PSD-95 deficiency disrupts PFC-associated function and behavior during neurodevelopment

Postsynaptic density protein-95 (PSD-95) is a major regulator in the maturation of excitatory synapses by interacting and trafficking N-methyl-D-aspartic acid receptors (NMDAR) and α-amino-3-hydroxy-5-methyl-4-isox-azoleproprionic acid receptors (AMPAR) to the postsynaptic membrane. PSD-95 disruption has recently been associated with neuropsychiatric disorders such as schizophrenia and autism. However, the effects of PSD-95 deficiency on the prefrontal cortex (PFC)-associated functions, including cognition, working memory, and sociability, has yet to be investigated. Using a PSD-95 knockout mouse model (PSD-95−/−), we examined how PSD-95 deficiency affects NMDAR and AMPAR expression and function in the medial prefrontal cortex (mPFC) during juvenile and adolescent periods of development. We found significant increases in total protein levels of NMDAR subunits GluN1, and GluN2B, accompanied by decreases in AMPAR subunit GluA1 during adolescence. Correspondingly, there is a significant increase in NMDAR/AMPAR-mediated current amplitude ratio that progresses from juvenile-to-adolescence. Behaviorally, PSD-95−/− mice exhibit a lack of sociability, as well as learning and working memory deficits. Together, our data indicate that PSD-95 deficiency disrupts mPFC synaptic function and related behavior at a critical age of development. This study highlights the importance of PSD-95 during neurodevelopment in the mPFC and its potential link in the pathogenesis associated with schizophrenia and/or autism.

and social impairments associated with neuropsychiatric disorders. However, the molecular and cellular effects in response to PSD-95 deficiency on the mPFC function have yet to be elucidated. We hypothesized that PSD-95 deficiency (Supplemental Figs 1 and 2) will disrupt synaptic maturation during critical periods of neurodevelopment due to an alteration in NMDAR/AMPAR-glutamatergic transmission that leads to impairments in mPFC development and function. By utilizing a PSD-95 knockout mouse model, this study characterizes changes in specific NMDAR-and AMPAR-subunit expression levels at the synapse, accompanied with fluctuations in SAP-102 and PSD-93 protein levels in response to PSD-95 deficiency in the mPFC. We also examine the effects of PSD-95 deficiency on synaptic function in an age-dependent manner during development, thus enabling us to identify a critical period at which PSD-95 deficiency disrupts synaptic maturation in the mPFC. Additionally, we assess mPFC synaptic function by recording NMDAR/AMPAR current and short-term plasticity measurements using whole-cell patch clamp recordings. And lastly, we evaluate mPFC-associated behavior such as cognition, working memory and sociability. This study provides great insight into the effects of PSD-95 deficiency on synaptic function and development of the mPFC and its potential implications in aberrant behavior associated with neuropsychiatric disorders.

Discussion
The mPFC is a major region within the cortex responsible for executive functions, such as cognition, working memory, and sociability. However, this region is highly susceptible-largely due to synaptic dysregulation-in patients diagnosed with neuropsychiatric disorders, thus contributing to symptoms that include cognitive deficits and sociability impairments 22,23 . PSD-95, a major component responsible for synaptic maturation, has recently been associated with neuropsychiatric disorders 4,24-27 . However, despite the importance of PSD-95 in excitatory synaptic function and impairment in diseases, how PSD-95 deficiency affects PFC-associated function remain unexplored. Previous studies using a PSD-95 knock-out model reveal decreases in AMPAR/NMDAR current ratio in the hippocampus at postnatal day 14, and within the visual cortex at postnatal day 30 9,10 ; thus, indicating region-specific developmental changes that occur due to PSD-95 deficiency. Our study, although revealing a similar reduction in the AMPAR/NMDAR current ratio, reports these changes occuring at postnatal day 35 in the mPFC of PSD-95 −/− mice.
