Adult intestinal stem cells: critical drivers of epithelial homeostasis and regeneration

Small populations of adult stem cells are responsible for the remarkable ability of the epithelial lining of the intestine to be efficiently renewed and repaired throughout life. The recent discovery of specific markers for these stem cells, together with the development of new technologies to track endogenous stem cell activity in vivo and to exploit their ability to generate new epithelia ex vivo, has greatly improved our understanding of stem cell-driven homeostasis, regeneration and cancer in the intestine. These exciting new insights into the biology of intestinal stem cells have the potential to accelerate the development of stem cell-based therapies and ameliorate cancer treatments.

The inner epithelial lining of the small intestine is a truly multitasking tissue. It must simultaneously achieve the efficient digestion (with the aid of enzymes released from the liver and pancreas) and the absorption of food contents released from the stomach while maintaining an effective barrier against potentially lethal micro organisms and carcinogens that are present in the intestinal lumen. Persistent aggression from the luminal content s induces a remarkably high rate of cell death, with up to 10 11 epithelial cells (~200 g) being lost every day in humans 1 . This imposes a requirement for daily self-renewal throughout life, which is driven by small populations of adult stem cells that reside within specialized niches. The ability of these stem cells to indefinitely self-renew while generating new functional epithelia makes them ideally suited for regenerative medicine applications. The same properties also make stem cells prime candidates for accumulating mutations that promote cancer growth -making them attractive therapeutic targets.
Over decades of intensive study, researchers have successfully deciphered many key aspects of intestinal biology, including the approximate size and location of the adult stem cell population within the epithelium. However, the identity of the stem cell population has long remained elusive owing to a lack of specific markers and suitable methodologies for rigorous evaluation of endogenous stem cell function 2,3 . The in ability to identify and isolate adult stem cells has been a major limitation to the study of their behaviour during homeostasis and disease. However, recent years have seen the discovery of robust adult intestinal stem cell markers, the development of clonal fate-mapping technologies (to evaluate stem cell behaviour in vivo), and the advent of near-physiological ex vivo culture systems (which support the long-term growth of functional epithelia from patient biopsies for use in diagnostic and/ or therapeutic applications) [4][5][6] . With these new tools, researchers have made impressive breakthroughs in understanding how intestinal stem cells interact with their local niche to maintain homeostasis in healthy intestinal epithelia, and how they might contribute to intestinal cancer initiatio n and progression.
Another crucial finding is the identification of 'reserve' stem cells that can be rapidly recruited to maintain epithelial homeostasis following injury 7,8 . This suggests that epithelial cells that are separate from the small pool of regular stem cells also harbour regenerative potential. Moreover, a nutrient-sensing mechanism within the stem cell niche (BOX 1) that regulates the size and activity of the intestinal stem cell pool was discovered 9 , and these findings could be exploited for therapeut ic applications.
This Review discusses the current models of stem cell-driven epithelial homeostasis and repair in the intestine, highlighting the contributions of pioneerin g research carried out in the 1970s and 1980s and of more recent breakthroughs driven by the identification of robust stem cell markers and the development of near-physiological epithelial culture systems.

Intestinal villus
A finger-like structure covered in simple columnar epithelium that projects into the intestinal lumen to maximize the surface area for digestion and absorption.

Duodenum
The proximal third of the small intestine, closest to the stomach. Characterized by the presence of long villi to ensure maximal nutrient digestion and absorption.

Intestinal crypt
Tubular invaginations of the epithelium harbouring the stem cells and their proliferating progeny; responsible for driving epithelial homeostasis and regeneration.

The adult intestinal epithelium
Three weeks after birth, gut development is complete and the mouse is ready to tackle its first solid meal. The organization of intestinal tissue is highly adapted to its function and to maintaining tissue integrity.

Cellular organization.
In the small intestine, each intestina l villus is encircled by at least six crypts of Lieberkühn, which house dedicated populations of stem and progenitor cells that self-renew to maintain epithelial function throughout life (FIG. 1a).
The organization of the epithelium is tailored to meet the specific functional requirements of the different regions of the intestinal tract. In the duodenu m, the villi are longest and the epithelium contains mainly absorptive enterocytes that also secrete hydrolytic enzymes to facilitate the efficient breakdown of partly digested food exiting the stomach and the absorption of nutrients. Also present are goblet cells, which secrete mucus, together with much smaller populations of enteroendocrine cells, which produce hormones (FIG. 1a). Goblet cells are most prevalent in the epithelium of the short villi of the distal small intestine, where they provide the extra lubrication needed to facilitate the passage of increasingly compact stool towards the colon. Intestinal crypt-resident Paneth cells, which secrete protective antimicrobial substances (such as cryptdins) and the hydrolytic enzyme lysozyme, are present throughout the small intestine. The epithelium contains at least three other cell types with poorly defined functions: cup cells, tuft cells and Peyer's patch-associated M (or microfol d) cells 10 .
In the colon, which specializes in compacting stool for rapid excretion, the inner epithelial lining is arranged into multiple crypts associated with a flat luminal surface. In contrast to that of the small intestine, this epithelium is characterized by a high density of goblet cells and the absence of Paneth cells (FIG. 1b).
Rapid epithelial turnover. In the mouse, the adult epithelium undergoes rapid renewal to maintain optimal function. Epithelial cells exposed to the harsh luminal environment die and are expelled from the tips of the villi through a complex cytoskeletal remodelling process that simultaneously seals the resulting gap to maintain a barrier function 11 . More than 300 million new epithelial cells must be generated daily in the small intestine alone to compensate for this high rate of cell death on the villi. Regeneration relies on the crypt base, where small populations of adult stem cells regularly divide to produce highly proliferative progenitors known as transitamplifying (TA) cells (FIG. 1). The nascent TA cells divide 2-3 times and gradually commit to the absorptive or secretory cell lineages while migrating upwards towards the base of the villi. Cell proliferation ceases when differentiated cells exit the crypt, and these epithelial cells then continue migrating upwards along the villi. This epithelial cell renewal cycle, which takes 3-5 days, is repeated several hundred times during the average 2-year lifespan of an inbred mouse.
Paneth cell turnover is the only exception to this rapid self-renewal. These cells are renewed every 3-6 weeks from dedicated secretory cell progenitors located at the base of the TA compartment, which mature into fully differentiated Paneth cells while following a downward migratory path to the crypt bottom 12,13 .
Origin of the epithelium and its stem cells Embryonic origin of the intestinal epithelium. In the mouse, the epithelium of the small intestine originates as a polarized, single-layered (pseudostratified) structure lining the inner surface of the primitive gut around embryonic day (E) 9.5 (REF. 14). At E14.5, following thickening of the epithelium, a major remodelling process initiates, resulting in the rapid conversion of the flat luminal surface into multiple villi. The regenerative capacity of the adult epithelium is established between E16.5 and postnatal day 7, when mature crypts harbouring adult stem cells and progenitor cells develop from shallow pockets of proliferative cells that are restricted to the base of the embryonic villi 15 .
Although the mechanics of these remodelling processes remain incompletely understood, bidirectional signal s between the epithelium and the underlying mesenchyme are considered to be crucial coordinators. In particular, Hedgehog signals from the developing epithelium are thought to direct embryonic villus formation and specify the site of crypt development at the villus base 16 . Moreover, WNT signalling has an essential role in establishing the regenerative capacity of the adult epithelium, as evidenced by crypt loss and impaired villu s formation following embryonic ablation of the major intestinal WNT effector protein, transcription factor 4 (TCF4) 17 .
Crypt numbers rapidly increase during postnatal development of the small intestine through a bifurcation mechanism known as crypt fission 18 . This poorly understood process may be triggered by a rapid increase in crypt size owing to increased stem and/or progenitor cell activity, which peaks approximatel y 2 weeks after birth in mice 18,19 .

