If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Stem cells are the crucial cells upon which the entire tissue is dependent. Here we define and discuss what is meant by and known about keratinocyte stem cells. One way in which these cells have been studied is by their ability to retain radioactivity labelled thymidine for long periods of time (label retaining cells, LRCs). The underlying mechanism has been assumed in the past to be slow cycling but a more likely explanation is the selective segregation of old and new DNA strands (Cairns's hypothesis). Experiments in the small intestine indicate that the stem cells here are selectively sorting their DNA and becoming LRCs. A possible role for p53 in stem cell biology is presented.
Skin has evolved to be a tough, flexible, protective barrier. The outer layers of the epidermis, the cornified layers, are composed of hard, thin, plates of keratinous material, sometimes referred to as squames, which represent the remains of individual cells which are constantly being sloughed from the surface. Beneath the cornified layers are several suprabasal strata of metabolically active but reproductively dead epidermal cells, which are tightly bound to one another, the lower layer of which, the spinous layer, has an intricate interdigitating membrane with many desmosomes (see Figure 4 in
) that bind these cells to the lowermost layer, the basal layer, within which cell proliferation occurs. In the mouse, the basal layer is relatively flat, in the human epidermis the undulations may be greater, and the number of nucleated cell layers in the spinous and granular layers may be greater. Beneath the epidermis, there is a tough, elastic, dermis that contains many collagen and elastain fibers. The skin may also be covered by hair or fur that provides thermal and UV protection and beneath the dermis there may be an adipose layer providing thermal insulation. The skin provides an effective protection again UV damage, and minor and major wounds. The large constant cell loss from the surface is compensated for in steady state by cell division activity in the basal layer. A question that has interested cutaneous biologists for many years is whether all the cells in the basal layer possess an equal potential in terms of proliferative activity, or whether sub-classes of proliferative cells can be identified.
The skin possesses a powerful, regenerative potential if injury occurs, and an ancillary question has been whether all the basal cells are equally efficient in their contribution to this regenerative wounding response. For many years, the basal layer was regarded as a homogeneous population with all cells being equipotential (Figure 1a); however, work in the early 1970s began to question this model and a more complex hierarchical, or lineage-based, concept became fairly rapidly accepted for the basal layer (Figure 1b). Here the basal layer was composed of a small number of self-maintaining stem cells that divided in an essentially asymmetric fashion, generating a daughter like themselves and a second daughter cell that entered a dividing transit population, a step that can be regarded as involving a differentiation event. In the dividing transit population, a series of cell divisions could occur before generating post-mitotic maturing cells that remained in the basal layer, until appropriate signals instigated the migration process from the basal layer to the suprabasal spinous layer. The number of generations in the dividing transit population is probably no more than three or four in mouse epidermis, but maybe somewhat greater in human epidermis where some cells of the last transit generation may be displaced into the first suprabasal layer. A fairly wide range of murine experiments provided evidence for this hierarchical organization. These ranged from classical cell kinetic experiments, such as the percent-labelled mitosis technique, analysis of clusters of tritiated thymidine-labelled cells with time, mathematical modelling studies, clonal regeneration following exposure to cytotoxic agents in vivo, clonal regeneration studies in vitro, studies of the retention of tritiated thymidine label in some basal cells when animals are labelled as babies and analyzed as adults (label-retention studies), considerations from chemical carcinogenesis experiments, spatial considerations and lineage tracking, studies into gap junctional communication and studies using transgenic models and, finally, concordance with other tissues, i.e., the acceptance of a unified scheme for cell replacement in all proliferating tissues.
