FMRP S499 Is Phosphorylated Independent of mTORC1-S6K1 Activity

Hyperactive mammalian target of rapamycin (mTOR) is associated with cognitive deficits in several neurological disorders including tuberous sclerosis complex (TSC). The phosphorylation of the mRNA-binding protein FMRP reportedly depends on mTOR complex 1 (mTORC1) activity via p70 S6 kinase 1 (S6K1). Because this phosphorylation is thought to regulate the translation of messages important for synaptic plasticity, we explored whether FMRP phosphorylation of the S6K1-dependent residue (S499) is altered in TSC and states of dysregulated TSC-mTORC1 signaling. Surprisingly, we found that FMRP S499 phosphorylation was unchanged in heterozygous and conditional Tsc1 knockout mice despite significantly elevated mTORC1-S6K1 activity. Neither up- nor down-regulation of the mTORC1-S6K1 axis in vivo or in vitro had any effect on phospho-FMRP S499 levels. In addition, FMRP S499 phosphorylation was unaltered in S6K1-knockout mice. Collectively, these data strongly suggest that FMRP S499 phosphorylation is independent of mTORC1-S6K1 activity and is not altered in TSC.


Introduction
Altered mTOR signaling is a shared feature of many neurodevelopmental disorders that display high rates of mental retardation with comorbid autistic features such as TSC and fragile X syndrome (FXS) [1]. TSC, the canonical mTORopathy, is a monogenetic disorder due to mutations in TSC1 or TSC2, which are upstream regulators of mTOR kinase activity in mTOR complex 1 (mTORC1). Most patients are born heterozygous for either TSC gene and experience additional inactivating mutations during development leading to loss of heterozygosity [2]. Subsequent runaway mTORC1 activity underlies cortical malformations and slow growing tumors. Although these malformations and associated seizure activity contribute to impaired cognition, imaging and neurocognitive studies suggest that they are not sufficient to fully explain the cognitive impairment in TSC patients [3,4]. This notion is supported by animal models, in which Tsc1 or Tsc2 heterozygosity is sufficient to impair neuroplasticity and learning and memory despite the absence of brain malformations and clinical seizures [5][6][7]. Learning and memory impairments in juvenile Tsc2 +/2 mice are rescued by rapamycin treatment indicating that these deficits are reversible and mTORC1dependent. In adult Tsc2 +/2 mice, an ERK inhibitor, but not rapamycin, rescued plasticity-dependent deficits, which is consistent with increased ERK activity in these mice [8]. Adult conditional Tsc1 +/2 mice, however, do not display increased ERK activity [9], suggesting cognitive deficits in these mice might be mTORC1-dependent. Collectively, these data suggest a biochemical contribution to cognitive deficits in TSC. One candidate molecule that is reportedly downstream of mTORC1 and involved in neuroplasticity is the fragile X mental retardation protein (FMRP), and could therefore contribute to TSC symptoms.
FMRP is an mRNA-binding protein that regulates the translation of ,4-6% of brain mRNAs, many of which are involved in neuroplasticity [10][11][12]. Mutation of FMR1 (the Xlinked gene encoding FMRP) results in FXS, the leading cause of inherited intellectual disabilities and autism [13]. In the absence of FMRP, FXS model mice exhibit elevated mTORC1 activity, which may contribute to cognitive deficits and altered plasticity [14]. Under normal conditions, FMRP's contribution to neuroplasticity is in part dictated by phosphorylation of serine 499 (S499), resulting in FMRP association with stalled polyribosomes and translational repression of synaptic mRNA [15,16]. Interestingly, the kinase responsible for S499 phosphorylation was identified as the mTORC1-dependent kinase S6K1 [17]. S6K1 is thus a pivotal kinase linking mTORC1 activity to FMRP phosphorylation and function.
