Stress, the hippocampus, and epilepsy

Authors


Address correspondence to Marian Joëls, Ph.D., SILS-CNS, University of Amsterdam, Kruislaan 320, 1098 SM Amsterdam, The Netherlands. E-mail: m.joels@uva.nl

Summary

Stress is among the most frequently self-reported precipitants of seizures in patients with epilepsy. This review considers how important stress mediators like corticotropin-releasing hormone, corticosteroids, and neurosteroids could contribute to this phenomenon. Cellular effects of stress mediators in the rodent hippocampus are highlighted. Overall, corticosterone—with other stress hormones—rapidly enhances CA1/CA3 hippocampal activity shortly after stress. At the same time, corticosterone starts gene-mediated events, which enhance calcium influx several hours later. This later effect serves to normalize activity but also imposes a risk for neuronal injury if and when neurons are concurrently strongly depolarized, for example, during epileptic activity. In the dentate gyrus, stress-induced elevations in corticosteroid level are less effective in changing membrane properties such as calcium influx; here, enhanced inhibitory tone mediated through neurosteroid effects on γ-aminobutyric acid (GABA) receptors might dominate. Under conditions of repetitive stress (e.g., caused from experiencing repetitive and unpredictable seizures) and/or early life stress, hormonal influences on the inhibitory tone, however, are diminished; instead, enhanced calcium influx and increased excitation become more important. In agreement, perinatal stress and elevated steroid levels accelerate epileptogenesis and lower seizure threshold in various animal models for epilepsy. It will be interesting to examine how curtailing the effects of stress in adults, for example, by brief treatment with antiglucocorticoids, may be beneficial to the treatment of epilepsy.

Individuals are continuously exposed to potential disturbances of the equilibrium in essential body functions. These potential disturbances (stressors) may result in a subjective state of stress, leading to a characteristic “stress response,” which aims to restore homeostatic control, variable to demand (also named allostasis; McEwen, 2007).

One of the first steps in the stress response is the activation of the autonomic nervous system, which gives the individual a means to quickly face a challenge (Fig. 1). This response is not only achieved by alterations of peripheral organ function, but also—indirectly—by raising central levels of noradrenaline, so that alertness is promoted as well as the choice of an appropriate behavioral strategy (Valentino & Van Bockstaele, 2008). Stress also leads to activation of the hypothalamo–pituitary–adrenal (HPA) axis. The initial step in this process is the elevation in the release of corticotropin-releasing hormone (CRH), in response to input from extrahypothalamic sources. This release, in turn, causes the secretion of adrenocorticotropic hormone (ACTH) from the anterior pituitary into the circulation, which subsequently leads to release of corticosteroid hormones (cortisol in humans; corticosterone in rodents) from the adrenal cortex. In general, corticosteroid hormones improve the restorative capacity after stress exposure and prepare the organism for future challenges (De Kloet et al., 2005). The HPA-axis activation is normalized approximately 2 h after stress exposure, through a negative feedback action at the level of the pituitary and hypothalamus. In addition to these steroids of adrenal origin, studies over the last decades have also shown that the brain itself can generate steroids (neurosteroids), which alter the function of nearby cells (Schumacher et al., 2003).

Figure 1.


Exposure to a stressful event is processed in limbic areas, which project to the paraventricular nucleus in the hypothalamus (Hy). From there the autonomic nervous system (ANS) is activated, resulting in the release of adrenaline; this in turn, can indirectly elevate the release of noradrenaline (NA) in brain, from projections originating in the locus coeruleus (LC). Stress also leads to the release of corticotropin-releasing hormone (CRH), which results in pituitary secretion of adrenocorticotropic hormone (ACTH). As a result, corticosterone is released from the adrenal cortex. Corticosterone acts on many peripheral organs but also feeds back on the pituitary and hypothalamus, thereby turning down the activity of the axis. Furthermore, corticosterone reaches extrahypothalamic brain regions such as the prefrontal cortex (P), amygdala (A), and hippocampus (Hi). Inset at top: Within the hippocampus, several stress-related hormones affect cellular function. In this review, the effects of CRH, the neurosteroid allotetrahydrodeoxycorticosterone (THDOC) and corticosterone (acting via mineralocorticoid receptor and glucocorticoid receptor, that is, MR and GR) are highlighted; effects of NA take place via α- and β-adrenoceptors and are mentioned only briefly.

