Discovered in Caenorhabditis elegans in 1993, microRNAs (miRNAs) make up a novel class of tiny, ~21–24 nucleotide, non-coding RNA species. Since its identification as a key component of a broadly conserved mechanism that regulates gene expression post-transcriptionally, the miRNA pathway has emerged as one of the most extensively investigated pathways of the past decade. Because of their potential to regulate a large number of protein-encoding genes, miRNAs have been implicated in numerous biological processes, including development, stem cell regulation and human diseases. In this Commentary, we focus on miRNAs and their roles in mammalian stem cells. Following an introduction to the miRNA biogenesis pathway with an emphasis on its regulatory features, we then discuss what is currently known about the roles that miRNAs have in the differentiation and maintenance of embryonic and somatic stem cells of diverse origins. In particular, their roles in stem cell differentiation have been well documented. Insights from these studies provide a paradigm for the function of miRNAs in facilitating cellular transitions during differentiation. By contrast, the roles that miRNAs have in the maintenance of stem cells are less well understood. However, with recent advances, their role as a rheostat that fine-tunes stem cell self-renewal has begun to emerge. Finally, we discuss future studies that will hopefully lead to a comprehensive understanding of the miRNA pathway in stem cells.
A hallmark of stem cells is their ability to self-renew in the long term to maintain their own population and to differentiate in order to generate daughter cells with specific physiological functions (Siminovitch et al., 1963). During embryonic development, pluripotent embryonic stem cells (ESCs) residing in the inner cell mass give rise to all tissues and organs of the body. Throughout their lifetime, tissue-specific, multipotent somatic stem cells maintain homeostasis of individual tissues and organs. Some of the most fascinating questions in biology focus on understanding how stem cells are specified, maintained and instructed to differentiate. Whereas external cues provided by, for example, the stem cell microenvironment or niche are required to regulate stem cell behavior, internal cues, such as transcriptional regulation, also govern the fate of stem cells. Ultimately, all regulatory networks converge on the regulation of gene expression.
MicroRNAs (miRNAs) comprise a novel class of non-coding regulatory RNAs that are widely expressed in both plants and animals (reviewed in Ambros, 2004; Bartel, 2009). They regulate mRNA stability and consequently protein production by recruiting the RNA-induced silencing complex (RISC) to its cognate target sites. They therefore have an important role in the post-transcriptional regulation of gene expression (reviewed in Bartel, 2009). The recognition of mRNA target sites, which are often located in the 3′ untranslated region (3′-UTR), by miRNAs is believed to be primarily mediated by base pairing between nucleotides 2–8 of the miRNA, a region also referred to as the seed sequence, and the cognate mRNA sequence (reviewed by Bartel, 2009). miRNA-mediated gene regulatory networks are rather complex because a single miRNA can recognize hundreds of targets and a single mRNA can be simultaneously regulated by multiple miRNAs. The complexity of miRNA-mediated regulation, its potential impact on the expression of a large number of proteins and its control over the output from the transcriptome have drawn increased attention to these tiny regulators. Over the past decade, much progress has been made in deciphering miRNA expression and function in animal development, including the differentiation and maintenance of stem cells (reviewed by Ambros, 2004; Bartel, 2009).
The investigation of the hitherto unappreciated layer of gene regulation mediated by miRNAs has also provided new insights into our understanding of stem cell biology and pointed to new directions for therapeutic applications of stem cells. In this Commentary, we review the most recent developments in our understanding of the biological role of miRNAs in the differentiation and maintenance of mammalian stem cells, discuss lessons learned from recent studies and offer our views on future challenges. Because of the considerable interest in this topic, many excellent reviews have been published in the past few years and we also refer our readers to these articles for a comprehensive overview of the field (Gangaraju and Lin, 2009; Ivey and Srivastava, 2010; Martinez and Gregory, 2010).
miRNA biogenesis and its regulation
In the past decade, the miRNA biogenesis pathway has been extensively investigated. We outline the biogenesis pathway in Fig. 1, but would also like to refer our readers to several excellent reviews for more detailed discussions (e.g. Kim et al., 2009; Krol et al., 2010). Here, we primarily explore the regulatory features of the miRNA biogenesis pathway and discuss the potential implications for the role of miRNAs in stem cells.
