An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Noebels JL, Avoli M, Rogawski MA, et al., editors. Jasper's Basic Mechanisms of the Epilepsies. 5th edition. New York: Oxford University Press; 2024. doi: 10.1093/med/9780197549469.003.0043
This chapter focuses on focal cortical dysplasia type II (FCDII) and mild malformation of cortical development with oligodendroglial hyperplasia in epilepsy (MOGHE), a mild malformation of cortical development, for which genetic advances have been remarkable in recent years. There is now clear evidence that FCDII is caused by brain somatic mutations in genes belonging to the mTOR pathway, and that MOGHE, a mild malformation of cortical development, is due to somatic mutations in the SLC35A2 gene, encoding the major Golgi-localized UDP-galactose transporter. Numerous rodent models using the in utero electroporation procedure have been used to model FCDII and recapitulate most neuropathological and clinical features. This chapter discusses the recent advances in using cerebral organoids to model neurodevelopmental disorders.
Malformations of cortical development (MCDs) are rare neurodevelopmental disorders that result from abnormal development of the cerebral cortex in utero (Oegema et al., 2020; Klingler et al., 2021). They comprise a heterogeneous spectrum of clinical, imaging, and molecular and histopathological entities. MCDs typically cause drug-resistant childhood epilepsy and developmental delay within the first years of life. Many MCDs are caused by an underlying genetic defect; typically by germline mutations affecting all cells of the body but also by somatic postzygotic mutations that affect only a subset of brain cells.
In this chapter, we focus on focal cortical dysplasia type II (FCDII) and MOGHE, a mild malformation of cortical development, for which genetic advances have been remarkable in recent years, revealing brain mosaicism as a major underlying cause.
Focal cortical dysplasia (FCD) encompasses a spectrum of cortical malformations defined by a consensus classification scheme by the International League against Epilepsy (ILAE) in 2011 (Blumcke et al., 2011). FCDs are among the most common MCDs and represent an important cause of refractory pediatric epilepsies that usually occur sporadically, with only rare familial cases described (Leventer et al., 2014). FCDs are typically detected using magnetic resonance imaging (MRI) following emergence of a seizure, but small lesions may only be identified after neuropathological investigations following epilepsy surgery. Despite the existence of a large diversity of antiseizure medications, patients with FCD are particularly prone to drug-resistant seizures. In most cases, treatment requires surgical resection of the epileptogenic zone defined as the area responsible for seizure generation (Blumcke et al., 2017). This provides access to pathologically relevant brain tissue that represents a unique opportunity for translational research. The size of the lesion varies from submicroscopic involvement of one sulcus to the involvement of an entire cerebral hemisphere (hemimegalencephaly). At the cellular level, FCD is the result of abnormal migration and differentiation of neurons during embryogenesis. The 2011 ILAE classification has identified three subtypes of FCD based on the presence or absence of features in addition to cortical dyslamination: type I (no abnormal cell types), type II (dysmorphic neurons with or without balloon cells), and type III (type I associated with another lesion such as hippocampal sclerosis or glioneuronal tumors) (Blumcke et al., 2011; Najm et al., 2018). The ILAE classification has recently be updated to include genetics finding (Najm et al., 2022). Recent genetic advances have shown clear evidence that FCDII is caused by brain somatic mutations in genes belonging to the mTOR signaling pathway (Fig. 43–1).
Hallmarks, concept, and genetics in FCDII. A. Left: MRI of a patient with FCDIIb (white arrow indicating the lesion); middle: hematoxylin-eosin staining showing a dysmorphic neuron (black arrow) and a balloon cell (blue arrow); right: pS6 and hematoxylin (more...)
mTOR (mechanistic target of rapamycin) is a serine/threonine protein kinase that is the core member and catalytic subunit of two complexes: mTORC1 (mTOR complex 1) and mTORC2 (mTOR complex 2). The mTOR pathway integrates various extracellular and intracellular stimuli, such as nutrients, hormones, and energy, through the phosphorylation of a myriad of downstream substrates to coordinate essential biological processes such as cell growth and proliferation, protein and lipid synthesis, mitochondrial and lysosome biogenesis, autophagy, and more broadly translation initiation. Therefore, mTOR acts as a mediator of vital anabolic and catabolic metabolism upon environmental cues. (For an extensive review of downstream processes, see Saxton and Sabatini, 2017.) There are two distinct upstream regulatory branches of the mTORC1 pathway: the PI3K/AKT/TSC branch and the GATOR1/GATOR2 branch (respectively thereafter referred to as “TSC pathway” and “GATOR pathway”). The TSC pathway involves the activation of tyrosine kinase receptors by growth factors such as insulin, as well as cytokines and hormones. Subsequently, the downstream signaling cascade is triggered starting from the phosphoinositide-3′/protein B kinases (PI3K-AKT) and extracellular signal-regulated kinases (ERKs), which in turn inactivate the hamartin/tuberin (TSC1/2) heterodimer. TSC2 acts as a GTPase-activating protein (GAP) toward Ras homolog enriched in brain (Rheb) protein, a direct activator mTORC1. Thus, the presence of extracellular stimuli leads to PI3K/AKT/TSC and ERK activation, which inhibits the TSC complex and leads to mTORC1 activation. In the absence of extracellular factors, the TSC complex inactivates Rheb, causing mTORC1 inhibition. The GATOR1 pathway is an amino-acid-sensing regulatory branch of the mTORC1 cascade (Bar-Peled et al., 2013). Sequentially, Sestrin2 and CASTOR1, which are sensors of leucine and arginine, respectively, both regulate the GAP activity toward Rags complex 2 (GATOR2), which, in turn, controls the GAP activity toward Rags complex 1 (GATOR1). GATOR2 is composed of five partners, namely, WDR24, WDR59, Mios, Sec13, and Seh1L. GATOR1 is composed of three partners, namely, DEPDC5, NPRL2, and NPRL3. DEPDC5 acts as a GTPase-activating protein toward the lysosomal RagA/B protein complex, responsible for mTORC1 activity state. GATOR1 therefore is the ultimate regulating unit of mTORC1 activity in this branch. Thus, in the presence of amino acids, Sestrin2 and CASTOR1 are inactive, GATOR2 inhibits GATOR1, and mTORC1 is active. Upon amino-acid deprivation, Sestrin2 and CASTOR1 inhibit GATOR2, and GATOR1 prevents mTORC1 activation via the inhibition of the Rag activity.