Stress is among the most frequently self-reported precipitants of seizures in patients with epilepsy (Frucht et al., 2000; Spector et al., 2000; Nakken et al., 2005; Haut et al., 2007; Sperling et al., 2008). Controlled studies are scarce, but the fact that soldiers in combat units have a higher incidence of seizures than soldiers working under less stressful conditions seems to support a role of stress in the occurrence of seizures (Moshe et al., 2008). Biochemically there is evidence that cortisol plays a role in seizure control (Gallagher et al., 1984). Indirect evidence for a role of stress hormones in the control of epilepsy also comes from recent work showing that mood disorders—which in most patients are accompanied by hyperactivity of the HPA axis (De Kloet et al., 2005)—may predispose to the development and/or progression of epilepsy (Hesdorffer et al., 2000; Alper et al., 2007). Cortisol and other stress hormones [ACTH, CRH, and deoxycorticosterone (DOC)] also play a role in pediatric epilepsy, for example, infantile spasms (Rogawski & Reddy, 2002; Baram, 2007) and nonepileptic seizures in children (Wood et al., 2004). Indeed, glucocorticoids, ACTH, and steroid manipulations have been shown to be effective in the management of pediatric epilepsies (Hrachovy et al., 1983; Baram et al., 1996; Schmidt & Bourgeois, 2000; Brunson et al., 2002; Gupta & Appleton, 2005). Studies with rodent seizure models have been interpreted to demonstrate the importance of stress for epilepsy/epileptogenesis. However, a superficial reading of the literature suggests that the findings are somewhat confusing and inconsistent, with pro- as well as anticonvulsive effects demonstrated (Table 1). This variability seems to depend on many factors, including the mediators involved in a particular stress response as well as the frequency and period of life in which stressful events are experienced (see subsequent text).

Table 1.   Overview of effects of stress in animal models of epilepsy/epileptogenesis
Hormone/stressEpilepsy modelMain effectReference
  1. ACTH, adrenocorticotropic hormone; ADT, after discharge threshold; ALLO, allopregnanolone; BLA, basolateral amygdala; DEX, dexamethasone; DOC, deoxycorticosterone; ECS, electroconvulsive shock; HIPP, hippocampus; icv, intracerebroventricular; Li–Pilo SE, status epilepticus by lithium–pilocarpine; MES, maximal electroshock; MET, metyrapone (steroid synthesis inhibitor); n.a., not applicable; PTZ, pentylenetetrazole; ST, seizure threshold.

CRH (icv in infant rats)n.a.Seizure activity, not blocked by ACTHBaram & Schultz, 1991, 1995
CRH (icv, adult rats)n.a.High CRH dose increases seizure activity and afterdischargeEhlers et al., 1983
CRH (icv, adult rats)n.a.Recurrent episodes of epileptic activity by CRHMarrosu et al., 1987, 1988
CRH (icv, adult rats)n.a.CRH induces seizure activity, worsened by 3 days DEXRosen et al., 1994
CORTWAG/Rij ratsRapid increase in discharges by CORTSchridde & van Luijtelaar, 2004
CORT (adult mice)KainateGreater convulsion susceptibility, via MRRoberts & Keith, 1994
CORT in adult male ratsIschemia, KainateEnhanced seizure rate, more damage by CORT, protection by METKrugers et al., 1999, 2000;  Smith-Swintosky et al.,1996
CORT in adult ADX female ratsBLA/HIPP kindlingIncreased kindled rates by CORTEdwards et al., 1999
CORT (>1 week, adult male rats)HIPP kindlingIncreased kindled rates by CORTKarst et al., 1999
CORT chronic low dose (rats)Amygdala kindlingIncreased kindled rates by CORT, via MR and GRTaher et al., 2005;  Kumar et al., 2007
CORT, DEX (long-term, adult rats)Cocaine kindlingIncreased kindled rate by DEXKling et al., 1993
DEX antenatally 3 daysHIPP kindling, ECS (P14, 15)Increased kindling threshold, rest unaffectedYoung et al., 2006
DOC in P15 ratsPTZ (15 min after DOC)Rapid anticonvulsant effects lost after pubertyEdwards et al., 2002a, 2002c, 2005
DOC in P15 ratsPTZAnticonvulsant activity via metabolitesPerez-Cruz et al., 2006
ACTH, DOC, DEX, P15 ratsPTZ, MES, HIPP kindlingNo effect of ACTH, DOC anticonvulsant DEX not anticonvulsantEdwards et al., 2002b
DOC (rats, mice)PTZ, kindling, picrotoxinDOC increases seizure threshold via DOC metabolitesReddy & Rogawski, 2002
Water submersion (adult rat)PTZDecreased severity of convulsionsAbel & Berman, 1993
Swim stress (adult mice)PTZAnticonvulsant effect, not via CORT or neurosteroids, role of α2-receptorPericic et al., 2000a, 2000b, 2001a
Swim stress (adult mice)PTZ, bicuculline, picrotoxin, 4AP, strychnine, KASwim stress raised dose needed for convulsions for all drugs but KAPericic et al., 2001b
Swim stress (adult rat)PTZIncreased seizure thresholdReddy & Rogawski, 2002
Prenatal stressDaily restrain motherIncreased kindled rate in infants and adult males, transiently lowered ADT; no effect postnatal stressEdwards et al., 2002a, 2002b, 2002c
Maternal deprivation, P2-9PTZ, P10-14Exacerbates cognitive impairment and neurodegeneration in adulthoodHuang et al., 2002
Maternal separation, P2-9Li-Pilo SE, P10Greater cognitive deficits and decreased ST; protection by METLai et al., 2006
Maternal separation, P2-14
versus daily handling
Amygdala kindling in adulthood
 
