Positive feedback regulation of Akt-FMRP pathway protects neurons from cell death


Address correspondence and reprint requests to Chan Young Shin, Department of Pharmacology, School of Medicine, Konkuk University, 1 Hwayang-Dong, Gwangjin-Gu, Seoul 143-701, Korea.
E-mail: chanyshin@kku.ac.kr


J. Neurochem. (2012) 123, 226–238.


Fragile X syndrome (FXS), the most common single genetic cause of mental retardation and autistic spectrum disease, occurs when FMR1 gene is mutated. FMR1 encodes fragile X mental retardation protein (FMRP) which regulates translation of mRNAs playing important roles in the development of neurons as well as formation and maintenance of synapses. To examine whether FMRP regulates cell viability, we induced apoptosis in rat primary cortical neurons with glutamate in vitro and with middle cerebral artery occlusion (MCAO) in striatal neurons in vivo. Both conditions elicited a rapid, but transient FMRP expression in neurons. This up-regulated FMRP expression was abolished by pre-treatment with PI3K and Protein Kinase B (Akt) inhibitors: LY294002, Akt inhibitor IV, and VIII. Reduced FMRP expression in vitro or in vivo using small hairpin Fmr1 virus exacerbated cell death by glutamate or MCAO, presumably via hypophosphorylation of Akt and reduced expression of B-cell lymphoma-extra large (Bcl-xL). However, over-expression of FMRP using enhanced green fluorescent protein (eGFP)-FMRP constructs alleviated cell death, increased Akt activity, and enhanced Bcl-xL production. The pro-survival role of Akt-dependent up-regulation of FMRP in glutamate-stimulated cultured neuron as well as in ischemic brain may have a clinical importance in FXS as well as in neurodegenerative disorders and traumatic brain injury.

Abbreviations used

protein kinase B


B-cell lymphoma-extra large




cyclin-dependent kinase


cytoplasmic polyadenylation element binding


enhanced green fluorescent protein


fragile X mental retardation protein


fragile X syndrome


glyceraldehyde 3-phosphate dehydrogenase


glycogen synthase kinase-3 beta


middle cerebral artery occlusion


3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide


nuclear localization signal


phosphate-buffered saline




propidium iodide


phosphatidylinositol 3-kinase


phosphatase 2A


terminal deoxynucleotidyl transferase dUTP nick end labeling

Fragile X syndrome (FXS) is a neurodevelopmental disorder with a wide spectrum of cognitive and behavioral problems [for review, see (Krueger and Bear 2011)]. Loss of fragile X mental retardation 1 (Fmr1) gene expression is an established cause of FXS (Krueger and Bear 2011). The expansion of CGG repeats (> 200) in 5′ untranslated regions of Fmr1 gene results in hypermethylation, transcriptional silencing, and loss of fragile X mental retardation protein (FMRP) (Chakrabarti and Davies 1997).

The Fmr1 gene encodes FMRP known to regulate translation of brain mRNAs which are related to dendritic growth (Darnell et al. 2001; Nimchinsky et al. 2001), synapse formation (Darnell et al. 2001), and monoamine synthesis (Berry-Kravis and Ciurlionis 1998). FMRP silencing induces abnormal regulation of brain mRNA translation (Brown et al. 2001; Napoli et al. 2008; Edbauer et al. 2010) and a disruption in the composition of the appropriate protein milieu, which mediates defects in neuronal development and synaptogenesis (Bassell and Warren 2008).

FMRP function is thought to be mediated largely by its ability to bind mRNAs through two well-known RNA-binding domains, like K-homology domains, and RGG (arg-gly-gly triplet) box (Siomi et al. 1993). Through these domains, FMRP recognizes, regulates (mainly represses) translation, and transports specific mRNAs (Brown et al. 2001; Zalfa et al. 2007; Bassell and Warren 2008; Dictenberg et al. 2008). In neurons, FMRP may modulate expression of mRNAs by controlling recognition, export, translational efficiency, and stability of target mRNAs (Darnell et al. 2001; De Rubeis and Bagni 2010). Meanwhile, it is also known that FMRP has a nuclear localization signal (NLS) and a nuclear export signal (Eberhart et al. 1996), suggesting that it shuttles between the nucleus and cytoplasm. In fact, FMRP is found in both nucleus and cytoplasm, an indication that it may have several independent functions within the cell such as the regulation of cell division, growth, and survival (Siomi et al. 1993). Among the various mRNAs which might be regulated by FMRP, some are related to the regulation of cell survival and death. These include components of Bcl-2 family-like Bcl-2-interacting protein (Bnip) and Bcl-2-interacting killer as well as signaling molecules such as NF-kB, PKC, and Mitogen-activated protein kinase (Brown et al. 2001; Chen et al. 2003).

Many studies have identified Ras-phosphatidylinositol 3-kinase (PI3K)- Protein Kinase B (Akt) as a major cell survival signaling pathway (Engelman et al. 2006; Chalhoub et al. 2009; Wagner-Golbs and Luhmann 2012), which makes the pathway an attractive target for therapeutics against many forms of cancer with misregulated cellular apoptosis (Vivanco and Sawyers 2002). Akt is a serine/threonine protein kinase that plays a key role in multiple cellular processes such as cell proliferation, apoptosis, and transcription (Datta et al. 1997). Among these various functions, Akt indirectly activates NF-kB transactivation by dissociation of phosphorylated Ik-B kinase, resulting in transcription of pro-survival genes like B-cell lymphoma-extra large (Bcl-xL) and trophic factors (Brunet et al. 2001). In addition, trophic events alleviate excitotoxic and ischemic injury by activating PI3K-dependent Akt activation. Collectively, Akt signaling pathway seems to be one of the major mediators or determinants of cellular survival or death.

