NeuroMolecular Medicine

, Volume 11, Issue 3, pp 200–207

Macro Role(s) of MicroRNAs in Fragile X Syndrome?


  • Xuekun Li
    • Department of Human GeneticsEmory University School of Medicine
    • Department of Human GeneticsEmory University School of Medicine
Review Paper

DOI: 10.1007/s12017-009-8081-2

Cite this article as:
Li, X. & Jin, P. Neuromol Med (2009) 11: 200. doi:10.1007/s12017-009-8081-2


Fragile X syndrome (FXS), the most common form of inherited mental retardation, is caused by the loss of functional fragile X mental retardation protein (FMRP). FMRP is an RNA-binding protein that can regulate the translation of specific mRNAs. It is known to regulate synaptic development through the regulation of local protein synthesis in synapses. MicroRNAs (miRNAs) are a class of small noncoding RNAs involved in almost every biological process. They exhibit spatiotemporal expression during brain development, and some miRNAs play important roles in neural development. A growing body of evidence now implicates the miRNA pathway in the molecular pathogenesis of FXS. Here we review the current state of knowledge about the microRNA pathway in neural development and the emergence of possible roles for miRNAs in FXS.


MicroRNAsFragile X syndromeSynaptic plasticityFMRP

Fragile X syndrome (FXS), which affects one in 4000 males and one in 8000 females, is one of the most common inherited mental retardation disorders (Turner et al. 1996). The clinical presentations of fragile X syndrome include mild to severe mental retardation, with IQs between 20 and 70, mild abnormal facial features, such as a prominent jaw and large ears, mainly in males, and macroorchidism in postpubescent males (Warren and Nelson 1994). The gene responsible for FXS, fragile X mental retardation 1 (FMR1), was first identified by positional cloning in 1991 (Verkerk et al. 1991).

FMR1 consists of 17 exons spanning around 38 kilobases (kb) to Xq27.3 (Ashley et al. 1993). Within the 4.4 kb of FMR1 transcript containing a 190-bp 5′ untranslated region (5′ UTR) and a 2281-bp 3′ UTR, a CGG trinucleotide repeat is located in the 5′ UTR (Eichler et al. 1993). This CGG repeat is highly polymorphic in length and content and is often interrupted by AGG trinucleotides among normal individuals. The normal repeat size ranges from 7 to 40, with 30 repeats found on the most common allele (Verkerk et al. 1991). A massive CGG repeat expansion has been associated with fragile X syndrome (Verkerk et al. 1991). In most affected fragile X patients, the CGG repeats expand to more than 230 (full mutation) and become abnormally hypermethylated, which result in silencing of the FMR1 gene (Fig. 1). Identification of other mutations in the gene, such as deletions and point mutations among patients with typical fragile X phenotypes, strongly suggests that FMR1 is the only gene involved in the pathogenesis of fragile X syndrome and that the functional loss of fragile X mental retardation protein (FMRP) is the cause of FXS (Kenneson et al. 2001).
Fig. 1

Schematic view of the FMR1 gene and FMRP. a. The FMR1 gene contains 17 Exons and spans 38 kb on the X chromosome. Highly polymorphic CGG repeats are located in the 5′ UTR of the FMR1 gene. Among normal individuals, the length of the CGG repeat is under 40 repeats. If the repeats expand to between 60 and 200 or to more than 200, this is called the premutation or full mutation allele, respectively. b. FMRP protein contains five well-known domains: NLS nuclear localization signal, KH1 and KH2 RNA-binding domains, NES nuclear export signal, and RGG RGG box, RNA binding

MicroRNAs (miRNAs) are a new class of small noncoding small RNAs that are ~22 nucleotides (nt) in length and generated from endogenous double-strand transcripts (Bartel 2004). Typically, miRNAs bind to the 3′ UTR of target mRNA, leading to the translational repression or mRNA degradation. Recent studies have shown that the miRNA pathway could be involved in the regulation mediated by FMRP (Jin et al. 2004a; Chang et al. 2009). In this review, we will summarize the current state of knowledge about the miRNA pathway and its role in brain development and discuss the potential role(s) of miRNAs in fragile X syndrome.