Because FMRP is absent in FXS and would be predicted to be hyperfunctional in TSC, it has been hypothesized that S6K1dependent FMRP S499 hyperphosphorylation in TSC might explain some of the opposite phenotypes observed in these two models of autism [6,18,19]. We thus set out to investigate S6K1 activity as well as FMRP S499 phosphorylation in TSC mouse models. Surprisingly, we found that phospho-FMRP S499 (pFMRP) levels are unchanged in heterozygous and conditional Tsc1 knockout mice despite significantly elevated mTORC1-S6K1 activity. Subsequent experiments revealed that neither up-nor down-regulating mTORC1-S6K1 signaling activity in vivo or in vitro has any effect on pFMRP levels, indicating that the mTORC1-S6K1 pathway plays no role in regulating S499 FMRP phosphorylation.

FMRP and pFMRP Antibody Validation
Prior to examining pFMRP levels, we validated the specificity of antibodies for total FMRP (tFMRP) and pFMRP S499 (referred to as pFMRP antibody). FMRP belongs to a small family of proteins that includes the fragile X-related proteins 1 and 2 (FXR1 and FXR2) and shares ,70-80% homology with FXR1/2 in the Nterminal region but essentially no homology in the C-terminal region [20,21]. Because some N-terminal antibodies can crossreact with FMRP-related proteins, we primarily utilized a Cterminal phospho-insensitive tFMRP antibody [15]. The tFMRP antibody recognized three distinct bands in cortical lysate from Fmr1 y/+ mice that were absent in Fmr1 y/2 mice ( Figure 1A). Upon longer exposure nonspecific bands (marked with asterisks) were revealed indicating equal loading between lanes ( Figure 1A). We tested two commercially available antibodies against pFMRP S499. One of the antibodies also recognized unphosphorylated FMRP and was not used further (data not shown). The second antibody (from PhosphoSolutions) has recently been used and validated [22]. We further characterized it as detailed below. The second antibody displayed a predominant pFMRP band that was absent in Fmr1 y/2 cortical lysate (arrow, Figure 1B). The pFMRP antibody also recognized two high molecular weight, non-specific bands (asterisks, Figure 1B). The same membrane was stripped (middle panel, Figure 1B) and reprobed with the tFMRP antibody indicating that the major pFMRP band was indeed FMRP (right panel, Figure 1B). To determine whether the pFMRP antibody is phospho-specific, Fmr1 y/+ cortical lysate was incubated with or without lambda phosphatase. The pFMRP antibody detected the FMRP-specific band in untreated lysate that was absent in phosphatase-treated lysate (left panel, Figure 1C). The same membrane was stripped and reprobed with the tFMRP antibody to ensure equivalent amounts of total FMRP in treated and untreated samples (right panel, Figure 1C). For additional verification the pFMRP antibody was tested against Neuro2a cell lysate (positive control, N2a) and recombinant human FMRP (rFMRP), which is devoid of post-translational modifications [23]. After running the gels for additional time to allow for better separation of FMRP isoforms, we found that the pFMRP antibody recognized two of four N2a FMRP bands, but did not recognize unmodified rFMRP ( Figure 1D). Of the 12 predicted murine FMRP isoforms, only two (isoforms 1 and 7) contain the S499 phosphorylation site [24]. To validate that the pFMRP 499 antibody specifically recognized the S499 site, we obtained a vector encoding GST-tagged FMRP and generated two S499 mutants, one with an alanine (S499A) and one with an aspartic acid (S499D) substitution. In transfected N2a cells, the tFMRP antibody recognized all three GST-tagged FMRP proteins, but the pFMRP antibody only recognized the S499, and not S499A or S499D, GST-tagged FMRP proteins ( Figure 1E). These data indicate that the pFMRP antibody specifically recognizes phosphorylated FMRP S499.
We next verified that S6K1 can phosphorylate FMRP S499 in vitro as previously reported [17]. Recombinant S6K1 or S6K2 were incubated with ATP and rFMRP. S6K1 displayed robust kinase activity towards FMRP as assessed by pFMRP immuno-blotting (n = 3, Figure 1F). S6K2 exhibited minor activity towards FMRP compared to S6K1.
