Potential roles for tumor necrosis factor and nuclear factor-κB in seizure activity


  • Benedict C. Albensi

    Corresponding author
    1. Department of Neurological Surgery, Cerebrovascular Research Center, The Cleveland Clinic Foundation, Cleveland, Ohio
    • Department of Neurological Surgery, NB20 Cerebrovascular Research Center, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195
    Search for more papers by this author

Numerous studies have shown that neurotrophins and cytokines can be neuroprotective in the brain (for review see Mattson and Scheff, 1994). Tumor necrosis factor (TNF)-α is a cytokine; however, TNF appears to have several functions, making its role difficult to understand. Moreover, not all investigators (Kelley et al., 2000; Venters et al., 2000; Bowie and O'Neill, 2000; Mattson and Barger, 2000) concur regarding what these functions may be, because TNF not only is involved in neuroprotection but also plays a role in cell death. Furthermore, studies have demonstrated TNF to be involved in the regulation of brain development, sleep regulation, and circadian rhythms (Merrill, 1992; Nistico et al., 1992; Floyd and Kruger, 1997), in addition to brain injury (Fig. 1). With regard to interactions linking cytokines and epilepsy, several studies provide evidence that these intercellular signaling proteins regulate seizure activity (for review see Jankowsky and Patterson, 2001). This review serves to survey briefly some of the pertinent animal and human data involving the role of TNF and nuclear factor-kappa B (NFκB) in seizure activity and in epileptic conditions.

Figure 1.

Signaling for TNF and NFκB and potential links to seizure activity among neurons and glial cells. After seizure activity, TNF can bind to TNF receptors. This binding in turn indirectly triggers the activation of NFκB, which in its inactive form is present as a three-subunit complex, i.e., p50, p65, and IκB (the inhibitory subunit) in the cytoplasm. The activation of NFκB is mediated by IκB kinase, which causes IκB phosphorylation (and eventual degradation) and thus frees the p50 and p65 dimer, making it active. This active dimer is then free to enter the nucleus, where it can bind to consensus κB sequences for κB-responsive genes. These genes can potentially express several different proteins that promote neuronal survival, such as manganese superoxide dismutase (MnSOD), Bcl-2, inhibitors of apoptosis proteins (IAPs), and calbindin. Other signals can also activate NFκB and include nerve growth factor, glutamate, ceramide (released from the cell membrane), increased intracellular calcium, and reactive oxygen species such as H2O2. However, the activation of Jun K pathways (in this example by ceramide) ultimately results in processes of apoptosis. Also following brain injury and/or seizure activity, NFκB can be activated in glial cells (microglial and astrocytes). Activation of NFκB in glial cells results in the production of cytokines such as TNF, excitatory amino acids (EAAs), and nitric oxide (NO). AP1, activator protein 1; ER, endoplasmic reticulum; JUN K, jun N-terminal kinase; Mito, mitochondria; Nuc, nucleus.


TNF can bind to two different receptors, p55 and p75 (Hohmann et al., 1990). Mice lacking the p55 receptor show increased neuronal degeneration following kainic acid (KA)-induced seizures and following ischemia (Bruce et al., 1996). TNF also activates NFκB, which in turn activates neuroprotective pathways and/or pathways involved in cell death (Baldwin, 1996). Inactivated NFκB exists in the cytoplasm as a transcription factor and is composed of a three-subunit complex, which includes p50, p65, and IκB (the inhibitory subunit). NFκB activation results from IκB phosphorylation mediated by IκB kinase. The activated dimer (p50 and p65), once liberated from IκB, then influences genes encoding neuroprotective proteins such as the calcium-binding protein calbindin-D28k and manganese superoxide dismutase (Mn-SOD). Other genes turned on by NFκB in glial cells include glial fibrillary acidic protein and the inducible form of nitric oxide synthase (Mattson and Camandola, 2001).


It has been proposed by several investigators (for review see McEachern and Shaw, 1999) that a continuum exists between processes of plasticity and neuropathology in part owing to the fact that many of the molecular processes involved in synaptic plasticity are the same as those activated during excitotoxic events in neurons. With this in mind, it becomes reasonable to explore also mechanisms of TNF and NFκB, as they relate to both plasticity and pathology.

Neurotrophic factors, such as TNF, may also be important in synaptic plasticity (for review see Albensi, 2001). Brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT3) were first found to potentiate synaptic transmission (Lu and Chow, 1999). Moreover, studies have demonstrated that BDNF plays a role in synaptic transmission and plasticity in the hippocampus (Lu and Chow, 1999). In cultured embryonic cells, TNF was found to enhance whole-cell Ca2+ currents and to reduce N-methyl-D-aspartate (NMDA)-induced currents by a mechanism involving NFκB (Furukawa and Mattson, 1998). In one study (Tancredi et al., 1992), the addition of exogenous TNF to hippocampal slices was found to enhance basal synaptic transmission but interfered with long-term potentiation (LTP) in a dose-dependent manner. Additionally, Albensi and Mattson (2000) found that stimulation of Schaffer collateral axons at a frequency of 1 Hz induced long-term depression (LTD) of synaptic transmission in CA1 of wild-type mice; however, LTD did not occur in slices from TNF receptor knockout mice. Stimulation at 100 Hz induced LTP in slices from both wild-type mice and mice lacking TNF receptors. Pretreatment of slices from wild-type mice with κB decoy DNA (an inactivator of NFκB) prevented induction of LTD and significantly reduced the magnitude of LTP. These data suggest important roles for TNF signaling pathways that modulate NFκB activity in regulation of synaptic plasticity.


