The N-methyl-D-aspartic acid (NMDA)-hypofunction theory of schizophrenia suggests that schizophrenia is associated with a loss of NMDA receptors, specifically on corticolimbic parvalbumin (PV)-expressing GABAergic interneurons, leading to disinhibition of pyramidal cells and cortical desynchronization. However, the presumed changes in glutamatergic inputs onto PV interneurons have not been tested directly. We treated mice with the NMDAR antagonist ketamine during the second postnatal week and investigated persistent cellular changes in the adult medial prefrontal cortex (mPFC) using whole-cell patch-clamp recordings and immunohistochemistry. Parvalbumin expression in the mPFC was reduced in ketamine-treated (KET) mice, and γ-aminobutyric acid release onto pyramidal cells was reduced in layers 2/3, but not layer 5. Consistent with pyramidal cell disinhibition the frequency of spontaneous glutamatergic inputs onto PV cells was also increased in KET mice. Furthermore, developmental ketamine treatment resulted in an increased NMDA:AMPA ratio in evoked synaptic currents and larger amplitudes of spontaneous NMDAR currents, indicating a homeostatic upregulation of NMDARs in PV interneurons. This upregulation was specific to NR2B subunits, without concomitant alterations in currents through NR2A subunits. These changes altered synaptic integration at PV cells during trains of excitatory postsynaptic potentials. These changes likely impact synaptic coincidence detection and impair cortical network function in the NMDAR antagonism model of schizophrenia.
Glutamate hypotheses of schizophrenia are based on the ability of N-methyl-D-aspartic acid (NMDA) receptor antagonists such as phencyclidine, ketamine, or MK-801 to induce a schizophrenia-like syndrome including positive and negative symptoms, and cognitive deficits in healthy subjects, and to exacerbate the symptoms of the disorder in schizophrenic patients (Javitt and Zukin 1991; Lahti et al. 1995; Malhotra et al. 1997; Krystal et al. 2002).
Subchronic blockade of NMDARs in animals also reduces the levels of the γ-aminobutyric acid (GABA)-synthesizing enzyme glutamic acid decarboxylase 67 (GAD67) and of the calcium-binding protein parvalbumin (PV) which labels fast-spiking (FS) interneurons (Cochran et al. 2003; Abekawa et al. 2007; Braun et al. 2007; Coleman et al. 2009). These changes replicate findings from postmortem studies in the brains of schizophrenics, suggesting that dysfunction of GABAergic interneurons is a core feature of schizophrenia (Eyles et al. 2002; Lewis et al. 2005; Akbarian and Huang 2006). The synchronous activity of PV cells generates gamma-oscillations, which correlate with performance in a variety of cognitive tasks, including the allocation of attention and working memory. Therefore, the functional loss of NMDARs on PV cells has been proposed to underlie the cognitive disturbances associated with schizophrenia (Winterer and Weinberger 2004; Minzenberg et al. 2010; Lewis et al. 2012).
Most of the evidence for the NMDAR-hypofunction theory is based on acute or subchronic administration of NMDAR antagonists in adult animals and humans (Javitt and Zukin 1991; Jentsch and Roth 1999; Homayoun and Moghaddam 2007). However, schizophrenia is a neurodevelopmental disorder and several recent studies suggest that NMDAR hypofunction may have different targets when induced in the adult state versus during development (Wang and Gao 2009; Belforte et al. 2010; Rotaru et al. 2011). NMDAR blockade or subunit deletion early in development when NMDARs are hypersensitive (Ikonomidou et al. 1989) and FS cells express high levels of NMDARs (Wang and Gao 2009; Zhang and Sun 2011) that reduces PV expression and might disrupt the maturation of FS connectivity (Belforte et al. 2010; Waites et al. 2005; Lewis et al. 2012). However, because levels of NMDARs decrease into adulthood (Wang and Gao 2009) the contribution of NMDARs to normal FS function in the adult cortex is still a matter of debate (Rotaru et al. 2013). More importantly, despite the purported importance of altered NMDAR function in FS interneurons, little is known about the changes in glutamatergic transmission onto these cells in NMDAR-hypofunction models.
To address several of these issues we treated mice with ketamine on postnatal days 7, 9, and 11 and performed whole-cell recordings in adult animals to study persistent changes in the firing properties and synaptic connections of FS cells in the medial prefrontal cortex (mPFC). Developmental ketamine treatment reduced PV expression and altered glutamatergic inputs onto FS cells. Recordings from FS cells and pyramidal cells indicated a persistent shift toward cortical disinhibition and an unexpected relative increase in total NMDA currents, as well as larger contribution of NR2B subunits in FS cells from ketamine-treated (KET) animals. The upregulation of NMDAR function in FS cells altered synaptic integration during repetitive synaptic stimulation, providing an alternative explanation for desynchronized network activity in the schizotypic cortex.
Materials and Methods
Animals and Ketamine Treatment
Male mice of the G42 line (CB6-Tg[Gad1-EGFP]G42Zjh/J; Jackson Laboratories, Bar Harbor, ME, RRID:IMSR_JAX:007677) expressing GFP in PV-positive interneurons (Chattopadhyaya et al. 2004) were used for the experiments. Saline or 30 mg/kg ketamine (Ketathesia HCl, Henry Schein, Dublin, OH) injections were administered at 10 mL/1 kg to pups on PND 7, 9, and 11 (Powell et al. 2012). Electrophysiological experiments were carried out on GFP-expressing hemizygous mice. Both WT and hemizygous animals were used for immunohistochemical analyses as indicated. All experiments were performed using adult animals (>PND 70). All procedures were approved by the Institutional Animal Care and Use Committee of The University of Texas at Dallas.
