ReportsNeuroscience

Thorase variants are associated with defects in glutamatergic neurotransmission that can be rescued by Perampanel

See allHide authors and affiliations

Science Translational Medicine  13 Dec 2017:
Vol. 9, Issue 420, eaah4985
DOI: 10.1126/scitranslmed.aah4985

Thorase ATPase, a multifunctional enzyme

Alterations in glutamatergic neurotransmission are thought to contribute to schizophrenia, a neuropsychiatric disease with multifactorial causes. Umanah et al. identified variants in the AAA+ ATPase Thorase in several patients with schizophrenia. They show in vitro that these variants disrupted the expression of glutamate receptors, altering their physiological function in mouse primary cortical neurons. Mice expressing these variants exhibited behavioral deficits. An FDA-approved drug, Perampanel, rescued these behavioral deficits, suggesting that Perampanel may be useful for treating diseases involving aberrant glutamatergic neurotransmission.

Abstract

The AAA+ adenosine triphosphatase (ATPase) Thorase plays a critical role in controlling synaptic plasticity by regulating the expression of surface α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs). Bidirectional sequencing of exons of ATAD1, the gene encoding Thorase, in a cohort of patients with schizophrenia and healthy controls revealed rare Thorase variants. These variants caused defects in glutamatergic signaling by impairing AMPAR internalization and recycling in mouse primary cortical neurons. This contributed to increased surface expression of the AMPAR subunit GluA2 and enhanced synaptic transmission. Heterozygous Thorase-deficient mice engineered to express these Thorase variants showed altered synaptic transmission and several behavioral deficits compared to heterozygous Thorase-deficient mice expressing wild-type Thorase. These behavioral impairments were rescued by the competitive AMPAR antagonist Perampanel, a U.S. Food and Drug Administration–approved drug. These findings suggest that Perampanel may be useful for treating disorders involving compromised AMPAR-mediated glutamatergic neurotransmission.

INTRODUCTION

ATAD1 encodes a protein we recently identified as Thorase, an AAA+ adenosine triphosphatase (ATPase) that plays a critical role in regulating the surface expression of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs). Thorase controls endocytosis and internalization of AMPARs through disassembly of the AMPAR/glutamate receptor–interacting protein 1 (GRIP1) complex, and thus contributes to the regulation of synaptic plasticity, and learning and memory (1). Altered glutamate signaling has recently been implicated in schizophrenia (27). ATAD1 is located on chromosome 10q23.31 (89,512,875 to 89,552,512 base pairs) in a region where our group has previously reported a strong linkage peak for schizophrenia in an Ashkenazi Jewish population (8). This region has been replicated in other independent studies (9) and has also been linked to bipolar disorder (10). We hypothesized that a good strategy to increase the probability of identifying functional variants in the ATAD1 gene would be to explore variants found in patients with schizophrenia. To do this, we used our cohort of 712 schizophrenia cases and 649 healthy controls of Ashkenazi Jewish origin.

RESULTS

We obtained bidirectional sequence data for the ATAD1 coding exons in 712 schizophrenia cases and 649 healthy controls and found three rare coding variants, all missense changes, in 5 individuals (fig. S1 and table S1). The variants R9H and D221H were present only in schizophrenia cases: R9H in a female and the D221H in a male. Another variant, E290K, was present in two male schizophrenia cases and one female control. Two of these variants, R9H and E290K, were seen at frequencies of 8.4 × 10−7 and 0.0001.1 × 10−4 in the Exome Aggregation Consortium (ExAC) database (exac.broadinstitute.org), which includes exomes of unscreened individuals. Whereas results from such small sample sizes do not provide proof that these variants may increase the risk for psychiatric disease, we considered that our strategy increased the odds that these variants may be implicated in schizophrenia and proceeded to examine their potential impact on the ATAD1 gene.

Thorase variants do not affect ATPase activity

The ATPase activity of wild-type Thorase and the variants R9H, D221H, and E290K was evaluated using purified recombinant His6 C-terminal–tagged proteins (fig. S1). Thorase E290K was not recognized by the anti-Thorase antibody, suggesting that the E290K Thorase variant has an altered epitope (fig. S1, B and C). Circular dichroism analyses suggested that there was no significant change in the structure of the R9H variant (fig. S1D). However, both D221H and E290K variants showed conformational changes. Both adenosine triphosphate (ATP)–binding (fig. S1E) and ATP hydrolysis (fig. S1, G and H) assays suggested that none of the variants affected ATPase activity.