miRNAs can be transcribed by both RNA polymerase II and III (Borchert et al., 2006; Cai et al., 2004). It is widely recognized that many highly expressed mammalian miRNAs are initially transcribed by RNA polymerase II in the form of a long, primary transcript (pri-miRNA) (Cai et al., 2004). Because transcription mediated by RNA polymerase II is under extensive regulation by transcription factors and diverse epigenetic mechanisms, a complex network of transcriptional regulation can potentially control miRNA expression and confer tissue- and cell-specific expression patterns at the transcriptional level. Indeed, numerous miRNAs are specifically regulated by transcription factors in stem cell lineages. In ESCs, key transcriptional regulators such as Oct4, Sox2, Nanog and Tcf3 specifically occupy the promoters of both active and silent miRNA genes to activate and repress their expression, respectively (Marson et al., 2008). In addition, specific histone marks associated with actively transcribed regions (such as histone H3 Lys4 trimethylation and histone H3 Lys36 trimethylation marks), as well as those associated with silenced regions (such as histone H3 Lys27 trimethylation marks), are also correlated with active and silent miRNAs, respectively (Marson et al., 2008). These observations provide strong evidence for an emerging scheme, in which master transcriptional regulators control a network of gene expression profiles, including the expression of both miRNAs and mRNAs. Subsequently, miRNAs modulate the expression level of their mRNA targets to further shape the output generated from the transcriptome.
In addition to transcriptional regulation, various post-transcriptional regulatory features have a role in miRNA production. During the formation of the primary transcript, the flanking sequences of the miRNA fold into a hairpin structure, called pre-miRNA, that is characteristic of miRNA-encoding genes. For most stereotypical miRNAs, the hairpin is sequentially processed, first by the Drosha–DiGeorge syndrome critical region gene 8 (Dgcr8) microprocessor complex in the nucleus and then by Dicer in the cytoplasm. Following Dicer processing, the mature miRNA is directly embedded with Argonaute proteins to form the RISC (Fig. 1) (Kim et al., 2009; Krol et al., 2010).
For many miRNAs, a co-transcriptional processing model for the pre-miRNA hairpin by the Drosha–Dgcr8 complex has been proposed (Morlando et al., 2008). However, several studies have shown that this step in the generation of the pre-miRNA can be tightly regulated by several proteins, as illustrated by the following examples (see also Fig. 1). Heterogeneous nuclear ribonucleoprotein A1 (hnRNP A1), for instance, specifically interacts with the stem-loop of the miR-18a pre-miRNA before further processing by Drosha and thereby enhances miR-18a maturation (Guil and Caceres, 2007). Furthermore, p53 and SMAD proteins enhance Drosha–Dgcr8 processing of pri-miRNA by interacting with the DEAD-box RNA helicase p68, a component of the Drosha–Dgcr8 microprocessor complex, and by recruiting the Drosha–Dgcr8 complex to a specific set of miRNAs (Davis et al., 2008; Suzuki et al., 2009). By contrast, Lin28 specifically inhibits the production of mature let-7 miRNA by interfering with pre-let-7 cropping from the primary transcript in the nucleus and destabilizing pre-let-7 in the cytoplasm (Heo et al., 2008; Viswanathan et al., 2008). Notably, the inhibition of let-7 production by Lin28 has an important role in the maintenance of ESCs (Melton et al., 2010; Viswanathan et al., 2008). These mechanisms, therefore, provide additional regulatory mechanisms by which cells can control miRNA production and function after the transcription of their primary transcripts.
After the miRNA duplex is released from the pre-miRNA hairpin by Dicer, one strand of the duplex is selectively incorporated with an Argonaute protein to form the RISC (Schwarz et al., 2003). In mammals, there are four Argonaute proteins (Ago1–Ago4, also known as eIF2c1–eIF2c4) that can associate with miRNAs to form the RISC. However, it is not completely clear whether all four Ago proteins associate with similar subset of miRNAs and have similar functions in the miRNA pathway during development and in stem cells. When the Ago proteins are knocked out individually in mice, only the loss of Ago2 results in embryonic lethality, whereas the loss of Ago1, Ago3 and Ago4 does not lead to discernible phenotypes during mouse development.
