Lgr5⁺ crypt base columnar cells (CBCs) are multipotent stem cells located in crypts of the small intestine and colon[14]. Lgr5 is an orphan G protein-coupled receptor expressed during embryogenesis and among epithelial stem cell populations in the adult intestine, hair follicles, stomach, mammary glands, and taste buds[18]. CBCs were first characterized in 1974 when an electron microscopy study identified a population of crypt cells that shared common secretory components with all differentiated epithelial cell lineages in the mouse intestine[19]. More recently, Barker et al[14] demonstrated that Lgr5-mediated activation of a permanent cell-labeling gene identified a line of cells originating from the intestinal crypt that yielded three differentiated cell types. The authors surmised that enteroendocrine cells were too rare to be detected among labeled cells[14]. A subsequent in vitro study demonstrated that organoids derived from single Lgr5⁺ cells form crypt domains containing all lineages of the adult intestinal epithelium including enteroendocrine and crypt paneth cells[20]. Taken together, these findings strongly suggest that multipotent Lgr5⁺ CBCs are true intestinal epithelial stem cells.

Quite contrary to expected stem cell behavior, evidence suggests that the expansion of Lgr5⁺ CBCs follows stochastic principles in which cells are equipotent and segregate chromosomes randomly[18,21,22]. Lgr5⁺ cells are also mitotically-active and demonstrate little asymmetric division[13,21]. Proliferation of these stem cells can at times approximate a square root growth curve, suggesting that they contain potential for rapid, yet very random clonal expansion[13,21,23]. As a likely consequence of their stochastic properties, Lgr5⁺ stem cells are subject to neutral drift, often resulting in monoclonal or oligoclonal populations in the intestinal crypt[21].

It seems dangerous for a stem cell to propagate in a manner dictated largely by chance. Random chromosomal segregation risks the introduction of genomic errors that can subsequently be passed to both daughters and self-perpetuating clones. Lgr5⁺ cells also seem to have little control over cell fate, suggesting that they are likely critically regulated by the surrounding milieu.

Quiescent DNA label-retaining intestinal stem cells (LRCs) have remained controversial since the 1970s when these mitotically-inactive cells were found at and around the +4 crypt position[24-26]. Although intestinal LRCs express a number of stem cell markers including Hopx, Tert, Lrig1, and Dclk1, they are widely identified by their expression of Bmi1, a member of chromatin-silencing polycomb-repressing complex 1[13,15,27]. Like Lgr5⁺ CBCs, Bmi1⁺ LRCs can form spheroids in vitro containing all differentiated epithelial cell types[13,20]. The multipotency of Bmi1⁺ LRCs has also been confirmed in vivo through lineage experiments[15]. In contrast to early reports of the radiation sensitivity of +4 position crypt cells, recent evidence suggests that quiescent stem cells are both resistant to and activated by moderate levels of radiation damage, thus suggesting a crucial role in recovery following intestinal injury[13,28]. Notably, Bmi1⁺ LRCs can single-handedly restore radiation-ablated mouse intestinal epithelium in the total absence of Lgr5⁺ stem cells[13].

Whether +4 quiescent LRCs are actually stem cells remains a matter of debate. Quiescent stem cells have only been found in the proximal small intestine and to date no presence has yet been found of a corresponding population in the colon[15,29]. Moreover, one study has identified quiescent LRCs not as stem cells, but rather as partially-differentiated secretory precursors[30]. Quiescent stem cell markers (including Bmi1, Tert, Hopx, and Lrig1) have also been found among Lgr5⁺ stem cells thereby questioning the validity of using such markers to identify a uniquely separate stem cell population[31].

