Mechanisms of Epileptogenesis: a Convergence on Neural Circuit Dysfunction

REVIEWS

Mechanisms of epileptogenesis: a convergence on neural circuit dysfunction

Ethan M. Goldberg and Douglas A. Coulter Abstract | Epilepsy is a prevalent neurological disorder associated with significant morbidity and mortality, but the only available drug therapies target its symptoms rather than the underlying cause. The process that links brain injury or other predisposing factors to the subsequent emergence of epilepsy is termed epileptogenesis. Substantial research has focused on elucidating the mechanisms of epileptogenesis so as to identify more specific targets for intervention, with the hope of preventing epilepsy before seizures emerge. Recent work has yielded important conceptual advances in this field. We suggest that such insights into the mechanisms of epileptogenesis converge at the level of cortical circuit dysfunction.

Epilepsy (BOX 1) is a common end point of many forms neuron-restrictive silencer factor (NRSF)) pathways in of acquired brain pathology (such as tumours, infection, acquired and genetic epilepsies. We discuss how dysfunc- stroke and traumatic brain injury). Epilepsy can also be tion of specific subtypes of neurons might lead to epilepsy, Dravet syndrome the result of the mutation of a single gene (genetic epi- with a focus on the theory of cortical GABAergic interneu- Previously referred to as severe lepsy), and it can be one component of a neurodevelop- ron dysfunction as the basis of the infantile-onset epileptic myoclonic epilepsy of infancy, mental disorder. The most common epilepsy syndrome encephalopathy of Dravet syndrome. Finally, we consider Dravet syndrome is a spectrum in adults — mesial temporal lobe epilepsy (TLE; see the pathology of discrete elements of neuronal circuits in of severe infantile-onset BOX 2) — is typically considered to be an acquired disor- mesial TLE. For a broader overview, see recent general epileptic encephalopathy that 5,6 is due to an underlying genetic der, resulting from injury to a previously normal brain. reviews of the mechanisms of acquired epileptogenesis . cause, most commonly However, there are also familial (presumed genetic) Seizures are network events that require re-entrant mutation of the voltage-gated forms of TLE. Although usually applied to the acquired activation of populations of neurons embedded within sodium channel subunit epilepsies, the concept of epileptogenesis might also be brain circuits; hence, epilepsy is inextricably a circuit-level NaV1.1. Typically, infants are initially normal, then develop relevant to the genetic epilepsies. An anti-epileptogenic phenomenon and cannot be understood outside this con- febrile and afebrile seizures, intervention could be used in the context of a broader text. It is likely that multiple and diverse mechanisms of progressing to myoclonus and neurodevelopmental abnormality that includes epilepsy epileptogenesis overlap to various extents among different multiple seizure types, and if an underlying defect could be identified and modu- epilepsy syndromes, both acquired and genetic. However, cognitive impairment. lated before seizures began. Conversely, there could be for a discrete genetic mutation or dysfunction of a molec- potentially modifiable genetic contributions to what are ular signalling pathway or specific cell type to produce classically considered to be acquired epilepsies, including epilepsy, it must do so by altering the function of circuits. 1–4 The Children’s Hospital of TLE . For example, an individual might have a genetic We suggest that a greater mechanistic understanding of Philadelphia, Division of predisposition to risk factors for epilepsy, such as status circuit function and circuit-level dysfunction in epilepsy Neurology, Colket epilepticus or hippocampal sclerosis, which in turn drive will lead to the development of successful and broadly Translational Research Building, 3501 Civic Center epileptogenesis. applicable interventions in epileptogenic processes, which Boulevard, Philadelphia, In this Review, rather than attempting to provide a remain a primary unmet need in epilepsy therapy. Pennsylvania 19104–4399, broad overview of epileptogenesis, we restrict our dis- USA. cussion to a set of emerging themes. We review the role Molecular signalling pathways in epileptogenesis Correspondence to: D.A.C. of specific large-scale molecular cell signalling pathways The possibility that a brain injury or genetic abnormal- e-mail: [email protected]. upenn.edu in epileptogenesis, highlighting the mammalian target ity might lead to the development of epilepsy through doi:10.1038/nrn3482 of rapamycin (mTOR) and repressor element 1 (RE1)- maladaptive activity of one or more large-scale molecu- Published online 18 April 2013 silencing transcription factor (REST; also known as lar signalling pathways is inherently attractive, as this

NATURE REVIEWS | NEUROSCIENCE VOLUME 14 | MAY 2013 | 337

Box 1 | Definitions: from seizure to epilepsy encoding the serine/threonine kinase AKT isoform that Seizures is predominant in the brain (AKT3), phosphatidylinosi- Seizures are abnormal, paroxysmal changes in the electrical activity of the brain. tol‑4,5‑bisphosphate 3‑kinase-α (PIK3CA) and MTOR 10,11,18–20 Seizures emerge from and are an inherent property of cerebral cortical structures. itself, produce syndromes that include epilepsy However, a (single) seizure can occur in an otherwise healthy individual; such (FIG. 1). In addition, manipulation of the mTOR pathway individuals are generally not considered to have epilepsy. can modulate acquired epilepsy in animal models. Epilepsy The role of the mTOR pathway in genetically deter- Epilepsy, by contrast, is a neurological disorder defined as a brain state that supports mined epilepsy11,18 is typified by tuberous sclerosis, which recurrent, unprovoked seizures. results from mutations in TSC1 and TSC2 (FIG. 1). The Epileptogenesis mechanisms of seizure generation in tuberous sclerosis are Epileptogenesis is the process by which the previously normal brain is functionally debated21, but seizures are thought to originate around or altered and biased towards the generation of the abnormal electrical activity that within cortical tubers22, which are foci of abnormal neu- subserves chronic seizures. The concept of a ‘mechanism of epilepsy’ refers to any ronal migration and cellular growth. Intracranial electro- biological feature of the brain that drives or supports recurrent, unprovoked seizures. encephalographic monitoring in patients with tuberous Examples of these biological features include molecular, anatomical or circuit level sclerosis has provided evidence that at least some tubers alterations, such as cell death or dysregulation of an ion channel or neurotransmitter are intrinsically epileptogenic and that they might in some receptor. By this definition, epilepsy mechanisms may be the downstream effectors of, cases be more epileptogenic than the perituberal cortex23. but are distinct from, a prior epileptogenic process. GFAP The process of epileptogenesis is classically thought to occur in three phases: first, the Tsc1 conditional knockout mice, in which Tsc1 is occurrence of a precipitating injury or event; second, a ‘latent’ period during which deleted from glia, show astrogliosis and dislamination, changes set in motion by the preceding injury act to transform the previously normal with pyramidal cell dispersion in the hippocampus24. brain into an epileptic brain; and third, chronic, established epilepsy. It is during the Glia in these mice show altered expression of gap junction latent period that the process of acquired epileptogenesis is thought to coalesce, and it proteins, potassium channels and astrocytic glutamate is at this point in the process that interventions might be used to prevent the transporters25. Treatment of these mice with the mTOR subsequent development of epilepsy. inhibitor rapamycin suppresses seizures; however, seizures The distinction between seizures, epilepsy and epileptogenesis is well illustrated by return if rapamycin is withdrawn26. the accumulating data indicating that classical anticonvulsants, which can terminate an Interestingly, selective deletion of the mTOR pathway ongoing seizure and decrease or prevent the occurrence of future seizures in epileptic individuals, are wholly ineffective in preventing the process of epileptogenesis169. gene Pten from a small subset of dentate granule cells in mice produces TLE27. The mechanism by which deletion of Pten in a subset of dentate granule cells might lead to epilepsy is unclear, but it caused cellular-level structural would suggest that directed manipulation of these path- abnormalities including increased spine density and ways might be able to modify the epileptogenic process granule cell hypertrophy, and this abnormal cell growth before the appearance of seizures. Research is motivated might itself be epileptogenic. Mossy fibre sprouting in these by an attempt to identify molecular signalling processes mice was the same from wild-type granule cells as from that might be involved in driving the formation of the those in which Pten had been deleted, which suggests aberrant neuronal circuitry seen in the epileptic brain, that mossy fibre sprouting might be a consequence rather including features such as axonal sprouting and cell death than a cause of epilepsy, at least in this mouse model. Circuits that develop following a delay after an initial brain insult. Somatic mosaicism for gain‑of‑function mutations Brain elements that are Important potential master regulators of epileptogenesis in the mTOR pathway components AKT3, PIK3CA composed of a collection of have emerged in recent years, including brain-derived and MTOR are responsible for some cases of the dra- neurons of various types and are connected to one another neurotrophic factor (BDNF)–tropomyosin-related matic focal cortical brain malformation that occurs in 7–9 28,29 to form a network or ensemble, kinase B (TRKB; also known as NTRK2) signalling , hemimegalencephaly . This condition is almost invari- the operation of which is the mTOR pathway10–12, the REST pathway13 and other ably associated with medically intractable epilepsy. Mice directed towards a particular transcriptional regulators14. Here, we focus on the roles harbouring an activating mutation in Akt3 have epilepsy function or domain of 30 information processing. Circuits of mTOR and REST given the recent and rapid accumula- and increased brain size . receive inputs, process tion of evidence implicating these pathways in temporal The mTOR pathway might also be involved in acquired information and produce lobe epileptogenesis. epilepsy. This has been investigated using rodent models outputs. of chronic acquired TLE after status epilepticus induced by Mammalian target of rapamycin. mTOR is a serine/ the chemoconvulsants kainate31 or pilocarpine32–34 (kainate- Tuberous sclerosis or pilocarpine-induced status epilepticus An autosomal dominant threonine protein kinase that is involved in multiple ), TLE induced by 35 multisystem disorder that is basic cellular functions, including cell growth, prolifera- electrical stimulation of the angular bundle or the amyg- caused by a mutation of the tion and survival. mTOR is regulated by pathways that dala36, post-traumatic epilepsy after brain injury37,38 and tumour suppressor genes respond to glutamatergic transmission and the energy models of infantile spasms39 and early hypoxic-ischaemic tuberous sclerosis complex 1 (FIG. 1) (TSC1; encoding hamartin) or state of the cell . In turn, mTOR regulates synaptic injury. Rapamycin treatment before kainate-induced sta- 15 16 TSC2 (encoding tuberin). It is plasticity , neuronal structure and the expression of tus epilepticus prevented the development of spontane- accompanied by the various ion channels and neurotransmitter receptors17. ous recurrent seizures in a subset of rats31; in those rats characteristic brain lesions of The mTOR pathway is strongly associated with epilepsy. that did develop epilepsy, rapamycin increased the latency cerebral cortical tubers and Genetic mutations in many components of the mTOR to seizure development and decreased the frequency of subependymal nodules, and by tuberous sclerosis the neurological sequelae of pathway, including complex 1 (TSC1) spontaneous seizures. When applied after the induction epilepsy, variable intellectual and TSC2, STE20‑related kinase adaptor-α (STRADA), of status epilepticus, rapamycin decreased cell death, disability and autism. phosphatase and tensin homologue (PTEN), the gene mossy fibre sprouting and aberrant dentate granule cell