Increased kindled rates in females, no difference in malesSalzberg et al., 2007
 
Adult social isolation (mice)PicrotoxinEnhanced seizure susceptibility via reduced endogenous ALLO levelsMatsumoto et al., 2003
Adult social isolation (rats)Bicuculline, PTZLower ST after bicuculline (not PTZ)Chadda & Devaud, 2004

The various stress mediators have very distinct cellular effects in limbic brain regions. To gain a better insight into the cellular mechanisms underlying the effects of stress and its mediators on epilepsy, this review first provides an update of the cellular actions of each of these potential mediators. The focus is primarily on hippocampal effects of corticosteroid hormones, and to a lesser extent on the rapidly acting CRH and metabolites of deoxycorticosterone (DOC) (neurosteroids); for extensive reviews on cellular actions of these rapidly acting stress hormones, the reader is referred to recent overviews (Belelli et al., 2006;Gallagher et al., 2008). Actions of sympathetic mediators such as noradrenaline will not be addressed, although it is well-established that hippocampal cell activity in general is enhanced via β-adrenoceptor activation (e.g., Stanton & Sarvey, 1985; Madison & Nicoll, 1986; Heginbotham & Dunwiddie, 1991; Gereau & Conn, 1994) and suppressed via α-receptors (Bergles et al., 1996; Scheiderer et al., 2004; Otmakhova et al., 2005). In agreement, pro- as well as anticonvulsant actions by noradrenaline via β- and α-receptors, respectively, have been described (e.g., Mueller & Dunwiddie, 1983; Jurgens et al., 2005). The overall effect of this monoamine will, therefore, depend largely on the circulating levels and thus specific receptor activation; in general, the net effect of noradrenergic compounds seems to be amelioration of epilepsy and epileptogenesis (for detailed review, see Giorgi et al., 2004).

In the final part of the review, an attempt is made to understand how the cellular effects of the various stress hormones may contribute to the effects of stress on epilepsy/epileptogenesis in rodents, and by inference in humans.

Cellular Effects of Stress Mediators

CRH

The 41-amino acid peptide CRH acts through G protein–coupled receptors, and has been reported to change neuronal function in a rapid and reversible manner. In the hippocampus, CRH almost without exception promotes excitatory transmission. Thus, in rat CA1 cells, the hormone increased spontaneous discharges and suppressed the hyperpolarization after bursts of action potentials (Aldenhoff et al., 1983). Particularly in infant rats, enhanced synaptic field-responses (Hollrigel et al., 1998) and bicuculline-induced bursts were observed (Smith & Dudek, 1994). In mice, CRH was shown to prime long-term potentiation (LTP) in the CA1 area (Blank et al., 2002), through strain-specific intracellular signaling pathways (Blank et al., 2003).

Comparable excitatory effects were also observed in the other hippocampal sub-areas. In rat CA3 neurons, CRH (particularly at higher concentrations) induced a small depolarization (Aldenhoff et al., 1983; Hollrigel et al., 1998). Similar to the effects in CA1 cells, the hyperpolarization after a burst of action potentials was decreased. The number of action potentials per burst and frequency of spontaneous excitatory postsynaptic currents were enhanced, and CRH caused large synchronized bursts in response to mossy fiber stimulation in the presence of bicuculline (Hollrigel et al., 1998). In the dentate gyrus, LTP was found to be facilitated by CRH (Wang et al., 1998, 2000).

In contrast to these excitatory effects of CRH, very recent studies have shown that CRH suppresses N-methyl-d-aspartate (NMDA)–induced currents in rat hippocampal neurons, via CRH-R1 (Sheng et al., 2008). This effect may relate to earlier reports of neuroprotective effects via CRH-R1 in hippocampal cultures (Facci et al., 2003). To what extent this protection is specific to the cells examined in these cultures awaits further examination.

Neurosteroids

Two decades ago it became evident that steroids can also be synthesized de novo in the brain in sufficient amounts to affect local neurotransmission (Schumacher et al., 2003). In particular metabolites of progesterone, such as 5α-pregnan-3α-ol-20-one, have been extensively studied and were found to potentiate the actions of γ-aminobutyric acid (GABA) via GABAA receptors (Rhodes et al., 2004; Belelli et al., 2006). Here the cellular actions of another steroid DOC and its metabolite allotetrahydrodeoxycorticosterone (THDOC) will be highlighted.

