Concise review: The Yin and Yang of intestinal (cancer) stem cells and their progenitors


  • Daniel E. Stange,

    1. Hubrecht Institute, Royal Netherlands Academy of Arts and Sciences and University Medical Center Utrecht, Utrecht, The Netherlands
    2. Department of General, Thoracic and Vascular Surgery, University Hospital Carl Gustav Carus, University of Dresden, Dresden, Germany
    Search for more papers by this author
  • Hans Clevers

    Corresponding author
    1. Hubrecht Institute, Royal Netherlands Academy of Arts and Sciences and University Medical Center Utrecht, Utrecht, The Netherlands
    • Correspondence: Hans Clevers, M.D., Ph.D., Hubrecht Institute, Royal Netherlands Academy of Arts and Sciences and University Medical Center Utrecht, Utrecht, The Netherlands. Telephone: +31 (0)30 212 18 31; Fax: +31 (0)30 212 18 01; e-mail:

    Search for more papers by this author

  • Author contributions: D.E.S. and H.C.: D.E.S. and H.C. wrote this review together.


The intestine has developed over the last few years into a prime model system for adult stem cell research. Intestinal cells have an average lifetime of 5 days, moving within this time from the bottom of intestinal crypts to the top of villi. This rapid self-renewal capacity combined with an easy to follow (mostly) unidirectional movement of cells offers an ideal site to conduct adult stem cell research. The delineation of the active pathways in the intestinal epithelium together with the development of molecular techniques to prove stemness laid the grounds for the identification of the intestinal stem cell. In vitro systems and transgenic mouse models broaden our knowledge on the role of the stem cell niche and those cells that reestablish homeostasis after perturbation of the system. These insights expedited also research on the role of normal adult stem cells in cancer initiation and the factors influencing the maintenance of cancer stem cells. Stem Cells 2013;31:2287–2295


The intestinal epithelium constitutes the fastest self renewing tissue of the body [1]. Constant mechanical damage inflicted on the gut by passing bowel content in combination with the chemical and biological assault by the luminal contents necessitates the nearly complete renewal of the epithelium every 4–5 days [2]. The gut is subdivided anatomically into two parts, the small intestine and the colon. While the small intestine's main duty is the absorption of nutrients and production of antimicrobial proteins, the colon mainly absorbs water back into the body while producing significant amounts of mucus. These two diverging functions are mirrored at the cellular level by a different anatomical setup. The small intestine's surface is maximized by millions of protrusions of the epithelium, called villi, and by invaginations into the submucosa, called crypts. Specialized absorptive cells (enterocytes) and mucus-secreting goblet and hormone-secreting enteroendocrine cells are the main cell types of villi. The origin of these differentiated cell types can be found in the crypts. Stem cells at the base produce rigorously dividing transit-amplifying (TA) cells. While proliferating, TA cells move upward and reach the crypt-villus junction after four to five cell divisions, after which they fulfill their specific function as specialized cells [3]. To compensate the continuous, conveyor-belt-like flow of cells along the crypt-villus axis, cells at the villus tip undergo apoptosis and are shed into the gut lumen. Only one cell type, the Paneth cell, escapes this upward movement. Pushed downward by the repulsive forces of ephrins and their EphB receptors, Paneth cells settle at the very base of crypts [4]. This cell type specializes in providing niche signals to stem cells and in the production of antimicrobial proteins, which keep the small intestine largely free of bacteria. Paneth cells have an average life span of 6–8 weeks, being the only differentiated cell type escaping the rapid self renewal [5]. While having an overall similar setup, the colon differs from the small intestine by the absence of villi, creating a flat surface epithelium. While typical Paneth cells are missing in the colon, deep-crypt-secretory cells have been proposed to represent their colonic counterpart [6]. Indeed, it has recently been shown that a c-Kit positive population of goblet cells resides directly adjacent to Lgr5+ stem cells at the colonic crypt bottom, that this cell produces Notch ligands and Egf similar to Paneth cells, and that it can functionally support Lgr5+ stem cells in organoid cultures [7]. Colonic TA cells differentiate toward the goblet and absorptive cell lineages.

