General anesthetic exposure in adolescent rats causes persistent maladaptations in cognitive and affective behaviors and neuroplasticity

Abstract

Accumulating evidence indicates that exposure to general anesthetics during infancy and childhood can cause persistent cognitive impairment, alterations in synaptic plasticity, and, to a lesser extent, increased incidence of behavioral disorders. Unfortunately, the developmental parameters of susceptibility to general anesthetics are not well understood. Adolescence is a critical developmental period wherein multiple late developing brain regions may also be vulnerable to enduring general anesthetic effects. Given the breadth of the adolescent age span, this group potentially represents millions more individuals than those exposed during early childhood. In this study, isoflurane exposure within a well-characterized adolescent period in Sprague-Dawley rats elicited immediate and persistent anxiety- and impulsive-like responding, as well as delayed cognitive impairment into adulthood. These behavioral abnormalities were paralleled by atypical dendritic spine morphology in the prefrontal cortex (PFC) and hippocampus (HPC), suggesting delayed anatomical maturation, and shifts in inhibitory function that suggest hypermaturation of extrasynaptic GABA_A receptor inhibition. Preventing this hypermaturation of extrasynaptic GABA_A receptor-mediated function in the PFC selectively reversed enhanced impulsivity resulting from adolescent isoflurane exposure. Taken together, these data demonstrate that the developmental window for susceptibility to enduring untoward effects of general anesthetics may be much longer than previously appreciated, and those effects may include affective behaviors in addition to cognition.

Introduction

General anesthetics (GAs) are clinically necessary to induce and maintain amnesia, unconsciousness and immobility during surgery. Numerous reports support a general absence of consequential effects resulting from GA exposure in adults. However, a growing body of literature suggests that the bookend periods of the lifespan may be times of increased vulnerability to negative postoperative GA effects. Postoperative cognitive dysfunction (POCD) is commonly associated with complications in elderly patients and is recognized as a major public health issue (Berger et al., 2015). Mounting clinical evidence also suggests that GA exposure during early developmental periods negatively impacts cognitive performance and increases the likelihood of behavioral disorders (Kotiniemi et al., 1997; Kalkman et al., 2009; Wilder et al., 2009; DiMaggio et al., 2011; Block et al., 2012; Stratmann et al., 2014). The factors that drive age-susceptibility to postoperative anesthesia sequelae remain unclear. While it is difficult to parse out direct GA effects from the underlying conditions that require surgery, as well as from accompanying surgical effects such as inflammation, animal models of early development allow for a more direct investigation of behavioral and neural adaptations induced through anesthesia alone.

Preclinical evidence has recapitulated early developmental postoperative effects in humans, including anxiety and persistent cognitive impairment (Jevtovic-Todorovic et al., 2003; Stratmann et al., 2010; Zhu et al., 2010; Paule et al., 2011; Seubert et al., 2013; Raper et al., 2015). Abnormal behaviors are associated with GA-induced neurotoxicity, reduced neural differentiation, impaired axonal pruning, and alterations in synaptogenesis and neurotransmission (Head et al., 2009; Tan et al., 2009; Briner et al., 2010; Lunardi et al., 2010; Mintz et al., 2013; Amrock et al., 2015; Ju et al., 2016 Obradovic et al., 2018). As a result, consequences of developmental GA exposure are now one of the most intensely studied areas in anesthesiology (Lin et al., 2017). However, it is likely that such GA effects may extend beyond neonatal-early childhood periods. Currently, the developmental window for vulnerability to enduring GA effects is not known.

Adolescence is estimated as years 12-22 in humans and postnatal days (P) 28-60 in rats and encompasses specific neuronal, hormonal, and behavioral changes that extend well beyond puberty into early adulthood (Spear 2000). Prior to adolescence, a wave of synapse production occurs that is followed by an extended period of extensive synaptic elimination and refinement (Bianchi et al., 2013). This latter period leads to decreased dendritic branches and fewer total, but more mature dendritic spines in adulthood, particularly in late-developing forebrain structures (Koleske, 2013; Kanjhan et al., 2016). This process typically augments stable neural network connectivity by eliminating unnecessary or weak connections. Synaptic pruning in late-developing regions is directly associated with expression of adolescent-typical behaviors (see: Spear, 2011; 2013 for review). Synaptic regulation of neurotransmitter receptors during adolescence is also important for behavioral maturation and closing the window of developmental neuroplasticity. In fact, the potential for neuroplasticity is maintained until GABAergic inhibition reaches a threshold, effectively reducing synaptic reorganization (Fagiolini and Hensch, 2000). Importantly, increases in GABA_A-R subunits are necessary for closing the critical period of neuronal development (Hensch, 2005). Extrasynaptic GABA_A-Rs in particular impact neuronal arborization in response to GAs (Mintz et al., 2013), are ontogenetically regulated during adolescence (Santerre et al., 2014) which coincide with this period of synaptic refinement, and are highly sensitive to potentiation by isoflurane (Jia et al., 2008).

Unfortunately, there is a dearth of information regarding enduring effects of adolescent GA exposure. Each year, millions more adolescents are exposed to GAs than children due to necessary or elective surgical procedures including third molar removal, self-image correction (plastic and bariatric surgery), as well as orthopedic procedures and trauma through sports or adverse events – events that are substantially more prevalent in adolescents than in younger children (ADA, 2007; Friedman, 2007; Tsai et al., 2007; ASPS, 2012; Kelleher et al., 2013; Inge et al., 2014). Therefore, we determined whether adolescent GA exposure influences behaviors associated with late developing brain structures such as the prefrontal cortex (PFC) and hippocampus (HPC). In addition, we assessed dendritic spine morphology and related neurotransmitter receptor maturation after adolescent GA exposure. Our findings support immediate and persistent alterations in affective responding along with a delayed onset in cognitive impairment. These effects are modulated in part by early-onset GABA_A receptor extrasynaptic inhibitory plasticity that effectively clamps typical adolescent maturation.

