|Home | About | Journals | Submit | Contact Us | Français|
Isoflurane and related anesthetics are widely used to anesthetize children, ranging from premature babies to adolescents. Concerns have been raised about the safety of these anesthetics in pediatric patients, particularly regarding possible negative effects on cognition. The purpose of this study was to investigate the effects of repeated isoflurane exposure of juvenile and mature animals on cognition and neurogenesis. Postnatal day 14 (P14) rats and mice, as well as adult (P60) rats, were anesthetized with isoflurane for 35mins daily for four successive days. Object recognition, place learning and reversal learning as well as cell death and cytogenesis were evaluated. Object recognition and reversal learning were significantly impaired in isoflurane-treated young rats and mice, whereas adult animals were unaffected, and these deficits became more pronounced as the animals grew older. The memory deficit was paralleled by a decrease in the hippocampal stem cell pool and persistently reduced neurogenesis, subsequently causing a reduction in the number of dentate gyrus granule cell neurons in isoflurane-treated rats. There were no signs of increased cell death of progenitors or neurons in the hippocampus. These findings show a previously unknown mechanism of neurotoxicity, causing cognitive deficits in a clearly age-dependent manner.
General anesthesia of children, ranging from premature babies to adolescents, is a common practice in modern anesthesiology, for surgery or relief from procedural pain. Isoflurane is commonly used to maintain general anesthesia in various types of surgery in patients as well as in animal models. The anesthetic action of isoflurane is thought to be mediated by multiple mechanisms, including agonist action on gamma-amino butyric acid (GABAA) receptors (Grasshoff et al, 2005) and antagonist action on the N-methyl--aspartate receptor (Harada et al, 1999). Isoflurane has long been considered safe and even neuroprotective in cell culture and animal models (Popovic et al, 2000; Sakai et al, 2007). Recent findings indicate that increased activation of GABAA receptors or decreased excitation of N-methyl--aspartate receptors could induce widespread neurodegeneration in the developing brain (Ikonomidou et al, 1999; Jevtovic-Todorovic et al, 2003). A recent population-based cohort study showed that early exposure to anesthesia was a significant risk factor for later development of learning disabilities (Wilder et al, 2009) and concerns have been raised regarding the use of anesthetics in pediatric patients. Neurons in the developing brain are particularly vulnerable to isoflurane (Jevtovic-Todorovic et al, 2003). Immature rats exposed to isoflurane displayed persistent memory and learning deficits, and this was associated with widespread caspase-3 activation and neuronal apoptosis (Culley et al, 2004; Jevtovic-Todorovic et al, 2003; Yon et al, 2005; Zhang et al, 2008). Postnatal generation of neurons occurs throughout life in certain areas of the brain, particularly the dentate gyrus of the hippocampus and the subventricular zone of the lateral ventricle walls (Curtis et al, 2007; Eriksson et al, 1998; Zhu et al, 2009). This process is more pronounced during early postnatal life and decreases with age (Qiu et al, 2007; Zhu et al, 2009). Neurogenesis in the hippocampus has been shown to be involved in memory acquisition (Shors et al, 2001), but the effects of anesthetic agents on neurogenesis are not well understood (Stratmann et al, 2009b). Serious concerns have been raised regarding the safety of certain anesthetic agents in pediatric patients, particularly patients subjected to extended and repeated surgery (Wilder et al, 2009). It is important to investigate if anesthesia induced by isoflurane and related agents may cause learning and memory deficits and, if so, to elucidate the biologic basis for this (Mellon et al, 2007).
Postnatal day 14 (P14) (n=56) and P60 (n=56) male Wistar rats or P14 male C57BL/6J mice (n=31) were randomly assigned to either isoflurane anesthesia or control. Isoflurane anesthesia was induced by placing the animal in a plastic chamber flushed continuously with either 1.7% isoflurane (Isoba, Schering-Plough Corporation, Kenilworth, NJ, USA) in a mixture of air and 100% oxygen (1:1) for 35mins daily for four successive days or the air–oxygen mixture alone at identical flow rates. The temperature of the chamber floor was controlled at 37°C with a heating pad under the chamber. Animals recovered rapidly from anesthesia and displayed no neurologic symptoms or signs of discomfort. There was no mortality during or after anesthesia. All animal experiment protocols were approved by the Gothenburg Committee of the Swedish Animal Welfare Agency (212-2005).