Our results showed that synaptosomal AMPAR-and NMDAR-subunits protein expression levels are altered during the adolescent, but not juvenile, age range in PSD-95 deficient mice, thus identifying a critical period at which PSD-95 is most susceptible in the mPFC. Accordingly, previous studies have shown that within the normal cortex, PSD-95 proteins levels increase from early life and peak during adolescence/early adulthood 28,29 , suggesting a development age range at which PSD-95 greatly influences synaptic maturation. We revealed a significant reduction in AMPAR-subunit GluA1, accompanied by a significant increase in NMDAR-subunits www.nature.com/scientificreports www.nature.com/scientificreports/ GluN1 and GluN2B in the absence of PSD-95 during adolescence (but not juvenile). This is contrary to normal synaptic development, as AMPARs demonstrate an increase in recruitment and stabilization compared to NMDARs 30 . We showed that the aberrant increase in NMDAR subunits observed in PSD-95 −/− mice was attributed to a compensatory increase in SAP-102 protein expression, and is corroborated with an increase in the interaction of GluN2B-containing NMDAR subunits. Therefore, we provide a novel mechanism describing a pivotal role of SAP-102 in response to PSD-95 deficiency and the dynamic responsibilities of MAGUK scaffolding proteins at the PSD. Interestingly, PSD-93, involved in the formation of heteromulterization complexes of PSD-95 and NMDARs 31 , is dramatically reduced in the mPFC of PSD-95 −/− mice. This data suggest a reduction in PSD-93/PSD-95 complexes at the PSD and further describes the dominant role of SAP-102 in the absence of PSD-95 and its effect on NMDARs. NMDARs consist of an obligatory GluN1 subunit and are involved in a GluN2B-to-GluN2A subunit switch that occurs from childhood-to-adulthood 32 in most cortical regions, whereas GluN2B protein levels remain high in the mPFC, and is thereby important for cognitive processes [33][34][35] .
Furthermore, SAP-102 preferentially binds GluN2B to form SAP-102/GluN2B complexes in immature synapses and are replaced by PSD-95/GluN2A complexes later in development 36 . Therefore, high expression levels of SAP-102/GluN2B due to PSD-95 deficiency would disrupt normal synaptic maturation within the PFC. GluN2B-containing NMDA receptors have slow kinetics and thus play a major role in calcium (Ca + ) influx at the postsynaptic membrane; however, an overabundance may result in a significant increase in Ca + conductance that could lead to excitotoxicity and neuronal damage 37 . Therefore, we expect the increase in GluN2B presence due to PSD-95 deficiency would impair mPFC tissue function, thus leading to behavioral deficits.
Our electrophysiology data revealed an increase in the NMDAR/AMPAR-mediated current ratio in layer V pyramidal neurons in PSD-95 −/− mice that progresses during development. We showed no differences in AMPAR-eEPSCs or NMDAR-eEPSCs at the juvenile age range (P21-25); and thus, is consistent with the observed AMPAR and NMDAR subunits protein expression levels in the mPFC of PSD-95 deficient mice. However, during the adolescence age range (P35-P39), we observed a significant decrease in AMPAR-eEPSCs, accompanied with an increase in the NMDAR-eEPSCs peak amplitude and decay time, that is indeed consistent with the observed decrease in AMPAR subunits GluA1 and increase in NMDAR GluN1 and GluN2B protein expression levels from western blot analysis. This data describe and corroborate that alterations in AMPAR and NMDAR protein www.nature.com/scientificreports www.nature.com/scientificreports/ expression levels at the PSD affect AMPA and NMDA-receptor function activity. Additionally, this data characterize the critical adolescent time point during development at which PSD-95 deficiency alters mPFC AMPAR and NMDAR protein expression levels and function.
We examined mPFC-associated behavioral phenotypes that assessed sociability, cognition, and working memory in PSD-95 −/− mice. In a previous study, a group characterized behavioral domains of mice with total PSD-95 deletion (Dlg4 −/− ), and showed Dlg4 −/− mice displayed repetitive behavior, cognitive deficits, disrupted motor coordination; however, no differences in unconditioned anxiety behaviors or social interaction 38 . In our study, PSD-95 −/− mice revealed no significant differences in locomotor activity, thereby displaying normal spontaneous locomotion compared to control mice. PSD-95 −/− mice showed a lack of social exploration and novelty, and thus describe reduced sociability as observed in SCZ patients 39,40 . However, it is plausible that regions other than the mPFC responsible for sociability are susceptible to PSD-95 deficiency, such as the nucleus accumbens (NAc), amygdala, and hypothalamus 41,42 . For instance, it was recently shown that the neuropeptide oxytocin regulates synaptic plasticity in the NAc during critical periods of neurodevelopment 42 . Therefore, a down-regulation of PSD-95 in the NAc during development is likely to disrupt synaptic plasticity that would lead to aberrant sociability. Moreover, the mPFC has a top-down influence on these subcortical regions 41 ; thus, PSD-95 deficiency may cause a reduction in mPFC output that would reduce sociability.
