Cognitive abnormalities in schizophrenia reflect deficits in prefrontal cortical function, which could be related to attrition of dendritic structures of prefrontal cortical neurons. Schizophrenia-related prefrontal deficits have been modeled in postpubertal neonatal ventral hippocampal lesioned (NVHL) rats, which displayed a loss of dendritic complexity and spines in layer 3 pyramidal neurons in the medial prefrontal cortex (mPFC). The influence of dendritic attrition on synaptic function and neuronal excitability in the mPFC remains poorly understood. Here, we performed electrophysiological recordings of layer 5 mPFC pyramidal neurons from postpubertal (postnatal 40–60 days) NVHL rats and sham-operated controls. We found that the dendritic length, complexity, and spine density of neurobiotin-labeled layer 5 mPFC pyramidal neurons in NVHL rats were significantly lower than those in sham-operated rats. However, the excitability of layer 5 mPFC pyramidal neurons remained unchanged after NVHL. We found no significant changes in the expression of vesicular glutamate and γ-aminobutyric acid transporters after NVHL. Intriguingly, NVHL increased the amplitude of action potential-independent miniature excitatory postsynaptic currents and decreased the frequency of miniature inhibitory postsynaptic currents. These opposing alterations in excitatory and inhibitory synapses, possibly shifting basal synaptic activity toward increased excitation, could be cellular substrates for mPFC functional deficits reported in NVHL rats.
Impairments in working memory, attention, and executive functions are common in schizophrenia (Goldman-Rakic 1994; Barch and Ceaser 2012). These cognitive impairments persist throughout the life of patients and are largely untreated by currently available neuroleptics (Goff et al. 2011). The pathophysiology of cognitive impairments could be related to aberrant neurodevelopment of cortico-limbic circuitry, resulting in a malfunctioning prefrontal cortex (PFC) (Marenco and Weinberger 2000; Volk and Lewis 2010). Indeed, functional magnetic resonance imaging studies of PFC activities during working memory tasks revealed significant alterations in PFC function in schizophrenia compared with controls (Manoach 2003; Reichenberg and Harvey 2007). Changes in PFC activity during working memory formation are related to alteration of PFC neuron firing, specifically in the dorsal lateral PFC (DLPFC). For example, during a primate delayed matching-to-sample task, the firing rate of PFC neurons increases during the delay period when cued information is held online in working memory, and the firing rate attenuates as working memory decays (Goldman-Rakic 1995; Fuster 2008). In rat medial PFC (mPFC), which is anatomically homologous to the primate DLPFC (Uylings and van Eden 1990; Van De Werd and Uylings 2008), firing of deep layer mPFC neurons also associates with distinct phases of various working memory tasks (Jung et al. 1998). These studies demonstrate the central role of mPFC neuronal function in working memory and raise the possibility that cognitive impairments observed in schizophrenia may arise from aberrant PFC neuron firing.
Prefrontal abnormalities related to schizophrenia have been modeled in rodents. For example, neonatal excitotoxic lesions of the ventral hippocampus (VH) in rat pups lead to postpubertal emergence of schizophrenia-like behaviors, including deficits in working memory (Lipska et al. 1993, 2002; Le Pen and Moreau 2002; Marquis et al. 2006; Kamath et al. 2008; Gruber et al. 2010). Neonatal VH lesioned (NVHL) rats are considered to be a putative neurodevelopmental model of schizophrenia (Marcotte et al. 2001; Tseng et al. 2009). The VH sends monosynaptic glutamatergic projections to the mPFC (Swanson 1981; Ferino et al. 1987), and as such, lesioning VH could reduce excitatory inputs to the mPFC. In parallel to this hypothesis, NVHL decreases the density of dendritic spines, which are generally considered to be postsynaptic structures of excitatory asymmetric synapses (Goldman-Rakic et al. 1989; Shepherd 1996), in layer 3 mPFC pyramidal neurons of postpubertal animals (Flores et al. 2005; Marquis et al. 2008). In addition, potassium-induced excitatory amino acid release is reduced by NVHL (Schroeder et al. 1999). Note that in the same study, NVHL increases glutamate binding, suggesting a compensatory increase in glutamate receptors after lesion. NVHL also affects inhibitory circuitry in the mPFC by decreasing mRNA expression of γ-aminobutyric acid (GABA)-synthesizing enzyme GAD67 (Lipska et al. 2003) and by delaying maturation of GABAergic neurons (Tseng et al. 2008). Nonetheless, the impact of NVHL on excitatory and inhibitory synaptic functions in the mPFC remains unclear. Notably, alterations in glutamatergic and GABAergic transmission in the mPFC after NVHL could affect the excitability of deep layer mPFC pyramidal output neurons. Since deep layer mPFC neuron firing has been associated with working memory formation (Jung et al. 1998), further studies are warranted to examine the influence of NVHL on the functional properties of synaptic inputs on these neurons.