338 | MAY 2013 | VOLUME 14 www.nature.com/reviews/neuro

Box 2 | Temporal lobe epilepsy mechanisms might include inhibition of mossy fibre sprouting, neuroprotection and maintenance of the Mesial temporal lobe epilepsy (TLE; also known as limbic epilepsy), refers to an integrity of the blood–brain barrier35. electroclinical syndrome in which seizures emanate from the temporal lobe. TLE is a prominent and prevalent epilepsy syndrome in children, adolescents and adults. This Repressor element 1‑silencing transcription factor. REST syndrome is often associated with mesial temporal sclerosis — a pattern of hippocampal damage, including atrophy and gliosis, seen in humans with TLE and in is a transcription factor that binds to DNA-binding animal models of chronic TLE. TLE is typically considered to be an acquired epilepsy motifs called neuron-restrictive silencer elements syndrome that is triggered by a precipitating factor, such as head trauma, infection or (NRSEs; also known as RE1 elements). REST negatively tumour, or an episode of prolonged, uncontrolled seizure (status epilepticus), although regulates the expression of many neuronal genes in non- there are known familial forms of mesial TLE1–4,170,171. Lateral TLE172 is a distinct syndrome neuronal cells and neuronal precursor cells. It also reg- that will not be discussed in this Review. From a basic science perspective, TLE has ulates neuronal gene expression in mature neurons42,43. generally been considered as an acquired condition; likewise, temporal lobe REST binds to various co-repressors, which recruit his- epileptogenesis has been considered relevant to the acquired epilepsies only. tone deacetylase 1 (HDAC1) and HDAC2 (FIG. 1); these Animal models of acquired TLE have been a particularly useful tool in the study of deacetylases regulate chromatin structure and repress epileptogenesis, because the animal model resembles the human condition in many key the expression of hundreds of neuronal genes44. Almost respects — chronic TLE can be induced in rodents using various types of brain injury 2,000 genes have REST-binding motifs44 (although this that also cause TLE in humans. Previous studies have identified changes in molecular 45 signalling pathways in animal models of acquired TLE, including alterations in number could be much greater ), including ~10% of 46 immediate-early gene expression and the expression of various ion channels and all neuronally expressed genes , some of which encode neurotransmitter receptors. Proposed circuit-level mechanisms include synaptic proteins that are key regulators of neuronal excitability reorganization, axonal sprouting and the generation of new, pathological, recurrent and have been independently proposed to be involved excitatory synaptic connections (for example, mossy fibre sprouting). Cellular-level in epilepsy mechanisms47. Such genes include type A mechanisms include inflammation, neuronal cell death (including the specific loss of GABA (GABAA) receptor β3 subunit (GABRB3), GABAA mossy cells and vulnerable populations of inhibitory GABAergic interneurons), reactive receptor δ-subunit (GABRD) and the ionotropic AMPA2 astrocytosis and dispersion of existing and the generation of new dentate granule cells glutamate receptor (GRIA2), BDNF and genes encoding in ectopic locations. Some of these changes are likely to be involved in and perhaps hyperpolarization-activated cyclic nucleotide-gated (HCN) required for the pathogenesis of epilepsy, whereas others might be reparative of or channel compensatory for the initial injury and ongoing chronic epileptic condition, and still subunits 1–4 (HCN1–HCN4). other changes might simply be epiphenomenal. More direct evidence that alterations in REST signalling might be involved in epileptogenesis continues to accumulate48. Status epilepticus in animal models neurogenesis, all of which are histopathological features increases the expression of REST in the hippocampus Mossy fibre sprouting associated with TLE. However, in mice, seizure fre- and alters the histone acetylation of prominent neuronally A process thought to be 42,49 triggered by cell death of quency was unchanged by rapamycin treatment given expressed genes . Status epilepticus induces promoter- hilar mossy cells, leading to after pilocarpine-induced status epilepticus, although specific acetylation of BDNF that is inhibited by HDAC de‑innervation of dentate the degree of mossy fibre sprouting was decreased34. inhibitors49,50. In the kindling model, inhibition of glycoly- granule cell dendrites, with Mossy fibre sprouting was inhibited by rapamycin sis induces the binding of REST to the BDNF promoter, subsequent ‘sprouting’ of mossy fibre collaterals and given before or after status epilepticus in the kainate- or which prevents BDNF upregulation and modulates epi- 31–34 pathological granule cell pilocarpine-induced model of TLE in rats and after leptogenesis, providing an interesting mechanistic link autoinnervation. Whether status epilepticus induced by electrical stimulation of the between glycolysis, BDNF–TRKB signalling (see above) this is a cause or angular bundle35 but not in a model of TLE induced by and the REST system51. The ultimate effector (or effectors) consequence of temporal electrical stimulation of the amygdala36. However, upon of this signalling cascade remain unknown. lobe pathology in temporal lobe epilepsy is controversial. treatment withdrawal, the degree of sprouting returned to The role of HCN channels in epileptogenesis has the levels seen in untreated epileptic animals32, and sei- been reviewed in REF. 52. Dendritic HCN channels in Hemimegalencephaly zures recurred after treatment removal33. Hence, whereas hippocampal pyramidal cells are dysregulated after pro- A malformation of cortical manipulation of mTOR signalling in most rodent mod- longed febrile seizures or after kainate- or pilocarpine- development that is characterized by the els of acquired epilepsy appears to modulate seizure fre- induced status epilepticus in rodents, leading to what pathological enlargement of quency and decrease the degree of various measures has been termed an ‘acquired HCN channelopathy’ (to one cerebral hemisphere and is of TLE neuropathology, such as mossy fibre sprouting differentiate this from a genetic channelopathy)53–55. This highly associated with and hilar cell loss, mTOR pathway modulation in such phenomenon might be partly due to REST-mediated developmental delay and models has not been shown to prevent the develop- transcriptional repression of Hcn156. Binding of REST intractable epilepsy. ment of epilepsy permanently. It seems that continued to Hcn1 is increased after the induction of status epilep- Kainate- or rapamycin administration is required to maintain any ticus in a rodent model of TLE, resulting in decreased pilocarpine-induced status anti-epileptogenic effects. Some authors have referred HCN1‑mediated ionic currents (Ih) and altered func- epilepticus to the effects of rapamycin as ‘antiseizure’ or ‘epilepto- tion of CA1 pyramidal cells. Interference with REST Standard and widely used models of acquired chronic static’, given that rapamycin has not been shown to enact binding to target genes (including Hcn1) using an focal epilepsy in rodents in a prevention of epilepsy development that persists after anti-REST decoy oligodeoxynucleotide was associated which status epilepticus is rapamycin withdrawal40,41. Despite these exciting recent with decreased seizure frequency in rats after kainate- elicited via the administration developments and the clear role of mTOR pathway mol- induced status epilepticus. These examples show that of a chemoconvulsant, either ecules in genetic epilepsy syndromes, the mechanism (or status epilepticus might activate the REST pathway to the glutamate agonist kainic acid (kainate) or the muscarinic mechanisms) by which manipulation of mTOR signalling suppress important regulators of neuronal excitability acetylcholine receptor agonist after brain injury (such as that due to status epilepticus) that have been independently proposed to be involved pilocarpine. might modulate epileptogenesis remain unclear. These in pathogenic epilepsy mechanisms. It will be important

NATURE REVIEWS | NEUROSCIENCE VOLUME 14 | MAY 2013 | 339

a mTOR pathway b REST pathway is activation of the mTOR or REST pathways involved in driving epileptogenesis rather than being the result IRS1 PI3K PtdIns- Status of this process? What are the circuit-level effects of the (4,5)P epilepticus Insulin 2 manipulation of key molecular signalling cascades such receptor PTEN as the mTOR pathway and REST after status epilepticus PtdIns- or other brain injury? What are the relevant molecular (3,4,5)P 3 REST effectors of mTOR pathway blockade or HDAC inhibition STRADA in the prevention of epileptogenesis in animal models of LKB1 PDK1 acquired TLE? And how widespread is the relevance of these signalling cascades across various forms of epilepsy AMPK AKT beyond TLE? HDAC CoREST Modulation of Extensive research in animal models and in humans TSC2 REST gene expression after brain injury using a range of prophylactic anticon- TSC1 RE1 Target vulsants, anti-inflammatory agents and neuroprotec- sequence genes tive agents has failed to demonstrate anti-epileptogenic Rapamycin effects5,6. What hope is there that targeting the mTOR RHEB • HCN channel subunit HCN1* and REST pathways will meet with greater success? As • BDNF* • Glutamate receptors (for example, opposed to classical anticonvulsants, for example, which mTOR GRIA2*) act on one or a few discrete targets, modulation of mTOR • GABA receptors (for example, GABRB3 or REST would have broad action. As there are likely to and GABRD) • Potassium channels (for example, KCNC2 be multiple divergent mechanisms of epilepsy develop- Protein Microtubule Mitochondrial and KCNC3) ment, perhaps polytherapy cocktails that target multiple, synthesis assembly metabolism • Transporters (for example, KCC2) parallel pathways will prove effective. Using epileptogenic circuit-level dysfunction (as discussed below) will allow Figure 1 | Mechanistic roles of the mTOR pathway and REST in epileptogenesis. us to place in context the downstream effects of mTOR a | A simplified version of the mammalian target of rapamycin (mTOR) pathway, and/or REST system blockade on epilepsy development and facilitate translational efforts towards therapy highlighting epilepsy-related molecules. For a more completeNature Reviews illustration | Neuroscience of the mTOR pathway, see REF. 12. b | A schematic depicting repressor element 1 (RE1)-silencing development. transcription factor (REST)-mediated gene silencing triggered by status epilepticus. REST co-repressor 1 (CoREST), a component of the histone deacetylase (HDAC) Cell type-specific dysfunction and epilepsy complex, interacts with REST to mediate transcriptional repression. The asterisk Previous work studying the basic mechanisms of limbic indicates genes for which there is specific experimental evidence suggesting regulation epilepsy has identified dysfunction at the cellular level by REST or HDAC inhibiton after status epilepticus49,56. AMPK, AMP-activated protein that might suggest potential targeted anti-epileptogenic kinase; BDNF, brain-derived neurotrophic factor; GABRB3; type A GABA receptor 3 β interventions. Here, we discuss specific subtypes of cor- subunit; GABRD; type A GABA receptor δ-subunit; GRIA2, ionotropic AMPA 2 glutamate receptor; HCN1, hyperpolarization-activated cyclic nucleotide-gated channel subunit tical inhibitory GABAergic interneurons for which a 1; IRS1, insulin receptor substrate 1; KCC2; potassium/chloride co-transporter 2; KCNC, mechanistic role in epilepsy is increasingly being recog- potassium voltage-gated channel subfamily C; LKB1, liver kinase B1 (also known as nized. A larger discussion of cortical interneuron sub- serine/threonine kinase STK11); mTORC1, mTOR complex 1; PDK1, 3‑phosphoinositide- types is beyond the scope of this Review and has been 57–60 dependent kinase 1; PI3K, phosphoinositide 3‑kinase; PtdIns(4,5)P2, phosphatidylinosi- addressed recently . More general roles of GABAergic tol‑4,5‑biphosphate; PtdIns(3,4,5)P3, phosphatidylinositol‑3,4,5‑triphosphate; PTEN, inhibition in epilepsy mechanisms have been reviewed phosphatase and tensin homologue; RHEB, RAS homologue enriched in brain; STRADA, in REFS 61–64, including the emerging potential use of STE20‑related adaptor protein-α; TSC, tuberous sclerosis complex. Part a is modified, GABAergic interneuron transplantation as a therapeutic with permission, from REF. 11 (2011) Elsevier. © treatment modality for epilepsy65–67.

to determine which gene targets of REST are up- or Somatostatin-positive hilar interneurons. Immuno- downregulated by this pathway after status epilepticus reactivity for the neuropeptide somatostatin defines vari- or other brain injury. This will prove challenging given ous overlapping subtypes of GABAergic interneurons in the broad activity of this pathway across the genome. If a the cerebral cortex. In the dentate gyrus, most GABAergic certain REST target was identified as particularly impor- interneuron somata that are immunopositive for somato- tant in driving epileptogenesis, would it be possible to statin are located in the hilus68; approximately 50% of all selectively interrupt status epilepticus-induced, REST- hilar GABAergic interneurons are somatostatin-positive69. Hyperpolarization-activated mediated suppression of specific targets? Can the REST The synaptic terminal fields of somatostatin-positive hilar cyclic nucleotide-gated system be safely modulated in humans after brain injury interneurons are most prominent in the outer molecular (HCN) channel An ion channel that is activated (for example, with HDAC inhibitors)? Although long- layer of the dentate gyrus, where the fibres of the perforant 70,71 by hyperpolarization and term treatment with such agents might be accompanied path terminate ; in addition, the dendrites of these cells mediates a mixed cationic by unacceptable toxicity, perhaps a single treatment or a are targeted by recurrent collaterals of the granule cell- current known as I . It has 72 h short course of treatment given immediately after injury derived mossy fibres . In turn, somatostatin-positive ter- multiple important functions in would prove safe. minals ramify primarily with granule cell dendrites. Thus, the regulation of neuronal excitability. HCN channels are The mTOR and REST pathways are exciting new hilar somatostatin interneurons are ideally positioned to encoded by four genes potential targets for intervention in the epileptogenic modulate the influence of the perforant path on dentate (HCN1–HCN4). process, but specific questions remain. To what extent granule cells (FIG. 2).

