The role of microRNAs in human neural stem cells, neuronal differentiation and subtype specification
- 2.6k Downloads
The impressive neuronal diversity found within the nervous system emerges from a limited pool of neural progenitor cells that proceed through different gene expression programs to acquire distinct cell fates. Here, we review recent evidence indicating that microRNAs (miRNAs) are critically involved in conferring neural cell identities during neural induction, neuronal differentiation and subtype specification. Several studies have shown that miRNAs act in concert with other gene regulatory factors and genetic switches to regulate the spatial and temporal expression profiles of important cell fate determinants. So far, most studies addressing the role of miRNAs during neurogenesis were conducted using animal models. With the advent of human pluripotent stem cells and the possibility to differentiate these into neural stem cells, we now have the opportunity to study miRNAs in a human context. More insight into the impact of miRNA-based regulation during neural fate choice could in the end be exploited to develop new strategies for the generation of distinct human neuronal cell types.
KeywordsmicroRNAs Neural development Cell fate specification Human neural stem cells Neuronal subtypes
Initially considered as “junk RNA”, non-coding RNAs are currently perceived as critical regulators of the cellular homeostasis (reviewed by Esteller 2011). In particular, microRNAs (miRNAs), which constitute a distinct class of small non-coding RNAs, have emerged as important post-transcriptional gene regulators. Mature miRNAs arise from large primary transcripts containing hairpin structures that are further processed by the sequential action of two ribonuclease (III) enzymes: Drosha and Dicer. The mature miRNAs are then incorporated into the RNA-induced silencing complex (RISC) and serve as guides to target mRNAs for translational inhibition or mRNA degradation. To date, more than 2,500 miRNAs have been annotated for the human genome (miRBase annotation v20; Kozomara and Griffiths-Jones 2011, 2013) and a large fraction of the known miRNAs is expressed in the human brain (Shao et al. 2010). Considering that each of these miRNAs is predicted to recognize several hundreds of targets, a large proportion of the transcriptome and consequently many cellular processes might be subjected to miRNA-based regulation (Lewis et al. 2005; reviewed by Esteller 2011). This is also the case for the mediation of cell fate decisions, where miRNAs act in synergy with other transcription regulators (transcription factors and epigenetic regulators) to establish gene regulatory networks (Herranz and Cohen 2010; Peláez and Carthew 2012; Arora et al. 2013). In this context, miRNAs and transcription factors can form feed-back or feed-forward loops. Feed-back regulation can be either negative (e.g., a transcription factor limits its own expression by inducing the expression of its own negative miRNA regulator) or positive (e.g., a miRNA reinforces its own expression by targeting its own negative transcription factor regulator). Double-negative feed-back loops, in which the miRNA and the transcription factor reciprocally repress each other, can function as bi-stable switches. Neuronal subtype decisions, in particular, often depend on pairs of cross-repressive transcription factors that might be regulated by miRNAs (e.g., Chen et al. 2011). Feed-forward loops are more complex and consist in two paths of regulation—one direct and one indirect—that can either act in the same (coherent) or in opposite directions (incoherent). MicroRNAs may be also components of feed-forward loops, whereby several different combinations are possible (for a detailed description see Peláez and Carthew 2012). On the one hand, miRNAs may help to ensure the robustness of a gene regulatory network by dampening perturbations and reducing noise. For instance, it was recently shown that miR-9 reduces the impact of genomic variations in Drosophila (Cassidy et al. 2013). On the other hand, miRNAs may also function as critical switches to canalize gene expression during cell fate decisions. This has been nicely demonstrated by the role of miRNAs in establishing chemosensory neuron asymmetry in C. elegans (reviewed by Alqadah et al. 2013).
In this review, we will discuss how miRNAs interact with gene regulatory motifs to regulate neuronal fate decisions. In the first part, we focus on the impact of miRNAs during neural induction and exemplarily highlight the interaction of miR-124 and miR-9 with important regulatory circuits and epigenetic regulators. In the second part, we describe how miRNAs interact with spatial and temporal fate determinants to generate the neuronal diversity found in the central nervous system (CNS). Finally, we will discuss how this knowledge could be harnessed to employ miRNA-based regulation for the derivation of specific neuronal subtypes from human pluripotent stem cells or neural stem cells.
MicroRNAs interact with gene regulatory motifs to regulate neural induction and neuronal differentiation
Many of the miRNAs expressed in the CNS are dynamically regulated both during physiological brain development and in vitro neural differentiation of stem cells, indicating a significant contribution to neural development and function (Krichevsky et al. 2003, 2006; Sempere et al. 2004; Miska et al. 2004; Smith et al. 2010; Liu et al. 2012). Indeed, the emerging picture is that miRNAs play critical roles throughout neural development from neural induction to neural progenitor expansion, differentiation and neuronal subtype specification (reviewed by Sun et al. 2013; Bian et al. 2013). Furthermore, miRNAs are also involved in regulating neuronal migration (e.g., Gaughwin et al. 2011; Rago et al. 2014) as well as neuronal function, neurite outgrowth and synaptic plasticity (reviewed by Siegel et al. 2011; McNeill and Van Vactor 2012). The overall impact of miRNAs as essential regulators of differentiation and neural development was first demonstrated by global loss-of-function experiments via deleting key components of the miRNA processing machinery, i.e., Dicer or Drosha co-factor DGCR8 (Kanellopoulou et al. 2005; Giraldez et al. 2005; Wang et al. 2007; Davis et al. 2008). Since then, several laboratories have taken advantage of the newly developed techniques to selectively modulate the activity of specific miRNAs in order to dissect their functions (reviewed by Akerblom et al. 2012). Nevertheless, considering the large numbers of miRNA species expressed in the CNS, knowledge on miRNA-based regulation during neurogenesis is still at its dawn. This is even more true for human neural development, which, until recently, was not accessible to standardized in vitro experimentation. With the increasing availability of human neural cell types from human pluripotent stem (hPS) cells, there is now the opportunity to study miRNAs in association with human physiology (reviewed by Benchoua and Peschanski 2013). A deeper insight into the role of miRNAs during human neural fate determination could, in the end, also be exploited to develop refined protocols for the generation of specific human neural subtypes.