Intestinal Stem Cells: Two Schools of Thought

As proliferation in the intestine is restricted to the crypts, it has long been envisioned that stem cells reside at the bottom of these epithelial invaginations [8, 9]. Two schools of thought have dominated the literature for decades: the “+4” stem cell model, proposed by Potten and colleagues [10] (Fig. 1A) and the stem cell zone model by Bjerknes and Cheng [11, 12] (Fig. 1B). The +4 cell was mainly put forward as a potential stem cell due to its specific characteristics: the +4 cell was reported to be unusually radiation sensitive, a property thought to be beneficial to stem cells as it would clear damaged stem cells from the stem cell pool [13]. The +4 cell was found to be dividing every 24 hours, at the same time being DNA label retaining by asymmetrically segregating its DNA [14, 15]. It is important to stress that the +4 stem cell as defined by Potten is not quiescent, while most subsequent studies on +4 stem cells have explicitly stressed their quiescent nature. Several markers of the +4 cell have been put forward, including Bmi1, Tert, Hopx, and Lrig1 [16-19]. These studies provided evidence that the +4 cell constitutes a rather quiescent stem cell, which is able to generate the rapidly cycling Lgr5+ stem cell [20]. An interconversion between +4 and Lgr5+ stem cells has also been documented [18]. On the other hand, the stem cell zone model puts forward crypt base columnar (CBC) cells [1, 21] as the intestinal stem cell, a cell type that is squeezed in between Paneth cells. These slender cells fulfill several requirements believed to be linked to stem cells: they are small, undifferentiated and cycling cells with a high nucleus-to-cytoplasm ratio [1]. In addition to morphological considerations, clonal marking by either radioactive phagosomes or chemical mutagenesis pinpointed CBC cells as the stem cells of the small intestine [21, 22].

Figure 1.

Two models for the intestinal stem cell. (A): The +4 stem cell model as proposed by Potten and adjusted by recent studies on the +4 cell. The +4 stem cell is capable of regenerating the Lgr5+ stem cell, which in turn is the work-horse of daily tissue regeneration, producing the four main lineages of the intestinal epithelium. Possibly, the Lgr5+ stem cell can also convert back into a +4 stem cell. (B): The stem cell zone model as proposed by Bjerknes and Cheng and adjusted by recent studies on the Lgr5+ stem cell. Lgr5+ stem cells reside in between Paneth cells, which define the stem cell zone. Once progeny of Lgr5+ stem cells loose contact to Paneth cells, they start to differentiated either toward the enterocyte lineage through a number of fast cell divisions as transit amplifying cells. Or they become a Dll1high secretory cell precursor, which differentiates toward the three secretory lineages (goblet, enteroendocrine and Paneth cells). A recently identified cell type is the Lgr5+ DNA label retaining cell, which can differentiate toward the Paneth and enteroendocrine cell lineage. The relationship between the Dll1high and the Lgr5+ DNA label retaining cell still needs to be determined.

Experimental Approaches to Assess Stemness

Adult stem cells are defined by just two characteristics: self-renewal or longevity, that is, stem cells can renew themselves and persist lifelong, and multipotency, that is, all differentiated cell types of a specific tissue originate from this stem cell [10]. As quiescent and actively cycling stem cell pools may coexist, even in the same tissue, defining stem cells based on cell cycle properties alone can be misleading [23]. Also, the mere expression of a stem cell marker in a given tissue that has been identified in a different tissue as stem cell-specific is insufficient, as a common signature of adult stem cells has not be defined up to today. In our view, definite proof of stemness can be achieved by two experimental approaches only: lineage tracing and (with limitations) transplantation of single cell-derived preparations [24, 25].

Lineage Tracing

Cre recombinase-based lineage tracing allows for fate mapping of stem cell progeny in vivo (see rev [24]. for an overview). This technique normally does not impact on the physiological behavior of the tissue under investigation, also recent publications pointed to a potential apoptotic effect of tamoxifen on certain cell types in certain tissues [26, 27]. Originally introduced by Cotsarelis and colleagues [28], the technique relies on the identification of a marker gene or a specific feature [29] for a specific stem cell pool. An inducible version of the Cre enzyme is then expressed from the marker gene's promoter. The Cre enzyme can excise DNA fragments that are flanked by a specific DNA sequence named LoxP (locus of X-over P1) sites. When fused to a fragment of the estrogen receptor, the Cre-ER fusion protein resides in the inactive state in the cytoplasm, but shuttles to the nucleus upon activation by the ER ligand tamoxifen. When crossed with a Cre-reporter line, for example, the R26R-LacZ mouse [30] or the multicolor R26R-confetti mouse [31], nuclear Cre can excise the LoxP-flanked transcriptional roadblock in front of the reporter, leading to expression of the reporter and irreversible marking of Cre-expressing cells. Upon cell division, the roadblock deleted DNA is carried over to the daughter cells, leading to reporter expression in all offspring. A refinement of this technique has been introduced by expressing both the Cre enzyme and a fluorescent protein from the same promotor using a bicistronic mRNA [32]. Just like the Cre enzyme, the fluorescent protein is only transcribed in cells originally expressing the marker gene. This allows to perform lineage tracing, while the cell of origin of the trace is visualized.