Epileptiform activity can be described as abnormally high-frequency bursting activity in distinct brain regions, such as what is seen in the hippocampus (Wheal et al., 1984). In association with this, abnormal regulation of γ-aminobutyric acid (GABA)-mediated inhibition is also often found. This includes an imbalance in the ratio of excitatory to inhibitory influences, resulting in seizure activity and altered firing patterns. Abnormal firing patterns have been further described as being due to the excessive firing of additional populations of pyramidal cells (normally regulated via GABA-mediated tonic inhibition). Some investigators (Bernard et al., 2000) hypothesize that deficits in GABAergic inhibition underlie most forms of epilepsy.

In addition, epileptiform activity is also thought to be the result of several structural alterations in the hippocampus such as neuronal cell loss and functional changes in glial cells (gliosis; Heinemann et al., 2000). It was also suggested (Pollen and Trachtenberg, 1970) many years ago that sites of epileptogenic foci are characterized by disruptions in potassium homeostasis. Furthermore, epileptic seizures involve other cell types such as endothelial, microvessels, and astroglial and microglial cells (de Bock et al., 1996). Glial cells are thought to limit abnormal increases in [K+]o, such that disturbances in glial cell function may promote epileptiform activity by not regulating excess potassium. In the case of severe seizures, not only do we see glial activation and neuronal damage, but investigators have also reported mossy fiber sprouting (Ben Ari et al., 1980).


Several animal studies have demonstrated the involvement of TNF and NFκB in seizure activity. For example, the induction of spontaneously recurring seizures involves the activation of inflammatory cytokines and related pro- and antiinflammatory genes in the rat hippocampus (De Simoni et al., 2000). Similarly, other investigators have found correlations between brain damage intensity and the release of TNF-α, suggesting production of this cytokine by macrophagic microglia. These investigators (de Bock et al., 1996) proposed a role for TNF-α and interleukin (IL)-6 in the adaptive phenomena that follow severe limbic seizures.

With regard to NFκB activation, data also support a role in seizure activity. In one study, significant induction of NFκB was observed at 4 hr after KA injection, and the maximal increase was reached at 8–16 hr posttreatment. Additionally, NFκB binding activities returned to control levels 5 days after injection (Rong and Baudry, 1996). Moreover, Won and colleagues (1999) provided evidence that the differential vulnerability of neurons in the rat and the hamster hippocampus to kainate is partly mediated by mechanisms involving NMDA-dependent activation of NFκB.


TNF is known to stimulate the proliferation of astrocytes and therefore may play a role in reactive gliosis resulting from brain injury (Selmaj et al., 1990; Balasingam et al., 1994). NFκB activity in glial cells has also been reported in several studies. In one such study, it was demonstrated that IL-1 causes persistent activation of NF-κB in glial cells (Bourke et al., 2000). Furthermore, recent results from Chang and others (2001) demonstrated that astrocytic TNF-α production was potently inhibited by K+, with 44% and 89% inhibition at 25 and 55 mM potassium, respectively. In contrast, astrocyte IL-6 inhibition required higher concentrations of K+ (≥75 mM). These results demonstrate a novel role for astrocyte potassium channel activity in modulation of glial cytokine production (Chang et al., 2001). Finally, it was also shown that NFκB activity is rapidly increased in hippocampal neurons within 4–16 hr following kainate-induced seizures, which is then followed by a delayed and sustained increase in NFκB activity in glial cells (Mattson and Camandola, 2001).


Over the last decade, several investigations have found evidence for cytokine involvement in neuron–glia interactions in human patients with epilepsy. The cytokine network is also activated in patients after recent tonic-clonic seizure (Peltola et al., 2000). In another study, two different human astrocytic cell lines derived from adult epilepsy surgical specimens were exposed (Barna et al., 1990) in vitro to 1–100 ng/ml recombinant TNF-α, and growth of adult human nonneoplastic astrocytes was stimulated by TNF alpha. Additionally, Liu et al. (2000) investigated IL-4Rα expression in specimens of nonneoplastic cerebral cortex removed for surgical treatment of intractable epilepsy compared with glial tumors and found that nonneoplastic epilepsy astrocytes express IL-4Rα in situ, confirming in vitro studies and implying IL-4 sensitivity in vivo.


Clearly a number of studies have implicated TNF and NFκB in seizure activity in both animal and human models. More specifically, interactions between TNF signaling and changes in channel conductances during conditions of seizure activity could be pursued in future work based on these prior studies. Furukawa and Mattson (1998) found that NFκB modulated voltage-dependent calcium channels and glutamate receptors. Therefore, it is feasible that changes in TNF levels can affect glutamate channel activity, affecting the overall balance of excitatory to inhibitory influences, ultimately resulting in epileptiform activity. Furthermore, these studies also showed that TNF increased calcium current while decreasing NMDA-induced current, suggesting that calcium influx though NMDA channels may play an important role in excitotoxicity. Signaling mechanisms such as these in neurons may be especially important for the induction and propagation of seizure activity. Additionally, similar signaling mechanisms may also exist in glial cells but have not been explored in detail. For example, Hinterkeuser et al. (2000) recently explored the functional properties of astrocytes in patients with temporal lobe epilepsy (TLE) who were pharmacoresistant to standard drug treatment. They found that a significant reduction of inward rectification of potassium current in human astrocytes was seen in sclerotic tissue compared with lesion-associated TLE. It would be interesting to perform additional experiments in this model to determine whether TNF is also involved in the modulation of potassium currents. Other future explorations might include glial–neuronal–vascular interactions in the presence of TNF.


There is substantial evidence that cytokines are involved in seizure activity and in human epilepsy. Animal and human studies have shown that cytokine-signaling proteins regulate seizure activity along various pathways. More recently, data showing the role of TNF and NFκB in seizure activity and human epilepsy have expanded, showing complex interactions among neurons and glial cell signaling.