Bioengineered Systems and Designer Matrices That Recapitulate the Intestinal Stem Cell Niche

Modern advances in understanding ISC biology are based largely on studies that define genetic pathways and mechanisms that govern ISC maintenance and differentiation. Among these are the wingless-related integration site (Wnt), bone morphogenetic protein (BMP), and Notch pathways, which classically are studied as key contributors to epithelial renewal in homeostasis, disease, and injury. Arguably, the Wnt/β-catenin signaling pathway has been a central focus of studies that have heavily influenced the current state in the field.^{,  ,}  Wnts are secreted ligands that bind their cognate receptors and function to regulate ISC maintenance and differentiation. Sox9 is largely a downstream Wnt target gene and shows a distinct expression gradient with higher expression at the base of the crypt in the ISC zone and lower expression through the TA zone, suggesting that Wnt signaling also is present in a gradient that mimics its downstream target genes.^{,  ,  ,}  In fact, 9 Wnts are expressed in the small intestine of mice and are regionally expressed along the crypt-villus axis. Contrary to popular assumptions, it appears that Wnt3 gradients may be formed not by simple diffusion, but rather by “plasma membrane dilution” as cells divide. A Wnt3-enhanced green fluorescent protein (EGFP) fusion transgenic mouse model enabled visualization of Wnt3 expression by proxy and showed high Wnt3 expression at Paneth cells, which produce Wnt3, and lower expression up the crypt axis (Table 1). Paneth cell–derived Wnt transfer involves direct contact between Paneth cells, which previously was suggested by in vitro ISC-Paneth cell co-cultures. However, it remains to be determined whether all 9 Wnts establish gradients from their cellular sources.^,  Complete understanding of Wnt gradient formation is challenging because there are many sources of Wnts and Wnt antagonists, including subepithelial myofibroblasts and nonmyofibroblast mesenchymal cells expressing Foxl1, Gli1, and CD34 within the ISC niche.^{,  ,  ,  ,}  In addition, potentiation of Wnt signaling by the R-spondin family of secreted factors recently was implicated as a major orchestrator of ISC maintenance and proliferation and suggests that R-spondin gradients also may play a role in ISC dynamics.

Aside from direct regulation of stem cell maintenance and differentiation, Wnt gradients are essential for physiologic cell compartmentalization and migration by regulating proper cell sorting along the crypt-villus axis. Cell sorting in the epithelial monolayer is regulated in a Wnt-dependent manner through the erythropoietin-producing human hepatocellular receptor (Eph)/Eph family receptor interacting proteins (Ephrin) receptor/ligand pairs, in which expression of EphB2 and EphB3 receptors and their ligand Ephrin-B1 are expressed inversely (Figure 1, Table 1).^{,  ,}  Genetic disruption of the EphB2/B3 receptors results in defects in cell sorting and manifests as a loss of cell compartmentalization with proliferative and differentiated cells intermingled. For instance, loss of EphB3 alone results in defective Paneth cell sorting observed by scattering of Paneth cells along the crypt-villus axis.

BMP gradients are found in an inverse orientation to Wnt gradients along the crypt axis, and correlate with reduced proliferative capacity and increased differentiation as cells migrate away from low BMP at the crypt base toward higher BMP at the crypt-villus junction (Figure 1, Table 1).^,  During development, hedgehog signaling occurs in localized regions, driving a BMP gradient to pattern differentiated cells along the developing villi.^,  BMP signaling is required for intestinal homeostasis because BMP inhibition causes abnormal villus morphogenesis, epithelial hyperplasia, and ectopic crypt formation. In the intestine, BMP2 and BMP4 are secreted by intravillus and intercrypt mesenchymal cells, whereas the BMP antagonists noggin and gremlin 1 are expressed mainly in mesenchymal cells beneath the crypt. Together these BMP agonist/antagonist gradients contribute to regulating the ratio of ISC self-renewal properties and differentiation along the crypt-villus axis (Figure 1).

