to the monoculture of either of these lines.²⁶ Recent studies also use intestinal organoids, which are derived directly from human patient intestinal biopsies, for in vitro assessment of intestinal function. While organoids in Matrigel droplet culture provide a 3D platform and provide improvements regarding physiological relevance and cell heterogeneity, they have limitations in terms of cost associated with maintenance.⁵⁵ Additionally, while organoids provide heterogeneity closer to that seen in vivo, this heterogeneity can make it difficult to compare results among experiments and labs. Factors such as Matrigel batch, media composition, and patient variability affect organoid growth and development and lead to confounding results.^56,57 Immortalized cell-line models can be grown rapidly, inexpensively, and reduce variability such that other factors like proliferation, differentiation, and polarization, can be examined for optimization of culture conditions.^58,59 As a preliminary study, we aimed to have low variability and low cost and therefore opted to use a combination of Caco2 and HT29-MTX to represent the epithelial layer of cells.

While our 3D scaffolds supported Caco2/HT29-MTX proliferation and differentiation, we recognize that the use of cancer cell lines limits the ability of the crypt and villi structures to fully mimic the native system. In vivo, stem cells inhibit the crypts of the intestinal epithelium and differentiate as they move upward into the villi. As Caco2 and HT29-MTX are transformed cell lines, they do not possess the ability to form proliferative crypts or differentiate into all major cell types of the intestinal epithelium. Additionally, despite displaying properties of enterocytes and goblet cells, respectively, they do not represent the native human intestine with respect to villi formation, marker gene expression, and drug permeability.^60,61,41 Therefore, while this study provides a foundation for the use of these 3D scaffolds with crypts and villi for future studies with intestinal organoids, it cannot yet be determined whether the crypts and villi in our 3D scaffolds provide a suitable niche for the biochemical gradients found in the native intestine that support a self-renewing epithelium.

Despite the limitations posed by the use of transformed cell lines, the present study, along with our recent study demonstrating intestinal organoids in co-culture with a human-derived microbiome within nonpatterned half-scaffolds⁶² provides a foundation for future studies. The goal of such studies will be to incorporate intestinal organoids into these half-scaffolds with crypts and villi to recreate the stem cell niche found in the crypts of the native intestine. As a biochemical gradient allows for continuous epithelial renewal, recreating this feature of the intestine may increase the culture time and relevance of our scaffold designs for self-renewing intestinal tissue to support the study of diseases that arise from disruptions to the natural biochemical gradient, such as cancer.⁶³ Furthermore, the present study was limited to the use of Transwells as controls, while future studies could explore in more depth the implications of scaffold architectures with a range of curvatures and without curvature and the 3D crypt and villi architectures to enable a more complete understanding of the correlation between morphology and biological functions.

4.4. Extension of Functional Culture Time.

In addition to including crypt and villi architectures in our scaffold design, a 3D design was used such that stromal cells could be incorporated into the bulk region of the scaffolds, mimicking the subepithelial layer found in vivo. 3D scaffolds present a more realistic environment to that found in vivo compared to 2D platforms.^10,18 The incorporation of a 3D scaffold design may, therefore, further explain the increased expression of all genetic markers and sustained expression of these markers over an 8-week period (Figure 5). In contrast, all genetic markers were expressed at lower levels and began to decrease in week 3 within Transwell cultures. It should also be noted that peak expression of most genes in the 3D scaffolds did not occur until week 5. This may be due in part to the fact that proliferation in the scaffolds likely took longer than in the Transwells due to the increased surface area and the requirement of the cells to migrate along the curved lumen and up the villi or down the crypts. As the present study focused on comparisons of 2D Transwells to 3D scaffolds with curvature and crypt/villi architecture, it remains in question which feature of the scaffolds (3D design, curvature, crypt/villi architecture) had the most significant impact on cell proliferation, differentiation, and culture time. For this reason, future studies may include more direct comparisons of scaffolds with various architectures (flat vs curved lumen, crypts/villi vs no substrate architecture, 3D vs Transwells).

The incorporation of stromal cells or signals mimicking those secreted by stromal cells can further contribute to sustained culture.^14,18,28 In vivo, the intestine is composed of many layers, the innermost layer being the mucosa which encompasses both epithelial cells and supporting stromal cells. Within the mucosa, stem cells within the crypts rely on signals from adjacent stromal cells to undergo proper proliferation and differentiation up the crypts.⁵⁷ For this reason, it has been increasingly common to incorporate stromal cell secretome signals in culture with epithelial cells, such as Caco2 or HT29-MTX.^14,57 While these stromal cell signals, such as Noggin, R-spondin, Wnt, and epidermal growth factor (EGF), contribute significantly to stem cell maintenance and function, there have also been various reports that direct co-culture with stromal cells provides additional support to the intestinal epithelium via secretion of extracellular matrix (ECM) proteins.^14,64 Direct co-culture of fibroblasts with Caco2 cells improved various morphological characteristics, including lateral membrane morphology and columnar shape polarization such that the cells more closely resembled in vivo enterocytes.⁵⁷ As some benefits of direct co-culture, may be lost in Transwell models where fibroblasts are cultured on the bottom of the basal compartment, it follows that a 3D culture system with direct contact of stromal and epithelial cells would present an avenue for increased culture time. Our 3D silk scaffold system incorporates myofibroblasts in a bulk compartment directly in contact with epithelial cells and a special organization mimicking that of the native intestine. This feature may also help to explain the increased culture time of Caco2 and HT29-MTX cells in our scaffolds compared to a 2D Transwell system (Figure 5).