For double-labeling of PV and NeuN, or PV and GFP, animals were perfused transcardially with warm 0.9% saline followed by 4% paraformaldehyde in 0.12 M phosphate buffer (PB; 4°C, pH 7.4). Brains were postfixed in paraformaldehyde with 30% sucrose for 1 h and were then transferred to 30% sucrose in PBS for ∼18 h at 4°C. Coronal slices (40 μm) were cut on a freezing microtome and collected in PBS containing 0.01% NaN3 as a preservative. Free-floating sections were incubated in combinations of the primary antibodies diluted in 0.3% Triton-X in PBS for 36 h at 4°C. We used the following primary antibodies: rabbit anti-parvalbumin (Swant, Switzerland, Cat# PV 25 RRID:AB_10000344; 1:2000 working dilution), mouse anti-NeuN Alexa 488 conjugated monoclonal antibody (Millipore, Billerica, MA, Cat # MAB377X, RRID:AB_2149209; 1:300 working dilution), and mouse anti-GFP monoclonal antibody (Abcam, Cambridge, MA, Cat# ab1218 RRID:AB_298911; 1:5000 working dilution). Sections were washed at least 3 times for 10 min each in PBS before they were incubated in the secondary antibodies for 2 h at room temperature. We used DyLight 594 Goat Anti-rabbit (Jackson ImmunoResearch, West Grove, PA, Cat # 111–515–144; 1:1000 working dilution) and FITC donkey anti-mouse (Jackson ImmunoResearch, Cat # 715–095–151; 1:2500 working dilution) in 0.3% Triton-X in PBS. Sections were again washed 3 times in PBS before they were mounted and cover-slipped using Prolong Gold Antifade with 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen, Grand Island, NY). For each animal a minimum of 4 sections including the prelimbic and infralimbic regions of the mPFC were imaged on a confocal microscope (Fluoview 1000, Olympus Corporation, Tokyo, Japan) at 20× or 60×. The acquired images were converted to TIFF format, and cell counts for PV, NeuN, GFP, and DAPI were performed on the 20× images using the thresholding function in ImageJ (NIH). The percentage of PV+ cells among NeuN, and the percentage of GFP-expressing cells among DAPI cells were calculated. Layer-specific alterations in PV expression were performed from the 20× images, by quantifying the percentage of PV expression among NeuN in the superficial and deep layers using ImageJ software (Paxinos and Franklin, 2008; Van de Werd et al. 2010). To measure somatic fluorescence intensity of PV+ interneurons we used images obtained at 60×, drew outlines around the somata of PV-stained cells and measured the median fluorescence intensity using the histogram function in ImageJ.
Mice were overdosed with urethane (3 g/kg body weight) and rapidly decapitated. Brains were removed and placed for 2 min in ice-cold oxygenated (95% O2–5% CO2) artificial cerebrospinal fluid (aCSF) containing the following (in mM): 110 choline, 25 NaHCO3, 1.25 NaH2PO4, 2.5 KCl, 7 MgCl2, 0.5 CaCl2, 25 dextrose, 11.6 Na+-ascorbate, and 3.1 Na+-pyruvate, bubbled with 95% O2–5% CO2. Coronal sections (350 µM thick) of the frontal cortex were cut on a vibratome (VT1000S, Leica). Slices were then transferred to an incubation ACSF for a minimum of 45–60 min prior to recording containing the following (in mM): 120 NaCl, 25 NaHCO3, 1.23 NaH2PO4, 3.3 KCl, 2.4 MgCl2, 1.8 CaCl2, 10 dextrose, bubbled with 95% O2–5% CO2; pH 7.4, 35°C. Whole-cell voltage-clamp recordings were obtained from cells in layer 2/3 or layer 5 of the mPFC as indicated at room temperature using recording ACSF containing (in mM): 126 NaCl, 2.5 KCl, 1.2 Na2HPO4, 25 Na2HCO3, 10 glucose, 2 CaCl2, and 1 MgCl2, bubbled with 95% O2–5% CO2. For voltage-clamp recordings, electrodes (3–5 MΩ open tip resistance for pyramidal cells, 6–8 MΩ for interneurons) were filled with (in mM): 130 CsCl, 20 TEA, 10 HEPES, 2 MgCl2, 0.5 EGTA, 4 Na-ATP, 0.3 Na-GTP, 14 phosphocreatine, and 2 QX-314. For current-clamp recordings electrodes (6–8 MΩ) were filled with a solution containing (in mM): 120 K-gluconate, 10 HEPES, 10 KCl, 10 NaCl, 4 ATP-Mg, 0.3 GTP-Na, 14 phosphocreatine and 0.04 Alexa 594, pH 7.2 (KOH). Recordings were performed on Axon Multiclamp 700B amplifier (Molecular Devices, Union City, CA) and data were acquired and analyzed using Axograph X (Axograph Scientific, New South Wales, Australia). Recordings were performed from cells either