Abstract
Recent lineage-tracing studies based on inducible genetic labelling have emphasized a crucial role for stochasticity in the maintenance and regeneration of cycling adult tissues. These studies have revealed that stem cells are frequently lost through differentiation and that this is compensated for by the duplication of neighbours, leading to the consolidation of clonal diversity. Through the combination of long-term lineage-tracing assays with short-term in vivo live imaging, the cellular basis of this stochastic stem cell loss and replacement has begun to be resolved. With a focus on mammalian spermatogenesis, intestinal maintenance and the hair cycle, we review the role of dynamic heterogeneity in the regulation of adult stem cell populations.
Introduction
In multicellular organisms, groups of cells specialize within tissues and organs to perform specific tasks and functions. In the course of adult life, these functional cells can become exhausted and progressively lost. To compensate for the ongoing loss of differentiated cells, new functional cells must be generated so that tissues remain in homeostasis. The maintenance and repair of cycling adult tissues usually rely upon the turnover of a small population of cells – termed adult stem cells – that possess the ability to self-renew, giving rise to differentiated cells while maintaining their number (Watt and Hogan, 2000; Fuchs and Chen, 2013).
The capacity of self-renewal has long been considered the defining feature of adult stem cells (Clermont and Leblond, 1952, 1953; Leblond and Stevens, 1948). To achieve homeostasis, stem cell proliferation and differentiation must be perfectly balanced, such that, following division, on average one daughter cell stays in the stem cell compartment, whereas the other differentiates either directly or through a limited series of divisions. Such fate asymmetry can be achieved as the invariant result of each and every stem cell division (termed ‘invariant asymmetry’). Alternatively, fate asymmetry may be orchestrated at the level of the population (termed ‘population asymmetry’), such that cell fate following each stem cell division is unpredictable or ‘stochastic’, and is specified only up to some defined probability (Klein and Simons, 2011; Simons and Clevers, 2011). These alternative models (Fig. 1A), both of which may be instructed by intrinsic (cell-autonomous) or extrinsic (environmental) cues, suggest very different regulatory mechanisms.
Proliferative hierarchies and patterns of stem cell self-renewal. (A) During tissue homeostasis, patterns of adult stem cell self-renewal can be grouped into four generic classes depending on: whether stem cell fate is regulated intrinsically (cell-autonomously) or whether it relies on extrinsic signals associated with the niche/microenvironment; and whether fate asymmetry is enforced at each and every stem cell division or whether it is achieved only at the level of the population (Simons and Clevers, 2011). (B) Traditionally, adult stem cell populations are thought to reside at the apex of linear (i.e. ‘one-way’) proliferative hierarchies in which they give rise to one or more types of transit-amplifying cell progeny with strictly limited proliferative potential. (C) Recent studies suggest a more flexible organization in which long-term self-renewal potential, fate bias and proliferative activity may be moderated by niche location and/or dynamical changes in transcriptional activity. In this scheme, stem cells form a ‘dynamically heterogeneous’ pool in which cells may transfer reversibly between ‘states’ of variable survival and fate potential. In addition, progenitors that are normally committed to differentiation may re-acquire long-term self-renewal potential in crisis or injury, following exposure to niche factors.
To address the factors that regulate stem cell self-renewal in adult tissues, attention has focused on defining the molecular mechanisms that control fate behaviour. By combining static marker-based assays with the transcriptional profiling of fixed samples, significant progress has been made in resolving key elements of the gene regulatory networks and signalling pathways that control stem cell activity and fate behaviour (Clements and Traver, 2013; Guruharsha et al., 2012; Holland et al., 2013; Laplante and Sabatini, 2012; Singh et al., 2012; Clevers and Nusse, 2012). However, it is becoming evident that stem cells function in dynamic and noisy environments in which levels of gene expression may adjust or fluctuate in response to promoter activity and extrinsic signals from the local microenvironment or niche (Morrison and Spradling, 2008). Therefore, to discriminate tissue stem cells from their more differentiated cell progeny and to define their functional behaviour, it is essential to address dynamic as well as static measures. Historically, the importance of such an approach was recognized prior to the genomics revolution. Charles Philippe Leblond, considered by many to be the father of modern stem cell biology, emphasized the ‘time dimension in histology’, and did much to advance early lineage-tracing methods using autoradiography and the incorporation of thymidine analogues (Leblond, 1965). However, it was not until the advent of transgenic animal models that it became possible to reliably trace the lineage of individual cells and their labelled progeny over time (Kretzschmar and Watt, 2012).
In recent years, pioneering studies using in vivo live-imaging platforms have begun to provide access to continuous-time lineage data (Bertrand et al., 2010; Boisset et al., 2010; Lam et al., 2010; Yaniv et al., 2006; Ritsma et al., 2013; Rompolas et al., 2012), whereas methods based on single-cell deep sequencing now offer the potential to resolve individual phylogenies, even in human tissues (Shapiro et al., 2013; Treutlein et al., 2014). By combining these lineage-tracing approaches with static marker-based assays, snapshots of clonal evolution over time can be integrated with population-level measures to reveal how stem and progenitor cells contribute to tissue maintenance. Efforts have also been made to develop statistical and mathematical methods that can resolve strategies of progenitor cell fate in development and tissue maintenance (Klein and Simons, 2011). When applied to actively cycling adult tissues in both human and model organisms, including the epidermis (Clayton et al., 2007; Mascré et al., 2012), oesophagus (Doupé et al., 2012), intestine (Simons and Clevers, 2011; Lopez-Garcia et al., 2010; Snippert et al., 2010; Hara et al., 2014; Amoyel et al., 2014) and germline (Klein et al., 2010; Sheng and Matunis, 2011), these studies show a preference for population asymmetric self-renewal, in which stem cells are continuously and stochastically lost and replaced by neighbours. This pattern of self-renewal results in ‘neutral drift’ dynamics, with the continual and stochastic loss of clones through differentiation compensated for by the expansion of others so that the overall stem cell population size remains constant. In some cases, these studies have overturned long-held paradigms and refocused the search for the molecular regulatory mechanisms that underpin stochastic fate behaviour. In particular, they have prompted the question of how the balance of stem cell proliferation and differentiation is regulated within dynamically changing environments (Simons and Clevers, 2011).