Methods:

Subjects and Isoflurane Exposure:

Male Sprague-Dawley rats were used for all studies. Juvenile rats were weaned on P21 and group-housed with same-sex littermates in a temperature-controlled (22°C) vivarium maintained on a 12 hr/12 hr light/dark cycle (lights on at 0700) with ad libitum access to food and water. All experiments were performed in accordance with guidelines for animal care established by National Institutes of Health and approved by Institutional Animal Care and Use Committees. Subjects were induced at P28 or P68 using individual induction chambers (VetEquip SOP-Compac 5, Pleasanton CA) for 40 min at a concentration of 3.0% and air flow rate of 1,000 cc/min with medical grade oxygen (100%) after visibly becoming immobile. Oxygen as our carrier gas was used as the risk of hypoxemia during exposure is significant (Ehrenfeld et al., 2010). Pulse oximetry measurements indicate O₂ saturation at 94-96%, and heart rats in the low 300s during anesthetic exposure (Physiosuite, Kent Scientific, Torrington, CT), and a reduction in respiratory rate of ~30–40%. In subjects without Physiosuite monitoring, anesthetic depth was continuously monitored via respiratory rate. All subjects were unresponsive to a pedal reflex. Animals not meeting this criterion (e.g., respiratory depression, pedal reflex, during surgery < 1%) were excluded from further analyses. Control animals were placed alone in a holding cage for the same length of time to eliminate restraint stress that could be induced by the size of the induction chambers. Parametric studies piloting animals in induction chambers with air flow alone for the same length of time as required for GA exposed subjects to become immobile (2-3 min) did not alter behavioral responses compared to animals remaining in their home cage (p = 0.45).

Novel Object Recognition:

Cognitive functioning was assessed using a rapid NOR paradigm employed elsewhere (Swartzwelder et al., 2012) The NOR task consists of three trials: habituation, familiarization, and novel object recognition. Habituation and familiarization occurred on the same day. Habituation consisted of rats being placed in a locomotor chamber (AccuScan Instruments, Columbus, OH) enclosed in a cabinet with a small fan for white noise for a 5-minute period without any objects. Rats were then removed and placed back in their respective cage for 5 minutes. Familiarization was subsequently conducted, where rats spent 5 minutes with two identical objects [spice jar (weighed down with sand) or triangular cup] located in the far left and right corners of the testing chamber. Subjects were then returned to the colony for 24 hours, after which subjects were tested for novel object recognition using one familiar object and one novel object for a duration of 5 minutes. Objects were counterbalanced for familiar and novel as well as for side placement. No significant object preference was noted (p > 0.10). For all three phases of testing, rats were placed in the box facing the front wall (back to objects) in order to control for biased exploration. Activity chamber assessments were done using VersaMax software (Omnitech Electronics, Columbus, OH). Thigmotaxic behavior was determined using percent time spent in the corners during habituation. Object avoidance (neophobia) was calculated as the percent time spent in the front quadrants away from objects during familiarization. Increased thigmotaxis and neophobia are facets of anxiogenic responding. Novel object recognition was calculated as the percent of time spent with the novel object over both objects.

Social Investigation:

Rats were exposed to isoflurane for 15 minutes or air at P24 and social investigation testing occurred 3 weeks later at late adolescence (P46). One day before testing (P45), each rat was habituated to the test apparatus for 30 minutes. The Plexiglas test apparatus (designed in-house; 30 × 20 × 20) contained clean pine shavings and was divided into two equally sized compartments by a partition that contained an aperture (7 × 5 cm) to allow movement between compartments. Rats were isolated for 30 minutes in the Plexiglas apparatus and exposed to a non-manipulated social partner (non-littermate) of the same age for 10 minutes. During the 10-minute test session, social investigation was recorded as we have done previously (Varlinskaya et al., 2009). Social investigation was considered sniffing of any part of the body or the foreign partner.

Electro-Foot Shock Aversive Water Drinking Test (AWDT):

The AWDT is a rapid task for assessing impulsivity in rodents (Yang et al., 2015). The testing chamber (designed in-house) was a 60 × 60 × 30 cm box divided into three compartments – a start area (front third of chamber) with smooth Plexiglas floor, open area with a mesh floor, and a small 15 cm square grid section in the back corner containing a water sipper tube. The AWDT consists of two phases: a two-day training phase followed by a testing day. Rats were water deprived for 24 hours during the three-day period, with the only other access to water during 20 minutes per day on training and testing phases. During training, rats were placed in the start area and monitored by a trained observer. Whenever rats continuously licked water from the water bottle for at least 5 seconds, they were placed back in the ‘start area’. If rats drank for less than 5 seconds, then they were not moved. The 5-second removal was employed to reduce satiation and allow for returned visits to the water bottle. During training phases, rats were never shocked. Each training day consisted of two 10-minute sessions separated by a 60-minute inter-trial interval. On the test day, rats were placed in the chamber for 10 minutes. Each time subjects drank from the water bottle for > 5 seconds, the shock grid was manually activated, delivering a 1 mA current and then immediately turned off. Impulsive responding was considered as the number of drinking bouts greater than 5 seconds, while ‘impulsive attempts’ is considered the total number of bouts.