Box 1 | A glance at the intestinal stem cell niche
Stem cell activity at the crypt base is strictly regulated during regular homeostasis by a complex array of signals delivered by neighbouring epithelial and stromal cells (FIG. 2b). Paneth cells, which are found in close association with the LGR5 + (Leu-rich repeatcontaining G protein-coupled receptor 5-expressing) crypt base columnar (CBC) stem cells at the crypt base, are an important source of various niche factors, including epidermal growth factor (EGF), WNT3A and Notch ligand 86 . These niche cells also enable the intestine to tailor the output of its stem cell compartment to nutrient availability 9 . Paneth cells respond to calorie restriction by reducing mTOR complex 1 (mTORC1) signalling, initiating a signal cascade that results in a rapid reduction in the size of the LGR5 + stem cell pool.
However, Paneth cells are unlikely to be the sole niche component in vivo. Indeed, neither Paneth cell ablation [90][91][92] nor epithelial-specific loss of WNT3A expression 93 is sufficient to block stem cell-driven epithelial renewal in vivo, suggesting the existence of redundant, non-epithelial sources of WNT and other key niche signals.
The colonic stem cell niche is less well defined. Although Paneth cells are generally absent from the majority of the colon, CD24 + and KIT + goblet cells that are located in close proximity to LGR5 + stem cells at the crypt base have been identified as probable niche components 94 . However, the major WNT source in the colon has yet to be identified.  [20][21][22] . Analysis of genetic marker expression patterns in newborn chimeric mice has shown that nascent crypts are polyclonal, as they develop from fetal stem cells derived from both parents 23,24 . During crypt morphogenesis in the first 2 weeks of life, these early stem cell populations initially undergo rapid expansion via symmetric cell division and then switch to an asymmetric cell division mode, which establishes the limited adult stem cell pool needed to sustain a functional epithelium 21 . During this period, a single progenitor cell gradually achieves dominance through a stochastic process known as purification, giving rise to a clonal pool of adult stem cells in each crypt that is maintained throughout life [25][26][27] . Additionally, analyses of random mitochondrial mutation patterns within healthy human small intestines and of Y-chromosome inheritance patterns in the colon of a XO/XY chimeric patient have established that adult crypts throughout the intestinal tract are clonal [28][29][30][31] . Despite this phenotypic clonality, it is important to note that adult crypts are not maintained by a single stem cell. Indeed, multiple stem cells sustain long-term epithelial homeostasis, although the exact number of them is not known (studies estimate between three and 16, depending on the methodology used) [32][33][34] . Perhaps the most precise calculation of crypt stem cell numbers was based on a continuous clonal labelling approach, which led to an estimate of five to seven stem cells driving daily epithelial renewal in the mouse small intestine 35 . Formal proof of the existence of multipotent, selfrenewing stem cells in the crypt of the adult intestine was obtained by mapping the inheritance of genetic markers introduced at low frequency in crypt cells via random somatic mutation 27,36 , as this led to the appearance of rare, long-lived clones containing all major epithelial lineages. However, these fate-mapping approaches were unable to reveal the location and identity of the stem cells within the crypt. Clues to the location of these stem cells were first obtained in the 1970s by radioisotope and 5-bromodeoxyuridine (BrdU) labelling of dividing intestinal cells [37][38][39] . The epithelial cell migration patterns indicated a common origin close to the crypt base, hinting at this being the location for multipotent stem cells 1 . In support of this, ablation of the crypt base populations through targeted irradiation completely abolished epithelial self-renewal in the intestine, which is consistent with loss of the stem cell compartment 40 . Despite these advances, the identity of adult stem cells remained frustratingly elusive for more than two decades and is today still the subject of intense debate.

Models of adult stem cell identity
Two models of intestinal stem cell identity have historically competed over the past four decades. The 'stem cell zone model' , by Leblond, Cheng and Bjerknes 27 , suggests that the columnar cells at the base of the crypt are the resident stem cells, whereas the '+4 model' , by Potten 48 , proposes that stem cells reside within a ring of 16 cells immediately above the Paneth cells. Despite the recent discovery of specific markers for these two candidate stem cell populations, it has proved difficult to definitively determine which of these models is correct. Instead, a unifying theory that incorporates aspects of both models is emerging, based on the existence of distinct stem cell pools involved in epithelial homeostasi s and regeneration.
The stem cell zone model. The identification of undifferentiated, mitotically active crypt base columnar (CBC) cells intercalated with Paneth cells at the bottom of crypts lead to the formulation of the 'Unitarian theory of the origin of the four epithelial cell types' . This model, outlined in publications by , proposed that the CBC cells gave rise to the four major cell lineages in the intestinal epithelium [41][42][43][44][45] . CBC cells were described as being actively phagocytic, helping to clear dead cells from the crypt base. Taking advantage of this trait, Cheng irradiated mice and showed that surviving CBC cells contained radiolabelled phagosomes as a consequence of engulfing neighbouring CBC cells that were killed via the incorporation of tritiated thymidine. Initially, only CBC cells were marked, but at later time points the cells from three major epithelial lineages became radiolabelled within the crypts. This was interpreted as CBC cells being the common ancestor of the differentiated epithelial cell types of the intestinal lining. However, because the three marked epithelial lineages were located in different crypts and no marked enteroendocrine cells were identified, formal identification of the CBC cells as multipotent stem cells was not possible.
Additional evidence in support of CBC stem cell identity was provided more than two decades later, when Bjerknes and Cheng used a mutation approach to introduce heritable somatic marks within the crypts 27 . A small proportion of the resulting epithelial clones comprised cells of all major lineages that were maintained over a long period, indicating that the original mutation had occurred within an unidentified self-renewing stem cell. Only these persistent, multilineage clones invariably contained a CBC cell, which was interpreted as further evidence of a CBC stem cell identity.
This prompted Cheng and Bjerknes to refine the original Unitarian theory of intestinal cell origins and publish the stem cell zone model in 1999 (REF. 27). In this model, CBC cells are adult stem cells that reside in a specialised niche (which was named the 'stem cell permissive zone') at the very base of the crypt. These cells proliferate to generate daughter cells that exit the niche and commit to multilineage differentiation at the 'common origin of differentiation' around position +5 (that is, the fifth cell position from the crypt base). Paneth cell progenitors mature into functional lysozyme-secreting cells as they migrate downwards to occupy the base of the crypt, whereas the majority of the non-Paneth cell lineages mature into functional epithelia while migrating upwards onto the villus. However, it is important to note that the crypt base is not exclusively populated by Paneth cells and CBC cells -rare enteroendocrine cells and goblet cells are also present, but they are post-mitotic and are therefore thought to be unresponsive to stem cell niche signals. Despite a substantial body of indirect evidence in support of the stem cell zone model, it failed to gain a general acceptance until 2007, when CBC cell-specific markers were first identified 46 . These markers enabled in vivo lineage tracing experiments 46 and ex vivo assays 47 to provide more direct evidence of CBC stem cell function during epithelial homeostasis and disease (see below and see Supplementary information S1 (Box)).

