Deficits in the activity of presynaptic γ-aminobutyric acid type B receptors contribute to altered neuronal excitability in fragile X syndrome

The behavioral and anatomical deficits seen in fragile X syndrome (FXS) are widely believed to result from imbalances in the relative strengths of excitatory and inhibitory neurotransmission. Although modified neuronal excitability is thought to be of significance, the contribution that alterations in GABAergic inhibition play in the pathophysiology of FXS are ill defined. Slow sustained neuronal inhibition is mediated by γ-aminobutyric acid type B (GABAB) receptors, which are heterodimeric G-protein-coupled receptors constructed from R1a and R2 or R1b and R2 subunits. Via the activation of Gi/o, they limit cAMP accumulation, diminish neurotransmitter release, and induce neuronal hyperpolarization. Here we reveal that selective deficits in R1a subunit expression are seen in Fmr1 knock-out mice (KO) mice, a widely used animal model of FXS, but the levels of the respective mRNAs were unaffected. Similar trends of R1a expression were seen in a subset of FXS patients. GABAB receptors (GABABRs) exert powerful pre- and postsynaptic inhibitory effects on neurotransmission. R1a-containing GABABRs are believed to mediate presynaptic inhibition in principal neurons. In accordance with this result, deficits in the ability of GABABRs to suppress glutamate release were seen in Fmr1-KO mice. In contrast, the ability of GABABRs to suppress GABA release and induce postsynaptic hyperpolarization was unaffected. Significantly, this deficit contributes to the pathophysiology of FXS as the GABABR agonist (R)-baclofen rescued the imbalances between excitatory and inhibitory neurotransmission evident in Fmr1-KO mice. Collectively, our results provided evidence that selective deficits in the activity of presynaptic GABABRs contribute to the pathophysiology of FXS.

