Inflammation Exacerbates Seizure-induced Injury in the Immature Brain


Address correspondence and reprint requests to Dr. Raman Sankar, Department of Pediatrics, Division of Neurology, 22-474 MDCC in CHS, David Geffen School of Medicine at UCLA, Los Angeles, CA 90095, U.S.A. E-mail:


Summary:  We examined the hypothesis that the introduction of an inflammatory agent would augment status epilepticus (SE)-induced neuronal injury in the developing rat brain in the absence of an increase in body temperature. Postnatal day 7 (P7) and P14 rat pups were injected with an exogenous provocative agent of inflammation, lipopolysaccharide (LPS), 2 h prior to limbic SE induced by either lithium-pilocarpine (LiPC) or kainic acid. Core temperature was recorded during the SE and neuronal injury was assessed 24 h later using profile cell counts in defined areas of the hippocampus. While LPS by itself did not produce any discernible cell injury at either age, it exacerbated hippocampal damage induced by seizures. In the LiPC model, this effect was highly selective for the CA1 subfield, and there was no concomitant rise in body temperature. Our findings show that inflammation increases the vulnerability of immature hippocampus to seizure-induced neuronal injury and suggest that inflammation might be an important factor aggravating the long-term outcomes of seizures occurring early in life.

There is a reciprocal relationship between seizures and inflammatory cytokines (Vezzani and Granata, 2005). On the one hand, status epilepticus (SE) induced by kainic acid (KA) enhanced the expression of interleukin-1β (IL-1β) (Minami et al., 1990; 1991; Eriksson et al., 2000), interleukin-6, and tumor necrosis factor-α (TNF-α) (Vezzani et al., 1999; De Simoni et al., 2000). On the other hand, IL-1β prolonged KA-induced seizures (Vezzani et al., 1999). IL-1β also lowered the threshold for hyperthermic seizures, and even produced seizures by itself (Dube et al., 2005). Production of cytokines has also been demonstrated in response to other central nervous system (CNS) insults such as ischemia or trauma (Allan and Rothwell, 2001). Proinflammatory cytokines may contribute to neuronal injury by acting in conjunction with activation of ionotropic glutamatergic receptors (Chao et al., 1995; Lawrence et al., 1998; Viviani et al., 2003; Cai et al., 2004; Bernardino et al., 2005). The neurodegenerative action of IL-1β when administered in combination with AMPA has been attributed to the increased seizure activity. Other cytokines, such as IL-6, may be neuroprotective (Ali et al., 2000).

The relationship between seizures and neuronal injury in experimental animals is specific to the stage of development and the model employed to produce experimental SE (Sankar et al., 1997, 1998, 2000; Holmes, 2002; Cilio et al., 2003). In these experimental models, younger animals show a decreased tendency to demonstrate neuronal injury (Sankar et al., 1997; Haas et al., 2001), especially in the CA3 and hilar regions of the hippocampus. The reasons for this relative resistance of the immature brain to the type of injury commonly seen in mature rats subjected to SE are not entirely clear. However, the P14 animals showed a striking level of CA1 injury (Sankar et al., 1998).

Furthermore, there is an apparent discrepancy between animal models and some clinical situations in terms of the outcome of SE. The occurrence of seizures contributed to additional neuronal injury in neonates who had sustained hypoxic-ischemic encephalopathy (Miller et al., 2002). The developmental outcome of neonates with electrographic seizures was worse than that of a cohort without seizures even though the two groups were matched for Apgar scores, initial cord blood pH, and base deficits (McBride et al., 2000). Magnetic resonance imaging evidence of hippocampal injury after prolonged and focal febrile SE has been presented (VanLandingham et al., 1998). More recently, results from a multicenter study evaluating children after febrile SE has noted hippocampal signal changes with predominant CA1 involvement (Shinnar et al., 2005), reminiscent of our findings with P14 rat pups subjected to LiPC SE (Sankar et al., 1998). It is conceivable that seizures in these settings occurred when inflammation was already present, an aspect not reproduced in laboratory experiments pertaining to the effect of seizures on the developing brain.