Phospho-FMRP S499 is not Increased in Tsc1 +/2 Mice Despite Elevated S6K1 Activity Because mTORC1 pathway activation has not been thoroughly characterized in Tsc1 +/2 mice, we first examined the phosphorylation levels of the mTORC1-dependent site on S6K1 T389 (pS6K1), which functions as an activating phosphorylation. We secondarily characterized phosphorylation of the S6K1-dependent sites on ribosomal protein S6 S240/244. We used the hippocampus of 2 month-old male Tsc1 +/2 mice to avoid potential variations in tFMRP levels between gender [25,26]. For all the conditions, we compared and graphed the levels of phosphoprotein divided by total protein (e.g., pFMRP:tFMRP). Despite significantly decreased levels of TSC1 and TSC2 in Tsc1 +/2 compared to Tsc1 +/+ mice (N = 4 for TSC1 and 6 for TSC2, Figure 2A and B), pS6K1 and pS6 levels were identical between Tsc1 +/+ and Tsc1 +/2 mice in whole cell lysates (N = 6-14, Figure 2C and D, dark grey bars in D). Significance (asterisks) is noted on each graph and the statistical tests are mentioned in each figure legend. We thus prepared synaptically enriched P2 fractions [27] to more specifically examine neuronal pS6K1 and pS6 levels. In P2 fractions, pS6K1 as well as pS6 were significantly elevated in Tsc1 +/2 mice compared to Tsc1 +/+ mice, suggesting that mTORC1-S6K1 activity is significantly elevated in neurons of Tsc1 +/2 mice (N = 6 or 7, Figure 2C and D, light grey bars in D). Surprisingly, however, despite elevated phosphorylation of mTORC1-dependent signaling markers, pFMRP levels were not significantly different between the P2 fractions of Tsc1 +/+ and Tsc1 +/2 mice ( Figure 2C and D). These data led us to question and further examine the mTORC1-S6K1 connection to FMRP phosphorylation.
Phospho-FMRP S499 is Not Altered in the Conditional Tsc1 2/2 Forebrain or Downstream of Hyperactive mTORC1 in vitro To determine whether FMRP phosphorylation might be increased in the Tsc1 2/2 state in vivo, we generated conditional heterozygous and knockout Tsc1 mice. Tsc1 fl/+ ;R26R-tdTomato (fl for floxed) mice were bred against Tsc1 wt/2 ;Emx1-Cre mice to yield mixed litters of forebrain-specific Tsc1 wildtype, heterozygous, and knockout mice. In Emx1-Cre mice, Cre recombinase is expressed in forebrain glial and glutamatergic progenitors beginning around E9.5 [28]. Forebrain-specific Cre-mediated recombination was verified by the expression of tdTomato fluorescence and PCR for Tsc1 ( Figure 3A). mTORC1-S6K1 signaling was dramatically elevated in the forebrain as demonstrated by significantly increased S6 S240/244 phosphorylation in Tsc1 fl/2 ;Emx1-Cre mice containing Tsc1 2/2 cells. Despite significantly elevated mTORC1-S6K1 activity, neither tFMRP nor pFMRP levels were altered in the cortex of these mice ( Figure 3B and C).
Because mTORC1 could regulate FMRP S499 phosphorylation independent of TSC1/2, we next used several manipulations to increase mTORC1 activity in N2a cells and assessed the impact on pFMRP levels. Serum is known to activate the mTORC1 pathway ( Figure 3D). N2a cells were thus exposed to 0, 2.5, 5 or 10% serum for 1 hour prior to sample collection. Using mTOR S2448 [29], S6K1 T389 and pS6 S240/244 as readouts of mTORC1 signaling, we found that mTORC1 pathway activity was highly correlated with serum concentration in N2a cells. In contrast, pFMRP levels were unaffected by serum (n = 4 per condition, Figure 3E and F). We then overexpressed different components of the mTORC1 pathway that are expected to increase mTORC1 activity: constitutively active AKT (AKT CA ), dominant negative TSC2 (TSC2 DN ), constitutively active Rheb (Rheb CA ), and hyperactive mTOR (mTOR hyper ) ( Figure 3G and H). All of these manipulations led to a significant increase in S6K1 T389 and pS6 S240/244 phosphorylation, however, pFMRP levels remained unchanged in these same samples (n = 4 per condition, Figure 3I and J).

