|Home | About | Journals | Submit | Contact Us | Français|
The in vitro analysis of intestinal epithelium has been hampered by a lack of suitable culture systems. Here we describe robust long-term methodology for small and large intestinal culture, incorporating an air-liquid interface and underlying stromal elements. These cultures showed prolonged intestinal epithelial expansion as sphere-like organoids with proliferation and multilineage differentiation. The Wnt growth factor family positively regulates proliferation of the intestinal epithelium in vivo. Accordingly, culture growth was inhibited by the Wnt antagonist Dickkopf-1 (Dkk1) and markedly stimulated by a fusion protein between the Wnt agonist R-spondin-1 and immunoglobulin Fc (RSpo1-Fc). Furthermore, treatment with the γ-secretase inhibitor dibenzazepine and neurogenin-3 overexpression induced goblet cell and enteroendocrine cell differentiation, respectively, consistent with endogenous Notch signaling and lineage plasticity. Epithelial cells derived from both leucine-rich repeat-containing G protein–coupled receptor-5–positive (Lgr5+) and B lymphoma moloney murine leukemia virus insertion region homolog-1–positive (Bmi1+) lineages, representing putative intestinal stem cell (ISC) populations, were present in vitro and were expanded by treatment with RSpo1-Fc; this increased number of Lgr5+ cells upon RSpo1-Fc treatment was subsequently confirmed in vivo. Our results indicate successful long-term intestinal culture within a microenvironment accurately recapitulating the Wnt- and Notch-dependent ISC niche.
The surface of the intestine is lined by a simple columnar epithelium that undergoes complete regeneration every 5–7 d1,2. Underlying this profound regeneration are ISC populations3, including pan-intestine Lgr5+ ISCs at the crypt base4 and small intestine Bmi1+ ISCs at approximately four cell positions directly above the Paneth cells5. These ISCs divide to produce transit amplifying cells, which migrate toward the lumen and differentiate into absorptive enterocyte, goblet, Paneth and enteroendocrine lineages2,6, followed by either extrusion into the luminal surface or Paneth cell phagocytosis7.
Stem cells are generally influenced by a microenvironmental niche, typically comprised of epithelial and mesenchymal cells and extracellular substrates, which instructs either self-renewal or selective adoption of a particular cell lineage8,9. The ISC niche is notable for myofibroblasts adjacent to the crypt base, which are believed to elaborate paracrine signals regulating the neighboring ISCs1,9,10. Extracellular Wnt signals are absolutely required within the ISC niche, as deduced from the rapid ablation of proliferation and secondary loss of differentiation observed with the secreted Wnt inhibitor Dickkopf-1 (Dkk1)11,12. Notch signals are similarly essential, with stimulation amplifying the progenitor pool and inhibition resulting in large-scale conversion to post-mitotic goblet cells13,14.
Many attempts have been made to produce culture systems that mimic normal intestinal epithelial growth and differentiation15–17. The most substantial obstacle has been the rapid initiation of apoptosis within a few hours after intestinal cells are removed from the basement membrane and underlying stroma18,19. Although explants of embryonic, neonatal or adult gut will develop successfully when transplanted into syngeneic or immunocompromised host animals20, the tissue is largely inaccessible for experimental manipulation and time-series observation. Several studies have demonstrated intestinal epithelial growth and differentiation in organ culture, but the three-dimensional (3D) architecture of the tissue was not preserved, and preservation of epithelial viability lasting more than 10 d is difficult21. Recently developed 3D embryonic organoid cultures allow cellular differentiation and intestinal morphogenesis of the tissue22. However, these in vitro methods have been restricted to embryonic tissue and maintain cellular viability for less than 14 d. Thus, to date, the study of intestinal stem and progenitor cells has been largely dependent on in vivo approaches, and primary culture methodology is not commonly employed in studies of the intestine.
Here we describe a robust long-term methodology for primary mouse intestinal culture allowing sustained intestinal proliferation and multi-lineage differentiation over a range of 30 to >350 d, using neonatal tissue as starting material. Defining characteristics include the use of an air-liquid interface coupled with a 3D culture matrix, as well as recapitulation of both the cellular myofibroblast architecture and the rigorous Wnt and Notch dependence of the ISC niche. We further exploit this methodology to show the presence of putative ISC populations within these cultures and their in vitro modulation by the Wnt agonist RSpo1-Fc. These studies describe a method to enable study of both ISCs and the ISC niche, as well as general investigations of intestinal biology.
