生命工学 - 九州大学（kyushu university）...パラチノースガイドブック ver.3.2 16...

生命工学

片倉喜範

特定の組織由来細胞を用いた研究

（腸管）

• 細胞株を用いる。

• プライマリー細胞を用いる。

• 細胞を樹立する。

• Ex vivo実験を行う。

• iPS細胞から分化誘導する。

• オルガノイドカルチャーを行う。

腸管由来細胞株

• 樹立細胞株

– Caco-2

– DLD-1, HCT116, SW480

• プライマリー株

– InEpC（LONZA社）

細胞株を用いる。

腸管由来細胞に特徴的なシグナル

プライマリー細胞を用いる。胎児

縦方向

• 腸管上皮＋pXMS-TAg-IRES-EGFP

細胞を樹立する。

Ex vivo実験を行う。（反転腸管試験）

パラチノースガイドブック Ver.3.2

16

8-2．血糖値上昇抑制メカニズム ①グルコース吸収阻害効果

パラチノースはグルコースに対して、濃度依存的な吸収阻害効果を発揮することがわかっています。

パラチノースがグルコースによる血糖上昇を抑制するメカニズムの解明のため、図 9-aで示してい

るような、ラットの腸管を反転させてグルコースの移行速度を測定する試験が行われています。この

試験の結果、パラチノースのグルコース吸収抑制効果が濃度依存的である（グルコースに対するパラ

チノースの割合が増すほど強くなる）ことがわかりました（図 9-b）27)。

A：　20mMグルコース C: 　20mMグルコース+2.5mMパラチノースB：　20mMグルコース+1.0mMパラチノース D: 　20mMグルコース+5.0mMパラチノース

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

A B C D

遊離

グル

コー

ス濃

度 (g/

L)

＊

図 9-b 小腸におけるパラチノースのグルコース吸収阻害効果

図 9-a 小腸におけるパラチノースのグルコース吸収阻害効果測定試験（概略図）

37に保温したビーカー内に 20 mMグルコースのみの標準バッファー、または 20 mM

グルコースに 1 mM、2.5 mM、5 mMいずれかの濃度のパラチノースを添加した標準バッ

ファーを入れます。これにラットの小腸を裏返したものをチューブに取り付け、内部を

標準バッファーで満たします。すると、時間とともにビーカー側（外液側）から反転腸

管内（つまり、小腸の内側から外側）にグルコースが移行する（吸収される）ため、反

転腸管内のグルコース濃度を経時的に測定することでグルコースの吸収速度と、パラチ

ノース濃度が小腸でのグルコースの吸収速度に及ぼす影響を調査することができます。

iPS cells for novel therapeutic strategies such as patient-specific or-gan transplantation.

Materials and methods

Growth and differentiation of iPS cells. Mouse iPS cell line(20D-17: RIKEN BRC, Tsukuba, Japan) was prepared for our study.This iPS cells was produced from mouse skin (fibroblast cells)by Dr. Shinya Yamanaka (Center for iPS Cell Research and Applica-tion, Institute for Integrated Cell-Material Sciences, Kyoto Univer-sity, Kyoto, Japan) and was carrying Nanog promoter-driven GFP/IRES/puromycin-resistant gene (Nanog-iPS cells) [3,4]. The iPScells were maintained on feeder layers of radiation-treatedmouse embryonic fibroblast (MEF: Millipore; Temecula, CA;http://www.Millipore.com) cells in Dulbecco’s modified Eagle’smedium (DMEM: Wako; Osaka, Japan; http://www.wako-chem.-co.jp/) containing 10% fetal bovine serum (FBS: EmbryoMax

!ES

Cell Qualifield FBS, Millipore), 0.1 mM nonessential amino acids(NEAA: Invitrogen, Carlsbad, CA; http://www.invitrogen.co.jp/),0.1 mM 2-mercaptoethanol (2-ME: Wako), 1 mM sodium pyruvate(Lonza; Walkersville, ML), and 50 U/ml penicillin and 50 lg/mlstreptomycin (Penicillin–Streptmycin liquid: Invitrogen) supple-mented with 1000 U/ml of leukemia inhibitory factor (LIF: ESGRO

!;

Millipore). To eliminate contaminating MEF cells, maintained iPScells were dissociated with 0.25% trypsin/EDTA (Mediatech;Manassas, VA; http://www.cellgro.com) and passaged twice (twodays for first passage and one day for second passage) on gelatin-coated dishes without feeder cells. To induce embryoid body (EB)formation, iPS cells were dissociated to a single cell suspensionwith 0.05% trypsin/EDTA (Mediatech), and then cultured in hang-ing drops, as previously described [15,16]. Briefly, the mediumfor EB formation is the same medium as for iPS-cell maintainedculture in the absence of LIF. The cell density of one drop was500 iPS cells per 15 ll of medium. After 6-days in a hanging dropculture, iPS cells aggregate and form a sphere structure, so calledEB. The resulting EBs were plated onto plastic 100-mm gelatin-coated dishes in the same medium as for EB formation and allowedto attach for the outgrowth culture. Cultures were maintained in ahumidified chamber in a 5% CO2/air mixture at 37 "C.

Immunohistochemistry. For immunohistochemical detection ofc-Kit, a transmembrane receptor that has tyrosine kinase activity,whole-mount preparations of iGut were fixed in acetone (4 "C,5 min), and for detection of smooth muscle actin and neurofila-ment (NF), the preparations were fixed in 4% paraformaldehyde(4 "C, 10 min). After fixation, the preparations were washed for30 min in phosphate-buffered saline (PBS) (0.01 M, pH 7.4), andthen incubated for 3 h at room temperature in 10% normal goatserum (Dako; Carpinteria, CA; http://www.dako.com) in PBS con-taining 0.3% (vol/vol) Triton X-100 (Calbiochem; san Diego, CA;http://www.emdbiosciences.com) to reduce non-specific antibodybinding. Tissues were then incubated at 4 "C for 1 day with a ratmonoclonal antibody raised against c-Kit protein (ACK2: 5 lg/mlin PBS; eBioscience, San Diego, CA; http://www.ebioscience.com),a rabbit monoclonal antibody against a-smooth muscle actin(1:250; Epitomics; Burlingame, CA; http://www.epitomics.com)and a rabbit polyclonal antiserum cocktail against primate andbovine low-(68–70 kDa), medium-(150 kDa), and high-molecular-weight NF (200–210 kDa) (1:500; Affiniti Research Products;Devon, UK; http://www.affiniti-res.com). This antiserum cocktailreacts with neuronal cell bodies, dendrites, and thick and thin ax-ons. The c-Kit immunoreactivity was detected using an Alexa Flour594-conjugated secondary antibody (1:200; Alexa Flour 594 goatanti-rat: Molecular Probes; Eugene, OR; http://probes.invitro-gen.com) for 2 h in the dark at room temperature, whereas smoothmuscle actin and NF immunoreactivity was detected using a Texas

Red-conjugated secondary antibody (1:100; Texas Red goat anti-rabbit: Invitrogen) for 1 h in the dark at room temperature. Tissueswere then examined using a laser confocal microscope (FV1000:Olympus; Tokyo, Japan; http://www.olympus.co.jp), which yieldedconfocal micrographs that were digital composites of Z-seriesscans of 10–15 optical sections through a depth of 10–20 lm. Finalimages were constructed using software (FV10-ASW Version 1.7:Olympus).

Electron microscopy. The iGut were fixed with 2% glutaraldehydein a 0.1 M phosphate buffer (pH 7.4) at 4 "C. They were post-fixedwith 2% OsO4 in a 0.1 M phosphate buffer, and subsequentlystained with uranyl acetate in a 0.05 M maleic acid buffer. Then,the specimens were dehydrated in a graded ethanol and embeddedin the epoxy resin. Ultrathin sections stained with uranyl acetateand modified Sato’s lead solution were submitted to TEM observa-tion (H-7600: Hitachi; Brisbane, CA; http://www.hitachi.com).

Results

iPS cell differentiation and mechanical activity

iPS cells were cultured for 6 days in a hanging drop culture sys-tem and allowed to form spherical multicellular aggregates,embryoid bodies (EBs). After 3–5 days in subsequent outgrowthculture, various types of cell clusters, including cardiac muscle-likebeating cell populations, emerged from the outgrowths of the at-tached EBs, as shown in video format (see Nara Medical Universityhome page http://www.naramed-u.ac.jp/~1sur/iHeart2.wmv). Onapproximately Day 7, the gut-like structure, iGut, which was differ-ent from the beating cells, began to elevate on the outgrowth ofEBs and contract spontaneously with an irregular rhythm(Fig. 1A). On approximately Day 10, the contracting iGut under-went a morphological transformation, which showed different sizeand shape including dome-like type and tubular type. The dome-like type iGut showed a hemispherical structure with a cavity thatcontained fluid and solids (see http://www.naramed-u.ac.jp/~1sur/iGut2.wmv) (Fig. 1B). The tubular type iGut proliferated to formmore prominent three-dimensional tubular structures with lumenthat extended its longitudinal axis in parallel to the dish bottom(Fig. 1C). On approximately Day 14, the iGut began rhythmic con-tractions (Fig. 1D). On approximately Day 21, the iGut showed dis-tinct and highly coordinated contraction patterns with regularrhythms. This mechanical activity was composed of periodic con-traction and relaxation (see http://www.naramed-u.ac.jp/~1sur/iGut1.wmv) (Fig. 1E–H). It was very similar to GI motility, i.e., peri-stalsis. The peristalsis-like contractions were accompanied by atransportation of contents (see http://www.naramed-u.ac.jp/~1sur/iGut3.wmv) (Fig. 1I–L). The iGut with peristalsis-like con-tractions exhibited the periodic movements of back and force inthe closed lumen (see http://www.naramed-u.ac.jp/~1sur/iGut4.wmv). The frequency of contraction was 6.9 ± 0.3 cyclesper minutes at 37 "C (n = 14 iGut). The frequency was stronglydependent on the temperature and decreased to 3.2 ± 0.5 cyclesper minutes at 20 "C (n = 14 iGut), which is one of the characteris-tics of mature GI tract.

