|Home | About | Journals | Submit | Contact Us | Français|
How transmitter receptors modulate neuronal signaling by regulating voltage-gated ion channel expression remains an open question. Here we report dendritic localization of mRNA of Kv4.2 voltage-gated potassium channel, which regulates synaptic plasticity, and its local translational regulation by fragile X mental retardation protein (FMRP) linked to fragile X syndrome (FXS), the most common heritable mental retardation. FMRP suppression of Kv4.2 is revealed by elevation of Kv4.2 in neurons from fmr1 knockout (KO) mice, and in neurons expressing Kv4.2-3′UTR that binds FMRP. Moreover, treating hippocampal slices from fmr1 KO mice with Kv4 channel blocker restores long-term potentiation (LTP) induced by moderate stimuli. Surprisingly, recovery of Kv4.2 after N-methyl-D-aspartate receptor (NMDAR)-induced degradation also requires FMRP, likely due to NMDAR-induced FMRP dephosphorylation, which turns off FMRP suppression of Kv4.2. Our study of FMRP regulation of Kv4.2 reveals a novel aspect of NMDAR signaling and a new FMRP target of potential relevance to FXS.
Mammalian central neurons rely on the dynamic interplay between transmitter receptors and voltage-gated ion channels on their dendrites for signal processing. For example, the A-type voltage-gated K+ channels (IA) on the dendrites of CA1 hippocampal pyramidal neurons regulate neuronal signaling by filtering fast synaptic potentials and regulating action potential back propagation, synaptic integration and long-term potentiation (LTP) (Kim and Hoffman, 2008). This IA derives primarily from Kv4.2 (Birnbaum et al., 2004; Cee et al., 2007; Chen et al., 2006). Enriched on the spines of CA1 pyramidal neurons, Kv4.2 is under the regulation of synaptic activity and it in turn contributes to the regulation of synaptic plasticity (Cee et al., 2007; Jung et al., 2008). Whether Kv4.2 mRNA is targeted to dendrites to present the opportunity of local regulation by synaptic activity is unknown.
How Kv4.2 regulation may help neurons to stay within the dynamic range of synaptic plasticity is another open question. Whereas the rapid down regulation of Kv4.2 upon N-methyl-D-aspartate receptor (NMDAR) activation due to its internalization and degradation (Cee et al., 2007; Lei et al., 2008; Liao et al., 2010) provides positive feedback to enhance excitation, the dendritic potassium channel level has to quickly recover after a barrage of synaptic activities, given that loss of Kv4.2 function causes enhanced induction of LTP (Chen et al., 2006) while increasing Kv4.2 expression abolishes the ability to induce LTP (Jung et al., 2008). Since alteration of Kv4.2 levels is associated with epilepsy and possibly Alzheimer's disease (Birnbaum et al., 2004) and the KCND2 gene coding for Kv4.2 is near rearrangement breakpoints in autism patients (Scherer et al., 2003), better understanding of the dynamic regulation of Kv4.2 by synaptic activities will help future analyses of the contribution of this potassium channel to neuronal signaling as well as its involvement in neurological and mental disorders.
The importance of local synthesis of dendritic proteins in synaptic plasticity (Kelleher III et al., 2004; Sutton and Schuman, 2005) has stimulated recent studies on trafficking of neuronal RNA granules (Kiebler and Bassell, 2006), regulation of local synthesis of synaptic proteins (Schuman et al., 2006; Sutton and Schuman, 2005) and mRNA transport (Sossin and DesGroseillers, 2006). One of the RNA binding proteins implicated is the fragile X mental retardation protein (FMRP) linked to Fragile X syndrome (FXS), the most common heritable mental retardation which is often associated with autism (Bagni and Greenough, 2005). Multiple symptoms of FXS patients including the altered spine morphology (Greenough et al., 2001; Hinton et al., 1991; Irwin et al., 2001) is recapitulated in fmr1 knockout (KO) mice (Comery et al., 1997; Nimchinsky et al., 2001), which also display compromised learning, abnormal behavior and altered synaptic plasticity (Penagarikano et al., 2007). This mouse model of FXS is therefore a suitable system for examining FMRP contribution to synaptic regulation of local translation.
FMRP can bind to its target mRNA directly or indirectly (Bagni and Greenough, 2005). It has multiple RNA-binding domains and may regulate mRNA localization (Dictenberg et al., 2008), mRNA stability (Zalfa et al., 2007) or mRNA translation (Muddashetty et al., 2007; Zalfa et al., 2003) in central neurons (Bassell and Warren, 2008). Since FMRP is localized to dendrites and spines (Antar et al., 2004), it could regulate local protein synthesis underlying spine development and synaptic plasticity (Antar and Bassell, 2003; Bagni and Greenough, 2005). The mGluR (metabotropic glutamate receptor) theory of FXS posits that FMRP normally acts downstream of Group1 mGluR as a repressor of translation; in the absence of FMRP, there is runaway protein synthesis that leads to exaggerated long-term depression (LTD) (Bear et al., 2004). Notably, LTP induced by moderate theta burst stimulation (TBS) is also impaired in the CA1 hippocampal field of fmr1 KO mice (Lauterborn et al., 2007), raising the question as to how FMRP might affect synaptic plasticity.