in layer 5 or layer 2/3 of the prelimbic and infralimbic regions of the mPFC (Paxinos and Franklin, 2008). Cells were recorded at a holding potential of −70 mV unless otherwise indicated. Access resistance of the recorded cells was monitored throughout the experiment and a <20% change was deemed acceptable. Synaptic potentials were evoked by focal stimulation with theta-glass pipettes (Warner Instruments, Hamden, CT), and a stimulus isolator (WPI, Sarasota, FL). Intrinsic properties of recorded cells were measured from recordings performed in current-clamp mode by injecting current pulses of 1 s duration, ranging from −100 to 300 pA in 20 pA steps. Measurements of these properties and their analyses were performed using Axograph X. Peak amplitude or area under the curve was calculated from averaged traces of at least 10 sweeps. The frequency and amplitude of spontaneous and miniature postsynaptic currents were measured from 100 s continuous recording using MiniAnalysis (Synaptosoft, Decatur, GA) with a threshold set at 2 times the RMS baseline noise. To study ketamine-induced changes in NMDA and α-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor-mediated currents, we calculated the NMDA:AMPA ratio as previously described (Bonci and Malenka 1999). In brief, a compound EPSC was first recorded at a holding potential of +40 mV. The AMPA component of the synaptic response was then pharmacologically isolated by bath application of 20 µM of the NMDAR blocker (±)-3-(2-carboxypiperazin-4-yl)propyl-1-phosphonic acid (CPP, Sigma-Aldrich, St. Louis, MO). This AMPA current was then digitally subtracted from the compound response in the absence of APV to yield the NMDAR current. The peaks of the isolated AMPA and NMDA responses were then divided to yield the NMDA:AMPA ratio and compared across treatment groups. Alterations in short-term synaptic dynamics were studied in current-clamp mode, holding cells at −70 mV. Excitatory postsynaptic potentials (EPSPs) were isolated by adding 75 µM PTX to the bath and a train of EPSPs (15 pulses at 20 Hz) was evoked by electrical stimulation. Contribution of NMDARs was blocked by bath application of 20 µM CPP as indicated. Averages of 10–20 sweeps were obtained and analyzed offline for peak amplitude and area under the curve of each EPSP. Data from the 2nd to the 15th EPSP in the trains were normalized to the first EPSP in the train and results are shown as percentage change of amplitude or area of the nth EPSP relative to the first EPSP in the train.
GABAA receptor-mediated currents were blocked by application of 75 µM picrotoxin to the bath in order to isolate glutamatergic events when recording spontaneous and miniature EPSCs. AMPA receptor-mediated events were blocked using 20 µM CNQX (6-cyano-7-nitroquinoxaline-2,3-dione disodium salt hydrate) when recording IPSCs. Miniature events were recorded in the presence of 1 µM tetrodotoxin (TTX, Alomone Labs, Jerusalem, Israel) in the bath. In experiments investigating the contributions of NR2A and NR2B subunits individually, recordings were performed at +40 mV with picrotoxin and CNQX in the bath and the subunits were pharmacologically isolated using 0.5 µM PEAQX (NVP-AAM077 tetrasodium hydrate, [[[(1S)-1-(4-bromophenyl)ethyl]amino](1,2,3,4-tetrahydro-2,3-dioxo-5-quinoxalinyl)methyl] phosphonic acid tetrasodium hydrate) and 0.5 µM Ro25-6981 ([R-(R*,S*)]-α-(4-hydroxyphenyl)-β-methyl-4-(phenylmethyl)-1-piperidinepropanol hydrochloride hydrate), respectively. Pharmacological agents were obtained from Sigma-Aldrich.
Differences between the 2 treatment groups in all experiments were assessed using 2-way ANOVAs, repeated-measures ANOVAs and Student's t-test as indicated. Post hoc comparisons of statistical significance used Student's t-tests. Differences in the cumulative distributions of the amplitudes or the interevent intervals of spontaneous and miniature events were analyzed using the 2-sample Kolmogorov–Smirnov test. For immunohistochemical experiments n indicates the number of animals per condition while in electrophysiological experiments n indicates the number of cells recorded. All data are represented as mean ± SEM, with P < 0.05 being considered statistically significant.
