Esophageal 3D Culture Systems as Modeling Tools in Esophageal Epithelial Pathobiology and Personalized Medicine

As a hollow muscular organ, the esophagus serves the passage of ingested food and liquid from the oral cavity to the stomach. Its luminal surface is lined by the mucosa, comprising stratified squamous epithelium and the underlying lamina propria and muscularis mucosa. The esophageal epithelium consists of proliferative basal keratinocytes and suprabasal keratinocytes, the latter undergoing postmitotic terminal differentiation, passive migration toward the luminal surface, and, ultimately, desquamation into the lumen. Through this dynamic process, a proliferation-differentiation gradient is generated while epithelial renewal occurs over a period of 2 weeks. Molecular markers defining basal keratinocytes include cytokeratins K5 and K14,^,  transcription factors p63, and Sry-related HMG box (SOX2), and cell surface molecules such as neurotrophin receptor p75NTR, integrin β1 (CD29), integrin α6 (CD49f), and transferrin receptor (CD71). Suprabasal keratinocytes are defined by differentiation markers such as cytokeratins K4 and K13, involucrin, and filaggrin,^{,  ,}  coupled with down-regulation of basal keratinocyte markers.

Doupe et al proposed that the esophageal epithelium is maintained by a single population of basal keratinocytes that give rise stochastically to proliferating and differentiating daughters with equal probability; however, functional cell heterogeneity has been postulated among basal esophageal keratinocytes. A minor subset of basal keratinocytes divide slowly or rarely and may have properties of quiescent stem cells. Such a cell population may provide an explanation as to how premalignant keratinocytes accumulate genetic alterations over years without being lost through epithelial renewal. Multiple cell surface and functional markers have been suggested to identify unique subsets of basal keratinocyte stem/progenitor cells, including neurotrophin receptor p75^NTR, integrins (β4, α6),^,  and ABCG2 gene product. Esophageal keratinocytes expressing these molecular markers have shown colony formation and self-renewal capabilities while also generating terminally differentiated progenitor cells.

Species differences exist between rodents and human beings with regard to anatomic esophageal structure. Foremost, the rodent esophagus lacks esophageal glands and papillae, both of which are present in the human esophagus. In addition, the rodent esophagus shows more explicit keratinization in the superficial cell layers, also known as stratum corneum of the squamous epithelium, as compared with its human counterpart. The rodent stomach consists of 2 compartments: the forestomach and distal stomach, featuring squamous epithelium and columnar epithelium, respectively, and some regard the forestomach as the counterpart of the human lower esophagus. This is important to note because Barrett’s esophagus (ie, intestinal metaplasia of the esophagus) is a human mucosal lesion involving the esophagogastric junction and has been modeled at the squamocolumnar junction within the murine stomach.^, 

One essential physiological function of the esophageal mucosa is to serve a barrier against thermal, physical, or chemical agents, and factors related to luminal contents, including microorganisms, food antigens, gastroduodenal acids, and alcohol, all of which may contribute to the pathogenesis of esophageal diseases. Unlike the stomach, duodenum, and intestine, the luminal surface of the esophagus is not densely covered by mucus layers. Given the lack of the stratum corneum in the human esophagus and the lack of esophageal glands in rodents, the epithelial barrier function of the esophagus is attributed mainly to intercellular junctional complexes including tight junctions, adherens junctions, and desmosomes formed by cell–cell adhesion molecules such as E-cadherin, p120 catenin, and claudins. The dysfunction of these adhesion molecules has been implicated in esophageal disease conditions.^{,  ,  ,} 

Throughout a long history of cell culture, various forms of 3D culture methodologies have been developed along with unique scaffolds, matrices, and cell culture media. In the esophagus, 3D culture systems have provided unique platforms to study multiple biological processes, including epithelial cell proliferation, differentiation, motility, stress response, and both homotypic and heterotypic cell–cell communications. Cellular interactions involve a variety of cell types (eg, fibroblasts, endothelial cells, and inflammatory cells) in the esophageal tissue microenvironment under homeostatic and pathologic conditions (eg, inflammatory milieu), and are mediated via cell surface molecules (eg, integrins and receptors such as Notch) as well as extracellular matrix proteins (eg, matrix metalloproteinases), as discussed in this review. The ability to experimentally manipulate 3D cultures has greatly enhanced our understanding of the molecular mechanisms and signaling pathways underlying esophageal physiology and pathophysiology.