340 | MAY 2013 | VOLUME 14 www.nature.com/reviews/neuro

a Loop structure of the b Circuit diagram: dentate gate c Circuit diagram: TA pathway hippocampus DG CA1 Molecular layer SO LPP CA1 Alveus LEC O-LM SC cell MPP RC CA3 SOM+ TA MEC SP PC Bi cell CA3 EC PP PV+ PV+ basket basket Granule SR cell layer GC CA3 MF PP PV+ DG basket

Hilus SOM+ HiPP cell SLM EC Mossy layer 3 cell TA pathway

CA3 Figure 2 | Embedded loop structures of the temporal lobe. a | The hippocampus receives two inputs, the perforant path (PP) and the temporoammonic (TA) pathway. The entorhinal cortex (EC) is in turn the target of hippocampal output, creating multiple excitatory loops. There is a long loop that originates from the EC and projects to the dentate gyrus (DG) via the PP, from the DG to area CA3 via mossy fibres (MFs), from area CA3 to area CA1Nature and Reviews from area | Neuroscience CA1 back to the subiculum (not shown) and/or EC. An intermediate-length loop originates from the EC and projects to area CA3 (via the PP pathway), then to area CA1 and from there to the subiculum and/or EC. A short loop projects from the EC to area CA1 (via the TA pathway) and back to the subiculum and/or EC. Recurrent collaterals (RCs) between CA3 pyramidal cells (PCs) are also shown. b | Simplified connectivity within the DG. Many connections and various inhibitory interneuron subtypes are omitted for clarity58,173,174. PCs and stellate cells within layer 2 of the medial EC (MEC) and lateral EC (LEC) form the PP to the middle and outer molecular layer, respectively, of the hippocampal DG, as well as to the distal dendrites of PCs in area CA3 (not shown)173. In the dentate, lateral PP (LPP) and medial PP (MPP) axons synapse onto dentate granule cells (GCs) and onto GABAergic interneurons (including parvalbumin-positive (PV+) basket cells and somatostatin-positive (SOM+) hilar PP-associated (HiPP) cells) as well as mossy cells in the dentate hilus. c | Selected connections of the TA pathway, which proceeds from layer 3 EC pyramidal and spiny stellate neurons (not shown) to the distal dendrites and distal apical tuft dendrites of CA1 neurons in stratum lacunosum-moleculare (SLM) in the short loop mentioned above as well as to the distal dendrites of CA2 (the intermediate-length loop, not shown). Extrinsic excitatory connections are shown in green; excitatory cells and intrinsic excitatory connections are shown in blue; and inhibitory cells and intrinsic inhibitory connections are shown in red and purple. Bi, bistratified cell; O‑LM, oriens lacunosum-moleculare. SC, Schaffer collateral; SO, stratum oriens; SR, stratum radiatum.

Selective loss of somatostatin-positive GABAergic Fast-spiking cells and epilepsy. Parvalbumin-positive hilar interneurons is a well-documented and consist- fast-spiking basket cells are key regulators of inhibition in ent finding in animal models of TLE73,74 and is found the cerebral cortex. Fast-spiking cells are the most numer- in the sclerotic hippocampus from humans with TLE75, ous subtype of cortical interneuron79,80. These cells form indicating that these neurons are highly and selectively powerful, proximally localized synapses on the soma and vulnerable to excitotoxic and ischaemic damage. In axon initial segment (AIS) of postsynaptic cells, exerting the kainate model of chronic TLE in mice, surviving fine control over the timing and probability of target cell somatostatin-positive hilar interneurons show axonal discharge. Fast-spiking cells mediate feedforward and Kindling sprouting to form new synapses with dentate granule feedback inhibition that is recruited by specific spatio- 76 81 A phenomenon whereby cells , which is consistent with increased somatosta- temporal patterns of excitatory activity . In addition, seizures themselves contribute tin immunoreactivity in the dentate molecular layer in fast-spiking cells form dense interconnected networks to epilepsy progression. The human TLE77. This might be a way to restore the com- through specific gap junctional coupling, which contrib- kindling model in experimental animals involves the generation promised levels of inhibition that are produced in part by utes to the function of these cells in the generation of fast of repeated, brief, focal seizures the loss of inhibitory interneurons. Interestingly, axonal oscillations within cortical structures. Modelling studies using electrical or chemical sprouting by somatostatin-positive hilar interneurons predict that perisomatically targeted inhibitory cell activ- stimulation, which subsequently is inhibited by rapamycin78. The consequences of the ity produces gamma oscillations82. Optogenetic activa- leads to permanently enhanced loss of somatostatin-positive hilar interneurons and the tion of parvalbumin-positive fast-spiking cells augments sensitivity to such stimuli and 83,84 longer, more severe seizures, changes that surviving neurons undergo in the context gamma-band oscillatory activity in the neocortex . and ultimately to chronic of hippocampal circuit function are subjects of ongoing In turn, dysfunction of interneuron-mediated gamma- epilepsy. investigation. band synchronization has been proposed to contribute

NATURE REVIEWS | NEUROSCIENCE VOLUME 14 | MAY 2013 | 341

to TLE mechanisms85. Specific optogenetic activation of mutation in Scn1a or bacterial artificial chromosome parvalbumin-positive interneurons in the hippocampus (BAC) transgenic mice expressing a mutation from a terminated seizures in a mouse model of TLE86. human patient with Dravet syndrome), although hap- Genetic manipulation of interneuron progenitors in loinsufficiency seems to be a common theme. Genetic the medial ganglionic eminence (MGE) that are des- background strain and developmental effects might also tined to become parvalbumin-positive fast-spiking cells affect the phenotype of mouse models of Dravet syn- causes epilepsy in mice87–89. In addition, genetic mutation drome. That interneuron-specific Scn1a knockout reca- or experimental manipulation of various ion channel pitulates the phenotype of the global null mutation shows genes that are prominently or exclusively expressed in that Scn1a deletion in cortical interneurons is sufficient fast-spiking cells (including the voltage-gated potassium to produce a Dravet-like phenotype in mice; however, channel genes KCNA1 (REFS 90–93) and KCNC2 (REF. 93)) this does not prove that forebrain interneurons are the causes epilepsy in humans and in mouse models. necessary and exclusive locus of cellular dysfunction in

Mutations of SCN1A (which encodes the voltage-gated Dravet syndrome or that the absence of NaV1.1 in fast-

sodium channel subunit NaV1.1) causes a spectrum of spiking cells is the mechanism of Dravet syndrome. As related epilepsy syndromes, including generalized epi- mentioned above, many perturbations directed towards lepsy with febrile seizures plus (GEFS+) and Dravet syn- fast-spiking cells produce severe epilepsy, at least in drome94–96. Initial functional studies of SCN1A mutations mouse models, so it is not surprising that deletion of a associated with GEFS+ yielded inconsistent results; how- major sodium channel that underlies action potential ever, similar studies of the more severe Dravet syndrome generation in these cells will cause hyperexcitability in phenotype showed a loss of function caused by haploin- cortical circuits with resultant epilepsy. Furthermore, sufficiency97. The loss of function caused by haploinsuf- although the SCN1A mutation is the most common cause ficiency was explained by evidence showing that mutation of Dravet syndrome, accounting for approximately 70% of SCN1A generates epilepsy by a selective reduction in of cases95,96, mutations in other genes can cause a Dravet the sodium current in cortical inhibitory GABAergic or Dravet-like phenotype. These genes include GABRG2 interneurons relative to pyramidal cells, which leads to (REF. 111), the gene encoding sodium channel-β1 subunit 112,113 114,115 decreased action potential firing frequency in interneu- (SCN1B) , the gene encoding NaV1.2 (SCN2A) rons98–103. This was a very attractive hypothesis as to the and protocadherin 19 (PCDH19)116, suggesting that mechanism of Dravet syndrome, given the important role multiple genetic mechanisms can produce Dravet or of fast-spiking cells in the regulation of cortical inhibi- Dravet-like syndrome. Whether these genetic mutations tion. The sodium current might be particularly reduced converge on fast-spiking cell dysfunction is unknown. in cerebral cortical parvalbumin-positive fast-spiking Dravet syndrome spectrum disorders might rep-

interneurons in which NaV1.1 appears to be prominently resent a point of convergence between acquired and and perhaps differentially expressed relative to pyramidal genetic mechanisms of epilepsy. Mutations of SCN1A cells and parvalbumin-negative interneurons104. Further (or other Dravet syndrome spectrum genes) generate a supporting this theory, interneuron-specific deletion of strong genetic predisposition to febrile and afebrile sei- Scn1a in the forebrain using Dlx1/2 Cre transgenic mice zures in humans. Patients with Dravet syndrome are not causes an epilepsy phenotype that is similar to global het- born with epilepsy, just as mice with Scn1a (or Scn1b117) erozygous deletion of Scn1a105. In addition, pathology in mutations do not exhibit spontaneous seizures until the these mice begins to appear at the same time that fast- late second or early third postnatal week. Like patients spiking interneurons acquire a mature electrophysiologi- with Dravet syndrome spectrum disease, mouse models cal phenotype106,107, which might be the developmental initially show temperature-sensitive seizures before the time point at which cortical structures begin to rely on emergence of spontaneous seizures98,100,104. As described fast-spiking cells as a major source of inhibition. However, above, prolonged febrile and afebrile seizures (particu- a published voltage-clamp recording of sodium currents larly status epilepticus) can trigger a secondary cascade from interneurons and principal cells in mouse models of of events that might compound the pathology, leading to Dravet syndrome exists only for dissociated hippocampal neuronal plasticity and epileptogenesis118,119. Interestingly, neurons in culture98 and neurons from the visual cortex104. focal TLE has been reported in patients with the SCN1A 120,121 Previous data showed that NaV1.1 is expressed at the mutation and GEFS+ . TLE has also been reported AIS, soma and dendrites of pyramidal cells108,109, but more in a small group of patients with GEFS+ that is due to

recent findings in adult rats suggest that NaV1.1 expres- the SCN1B mutation, including one patient without sion in the neocortex and hippocampus is restricted to known history of prolonged febrile seizure or status the AIS of parvalbumin-positive interneurons110. This epilepticus who had TLE with mesial temporal sclerosis unresolved issue complicates the interpretation of the (BOX 2) and who underwent successful temporal lobec- proposed cell type-specific dysfunction in mouse mod- tomy122. Of further interest, transgenic mice expressing els of Dravet syndrome. Such dysfunction might reflect a gain‑of‑function Scn2a mutation have epilepsy and 123 true cell type-specific differences in NaV1.1 expression mesial temporal sclerosis . Although MRI scans of or differential compensation by other sodium channel patients with Dravet syndrome are typically normal ini- genes in pyramidal cells as compared with interneurons. tially, hippocampal sclerosis develops over time in some This compensation might occur differently in the vari- patients124–126. This concept of epileptogenesis, linking ous mouse models of Dravet syndrome (such as Scn1a sodium channel mutations, febrile seizures, status epilep- knockout mice, knock-in mice harbouring a nonsense ticus and TLE, blurs the distinction between the classical

