SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

Recently, there has been resurgence of interest in the question of small intestinal stem cells, their precise location and numbers in the crypts. In this article, we attempt to re-assess the data, including historical information often omitted in recent studies on the subject. The conclusion we draw is that the evidence supports the concept that active murine small intestinal stem cells in steady state are few in number and are proliferative. There are two evolving, but divergent views on their location (which may be more related to scope of capability and reversibility than to location) several lineage labelling and stem cell self-renewing studies (based on Lgr5 expression) suggest a location intercalated between the Paneth cells (crypt base columnar cells (CBCCs)), or classical cell kinetic, label-retention and radiobiological evidence plus other recent studies, pointing to a location four cell positions luminally from the base of the crypt The latter is supported by recent lineage labelling of Bmi-1-expressing cells and by studies on expression of Wip-1 phosphatase. The situation in the human small intestine remains unclear, but recent mtDNA mutation studies suggest that the stem cells in humans are also located above the Paneth cell zone. There could be a distinct and as yet undiscovered relationship between these observed traits, with stem cell properties both in cells of the crypt base and those at cell position 4.


Stem cells

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

Stem cells can be defined as cells capable of extensive replication and self-maintenance throughout the whole life span of an organism (in spite of physiological or accidental removal or loss of cells from the population) which are capable of multilineage differentiation and flexible use of these options. In terms of their recognition, it has been proposed that the gold standard of stem cell identification in a steady state should involve marking putative stem cells to identify their niche and then performing lineage tracing to demonstrate that the proposed ‘stem cell’ has multipotentiality (1). However, this does not allow for the flexibility seen after damage or which may be required in a steady state. Thus, stem cells are capable of producing progeny that undergo differentiation events, after or as a consequence of cell division, and these differentiated cells develop into functional end-point cells for the tissue. The progeny of stem cells may differentiate down a variety of pathways, giving rise to the concept of multipotency. If the tissue is injured and stem cells are killed or removed, any surviving stem cell may be capable of altering its self-maintenance properties to enable the tissue to re-establish a full stem-cell complement. However, stem cells may also be replaced by daughter cells re-assuming stemness or re-occupying the stem cell niche. Any change in self-maintenance properties rarely involves more than a subtle change in favour of stem cell numbers because any greater change would result in deficiency in the differentiated compartment. This ability to change self-maintenance properties leads to the association of stem cells with capacity for regeneration, as seen when a tissue is damaged. This regeneration process often gives rise to clones or colonies of cells (depending on the situation, in vivo, in vitro, etc.), which has resulted in these regenerative stem cells often being referred to as clonogenic cells or clonogens (2).

Differentiation should be regarded as a qualitative change in cell phenotype that is the consequence of activation of a new gene or gene sets. A clear distinction should be made between any cyclic activation of genes and gene products such as those associated with cell cycle activity and, on the other hand, permanent changes in gene expression or proteins associated with function. Maturation, in contrast, should be regarded as a quantitative change in the cellular phenotype (proteins leading to functional competence or differentiation products). These terms were defined in detail by Lajtha in the late 1970s (3,4) and elaborated in a detailed article by Potten and Loeffler in 1990 (5).

In most tissues, notably renewing ones, the initial differentiation event results in production of a dividing transit cell that may itself represent a lineage precursor cell or it may divide further (and differentiate further) to provide such precursor cells. These precursors are sometimes referred to as committed precursors or progenitors. In the small intestine of the mouse, such progenitors may also be long-lived (6).

The crypt is a unique cell biological system as proliferation (transit cell lineages), differentiation and cell migration are all distributed linearly along the long axis of the crypt. This linear arrangement can be seen when well-oriented longitudinal sections of crypts are studied. Good longitudinal sections can be identified by the presence of Paneth cells at the crypt base and the presence of part, or all, of the crypt lumen. In such sections, the linear aspects of cell proliferation processes here can be studied by numbering cell positions along the side of the crypt starting from the centre of the crypt base (2,7–9). The fourth cell position (cp4) from the base is the average position of the first non-Paneth cell, but the distribution of Paneth cells can vary considerably from section to section (see 10). The quality of data obtained using crypt cell position analyses depends strongly on the quality of the histology (orientation and crypt selection to ensure good longitudinal crypt sections).

Stem cells in the crypts

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

It is now widely accepted that a small number of stem cells is ultimately responsible for all cell replacement on a day-to-day basis in small intestinal crypts and that these cells cycle endlessly with a cell cycle time of approximately 24 h, that is, they enter mitosis once a day. They are capable of producing progeny that enter a dividing transit population, which differentiates to generate at least four distinct differentiated cell lineages [columnar cells, mucus-secreting goblet cells, enteroendocrine cells and Paneth cells; in some specialized regions of the gut, yet other cell types may be derived from these stem cells (for example, M cells).

Various uncertainties remain and these include:

  •  Precise number of stem cells per crypt and what controls their numbers.
  •  Precise location and what defines the niche in which they sit.
  • Their distinguishing characteristics (including surface markers).
  •  Their relationship and interaction with the environmental niche.
  •  Whether their stem cell characteristics are intrinsically determined or determined by the environment in which they are located.

The intestinal stem cell niche

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

The importance of the stem cell microenvironment (niche) in regulating functions of the progenitor cells in different tissues is now widely recognized (11,12). In the intestinal crypt, the niche is likely to be made of epithelial, sub-epithelial and luminal components. Mucosal constituents of the niche include the permeable basement membrane (to which stem cells and their progeny adhere) and cells located beneath it. The latter include mesenchymal cells (myofibroblasts, fibroblasts and smooth muscle cells), endothelial cells, neural cells, macrophages and lymphocytes. Cells of the niche may regulate stem cell function via secreted products and possibly also by direct interactions with stem cells via pores in the basement membrane. Epithelial–mesenchymal interactions appear to be critical for normal development of intestinal tissue architecture, especially for crypts and villi. Wnt, Hedgehog, bone morphogenetic protein (BMP) and platelet-derived growth factor pathways are involved in these interactions (13,14). Recently, there has also been significant interest in interactions between stem cells and subepithelial myofibroblasts and other mesenchymal cells in the adult intestine (15–20).

The Wnt family of secreted glycoproteins is believed to be important as signalling molecules that mediate their biological effects by binding to Frizzled receptors and low-density lipoprotein receptor-related protein. Intestinal subepithelial myofibroblasts have been shown to express genes of the Wnt family (15,21). Studies also suggest that circular muscle (which is separated from stem cells by a thin layer of basement membrane) in Drosophila midgut constitutes the intestinal stem cell niche (18) in this species. These circular muscle cells express Wingless (Wg, a Drosophila Wnt homologue) that acts directly on intestinal stem cells to promote their self-renewal.

Numerous studies have demonstrated the importance of Wnt signalling in regulating intestinal epithelial stem cell function (13). These include regulation of stem cell maintenance, proliferation and differentiation into Paneth cells (22,23), at the crypt base. Wnt pathways also regulate cell position via their effects on expression of ephrin B ligands and Eph B receptors (24,25). Eph/Ephrin molecules have been reported to define tissue boundaries and cell migration (26,27). Thus, in mice with disruption of EphB2 and EphB3 genes, proliferative and differentiated cells intermingle and Paneth cells do not migrate to the crypt base, but are located along the crypt and villus (25).

In contrast to Wnt signalling, the BMP pathway is a negative regulator of crypt proliferation (28). BMP2 and BMP4 are both expressed by mesenchymal cells and their receptor (BMPR1A) is expressed in the epithelium (29,30). Studies of murine intestine suggest that expression of Noggin in mesenchymal cells in the vicinity of crypt bases protects the epithelium from action of BMPs, thereby enabling cell proliferation to continue (13). In the human colon, BMP antagonists GREM1 and GREM2 are preferentially expressed in crypt myofibroblasts and smooth muscle cells (16). Moreover, in vitro studies have demonstrated that GREM1 partially inhibited differentiation of human Caco-2 cells of the intestinal epithelial cell line.

One challenge will be to understand not only the operational modes of niche processing and what regulates signalling, but also spatial constraints. This must imply biomechanical features not discussed so far. Crypt fission and adenoma formation are complex biomechanical events, which are linked to niche formation and function and hence, stem cell functioning. The link between cytoskeleton alteration and proliferation brought about by the cadherin-β-catenin complex may be important in determining niche effects.

