Division of Pediatric Neurology, Department of Pediatrics, Duke University Medical Center, Durham, North Carolina, U.S.A
Department of Neurobiology, Duke University Medical Center, Durham, North Carolina, U.S.A
Address correspondence to Mohamad A. Mikati, Division of Pediatric Neurology, T0913 Children's Health Center, Duke University Medical Center, Box 3936, 2301 Erwin Road, Durham, NC 27710, U.S.A. E-mail: firstname.lastname@example.org
There is a gap in our knowledge of the factors that modulate the predisposition to seizures following perinatal hypoxia. Herein, we investigate in a mouse model the effects of two distinct factors: developmental stage after the occurrence of the perinatal insult, and the presence of a seizure predisposing mutation.
Effects of age: P6 (postnatal day 6) mouse pups were subjected to acute hypoxia down to 4% O2 over the course of 45 min. Seizure susceptibilities to flurothyl-induced seizures (single exposures) and to flurothyl kindling were determined at specific subsequent ages. Effects of mutation: Heterozygote mice, with deletion of one copy of the Kcn1a gene, subjected to P6 hypoxia were compared as adults to wild-type mice with respect to susceptibility to a single exposure to flurothyl and to the occurrence of spontaneous seizures as detected by hippocampal electroencephalography (EEG) and video recordings.
Effects of age: As compared to controls, wild-type mice exposed to P6 hypoxia had a shortened seizure latency in response to a single flurothyl exposure at P50, but not at P7 or P28 (p < 0.04). In addition, perinatal hypoxia at P6 enhanced the rate of development of flurothyl kindling performed at P28–38 (p < 0.03), but not at P7–17. Effects of mutation: Kcn1a heterozygous mice subjected to P6 hypoxia exhibited increased susceptibility to flurothyl-induced seizures at P50 as compared to Normoxia heterozygote littermates, and to wild-type Hypoxia and Normoxia mice. In addition, heterozygotes exposed to P6 hypoxia were the only group in which spontaneous seizures were detected during the period of long-term monitoring (p < 0.027 in all comparisons).
Our data establish a mouse model of mild perinatal hypoxia in which we document the following: (1) the emergence, after a latent period, of increased susceptibility to flurothyl-induced seizures, and to flurothyl induced kindling; and (2) an additive effect of a gene mutation to the seizure predisposing consequences of perinatal hypoxia, thereby demonstrating that a modifier (or susceptibility) gene can exacerbate the long-term consequences of hypoxic injury.
Neonatal hypoxia increases the risk for onset of epilepsy later in life (Glass et al., 2011). Despite important insights gained from rat models (Koh & Jensen, 2001; Rakhade et al., 2011; Zhou et al., 2011), mechanisms that can modify the long-term consequences of hypoxia are still not fully understood (Rakhade & Jensen, 2009). Additional insights are needed to understand why in the human condition some, but not all, individuals with neonatal hypoxia, even those with seemingly similar severity of insults, develop later life epilepsy (Allemand et al., 2009; Pisani et al., 2009), and to understand the factors that influence this process (Bergamasco et al., 1984; Rakhade et al., 2011).
The possibility that posthypoxia epileptogenesis is influenced by subsequent age and or by genetic background is an often-implicit assumption while dealing with the human condition. In support of this notion are the clinical data reporting that the patient's family history contributes to outcome after perinatal brain injury (Ottman et al., 1996; Berkovic et al., 2006), and that genetic factors, as evidenced by positive family history of epilepsy and of consanguinity, contribute to the development of symptomatic epilepsies (Choueiri et al., 2001). In addition, emerging data from genome-wide association studies suggest that mutations involving many genes including ion channel genes act as predisposing factors to various types of epilepsy, suggesting an interaction between such genes and environmental factors (Kasperaviciūte et al., 2010; Klassen et al., 2011; EPICURE Consortium et al., 2012).
