Functional Role of Inflammatory Cytokines and Antiinflammatory Molecules in Seizures and Epileptogenesis


Address correspondence and reprint requests to Dr. A. Vezzani at Lab Exp Neurol, Head, Department of Neuroscience, Istituto di Ricerche Farmacologiche Mario Negri, Via Eritrea 62, 20157 Milano, Italy. E-mail:


Summary:  Purpose: We investigated the changes in the expression of proinflammatory cytokines and related molecules in the rodent hippocampus after the induction of limbic seizures. We then studied the effects of pharmacologic intervention on the interleukin (IL)-1 system on limbic seizures and the susceptibility to seizures of transgenic mice overexpressing the naturally occurring antagonist of IL-1 (IL-1Ra) in astrocytes.

Methods: Limbic seizures were induced in rodents by intrahippocampal injection of kainic acid or bicuculline methiodide or by electrical stimulation of the hippocampus causing status epilepticus (SE). Seizure activity was recorded by EEG analysis and behavioral observation according to Racine's scale. Cytokine expression in the hippocampus was studied by reverse transcriptase–polymerase chain reaction (RT-PCR) followed by Southern blot quantification of the various messenger RNAs (mRNAs) and by immunocytochemistry.

Results: We found that limbic seizures rapidly and transiently enhanced IL-1β, IL-6, and tumor necrosis factor (TNF)-α mRNA in the hippocampus with a peak effect at 6 h after SE. Immunoreactivity of the various cytokines was increased in glia. The increase of IL-1Ra was delayed because the peak effect was observed at 24 h after SE. Moreover, IL-1Ra was not produced in large excess, as during peripheral inflammation but in a molar ratio to IL-1β of 1:1. Intrahippocampal injection of IL-1β worsened seizure activity, whereas IL-1Ra was a powerful anticonvulsant in various models of limbic seizures. Transgenic mice overexpressing IL-1Ra in astrocytes were less sensitive to bicuculline-induced seizures.

Conclusions: This study shows that limbic seizures in rodents rapidly and reversibly induce proinflammatory cytokines in glia and suggests that changes in the IL-1Ra/IL-1β ratio in brain may represent an effective physiopathologic mechanism to control seizures.

The proinflammatory cytokines and their related molecules are polypeptide hormones first identified as soluble mediators in the immune system (1). Recent evidence has shown that inflammatory cytokines and their receptors are present in various forebrain areas, and both neurons and glia are local sources of synthesis (2–4).

The involvement of cytokines in the pathogenesis of epilepsy has been recently suggested by the evidence that limbic seizures increase messenger RNA (mRNA) of inflammatory cytokines in rodent forebrain (5–9). In addition, the release of tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) from rat hippocampal slices is enhanced by seizures (10), and an increase in IL-1α immunoreactivity has been found in human epileptic tissue (11).

Kanemoto et al. (12) recently described a polymorphism in the promoter region of the IL-1β gene associated with temporal lobe epilepsy in patients with hippocampal sclerosis. This polymorphism appears to confer an enhanced ability to produce IL-1β after a proper stimulus.

The present study reports our findings on the cell populations in the rodent hippocampus synthesizing cytokines after limbic seizures, the time course of these changes, and the functional consequences that brain cytokines have on seizure activity.


Experimental animals

Male Sprague–Dawley rats (caesarean derived-caesarean originated barrier sustained; Charles River, Italy; 250–280 g) were used.

The IL-1 receptor antagonist (IL-1Ra) overexpressing mouse strain (B6/CBA F1) was generated and characterized as previously described (13). In brief, the secretable form of human recombinant (hr)IL-1Ra is expressed in astroglia under the glial fibrillary acidic protein (GFAP) promoter, leading to 10- to 23-fold increases of basal IL-1Ra levels in the central nervous system (CNS). In our experiments, we used the strain GILRA2, showing an increase of ∼15-fold in cerebrospinal fluid (CSF) IL-1Ra. In our experiments, we used male adult mice of 25–30 g.

Animals were housed with free access to food and water, under a 12-h light/dark cycle with constant temperature (21–23°C) and humidity (60 ± 5%). Procedures involving animals and their care were conducted in conformity with institutional guidelines that are in compliance with national and international laws and policies.