Among the Ago proteins, Ago2 is the only one with so-called ‘slicer’ activity, which is required for the cleavage of target mRNA by small interfering (si) RNA or perfectly matched miRNA (Liu et al., 2004). In mammalian cells, however, the requirement for the slicer activity of Ago2 was not immediately clear because very few endogenous siRNAs are produced and these are mostly restricted to the oocytes and ESCs (Tam et al., 2008; Watanabe et al., 2008). Furthermore, there is only one example of an miRNA that cleaves a single target, namely the cleavage of mRNA encoding HoxB8 by miR-196 (Yekta et al., 2004). Apart from this example, there is little evidence of mammalian miRNA-mediated cleavage of target mRNAs. However, in a hunt to decipher the slicer-dependent Ago2 function, miR-451, an miRNA highly expressed in terminally differentiated blood cells, was identified as a unique miRNA whose biogenesis unusually requires the ‘slicer’ activity of Ago2, but not the dicing activity of Dicer at the step of pre-miRNA processing (Fig. 1) (Cheloufi et al., 2010). Slicer-deficient Ago2 mutant mice are anemic, a phenotype that resembles the defect in erythropoiesis that was observed in miR-451-null mice (Cheloufi et al., 2010; Patrick et al., 2010; Rasmussen et al., 2010; Yu et al., 2010). However, the slicer-deficient Ago2 mutant pups died shortly after birth, showing a more severe phenotype than miR-451-null animals. These differences could be a sign that Ago2 is a unique Argonaute protein that has the ability to cleave RNAs in an miRNA-dependent or -independent manner (Karginov et al., 2010). Interestingly, a recent study showed that Ago2 itself is regulated by the mouse homolog of lin41 (mlin41, which in turn is a target of the let-7 miRNA) in several stem cells. Furthermore, the repression of Ago2 by mlin41 antagonizes the silencing of target mRNAs by let-7 and miR-124, which are both implicated in stem cell differentiation (see below) (Rybak et al., 2009). Together, these results suggest that the association of miRNAs with individual Ago proteins also has a role in regulating stem cell fate.
Whereas future studies will be needed to fully understand the extent to which the production of individual miRNAs and their association with individual Ago proteins are regulated, these results clearly indicate that miRNA biogenesis is under delicate control at multiple steps during and after transcription. In turn, although functional analyses of regulated miRNA biogenesis are still rare, such regulation could have a substantial impact on miRNA activity in stem cells and their lineages, as demonstrated by Lin28-mediated inhibition of let-7 in ESCs.
miRNA function in stem cell differentiation
When stem cells give rise to their progenies during differentiation, the transition at the cellular level resembles closely the developmental progression at the organismal level, that is, the progression from the inner cell mass (which comprises pluripotent ESCs) to the three germ layers (endoderm, mesoderm and ectoderm) and eventually to fully functional tissues and organs. Numerous pathways that have essential roles in stem cells, for example the Wnt and Notch signaling pathways, were initially characterized as a result of their roles during animal development. Similarly, the founding members of the miRNA family, namely lin-4 and let-7, were originally identified as heterochronic genes that govern developmental transitions in the nematode Caenorhabditis elegans (Lee et al., 1993; Pasquinelli et al., 2000; Reinhart et al., 2000).
Lin-4 and let-7 negatively regulate the expression of master regulators of differentiation, such as lin-14, lin-28 and lin-41. These master regulators are essential for the maintenance of an early developmental lineage, but must also be downregulated as the animal transitions to a later lineage. For example, lin-4 downregulates lin-14 through the first larval stage (L1) and let-7 downregulates lin-28 and lin-41 when the animal develops from the fourth larval stage (L4) to the adult stage (Lee et al., 1993; Reinhart et al., 2000).
Studies on lin-4 and let-7 miRNAs provided the first evidence that miRNAs have essential roles in development. The negative feedback regulatory network between lin-4 and let-7 and their mRNA targets provided the basis for a model in which miRNAs help to deplete protein production from mRNAs inherited from an earlier developmental stage. In doing so, miRNAs facilitate a precise and robust transition through the developmental program.
Following this paradigm, the roles for miRNAs in stem cell differentiation have been extensively documented in recent years. Fueled by a rapid development in miRNA profiling techniques, it quickly became evident that numerous miRNAs are differentially expressed as stem cells embark on differentiation (Houbaviy et al., 2003; Landgraf et al., 2007; Sempere et al., 2004; Suh et al., 2004). These observations provided the first clue and a molecular basis to explore the crucial roles that miRNAs have in stem cell differentiation. In the following sections, we will discuss current evidence for the important roles of miRNAs in embryonic and somatic stem cell differentiation.
miRNA in embryonic stem cell differentiation
Because of their ease to be cultivated as pluripotent stem cells and the ability to differentiate them along different lineages in vitro, ESCs have been extensively studied, yielding comprehensive knowledge on their underlying gene regulatory network. Therefore, it comes as no surprise that mouse ESCs have served as a fertile ground for the first observations of the functional significance of miRNAs in mammalian stem cells. The importance of miRNAs was highlighted when, following knockout of Dicer and subsequent loss of all miRNAs, ESCs failed to silence their self-renewal program and displayed severe defects in their ability to differentiate (Kanellopoulou et al., 2005; Murchison et al., 2005).
Because of the potential roles of Dicer in the biogenesis and function of other small RNA species, it was not clear at the time whether the loss of miRNAs in the Dicer-null ESCs was solely responsible for the phenotype observed. However, when the second major miRNA-processing gene, Dgcr8, was ablated in ESCs, similar defects in differentiation were observed (Wang et al., 2007). Together, these results unequivocally establish essential roles for miRNAs in controlling the balance between ‘stemness’ and differentiation in cultured ESCs.
