Dysfunctional epileptic neuronal circuits and dysmorphic dendritic spines are mitigated by platelet-activating factor receptor antagonism

Temporal lobe epilepsy or limbic epilepsy lacks effective therapies due to a void in understanding the cellular and molecular mechanisms that set in motion aberrant neuronal network formations during the course of limbic epileptogenesis (LE). Here we show in in vivo rodent models of LE that the phospholipid mediator platelet-activating factor (PAF) increases in LE and that PAF receptor (PAF-r) ablation mitigates its progression. Synthetic PAF-r antagonists, when administered intraperitoneally in LE, re-establish hippocampal dendritic spine density and prevent formation of dysmorphic dendritic spines. Concomitantly, hippocampal interictal spikes, aberrant oscillations, and neuronal hyper-excitability, evaluated 15–16 weeks after LE using multi-array silicon probe electrodes implanted in the dorsal hippocampus, are reduced in PAF-r antagonist-treated mice. We suggest that over-activation of PAF-r signaling induces aberrant neuronal plasticity in LE and leads to chronic dysfunctional neuronal circuitry that mediates epilepsy.

To analyze the effect of PAF-r antagonism on overall spine integrity, a correlation analysis was performed between the spine density and spine length of PAF-r antagonist-treated animals (Fig. 3). Dendritic spine density correlated positively with DS length in the OR, L-M and DG in LAU-0901-treated mice and vehicle-treated mice, such that lower spine density was associated with shorter spine length. This linearly-positive relation is justified by the fact that the slope of each regression line in the top and middle panels of Fig. 3 is significantly greater than 0 (every corresponding p value is < 0.05). Note that the Pearson Correlation Coefficients (VEH: OR r = 0.380 LM r = 0.386 DG r = 0.346 and LAU-0901: OR r = 0.291 LM r = 0.296 DG r = 0.399) indicate moderate correlations, by the rule of thumb, in large data sets. Conversely, DS density correlated negatively with DS length in LAU-09021-treated mice, such that higher densities were associated with shorter spine lengths, with the slopes of regression lines in the LM and DG being significantly less than zero (p < 0.05) but not significant in the OR (p = 0.117; Correlation Coefficients, OR r = − 0.204 LM r = − 0.366 DG r = − 0.291). In addition, we observed that dysmorphic filopodia-like formations from dendrites in vehicle-treated mice were prevented by LAU-0901 and LAU-09021 treatment (VEH: 4.36 ± 0.5; LAU-0901: 0.92 ± 0.2; LAU-09021: 1.46 ± 0.4; p < 0.0001 and 0.0002 respectively with VEH) (Figs 4a-c and 5).
At 105 days post-SE, animals were euthanized for spine density (Fig. 6c-e) and length ( Supplementary Fig. 4) measurements. At this time point, vehicle-treated animals continue to evidence a decrease in dentate gyrus spine density as compared to healthy controls. This loss in spine density appears to be attenuated in LAU-treated animals. These changes, at the major input region of the hippocampus, may aid in explaining the reduction in epileptiform spikes observed at this time point (additionally, LAU-09021 appears to modestly increase spine density in the stratum lacunosum moleculaire; no effects at 105 days were observed in the stratum oriens).