Pharmacological Inhibition of mTORC1 and S6K1 Activity does not Affect pFMRP S499 Levels in vitro
To determine whether apparent insensitivity of pFMRP levels to elevated mTORC1 signaling was simply an artifact due to a ceiling effect, we inhibited mTORC1-S6K1 signaling in vitro and in vivo. We also inhibited PP2a in vitro, which was previously shown to dephosphorylate FMRP [30]. In total, three drugs were utilized to inhibit mTORC1-S6K1 signaling: rapamycin, PF-4708671, and bisindolylmaleimide V (B5). Rapamycin inhibits the mTORC1 complex and has been extensively characterized. PF-4708671 and B5 are recently identified S6K1 inhibitors [31,32]. Okadaic acid was used to inhibit PP2a activity ( Figure 4D for signaling model). We consistently found that rapamycin, PF-4708671, and B5 significantly reduced the activity of mTORC1-S6K1 signaling as assessed with pS6 S240/244 even in conditions of serum stimulation ( Figure 4A, B, E and F). As previously reported, PF-4708671 induced a paradoxical hyperphosphorylation of S6K1 T389 [31] ( Figure 4A and E). Okadaic acid significantly increased the phosphorylation of the PP2a substrate ERK T202/Y204 ( Figure 4A and C). However, none of these blockers affected pFMRP or tFMRP levels ( Figure 4A-C and E-F).
Phospho-FMRP S499 is Unaltered by Pharmacological Inhibition of mTORC1 or S6K1 Activity in vivo, or in S6k1knockout Mice For in vivo experiments, we initially determined that PF-4708671 crosses the blood-brain-barrier, and found that it was cleared from the brain by 4 hours post-injection ( Figure 5A). We performed intraperitoneal injections of either rapamycin (1.5 mg/ kg) or PF-4708671 (75 mg/kg) while control mice were injected with a similar volume of vehicle (i.e, DMSO). Rapamycin injections were performed daily for 5 days and PF-4708671 mice were injected once and sacrificed two hours post-injection. Both rapamycin and PF-4708671 significantly decreased mTORC1-S6K1 activity as demonstrated by reduced pS6 S240/244 levels, however, pFMRP levels were unchanged ( Figure 5B and C).
It was previously reported that pFMRP S499 was absent in S6k1 KO mice [17]. We thus probed S6k1 +/+ and S6k1 2/2 cortical lysates with N-and C-terminal S6K1 antibodies to verify the absence of S6k1 full length and/or partial protein products. Despite a complete absence of S6K1 protein, neither pFMRP nor tFMRP levels were different between S6k1 genotypes, suggesting that S6K1 is dispensable for phosphorylation of native FMRP at S499 ( Figure 6A and B).