P14 rats were anesthetized with isoflurane (4% for induction and 1.7% for maintenance), placed on a heating pad (37°C), and rectal temperature was monitored (Physitemp Instruments Inc., Clifton, NJ, USA). Arterial pressure was recorded by using a Samba Preclin catheter inserted through the left carotid artery (Samba Sensor AB, Gothenburg, Sweden). The duration of the surgical procedure for each pup was about 10mins. Blood pressure (systolic, diastolic, and mean arterial pressure) was recorded in real-time with high resolution and precision. The pups were decapitated immediately after isoflurane anesthesia and mixed arterial and venous blood was collected from the neck vessels. Blood pH, PCO2, PO2, glucose (Glu), lactate (Lac), and bicarbonate (HCO3) were analyzed immediately in a blood analyzer (ABL 725, Radiometer A/S, Copenhagen, Denmark). For assessment of blood pressure after repeated isoflurane exposure, P14 pups were exposed to 1.7% isoflurane as described above 35mins each day for 3 days and blood pressure and blood gas analysis were performed on the fourth day (P17). The P60 rats were anesthetized for 35mins with 1.7% isoflurane and blood gas analysis was performed as above.
The thymidine analog 5-bromo-2-deoxyuridine (BrdU) (Roche, Mannheim, Germany, 5mg/mL dissolved in 0.9% saline) was prepared fresh before use and injected intraperitoneally (50mg/kg), once daily for four successive days immediately after each anesthesia.
Object recognition was tested on rats in an open field plastic arena (measuring 65cm × 48cm × 28cm with bedding material covering the floor) by an investigator blinded to the treatment of the animals (Bertaina-Anglade et al, 2006). The day before the test, each animal was subjected to a habituation session for 5mins. The test consisted of two trials with a 24h interval. During the first trial, animals were placed in the arena containing two identical objects for 5mins. The time spent exploring each object was video recorded. After each session, the objects were cleaned thoroughly, the bedding material was removed, and the test box was cleaned. The exploration of an object was defined as rearing on the object, sniffing at it at a distance of less than 2cm and/or touching it with the nose. For the second trial, one of the objects presented in the first trial was replaced by a novel object. The animals were returned to the arena for 5mins and the total time spent exploring each object was measured. Data are expressed as the recognition memory index, which corresponds to the difference between the time spent exploring the novel and the familiar objects, corrected for the total time exploring both objects.
IntelliCage is a novel and automated method for assessing spontaneous behavior and place learning in a home cage environment with social interaction (Galsworthy et al, 2005). P14 mice were randomized to isoflurane (n=16) or control (n=15) and isoflurane anesthesia was performed as described above. Mice from both groups were mixed and housed together. On the first day of the behavior test, the animals were transferred to the IntelliCages and housed in the same groups as previously. Immediately after the animals had been moved to the IntelliCages, recording of their behavior using the IntelliCage software was started. For the first 2 days of the experiment (acclimatization), the animals were free to drink from any of the water bottles available (eight per cage). For the following 2 days, the animals were required to open the doors covering the water bottle nipples by performing a nose poke to be able to drink (training). At the end of the two training days, the animals' corner activity was analyzed and the animals were assigned to the least visited corner for the following 4 days. During this period each animal could gain access to the two water bottles (one supplemented with fructose) in their assigned corner by performing a nose poke. If the animals performed a nose poke in an incorrect corner, a brief air puff was activated and the doors covering the water bottle nipples would remain closed (learning). After these 4 days of learning, the animals' corner activity was once again analyzed and the animals were once again assigned to the least visited corner as the correct corner (reversal learning). After four additional days with these settings, the animals were returned to their normal cages. Food was provided ad libitum during the whole experiment. The doors covering the bottle nipples were open for 7secs after a nose poke, after which the animals had to perform another nose poke to continue drinking. The doors were closed at the end of each visit. The data were extracted from the result files using the IntelliCage software, postprocessed in Microsoft Excel and then analyzed in SPSS (SPSS 17.0, SPSS Inc., Chicago, IL, USA). The daily activities of the animals were investigated by dividing the data into 24 h time periods.
The animals were deeply anesthetized and perfusion fixed with 5% formaldehyde in 0.1mol/L PBS, followed by immersion fixation in the same fixative overnight. Thereafter, one hemisphere was kept in 30% sucrose in 0.1mol/L PBS until it sank. Sagittal sections (30μm) were cut using a vibratome and sections were stored in tissue cryoprotectant solution (25% ethylene glycol, 25% glycerol, and 0.1mol/L phosphate buffer) at −20°C. These were used for cell proliferation (BrdU, P-H3) and differentiation (BrdU/NeuN, BrdU/S100ß) stainings as well as DAPI staining for counting of the total number of neuronal nuclei in the granule cell layer (GCL). The other hemisphere was dehydrated, paraffin embedded, and cut into 5μm sections. These sections were stained for cell death-related markers (apoptosis-inducing factor (AIF), terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL), active caspase-3, microtubule-associated protein 1, light chain 3 (LC3), and fodrin breakdown product (FBDP)) as described earlier (Zhu et al, 2005) and synapsin I.