Dendritic Spine Analysis:

6.5 weeks following GA or air exposure, rats were administered an overdose of sodium pentobarbital (100mg/kg) and immediately decapitated. Tissue was immediately sliced into 8mm cross-sections and Golgi stained: tissue was immersed in a mixture of potassium chromate, potassium dichromate, and mercuric chloride from the FD Rapid Golgi Stain Kit (FD NeuroTechnologies, Inc., Columbia, MD). The Golgi solution was refreshed within 24 hours following the initial submersion and absorbed for 2 weeks while protected from light. Tissue was then transferred to a 30% sucrose solution for 24 hours, flash frozen with isopentane and sliced into 100μm coronal sections. Brain slices containing the desired brain regions (PFC and HPC) were mounted onto 2% gelatinized microscope slides. 72 hours later, slides were immersed in an ammonium hydroxide solution from the kit and subsequently dehydrated with 50%, 75%, 95%, and 100% EtOH. Slides were cleared in xylene and coverslipped using Permount mounting medium (Fisher Scientific, Pittsburgh, PA). Pyramidal neurons in layer 5 of the rat infralimbic PFC and also the CA1 and CA3 of the HPC were analyzed for dendritic spine density. Pyramidal neurons were defined by the presence of a basilar dendritic tree, a distinct, single apical dendrite, and dendritic spines. Secondary and tertiary dendrites located at least 50 μm from the soma were randomly selected, and dendritic segments 10 μm in length were analyzed for dendritic spine density at 100x using Neurolucida software (MBF Bioscience, Williston, VT). The experimenter was blind to condition.

Western Blot Analysis:

One or 6.5 weeks following isoflurane or air exposure, animals were rapidly decapitated and brains micro-dissected. The cortex and HPC were isolated for one-week analysis, while persistent effects were assessed in the PFC and HPC at 6.5 weeks. Whole cell homogenates were isolated and processed as we have done previously (Santerre et al., 2014). Membranes were probed using antibodies directed against GABA_A receptor α4 (Millipore, Temecula, CA), δ (Novus, Littleton CO; SCBT, Dallas, TX) and protein loading verified using β-actin (Millipore). Densitometric analysis was conducted using NIH ImageJ.

Whole-Cell Recordings:

Hippocampal brain slices were prepared from adolescents and adults one week following isoflurane or air. Rats were briefly anesthetized with halothane vapor and decapitated. Slices were prepared and data assessed as we have done previously (Fleming et al., 2007). Potentiation of tonic current by the extrasynaptic GABA agonist 4,5,6,7- tetrahydroisoxazolo[5,4-c] pyridin-3-ol (THIP, 1μM) after isoflurane exposure was assessed via whole cell electrophysiological responses in dentate gyrus granule cells. The electrode solution consisted of (in mM) 130 CsCl, 10 HEPES, 4 NaCl, 0.2 EGTA, 10 Na₂CreatinePO₄, 4 MgATP, 0.3 TrisGTP, 6 QX-314, pH 7.2, 290 mOsM. THIP was diluted in ACSF and bath applied. Whole cell voltage-clamp recordings were performed using an Axopatch 200B amplifier (Axon Instruments). Signals were low-pass filtered at 2 kHz and digitized at 10 kHz using a National Instruments PCI-6251 DAQ board (Austin, TX) and WinWCP (University of Strathclyde) or Axon Instruments Digidata 1440A and Clampex 10.2 (Molecular Devices, Sunnyvale, CA). Series and input resistance were continuously monitored using a 200-ms, −10-mV pulse applied every 12.6 s. Tonic current measurement was performed using an in-house function written for MATLAB (V7.1, The Mathworks, Natik, MA). Transients from voltage steps were removed from the data. All-point histograms were generated for one min of data, and the center of this distribution, representing the mean holding current, were calculated by applying a Gaussian function f(x) = A·exp(-(x-μ)²/2σ²) to each histogram.

Stereotaxic surgery:

The GABA_AR δ -shRNA AAV (Serotype 5) was packaged by ViGene (Rockville, MD). A 21-nt siRNA sequence (5′- GGACGUGAGGAACGCCAUUGU-3′) of the rat GABA_AR δ -subunit mRNA was designed along with a control virus. Rats received isoflurane during site-specific vector delivery or were exposed to air at P28-30. Previous work has verified that the shRNA sequence efficiently and reliably knocks down GABA_A-R δ subunit, and remains active for 2 weeks when packaged in a lentiviral vector (Nie et al., 2011). Rats were randomly assigned to one of three groups: 1) air exposure for 40 minutes (Air), 2) isoflurane and AAV injection of scrambled sequence (Sham), or 3) isoflurane and AAV injection of GABA_A-R δ (δ shRNA). Anesthesia was maintained with 2.5-3.0% isoflurane for the duration of the surgery (40 min-1 hour). AAVs were infused bilaterally to either the infralimbic PFC (+3 AP, +/− 0.8 ML, and −4.0 DV) or the HPC (−3.3 AP, +/− 2.3 ML, −2.5 DV) with a syringe pump housing 10 μL gastight syringes (Hamilton Co., Reno NV) connected with polyethylene to 30-gauge stainless steel tubing. Vectors were bilaterally injected into the PFC or HPC at a flow rate of 0.2 μL/min, until a total of 1uL was infused. Needles remained in place following infusion for 2 minutes in order to prevent backflow into the needle tract. When surgeries were completed in less than 40 minutes, rats remained anesthetized until the desired 40-minute time point concluded. Behavioral experiments were conducted 2.5-3 weeks following surgery, after which brain tissue for the injected regions was harvested to determine knockdown efficiency.

Data Analysis:

Given the experimental design or a priori comparisons, datasets were analyzed using Student’s t-test with Bonferroni correction if needed. For all analyses, the criterion for significance was set at p < 0.05.

Results

Adolescent GA exposure persistently increases anxiety-like responding and precipitates delayed cognitive dysfunction.