For FMRP to be involved in dynamic regulation of dendritic mRNA, its grip on translational repression should be under synaptic regulation as well. How can FMRP repression be relieved? Phosphorylation of FMRP is a candidate mechanism (Ceman et al., 2003; Narayanan et al., 2007; Narayanan et al., 2008). Given that expression of FMRP with intact phosphorylation site causes an increase of stalled polyribosomes, dephosphorylation of FMRP may be associated with the release of polyribosomes from the stalled state (Ceman et al., 2003). Consistent with this scenario, Group1 mGluR activation causes transient dephosphorylation of FMRP (Narayanan et al., 2007; Narayanan et al., 2008), which may allow rapid mRNA regulation by FMRP. Whereas the involvement of mGluR in FMRP regulation has been investigated extensively, potential roles of other transmitter receptors in FMRP regulation still await analyses (Bassell and Warren, 2008), notwithstanding recent studies implicating NMDAR in FMRP regulation (Edbauer et al., 2010; Gabel et al., 2004b; Pilpel et al., 2009).
In this study, we show dendritic localization of Kv4.2 mRNA and FMRP suppression of local translation of the Dendra-Kv4.2 fusion protein in isolated dendrites. We demonstrate FMRP binding to the 3′UTR of Kv4.2 mRNA (Kv4.2-3′UTR) and identify domains of both FMRP and Kv4.2-3′UTR involved in their interaction, which likely contributes to FMRP suppression of Kv4.2 since expression of Kv4.2-3′UTR increases Kv4.2 level in neurons. Indeed, our analyses of fmr1 KO mice reveal that FMRP is important for Kv4.2 suppression in vivo. We further show that the deficit in LTP induction can be rescued by reducing Kv4 channel activity in hippocampal slices from fmr1 KO mice.
Unexpectedly, we found that NMDAR activation not only transiently reduces Kv4.2 protein level due to degradation as reported previously, but also increases Kv4.2 protein production in an FMRP-dependent process likely involving Kv4.2-3′UTR. We discovered that NMDAR activation causes dephosphorylation of mTOR and the downstream S6 kinase S6K1 as well as the ribosomal protein S6 and FMRP – both substrates of S6K1. We then expressed mutant FMRP in cultured hippocampal neurons from fmr1 KO mice lacking FMRP, and found that S499D but not S499A mutant FMRP retains the ability to dampen the exuberant Kv4.2 surface expression on the dendrites of hippocampal neurons from fmr1 KO mice, as well as the ability to suppress translation associated with Kv4.2-3′UTR. Taken together, our study identifies Kv4.2 mRNA as a new target of FMRP; whereas FMRP suppresses Kv4.2 in basal conditions, FMRP suppression is relieved by dephosphorylation upon NMDAR activation to increase Kv4.2 expression following synaptic activity, thereby maintaining neurons within the dynamic range of synaptic plasticity.
Given the involvement of dendritic Kv4.2 potassium channels in the regulation of synaptic plasticity, it is important to determine whether Kv4.2 mRNA resides in the dendrites. We could detect endogenous Kv4.2 mRNA in hippocampal neurons after 14 days in vitro (DIV14) by fluorescence in situ hybridization (FISH) using an anti-sense probe against the 3′UTR region of Kv4.2 mRNA; the control sense probe did not show any specific mRNA staining (Figure S1). We found Kv4.2 mRNA in cell bodies and along dendrites in a punctate distribution (Figure 1B). Moreover, Kv4.2 mRNA co-localized with dendritic marker MAP2 but not with axonal marker tau1 (Figure 1C). Moreover, we found dendritic localization of Kv4.2 mRNA in the CA1 dendritic field of the hippocampus (Figure 1A).