342 | MAY 2013 | VOLUME 14 www.nature.com/reviews/neuro

mechanisms of epilepsy that are driven either by acquired loops are important for normal hippocampal network brain injury or genetic mutation. Hence, it might be pos- operations but also render the temporal lobe particu- sible to intervene medically before the emergence of larly vulnerable to the generation of abnormal electrical seizures in at‑risk patients, in whom risk is defined by activity and underlie the predisposition of temporal lobe vulnerable genetic background in addition to exposure to structures to epileptogenesis132,133. Here, we focus our brain injury. Possible interventions might include dietary discussion on two important discrete circuit elements therapies (for example, the ketogenic diet127), stem cell- in the temporal lobe, both of which have been proposed based therapies to replace a dysfunctional cell type or tar- to be involved in the pathology of TLE: the dentate gate geted pharmacological or optogenetic agents tailored to between the entorhinal cortex and area CA3, and the a specific neuronal cell type. temporoammonic pathway between the entorhinal cor- The mechanisms by which the SCN1A mutation leads tex and area CA1. Large-scale dynamic imaging has pro- to epilepsy are highly controversial and are the subject of vided insights into the normal operation and dysfunction intense ongoing investigation. Recent data on Scn1b null of these circuit elements in epilepsy. mice might further complicate the picture. These mice show abnormal neuronal proliferation, neuronal migra- The dentate gate in limbic epileptogenesis. The dentate tion and axonal targeting at early developmental time gyrus has been a particular focus in the study of TLE for points, before the appearance of seizures128, suggesting various reasons. The dentate gyrus is the locus for the that neurodevelopmental abnormalities might contribute classical temporal lobe circuit rearrangement of mossy to epilepsy in these mice. As Scn1b regulates the physi- fibre sprouting, which is seen in human TLE and in ology, surface expression and subcellular targeting of animal models134,135; sprouting also occurs from CA3 Scn1a, perhaps Dravet syndrome represents a neurode- pyramidal cells and mossy cells in the dentate hilus in velopmental disorder of which epilepsy is but one promi- animal models of TLE136. The time course of the appear- nent component. That said, the delayed appearance of ance of mossy fibre sprouting, which occurs following epilepsy does not necessarily imply an underlying epilep- a delay after brain injury, is concordant with the latent togenic process; it could be that the circuit elements that period of temporal lobe epileptogenesis. However, the underlie seizure generation are simply not mature until role of mossy fibre sprouting in epileptogenesis remains a particular age (see above). This is clear from the fact controversial135. There are additional anatomical altera- that seizures appear days, months or years after birth in tions in this region in TLE, including dispersion of den- Dravet syndrome and in other genetic epilepsies. We sug- tate granule cells, abnormal granule cell neurogenesis, gest that a more complete understanding of these issues the appearance of granule cells in ectopic locations will require further investigation of the roles of fast-spik- and the loss of mossy cells and vulnerable populations ing cells in the normal function of cortical circuits and of inhibitory interneurons in the dentate gyrus and the mechanisms and consequences of fast-spiking cell hilar region. dysfunction in circuit operations in Scn1a mutant mice. Under normal conditions, the dentate gyrus strongly filters afferent input, relaying only a small component Circuit dysfunction in epileptogenesis of incoming, synchronous cortical synaptic inputs to its Seizures are network-level phenomena, and the incom- downstream target, hippocampal area CA3. This phe- plete knowledge of neuronal circuit function has left a nomenon has been termed the ‘dentate gate’132,133,137. gap in our understanding of how disruption at a molec- Dentate gating is thought to allow the dentate gyrus to ular or cellular level generates epilepsy in the intact filter entorhinal cortex input through the perforant path organism. We argue that bridging this gap will require to the hippocampus, selecting inputs and enacting a large-scale, dynamic, circuit-level analysis. Ongoing ‘sparse coding’ scheme that is thought to be important research in animal models of TLE has been facilitated for the computational functions of the dentate gyrus in by recent advances in network-level functional imag- pattern separation138. Dentate granule cells also have roles ing techniques such as voltage-sensitive dye (VSD) and in the modulation of CA3 place cell behaviour138 and in calcium imaging, which allow simultaneous monitor- cognitive processes such as learning and memory139,140. In ing of the activity of large numbers of neurons at high addition, the gate is thought to protect downstream area temporal resolution. The additional use of optogenetic CA3, which is vulnerable to metabolic challenges, from techniques allows researchers to manipulate defined excitotoxic damage that could be induced by aberrant subsets of neurons within these circuits with millisec- hypersynchronous entorhinal cortical activity. A signifi- ond precision86,129,130. For example, in an animal model cant contributor to this emergent network phenomenon is of post-stroke epilepsy, HCN channel dysregulation ren- the small percentage of dentate gyrus granule cells that are ders thalamic relay neurons hyperexcitable, and optoge- activated by perforant path stimulation (2–5% in vitro141). netic silencing of these cells interrupts seizures131. These A similarly small percentage (2%) of dentate granule cells techniques are ideally suited for the study of circuit-level are active during spatial navigation in animals in vivo142. phenomena such as seizures and the circuit-level dys- For these reasons, the assessment of gating is important functions that underlie epilepsy. as a network-level output measure of dentate gyrus func- Within the temporal lobe, there is an intimate rela- tion under normal conditions and of circuit dysfunction tionship between the hippocampus and entorhinal cortex in epilepsy. This phenomenon can readily be shown — two potentially epileptogenic brain regions — formed electrophysiologically in vitro, where epileptiform activ- by multiple re‑entrant circuit loops (FIG. 2). These synaptic ity in the entorhinal cortex induced by pharmacological

NATURE REVIEWS | NEUROSCIENCE VOLUME 14 | MAY 2013 | 343

Dentate gate TA pathway Aa Ab Ac Ba Current-clamp Bb VSD traces Current-clamp SO SR SLM Current-clamp Peak responses 20 ms TA stimulation 10 mV 0.1% nACSF 0.2% 10 mV Dendritic SO 50 ms VSD SR SO DG Hilus Field SLM SR * CA3 SLM 50 ms 0.5 mV 5 ms * = 30 ms %ΔF/F 0.1 0 –0.1 Ad Ae Af Bc Bd Current-clamp Current-clamp Current-clamp 10 mV 0.1% 5 μM PTX Gabazine (1 μM) and TA stimulation TA pathway 10 mV CGP 558545A (2 μM) 0.2% Dendritic

VSD DG SO Hilus Field SR ** CA3 0.5 mV SLM 50 ms 5 ms 50 ms ** = 40 ms %ΔF/F 0.1 0 –0.1 Figure 3 | Temporal lobe circuit elements under normal conditions. A | Dentate gating in the hippocampus under normal conditions (parts Aa–Ac) and after blockade of GABAergic inhibition (parts Ad–Af). Aa | A voltage-sensitive dye (VSD) image of depolarization that is restricted to the dentate gyrus granule cell layer in response to stimulation of the perforant path in an acute brain slice from the rat hippocampus. Ab | The VSD response is quantified in the dentate gyrus, hilus and area CA3. Ac | The top panel shows a current-clamp recording from a dentate gyrus granule cell and the botttom panel shows the field potential in the dentate gyrus. Ad | The collapse of the dentate after blockade of

GABAergic inhibition using 5 µM picrotoxin (PTX; a non-competitive type A (GABAA) receptor antagonist) is illustrated using VSD imaging. Ae | The quantification of the VSD response shows collapse of the dentate gate, with activation of upstream area CA3 after GABAergic blockade. Af | Blockade of inhibition elicits hyperactivation of dentate granule cells in response to perforant path activation, as shown by the current-clamp recording (top panel) and the field potential (bottom panel). B | Temporoammonic (TA) pathway function in area CA1 under normal conditions (parts Ba,Bb) and following GABAergic blockade (parts Bc,Bd). Ba | Activation within area CA1 in response to stimulation of the TA pathway is spatially restricted to the distal dendrites of CA1 pyramidal cells (indicated by the arrows), which is shown using VSD imaging. Bb | The top panel shows a current-clamp recording from a CA1 pyramidal cell dendrite in response to TA pathway activation, and the bottom panel shows the quantification of the VSD response in stratum oriens (SO), stratum radiatum (SR) and stratum lacunosum-moleculare (SLM) 30 ms after extracellular stimulation of the TA pathway, with evoked excitatory responses (indicated by the asterisk) in SLM. Bc | A VSD image shows that GABAergic

feedforward inhibition after application of the GABAA antagonist gabazine (1 μm) and the GABAB antagonist CGP 55845A (2 μm) spatially restricts TA pathway activation to the distal dendrites of CA1 pyramidal cells, with loss of spatial segregation within area CA1. Bd | The top panel shows a current-clamp recording from a CA1 pyramidal cell dendrite, and the theNature bottom Reviews panel shows | Neuroscience the quantification of the VSD response, with propagation of the response to SR and SO. F, fluorescence; nACSF, normal artificial cerebrospinal fluid. Part A is reproduced, with permission, from REF. 137 © (2007) Elsevier. Part B is reproduced, with permission, from REF. 154 © (2005) Society for Neuroscience.

manipulation does not propagate through the dentate research, as protecting or bolstering the gate after brain gyrus to area CA3 under control conditions (FIG. 3) but injury or the induction of status epilepticus could theo- does so, for example, after blockade of GABAergic inhi- retically modulate epileptogenesis. bition (FIG. 3) or in the kindled hippocampus143. Dentate The mechanisms of dentate gating are probably gating has also been demonstrated using large-scale VSD diverse but are likely to include the intrinsic electro- imaging of the dentate: stimulation of the perforant path physiological properties of dentate granule cells, unique produced an excitatory response in the inner molecular integrative features of granule cell dendrites145 and mul- layer of the dentate gyrus that subsequently spread to tiple forms of inhibition that operate within the dentate the granule cell layer of the dentate gyrus as well as the gyrus. There is potent feedforward and feedback inhi- proximal hilus but did not activate area CA3 (REF. 144). bition of perforant path-evoked dentate gyrus granule Understanding the mechanisms that underlie the break- cell activity (FIG. 2), and granule cells are also subjected down of dentate filter function is a goal of ongoing to powerful tonic inhibition. Although the granule cells

344 | MAY 2013 | VOLUME 14 www.nature.com/reviews/neuro

a Dentate gate during the latent period themselves rarely fire action potentials in response to Control perforant path activation, resident GABAergic interneu- DG rons are also targets of perforant path input (for exam- Hil ple, parvalbumin-positive fast-spiking basket cells in 146 CA3 the dentate gyrus granule cell layer and, perhaps, somatostatin-positive cells in the hilus147) and have 1mm been shown to activate readily, thereby constraining 5 ms 45 ms 200 ms granule cell activity through feedforward and feedback Latent inhibition. There is accumulating evidence that transient dentate gate dysfunction develops during the latent phase of TLE in rodent models. Such data suggest that this dysfunc-

tion involves GABAA receptor alterations, dysregulation of the maintenance of chloride gradients and acute loss 5 ms 45 ms 200 ms of GABAergic interneurons in the dentate hilus. The b Dentate gate in chronic epilepsy generation of acute brain slices prepared from epi- Control leptic rodents after the induction of status epilepticus DG combined with large-scale, high-speed dynamic opti- CA3 cal imaging of population activity has proven to be a powerful approach to the elucidation of circuit-level mechanisms of epileptogenesis. VSD imaging in rodent brain slices showed network- Epileptic 0.08 level dysfunction of the dentate gate during the latent 0.045 period of chronic TLE148. Dentate-mediated filtering of %ΔF/F perforant path activation, as measured by the depolarization of upstream area CA3, was compromised during the –0.045 latent period but recovered during the chronic phase of –0.08 TLE (FIG. 4). Hence, the dentate gate is ‘open’ for a brief 5 ms 25 ms 200 ms period of time corresponding to the latent period. One c TA pathway in chronic epilepsy mechanism underlying the breakdown of the dentate gate Control 0.08 is a depolarizing shift of the reversal potential for GABA-

