RESEARCH

BACKGROUND: In adult

mammalian organ-

isms, multiple tissues

—including the skin,

blood, stomach, and in-

testines—are entrapped

in a state of permanent

regeneration; older cells are constantly

shed, and the tissue is continuously being

regenerated from resident stem cells. This

phenomenon of “tissue renewal” was ap-

preciated by Leblond in 1956, but the un-

derlying mechanism has been unclear. It

is now evident that a class of extracellular

developmental signaling proteins, known

as Wnt signals, animate the continued

renewal of several mammalian tissues by

An integral program for tissue renewal and regeneration: Wnt signaling and stem cell control

STEM CELL SIGNALING

Hans Clevers,1 Kyle M. Loh,2 Roel Nusse2*

Multiple adult organs are in a state of continual regeneration. In tissues such as the skin,

intestines, brain, and mammary glands, Wnt signaling proteins sustain this constant regenera-

tion by inducing stem cells (green cells in the illustration) to grow. This leads to the robust sup-

ply of new cells (green) in order to replenish and maintain the tissue. [Image credits available

in the full article online.]

Skin

Intestines

Brain

Mammary glands

Wnt signaling fueling

tissue renewal and

stem cell activity in

diverse organs

Wnt

REVIEW SUMMARY

fueling stem cell activity. If the Wnt path-

way is inhibited, tissue renewal is crippled.

This signaling pathway is an ancient evo-

lutionary program dating from when Wnt

signals arose in the simplest multicellular

organisms, in which Wnts acted as primor-

dial symmetry-breaking signals crucial for

the generation of patterned tissues during

embryogenesis. In vertebrates, these signals

also function in pattern maintenance: They

sustain tissue renewal, enabling tissues

to be continuously replenished and main-

tained over a lifetime.

ADVANCES: In contrast to traditional “long-

range” developmental signals, Wnts seem to

act as short-range intercellular signals—act-

ing mostly between adjacent cells. Lending

credence to this notion, a membrane-teth-

ered Wnt protein variant can fulfill most

functions of a normal Wnt protein in Dro-

sophila. Likely explaining the short-range

nature of these signals, Wnt proteins are

attached to a lipid and therefore are hydro-

phobic; they cannot freely traverse the extra-

cellular space by themselves. This provides

insight into how tissue renewal is regulated.

It implies that Wnt signals emanating from

the stem cell microenvironment (the “niche”)

may influence adjacent stem cells without

affecting a broad field of cells located far-

ther away. The concept of an external niche,

however, may have to be refined because it

is clear that stem cells can sometimes act as

their own niche and have unexpected devel-

opmental self-organizing capacities. Last,

the widespread importance of Wnt signal-

ing in driving tissue re-

newal has been revealed

by the identification of

Axin2 and Lgr5, genes

expressed in cells that

are responding to Wnt

signals. Genetically la-

beling Axin2� or Lgr5� cells in a variety of

tissues has revealed that such cells fuel tissue

renewal in the intestines, mammary gland,

skin, and brain, among other organs.

OUTLOOK: The amazing continuous self-

regeneration of various mammalian tissues

over years and decades continues to be an

enigmatic terra incognita in biology. For

instance, visualization of stem cells in real-

time in vivo (through intravital microscopy)

has shown that when some stem cells are

ablated, they are replaced by more differ-

entiated cells that are recalled to the stem

cell niche, whereupon they regain stem cell

identity to effect tissue repair. Therefore,

lineage barriers between stem cell and dif-

ferentiated fates are not always stringent

and can be traversed during times of tissue

damage. Reactivated Wnt signals may be

instrumental in this process, and perhaps

such signals could be exploited in order to

enkindle tissue regeneration after injury or

disease. From a pragmatic perspective, Wnt

signals have already found practical use in

manipulating stem cells, enabling propaga-

tion of stem cells in vitro as self-renewing

cell populations and as organoids. ■

1Hubrecht Institute, Royal Netherlands Academy of Arts and Sciences (KNAW), University Medical Centre Utrecht and CancerGenomics.nl, 3584CT Utrecht, Netherlands.2Department of Developmental Biology, Howard Hughes Medical Institute, Stanford Institute for Stem Cell Biology and Regenerative Medicine, Stanford University, 265 Campus Drive, Stanford, CA 94305, USA.*Corresponding author. E-mail: rnusse@stanford.edu Cite this article as: H. Clevers et al., Science 346, 1248012 (2014). DOI: 10.1126/science.1248012

Read the full article at http://dx.doi.org/10.1126/science.1248012

ON OUR WEB SITE

SP

ECIAL SERIES: STEM C

EL

LS

54 3 OCTOBER 2014 • VOL 346 ISSUE 6205 sciencemag.org SCIENCE

Published by AAAS

REVIEW◥

STEM CELL SIGNALING

An integral program for tissuerenewal and regeneration:Wntsignaling and stem cell controlHans Clevers,1 Kyle M. Loh,2 Roel Nusse2*

Stem cells fuel tissue development, renewal, and regeneration, and these activities arecontrolled by the local stem cell microenvironment, the “niche.” Wnt signals emanatingfrom the niche can act as self-renewal factors for stem cells in multiple mammalian tissues.Wnt proteins are lipid-modified,which constrains them to act as short-range cellular signals.The locality of Wnt signaling dictates that stem cells exiting the Wnt signaling domaindifferentiate, spatially delimiting the niche in certain tissues. In some instances, stem cellsmay act as or generate their own niche, enabling the self-organization of patterned tissues.In this Review, we discuss the various ways by which Wnt operates in stem cell control and,in doing so, identify an integral program for tissue renewal and regeneration.

In a 1956 review entitled “Renewal of CellPopulations,” Leblond and Walker noted thatmultiple adult tissues, including the skin andintestines, accommodate numerous mitoticdivisions but seemingly do not undergo a com-

mensurate expansion in tissue size (1). The au-thors presciently concluded that “the cells of thetissue are said to undergo renewal” (1). Such tis-sues are perpetually being “recycled,” with cellsbeing extruded or lost and continually being re-placed by newly born cells.It is now evident that stem cells are required

for continuous tissue maintenance within di-verse organs. Cellular losses within these tissues(owing to either natural cellular attrition or in-jury) are persistently replenished by stem cells,which we define as cells that sustain continuedtissue formation by generating tissue progenywhile renewing themselves through division. Stemcell activity is often externally dictated by themicroenvironment (the niche) so that stem celloutput is precisely shaped to meet homeostaticneeds or regenerative demands.This Review details how a class of developmen-

tal signals, known as Wnt signals, control stemcell operation and are crucial for the continuedrenewal of multiple mammalian tissues. Such arole was presaged by a pivotal role for Wnt in thedevelopment and regeneration of the earliestanimals. Although a number of signals controlstem cell activity, Wnts are somewhat idiosyn-cratic in that they primarily seem to act as short-range cellular signals between adjacent cells. This

mode of spatially constrained signaling mightbear developmental and regenerative impor-tance, communicating a directive to nearby cellswithout influencing a broad domain.

Signaling by lipid-modified short-rangeWnt factors

A tenet of the stem cell niche model is the shortrange at which signals act, maintaining a lim-ited number of stem cells near the niche. By theirvery nature, Wnt proteins fit the bill.Wnts are secreted signaling proteins that by

virtue of their biochemical properties, seem prin-cipally to operate over short distances. All Wntproteins harbor a covalent lipid modification: apalmitate, appended by the palmitoyltransfer-ase Porcupine (Fig. 1A). This lipid group rendersthe Wnt protein hydrophobic and tethers it to cellmembranes or its cognate receptors. The trans-membrane proteinWntless (Wls) exclusively bindsonly lipidated Wnt proteins (2) and conveys themto the plasma membrane for secretion. Therefore,after secretion the lipid may be pivotal in limit-ing Wnt dispersion and its range of biologicalaction, a precept to which we return below.Once secreted, how Wnt signals are conveyed

to their target cells remains cryptic. Some Wntproteins may be incorporated into secretoryvesicles (3), in which Wls continues to bind Wntproteins (4) as a chaperone (Fig. 1B), perhapsavailing the presentation of lipidated Wnt pro-teins to their cognate receptors, known as Frizzledreceptors. Wnt signaling mediated by such vesi-cles would operate over a short distance, suchas at the neuromuscular junction (4) and also instem cell niches.Although it is sometimes assumed that Wnt

signals are long-range morphogens, there islittle evidence that this is the prevailing mode ofWnt action. Wnt signaling occurs mostly betweencells that are touching each other. Even in the

best studied example of long-range signaling bya Wnt—that is, by the Wnt ligand Wingless inDrosophila—recent evidence has made a case thatthe requirements for any function of Wingless canbe largely afforded by a nondiffusible, membrane-tethered form of the protein (Fig. 1C) (5) andthat Wingless does not act as a long-range mor-phogen in that context.Once delivered to their target cells, Wnt ligands

engage their cognate Frizzled receptors throughtheir palmitate group, which extends into thelipid-binding cysteine-rich domain (CRD) ofFrizzled receptors (6). Wnt ligands also bind theLrp5/6 transmembrane co-receptor, inducing itto form a complex with Frizzled (Fig. 2A). Thisinstills a conformational change in these recep-tors and enables phosphorylation by associatedprotein kinases. The phosphorylated cytoplasmicLrp tail subsequently inhibits glycogen synthasekinase 3 (GSK3) (7) and also binds theAxinprotein.In the absence of a Wnt signal, a destruction com-plex that includesAxin, anaphase-promoting com-plex (APC), and GSK3 phosphorylates b-catenin,continually targeting it for degradation by theproteasome. Inhibition of the destruction complex, aconsequence of Wnt–Frizzled–Lrp interactions, leadsb-catenin to accumulate in the nucleus (Fig. 2B).There, b-catenin governs transcriptional programsthrough association with Tcf/Lef transcription factors.In some instances, Wnt signals are transduced

independently of b-catenin—for example, duringmorphogenetic movements in vertebrate gastru-lation (8). In this pathway, Frizzled and an in-tracellular transduction component (Disheveled)are crucial, but not Lrp and b-catenin. This as-pect of Wnt signaling is evolutionarily ancientand may be involved in regulating stem cellpolarity and asymmetric division of stem cellswithin the confinement of the niche, as we dis-cuss below.Wnt signaling can be further augmented by

secreted R-spondin proteins (9, 10). R-spondins,acting through Lgr family receptors (11–13), in-hibit the transmembrane E3 ubiquitin ligasesRnf43/Znrf3 that ordinarily ubiquitinate and thusdegrade Frizzled receptors (14, 15). By antagoniz-ing Rnf43/Znrf3, R-spondins consequently stabi-lize surface Frizzled receptors and enhance Wntsignal strength (Fig. 2A) (14, 15).The fundamental core of the Wnt pathway

(Wnt, Frizzled, and downstream effectors) is evo-lutionarily ancient and is extant in the earliestmulticellular animals including ctenophores,sponge, and placozoans (16, 17), in which it me-diates basic axial patterning even in pre-bilateria(18, 19). In contrast, the R-spondin/Lgr axis isprincipally a vertebrate innovation (20). Was theR-spondin/Lgr pathway simply collateral to ver-tebrate speciation? Another possibility was thatit was evolutionarily co-opted to amplify Wnt sig-naling and thus sustain some types of adult stemcell in long-lived vertebrate species (20).

Wnt-driven transcriptional programs

In the nucleus, b-catenin interacts with Tcf/Leftranscriptional cofactors to regulate the transcrip-tion of Wnt target genes (Fig. 2B). Rather than

RESEARCH

SCIENCE sciencemag.org 3 OCTOBER 2014 • VOL 346 ISSUE 6205 1248012-1

1Hubrecht Institute, Royal Netherlands Academy of Artsand Sciences (KNAW), University Medical Centre Utrechtand CancerGenomics.nl, 3584CT Utrecht, Netherlands.2Department of Developmental Biology, Howard HughesMedical Institute, Stanford Institute for Stem Cell Biologyand Regenerative Medicine, Stanford University School ofMedicine, 265 Campus Drive, Stanford, CA 94305, USA.*Corresponding author. E-mail: rnusse@stanford.edu

conforming to a universal program, the tran-scriptional agenda imposed by b-catenin variesbetween lineages. However, several generalitiesmight exist. For instance, in Wnt-responsive stemcells it seems that b-catenin can directly inducetelomerase expression (21), causally explaining thelengthy telomeres of Wnt-driven intestinal stemcells (22) and pluripotent cells (21) and shieldingthem from genomic catastrophe.Although the phenotypic consequences of Wnt

signaling diverge between distinct lineages, sev-eral genes appear to represent generic Wnt tran-scriptional targets. Axin2 has emerged as onesuch Wnt target gene (23) that therefore servesas a reporter of ongoing Wnt signaling (24). Asdiscussed below, Axin2 (24) as well as a sec-ond gene, Lgr5 (25), can identify Wnt-responding

lineages in diverse tissues. Genetically labelingLgr5- or Axin2-expressing cells has revealed theirparticipation in tissue renewal in multiple or-gans, compellingly nominating such cells as stemcells in specific tissues. We summarize these celllabeling experiments in Table 1 and discuss threeexamples in more detail.

Intestinal stem cells

The small intestinal epithelium is the fastest pro-liferating tissue of adult mammals, being largelymade anew every 4 to 5 days (26). Villi protrudeinto the gut lumen and continually shed dif-ferentiated cells from their tips. These lossesare replenished by stem cells located in pro-liferative intestinal crypts that surround thevillus base (Fig. 3A). Wnt signals are pivotal

for the perennial renewal of the intestines, asshown by disruption of the pathway—whichleads to the abrupt cessation of proliferationin the intestinal crypts, consequently leadingto unabated loss of intestinal tissue and oftenmorbidity (27–29). Reciprocally, the Wnt co-agonist R-spondin potently stimulates intestinalproliferation in vivo (30).The crypt bottom harbors slender, cycling

“crypt base columnar” (CBC) cells (31), whichwere historically proposed to represent intes-tinal stem cells (32) (Fig. 3A). Exploiting theexpression of Wnt target gene Lgr5 in CBCs,genetic labeling of Lgr5+ crypt cells indeed dem-onstrated that these long-lived cells generate alldifferentiated intestinal cell types (25). Therefore,CBCs constitute multipotent intestinal stem cells(25) that require Wnt for proliferation (27, 33),perhaps explaining why Wnt is crucial for in-testinal renewal.Residing directly above the CBC stem cell zone

at the “+4” position is a potentially distinct popu-lation of slowly cycling cells [variously describedby molecular markers including Bmi1 (34),Hopx(35), Lrig1 (36, 37), and Tert (38, 39)] that also cangenerate all intestinal lineages (Fig. 3A).Instead of constituting irrevocably separated

lineages, it seems that Lgr5+ and +4 stem cellscan interconvert. The highly proliferative Lgr5+

CBCs appear to be the “workhorse” of daily in-testinal renewal (33). Yet, slowly cycling “reserve”+4 stem cells can be recalled to Lgr5+ status (40)and vice versa (35).Adding further complexity, the two stem cell

lineages may be partially overlapping. Lgr5+ cellscan coexpress +4 markers (such as Bmi1) (41–43).Indeed, whereas the majority of Lgr5+ cells areproliferative stem cells, a subset of Lgr5+ cellsare nondividing secretory precursors that co-express +4 markers (43). These precursors, typ-ically confined to secretory fates, can be promotedto multipotent stem cell status upon tissue dam-age to effect intestinal repair (43). This indicatesthat the developmental competence of precur-sors is not fixed but is rather labile, as we ex-plore further below.

Interfollicular epidermis

The interfollicular epidermis (IFE) is constantlyregenerated. Differentiated cells are shed fromthe surface and replaced by basal layer stemcells. Most basal layer cells transduce Wnt sig-nals, as visualized by a Wnt transcriptional re-porter and expression of Wnt target gene Axin2(44, 45). Axin2+ basal cells continuously producekeratinocytes for over 1 year in vivo and there-fore qualify as IFE stem cells (Fig. 3B) (44, 45).Certain evidence suggests that b-catenin is

crucial for epidermal proliferation and mainte-nance of IFE stem cells, both in vivo (44–46) aswell as in cell culture (47). However, extrapolat-ing a role for Wnt as an IFE self-renewal signalbased on these data has been complicated bythe fact that b-catenin operates dually in cell ad-hesion (48) as well as Wnt/b-catenin signaling.Implying a role for Wnt signaling specifically,simultaneous loss of Tcf3 and Tcf4 compromises

1248012-2 3 OCTOBER 2014 • VOL 346 ISSUE 6205 sciencemag.org SCIENCE

Fig. 1. Model of Wnt secretion, modification, and short-range signaling activity. Wnt proteins arelipid-modified by the Porcupine enzyme in the endoplasmic reticulum. Subsequently, lipid-modified Wntsare bound by the carrier proteinWls andmight be expelled in (B) secretory vesicles furnishingmembrane-boundWnt ligands or (A)might bedirectly presented as cell surface–boundWnt ligands. (C) InDrosophila,a constitutively membrane-tethered Neurotactin (Nrt)–Wingless fusion protein is able to execute allWingless functions, implying that Wnts need not be released from the membrane in order to signal.

RESEARCH | REVIEW

long-term IFE maintenance (49). Taken in col-lective, these findings suggest that IFE basal stemcell proliferation is controlled by Wnt signaling.Furthermore, basal cells produce their own Wntligands (44), implying autocrine (rather thanniche-dependent paracrine) regulation (Fig. 3B).This concept portends a type of “developmentalself-organization,” considered further below.

Mammary gland

The mammary gland constitutes another venueof tissue renewal because it undergoes cycles ofdynamic growth during puberty, pregnancy,and lactation. After lactation, the alveoli in thegland regress by involution and cell death, andthe tissue returns to a pre-pregnancy–like state.How are these cycles of regrowth continuallysustained?Initial transplantation (50) and subsequent

lineage-tracing experiments have establishedthat stem cells exist in the adult mammary epi-thelium that and they appear to be driven byWnt signaling (51) because they are designatedby Lgr5 (52–55) and Axin2 (24) in vivo and canbe expanded in vitro upon Wnt exposure (56).Axin2+ cells self-renew and continuously fuel cel-lular production during multiple cycles of preg-nancy, lactation, and involution (24), indicatingthat these cells (or a subset of them) are authenticstem cells.

Stochastic fate or invariant lineage?

The classical view of homeostatic stem cell self-renewal is exemplified by that of the hema-topoietic stem cell, which is believed to dividerarely and invariably in an asymmetric fashionto generate one new stem cell and one differen-tiated daughter. However, neither the intestinalcrypt nor the IFE abide by this rule of predeter-mined lineage choice. Each crypt contains a fixednumber of stem cells, determined by the size ofthe niche. Each of these stem cells divides everyday to generate two new “potential” stem cells.Chance decides which of these will stay withinthe niche at the crypt bottom and which arepushed out of the niche (57, 58). This process istermed “neutral competition” and ensures that(i) the number of available stem cells is constantand (ii) that damaged or lost stem cells are im-mediately replaced by healthy neighbors (59). Alsoin the skin, the Wnt-responding IFE stem cellsappear to divide stochastically to generate prolif-erating and differentiating daughter cells withequal probability (44, 60). Thus, whether any givenstem cell daughter will continue self-renewingis left to a throw of the dice—not destiny.

Plasticity within the stem cell hierarchy

In models of the hematopoietic hierarchy (61),all arrows “point away” from the stem cell, im-plying that once cells give up their stem cellidentity, there is no way back. Intestinal cells donot abide by this rule. Although Dll1+ secretoryprogenitors are typically short-lived precursorsthat are confined to secretory fates (Fig. 3A), ifcrypt stem cells are depleted, Dll1+ secretory pro-genitors can regain Lgr5+ stem cell status in vivo

SCIENCE sciencemag.org 3 OCTOBER 2014 • VOL 346 ISSUE 6205 1248012-3

Fig. 2.Wnt signalingmechanisms. (A) Wnt reception on the cell surface.Wnt ligands bind to the Frizzledand Lrp5/6 receptors, activating downstream signaling. The membrane proteins Znrf3 and Rnf43 areubiquitin ligases that continually down-regulate Frizzleds through ubiquitination. Binding of R-spondins toZnrf3 and Rnf43 and the Lgr4/5/6 receptor relieves Znrf3 and Rnf43 activity, thus stabilizing Frizzleds.(B). Wnt signaling in target cells. (Left) In the absence of Wnt, a destruction complex consisting of Axin,APC, and GSK3 resides in the cytoplasm, where it binds to and phosphorylates b-catenin, which is thendegraded. Dvl (Disheveled) is required for activating the pathway as well. In the nucleus,Tcell factor (TCF)is in an inactive state as the consequence of binding to the repressor Groucho. (Right) Binding ofWnt to itsreceptors induces the association of Axin with phosphorylated lipoprotein receptor-related protein (LRP).The destruction complex falls apart, and b-catenin is stabilized, subsequently binding TCF in the nucleus toup-regulate target genes, including Axin2 and Lgr5.

Table 1. Wnt-responsive tissue stem cells identified by means of lineage tracing.

Tissue Stem cell Marked by Reference

Intestine Crypt base columnar cell Lgr5 (25)Mammary gland Basal cell Axin2, Lgr5 (24, 50–53)Stomach Basal pyloric cell Lgr5 (85)Interfollicular epidermis Basal cell Axin2 (44, 45)Central nervous system Radial glial cell Axin2 (98)Hair follicle Outer bulge cell Lgr5 (99)Kidney Nephron segment-specific stem cell Lgr5, Axin2 (100, 101)Cochlea Tympanic border Axin2 (102)Ovary Hilum ovarian surface epithelial cell Lgr5 (103)Taste bud Circumvallate papilla stem cell

in posterior tongueLgr5 (104, 105)

RESEARCH | REVIEW

(62). In vitro, this process can be mimicked by apulse of high-dose Wnt3a (62). Similar observa-tions were reported for a noncycling secretoryprecursor (43). Therefore, lineage-restricted pro-genitors may gain an expansion of responsibilityupon injury, reacquiring multipotency and long-term self-renewal to perpetuate tissue repair. Thestem cell phenotype is not indelibly imprintedbut may be ordained unto other cell types duringthe regenerative response.

Wnt and tissue regeneration in theearliest animals

Even in the earliest animals, it seems that Wntcoordinates repair after injury in certain tissuesand imparts positional information crucial forshaping proper regeneration. Upon resectionof their tail, planarian flatworms regeneratetheir tail anew. Nonetheless, upon depletion ofb-catenin, a head is inappropriately regeneratedin lieu of the tail, leading to the generation ofmultiple heads (63, 64). Therefore, Wnt ensuresthat the original anatomic plan is faithfully re-stored after injury. Analogously, Wnt10a is up-regulated upon zebrafish tail resection and isnecessary for robust tail regeneration (65). Like-wise, Wnt3 is crucial for apical regeneration ofamputated hydra (66). Compellingly, in hydra theWnt source is apoptotic cells at the site of thewound, which provide Wnt3 to drive proliferationof underlying cells and thus regeneration (67).Therefore, Wnt elegantly links tissue loss withhow such tissue might be restored.

The sources of Wnt ligands: Redefinitionof the stem cell niche

Wnt signals, by virtue of their short-range na-ture, constitute ideal “niche factors,” controllingimmediately adjacent stem cells and thus per-mitting parsimonious command of cell fate.For instance, Lgr5+ CBCs in the crypt bottom

are evenly interspersed with Paneth cells (68)that, together with nonepithelial lineages includ-ing mesenchymal cells (69–71), supply Wnt pro-teins to maintain adjacent Lgr5+ CBCs (Fig. 3A).The localized spatial reach of Wnt dictates thatonly cells near the crypt bottom remain stemcells. Cells migrating upward out of the reach ofWnt signaling differentiate.This “Wnt-adjacency” model can also hold true

in regeneration. Upon bladder injury, stromalcells directly underlying the bladder basal epithe-lium up-regulate Wnt ligands, signaling to adja-cent basal stem cells to initiate bladder epitheliumregeneration (72). Therefore, stem and niche cellsare paired in both spatial location and function.Nevertheless, the past few years have seen a

revision to the monolithic notion that stem cellsneed always be controlled by an extrinsic niche.Axin2+ IFE stem cells express their own Wntligands, which they require for self-renewal (44).Therefore, they may continuously drive theirown self-renewal in an autocrine fashion (Fig.3B), akin to how Wnt3a-expressing axial stemcells in the early vertebrate embryo in essenceact as their own niche (73) to sustain their ownself-renewal during axis elongation and upon

1248012-4 3 OCTOBER 2014 • VOL 346 ISSUE 6205 sciencemag.org SCIENCE

Fig. 3. The provenance of Wnt ligands in the stem cell niche. (A) At the intestinal crypt bottom,Paneth cells and stromal cells supply Wnt ligands to sustain the self-renewal of Lgr5+ crypt stemcells, with which they are intercalated. The local Wnt signaling domain spatially delimits stem cellactivity to the crypt bottom. Cells moving upward begin to differentiate, although they may be re-stored to stem cell status upon returning to the crypt bottom. (B) Within the interfollicular epidermis,basal-layer stem cells express Wnt ligands and thus continuously induce their own self-renewal and actas their own niche. Basal stem cells also express long-range Wnt antagonists that diffuse to suprabasallayers, basally limiting the Wnt signaling field and “self-organizing” the stratified epidermal architecture.(C) Image of Dkk3 immunostaining (red) in epidermis of Axin2-CreERT2/Rosa26-mTmG mice exposedto Tamoxifen at P21 to induce labeled clones (green) and chased for 2 months (P77). (C) is courtesy ofX. Lim (44).

RESEARCH | REVIEW

serial transplantation (74, 75). In the case ofthe intestinal crypt, Lgr5+ CBCs generate Wnt-producing Paneth cells (25). This underpins whysingle Lgr5+ CBCs can form intestinal organoidsin vitro in the absence of niche cells (76)—becausestem cells can elaborate their own niche.

Developmental self-organization

These observations imply that in some contexts,stem cells can self-organize their own niche andautonomously perpetuate their activity. In thiscapacity, stem cells qualify as fundamental “unitsof development” (61) because they can incipientlyseed developing tissues anew. In the developingDrosophila intestine, the first cell division under-taken by the earliest intestinal stem cells is toasymmetrically generate a niche cell as well asanother stem cell (77). “Auto-niche generation”enables single stem cells to take root in the nascenttissue, expand to form islands of undifferentiatedstem cells, and subsequently fuel intestinal devel-opment (77).If stem cells can self-organize their own niche

and continue ever-expanding in vivo, this couldbe easily subverted to lead to tumorigenesis. Con-trary to this notion of unchecked stem cell ex-pansion, in each intestinal crypt there existsapproximately 14 Lgr5+ CBCs and 10 Panethcells per crypt bottom (57) and eight Lgr5+ stemcells per stomach pylorus pit (78). How is stemcell expansion so precisely constrained in thesteady state? By way of example, in the skin, IFEbasal stem cells produce not only their own Wntligands but also diffusible Wnt antagonists, in-cluding Dkk molecules (Fig. 3C) (44). Therefore,adjacent basal stem cells signal via Wnt to sus-tain one another in the basal compartment, yetDkk diffuses to the suprabasal layer to limit theWnt signaling field and likely to induce differen-tiation in that domain (44). Consequently, stemcell activity is spatially confined to the basal layer,and Dkk might prevent expansion of the stemcell territory beyond that layer (Fig. 3B). In sodoing, IFE stem cells might self-organize the strat-ified architecture of the epidermis.

Orienting asymmetric stem cell divisionsby Wnt signaling within the niche

Stem cell numbers also may be numerically lim-ited within the niche by Wnt-imposed asymmetric

stem cell divisions. Drosophila germline stemcells divide next to neighboring hub cells. Thedaughter cell closest to the hub cell remains astem cell, whereas the distal cell invariably dif-ferentiates; this asymmetric division is orientedby the Wnt signaling component APC (79). Ex-periments using a local source of Wnt in cellculture imply a conserved mechanism extendingto mammals. A localized Wnt signal can orient amouse embryonic stem cell (ESC) to divide asym-metrically by placing the centrosomes at oppositeends of the cell, thus orienting the mitotic spin-dle of the dividing cell (Fig. 4) (80). This gen-erates a Wnt-proximal and Wnt-distal daughtercell, the latter out of contact with the signal. Inthe Wnt-proximal cell, Wnt signaling maintainsthe stem cell fate, whereas the distal daughterdifferentiates (80). The orientation of stem celldivision is therefore coupled with the positionand fate of the dividing cell through the samesignal. Therefore, in some tissues Wnt signalsmay orient stem cell divisions within the nichein an asymmetric fashion, delimiting stem cellnumber and ensuring a proper ratio of stemcells to their committed progeny.

Growing Wnt-dependent stem cells

The roles of Wnt in stem cell self-renewal or lineage-specific differentiation in diverse tissues in vivoare manifold; therefore, Wnt signals have foundpractical use in manipulating stem cell devel-opmental programs in vitro. From a pragmaticperspective, because Wnt induces stem cell self-renewal in certain organs, it enables the in vitropropagation of such cells. For example, mam-mary gland stem cells can be expanded in vitroin the presence of Wnt protein and retain theirability to reconstitute the entire mammary organafter transplantation (56).Similarly, pluripotent naïve ESCs from the

rodent blastocyst may be cultivated in vitro indefined conditions by combining Wnt agonists[either Wnt protein (81) or GSK3 inhibitors (82)]with either leukemia inhibitory factor (LIF) sig-nals or mitogen-activated protein kinase (MAPK)/extracellular signal–regulated kinase (ERK) in-hibitors (83), as exemplified by the “2i” cultureregime for serum-free ESC culture (82).Because of the primacy of Wnt in instructing

the intestinal stem cell fate, Lgr5+ CBC stem

cells can be expanded in an R-spondin1–basedthree-dimensional culture system in ever-growingorganoids, or “mini-guts” (76), in which crypt andvillus domains are established containing normalratios of the appropriate cell types, whereas self-renewal kinetics closely resemble the in vivosituation (84). Comparable protocols have beenestablished for Lgr5+ cells derived from the sto-mach (85), liver (86), and pancreas (87). Whencells within organoids produce Wnt (for example,Paneth cells that secrete Wnt3 in small intestinalorganoids), the addition of R-spondin suffices.When organoids harbor no endogenous sourceof Wnt (for example, colon organoids), exoge-nous Wnt3a is added in addition to R-spondin(88). Transplantation of clonal (single Lgr5+ stemcell–derived) organoids derived from colon andliver has confirmed that the cultured organoidsretain their physiological functions (86, 89). Thisagain provides evidence for substantial develop-mental self-organization—namely, that single Lgr5+

intestinal stem cells carry the morphogenetic in-formation to create a structured tissue of com-plex architecture and diverse lineages.Proper lineage differentiation and crypt-villus

organization within small intestinal organoidsrelies on an interesting property of R-spondin1.Namely, it augments preexisting domains of Wntsignaling in the crypt bottom (68) rather thaninducing Wnt signaling de novo. Thus, when cellsexit the crypt bottom–like structures of mini-gutsand the spatial reach of Wnt, intestinal differ-entiation occurs normally (76), accounting forproper organoid architecture. In contrast, spa-tially uniform Wnt activation by GSK3 inhibi-tion captures a rather homogeneous populationof Lgr5+ stem cells in vitro in the absence of dif-ferentiated lineages (90).That being said, Wnt does not ubiquitously

instruct stem cell self-renewal and, in multiplecases, instead drives differentiation—for instance,Wnt instead stimulates primed pluripotent stemcells (including human ESCs) to differentiate intoprimitive streak (91, 92).

Concluding remarks

The emergent view is that lipid-modified Wntsignals predominantly act over short ranges tolocally control cell behavior, economically con-trolling stem cells within the spatial confines of

the niche. The short range of Wntaction implies a parsimonious mod-el of niche organization and tissuephysiology. Namely, in particular tis-sues it seems that Wnt-dependentstem cells are spatially restricted tothe vicinity of the Wnt-producingniche, physically delimiting the stemcell compartment and preventingunauthorized stem cell expansion.When a stem cell divides, chancemay dictate which (if any) of its suc-cessors are ousted from its niche, asin the intestines (57), stomach (78),and skin (44). In other lineages, Wntitself may orient stem cells to divideasymmetrically (80), conveniently

SCIENCE sciencemag.org 3 OCTOBER 2014 • VOL 346 ISSUE 6205 1248012-5

Fig. 4. A local Wnt signal induces asymmetric cell division. A cell exposed to a localWnt source distributes Wnt signaling components to the side of the cell where Wnttouches. This orients the mitotic spindle and centrosomes during division. The daughtercell close to the Wnt source maintains nuclear b-catenin and stem cell gene expression, whereas the distal cellaway from Wnt loses expression of such genes.

RESEARCH | REVIEW

anchoring Wnt-proximal stem cells to the nicheand ensuring proper spatial allocation of stemcells and differentiated progeny.In certain organs, stem cells exiting the niche

become deprived of Wnt and therefore differen-tiate. Nonetheless, developmental plasticity mayyet remain because early committed precursorscan flexibly regain stem cell status upon tissuedamage in vivo (43, 62, 93, 94) or Wnt3a treat-ment ex vivo, in some instances (62). This isprofound because it indicates that lineage po-tential is an amorphous property in vivo; lineage-restricted precursors can gain an expansion ofresponsibility upon injury and become fullyfledged multipotent stem cells once more. Intra-vital microscopy has documented that uponintestinal or hair follicle damage, precursors arespatially recalled to the stem cell niche (95, 96),upon which they reenter the niche signaling do-main and presumably become promoted to stemcell status as a consequence, although the respon-sible signals remain largely elusive. Therefore,lineage barriers between stem cell and progeni-tor states are not always stringent in vivo andcan be traversed during times of tissue damageand repair (43, 62, 93, 94). If stem cell and pro-genitor fates are interconvertible upon nichecontact (97), then stem cell status might not bean intrinsic entitlement but rather a positionalprivilege—reflectingwhether a cell is currently inthe embrace of the niche.Nonetheless the notion of a “niche” must be

refined because some stem cells may act as or es-tablish their own niche ab initio, portending un-expected developmental self-organization. Suchintrinsically programmed stemcell behavior couldunderpin emergence of complex patterned tissuesduring development and/or regeneration, as inthe Drosophila (77) and mouse (76) intestines.The above findings identify an integral pro-

gram for tissue generation, regeneration, andrenewal. In evolutionary antiquity, the core of theWnt pathway emerged in the simplest multicel-lular organisms (16, 17). Accruing evidence sug-gests that in the earliest metazoa, Wnt was anancestral “symmetry-breaking” signal that sep-arated otherwise-symmetric embryos into twohalves (the anterior versus the posterior domain)and in so doing enabled the evolutionary emer-gence of axially patterned animals (18, 19). Simplyput, the primordial role of Wnt signaling in theearliest animals was pattern formation (during tis-sue generation) and patternmaintenance (duringtissue regeneration), as evinced by howWnt estab-lishes a bodily pattern in hydra and planaria andenables the reconstitution of such pattern upontissue regeneration (63, 64, 66). In long-lived ver-tebrates, this ancestral pattern maintenance pro-gram has since been extended to tissue renewal,in which Wnt permits several tissues, includingthe skin and intestines, to be continuously re-plenished and thus maintained over a lifetime.

