Melatonin attenuates kainic acid-induced hippocampal neurodegeneration and oxidative stress through microglial inhibition


Address reprint requests to Seol-Heui Han, Department of Neurology, Chungbuk National University Hospital, Chungbuk National University Medical Research Institute, 62 Gaeshin-dong, Chungbuk 361-711, South Korea. E-mail:


Abstract: The antioxidant and anti-inflammatory effects of melatonin on kainic acid (KA)-induced neurodegeneration in the hippocampus were evaluated in vivo. It has been suggested that the pineal secretory product, melatonin, protects neurons in vitro from excitotoxicity mediated by kainate-sensitive glutamate receptors, and from oxidative stress-induced DNA damage and apoptosis. In this study, we injected 10 mg/kg kainate intraperitoneally (i.p.) into adult male Sprague-Dawley rats. This results in selective neuronal degeneration accompanied by intense microglial activation and triggers DNA damage in the hippocampus. We tested the in vivo efficacy of melatonin in preventing KA-induced neurodegeneration, oxidative stress and neuroinflammation in the hippocampus. Melatonin (2.5 mg/kg, i.p.) was given 20 min before, immediately after, and 1 and 2 hr after KA administration. Rats were killed 72 hr later and their hippocampi were examined for evidence of DNA damage (in situ dUTP end-labeling, i.e. TUNEL staining), cell viability (hematoxylin and eosin staining), and microglial (isolectin-B4 histochemistry) and astroglial responses (glial fibrillary acidic protein immunohistochemistry), as well as lipid peroxidation (4-hydroxynonenal immunohistochemistry). A cumulative dose of 10 mg/kg melatonin attenuates KA-induced neuronal death, lipid peroxidation, and microglial activation, and reduces the number of DNA breaks. A possible mechanism for melatonin-mediated neuroprotection involves its antioxidant and anti-inflammatory actions. The present data suggest that melatonin is potentially useful in the treatment of acute brain pathologies associated with oxidative stress-induced neuronal damage such as epilepsy, stroke, and traumatic brain injury.


Status epilepticus (SE) is a neurologic crisis with high morbidity and mortality [1]. This prolonged epileptic condition can produce selective but widespread neuronal degeneration, primarily in limbic structures. Kainic acid (KA), a potent excitotoxic, a rigid structural analog of glutamate, binds to specific presynaptic and postsynaptic kainate-type non-N-methyl d-aspartate (NMDA) receptors that occur abundantly in neurons in the cornu ammonus of the hippocampus, amygdala, and pyriform cortex, resulting in variable levels of neuronal death in the respective regions [2]. The pattern of neuronal damage induced by systemic KA administration approximates that seen following human temporal lobe epilepsy (TLE). Therefore, systemic administration of KA in rats has been used widely as an experimental model for human SE or TLE [3] and selective neuronal degeneration [4]. The described pathology ranges from neuronal shrinkage, perineuronal vacuolization, and astrocytic and dendritic swelling at early stages, to massive edema, cellular necrosis, reactive gliosis, oligodendritic loss, and demyelination at later stages [5, 6].

The exact molecular and cellular events responsible for these pathologic changes induced by KA are not yet well understood. One plausible mechanism for neuronal injury involves the excitotoxicity that results from the activation of presynaptic kainate receptors and the release of endogenous glutamate [7–9]. However, KA is also thought to mediate damage partly through an indirect mechanism that may involve the overproduction of free radicals, i.e. oxidative stress [10, 11]. It is suggested that the activation of microglia may lead to the production of reactive oxygen species (ROS) and their production can be significantly increased in response to injury [12].

Very recently, several lines of evidence have suggested that the pineal secretory product, melatonin, may have free-radical-scavenger and antioxidant properties [13]. It has been suggested that melatonin may also have strong anti-inflammatory effects [14–16].

