Isoflurane Disrupts Postsynaptic Density-95 Protein Interactions Causing Neuronal Synapse Loss and Cognitive Impairment in Juvenile Mice via Canonical NO-mediated Protein Kinase-G Signaling

Neurobasal media (21103-049), B-27 supplement (17504-044), penicillin and streptomycin (15140-122), Hank’s Balanced Salt Solution (4185-052), N-(2-hydroxyethyl) piperazine-N’-2-ethanesulfonic acid (15630080), glucose (A24940-01), horse serum (26050-088), and phosphate-buffered saline were obtained from Gibco (USA). Cytosine arabinoside (C6645), 8-bromoguanosine 3′,5′-cyclic monophosphate sodium salt (8-Br–cGMP [B1381]), 3-(5’-hydroxymethyl-2’-furyl)-1-benzyl indazole (YC-1 [Y102]), laminin (L2020), poly-D-lysine hydrobromide (P6407), bovine serum albumin, and paraformaldehyde were from Sigma-Aldrich (USA). DETA NONOate (82120) was from Cayman Chemicals (USA). 1H-[1,2,4] oxadiazolo[4,3-a]quinoxalin-1-1 (ODQ [0880]) and KT5823 (1289) were from Tocris Bioscience (United Kingdom). The Papain Dissociation kit (LK003150) was obtained from Worthington (USA). Anti-drebrin (ab11068) and Anti-synaptophysin (ab32594) antibodies were from Abcam (USA). microtubule-associated protein-2 (MAP-2; AB15452) and anti-NMDAR2A&B (AB1548) were from Millipore (USA). Mouse monoclonal antibody to PSD-95 (cat. 75-028) was from Antibodies Incorporated (USA) and anti-rabbit and anti-mouse horseradish peroxidase–conjugated secondary antibodies were purchased from Jackson ImmunoResearch Laboratories (USA). GG-12-1.5-Laminin–coated German coverslips were from Neuvitro Corporation (USA). Protein ladder, Alexa Fluor 568 (anti-rabbit) and 488 (anti-chicken), and ProLong Diamond Antifade Mountant (P36961) were purchased from Invitrogen (USA). VECTASHIELD antifade mounting medium with 4′,6-diamidino-2-phenylindole (H-1200-10) was purchased from Vectashield (USA). Pierce BCA Protein Assay Kit, Pierce Co-Immunoprecipitation Kit, and Halt Protease and Phosphatase Inhibitor Cocktail were from Thermo Fisher Scientific (USA). The purified fusion peptides, active Tat–PSD-95 wild-type PDZ2 peptide (referred to as PSD-95 wild-type PDZ2 peptides and wild-type PDZ2 peptide), and inactive Tat–PSD-95 mutant PDZ2 peptide (referred to as PSD-95 mutant PDZ2 peptide and mutant PDZ2 peptide in this manuscript) were purchased from Creative BioMart (USA) and used as previously described.¹⁹  10× Tris-buffered saline (1706435), 4× Laemmli sample buffer, 4 to 15% Criterio TG Precast Midi Protein Gels, and 10× Tris/glycine/sodium dodecyl sulfate buffer (1610732) were purchased from Bio-Rad (USA).

This study was implemented with approval from the Animal Care and Use Committee at Johns Hopkins University (Baltimore, Maryland) and was consistent with the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health (Bethesda, Maryland). Wild-type C57BL/6 mice of both sexes (The Jackson Laboratory, USA) were maintained on a 12:12 h light:dark cycle, on corncob bedding (7097 Teklad, Envigo, USA), with ad libitum access to chlorinated reverse-osmosis water and rodent diet (2018sx Teklad global 18% protein, Envigo, USA), and environmental enrichment in the form of Nestlets (Ancare, USA). Water and food were available ad libitum until mice were transported to the laboratory approximately 1 h before the experiments. For in vivo experiments, postnatal day 7 mice were randomly assigned to control and treatment groups. For in vitro experiments, postnatal day 0 to 1 litters were collected for preparation of primary neuronal cultures.

Neonatal mice at postnatal day 7 were exposed to 1.5% isoflurane in 50% O₂ for 4 h at a flow rate of 3 l/min, with airflow as the carrier (isoflurane group). Control animals were exposed to 50% O₂ (oxygen control group). All the mice were placed in a clear plastic cones and body temperature maintained by a heating pad set to 37°C. The purified fusion peptides, active PSD-95 wild-type PDZ2 peptide, or inactive PSD-95 mutant PDZ2 peptide at 8 mg/kg were injected into mice at postnatal day 7 intraperitoneally in 150 μl of phosphate-buffered saline and 10% glycerol, as previously described.¹⁹  Four hours after gas exposures or peptide injections, mice were decapitated, and hippocampal regions were dissected from the brain. The hippocampal tissue was pooled from two mice in every group for each independent experiment to attain sufficient protein for coimmunoprecipitation. For the behavior experiments, mice at postnatal day 7 were randomly assigned to four groups: (1) oxygen control + vehicle, (2) oxygen control + YC-1, (3) isoflurane + vehicle, and (4) isoflurane + YC-1. Mice in the oxygen control + YC-1 and isoflurane + YC-1 groups received 2 mg/kg of intraperitoneal injections of YC-1 at the time of isoflurane exposure. The mice in the oxygen control + vehicle and isoflurane + vehicle groups received 1% dimethyl sulfoxide. The solution of YC-1 was prepared in 1% dimethyl sulfoxide and was injected in the cohorts as previously described.^26,27 

