|Home | About | Journals | Submit | Contact Us | Français|
Fragile X syndrome (FXS), the most common inherited form of intellectual disability and prevailing known genetic basis of autism, is caused by an expansion in the Fmr1 gene that prevents transcription and translation of fragile X mental retardation protein (FMRP). FMRP binds to and controls translation of mRNAs downstream of metabotropic glutamate receptor (mGluR) activation. Recent work identified striatal-enriched protein tyrosine phosphatase (STEP) as an FMRP target mRNA. STEP opposes synaptic strengthening and promotes synaptic weakening by dephosphorylating its substrates, including ERK1/2, p38, Fyn, Pyk2, and subunits of NMDA and AMPA receptors. Here we demonstrate that STEP translation is dysregulated in Fmr1KO mice, resulting in elevated basal levels of STEP with a concomitant loss of mGluR-dependent STEP translation. We hypothesized that the weakened synaptic strength and behavioral abnormalities reported in FXS may be linked to excess levels of STEP. To test this hypothesis, we reduced or eliminated STEP genetically in Fmr1KO mice. In addition to attenuating audiogenic seizures and seizure-induced c-Fos activation in the periaqueductal gray, genetically reducing STEP in Fmr1KO mice reversed characteristic social abnormalities, including approach, investigation, novelty-induced hyperactivity and anxiety. Loss of STEP also corrected select non-social anxiety-related behaviors in Fmr1KO mice, such as open arm exploration in the elevated plus maze. Our findings indicate that genetically reducing STEP significantly diminishes seizures and restores social and non-social anxiety-related behaviors in Fmr1KO mice, suggesting that strategies to inhibit STEP activity may be effective for treating patients with FXS.
Individuals with fragile X syndrome (FXS) exhibit behavioral abnormalities that include intellectual disability, anxiety, hyperactivity, and seizures (Boyle & Kaufmann, 2010). They are also impaired socially, with 30% meeting the criteria for autism (Harris et al., 2008). The majority of cases are attributed to a CGG expansion in the Fmr1 5′ untranslated region (Kaufmann & Reiss, 1999) that becomes hypermethylated, leading to transcriptional silencing and ultimately diminished expression of fragile X mental retardation protein (FMRP) (Chen et al., 1995). FMRP binds select mRNAs and suppresses their translation (Siomi et al., 1993, Darnell et al., 2001, Laggerbauer et al., 2001, Zalfa et al., 2003, Darnell et al., 2011). In the absence of FMRP, translation of some of these mRNAs is upregulated (Zalfa et al., 2003, Lu et al., 2004, Hou et al., 2006, Westmark & Malter, 2007, Gross et al., 2010).
Huber and colleagues (2002) report that Fmr1 knockout (Fmr1KO) mice show exaggerated metabotropic glutamate receptor (mGluR)-dependent long-term depression (LTD), suggesting a role for FMRP in regulating protein synthesis-dependent synaptic plasticity. These results mark the inception of the mGluR theory of FXS, linking dysregulation of proteins normally suppressed by FMRP to mGluR-LTD. This theory posits that many FXS phenotypes, including elevated protein synthesis, increased glutamate receptor endocytosis, immature spine development, and behavioral abnormalities, originate in exaggerated mGluR signaling (Bear et al., 2004, Krueger & Bear, 2011). Consequently, genetic and pharmacological approaches that decrease mGluR signaling attenuate several of these abnormalities in Fmr1KO mice (Chuang et al., 2005, Yan et al., 2005, Dolen et al., 2007, De Vrij et al., 2008, Osterweil et al., 2010, Hays et al., 2011, Thomas et al., 2011).
Accordingly, another major focus in FXS research has been to discover alternative therapeutic targets by identifying mRNAs regulated by FMRP and/or mGluR-dependent signaling (Krueger & Bear, 2011). One excellent candidate is striatal-enriched protein tyrosine phosphatase (STEP), a brain-enriched phosphatase that regulates proteins associated with the post-synaptic density and downstream signaling, including NMDA receptors (NMDARs), AMPA receptors (AMPARs), Fyn, Pyk2, and the MAPK family of proteins ERK1/2 and p38 (Goebel-Goody et al., 2011). STEP has been coined an “LTD protein” because it is translated in response to mGluR activation and promotes mGluR-stimulated AMPAR endocytosis (Luscher & Huber, 2010, Zhang et al., 2008). In addition, STEP opposes synaptic strengthening by promoting NMDAR internalization and inactivating ERK1/2, Fyn, and Pyk2 (Nguyen et al., 2002, Pelkey et al., 2002, Paul et al., 2003, Snyder et al., 2005, Venkitaramani et al., 2011).
Cross-linking studies of FMRP-RNA complexes demonstrate that STEP mRNA binds to FMRP (Darnell et al., 2011). We hypothesized that STEP translation would be dysregulated in Fmr1KO mice and that excess STEP levels might contribute to behavioral phenotypes characteristic of Fmr1KO mice (Musumeci et al., 2000, Peier et al., 2000, Spencer et al., 2005, Yan et al., 2005, Liu & Smith, 2009, Mines et al., 2010, Spencer et al., 2011). We demonstrate that genetically modulating STEP levels corrects select behavioral abnormalities Fmr1KO mice, indicating STEP as viable target in FXS.
All experimental protocols were approved by the Yale University Institutional Animal Care and Use Committee and strictly adhered to the NIH Guide for the Care and Use of Laboratory Animals. Every effort was made to minimize the pain, discomfort, and quantity of mice used in this study. STEPKO mice (Venkitaramani et al., 2009) were crossed with Fmr1KO mice (courtesy of Dr. William T. Greenough, University of Illinois Urbana-Champaign) to generate STEP heterozygous (HT)/Fmr1HT female breeders, as well as STEPHT/Fmr1KO male breeders (Fig. 2a). In all crossings, animals were on the c57Bl/6 background. Breeders from different litters were mated to produce progeny with a selective reduction in STEP or Fmr1. Unless otherwise noted, mice were weaned and housed in groups of 2–5 in standard vented-rack cages in a 12:12 hour light:dark cycle with food and water available ad libitum. Mice weaned to solitary housing were excluded from the study.