Collectively, these data strongly suggest that FMRP S499 is not phosphorylated by S6K1 and demonstrate that S6K1 is not required for FMRP S499 phosphorylation. The specificity of the C-terminal, phospho-insensitive FMRP tFMRP antibody (Abcam #17722) was verified by immunoblotting whole cell cortical lysates from 2 month-old Fmr1 y/+ (WT) and Fmr1 y/2 (KO) mice. Short and long exposures (exp.) are shown. Arrows point to three FMRP-specific bands while asterisks point to nonspecific bands. Molecular weight is indicated by the colored ladders on the left which refer to panels A-D (B) Fmr1 y/+ and Fmr1 y/2 cortical lysates were immunoblotted with a pFMRP antibody and the film exposed for 1 minute (PhosphoSolution #p1125-499). The membrane was then stripped and reprobed with secondary only antibody (1 minute exposure) to verify removal of the primary antibody. The membrane was then probed with tFMRP antibody (30 second exposure). The arrow points to the FMRP-specific band while asterisks point to two nonspecific pFMRP bands. (C) Cortical lysate from Fmr1 y/+ mice was incubated with or without lambda (l) phosphatase for 30 minutes and immunoblotted with pFMRP, stripped and reprobed with tFMRP. (D) Neuro2a cell lysate (N2a) and unphosphorylated recombinant human FMRP (rFMRP) were immunoblotted with pFMRP. (E) GST-FMRP S499, S499A and S499D were transfected into N2a cells and lysates analyzed by immunoblot 24 hours later. tFMRP recognized all three isoforms (black, GST-labeled arrow) as well as endogenous FMRP (white arrows) however pFMRP only recognized S499 and endogenous FMRP from the same membrane. (F) Recombinant FMRP was incubated with ATP and with, or without (Neg), S6K1 or S6K2 and immunoblotted for pFMRP or total FMRP (tFMRP = 1C3 antibody here). The experiments were reproduced in triplicate. doi:10.1371/journal.pone.0096956.g001 mTORC1 Activity is not Required for mGluR I-dependent Phosphorylation of FMRP S499 Given that mGluR class I activity has been reported to regulate the phosphorylation of FMRP S499 [17], we investigated the dynamics of FMRP translation and phosphorylation following mGluR I activation in N2a cells which, have been used by others to study mGluR I-dependent FMRP signaling [33]. N2a cells were treated with (S)-DHPG (100 mM) for 0, 1, 2, and 5 minutes. Additional samples were collected 5 and 25 minutes post-washout. Importantly, DHPG elicited dynamic changes in pERK1/2 and mTOR-S6K1 signaling ( Figure 7A). tFMRP levels increased as early as 2 minutes following DHPG application and was accompanied by increased FMRP S499 phosphorylation, but there was no change in the ratio of pFMRP:tFMRP (n = 6, Figure 7A). In addition, experiments in the presence of rapamycin or PF-4708671 did not prevent FMRP phosphorylation accompanying FMRP synthesis (n = 3 each, Figure 7B).

Discussion
Our study demonstrates that mTORC1-S6K1 signaling does not contribute to phosphorylation of FMRP S499.
In order to study the mTORC1-S6K1-FMRP connection, we first examined whether mTORC1 pathway activity is increased in Tsc1 +/2 mice. Of note, our data represent the first biochemical detection of elevated S6K1 T389 phosphorylation in Tsc1 +/2 mice. In these mice, we only detected elevated S6K1 T389 in neuronally enriched P2 fractions supplemented with okadaic acid. Given that mTORC1-S6K1 signaling was found not to be elevated in the Tsc1 2/2 astrocytes [34], the lack of elevated S6K1 phosphorylation in whole hippocampal lysate could be due to a masking effect of glial mTORC1 signaling, which is presumably normal in the heterozygous state. Consistent with elevated S6K1 T389 phosphorylation, we also observed an increase in phosphorylation of the S6K1 substrate S6 S240/244 in P2 fractions. While one group reported elevated S6 S235/236 in Tsc2 +/2 mice [5], phosphorylation of S235/236 can be modulated independent of mTORC1 signaling [35]. Additionally, although pS6 S235/236 was elevated in the telencephalon of Tsc2DRG mice (dominant negative), this was found to be due to enhanced ERK-RSK signaling rather than mTORC1 activity [9]; ERK signaling, however, is not elevated in conditional Tsc1 +/2 telencephalon [9]. Although elevated pS6 S240/244 phosphorylation could have been due to decreased phosphatase activity, we found that expression of the main pS6 phosphatase, PP2a, was elevated in our P2 fractions making this explanation less likely (unpublished observations, Bartley and Bordey).  Importantly, despite evidence of elevated mTORC1-S6K1 activity in P2 fractions of Tsc1 +/2 mice and in the forebrain of conditional Tsc1 knockout mice, we could not detect elevated FMRP S499 phosphorylation. Because we found that S6K1 can phosphorylate recombinant human FMRP in vitro, we tested whether our inability to detect increased FMRP S499 phosphorylation may be because the mTORC1-S6K1 axis modulates FMRP S499 phosphorylation independent of the TSC pathway or because basal FMRP S499 phosphorylation is saturated. We therefore tested whether either overexpression of other inputs to mTORC1 or inhibition of mTORC1-S6K1 activity would alter FMRP S499 phosphorylation.