MicroRNAs regulating the transition of pluripotent stem cells to the neural lineage
When induced to enter neural differentiation, hPS cells undergo specific fate transitions reminiscent of in vivo neural development. This includes the transition of hPS cells to neuroepithelial cells, their segregation into distinct neural progenitors and terminal differentiation into specific neuronal and glial cell types. Furthermore, hPS cells respond to the same extracellular cues regulating neural development in vivo. For instance, during development, neural induction relies on the inhibition of the Activin/TGFβ-mediated pluripotency pathways and the anti-neural effects of BMP (reviewed by Stern 2005). Accordingly, pharmacological blockage of BMP/TGFβ signaling can be used to strongly promote the conversion of hPS cells towards the neural lineage (e.g., Lee et al. 2007; Smith et al. 2008; Chambers et al. 2009). This approach has been designated “dual SMAD inhibition”, since both BMP and Activin/TGFβ signaling converge on SMAD proteins as main signal transduction molecules (Chambers et al. 2009).
MicroRNAs regulating the balance between neural progenitor self-renewal and differentiation
The regulation of neural progenitor proliferation by miR-9 is based on a whole network of other interaction partners. Another node in this complex interplay is a feed-back loop with the orphan nuclear receptor TLX/NR2E1 (Fig. 2). This transcription factor regulates the maintenance and self-renewal of adult neural stem cells via recruitment of HDAC repressors to the promoters of target genes like p21 and Pten (Sun et al. 2007). TLX has been shown to repress miR-9 expression, while miR-9 reduces TLX protein levels in adult neural stem cells (Zhao et al. 2009). However, data from miR-9_2/miR-9_3 knock-out mice suggest that, depending on the developmental stage, miR-9 can associate with the RNA-binding protein Elavl1 to enhance TLX translation instead (Shibata et al. 2011). It has been shown that let-7b and let-7 days target TLX as well and thus might be able to trigger differentiation by de-repression of miR-9 (Zhao et al. 2010, 2013). In addition to its association with Elavl1, miR-9 competes with Elavl2 for binding to target mRNAs. Elavl2 binds to U-rich regions of FoxG1 mRNA, thereby dampening miR-9-mediated FoxG1 suppression (Shibata et al. 2011).
Notch signaling is one of the key pathways regulating neuronal development and expansion of neural progenitors. Both miR-124 and miR-9 have been shown to target several components of the Notch signaling cascade (Fig. 2). While miR-124 targets the Notch ligand Jag1 (Liu et al. 2011) and the Notch down-stream effector Sox9 (Cheng et al. 2009), miR-9 regulates Hes gene family members (Leucht et al. 2008; Bonev et al. 2011, 2012; Coolen et al. 2012). In turn, levels of miR-9 seem to depend on Notch signaling, building up yet another feed-back loop (Coolen et al. 2012; Bonev et al. 2012). In murine neural progenitors, this feedback induces an out-of-phase oscillation of pri-miR-9 and Hes1 levels (Bonev et al. 2012). However, mature miR-9 accumulates over time, limiting the oscillatory period. Presumably, this mechanism participates in timing the differentiation of neural progenitors, since neuronal differentiation is accompanied by high levels of miR-9 and low Hes1 levels (Bonev et al. 2011, 2012). Interaction of miR-9 with the Notch signaling pathway has also been found in Drosophila. Here, Drosophila homolog miR-9a impacts on Notch-mediated lateral inhibition during specification of sensory organ precursors by targeting of dLMO (Li et al. 2006; Biryukova et al. 2009) and senseless (Cassidy et al. 2013). However, these data have to be interpreted carefully, as, in Drosophila miR-9, expression is restricted to epithelial cells (Li et al. 2006) and, therefore, does not reflect the brain-enriched expression profile found in vertebrates (Sempere et al. 2004; Miska et al. 2004; Wienholds et al. 2005).
The data gathered so far illustrate that miR-124 and miR-9/9* add an additional level of intricacy to the highly orchestrated networks underlying neuronal differentiation. However, the list of miR-9 and miR-124 targets is still growing (for a detailed description of additional genes regulated by miR-9/9* and miR-124, please refer to the reviews by Coolen et al. 2013; Akerblom and Jakobsson 2013).
The role of microRNAs during neuronal subtype specification
MicroRNA–target interactions involved in neuronal subtype specification in vivo
ASE chemo-sensory neurons
Induction of ASEL identity
Induction of ASER identity
Mushroom body neurons
Temporal control of neuronal subtype specification
Promotion of cell cycle exit and differentiation, specific impact on survival of forebrain but not hindbrain neuronal progenitors
Xenopus; zebrafish; mouse
FoxG1, several other targets
Important for the generation of Cajal Retzius cells and proper cortical layer formation
MiRNA activity is necessary for proper generation of cortical layers
Saurat et al. 2013
miR-129, miR-155, miR-214, miR-222
Developmental timing of subtype specification
Decembrini et al. 2009
let-7, miR-125, miR-9
Acceleration of progenitor fate progression towards late-born neurons
La Torre et al. 2013
Restriction of DA neuron differentiation
de Chevigny et al. 2012
Delimiting the DV extent of the dopaminergic progenitor pool
Anderegg et al. 2013
Fgfr1, Canopy, Fgf8, Her5, Her9
Maintenance and correct positioning of the midbrain-hindbrain boundary
Leucht et al. 2008
Specification of the p2-pMN progenitor boundary
Chen et al. 2011
Spatial restriction of lumbar motor neuron identity
Asli and Kessel 2010
Specification of spatial MN identity (LMC and MMC column)
Specification of temporal MN identity (switch from early- LMCm to late-born LMCI)
Luxenhofer et al. 2014
The first evidence pointing to a modulatory role of miRNAs in programming neuronal identities came from studies in C. elegans (Johnston and Hobert 2003; Johnston et al. 2005). There are two classes of C. elegans, ASE chemosensory neurons, which are located at the right (ASER) or left (ASEL) side of the worm’s head. Although these neurons share many characteristics with regard to their projection and gene expression profiles, they are functionally divergent and react to different environmental cues. This left–right asymmetry is established by a pair of miRNAs (lys-6, miR-273) and their transcription factor targets (DIE-1, COG-1), which together form a cross-repressive loop. ASEL neurons show high expression levels of lys-6, which directly represses the ASER-promoting transcription factor COG-1. Low COG-1 expression levels allow for the expression of DIE-1 transcription factor, which induces the expression of ASEL genes including lys-6, while repressing the expression of ASER-associated genes. In turn, ASER neurons do not express lys-6 but high levels of COG-1, which induces the expression of miR-273. This miRNA targets DIE-1, thus leading to a de-repression of ASER genes (Hobert 2004; reviewed by Alqadah et al. 2013).