REFERENCES AND NOTES

1. C. P. Leblond, B. E. Walker, Renewal of cell populations.Physiol. Rev. 36, 255–276 (1956). pmid: 13322651

2. R. Najdi et al., A uniform human Wnt expression libraryreveals a shared secretory pathway and unique signaling

activities. Differentiation 84, 203–213 (2012). doi: 10.1016/j.diff.2012.06.004; pmid: 22784633

3. J. C. Gross, V. Chaudhary, K. Bartscherer, M. Boutros,Active Wnt proteins are secreted on exosomes. Nat. Cell Biol.14, 1036 (2012). doi: 10.1038/ncb2574

4. C. Korkut et al., Trans-synaptic transmission of vesicularWnt signals through Evi/Wntless. Cell 139, 393–404 (2009).doi: 10.1016/j.cell.2009.07.051; pmid: 19837038

5. C. Alexandre, A. Baena-Lopez, J. P. Vincent, Patterningand growth control by membrane-tethered Wingless.Nature 505, 180–185 (2014). doi: 10.1038/nature12879;pmid: 24390349

6. C. Y. Janda, D. Waghray, A. M. Levin, C. Thomas, K. C. Garcia,Structural basis of Wnt recognition by Frizzled. Science337, 59–64 (2012). doi: 10.1126/science.1222879;pmid: 22653731

7. J. L. Stamos, M. L. Chu, M. D. Enos, N. Shah, W. I. Weis,Structural basis of GSK-3 inhibition by N-terminalphosphorylation and by the Wnt receptor LRP6. eLife 3,e01998 (2014). doi: 10.7554/eLife.01998; pmid: 24642411

8. R. van Amerongen, R. Nusse, Towards an integrated view ofWnt signaling in development. Development 136, 3205–3214(2009). doi: 10.1242/dev.033910; pmid: 19736321

9. O. Kazanskaya et al., R-Spondin2 is a secreted activator ofWnt/beta-catenin signaling and is required for Xenopusmyogenesis. Dev. Cell 7, 525–534 (2004). doi: 10.1016/j.devcel.2004.07.019; pmid: 15469841

10. K.-A. Kim et al., R-Spondin family members regulate theWnt pathway by a common mechanism. Mol. Biol. Cell19, 2588–2596 (2008). doi: 10.1091/mbc.E08-02-0187;pmid: 18400942

11. K. S. Carmon, X. Gong, Q. Lin, A. Thomas, Q. Liu, R-spondinsfunction as ligands of the orphan receptors LGR4 andLGR5 to regulate Wnt/beta-catenin signaling. Proc. Natl.Acad. Sci. U.S.A. 108, 11452–11457 (2011). doi: 10.1073/pnas.1106083108; pmid: 21693646

12. W. de Lau et al., Lgr5 homologues associate with Wntreceptors and mediate R-spondin signalling. Nature 476,293–297 (2011). doi: 10.1038/nature10337; pmid: 21727895

13. A. Glinka et al., LGR4 and LGR5 are R-spondin receptorsmediating Wnt/b-catenin and Wnt/PCP signalling.EMBO Rep. 12, 1055–1061 (2011). doi: 10.1038/embor.2011.175; pmid: 21909076

14. H.-X. Hao et al., ZNRF3 promotes Wnt receptor turnover inan R-spondin-sensitive manner. Nature 485, 195–200(2012). doi: 10.1038/nature11019; pmid: 22575959

15. B. K. Koo et al., Tumour suppressor RNF43 is a stem-cellE3 ligase that induces endocytosis of Wnt receptors.Nature 488, 665–669 (2012). doi: 10.1038/nature11308;pmid: 22895187

16. J. F. Ryan et al., The genome of the ctenophore Mnemiopsisleidyi and its implications for cell type evolution. Science342, 1242592 (2013). doi: 10.1126/science.1242592;pmid: 24337300

17. M. Srivastava et al., The Trichoplax genome and the nature ofplacozoans. Nature 454, 955–960 (2008). doi: 10.1038/nature07191; pmid: 18719581

18. C. P. Petersen, P. W. Reddien, Wnt signaling and the polarityof the primary body axis. Cell 139, 1056–1068 (2009).doi: 10.1016/j.cell.2009.11.035; pmid: 20005801

19. T. W. Holstein, H. Watanabe, S. Ozbek, Signaling pathwaysand axis formation in the lower metazoa. Curr. Top. Dev. Biol.97, 137–177 (2011). doi: 10.1016/B978-0-12-385975-4.00012-7; pmid: 22074605

20. W. B. de Lau, B. Snel, H. C. Clevers, The R-spondin proteinfamily. Genome Biol. 13, 242 (2012). doi: 10.1186/gb-2012-13-3-242; pmid: 22439850

21. K. Hoffmeyer et al., Wnt/b-catenin signaling regulatestelomerase in stem cells and cancer cells. Science 336,1549–1554 (2012). doi: 10.1126/science.1218370;pmid: 22723415

22. A. G. Schepers, R. Vries, M. van den Born, M. van de Wetering,H. Clevers, Lgr5 intestinal stem cells have high telomeraseactivity and randomly segregate their chromosomes.EMBO J. 30, 1104–1109 (2011). doi: 10.1038/emboj.2011.26;pmid: 21297579

23. B. Lustig et al., Negative feedback loop of Wnt signalingthrough upregulation of conductin/axin2 in colorectaland liver tumors. Mol. Cell. Biol. 22, 1184–1193 (2002).doi: 10.1128/MCB.22.4.1184-1193.2002; pmid: 11809809

24. R. van Amerongen, A. N. Bowman, R. Nusse, Developmentalstage and time dictate the fate of Wnt/b-catenin-responsivestem cells in the mammary gland. Cell Stem Cell 11, 387–400(2012). doi: 10.1016/j.stem.2012.05.023; pmid: 22863533

25. N. Barker et al., Identification of stem cells in small intestineand colon by marker gene Lgr5. Nature 449, 1003–1007(2007). doi: 10.1038/nature06196; pmid: 17934449

26. C. P. Leblond, C. E. Stevens, The constant renewal of theintestinal epithelium in the albino rat. Anat. Rec. 100,357–377 (1948). doi: 10.1002/ar.1091000306;pmid: 18906253

27. J. H. van Es et al., A critical role for the Wnt effectorTcf4 in adult intestinal homeostatic self-renewal. Mol. Cell.Biol. 32, 1918–1927 (2012). doi: 10.1128/MCB.06288-11;pmid: 22393260

28. V. Korinek et al., Depletion of epithelial stem-cellcompartments in the small intestine of mice lackingTcf-4. Nat. Genet. 19, 379–383 (1998). doi: 10.1038/1270;pmid: 9697701

29. T. Fevr, S. Robine, D. Louvard, J. Huelsken, Wnt/b-catenin isessential for intestinal homeostasis and maintenance ofintestinal stem cells. Mol. Cell. Biol. 27, 7551–7559 (2007).doi: 10.1128/MCB.01034-07; pmid: 17785439

30. K. A. Kim et al., Mitogenic influence of human R-spondin1 onthe intestinal epithelium. Science 309, 1256–1259 (2005).doi: 10.1126/science.1112521; pmid: 16109882

31. J. Paneth, Ueber die secernirenden Zellen des Dünndarm-Epithels. Archiv für Mikroskopische Anatomie 31, 113–191 (1887).doi: 10.1007/BF02955706

32. H. Cheng, C. P. Leblond, Origin, differentiation and renewal ofthe four main epithelial cell types in the mouse smallintestine. V. Unitarian theory of the origin of the fourepithelial cell types. Am. J. Anat. 141, 537–561 (1974).doi: 10.1002/aja.1001410407; pmid: 4440635

33. K. S. Yan et al., The intestinal stem cell markers Bmi1 andLgr5 identify two functionally distinct populations. Proc. Natl.Acad. Sci. U.S.A. 109, 466–471 (2012). doi: 10.1073/pnas.1118857109; pmid: 22190486

34. E. Sangiorgi, M. R. Capecchi, Bmi1 is expressed in vivo inintestinal stem cells. Nat. Genet. 40, 915–920 (2008).doi: 10.1038/ng.165; pmid: 18536716

35. N. Takeda et al., Interconversion between intestinal stem cellpopulations in distinct niches. Science 334, 1420–1424(2011). doi: 10.1126/science.1213214; pmid: 22075725

36. A. E. Powell et al., The pan-ErbB negative regulator Lrig1 isan intestinal stem cell marker that functions as a tumorsuppressor. Cell 149, 146–158 (2012). doi: 10.1016/j.cell.2012.02.042; pmid: 22464327

37. V. W. Y. Wong et al., Lrig1 controls intestinal stem-cellhomeostasis by negative regulation of ErbB signalling.Nat. Cell Biol. 14, 401–408 (2012). doi: 10.1038/ncb2464;pmid: 22388892

38. D. T. Breault et al., Generation of mTert-GFP mice as a modelto identify and study tissue progenitor cells. Proc. Natl. Acad.Sci. U.S.A. 105, 10420–10425 (2008). doi: 10.1073/pnas.0804800105; pmid: 18650388

39. R. K. Montgomery et al., Mouse telomerase reverse transcriptase(mTert) expression marks slowly cycling intestinal stemcells. Proc. Natl. Acad. Sci. U.S.A. 108, 179–184 (2011).doi: 10.1073/pnas.1013004108; pmid: 21173232

40. H. Tian et al., A reserve stem cell population in smallintestine renders Lgr5-positive cells dispensable. Nature478, 255–259 (2011). doi: 10.1038/nature10408;pmid: 21927002

41. S. Itzkovitz et al., Single-molecule transcript counting ofstem-cell markers in the mouse intestine. Nat. Cell Biol. 14,106–114 (2012). doi: 10.1038/ncb2384; pmid: 22119784

42. J. Muñoz et al., The Lgr5 intestinal stem cell signature:Robust expression of proposed quiescent ‘+4’ cell markers.EMBO J. 31, 3079–3091 (2012). doi: 10.1038/emboj.2012.166; pmid: 22692129

43. S. J. Buczacki et al., Intestinal label-retaining cells aresecretory precursors expressing Lgr5. Nature 495, 65–69(2013). doi: 10.1038/nature11965; pmid: 23446353

44. X. Lim et al., Interfollicular epidermal stem cells self-renewvia autocrine Wnt signaling. Science 342, 1226–1230 (2013).doi: 10.1126/science.1239730; pmid: 24311688

45. Y. S. Choi et al., Distinct functions for Wnt/b-catenin in hairfollicle stem cell proliferation and survival and interfollicularepidermal homeostasis. Cell Stem Cell 13, 720–733 (2013).doi: 10.1016/j.stem.2013.10.003; pmid: 24315444

46. K. B. Jensen et al., Lrig1 expression defines a distinctmultipotent stem cell population in mammalian epidermis.Cell Stem Cell 4, 427–439 (2009). doi: 10.1016/j.stem.2009.04.014; pmid: 19427292

47. A. J. Zhu, F. M. Watt, Expression of a dominant negativecadherin mutant inhibits proliferation and stimulates terminal

1248012-6 3 OCTOBER 2014 • VOL 346 ISSUE 6205 sciencemag.org SCIENCE

RESEARCH | REVIEW

differentiation of human epidermal keratinocytes. J. Cell Sci.109, 3013–3023 (1996). pmid: 9004036

48. S. Beronja et al., RNAi screens in mice identify physiologicalregulators of oncogenic growth. Nature 501, 185–190(2013). doi: 10.1038/nature12464; pmid: 23945586

49. H. Nguyen et al., Tcf3 and Tcf4 are essential for long-termhomeostasis of skin epithelia. Nat. Genet. 41, 1068–1075(2009). doi: 10.1038/ng.431; pmid: 19718027

50. M. Shackleton et al., Generation of a functional mammarygland from a single stem cell. Nature 439, 84–88 (2006).doi: 10.1038/nature04372; pmid: 16397499

51. N. M. Badders et al., The Wnt receptor, Lrp5, is expressedby mouse mammary stem cells and is required to maintainthe basal lineage. PLOS One 4, e6594 (2009). doi: 10.1371/journal.pone.0006594; pmid: 19672307

52. K. E. de Visser et al., Developmental stage-specificcontribution of LGR5+ cells to basal and luminal epitheliallineages in the postnatal mammary gland. J. Pathol. 228,300–309 (2012). doi: 10.1002/path.4096; pmid: 22926799

53. V. Plaks et al., Lgr5-expressing cells are sufficient andnecessary for postnatal mammary gland organogenesis.Cell Reports 3, 70–78 (2013). doi: 10.1016/j.celrep.2012.12.017; pmid: 23352663

54. A. C. Rios, N. Y. Fu, G. J. Lindeman, J. E. Visvader, In situidentification of bipotent stem cells in the mammary gland.Nature 506, 322–327 (2014). doi: 10.1038/nature12948;pmid: 24463516

55. A. Van Keymeulen et al., Distinct stem cells contribute tomammary gland development and maintenance. Nature 479,189–193 (2011). doi: 10.1038/nature10573; pmid: 21983963

56. Y. A. Zeng, R. Nusse, Wnt proteins are self-renewal factorsfor mammary stem cells and promote their long-termexpansion in culture. Cell Stem Cell 6, 568–577 (2010).doi: 10.1016/j.stem.2010.03.020; pmid: 20569694

57. H. J. Snippert et al., Intestinal crypt homeostasis results fromneutral competition between symmetrically dividing Lgr5stem cells. Cell 143, 134–144 (2010). doi: 10.1016/j.cell.2010.09.016; pmid: 20887898

58. C. Lopez-Garcia, A. M. Klein, B. D. Simons, D. J. Winton,Intestinal stem cell replacement follows a pattern ofneutral drift. Science 330, 822–825 (2010). doi: 10.1126/science.1196236; pmid: 20929733

59. A. M. Klein, B. D. Simons, Universal patterns of stem cellfate in cycling adult tissues. Development 138, 3103–3111(2011). doi: 10.1242/dev.060103; pmid: 21750026

60. E. Clayton et al., A single type of progenitor cell maintainsnormal epidermis. Nature 446, 185–189 (2007). doi:10.1038/nature05574; pmid: 17330052

61. I. L. Weissman, Stem cells: Units of development, unitsof regeneration, and units in evolution. Cell 100, 157–168(2000). doi: 10.1016/S0092-8674(00)81692-X; pmid: 10647940

62. J. H. van Es et al., Dll1+ secretory progenitor cells revert tostem cells upon crypt damage. Nat. Cell Biol. 14, 1099–1104(2012). doi: 10.1038/ncb2581; pmid: 23000963

63. C. P. Petersen, P. W. Reddien, Smed-bcatenin-1 is requiredfor anteroposterior blastema polarity in planarianregeneration. Science 319, 327–330 (2008). doi: 10.1126/science.1149943; pmid: 18063755

64. K. A. Gurley, J. C. Rink, A. Sánchez Alvarado, b-Catenindefines head versus tail identity during planarianregeneration and homeostasis. Science 319, 323–327(2008). doi: 10.1126/science.1150029; pmid: 18063757

65. C. L. Stoick-Cooper et al., Distinct Wnt signaling pathways haveopposing roles in appendage regeneration. Development 134,479–489 (2007). doi: 10.1242/dev.001123; pmid: 17185322

66. T. Lengfeld et al., Multiple Wnts are involved in Hydraorganizer formation and regeneration. Dev. Biol. 330, 186–199(2009). doi: 10.1016/j.ydbio.2009.02.004; pmid: 19217898

67. S. Chera et al., Apoptotic cells provide an unexpectedsource of Wnt3 signaling to drive hydra head regeneration.Dev. Cell 17, 279–289 (2009). doi: 10.1016/j.devcel.2009.07.014; pmid: 19686688

68. T. Sato et al., Paneth cells constitute the niche for Lgr5stem cells in intestinal crypts. Nature 469, 415–418 (2011).doi: 10.1038/nature09637; pmid: 21113151

69. H. F. Farin, J. H. Van Es, H. Clevers, Redundant sourcesof Wnt regulate intestinal stem cells and promote formationof Paneth cells. Gastroenterology 143, 1518–1529.e7 (2012).doi: 10.1053/j.gastro.2012.08.031; pmid: 22922422

70. Z. Kabiri et al., Stroma provides an intestinal stem cell niche inthe absence of epithelial Wnts. Development 141, 2206–2215(2014). doi: 10.1242/dev.104976; pmid: 24821987

71. A. K. San Roman, C. D. Jayewickreme, L. C. Murtaugh,R. A. Shivdasani, Wnt secretion from epithelial cells andsubepithelial myofibroblasts is not required in the mouseintestinal stem cell niche in vivo. Stem Cell Rev. 2, 127–134(2014). doi: 10.1016/j.stemcr.2013.12.012

72. K. Shin et al., Hedgehog/Wnt feedback supports regenerativeproliferation of epithelial stem cells in bladder. Nature 472,110–114 (2011). doi: 10.1038/nature09851; pmid: 21389986

73. B. L. Martin, D. Kimelman, Brachyury establishes theembryonic mesodermal progenitor niche. Genes Dev. 24,2778–2783 (2010). doi: 10.1101/gad.1962910; pmid: 21159819

74. N. Cambray, V. Wilson, Axial progenitors with extensivepotency are localised to the mouse chordoneural hinge.Development 129, 4855–4866 (2002). pmid: 12361976

75. N. Cambray, V. Wilson, Two distinct sources for a populationof maturing axial progenitors. Development 134, 2829–2840(2007). doi: 10.1242/dev.02877; pmid: 17611225

76. T. Sato et al., Single Lgr5 stem cells build crypt-villusstructures in vitro without a mesenchymal niche. Nature 459,262–265 (2009). doi: 10.1038/nature07935; pmid: 19329995

77. D. Mathur, A. Bost, I. Driver, B. Ohlstein, A transient nicheregulates the specification of Drosophila intestinal stem cells.Science 327, 210–213 (2010). doi: 10.1126/science.1181958;pmid: 20056890

78. M. Leushacke, A. Ng, J. Galle, M. Loeffler, N. Barker, Lgr5+

gastric stem cells divide symmetrically to effect epithelialhomeostasis in the pylorus. Cell Reports 5, 349–356 (2013).doi: 10.1016/j.celrep.2013.09.025; pmid: 24209744

79. Y. M. Yamashita, D. L. Jones, M. T. Fuller, Orientation ofasymmetric stem cell division by the APC tumor suppressorand centrosome. Science 301, 1547–1550 (2003).doi: 10.1126/science.1087795; pmid: 12970569

80. S. J. Habib et al., A localized Wnt signal orients asymmetricstem cell division in vitro. Science 339, 1445–1448 (2013).doi: 10.1126/science.1231077; pmid: 23520113

81. D. ten Berge et al., Embryonic stem cells require Wntproteins to prevent differentiation to epiblast stem cells.Nat. Cell Biol. 13, 1070–1075 (2011). doi: 10.1038/ncb2314;pmid: 21841791

82. Q.-L. Ying et al., The ground state of embryonic stem cellself-renewal. Nature 453, 519–523 (2008). doi: 10.1038/nature06968; pmid: 18497825

83. J. Wray, T. Kalkan, A. G. Smith, The ground state ofpluripotency. Biochem. Soc. Trans. 38, 1027–1032 (2010).doi: 10.1042/BST0381027; pmid: 20658998

84. T. Sato, H. Clevers, Growing self-organizing mini-guts from asingle intestinal stem cell: Mechanism and applications.Science 340, 1190–1194 (2013). doi: 10.1126/science.1234852; pmid: 23744940

85. N. Barker et al., Lgr5+ve stem cells drive self-renewal inthe stomach and build long-lived gastric units in vitro.Cell Stem Cell 6, 25–36 (2010). doi: 10.1016/j.stem.2009.11.013; pmid: 20085740

86. M. Huch et al., In vitro expansion of single Lgr5+ liver stemcells induced by Wnt-driven regeneration. Nature 494,247–250 (2013). doi: 10.1038/nature11826; pmid: 23354049

87. M. Huch et al., Unlimited in vitro expansion of adult bi-potentpancreas progenitors through the Lgr5/R-spondin axis.EMBO J. 32, 2708–2721 (2013). doi: 10.1038/emboj.2013.204; pmid: 24045232

88. T. Sato et al., Long-term expansion of epithelial organoidsfrom human colon, adenoma, adenocarcinoma, andBarrett’s epithelium. Gastroenterology 141, 1762–1772 (2011).doi: 10.1053/j.gastro.2011.07.050; pmid: 21889923

89. S. Yui et al., Functional engraftment of colon epitheliumexpanded in vitro from a single adult Lgr5+ stem cell.Nat. Med. 18, 618–623 (2012). doi: 10.1038/nm.2695;pmid: 22406745

90. X. Yin et al., Niche-independent high-purity cultures ofLgr5+ intestinal stem cells and their progeny. Nat. Methods 11,106–112 (2014). doi: 10.1038/nmeth.2737; pmid: 24292484

91. K. M. Loh et al., Efficient endoderm induction from humanpluripotent stem cells by logically directing signals controllinglineage bifurcations. Cell Stem Cell 14, 237–252 (2014).doi: 10.1016/j.stem.2013.12.007; pmid: 24412311

92. K. C. Davidson et al., Wnt/b-catenin signaling promotesdifferentiation, not self-renewal, of human embryonic stemcells and is repressed by Oct4. Proc. Natl. Acad. Sci. U.S.A.109, 4485 (2012).

93. T. Nakagawa, M. Sharma, Y. Nabeshima, R. E. Braun,S. Yoshida, Functional hierarchy and reversibility within themurine spermatogenic stem cell compartment. Science328, 62–67 (2010). doi: 10.1126/science.1182868;pmid: 20299552

94. J. Cheng et al., Centrosome misorientation reduces stemcell division during ageing. Nature 456, 599–604 (2008).doi: 10.1038/nature07386; pmid: 18923395

95. L. Ritsma et al., Intestinal crypt homeostasis revealed atsingle-stem-cell level by in vivo live imaging. Nature 507,362–365 (2014). doi: 10.1038/nature12972; pmid: 24531760

96. P. Rompolas, K. R. Mesa, V. Greco, Spatial organizationwithin a niche as a determinant of stem-cell fate. Nature502, 513–518 (2013). doi: 10.1038/nature12602;pmid: 24097351

97. V. P. Losick, L. X. Morris, D. T. Fox, A. Spradling, Drosophilastem cell niches: A decade of discovery suggests a unifiedview of stem cell regulation. Dev. Cell 21, 159–171 (2011).doi: 10.1016/j.devcel.2011.06.018; pmid: 21763616

98. A. N. Bowman, R. van Amerongen, T. D. Palmer, R. Nusse,Lineage tracing with Axin2 reveals distinct developmental andadult populations of Wnt/b-catenin-responsive neural stemcells. Proc. Natl. Acad. Sci. U.S.A. 110, 7324–7329 (2013).doi: 10.1073/pnas.1305411110; pmid: 23589866

99. V. Jaks et al., Lgr5 marks cycling, yet long-lived, hair folliclestem cells. Nat. Genet. 40, 1291–1299 (2008). doi: 10.1038/ng.239; pmid: 18849992

100. N. Barker et al., Lgr5+ve stem/progenitor cells contribute tonephron formation during kidney development. Cell Reports2, 540–552 (2012). doi: 10.1016/j.celrep.2012.08.018;pmid: 22999937

101. Y. Rinkevich et al., In vivo clonal analysis reveals lineage-restricted progenitor characteristics in mammalian kidneydevelopment, maintenance, and regeneration. Cell Reports7, 1270–1283 (2014). doi: 10.1016/j.celrep.2014.04.018;pmid: 24835991

102. T. A. Jan et al., Tympanic border cells are Wnt-responsiveand can act as progenitors for postnatal mouse cochlearcells. Development 140, 1196–1206 (2013). doi: 10.1242/dev.087528; pmid: 23444352

103. A. Flesken-Nikitin et al., Ovarian surface epithelium at thejunction area contains a cancer-prone stem cell niche.Nature 495, 241–245 (2013). doi: 10.1038/nature11979;pmid: 23467088

104. K. K. Yee et al., Lgr5-EGFP marks taste bud stem/progenitorcells in posterior tongue. Stem Cells 31, 992–1000 (2013).doi: 10.1002/stem.1338; pmid: 23377989

105. N. Takeda et al., Lgr5 Identifies Progenitor Cells Capableof Taste Bud Regeneration after Injury. PLOS One 8,e66314 (2013). doi: 10.1371/journal.pone.0066314;pmid: 23824276

ACKNOWLEDGMENTS

We thank R. van Amerongen for insightful comments. The authorsare supported by the Howard Hughes Medical Institute and theCalifornia Institute for Regenerative Medicine (R.N.), the Fannieand John Hertz Foundation (K.M.L.), the U.S. National ScienceFoundation (K.M.L.), the Davidson Institute for Talent Development(K.M.L.), the European Union (H.C.), and the CancerGenomics.nlprogram (H.C.). H.C. is an inventor on several patent applicationsthat cover culturing methods for Wnt-dependent stem cells,filed by the Royal Netherlands Academy of Arts and Sciences.

10.1126/science.1248012

SCIENCE sciencemag.org 3 OCTOBER 2014 • VOL 346 ISSUE 6205 1248012-7

RESEARCH | REVIEW

RESEARCH

22 AUGUST 2014 • VOL 345 ISSUE 6199 889SCIENCE sciencemag.org

ILL

US

TR

AT

ION

: K

.SU

TL

IFF

/SCIENCE

BACKGROUND: Decades

o f l a b o r a t o r y a n d

clinical investigation

have led to success-

ful therapies using

hematopoietic stem

cells (HSCs), but few

other cell therapies have

transitioned from experimental to stan-

dard clinical care. Providing patients with

autologous rather than allogeneic HSCs re-

duces morbidity and mortality, and in some

circumstances broader use could expand

the range of conditions amenable to HSC

transplantation. The availability of a homo-

geneous supply of mature blood cells would

also be advantageous. An unlimited supply

of pluripotent stem cells (PSCs) directed

to various cell fates holds great promise

as source material for

cell transplantation

and minimally invasive

therapies to treat a va-

riety of disorders. In

this Review, we discuss

past experience and

challenges ahead and examine the extent

to which hematopoietic stem cell trans-

plantation and cell therapy for diabetes,

liver disease, muscular dystrophies, neuro-

degenerative disorders, and heart disease

would be affected by the availability of pre-

cisely differentiated PSCs.

ADVANCES: Although it is not yet possible

to differentiate PSCs to cells with character-

istics identical to those in the many organs

that need replacement, it is likely a matter

of time before these “engineering” problems

can be overcome. Experience with cell ther-

apies, both in the laboratory and the clinic,

however, indicate that many challenges re-

main for treatment of diseases other than

those involving the hematopoietic system.

Use of differentiated pluripotent stem cells in replacement therapy for treating disease

STEM CELL THERAPY

Ira J. Fox,* George Q. Daley, Steven A. Goldman, Johnny Huard, Timothy J. Kamp,

Massimo Trucco

REVIEW SUMMARY

Unlimited populations of differentiated PSCs should facilitate blood therapies and

hematopoietic stem cell transplantation, as well as the treatment of heart, pancreas,

liver, muscle, and neurologic disorders. However, successful cell transplantation will require

optimizing the best cell type and site for engraftment, overcoming limitations to cell migration

and tissue integration, and possibly needing to control immunologic reactivity (challenges

indicated in red). iPSC, induced PSC; ES cells, embryonic stem cells.

There are issues of immunity, separate from

controlling graft rejection, and identify-

ing the optimal cell type for treatment in

the case of muscular dystrophies and heart

disease. Optimization is also needed for the

transplant site, as in diabetes, or when deal-

ing with disruption of the extracellular ma-

trix in treating degenerative diseases, such as

chronic liver and heart disease. Finally, when

the pathologic process is diffuse and migra-

tion of transplanted cells is limited, as is the

case with Alzheimer’s disease, amyotrophic

lateral sclerosis, and the muscular dystro-

phies, identifying the best means and location

for cell delivery will require further study.

OUTLOOK: Considering the pace of progress

in generating transplantable cells with a ma-

ture phenotype, and the availability of PSC-

derived lineages in sufficient mass to treat

some patients already, the challenges to scal-

ing up production and eliminating cells with

tumor-forming potential are probably within

reach. However, generation of enough cells to

treat an individual patient requires time for

expansion, differentiation, selection, and test-

ing to exclude contamination by tumorigenic

precursors. Current methods are far too long

and costly to address the treatment of acute

organ injury or decompensated function. Im-

mune rejection of engrafted cells, however, is

likely to be overcome through transplanta-

tion of autologous cells from patient-derived

PSCs. Availability of PSC-derived cell popu-

lations will have a dramatic effect on blood

cell transfusion and the use of hematopoietic

stem cell transplantation, and it will likely

facilitate treatment of diabetes, some forms

of liver disease and neurologic disorders,

retinal diseases, and possibly heart disease.

Close collaboration between scientists and

clinicians—including surgeons and interven-

tional radiologists—and between academia

and industry will be critical to overcoming

challenges and to bringing new therapies to

patients in need. ■

The list of author affiliations is available in the full

article online.

*Corresponding author. E-mail: foxi@upmc.eduCite this article as I. J. Fox et al., Science 345, 1247391 (2014); DOI: 10.1126/science. 1247391

Read the full article at http://dx.doi.org/10.1126/science.1247391

ON OUR WEB SITE

SP

ECIAL SERIES: STEM C

ELL

S

Published by AAAS

REVIEW◥

STEM CELL THERAPY

Use of differentiated pluripotentstem cells in replacement therapyfor treating diseaseIra J. Fox,1* George Q. Daley,2,3,4 Steven A. Goldman,5,6 Johnny Huard,7

Timothy J. Kamp,8 Massimo Trucco9

Pluripotent stem cells (PSCs) directed to various cell fates holds promise as sourcematerial for treating numerous disorders. The availability of precisely differentiatedPSC-derived cells will dramatically affect blood component and hematopoietic stem celltherapies and should facilitate treatment of diabetes, some forms of liver disease andneurologic disorders, retinal diseases, and possibly heart disease. Although an unlimitedsupply of specific cell types is needed, other barriers must be overcome. This reviewof the state of cell therapies highlights important challenges. Successful celltransplantation will require optimizing the best cell type and site for engraftment,overcoming limitations to cell migration and tissue integration, and occasionallyneeding to control immunologic reactivity, as well as a number of other challenges.Collaboration among scientists, clinicians, and industry is critical for generating newstem cell–based therapies.

Induced pluripotent stem cells (PSCs) aregenerated by reprogramming somatic cellsto a pluripotent state by transient expres-sion of pluripotency factors. These cells canself-renew indefinitely and are able to dif-

ferentiate into any cell lineage (1, 2). The abilityto generate PSCs from individual patients anddifferentiate them into an unlimited supply oftissue and organ-specific cells capable of circum-venting immunologic rejection after transplan-tation could facilitate development of cell-basedtherapies for the treatment of a variety of de-bilitating disorders and dramatically change thepractice of medicine.Before these cells can be used in the clinic, a

variety of barriers must be overcome. For manydiseases, it is not yet possible to differentiatePSCs to cells with characteristics identical to those

in the organs that need replacement. There arealso challenges like scaling up production, elimi-nating cells with tumor-forming potential, anddecreasing the time needed for expansion, dif-ferentiation, selection, and testing. Furthermore,treatment of a genetic mutation using autolo-gous cells will often require genetic manipula-tion, which might result in changes that couldincrease cancer risk.Some form of immune suppression may also

be required to control cell loss after transplan-tation, whether due to rejection, an immuneresponse to a genetically corrected protein, orrecurrence of autoimmunity, with destruction ofthe transplant, as might be the case for diabetes.The standard signs of rejection used in solid or-gan transplantation are not likely to be usefulbecause the sensitivity of functional changes hasbeen shown, after islet transplantation, to beinadequate to diagnose rejection before dam-age to the engrafted cells is irreversible (3). Ofcourse, it might be possible to engineer PSC-derived grafts, with the usual caveats concern-ing activating oncogenes, so that they would beimmunologically inert and identifiable by anarray of imaging strategies.Although decades of laboratory and clinical

investigation have led to successful therapiesusing hematopoietic cells, few other cell ther-apies have transitioned from experimental tostandard clinical care. Here, we discuss thepresent state of cell therapy in the context ofhaving available differentiated PSC-derived cells.The “gold standard,” blood and hematopoieticstem cell (HSC) transplantation, is highlightedfirst, followed by an examination of cell ther-

apy for diabetes, liver disease, neurologic andretinal disorders, muscular dystrophies, andheart disease.

Hematopoietic cell-based therapies

Many of the principles of cell transplantationderive from our long experience with transfu-sion of blood products. Infused red blood cells(RBCs), platelets, and HSCs are the most widelyemployed cellular therapies in use today. Therelative ease of HSC transplantation (HSCT) de-rives in large part from the intrinsic potential ofHSCs to home to and integrate into native niches,give rise to differentiated progeny, and thereafterto egress into the circulation. Thus, HSCT avoidsthe challenges of restoring integrity and functionof more anatomically complex organs like thelung, heart, liver, and brain.Despite the successes of HSCT, isolated HSCs

cannot be expanded to the degree needed, al-though there has been limited success with cordblood. Furthermore, allogeneic HSCT is asso-ciated with considerable treatment-related mor-bidity and mortality. Thus, transplantation withautologous HSCs for the same indications wouldeliminate the major morbidities of immunemismatch and could potentially expand therange of conditions, including cancers, ame-nable to HSCT. One of the most promising ap-plications of somatic cell reprogramming isthe production of customized pluripotent stemcells followed by gene correction (4), differenti-ation into HSCs, and autotransplant with inten-tion to cure any one of dozens of inherited geneticdisorders of the blood-forming system. Such aproof of principle has been achieved for treatingmurine models of severe combined immune de-ficiency and sickle cell anemia (4, 5). HSCs havebeen derived from murine embryonic stem cellsthat manifest the cardinal features of clonal self-renewal and multilineage lymphoid-myeloid en-graftment in primary and secondary irradiatedhosts (6, 7). The derivation of HSCs from humanPSCs has proven elusive, although several exam-ples of low-level engraftment have been reported(8–10). Although true HSCs are not yet available,methods exist to produce RBCs (11, 12) and plate-lets (13) in vitro that are suitable for transfusion.To eliminate the costly and sometimes un-reliable system of volunteer blood supply, aswell as the risk of transmission of infectiousagents, a reliable method for generating an in-exhaustible, uniform supply of pathogen-freeblood products has tremendous appeal. Ulti-mately, advances in in vitro cell manufactureshould soon be able to reduce costs and enablean off-the-shelf supply.