The evidence above notwithstanding, an animal model demonstrating that an epilepsy predisposing gene may modify the long-term consequences of hypoxia has, to our knowledge, not been developed. This is particularly important, since combining two epilepsy predisposing factors, such as two epilepsy genes, may not necessarily result in additive effects, but rather, in some instances may even have a protective masking effect (Glasscock et al., 2007). Therefore, we reasoned that a mouse model with a gene mutation that predisposes to seizures but that, per se, does not cause epilepsy in the heterozygous animal will allow us to address the above questions, and could have the advantage of permitting subsequent investigations using the tools of mouse genetics. Kv1.1 voltage-gated potassium channels are expressed broadly in the brain, are concentrated along axons and axon terminals, and modulate action potential generation (Robbins & Tempel, 2012). We chose the Kcn1a gene because (1) it predisposes to seizures in humans (Zuberi et al., 1999); (2) its heterozygous mutations predispose to flurothyl-induced seizures in mice, but such mice do not have spontaneous recurrent seizures (Smart et al., 1998); and (3) Kv1.1 channels have a role in the processes that follow hypoxia in the developing brain (Deng et al., 2011).
The purpose of this study was, thus, to develop a mouse model of perinatal hypoxia with a long-term mild predisposition to seizures in order to investigate the potential added effects of developmental stage and of an epilepsy predisposing gene mutation on seizure susceptibility.
Reagents and animals
All chemical reagents were purchased from Sigma (St. Louis, MO, U.S.A.) and were of analytical grade or higher. C57BL/6 mice were obtained from Charles River Laboratories (Durham, NC, U.S.A.) and the Kcn1a mutant mice from the Jackson Laboratories (Bar Harbor, ME, U.S.A.). The initial heterozygous chimeric mice were mated with the C57BL/6J wild-type. Tail snippings were obtained before weaning and were genotyped to establish either heterozygosity or wild-type genetic background. DNA preparations from the tails of all offspring were analyzed by Southern blotting (Smart et al., 1998). Animals were maintained under a 12 h light-to-dark cycle with standard rodent chow provided ad libitum. All procedures were approved by the Duke University Institutional Animal Care and Use Committee. All litters in the Kcn1a experiments were genotyped. Genotyping of Kcn1a mice was performed with tail tissue, which was digested in a proteinase K buffer followed by column purification of genomic DNA (as described by Qiagen DNeasy Tissue kit, Qiagen, Valencia, CA, U.S.A.). An anchored three-primer polymerase chain reaction (PCR) technique was used to determine genotype. A common reverse primer 5′ GCT TCA GGT TCG CCA CTC CCC 3′ along with a Kcn1a-specific forward primer 5′ GCC TCT GAC AGT GAC CTC AGC 3′ and the Kcn1a knock-out-specific forward primer 5′ CCT TCT ATC GCC TTC TTG ACG 3′ were used as described by Smart et al. (1998).
Acute hypoxia procedure
Groups of mice were placed in the hypoxic chamber (approximately 50 cm × 25 cm × 20 cm) enclosed by an acrylic glass airtight lid with three evenly distributed small air outlets and a fourth reserved for the wire connected to the oximeter probe. Hypoxia was achieved through a regulated decrease of O2 and a simultaneous increase in N2 flow into the chamber. The air mixture was regulated according to twin digital gas flow meters (Aalborg Inc., Orangeburg, NY, U.S.A.), which maintain a minimum overall flow rate of at least 3 L/min and were used to decrease O2 levels in a two-stage manner. At the beginning of each experiment, mice were placed in the chamber with ambient O2 levels (approximately 21%) and the flow rate was initiated (time = 0) such that the first set point was to 10% O2 (i.e., 2.7 L/min N2 and 0.3 L/min O2). Once O2 levels reached 11%, the flow meters were adjusted to the next set point of 4% O2 (i.e., 2.88 L/min N2 and 0.12 L/min O2). After O2 levels reached 4%, the flow meters were again adjusted in a two-stage manner so as to return O2 levels to 10%, and then 21%. The entire treatment took approximately 45 min (descent and ascent were approximately 22 min each, and 4% O2 lasted for 1 min). To determine if there was a developmental effect of the Kcn1a mutation on growth, we sequentially measured weight during development in heterozygous mice and littermate wild-type controls. No statistical differences were found in weight between the two groups at any of the following postnatal dates: P2–13, p = 0.815; P6, p = 0.798; P > 21 weeks, p = 0.672 (Fig. S1).