To test whether the 3′UTR of Kv4.2 mRNA, which is relatively long (2.5 Kb) with its first 1.5 Kb of sequences highly conserved between human and rodents (Figure S2), can mediate dendritic targeting, we used the MS2 system for tracking the subcellular localization of RNAs (Fusco et al., 2003). We fused Kv4.2-S.3′UTR (sense) or Kv4.3-A.S.3′UTR (anti-sense) to MS2BS(6×) containing 6 tandem RNA hairpins that are binding sites for the RNA binding protein MS2 (Figure 2A). Co-transfection of DIV10-12 hippocampal neurons with MS2BS(6×)-Kv4.2-S.3′UTR and MS2-GFP-NLS (nuclear localization signal) made it possible to delineate the distribution of chimeric RNA containing MS2 binding sites (MS2BS) and Kv4.2-3′UTR via its association with GFP- tagged MS2. The control RNA containing only MS2 binding sites but no 3′UTR appeared exclusively inside the nucleus in all of the transfected neurons (Figure 2B) due to the presence of NLS in the GFP-tagged MS2 fusion protein. The control MS2BS(6×)-Kv4.2-A.S.3′UTR also yielded similar nuclear localization (Figure 2B). In contrast, MS2BS(6×)-Kv4.2-S.3′UTR appeared in the cytoplasm and also entered the dendrites, giving rise to a punctate pattern (Figure 2B), similar to the appearance of the MS2BS(6×)-Arc-S.3′UTR (Figure S3). Thus, the 3′UTR of Kv4.2 mRNA is sufficient for dendritic targeting.
Since FMRP is known to be present in multiple RNA-containing granules (Sossin and DesGroseillers, 2006), we examined neurons double-labeled for FMRP and Kv4.2 mRNA and found partial overlap of Kv4.2 mRNA and FMRP granules in proximal and distal dendrites (Figure 3B), similar to the overlapping distribution of FMRP and Arc mRNA, a target of FMRP (Figure S4A). We further show that Kv4.2 mRNA co-immunoprecipitated with FMRP from the adult mouse brain lysate (Figure 3A), similar to the co-immunoprecipitation of FMRP with PSD-95 mRNA, another target of FMRP. Finally, we monitored concerted movements of FMRP and Kv4.2-3′UTR by live imaging of neurons expressing MS2-GFP-NLS and MS2BS(6×)-Kv4.2-S.3′UTR together with fluorescently tagged FMRP following NMDAR activation, which enhanced the movement of these granules (Figure 3C). Taken together, these findings indicate that FMRP is associated with Kv4.2 mRNA in neuronal dendrites.
We then tested for binding of FMRP to the 3′UTR of Kv4.2 mRNA, because in silico analysis of this region has revealed the presence of U-rich stretches (Figure S2), a sequence motif for RNA binding to FMRP (Chen et al., 2003). By using streptavidin-beads to pull down proteins from brain lysates bound to biotinylated Kv4.2-3′UTR, we found FMRP binding of Kv4.2-S.3′UTR (Figure 3D) at a level comparable to that of Arc-3′UTR or PSD-95-3′UTR (Figure S4B and Figure S5A). This binding is specific to FMRP since Kv4.2-3′UTR showed no association with the RNA-binding protein staufen or non-RNA binding proteins such as mTOR, dynamin 1 and actin (Figure S5A). Furthermore, the binding is direct as evident from the interaction between bacterially expressed and purified FMRP and Kv4.2-S.3′UTR, at a level comparable to the interaction between FMRP and PSD-95-3′UTR (Figure 3E and Figure S5B). This binding is specific since FMRP binds Kv4.2-3′UTR but not GFP mRNA or Kv4.2-A.S.3′UTR (Figure 3D and Figure 3E).
Next, we examined the three domains of FMRP individually. Only the C-terminal domain of FMRP was specifically pulled-down with Kv4.2-3′UTR (Figure 3F). This domain contains an RGG box known to have an affinity for mRNAs. We then tested five RNA fragments that together encompass the entire 3′UTR of the mouse Kv4.2 mRNA, and found only fragment 2 and fragment 5 that contain U-rich sequences associated with the purified FMRP C-terminal domain (Figure 3G). Notably, fragment 2 includes an evolutionarily conserved U-rich sequence (Figure S2). Taken together, these studies show that the direct interaction between FMRP and Kv4.2-3′UTR is likely evolutionarily conserved.
We found the Kv4.2 mRNA level in the hippocampus of fmr1 KO mice was similar to that in wild-type (WT) mice (Figure 4A). We confirmed the gene targeting using primers that amplify exon 5 (or exon 1) of the fmr1 (or Kv4.2) gene that is interrupted by the neomycin resistance selection marker gene in the fmr1 (or Kv4.2) KO mice (Figure 4A); using other primers we found that these KO mice have some remnant, genetically altered, transcripts. Using the MS2 system to track the subcellular localization of MS2BS(6×)-Kv4.2-S.3′UTR in hippocampal neurons with or without FMRP, we found similar dendritic targeting (Figure 4B), indicating that FMRP is not required for dendritic targeting of Kv4.2-3′UTR. We further performed quantitative FISH analysis using an anti-sense probe against Kv4.2-3′UTR and examined the distribution of endogenous Kv4.2 mRNA in dendrites. We found similar dendritic localization and punctate pattern of Kv4.2 mRNA in WT and fmr1 KO neurons (Figure 4C), and similar Kv4.2 mRNA levels along dendrites of neurons with or without FMRP (Figure 4C). Taken together, these results suggest that FMRP is not essential for Kv4.2 mRNA dendritic targeting or stability in basal conditions.