In the present study, we used a combination of anatomical, electrophysiological, and protein expression approaches to examine the consequences of NVHL on synaptic structure and function of layer 5 mPFC pyramidal neurons in postpubertal rats (postnatal 40–60 days). We found that layer 5 mPFC pyramidal neurons exhibited a substantial loss of dendritic structures, but no changes in excitability after lesion. Interestingly, we observed opposing changes in the functional properties of glutamatergic and GABAergic synaptic inputs in the mPFC that could tilt the balance in favor of excitation of layer 5 mPFC pyramidal neurons. These compensatory alterations of glutamatergic and GABAergic synaptic inputs, which could be responsible for maintaining a stable excitability of layer 5 mPFC pyramidal neurons after NVHL-induced deafferentation, may be a mechanism underlying mPFC dysfunctions in these rats.
Materials and Methods
Care and use of animals was in accordance with the guidelines and policies of the Canadian Council on Animal Care and those of Facility Animal Care Committee at the Douglas Mental Health University Institute (DMHUI), McGill University. Eighteen pregnant female Sprague-Dawley rats at 15–18 days of gestation were obtained from Charles River Laboratories (Québec, Canada). They were housed in a temperature- and humidity-regulated environment at the animal facility of DMHUI. Animals were housed on a 12 h light–dark cycle with ad libitum food and water. Dams were individually housed until weaning [postnatal day (PD) 21]. On PD3, the litter size was culled to 8 pups per dam, with a maximum of 8 male pups which underwent NVHL or sham surgery on PD7.
Unless otherwise specified, all materials were purchased from Sigma-Aldrich (St Louis, MO, USA). Isoflurane was purchased from Baxter (Mississauga, Ontario, Canada). Borosilicate glass capillaries were bought from World Precision Instruments (Sarasota, FL, USA). Neurobiotin and avidin–biotin complex were purchased from Vector Laboratories (Burlingame, CA, USA). Tetrodotoxin (TTX) was purchased from Alomone Labs (Jerusalem, Israel). Reagents for the Bradford method and precast polyacrylamide gels were bought from BioRad (Hercules, CA, USA). Finally, antibodies against VGluT-1 and VGAT were purchased from Synaptic Systems (Göttingen, Germany).
VH lesion was performed according to the previously described procedure (Flores et al. 2005). On PD7, male pups (15–18 g) were anesthetized by hypothermia (placed on ice for 15–20 min) and immobilized to a stereotaxic apparatus (Kopf Instruments, Tujunga, CA, USA). A small incision was made with a surgical blade on the skin atop the pup's head. A blunted 30-gauge needle was connected to a 10 μL Hamilton syringe that was fixed to an infusion pump. The needle was lowered into the VH (coordinates relative to bregma: AP = 3.0 mm, ML = ± 3.5 mm, VD = 5.0 mm). Phosphate-buffered saline (PBS, 0.1 M, pH 7.4) with ibotenic acid (10 μg/μL, NVHL group) or PBS alone (sham group) was bilaterally injected into the VH at a rate of 0.15 μL/min for 2 min (0.3 μL total/side). The needle remained in place for 4 min to prevent diffusion of ibotenic acid along the needle path. The incision was then closed with surgical glue. After ear punching to designate group, pups were placed on a heating pad until fully revived and returned to their respective dams, where they remained until weaning (PD21). The rats were then separated by their lesion status and housed 2 per cage until sacrificed for experimental use between PD 40 and 60. A total of 33 sham-operated rats and 32 NVHL rats were used in this study.
Lesioned or sham rats were anesthetized with isoflurane before being decapitated by a guillotine. The brain was quickly removed from the skull and immersed in ice-cold carbogenated (95% O2 and 5% CO2) sucrose-substituted hyperosmotic artificial cerebrospinal fluid (S-ACSF) containing (in mM) 252 sucrose, 2.5 KCl, 0.1 CaCl2, 4 MgCl2, 10 glucose, 26 NaHCO3, and 1.25 NaH2PO4 (pH 7.35, 360–370 mOsm). The brain was trimmed by a razor blade into anterior and posterior halves, and the anterior half was used to prepare 400 μm thick coronal slices using a Vibratome (Leica, Concord, Ontario, Canada). Slices were immediately placed in carbogenated normal ACSF (125 mM NaCl instead of sucrose; 310–320 mOsm) at 32°C for 1 h and then maintained at room temperature for at least 30 min before electrophysiological recording. The remaining posterior half of the brain from lesioned rats was snap frozen and stored at −80°C for lesion verification.