SR 0.045 mediated postsynaptic potentials (EGABA) in dentate gyrus 148 %ΔF/F granule neurons : EGABA is typically very hyperpolarized –0.045 in granule cells under baseline conditions but shifts to markedly less hyperpolarized levels 1 week after the 5 ms 25 ms 55 ms 85 ms 180 ms –0.08 induction of status epilepticus, returning again to base- Eplileptic 0.08 line levels in the chronic phase of TLE. The basis of this SR 0.045 transient depolarizing shift in EGABA is decreased expres- %ΔF/F sion of potassium/chloride co-transporter 2 (KCC2; also 5 ms 25 ms 55 ms 85 ms 180 ms –0.045 known as SLC12A5), which normally extrudes chloride –0.08 from cells and sets EGABA at very hyperpolarized levels in granule cells; the result is altered synaptic integration Figure 4 | Circuit dysfunction in temporal lobe epilepsy. a | Dentate gating is in dentate granule cells and transient pathological hyper- impaired during the latent period (1 week after status epilepticus) in an animal model of excitability in response to perforant path activation. The chronic temporal lobe epilepsy (TLE). The top panel shows voltage-sensitive dye (VSD) developmental ‘GABA switch’ that renders GABA hyperpo- recordings of perforant path stimulation under control conditionsNature 5 Reviews ms after | stimulationNeuroscience (left), at peak (middle) and later (200 ms; right). The top right panel shows restriction of larizing involves the upregulation of KCC2 and the down- the perforant path-evoked depolarization to the molecular layer of the dentate gyrus regulation of sodium/potassium/chloride co‑transporter 1 (DG). Note that there is minimal area CA3 activation. During the latent period (bottom (NKCC1; also known as SLC12A2)149,150. The above obser- panel), identical stimulation reveals a delayed peak response as well as >200% increase vation is interesting in the context of the finding that KCC2 in activation (as measured by areal pixel activation) in area CA3. b | Gating function is expression can be suppressed by REST151 (as well as by retained in the chronic phase of acquired TLE in the rodent model. The two panels show neurotrophins152), possibly linking a molecular signalling VSD images of the DG 5 ms (left), 25 ms (middle) and 200 ms (right) after stimulation of cascade with the transient dentate gate dysfunction that is the perforant path in control rats (top panel) and in chronically epileptic rats (bottom seen during the latent period. panel). Under both conditions, there is a strong response in the DG that fails to propagate The potential anti-epileptogenic properties of the to area CA3. c | Temporoammonic (TA) pathway dysfunction in the chronic phase of NKCC1 antagonist bumetanide have been tested in the acquired TLE. VSD imaging of TA pathway function in control animals (top panel) and epileptic animals (bottom panel) at 5 ms, 25 ms, 55 ms, 85 ms and 180 ms after stimulation pilocarpine-induced status epilepticus model of TLE in of the TA pathway, producing spatially restricted activity in control animals that rats. Bumetanide, when administered alone after status aberrantly propagates to the strata radiatum and pyramidale in the epileptic condition. epilepticus, had no effect on the subsequent develop- Hil, hilus; F, fluorescence. Part a is reproduced, with permission, from REF. 148 © (2007) ment and progression of epilepsy, and the addition of Society for Neuroscience. Parts b and c are reproduced, with permission, from REF. 144 © bumetanide to a cocktail of diazepam and phenobarbi- (2006) Society for Neuroscience. tal administered after status epilepticus did not produce

NATURE REVIEWS | NEUROSCIENCE VOLUME 14 | MAY 2013 | 345

any additional effect on seizure frequency beyond that of above. Surprisingly, although dentate gate function diazepam and phenobarbital alone153. The development is impaired during the latent period, it recovers in the of other strategies to protect the dentate gate after status chronic phase of TLE in a rodent model144 (FIG. 4). In addi- epilepticus are needed to test the hypothesis that mainte- tion, activation of area CA1 is actually decreased in the nance of dentate gate function during the latent period chronic phase of TLE in response to the stimulation of could be anti-epileptogenic. the Schaffer collateral pathway. However, it is the processing of entorhinal cortex input by area CA1 that is Temporoammonic pathway dysregulation in limbic epi‑ markedly impaired in chronic TLE. Excitation in area leptogenesis. Under normal conditions, the excitatory CA1 that occurred in response to the stimulation of the postsynaptic potential (EPSP) generated by temporo- temporoammonic pathway was increased tenfold in ammonic input to area CA1 is restricted to the apical chronic TLE144. Combined, these dynamic changes cre- tuft of CA1 pyramidal cells by GABAergic inhibition154, ate a transient breakdown of the dentate gate, which is does not propagate to the cell soma154,155 and hence does followed by long-standing pathological hyperexcitability not typically elicit action potential firing154,155. Instead, within the temporoammonic pathway that allows aber- the soma primarily responds to temporoammonic rant entorhinal cortical input to enter the hippocampus pathway activation with an inhibitory postsynaptic through the disinhibited temporoammonic pathway. potential, which reflects proximally targeted disynaptic Thus, circuit-level analysis allows network excitability inhibition. This response of area CA1 to temporoam- to be compared between control and epileptic brains monic pathway stimulation is another circuit-level out- at multiple time points in the epileptogenic process, put measure of a component of the hippocampal circuit integrating the multitude of changes that occur during that can be used to assay large-scale network function limbic epileptogenesis and generating a large-scale phys- (or dysfunction). The mechanisms that underlie this iological output measure that correlates with changes in phenomenon include a high density of HCN channels the intact organism. in distal dendrites of CA1 hippocampal pyramidal cells156,157 and disynaptic feedforward inhibition that Conclusions might originate from oriens lacunosum-moleculare Here, we have presented a selected review of recent GABA switch In the developing brain, GABA (O‑LM) interneurons. Pharmacologic blockade of work on the mechanisms of epileptogenesis, citing spe- is an excitatory GABAergic inhibition results in the loss of this spatial cific examples chosen from a set of emerging themes neurotransmitter and segregation of temporoammonic excitatory input in the in the field, including dysregulation of molecular sig- depolarizes neurons. As the stratum lacunosum-moleculare, with subsequent propaga- nalling pathways, cell type-specific dysfunction and brain matures, a shift in the tion of temporoammonic EPSPs to the stratum radiatum pathology of discrete circuit elements. Some of the chloride potential of neurons stratum oriens154 causes the effects of GABA to and . Correspondingly, feedback inhibi- issues that have been discussed remain quite contro- become hyperpolarizing and tion driven by stimulation of GABAergic interneurons versial. Although we have largely restricted our discus- hence inhibitory. in the stratum oriens further constrains temporoam- sion to mechanisms of TLE and Dravet syndrome, this monic pathway-evoked EPSPs in CA1 hippocampal focus can and should be expanded to other brain areas Stratum lacunosum- moleculare pyramidal cells. and models of other epilepsy syndromes, both acquired A hippocampal layer that is Temporoammonic pathway dysfunction in rodent and genetic. We argue that the roles of individual mol- located superficially relative models of acquired chronic TLE results from disinhibi- ecules, particular cell types or a large-scale molecular to the stratum radiatum tion owing to acute loss of interneurons after status epi- signalling cascade in epileptogenesis will be best under- beneath the pial surface and lepticus, upregulation of T‑type calcium currents (I ) stood in the context of a circuit analyses. This can be contains the perforant path Ca,T input onto distal dendritic in CA1 hippocampal pyramidal cells and downregu- seen in the overlap between the themes that we have 158,159 tufts of CA1 pyramidal cells. lation of dendritic Ih .Overall, dendritic inhibition discussed. is decreased in CA1 hippocampal pyramidal cells in Epilepsy prevention was defined as Benchmark Stratum radiatum rodent models of TLE160. The A‑type potassium current Area I of the 2007 National Institute of Neurological A hippocampal layer located just superficially relative to the (IA) is downregulated in CA1 hippocampal pyramidal Disorders and Stroke (NINDS) Epilepsy Research 168 stratum lucidum in area CA3 cell dendrites in the pilocarpine model of chronic TLE Benchmarks . The data highlighted here illustrate 53 and to the pyramidal cell layer in rats . In the same model, ICa,T appears to be upregu- new avenues that hold promise of achieving this goal. in areas CA2 and CA1 that lated in dendrites of CA1 hippocampal pyramidal cells, Theorized mechanisms of epileptogenesis have moved contains the Schaffer collateral which leads to a change in firing mode from tonic to beyond alterations of ion channels and neurotransmit- axons from areas CA3 to CA1 159,161–163 and the dendrites of CA1 bursting . It is thought that CA1 hippocampal ter receptors to include cell type-specific dysfunction pyramidal cells. pyramidal neurons probably possess an intrinsic burst such as ‘interneuronopathies’ as well as dysregulation of

mechanism mediated by ICa,T that is normally sup- master regulatory pathways such as REST and mTOR. Stratum oriens pressed by I ; however, downregulation of dendritic I We suggest that these various themes all converge at the A relatively cell-free layer in the A A CA subfields of the in combination with upregulation of ICa,T unmasks and level of brain circuits, the defining neurophysiological 159 hippocampus that is located amplifies aberrant bursting in these cells . mediators of epilepsy. However, significant questions below the pyramidal cell layer Among temporal lobe structures, the dentate gyrus face the field as we move forward, and we have raised that contains the axons of is the least prone to the generation of spontaneous epi- many more questions than we have answered. This hippocampal pyramidal cells, leptiform discharges in vitro164. Inhibition in the dentate is perhaps necessarily the case as we are still without along with various types of GABAergic interneurons gyrus is enhanced rather than impaired in the chronic anti-epileptogenic therapies, although we do have new 165 166,167 including oriens-lacunosum phase in the kainate and pilocarpine rat models and qualitatively different treatment targets. Future moleculare interneurons. of TLE165, which is consistent with the results described research directions should explore the effects of genetic

346 | MAY 2013 | VOLUME 14 www.nature.com/reviews/neuro

manipulations, dysfunction of defined cell types and and therapeutic intervention in the epileptogenic pro- alterations in key cell signalling pathways on circuit cess. This concept provides support for the idea that function. Such analyses establish a framework within epileptogenesis is a clinical problem that is amenable which we can analyse potential molecular and cellular to medical intervention and hope for a future in which targets for pharmacological and genetic manipulation epilepsy can be prevented.