The levels of THDOC are normally low (1–5 nm), but can increase significantly after stress, up to levels as high as 15–30 nM (Reddy & Rogawski, 2002; Reddy, 2006). This is a range that was shown to cause within 30 min an upregulation of GABAA receptor δ-subunits in the dentate gyrus (Reddy & Rogawski, 2002); in the CA1 area where δ-subunits are expressed at very low levels, this effect of stress seems less relevant. Of interest to note, δ-subunits are located extrasynaptically and play a major role in tonic inhibition of dentate granule cells. In accordance, THDOC was reported to facilitate tonic inhibition in the dentate in the nanomolar range, whereas it affects phasic (synaptic) GABAergic signals only at a much higher concentration (Stell et al., 2003). In agreement, very high doses of THDOC largely enhance GABA-induced chloride currents in cultured hippocampal cells, and in fact can cause bicuculline-sensitive chloride currents even in the absence of GABA (Reddy & Rogawski, 2002). THDOC not only facilitates inhibitory responses, but also reduces the slope of the field excitatory postsynaptic potential (Stell et al., 2003). These data support that after stress, neurosteroids such as THDOC can shift the excitation/inhibition balance in the dentate gyrus toward more inhibition (Maguire & Mody, 2007).

Corticosterone

Nuclear receptors

In contrast to the mediators previously discussed, corticosterone so far is not thought to act through ionotropic or G protein–coupled receptors in the rodent hippocampus. Instead, corticosteroid hormones in brain act via two nuclear receptor types (De Kloet et al., 2005; McEwen, 2007): (1) the mineralocorticoid receptor (MR), which has a high affinity (subnanomolar range) for the endogenous rodent steroid corticosterone and is highly expressed in limbic areas such as the hippocampus; and (2) the glucocorticoid receptor (GR) with a 10-fold lower affinity for corticosterone, a receptor that is much more ubiquitously expressed but also has very high expression levels in the CA1 hippocampal area and the dentate gyrus. The difference in affinity has important implications for the receptor occupation throughout the day. Thus, corticosteroid hormones are released in ultradian pulses of 1–2 h duration (Young et al., 2004). The peak of these pulses varies in a circadian pattern, with high amplitudes just before the onset of the active period (morning in humans, evening in rats and mice) and low levels toward the end of that period. Stress leads to a premature and stronger pulse of corticosterone. In view of the high affinity, MRs will already be substantially occupied even with low hormone levels, such as circulate at the interpulse intervals (Conway-Campbell et al., 2007). By contrast, GRs are only partly activated under rest, but are substantially activated after stress or after an ultradian pulse with large amplitude. It is particularly this substantially activated GR state that could mediate delayed effects of stress on epilepsy-related processes.

The two corticosteroid receptor types belong to the family of nuclear receptors (Pascual-Le Tallec & Lombes, 2005; Zhou & Cidlowski, 2005). Upon binding of the hormone to the receptor, the complex translocates to the nucleus and regulates the transcription of responsive genes, an estimated 1–2% of the total genome (Morsink et al., 2006). As a consequence, corticosteroids affect the cellular protein content—also of proteins involved in neuronal excitability—with a delay of at least one hour, so that neuronal function is slowly but persistently altered. Therefore, corticosteroid hormones act primarily in a different (slower) time-domain than the hormones discussed previously, although it has been realized in recent years that corticosterone can also quickly alter neuronal function via a nongenomic mechanism (Tasker et al., 2006; Joëls et al., 2008).

Rapid effects of corticosterone in the CA1 area

Occasionally, rapid effects of corticosterone in the hippocampus have been described for several decades (e.g., Vidal et al., 1986; Dubrovsky et al., 1993), but only recently the intracellular signaling pathway has been resolved. In the CA1 area corticosterone quickly enhances the frequency (but not amplitude) of miniature excitatory postsynaptic currents (mEPSCs), which each represent the spontaneous release of one glutamate-containing vesicle (Karst et al., 2005); spontaneous inhibitory currents are not affected (Olijslagers et al., 2008). It was argued that the effect on mEPSC frequency results from an enhanced release probability of the vesicles, which is in line with in vivo microdialysis observations (Venero & Borrell, 1999). This effect critically depends on the MR gene and seems to be accomplished through a presynaptic signaling pathway involving ERK1/2 phosphorylation (Karst et al., 2005; Olijslagers et al., 2008). In addition to this presynaptic effect, corticosterone also induces a postsynaptic MR-dependent rapid reduction of the K-conductance IA. Collectively, these quick and rapidly reversible actions of corticosterone could signify that shortly after stress exposure corticosterone—in addition to other rapidly acting stress hormones such as noradrenaline and CRH—can promote excitatory transmission in the CA1 hippocampal area (Joëls et al., 2006, 2008). In line with these observations, it was found (in vitro) that corticosterone quickly facilitates LTP in the CA1 region induced by high frequency stimulation of the Schaffer collaterals (Wiegert et al., 2006).