The bacterial endotoxin lipopolysaccharide (LPS), a component of the outer membrane of Gram-negative bacteria, is recognized by cells of the immune system. Treatment with LPS increased susceptibility to both pentylenetetrazol convulsions (Yuhas et al., 2002; Sayyah et al., 2003) and hyperthermic seizures in mice (Heida et al., 2004). The effects of LPS in the CNS are conceivably related to stimulating production of several microglial cytokines (Lee et al., 1993; Turrin et al., 2001), although a direct effect through toll-like receptor (TLR) is also possible (Lehnardt et al., 2003). Pretreatment of postnatal day 17 (P17) rats with high-dose LPS prior to the administration of KA resulted in hippocampal cellular injury (Lee et al., 2000). Here, we report a systematic study of the effect of inflammation on seizure-induced injury produced by modest doses of LPS in immature rats, primarily utilizing a model of SE that had produced CA1-selective hippocampal injury in immature rats (Sankar et al., 1998).


Animals, injection of inflammatory factor, and induction of seizures

Wistar rat pups (Charles River Laboratories, Wilmington, MA, U.S.A.) of either sex were housed in standard laboratory conditions with controlled temperature/humidity, a 12-h light/dark cycle, and free access to food and water. Studies were approved by the Animal Research Committee at the University of California, Los Angeles.

At P6 or P13, animals were injected subcutaneously with 3 mEq/kg lithium chloride (Sigma, St. Louis, MO, U.S.A.). After 14–18 h, rats received i.p. injections of either LPS (10, 50, or 100 μg/kg, E. coli serotype 055:B5; Sigma), or vehicle. SE was induced 2 h later by s.c. injection of pilocarpine (PC, Sigma) in a dose of 100 mg/kg at P7, or 60 mg/kg at P14. Only rats that demonstrated behavioral manifestations of seizures progressing to forelimbs clonus were used for further studies. In order to investigate whether sustained inflammation affected ongoing delayed cellular injury, a separate group of 2-week-old animals were given 50 μg/kg of LPS or vehicle 2 h preceding the injection of PC (60 mg/kg), followed by repeated injections of 50 μg/kg of LPS or vehicle 24 and 48 h after the first LPS treatment.

Only P14 rats were used in studies utilizing KA (5 mg/kg, i.p., Sigma). They were used given a single bolus of LPS 2 h prior to KA. Control animals received vehicle in place of LPS. Experimental design is summarized in Table 1.

Table 1. Experimental design
Age at SENumber of animalsLPS μg/kgPC mg/kgKA mg/kg
  1. Note: 0 indicates vehicle. All animals were euthanized for histological studies 24 h after the last treatment.

  2. LPS, lipopolysaccharide; PC, pilocarpine; KA, kainic acid.

P73100  0 
5  0100 
5 50100 
P143100  0 
9  0 60 
6 10 60 
7 50 60 
8100 60 
9  0 5
14  50 5
P143100  0 
7  0 60 
7Day 1: 50 60 
 Day 2: 0 
 Day 3: 0 
7Day 1: 50 60 
 Day 2: 50 
 Day 3: 50 

Temperature recordings

Core temperature was acquired using a rectal probe and digital thermometer (Fine Science Tools, Foster City, CA, U.S.A.) every 20 min, beginning from 1 h prior to LPS injection until 5 h after the initiation of LiPC-SE (n = 5). Room temperature was maintained at 24°C throughout.

EEG recordings

Under isoflurane anesthesia, rats (n = 8 at P7, n = 9 at P14) were implanted with depth electrodes (Plastics1, Roanoke, VA, U.S.A.) into the cortex (AP: −2.5 mm; ML: 1.5 mm; V: 1 mm) from bregma. Electrographic activity was recorded using the MP150 – EEG100C acquisition system and AcqKnowledge 3.8 software (Biopac Systems Inc., Goleta, CA, U.S.A.). Time of the occurrence of the first spike and/or spike and wave and the total duration of SE were analyzed off-line.