1. Helbig, I., Scheffer, I. E., Mulley, J. C. & Berkovic, S. F. 22. Holmes, G. L. & Stafstrom, C. E. Tuberous sclerosis antiepileptogenic target. Epilepsia 53, 1119–1130 Navigating the channels and beyond: unravelling the complex and epilepsy: recent developments and future (2012). genetics of the epilepsies. Lancet Neurol. 7, 231–245 challenges. Epilepsia 48, 617–630 (2007). 42. Palm, K., Belluardo, N., Metsis, M. & Timmusk, T. (2008). 23. Mohamed, A. R. et al. Intrinsic epileptogenicity of Neuronal expression of zinc finger transcription factor 2. Berkovic, S. F. et al. Familial temporal lobe epilepsy: cortical tubers revealed by intracranial EEG REST/NRSF/XBR gene. J. Neurosci. 18, 1280–1296 a common disorder identified in twins. Ann. Neurol. monitoring. Neurology 79, 2249–2257 (2012). (1998). 40, 227–235 (1996). 24. Uhlmann, E. J. et al. Astrocyte-specific TSC1 43. Ballas, N., Grunseich, C., Lu, D. D., Speh, J. C. & 3. Kobayashi, E., Li, L. M., Lopes-Cendes, I. & Cendes, F. conditional knockout mice exhibit abnormal neuronal Mandel, G. REST and its corepressors mediate Magnetic resonance imaging evidence of hippocampal organization and seizures. Ann. Neurol. 52, 285–296 plasticity of neuronal gene chromatin throughout sclerosis in asymptomatic, first-degree relatives of (2002). neurogenesis. Cell 121, 645–657 (2005). patients with familial mesial temporal lobe epilepsy. 25. Zeng, L.‑H. et al. Abnormal glutamate homeostasis 44. Johnson, D. S., Mortazavi, A., Myers, R. M. & Wold, B. Arch. Neurol. 59, 1891–1894 (2002). and impaired synaptic plasticity and learning in a Genome-wide mapping of in vivo protein–DNA 4. Cendes, F., Lopes-Cendes, I., Andermann, E. & mouse model of tuberous sclerosis complex. interactions. Science 316, 1497–1502 (2007). Andermann, F. Familial temporal lobe epilepsy: a Neurobiol. Dis. 28, 184–196 (2007). 45. Otto, S. J. et al. A new binding motif for the clinically heterogeneous syndrome. Neurology 50, 26. Zeng, L.‑H., Xu, L., Gutmann, D. H. & Wong, M. transcriptional repressor REST uncovers large gene 554–557 (1998). Rapamycin prevents epilepsy in a mouse model of networks devoted to neuronal functions. J. Neurosci. 5. Loscher, W. & Brandt, C. Prevention or modification of tuberous sclerosis complex. Ann. Neurol. 63, 27, 6729–6739 (2007). epileptogenesis after brain insults: experimental 444–453 (2008). 46. Mortazavi, A., Leeper Thompson, E. C., Garcia, S. T., approaches and translational research. Pharmacol. Rev. 27. Pun, R. Y. K. et al. Excessive activation of mTOR in Myers, R. M. & Wold, B. Comparative genomics 62, 668–700 (2010). postnatally generated granule cells is sufficient to modeling of the NRSF/REST repressor network: from 6. Pitkänen, A. & Lukasiuk, K. Mechanisms of cause epilepsy. Neuron 75, 1022–1034 (2012). single conserved sites to genome-wide repertoire. epileptogenesis and potential treatment targets. Selective deletion of PTEN from a small subset of Genome Res. 16, 1208–1221 (2006). Lancet Neurol. 10, 173–186 (2011). dentate gyrus granule cells in adult mice is 47. Roopra, A., Huang, Y. & Dingledine, R. Neurological 7. McNamara, J. O. & Huang, Y. Z. & Leonard, A. S. sufficient to generate TLE. disease: listening to gene silencers. Mol. Interv. 1, Molecular signaling mechanisms underlying 28. Lee, J. H. et al. De novo somatic mutations in 219–228 (2001). epileptogenesis. Sci. STKE 2006, re12 (2006). components of the PI3K–AKT3–mTOR pathway cause 48. Roopra, A., Dingledine, R. & Hsieh, J. Epigenetics and 8. Binder, D. K. Neurotrophins in the dentate gyrus. hemimegalencephaly. Nature Genet. 44, 941–945 epilepsy. Epilepsia 53, 2–10 (2012). Prog. Brain Res. 163, 371–397 (2007). (2012). 49. Huang, Y., Doherty, J. J. & Dingledine, R. Altered 9. McNamara, J. O. & Scharfman, H. E. in Jasper’s Basic 29. Poduri, A. et al. Somatic activation of AKT3 causes histone acetylation at glutamate receptor 2 and brain- Mechanisms of the Epilepsies 4th edn hemispheric developmental brain malformations. derived neurotrophic factor genes is an early event (eds Noebels, J. L., Avoli, M., Rogawski, M. A., Neuron 74, 41–48 (2012). triggered by status epilepticus. J. Neurosci. 22, Olsen, R. W. & Delgado-Escueta, A. V.) 514–531 30. Tokuda, S. et al. A novel Akt3 mutation associated with 8422–8428 (2002). (Oxford Univ. Press, 2012). enhanced kinase activity and seizure susceptibility in 50. Huang, Y., Myers, S. J. & Dingledine, R. Transcriptional The classic textbook on basic mechanisms of mice. Hum. Mol. Genet. 20, 988–999 (2011). repression by REST: recruitment of Sin3A and histone epilepsy and epileptogenesis, which is now in its recently- 31. Zeng, L.‑H., Rensing, N. R. & Wong, M. The deacetylase to neuronal genes. Nature Neurosci. 2, released fourth edition. mammalian target of rapamycin signaling pathway 867–872 (1999). 10. Wong, M. Mammalian target of rapamycin (mTOR) mediates epileptogenesis in a model of temporal lobe 51. Garriga-Canut, M. et al. 2‑Deoxy‑D‑glucose reduces inhibition as a potential antiepileptogenic therapy: epilepsy. J. Neurosci. 29, 6964–6972 (2009). epilepsy progression by NRSF-CtBP-dependent from tuberous sclerosis to common acquired 32. Buckmaster, P. S., Ingram, E. A. & Wen, X. Inhibition of metabolic regulation of chromatin structure. Nature epilepsies. Epilepsia 51, 27–36 (2010). the mammalian target of rapamycin signaling pathway Neurosci. 9, 1382–1387 (2006). 11. Crino, P. B. mTOR: a pathogenic signaling pathway in suppresses dentate granule cell axon sprouting in a Pharmacologic inhibition of glycolysis during kindling developmental brain malformations. Trends Mol. Med. rodent model of temporal lobe epilepsy. J. Neurosci. epileptogenesis recruits REST to target genes 17, 734–742 (2011). 29, 8259–8269 (2009). including BDNF, leading to suppression of BDNF A prescient review of diverse types of focal brain 33. Huang, X. et al. Pharmacological inhibition of the expression, reduced neuronal excitability and malformation, suggesting that a spectrum of mammalian target of rapamycin pathway suppresses attenuated progression of kindling. disorders are due to aberrant mTOR pathway acquired epilepsy. Neurobiol. Dis. 40, 193–199 52. Noam, Y., Bernard, C. & Baram, T. Z. Towards an signalling (‘TORopathies’). (2010). integrated view of HCN channel role in epilepsy. Curr. 12. Laplante, M. & Sabatini, D. M. mTOR signaling in 34. Buckmaster, P. S. & Lew, F. H. Rapamycin suppresses Opin. Neurobiol. 21, 873–879 (2011). growth control and disease. Cell 149, 274–293 (2012). mossy fiber sprouting but not seizure frequency in a 53. Bernard, C. et al. Acquired dendritic channelopathy in 13. Qureshi, I. A. & Mehler, M. F. Epigenetic mechanisms mouse model of temporal lobe epilepsy. J. Neurosci. temporal lobe epilepsy. Science 305, 532–535 (2004). underlying human epileptic disorders and the process 31, 2337–2347 (2011). 54. Jung, S., Warner, L. N., Pitsch, J., Becker, A. J. & of epileptogenesis. Neurobiol. Dis. 39, 53–60 (2010). Rapamycin given to mice after pilocarpine-induced Poolos, N. P. Rapid loss of dendritic HCN channel 14. Brooks-Kayal, A. R., Raol, Y. H. & Russek, S. J. status epilepticus markedly decreased the extent expression in hippocampal pyramidal neurons Alteration of epileptogenesis genes. of mossy fibre sprouting, with no effect on the following status epilepticus. J. Neurosci. 31, Neurotherapeutics 6, 312–318 (2009). development of epilepsy or on seizure frequency. 14291–14295 (2011). 15. Tang, S. J. et al. A rapamycin-sensitive signaling 35. Van Vliet, E. a et al. Inhibition of mammalian target of 55. Chen, K. et al. Persistently modified h‑channels after pathway contributes to long-term synaptic plasticity in rapamycin reduces epileptogenesis and blood–brain complex febrile seizures convert the seizure-induced the hippocampus. Proc. Nat. Acad. Sci. USA 99, barrier leakage but not microglia activation. Epilepsia enhancement of inhibition to hyperexcitability. Nature 467–472 (2002). 53, 1254–1263 (2012). Med. 7, 331–337 (2001). 16. Jaworski, J., Spangler, S., Seeburg, D. P., 36. Sliwa, A., Plucinska, G., Bednarczyk, J. & Lukasiuk, K. 56. McClelland, S. et al. Neuron-restrictive silencer factor- Hoogenraad, C. C. & Sheng, M. Control of dendritic Post-treatment with rapamycin does not prevent mediated hyperpolarization-activated cyclic nucleotide arborization by the phosphoinositide-3ʹ‑kinase–Akt– epileptogenesis in the amygdala stimulation model of gated channelopathy in experimental temporal lobe mammalian target of rapamycin pathway. J. Neurosci. temporal lobe epilepsy. Neurosci. Lett. 509, 105–109 epilepsy. Ann. Neurol. 70, 454–464 (2011). 25, 11300–11312 (2005). (2012). In the kainate model of acquired chronic TLE in 17. Raab-graham, K. F., Haddick, P. C. G., Jan, Y. N. & 37. Erlich, S., Alexandrovich, A., Shohami, E. & Pinkas- rodents, interference with the interaction between Jan, L. Y. Activity- and mTOR-dependent suppression Kramarski, R. Rapamycin is a neuroprotective REST and target genes using antisense oligodeoxy- of Kv1.1 channel mRNA translation in dendrites. treatment for traumatic brain injury. Neurobiol. Dis. nucleotides prevented status epilepticus-induced Science 314, 144–148 (2006). 26, 86–93 (2007). REST-mediated repression of HCN1 and led to a 18. Crino, P. B. Focal brain malformations: seizures, 38. Chen, S. et al. Alterations in mammalian target of decrease in spontaneous seizure frequency. signaling, sequencing. Epilepsia 50, 3–8 (2009). rapamycin signaling pathways after traumatic brain 57. Freund, T. F. & Buzsáki, G. Interneurons of the 19. Striano, P. & Zara, F. Genetics: mutations in mTOR injury. J. Cereb. Blood Flow Metab. 27, 939–949 hippocampus. Hippocampus 6, 347–470 (1996). pathway linked to megalencephaly syndromes. Nature (2007). 58. Somogyi, P. & Klausberger, T. Defined types of cortical Rev. Neurol. 8, 542–544 (2012). 39. Raffo, E., Coppola, A., Ono, T., Briggs, S. W. & interneurone structure space and spike timing in the 20. McDaniel, S. S. & Wong, M. Therapeutic role of Galanopoulou, A. S. A pulse rapamycin therapy for hippocampus. J. Physiol. 562, 9–26 (2005). mammalian target of rapamycin (mTOR) inhibition in infantile spasms and associated cognitive decline. 59. Ascoli, G. A. et al. Petilla terminology: nomenclature preventing epileptogenesis. Neurosci. Lett. 497, Neurobiol. Dis. 43, 322–329 (2011). of features of GABAergic interneurons of the 231–239 (2011). 40. Galanopoulou, A. S. et al. Identification of new cerebral cortex. Nature Rev. Neurosci. 9, 557–568 21. Wong, M. Mechanisms of epileptogenesis in tuberous epilepsy treatments: issues in preclinical methodology. (2008). sclerosis complex and related malformations of Epilepsia 53, 571–582 (2012). 60. Defelipe, J. et al. New insights into the classification cortical development with abnormal glioneuronal 41. Galanopoulou, A. S., Gorter, J. a & Cepeda, C. Finding and nomenclature of cortical GABAergic interneurons. proliferation. Epilepsia 49, 8–21 (2008). a better drug for epilepsy: the mTOR pathway as an Nature Rev. Neurosci. 202–216 (2013).

NATURE REVIEWS | NEUROSCIENCE VOLUME 14 | MAY 2013 | 347

61. Meldrum, B. Physiological changes during prolonged 82. Bartos, M., Vida, I. & Jonas, P. Synaptic mechanisms soma and AIS of parvalbumin-positive basket cells; seizures and epileptic brian damage. Neuropadiatrie of synchronized gamma oscillations in inhibitory neocortical fast-spiking cells exhibit profoundly 9, 213–212 (1978). interneuron networks. Nature Rev. Neurosci. 8, abnormal physiology.