To test for FMRP regulation of Kv4.2 protein expression we compared Kv4.2 levels in the hippocampus from adult WT and fmr1 mutant mice, using Kv4.2 KO mice as control and Kv4.2 antibody specificity (Figure 5A and and5B).5B). We found ~1.5-2-fold increase of Kv4.2 immunoreactivity in the CA1 dendritic field of the hippocampus from 3 week-old and 2 month-old fmr1 KO mice (Figure 5A), and a similar increase of Kv4.2 protein levels in the hippocampus from adult fmr1 KO mice (Figure 5B).
Next, we performed surface biotinylation on cultured hippocampal neurons and used actin both as loading control and to confirm that our biotinylation protocol results in biotinylation of surface but not cytosolic proteins. We found higher surface as well as total Kv4.2 levels in DIV14 hippocampal neurons without FMRP (Figure 5C). Using antibody against an extracellular epitope of Kv4.2 for immunostaining of unpermeabilized DIV14 hippocampal neurons, we found higher levels of Kv4.2 on the dendritic surface of neurons without FMRP (Figure 5D). Since both total Kv4.2 protein levels and the surface expression of Kv4.2 on dendrites are elevated in the absence of FMRP while the relative proportions of surface and total Kv4.2 protein levels were not significantly altered, these findings indicate that FMRP suppresses Kv4.2 production.
To test whether FMRP binding to Kv4.2-3′UTR could suppress protein production, we performed an in vitro translation assay using Renilla luciferase transcript fused to Kv4.2-3′UTR together with firefly luciferase transcript for normalization, and included either purified GST as control or purified GST-mouse full-length FMRP, and nuclease-treated rabbit reticulocyte lysates. We found that FMRP suppressed Kv4.2-3′UTR-dependent translation by 60% (P<0.001, n=4) (Figure 5F). Moreover, expression of MS2BS(6×)-Kv4.2-S.3′UTR but not MS2BS(6×) alone led to an increase of surface Kv4.2 expression (Figure 5E), indicating that FMRP suppression of Kv4.2 is relieved by disruption of FMRP interaction with the 3′UTR of Kv4.2 mRNA via MS2BS(6×)-Kv4.2-S.3′UTR. Taken together, these findings support the notion that FMRP suppression of Kv4.2 protein expression in neuronal dendrites is due to translational repression via its association with Kv4.2-3′UTR.
To look for evidence of Kv4.2 local translation in the dendrites of cultured hippocampal neurons, we expressed Dendra-Kv4.2 in these neurons, severed a dendrite via UV illumination from the 2-photon microscope, photo-converted the existing Dendra-Kv4.2 from green to red and then monitored the appearance of newly synthesized green Dendra-Kv4.2 in the dendrite separated from the neuronal soma (Figure 5G). We found a nearly 2-fold increase of Dendra-Kv4.2 local translation in isolated dendrites from hippocampal neurons without FMRP (Figure 5G), indicating that Kv4.2 local translation is likely under the control of FMRP.
The hippocampus-dependent learning deficits of fmr1 KO mice are associated with an inability of moderate levels of theta burst stimulation to induce LTP as evident from field recording of the excitatory postsynaptic potential (fEPSP) (Lauterborn et al., 2007). As reported, five theta bursts of Schaffer collaterals stimulation induced LTP in hippocampal slices from WT mice (Figure 6A) but not fmr1 KO mice (Figure 6B) of postnatal day 14-21, whereas ten theta bursts were sufficient to induce LTP in both. Using the heteropodatoxin HpTx2 from a family of spider toxins specific for blocking Kv4 channels (Ramakers and Storm, 2002; Sanguinetti et al., 1997), we found HpTx2 dose-dependently restored LTP induction by five theta burst stimuli to slices from fmr1 KO mice but did not significantly alter LTP of control WT slices (Figure 6 C and D). It thus appears that hippocampal neurons from fmr1 KO mice have excessive Kv4 channel activity due to the lack of FMRP suppression, thereby compromising synaptic plasticity.