One of the earlier suggestions of this type of lineage organization arose out of studies into the spatial organization of murine and rodent epidermis, conducted by Christophers and colleagues, Mackenzie and colleagues, and Potten and colleagues, in the early 1970s (
). These studies indicated that the cornified cells, the cells of the granular and spinous layers, were all arranged into discrete columns of cells, with their lateral junctions being coincident. These cells tended not to be circular piles of plates, but hexagonal in surface profile when viewed from above (see Figure 2c). The piles of hexagonal cells minimally overlapped with other hexagonal columns and the edges of the cells had a regular alternating interdigitation at the boundaries (see Figure 2b). These structural features implied a considerable organization in the basal layer, immediately beneath the columns, that insured a steady cell production (cell migration) that was coordinated from one column to the next. The implications were that each column represented the historical proliferative record of a small group of proliferative basal cells that lay beneath the column, and that these proliferative cells were a stable and permanent proliferative unit beneath each column. This led to the concept of the epidermal proliferative unit (EPU) (see Figure 2 and Figure 3) (
We can summarize the EPU in the mouse as consisting of about 10 basal cells with about 10 suprabasal cells arranged in a column, 3–4 of which were nucleated. About 1 cell per d enters the column and, hence, 1 cell per d leaves. The average cell cycle time in the mouse basal layer is about 100 h. Clonal regeneration studies following irradiation suggested less than 1 regenerative cell per EPU (
). This suggests the sort of lineage that is illustrated in Figure 3 or Figure 1b, with essentially three divisions in the dividing transit population but other arrangements such as that in Figure 1c are also possible. Various observations based on cell kinetic changes following injury, the position of label-retaining cells (LRC), morphology, and mathematical modelling, have suggested that the central region of the cluster of 10 basal cells is the most likely location for the single stem cell.
Since the first description of the EPU, there has been considerable debate as to whether this concept has any application to the human. This is somewhat surprising since the epidermis in the majority of the body sites (abdomen, limbs, dorsum, etc.) are all characterized by a similar columnar organization in the cornified layers. The difficulty here is that the boundaries of the columns cannot be traced as distinctly to the basal layer as they can in the mouse; however, the existence of this level of organization in the upper strata implies a similar organizational structure lower down. The columnar organization and hexagonal surface view are often remarkably similar in human and mouse (Figure 4). Furthermore, β-galactosidase-transduced cells that were cultured and then transplanted back into skin as xenografts for humans or ordinary grafts for mouse, generated positively stained unitary structures, essentially indistinguishable for the two species and in both cases images that were highly consistent with the EPU concept (
One issue that needs to be defined is the question of what is meant by the term “stem cells”. A search through the literature quickly illustrates the fact that the concept of stem cells is remarkably context dependent. In 1990, a colleague and myself attempted to define what we meant by stem cells in the rapidly proliferating tissues of the body but basing our definition primarily on epithelial systems (
). The definition that we produced stated that stem cells are relatively undifferentiated, self-maintaining proliferative cells, capable of many rounds of cell division, during which they produce a variety of progeny that may differentiate down various pathways. They are also capable of regenerating the tissue after injury when they usually adapt and vary the probability of self-maintenance. The definition includes the term undifferentiated, which again is a term subject to considerable misuse. We viewed differentiation as a permanent (not cyclical) relative change in the appearance or expression of some molecule in the cell that distinguishes it from other cells. The accumulation of more of this cellular product can be regarded as maturation. So a stem cell may be undifferentiated relative to other cells in the tissue but may, or may not, be differentiated relative to embryonic stem cells. Here we touch on the topic of stem cell plasticity. A subject of considerable current debate. This also leads to another area of potential confusion, concerning the ability of stem cells to differentiate. Stem cells can produce daughter cells that can differentiate down many different pathways. The best example here being the bone marrow stem cell, which can produce daughters that differentiate down many diverse pathways resulting in the stem cell acquiring the attribute of pluripotency. A strange concept, since the stem cell always remains as a stem cell by definition and it is some of its daughters that acquire the ability to express certain differentiation markers. This debate in terms of stem cell plasticity centers round the issue of whether the stem cells can be induced to generate daughters that make even wider ranges of differentiated progeny perhaps any, or all, of the possible differentiated cell types of the body, in which case the stem cells should be thought of as being totipotent like embryonic stem cells. We should perhaps think more along the lines of the experimentalist possessing a totipotent capability in terms of manipulation of stem cell progeny!