The behavioral and anatomical deficits seen in fragile X syndrome (FXS) are widely believed to result from imbalances in the relative strengths of excitatory and inhibitory neurotransmission. Although modified neuronal excitability is thought to be of significance, the contribution that alterations in GABAergic inhibition play in the pathophysiology of FXS are ill defined. Slow sustained neuronal inhibition is mediated by ␥-aminobutyric acid type B (GABA B ) receptors, which are heterodimeric G-protein-coupled receptors constructed from R1a and R2 or R1b and R2 subunits. Via the activation of G i/o , they limit cAMP accumulation, diminish neurotransmitter release, and induce neuronal hyperpolarization. Here we reveal that selective deficits in R1a subunit expression are seen in Fmr1 knock-out mice (KO) mice, a widely used animal model of FXS, but the levels of the respective mRNAs were unaffected. Similar trends of R1a expression were seen in a subset of FXS patients. GABA B receptors (GABA B Rs) exert powerful pre-and postsynaptic inhibitory effects on neurotransmission. R1a-containing GABA B Rs are believed to mediate presynaptic inhibition in principal neurons. In accordance with this result, deficits in the ability of GABA B Rs to suppress glutamate release were seen in Fmr1-KO mice. In contrast, the ability of GABA B Rs to suppress GABA release and induce postsynaptic hyperpolarization was unaffected. Significantly, this deficit contributes to the pathophysiology of FXS as the GABA B R agonist (R)baclofen rescued the imbalances between excitatory and inhibitory neurotransmission evident in Fmr1-KO mice. Col-lectively, our results provided evidence that selective deficits in the activity of presynaptic GABA B Rs contribute to the pathophysiology of FXS.
Fragile X syndrome (FXS) 2 is the most common cause of inherited intellectual disability and the leading monogenic cause of autism spectrum disorders. Approximately one in every 4,000 males and one in every 6,000 females are diagnosed with FXS, and the patients suffer from developmental, cognitive, and neuropsychological complications throughout life (1,2). FXS originates from the silencing of the fragile X mental retardation 1 (FMR1) gene, caused by hypermethylation of the expanded CGG trinucleotide repeats in the 5Ј-untranslated region (UTR) of the gene (3)(4)(5)(6). FMR1 codes for fragile X mental retardation protein (FMRP), an RNA-binding protein essential in regulating protein translation and synaptic plasticity (7)(8)(9)(10)(11)(12). FMRP binds to numerous mRNAs important in neuronal network physiology, including calcium/calmodulin-dependent protein kinase type II ␣ chain, amyloid precursor protein, postsynaptic density protein 95 (DLG4), ␥-aminobutyric acid receptor subunits ␦ and ␤-1, and myelin basic protein (13).
Our understanding of the pathophysiology FXS has been greatly facilitated by the development of a mouse model that lacks expression of the FMR1 protein (Fmr1-KO) (14,15). Fmr1-KO mice share several phenotypes with FXS patients such as altered spine morphology, cognitive deficits, hyperactivity, sensory hypersensitivity, repetitive behaviors, and macroorchidism (15,16). Studies from Fmr1-KO mice have revealed profound alterations in excitatory neurotransmission that are thought to primarily arise from excessive activity of group 1 metabotropic glutamate receptors (mGluR1/5) (17). Increased mGluR1/5 signaling is believed to enhance the removal of ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors from excitatory synapses. This process leads to exaggerated long term depression, a well studied model of synaptic plasticity, which is believed to contribute to the deficits in dendritic morphology seen in FXS (9).
In addition to compromised glutamatergic neurotransmission, deficits in the efficacy of inhibitory neurotransmission are also implicated in FXS, leading to the concept that imbalances in the efficacy of excitatory and inhibitory neurotransmission (E/I balance) drive the behavioral and anatomical deficits (18 -22). Slow prolonged synaptic inhibition in the brain is mediated by GABA B Rs, which are heterodimeric G-protein-coupled receptors composed of GABA B R1 and GABA B R2 subunits that activate G i/o signaling pathways. Their activation leads to diminished neurotransmitter release and increased neuronal hyperpolarization (23)(24)(25). Two major isoforms of GABA B R1 are found in mammals, R1a and R1b (26). R1a-containing GABA B receptors are enriched predominantly in axonal terminals of glutamatergic neurons, whereas postsynaptic GABA B Rs contain R1b subunits (27)(28)(29)(30)(31). There has been considerable interest in the use of the selective GABA B R agonist baclofen as a treatment for FXS (32)(33)(34)(35)(36). However, it remains to be determined whether deficits in GABA B R signaling directly contribute to the pathophysiology of FXS.
In this study, we have compared GABA B R expression levels and function in the hippocampus of Fmr1-KO mice. We demonstrate selective down-regulation of the GABA B R1a isoform in Fmr1-KO mice and post-mortem hippocampi from FXS patients. This deficit compromised the ability of (R)-baclofen to reduce excitatory neurotransmission without compromising its ability to modify GABAergic neurotransmission. Furthermore, (R)-baclofen was able to alleviate the deficits in E/I balance seen in Fmr1-KO mice. Collectively, our results provide a rationale why correcting deficits in GABA B R signaling may be a useful therapeutic strategy in FXS.

Inactivation of the Fmr1 gene in mice leads to selective deficits in GABA B R1 subunit expression
To address whether alterations in GABA B receptor signaling contribute to the pathophysiology of FXS, we used Fmr1-KO mice maintained on a C57BL/6 background (14). We first assessed GABA B receptor expression levels in the hippocampus of male Fmr1-KO and age-matched male wild-type (WT) controls (Fig. 1, A and B) using immunoblotting. In SDS-soluble extracts, an antibody specific to GABA B R1 (University of California Davis/National Institutes of Health NeuroMab, N94A/49) detected two bands, R1a and R1b, respectively.
The efficacy of GABA B R signaling is subject to dynamic modulation via direct phosphorylation of serine residues 783 and 892 in the R2 subunit (37). Immunoblotting with the respective phosphospecific antibodies revealed that phosphorylation of both residues was comparable between WT and Fmr1-KO mice (Fig. 3, A and B).
As a means of further analyzing GABA B R expression in Fmr1-KO mice, hippocampal sections were stained with a GABA B R1 antibody, as specified, and imaged by confocal microscopy. Within CA1 (WT ϭ 69.4 Ϯ 12.2 and Fmr1-KO ϭ 41.6 Ϯ 2.7, t test, p Ͻ 0.01, n ϭ 18 -21 cells) and the dentate gyrus (WT ϭ 65.5 Ϯ 9.1 and Fmr1-KO ϭ 39.3 Ϯ 1.6, t test, p Ͻ 0.01, n ϭ 52-70 cells) decreases in the maximum expression level of the GABA B R1 subunit were seen in Fmr1-KO mice, consistent with the phenotype identified by immunoblotting (Fig. 3, C-F). Thus, Fmr1-KO mice exhibit selective deficits in GABA B R1a expression compared with WT mice.