Brain hippocampi were microdissected from postnatal day 0 to 1 pups. Then blood vessels and meninges were removed in ice-cold dissecting solution (1× Hank’s Balanced Salt solution, 1× penicillin/streptomycin, 1 mM sodium pyruvate, 10 mM N-2-hydroxyethylpiperazine-N-2-ethane sulfonic acid, and 30 mM glucose). The brain tissue was chopped into smaller pieces and digested using the Papain Dissociation kit according to the manufacturer’s instructions (Worthington, USA). Gently minced hippocampal tissue was incubated with Papain-deoxyribonuclease solution (20 U/ml papain and 0.005% deoxyribonuclease) for 30 min at 37°C with slow agitation. The suspension was triturated with a sterile fire-polished glass pipette and then centrifuged at 4°C for 5 min at 300g. The pellet was resuspended in Earle’s Balanced Salt Solution containing ovomucoid protease inhibitor with bovine serum albumin and deoxyribonuclease. Next, a discontinuous density gradient was prepared by pipetting the cell suspension onto a 5-ml layer of albumin-ovomucoid inhibitor solution. The gradient was centrifuged at 70g for 6 min at room temperature. The supernatant containing noncellular debris was discarded, and the cell pellet was resuspended in neurobasal medium supplemented with 1× GlutaMAX, 2% B27, 1% penicillin–streptomycin, and 1% heat-inactivated horse serum. Cells were seeded on poly-D-lysine– and laminin-coated glass chamber slides at a plating density of around 2.5 to 3.5 × 10⁴ cells per well and grown in a humidified carbon dioxide incubator at 37°C. Cytosine arabinoside (2 μM) was added to the cultures on the third day after plating to inhibit the proliferation of glial cells. Half of the culture medium was changed every alternate day. The purity of the primary neuronal cells was assessed with neuronal marker MAP-2.

At 7 days in vitro, neuronal cells were placed in humidified airtight gas chambers and exposed to 1.5% isoflurane in a gas of 25% O₂ and 5% CO₂ that was balanced with N₂. Control neurons were exposed only to carrier gas. The gas chambers were carefully monitored with a digital gas monitor. The chambers were placed in a standard 37°C incubator for 4 h. Neuronal cells were treated with pharmacologic inhibitors and activators to delineate the downstream signaling mechanisms in response to isoflurane or synaptic PSD-95 PDZ2 domain disruption.

Neuronal cell viability was measured with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay according to the manufacturer’s protocol (ab211091, Abcam, United Kingdom). Briefly, we seeded hippocampal neuronal cells around 1 to 2 × 10⁴ cells/well in triplicate onto Poly-D-Lysine-laminin–coated 96-well plates. At 7 days in vitro, cells were exposed and treated for 4 h with vehicle gas and 1.5% isoflurane or 1 μM inactive PSD-95 mutant PDZ2 peptide and 1 μM active PSD-95 wild-type PDZ2 peptide. Absorbance was measured at 590 nm on the Spark Multimode microplate reader (Tecan, Switzerland) to determine neuronal cell viability. Wells containing only media were used for blank absorbance reading. The cell viability was expressed as a percentage normalized relative to the viability in untreated cells.

Primary neuronal cells were fixed in 4% paraformaldehyde solution for 10 to 15 min, washed in phosphate-buffered saline three times (5-min intervals), and then blocked with blocking solution (2% normal goat serum + 1% bovine serum albumin + 0.1% Tween-20) in phosphate-buffered saline for 45 min at room temperature. Then cells were incubated with primary antibodies to MAP-2, (anti-chicken, 1:500), drebrin (anti-rabbit, 1:250), synaptophysin (anti-rabbit, 1:500), and PSD-95 (anti-mouse, 1:500) in blocking solution overnight at 4°C. Cells were washed three times in phosphate-buffered saline (5-min intervals) and incubated with Alexa Fluor–conjugated secondary antibody cocktails in phosphate-buffered saline containing 0.2% bovine serum albumin for 2 h at room temperature. After incubation, cells were washed five times in phosphate-buffered saline to remove excess secondary antibody and then mounted with ProLong diamond mounting medium. Images were captured using a LSM880-Airyscan fast super-resolution laser scanning confocal microscope (Zeiss, Jena, Germany) and Leica SPE DMI8 (USA).

Primary hippocampal neurons were fixed and coimmunostained with drebrin and MAP-2 antibodies. Images were taken using Zeiss LSM880-Airyscan fast superresolution laser scanning confocal microscope. All images were taken using Plan-Apochromat 63×/1.4 oil differential interference contrast. The image size scaling (per pixel) is around 0.06 μm × 0.06 μm × 0.19 μm. We analyzed drebrin-positive dendritic spines using Imaris software (Bitplane, Oxford Instruments, Bognor Regis, UK). Spines that emanated directly from the MAP-2–positive dendritic shaft were analyzed, and spines whose structure was not completely visible were excluded. For each neuron, spines along secondary or tertiary dendrites were imaged and selected for spine density quantification. Dendritic branches were reconstructed by using Imaris and were analyzed for spine density and length. Spine densities were calculated by quantifying the number of spines per 10 μm length of dendrite. Four to five independent cultures were performed per experimental condition (i.e., per group). In total, 25 to 40 neurons were analyzed per experimental condition for spine analysis. The results are expressed as mean spine density per 10 μm dendrite length. Blinding was performed such that all the groups were labeled differently at the time of imaging and analysis.