Interestingly, abnormal Ras-PI3K-Akt signaling cascades were reported in Fmr1 knockout animals (Hu et al. 2008). Although both Ras–MEK–ERK1/2 Mitogen-activated protein kinase cascades and Ras-PI3K–Akt cascades were normal in wild-type and Fmr1 knockout animals in basal condition, the stimulation-induced activation of PI3K–Akt was weakened and abnormal in Fmr1 knockout animals (Hu et al. 2008). Interestingly, standardized incidence ratio of cancer was reduced to 0.28 (95.0% ± 0.8) compared to control in a study of Dutch FXS patients (Schultz-Pedersen et al. 2001). These results suggest that down-regulation of FMRP may increase cell death via abnormal Ras-PI3K-Akt signaling in stimulated condition, which may misregulate downstream targets such as Bcl-xL. In this study, we hypothesized that FMRP may control cell survival and death in neurons via the regulation PI3K-Akt pathway and investigated the possibility using in vitro glutamate stimulation and in vivo transient focal ischemia (middle cerebral artery occlusion, MCAO) paradigm.

Materials and methods


Neurobasal medium (Gibco BRL, Grand Island, NY, USA) and B-27 supplement (Invitrogen, Carlsbad, CA, USA) were purchased from each vendor. The antibodies for western blotting of p-PI3K (regulatory p85 subunit of PI3K), PI3K, p-Akt (Ser 473), Akt, caspase 3 (total and cleaved form) were obtained from Cell Signaling (Beverly, MA, USA), and Bcl-xL antibody was from Santa Cruz Biotechnology (Santa Cruz, CA, USA). For western blot, FMRP antibody was from Millipore (MAB2160, Billerica, MA, USA) and for immunohistochemistry, FMRP antibody was purchased from Abcam (ab17722; Cambridge, MA, USA). Small hairpin RNA virus and related reagents were all purchased from Sigma (St. Louis, MO, USA). SYBR green mix was obtained from Fermentas (Glen Burnie, MD, USA) and Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay kit was obtained from Millipore. Annexin V assay kit was from Phoenix Flow Systems (San Diego, CA, USA). Transfection reagents were purchased from Invitrogen and Roche (Roche Diagnostics Corp., Indianapolis, IN, USA). All other reagents were purchased from Sigma (St. Louis, MO, USA). Dr. Darnell (The Rockefeller University, New York) kindly provided enhanced green fluorescent protein (eGFP)-empty vector and eGFP-tagged FMRP vector (Darnell et al. 2005).



For in vivo experiments, male Wistar rats (300 g, 10 weeks) were used and a total of 90 animals were killed throughout this study. For cell culture experiments, timed pregnant female rats were obtained from Dae Han Biolink (Seoul, Korea). Animal treatment and maintenance including anesthesia, surgery, and immunohistochemistry were approved by the Institutional Animal Welfare Committee of Konkuk University and carried out in accordance with the Principles of Laboratory Animal Care (NIH publication No. 85–23 revised 1985) and the Animal Care and Use Guidelines of Konkuk University, Korea.

Cell culture

Primary cortical neuron cultures were prepared from the cerebral cortex of fetal SD rat at embryonic day 16 as previously reported (Jeon et al. 2011a). Briefly, cortices were dissected, incubated in trypsin-EDTA, and dissociated with a Pasteur pipette. The dissociated cells were resuspended in B27-supplemented neurobasal medium and plated onto poly-d-lysine-pre-coated culture dishes. Cultured cortical neurons were kept for 10 days at 37°C in 10% CO2 environment before experiments.

In vitro cell treatments

For stimulation of cortical neurons, 50 μM glutamate was bath applied in the culture media for different time points, PI3K inhibitor (LY294002; 10 μM) and Akt inhibitors (IV, and VIII; final 10 μM) were pre-treated for 1 h, respectively. Unless otherwise specified, cells were harvested after 0.5-h glutamate treatment in most experiments.


Male Wistar rats were used for transient MCAO surgery under a general anesthesia, as we previously reported with a slight modification (Kim et al. 2011). In brief, silicon-coated nylon monofilament was introduced into right middle cerebral artery for 1 h and induced occlusion. After occlusion, filaments were removed allowing reperfusion of blood to the affected areas and lesions were sealed. For PCR and western blot analysis, the rats were anesthetized, decapitated and the striatum was dissected. For immunohistochemical analysis, the rats were anesthetized with ether and transcardially perfused with 4% PFA at 24 h post ischemia. Brain samples were post-fixed in 4% PFA for 24 h and equilibrated into 30% sucrose for three days prior to cryosectioning.

Real-time reverse transcription-polymerase chain reaction

Using Trizol reagent (Invitrogen), total RNA was extracted from the cell and converted to cDNA by Maxime RT PreMix Kit (iNtRON Biotechnology, Seoul, Korea). For real-time PCR, total cDNAs were diluted 1 : 10 in double-distilled water and Maxima® SYBR Green/ROX qPCR Master Mix (2X) (K0221, Fermentas). The primers for FMRP (accession number NM_002024.4) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH, accession number M17701) used for amplification reactions are as follows:

for FMRP,

forward primer : 5′-TTG GTA CCT TGC ACA CAT CA-3′

reverse primer : 5′-AAG TTA GCG CCT TGC TGA AT-3′

for GAPDH,

forward primer : 5′-TCC CTC AAG ATT GTC AGC AA-3′

reverse primer : 5′-AGA TCC ACA ACG GAT ACA TT-3′

The PCR protocol was as follows: 95°C for 30 s, 60°C for 8 s, 72°C for 15 s, and continued by a final step at 4°C for 10 s. After all the reactions were finished, data were compiled automatically by the equipment (7500 Real-Time PCR Systems; Applied Biosystems, Carlsbad, CA, USA). Results were circulated by comparative CT method (Livak and Schmittgen 2001).