Treatment with LAU-09021 during epileptogenesis reduced the spontaneous power of delta and theta waves in the OR, pyramidal (PYR), and stratum radiatum (RAD) layers; reduced power of beta waves in the OR, increased the power of beta and gamma waves in the DG (Fig. 7a); and reduced high frequency oscillations (HFOs; LAU-09021: 0.5 × 10 −7 ; VEH: 0.26 × 10 −6 ; p < 0.0001) in the DG (Fig. 7b) of the dorsal hippocampus 110 days after SE. Also, hippocampal spontaneous epileptiform spikes were reduced in LAU-09021-treated mice (LAU-09021: 1.8 ± 0.5; VEH: 5.3 ± 0.3, p:0.01) (Fig. 7c). Since neuronal hyper-excitability is a component of the altered neuronal network in epilepsy, we evaluated the degree of induced hyper-excitability following the PTZ test at sub-convulsive doses ( Fig. 6d-f, Supplementary Fig. 3). LAU-09021-treated mice displayed a trend of higher latency for seizures compared to vehicle-treated mice (Fig. 6e) and showed reduced seizure severity (Racine's score: LAU-09021: 1.4 ± 0.7; VEH: 5.2 ± 0.7, p < 0.0001) associated with attenuation of electrical discharges in all hippocampal layers (Fig. 7e,g).   (b) LFP from hippocampal CA1 region: stratum oriens (OR), pyramidal layer (PYR), stratum radiatum (RAD) and dentate gyrus (DG) at multiple time scales. In VEH-treated animals, high amplitude inter-ictal spikes, predominant in RAD, propagate to other sub-fields; these spikes are reduced in LAU-09021-treated mice. (c) Schematic indicating spine density measurement. (d) Golgi stained dendritic spines of non-epilepsy, epilepsy control-treated, and epilepsy LAU-treated mice (scale bar indicates 5 μ m (e) Spine density (avg. ± SEM) in various hippocampal regions at 105 days post-SE. In the dentate, vehicle-treated animals show reduced spine density as compared to healthy controls. LAU-09021 restores spine density to the levels observed in healthy controls. In the stratum lacunosum moleculaire, a modest but statistically significant increase in spine density is observed in LAU-09021-treated animals, as compared to vehicle-treated and healthy controls. No differences are observed in the stratum oriens.  Discussion PAF accumulation in the brain after seizures 16,24 or SE (Fig. 1) could sustain glutamate release 35 and activation of COX-2 gene expression 36 , inducing production of prostaglandin E2 (PGE2) 37 . PGE2 regulates neuronal membrane excitability 38 , which synergistically boosts intraneuronal calcium mobilization mediated by PAF 39 and facilitates a hyper-excitable state in the neuronal network during epileptogenesis. Therefore, anti-epileptogenesis mechanisms could be mediated by PAF-r antagonism, indicating a novel alternative therapeutic approach to modulating the COX-2 signaling cascade 40 .
Control of the input and output of the hippocampal network 41 takes place during the functional activation of DS and somatic inhibition 42,43 . Also, DS formation is widely assumed to reflect structural reorganization of synaptic connections 44 . Dendritic spines, sites for excitatory synaptic transmission, play a major role in neuronal plasticity in epilepsy. Models of TLE show DS loss 11,29 and increased dendritic length in pyramidal cells 45 , suggesting a cellular mechanism of recovery for increasing synaptic contacts. This is important in maintaining homeostatic synaptic function or formation of abnormal structural reorganization that promotes circuitry reorganization and epileptogenesis. Reduced spine density is associated with reduction of spine length and has been observed in other disorders 46,47 . Here, we observed a positive correlation between DS density and length in vehicle and LAU-0901-treated mice, suggesting plasticity as a consequence of SE (Fig. 3). Since LAU-0901-treated animals display greater spine density than vehicle-treated mice (in the OR and DG), LAU-0901 could promote faster recovery of dendritic spines in those hippocampal regions critical for hippocampal connectivity with other brain circuits. Interestingly, a negative correlation between DS density and length was observed in LAU-09021-treated mice. We hypothesize that this scenario reflects use of an alternative homeostatic plasticity mechanism, potentially reflecting different timescales of recovery associated with different pharmacokinetics of the two LAU compounds, a topic for further research.
PAF-r antagonism may elicit protection of neuronal circuitry 48 in epileptogenesis by increasing post-synaptic contacts in the DG (Fig. 6e). This could improve input from the entorhinal cortex, which may be the origin of the observed increase in DG neuronal network activity (Figs 6b and 7a) and retraction of post-synaptic contact in the stratum lacunosum-moleculare. Such an effect would limit CA3 input from Schaffer collaterals, as reflected by depressing neuronal network activity in the stratum radiatum (Fig. 7a). As a result of this restructuring of the hippocampal circuitry, PAF-r antagonism attenuates onset and propagation of epileptiform activity in the CA1 hippocampal activity (Figs 6b and 7a). On the other hand, LAU-09021 initially facilitates spine recovery, increasing the length rather than the number of spines at 5 days PSE. At 105 days PSE, however, variance in spine length appears diminished, whereas small but significant changes in spine density are still observed. Although a percentage of the observed differences are small in absolute terms, there may be synergistic effects that represent a cumulative benefit. Furthermore, while the biological significance of dendritic spine density and morphometry is well established, it is likely that these parameters vary over a small range such that epileptogenesis cannot instigate total DS loss, inherently limiting the signal window for PAF-r antagonist rescue. However, based on the current observations, we hypothesize that during epileptogenesis there is a dynamic restructuring of dendritic spines that could involve different waves of cellular mechanisms for mediating growth of dendritic spines (LAU-09021) as well as spinogenesis (LAU-0901). If epileptogenesis is interrupted by treatment, the long term cellular strategy of homeostatic plasticity may shift to spinogenesis (as compared to spine growth), as we observed at 105 days post-status epilepticus. Further studies at different time points in epileptogenesis will be required to determine if PAF-r antagonism mediates such dynamic changes in dendritic spines.
