Disruption of Hippocampal Neuregulin 1–ErbB4 Signaling Contributes to the Hippocampus-dependent Cognitive Impairment Induced by Isoflurane in Aged Mice

POSTOPERATIVE cognitive decline or dysfunction occurs frequently in the elderly after surgery and anesthesia, with the impairments in recent memory, attention, language comprehension, and social integration.^1–3  It may prolong the patients’ hospital stay, diminish the quality of life, and affect the recovery process.^1–4  In clinical practice, patients usually undergo surgery with general anesthesia; however, previous documents have suggested that patients receiving general anesthesia are associated with increased risk of developing postoperative cognitive dysfunction compared with the risk associated with patients receiving regional anesthesia.^5,6  Recently, accumulating evidence has shown that surgery induces cognitive impairments in rodents.^7–10  Anesthetics have also been reported to elicit cognitive impairment in rodents.^11–14  However, the effects of anesthetic isoflurane on cognitive function in aged mice and the underlying mechanisms remain largely to be determined. We therefore set out to assess the effects of isoflurane on learning and memory function in aged mice and to study the potential mechanism.

Neuregulin 1 (NRG1), a member of the epidermal growth factor family, acts by promoting the tyrosine kinase of ErbB receptor autophosphorylation.¹⁵  ErbB4, one of the ErbB receptors, has a high affinity to NRG1.¹⁶  Recent data suggest a key role of the NRG1–ErbB4 signaling in neurotransmission, synaptic plasticity, and synchronization of neuronal network activity in the cortex and hippocampus through raising precisely timed γ-aminobutyric acid release, which is of great importance for cognition, learning, and memory.^15–17 

Notably, the distribution of ErbB4 is largely restricted to specific classes of interneurons, particularly in the parvalbumin-positive interneurons.¹⁸  It has been reported that cognitive processes are regulated by the parvalbumin-positive interneurons, in which ErbB4 is required in the procedure of NRG1 modulating the pyramidal neuronal activity and the long-term potentiation in the adult mouse brain.¹⁹  In the ErbB4^−/− mice, parvalbumin-positive and glutamic acid decarboxylase (GAD) 67–positive interneurons are significantly decreased.²⁰  In the NRG1–ErbB4 signaling–mutant mice, the hippocampus-dependent memory was impaired notably.^19,21  Therefore, we hypothesized that the cognitive impairments induced by the isoflurane exposure in aged mice may be attributed to the disruption of hippocampal NRG1–ErbB4 signaling probably in the parvalbumin-positive interneurons by down-regulating the levels of parvalbumin and GAD67.

The animal care and the experiment were approved by the Ethics Committee of Jinling Hospital, Nanjing, China, and were performed according to the Guide for the Care and Use of Laboratory Animals approved by the National Institutes of Health of the United States.

Ninety male C57BL/6 mice with the age of 14-month old were purchased from the Animal Center of Jinling Hospital, Nanjing, China. The mice were housed five per cage in a 12-h light–dark cycle in a room of 22° ± 1°C with food and water available ad libitum. After 7 days of accommodation, the mice were randomly divided into five groups (n = 18): control group, dimethyl sulfoxide (vehicle) group, isoflurane group, isoflurane plus NRG1-β1 (a recombinant NRG1-1β; R&D Systems, Minneapolis, MN) group, and isoflurane plus NRG1-β1 plus AG1478 group (an ErbB4 inhibitor; Tocris Bioscience, Bristol, United Kingdom).

The isoflurane exposure was performed by placing mice with spontaneous breathing in an anesthesia chamber prefilled with 1.5% isoflurane carried by 100% O₂. The anesthesia was induced and maintained with 1.5% isoflurane carried by 100% O₂ of 1.5 l/min for 2 h. Temperatures of the mice were maintained at 37° ± 0.5°C during anesthesia by heating lamps and heating pads. In our preliminary experiment, we collected more than 0.5 ml of arterial blood from both the isoflurane-exposed mice and the 100% O₂–exposed mice immediately after the gas exposure with the same protocol of the current study. The blood gas analysis has shown no hypoxia or acidosis immediately after the isoflurane exposure or the 100% O₂ exposure in the mice. After the gas exposure, the mice were placed back into the home cage for emergence with 100% O₂ of 1.5 l/min supply. The mice in the control group and the vehicle group were not anesthetized but were given 100% O₂ of 1.5 l/min for 2 h in the identical chamber.