Considering the pro-oxidant and proinflammatory actions of KA and the possible importance of microglial cells in hippocampal neurodegeneration, it is crucial to determine the microglial response to KA-associated neurotoxicity. The purpose of the present investigation is to evaluate the contribution of melatonin to protection against neuronal damage and changes in microglial activation in the KA-induced seizure model. This was achieved by examining changes in neuronal degeneration in the hippocampus after systemic KA + vehicle injection or KA + melatonin co-administration, by comparing isolectin-B4 histochemistry (a microglial marker), immunoreactivity for glial fibrillary acidic protein (GFAP; an astroglial marker), 4-hydroxynonenal (4-HNE; a marker for lipid peroxidation) immunohistochemistry, and terminal transferase dUTP nick end-labeling (TUNEL; a marker for double-stranded DNA breaks).

Material and methods


Thirty adult male Sprague-Dawley rats weighing 320–350 g were used for the present study. The rats were obtained from Daehan Biolink (Chungbuk, Korea). Animals were housed under controlled environmental conditions with a 12-hr light–dark cycle (lights on at 07:00 hours) and were fed with standard laboratory food and tap water available ad libitum. All animals were handled in accordance with the Institutional Animal Care and Use Guidelines.

Drug treatment

Kainic acid and melatonin were purchased from Sigma Chemical Co. (St Louis, MO, USA). KA was dissolved in phosphate-buffered saline (PBS), and the pH adjusted to 7.4 with NaOH. Animals were divided into two groups: kainate + vehicle solution (5% ethanol) (KA-only group, n = 15), and kainate + melatonin (KA + melatonin group, n = 15). Animals in the KA-only group were injected intraperitoneally (i.p.) with a single 10 mg/kg dose of KA (0.1 mL/100 g) and observed for the onset of seizure activity. Melatonin was dissolved in 100% ethanol, further diluted in saline (the final ethanol concentration was 5%), and injected i.p. into rats of the KA + melatonin group at 2.5 mg/kg (0.1 mL/100 g) 20 min prior to, immediately after, and 1 and 2 hr after KA administration. Repeated injections of melatonin were essential for maintaining sufficiently high levels of melatonin to reduce neuronal death, as demonstrated previously by other investigators [17, 18]. Only those rats that exhibited full limbic seizures, i.e. forelimb clonus with rearing, were used in this study. Seizures occurred approximately 45 min after injection, typically lasted 2–3 hr, and were not observed to recur.

Tissue processing

Five KA-treated and two melatonin-treated animals died during an acute period of intoxication (mortality rates 33.3% versus 13.3%, respectively). Seventy-two hours after either KA + vehicle or KA + melatonin treatments, five surviving animals were selected randomly from each group, deeply anesthetized with sodium pentobarbital, and immediately perfused transcardially with 50 mL 0.9% saline followed by 500 mL 0.1 m neutral phosphate-buffered formaldehyde (4%). The brain was removed and postfixed at 4°C overnight in the same fixative and dehydrated, embedded in paraffin, and cut into serial 6-μm-thick coronal sections at the level of the dorsal hippocampus (2.5–4 mm posterior to bregma) [19]. Every sixth section was stained with hematoxylin and eosin (H & E) for anatomic orientation and histologic observation.

Lectin histochemistry

To assess microglia, the method described by Streit [20] was followed. Briefly, following deparaffinization and hydration, the sections were immersed in PBS containing cations (0.1 mm CaCl2, MgCl2, and MnCl2) for 10 min. Slices were then incubated in a moist chamber overnight at 4°C with isolectin-B4 from Griffonia simplicifolia (GSA I-B4) coupled to horseradish peroxidase (10 μg/mL; Sigma) in PBS containing 0.1% Triton X-100, 0.1 mm CaCl2, 0.1 mm MgCl2, and 0.1 mm MnCl2. After overnight incubation, the slices were washed three times for 5 min in PBS before they were reacted with 3,3′-diaminobenzidine (DAB; Sigma, St Louis, MO, USA) hydrogen peroxide (H2O2) substrate medium, which allows the lectin-binding sites to be visualized. Selected sections were counterstained with hematoxylin.