Primers used to detect the presence or absence of the WT or KO (neomycin) allele on the Fmr1 gene were as follows: 5′ WT primer-GGTTAAAAGTCATCCGTGGCTA-3′, 5′ KO primer-CTGAGCCCAGAAAGCGAA-3′, 5′ common primer-ACCACCACTGCCCTTCTGAT-3′. Fmr1 primers were combined and the following multi-primer PCR reaction used: 94°C for 5 min; 35 cycles of 94°C for 45 sec, 60°C for 1 min, 72°C for 1 min; final extension at 72°C for 10 min. Fmr1 products were electrophoresed on a 2% agarose gel to detect bands at 500 bp (KO), 350 bp (WT), or both (HT) (Fig. 2b). Primers used to detect the presence or absence of the WT or KO (neomycin) allele on the STEP gene were as follows: 5′-CCCTACTCTCATTCCTCCCTTCCC-3′, 5′ KO primer-CCACCAAAGAACGGAGCC-3′, 5′ common primer-GGCAGCAGATGCTGGTGGC-3′. STEP primers were combined and the following multi-primer PCR reaction used: 94°C for 5 min; 36 cycles of 95°C for 45 sec, 59°C for 1 min, 72°C for 1 min; final extension at 72°C for 10 min. STEP products were electrophoresed on a 2% agarose gel to detect bands at 400 bp (WT), 200 bp (KO) or both (HT) (Fig. 2b).
The antibodies, dilutions used, and their sources are listed in Table 1. Normal horse serum, ABC Vectastain Elite Kit, and DAB Peroxidase Substrate Kit were purchased from Vector Laboratories (Burglingame, CA, USA). Heparin was purchased from Hospira (Lake Forest, IL, USA) and Histoclear was obtained from National Diagnostics (Atlanta, GA, USA). (S)-3,5-Dihydroxyphenylglycine hydrate (DHPG) was obtained from Sigma-Aldrich.
Synaptoneurosomes (SNs) were obtained using a modified protocol (Hollingsworth et al., 1985). Hippocampi from 4–8 month STEPWT/Fmr1WT or STEPWT/Fmr1KO (three mice pooled for n=1) were homogenized with 15 passes in all-glass tissue grinders in ice-cold HEPES buffer containing protease inhibitors (Roche Applied Science, Indianapolis, IN, USA) and the following reagents (in mM): 124 NaCl, 3.2 KCl, 1.06 KH2PO4, 26 NaHCO3, 1.3 MgCl2, 2.5 CaCl2, 10 glucose, 10 HEPES (pH 7.4). Homogenates were passed through two 100 μm nylon mesh filters (Millipore) and then one 5 μm nitrocellulose filter (Millipore). Filtrates were spun at 1000 × g for 10 min at 4°C. Supernatants were collected for cytosolic fractions, and the resulting SN pellets were rinsed with HEPES buffer and centrifuged at 1000 × g for 10 min at 4°C. Supernatants were discarded and SNs pellets resuspended in HEPES buffer. After splitting in half, SNs were pre-incubated for 10 min at 37°C. One half was treated with 50 μM DHPG, while the other remained untreated (control), for 10 min at 37°C. After stimulation, SNs were lysed by adding SDS to a final concentration of 1%, sonicated, boiled for 5 min, and frozen at −80°C until further analysis. Protein concentrations were determined using a BCA protein assay kit (Pierce, Rockford, IL, USA) with bovine serum albumin (BSA) as a standard.
Proteins (10–40 μg) were resolved by 8 or 10% SDS-PAGE, transferred to polyscreen polyvinylidene difluoride membranes, blocked in 3% (w/v) BSA in Tris-buffered saline (140 mM NaCl, 20 mM Tris, pH 7.6) plus 0.1% Tween 20 (TBST) for 1 hr (RT), and incubated overnight at 4°C with primary antibodies. Following washes, membranes were incubated in secondary antibodies for 1 hr (RT). Immunodetection was accomplishes using chemiluminescence (SuperSignal Kits, Pierce) and visualized with a G:BOX imaging station using the GeneSnap image program (Syngene, Frederick, MD, USA). Blots were quantified using Image J software (version 1.43u; NIH). Protein signals were normalized to the average signal of all lanes on the same blot as described (Osterweil et al., 2010). This value was divided by the normalized GAPDH signal in the same lane and expressed as the mean of the normalized values ± s.e.m. All biochemical results in the text refer to STEP61, which is the membrane-associated and only isoform of STEP present in cortical and hippocampal tissue (Boulanger et al., 1995, Oyama et al., 1995, Bult et al., 1996).
Two cohorts of age-matched, sexually naïve mice were used for testing: cohort 1 for audiogenic seizures and c-Fos analysis, and cohort 2 for the behavioral battery. Each animal in cohort 2 was subject to the same number and order of tasks. The sequence of tasks was as follows: open field, marble burying, elevated plus maze, social choice test, social dominance tube test, and light/dark box. Details of the behavioral parameters analyzed and main effect differences are presented in Tables 2–5. Tests were conducted between the hours of 08:00 and 17:00 and, unless otherwise noted, the room was brightly lit with a modest and constant white background noise (60 dB). With the exception of audiogenic seizures, marble burying, and social dominance tube test, ANY-maze video tracking software (Stoelting, Wood Dale, IL, USA) was utilized with an overhead video camera system to automate behavioral testing and provide unbiased results. Investigators remained blind to the genotypes of mice during testing. All components of the behavioral apparatuses were wiped thoroughly with MB-10 (200 ppm) and 70% ethanol between tests to disinfect and eliminate odor trails between mice.
For audiogenic seizures (AGS), it was imperative that the mice were born and reared in static cages in a quiet room. Mice were weaned, tagged and tissue obtained for genotyping analysis at least 24 hrs prior to testing. At PND 20–24, male and female mice were habituated for 30 sec to a plexiglass test chamber (40 cm diameter × 60 cm high encased in sound-proof foam). The audio stimulus consisted of a 130 dB siren (modified personal alarm, PAL-1, Lane Self Defense, Roseville, MN, USA) powered by a DC converter and mounted on the chamber lid. Seizures were graded by two observers across a range of increasing intensity from no response to wild running, clonic/tonic seizures, or status epilepticus with respiratory arrest. The stimulus duration lasted until seizure onset or up to 5 min (whichever was first). No differences were observed between male and female mice in the incidence of AGS (data not shown); therefore the data were pooled together for statistical analysis.