None of the manipulations that altered mTORC1-S6K1 activity, either in vitro or in vivo, affected the degree of FMRP S499 phosphorylation. Finally, FMRP S499 phosphorylation is identical between S6k1 +/+ and S6k1 2/2 mice, indicating that S6K1 is dispensable for normal FMRP S499 phosphorylation. This finding is in contrast with data in a previous study [17]. Although our study used a different phospho-specific antibody, the antibody used by the other group was validated against FMRP S499A in two other studies [30,36]. It is possible that their antibody recognizes other, non-S499, phosphorylated motifs on FMRP. If a negative charge at S499 is required for phosphorylation of other sites on FMRP, as suggested by radiolabeling studies using S499, S499A and S499D [16], then the nonspecificity of their antibody would not be identified using FMRP S499A (which would be predicted to be completely unphosphorylated). In any event, consistent with findings from other groups, we found that mGluR I stimulation by DHPG does increase the level of tFMRP [37][38][39][40][41]. We also found that pFMRP S499 increased in parallel with tFMRP following DHPG stimulation, but was insensitive of mTORC1-S6K1 inhibition.
Collectively, these data demonstrate that S6K1 is not required for phosphorylation of endogenous FMRP at S499. Furthermore, these data indicate that the mTORC1-S6K1 pathway does not regulate the phosphorylation of FMRP S499 in any way. This negative finding is consistent with the fact that the sequence surrounding FMRP S499, EASNApS, contains none of the features that would render it a good S6K1 candidate. The preferred phosphorylation motif for S6 kinases is relatively well preserved, RXRXXpS (where X = any amino acid and p denotes the phosphorylated residue). With rare exception S6 kinase substrates deviate from this sequence [42,43], however, sequences that do diverge generally contain an arginine (R) in the 23 or 25 position [44]. Although we and another group found that S6K1 can phosphorylate FMRP in vitro [17], this may be an artifact of the high concentrations of isolated S6K1 and FMRP utilized in these kinase assays which could promote a nonphysiologic interaction between these two proteins. In the previous study, a phospho-specific antibody was used to verify that S6K1 can phosphorylate FMRP. However it should be noted that in this study activity-and S6K1-dependent changes in phosphorylated FMRP were only measured using radioactive phosphate. This radioactive method, however, cannot distinguish phosphorylated S499 from other phosphorylated sites, which could be S6K1dependent. We cannot explain, however, why pFMRP S499 was absent from S6K1 knockout mice using their phospho-specific antibody but present in our S6K1 knockout mice.
Our findings open clear questions. In particular, it is important to identify the kinase(s) responsible for FMRP S499 phosphorylation. The recent finding that phospho-mimetic FMRP but not phospho-dead FMRP can fully rescue dFmr1 drosophila highlights the biological significance of this phosphorylation site [45]. A previous study in drosophila reported that CK2, formerly casein kinase 2, phosphorylates a S499 homologous site, dFMRP S406 [46]; and this has been suggested in mice by [47] as well as in our hands (Bartley and Bordey, unpublished observations). If CK2 is the kinase for mammalian FMRP S499 this poses a particular conundrum for the activity-dependent regulation of FMRP S499 phosphorylation. CK2 is considered to be a constitutively active kinase which would suggest that S499 is regulated primarily by phosphatase activity or that phosphorylation sites other than S499 are regulated in an activity-dependent manner. In general, activity-dependent detection of changes in the phosphorylation of FMRP has been performed using radioactive phosphate or phospho-serine antibodies, which are incapable of distinguishing S499 phosphorylation from other phosphorylated residues.
Our initial interest was to investigate the contribution of dysregulated FMRP S499 phosphorylation to TSC. To our surprise we were unable to find a link between mTORC1-S6K1 and S499 phosphorylation under any condition. Although our findings strongly suggest that another kinase is responsible for the phosphorylation of FMRP S499, the mTORC1-S6K1 pathway may yet regulate the phosphorylation of other FMRP residues.