The next step forward in our understanding of keratinocyte stem cells arose from observations by Bickenbach and colleagues, and Morris and colleagues, in the early and mid-1980s that administration of tritiated thymidine to baby mice resulted in the retention of some of this radioactive label in a few isolated cells in the epidermis of adults (
). A wide range of subsequent studies suggested that these LRC were the slow cycling stem cells and various workers have subsequently attempted these approaches in other tissues. In the mouse, the studies involved a short series of repeated injections of tritiated thymidine into baby mice, with LRC being observed in the basal layer of the epidermis 3–12 wk later. As seen in sheets of epidermis, the LRC had a tendency to be located at the center of the EPU, where we believe the stem cells are located. They can be stimulated to divide by damage and they can also be shown to retain radiolabelled carcinogens for long periods of time (
). LRC have subsequently been seen in various regions of the hair follicle, particularly the upper outer root sheath, and this led to another significant advance in skin stem cell studies, namely the identification of some highly efficient stem cells in a specialized region of the outer root sheath called the bulge (
). The term “bulge” has led to some confusion since some small follicles such as the underfur (zigzag) follicles in the mouse lack a definitive enlarged area. The term has been applied to vibrissae follicles where there is a clear swelling in the upper regions of the follicle well away from the lower permanent areas of the follicle. It is often used to define areas that contain LRC, sometimes many more LRC than would be predicted on the basis of stem cell considerations. Those in the field should, perhaps, consider defining precisely what the bulge area is, how it is identified and how far up, or down, the follicle it reaches.
LRC can be seen in epidermis and in the hair follicle and for a variety of reasons it is believed that this approach provides a way of labelling keratinocyte stem cells; however, the underlying mechanisms involved here have not been extensively discussed or studied. The mechanism assumed by many to operate is that label incorporated into the DNA of stem cells in baby mice dilutes only very slowly because of the slow cell cycle time for these stem cells in the adult. But it is highly likely that in a 10–12-wk-old mice which contains LRC, the stem cells will have been through at least ten divisions, probably significantly more because the cell cycle time in the juvenile phase of skin development will be much shorter. This effectively rules out slow cell cycle as being the explanation.
There is commonly a somewhat puzzling use of the term quiescence for skin stem cells. If stem cells in epidermis or the outer root sheath are truly quiescent (i.e., non-cycling) then they cannot perform their function of producing cells for cell replacement.
An alternative mechanism for the label retention was provided by a hypothesis put forward by
, who suggested that stem cells would have evolved various protective mechanisms to conserve the genetic integrity of their DNA. DNA replication is probably one of the most hazardous events in the life of a cell. A simple and effective way of protecting stem cells against the risk of replication-induced DNA errors is for the cell to have devised a mechanism to segregate the old (template) from the newly synthesized DNA strands at mitosis, retaining the old template strands in the stem cells which would thus be genetically pure, and discarding the newly synthesized strands with any replication induced errors to the daughter cell destined for differentiation and cell loss. Recent evidence in support of this hypothesis has been obtained for the stem cells in the small intestine, which will be discussed later in this manuscript (
The next area to consider is the hair follicles. Hair follicles go through cyclic phases of active proliferation and hair production (the hair growth phase anagen) when the follicle is large with many cells. This is followed when there is a mature hair in the follicle, by a transition phase involving regression of the follicle, apoptotic events and a change in morphology (the transition phase being called catagen). This is followed by a quiescent phase when there is no proliferative activity to be observed in the lower regions of the much smaller follicle (the telogen phase). The hair growth cycle in mice is highly synchronized for the first two cycles, with the growth phase lasting 21 d and cell cycle activity in the germinal region being very rapid (cell cycle times of about 12 h). In humans the growth phase can last for many months and the cycle time in the germinal region of the follicle is probably about 36 h. There are a large number of cells produced per d and with this level of cell production and length of growth, the anagen follicle must contain some cells at the origin of all the cell movement that maintain the cell production at least for the duration of anagen, i.e., some self-maintaining stem cells. In some species such as Angora rabbits, Merino sheep, and some dogs, the hair growth cycle can be considerably lengthened and in humans it may last as long as a year. It is somewhat surprising that the lineage organization in the growing follicle has not been extensively studied but a cell lineage of the sort seen in the epidermis and in all other rapidly replacing tissues, including the small intestine, which is structurally somewhat similar, is most likely to apply in the growing hair follicle.
A major advance in our understanding of skin stem cell populations arose initially from early epidermal regeneration studies after various forms of injury (
), followed by elegant work from 1990 to today, looking at hair follicle formation and the activity of LRC in the upper outer root sheath under a wide range of very elegant follicle manipulation studies (
). These studies have shown that stem cells in the hair follicle bulge are remarkably potent reserve stem cells capable of forming hair follicles, sebaceous glands, and epidermis if those structures are injured; however, a question still remains, I believe, as to whether these bulge stem cells play any role in the normal hair growth cycles.