Total activity and center square exploration of male mice (2–5 months old) were assessed in an empty open field box constructed of white, opaque plexiglass (50 cm long × 50 cm wide × 30 cm tall) for 5 min. The center square was designated as the center 25% of the total area.
Male mice (2–5 months old) were assessed for repetitive behavior using a modified version of the marble burying assay (Thomas et al., 2009). After completing open field, mice were singly housed in static cages containing Alpha-Dri bedding (2.5 cm) for 48 hrs. On the test day, total activity was first assessed in home cages without marbles for 5 min. The mouse was briefly removed and 20 black glass marbles (15 mm diameter) were evenly arranged in a 4 × 5 pattern on the bedding. The mouse was returned to the cage and allowed to freely explore for 20 min. Upon completion, an overhead picture was captured. The percentage of each visible marble was quantified using ImageJ (version 1.43u; NIH). Images were converted to grayscale, thresholded, and a constant cutoff value established whereby anything darker than that value was considered to be a visible marble (i.e., eliminating shadows). Scales were normalized to the width of the cage. The area of each marble was then measured and compared to the area of a completely visible marble (control). Those with a visible area of ≤ 60% of the control marble (i.e., buried at least 40%) were considered buried.
After completing marble burying, male mice (2–5 months old) were rehoused with their littermates in groups of 2–5 in vented-rack cages for at least 48 hrs before conducting elevated plus maze (EPM). The apparatus was comprised of four arms (30 cm long × 7 cm wide) configured in a plus manner situated 40 cm above the floor. The two opposing closed arms were constructed with 15 cm tall walls. This task was performed in dim light (6–7 lux). Mice were individually placed in the center square and allowed to freely explore for 5 min while the investigator left the room. Any mice that fell from the maze were excluded. Entries were defined as 4 paws (or 75% of its body) in the zone. Videos were retroactively scored by an investigator for the number of risk assessing and/or exploratory head dips while in the open arms.
Male mice (5–10 months old) were tested in a modified three-chambered social choice task as described previously (Moy et al., 2009). The three-chambered apparatus consisted of an opaque rectangular plexiglass box (64 cm long × 43 cm wide × 38 cm tall) divided into three compartments (20 cm long × 43 cm wide) by clear plexiglass walls with small, retractable entryways. Each side chamber contained an inverted wire cup (10.5 cm tall × 10.5 cm diameter bottom × 7.6 cm diameter top, 1 cm bar spacing). The task consisted of three 10 min phases: (1) habituation, (2) socialization, and (3) social novelty. During habituation, the test mouse was allowed to freely explore the entire apparatus, with the side chambers containing only cups. During socialization, a stranger mouse was placed under a cup in one side while the other remained empty. The test mouse was allowed to then freely explore the apparatus. Activity in each chamber and in close proximity to either the mouse or cup (front 25% of the mouse ≤ 1 inch margin around the cup) was recorded. Socialization index was calculated as the time in the mouse chamber divided by total time in the mouse and cup chambers. During social novelty, a novel stranger mouse was positioned under the remaining empty cup, while the stranger introduced during the socialization phase remained (designated as familiar). The test mouse was allowed to then freely explore the apparatus. Activity in each chamber and in close proximity to either the novel or familiar mouse was recorded. Novelty index was determined as the time in the novel mouse chamber divided by the total time in the novel mouse and familiar mouse chambers. All stranger mice were non-littermate, age-matched mice of varying genotypes that had been previously habituated to being enclosed under the cup for 10 min on the same day of testing.
Male mice (8–10 months old) were tested in a social dominance tube test as described previously (Spencer et al., 2005). In this test, mice were evaluated for their number of ‘wins’ against STEPWT/Fmr1WT mice. Immediately prior to the match, mice were singly house and individually habituated to a clear plastic tube (30.5 cm long × 3.8 cm inner diameter) contained within the open field box for 5 min. Matches began when two opposing mice were placed head first at either end of the tube and released simultaneously. Matches concluded when one mouse completely backed out and retreated from the tube (designated as the loser). Matches lasting longer than 2 min or where mice crawled over each other were excluded. Each mouse performed 3 matches against non-littermate, age-matched STEPWT/Fmr1WT mice.
Male mice (8–10 months old) were examined in a partitioned light/dark box (38 cm long × 38 cm wide × 36 cm tall) composed of two equally-sized adjoining plexiglass chambers where one side was black and the other was clear. The room was brightly lit so that the clear side was fully illuminated. Mice were individually placed in the illuminated chamber and allowed to freely explore for 10 min. The total number of entries in the light and dark sides and overall activity within the illuminated side was recorded. Entries were defined as 4 paws (or 75% of its body) in the zone.
Thirty minutes after exposure to the audio stimulus, mice were anesthetized by sodium pentobarbital (i.p. 100 mg/kg) and perfused transcardially with 0.1 M sodium phosphate buffer (PB; pH 7.4) containing 5% sucrose and 1 unit/ml heparin followed by 4% paraformaldehyde in PB. Brains were post-fixed in the same solution for 12 hrs, then cryoprotected by equilibration in 20% and 30% sucrose in PB, embedded in a 0.9% saline solution containing 20% BSA and 0.5% gelatin, polymerized with 4.6% glutaraldehyde, and frozen. Fifty μm thick coronal free-floating section were obtained and stored at −20°C in cryoprotectant solution (30% glycerol and 30% ethylene glycol in 0.1M PB). Unless otherwise noted, all subsequent steps were completed under constant agitation at RT. Every sixth section from the anterior hippocampus to the posterior inferior colliculus was selected, resulting in a stained series at 300 μm intervals. Sections were rinsed in phosphate buffered saline (PBS), submerged in 3% H2O2 in PBS for 20 min to quench endogenous peroxidase activity, blocked in 3% normal horse serum in PBS with 0.3% Triton X-100 (blocking buffer) for 30 min, and incubated for 72 hrs in blocking buffer at 4°C containing anti-c-Fos antibody (1:500). Sections were washed with PBS plus 0.3% Triton X-100 and incubated for 1 hr in blocking buffer containing biotinylated horse anti-goat IgG secondary antibody (1:300). Following washes, sections were incubated in ABC Vectastain Elite Kit for 60 min, washed, and reacted for 3 min in Tris buffer (pH 7.5) containing diaminobenzidine (DAB), nickel substrate, and hydrogen peroxide according to manufacturer’s recommendations (DAB Peroxidase Substrate Kit). Sections were then washed in PBS, mounted, air dried, and nuclei were stained with methyl green. Sections were then dehydrated in ethanol, defatted in Histoclear, and mounted in DPX.