Unexpectedly, we observed aberrant filopodia-like spines in vehicle-treated mice (Fig. 4a); these spines were prevented by PAF-r antagonism in the LAU-09021-and LAU-0901-treated mice. At present there are no scientific reports showing that such aberrant dendritic spines occur in epileptogenesis; however, spine alterations are present in developmental and mental disorders 49,50 . The structural variation in spines observed in our studies could contribute to the aberrant formation of circuitry and co-morbidities of epilepsy, in addition to disruptive behavior and cognitive deficits. Moreover, prevention of the formation of these aberrant spines by PAF-r antagonism could be involved in reduction of seizure susceptibility and hippocampal hyperexcitability 17 (Figs 6 and 7). These results provide anatomical and electrophysiological evidence that aberrant dendritic spine formation in epileptogenesis could have an impact on late excitability of neuronal circuitry.
Spontaneous epileptiform events 51-57 are associated with brain hyper-excitability 58,59 and reflect aberrant neuronal projections present in TLE 32 . Delta-band activity (0-4 Hz) is characterized by slow waves during slow-wave sleep and drowsy brain states. The elevation of slow-wave activity, especially delta-band activity, is correlated with behavioral changes in simple-partial and complex-partial seizures 60,61 . Large amplitude (1-2 Hz) slow activity can also occur in the frontal and parietal neocortices during (ictal) and immediately following (postictal) temporal lobe seizures 61 . Power spectral densities for delta and theta waves were significantly lower in LAU-09021-treated mice three months after SE. This may signify that formation of an aberrant neuronal network in epileptogenesis is limited by PAF-r antagonism and, as a consequence, decreases the establishment of epileptic neuronal circuitry.
High-frequency oscillations at > 100 Hz have been recorded in cortical structures of humans and other animals, both under physiological conditions and during partial epilepsies 62 . Intracranial electroencephalography recordings obtained from pharmaco-resistant patients suffering from TLE have shown that HFOs are observed in tandem with an interictal spike 63,64 . LAU-09021 significantly reduced HFOs mainly in the DG, suggesting that LAU-09021-treated animals experience attenuation of hippocampal recurrent neuronal circuitry 65 .
Furthermore, beta and gamma waves are depressed after seizures 66 . We observed that LAU-09021 intervention (five days PSE) reduced spontaneous epileptiform activity. Beta and gamma oscillations in the hippocampal CA3 area are modulated by aberrant GABA-mediated neurotransmission from the DG 66 . LAU-09021-treated mice had an increase in beta and gamma oscillations in the DG compared to vehicle-treated animals. The possibility Scientific RepoRts | 6:30298 | DOI: 10.1038/srep30298 that LAU-09021 could facilitate GABAergic transmission from the DG to the CA3 67 after seizures in vivo should be explored more in depth.

Conclusion
We conclude that an increase in PAF activates PAF-r in the hippocampus during epileptogenesis, thus mediating neuronal network hyper-excitability and seizure susceptibility. By blocking PAF-r and using the PAF-r antagonist during epileptogenesis, aberrant connectivity in the hippocampus was limited and, as a consequence, onset of epilepsy was reduced. We speculate that PAF-r activity could mediate aberrant connectivity in epileptogenesis.
Taken together, our observations suggest that the neuronal circuitry in the epileptic brain 41 is enhanced by PAF-r over-activity during epileptogenesis. More experimental studies need to be conducted to elucidate the molecular and neurotransmission-related mechanisms involved in this process. Furthermore, understanding PAF antagonism and the potential therapeutic usefulness of PAF receptor antagonists is relevant to developing disease-modifying therapeutic interventions for patients at risk for epilepsy.