We next test whether NMDAR regulates Kv4.2 protein levels in DIV14-21 hippocampal neurons, and found that 5 min treatment with 100 μM NMDA induced first a robust decrease of Kv4.2 levels, which then quickly recovered 15 min after washing out NMDA (total time elapsed from the start of the NMDA treatment is 20 min) (Figure 7A). The NMDA-induced reduction of total Kv4.2 levels is attributed to degradation (Lei et al., 2008) and is dependent on calpain activity (Liao et al., 2010). We therefore pre-treated neurons with a mixture of calpain inhibitors (MDL+ALLN) for 15 min before applying NMDA for 5 min, and waited for another 15 min after washing out of NMDA to monitor NMDAR-mediated Kv4.2 regulation without the confounding effects of Kv4.2 degradation. In the presence of calpain inhibitors, NMDA treatment no longer caused a reduction of Kv4.2 levels, instead the Kv4.2 protein levels progressively increased by ~2-2.5-fold (P<0.01, n=4) following NMDAR activation (Figure 7B and Figure S6). These experiments reveal that NMDAR activation causes up regulation of Kv4.2 production concurrent with Kv4.2 degradation, to fine tune Kv4.2 levels following NMDAR activation and allow their restoration in due course.
We also performed a dual-luciferase reporter assay to look into the effect of NMDAR activation on translation associated with Kv4.2-3′UTR, and found that Kv4.2-3′UTR-dependent translation of luciferase increased shortly after NMDA treatment (Figure 7C), reaching a plateau 30 min after NMDA treatment before finally decreasing several hours later. Taken together, our results show that NMDAR activation induces up-regulation of Kv4.2 expression in hippocampal neurons likely in a process mediated by Kv4.2-3′UTR.
Having found that NMDAR activation causes up regulation of Kv4.2 expression, we next asked whether this regulation involves FMRP. Western analyses revealed that WT neurons showed robust recovery of Kv4.2 within 15 min, after the NMDAR-induced degradation caused a ~2-fold reduction (P<0.01, n=3) of Kv4.2 protein level. In contrast, the Kv4.2 levels of neurons from fmr1 KO mice remained reduced after NMDAR activation and showed no recovery (Figure 7D).
Next, we asked whether FMRP is required for NMDAR-induced up regulation of translation that is dependent on Kv4.2-3′UTR. Using the dual-luciferase reporter assay, we found that Kv4.2-3′UTR-dependent production of luciferase increased in response to NMDAR activation in WT neurons but not in neurons from fmr1 KO mice (Figure 7E). Given that in fmr1 KO mice there is excess basal Kv4.2 expression due to a lack of FMRP suppression of Kv4.2, the requirement of FMRP for NMDAR-induced up regulation of Kv4.2 production as well as Kv4.2-3′UTR-dependent translation raises the question whether this synaptic regulation could be due to a relief of FMRP suppression of Kv4.2.
How might FMRP suppression of Kv4.2 be turned off? FMRP may repress translation of its target mRNA by stalling ribosomes, which could be diminished by synaptic activity and dephosphorylation of FMRP (Ceman et al., 2003; Narayanan et al., 2007; Narayanan et al., 2008). To test whether NMDAR activation might turn off FMRP repression of Kv4.2, we examined FMRP phosphorylation at Serine 499 preceding the RGG box, a post-translational modification known to take place within 2-4 hr of FMRP synthesis (Ceman et al., 2003). Remarkably, we found rapid dephosphorylation of FMRP within 5 min exposure of DIV14-21 hippocampal neurons to NMDA (Figure 8A), accompanied with rapid dephosphorylation of mTOR, S6 kinase (S6K1) and S6 (Figure 8A) while the total protein levels of these proteins were unchanged. Given that the ribosomal S6 kinase S6K1 is the primary kinase for FMRP phosphorylation at S499 (Narayanan et al., 2008), FMRP dephosphorylation is likely a consequence of the inhibition of mTOR pathway shortly after NMDAR activation. As expected, treatment with the mTOR inhibitor rapamycin also resulted in FMRP dephosphhorylation (Figure S7).
We then tested the effects of phosphatase inhibitors. We treated neurons with 20 nM okadaic acid or 50 nM fostriecin to inhibit PP2A, 1 μM okadaic acid to inhibit PP1 and PP2A, or 10 μM cyclosporin A to inhibit PP2B/calcineurin. Whereas dephosphorylation of FMRP and mTOR was unaffected by treatment with 20 nM okadaic acid or 50 nM fostriecin, which inhibit PP2A, or the PP2B inhibitor cyclosporine A at 10 μM as compared with the DMSO carrier control, 1 μM okadaic acid greatly reduced dephosphorylation of both mTOR and FMRP following NMDAR activation (Figure 8B and Figure S8). These experiments reveal that PP1 is required for NMDAR-induced dephosphorylation of mTOR and its downstream effectors including S6 and FMRP. Thus, the signaling pathway for NMDAR-induced up regulation of Kv4.2 likely involves PP1 activation by NMDAR (Chung et al., 2009), leading to dephosphorylation of mTOR. This inhibition of the mTOR pathway then results in dephosphorylation of substrates of S6K1 downstream of mTOR, such as FMRP.