Three sections containing the periaqueductal gray (PAG) were selected at intervals of 300 μm, based on the Paxinos mouse brain atlas, corresponding approximately to stereotaxic coordinate Bregma −3.80 mm. The PAG region within the brain sections was projected at a 200X magnification onto paper using a microprojector (Ken-a-Vision 1000, Ken-A-Vision Manufacturing Company, Inc., Kansas City, MO, USA) and the locations of all c-Fos+ neurons were plotted. An investigator who was blind to genotype and AGS response quantified the total c-Fos+ neurons in the PAG from three sections per animal.
All data were presented as means ± s.e.m. (with the exception of audiogenic seizures and social dominance tube test) and significance set at P ≤ 0.05 as determined by SPSS software (IBM, New York, NY, USA). A 2×2 factorial ANOVA [Fmr1 genotype (WT, KO) x treatment (control, DHPG)] was used to determine main effect differences in protein levels. A one-tailed, unpaired Student’s t-test was used to calculate treatment differences in SNs compared to STEPWT/Fmr1WT mice because changes in one direction only were expected based on previous results (Zhang et al., 2008). Two-tailed Chi2 Fisher’s exact tests were conducted to determine significant differences in audiogenic seizures (genotype vs. STEPWT/Fmr1WT or genotype vs. STEPWT/Fmr1KO) and the social dominance tube test (genotype vs. STEPWT/Fmr1WT). A one-way ANOVA (genotype + response) with LSD post hoc comparisons was used to calculate differences in the number of c-Fos+ cells. For all other behavioral tests, a 3×2 factorial ANOVA [STEP genotype (WT, HT, KO) x Fmr1 genotype (WT, KO)] was performed to determine main effects. Post hoc comparisons were made using LSD following confirmation of a significant main effect of STEP. Where indicated, a planned contrast one-way ANOVA (STEP genotype) was performed when mice were Fmr1KO to determine if genetically reducing STEP significantly altered behaviors of Fmr1KO mice.
A recent study utilized high-throughput sequencing of RNA isolated by crosslinking immunoprecipitation (HITS-CLIP) to rigorously identify endogenous mRNA targets of FMRP in neurons (Darnell et al., 2011). This work revealed a highly significant association of STEP mRNA (Ptpn5) with FMRP across 7 independent experiments (p=9.97E-5), including use of two different FMRP antibodies and multiple biological replicates. One function of FMRP is to suppress translation (Siomi et al., 1993, Darnell et al., 2001, Laggerbauer et al., 2001, Zalfa et al., 2003, Darnell et al., 2011). Accordingly, we predicted that STEP expression would be upregulated at baseline in STEPWT/Fmr1KO mice. We also assessed whether mGluR-stimulated translation of STEP was disrupted. Hippocampal synaptoneurosomes (SNs) from 4–8 month old STEPWT/Fmr1WT and STEPWT/Fmr1KO mice were prepared and stimulated with the mGluR agonist DHPG (50 μM, 10 min) and analyzed by immunoblotting (Fig. 1). We first verified enrichment of PSD-95 in the SNs fraction, and confirmed that, while expressed in both fractions, STEP was most enriched in the SN fraction (Fig. 1a). No significant main effects of genotype or drug treatment were observed in GAPDH expression (Fig. 1b).
Main effect analysis showed that STEP expression was significantly greater in SNs from STEPWT/Fmr1KO than STEPWT/Fmr1WT mice (F(1, 8) = 6.177, P = 0.038) (Fig. 1c). As reported previously (Zhang et al., 2008), we found that DHPG treatment significantly increased STEP expression in STEPWT/Fmr1WT SNs (P = 0.048), but not in STEPWT/Fmr1KO SNs (P = 0.454) (Fig. 1c). No significant differences were observed between unstimulated cytosolic fractions from STEPWT/Fmr1WT and STEPWT/Fmr1KO mice (Fig. 1d) (P = 0.207), demonstrating that removal of FMRP selectively increases basal STEP expression in SNs. This increase was not limited to the hippocampus, because STEP levels were also significantly higher in whole brain extracts from STEPWT/Fmr1KO mice as compared to STEPWT/Fmr1WT (P = 0.04, data not shown). These findings indicated that in Fmr1KO mice, basal STEP expression was enhanced while mGluR-induced translation of STEP was impaired.
As a positive control, we found that PSD-95 expression was significantly higher in SNs from STEPWT/Fmr1KO than STEPWT/Fmr1WT mice (F(1, 8) = 20.715, P = 0.002) (Fig. 1e), consistent with exaggerated basal translation of PSD-95. Previous studies reported that DHPG stimulation significantly increases PSD-95 expression in STEPWT/Fmr1WT, but not STEPWT/Fmr1KO, neurons (Todd et al., 2003, Muddashetty et al., 2007). Similarly, we observed a modest, but not statistically significant, DHPG-induced increase in PSD-95 expression in STEPWT/Fmr1WT SNs (P = 0.089), which was absent in STEPWT/Fmr1KO SNs (P = 0.217).
We hypothesized that excessive STEP levels contribute to the manifestation of abnormal behavioral phenotypes normally present in Fmr1KO mice. To test this prediction, we genetically reduced or eliminated STEP in Fmr1KO mice (Fig. 2a). Examination of both STEPHT/Fmr1KO and STEPKO/Fmr1KO afforded the opportunity to increase therapeutic relevance by testing whether a 50% or 100% reduction in STEP protein (Venkitaramani et al., 2009) corrects behavioral phenotypes. We also examined STEPHT/Fmr1WT and STEPKO/Fmr1WT mice because their performance in the tasks used in this study had not yet been evaluated.