Diabetes

One of several approaches to resolving thelong-term complications associated with diabe-tes has been beta cell replacement by allogeneicwhole-pancreas or isolated pancreatic islet trans-plantation, using immune suppression in anattempt to control rejection and recurrent auto-immune destruction of the transplanted tissue.Cadaver pancreas transplantation has been

RESEARCH

SCIENCE sciencemag.org 22 AUGUST 2014 • VOL 345 ISSUE 6199 1247391-1

1Department of Surgery, Children’s Hospital of Pittsburghand McGowan Institute for Regenerative Medicine, Universityof Pittsburgh, Pittsburgh, PA, USA. 2Boston Children’sHospital and Dana Farber Cancer Institute, Boston, MA, USA.3Department of Biological Chemistry and MolecularPharmacology, Harvard Medical School Broad Institute,Cambridge, MA, USA. 4Howard Hughes Medical Institute,Chevy Chase, MD, USA. 5Center for TranslationalNeuromedicine, The University of Rochester Medical Center,Rochester, NY, USA. 6Center for Basic and TranslationalNeuroscience, University of Copenhagen, Denmark. 7StemCell Research Center, Department of Orthopaedic Surgery,University of Pittsburgh, School of Medicine, Pittsburgh, PA,USA. 8Stem Cell and Regenerative Medicine Center, Cellularand Molecular Arrhythmia Research Program, Department ofMedicine, School of Medicine and Public Health, University ofWisconsin, Madison, WI, USA. 9Division of Immunogenetics,Children’s Hospital of Pittsburgh, University of Pittsburgh,Pittsburgh, PA, USA.*Corresponding author. E-mail: foxi@upmc.edu

shown to reduce the complications associatedwith type 1 diabetes. However, it is a complexsurgical procedure associated with morbidityand measureable mortality (14). An importantlesson from this experience, however, is that con-ventional immune suppression is able to inhibitrecurrent destruction of insulin-producing cellswhile controlling allograft rejection.After years of laboratory and clinical investi-

gation, islet transplantation has become a real-istic alternative therapy, which, over the pastseveral decades has become increasingly moresuccessful (15–17). Experience with autologousislet transplantation after total or near-total pan-creatic resection for severe chronic pancreatitisand allogeneic islet transplantation for type1 diabetes have both been instructive (18). Thepreferred site for implantation is the liver, andengraftment is accomplished by minimally in-vasive transcutaneous catheter infusion throughthe liver into the portal vein. This approach isassociated with a small but measureable risk ofhemorrhage and partial portal vein thrombosis.More important, an immediate blood-mediatedinflammatory response and pathologic activa-tion of the coagulation system results in a largeloss of the infused islets as soon as they comeinto direct contact with blood (19). As the yieldof islets from a chronically injured pancreas isalready reduced, this constitutes a limiting fac-tor in trying to reach an islet mass sufficient forinsulin independence after autologous islet trans-plantation. It is also a serious issue after isletallotransplantation, where success often re-quires multiple infusions from different donorsand transplantation of a much greater islet massto achieve insulin independence. Although it isnot completely understood why so many moreislets are required for initial success in allo- thanfor autotransplants, this barrier to successfulinsulin replacement by cell therapy should beresolvable with an inexhaustible supply of do-nor beta cells.Failure to achieve long-term insulin indepen-

dence is more problematic. Five-year insulin in-dependence rates after islet allotransplantationin selected centers are approaching 50% (17, 20).These improved outcomes have resulted fromthe transplantation of a larger islet mass andmore effective control of rejection and auto-immunity, with late graft loss being attributedto toxicity to beta cells arising from medica-tions that suppress immunity and an inabilityto diagnose or treat recurrent disease or rejec-tion. Advances in donor cell imaging by geneticengineering of PSCs and the development of newstrategies for controlling the immune responseshould resolve some of these issues.However, experience with islet autotransplan-

tation indicates a more serious problem for long-term function of engrafted islet cells, independentof the need to control rejection or recurrent auto-immune destruction (21). Although short-terminsulin independence has been accomplished,almost all patients are back on insulin therapyafter 5 years. This loss of graft function may beattributable to chronic stimulation of an initially

marginal intrahepatic beta cell mass that producesmetabolic deterioration and loss of beta cells(18). Transplantation of a larger mass of isletsmay alleviate this problem and result inindefinite graft function. However, the site ofengraftment in the liver may also be respon-sible for poor long-term survival. Islets ectopi-cally engrafted in the liver are known to producea number of pathologic histologic abnormalities,including extensive amyloid deposition withinthe islets (22). Thus, it remains unclear whetherwhole islets can remain functionally intact in theliver over time.The use of PSC-derived beta cells could resolve

many of the above issues. Beta cells that havebeen dissociated and isolated from intact isletssuccessfully engraft within the liver lobule. Thisis in contrast to how intact islets engraft, whichis by partially remaining within the portal cir-culation (Fig. 1) and requiring neovasculariza-tion. Isolated cells may not be as susceptible toforming amyloid deposits, and transplantationof a large mass of individual beta cells might avoid

metabolic exhaustion and apoptosis. Dissociatedcells do not, however, engraft when infused asindividual beta cells but require reaggregation(23). In addition, individual beta cells functionless well when isolated. It has been reported thatmore insulin by a factor of 30 is released frombeta cells within intact islets as compared withthat released from purified single beta cells, andreaggregated beta cells and beta cells aggregatedwith alpha cells respond better to glucose chal-lenge by a factor of 4 than do single beta cellsalone (24). Thus, to treat type 1 diabetes, it may benecessary to transplant PSC-derived beta cells asaggregates or possibly to transplant them withPSC-derived non–beta cells.Whether PSC–derived surrogate beta cells will

have the same capacity to engraft in the liver-likeprimary beta cells is not known. Because theyare small, they might pass through the liver intothe lungs, as has been described after hepatocytetransplantation. Other sites of transplantation,like the gastric submucosal space, might offersome advantages over the portal vein (Fig. 2).

1247391-2 22 AUGUST 2014 • VOL 345 ISSUE 6199 sciencemag.org SCIENCE

Fig. 1. Comparison of intrahepatic engraftment of intact islets versus isolated beta cells afterintraportal transplantation. (A) Transplanted islet (in brown) within the blood vessel of the liverafter intraportal transplantation [From (28), with permission]. (B) Induced PSC-derived beta cells (inbrown) engraft as cells scattered throughout the liver parenchyma [From (150), with permission].

Fig. 2. Transplantation of insulin-producing cells. Illustration of traditional portal-vein infusion ofinsulin-producing cells for engraftment in the liver versus endoscopic placement of transplanted cellsin the gastric submucosal space.

RESEARCH | REVIEW

Surrogate beta cells could be implanted in thegastic submucosal space endoscopically, wherethey would be accessible for biopsy to monitorthe graft status (25).Without concomitant use of immune suppres-

sion, autoimmunity will mediate destruction ofgrafts composed of even autologous cells (26).Whether autoimmunity can be suppressed ina way that is less detrimental to the insulin-producing cells than conventional drug-basedallogeneic protection is as yet unknown. Strat-egies for potentially controlling this process byinducing tolerance are in development (27). Tomodulate immunologically mediated problems,transgenes might help confer long-term func-tion and reduce early loss of the graft in diabeticrecipients. Candidates for this approach are likethe ones targeting the tissue factor and co-agulation cascade, or exerting anti-inflammatoryactivity, that have already been tested in clonedpigs (28). An alternative approach is to encapsulatethe PSC-derived insulin-producing surrogates,providing a protective barrier from immune re-jection and autoimmunity while allowing freeexchange of nutrients, waste, and, most impor-tant, insulin and glucose. Unfortunately, thusfar these approaches have shown mixed results,at least after transplantation of intact islets indiabetic patients with low levels of circulatingC-peptide, and rarely achieve full and lastinginsulin independence (29). A variation on thisencapsulation and transplant strategy, usingPSC-derived beta cells, will soon be tested inclinical trials for the treatment of type 1 dia-betes (30).

Liver Disease

Hepatocyte transplantation holds great promiseas a therapy for individuals with life-threateningliver diseases, where organ transplantation isoften the only available treatment option. Pa-tients with both acute liver failure and liver-based inborn errors of metabolism, leading tolife-threatening extrahepatic complications, areideal candidates for cell therapy. Numerous studiesin rodents have shown that hepatocyte transplan-tation can reverse acute fulminant hepatic failure(31) and correct liver-based metabolic deficien-cies (32–37). Because the native architecture ofthe liver is intact in these diseases, the transplantprocedure involves simple injection of hepatocytesthrough the portal vein into the liver, where thecells integrate into the host liver and are indis-tinguishable from the native liver cells (38, 39).Infusion of hepatocytes is a minimally invasiveprocedure, so it can be performed on severely illpatients with relatively low risk. Because the na-tive liver is not removed, the transplanted hepato-cytes only need to improve liver function enoughto stabilize a patient with acute liver failure untiltheir own liver is able to regenerate or to replacethe enzyme deficiency that is missing in liver-based metabolic disorders, a goal similar to that ofgene therapy.Clinical trials of hepatocyte transplantation

have only demonstrated the long-term safety ofthe procedure (40–46). Transplanted hepatocytes

have not restored liver function enough to cir-cumvent the need for organ replacement inpatients with liver failure, and transplantationhas only resulted in partial correction of meta-bolic disorders (47). Efficacy has been limited byrelatively poor initial and long-term engraft-ment and an inability to monitor graft functionin real time, which makes diagnosing and treat-ing rejection nearly impossible. In acute liverfailure, the severity of liver dysfunction requiresthat the transplanted hepatocytes function im-mediately, and the lack of a clinically relevantdisease model means that the number of cellsthat need to engraft to reverse hepatic failure isessentially unknown (48, 49). Although animalmodels of metabolic liver disease recapitulate thehuman processes better, achieving an adequatelevel of engraftment is still a problem because thenumber of donor cells that can be safely trans-planted into the liver at any one time is small,usually less than 1% of the liver mass (50). Trans-plantation of a larger number of cells can leadto severe portal hypertension and translocationof cells into the systemic circulation with em-bolization to the lungs. Liver-directed radiationhas been proposed as a way to facilitate repopu-lation of the native liver by transplanted hepato-cytes, whose viability is relatively short onceisolated (51). Preparative radiation inhibits hosthepatocyte proliferation and induces postmitotichepatocyte death, allowing donor hepatocytesto preferentially proliferate and repopulate theirradiated host liver. This strategy has been em-ployed to completely correct a rodent model ofCrigler-Najjar syndrome (52).An immediately available, inexhaustible sup-

ply of functioning donor hepatocytes would al-low early intervention in patients with hepatic

failure and would allow hepatocytes to be infusedover a longer period of time. It is possible thatdaily large-scale portal-vein PSC-derived hepato-cyte infusions could provide the hepatocyte massnecessary to normalize the encephalopathy, co-agulation defects, and other life-threateningconsequences of hepatic failure and completelycorrect liver-based enzyme deficiencies. In addi-tion, unrestricted availability of donor hepato-cytes could allow treatment of patients withless severe, but debilitating, liver-based meta-bolic disorders, which are not now consideredcandidates for organ transplantation, such asphenylketonuria and partial urea cycle disor-ders. Whether acute hepatic failure is associatedwith changes in the local microenvironmentthat might interfere with engraftment and func-tion of transplanted hepatocytes is not yet knownand will need to be addressed with further clinicalexperience.Although PSC-derived hepatocytes may help

advance cell therapies for acute liver failure andmetabolic liver diseases, the vast majority of pa-tients who are in need of life-saving interven-tion are patients with end-stage cirrhosis andchronic hepatic failure. After infusion throughthe portal vein, hepatocytes have a difficult timeentering the hepatic cords through the patho-logically expanded extracellular matrix (Fig. 3)present in advanced cirrhosis (53). As a result,transplantation by this route generates severeportal hypertension and may produce portalthrombosis. Animal studies suggest that trans-plantation by direct injection into an extrahe-patic site, such as the spleen, can circumvent thisengraftment difficulty, improve liver function, andprolong survival in end-stage cirrhosis (43, 54, 55).Even then, however, transplanted hepatocytes

SCIENCE sciencemag.org 22 AUGUST 2014 • VOL 345 ISSUE 6199 1247391-3

Fig. 3. Cirrhosis and chronic liver failure. Normal control liver (behind) and liver with advancedcirrhosis (in front), highlighting the structural differences that make liver failure from cirrhosis difficult totreat by intrahepatic cell therapy [Images courtesy of R. Markin, University of Nebraska Medical Center].

RESEARCH | REVIEW

provide function for only a period of months.Because it is not clear that transplanted cellscan function for any sustained period of timein the abnormal environment produced by he-patic failure and portal hypertension, the samefate may await hepatocytes transplanted intoother extra-anatomic locations, such as thelymph node (56), although diversion of the por-tal circulation may be able to enhance survival(55, 57). For a variety of reasons, anecdotal re-ports of hepatocyte transplantation in humanswith end-stage cirrhosis has not produced anymeasurable level of success (58). Treatment ofchronic liver failure might benefit from trans-plantation of a large mass of hepatocytes into adecellularized human or animal liver scaffold(59). When repopulated with donor hepatocytesand nonparenchymal cells, the biohybrid graftmight then be vascularized through the portalcirculation as an engineered internal auxil-iary liver graft. Because this strategy wouldleave the native cirrhotic liver in place, it wouldstill leave unresolved the management of co-

existing portal hypertension and the risk ofdeveloping hepatocellular carcinoma in thenative liver.

Neurologic and retinal diseases

The brain is arguably the most difficult of or-gans in which to employ stem cell–based ther-apeutics; the myriad connections of its neuronsand their complex interdependency with mac-roglia, including astrocytes, oligodendrocytes,and glial progenitor cells, defy precise structuralreconstitution. Neurodegenerative disorders, inparticular, include diseases of both single andmultiple phenotypes, the heterogeneity of whichcan dictate how amenable each might be to cell-based therapeutics. Some prototypic degenera-tive dementias, such as Alzheimer’s disease andLewy body disease, involve a multitude of neu-ronal phenotypes—and in some cases glial as well,in multisystem atrophy. These multiphenotypicdisorders span anatomic and functional domainsand may exhibit both contiguous and trans-synaptic patterns of spread. For these reasons,

they remain poor targets for neuronal replace-ment strategies, at least for the near future.However, many diseases of the brain involve

single cell types, and these conditions lend them-selves to phenotype-specific cell replacement,whether by transplantation or by the inductionof endogenous neural stem or progenitor cells(60) (Fig. 4). Degenerative disorders in which theloss of single phenotypes predominate, especial-ly those in which a single region is differentiallyaffected—such as Parkinson’s disease (PD), inwhich nigrostriatal neurons are lost before otherneurons, and Huntington’s disease (HD), in whichmedium spiny neuronal loss and striatal atrophybecome apparent long before the onset of cor-tical neuronal loss—have proven more amenable tophenotype-specific cell replacement (61, 62). Mem-ory disorders, which can involve loss of the basalforebrain cholinergic neurons projecting to thehippocampus, have responded to PSC-derivedcholinergic neuronal replacement in rodents(63). Yet the memory loss of early Alzheimer’sand frontotemporal dementia typically heralds

1247391-4 22 AUGUST 2014 • VOL 345 ISSUE 6199 sciencemag.org SCIENCE

Fig. 4. PSC-derived glial and neural cells. Schematic representation of how PSC-derived glial and neural cells populations might be employed to treat avast array of neurologic and retinal diseases.

RESEARCH | REVIEW

deterioration across all cognitive modalities;thus, isolated replacement of cholinergic neu-rons is likely to prove effective in only veryselected cases of memory loss, which remaindifficult to identify at the onset. Nonetheless,for each of these subcortical neuronal pheno-types, animal studies have proven sufficientlypromising to justify both the production of cellsappropriate for clinical transplantation and thedesign of clinical trials by which to evaluatetheir safety and efficacy.Clinical trials of cell transplantation have al-

ready been performed in PD and HD (61, 62).These trials used fetal tissue dissected from theregions of interest, so that the specific cell typesneeded comprised but a fraction of the cells de-livered. Perhaps as a result, these early trialsyielded largely disappointing results with con-siderable variability in outcome and frequentuntoward side effects, the most prominent beingrefractory dyskinesias after fetal mesencephalictissue grafts in PD. Nonetheless, some trans-planted PD patients did experience durable ben-efit, providing at least a proof of principle infavor of dopaminergic neuronal transplanta-tion. The generation of midbrain dopaminergicneurons (64, 65) and medium spiny neurons(66, 67) from human PSCs (64–67) thus permita leap forward in the availability of well-definedengraftable neurons of relative phenotypic homo-geneity. This advance should enable a new gener-ation of therapeutic trials less compromised bydonor variability and compositional heterogeneity.Some disorders affect single neuronal types

but are so dispersed throughout the central ner-vous system (CNS) as to present problems of de-livery. Several primary epilepsies may derivefrom deficits in g-aminobutyric acid (GABAergic)interneuron numbers and function. A number ofgroups have asked whether transplantation ofhealthy, migration-competent interneurons intoepileptic cortex may provide benefit to patientswith medication-refractory epilepsy. The pro-duction of GABAergic interneurons from PSCs(68) has permitted assessment of this promis-ing approach to epileptic therapy. That said,GABAergic neurons comprise a plethora of func-tionally distinct phenotypes, thus rendering theeffects of their transplantation on host neuralcircuits unpredictable.Phenotype-specific disorders of the eye may

prove more straightforward targets for cell ther-apy. Loss of the retinal pigment epithelium (RPE)in age-related macular degeneration has beena particular target of interest in that efficientprotocols for generating RPE cells from PSCshave been developed (69, 70). PSC-derived RPEsare now in trials for macular degeneration. Asmore efficient protocols are developed for produc-ing specific retinal phenotypes from PSCs, a broadvariety of both intrinsic retinal disorders and opticneuropathies may prove appropriate targets forcell replacement.Other disorders of single phenotype are not

attractive as targets for cell-based therapy be-cause of their multicentric pathology, the non-migratory nature of potential replacement cells,

or both. The motor neuronopathies, such asamyotrophic lateral sclerosis (ALS) and spinalmuscular atrophy, are examples. Spinal motorneurons may be generated from PSCs (71–73),and yet their clinical utility has been limited bythe multisegmental nature of motor neuron lossin these diseases and by our current inability todirect long-distance axonal regrowth and targetreinnervation. Cell-based treatment approachesfor ALS have thus shifted away from neuronalreplacement toward the delivery of astrocytes,with the goal of correcting underlying glial meta-bolic deficiencies that may contribute to diseaseprogression in ALS (74). If successful, such studiesmay herald the use of PSC-derived glia for ALSand related disorders, in which neurons maybe the paracrine victims of glial dysfunction.Diseases of glia may prove especially accessi-

ble targets for cell-based therapy. The myelindiseases, which involve the loss or dysfunction ofoligodendrocytes, are among the most prevalentand disabling conditions in neurology and maybe particularly appropriate targets for replace-ment (75). They include multiple sclerosis andwhite matter stroke, the cerebral palsies, and thehereditary pediatric leukodystrophies. As a result,oligodendrocyte progenitor cells (OPCs), which cangive rise to myelinogenic oligodendrocytes, havebecome of great interest as potential therapeuticagents (76–78). Recently, transplantation of OPCsderived from human PSCs rescued otherwise le-thally hypomyelinated shiverermice (79). As such,one might expect that efforts under way usingtissue-derived cells to treat myelin disorders, suchas childhood Pelizaeus-Merzbacher disease (80)and adult progressive multiple sclerosis, maybe supplanted by PSC-derived OPCs.Because OPCs produce astrocytes as well as

oligodendrocytes and are highly migratory, theymight prove useful in rectifying the demyelination-associated enzymatic deficiencies of the lyso-somal storage disorders, such as Krabbe disease,metachromatic leukodystrophy, and Tay-Sachsdisease, as well as the astrocytic pathology ofvanishing white matter disease (81). In partic-ular, transcription activator-like effector nuclease(TALEN) and clustered regularly interspacedshort pallendromic repeats (CRISPR)/Cas geneediting technologies (82) may allow correction ofmutations in PSCs, thus allowing autologous andallogeneic therapy to be assessed across the en-tire range of hereditary leukodystrophies.Their promise notwithstanding, introducing

PSC-derived cells into the postnatal and matureCNS has its own set of risks, which include im-mune rejection, the induction of neuroepithelialneoplasms and teratomas (65, 80, 83), hetero-topic neuronal differentiation and epileptogen-esis, and the mass effect that may accompanyexuberant cell expansion. Neural stem cells havebeen reported to have escaped to the ventricularsystem and subarachnoid space, adventitiousgrowths that may present a risk of hydroceph-alus, syringomyelia, or surface venous compres-sion, as well as disruption of cerebrospinal fluidflow and waste clearance. The list goes on andserves to highlight the degree to which we must

always be concerned that any cell therapeuticnot result in unintended consequences.

Muscular Dystrophies

Duchenne muscular dystrophy (DMD) is a de-vastating, heritable X-linked muscle disease char-acterized by progressive muscle weakness dueto the lack of dystrophin expression at the sar-colemma of muscle fibers (84–87). Althoughvarious approaches have been investigated fordelivering dystrophin to dystrophic muscle, thereis still no effective treatment that alleviates pro-gression of the disease. Satellite cells play a keyrole in development of skeletal muscle duringembryogenesis and in regeneration of musclefibers during postnatal life. Upon isolation andculturing, satellite cells re-enter the cell cycle asmyoblasts and eventually fuse to form myotubesin vitro (88). Transplantation of normal myo-blasts into dystrophin-deficient muscle can createa reservoir of normal myoblasts capable of fusingwith one another and with dystrophic musclefibers, restoring dystrophin to the muscle (89–97).Myoblast transplantation in animal models and

DMD patients can transiently deliver dystrophinand improve the strength of injected dystrophicmuscle, but this approach is hindered by poorcell survival and limited migration of injectedcells from the original injection site (89–97). Re-searchers have used preparative irradiation orinjection of myonecrotic agents into the trans-plantation site to improve efficiency (89–98), andalthough these approaches have led to an im-provement in restoration of dystrophin in mdxmice, success remains limited.Many scientists consider satellite cells to be

the only myogenic cells responsible for musclegrowth, regeneration, and repair; however, re-ports suggest that other cell populations may beable to support muscle regeneration, making thema potential alternative source of cells (99–103).Cells from bone marrow (101, 103), blood vessels(100, 104–106), the neuronal compartment (99, 102),and connective tissue can differentiate toward amyogenic lineage, suggesting that transplanta-tion of these nonsatellite cell populations couldbe effective in improving muscle regeneration(107). Animal studies have shown that many havea very limited capacity to enhance muscle regen-eration after transplantation; however, mesoan-gioblasts, pericytes, and muscle-derived stem cells(MDSCs) survive postimplantation (Fig. 5) andrepair skeletal muscle after injury and disease(96, 100, 104–106, 108). As a result, their humancounterparts have been isolated and are under-going clinical trials for the treatment of stressurinary incontinence (108, 109).Most cell therapies for treating DMD employ

an intramuscular route of administration; how-ever, systemic delivery would be a far superiormethod because it is the heart, intercostal mus-cles, and diaphragm that are involved in the earlydeath of DMD patients and are almost impos-sible to reach by intramuscular injection. Systemicdelivery of stem cells can achieve widespreaddelivery and leads to dystrophin expression invarious muscle groups, although with varying

SCIENCE sciencemag.org 22 AUGUST 2014 • VOL 345 ISSUE 6199 1247391-5

RESEARCH | REVIEW

levels of success (110). This technology is stillunder development and is technically challeng-ing, especially when using small-animal models(110, 111).Although most cell transplant studies for DMD

have focused on the use of postnatal stem cells,the progeny of PSCs may eventually be usefulfor muscle regeneration and repair. A numberof stem cell lineages can be easily generatedin vitro, but differentiation into skeletal musclehas proven to be difficult (112–116). PSCs trans-fected with myogenic regulatory factors, in par-ticular Pax3 and Pax7, undergo partial myogenicdifferentiation and participate in skeletal mus-cle regeneration (117, 118), whereas MyoD andMyf5 have also been used to produce myoblastsfrom normal and dystrophic human PSCs (119).Intramuscular transplantation of human skele-tal myogenic progenitors derived from PSCsresults in durable engraftment, contributing tothe satellite cell pool (120). Furthermore, PSCs de-rived from patients with muscular dystrophies,differentiated into satellite and mesoangioblast-like cells, have been genetically corrected and,after reintroduction into dystrophic mice, ame-liorate their dystrophic phenotype (121, 122). Al-though PSCs have the potential to create aninexhaustible source of therapeutic cells for mus-cle repair, the availability of large numbers ofmuscle progenitor cells has not been a majorlimitation. Poor cell survival and the limited abil-ity of the cells to migrate from their initial site

of injection remain the two major hurdles tomuscle cell–based therapies, even using PSC-derived cell populations.Finally, despite the lack of dystrophin at birth,

the initiation of acute, severe muscle weaknessdoes not occur in DMD patients until they havereached the age of 4 to 8 years, which coincideswith gradual exhaustion of their muscle progen-itor cells (MPCs). Impairment in the myogenicpotential of the MPCs isolated from DMD mus-cle in various animal models has been described(123–125), and recent studies indicate that sparingof the extraocular muscles in DMD patients maybe related to the existence of a subpopulation ofMPCs that that do not become exhausted withage, indicating that muscle weakness in DMDpatients is related to MPC exhaustion duringdisease progression (126). These findings, takentogether, suggest that a relationship likely ex-ists between the rapid progression of musculardystrophy and stem cell exhaustion, and thedevelopment of cell therapeutic approaches todelay the exhaustion of MPCs may represent animportant alternative avenue to explore for de-laying the onset of pathologies associated withmuscular dystrophies.

Heart Disease

Heart disease is the most common cause of deathworldwide. Because the disease can result in thereplacement of contractile cardiomyocytes (CMs)with scar tissue, cellular and regenerative thera-

pies hold great promise. A wide variety of celltypes have been investigated for their abilityto repair the heart. These include skeletal myo-blasts, bone marrow–derived cells, cardiac stemcells, and mesenchymal stem cells (127). Mosteffort has focused on treating ischemic heartdisease by infusing cells intravascularly (intra-venous, intracoronary, or retrograde coronarysinus), by intramyocardial injection (transendo-cardial catheter–based injection or epicardialinjection), or by scaffold or patch-based epicar-dial delivery to the myocardium (127).When skeletal myoblasts are transplanted by

direct epicardial injection into the heart, theyform stable skeletal muscle grafts that do notcouple to the native cardiac muscle (128), andclinical experience in the setting of coronary arterybypass surgery has not demonstrated significantbenefit (129). Initial studies in a post–myocardialinfarction (MI) mouse model showed that injec-tion of bone marrow–derived c-kit+ and lineagenegative (lin–) cells dramatically regeneratedmyocardium and improved cardiac function(130). Subsequent studies, however, failed todemonstrate myocardial regeneration using thisapproach (131, 132). Nevertheless, clinical trialsevaluating the ability of bone marrow mono-nuclear cells (BMNCs) to repair the myocardiumwere initiated, and phase 1 trials of intracoro-nary BMNC delivery via the reperfused infarct-related artery hinted at efficacy. Larger phase 2trials, however, have produced mixed results,

ranging from no effect to amodest improvement in ejec-tion fraction (133). In smallphase 1 trials, mesenchymalstem cells, autologous cardiacstem cells, and cardiosphere-derived cells have similarlyshown possible benefit withrespect to left ventricular struc-ture and function in patientspost-MI with ischemic cardio-myopathy (133).Cell survival and engraft-

ment have been major limi-tations with adult stem cellsources, and the vast majorityof transplanted cells are lostwithin days in most investiga-tions (134). Thus, robust myo-cardial regeneration has notbeen observed, and the bene-ficial effects on cardiac func-tion and structure have beenattributed to othermechanisms,largely acting by paracrine sig-naling to preserve the borderzone around the infarction, re-duce apoptosis, blunt adverseremodeling, potentially stim-ulate endogenous stem cells,modulate inflammation, andpromote angiogenesis.PSCs can differentiate into a

range of cell types relevant forcardiac repair, including CMs

1247391-6 22 AUGUST 2014 • VOL 345 ISSUE 6199 sciencemag.org SCIENCE

Fig. 5. Importance of cell type for restoration of dystrophin-expressing myofibers. Engraftment efficiency formuscle-derived stem cells (left) and myoblasts (right) after transplant.

RESEARCH | REVIEW

(135), cardiac progenitors (136, 137), endothelialcells (ECs) (138), and smooth muscle cells (SMCs)(138), which have been tested in small-animalmodels. Protocols for efficient generation of CMsfrom human PSCs have been developed (139), andpurification of CMs has been accomplished usingcell surface markers or glucose-free culture con-ditions optimized for survival of CMs (140, 141).However, PSC-derived CMs are, in general, a mix-ture of nodal-like, atrial-like and ventricular-likecells (142) that have limited proliferative capac-ity and have a relatively immature phenotypebased on electrophysiological characteristics, con-tractile performance, and metabolic profile. De-spite these limitations, mouse PSC–derived CMsare capable of integration into the mouse myo-cardium, but this is an extremely rare event (113).Survival of transplanted PSC-derived CMs hasbeen a limitation, but by modulating multiple cel-lular processes a prosurvival cocktail was de-veloped, which enabled the survival of humanPSC-derived CMs as islands in athymic rat heartspost-MI, improving cardiac function (143). Trans-planted mouse PSC–derived cardiac progenitorsimprove the function of infarcted mouse hearts(144), and, in guinea pigs, where the heart exhib-its an intrinsic heart rate closer to that in man,human PSC–derived CMs engraft and function-ally couple to host CMs to improve left ventricularfunction and reduce ventricular arrhythmias (145).Preclinical studies are beginning to test PSC

cell therapy in large-animal models of heart dis-ease. Epicardial delivery of cell sheets composedof PSC-derived CMs in post-MI pig hearts hasshown encouraging results (146), and epicardialapplication of human PSC–derived ECs and SMCs,incorporated into a fibrin-based patch, post-MI,to the hearts of immune-suppressed pigs has re-duced infarction size and improved left ventric-ular function (Fig. 6) (147). In addition, directintramyocardial injection of human PSC–derived

CMs in immune-suppressed nonhuman primatehearts post-MI has produced large areas of en-graftment and electrical coupling, but no clearimprovement in cardiac function has been ob-served (148). However, a transient increase inventricular arrhythmias has occurred, raising apotential safety concern. Transplantation of hu-man PSC–derived cardiac progenitors has beenassociated with multilineage regeneration in theimmunosuppressed rhesus monkey heart post-MI (149). Thus, large-animal studies are begin-ning to define which cell preparations and deliverystrategies hold promise.Overall, clinical studies using adult cell sources

have not yet demonstrated robust clinical ben-efits. Animal studies using PSC-derived cardiaccells, however, have shown promising results, withsome evidence of improvement in left ventricularfunction and structure. Strategies that com-bine cells with bioengineered patches or decel-lularized cardiac matrices are now also beingexplored, especially for use in congenital heartdisease. At this time, however, the mechanismsby which different cell populations and deliverystrategies affect the various cardiac disease statesremain poorly understood, and the optimal cellpopulations to use and the best delivery strategiesfor clinical translation have not yet been defined.Further research is required.

Conclusions

As discussed above, the generation of an unlim-ited supply of specific cell types is crucial for celland regenerative therapies, because they offergreat hope for the treatment of a wide spectrumof diseases. However, numerous challenges re-main. Recurrent autoimmunity will probablyrequire immune suppression for some diseases,and it is still unclear which specific cell type willbe useful for the treatment of disorders such asDMD and heart disease. Optimization will also

be needed for the transplant site, as in diabetes,or when dealing with disruption of the extra-cellular matrix in treating degenerative diseases,as in chronic liver and heart disease. Finally, whenthe pathologic process is diffuse and migrationof transplanted cells is limited, as is the casewith Alzheimer’s disease, ALS, and the muscu-lar dystrophies, identifying the best means andlocation for cell delivery will require further studyin many cases. Using autologous cells and geneticengineering should help control rejection andautoimmunity and improve monitoring of short-and long-term engraftment. Despite existing chal-lenges, the availability of an unlimited supply ofclinically useful PSC-derived cell populations willfacilitate studies into the biology of those cellsand will likely assist in the treatment of diabetes,acute hepatic failure, metabolic liver diseases, re-tinal diseases, PD, HD, and possibly heart disease.Close collaboration between scientists and clini-cians, and between academia and industry, willbe critical to overcoming remaining challenges tobringing novel therapies to patients in need.

REFERENCES AND NOTES

1. K. Takahashi et al., Induction of pluripotent stem cells fromadult human fibroblasts by defined factors. Cell 131, 861–872(2007). doi: 10.1016/j.cell.2007.11.019; pmid: 18035408

2. J. A. Thomson et al., Embryonic stem cell lines derivedfrom human blastocysts. Science 282, 1145–1147 (1998).doi: 10.1126/science.282.5391.1145; pmid: 9804556

3. S. Lacotte, T. Berney, A. J. Shapiro, C. Toso, Immunemonitoring of pancreatic islet graft: Towards a betterunderstanding, detection and treatment of harmful events.Expert Opin. Biol. Ther. 11, 55–66 (2011). doi: 10.1517/14712598.2011.536530; pmid: 21073277

4. J. Hanna et al., Treatment of sickle cell anemia mousemodel with iPS cells generated from autologous skin. Science318, 1920–1923 (2007). doi: 10.1126/science.1152092;pmid: 18063756

5. W. M. Rideout 3rd, K. Hochedlinger, M. Kyba, G. Q. Daley,R. Jaenisch, Correction of a genetic defect by nucleartransplantation and combined cell and gene therapy. Cell109, 17–27 (2002). doi: 10.1016/S0092-8674(02)00681-5;pmid: 11955443

6. M. Kyba, R. C. Perlingeiro, G. Q. Daley, HoxB4 confersdefinitive lymphoid-myeloid engraftment potential onembryonic stem cell and yolk sac hematopoietic progenitors.Cell 109, 29–37 (2002). doi: 10.1016/S0092-8674(02)00680-3; pmid: 11955444

7. S. L. McKinney-Freeman et al., Surface antigen phenotypesof hematopoietic stem cells from embryos and murineembryonic stem cells. Blood 114, 268–278 (2009).doi: 10.1182/blood-2008-12-193888; pmid: 19420357

8. M. H. Ledran et al., Efficient hematopoietic differentiationof human embryonic stem cells on stromal cellsderived from hematopoietic niches. Cell Stem Cell 3,85–98 (2008). doi: 10.1016/j.stem.2008.06.001;pmid: 18593561

9. X. Tian, P. S. Woll, J. K. Morris, J. L. Linehan, D. S. Kaufman,Hematopoietic engraftment of human embryonic stemcell-derived cells is regulated by recipient innate immunity.Stem Cells 24, 1370–1380 (2006). doi: 10.1634/stemcells.2005-0340; pmid: 16456127

10. L. Wang, P. Menendez, C. Cerdan, M. Bhatia, Hematopoieticdevelopment from human embryonic stem cell lines.Exp. Hematol. 33, 987–996 (2005). doi: 10.1016/j.exphem.2005.06.002; pmid: 16140146

11. C. Mazurier, L. Douay, H. Lapillonne, Red blood cells frominduced pluripotent stem cells: Hurdles and developments.Curr. Opin. Hematol. 18, 249–253 (2011). doi: 10.1097/MOH.0b013e3283476129; pmid: 21519239

12. G. F. Rousseau, M. C. Giarratana, L. Douay, Large-scaleproduction of red blood cells from stem cells: What are thetechnical challenges ahead? Biotechnol. J. 9, 28–38 (2014).doi: 10.1002/biot.201200368; pmid: 24408610

13. S. Nakamura et al., Expandable megakaryocyte cell linesenable clinically applicable generation of platelets from

SCIENCE sciencemag.org 22 AUGUST 2014 • VOL 345 ISSUE 6199 1247391-7

Fig. 6. Fibrin patch–based delivery of human PSC–derived ECs and SMCs. (A) Cardiac magneticresonance image (MRI) from normal, MI, and cell-treatment groups at end systole (left) and enddiastole (right). (B) After infarction, the cell treatment produced significantly smaller infarct size (delayed-enhancement MRI) and (C) improved ejection fraction compared with MI alone. *P < 0.05 versus MI alone.Modified from (147), with permission.