Biogenesis and Functional Mechanism of miRNAs

After discovery of the first miRNA, lin-4, in Caenorhabditis elegans in 1993 (Lee et al. 1993), nearly 500 miRNA have since been identified, and more than 1000 miRNAs are predicted bioinformatically to exist in the human genome (Bartel 2009; Kim et al. 2009). MiRNAs account for about 1–5% of animal genes (Ambros 2004; Bartel 2004) and may regulate more than a third of protein-coding genes in humans (Miranda et al. 2006).

In mammals, the majority of endogenous miRNA genes are transcribed initially as primary transcripts (pri-miRNAs) that range from hundreds to thousands of nucleotides in length and contain one or more extended hairpin structures (Du and Zamore 2005). The nuclear RNase III enzyme Drosha, working with DGCR8, cleaves both strands near the base of the primary stem-loop and yields the precursor miRNA (pre-miRNA) (Fig. 2). After being exported to the cytoplasm by exportin-5/RanGTP, pre-miRNAs are further cleaved by the RNase III Dicer, along with a dsRNA-binding protein, TAR RNA-binding protein (TRBP) (Du and Zamore 2005). The Dicer-TRBP complex is also required for the processing of short hairpin RNA (shRNA) into small interference RNA (siRNA) of ~21 bp. After cleavage by Dicer and unwinding by RNA helicase, one strand of the miRNA/miRNA* or siRNA duplex (the antisense, or guide strand) is then preferentially incorporated into the RNA-induced silencing complex (RISC), while the other strand (the sense, or passenger strand) is degraded (Fig. 2). The RISC is a large and heterogeneous multi-protein complex. Components of the RISC that have been identified include Dicer, TRBP, and Argonaute 2 protein (AGO2) (Du and Zamore 2005). The RISC uses the guide RNA to find complimentary mRNA sequences via partial Watson-Crick base pairing, which leads to posttranscriptional gene silencing (PTGS) through inhibition of either translation initiation or elongation (Bartel 2004; Du and Zamore 2005). MiRNA could also negatively regulate protein expression through targeting of mRNA coding regions (Tay et al. 2008). The interaction between miRNAs and their targets could establish a feedback loop: miRNA inhibiting target expression, whereas target inducing miRNA transcription. Hence a single miRNA may simultaneously regulate the expression of multiple mRNA targets and thereby act as a rheostat to fine-tune protein expression (Kim et al. 2007; Baek et al. 2008; Selbach et al. 2008).
Fig. 2

A diagram illustrating the miRNA pathway and its interaction with FMRP. MiRNA gene is transcribed by RNA polymerase II, and the primary transcript is processed by Drosha. The produced precursor miRNA is exported into the cytoplasm and further processed by Dicer. Mature miRNA is loaded into the RISC and binds to target mRNA through base pairing. Current studies also support the following model: FMRP binds to some mRNAs, and miRISC could interact with FMRP-mRNA complex, thereby leading to translational inhibition

Besides translational suppression, miRNAs are also found to upregulate the translation of target mRNAs in a cell cycle-dependent manner, switching between translational suppression in proliferating cells to translational activation in quiescent cells (Vasudevan et al. 2007). Furthermore, miRNA can regulate transcription, as well (Obernosterer et al. 2006; Kim et al. 2008). MiR-320 is located in the promoter region of the cell cycle gene POLR3D. Overexpression of miR-320 induced the enrichment of Argonaute 1 (AGO1), Polycomb group component EZH2, and tri-methyl histone H3 lysine 27 to the promoter of POLR3D, suggesting miR-320 could indirectly regulate the transcription of POLR3D (Kim et al. 2008). However, whether these observations reflect the general function of miRNAs remains to be determined.

Role of the miRNA Pathway in Neural Development

Studies over the last several years have brought major advances in our understanding of the biological roles of miRNAs and revealed that miRNAs play essential functions in multiple biological pathways and disease conditions, such as developmental timing (Abbott et al. 2005; Giraldez et al. 2005), fate determination (Karp and Ambros 2005), apoptosis, metabolism (Chang and Mendell 2007; Bushati and Cohen 2008), immunity response (Baltimore et al. 2008), tumorigenesis (He et al. 2007), and neurodegenerative disorders (Nelson et al. 2008). Here we choose to focus on the roles of miRNA in the central nervous system.

