The immature brain differs from the adult brain in its susceptibility to seizures, seizure characteristics, and responses to antiepileptic drugs (AEDs) (1). Experimental evidence from animal models has revealed factors that may contribute to the age dependence of epileptic syndromes, and further suggests that early life seizures or treatment with AEDs can alter the normal maturation of brain function (1). To understand fully the age dependence of childhood epilepsies and to optimize treatments, it is necessary to consider maturational differences in the cellular and molecular mechanisms of epilepsy and how these may be functionally altered by early-life seizures or AED treatment. We discuss the developmental regulation of several factors and how these may contribute to seizure susceptibility and epileptogenesis in the developing brain.
SEIZURES IN THE IMMATURE HUMAN BRAIN
Seizure incidence is highest in the first year of life, and decreases with age throughout childhood and adolescence (2), and the majority of seizures in the first year of life are symptomatic (i.e., triggered by fever or hypoxia) (3). Early-life seizures can differ qualitatively from seizures in the adult. For instance, the electroencephalogram (EEG) in neonatal seizures frequently shows multifocal and asynchronous epileptiform activity (2), and behavioral seizures may be difficult to distinguish from spontaneous and voluntary movement (4). Although seizures later in childhood may resemble those in the adult, the etiologies of childhood epilepsies are different and can be highly age specific (5).
The immature brain can differ from the adult brain in its response to conventional AEDs (6). For example, neonatal seizures can be refractory to AEDs that are potent anticonvulsants in adults (7). Some types of seizure that are generally restricted to early childhood, such as absence or febrile seizures, respond to only a few of the AEDs that are commonly effective in adults (8,9). Children also show greater individual variation in responses and adverse cognitive side effects to AEDs (10). A major concern in neonatal epileptology is whether prolonged AED treatment may impair the normal activity-dependent maturation of brain function. An adverse effect of therapy is supported by a prospective study that revealed that children who were treated with phenobarbital (PB) for febrile seizures exhibited long-term cognitive and intellectual impairment (11,12). The long-term effects of other AEDs have yet to be systematically examined by similar prospective studies in humans.
In terms of prognosis, neonatal seizures represent a heterogeneous group (3,13), and it is difficult to identify which clinical characteristics of neonatal seizures determine a poor prognosis. Certain seizures can be benign, as in the case of neonatal hypocalcemic or hypoglycemic seizures in the first week of life (3), or simple febrile seizures in childhood (14). However, other early-life seizures may be related to long-term epilepsy. There is an incidence of later-life epilepsy in cases of symptomatic seizures in the neonate or complex and recurrent febrile seizures in childhood (2,15–18). A controversial question in clinical neonatal and pediatric epileptology is whether early-life seizures per se adversely alter brain maturation, leading to epilepsy or other neurologic syndromes. Patients with temporal lobe epilepsy (TLE) associated with mesial temporal sclerosis (MTS) have a reported increased incidence of previous neonatal or childhood febrile seizures (19–23). Van Landingham et al. (24) found that prolonged febrile convulsions in a small fraction of infants were associated with acute hippocampal edema and delayed hippocampal atrophy. These observations raise the possibility that childhood seizures that occur during a critical maturational period could alter brain development to increase the susceptibility to MTS and TLE. However, prospective and community-based studies have found that childhood febrile seizures are not more likely to be associated with TLE and MTS than are other types of generalized epilepsy (25–27). Thus, although febrile seizures are associated with an increased risk of epilepsy, it is not clear whether the seizures contribute to epileptogenesis or that they are merely indicative of an epilepsy-prone brain.
In summary, clinical experience suggests some major principles regarding the relationship between brain maturation and epilepsy. First, the immature brain is more susceptible to seizures compared with the adult brain. Second, the response of the immature brain to AEDs is more variable, and prolonged AED treatment in early life may adversely alter brain development. Third, certain types of seizure in early life are associated with increased likelihood of chronic epilepsy and may themselves contribute to epileptogenesis. Knowledge of age-specific mechanisms of epilepsy and the potential consequences of early-life seizures or AED treatment on brain maturation will help to understand better the age dependence of childhood epilepsies and to optimize age-appropriate treatments.