RESEARCH | REVIEW

human induced pluripotent stem cells. Cell Stem Cell14, 535–548 (2014). doi: 10.1016/j.stem.2014.01.011;pmid: 24529595

14. R. Bottino, M. Trucco, A. N. Balamurugan, T. E. Starzl,Pancreas and islet cell transplantation. Best Pract. Res.Clin. Gastroenterol. 16, 457–474 (2002). doi: 10.1053/bega.2002.0318; pmid: 12079269

15. C. B. Kemp, M. J. Knight, D. W. Scharp, P. E. Lacy,W. F. Ballinger, Transplantation of isolated pancreaticislets into the portal vein of diabetic rats. Nature 244, 447(1973). doi: 10.1038/244447a0; pmid: 4200461

16. C. Ricordi, P. E. Lacy, E. H. Finke, B. J. Olack, D. W. Scharp,Automated method for isolation of human pancreaticislets. Diabetes 37, 413–420 (1988). doi: 10.2337/diab.37.4.413; pmid: 3288530

17. E. A. Ryan et al., Five-year follow-up after clinical islettransplantation. Diabetes 54, 2060–2069 (2005).doi: 10.2337/diabetes.54.7.2060; pmid: 15983207

18. R. Alejandro et al., Natural history of intrahepatic canineislet cell autografts. J. Clin. Invest. 78, 1339–1348 (1986).doi: 10.1172/JCI112720; pmid: 3095376

19. B. Nilsson, K. N. Ekdahl, O. Korsgren, Control of instantblood-mediated inflammatory reaction to improve islets ofLangerhans engraftment. Curr. Opin. Organ Transplant. 16,620–626 (2011). doi: 10.1097/MOT.0b013e32834c2393;pmid: 21971510

20. F. B. Barton et al., Improvement in outcomes of clinical islettransplantation: 1999-2010. Diabetes Care 35, 1436–1445(2012). doi: 10.2337/dc12-0063; pmid: 22723582

21. D. E. Sutherland et al., Total pancreatectomy and isletautotransplantation for chronic pancreatitis. J. Am. Coll. Surg.214, 409–424, discussion 424–426 (2012). doi: 10.1016/j.jamcollsurg.2011.12.040; pmid: 22397977

22. G. T. Westermark, P. Westermark, C. Berne, O. Korsgren; NordicNetwork for Clinical Islet Transplantation, Widespread amyloiddeposition in transplanted human pancreatic islets. N. Engl. J.Med. 359, 977–979 (2008).doi: 10.1056/NEJMc0802893; pmid: 18753660

23. M. E. De Paepe, B. Keymeulen, D. Pipeleers, G. Klöppel,Proliferation and hypertrophy of liver cells surrounding isletgrafts in diabetic recipient rats. Hepatology 21, 1144–1153(1995). doi: 10.1002/hep.1840210438; pmid: 7705790

24. D. G. Pipeleers et al., Transplantation of purified islet cellsin diabetic rats. I. Standardization of islet cell grafts. Diabetes40, 908–919 (1991). doi: 10.2337/diab.40.7.908; pmid: 2060727

25. G. J. Echeverri et al., Endoscopic gastric submucosaltransplantation of islets (ENDO-STI): Technique andinitial results in diabetic pigs. Am. J. Transplant. 9,2485–2496 (2009). doi: 10.1111/j.1600-6143.2009.02815.x;pmid: 19775318

26. Worcester Human Islet Transplantation Group, Autoimmunityafter islet-cell allotransplantation. N. Engl. J. Med. 355,1397–1399 (2006). doi: 10.1056/NEJMc061530; pmid: 17005967

27. R. J. Creusot, N. Giannoukakis, M. Trucco, M. J. Clare-Salzler,C. G. Fathman, It’s time to bring dendritic cell therapy totype 1 diabetes. Diabetes 63, 20–30 (2014). doi: 10.2337/db13-0886; pmid: 24357690

28. D. J. van der Windt et al., Long-term controllednormoglycemia in diabetic non-human primates aftertransplantation with hCD46 transgenic porcine islets.Am. J. Transplant. 9, 2716–2726 (2009). doi: 10.1111/j.1600-6143.2009.02850.x; pmid: 19845582

29. S. W. Liao et al., Maintaining functional islets throughencapsulation in an injectable saccharide-peptide hydrogel.Biomaterials 34, 3984–3991 (2013). doi: 10.1016/j.biomaterials.2013.02.007; pmid: 23465491

30. T. C. Schulz et al., A scalable system for production offunctional pancreatic progenitors from human embryonicstem cells. PLOS ONE 7, e37004 (2012). doi: 10.1371/journal.pone.0037004; pmid: 22623968

31. A. A. Demetriou et al., Transplantation of microcarrier-attachedhepatocytes into 90% partially hepatectomized rats.Hepatology 8, 1006–1009 (1988). doi: 10.1002/hep.1840080505; pmid: 3047034

32. J. M. De Vree et al., Correction of liver disease by hepatocytetransplantation in a mouse model of progressive familialintrahepatic cholestasis. Gastroenterology 119, 1720–1730(2000). doi: 10.1053/gast.2000.20222; pmid: 11113093

33. C. G. Groth, B. Arborgh, C. Björkén, B. Sundberg,G. Lundgren, Correction of hyperbilirubinemia in theglucuronyltransferase-deficient rat by intraportalhepatocyte transplantation. Transplant. Proc. 9, 313–316(1977). pmid: 405772

34. A. J. Matas et al., Hepatocellular transplantation formetabolic deficiencies: Decrease of plasms bilirubin in Gunnrats. Science 192, 892–894 (1976). doi: 10.1126/science.818706; pmid: 818706

35. H. E. Rugstad, S. H. Robinson, C. Yannoni, A. H. Tashjian Jr.,Transfer of bilirubin uridine diphosphate-glucuronyltransferaseto enzyme-deficient rats. Science 170, 553–555 (1970).doi: 10.1126/science.170.3957.553; pmid: 4319199

36. J. C. Wiederkehr, G. T. Kondos, R. Pollak, Hepatocytetransplantation for the low-density lipoprotein receptor-deficient state. A study in the Watanabe rabbit.Transplantation 50, 466–471 (1990). doi: 10.1097/00007890-199009000-00021; pmid: 2144923

37. Y. Yoshida, Y. Tokusashi, G. H. Lee, K. Ogawa, Intrahepatictransplantation of normal hepatocytes prevents Wilson’sdisease in Long-Evans cinnamon rats. Gastroenterology 111,1654–1660 (1996). doi: 10.1016/S0016-5085(96)70029-X;pmid: 8942746

38. S. Gupta et al., Entry and integration of transplantedhepatocytes in rat liver plates occur by disruption of hepaticsinusoidal endothelium. Hepatology 29, 509–519 (1999).doi: 10.1002/hep.510290213; pmid: 9918929

39. K. P. Ponder et al., Mouse hepatocytes migrate to liverparenchyma and function indefinitely after intrasplenictransplantation. Proc. Natl. Acad. Sci. U.S.A. 88, 1217–1221(1991). doi: 10.1073/pnas.88.4.1217; pmid: 1899924

40. A. Dhawan, J. Puppi, R. D. Hughes, R. R. Mitry, Humanhepatocyte transplantation: Current experience and futurechallenges. Nat. Rev. Gastroenterol. Hepatol. 7, 288–298(2010). doi: 10.1038/nrgastro.2010.44; pmid: 20368738

41. C. M. Habibullah, I. H. Syed, A. Qamar, Z. Taher-Uz, Humanfetal hepatocyte transplantation in patients with fulminanthepatic failure. Transplantation 58, 951–952 (1994).doi: 10.1097/00007890-199410270-00016; pmid: 7940741

42. S. P. Horslen et al., Isolated hepatocyte transplantation in aninfant with a severe urea cycle disorder. Pediatrics 111,1262–1267 (2003). doi: 10.1542/peds.111.6.1262; pmid: 12777539

43. M. Mito et al., Morphology and function of isolatedhepatocytes transplanted into rat spleen. Transplantation 28,499–505 (1979). doi: 10.1097/00007890-197912000-00013;pmid: 516133

44. M. Mito, M. Kusano, Y. Kawaura, Hepatocyte transplantation inman. Transplant. Proc. 24, 3052–3053 (1992). pmid: 1466053

45. M. Muraca et al., Hepatocyte transplantation as atreatment for glycogen storage disease type 1a. Lancet359, 317–318 (2002). doi: 10.1016/S0140-6736(02)07529-3;pmid: 11830200

46. E. M. Sokal et al., Hepatocyte transplantation in a 4-year-old girlwith peroxisomal biogenesis disease: Technique, safety, andmetabolic follow-up. Transplantation 76, 735–738 (2003).doi: 10.1097/01.TP.0000077420.81365.53; pmid: 12973120

47. I. J. Fox et al., Treatment of the Crigler-Najjar syndrometype I with hepatocyte transplantation. N. Engl. J. Med. 338,1422–1427 (1998). doi: 10.1056/NEJM199805143382004;pmid: 9580649

48. P. N. Newsome, J. N. Plevris, L. J. Nelson, P. C. Hayes,Animal models of fulminant hepatic failure: A criticalevaluation. Liver Transpl. 6, 21–31 (2000).pmid: 10648574

49. J. Terblanche, R. Hickman, Animal models of fulminanthepatic failure. Dig. Dis. Sci. 36, 770–774 (1991).doi: 10.1007/BF01311235; pmid: 2032519

50. B. M. Bilir et al., Hepatocyte transplantation in acuteliver failure. Liver Transpl. 6, 32–40 (2000). doi: 10.1002/lt.500060113; pmid: 10648575

51. H. Zhou et al., Repopulation of individual liver lobes bytransplanted hepatocytes using regiospecific hepaticirradiation cures jaundice in the Gunn rat model ofCrigler-Najjar Syndrome type I. Hepatology 50, 303A (2009).

52. C. Guha et al., Normal hepatocytes correct serum bilirubin afterrepopulation of Gunn rat liver subjected to irradiation/partialresection. Hepatology 36, 354–362 (2002). doi: 10.1053/jhep.2002.34516; pmid: 12143043

53. S. Gagandeep et al., Transplanted hepatocytes engraft,survive, and proliferate in the liver of rats with carbontetrachloride-induced cirrhosis. J. Pathol. 191, 78–85 (2000).doi: 10.1002/(SICI)1096-9896(200005)191:1<78::AID-PATH587>3.0.CO;2-P; pmid: 10767723

54. N. Kobayashi et al., Hepatocyte transplantation in rats withdecompensated cirrhosis. Hepatology 31, 851–857(2000). doi: 10.1053/he.2000.5636; pmid: 10733539

55. J. Ribeiro et al., Intrasplenic hepatocellular transplantationcorrects hepatic encephalopathy in portacaval-shunted rats.Hepatology 15, 12–18 (1992). doi: 10.1002/hep.1840150104;pmid: 1727787

56. T. Hoppo, J. Komori, R. Manohar, D. B. Stolz, E. Lagasse,Rescue of lethal hepatic failure by hepatized lymph nodesin mice. Gastroenterology 140, 656–666.e2 (2011).doi: 10.1053/j.gastro.2010.11.006; pmid: 21070777

57. I. K. Schumacher et al., Transplantation of conditionallyimmortalized hepatocytes to treat hepatic encephalopathy.Hepatology 24, 337–343 (1996). doi: 10.1002/hep.510240209; pmid: 8690402

58. K. A. Soltys et al., Barriers to the successful treatment ofliver disease by hepatocyte transplantation. J. Hepatol.53, 769–774 (2010). doi: 10.1016/j.jhep.2010.05.010;pmid: 20667616

59. B. E. Uygun et al., Organ reengineering through developmentof a transplantable recellularized liver graft usingdecellularized liver matrix. Nat. Med. 16, 814–820 (2010).doi: 10.1038/nm.2170; pmid: 20543851

60. A. Benraiss et al., Sustained mobilization of endogenousneural progenitors delays disease progression in a transgenicmodel of Huntington’s disease. Cell Stem Cell 12, 787–799(2013). doi: 10.1016/j.stem.2013.04.014; pmid: 23746982

61. A. Benraiss, S. A. Goldman, Cellular therapy and inducedneuronal replacement for Huntington’s disease.Neurotherapeutics 8, 577–590 (2011). doi: 10.1007/s13311-011-0075-8; pmid: 21971961

62. O. Lindvall, Dopaminergic neurons for Parkinson’s therapy.Nat. Biotechnol. 30, 56–58 (2012). doi: 10.1038/nbt.2077;pmid: 22231097

63. Y. Liu et al., Medial ganglionic eminence-like cells derivedfrom human embryonic stem cells correct learning andmemory deficits. Nat. Biotechnol. 31, 440–447 (2013).doi: 10.1038/nbt.2565; pmid: 23604284

64. S. Kriks et al., Dopamine neurons derived from human EScells efficiently engraft in animal models of Parkinson’sdisease. Nature 480, 547–551 (2011).pmid: 22056989

65. N. S. Roy et al., Functional engraftment of human EScell-derived dopaminergic neurons enriched by coculturewith telomerase-immortalized midbrain astrocytes. Nat. Med.12, 1259–1268 (2006). doi: 10.1038/nm1495; pmid: 17057709

66. A. D. Carri et al., Developmentally coordinated extrinsicsignals drive human pluripotent stem cell differentiationtoward authentic DARPP-32+ medium-sized spiny neurons.Development 140, 301–312 (2013). doi: 10.1242/dev.084608;pmid: 23250204

67. The HD iPSC Consortium, Induced pluripotent stem cellsfrom patients with Huntington’s disease show CAG-repeat-expansion-associated phenotypes. Cell Stem Cell 11, 264–278(2012). doi: 10.1016/j.stem.2012.04.027; pmid: 22748968

68. C. R. Nicholas et al., Functional maturation of hPSC-derivedforebrain interneurons requires an extended timeline andmimics human neural development. Cell Stem Cell 12, 573–586(2013). doi: 10.1016/j.stem.2013.04.005; pmid: 23642366

69. D. E. Buchholz et al., Rapid and efficient directed differentiationof human pluripotent stem cells into retinal pigmentedepithelium. Stem Cells Transl. Med. 2, 384–393 (2013).doi: 10.5966/sctm.2012-0163; pmid: 23599499

70. J. H. Stern, S. Temple, Stem cells for retinal replacementtherapy. Neurotherapeutics 8, 736–743 (2011). doi: 10.1007/s13311-011-0077-6; pmid: 21948217

71. N. Singh Roy et al., Enhancer-specified GFP-based FACSpurification of human spinal motor neurons from embryonicstem cells. Exp. Neurol. 196, 224–234 (2005). doi: 10.1016/j.expneurol.2005.06.021; pmid: 16198339

72. S. Karumbayaram et al., Directed differentiation ofhuman-induced pluripotent stem cells generates activemotor neurons. Stem Cells 27, 806–811 (2009).doi: 10.1002/stem.31; pmid: 19350680

73. X. J. Li et al., Specification of motoneurons from humanembryonic stem cells. Nat. Biotechnol. 23, 215–221(2005). doi: 10.1038/nbt1063; pmid: 15685164

74. A. C. Lepore et al., Human glial-restricted progenitortransplantation into cervical spinal cord of the SOD1 mousemodel of ALS. PLOS ONE 6, e25968 (2011). doi: 10.1371/journal.pone.0025968; pmid: 21998733

75. S. A. Goldman, M. Nedergaard, M. S. Windrem, Glialprogenitor cell-based treatment and modeling of neurologicaldisease. Science 338, 491–495 (2012). doi: 10.1126/science.1218071; pmid: 23112326

76. D. R. Archer, P. A. Cuddon, D. Lipsitz, I. D. Duncan,Myelination of the canine central nervous system by glialcell transplantation: A model for repair of human myelindisease. Nat. Med. 3, 54–59 (1997). doi: 10.1038/nm0197-54;pmid: 8986741

77. M. S. Windrem et al., Fetal and adult human oligodendrocyteprogenitor cell isolates myelinate the congenitally dysmyelinated

1247391-8 22 AUGUST 2014 • VOL 345 ISSUE 6199 sciencemag.org SCIENCE

RESEARCH | REVIEW

brain. Nat. Med. 10, 93–97 (2004). doi: 10.1038/nm974;pmid: 14702638

78. M. S. Windrem et al., Neonatal chimerization with human glialprogenitor cells can both remyelinate and rescue theotherwise lethally hypomyelinated shiverer mouse. Cell StemCell 2, 553–565 (2008). doi: 10.1016/j.stem.2008.03.020;pmid: 18522848

79. S. Wang et al., Human iPSC-derived oligodendrocyteprogenitor cells can myelinate and rescue a mouse model ofcongenital hypomyelination. Cell Stem Cell 12, 252–264(2013). doi: 10.1016/j.stem.2012.12.002; pmid: 23395447

80. N. Gupta et al., Neural stem cell engraftment and myelinationin the human brain. Sci. Transl. Med. 4, 155ra137 (2012).doi: 10.1126/scitranslmed.3004373; pmid: 23052294

81. S. A. Goldman, Progenitor cell-based treatment of thepediatric myelin disorders. Arch. Neurol. 68, 848–856 (2011).doi: 10.1001/archneurol.2011.46; pmid: 21403006

82. M. Li, K. Suzuki, N. Y. Kim, G. H. Liu, J. C. Izpisua Belmonte,A cut above the rest: Targeted genome editing technologiesin human pluripotent stem cells. J. Biol. Chem. 289,4594–4599 (2014). doi: 10.1074/jbc.R113.488247;pmid: 24362028

83. M. H. Tuszynski et al., Neural stem cell dissemination aftergrafting to CNS injury sites. Cell 156, 388–389 (2014).doi: 10.1016/j.cell.2014.01.016; pmid: 24485445

84. E. P. Hoffman, R. H. Brown Jr., L. M. Kunkel, Dystrophin: Theprotein product of the Duchenne muscular dystrophy locus. Cell51, 919–928 (1987). doi: 10.1016/0092-8674(87)90579-4;pmid: 3319190

85. S. C. Watkins, E. P. Hoffman, H. S. Slayter, L. M. Kunkel,Immunoelectron microscopic localization of dystrophin inmyofibres. Nature 333, 863–866 (1988). doi: 10.1038/333863a0; pmid: 3290684

86. K. Arahata et al., Immunostaining of skeletal and cardiacmuscle surface membrane with antibody against Duchennemuscular dystrophy peptide. Nature 333, 861–863 (1988).doi: 10.1038/333861a0; pmid: 3290683

87. E. E. Zubrzycka-Gaarn et al., The Duchenne musculardystrophy gene product is localized in sarcolemma of humanskeletal muscle. Nature 333, 466–469 (1988). doi: 10.1038/333466a0; pmid: 3287171

88. R. Bischoff, Proliferation of muscle satellite cells on intactmyofibers in culture. Dev. Biol. 115, 129–139 (1986).doi: 10.1016/0012-1606(86)90234-4; pmid: 3516758

89. J. R. Beauchamp, J. E. Morgan, C. N. Pagel, T. A. Partridge,Dynamics of myoblast transplantation reveal a discreteminority of precursors with stem cell-like properties as themyogenic source. J. Cell Biol. 144, 1113–1122 (1999). doi:10.1083/jcb.144.6.1113; pmid: 10087257

90. Y. Fan, M. Maley, M. Beilharz, M. Grounds, Rapid death ofinjected myoblasts in myoblast transfer therapy. MuscleNerve 19, 853–860 (1996). doi: 10.1002/(SICI)1097-4598(199607)19:7<853::AID-MUS7>3.0.CO;2-8; pmid: 8965839

91. E. Gussoni et al., Normal dystrophin transcripts detected inDuchenne muscular dystrophy patients after myoblasttransplantation. Nature 356, 435–438 (1992). doi: 10.1038/356435a0; pmid: 1557125

92. J. Huard et al., Human myoblast transplantation: Preliminaryresults of 4 cases. Muscle Nerve 15, 550–560 (1992). doi:10.1002/mus.880150504; pmid: 1584246

93. G. Karpati et al., Dystrophin is expressed in mdx skeletalmuscle fibers after normal myoblast implantation. Am. J.Pathol. 135, 27–32 (1989). pmid: 2672825

94. J. R. Mendell et al., Myoblast transfer in the treatment ofDuchenne’s muscular dystrophy. N. Engl. J. Med. 333, 832–838(1995). doi: 10.1056/NEJM199509283331303; pmid: 7651473

95. J. E. Morgan, E. P. Hoffman, T. A. Partridge, Normalmyogenic cells from newborn mice restore normal histologyto degenerating muscles of the mdx mouse. J. Cell Biol.111, 2437–2449 (1990). doi: 10.1083/jcb.111.6.2437;pmid: 2277066

96. Z. Qu et al., Development of approaches to improve cellsurvival in myoblast transfer therapy. J. Cell Biol. 142,1257–1267 (1998). doi: 10.1083/jcb.142.5.1257;pmid: 9732286

97. J. T. Vilquin, E. Wagner, I. Kinoshita, R. Roy, J. P. Tremblay,Successful histocompatible myoblast transplantation indystrophin-deficient mdx mouse despite the production ofantibodies against dystrophin. J. Cell Biol. 131, 975–988(1995). doi: 10.1083/jcb.131.4.975; pmid: 7490298

98. Y. Torrente, E. El Fahime, N. J. Caron, N. Bresolin, J. P. Tremblay,Intramuscular migration of myoblasts transplanted aftermuscle pretreatment with metalloproteinases. Cell Transplant.9, 539–549 (2000). pmid: 11038070

99. D. L. Clarke et al., Generalized potential of adult neural stemcells. Science 288, 1660–1663 (2000). doi: 10.1126/science.288.5471.1660; pmid: 10834848

100. L. De Angelis et al., Skeletal myogenic progenitors originatingfrom embryonic dorsal aorta coexpress endothelial andmyogenic markers and contribute to postnatal musclegrowth and regeneration. J. Cell Biol. 147, 869–878 (1999).doi: 10.1083/jcb.147.4.869; pmid: 10562287

101. G. Ferrari et al., Muscle regeneration by bone marrow–derivedmyogenic progenitors. Science 279, 1528–1530 (1998).doi: 10.1126/science.279.5356.1528; pmid: 9488650

102. R. Galli et al., Skeletal myogenic potential of human andmouse neural stem cells. Nat. Neurosci. 3, 986–991 (2000).doi: 10.1038/79924; pmid: 11017170

103. M. F. Pittenger et al., Multilineage potential of adult humanmesenchymal stem cells. Science 284, 143–147 (1999).doi: 10.1126/science.284.5411.143; pmid: 10102814

104. M. Crisan et al., A perivascular origin for mesenchymal stemcells in multiple human organs. Cell Stem Cell 3, 301–313(2008). doi: 10.1016/j.stem.2008.07.003; pmid: 18786417

105. A. Dellavalle et al., Pericytes of human skeletal muscle aremyogenic precursors distinct from satellite cells.Nat. Cell Biol. 9, 255–267 (2007). doi: 10.1038/ncb1542;pmid: 17293855

106. B. Zheng et al., Prospective identification of myogenicendothelial cells in human skeletal muscle. Nat. Biotechnol.25, 1025–1034 (2007). doi: 10.1038/nbt1334; pmid:17767154

107. H. M. Blau, T. R. Brazelton, J. M. Weimann, The evolvingconcept of a stem cell: Entity or function? Cell 105,829–841 (2001). doi: 10.1016/S0092-8674(01)00409-3;pmid: 11439179

108. B. Péault et al., Stem and progenitor cells in skeletalmuscle development, maintenance, and therapy. Mol. Ther.15, 867–877 (2007). doi: 10.1038/mt.sj.6300145;pmid: 17387336

109. L. K. Carr et al., 1-year follow-up of autologous muscle-derived stem cell injection pilot study to treat stressurinary incontinence. Int. Urogynecol. J. Pelvic Floor Dysfunct.19, 881–883 (2008). doi: 10.1007/s00192-007-0553-z;pmid: 18204978

110. M. Sampaolesi et al., Mesoangioblast stem cells amelioratemuscle function in dystrophic dogs. Nature 444, 574–579(2006). doi: 10.1038/nature05282; pmid: 17108972

111. E. Bachrach et al., Systemic delivery of humanmicrodystrophin to regenerating mouse dystrophic muscleby muscle progenitor cells. Proc. Natl. Acad. Sci. U.S.A.101, 3581–3586 (2004). doi: 10.1073/pnas.0400373101;pmid: 14993597

112. M. Kennedy et al., A common precursor for primitiveerythropoiesis and definitive haematopoiesis. Nature 386,488–493 (1997). doi: 10.1038/386488a0; pmid: 9087406

113. M. G. Klug, M. H. Soonpaa, G. Y. Koh, L. J. Field, Geneticallyselected cardiomyocytes from differentiating embronicstem cells form stable intracardiac grafts. J. Clin. Invest. 98,216–224 (1996). doi: 10.1172/JCI118769; pmid: 8690796

114. M. Müller et al., Selection of ventricular-like cardiomyocytesfrom ES cells in vitro. FASEB J. 14, 2540–2548 (2000).doi: 10.1096/fj.00-0002com; pmid: 11099473

115. F. Rinaldi, R. C. Perlingeiro, Stem cells for skeletal muscleregeneration: Therapeutic potential and roadblocks. Transl.Res. 163, 409–417 (2014). doi: 10.1016/j.trsl.2013.11.006;pmid: 24299739

116. D. Vittet et al., Embryonic stem cells differentiate in vitroto endothelial cells through successive maturation steps.Blood 88, 3424–3431 (1996). pmid: 8896407

117. T. Barberi et al., Derivation of engraftable skeletal myoblastsfrom human embryonic stem cells. Nat. Med. 13, 642–648(2007). doi: 10.1038/nm1533; pmid: 17417652

118. J. Rohwedel et al., Muscle cell differentiation of embryonicstem cells reflects myogenesis in vivo: Developmentallyregulated expression of myogenic determination genes andfunctional expression of ionic currents. Dev. Biol. 164, 87–101(1994). doi: 10.1006/dbio.1994.1182; pmid: 8026639

119. S. Goudenege et al., Myoblasts derived from normal hESCsand dystrophic hiPSCs efficiently fuse with existing musclefibers following transplantation. Mol. Ther. 20, 2153–2167(2012). doi: 10.1038/mt.2012.188; pmid: 22990676

120. R. Darabi et al., Human ES- and iPS-derived myogenicprogenitors restore DYSTROPHIN and improve contractilityupon transplantation in dystrophic mice. Cell Stem Cell 10,610–619 (2012). doi: 10.1016/j.stem.2012.02.015; pmid: 22560081

121. A. Filareto et al., An ex vivo gene therapy approach to treatmuscular dystrophy using inducible pluripotent stem cells.

Nat. Commun. 4, 1549 (2013). doi: 10.1038/ncomms2550;pmid: 23462992

122. F. S. Tedesco et al., Transplantation of genetically correctedhuman iPSC-derived progenitors in mice with limb-girdlemuscular dystrophy. Sci. Transl. Med. 4, 140ra89 (2012).doi: 10.1126/scitranslmed.3003541; pmid: 22745439

123. G. Jasmin, C. Tautu, M. Vanasse, P. Brochu, R. Simoneau,Impaired muscle differentiation in explant cultures ofDuchenne muscular dystrophy. Lab. Invest. 50, 197–207(1984). pmid: 6694359

124. S. Liechti-Gallati, H. Moser, H. P. Siegrist, U. Wiesmann,N. N. Herschkowitz, Abnormal growth kinetics and5′-nucleotidase activities in cultured skin fibroblasts frompatients with Duchenne muscular dystrophy. Pediatr. Res.15, 1411–1414 (1981). doi: 10.1203/00006450-198111000-00004; pmid: 6272184

125. A. Sacco et al., Short telomeres and stem cell exhaustionmodel Duchenne muscular dystrophy in mdx/mTR mice.Cell 143, 1059–1071 (2010). doi: 10.1016/j.cell.2010.11.039;pmid: 21145579

126. K. M. Kallestad et al., Sparing of extraocular muscle in agingand muscular dystrophies: A myogenic precursor cellhypothesis. Exp. Cell Res. 317, 873–885 (2011). doi: 10.1016/j.yexcr.2011.01.018; pmid: 21277300

127. S. K. Sanganalmath, R. Bolli, Cell therapy for heart failure:A comprehensive overview of experimental and clinicalstudies, current challenges, and future directions. Circ. Res.113, 810–834 (2013). doi: 10.1161/CIRCRESAHA.113.300219;pmid: 23989721

128. B. Leobon et al., Myoblasts transplanted into rat infarctedmyocardium are functionally isolated from their host.Proc. Natl. Acad. Sci. U.S.A. 100, 7808–7811 (2003).doi: 10.1073/pnas.1232447100; pmid: 12805561

129. P. Menasché et al., The Myoblast Autologous Grafting inIschemic Cardiomyopathy (MAGIC) trial: First randomizedplacebo-controlled study of myoblast transplantation.Circulation 117, 1189–1200 (2008). doi: 10.1161/CIRCULATIONAHA.107.734103; pmid: 18285565

130. D. Orlic et al., Bone marrow cells regenerate infarctedmyocardium. Nature 410, 701–705 (2001). doi: 10.1038/35070587; pmid: 11287958

131. L. B. Balsam et al., Haematopoietic stem cells adoptmature haematopoietic fates in ischaemic myocardium.Nature 428, 668–673 (2004). doi: 10.1038/nature02460pmid: 15034594

132. C. E. Murry et al., Haematopoietic stem cells do nottransdifferentiate into cardiac myocytes in myocardialinfarcts. Nature 428, 664–668 (2004). doi: 10.1038/nature02446; pmid: 15034593

133. S. A. Fisher et al., Stem cell therapy for chronic ischaemicheart disease and congestive heart failure. CochraneDatabase Syst. Rev. 4, CD007888 (2014). pmid: 24777540

134. P. K. Nguyen, F. Lan, Y. Wang, J. C. Wu, Imaging: Guiding theclinical translation of cardiac stem cell therapy. Circ. Res.109, 962–979 (2011). doi: 10.1161/CIRCRESAHA.111.242909;pmid: 21960727

135. O. Caspi et al., Transplantation of human embryonic stemcell-derived cardiomyocytes improves myocardial performancein infarcted rat hearts. J. Am. Coll. Cardiol. 50, 1884–1893(2007). doi: 10.1016/j.jacc.2007.07.054; pmid: 17980256

136. A. Moretti et al., Multipotent embryonic isl1+ progenitorcells lead to cardiac, smooth muscle, and endothelial celldiversification. Cell 127, 1151–1165 (2006). doi: 10.1016/j.cell.2006.10.029; pmid: 17123592

137. L. Yang et al., Human cardiovascular progenitor cells developfrom a KDR+ embryonic-stem-cell-derived population.Nature 453, 524–528 (2008). doi: 10.1038/nature06894;pmid: 18432194

138. L. S. Ferreira et al., Vascular progenitor cells isolated fromhuman embryonic stem cells give rise to endothelial andsmooth muscle like cells and form vascular networksin vivo. Circ. Res. 101, 286–294 (2007). doi: 10.1161/CIRCRESAHA.107.150201; pmid: 17569886

139. C. L. Mummery et al., Differentiation of human embryonicstem cells and induced pluripotent stem cells tocardiomyocytes: A methods overview. Circ. Res. 111,344–358 (2012). doi: 10.1161/CIRCRESAHA.110.227512;pmid: 22821908

140. N. C. Dubois et al., SIRPA is a specific cell-surface markerfor isolating cardiomyocytes derived from human pluripotentstem cells. Nat. Biotechnol. 29, 1011–1018 (2011).doi: 10.1038/nbt.2005; pmid: 22020386

141. S. Tohyama et al., Distinct metabolic flow enableslarge-scale purification of mouse and human pluripotent

SCIENCE sciencemag.org 22 AUGUST 2014 • VOL 345 ISSUE 6199 1247391-9

RESEARCH | REVIEW

stem cell-derived cardiomyocytes. Cell Stem Cell 12,127–137 (2013). doi: 10.1016/j.stem.2012.09.013;pmid: 23168164

142. J. Q. He, Y. Ma, Y. Lee, J. A. Thomson, T. J. Kamp, Humanembryonic stem cells develop into multiple types of cardiacmyocytes: Action potential characterization. Circ. Res. 93,32–39 (2003). doi: 10.1161/01.RES.0000080317.92718.99;pmid: 12791707

143. M. A. Laflamme et al., Cardiomyocytes derived from humanembryonic stem cells in pro-survival factors enhance functionof infarcted rat hearts. Nat. Biotechnol. 25, 1015–1024(2007). doi: 10.1038/nbt1327; pmid: 17721512

144. N. Christoforou et al., Implantation of mouse embryonicstem cell-derived cardiac progenitor cells preservesfunction of infarcted murine hearts. PLOS ONE 5, e11536(2010). doi: 10.1371/journal.pone.0011536; pmid: 20634944

145. Y. Shiba et al., Human ES-cell-derived cardiomyocyteselectrically couple and suppress arrhythmias in injuredhearts. Nature 489, 322–325 (2012). doi: 10.1038/nature11317; pmid: 22864415

146. M. Kawamura et al., Feasibility, safety, and therapeuticefficacy of human induced pluripotent stem cell-derived

cardiomyocyte sheets in a porcine ischemic cardiomyopathymodel. Circulation 126 (suppl. 1), S29–S37 (2012).doi: 10.1161/CIRCULATIONAHA.111.084343; pmid: 22965990

147. Q. Xiong et al., Functional consequences of human inducedpluripotent stem cell therapy: Myocardial ATP turnover ratein the in vivo swine heart with postinfarction remodeling.Circulation 127, 997–1008 (2013). doi: 10.1161/CIRCULATIONAHA.112.000641; pmid: 23371930

148. J. J. H. Chong et al., Human embryonic-stem-cell-derivedcardiomyocytes regenerate non-human primate hearts.Nature 510, 273–277 (2014). doi: 10.1038/nature13233;pmid: 24776797

149. G. Blin et al., A purified population of multipotentcardiovascular progenitors derived from primate pluripotentstem cells engrafts in postmyocardial infarcted nonhumanprimates. J. Clin. Invest. 120, 1125–1139 (2010). doi: 10.1172/JCI40120; pmid: 20335662

150. Z. Alipio et al., Reversal of hyperglycemia in diabeticmouse models using induced-pluripotent stem (iPS)-derivedpancreatic beta-like cells. Proc. Natl. Acad. Sci. U.S.A.107, 13426–13431 (2010). doi: 10.1073/pnas.1007884107;pmid: 20616080

ACKNOWLEDGMENTS

We thank E. Tafaleng, N. Giannoukakis, and J. H. Cummins forproviding figures and for text editing. This work was supported inpart by NIH R01DK48794, P01DK096990, and R01DK099320(I.J.F.); National Institute of Neurological Disorders and Stroke,National Institute of Mental Health, the Mathers CharitableFoundation, the National Multiple Sclerosis Society, the AdelsonMedical Research Foundation, Child Health and DevelopmentInstitute, and the New York Stem Cell Research Program (S.A.G.);NIH 1P01AG043376 (J.H.); U01HL099773 (T.J.K.); and DODW81XWH-10-1-1055 (M.T.). I.J.F. is a member of the ScientificAdvisory Board of Regenerative Medicine Solutions, Inc.; G.Q.D.is a member of the Senior Advisory Board of MPM Capital,Inc.; S.A.G. is a consultant for Biogen Idec; T.J.K. is a foundingshareholder in Cellular Dynamics International; and J.H. is aconsultant for Cook Myosite. S.A.G. is an inventor on patentsowned by the University of Rochester and Cornell Universityconcerning the use of tissue- and iPSC-derived glialprogenitor cells.