The general procedures were described previously (Wong et al. 2000; Tse et al. 2011). Briefly, patch pipettes were pulled from borosilicate glass capillaries and filled with intracellular solution (pH 7.25, 280–290 mOsm) composed of (in mM): 110 Cs-gluconate (Cs-gluconate was replaced with 112.5 mM of additional CsCl for inhibitory postsynaptic current experiments and with 120 mM K-gluconate for current clamp experiments), 17.5 CsCl, 10 HEPES, 2 MgCl2, 0.5 ethylene glycol tetraacetic acid (EGTA), 4 ATP, 5 QX-314, and 0.5% neurobiotin (pH 7.2). Slices were transferred to a recording chamber perfused continuously with carbogenated ACSF. In experiments in which action potential-independent miniature synaptic activities were isolated, TTX (0.5 μM) was added to ACSF to block voltage-gated sodium channels. GABAA receptor antagonists, bicuculline (10 µM) and picrotoxin (10 μM), were used in some experiments to isolate excitatory synaptic activities. Alternatively, the AMPA receptor antagonist 6,7-dinitroquinoxaline-2,3-dione (20 μM) was added to ACSF to isolate inhibitory synaptic activities. The access resistance of the patch pipette was monitored throughout each experiment, and only recordings with stable and low access were analyzed. No electronic compensation for series resistance was used. After breakthrough, cortical neurons were injected with several 200 ms long hyperpolarizing pulses to estimate the input resistance of recorded neurons. Junction potential was corrected in current clamp experiments (Mody et al. 1992). The excitability of mPFC neurons was tested in the current clamp mode by injecting a series of depolarizing pulses in a stepwise manner. In voltage clamp experiments, neurons were held at −60 mV to isolate spontaneous synaptic activities. A Multi-Clamp 700B amplifier (Molecular Devices, Palo Alto, CA, USA) was used for all recordings. Recordings were low-pass-filtered at 2 kHz, sampled at 10 kHz, digitized, and stored in a PC using Clampex 10.1 (Molecular Devices). Electrophysiological data were analyzed off-line using either the Mini Analysis Program 6.0.3 (Synaptosoft, Decatur, GA, USA) for synaptic events examination or Clampfit (Molecular Devices) for input resistance estimation and firing properties analysis.
A total of 92 pyramidal neurons were recorded. Notably, there were no significant differences in access resistance (17.2 ± 1.3 MΩ in sham vs. 18.3 ± 0.8 MΩ in NVHL rats), input resistance (215.2 ± 16.5 MΩ in sham vs. 203.9 ± 9.9 MΩ in NVHL rats), or resting membrane potential (−68.6 ± 1.2 mV in sham vs. −70.4 ± 0.6 mV in NVHL rats) between sham and NVHL groups.
Pyramidal neurons were labeled by neurobiotin in the pipette solution during whole-cell recording (Wong et al. 2000). After recording, slices were quickly immersed in fixative containing 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (pH 7.4) overnight at 4°C. The tissue was then placed in fixative containing 30% sucrose for another 24 h for cryoprotection. Finally, fixed slices were embedded in optimal cutting temperature embedding medium (Sakura, Torrance, CA, USA) and resectioned into 50 µm-thick sections with a cryostat microtome (Leica). Free-floating cytochemical staining was performed in PBS with 0.2% Triton X-100 (PBST) for washing and diluting reagents throughout. Two 15 min PBST washes were performed between incubations. Endogenous peroxidase activity was quenched by incubating with 0.3% H2O2 in PBS for 15 min. To reveal neurobiotin-labeled pyramidal neurons, sections were incubated with avidin–biotin complex (1:1000) at room temperature for 1 h. After incubating sections in a mixture of 0.06% 3,3-diaminobenzidine tetrahydrochloride, 0.025% cobalt chloride, and 0.02% nickel ammonium sulfate in PBST for 15 min, 20 µL of 1% H2O2 was added (0.01% H2O2 final concentration) and the reaction was allowed to proceed for 2–3 min. After washing in PBS, sections were mounted on gelatin-coated glass slides, air-dried, dehydrated in ascending concentrations of ethanol, cleared with xylene, and cover-slipped with Permount mounting medium. Dendritic structures of labeled pyramidal neurons (6 NVHL and 6 sham neurons) were traced and examined using Neurolucida (MicroBrightField, Williston, VT, USA) that is connected to a motorized light microscope (Leica). The Sholl analysis was performed to measure the number of intersections between dendrites and concentric circles of 10–300 µm from the soma. Note that >97% of basal dendrites of the studied labeled neurons can be found within the 300 µm radius Sholl circle. Dendritic complexity of reconstructed neurons was further estimated by measuring the length, nodes (branching points), and endings (terminations) of basal dendrites within these concentric circles. For comparing dendritic spines, we counted all dendritic spines within a 300 µm radius Sholl circle from the soma under a ×100 objective. Spine density on apical and basal dendrites was estimated by dividing the total number of spines on these dendrites by their length within the Sholl circle.
