FMRP-Mediated Axonal Delivery of miR-181d Regulates ... - Cell Press

Article

FMRP-Mediated Axonal Delivery of miR-181d Regulates Axon Elongation by Locally Targeting Map1b and Calm1 Graphical Abstract

Authors Bin Wang, Lin Pan, Manyi Wei, ..., Xing-Yu Jiang, Xu Zhang, Lan Bao

Correspondence [email protected]

In Brief Wang et al. find that axon-enriched miR181d regulates axon elongation by targeting axonal Map1b and Calm1 in primary sensory neurons. FMRP mediates delivery of miR-181d and its targets to the axon. NGF triggers Map1b and Calm1 release from FMRP and miR181d-repressing granules in axons, controlling mRNA translation locally to induce axon elongation.

Highlights d

d

d

d

Axon-enriched miR-181d regulates axon elongation by targeting axonal Map1b and Calm1 FMRP mediates axon delivery of miR-181d, Map1b, and Calm1 NGF triggers Map1b and Calm1 release from FMRP and miR181d granules in axons FMRP and miR-181d mediate NGF-induced axon elongation by translational regulation

Wang et al., 2015, Cell Reports 13, 2794–2807 December 29, 2015 ª2015 The Authors http://dx.doi.org/10.1016/j.celrep.2015.11.057

Cell Reports

Article FMRP-Mediated Axonal Delivery of miR-181d Regulates Axon Elongation by Locally Targeting Map1b and Calm1 Bin Wang,1,5 Lin Pan,1,5 Manyi Wei,1 Qiong Wang,1 Wen-Wen Liu,2 Nuoxin Wang,2 Xing-Yu Jiang,2 Xu Zhang,3,4 and Lan Bao1,4,* 1State Key Laboratory of Cell Biology, CAS Center for Excellence in Molecular Cell Science, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200031, China 2National Center of Nanoscience and Technology, Beijing 100190, China 3Institute of Neuroscience and State Key Laboratory of Neuroscience, CAS Center for Excellence in Brain Science, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai 200031, China 4School of Life Science and Technology, ShanghaiTech University, Shanghai 200031, China 5Co-first author *Correspondence: [email protected] http://dx.doi.org/10.1016/j.celrep.2015.11.057 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

SUMMARY

Subcellular targeting and local translation of mRNAs are critical for axon development. However, the precise local control of mRNA translation requires investigation. We report that the Fmr1-encoded protein, FMRP-mediated axonal delivery of miR-181d negatively regulates axon elongation by locally targeting the transcripts of MAP1B (Map1b) and calmodulin (Calm1) in primary sensory neurons. miR-181d affected the local synthesis of MAP1B and calmodulin in axons. FMRP was associated with miR-181d, Map1b, and Calm1. Both FMRP deficiency in Fmr1I304N mice and Fmr1 knockdown impeded the axonal delivery of miR-181d, Map1b, and Calm1 and reduced the protein levels of MAP1B and calmodulin in axons. Furthermore, nerve growth factor (NGF) induced Map1b and Calm1 release from FMRP and miR-181d-repressing granules, thereby promoting axon elongation. Both miR-181d overexpression and FMRP knockdown impaired NGFinduced axon elongation. Our study reveals a mechanism for the local regulation of translation by miR-181d and FMRP during axon development. INTRODUCTION Neurons have specialized cellular structures that include dendrites and axons. Some mRNAs are known to traffic to dendrites and axons for local protein synthesis in response to rapid extracellular signals that are involved in neuronal differentiation, synapse formation, and axon growth during neuronal development (Jung et al., 2012; Lin and Holt, 2008). The molecular mechanisms underlying the regulation of local protein synthesis have been investigated. Recently, a class of non-coding small RNA

molecules, microRNAs (miRNAs), has been revealed to regulate mRNA stability and translation and to play important roles in a variety processes, including the development of the nervous system (Fineberg et al., 2009; McNeill and Van Vactor, 2012). Previous studies have reported the existence of a distinct population of miRNAs in the axons of primary sensory neurons located in the dorsal root ganglion (DRG) and that loss of Dicer, which produces mature miRNAs, resulted in defective peripheral nerve projections (Hancock et al., 2014). However, further functions and mechanisms of axonal miRNAs in the development of axons in primary sensory neurons require investigation. RNA binding proteins (RBPs) have been implicated as cytoskeletal adaptors that act via a form of ribonucleoprotein (RNP) granules that carry mRNAs into the dendrite or axon (Dictenberg et al., 2008; Lin and Holt, 2008). FMRP is an RBP that is encoded by Fmr1, the absence of which causes dysregulated neuronal disorders, including mental retardation and autism spectrum disorders (Bassell and Warren, 2008; Cheever and Ceman, 2009). Several reports provide solid evidence showing that FMRP controls protein synthesis by regulating mRNA stability or repressing protein synthesis by directly binding target genes in an activitydependent manner (Bassell and Warren, 2008; Wang et al., 2008). FMRP is also reported to be associated with mRNAs in the RNA-induced silencing complex (RISC) and to thereby potentially be involved in the miRNA-mediated regulation of activity-dependent protein synthesis (Muddashetty et al., 2011). Despite the role that FMRP plays in the trafficking of postsynaptic transcripts (Dictenberg et al., 2008), whether FMRP mediates the transport of miRNAs or FMRP mediates the regulation of local protein synthesis by miRNAs in response to extracellular signals in axons remains unexplored. In our study, by using a microfluidic culture of embryonic DRG neurons, we identified the relative enrichment of miR-181d in axons. miR-181d affected axon elongation by locally targeting the mRNAs of microtubule-associated protein 1B (MAP1B) and calmodulin, a crucial regulator of calcium signaling. FMRP was associated with miR-181d and its targets MAP1B mRNA

2794 Cell Reports 13, 2794–2807, December 29, 2015 ª2015 The Authors

(Map1b) and calmodulin mRNA (Calm1). The downregulation of FMRP in Fmr1I304N mice or in the cell bodies of cultured DRG neurons impaired the axonal delivery of miR-181d with Map1b and Calm1, leading to decreased protein levels of MAP1B and calmodulin in axons. Furthermore, nerve growth factor (NGF) increased the local synthesis of MAP1B and calmodulin in axons by releasing Map1b and Calm1 from FMRP and miR-181d-repressing granules. Thus, we reveal a mechanism for axon elongation by the FMRP-mediated axon delivery of miR-181d and its associated mRNAs and local regulation of protein synthesis. RESULTS miR-181d Is Enriched in Axons of Embryonic DRG Neurons To assess the distribution of miRNAs in axons, we used a microfluidic device containing separate cell body and axon compartments to culture DRG neurons (Figure S1A) (Chen et al., 2012). The DAPI signal was observed only in the cell body compartment, whereas the Tuj-1-positive signal was observed in both the cell body and the axon compartments (Figure 1A). RT-PCR also showed that g-actin was restricted to cell bodies, but b-actin and Impa1, the transcript of a key enzyme that is crucial for axonal maintenance, were observed in both the cell bodies and the axons of cultured embryonic day 13.5 (E13.5) DRG neurons (Figure S1B), consistent with previous reports (Andreassi et al., 2010). Furthermore, immunoblotting from the microfluidic cultures of DRG neurons detected MAP2 only in cell bodies, while Tau-1 was distributed in both cell bodies and axons (Figure S1C). These results thus confirm that the RNA and protein that were prepared from the axons of DRG neurons were devoid of nuclear components from the cell bodies, further indicating the successful compartmental culture of DRG neurons. Previous studies have demonstrated the importance of the Dicer-dependent miRNA pathway in sensory axon development and peripheral nerve regeneration (Hancock et al., 2014). We measured the amount of miRNA that was downregulated in the DRGs of the Dicer knockout mice (Zhao et al., 2010) in the cell bodies and axons of DRG neurons at E13.5. Quantitative realtime PCR (qPCR) showed that miR-181d was abundant in the axons of DRG neurons (Figures 1A and S1D). In situ hybridization using a lock nucleic acid (LNA)-modified specific probe showed that the miR-181d signal was present in both cell bodies and axons in punctate structures compared to the low signal of the scramble probe (Figure 1B). These data suggest that miR-181d is enriched and localized in axons of embryonic DRG neurons. Map1b and Calm1 Are Targeted by Axon-Enriched miR-181d in DRG Neurons To identify the possible downstream targets of miR-181d, bioinformatic tools including TargetScan and miRanda were used to analyze the axonal pool of thousands of mRNAs from the large-scale sequencing of developing and adult sensory neurons (Gumy et al., 2011). Nine candidate genes that were predicted by these software packages were selected (Figure S1E). To determine which genes might be regulated by miR-181d, we performed luciferase reporter assays in HEK293 cells after suc-

cessful transfection of a miR-181d inhibitor and its mimics in cultured E13.5 DRG neurons (Figure S1F). The overexpression of miR-181d with its mimics significantly decreased the 30 untranslated region (UTR) activity of Map1b and Calm1 but not of their mutants, which lost the ability to be bound by miR-181d (Figures 1C, S1G, and S1H). However, the overexpression of miR-497, which was not detected in the axons of sensory neurons (Hancock et al., 2014), did not affect the 30 UTR activity of Map1b and Calm1 (Figure S1I), demonstrating that the silencing pathway was not simply overloaded with exogenously expressed miRNAs. These data suggest that Map1b and Calm1 are potential targets of miR-181d. Although the transcript of MAP1B has been shown to localize in neurons, including dendrites and axons (Antar et al., 2006; Dictenberg et al., 2008), the distribution pattern of the MAP1B and calmodulin transcripts remains to be described in embryonic DRG neurons. RT-PCR showed the presence of Map1b and Calm1 in both the cell body and the axon compartments of microfluidic cultured E13.5 DRG neurons (Figure 1D). In situ hybridization further detected the granular appearance of Map1b and Calm1 expression in the cell bodies and axons of cultured E13.5 DRG neurons (Figure 1E). We also successfully monitored the 30 UTR transport of mRNAs using the MS2 system (Fusco et al., 2003), in which the b-actin-30 UTR, but not the g-actin-30 UTR, was detected in axons (Figure S1K). Plasmid transfection was performed on cultured post-natal day 14 (P14) DRG neurons, which are more resistant to electroporation than are E13.5 DRG neurons. The granules containing Map1b-30 UTR and Calm1-30 UTR were clearly observed and moved bidirectionally in the axons of cultured P14 DRG neurons (Figures 1F and 1G). These data verify the axonal localization and trafficking of MAP1B and calmodulin transcripts in DRG neurons. Furthermore, we assessed the protein levels of MAP1B and calmodulin in cultured E13.5 DRG neurons. Immunoblotting showed that the upregulation or downregulation of miR-181d bidirectionally regulated the protein level of MAP1B and calmodulin in cultured DRG neurons (Figures 1H and 1I), whereas qPCR showed no significant change in the levels of those mRNAs (Figure S1J), suggesting that miR-181d mediates the repression of MAP1B and calmodulin post-transcriptionally. However, the application of a miR-181d inhibitor and its mimics did not affect the protein level of b-actin, and the transfection of miR-497 mimics did not decrease the protein level of MAP1B and calmodulin in cultured E13.5 DRG neurons (Figures S2A–S2D). Taken together, these data suggest that axon-enriched miR-181d specifically targets Map1b and Calm1 in DRG neurons. miR-181d Regulates Axon Elongation by Affecting the Local Synthesis of MAP1B and Calmodulin The transport and local translation of mRNA are essential for axon development (Jung et al., 2012). MAP1B stabilizes the cytoskeleton during the elongation of central and peripheral axons in DRG neurons (Dajas-Bailador et al., 2012; MontenegroVenegas et al., 2010). Calmodulin mediates calcium signaling, which is involved in growth cone guidance and dendrite growth in the central nervous system (CNS) (McCue et al., 2010). Therefore, the miR-181d-mediated regulatory machinery might play a role in axon development in primary sensory neurons. We

Cell Reports 13, 2794–2807, December 29, 2015 ª2015 The Authors 2795

Figure 1. Identification of Axon-Enriched miR-181d and Its Targets, Map1b and Calm1, in Embryonic DRG Neurons (A) miR-181d was enriched in the axons of embryonic DRG neurons. Immunostaining for Tuj-1 demonstrated the successful establishment of microfluidic cultured E13.5 DRG neurons at 7 DIV (upper panel). A qPCR analysis of miR-181d showed a 5.4-fold increase in its axonal abundance compared to the cell bodies. The results are presented as mean ± SEM. *p < 0.05 versus the level of miR-181d in the cell body (n = 6). Scale bar, 300 mm. (B) Localization of miR-181d (red) in the axons and cell bodies of E13.5 DRG neurons at 4 DIV used LNA-modified probes. A scramble probe was used as the negative control. Immunostaining for Tuj-1 (green) indicated the neuronal profile. Scale bar, 5 mm. (C) Luciferase reporter assays showed that Map1b and Calm1 are potential downstream targets of miR-181d. The luciferase activity of the WT 30 UTR, but not the mutated Map1b and Calm1, was downregulated by miR-181d mimics in HEK293 cells. The results are presented as mean ± SEM. *p < 0.05 and **p < 0.01 versus the negative control (n = 3). (D) RT-PCR showed that Map1b and Calm1 were distributed in the cell bodies and axons of cells in the microfluidic cultures of E13.5 DRG neurons at 7 DIV. b-actin and g-actin were the positive and negative controls, respectively, for axonal distribution. (E) In situ hybridization showed that Map1b and Calm1 (red) were localized in the cell bodies and axons of cultured E13.5 DRG neurons at 4 DIV. Immunostaining for tubulin (green) was used to indicate the neuronal profile. Scale bar, 5 mm. (F and G) Live-cell imaging demonstrated the anterograde movement of Map1b-30 UTR (F) and Calm1-30 UTR (G) in the axons of cultured P14 DRG neurons at 4 DIV (arrowheads). Scale bar, 5 mm. (H and I) Immunoblotting showed that applying a miR-181d inhibitor (H) and its mimics (I) increased and decreased, respectively, the protein levels of MAP1B and calmodulin in cultured E13.5 DRG neurons at 4 DIV. Tubulin was used as a loading control. The results are presented as mean ± SEM. *p < 0.05 and **p < 0.01 versus the negative control (n = 3). See also Figures S1 and S2.


Figure 2. miR-181d Negatively Regulates Axon Elongation by Locally Targeting Axonal Map1b and Calm1 (A and B) Representative images of microfluidic cultured E13.5 DRG axons at 4 DIV (A) and quantitative results of axon growth rates (B) showed that the axonal application of a miR-181d inhibitor and its mimics increased and decreased axon elongation, respectively, in the microfluidic chamber. The results are presented as mean ± SEM. *p < 0.05 versus the negative control (n = 3). Scale bar, 300 mm.

(legend continued on next page)


successfully performed RNA transfection selectively in axons (Figures S2E and S2F) and began by detecting the effect of miR-181d on axon elongation. In microfluidic cultures of E13.5 DRG neurons in the presence of 100 ng/ml NGF, which kept normal neuron growth, knockdown of miR-181d by a specific inhibitor in the axon compartment significantly increased the axon growth rate, whereas overexpression of miR-181d by its mimics strongly decreased the axon growth rate (Figures 2A and 2B). In a control experiment, we showed that overexpression of miR497 by its mimics did not affect the axon growth rate (Figures S2G and S2H). Thus, miR-181d has a regulatory effect on axon elongation. We then evaluated the local synthesis of MAP1B and calmodulin in the presence of small interfering RNA (siRNA)-mediated knockdown of MAP1B and calmodulin in axons. In cultured E13.5 DRG neurons, transfection with siRNA dramatically reduced the endogenous expression levels of both MAP1B and calmodulin (Figures 2C and 2D). The application of siRNA exclusively in the axon compartment for 48 hr significantly decreased the axon growth rate by knocking down axonal MAP1B or calmodulin (Figures 2E and 2F), suggesting that the axonal synthesis of MAP1B and calmodulin contributes to axon elongation. Finally, to determine whether the effect of miR-181d on axon elongation involved the local translation of Map1b and Calm1, we used the synthesized and LNA-modified target protector (TP) to disrupt the interaction between miR-181d and Map1b or Calm1 (Figure S3A) (Staton and Giraldez, 2011). Transfection with 50 nM Map1b-TP and Calm1-TP markedly increased the protein levels of MAP1B and calmodulin (Figures S3B and S3C), whereas the levels of Map1b and Calm1, as well as the level of miR-181d, remained unchanged (Figure S3D). Moreover, the expression of Map1b-TP in the axon compartment significantly increased the amount of MAP1B but not the amount of calmodulin, whereas the expression of Calm1-TP increased the axonal synthesis of calmodulin but not that of MAP1B (Figure 2G). Using either Map1b-TP or Calm1-TP resulted in an increase in axon elongation (Figures 2H and 2I). Taken together, these data suggest that miR-181d regulates axon elongation by affecting the local synthesis of MAP1B and calmodulin. FMRP Is Associated with miR-181d, Map1b, and Calm1 RBPs mediate mRNA localization and translation in the invertebrate and mammalian nervous systems (Jung et al., 2012). Because the MAP1B mRNA has been shown to associate with

the FMRP-containing RNA granules in the dendritic spines and axons of neurons (Antar et al., 2006; Zalfa et al., 2003), we assessed the interaction of miR-181d and its targets with FMRP in P14 rat tissue of DRGs by performing co-immunoprecipitation using an FMRP antibody (Figure 3A). The qPCR showed that miR-181d, but not miR-9 or miR-132, was enriched in the FMRP-precipitated immunoprecipitates of DRG extracts (Figure 3B). In addition, RT-PCR showed that Map1b and Calm1 were co-immunoprecipitated by FMRP, as was Fmr1 (Figure 3C). To confirm a specific association between miR-181d and FMRP, we used a control—another RBP, cytoplasmic polyadenylation element-binding protein 1 (CPEB1), that binds to a specific cytoplasmic polyadenylation element sequence in the 30 UTR of its targets. Pheochromocytoma 12 (PC12) cells contain endogenous miR-181d, MAP1B, and calmodulin, and they are a good model system for detecting the association between FMRP and miR-181d, Map1b, and Calm1. We performed immunoprecipitation using a FLAG-tagged antibody in PC12 cells expressing FMRP-GFP-FLAG or CPEB1-GFP-FLAG (Figure 3D). The qPCR showed abundant miR-181d in the immunoprecipitates pulled down by FMRP-GFP-FLAG but not in those pulled down by CPEB1-GFP-FLAG (Figure 3E). However, miR-9 did not interact with FMRP-GFP-FLAG or CPEB1-GFP-FLAG (Figure 3E). We found that FMRP-GFP-FLAG, but not CPEB1GFP-FLAG, immunoprecipitated endogenous Map1b and Calm1 (Figure 3F). In addition, the interaction of FMRP with Map1b and Calm1 was not dependent on miRNA-181d, because knockdown of miRNA-181d by its inhibitor did not affect their association (Figures S4A and S4B). Therefore, these data provide biochemical evidence showing that FMRP associates with miR-181d, Map1b, and Calm1. We further explored the axonal co-localization of FMRP with miR-181d, Map1b, and Calm1. Immunostaining for FMRP combined with in situ hybridization for Map1b or Calm1 showed that these endogenous mRNAs were localized in the FMRP-positive granules in the axons of cultured P14 DRG neurons and approximately 53% of Map1b-positive granules and 36% of Calm1positive granules contained FMRP (Figures 3G and 3H). In situ hybridization for miR-181d, combined with immunostaining for FMRP, showed that approximately 42% miRNA-181d-positive granules contained FMRP (Figures 3G and 3H), and miR-181d was localized in approximately 27% FMRP granules containing Map1b-30 UTR and approximately 25% FMRP granules expressing Calm1-30 UTR in the axons of cultured P14 DRG neurons

(C and D) Representative immunoblots and quantitative data showed that si-Map1b and si-Calm1 decreased the protein levels of MAP1B (C) and calmodulin (D) in cultured E13.5 DRG neurons at 4 DIV. Tubulin was used as a loading control. The results are presented as mean ± SEM and normalized to the levels in the siRNA control. **p < 0.01 versus the siRNA control (n = 3). (E and F) Representative images of microfluidic cultured E13.5 DRG axons at 4 DIV (E) and quantitative results of axon growth rates (F) showed that the axonal application of si-Map1b or si-Calm1 reduced axon elongation. The results are presented as mean ± SEM. *p < 0.05 versus the siRNA control (n = 3). Scale bar, 300 mm. (G) Representative immunoblots and quantitative data showed that the axonal application of Map1b-TP or Calm1-TP blocked the interaction of miR-181d with the 30 UTR of Map1b or Calm1, respectively, and increased the protein levels of MAP1B and calmodulin in the axon, but not in the cell body, in microfluidic cultures of E13.5 DRG neurons at 4 DIV. Tubulin was used as a loading control. The results are presented as mean ± SEM and normalized to the mock control. *p < 0.05 and **p < 0.01 versus the negative control (n = 3). (H and I) Representative images of microfluidic cultured E13.5 DRG axons at 4 DIV (H) and quantitative results of axon growth rates (I) demonstrated that the axonal application of either Map1b-TP or Calm1-TP increased axon elongation in the microfluidic chamber. The results are presented as mean ± SEM. *p < 0.05 versus the negative control (n = 3). Scale bar, 300 mm. See also Figures S2 and S3.


Figure 3. FMRP Is Associated with miR181d, Map1b, and Calm1 (A) Immunoblotting showed the products of immunoprecipitation by a FMRP antibody in P14 DRGs. Immunoglobulin G (IgG) served as a control. (B) A qPCR analysis showed that miR-181d, but not miR-9, miR-132, or U6, was abundant in the FMRP immunoprecipitates of P14 DRGs. The results are presented as mean ± SEM. *p < 0.05 versus the IgG control (n = 3). (C) Map1b and Calm1 were detected in the FMRP immunoprecipitates of P14 DRGs using RT-PCR. Fmr1, which is known to be associated with FMRP, was used as a positive control. Gapdh was used as a negative control. (D) Immunoblots showing the immunoprecipitates that were pulled down by a FLAG antibody in PC12 cells expressing FMRP-GFP-FLAG or CPEB1GFP-FLAG. Tubulin or IgG served as the loading control. (E) A qPCR analysis showing that miR-181d was abundant in the FMRP but not the CPEB1 immunoprecipitates of PC12 cells. The results are presented as mean ± SEM. *p < 0.05 versus PC12 cells expressing GFP (n = 3). (F) Map1b and Calm1 were significantly detected in the FLAG immunoprecipitates from PC12 cells expressing FMRP-GFP-FLAG, but not those expressing CPEB1-GFP-FLAG, using RT-PCR. Fmr1 was used as the positive control for an association with FMRP, and Gapdh was used as the negative control. (G and H) Representative images (G) and quantitative results (H) showed co-localization between miR-181d, Map1b, or Calm1 (red) and FMRP (green) in the axons of cultured P14 DRG neurons at 4 DIV (arrowheads). The results are presented as mean ± SEM (n = 3). Scale bar, 10 mm. (I and J) Representative images (I) and quantitative results (J) showed miR-181d (green) co-localized with FMRP (blue) and Map1b-30 UTR or Calm1-30 UTR (red) in the axons of cultured P14 DRG neurons at 4 DIV (arrowheads). The results are presented as mean ± SEM (n = 3). Scale bar, 10 mm. See also Figure S4.

(Figures 3I and 3J). Taken together, our data suggest that FMRP associates with miR-181d, Map1b, and Calm1 in DRG neurons. FMRP Deficiency Causes Defects in the Axonal Delivery of miR-181d, Map1b, and Calm1 We assessed the function of FMRP in the axonal distribution of miR-181d, Map1b, and Calm1 in vivo. The qPCR showed that

the levels of miR-181d, Map1b, and Calm1 were lower in the sciatic nerves, but not in the DRGs, of Fmr1I304N mice at P14 than in those of wild-type (WT) mice (Figures 4A and 4B). We also used an adeno-associated virus (AAV) 2/8mediated technique to knockdown FMRP in the cell bodies of microfluidic cultured E13.5 DRG neurons (Figures 4C and 4D). Seven days after AAV-shFMRP-GFP-FLAG infection of cell body compartments, immunoblotting showed dramatically decreased protein levels of FMRP in both cell bodies and axons (Figures 4C and 4D), suggesting a high efficiency of knockdown by the AAV. The mRNA of FMRP (Fmr1) was rarely detected in the axons of cultured E13.5 DRG neurons, which is consistent with a previous


report (Gumy et al., 2011). The qPCR showed that the expression levels of miR-181d, Map1b, and Calm1 in the axons of DRG neurons were decreased, whereas their expression in the cell bodies remained unchanged (Figure 4E). Furthermore, we transfected MS2-mCherry with Map1b-30 UTR or Calm1-30 UTR into cultured P14 DRG neurons expressing either FMRP short hairpin RNA (shRNA) or a control shRNA. Knockdown of FMRP reduced the numbers of Map1b-30 UTR and Calm1-30 UTR granules in axons showing either anterograde or retrograde movement, in addition to increasing the number of granules showing non-significant movement—whether stationary, Brownian, or otherwise not showing displacement over long distances (Figures 4F–4H). However, the average anterograde and retrograde velocities of Map1b-30 UTR and Calm1-30 UTR granules were not changed (Figure 4I). Co-expression of shRNA-resistant FMRP (FMRPR) rescued the axonal transport of Map1b-30 UTR and Calm1-30 UTR granules by silencing FMRP (Figures 4F–4H), indicating that the defect caused by FMRP shRNA was not due to an offtarget effect. In addition, phase contrast images showed that the DRG neurons affected by knockdown of FMRP by either AAV infection or electroporation displayed morphologies similar to those of neurons that were transfected with control shRNA (Figures S5A and S5B), which excluded the effect of significant stress in neurons affected by FMRP shRNA. As a control, b-actin-30 UTR was not localized in the FMRP-GFP-positive granules, and the axonal transport of b-actin-30 UTR was unaffected by knockdown of FMRP (Figures S5C–S5F), suggesting the specificity of FMRP-mediated mRNA transport. Thus, FMRP is required for the axonal delivery of miR-181d, Map1b, and Calm1. FMRP Deficiency Decreases the Protein Level of MAP1B and Calmodulin We further assessed the protein levels of MAP1B and calmodulin in the FMRP-deficient axons of DRG neurons. Consistent with a previous report (Zang et al., 2009), we found that the protein level of FMRP was dramatically reduced in the DRGs and sciatic nerves of the Fmr1I304N mice (Figures 5A and 5B). We then showed that Fmr1I304N mice had decreased levels of MAP1B and calmodulin in sciatic nerves, whereas the levels of these proteins in DRGs were unchanged (Figures 5A and 5B). In microfluidic cultures of E13.5 DRG neurons infected with AAV-shFMRP-GFP-FLAG, the amounts of MAP1B and calmodulin in the axon compartment were significantly reduced (Figures 5C and 5D). Co-expression of AAV-FMRPR-mCherryFLAG rescued the decrease in the protein level of FMRP in the cell bodies and axons of cultured E13.5 DRG neurons induced by silencing FMRP (Figures 5C and 5D). The protein levels of MAP1B and calmodulin, which were reduced by silencing FMRP in axons, were markedly recovered by expression of FMRPR (Figures 5C and 5D). However, the protein levels of MAP1B and calmodulin in the cell bodies of cultured E13.5 DRG neurons were not affected by the knockdown or rescue of FMRP (Figure 5D). Furthermore, the stability and protein transport of MAP1B and calmodulin were not affected by silencing FMRP because the protein levels of MAP1B and calmodulin in cell bodies remained stable (Figure 5), and the fluorescence recovery after photobleaching (FRAP) of

MAP1B-GFP and calmodulin-GFP in axons was unchanged (Figures S5G–S5I). Taken together, these results suggest that the defects in the axonal delivery of mRNAs caused by FMRP deficiency influence the protein levels of MAP1B and calmodulin in the axons of DRG neurons. NGF Triggers the Release of Map1b and Calm1 from FMRP Granules Previous studies have shown that FMRP is a repressor of the transcripts that it binds and that their association can be affected by extrinsic signaling (Muddashetty et al., 2011). Given evidence supporting the FMRP-mediated axonal delivery of miR-181d and its targets, we next explored whether signals regulate their association with FMRP granules. NGF plays a critical role in axon growth during the development of primary sensory neurons because more than 80% of neurons express its receptor, tropomyosin receptor kinase A (TrkA), at E13.5 (White et al., 1996). To test the role of NGF, 5 ng/ml NGF was replaced to only maintain basal survival in microfluidic cultured DRG neurons for 12 hr. Treatment of the axon compartment with 100 ng/ml NGF for 2 hr significantly increased the protein levels of MAP1B and calmodulin in axons without affecting their levels in the cell bodies of microfluidic cultured E13.5 DRG neurons (Figures 6A and 6B). This increase was blocked by co-treatment of the axon compartment with cycloheximide (CHX), an inhibitor of protein synthesis (Figures 6A and 6B), which suggests the locally induced translation of Map1b and Calm1 by NGF. Neurotrophin-3 (NT-3) was unable to induce the axonal translation of Map1b and Calm1 (Figures S6A and S6B), suggesting that the effect of local NGF signaling on the protein synthesis of MAP1B and calmodulin is specific. These data suggest that NGF signaling promotes the local synthesis of MAP1B and calmodulin in axons. We then explored whether local NGF signaling regulates the levels of miR-181d, Map1b, and Calm1 in axons. Surprisingly, we found that miR-181d, Map1b, and Calm1 levels remained unchanged in both cell bodies and axons after NGF was applied to the axon compartment (Figures 6C and 6D). We hypothesize that NGF signaling may affect the association of FMRP with miR-181d and its targets. Therefore, we performed co-immunoprecipitation experiments to detect the interaction between FMRP and miR-181d, Map1b, and Calm1 in response to NGF. PC12 cells endogenously express high levels of the receptor TrkA, and they express miR-181d, Map1b, and Calm1. In PC12 cells exogenously expressing FMRP-GFP-FLAG, the amount of Map1b and Calm1 co-immunoprecipitated by a FLAG antibody was significantly decreased after treatment with 100 ng/ml NGF for 2 hr, whereas the amount of co-immunoprecipitated Fmr1 was not changed (Figures 6E and 6F). However, the level of miR-181d that was associated with FMRP was not altered in the presence of 100 ng/ml NGF (Figure 6G). Furthermore, the association of the P body component argonaute 2 (Ago2) with FMRP-GFP-FLAG was not affected by NGF treatment in PC12 cells (Figures 6H and 6I). Taken together, these results show that NGF triggers the release of Map1b and Calm1 from FMRP granules, leading to the escape of Map1b and Calm1 from translational inhibition by FMRP and miR-181d.


Figure 4. FMRP Deficiency Impairs Axonal Targeting of miR-181d, Map1b, and Calm1 (A and B) A qPCR analysis showed that miR-181d (A), as well as Map1b and Calm1 (B), was decreased in the sciatic nerves, but not in the DRGs, of P14 Fmr1I304N mice. The results are presented as mean ± SEM. *p < 0.05 and **p < 0.01 versus the WT (n = 5). (C) Schematic of the AAV-shFMRP-GFP-FLAG construct used for FMRP knockdown. (D) A representative image showed highly efficient infection of AAV-shFMRP-GFP-FLAG in a microfluidic culture of E13.5 DRG neurons at 7 DIV (upper panel). Immunoblotting revealed significant reductions in FMRP in the cell bodies and axons (lower panel). Scale bar, 300 mm. (E) A qPCR analysis showed that miR-181d, Map1b, and Calm1 were decreased in the axons, but not in the cell bodies, of microfluidic cultured E13.5 DRG neurons expressing AAV-shFMRP-GFP-FLAG at 7 DIV. The results are presented as mean ± SEM. *p < 0.05 and **p < 0.01 versus the control (Ctrl) shRNA (n = 4).

(legend continued on next page)


Figure 5. FMRP Deficiency Reduces the Levels of MAP1B and Calmodulin in the Axons of DRG Neurons (A and B) Representative immunoblots (A) and quantitative data (B) showed that levels of MAP1B and calmodulin were decreased in the sciatic nerves, but not in the DRGs, of P14 Fmr1I304N mice. FMRP was dramatically decreased in both DRGs and sciatic nerves in Fmr1I304N mice. The results are presented as mean ± SEM. *p < 0.05 and **p < 0.01 versus the WT (n = 5). (C and D) Representative immunoblots (C) and quantitative data (D) showed that infection with AAV-shFMRP-GFP-FLAG decreased the protein levels of MAP1B and calmodulin in the axons, but not in the cell bodies, of microfluidic cultured E13.5 DRG neurons at 7 DIV. These phenotypes were partially rescued by co-expression of AAV-FMRPmCherry-FLAG (FMRPR). The results are presented as mean ± SEM. *p < 0.05 and **p < 0.01 versus the control (Ctrl) shRNA, and #p < 0.05 and ##p < 0.01 versus the indicated treatments (n = 3). See also Figure S5.

miR-181d and FMRP Are Involved in NGF-Induced Axon Elongation by Regulating the Local Translation of Map1b and Calm1 Because NGF signaling regulated the association of Map1b and Calm1 with miR-181d and FMRP, we next investigated the functional role of miR-181d in NGF-induced axon elongation. After axonal application of 100 ng/ml NGF for 2 hr, we observed that the axon growth rate was faster than that observed under the basal condition of 5 ng/ml NGF in microfluidic cultures of E13.5 DRG neurons (Figures 7A and 7B). Transfection of miR-181d mimics into the axon compartment of microfluidic cultures of DRG neurons significantly attenuated the NGF-induced axon elongation (Figures 7A and 7B). Meanwhile, immunoblotting showed that overexpressing miR-181d mimics in axons blocked the NGF-induced local synthesis of MAP1B and calmodulin in axons (Figures 7C and 7D). Although axons lack Fmr1, we also excluded the possibility of a role for miR-181d in the regulation of FMRP expression. Immunoblotting showed that overexpression of miR-181d by its mimics did not affect the level of FMRP in cultured E13.5 DRG neurons (Figures S7A and S7B). We then explored the functional role of FMRP in NGF-induced axon elongation. In microfluidic cultures of E13.5 DRG neurons infected with AAV-shFMRP-GFP-FLAG, knockdown of FMRP did not change the axon growth rate in the presence of a low

concentration of 5 ng/ml NGF (Figures 7E and 7F), supporting that basal axon elongation in DRG neurons is not mediated by FMRP. Knockdown of FMRP significantly reduced the increase in axon growth rate that was induced by 100 ng/ml NGF for 2 hr (Figures 7E and 7F). Meanwhile, the NGF-induced local translation of Map1b and Calm1 in axons was abolished in microfluidic cultures of E13.5 DRG neurons infected with AAV-shFMRP-GFP-FLAG (Figures 7G and 7H). Taken together, these data suggest that miR-181d and FMRP mediate NGFinduced axon elongation by regulating the axonal translation of Map1b and Calm1. DISCUSSION The role of miRNAs is critical in the translational control of local mRNAs during axon development (Hancock et al., 2014; Kaplan et al., 2013). Here, we demonstrated that axon-enriched miR-181d contributes to axon elongation by regulating the local translation of Map1b and Calm1. miR-181d, together with Map1b and Calm1, is associated with FMRP in primary sensory neurons. FMRP deficiency impaired the axonal delivery of miR-181d with Map1b and Calm1 and thereby decreased the protein levels of MAP1B and calmodulin in axons. Moreover, NGF signaling triggers the release of Map1b and Calm1 from FMRP granules, leading to the escape

(F–I) Representative live-cell images (F and G) and quantitative data (H) showed that knockdown of FMRP by FMRP shRNA (green) reduced the numbers of Map1b-30 UTR and Calm1-30 UTR granules (red) moving in both anterograde and retrograde directions, as well as increasing the number of granules displaying non-significant movement in the axons of cultured P14 DRG neurons at 4 DIV. These phenotypes were rescued by co-expression of FMRPR-BFP-FLAG (blue). Mean velocity of anterograde and retrograde movements by Map1b-30 UTR and Calm1-30 UTR granules was not significantly changed (I). A 300 s video was collected, and a kymograph was created to demonstrate the movement of Map1b-30 UTR and Calm1-30 UTR granules. The results are presented as mean ± SEM. *p < 0.05 and **p < 0.01 versus the Ctrl shRNA, and #p < 0.05 and ##p < 0.01 versus the indicated treatments (n = 3). Scale bar, 10 mm. See also Figure S5.


Figure 6. NGF Signaling Regulates the Association of Map1b and Calm1 with FMRP (A and B) Representative immunoblots (A) and quantitative data (B) showed that axonal MAP1B and calmodulin were increased after 2 hr of NGF treatment in the axon compartment of microfluidic cultured E13.5 DRG neurons at 7 DIV. These phenotypes were abolished in the presence of CHX. Tubulin was used as a loading control. The results are presented as mean ± SEM. *p < 0.05 versus vehicle, and #p < 0.05 and ##p < 0.01 versus the indicated treatments (n = 3). (C and D) A qPCR analysis showed that miR-181d (C), as well as Map1b and Calm1 (D), levels were not changed in either cell bodies or axons after 2 hr of NGF treatment in the axon compartment of microfluidic cultured E13.5 DRG neurons at 7 DIV. The results are presented as mean ± SEM (n = 3). (E and F) RT-PCR (E) and qPCR analysis (F) showed that the association of Map1b and Calm1 with FMRP was decreased in the FLAG immunoprecipitates after NGF treatment in PC12 cells expressing FMRP-GFP-FLAG. The results are presented as mean ± SEM. *p < 0.05 versus vehicle (n = 4). (G) A qPCR analysis showed that the association of miR-181d with FMRP was not affected, as analyzed in FLAG immunoprecipitates, by NGF treatment in PC12 cells expressing FMRP-GFP-FLAG. The results are presented as mean ± SEM (n = 4). (H and I) Representative immunoblots (H) and quantitative data (I) showed that the association of FMRP with Ago2 was not affected by NGF treatment in PC12 cells expressing FMRP-GFP-FLAG. The results are presented as mean ± SEM (n = 3). See also Figure S6.

of Map1b and Calm1 from translational inhibition by miR-181d and FMRP. Thus, we have revealed a mechanism for FMRPmediated axonal delivery of miR-181d and its targets and local regulation of Map1b and Calm1 translation in response to NGF signaling. Axon-Enriched miR-181d Negatively Regulates Axon Elongation in Primary Sensory Neurons The spatiotemporal distribution of mRNAs and miRNAs is important for the functions of nervous systems (Fineberg et al., 2009; Jung et al., 2012; Kaplan et al., 2013). In our study, by using microfluidic cultures of E13.5 DRG neurons and qPCR, we dissected the relatively abundant miR-181d in axons from downregulated miRNAs in the primary sensory neurons of Dicer knockout mice. Recent studies have demonstrated the presence of miR-181a in the axons of cortical neurons (Sasaki et al., 2014) and miR-181d in the neurites of depolarized

human neuroblast cells (Goldie et al., 2014). High-throughput sequencing of RNAs immunoprecipitated by an Ago2 antibody showed that members of the miR-181 family were enriched in the brain (Chi et al., 2009). Despite reports that miR-181a in neurons is involved in synaptic plasticity (Saba et al., 2012) and cerebral ischemia (Ouyang et al., 2012), the functions of the miR-181 family in axon development, and the mechanisms underlying those functions, are far from clear. In microfluidic cultures of E13.5 DRG neurons, the application of a miR-181d inhibitor to the axon compartment increased axon elongation, whereas treatment with miR-181d mimics had the opposite effect. Using bioinformatic predictions derived from published axonal transcriptomes (Gumy et al., 2011), we screened out Map1b and Calm1 as downstream targets of miR-181d, suggesting the possibility that miR-181d negatively regulates axonal elongation by influencing the stabilization of microtubules and the activation of calcium signaling. In the CNS, Map1b has been reported to localize to the dendrites and axons of cortical neurons, and the local synthesis of MAP1B has been found to contribute to axon guidance, elongation, and neuronal migration via its effects on the stabilization


Figure 7. miR-181d and FMRP Are Involved in NGF-Induced Axon Elongation by Regulating Map1b and Calm1 Translation (A and B) Representative phase contrast images (A) and quantitative results of axon growth rates (B) demonstrated that axon elongation was promoted by 2 hr of treatment with NGF in the axon compartment of microfluidic cultured E13.5 DRG neurons at 4 DIV. This phenotype was abolished by the pre-application of miR-181d mimics in axons. The results are presented as mean ± SEM. *p < 0.05 versus vehicle, and #p < 0.05 versus the indicated treatment (n = 4). Scale bar, 10 mm. (C and D) Representative immunoblots (C) and quantitative data (D) showed that the NGF-induced increase of MAP1B and calmodulin in axons was abolished by the pre-application of miR-181d mimics in the axon compartment of microfluidic cultured E13.5 DRG neurons at 7 DIV. The results are presented as mean ± SEM. *p < 0.05 versus vehicle, and #p < 0.05 and ##p < 0.01 versus the indicated treatment (n = 3). (E and F) Representative images (E) and quantitative results of axon growth rates (F) demonstrated that the knockdown of FMRP by AAV-shFMRPGFP-FLAG decreased NGF-induced axon elongation in microfluidic cultured E13.5 DRG neurons at 7 DIV. The results are presented as mean ± SEM. **p < 0.01 versus the control (Ctrl) shRNA, and #p < 0.05 versus the indicated treatment (n = 3). Scale bar, 10 mm. (G and H) Representative immunoblots (G) and quantitative data (H) showed that FMRP knockdown by AAV-shFMRP-GFP-FLAG abolished the NGF-induced increases in MAP1B and calmodulin in the axons of cultured E13.5 DRG neurons at 7 DIV. The results are presented as mean ± SEM. *p < 0.05 and **p < 0.01 versus vehicle (n = 3). See also Figure S7.

of microtubules (Montenegro-Venegas et al., 2010; VillarroelCampos and Gonzalez-Billault, 2014). The local role of calmodulin-mediated calcium signaling has been investigated in axon development (Nakamuta et al., 2011); however, our data demonstrating the axonal distribution of Calm1 in fluorescence in situ hybridization and the transport of Calm1-30 UTR in livecell imaging are the evidence to support the presence of Calm1 in axons. The selective knockdown of miR-181d or the blockade of the interacting domain of Map1b and Calm1 with miR-181d in the axon compartment increased local levels of MAP1B and calmodulin and promoted axon elongation, suggesting a physiological role for miR-181d in the axons of primary sensory neurons. A recent study has also reported that miR-132 is abundant in the axons of sensory neurons and that it promotes axon extension by targeting the mRNA of the Ras guanosine triphosphatase activator Rasa1 (Hancock et al., 2014). Considering that miR-181d negatively regulates axon elongation by affecting Map1b and Calm1 translation, our data demonstrating that different miRNAs cooperatively affect axon elongation by targeting different downstream molecules are interesting.

FMRP Mediates the Axonal Delivery and the Responses to NGF Signaling of the miR-181d Machinery The presynaptic localization of FMRP has been demonstrated in the CNS (Antar et al., 2006; Klemmer et al., 2011). Our study reveals the importance of FMRP in the axonal delivery of miR181d and its targets. FMRP deficiency led to decreases in the level of miR-181d, as well as those of Map1b and Calm1, in the axons of primary sensory neurons. Consistent with these results, FMRP deficiency reduced the protein levels of MAP1B and calmodulin in axons. Previous studies showed that FMRP not only binds to the coding region or 50 UTR of mRNAs to block their translation by stalling ribosomal translocation (Darnell et al., 2011) but also interacts with Ago2 of the RISC in the miRNA pathway (Muddashetty et al., 2011). Our study provides evidence showing that miR-181d associates FMRP and represses the translation of Map1b and Calm1. During the longdistance axonal delivery of RNPs, FMRP may have dual roles wherein it anchors Map1b and Calm1 to cytoskeletal motors and functions in miR-181d-mediated RISC assembly to repress gene translation. In addition, other pathways may be involved in the axonal transport and local regulation of miR-181d and its targets, because the transport of these transcripts is only


partially affected by silencing FMRP, and approximately 42% of miR-181d-positive granules, 53% of Map1b-positive granules, and 36% of Calm1-positive granules contain FMRP in axons. Furthermore, our biochemical evidence supporting the association of FMRP with miR-181d in the cell bodies of DRG neurons implies that mature miRNAs can be transported from the cell body to supply the axonal abundance of miRNAs, although pre-miRNAs have also been found in the axons, where they generate mature miRNAs (Hancock et al., 2014). The precise molecular mechanisms underlying the FMRP-mediated axonal delivery of pre- or mature miRNAs require further investigation. It has been established that target-derived neurotrophin is required for the development of the peripheral nervous system (Patel et al., 2000). NGF signaling is crucial for neuronal survival and axon elongation in primary sensory neurons (Zhou et al., 2004). From our study, NGF triggers the axonal translation of Map1b and Calm1 in primary sensory neurons, indicating a mechanism by which neurotrophin is linked to local microtubule stabilization and calcium signaling. NT-3 is unable to affect the axonal translation of Map1b and Calm1, indicating the effector specificity of NGF-induced local translation. We provide a model that demonstrates that, in the presence of a low concentration of NGF or in its absence, the axonal translation of Map1b and Calm1 is suppressed by both FMRP and miR-181d in FMRP granules, whereas in the presence of 100 ng/ml NGF, Map1b and Calm1 are released from FMRP granules and made available for local translation. After 2 hr of NGF treatment, the re-association of mRNAs with FMRP and the replenishment of mRNAbound FMRP were not able to supply the same amount of released Map1b and Calm1. Because miR-181d is localized in approximately 27% FMRP granules containing Map1b-30 UTR and approximately 25% FMRP granules expressing Calm1-30 UTR in axons, the FMRP granules lack of miR-181d also contributes to NGF-induced local translation of Map1b and Calm1. Unlike the activity-dependent dissociation of mRNA and the RISC from the FMRP complex in postsynaptic regions (Muddashetty et al., 2011), miR-181d and Ago2 were still bound to the FMRP complex in response to NGF signaling following the release of Map1b and Calm1 for subsequent local translation. Signaling by metabotropic glutamate receptors phosphorylates FMRP at S499 and causes the release of mRNA and the RISC from the FMRP complex (Muddashetty et al., 2011); however, NGF signaling did not affect the phosphorylation of FMRP (data not shown). The mechanisms by which various signals trigger different elements to reorganize the FMRP complex require further investigation. The present study substantially expands our understanding of the molecular mechanisms of FMRP functions by showing that it cooperates with miR-181d to target Map1b and Calm1 in the axons of primary sensory neurons during axon development.