Recent studies also illustrate the capacity of the stem cell niche to regulate asymmetric cell division (31,32). In the testis of adult Drosophila, hub cells constitute the niche for germline stem cells. Asymmetric division of these occurs by keeping the maternal centrosome permanently localized within the portion of germline stem cell adjacent to hub cells of the niche. This ensures that cell divisions occur perpendicular to the niche, leading to generation of one germline stem cell (that remains adjacent to hub cells) and one differentiating germ cell (that moves away from hub cells) from each division. Interestingly, new germline stem cells were also shown to be produced by dedifferentiation of spermatogonia. As outlined elsewhere in this article, early transit cells in the intestine are also able to function as stem cells when the latter are removed by irradiation (33). It will be of interest to determine whether the small intestinal stem cell niche regulates asymmetric cell division, as has been reported for hub cells in Drosophila testis.

Intestinal stem cell function may also be regulated by components in the crypt lumen, which may be derived from epithelial cells or from the large number of bacteria present in the lumen. Antimicrobial peptides of the alpha-defensin family are secreted by Paneth cells (34) and may provide protection for small intestinal stem cells against bacterial infection. The mucin glycoprotein Muc2 also appears to play an important role in regulating intestinal stem cell function. Muc2-deficient mice have been reported to display aberrant intestinal crypt morphology and altered cell maturation and migration, together with development of small intestinal adenomas and adenocarcinomas (35).

Of the luminal bacterial products, ligands for toll-like receptor (TLR) 4 and TLR5 have been reported to provide protection to the intestinal epithelium against radiation injury (36,37). Ligands for TLR4 and TLR5 are lipopolysaccharide and the bacterial protein flagellin respectively. Administration of TLR5 ligand has been shown to be radio-protective in mice, as illustrated by preservation of crypt stem cells (36). Further studies are required to determine whether the radio-protective effects of TLR5 ligand are mediated by direct interactions with epithelial cells (including stem cells) or via other cells in the intestinal mucosa.

The potential for another bacterial product, a toxin secreted by enterotoxigenic strains of Bacteroides fragilis, to regulate intestinal stem cell function, is shown by the capacity of the purified toxin to activate T-cell factor-dependent β-catenin nuclear signalling in intestinal epithelial cells (38).

Long-term monoclonality of crypts

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

There is accumulating evidence that intestinal crypts, both in animals and in humans, are clonal populations, ultimately derived during development from a single cell. The evidence for this comes, in the case of experimental animals, from analysis of allophenic, tetraparental mice (39,40) and mice heterozygous for a defective glucose-6-phosphate dehydrogenase (G6pd) gene carried on the X chromosome and randomly expressed after Lyonization (40). In the case of the mouse chimaeras, tissues from the contributing strains were distinguished either by immunohistochemistry for strain differences in the H-2 locus or, more satisfactorily, using a polymorphism for binding of the lectin Dolichos biflorus agglutinin (DBA), which recognizes N-acetylgalactosamine residues on blood group markers (39) and is defined by the Dlb-1 locus carried on chromosome 1 (41). In mice with defective G6pd gene, enzyme histochemistry shows the distribution of cells expressing normal G6pd (42). In both the small intestine and the colon of the adult animal, crypts are always monophenotypic; in chimaeras, crypts are always composed wholly of cells from one component mouse strain and similar findings in the G6pd heterozygote show that this conclusion is indeed valid and not because of selective segregation of like tissues, the so-called ‘chimaeric artefact’. These observations demonstrate monoclonality of intestinal crypt and gastric gland systems in the adult animal and also that all contained cell lineages are, ultimately, progeny of a single cell which presumably established the crypt in the first place. Thus, the stem cell repertoire includes all intestinal crypt and gastric gland lineages (43). However, it is important to emphasize that while this conclusion is appropriate for adult animals, neonatal crypts show mixed phenotype, but within 2 weeks, these crypts have become monophenotypic, by a mechanism yet to be explained (40). Thus, in the neonatal mouse, intestinal crypts are polyclonal but, through a process known as ‘purification’, crypts become monoclonal by 2 weeks of age and remain so throughout adult life (39,44), by which time the stem cell population will have expanded to the numbers per crypt in the adult.

A model in which stem cell behaviour can be studied has been exploited by Winton et al. (41), Williams et al.(45) and Park et al. (46). When mice showing uniform staining for DBA or for G6pd are given a single dose of mutagen in the weeks that follow, crypts appear which are apparently composed of cells with a different, mutated phenotype. In these experiments, there is induction of a rapid, but transient, increase in crypts, which show a partial, or segmented, mutated phenotype. Later on, there is an increase in frequency of crypts showing a completely or wholly mutated phenotype, an increase which levels off at the same time as partially mutated or segmented crypts disappear. It is interesting that the small intestine and colon show a major difference in the timing of these events: the plateau is reached at between 5 and 7 weeks in the colon, but not until some 12 weeks in the faster proliferating small intestine; this time is called the time to monoclonal conversion (47), the clonal stabilization time (48). The emergence of partially mutated crypts and their replacement by wholly mutated crypts can be explained by a mutation at the Dlb-1 or G6pd locus in a single stem cell from which all lineages are derived. Thus, the partially mutated crypts are crypts in the process of being colonized by progeny from the mutated stem cell and this crypt will ultimately develop into a wholly mutated crypt. Alternatively, some of these partially mutated crypts could derive from mutations in non-stem proliferative cells and these would, of course, disappear as the mutated clone was lost through migration out of the crypt.

The reasons for this difference in timing between the small intestine and the colon could be explained by different durations of stem cell cycle time (41), presence of a stem cell ‘niche’ with differences in number of stem cells between the two tissues (45) and the possibility that crypt fission plays an important part in the genesis of the wholly mutated phenotype (46). Loeffler et al. (47,49) have also formulated a model which explains this phenomenon on the basis of several indistinguishable stem cells per crypt, which can replace each other. The hypothesis here is that the predominantly asymmetrically dividing stem cells occasionally divide symmetrically (<5% of the time). The ‘extra’ stem cell produced would have to be removed to ensure stability of the crypt. This would be achieved by random loss of any one of the stem cells (via differentiation or apoptosis). In this way, mutations can be propagated or lost and monoclonality achieved over a long period. If the loss is achieved via apoptosis, this may account for the low levels of spontaneous p53-independent apoptosis, which has been suggested to be part of the stem cell homeostasis mechanism. But importantly, these experiments again indicate that a single stem cell can give rise to all crypt lineages, in both the colon and the small intestine.

It is clear that events in the human colonic crypt after mutagenic insult are similar: approximately 9% of the Caucasian population secretes sialic acid lacking in O-acetyl groups, in which case, colonic goblet cells stain with the mild periodic acid Schiff (mPAS) technique and negatively, or weakly, with techniques which show O-acetyl sialic acid, such as the periodateborohydride/potassium hydroxide saponification/PAS (PB/KOH/PAS) method. This is explained by genetic variability in expression of the enzyme O-acetyltransferase (OAT) (50). Nine per cent of the population is homozygous for inactive OAT genes –oat/oat. The Hardy–Weinberg equilibrium then predicts that some 42% of the population is heterozygous –oat/oat+, but O-acetylation proceeds as there is one active OAT gene. Loss of this gene converts the genotype to oat/oat. In heterozygotes, this shows as crypt-restricted mPAS staining in a negative background. This is indeed seen in about 42% of the population, and again is most simply explained by mutation or loss of the gene by non-disjunction in the single crypt stem cell and colonization of this crypt by clonal progeny of the mutated stem cell. This unicryptal loss of heterozygosity (LOH) occurs randomly, is increased by age, as would be expected (50) and is also increased in individuals who have received pelvic irradiation 1 month before colectomy, (48,51). Longer follow-up showed that clonal stabilization time (or time to monoclonal conversion) was in the order of a year, reflecting differences in stem cell kinetics between mouse and human. One year, and not 1 month, seems to be needed for all cells to be converted to the progeny of a single cell (48). Interestingly, this method shows no increase in somatic mutation in hereditary non-polyposis colon cancer (HNPCC), where defects in DNA repair genes occur. This indicates that there is no generalized increase in mutation in colorectal mucosa, but a second somatic event is required, leading to increased propensity for somatic mutation only in the involved crypt (52). Neither is there any difference in wholly mutated crypts between the left and right colons despite the fact that left-sided carcinomas are the most common, indicating that each side shares the same levels of lifetime-accumulated stem cell mutational load (53). This somatic mutation at the OAT locus suggests that human crypts may be clonally derived and indeed Endo et al. (54), using human androgen receptor analysis (HUMARA), have shown that individual, micro-dissected human colonic crypts appear homotypic and clonal in origin.