So with these amazing advances which have occurred over the last few years involving some very precise and elegant experiments, we need to consider the stem cells in the epidermis, the resting and growing follicle, and the questions of how these are all inter-related during development, in steady state, during the hair growth cycles, and following injury. In 1969, we undertook a fairly detailed resin-embedded serial sectioning study during all the transition phases from telogen to anagen (
). Our interests were to identify the precursor cells (stem or melanoblast cells) for the anagen hair follicle melanocyte population. By tracing back from anagen 5 to telogen, it was possible to see a clear sequence of events from melanotic melanocytes through to dividing dendritic clear cells, back to individual clear cells in the follicle germ of many telogen follicles (see Figure 5). The conclusion from these experiments was that each telogen zigzag follicle germ contained 1, or at the maximum, 2 melanocyte stem cells (melanoblasts). These cells were also of interest to Bill Montana at that time (dendritic clear cells). The conclusions that were drawn from our study were consistent with observations from a series of radiobiological studies where hair follicles were irradiated and the consequences analyzed for the follicular pigment cell precursor population
) using a lineage specific marker. Interestingly, they are located in the lowermost permanent portion throughout the hair cycle (see Figure 5). In zigzag hair follicles (80% of pelage hair follicles), the hair bulge area where the arrector pili muscle is attached corresponds to the lowermost permanent portion. In telogen, clear cells were previously found in the lowermost portion of the follicle, sometimes called the hair germ and this is located immediately above the dermal papilla (see Figure 7). Indeed, the process of stem cell activation and proliferation Nishimura et al showed is similar to the one described for clear cells in 1969 (Silver et al). Interestingly, Nishimura et al found additional melanoblasts below the bulge area (which they called the sub-bulge area) in the larger types of pelage follicles (guard or over hair follicles). This area often corresponds to the larger hair germ in the telogen (guard-hair follicles) (Nishimura, personal communication), which includes the cluster of cells found just below the attachment site of the arrector pili muscle in these larger follicles (
) and indeed generated pictures very similar to those in our 1969 paper (see Figure 5). At the time of our studies into hair follicle melanocyte precursors, we regarded the region beneath a hair club in a telogen follicle as the follicle germ. This was the area where the melanoblasts were located. These stem cells were responsible, through an amplifying division scheme, for generating the clone of 30 or so melanocytes of the growing follicle (
). Using mice having a Cre-recombinase construct with a keratin 15 promoter crossed with R26R reporter mice expressing Lac Z Morris et al described some lineage tracking of cells in the lower permanent region of the follicle (referred to as the bulge).
The studies using isolated putative Lac Z or enhanced green fluorescent protein (EGFP) positive cells have shown that these lower follicle cells are capable of forming new anagen follicles, sebaceous glands, and epidermis when injected into scid mice; similar conclusions could be drawn from studies when Lac Z is permanently switched on in the mice.
These studies clearly show the potency of these cells; however, questions remain about the number of putative stem cells (extent of Lac Z staining) in telogen follicles and the influence of other cell types in the cell extraction approaches used.
Intuitively, one would expect a telogen follicle to contain only a few stem cells (say 4–6 in total) situated adjacent to the dermal papilla which on division form a cap over the developing anagen dermal papilla. Only a few of these would be expected to be visible in sections. Any more than this would imply some redundancy and raise questions about the function of the additional cells, whether they divide and if so where their daughters are placed, and most importantly, where do all the positive cells end up in the anagen follicle and what happens to them at the next catagen. There is a difficulty in discriminating between blue staining of cells and the functional role of each cell in three dimensions and over time in a dynamic system like telogen–anagen hair follicle.
One of the major observations in the Morris paper was the identification of gene expression profiles for these putative stem cells which, among other things, provides a means of identifying and manipulating these stem cells.