10.1126/science.1247391

1247391-10 22 AUGUST 2014 • VOL 345 ISSUE 6199 sciencemag.org SCIENCE

RESEARCH | REVIEW

RESEARCH

18 JULY 2014 • VOL 345 ISSUE 6194 283SCIENCE sciencemag.org

ILL

US

TR

AT

ION

: K

. S

UT

LIF

F/SCIENCE

BACKGROUND: Because

of their differentiation

potential, pluripotent

stem cells can gener-

ate virtually any cell

type and, as such, can

be used to model de-

velopment and disease

and even hold the promise of providing

cell-replacement therapies. Recently, struc-

tures resembling whole organs, termed or-

ganoids, have been generated from stem

cells through the development of three-

dimensional culture systems.

Organoids are derived from pluripotent

stem cells or isolated organ progenitors

that differentiate to form an organlike

tissue exhibiting multiple cell types that

self-organize to form a structure not unlike

the organ in vivo. This technology builds

upon a foundation of stem cell technolo-

gies, as well as classical developmental bi-

ology and cell-mixing experiments. These

studies illustrated two key events in struc-

tural organization during organogenesis:

cell sorting out and spatially restricted

lineage commitment. Both of these pro-

cesses are recapitulated in organoids,

which self-assemble to form the cellular

organization of the organ itself.

ADVANCES: Organoids have been gener-

ated for a number of organs from both

mouse and human stem cells. To date,

human pluripotent stem cells have been

coaxed to generate intestinal, kidney, brain,

and retinal organoids, as well as liver or-

ganoid-like tissues called liver buds. Deriva-

tion methods are specific to each of these

systems, with a focus on recapitulation of

endogenous developmental processes.

These complex structures provide a

unique opportunity to model human organ

development in a system remarkably simi-

lar to development in vivo. Although the

full extent of similarity in many cases still

remains to be determined, organoids are al-

ready being applied to human-specific bio-

logical questions. Indeed, brain and retinal

organoids have both been shown to exhibit

properties that recapitulate human organ

development and that cannot be observed

in animal models. Naturally, limitations

exist, such as the lack of blood supply, but

future endeavors will advance the technol-

ogy and, it is hoped, fully overcome these

technical hurdles.

OUTLOOK: The therapeutic promise of or-

ganoids is perhaps the area with greatest

potential. These unique tissues have the

potential to model developmental disease,

degenerative conditions, and cancer. Genetic

disorders can be modeled by making use of

patient-derived induced pluripotent stem

cells or by introducing disease mutations. In-

deed, this type of approach has already been

taken to generate organoids from patient

stem cells for intestine, kidney, and brain.

Furthermore, organoids that model

disease can be used as an alternative sys-

tem for drug testing that may not only

better recapitulate effects in human pa-

tients but could also cut down on animal

studies. Liver organoids, in particular,

represent a system with high expecta-

tions, particularly for

drug testing, because

of the unique meta-

bolic profile of the

human liver. Finally,

tissues derived in vi-

tro could be generated

from patient cells to provide alternative or-

gan replacement strategies. Unlike current

organ transplant treatments, such autolo-

gous tissues would not suffer from issues of

immunocompetency and rejection. ■

Organogenesis in a dish: Modeling development and disease using organoid technologies

ORGANOID GENERATION

Madeline A. Lancaster and Juergen A. Knoblich*

IMBA—Institute of Molecular Biotechnology of the Austrian Academy of Sciences, Vienna 1030, Austria.* Corresponding author. E-mail: juergen.knoblich@imba.oeaw.ac.at Cite this article as M. A. Lancaster and J. A. Knoblich, Science 345, 1247125 (2014). DOI: 10.1126/science.1247125

Read the full article at http://dx.doi.org/10.1126/science.1247125

ON OUR WEBSITE

Organoid generation and therapeutic potential. Organoids can be derived for a number

of organs from human pluripotent stem cells (PSCs). Like organogenesis in vivo, organoids

self-organize through both cell sorting out and spatially restricted lineage commitment of pre-

cursor cells. Organoids can be used to model disease by introducing disease mutations or

using patient-derived PSCs. Future applications could include drug testing and even tissue

replacement therapy.

PSCs

Diferentiation

Cell sorting out Spatially restrictedlineage commitment

Organoid

Patient-derivediPSCs

Diseasemutation

Genetic correction

Drugtesting

Organreplacement

Diseasemodeling

Therapeutic potential

REVIEW SUMMARY

SP

ECIAL SERIES: STEM C

ELL

S

Published by AAAS

REVIEW◥

ORGANOID GENERATION

Organogenesis in a dish: Modelingdevelopment and disease usingorganoid technologiesMadeline A. Lancaster1 and Juergen A. Knoblich1*

Classical experiments performed half a century ago demonstrated the immenseself-organizing capacity of vertebrate cells. Even after complete dissociation, cells canreaggregate and reconstruct the original architecture of an organ. More recently, thisoutstanding feature was used to rebuild organ parts or even complete organs fromtissue or embryonic stem cells. Such stem cell–derived three-dimensional cultures arecalled organoids. Because organoids can be grown from human stem cells and frompatient-derived induced pluripotent stem cells, they have the potential to model humandevelopment and disease. Furthermore, they have potential for drug testing and even futureorgan replacement strategies. Here, we summarize this rapidly evolving field and outline thepotential of organoid technology for future biomedical research.

Stem cell technologies hold promise formodeling development, analyzing diseasemechanisms, and developing potential ther-apies. Until quite recently, most stem cellmethods focused on pure populations of

particular stem cell–derived cell types (1), ratherthan the complete set of cell types of an organ.However, this is beginning to change with thedevelopment of three-dimensional (3D) culturesof developing tissues, called organoids.As organoid technology is on the verge of be-

coming an independent research field, a precisedefinition of the term becomes increasingly im-portant. The term organoid, simply defined as re-sembling an organ, has been used quite looselyfor a variety of structures, both in vitro and in vivo(2–4). The basic definition, however, implies sev-eral important features that are characteristics oforgans (Box 1). First, it must contain more thanone cell type of the organ it models; second, itshould exhibit some function specific to that organ;and third, the cells should be organized similarlyto the organ itself. This also implies similarity tothe manner in which the organ establishes itscharacteristic organization during development.Thus, we would like to define an organoid as con-taining several cell types that develop from stemcells or organ progenitors and self-organize throughcell sorting and spatially restricted lineage commit-ment, similar to the process in vivo (Box 1).

Self-organization: The foundationof organoid formation

Organoid methods build upon an extensivefoundation of classic developmental biology

and cell dissociation and reaggregation exper-iments (Fig. 1A). Two distinct approaches havebeen taken to understanding tissue patterning(5). In vivo examination of cell movements hasrevealed mechanisms of cell segregation intodiscrete domains during tissue morphogenesis(6). This process has been extensively examinedin, for example, the Drosophila wing disc wherethe anterior-posterior boundary is establishedthrough mutually repressive interactions (7). Asimilar process occurs during vertebrate embry-

onic development at the midbrain-hindbrainboundary, which then acts as an organizer forsubsequent tissue morphogenesis (8).The second approach to understanding tissue

patterning has been dissociation and reaggre-gation of tissues to examine relative morphoge-netic movements of cells in vitro (Fig. 2A). Thisapproach has been applied to virtually all de-veloping vertebrate organs in a number of classicstudies with embryonic chick tissues (9, 10) (Fig.1A). The results point to a general capacity ofcells to reorganize and segregate in a processtermed “cell sorting out” to form structures withmuch the same histogenic properties as thosein vivo (6, 11). For example, studies from theearly 1960s have utilized dissociated cells fromthe developing chick kidney (9) to form reaggre-gates that recapitulate virtually complete renaldevelopment.The basis of this organ self-assembly seems to

arise from segregation of cells with similar ad-hesive properties into domains that achieve themost thermodynamically stable pattern (Fig. 2A).Known as Steinberg’s differential adhesion hy-pothesis (12) (Fig. 1A), the theory is supportedby a range of in vitro cell-mixing experiments(13). Differential adhesion is mediated by cellsurface adhesion proteins, for example, in sepa-ration of vertebrate neural and epidermal ecto-derm (14, 15), where differential epithelial andneural and cadherin expression mediates cellsorting out.A second mechanism that can influence tis-

sue morphogenesis is proper spatially restrictedprogenitor fate decisions (Fig. 2B). An excellentexample is the developing vertebrate retina, whereneuroepithelial cells give rise to a complex lineagethat generates the various layers of the retina ina temporally and spatially restricted manner.

RESEARCH

SCIENCE sciencemag.org 18 JULY 2014 • VOL 345 ISSUE 6194 1247125-1

1IMBA—Institute of Molecular Biotechnology of the AustrianAcademy of Science Vienna 1030, Austria.*Corresponding author. E-mail: juergen.knoblich@imba.oeaw.ac.at

Cellsorting out

Lineagecommitment

Organoid

Differentiating PSCs

Box 1. Defining organoids.

Organoid n. Resembling an organ.

This implies:1. Multiple organ-specific cell types2. Capable of recapitulating some specific

function of the organ (eg. excretion, filtration,neural activity, contraction)

3. Grouped together and spatially organizedsimilar to an organ

Organoid formation recapitulates both majorprocesses of self-organization duringdevelopment: cell sorting out and spatiallyrestricted lineage commitment

Definition:A collection of organ-specific celltypes that develops from stem cells ororgan progenitors and self-organizesthrough cell sorting and spatiallyrestricted lineage commitment in amanner similar to in vivo.

This stratification depends upon proper stemcell division orientation, the interplay of sym-metric and asymmetric divisions, and migrationof differentiated daughter cells to defined lo-cations within the tissue (16, 17). Remarkably,this organization can also be recapitulated uponin vitro dissociation and reaggregation (18) butonly when retinal precursor cells are taken froma chick younger than embryonic day 6 (E6) (19, 20).This suggests that retinal layering depends notonly on cell sorting out but also proper execu-tion of lineage decisions by retinal progenitors.The combination of both sorting out and fate

specification in governing self-organization isparticularly evident in tumors called teratomas.Teratomas develop from pluripotent stem cells(PSCs) of the germ line and therefore display avariety of organized tissues (Fig. 1B). These in-clude epidermis, nervous tissue, gut, and bone,as well as eyes (21) and limbs (22). The sponta-neous development of these tissues from PSCs ispresumably because of a recapitulation of bothcell segregation and fate specification. Similarly,these two processes allow for the self-organizationseen in organoids (Box 1).In many ways, organoids represent the meth-

odological evolution of an in vitro system calledan embryoid body (EB) that is similar to anearly teratoma (Fig. 1B). EBs are 3D aggregatesof PSCs (23) that undergo initial developmentalspecification in much the same manner as thepregastrulating embryo (24). EBs can furtherdifferentiate to form various organized tissues,much like a teratoma. However, their growth invitro allows for the application of patterningfactors to drive particular identities. Not all or-ganoid methods make use of an initial EB stage,but they all involve exogenous tissue patterningand eventually reaggregation to form a 3D self-organized tissue, an organoid.Self-organization is possible in organoids be-

cause of a growing movement away from two-dimensional culture. This movement was triggeredby the discovery that epithelial cells, such as kid-ney (25) or breast epithelia (26), could developtubules and ducts when embedded in extracel-lular matrix hydrogels. Similarly, organoids oftenmake use of such gels, particularly the laminin-rich extracellular matrix secreted by the Engelbreth-Holm-Swarm tumor line (27), also called Matrigel.The resulting self-organizing structures exhib-

it typical tissue architecture, but note that theyare highly heterogeneous. Thus, each organoid

1247125-2 18 JULY 2014 • VOL 345 ISSUE 6194 sciencemag.org SCIENCE

1944

1907

1960

1964

2013

1961

1981

1998

2009

2012

2011

2014

2008

B Pluripotent stem cells

Teratoma Embryoid body Organoid

A

1987

Wilson demonstrates thepotential of dissociated sponge cells to self-organize to regenerate a whole organism (120).

Holtfreter performsdissociation-reaggregation

experiments with dissociatedamphibian pronephros (121).

Weiss and Taylor perform dissociation-reaggregation experiments with multiple organs from embryonic chick (9).

Pierce and Verney describe thedifferentiation of embryoid

bodies in vitro (122).

Thompson et al. isolate andculture the first human

embryonic stem cell linefrom human blastocysts (126).

Clevers and colleagues generategut organoids from adult

intestinal stem cells upon3D culture in Matrigel (34).

Retinal organoidsare generated fromhuman pluripotent

stem cells (65).

2013–2014:Kidney organoids are generated

by three independent groups,generating ureteric bud (68),

metanephric mesenchyme (29),or both (69).

Steinberg introduces the differential adhesion hypothesis (DAH) of cell sorting out (12).

Gut organoids are generated from human pluripotent stem cells in vitro (33). Later that year, retinal organoids are generated from mouse ES cells (64).

Sasai and colleagues generate 3D cerebral cortex tissue from pluripotent stem cells using the SFEBq method (54).

Brain organoids are generated from human pluripotent stem cells upon growth in Matrigel and with agitation (28).

Bissell and colleagues show that breast epithelia organize into 3D ducts and ductules when grown on Engelbreth-Holm-Swarm tumor ECM extract (27). Jennings and colleagues show similar structures with lung cells (125).

Evans establishespluripotent stem cells from

mouse embryos (123). Martin similarly isolates pluripotent

mouse cells and coins the term“embryonic stem cell” (124).

Fig. 1. History of organoid methodologies. (A)Key events in the history leading up to variousorganoid methodologies. (B) Comparison of para-digms of self-organization from pluripotent stemcells. Teratomas develop various tissues in vivo,either as spontaneous tumors that can arise inanimals and humans or from injected PSCs in arodent host. Embryoid bodies are 3D aggregatesof stem cells that self-organize to develop tissues,similar to teratomas in many ways, but formedin vitro. Organoids are similarly 3D in vitro–derivedtissues but are driven using specific conditionsto generate individual, isolated tissues.

RESEARCH | REVIEW

is unique and exhibits relative positioning oftissue regions that are often random, possiblybecause of a lack of embryonic axis formation. Forexample, brain organoids display various brainregions that individually develop quite similarly tothose in vivo (28) but are not reliably organizedrelative to one another because of a lack ofanterior-posterior and dorsal-ventral axes. Similarly,kidney organoids develop tubules corresponding toregions of the nephron segment (29), but they arerandomly positioned rather than displaying medul-lary and cortical organization. This heterogeneitymakes it difficult to generate pure populations ofsingle cell types, but it can be a powerful tool formodeling development and disease on a whole-organ scale.

Current organoid technologies

Organoids derived from human PSCs have so farbeen established for gut, kidney, brain, and retina,among others (Fig. 3). Many of the organs studiedhad already been demonstrated to self-organize inreaggregation experiments from embryonic tis-sues (Table 1), which suggests that organoidscould, in principle, be generated if organ progen-itors could be derived from PSCs. Below, we willdescribe the evolution of each of these organoidapproaches from this developmental foundation.

Gut organoids

The gastrointestinal (GI) tract develops primarilyfrom the endoderm (6), which forms an epithe-

lial tube that develops into three distinct por-tions, the foregut, midgut, and hindgut (30). Theforegut gives rise to the oral cavity, the pharynx,the respiratory tract, the stomach, the pancreas,and the liver. The midgut gives rise to the smallintestine and the ascending colon. The hindgutgives rise to the remaining portion of the colon,or large intestine, and the rectum. The separa-tion of these three domains involves the com-binatorial response to growth factors that haveanteriorizing or posteriorizing effects. In par-ticular, Wnt and fibroblast growth factor (FGF)signaling have been shown to inhibit anteriorgut fate and instead promote posterior fate,which can lead to midgut and hindgut identi-ties (31, 32).This knowledge of the posteriorizing effects

of Wnts and FGFs provides the foundation onwhich human intestinal organoids are built (33)(Fig. 3). Human PSCs can be driven toward ahindgut identity by initially applying activin A,a nodal-related molecule, to drive mesendoder-mal identity. The subsequent addition of pos-teriorizing Wnt3a and Fgf4 then specifies thehindgut, the precursor to the intestine. Orig-inally, this specification was performed in 2D,but surprisingly, the cells spontaneously formedhindgut tubes that budded off to form spheroids.This illustrates the remarkable self-organizingability of these progenitors, a property that al-lows them to generate complete 3D organoidswhen grown in a permissive environment. The

laboratory of Hans Clevers had previously shownthat adult intestinal stem cells could form organ-oids when cultured in 3D in Matrigel (34). Theseadult-derived organoids self-organized to form3D crypt-villus structures that mimicked the phys-iology and organization of the intestine and couldeven be transplanted into mice (35). Similarly,hindgut spheroids generated from human PSCscan be grown in Matrigel 3D growth condi-tions to further develop to mature intestinalorganoids (33).Intestinal organoids develop crypt-villus struc-

tures with stratified epithelium consisting of allthe major cell lineages of the gut (33, 34). Theseinclude columnar epithelial enterocytes with abrush border of apical microvilli. Furthermore,cell divisions occur at the base of the villus-likeprotrusions, and intestinal stem cells could beidentified by their expression of Lgr5 in moreadvanced organoids. Finally, these organoids dis-played intestinal functions including absorptiveand secretory activity.Although the intestine is the only gut region

so far generated from PSCs, other regions of thedigestive tract have been developed into organ-oids from adult stem cells. In particular, gastricorganoids have been generated from adult py-loric stem cells (36) or chief cells of the stomach(37). Lingual organoids have been establishedfrom adult tongue epithelium (38). These ap-proaches similarly use the 3D Matrigel envi-ronment, which suggests that Matrigel is a generalrequirement in GI tract organoid formation.Furthermore, the use of adult progenitor pop-ulations in this manner provides an additional,often more direct, route to the generation oforganoids.

Liver organoids

The liver is primarily derived from endoderm,developing from an outgrowth of ventral fore-gut epithelium that develops into a hepatic budstructure (39). This hepatic bud produces thehepatoblasts that generate both hepatocytes andbiliary epithelium, whereas adjacent mesoderm-derived mesenchyme contributes liver fibro-blasts and stellate cells. The growth of the liverbud involves extensive vascularization and eventu-ally it develops into the major fetal site ofhematopoiesis. Thus, liver development repre-sents a complex interplay of both endoderm- andmesoderm-derived tissues.Early reaggregation studies had shown that

dissociated chick embryonic hepatic tissue canreaggregate and organize into secretory unitstypical of the liver and consistent with forma-tion of functional bile ducts (9). More recently, aprogenitor population in adult mouse liver thatis activated after injury was identified that couldgenerate 3D liver organoids when grown inMatrigel (40). These adult-derived liver organ-oids display cells with biliary ductal identitiesand can be differentiated to form mature, func-tional hepatocytes. Finally, liver organoids canbe transplanted into mice and were shown topartially rescue mortality in a mouse model ofliver disease, pointing to their functionality.

SCIENCE sciencemag.org 18 JULY 2014 • VOL 345 ISSUE 6194 1247125-3

B Spatially restricted lineage commitment

A Cell sorting out

Cell surfaceadhesion proteins

Fig. 2. Principles of self-organization. (A) Cell sorting out describes the movement of cells intodifferent domains. Different cell types (purple or green) sort themselves because of different adhesiveproperties conferred by their differential expression of distinct cell adhesion molecules (shown as brownor orange bars). (B) Spatially restricted cell-fate decisions also contribute to self-organization in vivo andin organoids. Progenitors (green) give rise to more differentiated progeny (purple), which, because ofspatial constraints of the tissue and/or division orientation, are forced into a more superficial positionthat promotes their differentiation. These cells can sometimes further divide to give rise to moredifferentiated progeny (pink), which are further displaced.

RESEARCH | REVIEW

Although similar human liver organoids havenot yet been generated, a very different approachwas recently established to generate tissues re-miniscent of human liver buds (41). Beginningwith differentiation of human PSCs into hepaticendodermal cells in 2D, this method mixes threecell populations: the human PSC–derived hepaticcells, human mesenchymal stem cells, and hu-man endothelial cells. This mixed-cell populationmimics the early cell lineages of the developingliver. When mixed to a high density on a layer ofMatrigel, the cells spontaneously form a 3D ag-gregate. The liver bud–like aggregates display vas-cularization and can be ectopically transplantedinto mice to allow blood supply. Perhaps mostpromising is the finding that mice with transplantsof these liver bud tissues exhibit human-specificmetabolites in the blood. Furthermore, survivalof mice subjected to liver injury increased whenliver buds were transplanted into them.

Brain organoids

The vertebrate central nervous system derivesfrom the neural ectoderm (6). This tissue givesrise to the neural plate, which folds and fuses toform the neural tube, an epithelium with apical-basal polarity radially organized around a fluid-

filled lumen that eventually forms the brainventricles. Axes are established through the con-certed action of morphogen gradients, such asthe ventral-dorsal Shh-Wnt/Bmp axis, and therostral-caudal axis influenced by factors such asretinoic acid and FGF (42). These axes allow theepithelial tube to subdivide into four major re-gions, the forebrain, midbrain, hindbrain, andspinal cord. The forebrain gives rise to the ma-jority of the human brain, including the neo-cortex, hippocampus, and ventral telencephalicstructures, such as amygdala and hypothalamus.The midbrain gives rise to the tectum, whereasthe hindbrain gives rise to the cerebellum, pons,medulla, and brainstem.Generally, neurons are generated from neural

stem cells that reside next to ventricles (43). Neu-ral stem cells initially expand through symmetricproliferative divisions. During neurogenesis, stemcells switch to asymmetric divisions to give rise toself-renewing progenitors and more differentiatedcell types, including neurons and intermediateprogenitors (44). These more differentiated cellsmigrate outward to generate stratified structuressuch as the three layers of the medulla, the sevenlayers of the optic tectum, and the six layers ofthe cerebral cortex.

Although the final product of neural develop-ment is a highly complex interconnected brain,earlier reaggregation studies suggest that thisorgan has an intrinsic self-organizing capacity(45). When taken at early stages of brain devel-opment, chick neural progenitors self-organizedto form clusters of neuroepithelial cells orga-nized in a radial manner surrounding a lumen,reminiscent of the neural tube. The implicationof these classic experiments is that if neuroepi-thelium can be derived from PSCs, spontaneousself-organization is likely to occur.Numerous previous studies have made use

of in vitro–derived neural stem cells (NSCs) fromPSCs to study neural differentiation (46). How-ever, these homogeneous NSCs lack the character-istic apical-basal polarity and do not recapitulatethe complex lineage of NSCs in vivo. As an alter-native approach, neurospheres (47) are aggre-gates of NSCs that can be used to assess theirself-renewing capacity. However, neurospheresare likewise not well organized and, therefore,are limited in their capacity to model many as-pects of brain development.More recently, 2D neural tube–like structures

called neural rosettes were established from isolatedneuroepithelium or the directed differentiation

1247125-4 18 JULY 2014 • VOL 345 ISSUE 6194 sciencemag.org SCIENCE

Optic cup organoid

Hindgut endoderm

Mesendoderm

PSCs

Ectoderm/neuroectoderm

Intermediate mesoderm Neuroepithelium Retinal epithelium

Wnt3aFgf4

Bmp4Fgf9 20% KSR 1.5% KSR

Matrigel+

agitation

2% matrigelMatrigel RA

Minimal mediaActivin A

Cerebral organoidKidney organoidIntestinal organoid

Fig. 3. Overview of organoid methodologies.Organoid differentiation strategies developed so far from human PSCs. Conditions and growth factors areindicated for the derivation of progenitor identities. For neuroectoderm, minimal medium without serum is used. KSR is knockout serum replacement, aserum-free growth-promoting alternative. Limiting its use, along with a low concentration of Matrigel dissolved in the medium, promotes retinalneuroepithelium, whereas higher KSR and embedding in pure Matrigel promotes the formation of various brain regions. Renal organoids have beengenerated several ways, but growth factors in common are shown.

RESEARCH | REVIEW

of PSCs (48, 49). Because neural rosettes re-capitulate apical-basal polarity and exhibitspontaneous radial organization similar tothat of the neural tube, they are more capableof recapitulating many aspects of brain devel-opment. These include the production of inter-mediate progenitor types, as well as the timedproduction of layer identities similar to thosein vivo (50). However, because of the 2D na-ture of the method, it has many limitations inmodeling the overall organization of the devel-oping brain.Therefore, alternative 3D culture methods with

the potential to recapitulate brain tissue organ-ization have been used extensively for investiga-tions in the past several years. In particular, workfrom the lab of Yoshiki Sasai has focused ondeveloping various isolated brain regions in 3Dfrom mouse or human PSCs (51). Beginning withEB formation, particular brain region identitiescan be generated from neuroectoderm. Specifi-

cally, forebrain tissues are generated by platingmouse (52) or human (53) EBs in 2D and ex-amining adherent cells. However, aggregates de-velop more complex structures when allowed tocontinue growing in 3D (54), eventually generat-ing dorsal forebrain. This method has furtherbeen improved in a recent study (55) that alsoshowed neuronal layering reminiscent of earlycerebral cortical development.Other regions can also be generated by mimick-

ing endogenous patterning with growth factors.For example, Hedgehog signaling drives ven-tral forebrain tissue (56). In addition, cere-bellar identities can be generated by treatmentwith either Bmp4 and Wnt3a to generate gran-ule neurons (57) or Hedgehog inhibition to gen-erate Purkinje neurons of the cerebellum (58).Conversely, minimizing exogenous bioactive fac-tors, such as serum proteins, promotes hypo-thalamic identity (59). Thus, by stimulatingneuroectoderm through an EB stage followed

by the application of specific growth factors,organoids can be generated for a variety of in-dividual brain regions.More recently, heterogeneous neural organ-

oids were established, termed cerebral organ-oids, that contain several different brain regionswithin individual organoids (28) (Fig. 3). Theapproach similarly begins with EBs, but growthfactors are not added to drive particular brainregion identities. Instead, the method is influ-enced by the intestinal organoid protocol, name-ly, by embedding the tissues in Matrigel. Theextracellular matrix provided by the Matrigelpromotes outgrowth of large buds of neuroepi-thelium, which then expand and develop intovarious brain regions. Cerebral organoids canreach sizes of up to a few millimeters when grownin a spinning bioreactor, which improves nu-trient and oxygen exchange. This expansion al-lows the formation of a variety of brain regions,including retina, dorsal cortex, ventral forebrain,

SCIENCE sciencemag.org 18 JULY 2014 • VOL 345 ISSUE 6194 1247125-5

Table 1. Current state of the art for in vitro self-organizing tissues of various organs. mESCs, mouse embryonic stem cells.

Organ Early reaggregationexperiments

Identity derivationfrom PSCs

3D self-organizing structureor organoid

Endoderm Thyroid Embryonic chick thyroid(90), adult rat thyroid (91)

Thyroid progenitors frommESCs (92)

Functional thyroid organoid frommESCs (70)

Lung Embryonic chick lung (93) Lung progenitors from mESCsand hiPSCs (92, 94)

Bronchioalveolar structures frommouse adult lung stem cells (71)

Pancreas Mouse embryonicpancreas (95)

Pancreatic endocrine cellsfrom mESC (96) andhESCs (97)

Pancreatic organoids frommouse embryonic pancreaticprogenitors (72)

Liver Chick embryonic liver (9) Hepatocytes from mESCsand hESCs (98)

Liver organoids from adultstem cells (40); liver budsfrom human iPSCs (41)

Stomach Chick embryonic gizzard andproventriculus (99)

None Stomach organoids from adult stemcells (36, 37)

Intestine Rat embryonic intestine (100) Intestinal cells from mESCs(101) and hPSCs (33)

Intestinal organoids from humanPSCs (33)

Mesoderm Heart Chick (102) and rat (103)cardiac tissue

Spontaneous and directeddifferentiation of mESCsand hESCs (104)

Vascularized cardiac patch fromhESCs (105)

Skeletal muscle Embryonic chick legskeletal muscle (76)

Mesoangioblasts from humaniPSCs (106)

Anchored contracting skeletal musclein 3D matrix derived from myoblastprogenitors (107)

Bone Skeletal bone of chickembryonic leg (77)

Osteoblasts from mESCs (108)and hESCs (109)

Bone spheroids from human osteogeniccells (110)

Kidney Chick embryonic kidney (9) Intermediate mesoderm frommouse (111) and human(112) PSCs

Ureteric bud (68) and metanephricmesenchyme (29) renal organoids(69) from human and mouse PSCs

Ectoderm Retina Embryonic chick retina (61) Retinal progenitors frommouse (113) and humanPSCs (114)

Optic cup organoids from mouse(64) and human (65) PSCs

Brain Embryonic chickbrain cells (45)

Neural rosettes from mouseand human PSCs (48, 49)

Cerebral organoids from mouseand human PSCs (28, 55)

Pituitary Chick anterior pituitary (115) None Adenohypophysis organoidsfrom mouse PSCs (73)

Mammary gland Mammary gland fromadult virgin mice (75)

None 3D breast epithelia embeddedin Matrigel (116)

Inner ear Embryonic chick otocysts (117) Inner ear hair cells frommESCs (118)

Inner ear organoids from mESCs (74)

Skin Embryonic chick skin andfeather follicles (9)

Keratinocytes from mESCs (119) Stratified epidermis from keratinocytesderived from mESCs (119)

RESEARCH | REVIEW

midbrain-hindbrain boundary, choroid plexus,and hippocampus.

Retinal organoids

The retina is the light-receptive neural region ofthe eye and is derived from the neural ectoderm.Initially, the optic vesicle forms as an outgrowthof the diencephalon, the most posterior regionof the forebrain (60). Like the rest of the CNS,this optic vesicle begins as a pseudostratifiedneuroepithelium with a fluid-filled lumen. How-ever, concerted movements at two hinge re-gions force the vesicle to fold in on itself, formingthe optic cup. Thus, early in retinal development,two adjacent epithelial layers are established:the outer retinal pigmented epithelium and theinner neural retina. This trend of stratificationcontinues and eventually leads to a fully lami-nated tissue containing layers of photoreceptorsand supportive cell types, such as bipolar cells andamacrine cells.The retina has a long history of in vitro reag-

gregation studies and has been used as a modelof retinal layer formation for decades (18). Thefirst reaggregates were generated from chickretina in the early 1960s and demonstrated therobustness of retinal self-organization in vitro(61, 62). Subsequent studies have used retinalreaggregates to examine the relations betweendifferent cell types and their differentiation andorganization (63).The evolution to PSC-derived retinal organoids,

like other organoid approaches, is built upon afoundation rooted in developmental biology(Fig. 3). EBs are derived in minimal medium togenerate neuroectoderm (64). A nominal amountof Matrigel is dissolved in the medium at anearly stage to allow the formation of more rigidneuroepithelial tissues, a prerequisite of retinalpigmented epithelium formation. This promotesthe formation of buds of retinal primordial tissuesimilar to the optic vesicle. These buds are thencut away from the rest of the neuroepithelial tis-sues and maintained in a medium that supportsretinal tissue identity.The resulting optic cup organoids very closely

mimic early retina. They display proper mark-ers of neural retina and retinal pigmented epi-thelium, they display retinal stratification withproper apical-basal polarity, and they undergomorphological tissue shape changes that mimicthe stepwise evagination and invagination of theoptic cup in vivo.More recently, optic cup organoids were gener-

ated from human PSCs (65). These human retinalorganoids show many of the characteristics com-mon to mouse retina; however, they display a num-ber of human specific features as well. In particular,the human retinal organoids are larger than mouseorganoids, they require more time to develop, andthey display certain tissue morphological differ-ences, such as apical nuclear positioning.