Focal cortical dysplasia type II (FCDII) is characterized by cortical disorganization and the presence of abnormal cells: dysmorphic neurons (present in FCD type IIa and IIb) and balloon cells (only present in FCD type IIb) (Blumcke et al., 2011; Najm et al., 2018). These abnormal cells are intermixed with morphologically normal surrounding cells. Recent molecular discoveries have revealed that the focal and mosaic nature of FCDII is explained by the occurrence during brain development of somatic mutations. Originally, de novo germline and postzygotic somatic mutations in genes belonging to the PI3K/AKT3/mTOR pathway were found in a spectrum of related megalencephaly syndromes (Poduri et al., 2012; Lee et al., 2012; Riviere et al., 2012). Multiple subsequent studies then identified somatic mosaicism associated with smaller focal cortical malformations such as FCDII, in which genetic variants are present only in a subset of brain cells identified in resected tissues (D’Gama et al., 2015; Jansen et al., 2015; Nakashima et al., 2015; Lim et al., 2015; Moller et al., 2016; Mirzaa et al., 2016; D’Gama et al., 2017; Jamuar et al., 2014; Lim et al., 2017; Baldassari et al., 2019b; Gerasimenko et al., 2023). It is now well-recognized that FCDII is caused by brain-restricted variants in the mTOR pathway, establishing FCDII as part of the spectrum of “mTORopathies,” among other disorders such as tuberous sclerosis (TSC) (Iffland and Crino, 2017).
In the case of neurodevelopmental disorders, a pathogenic somatic mutation arises postzygotically in neural progenitors at different stage of development, thus producing clones of mutant neural cells of various sizes. Therefore, the size of the dysplastic lesion is likely correlated to the timing of occurrence of the postzygotic somatic variant. The variant allele frequency (VAF), defining the percentage of alleles carrying a given variant in deep bulk DNA sequencing, is used as a proxy to infer the proportion of mutant cells. VAFs can be as low as 1% (which corresponds to 2% of diploid mutated cells) in small FCDII such as bottom-of-the-sulcus dysplasia, and up to ~10% in extended FCDs; in hemimegalencephaly, VAF can be as high as 30%. Since somatic mutations associated with FCD are absent from the blood and are only present at low mosaic rate in the brain tissue (usually below 5%), their detection requires dedicated deep sequencing and bioinformatics tools (Marsan and Baulac, 2018). The diagnostic rate in FCDII varies across studies, from 38% (61/160) (Sim et al., 2019) to 41% (27/66) (D’Gama et al., 2017) and 63% (39/62 patients with FCDII or hemimegalencephaly; Baldassari et al., 2019b) (reviewed in Gerasimenko et al., 2023).
Single brain somatic mutations in the MTOR gene itself or in genes encoding upstream activators of the pathway (AKT3, PIK3CA, and RHEB) are an important cause of FCDII (Jansen et al., 2015; Nakashima et al., 2015; Lim et al., 2015, 2017; Moller et al., 2016; Baldassari et al., 2019b; Sim et al., 2019; Gerasimenko et al., 2023). These mutations are typically recurrent missense variants known in cancer and listed in the “Catalogue of Somatic Mutations in Cancer” (COSMIC). They act as gain of function and are sufficient to cause constitutive activation of the mTOR pathway. Analysis of pools of laser-captured micro-dissected FCDII dysmorphic neurons and balloon cells revealed both cell types harbored the mTOR-activating variant (Baldassari et al., 2019b; Lee et al., 2019, 2021). There is therefore a direct causal relationship between somatic mutations, mTOR signaling activation, and the generation of cytomegalic cells. In hemimegalencephaly, mutations were also detected in pools of micro-dissected glial cells (Baldassari et al., 2019b) or sorted NeuN-negative cells (D’Gama et al., 2017), pointing that these mutations arose in early neuroglial progenitors.