To ensure robust differentiation of ESCs towards a specific cell lineage, at least two distinct processes must be accurately controlled. First, the self-renewal program has to be properly silenced. Second, a robust program dictating the specific cell fate has to be established. Recent studies have begun to reveal roles for individual miRNAs in both events. miR-145 is expressed at low levels in self-renewing human ESCs, but it is substantially upregulated during differentiation (Xu et al., 2009). Acting as a key antagonist of ESC maintenance, miR-145 directly targets and suppresses the mRNAs encoding the transcription factors Oct4, Sox2 and Klf4, which are required to maintain pluripotency (Fig. 2A) (Xu et al., 2009).
Several other miRNAs follow a similar pattern during ESC differentiation. Among the most fascinating of these is the let-7 family, which is highly conserved from C. elegans to humans. In mice and humans, the let-7 family comprises eleven members dispersed over eight genomic loci. Let-7 antagonizes Myc and Lin28 protein production and, as a result, destabilizes the self-renewing capacity and promotes the differentiation of ESCs (Fig. 2A) (Melton et al., 2010). The potency of let-7 in promoting differentiation is further demonstrated by the observation that the inhibition of this miRNA family substantially enhances the reprogramming of somatic cells into induced pluripotent stem cells (Melton et al., 2010).
The ability to adjust culture conditions to coax ESCs along particular lineages has also provided new insights into how miRNAs function during the process of specifying lineage. Both miR-1 and miR-133 are muscle-specific miRNAs that are induced by serum response factor (SRF), a transcription factor that is activated when ESCs differentiate to form cardiomyocytes (Ivey et al., 2008; Zhao et al., 2005). When miR-1 and miR-133 are ectopically activated in murine ESCs, they promote mesoderm differentiation by repressing non-muscle gene expression through direct downregulation of Dll-1, a key ligand in the Notch signaling pathway (Fig. 2B) (Ivey et al., 2008). Interestingly, miR-1 and, to a lesser extent, miR-133 can rescue the abnormal differentiation phenotype observed in SRF-null embryoid body cells, arguing for a major role for miR-1 and miR-133 in the SRF-mediated differentiation program (Ivey et al., 2008). When taken together, these observations support the hypothesis that tissue-specific miRNAs preferentially target mRNAs that are expressed in other lineages, narrowing the gene expression profile towards the defined cell lineage and thereby maintaining tissue identity (Fig. 2B) (Farh et al., 2005; Stark et al., 2005).
Whereas restricting other lineages is one method by which miRNAs can ensure tissue selectivity of differentiating somatic stem cells, miRNAs can also achieve this by regulating the maintenance of somatic cells upon ESC differentiation. In this regard, the recent findings of Delaloy et al. offer new insights (Delaloy et al., 2010). This group showed that, when human ESCs are induced to form proliferative aggregates (neurospheres) of multipotent neuronal progenitor cells (hNPCs), they switch on the expression of a brain-specific miRNA, miR-9. Moreover, when antisense oligonucleotides are used to downregulate miR-9 activity in hNPCs, their proliferation is restricted and their migration is enhanced. These findings underline roles for miR-9 in specifically promoting the differentiation of ESCs into cells with a neural fate (Fig. 2C) (Delaloy et al., 2010).
Taken together, these observations show that miRNAs function in three distinct ways to promote ESC differentiation: (1) silencing the self-renewal program by targeting core pluripotency factors; (2) restricting the expression of genes in other lineages and, hence, dictating the differentiating ESCs to a defined lineage; (3) maintaining the state of the specified lineage (Fig. 2A–C).
miRNA in somatic stem cell differentiation
In parallel to work carried out in ESCs, numerous studies focusing on the differentiation of somatic stem cells have yielded considerable insights into the requirement for miRNAs in regulating homeostasis of somatic tissues (Fig. 2D–F).