MATURATIONAL FACTORS THAT MAY ENHANCE SEIZURE SUSCEPTIBILITY IN THE IMMATURE BRAIN
Animal studies indicate that the early postnatal period represents a critical developmental window in which synaptogenesis is ongoing and neuronal plasticity is increased compared with the adult (28–31). Excitatory synaptic transmission mediated by glutamate receptors is required for these processes and is enhanced in the immature brain compared with that of the adult (29,30). Synaptic density undergoes a postnatal overshoot before being “pruned” to adult levels (28,31), and this is paralleled by a relative overexpression of glutamate receptors (32,33). In rat, glutamate-receptor expression is highest in the second postnatal week (32,33), a developmental stage that is roughly analogous to the human term neonate with respect to a number of anatomic, physiologic, and biochemical parameters (34). Additional factors that govern synaptic transmission and neuronal excitability continue to change during this developmental window, including the expression and molecular composition of neurotransmitter receptors and transporters, neuromodulatory peptides and neuropeptide receptors, voltage-gated ion channels, and mechanisms of ionic homeostasis. These factors and their possible relationship to epilepsy in the developing brain are discussed later.
Ionotropic glutamate receptors
Glutamate is the major excitatory neurotransmitter in the brain, and there are several subtypes of glutamate receptor. These include the N-methyl-D-aspartate (NMDA), α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA), and kainate (KA) subtypes of ionotropic receptors (35), as well as different classes of metabotropic glutamate receptors (36,37). AMPA and KA receptors mediate fast excitatory signaling as they exhibit rapid activation and desensitization, and operate linearly near the resting membrane potential. NMDA receptors play a more modulatory role, as their activation requires concurrent glutamate binding and membrane depolarization and results in slower and longer-lasting excitation. NMDA-receptor channels are highly permeable to Ca2+ (in addition to Na+ and K+), and the influx of Ca2+ through NMDA receptors can trigger signaling pathways that regulate synaptic function and activity-driven synaptogenesis (38,39). The kinetics and permeation properties of NMDA receptors also underlie their key role in pathophysiologic processes such as ictal seizure discharges and hypoxic/ischemic neuronal injury (40–42).
Binding studies in the rat indicate that NMDA-receptor density peaks late in the first postnatal week in many forebrain structures, including hippocampus and neocortex (32)(Fig. 1). AMPA-receptor density appears to peak later in the second postnatal week around P10 (32), whereas kainate receptors gradually increase over the first few weeks of life (43). The first 2–3 postnatal weeks represent a window during which glutamate-mediated synaptic plasticity is enhanced (28), and certain brain regions exhibit heightened susceptibility to the epileptogenic and excitotoxic effects of glutamate-receptor agonists (33,44–46). Thus, the overshoot in expression of functional glutamate receptors is likely to play a major role in the increased excitability of the brain in early postnatal development.
The molecular composition of glutamate receptors also is developmentally regulated, and may contribute to the enhanced excitability of the immature brain. The functional properties of glutamate receptors are determined by the particular combination of molecular subunits that compose each receptor (35). In neocortex and hippocampus, expression of the NMDA-receptor subunit NR2B relative to the expression of NR2A is much higher during early postnatal development compared with that in adulthood (47,48). Recombinant NMDA receptors that contain predominantly NR2B exhibit slower decay times than do those that contain predominantly NR2A (48), and the developmental increase in NR2A expression results in a maturational shortening of NMDA receptor–mediated synaptic currents (49). Thus, synaptic excitability and plasticity may be increased in the immature brain by the increased duration of NMDA receptor–mediated excitation and increased Ca2+ influx.