In the vertebrate CNS, some miRNAs exhibit region-specific expression patterns indicating that the different neuronal subtypes residing in these regions may express distinct miRNA profiles (Kapsimali et al. 2007; Landgraf et al. 2007; Kim et al. 2007). Recently, He et al. (2012a) succeeded in analyzing the active miRNA repertoire at a neuron subtype-specific resolution in the adult mouse brain. For this purpose, they used miRNA-tagging and affinity-purification (miRAP), which relies on Cre-induced cell-specific tagging of Argonaute 2 (AGO2) and subsequent co-immunopurification of the tagged-AGO2 and its associated miRNAs (He et al. 2012a). Using this approach, they could demonstrate substantial differences between the expressed miRNA repertoire of glutamatergic neurons and GABAergic interneurons co-expressing either parvalbumin (PV) or somatostatin (SST). For instance, miR-133b and miR-187 were found to be higher expressed in GABAergic neurons than in glutamatergic pyramidal neurons, whereby miR-133b was enriched in the PV-expressing and miR-187 in the SST-expressing GABAergic neurons.
Several studies have addressed the impact of global miRNA loss on the development of specific brain regions by knocking-out Dicer, the key enzyme of miRNA biogenesis. Using the Cre/loxP recombination system, different mouse models for region- or cell-type-specific depletion of Dicer have been developed, e.g., for the retina (Georgi and Reh 2010; for an overview, see Cremisi 2013), cerebral cortex (De Pietri et al. 2008; Saurat et al. 2013; Cremisi 2013), hippocampus (Li et al. 2011a), midbrain (Kim et al. 2007; Huang et al. 2010; Pang et al. 2014) and spinal cord (Zheng et al. 2010; Chen and Wichterle 2012). Interestingly, the impact of Dicer knock-out was variable with regard to the different brain regions targeted and the neuronal subtypes affected. For instance, conditional knock-out of Dicer during late-stage dopaminergic differentiation of mouse ES cells led to a complete loss of dopaminergic neurons, while the number of GABAergic neurons was only reduced by 50 % (Kim et al. 2007). The same study further demonstrated that Dicer depletion in mouse postmitotic midbrain dopaminergic neurons using a DAT-Cre line results in the progressive loss of these cells, which is accompanied by the development of Parkinson’s disease-like symptoms (Kim et al. 2007). A similar loss of dopaminergic neurons due to increased apoptosis was also observed upon specific Dicer deletion in the midbrain of postnatal mice using adenovirus/AAV2-mediated Cre-delivery (Pang et al. 2014).
While Dicer ablation studies have revealed the overall importance of miRNAs in the development and maintenance of different neuronal cell types, further studies have led to the identification of specific miRNAs involved in neuronal subtype specification (Table 1). In the following paragraphs, we will delineate how miRNAs contribute to specifying neuronal subtypes by regulating the spatial or temporal identity of neural progenitor cells. We will further focus on the impact of miRNA-based regulation on neuronal subtype specification in the spinal cord. Finally, we will discuss how miRNAs could be used as tools to modulate the generation of distinct neuronal cell types directly from hPS cells or from hPS cell-derived neural stem cells.
MicroRNAs regulating the spatial identity of neural progenitors
The refinement of the CNS into its main subdivisions along the AP axis (forebrain, midbrain, hindbrain and spinal cord) is regulated by local organizing centers (Lumsden and Krumlauf 1996; Kiecker and Lumsden 2012). One of them, the midbrain–hindbrain boundary (MHB), also called isthmus, regulates the patterning of the midbrain and the anterior hindbrain via Wnt and FGF signaling (Wurst and Bally-Cuif 2001). In Zebrafish, levels of MHB effectors like fgfr1, fgf8 and canopy as well as genes preventing MHB neurogenesis, i.e., her5 and her9, are regulated by miR-9 (Leucht et al. 2008). Interestingly, the MHB is the only part of the Zebrafish neural tube where miR-9 cannot be detected (Leucht et al. 2008). Gain- and loss-of-function studies underlined the importance of miR-9 activity for the maintenance and correct positioning of this organizing center. While ectopic miR-9 expression delimits the spatial extent of the MHB, inhibition of miR-9 causes its expansion along the AP axis (Leucht et al. 2008). Protection of fgfr1 from miR-9 targeting was sufficient to partially rescue MHB formation, underlining the importance of the FGF signaling pathway during this process. However, the effect of miR-9 overexpression on MHB marker expression occurred earlier and was even more pronounced than the changes observed in the Zebrafish fgf8-mutant ace indicating that more targets are at play (Leucht et al. 2008). Accordingly, target protection of the Hes homolog Her5 alone in the presence of miR-9 was also able to rescue the expression of MHB markers (Leucht et al. 2008).