Given prior findings revealing cognitive impairments following early developmental GA exposure, we initially assessed memory performance using the novel object recognition task. Novel object recognition was initially assessed one week following GA exposure while subjects were still in adolescence (P35). At this time point, isoflurane-exposed subjects did not differ from controls in time spent exploring the novel object (Figure 1A). However, they did display greater anxiety-related responding during the training portion of the task. During habituation to the testing apparatus (without objects), isoflurane-exposed subjects displayed a strong trend toward greater thigmotaxis (staying near the walls/corners) relative to controls (t₂₁ = 2.025, p = 0.055, Figure 1C). Isoflurane exposure also increased neophobic-like responding, as noted by time spent away from objects during familiarization relative to controls (t₂₁ = 2.354, p = 0.028, Figure 1E). We next tested behavioral responding in a separate cohort 6.5 weeks later when rats had aged into adulthood. In contrast to one-week, adolescent isoflurane-exposed rats tested as adults exhibited enduring memory impairment, as indicated by decreased time spent exploring the novel object compared to controls (t₁₃ = 3.990, p = 0.0008, Figure 1B). While thigmotaxic responding did not differ (Figure 1D), cognitive impairment was accompanied by an almost two-fold increase in neophobic responding (t₁₃ = 4.143, p = 0.0006, Figure 1F). Adults exposed to isoflurane did not differ from controls on novel object recognition (not shown). Given that adolescent GA exposure markedly and persistently enhances nonsocial anxiety-related responding, we next determined whether heightened anxiety was specific or more generalized. Using a social interaction task, which is more sensitive to anxiety-related perturbations, rats that received isoflurane for even a shorter duration (15 minutes) during adolescence and tested into early adulthood also displayed a slight but significant reduction in social investigation (reduced 9±3% in isoflurane-exposed subjects compared to controls, Figure 2A).

Figure 1.

Adolescent isoflurane exposure altered cognitive and anxiety-related behaviors into adulthood. There was no impact on novel object recognition (NOR) 1 week following adolescent GA exposure (A), but NOR was impaired in GA-exposed adolescent rats once they had aged into adulthood (B). 50% represents chance (no recognition). Anxiety-like behaviors were also assessed during habituation and familiarization phases of the NOR task. During habituation, thigmotaxic behavior was increased 1 week following adolescent GA exposure (C), although this effect did not persist into adulthood (D). During familiarization, adolescent GA exposure lead to an overall avoidance of objects that was observed 1 week (E) and was present at least 6.5 weeks post exposure (F). All data are compared to age-matched controls. Data are represented as mean ± SEM. * = p < 0.05; n = 7-8/group.

Figure 2.

Adolescent isoflurane exposure reduced social activity and persistently altered impulsive-like behavior. Social interaction was slightly reduced following 3 weeks following a short 15-min adolescent GA exposure (A). The AWDT measures the willingness of a water-deprived rat to risk the exposure to shock in order to gain access to water. Drinking bouts resulting in shock are considered impulsive. One week following adolescent GA exposure, rats exhibited a decrease in the ratio of shocked bouts (B). However, when animals had aged into adulthood, an increase in impulsive-like responding was observed following adolescent GA exposure (C). Data are compared to air-exposed age matched controls. Data are represented as mean ± SEM. * = p < 0.05, ** = p < 0.01; n = 6-7/group.

Adolescent GA exposure maintains adolescent impulsivity into adulthood.

Impulsivity measures (e.g., delayed discounting) require extensive training, therefore we employed the Aversive Water Drinking Task adapted from Kim et al (2012) and Yang et al (2015). In this task impulsivity is defined as the willingness of water-deprived subjects to potentially expose themselves to an aversive shock in order to gain access to water. We found that control adolescents experienced two times as many drinking bouts that received a shock than adults (t₁₁ = 2.312, p = 0.041, comparison of control groups in Figures 2B and 2C) without a difference across ages in total drinking attempts (not shown). Interestingly, adolescent subjects tested one week following isoflurane exposure had a lower proportion of total bouts that received a shock compared to controls (t₁₂ = 3.731, p = 0.003, Figure 2B). Strikingly, GA-exposed adolescents displayed a nearly three-fold increase in total drinking attempts as measured by the total number of drinking times during testing compared to controls (t₁₂ = 4.688, p = 0.0005, Control = 20.29 ± 5.71, Isoflurane = 57.14 ± 5.405) as well as a strong trend towards increased attempts during task training (not shown). Given these findings, we predicted that impulsive responding would be increased into adulthood following GA exposure. We used the same task on a separate cohort that was tested in adulthood following adolescent isoflurane exposure. The proportion of total bouts resulting in a shock was greater in subjects exposed to isoflurane during adolescence compared to controls (t₁₀ = 2.153, p = 0.028, Figure 2C), and total impulsive attempts did not differ during training or testing (p = 0.08 and p = 0.60, respectively). Total drinking attempts did not differ in adulthood between controls and adolescent-exposed groups (not shown). Taken together, these data suggest that adolescent GA exposure unmasks impulsive tendencies that are manifested in adulthood.

Dendritic spine maturation is blunted in late-developing regions following adolescent GA exposure.

Given that multiple brain regions undergo synaptic elimination and refinement throughout adolescence, we next determined if persistent GA-induced behavioral abnormalities were associated with changes in dendritic spine morphology (Figure 3A). We found that dendritic spine density was increased in layer 5 pyramidal neurons of the medial PFC in adulthood following adolescent GA exposure (t₁₂ = 2.163, p = 0.026, Figure 3B). In addition, pyramidal cell dendritic spine density was also increased in the CA1 region of the hippocampus (t₁₆ = 2.344, p = 0.016, Figure 3C), though not in area CA3 (Figure 3D).

Figure 3.

Adolescent isoflurane exposure increased dendritic spine density on pyramidal cells in adulthood. Representative images of PFC dendritic spines (A). Overall spine density was increased on secondary and tertiary apical dendrites in layer 5 of the infralimbic PFC (B), and the CA1 region of the HPC (C), 6.5 weeks following adolescent GA exposure. There were no differences in spine density in the CA3 (D). Data are compared to air-exposed age matched controls. Data are represented as mean ± SEM. * = p < 0.05, ** = p < 0.01; n = 8-9/group.

Adolescent GA exposure results in premature increases in extrasynaptic GABAA receptors associated with tonic inhibition.