A second mechanism, although less frequent, accounting for FCDII is the occurrence of a “double-hit,” germline and somatic mutations, as previously shown in the TSC1/2 genes causing tuberous sclerosis complex (TSC) (Martin et al., 2017). Heterozygous germline mutations in mTOR repressor GATOR1 complex genes (DEPDC5, NPRL2, and NPRL3) are frequently reported in patients with focal epilepsy phenotypes, ranging from nonlesional (often familial) (Ishida et al., 2013; Dibbens et al., 2013) focal epilepsies to cases associated with an FCDII (Baulac et al., 2015; Ricos et al., 2016; Baldassari et al., 2019a). Somatic second-hit events of the wild-type allele leading to biallelic gene inactivation of DEPDC5 have been reported in a subset of FCDII cases, establishing the two-hit model in FCDII, according to Knudson’s two-hit model of tumorigenesis (Baulac et al., 2015; Mirzaa et al., 2016; Ribierre et al., 2018; Baldassari et al., 2019b; Sim et al., 2019; Lee et al., 2019). The occurrence of a brain somatic “second-hit” mutation on top of a heterozygous germline variant leads to constitutive mTOR activation in a subset of cells (i.e., dysmorphic neurons). DEPDC5 mutations are predominantly associated with FCDIIa (without balloon cells) subtypes.
Two studies recently reported the proof of principle that brain variants causing MCDs can be detected in the circulating cell-free DNA from cerebrospinal fluid liquid (CSF) (Kim et al., 2021; Ye et al., 2021). This finding offers future opportunities to achieve a genetic diagnosis before neurosurgery, or in patients not eligible for surgery, or when brain tissue is not available.
Mild malformations of cortical development (mMCD) are a subtype of epileptogenic focal MCDs that lacks clear histopathological evidence for cytoarchitectural abnormalities. Mosaic variants in the SLC35A2 gene have emerged as an important genetic etiology in a spectrum of focal epilepsies ranging from nonlesional focal epilepsies, mild malformation of cortical development (mMCD), and mild malformation of cortical development with oligodendroglial hyperplasia in epilepsy (MOGHE) (Sim et al., 2018; Winawer et al., 2018; Bonduelle et al., 2021; Miller et al., 2020). SLC35A2 somatic mutations are notably frequently encountered (45% of cases) with the MOGHE phenotype (Barba et al., 2023; Bonduelle et al., 2021). MOGHE is a clinico-pathological entity associated with pediatric drug-resistant focal epilepsy, predominantly manifesting in the frontal lobe and subjected to epilepsy surgery (Hartlieb et al., 2019; Schurr et al., 2017). Histological features of MOGHE consist of increase of Olig2-, and PDGFR-alpha-immunoreactive oligodendroglia in white matter and deep cortical layers, and heterotopic neurons in white matter. The finding of a non-mTOR-related gene in mMCD is consistent with the neuropathology features consisting of focal neuronal migration defects or focal cortical dyslamination in the absence of dysmorphic neurons and balloon cells. The SLC35A2 gene encodes for a protein that transports uridine-diphosphate-galactose (UDP) from the cytosol to the Golgi, where it functions in asparagine (N-linked) glycosylation, an essential posttranslational modification of proteins and sphingolipids. SLC35A2 mutations act as loss of function, also called inactivating mutations, and result in aberrant N-glycan patterns in mutation-carrying brain tissues (Sim et al., 2018).
The transition from identifying novel disease-causing mutations to working out appropriate treatment requires the development of relevant cellular and animal models. The path toward understanding the molecular basis of pathogenesis most frequently initiates with the development of genetically modified animal models. Numerous constitutive and conditional knockout and knock-in mice and rat strains have been generated for Tsc1/Tsc2 (an exhaustive list of animal models of TSC genes and GATOR1 complex genes is reviewed in Switon et al., 2017; Nguyen and Bordey, 2021; Lasarge and Danzer, 2014; Marsan and Baulac, 2018). Altogether, these studies have demonstrated the key role of each individual gene in early embryonic development and have been instrumental in demonstrating in vivo the importance of TSC and GATOR1 genes in regulating mTORC1 activity and husking physiological pathways mediating brain malformation and epileptogenesis. However, the lack of robust neuropathological and behavioral phenotypes (i.e., spontaneous seizures) expected from modeling refractory neurodevelopmental epilepsy prompted the scientific community to establish animal models accurately mimicking the genetic etiology of the disorder. Since the early evidence of disease-causing somatic mTOR-pathway gene mutations in cortical malformations and focal epilepsy, several studies focused on overexpressing or downregulating these target genes in a focal area of the developing brain cortex via in utero electroporation to reproduce somatic brain mosaicism.
In utero electroporation (IUE) is an efficient approach to modulate gene expression in the developing rodent brain. It consists in a swift two-step procedure. First, negatively charged molecules such as nucleic acids are injected in utero into the lateral ventricle of embryos. Then, an electric field oriented toward the targeted cortical area promotes the entry of injected molecules into neural progenitor cells lining the ventricular wall. Electroporated embryos can then be examined at any developmental or adult stage. By controlling the embryonic day of development at which IUE is performed and the orientation of electrodes, this technique allows precise spatial and temporal control of exogenous nucleic acid expression. Furthermore, the use of specific promoters provides an additional layer of specificity to the technique. As a result, the IUE procedure is such that only a defined subset of cells will be targeted, leading to somatic mosaicism in the region of electroporation. To model FCDII, IUE is typically performed at embryonic day (E) 14.5, a developmental stage at which progenitors of superficial layer 2/3-destined pyramidal neurons are targeted. One reason is the ability to observe a migration defect, reminiscent of cortical dyslamination, between the initial location of progenitor cells lining the ventricular wall and the final location of their daughter cells expected to reach the upper cortical plate at E 18.5, or layer 2/3 at the postnatal stage (Fig. 43–2).
In utero electroporation experiment workflow. Timed pregnant mice are anesthetized, and uterine horns are exposed to inject a mixture of nucleic acids into the brain ventricle of embryos. Electrodes are then placed around the head to target the cortical (more...)