ISCs are located in the crypt base and persistent Notch signaling is required for their maintenance. Differentiation of ISCs into secretory lineages occurs when Notch signaling is reduced. Notch is a membrane-bound receptor that is stimulated by ligands Jagged and Delta-like ligands (Dll1/4) found on adjacent cells (Table 1). Under homeostatic Notch signaling, the majority of epithelial cells differentiate toward the absorptive lineage. When Notch signaling is reduced, ISC proliferation is reduced and ISCs differentiate toward a secretory cell fate (Paneth, goblet, enteroendocrine, or tuft cells).^,  Importantly, opposing activities of Notch and Wnt signaling regulate the balance between ISC maintenance and secretory lineage allocation. Although strict gradients of Notch and its receptors have not been formally established along the crypt-villus axis, evidence suggests that Notch activity may regulate the Wnt gradient influence on cell fate and highlights the critical interplay between different ISC regulatory pathways and the ultimate impact of factor gradients.

In close proximity to the ISC niche, ECM proteins and the biomechanical properties they provide have long been known to exert a powerful influence on ISC behavior and differentiation. Not surprisingly, gradients of ECM span the crypt axis, producing alterations in the matrix chemical and mechanical properties (Figure 1, Table 1). The crypt base is enriched for a number of ECM components within the basement membrane, including isoforms of fibronectins, laminins, collagens, and glycosaminoglycans, suggesting specific ECM component isoforms play a role in the regulation of ISC maintenance, proliferation, and differentiation.^{,  ,  ,}  ECM also is found diffusely throughout the lamina propria, providing supportive biomechanical properties for epithelial and mesenchymal cells along the crypt-villus axis because cell behavior has been shown to respond to and depend on ECM stiffness.

A physical scaffold of ECM supports the ISC niche and is in contact with all of the cellular components. Both epithelial and mesenchymal compartments deposit this ECM; however, the ECM proteins secreted by specific cell types have not been fully characterized.^,  The basolateral membrane of the epithelium directly interacts through integrin binding with the basement ECM, a thin sheet of matrix composed mostly of laminins and collagen IV. The ECM found in the niche can impact cell behavior by modulating signaling pathways that control cellular proliferation, growth, and death. This influence can occur through physical anchoring of cells, regulating biomechanical stiffness, and creating reservoirs of soluble factors. The latter is particularly interesting because it is conceivable that factors can be focused to microniches to exert a particular effect. For example, heparan sulfate proteoglycans, found on many ECM proteins, stabilize growth factors such as Wnt. Heparan sulfate proteoglycans are found on collagen XVIII in the basement membrane of the ISC niche, and although not formally tested, may interact with Wnt to increase the Wnt concentration toward the base of the crypt.

Laminins are the most abundant protein in the basement membrane and different laminins are regionally expressed along the crypt-villus axis. The localized distribution suggests a relationship between regional ECM composition and epithelial cell function. Laminin α2 is found surrounding the base of the crypts, whereas laminins α3 and α5 are found in the villi. Reduction of laminin α5 in the small intestine resulted in expression of colonic-type isoforms and transformation to colonic mucosal architecture, which was associated with increases in cell proliferation and migration. This study suggests that regionalization of ECM deposition fundamentally changes tissue morphology and cell fate. Other ECM proteins have been found to be distributed regionally, including tenascin, found at the base of the villus and surrounding the crypt, and fibronectin, surrounding the crypt.

Laminins interact with integrins found on epithelial cells, anchoring the cells to the ECM, which is essential for ISC survival because apoptosis occurs because of a lack of epithelial β1 integrin interaction. Consequently, β1 integrin is found homogenously along the crypt-villus axis. Regional expression of the α subunit of integrins exists, with α2 enriched in the lower crypt and α3 enriched in the upper crypt and villus regions.^,  These integrin interactions with laminins suggest a role of ECM in ISC proliferation and maintenance, further highlighting the role of the niche environment on ISCs.