In mammalian skin, miR-203 is the most abundantly expressed miRNA (Yi et al., 2008). When the spatiotemporal expression of miR-203 was visualized by in situ hybridization, it was shown that this miRNA is rapidly switched on when epidermal stem cells begin to differentiate (Yi et al., 2008). Illustrating the functional significance of this expression pattern, premature activation of miR-203 in epidermal stem cells depletes the stem cell pool by restricting their proliferative potential and promoting their exit from the cell cycle. Conversely, transient inhibition of endogenous miR-203 by a chemically modified antisense oligonucleotide (an antagomir) resulted in normally differentiating suprabasal cells that proliferated and showed upregulation of the key miR-203 target ΔNp63α (Fig. 2D) (Yi et al., 2008). ΔNp63α is a transcription factor essential for the maintenance of epidermal stem cells (Mills et al., 1999; Senoo et al., 2007; Yang et al., 1999). Although it is not yet clear whether ΔNp63α is the only target that mediates the function of miR-203, this study provides an important example of miRNAs regulating the developmental transition by targeting a key stem cell regulator at the point at which differentiation is induced (Yi et al., 2008). Additionally, it is intriguing that, whereas miR-203 potently represses epidermal stem cell proliferation, terminal differentiation markers such as Krt1, Krt10, loricrin and involucrin were not activated when miR-203 was precociously induced in the stem cells (Yi et al., 2008). This latter finding suggests that, at least for the epidermis, the exit from the cell cycle and the induction of terminal differentiation in somatic stem cells can be regulated by distinct mechanisms.
In the adult brain, miR-9 has effects on neuronal stem cells that differ from those that miR-203 has on epidermal stem cells. In the brain, these neural stem cells reside at the subventricular zone (SVZ). miR-9 functions in neural progenitor maintenance, in part by directly targeting the nuclear receptor NR2E1 (nuclear receptor subfamily 2, group E, member 1, previously known as TLX), which is an essential governor of self-renewal of neural stem cells (Zhao et al., 2009). Interestingly, NR2E1 also antagonizes the expression of miR-9 by directly reducing the transcription of pri-miR-9 (Zhao et al., 2009). Thus, NR2E1 and miR-9 form a negative feedback regulatory network to balance both proliferation and differentiation of neural stem cells.
By contrast, miR-124, which is one of the most specific and abundant miRNAs in the brain, more closely mirrors the effects shown by miR-203. Like miR-203, miR-124 is expressed at low levels in the stem cell compartment in the SVZ, but is sharply upregulated in mature granule and periglomerular neurons (Cheng et al., 2009). Similarly, gain of function of miR-124 induces cell cycle exit, whereas inhibition of miR-124 by antagomir in vivo results in an increase in the population of precursor cells in the SVZ. Moreover, Sox9, a key transcription factor whose downregulation is required for neural differentiation, has been identified as a direct target of miR-124 (Fig. 2E) (Cheng et al., 2009).
A recent study of skeletal muscle satellite cells provided another interesting parallel to the skin stem cell lineage. Analogous to ΔNp63α in epidermal stem cells, Pax7 is a crucial transcription factor that is highly expressed in quiescent muscle stem cells (satellite cells) and is required to maintain the stem cell population. Upon injury, satellite stem cells are activated to repair the wound. In such an event, Pax7 is rapidly downregulated to allow these cells to enter differentiation and contribute to muscle regeneration. Both miR-1 and miR-206 are markedly upregulated concomitant with downregulation of Pax7 and satellite cell differentiation; both of these miRNAs target mRNA encoding Pax7 (Fig. 2F) (Chen et al., 2010). Conversely, specific inhibition of miR-1 and miR-206 by antagomirs delays Pax7 downregulation and interferes with the differentiation program (Chen et al., 2010). Together, these findings suggest that some somatic miRNAs are primarily required for transitioning stem cells along a differentiation lineage.
The hematopoietic system, with its well-defined cell lineages, offers an ideal system to map the lineage-specific expression patterns of miRNAs and to decipher their roles in differentiation. In one of the earliest studies of mammalian miRNAs, miR-181, miR-223 and miR-142 were all found to be upregulated when bone marrow progenitor cells differentiate towards B lymphocytes (Chen et al., 2004). Moreover, forced expression of miR-181 in those progenitors enhanced B-lymphocyte development at the expense of T-lymphocyte differentiation. In another important study, it was discovered that miR-155 is expressed in mature B and T lymphocytes, and that it is also substantially upregulated in a variety of lymphomas (Eis et al., 2005). Loss-of-function studies in mice revealed that miR-155 is a crucial regulator of specific differentiation processes in the immune response in vivo (Rodriguez et al., 2007; Thai et al., 2007). In an effort to identify the targets of miR-155, the mRNAs upregulated in miR-155-null T lymphocytes were analyzed for the presence of seed matches to miR-155 (Rodriguez et al., 2007). Even though ~65% of the upregulated genes contained miR-155 seed matches in their 3′-UTRs, the key miR-155 targets seemed to be mRNAs encoding cytokines, providing an explanation for the remarkable impact and specificity of this miRNA on immune cell lineages.