We first tested whether genetically reducing STEP decreased audiogenic seizures (AGS) in Fmr1KO mice. As expected, none of the STEPWT/Fmr1WT mice had seizure activity in response to the audio stimulus, and neither did STEPHT/Fmr1WT or STEPKO/Fmr1WT mice (Fig. 3a, Table 2). Consistent with other reports (Musumeci et al., 2000, Yan et al., 2005, Dolen et al., 2007), 76.2% of STEPWT/Fmr1KO mice showed seizure activity (P = 0.0001, vs. STEPWT/Fmr1WT mice). This value dropped to 53.3% in STEPHT/Fmr1KO mice and 43.5% in STEPKO/Fmr1KO mice (Fig. 3a, Table 2). Of relevance, the incidence of AGS was significantly lower in STEPKO/Fmr1KO mice compared with STEPWT/Fmr1KO mice (P = 0.036), demonstrating that STEP deficiency significantly reduced AGS in Fmr1KO mice.
AGS are mediated through brainstem circuits involving the periaqueductal gray (PAG) (Faingold, 1999, Faingold & Randall, 1999, N’gouemo & Faingold, 1999). To evaluate possible genotypic differences in neuronal activity induced by the audio stimulus, we examined expression of c-Fos in the PAG shortly after stimulation (Fig. 3, b–h). In the absence of seizure-related behaviors, no differences in the number of c-Fos+ PAG cells were found (Fig. 3h). However, following a tonic/clonic seizure, c-Fos+ cells increased significantly in the PAG, regardless of genotype (P = 0.0001) (Fig. 3, b–h). STEPKO/Fmr1KO mice consistently showed fewer seizure-induced PAG c-Fos+ cells than STEPWT/Fmr1KO mice (P = 0.007) (Fig. 3, c–d, f–h), indicating that genetic elimination of STEP in Fmr1KO mice attenuated neuronal activation in the PAG when seizures were present.
We next investigated whether genetically reducing STEP altered socialization in Fmr1KO mice. One way socialization is evaluated is using a three-chambered social choice task (Crawley, 2007, Moy et al., 2009). There are three phases to this task: (1) habituation, (2) socialization, and (3) social novelty. A 3×2 factorial ANOVA was conducted to examine main effects of Fmr1 and STEP genotype on various parameters in these phases (Figs. 4–5, Table 2).
Mice are typically social beings that prefer to spend time sniffing and interacting with other mice (Murcia et al., 2005). During the socialization phase, no main effects of either Fmr1 or STEP were observed for the socialization index (Fmr1: F(1, 46) = 2.940, P = 0.093; STEP: F(2, 46) = 0.595, P = 0.556) (Table 3), indicating similar preference for the mouse. We also analyzed main effects of Fmr1 and STEP in close proximity to the mouse or the cup (Fig. 4, b–e; Table 3). Fmr1 and STEP did not impact the time spent in close proximity to the stranger mouse (Fmr1: F(1, 46) = 3.887, P = 0.055; STEP: F(2, 46) = 0.810, P = 0.451) (Fig. 4c, Table 3) or the cup (Fmr1: F(1, 46) = 0.083, P = 0.774; STEP: F(2, 46) = 0.394, P = 0.677) (Table 3).
Social approach during the socialization phase was determined by the number of entries in close proximity to the stranger mouse, a method providing similar measurements to number of nose contacts (Mines et al., 2010). Main effect analysis indicated that Fmr1KO mice enter the area in close proximity to the stranger mouse more times than Fmr1WT mice (F(1, 46) = 13.044, P = 0.001) (Fig. 4d, Table 3). When comparing Fmr1KO mice with varying levels of STEP, we found that STEPHT/Fmr1KO mice had significantly fewer visits with the stranger mouse than STEPWT/Fmr1KO mice (P = 0.008, planned contrast) (Fig. 4d). These results demonstrated that genetically reducing STEP significantly decreased the number of visits in close proximity, thereby reversing this Fmr1KO phenotype.
We also found a significant main effect of STEP genotype on the duration of each visit in close proximity to the mouse, where STEPHT and STEPKO mice spent significantly more time per entry than STEPWT mice (F(2, 46) = 4.50, P = 0.016) (Fig. 4e, Table 3). These results revealed that STEPHT and STEPKO mice maintained each social interaction longer. Altogether, data from the socialization phase indicated that STEPWT/Fmr1KO mice visited the stranger mouse more frequently, and genetically reducing STEP reversed this phenotype. Decreasing STEP levels in Fmr1KO mice also increased the duration of each visit, possibly enhancing the length of the interaction.
No significant differences in the total distance traveled were observed among genotypes in the socialization phase, thus excluding the possibility that other aspects were driven by locomotion differences (Fmr1: F(1, 46) = 1.906, P = 0.174; STEP: F(2, 46) = 1.689, P = 0.196) (Fig. 4b, Table 3). Moreover, we found that neither Fmr1 nor STEP had an impact on parameters with the cup alone or chamber during the socialization phase (Table 3). These findings indicated the observed differences were specific for the social aspect of the task.
We subsequently analyzed main effects of Fmr1 and STEP during the social novelty phase, where a novel mouse is introduced to the task (Fig. 5a). Striking main effect differences were observed in locomotion during this phase for both Fmr1 and STEP (Fig. 5b, Table 3). Fmr1KO mice traveled significantly more than Fmr1WT mice (F(1, 46) = 6.790, P = 0.012), indicating that Fmr1KO mice become hyperactive in the presence of the novel mouse and/or when two mice are in the apparatus. In contrast, STEPHT and STEPKO mice traveled significantly less compared to STEPWT mice (F(2, 46) = 5.871, P = 0.005), suggesting decreased activity. The net result was that genetically reducing STEP reversed the social novelty-induced hyperactivity normally observed in Fmr1KO mice.
No significant main effects of Fmr1 or STEP were observed for the novelty index (Fmr1: F(1, 46) = 0.004, P = 0.949; STEP: F(2, 46) = 0.129, P = 0.879) (Table 3), demonstrating similar preference for the novel chamber over the familiar chamber. Nonetheless, main effect analysis revealed that Fmr1KO mice spent significantly more time (F(1, 46) = 4.996, P = 0.030) (Fig. 5c, Table 3) and had more entries (F(1, 46) = 4.549, P = 0.038) (Fig. 5d, Table 3) in close proximity to the novel mouse than Fmr1WT mice. One likely explanation for these results was that the hyperactivity of Fmr1KO mice was causing them to spend more time and have more visits to the novel mouse. Neither Fmr1 nor STEP genotype impacted the duration of each entry for either (Fig. 5e, Table 3). For all parameters close to the novel mouse, no significant main effects of STEP were observed (Fig. 5, c–e; Table 3), demonstrating that genetically reducing STEP in Fmr1KO mice did not affect behaviors in close proximity to the novel mouse during this phase.