We implanted a stainless steel cannula into the left lateral ventricle (0.45 mm posterior, 1.08 mm lateral to bregma, and 2.50 mm below dura) during the anesthesia with intraperitoneal injection of 60 mg/kg sodium pentobarbital except for the mice in the control group. Then, the mice were given a period of 1 week for postoperative recovery.

NRG1-β1 (10 μM) and/or AG1478 (5 mM) were intracerebroventricularly injected 1 h before the behavior tests for 2 days. The doses of NRG1-β1 and AG1478 were selected according to the previous studies.^12,18  The time points for the drug injection were selected on the basis of the results of our preliminary experiment, showing that the hippocampal p-ErbB4 levels peaked 1 h after the intracerebroventricular injection of NRG1-β1 in mice (fig. 1). The total injection volume was 2 μl. The mice in the vehicle group and the isoflurane group received an equivalent volume of vehicle.

The locomotor and exploratory activities of the mice (n = 10 for each group) were evaluated by the open-field test 24 h after the gas exposure. Each mouse was released in the center of the white plastic chamber (30 cm × 30 cm× 30 cm, XR-XZ301; Shanghai Softmaze Information Technology Co., Ltd., Shanghai, China) and left to explore it for 5 min while the activities were automatically recorded by a video-tracking system. At the end of testing, the arena was cleaned with 75% alcohol to avoid the presence of olfactory cues.

The fear conditioning test is a simple and sensitive test of hippocampus-dependent and hippocampus-independent memory function. After the open-field test, the mouse was placed in the conditioning chamber (XR-XC404; Shanghai Softmaze Information Technology Co., Ltd.) for 3 min as an accommodation period and then one tone-foot-shock pairing (tone, 30 s, 65 dB, 1 kHz; foot-shock, 2 s, 0.75 mA) was delivered. The mouse was allowed to explore the chamber for another 30 s after the shock to study postshock freezing. The contextual fear conditioning test was assessed 24 h after the training by placing the mice back in the same test chamber for 5 min. The tone fear conditioning test was assessed 2 h after the contextual fear conditioning test in a novel chamber changed in the shape, color, and smell and the training tone was delivered for 3 min. Freezing behavior, defined as the absence of all visible movement of the body except the movement necessitated by respiration, was scored by an observer software. At the end of testing, the chamber was cleaned with 75% alcohol to avoid the presence of olfactory cues.

The brain tissues of eight mice without any behavior tests from each group were harvested 48 h after the gas exposure to determine the levels of NRG1, ErbB4, p-ErbB4, parvalbumin, and GAD67 in the hippocampus using Western blotting, enzyme-linked immunosorbent assay, and immunofluorescence.

The levels of ErbB4, p-ErbB4, parvalbumin, and GAD67 in the hippocampus of mice (n = 4 for each group) were assessed by the Western blotting 48 h after the gas exposure. The normalized protein samples were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis and then were transferred onto polyvinylidene difluoride membranes. Membranes were blocked with 5% skim milk in Tris-buffered saline tween for 1 h and then incubated with rabbit anti-ErbB4 (1:500; Santa Cruz Biotechnology, Dallas, TX), rabbit anti-p-ErbB4 (1:500; Santa Cruz Biotechnology), rabbit anti-β-actin (1:1,000; Cell Signaling Technology, Boston, MA), rabbit anti-parvalbumin (1:1,000; Abcam, Cambridge, United Kingdom), and mouse anti-GAD67 (1:500; Santa Cruz Biotechnology) overnight in a room at 4°C temperature. After thorough washing, membranes were incubated in Tris-buffered saline tween with the secondary antibody (goat anti-rabbit and goat anti-mouse; Santa Cruz Biotechnology) diluted 1:1,000 for 1 h at room temperature. Bands were visualized by the enhanced chemiluminescence and quantitated with the Image Quant Software (Syngene, Cambridge, United Kingdom).

The levels of NRG1 in the hippocampus of mice (n = 4 for each group) were determined by the enzyme-linked immunosorbent assay kits following the protocol provided by the manufacturer (BGI, Shenzhen, China). The readings were normalized to the amount of standard protein.