Organ (explant) culture was a major tool for in vitro live esophageal tissue analyses before primary esophageal epithelial cell culture and esophageal cancer cell lines became available in late 1970s and early 1980s, respectively. The foremost advantage of organ culture is the maintenance of natural tissue architecture in situ. Organ culture may be used to study cross-talk between epithelial cells and nonepithelial cells in a live tissue-like context. Given the potential importance of a variety of cell types (ie, epithelial cells, fibroblasts, nerve cells, immune cells, and endothelial cells) present in the tissue microenvironment, organ culture indeed may be more physiologically relevant than other 3D culture systems because co-culturing multiple cell types remains difficult.

Early organ culture studies have shown that esophageal explants from animals and human beings remained viable for 3–14 days ex vivo and provided substantial insight into esophageal physiological functions, including secretory and absorptive activities by basal and differentiating prickle cells, as well as endocytosis mediated by prickle cells and terminally differentiated superficial cells. In fetal human esophagi, organ culture detected not only epithelial cell proliferation but also replacement of columnar ciliated epithelium with stratified squamous epithelium, recapitulating the epithelial changes occurring during esophageal development.

Organ culture also has been used to study esophageal pathologies. Production of esophagitis-relevant cytokines was shown in patient-derived squamous epithelial explants. Retinoic acid induced BE-like glandular differentiation in explants derived from squamous esophageal epithelium. Explanted patient-derived mucosal biopsy specimens of Barrett’s esophagus showed increased proliferation and cyclooxygenase (COX)-2 expression in response to bile salt exposure.^,  Likewise, acid pulse induced a hyperproliferative response in Barrett’s esophagus explants via altered Na⁺/H⁺ exchange. Radiation sensitivity was tested in tumor explants from ESCC and EADC.

Although the use of chemically defined medium containing insulin and hydrocortisone enabled the maintenance of live human esophageal tissues in organ culture for up to 6 months, most experiments in organ culture are performed for short time periods (24–72 h) only. This is in part because the proliferation kinetics of esophageal keratinocytes in organ culture are not necessarily reflective of those occurring in vivo. In addition, monolayer outgrowth of keratinocytes has been observed in explant culture. Organ culture is inevitably more complicated than the 3D culture systems discussed later. Other limitations of this method include the difficulty of genetic manipulation of human tissues and the inability to passage organ cultures. Esophageal explants have yet to be derived from genetically engineered mouse models. To this end, Cre-mediated gene deletion can be performed ex vivo in organ culture with tissues isolated from transgenic mice carrying tamoxifen-inducible cell type–specific Cre and a floxed gene of interest.

Unlike organ culture, most 3D culture systems involve dissociation of originating tissue samples or preparation of cells grown in monolayer cultures before 3D reconstitution in vitro. Such 3D culture systems can be largely classified into 2 categories. One category is represented by OTC in which esophageal epithelial cells are grown over a collagen matrix containing fibroblasts that mimics the subepithelial lamina propria. The second category features spherical 3D structures of esophageal epithelial cells typically generated under submerged conditions. This includes multicellular spheroid culture, sphere formation assays, and 3D organoids.

OTC, also known as raft culture, is a form of tissue engineering with recapitulation of esophageal physiology and pathology (Figure 1). OTC was first established in the field of skin biology in the 1980s when the dermal equivalent comprising contracted type I collagen and fibroblasts was submerged in liquid medium and used as a raft-like platform to grow multilayered epidermal keratinocytes at the air–liquid interface, mimicking skin architecture as found in vivo.^{,  ,  ,}  The air–liquid interface triggers epithelial stratification and cornification by inducing terminal differentiation in the superficial cell layer with induction of a variety of squamous cell differentiation markers including cytokeratins, transglutaminase, involucrin, and filaggrin. Although normal keratinocytes generate well-organized epidermal architecture, keratinocytes transformed by oncogenic viruses such as human papilloma virus form disorganized epithelia displaying aberrant proliferation, differentiation, and invasion into the dermal compartment.^{,  ,  ,  ,} 

We adopted this system for normal and genetically engineered esophageal keratinocytes, optimizing culture conditions and evaluating the influence of a variety of sources of fibroblasts,^,  including the skin and the esophagus of adults, children, and embryos as well. Fibroblasts evaluated in OTC include those isolated from the skin and the esophagus of adults, children, and embryos, as well as cancer-associated fibroblasts.^{,  ,}  Although multiple esophageal cell lines were used successfully in OTC (Table 1), telomerase-immortalized normal human esophageal keratinocytes (EPC2-hTERT) and primary human fetal esophageal fibroblasts (FEF3) have served as standard and quality-control cells. EPC2-hTERT cells show complete stratified epithelia in OTC. Of note, commonly used HET1A, an oncogenic simian-virus-40 T-antigen-immortalized human esophageal epithelial cell line, did not recapitulate normal squamous esophageal epithelium in OTC. With detailed protocols published by Kalabis et al, we have interrogated gene and molecular functions as well as cell–cell and cell–extracellular matrix interactions in broad contexts related to the esophageal physiological and pathologic microenvironment,^{,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,}  as detailed later in this article.