As an additional measure of social anxiety, we utilized the social dominance tube test (Lindzey et al., 1961, Spencer et al., 2005) (Fig. 6). As published elsewhere (Spencer et al., 2005), STEPWT/Fmr1KO mice won only 34.5% of the time (10/29 matches) (P = 0.035). The percentage of wins increased to 56.6% in STEPHT/Fmr1KO mice (13/23 matches), demonstrating that STEPHT/Fmr1KO and STEPWT/Fmr1WT mice retreat from the tube at a similar frequency (P = 0.556). Taken further, STEPKO/Fmr1KO mice won 76.9% of the time (10/13 matches) (P = 0.017). These findings revealed that genetically reducing STEP in Fmr1KO mice increased the number of wins in the social dominance tube test.
Given that the number of wins was greater for STEPKO/Fmr1KO mice than STEPWT/Fmr1WT mice, we next investigated whether loss of the STEP gene alone caused the mice to win more often. We found that STEPKO/Fmr1WT mice did not win significantly more than STEPWT/Fmr1WT (61.1% wins; 11/18 matches) (P = 0.318; data not shown), demonstrating that loss of STEP alone was not causing STEPKO/Fmr1KO mice to win more often. Altogether, these findings supported the hypothesis that genetically reducing STEP decreased social anxiety in Fmr1KO mice.
To address whether genetically reducing STEP also reversed non-social anxiety-related phenotypes in Fmr1KO mice, we first evaluated main effect differences in the elevated plus maze (EPM) (Fig. 7, Table 4). In agreement with studies showing reduced anxiety in Fmr1KO mice (Peier et al., 2000, Qin et al., 2002, Spencer et al., 2005, Qin & Smith, 2008, Mines et al., 2010, Yuskaitis et al., 2010, Spencer et al., 2011, Liu et al., 2011, but see also Restivo et al., 2005, Bilousova et al., 2009), main effect analysis revealed that Fmr1KO mice entered the open arms more times (F(1, 53) = 4.720, P = 0.034) (Fig. 7a, Table 4) and spent a greater percentage of total arm time in the open arms (F(1, 53) = 4.134, P = 0.047) (Table 4) than Fmr1WT mice. Conversely, main effect analysis showed that STEPKO mice entered the open arms significantly less than STEPWT or STEPHT mice, suggesting that STEPKO mice have greater anxiety in the EPM (F(2, 53) = 3.560, P = 0.035) (Fig. 7a, Table 4). The net result was that genetically eliminating STEP normalized the effect of Fmr1 genotype on open arm exploration. While the percentage of open arm time showed a similar pattern for a main effect of STEP, these results did not reach statistical significance (F(2, 53) = 0.725, P = 0.489) (Table 4).
We observed that neither Fmr1 nor STEP altered the number of total arm entries (Fmr1: F(1, 53) = 1.290, P = 0.261; STEP: F(2, 53) = 2.147, P = 0.127) (Fig. 7b, Table 4), total distance traveled (Fmr1: F(1, 53) = 0.044, P = 0.835; STEP: F(2, 53) = 1.075, P = 0.349) (Table 4), or number of closed arm entries (Fmr1: F(1, 53) = 0.136, P = 0.713; STEP: F(2, 53) = 1.207, P = 0.307) (Table 4) in the EPM. These findings demonstrated that the observed main effects of Fmr1 and STEP on open arm entries were specific for exploration of the open arms and were not accounted for by differences in overall locomotion.
Time spent in the center square of the EPM has been interpreted as decision making time and/or motivation for exploratory behavior (Cruz et al., 1994, Hilber & Chapillon, 2005, Rodgers & Johnson, 1995). Duration therefore is thought to reflect the animal’s choice to either approach or avoid conflict and/or their motivation to explore novel environments (Lister, 1990). We discovered that Fmr1KO mice spend a greater percentage of total time in the center square (F(1, 53) = 7.646, P = 0.008) and spend more time per entry (F(1, 53) = 7.337, P = 0.009) than Fmr1WT mice (Fig. 7, c–d; Table 4). Moreover, Fmr1KO mice performed more exploratory and/or risk assessing head dips than those that were Fmr1WT (F(1, 49) = 7.044, P = 0.011) (Fig. 7e, Table 4), providing additional support for increased exploratory behavior. In all of these parameters (Fig. 7c–e), STEP genotype did not have a significant effect on these parameters (Fig. 7, Table 4), suggesting that the reversal of open arm exploration in STEPKO/Fmr1KO mice (Fig. 7a) occurred without changes in behavioral indicators of decision making and/or exploration.
Corroborating our EPM results, there was a significant main effect of Fmr1 in the light/dark box, where the time (F(1, 49) = 11.225, P = 0.002) and total distance traveled (F(1, 49) = 6.431, P = 0.014) in the illuminated side was increased (Fig. 8, a–b; Table 4), suggesting reduced anxiety. Although no significant main effects of STEP were observed for these parameters, STEPHT/Fmr1KO mice tended to spend less time and travel less distance in the light side than STEPWT/Fmr1KO mice (Fig. 8, a–b), but these differences did not reach statistical significance (P = 0.061 and P = 0.072, respectively, planned contrast).
However, a significant main effect of STEP was observed for the number of total light and dark entries (F(2, 49) = 6.247, P = 0.004), a value comparable to number of transitions (Fig. 8c, Table 4). Specifically, STEPHT and STEPKO mice had significantly fewer total entries than STEPWT mice, suggesting increased anxiety following genetic reduction of STEP. Although light side exploration and total number of light/dark entries (i.e., transitions) are often highly correlated (Crawley & Goodwin, 1980), we saw no significant main effect of Fmr1 on the total number of entries (F(1, 49) = 2.216, P = 0.134) (Fig. 8c, Table 4).