Behavioral scoring of seizures during hypoxia
Our initial experiments were aimed at identifying the perinatal age in C57BL/6 mice of increased susceptibility to develop behavioral seizures during acute hypoxia. Mice were studied at P2 (n = 10), P4 (n = 12), P6 (n = 13), P8 (n = 10), P10 (n = 11), P13 (n = 11), and as adults (n = 14). Behavioral changes induced by hypoxia were scored according to a modified Morrison scale (Morrison et al., 1996). This scale, which was developed specifically for mice, has been successfully used in scoring ischemia-carotid ligation-induced seizures in immature mice (Comi et al., 2004). The scale consists of the following: stage 0, normal behavior; stage 1a, immobility; stage 1b, immobility with myoclonic jerks (MJs); stage 2, rigid posture; stage 3, repetitive swimming and pedaling movements, circling, or head bobbing; stage 4, forelimb clonus, loss of posture; stage 5, exhibit stage 4 repeatedly. Animals were videotaped during the experiments, and the videos were subsequently reviewed and scored in a blinded manner.
Electrode implantation and EEG recordings
To determine whether hypoxia-induced seizure behaviors were associated with electrographic seizures, mice of different age groups (P2–4, n = 10; P6, n = 12; P8–10, n = 18; P12–14, n = 10) were recorded with hippocampal depth electrodes during the acute hypoxia procedure. Animals were implanted bilaterally with monopolar steel electrodes under isoflurane anesthesia with stereotactic guidance. Electrodes were implanted 0.5 mm below the dura in the ventral hippocampus at the following coordinates: 1.0, 1.5, 2.0, 2.5, and 3 mm lateral to the midline and 1.0, 1.5, 2.0, 2.5, and 3 mm rostral to lambda in P2–4, P6, P8–10, P12–14, and adult mice, respectively. Ground wires were implanted subcutaneously and held in place next to the electrode wires via a cement head cap. Electrocardiography was recorded via self-adhesive surface electrodes (Rhythmlink International, Columbia, SC, U.S.A.). Animals were allowed to recover on a heating pad for a minimum of 2 h. Such a period of time has been shown to be sufficient to allow animals to recover from isoflurane anesthesia, thus avoiding affecting seizure susceptibility (Eger & Johnson, 1987; Joksovic et al., 2009). Subsequently, animals were gently placed in the hypoxia chamber with supplemental heat provided.
Experiment I: Effect of age on flurothyl seizure latency
To investigate the influence of age, we studied three groups of P6 hypoxia mice that were tested, as compared to three parallel Normoxia control groups, for seizure threshold: the first at P7 (n = 13), the second at P28 (n = 8), and the third at P50 (n = 14; Fig. 1A). We compared flurothyl seizure latency to control littermates (n = 13, 10, and 14, respectively). The procedure was based on that described by Samoriski and Applegate (1997). Mice were placed individually in a 2.4 L closed acrylic glass chamber. Generalized seizures were elicited using a 10% solution of flurothyl (2,2,2-trifluoroethyl ether; Sigma Aldrich Chemical, St. Louis, MO, U.S.A.) in 95% ethanol. Flurothyl was delivered by infusion (0.15 ml/min) using a 5 ml B-D Multifit syringe driven by a Harvard apparatus infusion pump. Flurothyl was delivered onto a Whatman filter pad (Millipore, Billerica, MA, U.S.A.) that was suspended at the top of the chamber and was changed after each exposure. Infusion was terminated and the chamber was opened to room air with the onset of sustained generalized seizure activity. Onset of generalized seizures was defined as a sustained loss of postural control (>2 s). Seizure tests were conducted at the same time of day to minimize circadian influences on flurothyl-induced seizure thresholds (Davis & Webb, 1963). Latencies (in seconds) from the start of infusion to the onset of each generalized seizure were recorded and compared between the Hypoxia groups and Normoxia controls. Duration of these seizures in these and in the kindling experiments varied between 10 and 40 s.