Histological Lesion Verification
The snap frozen posterior half of NVHL brains containing the hippocampus and the surrounding structures was coronally sectioned (50 μm) using a cryostat microtome (Leica). Sections were mounted on gelatin-precoated glass slides and stained with cresyl violet (0.5%). The lesion was deemed acceptable if it exhibited bilateral neuronal loss, retraction, and cavitations in the VH. Additionally, the dorsal half of the hippocampus and the adjacent nuclei (i.e. amygdala and thalamus) had to be intact for the lesion to be acceptable.
mPFC synaptosomal fractionations were prepared from PD40–60 sham and NVHL rats as described (Hallett et al. 2008). Briefly, slices were homogenized in lysis buffer (in mM: 10 Tris base, 5 NaF, 1 Na3VO4, 1 ethylenediaminetetraacetic acid, 1 EGTA, pH 7.4) with 320 mM sucrose. Homogenate was centrifuged at 800 × g for 10 min at 4°C, and the supernatant was centrifuged at 9200 × g for 15 min at 4°C. The resulting P2 pellet was resuspended in lysis buffer containing 35.6 mM sucrose and incubated on ice for 30 min. Resuspended P2 was then centrifuged at 25 000 × g for 30 min at 4°C to obtain the LP1 pellet, which contains synaptosomal membranes. The LP1 pellet was resuspended in lysis buffer containing 1% sodium dodecyl sulfate and sonicated. After centrifuging at 15 000 × g for 5 min, the pellet was discarded and the supernatant, containing solubilized synaptosomal proteins, retained. Protein concentration was determined by the Bradford method, and synaptosomal proteins were used for estimating synaptic expression of the vesicular neurotransmitter transporter.
Equal amounts of synaptosomal mPFC proteins were separated on a 4–20% precast polyacrylamide gel, transferred to a polyvinylidene difluoride membrane, and immunoblotted with antibodies against VGluT-1 (1:4000) and VGAT (1:10 000). The same membrane was stripped and reprobed with antibodies against actin (1:1000). Optical density (OD) of VGluT-1 and VGAT, which was quantified using ImageJ (NIH, USA), was normalized by the OD of actin to control for protein loading.
All data are presented as mean ± SEM. Repeated-measures analysis of variance (ANOVA) was used to compare dendritic parameters in the Sholl analysis. Both the effect of lesion (sham vs. NVHL) and the effect of Sholl circle radius on various dendritic parameters were examined. Student's t-tests were used for comparisons between sham and NVHL groups.
We examined whether NVHL reduces the complexity and spine density of basal dendrites in layer 5 mPFC neurons that were labeled by neurobiotin during electrophysiological recording (Fig. 1A). Using Sholl's analysis in Neurolucida, we found that the number of contacts between basal dendrites of reconstructed neurons from NVHL rats (n = 6) and Sholl circles (number of intersections) was significantly lower compared with sham rats (Fig. 1B; n = 6). Repeated-measures ANOVA revealed significant effect of lesion (F1,10 = 10.5, P= 0.008), Sholl radius (F29,290= 39.7, P< 0.001), and, importantly, significant interaction between the effects of lesion and Sholl radius on the number of dendritic intersections (F29,290= 2.59, P< 0.001). Compared with sham rats, reconstructed layer 5 mPFC pyramidal neurons from NVHL rats also displayed shorter dendritic length (3.54 ± 0.35 mm in sham vs. 1.84 ± 0.40 mm in NVHL rats; P= 0.010), fewer number of nodes (i.e. branching points: 20.5 ± 1.69 in sham vs. 10.2 ± 2.30 in NVHL rats; P= 0.005), and endings (i.e. termination of dendrites: 29.8 ± 2.41 mm in sham vs. 16.5 ± 2.78 mm in NVHL rats; P= 0.005), suggesting a lower dendritic complexity of layer 5 mPFC pyramidal neurons in NVHL rats (Fig. 1C). In addition, we compared the number of spines within the same Sholl circles we used for dendritic complexity analysis (Fig. 1D). Repeated-measures ANOVA revealed significant effect of lesion (F1,10= 44.3, P< 0.001), Sholl radius (F29,290= 30.4, P< 0.001), and interaction between the effects of lesion and Sholl radius on the number of dendritic spines (F29,290= 5.17, P< 0.001). By dividing the total number of spines on apical and basal dendrites by their length within these Sholl circles (Fig. 1E), we found that the spine densities of both apical (6.8 ± 0.2 spines/10 µm in sham vs. 4.6 ± 0.3 spines/10 µm in NVHL rats; P< 0.001) and basal dendrites (5.9 ± 0.1 spines/10 µm in sham vs. 3.9 ± 0.4 spines/10 µm in NVHL rats; P= 0.003) were significantly reduced in NVHL rats. Notably, the decrease in the number of spines (Fig. 1D) is more pronounced than the decrease in the dendritic length (Fig. 1B), probably due to the combined effects of reduced dendritic length and spine density caused by NVHL.
