Cognition in Epilepsy: ΔFosB Takes Center Stage (and May Star in the Prologue)

Commentary

Epigenetic Suppression of Hippocampal Calbindin-D28k by ΔFosB Drives Seizure-Related Cognitive Deficits.

You JC, Muralidharan K, Park JW, Petrof I, Pyfer MS, Corbett BF, LaFrancois JJ, Zheng Y, Zhang X, Mohila CA, Yoshor D, Rissman RA, Nestler EJ, Scharfman HE, Chin J. Nat Med 2017;23:1377–1383.

The calcium-binding protein calbindin-D28k is critical for hippocampal function and cognition, but its expression is markedly decreased in various neurological disorders associated with epileptiform activity and seizures. In Alzheimer's disease (AD) and epilepsy, both of which are accompanied by recurrent seizures, the severity of cognitive deficits reflects the degree of calbindin reduction in the hippocampal dentate gyrus (DG). However, despite the importance of calbindin in both neuronal physiology and pathology, the regulatory mechanisms that control its expression in the hippocampus are poorly understood. Here we report an epigenetic mechanism through which seizures chronically suppress hippocampal calbindin expression and impair cognition. We demonstrate that ΔFosB, a highly stable transcription factor, is induced in the hippocampus in mouse models of AD and seizures; it binds and triggers histone deacetylation at the promoter of the calbindin gene (Calb1) and downregulates Calb1 transcription. Notably, increasing DG calbindin levels—either by direct virus-mediated expression or inhibition of ΔFosB signaling—improves spatial memory in a mouse model of AD. Moreover, levels of ΔFosB and calbindin expression are inversely related in the DG of individuals with temporal lobe epilepsy (TLE) or AD and correlate with performance on the Mini-Mental State Examination (MMSE). We propose that chronic suppression of calbindin by ΔFosB is one mechanism through which intermittent seizures drive persistent cognitive deficits in conditions accompanied by recurrent seizures.

Cognitive impairment is a relatively frequent comorbidity in epilepsy (1). Although the severity varies widely among individuals, memory and learning difficulties can significantly reduce quality of life in those affected (2). Knowledge about causes and trajectories of cognitive impairment in epilepsy is incomplete and, more importantly, treatments to improve cognition in epileptic patients are scarce. In a comprehensive study using mouse models and human tissue, Chin and colleagues have recently provided interesting clues about potential underlying molecular mechanisms that may lead to the development of mechanism-targeted treatment strategies to improve cognition in epilepsy in the future.

The study is based on a long-known observation in experimental and human epilepsy and Alzheimer's Disease (AD): reduction of expression of the calcium-binding protein calbindin-D28k in the dentate gyrus. Both conditions, epilepsy and AD, are associated with impaired cognition and seizures, suggesting overlapping pathological mechanisms (3). Because of calbindin-D28k's prominent role in synaptic function and cognition (4), the authors speculated that reduced calbindin-D28k may contribute to cognitive defects. To test this hypothesis, they first set out to answer the question of what could cause this loss of calbindin immunoreactivity in epilepsy and AD. Theoretically, at least two scenarios come to mind: In the first scenario, calbindin protein expression itself is reduced due to decreased gene transcription, mRNA translation or protein stability. In the second scenario, the number of cells expressing calbindin is diminished. The latter is plausible, as seizures and epilepsy are often accompanied by neuronal cell loss. However, several studies have shown that the reduction of calbindin-D28k immunoreactivity in the dentate gyrus is not due to neuronal cell loss but may rather reflect a change in neuronal excitability. Indeed, grafting fetal cells into the CA3 region of a rat model of epilepsy to overcome loss of CA3 and hilar neurons restores calbindin staining in the dentate gyrus, suggesting that normalization of hippocampal network activity also normalizes calbindin (5).

In their study, You et al. go a step further to provide evidence for a molecular mechanism underlying reduced calbindin after seizure. They show that protein levels of the transcription factor ΔFosB are increased in neurons in which calbindin is reduced in two different mouse models of epilepsy with cognitive deficits, a seizure-prone AD mouse and the pilocarpine model of temporal lobe epilepsy. Elevated ΔFosB is accompanied by decreased histone deacetylation and hypermethylation of the Calb1 promoter in these mice, suggesting epigenetic suppression of calbindin-D28k. Using virus-mediated overexpression or inhibition of ΔFosB, the authors elegantly show that elevated ΔFosB is both necessary and sufficient for reduction in calbindin-D28k and cognitive impairment. Moreover, viral calbindin overexpression also improved spatial memory in the AD mouse, supporting a direct causal role of reduced calbindin in mediating cognitive impairment.