The immunofluorescence of ErbB4, parvalbumin, and GAD67 in brain sections of mice (n = 4 for each group) was assessed 48 h after the gas exposure. The brains were cut coronally. Five sections (10 μm thickness) per mouse were chosen by randomization for the immunohistochemistry studies. There was a 30-μm interval between the two sections. The tissue sections were incubated in 1% bovine serum albumin for 1 h and followed by primary antibodies: rabbit anti-ErbB4 (1:50; Santa Cruz Biotechnology), mouse anti-parvalbumin (1:500; Abcam), and mouse anti-GAD67 (1:100; Santa Cruz Biotechnology) in 1% bovine serum albumin at 4°C overnight. After washing with phosphate buffered saline, the sections were incubated in goat anti-mouse IgG-fluorescein isothiocyanate, anti-mouse IgG-Cy3, and goat anti-rabbit IgG-Cy3 (1:200; Bioworld Technology, St. Louis Park, MN) for 1 h at room temperature. Fluorescent images were captured by a confocal microscope. Coexpression of ErbB4 with GAD67 or parvalbumin was determined by counting only the neurons that were independently, clearly, and identifiable in both the red and green color channels with 4’, 6-diamidino-2-phenylindole staining, rather than solely on appearance of yellow pixels in the merged images. The z-stack high-magnification colocalization images for the ErbB4 and GAD67 or parvalbumin were obtained. The slices were imaged every 1 μm across the entire slice and then collapsed together to form one image of the resulting z-stack picture. For immunohistochemistry analysis, the mean value of eight random regions (×40) across hippocampus in each section was calculated. The numerical density (cells per square millimeter) was calculated using the confocal images with the Image J, version 1.26t (National Institutes of Health, Bethesda, MD).

Data are presented as mean ± SD. There were four samples in each group. The sample size was guided by the experience from the pilot studies. Normality of data was analyzed by using Shapiro–Wilk test, and the data were found in Gaussian distribution. Therefore, the comparisons among the groups were performed by using one-way ANOVA followed by the Tukey test (two-tailed). The data were analyzed by using the SPSS software (version 16.0; SPSS Inc., Chicago, IL). P value less than 0.05 was considered statistically significant.

Both intracerebroventricular cannulation and drug interventions did not significantly affect the total ambulatory distance traveled and the time spent in the center during the open-field test (fig. 2, A and B). The isoflurane exposure reduced the freezing time to context from 50.28 ± 11.53% to 30.82 ± 10.00% compared with that in the control group (n = 10; P = 0.002; fig. 2C). NRG1-β1 increased the freezing time to context to 46.45 ± 9.88% in the isoflurane plus NRG1-β1 group compared with that in the isoflurane group (n = 10; P = 0.017; fig. 2C), which was abolished by AG1478 in the isoflurane plus NRG1-β1 plus AG1478 group (31.97 ± 10.94%; n = 10; P = 0.032; fig. 2C). No significant difference was detected in respect of the freezing time to tone among the five groups (fig. 2D).

The hippocampal concentration of NRG1 decreased to 112.92 ± 16.65 ng/g in the isoflurane group, but not in the isoflurane plus NRG1-β1 and isoflurane plus NRG1-β1 plus AG1478 groups, compared with that in the control group (172.79 ± 20.85 ng/g; n = 4; P = 0.007; fig. 3A). The hippocampal p-ErbB4/ErbB4 decreased to 42.26 ± 9.71% in the isoflurane group compared with that in the control group (69.15 ± 12.20%; n = 4; P = 0.024; fig. 3, B and C). The NRG1-β1 injection reversed the decrease in the hippocampal p-ErbB4/ErbB4 induced by the isoflurane exposure in the isoflurane plus NRG1-β1 group (up to 73.69 ± 10.77%, n = 4; P = 0.007; fig. 3, B and C), which was abolished by the AG1478 injection in the isoflurane plus NRG1-β1 plus AG1478 group (reduction to 49.16 ± 10.86%, n = 4; P = 0.043; fig. 3, B and C).

In the current study, the detected levels of NRG1, p-ErbB4, and ErbB4 were from the whole hippocampus, rather than from the GAD67- or parvalbumin-positive interneurons in the hippocampus, we therefore were unable to confirm the relation between the NRG1–ErbB4 signaling and the parvalbumin-positive interneurons. Then, we used the double-immunofluorescence to detect the colocalization of ErbB4 and γ-aminobutyric acidergic interneuron marker GAD67, and ErbB4 and parvalbumin in the hippocampus. The results indicated that ErbB4 was primarily expressed in the GAD67-immunoreactive interneurons (fig. 4A). The average coexpression of parvalbumin and ErbB4 in parvalbumin-positive interneurons was 82.73 ± 11.37% and in ErbB4-positive neurons was 63.73 ± 10.05% across the hippocampus (fig. 4, B and C; n = 4).