Immunohistochemistry

Immunohistochemistry was used to identify the cellular com-ponents of the contracting iGut. Immunoreactivity for smoothmuscle actin was observed in the wall surrounding the lumenand spreading out from the iGut (Fig. 2A). Immunoreactivity forc-Kit, a marker for ICC [17], was observed in the iGut that showedspontaneous contractions. The c-Kit immunopositive (c-Kit+) cellswere mostly multipolar, and formed a distinct and dense network

T. Ueda et al. / Biochemical and Biophysical Research Communications 391 (2010) 38–42 39

iPS細胞から分化誘導する。(iGutの作製）

Generation of functional gut-like organ from mouse induced pluripotentstem cells q

Takeshi Ueda a, Takatsugu Yamada a,*, Daisuke Hokuto a, Fumikazu Koyama a, Shogo Kasuda b,Hiromichi Kanehiro a, Yoshiyuki Nakajima a

a Department of Surgery, Nara Medical University, 840 Shijo-cho, Kashihara, Nara 634-8522, Japanb Department of Legal Medicine, Nara Medical University, 840 Shijo-cho, Kashihara, Nara 634-8522, Japan

a r t i c l e i n f o

Article history:Received 27 October 2009Available online 4 November 2009

Keywords:Induced pluripotent stem (iPS) cellsInduced gut (iGut)Organ regenerationMotor functionPeristalsisHanging drop culture

a b s t r a c t

Induced pluripotent stem (iPS) cells have the pluripotency to differentiate into broad spectrum deriva-tives of all three embryonic germ layers. However, the in vitro organ differentiation potential of iPS cellsto organize a complex and functional ‘‘organ” has not yet been demonstrated. Here, we demonstrate thatmouse iPS cells have the ability to organize a gut-like organ with motor function in vitro by a hangingdrop culture system. This ‘‘induced gut (iGut)” exhibited spontaneous contraction and highly coordinatedperistalsis accompanied by a transportation of contents. Ultrastructural analysis identified that the iGuthad large lumens surrounded by three distinct layers (epithelium, connective tissue and musculature).Immunoreactivity for c-Kit, a marker of interstitial cells of Cajal (ICCs, enteric pacemaker cells), wasobserved in the wall of the lumen and formed a distinct and dense network. The neurofilament immuno-reactivity was identified to form large ganglion-like structures and dense neuronal networks. The iGutwas composed of all the enteric components of three germ layers: epithelial cells (endoderm), smoothmuscle cells (mesoderm), ICCs (mesoderm), and enteric neurons (ectoderm). This is the first report todemonstrate the in vitro differentiation potential of iPS cells into particular types of functional ‘‘organs.”This work not only contributes to understanding the mechanisms of incurable gut disease through dis-ease-specific iPS cells, but also facilitates the clinical application of patient-specific iPS cells for noveltherapeutic strategies such as patient-specific ‘‘organ” regenerative medicine in the future.

! 2009 Elsevier Inc. All rights reserved.

Introduction

Induced pluripotent stem (iPS) cells have been establishedfrom mouse and human adult somatic cell cultures throughreprogramming by transduction of four defined transcription fac-tors (Oct3/4, Sox2, Klf4, and c-Myc) [1–4]. The iPS cells have thepluripotency to differentiate into broad spectrum derivatives ofall three embryonic germ layers. Recent studies have reportedthat mouse and human iPS cells can differentiate into particulartypes of ‘‘cells,” such as hematopoietic cells (dendritic cells andmacrophages), cardiomyocytes, vascular cells, motor neurons,and insulin-producing cells under appropriate conditions [5–14].However, the pluripotent ability of iPS cells to organize a complex

and functional ‘‘organ,” which is composed of a variety of celltypes in an orderly manner and exhibits systematic function,has not yet been demonstrated.

To induce an in vitro organ differentiation potential of iPS cells,we used a three-dimensional hanging drop culture system to formembryoid bodies (EBs) [15,16]. In the present study, we demon-strate that mouse iPS cells have the ability to organize a gut-likeorgan with motor function in vitro. This functional gut-like organinduced from mouse iPS cells, which we designated ‘‘iGut (inducedgut),” is organized by enteric components of all three germ layers,including epithelial cells (endoderm), smooth muscle cells (meso-derm), interstitial cells of Cajal (ICCs) (mesoderm), and entericneurons (ectoderm). Notably, iGut represents a three-dimensionalstructure with lumen and exhibits mechanical activity such asspontaneous contraction and highly coordinated peristalsis accom-panied by transportation of contents. This is the first report todemonstrate the in vitro differentiation potential of iPS cells intoparticular types of ‘‘organs,” which consist of derivatives of allthree embryonic germ layers. This work not only contributes toour understanding of the mechanisms of gut development and dis-ease, but also facilitates the clinical application of patient-specific

0006-291X/$ - see front matter ! 2009 Elsevier Inc. All rights reserved.doi:10.1016/j.bbrc.2009.10.157

Abbreviations: EB, embryoid body; ES cell, embryonic stem cell; ICCs, interstitialcells of Cajal; iGut, induced gut; iPS cell, induced pluripotent stem cell; LIF,leukemia inhibitory factor; NF, neurofilament.

q T.U., T.Y. and D.H.: study concept and design, acquisition of data, analysis andinterpretation of data; F.K., S.K., and H.K.: technical support; N.Y.: study supervi-sion.

* Corresponding author. Fax: +81 744 24 6866.E-mail address: [email protected] (T. Yamada).

Biochemical and Biophysical Research Communications 391 (2010) 38–42

Contents lists available at ScienceDirect

Biochemical and Biophysical Research Communications

journal homepage: www.elsevier .com/locate /ybbrc

(Fig. 2B). Immunoreactivity for neurofilament (NF) was detected inthe iGut that exhibited peristalsis-like contractions. The NF immu-nopositive (NF+) cells were mostly multipolar with prominentdendrites and axon-like processes (Fig. 2C), and formed large

ganglion-like structures (Fig. 2D). NF+ cells were distributed notonly outside the iGut but also in the wall, and NF+ fibers from gan-glions located close to the iGut were often seen projecting into thecontracting iGut (Fig. 2E).

Fig. 1. The iGut began to elevate on the outgrowth of EBs and contract spontaneously with an irregular rhythm after 7 days in outgrowth culture (A). Dome-like type iGut onDay 10. They showed a hemispherical dome-like structure with a cavity that contained fluid and solids (B). Tubular type iGut on Day 10. They proliferated to form moreprominent three-dimensional tubular structures with lumen that extended its longitudinal axis in parallel to the dish bottom (C). iGut on Day 14. They began rhythmiccontractions (D). iGut on Day 21. They showed distinct and highly coordinated peristalsis-like contraction patterns with regular rhythms (E–H) This mechanical activity wascomposed of periodic contraction (E, F) and relaxation (G, H). The peristalsis-like contractions were accompanied by transportation of contents (arrowhead) (I–L). Scale barsrepresent the following sizes: 300 lm, (A); 500 lm, (B–J); and 200 lm, (K, L).

Fig. 2. Immunohistochemistry for iGut.: The contracting iGut showing a-smooth muscle actin immunoreactivity in the wall of the iGut surrounding the lumen and someimmunopositive cells spreading out from the iGut (A). A large number of c-Kit+ cells were identified in the wall of the lumen, and formed a distinct and dense network. The c-Kit+ cells were mainly multipolar with prominent dendrites (B). NF+ cells located in the wall of iGut. They were mostly multipolar with prominent dendrites and axon-likeprocesses (C). NF+ cells formed large ganglion-like structures outside the iGut (D). The neurofilament immunopositive (NF+) cells were observed in the wall and outside theiGut, which exhibited peristalsis-like contractions. They formed large ganglion-like structures inside the wall and outside the iGut (arrowhead). The NF fibers from ganglionslocated close to the iGut were often seen projecting into the contracting iGut (broken line) (E). Scale bars represent the following sizes: 200 lm, (A); 80 lm, (B); 30 lm, (C);30 lm, (D) and 100 lm, (E).

40 T. Ueda et al. / Biochemical and Biophysical Research Communications 391 (2010) 38–42

Ultrastructural characteristics of iGut

Ultrastructural analysis confirmed the cellular components ofthe iGut. The wall of the iGut was composed of three distinct lay-ers: epithelium (EP), connective tissue (CT), and muscle layer (ML)(Fig. 3A). The innermost was a flat and single layer of epithelium,which did not form any plica, intestinal villus, or crypt. Most ofthe epithelial cells were columnar type, with fewer and shortermicrovilli than those in epithelia in the mouse GI tract (Fig. 3B).Well-developed goblet cells and tuft cells were observed amongthe columnar cells (Fig. 3C and D). The second layer was connectivetissue composed of fibroblasts and collagen fibers. Both blood andlymphatic vessels were absent. The outermost layer was composedof prominent smooth muscle cells, filled with myofilaments, cave-olae, and dense bodies in their cytoplasm (Fig. 3E, 3F). The outersurface of the iGut was lined with a flat and thin monolayer, sim-ilar to the serosa.

Discussion

In the current study, we succeeded in inducing differentiation ofmouse iPS cells into a functional gut-like organ from embryoidbodies (EBs). This functional gut-like organ induced from iPS cells,which we designated ‘‘iGut (induced gut),” was composed of theenteric derivatives of all three embryonic germ layers: epithelialcells (endoderm), smooth muscle cells and ICCs (mesoderm), andenteric neurons (ectoderm), and exhibited mechanical activitysuch as spontaneous contraction and peristalsis-like motion. TheiPS cells are identified to have pluripotency by ES cell-marker geneexpression, ES cell-specific surface antigens and teratoma forma-tion after subcutaneous transplantation into immune-deficient

mice [1–3]. However, the iPS cells are not completely identical toES cells in the global gene-expression patterns and DNA methyla-tion status [1–3]. Although the iPS cells were confirmed to differ-entiate into all three germ layers in teratomas, including neuraltissue, cartilage, smooth muscle, adipose and gut-like epithelium,the differentiation potential of iPS cells to organize a complexand functional ‘‘organ” has not been demonstrated [1–4]. This isthe first report to demonstrate the in vitro differentiation potentialof iPS cells into particular types of ‘‘organs,” which not only consistof a variety of derivatives of all three germ layers but also exhibitsystematic functions.