Transplantation is the historical gold standard for assessing stemness. If performed with a single cell, it can reveal self-renewal and multipotency [33-36]. A prerequisite of the method is the possibility to isolate single stem cells. In addition, a suitable assay system needs to be established, for example, irradiation for hematopoietic or muscle stem cells or the “cleared fat pad” method for mammary gland stem cells. One has to bear in mind that these assay systems in many cases substantially alter the receiving tissue and the transplanted cell finds itself in a different microenvironment. A transplanted cell may therefore elicit a wider contribution toward the different lineages of its origin when compared with its behavior in normal homeostasis. Good examples for this are hair follicle stem cells, that normally only contribute to the hair follicle, but after transplantation can generate all components of the skin, that is, epidermis and sebaceous glands [28, 37-39]. The “actual” stemness behavior in normal homeostasis therefore has to be distinguished from the stemness potential [24, 40]. Transforming thymic epithelial cells into hair follicle stem cells upon transplantation intriguingly have shown this difference [41]. Transplantation assays therefore complement lineage tracing to document both the actual and potential stemness potential.

Identification of Intestinal Stem Cells

As discussed above, a prerequisite to prove stemness of a certain cell is the identification of a specific marker or a specific feature: either for the generation of lineage-tracing mice or for isolation purposes prior to transplantation. In the case of the CBC cell, this has been accomplished step by step. At the basis lies the insight, that the Wnt pathway is the major driving force of proliferation and tissue maintenance in the intestinal crypt. Removal of Tcf4 (Tcf7l2), one of the main downstream effectors of the Wnt pathway, blocks proliferation and depletes the intestinal stem cell compartment [42]. Subsequent studies have identified the core transcriptional Wnt pathway program in the intestine [43, 44]. Genes within this pathway can be grouped by their expression pattern within the intestinal crypt: genes that are expressed throughout the proliferative compartment including the TA compartment, genes expressed in Paneth cells, and a small group of genes with a restricted non-Paneth cell expression at the bottom of the crypt. Members of the latter group are Ascl2 and Lgr5. To confirm the expression and - bearing in mind the stem cell zone model - to generate a tool to perform in vivo lineage tracing, an Lgr5-EGFP-ires-CreERT2 (Lgr5-ki) mouse was established [32]. EGFP expression in this mouse model was confined to CBC cells, faithfully recapitulating the endogenous expression pattern of Lgr5. Immediately upon induction of Cre activity in Lgr5-ki x R26R-LacZ mice, single blue CBC cells could be detected. These cells over time produced offspring that formed ribbons along the crypt-villus axis, including all differentiated cell types of the intestine. Tracings persisted over the lifetime of mice. CBC cells of the intestine therefore fulfill the requirements for stem cells: multipotency and longevity.

The first lineage tracing from the +4 cell position was reported using Bmi1 as a Cre driver [16]. Bmi1 is a member of a polycomb transcriptional repressor complex found to be involved in self-renewal of neuronal and hematopoietic stem cells [45, 46]. Bmi1 driven lineage tracing resulted in the labeling of complete crypts. All differentiated cell types were contained within the Bmi1 progeny and tracings persisted long-term, thus proving that Bmi1 is expressed in pluripotent and self-renewing stem cells. Nevertheless, single molecule in situ hybridization indicated, that Bmi1 is expressed throughout the crypt, including the Lgr5+ stem cells compartment [47]. This was confirmed in a repeat of the original experiment, where lineage tracing was found not to initiate specifically at the +4 position, but throughout the crypt including in the CBC stem cells [48]. Bmi1 is a good example of the pitfalls of lineage tracing: this technique needs to be accompanied by detailed analyses of the expression domain of the Cre driver and the site of initiation of lineage tracing [49].