To investigate clonality directly in the intestinal mucosa, Novelli et al. (55) searched for tissue from human XO/XY mosaics. It is possible to differentiate between XO and XY cells using in situ hybridization with Y chromosome-specific probes and thus the Y chromosome can be used to determine patch size and tissue clonality, as in allophenic XX/XY chimaeric mice (56). The XO/XY phenotypic male patient was of short stature, but with no other stigmata of Turner’s syndrome. By coincidence, this individual also had familial adenomatous polyposis coli, with a frameshift mutation at codon 1309. Fluorescence in situ hybridization (FISH) showed that approximately 20% of peripheral blood lymphocytes were XO. Non-isotopic in situ hybridization (NISH) was performed on histological sections of small and large intestine, using Y chromosome-specific probes, and intestinal crypts were composed almost exclusively of either XY or XO cells. All indigenous epithelial lineages could be visualized as XO or XY, but a combination of immunostaining and NISH was required to show that crypt neuroendocrine cells shared the same genotype as other resident crypt cells. The patches of XO crypts were irregular in shape and patch size varied widely (mean 1.4 crypts, range 1–14). Percentage of XO crypts varied between sections with a mean of 9%. Crypts at patch borders showed no mixed XO/XY crypts; however, out of 12 000 crypts, four crypts were seen in an XY patch in which the XO cells formed one hemi-crypt, while XY cells occupied the other. Such a remarkable symmetrical distribution is very similar to the partial crypt changes described in the human colon with OAT loss (48) and in mice given mutagens, using markers for the Dlb-1 or G6pd locus (46). This rare event suggests a mutation in the crypt, with loss of the Y chromosome in one of the stem cells, although another possibility is that there has been failure of ‘purification’ of this single crypt.

Thus, the available evidence is overwhelmingly in favour of clonal derivation of colonic crypts. However, perhaps it is important to emphasize that establishing that crypts are indeed clonal does not imply that crypts contain but a single stem cell; in fact, the prevailing view, based mainly on radiobiological studies of crypt survival, indicates a multiplicity of stem cells, with perhaps as many as 4–16 actual stem cells (ASCs) and 30–40 potential stem cells (PSCs) per small intestinal crypt (5,57). It has been suggested that colonic crypts contain as few as three or four ASCs.

Thompson et al. (56) were the first to show that gastric glands are also clonal populations, using XO/XY mice and localizing the male nuclei with a Y chromosome-specific probe. They also showed that gastric neuroendocrine cells shared the same clonal derivation, possibly the ‘final nail’ in that portion of the APUD (amine precursor and decarboxylation) hypothesis proposing that endocrine cells are of neural crest origin (58). In the human stomach, the situation may be more complex in that while antral gastric glands are clonal in derivation, in body glands, foveolae and bases of the glands appear homotypic (59). The isthmus and neck zone, the area which contains the putative gastric stem cell population, is heterotypic on HUMARA analysis and thus may be of polyclonal structure, possibly reflecting a complex stem cell organization. It is, of course, possible that differences in timing of methylation might be important here, if they occur after the formation of the gastric gland primordia in humans (59).

Crypt base columnar cells are the stem cells

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

The controversy concerning location of stem cells in small intestinal crypts originated in a series of ground-breaking articles published by Cheng and Leblond in 1974 (60–63). The study was based on an analysis of some impressive electron micrographs and a simple lineage tracking study. In the last of the series of articles (64), they presented the Unitarian Theory of origin of epithelial cell types in the intestine. This theory concluded that all four differentiated cell types in the small intestine are derived from a small population of undifferentiated elongated intercalated cells located amongst the Paneth cells near the base of the crypt. These cells were called crypt base columnar cells (CBCCs). This hypothesis was rapidly accepted by the scientific community as it agreed with preconceived ideas of hierarchically organized cell populations in various tissues, and placed the intestine on a similar cellular organization to the more extensively studied haematopoietic system. However, data supporting this hypothesis were not strong. The study involved inducing cell death in proliferating cells in the crypt using internal β irradiation of DNA after incorporating tritiated thymidine ([3H]dT) (10 μCi/g). This DNA precursor is incorporated into DNA of S-phase cells. The very weak β particles internally irradiate the DNA inducing cell death. The dead radioactive cells fragment and these fragments are engulfed (phagocytosed) by neighbouring cells. This was recognized retrospectively as a type of apoptosis. These apoptotic fragments, some of which contained radioactive DNA, were incorporated into phagosomes in neighbouring epithelial cells, thus generating a radioactive cytoplasmic marker. The progeny of such marked cells could be followed and radioactive-labelled phagosomes were subsequently observed in three out of the four differentiated cell types of the crypt; only a single endocrine cell was found with an unlabelled phagosome. As the initial cell death occurred near the crypt base, it was concluded that this was the location of the cells from which everything was derived, that is, the stem cells. The study was performed using electron microscopy, a technique more suited to qualitative rather than quantitative observations; the result stated that the first phagosomes were observed in the CBCCs. However, the data presented actually showed more cell death occurring at cell positions 4–6 than at the crypt base (64) which is what was subsequently observed after external radiation of the crypt (65–67).

The above work illustrates one of the problems commonly faced by pathologists and anatomists, namely attempting to interpret a series of static images of different cells or tissues from different animals or subjects to explain dynamic processes. It is, to some extent, typified by figure 14 in Leblond and Cheng’s 1976 study (68), which shows a clear crypt base columnar cell with two early Paneth granules. This is clearly a Paneth cell precursor showing morphology similar to an intercalated CBC cell, but it remains unclear whether it is derived from an adjacent CBC stem cell or a stem cell situated elsewhere in the crypt (for example, at cp4) and has then undergone differentiation.

A later study from Bjerkness and Cheng led to the proposal that all differentiated offspring of the common epithelial stem cell originate in position 5 and above (69). Most columnar, mucous and enteroendocrine cells originate in positions 5 and above and migrate upwards. However, not only some progeny migrate downwards and into the Paneth cell zone as Paneth precursors but also some endocrine cells and goblet cells were found to migrate downwards, but the latter did not divide further. It was proposed that all undifferentiated cells found in positions 1–4 (intercalated in the Paneth cell zone) are stem cells (CBCCs), which proliferate, but do not differentiate in positions 1–4 except for Paneth differentiation (see above). The crypt base (cell positions 1–4) was referred to as the stem cell zone.

Recent studies from the Clevers’ laboratory (70,71), using one of up to 80 Wnt pathway genes, (Leucine-rich-repeat containing G-protein-coupled receptor 5), also provided evidence in support of the CBCCs being the actual functioning stem cells. Molecular function of the gene remains largely unknown, but it is expressed strongly in intercalated cells. Most of the intercalated cells are positive for Lgr5 and most were convincingly shown to be cycling with a cell cycle time of about 24 h. Using a tamoxifen-inducible CRE knock-in allele and the Rosa 26 Lac Z reporter, lineage tracking experiments were presented and Lac Z columns of positive cells were demonstrated at 5 and 60 days post-induction, in which all cell lineages were represented. The columns showed positive cells at all positions from the base of the crypt on to the villus with some scattered, apparently unstained cells, in the 60-day sample. The main difficulty with interpretation of these results is knowing whether the expression is variable (on and off, weak and strong) in individual cells with time and with differentiation/maturation/cell cycle status, and whether there is some level of expression in the cells at cell position 4. These considerations are complicated by possible variability in the half-life of the protein. It is difficult to tell from the pictures presented whether there might be some expression levels at cell position 4, but the quantitative data presented do show 10% positive cells at this cell position, a frequency which could be consistent with stem cell frequencies suggested by others (see below). A recent picture (figure 4b in 28) shows not only strong staining in the CBCCs but also weaker staining at cp 5–6 immediately above the Paneth cells.