Deficits in GABA B R expression are evident in post-mortem tissue from FXS patients
To examine the translational significance of our measurements in mice, we compared the expression levels of GABA B Rs in posterior hippocampal tissue from three FXS patients and three controls ( Fig. 3; male, Caucasian, and 72-86 years old). Two FXS patients showed weak expression of R1a isoform but comparable R1b isoform expression relative to control individuals (Fig. 4A, red square). One FXS patient did not show any change in R1a or R1b expression. GABA B R2 subunit expression was similar across the subjects (Fig. 4B). Quantitative PCR revealed that levels of the R1a, R1b, and R2 subunit mRNAs were unaltered in FXS patients (Fig. 2B). The respective patient demographics are shown in Fig. 4D. Given the limited availability of post-mortem tissue, our results suggest that, in common with Fmr-1-KO mice, a similar trend of decreased GABA B R1 subunit is seen in a subset of FXS patients.

GABA B R-dependent modulation of excitatory transmission is reduced in Fmr1-KO mice
GABA B Rs exert profound pre-and postsynaptic effects on neuronal excitability via the activation of G i/o . More specifically, they reduce neurotransmitter release via inhibiting voltage-gated Ca 2ϩ channels, whereas postsynaptically they induce hyperpolarization via the activation of G-protein-activated inwardly rectifying potassium (GIRK) channels (24,25).
To assess whether Fmr1-KO mice exhibit impaired GABA B R signaling, we first examined their ability to modulate glutamatergic neurotransmission. To do so, we examined the effects of the GABA A R agonist (R)-baclofen on the properties of evoked excitatory postsynaptic currents (eEPSCs) in the CA1 region of the hippocampus following activation of the Schaffer collateral pathway (Fig. 5A). Consistent with published studies, (R)baclofen decreased eEPSC amplitude in WT mice (Fig. 5B). Interestingly, in Fmr-1KO mice, 5 M (R)-baclofen was less effective in suppressing eEPSCs (Fig. 5C) as the residual current observed in the presence of this drug was significantly higher than that seen in WT ( Fig. 5D; WT ϭ 0.13 Ϯ 0.01 and Fmr1-KO ϭ 0.27 Ϯ 0.04, t test, p Ͻ 0.01, n ϭ 3 animals per group, 11-31 recordings from each animal). To control for the specificity of the effects seen with (R)-baclofen, we assessed whether the ability of adenosine A1 receptors to modulate eEPSCs was altered in Fmr1-KO mice. In common with GABA B Rs, A1 receptors modulate eEPSCs via activation of G i/o (38,39). In contrast to (R)-baclofen, the effects of adenosine on eEPSCs were comparable between genotypes (Fig. 5D).
Collectively, our results suggest that the ability of GABA B Rs to modulate excitatory neurotransmission is compromised in Fmr1-KO mice, which correlates with reduced R1a subunit expression as shown in Fig. 1. Consistent with our results, R1adeficient mice show selective deficits in the ability of GABA B Rs to modulate excitatory neurotransmission (27,28).