Mice were euthanized by decapitation, and hippocampal regions were dissected from the brain under a dissecting microscope. Total protein was extracted and subjected to coimmunoprecipitation with the Pierce Co-Immunoprecipitation Kit (Thermo Scientific, USA) according to the manufacturer’s instructions. In brief, the hippocampal tissue was dissected and homogenized in a bead homogenizer in the presence of ice-cold Radio-Immunoprecipitation Assay buffer (0.025 M Tris, 0.15 M NaCl, 0.001 M EDTA, 1% NP-40, 5% glycerol, pH 7.4) supplemented with a protease and phosphatase inhibitor cocktail. The crude homogenates were centrifuged at 16,000g for 20 min at 4°C. The lysates were collected, and the protein concentration estimated with the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, USA), according to the manufacturer’s instructions. The protein lysates were precleaned in a column of agarose resin that was incubated at 4°C for 45 min with gentle end-over-end mixing. Next, the column was centrifuged at 1,000g for 1 min. The flow-through containing the cleaned protein lysate was collected and the column discarded. Equal amounts of protein were incubated with 5 μg of affinity-purified rabbit-NMDAR2A&B antibody for 4 h and then immunoprecipitated with protein A/G agarose beads. Beads were washed three times with lysis buffer for 1 min each before immunoprecipitation. The protein was then resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotted through Western blotting. Briefly, proteins from the gel were transferred to polyvinylidene difluoride membranes (Millipore, USA). The membranes were blocked in Tris-buffered saline + 0.1% Tween-20 containing 5% nonfat milk for 1 h at room temperature, and then incubated with primary antibodies to PSD-95 (1:1,000) or NMDAR2A&B (1:1,000) in Tris-buffered saline + 0.1% Tween-20 containing 5% nonfat milk overnight at 4°C. Membranes were washed three times for 10 min each in Tris-buffered saline + 0.1% Tween-20 and then incubated for 2 h with horseradish peroxidase–conjugated anti-rabbit or anti-mouse immunoglobulin at a dilution of 1:5,000. Finally, the membranes were washed to remove excess secondary antibody. Proteins were visualized with enhanced chemiluminescence (Amersham, USA) and images captured with the ChemiDoc Imaging System (Bio-Rad, USA). PSD-95 coimmunoprecipitate was quantified relative to NMDAR2A&B precipitate using Bio-Rad Quantity One image analysis software.

cGMP-dependent protein kinase activity was determined with the CycLex assay kit (MBL International, USA) according to the manufacturer’s protocol. Briefly, primary hippocampal neurons were plated in six-well plates at 1.5 to 2 × 10⁶ cells per well and treated with gases or peptides along with pharmacologic activators and inhibitors. Then the neuronal cells were lysed with freeze-thaw cycles before being centrifuged for 20 min at 14,000g. Finally, the lysates were added to wells coated with recombinant G-kinase substrates. The absorbance was measured from 450 nm to 550 nm on the Spark Multimode microplate reader.

Synapse quantification was performed using custom-written plugin Puncta Analyzer for quantification of colocalized synaptic puncta as described in a study published by Ippolito and Eroglu.²⁸  Briefly, primary neuronal cells were fixed in 4% paraformaldehyde and coimmunostained for presynaptic marker synaptophysin, postsynaptic marker PSD-95, and neuronal marker MAP-2. Confocal images were taken using Zeiss LSM880-Airyscan fast super-resolution laser scanning confocal microscope. All images were taken using Plan-Apochromat 63×/1.4 Oil differential interference contrast with scaling x = 0.043 μm and y = 0.043 μm. The image size was x, 73.9 μm, and y, 73.9 μm in single plane. MAP-2–positive neurons with soma and dendrites were centered in the field of view and were imaged. Synapses were quantified by analyzing colocalization of synaptophysin and PSD-95 puncta using a plugin puncta analyzer in ImageJ (National Institutes of Health, USA). The background of all the images was removed separately from the red and green channel using the rolling ball radius of 50. Next, thresholding was performed for detecting discrete puncta, and puncta were identified in red and green channels using the Puncta Analyzer plugin. Colocalized puncta were counted as a synapse.²⁸  Four independent cultures were performed per experimental condition (i.e., per group). In total, 30 to 40 neurons were analyzed per experimental condition for synapse quantification. For each independent culture, similar regions of interest including soma and all the dendritic fragments of neurons were assessed for colocalization as previously described.²⁹ 

The novel object recognition procedure was performed as reported in our previous studies^17,18  and was used to evaluate nonspatial hippocampal memory.^30,31  Briefly, it was composed of two sessions, a familiarization session during which mice were allowed to freely explore two similar objects within an opaque box (40 cm width × 40 cm length × 34 cm height) for 10 min, followed by a second session after 2 h, which was a test session, in which one of the objects was replaced by a novel or unacquainted object. Mice innately prefer to explore the novel object rather than the familiar one, and a preference for the novel object demonstrates intact memory for the familiar object. After 2 h, the object recognition was tested, where a novel object was substituted for one of the familiar training objects and then mice were allowed to explore for 5 min. Object investigation time was determined by the amount of time the mouse spent in the zone immediately surrounding the object. Only mice that investigated any object for at least 10 s (criterion) were taken into consideration. During the testing phase, mice (n = 8) did not meet the criterion and so were not included in the analysis. Data were recorded with a video camera, and time spent with each object was analyzed by ANYmaze software (Stoelting, USA).