Rats were euthanized with pentobarbital (100 mg/kg, i.p.) and underwent transcardiac perfusion-fixation with saline followed by 4% paraformaldehyde 24 h after onset of SE. Animals given repeated LPS injections were perfused 72 h after PC injection. Brains were removed, dehydrated, embedded in paraffin, cut at 8-μm thick coronal sections, and stained with either hematoxylin & eosin (H&E, Sigma) and examined under fluorescent light (Leica DLMB microscope; Leica Microsystems CMS GmbH, Wetzlar, Germany) using a fluorescein filter (Sankar et al., 1997; Schmued et al., 1997).

All injured neurons, identified by their eosinophilic (acidophilic) cytoplasm and pyknotic nuclei, were counted bilaterally in the CA1, CA3, hilus, and the dentate granule cell layer in sections every 80 μm apart throughout the entire dorsal hippocampus, inclusive of the area represented by 3.0 to 4.3 mm posterior to bregma (Paxinos and Watson, 1982) resulting in 9–12 sections per animal. In preliminary studies, injured neuronal counts were corroborated with 90% agreement by two investigators, and then performed individually in subsequent experiments by one of the investigators (SA) blinded to treatment conditions.

To address the concern of specificity of markers of neuronal injury, two additional histological stains, Fluoro-Jade B (F-JB, Histo-chem Inc., Jefferson, AR, U.S.A.) or terminal deoxynucleotidyl transferase-mediated biotinylated UTP nick end labeling (TUNEL, NeuroTACS, Trevigen, Gaithersburg, MD, U.S.A.), were applied to alternate sections for qualitative corroboration of H&E staining. Sections for F-JB were deparaffinized, rehydrated, incubated with KMnO4 followed by 0.001% F-JB. Sections for TUNEL (LPS and vehicle treated samples were processed concurrently) were also deparaffinized, then incubated with proteinase K (Sigma) followed by terminal deoxynucleotidyl transferase and biotinylated-16-dUTP. The reaction product was visualized with diaminobenzidine (Sigma).

Statistical analysis

Data are expressed as the mean ± standard error of the mean (SEM). Kruskall–Wallis one-way analysis of variance (ANOVA) with post hoc Dunn's test, or Mann–Whitney rank sum test were performed using GraphPad Prism4 software (GraphPad Software Inc, San Diego, CA, U.S.A.). p < 0.05 was considered significant.


All rats given PC at P7 showed behavioral manifestations of seizures progressing to stage 3, but did not show prolonged forelimb clonus like those in the older age group. Ninety percent of P14 rats progressed to stage 3 or beyond and only these responders were used for further studies. In 1-week-old rats, the time to onset (initiation) of SE as well as the total duration of seizures were not significantly different (Fig. 1A). However, in 2-week-old rats, the onset of seizures after PC was significantly delayed as compared to those given vehicle (12.7 min ± 4.3 vs. 8.7 min ± 4.5, p < 0.05). However, the total duration of the SE was not different. Pretreatment with LPS (100 μg/kg) did not affect baseline (data not shown) or ictal EEG patterns at either age (Fig. 1B). In 2-week-old rats, the core temperature did not deviate beyond 37.4–37.6°C during the baseline recording (Fig. 1C). After the injection of LPS (100 μg/kg), but prior to SE, body temperature decreased and was significantly lower than that of those given vehicle. After the onset of SE, core temperature did not differ between groups. In 1-week-old rats, both groups (SE with and without LPS treatment) displayed a decrease in temperature over time, but did not vary from one another with the exception of a spike at one time point in vehicle-treated SE rats.

Figure 1.