Methods
Animals. Studies were performed according to National Institutes of Health guidelines and in accordance with nationally accepted principles in the care and use of experimental animals. The Institutional Animal Care and Use Committee (IACUC) at the Louisiana State University Health Sciences Center (LSUHSC), New Orleans, approved the animal protocols used for this study. Water and food were available for ad libitum consumption. All efforts were made to minimize pain and suffering and to reduce the number of mice used in these experiments. For euthanasia, animals were deeply anesthetized with ketamine hydrochloride and xylazine (200 mg/kg + 10 mg/kg; i.p.) prior to death by decapitation.
A total of 122 C57BL/6 adult male mice (20-25 g; Charles River Labs, Wilmington, MA) and 12 adult male PAF-r ( −/− ) and 5 wild type were used in this study. For PAF-r −/− : Donor strain: 129P2/OlaHsd via E14.1 ES cell line, genotype a/aB/BC/C. The mice were developed by Dr. Satoshi Ishii (University of Tokyo, 1998). A neomycin cassette was inserted into the open reading frame of the Ptafr gene. The mutant mice were backcrossed to C57BL/6. Colony maintenance was conducted at the LSUHSC-Neuroscience Center of Excellence and backcrossed to C57BL/6 (Heterozygote x C57BL/6NCrlCrlj). PAF-r −/− mice were bred in-house in our animal facility. For this study, mice were anesthetized, tailed and genotyped as described below. Briefly, genomic DNA preparation was performed by digesting 5 mm tail tips in a mixture of proteinase K and DirectPCR (Tail) reagent (Viagen, Los Angeles, CA) following the manufacturer's directions. The lysates were spun down and 2 μ l of each genomic preparation were used for PCR analysis. PCR were carried out using illustra PuReTaq Ready-To-Go PCR beads (GE Healthcare, Buckinghamshire, UK) with the following conditions: 94 °C 180 s (1 cycle); 94 °C 30 s, 56 °C 30 s, 72 °C 60 s (30 cycles) and, 72 °C 300 s (1 cycle). Primers used for the analysis were as follows: mPAFR Forward 5′ -CTCCCACTGTGGATTGTCTACTACT-3′ ; mPAFR Reverse 5′ -AAGATAAGGAAGAAGACGAGGAAGA-3′ ; Neo CASSETTE 5′ -CTATCAGGACATAGCGTTGGCTAC-3′ . The combination of the primers used to detect wild-type allele were mPAFR Forward and Reverse, and for the PAF-r −/− allele was mPAFR Forward and Neo CASSETTE. PCR for both alleles was run in separate reactions retrieving a band of approximately 400 bp for WT and 1000 bp for Mutant revealed in a 1% agarose gel stained with Ethidium Bromide.
Pentylenetetrazol test. Systemic administration of pentylenetetrazol (PTZ), a model used to test potential anti-convulsive effects 27 , was used as an initial step to further evaluate the spectrum of activity of PAF-r antagonist compounds. Based on our preliminary experiments, we observed that: a) intraperitoneal (i.p.) administration of PTZ at 35 mg/kg is useful as a sub-convulsive dose to evaluate non-convulsive seizures, and b) the PAF-r antagonist LAU-0901 has a physiological effect on the brain for 4-6 hours after injection and has no apparent adverse effects 17,68 . Animals were placed in individual Plexiglass cages (28 × 28 × 37.5 cm) and given a single dose of PTZ (Sigma, St Louis, MO) at 35 mg/kg/i.p. or 70 mg/kg/i.p. to evaluate effectiveness of the PAF-r antagonist compounds against seizures or to determine chronic hippocampal hyper-excitability and seizure susceptibility as a consequence of LE. Animal recordings (90-minutes in length) were made using a video-recording system (Handycam Sony); recordings began immediately after PTZ injection. At the end of the experiment, animals were euthanized. Locomotor seizures were quantified according to a modified Racine's Score 69 , as follows: 0, normal behavior-walking, exploring, sniffing, grooming; 1, immobile, staring, jumpy, curled-up posture; 2, automatisms-repetitive blinking, chewing, head bobbing, vibrissae twitching, scratching, face washing, "star gazing"; 3, partial-body clonus, occasional myoclonic jerks, shivering; 4, whole-body clonus, "corkscrew" turning and flipping, loss of posture, rearing, falling; 5, non-intermittent seizure activity; and 6, wild running, bouncing, tonic-clonic seizures.