Three additional markers have been proposed to mark an Lgr5 independent (quiescent) stem cell population based on lineage tracing experiments. Expression of telomerase reverse transcriptase (Tert) has been reported to occur in a rare cell population [50], and lineage tracing from the Tert promotor revealed full gland labeling [17]. Yet, in an independent study, Tert mRNA expression was found consistently in every cell throughout the whole crypt [47, 48]. Telomerase activity was found at significant levels in Lgr5 progeny, while being highest in Lgr5+ stem cells itself [51]. Hopx is an atypical homeodomain-containing protein, which has been proposed to constitute an Lgr5 independent quiescent stem cell marker [18]. Nevertheless, expression analyses of sorted Lgr5 stem cells have detected robust levels of Hopx mRNA and Hopx protein, indicating at least a partial overlap [48]. This was complemented by single molecule in situ hybridization and immunohistochemistry, both demonstrating a broad expression domain of Hopx including the CBC stem cell zone. The pan-ErbB inhibitor Lrig1 is the fourth gene proposed to mark a Lgr5 independent stem cell population in the intestine [19]. This conclusion was based on lineage tracing results and expression domain analysis using a newly generated Lrig1 antibody. Conversely, in a parallel study using a commercially available and previously validated antibody [37], a different staining pattern could be observed [52]. Lrig1 staining was observed consistently in every crypt, and a gradient of expression was observed throughout the crypt with highest expression at the bottom both in small intestine and in colon.

Two recent studies have directly assessed the identity of DNA label retaining cells. Fodde and colleagues [53] used a pulse of the chromatin marker H2B-GFP and found that rare cells still retained this label after 50–60 days. These cells appeared to represent various precursors and mature versions of Paneth lineage cells. Winton and colleagues [29] (see also below) then showed that some of these cells coexpress Lgr5 and all +4 markers and that—under conditions of damage—these cells could revert to become Lgr5+ stem cells. Thus, these observations may unify the two competing stem cell models [79].

Primary Intestinal Cultures from Intestinal Stem Cells

The identification and molecular characterization of the intestinal stem cell stimulated efforts to culture them. The establishment of an in vitro culture for the intestine has been pursued for a long time. The main issue has been the survival of the cultures, with most not being sustainable for more then a few days [9]. In vitro cultures would offer many advantages over in vivo models, including ease of experimental manipulation and long-term live imaging. Two independent approaches have overcome this obstacle: both systems rely on a three-dimensional (3D) collagen/laminin-rich gel, but one includes a mesenchymal niche, while the other is purely epithelial. Ootani et al. established a 3D culture by using intestinal tissue fragments implanted within a collagen gel with an air-liquid interface [54]. Intestine from neonatal mice could be cultured as spheres including crypt- and villus-like structures for prolonged time periods. Multilineage differentiation from a proliferating compartment including Lgr5 stem cells could be observed. The cultures were propagated without the addition of growth factors to the culture medium, indicating that the cocultured mesenchyme provides all necessary factors. This was substantiated by the fact that long-term cultures were only viable when associated with a robust underlying stroma. Taken together, this culture method recapitulates the complex nature of epithelial-mesenchymal interaction in the intestine and allows for the multilineage growth of intestinal epithelium. As cultures regularly survive up to 30 days, in vitro manipulation of the system is possible and can recapitulate accurately the in vivo situation, as demonstrated by Ootani et al. by the interference with the Notch and Wnt pathway.