To date, around 70% of known miRNAs are found to be expressed in the central nervous system, with some of them enriched in the brain, and the expression of microRNAs in the brain exhibits dynamic spatial and temporal patterns (Cao et al. 2006; Kosik 2006). Profiling studies have discovered that some miRNAs are enriched in the developmental and adult central nervous system (Kapsimali et al. 2007; Makeyev et al. 2007), suggesting that they might play important roles in brain development and normal brain function. So far, studies have shown that miRNAs regulate neuronal development (Pasquinelli and Ruvkun 2002; Visvanathan et al. 2007; Leucht et al. 2008) and neurogenesis (Krichevsky et al. 2006; Li et al. 2006; Choi et al. 2008; De Pietri Tonelli et al. 2008; Ferretti et al. 2008), and also play a part in neurological disorders (Kim et al. 2007; Cuellar et al. 2008; Wang et al. 2008).

Considering the key role of Dicer in the miRNA pathway, some studies have investigated the function of miRNA by removing Dicer in animals. In Drosophila, there are two Dicer isozymes, Dicer-1 and Dicer-2; Dicer-1 is required for the miRNA pathway, whereas Dicer-2 is required for the siRNA pathway (Lee et al. 2004). Through examination of the mutant flies, Dicer-1 was found to modulate the wiring specificity of olfactory projection neurons (Berdnik et al. 2008). The dendrites of Dicer-1 mutant projection neurons could not target to the specific glomerulus, and axonal terminations were altered (Berdnik et al. 2008). In zebrafish, Dicer mutation arrests embryonic development around 10 days (Wienholds et al. 2003), and mutant zebrafish display abnormal morphogenesis during gastrulation, brain formation, somitogenesis, and heart development (Giraldez et al. 2005), whereas injection of miR-430 miRNAs rescues the brain defects, indicating a function for miR-430 in brain development (Giraldez et al. 2005). Since the deletion of Dicer in mice led to lethality (Bernstein et al. 2003), researchers generated conditional knockouts of Dicer for further study. Dicer conditional knockout ES cells displayed severe defects, and re-expression of Dicer could rescue these phenotypes (Kanellopoulou et al. 2005). Conditional knockout of Dicer in postnatal mice leads to increased cortical apoptosis and enlarged lateral ventricle size, and decreased brain weight and dentate gyrus size (Davis et al. 2008). The development of their hippocampal neurons is also disrupted; including decreased dendritic branching and hippocampal formation size and increased dendritic spine length (Davis et al. 2008). Furthermore, specific ablation of Dicer in the dorsal telencephalon during their embryonic development results in abnormal brain development, including a smaller postnatal cortex, disrupted neuronal layer formation of the cortex, and increased apoptosis (De Pietri Tonelli et al. 2008). Overall, these studies indicate that the miRNA pathway is essential for development of the central nervous system.

A series of studies have also revealed roles of individual miRNAs in neural development. MiRNA profiling studies found that specific miRNAs, particularly miR-124 and miR-9, significantly increased upon neuronal differentiation (Krichevsky et al. 2006). MiR-124 is a brain-specific miRNA, which is conserved from invertebrates to mammals. It can directly inhibit small C-terminal domain phosphatase 1 and promotes neuronal differentiation in the developing spinal cord of chick (Visvanathan et al. 2007). MiR-124 can also repress PTBP1, a global repressor of alternative pre-mRNA splicing in non-neuronal cells, and induce nervous system-specific alternative splicing in neuroblastoma cell lines, which will promote neuronal differentiation (Makeyev et al. 2007). MiR-9 is highly expressed in the developing medial pellium of mouse, and the ectopic overexpression of miR-9 in mouse embryo through in utero electroporation can interrupt neuronal differentiation in the cortex (Shibata et al. 2008). Furthermore, miR-132 expression could be activated by the calcium response element binding protein and could directly suppress the translation of P250GAP (Vo et al. 2005). Lastly, miRNAs and their precursors have also been identified that are localized in synapses (Schratt et al. 2006; Lugli et al. 2008; Siegel et al. 2009). One of them, miR-134, could negatively regulate the size of dendritic spines and spine development through inhibiting LIMK 1 (Lim-domain containing protein kinase 1) (Schratt et al. 2006). Further studies have revealed that myocyte enhancing factor 2 (Mef-2) regulates the transcription of miR379-410 cluster, including miR-134 (Fiore et al. 2009). MiR-134 could regulate neuronal activity-induced dendritic development by inhibiting Pumilio2. Another miRNAs, miR-138, was also found in synaptosomes and the inhibition of miR-138 activity led to a significant increase of spine volume, but not affecting spine density by targeting on acyl-protein thioesterase 1 (APT1) (Siegel et al. 2009). These studies demonstrated how activity-dependent gene expression, global transcriptional control and local post-transcriptional control orchestrate in a neuron. These observations together indicate that miRNAs provide a new dimension to the regulation of neuronal development.