In vivo transplant culture has served as another platform to reconstitute esophageal epithelium in 3D. According to this method, which may not be categorized as cell culture in a strict sense, epithelial cells and fibroblasts are injected into the devitalized and denuded trachea tube of rats, which then is further transplanted into immunodeficient mice. Esophageal keratinocytes are allowed to grow for 4 weeks inside the xenografted trachea and the resulting epithelial structure may be analyzed morphologically. Wang et al^,  used this system successfully to characterize the contribution of Hedgehog signaling to esophageal development and metaplasia with esophageal keratinocytes isolated from genetically engineered mice.

Multicellular spheroid culture emerged in the early 1970s, stemming from dissociation-aggregation experiments in which dissociated tissue-derived cells undergo aggregation and self-organization under free-floating conditions with gentle stirring via spinner flasks or roller bottles. This technique was applied to a variety of cell types with primary culture and established cell lines recapitulating histologic tissue architecture of the originating organ.^{,  ,}  This platform has been used to study cross-talk between tumor cells and other cell types (eg, immune cells and endothelial cells) in co-culture experiments.^{,  ,}  Although multicellular spheroid culture was performed with nontransformed bovine esophageal epithelial cells as well as ESCC cell lines (Table 1), these studies were limited to morphologic comparison or analysis of cell viability in resulting spheroids.^, 

In the history of 3D culture, the importance of extracellular matrix components for self-organization of functional epithelial cells has been recognized by studies using mammary epithelial cells grown under floating conditions in type I collagen or Matrigel (Corning, Tewksbury, MA) rather than on a plastic surface. Under such conditions, mammary epithelial cells showed not only polarity but also luminal formation with milk protein secretion, which occurred more efficiently in the presence of Matrigel compared with type I collagen. Andl et al embedded esophageal epithelial cell aggregates into either type I collagen or Matrigel (Table 1) to observe that cell migration and invasion from the resulting spheroids were coupled with loss of E-cadherin–mediated cell adhesion. They further identified activin in the tumor microenvironment as a regulator of spheroid growth and invasion.

Sphere formation assays gained popularity in the 2000s as an excellent tool to characterize stem cells from multiple tissue types, including tumors (see Weiswald et al for an excellent review regarding tumor spheres and other spherical cancer models). Sphere formation assays were first used to show proliferation, self-renewal, and multipotent capabilities of neural stem cells in serum-free medium supplemented with epidermal growth factor (EGF) and differentiation-inducing other growth factors. To form 3D spherical structures, a single cell or a small number (<10¹–10²) of cells are inoculated per well and allowed to proliferate under free-floating conditions. This permits cell fate determination by tracking cell lineage within resulting spherical 3D structures. Unlike multicellular spheroid culture, cell aggregation is deliberately avoided at the onset of sphere formation assays to ensure a proliferative expansion of single-cell derivatives. In cancer research, sphere formation was used to explore cancer stem cells (CSCs) or tumor-initiating cells in conjunction with a variety of putative CSC markers and evaluation of therapeutic sensitivity for such cell populations.

As summarized in Table 1, sphere formation assays have been used to characterize EADC and ESCC CSCs expressing a variety of markers, including CD54 (intercellular adhesion molecule 1), CD49f, CD44, CD271, CD90, and aldehyde dehydrogenase, either alone or in combination.^{,  ,  ,  ,  ,  ,}  Increased sphere formation corroborated CSC attributes such as chemoresistance, invasiveness, and tumorigenicity, and expression of stemness markers (SOX2, ALDH1A1, and KLF4) induced by transforming growth factor-β1. Sphere formation assays were used to document the requirement of H3K4 demethylase Jumonji/Arid1b (Jarid1b) in the maintenance of ESCC CSCs and the role of Hippo coactivator yes-associated protein 1-mediated transcriptional regulation of SOX9 in EADC CSCs.

Sphere formation also has been used to test pharmacologic therapeutic effects upon putative esophageal CSCs. Metformin appeared to inhibit sphere formation by Aldehyde dehydrogenase (ALDH)1⁺ EADC CSCs by targeting phosphatidylinositol 3-kinase/AKT and mammalian target of rapamycin. Pharmacologic inhibition of Notch signaling by γ-secretase inhibitors impaired tumor initiation as well as sphere formation by EADC CSCs, and Notch appeared to regulate genes such as SOX2, which is essential in stemness. The plant-derived agent curcumin also was found to decrease the sphere formation capability of ESCC cell lines.