Experiment II: Effect of age on flurothyl kindling
To investigate the effects of age on flurothyl kindling we investigated two groups of P6 hypoxia mice, one starting kindling at P7 (n = 13 for Hypoxia group and n = 13 for Normoxia group) and the other at P28 (n = 10 for Hypoxia group and n = 8 for Normoxia group), and compared them to two groups of Normoxia controls (Fig. 1B). The kindling sequence followed the above procedures and the protocol of Samoriski and Applegate (1997) with eight daily, once per day at 8–10 a.m. , flurothyl-induced seizure exposures. We also compared flurothyl seizure thresholds of the above groups to parallel hypoxia and Normoxia groups (n = 14 for both groups) that were sham kindled starting at the same ages of P7 (one hypoxia and one Normoxia groups), and at P28 (one hypoxia and one Normoxia group), to determine whether any changes in seizure threshold that may be observed at the end of the kindling process were due to the effect of age or due to the process of kindling.
Experiment III: Effect of gene mutation on flurothyl seizure latency
To investigate the influence of a Kcn1a gene mutation we tested four groups of mice (Wild Type-Normoxia n = 12, Wild Type-Hypoxia n = 17, Heterozygous-Normoxia n = 14, Heterozygous-Hypoxic n = 16) at P50 for flurothyl seizure latency using the above-described procedures (Fig. 1C).
Experiment IV: Effect of gene mutation on occurrence of spontaneous seizures
To investigate the influence of a Kcn1a gene mutation on the occurrence of spontaneous seizures, we studied 26 mice as of the age of 42 days (litter mate matched controls; Fig. 1D) with video–electroencephalography (EEG) monitoring (Heterozygous-Hypoxia, seven animals; Wild Type-Hypoxia, seven; Heterozygous-Normoxia, five; Wild Type-Normoxia, seven) that was reviewed for spontaneous seizures (50 h of recording time per animal, with the overall goal of approximately 12 h recording per day for a period of 4 days). This occasionally was modified owing to mice self-disconnection from the recording equipment). These animals were implanted with monopolar electrodes into the hippocampus bilaterally and allowed 3 days to recover before recording was begun (3 mm below the dura, 3 mm lateral of the midline, and 3 mm rostral to lambda). Mice were then allowed to recover for 3 days before being monitored over the subsequent weeks with concurrent video and EEG recordings. To ensure objectivity, EEG recordings were reviewed for occurrence of electrographic seizures and for occurrence of stage IV or higher behavioral seizure (Morrison et al., 1996) by an individual unaware of prior treatment. Only electrographic seizures associated with behavioral seizures of stage IV or higher were counted. Videos were reviewed for the occurrence of behavioral seizures with the reviewer blinded to the genetic or treatment/no treatment status of the animal.
Groups were compared using the chi-square test (occurrence of seizures), nonpaired t-test (seizure latency), analysis of variance (ANOVA) for two groups with repeated measures (kindling data), and ANOVA for multiple groups (seizure latency) unless the data did not fit the normal distribution, in which case the Mann-Whitney and Kruskal-Wallis tests were used.
Behavioral seizures in the P6 mouse and comparison to other ages
Staging of behavioral seizures revealed that P6 was a particularly vulnerable age. Neonatal mice aged P2–8 were more likely to achieve stage IV seizures than older mice (Table 1, p < 0.0001 for trend and for comparison of P6 with adult animals). In addition, they exhibited a lower mortality rate (Table 1). Among the different age groups tested, stage IV behaviors were seen in nearly all (12/13) of the P6 mice. During the hypoxia procedure, P6 mice typically exhibited a loss of posture below 10% O2, which remained the case except when it was intermittently interrupted by episodes of forelimb clonus. All animals regained sternal recumbency at the end of the hypoxia treatment.
Table 1. Number of mice achieving stage IV seizures and mortality during P6 hypoxia
Wild type C57BL/6 mice (Charles River)
Chi-square for trend p < 0.0001. Chi-square comparing P6 and Adult p < 0.0001. Animals represented in this table were all nonimplanted animals.