3D culture of either small or large intestine from neonatal mice within a collagen gel with an air-liquid interface (Supplementary Fig. 1a online) yielded expanding cystic structures (termed intestinal spheres) on gross inspection within 7 d, following initial outer spindle cell growth (Fig. 1a and Supplementary Fig. 1b). Virtually all of these cultures showed growth for 30 d, with some growing to >350 d in vitro (the latest time point examined) (Fig. 1b–e and Supplementary Figs. 2a and 3a online). The wall of the intestinal spheres consisted of a polarized epithelial monolayer with an apical, inner luminal surface and a basal outer surface in close proximity to myofibroblasts and the collagen matrix (Fig. 1c,e). The intestinal epithelial cells not only showed highly proliferative activity at extended time points (Fig. 1c,e) but also expressed numerous markers for multilineage differentiation to the absorptive enterocyte (lactase, maltase, sucrase and Na+-K+ ATPase), goblet (mucin-2), enteroendocrine (chromogranin A, serotonin and glucagon-like peptide-2) and Paneth cell (lysozyme, cryptdin and matrix metalloproteinase-7) lineages (Fig. 1c and Supplementary Fig. 2a). Underlying myofibroblasts expressed α-smooth muscle actin (Supplementary Fig. 2b). Ultrastructural examination revealed the fully differentiated microstructures of cultured intestinal epithelial cells, including microvilli, mucus granules and endocrine granules, as well as intracellular connections of junctional complexes (Fig. 1f). We have also been able to use small and large intestine from juvenile or adult mice up to 26 weeks of age (the oldest age evaluated) as starting material (Fig. 2). Although we have less experience with cultures of adult intestine, our preliminary studies indicate that their viability is much less extensive than with neonatal cultures.
Regardless of the age of the mouse cells used for the intestinal culture, both proliferative zones and differentiated zones were present (Supplementary Fig. 3b,c). Whereas proliferative zones were commonly observed within areas of monolayer (Supplementary Fig. 3b) within 2 weeks, crypt-like structures were also often produced within both small and large intestinal spheres (Figs. 1e and and22 and Supplementary Fig. 3a,c). Furthermore, villus-like protrusions were occasionally present in the jejunal spheres (Fig. 2b). The crypt-like structures showed marked proliferative activity; in contrast, the villus-like structures or differentiated zones were devoid of proliferating cell nuclear antigen (PCNA)-positive cells (Fig. 2c,d and Supplementary Fig. 3b,c). Accumulation of apoptotic sloughed cells positive for single-stranded DNA in the sphere lumen (Supplementary Fig. 2c) and BrdU pulse labeling (Supplementary Fig. 2d) revealed the rapid turnover and proliferation of intestinal epithelial cells in culture. Some of the intestinal spheres showed autonomous contraction within the outer surrounding muscle layer during culture days 5–14 (Supplementary Video 1 online). The initial growth of myofibroblasts and the air-liquid interface microenvironment were essential for intestinal epithelial growth in this system (Supplementary Fig. 1). Although we did observe proliferative zones and viable tissue at >200 d of culture, they became more and more sporadic over time, as opposed to a more generalized presence in cultures <30 d old. Notably, viable intestinal tissue at such extended time points seemed to be obligately associated with a robust underlying stroma (Fig. 1c,e).