All data were tested for normality using the Shapiro–Wilk test and then presented as mean ± SD. The paired or independent unpaired two-tailed Student’s t test or one-way ANOVA was used for statistical analysis. For multiple testing corrections, the Tukey post hoc test was used when data were compared between all the groups. The Bonferroni post hoc test was used when data between selected groups were compared; coimmunoprecipitation analysis (comparing control, isoflurane, and PSD-95 mutant PDZ2 peptide, PSD-95 wild-type PDZ2 peptide, which included one family, four treatments, and two comparisons), total spine density, and relative cGMP-dependent protein kinase activity (comparing control, isoflurane, control + DETA NONOate, control + DETA NONOate + ODQ, isoflurane + DETA NONOate, isoflurane + DETA NONOate + ODQ, which included one family, six treatments, and four comparisons) and (PSD-95 mutant PDZ2 peptide, PSD-95 wild-type PDZ2 peptide, PSD-95 mutant PDZ2 peptide + DETA NONOate, PSD-95 mutant PDZ2 peptide + DETA NONOate + ODQ, PSD-95 wild-type PDZ2 peptide + DETA NONOate, PSD-95 wild-type PDZ2 peptide + DETA NONOate + ODQ, which included one family, six treatments, and four comparisons), (comparing control, isoflurane, control + 8-Br–cGMP, control + 8-Br–cGMP + KT5823, isoflurane + 8-Br–cGMP, isoflurane + 8-Br–cGMP + KT5823, which included one family, six treatments, and four comparisons) and (comparing PSD-95 mutant PDZ2 peptide, PSD-95 wild-type PDZ2 peptide, PSD-95 mutant PDZ2 peptide + 8-Br–cGMP, PSD-95 mutant PDZ2 peptide + 8-Br–cGMP + KT5823, PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP, PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP + KT5823, which included one family, six treatments, and four comparisons); for synapse puncta quantification (comparing control, isoflurane, control + 8-Br–cGMP, control + 8-Br–cGMP + KT5823, isoflurane + 8-Br–cGMP, isoflurane + 8-Br–cGMP + KT5823, control + YC-1, isoflurane + YC-1, which included one family, eight treatments, and five comparisons) and (comparing PSD-95 mutant PDZ2 peptide, PSD-95 wild-type PDZ2 peptide, PSD-95 mutant PDZ2 peptide + 8-Br–cGMP, PSD-95 mutant PDZ2 peptide + 8-Br–cGMP + KT5823, PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP, PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP + KT5823, PSD-95 mutant PDZ2 peptide + YC-1, PSD-95 wild-type PDZ2 peptide + YC-1, which included one family, eight treatments and five comparisons). Sample sizes were chosen based on preliminary data and/or published studies. A sample size of n = 3 was chosen for coimmunoprecipitation experiment because n = 3 was sufficient to show the disruption of interactions between NMDAR NR2 subunits and PSD-95 in response to cell-permeant fusion peptide.¹⁹  A sample size of n = 4 to 5 independent culture experiments in which a total of 25 to 40 neurons were analyzed per experimental conditions for spine analysis was chosen because n = 3 to 6 experiments in which 30 dendritic segments in total were sufficient to show significant differences in spines density in primary neuronal cultures.^29,32  A sample size of n = 4 independent culture experiments in which a total of 30 to 40 neurons were analyzed per experimental condition was chosen for synapse quantification because around 30 neurons were sufficient to show significant differences in synapse density.³³  A sample size of n = 3 was chosen for studying the cGMP-dependent protein kinase activity because n = 3 was sufficient to show the effects of icariside II against oxygen glucose deprivation–induced primary hippocampal neuron injury to enhance cGMP level and PKG activity.³⁴  All the n numbers (n = number of independent culture or experiments) are reported in the figure legends. In the novel object recognition test, male and female data were separated and analyzed in a one-way ANOVA followed by Bonferroni post hoc test (comparing oxygen control + vehicle, oxygen control + YC-1, isoflurane + vehicle, and isoflurane + YC-1, which included one family, four treatments, and three comparisons). Because all animals survived, there were no missing data in this study. Outlier value was detected by GraphPad Prism (robust regression and outlier removal Q = 1%), so it was removed. Using means and SD of recognition index from mixed sex control and isoflurane group, we determined the sample size needed for 90% power and α = 0.05 was n = 9 mice. Thus, power is greater than 90% with males (n = 15; Cohen’s d = 1.29) and females (n = 14; Cohen’s d = 1.45). Statistical significance was set at P < 0.05. All statistical analyses were carried out with GraphPad Prism version 8.4.3 (GraphPad Inc., USA).

To examine whether isoflurane exposure disrupts synaptic PDZ2 domain–mediated protein interactions in the developing brain hippocampus, we exposed postnatal day 7 mouse pups to isoflurane or control conditions (50% oxygen) for 4 h. A separate cohort was administered intraperitoneal injection of fusion peptides, active PSD-95 wild-type PDZ2 peptide (mimics isoflurane), or inactive PSD-95 mutant PDZ2 peptide (an inactive control). Under normal conditions, the second PDZ domain of PSD-95 interacts with the PDZ domain of nNOS or the C-terminal tails of the NR2A and NR2B subunits.¹⁶  Through the coimmunoprecipitation assay, we found that isoflurane and PSD-95 wild-type PDZ2 peptide markedly disrupted the interactions between NR2A/2B subunits and PSD-95 in the hippocampal lysates. However, in control and PSD-95 mutant PDZ2 peptide groups, this interaction was retained (fig. 1, A and B). Data were analyzed by using one-way ANOVA, which showed that the levels of PSD-95 coimmunoprecipitate relative to NR2A/2B precipitate (mean ± SD [in percentage of control]) were decreased in isoflurane (54.73 ± 16.52) compared with control, P = 0.001; and in PSD-95 wild-type PDZ2 peptide (51.32 ± 12.93) compared with PSD-95 mutant PDZ2 peptide, P = 0.001 (n = 3).