Achieved stage IV
Electrographic seizures in the P6 mouse and comparison to other ages
Baseline EEG recordings showed the expected low amplitude intermixed activity in all age groups (Fig. 2). During the initial phase of hypoxia (20–10% O2), there was suppression of the background EEG activity in all animals in each of the age groups. As the O2 level approached 4%, hippocampal EEG seizures were observed in a subset of P6 mice (three of 12 pups recorded). These electrographic seizures were characterized by rhythmic sharp activity in a waxing and waning pattern that usually lasted for approximately 60 s per animal (Fig. 2). Electrographic seizure activity was detected in the hippocampus only at P6, thereby supporting that the C57BL/6 mouse was sensitive at the age of P6 to this mild hypoxia paradigm.
Experiment I (effect of age): P6 hypoxia leads to subsequent reduced flurothyl seizure threshold in P50 but not in P7 or P28 mice
P6 hypoxia did not affect the response of the younger animals (P7 and P28) to a single exposure to flurothyl (p > 0.05), but did result in a significant decrease in flurothyl latency in P50 animals (Fig. 3, p < 0.05). The means in seconds and standard deviations (SDs) for each the groups were as follows: P7 Normoxia 660 ± 83, P7 Hypoxia 648 ± 84, P28 Normoxia 336 ± 33, P28 Hypoxia 340 ± 31, adult P50 Normoxia 422 ± 52, Hypoxia 369 ± 61. Given the number of mice, mean and SD of each group, the power, 1−β, to detect a similar decrease of 12.53% was 0.795 in P7 mice and 0.823 in P28 mice. These findings indicate that there is a delay in the effects of perinatal hypoxia on flurothyl-induced seizure susceptibility.
Experiment II (effect of age): Perinatal P6 hypoxia enhances flurothyl kindling in P28–38 mice but not in P7–17 mice
During kindling at P7–17 there was no difference between the hypoxia and Normoxia groups, indicating that prior P6 hypoxia did not predispose to kindling during this period (Fig. 4A, p > 0.05). There was a day effect with a progressively decreasing threshold indicating either that kindling is occurring in both groups equally or that there is an age effect (Fig. 4A, p < 0.0001). Upon further analysis this was considered to be due to an age effect because when the above kindled P17 mice (with or without preceding P6 hypoxia) were compared on day P17 to sham-kindled P17 mice (with or without preceding P6 hypoxia), there were no differences in latency to flurothyl-induced generalized seizures in any of the comparisons (Fig. 4B, p > 0.05). This age effect is consistent with the prior observations of Sperber and Moshé (1988).
When flurothyl kindling was begun on P28, the P6 Hypoxia group kindled faster than the Normoxia group (Fig. 5A, p < 0.03). In addition, there was a day effect (Fig. 5A, p < 0.0001), which was attributable to kindling, since comparison of the above two kindled groups at P38 with the sham-kindled groups at the same age of P38 showed that the kindled groups had shorter latencies than the sham-kindled ones (Fig. 5B, p < 0.01).
Experiment III (effect of mutation): After P6 hypoxia, the presence of a Kcn1a heterozygous mutation increases the predisposition to flurothyl-induced seizures at P50
We recorded at P6 the acute behavioral seizure manifestations of wild-type and heterozygous animals during exposure to hypoxia: when heterozygous (n = 16) and wild-type pups (n = 13) were observed during P6 hypoxia, the latency to and duration of stage 4 seizures were not different between the two groups: Latency, 1,043 ± 187 and 1,001 ± 134 s, respectively, p = 0.85; Duration, 132 ± 26.6 and 101 ± 29.9 s, respectively, p = 0.46; Fig. 6). Power analysis demonstrated the ability to discern a 20% difference between groups with an alpha error of 0.05 and power (1−β) >0.80.
At P50 (Fig. 6) the Heterozygous-Hypoxia group had a significantly shorter latency to flurothyl-induced seizures compared to all other groups: the Wild Type-Normoxia group (24.8% difference, p < 0.01, mean 595, SD ± 98), to the Heterozygous-Normoxia group (12.3% difference, p < 0.05, mean 532, SD ± 92), and to the Wild Type-Hypoxia group (12.2% difference, p < 0.05, mean 534, SD ± 91). In addition, both the Wild Type-Hypoxia and the Heterozygous-Normoxia groups had shorter latency to flurothyl seizures than the Wild Type-Normoxia group (p < 0.05, in both comparisons), but were not different from each other (p > 0.05).