BrdU staining was performed on free-floating sections. The halogenated pyrimidine was exposed through DNA denaturation. This was achieved by incubating the sections in 2mol/L HCl for 30mins at 37°C, rinsing in 0.1mol/L borate buffer (pH 8.5) for 10mins, followed by several rinses in Tris-buffered saline (TBS: 0.08mol/L Trizma-HCl, 0.016mol/L Trizma-Base, 0.15mol/L NaCl, pH 7.5). Sections were incubated in blocking solution for 30mins (3% donkey serum and 0.1% Triton X-100 in TBS) and then incubated with primary antibody (rat anti-BrdU, 1:100, 5μg/mL; clone: BU1/75, Oxford Biotechnology Ltd, Oxfordshire, UK) in blocking solution for 16h at 4°C. After rinsing with TBS, sections were incubated for 2h at room temperature with biotinylated donkey anti-rat IgG (1:1,000, Jackson ImmunoResearch Lab, West Grove, PA, USA), followed by avidin-peroxidase for 1h (ABC kit, Vectastatin Elite, Vector Laboratories, Burlingame, CA, USA), and then detection solution (26.5mg/mL diaminobenzidine, 0.01% H2O2, 0.04% NiCl). For the phospho-histone H3 staining, sections (30μm) were denatured in 10mmol/L sodium citrate solution (pH 9.0) for 30mins at 80°C and endogenous peroxidase activity was removed by 0.6% H2O2 for 30mins. After rinsing, the sections were blocked with 3% donkey serum and then incubated with rabbit anti-phospho-histone H3 polyclonal IgG (1:200, 5μg/mL, Upstate, Lake Placid, NY, USA) in blocking solution for 16h at 4°C, followed by 2h at room temperature with biotinylated donkey antirabbit IgG antibody (1:1,000, Jackson ImmunoResearch Lab).
The phenotypes of BrdU-labeled cells were determined using triple immunofluorescent staining. Antibodies against NeuN and S-100β were used to detect mature neurons and astrocytes, respectively. DNA denaturation was performed as above, followed by incubation with rat anti-BrdU (1:250, 2μg/mL; clone: BU1/75, Oxford Biotechnology Ltd) together with mouse anti-NeuN monoclonal antibody (1:200, 5μg/mL; clone: MAB377, Chemicon, Temecula, CA, USA) and rabbit anti-S-100ß (1:1,000; Swant, Bellinzona, Switzerland) in TBS at room temperature for 60mins. After washing, the sections were incubated with secondary antibodies, Alexa 488 donkey anti-rat IgG (H+L) combined with Alexa 555 donkey anti-mouse IgG (H+L) and Alexa 647 donkey anti-rabbit IgG (H+L) at room temperature for 60mins. All secondary antibodies were from Jackson ImmunoResearch Lab, diluted 1:1000. Sections were mounted using Vectashield mounting medium.
For double labeling of SOX-2 and GFAP, free-floating sections were incubated with goat anti-SOX-2 (1:200, 1μg/mL; sc-17320, Santa Cruz Biotechnology, Santa Cruz, CA, USA) and mouse anti-GFAP (1:500, clone: GA5, MAB 360, Chemicon International) in TBS containing 3% donkey serum and 0.1% Triton X-100 at 4°C overnight. After washing, sections were incubated with secondary antibodies, Cy5 donkey anti-rat combined with Alexa 555 donkey anti-mouse and Alexa 488 donkey anti-goat, 1:1000 for 2h at room temperature.
