|Home | About | Journals | Submit | Contact Us | Français|
Studies of febrile seizures have been driven by two major enigmas: first, how these most common of human seizures are generated by fever has not been known. Second, epidemiological studies have linked prolonged febrile seizures with the development of temporal lobe epilepsy, yet whether long or recurrent febrile seizures cause temporal lobe epilepsy has remained unresolved. To investigate these questions, a model of prolonged (complex) febrile seizures was developed in immature rats and mice, permitting mechanistic examination of the potential causal relationships of fever and seizures, and of febrile seizures and limbic epilepsy. Although the model relied on hyperthermia, it was discovered that the hyperthermia-induced secretion of endogenous fever mediators including interleukin-1β, which contributed to the generation of these ‘febrile’ seizures. In addition, prolonged experimental febrile seizures provoked epilepsy in a third of the animals. Investigations of the mechanisms of this epileptogenesis demonstrated that expression of specific ion (HCN) channels and of endocannabinoid signaling, may be involved. These may provide novel drug targets for intervention in the epileptogenic process.
Febrile seizures (FS), generally defined as seizures taking place during fever, and without an obvious central nervous system (CNS)-invasive infection, are the most common type of convulsive events in infants and young children . FS lasting less than 10 [2,3] or 15 min  have not been associated with subsequent epilepsy or cognitive deficits in prospective or retrospective studies [5–7]. However, the consequences of prolonged FS, one of the forms of complex FS, are controversial [2,8]. Retrospective studies have linked a history of prolonged FS and subsequent temporal lobe epilepsy (TLE) [9–12]. However, prospective studies failed to implicate prolonged FS as a strong candidate for causing TLE .
If prolonged FS are responsible for TLE, then a strong incentive exists for preventing them. Prevention of these seizures would benefit from an understanding of the means by which fever leads to seizure generation. This need to understand the pathophysiology of FS, and the controversy over the clinical outcome of prolonged FS have provided an incentive for the development of animal models for these seizures. Such models enable direct examination of the potential mechanisms for the generation and consequences of these seizures, using a panoply of investigational tools.
Febrile seizures occur in a strongly age-specific manner, supporting the strong contribution of factors that selectively characterize specific stages of brain development [14,15] (see Table 1 for the mechanisms of febrile seizures generation). The seizures are familial in some cases and sporadic in others, suggesting that both genetic and environmental elements contribute to their generation . The contribution of environment, namely the increase in brain temperature (hyperthermia) to seizure generation, has become evident from the use of animal models, where hyperthermia leads to seizures in practically all rats or mice. This suggests that genetic susceptibility is not necessarily required for induction of such seizures [17–20]. Conversely, the fact that different mouse strains vary in the temperature required to generate seizures (seizure-threshold temperature; ), strongly implies that the genetic background may influence the susceptibility to developing a seizure with fever (although when the temperature is high enough, all tested strains develop seizures).
Several genes have been implicated in the susceptibility to febrile seizures, including those coding sodium channels, [22,23], GABAA receptors [24–26], and interleukins [27,28]. In addition, interactions among several genes might contribute to the occurrence of these seizures in a more complex manner.
Elevating brain temperature in itself alters many neuronal functions, including several temperature-sensitive ion channels [29,30]. This should influence neuronal firing and the probability of generating massive synchronized neuronal activity, i.e., seizures. Remarkably, hyperthermia provoked by medication overdose or hot baths often provoke seizures in young children , indicating that increase in brain temperature may suffice to generate seizures. Obviously, fever involves, in addition to increased brain temperature, also an inflammatory process including secretion of cytokines in the periphery as well as in the brain [32,33]. Indeed, it was discovered that fever and hyperthermia share common mechanisms to provoke seizures: The fever-promoting, pyrogen, interleukin-1β contributes to fever generation and, conversely, fever leads to the synthesis of this cytokine in hippocampus [34–37]. In addition, interleukin-1β, has been shown to increase neuronal excitability, acting via both glutamate and GABA . In vivo, these actions of interleukin-1β enhance the actions of seizure-provoking agents . In support of an important role for endogenous interleukin-1β in the generation of febrile seizures has come from studies in mice lacking the receptor for this cytokine. Much higher temperatures were required to elicit hyperthermic seizures in these mice , and interleukin-1β provoked seizures in immature rats and mice when given directly into the brain . In addition, using lipopolysaccharide (LPS), a bacterial toxin, to induce release of endogenous interleukin-1β 1 in rodents lowers the threshold to kainic acid, and combining LPS with low-dose kainic acid results in seizures [39,40].