Experiment IV (effect of mutation): After P6 hypoxia the presence of Kcn1a heterozygous mutation predisposes to the development of spontaneous seizures
Three of the seven animals in the Heterozygous-Hypoxia group had stage IV seizures as confirmed by video and by EEG recordings from both hippocampi (Fig. 7). Two of the three animals had one stage IV seizure each, and one had two such seizures. No spontaneous seizures were seen in any of the other three comparison groups (Hypoxic-Heterozygous group n = 7; Hypoxic-Wild Type n = 7; Normoxic-Heterozygous n = 5; Normoxic-Wild Type n = 7; p = 0.026, Kruskal-Wallis test comparing all four groups).
Kv1.1 voltage-gated potassium channels are expressed broadly in the brain, are concentrated along axons and axon terminals, and modulate action potential generation (Robbins & Tempel, 2012). We chose the Kcn1a gene because (1) it predisposes to seizures in humans (Zuberi et al., 1999); (2) its heterozygous mutations predispose to flurothyl-induced seizures in mice, but such mice as we observed and as reported previously do not have spontaneous recurrent seizures (Smart et al., 1998); and (3) Kv1.1 channels have a role in the processes that follow hypoxia in the developing brain (Deng et al., 2011). In terms of differences in development between heterozygotes and litter mate controls, we observed no discernible difference in weight between heterozygotes and wild-type mice either during the perinatal period or as adults (Fig. S1). Our data demonstrate two notable findings in a mouse model of mild perinatal hypoxia: (1) the emergence, after a latent period, of increased susceptibility to flurothyl-induced seizures and to flurothyl kindling; (2) the seizure predisposing effects of perinatal hypoxia were additive to those of genetic susceptibility, thereby demonstrating that a modifier gene can influence long-term responses to hypoxic injury and that the added effects of both can increase the risk of spontaneous seizures.
Potential relevance of our model
Although using different neuroanatomic, cellular, synaptic, and molecular milestones may give different corresponding age equivalence between rodent and human brains, overall the predominance of the recent evidence points that the P5–9 mouse is a close comparison to the human neonatal brain (Avishai-Eliner et al., 2002; Clancy et al., 2007). A significant percentage of neonates experience acute seizures in association with hypoxia (Murray et al., 2006). In addition, when neonatal seizures do occur there is often a lack of EEG correlation, whereas at other times the same manifestations may be associated with EEG seizures. This is similar to our observations in the P6 mouse and to that of Jensen et al. (1991) in the P10 rat. Of note is that we investigated mice for seizures during acute hypoxia and that most of the human neonatal studies reported on posthypoxia seizure activity. It is possible that the behavioral seizures we observed are either nonepileptic in nature, or that our hippocampal recordings missed electrographic seizures in other areas of the brain. The use of monopolar electrodes, one in each hemisphere, in our study should have reduced, but not totally eliminated, that possibility. There have been many studies describing in various animal models the consequences of perinatal hypoxia and their underlying pathophysiology. Each model provided specific advantages and added insights into the effects of perinatal hypoxia (Chiba, 1985; Matsumoto, 1990; Jensen et al., 1991, 1992; Koh and Jensen, 2001; Mikati et al., 2005, 2011; Talhouk et al., 2008; Zeinieh et al., 2010; Dudek & Stanley, 2011). The hypoxia paradigm we used is one that is milder than many of the other previous models mentioned earlier. Although this paradigm was associated with only a modest effect on predisposition to flurothyl-induced seizures and flurothyl kindling, it had the advantage of allowing us to study the effects of age and of a modifier gene mutation on the consequences of hypoxia. Previous work by Rakhade et al. (2011) in a rat model of hypoxia-induced neonatal seizures showed a higher prevalence of spontaneous adult seizures compared to our results presented here. This is likely to be due to a number of factors including the degree of hypoxic injury, age at time of insult, genetic background, species, and to the fact that in our study we counted only stage IV seizures, which would underestimated the total number of seizures (Rakhade et al., 2011). Our mouse model is similar to mild perinatal hypoxia in humans, rather than to more severe global hypoxia ischemia or to acute perinatal stroke. Having established this model in a mouse could potentially allow the utilization of the powerful tools of mouse genetics to study the additional factors that predispose to epileptogenesis after less severe forms of perinatal hypoxia in future studies.