However, these data do clearly show that some Lgr5 cells near or at the base of the crypt exhibit the ability to generate all intestinal cell lineages (multilineage differentiation). Sato et al. (72) from the same laboratories have recently shown, somewhat surprisingly, that single small intestinal crypts can be grown in culture and these undergo proliferation and crypt fission to produce organoids containing several crypt-like structures and even an intestinal-like lumen. These authors have also shown that single isolated Lgr5-positive cells can be grown to produce similar organoids over a period of about 2 weeks and that the cells can be subcultured. The surprising feature of these studies is that attempts to grow intestinal cells have been made, and largely failed, over the past 50 years or more. Perhaps the feature that provided success in these studies was the use of polymerized matrigel and EGF, R-respondin and Noggin plus the Rho-kinase inhibitor Y-27632 which reduced anoikis and cell death. It was also unusual that the authors could not grow colonic crypts, which have been grown by others with limited success (73,74). The final surprising feature of these studies was that this proliferation and spatial organization into organoids did not apparently require a mesenchymal niche. However, matrigel is a solubilized basement membrane preparation extracted from Engelbreth–Holm–Swarm (EHS) mouse sarcoma. Its major component is laminin followed by collagen IV, both of which are expressed in myofibroblasts (75). They also used Noggin, a BMP inhibitor, which is also expressed by mesenchymal cells (16). Matrigel also contains various growth factors (EGF, insulin-like growth factor and tissue plasminogen activator) as well as TGF-beta, which is expressed by myofibroblasts (76). Thus, it could be argued that although the physical presence of mesenchymal niche cells may not be required, their secretory components are necessary.

A recent study using immunohistochemical approaches to look at the distribution of Lgr5 in normal and pre-malignant human intestine (77) showed patterns in normal small intestinal crypts that were presented as supporting the CBCC stem cell specificity of Lgr5. However, the picture for the small intestine (Fig. 1c) showed weak cytoplasmic staining in the Paneth region and some clearly positive cells higher up the crypt. The authors pointed out that the number of positive cells varied and that many crypts were totally negative. Staining was seen of some cells in the lamina propria and the patterns of staining in adenomas were complex. Thus, the situation may be more complex than previously thought. Perhaps the actual stem cells can acquire Lgr5 for a time and then lose it later.

image

Figure 1.  Label retaining cells in small intestinal crypts. (a) BrdU pulse labelling showing the distribution of S-phase cells. (b) [3H]dT label retaining cell at about cell position 5. Paneth cells are clearly visible and centripetally located mitotic figures can be seen. Mice were irradiated with 12 Gy and then received 12-hourly injections of [3H]dT over days 1–6. The mice were killed and the intestines fixed and prepared for histology on day 21 (15 days and 15 divisions after the last thymidine injection). (c) BrdU label retaining cell at about cell position 4. Mice were irradiated as above and were given bromodeoxyuridine ad libitum in the drinking water for 6 days before being killed and treated as above on day 13 (7 days post-BrdU. (d) A further example of [3H]dT label retaining cells at cell position 4–5. I am grateful to Dr Kee Woei Ng for this photomicrograph.

Download figure to PowerPoint

Some of the difficulties encountered in the study are first, that the tamoxifen-induction process may vary from cell to cell (proximity of blood vessels, drug availability, intracellular levels, etc.) and secondly, there may be differences in the recombination kinetics in different cell types (CBCC and cp 4 cells for example). These effects could result in the patterns of expression seen after appropriate times, but would not necessarily define the cells that represent origins of the various cell lineages and differentiation events. The fact that there are 10% Lgr5-positive cells at cell position 4 suggests that the patterns seen for Lgr5 expression at various times after induction and the relationship to adenoma formation and the organoids in culture could all be derived from the 10% Lgr5-positive cells seen at cell position 4. For reasons that that are unclear at present (but presumably related to the as yet unknown function of Lgr5) CBCCs have high constitutive levels of Lgr5. Alternatively, the 10% Lgr5 cells at cp 4 could be expressing Lgr5 because of the length of the Lgr5 half-life.

A recent study (78) using Lgr5 null/LacZ-NeoR knock-in mice and studying early development of the ileum showed that Lgr5 becomes restricted to intervillus regions and that it leads to premature differentiation of Paneth cells, but no detectable effect on other cell lineages of the intestine or any changes in proliferation of migration. There was apparently inappropriate activation of the Wnt signalling pathway and the authors concluded that Lgr5 acts as a negative regulator of the Wnt pathway in the developing gut. The observations on Paneth cell differentiation without changes in other differentiation pathways would seem to suggest that Lgr5 is more closely involved in the Paneth cell pathway (possibly with Paneth precursors) than with multilineage differentiation associated with stem cells.

If CBCCs are the ASCs of the system, the predicted cell migratory pathways will be complex. Intestinal epithelium is a simple columnar epithelium, which effectively protects the body from hazardous materials in the external environment, that is, the lumen of the gut. This is achieved by maintaining close cellular interactions, tight cell junctional complexes and a requirement for controlled cell movements. The CBC stem cell theory, including the stem cell zone hypothesis, suggests that these CBC stem cells divide with a cycle of 24 h and produce progeny that move to cell position 4–5 where they continue to divide with a daily cycle. At cp 4, they subsequently generate the transit lineages that move from crypt to villus. Occasionally, a cell from cp 4 will move downwards to replace Paneth cells (a Paneth cell precursor). Indeed, after a single injection or continuous infusion of [3H]dT, the first labelled Paneth cells to appear were the highest Paneth cells in the crypt. Later, labelled Paneth cells became more numerous in lower positions and eventually appeared in position 1. Moreover, the size of granules in labelled Paneth cells increased with time. Bjerknes and Cheng proposed that Paneth cells originate in position 5 or above and then migrate downwards, again consistent with the stem cell zone hypothesis – stem cells in positions 1–4 receive no signal to differentiate and only those stem cells that migrate up out of the stem cell zone into position 5 will be so induced. These observations were confirmed by Clevers et al. (personal communication) where Lgr5 positive Paneth cells appear at the top of the Paneth cell zone and move downwards. Thus, in many ways, the recent data from the study of Clevers support the earlier stem cell zone hypothesis of Bjerknes and Cheng.

This model suggests turbulent (69) movement of cells around and between Paneth cells to reach first cell position 4–5, then to move back down as a Paneth precursor to replace lost Paneth cells either by further division or by differentiation (79) – a complex process bearing in mind the three-dimensional architecture of the crypt base, the shape and size of the Paneth cells and the necessity to maintain integrity of the epithelium. There is no direct evidence supporting such complex migratory activity in the first 1–4 cell positions i.e. the progeny of CBC stem cells migrating upwards, while Paneth cell precursors and possibly other cell types migrating downwards. This implies that cells move past each other in the opposite direction within the confines of the Paneth zone. It is perhaps significant that there are no images of the bipolar, spindle-shaped CBCCs (or their daughters) migrating or dividing (with a short cell cycle one might expect to see an occasional mitosis) or the occurrence of doublets. The only image of a cell that relates to these issues is the CBCC showing two early Paneth granules that suggest it is en route to being a Paneth cell.

Occam’s razor comes to mind here (William of Occam 1285–1349, a Franciscan monk who was a Philosopher of Science, proposed that if one has two competing theories, select the simplest: his ‘razor’ cuts through complexity to the simplest explanation). The simplest model proposes that the stem cells are located at cp 4–5 and they occasionally produce Paneth precursors that migrate downwards, but mostly produce precursors for all the other lineages that migrate upwards. If the CBCCs are not Paneth precursors, where are the precursors for these cells? If the CBCCs are the precursors, their cycle time would seem to be too fast for Paneth cells with their 2- to 3-week turnover (80).

Lgr5 crypt stem cells as the origins of adenomas

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

There has been considerable discussion concerning whether intestinal stem cells are the cells, which give rise to tumours. Certainly, in the small bowel, this seems to be so; it has recently been shown that deletion of Apc in Lgr5-positive cells leads to their transformation within days. Transformed stem cells remain located at the crypt base and lead to development of microadenomas, which develop into macroscopic adenomas within 3–5 weeks. A stem cell/progenitor cell hierarchy is maintained in early neoplastic lesions. However, in contrast, Apc deletion in transit-amplifying cells leads to very limited adenoma growth (81).

Similarly, prominin 1 (PROM1, also called CD133) has been extensively used to isolate putative cancer stem cells. Lineage-tracing in adult Prom1+/C-L mice containing the Rosa26-YFP reporter allele found that Prom1+ cells are located at the base of the crypts in the small intestine and these co-express Lgr5. These cells can generate the entire small intestinal epithelium. Activation of Wnt signalling in Prom1+/C-L mice by a tamoxifen-inducible CRE-dependent mutant allele of Ctnb1, then tracing the lineages using the Rosa26-YFP allele, resulted after 10 days, in streams of marked cells emanating from the crypts and at later times (60 days), in replacement of the mucosa of the entire small intestine with neoplastic cells with high-grade intraepithelial neoplasia and adenoma formation (82).