It might be noted that fever of specific infectious etiologies, and specifically HHV6 might influence the probability of generation of febrile seizures [41,42]. Finally, hyperthermia-induced hyperventilation and alkalosis have been proposed as a pivotal element of febrile seizure generation (see elsewhere in this issue). As discussed more fully elsewhere , alkalosis of the brain has been shown to provoke neuronal excitability [44,45], and contributes to seizure pathophysiology in models where the latency between fever and seizure onset is long (30 min; ). Remarkably, human conditions associated with severe alkalosis, including prolonged crying and pyloric stenosis of infants, are not associated with the generation of seizures.
In children, a large behavioral spectrum of seizure is provoked by fever. Classical febrile seizures are short and lack motor phenomena that point to a focal origin. In other words, simple febrile seizures do not involve movements restricted to a single or unilateral limbs. However, in children, the onset of febrile seizure may not be recognized prior to the evolution of motor movement, so that subtle early components of the seizures may be missed (See Neville BG, elsewhere in this issue). Specifically, behavior arrest, confusion, dazed look or altered consciousness may be missed. These may point to origins of the seizures in the limbic system – the brain region most susceptible to seizures.
In rodent models, the behavioral and the EEG onset of the seizures can be more readily defined, and clearly originate in the limbic circuit: initial seizure behaviors of immature mice or rats involve arrest of movement, and this freezing is associated with loss of responsiveness to environmental stimuli. The next phase consists of oral automatisms, as typical for limbic human and rodent seizures . Notably, whereas EEG recoding is practically unavailable for the onset of febrile seizures in children, EEG traces from rodents with experimental febrile seizures suggest onset of these seizures in amygdala and hippocampus. Thus, accompanying the Racine stages 0–3 propagation of the behavioral seizures, bipolar electrode recordings from amygdala, hippocampus and cortex showed spike-trains in hippocampus and amygdala with progressively increasing amplitude [20,48,49].
The overwhelming evidence from human and animal models suggest that the outcome of short febrile seizures is benign. However, whether prolonged febrile seizures and febrile status epilepticus lead to epilepsy has been more difficult to sort out. Generally, prospective epidemiological evaluations have provided little evidence for epileptogenesis [2,4,8], although those looking at longer time-scales have shown an increasing probability for epilepsy development [50–53]. In contrast, retrospective analyses have linked a history of complex, and particularly of prolonged febrile seizures to temporal lobe epilepsy [9–12] suggesting a potential contribution of febrile seizures to epileptogenesis. The inconsistent nature of these data promoted the use of animal models, to study this issue more directly.
In animal models of long febrile seizures, the seizures are induced in a normal rodent brain (e.g., [17–21,39,40,45,48,54–57]), or in animals that have been subjected to prior insults [58,59]. Initial studies using an experimental model of prolonged febrile seizures in immature rats without genetic or imposed predisposing factors or insults, employed intermittent observation and EEG recording during daytime , and did not demonstrate the onset of spontaneous seizures in rats that had experienced experimental febrile seizures. More recently, nocturnal simultaneous video-EEG recordings demonstrated that these seizures led to limbic (TLE) in ~30% of animals . In addition, inter-ictal epilepti-form discharges were recorded in 15 (88.2%) of these animals exposed to prolonged experimental febrile seizures. Neither the EEG (>400 recorded hours) nor the behavior of normothermic controls and of hyperthermic controls, that experienced hyperthermia but in whom the seizures were prevented, demonstrated any abnormalities [43,60]. These studies support the epileptogenic nature of prolonged febrile seizures [43,60].