Ethics Statement
All animal research protocols were approved by the Institutional Animal Care and use Committee, Yale University.
Tsc1 fl/+ mice (Jackson Labs) were also generated by David J. Kwiatkowski. With the exception of Tsc1:Emx1-Cre transgenic mice, all mice used in this study were 2 months old. Because all Tsc1 flox/2 mice die by P20 due to seizure, only P7 mice were used in this study.
Genotyping was performed either in house or using Genetyper services (www.genetyper.com). For in-house and Genetyper-assisted genotyping we used a three-primer protocol that allows for the simultaneous detection of wild type, mutant and floxed Tsc1 alleles: TSC1F4536:59 -AGG AGG CCT CTT CTG CTA CC; Figure 4. Inhibiting mTORC1 and S6K1 activity has no effect on pFMRP S499 levels in vitro. (A) N2a cells were treated with vehicle 1 (DMSO), vehicle 2 (ethanol), rapamycin (Rapa), PF-4708671 (PF), bisindolylmaleimide V (B5) or okadaic acid (OA) for 1 hour prior to cell lysis followed by immunoblotting for pFMRP, pS6K1, pS6, pERK T202/Y204 (readout for OA), and their total protein counterparts. Laddered bracket to right of tFMRP indicates FMRP isoforms and the asterisk denotes a nonspecific band. S6K1 isoforms are marked by p80 and p75. (B and C) Statistical verification of stable pFMRP:tFMRP across all conditions despite a significant decrease in pS6:S6 subsequent to mTORC1-S6K1 inhibition (B) and increase in pERK:ERK subsequent to PP2a inhibition (C). One way ANOVA with post-hoc Dunnett's test (N = 4 per condition. Error bars = SEM). (D) Model of pathway and effect of pharmacological inhibitor. (E) N2a cells. Immunoblotting for pFMRP, tFMRP and mTORC1 pathway components from N2a cells maintained in 5% serum, transferred to increasing concentrations of serum (0, 2.5, 5 or 10%), and treated with vehicle (DMSO), rapamycin (Rapa) or PF-4708671 (PF) for 24 hours. Laddered bracket to the right of tFMRP blot indicates FMRP isoforms and the asterisk a nonspecific band. S6K1 isoforms are indicated by p85 and p70. The asterisk to the right of the total S6K1 blot indicates residual tFMRP signal from the blot above. (F) Bar graphs of (E). Statistical analysis: unmatched two-way ANOVA corrected for multiple comparisons with a post-hoc Tukey's test. N = 4 per condition. Error bars = SEM. doi:10.1371/journal.pone.0096956.g004

Tissue Preparation
Mice were anesthetized with isoflurane followed by decapitation. Brains were acutely dissected in ice-cold Hank's Balanced Salt Solution (HBSS, GIBCO Cat. No. 14170-112), snap frozen in liquid nitrogen and stored at 280uC. For whole cell lysates, samples were homogenized in RIPA buffer supplemented with DNase I (8 U/10 ml), 1x HALT protease/phosphatase inhibitor cocktail (Pierce #78443), and okadaic acid (100 nM). Protein concentrations were quantified using a standard BCA protein assay (Pierce #23225). For a given experiment, samples were diluted to the same concentration with lysis buffer and then boiled in an equal volume of 2x Laemmli buffer at 99uC for 5 minutes.

P2 Fractionation
Tissues were homogenized in P2 lysis buffer (4 mM HEPES, 0.32 M Sucrose, 1x HALT, 5 mM EDTA, and 100nM Okadaic Acid) with micro-pestles (RPI #199222); 20 strokes/sample. Homogenates were centrifuged at 1000 g for 10 minutes at 4uC and the pellet discarded. Supernatants were recentrifuged at 10,000 g x 15 minutes at 4uC and the supernatant set aside as the cytoplasmic fraction. The pellet (P2 fraction) was resuspended in P2 buffer and centrifuged again at 10,000 g for 15 minutes at 4uC. The supernatant was discarded and the P2 fraction was resuspended in 50 mM Tris-H 2 O supplemented with HALT. Cytoplasmic and P2 fraction protein concentrations were quantified and the samples were boiled in an equal volume of 2x Laemelli Sample Buffer. We validated that P2 fractions were enriched for synaptic proteins PSD95 and SAPAP3 and relatively depleted of nonsynaptic proteins such as a-tubulin and the glial protein GFAP [48] (data not shown).