Corroborating these results, center square exploration in the open field task indicated decreased anxiety in Fmr1KO mice (Table 5). Main effect analysis demonstrated that Fmr1KO mice had greater number of center entries (F(1, 73) = 4.975, P = 0.029), increased distance traveled in center (F(1, 73) = 5.580, P = 0.021), and increased center/total distance traveled ratio (F(1, 73) = 3.866, P = 0.05). In line with previous reports that Fmr1KO mice are more hyperactive in open field tests (Peier et al., 2000, Spencer et al., 2011), we found that the total number of center/border entries were significantly higher in Fmr1KO mice than those that were Fmr1WT (F(1, 73) = 4.654, P = 0.034); however, the total distance traveled was not different (F(1, 73) = 1.416, P = 0.238) (Table 5). Discrepancies in this measure could be accounted for based on methodological differences, as we assessed open field activity for only 5 min and other studies have used 30 min tests (Peier et al., 2000, Spencer et al., 2011). Of note, there were no significant main effects of STEP genotype on any parameter in the open field task (Table 5), demonstrating that genetically reducing STEP did not reverse behaviors in this particular task.
To determine if the abnormalities we observed in Fmr1KO mice were due to increased perseverative and/or repetitive behaviors, we next examined marble burying behaviors. While marble burying was initially characterized as a measure of anxiety (Njung’e & Handley, 1991), more recent studies indicate that it is a reflection of repetitive and/or perseverative behaviors (Thomas et al., 2009). In an effort to reduce inter-investigator variability and subjectivity, we devised a new method to quantitate the percentage of each marble buried. Nonetheless, we observed no significant main effects of Fmr1 or STEP on the number of marbles buried (Fmr1: F(1, 68) = 0.294, P = 0.590; STEP: F(2, 68) = 1.549, P = 0.220) (Table 5). These findings revealed that observed behavioral abnormalities in social and non-social anxiety-related domains in Fmr1KO mice were not due to increased perseverative and/or repetitive behaviors.
Much FXS research is currently focused on discovering new target proteins whose translation is controlled by GluRs and FMRP and dysregulated in Fmr1KO mice (Bear et al., 2004, Krueger & Bear, 2011). Here we describe and validate STEP as a novel target. We establish increased basal STEP expression and abolished mGluR-induced STEP translation in Fmr1KO mice. Genetic modulation of STEP corrects select behavioral abnormalities normally present in Fmr1KO mice. Specifically, Fmr1KO mice null for STEP exhibit fewer AGS and show a corresponding reduction in PAG neuronal activity following seizures. Moreover, most social and non-social anxiety-related abnormalities associated with Fmr1KO mice are reversed by genetically reducing STEP. Altogether, these findings highlight the benefit of reducing STEP in Fmr1KO mice and suggest that developing STEP inhibitors may be worthy of clinical consideration for FXS.
FMRP suppresses mRNA translation by stalling ribosomal translocation (Muddashetty et al., 2007, Darnell et al., 2011). Accordingly, the expression of some proteins encoded by FMRP target mRNAs is upregulated in Fmr1KO mice (Zalfa et al., 2003, Lu et al., 2004, Hou et al., 2006, Westmark & Malter, 2007, Gross et al., 2010). Given that STEP has been identified as a FMRP target (Darnell et al., 2011), we predicted that its translation would be dysregulated in Fmr1KO mice. We establish that STEP levels are basally elevated in synaptoneurosomal, but not cytosolic, fractions, suggesting a selective increase in dendrites. Additionally, mGluR-stimulated translation of STEP is absent in Fmr1KO mice, consistent with impaired mGluR-induced translation of FMRP targets (Zalfa et al., 2003, Lu et al., 2004, Hou et al., 2006, Muddashetty et al., 2007, Westmark & Malter, 2007, Park et al., 2008). This disruption is likely caused by improper recruitment of mRNAs to active polyribosomes in Fmr1KO mice (Weiler et al., 2004, Muddashetty et al., 2007).
We hypothesized that increased STEP levels contribute to synaptic deficits and behavioral abnormalities in Fmr1KO mice. STEP is found in dendrites and has emerged as critical modulator of synaptic plasticity and key player in several neuropsychiatric disorders (Boulanger et al., 1995, Pelkey et al., 2002, Zhang et al., 2008, Zhang et al., 2010, Goebel-Goody et al., 2011). Substrates include NMDARs, AMPARs, Fyn, Pyk2, ERK1/2 and p38 (Nguyen et al., 2002, Pelkey et al., 2002, Munoz et al., 2003, Paul et al., 2003, Snyder et al., 2005, Zhang et al., 2008, Venkitaramani et al., 2011). STEP-mediated dephosphorylation of Fyn, Pyk2, ERK1/2, and p38 inactivates these enzymes, whereas dephosphorylation of surface NMDARs and AMPARs promotes endocytosis. Thus, STEP appears to oppose synaptic strengthening and promote synaptic weakening. In particular, STEP is translated following mGluR stimulation and is required for mGluR-mediated AMPAR internalization (Zhang et al., 2008), suggesting a role in mediating mGluR-LTD (Moult et al., 2006). Given that STEP is among a handful of “LTD proteins” dysregulated in Fmr1KO mice (Zalfa et al., 2003, Lu et al., 2004, Hou et al., 2006, Westmark & Malter, 2007, Park et al., 2008), we propose that the complex interaction of STEP with other affected proteins in FXS leads to behavioral abnormalities.
Using a comprehensive battery of tests, we evaluated whether genetically reducing STEP ameliorates abnormalities in Fmr1KO mice. One consistent phenotype of Fmr1KO mice is their increased AGS susceptibility (Musumeci et al., 2000, Yan et al., 2005). Motivation for assessing AGS in our study stems from prior work showing a striking seizure resistance in STEPKO mice (Briggs et al., 2011). Here we establish that loss of STEP in Fmr1KO mice significantly reduces AGS. Notably, genetic reduction of mGluR5 and APP in Fmr1KO mice leads to similar magnitude changes in AGS frequency (Dolen et al., 2007) (Westmark et al., 2011), highlighting common signaling pathways that may exist for attenuating AGS. Although the mechanism behind this reduced susceptibility in STEPKO/Fmr1KO mice is still unknown, studies point to altered excitatory/inhibitory balance. Specifically, Fmr1KO mice exhibit decreased neocortical inhibition, leading to hyperexcitability (Gibson et al., 2008, Olmos-Serrano et al., 2010, Paluszkiewicz et al., 2011) and increased seizure susceptibility (Musumeci et al., 2000, Yan et al., 2005). In contrast, loss of STEP increases hippocampal inhibitory excitability and decreases excitation, thereby reducing seizure susceptibility (Briggs et al., 2011). Further experiments are required to determine whether normalized excitatory/inhibitory balance contributes to decreased AGS in STEPKO/Fmr1KO mice.