The keratinocyte stem cell populations may be related in the way that is illustrated in Figure 6 and Figure 7. Diagrams similar to Figure 7 have been published before (
). What, in my view, remains uncertain is the amplification divisions between the hair follicle bulge reserve stem cells and the steady-state stem cells in epidermis and any stem cells in the growing hair follicle. It has been proposed by some that at each new hair growth cycle, stem cells or stem cell progeny migrate from the bulge region to the follicle germ, to initiate a new hair growth cycle, a complicated concept that involves difficult cell migratory processes, complex regulatory mechanisms, and possible stem divisional activity in a tissue that has evolved to be tough with extensive cell-to-cell adhesion. This clearly can happen if the follicle is injured but it seems unnecessarily complicated to evoke such a mechanism for the steady-state unperturbed series of hair growth cycles when the anagen precursor stem cell could reside in the telogen germ region. But I do note that in some recent publications the bulge concept has been expanded to incorporate all the permanent lower regions of the follicle, to include what was previously regarded as the telogen germ. (See also
.) So, there is an issue of how the bulge is defined. Is it a region of the upper outer root sheath beneath the sebaceous gland or is it the entire permanent region of the follicle that includes the follicle germ. If the latter, then the role of the upper outer root sheath cells in hair cycles becomes less clear.
The different stem cell locations suggested in Figure 7 imply specific local environments (niches) for the stem cells. In the EPU, this may be the adjacent, more immature dividing transit cells, with some ill-defined influence of the underlying dermal tissue. For the bulge, there is clearly also the possibility of a niche environment being supplied by the local keratinocytes (of uncertain lineage status) while, in the telogen follicle, a niche will be provided by the neighboring quiescent keratinocytes, together with an important and known influence emanating from the dermal papilla. In the growing follicle, the nature of a niche is less clear since the location of the stem cells is less clear. But it is also likely to involve neighboring immature transit cells and a dermal papilla component.
What can we learn from the stem cell organization in other tissues? A tissue with some similarity to the epidermis is the dorsal surface of the tongue. In mice this has been studied in some detail to define a tongue proliferative unit which is directly analogous to the EPU. Here the architecture is much more complex but the organization and lineage structure is essentially the same (
). The crypts in the small intestine are another tissue organized in a very similar fashion with cell lineages and lineage ancestor stem cells. Similar lineages have been described for the glandular elements of the stomach, the crypts in the large intestine and although the structural relationships between the lineages and the tissue are ill defined, similar lineage organizations are clearly present during spermatogenesis and bone marrow hematopoiesis. The thing that differs mainly in these tissues is the number of generations of cells that are seen in the dividing transit compartment. The number of generations determines the degree of amplification in cell production numbers each time a stem cell divides and it also inversely determines the frequency with which stem cells can be found in the proliferating regions of the tissue. These range from about 10% of the proliferating cells in some epithelia to a fraction of 1% in the bone marrow or testis.
One complication that needs to be considered in relation to stem cell populations is the point of commitment between the stem cell compartment and the dividing transit compartment. In Figure 1a–c, this is shown at the point of division of the stem cell which, as a consequence, is an asymmetric division; the two daughters differing at the point of division. But it is possible that this commitment may be delayed one or a few divisions down the cell lineage as illustrated in Figure 1d. This provides an hierarchy, or age structure, within a stem cell compartment and a version of this scheme is believed to occur in the bone marrow lineage and in the small intestinal crypt. Is this what we have in the skin? (see Figure 6). It is interesting to note that in Figure 1d the cells within the stem cell compartment are now dividing in an essentially symmetric fashion.
There have been some impressive advances recently, stemming from the work of Kaur and colleagues, and Morris and colleagues, in the identification and characterization of stem cells in epidermis and hair follicle (
). From these and other studies, it is possible to draw up a list of features that are associated with keratinocyte stem cells and these are shown in Table I. These recent advances are important in that they allow the stem cells to be identified, potentially isolated and, as a consequence, characterized better, leading to the possibility of manipulation and expansion of these cells in culture.
Table IProperties of some keratinocyte stem cells
Have a high colony-forming ability in vitro. Holoclones
Located in interfollicular epidermis, specific locations in a rete ridge configuration, in the bulb or matrix of growing hair follicles. In the germ of telogen follicles, and in the bulge or upper outer root sheath
Fixed cells (strong adherence or anchorage)
Long-term proliferative capability
Low proportion of cells in S+G2/M (long cell cycle) (quiescent?)
Present in total keratinocyte population at 4%–8+%.
