|Home | About | Journals | Submit | Contact Us | Français|
Fragile X Syndrome is caused by the loss of FMRP, which represses and reversibly regulates the translation of a subset of mRNAs in dendrites. Protein synthesis can be rapidly stimulated by mGluR-induced and PP2A-mediated dephosphorylation of FMRP, which is coupled to the dissociation of FMRP and target mRNAs from miRISC complexes. Here, we report the rapid ubiquitination and UPS mediated degradation of FMRP in dendrites upon DHPG stimulation in cultured rat neurons. Using inhibitors to PP2A and FMRP phosphomutants, degradation of FMRP was observed to depend on its prior dephosphorylation. Translational induction of an FMRP target, PSD-95 mRNA, required both PP2A and UPS. Thus, control of FMRP levels at the synapse by dephosphorylation-induced and UPS mediated degradation provides a mode to regulate protein synthesis.
Protein homeostasis is of universal importance in processes ranging from cell cycle to immune response to synaptic plasticity. Targeted degradation of crucial proteins by the ubiquitin proteasome system(UPS) can provide a basis for spatiotemporal regulation of various functions(Hershko and Ciechanover, 1998; Rechsteiner, 1987). UPS action can be rapid and regarded as a type of signaling event(Pierce et al., 2009; Rinetti and Schweizer, 2010). The role of UPS is particularly important in post-mitotic neurons having to continuously modify protein composition at synapses as evidenced by the number of ‘proteinopathies’ in the nervous system caused by susceptibility of neurons to UPS abnormalities(Ross and Poirier, 2004). Angelman syndrome results from mutations in a ubiquitin ligase and impaired ubiquitination of Arc necessary for AMPA receptor endocytosis(Greer et al., 2010). Inhibition of UPS leads to defects in LTP(Fonseca et al., 2006; Hegde et al., 1993; Karpova et al., 2006) as well as LTD(Colledge et al., 2003; Hou et al., 2006). The types of synaptic proteins regulated by ubiquitination include receptors and scaffolding proteins(Bingol and Schuman, 2004; Colledge et al., 2003; Ehlers, 2003).
The regulated degradation of synaptic proteins by UPS is coordinated with activity mediated synthesis of synaptic proteins(Steward and Schuman, 2003). Local protein synthesis provides a means to affect synaptic protein content and regulate plasticity(Wang et al., 2010). Proteins that regulate synaptic protein synthesis may also be degraded by UPS, providing additional control to regulate protein abundance at synapses. For example, NMDAR-induced degradation of MOV10, an argonaute protein, leads to disinhibition of translation by microRNAs(Banerjee et al., 2009).
Fragile X Syndrome, the most common form of inherited intellectual disability, is caused by loss of Fragile X Mental Retardation protein, FMRP, which often is a negative regulator of target mRNA translation important for synaptic function(Zalfa et al, Cell 2003; Muddashetty et al, 2007; Darnell et al Cell 2011). Activation of gp1 metabotropic glutamate receptors(mGluRs) normally leads to a rapid and protein synthesis-dependent LTD, which is exaggerated and protein synthesis-independent in Fmr1 KO, likely due to excess and dysregulated translation(Waung and Huber, 2009). Surprisingly, mGluR activation leads to both FMRP loss at synapses(Antar et al., 2004), and FMRP synthesis at synapses (Weiler et al., 1997). FMRP synthesis and UPS mediated degradation of FMRP appear to be coordinately regulated in hippocampal slices(Hou et al., 2006), but the mechanism and function of activity-induced UPS-mediated degradation of FMRP are unclear. mGluR-induced protein synthesis-dependent LTD and mGluR-induced epileptogenesis required the UPS in WT neurons which was altered in Fmr1 KO(Hou et al., 2006; Zhao et al., 2011). Since mGluR-induced translation is initiated by rapid and transient dephosphorylation of FMRP by protein phosphatase 2a(PP2A)(Narayanan et al., 2007), we sought to investigate a link with UPS-mediated degradation of FMRP. Here we show that mGluR-induced dephosphorylation of FMRP facilitates its ubiquitination and UPS-mediated degradation, which may provide a switch for rapid translation induction by mGluRs.
Neurons were cultured from E18 Sprague-Dawley rat embryos of either sex as described previously(Antar et al., 2004).
FLAG-GFP-FMRP constructs mutated at Serine499 and F-luciferase-PSD-95UTR have been described(Narayanan et al., 2007). FLAG3XmCherryFMRP constructs and PAGFP-FMRP constructs were generated at the Emory University Custom Cloning Core Facility. Neurons were transfected by Neuromag(OZ biosciences) and Neuro2a cells with Lipofectamine 2000(Invitrogen).