We previously established an in vitro organ differentiation sys-tem that can reproduce a contracting gut-like organ from mouse EScells by a three-dimensional hanging drop culture [15,16]. We ap-plied this system to iPS cells and investigated the organ differenti-ation properties of iPS cells into a gut-like organ. Ultrastructuralanalysis revealed that the iGut had large lumens surrounded bythree layers (i.e., epithelium, connective tissue and musculature).Immunoreactivity for c-Kit, a marker for ICC (enteric pacemakercell), was distributed to form a distinct and dense network in thewall of the iGut that showed spontaneous contractions. Thisnetwork structure was similar to that of ICC at the level of themyenteric plexus in a murine embryo or neonate [18]. Notably,the neurofilament (NF) immunopositive cells formed large gan-glion-like structures and dense neuronal networks not only inthe wall of the lumen but also outside the iGut showing highlycoordinated peristalsis, suggesting that innervations to the iGutgoverns the peristalsis-like contractions. Interestingly, the neuro-nal fibers from ganglions located close to the iGut were often seenprojecting into the wall of the iGut. These findings raise the possi-bility that the in vitro neuronal differentiation process in the iGut

Fig. 3. Electron micrographs of contracting iGut.: A cross section of the wall clearly showing three layers: EP (epithelium), CT (connective tissue), and ML (muscle layer) (A).Most of the epithelial cells were columnar type, with fewer and shorter microvilli than those in epithelia in the mouse GI tract (B). A goblet cell with many mucous globules inthe apical cytoplasm is prominent in the EP (C). Tuft cells were observed among columnar cells in the EP, and were characterized by microvilli with long bundles of straightfilaments extending from the core of the microvilli and small vesicles in the apical cytoplasm (D). Smooth muscle cells in the muscle layer showing typical features offilament-like structures (i.e., thick, thin, and intermediate filaments), and dense bodies (arrowheads) (E). Smooth muscle cells filled with myofilaments, caveolae (arrows) anddense bodies (arrowheads) in their cytoplasm (F). Scale bars represent the following sizes: 6 lm, (A); 1 lm, (B); 3 lm, (C); 2 lm, (D); 1 lm, (E); and 500 nm, (F).

T. Ueda et al. / Biochemical and Biophysical Research Communications 391 (2010) 38–42 41

www.nature.com/scientificreports/

2Scientific RepoRts | 5:16479 | DOI: 10.1038/srep16479

of drug absorption after oral administration, the small intestine plays a major role in drug absorption and metabolism2. Specifically, cytochrome P450 3A4 (CYP3A4, the dominant drug metabolizing enzyme in the human small intestine), peptide transporter 1 (PEPT1), and P-glycoprotein (P-gp) are expressed at high levels in the enterocytes3,4. Because CYP3A4 interacts with absorbed drugs in the enterocytes, CYP3A4-mediated intestinal metabolism is a significant factor in oral drug bioavailability5,6. In addition, CYP3A4 and P-gp share not only many substrates (e.g., dexamethasone and etoposide) and inhibitors (e.g., quinidine and testosterone), but also inducers such as rifampicin7. Therefore, a model that could evaluate both drug absorption and metabolism would greatly facilitate the development of safer and more effective drugs.

Animal models are used for evaluation of drug absorption in the small intestine. However, it is known that there are species differences in small intestinal drug absorption and the first-pass effect. Because human primary enterocytes cannot be obtained in large numbers, Caco-2 cells (a human colorectal car-cinoma cell line) monolayers are widely utilized for drug absorption studies in human small intestine8–10. Polarized Caco-2 cell monolayers can form a physical and biochemical barrier that reproduces the enterocyte barrier of the human small intestine. Although the Caco-2 cell monolayer is a useful model, it does have some significant drawbacks. First, it is difficult to accurately evaluate CYP3A4-mediated drug metabolism and the CYP3A4 induction potency of drugs because the CYP3A4 expression levels in Caco-2 cells are much lower than those in the enterocytes11,12. Consequently, it is difficult to evaluate both drug absorption and metabolism by using the Caco-2 cell monolayer model. Secondly, the perme-ability level of hydrophilic drugs, which are absorbed via the paracellular route, in Caco-2 cells is lower than that in the human small intestine10,13. Therefore, it is also difficult to evaluate the hydrophilic drug absorption by using the Caco-2 cell monolayer model. There is thus, need of a novel model to resolve these issues.

Human induced pluripotent stem (iPS) cells14 have the potential to self-replicate and differentiate into multiple types of body cells, including enterocytes. In this study, we aimed to generate enterocyte-like cells from human iPS cells (hiPS-ELCs) which could evaluate both drug absorption and metabolism. Recently, some groups have reported that intestinal tissues and intestinal organoids, which are con-sist of all four intestinal cell types (paneth cells, goblet cells, enterocytes, and enteroendocrine cells), could be differentiated from human pluripotent stem cells in vitro15–18. However, it might be difficult to apply the intestinal tissues to drug absorption and metabolism studies because they are generated under 3-dimensional culture conditions. Although some researchers have demonstrated that enterocyte-like cells could be generated under 2-dimensional culture conditions, their drug absorption and metabo-lism capacities have not been sufficiently characterized19–21. Therefore, we decided to generate enterocyte monolayers, which have drug absorption and metabolism capacities, from human iPS cells.

Materials and MethodsReagents for compound screening. PMA was purchased from BIAFFIN GMBH & CO KG. TGFβ 1, EGF, Follistatin, and R-spondin 1 were purchased from R&D systems. Wortmannin and Pentagastrin were purchased from Sigma. Noggin was purchased from PeproTech. SB431542 was obtained from Wako. To generate Wnt3A-conditional medium, L-Wnt-3A cells (ATCC, CRL2647) were cultured with the DMEM-High Glucose medium (Invitrogen), which contains 10% Knockout Serum Replacement, 1% Non-Essential Amino Acid Solution, Penicillin-Streptomycin, 2 mM L-Glutamine, and 100 µ M β -mer-captoethanol, for 24 hr, and then the conditioned medium was collected.

Human iPS cells culture. A human iPS cell line, Tic22,23 (provided by Dr. A. Umezawa, National Center for Child Health and Development), was maintained on a feeder layer of mitomycin C-treated mouse embryonic fibroblasts (Millipore) with ReproStem medium (ReproCELL) supplemented with 10 ng/ml fibroblast growth factor 2 (FGF2, KATAYAMA CHEMICAL INDUSTRIES).

In vitro differentiation. Before the initiation of enterocyte differentiation, human iPS cells were dissociated into clumps by using dispase (Roche) and plated onto BD Matrigel Basement Membrane Matrix (BD Biosciences). These cells were cultured in the MEF-conditioned medium for 2–3 days. The differentiation protocol for the induction of definitive endoderm cells was described previ-ously24. Briefly, for the definitive endoderm differentiation, human iPS cells were cultured for 4 days in L-Wnt3A-expressing cell-conditioned RPMI1640 medium (Sigma) containing 100 ng/ml Activin A (R&D Systems), 4 mM L-Glutamine, 0.2~0.5% FBS, and 1 × B27 Supplement Minus Vitamin A (Life Technologies). For the induction of intestine-like cells, the definitive endoderm cells were cultured for 15 days in DMEM-High Glucose medium (Invitrogen) containing 5 µ M 6-Bromoindirubin-3′ -oxime (BIO; Calbiochem), 10 µ M N-[(3,5-difluorophenyl) acetyl]-L-alanyl-2-phenyl-1, 1-dimethylethyl ester-glycine (DAPT; Peptide Institute), 10% Knockout Serum Replacement (Invitrogen), 1% Non-Essential Amino Acid Solution (Invitrogen), Penicillin-Streptomycin, 2 mM L-Glutamine, and 100 µ M β -mercaptoethanol. For the induction of hiPS-ELCs, the intestine-like cells were cultured for 15 days in L-Wnt3A-expressing cell-conditioned DMEM-High Glucose medium (Invitrogen) containing 5 µ M BIO, 10 µ M DAPT, 10 µ M SB431542 (Wako), 250 ng/ml EGF (R&D systems), 10% Knockout Serum Replacement, 1% Non-Essential Amino Acid Solution, Penicillin-Streptomycin, 2 mM L-Glutamine, and 100 µ M β -mercaptoethanol.


www.nature.com/scientificreports

Generation of enterocyte-like cells

from human induced pluripotent

stem cells for drug absorption

and metabolism studies in human

small intestine

Tatsuya Ozawa1,2,*, Kazuo Takayama1,2,3,*, Ryota Okamoto1,2, Ryosuke Negoro1,

Fuminori Sakurai1,4, Masashi Tachibana1, Kenji Kawabata5,6 & Hiroyuki Mizuguchi1,2,3,7

Enterocytes play an important role in drug absorption and metabolism. However, a widely used

enterocyte model, Caco-2 cell, has difficulty in evaluating both drug absorption and metabolism because the expression levels of some drug absorption and metabolism-related genes in these cells

differ largely from those of human enterocytes. Therefore, we decided to generate the enterocyte-like cells from human induced pluripotent stem (iPS) cells (hiPS-ELCs), which are applicable to drug

absorption and metabolism studies. The efficiency of enterocyte differentiation from human iPS cells was significantly improved by using EGF, SB431542, and Wnt3A, and extending the differentiation period. The gene expression levels of cytochrome P450 3A4 (CYP3A4) and peptide transporter 1 in the

hiPS-ELCs were higher than those in Caco-2 cells. In addition, CYP3A4 expression in the hiPS-ELCs was induced by treatment with 1, 25-dihydroxyvitamin D3 or rifampicin, which are known to induce CYP3A4 expression, indicating that the hiPS-ELCs have CYP3A4 induction potency. Moreover, the transendothelial electrical resistance (TEER) value of the hiPS-ELC monolayer was approximately

240 Ω*cm2, suggesting that the hiPS-ELC monolayer could form a barrier. In conclusion, we succeeded in establishing an enterocyte model from human iPS cells which have potential to be

applied for drug absorption and metabolism studies.