To investigate the possible mechanism underlying the isoflurane exposure–induced cognitive impairment mediated by NRG1–ErbB4 signaling, we detected the hippocampal levels of parvalbumin and GAD67 using the Western blotting. As shown in figure 5, the levels of parvalbumin and GAD67 decreased to 75.89 ± 10.26% and 73.87 ± 16.89%, respectively, in the isoflurane group compared with that in the control group (parvalbumin: 101.68 ± 11.21%, n = 4, P = 0.011; GAD67: 104.71 ± 6.85%, n = 4, P = 0.014). The NRG1-β1 injection reversed these reductions in the isoflurane plus NRG1-β1 group (all P < 0.05), which was abolished by the AG1478 injection in the isoflurane plus NRG1-β1 plus AG1478 group (all P < 0.05). These data were consistent with the results of the immunofluorescence (all P < 0.05; fig. 6).

The current study demonstrated that the isoflurane-exposed mice presented poorer performance in contextual fear conditioning test, which was consistent with the recent findings showing isoflurane anesthesia–induced cognitive impairments in rodents.^12,22  More importantly, we found for the first time that NRG1-β1 increased the freezing time to context 48-h postisoflurane exposure, which was abolished by an ErbB4 receptor antagonist AG1478. These results indicate that dysfunction of hippocampal NRG1–ErbB4 signaling plays a key role in the isoflurane exposure–induced hippocampus-dependent cognitive impairment in aged mice.

Isoflurane, a commonly used inhalation anesthetic in clinical practice, has been shown to induce cognitive impairments after the long-time exposure in the rodent models, although it is also reported to improve cognitive impairments in the ischemia animal models.^23,24  Many neuropathology mechanisms such as the disruption in the neurotransmission, the alteration of intracellular calcium homeostasis, and the damage of synaptic plasticity have been suggested to play key roles in the isoflurane-induced cognitive impairments.^25–27  However, the underlying mechanisms of these cognitive impairments remain largely to be determined.

The hippocampus is important to learning and memory, in which anesthetic-induced acute memory impairment parallels the inhibition of synaptic plasticity.^28,29  In the current study, isoflurane exposure and the drug interventions did not affect the locomotor activity and the exploratory behavior. Moreover, the isoflurane-exposed mice had decreased freezing time to context, but no difference in the freezing time to tone compared with that in the control mice, suggesting that the isoflurane exposure impairs the early hippocampus-dependent learning and memory in aged mice. These behavior results were still somewhat different from the results of the studies by Zhang et al.¹³  and Lin et al.,³⁰  which may be attributed to the differences in the animal species, the methods of the anesthetic exposure, the time points of the performance of the memory tests, and especially the animal age which has been increasingly reported to contribute to postoperative brain dysfunction in the previous works.^1,13 

NRG1–ErbB4 signaling is critical in various neurophysiology processes including neuron development and migration, synapse plasticity, and synthesis and secretion of neurotransmitter.¹⁵  In the hippocampus, NRG1–ErbB4 signaling has been implicated in the modulation of cognitive processes through regulating the activity of neural networks and synchronizing the pyramidal cells,³¹  whereas the disturbance of NRG1–ErbB4 signaling is associated with specific impairments in the hippocampus-dependent memory.³²  In the current study, the isoflurane exposure elicited the hippocampus-dependent cognitive impairment and reduced the hippocampal NRG1 level significantly in aged mice, whereas NRG1-β1 administration reversed this cognitive deficit. Furthermore, an ErbB4 receptor antagonist AG1478 was shown to abolish the beneficial effects of NRG1-β1, implying the involvement of ErbB4 in this procedure. These results suggest that NRG1–ErbB4 signaling contributes to the hippocampus-dependent cognitive impairment induced by isoflurane exposure in aged mice.