The ability of GABA B Rs to modulate inhibitory neurotransmission is comparable in WT and Fmr1-KO mice
To assess whether Fmr1-KO mice exhibit global deficits in GABA B R signaling, we compared the ability of baclofen to modulate inhibitory neurotransmission between genotypes. First we compared the ability of presynaptic GABA B Rs to modulate inhibitory neurotransmission between strains. To do so, we compared the effects of baclofen and adenosine to regulate the evoked inhibitory synaptic currents (eIPSCs) in CA1 neurons following activation of the Schaffer collateral pathway. The ability of baclofen and adenosine to modulate eIPSCs was comparable between genotypes (Fig. 6, A-C; WT (R)-Bac ϭ 0.36 Ϯ 0.07 and Fmr1-KO (R)-Bac ϭ 0.38 Ϯ 0.09, t test, p ϭ 0.82, n ϭ 6 -7 slices). Next we analyzed the efficacy of postsynaptic GABA B R signaling by comparing the ability of (R)-baclofen to activate GIRK channels. (R)-Baclofen exerted similar effects in WT and Fmr1-KO mice (Fig. 6, D and E; WT ϭ 3.35 Ϯ 0.57 pA/picofarad and Fmr1-KO ϭ 3.31 Ϯ 0.59 pA/picofarad, t test, p ϭ 0.96, n ϭ 5 slices). Thus, in contrast to our observations on excitatory neurotransmission, the ability of GABA B R to modulate inhibitory synaptic transmission was unaltered in Fmr1-KO mice.

Comparing the properties of spontaneous EPSCs and IPSCs and their modulation by baclofen in Fmr-1-KO mice
To ascertain the significance of our findings with GABA B Rs for the pathophysiology of FXS, we compared the properties of spontaneous excitatory and inhibitory synaptic currents in CA1 neurons from WT and Fmr1-KO mice using voltage clamp recordings (Fig. 7). Spontaneous EPSCs (sEPSCs) were recorded at Ϫ70 mV in the presence of 50 M picrotoxin, whereas spontaneous inhibitory postsynaptic currents (sIPSCs) were measured at the same potential in the presence of 50 M 2-amino-5-phosphonopentanoic acid and 10 M 6,7-dinitroquinoxaline-2,3-dione. sEPSC frequency was significantly lower in Fmr1-KO mice ( Fig. 7A; WT ϭ 6.69 Ϯ 0.11 Hz and Fmr1-KO ϭ 6.04 Ϯ 0.10 Hz, Dunn's post hoc test, p Ͻ 0.01, n ϭ 12-14 cells). This decrease in sEPSC frequency is consistent with the reductions in presynaptic GABA B R-dependent modulation of eEPSCs and increased PPF seen in Fmr1-KO mice (Fig. 4). In addition to this, Fmr1-KO mice exhibited lower sIPSC frequency ( We also compared sEPSC and sIPSC amplitudes between genotypes. Fmr1-KO mice exhibited larger sEPSC amplitudes ( Finally, the relative charge transfer of spontaneous activity was determined by multiplying peak event area by frequency (43). The ratio of sEPSC/sIPSC charge transfer or "E/I balance" was then compared between genotypes (Fig. 7E). Changes in both frequency and amplitude contributed to a net increase of E/I balance charge transfer in Fmr1-KO mice ( Fig. 7E; WT ϭ