Notably, most AMPA and kainate receptors in the mature brain show little relative permeability to Ca2+, but a larger proportion of these channels may exhibit Ca2+ permeability in the immature brain (50). AMPA receptors that lack a GluR2(B) subunit exhibit significantly higher permeability to Ca2+ and other divalent cations than do those that contain a GluR2(B) subunit (51–54), and the ratio of expression of GluR2 subunits to that of other AMPA-receptor subunits appears to be significantly lower in immature neocortex and hippocampus compared with those of the adult (55). This suggests that a larger proportion of AMPA receptors are permeable to Ca2+ in immature neurons in these brain regions, and may therefore mediate pathophysiologic events in early postnatal life, similar to NMDA receptors (50,56). Consistent with this notion, transient knockdown of GluR2 expression using hippocampal microinfusions of antisense mRNA was recently shown to cause spontaneous seizure-like behaviors in young rats (57).
Metabotropic glutamate receptors
There are at least eight cloned metabotropic glutamate receptors (termed mGluR1–mGluR8) (36). These have been classified into three groups based on sequence homology, coupling to second-messenger systems, and pharmacologic sensitivities. Group I receptors are coupled to phosphoinositide (PI) hydrolysis that leads to Ca2+ mobilization from intracellular stores, whereas group II and III receptors are negatively coupled to adenylyl cyclase (AC) activity. Although the consequences of mGluR activation vary depending on receptor type, neuronal type, or brain region, some general principles regarding the effects of mGluR activation in relation to seizures have emerged (37). Postsynaptic group I mGluR activation in general causes an increase in the intrinsic excitability of principal neurons (particularly in hippocampal CA1 and CA3 subfields), mainly by downmodulation of voltage-gated potassium channels (58), and therefore, activation of PI-coupled mGluRs is likely to promote seizure activity. Conversely, presynaptic group II and III receptor activation tends to depress excitatory synaptic transmission by inhibiting glutamate release (59), and therefore, activation of AC-coupled mGluRs is likely to inhibit seizure activity.
Given these potential roles for mGluRs in the generation or inhibition of seizure activity, available evidence suggests that the developmental regulation of mGluR function also may contribute to the increased seizure susceptibility of the immature brain. mGluR agonist-stimulated PI turnover has been shown to be relatively robust in slices of immature rat brain, increasing from age P1 to P7–10 before gradually decreasing to adult levels at around P24 (60). This appears to contrast with the activity of mGluRs negatively coupled to AC, as cyclic AMP accumulation induced by the AC activator forskolin was shown to be inhibited by the nonspecific mGluR agonist 1S,3R(ACPD) in adult but not in neonatal (P1–15) rat hippocampus (61,62). Interestingly, mGluRs negatively coupled to AC are expressed in early postnatal development, but nonspecific mGluR activation in the neonatal hippocampus increases basal cyclic AMP levels (63). Thus, regardless of mGluR gene expression, the developmental pattern of mGluR function may favor a hyperexcitable state as the activity of postsynaptic mGluRs that promote increased intrinsic neuronal excitability can predominate over mGluRs that presynaptically regulate neurotransmitter release.
The expression of glutamate transporters also is developmentally regulated and may play a role in the enhanced excitability of the immature brain. In animal models, decreased expression of glutamate transporters can lead to seizures or lower seizure thresholds (64). For example, knockout mice lacking the glutamate transporter GLT-1 display lethal seizures that are related to elevated extracellular glutamate levels (65). Additionally, mice deficient in the glutamate transporter GLAST do not exhibit spontaneous seizures, but show shorter onset latency and more severe stages of pentylenetetrazol-induced seizures compared with wild-type mice (66). Notably, the expression of both of these astrocytic glutamate transporters gradually increases during postnatal development in the rat (67). Thus, a relatively lower activity of certain glutamate transporters during development could contribute to enhanced seizure susceptibility in the immature brain. However, the developmental profile of glutamate-transporter expression and regulation differs across transporter subtypes and brain regions (67,68), and the specific contributions of each subtype during development are yet to be fully elucidated.