Hippocampal neurons(DIV14-21) were stimulated with DHPG (50 μM) with or without pretreatment of okadaic acid(10nM) or MG132(25 μM) for 20′. Neurons were fixed with 4% paraformaldehyde and processed for immunofluorescence to detect total and phospho-FMRP(Antar et al 2004; Narayanan et al 2007). Images were acquired on a Nikon-TE-ECLIPSE inverted microscope with a 60X DIC-oil objective. Multiple Z-slices were acquired with a cooled CCD camera(Cascade, Photometrics) and intensity in distal regions(>45 μm) of dendrites were quantitated.
Hippocampal neurons(9DIV) were co-transfected with PAGP-FMRP constructs and mCherry for morphologic identification of transfected cells. A distal region of the dendrite was selected for activation (405 laser, 6 pulses, 40% laser power, Nikon A1R confocal microscope). A brief pause (2′) was allowed for maturation of PAGFP. Subsequently, DHPG was added to the medium and images were immediately acquired (every 10 seconds for 8 minutes). The PAGFP particles were automatically tracked by IMARISTrack software(Bitplane) and intensities of the ‘spots’ at each point were background subtracted. An unstimulated region proximal to the cell body was also measured to quantify the amount of fluorescence lost due to transport. ‘Spots’ were classified as degrading if there was at least 20% decrease in final intensities relative to initial fluorescence intensity. Relative fluorescence was calculated by normalizing the intensity to initial intensity. The cumulative relative intensities were fitted to an exponential decay and the degradation rates were extracted.
Immunoprecipitation was performed as before (Muddashetty et al., 2007) on high density cortical neurons at DIV14 incubated with MG132 (25μM, 12 hrs) and treated with DHPG (1min) before wash and lysis with IP buffer, with or without pretreatment with okadaic acid(10nM). For ubiquitination assays, Neuro2a cells were co-transfected with GFP-FLAG or mCherry 3X FLAG fused FMRP constructs and HA-tagged ubiquitin (24 hours). After pre-incubation with MG132 (4 hrs), cells were lysed and used for immunoprecipitation followed by western blotting. Band intensities were quantified using ImageJ.
Relative luciferase activity of a reporter fused with 3′UTR of PSD-95 was measured by dual luciferase assay as described (Muddashetty et al., 2011). High density cortical neurons were were stimulated with DHPG for indicated times with or without okadaic acid and MG132 pretreatment.
While UPS activity has been implicated in mGluR-induced hippocampal LTD(Hou et al., 2006) and epileptogenesis(Zhao et al., 2011), and FMRP is degraded after mGluR stimulation, it is not known whether FMRP degradation is a necessary step in this cascade to stimulate synaptic protein synthesis. We asked whether mGluR-induced signaling of translation requires FMRP ubiquitination and degradation. Cultured hippocampal neurons were treated with DHPG, a gp1 mGluR agonist, for 30 seconds, and FMRP levels were quantified in distal dendrites by immunofluorescence. mGluR activation decreased FMRP signal in dendrites (33.1 ± 3.629 %) (Fig. 1A,B,C). Pretreatment of neurons with either a PP2A inhibitor, okadaic acid(Fig1.C), or UPS inhibitors, MG132(Fig.1B) or Bortezomib(Fig.1E), occluded mGluR-induced loss of FMRP, suggesting that FMRP is degraded by the UPS system and requires FMRP dephosphorylation by PP2A(Fig 1B,C,E). The rapid mGluR-induced loss of dendritic FMRP was detected with two different antibodies (data not shown). Quantitative immunofluorescence with a phosphospecific antibody showed that PP2A mediated dephosphorylation of FMRP was not dependent on UPS since DHPG-induced loss of phospho-FMRP signal was not affected by pretreatment of neurons with MG132(Fig.1D). These results indicated that dephosphorylation and degradation of FMRP are distinct events that can be uncoupled, and further suggest that PP2A activity is required for degradation.
mGluR activation was previously shown to result in a loss of FMRP signal from postsynaptic compartments of cultured neurons(Antar et al., 2004). Here, we used isolated synaptoneurosomes(SNS) to examine the effect of mGluR activity on synaptic FMRP levels in the presence of anisomycin to inhibit protein synthesis. Western blot analysis of FMRP from SNS indicated that, by 2 and 5 minutes after DHPG treatment, FMRP levels were significantly decreased(28.4±4.7% at 2 min, 45.14±15% at 5 min) in the presence anisomycin(Fig.1F,G). However, DHPG-induced loss of FMRP was absent in SNS treated with MG132, suggesting a role for UPS(Fig. 1F,G). DHPG induced loss of FMRP was dependent on signaling through mGluR5, since pretreatment of synaptoneurosomes with MPEP abolished FMRP loss (Fig1.H,I).