Hence, the CBCC Lgr5/prominin1+ cells appear to be the target cells for carcinogenesis in the small intestine, consistent with the ‘bottom up’ hypothesis proposed for adenoma formation in the human (83,84).

However, expression patterns and use of Prom1/CD133 knock-in mice showed that Prom1 mRNA was expressed throughout the lower half of the crypt and was not specifically associated with the Lgr5 cells, but may mark stem cells as well as early transit amplifying cells (CBCC as well as cp 4 cells?) (85).

Some cell position 4 cells as the stem cells

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

Now let us consider evidence in support of the concept that the stem cells are located at cell position 4–5. The first point to make in this context is that this model suggests that stem cells are located immediately above the highest Paneth cell for the small intestine, but, in other regions, may be located either at the very base of the crypt where no Paneth cells exist (for example, in the large intestine) or even at higher positions in some sites where there is clearly significant movement both up and down the crypts (or glands) as is seen in some regions of the stomach. However, Paneth cells have a very irregular three-dimensional profile when considering crypt circumference and in longitudinal sections of the crypt. Paneth cells may occupy one or two positions from the base, or as many as seven positions from the base (9,69,86). The average position of the first non-Paneth cell is cell position 4 or 5 from the base. The crypt is a flask-shaped structure with a hollow centre and is made up of a series of annuli of cells. With the exception of the first few cell positions, the annuli in mice contain 16 cells in circumference. The model proposes that between four and six cells in this annulus represent the ultimate stem cells. These could, in fact, be located anywhere between cell position 3 and as high as cell position 8 or 9 (see figure 4 in 10). With only 25% of the cells (that is, four stem cells) in this undulating ring of cells, at cell position 4, only occasional longitudinal sections will ‘hit’ a stem cell. Thus, for longitudinal crypt sections, many will not contain a stem cell; some will contain one stem cell on one side and a few will have one stem cell on each side.

The number of stem cells per crypt also remains somewhat uncertain. Complex mathematical modelling of the crypt (47,86–89) suggests that the numbers in steady state could be between four and 16 per crypt, but if the system is damaged, many more cells may be capable of assuming stem cell functions (see below and the mathematical modelling studies of Paulus et al. (90) and Gerike et al. (91)). Numbers close to one would seem unlikely as they would inevitably generate significant asymmetry in the crypt in terms of proliferative lineage and could result in crypt extinction when that stem cell was killed or deleted. Taking into account a wide and diverse range of observations, the most likely number would appear to be close to four per crypt. If the number of stem cells per crypt is four, random longitudinal sections through the crypt would only hit the stem cells approximately 25% of the time. Thus, any marker associated with stem cells, assuming it is expressed permanently, that is, not cell cycle regulated, would only show expression in occasional half crypt sections and even less frequently as single cells on both sides of the crypt section (in fact this is what is seen, at least in the proximal small bowel, if Bmi1 is used as a marker (92)).

Actual and potential stem cells

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

A further complication to the issue of stem cell numbers arises from extensive studies on the response of intestinal crypts to high doses of radiation and cytotoxic drugs (93–100). This rarely or never occurs in nature, but only in experimental laboratory conditions. It is clear that many more than four cells per crypt possess a regenerative potential. The best estimate for the number of these regenerative cells is about 30–40 per crypt (57,90,93,99–105). However, an alternative explanation for the survival curves from which these numbers were generated involved a second cytotoxicity component that reduced crypt survival further at high radiation doses. This model actually provided a better fit to the survival curves, but implied lower numbers of radio-resistant clonogenic cells (106). These observations along with many other studies (see 97) have led to the suggestion that the initial differentiation event that distinguishes non-stem cells from their stem cell parents occurs not at the time of the stem cell division, but a few generations down the transit lineage, giving rise to the idea of a stem cell hierarchy or age structure. This suggests that there are about four stem cells that perform the day-to-day function of cell production (the ASCs presumably associated with a specific niche) and up to 30 or 40 cells that can regenerate the tissue if it is experimentally damaged by high doses of radiation (the PSCs) (10,97,107,108) providing protection if all ASCs are killed (see Fig. 2). Figure 6 in Potten et al. (107) shows the theoretical spatial distribution of actual and PSCs based on mathematical modelling studies. Stem cells (actual and potential) could be found over the range of about 2–14 cell positions.

image

Figure 2.  Apoptotic cells at cell position 4–6 after exposure to 1 Gy of radiation. (a) Haematoxylin and eosin. (b) TUNEL staining. (c and d) Caspase-3 staining.

Download figure to PowerPoint

These data reflect recent studies in Drosophila, where in niches that contain multiple stem cells (such as those maintaining the germ cells) lost stem cells are replaced by division of neighbouring stem cells or by reversion of transit cells, even from some distance away (109). After experimental emptying of the germ-line stem cell niche, follicle cell progenitors, including somatic stem cells, eventually enter the niche and proliferate as long as the niche cap or hub cells remain. Thus, empty niches can signal incoming cells and support ectopic proliferation (110).

Although most crypts are uniform in size and indistinguishable from one another, one cannot rule out the possibility that there are occasional ‘master’ crypts with stem cell numbers and characteristics that differ from the majority of crypts. Some possible support for the master crypt concept comes from some studies involving repeated (6 hourly) injections of the S-phase-specific cytotoxic drug hydroxyurea (HU) (93). After 5–10 repeated injections, a plateau in crypt killing appeared to be reached with about 30% of the crypts being resistant to further HU doses (or other cytotoxics such as high doses of [3H]dT or cytosine arabinoside) (see fig. 3.23 in 93). However, these crypts contained highly potent clonogenic cells, which can restore the crypt population back to 70% of normal, at which time, further doses of cytotoxics can again reduce crypt survival back to about 30%.

image

Figure 3.  Schematic diagram of the lower region of small intestinal crypt in longitudinal section and a transverse section at about cell position 4. The position of the actual (ASC) and potential (PSc) stem cells is shown together with the possible relationship to the early transit generation (T1–T3). Some of the features and characteristics (based on various publications cited in the text) associated with ASCs are shown. The Paneth cell precursors (PCPs) may alternatively be regarded as ASCs, or part of the ASC population, and in the lower regions of the crypt (crypt base or cp 1–4) have been termed crypt base columnar cells (CBCCs). The pericryptal fibroblasts (PCF), which may undergo endoreduplication (160), could be important elements for the intestinal niche.

Download figure to PowerPoint

Evidence in support of cp 4 stem cells

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

The first observations suggesting an origin for all cell replacement in the crypt to be at cell position 4–5, were a type of cell migration tracking experiment (first performed by Cairnie et al. (7)), involving movement of [3H]dT-labelled cells up the crypt and on to the villus. In this case, living radioactively marked cells rather than marked phagosomes were tracked, as the dose of radioactive DNA precursor was much lower than in the Cheng & Leblond study. Initially, [3H]dT was taken up by S-phase cells, which subsequently divided and migrated. The number of radioactive-labelled cells was counted at each cell position generating a cell position labelling index distribution (111–113). The leading edge of these cell position distributions enabled a cell migration velocity to be determined (about 1 cell position per hour out of the crypt) and an analysis of the changing velocity with cell position, enabled the origin of this cell migration to be determined by back extrapolating velocity versus cell position distribution to cell position 4 (114). These observations were clearly consistent with the idea that all the cells of the crypt were ultimately derived from cells located at about cell position 4. A more detailed version of this type of primitive lineage tracking was performed by Qiu et al. (112) and Potten (113), in which case, the back extrapolate for all cell migration was to cell position 2–5 for the small intestine. This study clearly showed difference in the colonic region of the large intestine (no Paneth cells) where the back extrapolate was to the base of the crypt (cell positions 1–2). This migration velocity exhibited a strong circadian rhythm and at some times of day, there was essentially no movement or even negative movement (112,114). This ‘negative’ migration may be important as it provides an explanation for how the intercalated CBCCs might be produced from cp 4. Although the back extrapolates in these studies arrived at values that are consistent with some views of the crypt organization, this may be purely fortuitous as such back extrapolates have inherently large uncertainties.