Surgical procedures and drug injection

For intrahippocampal injections, mice or rats were anesthetized with Equithesin (1% phenobarbital/4% chloral hydrate; 3.5 ml/kg, i.p.), and an injection-guide cannula (22 gauge) was unilaterally positioned on top of the dura. For simultaneous EEG recordings, nichrome-insulated bipolar depth electrodes were implanted in the injected and contralateral hippocampus [coordinated from bregma: (mm) (nose bar 0); AP, 1.9/2.5; L, ±1.5/2.5; 1.5/2.9 below dura), and a ground lead was positioned over the nasal sinus. Cannula and electrodes were connected to a multipin socket and secured to the skull with acrylic dental cement. Animals were allowed 3–5 days to recover from the surgical procedure before the start of the study. All the pharmacologic experiments were carried out between 9 a.m. and 1 p.m.

Kainic acid or bicuculline methiodide (Sigma, St. Louis, MO, U.S.A.) was dissolved in phosphate-buffered saline (pH 7.4) and unilaterally injected in the dorsal hippocampus at the doses reported in the Results section. The dose of convulsant drugs was adjusted in the various experiments to induce seizure activity of comparable severity in 100% of naive animals with ≤20% mortality.

Human recombinant (hr)IL-1Ra (inhibitory activity in murine thymocyte proliferation assay, 1.7 × 106 U/mg) and hrIL-1β (bioactivity in murine thymocyte stimulation assay, ∼3 × 107 U/mg; D. Boraschi, Dompé, L'Aquila, Italy) were dissolved in sterile saline and injected in 0.5 μl intrahippocampally or 3 μl intracerebroventricularly.

Animals injected with corresponding amounts of heat-inactivated cytokines before the convulsant stimulus were used as controls.

Electroencephalographic (EEG) recordings

EEG recordings were carried out in the hippocampus of freely moving animals as previously described (14,15). A 15-min baseline recording was done to assess the spontaneous EEG pattern. Drugs were injected slowly (60 s) through an injection needle (28 gauge) that extended below the guide cannula to reach the dorsal hippocampus. The needle was left in place for an additional minute before removal. The EEG recordings were made continuously for 90–120 min after drug injection.

Ictal episodes were characterized by high-frequency and/or multispike complexes and/or high-voltage synchronized spikes simultaneously occurring in both hippocampi (14,15).

Motor seizures

Motor seizures induced by bicuculline methiodide in mice were visually observed by two independent investigators unaware of the identity of the experimental groups. They were quantified in experimental and matched control mice using the following parameters: (a) the time to onset of the first seizure (either clonic or tonic); (b) the duration of the clonic and tonic components of seizures; (c) the number of motor seizures; and (d) the number of mice showing motor seizures.

Clonic seizures consisted of a rhythmic contraction of forelimbs and/or hindlimbs and/or the back muscles. A tonic seizure consisted of a rigid extension of the fore-and/or hindlimbs with or without loss of posture. The time of observation was of 120 min.

Self-sustained limbic status epilepticus (SE)

Rats underwent “continuous” hippocampal stimulation as previously described in detail (16). In brief, animals were implanted bilaterally with bipolar insulated nichrome electrodes (60 μm) in the CA3 region of the ventral hippocampus. Bilateral screw electrodes were also positioned over the parietal cortex. One week after electrode implantation, the afterdischarge threshold was determined for each rat, and only animals with threshold values ≤250 μA were studied further. Then the rats were stimulated in the left or right ventral hippocampus at 400 μA (50-Hz, 1-ms biphasic square waves in 10-s trains applied every 11 s) to override postictal refractoriness, for 60 min, according to a previously published protocol (17). Chart recordings of EEG activity were obtained before, during (every 10 min for 1 min in the absence of electrical stimulation; i.e., the “stimulus off” period) and after stimulation, and the EEG responses were visually inspected. Behavior was observed during the stimulation protocol and scored using Racine's scale (18).

RNA extraction and RT-PCR analysis

The experimental animals and their respective controls (n = 5 for each experimental group) were killed at various times (2 h to 60 days) after the end of the electrical stimulation. Rats were decapitated, and their hippocampi were rapidly dissected out, immediately frozen on dry ice, and kept at –20°C until assay. The stimulated and contralateral hippocampi were analyzed separately. Total RNA was extracted from the hippocampi, and reverse transcriptase–polymerase chain reaction (RT-PCR) analysis was carried out as previously described in detail (15). A portion of the PCR mixture (7 μl) was electrophoresed in a 1.5% (wt/vol) agarose gel, stained with ethidium bromide, photographed, and transferred to Gene Screen Plus membrane (NEN Life Science Products, Boston, MA, U.S.A.). Membranes were prehybridized in sodium chloride, sodium citrate (5× SSC) containing 1% sodium dodecylsulfate (SDS), 0.5% nonfat milk, and 100 μg/ml of denatured salmon sperm DNA at 37°C for 4 h. Oligonucleotide probes internal to the PCR primers were radiolabeled with [32P]-adenosine triphosphate (ATP) by T4 polynucleotide kinase (Promega, Madison, WI, U.S.A.). All labeled probes were purified through G-50 Sephadex Quick Spin Columns (Boehringer Mannheim SpA, Monza, Italy). Membranes were hybridized overnight at 37°C, washed first in 6× SSC at room temperature, and then in 3 M tetraethyl ammonium chloride, 50 mM Tris HCl, pH 8, and 0.1% SDS at 58°C. Subsequently, membranes were washed in 2× SSC and dried before autoradiography.