The purity of primary neuronal cells was assessed with MAP-2/4′,6-diamidino-2-phenylindole immunostaining, and in total, 85% of the cells were positive for MAP-2 proteins (fig. 2A). At 7 days, in vitro cells were exposed and treated for 4 h, and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay showed that neither 1.5% isoflurane nor 1 μM active PSD-95 wild-type PDZ2 peptide reduced viability in the hippocampal neurons when compared to that of respective control or untreated neurons (fig. 2, B and C). Data were analyzed using one-way ANOVA, which showed no significant differences in cell viability (in percentage) between untreated and experimental groups (n = 3 independent culture experiments). The results were as follows: untreated versus control, P = 0.909; untreated versus isoflurane, P = 0.783; control (94.53 ± 24.93) versus isoflurane (91.12 ± 12.18), P = 0.963; untreated versus PSD-95 mutant PDZ2 peptide, P = 0.994; untreated versus PSD-95 wild-type PDZ2 peptide, P = 0.960; and PSD-95 mutant PDZ2 peptide (99.1 ± 9.97) versus PSD-95 wild-type PDZ2 peptide (97.49 ± 16.76), P = 0.983. PSD-95 is a fundamental scaffolding protein during synaptogenesis and neurodevelopment, and its dysfunction during development may alter synaptic plastic events at the dendritic spines, causing neurologic deficits.³⁵  Therefore, we examined whether isoflurane or loss of synaptic PDS95-PDZ2 and NR2A/2B protein interaction affects early dendritic spines in developing hippocampal neuronal cells. After exposure to 4 h of 1.5% isoflurane or 1 μM active PSD-95 wild-type PDZ2 peptide, neurons exhibited significant spine loss in comparison with their respective controls (mean ± SD [spine density per 10 μm]) as follows: control (5.28 ± 0.56) versus isoflurane (2.23 ± 0.67), P<0.0001; PSD-95 mutant PDZ2 peptide (4.74 ± 0.94) versus PSD-95 wild-type PDZ2 peptide (1.47 ± 0.87), P<0.001; fig. 2, D to G). Additionally, we observed a significant increase in spine length in response to PSD-95 wild-type PDZ2 peptide exposure (PSD-95 mutant PDZ2 peptide [1.36 ± 0.39] versus PSD-95 wild-type PDZ2 peptide [2.28 ± 0.48], P = 0.011) and a trend toward increased spine length in response to isoflurane exposure (control [1.48 ± 0.25] versus isoflurane [1.99 ± 0.49], P = 0.076), although the latter change was not statistically significant (fig. 2, H and I). Data were analyzed by using independent two-tailed unpaired t-test.

NO signaling plays an important role in regulating the development of excitatory synapses, and impairment in NO production interferes with the development of spines and synapses.³⁶  We investigated the effects of the NO donor DETA NONOate on isoflurane- and PSD-95 wild-type PDZ2 peptide–induced loss of early dendritic spines in hippocampal neurons. First, we determine the noncytotoxic concentration of DETA NONOate and sGC inhibitor ODQ in neuronal cells (data not shown). Addition of 150 μM DETA NONOate at the time of isoflurane or PSD-95 wild-type PDZ2 peptide exposure attenuated the loss of dendritic spines in neuronal cells (mean ± SD [spine density per 10 μm]). Data were analyzed using one-way ANOVA, and total spine density was measured in response to DETA NONOate and ODQ in isoflurane- and PSD-95 wild-type PDZ2 peptide–treated neuronal cells as follows: isoflurane (1.82 ± 0.83) versus isoflurane + DETA NONOate (3.25 ± 0.31), P = 0.037; PSD-95 wild-type PDZ2 peptide (1.69 ± 0.73) versus PSD-95 wild-type PDZ2 peptide + DETA NONOate (3.14 ± 0.5), P = 0.030; (fig. 3, A to D). To determine whether activation of sGC by NO is specific and sufficient to prevent isoflurane- and PSD-95 wild-type PDZ2 peptide–induced loss in spine density, we treated neuronal cells with ODQ (100 μM) along with DETA NONOate. ODQ blocked the ability of DETA NONOate to attenuate isoflurane and PSD-95 wild-type PDZ2 peptide–induced dendritic spine loss as follows: isoflurane + DETA NONOate versus isoflurane + DETA NONOate + ODQ (1.21 ± 0.2), P = 0.002; isoflurane versus isoflurane + DETA NONOate + ODQ, P = 0.903; PSD-95 wild-type PDZ2 peptide + DETA NONOate versus PSD-95 wild-type PDZ2 peptide + DETA NONOate + ODQ (1.13 ± 0.17), P = 0.002; and PSD-95 wild-type PDZ2 peptide versus PSD-95 wild-type PDZ2 peptide + DETA NONOate + ODQ, P > 0.999; (fig. 3, A to D).

Multiple effects of NO are mediated through its canonical receptor, sGC, and the secondary messenger cGMP.³⁷  Therefore, we investigated the effects of the cGMP analog 8-Br–cGMP on isoflurane- and PSD-95 wild-type PDZ2 peptide–induced dendritic spine loss. First we determined the noncytotoxic concentration of 8-Br–cGMP and PKG inhibitor KT5823 in neuronal cells (data not shown). The addition of 200 μM 8-Br–cGMP at the time of isoflurane or PSD-95 wild-type PDZ2 peptide exposure attenuated the loss of dendritic spines in the hippocampal neuronal cells (mean ± SD [spine density per 10 μm]). Data were analyzed using one-way ANOVA, and spine density were measured in response to 8-Br–cGMP and KT5823 in isoflurane- and PSD-95 wild-type PDZ2 peptide–treated cells. Isoflurane (1.69 ± 0.49) versus isoflurane + 8-Br–cGMP (3.52 ± 0.8), P = 0.045, (fig. 4, A and C). 8-Br–cGMP also attenuated PSD-95 wild-type PDZ2 peptide–induced loss of spine density, as follows: PSD-95 wild-type PDZ2 peptide (1.52 ± 0.61) versus PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP (3 ± 0.42), P = 0.048, (fig. 4, B and D). Next, to determine whether activation of PKG by 8-Br–cGMP is specific and sufficient to prevent isoflurane and PSD-95 wild-type PDZ2 peptide–induced loss in spine density, we added 1 μM KT5823 along with 8-Br–cGMP. We found that KT5823 blocked the ability of 8-Br–cGMP to prevent isoflurane-induced spine loss, as follows: isoflurane + 8-Br–cGMP versus isoflurane + 8-Br–cGMP + KT5823 (1.66 ± 0.43), P = 0.041; isoflurane versus isoflurane + 8-Br–cGMP + KT5823, P > 0.999, (fig. 4, A and C); and PSD-95 wild-type PDZ2 peptide–induced spine loss as follows: PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP versus PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP + KT5823 (1.49 ± 0.56), P = 0.042; PSD-95 wild-type PDZ2 peptide versus PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP+KT5823, P > 0.999, (fig. 4, B and D), indicating that this pathway is PKG-dependent. cGMP-dependent protein kinase is involved in long-term potentiation and in synaptic plasticity,³⁸  and cGMP-dependent protein kinase type I is detectable in dissociated cultures of hippocampal neurons.³⁹  To determine changes in cGMP-dependent protein kinase activity in response to isoflurane or synaptic PDZ2 disruption, we exposed hippocampal neurons to isoflurane or PSD-95 wild-type PDZ2 peptide in the presence/absence of 8-Br–cGMP and KT5823. Both isoflurane and PSD-95 wild-type PDZ2 peptide caused a decrease in cGMP-dependent protein kinase activity. Data were analyzed by one-way ANOVA to compare relative cGMP-dependent protein kinase activity between control and experimental groups (mean ± SD). A decrease in the relative levels of cGMP-dependent protein kinase activity in isoflurane (0.44 ± 0.05) versus control, P < 0.001; and PSD-95 wild-type PDZ2 peptide (0.45 ± 0.06) versus PSD-95 mutant PDZ2 peptide, P = 0.002 (fig. 4, E and F). However, 8-Br–cGMP ameliorated the loss in cGMP-dependent protein kinase activity: isoflurane versus isoflurane + 8-Br–cGMP (0.77 ± 0.16), P = 0.032; and PSD-95 wild-type PDZ2 peptide versus PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP (0.81 ± 0.08), P = 0.037. Conversely, adding PKG inhibitor KT5823 further decreased the cGMP-dependent protein kinase levels in the hippocampal neuronal cells: isoflurane + 8-Br–cGMP versus isoflurane + 8-Br–cGMP + KT5823 (0.31 ± 0.15), P = 0.003; isoflurane versus isoflurane + 8-Br–cGMP + KT5823, P = 0.918; PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP versus PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP + KT5823 (0.44 ± 0.14), P = 0.033; and PSD-95 wild-type PDZ2 peptide versus PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP + KT5823, P > 0.999 (fig. 4, E and F).