Effects of age: analysis of our observations and comparison with previous literature
Our findings that the enhancing effects of perinatal hypoxia on flurothyl seizures and flurothyl kindling were age dependent extend the findings of earlier studies. Previous investigations examining the development of electrical kindling after perinatal hypoxia have presumably revealed apparently contradictory results. Some studies have observed accelerated amygdalar kindling (Chiba, 1985), and hippocampal kindling (Sato et al., 1994), of adult rats that were previously subjected to hypoxia at P10–12. On the other hand, other studies reported that the rate of amygdalar kindling was not different after hypoxia treatment in juvenile (Moshé & Albala, 1985), or adult rats (Applegate et al., 1996). Our findings in flurothyl kindling are consistent with those of the former studies of Chiba (1985), Sato et al. (1994), and have the advantage of studying two different age groups in the same study. Our results are also consistent with the previously reported age effects on electrical kindling: As compared to adult animals, in the nonhypoxic developing brain the rates of kindling epileptogenesis have generally been shown to be either equivalent (Moshé et al., 1981; Baram et al., 1993) or reduced (Gilbert & Cain, 1982). In addition, the work of Lothman demonstrated that although rapid kindling occurred at P7–28, P7 rats were less capable of retaining the kindling effect than were older rats (Michelson & Lothman, 1991).
Our findings of an apparent latent period in the predisposition to flurothyl-induced seizures and kindling warrant further comments. Previous studies about the presence or absence of a latent period after perinatal hypoxia have concentrated on studying the time course of spontaneous seizures, which we did not do, and have reported seemingly contradictory results. It is worthwhile to examine here the results of these studies because the presence of such a period may allow for interventions that may interfere with the processes leading to epilepsy during development (Mikati et al., 1994, 2007; Bolanos et al., 1998; Cilio et al., 2001). Rakhade et al. (2011) reported a lack of seizures during the P13–20 time period in rats that had undergone a relatively mild hypoxic insult at P10. On the other hand, Dudek and Stanley (2011) and Kadam et al. (2010)) demonstrated a continuous and progressive increase in spontaneous seizures, without a latent period, in rats exposed neonatally to the more severe insult of hypoxia and carotid artery ligation. We did not monitor continuously starting directly after P6 hypoxia for spontaneous recurrent seizures and, thus, cannot comment on the presence or absence of a latent period for development of spontaneous seizures in our model. However, we did observe a delay in the predisposition to flurothyl seizures and flurothyl kindling. Taken together, all the previously mentioned studies, as well as ours, raise the possibility that milder purely hypoxic insults may be associated with latent periods for predisposition to seizures, whereas more severe ones may not be. Whether the occurrence of an apparent latent period in some models is caused by the specific developmental stage or, merely, to the passage of time irrespective to initial age, or to both, is an interesting question that needs to be addressed in future studies.
Effects of age: potential underlying mechanisms
All the P6 pups that were treated with hypoxia were included in the analysis. The reasoning for this is that our goal was to determine the effects of hypoxia on seizure threshold and spontaneous seizure frequency in adulthood, rather than to compare the effects of hypoxia with seizures with those of hypoxia without seizures. The majority of pups, both heterozygotes and wild type, had acute hypoxic seizures, most of which were behavioral and at least some were electrographic, which mimics the human condition. The potential mechanisms that may underlie our observations should be the subject of future studies, but currently we know that the acute and subacute changes in excitability that follow hypoxia in the rat model have been demonstrated to be largely due to calcium permeable AMPA (alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptors (Talos et al., 2006; Zhou et al., 2011), reduced inhibition (Stafstrom, 2009), and mossy fiber sprouting (Rakhade et al., 2011). Additional potential factors in our model may involve age-dependent mechanisms given our observations on the effects of age. One such candidate factor may be brain-derived neurotrophic factor (BDNF), which has been shown to increase after perinatal hypoxia and which contributes to adult epileptogenesis in models of temporal lobe epilepsy (Han & Holtzman, 2000).