For semiquantitative RT-PCR, the amplification reaction was performed using Maxime PCR premix Kit (iNtRON Biotechnology) as previously reported (Jeon et al. 2011b). The expected, size of the amplified DNA fragments was 486 base pairs for FMRP and 308 base pairs for GAPDH, respectively.

Western blot

After treatment, cultured cells were lysed with 2X sample buffer (4% w/v sodium dodecyl sulfate, 20% glycerol, 200 mM dithiothreitol, 0.1 M Tris-HCl, pH 6.8, and 0.02% bromophenol blue). For in vivo experiments, rats were killed and briefly perfused with ice-cold saline (pH 7.4). Brains were rapidly removed and dissected to prefrontal cortex and striatum. Brain tissues were homogenized in 1 mL Radioimmunoprecipitation assay buffer containing 62.5 mM Tris–HCl (pH 6.8), 2% w/v SDS, 10% glycerol, 10 mM 2–mercaptoethanol, 1 mM sodium orthovanadate, 100 μM phenylmethylsulfonyl fluoride, 1 μg/mL aprotinin, and 1 μg/mL leupeptin. After centrifugation for 30 min at 4°C, the concentration of protein extracts were measured by using bicinchoninic acid protein assay. In some cases, cultured cells were fractionated into nuclear and cytoplasm fractions as previously reported with a slight modification (Schreiber et al. 1990). In brief, cells were harvested with hypotonic buffer [10 mM HEPES-KOH (pH 7.9), 10 mM KCl, 1.5 mM MgCl2, 1 mM dithiothreitol, 0.1% Nonidet P40, and protease inhibitor stock III (535140, Calbiochem)] and centrifuged (7200 g, 5 min, 4°C) to obtain cytoplasmic fraction. The pellet was resuspended with extraction buffer [10 mM HEPES-KOH (pH 7.9), 400 mM KCl, 0.1 mM EDTA, 25% glycerol, and protease inhibitor stock III (535140, Calbiochem)] and incubated on ice for an hour. Samples were then centrifuged 12 000 g for 10 min, at 4°C and the supernatant was collected as nuclear fraction. After preparation, samples were run through an 8–12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred to nitrocellulose membrane and western blot analysis was performed as described (Jeon et al. 2011b). Appropriate protein bands were detected by enhanced chemiluminescence (Amersham, Buckinghampshire, UK).


Cortical neurons on poly-d-lysine pre-coated cover glasses (Fisher Scientific, Pittsburgh, PA, USA) were fixed with 4% paraformaldehyde (PFA) and incubated at 4°C overnight with an appropriate primary antibodies such as NeuN {1 : 500 diluted in blocking buffer [1% bovine serum albumin, 5% fetal bovine serum contained in phosphate-buffered saline (PBS)]}, FMRP (1 : 500 diluted in blocking buffer), p24 (1 : 500 diluted in blocking buffer). Next day, samples were incubated with an appropriate secondary antibody (TMRE or FITC conjugated, 1 : 500 diluted in blocking buffer), then mounted using Vectashield (Vector laboratories, Burlingame, CA, USA) and visualized with a confocal fluorescence microscope (TCS-SP; Leica, Heidelberg, Germany). Each treatment was performed in triplicate and assessment was performed on five regions of interest/coverslip selected at random.


Rat brains were perfused with 4% PFA and were cryosectioned using Cryocut microtome (CM-1850; Leica, McHenry, IL, USA) to a thickness of 40 μm and incubated with specific primary antibodies such as NeuN (1 : 500) and FMRP (1 : 500) in blocking buffer overnight. After washing three times with PBS, samples were incubated with an appropriate secondary antibody (1 : 500 diluted with blocking buffer) at around 20°C for 2 h. Samples were mounted using Vectashield (Vector laboratories), dried another 24 h, and observed with a confocal fluorescence microscope (TCS-SP; Leica).

Silencing of FMRP expression

shRNA virus were purchased as glycerol stocks (Sigma, SHGLY TRCN0000059759) and prepared using maxi-prep kit (QIAGEN, Valencia, CA, USA). ShRNA vector, packaging vector (Sigma), and FuGENE 6 (Roche) were mixed according to the manufacturer’s recommendation (Jeon et al. 2011b). The resulting shFmr1 virus targets Fmr1 gene (NM_002024) and the sequence composed of sense, loop, and antisense strands as follows:


As a control, non-target shRNA control vector with following sequence was used (SHC002, Sigma):


shRNA virus was used at 50 multiplicity of infection (MOI) for transduction in primary cortical neurons in serum-free media for 48 h (Jeon et al. 2011b) and in vivo microinjection into rat striatum. Briefly, anesthetized male rats were positioned in a Stoelting stereotaxic apparatus (51639; Stoelting Co, Wood Dale, IL, USA) and shRNA virus introduced into the striatum (+0.7 mm anteroposterior, ± 2.1 mm mediolateral, 5.0 mm dorsoventral from bregma, computed value according to brain MAPS) by microinjection using 10 μL Hamilton syringe. Either non-targeting shRNA virus (CV) or shFmr1RNA virus (FV) was introduced into the right hemisphere of striatum and PBS was introduced into left hemisphere of striatum as a control (1.5 μL and 0.5 μL/min). The injected regions were sealed and sterilized. Four days post injection, the middle cerebral artery occlusion (1 h) and 24-h reperfusion was performed on the either sham or shRNA virus-injected rats. For PCR and western blot analysis, the rats were anesthetized by ether, and decapitated, and the striatum was dissected. After dissociation, tissues were lysed using Trizol (PCR) and sample buffer (Western), respectively. For immunohistochemical analysis, the rats were anesthetized with ether and transcardially perfused with 4% PFA. Brain samples were post-fixed in 4% PFA for 24 h and equilibrated into 30% sucrose for 3 days prior to cryosectioning.