Seizure activity can be mapped by examining expression of the immediate early gene, c-Fos, a transcription factor that marks cellular activation (Morgan et al., 1987, Herrera & Robertson, 1996). We report that AGS significantly increase c-Fos activation in the PAG, a modulatory structure in the AGS efferent pathway (Faingold, 1999, Faingold & Randall, 1999, N’gouemo & Faingold, 1999, Ross & Coleman, 2000). Of significance, AGS-induced c-Fos staining is significantly less in STEPKO/Fmr1KO mice compared to STEPWT/Fmr1KO mice, suggesting attenuated PAG activation. Changes in c-Fos number and/or intensity are associated with different seizure stages and varying degrees of severity (Morgan et al., 1987), indicating that loss of STEP may reduce the severity of AGS in Fmr1KO mice. Alternatively, it is possible that patterning and neuronal output of AGS is modified in STEPKO/Fmr1KO mice. While it is likely that other subcortical auditory nuclei are affected (Le Gal La Salle & Naquet, 1990), our findings provide compelling evidence that loss of STEP significantly reduces AGS frequency and PAG activation in Fmr1KO mice.
Individuals with FXS exhibit socially anxious behaviors with abnormal approach and interaction (Wolff et al., 1989, Cornish et al., 2008, Harris et al., 2008). These impairments are attributed to anxiety and hyperarousal to social stimuli, since the motivation for the interaction is present but is too overwhelming to maintain (Cornish et al., 2008). Aberrant social behaviors are also found in Fmr1KO mice (Spencer et al., 2005, Mineur et al., 2006, McNaughton et al., 2008, Liu & Smith, 2009, Moy et al., 2009, Mines et al., 2010, Liu et al., 2011). Using the social choice task and social dominance tube test, we demonstrate increased social anxiety alongside heightened interest and hyperactivity in Fmr1KO mice that is reversed by genetically manipulating STEP. Our findings are in accord with Fmr1KO mice displaying more active social behaviors in direct interaction tests (Spencer et al., 2005) and spending less time per nose contact with mice in the social choice task (McNaughton et al., 2008, Mines et al., 2010). One speculation is that these abnormalities reflect hyperarousal to social stimuli concomitantly with social anxiety, as seen in FXS humans. Disparities between our work and those finding diminished social approach and novelty preference in Fmr1KO mice may be explained by the age of the mice and methodology used to quantify nose contacts (Liu et al., 2011, Liu & Smith, 2009, Moy et al., 2009).
Another well-characterized behavioral phenotype of Fmr1KO mice is reduced anxiety in non-social paradigms (Peier et al., 2000, Qin et al., 2002, Spencer et al., 2005, Qin & Smith, 2008, Mines et al., 2010, Yuskaitis et al., 2010, Liu et al., 2011, Spencer et al., 2011). Using three independent non-social anxiety tasks, we confirm that Fmr1KO mice are less anxious than Fmr1WT mice. An alternative interpretation is that Fmr1KO mice show increased exploratory behaviors in novel environments, since EPM center square time and number of head dips are increased and these behaviors appear to rely heavily on exploratory behavior (Rodgers & Johnson, 1995). Whether the abnormality associated with Fmr1KO mice is due to reduced anxiety and/or increased exploratory behavior, we find that loss of STEP in Fmr1KO mice reverses open arm exploration in the EPM, thereby normalizing this phenotype.
We also report novel social and non-social anxiety-related phenotypes of STEPHT and STEPKO mice. First, they maintain each social interaction longer in the socialization phase of the social choice task, so it is interesting to consider that the quality of each interaction is enhanced. Prior studies report that STEPKO mice exhibit more dominance behaviors in direct interaction tasks (Venkitaramani et al., 2011), so another explanation might be that increased aggression leads to longer visits. However, our results show that STEPKO/Fmr1WT and STEPWT/Fmr1WT mice retreat from the social dominance tube test at a similar frequency, reducing the likelihood of this possibility. Second, in both the EPM and light/dark box tasks, there are indications that loss of STEP causes mice to be more anxious in non-social paradigms and/or have decreased exploratory behavior. While the basis for these phenotypes remain unknown, several studies have found altered social, anxiety, and depressive-like behaviors following genetic modulation of some STEP substrates, such as NMDARs and ERK2, (Boyce-Rustay & Holmes, 2006, Satoh et al., 2011).
In conclusion, the present study establishes that genetically reducing STEP in Fmr1KO mice, either fully or partially, reduces AGS susceptibility and normalizes select social and non-social anxiety-related behaviors. Our results may therefore suggest a potential therapeutic role of STEP in reversing characteristic phenotypes of FXS. Further investigation is required to uncover the molecular basis and/or downstream pathways responsible for attenuating Fmr1KO behaviors when STEP is reduced. Moreover, because STEP is differentially modulated in several neuropsychiatric disorders (Goebel-Goody et al., 2011), future work must address how to inhibit STEP in particular brain regions affected in FXS.
We thank Dr. Pradeep Kurup, Dr. Niki Carty, and other laboratory members for helpful discussions and advice on biochemical experiments; Deborah Hall, Jeesun (Iris) Park, Carole Nasrallah, Faten Sayed, Linda Li, and Matthew Baum for assistance with behavioral experiments and genotyping of mice; and Nicholas Woods and Elizabeth Litvina for performing c-Fos quantification. We also wish to thank Dr. Michael Tranfaglia, Dr. Peter Olausson, Dr. Natalina Salmaso for advice on conducting and analyzing the behavioral experiments. This research was supported by a FRAXA Research Foundation Fellowship and NIH grant 5T32MH018268 to S.M.G.-G., and by NIH grants MH52711 and MH091037 to P.J.L. Support for J.R.N. was provided by XXXX.