Previous studies showed that isoflurane impairs synaptic plasticity in the mouse hippocampus.⁴⁰  NO-cGMP-PKG signaling is involved in the modulation of synaptic plasticity and memory formation in various brain regions.^41,42  Therefore, we first investigated effects of isoflurane or PSD-95 wild-type PDZ2 peptide on synapse in primary neurons. We exposed primary neurons with isoflurane and peptides for 4 h at 7 days in vitro, in the presence and absence of pharmacologic drugs of the pathway. The cells were fixed for immunocytochemistry at 14 days in vitro, and the concentration of all the drugs was maintained during media changes. Data were analyzed using one-way ANOVA, in which we found a significant reduction in the synaptic puncta (mean ± SD [in percentage]) in the primary neurons exposed to isoflurane (41.10 ± 14.38) in comparison to control, P = 0.001 (fig. 5, A and B); and PSD-95 wild-type PDZ2 peptide (50.49 ± 14.31) in comparison to PSD-95 mutant PDZ2 peptide, P < 0.001 (fig. 6, A and B). We next demonstrated the effects of the cGMP analog 8-Br–cGMP and YC-1 on isoflurane or PSD-95 wild-type PDZ2 peptide–induced synapse loss. Addition of 200 μM 8-Br–cGMP and 10 μM YC-1 at the time of isoflurane exposure attenuated the loss of synapse as follows: isoflurane versus isoflurane + 8-Br–cGMP (89.57 ± 26.98), P = 0.011; and isoflurane versus isoflurane + YC-1 (88.86 ± 29.87), P = 0.013 (fig. 5, A and B). 8-Br–cGMP and YC-1 also attenuated PSD-95 wild-type PDZ2 peptide–induced loss of synapse as follows: PSD-95 wild-type PDZ2 peptide versus PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP (80.27 ± 12.04), P = 0.045; and PSD-95 wild-type PDZ2 peptide versus PSD-95 wild-type PDZ2 peptide + YC-1 (79.96 ± 9.04), P = 0.048, (fig. 6, A and B). Next, to validate whether activation of PKG by 8-Br–cGMP is specific and sufficient to prevent isoflurane- and PSD-95 wild-type PDZ2 peptide–induced loss in synaptic puncta, we added KT5823 along with 8-Br–cGMP. We found that KT5823 blocked the ability of 8-Br–cGMP to prevent isoflurane and PSD-95 wild-type PDZ2 peptide–induced synapse loss as follows: isoflurane + 8-Br–cGMP versus isoflurane + 8-Br–cGMP + KT5823 (53.42 ± 18.89), P = 0.090; isoflurane versus isoflurane + 8-Br–cGMP + KT5823, P > 0.999 (fig. 5, A and B); PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP versus PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP + KT5823 (60.65 ± 18.65), P = 0.368; and PSD-95 wild-type PDZ2 peptide versus PSD-95 wild-type PDZ2 peptide + 8-Br–cGMP + KT5823, P > 0.999 (fig. 6, A and B), indicating that this pathway is PKG-dependent.