Slow effects of corticosterone on ion currents in the CA1 area

Most corticosteroid effects, however, develop much slower. One of the conspicuously altered cell properties is the high-voltage activated calcium current (Kerr et al., 1992; Karst et al., 2000). Exposure of hippocampal CA1 cells to a brief pulse of corticosterone causes several hours later a GR-dependent long-lasting increase in L-type Ca-current amplitude (Chameau et al., 2007). It appeared that the number of L-type channels in the membrane rather than the single-channel conductance is increased. The enhancement of high-voltage activated calcium current amplitude involves DNA binding of GR homodimers (Karst et al., 2000). After chronic stress, the increase in sustained Ca-current amplitude is seen even when corticosteroid levels are low (Karst & Joëls, 2007).

It is notable that the enhanced influx of calcium through high-voltage activated calcium channels appears to be linked to a less efficient calcium efflux (Joëls et al., 1998; Bhargava et al., 2002). This would predict that processes critically depending on the intracellular calcium level are affected by corticosteroids. In accordance, corticosterone via GRs slowly and long-lastingly increases cell-firing frequency accommodation during a brief period of depolarization, a cell property that depends on activation of a calcium-dependent K-conductance (Joëls & De Kloet, 1989; Kerr et al., 1989).

The functional implications of corticosteroid actions on calcium (dependent) signaling could be manifold (Fig. 2). Under “normal” circumstances, it could help to gradually attenuate hippocampal firing after the initial arousal shortly after stress through rapidly acting stress hormones (including corticosterone). This would fit with the general concept of corticosteroids slowly promoting the restorative capacity of the body. However, in the long term, enhanced calcium exposure could also impose an added risk factor on hippocampal CA1 cells, particularly when these cells experience a prolonged period of excitation during which L-type calcium channels stay open, such as may occur during concurrent seizure activity. Similarly, a period of repetitive (chronic) stress—for example, as a result of frequent unpredictable seizures—will enhance the vulnerability of CA1 neurons.

Figure 2.


Corticosterone levels rise after stress, but are gradually normalized because of the negative feedback action of corticosterone at the level of the pituitary gland and the hypothalamus (top). The text below the top graph summarizes the main effects of corticosterone (1) shortly after stress (i.e., when hormone levels are high) and (2) several hours later (via delayed gene-mediated pathways), for the CA1/CA3 hippocampal region and the dentate gyrus. For each region and time-domain the putative consequences for seizure activity are hypothesized in bold. Text at the bottom formulates what the overall implication of stress hormone actions might be during seizure activity, after acute or chronic stress conditions.

Slow effects of corticosterone on neurotransmitter actions in the CA1 area

Corticosteroid hormones do not only target genes involved in ion channel function, but also influence the efficacy of neurotransmitters. Slow gene-mediated actions on transmission through glutamate, the main excitatory transmitter in the CA1 region, have been addressed in a number of studies, but the data so far are not unequivocal. Most studies report a steady glutamatergic transmission under conditions of predominant MR activation, which is suppressed when GRs become activated (see for review Joëls et al., 2007). It is not clear to what extent the suppression of excitatory transmission is really gene-mediated, because some effects have been observed within 20 min (e.g., Vidal et al., 1986), which is rapid for a mechanism involving transcription and translation. Slow corticosteroid influences on GABAergic transmission have not been examined extensively.

Although the actions of corticosterone on low-frequency synaptic stimulation in the CA1 area are thus not fully understood, the effects on high-frequency stimulation are very consistent. In the CA1 area, corticosterone exerts a delayed suppressive effect on the induction of NMDA-receptor dependent forms of LTP (extensively reviewed in Kim & Diamond, 2002; Diamond et al., 2005); potentiation linked to activation of voltage-dependent calcium channels (which occurs with extremely strong stimulation protocols) is enhanced (Krugers et al., 2005). All in all, through GR-dependent signaling the activity of CA1 neurons is likely to be suppressed several hours after stress exposure; at that time it is very difficult to induce LTP. Interestingly, after chronic stress, LTP in the CA1 area is very much impaired, even when corticosteroid levels are low (see for extensive review Joëls et al., 2007).

Corticosteroid hormones also slowly influence monoaminergic transmitter function. For instance, shortly after stress, noradrenaline (NA) levels are elevated (Roozendaal, 2003). The overall effect on excitability depends on the concentration of NA and the types of receptors that will be activated. Behavioral studies have pointed to the involvement of at least β-receptors in the effects of stress in limbic regions (see e.g., Roozendaal, 2003). Through β1-adrenoceptors, hippocampal CA1 neurons will be activated, since the firing frequency accommodation and afterhyperpolarization amplitude are attenuated via a cAMP-dependent mechanism (Madison & Nicoll, 1986). It was found that GR activation slowly reverses these excitatory actions mediated by the β1-adrenoceptors (Joëls & De Kloet, 1989). Conversely, hyperpolarizing effects mediated via serotonin-1A receptors are promoted several hours after GR activation (Beck et al., 1996; reviewed in Joëls et al., 2007). Both phenomena are compatible with the view that corticosterone via GRs slowly normalizes hippocampal excitability after a period of stress.