Parvalbumin-positive interneurons, the major subpopulation of γ-aminobutyric acidergic interneurons, express the calcium-binding parvalbumin and are characterized by the fast-spiking discharge pattern.³³  They are the main recipient of recurrent glutamatergic innervations and phase the output of pyramidal innervations in the hippocampal circuitry.^34,35  Accumulating evidence suggests that parvalbumin-positive interneurons are the major cellular target of NRG1–ErbB4 signaling in regulating the neuron network activity,^36,37  which can be confirmed by the findings that parvalbumin-Cre; ErbB4^−/−mice have contextual fear conditioning impairment, in which ErbB4 is knocked out specifically in parvalbumin-positive interneurons, whereas the CaMKII-Cre; ErbB4^−/− mice display normal freezing response in which ErbB4 is knocked out in excitatory neurons.^19,38  To determine whether the disruption of NRG1–ErbB4 signaling in the parvalbumin interneurons contributes to the cognitive impairments induced by the isoflurane exposure, we investigated the distribution of ErbB4 in the hippocampus. The results indicated the colocalization of ErbB4 and γ-aminobutyric acidergic interneuron marker GAD67 and parvalbumin, two markers of parvalbumin interneurons, supporting the view that ErbB4 is largely expressed in the parvalbumin-positive interneurons in the hippocampus.^18,39  In this study, both Western blotting and immunofluorescence results suggest that NRG1-β1 improves the cognitive impairment induced by the isoflurane exposure probably by the up-regulation of parvalbumin and GAD67 in the hippocampus.

The studies by Terrando et al.⁷  and Cibelli et al.⁸  have suggested that anesthesia may not strongly influence memory dysfunction in rodents, rather the surgery-induced neuroinflammation leads to the cognitive impairment. However, other studies have shown that anesthetic isoflurane and sevoflurane induce neuroinflammation, leading to cognitive impairment.^13,14  The data from the current studies also suggest that anesthesia with 1.5% isoflurane for 2 h induces cognitive impairment in aged mice. Note that the anesthetics in the studies by Terrando et al. were 2.1% isoflurane and buprenorphine (0.1 mg/kg).⁷  The anesthetics in the other studies were 1.4 or 1.5% isoflurane for 2 h once or 3.0% sevoflurane for 2 h daily for 3 days.^14,40,41  It is conceivable that different anesthesia regimens could have different effects on learning and memory function in rodents. The future studies may include determining whether the surgery and anesthesia can potentiate each other, leading to a greater cognitive impairment. These studies would lead to a better understanding of the underlying mechanism of postoperative cognitive dysfunction observed in patients.

There are limitations in the current study. First, although most of the ErbB4 are expressed in the parvalbumin-positive interneurons in the current study, we still cannot verify whether isoflurane exposure affects the NRG1–ErbB4 signaling especially in the parvalbumin-positive interneurons. Second, for the technical reasons, we cannot identify the disturbance in NRG1–ErbB4 signaling in the parvalbumin interneurons as a mechanism underlying isoflurane exposure–induced cognitive decline. Therefore, further studies using parvalbumin-Cre; ErbB4^−/−mice and CaMKII-Cre; ErbB4^−/− mice or electrophysiological techniques are needed to demonstrate whether isoflurane exposure affects the NRG1–ErbB4 signaling activity selectively in the parvalbumin-positive interneurons. In addition, the data in the current experiment were obtained after exposure to isoflurane carried by 100% O₂ in the aged mice. The future studies to investigate the effects of isoflurane or other inhalational anesthetics carried by different concentrations of oxygen on the NRG1–ErbB4 signaling in parvalbumin-positive interneurons are warranted.

In conclusion, exposure to isoflurane carried by 100% O₂ impairs the early hippocampus-dependent cognitive function and decreases the hippocampal levels of NRG1, p-ErbB4, parvalbumin, and GAD67 in aged mice, which is reversed by the NRG1-β1. All the effects of NRG1-β1 are blocked by the ErbB4 inhibitor AG1478. We provided additional evidence that the disruption of NRG1–ErbB4 signaling in parvalbumin-positive interneurons contributes, at least partially, to the isoflurane-induced hippocampus-dependent cognitive impairment after exposure to isoflurane carried by 100% O₂ in aged mice, pending further studies.

This work was supported by the grants from the National Natural Science Foundation of China, Beijing, China (Numbers 81271216 and 81222013) and the Natural Science Foundation of Jiangsu Province, Nanjing, China (Number BK2012778) and R01 GM088801 and R01 AG041274 from National Institutes of Health, Bethesda, Maryland, and Cure Alzheimer’s Fund, Wellesley, Massachusetts.

The authors declare no competing interests.