To determine whether neonatal isoflurane exposure or synaptic PSD-95 PDZ2 disruption contributes to cognitive impairment at later stages, we investigated the effect of isoflurane on recognition memory by assessing hippocampal-dependent object recognition at postnatal day 35 in both the male and female mice, 4 weeks after exposure at postnatal day 7. We observed that oxygen control + vehicle, oxygen control + YC-1, and isoflurane + YC-1 groups were able to discriminate between novel and known objects as calculated by significantly increased amounts of time investigating the novel object over the known object (fig. 7, A and C) mean ± SD [seconds]) for novel versus known: male, oxygen control + vehicle [21.36 ± 9.42 versus 12.07 ± 5.2], P = 0.001 (n = 15); oxygen control + YC-1 [16.12 ± 5.5 versus 10.12 ± 5.17], P = 0.001 (n = 15); isoflurane + YC-1 [17.62 ± 4.82 versus 13.16 ± 5.9], P = 0.014 (n = 15) female, oxygen control + vehicle [21.87 ± 6.5 versus 11.53 ± 6.62], P < 0.0001 (n = 14); oxygen control + YC-1 [19.04 ± 9.55 versus 11.42 ± 5.05], P = 0.007 (n = 15); isoflurane + YC-1 [18.66 ± 5.58 versus 13.05 ± 4.66], P = 0.009 (n = 15). In contrast, isoflurane + vehicle group depicts no significant increase in investigation time between the novel and known objects: male, isoflurane + vehicle [18.42 ± 11.69 versus 18.66 ± 10.1], P = 0.940 (n = 15); and female; isoflurane + vehicle [16.38 ± 4.78 versus 14.1 ± 4.48], P = 0.091 (n = 14), with n = number of mice. Treatment with YC-1 prevents the impairment in novel object recognition caused by isoflurane exposures as shown by the increased discrimination in novel object recognition. The discrimination or recognition index was the time investigating a novel object over the time investigating a novel object plus a familiar object × 100)⁴³  (fig. 7, B and D). Data were analyzed using one-way ANOVA, which indicated that the isoflurane + vehicle group performed less well than the other groups in both the male and female mice (fig. 7, B and D; mean ± SD [recognition index]): males, oxygen control + vehicle [64.08 ± 10.57] versus isoflurane + vehicle [48.49 ± 13.41], P = 0.001; oxygen control + vehicle versus oxygen control + YC-1 [62.13 ± 10.88], P > 0.999; oxygen control + vehicle versus isoflurane + YC-1 [58.24 ± 9.99], P = 0.486; females: oxygen control + vehicle [67.13 ± 11.17] versus isoflurane + vehicle [53.76 ± 6.64], P = 0.003; oxygen control + vehicle versus oxygen control + YC-1 [62.4 ± 11.97], P = 0.667; oxygen control + vehicle versus isoflurane + YC-1 [58.78 ± 10.46], P = 0.100.

In the current study, we examined the effects of isoflurane on synaptic PDZ2 interaction in the mouse hippocampus and on dendritic spines and synapses in the primary hippocampal neurons and cognitive functions in mice. We previously showed that inhalational anesthetics at clinically relevant concentrations disrupt PDZ domain–mediated protein–protein interactions between PSD-95 or PSD-93 and the NMDAR NR2 subunits or nNOS.^15,16,44  We speculated that disruption of synaptic PDZ2 interactions during the critical period of synaptogenesis could impair neuronal plasticity and lead to long-term cognitive dysfunction or neurologic impairments. Mostly, neurologic disorders converge into a common phenotype of impaired synaptic connectivity. We found that 4 h exposure to isoflurane or a PSD-95 wild-type PDZ2 peptide disrupted the interaction between NR2A/2B subunits and PSD-95 PDZ2 domain in the hippocampus of neonatal mouse brain. Additionally, in cultured hippocampal neurons, isoflurane and PSD-95 wild-type PDZ2 peptide caused loss of early dendritic spines, loss of mature synapses (the overlap of pre- and postsynaptic markers can be used as a representation of the mature synapse⁴⁵ ), and decrease in cGMP-dependent protein kinase activity. Introduction of NO donor or cGMP analog at the time of exposure prevented the loss in early dendritic spines, mature synapses, and the decrease in cGMP-dependent protein kinase activity in response to isoflurane or PSD-95 wild-type PDZ2 peptide. Furthermore, introducing of sGC activator YC-1 in mice during the neonatal isoflurane exposure prevented the loss in object recognition memory at 5 weeks. These data suggest that dysregulation of NMDAR NR2–PSD-95–PDZ2–nNOS ternary complexes upstream after early anesthetic exposure impairs downstream components NO/sGC/cGMP–mediated PKG signaling pathway (fig. 8).

The hippocampus is the main region for learning and memory and has been implicated in the amnestic and delayed cognitive effects of anesthetics.⁴⁸  Therefore, we endeavored to understand the downstream signaling components that are perturbed in response to this disruption of synaptic PDZ2 interactions in the hippocampus. To that end, we employed primary neuronal cultures of cells dissociated from the brain hippocampus for studying the complex downstream signaling mechanisms. Previous studies from others have shown that 4 h of exposure to 1.4% isoflurane decreases dendritic filopodial spines in neonatal primary neuronal cultures at 5 days in vitro.⁴⁹  We explored the effects of isoflurane and PSD-95 wild-type PDZ2 peptide on dendritic spines at 7 days in vitro and synapse at 14 days in vitro since the functional polarization begins approximately a week after neuronal cells are plated. In addition, 14 days in vitro is a time point at which synaptic connections are well established.

We found that 4 h of isoflurane or PSD-95 wild-type PDZ2 peptide (mimics isoflurane) exposure at 7 days in vitro caused loss of early dendritic spines. These in vitro neuronal culture findings coincide with our laboratory’s recently published studies in intact animals, which showed that isoflurane and PSD-95 wild-type PDZ2 peptide caused loss of spine density.^17,18  Excitatory synapses are found on dendritic spines; thus, the loss of early dendritic spines could reduce excitatory neurotransmission and also lead to loss of synapses.⁵⁰  Furthermore, we also observed that 4 h of isoflurane or PSD-95 wild-type PDZ2 peptide at 7 days in vitro caused a decrease in the synapses at 14 days in vitro in the primary neurons. Many of the PDZ domain–mediated protein–protein interactions linking the ion channels and receptors to their downstream signaling pathways have been shown to be impeded by inhalational anesthetics. The disruption of synaptic PDZ2 domains by isoflurane provides a plausible mechanism for the dissociation of the PSD-95–NMDAR–nNOS ternary complex in the neonatal brain hippocampus and cognitive dysfunctions in mice, loss of early dendritic spines, mature synapse, and cGMP-dependent protein kinase activity in the hippocampal neurons.