Effects of corticosterone in the CA3 area and dentate gyrus

In contrast to the CA1 hippocampal region, cellular corticosteroid actions in the CA3 region and dentate gyrus are less well-investigated and, so far, more equivocal. The effects of corticosterone on electrical properties of CA3 cells largely resemble those described for CA1 neurons. For example, high concentrations of corticosterone, occupying both MRs and GRs, enhance Ca-current amplitude and Ca-dependent phenomena similar to those seen in the CA1 region (Kole et al., 2001). Moreover, like the CA1 area, predominant MR activation in the CA3 region is associated with efficient LTP (via NMDA-receptors), whereas GR agonists suppress LTP (Pavlides & McEwen, 1999). CA3 neurons, in contrast to CA1 cells, normally display a burst firing pattern of activity. It was shown that predominant MR activation is associated with a higher incidence of bursting cells, whereas high levels of corticosterone result in fewer bursting cells (Okuhara & Beck, 1998). Overall, corticosterone appears to promote excitatory transmission in CA3 cells (as in CA1) via MR, and gradually reverses excitation via GR. Interestingly, CA3 neurons are extremely sensitive to chronic stress. Extensive dendritic retraction has been described in response to chronic stress, although the precise consequences of these dendritic changes for physiologic function are not known (for review see McEwen & Magarinos, 1997). Electrophysiologic investigations showed that after chronic stress, NMDA-receptor mediated responses are enhanced (Kole et al., 2002), whereas LTP is largely impaired (Pavlides et al., 2002).

In the dentate gyrus, a complicating factor in the interpretation of stress effects is that stress (chronic as well as acute) and corticosterone are known to suppress neurogenesis (Fuchs et al., 2006; Lucassen et al., 2006; Mirescu & Gould, 2006; Paizanis et al., 2007). Although the extent of this phenomenon is unlikely to change dentate function after acute stress, long-term exposure to stress (such as may occur during epilepsy) could shift the neuronal composition of the dentate. Because epilepsy by itself triggers neurogenesis (Scharfman, 2004; Parent, 2007), it is not easy to predict what the functional outcome of these opposing processes could be.

Rapid single-cell effects of corticosterone have not been studied in much detail so far in the dentate gyrus. At the network level, though, it was found that stress via MRs nongenomically facilitates the step from transient to stable LTP (Korz & Frey, 2003). This phenomenon is mitogen-activated protein kinase dependent and requires intact input from the amygdala (Ahmed et al., 2006). In general, the effect of stress on LTP in the dentate is very sensitive to the nature, timing, and the severity of the stress, so that next to facilitation of LTP stress has also been described to induce suppression or be ineffective (reviewed in Joëls et al., 2007). In a direct comparison between CA1 and dentate gyrus, stress was found to suppress LTP in the CA1 but enhance it in the dentate (Kavushansky et al., 2006).

Preliminary evidence supports the view that at the single-cell level, dentate granule cells may be less responsive to GR activation than CA1 neurons are (Joëls, 2006). This feature of granule cells was observed (in handled rats) with respect to glutamate responsiveness and calcium current amplitude (Karst & Joëls, 2003; Van Gemert & Joëls, 2006). Why dentate, as opposed to CA1, cells are less responsive to a high dose of corticosterone—despite their very high expression level of GRs—is still unclear. Interestingly, these cells do respond to a pulse of corticosterone with enhancement in glutamate responses and calcium currents after a 3-week period of unpredictable stress (Karst & Joëls, 2003; Van Gemert & Joëls, 2006). Overall (and in the absence of signals recorded in freely moving animals), it is not easy to generalize the effects of stress on dentate gyrus activity. It seems, however, that the slow GR-dependent normalization of neuronal activity seen in the CA1 area several hours after stress is less likely to occur under “normal” circumstances in the dentate gyrus. One could speculate that in the dentate gyrus neurosteroid-mediated enhancement of tonic inhibition is one of the more conspicuous effects after acute stress.

Implications of HPA-axis related stress mediators for hippocampal function

How will stress affect overall hippocampal function? At present, data are scarce regarding the effects of stress or corticosterone on hippocampal cell and network activity in freely moving animals (e.g., Pfaff et al., 1971). However, based on the in vitro observations described in the previous sections, it is possible to make some predictions. Most likely a single episode of stress will initially enhance CA1 and CA3 cell activity, owing to the joint effects of noradrenaline (via β-receptors), CRH, and corticosterone itself (Joëls et al., 2006; Fig. 2). There is even evidence that these stress hormones can facilitate each other’s function (Pu et al., 2007). As an adaptive mechanism, this may help to encode the information that is attached to the stressful event. It also introduces, however, a window of enhanced excitability, which could potentially predispose to or exacerbate epileptic activity.