These studies have also begun to question our understanding of stem cell identity in adult tissues. When considering the identity of stem cells, two key assumptions are usually implicit, but rarely challenged. First, it is presumed that stem cells are defined by the signature expression of molecular markers, distinct from their more differentiated cell progeny. Second, in the course of tissue turnover, stem cells and their progenitor cell progeny are thought to move irreversibly through a differentiation hierarchy (Fig. 1B). However, with the advent of more-refined lineage-tracing approaches, both of these assumptions have been called into question. Increasing evidence suggests that expression levels of key fate determinants are not fixed, but drift over time or fluctuate in response to transcriptional activity and to extrinsic cues from the local microenvironment (Chambers et al., 2007; Chang et al., 2008; Pina et al., 2012; Levine et al., 2013; Imayoshi et al., 2013). Furthermore, progenitors expressing the same putative stem cell marker can exhibit heterogeneous fate choices (Graf and Stadtfeld, 2008), while stem cells with different expression profiles may behave similarly in the long term. Finally, recent studies in disparate tissues have also shown that cells normally committed to differentiation can, in the course of regeneration following the targeted ablation of endogenous stem cell populations, re-acquire the hallmark properties of tissue stem cells, including the potential for long-term self-renewal (Rompolas et al., 2012; van Es et al., 2012; Tata et al., 2013; Yanger et al., 2014; Tetteh et al., 2015).
Together, these findings question the traditional view of adult stem cell populations as discrete entities comprising functionally equivalent cells. Instead, gathering evidence suggests that, in some tissues, stem cells may transit reversibly between discrete or a continuum of states in which they become temporarily biased towards specific fates, but the final decision is made stochastically or governed locally by cell-extrinsic factors (Enver et al., 2009, 1998; Chalancon et al., 2012). In this way, a transcriptionally heterogeneous cell population may function, long-term, as a single equipotent stem cell pool (Fig. 1C). Here, we review case studies from three canonical cycling adult tissues types – the mammalian germline, intestine and hair follicle – that exemplify the role of heterogeneity and stochasticity in the regulation of adult stem cell behaviour, as well as the conservation of self-renewal strategies between seemingly disparate tissue types. These studies highlight the value of a multifaceted approach to the study of tissue maintenance that places emphasis on quantitative and dynamic measures of fate behaviour.
Examples of dynamic stem cell heterogeneity
Mammalian spermatogenesis
In mammals, spermatogenesis takes place in the seminiferous tubules of testes (Russell et al., 1993). In common with other cycling adult tissues, the testes contain adult stem cells – termed germline stem cells (GSCs) – that continually self-renew throughout adult life and are capable of rapid regeneration following injury. Throughout all stages of their development, germ cells are nourished by large somatic Sertoli cells, which support a network of tight junctions that separate the basal and adluminal compartments of the seminiferous tubule (Fig. 2A). Spermatogonia (mitotic germ cells that include GSCs) lie in close association with the basement membrane of the seminiferous tubule, and form the basal germ cell compartment. When meiosis begins, cells detach from the basement membrane, and translocate across the tight junctions. They then undergo meiotic divisions and spermiogenesis (Fig. 2B) before their release into the lumen as mature sperm. In mice, spermatogonia are subdivided into ‘undifferentiated’ and ‘differentiating’ populations, with the differentiated cells expressing the receptor tyrosine kinase Kit. Furthermore, undifferentiated spermatogonia can exist as singly isolated cells (termed Asingle or As) or as syncytial chains of cells connected by cellular bridges, consisting mainly of two (Apair or Apr), four (Aaligned-4 or Aal-4), eight (Aal-8) or 16 (Aal-16) cells (Fig. 2C) (de Rooij and Russell, 2000).
Stem cell dynamics during mammalian spermatogenesis. (A) Schematic showing the architecture and cellular organization of the mammalian testis. Spermatogonia lie in close association with the basement membrane of the seminiferous tubule. When meiosis begins, they detach from the basement membrane, translocate across the tight junctions between supporting Sertoli cells and undergo meiotic divisions and differentiation before their release into the lumen as mature sperm. (B) Spermatogonia progress through a differentiation hierarchy while migrating from the basement membrane to the lumen. (C) In the undifferentiated compartment, spermatogonia can exist as singly isolated cells (termed Asingle or As) or as syncytial chains of two (Apair or Apr), four (Aaligned-4 or Aal-4), eight (Aal-8) or 16 (Aal-16) cells. Undifferentiated spermatogonia are characterized by heterogeneous and complementary expression of GFRα1 and Ngn3, with As and smaller syncytial chains biased towards GFRα1 expression. Following upregulation of Ngn3, spermatogonia are competent to transfer to the differentiated Kit+ compartment, in concert with the periodic seminiferous cycle. (D) Whole mount (top panels) of a seminiferous tubule showing GFRα1 expression (magenta) and GFP-labelled clones (green) at 14 days post-clonal induction. The fragmentation of an Aal-4 syncytial chain results in two Apr chains. Reproduced, with permission, from Hara et al. (2014). Schematic (bottom) showing the cellular basis for germ line stem cell maintenance: chance stem cell loss through differentiation, signalled by the downregulation of GFRα1, is perfectly compensated for by stem cell duplication achieved through the fragmentation of neighbouring GFRα1+ syncytia. Through this ongoing process of stem cell loss through differentiation and replacement, stem cell-derived clones follow a ‘quasi’ one-dimensional pattern of ‘neutral drift’ where their chance extinction is compensated for by the expansion of neighbours along the seminiferous tubule.