We also assessed the recognition memory of PSD-95 −/− mice during the novel object recognition task. Recognition memory involves the mPFC and hippocampus, as the task includes memory consolidation and retrieval 43,44 , although some studies suggest the mPFC is not a requirement for simple recognition tasks 45,46 . Interestingly, PSD-95 −/− mice showed no differences compared to control mice in discrimination ratio during the tasks, although displayed recognition memory deficits to the novel object. Therefore, since PSD-95 −/− mice www.nature.com/scientificreports www.nature.com/scientificreports/ exhibit only partial novel object recognition deficits, it could suggest that hippocampus-mPFC projections remain intact. However, when performing the T-maze task to assess learning and working memory, PSD-95 −/− mice could not reach criterion (70% correct responses) even after 14 days compared to control mice beginning at day 6, indicating severe learning deficits. Not surprisingly, PSD-95 −/− mice performed significantly worse during delay tasks at 5 sec, 15 sec, and 60 sec (≤55% correct responses), suggesting working memory impairments. No significant differences were observed at 30 sec, and is likely attributed to the slight decline from control mice (63.89%). We associate these learning deficits with our electrophysiological data, more specifically, an attenuation in AMPAR-eEPSPs following a 10-pulse stimulation, further illustrating the effects of a reduction in GluA1 protein expression in the mPFC of PSD-95 −/− mice. Together, these results suggest that PSD-95 deficiency severely impairs mPFC-associated behavior that includes deficits in sociability, learning, and cognition. However, a discord of interpreting these results is the utilization of a whole-brain PSD-95 knock-out model rather than an mPFC-specific PSD-95 knockout. For instance, other regions containing mPFC projections such as the ventral hippocampus (memory consolidation and retrieval), basolateral amygdala (emotional control), and mediodorsal thalamus (working memory), may severely influence behavioral attributes of PSD-95 −/− mice 47,48 . Therefore, utilizing optogenetics techniques to understand the circuits involved would likely further elucidate the behavioral phenotypes of PSD-95 deficient mice.
We conclude that PSD-95 deficiency disrupts NMDAR/AMPAR balance and attenuates glutamatergic transmission within the mPFC. Since NMDARs are important for synaptic plasticity and cortical development, and involved in learning and memory, PSD-95 deficiency would alter these processes due to an aberrant upregulation of NMDARs at the synapse. Additionally, a reduction in AMPARs at the synapse may delay synaptic plasticity, and thus will affect learning and memory. Altogether, our results revealed alterations at the molecular, physiological, and behavioral levels in PSD-95 deficient mice during the adolescent age range; however, whether these changes persist into adulthood have yet to be investigated. Based upon our data, that reveal a progressive increase in NMDAR/AMPAR ratio from P21-to-P35, we expect an exaggerated decrease in GluA1 and AMPAR function, accompanied with dramatic increases in GluN1 and GluN2B and NMDAR function in adult (>postnatal day 70) PSD-95 deficient mice, thereby exacerbating behavioral phenotypes. Additionally, in Dlg4 −/− mice, cognitive deficits were observed during adulthood, therefore indicate persistent physiological aberrations 38 . Furthermore, examining spine density within layer 5 pyramidal neurons would be critical for understanding the structural effects of PSD-95 deficiency. Acquiring this data would be important for applying treatments such as memantine (an NMDAR antagonist) or LY395756 (mGluR2 agonist and mGluR3 antagonist), which were shown to balance NMDAR/AMPAR expression and function indirectly, and could rescue mPFC-associated behavioral deficits 49,50 . Thus, the data from our study may provide future therapeutic options that alleviate glutamatergic dysfunction in response to PSD-95 deficiency.