A series of comprehensive cell kinetic studies (vincristine metaphase accumulation, continuous labelling and per cent labelled mitosis) all conducted on a cell position basis showed that cells at cell positions 4–5 had a cell cycle time of about 24 h in comparison to cells in the mid-region of the crypt where the cycle times were closer to 12 h (115–117). Continuous labelling studies suggested that most cells in the crypt were in cycle, that is, there were no quiescent cells present except differentiated Paneth cells. However, it is difficult to rule out totally the possibility that crypts contain one or two quiescent undifferentiated cells. Yet other studies have indicated that cells at position 4–5 show a cycle time more strongly linked to circadian rhythm than those in the mid-crypt region, suggesting that these daily rhythms originated from some daily synchronization of stem cell cycle activity as has been seen clearly in filiform papillae of the tongue (118).

There remains some debate concerning the issue of whether stem cells are intrinsically different from their transit progeny, or their stemness is derived extrinsically from their micro-environment/niche. It is possible to build up a dynamic single-cell-based spatial model of the crypt to accommodate most cell proliferation studies, clonal conversion studies and gain and loss of function manipulations, on the basis of cell–cell and cell–niche interactions, with essentially all the proliferative cells of the crypt being equal and their temporary actual stem cell behaviour being determined by their spatial position and the ultimate restriction of division potential being determined by some proliferation cut-off at a particular position (Loeffler, unpublished data). However, there are a number of diverse and independent observations that suggest that stem cells do indeed have some specific intrinsic characteristics (see below).

Under normal circumstances, proliferative cells at the top of the crypt undergo one mitosis before emigration from the crypt, as shown by grain density (in autoradiographs) of the migrating cells (113). One division is hardly a satisfactory criterion for stemness. The same argument can be applied to the next cell down, which might divide twice before emigration. The only cells to which this argument does not apply are those at the origin of all migration at the crypt base - cells which are anchored (in a niche) and which divide endlessly, that is, satisfy the criteria for stem cells.

There is a major distinction to be made here between what happens in a crypt under normal circumstances and what may be possible in very abnormal laboratory experimental conditions. It is quite clear that under some, often extreme, experimental criteria, cells can be demonstrated to be capable of amazing proliferative and differentiative capabilities that are never achieved under normal in vivo circumstances, particularly steady state conditions. If one is clever enough, and luck is on one’s side, almost anything can be achieved with a nucleated cell, that is, from crypt regeneration, to hair follicle reconstruction, to tadpole production and even to Dolly the sheep. Such remarkable end-points should, however, be interpreted with some caution.

Immortal strand retention

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

The most compelling of possible intrinsic specific stem cell features is the fact that a few cells at cell position 4–5 appear to be able to retain [3H]dT or bromodeoxyuridine (BrdU) label through many rounds of division (up to 10-15) are label retaining cells (LRCs) (Fig. 1), if they are exposed to the label when making new stem cells, for example, during development or regeneration after cytotoxic insult. The ability to generate LRCs has been repeated independently in two separate unpublished studies (see Fig. 1) and also has been reported by Demidov et al. (119). The ability to retain label in this way, or that a few LRCs can be seen at cell position 4, does not mean that they are quiescent cells as has been suggested recently in the literature (14), because it has been clearly shown that [3H]dT-LRCs can subsequently incorporate BrdU, that is, they are passing through the cell cycle once a day (120). Such double-labelled cells differentially segregate these two labels. BrdU disappears from the LRCs after the second division, while the [3H]dT persists through many cell divisions (120). These observations provide strong support for the immortal strand hypothesis proposed by Cairns in 1975 (121,122). This suggested that to minimize the risk of incurring DNA replication-induced errors in the crucial stem cells of tissues, these cells would have evolved a mechanism for selectively retaining the old, or template, strands of DNA in such daughter cells destined to remain a stem cell (by virtue of their niche or some intrinsic properties). Newly synthesized strands with potential replication-induced errors would be passed to the daughter cell destined to enter the dividing transit population and to be shed from the tip of the villus 3–5 days later. This process of selective segregation of DNA strands was convincingly shown for small intestinal crypts (120) and has subsequently been observed in breast cells in vitro (123), breast stem cells in in vivo fat-pad transplants (124), some muscle satellite cells (125), muscle progenitor cells (126) and some CNS cells (127), but apparently not in haematopoietic stem cells (128). It was also inferred from studies of tongue epithelium (129) and can be concluded to be the mechanism behind label retention in other tissues such as epidermis, but probably not the hair follicle bulge region. It indicates a unique and specific characteristic of small intestinal stem cells. It seems that this selective DNA strand segregation process does not operate in the large bowel - it has proved difficult to generate LRCs in the colon (20,130). This has been recently confirmed by some extensive studies using more appropriate radiation doses and labelling protocols for the large bowel than in earlier work (Gandara, Mahida & Potten unpublished data). For the ileum, average non-Paneth LRCs index at cell position 4–6 was 1.5% at 7 days after labelling and 1.1% at 15 days (12 Gy followed by 5 days of labelling starting at 24 h). For the mid-colon, LRCs index at cp 1–4 was 7.5% at 7 days, but had fallen to 0.2% by day 15. However, some other approaches have shown the presence of some long-lived BrdU-labelled LRCs in the colon (131). However, it is not clear whether these LRCs are stem cells.

The selective strand segregation hypothesis would suggest that sister chromatid exchange phenomena are prohibited in these cells (why mix the strands again?) (97,122,132) and as some of the enzyme systems for sister chromatid exchange and excision repair are common, it might be inferred that these cells would be very sensitive to DNA damaging agents such as radiation. This does indeed seem to be the case. A wide range of studies on the induction of apoptosis using small doses of radiation indicated that a few cells around cell position 4–5 were exquisitely sensitive to radiation damage and committed an altruistic form of apoptosis or cell deletion within 3–6 h of irradiation (see earlier) (Fig. 2). Doses of 5 cGy killed some of these cells and 100 cGy appeared to kill them all. As the numbers of LRCs and radiation-sensitive cells were similar, and they occurred at the same cell position, it was concluded that they represent the same cells and that they were indeed the ASCs (97). This is yet to be proven by looking for radiolabelled apoptotic cells after irradiation of animals with LRCs in a quantitative way, followed by demonstration of co-expression of some other stem cell-specific markers for which lineage determination has been seen (for example, Lgr5, Wip 1 phosphatase, Bmi 1).

Extreme sensitivity of some cells at cp4 to very low doses of radiation suggests a specific subpopulation of apoptosis-susceptible cells at this position in the crypt (Fig. 2). The most likely interpretation of these observations is that these are the stem cells that also selectively segregate their DNA. The apoptosis radiosensitivity of these cells has a circadian rhythm that can be related to cell-cycle sensitivities (133) with the greatest sensitivity (highest yield) in late S, G2 and M (03:00–09:00 hours) and the lowest sensitivity in G1 (18:00–21:00 hours). This circadian radiosensitivity is also seen for clonogenic cells (134). A dose of about 1 Gy kills all these apoptosis susceptible cells in every crypt, but 2 days later, they are re-established (33), presumably from more radio-resistant early transit cells (PSCs). These more resistant early transit cells then would have to restructure their molecular apoptosis mechanisms to become susceptible again. Apoptotic cells and fragments are engulfed by neighbouring cells that migrate with time, carrying the apoptotic fragment-containing phagosomes with them (see 64). These phagosomes are counted as apoptotic cells. After irradiation, dying cells fragment into between one and about 10 fragments and these have a half-life in the crypt of about 12 h (135). The precise distribution of apoptotic cells peaks at cp 4–5, but spreads up to about cp 20 at 4.5 h post-exposure (135). There will have been some movement in 4.5 h and the apoptosis distribution is greater than the theoretical distribution of stem cells. By measuring the movement of medians of the plots of the apoptotic yield versus cell position over time, it is possible by back extrapolation to determine the position of the apoptosis-susceptible cells at time zero (136). For internal and external irradiation, the apoptosis-susceptible cells are located at cp 4 and 5 respectively. Drugs such as isopropyl methane sulphonate (IMS), bleomycin (BLM) and adriamycin (ADR) also target cells at about cp 5 (136–138), while most other cytotoxic anti-cancer drugs target cells much higher up the crypt. For a range of chemical carcinogens, which also induce apoptosis to remove carcinogenic DNA lesions in the small intestine, target cells were also found to be at about cp 4 (cp 4.2 for ENU and NMU, 3.7 for DMH and 5.1 for NDMA (139), (N-nitroso-N-ethylurea, N-nitrosodimethylamine, 1-2-dimethylhydrazine and N-nitroso-N-methylurea). Corresponding positions of carcinogen target cells in the colon were 8.0, 4.9, 8.4 and 10.5 respectively, while the stem cells are located right at the bottom of the crypt (cp 1–2) (140). This is consistent with the concept that carcinogen target cells (CTCs) are the ASCs in the small intestine, which are protected from tumour induction in the small bowel by an altruistic suicide of cells bearing carcinogenic lesions, while the CTCs in the large bowel are not protected by this mechanism and hence can be the initiation point for cancers (140,141). Apoptosis in the large bowel is prevented by expression of the survival (anti-apoptotic) gene Bcl-2 (142). Bcl-2 is not normally expressed in the small intestine at high levels, however, recent data referred to above indicate that Lgr5/prominin 1+cells and Bmi1-positive cells can both lead to tumour formation in the small bowel. Cell position of apoptosis-susceptible cells in the small intestine, obtained by back-extrapolation of changes in the median of apoptosis yield at each cell position with time, and the back-extrapolate for movement of the leading edge of the labelling index distribution, is shown in Table 1. These data have inherently large confidence limits, but remarkably give a fairly consistent result.