A purely epithelial system was established by Sato et al. using isolated intestinal crypts grown in a 3D laminin-rich gel (Matrigel) [55]. The identification of the growth factors necessary to maintain the culture is based on genetic studies, delineating the importance of the Egf [56, 57] and Wnt pathway [42] for the maintenance of the intestinal architecture and the effect of the Bmp pathway on the terminal differentiation of intestinal cells [58]. In this culture system, no mesenchyme is included, while the essential components of the culture medium are Egf (to stimulate the Egfr pathway), Noggin (to block Bmp signaling), and Rspondin1 (to enhance Wnt activity). Under these conditions, isolated single crypts quickly form cystic structures, extending crypt-like protrusions within a few days. The crypt domains contain Lgr5 positive stem cells interspersed between Paneth cells, and rapidly proliferating TA cells. The crypts produce offspring that move into the cystic body of the organoid, where all differentiated cell types of the intestine can be found. Apoptotic cells are shed into the central lumen, which corresponds to the lumen of the gut. Of note, organoids can be established at low efficiency (6%) from single Lrg5 positive cells, while the culture efficiency of Lgr5 negative cells is negligible. The organoids continuously and indefinitely expand and require regular splitting, which in effect yields an ever-expanding culture that produces enough cells to enable regenerative approaches. In fact, colonic organoids grown from a single stem cell have been successfully transplanted into damaged colon of mice [59]. The transplanted cells fully integrated into the epithelium, while no signs of metaplastic change became obvious. Further optimization of the culture conditions now allows the growth of human intestinal organoids, opening up new areas of research on human intestinal diseases [60, 61].

Several conclusions can be drawn from the mesenchyme free culture system: first, a mesenchymal niche is dispensable to establish a well orientated crypt-villus axis, with stem cells and Paneth cells located at the bottom fueling the TA compartment and subsequently the villus domain. Secondly, the cells surrounding the crypts in vivo appear to provide essential factors to the epithelium, as these factors have to be provided to the purely epithelial cultures in the form of Egf, Noggin and Rspondin in vitro, while being dispensable in the mesenchymal niche containing culture system,

The Intestinal Stem Cell Niche

One could suspect that providing these essential niche factors ubiquitously in the culture medium would lead to a more or less pure stem cell culture, as all the cells are in contact with these factors. As demonstrated by Sato et al. [55], the architecture of the crypts is set up by the epithelial cells themselves. What are these intrinsic factors that allow for this self-organization without any cue from the environment? Imaging of growing organoids from Lgr5-ki mice by live confocal microscopy revealed that whenever a new bud was formed from the cystic main body, an Lgr5-GFP positive stem cell was located at the tip of the bud [62]. Interestingly, this stem cell was always associated with a Paneth cell next to it. Upon further growth, with every stem cell also a Paneth cell was generated. The nearly geometrical distribution of the two cell types can also be observed in vivo, where stem cells and Paneth cells try to maximize their heterotypic contact area. Close to 90% of the cell membrane area at the bottom of crypts is occupied by direct contact of Paneth and stem cells. Plating either single sorted stem cells alone or a mixture of stem and Paneth cells demonstrated that stem cells indeed require this intimate contact. By plating the mixture, the organoid forming capability increased 10-fold. Transcriptional profiling of Paneth cells revealed that they express Wnt pathway ligands (Wnt3 and Wnt11), Egf, and Notch signals (Dll1 and Dll4). Paneth cells therefore provide these essential signals to stem cells. When Wnt proteins are ubiquitously available by adding them to the culture medium, organoids convert into round cysts of undifferentiated cells. Thus, the local Wnt source, further amplified by Rspondin supplied by the medium, is responsible for the development of a correct crypt axis.

