© 2004, AAAS

In a normal neuron (top), action potentials progress down the axon and there is a less robust dendritic echo. In a person with sprouted axons (middle), action potentials are rerouted back to neighboring dendrites. But action potential reverberation can occur without sprouting (bottom). Blocked A-type K+ channels reduce inhibitory A currents allowing larger echoes. (From: K. Staley, Science, 305:482–3, July 23, 2004.)

Epilepsy often develops after the brain is damaged, and patients commonly must take anticonvulsant drugs for a lifetime despite unpleasant side effects. Such drugs target the seizures but not the underlying cause. Now, new theories promise to untangle the mechanisms of epileptogenesis and presage the possibility of a new generation of drugs that treat the initial brain damage and prevent epilepsy from developing.

In roughly half of all patients with epilepsy, the condition develops later in life...


A generally held theory suggests that something upsets the balance of excitatory and inhibitory signals in the brain, leading to overall hyperexcitability. This theory rests on observations that inhibitory cells become less active while excitatory pathways multiply, implying "a simple balance of inhibition and excitation," according to John Duncan of the Institute of Neurology in London. He says that this theory is probably wrong, or at least incomplete. "The reality is clearly more complicated, as neurons form intricate networks and interconnections."

New research suggests that subtle structural changes occur within individual neurons, at the level of ion channels, which affect neuronal functioning in multiple ways. Researchers say this has radical implications for development of anti-epilepsy drugs. Beyond simply boosting inhibition or reducing excitation, one must tackle the underlying causes, which may include neuronal loss or ion-channel disruption. Until the underlying cause is found, epilepsy cannot be prevented.


Researchers continue to search for answers, especially in the hippocampus. In temporal lobe epilepsy, the most common form of the disease in humans, the hippocampus plays a key role in initiating seizures. The dentate gyrus in the hippocampus seems to be most susceptible. Granule cells in the dentate gyrus receive excitatory signals from outside the hippocampus and contribute to hippocampal output. This output is kept in check by inhibitory interneurons within the hippocampus.

In 1991, Bob Sloviter of the University of Arizona proposed a theory of epileptogenesis that is still probably the most influential today.1 His theory of deficit inhibition says that the inhibitory mechanisms regulating granule cell output can become dormant, leading to the frenzy of electrical activity that constitutes a seizure. That theory's main rival proposes that granule cell excitation becomes increased in the epileptic brain.

Following damage, the brain attempts to rewire itself. Neurons sprout new axons and make new connections. But new axons could grow into existing networks, creating positive feedback loops that amplify excitatory signals.


In July, a group led by Christophe Bernard of INSERM in Marseilles, France, discovered a previously unidentified feedback mechanism that could also contribute to epileptogenesis.2 "You can have a form of feedback that doesn't involve the sprouting of new axons," says Kevin Staley, a neurologist at the University of Colorado Health Sciences Center in Denver. "It just involves allowing the signal to echo back into the dendrites more freely."

Dendrites, the information processors of a neuron, receive incoming signals and transmit them to the soma or cell body. Normally the outflow of positively charged potassium ions through the neuronal membrane acts as an inhibitory signal. And in the dendrites, potassium channels prevent the signals coming into the soma from reverberating back to the dendrites and reinforcing themselves.

Bernard's group studied these channels in a rat model of epilepsy. In rats, the drug pilocarpine triggers a prolonged seizure known as status epilepticus. The rat recovers, only to have spontaneous seizures develop three weeks later. The hippocampal tissue shows dramatically reduced numbers of potassium channels.2 Those left, they found are phosphorylated and less active than usual. This channelopathy allows the signal to echo unchecked in the dendrites.

Bernard says that demonstrating an acquired channelopathy in the dendrite is unique. "Up to now we were just talking about channelopathies that were genetic." The research on acquired channelopathies has largely focused on the cell body; few had thought to look in the dendrites.

Though it is not yet possible to reverse the channelopathy, Bernard's group has experimented with MEK inhibitors that interfere with the phosphorylation of potassium channels and produce partial recovery.