Early lineage cells express α2β1 integrins
Strong adhesion to basal lamina extracellular matrix, type IV collagen, or fibronectin
Express keratin 19, keratin 15, CD34
Express integrin α6 strongly but weak expression of CD71
A cell surface proliferation marker recognized by monoclonal antibody 10G7, which turns out to be transferrin receptor. Information taken fromPotten and Booth (2002) but see alsoBraun et al (2003), Trempus et al (2003), Morris et al (2004). LRC, label-retaining cell; EPU, epidermal proliferative unit.
Retain 3HTdR label when treated as babies (LRC)
Small cells with a high nuclear to cytoplasmic ratio
The LRC can form colonies in vitro
are predominantly α6bright CD71dim
divide when skin is injured
tend to be at center of EPU
retain radiolabelled skin carcinogens
aA cell surface proliferation marker recognized by monoclonal antibody 10G7, which turns out to be transferrin receptor. Information taken from
The small intestinal epithelium is a model system that may provide some insight into the mechanisms underlying LRC, some confirmation of their stem cell status, and an indication of one way in which cancer induction is minimized in the epidermis. The small intestinal epithelium is a rapidly proliferating tissue with a high degree of structural polarity. Cell function is achieved by differentiated cells on finger-like projections into the lumen of the intestine, the villi, whereas cell proliferation to counterbalance cell death and loss from the villus is achieved by cell division activity in small closed proliferative units at the base of the villi, called crypts. Each crypt is a flask-shaped structure with about 250 cells in total, these being arranged as a series of annuli of about 16 cells. The bulk of the cells divide with an average intermitotic time of 12 h in the mouse, the same as the average cell cycle time in a growing hair follicle, which has a similar degree of polarity. It is now widely accepted that cell proliferation in the crypt is achieved within a series of about 5 cell lineages, each lineage with its own self-maintaining lineage ancestor stem cell. There are about 6 generations in the cell lineage in the small intestine, possibly 7 or 8 in the large intestine where similar numbers of lineages and, hence, lineage ancestor stem cells are believed to exist. In the absence of stem cell specific markers (although some are now in the process of being developed), the small intestinal epithelium represents a unique and valuable cell biological model system, since the position of a cell in the cell lineage can be directly related to its topographical position within the crypt. A wide variety of experiments have indicated that the lineage ancestor cells are to be found in the annulus of 16 cells positioned about 4 or 5 cell positions from the base of the crypt immediately above a functional differentiated component called the Paneth cells. The lineage that best explains the cell replacement process is the one illustrated in Figure 1c with the commitment to differentiation that separates stem cells from non-stem cells occurring not at the level of the lineage ancestor cell divisions, but 1 or 2 generations down the lineage. The cells at the stem cell location in the small intestinal crypts of mouse appear to have a cell cycle time twice as long as the majority of the crypt cells, i.e., they pass through a cycle once a day and, hence, in the lifespan of the laboratory mouse these cells may divide a thousand times. Alternatively, more complex models of the hierarchy can be envisaged with master stem cells dividing very rarely, producing progeny that divide rapidly before undergoing differentiation into the dividing transit compartment although there is no experimental evidence to support such a model. Once a cell leaves the proliferative compartment in the crypt, it has a life expectancy of about 3–4 d before being shed as a worn-out cell from the tip of the villus (see Figure 8 and
The small intestine of both mouse and man has a greater mass than in the large intestine by factors of between 3 and 4, the stem cells are proliferating 1.5 times faster in the small intestine, and there are two to three times more total stem cells in the small intestine than the large intestine and three to four times more total stem cell divisions in a lifespan. All of these numbers would suggest that the stem cells in the small intestine should have a significantly higher risk of incurring the sequence of genetic changes that lead to cancer than the stem cells in the large intestine. But the cancer incidence data clearly show that the small intestine has about 70 times less cancer than the large intestine, which implies that the stem cells in the small bowel are very efficiently protected against the genetic risks associated with cancer development.