What remains to be fully understood are the complex mechanisms that regulate ECM deposition and ECM-dependent signaling mechanisms that regulate bidirectional interactions, or dynamic reciprocity, between the epithelium and the mesenchyme. Although ECM clearly is appreciated as playing an essential role in intestinal epithelial biology, a more comprehensive understanding of the ECM complexity is limited by constraints in ECM and isoform detection, a low-resolution working knowledge of the different cell types that deposit ECM, and a lack of animal models to test ECM-dependent mechanisms.

Mechanical cues from the local environment and ECM, deriving from physical properties such as stiffness, shape, and adhesion, guide ISC fate through regulated sensing pathways and biochemical signal transducers.^{,  ,}  Although gradients of stiffness have not been directly measured in the different stromal and matrix regions surrounding the crypt/villus because of technical limitations, inferences about their respective properties can and have been made. For example, the mucosa, which supports the epithelium and crypt-villus architecture, has been shown to have 0.25–0.5 times the collagen content of the submucosa, suggesting that a stiffness gradient may exist on the millimeter scale between these layers. In addition, in vitro culture systems have shown optimal ISC growth on stiff matrices and optimal differentiation on soft matrices. Extension of this gradient to the micron scale of the crypt-villus axis is plausible, although this has yet to be confirmed in the in vivo intestine.

One of the main functions of the intestinal epithelium is to provide a barrier to luminal contents, including the large population of gut microbiota found in greatest abundance in the distal small intestine and colon. The co-evolution of animals and microbes has led to the majority of metabolic pathways within the host gut lumen involving, or even relying, on the gut microbiome, highlighting the importance of microbial product consideration when profiling the gradients of metabolites across the epithelium, especially in the colon.^{,  ,} 

The assignment of chemicals found in the gut lumen to microbial vs host production suggests that primary amines, polyamines, and short-chain fatty acids (SFCAs) are largely produced by the microbiome.^{,  ,}  The SCFAs produced from the fermentation of dietary fiber by microbes are perhaps the most well-studied microbial metabolite because they regulate energy metabolism and impact ISC proliferation.^{,  ,}  The SCFA that has garnered the greatest attention is butyrate, which differentiated colonocytes use as their primary energy source and is essential for undergoing healthy mitochondrial respiration. In contrast, butyrate acts as an inhibitor of ISC proliferation, with typical luminal concentrations (approximately 70 mmol/L in human being) capable of inducing ISC apoptosis through a Foxo3-related mechanism.^,  During homeostasis, the physical structure of the intestinal crypt forms an effective barrier against butyrate from reaching ISCs at the base of the crypt, resulting in a sharp gradient between the crypt base and the intestinal lumen where differentiated cells are actively metabolizing butyrate (Table 1).

Although not as rigorously characterized as SCFAs, other microbial metabolites also have been studied for their impact on the gut epithelium and on overall organismal health. Polyamines, which include putrescine, spermidine, and spermine, are produced by luminal microbiota and have been shown to be essential for growth and development of the small intestinal and colonic mucosa.^{,  ,  ,}  Free catecholamines, such as the neurotransmitters dopamine and norepinephrine, are produced within the gut lumen by microbes with high β-glucuronidase activity. Recently, the microbial product desaminotyrosine, produced from the digestion of plant flavonoids, was shown to enter the bloodstream and trigger type 1 interferon signaling, acting as a protective measure against influenza. These examples suggest that microbial metabolites produced in the gut lumen have both local impacts on epithelial proliferation and barrier function as well as far-reaching impacts on overall host health.