We next investigated whether neonatal anesthetic exposure or disruption of synaptic PSD-95 PDZ2 interactions in the hippocampus affects the memory and cognitive functions in mice once they reached a certain age. Several groups including ours have measured cognitive performance in mice using a novel object recognition behavior test that has previously indicated impairment after early exposure to anesthesia and involves hippocampal-dependent learning and memory including nonspatial recognition memory in both male and female mice.^17,18  Previously published studies from our laboratory demonstrated that 3- and 6-week-old mice exposed early to isoflurane or injected with PSD-95 wild-type PDZ2 peptide exhibited reduced recognition memory performance.^17,18  However, to validate whether this cognitive memory impairment is due to the impaired downstream sGC/cGMP-dependent signaling. We have administered sGC activator–YC-1 in mice at the time of isoflurane exposure. We found that YC-1 treatment mitigated the impairment of recognition memory in response to isoflurane, which is also similar to the findings from another group that has reported neuroprotective efficacy of YC-1 on cognitive dysfunctions in response to different anesthetics.²⁶  This also implicates the involvement of the sGC/cGMP–dependent signaling in early anesthetic exposure–produced cognitive impairments.

Brain development is orchestrated through the plasticity- and activity-dependent mechanisms that control spines and synapse formations and eliminations.³⁶  Lines of evidence suggest that activity-dependent spines and synapse formation are interceded by a postsynaptic signaling cascade implicating NO, cGMP, and vasodilator-stimulated phosphoprotein phosphorylation.^36,51  Our recent findings demonstrate that using the NO donor prevents neonatal isoflurane-induced impairments in synaptic plasticity and memory.¹⁷  However, the pathways downstream of NO that are affected by the disruption of the synaptic PSD-95–PDZ2 domain interactions and that are responsible for the loss in early dendritic spines have not been uncovered. Therefore, to identify the downstream components of the NO-mediated signaling, we used several pharmacologic activators and inhibitors. sGC is a known receptor for NO, and its activation causes a subsequent rise in cGMP levels.⁵²  Here we found the components of the NO signaling pathway that sequentially activate PKG are potentially involved in the mechanisms underlying loss of early dendritic spine density and synapses in hippocampal neuronal cells in response to the isoflurane or wild-type PDZ2 peptide exposure. Pretreating the neuronal cells with NO donor DETA NONOate at the time of isoflurane and PSD-95 wild-type PDZ2 peptide exposure prevented the loss in dendritic spine density. However, this effect was reversed by using a selective sGC inhibitor (ODQ). When NO donor was added in the presence of sGC inhibitor, it did not prevent isoflurane-induced spine loss. These results implicate a role for sGC in mediating the action of NO in response to isoflurane or synaptic PSD-95 PDZ2 disruption. Hence, activation of sGC by NO is specific and sufficient to prevent isoflurane-induced dendritic spine loss in the hippocampal neuronal cells.

The NO-sGC-cGMP pathway plays an important role in modulating synaptic transmission, plasticity, and cognitive functions in the hippocampus and cerebral cortex.⁵³  We also observed that the effect of DETA NONOate was mimicked by the cGMP analog 8-Br–cGMP. Pretreatment of 8-Br–cGMP prevented the isoflurane- and PSD-95 wild-type PDZ2 peptide–induced loss in early spine density and synapses. To validate whether activation of PKG by cGMP is specific and sufficient to prevent isoflurane or synaptic PSD-95 PDZ2 disruption–induced spine loss, we also added 8-Br–cGMP in the presence of PKG inhibitor (KT5823) to the hippocampal neurons. However, in the presence of KT5823, 8-Br–cGMP did not prevent isoflurane-induced loss of spines. These results indicate that cGMP acts through PKG to prevent isoflurane and PSD-95 wild-type PDZ2 peptide–induced spine loss in the hippocampal neuronal cells. Furthermore, adding NO-independent sGC activator YC-1 also prevents the loss in synapse in response to isoflurane or PSD-95 wild-type PDZ2 peptide in the hippocampal neuronal cells.

Major limitations of the current study are that we have not studied the sex-specific effects in the coimmunoprecipitation and in vitro primary neuron culture experiments in response to the isoflurane anesthesia or PSD-95 wild-type PDZ2 peptide, given the known sex-specific sensitivity to anesthetics during distinct stages of development. Nevertheless, in response to peer review, we have performed additional experiments and have reported the sex-specific effects of neonatal isoflurane exposure in male and female mice at 5 weeks on object recognition memory (fig. 7).

Our findings show that clinically relevant concentration of isoflurane markedly disrupted the interactions between NMDAR NR2A/2B subunits and PSD-95 in the neonatal mouse brain hippocampus. We also found that neonatal isoflurane or synaptic PSD-95 PDZ2 disruption cause impairment in the recognition memory in both male and female mice at 5 weeks and loss of early dendritic spines and synapses in vitro without inducing significant neuronal death. Our results strongly suggest that introducing an NO donor or cGMP analog in vitro prevents the loss in early dendritic spines and synapses. In addition, introducing sGC activator YC-1 at the time of isoflurane exposure has prevented the loss in object recognition memory in both the sexes. Molecular dissection of these intricate mechanisms has added to our understanding of how isoflurane or disruption of the PDZ2 domain of the PSD-95 protein in the neonatal hippocampus upstream perturbs these downstream signaling pathways. Additionally, our work expands several avenues for therapeutic interventions in which targeting these molecular events might treat neurologic dysfunctions including inhalational anesthetic-related cognitive impairment.

The authors thank Claire Levine, M.S., E.L.S. (Johns Hopkins University, Baltimore, Maryland), for her excellent scientific editing.

This work was supported by National Institutes of Health, National Institute of General Medical Sciences (Bethesda, Maryland) grant No. R01GM110674 to Dr. Johns.

The authors declare no competing interests.