Our results show that hyperthermia increases the intrinsic excitability of both excitatory and inhibitory neurons of the hippocampus. Hyperthermia strongly affected the physiology of all three major excitatory neuron populations. Hyperthermia-induced spiking was most prominent in CA3 pyramidal cells, which usually exhibited a burst-firing pattern. The hypersensitivity of CA3 is consistent with previous studies of other types of seizure models, which demonstrated that CA3 pyramidal cells are particularly susceptible to epileptiform activity (Schwartzkroin and Prince, 1977
; Traub and Wong, 1982
; Wong and Traub, 1983
; Tancredi et al., 1990
; Jensen and Yaari, 1997
; Rutecki and Yang, 1998
; Derchansky et al., 2006
; Isaev et al., 2007
). Studies of the long-term effects of hyperthermic seizures have focused on the CA1 area (Chen et al., 1999
; Dube et al., 2000
; Brewster et al., 2002
; Han et al., 2006
); post-febrile seizure modifications induced in CA3 should also be explored.
GIN (O-LM) inhibitory interneurons were even more sensitive to febrile temperatures than pyramidal cells. The interneurons robustly fired spikes in response to heat in both CA1 and CA3. Unlike pyramidal cells, GIN cells in CA1 were as likely to generate hyperthermia-induced spikes as those in CA3, although the maximal firing frequency reached at 41°C in CA3 cells was twice as high in that in CA1 cells. O-LM cells are known to preferentially fire in the theta frequency range (3–11 Hz), both in vivo
and in vitro
(Fanselow et al., 2008
; Klausberger and Somogyi, 2008
). O-LM cells are thought to be important for generating and maintaining theta oscillations in the hippocampus (Gloveli et al., 2005
). Cortical oscillations and seizures have a close relationship. Cortical oscillation frequencies can change as a result of seizures, such as a change from theta to gamma frequencies after temporal lobe seizure induction (Dugladze et al., 2007
), and seizures can also result from changes in cortical oscillations frequencies, such as from sleep spindle (6–14 Hz) to absence seizure (3–4 Hz) frequencies or (Blumenfeld and McCormick, 2000
). Hyperthermia may similarly increase the firing frequency of O-LM cells, thereby affecting synchronization states within the hippocampus. Previous reports have shown that when O-LM cells increase their firing frequency, there appears to be an increase in gamma coherence across large distances (Tort et al., 2007
). In turn, this may encourage hypersynchrony and facilitate the development of seizures.
Significant changes of intrinsic membrane properties accompanied hyperthermia-induced increases in spiking activity. Membrane potential depolarized and input resistance decreased in all cell types. Previous reports indicate that depolarization of cortical pyramidal cells increases apparent input resistance, as measured in current-clamp, due to the activation of persistent sodium conductances (Connors et al., 1982
; Stafstrom et al., 1985
; Spruston and Johnston, 1992
; Cruikshank et al., 2007
). Indeed, we found that depolarization alone, at 30°C, increases input resistance. Thus, if anything, we are probably underestimating the degree to which input resistance is reduced by high temperatures. Another surprising finding is that depolarization block of GIN interneurons occurred at more hyperpolarized potentials during hyperthermia compared to 30°C. These data imply that the mechanism by which hyperthermia leads to depolarization is probably distinct from its mechanisms of altering input resistance and intrinsic excitability.
Our results suggest that studies targeting CA3 populations of neurons may provide the most insight into the mechanisms of heat-induced excitability, which we did not address here. Temperature affects all biochemical processes. Input resistance decreased during hyperthermia in principal cells as well as O-LM interneurons. The combination of depolarization and decreased input resistance suggests that hyperthermia increases membrane sodium and/or calcium conductance. Heat-activated channels, such as transient receptor potential vanilloid 1 (TRPV1) receptors (Dhaka et al., 2006
; Gibson et al., 2008
), could influence excitability. The effects of interleukin-1β application on different cell types or recordings of mice with an IL-1ra deletion, previously shown to affect febrile seizure threshold would be interesting to investigate as well. Mutations of GABAA
channels can lead to familial forms of generalized epilepsy with complex febrile seizures [GEFS+; (Scheffer and Berkovic, 1997
; Spampanato et al., 2004
; Nakayama, 2009
)]. Febrile seizures lead to long-term changes in hyperpolarization-activated cyclic nucleotide-gated (HCN) channel and endocannabinoid channel signaling (Chen et al., 2001
; Brewster et al., 2002
); these processes might also be important in the generation of febrile seizures. A recent study identified a temperature sensitive Na+
channel, Nav1.2, that is specifically expressed in the axon initial segment of neurons and may contribute to febrile seizure mechanisms (Thomas et al., 2009
). Nav1.1 channels, mutated in some cases of severe myoclonic epilepsy of infancy (Oakley et al., 2009
), would also be interesting to explore.