Oxygen is essential for epithelial cell survival; therefore, steep oxygen gradients exist across the intestinal mucosa and gut lumen. Partial pressures of oxygen reach 80 mm Hg deep in the submucosa, where intestinal cells are oxygenated by the oxygen-rich vasculature extending throughout intestinal tissue, but decrease to 0.1–1 mm Hg in the large intestinal lumen where anaerobic microbes thrive and facultative anaerobes act as oxygen scavengers (Table 1).^{,  ,  ,  ,}  Oxygen levels within the vasculature also may influence the microbiota composition, with more oxygen-tolerant microbes found closer to the intestinal epithelium. Production of nitric oxide by NO synthase within the vasculature and supporting stroma, potentially creating an NO gradient across the epithelium (Table 1).^{,  ,  ,}  Inversely, the luminal microbiota is the source of many gases including methane, H₂, and hydrogen sulfide, which may form gradients impacting the intestinal epithelium and mucosa (Table 1).^{,  ,}  For example, hydrogen sulfide is increased in intestinal tissue from patients with ulcerative colitis, suggesting that this gas can diffuse into the epithelium.^,  The rapid diffusion of gases through tissue in combination with relatively short lifetimes may enable gradients to be rapidly modulated and act as first responders or sentinel molecules to changing intestinal conditions or insults.

Although it is well recognized that many growth factor and ECM gradients exist, few studies have re-created these gradients for proper epithelial polarization and recapitulation of the in vivo crypt-villus architecture in vitro. Primary ISCs cultured with exogenous factors such as Wnt, Notch, and R-spondin form organoids with crypt-like buds possessing increased ISC numbers. Because of their 3D morphology, it has been technically challenging to re-create spatial gradients to form separate proliferative and differentiated cell zones within organoids. The requirement for exogenous ISC supporting factors suggests that ISCs alone are unable to produce an adequate niche environment, including the gradients associated with a fully polarized crypt/villus structure.^,  The use of microfluidic systems has enabled a linear gradient of Wnt3A and R-spondin 1 to be placed over colonic murine organoids to create a distinct stem/proliferative and differentiated cell zone within a single organoid and shows that growth factors are integral in re-creating the cell compartments found in the crypt.

Organoids typically are fully embedded within a hydrogel, possess an inaccessible lumen, show a nonphysiologic shape and size, and do not show the highly segregated spatial cell location seen within in vivo crypts. These limitations make conclusions from the manipulation of gradients in organoid systems challenging to interpret. Fortunately, a multitude of gradient-making systems exist for the task of applying controlled surface- and solution-based chemical gradients to cells. One of the most simple and robust gradient systems, the Transwell, represents a facile and robust strategy to form gradients across intestinal epithelial monolayers. A modified version has been paired with a shaped collagen scaffold to re-create arrays of polarized in vitro crypts and villi (Figure 2A and B).^{,  ,  ,}  Microwells (crypts) and posts (villi) were fabricated into a collagen scaffold to mimic the shape, size, and density of crypts in vivo. Primary human and murine epithelial cells were expanded as monolayers and then placed under a gradient of growth factors including Wnt, R-spondin, and Noggin, resulting in a polarized epithelium with stem/proliferative cells isolated to the base of the microwells and differentiated cells localized to the luminal surface or villi of the scaffolding. Importantly the epithelium in this culture system showed key features of the in vivo intestine, including the production of alkaline phosphatase by enterocytes and mucous production and secretion by goblet cells. Titrating the growth factor gradient moved the stem/proliferative cell regions up or down the in vitro crypt axis showing that the cells were fully responsive to the microenvironment.

Microfluidic devices have been used extensively to create solution gradients of many molecules including growth factors with a range of device designs, parameters, and operating conditions.^{,  ,}  With respect to soluble gradients of proteins relevant to the intestine, a 2-chamber microfluidic chip with an intervening porous membrane yielded a bipolar gradient of Wnt3A and Dickkopf-related protein-1 that could be turned on or off on-demand and featured programmable gradient features. Although gradients can be created in microdevices, significant hurdles still exist in integrating primary cells. Although the devices generally are fabricated from biocompatible substances, the materials do not recapitulate the chemical or physical properties required to support primary cells, including surface properties such as stiffness and porosity.