Thorase variants D221H and E290K result in oligomer disassembly defects

Most AAA+ ATPase family members form oligomers, which are critical for assembling and disassembling protein complexes (8). To determine whether the Thorase variants formed oligomers upon ATP binding, glutaraldehyde cross-linking of purified Thorase was examined in the presence of either adenosine diphosphate (ADP) or ATP. Most of the ADP-bound Thorase was dimeric, whereas ATP-bound Thorase formed large oligomeric complexes of masses greater than 250 kDa (Fig. 1A). Upon ATP hydrolysis, the large Thorase oligomeric complexes disassembled into dimers. Wild-type Thorase and the Thorase variants all formed large oligomeric complexes. However, the Thorase variants D221H and E290K showed defects in disassembly (Fig. 1B).

Fig. 1. Thorase variants show defects in Thorase oligomer and GluA2-GRIP1 complex disassembly.

(A) Immunoblot analyses of oligomer formation by wild-type (WT) Thorase and the Thorase variants. The samples were cross-linked by glutaraldehyde in the presence of 1 mM adenosine diphosphate (ADP) or 1 mM adenosine triphosphate (ATP). The ATP-treated samples were incubated at 4°C for Thorase oligomer formation and then at 37°C for ATP hydrolysis to enable disassembly of the Thorase oligomers. Samples were collected every 15 min for 150 min (lanes 1 to 10) during the incubation for cross-linking. (B) Graphical representation of the percentages of the oligomer states of Thorase in different fractions collected at 15-min intervals. (C) Immunoblot analyses of FLAG-tagged Thorase immunoprecipitations (IP) of mouse primary cortical neurons from heterozygous Thorase-deficient mice expressing WT Thorase or the Thorase variants in the presence of different nucleotides. The samples were resolved on 10% SDS–polyacrylamide gel electrophoresis (PAGE) and immunoblotted with anti-Thorase antibody, anti-FLAG–horseradish peroxidase antibody, anti-GluA2 antibody, or anti–glutamate receptor-interacting protein 1 (GRIP1) antibody. (D) Normalized percent bound GluA2 and GRIP1 in the Thorase FLAG-tagged immunoprecipitated samples in (C). (E) Immunoblot analyses of GluA2 immunoprecipitation of mouse primary cortical neurons from heterozygous Thorase-deficient mice expressing FLAG-tagged WT Thorase or the Thorase variants in the presence of different nucleotides. The samples were separated on 10% SDS-PAGE and immunoblotted with anti-FLAG antibody, anti-GluA2 antibody, or anti-GRIP1 antibody. (F) Normalized percent bound Thorase and GRIP1 in the GluA2 immunoprecipitated samples in (E). (G) Percent of Thorase and GRIP1 disassembled (upon ATP hydrolysis) from the GluA2 complexes in (E). Means ± SEM (n = 3). *P < 0.10; n.s. (nonsignificant), P > 0.10, Holm-Sidak post hoc test compared with WT. Power (1 − β error probability) = 0.8 to 1.0.

Thorase is known to regulate AMPAR trafficking. Thus, we investigated the interaction of the Thorase variants with the AMPAR subunit GluA2 and GRIP1 in mouse primary cortical neurons (Fig. 1, C and D) using recombinant protein pulldown assays (fig. S2, A to F). The Thorase variant R9H bound to GluA2 and GRIP1 in a similar manner to wild-type Thorase; however, the D221H and E290K variants displayed defective binding to GluA2 and GRIP1 (Fig. 1, C and D). In the presence of ATP, wild-type Thorase and the Thorase variant R9H disassembled GRIP1 from GluA2, whereas the Thorase variants D221H and E290K showed impaired disassembly of GRIP1 from GluA2 (Fig. 1, E to G, and fig. S2, G to J).

Thorase variants reduce endocytosis of GluA2 in mouse primary cortical neurons

Several lines of evidence from neurodevelopmental, neuropathological, genetic, and behavioral pharmacological data indicate that AMPAR-mediated neurotransmission is compromised in patients with schizophrenia (9). Altered expression of the AMPAR subunit GluA2 in the prefrontal cortex (3, 5) and in the nucleus accumbens (4) has been reported in patients with schizophrenia. We therefore examined the surface expression of GluA2 in mouse primary cortical neurons expressing wild-type Thorase or the Thorase variants in vitro. There was an increase in GluA2 surface expression in neurons expressing the Thorase variants D221H and E290K compared to neurons expressing wild-type Thorase or the R9H variant (Fig. 2, A to C). This was further confirmed by surface protein bis(sulfosuccinimidyl)-suberate (BS3) cross-linking (Fig. 2, D and E) and biotinylation assays (fig. S3A). In response to N-methyl-d-aspartate (NMDA) and glycine, there was a significant internalization of surface GluA2 in wild-type Thorase–expressing neurons (P < 0.05), whereas neurons expressing the three Thorase variants displayed less internalization of surface GluA2 (Fig. 2, D and E).