Developmental Ketamine Treatment Reduces PV Expression in the mPFC of Adult Mice
A loss of PV expression in FS interneurons is a hallmark of the NMDAR-hypofunction model (Behrens et al. 2007; Braun et al. 2007; Zhang et al. 2008) and several other rodent models of schizophrenia (Hikida et al. 2007; Lodge et al. 2009) because it mimics changes in subsets of PV interneurons observed in postmortem brains of schizophrenic subjects (Beasley and Reynolds 1997; Eyles et al. 2002; Lewis et al. 2005; Akbarian and Huang 2006). The reductions in GAD67 and PV levels that result from subchronic NMDAR blockade are believed to represent maladaptive homeostatic mechanisms in a faulty effort to maintain a balance of excitation and inhibition in the cortical network (Behrens et al. 2007; Lisman et al. 2008). We performed immunohistochemical experiments in the mPFC of adult male mice to quantify the loss of PV expression that results from developmental ketamine treatment, relative to control littermates that received saline injections in the second postnatal week. GABAergic interneurons make up ∼20% of the neuronal population in rodents, and cells positive for PV comprise of ∼50% of those interneurons (Kinney et al. 2006; Povysheva et al. 2008). Consistent with previous studies using perinatal NMDAR antagonism (Nakatani-Pawlak et al. 2009; Powell et al. 2012) the number of PV+ interneurons in KET animals was reduced when expressed as the percentage of NeuN stained neurons across all cortical layers (Fig 1A,C left) (saline-treated [SAL] = 8.29 ± 0.79%, n = 10 animals; KET = 4.52 ± 0.69%, n = 8; P = 0.003). Next, we investigated whether the loss of PV expression differed between the superficial and deep layers of the mPFC (Fig 1C, middle and right). A 2-way ANOVA revealed a main effect of treatment (F1,32 = 33.163, P < 0.0001), indicating that KET animals showed a significant reduction in PV immunoreactivity in both layers 2/3 and 5, respectively, expressed as a percentage of NeuN stained cells (L 2/3: SAL = 6.48 ± 0.44%, n = 10 animals; KET = 1.94 ± 0.28%, n = 8 animals; P < 0.0001; L 5: SAL = 10.81 ± 1.09%; KET = 5.69 ± 1.14%; P = 0.005). However, no significant layer X treatment interaction was found (F1,32 = 0.122, P = 0.73), showing that the relative reduction of PV+ cells was similar across the layers. In a separate group of animals, we also quantified the fluorescence intensity in the somata of PV+ interneurons. In PV+ cells from KET mice the median pixel intensity was reduced (Fig 1B right, D) (SAL = 33.12 ± 4.16, 1,123 PV+ cells from 4 animals; KET = 19.23 ± 3.29, 665 PV+ cells from 4 animals; P = 0.039). In contrast, the number of GFP+ interneurons was not reduced (Fig 1B middle, E) (SAL = 1.76 ± 0.12%, n = 7; KET = 1.37 ± 0.17, n = 7; P = 0.085), suggesting that a majority of the PV+ interneurons was viable and allowing us to visually identify GFP+ cells during electrophysiological recordings in acute slices. Importantly, the reduction of PV expression among the subset of GFP+ cells (animals of the G42 line express GFP in only ∼50% of all the PV+ cells; Chattopadhyaya et al. 2004) was similar to that seen in the total population of PV+ cells. Thus, ketamine treatment reduced the percentage of PV-GFP colocalization to 42.5% (479 PV+ cells among 1125 GFP+ cells counted from n = 7 animals).
Ketamine Administration Reduces GABAergic Inputs to Layer 2/3 but not Layer 5 Pyramidal Cells
A central assumption of the NMDA-hypofunction theory of schizophrenia is that NMDAR blockade exerts a selective effect on GABAergic interneurons, leading to disinhibition of glutamatergic principal cells, which may further impact the GABAergic interneurons (Olney et al. 1999; Behrens et al. 2007; Nakazawa et al. 2012). To test these assumptions we examined whether developmental ketamine treatment resulted in persistent changes in synaptic transmission between pyramidal (PYR) cells and FS interneurons, respectively. We first recorded changes in GABAergic inputs onto layer 5 pyramidal cells in adult brain slices of the mPFC from KET and SAL animals (Fig 2A). We observed no differences in the average amplitude of spontaneous inhibitory postsynaptic currents (sIPSCs) between the 2 treatment groups (SAL = 19.32 ± 3.46 pA, n = 8; KET = 17.74 ± 2.6 pA, n = 8; P = 0.72). Ketamine treatment also did not change the average frequency of sIPSCs (SAL = 8.07 ± 0.64 Hz, KET = 9.82 ± 0.87 Hz, P = 0.13) (Fig 2B). Similarly, ketamine treatment did not alter the properties of action-potential independent miniature IPSCs (mIPSCs) recorded in the presence of TTX in the bath (amplitude: SAL = 19.26 ± 2.4 pA, n = 9; KET = 15.73 ± 1.17 pA, n = 9; P = 0.2; frequency: SAL = 7.5 ± 0.95 Hz, KET = 6.97 ± 1.25 Hz; P = 0.73), indicating that ketamine treatment had no effect on GABA release probability in layer 5 of the adult brain (Fig 2D). However, results from a recent study using perinatal PCP treatment in rats suggest that changes in GABAergic transmission may be layer-specific, with reductions in GABA release probability seen in layer 2/3, but not in layer 5 (Kjaerby et al. 2014). Therefore, we tested whether the effects of our perinatal ketamine treatment show similar layer specificity (Fig 3). Consistent with our results obtained in layer 5, the average frequency (SAL = 5.35 ± 1.04 Hz, n = 7; KET = 6.02 ± 0.39 Hz, n = 7; P = 0.56) of sIPSCs recorded from layer 2/3 pyramidal cells did not differ between the 2 treatment groups (Fig 3A). Similarly, the average amplitude of sIPSCs (SAL = 24.77 ± 2.87 pA, n = 7; KET = 28.34 ± 6.66 pA, n = 7; P = 0.63) was also unchanged by ketamine treatment (Fig 3A). In contrast, consistent with the report by Kjaerby et al. (2014), we found that ketamine treatment significantly reduced the average frequency of mIPSCs (SAL = 5.02 ± 0.65 Hz, n = 10; KET = 3.08 ± 0.31 Hz, n = 10; P = 0.015). These changes in frequency occurred without changes in the average amplitude of the mIPSCs (SAL = 18.17 ± 0.75 pA; KET = 19.62 ± 1.96 pA; P = 0.49) (Fig 3C,D). Taken together, these results suggest that developmental ketamine treatment leads to presynaptic and layer-specific alterations in GABAergic transmission onto mPFC pyramidal cells.