Discussion
In this study, we compared GABA B R expression levels and signaling in the brains of Fmr1-KO mice. In this model of FXS, we establish that GABA B R1a subunit protein expression was significantly decreased in multiple brain regions. In contrast, the expression levels of GABA B R1b and GABA B R2 subunits were comparable between genotypes. Significantly, the levels of the mRNAs encoding GABA B Rs were comparable in Fmr1-KO mice with those in WT, suggesting that FMRP does not act to modify the stability and or transcription of the respective gene transcripts. Interestingly, similar trends in GABA B R1 expression were seen in a small sample of post-mortem brain samples obtained from male FXS patients. However, the paucity of suitable human tissue precluded further investigation of these observations.
Evidence accrued from isoform-specific knock-out mice strongly suggests that the sushi domains at the N terminus of GABA B R1a subunits are critical in determining the accumulation of heterodimeric GABA B Rs on presynaptic glutamatergic terminals. GABA B Rs on these structures are believed to play a key role in limiting glutamate release and in modulating synaptic plasticity. In keeping with our biochemical deficits, the ability of (R)-baclofen to suppress eEPSCs was decreased in Fmr1-KO mice. Likewise, a selective deficit in PPF was observed in Fmr1-KO mice. These deficits are likely to arise from reduced levels of GABA B R expression and not GPCR effector coupling as the ability of adenosine receptors to suppress EPSCs was not compromised in the Fmr1-KO mice. GABA B Rs composed of R1b/R2 are found on GABAergic presynaptic specializations and postsynaptic sites adjacent to dendritic spines where they regulate GABA release and neuronal hyperpolarization, respectively. In agreement with our biochemical studies, the ability of GABA B Rs to modulate eIPSCs and induce GIRK channel-mediated hyperpolarization was comparable in Fmr1-KO and WT mice. The frequency of spontaneous synaptic activity in Fmr1-KO mice did not change upon (R)-baclofen application, further demonstrating deficient GABA B R modulation in axonal termi-nals. Thus, collectively, our results suggest that Fmr1-KO mice have selective deficits in the function of presynaptic GABA B Rs on principal neurons. This in turns leads to a reduction in the ability of (R)-baclofen to regulate excitatory neurotransmission.
It is widely believed that alterations in E/I balance are central to the behavioral and anatomical deficits seen in FXS. Therefore, we examined whether the deficits in GABA B R signaling seen in Fmr1-KO mice impact these respective parameters. Consistent with our biochemical and electrophysiological measurements, which suggest selective deficits in GABA B R receptor signaling in glutamatergic nerve terminals, the deficits in E/I balance are reversed by (R)-baclofen. This highly stable and potent GABA B R agonist may be more effective in activating the remaining populations of GABA B Rs in Fmr1-KO compared with the endogenous neurotransmitter GABA. Consistent with this notion, there is evidence that GABA levels are also diminished in FXS (1)(2)(3)(4)20). Collectively, these results suggest that the deficits in R1a subunit expression levels and subsequent signaling seen in our experiments contribute to altered neuronal excitability seen in FXS. In keeping with this view, it is interesting to note that R1a knock-out mice share some phenotypes in common with Fmr1-KO mice. In particular, both strains exhibit cognitive deficits, social avoidance, and sleep disturbance (27, 44 -49). Thus, some of the phenotypes seen in FXS are likely to arise from deficits in the activity of GABA B Rs containing R1a subunits.
The mechanisms by which FMRP impacts GABA B R function remain to be determined. However, our results strongly suggest that FMRP does not modulate the expression levels of the mRNAs encoding these receptors. A study using a cross-linking-immunoprecipitation assay revealed that FMRP binds to GABA B R1 and GABA B R2 in the mouse brain; however, this result has been difficult to reproduce in subsequent studies (13,50). It is also emerging that GABA B R activity itself may play a role in promoting FMRP expression in neurons (51). FMRP is enriched in the soma and dendrites via its interaction with the messenger ribonucleoprotein complex (10). Less is understood about the role FMRP plays in axonal terminals; however, it is implicated in axon guidance and pre- synaptic short term plasticity (10,(52)(53)(54). It has been estimated that upward of 30% of the presynaptic proteome is potentially regulated by FMRP (55). However, it is unclear which of these potential binding partners influence the function of presynaptic GABA B Rs in glutamatergic nerve terminals. It may be possible that FMRP binds directly to GABA B Rs to regulate their functions as similar modulation of big potassium (BK) channels by FMRP has recently been reported in CA3 pyramidal neurons (56).
In summary, our results highlight the importance of GABA B R dysfunction in FXS. Further investigation of the mechanism by which GABA B Rs are modulated or controlled by FMRP could aid the development of improved therapeutic strategies that target GABA B Rs for the treatment of FXS.