4.5. Oxygen Profiles.

The native gut includes proximal-to-distal and radial oxygen gradients.⁴⁰ The mucus on the apical side of the in vivo lumen has oxygen concentrations of 0.1–1% (ca. 1–6 mmHg), capable of supporting obligate anaerobic microbes; however, moving across the intestinal wall to the vascularized submucosa, the oxygen concentration increases to about 6% (~42 mmHg) to achieve tissue oxygenation. Finally, the muscle wall of the intestine has the highest oxygen concentrations, ~7 to 10% (42–71 mmHg) oxygen.⁶⁵ This feature is crucial to maintaining a healthy intestinal epithelium while also allowing gut microbes to thrive. The gastrointestinal tract supports a large array of microorganisms, having the highest abundance and diversity within the human body. The oxygen gradients of the intestine allow for support of both aerobic and anaerobic populations of commensal organisms. This diversity of microbes plays a large role in normal immune response, nutrient absorption, pathogen protection, and drug metabolism. Changes in oxygen tension can tip the balance of microbial communities in the intestine to favor pathogenic bacteria which can lead to the development of various diseases both inside and outside the gastrointestinal tract, including inflammatory bowel disease, obesity, diabetes, cancer, and even neurological disorders such as autism and depression. It is therefore critical to maintain balance within the intestinal microbial populations for retention of gut homeostasis and human health.^66,67

For this reason, intestinal models capable of producing oxygen tensions which can support the wide range of microbes in the human gut along with human epithelial cells are key. Current in vitro models, such as those carried out in 2D culture or with Transwells, have been utilized to investigate host–microbiome interactions, but these studies cannot be carried out longer than a few hours before cell injury and death occur as a result of bacterial overgrowth.^68,69 Further, to generate Transwell systems with sustained low oxygen tension to support obligate anaerobes, it is necessary to use anaerobic chambers and to perfuse oxygenated basolateral media, both of which increase both complexity and cost of such systems.^70,71 Organoid cultures provide new opportunities for studying host–microbiome interactions; however, they also cannot be co-cultured with bacteria for extended periods of time and cannot sustain low luminal oxygen levels, which is necessary to co-culture certain obligate anaerobes.⁷² The mucosal-simulator of the human intestinal microbial ecosystem (M-SHIME), is a specialized bioreactor model which was developed with anaerobic conditions to support the growth of gut microorganisms in vitro; however, no living human cell is included in this system.⁷³

Some newer systems have made strides in incorporating both microbes and human intestinal cells for longer time frames. One recent study employed an organotypic slice model of human intestinal tissue taken from a colon biopsy to exploit the impact of oxygenation on submucosal function and to investigate host–microbe interactions following treatment with antibiotics. While this study made strides in increasing physiological relevance by incorporating immune cell types and sustaining bacterial populations through 72 h, the method of direct ex vivo tissue culture makes reproducibility and scale-up difficult.⁶⁵ Other studies have focused on generating oxygen gradients in gut-on-a-chip models which can maintain anaerobic microbial communities alongside a human epithelium for 72 h. However, these culture systems require flow to maintain these cultures, increasing complexity, and do not include 3D designs or crypt and villi architectures.^74,75 As a major component of the in vivo radial oxygen gradient is the crypt and villi architecture, this limits the physiological relevance of the studies.⁷⁶ Some recent studies have gone a step further, incorporating either crypt or villi features; however, neither of these systems incorporated both features and the systems remain 2.5D designs rather than a complete 3D design and were without the curvature.^61,66,71

The system presented in the current study enables replication of the in vivo oxygen gradients along both the radial and proximal to distal axes simply by placing the scaffolds with the luminal space facing up (Figure 6A) or down (Figure 6B) in culture. Positioning the scaffolds facing up resulted in pO₂ between 5 and 7% within the lumen while positioning the scaffolds facing down resulted in pO₂ between 1 and 2% in the lumen. These oxygen profiles stabilized after about half an hour and remained for up to 48 h. As our tissue model provides a simple and user-friendly approach to control oxygen profiles it can be used to study links between intestinal disease and disruption of oxygen homeostasis. Future studies can investigate the consequences of modifying the oxygen gradients within our current model on both human and microbial cell populations.

5. CONCLUSIONS

The intricate architecture of the human intestine is key to many functions, providing both biochemical and biophysical cues.¹⁰ Crypts provide a niche for stem cells to ensure the epithelium is renewed as quickly as it is lost to aging and to the shear stress of intestinal flow. Villi provide increased surface area for adequate absorption of nutrients.⁷ The cylindrical shape with a hollow lumen supports the oxygen gradients necessary to maintain both bacterial and human cell populations in close proximity.^8,74 While many attempts have been made to reproduce these complex features of the intestine in vitro, none have been able to faithfully recapitulate these intricate structures. Here, we demonstrated a unique 3D platform to generate scaffolds with crypts and villi that support intestinal epithelial cells and sustained low oxygen tension to support a human microbiome community. The intestinal epithelial cells were able to survive, proliferate, and differentiate properly both within the crypts and on the villi, indicating that the architecture of the scaffolds was conducive to a functional epithelium. Future directions include the use of this 3D scaffold system as a supporting structure to provide intestinal epithelial stem cells (repetitive compartmentalized structures of crypt-villus units in scaffold lumen) with the presentation of a set of biochemical and biophysical cues (scaffold bulk space) to enable the epithelium to maintain an ability to generate different native cell types and to constantly renew themselves to restock the stem cell pool and replace dead epithelial cells to support cell turnover when cultured in vitro. This goal would enable long-lived intestinal tissue systems in vitro, to support studies related to chronic conditions such as disease development, drug treatments nutritional regimens, and mimics of surgical interventions, among other needs.