Mosaic hyperactivation of the mTOR pathway in the developing rodent neocortex in vivo by IUE has been broadly used in the recent years to generate models of FCDII (reviewed in Nguyen and Bordey, 2021). Several rodent models recapitulating mosaic mTOR-activating variants have been established to study the impact of mTOR hyperactivation on brain development and epilepsy. To date, mouse models consist of the overexpression of the humanized mutated genes of MTOR (Lim et al., 2015; Nakashima et al., 2015; Park et al., 2018; Koh et al., 2021; Pelorosso et al., 2019), AKT3 (Baek et al., 2015), PIK3CA (Yu et al., 2020; Zhong et al., 2021), and RHEB (Hsieh et al., 2016, 2020; Zhao et al., 2019; Nguyen et al., 2019; Proietti Onori et al., 2021).
Although less frequently found in patients, loss-of-function mutations in repressors of the mTOR pathway (DEPDC5, NPRL3, and TSC1/2) have also been reproduced in rodents. Most studies aimed to generate a complete gene knockout, either by expressing a Cre-recombinase (Tsc1, Feliciano et al., 2011; Depdc5, Dawson et al., 2020) or CRISPR/Cas9 constructs (Tsc1, Lim et al., 2017; Depdc5, Ribierre et al., 2018; Hu et al., 2018); few have selected a knock-down approach (Tsc2, Tsai et al., 2014; Moon et al., 2015), and one study has successfully generated a somatic knock-in of loss-of-function mutations in Tsc1 and Tsc2 (Lim et al., 2017).
Regardless of the mutant gene, all rodent models faithfully displayed an architectural disorganization of the neocortex due to abnormal neuronal migration reminiscent of cortical dyslamination. They also consistently exhibited cytomegalic neurons with dendritic hypertrophy and increased branching and mTORC1 hyperactivity, reminiscent of dysmorphic neurons. Curiously, none of the rodent models reproduced typical enlarged balloon cells with neuroglial immunoreactivity. This could be explained by the targeted cell type by IUE procedure or could reveal a human-specific feature not reproducible in mice.
One major difference between models is the variability in the epileptic phenotype. Indeed, while most models show the presence of spontaneous epileptic seizures occurring at various percentages from 30% to nearly all mice, some models only exhibited an increased susceptibility to induced seizures. This is most striking in models of Tsc1 and Depdc5 knockout for which IUE was performed at the same developmental stage and either led to the presence of spontaneous epileptic seizures (Tsc1, Lim et al., 2017; Depdc5, Ribierre et al., 2018; Hu et al., 2018) or no seizures (Tsc1, Feliciano et al., 2011; Depdc5, Dawson et al., 2020). This discrepancy could be attributable to the variability in the IUE procedure itself, as suggested by a study showing that IUE of a RHEB variant in the somatosensory cortex did not generate epileptic seizures, unlike IUE in the medial prefrontal cortex (Hsieh et al., 2016).
One peculiarity was reported specifically in the Depdc5 focal knockout mouse model (Ribierre et al., 2018). In addition to the set of FCDII neuropathological hallmarks, some mice presented with a single cluster of severe seizures terminating with a sudden unexpected death in epilepsy (SUDEP)-like event. This terminal phenomenon resembled SUDEP, a rare but tragic outcome in epileptic patients. This observation fits well with the clinical observation that pathogenic DEPDC5 variants are linked to an increased risk of SUDEP (Nascimento et al., 2015; Bagnall et al., 2016; Weckhuysen et al., 2016; Baldassari et al., 2019a) and a recent Depdc5 neuronal KO (Syn1 cKO) mouse model (Bacq et al., 2022).
In summary, “mTORopathy” mouse models of FCDII with Depdc5, Nprl3, or Tsc1/2 loss of function or Mtor, Pik3ca, Akt3, and Rheb gain of function consistently display typical cytoachitectural and behavioral abnormalities supporting impaired cortical circuit formation underlying epileptogenesis (Fig. 43–3). Moreover, mTOR hyperactivation is associated with enhanced synaptic transmission due to alterations in ion channel and synaptic protein expression (reviewed in Nguyen and Bordey, 2021; Lasarge and Danzer, 2014).
Main histopathological and behavioral hallmarks in FCDII rodent models. In utero electroporation (IUE) is typically performed at E14.5. At E18.5, layer II/III-destined progenitor cells already show a marked migration delay (DAPI/GFP images) (Ribierre (more...)