Lgr5-EGFP-ires-CreERT2/ R26R-lacZ mouse model
Generated by crossing Lgr5-EGFP-ires-CreERT2 (which encodes a tamoxifen-activatable Cre enzyme that catalyzes recombination across DNA sequences called loxP sites) and R26R (Rosa26 reporter construct)-lacZ mouse strains. Facilitates the conditional activation of the lacZ reporter gene in LGR5-expressing cells in living tissues to evaluate their stem cell identity via lineage tracing.
The +4 model. The first indication that intestinal epithelial cells may occupy position +4 (the fourth position from the crypt base), immediately above the Paneth cell compartment, came from cell-tracking experiments evaluating the migration rates of radiolabelled cells at various positions within the crypt 37,38 . Potten and colleagues later documented the existence of cells with stem cell attributes at a similar location 48 . These cells were actively dividing (every 24 hours), but they were capable of retaining labels incorporated into their DNA during crypt neogenesis (when new stem cells are being generated). Although this label retention is generally considered to be a trait characteristic of cellular quiescence, the +4 label-retaining cells (LRCs) could be labelled subsequently with BrdU, confirming their active proliferation status 3 . These conflicting observations were attributed to the ability of the +4 cells to retain labelled template DNA strands selectively during mitosis while segregating newly synthesized DNA strands, which could contain potentially dangerous replication errors, to the short-lived TA daughter cells during asymmetric division 3 . Originally described in the 'immortal strand hypothesis' (REF. 39), this phenomenon was proposed to limit the accumulation of DNA damage in long-lived adult stem cells. However, this model has been challenged, as it makes several unfounded assumptions about +4 cells, including the presence of asymmetric cell division and DNA exchange between sister chromatids (that is, between template DNA and newly-synthesized DNA, as typically occurs in somatic cells) 49 .
Moreover, cells that were highly sensitive to ionizing radiation, a surrogate stem cell trait believed to ensure that long-lived cells do not survive DNA damage that could eventually lead to cancer, were identified around the +4 position 48 . However, formal proof that label retention and radiation sensitivity can be attributed to the same +4 cell population is currently lacking. Studies to validate and refine the +4 stem cell model through the identification of specific marker genes that facilitate a direct evaluation of endogenous stem cell identity have been reported. However, their observations are difficult to reconcile with the original +4 model (as discussed later). In vivo lineage tracing using the Lgr5 marker. Lineage tracing is a cell fate-mapping tool used for evaluating the stem cell identity of candidate populations in their native microenvironments 4 . It involves introducing permanent, heritable genetic marks into candidate stem cells. As the descendants of these cells inherit the marks, this enables their characterization within the epithelium. If all differentiated cell lineages can be traced back to having derived from a single cell, this cell is considered to be a multipotent stem cell. Long-term generation of marked cell lineages within the epithelium indicates that the candidate stem cell has self-renewal capacity. Any candidate cell demonstrating both multipotency and self-renewal capacity fulfils the minimal definition of an epithelial stem cell.

Validation of CBC cells as stem cells
The first marker for CBC cells, Lgr5 (Leu-rich repeatcontaining G protein-coupled receptor 5), was identifie d as a WNT target gene selectively expressed at the base of adult intestinal crypts 46 (FIG. 2a). CBC-specific expression of LGR5 was confirmed using Lgr5-LacZ and Lgr5-EGFP reporter mouse models. In these mice, approximately 14 LGR5-expressing (LGR5 + ) CBC cells were found at positions 1-4 in the crypt base, evenly distributed between Paneth cells.
The stem cell identity of LGR5 + CBC cells was confirmed via lineage tracing using an Lgr5-EGFP-ires-Cre-ERT2/R26R-lacZ mouse model. Stochastic activation of the lacZ reporter gene at low frequency in LGR5 + CBC cells

