|Home | About | Journals | Submit | Contact Us | Français|
The functional significance of microRNA-9 (miR-9) during evolution is evidenced by its conservation at the nucleotide level from flies to humans but not its diverse expression patterns. Recent studies in several model systems reveal that miR-9 can regulate neurogenesis through its actions in neural or non-neural cell lineages. In vertebrates, miR-9 exerts diverse cell-autonomous effects on the proliferation, migration and differentiation of neural progenitor cells by modulating different mRNA targets. In some developmental contexts, miR-9 suppresses apoptosis and is misregulated in several types of cancer cells, influencing proliferation or metastasis formation. Moreover, downregulation of miR-9 in postmitotic neurons is also implicated in some neurodegenerative diseases. Thus, miR-9 is emerging as an important regulator in development and disease through its ability to modulate different targets in a manner dependent on the developmental stage and the cellular context.
MicroRNAs (miRNAs) are endogenous, noncoding RNAs (~21–23 nucleotides) that destabilize or inhibit the translation of target mRNAs, mostly by binding to their 3′ untranslated regions (3′UTRs).1,2 Owing to their ability to fine-tune the levels of many target proteins, miRNAs are excellent candidates to regulate many complex biological processes. Indeed, recent advances in our understanding of miRNAs have firmly established the notion that this unique class of regulators of gene expression is critically important in many aspects of development and disease.3–9 On the other hand, numerous questions remain to be addressed. For instance, many miRNAs are enriched in the brain and their expression is spatiotemporally regulated.10–13 However, we still do not fully understand how evolutionarily conserved and species-unique miRNAs contribute to the morphogenesis and function of the diverse nervous systems in the animal kingdom.
In this review, we summarize the latest findings regarding the diverse roles of miR-9 in neuronal development and tumor formation in various model systems, including fruit fly, frog, zebrafish, chick, mouse and human embryonic stem cells. These findings illustrate the emerging theme that miR-9, and likely many other miRNAs as well, fulfills diverse functions that are dependent on the developmental stage and cellular context.
The mature miR-9 sequence is identical in insects and humans (Fig. 1A). Of 22 vertebrate species annotated in miRbase14 as containing miR-9, 17 have multiple copies of the gene. All vertebrate miR-9 orthologs are identical in their mature sequence, with the exception of Xenopus tropicalis, which possesses, in addition to three copies identical to those in the other vertebrates, a fourth miR-9 copy containing a single mutation at nucleotide 11 of the mature sequence (Fig. 2). However, this sequence annotated in miRbase needs to be confirmed and validated experimentally.
Between invertebrates, the mature miR-9 sequence is less stringently conserved. Of 30 invertebrates annotated in miRbase14 as containing miR-9 sequences identical to those in vertebrates (named as miR-9a), 22 also possess miR-9 copies that differ in their mature sequence from human miR-9. Based on precursor sequences (Fig. 1B), the relative similarities of different miR-9 homologs can be analyzed using the Close-Neighbor-Interchange algorithm16 (Fig. 2). Although it is difficult to construct a reliable evolutionary tree using relatively short miR precursor sequences, it does seem that different vertebrate pre-miR-9 homologs cluster into three sister groups (red, light blue and purple lines in Fig. 2), which likely have the same common ancestor as that of invertebrate pre-miR-9a (yellow lines in Fig. 2).
Vertebrate miR-9 precursors also produce a miR-9 star strand (miR-9*) that may behave as a functional miRNA (Fig. 1B). Interestingly, the sequence of the miR-9* seed region is identical to that of the invertebrate-specific miR-79. MiR-9/miR-9* and miR-79/miR-79* may share a common ancestor that duplicated and changed its strand expression preference multiple times during evolutionary history, while maintaining the function of both strands in vertebrates and insects.18
Despite the sequence conservation of miR-9, its expression patterns differ strikingly among species. At early embryonic stages in Drosophila, miR-9a is expressed in most epithelial cells except those in the ventral ectoderm. At stage 12, miR-9a is expressed in the ectoderm but not in the central nervous system (Fig. 3A).19,20 In sharp contrast, miR-9 expression in vertebrates appears to be largely confined to the nervous system (Fig. 3).