Functional targeting of Hes genes by miR-9 was shown to be conserved in Xenopus and mouse (Bonev et al. 2011, 2012; Coolen et al. 2012). In both organisms, miR-9 acts as a fine-tuner of neurogenesis as part of a negative feed-back loop with Hes genes as described in the first part of this review (Coolen et al. 2012; Bonev et al. 2012). In Xenopus, miR-9 loss causes a failure in neurogenesis along the AP axis by de-repression of the Hes1 homolog hairy1 (Bonev et al. 2011, 2012). The resulting elevated levels of hairy1 were found to promote proliferation through Fgf8, Zic1 and CyclinD1 (Bonev et al. 2011). However, besides its general role in cell cycle exit, the impact of miR-9 on Xenopus neuronal progenitors differed, dependent on the region analyzed (Bonev et al. 2011, 2012). In the Xenopus hindbrain, miR-9 expression is restricted to neural progenitors, which expand upon its loss. In the forebrain, miR-9 is expressed in progenitors as well as developing neurons. Here, loss of miR-9 induces p53-mediated apoptosis, which counteracts the increase in proliferation leading to an unexpected reduction in the total number of neural progenitors (Bonev et al. 2011, 2012). This regional specificity might explain the so far contradictory data gathered on the function of miR-9 with regard to neural progenitor expansion (Zhao et al. 2009; Delaloy et al. 2010; Shibata et al. 2011).
The spatial identity of neural progenitors along the DV axis is also influenced by miRNAs. MiR-7a regulates adult neurogenesis in the olfactory bulb and is expressed in a dorso-ventral gradient in the ventricle walls (de Chevigny et al. 2012). Based on the segmentation of the lateral ventricle walls, defined types of olfactory bulb neurons are generated. Dopaminergic neurons are predominantly generated from progenitors located in the dorsal periventricular zone–a region exhibiting relatively low miR-7a expression (de Chevigny et al. 2012). Dopaminergic specification depends on the transcription factor Pax6, whose 3’UTR carries a functional binding site for miR-7a. Inhibition of miR-7a leads to an increased ventral Pax6 expression and a higher rate of differentiated dopaminergic neurons in the olfactory bulb (de Chevigny et al. 2012).
Recently, it was shown that miR-135a delimits the dorso-ventral extent of dopaminergic progenitors by targeting Lmx1b during murine midbrain development (Anderegg et al. 2013). The FoxA2/Lmx1a/b expression domain was markedly reduced upon ectopic expression of miR-135a–concomitantly with an impaired generation of TH-positive dopaminergic neurons. Furthermore, the dimension of the Wnt1 expression domain and the overall Wnt activity in the developing midbrain were reduced. Overexpression of Lmx1b induced opposite effects, in that the midbrain dopaminergic progenitor domain was expanded and Wnt activity was increased. Interestingly, ectopic expression of Lmx1b increased miR-135a, while depletion of Lmx1b decreased the expression of this miRNA. Thus, miR-135a and Lmx1b might be engaged in a negative feed-back loop in fine-tuning Wnt activity, midbrain progenitor allocation and midbrain size. The impact of miR-135a on Wnt signaling might be in part mediated by the induced down-regulation of Lmx1b. However, miR-135a might also interfere directly with Wnt signaling, since several Wnt molecules were identified as potential miR-135a targets (Anderegg et al. 2013).
Together, these data illustrate how neural identities along the spatial coordinates within the nervous system are modulated by miRNAs fine-tuning the expression of important fate determinants and modulating morphogen signaling.
MicroRNAs regulating temporal fate specification of neural progenitors
During retinal and cerebral cortex development, neural progenitor cells (retinal progenitors and radial glial cells, respectively) proceed through different competence states. This results in the successive emergence of distinct neuronal cell types, which are organized in a laminar pattern according to their neuronal birth order (corticogenesis: reviewed by Greig et al. 2013; retinogenesis: reviewed by Centanin and Wittbrodt 2013). The shift in progenitor competence over time is controlled by several transcription factors and there is mounting evidence that miRNAs play a regulatory role as well (reviewed by Cremisi 2013). Conditional deletion of Dicer during early mouse retinal development resulted in increased and prolonged production of early-born ganglion cells, while the production of late-born cell types was impaired (Georgi and Reh 2010). Similarly, Dicer-null mouse cortical stem cells were only able to produce early-born deep layer projection neurons and failed to generate late-born upper layer neurons (Saurat et al. 2013). The authors of this study proposed that the production of the late-born neurons might critically depend on an active miRNA system. Noteworthy to mention in this context is that the production of Cajal Retzius cells, which emit instructive cues for proper cortical development, is impaired in miR-9_2/9_3 double knock-out mice (Shibata et al. 2011). Accordingly, inhibition of miR-9 using an antisense oligonucleotide resulted in an abnormal development of cortical layers (Shibata et al. 2008, 2011). For more information on the impact of miR-9 during cortical neurogenesis, please refer to the review by Abernathy and Yoo in this Special Issue.
During retinal development, a specific subset of miRNAs has been shown to modulate the fate of neural progenitors by ensuring the correct temporal expression of key transcription factors (reviewed by Cremisi 2013). In Xenopus, the homeobox genes Xotx2 and Xvsx1 are necessary for the generation of bipolar neurons, the last neuronal cell type produced. Although the respective transcripts are already present in early retinal progenitors, Xotx2 and Xvsx1 protein is only detected at later stages due to the action of four miRNAs. These miRNAs, i.e., miR-129, miR-155, miR-214 and miR-222, are down-regulated during retinal development allowing the translation of Xotx2 and Xvsx1 (Decembrini et al. 2009). Inhibition of these miRNAs by transfection of respective antisense oligonucleotides into the optic vesicle resulted in an ectopic generation of bipolar neurons. Interestingly, their expression level is coupled to progenitor cell cycle length, which increases during retinal development and might serve as an intrinsic timer of neural progenitor age (Ohnuma et al. 2002; Decembrini et al. 2009; Pitto and Cremisi 2010). Lengthening of the cell cycle by inhibition of SHH signaling resulted in a decreased expression of miR-129, miR-155, miR-214 and miR-222, while speeding-up of cell cycle progression had the opposite effect (Decembrini et al. 2009). Another group of miRNAs, i.e., let-7, miR-125 and miR-9 has been shown to promote the progression of murine retinal progenitors from early to late fates via targeting Protogenin (Ptrg) and Lin28b, two factors involved in early retinal progenitor competence (La Torre et al. 2013). Overexpression of let-7, miR-125 and miR-9 accelerated progenitor fate progression and development of late-born neurons. In contrast, overexpression of their target genes–Ptrg and Lin28b—retained the progenitor cells at an early competence state. Similar to their function as developmental indicators of retinogenesis, recent findings have pointed to a role of let-7 and miR-125 during temporal fate specification in Drosophila mushroom body (MB) neurons (Wu et al. 2012; Kucherenko et al. 2012). Here, let-7 and miR-125 contribute to the progressive down-regulation of chinmo, which controls MB subtypes specification in a concentration-dependent manner (Wu et al. 2012). In addition to chinmo, let-7 and miR-125 also target abrupt, another temporal regulator of MB subtype specification (Kucherenko et al. 2012). Very recently, it was shown that olfactory bulb (OB) interneurons generated during embryogenesis show no miR-125b expression, while OB interneurons generated during adult neurogenesis exhibit miR-125b expression (Akerblom et al. 2014). Hence, the lack of miR-125b expression appears to distinguish OB interneuron subpopulations generated during different time periods suggesting that miR-125b might be implicated in regulating the temporal appearance of distinct neuronal subtypes. Interestingly, both let-7 and miR-125b regulate temporal fate progression of different lineages during C. elegans development (Olsen and Ambros 1999; Ambros et al. 2003; Ambros 2011). Together, these findings indicate that, although the cellular context is different, similar factors, such as let-7 and miR-125b, take part in controlling developmental fate transitions.