Status epilepticus (SE) with or without lipopolysaccharide (LPS). A: Mean (± SEM) of both time to onset (open bar) and total duration (dark bar) of SE after pilocarpine (PC) treatment following a prior injection of vehicle (LPS0) or 100 μg/kg LPS (LPS100). B: EEG tracings show representative cortical recordings during SE corresponding to respective age plus treatment in A. C: Core temperature is not elevated after LPS. Baseline recordings in P14 rats (filled symbols) are lower after LPS than in those given vehicle and equalizes during SE. LPS does not affect core temperature in P7 rats (open symbols), and results in a brief drop during SE. *p < 0.05.

LPS (100 μg/kg) alone did not induce any discernible injury to the hippocampus or any extrahippocampal areas examined (including the thalamus or piriform cortex, data not shown) at either age. However, neuronal injury after LiPC-SE was significantly more severe in the CA1 of both 1-week-old (Fig. 2) and 2-week-old (Fig. 3) pups pretreated with LPS than in those injected with vehicle or given PC alone. Alternate sections stained with F-JB and TUNEL were qualitatively similar with exacerbated injury following LPS pretreatment and SE (Fig. 2C–F). Repeated injections of LPS (50 μg/kg) 24 and 48 h after LiPC-SE further increased CA1 neuronal injury as compared to a single LPS injection in 2-week-old animals (Fig. 4), but the overall effect of repeated 50 μg/kg doses of LPS did not produce a more severe injury as compared to a single dose of 100 μg/kg given prior to the induction of SE (Fig. 3C).

Figure 2.

CA1 neuronal injury in 1-week-old rats. Scattered labeling is seen in vehicle treated animals followed by 100 mg/kg pilocarpine (A,C,E). Significant injury results from an injection of 100 μg/kg lipopolysaccharide (LPS) prior to seizure induction (B,D,F). Cell damage was assessed by H&E fluorescence (A,B), F-JB (C,D), or TUNEL (E,F). G: H&E section of LPS (100 μg/kg) without pilocarpine does not result in discernible injury. Scale bar is 200 μ. H shows hippocampal counts of eosinophilic neurons 24 h after varying doses of LPS (0, 50, or 100 μg/kg) + 100 mg/kg LiPC. *p < 0.05 LPS0 vs. LPS50 or LPS100. H, hilus; DG, dentate granule cell layer.

Figure 3.

Lipopolysaccharide (LPS) plus pilocarpine (PC) induced hippocampal injury in 2-week-old rats. (A) LPS 100 μg/kg + vehicle lacks discernible CA1 neuronal injury; (B) vehicle + 60 mg/kg PC results in cell damage that becomes significantly more pronounced with (C) LPS 100 μg/kg + 60 mg/kg PC. High magnification inset in C demonstrates injured CA1 neurons with eosinophilic cytoplasm, irregularly shaped cell bodies and pyknotic nuclei. Scale bar is 200 μ for A–C, 40 μ for inset. Graph (D) shows the mean hippocampal counts (± SEM) of damaged neurons with varying doses of LPS (0, 10, 50, and 100 μg/kg) + 60 mg/kg PC 24 h after SE. *p < 0.05 LPS0 vs. LPS50 or LPS100. H, hilus; DG, dentate granule cell layer.

Figure 4.

Eosinophilic hippocampal cell counts 72 h after repeated injections of lipopolysaccharide (LPS) followed by pilocarpine (PC). Two-week-old rats were given either vehicle (LPS0) or 50 μg/kg LPS (LPS50, LPS50R) followed 2 h later with PC (60 mg/kg), and then injected with either vehicle (LPS0, LPS50) or 50 μg/kg LPS (LPS50R) at 24 and 48 h after the first LPS injection. Data represent means ± SEM. H, hilus; DG, dentate granule cells. *p < 0.05 LPS0 vs. LPS50, LPS50R and LPS50 vs. LPS50R.