To test whether regulation of FMRP phosphorylation at S499 might account for the regulation of Kv4.2-3′UTR-dependent translation, we compared the wild-type (WT) form of FMRP with mutant FMRP with S499 replaced by Alanine (S499A) or Aspartate (S499D) (Ceman et al., 2003), and performed a dual-luciferase reporter assay by co-transfecting HEK293 cells with Renilla luciferase-Kv4.2-3′UTR together with firefly luciferase, plus GFP-tagged FMRP (WT, S499A, S499D), or GFP alone as control. In contrast to the suppression of Kv4.2-3′UTR-dependent luciferase production by FMRP-WT and FMRP-S499D (Figure 8C), FMRP-S499A showed much less suppression (Figure 8C). Thus, FMRP phosphorylation at S499 appears to be crucial for FMRP suppression of translation associated with Kv4.2-3′UTR.
To test whether regulation of FMRP phosphorylation affects Kv4.2 channel density on neuronal dendrites, we transfected cultured hippocampal neurons from fmr1 KO mice with GFP-tagged FMRP (WT, S499A, S499D), or GFP alone as control, and used antibody against extracellular epitope of Kv4.2 to assess its surface expression level. In control experiments involving transfecting hippocampal neurons from WT or fmr1 KO mice with GFP, we found higher levels of surface expression of Kv4.2 on the dendrites of neurons from fmr1 KO mice (Figure 8D). By introducing wild-type or mutant FMRP into hippocampal neurons from fmr1 KO mice, we found that neurons expressing FMRP-WT or FMRP-S499D had similar levels of Kv4.2 surface expression whereas neurons expressing FMRP-S499A had significantly increased Kv4.2 protein levels on the surface of their dendrites (Figure 8D), indicating that the S499D but not S499A mutant form of FMRP retains the ability to suppress Kv4.2. Taken together, our results suggest Kv4.2-3′UTR-dependent protein synthesis as well as Kv4.2 channel density on neuronal dendrites depends on the status of FMRP phosphorylation.
This study provides evidence for dendritic targeting of mRNA of the Kv4.2 dendritic voltage-gated potassium channel that is important for controlling dendritic excitability and synaptic plasticity. FMRP suppresses Kv4.2 expression in basal conditions, and is also involved in NMDAR-mediated Kv4.2 up regulation due to its dephosphorylation. Our study thus implicates a new signaling pathway linking FMRP with dendritic Kv4.2 regulation by synaptic activity, and provides a new lead for consideration regarding the etiology of FXS.
In addition to the interaction with FMRP for translation suppression, Kv4.2-3′UTR also mediates dendritic targeting and increases steady state levels of mRNA. Taken together with the previous report on local translation of Kv1.1 mRNA in dendrites (Raab-Graham et al., 2006), voltage-gated ion channels now join the rank of postsynaptic scaffolding proteins such as PSD-95 and SAPAPs, activity-dependent synaptic proteins such as CaMKIIα, Arc and MAP 1b, and ligand-activated ion channels such as GluR1/2 and GABAARδ (Bassell and Warren, 2008) as dendritic proteins with their mRNAs localized in neuronal dendrites and under the regulation of synaptic activity.
FXS, the most common heritable mental retardation often associated with autism, is caused by the loss of FMRP function (Bagni and Greenough, 2005). Our finding of Kv4.2 mRNA association with FMRP in neuronal dendrites and direct binding of FMRP to Kv4.2-3′UTR led us to discover that Kv4.2 is under the control of FMRP. Whereas loss of FMRP resulted in no significant changes in Kv4.2 mRNA level or dendritic localization, it caused a dramatic increase of total Kv4.2 levels in the CA1 dendritic field of the hippocampus and in cultured hippocampal neurons from fmr1 KO mice. Similar elevation of Kv4.2 levels was also found for surface expression of Kv4.2, especially on distal dendrites, revealing FMRP suppression of Kv4.2 in vivo.
Whereas we found elevated Kv4.2 in the hippocampal dendritic field of 3 week-old as well as 2 month-old fmr1 KO mice (Figure 5A), a recent study reports Kv4.2 levels are reduced in fmr1 KO mice (Gross et al., 2011), however, this conclusion is based on Kv4.2 immunostaining that shows a different pattern from the documented Kv4.2 expression in stratum radiatum but low in stratum lacunosom moleculare (Menegola and Trimmer, 2006) thus raising question about the specificity of the immunostaining. This study also reports that GFP-Kv4.2 3′ UTR is associated with mCherry-FMRP but not mCherry-RGG or other FMRP fragments that contain one or both RNA-binding domains (Gross et al., 2011). In contrast, we found direct binding of Kv4.2 3′UTR to FMRP as well as its RGG-containing C-terminal domain (Figure 3E--GG).