The dependence of Lgr5+ stem cells on Paneth cells could also be observed in three different models of (transient) Paneth cell reduction in vivo. Two models, Gfi1−/− and CR2-tox176 mice (which express diphtheria toxin A under the Paneth cell specific cryptdin 2 promotor) show a large reduction of Paneth cell numbers [63, 64]. Nevertheless, 90% of crypts contain at least one Paneth cell. Stem cells are concomitantly reduced, but are explicitly present wherever Paneth cells can be found. In the third model, Sox9 was conditionally deleted in adult mice in all intestinal cells except Paneth cells using Ah-cre mice [65]. Preexisting Paneth cells are unaffected in this system. As Sox9 is essential for Paneth cell development [66, 67] and Paneth cells have a half live of about 8 weeks [5], a successive loss of Paneth cell numbers was observed, which cumulated in their virtually complete absence at 7–8 weeks. In parallel, stem cells also dwindled. A few crypts escaped the Sox9 deletion and contained normal numbers of Paneth and stem cells. These escaper crypts were able to rapidly repopulate the intestine, likely via crypt-fission. Of note, these three models do not show a complete loss of Paneth cells. Atoh1 is a basic helix–loop–helix transcription factor important for determining the secretory cell fate in the intestine. Expression of Atoh1 is suppressed by Notch signaling. If all Paneth cells are deleted abruptly via deletion of Atoh1, stem cells survive and show no major phenotype [68, 69]. The loss of Paneth cells eliminates Notch signaling to stem cells, as Notch signaling requires membrane-bound ligand-receptor interactions between neighboring cells. If Notch signaling is blocked by the γ-secretase inhibitor dibenzazepine (DBZ), all intestinal stem cells are lost and converted to secretory cells [70]. Similarly, simultaneous deletion of the Notch-receptor ligands Dll1 and Dll4 results in stem cell loss [71]. Nevertheless, if Notch signaling is blocked by DBZ and Atoh1 deleted simultaneously, intestinal stem cells are protected from differentiation into secretory cells [70]. Similarly, deletion of Atoh1 in all epithelial gut cells including stem cells as accomplished in studies [68, 69] results in a complete loss of Paneth cells, but at the same time relieves the stem cells from their dependency on Notch signals, which normally suppress Atoh1. Interestingly, Atoh1-deleted and thus Paneth cell-depleted crypts can not survive in vitro in mesenchyme-free organoid cultures, unless Wnt and Egf are added to the culture medium [69]. Other sources must therefore redundantly produce these Paneth cells products in vivo. Indeed, the crypt surrounding mesenchyme produces Wnt, mainly Wnt2b [72] and Egf [73]. Paneth cells and the surrounding mesenchyme therefore act as a double safety circuit for intestinal stem cells in vivo (Fig. 2). Only the Notch signal, transmitted by direct cell-to-cell contact, is uniquely provided by Paneth cells to stem cells.

Figure 2.

The intestinal stem cell niche. Intestinal stem cells depend on three signaling pathways: the Wnt, Egf, and Notch pathway. Notch signaling is provided to stem cells by Paneth cells by direct cell-to-cell contact. Notch receptors are hereby present on the cell membrane of stem cells, while the Notch ligands are presented by Paneth cells. The ligands Egf and Wnt (mainly Wnt2b) are provided redundantly by Paneth cells and by mesenchymal cells surrounding the crypt. Rspondins are probably delivered to the crypts via blood vessels, their place of production is currently unknown.

Plasticity of Intestinal Stem Cells and Progenitors

How is the stem cell zone maintained in homeostasis? Two recent studies have shed light on this topic [31, 74]. Both the studies used lineage tracing to show that intestinal stem cells divide symmetrically and that the fate of the two equipotent daughter cells is entirely niche-dependent. If both daughter cells can find a place in the restricted stem cell niche, under normal conditions defined by contact to Paneth cells, this will lead to the generation of two new stem cells. If space restriction does not allow both cells to stay in the niche, one or both lose contact to a Paneth cell, to be pushed out of the stem cell niche at the crypt bottom and to become early progenitors of the different gut lineages. This model of “neutral competition” between stem cells implies that the stem cell population and not individual stem cells maintain homeostasis over time. What happens if this homeostasis is perturbed, for example, in the case of stem cell loss? In an elegant study, Tian et al. killed Lgr5+ stem cells by injecting diphteria toxin to a mouse expressing the diphteria toxin receptor from the Lgr5 locus [75]. Of note, Lgr5+ stem cells can only be depleted for 10 days in this system, due to side effects of the toxin. Nevertheless, within this time frame, the intestinal architecture stayed intact. As the upward migration of cells still continued, this experiment clearly indicated that the small intestine continued to renew. Lgr5+ cells quickly reappeared within 48 hours after cessation of toxin injection. Using lineage tracing from the Bmi1 locus, the authors could show that reappearing Lgr5+ stem cells derive from Bmi1 positive cells. As discussed above, experimentation has indicated that Bmi1 is expressed throughout the crypt [48]. Which cells can therefore regenerate these new Lgr5+ stem cells? Recently, Dll1 was shown to be expressed highest in early progenitors of Lgr5+ stem cells [76, 77] (Fig. 1B). These cells appear around cell position +5, the first cell position from the bottom of the crypt that has no contact with Paneth cells. Under normal circumstances, lineage tracing from Dll1+ cells results in short-lived clones containing all secretory cell types. Thus, Dll1 marks a progenitor cell of the secretory lineage. After depleting Lgr5+ stem cells by irradiation, it could be shown that the Dll1+ progenitor cells can revert to Lgr5+ stem cells.