Fig. 2. Thorase variants cause impaired GluA2 endocytosis and trafficking in mouse primary cortical neurons.

(A) Representative images of unstimulated mouse primary cortical neurons (control) or mouse primary cortical neurons stimulated with N-methyl-d-aspartate (NMDA) to induce GluA2 endocytosis. Mouse primary cortical neurons from heterozygous mice deficient in Thorase expressed green fluorescent protein (GFP) alone (Het-GFP; control), WT Thorase-GFP (Het-WT), or the Thorase-GFP variants R9H, D221H, or E290K. Lower panels are high-resolution images. Scale bars, 5 μm. (B) Quantification of the ratio of surface GluA2 (sGluA2) to internalized GluA2 (iGluA2) for mouse primary cortical neurons in (A). (C) GluA2 internalization index in (A) and (B) was measured as the ratio of the fluorescence intensities of iGluA2 to total GluA2 (iGluA2 plus sGluA2). (D) Immunoblot analyses of bis(sulfosuccinimidyl)-suberate (BS3) cross-linking of sGluA2 in mouse primary cortical neurons derived from heterozygous Thorase-deficient mice expressing FLAG-tagged WT Thorase (Het-WT) or the Thorase variants. The samples were separated on 4 to 12% gradient SDS-PAGE and immunoblotted with anti-GluA2 or anti-FLAG antibody. (E) Optical densitometry quantification of sGluA2 in (D). Means ± SEM (n = 3). *P < 0.10; n.s., P > 0.10, analysis of variance (ANOVA) with Holm-Sidak post hoc test when compared to Het-WT. Power (1 − β error probability) = 0.8 to 1.0. (F) Time trace of surface pH-GluA2 fluorescence changes in mouse primary cortical neurons in response to NMDA treatment. (G) Maximum amplitudes of pH-GluA2 fluorescence intensity changes in response to NMDA stimulation. (H) Average recycling half-life (T1/2), which is the time taken from maximum endocytosis to 50% recycling of GluA2. Means ± SEM (n = 7). ***P < 0.05; n.s., P > 0.10, ANOVA with Holm-Sidak post hoc test when compared with Het-WT (black). Power (1 − β error probability) = 1.0.

To further evaluate the effects of the Thorase variants on AMPAR trafficking, pH-sensitive green fluorescent protein (GFP) fused to the N-terminal extracellular domain of GluA2 was used to examine AMPAR distribution at the extracellular surface, as previously described (1, 10). In mouse primary cortical neurons expressing the Thorase variants, NMDA stimulation led to a significantly reduced internalization of GluA2, and recovery was significantly faster compared to neurons expressing wild-type Thorase (P < 0.05; Fig. 2, F to H, and fig. S3B). These results suggested that the Thorase variants disrupted AMPAR internalization and trafficking. Recently, we showed that Thorase is also part of the mitochondrial protein quality control system and thereby plays an important role in maintaining mitochondrial integrity (11). Results of mitochondrial function analyses assessed in mouse embryonic fibroblasts expressing the Thorase variants suggested that the variants had no significant effects on mitochondrial function (fig. S3, C and D).

Thorase variants increase the frequency and amplitude of sEPSCs and mEPSCs

To study the physiological and behavioral consequences of these Thorase variants, an adeno-associated virus 2 (AAV2) vector expressing GFP- or FLAG-tagged mouse wild-type Thorase or its variants was delivered by intracerebroventricular injection into the brains of postnatal day 2 mice that were heterozygous for loss of Thorase (12, 13). Immunostaining of brain sections from mice expressing the AAV2-Thorase-FLAG construct indicated the expression of wild-type or variant Thorase-FLAG throughout the mature mouse brain (fig. S4, A and B). Immunoblot analysis revealed similar expression of wild-type Thorase-FLAG and variant Thorase-FLAG compared to endogenous Thorase (fig. S4, C and D). Several lines of evidence suggest that behavioral core symptoms like reduced social drive or deficits in learning and memory, commonly observed in psychiatric disorders such as schizophrenia and autism, arise from a dysfunctional prefrontal cortex (14, 15). We therefore performed electrophysiology on brain slices of murine prefrontal cortex from heterozygous Thorase-deficient mice that contained deep-layer prelimbic pyramidal neurons (16). The Thorase variants did not alter intrinsic excitability of prelimbic pyramidal neurons in the brain slices (fig. S5, A to C). However, brain slices from heterozygous mice expressing FLAG-tagged Thorase variants did show an increase in both frequency and amplitude of spontaneous glutamate-mediated excitatory postsynaptic currents (sEPSCs; Fig. 3, A to E). However, the sEPSC rise and decay times did not differ between groups (fig. S5, D and E). We observed an increase in both amplitude and frequency of glutamate-mediated miniature excitatory postsynaptic currents (mEPSCs) in brain slices from mice expressing the Thorase variants D221H and E290K (P < 0.05; Fig. 3, F to J). No changes were detected in the rise and decay times of mEPSCs between any of the groups (P > 0.05; fig. S5, F and G).