In another elegant study of hematopoietic-specific miRNAs, Xiao et al. focused on miR-150 and provided an example of a single mRNA–miRNA regulatory pair that has crucial functions in lymphocyte development (Xiao et al., 2007). Like miR-155, miR-150 is primarily expressed in mature lymphocytes rather than in the progenitor cells. In loss- and gain-of-function studies, miR-150 was shown to regulate lymphocyte terminal differentiation (Xiao et al., 2007). Importantly, the expression of c-Myb, a key transcription factor and the top predicted target of miR-150, was inversely correlated with that of miR-150, suggesting that miR-150 might be both necessary and sufficient to control c-Myb expression. Interestingly, the c-Myb heterozygous knockout mice showed phenotypes similar to those observed in miR-150 transgenic mice, thereby lending support to the argument that miR-150 carries out its function through the accurate control of c-Myb expression (Xiao et al., 2007).
Studies of the erythroid lineage have also revealed an inverse relationship between miRNA expression and stem cell progenitor maintenance. When erythroid progenitors begin to differentiate into red blood cells, the miR-144–miR-451 locus is directly activated by a key erythroid transcription factor, Gata-1 (Dore et al., 2008). Loss-of-function studies in both zebrafish and mouse revealed that this conserved miRNA locus is required for erythroblast maturation (Dore et al., 2008; Rasmussen et al., 2010). Interestingly, the mRNA of 14-3-3ζ, a key regulator of cytokine signaling, was identified as a direct target of miR-451. More importantly, knocking down 14-3-3ζ in miR-451-null hematopoietic stem cells rescued these defects during erythroid differentiation in vitro, providing another example whereby a single target might be primarily responsible for mediating the effects of an miRNA in a defined cellular context (Patrick et al., 2010; Yu et al., 2010).
Overall, the emerging literature on miRNAs in somatic tissues provides fertile ground for additional studies into the mechanisms by which miRNAs control the balance between somatic stem cells and their differentiating lineages (Fig. 2D–F).
Role of miRNAs in stem cell self-renewal
Compared with their function in differentiation, the roles of miRNAs in stem cell maintenance are poorly understood. Nonetheless, studies on ESCs have shown that loss of either Dgcr8 or Dicer results in proliferation defects. Because Dgcr8 acts as a cofactor for Drosha in the nucleus during miRNA biogenesis, direct transfection of an miRNA duplex into the cell can bypass the requirement for Drosha processing and thereby provide a means to examine the role of an individual miRNA or an miRNA family in the Dgcr8-null background. With this approach, it was shown that members of the miR-290 family, an ESC-specific miRNA family, are able to rescue the proliferation defects caused by loss of Dgcr8 function by directly controlling the expression of key regulators of the cell cycle pathway, including Cdkn1a, Rbl2 and Lats2, which are negative regulators of the cell cycle in ESCs (Wang et al., 2008) (Fig. 3A).
Interestingly, the regulation of Rbl2 by the miR-290 family is also a key link to the DNA methylation defects observed in Dicer-null ESCs (Fig. 3A) (Benetti et al., 2008; Sinkkonen et al., 2008). Derepression of Rbl2 in Dicer-null ESCs results in transcriptional repression of DNA methyltransferases and, in turn, leads to a decrease in de novo DNA methylation (Benetti et al., 2008; Sinkkonen et al., 2008).
The miR-290 family also has an essential role in antagonizing the effects of differentiation-related miRNAs, such as the let-7 family (Melton et al., 2010). The ectopic expression of let-7 family members in Dgcr8-null but not in wild-type ESCs was able to silence the self-renewal program. This observation suggests that other miRNAs that are highly expressed in ESCs are capable of blocking the effect of let-7 on self-renewal. Indeed, co-expression of the miR-290 and let-7 family members largely abolished the inhibition of let-7 miRNAs on self-renewal (Melton et al., 2010). Although the direct targets of the miR-290 family that mediate the antagonizing effects on the let-7 miRNAs remain elusive, it has become increasingly clear that an underlying network centered on the miR-290 family has a pivotal role in governing self-renewal and cell cycle progression in ESCs (Fig. 3A). The systematic identification of the targets of this family and functional dissection of their contribution should provide substantial new insights into our understanding of the self-renewal program.
In somatic stem cells, evidence for the involvement of miRNAs in the maintenance of self-renewal has just begun to surface. Notably, miR-205 was identified as the most abundantly expressed miRNA in mammary gland progenitor cells (Ibarra et al., 2007). When miR-205 was overexpressed in a mammary epithelial cell line with a heterogeneous progenitor population, it expanded the progenitor population, enhancing proliferation and leading to increased colony formation (Greene et al., 2010). Intriguingly, PTEN, a well-characterized tumor suppressor gene, was identified as an miR-205 target (Greene et al., 2010). Because the function of PTEN has an unusually intimate correlation with its expression level (Alimonti et al., 2010), it is of particular interest to speculate that the modulation of PTEN expression by miR-205 might have a crucial role in balancing self-renewal of the stem cells of the mammary gland and other tissues in which this miRNA is expressed (Fig. 3B).