In contrast to current single-lineage stem cell theories, the coexistence of two putative intestinal stem cell types may suggest a more complex pathway for the development of the intestinal epithelium (Figure 1)[10,32]. On one hand, evidence exists supporting the subordinancy of LRCs to LGR5⁺ cells: LRCs have been characterized as secretory precursors and may not share markers unique from Lgr5⁺ cells[30,31,33]. On the other hand, evidence also exists conversely that Lgr5⁺ cells may be subordinate to LRCs: Bmi1⁺ LRCs restore radiation-ablated Lgr5⁺ cell populations[13,29]. These findings when taken together suggest that LRCs likely interconvert with Lgr5⁺ CBCs, regardless of whether LRCs are actually stem cells. Such findings suggest that intestinal epithelial development is neither as hierarchical nor as unidirectional as once thought, though the extent of which is not yet known.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Origin and development of normal intestinal stem cells. Lgr5⁺ CBCs and +4 LRCs coexist in the crypt. Each stem cell is fully multipotent. Lgr5⁺ cells likely maintain intestinal homeostasis under normal conditions. Following intestinal injury, the reserve population comprised of +4 LRCs and Dll1⁺ secretory progenitors restore both the epithelium and Lgr5⁺ CBCs. Tuft cells are Bmi1⁺ cells that may be synonymous with or descendants of +4 LRCs. CRC: Colorectal cancer; CBCs: Crypt base columnar cells; LRCs: Label-retaining intestinal stem cells.

Based on the discussion thus far, perhaps the actions of the stem cell pool as we currently understand it are comprised of the combined properties of Lgr5⁺ and quiescent stem cells in the crypt (Figure 1). Under normal conditions, Lgr5⁺ stem cells could function to self-sufficiently maintain epithelial homeostasis through high-output cell production in response to trophic niche signals (e.g., Wnt)[34,35]. However, Lgr5⁺ CBCs are likely as sensitive to genetic damage as they are to injury. In these situations, the quiescent LRC population may assist with recovery from intestinal injury, either directly or by restoring Lgr5⁺ stem cells.

A multitude of signals in the intestinal crypt affect the function and growth of intestinal stem cells (Figure 2A and B)[37]. Of these, Wnt proteins are one of the most crucial for maintaining stem cell homeostasis[34,35,37,38]. Wnt promotes both cellular dedifferentiation and proliferation during embryogenesis and in many adult animal tissues[39-42]. Inhibition of the Wnt pathway results in crypt loss and a marked reduction in epithelial proliferation[43]. Among mice with inducible APC-knockouts, Wnt results in intestinal mucosa populated by undifferentiated cells[44]. Wnt activity is also among the essential signals for the formation of crypt structures from single stem cell cultures as well as for the reprogramming of somatic cells into induced pluripotent stem cells (iPSCs)[20,34,39,41]. Cell-proliferative genes are activated by Wnt via nuclear β-catenin intermediaries and include cell migration controllers (EPH), proliferative signals (c-myc, cyclin D1), and stem and cancer cell markers (Lgr5, Bmi1)[10,14,35,45-47].

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Niche regulation of the normal intestinal epithelium and colorectal cancer. A: Intestinal subepithelial myofibroblasts (ISEMFs) surround the crypt. Along with paneth cells (P), they supply the stem cell niche with trophic signals. Developing intestinal cells migrate upwards towards the villus apex, during which time they are subject to niches among the various strata in the crypt. B: Redundant mediators expressed by ISEMFs and Paneth cells contribute to the preservation of the stem cell compartment and normal intestinal proliferation. C: The local niche immediately around CRC likely fosters tumor growth by activating stem cell pathways. CRC cells lacking proliferative ability may re-awaken upon re-entry into the niche. CRC: Colorectal cancer; BMP: Basic metabolic panel.

The Wnt pathway is also a highly influential mediator of cancer (Figure 2C). APC mutations facilitate Wnt activity by dysregulating β-catenin-mediated gene expression[45,48]. APC mutations are common, occurring in over 80% of sporadic colorectal cancer[48]. Vermeulen et al[49] showed that primary spheroidal cultures derived from human CRCs are regulated by Wnt signals in the surrounding microenvironment, such as those secreted by intestinal myofibroblasts. They also demonstrated that extrinsic Wnt pathway activation was an important determinant in the cellular acquisition of cancer stem cell features (e.g., formation of tumors when injected into immune-deficient mice and in vitro recapitulation of xenograft isolate behavior to that of the original tumor)[49].