Over-expression of FMRP

Cells were transfected with eGFP-empty or eGFP-FMRP expression vectors using Lipofectamine 2000 (Invitrogen) in Opti-minimal essential medium for 6 h and recovered for additional 18 h as suggested by the manufacturer with slight modification (Jeon et al. 2011b). Cells were treated with 50 μM glutamate for 30 min and observed using a TCS-SP microscope (Leica) after propidium iodide (PI) staining or harvested for the examination of cellular proteins by western blot.

Cell viability measurement

MTT assay.  Primary cortical neurons were treated for indicated times and incubated with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) dye (5 mg/mL) at 37°C. After 30-min incubation, cells were solubilized with dimethylsulfoxide and the absorbance of the sample was read at 590 nm with a microplate reader (Molecular Devices, Sunnylvale, CA, USA). Data were expressed as the percentage (%) of control (untreated cells).

PI staining.  After treatment, cultured cortical neurons were incubated with PI (10 μg/mL, contained with 10 μg/mL RNase) for 5 min at around 20°C followed by fixation using ice-cold methanol for 0.5 h at −20°C and mounted. PI is a marker for cells that have lost their membrane integrity and has been used to identify apoptotic cells in vitro (Jeon et al. 2011b). PI-positive cells were counted in five random fields per sample by an individual blind to condition.

TUNEL assay.  After treatment, cells were fixed and permeabilized by pre-cooled ethanol :  acetic acid (2 : 1) for 5 min at −20°C as suggested by the manufacturer of the TUNEL assay kit (Millipore). After washing, equilibration buffer was applied and incubated for 10 s at around 20°C. Working strength TdT enzyme was added immediately and incubated in a humidified chamber at 37°C for 1 h. The reaction was stopped for 10 min and samples were incubated with anti-digoxigenin-conjugated rhodamine in a humidified chamber for 0.5 h at around 20°C. Specimens were mounted under a glass coverslip and observed. TUNEL-positive cells were counted in five random fields per sample by an individual blind to condition.

Annexin V cell death assay.  Annexin V cell death analysis was performed using a commercial kit (AN-1001; Phoenix Flow Systems Inc., San Diego, CA, USA). After glutamate treatment, cultured neurons were washed two times with cold PBS. Cells were resuspended with 1X binding buffer and added with staining reagents (Annexin V and PI: 5 μL, respectively) as recommended by the manufacturer for 20 min at around 20°C. After staining, 400 μL of binding buffer was added to each tube and samples were measured using a flow cytometer (FACSCalibur; BD, San Jose, CA, USA). Results were analyzed using FlowJo program (Tree Star Inc, Ashland, OR, USA).


Data are expressed as the mean ± SEM and analyzed for statistical significance by using one-way anova followed by Newman–Keul’s post hoc test and a p-value < 0.01 was considered significant.


Phosphorylation of Akt is a prerequisite in promoting cell viability after glutamate stimulation via transient induction of FMRP

To induce neuronal cell death, we treated cultured rat primary cortical neuron with 50 μM glutamate (Dong et al. 2009). Consistent with previous findings, glutamate induced phosphorylation of both PI3K and Akt within 10 min (Fig. 1a) and longer treatment induced neuronal death as determined by Annexin V staining, MTT assay (Figure S1), TUNEL staining (Fig. 3b), and PI staining (Figure S3). Level of cleaved caspase 3, which is often associated with cell death, was also increased by glutamate treatment (Fig. 1b). Surprisingly, glutamate also induced a rapid induction of FMRP protein that had a remarkably similar time-course to the activation (phosphorylation) of PI3K and Akt (Fig. 1c). At 3 h post glutamate treatment, cellular FMRP expression started to gradually decrease until 24 h. Pre-treatment of the cells with Akt inhibitor IV to block Akt activation resulted in a concentration-dependent inhibition of the glutamate-induced FMRP induction (Fig. 1d). As the PI3K and Akt pathway is well known for their cellular pro-survival role (Cantley 2002; Liu et al. 2010), it was not surprising that Akt inhibitor IV pre-treatment also increased cell death following glutamate treatment (Fig. 1e) and decreased the level of the pro-survival protein Bcl-xL (Fig. 1f). These results suggest that Akt phosphorylation is required for the induction of FMRP protein and the regulation of neuronal survival following glutamate treatment. Treatment of a PI3K inhibitor, LY294002, or another Akt inhibitor, Akt inhibitor VIII, also inhibited glutamate-induced expression of FMRP and Bcl-xL, and increased cell death (Figure S2a–f). As a downstream target of PI3K-Akt signaling, we examined Glycogen synthase kinase-3 beta (GSK3-β) phosphorylation after glutamate treatment, which is also known as a modulator of cell survival, differentiation, and proliferation in neural system (Liang and Slingerland 2003; Bhat et al. 2004). Consistent with the role of GSK3-β as a downstream target of Akt signaling pathway in cellular survival system against neurological insult condition such as MCAO (de la Torre et al. 2012; Ye et al. 2012), GSK3-β was phosphorylated by glutamate and treatment with Akt inhibitors like Akt inhibitor IV and VIII inhibited glutamate-induced GSK3-β phosphorylation (Figure S2g–i). These results suggest that GSK3-β might be functional downstream of Akt signaling pathway involved in the regulation of cell survival response against glutamate-induced neurotoxicity.

Figure 1.