Fig. 3. Thorase variants increase the frequency and amplitude of sEPSCs and mEPSCs.

(A) Representative traces of spontaneous glutamate-mediated excitatory postsynaptic currents (sEPSCs) recorded in heterozygous Thorase-deficient mice expressing WT Thorase (Het-WT; black) or the Thorase variants R9H, D221H, and E290K. (B to E) Mean cumulative probability distributions for sEPSC amplitude (B) and sEPSC frequency (D) recorded in prelimbic pyramidal neurons in brain slices from the prefrontal cortex of heterozygous Thorase-deficient mice expressing WT Thorase (Het-WT; black) or one of the three Thorase variants. (C and E) Mean sEPSC amplitude and sEPSC frequency, respectively. Means ± SEM (n ≥ 5). *P < 0.05; n.s., P > 0.10, ANOVA with Holm-Sidak post hoc test when compared to Het-WT mice (black). Power (1 − β error probability) = 0.8 to 1.0. (F) Representative traces of glutamate-mediated miniature excitatory postsynaptic currents (mEPSCs) recorded in WT mice (brown) or heterozygous Thorase-deficient mice expressing WT Thorase (Het-WT; black) or the Thorase variants R9H (orange), D221H (blue), and E290K (green). (G to J) Mean cumulative probability distributions for mEPSC amplitude (G) and mEPSC frequency (I) for prelimbic pyramidal neurons recorded in brain slices from the prefrontal cortex of WT mice (brown) or heterozygous Thorase-deficient mice expressing WT Thorase (Het-WT; black) or the Thorase variants R9H (orange), D221H (blue), and E290K (green). (H and J) Mean mEPSC amplitude and mEPSC frequency, respectively. Means ± SEM (n ≥ 4). *P < 0.05; n.s., P > 0.10, ANOVA with Holm-Sidak post hoc test when compared to Het-WT mice (black). Power (1 − β error probability) = 0.8.

The increase in mEPSC amplitude and frequency in the mouse prelimbic pyramidal neurons expressing the Thorase variants D221H and E290K suggested a more efficient unitary conductance (17) or a higher density of AMPARs at the synapses (18) compared to prelimbic pyramidal neurons expressing wild-type Thorase. The latter scenario would be consistent with biochemical and immunostaining data (Figs. 1 and 2), suggesting an increased surface expression of AMPARs in neurons expressing the Thorase variants. Increased glutamate-mediated sEPSCs observed in the mouse prelimbic pyramidal neurons expressing the Thorase variants might suggest a dysfunctional organization of glutamatergic synaptic transmission that would become evident when action potential–dependent synaptic transmission was intact. Alterations in the synaptic plasticity of neurons expressing the Thorase variants were further confirmed by showing an increase in long-term potentiation (LTP) and elimination of long-term depression (LTD; fig. S6, A to F).

Mice expressing the Thorase variants show psychostimulant hypersensitivity and behavioral changes

To determine whether the Thorase variants could lead to behavioral changes, a battery of behavioral tests was performed on heterozygous mice expressing wild-type Thorase or the Thorase variants. Several studies have shown that the disruption of NMDA receptors (NMDARs) with NMDAR antagonists impairs cognitive function in healthy subjects and intensifies psychotic symptoms in patients with schizophrenia (19, 20). NMDARs have been shown to modulate surface expression of AMPARs (1, 10). We therefore used the psychostimulant MK-801, a noncompetitive NMDAR antagonist (19, 20), to evaluate the sensitivity of the mice carrying the Thorase variants to this psychostimulant. Heterozygous mice expressing the Thorase variants showed deficits in exploration in a novel environment (Fig. 4, A to D, and fig. S7, A to C) (21, 22) and an enhanced locomotor response after treatment with MK-801 (Fig. 4, A and B) compared to heterozygous mice expressing wild-type Thorase. The mice expressing the Thorase variants were also hypersensitive to the psychostimulant amphetamine (fig. S7D).