Animals
Fmr1-KO mice were originally purchased from The Jackson Laboratory (B6.129P2-Fmr1tm1Cgr/J) and then bred in house (homozygous female ϫ hemizygous male). Fmr1-KO and WT C57BL/6 mice were housed under constant temperature and humidity on a 12-h light/dark cycle with standard rodent food and water ad libitum. Male mice were used in the current study. All animal protocols were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by Institutional Animal Care and Use Committee of Tufts University.

Human tissue
Post-mortem hippocampal brain tissues from FXS patients and control subjects were gifted from the Lieber Institute for Brain Development (Baltimore, MD). For the control subjects, consent for the use of post-mortem human brains was obtained through the Lieber Institute for Brain Development under Maryland Department of Health and Mental Hygiene protocol number 12-24 and Western Institutional Review Board protocol number 20111080. Clinical characterization, macroscopic and microscopic neuropathological examinations, toxicological analyses, and quality control measures were performed as described previously (59,60). Each subject was diagnosed retrospectively by two board-certified psychiatrists according to criteria in the Diagnostic and Statistical Manual of Mental Disorders, 4th Ed. A psychiatric narrative summary was compiled for every case from a combination of data from a telephone screening on the day of donation with next of kin; from police, autopsy, and toxicology reports; from psychiatric records; from family informant interviews with next of kin (a psychological autopsy interview and the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, 4th Ed.); and from interviews with psychiatric treatment providers (59). The healthy controls had no known history of psychiatric illness or substance abuse or dependence and were screened for drug intoxication at time of death. The information about neuroleptic and antidepressant medications in subjects with psychiatric disorders was acquired from toxicology testing as described previously (60). The

Immunohistochemistry
Five-week-old male mice were intracardially perfused with 4% (w/v) paraformaldehyde. Brains were extracted and immersed in 30% (w/v) sucrose for 72 h. Coronal sections of 40-m thickness were prepared, and free-floating sections were rinsed in BupH-PBS (Thermo Fisher Scientific), blocked for 1 h in 10% (w/v) BSA with 0.3% Triton X-100 in PBS, and incubated overnight at room temperature with the indicated antibodies. The following day, the tissue was rinsed in PBS and incubated for 2 h in Alexa Fluor 488 secondary antibodies (Life Technologies, Thermo Fisher Scientific). After extensive rinsing with PBS, tissues were mounted, dried, and placed on coverslips with ProLong Gold antifade reagent (Invitrogen). Images were acquired with a Nikon Ti microscope and analyzed using NIH ImageJ. Images were recorded using the same microscope settings for each treatment and were taken within the linear range as detailed previously (61). Controls were obtained by incubating slices in the absence of primary antibody. Data were presented as mean Ϯ S.E.

Western blotting
Hippocampal, prefrontal cortical, and midbrain tissues were dissected from isoflurane-anesthetized Fmr1-KO and WT mice (3-16 weeks), flash frozen, and stored at Ϫ80°C. Alternatively, post-mortem hippocampal tissues from FXS and control subjects were used. On the day of experiment, the brain tissue was homogenized in lysis buffer containing 20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 5 mM EDTA, 10 mM NaF, 2 mM Na 3 VO 4 , and 10 mM sodium pyrophosphate plus 1:1000 protease inhibitors 4-(2-aminoethyl)benzenesulfonyl fluoride, antipain, aprotinin, leupeptin, and pepstatin. After incubating for an hour at 4°C, the brain lysate was centrifuged at 13,200 rpm at 4°C for 15 min. Using standard Western blotting techniques, the supernatant was subjected to SDS-PAGE (8 -10% gel) and transferred to a nitrocellulose membrane overnight at 4°C. The membrane was blocked with 5% nonfat milk in PBST for an hour. Subsequently, it was incubated in a specific primary antibody in 5% milk in PBST. After rigorous washes, a specific horseradish peroxidase-conjugated secondary antibody (in 5% milk in PBST) was incubated with the membrane for an hour at room temperature. The blots were developed using enhanced chemiluminescence systems (Amresco and Thermo Fisher Scientific), imaged using the charge-coupled device-based LAS 3000 system (FujiFilm), and quantified with NIH ImageJ software. Data were presented as mean Ϯ S.E.