Materials and Methods
Animals. To model PSD-95 deficiency, we acquired PSD-95 knock-out mice from Jackson laboratories (B6.129-Dlg4 tm1Rlh /J). Standard breeding procedures were used to generate homozygous PSD-95 −/− mice. All mice were genotyped using PCR techniques described in the Supplemental information ( Supplementary Fig. 1). Both male and female C57/BL6 mice were used for all experiments and were divided into juvenile (Postnatal days, P14-21) and adolescent (Postnatal days, P35-55) groups as reported in prior studies [51][52][53] . The mice were cared for according to the National Institutes of Health (NIH) guidelines. Our animal experiment protocol was approved by the Institute Animal Care and Use Committee (IACUC) at Drexel University College of Medicine.
Tissue collection and synaptosomal protein preparation. Mice aged P21 and P35 were anesthetized with euthasol (0.2 mg/kg, i.p.). Once unresponsive to toe-and tail-pinch, mice were transcardially perfused with ice-cold perfusion buffer. The medial prefrontal cortex was micro-dissected and isolated from mice and homogenized in a sucrose buffer (320 mM sucrose, 4 mM HEPES-NaOH buffer, pH 7.4, 2 mM EGTA, 1 mM Na 3 VO 4 , 0.1 mM PMSF, 50 mM NaF, 10 mM Na 4 P 2 O 7 , 20 mM C 3 H 9 O 6 P, 1 μg/mL leupeptin, and 1 μg/mL aprotinin). Homogenates were centrifuged at 1,000 g for 10 min at 4 °C to remove nuclear materials and large cell fragments. The supernatant was centrifuged at 15,000 g for 15 min to yield cytoplasmic proteins and the pellet was hypoosmotically lysed and re-suspended in homogenization buffer. The suspension was incubated for 30 minutes at 4 °C with continuous mixing, and then centrifuged at 25,000 g for 30 min to isolate synaptosomal protein fractions. Protein concentrations were measured using the Magellan protein assay machine (Tecan). Protein samples were made with lysis buffer, Laemlli sample buffer and b-mercaptoethanol to a total volume of 10 µg of protein.
Western Blot and co-immunoprecipitation. The western blot and co-immunoprecipitation procedure were performed as previously described 54,55 . The mPFC protein samples were boiled at 95 °C for 5 min and then loaded into an SDS-PAGE gel for electrophoresis. Following electrophoresis, gels were transferred to polyvinylidene difluoride (PVDF) membranes (Millipore Billerica, MA) for 1 hr at 100 V, 4 °C. Membranes were blocked with 5% non-fat dry milk in TBST (0.05% Tween-20 in 1X Tris-buffered saline) for 1 hr and incubated in the following dilutions of primary antibodies for 1 hr: mouse monoclonal anti-GluN1 www.nature.com/scientificreports www.nature.com/scientificreports/ generate control bands. Blots will be rinsed with TBST 3x 20 minutes, and membranes will be incubated in horseradish peroxidase (HRP)-conjugated goat anti-mouse or rabbit IgG (mouse, Vector Laboratories Cat# PI-2000, RRID: AB_2336177; rabbit, Cat# PI-1000, RRID: AB_2336198) at 1:10000 for 1 hr. Protein bands were detected with the ECL Western Blotting System (Amersham ECL Western Blotting Detection Reagent, RPN2106). Membranes were then exposed to films and band densities measured with Image J (NIH). Data were normalized to levels of β-actin.
The mPFC tissue protein samples were further used for co-immunoprecipitation assays. 25 µg/µl of synaptosomal proteins were incubated overnight with 2.5 µg of anti-SAP-102 or anti-PSD-93. The immunocomplexes were isolated with 100 µl of Protein G MagBeads (GenScript, Cat# L00274) and incubated for 1-2 h at room temperature. Immunoprecipitates were washed three times with 1x phosphate buffered saline (PBS) and resuspended in Laemlli sample buffer, and boiled for 5 min. The supernatant was collected and immunoprecipitates were identified using standard western blot procedure as described above. Antibodies were directed against GluN2B, and against GluA1 as a negative control. Samples from each animal were run at least 3 times to minimize interblot variance. Results are presented as mean ± SEM and significance determined with Student's t-test.