Table 1.   Cell position for target cells in the small intestinal crypts based on back extrapolation of data obtained at various times
 Cell position
  1. Based on data presented in various publications cited in the text.

  2. It is surprising that with the high inherent confidence limits on such back extrapolation approaches these studies give such consistent results (overall mean cp 4.4)

Labelling index (cells at origin of cell proliferation, i.e. stem cells)4
2–5 (3.5)
Median of apoptotic distribution (apoptosis-susceptible cells)
 Cytotoxics
  Internal irradiation4
  External irradiation5
  IMS5
  BLM5
  ADR5
 Chemical carcinogens
  ENU4.2
  MNU4.2
  DMH3.7
  NDMA5.1

Some cells in lower cell positions (cp 4–6) of the crypt respond to induction of apoptosis in other cells in this region (after very low doses of various types of radiation, including high doses of β-irradiation from [3H]dT, and after very low doses of neutrons) by rapid and dramatic shortening of their cell cycle (143,144) suggesting that the stem cells are very sensitive to changes in their numbers and have a rapid response mechanism to compensate for that.

Other possible stem cell markers

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

Mushashi-1 is a gene that encodes an RNA-binding protein that was initially found to be associated with early asymmetric divisions in Drosophila sensory organ precursor cells (145,146). It is believed to influence Notch signalling pathways by suppressing mRNA needed for Numb synthesis; Numb protein is asymmetrically distributed in neural progenitor cells. An antibody to Mushashi-1 (Msi-1) has been used to determine its distribution in developing crypts, adult crypts and adenomas (147). This antibody positively stains a small number of cells distributed at cell position 4–5 in adult small intestine and a few cells at the base of the crypt in the large intestine (19,147). Kayahara et al. (148) have also demonstrated Msi-1 staining in cells at cp 4. They also observed staining in CBCCs. Hes-1, a transcription factor regulated by Notch signalling, was also expressed in the lower crypt, but with lower levels of specificity for cp 4.

Mushashi-1 is also strongly expressed in developing crypts (2-day-old mice), which would be predicted to be enriched with stem cells. It is also strongly expressed in post-irradiation regenerating crypts and in early adenomas: both are situations where stem cell numbers might be expected to be elevated (147). However, lineage labelling has not yet been demonstrated for this marker and other studies have shown its expression in terminally differentiated cells such as parietal cells (149). It has been found to be strongly expressed in Sertoli cells in the testes and in taste buds on the tongue (Potten, unpublished data).

Bmi-1 is a polycomb group protein involved in self-renewal in neural, haematopoietic and leukaemic cells (92). Using a tamoxifen-inducible CRE from the Bmi-1 locus, it has been shown that Bmi-1 is expressed in a small number of cells located near the bottom of crypt in the small intestine, predominantly at about cell position 4 from the base. Occasional CBCCs are also positive (92,150). Using an approach where tamoxifen-activated CRE recombinase was knocked into the Bmi1 locus, also produced long-lived clones populated by all four cell lineages. Moreover in this case, these Bmi-1-positive cp 4 cells do proliferate and expand, are self-renewing and give rise to all the differentiated cell lineages of the small intestine epithelium. As with the Lgr5/prominin1+ cells referred to above, induction of a stable form of β-catenin in these cells was sufficient to generate adenomas rapidly. Thus, these data support the existence, at least in the proximal small bowel, of self-renewing stem cells at cp 4 with multilineage capacity.

The microtubule-associated kinase, DCAMKL-1 (doublecortin and CaM Kinase-like1) is expressed in scattered cells around cp 4 in normal small intestinal crypt sections (151). Several examples of the staining pattern were presented in accurate, clean immunohistochemically stained sections. These cells were also Musashi-1 positive although the Msi-1 staining was not as cp 4 specific as has been reported elsewhere (147). Curiously, DCAMKL-1-positive cells were not the apoptosis-susceptible cells and early post-irradiation regenerative crypts showed no DCAMKL-1 staining. DCAMKL-1 expression was shown at around cp 4 by Giannakis et al. (152) who also noted MapK14 expression at this point in the crypt, which is linked to IL-6 signalling. However, the problem with these markers is that none of them has been proven to relate to self-renewing cells, which are capable of multilineage differentiation, in the way in which Lgr5/prominin 1 and Bmi 1 have been so demonstrated (see above). Even with the more promising markers, a potential problem is that neither function nor expression mechanism is as yet understood, in some cases making definitive conclusions difficult.

There is a large number of genes and their proteins associated with the Wnt and Hedgehog signalling pathways that are also being studied. These include the BMP and its type one receptor BMPR1A, phosphorylated (P)-phosphatase and tensin homologue (PTEN) (30) and phosphorylated-β-catenin, Wip1 phosphatase and phosphorylated Akt (P-Akt). All have been observed to have some level of specificity for small numbers of cells at around cell position 4 (14,153). Besides this, it has been recently shown that telomerase reverse transcriptase (mTert) as identified in GFP transgenic mice, stains a few cells at cell position 4(154). We have also observed occasional telomerase-expressing cells, using a telomerase anti-Tert polyclonal antibody, at cell position 4 (97). The selective DNA strand segregation hypothesis (121) would suggest that telomeres of the ‘old’ template strands would not be subject to telomere erosion. Thus, stem cells would not be expected to age as a consequence of telomere shortening. However, telomerase may be required to maintain telomere length in the immediate transit cell daughter and its subsequent progeny and may be important for counting the correct number of cell generations in the transit population, a crucial parameter determining cell output. Telomere length varies from species to species and mice have particularly long telomeres.

These results on expression of putative markers were summarized in a comprehensive review in 2008 (14). However, it remains unclear to what extent these represent stem cell markers, as some have a fairly broad expression pattern. The following genes have also been linked to the stem cell position (cp 4) or lower regions of the crypt using in situ hybridization: Apex1, Ascl2, Diap3, Gemin4, Rhobtb3, Sox4 and Wdrl2. (155,156). Gregorieff et al. (15) looked at the expression levels of all Wnts, Frizzled and their antagonists in the gut and found (using in situ hybridization) an interesting pattern of expression of Frizzled-related protein 5 (sFRP5) at about cp 4 in the small intestine and at the base of the crypts in the colon. Demidov et al. (119) presented extensive studies on label-retaining cells (LRCs), apoptosis and gene expression (in situ hybridization (ISH) for mRNA and immunohistochemistry (IHC) for protein expression). Expression of various markers showed some striking specificity for cp 4. Wip 1 phosphatase was expressed at cp 4 (ISH and IHC). It was co-expressed with phosphor-PTEN and was also expressed in LRCs. They also observed apoptosis with specificity for cp 4 and these apoptotic cells were shown to express phosphor-PTEN, Sox4 and Ascl 2. Furthermore, they showed that the apoptosis was associated with the LRCs. Studies in ApcMin mice showed a relationship between Wip 1 phosphatase and polyp formation (5-fold increased mRNA expression in polyps). Accepting the remarkably tight cell position distributions (much tighter that commonly observed), these studies clearly link Wip 1 phosphatase, p-PTEN, apoptosis-susceptible cells and LRCs to cell position 4 and to polyp induction.

Incidental observations have also shown some cells strongly expressing wild-type p53 protein at early times post-irradiation. These few cells occur at cell position 4 and are interpreted to be surviving regenerative stem cells (the early PSCs) that are using the p53/p21 DNA repair pathway prior to regenerative divisions to restore stem cell numbers. In similar experiments 24 h after a dose of radiation, which induces a cell cycle blockage, the earliest cells to enter DNA synthesis can be seen at cell position 4. They can be labelled with BrdU and are assumed to be the same regenerative stem cells (PSCs), presumed to have successfully passed through the p53/p21 repair process (97).