Box 2 | Additional CBC stem cell markers
The WNT target gene Sox9 encodes a member of the sex-determining region Y (SRY) family of transcription factors, which regulate cell proliferation in the intestine 95,96 . Using transgenic mice expressing enhanced GFP (EGFP) under the control of the Sox9 promoter (Sox9-EGFP), two distinct cell populations, expressing EGFP at low (Sox9-EGFP low ) and high (Sox9-EGFP hi ) levels, were identified at the crypt base 97 . EGFP hi cells were enteroendocrine cells and candidate +4 stem cells, whereas EGFP low cells overlapped substantially with the LGR5 + (Leu-rich repeat-containing G protein-coupled receptor 5-expressing) crypt base columnar (CBC) cell compartment 97,98 . Sox9-ires-EGFP and Sox9-ires-lacZ knock-in mice showed a similar Sox9 expression profile at the crypt base but also reported endogenous Sox9 expression in Paneth cells and, less frequently, in villus cells 99 . SOX9 low CBC cells were confirmed to have stem cell identity by their ability to generate multilineage intestinal 'organoids' in an ex vivo culture assay 98 . By contrast, SOX9 hi cells were only able to proliferate and generate intestinal organoids ex vivo following irradiation, prompting speculation that this population includes damage-responsive 'reserve' +4 stem cells. The most compelling evidence in support of Sox9-expressing cells being stem cells was provided by in vivo lineage tracing using a Sox9-ires-CreERT/R26R-lacZ model, when typical stem cell-driven tracing clones were generated and maintained throughout the small intestine and colon 99 . Although the expression of Sox9 in various differentiated epithelial lineages reduces its value as a selective marker of endogenous intestinal stem cells, expression profiling of the SOX9 low population did reveal CD24 as a cell-surface marker that can be used to enrich for CBC stem cells through cell cell-sorting approaches 100 .
Musashi homologue 1 (Msi1) encodes an RNA-binding protein implicated in neural stem cell homeostasis. It was originally proposed as a CBC cell marker on the basis of antibody staining patterns, which labelled the crypt base in the small intestine and colon 101,102 . However, both single-molecule mRNA fluorescence in situ hybridization (FISH) analyses and in vivo reporter gene analyses using an Msi1-EGFP transgenic line indicate that Msi1 expression may extend to the lower transit-amplifying compartment 51,78,103 . Definitive proof of the stem cell identity of MSI1 + cells could be obtained with in vivo lineage tracing experiments, which should now be feasible using a recently described Msi1-CreERT2-KI line 104 .
Prominin 1 (Prom1; also known as CD133 in humans), which encodes a cell-surface glycoprotein, is a gene that has been controversially used as a marker of cancer stem cells in a variety of epithelial cancers 105,106 . It has also been reported as a stem cell marker of in healthy adult tissues, including the bone marrow and intestine 107,108 . In the intestine, in situ hybridization analysis revealed PROM1 expression at the crypt base, overlapping with the CBC stem cells. In vivo lineage tracing using a Prom1-CreERT2ires-nLacZ/R26R-yellow fluorescent protein model identified at least some of these PROM1 + cells as being bona fide adult stem cells. However, a subsequent study using independent antibody staining and in vivo lineage tracing analyses found that Prom1 is expressed throughout the proliferative zone of the crypt (including the CBC cell compartment) and that the majority of marked cells originated from short-lived transien t-amplifying cells and thus were rapidly lost 109 .
generate self-renewing 'organoids' that recapitulated the architecture and cellular composition of a functional intestinal epithelium 47 . Of note, cells not expressing LGR5 (LGR5 − cells) lacked the ability to generate these epithelial organoids, highlighting the value of this cultur e system to assay endogenous 'stemness' . The base of human intestinal crypts contains CBClike cells, but the lack of antibodies against LGR5 has precluded a formal evaluation of their expression profile and their stem cell identity. However, epithelial cells from the very bottom of the crypt in the human colon have been isolated using antibodies against the surface-expressed WNT target gene ephrin type B receptor 2 (EPHB2). Cells expressing high levels of this gene (EPHB2 hi cells) behaved as stem cells in ex vivo organoid culture assays and expressed high levels of LGR5, suggesting a conservation of intestinal stem cell identity in mice and humans 50 .

The LGR5 + CBC stem cell expression signature.
Combinatorial microarray and proteomic approaches have established an accurate molecular signature for the LGR5 + CBC stem cells 51,52 . A comparison between stem cells and their progeny isolated by FACS from Lgr5-EGFP reporter mice (exploiting different levels of EGFP: high in stem cells and low in progeny) revealed approximately 500 genes preferentially expressed in stem cells. Consistent with the known function of WNT signalling on stem cell-driven epithelial homeostasis in the intestine 17 , a strong WNT signature was present, including many WNT target genes, such as Sox9, Achaete-Scute homologue 2 (Ascl2), EphB2, Troy (also known as Tnfrsf19) and axis inhibition 2 (Axin2). This approach also revealed novel markers of the LGR5 + CBC stem cells, including olfactomedin 4 (Olfm4), SPARC-related modular calcium-binding 2 (Smoc2) and ring finger 43 (Rnf43) (FIG. 2). Selective expression of the bone morphogenetic protein inhibitor Smoc2 on the stem cell compartment was validated through in vivo lineage tracing 51 . Surprisingly, many +4 stem cell markers (including Bmi1, Leu-rich repeats and immunoglobulin-like domains 1 (Lrig1), telo merase reverse transcriptase (Tert) and homeodomain-only (Hopx)) were strongly expressed in LGR5 + stem cells, casting doubt onto whether these are truly independent intestina l stem cell pools (as discussed below).

Regulation of CBC stem cell activity and fate.
A closer look at the LGR5 + stem cell expression signature has provided new mechanistic insight into the regulation of intestinal stem cell activity and fate in vivo (FIG. 2). ASCL2, a transcription factor that promotes neuroblast differentiation, was identified as a critical regulator of intestinal stem cell fate, as its conditional ablation in the intestinal epithelium resulted in a rapid and selective loss of the LGR5 + stem cell compartment 53 . Lgr5, which encodes a receptor with seven transmembrane segments, was itself linked to stem cell homeostasis, as conditional ablation of Lgr5 and its close homologue Lgr4 resulted in the suppression of WNT signalling and stem cell death 54 . The direct role for LGR5 in modulating WNT signalling in intestinal stem cells was confirmed when LGR5 was identified as a facultative component of the WNT signalling complex at the plasma membrane.
LGR5 recruits secreted WNT agonists (roof plate-specific spondin (R-spondin) 1-4), thus amplifying canonical WNT signalling and ensuring stem cell homeostasis in vivo [54][55][56][57] . Two other WNT target genes that are highly expressed in stem cells, Rnf43 and Troy, also regulate endogenous WNT signalling. RNF43 suppresses WNT signalling in stem cells by ubiquitylating Frizzled receptor proteins to promote their degradation 58,59 , whereas TROY exerts an inhibitory function by destabilizing the low-density lipoprotein receptorrelated protein 6 (LRP6) co-receptor protein (after it is recruited to the WNT receptor complex by LGR5) 60 . These observations highlight the importance of maintaining an optimal level of WNT signalling for intestinal stem cell homeostasis in vivo (see Supplementary information S1 (Box)).The recent elucidation of the structure of the WNT receptor complex containing LGR5, R-spondin and other components should increase our understanding of the mechanism of action of LGR5 in intestinal stem cells [61][62][63] .