γ-Aminobutyric acid (GABA) is the predominant inhibitory neurotransmitter in the brain, and the expression and function of GABA receptors also are developmentally regulated. GABAA receptors, which mediate postsynaptic responses to GABA in central neurons, are expressed at embryonic stages (69). However, in the first postnatal week, activation of GABAA receptors causes membrane depolarization rather than the hyperpolarization typical of mature GABA-ergic synapses (see Fig. 1) (28,70). This difference is not due to receptor composition, but rather results from maturational changes in the transmembrane chloride ion gradient, as this largely governs the equilibrium potential for GABAA channels (71). Inhibitory (hyperpolarizing) GABAA receptor–mediated potentials gradually appear over the first 3 postnatal weeks (70)(Fig. 1) and are temporally correlated with the induction of expression of the neuronal K+/Cl– cotransporter KCC2, which extrudes Cl– from cells (72). Thus, although functional GABA receptors are present very early in development, the delayed onset of functional GABA-ergic inhibition may contribute to the enhanced excitability of the immature brain.
The molecular composition of GABAA receptors also is developmentally regulated with associated changes in functional properties (73,74). For example, the expression of GABAA receptor α1 subunits is low at birth and gradually increases with maturation (69,75–77). This maturational change is associated with a gradual shift toward the more rapid kinetics and increased sensitivity to certain benzodiazepines [e.g., diazepam (DZP)] that are characteristic of most GABAA receptors in the adult brain (76,78,79). Such a developmental change in the pharmacologic properties of GABAA receptors suggests that the neonatal brain may respond differently from the adult brain to AEDs that act by enhancing GABAA-receptor function.
Unlike GABAA receptors, the G protein–coupled GABAB receptors are activated both pre- and postsynaptically, with opposite effects on synaptic transmission (80). Postsynaptic GABAB receptors mediate relatively slowly activating and long-lasting membrane hyperpolarization through the activation of a K+ conductance, whereas the activation of presynaptic GABAB receptors decreases neurotransmitter release through the inhibition of Ca2+ channels (80). In animal models of epilepsy, GABAB receptors can have varied roles depending on the particular model and brain region involved. For example, infusion of a GABAB-receptor antagonist into the substantia nigra decreased thresholds for flurothyl-induced seizures in immature rats (81), whereas GABAB-receptor antagonists abolish seizures in several animal models of absence epilepsy (82,83).
GABAB-receptor binding in rat brain increases during postnatal development and peaks in a regionally specific manner during the first 3 weeks, before declining to adult levels (84). In neocortex and thalamus, binding gradually increases to an overshoot at P14 before decreasing to mature levels, whereas peak binding occurs within the first postnatal week in hippocampus (84). Notably, in hippocampus, the presynaptic effects of GABAB-receptor activation are observable earlier in development than are the postsynaptic effects (80). If the activation of presynaptic GABAB receptors depresses GABA release, then the earlier appearance of presynaptic GABAB-receptor function would be expected to promote synaptic excitability. Although a clear role or age dependence of GABAB receptors in the generation or suppression of limbic seizures has not been established (85), GABAB-receptor activation in the thalamus is clearly required for the generation of absence seizures in most animal models, including those that mimic the age dependence of clinical absence epilepsy (83). Thus, it has been suggested that the age dependence of absence seizures may derive from a mismatch in the rates of maturation of GABAB-mediated and glutamate receptor–mediated transmission in thalamocortical circuits such that the maturation of NMDA-receptor function lags behind that of GABAB receptors before adulthood (83).
The developmental regulation of certain neuromodulatory peptides also may influence the excitability of the immature brain. The excitatory neuropeptide corticotropin-releasing hormone (CRH) is the most potent epileptogenic peptide (86) and may play a critical role in the “triggering” of seizures (i.e., by fever or hypoxia) in the immature brain (87). In immature rat amygdala, the expression of CRH receptors reaches twice that of the adult level in the second postnatal week, a developmental stage during which the immature brain is most susceptible to CRH-induced seizures (88,89). Other neuromodulatory peptides, such as neuropeptide Y and somatostatin, can modulate seizures (86), and their function and expression are developmentally regulated (90,91). However, specific roles for these neuropeptides in seizures in the immature brain have not been investigated extensively.