In early studies, detailed analyses of fixed specimens led to the conjecture that stem cell activity is limited to the population of As spermatogonia, whereas interconnected Apr and Aal syncytia were irreversibly committed to differentiation, a hypothesis known as the ‘As model’ (Huckins, 1971; Oakberg, 1971). Consistent with this model, post-transplantation colony formation and regeneration assays confirmed that the vast majority of stem cell activity is restricted to the population of undifferentiated (Kit-negative) spermatogonia (Shinohara et al., 2000). More recently, the identification of genetic markers that are enriched in or restricted to As spermatogonia, including the transcriptional repressor ID4, the polycomb complex protein Bmi1 and the paired-box protein Pax7, allowed the potency of individual As cells to be assessed (Aloisio et al., 2014; Komai et al., 2014; Oatley et al., 2011). These studies confirm that at least a fraction of As cells retains long-term self-renewal potential, lending further support to the As model paradigm.
However, recent lineage-tracing studies have questioned the validity of the As model and offer a new perspective on the identity and function of adult stem cell populations in the germline and indeed in other adult tissues. These studies focused on the fate and behaviour of two separate compartments of undifferentiated spermatogonia, characterized by the expression of the glial cell-derived neurotrophic factor family receptor α1 (GFRα1) and the transcription factor neurogenin 3 (Ngn3). In undisturbed testes, these factors are expressed heterogeneously, with GFRα1 expressed more widely in As cells and shorter syncytia (Apr and a few Aal), whereas Ngn3 is expressed in a largely complementary manner (Fig. 2C). By developing an inducible genetic labelling transgenic mouse model based on the Cre-loxP recombination system with an Ngn3 promoter, the Yoshida lab showed that the vast majority of Ngn3-expressing cells proceed rapidly to differentiation, maturation and loss, but that a small minority of cells retain long-term self-renewal potential (Nakagawa et al., 2007). Furthermore, through the development of long-term ‘scaling’ properties of the measured clone size distribution, a follow-up study showed that GSCs are not individually long-lived, but are stochastically lost through differentiation and replaced by neighbouring GSCs, leading to neutral drift dynamics of the surviving clone size (Klein et al., 2010). Although these results are seemingly compatible with the As model, a subsequent in vivo live-imaging study by the same group revealed that the cellular bridges that connect cells within syncytia can break down, leading to the infrequent ‘fragmentation’ of Ngn3-expressing syncytia into single cells or shorter syncytia (Nakagawa et al., 2010). Such flexible behaviour of Ngn3-expressing spermatogonia questions the premise of the As model that syncytia are irreversibly committed to differentiation. Instead, these results suggest that the entire pool of undifferentiated spermatogonia may contribute to stem cell activity.
To address this issue, Yoshida and colleagues then combined detailed in vivo live imaging with long-term genetic lineage tracing using a pulse-labelling assay to follow the fate of individual GFRα1+ spermatogonia and their differentiating progeny (Hara et al., 2014). Continuous live-imaging data totalling more than 1 year of filming revealed that just 5% of GFRα1-expressing As cell divisions are complete, with the vast majority leading to the generation of Apr syncytia. Therefore, if the transition from As to Apr indeed signalled commitment to differentiation, as conjectured by the As model, the GFRα1+ As population would become rapidly depleted over time. However, alongside the cell division rate of around once per 10 days for GFRα1-expressing cells (independent of syncytial length), the live-imaging study also revealed fragmentation of GFRα1-expressing syncytia at a rate of around once per 20 days per interconnecting bridge, providing a possible route to replenish the As compartment.
Together, these findings suggest a revised model of GSC maintenance in which a morphologically heterogeneous cell population, comprising predominantly GFRα1-expressing spermatogonia (including As and syncytial chains), functions long term as a single stem cell pool. In this paradigm, germ cell production involves a coordinated process in which the commitment of cells to differentiation (signalled by the downregulation of GFRα1 expression and upregulation of Ngn3) is perfectly compensated for by the fragmentation of neighbouring GFRα1-expressing syncytia (Fig. 2D). To test this hypothesis, the measured rates of cell division and syncytial fragmentation were used to predict the medium (weeks to months) and long-term (months to over 1 year) clonal evolution of labelled GFRα1-expressing cells and their differentiating progeny. By collecting clone size and compositional data at single cell resolution, compelling quantitative evidence was obtained in support of the new model for germline maintenance. Through continual GSC loss and replacement, clones undergo a neutral drift process in which their chance expansion through syncytial fragmentation is perfectly compensated for by the contraction or loss of others through differentiation. At the same time, this study established a cellular basis to understand the process of GSC loss and replacement that was revealed by the long-term scaling behaviour of the clone size distribution reported in the earlier Ngn3 lineage-tracing study.
Although the cellular organization of the mammalian germline is of course unusual, these studies highlight several important features of stem cell dynamics that may translate to other stem cell-supported cycling adult tissues. First, maintenance of the stem cell compartment involves the continual stochastic loss and replacement of stem cells, leading to a progressive consolidation of clonal diversity. Second, stem cell competence is not restricted to a homogeneous cell population, defined by a signature expression of molecular markers. Instead, through the reversible transfer of cells between morphologically and genetically distinct states with differential survival probability, a heterogeneous population is able to function long-term as a single equipotent stem cell pool (see Fig. 1C). Third, it is only through quantitative analysis that ‘neutral’ competition between equipotent stem cells can be discriminated from ‘non-neutral’ clonal dominance associated with engrained (i.e. long-term) heterogeneity of fate potential. Fourth, the elucidation of short-term heterogeneity, and the cellular basis for stem cell loss and replacement, is facilitated by access to continuous-time in vivo live imaging.