MiRNA Pathway in Fragile X Syndrome

FMRP and its autosomal paralogs, the fragile X-related proteins FXR1P and FXR2P, constitute a small family of RNA-binding protein (the fragile X-related gene family) (Siomi et al. 1995; Zhang et al. 1995). These proteins share over 60% amino acid identity and contain two types of RNA-binding motif: two ribonucleoprotein K homology domains (KH domains) and a cluster of arginine and glycine residues (RGG box). FMRP is associated with polyribosomes via messenger ribonucleoprotein (mRNP) particles (Ashley et al. 1993; Feng et al. 1997b). FMRP is expressed in many tissues, including brain and testis. Particularly, FMRP is highly expressed in neuron, and localized in dendrites and spines (Feng et al. 1997a; Antar et al. 2004). FMRP is known to play a role in translational control and can suppress translation both in vitro (Laggerbauer et al. 2001; Li et al. 2001) and in vivo (Muddashetty et al. 2007). The current working model is that FMRP binds target mRNA in the nucleus to form a ribonucleoprotein complex which is transported to dendrites and spines, where mRNAs are locally translated in response to stimuli and FMRP involves in the regulation of protein synthesis (Bassell and Warren 2008). Neuroanatomical studies have found that neurons of FXS patients have more long spines, but fewer short spines, and more immature (long and thin) spines and increased spine density well, suggesting neurons in FXS have impaired synapse development (Irwin et al. 2000; Irwin et al. 2001). These abnormalities of neuronal development have also been found in Fmr1 knockout (KO) mice (Comery et al. 1997; Nimchinsky et al. 2001; Grossman et al. 2006). Proteomic studies showed that in Fmr1 KO mice more than 100 proteins displayed the altered expression, and these proteins are involved in regulating synaptic structure, neurotransmission, dendritic mRNA transport (Liao et al. 2008). Based on these observations, FMRP is proposed to be involved in synaptic plasticity via regulation of mRNA transport and local protein synthesis of specific mRNAs at synapses. Due to space limitations, the next we will focus on the potential link between the miRNA pathway and fragile X syndrome (For a detailed discussion on the molecular pathogenesis of fragile X syndrome, we refer the reader to the recent review by Bassell and Warren) (Bassell and Warren 2008).

FMRP interacts biochemically and genetically with known components of the miRNA pathway. Experiments in Drosophila revealed specific biochemical interactions between dFmrp and functional RISC proteins, including dAGO1, dAGO2, and Dicer (Caudy et al. 2002; Ishizuka et al. 2002; Xu et al. 2004). dFmr1 has a strong genetic interaction with dAGO1, and dAGO1 dominantly interacted with dFmr1 in both dFmr1 overexpression and loss-of-function models (Jin et al. 2004b). Furthermore, dFmr1 also interacts genetically with AGO2, as exemplified by their ability to coregulate ppk1 mRNA levels (Xu et al. 2004). Other studies yield further evidence for the involvement of FMRP in miRNA-containing RISC and P body-like granules in Drosophila neurons (Barbee et al. 2006). Recombinant human FMRP is able to act as an acceptor for Dicer-derived miRNAs, and importantly, endogenous miRNAs themselves are associated with FMRP in both flies and mammals (Caudy et al. 2002; Ishizuka et al. 2002; Jin et al. 2004b). In the adult mouse brain, Dicer and eIF2c2 (the mouse homolog of AGO1) interact with FMRP at postsynaptic densities (Lugli et al. 2005). Presumably, this interaction works to regulate translation of target mRNAs in an activity-dependent manner. Recently it was shown that the phosphorylation of FMRP could increase its association with miRNA precursors (Goldblatt et al. 2009). Based on these findings, RISC proteins, including Argonaute (AGO) and Dicer, may possibly interact with FMRP and use the loaded guide miRNA(s) to interact with target sequences within the 3′ UTR of RNA bound to FMRP and suppress its translation (Jin et al. 2004a). In this model, FMRP facilitates the interaction between miRNAs and their target mRNA sequence, ensuring proper targeting of guide miRNA-RISC within the 3′ UTRs and proper translational suppression (Fig. 2). Intriguingly, another member of the fragile X-related (FXR) protein family, FXR1, has also been implicated in miRNA-mediated translational upregulation through an association with AGO2 on AU-rich 3′ UTRs in quiescent cells (Vasudevan and Steitz 2007; Vasudevan et al. 2007). Nonetheless, the relevance of these observations to FMRP-mediated translational regulation has not been established.