VOLTAGE-GATED ION CHANNELS AND ION HOMEOSTASIS
The expression of epileptiform activity requires synaptic communication among neurons, and therefore, the discussion heretofore has focused on the role of synaptic mechanisms in seizures and epileptogenesis. However, the intrinsic excitability of postsynaptic and axonal membranes is largely determined by the activity of voltage-gated ion channels, and the developmental regulation of voltage-gated channels and ionic homeostasis may contribute to the increased seizure susceptibility of the immature brain.
The developmental patterns of expression and function of voltage-gated ion channels vary across cell type and brain region. In general, these patterns promote increased intrinsic membrane excitability during embryogenesis and early postnatal development, as spontaneous electrical activity is necessary for cell differentiation, migration, and synaptogenesis (92). Action potentials in immature neurons are of significantly longer duration compared with their adult counterparts, largely due to slower activation rates of delayed rectifier K+ channels that repolarize the action potential (92,93). During an action potential, the amount of Ca2+ influx through voltage-gated Ca2+ channels is determined largely by action-potential duration, and therefore, the longer action potentials in immature neurons play a key role in activity-driven Ca2+-dependent developmental events. For example, premature shortening of the action potential by overexpression of a delayed-rectifier K+ channel (Kv1.1) in embryonic Xenopus spinal neurons caused a decrease in the number of neurons achieving terminal differentiation (94). Notably, knockout mice lacking Kv1.1 exhibit spontaneous seizures beginning in the third postnatal week (95). Clearly, the increased intrinsic membrane excitability in immature neurons is critical for normal development, but at the apparent expense of enhancing network excitability while development is ongoing.
The developmental regulation of voltage-gated channels in axons and at presynaptic terminals may be particularly critical in the generation of seizure activity through their influence on neurotransmitter release. In epileptic DBA/2J mice, a strain genetically prone to audiogenic seizures, binding of a radiolabeled N-type Ca2+-channel toxin (ω-conotoxin GVIA) indicated that the expression of these presynaptic Ca2+ channels was abnormally increased between postnatal days 2–8 in parallel with seizure susceptibility (96). In contrast, nonepileptic mice showed no change in ω-GVIA binding during the same period, but exhibited a rapid increase in binding between postnatal days 11 and 14 (97). Notably, although these mice are not epileptic, this period of N-channel proliferation coincides with the period of heightened susceptibility in rodents to seizures induced by hypoxia or hyperthermia, as discussed later. Thus, the developmental change in presynaptic Ca2+ channels that govern neurotransmitter release may be a critical component in determining the early postnatal susceptibility to seizure.
Mechanisms of ion homeostasis also are changing during development. As mentioned earlier, the neuronal expression of chloride transporters has a postnatal onset, and in their absence, the transmembrane Cl– gradient is such that Cl–-mediated currents are depolarizing. The major neuronal Na+/K+ adenosine triphosphatase (ATPase) also is less abundant in the immature brain (98), and thus, moderate increases in neuronal activity could regeneratively cause extracellular K+ to increase to epileptogenic levels. Intracellular Ca2+ homeostasis also may be developmentally regulated. For example, the calcium-binding protein calbindin D28K is not expressed in immature hippocampal dentate granule cells, but is expressed by these same cells as they mature (99).
ANIMAL MODELS OF SEIZURE AND STATUS EPILEPTICUS IN THE IMMATURE BRAIN
A number of experimental seizure models have at least in part recapitulated components of the age-specific seizure syndromes (100–102). Symptomatic early-onset seizures can be modeled in the developing brain by exposure to chemoconvulsants such as kainate, pilocarpine, flurothyl, and tetanus toxin, or by repeated electrical stimulation (kindling). Seizure thresholds to chemoconvulsant and kindling stimuli are much lower in immature animals (100,103). Additionally, certain pathophysiologic conditions such as hypoxia or hyperthermia may precipitate seizures in the immature animal, while having no epileptogenic effect in the adult (104,105). As mentioned earlier, CRH can trigger spontaneous seizures in immature animals, but not in adults (87).