LPS (50 μg/kg) also exacerbated hippocampal injury after KA induced seizures in 2-week-old animals. The number of labeled cells significantly increased in both the CA1 (p < 0.05) and CA3 (p < 0.005) in LPS + KA as compared to vehicle + KA groups (Fig. 5).

Figure 5.

Lipopolysaccharide (LPS) and kainic acid (KA)-induced neuronal injury in 2-week-old rats. Vehicle (LPS0) or 50 μg/kg LPS (LPS50) was followed 2 h later with KA (5 mg/kg). No labeled cells were detected in the dentate granule cell layers or hilus. Data represent means ± SEM. *p < 0.05 CA3 LPS0 vs. LPS50, **p < 0.005 CA1 LPS0 vs. LPS50.


Our principal finding is that region-specific neuronal injury resulting from SE was exacerbated by induction of inflammation. In 2-week-old rat pups, a significant increase in neuronal injury was seen in the CA1 that was not accompanied by injury in the dentate granule cells, the polymorphic cells in the hilus, or the pyramidal cells of sector CA3. Enhanced injury to the CA1 pyramidal cells of 1-week-old rat pups further supports this finding as this age typically demonstrates little or no injury following LiPC SE (Sankar et al., 1997). We did not find any neuronal injury in control rat pups that received LPS without seizures, consistent with the observation by Cai et al. (2003) and Yang et al. (2004) who did not find gray matter injury even at much higher doses of LPS (1 mg/kg).

The KA model, meanwhile, has been well documented to result in a lack of injury to 2-week-old rats (Albala et al., 1984; Sperber et al., 1992). We were interested in seeing if administration of LPS would significantly influence the pattern or the extent of KA-induced injury, and there was modest, but nevertheless, discernible injury as a result in both CA1 and CA3. The observed changes with 50 μg/kg of LPS were qualitatively similar to the exacerbated injury involving both CA1 and CA3 reported by Lee et al. (Lee et al., 2000) who employed a sixfold dose of LPS (300 μg/kg). However, because the total amount of CA1 injury in P14 animals treated with KA was minor compared to the LiPC model, we did not perform parallel studies with KA in the P7 animals.

Lee et al. (2000) observed an increase in temperature after administration 300 μg/kg of LPS. Potential mechanisms mediating enhancement of seizure-induced injury could include an elevation in temperature in response to pretreatment with LPS and/or an enhancement of seizures by LPS. We employed lower doses of LPS and found that 100 μg/kg of LPS in the P14 pup decreased the core temperature, consistent with the report of Heida et al. (2004). The baseline core temperature and that after administration of LPS were both lower in the P7 animals which lack body hair and have a less fully developed thermoregulatory system. In the P7 pups, the difference in temperature after the induction of SE was different between the LPS- and vehicle-treated animals only at one time point at which the LPS-treated animals displayed a lower core temperature. Nevertheless, our data demonstrate significant enhancement of CA1 injury under these conditions. Yang et al. found an enhancement of hypoxic-ischemic brain damage in rat pups by LPS pretreatment without accompanying alterations in temperature (Yang et al., 2004). Additionally, our results show a slight delay in the onset of seizures after PC injection in the LPS rats, and no difference in the total duration of seizures. Thus, neither a difference in core temperature nor a difference in severity or duration of seizures can be attributed to the difference in the extent of neuronal injury seen between animals treated with vehicle versus LPS.

The effect of LPS on SE in the P7 pups was especially surprising, since they do not show significant injury in response to LiPC SE (Sankar et al., 1997). Therefore, we employed additional techniques (Fluoro-Jade B and TUNEL) to confirm qualitatively the dramatic effect of LPS. We readily acknowledge that the use of TUNEL stain does not confirm or deny the mode of death as to whether it is necrotic or apoptotic, since caspase activation, TUNEL staining, and DNA laddering have been associated with a necrotic morphology under certain conditions (Niquet et al., 2003; Niquet et al., 2004). However, the primary focus of our study was to examine the exacerbation of seizure-induced injury per se, rather than the mechanisms of cell death involved in the process.

The ability of LPS to induce cytokines in the immature brain has been demonstrated (Lee et al., 1993; Cai et al., 2000; Turrin et al., 2001). In primary cultures of hippocampal neurons, addition of IL-1β increased the NMDA-mediated calcium increase and increased cell death by approximately 30% (Viviani et al., 2003). While augmentation of NMDA receptors by IL-1β could have contributed to the enhancement of injury from LiPC-SE observed, other factors such as non-NMDA receptor mediated actions can be quite important in explaining the special vulnerability of the CA1 neurons early during development. Bernardino et al. (2005) found that addition of IL-1β or higher concentrations of TNF-α to an organotypic hippocampal slice culture resulted in augmentation of AMPA-induced excitotoxicity. This observation is particularly relevant to our demonstration of the specific pattern of the effect involving CA1 pyramidal cells. In P10 rats, AMPA receptors in the CA1 show a paucity of GluR2 subunits and, as a result, are highly calcium permeant (Sanchez et al., 2001). Further, clearance of glutamate by transporter uptake has been shown to be slower in P12-P14 astrocytes from the CA1 compared to that in astrocytes from adult rats (Diamond, 2005). The role of nitric oxide in mediating the contribution of cytokines IL-1β and TNF-α has been described (Chao et al., 1995; Hu et al., 1997; Yang et al., 2005). Region-specific pattern of substance P expression in the CA1 of P14 rats (Liu et al., 2000), which contributed to enhanced excitotoxicity (Liu et al., 1999a, 1999b), has also been demonstrated. A mutually reinforcing interaction between cytokines and substance P could have contributed to our observations. Stimulation by IL-1β and TNF-α produced cultured brain endothelial cells to secrete substance P (Cioni et al., 1998), whereas stimulation of an astrocytoma cell line by substance P caused them to secrete cytokines (Palma and Manzini, 1998).

Repeated daily injections of LPS further exacerbated injury from an episode of LiPC-SE. The blood-brain barrier is rendered vulnerable by LPS injection (Ivey et al., 2005; Osuchowski et al., 2005) and exposure of LPS-treated P14 rat pups to glutamate administered by a microdialysis cannula in the striatum produced an enhancement in the release of OH radicals, a reactive oxygen species (Cambonie et al., 2004) that can contribute to neuronal injury. Incubation of cultured rat cerebellar granule neurons with LPS increased action potential firing by compromising both delayed rectifier and transient A-type K+ conductances (Mezghani-Abdelmoula et al., 2003).

Seizures are a common sequel of conditions where there is preexisting inflammation such as meningitis, and inflammatory mediators may contribute to enhanced susceptibility to prolonged febrile convulsions (Dube et al., 2005). Yoon et al. (2003) have reviewed the research describing the important role played by intrauterine infection and inflammation in producing cerebral palsy. Intrauterine infection leads to premature delivery and the offspring is highly vulnerable to hypoxia-ischemia. Seizures in this setting of inflammation have been demonstrated to contribute to additional injury (Miller et al., 2002).

A multicenter study, evaluating children who have undergone prolonged febrile seizures, has described CA1-predominant signal changes in the MRI (Shinnar et al., 2005). Our experiments combining inflammation and prolonged seizures by employing LPS and PC have reproduced in rat pups a major aspect of the described pathology in children who have undergone prolonged febrile convulsions. Our experiments have duplicated an important component of the pathophysiology without employing temperature as the stimulus. This model may thus represent a “bedside-to-bench” translation and provide an important basis for future studies on neuroprotection as well as epileptogenicity.


Acknowledgments:  This study was supported by NS046516 (RS), AEAC Association (SA), and the DAPA Foundation. We thank Gene Gurkoff for helpful comments.