GFAP immunohistochemistry

Briefly, after deparaffinization and rehydration the sections were incubated with 0.25% H2O2 in methanol for 20 min to block endogenous peroxidase activity. The sections were washed twice in PBS and incubated with monoclonal antibodies directed against GFAP (clone G-A5; dilution 1:2000; Boehringer Mannheim, Mannheim, Germany) in PBS containing 1% bovine serum albumin (BSA) and 0.3% Triton X-100 for 24 hr at 4°C. After thorough rinses in PBS (3 × 10 min), the sections were incubated in biotinylated anti-goat immunoglobulin G in PBS (dilution 1:200; Vector Laboratories, Burlingame, CA, USA) for 30 min at 37°C and washed twice in PBS. The ABC protocol (Vectastain ABC Kit, Vector Laboratories, Burlingame, CA, USA) and the DAB reaction were used to detect GFAP-immunopositive structures.

Terminal transferase dUTP nick end-labeling

One series of sections was used to identify DNA fragmentation with TUNEL. DNA strand breaks were identified by labeling the 3′-OH ends with terminal deoxynucleotidyl transferase using peroxidase-12-UTP and an in situ Cell Death Detection Kit, POD (Boehringer Mannheim, Indianapolis, IN, USA) according to the manufacturer's instructions with some modifications. Briefly, slides were deparaffinized and endogenous peroxidases were inactivated with 1% H2O2 in PBS/methanol (1:1) for 10 min at room temperature. Slides were then rinsed three times with PBS and preincubated with 0.1 m sodium cacodylate [terminal deoxynucleotidyl transferase (TdT) buffer] at pH 6.0 for 10 min at 37°C. The slides were then exposed for 20 min to the reaction mixture (50 U TdT, 10 nm biotin-dUTP, 25 mm cobalt chloride in TdT buffer) at 37°C. The reaction was stopped by incubating the slides for another 10 min with 0.1 m sodium citrate. After blocking with 2% BSA, slides were covered for 30 min with in situ Cell Death Detection Kit, POD for 30 min at room temperature, rinsed with PBS, and labeled with DAB substrate. As a positive control, tissue sections were treated with DNase (1 mg/mL in potassium cacodylate buffer, pH 7.2).

4-Hydroxynonenal immunohistochemistry

Deparaffinized coronal sections were refixed in Bouin's solution for 20 min. After pretreatment with 3% H2O2 in methanol, the sections were incubated with 10% normal goat serum in 0.01 m PBS containing 2% BSA for 10 min at room temperature to block non-specific binding, followed by overnight incubation with monoclonal anti-4-HNE antibody (1:100; Oxis International, Portland, OR, USA). The sections were then incubated with ABC reagents (Vectastain ABC Kit, Vector Laboratories) for 1 hr and staining was completed by placing the sections in 0.003% H2O2 containing DAB chromogen to visualize the peroxidase-catalyzed reaction product.

Assessment of hippocampal damage

The percentage of surviving neurons in the CA1 and CA3 regions in three separate sections from each animal was calculated at ×400 magnification by an investigator unaware of the treatment that the animal had received. Neuronal survival was quantified within 250-μm2 (before magnification) areas selected in each hippocampal subfield.

Statistical analysis

The values derived from histologic analysis are given as the average percentage of degenerating neurons in a particular hippocampal subfield, expressed as the mean ± S.D., and compared using the Wilcoxon rank sums test. Differences were deemed significant at P < 0.05.


