Skip to main content

Development and Epilepsy - Strategies for Innovative Research to improve diagnosis, prevention and treatment in children with difficult to treat Epilepsy

Final Report Summary - DESIRE (Development and Epilepsy - Strategies for Innovative Research to improve diagnosis, prevention and treatment in children with difficult to treat Epilepsy)

Executive Summary:
Within DESIRE, we collected and assessed clinically, neurophysiologically and with neuroimaging 2431 patients/families, 1042 with focal or diffuse malformations of cortical development (MCD) and 1389 with epileptic encephalopathies of seemingly non-lesional origin. Combining array CGH, massive parallel targeted resequencing, whole exome sequencing and functional studies in cellular and animal models, we to identified novel causative genes for epileptogenic developmental disorders (EDD) and extend knowledge on genotype/phenotype correlations for many known causative genes. Studying mechanisms of epileptogenesis, we demonstrated that, in animal models (freeze lesion model), seizures start in somatosensory cortex, preferentially close to the lesion, and that lacosamide has relatively specific anti-epileptic effects.
In human brain slices, we found that the seizure onset site is the same in consecutive seizure like events and that acetazolamide is effective in blocking SLEs already at low concentrations.
The study of pathological mechanisms in Epileptic Encephalopathies (EE), intended as gene-environment(s) interactions, has been another main objective of DESIRE. For this topic, we focused on Dravet syndrome (DS) and on the epilepsy-aphasia spectrum (EAS) disorders, which represent two extremes of the EE spectrum. We compared the results of clinical studies with those obtained from reliable experimental models. Analyses already performed and ongoing show that, in addition to the type of mutation, the age at onset and the type and frequency of seizures contribute in determining the EE outcome. Our data also indicate that identifying the initial causes (e.g. genetic mutations) and homeostatic and pro-pathologic responses induced by interactions with the internal and the external environment is essential for identifying predictive biomarkers for disease course and targets for therapies, a necessary step to head towards personalised disease treatments.
High frequency oscillations (HFO) are oscillatory electrical neural events closely linked to epileptogenicity. SMEI Micromed developed an advanced hardware and software prototype for HFO analysis. Using this prototype, we found a correlation between scalp and intracranial ripples, seizure onset zone (SOZ) and postoperative outcome. We also demonstrated that Ultra-High Field (UHF) 7T MRI is a useful tool for revealing hidden epileptogenic lesions and has the potential to improve the assessment of FCD compared to lower field MRI.
Analysing our unique data collection of more than 10,000 patients, i.e. the European Epilepsy Brain Bank, we identified histopathological diagnosis, age at surgery, and disease duration as significant predictors of outcome. We also showed that convergent concordance among electro-magnetic, metabolic and hemodynamic data with stereoelectroencephalography (SEEG) invariably predicts a positive outcome after surgery. In addition, we developed new tools to localize the epileptogenic zone (EZ). Attempting to discover molecular signatures that can be translated into clinically useful classification tools, we extended the evidence for disease-specific methylation signatures towards focal epilepsies. We also established a low input miRNA sequencing protocol, implementing SMEI CeGaT portfolio, which made it possible to detect miRNA expression patterns in microvesicles, platelets, and neuronal and non-neuronal cells.
To improve epilepsy treatment in children with FCD ILAE Type II, we planned to setup a clinical trial assessing whether pre-treatment of children with epilepsy undergoing surgical resection of FCD improved seizure free outcome. The appropriate trial documentation underwent a translation process in four languages (Italian, German, French and Czech). Out of the 14 participating sites, 5 opened to recruitment and an additional 4 received on-site training. Although we undertook work to open further sites across Europe, there were continued delays with contracting and approval processes. In opened sites, it was also clear that numbers required for the primary outcome would not be achieved. In total, we have been able to submit 4 samples for methylation studies.
Within activities we performed for identifying innovative strategies for treatment and prevention of EDD related epilepsy and its consequences, we developed engineered nanoparticles that represent a promising first step for implementing a novel, non-invasive therapeutic and diagnostic tool for Tuberous Sclerosis. In addition, we successfully tested two different approaches to switch off hyperactive neurons. These nanoparticles represent a valid tool to be implemented for pre-clinical and clinical studies. We also setup protocols to selectively kill sick neurons within the intact brain, prevent epileptic manifestations by inhibiting the activity of specific neurons, control gene transcription in the epileptic environment through optogenetics and pharmacological approaches, and develop novel cell therapy approaches reprogramming fibroblasts into functional inhibitory neurons.