Gradients of immobilized growth factors are more simple and often easier to operate than soluble gradients. These gradients also can be engineered to release proteins over time, yielding both temporal and/or spatial gradients to manipulate the stem cell microenvironment or direct differentiation. This is best exemplified in human mesenchymal stem cell systems, in which microparticles with controlled concentrations of BMP-2 and insulin-like growth factors are placed into alginate and silk scaffolds to form a gradient. Similarly, an opposing gradient of platelet-derived growth factor and BMP-2 was formed on a membrane, regulating adipose-derived stem cell differentiation. Lateral gradients of BMP-2 and/or BMP-7 applied to a biocompatible film using a microfluidic device were competent to alter myofibroblast gene expression and enzyme activity while covalent immobilization of Wnt3A on aldehyde-functionalized surfaces directs human mesenchymal stem cell differentiation.^,  In the future, select positioning of growth factor–producing cells with respect to intestinal epithelial cells also may be used to provide both soluble and surface-based gradients to more accurately mimic the in vivo conditions.

In vitro platforms have provided simplified and well-controlled approaches to study cell behaviors under a gradient of surface-bound cues.^{,  ,}  Gradients of collagen, laminin, or fibronectin have been formed on surfaces comprising poly(D,L-lactic acid), gold films, poly(ethylene glycol) hydrogel, and other surfaces.^{,  ,}  Although gradients are generated most easily on planar substrates, gradients on 3D scaffolds may be more physiologically relevant albeit technically more difficult. A gradient of BMP-7, transforming growth factor-β₂, and vascular endothelial growth factor₁₆₅ was formed along the longitudinal direction of polycaprolactone/pluronic F127 cylindric scaffolds by first creating a gradient of fibril-like polycaprolactones for the attachment of proteins. Although a multitude of cell types (both primary and tumor cells) have been paired with surface-bound protein gradients, none have included primary intestinal cells owing to their complex culture requirements.^,  A self-renewing monolayer culture system for primary intestinal epithelial cells recently was realized on a collagen hydrogel and now offers the possibility of acting as substrate for gradients of surface-bound proteins.

Methods to create a continuous gradient of stiffness in hydrogels for cell culture have been reviewed recently. These strategies have created lateral stiffness gradients within the range of 1 kPa to 3.1 MPa using a variety of materials.^{,  ,  ,  ,}  Polyacrylamide gels have a unique advantage because their surface is easily conjugated with ECM proteins so that the ECM density is independent of substrate stiffness. In addition to planar gradients, stiffness variations along the longitudinal direction of a cylindrical hydrogel were created using a gradual freezing and thawing method. In a recent breakthrough, a synthetic hydrogel-based system of variable stiffness was developed to support organoid-based culture of ISCs. Intestinal cell expansion was optimal within matrices of intermediate stiffness (1.3 kPa), whereas stem cells embedded in soft matrices (300 Pa) proliferated poorly (Figure 2E). A platform possessing a gradient of stiffness but with a constant density of ECM will be critical to fully recapitulate all architectural and physiological features of the in vivo intestinal epithelium.

Reproducing the gradient profile of microbial products across organoids is technically difficult because these constituents are at their greatest concentration within the lumen, which in organoids is enclosed and not easily accessed through a hydrogel layer.^{,  ,}  In a tour de force effort, induced pluripotent stem cell (iPSC)-derived organoids were individually microinjected with Escherichia coli to investigate the interaction of the intestinal epithelium with the microbes. Colonization with E coli induced functional maturation of the immature organoids, increasing antimicrobial peptide production, mucus layer maturation, and improved barrier function. To circumvent the challenge of microinjecting single Matrigel (BD Biosciences, San Diego, CA)-embedded organoids, a strategy based on the preferred adhesion of cells to ECM rather than nonadhesive microstructures was developed, enabling cell self-organization around a pH-sensing particle. Refinements in strategies such as this may enable future delivery of analyte-sensing or drug-releasing microparticles and/or microbes to the lumen of organoids.