Intestinal subepithelial myofibroblasts (ISEMFs), located underneath the basement membrane in the crypt, are stromal cells widely known to promote stem cell self-renewal and differentiation (Figure 2A and B)[20,34,35]. ISEMFs originate from regional intestinal fibroblasts and possibly trans-differentiated bone marrow cells[50]. Intestinal myofibroblasts function as anchors for cell adhesion and provide trophic signals to stem cells via cell-cell interactions and secreted mediators[51]. ISEMFs also contribute to wound healing, mucosal protection, fluid and electrolyte transport, and growth of the basement membrane[50,52]. Secreted myofibroblast mediators are numerous: Wnt proteins, hepatocyte growth factor, fibroblast growth factor, TGF-β, keratinocyte growth factor, matrix metalloproteinases, stem cell factor, VEGF, and numerous interleukins, to name a few[52,53].

ISEMFs have long been implicated in promoting colorectal cancer growth and invasion (Figure 2C)[51]. Little clarity exists regarding whether peri-CRC myofibroblasts are derived from normal ISEMFs. Based on knowledge gleaned from other cancer systems, functional differences between normal and CRC fibroblasts do likely exist[54]. Still, even normal myofibroblasts are capable of facilitating CRC growth. Vermeulen et al[49] found that normal colonic myofibroblasts prevented both the morphological and molecular differentiation of co-cultured colorectal cancer cells. Furthermore, these myofibroblasts were shown to re-induce tumorigenic potential in subpopulations of CRC cells with low degree of proliferative activity[49].

Paneth cells are terminally-differentiated secretory cells intermingled between Lgr5⁺ CBCs at the base of crypts in the small intestinal mucosa[55]. Though unclear why no Paneth cells have been found elsewhere in the intestine, a population of c-kit⁺/CD117⁺ goblet cells in the colon may perhaps function analogously[33,56]. Co-culture of c-kit⁺ cells with Lgr5⁺ stem cells promotes the growth of organoids in similar fashion to those produced from Paneth/Lgr5⁺ cell co-cultures[55,56].

Paneth cells contribute to the preservation of the stem cell compartment through the expression of Wnt proteins and other secreted signals such as epidermal growth factor and Notch ligands, all important in the maintenance of the Lgr5⁺ CBC population[55]. Paneth cells also secrete antimicrobial peptides[57]. Furthermore, they facilitate epithelial repair by deactivating paneth-specific genes and converting to a phase that promotes Bmi1⁺ cell proliferation[58].

Paneth cells seemingly serve a redundant role in the intestinal crypt. Wnt proteins released from Paneth cells are also derived from other sources in and around the intestinal crypt[59]. Notably, the complete removal of paneth cells in mouse model systems has not been shown to affect the proliferation of Lgr5⁺ CBCs[60].

Is there a population of cells in the intestinal epithelium that reliably serves as the source for most, if not all of colorectal cancers? Intestinal stem cells are prime suspects due to their pre-existing proliferative and self-restorative behavior, making them perhaps more sensitive to overt carcinogenesis[10,35]. In support of this notion, Barker et al[61] demonstrated that APC deletions only among Lgr5⁺ stem cells (6.5% of tumor mass) promoted the formation of adenomas, even in the setting of uniform tumor Wnt target gene activation. Barker and colleagues concluded that Lgr5⁺ stem cell transformation-especially via loss of APC function-is a highly efficient pathway to neoplasia[61]. Multi-color reporter lineage retracing experiments by Schepers et al[62] have also confirmed that early adenomas are mostly of monoclonal origin, though occasionally oligoclonal. Schepers et al[62] also identified stem-like Lgr5⁺ tumor origin cells at the base of adenomas that shared organizational resemblances to normal stem cells and were 20-fold more efficient at forming cell colonies in vitro than Lgr5-poor cells derived from the same population.