In order to study DHPG –induced loss of FMRP while excluding potential effects arising from FMRP transport and synthesis, or antibody accessibility, we used photoactivatable FMRP in live neurons. PAGFP-FMRP was expressed in hippocampal neurons and a small region of distal dendrites was activated using a 405 nm laser to reveal GFP-FMRP puncta in dendrites. The photoactivated GFP signal was monitored for 500 sec following DHPG stimulation or vehicle. Loss of PAGFP-FMRP signal was detectable as soon as 2 min following photo-activation. 21.3% of GFP-FMRP particles showed a decrease in signal intensity during time lapse in vehicle treated neurons (Fig.1 J-M, table1), whereas addition of DHPG increased the number of degrading particles to 76.27 %. The time course of changes in relative intensities upon DHPG treatment could be fitted to a one phase exponential decay with a half life of 4.3 min(Fig.1M, table 1). Analysis of fluorescence intensities at a proximal(0-5 μm from cell body) non-activated ROIs showed no difference between the fluorescence intensities of basal and DHPG treated neurons (data not shown), suggesting that DHPG-induced loss in the locally photoactivated FMRP signal in distal dendrites was not due to transport of photoactivated FMRP within dendrite during this very brief time period. Pretreatment of neurons with either okadaic acid or MG132 occluded the loss in dendritic FMRP signal following DHPG treatment(Fig. 1M). This finding demonstrates that UPS mediated dendritic FMRP degradation is increased by DHPG and requires PP2A activity.
We then investigated whether loss of FMRP is due to FMRP ubiquitination. High density cortical neurons pretreated with MG132 were used to immunoprecipitate endogenous FMRP. Stimulating neurons with DHPG (1′) before lysis and immunoprecipitation led to an increase in the accumulation of a high molecular weight species of FMRP which was also reactive to an anti-ubiquitin antibody(Fig.2A). The ~100kDa endogenous ubiquitinated FMRP species which appears upon DHPG treatment is in agreement with earlier reports using hippocampal slices (Hou et al., 2006). However, we could detect ubiquitination after very brief DHPG stimulation(1 minute as opposed to 10 min in the previous study). This supports a model of immediate early UPS mediated degradation of FMRP. These ubiquitinated FMRP bands could be detected by multiple FMRP antibodies. To further confirm this result, FLAG tagged FMRP was expressed in Neuro2a cells which have previously been shown responsive to mGluR activation(Liu et al., 2002). Preincubation of transfected cells with MG132 led to an accumulation of high molecular weight species labeled by anti-FLAG antibody in cells expressing FLAGmCherry-FMRP but not FLAGmCherry alone(Fig.2C). DHPG treatment further increased the proportion of FMRP in the high molecular weight smear(73 ± 19.8%)(Fig. 2C, D). In order to assess the role of PP2A activity in polyubiquitination, cortical neurons from WT and Fmr1 KO neurons were pretreated with MG132, and with or without okadaic acid, before stimulating with DHPG. The endogenous high molecular weight FMRP species, which appeared only in the WT but not Fmr1 KO neurons, was significantly reduced in the presence of okadaic acid(Fig.2B). We then analyzed if the regulation of FMRP ubiquitination by PP2A was due to a change in FMRP phosphorylation by performing ubiquitination assays with FMRP phosphomutants expressed in Neuro2a cells. FLAG-tagged phosphomutants of FMRP which mimic the phosphorylated form(S499D) or dephosphorylated form(S499A)(Ceman et al., 2003) were co-expressed with HA-ubiquitin and immunoprecipitated with an anti-FLAG antibody. Western blot analysis with an antibody to HA revealed the accumulation of high molecular weight HA smear due to polyubiquitination in the anti-FLAG-FMRP immunoprecipitate(Fig. 2E). These high molecular bands were also detected with antibody to FLAG(data not shown). Dephospho-mimic S499A showed a significant increase in accumulation of polyHA-ubiquitin(2.1 ± 0.3 fold) as compared to the WT or S499D(Fig. 2E, F). This result suggests that the activity-induced UPS-mediated degradation of FMRP is increased by dephosphorylation.
Postsynaptic density-95(PSD-95) mRNA is a well studied FMRP target whose expression and translation has been shown to increase rapidly in response to mGluR activity (Muddashetty et al., 2007; Todd et al., 2003). We measured the endogenous levels of PSD-95 in SNS by western blot(Fig. 3A) and further showed that mGluR stimulation results in an increase in endogenous PSD-95 protein expression in isolated SNS, occuring within 2 minutes(39.3 ± 11.6% at 2 min, 48.5 ± 9.06% at 5 min) (Fig. 3B), which is the same timescale as observed for FMRP degradation(Fig. 1E,F). Translation of a firefly luciferase reporter fused to the 3′UTR of PSD-95 was shown to be dependent on the presence of FMRP(Muddashetty et al., 2011) and FMRP phosphorylation status, without affecting the steady state levels of PSD-95 mRNA. The effect of inhibiting UPS and PP2A on this PSD-95 3′UTR reporter was examined. By 2 minutes of stimulation with DHPG, luciferase activity was increased 1.5 fold(Fig.3C). Removal of the 3′UTR rendered the reporter unresponsive to DHPG, without alteration in steady state mRNA levels (data not shown). Pretreament of neurons with okadaic acid or MG132 occluded the increased translation of PSD-95 mRNA in response to mGluR stimulation(Fig. 3C). Collectively, these data strongly suggest that translation of PSD-95 is tightly regulated by mGluR mediated dephosphorylation and degradation of FMRP.