The cell death-related markers include active caspase-3, AIF, FBDP, microtubule-associated protein 1 LC3, and detection of DNA strand breaks by TUNEL. Sections (5μm) were deparaffinized and rehydrated. Antigen retrieval was performed by heating the sections in 10mmol/L boiling sodium citrate buffer (pH 6.0) for 10mins. Nonspecific binding was blocked for 30mins with 4% goat or horse serum (depending on the species used to raise the secondary antibody) in PBS. Rabbit anti-active caspase-3 (1:100, 10μg/mL, BD Pharmingen, Franklin Lakes, NI, USA), anti-FBDP (1:50), anti-LC3 (1:300, #2775, Cell Signaling, Danvers, MA, USA), goat anti-AIF (1:100, 2μg/mL, sc-9416, Santa Cruz), or anti-synapsin I a/b (1:150, 1.3μg/mL, sc-7379, Santa Cruz) were incubated for 60mins at room temperature, followed by the appropriate, biotinylated goat anti-rabbit (1:150, for active caspase-3, FBDP, and LC3) or biotinylated horse anti-goat (1:200, for AIF and synapsin I) secondary antibody for 60mins at room temperature. Endogenous peroxidase activity was blocked with 3% H2O2 for 5mins. Visualization was performed using Vectastain ABC Elite with 0.5mg/mL 3,3′-diaminobenzidine enhanced with 15mg/mL ammonium nickel sulfate, 2mg/mL β- glucose, 0.4mg/mL ammonium chloride, and 0.01mg/mL β-glucose oxidase (all from Sigma, Stockholm, Sweden). The TUNEL staining was performed according to the instructions of the manufacturer (Roche). After deparaffinization, rehydration, and antigen retrieval, sections were incubated with 3% bovine serum albumin in 0.1mol/L Tris-HCl (pH 7.5) for 30mins followed by 50μL of TUNEL reaction mixture (terminal deoxynucleotidyl transferase fluorescein-2′-deoxyuridine, 5′-triphosphate (dUTP) and deoxynucleotide triphosphate) on each sample for 60mins at 37°C. After washing, slides were incubated with 0.3% H2O2 in methanol for 10mins, followed by 3% bovine serum albumin in 0.1mol/L Tris-HCl (pH 7.5) for 30mins at room temperature. Incubation with peroxidase-conjugated anti-fluorescein (diluted 1:5) at 37°C for 30mins was followed by visualization using 3,3′-diaminobenzidine as above.
The numbers of BrdU-labeled cells and phospho-histone H3-positive cells were counted in the GCL, including the subgranular zone using unbiased stereological counting techniques. All cells were counted in every 12th section by using a semiautomatic stereology system (StereoInvestigator, MicroBrightField Inc., Magdeburg, Germany). After outlining the borders of the GCL, the area is calculated automatically. The computer will overlay a grid system with counting frames in the outlined area. Cells within the counting frames as well as cells touching two out of four predetermined sides of the counting box were counted. The cells were also counted in the z axis. The counting volume was calculated by multiplying the average of all the traced areas from each section with the section thickness. The cell density was calculated from the sum of all cells counted divided by the counting volume. To calculate the total number of cells, the density was multiplied with total volume, which was calculated as counting volume multiplied by 12, the sampling frequency of sections in this study. For the immunofluorescence stainings, at least 50 BrdU-positive cells in the dentate gyrus (DG) were phenotyped using a confocal laser scanning microscope (Leica TCS SP, Heidelberg, Germany) and the ratio of Brdu/NeuN or BrdU/S100β double-labeled cells was calculated for each sample. The total number of newborn neurons (BrdU/NeuN positive) and astrocytes (BrdU/S100β positive) in each sample was calculated based on the number of BrdU-positive cells and the ratio of double labeling. The total number of neurons in the GCL was assessed by counting the neuronal nuclei stained with DAPI, using stereological counting techniques. Double-labeled SOX-2/GFAP-positive cells were counted in the entire GCL. All counting was performed by investigators blinded to group assignment.
Quantitative analysis of the synapsin I optical density was performed by using image analysis. The images were captured using a light microscope equipped with a CCD camera (Olympus, Tokyo, Japan). Images were calibrated using a known optical density grayscale in the Image Gauge software (Fujifilm, Tokyo, Japan). Background optical densities were measured in adjacent sections stained without primary antibody and subtracted from the raw optical densities of the examined tissue. Average measurements from three sections per animal were used in the statistical analysis.
All data are expressed as mean±s.e.m. Student's unpaired t-test was used to compare the numbers of cell death-related markers, GCL volumes, total neuron numbers, BrdU-, and P-H3-positive cell numbers, numbers of newborn neurons and astrocytes, the densities of synapsin I as well as adult rat blood gas parameters between the two groups. ANOVA with Bofferroni/Dunn post hoc test was used for young rat blood gas and blood pressure data. The Mann–Whitney U-test was used for comparing the object recognition index between the two groups. For the IntelliCage analysis, generalized estimating equations were used. A Poisson model was chosen for integer values, a binary logistic model for ratios, and a normal log model for other values. Data were first analyzed for interaction effects (treatment × time), if no significance was found, the data were reanalyzed for effects of treatment and time separately. Significance was assumed when P<0.05.
The mean arterial blood pressure (MAP) was 48.6±3.6mmHg after 15mins and 46.8±3.4mmHg after 35mins isoflurane inhalation in the P14 rats (Figure 1A). Repeated isoflurane exposure appeared to reduce MAP slightly when compared with single exposure, but not significantly. MAP remained stable during the 35mins of isoflurane exposure, both during the first and the fourth exposure (Figure 1B). The body temperature was stable during anesthesia in the P14 pups (Figure 1C). Blood gas analysis showed that isoflurane inhalation induced a slight respiratory acidosis and hyperglycemia, and that this was not different between the two age groups (Figure 1D).