As Wnt signaling promotes maintenance of epithelial stem cells and early progenitor compartments23, we hypothesized that the long-term growth of the intestinal cultures would be modulated by alteration of Wnt signals in vitro. We previously showed that Dkk1-dependent antagonism of extracellular Wnt signaling produces rapid cessation of intestinal epithelial proliferation and crypt loss in the adult mouse12, consistent with findings in nonconditional villin-Dkk1–expressing transgenic mice11. Accordingly, addition of recombinant Dkk1 into the culture medium at the time of plating resulted in dose-dependent growth inhibition (Fig. 3a,b). Similarly, when we preestablished small or large intestine spheres without Dkk1 for 28 d followed by subsequent Dkk1 treatment for 5 d, we observed rapid degeneration of the epithelial layer (Fig. 3c). Conversely, to achieve gain-of-function Wnt activation, we used recombinant RSpo1-Fc, as studies have shown that it strongly augments intestinal proliferation in vivo24. Treatment with an RSpo1-Fc fusion protein (Supplementary Fig. 4 online) produced a significant increase in the number and size of intestinal spheres, with a marked increase in the number of PCNA+ proliferating cells with undifferentiated features, whether in neonatal cultures (Fig. 3d–h) or those derived from adult 3- to 26-week-old mice (Fig. 2b,d,h and Supplementary Fig. 5a online). Furthermore, RSpo1-Fc treatment extended the ability to culture adult intestine to 28 d (Fig. 2i,j), which seemed to be otherwise restricted to approximately 7–10 d without RSpo1-Fc treatment (data not shown). Expression of the β-catenin and T cell factor target genes Cd44 and Myc in cultured intestinal epithelial cells was strongly induced by RSpo1-Fc, confirming functional stimulation of Wnt signaling (Fig. 2e,f and Supplementary Fig. 5b). Thus, Wnt gain- and loss-of-function studies indicated both accurate recapitulation of the Wnt-dependent ISC niche in culture, as well as Wnt responsiveness of the cultured intestinal epithelium.
Notch genes encode large, single-transmembrane receptors regulating a broad spectrum of cell fate decisions; in the intestine, Notch governs secretory lineage fate and maintains the proliferative progenitor state13,14. Notch inhibition with either γ-secretase inhibitors25 or by conditional targeting of the Notch pathway transcription factor recombination signal-binding protein for immunoglobulin κ J region13 induce marked goblet cell hyperplasia in vivo. Accordingly, treatment of preestablished small intestine cultures with the γ-secretase inhibitor dibenzazepine25 for 5 d produced complete conversion of the epithelial layer into terminally differentiated goblet cells, as determined by morphology, periodic acid–Schiff (PAS) staining and the absence of mitotic PCNA staining (Fig. 4a and Supplementary Fig. 6 online). We observed identical results with dibenzazepine treatment of large intestinal cultures (Supplementary Fig. 6). These results suggested accurate ex vivo recapitulation of the Notch-dependent ISC niche with preservation of endogenous Notch signaling within the intestinal sphere culture.
We further examined the plasticity of the intestinal sphere cultures with respect to the enteroendocrine lineage. The helix-loop-helix transcription factor neurogenin-3 (Ngn3) regulates enteroendocrine fate, with overexpression increasing intestinal enteroendocrine cell number26–28. Adenoviral Ngn3 overexpression in jejunal cultures was sufficient to induce an approximately threefold increase in the number of chromogranin A–positive enteroendocrine cells versus a control adenovirus expressing an antibody IgG2α Fc fragment12 (Fig. 4b). Both Ngn3-mediated enteroendocrine cell differentiation and dibenzazepine-mediated goblet cell differentiation argue for the substantial plasticity of the cultured epithelium and the potential for its modulation by viral or small molecule approaches.
Mosaic analyses of developing intestine have revealed that intestinal epithelium is initially generated as a well mixed population of multiple progenitors at birth that becomes exclusively monoclonal within each crypt by postnatal day 14, suggesting specification of ISCs by the niche during this period29,30. Accordingly, intestinal cultures from tetrachimeric mice with mosaic expression of distinct fluorescent proteins30 showed progressively demarcated clonal fluorescence domains consistent with ISC specification (Supplementary Fig. 7 online).
The prolonged expansion and proliferation within the intestinal sphere cultures (>350 d) strongly suggested the presence of functional ISCs. We sought to demonstrate the presence of Lgr5+ and Bmi1+ epithelial cell populations, which are putative ISCs4,5, in the cultures as well as to exploit this system to explore their previously unexamined regulation by extracellular Wnt signals4,5. We thus exposed jejunal cultures from Bmi1Cre-ER/+;Rosa26LacZ/+ mice5 to 7 d of tamoxifen. In this model, tamoxifen temporally induces expression of Cre recombinase in Bmi1-expressing cells while simultaneously inducing expression of β-galactosidase from the ubiquitously active Rosa26 locus. Consequently, cells expressing Bmi1 and their progeny are permanently labeled upon tamoxifen treatment. Under these conditions, RSpo1-Fc treatment, as opposed to vehicle controls, robustly induced detectable LacZ+ epithelial clusters corresponding to Bmi1+ cells and their progeny, consistent with a previously undocumented responsiveness of the Bmi1+ lineage to extracellular Wnt signals (Fig. 5a,b).