Translational activation of FMRP target mRNAs has been found to be very rapid and depend on FMRP dephosphorylation(Narayanan et al., 2007). Here we show that dephosphorylation of FMRP by PP2A was required for FMRP degradation, which suggests coordinated molecular events that regulate the early steps of mGluR induced translational induction. While other studies implicate FMRP degradation in response to mGluR activation(Hou et al., 2006; Zhao et al., 2011), it was not clear if these molecular events, dephosphorylation and degradation, are coordinated and what concerted role they may play to rapidly activate translation. By detecting the accumulation of ubiquitinated FMRP and observing FMRP degradation in dendrites of live neurons in the same timescale as rapid PP2A induced dephosphorylation and PSD95-translation induction, we provide evidence for rapid FMRP ubiquitination and degradation as a component of the early signaling events lead to rapid protein synthesis.
Our recent work also shows that FMRP dephosphorylation leads to disinhibition of translation by microRNA by means of RISC dissociation from mRNA(Muddashetty et al., 2011). A recent study has reported that FMRP inhibits translation by stalling ribosomes(Darnell et al., 2011). It is possible that regulation of translation by FMRP may involve multiple sequential and coordinated steps, which may be elicited first by dephosphorylation mediated polyubiquitination and miRNA-RISC release from mRNA followed by degradation of FMRP. We speculate that ubiquitination triggers release of RISC, removing translational inhibition within the microRNP complex, yet FMRP degradation is needed to remove ribosome stalling. Further work is needed to study the coordination and interrelationships of these molecular events.
Our findings reveal a non-canonical mode of dephosphorylation induced degradation of FMRP. The conventional degradation of phosphodegrons by SCF-Ring finger ubiquitin ligases proposes that phosphorylation in a PEST sequence leads to degradation via ubiquitination of adjacent lysines(Rogers et al., 1986). Although a putative PEST site, a motif for phosphorylation based degradation exists in FMRP, overlapping with a phosphorylation site (Serine499)(Ceman et al., 2003), the half-life of FMRP expressed in fibroblasts did not seem to be changed by phosphorylation. It remains to be understood how dephosphorylation facilitates polyubiquitination as shown in our study. These questions could be answered from structural and proteomic studies. The mechanism may involve other UPS interacting proteins whose interaction may be increased by dephosphorylation. There are other studies that suggest phosphorylation prevents degradation. An N-terminal Proline in the c-mos degron promotes degradation by reducing phosphorylation at a downstream serine, which is a motif for a yet unknown ligase(Hunter, 2007; Sheng et al., 2002). Dephosphorylation may cause a conformational change exposing the ligase recognition motif in FMRP or may cause changes in phosphorylation status of other residues. Moreover, PEST sites are highly variable in conformation and are involved in multiple cellular functions(Sandhu and Dash, 2006).
Protein homeostasis at the synapse and control of long term plasticity involves the coordinate synthesis and degradation of synaptic proteins(Cajigas et al., 2010). mGluR-dependent synaptic plasticity appears to involve synthesis of proteins like Arc/Arg3.1, MAP1B and PSD-95 mRNAs which are translationally regulated by FMRP. Herein we report a novel mechanism for UPS in dendrites linking the dephosphorylation-dependent degradation of FMRP to translational disinhibition of an FMRP-target mRNA, PSD-95. These findings suggest a possible general mechanism to regulate protein synthesis dependent synaptic plasticity by the degradation of a translational repressor.
We thank Lei Xing for making the PAGFP-FMRP construct, Xiaodi Yao for the FLAG-mCherry-FMRP construct and Wilfred Rossoll for the HA-ubiquitin construct. We thank Oscar Laur and the Emory Custom Cloning Core facility. We thank Alexa Mattheyses for technical support in the microscopy core (Emory Neuroscience NINDS Core Facilities P30NS055077). We thank Stephanie Ceman for the phospho-FMRP antibody. We appreciate technical assistance by Laura Griffin, Nathan Sivanasundarum and Christopher Newhouse. This research was supported by the National Institutes of Health grant MH086405 (G.J.B) and a FRAXA postdoctoral fellowship (V.N).