GR-dependent genomic actions are thought to slowly reverse the enhanced activity, via attenuation of cell firing, reduction of noradrenergic actions, and potentiation of inhibitory 5-HT effects; patterned input reaching the CA1 area at that time will meet a heightened threshold for synaptic strengthening, which can be surpassed only if the input is sufficiently salient. These slow GR-dependent effects then serve as a means to normalize earlier raised activity, restore the prestress situation, and preserve the encoded information. However, this normalization comes at a cost. Although enhanced Ca-influx in the short-term may be adaptive and help to normalize activity, it also imposes an added risk for neurologic damage, especially during periods of enhanced excitation such as may occur during epileptic activity. Exposure to a high Ca-load may particularly occur after repetitive stress, which may thus enhance vulnerability.

Changes in the dentate gyrus after stress are less easy to predict. The role of corticosteroids is probably rather limited, although excitatory actions can become uncovered after a period of repetitive and unpredictable stress. Neurosteroids most likely will suppress activity via enhancement of tonic inhibition.

Effect of stress and stress mediators in animal models of epilepsy

The HPA-axis–related stress mediators have all been tested separately, with respect to their pro- or anticonvulsant activity, in rodent models of epilepsy/epileptogenesis (Table 1). In discussing these results, it is important to point out that in all cases these drugs were administered peripherally or intracerebroventricularly—so that not only hippocampal cells but also neurons in other regions (which may indirectly influence hippocampal function, like cells of the basolateral amygdala) will be affected. Still, the findings can be largely understood from the in vitro cellular actions described in the previous sections.

First, administration of CRH was consistently found to induce seizure activity (Ehlers et al., 1983; Marrosu et al., 1987, 1988; Baram & Schultz, 1991, 1995). The effects of CRH on seizure activity were amplified when CRH administration was combined with several days of pretreatment with the synthetic glucocorticoid dexamethasone (Rosen et al., 1994). This result can be fully explained by the excitatory cellular effects of CRH in all hippocampal subfields (see preceding text). Second, all studies agree that DOC, through its metabolites, has anticonvulsant properties (Edwards et al., 2002b, 2002c, 2005; Reddy & Rogawski, 2002; Perez-Cruz et al., 2006), particularly in young animals (Edwards et al., 2002c). As argued THDOC accomplishes this anticonvulsant activity by promoting tonic inhibition in the dentate gyrus.

As for corticosterone, more discharge activity, greater convulsion susceptibility, and increased seizure rate were reported (in various rodent models), even after a single corticosteroid administration (Roberts & Keith, 1994; Smith-Swintosky et al., 1996; Krugers et al., 1999, 2000; Schridde & van Luijtelaar, 2004). This effect may involve rapid (Schridde & van Luijtelaar, 2004) actions via MR (Roberts & Keith, 1994), as well as enhanced calcium load via GRs, resulting in more neurodegeneration (Krugers et al., 1999, 2000). The remaining studies with corticosterone used some variation of the kindling paradigm; the hormone was (repetitively) administered during part of, or the entire, period of kindling. These studies unequivocally show that kindling rate is increased by corticosterone (Kling et al., 1993; Edwards et al., 1999; Karst et al., 1999; Taher et al., 2005; Kumar et al., 2007); the only exception is a study in which dexamethasone was administered antenatally, leading to increased kindling threshold in infancy (Young et al., 2006). Cellular influences of the hormone treatment were not studied during the kindling period. However, it seems likely that at least the corticosteroid-induced enhanced calcium load plays a role in increasing the kindling rate, since CA1 cells of fully kindled animals (animals that had been treated with corticosterone some weeks earlier, during epileptogenesis) still displayed increased Ca-current amplitudes (Karst et al., 1999).

The results with exogenous administration of these stress hormones are therefore clearcut. But what happens when the hormones are released from endogenous sources, after stress? A limited number of studies have investigated the effect of a single, acute stressor on seizure activity (Abel & Berman, 1993; Pericic et al., 2000a, 2000b, 2001a, 2001b; Reddy & Rogawski, 2002). This type of stimulation mimics the activation that a patient with epilepsy encounters acutely—a single, moderately stressful situation. The animal studies report decreased severity of convulsions and increased seizure threshold shortly after swim stress. It was argued that swim stress suppresses seizure activity through a THDOC-dependent strengthening of GABAergic inhibition in the dentate gyrus (Reddy & Rogawski, 2002; Reddy, 2006). This mechanism may fail after chronic epilepsy, owing to reduced sensitivity and downregulation of δ-subunits (Zhang et al., 2007). It is important to realize, however, that only one type of stress (water exposure) has been explored in these studies, and the effects observed with swim stress cannot be easily extrapolated to other types of stress. For instance, it was found that swim stress (a combination of physical and psychological stress) exerts opposite effects on LTP in the dentate gyrus than a brief period of handling (Korz & Frey, 2003), or a psychological stressor—like novelty exposure—during a particular phase of LTP (Straube et al., 2003). It is, therefore, certainly possible that other acute single stressors exert pro- rather than anticonvulsant influences.