The use of a Transwell insert also provides a robust strategy to form microbial-product gradients across intestinal epithelial monolayers; however, the vast majority of these studies have focused on the response of tumor cells to microbial metabolites.^{,  ,  ,}  Although these models provide useful insights, recently established primary intestinal epithelial monolayer cultures in these inserts are expected to replace the intestinal cancer models.^{,  ,  ,}  For example, a mouse primary colonic monolayer was used to show that interferon-γ (IFN-γ) did not increase IgA-receptor expression or transcytosis as IFN-γ did in tumor cells.

Perhaps the most straightforward strategy to establish a gradient of microbial products across an in vitro epithelium is to co-culture commensal microbes on the luminal surface of a monolayer. Because most intestinal microbes are obligate anaerobes, these model systems must supply an appropriate deoxygenated environment for the microbes in concert with an oxygen source to maintain intestinal epithelial cell viability. This demanding co-culture environment is most easily achieved using tumor models. A microfluidic model, the HuMix system (University of Luxembourg, Luxembourg City, Luxembourg), uses 3 stacked compartments, a deoxygenated microbial chamber, an apical cell chamber, and a basal oxygenated chamber to re-create mammalian cell-microbe cross-talk.^,  Mixtures of facultative (Lactobacillus rhamnosus GG) and obligate (Bacteroides caccae) anaerobes were co-cultured with mammalian cells, showing the potential of this technology.

To truly understand the interactions of microbes and their gradients within intestinal tissue, primary tissue constructs with the architecture and tissue polarity of in vivo colon crypts must ultimately be used for gradient formation and microbial co-culture. Primary human crypts paired with a physiologic SCFA gradient along the crypt axis showed that butyrate, but not acetate or propionate, suppressed the proliferative cell population in a dose-dependent manner, diminishing the size of the ISC niche (Figure 2C). The SCFA gradients also enhanced epithelial cell differentiation into the absorptive colonocyte lineage as is shown to occur in vivo. The ability to recapitulate complex architectural features and physiologic responses shows the value of these emerging organ-on-chip systems.

The intestine and surrounding lymphoid system are a major site for immune system education regarding potential pathogenic microbes as well as a site for maturation of immune cells.^{,  ,  ,}  Gradients of tumor necrosis factor-α and IFN-γ applied to human in vitro crypt arrays reduced the alkaline phosphatase activity of differentiated cells as well as the size of the stem/proliferative cell compartment, profoundly altering crypt properties. Model systems to identify the differential impact of these inflammatory mediators and immune cells along the length of the crypt axis will be critical to understanding the impact of inflammation on the initiation or exacerbation of disease states.

Platforms capable of spatiotemporal oxygen control have the potential to enable the study of intestinal physiology, microbial ecology, and their interactions in vitro. Oxygen gradients with arbitrary profiles (eg, linear, exponential, and nonmonotonic) have been generated as stable or reconfigurable gradients.^,  Oxygen gradients compatible with an open-well culture also have permitted direct observation of cell behavior in response to varying oxygen levels. Vertical oxygen gradients over Caco-2 cells established by simultaneous perfusion of oxygenated and deoxygenated media through 2 separated microchannels permitted co-culture of Caco-2 cells with an obligate anaerobe under conditions representative of the gastrointestinal human–microbe interface. These types of controllable gradients may in the future be used to mimic hypoxic insults during loss of blood flow or epithelial integrity in the intestine.

Although gas control with microfluidics offers unique advantages, these systems typically require bulky or sophisticated off-chip equipment such as pumps, valves, and control systems. A spontaneous (cell-mediated) oxygen gradient was generated using cell respiration to deplete oxygen. Intraluminal oxygen gradients were spontaneously established when Caco-2 cells were cultured on a porous scaffolding containing a shaped, high-surface-area lumen (Figure 2D). In another example, rapid, predictable oxygen gradients were established by embedding cells in gelatin-based O₂-controllable hydrogels. Although numerous platforms are available, the effect of gas gradients on primary intestinal epithelial cells and host-microbe interactions has not been studied extensively, presenting an opportunity for future research.