The issue of stem cell markers is difficult because it is not clear if they are secondary or indirect features related to a cell, or if they are primary features that determine cell function. Their expression may vary with time and with functional status of the cells (steady state, injury response).

The human small intestinal crypt

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

Little or nothing is known about stem cells in the human small intestinal crypt. However, very recent studies have taken advantage of the observation that mutations in cytochrome c oxidase, a component of respiratory complex IV, which is encoded by the mitochondrial genome, occur in intestinal stem cells (157). Multiple stem cells in the human small bowel crypt were demonstrated by identifying both entirely cytochrome c oxidase-deficient and also partially deficient crypts (158). It was shown that every cell within a mutated crypt is derived from a single stem cell and is clonal; laser-capture microdissected individual cells all the way up from the base of the crypt on to the villus contained the same mtDNA mutation, throughout the crypt. All lineages were mutated, which confirmed multilineage differentiation for the small intestinal stem cells. In mutated patches of crypts, every cell captured from a mutated crypt within a patch contained an identical mutation, while neighbouring cytochrome c oxidase positive crypts showed wild-type genotype – good evidence for the presence of a common stem cell. These crypts must have been divided by fission for a patch to develop.

In partially mutated crypts, stem cells and spread of their progeny can be mapped within individual human intestinal crypts. In the small intestine, however, a mutated clone appears to begin above the level of Paneth cells (fig. 3d in 159), with no mutated cells being seen within the Paneth cell zone itself, suggesting that in humans, the niche is indeed sited immediately above the Paneth cells, with no mutated cells between Paneth cells.

Conclusion

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References

It is important that any model to explain a stem cell population and its behaviour under a variety of conditions should accommodate as many available and diverse experimental data as possible, rather than (as is often the case) a single set of observations. There are now two excellent studies, which show that Lgr5-positive CBCCs or cp 4 cells show both self-renewal and multilineage potentiality. These cells have been shown to be the cells of origin in neoplastic transformation, and they can also be grown in culture to produce all cell lineages and spatially organized organoids. However, cells at position 4+ can also show lineage labelling, self-renewal and to be the origin of adenomas (Bmi-1 and Wip-1 phosphatase studies). Previous cell kinetic and radiobiological studies also support this viewpoint. The latter data would support the hypothesis that there is a small number of cells at cell position 4–5 that express a range of specific features, many of which can be associated with stem cells or stem cell function. These observations are interpreted by the model of a few ultimate ASCs in the undulating annulus of 16 cells that occurs at around cell position 4 to 5 with a larger population of PSCs occurring over cell positions 4–6. Theoretical distributions and percentages for the ASCs and PSCs in terms of cell position have been presented by Potten et al. (107) and span approximately cell positions 2–8 and 2–14 respectively. This concept has been repeatedly presented and refined over the last 10–15 years (10,97,107,108,113).

The model suggests that the undulating annulus of 16 cells at cell position 4 contains in the order of four ASCs that are responsible for all the day-to-day cell replacement. These cells must generate daughters that initially fill the cp 4 annulus and then predominantly enter the mid-crypt compartment. Occasionally, they produce a Paneth cell precursor and possibly precursors of other cell types, that migrate downwards. Both stem cells and Paneth cell precursors (PCP or CBCs) divide about once a day, but Paneth cell precursors divide a very limited number of times, perhaps only once. The four ASCs selectively sort their old and new DNA strands and if labelled with [3H]dT when making new stem cells (during development or post-injury) can become permanently labelled (LRCs). This is a way of marking these crucial cells. As a consequence of the DNA segregation process, these ultimate stem cells are exquisitely radio-sensitive and readily die when injured by radiation, via apoptosis. When all are killed in this way, early transit cells have the ability to ‘revert’ to function as stem cells, regenerate the stem cell compartment and re-establish the radio-sensitive compartment within 48 h (33). These early transit cells are relatively radio-resistant, use the p53/p21 repair pathway and enter their first regenerative divisions about 24 h after irradiation (see Table 2). Ultimate or ASCs may express a range of specific markers as described above (see Table 3). However, the rigour of these as stem markers remains to be tested and validated, especially with respect to the ability to label cells, which have self-renewal capability and multilineage capacity.

Table 2.   Some properties or attributes associated with stem cells at cell position 4 (based on various publications cited in the text)
Properties/attributesCrypt baseCBCC/PcP (ASc)CP4 (ASc)PSc
  1. CBCC, crypt base columnar cells; PcP, Paneth cell precursors; ASc, actual stem cell; PSc, potential stem cell; cp, cell position; LRCs, label retaining cells.

Cell cycle (h)24242412–24
Multilineage differentiation  
Selective DNA segregation   
Extreme radio-sensitivity   
Label retention   
Altruistic apoptosis (p53 dependent)   
Spontaneous apoptosis (p53 independent) (stem cell homeostasis)   
Circadian synchronization   
Carcinogenic target cells   
Adenoma formation  
P53/p21 damage response   
Telomerase/(hTERT)   ?
Table 3.   Some genes or molecules associated with the crypt base, crypt base columnar cells or cells at cell position 4 (based on publication cited in the text)
Genes/moleculesSelf-renewalMultilineage differentiationAdenoma formationCrypt baseCBCC/PcP (ASc)CP4 (ASc)
Lgr 5 ✓ (weak)
Prominin 1 (CD133) 
Ephrin B3   
Mushashi-1 (Msi-1) (Notch/numb/Hes-1)    ?
Hedgehog/Wnt/   ??
Frizzled signalling      
 Hes-1    ?
 Bmi-1 ?
 BMP     
 PTEN     
 sFRP-5     
 β-catenin/TCF/Lef/APC    ?
 Wip-1 phosphatase   
 P-Akt     
 Dcamkl-1     
IL-6 signalling/Mapk 14     
Apex-1   ???
Ascl-2   ???
Diap-3   ???
Gemin-4   ???
Rhobtb-3   ???
Sox-4   ???
Wdrl-2   ???

On the other hand, it must be conceded that CBCCs, which express Lgr5 and prominin 1, have been proven to self-renew and be capable of multilineage differentiation. These cells are also capable of neoplastic transformation. However, it remains unclear to what extent Lgr5/prominin 1 cells are exclusively located at the crypt base. It is possible that multilineage differentiation and association with adenoma formation could all be linked to the 10% Lgr5-positive cells at cell position 4. Prominin-related cells seem to be more associated with Paneth cell lineage than multilineage differentiation. Furthermore, Bmi1 was expressed just above the Paneth cells at around cell position 4, and that these cells are also capable of multilineage differentiation and self-renewal. The relationship between crypt base cells and cells at position 4 – both of which have been shown to have stem cell properties – including lineage labelling capacity, is intriguing and likely to be complex. Could they both be part of a more complex stem cell compartment? Thus, in the mouse small intestine, the exact location of the niche and its stem cells is still in dispute and further study must address this question.

Footnotes
  • A recent paper looking at DCAMKL-1 and LGR-5 in small intestinal crypts (161) has concluded that DCAMKL-1 is predominantly expressed in the lower regions of the crypt (49% Of the positive cells were at cp 4). These cells self-renew and can give rise to sheroids in culture and glandular structures when implanted subcutaneously in mice. The structures express many gut lineage markers (i.e. the cells exhibit multi-lineage differentiation). These DCAMKL-1 positive cells at cp 4 also retain BrdU label (i.e. are LRCs). The authors state that the cells are largely quiescent (PCNA negative) although there are many other studies showing cell cycle activity in this region of the crypt. In contrast, LGR-5 appears to be expressed in different cells at the crypt base (i.e. no co-localisation of DCAMKL-1 and LGR-5) and the LGR-5 cells may represent a different precursor population. The DCAMKL-1 positive cells do not express and other putative stem cell markers.

References

  1. Top of page
  2. Abstract
  3. Stem cells
  4. Stem cells in the crypts
  5. The intestinal stem cell niche
  6. Long-term monoclonality of crypts
  7. Crypt base columnar cells are the stem cells
  8. Lgr5 crypt stem cells as the origins of adenomas
  9. Some cell position 4 cells as the stem cells
  10. Actual and potential stem cells
  11. Evidence in support of cp 4 stem cells
  12. Immortal strand retention
  13. Other possible stem cell markers
  14. The human small intestinal crypt
  15. Conclusion
  16. Acknowledgements
  17. References