Kidney organoids

The kidney arises from an early embryonic tissuecalled the intermediate mesoderm, a subdivisionof mesodermal identity that develops from the

primitive streak (66). In vivo, the primitive streakdisplays opposing gradients of Bmp4 and activinA, which combinatorially specify the endodermor mesoderm. The intermediate mesoderm isfurther subspecified through the action of Fgfand Wnt signaling. This tissue then develops intotwo closely interacting domains, the ureteric budand the metanephric mesenchyme, which pro-mote each other’s growth and branching to de-velop early renal tubules.Like many of the tissues for which organoids

have so far been developed, evidence that kid-ney tissue might be capable of self-organizationcomes from early reaggregation experiments ofchick embryonic kidney (9). The resultant tis-sues displayed various segments of the nephron,including collecting duct, distal and proximaltubules, and glomeruli formed by the interac-tion with allantoic vessels upon transplantationon the chick allantoic membrane. Furthermore,the tissues could develop the stereotypic orga-nization of the kidney with cortical and medul-lary region. These experiments suggest that, ifkidney progenitors can be made from PSCs, thesewould, in principle, be capable of forming orga-nized tissues if grown in a permissive environment.This principle is what has now been demonstratedby three independent studies (67) (Fig. 3).Each of the recently published methods uses

various combinations of growth factors to mimicendogenous signaling to drive renal differenti-ation. Specifically, ureteric bud identity can begenerated by exposing human induced PSCs in2D to Bmp4 and Fgf2 to drive mesodermal iden-tities (68), followed by subsequent applicationof retinoic acid, Bmp2, and activin A. Such ure-teric bud cells can be cocultured with dissociatedmouse embryonic kidney to self-organize withinthe mouse aggregate and populate 3D uretericbud structures.The second major renal precursor tissue, the

metanephric mesenchyme, can instead be gener-ated beginning with an initial EB stage frommouseand human PSCs (29). Sequential application ofactivin A followed by Bmp4 and the Wnt agonistCHIR99021 then induces posterior mesoderm,the precursor to the intermediate mesoderm. Fi-nally, application of retinoic acid followed by Fgf9then stimulates the tissues to take on ametanephricmesenchyme identity. By coculturing with spinalcord tissue, a known nephric inducer, this tissuecan produce well-organized nephric tubules andeven nascent glomeruli.Finally, both principal lineages of the kid-

ney can be generated together (69) by applyingactivin A and Bmp4 to human embryonic stemcells (hESCs) grown in 2D to generate primitivestreak identity. These cells transition to an in-termediate mesoderm identity upon exposureto Fgf9 and spontaneously develop further intoureteric bud and metanephric mesenchyme inthe absence of further growth factors. Althoughthese specification events were initially per-formed in 2D, the cells take on 3D morphol-ogies by either growing at low density to allowdome-like colonies to form or when coculturedwith mouse kidney reaggregates. In both cases,

more complex tissues arise in 3D with structuresresembling ureteric epithelium and proximaltubules.

Organoids from model organisms

Despite the relatively few human-derived organo-ids so far described, several others have alreadybeen established from mouse PSCs or adult tis-sue stem cells. These include the endoderm-derivedthyroid, lung, and pancreas. Thyroid organoidscan be produced by overexpression of two fac-tors important for thyroid specification, Nkx2.1and Pax8, followed by treatment of EBs withthyroid-stimulating hormone (70). Lung organoidscan develop from cocultured adult bronchioal-veolar stem cells and lung endothelial cells (71)in Matrigel. And finally, pancreatic organoidscan develop from simply plating pancreatic pro-genitor cells in Matrigel (72). All three systemsgive rise to self-organized characteristic epithe-lia and, in the case of thyroid organoids, evensynthesis and secretion of functional thyroidhormone.Organoids have also been derived from ectoderm-

derived pituitary and inner ear. Specifically, thetwo identities of the developing pituitary, theneural ectodermal infundibulum and the ade-nohypophysis, could be generated in large EBsgrown under ectoderm-promoting conditions(73). Remarkably, these pituitary organoids canmature and synthesize pituitary hormones, suchas growth hormone, follicle-stimulating hor-mone, and thyroid-stimulating hormone. Addi-tionally, sensory epithelia of the inner ear canbe generated from EBs grown under ectoderm-promoting conditions with subsequent treat-ment with Bmp4 followed by Fgf2 to drive oticidentity (74). The resulting otic vesicles generatefunctional inner ear sensory epithelia with stereo-cilia and kinocilia.

The future of organoid technologies

The generation of 3D organoids from human PSCsis currently in its infancy, but the field is rapidlyevolving. In the near future, human organoidsmay be generated for those organs that have al-ready been established in mouse or where aprinciple of self-organization has already beendemonstrated in reaggregation studies. Theseinclude skin (9), mammary gland (75), muscle(76), and bone (77), to name a few.

Paradigms of organ development

Because organoids represent an easily accessiblemodel system, they have the potential to opendoors to developmental questions that havebeen difficult or impossible to answer using tra-ditional techniques. This is particularly true forbiological principles that are specific to humans.For example, human brain organoids have al-ready been used to examine the unique divisionmode of human neural stem cells (28). Similarly,retinal organoids have been used to test differ-ences between human and rodent tissue morpho-genesis and timing (65). Additionally, organoidsfor the GI tract can be applied to the study ofcoordinated development of GI organs, a process

1247125-6 18 JULY 2014 • VOL 345 ISSUE 6194 sciencemag.org SCIENCE

RESEARCH | REVIEW

that exhibits important differences in humanscompared with laboratory animals (78).Furthermore, organoids hold the potential to

model adult homeostasis as well. Indeed, intes-tinal organoids have already been used to ex-amine the role of the crypt niche in stem cellself-renewal and differentiation (79, 80). Thisis particularly true for organoids that have beenderived from adult progenitors, such as liverand stomach, that closely recapitulate regen-erative events seen in the adult organ.

Therapeutic potential

Disease modeling will likely be a primary focusof future organoid studies (Fig. 4). These canrange from developmental disorders, cancer, in-fectious disease, and degeneration. For example,gut organoids are already being used to examineinfectious diseases (81, 82); tumor biology (83, 84);and genetic conditions (85, 86).Along these lines, patient iPSCs will be a val-

uable tool in future disease modeling. Recently,kidney organoids were generated from iPSCsderived from a patient with polycystic kidneydisease (68). Although the method did not testfor a phenotype, this will likely represent an im-

portant tool in studying this and other genetickidney disorders. Similarly, retinal organoids havethe potential to model human genetic disordersthat lead to blindness, such as retinitis pigmen-tosa. These types of disorders can be modeledby making use of patient iPSCs or, alternatively,through the introduction of patient mutationsinto human PSCs using modern genome-editingtechnologies.Brain organoids, in particular, have huge po-

tential in this respect. They could, in principle,be used to model various neurodevelopmentaldisorders that have been difficult or impossibleto model in animals. Indeed, brain organoidswere the first organoids to make use of patientiPSCs in this manner and to model the develop-mental disorder microcephaly (28). In the future,cerebral organoids may even have the potentialto model disorders such as autism, schizophrenia,or epilepsy, and perhaps even adult-onset disor-ders like neurodegenerative diseases.Organoids also have the potential to be used for

testing efficacy and toxicity of drug compounds(Fig. 4). This could be applied to organoids thatmodel degenerative conditions—for example, liverfibrosis or cystic kidney diseases—where one could

screen for effective treatments. If successful, thisapproach could even cut down on the use of ani-mal testing, reserving it for studies requiringwhole-organism readouts. For this, developmentof human liver organoids would be of particularrelevance (Fig. 4), because the human liver oftenmetabolizes drugs in a manner distinct fromanimals’ metabolism. Drugs can be removed atearly stages of screening when they could other-wise be functional in humans, or more drastically,toxic metabolites can be produced specifically inhumans but not in tested animals. Methods toscreen compounds in an in vitro human livermodel are therefore being investigated as analternative in the drug discovery process (87).Human liver buds have already been shown toproduce human-specific metabolites (41), whichsuggests that liver organoids could represent anideal system to perform these types of studies.Finally, organoids have the potential to pro-

vide alternative approaches to cell or even whole-organ replacement strategies in the clinic (Fig. 4).Organoids could provide a source of autologoustissue for transplantation. In this respect, renalorganoids hold enormous therapeutic potential asthis is the organ with the highest rate of end-stagefailure leading to the highest organ demand fortransplants. Already, Taguchi et al. succeeded intransplanting kidney organoids under the renalcapsule of adult mice, which led to vasculari-zation, a promising step toward a replacementstrategy (29).Additionally, retinal organoids could be used

to treat certain types of retinal degeneration andblindness. Indeed, stem cell–therapy clinical trialsare already under way to replace certain degen-erating retinal cell types (88). Retinal organoidscould provide an alternative approach that maybetter recapitulate development and, therefore,produce particular cell types of interest for trans-plantation. Finally, intestinal organoids could pro-vide a treatment option for replacement of damagedcolon after injury or following removal of dis-eased tissue. Remarkably, intestinal organoid trans-plantation has already been performed inmice andcan contribute to colon repair after injury (35, 89).Organoid approaches could even allow for genecorrection in the case of genetic defects, usingmodern genome-editing technologies to replacedamaged organ with repaired tissue.Although it is clear that there are many po-

tential uses for organoids, it is important to keepin mind their current limitations. In particular,all of the organoid systems so far established re-main to be thoroughly characterized with regardto the extent of recapitulation of in vivo develop-ment. For example, although retinal organoidsnicely display typical laminar organization, outersegments fail to form; for example, photorecep-tors fail to fully mature to become light-sensitive.Likewise, cerebral organoids recapitulate fairlyearly events in brain development, but later fea-tures, such as cortical plate layers, fail to fully form.The issue of maturation seems to be a com-

mon hurdle in organoid technologies, and it re-mains to be seen whether this will significantlyaffect their therapeutic and research potential.

SCIENCE sciencemag.org 18 JULY 2014 • VOL 345 ISSUE 6194 1247125-7

Disease modeling

Drug efficacy testing

Organ replacement therapy

Drug safetytesting

Fig. 4. Therapeutic potential of organoids. Organoids can be used to model diseases (beige box),for example modeling neurodevelopmental disorders with cerebral organoids. These types of diseasemodels can then be used for testing drug efficacy in vitro before moving to animal models (greenbox). Drug compounds can be tested for toxicity and metabolic profile in liver organoids (gray box).And finally, organoids could be made from patient cells to provide autologous transplant solutions(pink box).

RESEARCH | REVIEW

Human intestinal organoids have been shownto display characteristics of mature intestine,producing Lgr5+ adult stem cells (33). Otherorganoids could perhaps be coaxed to fully ma-ture once transplanted, either ectopically or fortherapeutic purposes. These studies will likelybe a primary focus of future organoid research.Finally, the lack of vascularization is generally

an issue with organoids in vitro. Because of lim-itations in nutrient supply, organoids have alimited growth potential, which can also affecttheir maturation. Vascularization is an issue intissue engineering as a whole, and various ap-proaches have been taken to address it. In thecase of organoids, spinning bioreactors can pro-vide better nutrient exchange allowing for sizes ofup to a few millimeters (28). Alternatively, cocul-ture with endothelial cells can generate vascular-like networks (41). Perhaps the most promisingsolution, however, is the transplantation of thesetissues, as has been done for liver buds and kid-ney organoids, which stimulates invasion fromhost vasculature (29).Overall, organoids have enormous potential to

model development and disease, as a tool for drugtesting, and as a therapeutic approach. Futureefforts will no doubt bring them closer to reach-ing that potential.

REFERENCES AND NOTES

1. R. Lanza, J. Gearhart, B. Hogan, D. Melton, R. Pedersen, Eds.,Essentials of Stem Cell Biology (Elsevier, San Diego, CA, ed. 2, 2009).

2. Y. Hachitanda, M. Tsuneyoshi, Neuroblastoma with a distinctorganoid pattern: A clinicopathologic, immunohistochemical,and ultrastructural study. Hum. Pathol. 25, 67–72 (1994).doi: 10.1016/0046-8177(94)90173-2

3. B. Zimmermann, Lung organoid culture. Differentiation 36,86–109 (1987). doi: 10.1111/j.1432-0436.1987.tb00183.x;pmid: 3446562

4. C. Xinaris et al., In vivo maturation of functional renalorganoids formed from embryonic cell suspensions. J. Am.Soc. Nephrol. 23, 1857–1868 (2012). doi: 10.1681/ASN.2012050505; pmid: 23085631

5. P. B. Armstrong, Cell sorting out: The self-assembly oftissues in vitro. Crit. Rev. Biochem. Mol. Biol. 24, 119–149(1989). doi: 10.3109/10409238909086396; pmid: 2651008

6. S. F. Gilbert, Ed., Developmental Biology (Sinauer Associates,Sunderland, MA, ed. 6, 2000).

7. C. Dahmann, A. C. Oates, M. Brand, Boundary formation andmaintenance in tissue development. Nat. Rev. Genet. 12,43–55 (2011). doi: 10.1038/nrg2902; pmid: 21164524

8. C. Kiecker, A. Lumsden, Compartments and their boundaries invertebrate brain development. Nat. Rev. Neurosci. 6, 553–564(2005). doi: 10.1038/nrn1702; pmid: 15959467

9. P. Weiss, A. C. Taylor, Reconstitution of complete organs fromsingle-cell suspensions of chick embryos in advanced stages ofdifferentiation. Proc. Natl. Acad. Sci. U.S.A. 46,1177–1185 (1960). doi: 10.1073/pnas.46.9.1177; pmid: 16590731

10. A. Moscona, Rotation-mediated histogenetic aggregation ofdissociated cells. A quantifiable approach to cell interactionsin vitro. Exp. Cell Res. 22, 455–475 (1961). doi: 10.1016/0014-4827(61)90122-7; pmid: 13773012

11. L. Wolpert, Ed., Principles of Development (Oxford UniversityPress, Oxford, ed. 3, 2007).

12. M. S. Steinberg, in Cellular Membranes in Development, M. Locke,Ed. (Academic Press, New York, 1964), pp. 321–366.

13. R. A. Foty, C. M. Pfleger, G. Forgacs, M. S. Steinberg, Surfacetensions of embryonic tissues predict their mutual envelopmentbehavior. Development 122, 1611–1620 (1996).pmid: 8625847

14. T. Fujimori, S. Miyatani, M. Takeichi, Ectopic expression ofN-cadherin perturbs histogenesis in Xenopus embryos.Development 110, 97–104 (1990).pmid: 2081473

15. R. J. Detrick, D. Dickey, C. R. Kintner, The effects ofN-cadherin misexpression on morphogenesis in Xenopusembryos. Neuron 4, 493–506 (1990). doi: 10.1016/0896-6273(90)90108-R; pmid: 2322458

16. M. Agathocleous, W. A. Harris, From progenitors todifferentiated cells in the vertebrate retina. Annu. Rev. CellDev. Biol. 25, 45–69 (2009). doi: 10.1146/annurev.cellbio.042308.113259; pmid: 19575661

17. M. Žigman et al., Mammalian inscuteable regulates spindleorientation and cell fate in the developing retina. Neuron 48,539–545 (2005). doi: 10.1016/j.neuron.2005.09.030;pmid: 16301171

18. P. G. Layer, A. Robitzki, A. Rothermel, E. Willbold, Of layersand spheres: The reaggregate approach in tissue engineering.Trends Neurosci. 25, 131–134 (2002). doi: 10.1016/S0166-2236(00)02036-1; pmid: 11852139

19. G. Vollmer, P. G. Layer, Reaggregation of chick retinal andmixtures of retinal and pigment epithelial cells: The degree oflaminar organization is dependent on age. Neurosci. Lett. 63, 91–95(1986). doi: 10.1016/0304-3940(86)90019-4; pmid: 2419803

20. A. Rothermel, E. Willbold, W. J. Degrip, P. G. Layer,Pigmented epithelium induces complete retinal reconstitutionfrom dispersed embryonic chick retinae in reaggregationculture. Proc. Biol. Sci. 264, 1293–1302 (1997). doi: 10.1098/rspb.1997.0179; pmid: 9332014

21. C. Sergi, V. Ehemann, B. Beedgen, O. Linderkamp, H. F. Otto,Huge fetal sacrococcygeal teratoma with a completelyformed eye and intratumoral DNA ploidy heterogeneity.Pediatr. Dev. Pathol. 2, 50–57 (1999). doi: 10.1007/s100249900089; pmid: 9841706

22. N. Kuno et al., Mature ovarian cystic teratoma with a highlydifferentiated homunculus: A case report. Birth Defects Res.A Clin. Mol. Teratol. 70, 40–46 (2004). doi: 10.1002/bdra.10133; pmid: 14745894

23. M. Evans, Discovering pluripotency: 30 years of mouseembryonic stem cells. Nat. Rev. Mol. Cell Biol. 12, 680–686(2011). doi: 10.1038/nrm3190; pmid: 21941277

24. G. Weitzer, Embryonic stem cell-derived embryoid bodies: An invitro model of eutherian pregastrulation development and earlygastrulation. Handbook Exp. Pharmacol. 2006, 21–51 (2006).

25. R. Montesano, G. Schaller, L. Orci, Induction of epithelialtubular morphogenesis in vitro by fibroblast-derived solublefactors. Cell 66, 697–711 (1991). doi: 10.1016/0092-8674(91)90115-F; pmid: 1878968

26. C. Hagios, A. Lochter, M. J. Bissell, Tissue architecture: Theultimate regulator of epithelial function? Philos. Trans. R. Soc.Lond. B Biol. Sci. 353, 857–870 (1998). doi: 10.1098/rstb.1998.0250; pmid: 9684283

27. M. L. Li et al., Influence of a reconstituted basementmembrane and its components on casein gene expressionand secretion in mouse mammary epithelial cells. Proc. Natl.Acad. Sci. U.S.A. 84, 136–140 (1987). doi: 10.1073/pnas.84.1.136; pmid: 3467345

28. M. A. Lancaster et al., Cerebral organoids model human braindevelopment and microcephaly. Nature 501, 373–379 (2013).doi: 10.1038/nature12517; pmid: 23995685

29. A. Taguchi et al., Redefining the in vivo origin of metanephricnephron progenitors enables generation of complex kidneystructures from pluripotent stem cells. Cell Stem Cell 14, 53–67(2014). doi: 10.1016/j.stem.2013.11.010; pmid: 24332837

30. J. K. Heath, Transcriptional networks and signaling pathwaysthat govern vertebrate intestinal development. Curr. Top. Dev.Biol. 90, 159–192 (2010). doi: 10.1016/S0070-2153(10)90004-5; pmid: 20691849

31. J. Dessimoz, R. Opoka, J. J. Kordich, A. Grapin-Botton,J. M. Wells, FGF signaling is necessary for establishing guttube domains along the anterior-posterior axis in vivo. Mech.Dev. 123, 42–55 (2006). doi: 10.1016/j.mod.2005.10.001;pmid: 16326079

32. V. A. McLin, S. A. Rankin, A. M. Zorn, Repression of Wnt/beta-catenin signaling in the anterior endoderm is essential for liverand pancreas development. Development 134, 2207–2217(2007). doi: 10.1242/dev.001230; pmid: 17507400

33. J. R. Spence et al., Directed differentiation of human pluripotentstem cells into intestinal tissue in vitro. Nature 470, 105–109(2011). doi: 10.1038/nature09691; pmid: 21151107

34. T. Sato et al., Single Lgr5 stem cells build crypt-villusstructures in vitro without a mesenchymal niche. Nature 459,262–265 (2009). doi: 10.1038/nature07935; pmid: 19329995

35. S. Yui et al., Functional engraftment of colon epitheliumexpanded in vitro from a single adult Lgr5+ stem cell.Nat. Med. 18, 618–623 (2012). doi: 10.1038/nm.2695;pmid: 22406745

36. N. Barker et al., Lgr5(+ve) stem cells drive self-renewal in thestomach and build long-lived gastric units in vitro. Cell StemCell 6, 25–36 (2010). doi: 10.1016/j.stem.2009.11.013;pmid: 20085740

37. D. E. Stange et al., Differentiated Troy+ chief cells act asreserve stem cells to generate all lineages of the stomachepithelium. Cell 155, 357–368 (2013). doi: 10.1016/j.cell.2013.09.008; pmid: 24120136

38. H. Hisha et al., Establishment of a novel lingual organoidculture system: Generation of organoids having maturekeratinized epithelium from adult epithelial stem cells.Sci Rep 3, 3224 (2013). doi: 10.1038/srep03224;pmid: 24232854

39. K. S. Zaret, Regulatory phases of early liver development:Paradigms of organogenesis. Nat. Rev. Genet. 3, 499–512(2002). doi: 10.1038/nrg837; pmid: 12094228

40. M. Huch et al., In vitro expansion of single Lgr5+ liver stemcells induced by Wnt-driven regeneration. Nature 494, 247–250(2013). doi: 10.1038/nature11826; pmid: 23354049

41. T. Takebe et al., Vascularized and functional human liverfrom an iPSC-derived organ bud transplant. Nature 499,481–484 (2013). doi: 10.1038/nature12271; pmid: 23823721

42. T. J. Petros, J. A. Tyson, S. A. Anderson, Pluripotent stemcells for the study of CNS development. Front Mol Neurosci 4,30 (2011). doi: 10.3389/fnmol.2011.00030; pmid: 22016722

43. M. K. Lehtinen, C. A. Walsh, Neurogenesis at the brain-cerebrospinal fluid interface. Annu. Rev. Cell Dev. Biol. 27,653–679 (2011). doi: 10.1146/annurev-cellbio-092910-154026; pmid: 21801012

44. M. A. Lancaster, J. A. Knoblich, Spindle orientation inmammalian cerebral cortical development. Curr. Opin.Neurobiol. 22, 737–746 (2012). doi: 10.1016/j.conb.2012.04.003; pmid: 22554882

45. K. Ishii, Reconstruction of dissociated chick brain cells inrotation-mediated culture. Cytologia (Tokyo) 31, 89–98(1966). doi: 10.1508/cytologia.31.89; pmid: 5959228

46. L. Conti, E. Cattaneo, Neural stem cell systems: Physiologicalplayers or in vitro entities? Nat. Rev. Neurosci. 11, 176–187(2010). doi: 10.1038/nrn2938; pmid: 20107441

47. B. A. Reynolds, S. Weiss, Generation of neurons andastrocytes from isolated cells of the adult mammalian centralnervous system. Science 255, 1707–1710 (1992).doi: 10.1126/science.1553558; pmid: 1553558

48. Y. Elkabetz et al., Human ES cell-derived neural rosettesreveal a functionally distinct early neural stem cell stage.Genes Dev. 22, 152–165 (2008). doi: 10.1101/gad.1616208;pmid: 18198334

49. S. C. Zhang, M. Wernig, I. D. Duncan, O. Brüstle, J. A. Thomson,In vitro differentiation of transplantable neural precursors fromhuman embryonic stem cells. Nat. Biotechnol. 19, 1129–1133(2001). doi: 10.1038/nbt1201-1129; pmid: 11731781

50. N. Gaspard et al., An intrinsic mechanism of corticogenesisfrom embryonic stem cells. Nature 455, 351–357 (2008).doi: 10.1038/nature07287; pmid: 18716623

51. D. Cyranoski, Tissue engineering: The brainmaker. Nature488, 444–446 (2012).doi: 10.1038/488444a; pmid: 22914148

52. K. Watanabe et al., Directed differentiation of telencephalicprecursors from embryonic stem cells. Nat. Neurosci. 8,288–296 (2005). doi: 10.1038/nn1402; pmid: 15696161

53. K. Watanabe et al., A ROCK inhibitor permits survival ofdissociated human embryonic stem cells. Nat. Biotechnol. 25,681–686 (2007). doi: 10.1038/nbt1310; pmid: 17529971

54. M. Eiraku et al., Self-organized formation of polarized corticaltissues from ESCs and its active manipulation by extrinsicsignals. Cell Stem Cell 3, 519–532 (2008). doi: 10.1016/j.stem.2008.09.002; pmid: 18983967

55. T. Kadoshima et al., Self-organization of axial polarity, inside-out layer pattern, and species-specific progenitor dynamics inhuman ES cell-derived neocortex. Proc. Natl. Acad. Sci. U.S.A. 110,20284–20289 (2013). doi: 10.1073/pnas.1315710110;pmid: 24277810

56. T. Danjo et al., Subregional specification of embryonic stemcell-derived ventral telencephalic tissues by timed andcombinatory treatment with extrinsic signals. J. Neurosci. 31,1919–1933 (2011). doi: 10.1523/JNEUROSCI.5128-10.2011;pmid: 21289201

57. H.-L. Su et al., Generation of cerebellar neuron precursorsfrom embryonic stem cells. Dev. Biol. 290, 287–296 (2006).doi: 10.1016/j.ydbio.2005.11.010; pmid: 16406324

58. K. Muguruma et al., Ontogeny-recapitulating generationand tissue integration of ES cell-derived Purkinje cells.Nat. Neurosci. 13, 1171–1180 (2010). doi: 10.1038/nn.2638;pmid: 20835252

59. T. Wataya et al., Minimization of exogenous signals in ES cellculture induces rostral hypothalamic differentiation. Proc.Natl. Acad. Sci. U.S.A. 105, 11796–11801 (2008).doi: 10.1073/pnas.0803078105; pmid: 18697938

1247125-8 18 JULY 2014 • VOL 345 ISSUE 6194 sciencemag.org SCIENCE

RESEARCH | REVIEW

60. W. Heavner, L. Pevny, Eye development and retinogenesis.Cold Spring Harb. Perspect. Biol. 4, a008391 (2012).doi: 10.1101/cshperspect.a008391; pmid: 23071378

61. A. Stefanelli, A. M. Zacchei, V. Ceccherini, Retinalreconstitution in vitro after disaggregation of embryonicchicken eyes. Acta Embryol. Morphol. Exp. 4, 47–55 (1961).

62. A. A. Moscona, Analysis of cell recombinations in experimentalsynthesis of tissues in vitro. J. Cell. Comp. Physiol. 60 (suppl. S1),65–80 (1962). doi: 10.1002/jcp.1030600407

63. S. Nakagawa, S. Takada, R. Takada, M. Takeichi,Identification of the laminar-inducing factor: Wnt-signal fromthe anterior rim induces correct laminar formation of theneural retina in vitro. Dev. Biol. 260, 414–425 (2003).doi: 10.1016/S0012-1606(03)00320-8; pmid: 12921742

64. M. Eiraku et al., Self-organizing optic-cup morphogenesis inthree-dimensional culture. Nature 472, 51–56 (2011).doi: 10.1038/nature09941; pmid: 21475194

65. T. Nakano et al., Self-formation of optic cups and storablestratified neural retina from human ESCs. Cell Stem Cell 10, 771–785(2012). doi: 10.1016/j.stem.2012.05.009; pmid: 22704518

66. M. H. Little, A. P. McMahon, Mammalian kidney development:Principles, progress, and projections. Cold Spring Harb.Perspect. Biol. 4, a008300 (2012). doi: 10.1101/cshperspect.a008300; pmid: 22550230

67. B. D. Humphreys, Kidney structures differentiated from stemcells. Nat. Cell Biol. 16, 19–21 (2014). doi: 10.1038/ncb2904;pmid: 24366031

68. Y. Xia et al., Directed differentiation of human pluripotent cells toureteric bud kidney progenitor-like cells. Nat. Cell Biol. 15,1507–1515 (2013). doi: 10.1038/ncb2872; pmid: 24240476

69. M. Takasato et al., Directing human embryonic stem celldifferentiation towards a renal lineage generates a self-organizing kidney. Nat. Cell Biol. 16, 118–126 (2014).doi: 10.1038/ncb2894; pmid: 24335651

70. F. Antonica et al., Generation of functional thyroid fromembryonic stem cells. Nature 491, 66–71 (2012).doi: 10.1038/nature11525; pmid: 23051751

71. J.-H. Lee et al., Lung stem cell differentiation in mice directed byendothelial cells via a BMP4-NFATc1-thrombospondin-1 axis. Cell156, 440–455 (2014). doi: 10.1016/j.cell.2013.12.039; pmid: 24485453

72. C. Greggio et al., Artificial three-dimensional niches deconstructpancreas development in vitro. Development 140, 4452–4462(2013). doi: 10.1242/dev.096628; pmid: 24130330

73. H. Suga et al., Self-formation of functional adenohypophysisin three-dimensional culture. Nature 480, 57–62 (2011).doi: 10.1038/nature10637; pmid: 22080957

74. K. R. Koehler, A. M. Mikosz, A. I. Molosh, D. Patel, E. Hashino,Generation of inner ear sensory epithelia from pluripotentstem cells in 3D culture. Nature 500, 217–221 (2013).doi: 10.1038/nature12298; pmid: 23842490

75. F. Prop, G. J. Wiepjes, in Tissue Culture: Methods andApplications, P. F. Kruse Jr. and M. K. Patterson Jr., Eds.(Academic Press, New York, 1973) pp. 21–24.

76. I. R. Konigsberg, Clonal analysis of myogenesis: Its relevanceto the general problem of the stability of cell-type in culturedanimal cells is discussed. Science 140, 1273–1284 (1963).doi: 10.1126/science.140.3573.1273; pmid: 14034592

77. A. Moscona, H. Moscona, The dissociation and aggregationof cells from organ rudiments of the early chick embryo.J. Anat. 86, 287–301 (1952).pmid: 12980879

78. E. Lévy, E. Delvin, D. Ménard, J.-F. Beaulieu, Functionaldevelopment of human fetal gastrointestinal tract. Methods Mol.Biol. 550, 205–224 (2009). doi: 10.1007/978-1-60327-009-0_13; pmid: 19495706

79. T. Sato et al., Paneth cells constitute the niche for Lgr5 stemcells in intestinal crypts. Nature 469, 415–418 (2011).doi: 10.1038/nature09637; pmid: 21113151

80. A. Durand et al., Functional intestinal stem cells after Panethcell ablation induced by the loss of transcription factor Math1(Atoh1). Proc. Natl. Acad. Sci. U.S.A. 109, 8965–8970 (2012).doi: 10.1073/pnas.1201652109; pmid: 22586121

81. S. R. Finkbeiner et al., Stem cell-derived human intestinalorganoids as an infection model for rotaviruses. MBio 3,e00159-12 (2012). doi: 10.1128/mBio.00159-12; pmid: 22761392

82. A. Castellanos-Gonzalez, M. M. Cabada, J. Nichols, G. Gomez,A. C. White Jr., Human primary intestinal epithelial cells as animproved in vitro model for Cryptosporidium parvum infection.Infect. Immun. 81, 1996–2001 (2013). doi: 10.1128/IAI.01131-12;pmid: 23509153

83. T. M. Yeung, S. C. Gandhi, J. L. Wilding, R. Muschel,W. F. Bodmer, Cancer stem cells from colorectal cancer-derived cell lines. Proc. Natl. Acad. Sci. U.S.A. 107, 3722–3727(2010). doi: 10.1073/pnas.0915135107; pmid: 20133591

84. K. Onuma et al., Genetic reconstitution of tumorigenesis inprimary intestinal cells. Proc. Natl. Acad. Sci. U.S.A. 110, 11127–11132 (2013). doi: 10.1073/pnas.1221926110; pmid: 23776211

85. A. E. Bigorgne et al., TTC7A mutations disrupt intestinalepithelial apicobasal polarity. J. Clin. Invest. 124, 328–337(2013). doi: 10.1172/JCI71471; pmid: 24292712

86. J. F. Dekkers et al., A functional CFTR assay using primarycystic fibrosis intestinal organoids. Nat. Med. 19, 939–945(2013). doi: 10.1038/nm.3201; pmid: 23727931

87. C. Luni, E. Serena, N. Elvassore, Human-on-chip for therapydevelopment and fundamental science. Curr. Opin.Biotechnol. 25, 45–50 (2014). doi: 10.1016/j.copbio.2013.08.015; pmid: 24484880

88. C. M. Ramsden et al., Stem cells in retinal regeneration: Past,present and future. Development 140, 2576–2585(2013). doi: 10.1242/dev.092270; pmid: 23715550

89. R. P. Fordham et al., Transplantation of expanded fetalintestinal progenitors contributes to colon regeneration afterinjury. Cell Stem Cell 13, 734–744 (2013). doi: 10.1016/j.stem.2013.09.015; pmid: 24139758

90. S. R. Hilfer, L. B. Iszard, E. K. Hilfer, Follicle formation in theembryonic chick thyroid. II. Reorganization after dissociation.Z. Zellforsch. Mikrosk. Anat. 92, 256–269 (1968).doi: 10.1007/BF00335651; pmid: 4907211

91. J. M. Mallette, A. Anthony, Growth in culture of trypsindissociated thyroid cells from adult rats. Exp. Cell Res. 41, 642–651(1966). doi: 10.1016/S0014-4827(66)80115-5; pmid: 4952051

92. T. A. Longmire et al., Efficient derivation of purified lung andthyroid progenitors from embryonic stem cells. Cell Stem Cell 10,398–411 (2012). doi: 10.1016/j.stem.2012.01.019; pmid: 22482505

93. J. W. Grover, The enzymatic dissociation and reproduciblereaggregation in vitro of 11-day embryonic chick lung. Dev.Biol. 3, 555–568 (1961). doi: 0.1016/0012-1606(61)90032-X;pmid: 13902340

94. H. Mou et al., Generation of multipotent lung and airwayprogenitors from mouse ESCs and patient-specific cysticfibrosis iPSCs. Cell Stem Cell 10, 385–397 (2012).doi: 10.1016/j.stem.2012.01.018; pmid: 22482504

95. M. R. Bernfield, P. E. Fell, Separation of the proliferating anddifferentiating cell populations of cultured embryonicpancreatic epithelium. Proc. Natl. Acad. Sci. U.S.A. 58, 2227–2234(1967). doi: 10.1073/pnas.58.6.2227; pmid: 5242199

96. N. Lumelsky et al., Differentiation of embryonic stem cells toinsulin-secreting structures similar to pancreatic islets.Science 292, 1389–1394 (2001). doi: 10.1126/science.1058866; pmid: 11326082

97. K. A. D’Amour et al., Production of pancreatic hormone-expressing endocrine cells from human embryonic stemcells. Nat. Biotechnol. 24, 1392–1401 (2006). doi: 10.1038/nbt1259; pmid: 17053790

98. H. Shirahashi et al., Differentiation of human and mouseembryonic stem cells along a hepatocyte lineage. CellTransplant. 13, 197–211 (2004). doi: 10.3727/000000004783984016; pmid: 15191158

99. K. Urase, S. Yasugi, Induction and inhibition of epithelialdifferentiation by the mixed cell aggregates of themesenchymes from the chicken embryonic digestive tract.Dev. Growth Differ. 35, 33–40 (1993). doi: 10.1111/j.1440-169X.1993.00033.x

100. R. K. Montgomery, H. M. Zinman, B. T. Smith, Organotypicdifferentiation of trypsin-dissociated fetal rat intestine. Dev.Biol. 100, 181–189 (1983). doi: 10.1016/0012-1606(83)90209-9; pmid: 6617990

101. L. Cao et al., Intestinal lineage commitment of embryonicstem cells. Differentiation 81, 1–10 (2011). doi: 10.1016/j.diff.2010.09.182; pmid: 20934799

102. D. A. Fischman, A. A. Moscona, in Cardiac Hypertrophy, N. K. Alpert,Ed. (Academic Press, New York, 1971), pp. 125–139.