FCDII is specified into two histopathological entities, with (FCDIIb) or without (FCDIIa) balloon cells that contribute to MRI alterations, including transmantle sign and increased cortical thickness and gray/white blurring: balloon cells are more often associated with hypomyelination and presence of transmantle sign, while increased cortical thickness and blurred grey/white junction (due to increased neuronal heterotopia) are more commonly found in FCDIIa lesions (Muhlebner et al., 2012; Tassi et al., 2012; Colombo et al., 2012). Yet there is no clear distinguishing evidence during the clinical course, and both FCD subtypes are associated with intractable epilepsy. Dysmorphic neurons and balloon cells, although detected in a same macroscopic region, may show different distribution with balloon cells showing higher densities in subcortical regions versus cortical areas, suggesting migratory arrest during development. Dysmorphic neurons are identified as immature cytomegalic neurons with abnormal shape, enlarged soma, Nissl substance cytoplasmic aggregates, and enrichments for neurofilament proteins. Balloon cells are very large opalescent cells with eccentric nuclei, expressing glial markers, and possibly neuronal markers. The two canonical immunohistochemistry markers are Smi32, a pan-neuronal neurofilament H, for dysmorphic neurons, and vimentin for balloon cells (Blumcke et al., 2021; Najm et al., 2018). Both cell types also express hyperphosphorylated S6K1 and S6 proteins indicative of hyperactive mTORC1 signaling. Balloon cells express stem cell markers (such as SOX2, Oct-4, c-Myc, FOXG1, KLF4, Nanog, SOX3, and b1 integrin) and progenitor cell markers (such as vimentin, CD133, CD34, and Nestin), suggesting they are derived from radial glial progenitor cells (Orlova et al., 2010; Lamparello et al., 2007; Yasin et al., 2010). A subpopulation of balloon cells expressing Mcm2 may remain in the G1 phase, indicating a cell cycle defect (Thom et al., 2007). Recent genetic findings indicate that dysmorphic neurons and balloon cells carry the same identical somatic mutation; suggesting they originate from a common progenitor cell that acquired a somatic mutation (Baldassari et al., 2019b; Lee et al., 2021). How the same progenitor cell can give rise to two distinct pathological cell entities is still an open question. Thus, currently, the developmental origin and lineage relationship between dysmorphic neurons and balloon cells are still enigmatic. We review here current knowledge on their contribution to the cellular and neuronal circuits involved in epileptogenesis.
The exact contribution of dysmorphic neurons and balloon cells to the mechanisms of epileptogenesis in FCD is still largely unknown. Dysmorphic neurons are thought to play a central role in the initiation and propagation of epileptic discharges. Dysmorphic cytomegalic pyramidal neurons have abnormal passive membrane properties such as increased cell capacitance, consistent with increased membrane area, longer time constant, and lower input resistance compared with normal-appearing pyramidal neurons. There is evidence from electrophysiological recordings from freshly resected FCD tissue that dysmorphic neurons show intrinsic hyperexcitability and may be pro-epileptogenic (Abdijadid et al., 2015; Cepeda et al., 2003, 2005). In contrast, balloon cells are neurophysiologically rather silent and do not display active voltage- or ligand-gated currents, nor do they receive synaptic inputs (Cepeda et al., 2003). Moreover, dysmorphic neurons show substantial correlations to both seizure onset and interictal markers, while seizures do not seem to originate in areas with balloon cells. Co-registration of intracerebral electroencephalogram (EEG) recording with neuropathology findings indicated that seizure onset is correlated with the presence of dysmorphic neurons, but not balloon cells. Dysmorphic neurons are thought to be the source for repetitive epileptogenic discharges and ictal-onset patterns. Boonyapisit et al. reported that balloon cell–containing areas are less epileptogenic than those with dysmorphic neurons, possibly due to a severe disruption of the neuronal circuits (Boonyapisit et al., 2003). Rampp and colleagues provided evidence for a contribution of dysmorphic neurons to interictal spikes, fast gamma activity, and high-frequency oscillations (HFOs) by co-registration of intracerebral EEG and histology (Rampp et al., 2021).
Conceptualization of the epileptogenic zone is now evolving to integrate the mutation load, the dysmorphic neuron density, and cell-type-specific genetic data. A variant gradient was shown in surgical specimens with a DEPDC5 mutation, with a higher mosaic rate of the somatic variant in the seizure-onset zone than compared with the surrounding epileptogenic zone (Ribierre et al., 2018). In another study, the somatic mutation load correlated with both dysmorphic neuron density and the epileptogenic zone (Lee et al., 2019). Therefore, seizures originate from areas with the highest density of dysmorphic neurons harboring the somatic mutation. How clones of abnormal cells spatially distribute to form a focal lesion is strictly related to the way neural cells are born and migrate during normal brain development. Recent studies that used somatic variants as clonal markers suggest that there might be a high degree of clonal mixing in the normal human brain, but its full complexity is far from understood (Huang et al., 2020; Bizzotto et al., 2021; Coorens et al., 2021).
Clinically, it is still controversial whether the seizure-onset zone is located within the center of dysplastic area or at the periphery of lesions. Recent molecular findings revealing a correlation between mutation gradient, dysmorphic neuron density, and epileptogenicity support the likelihood of the epileptogenic seizure-onset zone being intralesional rather than “perilesional” (Lee et al., 2019).
One of the most consistent histological findings in FCDII rodent models is the neuronal migration delay and presence of mTOR-active cytomegalic neurons. Rapamycin can rescue the migration defect when administered prenatally (Tsai et al., 2014; Baek et al., 2015; Ribierre et al., 2018; Lim et al., 2017) while it can reverse the increased soma size of electroporated neurons, including when administered during adulthood (Tsai et al., 2014; Ribierre et al., 2018; Lim et al., 2015; Hsieh et al., 2016; Park et al., 2018; Hu et al., 2018; Iffland et al., 2020). Interestingly, neuronal soma size increase appears to positively correlate with mTORC1 activity levels (Nguyen et al., 2019). Hence, mTOR-pathway hyperactivity emerges as the main contributor to these defects, although which of the multiple downstream intracellular pathways are involved remains unclear. While mTORC1 signaling has a role in radial glia scaffolding and actin polarity (Liu et al., 2010), one study demonstrated that neuronal migration defect was not solely dependent on radial glia abnormalities but can result from intrinsic neuronal defects (Lin et al., 2016). This observation is further supported by a study reporting an mTOR-dependent autophagy defect leading to disrupted ciliogenesis in neurons themselves, responsible for their neuronal misplacement (Park et al., 2018). Other brain cytological abnormalities attributed to mTORC1 hyperactivity are commonly reported in FCDII models (Lasarge and Danzer, 2014). Among them, dendrite hypertrophy, dendritic spine swelling, and dendritic tree patterning are frequently described (Ribierre et al., 2018; Hsieh et al., 2016; Dawson et al., 2020; Zhang et al., 2020). Perturbed axonal growth and guidance has also been reported and was rescued by modulating 4E-BP1 or S6K1/2 pathways downstream of mTORC1 (Gong et al., 2015). Interestingly, multiple studies showed that ectopic neurons retain their molecular identity (Baek et al., 2015; Park et al., 2018; Iffland et al., 2020; Zhong et al., 2021), suggesting that mTORC1 hyperactivity from E14 perturbs migration yet does not alter cell fate.