62. Bernard, C., Cossart, R., Hirsch, J. C., Esclapez, M. 45–56 (2007). 105. Cheah, C. S. et al. Specific deletion of NaV1.1 sodium & Ben-Ari, Y. What is GABAergic inhibition? How is it 83. Cardin, J. A. et al. Driving fast-spiking cells induces channels in inhibitory interneurons causes seizures modified in epilepsy? Epilepsia 41, S90–S95 gamma rhythm and controls sensory responses. and premature death in a mouse model of Dravet (2000). Nature 459, 663–667 (2009). syndrome. Proc. Nat. Acad. Sci. USA 109, 63. Mann, E. O. & Mody, I. The multifaceted role of 84. Sohal, V. S., Zhang, F., Yizhar, O. & Deisseroth, K. 14646–14651 (2012). inhibition in epilepsy: seizure-genesis through Parvalbumin neurons and gamma rhythms enhance 106. Jonas, P., Bischofberger, J., Fricker, D. & Miles, R. excessive GABAergic inhibition in autosomal dominant cortical circuit performance. Nature 459, 698–702 Interneuron diversity series: fast in, fast out--temporal nocturnal frontal lobe epilepsy. Curr. Opin. Neurol. 21, (2009). and spatial signal processing in hippocampal 155–160 (2008). 85. Avoli, M. & De Curtis, M. GABAergic synchronization interneurons. Trend. Neurosci. 27, 30–40 (2004). 64. Galanopoulou, A. S. Mutations affecting GABAergic in the limbic system and its role in the generation of 107. Goldberg, E. M. et al. Rapid developmental signaling in seizures and epilepsy. Pflügers Arch. 460, epileptiform activity. Prog. Neurobiol. 95, 104–132 maturation of neocortical FS cell intrinsic excitability. 505–523 (2010). (2011). Cererb. Cortex 21, 666–682 (2011). 65. Sebe, J. Y. & Baraban, S. C. The promise of an 86. Krook-Magnuson, E., Armstrong, C., Oijala, M. & 108. Westenbroek, R. E., Merrick, D. K. & Catterall, W. A. interneuron-based cell therapy for epilepsy. Dev. Soltesz, I. On‑demand optogenetic control of Differential subcellular localization of the RI and RII Neurobiol. 71, 107–117 (2011). spontaneous seizures in temporal lobe epilepsy. Na+ channel subtypes in central neurons. Neuron 3, 66. Anderson, S. A. & Baraban, S. C. in Jasper’s Basic Nature Commun. 4, 1376 (2013). 695–704 (1989). Mechanisms of the Epilepsies 4th edn (eds 87. Cobos, I. et al. Mice lacking Dlx1 show subtype- 109. Gong, B., Rhodes, K. J., Bekele-Arcuri, Z. & Noebels, J. L., Avoli, M., Rogawski, M. A., Olsen, R. W. specific loss of interneurons, reduced inhibition and Trimmer, J. S. Type I and type II Na+ channel α-subunit & Delgado-Escueta, A. V.) 1122–1126 (Oxford Univ. epilepsy. Nature Neurosci. 8, 1059–1068 (2005). polypeptides exhibit distinct spatial and temporal Press, 2012). 88. Butt, S. J. B. et al. The requirement of Nkx2‑1 in the patterning, and association with auxiliary subunits in 67. Maisano, X. et al. Differentiation and functional temporal specification of cortical interneuron rat brain. J. Comp. Neurol. 412, 342–352 (1999). incorporation of embryonic stem cell-derived GABAergic subtypes. Neuron 59, 722–732 (2008). 110. Lorincz, A. & Nusser, Z. Cell-type-dependent molecular interneurons in the dentate gyrus of mice with temporal 89. Batista-Brito, R. et al. The cell-intrinsic requirement of composition of the axon initial segment. J. Neurosci. lobe epilepsy. J. Neurosci. 32, 46–61 (2012). Sox6 for cortical interneuron development. Neuron 28, 14329–14340 (2008).

68. Houser, C. R. Interneurons of the dentate gyrus: an 63, 466–481 (2009). 111. Harkin, L. A. et al. Truncation of the GABAA-receptor

overview of cell types, terminal fields and 90. Smart, S. L. et al. Deletion of the KV1.1 potassium γ2 subunit in a family with generalized epilepsy with neurochemical identity. Prog. Brain Res. 163, channel causes epilepsy in mice. Neuron 20, 809–819 febrile seizures plus. Am. J. Hum. Genet. 70, 217–232 (2007). (1998). 530–536 (2002). 69. Jinno, S. & Kosaka, T. Patterns of expression of 91. Eunson, L. H. et al. Clinical, genetic, and expression 112. Patino, G. a et al. A functional null mutation of SCN1B neuropeptides in GABAergic nonprincipal neurons in studies of mutations in the potassium channel gene in a patient with Dravet syndrome. J. Neurosci. 29, the mouse hippocampus: quantitative analysis with KCNA1 reveal new phenotypic variability. Ann. Neurol. 10764–10778 (2009). optical disector. J. Comp. Neurol. 461, 333–349 48, 647–656 (2000). 113. Ogiwara, I. et al. A homozygous mutation of voltage- + (2003). 92. Goldberg, E. M. et al. K channels at the axon initial gated sodium channel βI gene SCN1B in a patient with 70. Bakst, I., Avendano, C., Morrison, J. H. & segment dampen near-threshold excitability of Dravet syndrome. Epilepsia 53, e200–e203 (2012). Amaral, D. G. An experimental analysis of the origins neocortical fast-spiking GABAergic interneurons. 114. Kamiya, K. et al. A nonsense mutation of the sodium of somatostatin-like immunoreactivity in the dentate Neuron 58, 387–400 (2008). channel gene SCN2A in a patient with intractable gyrus of the rat. J. Neurosci. 6, 1452–1462 (1986). 93. Lau, D. et al. Impaired fast-spiking, suppressed epilepsy and mental decline. J. Neurosci. 24, 71. Katona, I., Acsády, L. & Freund, T. F. Postsynaptic cortical inhibition, and increased susceptibility to 2690–2698 (2004). targets of somatostatin-immunoreactive interneurons seizures in mice lacking Kv3.2 K+ channel proteins. 115. Shi, X. et al. Missense mutation of the sodium channel in the rat hippocampus. Neuroscience 88, 37–55 J. Neurosci. 20, 9071–9085 (2000). gene SCN2A causes Dravet syndrome. Brain Dev. 31, (1999). 94. Escayg, A. et al. Mutations of SCN1A, encoding a 758–762 (2009). 72. Acsády, L., Kamondi, A., Sík, A., Freund, T. & neuronal sodium channel, in two families with 116. Depienne, C. et al. Sporadic infantile epileptic Buzsáki, G. GABAergic cells are the major postsynaptic GEFS+2. Nature Genet. 24, 343–345 (2000). encephalopathy caused by mutations in PCDH19 targets of mossy fibers in the rat hippocampus. 95. Claes, L. et al. De novo mutations in the sodium- resembles Dravet syndrome but mainly affects J. Neurosci. 18, 3386–3403 (1998). channel gene SCN1A cause severe myoclonic epilepsy females. PLoS Genet. 5, e1000381 (2009). 73. Sloviter, R. S. Decreased hippocampal inhibition and a of infancy. Am. J. Hum. Genet. 68, 1327–1332 117. Chen, C. et al. Mice lacking sodium channel β1 selective loss of interneurons in experimental epilepsy. (2001). subunits display defects in neuronal excitability, Science 235, 73–76 (1987). 96. Harkin, L. A. et al. The spectrum of SCN1A‑related sodium channel expression, and nodal architecture. 74. Buckmaster, P. S. & Dudek, F. E. Neuron loss, granule infantile epileptic encephalopathies. Brain 130, J. Neurosci. 24, 4030–4042 (2004). cell axon reorganization, and functional changes in the 843–852 (2007). 118. McClelland, S., Dubé, C. M., Yang, J. & Baram, T. Z. dentate gyrus of epileptic kainate-treated rats. 97. Ohmori, I., Kahlig, K. M., Rhodes, T. H., Wang, D. W. & Epileptogenesis after prolonged febrile seizures: J. Comp. Neurol. 385, 385–404 (1997). George, A. L. Nonfunctional SCN1A is common in mechanisms, biomarkers and therapeutic 75. De Lanerolle, N. C., Kim, J. H., Robbins, R. J. & severe myoclonic epilepsy of infancy. Epilepsia 47, opportunities. Neurosci. Lett. 497, 155–162 (2011). Spencer, D. D. Hippocampal interneuron loss and 1636–1642 (2006). 119. Shinnar, S. et al. MRI abnormalities following febrile plasticity in human temporal lobe epilepsy. Brain Res. 98. Yu, F. H. et al. Reduced sodium current in GABAergic status epilepticus in children: the FEBSTAT study. 495, 387–395 (1989). interneurons in a mouse model of severe myoclonic Neurology 79, 871–877 (2012). 76. Zhang, W. et al. Surviving hilar somatostatin epilepsy in infancy. Nature Neurosci. 9, 1142–1149 120. Abou-Khalil, B. et al. Partial and generalized epilepsy interneurons enlarge, sprout axons, and form new (2006). with febrile seizures plus and a novel SCN1A mutation. synapses with granule cells in a mouse model of 99. Ragsdale, D. S. How do mutant Nav1.1 sodium Neurology 57, 2265–2272 (2001).

temporal lobe epilepsy. J. Neurosci. 29, channels cause epilepsy? Brain Res. Rev. 58, 121. Sugawara, T. et al. Nav1.1 mutations cause febrile 14247–14256 (2009). 149–159 (2008). seizures associated with afebrile partial seizures. In a mouse model of chronic TLE, surviving somato- 100. Oakley, J. C., Kalume, F., Yu, F. H., Scheuer, T. & Neurology 57, 703–705 (2001). statin-positive hilar interneurons exhibit axonal Catterall, W. A. Temperature- and age-dependent 122. Scheffer, I. E. et al. Temporal lobe epilepsy and GEFS+ sprouting and form functional inhibitory seizures in a mouse model of severe myoclonic phenotypes associated with SCN1B mutations. Brain connections with dentate gyrus granule cells at epilepsy in infancy. Proc. Nat. Acad. Sci. USA 106, 130, 100–109 (2007). increased frequency as demonstrated anatomically 3994–3999 (2009). An interesting case series including a patient with and physiologically. 101. Mullen, S. A. & Scheffer, I. E. Translational research in GEFS+ due to the SCN1B mutation but no history 77. Mathern, G. W., Babb, T. L., Pretorius, J. K. & epilepsy genetics: sodium channels in man to of febrile or afebrile status epilepticus who Leite, J. P. Reactive synaptogenesis and neuron interneuronopathy in mouse. Arch. Neurol. 66, 21–26 developed TLE with classic mesial temporal densities for neuropeptide Y, somatostatin, and (2009). sclerosis and become seizure-free after unilateral glutamate decarboxylase immunoreactivity in the 102. Martin, M. S. et al. Altered function of the SCN1A temporal lobectomy. epileptogenic human fascia dentata. J. Neurosci. 15, voltage-gated sodium channel leads to γ-aminobutyric 123. Kearney, J. A. et al. A gain‑of‑function mutation in the 3990–4004 (1995). acid-ergic (GABAergic) interneuron abnormalities. sodium channel gene Scn2a results in seizures and 78. Buckmaster, P. S. & Wen, X. Rapamycin suppresses J. Biol. Chem. 285, 9823–9834 (2010). behavioral abnormalities. Neuroscience 102, axon sprouting by somatostatin interneurons in a 103. Escayg, A. & Goldin, A. L. Sodium channel SCN1A and 307–317 (2001). mouse model of temporal lobe epilepsy. Epilepsia 52, epilepsy: mutations and mechanisms. Epilepsia 51, 124. Striano, P. et al. Brain MRI findings in severe 2057–2064 (2011). 1650–1658 (2010). myoclonic epilepsy in infancy and genotype–phenotype

79. Gonchar, Y., Wang, Q. & Burkhalter, A. Multiple 104. Ogiwara, I. et al. Nav1.1 localizes to axons of correlations. Epilepsia 48, 1092–1096 (2007). distinct subtypes of GABAergic neurons in mouse parvalbumin-positive inhibitory interneurons: a 125. Siegler, Z. et al. Hippocampal sclerosis in severe visual cortex identified by triple immunostaining. circuit basis for epileptic seizures in mice carrying an myoclonic epilepsy in infancy: a retrospective MRI Front. Neuroanat. 1, 3 (2007). Scn1a gene mutation. J. Neurosci. 27, 5903–5914 study. Epilepsia 46, 704–708 (2005). 80. Xu, X., Roby, K. D. & Callaway, E. M. Immunochemical (2007). 126. Guerrini, R., Striano, P., Catarino, C. & Sisodiya, S. M. characterization of inhibitory mouse cortical neurons: Knock‑in mice harbouring an Scn1a nonsense Neuroimaging and neuropathology of Dravet three chemically distinct classes of inhibitory cells. mutation that is also found in patients with Dravet syndrome. Epilepsia 52, 30–34 (2011). J. Comp. Neurol. 518, 389–404 (2010). syndrome develop temperature-sensitive and 127. Nangia, S., Caraballo, R. H., Kang, H.‑C., Nordli, D. R. 81. Pouille, F. & Scanziani, M. Enforcement of temporal spontaneous seizures resembling the human & Scheffer, I. E. Is the ketogenic diet effective in fidelity in pyramidal cells by somatic feed-forward condition. Immunohistochemistry indicates specific epilepsy syndromes? Epilepsy Res. 100,

inhibition. Science 293, 1159–1163 (2001). restricted expression of NaV1.1 protein at the 252–257 (2012).