NVHL does not Affect the Excitability of Layer 5 mPFC Pyramidal Neurons
We next examined the excitability of layer 5 mPFC pyramidal neurons in sham (n = 14) and NVHL rats (n = 17) (Fig. 2). Examining the firing properties of these neurons at the current clamp mode revealed no differences in action potential threshold (−43.4 ± 1.8 mV in sham vs. −41.0 ± 1.4 mV in NVHL rats; P= 0.22). We also found no differences in action potential qualities including mean spike amplitude (56.0 ± 7.7 mV in sham vs. 65.2 ± 3.4 mV in NVHL rats; P= 0.17), first spike amplitude (66.9 ± 8.2 mV in sham vs. 75.5 ± 4.6 mV in NVHL rats; P= 0.26), and maximum spike amplitude (75.3 ± 4.9 mV in sham vs. 81.7 ± 2.4 mV in NVHL rats; P= 0.14). In addition, latencies of first spikes (365.4 ± 104.0 ms in sham vs. 382.4 ± 62.2 ms in NVHL rats; P= 0.87) of these 2 groups were similar. Finally, we found no changes in the repolarization parameter, slow after-hyperpolarization amplitude (−6.6 ± 0.6 mV in sham vs. −5.4 ± 1.1 mV in NVHL rats; P= 0.52).
NVHL does not Affect Spontaneous Action Potential-Dependent Excitatory and Inhibitory Postsynaptic Currents Recorded from Layer 5 mPFC Pyramidal Neurons
Excitability is influenced by the balance between excitatory and inhibitory synaptic inputs. We surmised that the functional consequences of dendritic spine reduction might be evident after isolating the excitatory and inhibitory components. To test this, we compared the amplitude and frequency of spontaneous action potential-dependent excitatory postsynaptic currents (sEPSCs) and inhibitory postsynaptic currents (sIPSCs) recorded from layer 5 mPFC pyramidal neurons of sham and NVHL rats. Despite the loss of dendritic spines caused by NVHL, neither the amplitude (14.1 ± 1.4 pA in sham vs. 14.5 ± 1.0 pA in NVHL rats; P= 0.83) nor the frequency (7.8 ± 0.7 Hz in sham vs. 6.5 ± 1.2 Hz in NVHL rats; P= 0.33) of sEPSCs recorded from sham rats (n = 13) was different from that of NVHL rats (n = 11) (Fig. 3). When we compared the amplitude (37.3 ± 1.54 pA in sham rats vs. 44.9 ± 11.4 pA in NVHL rats; P= 0.52) and the frequency (6.37 ± 0.52 Hz in sham rats vs. 5.92 ± 0.58 Hz in NVHL rats; P= 0.28) of sIPSCs between sham (n = 8) and NVHL rats (n = 8), we again found no significant difference (Fig. 4).
Increase in the Amplitude of Miniature EPSCs (mEPSCs) in Layer 5 mPFC Pyramidal Neurons of NVHL Rats
Given that NVHL alters the dendritic spine density of layer 5 mPFC pyramidal neurons, the stable action potential-driven sEPSCs we observed in lesioned rats are somewhat surprising. One potential explanation for the stable level of network-driven synaptic activities after synaptic loss is an up-regulation of function at individual excitatory synapses. To test this possibility, we compared TTX-insensitive mEPSCs recorded from layer 5 mPFC pyramidal neurons of sham (n = 10) and NVHL rats (n = 10) (Fig. 5). mEPSCs are action potential-independent and are not driven by neural network, but instead depend on spontaneous quantal release of glutamate from individual synaptic vesicles. Unlike action potential-dependent sEPSCs, we found that the amplitude of mEPSCs recorded from layer 5 mPFC pyramidal neurons of NVHL rats was higher than that of controls (7.4 ± 0.5 pA in sham vs. 8.9 ± 0.4 pA in NVHL rats; P= 0.043). Comparing the frequency (4.5 ± 0.7 Hz in sham vs. 6.6 ± 0.9 Hz in NVHL rats; P= 0.101) and the interspike interval of mEPSCs recorded from sham and NVHL rats revealed no significant changes. Finally, the kinetic properties of mEPSCs, including their rise time and decay time constant, remained unchanged after NVHL.