We next investigated expression of the δ subunit that is exclusively associated with extrasynaptic GABA_A receptor subtypes and often pairs with α4 subunits (Kumar et al., 2009; Brickley and Mody, 2012). As a positive control, we first verified typical ontogenetic levels of extrasynaptic subunit expression. GABA_A receptor δ subunit levels were 56% (t₉ = 6.601, p < 0.0001) and 29% (t₉ = 2.900, p = 0.009) lower in the cortex and HPC, respectively, of adolescents compared to adults. Reduced delta subunit expression during adolescence is in line with our previous findings (Fleming et al., 2007; Santerre et al., 2014). Interestingly, one week following adolescent GA exposure α4 and δ subunits were increased in the cortex (t₁₀ = 3.014, p = 0.007 and t₉ = 2.165, p = 0.029, Figure 4A), whereas δ, but not α4, was increased in the HPC (t₁₂ = 3.583, p = 0.002, Figure 4B). No effects of GA exposure were noted in extrasynaptic subunit expression in adults one week following isoflurane exposure in adult animals (not shown).

Figure 4.

Adolescent isoflurane exposure increased inhibitory extrasynaptic GABA_A receptor subunit expression and function. Cortical δ and α4 subunits were upregulated one week following adolescent isoflurane exposure (A). δ subunits were increased in the HPC one week following exposure, though α4 was not (B). Potentiation of tonic current by THIP was increased in the HPC one week following adolescent GA exposure. Adult GA exposure did not alter THIP potentiation. Representative tracings were chosen from randomly selected recordings (C). Data are represented as mean ± SEM. * = p < 0.05, ** = p < 0.01; n = 6-8 animals/group, n = 18-22 cells/group.

To further corroborate increased extrasynaptic GABA_A receptors following adolescent, but not adult GA exposure, we examined tonic current potentiation by THIP, a highly selective extrasynaptic GABA_A receptor agonist. We restricted responding to HPC dentate gyrus granule cells, as tonic current within these cells is predominantly mediated by δ-containing receptors (Brickley and Mody, 2012). Similar to our protein analyses, potentiation of tonic current was again lower in adolescents versus adults (t₂₄ = 1.783, p = 0.044, a priori comparison of controls, Figure 4C). THIP further potentiated tonic current one-week later in subjects exposed to isoflurane during adolescence, compared to age-matched controls (t₃₈ = 2.104, p = 0.042), but not when isoflurane exposure occurred in adulthood. Given that these robust increases were observed only in adolescents following GA exposure, we questioned whether elevations in extrasynaptic GABA_A receptor inhibition persisted into adulthood. No differences were noted in the PFC for either α4 or δ subunit levels, whereas δ, but not α4, remained elevated in the HPC (t₁₃ = 2.416, p = 0.016, Figure 5). Taken together, these findings suggest that reduced dendritic spine maturation is associated with premature increases in GABA_A receptor extrasynaptic receptors following adolescent GA exposure.

Figure 5.

Adolescent isoflurane exposure persistently increased hippocampal inhibitory extrasynaptic subunits. Cortical levels of GABA_A-R subunits δ and α4 were unaltered 6.5 weeks following adolescent isoflurane exposure (A). In the HPC, increased δ expression was sustained into adulthood, while there was no difference in α4 expression (B). Data are represented as mean ± SEM. * = p < 0.05; n = 6-9/group.

Mitigating premature increases in extrasynaptic GABA_A receptors following adolescent GA exposure restores typical impulsivity responding

To determine whether GA-induced increases in δ-containing extrasynaptic receptors modulated behavioral responding, we infused an adenoviral-associated vector (AAV, serotype 5) containing a previously verified shRNA sequence directed to the δ subunit (Nie et al., 2011) directly into the infralimbic PFC or HPC for impulsivity or cognition tasks, respectively. Infusions occurred at the time of isoflurane exposure, and impulsive and cognitive responses were then tested three weeks later, respectively. Positive (sham) control subjects also experienced isoflurane during the surgical procedure but were instead injected with a sham AAV5 viral vector. Similar to studies above, a third group that did not receive isoflurane/surgery was included as a control. Sham-exposed subjects again displayed an increase in impulsive responding as evident by the increased ratio of shocked bouts (t₁₀ = 2.345, p = 0.041, Figure 6A), whereas δ shRNA subjects did not differ from controls. Total impulsive attempts did not differ for either group. Regression analysis of impulsive responding relative to the expression of delta subunit levels for δ shRNA-exposed subjects were positively correlated (R² = 0.8423, p = 0.004, Figure 6E). For novel object recognition, both sham- and δ shRNA-exposed subjects displayed reduced novel object preference when compared to controls [t₁₅ = 2.925, p = 0.011 (Sham), t₁₃ = 2.871, p = 0.013 (shRNA), Figure 6B], and a strong trend for an increase in neophobic-like responding (object avoidance, both t_13-15 = 1.611, p = 0.064), but no difference in thigmotaxis (Figure 6C-D). Notably, these findings indicate that adolescent GA exposure gives rise to cognitive deficits in as early as three weeks post exposure.

Figure 6.

Inhibiting δ subunit upregulation in the PFC reverses persistent impulsivity following adolescent GA exposure. All data are compared to control values (indicated via dotted line). While sham rats exhibited greater impulsivity following GA exposure, δ shRNA knockdown in the PFC restored impulsivity to control levels (A). Knockdown of δ in the HPC did not reverse GA-induced memory deficits; both sham and δ shRNA rats exhibited memory impairment 3 weeks following adolescent GA exposure (B). GA-induced anxiety behaviors were not altered following HPC δ knockdown (C & D). Impulsivity Ratios were correlated to the amount of GABA_A receptor δ subunit expression (E). Data are represented as mean ± SEM. * = p < 0.05; n = 6-9/group.