It is known that poor pharmacokinetics and poor bioavailability are responsible for approximately 10% of drug withdrawal1. Various organs, including the small intestine, play an important role in pharma-cokinetics (absorption, distribution, metabolism, and excretion) and bioavailability. Because the drug transporters and metabolism enzymes are strongly expressed in enterocytes, which are the primary site

1Laboratory of Biochemistry and Molecular Biology, Graduate School of Pharmaceutical Sciences, Osaka ni ersit Osa a 565 0871 apan. a orator of epatoc te e u ation ationa Institute of iome ica

Inno ation ea t an utrition Osa a 567 0085 apan. 3i e ase esearc ro ect on epatic o icit an eta o ism ra uate c oo of armaceutica ciences Osa a ni ersit Osa a 565 0871 apan. 4Laboratory

of e u ator ciences for O i onuc eoti e erapeutics inica Dru De e opment ro ect ra uate c oo of armaceutica ciences Osa a ni ersit Osa a 565 0871 apan. 5 a orator of tem e e u ation ationa

Institute of iome ica Inno ation ea t an utrition Osa a 567 0085 apan. 6Laboratory of Biomedical Inno ation ra uate c oo of armaceutica ciences Osa a ni ersit Osa a 565 0871 apan. 7Global Center for e ica n ineerin an Informatics Osa a ni ersit Osa a 565 0871 apan. * ese aut ors contri ute e ua to t is wor . orrespon ence an re uests for materia s s ou e a resse to . . emai : mi u ucp s.osa a u.ac. p

ecei e : 7 arc 015

accepte : 14 Octo er 015

Pu is e : 1 o em er 015

OPEN



Figure 1. Promotion of enterocyte differentiation by combination treatment with three compounds and differentiation period extension. (A) The procedure for enterocyte differentiation from human iPS cells by treatment of compounds is presented. From day 19 to 24, the human iPS-derived intestinal cells were treated with the test compounds. (B) The gene expression levels of the enterocyte marker ANPEP in the test compound-treated human iPS-derived intestinal cells were measured by real-time RT-PCR analysis on day 24. On the y axis, the gene expression levels in “Control (untreated hiPS-ELCs)” were taken as 1.0. (C) On day 24, the gene expression levels of the enterocyte marker VILLIN in the PMA, Wortmannin, SB431542, EGF or Wnt3A-treated human iPS-derived intestinal cells were measured by real-time RT-PCR analysis. On the y axis, the gene expression levels in “Control” were taken as 1.0. (D) Temporal gene expression levels of ANPEP in the human iPS cell-derived intestinal cells (day 24, 29, and 34) were measured by real-time RT-PCR analysis. On the y axis, the gene expression levels in Adult Intestine were taken as 1.0. (E) The modified enterocyte differentiation protocol is illustrated. (F) A morphological image of human iPS-derived enterocyte-like cells is represented. Scale bar represents 100 µ m. (G) Human iPS cell-derived enterocyte-like cells were assayed for the expression of intestinal marker CDX2 (Red) by immunohistochemistry. Nuclei were stained with DAPI (Blue). Scale bar represents 40 µ m. (H) Percentages of VILLIN-positive cells in the SB431542, EGF, and Wnt3A-treated enterocyte-like cells were analyzed by flow cytometry analysis on day 24 and 34. Data are represented as the means ± S.E. (n ≧ 3). Statistical analysis was performed using the unpaired two-tailed student’s t-test. *P < 0.05.



Figure 1. Promotion of enterocyte differentiation by combination treatment with three compounds and differentiation period extension. (A) The procedure for enterocyte differentiation from human iPS cells by treatment of compounds is presented. From day 19 to 24, the human iPS-derived intestinal cells were treated with the test compounds. (B) The gene expression levels of the enterocyte marker ANPEP in the test compound-treated human iPS-derived intestinal cells were measured by real-time RT-PCR analysis on day 24. On the y axis, the gene expression levels in “Control (untreated hiPS-ELCs)” were taken as 1.0. (C) On day 24, the gene expression levels of the enterocyte marker VILLIN in the PMA, Wortmannin, SB431542, EGF or Wnt3A-treated human iPS-derived intestinal cells were measured by real-time RT-PCR analysis. On the y axis, the gene expression levels in “Control” were taken as 1.0. (D) Temporal gene expression levels of ANPEP in the human iPS cell-derived intestinal cells (day 24, 29, and 34) were measured by real-time RT-PCR analysis. On the y axis, the gene expression levels in Adult Intestine were taken as 1.0. (E) The modified enterocyte differentiation protocol is illustrated. (F) A morphological image of human iPS-derived enterocyte-like cells is represented. Scale bar represents 100 µ m. (G) Human iPS cell-derived enterocyte-like cells were assayed for the expression of intestinal marker CDX2 (Red) by immunohistochemistry. Nuclei were stained with DAPI (Blue). Scale bar represents 40 µ m. (H) Percentages of VILLIN-positive cells in the SB431542, EGF, and Wnt3A-treated enterocyte-like cells were analyzed by flow cytometry analysis on day 24 and 34. Data are represented as the means ± S.E. (n ≧ 3). Statistical analysis was performed using the unpaired two-tailed student’s t-test. *P < 0.05.



Figure 2. Expression analyses of intestinal transporters in the human iPS-derived enterocyte-like cells. Human iPS-derived enterocyte-like cells (hiPS-ELCs) were differentiated according to the protocol described in Fig. 1E. (A) The gene expression levels of PEPT1 in Caco-2 cells, hiPS-ELCs, and Adult Intestine were measured by real-time RT-PCR analysis. On the y axis, the gene expression levels in Caco-2 cells were taken as 1.0. (B) The hiPS-ELCs were assayed for expression of PEPT1 (Red) by immunohistochemistry. Nuclei were stained with DAPI (Blue). Scale bars represent 40 µ m. (C,D) The gene expression levels of apical transporters (C) and basolateral transporters (D) in the Caco-2 cells, hiPS-ELCs, and Adult Intestine were measured by real-time RT-PCR analysis. On the y axis, the gene expression levels in Adult Intestine were taken as 1.0. Data are represented as the means ± S.E. (n ≧ 3). Statistical analysis was performed using the unpaired two-tailed student’s t-test. *P < 0.05. (E) After the enterocyte differentiation, the hiPS-ELCs were treated with or without 100 µ M captopril for 24 hr. The hiPS-ELCs were treated with D-Ala-Leu-Lys-AMCA (blue) for 4 hr. After the uptake of D-Ala-Leu-Lys-AMCA, the cells were fixed, and stained with anti-VILLIN antibodies (green). Scale bars represent 50 µ m.



Figure 2. Expression analyses of intestinal transporters in the human iPS-derived enterocyte-like cells. Human iPS-derived enterocyte-like cells (hiPS-ELCs) were differentiated according to the protocol described in Fig. 1E. (A) The gene expression levels of PEPT1 in Caco-2 cells, hiPS-ELCs, and Adult Intestine were measured by real-time RT-PCR analysis. On the y axis, the gene expression levels in Caco-2 cells were taken as 1.0. (B) The hiPS-ELCs were assayed for expression of PEPT1 (Red) by immunohistochemistry. Nuclei were stained with DAPI (Blue). Scale bars represent 40 µ m. (C,D) The gene expression levels of apical transporters (C) and basolateral transporters (D) in the Caco-2 cells, hiPS-ELCs, and Adult Intestine were measured by real-time RT-PCR analysis. On the y axis, the gene expression levels in Adult Intestine were taken as 1.0. Data are represented as the means ± S.E. (n ≧ 3). Statistical analysis was performed using the unpaired two-tailed student’s t-test. *P < 0.05. (E) After the enterocyte differentiation, the hiPS-ELCs were treated with or without 100 µ M captopril for 24 hr. The hiPS-ELCs were treated with D-Ala-Leu-Lys-AMCA (blue) for 4 hr. After the uptake of D-Ala-Leu-Lys-AMCA, the cells were fixed, and stained with anti-VILLIN antibodies (green). Scale bars represent 50 µ m.



and protein expression levels of ZO-1 in the hiPS-ELCs were measured by real-time RT-PCR and immu-nohistochemical analysis (Fig. 4A,B, respectively). The gene expression levels of ZO-1 in the hiPS-ELCs were similar to those in the human adult intestine, and were lower than those in Caco-2 cells. In addi-tion, both the hiPS-ELCs and Caco-2 cells were positive for ZO-1. Barrier function in the hiPS-ELC and Caco-2 cell monolayers was analyzed by TEER measurements and FD-4 permeability tests (Fig. 4C and S6, respectively). The TEER values in the hiPS-ELC and Caco-2 cell monolayers were approximately 240

Figure 3. CYP3A4 expression level and induction potency in the human iPS-derived enterocyte-like cells. Human iPS cell-derived enterocyte-like cells (hiPS-ELCs) were differentiated according to the protocol described in Fig. 1E. (A) The gene expression levels of CYP3A4 in Caco-2 cells, hiPS-ELCs, and Adult Intestine were measured by real-time RT-PCR analysis. On the y axis, the gene expression level in Caco-2 cells was taken as 1.0. (B) The CYP3A4 protein expression levels in the Caco-2 cells and hiPS-ELCs were measured by western blotting analysis. (C) The CYP3A4 activity levels in the Caco-2 cells and hiPS-ELCs were measured by CYP3A4-Glo assay kit. (D,E) The CYP3A4 induction potency was examined in the hiPS-ELCs and Caco-2 cells. The hiPS-ELCs and Caco-2 cells were treated with 100 nM 1,25-dihidroxyvitamin D3 (VD3) (D) or 20 µ M rifampicin (RIF) (E) for 24 hr or 48 hr, respectively, and then the gene expression levels of CYP3A4 were measured by real-time RT-PCR analysis. On the y axis, the gene expression levels of CYP3A4 in Caco-2 cells treated with DMSO (Solvent) were taken as 1.0. (F,G) The gene expression levels of VDR (F) and PXR (G) in the hiPS-ELCs and Caco-2 cells were examined by real-time RT-PCR analysis. On the y axis, the gene expression levels of VDR and PXR in Caco-2 cells were taken as 1.0. The data are represented as the means ± S.E. (n ≧ 3).



and 400 Ω *cm2, respectively. Moreover, the hiPS-ELC monolayers had higher FD-4 permeability than Caco-2 cells, but lower than undifferentiated human iPS cells. These results suggest that the hiPS-ELC monolayers have weaker barrier function than the Caco-2 cell monolayers. Importantly, it is known that the TEER values in Caco-2 cell monolayers are higher than those in the small intestine13,36. Therefore, the hiPS-ELC monolayers might be a more suitable in vitro model for evaluating the absorption of hydro-philic drugs than the Caco-2 cell monolayers.