Decrease in the Frequency of Miniature IPSCs (mIPSCs) in Layer 5 mPFC Pyramidal Neurons of NVHL Rats
The increase in the mEPSC amplitude could functionally compensate for the loss of dendritic spines in NVHL rats. If these changes reflect compensatory mechanisms to maintain excitability, we predicted that NVHL rats would also exhibit attenuated inhibition. To test this, we compared the functional properties of mIPSCs recorded from layer 5 mPFC pyramidal neurons of sham (n = 9) and NVHL rats (n = 8) (Fig. 6). While we did not find changes in the amplitude of mIPSCs between these 2 groups (25.7 ± 2.3 pA in sham vs. 21.9 ± 1.6 pA in NVHL rats; P= 0.21), we found a significant decrease in the frequency of mIPSCs in NVHL rats when compared with sham rats (7.4 ± 0.4 Hz in sham vs. 5.8 ± 0.5 Hz in NVHL rats; P= 0.048). Nonetheless, we observed no changes in the kinetic properties of mIPSCs between sham and NVHL rats.
NVHL does not Affect the Expression of Presynaptic Markers of Glutamatergic and GABAergic Synapses in the mPFC
Finally, we asked whether the expression of presynaptic markers of glutamatergic and GABAergic synapses in the mPFC is altered by NVHL. Since vesicular transporters of glutamate (VGluT-1) and GABA (VGAT) accumulate glutamate and GABA into synaptic vesicles, respectively, these transporters are reliable presynaptic markers for excitatory and inhibitory synapses. Comparing the expression of these presynaptic markers in the synaptosomal fraction of mPFC lysate from sham and NVHL rats (Fig. 7), however, revealed no differences.
We found significant alterations in the dendritic morphology of layer 5 mPFC pyramidal neurons in NVHL rats, which included attrition of dendritic structures and reduced dendritic complexity. In addition, we observed a significant dendritic spine loss after VH lesion, suggesting a reduction in excitatory synapses. In keeping with these morphological deficits, we identified significant differences of activity in individual mPFC synapses in NVHL rats. The amplitude of mEPSCs was significantly increased, whereas the frequency of mIPSCs was significantly decreased. These alterations in basal synaptic activity may functionally compensate for the reduction of excitatory inputs in the mPFC after VH lesion, thus preserving normal excitability and action potential-driven synaptic activities in layer 5 mPFC pyramidal neurons.
Loss of Dendritic Structures of Layer 5 mPFC Pyramidal Neurons in NVHL Rats
The loss of dendritic structures and spines we observed in layer 5 pyramidal neurons of NVHL rats is consistent with the reported alterations in dendritic morphology of layer 3 pyramidal neurons in the mPFC of these rats (Flores et al. 2005; Marquis et al. 2008). The loss of dendrites in NVHL rats reduces postsynaptic membrane for both excitatory and inhibitory synapses. Since dendritic spines represent sites of excitatory synapses, a reduction in spine density would mean fewer excitatory synaptic inputs on mPFC neurons. While changes in dendritic morphology we observed in layer 5 pyramidal neurons share similarity with dendritic alterations in layer 3 pyramidal neurons in the mPFC of NVHL rats, our new findings have additional implications. Layer 5 pyramidal neurons provide the principal efferent of the cortex (McGuire et al. 1991; Berendse et al. 1992). Changes in dendritic structures of layer 5 mPFC pyramidal neurons could affect synaptic inputs to projected areas. Superficial and deep layers of mPFC also receive distinct synaptic inputs. For instance, while the layer 3 of mPFC receives primarily noradrenergic inputs, dopaminergic inputs are more common in the layer 5 of mPFC (Bunney and Aghajanian 1976). The attrition of basal dendrites of layer 5 mPFC pyramidal neurons could be responsible for the alterations of dopaminergic regulation of these neurons in NVHL rats (Tseng et al. 2007). Notably, reduced dendritic structures were reported not only in layer 3 pyramidal neurons of schizophrenia patients (Glantz and Lewis 2000), but also in layer 5 PFC pyramidal neurons (Black et al. 2004). Our findings therefore further support the use of NVHL rats to model deficits of prefrontal neuronal morphology in schizophrenia. Whether these structural changes contribute to the impairment of mPFC-related cognitive functions in NVHL rats (Lipska et al. 1995; Grecksch et al. 1999; Le Pen and Moreau 2002) remains unknown. Dendritic length of layer 5 PFC pyramidal neurons undergoes rapid growth during early postnatal periods in rats [first postnatal month (Zhang 2004)] and humans [from birth to 2.5 years (Petanjek et al. 2008)]. Our findings strongly suggest that lesioning the VH during dendritic development results in a lasting reduction in the dendritic length and complexity of layer 5 mPFC pyramidal neurons in postpubertal NVHL rats. In addition, we also provided evidence that reduction in dendritic structures in NVHL rats associates with changes in synaptic function.