A model of CBC stem cell-driven homeostasis
Insights into how actively cycling LGR5 + CBC cells balance stem cell maintenance and differentiation to maintain epithelial homeostasis in vivo were obtained using a multicolour lineage tracing approach 64 . These clonal fate-mapping experiments unexpectedly revealed that individual LGR5 + crypt cells typically undergo symmetrical division to generate progeny with identical fates (either two stem cells or two TA cells), rather than undergoing asymmetric division like most somatic stem cells. Although this division mode suggests a regular loss of individual LGR5 + stem cells, a balanced supply of new stem cells and TA cells is achieved at the population level. Over time, the multicoloured LGR5 + stem cell populations gradually became monochromatic, through a process of stochastic refinement known as neutral drift (FIG. 2c). These observations indicated that homeostasis of the epithelial and stem cell compartments is achieved by neutral competition among 14 symmetrically dividing stem cells for restricted niche space at the crypt base. A separate mathematical modelling approach provided additional support for this model 65 . By contrast, analysis of the relationship between mitotic spindle orientation and DNA segregation during LGR5 + stem cell division indicated that asymmetric division predominates, with template strands being preferentially distributed to progeny adopting a stem cell fate during mitosis 66 . However, these results were challenged in later studies analysing spindle orientation in crypt base cells 67 and DNA label distribution kinetics in dividing LGR5 + CBC stem cells, which suggested that chromosomes randomly segregate to daughter cells 68,69 . More recently, highly sensitive multi-isotope imaging mass spectrometry analyses failed to identify any evidence of label retention during crypt cell division, confirming a model of random DNA strand segregation, which is inconsistent with the immortal strand hypothesis 70 .

Ires
(Internal ribosome entry site). A ribosome-binding site present in the middle of an mRNA that facilitates internal translation initiation to generate an independent protein.
Although symmetrical division seems to predominate in the LGR5 + CBC stem cell compartment, it is currently unknown whether the choice of adopting stem cell or TA cell fates is truly stochastic. It is plausible that local niche and biomechanical factors could direct post-mitotic fate choices of crypt LGR5 + stem cells in a position-dependent manner. In theory, this could be investigated by fate-mapping LGR5 + stem cells at different crypt locations through multicolour lineage tracing, but this is technically challenging. Alternatively, robust, predictive computer models of crypt homeostasis can be used to evaluate experimental predictions of in vivo stem cell behaviour 71,72 .
Recent studies are shedding light onto the regulation of CBC stem cell fate choices in vivo. Surprisingly, epigenetic regulation via DNA methylation does not seem to play a major part in directing their fate 73 . Conversely, the GTPases CDC42 and RAB8A, together with the endoplasmic stress response pathway, seem to be important mediators of both stem cell division and differentiation 74,75 .

Validation of +4 cells as stem cells
Several genes have been reported to be selectively expressed in the candidate +4 stem cells (BOX 3; FIG. 2a). Although some of these markers have been validated by in vivo lineage tracing, many of them seem to mark distinct epithelial cell populations with different cellular features distinct from those of the LRC population originally described by Potten. This may indeed reflect the existence of multiple phenotypically distinct stem cell populations residing at the +4 position. However, it is also becoming increasingly apparent that some mouse models do not faithfully report endogenous expression patterns of candidate stem cell markers, and this has caused some controversy regarding the validity of some of the +4 stem cell markers, which we briefly discuss below. Here, we discuss findings pertinent to markers evaluated by in vivo lineage tracing.

BMI1.
The Bmi1 gene encodes a component of the Polycomb repressor complex that is implicated in the regulation of haematopoietic and neural stem cell replication 76 . In situ hybridization analyses documented Bmi1 expression predominantly at the +4 position within the proximal small intestine, which was confirmed using a Bmi1-EGFP reporter mouse model 8 . In vivo fate-mappin g experiments using a Bmi1-ires-CreERT2/ R26R-lacZ mouse model revealed that these proximal BMI1 + populations harbour self-renewing, multipotent stem cells contributing to long-term epithelial homeostasis. Additional support for these BMI1 + cells having stem cell identity was provided by in vivo ablation of the BMI1 + population, which blocked epithelial renewal, and by showing that isolated BMI1 + cells can generate epithelial organoids in culture that are reminiscent of functional intestina l tissue 8 , 77 .
The BMI1 + population was found to be relatively quiescent, radiation resistant and not regulated by WNT signalling, prompting the conclusion that BMI1 marks a +4-restricted reserve stem cell population that is different from that in the LGR5 + CBC compartment at the crypt base. However, further expression analyses have documented robust Bmi1 expression throughout the proliferative zone of the crypt, including the LGR5 + CBC compartment 51,[78][79][80][81] . In agreement with such a broad endogenous expression pattern, a recent fate-mapping study using a Bmi-CreERT2 transgenic model noted the initiation of Bmi1-driven lineage tracing at random locations throughout the crypts 8 . Moreover, efforts to reproduce the original findings using the Bmi-ires-CreERT2 mouse model led to similar conclusions: in contrast to previous reports of lineage tracing being predominantly initiated at the +4 position, tracing initiation occurred at random positions throughout the crypt, including the CBC compartment 51 . Over time, most of these lineage tracing events were lost, which is consistent with them having originated within the short-lived TA progenitor cell compartment. Long-term tracing events were postulated to arise from CBC stem cells expressing both Lgr5 and Bmi1 (REF. 51).
Taken together, these latest findings are incompatible with BMI1 being a selective marker of the +4 stem cells in the small intestine, casting doubt on the value of using Bmi1-driven lineage tracing for studying the contribution of +4 reserve stem cells to crypt regeneration and disease.

HOPX.
The Hopx gene encodes an atypical homeobox protein predominantly expressed at the +4 positio n throughout the entire intestine, as shown using