Most recently, miR-125b was identified as an miRNA specifically expressed in skin stem cells (Zhang et al., 2011). When the stem cells differentiate into outer root sheath (ORS) progenitor cells, miR-125b is rapidly downregulated. When miR-125b is elevated in both stem cells and ORS progenitor cells through an inducible, transgenic strategy in mice, the balance between maintenance of stem cells and differentiation into hair follicle lineages is shifted towards the stem cell fate. More stem-cell-like cells are generated and their differentiation is impaired. Intriguingly, the maintenance of stem cells was intact and, upon withdrawal of doxycycline to decrease the levels of miR-125b, the differentiation block was lifted even after up to 4 months of continuous miR-125b induction. These observations suggest an interesting new model for the roles of miRNA in stem cells as a rheostat mechanism precisely governing the input of stem cells to the differentiation program (Fig. 3C).
New roles for miRNA regulation in injury and aging
In addition to the roles of miRNAs in controlling stem cell differentiation, which have become increasingly well established, several new twists have emerged in the miRNA world. Among them is the notion that miRNAs function in injury or stress situations, as alluded to earlier when discussing the roles of Pax7. Furthermore, knockout mouse models have yielded some intriguing surprises. In a remarkable series of studies on miRNAs in cardiac development, individual ablation of the genes encoding miR-133, miR-143, miR-145, miR-206, miR-208 and miR-499 did not manifest itself in discernible phenotypes during normal development and tissue homeostasis (reviewed by Liu and Olson, 2010). However, in response to various stress conditions, all of these mutant mice displayed strong phenotypes (Liu and Olson, 2010). Such observations reveal that, even though compensatory networks might be able to partially rescue the loss of individual miRNAs in a well-controlled environment in vivo, when exposed to more stringent conditions, these backup circuits no longer suffice to maintain normal tissue function. They also raise the interesting possibility that the miRNA pathway is particularly important for the proper maintenance and differentiation of stem cells when they are under stress.
Culture conditions can present both a stressful situation to cells and a platform for manipulating the microenvironment. As such, systems making use of cultured stem cells often generate new insights into miRNA functions. Examples of such a case are the bi-cistronic miR-143 and miR-145 loci, which are substantially upregulated in multipotent cardiac progenitors in vivo (Cordes et al., 2009). When specifically examined during the reprogramming of fibroblasts into vascular smooth muscle cells (VSMCs) in vitro, miR-145 but not the clustered miR-143, which shares no sequence similarity with miR-145, was able to potentiate the reprogramming effects of myocardin (Myocd), an essential component of the molecular switch that activates genes encoding contractile muscle proteins (Cordes et al., 2009). Furthermore, miR-145 was sufficient to guide ~75% cells into VSMCs, when the reprogramming was performed with neural crest stem cells (Cordes et al., 2009).
Whereas both assays demonstrated an essential role for miR-145 in controlling VSMC fate, the most remarkable finding was that, instead of inhibiting Myocd, whose mRNA contains recognition sites for miR-145, miR-145 activated the expression of a luciferase reporter containing the 3′-UTR of Myocd by more than 100-fold (Cordes et al., 2009). miRNAs have been shown previously to be capable of promoting protein production of their target mRNA in cultured cells when these are under cell cycle arrest (Vasudevan et al., 2007). However, a second luciferase reporter containing the 3′-UTR of another miR-145 target, Klf4, was specifically repressed by miR-145 under the same experimental conditions. These findings indicate that miR-145 can either activate or repress its targets, and that the effect is probably determined by specific interactions between miR-145 and individual mRNA targets (Cordes et al., 2009).
Exactly how miR-145 exerts its effects at the mechanistic level is not yet clear. One possibility is that miR-145 could prevent the binding of a context-dependent repressive RNA-binding protein (Cordes et al., 2009). Irrespective of the underlying mechanism, another question to be addressed in the future is the extent to which this presently unconventional regulatory mechanism emerges to take a place in the multifaceted toolbox of miRNAs.
Finally, it is worth mentioning that miRNAs have also been implicated in the regulation of stem cell aging. In this regard, Hmga2, a key transcription factor for the self-renewal of neural stem cells, is highly expressed in fetal neural stem cells in mice, but its levels decline by more than 99% during their lifetime (Nishino et al., 2008). The decreased level of Hmga2 reduces the self-renewal capacity of neural stem cells and seems to be, in part, caused by an approximately 30-fold age-induced increase in the expression of let-7b, which is known to target and inhibit Hmga2 (Mayr et al., 2007; Nishino et al., 2008). The specific disruption of Hmga2 regulation using let-7b with a truncated 3′-UTR showed substantial rescue of the self-renewal capacity in an in vitro culture assay, supporting the causative role of let-7b in the downregulation of Hmga2 (Nishino et al., 2008).