Rat recognition memory was impaired 4 weeks after isoflurane inhalation starting at P14, as indicated by the recognition index, which was significantly lower (26.3%±10.1%) in the P14 to P45 isoflurane group compared with the control group (56.4%±9.4%) (P=0.022) and it was further aggravated after 10 weeks (P14 to P90) (5.2%±2.8% versus 42.1%±4.7%, P=0.008). There was no significant effect on recognition index after isoflurane inhalation starting at P60, neither after 4 weeks (P60 to P91; 42.9%±5.2% versus 48.9%±9.4%, P=0.631) nor after 10 weeks (P60 to P120; 32.5%±7.3% versus 47.8%±5.7%, P=0.096) (Figures 2A and 2B).
In the IntelliCage test, the mice displayed an incorrect visit ratio around 70% (random corner visits yield 75%) during the learning phase and both groups showed improved learning over time (P<0.001), with an incorrect visit ratio around 40% on day 4, but with no difference between the isoflurane-treated and the control mice (Figure 3B). On the first day of the reversal learning phase, the incorrect visit ratio was over 85% for both groups, as expected when the mice keep trying to drink in the previously assigned correct corner (Figure 3B). However, though the control animals quickly learned to find their new assigned corner (P<0.01) the isoflurane-treated animals stabilized around 75%, indicating a dramatically impaired reversal learning (Figure 3B). There were no significant differences in the total number of visits (data not shown). The numbers of correct and incorrect visits are shown in Figures 3C and 3D, respectively. Both groups showed an increasing number of correct visits during the learning phase, but during the reversal learning phase the isoflurane-treated mice performed an average of 49% fewer correct visits (P<0.01) compared with control animals (Figure 3C). During the 4 days of both learning and reversal learning, the average increase of correct visits was 41% (P<0.001) and 40% (P<0.001), respectively, indicating that all mice learn where to find water, but that the isoflurane-treated mice have difficulties during reversal learning (P<0.01) (Figure 3C). The number of incorrect visits revealed significant interaction between time and treatment during the learning phase, and both groups appear to make fewer incorrect visits from day 4 of the learning phase. During the reversal learning phase, there was no difference in the number of incorrect visits between the groups (Figure 3D). Both control animals and isoflurane-treated animals showed an increase in the amount of time spent per visit in the correct corners (Figure 3E) during the learning phase (63%, P<0.001) and reversal learning phases (37%, P<0.001). In the reversal learning phase, isoflurane-treated animals spent an average of 58% (P<0.01) less time in the correct corner per visit compared with controls (Figure 3E). No differences were seen in average time spent in incorrect corners (Figure 3F). Control animals and isoflurane-treated animals showed an increase in the number of correct nose pokes during the learning (101%, P<0.001) and reversal learning (56%, P<0.001) phases (Figure 3G). Treated animals performed an average of 73% (P<0.01) fewer correct nose pokes during the reversal learning phase (Figure 3G). During the learning phase, there was an average decrease in the number of incorrect nose pokes (15%, P<0.05, Figure 3H). Both groups showed an increase in the average number of nose pokes per visit in correct corners (Figure 3I) during the learning phase (82%, P<0.001) and reversal learning phase (52%, P<0.01), but in the reversal learning phase, isoflurane-treated animals performed 71% (P<0.01) fewer nose pokes when entering a correct corner (Figure 3I).
The total number of neurons (Figures 4A and 4B) and volume of the GCL (Figure 4C) increased with brain development, but there were no significant differences 4 weeks after isoflurane exposure compared with the controls (Figures 4A and 4C). However, if waiting until P90 or P120, respectively, the total number of neurons was significantly lower in the GCL if isoflurane exposure occurred at P14 (2.79±0.14 × 105 versus 3.21±0.09 × 105, P=0.031) but not at P60 (Figure 4B). Neither the GCL volume nor the density of neurons in the GCL was significantly different between isoflurane and control groups in any age group (Figure 4C). Hence, isoflurane-induced memory impairment was aggravated as the rats grew older, and correlated with lower neuronal numbers in the GCL.