Box 3 | Other reported +4 markers
Many other purported markers of intestinal stem cells, often collectively referred to as +4 stem cell markers despite their poor characterization, have been published over the past decade 110 . Unlike Bmi1, Leu-rich repeats and immunoglobulin-like domains 1 (Lrig1) and telomerase reverse transcriptase (Tert), these have not been validated by in vivo lineage tracing but have instead been proposed on the basis of their localization within the crypt or their selective expression on cells displaying presumptive stem cell attributes, such as DNA label retention or cellular quiescence.
Immunohistochemistry analyses have detected enriched phospho-phosphatase and tensin homologue expression on label-retaining cells (LRCs) residing just above the Paneth cell compartment around positions +4 and +5 (that is, the fourth and fifth position from the crypt base) 111,112 . However, these findings were subsequently questioned when the same antibody was shown to detect a subset of post-mitotic enteroendocrine cells at the crypt base 113 .
Expression of the phosphatase Wip1 (also known as Ppm1d) was also reported to be predominantly restricted to position +4, although WIP1 + cells intercalated with the Paneth cells were also readily apparent 114 . Somewhat surprisingly, the observed depletion of this candidate stem cell population in Wip1-knockout mice had no discernible effect on epithelial homeostasis.
Interest in doublecortin-like and CAM kinase-like 1 (Dcamkl1; also known as Dclk1) as a potential stem cell marker resulted from its identification as a highly expressed gene in stem cell-enriched zones micro-dissected from the crypt base 115 . Immunohistochemical analysis subsequently documented expression on rare, quiescent cells predominantly located around position +4 (REFS 116,117). Their potential stem cell status was further enhanced when lineage marker studies failed to document any overlap with the major differentiated epithelial cell types. However, more recent studies detected DCAMKL1 + cells throughout the crypt and villus epithelium and phenotyped them as a rare population of tuft (caveolated) cells, a differentiated cell type displaying thick tufts of microvilli the in vivo function of which is still poorly understood 10,95,118 . This was confirmed by in vivo lineage tracing using a new Dcamkl1-driven CreERT2 allele 119 .
Hopx-lacZ reporter mice (in which LacZ is inserted in the endogenous Hopx locus) 81 . These HOPX + cells are relatively quiescent and radiation resistant, and they harbour the capacity to rapidly proliferate in response to irradiation-induced injury. A formal demonstration of HOPX + stem cell identity was provided by in vivo lineage tracing using a Hopx-ires-CreERT2 mouse model, when reporter gene activation at the +4 position produced persistent, multipotent stem cell signature tracing s throughou t the intestine.
Comparative expression profiling of lasermicrodissect ed HOPX + cells and their immediate crypt base progeny showed that HOPX + cell descendants have higher expression of Lgr5 and other CBC marker genes. Conversely, organoid culture assays confirmed that isolated LGR5 + CBC stem cells give rise to HOPX + cells ex vivo 81 .
These observations support a model in which proliferating LGR5 + stem cells and quiescent HOPX + stem cells located at distinct anatomical locations within the crypt efficiently interconvert during epithelial homeostasis. Such functional interplay between +4 cells and CBC cells at lower crypt positions helps to unify the two models of intestinal stem cell identity. However, it should be noted that a different Hopx expression profile was reported in a separate study, in which single-molecule RNA fluorescence in situ hybridization (FISH) and LGR5 + CBC expression profiling analyses detected endogenous Hopx expression throughout the crypt, with highest levels in the LGR5 + CBC cells at the crypt base 51 .

LRIG1.
The Lrig1 gene encodes a single-pass transmembrane receptor that functions as a conditional inhibitor of ERB proteins in several adult tissues. Its discovery as a marker and proliferation regulator of stem cells in the epidermis 82 prompted efforts by two groups to investigat e its value as a marker of intestinal stem cells.
In one study, in vivo lineage tracing using an Lrig1ires-CreERT2 mouse model generated long-term tracing units typical of multipotent stem cell output throughout the small intestine 80 . Reporter gene activation was most commonly observed within positions +2 to +5, although tracing was also initiated throughout the lower portion of the TA compartment, which is consistent with a gradien t of Lrig1 expression emanating from the crypt base.
In the colon, Lrig1 expression marked a small population of cells at the very bottom of the crypt, a minority of which co-expressed Lgr5. Although this was interpreted as evidence of very limited overlap between the LRIG1 + and LGR5 + stem cell populations, the mosaic expression of the Lgr5-EGFP reporter allele in the colon invalidates this claim. Fate mapping indicated that at least some of the cells were actively cycling stem cells contributing to daily epithelial homeostasis. However, a proportion of LRIG1 + cells that underwent Cre-mediated reporter gene activation failed to generate labelled progeny over a long period, indicating that these might be quiescent. These quiescent cells were induced to proliferate and contribute epithelial progeny following irradiation-induced damage to the colonic epithelia.
Comparative expression profiling of LRIG1 + and LGR5 + populations from the colon yielded markedly different transcriptomes, providing further evidence of their independence. Although both populations expressed equivalent levels of reported +4 markers, including BMI1, TERT and prominin 1, the LRIG1 + cells selectively expressed gene signatures involved in oxidative damage responses and negative regulation of cell proliferation. LRIG1 was itself identified as tumour suppressor in the intestine when genetic ablation resulted in adenoma formation within 6 months 80 . This prompted speculation that LRIG1 functions to regulate ERBB signalling on the intestinal stem cell compartment to prevent aberrant stem cell activities that could lead to cancer formation.
A second study yielded markedly contrasting conclusions in the small intestine 83 . In situ and immunohistochemistry analyses confirmed LRIG1 expression throughout the lower third of the small intestine crypts, but the LGR5 + stem cell transcriptome revealed a significant enrichment of Lrig1 expression within this CBC population 51 . An independent study using singlemolecule mRNA FISH also documented Lrig1 as being expressed in a broad gradient, with the highest levels being present in LGR5 + CBC stem cells 51 . This extensive overlap of the LRIG1 + and LGR5 + crypt populations was confirmed at the protein level through flow cytometric analysis using an independent LRIG1-specific antibody 51 . The LRIG1 + cells were also characterized as being actively proliferating -again, in stark contrast to the colonresident LRIG1 + cells that were documented as being quiescen t by Powell and colleagues.
A clue to the function of LRIG1 in LGR5 + stem cells was revealed by phenotypic analysis of Lrig1-knockout mice intestines. Extensive crypt hyperplasia was evident soon after birth, and the LGR5 + stem cell compartment was markedly expanded, which was directly linked to deregulated Erbb expression in the stem cell compartment (as treatment of Lrig1-knockout mice with an ERBB inhibitor restored crypt homeostasis).
Collectively, these studies suggest that LRIG1 has important functions in regulating ERBB signalling in the intestinal stem cells to prevent disease. However, the broad expression gradient of Lrig1 in crypts of the small intestine and its enrichment in the LGR5 + CBC cell pool seem to disqualify it as a specific marker for intestinal stem populations.
TERT. Elevated telomerase expression is considered to be a stem cell trait that protects against replication-induced senescence. Using Tert-GFP reporter mice, rare, predominantly quiescent Tert + cells were detected around the +4 position 84 . These TERT + cells were later shown to be independent of the LGR5 + CBC stem cells and other +4 stem cell pools, and resistant to ionizing irradiation 79 . In vivo lineage tracing demonstrated that a small fraction of the TERT + crypt cells were actively cycling stem cells contributing to epithelial homeostasis in both the small intestine and colon 79 . The quiescent TERT + cells that did not participate in daily tissue homeostasis could be activated and proliferate to contribute to damage-induced epithelial regeneration. Collectively, these observations seem to support TERT as being a marker of an independent, quiescent, damage-inducible pool of intestinal stem cells. However, other studies evaluating endogenous telomerase expression and activity levels in the crypt have reached contrasting conclusions 69,78 . Tert mRNA and telomerase activity were readily detected in all proliferative cells of the crypt, with highest levels present within LGR5 + CBC stem cells sorted by FACS. Single-molecule mRNA FISH analysis confirmed this expression pattern in proximal small intestine. Such an elevated telomerase activity would probably benefit the highly proliferative LGR5 + stem cells, but it remains unclear what advantage this would bestow on quiescent populations that by definition rarely divide.