DiscussionThe main purpose of this study was to generate hiPS-ELCs that could be used for drug absorption and metabolism studies. The efficiency of enterocyte differentiation from human iPS cells was significantly improved by using EGF, SB431542, and Wnt3A treatment, and by the extension of differentiation period (Fig. 1). The hiPS-ELCs expressed CYP3A4 and PEPT1 at higher levels than Caco-2 cells (Figs 2 and 3, respectively). In addition, the hiPS-ELCs showed CYP3A4 and P-gp induction potency (Fig. 3 and S3, respectively) and barrier forming potency (Fig. 4).

We identified that treatment with the combination of EGF, SB431542 (TGFβ signal inhibitor) and Wnt3A treatments could enhance the enterocyte differentiation efficiency (Fig. 1). At the end of differen-tiation (day 34), the LGR5 expression was maintained at a high level (Fig. S3B), and approximately 80% of VILLIN-positive hiPS-ELCs were proliferative (Fig. S3D). Because it is known that EGF and Wnt3A are essential for maintenance of the self-renewing intestinal organoids28,37,38, EGF and Wnt3A treatments might support the proliferation of the human iPS-derived intestinal stem/progenitor cells. Therefore, EGF and Wnt3A treatment might have increased the enterocyte differentiation efficiency by enhancing the proliferation of human iPS-derived intestinal stem/progenitor cells. In addition, the removal of SB431542 resulted in a significant decrease of proliferating VILLIN-positive cells, and an increase of apoptotic VILLIN-positive cells (Figs S3D and S3E, respectively). Thus the inhibition of TGFβ signals might also have contributed to the survival and proliferation of human iPS-derived intestinal stem/progenitor cells.

Figure 4. Analysis of barrier formation capacity in the human iPS-derived enterocyte-like cell monolayers. (A) The gene expression levels of ZO-1 in Caco-2 cells, human iPS-derived enterocyte-like cells (hiPS-ELCs) and Adult Intestine were examined by real-time RT-PCR analysis. On the y axis, the gene expression levels in Caco-2 cells were taken as 1.0. (B) Immunostaining analysis of ZO-1 (Green) in the hiPS-ELCs and Caco-2 cells was performed. Nuclei were stained with DAPI (Blue). Scale bars represent 40 µ m. (C) TEER values of Caco-2 cell monolayers and hiPS-ELC monolayers were measured by Millicell-ERS. All data are represented as the means ± S.E. (n ≧ 3).

Corning® パーミアブルサポートとは

パーミアブルサポート、またはセルカルチャーインサートは、接着系細胞と浮遊系細胞の研究に欠かすことができないツールです。

セルカルチャーインサートは以下の用途でお使いいただけます：

w in vivo により近い細胞培養環境を実現するようにデザインされています。

w 上層と下層側から培地提供できるので自然な方法で極性細胞が代謝活性を得ることができます。

w 細胞と細胞の接触なく共培養を行うことができます。

w 多様な実験条件に応じて、様々なポアサイズ、メンブレンとコーティングをご使用いただけます。

このセレクションガイドは、希望する in vivo の環境により近い細胞培養環境を実現するためのメンブレンのタイプ、ポアサイズ、フォーマットと表面処理の最適な組み合わせの選択をサポートします。

Create a More Natural Environment for Your Cells

Coaster® トランズウェルパーミアブルサポート：スタンダードモデル

トランズウェルインサートはマルチウェルプレートにあらかじめセットされている、便利な ready-to-use のパーミアブルサポート製品です。このウェルが中央にくる吊り下げ式の独自デザインは、インサートとウェル側壁間での培地の毛細管現象を防ぎます。トランズウェルインサートには様々なサイズ、メンブレンタイプ、形状があります。広範囲に及ぶ引用、プロトコール、テクニカルサポートにより支えられており、25年以上にわたってセルカルチャーインサートのリーディングブランドとなっています。

Falcon® と Corning® BioCoat™ インサート：さらなる選択肢の提供

Falcon、Corning BioCoat インサートが加わり、さらに広範囲のパーミアブルサポート製品を提供しています。中には遊走アッセイに使用される Corning フルオロブロック™ インサートシステムや、細胞接着や増殖、分化を促進する BioCoat インサートもあります。

Falcon インサートはポリエチレンテレフタレート（PET）製で、個別包装されており、様々なポアサイズや形状があります。最良の結果を得るためには、Falcon、BioCoat とフルオロブロックインサートには Falcon セルカルチャーコンパニオンプレートを使用してください。このプレートは、インサートに細胞を播種したり、培養するためにインサートの位置を固定できます。

多孔性メンブレンマルチウェルプレート内

インサート

インサート内

トランズウェルは、独自の真ん中に中

心がくる吊り下げ式のデザインなの

で、インサートと外側のマルチプレー

トウェル間の毛細管現象を防ぎます。

また、インサート壁の隙間を通してマ

ルチウェルプレート内にアクセスが可

能で、マルチプルウェルプレート底部

の細胞に損傷を与えることなく共培養

を行うことができます。

LETTERS

Single Lgr5 stem cells build crypt–villus structures invitro without a mesenchymal nicheToshiro Sato1, Robert G. Vries1, Hugo J. Snippert1, Marc van de Wetering1, Nick Barker1, Daniel E. Stange1,Johan H. van Es1, Arie Abo2, Pekka Kujala3, Peter J. Peters3 & Hans Clevers1

The intestinal epithelium is the most rapidly self-renewing tissue inadult mammals. We have recently demonstrated the presence ofabout six cycling Lgr51 stem cells at the bottoms of small-intestinalcrypts1. Here we describe the establishment of long-term cultureconditions under which single crypts undergo multiple cryptfission events, while simultanously generating villus-like epithelialdomains in which all differentiated cell types are present. Singlesorted Lgr51 stem cells can also initiate these crypt–villusorganoids. Tracing experiments indicate that the Lgr51 stem-cellhierarchy is maintained in organoids. We conclude that intestinalcrypt–villus units are self-organizing structures, which can be builtfrom a single stem cell in the absence of a non-epithelial cellularniche.

The self-renewing epithelium of the small intestine is ordered intocrypts and villi2. Cells are newly generated in the crypts and are lost byapoptosis at the tips of the villi, with a turnover time of 5 days in themouse. Self-renewing stem cells have long been known to reside nearthe crypt bottom and to produce the rapidly proliferating transitamplifying (TA) cells. The estimated number of stem cells is betweenfour and six per crypt. Enterocytes, goblet cells and enteroendocrinecells develop from TA cells and continue their migration in coherentbands along the crypt–villus axis. The fourth major differentiated celltype, the Paneth cell, resides at the crypt bottom. We have recentlyidentified a gene, Lgr5, that is specifically expressed in cycling cryptbase columnar (CBC) cells that are interspersed between the Panethcells1. Using a mouse in which a green fluorescent protein (GFP)/tamoxifen-inducible Cre recombinase cassette was integrated intothe Lgr5 locus, we showed by lineage tracing that the Lgr51 cellsconstitute multipotent stem cells that generate all cell types of theepithelium1, even when assessed 14 months after induction of Cre3.

Although a variety of culture systems have been described4–7, nolong-term culture system has been established that maintains basiccrypt–villus physiology2. We attempted to design such a culture systemby combining previously defined insights in the growth requirementsof intestinal epithelium. First, Wnt signalling is a pivotal requirementfor crypt proliferation8–10 and the Wnt agonist R-spondin 1 inducesmarked crypt hyperplasia in vivo11. Second, signalling by epidermalgrowth factor (EGF) is associated with intestinal proliferation12.Third, transgenic expression of Noggin induces an expansion of cryptnumbers13. Fourth, isolated intestinal cells undergo anoikis outside thenormal tissue context14. Because laminin (a1 anda2) is enriched at thecrypt base15, we explored the use of laminin-rich Matrigel to supportintestinal epithelial growth. Matrigel-based cultures have been usedsuccessfully for the growth of mammary epithelium16.

Mouse crypt preparations were suspended in Matrigel. Crypt growthrequired EGF and R-spondin 1 (Supplementary Fig. 1a). Passagingrevealed a requirement for Noggin (Supplementary Fig. 1b). The

cultured crypts behaved in a stereotypical manner (Fig. 1a; Supple-mentary Movie 1). The upper opening rapidly became sealed, andthe lumen filled with apoptotic cells. The crypt region underwentcontinuous budding events, reminiscent of crypt fission17. Paneth cellswere always present at the bud site. Most crypts could be cultured(Fig. 1b). Further expansion created organoids, comprising more than40 crypt domains surrounding a central lumen lined by a villus-likeepithelium (‘villus domain’) (Fig. 1c–e). Staining with E-cadherinrevealed a single cell layer (Supplementary Fig. 2). At weekly intervals,organoids were mechanically dissociated and replated at one-fifth ofthe pre-plating density. Organoids were cultured for more than 8months without losing the characteristics described below.Expression analysis by microarray revealed that organoids remainedhighly similar to freshly isolated small-intestinal crypts, for instancewhen compared with fresh colon crypts (Supplementary Fig. 3).Moreover, no significant induction of stress-related genes was observed(Supplementary Table 1).

Culture of Lgr5–EGFP–ires–CreERT2 crypts revealed Lgr5–GFP1

stem cells intermingled with Paneth cells at the crypt base. Wntactivation, as demonstrated by nuclear b-catenin (SupplementaryFigs 4a and 9) and expression of the Wnt target genes Lgr5 (Fig. 1d)and EphB2 (ref. 18) (Supplementary Fig. 4b), was confined to thecrypts. Apoptotic cells were shed into the central lumen, a processreminiscent of the shedding of apopotic cells at villus tips in vivo(Supplementary Fig. 4c). Metaphase spreads of organoids more than3 months old consistently revealed 40 chromosomes in each cell(n 5 20) (Supplementary Fig. 4d). We found no evidence for thepresence of myofibroblasts or other non-epithelial cells (Supple-mentary Fig. 5).