MicroRNAs regulating neuronal subspecification in the spinal cord
Additional evidence for the importance of miRNAs as spatial and temporal regulators can be drawn from data on spinal cord development. The spinal cord, which is subdivided into (11) discrete neuronal progenitor domains along its dorso-ventral axis, is a well-characterized example of spatial neural patterning. Depending on the combinatorial transcription factor code expressed in the progenitors, each domain gives rise to a distinct set of neuronal subtypes, i.e., several classes of interneurons or motor neurons (MN) (reviewed by, e.g., Jessell 2000; Dessaud et al. 2008). Recently, three miRNAs (miR-17-3p, miR-196 and miR-9) have been shown to be involved in the subtype specification of spinal cord neurons.
The first example is miR-17-3p, which is implicated in the DV patterning of mouse spinal cord and affects motor neuron generation (Chen et al. 2011). Patterning of the developing spinal cord along the DV axis is mediated by the combinatorial action of SHH, retinoic acid (RA), BMP and Wnt signaling (reviewed by, e.g., Dessaud et al. 2008; Le Dréau and Martí 2012). These signals lead to the successive induction of key transcription factor determinants including pairs of transcriptional co-regulators. These pairs of cross-repressive transcription factors act as genetic switches in order to ensure unambiguous progenitor cell identity (Briscoe et al. 2000; Dessaud et al. 2010). The boundary between the motor neuron (pMN) and the V2 interneuron (p2) domain is specified by a cross-repressive loop between Olig2 and Irx3, the balance of which is regulated by miR-17-3p. Olig2 is transiently expressed in the early p2 domain and subsequently down-regulated in order to allow Irx3 expression and consolidation of p2 identity (Dessaud et al. 2010; Chen et al. 2011). MiR-17-3p is expressed in the p2 domain and represses Olig2 by direct interaction with its 3’UTR. Loss of this miRNA resulted in an impaired production of V2 interneurons and an expansion of the pMN domain due to persistent Olig2 expression in the p2 progenitors.
The post-mitotic motor neurons generated from the common Olig2-positive pMN domain are further diversified into different motor neuron subtypes, which are arranged in longitudinally oriented columns, i.e., median (M), hypaxial (H), preganglionic (P) and lateral (L) motor columns. The motor neuron columns are characterized by their unique axonal projection patterns to the musculature. For instance, the lateral motor column (LMC) innervates the muscles in the limb, whereas the median motor column (MMC) projects to axial muscles (Jessell 2000; Dasen and Jessell 2009; Philippidou and Dasen 2013). The lateral motor column (LMC) is further split into two subcolumns: the lateral LMCl and medial LMCm subcolumns. The definition of the different motor neuron identities is directed by Hox genes and accessory transcription factors, which themselves may be subject to miRNA-based regulation, e.g., by miR-196 and miR-9 (Yekta et al. 2008; Asli and Kessel 2010; Otaegi et al. 2011). During spinal cord development, miR-196 and Hoxb8 exhibit a mutually exclusive expression pattern along the AP axis. MiR-196 was proposed to act as a post-transcriptional regulator of Hoxb8 ensuring the absence of Hoxb8 expression in the lumbar motor neuron segment (Asli and Kessel 2010). Ectopic expression of Hoxb8 in the lumbar area resulted in an impaired generation of motor neurons. Although inhibition of miR-196 in the lumbar motor neuron column recapitulated the effect of Hoxb8 overexpression, it did not lead to an up-regulation of Hoxb8 protein. Therefore, the authors argued that miR-196 might merely act as a fail-safe mechanism to prevent inappropriate Hoxb8 expression secondary to the direct transcriptional regulation of the Hoxb8 locus. A critical co-factor of Hox-dependent regulation of spatial motor neuron identities is FoxP1. Its expression is restricted to lateral (LMC) and preganglionic (PGC) motor neurons, which are—in contrast to hypoaxial (HMC) and medial (MMC) motor neurons–Hox-sensitive (Dasen and Jessell 2009). Mis-expression or depletion of FoxP1 strongly affects motor neuron diversification and their columnar organization (Dasen et al. 2008; Rousso et al. 2008). In LMC, motor neurons of the chicken spinal cord levels of FoxP1 are fine-tuned by overlapping miR-9 expression (Otaegi et al. 2011). Ectopic expression of miR-9 even switches LMC into MMC, thereby altering the targets of their axonal projections (Otaegi et al. 2011). This effect was counteracted by ectopic FoxP1 expression further indicating that FoxP1 is an important target of miR-9 in MN subspecification (Otaegi et al. 2012). While electroporation of a competing FoxP1 3’UTR induced elevated FoxP1 and reduced HB9 levels, it did not alter the expression of MMC specific Lhx3 (Otaegi et al. 2011). However, a specific miR-9 sponge caused a mild reduction in Lhx3-positive neurons (Otaegi et al. 2012).