Not only is FMRP required for suppression of dendritic Kv4.2, it is also essential for NMDAR-induced Kv4.2 protein production that enables Kv4.2 level to fully recover after its degradation and down regulation induced by NMDAR activation. FMRP thus plays a crucial role in tuning the dendritic Kv4.2 channel density and permitting dynamic regulation of Kv4.2 during synaptic activities. We found the elevated Kv4.2 level in fmr1 KO mice contributes to the LTP deficits (Lauterborn et al., 2007), because the Kv4 channel blocker HpTx2 dose-dependently restored LTP induction by five theta bursts (Figure 6). Given that hippocampal CA1 neurons lacking FMRP can exhibit LTP in response to strong stimuli (ten theta bursts) (Lauterborn et al., 2007), Kv4.2 suppression by FMRP appears to be important for maintaining these neurons within the dynamic range for synaptic plasticity. Moreover, concurrent with NMDAR-induced Kv4.2 internalization and degradation (Kim et al., 2007; Lei et al., 2010), NMDAR causes de- repression of Kv4.2 production by inducing FMRP dephosphorylation to restore the Kv4.2 level within 20 min (Figure 7), so as to terminate the positive feedback regulation mediated by Kv4.2 down regulation.
Whereas chemical LTP causes Kv4.2 internalization and redistribution (Cee et al., 2007) and NMDAR activation causes significant reduction of Kv4.2 channels in a reversible manner (Liao et al., 2010), our finding of elevated Kv4.2 levels due to NMDA treatment in the presence of calpain inhibitors, taken together with the luciferase assay showing NMDAR-induced up regulation of translation associated with Kv4.2-3′UTR, strongly suggests that NMDAR activation causes increased production of Kv4.2. Because new protein synthesis is clearly required for long-lasting activity-dependent changes in synaptic transmission, the manner by which neuronal activity engages the translational machinery is key to our understanding of long-term information storage. In addition to the rapid and bidirectional remodeling of synaptic NMDAR subunit composition by A-type K+ channel activity (Jung et al., 2008), the activity-dependent regulation of Kv4.2 expression uncovered in our study provides a mechanism for rapid recovery of Kv4.2 after NMDAR-induced degradation. Whereas immediate down regulation of Kv4.2 upon NMDAR activation corresponds to positive feedback regulation important for synaptic plasticity, NMDAR-induced up regulation of Kv4.2 provides a means for negative feedback regulation for homeostasis.
Both metabotropic and ionotropic glutamate receptors are known to regulate local protein translation. With a requirement of local protein synthesis for mGluR-dependent LTP and LTD, mGluR activation rapidly increases dendritic local protein synthesis (Sutton and Schuman, 2005). As to NMDAR-mediated translational regulation, NMDA treatment initially causes repression of overall protein synthesis (within 5 min), followed with preferential translation of specific targets such as CaMKIIα (Scheetz et al., 2000). In this study, we show that NMDAR signaling affects translation associated with Kv4.2-3′UTR and causes up regulation of Kv4.2 in an FMRP-dependent manner.
Several studies have linked FMRP to NMDAR signaling, including dynamic dendritic FMRP localization in response to visual experience (Gabel et al., 2004a), accumulation of the mRNA encoding Arc/Arg3.1, a target of FMRP, in regions of activated synapses (Steward and Worley, 2001), and NMDA-induced total protein synthesis in synaptosomes (Muddashetty et al., 2007). We found that Kv4.2 up regulation by NMDAR is due to NMDAR-induced dephosphorylation of FMRP for de-repression of Kv4.2. It remains to be determined whether other transcripts besides Kv4.2 mRNA are regulated by NMDAR via the same signaling pathway.
Dephosphorylation of FMRP may lead to the release of polysomes from the stalled state (Ceman et al., 2003). Activation of Group1-mGluR causes rapid FMRP dephosphorylation through PP2A activation, associated with a burst of translation of FMRP targets in hippocampal neurons (Narayanan et al., 2007). Here, we show that NMDAR activation leads to rapid dephosphorylation of FMRP in a process dependent on PP1 but not PP2B, consistent with previous findings of NMDAR activation of PP1 in hippocampal neurons (Chung et al., 2009). We further asked whether NMDAR-induced up regulation of Kv4.2 might involve FMRP dephosphorylation, by testing FMRP mutants (S499A or S499D). The S499A mutation abolishes the ability of FMRP to suppress Kv4.2-3′UTR-dependent translation in luciferase assay as well as surface Kv4.2 levels, while the S499D mutation preserves the functions of FMRP (Figure 8). Our study thus provides evidence for a role of the FMRP phosphorylation status on FMRP regulation of its target mRNA.