Discussion

Given the immense attention that neonatal GA exposure has received, surprisingly little work has investigated the impact of GA exposure during adolescence. Further, the impact of developmental GA exposure on receptor plasticity is not completely understood. The present study utilized a 40 minute exposure to model the duration of a short, one-time less-invasive surgical procedure commonly used in humans (Garip et al., 2007). These findings demonstrate that a brief GA exposure during adolescence is sufficient to induce persistent abnormalities in cognition and affect-related responding, as well as in dendritic spine density and synaptic inhibitory function. The FDA has recently issued a warning about the use of GAs in pregnant women and children under the age of three (FDA, 2017). The present results, when combined with existing studies, suggest that this recommendation should be modified to include adolescence as a potentially vulnerable period. As an increasing number of studies indicate more persistent effects of GAs, further work into defining the age-related markers for GA vulnerability is essential, particularly as related to adolescence. Given the massive longitudinal Adolescent Brain Cognitive Development (ABCD) study being conducted by multiple NIH institutes, it is conceivable that surgical experience could also be tracked. This need is underscored by the fact that adolescence is a period during which multiple neuropsychiatric disorders often emerge, including substance use disorders and their precursor behaviors. In fact, a recent JAMA report linked a three-fold increase in opioid prescription refills in young individuals to third molar removal (Harbaugh et al 2018).

Some recent publications have called into question the persistence of cognitive deficits following early childhood exposure to GAs (e.g., Sun et al., 2016), but the extant preclinical and clinical literature overwhelmingly supports the occurrence of such abnormalities after GA exposure (see: Lin et al., 2017 for review). Such outcomes are particularly obvious in studies that use animal models to eliminate the confound of pre-existing conditions and clinical complications – a major limitation of post-surgical procedure evaluations in young children. Nonetheless, it should be acknowledged that discrepancies continue to be published. While the underlying factors contributing to these differences are not yet verified, there are some important considerations. First, there may be a ‘cut-off’ age in exposure timing such that young children demarcating vulnerability/resilience to a single exposure as noted in recent multi-site studies (Davidson, et al., 2016; Sun et al., 2016). Second, select developmental periods may be resilient to a single exposure, but develop cognitive abnormalities only following repeated exposures (Coleman et al., 2017), thereby unmasking of developmental vulnerability. Finally, surgical contributors, such as a systemic inflammatory events that contribute to neuroinflammation, may be required to unmask developmental exposure parameters (Terrando et al., 2011; Vacas et al., 2013). Apart from early childhood exposures, to our knowledge, this is the first study to directly assess effects of anesthesia in adolescence, a second developmental period that is arguably as critical as early childhood neurogenesis due to the pronounced brain maturation that occurs within and well after the pubertal period (see: Spear 2018). The current findings further support the need for more parametric studies to define developmental periods of vulnerability and resilience. Moreover, given that subjects in the current study that experienced a surgical procedure not only replicated cognitive deficits at an earlier time point as those without surgery highlight the need to further characterize post-exposure consequences, and whether post-operative inflammation acts as an accelerant.

Most studies of developmental GA exposure focus on memory and cognitive tasks. The present findings both add to that accumulating literature and provide a mechanistic framework for interpreting those findings and generating hypotheses for future behavioral analyses. For instance, it may be valuable for future studies to focus on tasks that assess specific aspects of cognitive functioning (e.g., set-shifting or delayed discounting) rather than methods that assess only coarse changes in general aspects of cognition. For example, in the present study, we used a modified novel object preference task as used previously (Swartzwelder et al., 2012) that allows for distinctions in memory consolidation and/or retrieval, as well as non-cognitive measures. To further refine our understanding of the enduring effects of GA exposure, future behavioral studies could be selected based on the typical neurobiological events undergoing plasticity/maturation within the defined developmental periods (see: Spear, 2000). Additionally, it is possible that in previously reported negative results, a longer incubation period between exposure and testing may have been required.

One notable behavioral finding is that isoflurane exposure during adolescence resulted in increased impulsivity-like responding. Adolescent animals are more impulsive than adults as noted by greater risk-taking. Adolescent impulsivity is an evolutionarily conserved phenotype that typically diminishes during the transition to adulthood (Spear, 2000). Interestingly, different facets of impulsivity were affected depending on the incubation period post-exposure, with increased number of total attempts as a measure of compulsive responding (see: Jonkman et al., 2012; Robbins et al., 2012) first being increased, which then transitioned towards impulsive responding into early adulthood. Irrespective, it appears that adolescent GA exposure may lead to persistence of this adolescent-typical behavior. Persistence of adolescent typical characteristics into adulthood has been observed previously in rodents after adolescent intermittent ethanol exposure, including measures of acute ethanol sensitivity, neocortical EEG features, and GABA_A receptor-mediated function (Spear and Swartzwelder, 2014). Although more studies of adolescent GA exposure specifically designed to assess this ‘lock-in’ effect with GAs are needed, this initial finding raises the important question of how such exposure may alter neurobehavioral developmental trajectories. The aversive water-drinking task was chosen over classical measures of impulsivity due to its minimal training period, which allowed for testing during adolescence. Given the current age-dependent findings, a more nuanced assessment of impulsivity into adulthood are also needed.

Although the current findings indicate that adolescent GA exposure enhances anxiety and impulsivity as well as impairs cognition, it is possible this cluster of behaviors is interactive. For example, the shift in novel object recognition that we conservatively interpret as a change in memory function could be mediated, in part, by changes in anxiety- or impulsivity-related behaviors. Similarly, increased anxiety could also alter ‘compulsive drinking’ on the aversive water drinking task. Conversely, cognitive impairment may reduce GA subjects’ ability to process shock with length of drinking during impulse testing. However, regardless of the possible interactions between GA-induced behavioral effects, the fact that animals were functionally impaired following exposure indicates the need for future studies utilizing multiple tasks to help delineate the specific domains in which enduring sequelae are manifested. From a neural perspective, it is noteworthy that we observed several distinct but related effects of adolescent GA exposure that indicate adolescent brain vulnerability. But, as is often the case, the relationship of neural findings to the behavioral outcomes is not necessarily clear. Indeed, extrasynaptic GABA_A receptor knockdown in the dorsal HPC did not impact novel object recognition. The HPC was selected as it is a convergent region in cognitive processing, but novel object recognition involves a complex interaction of multiple brain regions, and it is highly likely that non-hippocampal regions such as the PFC or entorhinal cortex may be influencing GA-induced cognitive alterations in adolescents (Benn et al., 2016; Chao et al., 2016; Fernandez et al., 2016). While δ-containing receptors are implicated in memory processes (Whissell et al., 2016), alternatively, future studies should investigate other extrasynaptic GABA_A receptor subtypes, such as those containing α5, known to play a role in short-term memory (Hausrat et al., 2015). Nonetheless, it is also not surprising that anxiety-related responding was not altered following δ knockdown, as the dorsal HPC has a limited role in anxiety circuitry.