Ketamine Administration Does not Alter Intrinsic Firing Properties of Layer 5 FS Cells
Repeated NMDAR antagonism in adult animals can induce changes in the membrane properties of cortical FS interneurons (Wang and Gao 2012). Therefore, we first examined the effects of developmental ketamine administration on basic membrane properties of FS neurons in layer 5 of the mPFC (Fig 4A). FS interneurons were targeted based on their GFP fluorescent signal (Fig 4B), taking advantage of the fact that ketamine treatment reduced PV expression without similarly affecting the GFP signal (cf. Fig. 1E). All cells showed high-frequency nonadapting firing patterns typical of cortical FS cells. The basic membrane properties of cells from saline- and KET animals showed no significant differences (Table 1), and cells fired similar numbers of action potentials across a range of injected currents (Fig 4C) (2-way ANOVA: SAL n = 16, KET n = 15 cells, F1,29 = 0.961, P = 0.34). Thus the large reduction in PV expression following ketamine administration did not significantly alter the intrinsic properties of GFP+ (PV-expressing) cells, and specifically did not affect their fast-spiking phenotype which arises from voltage-gated potassium channels containing Kv3.1 subunits (Martina et al. 1998; Kawaguchi and Kondo 2002).
|Saline (n = 16)||Ketamine (n = 15)||P value|
|AP threshold (mV)||−41.75 ± 1.61||−39.76 ± 1.24||0.34|
|AP amplitude (mV)||58.21 ± 3.51||58.66 ± 3.07||0.92|
|AP half width (ms)||0.76 ± 0.03||0.74 ± 0.02||0.55|
|Input resistance (MΩ)||241.23 ± 24.8||235.78 ± 28.77||0.89|
|fAHP amplitude (mV)||16.12 ± 1.13||18.32 ± 1.13||0.18|
|Saline (n = 16)||Ketamine (n = 15)||P value|
|AP threshold (mV)||−41.75 ± 1.61||−39.76 ± 1.24||0.34|
|AP amplitude (mV)||58.21 ± 3.51||58.66 ± 3.07||0.92|
|AP half width (ms)||0.76 ± 0.03||0.74 ± 0.02||0.55|
|Input resistance (MΩ)||241.23 ± 24.8||235.78 ± 28.77||0.89|
|fAHP amplitude (mV)||16.12 ± 1.13||18.32 ± 1.13||0.18|
Ketamine Treatment Increases Excitatory Inputs Onto FS Interneurons in Layer 5
The data presented above and previous evidence from recordings in vivo (Homayoun and Moghaddam 2007) and in vitro (Kjaerby et al. 2014) suggest that NMDAR antagonist treatment can alter GABAergic synaptic transmission, leading to cortical disinhibition and eventual loss of PV expression as a homeostatic response to continued overexcitation (Behrens et al. 2007). However, whether and how excitatory inputs onto FS cells are altered has not been directly tested yet. To address this question we performed whole-cell patch-clamp recordings from layer 5 FS interneurons of the mPFC to study changes in AMPA-mediated glutamatergic inputs onto FS cells from KET animals. Pharmacologically isolated sEPSCs were recorded from PV cells (Fig 5A). Ketamine treatment during the second postnatal week significantly increased the average frequency of sEPSCs in PV cells in adult animals (SAL = 10.16 ± 2.48 Hz, KET = 18.3 ± 2.06 Hz; P = 0.03). This increase in frequency was independent of changes in the average amplitude of sEPSCs (SAL = 14.16 ± 1.31 pA, n = 6; KET = 12.04 ± 0.73 pA, n = 6; P = 0.19).
Because sEPSCs represent both action potential-dependent and -independent release of glutamate, we next recorded miniature EPSCs (mEPSCs) in the presence of TTX (1 µM) to eliminate the contribution of action potential-mediated release events (Fig 5B). Neither the average amplitude (SAL = 10.43 ± 1.06 pA, n = 7; KET = 11.65 ± 1.02 pA, n = 6, P = 0.43) nor the frequency (SAL = 6.78 ± 0.95 Hz, KET = 7.97 ± 0.97 Hz, P = 0.4) of mEPSCs were altered by ketamine administration. Thus these results indicate that ketamine treatment results in persistent increases in glutamatergic inputs to FS cells consistent with disinhibition of pyramidal neurons.