As the Lgr5 gene, encoding an orphan G protein–coupled receptor, has been validated as a robust panintestinal ISC marker4, we examined Lgr5-expressing epithelial cell populations in both small and large intestinal cultures using fluorescent in situ hybridization (Fig. 5c). This revealed Lgr5+ cells in both jejunum and colon cultures that were more numerous with RSpo1-Fc treatment (Fig. 5d), paralleling increased mitotic index and growth of the cultured intestinal spheres (Fig. 3d–h). Confirming the predictive nature of the culture system, we further observed increased numbers of Lgr5+ cells in small and large intestinal crypts in vivo after systemic adenoviral RSpo1-Fc expression (Fig. 5e and Supplementary Fig. 8a online), paralleling marked crypt hyperplasia and increased mitotic index (Fig. 5f and Supplementary Fig. 8b,c).
This study describes a robust long-term methodology for either small or large intestinal culture, allowing sustained intestinal proliferation and multilineage differentiation for a range of 30 d to >350 d. The lack of long-term methodology for primary intestinal culture has been a substantial obstacle to exploration of intestinal stem and progenitor cell biology and more general questions of physiology, despite attempts to recapitulate normal intestinal epithelial growth and differentiation in vitro15–17. Monolayer cultures have been complicated by rapid apoptosis18,19, whereas organoid cultures have historically been unable to preserve viability lasting more than ~10 d21. More recently developed 3D organoid cultures have been restricted to embryonic tissue and again have shown restricted viability, typically less than 14 d22. Consequently, implantation of dissociated intestinal material into syngeneic hosts or immuno-compromised mice, either directly or with an intervening culture period, has been the only available tool with which to demonstrate the presence and potential of stem cells and progenitors in cultured intestine17.
To overcome the difficulties described above, we have maintained minced intestinal fragments in a 3D matrix scaffold of type I collagen gel under an air-liquid interface. With growing evidence that the 3D matrix environment has a crucial role in facilitating the behavior of stem cells and tissue morphogenesis, organotypic 3D cultures have great advantages over both conventional two-dimensional cell approaches and animal models31–33. The major features of 3D collagen gel models are derived from their ability to mimic normal tissue organization to induce appropriate polarity of epithelial cells, to induce behavior of fibroblasts including extracellular matrix remodeling and to induce signaling between cell-cell and cell-matrix interactions34,35. Additionally, air-liquid interface methods allow long-term culture of various epithelial cell types via improved oxygenation in vitro36–38. In the current method, substitution of conventional immersed conditions for an air-liquid interface markedly decreases viability, although sporadic short-term sphere formation and growth can be observed. Our results suggest that our 3D culture system enables mouse intestinal fragments to recapitulate intestinal epithelial growth and differentiation within the microenvironment of an in vitro ISC niche.
Adult or somatic stem cells generally have limited function without their niche8. Our method allows primary intestinal epithelium to be cultured in close apposition to myofibroblasts, which have been proposed to be the candidate niche supporting ISCs and influencing intestinal epithelial growth1,9,10. Indeed, the growth of individual intestinal spheres was highly correlated with the preceding growth of myofibroblasts, which lined the basal surface of the cultured intestinal epithelium. Thus, appropriate stromal cell growth seems to permit long-term culture and multilineage differentiation of intestinal epithelial cells without implantation into syngenic or immunocompromised host animals. Our results by no means exclude the possibility of culturing intestinal epithelium without a cellular niche, which could certainly be achieved with appropriate extracellular signals.
Cultures <30 d old were predominantly comprised of viable, robustly proliferating tissue, which was present but increasingly sporadic at extremely long culture durations of >200 d. Although we obtained the most prolonged culture with neonatal intestine, we observed limited culture for 7 to 10 d without RSpo1-Fc treatment and up to 4 weeks with RSpo1-Fc treatment using adult mouse intestine as starting material. These aforementioned limitations of long-term and adult culture may be secondary either to progressive deficits in either the stromal myofibroblasts or other niche components in vitro or to cell-intrinsic deficits in ISCs, transit amplifying cells or both.