As well as presenting a new perspective on the identity of GSCs and the cellular basis for stem cell self-renewal, these studies also raise new mechanistic questions. What is the molecular regulatory basis for stochastic GSC loss and replacement? What role is played by the periodic seminiferous cycle that orchestrates the progression of spermatogonia and spermatocytes through the differentiation pathway? How is the fragmentation of GFRα1+ syncytia so exquisitely correlated with the commitment to differentiation of neighbours when GSCs are separated by legions of differentiating progeny? And, functionally, given the singular role of the germline in the propagation of genetic information to the next generation, what are the implications of neutral drift clone dynamics for the inheritance of congenital disorders due to the acquisition and spread of de novo mutations in GSCs (Goriely et al., 2003; Giannoulatou et al., 2013)? But before speculating on potential regulatory mechanisms and their implications, it is instructive to look for parallels of these dynamics. To this end, we consider a second mammalian epithelial tissue that is characterized by a high degree of turnover.
Intestinal maintenance
The epithelium of the mammalian small intestine is organized into large numbers of self-renewing crypt-villus units (Fig. 3A). Villi form finger-like protrusions of the gut wall that project into the lumen to maximize available absorptive surface area. The villi are covered by a simple post-mitotic epithelium, beneath which capillaries and lymph vessels mediate transport of absorbed nutrients into the body. The base of each villus is surrounded by multiple epithelial invaginations, termed crypts of Lieberkühn. The crypts play host to a population of rapidly proliferating intestinal stem cells (ISCs), which fuel the active self-renewal of the epithelium throughout adult life (Clevers, 2013). These multipotent cells give rise to lineage-restricted transit-amplifying cell progeny, which migrate along the walls of the intestinal crypt and generate the various differentiated absorptive and secretory cell types.
Stem cell dynamics during intestinal maintenance. (A) Schematic showing the cellular organization of the mammalian small intestine. In adults, stem cells at the intestinal crypt base exhibit multi-lineage potential, giving rise to transit-amplifying cell progeny, which migrate along the walls of the crypt and differentiate into functional secretory and absorptive cell types. (B) On the basis of genetic lineage-tracing assays, the intestinal stem cell compartment has been associated with several molecular markers, including Lgr5, Bmi1 and Hopx, that are expressed heterogeneously within the crypt. (C) Time-lapse in vivo clonal data depicting the process of dynamic heterogeneity. Upper panels: following genetic labelling, a clone marked by RFP containing three Lgr5-positive cells (GFP) at the base of the niche expands over the next 3 days to occupy both border and niche base regions. Lower panels: a clone containing two Lgr5-positive cells all at the niche border expands to occupy both border and base regions. Reproduced, with permission, from Ritsma et al. (2014). Through this process of loss and replacement, stem cell-derived clones follow a quasi one-dimensional pattern of neutral drift in which their chance extinction is perfectly compensated for by the expansion of neighbours around the collar of the crypt, leading to scaling of the clone size distribution.
The identity, multiplicity and behaviour of ISCs have remained the subject of continuing debate and controversy. Beginning with the early pioneering studies of Leblond, which were based on incorporation of thymidine analogues, ISCs have been localized to the base region of the crypt (Cheng and Leblond, 1974a,b). However, these early studies could not assess the long-term potency or heterogeneity of individual crypt base progenitors. Subsequent lineage-tracing studies of individual clones marked by a chemical mutagen (ethylnitrosourea) showed that, in the course of turnover, the entire intestinal epithelium could be derived from a single marked cell (Winton et al., 1988). This result provided strong evidence that the intestinal epithelium is maintained by multipotent progenitor cells, which reside at the apex of a proliferative hierarchy. Later, with the advent of transgenics, long-term lineage-tracing studies based on the clonal marking of targeted cell populations identified several putative ISC markers, including the leucine-rich repeat-containing G-protein-coupled receptor 5 (Lgr5), the polycomb complex protein Bmi1 and the homeodomain protein Hopx (Barker et al., 2007; Sangiorgi and Capecchi, 2008; Takeda et al., 2011). These markers have been associated with different subpopulations of ISCs (Fig. 3B). Lgr5 expression is enriched in ‘crypt base columnar cells’, which lie interspersed between large mature secretary cells known as Paneth cells (Barker et al., 2007). By contrast, Bmi1 is more widely expressed in the crypt base region, with a peak of expression around the boundary of the Paneth cell compartment (row+4 from the base of the crypt), whereas Hopx is more tightly expressed in the same region (Muñoz et al., 2012).
Although these studies confirm that the ISC compartment contains cells expressing Lgr5, Bmi1 and/or Hopx, such qualitative studies cannot define the range, potential short-term heterogeneity and fate behaviour of ISCs. Once again, long-term lineage-tracing studies, allied with short-term in vivo live imaging, have provided the means to address the identity and functional properties of the ISC compartment. The first of these studies, a long-term lineage-tracing investigation based on the inducible genetic labelling of intestinal cells using a Cre-loxP recombinase system under the control of a ubiquitous promoter, showed that, in common with GSCs, ISCs follow a pattern of population asymmetric self-renewal (as evidenced by scaling behaviour of the clone size distribution) in which ISC loss through differentiation is perfectly compensated for by the duplication of a neighbouring ISC (Lopez-Garcia et al., 2010). Through this process of stochastic ISC loss and replacement, stem cell-derived clones undergo neutral drift dynamics, expanding or contracting around the crypt base until individual clones become lost, or the crypt becomes monoclonal.
Although this study provided insight into the functional behaviour of the ISC compartment, by focusing on medium-term (weeks) and long-term (months to 1 year) clonal dynamics, the size, molecular identity and short-term potential of the ISC compartment could not be resolved. However, subsequent pulse-chase lineage-tracing studies based on Lgr5 expression (Snippert et al., 2010), combined with studies of the colony-forming efficiency of Lgr5-expressing cells co-cultured with Paneth cells (Sato et al., 2009), led to the conjecture that stem cell competence may be linked to Wnt factors, which signal through Lgr5 (de Lau et al., 2011). Thus, through ‘neutral’ competition for Paneth cell contact following cell division, ISCs become displaced from the niche environment and enter into a differentiation pathway (Fig. 3C).
Although, in principle, the short-term potency of crypt base progenitors can be assessed through the use of targeted promoters, difficulties associated with the toxicity and delayed action of the Cre recombinase, effects of the inducing agent and the slow acquisition of fluorescent reporters make a definitive assessment problematic. Instead, to resolve potential heterogeneity of the stem cell compartment, medium- and long-term lineage-tracing assays were combined with short-term in vivo live imaging of clonally labelled tissue (Ritsma et al., 2014). By following the fate of marked Lgr5-expressing cells and their differentiating progeny over several days of time-lapse imaging, van Rheenen and colleagues showed that cells positioned at the base of the crypt (rows 0 to +2) experience a survival advantage over cells positioned near the border of the Paneth cell niche (rows +3 to +4). Yet, through the reversible transfer of cells between the border and base regions, the heterogeneous population of ISCs functions long-term as a single equipotent stem cell pool (compare with Fig. 1C) (Lopez-Garcia et al., 2010; Kozar et al., 2013). Whether the short-term potency of ISCs correlates with the expression of the putative stem cell markers remains an intriguing unresolved issue.
Together, these findings highlight the fact that, despite obvious differences in anatomy and cellular organization, the dynamics and fate behaviour of GSCs and ISCs, as well as the means through which they were elucidated, show striking and unexpected parallels. In both cases, the stem cell compartment is heterogeneous, with cells transferring reversibly between ‘states’ that are temporarily biased or ‘primed’ for proliferation or differentiation. Yet, once a clone comprising an individually labelled stem cell and its progeny reflects the composition of the heterogeneous stem cell pool (a situation that will prevail on time scales comparable with the time of transfer between different primed states), the subsequent clonal evolution will become statistically indistinguishable from that of an effective single equipotent stem cell pool (Fig. 1C). In this scenario, quantitative clonal analyses of both the germline and intestine reveal a process of stochastic stem cell loss and replacement that leads to neutral drift dynamics in which chance clonal contraction and loss is compensated for by the expansion of neighbouring clones (Klein and Simons, 2011).
Hair follicle cycling
We have emphasized the functional similarities of stem cell maintenance in the germline and intestine. But to what extent does their behaviour provide insight into other cycling tissue types? As mentioned above, quantitative clonal analyses based on genetic lineage-tracing approaches have provided evidence that population asymmetric stem cell self-renewal may be an ubiquitous feature of adult tissue maintenance, at least in actively cycling epithelial tissues (Klein and Simons, 2011; Simons and Clevers, 2011). In each documented case in which quantitative data on long-term clonal evolution have been available, their analysis is seen to be consistent with the steady turnover of an equipotent stem cell population. However, as illustrated by the examples of the mammalian testis and intestinal crypt, long-term steady-state behaviour may mask the presence of short-term dynamic heterogeneity and fate priming of the stem cell pool.
In this context, the mammalian hair follicle provides an interesting case study. The hair follicle is unusual as a cycling tissue because it is not generated at a constant steady rate, but undergoes periodic bouts of regression and regeneration throughout adult life. On the basis of label-retaining assays and lineage-tracing studies using targeted promoters (Tumbar et al., 2004; Cotsarelis et al., 1990), stem cell identity has been localized to a permanent and discrete region of the hair follicle known as the bulge (Fig. 4A). Otherwise dormant stem cells residing in the bulge region sporadically enter into the cell cycle in response to signals derived from the base of the niche, and give rise to progeny that repopulate the hair follicle. Alongside putative stem cell markers such as keratin 15 and the transcription factor Sox9, which are expressed throughout the bulge region, other markers are expressed heterogeneously, such as the hematopoietic progenitor cell antigen CD34, which is expressed more strongly in the distal region while Lgr5 is enriched proximally (Rompolas and Greco, 2014) (Fig. 4B).
Stem cell dynamics in the hair follicle. (A) Schematic of a mouse hair follicle during the resting phase of the hair cycle. Stem cells reside in the bulge and hair germ, whereas other epithelial cell populations occupy the compartments located above the bulge. Clones derived from stem cells situated in different parts of the niche have been observed to follow different fates, with cells in the hair germ primed for differentiation, while cells in the upper part of the bulge are mostly quiescent. (B) Hair growth and stem cell self-renewal in the bulge is driven by a heterogenous progenitor population. While keratin 15 and Sox9 are expressed throughout the compartment, CD34 is expressed more strongly towards the upper end of the bulge, and Lgr5 is enriched proximally. (C) Following ablation of either the hair germ (left) or the bulge region (centre), the entire niche is repopulated through proliferation and migration of the remaining stem cells. In response to ablation of the entire stem cell niche (right), non-hair epithelial cells migrate downwards to regenerate the stem cell compartment.
To trace the dynamics of hair follicle stem cells during the phase of regeneration, the Greco lab has recently employed a novel two-photon in vivo live-imaging approach, allowing deep penetration into the tissue (Rompolas et al., 2012). When combined with genetic lineage tracing, this method has enabled individual stem cell lineages to be followed from their exact place of origin throughout the process of regeneration (Rompolas et al., 2013). This study found that stem cells located in the upper half of the hair follicle niche were more likely to remain quiescent or to proliferate without committing to a specific fate (Fig. 4A). By contrast, stem cells situated in the lower bulge region were more likely to proliferate in response to activating stimuli from the niche base, undergoing limited amplification before differentiating. These observations suggest that, in common with the intestinal crypt, the location of a stem cell within the niche at the onset of a new regeneration cycle dictates its fate during the cycle (Rompolas et al., 2013). Whether stem cells in the lower bulge region continue to harbour long-term self-renewal potential, as do stem cells at the niche border of crypt, or whether they have irreversibly entered a differentiation pathway remains unclear. However, the flexibility and regenerative capacity of the progenitor populations have been tested under injury conditions.
Using laser-induced cell ablation to specifically remove either the bulge stem cells or the hair germ (stem cell progeny) at the onset of hair growth (Rompolas et al., 2013), further studies by the Greco lab showed that, remarkably, in both cases the hair follicle niche recovered the lost cell population, restored its anatomical features and proceeded normally through the hair cycle (Fig. 4C). Differentiating hair follicle cells can thus regain stem cell competence in response to injury. Surprisingly, distant epithelial cells located above the bulge were also observed to become proliferative, and some descended rapidly into the niche (Fig. 4C). By limiting genetic labelling to epithelial cells outside of the niche, it was confirmed that loss of the stem cell pool due to injury can mobilize cells that do not normally participate in hair regeneration to repopulate the niche and sustain hair growth. Indeed, once these cells entered the niche, they displayed characteristics consistent with the fate of endogenous stem cells in their new locations.
The hair follicle study therefore provides evidence for both stem cell heterogeneity and flexibility under conditions of stress. Although, in this case, the recruitment of differentiating cells to the stem cell niche has not yet been confirmed under normal physiological conditions, the conversion of epithelial cells to bulge stem cells in response to crisis suggests that cells seemingly committed to a differentiation lineage are able to ‘reprogramme’ and assume long-term stem cell fate identity. Future studies will reveal whether stem cell heterogeneity and the flexibility of differentiating progeny represent a more ubiquitous feature of this and other adult stem cell populations.
Questioning stem cell identity
The emergence of stochastic stem cell fate behaviour, stem cell heterogeneity and priming in tissue maintenance questions our understanding of adult stem cell identity and the definition of commitment. Even within an equipotent stem cell population, although all cells retain long-term self-renewal potential, chance stem cell loss and replacement mean that only a diminishing minority of clones actually persist long term. Yet it would make no sense to segregate cells prospectively according to their eventual long-term fate. Similarly, in a dynamic heterogeneous stem cell population, the long-term survival potential of individual cells may itself vary over time. For example, in the intestinal crypt, an ISC positioned at the border of the niche has a long-term survival probability that is several times smaller than an ISC positioned towards the crypt base (Ritsma et al., 2014). However, if the border ISC or its ISC progeny transfers to the base region, the survival probability is proportionately readjusted. It would therefore seem inappropriate to designate only the base population as a stem cell type; instead, the entire compartment functions as just one heterogeneous population.
Furthermore, in defining stem cell behaviour, much of the discussion in the literature has centred on the mode of division, and in particular on whether the fate outcome is symmetric or asymmetric (Watt and Hogan, 2000; Morrison and Kimble, 2006; Fuchs and Horsley, 2011; Watt and Huck, 2013). However, this designation is useful only if fate behaviour is defined shortly prior to, or on, division. If fate outcome is linked to the proximity of daughter cells to the niche following division, as implicated in the germline and intestine, the division mode may not be the primary determinant of daughter cell fate. In the search for a mechanism, it would therefore be expedient to focus more on local environmental cues that instruct fate behaviour.
The potential for ambiguity in the definition of an adult stem cell does not end there. Alongside the innate regenerative capacity of the endogenous stem cell population, evidence from regenerative studies of hair follicles shows that cells normally committed to differentiation in steady state are able to repopulate the stem cell compartment and re-acquire long-term self-renewal potential in response to injury or stress. Indeed, such behaviour is far from unique (Tetteh et al., 2015). Studies have shown that, following the ablation of spermatogonia through busulphan administration, the recovery of the GSC compartment involves the large-scale transfer of Ngn3-expressing cells to the GFRα1-expressing stem cell compartment, as well as the expansion of the surviving GFRα1-expressing cell population (Hara et al., 2014; Nakagawa et al., 2010). Similarly, the targeted genetic ablation of Lgr5-expressing cells following exposure to diphtheria toxin leads to the transfer of differentiating cells back into the stem cell compartment, and the regeneration of the stem cell pool (Tian et al., 2011). Furthermore, independent studies show that cells positive for the Notch ligand Dll1, as well as quiescent Lgr5-expressing cells, which are both largely committed to differentiation into the secretory cell lineage in conditions of normal homeostasis, can re-establish multipotency and contribute to long-term self-renewal following the ablation of ISCs through radiation damage (van Es et al., 2012; Buczacki et al., 2013). Finally, the regeneration of trachea following the genetic ablation of basal cells (which include the resident stem cell population) involves the de-differentiation of differentiated cells known as club cells (Tata et al., 2013). Together, these results suggest that the entry of cells into a differentiation pathway may not involve an abrupt ‘binary’ decision but may occur progressively, with cells retaining stem cell potential ready to be mobilized under appropriate cues. Such flexibility may strengthen the resilience of tissues to crisis or injury, enabling the ensemble of differentiating progeny to function as a ‘reserve’ stem cell population (compare with Potten and Loeffler, 1990).
Taken together, these studies suggest that the fate potential of stem and progenitor cells may not be organized into a strict classical ‘one-way’ proliferative hierarchy involving functionally discrete cell populations. Rather, the arrangement of cell types may be more accurately represented as a continuum, in which both the proliferative and fate potential becomes gradually restricted. In such cases, transitions between different cell ‘states’ may occur reversibly even under physiological conditions, in response to niche-dependent factors, and can be promoted through injury or stress.
Stem cell-niche interactions
The intestinal crypt and hair follicle bulge highlight the crucial role played by interactions with the local microenvironment in defining the proliferative capacity and fate behaviour of stem cells. In the intestinal crypt, the balance between stem cell loss and replacement, and size of the stem cell pool, are regulated by exposure to Paneth cells as well as to factors from the adjacent stromal tissue. Through competition for limited niche access, ISCs are able to self-renew, and they can recover their number during the regeneration of tissue following the partial ablation of the stem cell compartment by radiation damage or other forms of injury (Tian et al., 2011; Metcalfe et al., 2014).