 Glutamate-induced alteration of PI3K/Akt signaling pathway and fragile X mental retardation protein (FMRP) induction in rat cortical neurons. Rat primary cortical neurons were treated with 50 μM glutamate for indicated times (a–c) or for 30 min (d, f) and were harvested for western blot analysis. For cell viability analysis in (e), cells were treated with 50 μM glutamate for 3 h. (a) Treatment of cortical neurons with 50 μM glutamate elicited a transient PI3K and Akt phosphorylation (Ser 473). (b) Caspase 3 expression was determined by western blot and the level of cleaved caspase-3, an apoptosis marker, was also determined. Graph indicates the ratio of cleaved versus total caspase 3. (c) Over the same time course a change in FMRP expression was also detected by western blot. (d) Pre-treatment of an Akt activation inhibitor IV (1, 5 μM, 1 h) decreased glutamate-induced FMRP expression in neurons. (e) Pre-treatment with Akt activation inhibitor IV 1 h before 50 μM glutamate treatment further decreased cell viability 3 h after glutamate treatment compared to glutamate alone. Cell viability was measured by MTT assay as described in method. (f) The expression level of pro-survival protein Bcl-xL was examined by western blot. Bar graph represents the means ± SEM. of the data from four separate experiments and β-actin was used as a loading control (c, d, and f). *Significantly different compared with control and #significantly different compared with glutamate-treated samples (< 0.01, = 4).

Induction of FMRP expression in striatum after transient focal cerebral ischemia

Brain injury, resulting in neuronal loss, was induced in male Wistar rats by MCAO. Fmr1 mRNA expression was measured by real-time PCR 1 h post reperfusion in striatum, an area directly impacted by the occlusion (Fig. 2a). Fmr1 mRNA levels spiked at 1 h (Fig. 2a), as did FMRP protein expression (Fig. 2b). This type of expression profile was not detected for other cellular proteins including β-actin, α-CaMKII, and histone H3 (Fig. 2b). Although the increase in FMRP protein was detected in both the cytosolic and nuclear fractions, the increase was more prominent in the nuclear fraction (Fig. 2b). Similar modest increase of FMRP in cytosolic compartments (Fig. 2c and d, arrow head) and larger increase in nuclear compartments (Fig. 2c and d, arrow) was also observed in histological examination of striatal tissues after MCAO. When we analyzed histology samples using Image J program (http://rsb.info.nih.gov/ij/, NIH, Bethesda, MD, USA), the increase in immunostaining intensity in MCAO brain was 409.0 ± 28.1% (= 3, < 0.01) as compared with control. In our immunohistochemistry data, not all neurons expressed FMRP. Although most of the neurons in striatum expressed FMRP, there was a small percentage of cells which did not express FMRP. The nuclear localization of FMRP might suggest a role of FMRP in mRNA processing such as binding, stability control, and transport.

Figure 2.

 Middle cerebral artery occlusion (MCAO)-induced fragile X mental retardation protein (FMRP) expression in rat striatum. (a) Following MCAO surgery and reperfusion for times indicated, the striatum was dissected and total RNA was extracted, reverse transcribed, and level of Fmr1 and GAPDH mRNA was determined by real-time PCR as described in methods (= 3). (b) At various times post reperfusion, the striatum was dissected. Western blots were performed using the cytosolic (cytosol) and nuclear fractions against FMRP, Histone H3, α-CaMKII, and β-actin. *Significantly different as compared with control (< 0.01, = 3). (c) In a separate group of rats, the brains were prepared for immunohistochemistry after 1-h reperfusion (see Methods). Samples were costained with FMRP (ab17772, abcam, green fluorescence) and the nuclear maker TOPRO3 (blue fluorescence). Sections were observed using a confocal fluorescence microscope (40× objective) with 3× digital zoom (LSM710, Carl Zeiss). Arrows indicate examples of colocalization of FMRP with nuclear marker and arrowheads represent some of FMRP staining in cytosolic space. Magnification of boxed region of merged picture was also shown in right-most panel. Data are representative of three independent experiments. Scale bar represent 20 μm. (d) Samples were costained against FMRP (red fluorescence) and the neuronal nuclear maker NeuN (green fluorescence). Arrows indicate colocalization of FMRP with nuclear marker and arrowheads represent FMRP staining majorly in cytosolic space. Data are representative of three independent experiments.

Inhibition of FMRP expression augmented neuronal apoptosis both in vitro and in vivo

To investigate the role of FMRP in the regulation of cell death, we modulated the level of cellular FMRP using shRNA virus. After Fmr1 small hairpin lentiviral infection (Fmr1 virus, FV), the expression of both Fmr1 mRNA and FMRP was almost completely eliminated in rat primary cortical neurons (Fig. 3a). In contrast, non-targeted shRNA virus (control virus, CV) had no effect on either Fmr1 mRNA or FMRP expression level (Fig. 3a).

Figure 3.