In general, these models exhibit age dependence not only in their efficacy for inducing seizures, but also in their long-term effects on brain function. Unlike that in the adult, the immature brain appears to be relatively resistant to seizure-induced neuronal injury (1,102). Kainate injections induce status epilepticus in immature animals with no delayed neuronal loss despite ≥2 h of continuous seizures (106–109). More severe epileptogenic conditions have been found to both induce seizures and cause neuronal injury in the immature brain (110), and pilocarpine status-induced neuronal death in the immature brain has been reported (111). However, the extent of death is significantly less than that seen after status in the adult brain (112). These data suggest that although the resistance of the immature brain to chemoconvulsant-induced neuronal death is not absolute, it is much greater relative to the adult.
Despite the lack of significant injury in the immature brain in these experimental models, other evidence suggests that neonatal seizures may adversely alter the function of surviving neurons and neuronal circuitry to reduce seizure thresholds permanently or to promote epileptogenesis. In rat pups at age P10, global hypoxia induces spontaneous seizures and results in chronically decreased thresholds to chemoconvulsant-induced seizures (113) and increased excitability in hippocampal slices (114), despite no apparent neuronal injury (115). Similarly, hyperthermia-induced convulsions in the P10 rat do not cause acute cell death, yet result in long-lasting changes in inhibitory synaptic transmission (116), and permanently decreased seizure threshold and increased hippocampal excitability (104). Intrahippocampal tetanus toxin injections in P10 rat result in the development of acute seizures, which lead to spontaneous recurrent seizures later in life and aberrant mossy fiber sprouting in the dentate gyrus (117,118). Recurrent seizures induced by repeated flurothyl or pentylenetetrazol administration beginning in the second postnatal week result in long-term increases in seizure susceptibility with selective mossy fiber sprouting in hippocampal subfields (119,120). Notably, animals that experienced repeated flurothyl-induced seizures also showed modest but significant impairment in learning and memory in adulthood (119). The cellular and molecular events that mediate these morphologic and functional changes are still under investigation, but these data reiterate the concept that seizures can induce long-lasting, potentially adverse functional changes in the immature brain that may not appear acutely as injury.
Studies of the epileptic mature brain have revealed the contributions of various cellular and molecular factors to seizure expression and epilepsy. Given this information, we can now look at how the functional properties of many of these factors change with maturation, and infer the potential consequences for seizure susceptibility at different developmental stages. Although developmental changes can alter seizure susceptibility, seizures also may alter the normal pattern of brain development. To devise optimal age-specific therapeutic strategies for the treatment of neonatal and childhood seizures, it will be necessary to identify developmentally regulated factors that are critical for acute and long-term seizure susceptibility, and to understand how brain maturation may be altered by seizures.
The developmental profile of KA receptor function depicted in Fig. 1 was based purely on receptor binding studies. In situ hybridization has shown a more varied maturational pattern of KA receptor subunit gene expression, with GluR5 gene expression peaking at birth in sensory cortex, thalamus, and hippocampal CA1 interneurons (Bahn et al., J Neurosci 1994;14:5525–47). Recent electrophysiological evidence indicates that low affinity KA receptors are synaptically activated at thalamocortical synapses during early postnatal development (Kidd and Isaac, Nature 1999;400:569–73), and therefore, may have a role in synaptic signaling and plasticity in the immature brain. Speculation on possible roles for these receptors in determining seizure susceptibility await further study.
Acknowledgment: This work was supported by Public Health Service grants T32 AG00222 from the National Institute on Aging (RMS), R01 N531718 from the National Institute of Neurological Disorders and Stroke (FEJ), and the Children's Hospital Mental Retardation Research Center grant P30 HD18655 from the National Institute of Child Health and Development.