Lambda Phosphatase Assay
A single CD1 adult mouse hippocampus was homogenized in 300 ml phosphatase assay lysis buffer (4 mM HEPES, 0.5% Triton-X-100, 120 mM NaCl, and 2 Roche protease inhibitor tablets per 10 ml). The sample was centrifuged at 13,000 rcf for 10 minutes at 4uC. 800U lambda phosphatase (NEB #P0753S) was added to 100 ml supernatant and the sample incubated at 37uC for 30 minutes. The reaction was terminated by addition of an equal volume of 2x Laemmli buffer.

Antibodies
Antibodies and usage parameters are listed in Table 1.

Plasmid and Transfection
Transfections were formed using PolyJet transfection reagent (SignaGen) according to the manufacturer's protocol. Vectors and their sources are listed in Table 2. Kinase Assay with Recombinant (r) FMRP Kinase assays were performed by Kinexus (Canada). Recombinant FMRP S500 and S500D were generated as previously described [23].

Neuro2a (N2a) Cell Culture
In general, N2a cells were cultured in complete media (Dulbecco's Modified Eagle Medium (DMEM) (Gibco 11965-092), 5% fetal bovine serum (FBS) (Gibco 16140-071), and 1% Penicillin-Streptomycin (Gibco 15140-122) in a 37uC incubator at 5% CO 2 . When cells reached approximately 70% confluence in six-well plates, the medium was replaced with pre-warmed complete media and treatment was begun one hour later. For in vitro experiments, individual treatments and transfections were performed between two and six times prior to performing the combined experiments represented in figures 2 and 3 (N = 4 for each condition in in vitro experiments). Cell lysis was performed on ice. Cells were rinsed twice with ice-cold 1X phosphate-buffered saline Laemmli and lysed in N2a lysis buffer (RIPA, 1x HALT protease/phosphatase inhibitor cocktail, 8U/10ml DNase I, 100nM okadaic acid). Cells were then scraped from the wells and lysates centrifuged at 16,000 RCF for 20 minutes at 4uC. The supernatant was added to 6X Laemmli sample buffer to a final concentration of 1X sample buffer and boiled for 5 minutes at 99uC.

Immunoblotting
All western blots were performed using 10% tris-gylcine gels and protein transferred to PVDF membranes according to a standard wet transfer protocol. In cases where cross-blot normalization was required a standard sample was loaded on each gel to account for inter-gel variability. In general, the optimal linear range for each antibody was determined using the appropriate sample type (P2, whole cell etc.) prior to experimental immunoblot assays. In some cases, limited linear ranges were run on the same gel (that is 80% and 120% of a control sample were loaded in end lanes) to ensure detectability of minor changes in protein signals. Densitometry was performed using Image J without background correction or rolling ball adjustments. All phospho-protein signals were normalized to total protein signals from the same blot. For phospho-proteins, adequate removal of phospho-antibody was verified by probing with secondary alone after stripping the membrane. Raw ratios of phospho-protein normalized to total protein or total protein normalized to loading control (generally ERK unless otherwise stated) were calculated in Microsoft Excel.

Statistical Analysis
Statistical analysis was performed on raw densitometric ratios using GraphPad 6. For data presentation, values were normalized to control data such that control groups were always = 1. For in vitro experiments, the data is represented as the % change from the control lane on the same membrane; as such, control lanes are without error bars. Statistical significance was determined using Mann-Whitney U, one-way ANOVA or two-way ANOVA using Dunnett's and Turkey's post-hoc tests where indicated. P,0.05 was considered significant. Data are shown as mean 6 standard of the mean (SEM) unless otherwise specified.