Fig. 4. Perampanel ameliorates behavioral deficits in mice expressing the Thorase variants.

(A to D) Open-field test assessments in WT mice (brown) and heterozygous Thorase-deficient mice expressing WT Thorase (Het-WT; black) or the Thorase variants R9H (orange), D221H (blue), and E290K (green). (A) Representative pathview images of the open-field activities of the mice treated with vehicle, the psychostimulant MK-801, or Perampanel showing characteristics of the patterns of locomotor activity. (B) Graphical representation of the total distance traveled by mice in the open-field test after treatment with vehicle or MK-801. (C) Graphical representation of the total distance traveled by mice in the open-field test after treatment with vehicle or Perampanel. Means ± SEM (n = 10). ***P < 0.01; **P < 0.05; *P < 0.1; n.s., P > 0.10, ANOVA with Holm-Sidak post hoc test when compared to Het-WT (black). Power (1 − β error probability) = 0.8 to 1.0. (D) Percent central and peripheral activities in the open-field test after treatment with vehicle or Perampanel. (E) Graphical representation of percent correct spontaneous alternation in the Y-maze spatial working memory test after treatment with vehicle or Perampanel. (F) Prepulse inhibition (PPI) in the acoustic startle response test. The graphs represent the percent PPI at different prepulse intensities after treatment with vehicle or Perampanel. Means ± SEM (n = 7 to 10). ***P < 0.01; **P < 0.05; *P < 0.1; n.s., P > 0.10, ANOVA with Holm-Sidak post hoc test when compared to Het-WT (black). Power (1 − β error probability) = 0.9 to 1.0. (G) Assessment of social interactions by WT mice (brown), heterozygous Thorase-deficient mice expressing WT Thorase (Het-WT; black), or the Thorase variants R9H (orange), D221H (blue), and E290K (green). Graph represents number of mouse visits to a chamber containing a new unfamiliar mouse versus an empty chamber. (H) Evaluation of social memory in WT mice (brown), heterozygous Thorase-deficient mice expressing WT Thorase (Het-WT; black), or the Thorase variants R9H (orange), D221H (blue), and E290K (green). Graph represents number of mouse visits to a chamber containing either a familiar mouse or a new unfamiliar mouse. (I and J) Trace fear conditioning to assess contextual memory and associative learning in WT mice (brown), heterozygous Thorase-deficient mice expressing WT Thorase (Het-WT; black), or the Thorase variants R9H (orange), D221H (blue), and E290K (green). (I) Mean percent freezing by mice during the test to assess contextual memory. (J) Mean percent freezing by mice during the cue test. Graphical representation of the average freezing time for the first two audible tones is shown. Means ± SEM (n = 12). ***P < 0.005; **P < 0.01; *P < 0.05, two-way ANOVA with Tukey-Kramer post hoc test compared to Het-WT (black). Power (1 − β error probability) = 1.0.

Mice expressing the Thorase variants show impaired memory and social behavior

Assessment of spatial working memory using spontaneous alternation in a Y-maze continuous trial test (Fig. 4E and fig. S7, E and F) indicated that the Thorase variants led to deficits in both spatial working memory and spatial recognition memory. Several studies have shown that patients with psychiatric disorders such as schizophrenia exhibit deficits in prepulse inhibition of the acoustic startle response; similar deficits can be measured in mice (23, 24). Heterozygous mice expressing the Thorase variants exhibited deficits in prepulse inhibition (Fig. 4F and fig. S7, G and H) compared to heterozygous mice expressing wild-type Thorase.

Next, we examined social behavior in mice expressing the Thorase variants. Heterozygous Thorase-deficient mice or heterozygous mice expressing wild-type Thorase made more visits and spent more time with new unfamiliar mice introduced to their cages compared to mice carrying the Thorase variants (Fig. 4G and fig. S7I). In the social novelty preference test, heterozygous Thorase-deficient mice or heterozygous mice expressing wild-type Thorase made more visits and spent more time with new mice compared to familiar animals than did heterozygous mice expressing the Thorase variants (P < 0.05; Fig. 4H and fig. S7, J and K). In contrast, Thorase variant mice spent equal time and made equal visits with both new and familiar mice (Fig. 4H and fig. S7, J and K). Together, these data reveal deficits in social behavior in heterozygous mice expressing the Thorase variants.