The channelopathy is probably not the whole story, but those working on molecules designed simply to boost inhibition in the brain are "barking up the wrong tree," Bernard says. "It will take time to change mentalities on this issue, because most of epilepsy research is based on a deficit of GABA-ergic inhibition."

"[Bernard's] theory may ultimately turn out to be true," says Donald Weaver of Dalhousie University, Nova Scotia. "But like most things in science, it will have to be replicated in many different labs before acceptance is widespread." He points out that the study of epileptogenesis is still in its infancy.



Courtesy of Christophe Bernard, INSERM

In this CA1 pyramidal cell superimposed on its environment, action potentials are generated in the perisomatic region and their amplitude reaches about 100 mV. Backpropagated action potential (bAP) decreases with distance from the soma in control tissue (green bars) to levels of 2–3 mV. In epileptic tissue (red bars), the bAP amplitude remains around 40 mV.

Weaver has dedicated his own research to finding pharmacological interventions to epileptogenesis. "We're working on the notion that there is a concomitant excessive excitation combined with a loss of inhibition," he says. A prime target: β-alanine, which binds to both inhibitory GABA receptors and excitatory NMDA receptors, stimulating the former and acting as an antagonist to the latter. "This one molecule has feet in two camps," he explains. A Canadian biotech company, Neurochem, is now developing β-alanine analogues. To date it has close to 400 and is testing them in rats.

Meanwhile, at Duke University in Durham, NC, James McNamara has a different approach. He is interested in brain-derived neurotrophic factor (BDNF), which is known to promote axon sprouting and has been shown, in animals, to increase tenfold after a brain injury. His group created knockout mice lacking either BDNF or its receptor, TrkB, and tested their susceptibility to the kindling model of epilepsy. In kindling, the animals are given small electric shocks or small doses of toxic chemicals at regular intervals. After about two weeks the treatments trigger seizures.

McNamara's group recently found that mice lacking the TrkB receptor – but interestingly not those lacking its ligand, BDNF – are resistant to kindling.3 The mice don't display the gradual lowering of seizure threshold. "Of all the studies that have been done in that model for the last 35 years, no single perturbation, be it pharmacological or genetic, has actually prevented the [kindling] process," says McNamara. "The fact that this one did argues that TrkB in particular is essential, not merely regulatory."

He admits that the findings need to be replicated in other animal models, such as the status epilepticus model, because with kindling the animals never experience spontaneous seizures. Many consider it a less than ideal model of human epileptogenesis.

At the Institute of Experimental Medicine in Budapest, Hungary, Tamás Freund is developing a new epilepsy model. Owing to the endocannabinoid receptors on interneurons in the hippocampus, cannabinoid signaling might modulate interneuron signaling. So, in hopes of testing new drug targets, Freund is creating a transgenic mouse in which endocannabinoid receptors are expressed wherever inhibitory neurons are found in the brain, rather than in just a subset.

The interneuron also is coming under more scrutiny. Epilepsy has long been considered a disorder of granule cell output. Bernard's group has shown, however, that temporal lobe epilepsy is associated with the loss of a particular type of interneuron that projects to the dendrites of granule cells.4

The massive electrical discharge of granule cells during a seizure is highly synchronized, just like their normal activity. Since inhibitory interneurons that regulate granule cell output are also responsible for synchronizing output across large cell populations, impairing their function should disrupt that synchronicity, if epilepsy is truly a general breakdown of inhibition.

Bernard suggests that only a subset of interneurons is affected, and epileptogenesis results from a disruption of input to the granule cells via the dendrites, not their output. This finding has since been replicated in rats and humans by Freund.5 Working with tissue surgically removed from patients' temporal lobes in an attempt to eliminate seizure foci, he and his colleagues found that the number of neurons in that same subset was reduced. "The major message of our work is that the perisomatic inhibitory cells, which are responsible for the synchronization of large cell populations, are completely intact," says Freund.

He adds that the results suggest new drug targets. Therapies could protect the vulnerable interneurons, for instance. And as more of the basic mechanisms of epileptogenesis are revealed, it may be only a matter of time before drugs can prevent epilepsy, rather than just manage it.

Laura Spinney (lspinney@the-scientist.com)

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!