Post-status epilepticus model of epileptogenesis. Status epilepticus was induced by a single dose of pilocarpine hydrochloride (250 mg/kg) (Sigma Aldrich, St. Louis, MO) administered i.p. 30 minutes after methyl scopolamine nitrate (1 mg/kg; i.p., Sigma Aldrich). Animals were placed in individual Plexiglass cages and monitored by laboratory personnel during and after SE. Seizures were rated according Racine's score. Non-intermittent seizure activity, stages 3 and/or 4, for each mouse was limited to 90 minutes using a single dose of diazepam (10 mg/kg, i.p., Sigma Aldrich). Each animal was monitored by trained laboratory personnel in a temperature-controlled surgical room until full locomotor recovery was observed (2-4 hours). Surviving animals (60%) were randomized by number assignment and placed in individual cages in an animal room with an artificial 12-hour light/dark cycle with access to food and water ad libitum.
Local field potential recordings and analysis. For local field potential (LFP) analysis, a silicon probe with 16 electrodes (spacing 100 μ m, NeuroNexus, Ann Arbor, MI) was implanted in the right dorsal hippocampus of each mouse (from Bregma: 1.80 mm posterior; 1-1.5 mm lateral and 2.70 mm depth) 98 days after SE under anesthesia induced by a mixture of ketamine (200 mg/kg) and xylazine (10 mg/kg) (Vedco Inc., Saint Joseph, MO) using a surgical microscope and sterilized neurosurgical instruments. Briefly, during surgery the probe was placed on superficial layers of the cortex and then moved inward slowly with the aid of stereotaxic equipment (Kopf Instruments, Tujunga, CA). The resulting hole was covered by Surgicel (Ethicon Inc., San Angelo, TX) and saturated with a sterile cerebral spinal fluid (Harvard Apparatus, Holliston, MA). A stainless steel screw (Plastic One, Roanoke, VA) was implanted in the occipital bone as a ground wire for the silicon probe. Plastic One gel (Plastic One) was used to attach the probe and screw it to the skull. After recovery from anesthesia, mice were placed for recordings in individual Plexiglass cages and allowed to explore freely; food and water were provided ad libitum. LFP from the hippocampi were recorded from the headstage connected with the probe, amplified (1000× ), band-pass filtered (0.1-300 Hz), and digitized at 1 KHz using MAP system data acquisition 7 days after surgery. Briefly, continuous LFP activity (4-5 minutes) from each freely-moving mouse was recorded and sampled (10-12 samples/hour) every 5 minutes from 10:00 a.m. to 4:00 p.m. using a MAP (Plexon) and video-recorder system for 5 consecutive days. LFPs were analyzed using NeuroExplorer (Next Tech Solutions, Inc., Austin, TX). Delta epochs (3-6 seconds each) from each LFP were determined by calculating the ratio of delta and theta frequency bands in the hippocampal CA1 region 70 , without artifacts or noise, by an investigator blinded to the treatment and confirmed during a period of immobility by visual inspection of video-LFP recordings. Band frequencies for delta (0.1-3.9 Hz), theta (4-8 Hz), beta (13-20 Hz), low gamma (21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38)(39)(40), and bands from 200-300 Hz were selected from the LFPs using power spectral density (PSD) analysis (NeuroExplorer, Next Technologies, Madison, AL) 14 . Then, average values of PSD of each band were compared from samples (n = 42 each group) of the PAF-r antagonist LAU-09021-and vehicle-treated animals. For automatic assessment of high frequency spikes the signal was filtered at 250-300 Hz band and analyzed using the offline sorter threshold function. Successive spike amplitude activity above the baseline was quantified using NeuroExplorer combined with offline sorter software. Artifacts such as head movement or grooming were excluded by visual inspection of video-LFP recordings. At the end of the experiment, verification of the probe placement in the dorsal hippocampus was confirmed by histology 13 for anatomic-physiological correlation of hippocampal layers.
Rapid kindling model of epileptogenesis. Bipolar electrode units (Plastic One Inc., Roanoke, VA, U.S.A.) were implanted in the dorsal right hippocampus (coordinates: 2.3 mm caudal to bregma; 1.75 mm lateral to midline; 2.00 mm ventral to dura) and ground wire was placed on the occipital bone, guided by stereotaxic procedure under anesthesia induced by intraperitoneal injection of a mixture of ketamine hydrochloride: xylazine (200 mg/kg: 10 mg/kg; Vedco). One week after surgery, kindling was achieved by sub-convulsive electrical stimulation at 30-min intervals (six stimulations per 10-s train containing 50-Hz biphasic pulses of 100-lA amplitude. To test the seizure susceptibility responses, sub-convulsive electrical stimulations (as during kindling) were given 1 week after kindling (rekindling).