Effects of gene mutation: analysis of our findings and comparison with previous literature
Our finding that the presence of an epilepsy predisposing gene mutation modifies the consequences of perinatal hypoxia is consistent with clinical studies, which indicate that genetic as well as environmental factors help to determine the epileptic phenotype (Scheffer & Berkovic, 1997; Choueiri et al., 2001). In addition, several animal studies have demonstrated that mouse strain can influence seizure threshold (McKhann et al., 2003; Papandrea et al., 2009). However, combining two factors that predispose to epilepsy does not necessarily enhance epileptogenicity (Glasscock et al., 2007). There have been a few studies that have investigated the effect of gene mutations on the consequences of hypoxia. Yamada et al. (2001) demonstrated that mutant mice lacking the Kir6.2 subunit of KATP channels (knockout mice) were susceptible to generalized seizures directly after brief hypoxia. However, unlike ours, their study was not on neonatal mice and did not investigate the long-term effects of hypoxia in these mice. Fung et al. (2010) found that hypoxic-ischemic brain injury exacerbates neuronal apoptosis and precipitates spontaneous seizures in glucose transporter isoform 3 heterozygous null mice, but the effect of the gene mutation appeared to be more related to the underlying metabolic problem and exacerbation of the acute injury rather than to subsequent processes.
The choice to examine a relatively moderate to mild hypoxia paradigm may have made it more likely to observe an additive effect. However, the effects of a gene mutation may be different with more severe hypoxia, and further studies will need to be conducted to determine those effects. Consistent with the previous literature (Smart et al., 1998), we did not observe any spontaneous seizures in the Heterozygous-Normoxia group, but we did observe an additive effect of hypoxia and the gene mutation. Of course our data do not rule out the possible occurrence of seizures at other times in the Wild Type-Hypoxia group if longer video-EEG recordings were performed and reviewed (Dudek & Stanley, 2011; Rakhade et al., 2011).
Effect of gene mutation: potential underlying mechanisms
Kv1.1 affects neuronal excitability and neurotransmitter release is involved in cytokine release (Deng et al., 2011) and in BDNF regulation (Lavebratt et al., 2006). Any of these or other related mechanisms may potentially be involved in the long-term additive effect of the Kcn1a gene mutation to those of perinatal hypoxia. Another, less likely, possible explanation is that the presence of the mutation resulted in a more severe acute injury during P6 hypoxia. Investigation of this possibility would require extensive further studies, which were beyond the goals of our study. The fact that the gene is not expressed at P6 argues against this possibility (http://developingmouse.brain-map.org).
Conclusions and Implications
Our findings support the concept that the presence of epilepsy predisposing gene mutations could account for at least part of the heterogeneity of clinical outcomes after perinatal hypoxia. Our data also suggest that following at least certain types of hypoxic injury, there is a delay in the development of predisposition to seizures and to kindling. This raises the possibility that manipulation of the mechanisms that may be active during such a delay could, in principle, prevent development of epilepsy. In addition, our data suggest that genome-wide association studies and exome/genome sequencing investigations in humans exposed to perinatal hypoxia may prove to be able to improve our ability to prognosticate and predict long-term outcome (Calkavur et al., 2011). Such knowledge may eventually help target intervention studies in neonates to those infants who are at higher genetic risk for long-term sequelae.
We would like to Dr. James O. McNamara for his invaluable contributions during the planning of the study, material support from his lab during the performance of the study and for his review of the manuscript. We also thank the staff in the Division of Pediatric Neurology and the Clinical Neurophysiology Program, including Dr. Christa Swisher, Lindsay Johnson, Michael Pompliano, and Dr. Piyush Kumar Singh. Finally we thank the whole of the McNamara laboratory, especially Wei Qian. Supported by the J. Gary and Dawn P. Burkhead fund #391-4486 (MAM).
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.