Etiology and pathogenesis
In approximately 50% of families, a molecular diagnosis can be determined through identification of mutations in the LGI1 as well as RELN genes being most often described. Other less frequently implicated genes have also been described but they account for a minority of families in whom a mutation has been identified.
The LGI1 gene is mutated in approximately 30% of families with familial lateral temporal lobe epilepsy (10; 74). A large Italian study identified LGI1 mutations in only 30% of kindreds (67). Although LGI1 mutations appear to be specific for this type of temporal lobe epilepsy, the identification of LGI1 mutations in only one half of families presenting the typical phenotype (10; 74) suggests genetic heterogeneity in the families. A linkage to chromosome 19q13.11-q13.31 has been found in a large Brazilian kindred presenting with auditory auras (15).
A comprehensive review on LGI1 mutations in lateral temporal lobe epilepsy can be found in a paper by Nobile and colleagues (71) and in a review on pathogenic mechanisms in LGI1 mutations (99). The LGI1 gene was cloned from a glioblastoma cell line, and although previous studies have suggested that LGI1 represents a tumor suppressor gene (26), a more recent study was unable to establish a correlation between the gene and malignant glioma suppression (45). LGI1 (also known as Epitempin) is a secreted neuronal protein characterized by a central leucine-rich repeat region (49), which is involved in regulation of cell growth, adhesion, and migration.
The role of LGI1 in the epileptogenic process began to be clarified in the past decade. Fukata and colleagues demonstrated that LGI1 binds to the postsynaptic membrane protein ADAM22, thus, linking LGI1 with modulation of glutamate-AMPA synaptic transmission (43). In addition, defective LGI1 could lead to presynaptic changes in inactivation kinetics of A-type potassium channels through an effect on voltage-gated potassium channel subunit (Kv1.1), thus, increasing excitability (83).
Two protein isoforms of LGI1 are differentially expressed in the human brain, with a much higher concentration in the temporal neocortex than in the hippocampus, which could explain the characteristic auditory auras in patients with LGI1 mutation (44). The association of lateral temporal malformation patterns in some families may be indicative of a probable role of LGI1 in the development of the temporal lobes (56; 90).
At least 43 LGI1 mutations have been described today. LGI1 mutations in familial lateral temporal lobe epilepsy patients are associated with loss-of-function, with the LGI1 protein either not secreted, misfolded, or unstable (77; 85; 59). Exceptions are the mutations described by Striano and colleagues and Di Bonaventura and colleagues, as they do not affect LGI1 protein secretion (34; 87). Striano and colleagues also reported that the mutation found in their family does not induce large structural rearrangements although could destabilize its interactions with target proteins. Investigating LGI1 mutations described in familial lateral temporal lobe epilepsy as well as epilepsy due to LGI1 autoantibodies, Dazzo and colleagued have validated and introduced the concept that LGI1 mutations can be secretion-defective or secretion-competent, the latter resulting in impaired interaction between secreted LGI1 with neuronal receptors ADAM22 and ADAM23 (29). For a review of the interactions between LGI1 and ADAM receptors please refer to Yamagata and Fukai (99).
Many different mutations have been described in different kindreds worldwide, all located in the coding region or exon splice sites (46; 51; 70; 41; 56; 68; 77; 10; 18; 48; 74; 76; 25; 71; 53; 34; 87; 31; 52; 61). Bovo and colleagues analyzed the promoter region of the LGI1 gene in sporadic and familial lateral temporal lobe epilepsy patients and found no mutations in the promoter sequence (19).
LGI1 microdeletions have been identified in families that tested negative for exon sequencing, suggesting that copy number variation analysis could help further understand mutation negative patients or families (40; 39). This has not been confirmed in a series of 8 families, and 20 sporadic patients tested negative for known LGI1 mutations (63).