A similar strategy to regulate the size of the stem cell compartment may operate in the germline. Although studies have not yet identified a localized niche structure in the mammalian testis, the association of undifferentiated spermatogonia with the vasculature (Yoshida et al., 2007) suggests that intratubular domains may play host to a somatic cell type (or types) that creates a niche environment to support GSCs, much as Paneth cells do in crypts. Competition of spermatogonia for access to these limited niche domains may provide a simple and robust mechanism to regulate both the balance between syncytial fragmentation and differentiation, and the total size of the stem cell pool. Furthermore, if GFRα1 expression is linked to proximity to the niche, then the fragmentation of GFRα1+ syncytia may displace their neighbours from niche-maintaining sites, leading to loss of GFRα1 expression and the upregulation of differentiation markers such as Ngn3.
In contrast to the mammalian testis, the effects of localized niche factors on stem cell regulation have been defined in the Drosophila ovary and testis. In these cases, stem cell identity is traditionally thought to be restricted to the population of GSCs that directly contact a central hub of stromal cells (Hardy et al., 1979; Kiger et al., 2000). These cells remain closely associated with their niche during the cell cycle through cadherin-mediated cell adhesion (Song and Xie, 2002). Through a regulated process of spindle orientation, GSC division leads to a predominantly asymmetric fate outcome in which one daughter cell remains anchored to the hub and retains GSC identity, while another is displaced from the niche and enters into a differentiation pathway (Hardy et al., 1979). However, static lineage-tracing studies and ex vivo live imaging in the Drosophila testis and ovary show that, even under normal physiological conditions, sporadic stem cell loss from the hub may be compensated for by the symmetric duplication of neighbouring stem cells, and vice versa, leading to neutral drift dynamics of the clonal population (Sheng and Matunis, 2011; Kronen et al., 2014). Whether these rare events are associated with chance loss or active displacement of ‘inferior’ GSCs, or whether infrequent symmetric divisions are a routine part of the normal program of homeostatic turnover remains unclear. In this context, it is interesting to note that the second resident stem cell population in Drosophila testis, the somatic cyst stem cells (CySC) that give rise to the cyst cells ensheathing developing germ, undergo loss and replacement at a much higher rate (Amoyel et al., 2014).
Alongside the ability of Drosophila GSCs to undergo symmetric as well as asymmetric cell divisions in normal homeostasis, differentiating germ cells also retain the ability to re-establish contact with the hub and re-acquire stem cell function in the course of regeneration following the depletion of the GSC pool by protein starvation or genetic ablation (Brawley and Matunis, 2004; Kai and Spradling, 2004). Although such behaviour has been traditionally associated with a process of de-differentiation, and is distinct from GSC renewal through symmetric cell division, it is interesting to note that the process of spindle orientation and division asymmetry may not be essential for germline maintenance. In particular, studies based on the targeted depletion of stat in GSCs, which leads to their detachment from the hub, show that contact with CySCs alone is sufficient to maintain GSC self-renewal and spermatogenesis (Leatherman and Dinardo, 2010). Indeed, under these conditions, the maintenance of Drosophila germ line may in fact parallel the process of dynamic heterogeneity that characterizes the mammalian system.
Dynamic interactions with the niche may thus serve both to constrain stem cell identity under physiological conditions and orchestrate the regeneration of the stem cell compartment following injury. Future studies might address the extent to which recruitment to the stem cell pool upon injury is a reflection of underlying cell fate heterogeneity, or instead is the consequence of active cell fate reprogramming following catastrophic stem cell loss.
The role of stem cell quiescence
Although lineage-tracing assays provide a powerful read-out of the behaviour and dynamics of cycling cells, they are notoriously insensitive to the existence and potential function of long-term quiescent (slow-cycling) or dormant cell populations. Indeed, both the germline and intestinal crypt have been associated with a quiescent progenitor cell population. In humans and other primates, detailed analyses of fixed specimens have identified a subpopulation of singly isolated spermatogonia, termed Adark on the grounds of their histological appearance (de Rooij and Russell, 2000). It has been speculated that this minority cell population may play a special role in the long-term maintenance of tissue, supporting the more rapidly cycling but transient spermatogonial cell population (Hermann et al., 2010). Similarly, studies of the mouse intestinal crypt have identified a population of quiescent cells marked by the expression of telomerase reverse transcriptase (Tert) or Lgr5 (Muñoz et al., 2012; Montgomery et al., 2011).
It is difficult to identify the potential function and significance of these minority quiescent cell populations, particularly in tissues such as the germline and intestinal crypt, where active cycling cells are seen to maintain life-long self-renewal (at least in mice). As shown by a recent study of intestinal crypts, quiescence may not in itself be a signature of stem cell function, at least under conditions of normal maintenance (Buczacki et al., 2013). However, in long-lived organisms, it may be advantageous to hold a dedicated slow-cycling or dormant stem cell population in reserve so that it may ‘drip-feed’ the cycling stem cell pool to compensate for progressive chance loss or ageing.
Alternatively, the reversible transfer of stem cells between an active and quiescent state under physiological conditions (Fig. 1C), itself a manifestation of dynamic heterogeneity, may provide a robust mechanism to maintain a stem cell pool such that the overall turnover rate of the tissue is steady but slow. Equally, the sporadic entry of stem cells into a quiescent or dormant state in a cycling tissue may provide an insurance mechanism to shield the wider population from demands experienced by actively cycling cells, and thereby protect the long-term integrity of the tissue. Such behaviour would mirror the strategy of phenotypic switching observed in bacterial populations (Balaban et al., 2004).
Conclusions
Taken together, these observations highlight the requirement to develop an extended definition of stem cell identity, one that adequately captures the heterogeneity of tissue stem cells and the flexibility of their progeny. In the studies discussed above, stem cell identity consistently emerges not as the property of a discrete population, but as a functional state that a wider population of cells may enter, exit and re-enter according to the demands of the tissue. All cells belonging to this wider population therefore have the capacity for long-term self-renewal, but their proliferative potential at any given time may depend on their precise location, on signals from the niche environment, and on other extrinsic and intrinsic factors.