Closing remarks and future challenges
Functional characterization of miRNAs in mammalian stem cells is still in its infancy. So far, many studies used either in vitro cell culture models or transgenic mouse models, as well as in vivo delivery of antagomirs, to address the roles of miRNAs in stem cells. To rigorously investigate the roles that miRNAs have in governing stem cell fate in vivo and to examine their long-term functions, knockout mouse models will be indispensable for future studies. The development of accurate stem cell identification and isolation, highly sensitive miRNA profiling techniques at a single-cell level, and rapidly generated conditional knockout mouse models for individual miRNAs will hopefully allow us to address several fundamental questions concerning miRNAs and stem cells in the near future.
Among the myriad key issues that remain unresolved is the question of how a single miRNA or miRNA family regulates stem cells in response to various endogenous and exogenous stimuli. When C. elegans miRNAs are knocked out individually or in combination as a whole family that shares the same seed sequence, only a few exhibit developmental defects (Alvarez-Saavedra and Horvitz, 2010; Miska et al., 2007). Similarly, with more than 30 miRNAs having been knocked out individually without effect in mice, it is evident that the loss of a single miRNA or even a whole family might not cause the severe developmental defects in cell fate specification that would be expected from ablation of a gene encoding a master regulator. However, when these mutant mice are subject to physiological stresses such as injury or DNA damage, substantially disrupted phenotypes appear (reviewed by Leung and Sharp, 2010; Liu and Olson, 2010). These observations also suggest that miRNA-mediated regulation has an important role in managing the robustness of the biological system (reviewed by Herranz and Cohen, 2010).
These observations begin to link the miRNA pathway, one of the most ancient pathways (Christodoulou et al., 2010), to stress responses (reviewed by Leung and Sharp, 2010). It is worth noting that, because of the central role of stem cells during the development and homeostasis of adult tissues, a defective stress response in stem cells could be magnified and manifested as defects in their differentiated daughters. This illustrates the crucial need to investigate how stress signals regulate miRNA expression and how miRNA-mediated regulation in turn balances the output of gene expression and protects stem cells from the various and diverse stresses that they encounter throughout their life.
A second key issue to address will be how a single miRNA can execute its physiological function by regulating its targets. Directly related to this is the question whether a single miRNA can perform different functions when placed in distinct cellular contexts. As miRNA functions in mammalian stem cells begin to unfold, some of the most challenging questions to be answered are which mRNAs are targeted by a specific miRNA and which of its many targets are the most important for a specific miRNA in a given cellular context.
To answer these questions, the complexities of miRNA-mediated regulation must be considered. First, a single miRNA can target more than 100 genes in a defined cellular context. Second, a single gene can be targeted by multiple, potentially unrelated miRNAs. Third, the effect of an miRNA on its targets is somewhat subtle, for example, <2-fold downregulation, as demonstrated by recent studies using genome-wide approaches to quantify the impact of miRNA on protein output (Baek et al., 2008; Selbach et al., 2008). However, in both studies, the number of proteins analyzed was limited to ~3000 of the most abundantly expressed proteins. It is possible that miRNA targets, which are expressed at a relatively low level, can be more potently regulated, as demonstrated by numerous individual studies. Recently, experimental identification approaches have been applied to identify RISC-associated mRNA species as miRNA target candidates (Chi et al., 2009; Hafner et al., 2010). With future development in these promising techniques, especially given the reduced requirement for input materials, these approaches should provide exciting insights into the spectrum of miRNA targets and the extent to which they are regulated by miRNAs in mammalian stem cells.
A third key area to be addressed in the future is the question of how miRNA-mediated regulation integrates with other regulatory mechanisms, for example, transcriptional regulation, to modulate stem cell fate. Although biologists now appreciate the importance of the regulatory layer provided by miRNAs, the interactions between the miRNA pathway and other regulatory mechanisms must be elucidated to fully understand how miRNAs work. Furthermore, it will be important to unveil the characteristics of the miRNA pathway that are distinct from other mechanisms such as cellular-context-dependent function. The answers to these fascinating questions are certainly going to not only provide substantial new insights into the functions of miRNAs, but also point to new directions in how to use and target miRNAs in the manipulation of stem cells for regenerative medicine.
This publication was made possible by grants R00AR054704 and R01AR059697 (R.Y.), and R01AR031737 (E.F.) from the NIH. E.F. is a Howard Hughes Medical Institute Investigator. Deposited in PMC for release after 12 months.
- © 2011.