Long-term memory and associative learning deficits in heterozygous mice expressing the Thorase variants could be reversed by Perampanel

Associative learning in heterozygous mice expressing the Thorase variants was assessed by subjecting them to trace fear conditioning using an audible tone and electric foot shock. After training, the animal’s ability to associate the training context to the shock and the tone to the shock was tested. No significant difference in the mean percent freezing between the various groups of mice during training was observed (fig. S7L), suggesting that the hearing of mice expressing the Thorase variants and their response to the foot shock were not impaired. However, there was significantly less freezing during the contextual memory test in heterozygous mice carrying the Thorase variants R9H and E290K compared to heterozygous mice expressing wild-type Thorase (P < 0.05). There was a trend toward less freezing in mice carrying the Thorase D221H variant compared to mice expressing wild-type Thorase (Fig. 4I and fig. S7M). In addition, heterozygous mice carrying any of the three Thorase variants displayed a significantly lower average percent freezing response to the first two audible tones compared to heterozygous mice expressing wild-type Thorase (P < 0.05, P < 0.01; Fig. 4J and fig. S7N). Together, these results indicate that heterozygous mice carrying the Thorase variants showed deficits in both context- and cue-dependent fear conditioning, suggesting impaired associative learning and memory. A summary of the biochemistry, physiology, and behavior results is listed in table S2.

Psychosis is normally treated with antipsychotic medications that are only partially effective, and most patients do not resume normal activities after hospitalization (9). Allosteric modulation of proteins linked to schizophrenia has been suggested as the best approach for the successful development of new drugs. Current treatment emphasis is now shifting to addressing the cognitive and negative symptoms of schizophrenia using allosteric modulators (9). We therefore evaluated the effects of the U.S. Food and Drug Administration–approved AMPAR antagonist Perampanel on heterozygous mice expressing the Thorase variants. We evaluated the effects of Perampanel (0.5 mg/kg) on locomotor activities in the open-field test, on spatial working memory in the Y-maze test, and on prepulse inhibition. The open-field test results showed that heterozygous mice expressing the Thorase variants were more sensitive to Perampanel than heterozygous mice expressing wild-type Thorase and that treatment normalized their open-field activities (Fig. 4, A to D). Treatment with Perampanel also normalized deficits in exploration in a novel environment (Fig. 4D) and restored spatial working memory in heterozygous mice expressing the Thorase variants (Fig. 4E). The prepulse inhibition deficits in heterozygous mice expressing the Thorase variants were reversed by Perampanel treatment (Fig. 4F). Perampanel was also able to ameliorate some MK-801–induced behavioral deficits in the heterozygous mice expressing the Thorase variants (fig. S8, A to E).

DISCUSSION

Recently, it has been appreciated that schizophrenia spectrum disorders and autism spectrum disorders may share common genetic and biological components that unfold over time. Thorase has been implicated in an autism spectrum disorder. The deletion of chromosome 10q23.31 containing ATAD1, the gene encoding Thorase, in a patient was associated with Bannayan-Riley-Ruvalcaba and Cowden syndromes characterized by developmental delay and learning disabilities (25, 26). Although some of the patient’s symptoms were linked to the deletion of PTEN (also located on 10q23.31) (26), the authors could not explain other symptoms observed in this patient. The developmental delay and learning disabilities in this patient were consistent with symptoms observed in mice lacking ATAD1 (1). We have proposed that the deletion of Thorase in this patient accounted for some of the symptoms linked to the deletion of 10q23.31 (1). Because of possible commonalities between autism and schizophrenia, we wondered whether deletion of or mutations in Thorase could be associated with schizophrenia in our Ashkenazi Jewish cohort. We found three Thorase variants in this patient population. However, the small sample size did not provide sufficient evidence that these variants increased the risk for psychiatric disease. Future studies will need to be conducted in a larger patient cohort to determine the relative contribution of the Thorase variants to human neuropsychiatric disease.

We evaluated the biological, physiological, and behavioral effects of these Thorase variants in mouse primary cortical neurons and in heterozygous Thorase-deficient mice expressing either the Thorase variants or wild-type Thorase. Our results suggested that altered AMPAR-mediated synaptic activities in heterozygous mice expressing the Thorase variants may be due to dysfunctional AMPAR trafficking at synapses. The elevated surface expression of AMPARs in the prefrontal cortex of heterozygous mice carrying the Thorase variants may have contributed to behavioral deficits in these animals. Inhibition of excess surface expression of AMPARs by Perampanel normalized several aberrant behaviors in heterozygous mice carrying the Thorase variants. Perampanel treatment to modulate surface expression of AMPARs could offer a potential therapeutic opportunity for treating disorders associated with abnormal AMPAR-mediated neurotransmission.