In 1975, Cairns proposed that a very simple but totally effective way of protecting a stem cell against DNA replication-induced error would be to have evolved a mechanism (completely undefined or studied) for sorting the template strands of DNA which would be error-free from the newly synthesized strands of DNA that might contain replication-induced errors. The template strands would be conserved within the cell destined to remain as a stem cell, while the newly synthesized strands with potential errors would be discarded to the daughter cell destined for differentiation and loss from the tissue which, in the murine small intestine, would occur 5–7 d after birth from a stem cell (the villus transit time plus the transit time through the cell lineages). If this hypothesis is valid it should be possible to label the template strands at times when stem cells are making new stem cells and, hence, new template strands, ideally at the penultimate divisions before the stem cells assume their steady-state asymmetric division mode. Stem cells make more stem cells in two situations, one is during tissue development in the postnatal animal and the second is following cytotoxic insult such as radiation where stem cells are killed. Figure 9 shows that LRC can be observed in both these situations and that for the developmental scenario the optimum time for labelling is about 6 wk of age, after which LRC can be observed at 11 wk which represents a time interval during which the stem cells would be predicted to have undergone between 20 and 40 cell divisions. Also shown in Figure 9 is the appearance of these LRC and the fact that they appear with the highest frequency at cell positions 4–5 (the stem cell location). The post-irradiation regeneration situation tends to generate higher yields of LRC but they still occur with the highest frequency at the stem cell position (
). These LRC can be seen to undergo mitosis. Subsequent bromodeoxyuridine labelling (a series of BrdU injections) to a 11-wk mouse after generating LRC by labelling with tritiated thymidine at 6 wk or, a series of bromodeoxyuridine injections on day 8 after a dose of radiation when LRC were generated by tritiated thymidine administration over days 1–4, can result in labelling all the LRC with bromodeoxyuridine. This shows that all the LRC at the stem cell position are cycling. This allowed a more definitive experiment to be undertaken to address the issue of DNA strand segregation. In these experiments, which have been conducted more than once for both the juvenile labelling and for the post-irradiation regeneration labelling, LRC (tritiated thymidine) have subsequently had their newly synthesized strands labelled with BrdU. Immediately after the BrdU labelling, all of the LRC at the stem cell position are doubly labelled. Over a period of 2 d (two divisions since there is a 24 h cell cycle) the doubly labelled cells become singly labelled with tritiated thymidine which persists in the LRC for the duration of the experiment (up to about 10 d) (see Figure 10). These experiments strongly suggest, since the two labels segregate differently, that selective strand segregation as described by the Cairns's hypothesis must be operating. This, of course, has fairly major implications in several areas of cancer biology, cancer genetics and ageing (telomere shortening).
p53 is a tumor suppressor gene that can be regarded as the guardian of the genome (
) for adult stem cells and for embryos. Wild-type p53 protein keeps cells normal and the gene is mutated in many tumors. It can be shown to play an important role in deleting genetically abnormal embryos and cells in embryos (
). It is implicated in the initiation of apoptosis or in cell cycle arrest and repair processes. It may be implicated in the permanent cell cycle arrest (premature differentiation) post-irradiation.
Recent work using a fibroblast cell line, where p53 has been deleted but can be switched on by a zinc responsive promoter, has shown, using time-lapse studies, that when p53 is switched on in these otherwise p53 null cells, an asymmetric cell division mode is assumed with one daughter remaining proliferatively active and the other undergoing a differentiation event. It was subsequently shown that in these asymmetrically dividing cells, the DNA strands were asymmetrically distributed (using bromodeoxyuridine labelling and DNA quantification on cesium chloride gradients). There was also an asymmetric distribution of the bromodexoyuridine label in binucleate cells generated by cytoclasium D (
). What these studies suggest is that p53 may be involved in the asymmetric divisions that occur in stem cells and in the selective DNA strand segregation process that we have just described. The interesting feature illustrated by the small intestine is that some of these processes regulated by p53 differ in the stem cell and the dividing transit compartment. p53-dependent apoptosis appears to be exclusively a property associated with stem cells, whereas cell cycle arrest (G2 in these cells) is something exclusively associated with the dividing transit cells (see
). Asymmetric divisions and selective strands segregation may also be an exclusive property of the stem cell compartment and one regulated in some way by the p53 gene, see Figure 11.
A secondary corollary to the Cairns' hypothesis was that the stem cells would have to have a prohibition of sister chromatid exchange phenomenon since this would mix the strands. Such a prohibition may result in compromising excision repair processes, since some of the enzymes are common to the two processes and, as a consequence, the stem cells selectively segregating their DNA might be predicted to be extremely radiosensitive.