In addition to the effect of a single stressor, the influence of long-term disturbances in the stress system has been examined with respect to animal models for epilepsy. Chronic stress, in general, imposes a risk on the integrity of hippocampal cells and cell function (see for reviews Fuchs et al., 2006; Joëls et al., 2007; Lucassen et al., 2006; Mirescu & Gould, 2006; Paizanis et al., 2007). Likewise, long-lasting aberrations of the stress system worsen the outcome in models of epilepsy/epileptogenesis. For instance, long-term social isolation in adulthood (a model for chronic mild stress) was found to enhance seizure susceptibility and lower seizure threshold (Matsumoto et al., 2003; Chadda & Devaud, 2004). A special type of “chronic stress” is induced by perinatal stress. In all cases reported to date, perinatal stress was found to worsen the outcome of epilepsy/epileptogenesis in infancy, stretching into adulthood (Edwards et al., 2002a; Huang et al., 2002; Lai et al., 2006; Salzberg et al., 2007). The explanation of these findings is complex. Profound stress experienced in this young, vulnerable period is known to lead to hyperactivity of the HPA axis in adulthood (Heim & Nemeroff, 2001; Meaney & Szyf, 2005; Pryce et al., 2005; Maccari & Morley-Fletcher, 2007); therefore, a history of perinatal stress may lead to release of more stress hormones when the individual experiences stress in adulthood. Moreover, a pulse of corticosterone such as occurs after stress may lead to entirely different effects in adults with or without a prior history of early life stress (Champagne et al., 2008). Finally, stress during a vulnerable period of brain development may lead to differences in brain circuits, cellular properties, and synaptic connections, rendering the individual more vulnerable to the onset of epilepsy.

In conclusion, stress hormones alter cellular activity in all hippocampal subfields in a region-specific way. Generally these hormones promote excitatory activity, although the neurosteroid THDOC leads to enhanced tonic inhibition, particularly in the dentate gyrus. During an acute single stress exposure, the balance between the various stress hormones involved may depend on the specific brain regions and circuits involved in that type of stressor; those brain circuits are certainly not limited to the hippocampus but will involve many other regions, for example, the amygdala. Current animal studies are limited in that only one type of stressor has been investigated, and controlled studies in humans are difficult to perform. There is general consensus, however, with respect to the consequences of early life stress or chronic mild stress in adulthood (unrelated to the experience of unpredictable seizures itself) for epilepsy/epileptogenesis. All animal studies show that such conditions of stress increase kindling rate and/or exacerbate neurodegeneration as well as lower seizure threshold. In view of the latter it is not surprising that periods of stress increase the likelihood for seizures to occur, as is indeed self-reported by epilepsy patients (Frucht et al., 2000; Spector et al., 2000; Nakken et al., 2005; Haut et al., 2007; Sperling et al., 2008).

Animal studies investigating the effect of chronic stress have used mildly stressful conditions that are not directly related to epilepsy itself (e.g., social isolation). One should realize, however, that experiencing frequent and unpredictable seizures in adulthood will in and of itself cause a situation of chronic stress to the individual. Consequently, the ongoing process of epileptogenesis and the course of epilepsy (including changes in seizure threshold and the process of neurodegenerative changes) might be negatively influenced by the stress associated with the disease itself—in addition to influences of other stressful exposures. This process may constitute a negative loop, in which periods of stress (e.g., early in life) promote epileptogenesis in predisposed individuals or lower seizure threshold in epilepsy patients, thereby increasing the likelihood of exposure to frequent unpredictable and uncontrollable stress associated with resultant seizure activity, which in turn exacerbates the disease. It will, therefore, be of interest to examine how curtailing this loop by brief treatment with GR-antagonists or corticosteroid-synthesis inhibitors—which appear to exert protective effects in animal models of epilepsy and chronic stress (Smith-Swintosky et al., 1996; Krugers et al., 1999, 2000, 2006; Karst & Joëls, 2007; Kumar et al., 2007; but also see Borowicz et al., 2002)—may be beneficial to the treatment of epilepsy.

Acknowledgment

MJ is supported by grant # 912-04-042 from The Netherlands Organization for Scientific Research NWO.

I confirm that I have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Disclosure: The author has no conflict of interest.

Ancillary