A current view of adult stem cell identity
The intestinal stem cell field has seen major advances over the past 5 years, mostly driven by the identification of new stem cell markers and the development of nearphysiological assays to rigorously evaluate endogenous stem cell identity in vivo. Despite these technological breakthroughs, there is still no consensus on the true identity (+4 cells versus LGR5 + CBC cells) of adult intestinal stem cells. Recent attempts to reconcile the opposing views have led to the formulation of a more plastic model of stem cell identity. This model describes the intestinal crypt base as a specialized niche environment harbouring both dedicated active stem cells (the CBC cells), responsible for daily epithelial homeostasis, and more quiescent 'reserve' stem cells (collectively known as +4 cells) that can be activated to effect tissue repair following injury. This new model also suggests that cells residing in the lower positions of the TA cell compartment can acquire stem cell identity to ensure maintenance of epithelial homeostasis in the eventuality of a catastrophic loss of stem cell populations resident in the crypt base (FIG. 3).

Epithelial regeneration and crypt plasticity
The intestinal epithelium has a remarkably ability to survive acute injury and to effect rapid regeneration to restore function. Although not strictly physiological, irradiation-induced injury models have provided important mechanistic insight into the regeneration process, revealing the existence of 'regular' (or 'actual') and 'reserve' (or 'potential') stem cell populations within the lower regions of the crypt. It has been proposed that many of the +4 markers discussed above define dedicated pools of damage-resistant reserve stem cells that are selectively activated in response to acute damage to the regular stem cell pool. However, there is growing support for a more general model of crypt plasticity, which describes a niche-induced conversion of committed progenitors into functiona l stem cells following tissue damage.

Dedicated reserve stem cells. Actively proliferating
LGR5 + CBC cells drive intestinal epithelial homeo stasis under physiological conditions 46 (FIG. 3a). However, a study using a targeted conditional ablation strategy has shown that the intestine can survive short-term loss of the CBC stem cell pool in vivo, despite CBC cells having such a crucial role, hinting at the existence of additional, damage-resistant stem cells 8 (FIG. 3b).
Based on traits such as cellular quiescence and resistance to radiation, +4-resident cells are reserve stem cell candidates 77 . Although the +4 populations marked by HOPX and LRIG1 remain to be functionally identified as reserve stem cells, the exquisite radiation-sensitivit y of the +4 LRC population originally described by Potte n would seem to disqualify it from serious consideration 48 . Strong support for the existence of a BMI + reserve stem cell pool was provided by in vivo lineage tracing from the Bmi1 locus following acute ablation of the LGR5 + CBC stem cell compartment, revealing that Bmi1 + cells contribute to the repopulation of the LGR5 + stem cell pool during the subsequent regeneration phase 8 (FIG. 3c). However, the fact that Bmi1 is expressed throughout (and traces from) the entire proliferative zone of the crypt precludes a formal identification of these as reserve stem cells 51 .

Committed progenitors as potential stem cells.
An alternative, although not necessarily mutually exclusive, explanation for the robust regenerative capacity of the intestinal epithelium may be related to the plasticity of early stem cell progeny. Clonal regeneration assays have identified up to 30-40 cells in the lower third of the intestinal crypts with the ability to survive exposure to varying doses of radiation or cytotoxic drugs and contribute to crypt regeneration 40,85 . These observations hinted at the existence of a stemness hierarchy in the lower crypt region, with early LGR5 + CBC cell progeny gradually losing their stem cell features during successive rounds of cell division as they progressively switch on differentiation programs during their migration up the crypts. Following acute loss of the regular stem cell pool, the surviving progenitor populations are thought to fall back into the vacant stem cell niche at the crypt base, where they quickly re-acquire stem cell identity and re-establish epithelial renewal 48 . Such niche-driven plasticity has been demonstrated using ex vivo organoid cultures (BOX 4). Under normal conditions, LGR5 − TA cells isolated from intestinal crypts are incapable of generating epithelial organoids in culture. However, brief exposure to one of the major in vivo niche signal s, WNT3A, efficiently converts these LGR5 − cells into organoid-proficient LGR5 + stem cells 86 .
Strong evidence in support of this model was recently provided by approaches using inducible histone H2B-GFP (or histone H2B-yellow fluorescent protein) to identify and isolate LRCs in the small intestine 7,87 . Long-lived LRCs (4-9 weeks old) comprised exclusively mature Paneth cells, which is consistent with the lifespan (6-8 weeks) of this terminally differentiated population 13 . However, a different, shorter lived (<4 weeks) LRC population presented a mixed phenotype, expressing a collection of markers for Paneth cells, enteroendocrine cells and LGR5+ CBC and +4 stem cells. Using an elegant split-Cre recombinase strategy to carry out lineage tracing from these LRCs in vivo, it was shown that under physiological conditions the mixed-phenotype LRC population functioned as bipotent progenitors of the mature Paneth cell and enteroendocrine cell lineages 7 . However, following irradiation-induced damage,