A second cell type has in the meanwhile been identified to be able to revert to Lgr5+ stem cells: a Paneth/enteroendocrine cell precursor [29]. This cell type is unique in its coexpression pattern, which consists of expression of Lgr5 and Paneth cell genes. Paneth cells constitute the only nondividing, long-lived cell population in the intestine. Consistent with this, it has been shown that Paneth cells can retain DNA label over prolonged time periods (Roth et al. 2012). Paneth cell precursors can due to their slow turnover (2–3 weeks) also retain DNA label, but are distinguishable from terminally differentiated Paneth cells by their expression of Lgr5. Winton and coworkers designed an elegant system to selectively induce lineage tracing in these Lgr5+ label retaining cells [29]. Under normal circumstances, Paneth cell precursors are not able to generate ribbons in a lineage trace, but do so after damaging the intestinal epithelium. In these situations, the Paneth cell precursors can revert to stem cells, capable of producing all differentiated cell types of the intestine (Fig. 1B). The proposed +4 cell markers Hopx, Lrig1, and Tert are (as described above) expressed throughout the lower crypt including the label-retaining Paneth cell precursor. This fact might explain the previous findings of lineage tracing from slowly dividing cells [78]. For example, Hopx lineage tracing results in 50% of the cases in labeling of single Paneth cells after a two month trace [18]. This finding indicates that Hopx marks, among other cells, Paneth cell precursors. These cells are not true stem cells, as they have a limited lifespan and are determined to terminally differentiate into Paneth cells. Yet, as a pool of cells they are resistant to damage and capable of replenishing the stem cell pool. Thus, they might be considered functionally to constitute a pool of “reserve stem cells”.

We propose that probably all progenitors that come into contact with Paneth cells for whatever reason, will inevitably become Lgr5+ stem cells (Fig. 3). In this model, intestinal stem cells and progenitors behave like yin and yang; they are inseparably interconnected, complementary and form a dynamic system around the Paneth cell.

Figure 3.

Plasticity of the intestinal stem cell. Lgr5+ intestinal stem cells can be depleted, e.g. using irradiation while Paneth cells are relatively resistant and unharmed by this procedure. Subsequently, niche space between Paneth cells becomes available for progenitor cells like the Dll1high or the Lgr5+ DNA label retaining cell. These cells convert back to Lgr5+ stem cells upon intimate contact to a Paneth cell. Sequential cell divisions of Lgr5+ stem cells produce daughter cells, which fill the empty niche space between Paneth cells.

Cancer Stem Cells and Their Niche in Intestinal Tumors

Research in cancer stem cells has been revived in recent years [79]. The “gold standard” to test whether certain tumor cells posses cancer stem cell properties is xenotransplantation of sorted subpopulations of tumor cells into immunecompromised mice [80]. Xenotransplantation requires the availability of specific antibodies to isolate the population of interest. For leukemic stem cells, the starting point for the detection of cancer stem cells were markers previously established to mark normal hematopoietic stem cells [81, 82]. Subsequently, cancer stem cells were also isolated from solid tumors including breast [83], brain cancer [84], colon cancer [85-88]. In contrast to leukemic stem cells, markers for normal stem cells in these tissues were not available. Thus, antibody selection was based on the capability of the antibody to differentiate subpopulations within the tumor, and not on biological reasoning. This might be one of the reasons why conflicting data subsequently emerged, negating the specificity of the used markers for cancer stem cells [89, 90].

Xenotransplantation assays also have the inherent drawback that human cancer cells are transplanted into hosts without an immune system and often into heterotopic sites. The transplanted cells thus end up in a completely different environment. Three recent articles have circumvented these problems by following the fate of individual cancer cells, two of them by genetic lineage tracing [91-93]. Earlier, it could be shown that Lgr5+ positive stem cells are very efficient in generating intestinal adenomas, while non-stem cells were not [94]. Additionally, Cd133 (Prom1) positive cells, expressed highest in Lgr5+ stem cells, also very efficiently transformed into tumors [95]. These findings established that the intestinal stem cell is the cell of origin of intestinal adenomas. Interestingly, both Lgr5 and Cd133 are not expressed in all adenoma cells, but only in a small fraction (7% for both Cd133 and Lgr5). This prompted further research into a possible hierarchy within adenoma cells with the idea that this rare Lgr5/Cd133-positive cell population could again function as a stem cell population, but this time in tumors. Taking advantage of a recently generated Cre multicolour reporter line (R26R-confetti) [31], which has the unique ability to be twice-inducible, this could indeed be established. When tracing in this line is initiated, the induced cell and all progeny are labeled with one of four possible colors. A second tamoxifen pulse can change this color into a different one, making it possible to retrace an already traced clone. Now, if adenomas are initiated in Lgr5+ stem cells by deletion of the Apc tumor-suppressor gene in Lgr5-ki/Apcfl/fl/R26R-confetti compound mice, the resulting adenomas are expressing one color. Included in this colored clone are Lgr5+ adenoma cells. Induction of retracing in Lgr5+ adenoma cells resulted in growing clones with a new color within the initially traced adenoma. Within these retraced adenoma clones differentiated cells such as TA-like cells and Paneth cell-like cells could be detected. This experiment established that Lgr5+ adenoma cells are again behaving as stem cells within the adenoma.