Several reports link alterations in potassium channel expression with neurological and mental disorders. Alteration of Kv4.2 levels may be related with epilepsy and perhaps also Alzheimer's disease (Birnbaum et al., 2004). The Kv4 channel beta subunits DPP6 and DPP10 are implicated in autism susceptibility (Marshall et al., 2008) and the KCND2 gene coding for Kv4.2 is near rearrangement breakpoints of unrelated autism patients (Scherer et al., 2003). FMRP is crucial for maintaining Kv3.1b tonotopicity and its up regulation by acoustic stimulation (Strumbos et al., 2010), and mutations in KCNC3 are responsible for spinocerebellar ataxia (SCA) in two families (Waters et al., 2006). FMRP may also control gating of the Na+-activated K+ channel Slack by protein-protein interaction (Brown et al., 2010). Our study showing dysregulation of Kv4.2 on hippocampal neuronal dendrites and inability of NMDAR to up regulate Kv4.2 production in fmr1 KO mice indicates that an imbalance in the spatial and temporal regulation of Kv4.2 likely affects synaptic plasticity, and may contribute to impairments of neuronal signaling in FXS.
C57BL6/J, FVB.129P2-Pde6b+ Tyrc−ch/AntJ (control mice for fmr1 KO), FVB.129P2-Fmr1tm1Cgr/J (fmr1 KO) were from the Jackson Laboratory and Kv4.2 KO mice were kindly provided by Dr. Tom Schwarz and Dr. Jeanne M. Nerbonne. The use and care of animals in this study follows the guidelines of the UCSF Institutional Animal Care and Use Committee.
Hippocampal neurons isolated from embryonic day 17 mouse brains were plated at a density of 1-3 × 105 cells/well as described previously (Fu et al., 2007). HEK293 cells were maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 110 μg/ml sodium pyruvate and 2 mM L-glutamine. Cells were kept at 37°C in a humidified CO2-controlled (5%) incubator and were transfected using Lipofectamine™ 2000.
Hippocampal neurons grown on coverslips were immunostained with or without prior transfection. Cells were washed with phosphate-buffered saline (PBS), fixed in 4% formaldehyde, and incubated in blocking buffer (1% goat serum in PBS containing 0.1% Triton X-100), then incubated with primary antibody for 4 hrfollowed with secondary antibody for 1 hr at room temperature, and examined under a Leica TCS SP2 confocal microscope (Leica Microsystems, Bannockburn, IL).
Proteins denatured by heating for 15 min at 60°C in Laemmli sample buffer (Laemmli, 1970) were separated by SDS-PAGE and transferred to nitrocellulose membranes. After blocking in TTBS buffer (10 mM Tris-HCl, pH7.5, 150 mM NaCl, 0.05% Tween 20, and 5% skim milk powder), the membranes were incubated with primary antibody overnight at 4°C and then with horseradish peroxidase (HRP)-secondary antibody for 1 hr at room temperature. Detection was performed using an enhanced chemiluminescence (ECL) kit and hyperfilm MP, and quantified using a Fuji BAS-2000 image analyzer.
Hippocampal neurons (DIV14-21) were incubated with neurobasal medium for 5 min (control) or with neurobasal medium containing 100 μM NMDA for 5 min at 37°C, washed twice with neurobasal medium and then maintained in the medium for 15 min or time indicated in results.
Additional information on materials, antibodies, DNA constructs, protein purification, slice in Situ hybridization, FISH in cultured hippocampal neurons, tracking mRNA using MS2 system, live imaging using Dendra-Kv4.2, immunoprecipitation of FMRP in brain lysates, quantitative RT-PCR, biotinylated RNA and protein binding assay, immunohistochemistry, surface biotinylation, surface staining, in vitro translation assay, luciferase reporter assay and LTP experiment are described in Supplemental Experimental Procedures.
We thank Drs. Tom Schwarz and Jeanne M. Nerbonne for kindly providing Kv4.2 KO mice, Dr. Lynn Regan for kindly providing mouse full-length FMRP construct, Dr. Stephanie Ceman for kindly providing GFP-FMRP constructs, Dr. Marc I. Diamond for kindly providing pcDNA3-mouse H1d construct and Dr. Seung Key Jang for sharing MS2-GFP-NLS and MS2BS(6×) constructs. We also thank Dr. Sila Konur for help with imaging and for critical reading of the manuscript, and Dr. Desiree Thayer for critical reading of the manuscript and many helpful discussions. This work was supported by the National Institute of Mental Health (NIMH) to Lily Yeh Jan. Lily Yeh Jan and Yuh Nung Jan are investigators of the Howard Hughes Medical Institute (HHMI).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.