Pulse-chase
A method for detecting quiescent label-retaining cells (LRCs) in vivo. A nucleotide analogue (the label) is administered to the mouse for a short period (the pulse); this is followed by an extended period when no further nucleotide analogue is given (the chase). Actively dividing cells rapidly dilute out the label during DNA replication, and labelled cells are lost within 3-4 rounds of cell division. Non-dividing (quiescent) cells do not dilute out the label and thus remain detectable as LRCs.
cells of the same LRC population were rapidly converted into self-renewing, multipotent stem cells contributing to epithelial regener ation (FIG. 3d). This LRC population was further characterized as being a subset of the LGR5+ CBC cells, constituting 20% of the total LGR5+ pool at the crypt base.
A major implication of this finding is that LGR5 is marking a mixed population of regular and reserve stem cells in the small intestine. However, if this were true then Lgr5-driven lineage tracing would result in the generation of a substantial proportion of Paneth cell-restricted clones under physiological conditions -this has never been reported. As the regeneration capacity of the small intestinal epithelium is maintained following short-term ablation of the entire LGR5-expressing crypt population, including both regular CBC and reserve LRC stem cell pools, other LGR5 − cell populations capable of functioning as stem cells must exist.
In a separate study, LRCs were characterized after 7 weeks of pulse-chase. In agreement with the findings described above, these long-lived LRCs were differentiated Paneth cells 87 . These label-retaining Paneth cells did not express LGR5 and could be activated to proliferate in response to irradiation injury, generating BMI1 + cells proposed to function as stem cells during the subsequent tissue regeneration process. These findings are in contrast to those of the earlier study, which documented the Paneth cell LRCs as being a terminally differentiated population incapable of injury-induced stimulation 7 . The colonic epithelium, which is devoid of Paneth cells and LRC populations, is also capable of surviving acute injury, suggesting the existence of actively cycling cells capable of functioning as stem cells during regeneration.
In another study, in vivo lineage tracing was used to demonstrate the plasticity of early stem cell progeny expressing the Notch ligand Delta-like 1 (DLL1) 88 . Under physiological conditions, DLL1 + cells at the +5 position (the proposed 'common origin of differentiation') function as secretory progenitors, generating short-lived clones comprising goblet cells, enteroendocrine cells, Paneth cells and tuft cells. However, irradiation-induced depletion of the LGR5 + CBC stem cell compartment resulted in conversion of the DLL1 + cells into multipotent LGR5 + CBC stem cells contributing to the subsequent epithelial regeneration process 88 (FIG. 3d).  LGR5-GFP -/RFP + colonic crypt Ex vivo culture systems, capable of efficiently and safely maintaining the long-term regenerative capacity of purified stem cells following their extraction from endogenous niches, are of crucial importance for realizing the clinical potential of adult stem cells. Major advances have been made in establishing culture systems that support the long-term growth and expansion of near-physiological intestinal epithelia from purified stem cells 20,22,47,120 . To date, several methods have been used to grow 'organoids' from the small intestine 121 , but the most successful method is a Matrigel-based three-dimensional culture system that supports the growth of self-renewing, near-native intestinal epithelia in the absence of stromal niche components 47 (see the figure, part a). In this system, intact intestinal crypts or purified LGR5 + (Leu-rich repeat-containing G protein-coupled receptor 5-expressing) crypt base columnar (CBC) stem cells are plated into laminin-rich Matrigel supplemented with a cocktail of growth factors found in the endogenous stem cell niche, including the bone morphogenetic protein inhibitor Noggin, the WNT agonist roof plate-specific spondin 1 (R-spondin1), epidermal growth factor (EGF) and Notch ligand 47 . This generates self-renewing epithelial organoids organized into discrete crypts harbouring intercalated stem cells and Paneth cells at their base and associated villus-like regions comprising the various differentiated cell lineages. This system was adapted for use in culturing colonic epithelia, although this required exogenous WNT3A, probably reflecting subtle differences in the endogenous stem cell niches of the small intestine and colon 120 (see the figure part b, which depicts the generation of red fluorescent protein-expressing (RFP + ) colonic organoids from a single LGR5 + stem cell carrying a permanent, heritable RFP mark). In a beautiful example of their clinical potential, both mouse colonic organoids and intestinal 'spheroids' were successfully used to repair damaged colonic epithelia in vivo when delivered via a simple enema 22,122 (see the figure, part c, which shows that cultured RFP + colonic organoids can be used to effectively repair damaged colonic epithelia in vivo via orthotopic transplantation in the mouse, as supported by the appearance and maintenance of patches of RFP + colonic epithelia in the recipient wild-type mouse). Importantly, the organoid culture system has recently been successfully adapted for routinely growing human epithelia from the small intestine and colon 50,120 . This development opens up many exciting translational opportunities in the biomedical sectors, including the de novo growth of isotype-matched human epithelia for treating human gastrointestinal (GI) tract diseases, such as ulcers, and the expansion of matched healthy and tumour epithelia from patient biopsies for use in basic research into epithelial biology and various biomedical applications, including drug screening and deep-sequencing efforts as a prerequisite to the development of personalized treatment regimens (see the figure, part d).
An alternative source of intestinal stem cells for therapeutic applications was revealed recently, when it was shown that abundant progenitor cells isolated from both mouse and human fetal intestine could be expanded as self-renewing spheroids in an ex vivo culture system. Exposure to exogenous WNT ligands efficiently converted these progenitor cultures into mature organoid structures, highlighting their value as a source of functional intestinal epithelia 22 .

Box 4 | Biomedical applications of intestinal stem cells: the epithelial organoid culture system
This plasticity of the lower crypt cell populations during epithelial regeneration is probably dependent on instructive signals emanating from the surviving stem cell niche. Such niche-driven plasticity may also exist within gastrointestinal tumours, which would have important implications for therapies targeting the elimination of cancer stem cell populations (see Supplementary information S1 (Box)). The intestinal niche is a critical component in governing stem cell behaviour not only during epithelial homeostasis but also during tissue regeneration following injury.

Summary and future outlook
The intestinal stem cell field has seen impressive advances over the past decade, fuelled by the identification of specific biomarkers, better characterization of the endogenous niche and the development of new in vivo and ex vivo models for rigorously evaluating stem cell identity and function during homeostasis and disease (BOX 4). After several years of intense debate over the precise identity and function of the intestinal stem cell, it is becoming apparent that there is perhaps no definitive answer. Rather than relying on a single, hard-wired stem cell compartment to maintain epithelial homeostasis and effect tissue regeneration following injury, the intestine seems capable of drawing on several pools of highly plastic, reserve stem cell populations in the lower regions of the crypt (FIG. 3). Such plasticity is likely to be endowed by the specialized niche environment at the crypt base (BOX 1; FIG. 2b), which provides the requisite signals to efficiently convert committed progenitor populations into multipotent adult stem cells.
Such niche-driven plasticity would explain the impressive ability of the intestinal epithelium to survive and regenerate following major injury and could potentially be exploited for regenerative medicine applications in the clinic.
The next major challenge for the intestinal stem cell field is to rapidly translate our knowledge of mouse stem cells into the human arena to ensure that the clinical potential of intestinal stem cells is quickly realized. Efforts to exploit intestinal stem cells for regenerative medicine have accelerated over the past decade, driven mostly by the identification of cell-surface markers facilitating the isolation of pure stem cell populations and the development of near-physiological culture methods supporting stem cell expansion and epithelial growth (BOX 4).
LGR5 is one of the few cell-surface stem cell markers that could potentially be used to isolate human intestinal stem cells, but it has proved to be a difficult targe t for antibody generation. However, next-generatio n strategies such as genetic immunization are starting to deliver antibodies capable of effecting the isolation of live human intestinal stem cells 89 . The available human intestinal organoid culture systems will be instrumental in evaluating and characterizing these candidate human stem cells as a prerequisite to exploiting their clinical potential in the near future. Recent characterization of the complex niche signals that regulate stem cell function in vivo has also been of fundamental importance to these translational efforts, ensuring maintenance of the regenerative capacity of isolated stem cells and minimizing the risk of introducing undesirable heritabl e traits that could compromise the safety of patients.