Developmental Ketamine Administration Leads to Homeostatic Upregulation of NMDARs
The contribution of NMDAR currents to synaptic signaling at PV interneurons changes throughout development (Wang and Gao 2009; Zhang and Sun 2011) and this may explain increased vulnerability of FS cells to NMDAR antagonism during development (Kaalund et al. 2013). However, previous reports suggest that the overall NMDAR current contribution to PV interneurons may be much lower than in other types of cortical cells (Wang and Gao 2009; Rotaru et al. 2011). Therefore we tested whether ketamine treatment induced alterations in NMDAR currents in layer 5 PV interneurons in adulthood. We performed voltage-clamp recordings from GFP+ cells at +40 mV and isolated AMPA and NMDA current components to calculate changes in the NMDA:AMPA ratio (Fig 6A). Surprisingly, PV interneurons from KET animals showed an increased NMDA:AMPA ratio compared with cells in SAL animals (SAL = 0.69 ± 0.15, n = 8; KET = 1.45 ± 0.24, n = 11; P = 0.02). These results from electrically evoked EPSCs indicated a relative increase in NMDARs relative to the AMPAR component in glutamatergic transmission, but they did not the address whether ketamine treatment altered the absolute contribution of NMDARs. If ketamine treatment resulted in upregulation of NMDARs then spontaneous NMDAR events onto FS cells should also show increased amplitudes, consistent with a postsynaptic effect. Therefore, we recorded spontaneous NMDAR currents (sEPSCNMDAR) from PV interneurons at −70 mV in nominally Mg2+-free ACSF, containing 20 µM CNQX and 75 µM PTX to block AMPA and GABAA mediated currents, and measured the amplitude, frequency, and decay times of sEPSCNMDAR (Fig 6B). The average amplitude of sEPSCNMDAR was significantly increased in cells from KET animals (SAL = 6.3 ± 0.6 pA, n = 7; KET = 8.17 ± 0.49 pA, n = 9; P = 0.03). In contrast, the frequency of sEPSCNMDAR did not differ between treatment groups (SAL = 2.55 ± 0.37 Hz; KET = 2.31 ± 0.22 Hz; P = 0.56), further indicating a postsynaptic locus of the observed changes in ketamine animals. Finally, sEPSCNMDAR in PV cells from KET animals had significantly slower decay times (SAL = 2.53 ± 0.33 ms; KET = 4.1 ± 0.32 ms; P = 0.004). Taken together these results suggest that developmental ketamine administration caused an upregulation of NMDARs in PV interneurons.
NMDAR Currents in PV Interneurons From Ketamine Animals Have Increased Contribution From NR2B Subunits
The contribution of NR2A and NR2B subunits to the total NMDAR current in PV interneurons changes during development. In young animals there is a larger contribution of NR2B subunits, while in adult animals NR2A subunits dominate (Zhang and Sun 2011). Because we observed an increase in the NMDA:AMPA ratio in KET animals, we next tested whether this increase was accompanied by a shift in the relative contribution of the NR2A or NR2B subunits to the NMDA current. To this end we recorded pharmacologically isolated NR2A- or NR2B-specific currents from PV interneurons. First a compound NMDAR current was recorded at +40 mV in ACSF that contained 20 µM CNQX and 75 µM PTX. Next we bath applied either the NR2B-specific blocker Ro25-6981 (0.5 µM) or the NR2A-specific blocker PEAQX (NVP-AAM077, 0.5 µM) and quantified the percentage change of the subunit-specific current relative to the total NMDAR current (Fig 7). In the presence of Ro25-6981 PV cells from KET animals showed a larger reduction in the current (SAL = −20.49 ± 4.4%, n = 7; KET = −35.35 ± 5.22%, n = 8; P = 0.03), indicating that NR2B subunit contribution was relatively increased in the ketamine group (Fig 7A). This observation is consistent with the slower sEPSCNMDAR seen in the KET animals. In contrast, application of PEAQX reduced the total NMDA currents to similar extents (SAL = −46.38 ± 5.21%, n = 5; KET = −50.81 ± 4.01%, n = 5; P = 0.52), indicating that NR2A subunit contribution did not differ between treatment groups (Fig 7B). Bath application of 20 µM CPP at the end of the experiment completely abolished the response in all cells tested (data not shown). Our data indicate a shift in the NR2A to NR2B balance in PV cells of KET animals that is due to upregulation of the NR2B subunit.
Developmental Ketamine treatment Alters EPSP Integration in Adult Animals
In PFC pyramidal cells the summation of EPSPs during repetitive stimulation is NMDAR dependent (Wang et al. 2008). However, glutamatergic inputs onto FS interneurons show a characteristic synaptic depression to trains of inputs, indicating a relatively small contribution of NMDARs at hyperpolarized potentials (Gonzalez-Burgos et al. 2005; Rotaru et al. 2011). Here we observed that developmental ketamine administration resulted in an upregulation of NMDARs in PV interneurons. Therefore, we tested whether increased NMDAR contribution alters temporal summation in FS cells from ketamine animals. Cells from both SAL (n = 13 cells) and KET (n = 13 cells) animals showed the characteristic pattern of synaptic depression in response to a train of 15 pulses at 20 Hz, but the cells from ketamine animals showed relatively enhanced summation (Fig 8A). Comparison of the normalized amplitude and area under the curve of the EPSP trains revealed a significant main effect of ketamine treatment for amplitudes (2-way ANOVA: F1,360 = 54.55, P < 0.0001) and area (F1,360 = 66.92, P < 0.0001). No significant treatmentXEPSP interaction was observed for amplitude (F14,360 = 1.02, P = 0.44) or area (F14,360 = 1.142, P = 0.32). The enhanced synaptic summation in ketamine-animals may be indicative of increased NMDAR contribution. In a subset of these cells (n = 7 SAL and 6 KET) we applied the NMDAR antagonist CPP to the recording bath to confirm that the changes in the ketamine-group were NMDAR mediated (Fig 8B). Changes in amplitudes of the EPSPs (normalized to the amplitude of the first EPSP, see Materials and Methods) were compared for each treatment group using a within-group repeated-measures ANOVA. No main effect of CPP application was observed in the saline group (F1,6 = 2.095, P = 0.198) or drugXEPSP interaction (F14,84 = 0.831, P = 0.634) were observed. However, in the ketamine-group CPP application caused a significant reduction in EPSP amplitudes (main effect: F1, 5 = 14.34, P = 0.013; drugXEPSP interaction: F14, 70 = 3.87, P < 0.0001). These results confirmed that developmental ketamine administration resulted in an upregulation of NMDARs in FS cells that alter synaptic summation in adult mice.