Of course, the present results must be interpreted conservatively within the clinical framework. Importantly, it should be stated that the current work should not discourage the use of GAs within vulnerable age periods when absolutely necessary. However, these findings, combined with the emerging related literature, do suggest that adolescence may represent a developmental period with elevated sensitivity to enduring GA effects relative to adulthood. We believe that the present findings will provide a specific and mechanistic backdrop for the emergence of additional clinical and pre-clinical studies to address this issue from a directly translational standpoint, particularly as recent efforts seek to mitigate the consequences of exposure during periods of vulnerability (e.g., Tachibana et al., 2012). As adolescence is complex, future work also needs to address late adolescence and emerging adult periods to further pinpoint the window of vulnerability (Spear, 2015). The current findings may also impact future developmental studies that require surgical manipulations. As GA exposure produced neural and behavioral effects that differ from non-exposed age-matched subjects, inclusion of non-manipulated controls may become necessary in experimental designs. Although this inclusion increases animal numbers, it is important to know whether experimental results are independent of GA effects. It may also be beneficial to perform retrospective analysis of datasets to determine whether GA exposure could have influenced outcomes.

Some methodological points regarding the current study should be taken into consideration. First, it should be pointed out that the concentration of isoflurane in the current study is somewhat higher than the typical minimum alveolar concentration (MAC – concentration in which 50% of subjects are immobile to a noxious stimulus) for isoflurane in rodents (~1.5%, Sonner et al., 2007; Eger et al., 2008). The exposure employed is not atypical in rodent studies (see: Howard et al., 2008), and ensures that all subjects achieved immobility during exposure, as concentrations that fail to eliminate responses to noxious stimuli are less than ideal (Antognini & Carstens, 2005). Developmental periods may require higher effective doses than adults (see: Fang et al., 1997). In fact, MAC in early developing male Sprague-Dawley rats, for the first hour of exposure is ~3.2% isoflurane (Lee et al., 2014). The current study also had a duration of exposure that was substantially less than typical neonatal exposures, (~5.5 hours, Lin et al., 2017). Second, while the respiratory rate was slower during exposure, it is unlikely that hypoxemia or hypoperfusion was a contributing factor as isoflurane was supplemented with oxygen and pulse oximetry measures were in the mid 90-percentile. Similarly, hyperoxemia, of neuronal tissue is also less likely to contribute to age-dependent effects given the lack of behavioral or biochemical alterations in adults. Third, while isoflurane can cause body temperature dysregulation, findings are unlikely related to hypothermia. During surgical manipulations, normothermic limits were maintained with a circulating water bath, and sham subjects (Figure 6) displayed impaired cognition along with anxiogenic and impulsive-like responding. Thus, it is more than likely that observed behavioral effects are primarily due to isoflurane.

Apart from these, all experiments were conducted exclusively in males. Current investigations needed to be restricted to males as extrasynaptic GABAA receptor expression and function are impacted across the estrous cycle and early adolescence (Wu et al., 2013; Barth et al., 2014; Cushman et al., 2014; Afroz et al., 2016).Studies have utilized both male and female neonatal rats and found similar behavioral and neurobiological maladaptations (see: Jevtovic-Todorovic et al., 2003; Scallet et al., 2004; Briner et al., 2011), whereas other research indicates that only males (Lee et al., 2014; Tan et al., 2014) or only females (Aligny et al., 2014; Takaenoki et al., 2014) exhibit deficits in cognitive/affective behaviors following early GA exposure. Elsewhere, while cognitive/affective impairments are present in both sexes following neonatal GA exposure, the effect is greater in males (Rothstein, Simkins, and Nunez, 2008). More studies investigating sex differences following developmental GA exposure are clearly needed.

The present findings also open the possibility that the enduring effects of adolescent GA exposure may depend upon the specific age at which exposure occurs. In the rat, there is an extensive period of synaptogenesis which occurs until approximately P14 centered around synaptic growth (for review: Anderson, 2003). Once this period of synaptogenesis has ended, adolescent neural maturation depends on pruning of unnecessary synaptic connections in order to increase efficiency of neural connections, a process critical for typical behavioral maturation (Grutzendler et al., 2002; Duan et al., 2003). The PFC and HPC in particular undergo enhancement and refinement of neural networks and dendritic spines throughout the adolescent period (Gogtay et al., 2004; Casey et al., 2005; Petanjek et al., 2011). For instance, approximately 96% of adult spines compared to 73% during early adolescence in rodent cortical layer-5 pyramidal neurons are considered stable, with the remaining spines undergoing elimination (Grutzendler et al., 2002). Changes in PFC as well as HPC dendritic spines are also strongly associated with impulsivity and cognitive functioning (Chen et al., 2013; Benn and Robinson, 2014; Risher et al., 2015). Thus, the underlying neurodevelopmental state likely plays an important role on the precise impact of developmental GA exposure thereby resulting in developmentally dissociable effects. Supporting this, during the juvenile/pre-adolescent period, evidence suggests that GA exposure increases, rather than decreases dendritic spine density, whereas levels are reduced following earlier neonatal exposure (Briner et al., 2010; 2011). However, unlike neonates, juvenile rats do not appear to exhibit neurodegeneration following GA exposure (Ikonomidou et al., 2000; Briner et al., 2010). Developmental dissociable effects are further supported by ethanol, which shares common neurobiological substrates as GAs (Howard et al., 2014). Adolescent intermittent ethanol exposure leads to altered LTP, increased dendritic spines, coupled with cognitive impairment and anxiety (Spear and Swartzwelder, 2014; Risher et al., 2015). Nonetheless, further work needs to delineate whether adolescent GA-induced retention of dendritic spines are the result of inability to prune, or continuation of adolescent typical spine dynamics into adulthood.