LGI1 mutations have been identified in about 2% of nonfamilial cases (71). No mutations in LGI2, LGI3, or LGI4 have been identified in 71 families with various types of temporal lobe epilepsy, including 4 with familial lateral temporal lobe epilepsy (10; 07). Finally, no mutations related to the ADAM22 gene (23; 32) or Kv1 channel genes (32) have been identified in kindreds with familial lateral temporal lobe epilepsy without LGI1 mutations.
Anderson has proposed, based on evidence from in vitro studies in transgenic mice (100), that LGI1 modulates a mechanism that could lead to epileptogenesis via persistent immaturity of glutamatergic circuitries (06). Interestingly, LGI1 has been demonstrated as the autoantigen for antibodies found in patients with limbic encephalitis, which had been previously attributed to voltage-gated potassium channels (58). These 2 observations point to a role of LGI1 in the maturation of glutamatergic neurotransmission and also in the etiology of acquired epilepsies.
Further insights about the mechanisms underlying epileptogenicity derived from LGI1 mutations have been unraveled in vitro using hippocampal CA3 neurons (84). LGI1 deletion resulted in downregulation of axonal Kv1.1 and Kv1.2 channels, thus, impairing the capacity of axonal D-type current to limit glutamate release. This could be a potential mechanism through which LGI1 regulates intrinsic excitability of neurons and promotes development of epileptogenicity.
Heterozygous mutations in the gene encoding for the secreted protein reelin (RELN) on chromosome 7q have been found in 17% of kindreds that tested negative for LGI1 mutations, providing an interesting new venue for understanding the pathophysiology of this epilepsy syndrome (28; 69). Homozygous RELN mutations cause lissencephaly with cerebellar hypoplasia, and as the authors found an inhibitory effect of mutations on protein secretion and co-localization of LGI1 and RELN in many rat brain regions, they bring to discussion the regulatory role these 2 proteins exert in brain development.
Pippucci and colleagues used whole exome sequencing to investigate probands who tested negative for known LGI1 mutations and found CNTNAP2 intragenic deletion, 2 truncating mutations of DEPDC5 gene, and a missense SCN1A change (75). Interestingly, CNTNAP2 encodes CASPR2 pertaining to the voltage-gated potassium channel complex with which LGI1 interacts, whereas DEPDC5 has been originally identified in familial focal epilepsy with variable foci and now in about 10% of focal epilepsies, including patients with an underlying malformation of cortical development. No DEPDC5 mutations were found in another cohort of familial lateral temporal lobe epilepsy patients who tested negative for LGI1 mutations (88).
More recently, Leonardi and colleagues screened 28 familial lateral temporal lobe epilepsy kindreds that had tested negative for LGI1 and RELN mutations and found no CNTNAP2 (contactin associated protein like 2) mutations with next-generation sequencing and copy number variation analyses (60).
Using whole exome sequencing combined with genome-wide single-nucleotide polymorphism-array linkage analysis, Dazzo and colleagues described 2 variants in the MICAL-1 (microtubule-associated monooxy-genase, calponin, and lim domain containing 1) gene in 2 different families with familial lateral temporal lobe epilepsy (30). Dysregulation of the actin cytoskeleton dynamics was proposed by the authors as a possible mechanism for these pathogenic variants.
Finally, 2 possibly pathogenic missense variants in the SCN1A gene were identified in 3.8% of sporadic patients in a cohort of lateral temporal lobe epilepsy with no antecedent of febrile seizures (13). Patients within GEFS+ families could present with typical semiology with auditory features and carry SCN1A mutations.
In the most recent and large genetic study in lateral temporal lobe epilepsy including 112 unrelated cases (of whom 33 were familial patients), Bisulli and colleagues applied next-generation sequencing and confirmed an underlying genetic abnormality in 8% of individuals (16). Pathogenic or likely pathogenic variants were identified in LGI1 (2.7%), RELN (1.8%), SCN1A (2.7%), and DEPDC5 (0.9%).