Besides the spatial regulation, the generation of the different motor neuron types from their common MN progenitor pool also depends on the temporal progenitor identity. Early progenitors give rise to MMC and early-born medial LMCm motor neurons followed by the production of late-born lateral LMCl motor neurons (Jessell 2000). The motor neurons originate from a unique progenitor pool that is diversified by transcription factors (Isl1, Lhx genes, OC1/Onecut1) as well as secreted molecules like retinoic acid (RA). RA is secreted by earlier-born motor neurons and was shown to induce the expression of specific miRNAs, including miR-9 (Kutty et al. 2010; Laneve et al. 2010). As recently shown, miR-9 might be involved in regulating the transition of progenitor competence from earlier-born to later-born motor neurons by targeting OC1 (Luxenhofer et al. 2014). Inhibition of miR-9 leads to increase of the OC1-positive earlier-born LMCm population resembling the mutually exclusive expression pattern of OC1 and miR-9 in the chick spinal cord (Luxenhofer et al. 2014). OC1 retains the expression of Isl1 favoring the generation of earlier-born Isl1/FoxP1 double-positive motor neurons (Roy et al. 2012). MiR-9-mediated suppression of OC1 allows the induction of later-born motor neuron fate by relieving the Isl1-mediated repression of Lhx1. Therefore, earlier-born motor neurons were reduced upon ectopic miR-9 expression, while the rate of later-born motor neurons positive for Lhx1 and Hb9 was increased (Luxenhofer et al. 2014). Accordingly, a higher rate of earlier-born neurons was produced upon sponge-mediated miR-9 inhibition (Luxenhofer et al. 2014). These findings underline that the development of different neuronal subpopulations in the spinal cord depends on miRNA action in both spatial and temporal dimensions.
MicroRNAs as tools to modulate cell fate and neuronal subtype decisions in vitro
MicroRNAs impacting on in vitro dopaminergic differentiation
Inhibition of the generation of TH-positive neurons (no impact on DA neuron development in miR-133 knock-out mice)
Mouse ES cells
Inhibition of the generation of TH-positive neurons
Mouse ES cells
Yang et al. 2012
Promotion of the generation of TH-positive neurons
Human ES cel- derived lt-NES cells
Stappert et al. 2013
Inhibition of the generation of TH-positive neurons
Using human neural stem cells to study microRNAs in a human context
Most of the findings discussed above rely on experiments in animal model systems and may not always be transferrable to human neural cells (Gao 2009). In order to use miRNAs as tools to generate other neuronal cell types such as cortical, retinal and motor neuron subtypes, there is increasing interest to translate and extend the findings to in vitro differentiation protocols using human pluripotent stem cells (see also Benchoua and Peschanski 2013). However, the generation of mature neuronal cell types from hPS cells via so-called run-through protocols is prone to variability. Proliferative neural stem cells (NSCs) that can be derived from hPS cells as a stable intermediate might be used to minimize this variability. There are several protocols available to derive different NSC populations from hPS cells, such as primitive pre-rosette neuroepithelial stem cells (Li et al. 2011b; Reinhardt et al. 2013), rosette-forming neuroepithelial stem cells (Elkabetz et al. 2008; Koch et al. 2009b) and radial-glial like neural stem cells (Conti et al. 2005). These NSC populations can be distinguished by their morphology, self-renewal capacity and differentiation potential and are likely to represent different developmental stages similar to the range of NSCs generated in vivo. For a comparison of the different NSC populations please refer to the reviews by Conti and Cattaneo (2010) and Karus et al. (2014). Intriguingly, NSCs with similar properties have been successfully isolated from mouse (Hitoshi et al. 2004; Elkabetz et al. 2008) and even human brain (Tailor et al. 2013), indicating that these in vitro-generated NSCs might be a valuable model system for early neural development. Furthermore, given that NSC production and maintenance relies on inhibition of BMP/TGFβ signaling and activation of Wnt and Notch signaling among other signals (e.g., Borghese et al. 2010; Li et al. 2011b; Reinhardt et al. 2013), it is likely that miRNAs regulating these pathways (as discussed above) might also influence NSC fate. Hence, we and others used hPS cell-derived NSCs to first assess stage-dependent miRNA signatures during human neuronal differentiation, a topic that has been difficult to address due to the limited access to primary human neural tissue (Wu et al. 2007; Liu et al. 2012; Stappert et al. 2013). The identified miRNA expression patterns in many cases overlapped with data from previous miRNA profiling analyses performed in rodent models, mouse ES cells or immortalized cell lines (Sempere et al. 2004; Krichevsky et al. 2006; Smith et al. 2010), indicating that many miRNA functions might be conserved between species.
With regard to neuronal subtype specification, we identified two miRNAs that promote the generation of dopamine-like neurons from lt-NES cells (Stappert et al. 2013). Lt-NES cells have a strong differentiation bias to GABAergic neurons. However, they can also give rise to other neuronal phenotypes, such as motor neurons and dopaminergic neurons, when cultured in the presence of patterning cues. For instance, culturing lt-NES cells in the presence of SHH and FGF8b, two morphogenes that are important for dopaminergic neuron specification (Ye et al. 1998), leads to the generation of TH-positive dopamine-like neurons (Koch et al. 2009b; Falk et al. 2012). Furthermore, neuronal subtype specification of differentiating lt-NES cells may be influenced by specific miRNAs (Stappert et al. 2013). By gain- and loss-of-function experiments we could show that miR-181a and miR-125b specifically promote the emergence of TH-positive dopamine-like neurons from lt-NES cells. Interestingly, miR-181a* inhibited the formation of this neuronal population, indicating an intrinsic regulatory mechanism of the bifunctional miR-181a/a* on dopaminergic differentiation. This might also be reflected by the expression levels of miR-181a versus miR-181a*, the ratio of which is increased in human fetal midbrain compared to human whole fetal brain extracts. Moreover, transient delivery of the respective miRNA mimics and inhibitors was sufficient to affect neuronal subtype specification of lt-NES cells and could thus be used to augment the treatment with patterning signals. Thus, it would be interesting to combine miRNA modulation with the recently improved protocols specifically tailored towards efficient generation of midbrain dopaminergic neurons from human pluripotent stem cells (Kriks et al. 2011; Kirkeby et al. 2012; Xi et al. 2012).