The current study also suggests that tonic inhibition via extrasynaptic GABA_A receptors following GA exposure plays a critical role in developmental synaptic plasticity. Relatively little is known about the role of extrasynaptic GABA_A-R subunits on synaptic pruning. One study found that knockout of α4 subunits prevented the synaptic pruning phase (Afroz et al., 2016). Wild-type female mice exhibited a 50% decrease in hippocampal dendritic spines across puberty, an effect that was dependent upon inhibition of NMDA-Rs via α4βδ-containing GABA_A-Rs. Inhibition of NMDA was not observed following α4 knockouts, and synaptic pruning did not occur. Therefore, in the CA1 region of the HPC, increased GABA is associated with synaptic pruning, particularly through remodeling of cytoskeletal elements. However, this study was done in female mice, during a time at which α4βδ GABA_A-Rs fluctuate throughout the estrous cycle (Wu et al., 2013; Barth et al., 2014). Further, while delta subunits are found in the CA1, tonic inhibition in these cells are primarily mediated by α5 extrasynaptic receptors (Caraiscos et al., 2004). Similarly, GABA_A receptors in the prefrontal cortex are also more heterogeneous. As current electrophysiological approaches focused on the HPC dentate granule cells, as tonic current is largely attributed to α4/δ-containing extrasynaptic receptors that are robustly expressed in this cell type, the present findings only indirectly link extrasynaptic inhibitory activity to dendritic maturation. Nevertheless, it seems that more attention to the question of GABA_A receptor mediation of early and adolescent neural circuit development would be valuable.

Extrasynaptic GABA_A receptors have been associated with cognition and anxiety. Acutely increasing activation of δ-containing receptors through THIP administration was sufficient to induce loss of memory using several memory tasks, including NOP, and reduced LTP that was absent in δ knockout mice (Whissell et al., 2013). The GABAergic system is well described as being critical for anxiety, and disinhibition as a result of extrasynaptic GABA_A receptor activity is suggested to contribute to anxiety/fear responding (Liu et al., 2017). While the GABAergic system also appears to play a role in regulation of impulsive behaviors (Lane et al., 2005; Lee et al., 2009; Hayes et al., 2014), the present study may be one of the first to directly implicate PFC extrasynaptic δ subunits in impulsive behavior. However, previous studies administering the GABA_A-R agonist muscimol into the PFC induced impulsive-like behavior using the delay-discounting task (Murphy et al., 2012), and other work with knockout models indicates that muscimol may be more selective for extrasynaptic receptors than previously thought (Chandra et al., 2010).

Although our data strongly suggest a role for extrasynaptic receptors in mediating adolescent GA effects, their involvement may co-occur with other molecular events. For example, epigenetic mechanisms such as hypermethylation of BDNF and Reelin genes have been observed following a sustained neonatal exposure to sevoflurane, and were associated with memory impairment and decreased dendritic spine density that were reversed by a DNA methyltransferase (DNMT) inhibitor (Ju et al., 2016). BDNF and c-Fos hypoacetylation has also been noted in early postnatal organotypic cultures (Dalla Massara et al., 2016). Other studies have also found that DNA methylation is altered in aged rats, an effect that associates with long-lasting spatial memory impairment (Culley et al., 2004). Future studies should assess the putative epigenetic mechanisms involved in adolescent GA exposure. Further, synaptic pruning also occurs through either retraction of synapses or removal by glial cells (Schafer et al., 2012; Chung et al., 2013). Whether glial-mediated synaptic pruning is reduced following adolescent GA exposure has yet to be determined.

In summary, the current findings further expand the developmental window of GA vulnerability to include the adolescent period, and provide mechanistic insight into the enduring effects of adolescent GA exposure on the brain and behavior. Specifically, we show that GA exposure hinders typical dendritic maturation and is associated with premature increases of inhibitory function during adolescence, which, when reversed, are able to modulate the emergence of some atypical behavioral responding from adolescent GA exposure.

Significance Statement.

• Adolescence is a critical developmental period, during which the consequential effects of general anesthetics are not well defined.

• The present work demonstrates that adolescent anesthetic exposure alters cognition, impulsivity and anxiety related behaviors in adulthood.

• In parallel with those enduring behavioral alterations, adolescent anesthetic exposure also altered dendritic spine morphology and physiological inhibitory function in the hippocampal formation and frontal cortex.

• These results demonstrate enduring neurobehavioral effects of adolescent anesthetic exposure and provide mechanistic insight into their possible causes, thus informing our understanding of developmental anesthetic vulnerability and possible neuropsychiatric disorders that commonly precipitate during adolescence.

Highlights.

• Adolescence is a critical developmental period, during which the consequential effects of general anesthetics are not well defined.

• The present work demonstrates that adolescent anesthetic exposure alters cognition, impulsivity and anxiety related behaviors in adulthood.

• In parallel with those enduring behavioral alterations, adolescent anesthetic exposure also altered dendritic spine morphology and physiological inhibitory function in the hippocampal formation and frontal cortex.

• These results demonstrate enduring neurobehavioral effects of adolescent anesthetic exposure and provide mechanistic insight into their possible causes, thus informing our understanding of developmental anesthetic vulnerability and possible neuropsychiatric disorders that commonly precipitate during adolescence.

Abbreviations footnote:

GABA

γ-aminobutyric acid

GABA_A-Rs

GABA type A receptors

DG

dentate granule cells