Cells positive for markers of degeneration, including necrosis (calpain activation, fodrin breakdown product (FBDP)), DNA strand breaks (TUNEL), autophagy (LC3) as well as caspase-dependent (active caspase-3) and -independent (AIF) apoptosis (Zhu et al, 2005), were counted in the DG and CA1 24h after the last isoflurane inhalation (Figure 5A) in both P14 and P60 rats. The number of cells positive for the markers listed was not different between isoflurane-treated and control animals, neither in the DG nor in the CA1 of the P14 group (Figure 5B) or the P60 group (data not shown). Furthermore, the frequency of degenerating newborn cells, as judged by the number of cells positive for both TUNEL and BrdU, was not increased by isoflurane in the GCL of P14 rats (Figures 5C and 5D). This indicates that repeated isoflurane exposure had no obvious effect on cell death in the hippocampus, neither on newly formed neuronal progenitors nor on existing neurons.
To investigate if there was a relationship between synapse density and learning, synapsin I immunostaining was performed in the P14 to P90 rats (Figure 6A). Synapsin I was expressed throughout the hippocampus with high density in the mossy fiber layer, the hilus of the DG, and the stratum lucidum of the CA3 region (Figure 6B). The density of synapsin I did not change significantly after isoflurane exposure, neither in the DG nor in the CA3 region (Figure 6C).
To evaluate the effects of isoflurane on cell proliferation and differentiation in the hippocampus, P14 and P60 rats received BrdU injections once daily for 4 consecutive days (Figure 7A). Proliferation was evaluated 24h or 4 weeks after the last isoflurane inhalation using BrdU or phospho-histone H3 (P-H3) immunostaining. Proliferating cells were detected in clusters in the subgranular zone of the DG (Figure 7B). The number of BrdU-labeled cells in P14 control rats was 13 times higher than those of P60 control rats (56,607±2,400/mm3 versus 4,294±520/mm3, P<0.0001), as expected. Isoflurane inhalation reduced proliferation immediately, evaluated 24h after the last inhalation, in P14 (21% reduction, P=0.016), but not in P60 rats (Figure 7B). Surprisingly, the inhibition of proliferation persisted and was even more pronounced 4 weeks later as detected by P-H3 staining. The number of P-H3-positive cells was 71% lower in the isoflurane-treated P14 to P45 rats (1,876±165/mm3 versus 542±58/mm3, P<0.0001), but unchanged in P60 to P91 rats (Figure 7B). Newly generated, BrdU-labeled cells were phenotyped 4 weeks after exposure. The number of BrdU-labeled cells in P14 to P45 rats was 39% lower in isoflurane-treated rats (36,003±2,613/mm3 versus 20,470±3,349/mm3 P=0.0181), but unaffected in P60 to P91 rats (Figure 7D). More than 90% of the newborn cells in the GCL developed into neurons in controls. Isoflurane reduced neuronal production, as judged by BrdU/NeuN colabeling, significantly in both the P14 to P45 (34,116±2,849/mm3 versus 17,846±2,806/mm3 P=0.0097) and P60 to P91 groups (3,340±194/mm3 versus 2,465±183/mm3 P=0.0384) (Figure 7D), most pronounced in the P14 to P45 rats (43% reduction). Conversely, the numbers of newborn astrocytes increased significantly in both the P14 to P45 (1,071±115/mm3 versus 1,968±441/mm3 P=0.023) and P60 to P91 groups after isoflurane exposure (183±32/mm3 versus 475±94/mm3 P=0.0104) (Figure 7D).
The radial glia-like stem cells were assessed by double labeling of SOX-2 and glial fibrillary acidic protein (GFAP) Naylor et al, 2008; Steiner et al, 2006). A significant decrease in SOX-2/GFAP-positive cells was observed in isoflurane-exposed young animals compared with controls (3,966±434 versus 7,007±327, P=0.0002). In contrast, there was no difference in double-positive cells in the adult brains exposed to isoflurane (Figure 8).
Our initial hypothesis was that repeated isoflurane exposure during brain development would cause widespread neurodegeneration with subsequent memory and learning disabilities, more pronounced in the immature brain. This hypothesis was based on recent work showing that isoflurane exposure during early postnatal development resulted in increased apoptosis and subsequent behavioral impairment (Jevtovic-Todorovic et al, 2003) and a recent study showing an association between the cumulative duration of anesthesia exposure and later development of learning disabilities in children younger than 4 years of age (Wilder et al, 2009). To our surprise, repeated isoflurane exposure did induce long-lasting, even progressive, impairment of recognition and reversal learning, but had no effect on cell death of neurons or neuronal progenitors in the young, still growing brain.