The generation of authentic neuronal cell types that fully resemble their in vivo counterparts is still one of the main challenges in stem cell research. For some neuronal cell types, it is even unclear which markers are required to demonstrate the authenticity of the desired cell types. This problem is further aggravated by the fact that the overall knowledge on the gene regulatory circuitries contributing to neuronal diversification and neuronal transmitter-specific functions is rather limited (reviewed by, e.g., Ernsberger 2012; Sandoe and Eggan 2013). The data discussed above demonstrate that miRNAs play important roles in regulating neural differentiation and conferring neuronal identities and should therefore be taken into consideration when annotating neuronal subtype-specific marker gene expression profiles. In this context, subtype-specific reporter cell lines and RNA sequencing analysis could be used to assess the coding and non-coding transcriptome of specific neuronal cell types on a global scale. This information could, in the end, be harnessed to further refine in vitro differentiation paradigms.
Recent findings have placed miRNAs in the midst of gene regulatory networks involved in neural induction, neuronal differentiation and fate specification. MicroRNAs contribute to the establishment of transcriptional codes determining the ground-state of cellular identity. However, knowledge on the impact of miRNA-based regulation during human neural development is still limited–a gap that could be closed by the increasing availability of human neural cell types generated from human pluripotent stem cells. In this context, well-defined populations of human neural stem cells, such as lt-NES cells, could be used to study miRNAs with regard to early human neural development. As indicated by a few pioneer studies, miRNAs could be envisioned as tools to direct the differentiation of pluripotent stem cells and derived neural stem cells towards medically relevant neuronal subtypes. Each miRNA may have numerous mRNA targets and modulating a single miRNA may thus alter the entire differentiation process, making miRNA-based regulation an attractive approach for in vitro specification of neuronal cell fates. Furthermore, miRNA activity may be transiently modulated by applying synthetic miRNA mimics and inhibitors, which could be easily combined with other patterning cues. Finally, deregulation of miRNA activity is associated with many neurodegenerative diseases. Therefore, miRNAs may represent promising targets to develop novel therapeutic approaches (reviewed by, e.g., Junn and Mouradian 2012; Maciotta et al. 2013), whose potential might be evaluated using the iPS cell technology. Altogether, connecting miRNAs to specific functions during human neural development has a great value for the deeper understanding of both physiological and pathological processes in the CNS.
We thank PD Dr. Sandra Blaess, Nityaa Venkatesan and Nils Christian Braun for critical reading of the manuscript. We also thank Johannes Jungverdorben for providing immunofluorescence pictures. Work in the laboratory of Oliver Brüstle is supported by the EU (HEALTH-F4-2013-602278-NeuroStemCellRepair, Eurostemcell HEALTH-F1–2010-241878, FP7-HEALTH-2010-266753-SCR&Tox, COLIPA, 115582-EBiSC), the German Federal Ministry of Education and Research (BMBF; grants 01GS0860-DiGToP, 01GN1008C, 01GN1009B, 0315799-BIODISC), the German Federal Ministry of Education and Research (BMBF) within the framework of the e: Med research and funding concept (grant 01ZX1314A-IntegraMent), BIO. NRW (project StemCellFactory, z0911bt027i), California Institute of Regenerative Medicine and the BMBF (RFA 10–01 CIRM Early Translational II Research Awards; 0316020; VDI: 1316020), BONFOR and the Hertie Foundation.
Conflict of interest
O.B. is co-founder of and has stock in LIFE & BRAIN GmbH. Lt-NES® is a registered trademark of LIFE & BRAIN GmbH. The other authors have no financial interests to disclose.
- Akerblom M, Petri R, Sachdeva R, Klussendorf T, Mattsson B, Gentner B, Jakobsson J (2014) microRNA-125 distinguishes developmentally generated and adult-born olfactory bulb interneurons. Development. doi: 10.1242/dev.101659
- Alqadah A, Hsieh Y-W, Chuang C-F (2013) MicroRNA function in left-right neuronal asymmetry: perspectives from C. elegans. Front Cell Neurosci 7:158Google Scholar
- Anderegg A, Lin H-P, Chen J-A, Caronia-Brown G, Cherepanova N, Yun B, Joksimovic M, Rock J, Harfe BD, Johnson R, Awatramani R (2013) An Lmx1b-miR135a2 regulatory circuit modulates Wnt1/Wnt signaling and determines the size of the midbrain dopaminergic progenitor pool. PLoS Genet 9:e1003973PubMedCentralPubMedGoogle Scholar
- Borghese L, Dolezalova D, Opitz T, Haupt S, Leinhaas A, Steinfarz B, Koch P, Edenhofer F, Hampl A, Brüstle O (2010) Inhibition of Notch signaling in human embryonic stem cell-derived neural stem cells delays G1/S phase transition and accelerates neuronal differentiation in vitro and in vivo. Stem Cells 28:955–964PubMedGoogle Scholar
- Centanin L, Wittbrodt J (2013) Retinal neurogenesis. Development 141:241–244Google Scholar
- Coolen M, Katz S, Bally-Cuif L (2013) miR-9: a versatile regulator of neurogenesis. Front Cell Neurosci 7:220Google Scholar
- Cremisi F (2013) MicroRNAs and cell fate in cortical and retinal development. Front Cell Neurosci 7:1–8Google Scholar