Still, evidence suggests that colorectal cancer may also arise from non-stem cells, supporting the idea that ultimately any cell harbors the potential to foster neoplasia. Early observations by Cole et al[63] reveal that early adenomatous polyps are positioned at the top of colonic crypts without contact with the stem cell compartment. Schwitalla et al[64] have also demonstrated that Wnt-constitutive intestinal cells can re-acquire stem cell properties in an NF-_KB dependent manner and lead to tumor formation. These findings are congruent with iPSC research through which differentiated somatic cells have been reprogrammed back to proliferative stem-like states on account of key genetic alterations[41]. As with other non-intestinal cancers, no clear distinction yet exists identifying which CRCs, if any, are derived from non-stem cells[65].

What are the triggers that stimulate a cell to progress to cancer? Based on the discussion thus far, the neoplastic potential of a cell might be directly correlated with the combined disruptive impact of affected genes. However, One might imagine a situation in which a cell lacking sufficient functional derangement can be driven to cancer in response to external stimuli. Signals may come from cell placement in a Wnt-rich intestinal crypt, or in response to inflammation in light of concurrent genetic Wnt derangements as Schwitalla et al[64] have explored.

Not surprisingly, many normal stem markers such as Lgr5, DCLK1, CD133, CD44, CD24, and ALDH1 have also been found among highly proliferating fractions of colorectal cancers[10,52,66,67]. Given the apparent genetic heterogeneity among CRC[7,8], very few, if any, markers are both specific to CRC stem cells and ubiquitous among all CRCs[9]. Table 1 lists putative CRC stem cell markers as previously covered by other authors[10,68-71]. What remains unclear is whether such markers reflect carry-over from intestinal stem cell precursors as with other cancers (e.g., leukemia)[35] or else a re-activation of stem cell pathways. Regardless of the underlying reason, that CRC and normal intestinal epithelial stem cells express many of the same cell surface markers poses a challenge to the isolation of tumor stem cells.

One putative stem cell marker, Doublecortin-like kinase 1 (Dclk1), may be a useful marker for both normal and neoplastic intestinal stem cells. Dclk1 is a complex multi-splicoform transmembrane serine-threonine kinase involved in embryonic neuronal migration through intracellular signaling pathways[72,73]. In the digestive tract, Dclk1⁺ cells have been found in the stomach and at the +4 position of the intestinal crypt[74,75]. Intestinal Dclk1⁺ cells are functionally akin to quiescent stem cells via their label retention and radiation-induced activity[74,76]. Some studies contend that Dclk1⁺ cells are not intestinal stem cells at all. Dclk1 expression may be shared not only by stem cells but also among the enteroendocrine lineage[77]. Alternatively, Gerbe et al[78] propose that Dclk1⁺ cells are actually novel differentiated tuft cells with unidentified function.

Interestingly, cells aberrantly expressing Dclk1 have been found among both mouse intestinal adenomas and human colorectal cancers, suggesting a potential role for Dclk1 to identify neoplastic stem-like intestinal cells[74,79]. Nakanishi et al[80] recently demonstrated that Dclk1 specifically identifies abnormal intestinal mucosa found among tumors in the small intestine of APC^min/+ mice. Not only did Dclk1⁺ tumor cells co-express Lgr5, they also demonstrated higher expression of other cancer stem cell markers versus non-tumor cells[80]. Furthermore, ablation of Dclk1⁺ cells led to regression of the containing polyps without apparent effect to normal intestine[80]. These results concur with findings from our group showing that siRNA-based Dclk1 interference leads to growth arrest of xenoplanted CRC[81,82]. Also notable is a recent study by Li et al[67] demonstrating increased Dclk1⁺ expression among cell fractions with a higher percentage stem-like HCT116 human CRC cells. Taken together, these findings support the notion that Dclk1⁺ cells can identify colorectal cancer stem cells and that Dclk1 is critical for tumor growth.