103. S. P. Halbert, R. Bruderer, T. M. Lin, In vitro organization ofdissociated rat cardiac cells into beating three-dimensionalstructures. J. Exp. Med. 133,677–695 (1971). doi: 10.1084/jem.133.4.677; pmid: 4926207

104. K. R. Boheler et al., Differentiation of pluripotent embryonicstem cells into cardiomyocytes. Circ. Res. 91, 189–201 (2002).doi: 10.1161/01.RES.0000027865.61704.32; pmid: 12169644

105. K. R. Stevens et al., Physiological function and transplantation ofscaffold-free and vascularized human cardiac muscle tissue. Proc.Natl. Acad. Sci. U.S.A. 106, 16568–16573 (2009). doi: 10.1073/pnas.0908381106; pmid: 19805339

106. F. S. Tedesco et al., Transplantation of genetically correctedhuman iPSC-derived progenitors in mice with limb-girdlemuscular dystrophy. Sci. Transl. Med. 4, 140ra89 (2012).doi: 10.1126/scitranslmed.3003541; pmid: 22745439

107. D. W. J. van der Schaft et al., Engineering skeletal muscle tissuesfrom murine myoblast progenitor cells and application of electricalstimulation. J. Vis. Exp. 2013, e4267 (2013).pmid: 23542531

108. L. D. Buttery et al., Differentiation of osteoblasts and in vitrobone formation from murine embryonic stem cells. TissueEng. 7, 89–99 (2001). doi: 10.1089/107632700300003323;pmid: 11224927

109. R. C. Bielby, A. R. Boccaccini, J. M. Polak, L. D. K. Buttery,In vitro differentiation and in vivo mineralization of osteogenic cellsderived from human embryonic stem cells. Tissue Eng. 10, 1518–1525(2004). doi: 10.1089/ten.2004.10.1518; pmid: 15588411

110. S. Kale et al., Three-dimensional cellular development is essentialfor ex vivo formation of human bone. Nat. Biotechnol. 18, 954–958(2000). doi: 10.1038/79439; pmid: 10973215

111. M. Nishikawa et al., Stepwise renal lineage differentiation ofmouse embryonic stem cells tracing in vivo development.Biochem. Biophys. Res. Commun. 417, 897–902 (2012).doi: 10.1016/j.bbrc.2011.12.071; pmid: 22209845

112. S. Mae et al., Monitoring and robust induction of nephrogenicintermediate mesoderm from human pluripotent stem cells.Nat. Commun. 4, 1367 (2013). doi: 10.1038/ncomms2378;pmid: 23340407

113. H. Ikeda et al., Generation of Rx+/Pax6+ neural retinalprecursors from embryonic stem cells. Proc. Natl. Acad. Sci.U.S.A. 102, 11331–11336 (2005). doi: 10.1073/pnas.0500010102; pmid: 16076961

114. D. A. Lamba, M. O. Karl, C. B. Ware, T. A. Reh, Efficientgeneration of retinal progenitor cells from human embryonicstem cells. Proc. Natl. Acad. Sci. U.S.A. 103, 12769–12774(2006). doi: 10.1073/pnas.0601990103; pmid: 16908856

115. H. Sobel, The behaviour in vitro of dissociated embryonicpituitary tissue. J. Embryol. Exp. Morphol. 6, 518–526 (1958).pmid: 13575665

116. G. Y. Lee, P. A. Kenny, E. H. Lee, M. J. Bissell, Three-dimensional culture models of normal and malignantbreast epithelial cells. Nat. Methods 4, 359–365 (2007).doi: 10.1038/nmeth1015; pmid: 17396127

117. M. F. Orr, Histogenesis of sensory epithelium in reaggregatesof dissociated embryonic chick otocysts. Dev. Biol. 17, 39–54(1968). doi: 10.1016/0012-1606(68)90088-2; pmid: 5642021

118. H. Li, G. Roblin, H. Liu, S. Heller, Generation of hair cells bystepwise differentiation of embryonic stem cells. Proc. Natl.Acad. Sci. U.S.A. 100, 13495–13500 (2003). doi: 10.1073/pnas.2334503100; pmid: 14593207

119. C. Coraux et al., Reconstituted skin from murine embryonicstem cells. Curr. Biol. 13, 849–853 (2003). doi: 10.1016/S0960-9822(03)00296-3; pmid: 12747834

120. H. V. Wilson, A new method by which sponges may beartificially reared. Science 25, 912–915 (1907). doi: 10.1126/science.25.649.912; pmid: 17842577

121. J. Holtfreter, Experimental studies on the development of thepronephros. Rev. Can. Biol. 3, 220–250 (1944).

122. G. B. Pierce Jr., E. L. Verney, An in vitro and in vivo study ofdifferentiation in teratocarcinomas. Cancer 14, 1017–1029(1961). doi: 10.1002/1097-0142(196109/10)14:5<1017::AID-CNCR2820140516>3.0.CO;2-P; pmid: 13735537

123. M. Evans, Origin of mouse embryonal carcinoma cells and thepossibility of their direct isolation into tissue culture. J. Reprod. Fertil.62, 625–631 (1981). doi: 10.1530/jrf.0.0620625; pmid: 7019433

124. G. R. Martin, Isolation of a pluripotent cell line from earlymouse embryos cultured in medium conditioned byteratocarcinoma stem cells. Proc. Natl. Acad. Sci. U.S.A. 78, 7634–7638(1981). doi: 10.1073/pnas.78.12.7634; pmid: 6950406

125. J. M. Shannon, R. J. Mason, S. D. Jennings, Functionaldifferentiation of alveolar type II epithelial cells in vitro:Effects of cell shape, cell-matrix interactions and cell-cellinteractions. Biochim. Biophys. Acta 931, 143–156 (1987).doi: 10.1016/0167-4889(87)90200-X; pmid: 3663713

126. J. A. Thomson et al., Embryonic stem cell lines derived fromhuman blastocysts. Science 282, 1145–1147 (1998).doi: 10.1126/science.282.5391.1145; pmid: 9804556

ACKNOWLEDGMENTS

We are grateful to N. Corsini and M. Renner for critical feedbackas well as other members of the Knoblich lab for discussions.M.A.L. received funding from the European Molecular BiologyOrganization (EMBO), the Helen Hay Whitney Foundation, and aMarie Curie post-doctoral fellowship. Work in J.A.K.’s laboratory issupported by the Austrian Academy of Sciences, the AustrianScience Fund (FWF) (projects Z153-B09 and I552-B19), and anadvanced grant from the European Research Council.

10.1126/science.1247125

SCIENCE sciencemag.org 18 JULY 2014 • VOL 345 ISSUE 6194 1247125-9

RESEARCH | REVIEW

RESEARCH

SCIENCE sciencemag.org 13 JUNE 2014 • VOL 344 ISSUE 6189 1243

BACKGROUND: At the

surface of body or-

gans, epithelial tis-

sues must withstand

harsh external envi-

ronments. To do so,

they rely heavily upon

stem cells to replenish

and repair wounds and replace the many

cells that die from this wear and tear. To

maintain tissue size, the number of cells

lost must be compensated by cell divisions.

Tissue homeostasis and wound repair are

ensured by stem cells, located within spe-

cialized microenvironments, referred to as

niches. Each niche is tailored to accommo-

date the regenerative needs of its tissue.

Some tissues—for instance, skin epithe-

lium—harbor multiple stem cell niches,

each with their own responsibility for

maintaining cellular balance within their

particular domain. Governance of discrete

tissue units has ancient origins and is also

seen in Drosophila gut epithelium.

Identifying stem cells and tracking their

progeny is accelerated by lineage tracing, a

technique in which a stem cell is genetically

marked in its niche and in a way such that

their subsequent progeny retain marker ex-

pression. Although interpretation of these

experiments has been complicated by the

lack of specificity of most stem cell mark-

ers, this method can be helpful in evaluat-

ing the contribution of stem cells to tissue

homeostasis and wound repair. Additional

tools include live imaging of marked stem

cells and ablating stem cells in situ either

by laser or by targeted expression of diph-

theria toxin/receptor in stem cells.

ADVANCES: Accumulating evidence on

bone marrow, intestinal stem cell crypts,

and hair follicles suggests that stem cells

often exist in two distinct states based upon

their relative activity and/or their ease of ac-

tivation during homeostasis and/or wound-

induced regeneration. Recent studies on

the hair follicle reveal that signals emanat-

ing from both heterologous niche cells and

from lineage progeny influence the tim-

ing and length of stem cell activity. This in

turn can profoundly affect the amount of tis-

sue regenerated. Stem

cell ablation studies on

both intestinal and hair

follicle stem cell niches

further show that the

two states are intercon-

vertible, perhaps best

exemplified by the ability of a single intes-

tinal stem cell to eventually outcompete its

siblings during rounds of turnover within

an intestinal villus.

Additional new findings suggest that

fates and multilineage potentials of epi-

thelial stem cells can change, depending

upon whether a stem cell exists within its

resident niche and responds to normal tis-

sue homeostasis, whether it is mobilized

to repair a wound, or whether it is taken

from its niche and challenged to de novo

tissue morphogenesis after transplantation.

In this Review, we discuss how naturally

lineage-restricted populations of stem cells

and committed progenitors can display

such remarkable plasticity under these

different conditions.

OUTLOOK: Although the molecular mech-

anisms underlying cellular plasticity, fate

conversion, and reacquisition of stem cell

properties in committed and/or differenti-

ated cells still remain poorly understood,

this cellular plasticity and lineage revers-

ibility may represent adaptive mechanisms

for the self-preservation of epithelia to re-

pair body surfaces and linings in whatever

ways possible following injuries. When gone

awry, these repertoires become the curse

of epithelial stem cells, contributing in

major ways to human cancers.

Plasticity of epithelial stem cells in tissue regeneration

STEM CELL PLASTICITY

Cédric Blanpain1,2* and Elaine Fuchs3*

1Institut de Recherche Interdisciplinaire en Biologie Humaine et Moléculaire (IRIBHM), Université Libre de Bruxelles, Brussels B-1070, Belgium. 2Walloon Excellence in Life Sciences and Biotechnology (WELBIO), Université Libre de Bruxelles (ULB), Brussels B-1070, Belgium. 3Howard Hughes Medical Institute, The Rockefeller University, New York, NY 10065, USA.*Corresponding author. E-mail: fuchslb@rockefeller.edu (E.F.); Cedric.Blanpain@ulb.ac.be (C.B.) Cite this article as C. Blanpain and E. Fuchs, Science 344, 1242281 (2014). DOI: 10.1126/science.1242281

Read the full article at http://dx.doi.org/10.1126/science.1242281

ON OUR WEBSITE

Coordinating stem cell activity to match tissue output. Stem cells (purple) often exist in

two states, one more quiescent than the other. Primed stem cells are closer to activating niche

signals (green). They typically respond faster and generate shorter-lived progenitors (orange),

which also signal, fueling tissue production. Each stem cell niche must be responsive to the

regenerative demands of tissue homeostasis and wound repair and adjust niche activating and

inhibitory signals as necessary.

Activating

signals

Tissue

regeneration

Active

Short-lived

Primed

Quiescent

Interconvertible

stem cell states

REVIEW SUMMARY

ILL

US

TR

AT

ION

: K

. S

UT

LIF

F/SCIENCE

SP

ECIAL SERIES: STEM C

ELL

S

Published by AAAS

REVIEW◥

STEM CELL PLASTICITY

Plasticity of epithelial stem cells intissue regenerationCédric Blanpain1,2* and Elaine Fuchs3*

Tissues rely upon stem cells for homeostasis and repair. Recent studies show that the fateand multilineage potential of epithelial stem cells can change depending on whether astem cell exists within its resident niche and responds to normal tissue homeostasis,whether it is mobilized to repair a wound, or whether it is taken from its niche andchallenged to de novo tissue morphogenesis after transplantation. In this Review, wediscuss how different populations of naturally lineage-restricted stem cells and committedprogenitors can display remarkable plasticity and reversibility and reacquire long-termself-renewing capacities and multilineage differentiation potential during physiological andregenerative conditions. We also discuss the implications of cellular plasticity forregenerative medicine and for cancer.

Epithelia are cellular sheets often residingat the interface between the external envi-ronment and body organs, including skin,gut, airway tracts, kidney, liver, mammaryglands, and prostate. They perform a di-

verse array of physiological functions, includingthe ability to retain body fluids, absorb nutrients,filter and eliminate toxic by-products of metab-olism, and regulate body temperature. Each epi-thelium is morphologically and molecularly suitedto its particular task, a feature that necessitatesspecialized cell lineages.Most epithelia replenish themselves through

a process called tissue homeostasis, in which thenumber of cell divisions within a tissue com-pensates for the number of cells lost (1). Tissuehomeostasis is ensured by the existence of stemcells (SCs) located within specialized microen-vironments, referred to as niches. Each niche istailored to accommodate the regeneration needsof the tissue (2).The skin epidermis and its appendages (hair

follicles, sebaceous glands, and sweat glands)harbor spatially distinct SC niches. The inner-most (basal) layer of interfollicular epidermis(IFE) harbors proliferative progenitors, whichgenerate the stratified layers of the skin barrier.Every few weeks, the IFE renews itself almostentirely, placing a constant demand on its SCs.Sebaceous glands (SGs) also turnover continu-ously during adult homeostasis. By contrast, hairfollicles (HFs) cycle through bouts of hair growth

and degeneration, necessitating only periodic useof SCs, whereas sweat gland (SwG) cells are most-ly quiescent (Fig. 1A).Other epithelia also have distinct require-

ments for tissue homeostasis, which must bemet by their resident SCs. In the small intestine,the epithelium is organized into a crypto-villusunit (Fig. 1B). The crypt is composed of co-lumnar basal cells (CBCs) intermingled withPaneth cells at the crypt base; an overlying com-partment of transit-amplifying (TA) cells dividesseveral times and then terminally differentiatesto generate the absorptive and secretory cells

of the villus. Villus cells are subsequently shedinto the lumen (3), which results in continualturnover of the entire crypt every 3 to 5 days.CBCs, now known to be SCs, fuel the process.

Functionally validating stemnessof epithelial cells in vitro

Different methods have been elaborated through-out the years to study the fate, renewal, and dif-ferentiation potential of epithelial SCs. The firstfunctional demonstration of an epithelial SCwas made when methods were identified toculture human epidermal keratinocytes underconditions where they could be maintained andpropagated for hundreds of generations with-out losing stemness (4). When grown from anunaffected region of a burn patient, expandedepidermal cultures could be stably engraftedonto the damaged skin (5). Engrafted epider-mis did not develop cancer or other abnor-malities, which indicated that, under the rightconditions—in this case, coculture with irradiateddermal fibroblasts—in vitro SC expansion anddifferentiation can be achieved without delete-rious consequence.The requirement of dermal neighbors for suc-

cessful culturing of epidermal SCs highlights thereliance of SCs on cross-talk with their nichemicroenvironment. Indeed, by elucidating keyheterologous niche components and/or thecross-talk involved, SCs from many differentepithelia have since been successfully cultured.For intestinal stem cells (ISCs), it took BMP andNotch inhibition together with Wnt activationto recapitulate in vitro the long-term prolifera-tive capacity and multipotency normally con-ferred to ISCs by their niche (6). These studiesunderscore the complexities of signaling cir-cuitry governing SC behavior and the need to

RESEARCH

1Institut de Recherche Interdisciplinaire en Biologie Humaineet Moléculaire (IRIBHM), Université Libre de Bruxelles,Brussels B-1070, Belgium. 2Walloon Excellence in LifeSciences and Biotechnology (WELBIO), Université Libre deBruxelles (ULB), Brussels B-1070, Belgium. 3Howard HughesMedical Institute, The Rockefeller University, New York, NY10065, USA.*Corresponding author. E-mail: fuchslb@rockefeller.edu (E.F.);cedric.blanpain@ulb.ac.be (C.B.)

Dermal papilla (DP)

Matrix transit-amplifying (TA) cells

Outer root sheath (ORS)

Hair shaft

Bulge SC

Sebaceous gland (SG)

IsthmusInfundibulum

Interfollicular epidermis (IFE)

A

B

DD

Villus

Crypt

Enterocytes

Goblet cells

Neuroendocrine cells

+4 Border SC (Lgr5/Hopx/Bmi1)

TA cells (Bmi1/Hopx/Dll1)

Paneth cells

Lgr5+ CBCs+1+2

+3+4

Fig. 1. Skin and intestinal epithelia: paradigms for epithelial stem cell biology. (A) Schematicillustrating the epithelial lineages of hairy skin, color-coded here, which derive from at least four distinctstem cell populations. (B) Schematic illustrating the location of intestinal crypt stem cells (green), givingrise to TA cells and, in turn, four distinct cell types, three in the villus and one in the crypt.

SCIENCE sciencemag.org 13 JUNE 2014 • VOL 344 ISSUE 6189 1242281-1

understand this to maintain SCs in the absenceof other heterologous cell types in vitro.

Identifying epithelial SCs in vivoand probing their roles intissue homeostasisHF homeostasis

Lineage tracing entails the genetic marking ofone or a group of cells in their normal physiolog-ical context in a way that their subsequent pro-geny retain marker expression. This method ispowerful in evaluating the contribution of SCs totissue homeostasis (1). The fluctuations of HFsthrough synchronized bouts of hair growth andinactivity present an interesting variation on thistheme (Fig. 2A). Before modern-day genetics, cellswith proliferative potential that spent extendedperiods in quiescence were marked and monitoredby nucleotide analog pulse-chase experiments.Such label-retaining cells (LRCs) reside at thebase of the resting HF, a region now referred toas the bulge and its associated hair germ (HG)(7). LRCs are SCs, as demonstrated by using a reg-ulatable fluorescent histone to label LRCs andmonitor their cell divisions, as well as lineagetracing to follow their fate (8–12) (Fig. 2B).Both bulge and HG share many molecular fea-

tures of stemness, including expression of Lgr5and Sox9 (12, 13). However, HG cells are alwaysthe first to be activated at the start of each newhair cycle, and they undergo more divisions thanbulge cells (13). Their close proximity to the un-derlying mesenchymal signaling center, thedermal papillae (DP), functions in dictating thisearly response.Activated HG cells do not maintain stemness

in vitro (13), and in vivo, they generate the TAcells that produce the hair and its channel (14, 15).By contrast, once the new hair cycle initiates,some bulge cells leave their niche and form aninverse proliferative gradient along the emerg-ing outer root sheath (ORS). Early in the hair-growth phase, TA cells stimulate remainingbulge cells to proliferate and replenish the niche(15). ORS cells closest to the bulge return toquiescence soon thereafter and form a newbulge and HG for the next cycle (12, 16). Theability of bulge and HG SCs to generate theseven different HF lineages underscores theirmultilineage potency. Additionally, even thoughbulge normally gives rise to HG, HG can re-plenish an empty bulge niche, as shown by laserablation and live imaging (16), which under-scores their close relation and capacity to inter-convert when necessary (see below).Although the above studies disclose insights

into the behavior and maintenance of cyclingHFs, lineage tracings reveal the existence of atleast two additional SC populations—SG andinfundibulum—within the noncycling HF segment.SGs are maintained by unipotent Lgr6+Lrig1+

SCs that arise from Blimp1-expressing progen-itors (17). In adults, Lgr6-expressing cells markand sustain SGs (18, 19), whereas Lrig1 expres-sion extends to SCs fueling infundibulum ho-meostasis (19) (Fig. 2A). One other SC populationin the upper bulge region has been suggested on

IFE SC tracing

Infundibulum SC tracing

IFEprogenitortracing

Bulge SCtracing

Bulge SClabeling

SG SC tracing

HGDP

Bulge SC transplantation

FACS isolation of bulge SC

a6

CD

34

BA

Fig. 2. Epidermal homeostasis is achieved through distinct pools of stem cells. (A) Schematic illus-trating the outcome of five separate lineage tracings of Rosa26-floxed-stop-floxed–reporter mice. In eachexperiment, a different inducible Cre recombinase was expressed in the desired SC or progenitor com-partment. Because the Rosa26 promoter is generic, once Cre is activated and the stop codon is excised,the marked cells and all their downstream progeny express the reporter. The results shown here illustratethat each SC compartment is responsible for sustaining tissue homeostasis within a discrete skin domain.(B) We purified fluorescently marked bulge SCs (green) by fluorescence activated cell sorting (FACS) andcultured them as individual colonies of cells before transplanting the cells to a hairless mouse. Theexperiment illustrated that a clone from a single bulge SC can regenerate the entire skin epithelium, whichdocuments the stemness and multipotency of the cells (9, 69, 70). We now know that when taken out oftheir native niche and engrafted, epithelial SCs are often less restricted in their fates.

1242281-2 13 JUNE 2014 • VOL 344 ISSUE 6189 sciencemag.org SCIENCE

RESEARCH | REVIEW

the basis of its encasement by sensory nervesheaths (20). Whether these cells represent anindependent pool of functional SCs remainsunresolved.A sharp boundary exists between infundibulum-

derived Lrig1+ cells and IFE (19), and little if anycontribution to the IFE has been observed bythe various adult SCs thus far identified in theHF (9, 10, 19, 21–23) (Fig. 2A). This argues againstthe prior view that a single “master” SC pop-ulation presides over all skin lineages, as initially

postulated based upon embryonic Lgr6-Cre lin-eage tracing (18). Indeed, the paradigm forsegmental-tissue governance by SC units hasancient origins, as, like the HF, Drosophila in-testinal epithelium is also compartmentalizedinto discrete units maintained by separate SCpopulations (24).

IFE homeostasis

The IFE is maintained by juxtaposition of smallunits of proliferation containing stem and/or

progenitor cells (1). During embryogenesis, thesingle layer of K14+ epidermal basal progenitorsundergoes a spindle orientation shift from >90%symmetric to ~70% asymmetric cell divisions,which leaves one daughter in the basal layerand one suprabasal differentiating daughter cell(25). Postnatally, SCs and transient progenitorscoexist within the IFE basal layer, and bothexpress K14 but can be distinguished by theirsurvival rate, mode of division, gene expres-sion, and ability to respond to tissue damage

Crypt colonization by border cells

Lgr5+ CBC lineage tracing during homeostasis

Dynamic of Lgr5+ bottom/centre cells

Short-livedclones

Border cells

Weeks Months

A

CB

D

Bottom/center cells

Dynamic of Lgr5+ border cells

Intestinal homeostasis following Lgr5 cell ablation

Bmi1 tracing

Lgr5+ ablation

E

Labeling

+1

+2

+3

+4

+1

+2

+3

+4

+1

+2

+3

+4

+1

+2

+3

+4

Monoclonal drift of Lgr5+ CBCs

Fig. 3. Interconversion and monoclonal drift of intestinal stem cells. (A) Lineage tracing of Lgr5+ cells (green) showing that these crypt cells giverise to all intestinal lineages during homeostasis (38). (B and C) Intravital microscopy showing the colonization of the crypt from Lgr5 cells at bottomcenter. Bmi1+ border (+4) cells either colonize the bottom of the crypt or give rise to TA cells (red) (42). (D) Lineage ablation of Lgr5+ (yellow X’s)prompts Bmi1+ cells (red) to convert into Lgr5+ crypt cells, and thus gut homeostasis is not impaired (43). (E) Multicolor lineage tracing rapidly leads tounicolor crypts, which demonstrate the monoclonal drift of ISCs (49).

SCIENCE sciencemag.org 13 JUNE 2014 • VOL 344 ISSUE 6189 1242281-3

RESEARCH | REVIEW

(26) (Fig. 2A). Basal progenitors targeted byAh-CreER (27, 28), Inv-CreER (26), and possiblyAxin2-CreER (29), divide mostly asymmetrical-ly, whereas K14+ basal SCs are integrin-richand divide mostly symmetrically to generatetwo long-lived daughters (26, 30). Althoughthe exact nature of the imbalance between SCand progenitor division is not yet clear, each SC-progenitor division must also be accompaniedby some differentiation, driven in part by Notchsignaling (31–34).

Crypt homeostasis:A one-cell–winner competition

Ultrastructural analyses and proliferative ca-pacity of the intestinal crypt led to the initialhypothesis that CBCs are ISCs (35) (Fig. 1B).Subsequent assignment of stemness favoredcells at the +4 position, given their mode ofchromosome segregation (36) and higher re-sistance to DNA damage–induced cell death(37). Lineage tracings of +4 CBCs with Bmi1,mTER, and Hopx-CreER and 0→+3 CBCs withLgr5-CreER revealed that all crypt CBCs behave asinterconvertible multipotent ISCs (38–42) (Fig. 3,A to C). This is further exemplified by diphtheriatoxin (DT)–targeted ablation of Lgr5-expressingcells, which does not impair intestinal homeost-asis (43) (Fig. 3D). Thus, despite their markedlydifferent regenerative demands, both HF andintestine have spatially discrete interconvertibleSCs existing in quiescent and primed and/oractivated states (bulge and HG versus +4 and0→+3 crypt cells).Although it was initially proposed that all

Lgr5+ ISCs cycle rapidly (38), a recent studyusing yellow fluorescent protein and histoneH2B label–retention assays reveals that ~20%of Lgr5-expressing cells cycle less frequently,exhibit a mixed ISC–Paneth cell transcriptionalprofile, and differentiate into Paneth and neuro-endocrine cells (44). Although these slow-cyclingcells do not contribute to crypt homeostasis dur-ing physiological conditions, they can form organ-oids in vitro with comparable efficiencies as rapidlycycling Lgr5+ ISCs and can mediate crypt regen-eration after injuries (44).Despite these behavioral distinctions among

ISCs, their cellular dynamics within the cryptsystematically drift toward monoclonality (45–49).Thus, over time, each crypt-villus unit derivesfrom a single ISC (Fig. 3E). The mechanismleading to crypt monoclonality is thought to de-rive from neutral competition between an equi-potent pool of ISCs that includes both Lgr5 andBmi1-Hopx ISCs (49). In contrast to epidermis(in which progenitors divide mostly asymmet-rically), ISCs are thought to divide symmetrical-ly and compete for niche space (48, 49). Basedinitially on Lgr5 expression and mathematicalmodeling (48, 49) and subsequently on a novelmethod of continuous labeling (50), it is esti-mated that between 5 and 16 Lgr5+ ISCs com-pete with each other for niche space in a neutraldrift manner.Live imaging of Lgr5-CreER lineage tracing

has recently enabled the visualization of these

displacements during ISC divisions. Ironi-cally, with each division, ISCs reorganize theirposition within the crypt, which underscores theirinterconvertibility (42) (Fig. 3, B and C). In theend, one ISC outcompetes the others. It will beinteresting to see in the future whether such com-petition happens in other SC niches and howthe competition unfolds at a molecular level.Crypt monoclonality underscores the multi-

lineage potential of ISCs. Increasing evidencesuggests that their fate choices are rooted at thetranscriptional level. Thus, equipotent progenitorsundergoing Notch-mediated lateral inhibitionquickly enable distinct—in this case, reversible—cell fates to establish progenitor cell lineages aseither absorptive or secretory. Moreover, Atoh1,a secretory-specific transcription factor expressedby ISCs, controls lateral inhibition through Dllgenes and also drives expression of secretorylineage genes, which suggests that intestinal

crypt lineage plasticity involves a lineage-restrictedtranscription factor expressed by multipotentISCs (51).

Switch from multipotency to unipotencyin glandular epithelia

Mammary glands (MGs), SwGs, and prostate glandsare composed of an inner luminal layer, surroundedby an outer layer of myoepithelial and/or basalcells. Their morphogenesis begins late in embryo-genesis and is completed postnatally.As judged by lineage tracing, both MGs and

SwGs and their associated ducts originate fromK14-expressing multipotent embryonic epidermalprogenitors (52–54). Although it was recentlysuggested that some bipotent SCs persist withinthe myoepithelial layer (55), myoepithelial andluminal lineages of MGs, SwGs, and prostate arelargely maintained postnatally by distinct pools ofunipotent SCs (52–54, 56–60) (Fig. 4A).

Unipotentmyoepithelial cells

Unipotent luminal cells

Unipotent luminal alveolar progenitors

Mammary and sweat glands during regeneration

Mammary gland during puberty and lactation

Luminal cell tracing

Myoepithelial cell tracing

MyoepithelialSC tracing

FACS isolation of myoepithelial SC

Transplantation into mammary mesenchyme

A

B

CD29

CD

24

Fig. 4. Plasticity ofglandular epithelium duringregeneration. (A) Lineagetracing reveals that duringpuberty and pregnancy, MGexpansion is sustained largely by unipotent myoepithelial cells (red) andluminal cells (green) (52). (B) After transplantation into mammary me-senchyme, unipotent myoepithelial cells (red) from the MG or the SwGacquire multipotency and reform a new gland replete with basal and luminal cells (52, 53).

1242281-4 13 JUNE 2014 • VOL 344 ISSUE 6189 sciencemag.org SCIENCE

RESEARCH | REVIEW

In the adult, both myoepithelial and luminalepithelial SwG SCs display very little turnoverduring homeostasis (53). By contrast, MG’s SCs

exert tremendous tissue-generating potential dur-ing puberty and pregnancy, making them espe-cially well suited for studying glandular SC biology

(52, 54). Heterogeneity within luminal and alveolarcompartments has been seen with Notch2-CreER

and Notch3-CreER lineage tracing (59, 60). Whether

A

C

B

D

Labeling Wounding Short-term repair Long-term repair

IFE SC

IFE progenitor

Bulge repair

Bulge ablation Hair cycle

Infundibulum SC

Bulge SC

HG labeling

xx x

xxx

x

Fig. 5. Plasticity of epidermal cells during tissue repair. (A) Lineage tracing of IFE SCs (blue)and progenitors (grey) during wound healing showing that SCs stably contribute to epidermalrepair while progenitor contribution is only transient (26). (B andC) Lineage tracing of bulge (B)and infundibulumSCs (C)demonstrate that adult HFSCs are rapidly recruited to IFEduringwounding, but very fewcells survive and contribute to IFE homeostasisafter wound repair (19, 21). (D) After ablation of bulge cells (red X’s), hair germ (HG) cells (green) recolonize the bulge niche and mediate hair regeneration (16).

SCIENCE sciencemag.org 13 JUNE 2014 • VOL 344 ISSUE 6189 1242281-5

RESEARCH | REVIEW

these two luminal populations can interconvertremains unknown.During prostate development, clonal analyses

also suggest heterogeneity, this time in the basalcompartment. Bipotent and unipotent basal pro-genitors have been identified, as well as basal cellsalready committed to the luminal lineage (61).Whether this apparent cellular heterogeneity re-flects the existence of distinct progenitors or,alternatively, stochastic fate decisions of a singlemultipotent progenitor remains to be deter-mined (61).Altogether, lineage-tracing experiments per-

formed in different glandular epithelia show thatthey initially develop from multipotent progeni-tors which are progressively replaced by unipo-tent SCs for adult tissue homeostasis and repair(52, 53, 56, 61). However, despite similar histolo-gies and SC behaviors, their multipotency →unipotency switch occurs at different times duringdevelopment (52, 53, 56–58, 61).

Transient plasticity of epithelialSC during tissue repair

Over evolution, homeostasis has been optimizedfor different SC compartments to replace localcells that die. However, if one SC compartmentis damaged, other SCs must be recruited to re-pair the injury. A series of recent studies revealsthat the fate and differentiation potential of epi-thelial cells can broaden during tissue regenerationafter wounding. In some cases, unipotent progenitorsacquire multipotency, whereas, in others, nor-mally committed cells revert back to a SC-likestate to ensure tissue regeneration. The cellularplasticity and reversibility observed in adultepithelial tissues have not been associatedwith “transdifferentiation” into completely un-related fates but rather with contribution to therepair of the tissue from which the cells orig-inated. In this regard, the plasticity seems to arisethrough a process of dedifferentiation and/orredifferentiation.How SCs respond to injuries and repair tissue

wounds varies dramatically depending not onlyon the particular SC niche but also its proximityto the wound. In SwG cells, for example, wherefour different unipotent progenitors exist (53),luminal and myoepithelial progenitors are mo-bilized, but these SCs act unipotently in mediat-ing tissue regeneration, at least under conditionswhere luminal or myoepithelial progenitors areselectively killed (53). Although these findingsillustrate the ability of different SC compartmentsto mobilize in response to different types of in-juries, each SC niche knows its own job and doesnot carry out the job of other resident niches.By monitoring the fate of early IFE prog-

eny during wound repair, signs of transientplasticity begin to surface. Thus, although long-lived IFE SCs are recruited to the wound regionand stably contribute to reepithelialization, short-livedinvolucrin+ IFE progenitors also migrate to woundsites. Within a month, most involucrin+-derivedprogeny terminally differentiate (26), which sug-gests that lineage reversion is not sustained long-term (Fig. 5A). The apparent transient nature of

lineage reversion observed in IFE contrasts withesophagus, where progenitors seem to changetheir mode of proliferation in repairing inci-sional wounds (62). Whether this difference isattributed to the type and/or severity of wound(incisional versus full thickness) or a funda-mental difference in SC behaviors remains to beaddressed.Transient plasticity has also been reported

for adult HF SCs in response to injury. In super-

ficial skin wounds, bulge and infundibulum SCsmigrate upward, proliferate, and participate inthe epidermal repair process (8, 19, 21, 22, 63)(Fig. 5, B and C). Through mechanisms presentlyunknown, migrating HF SCs lose HF markersand adopt an IFE differentiation program. How-ever, unlike neonatal skin, most of these cellsdo not seem to persist long-term within IFE(19, 63, 64). In this regard, they act more likea cellular bandage, which perhaps analogously

Fate of Dll1 progenitor during homeostasis

Fate of Dll1 progenitor after ionizing radiation

A

B

Intestinal regeneration following Lgr5 cell ablation C

Lgr5+ ablation

Dll1 tracing

Dll1 tracing

Days Weeks

x

xx

xx

Ionizing radiation

x

xx

Ionizing radiation

Tissue degeneration

Fig. 6. Plasticity and interconversion into SCs during intestinal regeneration. (A and B) Dll1 lineagetracing showing that, although Dll1+ cells (red) are transient and typically only differentiate into secretorycells (black; interspersed in villus) during homeostasis (A), upon g-radiation–induced cell death (blueX’s), Dll1+ TA cells revert and colonize the crypt (B) (82). (C) When intestine is depleted of Lgr5+ cells(yellow X’s) and then exposed to g-radiation, regeneration is impaired, revealing a critical role for Lgr5+

cells in repairing extensive tissue damage (83).

1242281-6 13 JUNE 2014 • VOL 344 ISSUE 6189 sciencemag.org SCIENCE

RESEARCH | REVIEW

to involucrin+ IFE progenitors (26), are quick torespond but are eventually replaced by IFE SCsand their progeny.Two relatively recent strategies to kill resident

SCs—either laser-ablating them or ablating themthrough DT expression—have proven to be power-ful methods to extricate SCs from their niches andexamine the consequences. Initially shown forDrosophila germ SCs (65, 66), it is now wellestablished that when mammalian epithelialSCs are ablated, the empty niches can recruitand induce normally committed cells to prolif-erate and revert back to a stemlike state.A particularly elegant demonstration of this

paradigm was made by coupling live imagingwith laser-mediated cell lineage ablation ofdifferent HF populations (14, 16). The cellularplasticity within the bulge HF SC niche wasdocumented by illustrating that bulge and HGcells can interconvert when one of these com-partments is emptied (16, 67) (Fig. 5D). It isnoteworthy that cells located in the upper bulgeregion, the so-called “junctional zone” SCs, couldalso replenish the bulge niche after bulge SCablation (16). Although future studies will be nec-essary to more closely examine the long-termcapacity to interconvert into each other’s fate andrestore tissue function after injury, these findingscapture the plasticity displayed by distinct skinepithelial SC compartments after injuries.