Systemic KA administration produced the well-described sequential behavioral changes. At earlier time points, rats exhibited immobility and rigid postures that were replaced by ‘staring spells’ after approximately 45 min, followed by repetitive head nodding, ‘wet dog shakes’, and subsequent rearing and falling. Eventually, the rats developed generalized tonic–clonic seizures with continuous convulsions (SE), lasting for several hours. In the present study, co-treatment with melatonin did not significantly alter the clinical course of KA-induced limbic seizures.

Consistent with the existing literature, systemic kainate injection resulted in a loss of pyramidal neurons in fields CA1 and CA3 of the hippocampus, while the CA2 region of the hippocampus was relatively resistant to seizure activity. Typical injured eosinophilic neurons were visualized by H & E staining. Degenerating cells become intensively eosinophilic while the nucleus became pyknotic, losing its affinity for hematoxylin stain (red neurons). In contrast, rats treated with KA and melatonin had substantially reduced levels of KA-induced cell loss in the CA1 and CA3 cell layers (Fig. 1, Table 1).

Figure 1.

Representative photomicrographs of hematoxylin and eosin stained sections in the hippocampal CA1 (A, B) and CA3c (C, D) subfields 72 hr after treatment with kainic acid (KA) with vehicle (KA; A, C) and kainic acid with melatonin co-treatment (KA + melatonin; B, D). The number of acidophilic degenerated neurons is dramatically reduced by melatonin co-treatment. Scale bar, 65 μm.

Table 1.  Percentage of acidophilic neurons after KA with vehicle and KA with melatonin co-treatment
Hippocampal subfieldsTreatments
KA + vehicle (%)KA + melatonin (%)
  1. Values represent the mean ± S.D., percentages of degenerated red neurons deriving from three adjacent coronal sections in each animal. The number of rats for each group was five. Data were analyzed by Wilcoxon rank sums test. *P < 0.05 when compared with KA + vehicle group.

CA149± 6.024± 2.9*
CA351± 3.727± 7.4*

Kainic acid-treated rats showed vigorous microglial activation. These microglial changes are dependent on the severity of neuronal degeneration. Melatonin treatment successfully prevented both neurodegeneration and microglial activation, and only a few ramified resting microglia were scattered throughout the hippocampal subfields of melatonin-treated rats (Fig. 2). Even in KA-only-treated rats, there was no microglial activation in the CA2 region, where no neuronal degeneration had occurred (data not shown).

Figure 2.

Representative photomicrographs showing changes in isolectin-B4 expression in the hippocampus. Robust microglial activation is evident in the kainic acid (KA) + vehicle treatment group (A, C), whereas a few ramified resting microglia are seen in the KA + melatonin group (B, D). Scale bar, 65 μm.

The number of GFAP-positive cells was similar in both the KA-only and KA + melatonin groups. Morphologically, KA caused a marked increase in the size, arborization, and stainability of GFAP-immunoreactive cells. These parameters were markedly attenuated when KA was co-administered with melatonin (Fig. 3).

Figure 3.

Immunohistochemical detection of glial fibrillary acidic protein (GFAP) in the hippocampal CA1 (A, B) and CA3c (C, D) subfields. The aborization and stainability of GFAP-positive cells was attenuated with melatonin co-treatment. Scale bar, 65 μm.

4-Hydroxynonenal immunohistochemistry was performed to assess whether KA-induced lipid peroxidation occurs in our model. Using the same procedure performed in the assessment of hippocampal damage (H & E staining), the percentage of 4-HNE positive neurons was calculated in the CA1 hippocampal subfield (KA-only, 25.90 ± 3.54%; KA + melatonin, 9.28 ± 1.38%; P < 0.06, Wilcoxon signed ranks test). The unequivocal 4-HNE immunoreactivity was detected only in degenerating neurons and melatonin co-administration ameliorated the levels of this KA-induced lipid peroxidation product. The 4-HNE reaction products obtained by DAB consisted of brown punctate granules around neuronal membranes (Fig. 4).

Figure 4.