Densitometric analysis

The relative intensity of the bands on autoradiograms was measured with an IBAS 2 image analyzer (Kontron-Zeiss, Milano, Italy). We determined the ratio between the optical density (OD) of the bands corresponding to cytokines and that of the corresponding β-actin bands (used as an internal standard).

Tissue preparation for immunocytochemistry

Experimental animals and their respective controls (n = 5) were killed 18 h to 60 days after the end of the electrical stimulation or 3 h after kainate injection. Rats were deeply anesthetized with Equithesin and perfused via the ascending aorta with 100 ml cold phosphate-buffered saline (PBS; 50 mM, pH 7.4) followed by ∼300 ml chilled 4% paraformaldehyde in PBS, as previously described (14). After we carefully removed the brains from the skull, they were postfixed in the same fixative for 90 min at 4°C and then transferred to 20% sucrose in PBS and kept there for 24 h at 4°C. The brains were then immersed in –50°C isopentane for 3 min. They were stored in sealed counting vials at –70°C until use.

Serial cryostat sections (40 μm) were cut horizontally and coronally from all brains throughout the septal and temporal extension of the hippocampus. Adjacent sections were collected for staining of IL-1β, TNF-α, and IL-6. Immunostaining was done as previously described (14,15).

Sections were taken at comparable anteroposterior and mediolateral levels in controls and epileptic rats. Nissl staining was carried out with cresyl violet in representative sections of controls and epileptic rats to check for electrode placement and to assess neurodegeneration.

Statistical analysis of data

Kruskal–Wallis one-way analysis of variance (ANOVA) was used to evaluate the overall change in mRNA expression at different times compared with controls. One-way ANOVA followed by Tukey's test was used in all other experiments.


Effect of limbic seizures on endogenous cytokine expression

Limbic seizure induced by status epilepticus (SE) caused a time-dependent increase of IL-1β, IL-1Ra, IL-6, and TNF-α mRNA in the hippocampus (Fig. 1). The maximal increase was reached at 6 h for IL-1β (445%), IL-6 (405%), and TNF-α (264%), whereas at 24 h, IL-1Ra was 494%, as assessed by Southern blot quantification. All cytokine mRNAs returned to control level by 7 days except for IL-1β mRNA, which was still increased by 241% 60 days after SE.

Figure 1.

Cytokines messenger RNA (mRNA) expression in the rat hippocampus at different times after SE. Time 0, control group (rats implanted with electrodes but not stimulated). The amount of cytokine mRNA is expressed as the ratio of densitometric measurements (OD) of the samples to the corresponding internal standard (β-actin). Data expressed as the mean ± SEM of five rats. P < 0.01 for each cytokine by Kruskal–Wallis.

Immunocytochemical analysis at selected times after SE showed that cytokines were enhanced in glial cells (Fig. 2) in areas of intense microglia activation (not shown; see 14,15). Scattered neurons were immunopositive for IL-1β and IL-6 in control sections.

Figure 2.

High-magnification photomicrographs showing interleukin (IL)-1β, IL-6, and tumor necrosis factor (TNF)-α immunoreactivity, respectively, in control sections from rats implanted with electrodes but not stimulated (A, C, D) and 18 h after SE (B, D, F). Arrows depict glia, and arrowheads depict neurons. Note the enhanced immunoreactivity in glial cells in B, D, and F after SE compared with respective control panels (A, C, D). GC, granule cells; CA3, CA3 pyramidal cells; sl, stratum lacunosum; ml, stratum moleculare; h, hilus. Scale bar, 50 μm.

Sixty days after SE, CA3c pyramidal neurons and hilar interneurons showed enhanced immunoreactivity for IL-1β (not shown; see 15).