Miniature excitatory postsynaptic currents (mEPSCs).
To isolate miniature AMPA receptor currents, the membrane potential was held at −70 mV with Cs + -containing solution intracellular solution in the presence of picrotoxin and tetrodotoxin (TTX, 0.5 µM, Hello Bio, Princeton, NJ, Cat# HB1034). The mEPSCs from layer V pyramidal neurons were recorded for 5 min, and the frequency and amplitudes were measured by averaging 5 sweeps from the on-set of recording.
10-pulse temporal summation of evoked excitatory postsynaptic potentials (eEPSPs). Whole-cell current clamp recordings were used to measure eEPSPs. Patch electrodes were filled with potassium gluconate internal solution and a bipolar electrode was used to stimulate layer II/III and record from layer V pyramidal neurons in the mPFC as similarly described above. A train of 10 pulses/stimuli were delivered at 20 Hz to record eEPSPs. The membrane potentials of the recording cells were adjusted within −68 mV to −73 mV through a small holding current.
Data analysis. Electrophysiological experiments were conducted with the Axon MultiClamp 700B amplifier (Molecular Devices). Data acquisition was from pCLAMP 9.2 software and analyzed using Clampfit 9.2 (Molecular Devices). The eEPSCs amplitudes were measured by averaging 30 sweeps from the onset to the peak amplitude of the EPSCs. The NMDAR/AMPAR-mediated current ratio was calculated by measuring AMPAR peak value at −60 mV and NMDAR peak value at +60 mV 50-ms post-stimulus that is illustrated by a yellow circle on EPSC trace examples. The kinetics of NMDAR-EPSC decay (τ) were measured by standard exponential fitting 63% of the EPSC peak amplitude and is represented in milliseconds (ms). To analyze sEPSCs, we select a sample sEPSC as a template within the 5 min data acquisition period. The frequency (number of events/300 = Hertz) and amplitude (peak) of the individual events were examined with a threshold in Clampfit. Neurons that produced stable baseline EPSCs were used for analysis. EPSC data were analyzed with the two-tailed Student t-test for statistical significance and were presented as mean ± S.E.M. The 10-pulse AMPAR-eEPSPs were recorded in current clamp mode, and peak voltages were measured and normalized to calculate temporal summation. Repeated-measures analysis of variance (ANOVA) with paired t-test was used to determine significance. The paired-pulse ratio was determined as the peak voltage of EPSP2/EPSP1. www.nature.com/scientificreports www.nature.com/scientificreports/ Sociability tasks. The sociability test was performed on mice aged P35 within a 3-chamber apparatus (box measures 62 × 43 × 20 cm; individual 3 chambers 19.5 × 43 cm). The tasks were divided into two sessions assessing social approach and social exploration/novelty 58,59 ; and is further described in the Supplemental information. Both sessions were 10 min, and the time in each chamber and sniffing time (nose pokes) were measured. ANOVA followed up with a paired t-test were used for statistical analysis. p < 0.05 was considered statistically significant.
Cognition tasks. First, to evaluate recognition memory, a novel object recognition task was performed on mice aged P35. Object sniffing time was measured between the familiar object and novel object, and discrimination ratio (novel object interaction/total interaction with both objects) was calculated to score object recognition. A one-sample t-test was used to compare individual groups to the discrimination ratio of 0.5 (50%). To compare the groups a two-tailed Student t-test was for statistical significance and presented as mean ± S.E.M. p < 0.05 was considered statistically significant. Next, to evaluate learning and working memory, a discrete paired-trial delayed alternation training task using a T-maze apparatus (50 × 72 cm) was used as previously reported 60,61 . Mice were required to reach criterion (70% correct) to begin testing, which consisted of variable intra-trial delays of 5 s, 15 s, 30 s, and 60 s. Repeated measures of ANOVA with Tukey-Kramer post hoc test were used for statistical analysis of the working memory task. Details of novel object recognition and T-maze working memory tasks are in the Supplemental information.