Representative photomicrographs demonstrating immunohistochemical detection of hydroxynonenal (HNE)-modified protein in the hippocampal CA1 subfield, counterstained with hematoxylin. 4-HNE-immunoreactivity was observed in shrunken or triangular neurons (A). The 4-HNE reaction products obtained by 3,3′-diaminobenzidine consisted of brownish punctate granules around neuronal membranes. There are fewer 4-HNE-positive neurons present after co-administration of KA with melatonin (B). Scale bar, 65 μm.

In both KA-only and KA + melatonin groups, TUNEL-positive neurons were detected in the CA1 and CA3c regions. There was a tendency of fewer presence of TUNEL-positive cells with melatonin co-treatment (Fig. 5). However, TUNEL-positivity, regarded as nuclear or DNA fragmentation, can be seen in necrotic as well as apoptotic neurons and is not specific for apoptosis, statistical comparison was not made in this investigation.

Figure 5.

Comparison of terminal transferase dUTP nick end-labeling (TUNEL)-stained sections in the hippocampal CA3c subfield. The number of TUNEL-labeled neurons was reduced in kainic acid (KA) + melatonin-treated animals (B) compared with KA-only-treated animals (A). Scale bar, 65 μm.


Consistent with other investigations, systemic administration of KA (10 mg/kg) to adult rats resulted in well-characterized complex seizure activity starting with mild head nodding, ‘wet dog shakes’, and ‘piano playing state’, and culminating in severe limbic seizures. Although melatonin co-treatment did not significantly alter the neurobehavioral symptoms, it improved the survival rate [33.3% (5/15) KA-treated rats died while 13.3% (2/15) KA + melatonin-treated rats died] in melatonin-co-treated rats.

While KA has been used widely as a model of human TLE and selective hippocampal neurodegeneration, few attempts have been made to characterize the microglial activation associated with this limbic seizure model. Moreover, the sensitivity of hippocampal subfields to KA depends on the route of administration. Intraperitoneal injection results in marked neuronal injuries in the CA4, CA3, and CA1 subfields, and to a lesser extent in CA2, whereas intraventricular and intra-amygdala administration predominantly involves the CA3 subfield. Therefore, the CA2 hippocampal region is relatively resistant to seizure activity and concomitant neuronal damage. In this investigation, we demonstrated that KA-induced hippocampal neurodegeneration was significantly attenuated by melatonin co-treatment compared with the vehicle-treated control group. The observed neuroprotective effects of melatonin were accompanied by attenuated microglial and astroglial reactions, as well as a reduced number of TUNEL-positive and 4-HNE-immunoreactive cells.

Microglia are thought to be the first line of defense in the central nervous system, and they are rapidly activated even in response to minor pathologic changes in the brain [21]. Intracerebroventricular injection of KA induces neuronal loss and the activation of glial cells in the hippocampal CA3 region [22]. It is evident from our study that robust microglial activation occurs primarily in the vulnerable brain regions following KA-induced SE; that is, in the CA1, CA3, and hilar areas, where evidence of astrogliosis, DNA breaks, and lipid peroxidation were also noted. Reactive astrocytes with swollen cell bodies were seen adjacent to degenerated neurons.

It has been suggested that KA-induced neurotoxicity is partly mediated by oxidative stress. Once generated in the microglia, ROS may mediate neuronal death directly via oxidative damage to neuronal membranes, DNA, or organelles [9, 23]. Alternatively, glial-derived ROS may contribute to neuronal death indirectly by increasing microglial activation and cytokine production [24]. It is particularly interesting that oxygen free radicals have the capacity to initiate the destruction of cell membranes by inducing lipid peroxidation and altering membrane phospholipids, with the subsequent generation of reactive aldehydes including malondialdehyde and 4-HNE. HNE can be cytotoxic at the concentrations generated in cells. Furthermore, it renders neurons vulnerable to excitotoxicity [25]. It has also been suggested that HNE is the most reliable index of lipid peroxidation [26]. Our data demonstrate strong HNE immunoreactivity, suggesting the involvement of oxidative stress, in the degenerated hippocampal pyramidal cells 3 days after KA treatment.