Two to three hours after intrahippocampal injection of kainic acid or bicuculline methiodide, IL-1β staining was significantly enhanced in glia (not shown; see 14,19).

Seizure modulation by pharmacologically affecting the IL-1 system

Intrahippocampal injection in rats of 1 ng/0.5 μl hrIL-1β significantly enhanced by twofold on average the time spent in EEG seizures induced by a local application of 0.2 nmol kainic acid (Table 1). This effect was blocked by co-injection with IL-1Ra or (R)-CPP, a selective blocker of N-methyl-d-aspartate (NMDA) receptors (not shown, see 14). Intraventricular application of 0.1 μg hrIL-1Ra reduced by two- to fourfold on average the number of seizures and the time in seizure activity induced by kainate when given 10 min before and 10 min after the convulsant (Table 1).

Table 1.  Seizure modulation
DrugOnset (min)Number of seizuresTime in seizures (min)
  • Data expressed as the mean ± SEM. Interleukin (IL)-1β (1 ng/0.5 μl; n = 9) or IL-1Ra (0.1 μg/0.5 μl; n = 9) was injected intrahippocampally 10 min before 0.2 nmol/0.5 μl kainic acid. IL-1Ra was injected i.c.v. 10 min before and 10 min after kainate. Controls (n = 37) were injected with the respective heat-inactivated cytokines before kainic acid.

  • a

     p < 0.05.

  • b

     p < 0.01 vs. control by Tukey's test.

Control11.4 ± 2.416.0 ± 2.024.0 ± 2.5
IL-1β12.3 ± 2.916.0 ± 1.052.3 ± 4.9b
IL-1Ra21.7 ± 6.28.0 ± 2.0a13.4 ± 2.6b

hrIL-1Ra (0.5 μg/3 μl) intracerebroventricularly administered 10 min before and 20 and 40 min after the beginning of electrical stimulation, significantly reduced behavioral seizure without modifying the epileptic activity in the stimulated hippocampus. Thus, the behavioral score (18) after the second and third IL-1Ra injection was 1.8 ± 0.3, whereas in control rats receiving the heat-inactivated cytokine, it was 2.8 ± 0.3 (p < 0.01; n = 12–13 rats).

We inhibited MEK1 protein kinase (P38 MAP kinase) activity by impairing its phosphorylation with SB 253080. This kinase is crucially involved in the effects of IL-1β on synaptic transmission (20,21). SB 253080 (400 μM in 0.5 μl) or its vehicle (10% DMSO), given intracerebroventricularly to rats 5 min before kainate, delayed by 80% on average the time to onset of EEG seizures [control, 5 ± 1 min; SB, 9.2 ± 1.6 (n = 10); p < 0.05] without modifying the other seizure parameters.

Transgenic mice overexpressing IL-1Ra

Unilateral intrahippocampal injection of 50 ng hrIL-1β 5 min before a local injection of 0.06 nmol bicuculline methiodide in B6/CBA wild-type mice worsened the behavioral seizure pattern by reducing the latency to clonus by 52% (p < 0.01) and increasing the time in clonic and tonic seizures by 47 and 250%, respectively (p < 0.01) compared with mice receiving the heat-inactivated cytokine.

Conversely, unilateral injection of 0.3 nmol hrIL-1Ra effectively reduced behavioral convulsions induced by 0.08 nmol bicuculline methiodide by delaying the onset time to tonic–clonic seizures by two- to fourfold (p < 0.01) and reducing the time spent in clonic and tonic seizures by 50 and 90%, respectively (p < 0.01). IL-1Ra was ineffective on seizures when injected in mice lacking IL-1 type I receptors (not shown, see 19).

Table 2 shows decreased susceptibility to bicuculline methiodide–induced seizures in transgenic mice selectively overexpressing IL-1Ra in glia. We used the strain GILRA2, showing an increase of ∼15-fold in CSF IL-1Ra. The duration of clonic and tonic seizures was reduced by 30 and 76%, respectively. A significant delay was found in the time to onset of clonic seizures. Simultaneous EEG recordings of hippocampal epileptic activity showed that the duration of ictal episodes was significantly reduced by 67% on average (p < 0.05) in GILRA2 mice compared with wild-type littermates.

Table 2.  Susceptibility to bicuculline methiodide-induced seizures in transgenic mice overexpressing the human soluble form of IL-1Ra in astrocytes (GILRA2)
 Behavioral seizures (min)EEG seizures (min)
  • Data expressed as the mean ± SEM (n = 7–11). Bicuculline methiodide (0.24 nmol in 0.5 μl) was unilaterally injected in the dorsal hippocampus of B6/CBA GILRA2 mice overexpressing the human soluble form of IL-1Ra in astrocytes or in their wild-type littermates.

  • a

     p < 0.05.

  • b

     p < 0.01 vs. wild-type by Student's t test.

Wild-type2.9 ± 0.46.9 ± 1.32.9 ± 0.768.9 ± 4.22.0 ± 0.88.4 ± 1.5
GILRA22.5 ± 0.544.7 ± 16b0.7 ± 0.3a22.4 ± 7.9b7.7 ± 4.42.8 ± 2.6a


Our study shows that limbic seizures in rodents rapidly and reversibly induce inflammatory cytokines in glia in the hippocampus.

The early induction (3–18 h) after the onset of seizures strictly overlaps with the hippocampal areas showing activated microglia cells. Astrocytes are likely to contribute to cytokine production in the late phases after seizures (18 h onward) when they appear to be functionally activated (see GFAP and morphologic changes, not shown).

With low convulsant doses of intrahippocampal bicuculline in rats as a nonlesional model of seizures (see 14), we found a milder and less widespread induction of IL-1β than that observed after kainate or SE (present study), suggesting that degenerating neurons play a significant role in priming inflammatory cytokine production.

Our pharmacologic findings using different models of limbic seizures in rats and mice consistently showed that intrahippocampal application of IL-1β has proconvulsant actions, whereas IL-1Ra acts as a powerful anticonvulsant agent. It is worth noting that the maximal expression of IL-1Ra in the rodent brain occurred later than that of the inflammatory cytokines (24 vs. 6 h), and the molar ratio between IL-1β and its receptor antagonist was ∼1:1. This is different from what occurs during peripheral inflammatory processes, in which circulating IL-1Ra levels increase simultaneous with and are in large excess (10- to 100-fold) of those of IL-1β(22). Thus, the brain appears to be less effective at rapidly dampening IL-1β actions than the periphery.

With regard to the mechanism(s) involved in the facilitation of seizures by IL-1β, our data suggest that this cytokine plays a permissive role on glutamatergic neurotransmission. Thus, selective blockade of NMDA receptors impairs the proconvulsant effect of this cytokine. It has been shown that IL-1β markedly attenuates astrocytic glutamate uptake (23). It may also affect NMDA-receptor function because IL-1β receptors are associated with signal-transduction pathways that are known to affect the response of NMDA receptors to endogenous ligands (1). IL-1β may also act by impairing γ-aminobutyric acid (GABA)-mediated neurotransmission. Zeise et al. (24) and Wang et al. (25), respectively, reported decreased synaptic inhibition in CA3 and decreased peak magnitude of GABA-elicited currents in hippocampal neurons by IL-1β at concentrations encountered in pathophysiologic conditions.

Finally, the effect of IL-1β may involve other cytokines as well. IL-1 induces the synthesis of IL-6 and TNF-α in astrocytes and microglia (1–3), and many actions of IL-1 in the CNS are mediated by these cytokines (25).

Our results are only apparently at variance with the electrophysiologic evidence showing that IL-1 inhibits long-term potentiation (LTP) (25–28). The effect of inflammatory cytokines depends on the functional state of neurons, their site of release, and particularly their concentration and the duration of tissue exposure.

It is worth noting that both the lack of IL-1 type I receptors (see 19) and the inhibition of p38 MAPK (present study) significantly delay the onset of EEG and/or behavioral limbic seizures, thus supporting a modulatory role of the endogenous IL-1 system in ictogenesis and seizure propagation.

Moreover, the anticonvulsant activity of IL-1Ra and the reduced susceptibility to seizures in mice overexpressing IL-1Ra in astrocytes strongly support a functional role of IL-1β released from glia during seizures.

Changes in the IL-1Ra/IL-1β ratio in the brain may therefore represent an effective physiopathologic mechanism to control seizures. IL-1Ra seems to be a well-tolerated endogenous protein that is highly inducible (22), so pharmacologic means that increase this ratio may be a novel strategy for inhibiting seizures.

Acknowledgment: This work was supported by Telethon Onlus Foundation, GP0285/01.

We thank Prof. K. Iverfeldt (University of Stockholm, Stockholm, Sweden) and Prof. T. Bartfai (Scripps Research Institute, La Jolla, California, U.S.A.) for their invaluable contribution to this study.

We also thank Mr. Simone Agosteo for his help in editing the manuscript.