Accurate recapitulation of the in vivo Wnt and Notch dependence of the ISC niche represents a prominent feature of the intestinal sphere cultures. Indeed, Dkk1 and dibenzazepine treatment phenocopied the intestinal effects of in vivo Wnt and Notch inhibition on proliferation and goblet cell differentiation, respectively11–13,25. At the same time, the endogenous Wnt and Notch signaling within the intestinal cultures was sufficient to support vigorous expansion, and exogenously added RSpo1-Fc conferred a further induction of intestinal epithelial growth, even allowing limited expansion of adult intestinal cultures. In this regard, RSpo1-Fc induced expansion of transit amplifying cell populations, which express the Wnt target genes Cd44 and Myc in concert with PCNA+ proliferative activity. This preservation of the ISC niche probably underlies the successful support of long-term proliferation and differentiation observed in the current studies. These studies further illustrate the utility of extracellular Wnt agonists, particularly R-spondin-1, as a potent growth factor that markedly enhances the efficacy of intestinal culture.
The observed prolonged growth and differentiation (>350 d) suggest the presence of ISCs or extremely long-lived transit amplifying cells in the sphere cultures. Our detection of both Lgr5+ cells as well as cells derived from the Bmi1+ lineage within the intestinal cultures is consistent with the former possibility. The potential regulation of ISCs by extracellular Wnt signals has been a plausible but previously untested hypothesis. The observed increase in both Lgr5+ lineage– and Bmi1+ lineage–derived cells by RSpo1-Fc treatment in vitro is consistent with the direct regulation of ISCs by Wnt signals, although we can not exclude the alternative possibilities that transit amplifying cells expressing these markers may have arisen in culture, or that RSpo1-Fc–induced Lgr5+ cells merely reflect increased Wnt-dependent gene expression rather than increased ISC number. We observed similar increases in crypt Lgr5+ cells after in vivo adenovirus RSpo1-Fc treatment, demonstrating the predictive nature of the in vitro culture system, although in both cases our in situ hybridization analyses only allow conclusions regarding Lgr5 mRNA and not Lgr5 protein.
The availability of a robust intestinal culture system supporting the expression of populations expressing ISC markers, accurately recapitulating the ISC niche in both small and large intestine and applicable to both neonatal and adult tissues should greatly facilitate the study of intestinal stem cells and niche-ISC interactions. This model could also potentially be used to study intestinal epithelial interactions with other heterologous cell types, including neurons, smooth muscle, endothelial cells and immune cells. Furthermore, the enablement of primary intestinal culture should have widespread application to general studies of intestinal biology, including investigations of physiology, host-pathogen interactions, neoplasia, tissue engineering and regenerative medicine.
We are grateful to members of the Kuo laboratory, C. Chartier, R. Nusse, L. Chia, M. Lee and M. Pech for helpful discussions. We are indebted to M. Amieva for video recording, J. Yuan, P. Chu, H. Ideguchi and S. Nakahara for technical assistance, M. Kay (Stanford University) for adenovirus expressing Ngn3, P. Soriano (Mount Sinai School of Medicine) for Rosa26-1 vector, J. Chen (Stanford University) for Wnt reporter cells, R. Nusse (Stanford University) for Wnt3a L cells, and M. Selsted (University of California–Irvine) and E. Sibley (Stanford University) for antibodies specific for cryptidin and intestinal hydrolases, respectively. A.O. was a California Institute for Regenerative Medicine Scholar. X.L. was supported by a Dean's Fellowship from Stanford University, and H.U. was supported by the Floren Family Fund. This work was supported by US National Institutes of Health grant R01 DK069989-01, California Institute for Regenerative Medicine RS1-00243-1 and the Broad Medical Research Foundation (C.J.K.), US National Institutes of Health grant R01 CA86065-06 and the Smith Family Fund (I.L.W.), the Ichiro Kanehara Foundation, Kato Memorial Trust for Nambyo Research and Nagao Memorial Trust (A.O.), and Grants-in-Aid from the Japanese Ministry of Education, Culture, Sports, Science and Technology for Scientific Research No. 20592023 (S.T.). We acknowledge the generous support of the Stanford University Digestive Disease Center (US National Institutes of Health grant P30 DK56339; H.U. and C.J.K.).
Note: Supplementary information is available on the Nature Medicine website.
COMPETING INTERESTS STATEMENT
The authors declare competing financial interests: details accompany the full-text HTML version of the paper at http://www.nature.com/naturemedicine/.
Reprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions/