miR-9 and miR-124 synergistically affect regulation of dendritic branching via the AKT/GSK3β pathway by targeting Rap2a

A single microRNA (miRNA) can regulate expression of multiple proteins, and expression of an individual protein may be controlled by numerous miRNAs. This regulatory pattern strongly suggests that synergistic effects of miRNAs play critical roles in regulating biological processes. miR-9 and miR-124, two of the most abundant miRNAs in the mammalian nervous system, have important functions in neuronal development. In this study, we identified the small GTP-binding protein Rap2a as a common target of both miR-9 and miR-124. miR-9 and miR-124 together, but neither miRNA alone, strongly suppressed Rap2a, thereby promoting neuronal differentiation of neural stem cells (NSCs) and dendritic branching of differentiated neurons. Rap2a also diminished the dendritic complexity of mature neurons by decreasing the levels of pAKT and pGSK3β. Our results reveal a novel pathway in which miR-9 and miR-124 synergistically repress expression of Rap2a to sustain homeostatic dendritic complexity during neuronal development and maturation.

the total number of dividing cells 21 . Furthermore, miR-124 and miR-9 regulate neural lineage differentiation in embryonic stem cells in vitro 22 .
Synergism between miR-9/9 * and miR-124 mediates the conversion of human fibroblasts to neurons, but separate expression of these miRNAs has no effect [23][24][25] . MiR-9* and miR-124 reduce proliferation of neural progenitors by repressing the Brg/Brm-associated factor BAF53a, which in turn represses its neuron-specific homolog BAF53b 26,27 , a critical factor in dendritic development. Although miR-9 and miR-124 have some distinct targets, their synergistic effects on neuronal development are still not clear and merit further investigation. In this study, we identified Rap2a as a common target gene of miR-9 and miR-124. Moreover, we found that repression of Rap2a by miR-9 and miR-124 affects the activation of AKT and GSK3β , which control neuronal differentiation and dendritic branching. Our findings reveal a novel pathway that governs dendritic branching via the synergistic effects of miR-9 and miR-124.

Results
MiR-9 and miR-124 synergistically promote dendritic branching of differentiated neurons, and Rap2a is predicted to be a common target of both miRNAs. Previous studies demonstrated that miR-9 and miR-124 play crucial roles in determining neuron fate. In addition, both of these miRNAs start to be expressed at almost the same time, and their levels gradually increase over the course of neuronal development 22,28,29 . These observations suggest that miR-9 and miR-124 have synergistic effects on neural development. Therefore, we transfected NSCs in vitro with lentiviruses that overexpress miR-9, miR-124, or both ( Fig. 1A and Supplementary Fig. S1B). Surprisingly, MAP2-positive neurons derived from NSCs co-overexpressing of miR-9 and miR-124 for 7 days had many more dendritic branches than those transfected with control virus or virus expressing miR-9 or miR-124 alone (Fig. 1A). These results suggest that miR-9 and miR-124 can synergistically regulate neurites morphology and promote dendritic branching.
To screen for target genes of miR-9 and miR-124, we used the online prediction tools TargetScan and PicTar [30][31][32] . Several Ras superfamily members were predicted to be the targets of miR-9 or miR-124 (Table 1). Among them, Rhog was previously verified as a target of miR-124 and shown to control axonal and dendritic branching 33,34 . This observation suggested that miR-9 and miR-124 regulate dendritic branching through the Ras superfamily members. Both algorithms strongly predicted that Rap2a is a common target of miR-9 and miR-124 (Table 1). Sequence analysis revealed that the 3′ UTR of Rap2a contains regions complementary to the seed regions of miR-9 and miR-124 (Fig. 1B), i.e., that the Rap2a mRNA has putative miR-9 and miR-124 binding sites in its 3′ UTR (Fig. 1B).
To determine the expression patterns of miR-9, miR-124, and Rap2a, we measured the levels of miR-9 and miR-124 in NSCs, the undifferentiated multipotent neural progenitor cell line C17.2, and mature neurons. The levels of miR-9 and miR-124 were considerable higher in postmitotic neurons than in NSCs or C17.2 cells (Fig. 1C,D). On the contrary, the level of Rap2a was much lower in postmitotic neurons than in NSC and C17.2 cells (Fig. 1E,F). Mature neurons contained a higher level of Tuj1 and lower level of nestin than NSC and C17.2 (* * P < 0.01; * * * P < 0.001). cells (Fig. 1G,H). The inverted expression patterns of miR-9/-124 and Rap2a supported our hypothesis that Rap2a is a common target of both of these miRNAs.
We also analyzed the dendritic complexity of differentiated neurons following transfection with LV-miR-EPs. The complexity of dendritic branching was analyzed in terms of in morphology, number of dendritic intersections (NDIs), and the total number of dendritic end tips (TNDEPs) (Fig. 3C-E). MAP2-positive neurons derived from NSCs had more dendritic branches, NDIs, and TNDEPs in the LV-miR-9-miR-124 (hi) group than in the LV-Ctrl and LV-miR-9-124 (lo) group ( Fig. 3C-E). Rap2V12 decreased the dendritic complexity of miR-124 miR-9

Rreb1
Ras repressor protein 1 Ras-GTPase-activating protein SH3domain binding protein 1 Table 1. Members of the Ras superfamily were predicted as conserved targets of miR-9 and miR-124 by the online prediction tools TargetScan and PicTar.
neurons transfected with LV-miR-9-124 (hi) (Fig. 3C-E). These findings suggest that miR-9 and miR-124, in a concentration-dependent manner, synergistically regulate the neuronal differentiation of NSCs and dendritic complexity of differentiated neurons. Furthermore, increasing the activity of Rap2a can diminish the synergistic effects of miR-9 and miR-124 on neuronal differentiation and dendritic branching. Next, we investigated the influence of culture time on the synergistic effects of miR-9 and miR-124 in NSCs. Both 3 and 7 days after transfection with LV-miR-9-124 [miR-9-124 (3d) and miR-9-124 (7d), respectively], NSC cultures contained more MAP2-positive cells than controls (Fig. 3F,G). In addition, dendritic complexity of MAP2-positive cells increased over time following miR-9-124 transfection ( Fig. 3H-J). However, LV-Rap2V12 also significantly decreased (P = 0.008) the number of MAP2-positive cells three days after LV-miR-9-124 transfection (Fig. 3F,G). These results suggest that miR-9 and miR-124 synergistically regulate the neuronal differentiation of NSCs and dendritic complexity of differentiated neurons in a time-dependent manner. However, elevated Rap2a activity could also diminish the synergistic effects of miR-9 and miR-124 on the dendritic complexity of MAP2-positive differentiated neurons. Thus, our results demonstrate that miR-9 and miR-124 promote neuronal differentiation of NSCs and increase dendritic branching by inhibiting Rap2a protein.  Fig. S1C), respectively. Seven days after transfected, the postmitotic neurons transfected with LV-Rap2N17 maintained dendritic branch morphology similar to that of LV-Ctrl-transfected neurons (Fig. 4A, left panel and middle panel). Dendritic analysis revealed that neither NDIs nor TNDEPs differed between LC-Ctrl-and LV-Rap2N17-transfected neurons (Fig. 4B,C). In LV-Rap2V12-transfected cells (Fig. 4A, right panel), the number of neuronal dendritic branches was strikingly reduced relative to those in LV-Ctrl-and LV-Rap2N17-transfected cells (Fig. 4A-C). These results suggested that inhibition of Rap2a is indispensable for dendritic branching and complexity of mature neurons.
AKT-GSK3β signal pathway is involved in the regulation of dendritic complexity of mature neurons by Rap2a. To identify the signaling pathway(s) involved in the regulation of dendritic complexity by Rap2a, we overexpressed miR-9-124, Rap2N17, and Rap2V12 in neurons. LV-Rap2V12 transfection considerable decreased the level of pAKT in mature neurons relative to LV-Ctrl, LV-miR-9-124, and LV-Rap2N17 transfection (Fig. 5A,B). Thus, Rap2a, but not miR-9 or miR-124, can change the level of pAKT, as mature neurons maintained high levels of miR-9 and miR-124 and a low level of Rap2a (Fig. 1C-F). This result also suggests that the AKT signaling pathway is involved in the regulation of dendritic complexity of mature neurons by Rap2a.