We cultured crypts from Lgr5–EGFP–ires–CreERT2 mice crossedwith the Cre-activatable Rosa26–LacZ reporter to allow lineage tracing.Directly after induction with low-dose tamoxifen, we noted singlelabelled cells (Supplementary Fig. 4e, g). More than 90% of thesegenerated entirely blue crypts (Supplementary Fig. 4e–g), implying thatthe Lgr5–GFP1 cells did indeed retain stem cell properties. Crypts fromthe Cre-activatable Rosa26–YFP reporter19,20 mouse allowed lineagetracing by confocal analysis. Directly after treatment with tamoxifen,we noted single labelled cells that induced lineage tracing over thefollowing days, both in freshly isolated crypts (Supplementary Fig.6a–c) and in established organoids (Supplementary Fig. 6d).Supplementary Movie 2 represents four days of lineage tracing, reveal-ing green Lgr51 cells and YFP1 offspring (pseudocolour red) againstthe backdrop of a growing organoid.

Recently, mammary gland epithelial structures were establishedfrom single stem cells in vitro 21. When single Lgr5–GFPhi cells weresorted, these died immediately. The Rho kinase inhibitor Y-27632,which inhibits anoikis of embryonic stem cells22, significantly

1Hubrecht Institute and University Medical Center Utrecht, Uppsalalaan 8, 3584CT Utrecht, The Netherlands. 2Nuvelo, Inc., 201 Industrial Road, Suite 310, San Carlos, California94070-6211, USA. 3The Netherlands Cancer Institute, Antoni van Leeuwenhoek Hospital, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands.

Vol 459 | 14 May 2009 | doi:10.1038/nature07935

262 Macmillan Publishers Limited. All rights reserved©2009

decreased this cell death. Because cell-to-cell Notch signalling isessential to maintain proliferative crypts23, we also provided aNotch-agonistic peptide24. Under these conditions, significant num-bers of Lgr5–GFPhi cells survived and formed large crypt organoids.Organoids formed rarely when GFPlow daughter cells were seeded(Fig. 2d). Multiple Lgr5–GFPhi cells were intermingled with Panethcells at crypt bottoms (Fig. 2e, f). Incorporation of 5-ethynyl-29-deoxyuridine (EdU, a thymidine analogue) revealed S-phase cellsin the crypts (Fig. 2g).

We sorted cells at one cell per well, visually verified the presence ofsingle cells and followed the resulting growth. In each of four individualexperiments, we identified and followed 100 single cells. On average,

about 6% of the Lgr5–GFP hi cells grew out into organoids, whereas theremaining cells typically died within the first 12 h, presumably as aresult of physical and/or biological stress inherent in the isolationprocedure. GFPlow cells rarely grew out (Fig. 3a). Figure 3b andSupplementary Fig. 7 illustrate the growth of an organoid from a singleLgr5–GFPhi cell. By four days of culture, the structures consisted ofabout 100 cells, which is consistent with the 12-h cell cycle of pro-liferative crypt cells25 (Fig. 3c). After 2 weeks, the organoids were dis-sociated into single cells and replated to form new organoids (Fig. 3d).This procedure could be repeated at least four times on a two-weeklybasis, without apparent loss of replating efficiency.

The organoids derived from single stem cells were indistinguishablein appearance from those derived from whole crypts. Paneth cells andstem cells were located at crypt bottoms (Figs 2e, f and 4c, g). Fullypolarized enterocytes, as demonstrated by villin1 mature brushborders and apical alkaline phosphatase, lined the central lumen(Fig. 4a, e, i). Goblet cells (Muc21, Fig. 4b; periodic acid–Schiff(PAS)1, Fig. 4f) and enteroendocrine cells (chromogranin A1,Fig. 4d; synaptophysin1, Fig. 4h) were scattered throughout the orga-noid structure. Four types of mature cell were recognized by electronmicroscopy (Fig. 4i–l). Non-epithelial (stromal/mesenchymal) cellswere absent, an observation confirmed by electron-microscopicimaging (Fig. 4i–p and Supplementary Fig. 8c–g). Both the crypts(Fig. 4m–o) and the central luminal epithelium (Fig. 4p) consistedof a single layer of polarized epithelial cells resting directly onthe Matrigel support. High-resolution images of these electron-microscopic pictures are given in Supplementary Fig. 9. We frequentlynoted small intercellular vacuoles, possibly an indicator of culture-induced or fixation-induced stress (Fig. 4i–p and SupplementaryFig. 8).

Crypt domain

Villusdomain

Lumen

Day 1.5 Day 2

Day 3.5Day 3

Day 4.5 Day 5

Day 1

Day 2.5

Day 4

e

c

a

d

b

Figure 1 | Establishment of intestinal crypt culture system. a, Time courseof an isolated single crypt growth. Differential interference contrast imagereveals granule-containing Paneth cells at crypt bottoms (arrows). b, c, Singleisolated crypts efficiently form large crypt organoids within 14 days; b, onday 5; c, on day 14. d, Three-dimensional reconstructed confocal image after3 weeks in culture. Lgr5–GFP1 stem cells (green) are localized at the tip ofcrypt-like domains. Counterstain, ToPro-3 (red). e, Schematicrepresentation of a crypt organoid, consisting of a central lumen lined byvillus-like epithelium and several surrounding crypt-like domains. Scale bar,50 mm.

GFPhi

GFPlow

GFP–

GFP

d

e f g

a b c

FSC

Figure 2 | Single Lgr51 cells generate crypt–villus structures.a, Lgr5–GFP1 cells from an Lgr5–EGFP–ires–CreERT2 intestine (bottom);wild-type cells (top). Two positive populations, GFPhi and GFPlow, arediscriminated. FSC, forward scatter. b, Confocal analysis of a freshly isolatedcrypt. Black arrowheads, GFPhi; white arrowheads, GFPlow. c, Sorted GFPhi

cells. d, 1,000 sorted GFPhi cells (left) and GFPlow cells (right) after 14 days inculture. e, f, Fourteen days after sorting, single GFPhi cells form cryptorganoids, with Lgr5–GFP1 cells and Paneth cells (white arrows) located atcrypt bottoms. Scale bar, 50 mm. f, Higher magnification of e. g, Organoidscultured with the thymidine analogue EdU (red) for 1 h. Note that only cryptdomains incorporate EdU. Counterstain, 4,6-diamidino-2-phenylindole(DAPI; blue).

NATURE | Vol 459 | 14 May 2009 LETTERS

263 Macmillan Publishers Limited. All rights reserved©2009

オルガノイドカルチャーを行う。

REVIEW

Growing Self-Organizing Mini-Gutsfrom a Single Intestinal Stem Cell:Mechanism and ApplicationsToshiro Sato1* and Hans Clevers2*

Recent examples have highlighted how stem cells have the capability to initiate morphogenesisin vitro; that is, to generate complex structures in culture that closely parallel their in vivocounterparts. Lgr5, the receptor for the Wnt-agonistic R-spondins, marks stem cells in multipleadult organs of mice and humans. In R-spondin–based three-dimensional cultures, these Lgr5stem cells can grow into ever-expanding epithelial organoids that retain their original organidentity. Single Lgr5 stem cells derived from the intestine can be cultured to build epithelialstructures that retain hallmarks of the in vivo epithelium. Here, we review the mechanisms thatsupport this notable example of self-organization and discuss applications of this technology forstem cell research, disease modeling (e.g., for colorectal cancer and cystic fibrosis), andregenerative medicine.

The epithelium of the small intestine has ahigher self-renewal rate than any othermam-malian tissue, with a turnover time of less

than 5 days. Intestinal stem cells reside near thebottom of the intestinal crypt. Their rapidly di-viding, transit-amplifying (TA) daughter cells oc-

cupy the remainder of the crypts and flow ontothe flanks of the villi, where they differentiate,absorb nutrients, and eventually die at the villustips (Fig. 1A). The differentiated cell types includeabsorptive enterocytes, multiple secretory cells(Paneth cells, goblet cells, enteroendocrine

cells, and tuft cells), and the M cells of Peyer’spatches (1).

Stem Cells of the Intestinal CryptCheng and Leblond (2) were the first to describethe slender crypt base columnar (CBC) cells atthe crypt bottom, intercalated between the post-mitotic Paneth cells (Fig. 1B). The Wnt targetgene Lgr5 is an excellent marker for CBC cells,in turn allowing the assessment of their positionin the self-renewal hierarchy by genetic lineagetracing (3). Marked Lgr5+ cells persist for the life-time of a mouse, whereas their progeny includeall differentiated cell lineages of the epithelium.Thus, Lgr5+ cells represent cycling, long-lived,mul-tipotent stem cells. A second, quiescent stem celltype is proposed to reside at position four (direct-ly above the Paneth cells): These “+4 cells” (1)were originally identified as DNA label-retainingcells (4). Several markers have been described forthese cells, such as Bmi-1, Hopx,mTert, and Lrig1(5–8). Because the stem cell niche (microenvi-ronment) and culture condition for +4 cells have

1Department of Gastroenterology, Keio University School of Med-icine, Tokyo 160-8582, Japan. 2Hubrecht Institute–KoninklijkeNederlandse Akademie van Wetenschappen and University Med-ical Centre Utrecht, Uppsalalaan 8, 3584CT, Utrecht, Netherlands.

*Corresponding author. E-mail: [email protected] (T.S.);[email protected] (H.C.)

Smad1/5/8

Ras Raf MEK

Notch NICD

RBP-J

ERK

Stemness

EGF

Wnt

BMPR-spondin

EGFR

LRP5/6

Lgr4/5

BMPR

β-cateninFrizzled

DII1/4

cut cut

A C

B

Wnt

Crypt

Lgr5 CBC

+4 cell+4 cell

Panethcells

Villus

TAcells

BMP EGF

Tcf4

Smad4

Paneth cell

Stem cell

Fig. 1. Histological locationandbiological interactionof intestinal stemcells and their niche. (A) Scheme of intestinal epithelial structure and stem cells.Spatial gradients of Wnt, BMP, and EGF signals are formed along the crypt axis. (B)Cartoonof the stemcell niche. Lgr5+ intestinal CBC cells intimately adhere to Paneth

cells and receive signals for stem cell maintenance. (C) Three signals (EGF, Notch,and Wnt) are essential for intestinal epithelial stemness, whereas BMP negativelyregulates stemness. For full Wnt activation in the intestinal epithelium, R-spondin–Lgr4/5 signal is required. Currently, the source of R-spondin is unknown.