In developing Xenopus embryos, miR-9 expression is restricted to the anterior nervous system and is absent from the spinal cord (Fig. 3E and F).21 In zebrafish, expression begins in the telencephalon as early as 24 h after fertilization and expands to most regions of the nervous system, including the spinal cord (Fig. 3C and D).22 Nonetheless, some areas such as the midbrain-hindbrain boundary (MHB) in zebrafish and Xenopus brains are devoid of miR-9.21,22 Although information about miR-9 during chick development is incomplete, it is strongly expressed across the nervous system, including the spinal cord (Fig. 3G and H).23,24
In mice, miR-9 is readily detected in the forebrain and hindbrain starting on embryonic day (E) 9.5.25 Initially, miR-9 expression is restricted to the medial pallium and caudal pallidum (E10.5) and subsequently extends to the ventricular zone of the ganglionic eminences and the cortical marginal zone (E11.5) (Fig. 3I and J). From E12.5, miR-9 is also expressed in the septum and in the differentiated zone corresponding to the prospective cortical plate.25 MiR-9 expression peaks between E13.5 and E15.5 and declines afterward but is detectable in both neurons and astrocytes.26,27 MiR-9 expression is retained in the adult subventricular zone, a neurogenic area of the mouse brain.12
As stated above, major spatial differences are observed across vertebrates, suggesting a high level of specialization in miR-9 functions. Thus, along the anteroposterior axis, miR-9 expression is restricted to the anterior nervous system in Xenopus21 but extends also to the spinal cord in the zebrafish, chicken and mouse13,23,25 (Fig. 3). Even in the same domain of expression, miR-9 can be differentially expressed in different cell populations. In developing zebrafish and mouse spinal cord, miR-9 is expressed in proliferating neural precursors.12,13 In the chick, miR-9 is transiently expressed in a pool of spinal motor neurons.24 Xenopus miR-9 seems to be confined to neural progenitor cells (NPCs) in both the forebrain and hindbrain but is expressed only in differentiated forebrain neurons.21
In Drosophila, sensory organ precursor (SOP) cells arise from proneural clusters that express proneural genes and prevent neighboring cells from adopting the SOP cell fate through a well-studied process called lateral inhibition.28 MiR-9a mutant flies have subtle increases in the number of sensory bristles and sensory neurons, which are differentiated from SOP cells, indicating that miR-9a negatively regulates early neurogenesis and ensures the precise specification of SOPs in flies.19 It does so by downregulating Senseless, a transcription factor that promotes SOP specification, in non-SOP epithelial cells.29 Thus, in Drosophila, miR-9 ensures accurate specification of NPCs through its actions in non-neural cell lineages. Drosophila also encodes miR-9b and miR-9c, whose seed sequences resemble those of miR-9a (Fig. 1A). It is not known whether their functions overlap with those of miR-9a.
In zebrafish embryonic brain, the MHB is an organizing center that patterns midbrain and anterior hindbrain development. Fibroblast growth factors secreted from the MHB are involved in the patterning of the surrounding neural tissue, and their activity is tightly controlled by negative feedback inhibition.30 MiR-9 is absent in MHB but expressed in adjacent neural tissue and targets several components of the fibroblast growth factor pathway, thereby limiting the organizing activity of the MHB.22 Loss of miR-9 leads to the expansion of the MHB. Furthermore, miR-9 promotes neurogenesis in adjacent regions by suppressing her5, an antineurogenic basic helix-loophelix Hairy/E(spl) transcription factor.22 Thus, miR-9 prevents the brain region adjacent to the MHB from adopting the MHB fate.
In vertebrates, miR-9 is highly expressed in NPCs, and its context-dependent functions in proliferation have been extensively studied in several model systems. In zebrafish embryos, although miR-9 over-expression leads to a pronounced decrease in proliferation, miR-9 knockdown does not seem to affect the proportion of mitotic cells positive for phosphorylated histone H3 in the ventricular zone.22 Both loss- and gain-of-function approaches reveal that miR-9 suppresses the proliferation of adult NPCs by downregulating the orphan nuclear receptor TLX, an essential regulator of neural stem cell self-renewal, by binding to its 3′UTR.31 A consensus remains to be reached whether miR-9 regulates TLX expression at E13.5.25,31