Glycogen synthase kinase 3 beta (GSK3β ) acts downstream of Akt, and its activity is inhibited via phosphorylation of its serine 9 residue (Ser9) by pAKT, leading to control of neurogenesis, neuronal polarization, and axonal outgrowth 35 . To further detect the influence of Rap2a on the activity of AKT and GSK3β , we forced mature neurons to overexpress Rap2a. Compared to the LV-Rap2N17 control, overexpression of Rap2V12 resulted in greater reductions in the levels of pAKT and pGSK-3β (Fig. 5C,D). This inhibition pattern was also apparent in LV-Rap2V12-transfected neurons cultivated for longer periods (Fig. 5E,F). Because miR-9 and miR-124 synergistically inhibited Rap2a translation, and NSCs contained low levels of miR-9 and miR-124 and high level of Rap2a (Fig. 1E,F), we wondered whether miR-9 and miR-124 could synergistically alter the levels of pAKT and pGSK-3β in NSCs. Neither miR-9 nor miR-124 could change the levels of pAKT or pGSK-3β in NSCs following transfection with LV-miR-EPs (Fig. 5G,H); only LV-miR-9-124 transfection could significantly increase the levels of pAKT (P = 0.0009) and pGSK-3β (P = 0.0008) in NSCs (Fig. 5G,H). These results further demonstrate that Rap2a, the common target of miR-9 and miR-124, exerts its physical roles in NSCs and neurons by regulating the activity of AKT and GSK3β .

Discussion
Relationships between miRNAs and targets can be both one-to-many and many-to-one, i.e., one miRNA can repress many proteins, and one protein can be regulated by many miRNAs. For example, miR-155 can target the bone morphogenetic protein (BMP)-responsive transcriptional factors SMAD2 and SMAD5, nuclear factor κ B (NF-κ B) inhibitor κ B-Ras1, and MyD88 to modulate macrophage responses, lymphomagenesis, hematopoiesis, and inflammation [36][37][38][39] . On the other hand, miR-15 and miR-16 control apoptosis by targeting BCL-2 mRNA 40 . MiR-224 and miR-203 downregulate NPAS4 (Neuronal Per-ARNT-SIM homology domain 4) expression through its 3′ UTR 41 . This characteristic of miRNAs and their targets has drawn increasing attention to the synergistic effects of miRNAs. For instance, miR-499 and miR-133 synergistically promote cardiac differentiation 42 . Likewise, the combined action of miR-106b, miR-93, and miR-25 effectively repress expression of PTEN transcripts in prostate cancer 43 . In this study, we observed that co-overexpression of miR-9 and miR-124 in NSCs promoted neuronal differentiation and dendritic branching, whereas neither miRNA had an effect, strongly suggesting that miR-9 and miR-124 exert synergistic effects on neuronal differentiation and dendritic tree complexity. Recent studies report that genetic switches responsible for control of neuronal gene expression are targets of both miR-9 and miR-124. MiR-9 targets repressor-element-1-silencing transcription factor (REST), and miR-9* targets CoREST 44 . MiR-124 also targets CoREST to regulate intrinsic temporal changes in RGC growth cone sensitivity and radial migration of projection neurons 45,46 . Although these studies proposed that miR-9 and miR-124 play crucial roles in neuron fate, they did not clearly elucidate the synergistic effects. Here, we showed that miR-9 and miR-124 play synergistic roles in neuron fate, and that Rap2a is their common target.
Scientific RepoRts | 6:26781 | DOI: 10.1038/srep26781 addition, overexpression of Rap2V12 could not completely offset the synergistic effects of miR-9 and miR-124, leading us to speculate that miR-9 and miR-124 may regulate neuron fate via another mechanism.
The multifunctional serine/threonine kinase GSK3β plays a variety of roles in activity-dependent regulation of dendritic development and maintenance 52,53 . Phosphorylation of GSK3β on Tyr216 leads to activation, whereas phosphorylation of Ser9 by AKT results in inactivation 35,54 . We found that levels of pAKT (phosphorylation of Ser473) and pGSK3β (phosphorylation of Ser9) were dramatically downregulated by overexpression of Rap2a in mature neurons (Fig. 5A,B). Thus, the AKT/GSK3β signaling pathway is regulated by Rap2a, and miR-9 and miR-124 can control AKT/GSK3β signaling pathway by targeting Rap2a. It is reported that in B cells Rap2V12 reduces Akt activity via PI3K inhibition 55 . Our results proved that Rap2V12 can also repress Akt activity to inhibit neuronal differentiation and dendritic branching in nervous system. Although Rap2a is involved in the JNK and ERK signaling pathways 56,57 , we did not detect obvious changes in the levels of pERK or pJNK upon overexpression of miR-9 and miR-124 in NSCs (data not shown). As homologous proteins of Rap2a, Rap2b was reported to closely correlate with cancer 58 . The biological function of Rap2c was still unclear. The roles of both Rap2b and Rap2c have not yet been reported in nervous system. Considering the vital function of Ras superfamily in nervous system, Rap2b and Rap2c may have some novel roles in differentiation of NSCs, which still need to investigate further.