Somatic mutations in mTOR-pathway genes lead to a continuum of disorders from FCDII to HME, depending on the percentage of mutated cells. It remains intriguing how as little as 1%–5% of mutated cells within the dysplastic zone can cause such severe and refractory focal epilepsies. Therefore, putative non-cell-autonomous effects of mutated neurons on their neighboring cells have been investigated across multiple studies. Local paracrine mTORC1 hyperactivity has been described in the vicinity of the electroporated area in Depdc5 and Mtor FCDII mouse models (Hu et al., 2018; Ribierre et al., 2018; Kim et al., 2019). Other studies showed a non-cell-autonomous effect on the migration of apparently healthy neurons through the secretion of reelin in an Akt3 mouse model (Baek et al., 2015) or the secretion of Rheb RNA-containing exosomes (Kim et al., 2019). Nonetheless, regarding morphological changes, two studies demonstrated that while mutated neurons exerted a self-reinforcement effect on their own abnormal growth, they did not affect neighboring nonmutated cells in a Pten mouse model (Arafa et al., 2019; LaSarge et al., 2019). Finally, a recent study described a paracrine effect of mTOR hyperactivity on the density of interneurons within the electroporated area (Zhong et al., 2021), as supported by histological stainings in human resected tissue showing decreased densities of interneurons, mainly parvalbumin-expressing (Blauwblomme et al., 2019; Nakagawa et al., 2017; Liang et al., 2020).
In utero electroporation-based rodent models, which often exhibit spontaneous seizures, allow one to investigate neuronal cortical network excitability changes directly underlying epileptogenesis and preceding the onset of spontaneous recurrent seizures.
Intrinsic electrophysiological properties of electroporated neurons show common features across genotypes (i.e., regardless of the mutated gene). Consistent with increased cell size, mutated neurons have increased capacitance and decreased input resistance, together accountable for an increased action potential threshold (rheobase), reflecting the input needed for action potential firing (Ribierre et al., 2018; Hu et al., 2018; Hsieh et al., 2016; Koh et al., 2021). Per se, this observation seems counterintuitive as an increased rheobase implies decreased excitability. However, increased excitability in electroporated neurons was shown once reaching their firing threshold (Ribierre et al., 2018; Hu et al., 2018). Hence, one hypothesis to explain neuronal hyperexcitability includes a greater dendritic integration of synaptic input due to extensive branching, which compensates for increased rheobase and therefore translates to increased firing of action potentials.
Regarding the synaptic properties of electroporated neurons, too few studies report synaptic function characterization, and no consensus emerges, stressing the need to systematically investigate these properties in future work. While the spontaneous excitatory postsynaptic current frequency but not amplitude was increased in a Pten model (Chen et al., 2015), no change in amplitude but a decrease in frequency was reported in a Rheb model (Lin et al., 2016), and no change in frequency but a slight increase in amplitude was reported in a Depdc5 model (Ribierre et al., 2018). Hypotheses explaining these discrepancies include the age at which recordings were performed (with regard to critical periods of spine autophagy and synaptic maturation), the cortical region and layer in which ectopic electroporated neurons are located (with regard to their input connectivity), and possibly the genotype (with regard to their impact of the levels of mTORC1 hyperactivation and retrograde regulation loops).
Although the mechanisms underlying neuronal excitability at the cellular level remain unclear, a common denominator of FCDII rodent models is the localized cortical network excitability triggering epileptic seizures. Multiple studies in human tissues have shown a clear correlation between the density of mutated cells, the extent of the dysplasia, and tissue epileptogenicity in human surgical specimens (Ribierre et al., 2018; Lee et al., 2021). Recently, novel perspectives have been brought on the relationship between electroporated neurons and the surrounding unaffected brain to explain cortical hyperexcitability and seizure generation.
One study in an Mtor mouse model reported that hyperexcitability underlying epileptogenesis was derived from nonmutated neurons directly surrounding electroporated neurons in which they found increased expression of adenosine kinase (ADK) gene. To explain this finding, authors suggested that increased adenosine kinase activity could lead to extracellular depletion of adenosine resulting in local neuronal hyperexcitability of neurons, as shown by the rescue effect of a selective adenosine A1 receptor agonist (Koh et al., 2021).