348 | MAY 2013 | VOLUME 14 www.nature.com/reviews/neuro

128. Brackenbury, W. J., Yuan, Y., Malley, H. A. O., granule cell dendrites relative to other principal cells 162. Su, H. et al. Upregulation of a T‑type Ca2+ channel Parent, J. M. & Isom, L. L. Abnormal neuronal in the cerebral cortex may contribute to dentate causes a long-lasting modification of neuronal firing patterning occurs during early postnatal brain gating. mode after status epilepticus. J. Neurosci. 22, development of Scn1b‑null mice and precedes 146. Ewell, L. A. & Jones, M. V. Frequency-tuned 3645–3655 (2002). hyperexcitability. Proc. Nat. Acad. Sci. USA 110, distribution of inhibition in the dentate gyrus. A thorough review of ion channel alterations in 1089–1094 (2013). J. Neurosci. 30, 12597–12607 (2010). epilepsy and other neurological disorders. 129. Tye, K. M. & Deisseroth, K. Optogenetic This paper shows that the dentate gate is 163. Yaari, Y., Yue, C. & Su, H. Recruitment of apical investigation of neural circuits underlying brain subserved at least in part by feedforward inhibition dendritic T‑type Ca2+ channels by backpropagating disease in animal models. Nature Rev. Neurosci. 13, by fast-spiking interneurons onto dentate gyrus spikes underlies de novo intrinsic bursting in 251–266 (2012). granule cells. However, granule cell activity in hippocampal epileptogenesis. J. Physiol. 580, 130. Bernstein, J. G. & Boyden, E. S. Optogenetic tools for response to stimulation of the perforant path is 435–450 (2007). analyzing the neural circuits of behavior. Trend. Cog. frequency-dependent. 164. Bear, J. & Lothman, E. W. An in vitro study of focal Sci. 15, 592–600 (2011). 147. Leranth, C., Malcolm, A. J. & Frotscher, M. Afferent epileptogenesis in combined hippocampal- 131. Paz, J. T. et al. Closed-loop optogenetic control of and efferent synaptic connections of somatostatin- parahippocampal slices. Epilepsy Res. 14, 183–193 thalamus as a tool for interrupting seizures after immunoreactive neurons in the rat fascia dentata. (1993). cortical injury. Nature Neurosci. 16, 64–70 (2013). J. Comp. Neurol. 295, 111–122 (1990). 165. Wu, K. & Leung, L. S. Increased dendritic excitability in 132. Heinemann, U. et al. The dentate gyrus as a regulated 148. Pathak, H. R. et al. Disrupted dentate granule cell hippocampal ca1 in vivo in the kainic acid model of gate for the propagation of epileptiform activity. chloride regulation enhances synaptic excitability temporal lobe epilepsy: a study using current source Epilepsy Res. Suppl. 7, 273–280 (1992). during development of temporal lobe epilepsy. density analysis. Neuroscience 116, 599–616 (2003). 133. Lothman, E. W., Stringer, J. L. & Bertram, E. H. The J. Neurosci. 27, 14012–14022 (2007). 166. Gibbs, J. W., Shumate, M. D. & Coulter, D. A. dentate gyrus as a control point for seizures in the 149. Ben-Ari, Y. Excitatory actions of gaba during Differential epilepsy-associated alterations in

hippocampus and beyond. Epilepsy Res. Suppl. 7, development: the nature of the nurture. Nature Rev. postsynaptic GABAA receptor function in dentate 301–313 (1992). Neurosci. 3, 728–739 (2002). granule and CA1 neurons. J. Neurophysiol. 77, 134. Dudek, F. E. & Sutula, T. P. Epileptogenesis in the 150. Owens, D. F. & Kriegstein, A. R. Is there more to GABA 1924–1938 (1997). dentate gyrus: a critical perspective. Prog. Brain Res. than synaptic inhibition? Nature Rev. Neurosci. 3, 167. Cohen, A. S., Lin, D. D., Quirk, G. L. & Coulter, D. A.

163, 755–773 (2007). 715–727 (2002). Dentate granule cell GABAA receptors in epileptic 135. Buckmaster, P. S. in Jasper’s Basic Mechanisms of the 151. Yeo, M., Berglund, K., Augustine, G. & Liedtke, W. hippocampus: enhanced synaptic efficacy and altered Epilepsies 4th edn (eds Noebels, J. L., Avoli, M., Novel repression of Kcc2 transcription by REST–RE‑1 pharmacology. Eur. J. Neurosci. 17, 1607–1616 (2003). Rogawski, M. A., Olsen, R. W., & Delgado- controls developmental switch in neuronal chloride. 168. Kelley, M. S., Jacobs, M. P. & Lowenstein, D. H. The Escueta, A. V.) 416–431 (Oxford Univ. Press, 2012). J. Neurosci. 29, 14652–14662 (2009). NINDS epilepsy research benchmarks. Epilepsia 50, 136. Zhang, W., Huguenard, J. R. & Buckmaster, P. S. 152. Rivera, C. et al. Mechanism of activity-dependent 579–582 (2009). Increased excitatory synaptic input to granule cells downregulation of the neuron-specific K‑Cl 169. Temkin, N. R. Preventing and treating posttraumatic from hilar and CA3 regions in a rat model of cotransporter KCC2. J. Neurosci. 24, 4683–4691 seizures: the human experience. Epilepsia 50, 10–13 temporal lobe epilepsy. J. Neurosci. 32, 1183–1196 (2004). (2009). (2012). 153. Brandt, C., Nozadze, M., Heuchert, N., Rattka, M. & 170. Vadlamudi, L., Scheffer, I. E. & Berkovic, S. F. Genetics 137. Coulter, D. A. & Carlson, G. C. Functional regulation of Löscher, W. Disease-modifying effects of phenobarbital of temporal lobe epilepsy. J. Neurol. Neurosurg. the dentate gyrus by GABA-mediated inhibition. Prog. and the NKCC1 inhibitor bumetanide in the Psych. 74, 1359–1361 (2003). Brain Res. 163, 235–243 (2007). pilocarpine model of temporal lobe epilepsy. 171. Ribeiro, P. et al. Expression profile of Lgi1 gene in 138. Leutgeb, J. K., Leutgeb, S., Moser, M.‑B. & Moser, E. I. J. Neurosci. 30, 8602–8612 (2010). mouse brain during development. J. Mol. Neurosci. Pattern separation in the dentate gyrus and CA3 of 154. Ang, C. W., Carlson, G. C. & Coulter, D. A. 35, 323–329 (2008). the hippocampus. Science 315, 961–966 (2007). Hippocampal CA1 circuitry dynamically gates direct 172. Michelucci, R., Pasini, E. & Nobile, C. Lateral temporal 139. Clelland, C. D. et al. A functional role for adult cortical inputs preferentially at theta frequencies. lobe epilepsies: clinical and genetic features. Epilepsia hippocampal neurogenesis in spatial pattern J. Neurosci. 25, 9567–9580 (2005). 50, 52–54 (2009). separation. Science 325, 210–213 (2009). 155. Soltesz, I. Brief history of cortico-hippocampal time 173. Amaral, D. G., Scharfman, H. E. & Lavenex, P. The 140. McHugh, T. J. et al. Dentate gyrus NMDA receptors with a special reference to the direct entorhinal input dentate gyrus: fundamental neuroanatomical mediate rapid pattern separation in the hippocampal to CA1. Hippocampus 5, 120–124 (1995). organization (dentate gyrus for dummies). Prog. Brain network. Science 317, 94–99 (2007). 156. Tsay, D., Dudman, J. T. & Siegelbaum, S. A. HCN1 Res. 163, 3–22 (2007). 141. Yu, E. P. et al. Protracted postnatal development of channels constrain synaptically evoked Ca2+ spikes in 174. Morgan, R. J., Santhakumar, V. & Soltesz, I. Modeling sparse, specific dentate granule cell activation in the distal dendrites of CA1 pyramidal neurons. Neuron. the dentate gyrus. Prog. Brain Res. 163, 639–658 mouse hippocampus. J. Neurosci. 33, 2947–2960 56, 1076–1089 (2007). (2007).

(2013). 157. Magee, J. C. Dendritic lh normalizes temporal 142. Chawla, M. K. et al. Sparse, environmentally selective summation in hippocampal CA1 neurons. Nature Acknowledgements expression of Arc RNA in the upper blade of the Neurosci. 2, 508–514 (1999). We thank L. Isom, E. Marsh and three anonymous referees rodent fascia dentata by brief spatial experience. 158. Coulter, D. A. et al. Hippocampal microcircuit for assistance in the preparation of this manuscript. Hippocampus 15, 579–586 (2005). dynamics probed using optical imaging approaches. 143. Behr, J., Lyson, K. J. & Mody, I. Enhanced propagation J. Physiol. 589, 1893–1903 (2011). of epileptiform activity through the kindled dentate 159. Beck, H. & Yaari, Y. Plasticity of intrinsic neuronal Competing interests statement gyrus. J. Neurophysiol. 79, 1726–1732 (1998). properties in CNS disorders. Nature Rev. Neurosci. 9, The authors declare no competing financial interests. 144. Ang, C. W., Carlson, G. C. & Coulter, D. A. Massive and 357–369 (2008). specific dysregulation of direct cortical input to the 160. Cossart, R. et al. Dendritic but not somatic GABAergic FURTHER INFORMATION hippocampus in temporal lobe epilepsy. J. Neurosci. inhibition is decreased in experimental epilepsy. Douglas A. Coulter’s homepage: http://stokes.chop.edu/ 26, 11850–11856 (2006). Nature Neurosci. 4, 52–62 (2001). web/coulter/ 145. Krueppel, R., Remy, S. & Beck, H. Dendritic 161. Sanabria, E. R., Su, H. & Yaari, Y. Initiation of network The Children’s Hospital of Philadelphia Division of Neurology: integration in hippocampal dentate granule cells. bursts by Ca2+-dependent intrinsic bursting in the rat http://www.chop.edu/service/neurology/home.html Neuron 71, 512–528 (2011). pilocarpine model of temporal lobe epilepsy. ALL LINKS ARE ACTIVE IN THE ONLINE PDF Unique synaptic integration by dentate gyrus J. Physiol. 532, 205–216 (2001).

NATURE REVIEWS | NEUROSCIENCE VOLUME 14 | MAY 2013 | 349