In contrast, miR-9 increases the proliferation of NPCs derived from cultured human embryonic stem cells at the neurosphere stage.32 Loss of miR-9 decreases proliferation capacity but results in precocious migration of hNPCs or rat embryonic NPCs out of neurospheres without affecting their progenitor identity.32 When hNPCs were transplanted into medial ganglionic eminence (MGE) of brain slices from E14.5 C57BL/6 mouse embryos or the striatum of immunodeficient adult mice 1 week after induction of permanent focal ischemia, more miR-9-deficient hNPCs migrated farther away than control hNPCs from the injection site toward the neocortex or the injury site.32 In NPC proliferation and migration, stathmin, a developmentally regulated cytosolic phosphoprotein that has catastrophe-promoting microtubule-depolymerization activity,33 is one of the key downstream targets.32
During brain development in X. tropicalis, miR-9 exhibits region-specific expression patterns. It is expressed in both NPCs and developing neurons in the forebrain, but is expressed only in NPCs in the mid- and hindbrain.21 Along the anteroposterior axis, endogenous miR-9 limits NPC proliferation through a single key target, hairy 1, a member of the Hes family of transcriptional regulators that is specifically expressed in neurogenic regions of the brain.21
All the studies described above used acute knockdown or overexpression approaches. A recent genetic deletion study in mice revealed further complexity in the context-dependent functions of miR-9 in vivo.26 In mammals, miR-9 is encoded by three genes that also produce miR-9* (Figs. 1 and and22). miR-9-2 or miR-9-3 knockout mice are viable and grossly normal; however, miR-9-2/3 double-knockout mice die within a week after birth.26 From E12.5 to E13.5, the number of mitotic cells that were positive for phosphorylated histone H3 and labeled by a 30 min pulse of bromodeoxyuridine is increased in the subventricular and ventricular zones, indicating increased proliferation, which correlated with the upregulation of Foxg1, a direct target of miR-9. However, by E16.5, Foxg1 is no longer regulated by miR-9, probably because of interference by Elavl2/HuB, an AU-rich RNA-binding protein whose expression is increased at this developmental stage.26 The absence of apoptosis and the reduced neuronal differentiation would lead one to expect that the total number of progenitor cells would be significantly higher at E16.5 in miR-9-2/3 double-knockout mice, although this was not examined directly. Interestingly, from E15.5 to E18.5, NPC proliferation in the pallium of miR-9-2/3 double knockout mice was significantly suppressed, which correlated with elevated expression of Nr2e1, also known as TLX.26 The regulation of Nr2e1/TLX by miR-9 at E16.5 but not at E13.5 seems to result from a difference in the cellular context due to elevated expression of Elavl1/HuR at E16.5.26 Thus, this genetic analysis suggests that miR-9 can either promote or suppress NPC proliferation through different targets at different stages of brain development.
In miR-9-2/3 double-knockout mice, the tangential migration of interneurons into the pallium was compromised,26 which is consistent with the observation that overexpression of miR-9 promoted outward migration of apparently differentiated neurons from embryonic ventricular zone.31 It remains to be determined whether and how the migration of other NPCs or newly generated neurons is regulated by miR-9 in vivo.
A key function for miR-9 is to regulate the generation of postmitotic neurons from NPCs. Ectopic expression of miR-9 in mouse cerebral cortex at E11.5 induces premature differentiation of Cajal-Retzius cells. Acute miR-9 knockdown at E12.5 reduces the number of these cells but has little effect on the number of cortical NPCs.25 This finding was largely confirmed in miR-9-2/3 knockout mice, which have greatly reduced numbers of Cajal-Retzius cells and other early born neurons.26 Although the total number of NPCs was not analyzed at this developmental stage in the knockout mice, the miR-9 phenotype in both studies suggests a role in neuronal differentiation, since Foxg1, which suppresses Cajal-Retzius cell differentiation, is a key direct target of miR-9 and is upregulated at E13.5.26 Consistent with this finding, miR-9 overexpression in the ventricular zone at E13.5 causes precocious neuronal differentiation,31 and miR-9 knockdown in mouse embryonic stem cells increases the number of glial cells at the expense of neuron production.34
MiR-9 is expressed not only in mouse embryonic NPCs but also in adult NPCs. Indeed, overexpression of miR-9 in adult NPCs promotes neuronal differentiation, which is mediated by the interaction between miR-9 and the mRNA encoding the nuclear receptor TLX.31 The effect of overexpressed miR-9 on TLX expression is relevant only in adult NPCs shortly before terminal differentiation is induced with retinoic acid or forskolin, suggesting that miR-9 alone rather facilitates, but cannot induce, neural differentiation of adult NPCs. Because acute inhibition of miR-9 activity does not affect the neuronal differentiation of adult NPCs,31 the exact physiological functions of miR-9 in adult mouse brain remain to be clarified.