A major new finding presented here is that melatonin co-treatment markedly attenuates KA-induced microglial activation (Fig. 2) and lipid peroxidation (Fig. 4) associated with hippocampal neurodegeneration (Fig. 1), and to a lesser extent with astrogliosis (Fig. 3). These findings are consistent with previous reports that melatonin prevents KA-induced neuronal cell death and reduces lipid peroxidation products in rats and mice in vivo [16, 17, 27]. Although the administration of melatonin did not completely inhibit KA-induced seizures, melatonin seems to be effective in attenuating oxidative injuries in vulnerable neuronal cells.

Neuronal injury following SE has traditionally been discussed in the context of excitotoxicity and necrosis [28]. However, increasing evidence suggests that an apoptotic mechanism may be involved during epilepsy [29]. In the present study, many degenerating neurons exhibited positive TUNEL labeling, which is indicative of fragmentation and condensation of chromatin, i.e. DNA breaks in the affected neurons. There are markedly fewer TUNEL-positive cells in rats co-treated with melatonin (Fig. 5), which may suggest an antiapoptotic activity for melatonin, although not all TUNEL-positive cells are apoptotic. However, this conjecture should be confirmed through further investigations employing other technology such as gel electrophoresis of DNA extract, electron microscopy, and analysis of bax and caspase-3 protein expression or combinations of thereof.

Melatonin is an indole that is synthesized in and secreted from the pineal gland primarily during the night. Its lipophilicity ensures that melatonin rapidly crosses the blood–brain barrier and enters cells, where it may accumulate in the nucleus [30]. Recently, it was demonstrated that melatonin is a free radical scavenger [14, 31], an antioxidant that protects cells against the damage induced in various pathologic conditions [32]. Moreover, many investigation demonstrated that melatonin protects the central nervous system from damage caused by free radicals by a variety of toxins including 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine [33], phosphine [34], amyloid-β peptide [35, 36], and ischemia/reperfusion injury [37]. Furthermore, melatonin is also a scavenger of peroxynitrite and it inhibits the production of NO [15]. It also inhibits the occurrence of DNA single-strand breaks in response to peroxynitrite and reduces the suppression of mitochondrial respiration in cultured J774 macrophages [14]. These actions may contribute to the antioxidant and anti-inflammatory effects of melatonin in various pathophysiologic conditions. Thus, the neuroprotective effects of melatonin, as demonstrated in the present study, may reflect its role as a free radical scavenger, an antioxidant, an antiapoptotic, or an anti-inflammatory agent. Moreover this endogenous agent has no known side-effects and readily crosses the blood–brain barrier. Since we did not investigate EEG recordings in this experiment, it is possible to question whether the melatonin-treated group sustained less severe clinical seizures and thus caused less neuronal damages. However, actual neuronal degeneration might contribute to the severity of the clinical phenotype, we speculate that the neuroprotective effect of melatonin is not mainly attributable to decreased seizure activity but to direct antioxidative and/or anti-inflammatory properties. Further investigations using other antioxidants and anti-inflammatory agents during SE should be carried out to elucidate whether oxidative stress and neuroinflammation are directly involved in the pathogenesis of KA-induced hippocampal neurodegeneration.

In conclusion, the observed protective effects of melatonin on KA-induced neurodegeneration is of potential clinical interest in experimental therapy for acute brain pathologies that involve oxidative stress or neuroinflammation, such as epilepsy, stroke, and traumatic brain injury.


The authors thank KJ Lee for his excellent technical assistance in carrying out the histologic procedure. This work was financially supported by Clinical Research Fund from Our Lady of Mercy Hospital and the Korea Science and Engineering Foundation (KOSEF) through the Brain Disease Research Center at Ajou University.