Epileptic seizures in rodent models of FCDII rapidly generalize (Ribierre et al., 2018; Kao et al., 2021), and often their focal initiation cannot be observed. This can be explained first by the extent of the electroporation itself, and second by the low spatial coverage of intracortical electrode devices that is unable to catch a timely difference in the ictal event prior to generalization. To explain the phenomenon of rapid generalization of seizures, one study reported enhanced axonal innervation and synaptic connectivity in the contralateral hemisphere of a Rheb mouse model (Proietti Onori et al., 2021). However, in contrast to Koh et al. (2021), authors demonstrated the absence of changes in electrophysiological properties and excitability of neighboring proximal nonmutated neurons, while striking changes were observed in distally connected nonmutated neurons. They therefore conclude that contralateral nonmutated neurons innervated by electroporated neurons of the opposite hemisphere might be the main driver of epilepsy in this model.
Interestingly, evidence from the work of Kao et al. (2021) and Koh et al. (2021) demonstrates that tissue excitability and epileptogenesis emerge from the electroporated region, in line with studies in humans showing co-registration of MRI/PET combined to SEEG describing a clear correlation between the seizure-onset zone and the heart of the dysplastic lesion (Desarnaud et al., 2018). Nonetheless, independent epileptogenesis emerging from the contralateral hemisphere following FCD epilepsy surgery has been reported (Guerrini et al., 2021), which could be compatible with the hypothesis of a long-range non-cell-autonomous effect as described in Proietti Onori et al. (2021).
Altogether, the recent studies have opened the path to multiple hypotheses and prompt a deeper investigation of the many cellular mechanisms involved in cortical tissue hyperexcitability.
Access to human brain samples from patients with central nervous system disorders is a major limitation in neurological disease research. Almost uniquely for neurologic disorders, fresh postoperative tissues from refractory focal epilepsies are available for genetic, biochemical or electrophysiological analyses of pathophysiologic-relevant brain structures. However, limited information exists regarding the developmental mechanisms contributing to epileptogenesis in the context of a cortical malformation. To overcome the many hurdles inherent to human tissue availability, and investigate developmental processes, human pluripotent stem cell (hiPSC) technology has been introduced in three-dimensional self-organized structures, referred to as brain or cerebral organoids. Brain organoids recapitulate some features of human cortical development, mainly the typical progenitor zone organization with outer radial glia (oRG) progenitors (Velasco et al., 2020). They are increasingly being used as model systems to gain new insights into neurodevelopmental disorders since they bridge the gap between two-dimensional human cell cultures and nonhuman animal models. Single-cell RNA sequencing approaches have led to significant advances in understanding the cell type composition, developmental time course, and molecular taxonomy of brain organoids, while electrophysiological recordings and live imaging allow functional analysis of network wiring. Most conventional electrophysiology techniques can be applied to cerebral organoids such as whole-cell patch-clamp for high temporal resolution of neural activity and live calcium imaging (for spatial resolution). Microelectrode array (MEA) recordings have been increasingly adopted due to the ability to provide both high temporal and network-scale resolution (reviewed in Passaro and Stice, 2020). With the advent of gene-editing tools, most notably CRISPR/Cas9, allowing rapid and efficient manipulation of gene expression, organoids are in principle an ideal system to model somatic mosaicism, such as in FCDII, during cortical development.
Cortical organoids have been shown to be useful for its application to epilepsy studies since they display electrical activity and structured network activity (reviewed in Nieto-Estevez and Hsieh, 2020). Indeed, several techniques including whole-cell patch-clamp, MEA recordings, and live calcium imaging can detect spontaneous neuronal activity and complex network activities in organoid-derived cortical neurons (Khan et al., 2020; Sun et al., 2019), therefore holding great promises in deciphering the molecular and network dynamics underlying epileptogenesis and neurodevelopmental defects. Trujillo and colleagues showed that human cortical organoids exhibit electrical activity and structured network activity that increases during the maturation process. In this protocol, cortical organoids begin to exhibit highly synchronous and stereotypical network activity at 2 months, which transitions into rhythmic activity by 4–6 months. More mature cortical organoids (6–10 months) exhibit periodic and highly regular nested oscillatory network events that are dependent on glutamatergic and GABAergic signaling (Trujillo et al., 2019). Moreover, complex network dynamics such as multifrequency oscillations can be detected using calcium sensor imaging and extracellular recording of local field potentials (LFPs) in fusion organoids consisting of an integrated cortex–ganglionic eminence organoid containing a mixture of excitatory and inhibitory neurons (also called assembloids). Such fusion organoids when derived from iPSCs from individuals with Rett syndrome, a neurodevelopmental disorder caused by MECP2 mutations, displayed epileptiform-like activity that could be rescued with a neuroregulatory TP53 target inhibitor drug, pifithrin-α (Samarasinghe et al., 2021). These findings illustrate the potential value of the organoid modeling approach in personalized drug discovery. Sakaguchi and colleagues used calcium imaging to show that cerebral organoids display self-organized and complex human neural network activities that include synchronized and nonsynchronized patterns (Sakaguchi et al., 2019).
Outer radial glial (oRG) cells are a major progenitor population during the neurogenic period of human cortical development and show specific activation and function of the mTOR signaling activity at this stage. Andrews and colleagues showed in human organoids that mTOR signaling pathway regulates oRG morphology and migration by changing the actin cytoskeleton through the activity of the Rho-GTPase, CDC42 (Andrews et al., 2020). This study emphasizes the key role of mTOR during human cortical development and further validates cerebral organoids as a valuable model to investigate mTORopathies.
Tuberous sclerosis complex (TSC) is part of the spectrum of “mTORopathies”. TSC is a multisystem developmental disorder caused by germline mutations in the TSC1/2 genes, whose protein products are repressors of mTORC1, therefore leading to mTORC1 hyperactivation. Somatic second-hit mutations in TSC1/2 genes are frequently reported in TSC-related tumors (hamartomas) and have been detected in a subset of cortical tubers, which are dysmorphic-containing regions associated with epileptogenesis (Martin et al., 2017). Blair and colleagues used CRISPR/Cas9 gene editing to insert loss-of function mutations in TSC1 and TSC2 human embryonic stem cells (hESCs) and to generate the “two-hit” event (constitutive loss-of-function TSC2 mutation in one allele and a Cre-inducible conditional inactivation in the second allele) in neural progenitors (Blair et al., 2018). The study brings evidence that second-hit mutations are necessary and sufficient to generate mTOR-hyperactive dysplastic cells and induce a strong bias toward an astroglial cell fate in developing human cortical spheroids. Chronic treatment of cortical organoids with rapamycin prevented mTORC1 hyperactivation and cellular hypertrophy of TSC2–/– cells.
Cortical organoids have also been used to model other neurodevelopmental disorders, including MCDs associated to epilepsy such as microcephaly syndromes (Lancaster et al., 2013; Esk et al., 2020), lissencephaly (Bershteyn et al., 2017), periventricular heterotopia (Klaus et al., 2019), or 22q11 syndrome (Khan et al., 2020).
Although the organoid field is still young and needs technological improvement to increase organoid formation efficiency, limit variability, end-point morphology, and function, brain organoids derived from human stem cells present enormous potential for disease modeling and understanding human neurodevelopment. They also represent an ideal setting to study the molecular alterations responsible for cortical malformations and epilepsy. Furthermore, they will allow for uncoupling malformation from epilepsy and ultimately exploring/screening mTOR-independent functions. At last, these models will soon be amenable to large-scale production and phenotyping, for translational studies such as drug screening prior to drug testing in more complex animal models. These exciting models will accelerate our knowledge of human brain development and likely lead to new strategies for treating epilepsy and personalized medicine.
The debilitating effects of seizures in refractory focal epilepsy are a burden to patients and their families. So far, the only therapeutic option is the surgical removal of the seizure-onset zone, which can represent a small area of a gyrus up to a whole hemisphere. This invasive approach yields a high success rate of worthwhile improvement. However, many patients are ineligible for epilepsy surgery, while others unfortunately are not seizure-free after the intervention. Thus, precision medicine is greatly needed for these patients. mTOR hyperactivity has been studied in the context of other disorders, including cancers, such that small-molecule inhibitors of PI3K, mTOR, and AKT have been developed. Besides clear oncological indications, these drugs might therefore be useful for the treatment of malformation- associated lesional epilepsies. Promising inhibitors include sirolimus and everolimus (mTOR inhibitors), duvelisib inhibitors (PI3K inhibitor), or Miransertin (AKT inhibitor). Many open questions regarding tolerability, patient selection, sensitivity markers, development of resistances, and toxicology remain to be clarified for these and other emerging molecules. Moreover, no data have demonstrated yet whether seizure control and cognitive development can be improved with an mTOR pathway-targeting compound to address an early developmental lesion.
Most functional studies in rodents have demonstrated the potency of rapamycin on preventing or reversing the main pathological hallmarks of FCDII. In human, the use of rapalogs (analogs of rapamycin) is under several clinical trials. A recent clinical trial administered sirolimus to a cohort of 16 FCDII patients and reported that the reduction of focal seizures did not meet the predetermined level of statistical significance. However, focal seizure frequency reduced over time in sirolimus-treated patients (Kato et al., 2022).
Recently, a few studies have tackled the challenge of precision medicine, either by identifying gene-specific deregulations downstream of mTORC1 or by leveraging precise administration of therapeutic agents. Recently, two candidates have emerged for putative precision therapy in a Rheb mouse model by modulating filamin A (FLNA) (Zhang et al., 2020) or HCN4 (Hyperpolarization Activated Cyclic Nucleotide Gated Potassium Channel 4) gene expression (Hsieh et al., 2020). More specifically, HCN4 was found specifically expressed in pathogenic cells (and not in normal cells). Specificity is key for precision medicine in FCD, and therefore HCN4 appears as a promising targetable biomarker of epileptogenic cells. Another study reported a promising effect of antisense oligonucleotide targeting Rictor, a component of mTOR complex 2, in a Pten knockout epileptic model (Chen et al., 2019). These approaches might necessitate chronic administration of therapeutic agents and remain to be further tested in different rodent models of FCD to define whether they are specific to an etiology or applicable regardless of the mutated gene.
Altogether, the future of precision medicine for FCDII lies in the identification of specific robust targetable biomarkers regardless of the mutated gene. Ideally, therapeutic agents would selectively remove (by triggering apoptosis or recruiting immune cells for clearance) rather than modulate the phenotype of pathogenic cells, which would therefore be in favor of acute therapy, in opposition to the need of chronic administration and lifelong therapy to control seizures.
The authors declare no conflicts of interest.
This work was supported by Ligue Française contre l’épilepsie, AXA Research Funds and Fondation Maladies rares (FRM) to TR.
This is an open access publication, available online and distributed under the terms of a Creative Commons Attribution-Non Commercial-No Derivatives 4.0 International licence (CC BY-NC-ND 4.0), a copy of which is available at https://creativecommons.org/licenses/by-nc-nd/4.0/. Subject to this license, all rights are reserved.
Your browsing activity is empty.
Activity recording is turned off.
See more...