Modification of the Functional Properties of Excitatory and Inhibitory Synapses in the mPFC After NVHL
The loss of dendritic structures and spines in layer 5 mPFC pyramidal neurons in NVHL rats coincided with an increase in the mEPSC amplitude. This functional change could have a compensatory role. Despite the loss of synaptic structures, both the action potential-dependent sEPSCs we recorded from layer 5 mPFC pyramidal neurons and the excitability of these neurons in NVHL rats remain unaltered. An increase in the amplitude of spontaneous mEPSC from remaining synapses could be responsible for maintaining a normal level of excitatory function in layer 5 mPFC pyramidal neurons after lesion. The increase in the mEPSC amplitude could be due to an increase in postsynaptic glutamate receptors, consistent with a report of increased glutamate binding in the frontal cortex of NVHL rats (Schroeder et al. 1999). While we observed no changes in the expression of a presynaptic marker VGluT-1 between sham and NVHL rats, a potential increase in the presynaptic function of glutamate synapses in NVHL rats is supported by a trend (P= 0.101) of increased mEPSC frequency in these rats. Although speculative, reduction of glutamate synapses as reflected by decreased dendritic spines and no changes in the overall VGluT-1 expression may suggest an increase in VGluT-1 in remaining mPFC glutamate synapses in NVHL rats, which could enhance presynaptic release of glutamate. While most of the excitatory synapses are formed on dendritic spines in the adult brain (Harris and Kater 1994), up to 50% synapses are formed on dendritic shaft in developing brain (Boyer et al. 1998). The unaltered or possibly increase in mEPSC frequency we observed in NVHL rats could be due to an increase of shaft synapses on layer 5 mPFC pyramidal neurons after lesion. It is presently unclear whether the deafferentation caused by NVHL affects the developmental profile of shaft synapses.
In addition to the increase in the basal excitatory tone, we found a significant decrease in the frequency of mIPSCs recorded from layer 5 mPFC pyramidal neurons in NVHL rats. This finding implies a state of reduced inhibition in the mPFC after lesion. Decreases in a number of GABAergic markers have been observed in the PFC of schizophrenia patients (Lewis et al. 2005). A decrease in the mIPSC frequency in layer 5 pyramidal neurons could be due to a loss of functional inhibitory presynaptic inputs after dendritic attrition, which reduces available postsynaptic membrane for making synaptic contacts in the mPFC of NVHL rats. Note that a previous report suggested delayed maturation of GABAergic interneurons in NVHL rats (Tseng et al. 2008). Whether the loss of inhibitory synaptic function we found in NVHL rats is related to developmental abnormality of interneurons remains unclear. Nonetheless, the expression of VGAT, a presynaptic marker of inhibitory synapses, in sham and NVHL rats was similar, suggesting that a loss of GABA-containing vesicles is not likely responsible for the reduction in mIPSC frequency in NVHL rats.
Impact of NVHL on the Excitability of Layer 5 mPFC Neurons
Our finding of unaltered excitability of layer 5 mPFC pyramidal neurons after NVHL parallels findings from a previous report (O'Donnell et al. 2002). The seemingly stable excitability of mPFC neurons after a loss of dendrites and spines in NVHL rats could be compensated by several mechanisms. As discussed previously, the opposing changes of miniature excitatory and inhibitory synaptic currents, which represent postsynaptic responses from quantal release of glutamate and GABA, respectively, could compensate the loss of synaptic structures in NVHL rats. However, other mechanisms could participate. The finding that action potential-dependent postsynaptic activities in NVHL rats remain largely intact when compared with sham animals (Figs 3 and 4) supports compensatory changes at the network level. For example, the loss of functional synapses could lead to higher activity-driven releases from remaining synapses. It should be noted, however, that since we measured the excitability of layer 5 mPFC pyramidal neurons in brain slices, it is not clear whether stable excitability of these neurons in NVHL rats can be maintained in vivo in the presence of extrinsic inputs from other brain regions.