The NMDA hypofunction theory of schizophrenia (Olney and Farber 1995; Olney et al. 1999) provides a framework for the complex interactions of the GABA, glutamate, and dopamine neurotransmitter systems in schizophrenia. The theory suggests that schizophrenia is associated with a loss of NMDARs, particularly on GABAergic interneurons, which leads to a loss of inhibition and cortical desynchronization. Here we studied the persistent changes in mPFC network function that result from subchronic NMDAR blockade during the second postnatal week. Our results are the first to directly study changes in glutamatergic transmission and NMDAR-mediated currents in FS cells that are thought to precipitate the disinhibition of cortical pyramidal cell networks. We focused our analysis on microcircuits of the deep layers of the mPFC which favor the generation of persistent activity required for working memory (Sanchez-Vives and McCormick 2000). In addition, NMDA hypofunction models suggest that dysfunctional cortical outputs from layer 5 could be responsible for dysregulation of downstream monoaminergic transmitter systems in schizophrenia (Nakazawa et al. 2012).
Administration of PCP and ketamine induce schizophrenia-like symptoms in healthy subjects (Javitt and Zukin 1991; Krystal et al. 1994; Lahti et al. 1995; Newcomer et al. 1999) and therefore acute systemic administration of NMDAR antagonists in adult animals is widely used as a model to study the behavioral and neurochemical changes that may occur in the disease (Jentsch and Roth 1999; Homayoun and Moghaddam 2007; Mouri et al. 2007). However, schizophrenia is a developmental disorder which manifests itself in late adolescence or early adulthood and thus animal models of the disorder are needed that replicate specific changes during development (Lewis and Lieberman, 2000; Gonzalez-Burgos and Lewis 2012). In rodents the first 2 weeks after birth correspond to the second trimester of pregnancy in humans, during which time transient epigenetic factors or the exposure to environmental insults increase the probability of developing schizophrenia as an adult (Bayer et al. 1993; Brown and Derkits 2010). During early postnatal development FS cells express high levels of hypersensitive NMDARs (Ikonomidou et al. 1989; Wang and Gao 2009; Zhang and Sun 2011) which shape the maturation of these cells. Accordingly, NMDAR blockade or subunit deletion early in development can affect behavior into adulthood by disrupting the maturation of normal FS connectivity (Waites et al. 2005; Belforte et al. 2010; Lewis et al. 2012; Nakatani-Pawlak et al. 2009; Jeevakumar et al. in preparation). Throughout development the numbers of NMDARs gradually decline to levels that are lower than those found in pyramidal cells (Wang and Gao 2009; Rotaru et al. 2011), and thus there is considerable debate whether NMDAR antagonism in adult animals can preferentially affect FS cells to induce disinhibition of excitatory pyramidal cells as the NMDAR-hypofunction theory predicts (Rotaru et al. 2011, 2013) and as the available in-vivo data using acute NMDAR blockade would suggest (Homayoun and Moghaddam 2007).
In agreement with previous studies using perinatal NMDAR antagonist application (Powell et al. 2012; Kaalund et al. 2013) we observed a reduction in PV expression in the adult mPFC. These changes parallel the loss of GAD67 and PV observed in the cortex of postmortem brains of schizophrenic patients (Benes and Berretta 2001; Hashimoto et al. 2003; Lewis et al. 2005). Our results also provide evidence for reduced GABA release in layers 2/3, as well as increased action potential-dependent glutamate release onto FS cells in layer 5, 2 observations that are consistent with the idea of disinhibition of pyramidal cells as a result of persistent alterations in the GABAergic network. The changes in GABAergic transmission showed apparent layer specificity—while we observed no significant changes in either sIPSCs or mIPSCs in somatic recordings from layer 5 pyramidal cells we saw a reduction in the frequency of mIPSCs onto layer 2/3 pyramidal cells, indicating decreased action potential-independent release of GABA in the superficial layers. These findings are consistent with results from a recent study that described similar layer-specific alterations in mIPSCs, but not sIPSCs, following perinatal PCP treatment in rats (Kjaerby et al. 2014). Pyramidal cells in the superficial layers send excitatory projections to interneurons in the deep layers (Dantzker and Callaway 2000; Thomson and Bannister 2003) and thus disinhibition of pyramidal cells in layer 2/3 may also explain the increased excitatory drive onto layer 5 FS interneurons that we observed. However, the mechanisms that underlie the apparent layer-specificity in the ketamine modulation of GABA release are still unclear but may be a consequence of the distribution of PV interneurons across layers of the cortex. In the rodent cortex the percentage of PV cells is highest in layer 5 (Gabbott et al. 1997; Yuan et al. 2011). Thus a similar relative loss in the number of PV cells may have more pronounced functional consequences in layers 2/3 than in layer 5. In support of this idea the average frequency of mIPSCs in SAL animals was 5.01 Hz in layers 2/3 and 7.5 Hz in layer 5 (Figs 2D and 3D). A possible explanation for the layer-specificity of ketamine-induced changes in mIPSCs is that these changes are more likely to reach a critical level that can be detected with our recording techniques in layers 2/3. While these results suggest a primary locus of cortical disinhibition in the superficial layers, it is important to note that the number of PV-immunopositive cells was significantly reduced throughout all cortical layers, suggesting that NMDAR antagonism and the resultant changes in network function affect FS interneurons in all cortical layers in a similar fashion. Here, we focused our analysis on functional changes in layer 5 because disruptions in (prefrontal) cortical output may be responsible for downstream dysregulation of dopaminergic nuclei in schizophrenia (Nakazawa et al. 2012).