mimics and inhibitors, siRNAs, and TPs are provided in the Supplemental Experimental Procedures. The construction of plasmids Luc-MS2-24XbsMap1b-30 UTR, Luc-MS2-24Xbs-Calm1-30 UTR, and other plasmids used for luciferase reporter or shRNA assays and for FMRP and scrambles is also provided in the Supplemental Experimental Procedures. In Situ Hybridization A detailed description of the procedure is provided in the Supplemental Experimental Procedures. miR-181d and its targeting mRNAs were detected as reported, with some modifications (Bicker et al., 2013; Li et al., 2012). Live-Cell Imaging for mRNA Granules Electroporated P14 DRG neurons at 4 days in vitro (DIV) were cultured on glass-bottom dishes on a temperature-controlled workstation (37 C) and imaged using a UltraView Vox system (PerkinElmer). Regions of axon segments 100–150 mm from the cell body were selected, and 150 frames were collected during a 5 min period. A kymograph was generated using ImageJ 1.4 software (NIH). The number and average velocity of RNA granules in a kymograph were analyzed using Image-Pro Plus 5.1 software (Media Cybernetics). Axon Elongation Assay A detailed description of the procedure is provided in the Supplemental Experimental Procedures. We detected the axon growth rate as described previously, with some modifications (Ye et al., 2003). Briefly, in microfluidic cultures of E13.5 DRG neurons at 7 DIV, the clear and longest distal axons in the axon compartment were selected before and 48 hr after transfection with synthetic siRNA, miRNA mimics, or an inhibitor. The increase in axon length over 48 hr was analyzed using Neurolucida software (MBF Bioscience). For NGF-induced axon growth assays, axons from microfluidic chambers were monitored after the indicated treatment or after AAV-mediated infection using 10 min intervals for a total of 60 min using a live-cell imaging microscope, as previously reported (Hengst et al., 2009). Axon length was analyzed using an ImageJ plug-in (NIH). RNA Immunoprecipitation A detailed description of the procedure is provided in the Supplemental Experimental Procedures. Briefly, the supernatants from DRG tissues and cultured PC12 cells were pre-cleared using protein G sepharose and then incubated with 5 mg FMRP antibody (Abcam) conjugated to immobilized beads, as previously reported (Zalfa et al., 2003). Total RNA was extracted from the beads using Trizol and prepared for the subsequent detection of FMRP-associated mRNAs and miRNAs. Cell Culture, Transfection, and Immunostaining; RNA Extraction, RT-PCR, and qPCR; Luciferase Reporter Assay; Immunoblotting; Drug Treatment; and FRAP Detailed descriptions of these procedures are provided in the Supplemental Experimental Procedures. Statistical Analysis All data are presented as mean ± SEM. Statistical analyses were performed using two-tailed paired Student’s t tests in Prism (GraphPad). Differences were considered significant at p < 0.05. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, seven figures, and four tables and can be found with this article online at http://dx.doi.org/10.1016/j.celrep.2015.11.057.

EXPERIMENTAL PROCEDURES Animals, Plasmids, miRNA Mimics and Inhibitors, siRNAs, and TPs All animal experiments were approved by the Committee of Use of Laboratory Animals in the Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. Descriptions of animal information and the synthesis of miRNA

AUTHOR CONTRIBUTIONS L.B., B.W., L.P., and X.Z. designed the research. B.W. and L.P. performed the majority of the experiments with the help of Q.W. and M.W. and analyzed the


data. W.-W.L. and N.W. made the microfluidic devices as instructed by X.-Y.J. L.B., B.W., and L.P. wrote the manuscript.

of embryonic and adult sensory axons reveals changes in mRNA repertoire localization. RNA 17, 85–98.

ACKNOWLEDGMENTS

Hancock, M.L., Preitner, N., Quan, J., and Flanagan, J.G. (2014). MicroRNA132 is enriched in developing axons, locally regulates Rasa1 mRNA, and promotes axon extension. J. Neurosci. 34, 66–78.

We thank Dr. Mofang Liu, Dr. Hong Cheng, and Minghua Lu for technical help to set up the experiments and for comments on the project, and we thank Dr. Zilong Qiu for the Fmr1I304N mice. This work was supported by grants from the National Natural Science Foundation of China (31271141 and 31330046), National Basic Research Program of China (2014CB942802), China Postdoctoral Science Foundation (2012M510096 and 2013T60469), Postdoctoral Foundation from Shanghai Institutes for Biological Sciences (2012KIP505), and Shanghai Postdoctoral Science Foundation (12R21417300). Received: June 15, 2015 Revised: October 15, 2015 Accepted: November 17, 2015 Published: December 17, 2015 REFERENCES Andreassi, C., Zimmermann, C., Mitter, R., Fusco, S., De Vita, S., Saiardi, A., and Riccio, A. (2010). An NGF-responsive element targets myo-inositol monophosphatase-1 mRNA to sympathetic neuron axons. Nat. Neurosci. 13, 291–301. Antar, L.N., Li, C., Zhang, H., Carroll, R.C., and Bassell, G.J. (2006). Local functions for FMRP in axon growth cone motility and activity-dependent regulation of filopodia and spine synapses. Mol. Cell. Neurosci. 32, 37–48. Bassell, G.J., and Warren, S.T. (2008). Fragile X syndrome: loss of local mRNA regulation alters synaptic development and function. Neuron 60, 201–214. Bicker, S., Khudayberdiev, S., Weiß, K., Zocher, K., Baumeister, S., and Schratt, G. (2013). The DEAH-box helicase DHX36 mediates dendritic localization of the neuronal precursor-microRNA-134. Genes Dev. 27, 991–996. Cheever, A., and Ceman, S. (2009). Translation regulation of mRNAs by the fragile X family of proteins through the microRNA pathway. RNA Biol. 6, 175–178. Chen, X.Q., Wang, B., Wu, C., Pan, J., Yuan, B., Su, Y.Y., Jiang, X.Y., Zhang, X., and Bao, L. (2012). Endosome-mediated retrograde axonal transport of P2X3 receptor signals in primary sensory neurons. Cell Res. 22, 677–696.

Hengst, U., Deglincerti, A., Kim, H.J., Jeon, N.L., and Jaffrey, S.R. (2009). Axonal elongation triggered by stimulus-induced local translation of a polarity complex protein. Nat. Cell Biol. 11, 1024–1030. Jung, H., Yoon, B.C., and Holt, C.E. (2012). Axonal mRNA localization and local protein synthesis in nervous system assembly, maintenance and repair. Nat. Rev. Neurosci. 13, 308–324. Kaplan, B.B., Kar, A.N., Gioio, A.E., and Aschrafi, A. (2013). MicroRNAs in the axon and presynaptic nerve terminal. Front. Cell. Neurosci. 7, 126. Klemmer, P., Meredith, R.M., Holmgren, C.D., Klychnikov, O.I., Stahl-Zeng, J., Loos, M., van der Schors, R.C., Wortel, J., de Wit, H., Spijker, S., et al. (2011). Proteomics, ultrastructure, and physiology of hippocampal synapses in a fragile X syndrome mouse model reveal presynaptic phenotype. J. Biol. Chem. 286, 25495–25504. Li, L., Wei, D., Wang, Q., Pan, J., Liu, R., Zhang, X., and Bao, L. (2012). MEC-17 deficiency leads to reduced a-tubulin acetylation and impaired migration of cortical neurons. J. Neurosci. 32, 12673–12683. Lin, A.C., and Holt, C.E. (2008). Function and regulation of local axonal translation. Curr. Opin. Neurobiol. 18, 60–68. McCue, H.V., Haynes, L.P., and Burgoyne, R.D. (2010). The diversity of calcium sensor proteins in the regulation of neuronal function. Cold Spring Harb. Perspect. Biol. 2, a004085. McNeill, E., and Van Vactor, D. (2012). MicroRNAs shape the neuronal landscape. Neuron 75, 363–379. Montenegro-Venegas, C., Tortosa, E., Rosso, S., Peretti, D., Bollati, F., Bisbal, M., Jausoro, I., Avila, J., Ca´ceres, A., and Gonzalez-Billault, C. (2010). MAP1B regulates axonal development by modulating Rho-GTPase Rac1 activity. Mol. Biol. Cell 21, 3518–3528. Muddashetty, R.S., Nalavadi, V.C., Gross, C., Yao, X., Xing, L., Laur, O., Warren, S.T., and Bassell, G.J. (2011). Reversible inhibition of PSD-95 mRNA translation by miR-125a, FMRP phosphorylation, and mGluR signaling. Mol. Cell 42, 673–688.

Chi, S.W., Zang, J.B., Mele, A., and Darnell, R.B. (2009). Argonaute HITS-CLIP decodes microRNA-mRNA interaction maps. Nature 460, 479–486.

Nakamuta, S., Funahashi, Y., Namba, T., Arimura, N., Picciotto, M.R., Tokumitsu, H., Soderling, T.R., Sakakibara, A., Miyata, T., Kamiguchi, H., and Kaibuchi, K. (2011). Local application of neurotrophins specifies axons through inositol 1,4,5-trisphosphate, calcium, and Ca2+/calmodulin-dependent protein kinases. Sci. Signal. 4, ra76.

Dajas-Bailador, F., Bonev, B., Garcez, P., Stanley, P., Guillemot, F., and Papalopulu, N. (2012). MicroRNA-9 regulates axon extension and branching by targeting Map1b in mouse cortical neurons. Nat. Neurosci. 15, 697–699.

Ouyang, Y.B., Lu, Y., Yue, S., Xu, L.J., Xiong, X.X., White, R.E., Sun, X., and Giffard, R.G. (2012). miR-181 regulates GRP78 and influences outcome from cerebral ischemia in vitro and in vivo. Neurobiol. Dis. 45, 555–563.

Darnell, J.C., Van Driesche, S.J., Zhang, C., Hung, K.Y., Mele, A., Fraser, C.E., Stone, E.F., Chen, C., Fak, J.J., Chi, S.W., et al. (2011). FMRP stalls ribosomal translocation on mRNAs linked to synaptic function and autism. Cell 146, 247–261.

Patel, T.D., Jackman, A., Rice, F.L., Kucera, J., and Snider, W.D. (2000). Development of sensory neurons in the absence of NGF/TrkA signaling in vivo. Neuron 25, 345–357.

Dictenberg, J.B., Swanger, S.A., Antar, L.N., Singer, R.H., and Bassell, G.J. (2008). A direct role for FMRP in activity-dependent dendritic mRNA transport links filopodial-spine morphogenesis to fragile X syndrome. Dev. Cell 14, 926–939.

Saba, R., Sto¨rchel, P.H., Aksoy-Aksel, A., Kepura, F., Lippi, G., Plant, T.D., and Schratt, G.M. (2012). Dopamine-regulated microRNA miR-181a controls GluA2 surface expression in hippocampal neurons. Mol. Cell. Biol. 32, 619–632.

Fineberg, S.K., Kosik, K.S., and Davidson, B.L. (2009). MicroRNAs potentiate neural development. Neuron 64, 303–309.

Sasaki, Y., Gross, C., Xing, L., Goshima, Y., and Bassell, G.J. (2014). Identification of axon-enriched microRNAs localized to growth cones of cortical neurons. Dev. Neurobiol. 74, 397–406.

Fusco, D., Accornero, N., Lavoie, B., Shenoy, S.M., Blanchard, J.M., Singer, R.H., and Bertrand, E. (2003). Single mRNA molecules demonstrate probabilistic movement in living mammalian cells. Curr. Biol. 13, 161–167.

Staton, A.A., and Giraldez, A.J. (2011). Use of target protector morpholinos to analyze the physiological roles of specific miRNA-mRNA pairs in vivo. Nat. Protoc. 6, 2035–2049.

Goldie, B.J., Dun, M.D., Lin, M., Smith, N.D., Verrills, N.M., Dayas, C.V., and Cairns, M.J. (2014). Activity-associated miRNA are packaged in Map1b-enriched exosomes released from depolarized neurons. Nucleic Acids Res. 42, 9195–9208.

Villarroel-Campos, D., and Gonzalez-Billault, C. (2014). The MAP1B case: an old MAP that is new again. Dev. Neurobiol. 74, 953–971.

Gumy, L.F., Yeo, G.S., Tung, Y.C., Zivraj, K.H., Willis, D., Coppola, G., Lam, B.Y., Twiss, J.L., Holt, C.E., and Fawcett, J.W. (2011). Transcriptome analysis

Wang, H., Dictenberg, J.B., Ku, L., Li, W., Bassell, G.J., and Feng, Y. (2008). Dynamic association of the fragile X mental retardation protein as a messenger ribonucleoprotein between microtubules and polyribosomes. Mol. Biol. Cell 19, 105–114.


White, F.A., Silos-Santiago, I., Molliver, D.C., Nishimura, M., Phillips, H., Barbacid, M., and Snider, W.D. (1996). Synchronous onset of NGF and TrkA survival dependence in developing dorsal root ganglia. J. Neurosci. 16, 4662– 4672.

Zang, J.B., Nosyreva, E.D., Spencer, C.M., Volk, L.J., Musunuru, K., Zhong, R., Stone, E.F., Yuva-Paylor, L.A., Huber, K.M., Paylor, R., et al. (2009). A mouse model of the human fragile X syndrome I304N mutation. PLoS Genet. 5, e1000758.

Ye, H., Kuruvilla, R., Zweifel, L.S., and Ginty, D.D. (2003). Evidence in support of signaling endosome-based retrograde survival of sympathetic neurons. Neuron 39, 57–68.

Zhao, J., Lee, M.C., Momin, A., Cendan, C.M., Shepherd, S.T., Baker, M.D., Asante, C., Bee, L., Bethry, A., Perkins, J.R., et al. (2010). Small RNAs control sodium channel expression, nociceptor excitability, and pain thresholds. J. Neurosci. 30, 10860–10871.

Zalfa, F., Giorgi, M., Primerano, B., Moro, A., Di Penta, A., Reis, S., Oostra, B., and Bagni, C. (2003). The fragile X syndrome protein FMRP associates with BC1 RNA and regulates the translation of specific mRNAs at synapses. Cell 112, 317–327.

Zhou, F.Q., Zhou, J., Dedhar, S., Wu, Y.H., and Snider, W.D. (2004). NGFinduced axon growth is mediated by localized inactivation of GSK-3beta and functions of the microtubule plus end binding protein APC. Neuron 42, 897–912.


Supplemental Information FMRP-Mediated Axon Delivery of miR-181d Regulates Axon Elongation by Locally Targeting Map1b and Calm1 Bin Wang, Lin Pan, Manyi Wei, Qiong Wang, Wen-Wen Liu, Nuoxin Wang, Xing-Yu Jiang, Xu Zhang, Lan Bao

Figure S1. Related to Figure 1, Axon-Enriched miR-181d Targets Map1b and Calm1 in Embryonic DRG Neurons. (A) Schematic of cultured DRG neurons in microfluidic chamber. The cell bodies of cultured DRG neurons are located in the cell body compartment and the axons cross microchannels to the axon compartment. The length of microchannels is 300 m. (B) RT-PCR showed that -actin was exclusively distributed in the cell bodies of E13.5 DRG neurons at 7 DIV in the microfluidic chamber, whereas Impa1 and -actin were observed in both cell bodies and axons. (C) Immunoblotting showed that MAP2 was

only detected in the cell bodies of E13.5 DRG neurons at 7 DIV in the microfluidic chamber, whereas Tau-1 was expressed in both cell bodies and axons. Tubulin was used as a loading control. (D) Relative expression level of miRNAs in the axons versus cell bodies using qPCR analysis showed that miR-181d was the most enriched microRNA in the axon compartment of cultured E13.5 DRG neurons at 7 DIV. The results are presented as the mean ± SEM (n = 3). (E) Venn diagram showing nine mRNAs (white area) of predicated miR-181d targets using TargetScan and miRanda softwares from the known transcripts (Raw value >100) existed in axons. (F) A qPCR analysis showed that miR-181d was upregulated by miR-181d mimics and downregulated by miR-181d inhibitor, but not changed by miR-21 inhibitor in cultured E13.5 DRG neurons at 4 DIV. The results are presented as the mean ± SEM. ** P < 0.01 and *** P < 0.001, versus the negative control (n = 3). (G) Luciferase reporter assay showed that Map1b and Calm1 were potential downstream targets of miR-181d. The luciferase 3’ UTR activity of Map1b and Calm1 were downregulated by miR-181d mimics in HEK293 cells, whereas the luciferase 3’ UTR activity of Rab11A, Arhgef3, Eif4a2, Dpp8, Pdap1, Ppp4r2 and Ankrd13c remained unchanged. The results are presented as the mean ± SEM. * P < 0.05, versus negative control (n = 3). (H) Sequences of the WT and mutanted (mut) 3’ UTR for Map1b and Calm1. The miR-181d seed sequence is highlighted in red and the seed match sequences of 3’ UTR for Map1b and Calm1 are highlighted in blue. (I) Luciferase reporter assay showed that Map1b and Calm1 were not the potential downstream targets for miR-497. The luciferase 3’ UTR activity of Map1b and Calm1 were unchanged by

miR-497 mimics in HEK293 cells. The results are presented as the mean ± SEM (n = 3). (J) A qPCR analysis showed that the mRNA levels of Map1b and Calm1 as well as

-actin were not changed by miR-181d inhibitor and mimics in cultured E13.5 DRG neurons at 4 DIV. The results are presented as the mean ± SEM (n = 3). (K) The representative images showed that -actin (red) was observed in both cell bodies and axons (arrowheads) in cultured P14 DRG neurons at 4 DIV, whereas -actin (red) was only observed in the cell bodies. GFP (green) was used to indicate the neuronal profile. Scale bar, 10 m.

Figure S2. Related to Figures 1 and 2, Specificity of MAP1B and Calmodulin Affected by miR-181d in Embryonic DRG Neurons. (A and B) Representative immunoblots (A) and quantitative data (B) showed that applying a miR-497 mimics did not affect the protein levels of MAP1B and calmodulin in cultured E13.5 DRG neurons at 4 DIV. Tubulin was used as a loading control. The results are presented as the mean ± SEM (n = 3). (C and D) Representative immunoblots (C) and quantitative data (D) showed that applying a miR-181d inhibitor and its mimics did not change the protein level of -actin in cultured E13.5 DRG neurons at 4 DIV. Tubulin was used as a loading control. The

results are presented as the mean ± SEM (n = 3). (E) Schematic of miRNA transfection (control miRNA NC with Cy3 dye) in the axon compartment of microfluidic chamber. (F) The phase contrast images showing that Cy3 dye was selectively observed in the axons but not the cell bodies of E13.5 DRG neurons at 7 DIV after co-transfection of NC with Cy3 dye in the axon compartment of microfluidic chamber. Scale bar, 10 m. (G and H) Representative images of microfluidic cultured E13.5 DRG axons at 4 DIV (G) and quantitative result of axon growth rates (H) showing that the axonal application of a miR-497 mimics did not affect axon elongation. The results are presented as the mean ± SEM (n = 3). Scale bar, 300 m.

Figure S3. Related to Figure 2, Map1b-TP and Calm1-TP Increase the Protein Levels of MAP1B and Calmodulin in Embryonic DRG Neurons. (A) Schematic model showing that the interaction between the 3’ UTR of target and RNA-induced silencing complex (RISC) was disrupted by target protector (TP) (upper panel). Sequences of Map1b-TP and Calm1-TP (highlighted in blue) matched the sequences of 3’ UTR for Map1b and Calm1 as well as mature miR-181d (highlighted in red) (down panel). (B and C) Representative immunoblots (upper panel) and quantitative result (down panel) showed that MAP1B (B) and calmodulin (C) were dose-dependently increased by application of Map1b-TP and Calm1-TP, respectively, in cultured E13.5 DRG neurons at 4 DIV. Tubulin was used as a loading control. The results are presented as the mean ± SEM (n = 3). (D) A qPCR analysis showed that Map1b, Calm1 and -actin as well as miR-181d were not changed after transfection with Map1b-TP or Calm1-TP in cultured E13.5 DRG neurons at 4 DIV.

The results are presented as the mean ± SEM (n = 3).

Figure S4. Related to Figure 3, Association of FMRP with Map1b and Calm1 Is Not Dependent on miR-181d. (A and B) RT-PCR (A) and qPCR analysis (B) showing that the association of Map1b and Calm1 with FMRP was not dependent on miR-181d in cultured PC12 cells expressing FMRP-GFP-FLAG. The results are presented as the mean ± SEM (n = 3).

Figure S5. Related to Figures 4 and 5, The Neuronal Morphology, the Axonal Transport of -actin, MAP1B and Calmodulin Are Not Affected by Knockdown of FMRP in Cultured DRG Neurons. (A and B) Phase contrast images of AAV-infected E13.5 DRG neurons at 7 DIV (A) and electroporated P14 DRG neurons at 4 DIV (B) affected by FMRP shRNA

displayed morphologies similar to those of neurons transfected with control (Ctrl) shRNA. FMRP (red) was immunostained to detect the efficiency of FMRP knockdown and GFP (green) was used to indicate the neuronal profile. Scale bar, 10 m. (C) -actin-3’ UTR was not co-localized with FMRP-GFP in the axon of cultured P14 DRG neurons at 4 DIV. Scale bar, 10 m. (D-F) Representative live-cell images (D) and quantitative results (E) showed that knockdown of FMRP by FMRP shRNA (green) did not affect the number of -actin-3’ UTR granules (red) in axons showing both anterograde and retrograde movement as well as non-significant movement in cultured P14 DRG neurons at 4 DIV. The mean velocities of anterograde and retrograde movement for -actin-3’ UTR granules were not significantly changed (F). A 300 s video was collected and a kymograph was created to demonstrate the movement of -actin-3’ UTR granules (D). The results are presented as the mean ± SEM (n = 3). Scale bar, 10 m. (G-I) Representative FRAP images (G and H) and quantitative data (I) showed that the fluorescence recovery (green) of MAP1B-GFP (G) and calmodulin-GFP (H) were not affected after knockdown of FMRP by FMRP shRNA (red) in the axons of cultured P14 DRG neurons at 4 DIV. The results are presented as the mean ± SEM (n = 3). Scale bar, 10 m.

Figure S6. Related to Figure 6, NT-3 Is Unable to Induce Axonal Synthesis of MAP1B and Calmodulin. Representative immunoblots (A) and quantitative data (B) showed that axonal application of 100 ng/ml NT-3 was unable to affect the axonal synthesis of MAP1B and calmodulin in microfluidic cultures of E13.5 DRG neurons at 7 DIV. The results are presented as the mean ± SEM (n = 3). Tubulin was used as a loading control.

Figure S7. Related to Figure 7, FMRP Is Not Regulated by miR-181d Mimics. Representative immunoblots (A) and quantitative data (B) showed that the protein level of FMRP was not affected by miR-181d mimics in cultured E13.5 DRG neurons at 4 DIV. The results are presented as the mean ± SEM (n = 3). Tubulin was used as a loading control.

Table S1. Sequences of miRNAs, siRNAs, shRNAs and Target Protectors, Related to Experimental Procedures. miRNA/siRNA/ shRNA/TP mature miR-181d mature miR-21 negative control si-Map1b si-Calm1 si-control FMRP shRNA

Sequence (5’3’)

AACAUUCAUUGUUGUCGGUGGGU UAGCUUAUCAGACUGAUGUUGA UUUGUACUACACAAAAGUACUG TTTATCGGTATGAAGCCCGTT TTCTTCTTCGCTATCTGTGTT ACGTGACACGTTCGGAGAATT CCGGGTGATGAAGTTGAGGTTTATTCAAGAGATAAACC TCAACTTCATCACTTTTTTG scrambled shRNA CCGGTTCTCCGAACGTGTCACGTTTCAAGAGAACGTG (Ctrl shRNA) ACACGTTCGGAGAATTTTTTG Map1b-TP TGTACATTCAAGTCACTTCCTAAG Calm1-TP GTACTTTTGACATTCAGTTTTGCTA scrambled TP GTGTAACACGTCTATACGCCCA (negative control)

Table S2. Companies and Working Concentration of Antibodies, Related to Experimental Procedures. Antibody

Company

Ago2 calmodulin ERK GFP FMRP MAP1B MAP2 pERK Tau-1 tubulin Tuj-1 Tuj-1

Sigma Millipore Santa Cruz Roche Abcam Abcam Sigma Santa Cruz Chemicon Sigma Chemicon Abcam

Concentration Immunostaining

1:2000 1:1000 1:10000

for Concentration Immunoblotting 1:1000 1:2000 1:2000 1:2000 1:2000 1:2000 1:1000 1:2000 1:1000 1:10000

for

Table S3. Primer Sequences for RT-PCR Detection, Related to Experimental Procedures. Gene (Rat) Forward Primer Map1b CCCCGTTCAGAACTCTCAGA Calm1

-actin -actin Impa1 c-Jun Fmr1 Gapdh

Reverse Primer TGCAACTTATCTGGCT GAGA CACTGGGTCAGAACCCAACAG GGACCACCAACCAAT ACATGC AGCCATGTACGTAGCCATCC’ CTCTCAGCTGTGGTGG TGAA ATGTGGCTCGGTCACTTGG TTTACAGGTGTCGATG CAAACG AGTGATCCTCGCAAGACAAGCTGG CGCAAAGTCTCCGGCT TTCTGG AGGCTAGATTGCGGATGAACTCC GAGGGACTACAGGCT CTCACC GGAACTAATTCTGAAGCATCAAATG CGTCATTTCCTTTGAA ACCTCC GGCAAGTTCAACGGCACAG CGCCAGTAGACTCCAC GAC

Table S4. Primer Sequences for Detection of qPCR, Related to Experimental Procedures. Gene/miRNA Forward Primer Map1b (Rat) Map1b (Mouse) Calm1 (Rat/Mouse) Fmr1 (Rat) Fmr1 (Mouse) -actin (Rat/Mouse) Gapdh (Rat) Gapdh (Mouse) miR-181d (Rat/Mouse) miR-21 (Rat/Mouse) miR-9 (Rat/Mouse) miR-132 (Rat/Mouse) miR-19b (Rat/Mouse) miR-365 (Rat/Mouse) miR-193 (Rat/Mouse) miR-377

Reverse Primer/ Reverse Transcription Primer CTGGAGATGTACGTGCTTAA GTCAGGTAGGAAATCGG GATG GCTTGAGATGTACGTGCTTAA AGGTTAAGTAGGAAATC C GGGATG GTCATGCGGTCACTGGGTCAG CAAGAACTCTGGGAAGT CAATGG GGAACTAATTCTGAAGCATCA CGTCATTTCCTTTGAAAC AATG CTCC GGAACTAATTCTGAAGCATCA CGTCGTTTCCTTTGAAGC AATG CTCC GCAAGCAGGAGTACGATGAG TGCGCAAGTTAGGTTTTG TC TCAAAG GGCAAGTTCAACGGCACAG CGCCAGTAGACTCCACG AC TATGTCTTGGAGTCTACTGGT GTTGTCATATTTCTCGTG GTCTTCACC GTTCACACCC ACACTCCAGCTGGGAACATTC TCAACTGGTGTCGTGGA ATTGTTGTC GTCGGCAATTCAGTTGAG ACCCACCG ACACTCCAGCTGGGTAGCTTA CTCAACTGGTGTCGTGG TCAGACTGA AGTCGGCAATTCAGTTGA GTCAACATC ACACTCCAGCTGGGTCTTTGG TCAACTGGTGTCGTGGA TTATCTAGCT GTCGGCAATTCAGTTGAG TCATACAG ACACTCCAGCTGGGTAACAGT CTCAACTGGTGTCGTGG CTACAGCCA AGTCGGCAATTCAGTTGA GCGACCATG ACACTCCAGCTGGGTGTGCA TCAACTGGTGTCGTGGA AATCCATGCA GTCGGCAATTCAGTTGAG TCAGTTTT ACACTCCAGCTGGGtTAATGC TCAACTGGTGTCGTGGA CCCTAAAAAT GTCGGCAATTCAGTTGAG ATAAGGAT ACACTCCAGCTGGGAACTGG TCAACTGGTGTCGTGGA CCTACAAAG GTCGGCAATTCAGTTGAG ACTGGGAC ACACTCCAGCTGGGTGAATCA TCAACTGGTGTCGTGGA

(Rat/Mouse)

CACAAAGGC

miR-7a (Rat/Mouse)

ACACTCCAGCTGGGTGGAAG ACTAGTGATT

miR-181a (Rat/Mouse)

ACACTCCAGCTGGGAACATTC AACGCTGTC

miR-181b (Rat/Mouse)

ACACTCCAGCTGGGAACATTC ATTGCTGTC

miR-196a (Rat/Mouse) miR-345-5P (Rat/Mouse)

miR-712* (Rat/Mouse) U6 (Rat/Mouse) 5S (Rat/Mouse) universal reverse primer (Rat/Mouse)

GTCGGCAATTCAGTTGAG AAAAGTTG TCAACTGGTGTCGTGGA GTCGGCAATTCAGTTGAG ACAACAAA TCAACTGGTGTCGTGGA GTCGGCAATTCAGTTGAG ACTCACCG

TCAACTGGTGTCGTGGA GTCGGCAATTCAGTTGAG ACCCACCG ACACTCCAGCTGGGTAGGTAG TCAACTGGTGTCGTGGA TTTCATGTT GTCGGCAATTCAGTTGAG CCCAACAA ACACTCCAGCTGGGTGCTGAC TCAACTGGTGTCGTGGA CCCTAGTCC GTCGGCAATTCAGTTGAG GCACRGGA ACACTCCAGCTGGGTGCGAG TCACCCCCGG

TCAACTGGTGTCGTGGA GTCGGCAATTCAGTTGAG CAACACC ACACTCCAGCTGGGCGCAAAT CTCAACTGGTGTCGTGG TCGTGAAGC AGTCGGCAATTCAGTTGA GAAAAATAT ACACTCCAGCTGGGGCCTGG CTCAACTGGTGTCGTGG GAATACCGG AGTCGGCAATTCAGTTGA GAAAGCCTA CTCAACTGGTGTCGTGG AGTCGG

Supplemental Experimental Procedures Animals All experiments were approved by the Committee of Use of Laboratory Animals in the Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. The Sprague Dawley rats at E13.5 and P14 were provided by Shanghai Laboratory Animal Center, Chinese Academy of Sciences. Fmr1I304N mice (strain B6.129-Fmr1tm1Rbd/J) were purchased from Jackson Laboratory (Zang et al., 2009).

Plasmids pcDNA-MS2-mcherry, Luc-MS2-24Xbs and Luc-MS2-24Xbs-Map1b-3’ UTR were provided by Dr. Sossin (McGill University, Canada). MAP1B-GFP was provided by Dr. Rashmi Kothary (University of Ottawa, Canada). The full length of Calm1-3’ UTR, -actin-3’ UTR and γ-actin-3’ UTR were amplified from rat DRG cDNA. These fragments were cloned into Luc-MS2-24Xbs to produce Luc-MS2-24Xbs-Calm1-3’ UTR, Luc-MS2-24Xbs--actin-3’ UTR and Luc-MS2-24Xbs-γ-actin-3’ UTR. The full length of calmodulin was amplified from rat DRG cDNA and cloned in pEGFP-N3 to produce calmodulin-GFP. The full length of FMRP and CPEB1 was amplified from rat DRG cDNA and cloned into pAOV.CMV.GFP.FLAG to produce FMRP-GFP-FLAG and CPEB1-GFP-FLAG. The 3’ UTR of Map1b, Calm1, Rab11a, Arhgef3, Eif4a2, Dpp8, Pdap1, Ppp4r2 and Ankrd13c containing the predicted binding site of miR-181d were amplified from rat DRG cDNA and were cloned into luciferase reporter using XhoI and NotI. The mutanted 3’ UTR of Map1b and Calm1

with the seed region was obtained by using a KOD site-mutagenesis kit (Toyobo, Japan). For AAV construction and infection, the shRNA for FMRP and scrambled shRNA were cloned into the pAKD.CMV.bGlobin.eGFP.H1.shRNA vector and the full

length

of

shRNA-resistant

pAOV.CMV.mcherry.FLAG

and

FMRP

(FMRPR)

was

pAOV.CMV.BFP

cloned to

into

the

produce

AAV-FMRPR-mCherry-FLAG and FMRPR-BFP-FLAG (Obio Technology Co., Ltd. Shanghai, China). The AAV2-based vector pseudotyped with AAV8 serotype capsid (AAV2/8) was purchased and supplied in titers 1-3 1012 (Obio Technology Co., Ltd. Shanghai, China). The sequences of FMRP shRNA and scrambled shRNA (control shRNA) were indicated in Table S1.

miRNA Mimics and Inhibitor, siRNA and Target Protectors The miRNA mimics and inhibitor (Guangzhou RiboBio Co. Ltd., China), and siRNA (Shanghai GenePharma Co. Ltd., China) were synthesized. The miRCUYR LNATM microRNA Power Inhibitors (TP) (Exiqon, Denmark) were custom-designed for competing the miR-181d effects on Map1b and Calm1. The sequences of siRNAs and TPs were indicated in Table S1.