7 JUNE 2013 VOL 340 SCIENCE www.sciencemag.org1190

Morphogenesis

REVIEW

Growing Self-Organizing Mini-Gutsfrom a Single Intestinal Stem Cell:Mechanism and ApplicationsToshiro Sato1* and Hans Clevers2*

Recent examples have highlighted how stem cells have the capability to initiate morphogenesisin vitro; that is, to generate complex structures in culture that closely parallel their in vivocounterparts. Lgr5, the receptor for the Wnt-agonistic R-spondins, marks stem cells in multipleadult organs of mice and humans. In R-spondin–based three-dimensional cultures, these Lgr5stem cells can grow into ever-expanding epithelial organoids that retain their original organidentity. Single Lgr5 stem cells derived from the intestine can be cultured to build epithelialstructures that retain hallmarks of the in vivo epithelium. Here, we review the mechanisms thatsupport this notable example of self-organization and discuss applications of this technology forstem cell research, disease modeling (e.g., for colorectal cancer and cystic fibrosis), andregenerative medicine.

The epithelium of the small intestine has ahigher self-renewal rate than any othermam-malian tissue, with a turnover time of less

than 5 days. Intestinal stem cells reside near thebottom of the intestinal crypt. Their rapidly di-viding, transit-amplifying (TA) daughter cells oc-

cupy the remainder of the crypts and flow ontothe flanks of the villi, where they differentiate,absorb nutrients, and eventually die at the villustips (Fig. 1A). The differentiated cell types includeabsorptive enterocytes, multiple secretory cells(Paneth cells, goblet cells, enteroendocrine

cells, and tuft cells), and the M cells of Peyer’spatches (1).

Stem Cells of the Intestinal CryptCheng and Leblond (2) were the first to describethe slender crypt base columnar (CBC) cells atthe crypt bottom, intercalated between the post-mitotic Paneth cells (Fig. 1B). The Wnt targetgene Lgr5 is an excellent marker for CBC cells,in turn allowing the assessment of their positionin the self-renewal hierarchy by genetic lineagetracing (3). Marked Lgr5+ cells persist for the life-time of a mouse, whereas their progeny includeall differentiated cell lineages of the epithelium.Thus, Lgr5+ cells represent cycling, long-lived,mul-tipotent stem cells. A second, quiescent stem celltype is proposed to reside at position four (direct-ly above the Paneth cells): These “+4 cells” (1)were originally identified as DNA label-retainingcells (4). Several markers have been described forthese cells, such as Bmi-1, Hopx,mTert, and Lrig1(5–8). Because the stem cell niche (microenvi-ronment) and culture condition for +4 cells have

1Department of Gastroenterology, Keio University School of Med-icine, Tokyo 160-8582, Japan. 2Hubrecht Institute–KoninklijkeNederlandse Akademie van Wetenschappen and University Med-ical Centre Utrecht, Uppsalalaan 8, 3584CT, Utrecht, Netherlands.

*Corresponding author. E-mail: [email protected] (T.S.);[email protected] (H.C.)

Smad1/5/8

Ras Raf MEK

Notch NICD

RBP-J

ERK

Stemness

EGF

Wnt

BMPR-spondin

EGFR

LRP5/6

Lgr4/5

BMPR

β-cateninFrizzled

DII1/4

cut cut

A C

B

Wnt

Crypt

Lgr5 CBC

+4 cell+4 cell

Panethcells

Villus

TAcells

BMP EGF

Tcf4

Smad4

Paneth cell

Stem cell

Fig. 1. Histological locationandbiological interactionof intestinal stemcells and their niche. (A) Scheme of intestinal epithelial structure and stem cells.Spatial gradients of Wnt, BMP, and EGF signals are formed along the crypt axis. (B)Cartoonof the stemcell niche. Lgr5+ intestinal CBC cells intimately adhere to Paneth

cells and receive signals for stem cell maintenance. (C) Three signals (EGF, Notch,and Wnt) are essential for intestinal epithelial stemness, whereas BMP negativelyregulates stemness. For full Wnt activation in the intestinal epithelium, R-spondin–Lgr4/5 signal is required. Currently, the source of R-spondin is unknown.


Morphogenesis

Science, 340(6137): 1190-1194 (2013)

not yet been defined, we will not discuss themfurther. Very recently, genetic marking of DNAlabel-retaining cells has identified a rare, nondi-viding secretory precursor that coexpresses Lgr5and all +4 markers. Located near crypt bottoms,this precursor undergoes terminal differentiationover periods of weeks. Upon tissue damage, itreverts into a cycling, Lgr5+ stem cell CBC cell(9). Hereafter, we refer to the dividing Lgr5+ stemcells as Lgr5-CBC cells to distinguish them fromthese much rarer, label-retaining, nondividingLgr5+ cells.

Every murine crypt contains ~15 Lgr5-CBCcells (10).AnLgr5-CBCcell divides every 24hoursand, through its TA daughters, generates 16 to 32differentiated epithelial cells per day. Cell divi-sion generally occurs in a symmetrical fashion,after which individual daughter cells stochas-tically adopt stem cell or TA cell fates, dependingon available niche space (10, 11).

The Intestinal Stem Cell NicheIntestinal homeostasis is tightly controlled by fourwell-characterized signaling pathways (Fig. 1,A and C). Wnt constitutes the key pathway tomaintain stem cell fate and drive proliferation ofstem- and TA cells (12). Paradoxically, Wnt alsodrives terminal differentiation of Paneth cells, that

are always in direct contact with the stem cells(13). Wnt factors, when engaging their Frizzled-Lrp5/6 co-receptors, will induce stabilization ofb-catenin. The latter molecule can then bind andactivate the transcription factor Tcf4, thus activat-ing a genetic program that supports stemness(Fig. 1C). Notch is also essential to maintain theundifferentiated state. When Notch signaling isblocked in proliferative stem and TA cells, thesecells differentiate into secretory lineage cells(14). Dll1+ Dll4+ Paneth cells trigger signaling byNotch1/2 on neighboring stem cells, thus keepingthem from secretory differentiation (15). Each day,stem cell daughters lose contact with the Dll1/4-expressing Paneth cells. Some of these cells down-regulate Notch and up-regulate Dll1, setting theirown secretory fate (16). Such a secretory precur-sor presents Dll1 to six to eight neighboring Notch+

TA cells, which will experience active Notch sig-nals and stay fated toward the enterocyte line-age. In sum, Notch lateral inhibition controls theenterocyte-secretory switch. Figure 1C depictshow engagement of Notch by its ligand Dll1/4leads to release of the Notch intracellular domain,which subsequently interacts with the nuclear ef-fector RBP-J to suppress the gene program forsecretory differentiation. Epidermal growth fac-tor (EGF) signals exert strong mitogenic effects

on stem and TA cells upon engagement of theirEGF receptors (EGFRs). Indeed, the Ras/Raf/Mek/Erk signaling axis is active in crypt epi-thelium. Inhibition of Mek ablates intestinal stemcells (17). Finally, bone morphogenetic protein(BMP) signals are active in the villus compartment.When BMP signaling in the villus is inhibitedby transgenic Noggin, cryptlike structures appearalong the flanks of the villi (18), implying thatBMP inhibition creates a crypt-permissive envi-ronment. Figure 1C shows how engagement ofBMP receptors by BMP leads to complexes be-tween Smad1/5/8 and Smad 4 to repress stem-ness genes in the nucleus.

The Lgr5-CBC cells are in intimate contactwith Paneth cells. Paneth cells secrete bactericid-al products as well as EGF andWnt3. Moreover,they express the Notch ligands Dll1 and Dll4on their surface (Fig. 1C). Paneth cell depletionin vivo results in concomitant loss of Lgr5-CBCcells in several animal models (19, 20), with theexception of a conditional Math1 mutant, whicheliminates Paneth cells yet retains its stem cells(21). It was then pointed out thatMath1 deletionin Lgr5-CBC cells relieves the in vivo dependenceon Notch ligands, which are normally providedby Paneth cells and that are essential for Lgr5-CBC cell maintenance (22). Durand et al. noted

A C

B

CryptSingle cells

Organoids

Wound bed

Wnt5A+ cells

FACS

Culture medium

Fig. 2. Mini-gut culture system. (A) Lgr5+ CBC cells genetically labeled byEGFP are sorted and embedded in Matrigel. The culture medium consists ofEGF, Noggin, and R-spondin. FACS, fluorescence-activated cell sorting. (B) Timecourse of organoid growth. A single stem cell forms a symmetric cyst structure.The symmetry is broken by bud formation. The budding structure resembles a

crypt. Lgr5+ CBC cells are depicted in yellow, and Paneth cells are shown inblue. (C) Scheme showing the engraftment of intestinal oganoids. Organoidsadhere to a de-epithelialized wound bed. Organoids form a flat epithelial layer,followed by crypt reconstruction. Wnt-5A+ mesenchymal cells support cryptstructure formation in the damaged area. Donor organoids are depicted in red.

www.sciencemag.org SCIENCE VOL 340 7 JUNE 2013 1191

SPECIALSECTION

that Math1-mutant mini-guts (see below) fail togrow in vitro, which implies a crucial in vitro de-pendence on another signal provided by Panethcells, probably Wnt3. Indeed, Wnt3−/− crypts grownormally in vivo but fail to grow in vitro (13).Together, Paneth cells fulfill a stem cell nichefunction. In vivo, Wnts are also provided by sur-rounding mesenchyme (13), whereas R-spondinsand BMP inhibitors are distinctly provided fromnonepithelial sources.

Epithelial Mini-Gut CultureIt has generally been thought impossible to es-tablish long-term cultures from primary adult tis-sues without inducing genetic transformation.Indeed, the Hayflick limit indicates that somaticcells have a limited proliferative potential (23).Yet, we have recently established an in vitro cul-ture system to grow three-dimensional (3D) intes-tinal epithelial organoids (“epithelial mini-guts”)from a single Lgr5-CBC cell for periods greaterthan 1.5 years (24). We employed R-spondin-1, aWnt signal enhancer, that was later discovered asthe physiological ligand of Lgr5: The R-spondin–Lgr5 interaction augments Wnt signal strengthinitiated by the interaction ofWnt with Frizzledand Lrp (25, 26) (Fig. 1C). R-spondins are cru-cial in vivo. Gut-specific depletion of Lgr4 and -5leads to the demise of crypts (26, 27), whereasexogenous R-spondin-1 induces hyperplasticity ofcrypts (28). The isolated crypts require Matrigel(BD Biosciences, San Jose, CA), a 3D laminin-and collagen-rich matrix that mimics the basallamina. A cocktail of R-spondin, EGF, andNogginrepresents the minimal, essential stem cell main-tenance factor cocktail. For colon crypt culture,Wnt ligand is an additional factor required tomaintain Lgr5-CBC cells, because the epithelium

makes little, if any, Wnt. Of note, this approachalso allows derivation of intestinal organoids frominduced pluripotent stem cells (29). In an alter-native approach, intestinal fragments containingepithelial and mesenchymal components fromneonatal mice were grown for several monthsin a collagen gel with air-liquid interface in thepresence of serum. The expanding cystic struc-tures consisted of a simple epithelium of all celltypes surrounded by myofibroblasts and wereresponsive to R-spondin-1 and Notch inhibition.They did not develop defined crypt- and villus-like domains (30).