We found that subchronic ketamine treatment during development resulted in an apparent homeostatic upregulation of NMDARs in FS interneurons. This was evident as an increased ratio of NMDAR to AMPAR-mediated currents in electrically evoked EPSCs, as well as an increase in the amplitude of spontaneous NMDAR-mediated events. In addition to a relative increase in the total NMDAR current we found that in FS interneurons from KET animals the contribution of currents through NR2B subunits was increased. This was evident in recordings of pharmacologically isolated subunit-specific currents, as well as slower decay kinetics of spontaneous NMDAR currents in Mg2+ free solution, consistent with the slower kinetics of NR2B subunits (Paoletti et al. 2013). PV interneurons in the PFC of naïve animals express several subtypes of NMDA receptors, with a high contribution from NR2A subunits and a comparatively lower number of NR2B subunits (Kinney et al. 2006; Wang and Gao 2009; Xi et al. 2009). Postmortem studies in the brains of schizophrenic subjects indicate that GABA interneurons (Woo et al. 2004), and PV cells specifically (Bitanihirwe et al. 2009), show reduced levels of NR2A mRNA. However, Western blot analyses show that perinatal NMDAR antagonism can upregulate both NR2A and NR2B subunits in the PFC (Owczarek et al. 2011). Similar changes have been observed in the hippocampus of adult rats treated subchronically with MK-801 (Rujescu et al. 2006), in the frontal cortex of perinatally KET rats (Liu et al. 2011), in rat forebrain cultures (Liu et al. 2013), and in the prefrontal cortex of adult mice chronically treated with ketamine (Chatterjee et al. 2012). Our study is the first to provide direct evidence for NMDA-R upregulation in such a model in PV interneurons. In the development of FS interneurons a shift in subunit composition occurs so that the number of NR2B subunits decreases with age, leaving NR2A subunits to predominate in adult animals, and this switch from NR2B to NR2A subunits occurs after the first postnatal week, peaking during young adulthood (Monyer et al. 1994; Liu et al. 2004; Paoletti et al. 2013). The switch depends on, among other factors, the activity of NMDARs and the release of internal Ca2+ from IP3Rs (Matta et al. 2011). Thus repeated blockade of NMDARs during the second postnatal week might interfere with the mechanisms governing Ca2+ release (Taylor and Tovey 2010), leading to disruption of the subunit switch.
The changes in NMDAR transmission onto FS cells that we observe here might also contribute to cortical desynchronization and the behavioral or sensory deficits observed in perinatal NMDAR antagonism models. FS cells from various species and across different brain regions show a characteristic pattern of strong synaptic depression in response to consistent inputs (Beierlein and Connors 2002; Gonzalez-Burgos et al. 2005; Rotaru et al. 2011). The synapses of FS cells contact the perisomatic regions of their target neurons (Somogyi 1977; Halasy et al. 1996; Somogyi and Klausberger 2005), placing them in a position to tightly regulate the activity of pyramidal cells either via perisomatic inhibition, or as in the case of the axo-axonic chandelier cells, via precisely timed excitatory inputs (Szabadics et al. 2006; Woodruff et al. 2009). Thus one of the main functions of PV basket cells in cortical circuits is to synchronize the activity of large numbers of pyramidal cells (Sohal et al. 2009; Gonzalez-Burgos and Lewis 2012 for review). FS interneurons are thought to be particularly suited to integrate coincident inputs from presynaptic pyramidal cells because excitatory inputs onto these cells show fast kinetics (Galarreta and Hestrin 2001) with relatively low NMDAR contribution in adult animals (Fig. 6; Gonzalez-Burgos et al. 2005; Wang and Gao 2009; Zhang and Sun 2011). However, after developmental ketamine treatment FS interneurons displayed slower, longer lasting responses leading to enhanced synaptic integration, which was abolished following bath application of the NMDAR blocker CPP. The stronger NMDAR drive and the resulting prolonged EPSP integration is likely to have direct consequences on coincidence detection, thereby impairing the network's ability to generate and maintain normal levels of oscillations during cognitive tasks (Rotaru et al. 2013). Thus these changes may further alter the balance of excitation and inhibition in prefrontal cortical networks, resulting in impaired gamma rhythms and cognitive deficits in schizophrenic subjects (Lewis et al. 2012).
Taken together, our findings suggest that developmental NMDAR blockade induces several key neurochemical and electrophysiological changes that the NMDAR-hypofunction model of schizophrenia predicts, but that it also leads to homeostatic compensatory changes in NMDAR currents on FS interneurons. The alterations in NMDAR-mediated transmission seen in the FS cells may result in cortical desynchronization and lead to the behavioral aberrations seen in NMDAR antagonism models of schizophrenia. The mechanisms underlying these changes and their potential relationship to the pathophysiology of schizophrenia clearly warrant further investigation.
We thank Millad Sobhanian, Christopher Driskill, and Jaime DeLeon for additional support in immunohistochemical experiments. S.K. wishes to acknowledge institutional support from The University of Texas at Dallas. Conflict of Interest: The authors declare no competing financial interests.