 shFmr1 virus-induced functional silencing of fragile X mental retardation protein (FMRP) aggravates cell death via reduced Akt phosphorylation. (a) Expression of Fmr1 mRNA (upper panel) and FMRP protein (lower panel) was evaluated by semiquantitative RT-PCR and western blot with GAPDH and β-actin as loading control, respectively. Transduction with lentiviral particle-harboring shRNA specific for Fmr1 (FV) demonstrated specific knockdown of Fmr1 mRNA and FMRP protein expression in primary cortical neurons. Non-target shRNA control virus (CV) was used as a control. Bar graphs represent mean ± SEM. (= 4). (b) In cultured cortical neurons, shRNA virus was transduced for 48 h as described above and cells were changed with fresh media. FV viral-transduced cells showed increased TUNEL staining after glutamate treatment (50 μM, 3 h) compared with CV-transduced neurons. Graph represents the quantification of the number of TUNEL-positive cells. (c) PI3-K and Akt phosphorylation status and (d) Bcl-xL and cleaved caspase 3 expressions in CV and FV-transduced neurons following glutamate application for 30 min. FV transduction decreased the glutamate-induced PI3-K and Akt phosphorylation response when compared with untreated (control) or CV-transduced cells treated with glutamate. Bar graph represents mean ± SEM. (e) Brain sections of striatum microinjected with shRNA were immunostained against FMRP (red) and p24 viral core protein (green). CV-injected rats showed abundant coexpression of FMRP and p24 viral core protein (white filled arrows, yellow cells). In FV-injected striatum, cells expressing p24 viral protein were devoid of FMRP expression (white empty arrows, green cells). (f) TUNEL staining of CV or FV-transduced sections of striatum after 1-h MCAO and reperfusion (1 h). Graph represents the quantification of TUNEL-positive cells. *Significantly different compared with con (con vs.glutamate) and #significantly different compared with CV (CV vs. FV) (< 0.01, = 4).

When we analyzed cell death using TUNEL staining, rat primary cortical neurons expressing FV but not CV showed increased basal and glutamate-induced cell death (Fig. 3b). Similarly, the number of PI-positive cell (arrow head) after glutamate treatment was increased by the loss of FMRP in cortical neurons (Figure S3a). Importantly, cells with high level of FMRP (arrow) was not usually PI positive even after glutamate treatment (Figure S3a), cells with no or low FMRP expression (arrow head) were mostly PI positive. We also examined cell viability by MTT assay after glutamate treatment (50 μM, 3 h). Consistent with TUNEL and PI-staining experiments, FV group (FMRP silenced) was more vulnerable to cell death (Figure S3b). Altogether, these results suggest that FMRP plays a protective role against glutamate-induced cell death in cultured neurons.

To determine if the PI3K signaling pathway is affected by FMRP knockdown, we next examined the status of PI3K-Akt phosphorylation and Bcl-xL expression in FV neurons. In cultured neurons with reduced FMRP expression (FV), the glutamate-induced activation of PI3K and Akt was reduced compared to control (no virus con- nontreated samples- or CV) (Fig. 3c). Interestingly, the expression of the pro-survival protein Bcl-xL was also decreased in FV group (Fig. 3d). In contrast, the level of pro-apoptotic cleaved caspase 3 was increased in glutamate-treated FV group compared with control or CV group (Fig. 3d).

In agreement with the in vitro experiments, stereotactic injection of FV shRNA into rat striatum effectively silenced FMRP expression (Fig. 3e). In striatal sections of CV brains, there is abundant double labeling of FMRP and p24 core viral protein, (white filled arrows, Fig. 3e, middle panel) suggesting that CV did not induce down-regulation of FMRP. However, in Fmr1 shRNA injected striatum (FV), numerous cells expressing p24 protein (green) were without detectable FMRP expression (white empty arrows, Fig. 3e right panel). There are cells still expressing FMRP (red), but only in cells without viral expression (green).

To confirm that knockdown of FMRP protein resulted in an increase in cell death in vivo, we injected FV or CV into the striatum prior to MCAO and analyzed apoptotic cell death by TUNEL staining. We observed the number of TUNEL-positive cells was higher in FV rats following MCAO than that of CV-injected MCAO rats (Fig. 3f). These results suggest that knockdown of FMRP expression adversely affects cell survival presumably via decreased Akt phosphorylation leading to the alterations in apoptotic modulators such as Bcl-xL and caspase-3.

Over-expression of FMRP enhanced cellular survival

If knockdown of FMRP results in enhanced cell death, it is reasonable assumption that over-expression of FMRP could spare neurons from cell death following toxic stimuli. Therefore, we next investigated the effect of over-expression of FMRP on cell survival. Over-expression of FMRP was confirmed by western blot (Fig. 4a). Rat primary cortical neurons over-expressing eGFP-FMRP were less sensitive to glutamate-induced cell death (Fig. 4b). Compared with eGFP-FMRP-expressing cells, eGFP-expressing cells showed 3.76 ± 1.12-fold more PI-positive cells, suggesting that over-expression of FMRP protected cells from glutamate-induced cell death. Consistent with the role of FMRP in the PI3K pathway following a toxic stimulus, over-expression of FMRP significantly increased the phosphorylation of both PI3K and Akt (Fig. 4c). The level of Bcl-xL was also significantly increased when FMRP was over-expressed, again suggesting a protective role for FMRP in cell survival (Fig. 4d).

Figure 4.

 Neurons over-expressing fragile X mental retardation protein (FMRP) were protected from glutamate stimulation. (a) After mock (Con), eGFP, or eGFP-FMRP transfection, cellular protein level of FMRP, GFP, and β-actin was detected by western blot. In right panel, high molecular weight, exogenous eGFP-FMRP expression was quantified by densitometry as compared with endogenous FMRP in con and eGFP group. Data represent the means ± SEM of the data from three separate experiments. #Significantly different as compared with control or eGFP-transfected cells (con or eGFP vs. eGFP-FMRP) (< 0.01, = 3). (b) Either eGFP or eGFP-FMRP-transfected cortical neurons in culture were treated with glutamate (50 μM) for 3 h and analyzed by PI staining. Cells with PI (red) and GFP (green) were costained (yellow cells, white arrows) indicating transfected cells undergoing cell death. Graph represents the relative number of PI-positive cells as compared with control within the population of GFP-expressing cells per field. (c, d) After treatment with 50 μM of glutamate for 30 min, eGFP- or eGFP-FMRP-transfected primary neurons were analyzed by western blot. Glutamate-induced PI3-K and Akt phosphorylation (c) or Bcl-xL expression (d) was enhanced in eGFP-FMRP-transfected cells compared with untransfected (con) or eGFP-transfected cells. *Significantly different from con (con vs. glutamate), #significantly different compared with eGFP-transfected cells (eGFP vs. eGFP-FMRP) (< 0.01, = 4).