If stem cell identity is indeed a state that is accessible to a wide population of progenitors, it becomes crucial to determine how recruitment to and exit from the stem cell pool is regulated at the molecular level. The dependence of self-renewal potential on the position of a cell within the niche that was observed in intestine and hair follicle suggests that spatially localized signals may play an important role in determining the state of progenitors; these could result from direct interactions with surrounding cells of the same or different types, extracellular matrix components or soluble paracrine mediators (Jones and Wagers, 2008). Non-local signals, including metabolic and endocrine factors, may further contribute to aligning overall niche stem cell activity with the requirements of different tissues (Scadden, 2006).
In maintaining the stem cell pool at a constant size, it is not clear whether the aberrant loss of a stem cell triggers recruitment of a differentiating progenitor back to the niche, or whether stem cells exit the compartment in response to the fate reversal of differentiating cells. Although the mechanisms that govern regeneration after injury may differ substantially from those operative in homeostasis, the examples above, however, suggest that stem cell recruitment may occur in response to stem cell loss from the niche. As repopulation of the niche has been observed even following the ablation of the entire stem cell pool, it is conceivable that stem cell identity can be initiated by factors derived from supporting cells or the extracellular matrix, rather than from other stem cells. The physical structure of the niche, which is independently maintained by the extracellular matrix, associated vasculature and supporting cells, may therefore play a more active role in regulating stem and progenitor cell fate than previously appreciated.
The observation that the recruitment of differentiating progenitors back to the stem cell pool occurs under normal physiological conditions opens up an intriguing new possibility for the mechanism of ageing in tissues with high cellular turnover. Although the accumulation of mutations in nuclear and mitochondrial DNA is considered to be the most fundamental and irreversible cause of ageing, declining homeostasis in cycling tissues has also been associated with changes in the numbers or properties of stem cells and their niches (Liu and Rando, 2011; Jones and Rando, 2011). In particular, tissue ageing may be accompanied by a gradual loss of functional stem cells, which has been thought to result from increased rates of stem cell death, quiescence or differentiation (Nijnik et al., 2007; Inomata et al., 2009). However, paradigms that could explain how the global process of ageing leads to the gradual loss of only a small fraction of stem cells at any one time have so far been lacking. Following the discoveries of heterogeneity within stem cell compartments and the ongoing interconversion between stem cells and their differentiating progeny, it may be that ageing is a result not (only) of an increase in the active loss of stem cells, but (also) of a decrease in the recruitment of differentiating cells back to the niche. This would be consistent with a number of studies showing that stem cells from ageing animals can continue to function normally when maintained in a young niche or provided with young systemic factors (Katsimpardi et al., 2014; Ryu et al., 2006).
Over the past few years, the co-existence of distinct progenitor cell behaviours has been reported across a wide range of tissues, often through genetic lineage-tracing approaches (Challen et al., 2010; Lu et al., 2012; Van Keymeulen et al., 2011). It remains an unresolved issue whether these observations reflect an underlying heterogeneity within a single stem cell compartment, and whether the continual inter-conversion of long-term stem cells and their differentiating progeny occurs in these tissues under physiological conditions. Static lineage tracing alone can provide important clues to the degree of progenitor cell plasticity in some tissues, but owing to the lack of tools for the targeted labelling of specific subpopulations of cells, it is frequently impossible to identify unambiguously the source of labelled clones. To unravel stem cell heterogeneity, lineage hierarchies and the de-differentiation capacity of differentiating cells, in vivo live-imaging approaches will therefore be indispensable. In resolving the biological significance of heterogeneity within the stem cell compartment, we expect that stem cell diversity and lineage plasticity will emerge as ubiquitous features of adult tissue stem cell populations.
Acknowledgements
We are grateful to Cedric Blanpain, Hans Clevers, Tariq Enver, Valentina Greco, Kenshiro Hara, Allon Klein, Anna Philpott, Jacco van Rheenen, Shahragim Tajbakhsh, Doug Winton, Shosei Yoshida and all members of the Simons group for illuminating discussions.
Footnotes
Competing interests
The authors declare no competing or financial interests.
Funding
B.D.S. acknowledges the financial support of the Wellcome Trust [098357/Z/12/Z] as well as core grants from the Wellcome Trust [092096] and Cancer Research UK [C6946/A14492].
- Received December 14, 2014.
- Accepted February 24, 2015.
- © 2015. Published by The Company of Biologists Ltd
References
Article navigation
Related articles
Cited by...
More in this TOC section
Similar articles
Other journals from The Company of Biologists
Kathryn Virginia Anderson (1952-2020)
Developmental geneticist Kathryn Anderson passed away at home on 30 November 2020. Tamara Caspary, a former postdoc and friend, remembers Kathryn and her remarkable contribution to developmental biology.
Zooming into 2021
In a new Editorial, Editor-in-Chief James Briscoe and Executive Editor Katherine Brown reflect on the triumphs and tribulations of the last 12 months, and look towards a hopefully calmer and more predictable year.
Read & Publish participation extends worldwide
Over 60 institutions in 12 countries are now participating in our Read & Publish initiative. Here, James Briscoe explains what this means for his institution, The Francis Crick Institute. Find out more and view our full list of participating institutions.
Upcoming special issues
Imaging Development, Stem Cells and Regeneration
Submission deadline: 30 March 2021
Publication: mid-2021
The Immune System in Development and Regeneration
Guest editors: Florent Ginhoux and Paul Martin
Submission deadline: 1 September 2021
Publication: Spring 2022
Both special issues welcome Review articles as well as Research articles, and will be widely promoted online and at key global conferences.
Development presents...
Our successful webinar series continues into 2021, with early-career researchers presenting their papers and a chance to virtually network with the developmental biology community afterwards. Sign up to join our next session:
10 February
Time: 13:00 (GMT)
Chaired by: preLights