The relevance of findings by You and colleagues for the human disease was illustrated by experiments in postmortem dentate gyrus tissue from individuals with AD or mild cognitive impairment (MCI), and in surgically resected dentate gyrus tissue from people with temporal lobe epilepsy. As in the mouse models, increased ΔFosB correlated with decreased calbindin. Moreover, ΔFosB expression levels correlated with cognitive performance in individuals with MCI. These kinds of studies in human tissue can provide only correlative but not causal evidence. Nevertheless, these findings support the hypothesis that similar mechanisms occur in human disease as in mouse models.

How could translational epilepsy research and drug development take advantage of these findings? One could imagine two potential rescue strategies to overcome the reduced calbindin-staining and resulting cognitive impairments: replenish the lost protein calbindin, or reduce the upstream regulator and apparent culprit, ΔFosB. The latter may be more effective because increased ΔFosB is expected to alter the expression of many other downstream targets, in addition to calbindin. Although You et al. show that calbindin overexpression improves spatial memory in the AD mouse model, they did not show if that is also the case in temporal lobe epilepsy. Importantly, it is also unclear if calbindin overexpression completely restored cognition and behavior or just improved them. There are a few potential targets of ΔFosB that are critical for synaptic plasticity and may play a role in cognition. The AMPA receptor subunit GluA2, for example, contains a ΔFosB-sensitive site in its promoter region and is elevated after ΔFosB overexpression in the nucleus accumbens (6). GluA2 is the only AMPA receptor subunit that is calcium-impermeable and therefore has a strong influence on AMPA receptor conductance and neuronal function. Another interesting ΔFosB downstream target that has been implicated in epilepsy and AD is cyclin-dependent kinase 5 (Cdk5; 7–9). Before translating findings by You et al. into clinical trials, more research is needed to understand if calbindin is indeed the major downstream target of ΔFosB mediating cognitive defects—or if other targets, such as GluA2 or Cdk5, contribute as well.

It is well known that elevated synaptic activity and seizures induce ΔFosB expression. But can ΔFosB overexpression itself alter neuronal activity and seizure susceptibility? Notably, You et al. showed that overexpression of ΔFosB impaired synaptic plasticity, as evident by deficiencies in spatial memory, but did not evoke spontaneous seizures. This would have important implications if the findings of this paper were developed into a therapeutic strategy. It could mean that ΔFosB-targeted treatments would correct only the cognitive deficits while leaving the seizure phenotype unchanged. Brain activity was recorded using hippocampal and cortical EEG for a 2-day period, and no abnormalities were observed. Although this is strong support for the hypothesis that ΔFosB overexpression per se does not induce seizures, follow-up studies are needed to further assess the causal or consequential relationship between ΔFosB and seizures. It is conceivable that increased ΔFosB in the hippocampus is a consequence of neuronal hyperactivity but, at the same time, causes a sustained hyperexcitable state of the brain, making subsequent seizures more likely. This could be tested by assessing susceptibility to chemically provoked seizures. This is particularly important as the AD mouse model analyzed in this study does not only show spontaneous nonconvulsive seizures but also has higher susceptibility to chemically induced seizures (10). The effect of ΔFosB overexpression on spontaneous or provoked seizure activity in the AD mice was not assessed in this study but will be essential to fully understand if increased ΔFosB is only a consequence of seizure activity or whether it may be both cause and consequence.

Although difficult to comprehensively assess in large cohorts, there is evidence that in certain epilepsy disorders cognitive and behavioral impairments start before seizure onset (1). The findings of Chin and colleagues have suggested that, at least in the mouse models studied, cognitive impairment in epilepsy is not a direct consequence of seizure activity but rather mediated by molecular changes, such as increased ΔFosB, that lead to reduced calbindin expression. It will be interesting to assess if ΔFosB-mediated reduction of calbindin-Dk28 may be a contributing factor to cognitive impairment before the onset of convulsive seizures in affected humans.