Two interesting points remain to be established: first, that Lgr5 negative cells within the adenomas cannot retrace. At least in vitro, Lgr5+ cells have a substantially higher capability to grow and form colonies than Lgr5 negative cells [61, 93]. Second, adenomas are only the first step in the development of colorectal cancer [96]. It needs to be established whether the same hierarchy exists in fully developed colorectal cancers, ideally by the methods applied by the studies cited here. Analyses of single cells isolated from human colorectal cancers established that a subset of cells within a certain tumor expressed a signature similar to the Lgr5+ stem cell signature [97]. Interestingly, other cells within the same tumor could be detected that express a signature similar to each of the differentiated cell types present in the normal colon. This is in agreement with a second study that found two complementary expression domains within colorectal cancers: one is expressing markers of intestinal stem cells and another is expressing markers of differentiated cells [98]. This heterogeneity within the tumor could be generated from a single sorted colorectal cancer cell, indicating that the heterogeneity within a certain tumor originates from the differentiation of a cancer stem cell and not from the existence of multiple divergent clones [88]. Cancer cells positive for intestinal stem cell markers also displayed the highest tumor-initiating potential in xenograft transplantation assays [98, 99]. Thus, although not genetically tested yet, these studies suggest a hierarchy within colorectal cancers, consisting of cancer stem cells and more differentiated progeny.

What induces this hierarchy in cancers? It has been shown that myofibroblasts can restore high Wnt activity and thus a stem cell phenotype in more differentiated colorectal cancer cells [99]. Similar to the situation in the normal crypt, the stem cell niche in tumors seems to be not only comprised by surrounding mesenchymal cells, but also by a subset of tumor cells. Lineage tracing within adenomas has unveiled the presence of Paneth cell-like cells next to Lgr5+ adenoma cells [93]. These Paneth-cell like cells migh indeed act as niche cells. Deletion of the E3 ubiquitin ligases Rnf43 and Znrf3, which normally target frizzled receptors, renders stem cells hypersensitive to Wnt signals. This results in an expansion of the stem cell compartment and ultimately leading to adenoma formation [100]. Adenomas induced in this way consist mainly of stem cells and Paneth cells. Inhibiting the secretion of Wnt by these Paneth cell-like cells in vitro resulted in the rapid death of adenoma cells. Thus, growth of these adenomas depends on the paracrine secretion of Wnts by adenoma Paneth cells. Although Apc mutant adenomas are independent of an exogenous source of Wnts, the close proximity of Lgr5+ adenoma stem cells to Paneth cell-like cells is pointing to an important role of these cells also for Apc mutant adenomas and potentially carcinoma growth. Of note, a certain niche factor can theoretically always be replaced by a specific, the niche factor mimicking mutation in the cancer stem cell. The moment, a cancer stem cell becomes independent of all niche factors, the tumor will undergo a final clonal expansion and consist of a homogeneous mass of cancer (stem) cells [80]. Until this ultimate step, of which it is unclear when and if it happens in colorectal cancer, as most if not all colorectal cancers show heterogeneity, cancer stem cells are likely to dependent on niche cells. This notion has important consequences for new cancer treatment strategies. If the cancer stem cell phenotype is not “hard-wired,” but similar to the normal crypt stem cells imposed by the environment, treatments targeting the cancer stem cell are condemned to fail. Rather, targets on niche cells could be the new “holy grail”.


We would like to thank Robert Siegmund for help with figure design.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.