FMRP has been implicated in various neurological diseases including FXS, autism, epilepsy, and attention deficit/hyperactivity disorder. As such, research efforts have been focused on understanding the function of FMRP during development and the regulation of synaptic protein expression. However, even in this relatively intensely studied field, the exact regulatory mechanisms and functions of FMRP, especially in the context of normal and stressed condition, have yet to be fully elucidated.

In the work presented here, we described a pro-survival function of FMRP in primary cortical neurons. Using loss- and gain- of functional analysis, we showed that FMRP promoted cell survival under basal states and protects against cell death following toxic stimuli. We employed two common methods to induce neuronal stress, excessive glutamate stimulation in vitro and ischemia in vivo. In both cases, cell death was amplified in the absence of FMRP and tempered following over-expression of FMRP.

After bath application of [(S)-3,5-Dihydroxyphenylglycine], which is a metabotropic glutamate receptor 1/5 agonist, the level of FMRP was biphasically regulated by ubiquitin-dependent proteolysis system and translational control of FMRP as well as other regulator proteins (Hou et al. 2006; Zhao et al. 2011). Similar biphasic regulation of FMRP level was also observed in our study, in this case with initial increase of FMRP until 1 h, followed by down-regulation to basal level, although it remains to be determined how it can be reconciled with the observed changes in the activity of PI3K-Akt pathway. The rapid induction of FMRP in our study may also suggest translational control mechanism of FMRP expression at least in the early stage of cellular stress. Actually, glutamate-induced rapid up-regulation of FMRP in cultured primary neuron was not affected by the pre-treatment of a transcriptional inhibitor actinomycin D, however, the pre-treatment of a translational inhibitor anisomycin significantly reduced FMRP expression in control and glutamate-stimulated cells (our unpublished results). Similarly, other researchers reported that mGluR-dependent rapid FMRP expression is regulated by translational control (Weiler et al. 1997; Narayanan et al. 2008), which might be regulated by activation of PI3K-Akt-mTOR pathway (Hou and Klann 2004).

Previously, Khalil et al. showed a modest but significant decrease in HeLa cell proliferation after Fmr1 siRNA-induced knockdown of FMRP (Khalil et al. 2008) and we reported a decrease in HeLa cell viability following Fmr1 siRNA in an etoposide-stimulated cell toxicity model (Jeon et al. 2011b). Similarly, normal hippocampal neuronal viability and development was reduced in Fmr1 knockout mice compared with wild type (Jacobs and Doering 2010), and neural stem cells from Fmr1 knockout mice showed increased cell death (Castren et al. 2005; Castren 2006). Taken together with our results, these data strongly implicate an additional role of FMRP as a pro-survival protein.

Regarding the pro-survival role, the possible involvement of FMRP in the Ras pathway was first described in lymphocyte of FXS patients, which showed decreased expression of Ras-GTPase-Activating protein SH3-domain-binding protein, an effector of Ras signal transduction pathway (Zhong et al. 1999). Ras-GTPase-Activating protein SH3-domain-binding protein is expressed in brain (Atlas et al. 2004) and affects proliferation, differentiation, and survival of cell (Zhang and Shao 2010). Hu et al. reported that Ras-PI3K-Akt signaling is muted in FMRP knockout animals, suggesting the possibility that FMRP regulates Ras-PI3K signaling pathway (Hu et al. 2008). FXS patients showed lower cancer incidence compared with normal subjects (Schultz-Pedersen et al. 2001), and it is speculated that altered cell death regulatory mechanism, possibly via misregulated Ras-PI3K-Akt signaling in FXS patients, might underlie the lower cancer incidence (Hu et al. 2008).

In this study, FMRP expression was regulated by PI3K or Akt activation. In turn, increased level of FMRP seems essential for the up-regulation of Akt activity as evidenced by shRNA and over-expression studies, which forms a positive feedback loop. It is unclear how transient increase in FMRP may lead to activation of PI3K in our condition. In Fmr1 knockout hippocampal CA1 cells, defective histamine or acetylcholine-induced activation of PI3K-Akt pathway has been reported, albeit with the increased basal Ras activity (Hu et al. 2008). Interestingly, the basal level of Ras-dependent phosphorylation of Akt between WT and Fmr1 knockout mice was not different, which suggests that the expression of FMRP might be important in the stimulus-dependent but not basal level of activation of PI3K, at least to the downstream of Ras pathway (Hu et al. 2008). Similarly, it has been reported that [(S)-3,5-Dihydroxyphenylglycine]-induced phosphorylation of the downstream effectors of PI3K, such as phosphoinositide-dependent protein kinase 1, Akt, mTOR, and ribosomal p70S6 kinase, but not the basal level of phosporylation is impaired in Fmr1 knockout mice (Ronesi and Huber 2008), which may result from the reduced mGluR-long Homer interaction in Fmr1 knockout mice. Although other researchers reported up-regulation of PI3K activity in Fmr1 knockout mice (Gross et al. 2010; Sharma et al. 2010), these results may suggest a hypothesis that transient increase in FMRP may increase stimulus-dependent coupling of PI3K to upstream regulators by mechanisms hitherto unknown. Recently, it was reported that PTEN, a PI3K inhibitor and its mRNA is a putative target of FMRP, was dephosphorylated in Fmr1 knockout mice, although it is not sufficient to modulate PI3K activity (Sharma et al. 2010). Whether transient induction of FMRP in our condition may affect PTEN activity or expression to sufficient extent to modulate PI3K activity remains to be determined. Another candidate molecule which might be mediating the regulation of PI3K-Akt pathway through FMRP is protein phosphatase 2A (PP2A). The mRNA for PP2A has been suggested as a target of FMRP (Waggoner and Liebhaber 2003) and FMRP inhibited the translation of PP2A to act as a negative regulator of PP2A (Castets et al. 2005). Considering the role of PP2A in the inhibitory regulation of Akt activity among others (see a recent review by Bononi et al. 2011), it is also possible that the changes in the level of FMRP in our condition may exert its effects on PI3K-Akt pathway by modulating the level and activity of PP2A, which again needs experimental evidences in the future.