In situ hybridization. Mice were deeply anesthetized and brains were dissected according to previous procedures 17 , and a 385 bp fragment from the coding region of the PAF-r mRNA was amplified from mouse brain cDNA (Clonotec) by PCR and inserted into pCRrII TOPO (Invitrogen). 35 S-labeled riboprobes were transcribed in sense and antisense directions.
Liquid chromatography tandem mass spectrometry. PAF concentration in hippocampal samples was calculated as follows: brain enzymes were inactivated using high-powered microwave irradiation (10 KW, 400 V, 750 ms) focused on the head after anesthesia induced by isofluorane, followed by rapid immersion of the mouse head in ice water. Protein precipitates were separated by centrifugation, and solvent extracts were pre-equilibrated at pH 3.0 in 10% methanol/water, loaded to 500 mg C18 columns (Varian, Palo Alto, CA, U.S.A.), and then eluted with 1% methanol/ethyl acetate. Eluates were concentrated on an N 2 stream evaporator. Samples were loaded to a liquid chromatograph-tandem mass spectrometer (LC-MS-MS; LC-TSQ Quantum, Thermo-Finnigan, Waltham, MA, U.S.A.) installed with a Biobasic-AX column (Thermo-Hypersil-Keystone, Bellefonte, PA, U.S.A.) (100 · 2.1 mm, 5-lm particle sizes). Samples were eluted in a linear gradient [100% solution A (40:60:0.01 methanol/water/acetic acid pH 4.5) to 100% solution B (99.99:0.01 methanol/acetic acid)] at a flow rate of 300 ll/min for 30 min. LC effluents were diverted to an electrospray-ionization probe (ESI) on a TSQ Quantum (Thermo-Finnigan) triple quadrupole mass spectrometer 13 ; lipid standards (Cayman Chem., Ann Arbor, MI, U.S.A.) were used for tuning, optimization, and calibration curves. The instruments were set on full-scale mode to detect parent ions and also were set to selected-reaction mode (SRM) for quantitative analysis to detect product ions simultaneously. The selected parent/product ions (m/z) and collision energy (v) obtained on negative ion detection mode were 370.0/171.3/2 for PAF 17 . Dendrite spine detection and analysis. Brains were processed following established procedures according to the manufacturer's instructions (FD Rapid GolgiStain Kit, FD Neurotechnologies, Inc., Columbia, MD). Coronal sections (40 μ m) were made and then mounted, air-dried, dehydrated in alcohol, cleared in xylene and cover-slipped. Coronal sections were scanned using brightfield microscopy and magnified at 100X. In addition, individual areas were photographed at 40× and then apical and basal dendrites from Golgi-impregnated neurons were selected from treated mice using imaging application software (OlyVIA, Olympus, Center Valley, PA). Z-stacks (step size = 0.3 μ m) of dendrites from the stratum oriens (OR), lacunosum-moleculare layers (L-M) of the CA1 and outer molecular layer (OM) from the dentate gyrus (DG) were captured using a 100× /oil objective. Images were recorded using an Axioplan 2 microscope (Carl Zeiss Inc., Thornwood, NY) coupled with AxioCam and Axiovision software (Carl Zeiss Inc.). Dendrite density and length of dendritic spines were quantified with Image J (National Institute of Health). A minimum of 10 dendrites were calculated per animal for the following regions: OR, L-M and DG. We then calculated the number of dendritic spines per segment of individual dendrites per hippocampal subfield for each treated group of animals.
Statistics. The data retrieved from each experiment were averaged and expressed as mean ± S.E.M. For statistical significance, one time point (or more than two observations) was analyzed using Student's t-Test and ANOVA followed by post-hoc tests (Tukey-Kramer and Hsu'MCB). Correlation analysis was performed for spine density (number of spines per 10 μ m of dendrite segment cell and dendritic spine length) using Pearson's correlation analysis. A p-value < 0.05 was considered significant. All data analyses, including sample size were conducted using JMP 8.0 statistical software from SAS (Cary, NC).