Despite the strong evidence suggesting that only a small fraction of colorectal tumor cells is responsible for maintaining tumor growth, the isolation of “pure” colorectal cancer stem cells has remained an ongoing challenge due to numerous theoretical and practical reasons. In fact, the term “cancer stem cell” may be somewhat of a misnomer. There is no expectation that a dysregulated colorectal cancer cell follows the exact biochemical principles of a normal intestinal epithelial stem cell, even if they share common signaling pathways. So long as the phrase “cancer stem cell” is used loosely to refer to cells in control of the proliferative hierarchy demonstrated by CRC, there is no perceived problem.

The first studies documenting a tumor-initiating CRC subfraction came in 2007 with the identification of CD133⁺ cells comprising 2.5% of tumor mass[83,84]. However, the significance of CD133 as a specific CRC marker has subsequently been debated[52]. Other markers have further assisted in the enrichment of CRC stem cell fractions (Table 1). Kemper et al[66] found that Lgr5⁺ cells comprised only 1.9%-11.1% of putative stem cells already marked by Epcam, although admittedly the Lgr5⁺ fraction was more highly clonogenic. Isolation of DCLK1 among tumor stem cells has been previously discussed, but even Nakanishi et al[80] did not find DCLK1 universally among all tumors in their mouse experiments.

The current methods employed to identify CRC stem cells are derived from non-exclusive properties shared by all intestinal stem cells. These methods include: DNA label retention, in vitro and in vivo proliferation assessments, and detection of cell surface markers[10]. Consequently, the isolation of CRC stem cells is fraught with as much, controversy as normal intestinal stem cells. Not the least of which, subtle differences between humans and animal models may consequently make experimental findings difficult to generalize. The apparent genetic heterogeneity of CRC lends further worry that finding a universal identification standard for CRC stem cells may long remain a daunting task[7,8].

It is becoming increasingly apparent that both the normal intestine and colorectal cancer are subject to “plasticity” processes that convert cells back to less-differentiated forms. Conventional stem cell theory holds that cellular development follows a unidirectional and irreversible hierarchy through semi-differentiated intermediates and concludes with terminal differentiation[85]. The implied goal of such a model is to produce cells capable of specialized organ functions[86]. In the intestine, recent evidence has revealed that short-lived Dll1⁺ secretory progenitors can readily revert to Lgr5⁺ stem cells following radiation injury (Figure 1)[87,88]. The apparent conversion of quiescent Bmi1⁺ LRCs to Lgr5⁺ stem cells is another clear demonstration of cellular plasticity[30,31]. That differentiated somatic cells, too, can fate-reprogram into iPSCs carries profound implications regarding the exclusivity of stem cell traits and the potential for any cell in an organism to participate in tissue regeneration[41].

Cellular plasticity processes may also depend largely on the cellular microenvironment. For example, extrinsically-derived Wnt signals can sufficiently replace Myc gene mutations during iPSC creation[39]. Also, non-proliferating CRC cells possessing low Wnt activity have been shown to regain proliferative tumorigenic potential when co-cultured with colonic myofibroblasts or the conditioned medium derived from myofibroblast cultures[49]. These results indicate that extrinsic signals -notably activators of the Wnt pathway- are perhaps sufficient to induce behavioral reprogramming, especially in CRC[49].

That fate-reversal occurs in CRC suggests that CRC expansion adheres to a proliferative pattern somewhere in between the classical hierarchical and stochastic growth models[49,85]. Admittedly, however, it is not known to what degree cellular plasticity plays a role in the proliferation of colorectal cancer. Perhaps even among different CRCs there is variation in functional dependence on extrinsic signals, ultimately affecting the growth patterns and behavior of the neoplastic phenotype. In this way, perhaps an extreme disturbance of either genetic derangement or environmental signals alone would also be a sufficient trigger for carcinogenesis[36].