Single crypts can be readily isolated frommouseor human intestine by EDTA-based Ca2+/Mg2+

chelation. Such crypts grow into 3D organoidsunder the culture conditions described. Culturingof single stem cells is inefficient at 1 to 2% plat-ing efficiency, whereas up to half of the stemcell–Paneth cells doublets form organoids in vitro(19). This doublet assay has allowed demonstra-tion that Paneth cells monitor the metabolic stateto fine-tune stem cell activity (31).

In vitro–generated organoids occur as cystswith a central lumen flanked by a simple, highlypolarized villus epithelium.Multiple cryptlike struc-tures project outward (Fig. 2B). The basal side ofthe cells is oriented toward the outside, touchingthe matrigel, whereas enterocyte brush bordersform the luminal surface. Secretion by Paneth andgoblet cells occurs toward the lumen. The or-ganoids can be passaged weekly at a 1:5 ratio forat least 1.5 years, with a phenotype and karyotypethat remain unchanged. Mechanically disruptedorganoids rapidly reseal. Self-renewal kinetics andcell-type composition closely resemble the in vivosituation. Notably, a timer of unknown molecularnature remains active in the absence of the in vivo

wear-and-tear: 2 to 3 days after terminal differen-tiation, the cells exfoliate into the lumen.

How normal are the epithelial mini-guts? Totest this, large numbers of organoids were grownin vitro from a single Lgr5 stem cell (Fig. 2B) thatwas marked by a red fluorescent protein in theadult colon (32). These in vitro–expanded colonorganoids were reintroduced per anum into the co-lons of multiple mice with chemical-induced mu-cosal lesions (Fig. 2C). The engrafted epithelialmini-guts regenerated epithelial patches that wereindiscernible from surrounding recipient epithe-lium. The patches persisted for at least 6 monthswithout changing their histologic appearance.

How Does an Lgr5 Stem Cell Establish EpithelialMini-Gut Architecture?When epithelial mini-guts derived from a Wnt re-porter mouse (Axin2-LacZ) are grown under stan-dard conditions, strong Wnt signals only occuradjacent to Paneth cells (19) (Fig. 3A). This fitswith the notion that Paneth cells are the onlysource ofWnt in culture (13) and that R-spondin-1 that is ubiquitously present in the medium en-hances these focal Wnt signals. Thus, sharp Wntgradients surround Paneth cells. In a medium con-taining Wnt3A in addition to R-spondin-1, theWnt gradient is lost and epithelial mini-guts be-come symmetric, round cysts, consisting of a homo-geneous population of stem and progenitor cells(19). Epithelial mini-guts grown from adenoma-tous polyposis coli (APC)–mutant adenoma cellsdisplay the same symmetric shape (19), which isnot surprising, because APC loss leads to consti-tutive Wnt pathway activation. These observa-tions imply that the typical crypt-villus architectureis suppressed under conditions of homogeneous(rather than focal) Wnt signaling.

A B

Wnt

EphB+EphB+

EphrinB+ EphrinB+

Fig. 3. Mechanism of the self-organizing architecture of a mini-gut.(A) An organoid derived from Axin2-LacZ knock-in mice. Axin2-LacZ (blue)expression recapitulates Wnt activation. (B) Wnt activation induces local pro-

liferation and EphB expression. Local cell expansion and Eph-Ephrin repulsiveforce generate bud formation. Wnt-producing Paneth cells are depicted inblue; Wnt-activated cells are shown in pink.


Morphogenesis

When a single cultured stem cell is followedover time, a small symmetric cyst forms. Thestochastic appearance of a Paneth cell constitutesthe “symmetry-breaking” event (Fig. 3B): A budforms around the cell; within 2 to 3 days, this buddevelops into a cryptlike structure with stem andPaneth cells. Every nondifferentiated cell thattouches a Paneth cell is (or becomes) an Lgr5-CBC cell (16, 19), presumably driven by potentWnt and Notch signals from the Paneth cell.

The proliferative stem cell–Paneth niche pushesitself outward from the central cyst (movie S1).This process is most likely driven by repulsiveEphB-EphrinB interactions, as described in vivo(33). EphB2 and -B3 are Wnt target genes, ex-pressed by Paneth and stem cells. When TA cellsmove toward the villus in vivo, they travel downthe Wnt gradient. As a consequence, they differ-entiate and gradually replace EphB expression byexpression of the counterstructures, EphrinBs.Translated to the situation in the mini-gut, cellsthat sit at the Wnt source (Paneth cells and theirdirect neighbors, the Lgr5-CBC cells) are EphB+.Because all other cells in the central cyst do notexperience Wnt signals and express the counter-structure EphrinB, the EphB+ stem cells and Panethcells are expelled from the cyst and create a bud(Fig. 3B). In these cryptlike buds, the Wnt3-producing (and, therefore, EphB3+) Paneth cellsare sorted toward the bottom. Thus, aWnt-gradientis automatically created along the crypt axis. Pro-liferating TA cells are mechanically pushed towardthe lumen by younger TA cells, thus experiencingrapidly decreasing Wnt levels. This drives their

terminal differentiation into one of the villus-epithelial cell types, with Notch lateral inhibitionacting as the enterocyte-secretory fate switch.

How Stem Cell Numbers Are ControlledTouching a Paneth cell appears necessary andsufficient to acquire and maintain the Lgr5-CBCcell phenotype. Thus, a key parameter that con-trols crypt homeostasis is the number of Panethcells. StrongWnt signals, in the absence of Notchsignals, drive formation of new Paneth cells(13, 34). Because Paneth cell–derivedWnt can,in turn, generate more stem and Paneth cells, aWnt-driven positive-feedback loop could poten-tially induce ever-expanding crypts. Two E3 li-gases (Rnf43 and Znrf3) are encoded by stemcell–specific Wnt target genes (35). These two E3ligases down-regulate Wnt receptors, thus servingas feedback inhibitors of the Wnt pathway(35, 36). When this negative-feedback loop isbroken by deletion of the two E3 ligase genes, arapidly expanding Paneth–stem cell zone is ob-served (35). Thus, Rnf43 and Znrf3 serve in thehomeostatic control of crypt size by negativelyregulating Wnt signal strength. The strength ofthe EGF signal also affects stem cell numbers.Deletion of Lrig1, a negative regulator of EGFRfamily members, leads to increased stem cell num-bers in vivo and in culture (17).

Epithelial Mini-Guts as Experimental ToolsThe ability to grow epithelial mini-guts from sin-gle cells can be used as a surrogate marker ofstemness, allowing the study of (the interrelation-

ship of) intestinal stem cell types (6, 37–39, 40).Although cycling Lgr5-CBC cells grow understandard conditions (24), the addition of Wnt3Acan coerce proliferative progenitors, such as theDll1+ secretory precursor (16), or even the non-dividing Paneth precursors (9, 41), to revert to a stemcell phenotype and generate epithelial mini-guts.

Organoids that are grown from mouse intes-tine are amenable to all standard experimentalmanipulations used for cell lines, including long-term storage by freezing, transfection of DNAandof small interfering RNA, and infection with re-combinant retro- and lentiviruses (42). They canbe analyzed by immunohistochemistry and con-focal immunofluorescence, gene expressionmicro-array, and mass spectrometry. Floxed alleles canbe deleted in culture (for instance, using CreERT2inconjunctionwith tamoxifen) and the effect followedin real time. The identification of Lgr4 and -5 as re-ceptors ofR-spondins has restedon thegeneration oforganoids mutant in these genes (26, 27, 43).Other genes studied in organoids include Toll-like receptor 4 (44), Troy (45), and YAP (46).

The organoid technology is well suited to thestudy of cell-biological phenomena that requirea closed epithelial structure with a physiologi-cal, polarized topology. Mizutani et al. studiedP-glycoprotein (an efflux transporter acting atthe intestinal epithelium) in mini-guts. The trans-porter was observed at the apical (luminal) mem-brane in organoids and actively pumped rhodamine123 toward the luminal space (47).

Organoids recapitulate the complete stem celldifferentiation hierarchy and allow the in vitro

Fig. 4. Basic and clinical applications of an epithelial mini-gut. Anepithelial mini-gut is efficiently established from a single (3 to 5 mm2) endo-scopic biopsy sample. EDTA chelation releases ~3000 crypts from a biopsysample. An epithelial mini-gut grows logarithmically and expands 1000-foldwithin amonth. Three applications of epithelial mini-guts are as follows: (i) As anexperimental tool. Genetic manipulation, gene expression analysis, live imaging,

and other standard biological analyses can be employed for normal and patient-derived epithelial mini-guts. (ii) As a diagnostic tool. Patient-derived epithelialmini-guts recapitulate in vivo intestinal epithelial functions and genetic sig-natures. Efficient expansion of pure epithelial cells provides a high-quality sourcefor deep sequencing or functional assays. (iii) As a therapeutic tool. Epithelialmini-gut transplantation may become a feasible regenerative therapy.

Experimental tool Diagnostic tool Therapeutic tool

Research for Intestinal stem cell Intestinal differentiation Epithelial functions • • • • •

Potential regenerative therapy for Microvillus inclusion disease Ulcerative colitis Post-endoscopic mucosal resection • • • • •

Cystic Fibrosis Mutational analysis in CRC Drug absorption/metabolism • • • • •

Epithelial Mini-Gut

Crypts Biopsysample

Colonoscopy

www.sciencemag.org SCIENCE VOL 340 7 JUNE 2013 1193

SPECIALSECTION

生命工学 - 九州大学（kyushu university）...パラチノースガイドブック ver.3.2 16...

Documents