Apoptosis is easily recognized in small intestinal crypts both at the level of the transmission electron microscope and also in good-quality hematoxylin and eosin paraffin sections. In the mid-1970's (
) we instigated a program of work to quantitate the time course, the cell positional distribution and the dose response for apoptosis in the small intestinal crypts of the mouse. These experiments have been expanded and reviewed in several publications (
Following a dose of radiation, apoptotic cells can be observed in the crypt within a period of 1–2 h, with maximal values occurring at between 3 and 6 h, after which there is a steady return to control levels.
When the yield of apoptosis following a dose of radiation is studied on a cell positional basis, it is clear that the cells that die at 3–6 h post-irradiation occur with the highest probability at cell positions 4–5. It is not the rapidly proliferating cells that die via apoptosis, but a small cohort that are located at the 4th–5th position in the crypt, i.e., at the stem cell position. The majority of the rapidly dividing cells do not enter apoptosis particularly at early times, but appear to undergo a permanent cell cycle arrest (a process we refer to as premature differentiation,
The dose–response curve shows that these apoptosis susceptible cells at cell position 4–5 are exquisitely radiosensitive. Increases in the number of apoptoses per crypt above the spontaneous levels can be detected after doses as low as 0.01–0.05 Gy. The dose–response saturates at around 0.5–1 Gy at which point about 5 cells at the stem cell location die per crypt. This apoptotic response is independent of dose rate and linear energy transfer, suggesting a lack of an ability to repair the damage (
) and appears to be restricted to about 5 cells at the stem cell position. This suggests that these cells have an extremely rapid and efficient damage detection mechanism, which results in the activation of an altruistic cell suicide thus removing the damage. This of course is an efficient way of deleting any random errors that might occur in the template strands conserved in the stem cells.
Thus, the small intestinal crypts are protected by essentially three mechanisms. Firstly, selective retention of template strands in the stem cells eliminating the risk of replication-induced errors. Secondly, an altruistic apoptosis in stem cells that incur any random errors in their template strands and, thirdly, by the presence of a stem cell hierarchy such as illustrated in Figure 1d, so that if all stem cells are killed, their daughters simply assume stem cell status. The killing of all the stem cells is an unlikely scenario in nature but can be engineered in the laboratory.
These studies in the intestine indicate that LRC can be generated and that these cells appear to be selectively segregating their DNA. This suggests that the underlying mechanism for label retaining cells in the epidermis may also be selective DNA strand segregation. Some early studies implicated selective strand segregation also in the stratifying epithelium on the dorsal surface of the tongue (
), so this may be a general phenomenon associated with epithelial stem cells but it remains to be seen whether it applies in other epithelial and non-epithelial systems. Label retention provides a valuable new approach for marking stem cells in a semi-permanent way and this has been used to associate other characteristics with these crucial cells.
Stem cell biology is currently a rapidly expanding field and stem cells probably exist in all the tissues of the body. Markers for these cells are being developed and this will facilitate isolation, purification, characterization and will raise the possibility of selective stem cell expansion in vitro. The major issues that face stem cell biologists for the future are to identify the regulatory factors that determine whether epithelial stem cells proliferate (expand their numbers, or divide asymmetrically), remain quiescent, differentiate, or die. Ex vivo growth and expansion techniques need to be developed and optimized, and the issue of stem cell plasticity needs to be addressed. This is linked to the necessity of being able to control the differentiation routes down which stem cells or, more correctly, their progeny, move. It is also important to have a better understanding of the role of adult tissue stem cells in various diseases but most notably cancer. The final area for research over the next decade is the role that stem cells may play in the aging process of tissues. Do we age because of loss of stem cells or do we age because the existing stem cells lose full functional competence. With the rapidly expanding aging population of the world, this is going to become a major socio-economic problem for the future.
I should particularly like to thank Christine Sutcliffe for her help with this and many other manuscripts and reports. I am also extremely grateful to Dawn Booth for her amazing skill in powerpoint. Many highly competent scientists and technicians have worked with me over the years and I am indebted to them. I am also particularly grateful to Drs Ian Mackenzie and Emi Nishimura for permission to reproduce their figures and Emi Nishimura for helpful comments.
Al Barwari S.E.
Regeneration and dose–response characteristics of irradiated mouse dorsal epidermal cells.