The mechanisms by which febrile status epilepticus or prolonged febrile seizures might contribute to the development of temporal lobe epilepsy are not known. The use of animal models might provide some helpful information in this context as well. The prolonged experimental febrile seizures led to transient neuronal injury . Interestingly, the injured neurons were located in the distribution of the cell loss and gliosis found in humans with mesial temporal sclerosis (MTS). However, the involved neurons did not die, as supported by neuronal counts [19,60,61], and acute apoptosis was not observed even after 60 min-long seizures . Neurogenesis was also not observed after these seizures [56,61–63], and mossy fiber sprouting was minimal  and not likely the source of the epileptogenic process.
Several molecular and functional changes took place following the experimental prolonged febrile seizures, and might provide the mechanisms for seizure-evoked hippocampal hyperexcitability [20,64,65]. The full spectrum of the molecular changes induced by the seizures in animal models is under study; however, already established are persistent changes in the expression of specific genes such ion channels and endocannabinoid receptors .
Experimental prolonged febrile seizures rapidly led to altered calcium signaling in hippocampal neurons, via the formation of calcium-permeable AMPA channels devoid of the GluR2 subunit . This altered route of calcium entry has been shown to promoted significant number of intracellular cascades, culminating in changes of gene expression [67–69]. One consequence was an alteration the expression of the ions channels governing the properties of Ih, a hyperpolarization-triggered cationic current that contributes to the maintenance of neuronal membrane potential, subthreshold oscillations and dendritic integration [46,65]. This change in Ih promoted frequency-dependent rebound depolarization in response to hyperpolarizing input, which was augmented after the seizures [64,65]. At the molecular level, the changes in Ih appeared to result from persistent altered expression of hyperpolarization-activated cyclic-nucleotide gated (HCN) channels that conduct this current. A reduction of the expression of the HCN1 iso-form was observed, as well as increased formation of HCN1/HCN2 heteromeric channels, which were >200% higher in the CA1 hippocampal region of animals experiencing prolonged febrile seizures compared to controls [49,70]. The relevance of the alterations in HCN channels and Ih to human epileptogenesis is not clear, but HCN1 channel expression was found to be changed also in hippocampi removed from patients with temporal lobe epilepsy and mesial temporal sclerosis, often with a history of early life seizures . This suggests that HCN channel expression is influenced also in human temporal lobe epilepsy. Similarly, it might be speculated that mutations in HCN channel genes which will alter Ih should be discovered in individuals with epilepsy.
A second change provoked by prolonged experimental febrile seizures that promoted hyperexcitability involved altered endocannabinoid signaling. In essence, the seizures increased the number of presynaptic can-nabinoid type 1 receptors, which increased retrograde inhibition of GABA release, promoting hyperexcitability . It is proposed that many other persisting alterations in gene expression will be found after experimental febrile seizures, and perhaps human prolonged febrile seizures. Potentially, common regulatory mechanisms will drive these changes, and the resulting alteration of intrinsic neuronal excitability and the responses of neurons to network input will contribute to the generation of a hyperexcitable state associated with spontaneous seizures.
Seizures provoked by fever are common, and teach us about mechanisms of seizure generation early in life. Whereas simple febrile seizures are benign, the pathophysiology of febrile seizures should be studied so that prolonged seizures and their potential consequences will be better understood. Animal models offer the hope of providing the mechanisms for simple and prolonged febrile seizures, as well as enabling an understanding of the proepileptogenic consequences of prolonged febrile seizures.
The authors thank Ms. Joy Calara for excellent editorial help. Supported by NIH Grant R37 NS35439.