Potential Impact of Changes in Miniature Synaptic Activities on the mPFC Function of NVHL Rats
In spite of their potential compensatory role, changes in miniature synaptic activities caused by NVHL could have a negative impact on mPFC function. Changes in miniature synaptic function could contribute to the altered prefrontal cortical neuron firing in response to neuromodulators in NVHL rats. For instance, the enhanced function of individual excitatory synapses as reflected by a rise in the mEPSC amplitude in NVHL rats could contribute to the enhanced excitability of mPFC in response to stimulation of the ventral tegmental area (O’ Donnell et al. 2002), which sends dopaminergic inputs to the mPFC that normally inhibit the firing of mPFC neurons (Lewis and O'Donnell 2000). Alternatively, the increase in the mEPSC amplitude may affect the formation of synaptic plasticity such as long-term potentiation (LTP). LTP in the mPFC has been implicated in various cognitive functions (Goto et al. 2010). Enhanced basal excitatory function after NVHL could limit the room for further strengthening of synaptic activity, which impairs LTP formation. The reduced basal activities of inhibitory synapses, however, suggest a reduction in functional inhibitory synaptic input in the mPFC after lesion. Although changes in miniature inhibitory synaptic inputs in NVHL rats in the layer 5 region seem to be compensated at the network level such that we observed no apparent difference in action potential-dependent sIPSCs between sham and NVHL rats, functional impairments caused by the reduction of basal inhibitory inputs may surface if extra inhibition is needed. For example, dopamine enhances the inhibitory postsynaptic responses in the mPFC (Seamans et al. 2001), and this dopamine-induced modulation is affected in NVHL rats (Tseng et al. 2008). Possibly, the reduced functional inhibitory synaptic inputs we observed in NVHL rats may contribute to the impairment of dopamine-induced enhancement of inhibitory synaptic function in these rats.
Implication of Changes in Basal Synaptic Activities in the mPFC After NVHL for Schizophrenia
The PFC in primates and humans undergoes extended period of developmental changes, which begin with overproduction in synapses in early postnatal development and are followed by pruning of synapses from puberty to adulthood (Rakic et al. 1994; Lewis 2009; Petanjek et al. 2011). These changes associate not only with developmental alterations in gene expression and metabolites in the PFC (Somel et al. 2009; Fu et al. 2011), but also with the maturation of cognitive processes such as working memory and cognitive control in adulthood (Luna et al. 2004; Bunge and Wright 2007). Available evidence suggests that developmental abnormality of the PFC could be responsible for the pathogenesis of cognitive deficits in schizophrenia, whose age of onset is close to puberty when the PFC reaches maturity. For instance, while the number of neurons in the PFC of schizophrenia patients remains unaltered (Akbarian et al. 1995; Thune et al. 2001), PFC pyramidal neurons show neuropil shrinkage, resulting in higher neuronal density and reduced cortical thickness (Selemon et al. 1998). These morphological changes are related to reduction in synaptic protein expression (Glantz and Lewis 1997) and spine density (Garey et al. 1998; Glantz and Lewis 2000). These findings are consistent with a hypothesis that excessive pruning of excitatory synapses could be responsible for the pathogenesis of schizophrenia (Keshavan et al. 1994). Using NVHL rats, we not only found that both layer 3 (Flores et al. 2005) and layer 5 pyramidal neurons (the present study) exhibit excessive pruning of dendritic structures and spines in the mPFC, but also revealed enhancement and suppression of the function of remaining excitatory and inhibitory synapses, respectively, in layer 5 pyramidal neurons.
What is the implication of opposing changes of miniature excitatory and inhibitory synaptic activities in the mPFC of NVHL animals for the cognitive deficits in schizophrenia? During the performance of tasks with low working memory load, schizophrenia patients have been shown to exhibit normal or even hyperactivation of PFC when compared with controls (Callicott et al. 2003; Manoach 2003), which could be due to opposing changes in basal synaptic function toward excitation that is similar to our findings obtained from NVHL rats. Nonetheless, high basal levels of excitation and a loss of inhibitory inputs in the mPFC may increase the levels of synaptic noise, which could affect the ability of mPFC neurons to detect extrinsic excitatory inputs conveying working memory task-related information. This reduction in signal-to-noise ratio in the mPFC could make NVHL rats, and potentially schizophrenia patients, prone to working memory errors.
This work was supported by a grant from the Canadian Institutes of Health Research (MOP68922) to L.K.S. and T.P.W.
The authors would like to thank Dr Rosemary Bagot for her comments on the manuscript and Drs Colleen Manitt and Xianglan Wen for their technical assistance of using “Neurolucida.” Conflict of Interest: None declared.