The fact that FMRP is associated with Dicer, miRNAs, and specific mRNA targets raises the question of whether FMRP is associated with specific miRNAs and modulates their processing. To address this question, the expression and processing of miRNAs were examined in Drosophila dFmr1 mutants; in the fly brain, dFmrp was specifically associated with miR-124a, a nervous-system-specific miRNA (Xu et al. 2008). dFmrp is required for the proper processing of pre-miR-124a, whereas the loss of dFmr1 leads to a reduced level of mature miR-124a and an increased level of pre-miR-124a. These results suggest a modulatory role for dFmrp to maintain proper levels of miRNAs during neuronal development (Xu et al. 2008). In our own studies, we have shown that dFmr1, the Drosophila ortholog of the FMR1 gene, plays a role in the proper maintenance of germline stem cells in ovary, potentially through the miRNA pathway (Yang et al. 2007). To test this, we recently used an immunoprecipitation assay to reveal that specific microRNAs (miRNAs), particularly the bantam miRNA, are physically associated with dFmrp in ovary (Yang et al. 2007). Bantam has been shown to be regulated by hippo pathway and controls cell proliferation and apoptosis in Drosophila (Brennecke et al. 2003; Thompson and Cohen 2006). We found that, like dFmr1, bantam miRNA is not only required for repressing primordial germ cell differentiation, but also contributes to the maintenance of germline stem cell (Yang et al. 2007). Furthermore, we showed that bantam miRNA genetically interacts with dFmr1 to regulate the fate of germline stem cells (Yang et al. 2007). Collectively, our results support the notion that the FMRP-mediated translational pathway functions through specific miRNAs to control stem cell regulation; however, we saw no effect of dFmrp on the biogenesis of the bantam miRNA. A recent study showed that the loss of dFmr1 did not alter the expression of hid, an mRNA target of bantam miRNA. Given that dFmrp is not associated with hid mRNA, this finding might suggest that dFmrp could be involved in the miRNA-mediated gene regulation of specific mRNAs (Zarnescu et al. 2005; Cziko et al. 2009). Whether FMRP is associated with specific miRNAs in mammalian cells remains to be determined.

We have long known that the formation of long-term memory requires normal synaptic connections and synaptic protein synthesis, including Ca2+/calmodulin-dependent protein kinase II (CAMK II) and the transport of new synthesized proteins (Bailey et al. 2001). It has since been shown that activity-dependent synaptic protein synthesis is regulated by the RISC pathway, and the binding sites for some miRNAs in the 3′ UTR of CAMKII are also involved in this regulation (Ashraf et al. 2006). Using Drosophila olfactory memory as a model, dFmrp was found to be acutely required and interacted with AGO1 in the formation of long-term memory (Bolduc et al. 2008). The absence of dFmrp led to excess protein synthesis, which impairs long-term memory (Bolduc et al. 2008). This observation supports the idea of an interaction between the miRNA pathway and FMRP playing an important role in learning and memory.


Recent discoveries of different small regulatory RNAs, including miRNAs, piRNAs, and endo-siRNAs, have revealed a new layer of gene regulation. These “micro” regulatory RNAs could play “macro” roles in shaping diverse cellular pathways. Studies in recent years have demonstrated the interaction between FMRP and the miRNA pathway at multiple levels; however, a complete molecular understanding of how FMRP uses the miRNA pathway and specific miRNAs to regulate translation is still lacking. Such an understanding is vital not only to reveal the molecular pathogenesis of fragile X syndrome, but also it could help us better understand the pathogenesis of mental retardation in general.


We would like to thank C. Strauss for critical reading of the manuscript. This work was supported in part by NIH grant (R01 MH076090). X.L. is supported by FRAXA Postdoctoral Fellowship. P.J. is a recipient of the Beckman Young Investigator Award and the Basil O’Connor Scholar Research Award, as well as an Alfred P Sloan Research Fellow in Neuroscience.

Copyright information

© Humana Press Inc. 2009