MATERIALS AND METHODS

Study design

We evaluated the functional effects of three Thorase variants identified in a cohort of schizophrenia patients in an Ashkenazi Jewish population. We used biochemical, electrophysiology, and behavioral studies to determine the effects of these Thorase variants on glutamatergic neurotransmission in heterozygous mice deficient in Thorase expressing one of the three Thorase variants or wild-type Thorase (2, 7). The purity of recombinant proteins used to determine the biochemical activities of the Thorase variants was assessed by SDS–polyacrylamide gel electrophoresis Coomassie staining. Western blots were probed with anti-Thorase antibody and anti-His6 antibody. In studies of mouse primary cortical neurons, neurons expressing GFP were used as a control. Several well-established behavioral tests (19, 20) were used to assess the behavior of heterozygous mice expressing the Thorase variants compared to heterozygous mice expressing wild-type Thorase. For all tests, the effect size was first calculated on the basis of preliminary experiments and then used in power analysis by the G*Power software to determine sample size by given error probability and power (table S3). Mice were toe-clipped for identification; however, they were randomly selected and then assigned numbers during behavioral tasks. A blinded observer assessed the outcomes, and all data collected were included in the final analysis. Imaging and analysis of signal intensity were performed by automated software.

Antibodies

All antibodies were acquired commercially: Thorase [monoclonal antibody (mAb), NeuroMab, RRID: AB_2564836], GluA2-N/C (mAb, Millipore-Chemicon, RRID: AB_2113875 and RRID: AB_2247874), and GRIP1 [polyclonal antibody (pAb), Millipore-Chemicon, RRID: AB_11210079]. Anti–glutathione S-transferase (GST)–horseradish peroxidase (HRP) (RRID: AB_771429), anti-His6 antibody (RRID: AB_771435), and actin-HRP were purchased from GE Healthcare, and anti-FLAG-HRP was purchased from Sigma. Perampanel was provided by Eisai Co. Ltd. Dizocilpine (MK-801) and NMDA were purchased from Sigma-Aldrich.

Drug preparations and treatments

Perampanel was reconstituted to 0.4 mg/ml (in sterile 0.1 M HCl in saline) (27), and MK-801 was reconstituted to 1 mg/ml in sterile saline. Drugs or vehicle (sterile 0.1 M HCl in saline) were administered by intraperitoneal injections in equal volumes.

Animals

Thorase heterozygous (Thorase +/−) mice were obtained by crossing heterozygous (Thorase +/−) mice (1). Animal experiments were performed in compliance with the regulations of the Animal Ethical Committee of the Johns Hopkins University Animal Care and Use Committee. Mouse behavioral testing occurred in the Johns Hopkins University School of Medicine Behavioral Core using protocols approved by and standard protocols found in the behavior core manual.

Sequencing of ATAD1 coding exons

To identify the Thorase (ATAD1) variants associated with schizophrenia cases of Ashkenazi Jewish origin, we obtained bidirectional sequence of the ATAD1 coding exons in 712 schizophrenia cases and 649 controls of Ashkenazi Jewish origin, as previously described (28). The functional activities of all three coding variants—R9H, D221H, and E290K—identified in schizophrenia cases were determined.

ATPase activity assay

ATP binding of the variants was determined by ultraviolet light–induced cross-linking of radiolabeled [α-P32]ATP, and glutaraldehyde chemical cross-linking was used to evaluate oligomerization upon ATP binding, as described by Babst et al. (29). Structural and conformational changes were examined using circular dichroism. Details of procedures are described in Supplementary Materials and Methods.

Thorase variants and their interaction with GluR2 and GRIP1

For the in vitro Thorase-GluR2-GRIP1 interactions assays, purified GST-GluR2 and GST-GRIP1 recombinant proteins bound to glutathione beads were incubated with purified His6-tagged Thorase proteins. In a separate experiment, GST-Thorase recombinant proteins bound to glutathione beads were incubated with purified Glu2 C-terminal fragment (GluR2C) and GRIP1 PDZ 4/5 domain fragment. The GST beads were extensively washed to evaluate Thorase-GluR2-GRIP1 interactions. Immunoprecipitation of Thorase-FLAG in the presence of primary cortical cultures expressing the Thorase variants was used to determine Thorase-GluR2-GRIP1 interactions in neurons. Details of procedures are described in Supplementary Materials and Methods.