In contrast to our results, other researchers reported excess PI3K activation in neurons from Fmr1 knockout mice (Gross et al. 2010; Sharma et al. 2010). At present, the reason underlying those different results are not clear. Because we used shRNA instead of cells derived from knockout animals, the most evident difference would be the transient and acute nature of the FMRP down-regulation in our condition as compared with knockout animals. In situations where long-term changes in FMRP expression happen (knockout mice and human patients), compensatory responses may occur to minimize the deleterious effects to cells. Second, although the exact types of subpopulation of cells more vulnerable to FMRP loss is not obvious from this study, susceptible cells might be lost already in FMRP knockout mice during development, which may make it difficult to observe massive cell death in FMRP knockout mice. Investigating such cell types, for example, neural progenitor cells in specified loci and developmental stage, would be an interesting topic.

Recently, FMRP has been associated with the regulation of cellular mitosis in neural stem cells. Luo et al. showed almost 52.0% increase in Bromodeoxyuridine (BrdU) (S phase marker) positive neural stem/progenitor cells in Fmr1 knockout animals and suggested the mechanism by translational control of cyclin D1 and cyclin-dependent kinase 4 (CDK4) (Luo et al. 2010; Callan and Zarnescu 2011; Guo et al. 2011), both of which are well-known partners of cell cycle progression enhancers. In contrast to above results, our results suggest that knockdown of FMRP in cultured neuron decreases survival of neurons. The innate property of neuron versus neural stem cell may explain the discrepancy. Unlike to neural stem cell, the overt activation of cell cycle progression in neuron, which is post-mitotic, has been implicated in ectopic cell death in pathological conditions including AD, PD, and stroke [for a review, see (Lopes and Agostinho 2011)]. Neurological insults such as glutamate excitotoxicity and stroke induce oxidative stress, which result in overt activation of cyclin-dependent kinase 5 (CDK5) and CDK4 in neurons. CDK4 over-expression (activation) led neurons to re-enter the cell cycle (from G1 to S and G2 phase). However, the re-entry into cell cycle does not reach the M phase and initiates degenerative process (apoptosis) by activating caspase 3 and pro-apoptotic factors of Bcl-2 (Lopes and Agostinho 2011). Therefore, the same molecular events mediated by down-regulation of FMRP, i.e. over-activation CDK5/4 may mediate mitosis in neural stem cell, but susceptibility to cell death in neuron. Interestingly, kinetic study of BrdU-positive cells in brain revealed that the survival of BrdU-labeled cells over 4-weeks span was lower in FMRP knockout mice suggesting that FMRP deficiency might reduce cell survival of young neurons (Luo et al. 2010), which is consistent with our results. Taken together, these results suggest that the loss of FMRP expression may reduce cell survival of (young) neuron, although it may induce cell proliferation of neural stem cells.

Even though the primary localization of FMRP is cytoplasm in immunocytochemical staining or western blot with biochemically fractionated samples (Devys et al. 1993; Feng et al. 1997), it contains a functional, non-classical NLS near its N terminus (Eberhart et al. 1996; Sittler et al. 1996; Bardoni et al. 1997) with occasional nuclear localization (Feng et al. 1997; Zhang et al. 2007; Kim et al. 2009). FMRP-GFP- or eGFP-FMRP-transfected cells showed strong nuclear localization with bound mRNAs (De Diego Otero et al. 2002) and shuttles between cytoplasm and nucleus. In an experiment using a series of mutant with NLS and nuclear export signal, it has been suggested that FMRP primarily binds target mRNAs in nucleus although it is effectively transported out by interaction with RNAs and Tap/NXF1, a bulk mRNA exporter (Kim et al. 2009). Silencing of Tap/NXF1 increased nuclear localization of eGFP-FMRP (Kim et al. 2009). These results raise interesting possibility that either retention of RNAs in the nucleus or modulation of FMRP interacting proteins such as Tap/NXF1 or nuclear FMRP-interacting protein 1 (Bardoni et al. 2003) may retain FMRP in nucleus. Recently, nuclear retention and survival role of other RNA-binding proteins has also been suggested. RNA-binding proteins, cytoplasmic polyadenylation element binding (CPEB) 1, 3, and 4, are accumulated in nucleus of neurons after treatment with ionotropic glutamate receptor agonists in intracellular calcium and alpha-CAMKII-dependent manner (Kan et al. 2010). Among those CPEB isotypes, CPEB4 has been suggested to be important in cell survival during focal ischemia in vivo and oxygen–glucose deprivation in vitro (Kan et al. 2010). With these several intriguing possibilities including the role of PI3K/Akt pathway in the regulation of nuclear localization of FMRP in mind, we are actively investigating the mechanism and role of nuclear FMRP translocation in the regulation of neuronal survival.

In conclusion, our experimental results suggest that physiological regulation of FMRP may enhance cellular survival against neurotoxic stimulations such as ischemic stroke. Investigating whether endogenous expression of FMRP may perform neuroprotective roles against other neurodegenerative disorders might provide a better understanding and therapeutic targeting of these diseases.


This work was supported by a grant (2011-0014258) from National Research Foundation of Korea.