The microenvironment controls thefate of epithelial SCs

It has long been observed that when SCs aretaken out of context and transplanted, eitherdirectly or after cell culture, they exhibit greatermultipotency in their new microenvironment.Thus, upon engraftment to immunocompromised

mice, freshly isolated bulge cells (9, 68) or clonalprogeny of single bulge cells (69, 70) each generatenot only HFs, but also IFE and SGs long-term(Fig. 2B). This is also true for isthmus and SG SCs(71, 72). Analogously, when normally unipotentSwG, MG, or prostate basal or myoepithelial SCsare purified and engrafted de novo, they generateentire functional glands (52, 53, 73–76).When unipotent MG myoepithelial cells are

transplanted into mammary mesenchyme ofpregnant mice, they can reform a functional MG(52) (Fig. 4B), which demonstrates the plasticityof unipotent myoepithelial cells during regener-ative conditions. Note that MG myoepithelialcells can also generate MGs when engrafted toshoulder pads, whereas SwG myoepithelial cellsgenerate SwGs in virgin mammary fat pads (53).These findings suggest that for some adult pro-genitors, once identity is established, they takelonger to respond to environmental and systemicprogramming factors. By contrast, when progen-itors form tissue de novo during embryonic de-velopment, they have yet to receive the epigeneticmarks that restrict their fates.Similarly, after culture in vitro, marked thymic

epithelial cells can be mixed with embryonicthymus and transplanted underneath the kid-ney capsule, where they integrate into the thymicnetwork and differentiate into functional thymicepithelial cells (77). However, when the samecultured thymic epithelial cells are transplantedtogether with skin mesenchyme onto back skin,they differentiate into all epidermal lineagesincluding HF and IFE (77). This plasticity in SCbehavior appears to become more permanentwith subsequent transplantations, illustratinghow the microenvironment can instruct thesecells to adopt very different fates.

A hint that adult epithelial cells may be ableto undergo permanent fate conversions in vivocomes from monitoring IFE behavior after mas-sive wounding. In this case, the IFE was reportedto regenerate HFs, which is something it neverdoes during homeostasis (78). It has long beenknown that transgenic b-catenin stabilization, theoutput of a Wnt signal, is sufficient to reprogramK14+ IFE into HFs replete with their own DP (79).Overexpressing the hedgehog pathway also stim-ulates IFE to HF progenitor reprogramming, butin this case, differentiation becomes suppressed atthe expense of hyperproliferation, which leads tobasal cell carcinoma (80, 81).

Reversing fates: Converting committedprogeny to SCs

Although the ability of adult epithelial SCs toacquire different lineage fates seems remarkable,several studies have recently suggested thatcommitted epithelial lineage cells may havethe capacity to acquire stemness. During normalhomeostasis in the intestine, Delta-like 1 (Dll1)–expressing cells (82), or slow-cycling Lgr5+ cells(44), are both short-lived Lgr5-derived progenycommitted to the secretory lineage. However,after g-irradiation–induced tissue damage, thesenormally committed Dll1+ progenitors appear torevert back to ISCs (82) and contribute to in-testinal regeneration (Fig. 6, A and B). Similarly,when Dll1+ progenitors are purified and placedin Wnt3a-supplemented cultures, they formgut organoids containing Lgr5+ SCs and allintestinal lineages (82), which supports theidea that they revert into a stemlike state. HowWnt signaling might influence the reversionprocess in vivo is a yet-unaddressed intriguingquestion. Whether these reserve cells are suffi-

cient to be functionally relevant inthe context of tissue repair is stillunclear, as g-irradiation–inducedintestinal epithelial regenerationdoes not occur after Lgr5 ablation(Fig. 6C) (83).Another example of plasticity

stems from recent lineage tracingof committed secretory cells inthe lung (84), which can revertinto stable and functional basalSCs in vivo if all airway SCs areablated (85) (Fig. 7). In this case, itwas shown that these dediffer-entiated cells can respond to epi-thelial injury and repair injuriesequivalently to their endogenousSC counterparts. By contrast, directcontact with a single basal SC wassufficient to prevent secretory celldedifferentiation, suggestive of neg-ative cross-talk between SCs andcommitted progeny. Overall, thepropensity of committed cells to de-differentiate is typically inverselycorrelated to their state of maturity.The ability of a priori differenti-

ated cells to be reprogrammed andinterconvert into SCs has also been

Trachea during homeostasis and regeneration

Clara cell tracing

Clara cells

Basal cells

Ciliated cells

Basal cell tracing

Clara cell tracing

Basal cell ABLATION

Weeks Months

xx x x

Fig. 7. Plasticity and interconversion into SCs during tracheal regeneration. During tracheal homeostasis,basal cells (green) give rise to TA Clara cells (pink) and terminally differentiated ciliated cells (white). Lineageablation of basal cells (red X’s) induces the interconversion of Clara and/or ciliated cells into basal SCs (85).

SCIENCE sciencemag.org 13 JUNE 2014 • VOL 344 ISSUE 6189 1242281-7

RESEARCH | REVIEW

illustrated for stomach (86). The stomach epi-thelium is composed of an upper part of rapidlyrenewing cells, a middle zone, the isthmus thatactively proliferates, and a bottom zone that con-tains two cell types (parietal and chief cells) withvery low cellular turnover (87). Lineage tracingrevealed that Sox2-expressing cells in the isth-mus region are responsible for the homeostasisof the glandular stomach, giving rise to all sto-mach lineages (88). Through lineage tracingusing Troy-CreER to target differentiated pa-rietal and chief cells (86), it was reported thatprogeny of some Troy cells slowly expandand reach the top of the gland after 6 monthsof chase, which shows that these cells play onlya very minor role during homeostasis. However,the Troy cells can be cultured long-term asmultipotent organoids in vitro and expand several-fold after tissue damage in vivo, suggestive oftheir ability to aid in repair of stomach inju-ries (86).After acute injuries, liver and pancreatic beta-

cell regeneration seems to involve self-duplicationof differentiated cells (89, 90). In contrast, chron-ic and severe hepatic injuries stimulate maturehepatocytes and/or biliary cells to dedifferen-tiate into bipotent progenitor state–expressingSC markers, such as Lgr5, that mediate liverregeneration through their proliferation andredifferentiation (91).Altogether, these remarkable studies point to

the view that, under certain nonhomeostatic con-ditions, differentiated cells dedifferentiate, re-vert back to a SC-like fate, and participate intissue repair. In particular, this seems to happenafter severe injury, a situation where the tissuemust respond quickly and creatively to ensureanimal survival.

Reversibility of lineage differentiationand SC plasticity during tumorigenesis

The plasticity of epithelial lineage commitmentand the ability of committed progeny to revertback to SCs may have important implicationsfor tumorigenesis. In 1990, this notion was ini-tially postulated by Bailleul et al., who observedthat mice expressing an oncogenic Hras drivenby a differentiation-specific promoter developpapillomas after wounding (92). In an inter-esting variation to this theme, normally fate-restricted, unipotent basal and luminal SCs ofglandular epithelia reacquire certain featuresof multipotent SCs during tumor progression.For instance, tumor suppressor inactivation inluminal MG cells can lead to the formation ofbasal-like breast cancer (93), replete with het-erogeneous expression of both basal and lu-minal markers.Similar observations have been made for

prostate cancer, where ablation of a tumorsuppressor gene in luminal SCs induces tumorformation (57). Basal progenitors seem intrin-sically more resistant to tumorigenesis, andeven when they undergo a fate transition intoluminal cells, the tumorigenic lesions that appearare less aggressive than those originating directlyfrom luminal cells (56, 58).

Irrespective of underlying cause or mecha-nism, the plasticity within the tissue hierarchicalorganization is likely to have broader impli-cations for tumor initiation and maintenance.In the intestine, for instance, adenomas arisefrom activating mutations in the Wnt/b-cateninpathway. After a single oncogenic hit, only Lgr5/Bmi1/prominin–expressing ISCs initiate tumorformation (39, 94, 95), whereas targeting TAprogeny have either no effect or induce onlymicroadenomas (94). However, concomitantactivation of the Wnt pathway and another on-cogenic hit cause normally committed TA cellsto revert to a SC-like state and induce tumorformation (96).Once initiated, these tumors may display hi-

erarchical organization, replete with tumor-propagating cells (so-called cancer SCs), definedfunctionally by their ability upon serial trans-plantation to induce secondary tumors that re-semble the parental tumor. Distinct populationsof cells with tumor-propagating capacity capa-ble of interconversion have also been identifiedwithin cancers (96–99), which raises the possi-bility that upon transplantation, more committedcells within a heterogeneous cancer may reac-quire SC properties, analogous to the plasticityobserved in normal SCs after transplantation.Consistent with this notion, non-SCs of humanbasal breast cancers can switch to SC state,depending on ZEB1, a regulator of the epithelial-mesenchymal transition (100). This result suggestsa dynamic model where interconversion betweenlow and high tumorigenic states can occur, whichincreases the potential for cancer progression.Further studies will be required to define theextent to which extent cell plasticity influencescancer growth and relapse after therapy.

Conclusion

The examples provided in this Review havehighlighted the hierarchical and spatial orga-nization of epithelial tissue homeostasis and theimportant plasticity of progenitors and differ-entiated cells during regenerative conditions.This cellular plasticity and lineage reversibil-ity may represent adaptive mechanisms forthe self-preservation of epithelia to repair bodysurfaces and linings in whatever ways possibleafter injuries. Across many different epitheliasubjected to a diverse array of injuries, the par-adigm emerging is that the minimum numberof SCs needed to repair injuries will be activatedand recruited during the healing process. As in-juries become more severe, and greater numbersof SCs are depleted from their niches, more SCsbecome mobilized to participate in wound repair.When all SCs are exhausted, early progeny be-come recruited, until eventually, with massive in-juries, the tissue can no longer cope with repair.Although the molecular mechanisms underlyingcellular plasticity, fate conversion, and reacqui-sition of stem cell properties in committed and/ordifferentiated cells still remain poorly understood,these versatile built-in programs have major impli-cations for regenerative medicine. On the flip sideof this coin, however, is that when gone awry,

these repertoires become the curse of epithe-lial SCs, most of which contribute in major waysto the most life-threatening of human cancers.

REFERENCES AND NOTES

1. C. Blanpain, B. D. Simons, Unravelling stem cell dynamicsby lineage tracing. Nat. Rev. Mol. Cell Biol. 14, 489–502(2013). doi: 10.1038/nrm3625; pmid: 23860235

2. S. J. Morrison, A. C. Spradling, Stem cells and niches:mechanisms that promote stem cell maintenance throughoutlife. Cell 132, 598–611 (2008). doi: 10.1016/j.cell.2008.01.038; pmid: 18295578

3. H. Clevers, The intestinal crypt, a prototype stem cellcompartment. Cell 154, 274–284 (2013). doi: 10.1016/j.cell.2013.07.004; pmid: 23870119

4. J. G. Rheinwald, H. Green, Serial cultivation of strains ofhuman epidermal keratinocytes: the formation of keratinizingcolonies from single cells. Cell 6, 331–343 (1975).doi: 10.1016/S0092-8674(75)80001-8; pmid: 1052771

5. G. G. Gallico 3rd, N. E. O’Connor, C. C. Compton, O. Kehinde,H. Green, Permanent coverage of large burn wounds withautologous cultured human epithelium. N. Engl. J. Med. 311,448–451 (1984). doi: 10.1056/NEJM198408163110706;pmid: 6379456

6. T. Sato et al., Single Lgr5 stem cells build crypt-villusstructures in vitro without a mesenchymal niche. Nature 459,262–265 (2009). doi: 10.1038/nature07935; pmid: 19329995

7. G. Cotsarelis, T. T. Sun, R. M. Lavker, Label-retaining cellsreside in the bulge area of pilosebaceous unit: Implicationsfor follicular stem cells, hair cycle, and skin carcinogenesis.Cell 61, 1329–1337 (1990). doi: 10.1016/0092-8674(90)90696-C; pmid: 2364430

8. T. Tumbar et al., Defining the epithelial stem cell niche inskin. Science 303, 359–363 (2004). doi: 10.1126/science.1092436; pmid: 14671312

9. R. J. Morris et al., Capturing and profiling adult hair folliclestem cells. Nat. Biotechnol. 22, 411–417 (2004).doi: 10.1038/nbt950; pmid: 15024388

10. V. Levy, C. Lindon, B. D. Harfe, B. A. Morgan, Distinct stemcell populations regenerate the follicle and interfollicularepidermis. Dev. Cell 9, 855–861 (2005). doi: 10.1016/j.devcel.2005.11.003; pmid: 16326396

11. Y. V. Zhang, J. Cheong, N. Ciapurin, D. J. McDermitt,T. Tumbar, Distinct self-renewal and differentiation phasesin the niche of infrequently dividing hair follicle stem cells.Cell Stem Cell 5, 267–278 (2009). doi: 10.1016/j.stem.2009.06.004; pmid: 19664980

12. Y. C. Hsu, H. A. Pasolli, E. Fuchs, Dynamics between stemcells, niche, and progeny in the hair follicle. Cell 144,92–105 (2011). doi: 10.1016/j.cell.2010.11.049; pmid: 21215372

13. V. Greco et al., A two-step mechanism for stem cell activationduring hair regeneration. Cell Stem Cell 4, 155–169 (2009).doi: 10.1016/j.stem.2008.12.009; pmid: 19200804

14. P. Rompolas et al., Live imaging of stem cell and progenybehaviour in physiological hair-follicle regeneration. Nature 487,496–499 (2012). doi: 10.1038/nature11218; pmid: 22763436

15. Y. C. Hsu, L. Li, E. Fuchs, Transit-amplifying cells orchestratestem cell activity and tissue regeneration. Cell 157, 935–949(2014). doi: 10.1016/j.cell.2014.02.057; pmid: 24813615

16. P. Rompolas, K. R. Mesa, V. Greco, Spatial organization withina niche as a determinant of stem-cell fate. Nature 502,513–518 (2013). doi: 10.1038/nature12602; pmid: 24097351

17. V. Horsley et al., Blimp1 defines a progenitor population thatgoverns cellular input to the sebaceous gland. Cell 126,597–609 (2006). doi: 10.1016/j.cell.2006.06.048;pmid: 16901790

18. H. J. Snippert et al., Lgr6 marks stem cells in the hair folliclethat generate all cell lineages of the skin. Science 327, 1385–1389(2010). doi: 10.1126/science.1184733; pmid: 20223988

19. M. E. Page, P. Lombard, F. Ng, B. Göttgens, K. B. Jensen, Theepidermis comprises autonomous compartments maintainedby distinct stem cell populations. Cell Stem Cell 13, 471–482(2013). doi: 10.1016/j.stem.2013.07.010; pmid: 23954751

20. I. Brownell, E. Guevara, C. B. Bai, C. A. Loomis, A. L. Joyner,Nerve-derived sonic hedgehog defines a niche for hair folliclestem cells capable of becoming epidermal stem cells.Cell Stem Cell 8, 552–565 (2011). doi: 10.1016/j.stem.2011.02.021; pmid: 21549329

21. M. Ito et al., Stem cells in the hair follicle bulge contribute towound repair but not to homeostasis of the epidermis.Nat. Med. 11, 1351–1354 (2005). doi: 10.1038/nm1328;pmid: 16288281

1242281-8 13 JUNE 2014 • VOL 344 ISSUE 6189 sciencemag.org SCIENCE

RESEARCH | REVIEW

22. V. Jaks et al., Lgr5 marks cycling, yet long-lived, hair folliclestem cells. Nat. Genet. 40, 1291–1299 (2008). doi: 10.1038/ng.239; pmid: 18849992

23. K. K. Youssef et al., Identification of the cell lineage at theorigin of basal cell carcinoma. Nat. Cell Biol. 12, 299–305(2010).pmid: 20154679

24. A. Marianes, A. C. Spradling, Physiological and stem cellcompartmentalization within the Drosophila midgut. eLife 2,e00886 (2013). doi: 10.7554/eLife.00886; pmid: 23991285

25. T. Lechler, E. Fuchs, Asymmetric cell divisions promotestratification and differentiation of mammalian skin. Nature437, 275–280 (2005). doi: 10.1038/nature03922;pmid: 16094321

26. G. Mascré et al., Distinct contribution of stem and progenitorcells to epidermal maintenance. Nature 489, 257–262 (2012).doi: 10.1038/nature11393; pmid: 22940863

27. E. Clayton et al., A single type of progenitor cell maintainsnormal epidermis. Nature 446, 185–189 (2007).doi: 10.1038/nature05574; pmid: 17330052

28. D. P. Doupé, A. M. Klein, B. D. Simons, P. H. Jones, Theordered architecture of murine ear epidermis is maintainedby progenitor cells with random fate. Dev. Cell 18, 317–323(2010). doi: 10.1016/j.devcel.2009.12.016; pmid: 20159601

29. X. Lim et al., Interfollicular epidermal stem cells self-renewvia autocrine Wnt signaling. Science 342, 1226–1230 (2013).doi: 10.1126/science.1239730; pmid: 24311688

30. P. H. Jones, F. M. Watt, Separation of human epidermal stemcells from transit amplifying cells on the basis of differencesin integrin function and expression. Cell 73, 713–724 (1993).doi: 10.1016/0092-8674(93)90251-K; pmid: 8500165

31. A. Rangarajan et al., Notch signaling is a direct determinantof keratinocyte growth arrest and entry into differentiation.EMBO J. 20, 3427–3436 (2001). doi: 10.1093/emboj/20.13.3427; pmid: 11432830

32. C. Blanpain, W. E. Lowry, H. A. Pasolli, E. Fuchs, Canonicalnotch signaling functions as a commitment switch in theepidermal lineage. Genes Dev. 20, 3022–3035 (2006).doi: 10.1101/gad.1477606; pmid: 17079689

33. M. Moriyama et al., Multiple roles of Notch signaling in theregulation of epidermal development. Dev. Cell 14, 594–604(2008). doi: 10.1016/j.devcel.2008.01.017; pmid: 18410734

34. S. E. Williams, S. Beronja, H. A. Pasolli, E. Fuchs, Asymmetriccell divisions promote Notch-dependent epidermaldifferentiation. Nature 470, 353–358 (2011). doi: 10.1038/nature09793; pmid: 21331036

35. H. Cheng, C. P. Leblond, Origin, differentiation and renewal ofthe four main epithelial cell types in the mouse smallintestine. I. Columnar cell. Am. J. Anat. 141, 461–479(1974). doi: 10.1002/aja.1001410403; pmid: 4440632

36. C. S. Potten, W. J. Hume, P. Reid, J. Cairns, The segregationof DNA in epithelial stem cells. Cell 15, 899–906 (1978).doi: 10.1016/0092-8674(78)90274-X; pmid: 728994

37. A. J. Merritt et al., The role of p53 in spontaneous andradiation-induced apoptosis in the gastrointestinal tract ofnormal and p53-deficient mice. Cancer Res. 54, 614–617(1994).pmid: 8306319

38. N. Barker et al., Identification of stem cells in small intestineand colon by marker gene Lgr5. Nature 449, 1003–1007(2007). doi: 10.1038/nature06196; pmid: 17934449

39. E. Sangiorgi, M. R. Capecchi, Bmi1 is expressed in vivo inintestinal stem cells. Nat. Genet. 40, 915–920 (2008).doi: 10.1038/ng.165; pmid: 18536716

40. R. K. Montgomery et al., Mouse telomerase reversetranscriptase (mTert) expression marks slowly cyclingintestinal stem cells. Proc. Natl. Acad. Sci. U.S.A. 108,179–184 (2011). doi: 10.1073/pnas.1013004108; pmid: 21173232

41. N. Takeda et al., Interconversion between intestinal stem cellpopulations in distinct niches. Science 334, 1420–1424(2011). doi: 10.1126/science.1213214; pmid: 22075725

42. L. Ritsma et al., Intestinal crypt homeostasis revealed atsingle-stem-cell level by in vivo live imaging. Nature 507,362–365 (2014). doi: 10.1038/nature12972; pmid: 24531760

43. H. Tian et al., A reserve stem cell population in small intestinerenders Lgr5-positive cells dispensable. Nature 478, 255–259(2011). doi: 10.1038/nature10408; pmid: 21927002

44. S. J. Buczacki et al., Intestinal label-retaining cells aresecretory precursors expressing Lgr5. Nature 495, 65–69(2013). doi: 10.1038/nature11965; pmid: 23446353

45. B. A. Ponder et al., Derivation of mouse intestinal crypts fromsingle progenitor cells. Nature 313, 689–691 (1985).doi: 10.1038/313689a0; pmid: 3974703

46. D. F. Griffiths, S. J. Davies, D. Williams, G. T. Williams,E. D. Williams, Demonstration of somatic mutation and

colonic crypt clonality by X-linked enzyme histochemistry.Nature 333, 461–463 (1988). doi: 10.1038/333461a0;pmid: 3374587

47. D. J. Winton, M. A. Blount, B. A. Ponder, A clonal markerinduced by mutation in mouse intestinal epithelium. Nature 333,463–466 (1988). doi: 10.1038/333463a0; pmid: 3163778

48. C. Lopez-Garcia, A. M. Klein, B. D. Simons, D. J. Winton,Intestinal stem cell replacement follows a pattern of neutraldrift. Science 330, 822–825 (2010). doi: 10.1126/science.1196236; pmid: 20929733

49. H. J. Snippert et al., Intestinal crypt homeostasis results fromneutral competition between symmetrically dividing Lgr5stem cells. Cell 143, 134–144 (2010). doi: 10.1016/j.cell.2010.09.016; pmid: 20887898

50. S. Kozar et al., Continuous clonal labeling reveals smallnumbers of functional stem cells in intestinal crypts andadenomas. Cell Stem Cell 13, 626–633 (2013). doi: 10.1016/j.stem.2013.08.001; pmid: 24035355

51. T. H. Kim et al., Broadly permissive intestinal chromatinunderlies lateral inhibition and cell plasticity. Nature 506,511–515 (2014). doi: 10.1038/nature12903; pmid: 24413398

52. A. Van Keymeulen et al., Distinct stem cells contribute tomammary gland development and maintenance. Nature 479,189–193 (2011). doi: 10.1038/nature10573; pmid: 21983963

53. C. P. Lu et al., Identification of stem cell populations in sweatglands and ducts reveals roles in homeostasis and woundrepair. Cell 150, 136–150 (2012). doi: 10.1016/j.cell.2012.04.045; pmid: 22770217

54. R. van Amerongen, A. N. Bowman, R. Nusse, Developmentalstage and time dictate the fate of Wnt/b-catenin-responsivestem cells in the mammary gland. Cell Stem Cell 11,387–400 (2012). doi: 10.1016/j.stem.2012.05.023;pmid: 22863533

55. A. C. Rios, N. Y. Fu, G. J. Lindeman, J. E. Visvader, In situidentification of bipotent stem cells in the mammary gland.Nature 506, 322–327 (2014). doi: 10.1038/nature12948;pmid: 24463516

56. N. Choi, B. Zhang, L. Zhang, M. Ittmann, L. Xin, Adult murineprostate basal and luminal cells are self-sustained lineagesthat can both serve as targets for prostate cancer initiation.Cancer Cell 21, 253–265 (2012). doi: 10.1016/j.ccr.2012.01.005; pmid: 22340597

57. X. Wang et al., A luminal epithelial stem cell that is a cell oforigin for prostate cancer. Nature 461, 495–500 (2009).doi: 10.1038/nature08361; pmid: 19741607

58. Z. A. Wang et al., Lineage analysis of basal epithelial cellsreveals their unexpected plasticity and supports a cell-of-origin model for prostate cancer heterogeneity. Nat. Cell Biol.15, 274–283 (2013). doi: 10.1038/ncb2697; pmid: 23434823

59. S. Šale, D. Lafkas, S. Artavanis-Tsakonas, Notch2 genetic fatemapping reveals two previously unrecognized mammaryepithelial lineages. Nat. Cell Biol. 15, 451–460 (2013).doi: 10.1038/ncb2725; pmid: 23604318

60. D. Lafkas et al., Notch3 marks clonogenic mammary luminalprogenitor cells in vivo. J. Cell Biol. 203, 47–56 (2013).doi: 10.1083/jcb.201307046; pmid: 24100291

61. M. Ousset et al., Multipotent and unipotent progenitorscontribute to prostate postnatal development. Nat. Cell Biol. 14,1131–1138 (2012). doi: 10.1038/ncb2600; pmid: 23064263

62. D. P. Doupé et al., A single progenitor population switchesbehavior to maintain and repair esophageal epithelium.Science 337, 1091–1093 (2012). doi: 10.1126/science.1218835; pmid: 22821983

63. V. Levy, C. Lindon, Y. Zheng, B. D. Harfe, B. A. Morgan,Epidermal stem cells arise from the hair follicle afterwounding. FASEB J. 21, 1358–1366 (2007). doi: 10.1096/fj.06-6926com; pmid: 17255473

64. J. A. Nowak, L. Polak, H. A. Pasolli, E. Fuchs, Hair follicle stemcells are specified and function in early skin morphogenesis.Cell Stem Cell 3, 33–43 (2008). doi: 10.1016/j.stem.2008.05.009; pmid: 18593557

65. T. Kai, A. Spradling, An empty Drosophila stem cell nichereactivates the proliferation of ectopic cells. Proc. Natl. Acad.Sci. U.S.A. 100, 4633–4638 (2003). doi: 10.1073/pnas.0830856100; pmid: 12676994

66. T. Kai, A. Spradling, Differentiating germ cells can revert intofunctional stem cells in Drosophila melanogaster ovaries.Nature 428, 564–569 (2004). doi: 10.1038/nature02436;pmid: 15024390

67. M. Ito, K. Kizawa, K. Hamada, G. Cotsarelis, Hair follicle stemcells in the lower bulge form the secondary germ, abiochemically distinct but functionally equivalent progenitorcell population, at the termination of catagen. Differentiation

72, 548–557 (2004). doi: 10.1111/j.1432-0436.2004.07209008.x; pmid: 15617565

68. H. Oshima, A. Rochat, C. Kedzia, K. Kobayashi, Y. Barrandon,Morphogenesis and renewal of hair follicles from adultmultipotent stem cells. Cell 104, 233–245 (2001).doi: 10.1016/S0092-8674(01)00208-2; pmid: 11207364

69. C. Blanpain, W. E. Lowry, A. Geoghegan, L. Polak, E. Fuchs,Self-renewal, multipotency, and the existence of two cellpopulations within an epithelial stem cell niche. Cell 118,635–648 (2004). doi: 10.1016/j.cell.2004.08.012;pmid: 15339667

70. S. Claudinot, M. Nicolas, H. Oshima, A. Rochat, Y. Barrandon,Long-term renewal of hair follicles from clonogenicmultipotent stem cells. Proc. Natl. Acad. Sci. U.S.A. 102,14677–14682 (2005). doi: 10.1073/pnas.0507250102;pmid: 16203973

71. U. B. Jensen et al., A distinct population of clonogenic andmultipotent murine follicular keratinocytes residing in theupper isthmus. J. Cell Sci. 121, 609–617 (2008).doi: 10.1242/jcs.025502; pmid: 18252795

72. J. G. Nijhof et al., The cell-surface marker MTS24 identifies anovel population of follicular keratinocytes withcharacteristics of progenitor cells. Development 133,3027–3037 (2006). doi: 10.1242/dev.02443; pmid: 16818453

73. M. Shackleton et al., Generation of a functionalmammary gland from a single stem cell. Nature 439,84–88 (2006). doi: 10.1038/nature04372; pmid: 16397499

74. J. Stingl et al., Purification and unique properties ofmammary epithelial stem cells. Nature 439, 993–997(2006).pmid: 16395311

75. A. S. Goldstein et al., Trop2 identifies a subpopulation ofmurine and human prostate basal cells with stem cellcharacteristics. Proc. Natl. Acad. Sci. U.S.A. 105,20882–20887 (2008). doi: 10.1073/pnas.0811411106;pmid: 19088204

76. L. Xin, R. U. Lukacs, D. A. Lawson, D. Cheng, O. N. Witte,Self-renewal and multilineage differentiation in vitro frommurine prostate stem cells. Stem Cells 25, 2760–2769(2007). doi: 10.1634/stemcells.2007-0355; pmid: 17641240

77. P. Bonfanti et al., Microenvironmental reprogramming ofthymic epithelial cells to skin multipotent stem cells. Nature466, 978–982 (2010). doi: 10.1038/nature09269;pmid: 20725041

78. M. Ito et al., Wnt-dependent de novo hair follicle regenerationin adult mouse skin after wounding. Nature 447, 316–320(2007). doi: 10.1038/nature05766; pmid: 17507982

79. U. Gat, R. DasGupta, L. Degenstein, E. Fuchs, De novo hairfollicle morphogenesis and hair tumors in mice expressing atruncated beta-catenin in skin. Cell 95, 605–614(1998). doi: 10.1016/S0092-8674(00)81631-1;pmid: 9845363

80. S. H. Yang et al., Pathological responses to oncogenicHedgehog signaling in skin are dependent on canonicalWnt/beta3-catenin signaling. Nat. Genet. 40, 1130–1135(2008). doi: 10.1038/ng.192; pmid: 19165927

81. K. K. Youssef et al., Adult interfollicular tumour-initiating cellsare reprogrammed into an embryonic hair follicle progenitor-like fate during basal cell carcinoma initiation. Nat. Cell Biol. 14,1282–1294 (2012). doi: 10.1038/ncb2628; pmid: 23178882

82. J. H. van Es et al., Dll1+ secretory progenitor cells revert tostem cells upon crypt damage. Nat. Cell Biol. 14, 1099–1104(2012). doi: 10.1038/ncb2581; pmid: 23000963

83. C. Metcalfe, N. M. Kljavin, R. Ybarra, F. J. de Sauvage,Lgr5+ stem cells are indispensable for radiation-inducedintestinal regeneration. Cell Stem Cell 14, 149–159 (2014).doi: 10.1016/j.stem.2013.11.008; pmid: 24332836

84. E. L. Rawlins et al., The role of Scgb1a1+ Clara cells in thelong-term maintenance and repair of lung airway, but notalveolar, epithelium. Cell Stem Cell 4, 525–534 (2009).doi: 10.1016/j.stem.2009.04.002; pmid: 19497281

85. P. R. Tata et al., Dedifferentiation of committed epithelialcells into stem cells in vivo. Nature 503, 218–223 (2013).pmid: 24196716

86. D. E. Stange et al., Differentiated Troy+ chief cells act asreserve stem cells to generate all lineages of the stomachepithelium. Cell 155, 357–368 (2013). doi: 10.1016/j.cell.2013.09.008; pmid: 24120136

87. N. Barker, S. Bartfeld, H. Clevers, Tissue-resident adult stemcell populations of rapidly self-renewing organs. Cell StemCell 7, 656–670 (2010). doi: 10.1016/j.stem.2010.11.016;pmid: 21112561

88. K. Arnold et al., Sox2+ adult stem and progenitor cells areimportant for tissue regeneration and survival of mice. Cell

SCIENCE sciencemag.org 13 JUNE 2014 • VOL 344 ISSUE 6189 1242281-9

RESEARCH | REVIEW

Stem Cell 9, 317–329 (2011). doi: 10.1016/j.stem.2011.09.001; pmid: 21982232

89. A. W. Duncan, C. Dorrell, M. Grompe, Stem cells and liverregeneration. Gastroenterology 137, 466–481 (2009).doi: 10.1053/j.gastro.2009.05.044; pmid: 19470389

90. Y. Dor, J. Brown, O. I. Martinez, D. A. Melton, Adult pancreaticbeta-cells are formed by self-duplication rather thanstem-cell differentiation. Nature 429, 41–46 (2004).doi: 10.1038/nature02520; pmid: 15129273

91. M. Huch et al., In vitro expansion of single Lgr5+ liver stemcells induced by Wnt-driven regeneration. Nature 494,247–250 (2013). doi: 10.1038/nature11826; pmid: 23354049

92. B. Bailleul et al., Skin hyperkeratosis and papilloma formationin transgenic mice expressing a ras oncogene from asuprabasal keratin promoter. Cell 62, 697–708 (1990).doi: 10.1016/0092-8674(90)90115-U; pmid: 1696852

93. G. Molyneux et al., BRCA1 basal-like breast cancers originatefrom luminal epithelial progenitors and not from basal stemcells. Cell Stem Cell 7, 403–417 (2010). doi: 10.1016/j.stem.2010.07.010; pmid: 20804975

94. N. Barker et al., Crypt stem cells as the cells-of-origin ofintestinal cancer. Nature 457, 608–611 (2009). doi: 10.1038/nature07602; pmid: 19092804

95. L. Zhu et al., Prominin 1 marks intestinal stem cells that aresusceptible to neoplastic transformation. Nature 457,603–607 (2009). doi: 10.1038/nature07589; pmid: 19092805

96. S. Schwitalla et al., Intestinal tumorigenesis initiated bydedifferentiation and acquisition of stem-cell-like properties. Cell152, 25–38 (2013). doi: 10.1016/j.cell.2012.12.012; pmid: 23273993

97. E. Quintana et al., Phenotypic heterogeneity amongtumorigenic melanoma cells from patients that is reversibleand not hierarchically organized. Cancer Cell 18, 510–523(2010). doi: 10.1016/j.ccr.2010.10.012; pmid: 21075313

98. M. Schober, E. Fuchs, Tumor-initiating stem cells ofsquamous cell carcinomas and their control by TGF-b andintegrin/focal adhesion kinase (FAK) signaling. Proc. Natl.Acad. Sci. U.S.A. 108, 10544–10549 (2011). doi: 10.1073/pnas.1107807108; pmid: 21670270

99. G. Lapouge et al., Skin squamous cell carcinoma propagatingcells increase with tumour progression and invasiveness.

EMBO J. 31, 4563–4575 (2012). doi: 10.1038/emboj.2012.312; pmid: 23188079

100. C. L. Chaffer et al., Poised chromatin at the ZEB1 promoterenables breast cancer cell plasticity and enhancestumorigenicity. Cell 154, 61–74 (2013). doi: 10.1016/j.cell.2013.06.005; pmid: 23827675

ACKNOWLEDGMENTS

We thank our colleagues in epithelial stem cell biology for theexcitement and major advances they’ve contributed. We apologizefor works not cited because of space constraints. E.F. is a HHMIinvestigator and supported by grants from the NIH, Empire StateStem Cell Board (NYSTEM), and Ellison Foundation. C.B. is aWELBIO investigator and is supported by the Fond National de laRecherche Scientifique (FNRS), the European Research Council(ERC), the ULB Foundation, the Fondation contre le Cancer, andthe Foundation Bettencourt Schueller.

10.1126/science.1242281

1242281-10 13 JUNE 2014 • VOL 344 ISSUE 6189 sciencemag.org SCIENCE

RESEARCH | REVIEW