In sphere formation assays, microRNA miR-181b was implicated in signal transducer and activator of transcription-3 (STAT3)-mediated transcriptional regulation of CSCs, whereas miR-377–mediated regulation of CD133+ ESCC CSCs has been shown. It must be noted that sphere formation assays in the studies described earlier were performed with cell lines, but not cancer cells isolated from primary tumors. In 1 study, sphere formation assays failed to detect ESCC CSCs expressing CD44 despite the high tumor-initiating capability of these cells being validated in mice. In addition, high densities (10³–10⁴ cells/well) of cells were seeded to form spheres in many of the described studies,^{,  ,  ,  ,  ,  ,  ,}  while several studies failed to provide information as to how spheres were generated,^{,  ,}  precluding determination of whether resulting spherical 3D structures represented cell aggregates or single-cell–derivative products. Because this represents a deviation from the original neurosphere assay principle, data obtained in these studies must be interpreted carefully.

The 3D organoid system has emerged in the past several years as a robust tool in basic research with the potential for personalized medicine. 3D organoids have been defined as a miniature organ-like 3D structure derived from single cells or a small number of cell clusters containing stem/progenitor cells grown in basement membrane (ie, Matrigel) under submerged conditions (Figure 1). Overcoming the difficulty to culture intestinal cell types, Sato et al^,  were successfully able to grow stem and progenitor cells from either isolated intestinal crypts or single-cell suspensions prepared from the small or large intestine into lobulated 3D structures containing crypt and villus compartments, the latter containing terminally differentiated secretory and absorptive cell lineages. By passaging dissociated primary structures to generate secondary 3D organoids, this system has been used to validate the self-renewal activities of putative stem cells. Because this can be performed using live tissue pieces from biopsy specimens or even frozen tissues, this novel cell culture method has been transformative with great potential to advance personalized medicine, for example, by testing the chemotherapeutic sensitivity of colorectal cancer–derived 3D organoids from individual patients. Intestinal organoids have been coupled successfully with genetically engineered mouse models of colorectal cancer, facilitating functional studies into the biology of this organ in the context of health and disease. Application of the 3D organoid system has been extended to a variety of cell types from digestive (intestines, stomach, liver, pancreas) and nonintestinal organs (eg, brain, lung, breast, kidney, prostate, and ovary) from both human and murine tissue sources as well as embryonic or induced pluripotent stem cells.^,  Besides colorectal cancer,^{,  ,}  patient-derived tumor organoids have been generated successfully from pancreas,^,  prostate, breast, and uterine cancer cells in primary and metastatic lesions and blood samples, with a goal of translation into personalized therapy.

3D organoids have been cultured in Dulbecco's modified Eagle medium:nutrient mixture F12-based serum-free medium supplemented with transferrin, selenium, ethanolamine, insulin, antioxidants, and vitamins. This medium is supplemented with pharmacologic agents, such as transforming growth factor-β kinase/activin receptor-like kinase inhibitor A83-01, the p38 mitogen-activated protein kinase inhibitor SB202190 and the rho-associated kinase inhibitor Y-27632 to facilitate the establishment of organoids. Besides Matrigel, addition of growth factors and hormones, including noggin, EGF, Wnt3A, R-spondin, and gastrin, to 3D organoid media provides essential niche factors present in the tissue microenvironment in situ. Unique niche factors may influence organoid formation differentially from different cell types as pioneered by Sato et al in intestinal cell types. For example, Wnt3A is essential for stem cell maintenance in colonic, but not small intestinal, 3D organoids, in mice in which withdrawal of Wnt3A facilitates differentiation in murine colonic organoids. Optimization of 3D organoid culture medium has broadened the cell types of human and rodent origin that can be grown successfully and passaged as 3D organoids.

3D organoids featuring the stratified squamous epithelium of the esophagus were first established from murine esophageal mucosa by DeWard et al to investigate esophageal stem and progenitor cells. This group used medium similar to that used for human intestinal 3D organoids. We have used simplified medium components to generate single-cell–derived murine esophageal 3D organoids for multiple passages,^,  indicating that certain agents and factors are dispensable for murine esophageal 3D organoid culture. The conditions that we have optimized were permissive for generation of murine 3D organoids from normal esophageal epithelium as well as chemical carcinogen-induced precancerous and ESCC lesions.