Disassembly and endocytosis of GluR2/GRIP1 complexes

The ability of the Thorase variants to interact with and disassembly the GluR2-GRIP1 complex was evaluated by immunoprecipitation of GluR2 in the presence of ADP, ATP, or ATPγS from primary cortical cultures expressing the Thorase variants, as previously described (1). An antibody-feeding internalization assay for endocytosis of surface GluR2 receptors and GluR2 surface protein cross-linking were used to evaluate GluR2 surface expression in the presence or absence of NMDA. pHluorin-tagged GluR2 fusion protein was used to analyze the membrane trafficking of GluR2 after NMDAR activation in primary hippocampal cultures expressing the Thorase variants. Details of procedures are described in Supplementary Materials and Methods.

Electrophysiology

Schizophrenia-linked Thorase mutants were expressed in mice brain delivering AAV2 viruses carrying Thorase-FLAG wild-type or schizophrenia-linked variants into newborn Thorase heterozygous mice by intracerebroventricular injection (12). To determine whether intrinsic excitability in neurons is altered in mice expressing the schizophrenia-linked Thorase variants, ex vivo electrophysiology was performed in the prelimbic pyramidal neuron portion of the medial prefrontal cortex. Details of procedures are described in Supplementary Materials and Methods.

Behavioral experiments

Behavioral experiments associated with schizophrenia were performed using 6- to 8-month-old mice. The tests were performed in the following order: open-field, Y-maze, social interaction, prepulse inhibition, trace fear conditioning MK-801, and Perampanel challenge tests. Details of procedures for individual behavior tests are described in Supplementary Materials and Methods.

Data analyses and statistics

All experiments were repeated at least three times, and quantitative data are presented as means ± SEM performed by GraphPad Prism 6 software (InStat, GraphPad Software). Statistical significance was assessed by one- or two-way analysis of variance (ANOVA). The significant differences were identified by post hoc analysis using the Holm-Sidak post hoc test for multiple comparisons. Assessments were considered significant with P < 0.05 and nonsignificant with P > 0.05. Power analysis and sample size calculation for all experiments were determined using G*Power 3.1 and SigmaStat statistics software. Power was calculated as 1 − β, assuming α = 0.05.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/9/420/eaah4985/DC1

Materials and Methods

Fig. S1. Thorase variants show no defects in ATPase activity but do show changes in structure.

Fig. S2. Thorase variants impair GluA2 and GRIP1 interactions.

Fig. S3. Thorase variants impair GluA2 trafficking.

Fig. S4. AAV2 Thorase-FLAG expression in mice.

Fig. S5. Membrane properties and biophysical features of AMPARs are not affected in prelimbic cortex pyramidal neurons of mice expressing the Thorase variants.

Fig. S6. Defects in LTP/LTD in Thorase variant mice.

Fig. S7. Impaired social behavior and learning deficits in Thorase variant mice.

Fig. S8. Fear conditioning and the effects of Perampanel in Thorase variant mice.

Table S1. Summary of ATAD1 variants identified in patients with schizophrenia of Ashkenazi Jewish origin.

Table S2. Summary of activities of the Thorase variants.

Table S3. Statistical analyses.

REFERENCES AND NOTES

  1. Acknowledgments: We thank Eisai Co. Ltd. (Tokyo, Japan) for providing Perampanel as noted in Materials and Methods. Funding: This work was supported by grants from the NIH DA000266 to T.M.D. and V.L.D. and the Simon’s Foundation Autism Research Initiative to T.M.D. Author contributions: G.K.E.U. contributed to all aspects of the project. M.P. and X.Y. conducted electrophysiology experiments. R.C. generated virus and conducted biochemical experiments. J.C., S.N., and N.G. conducted behavior and biochemistry experiments. L.S., A.A.B., M.C., E.M., J.W.K., and C.C.C. conducted biochemistry experiments. X.M. and L.C. conducted biochemistry and cell culture experiments. S.A.A. performed mitochondrial function experiments. M.V.P., A.B., A.E.P., D.A., and D.V. performed the sequencing of the ATAD1 coding exons in schizophrenia cases. G.K.E.U., T.M.D., and V.L.D. designed experiments and wrote the paper. The study was conceived and scientifically directed by T.M.D. and V.L.D. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Perampanel was made available by Eisai Pharmaceuticals through a materials transfer agreement. Requests for ATAD1 sequencing data will be managed by the Johns Hopkins School of Medicine Office of Research Administration. (https://www.hopkinsmedicine.org/research/resources/offices-policies/ora/index.html).
View Abstract

Navigate This Article