The functional significance of miR-9 in neuronal differentiation has been demonstrated in other experimental systems. In zebrafish embryos, miR-9 knockdown reduces the production of HuC-positive postmitotic neurons, an action mediated in part by her5 and her9, 2 proneural transcription factors targeted by miR-9.22 During brain development in X. tropicalis, miR-9 is required for neuronal differentiation in all regions along the anteroposterior axis.21 In chick spinal cord, miR-9 has a dynamic expression pattern and regulates the differentiation of motor neuron subtypes by fine-tuning the expression of FoxP1, which defines motor neuron identity in lateral, medial and preganglionic motor columns.24 While miR-9 plays a central role in the differentiation of motor neuron subtypes, it does not seem to affect overall neurogenesis, as the number of NPCs and differentiated neurons remains normal,24 which is in stark contrast to its roles in the brains of several model organisms.
MiR-9 and miR-9* also seem to regulate the morphological differentiation of postmitotic neurons. Ectopic expression of miR-9a in Drosophila larval sensory neurons significantly increases dendritic branching.35 In mice, miR-9* targets the 3′UTR of BAF53a, a key component of ATP-dependent chromatin-remodeling complexes.36 Expression of BAF53a does not affect activity-dependent dendritic growth of cultured rodent hippocampal neurons. However, BAF53a with a mutant 3′UTR lacking miR-9* and miR-124 binding sites was not regulated simultaneously by these miRNAs and inhibited this differentiation process.36 Thus, miR-9* together with miR-124 may contribute to activity-dependent dendritic growth by downregulating BAF53a.
A potential role for miR-9 in apoptotic cell death has been suggested by studies in some but not other model systems. Acute knockdown of miR-9 does not affect the survival of hNPCs in newly formed neurospheres or embryonic rat NPCs, as shown by the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) analysis and a cytotoxicity assay.32 Similarly, the number of TUNEL-positive or active caspase3-positive cells was unchanged in the developing telencephalon of miR-9-2/3 double knockout mice.26 In contrast, treatment with antimiR-9 morpholino in the forebrain but not the hindbrain of stage 30 Xenopus embryos significantly increased apoptosis of NPCs.21 How miR-9 specifically affects apoptosis only in the forebrain is unknown, although it likely involves the p53 pathway.21 These findings suggest that, under certain circumstances, miR-9 helps to suppress apoptotic cell death in vertebrate NPCs.
A context-dependent role in apoptosis seems to be an ancient feature of miR-9 function. A prominent phenotype in miR-9a knockout flies is a completely penetrant notching defect in the posterior wing margin.19,37,38 MiR-9 is highly expressed in most, if not all, epithelial cells in the wing disc,19 but the notching phenotype is much more pronounced at the posterior than the anterior wing margin.19,37,38 The defect is caused by apoptotic cell death and the transcriptional regulator Drosophila LIM only (dLMO) is the key target of miR-9 in this process.37,38 Loss of one copy of dLMO rescued the apoptosis and wing margin defects in miR-9a mutants.37,38 It is not known why the posterior wing margin is more sensitive to miR-9a activity than the anterior wing margin.
Although much has been learned about the cellular functions of miR-9, we still do not understand how it contributes to brain function and dysfunction. In adult rat brain, alcohol increases miR-9 expression in supraoptic nucleus neurons and striatal neurons.39 A key factor in the development of alcohol tolerance is the large conductance calcium- and voltage-activated potassium channel. MiR-9 contributes to drug adaptation and adult brain plasticity by downregulating specific mRNA splice variants of this channel.39
The potential involvement of miRNAs in age-dependent neurodegeneration is suggested by findings in conditional dicer knockout mice in which specific types of neurons are lost;40–42 however, the exact mechanisms are unclear. In principle, loss of miRNAs can cause accumulation of toxic proteins and subsequent neuronal loss and might affect neuronal survival by changing the levels of trophic or prosurvival factors. Recent evidence suggests that miR-9 may play a role in neurodegenerative diseases. Increased miR-9 levels have been found in postmortem Alzheimer disease brains.43 MiR-9 and miR-9* are downregulated in motor neurons differentiated from mouse embryonic stem cells (mESCs) carrying the SMN1 mutation. MiR-9 downregulation leads to increased expression of heavy neurofilament, which alters the intermediate filament dynamics in this model of spinal muscular atrophy.44 Interestingly, miR-9 levels are dramatically decreased after ischemia due to middle cerebral artery occlusion in rats.45
In neuronal cells, RE1-silencing transcription factor (REST) is sequestered in the cytoplasm in part as a result of binding to huntingtin, the protein in which an abnormal polyglutamine expansion causes Huntington's disease. In non-neuronal cells, REST suppresses neuronal gene expression by binding to RE1 consensus sequences and recruiting CoREST and other corepressors.46 In Huntington's disease, polyglutamine expansions in huntingtin abolish binding to REST, which then translocates to the nucleus and represses the expression of neuronal genes, including miR-9-1, miR-9-2 and miR-9-3.46,47 Indeed, miR-9 and miR-9* levels are reduced in the early stages of Huntington's disease. Interestingly, miR-9 targets REST and miR-9* targets CoREST, providing a negative feedback loop between the REST silencing complex and miRNAs.46,47 It remains to be determined whether the deregulated expression of miRNAs is directly involved in disease progression and contributes to neuronal cell death.