Our results reveal the mechanism by which miR-9 and miR-124 synergistically promote neuronal differentiation and dendritic branching (Fig. 6). Rap2a decreases phosphorylation levels of AKT, thereby inactivating it. MiR-9 and miR-124 repress Rap2a by binding to specific sites in the Rap2a 3′ UTR, thereby releasing the inhibition of AKT, ultimately resulting in inactivation of GSK3β by phosphorylation on Ser9. Inactivation of GSK3β boosts neuronal differentiation and dendritic branching. In short, the results suggest that the synergistic effects of miR-9 and miR-124 control AKT/GSK3β signaling to regulate neuronal differentiation and dendritic complexity by inhibiting Rap2a.
The results of this study reveal a previously unknown interaction between miR-9, miR-124 and Rap2a, and emphasize the synergistic effects of miR-9 and miR-124 on neuronal differentiation and dendritic complexity.
Luciferase assay. HEK293 cells were seeded in 24-well plates and transfected the next day with 0.4 μ g of miRNA expression vector, 0.4 μ g of firefly luciferase reporter vector, and 0.08 μ g of the control vector pRL-TK (Promega, Madison, USA), which contains Renilla luciferase. Transfections were performed using Lipofectamine 2000 (Invitrogen). Each treatment was performed in triplicate in three independent experiments, and the activities of firefly and Renilla luciferase were measured consecutively using dual-luciferase assays (Promega) 24 h after transfection.
Cell transfection and transduction. HEK293 cells and C17.2 cells were seeded in 24-well plates and transfected the next day with miRNA expression vectors with or without miRNA sponges, Transfections were performed using Lipofectamine 2000. The cells were then incubated for 48 h.
For virus transduction, NSCs were digested into single-cell suspensions, and then seeded in poly-L-lysine-coated 24-well plates at 1 × 10 5 cells/cm 2 . The next day, low (5 μ L, titer: 1 × 10 8 TU/mL) or high amounts (10 μ L, titer: 1 × 10 8 TU/mL) of viral supernatants were added to the cells. The medium containing virus was removed and discarded 24 h after transduction and replaced with fresh growth medium of NSCs. Neurons derived from cortex of E14-E16 C57BL/6 mice were plates at 1 × 10 5 cells/cm 2 and cultured for 3 days. On the fourth day, low (5 μ L, titer: 1 × 10 8 TU/mL) or high amounts (10 μ L, titer: 1 × 10 8 TU/mL) of viral supernatant were added to the cells. The medium containing virus was removed and discarded 24 h after transduction and replaced with fresh growth medium of neurons. The cells were incubated for 3 or 7 days, and then harvested or immunostained.
Immunocytochemistry. Cells were fixed in 4% paraformaldehyde for 30 min, and then blocked for 1 h with 1% bovine serum albumin containing 0.3% Triton X-100. Blocked cells were incubated overnight at 4 °C with Rabbit polyclonal antibody to MAP2 (Millipore) and Rabbit polyclonal antibody to NeuN antibody (Millipore), and then for 2 h at room temperature with the relative secondary antibodies (DyLight 488-conjugated AffiniPure Donkey anti-rabbit IgG, Jackson ImmunoResearch Laboratories, West Rove, PA, USA). Images were acquired using an IX71 inverted microscope (Olympus, Japan).
Proteins were separated on 10% or 15% (for Rap2a) SDS-PAGE gels at a constant 100 mV voltage and transferred to Polyvinylidene Difluoride (PVDF) membranes at 300 mV for 1 h. PVDF membranes were blocked in 5% nonfat milk for 1 h; incubated overnight at 4 °C with primary antibodies against Rap2a (Proteintech, Wuhan, China), nestin (Sigma-Aldrich, St. Louis, MO, USA), Tuj1 (Sigma-Aldrich), p-AKT (Ser473) (Cell Signaling Technology, Boston, MA), p-GSK3β (Ser9) (Cell Signaling Technology), or β -actin (Sigma-Aldrich); and then incubated for 2 h at room temperature with the relative secondary antibodies conjugated with horseradish peroxidase (Abcam). Immunoreactive bands were visualized using an enhanced chemiluminescence kit on a Bio-Rad Image Lab system. Scientific RepoRts | 6:26781 | DOI: 10.1038/srep26781 Statistical analysis. All statistical analyses of experimental data were performed using GraphPad Prism 5.0 (GraphPad) and are presented as group mean ± SEM. All experiments were repeated at least three times. Comparison of the two groups was performed using independent two-tailed Student's t tests, and P values < 0.05 were considered significant.