We also observed changes in synaptic activity during hyperthermia. CA3 pyramidal cells have recurrent excitatory synapses (Amaral et al., 1990
; Ishizuka et al., 1990
; Li et al., 1994
; Bains et al., 1999
; McIntyre and Schwartzkroin, 2008
), so enhanced glutamate release might be particularly important for seizures generated there. Synaptic activity in GIN cells was noticeably increased at hyperthermic temperatures in both CA1 and CA3. However, we found that experimental blockade of ionotropic glutamate and GABA receptor-mediated transmission did not significantly alter the rates of hyperthermia-induced increases in spontaneous spiking. This does not mean that synaptic activity and connectivity are unnecessary for the generation of febrile seizures, but it does suggest that the high temperature increases the intrinsic excitability of pyramidal cells and interneurons.
High temperatures can increase tissue metabolic rate and decrease oxygen solubility, potentially leading to hypoxia, particularly in tissue preparations in vitro
where oxygenation may already be marginal (Hajos and Mody, 2009
). Hypoxia-induced injury and seizures are well-described phenomena, especially in the developing brain (Jensen and Wang, 1996
; Sanchez and Jensen, 2005
). One concern might be that the effects of hypoxia confound the results we described here. However, the time-course of the effects of hypoxia vs. hyperthermia are distinctly different. Membrane depolarization does not occur until about 5 min after the start of acute hypoxia (Jiang and Haddad, 1992
; Bhave et al., 2003
; Richard et al., 2010
). We found that hyperthermia induces membrane depolarization within 1 min of exposure, however. Additionally, hypoxia-induced changes such as spreading depression and seizures more strongly affect CA1 than CA3 (Kawasaki et al., 1990
; Kreisman et al., 2000
; Sanchez and Jensen, 2005
). In contrast, we found that hyperthermia more strongly influenced neurons in the CA3 area. We minimized effects of hypoxia by keeping superfusion rates high and measuring neuronal effects only during the first few minutes at high temperature, but we cannot entirely eliminate the possibility that hypoxia contributed to our results. We note that many neurons did recover fully after a test period of hyperthermia.
GIN cells belong to the class of somatostatin-expressing inhibitory interneurons that have been implicated in several aspects of epilepsy. Some studies have suggested that activation of somatostatin-positive interneurons, and the resulting release of somatostatin, have antiepileptogenic effects (Tallent and Siggins, 1997
; Vezzani and Hoyer, 1999
; Tallent and Qiu, 2008
). However, somatostatin knockout mice were only mildly more prone to kainic acid-induced seizures than controls (Buckmaster et al., 2002
). Although GIN interneurons increased their firing rates in response to hyperthermia, their activity did not reduce the firing of CA3 pyramidal cells. Further experiments are needed to explore whether somatostatin release itself has an effect on excitability under febrile conditions. Although decreased IPSCs were noted in response to hyperthermia (Qu et al., 2007
; Qu and Leung, 2008
), direct interneuron recordings were lacking. Both increases and decreases in inhibition may increase overall network excitability via inhibitory-inhibitory interactions. Investigations of inhibitory-inhibitory connections would provide further insight into the hyperthermia-driven circuit dynamics.
Gap junctions can synchronize interneurons very effectively (Galarreta and Hestrin, 1999
; Gibson et al., 1999
; Deans et al., 2001
). In the hippocampus, neuronal gap junctions have been implicated in the generation and maintenance of gamma oscillations (Hormuzdi et al., 2001
; Buhl et al., 2003
), and in various forms of epileptogenesis (Carlen et al., 2000
; Perez Velazquez and Carlen, 2000
; Nemani and Binder, 2005
; Gajda et al., 2006
). However, the role of interneuronal gap junctions in seizure discharges has not been directly tested in the hippocampus. Gap junctions can be modulated by pH and hypoxia (Peracchia, 2004
; Gonzalez-Nieto et al., 2008
; Talhouk et al., 2008
), but the effects of hyperthermic temperatures on neuronal gap junctions have not been investigated. We found that electrical coupling between GIN cells was maintained at hyperthermic temperatures. It may be that the role of gap junctions becomes more important for network synchrony during hyperthermic seizures when chemical transmission appears less effective.
Our results show that hyperthermia significantly increases the intrinsic excitability and spontaneous firing rates of neurons. This alone might lead to the generation of febrile seizures. CA3 neurons were most susceptible to heat, suggesting that the CA3 area is a likely site for the initiation of febrile seizures. To better understand febrile seizure mechanisms and treatment, it will be important to determine why some neuron populations are more sensitive to hyperthermia than others. Temperature-induced changes in excitability may also have beneficial effects. Anecdotal and epidemiological evidence suggests that fevers leading to body temperatures of 1.5–2.5°C above normal can acutely improve the behavior of children with autism spectrum disorders (Cotterill, 1985
; Torres, 2003
; Curran et al., 2007
). The mechanisms of this effect are entirely unknown and indeed may not involve temperature changes directly (Mehler and Purpura, 2009
). Nevertheless, it would not be surprising if the potent effects of modestly high temperatures on neuronal excitability cause significant changes in behavior.
Conflict of interest statement
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.