Isoflurane exposure has been associated with a persistent deficit in the acquisition and performance of spatial memory tasks in both young and aged rats, indicating that this was not an age-specific phenomenon (Culley et al, 2004). One study showed that immature brains, especially during intense synaptogenesis, are more sensitive to isoflurane toxicity, as judged by persistent memory deficits (Jevtovic-Todorovic et al, 2003). Our data failed to find major differences in synapsin immunoreactivity, but could not rule out functional synaptic deficits. The age-related long-lasting effect of isoflurane on memory function is matter of debate (Culley et al, 2004; Li et al, 2007). Our study showed clearly age-related recognition memory impairment. The relative contributions of neuronal cell death and neurogenesis to recognition memory are largely unknown. Significant neuronal loss and/or perturbation of synaptic proteins during brain development are thought to be, at least in part, responsible for impaired cognitive development (Fournier et al, 2008). Several studies have correlated alterations in neuronal cell death (Jevtovic-Todorovic et al, 2003) and neurogenesis (Denis-Donini et al, 2008; Kee et al, 2007) with cognitive performance. In a recent study, P7 and P60 rats were anesthetized for 4h (one session) and showed moderately decreased progenitor proliferation after anesthesia in P7 rats and a delayed-onset, progressive deficit in fear conditioning and spatial reference memory tasks. In P60 rats, however, they found increased early neuronal differentiation and subsequently increased progenitor proliferation 5 to 10 days after anesthesia, followed by improved spatial reference memory (Stratmann et al, 2009b). Our data confirm their findings that the young brain is more susceptible to isoflurane than the adult brain and that functional deficits worsen with time, but whereas they found widespread cell death caused by 2 and 4h of isoflurane (but not 1h) (Stratmann et al, 2009a) and moderately decreased (P7) or even increased (P60) proliferation, we found no cell death and dramatic, persistent reductions of proliferation in the young (P14) but not in the adult rats. Possible reasons for these discrepancies include different ages used (P14 versus P7), different doses of isoflurane (1.7% versus 3.5% to 4%), number of sessions (35mins daily for 4 days versus 4h in one session), different learning tasks (object recognition and IntelliCage versus fear conditioning and Morris water maze), different BrdU labeling paradigms, and perhaps different rat strains used (we used Wistar rats but the strain is not indicated in the Stratmann paper).
Isoflurane has neuroprotective potential by reducing apoptosis in in vitro hypoxia/ischemia and animal stroke models (Popovic et al, 2000; Sakai et al, 2007), presumably attributed to enhancement of GABA neurotransmission. However, another study showed that isoflurane just postponed cell death in a model of ischemia (Kawaguchi et al, 2004). Other studies have reported that isoflurane exposure for 6h triggers widespread neuro-apoptotic degeneration in P7 rats dose dependently (Jevtovic-Todorovic et al, 2003). One study found that exposure to isoflurane for more than 1h led to respiratory and metabolic derangements in neonatal mice (Loepke et al, 2006, 2009; Stratmann et al, 2009a). The widespread neuronal injury after anesthesia exposure could be partly attributed to hypercarbia. To avoid anesthesia-related effects on blood gasses, we used shorter exposures, 35mins per session, which presumably is equivalent to a much longer time span in a human setting, taking differences in the rates of brain development and life span into account. It would be interesting to compare the effects of four daily 35min exposures with a single 140min exposure in our paradigm. In a recent report, isoflurane exposure had no significant effect on cell death (Sall et al, 2009), neither in mature nor in newborn neurons, neither in young nor in adult brains, in accordance with our findings. These differences in detection of cell death might be attributable to the use of different cell lines or differences in the duration and concentration of isoflurane, different time points for evaluation of cell death after isoflurane exposure, as well as variations in the physiologic or pathologic conditions of animals during isoflurane exposure (Kawaguchi et al, 2004; Wei et al, 2008; Xie et al, 2006). Isoflurane is used clinically at a wide range of concentrations (about 0.2% to 3%), depending on the combination with other kinds of anesthetics and the duration of surgery. It has been reported that the neurotoxicity of isoflurane is concentration- and duration-dependent (Jevtovic-Todorovic et al, 2003; Stratmann et al, 2009a). Loepke et al (2006) found that when infant mice were exposed to isoflurane (3% for 30mins followed by 1.8% for 1h) without assisted ventilation, the MAP remained stable and blood gas values did not vary markedly from unanesthetized controls. Johnson et al (2008) found that 2% isoflurane for 1h had no effect on blood glucose in immature mice and the minimum alveolar concentration of isoflurane is 1.6% for human neonates (LeDez and Lerman 1987). In this study, we found that neither single nor repeated (up to 4) isoflurane exposures for 35mins, without assisted ventilation, had any apparent effect on the MAP. A slight respiratory acidosis and hyperglycemia were noticed in the anesthetized rats, but it was not different between the two age groups. There was a tendency toward hypercarbia in both age groups, which probably explains the acidosis, but it was not statistically significant. Therefore, respiratory or metabolic derangements cannot explain the great differences between the young and adult animals observed for neural stem cell numbers and neurogenesis after isoflurane exposure.
This is the first study to our knowledge showing a reduction in undifferentiated neural stem cells after treatment with anesthetic agents, a reduction not related to cell death. Environmental stimuli may either enhance (Kempermann et al, 1997) or suppress (Gould et al, 1998) stem/progenitor cell proliferation, depending on the relative predominance of enriching or noxious qualities. Anesthetics clearly alter the environmental stimuli, including neuronal signaling, and apparently have age-dependent effects on cell proliferation in the DG (Stratmann et al, 2009b). In the absence of increased cell death, we explored the possibility of a reduced cell cycle rate or cell cycle exit, explaining the neural stem cell reduction. The frequency of BrdU-labeled cells in the subgranular zone colabeled with Ki67 or P-H3 4 days after the first BrdU labeling, indicating continuous proliferation of precursor cells, was, however, not different between controls and isoflurane-treated mice (data not shown). It remains to be shown if changes in the neurogenic niche can explain the reduction in the stem cell pool and the persistently reduced neurogenesis in the young, but not the adult, rodent hippocampus.
Evidence indicates that new neurons produced in the hippocampus have a functional role in cognitive processes such as learning and memory (Aimone et al, 2006; Kempermann et al, 1997; Shors et al, 2001). In our study, a stronger reduction in neurogenesis was correlated with more pronounced memory impairment. The DG develops in three distinctive phases and the peak of DG neuron expansion occurs around P7 (Altman and Bayer 1990). The GCL is still growing at P14 (Fukuda et al, 2005) but the quality and quantity of stimuli passing through the DG will be very different at P7 (when the eyes are still closed) and P14 (when the pups move around freely), and the speed of maturation of progenitor cells and young neurons might differ (Overstreet-Wadiche et al, 2006). This might, at least partly, explain why we find more profound learning impairment after isoflurane exposure of P14 rats than that demonstrated for P7 rats (Stratmann et al, 2009b). The IntelliCage test provided us with important information regarding the nature of the learning deficits observed. Over the first 4 days, there was no difference between the control and isoflurane-treated mice in their ability to learn to drink from the assigned corner, but when they were assigned a new corner (reversal learning), only the control mice managed to learn this task, whereas the isoflurane-treated mice displayed an incorrect visit ratio around 75% after 4 days of testing, indicating that corners were visited randomly, without any learning taking place. In addition, when these animals did find the correct corner, they spent less time there and performed fewer nose pokes compared with control animals, further indicating difficulties in ‘delearning' the old task and learning the new one. These data suggest a difference in the animals' ability to replace one declarative memory (learning) with a new, modified one (reversal learning). This is analogous to the Morris Water Maze, which entails both a procedural part (swimming to find the platform) and a declarative part (finding the correct quadrant) and where reversal learning involves changing the declarative memory to a new one (when the position of the platform is changed) while maintaining the procedural memory (Rossato et al, 2006). In the IntelliCage, both control and isoflurane-treated mice continuously learn to perform correct visits and improve over time, both in the learning and reversal learning phases (Figure 3C), indicating no impairment of the procedural memory. Declarative memory, including spatial memory, is linked to the hippocampus (Manns and Eichenbaum 2006), but we cannot exclude that other parts (e.g., entorhinal cortex, amygdala, and striatum) also take part in this specific type of learning task.
In summary, our findings indicate that if the developing brain is exposed to the common anesthetic isoflurane, a persistent decrease in the hippocampal neural stem cell pool and neurogenesis will ensue, and that this correlated with impaired memory function. The adult brain, however, where neurogenesis is more limited, displayed only moderate changes and no significant memory impairment. This is in accordance with the hypothesis that postnatal hippocampal neurogenesis is important for memory and learning (Aimone et al, 2006). Furthermore, the findings strongly indicate that anesthetic agents like isoflurane should be carefully evaluated and probably used with caution when anesthetizing pediatric patients.
This work was supported by the Swedish Research Council, the Swedish Childhood Cancer Foundation, the Torsten and Ragnar Söderberg Foundation, the King Gustav V Jubilee Clinic Research Foundation (JK-fonden), The Frimurare Barnhus Foundation, Swedish governmental grants to scientists working in health care (ALF), the Wilhelm and Martina Lundgren Foundation, Edit Jacobssons Foundation, the Swedish Brain Foundation (Hjärnfonden), the Gothenburg Medical Society, the Swedish Research Links Program through the Swedish International Development Cooperation Agency (SIDA), the Ministry of Education (211 project), and the National Nature Science Foundation of China (30870883 to CZ). We thank the Center for Mouse Physiology and Bio-Imaging, University of Gothenburg, for help with blood pressure measurements.
The authors declare no conflict of interest.