Aberrant miR-9 levels have been reported in many types of cancer, suggesting that miR-9 is involved in tumor formation or progression. However, by regulating various mRNA targets, miR-9 may have opposing effects on proliferation in different types of cancer cells.48–63 For instance, miR-9 is overexpressed in human Hodgkin's lymphoma cells,51 primary brain tumors53,57 and CDX2-negative gastric cancer cells.63 MiR-9 knockdown inhibits the proliferation of human gastric cancer cells and overexpression of CDX2, a direct target of miR-9, has a similar effect.63 In contrast, miR-9 is downregulated in human ovarian tumor cells and overexpression of miR-9 suppresses their proliferation, in part by downregulating NFkappaB1.50,56 Overexpression of miR-9 in human neuroblastoma and medulloblastoma cells also inhibits cell growth.47,54
MiR-9 seems to be a useful marker for tumor metastasis, but its role in this process is also dependent on the type of cancer. For instance, miR-9-3 is downregulated in breast cancers with vascular invasion or lymph node metastases.48 MiR-9 expression is also significantly lower in metastatic than primary brain tumors.53 Reduced miR-9 expression due to hypermethylation is associated with metastasis of cancer cells in the lymph node52 and metastatic recurrence in patients with clear-cell renal cell carcinoma.61 In contrast, miR-9 expression is significantly higher in breast cancers with metastases than in those without metastases and seems to promote metastasis.60
MiR-9 induces epithelial mesenchymal transition in human epithelial cells and breast carcinoma cell lines by targeting the adherens junction protein E-cadherin.60 Cells overexpressing miR-9 lose contact with each other, and their motility and migration are enhanced in vitro. MiR-9 also induces transcription of vascular endothelial growth factor, in part by downregulating E-cadherin, which leads to release of β-catenin from the junctions and translocation into the nucleus. However, additional miR-9 targets might also be responsible for increased transcription of vascular endothelial growth factor and miR-9-induced migration, since E-cadherin knockdown does not mimic the miR-9 overexpression phenotype in vitro. When breast cancer cells overexpressing miR-9 are implanted in mice, the tumors show enhanced angiogenesis and grow faster than those formed by cells with low levels of miR-9. These findings imply that miR-9 promotes metastasis of epithelial tumors. Indeed, forced expression of miR-9 in nonmetastatic breast tumors facilitates the formation of pulmonary micrometastases and miR-9 knockdown inhibits metastasis of highly metastatic cells.60
There is strong evidence that miR-9 acts at different stages of neurogenesis and coordinates the complex genetic programs necessary for the proper generation of postmitotic neurons. The miR-9-2/3 double-knockout mice recapitulate many defects in cortical development observed in a mouse model in which cortical NPCs are devoid of all mature miRNAs,64 confirming the prominent role of miR-9 as an intrinsic positive regulator in this process. In contrast, Drosophila miR-9a seems to negatively regulate early neurogenesis through its actions in non-neural cell lineages.
MiR-9 can exert opposite effects on NPC proliferation, migration and differentiation, depending on the developmental stage and cellular context. Although miR-9 is enriched in vertebrate brain, its expression pattern is spatially and temporally regulated. Accordingly, miR-9 uses different targets to mediate its diverse functions. The recent finding that some RNA binding proteins can affect the interaction between miR-9 and its targets might also explain in part how different cellular contexts can influence miR-9 function.
In addition to its role in neurogenesis, miR-9 seems to be important for apoptosis in some developmental contexts. Moreover, aberrant expression of miR-9 in many cancer types makes it an important candidate as a diagnostic and prognostic marker. MiR-9 has also been implicated in other human pathological conditions such as neurodegeneration. However, it remains to be seen whether miR-9 contributes significantly to pathogenic processes. Further understanding of the functions of miR-9 and its target interactions in various cellular contexts will shed light on both development and disease.
We thank S. Ordway for editorial assistance, A. Wilson for help with graphics and lab members for discussions. We also thank Drs. S. Aizawa, L. Bally-Cuif and T. Sun for sharing the original high-resolution images reproduced in Figure 3. This work was supported by grant RO1 NS066586 from the National Institutes of Health (F.B.G.).