Cell Culture, Transfection and Immunostaining HEK293 cells and PC12 cells were obtained from American Type Culture Collection. HEK293 cells were cultured in Minimum Essential Medium (Invitrogen, Carlsbad, CA) containing 10% fetal bovine serum (FBS; Invitrogen) and antibiotics. PC12 cells

were cultured in RPMI medium 1640 (Invitrogen) containing 15% horse serum and 2.5% FBS. Transient expression with 100 nM miR-181d mimics, 100 nM miR-181d inhibitor, 100 nM synthetic siRNA or 50 nM TP was performed with RNAiMAX according to the manufacturer’s protocol (Invitrogen). Transfection for plasmids was performed with lipofectamine 2000 according to the manufacturer’s protocol (Invitrogen). For compartmental culture of DRG neurons, the microfluidic chamber was fabricated and dissociated DRG neurons from rats at E13.5 were plated into the cell body compartment and cultured in the Neurobasal medium containing 2% B27 supplement, 2 mM L-Glutamine (Invitrogen) and 100 ng/ml NGF (Invitrogen). Meanwhile, 10 M 5-fluoro-2’-deoxyuridine (Sigma St. Louis, MO) and 10 M uridine (Sigma) were used to inhibit the proliferation of non-neuronal cells. Axons crossed microchannels and reached the axon compartment within 24 hr. For transfection experiment, dissociated DRG neurons from rats at P14 were electroporated by Nucleofector II (Amaxa, Germany) using a neuron nucleofection kit and then plated on glass-bottom dish with DMEM containing 10% FBS. After 6 h, the culture medium was replaced with DMEM/F12 (1:1) containing 1% N2 supplement, and the neurons were maintained for 48 hr for live-cell imaging and immunostaining. For immunostaining, neurons grown on glass-bottom dish were fixed in 4% paraformaldehyde for 15 min. Then the cells were incubated with the indicated primary antibodies (Table S2) overnight at 4°C followed by secondary antibodies conjugated with FITC (1:100; Jackson Immunoresearch, West Grove, PA) for 45 min

at 37 °C, and scanned with the Leica SP8 confocal microscopy (Leica, Germany).

RNA Extraction, RT-PCR and qPCR The total RNA from tissues, cell lysates or beads was extracted using TRIzol reagent (Invitrogen). For mRNA, the first-strand cDNA was reversely transcribed using superscript III reverse transcriptase (Invitrogen). For RT-PCR, the PCR product was analyzed on 1.5% agarose gels. The qPCR with SYBR Green detection was performed using an ABI PRISM 7500 sequence detection system (Applied Biosystems, Foster City, CA). For miRNA detection, the miRNAs were reversely transcribed with gene specific stem-loop primer. All primers were indicated in Table S3 and S4. The qPCR results were analyzed and expressed as relative mRNA or miRNA level of the cycle threshold value, which were then converted to fold change.

Luciferase Reporter Assay HEK293 cells were plated into 24 well plates and co-transfected with the luciferase reporter and negative control or miR-181d mimics. Cells were harvested after transfection for 24 hr, and the firefly and renilla luciferase activity were measured using the Dual-Luciferase Reporter Assay System (Promega, Madison City, WI) with an Envision Luminometer (PerkinElmer Inc., Waltham, MA).

Immunoblotting Tissues, cell lysates or beads were incubated in SDS-PAGE loading buffer for 20-30

min at 50°C. The samples were separated on SDS-PAGE, transferred, probed with the indicated antibodies (Table S2), and visualized with enhanced chemiluminescence (Amersham Biosciences, Litter Chalfont, UK). The intensity of immunoreactive bands was analyzed with the Image-Pro Plus 5.1 software.

In situ Hybridization The Locked Nucleic Acid (LNA)-modified 5’-digoxigenin (DIG) and 3’-DIG labeled probes (25 nM; Exiqon) were used to specifically detect mature miR-181d. Scrambled probe (25 nM; Exiqon) served as a negative control. The detection of miR-181d was performed as previously described with some modifications (Bicker et al., 2013). Cultured E13.5 DRG neurons were fixed in 4% paraformaldehyde and hybridized with DIG-labeled miR-181d probe at 55°C overnight. After incubation with anti-DIG-POD (Roche, Indianapolis, IN) at 4°C overnight, the signal was amplified by the TSATM Plus fluorescein and Cy3 System (PerkinElmer Inc.) according to the manufacturer’s protocol. The mRNA detection was performed as previously described (Li et al., 2012). Briefly, antisense probe for Map1b and Calm1 were amplified with PCR primers and labeled with DIG. The cultured E13.5 DRG neuron was fixed in 4% paraformaldehyde and hybridized at 65°C overnight. After incubation with anti-DIG-POD (Roche) at 4°C overnight, the signal was amplified by TSATM Plus Cy3 System (PerkinElmer Inc.). For immunostaining experiments, the neurons were incubated with the indicated primary antibody (Table S2) overnight at 4°C followed by secondary antibodies for 45 min at 37 °C before TSA amplification. The cells were

scanned by LSM 710 confocal microscopy (Zeiss, Germany) and Leica SP8 confocal microscopy.

Axonal Elongation Assays We detected the axon growth rate as described previously with some modifications (Ye et al., 2003). In microfluidic cultures of E13.5 DRG neurons in presence of 100 ng/ml NGF, the clear and longest distal axons were selected before indicated treatment. The images of distal axons were captured by the PerkinElmer UltraView Vox system using a 10 lens. Then the photographs from axon compartment were stitching to produce an integrated image for whole distal axon compartment. After 48 hr transfection with synthetic siRNA, miRNA mimics or inhibitor on axon compartment, the second image was captured at the same region. The increase in axon length over 48 hr was analyzed using Neurolucida software (MBF Bioscience, Williston, VT). At least 5 chambers were observed in one experiment and three independent experiments were performed for statistical analysis. For live-cell imaging of the NGF-induced axon growth, before NGF treatment the medium in the axon compartment of microfluidic chamber was replaced with low concentration of 5 ng/ml NGF for 12 hr, which was used to maintain neuronal survival and keep axons with a slow growth rate as previously reported (Hengst et al., 2009). After application of 100 ng/ml NGF for 2 hr in the axon compartment, the axon growth was monitored for another 1 hr by live-cell imaging. To investigate the effect of miR-181d on the NGF-induced axon growth, miR-181d mimics were

selectively transfected in the axon compartment for 48 hr before NGF treatment. To track more axons, the phase contrast images were captured by using a 40  lens from UltraView Vox system (PerkinElmer Inc.) through multi point scanning on one microfluidic chamber. At least 20 axons of a microfluidic chambers were monitored with a 10 min intervals for a total of 60 min. For the AAV-mediated transfection, the lowest laser intensity was used to monitor the growth of GFP-positive axons to minimize the photo toxicity. The DIC and fluorescent images were captured from AAV-infected chambers. Axon length was analyzed using an Image J plug-in (NIH, Bethesda, MA). At least 20 axons for a chamber, 5 chambers for one experiment and three independent experiments were performed for statistical analysis.

Photobleaching after Fluorescence Recovery Photobleaching after fluorescence recovery (FRAP) was performed as previously described (Funahashi et al., 2013). Briefly, we transfected MAP1B-GFP or calmodulin-GFP into cultured P14 DRG neurons expressing either FMRP shRNA or control shRNA. At 4 DIV, the axons of GFP-positive DRG neurons were photobleached with laser irradiations and then the axons were monitored with 5-s intervals for a total of 4 min for MAP1B-GFP and 0.5 s intervals for a total of 2 min for calmodulin-GFP. Fluorescent images were captured by using a 40  lens from UltraView Vox system and the fluorescent intensity was analyzed by Volocity for FRAP (PerkinElmer Inc.).

RNA Immunoprecipitation DRG from rats at P14 and PC12 cells were used for immunoprecipitation as previously described with some modifications (Zalfa et al., 2003). Briefly, the tissues or cultured cells were lysed in the RIPA buffer (20 mM Tris 7.5, 150 mM NaCl, 5 mM MgCl2, 1%NP40) with protease and RNase inhibitors (Promega). The resulting supernatants were pre-cleared with protein G sepharose and then incubated with 5 g FMRP antibody (Abcam, Cambridge, UK) conjugated to immobilized beads. The same amount of IgG was used as control. For cultured cells, 5 g FLAG antibody (Sigma) was incubated with supernatant from PC12 cells transfected with GFP-FLAG, FMRP-GFP-FLAG or CPEB1-GFP-FLAG in the absence or presence of 100 ng/ml NGF. The beads were washed in RIPA buffer with 150, 250, 350, 450 and 550 mM NaCl. Total RNA was extracted from the beads using Trizol and prepared for the subsequent detection of FMRP-associated mRNAs and miRNAs.

Drug Treatment For inhibition of protein synthesis, 100 g/ml cycloheximide (CHX; Sigma) was added for 2 hr followed with 100 ng/ml NGF treatment in the axon compartment of microfluidic cultured E13.5 DRG neurons. In all experiments, the control group was treated with the vehicle used for drug preparation.

Supplemental References Bicker, S., Khudayberdiev, S., Weiss, K., Zocher, K., Baumeister, S., and Schratt, G.

(2013). The DEAH-box helicase DHX36 mediates dendritic localization of the neuronal precursor-microRNA-134. Genes Dev 27, 991-996. Funahashi, Y., Namba, T., Fujisue, S., Itoh, N., Nakamuta, S., Kato, K., Shimada, A., Xu, C., Shan, W., Nishioka, T., et al. (2013). ERK2-mediated phosphorylation of Par3 regulates neuronal polarization. J Neurosci 33, 13270-13285. Hengst, U., Deglincerti, A., Kim, H.J., Jeon, N.L., and Jaffrey, S.R. (2009). Axonal elongation triggered by stimulus-induced local translation of a polarity complex protein. Nat Cell Biol 11, 1024-1030. Li, L., Wei, D., Wang, Q., Pan, J., Liu, R., Zhang, X., and Bao, L. (2012). MEC-17 deficiency leads to reduced alpha-tubulin acetylation and impaired migration of cortical neurons. J Neurosci 32, 12673-12683. Ye, H., Kuruvilla, R., Zweifel, L.S., and Ginty, D.D. (2003). Evidence in support of signaling endosome-based retrograde survival of sympathetic neurons. Neuron 39, 57-68. Zalfa, F., Giorgi, M., Primerano, B., Moro, A., Di Penta, A., Reis, S., Oostra, B., and Bagni, C. (2003). The fragile X syndrome protein FMRP associates with BC1 RNA and regulates the translation of specific mRNAs at synapses. Cell 112, 317-327. Zang, J.B., Nosyreva, E.D., Spencer, C.M., Volk, L.J., Musunuru, K., Zhong, R., Stone, E.F., Yuva-Paylor, L.A., Huber, K.M., Paylor, R., et al. (2009). A mouse model of the human Fragile X syndrome I304N mutation. PLoS Genet 5, e1000758