- Dasen JS, Jessell TM (2009) Hox networks and the origins of motor neuron diversity. Curr Top Dev Biol 88:169–200Google Scholar
- De Pietri TD, Pulvers JN, Haffner C, Murchison EP, Hannon GJ, Huttner WB (2008) miRNAs are essential for survival and differentiation of newborn neurons but not for expansion of neural progenitors during early neurogenesis in the mouse embryonic neocortex. Development 135:3911–3921Google Scholar
- Falk A, Koch P, Kesavan J, Takashima Y, Ladewig J, Alexander M, Wiskow O, Tailor J, Trotter M, Pollard S, Smith A, Brüstle O (2012) Capture of neuroepithelial-like stem cells from pluripotent stem cells provides a versatile system for in vitro production of human neurons. PLoS ONE 7:e29597PubMedCentralPubMedGoogle Scholar
- Gao F-B (2009) Context-dependent functions of specific microRNAs in neuronal development. Neural Dev 5:25Google Scholar
- Inui M, Martello G, Piccolo S (2010) MicroRNA control of signal transduction. Nat Rev Mol Cell Biol 11:264–275Google Scholar
- Koch P, Breuer P, Peitz M, Jungverdorben J, Kesavan J, Poppe D, Doerr J, Ladewig J, Mertens J, Tüting T, Hoffmann P, Klockgether T, Evert BO, Wüllner U, Brüstle O (2011) Excitation-induced ataxin-3 aggregation in neurons from patients with Machado-Joseph disease. Nature 480:543–546PubMedGoogle Scholar
- Koch P, Tamboli IY, Mertens J, Wunderlich P, Ladewig J, Stuber K, Esselmann H, Wiltfang J, Brüstle O, Walter J (2012) Presenilin-1 L166P mutant human pluripotent stem cell-derived neurons exhibit partial loss of gamma-secretase activity in endogenous amyloid-beta generation. Am J Pathol 180:2404–2416PubMedGoogle Scholar
- Kriks S, Shim J-W, Piao J, Ganat YM, Wakeman DR, Xie Z, Carrillo-Reid L, Auyeung G, Antonacci C, Buch A, Yang L, Beal MF, Surmeier DJ, Kordower JH, Tabar V, Studer L (2011) Dopamine neurons derived from human ES cells efficiently engraft in animal models of Parkinson’s disease. Nature 480:547–551PubMedCentralPubMedGoogle Scholar
- Landgraf P, Rusu M, Sheridan R, Sewer A, Iovino N, Aravin A, Pfeffer S, Rice A, Kamphorst AO, Landthaler M, Lin C, Socci ND, Hermida L, Fulci V, Chiaretti S, Foa R, Schliwka J, Fuchs U, Novosel A, Muller RU, Schermer B, Bissels U, Inman J, Phan Q, Chien M, Weir DB, Choksi R, De Vita G, Frezzetti D, Trompeter HI, Hornung V, Teng G, Hartmann G, Palkovits M, Di Lauro R, Wernet P, Macino G, Rogler CE, Nagle JW, Ju J, Papavasiliou FN, Benzing T, Lichter P, Tam W, Brownstein MJ, Bosio A, Borkhardt A, Russo JJ, Sander C, Zavolan M, Tuschl T (2007) A mammalian microRNA expression atlas based on small RNA library sequencing. Cell 129:1401–1414PubMedCentralPubMedGoogle Scholar
- Le Dréau G, Martí E (2012) Dorsal-ventral patterning of the neural tube: A tale of three signals. Devel Neurobio 72:1471–1481Google Scholar
- Lewis BP, Burge CB, Bartel DP (2005) Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell 120:6–6Google Scholar
- Li W, Sun W, Zhang Y, Wei W, Ambasudhan R, Xia P, Talantova M, Lin T, Kim J, Wang X, Kim WR, Lipton SA, Zhang K, Ding S (2011b) Rapid induction and long-term self-renewal of primitive neural precursors from human embryonic stem cells by small molecule inhibitors. Proc Natl Acad Sci U S A 108:8299–8304PubMedCentralPubMedGoogle Scholar
- Liu XS, Chopp M, Zhang RL, Tao T, Wang XL, Kassis H, Hozeska-Solgot A, Zhang L, Chen C, Zhang ZG (2011) MicroRNA profiling in subventricular zone after stroke: miR-124a regulates proliferation of neural progenitor cells through Notch signaling pathway. PLoS ONE 6:e23461PubMedCentralPubMedGoogle Scholar
- Liu J, Githinji J, Mclaughlin B, Wilczek K, Nolta J (2012) Role of miRNAs in neuronal differentiation from human embryonic stem cell-derived neural stem cells. Stem Cell Rev Rep 8:1129–1137Google Scholar
- McLaren D, Gorba T, Marguerie de Rotrou A, Pillai G, Chappell C, Stacey A, Lingard S, Falk A, Smith A, Koch P, Brüstle O, Vickers R, Tinsley J, Flanders D, Bello P, Craig S (2013) Automated large-scale culture and medium-throughput chemical screen for modulators of proliferation and viability of human induced pluripotent stem cell-derived neuroepithelial-like stem cells. J Biomol Screen 18:258–268PubMedGoogle Scholar
- Mertens J, Stüber K, Wunderlich P, Ladewig J, Kesavan JC, Vandenberghe R, Vandenbulcke M, van Damme P, Walter J, Brüstle O, Koch P (2013) APP processing in human pluripotent stem cell-derived neurons is resistant to NSAID-based γ-secretase modulation. Stem Cell Rep 2013:1–8Google Scholar
- Ono Y, Nakatani T, Sakamoto Y, Mizuhara E, Minaki Y, Kumai M, Hamaguchi A, Nishimura M, Inoue Y, Hayashi H, Takahashi J, Imai T (2007) Differences in neurogenic potential in floor plate cells along an anteroposterior location: midbrain dopaminergic neurons originate from mesencephalic floor plate cells. Development 134:3213–3225PubMedGoogle Scholar
- Peláez N, Carthew RW (2012) Biological robustness and the role of microRNAs. Curr Top Dev Biol 99:237–255Google Scholar
- Reinhardt P, Glatza M, Hemmer K, Tsytsyura Y, Thiel CS, Höing S, Moritz S, Parga JA, Wagner L, Bruder JM, Wu G, Schmid B, Röpke A, Klingauf J, Schwamborn JC, Gasser T, Schöler HR, Sterneckert J (2013) Derivation and expansion using only small molecules of human neural progenitors for neurodegenerative disease modeling. PLoS ONE 8:e59252PubMedCentralPubMedGoogle Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution License, which permits any use, distribution and reproduction in any medium, provided the original author(s) and the source are credited.