Project Context and Objectives:
Epilepsy and epileptiform disorders impose a major medical and socio-economic problem due to their high prevalence and severity.
DESIRE focuses on epileptogenic developmental disorders EDD, i.e. early onset epilepsies whose origin is closely related to developmental brain processes. A major cause of EDD are malformations of cortical development (MCD), either macroscopic or subtle. EDD are often manifested as epileptic encephalopathies (EE), i.e. conditions in which epileptic activity itself may contribute to severe cognitive and behavioural impairments. EDD are the most frequent drug-resistant pediatric epilepsies carrying a lifelong perspective of disability and reduced quality of life.
Updated and comprehensive 2010 estimates for European countries (Olesen et al, 2012; Eur J Neurol, 19:155-62) revealed an annual economic cost of about 13.8 billion euros for epilepsy, and drug-resistant epilepsies would consume the most significant portion.
Although EDD collectively represent a major medical and socio-economic burden, their molecular diagnosis, pathogenic mechanisms (PM) and rationale treatment are poorly understood.
Lack of knowledge and of targeted therapeutic interventions create an enormous burden to families confronted with such disorders.
As many patients with EDD develop drug resistance and cognitive decline, improved diagnostic strategies are mandatory for personalized therapy.
Specific objectives of DESIRE are to advance the state of the art with respect to:
(1) the genetic and epigenetic causes and PM of EDD, particularly epileptogenic MCD, to elucidate molecular networks and disrupted protein complexes and search for common bases for these apparently heterogeneous disorders.
(2) the diagnostic tools (biomarkers) and protocols through the study of a unique and well-characterized cohort of children to provide standardized diagnosis for patient stratification and research across Europe.
(3) treatment of EDD using randomized, multidisciplinary clinical protocols and testing preclinical strategies in experimental models to also address novel preventative strategies.

Project Results:
WP1
The main objectives of WP1 were: 1) To identify novel genetic determinants of epileptogenic malformations of cortical development (MCD) and epileptic encephalopathies with seemingly non-lesional origin; 2) to understand pathomechanisms underlying epileptogenic MCD, especially for the genes exhibiting a very low frequency of mutations and those implicated in focal cortical dysplasia (FCD); 3) To elucidate epigenetic regulatory mechanisms of gene expression in FCD subtypes and discriminate epigenomic changes intersecting transcriptome and genome methylation with metabolic intervention models; 4) To provide insights into the biological and cellular processes underlying MCD in cellular and animal models; 5) To complement functional data obtained from animal models performing pharmacological and metabolic characterization of human acute dysplastic neocortical slices.

Within this WP, we collected 2431 patients/families (1042 with focal or diffuse malformations of cortical development and 1389 with epileptic encephalopathies with seemingly non-lesional origin). We assessed each patient enrolled in the study according to the protocol established in DESIRE and stored clinical history, neurological examination, MRI evaluation, video-EEG recordings, neuropsychological assessment, and genetic data in dedicated databases in each centre involved in this task.

Combining array CGH analysis, massive parallel targeted resequencing, and functional studies, we contributed to extend knowledge on genotype/phenotype correlations for several epileptogenic developmental disorders (EDD). In particular, we: 1) confirmed the causative role of somatic duplications involving AKT3 in focal malformations of cortical development; 2) identified a single mutational event in FLNA, resulting in co-occurring gain-of and loss-of-function effects, associated with a novel complex phenotype including PNH, epilepsy and Melnick-Needles syndrome; 3) confirmed the causative role of mutations in CHRNA2 in ADNFLE and identified a novel (loss of function) disease mechanism for mutations in this gene; 4) demonstrated that GABRA1 gene should be considered in patients with generalised EE, especially those with infantile-onset, prominent tonic-clonic and myoclonic seizures, generalized spike-and-wave discharges, and a photoparoxysmal response on the EEG; 5) proved that a hyperkinetic–dyskinetic movement disorder should be considered a distinctive feature of the FOXG1-related phenotype and that, in children with epileptic encephalopathy with migrating partial seizures, concomitant multisystem involvement should prompt investigations for Congenital Disorders of Glycosylation (CDG); 6) contributed to demonstrate that SPTAN1-related disorders comprise a wide spectrum of neurodevelopmental phenotypes ranging from mild to severe and progressive; 7) proved that KCNB1 mutations are associated with diffuse brain dysfunction combining seizures, motor, and cognitive impairment; 8) broadened the AMPD2-related clinical spectrum by describing individuals with and without microcephaly and the characteristic “figure 8” shape of the midbrain; 9) showed that the imaging phenotype associated with variants in TUBG1 is more in line with that resulting from variants in LIS1 instead of those caused by variants in other tubulin genes; 10) contributed to demonstrate the causative role of GNAO1 mutations in an expanded spectrum of early-onset epilepsy and movement disorders, frequently exacerbated by specific triggers and at times associated with self-injurious behaviour; 11) contributed to demonstrate that GABRB3 mutations are associated with a broad phenotypic spectrum of epilepsies and that reduced receptor function causing GABAergic disinhibition represents the relevant disease mechanism, 12) demonstrated that LIS1 mutations can cause familial epilepsy associated with mild pachygyria in the posterior brain.
Taken together, these results emphasize the benefit of accurate genetic diagnosis in patients with neurodevelopmental disorders with epilepsy. We also demonstrated that panels targeting about 100 genes represent the best cost-effective diagnostic option in paediatric drug-resistant epilepsies. Indeed, analysing panels targeting such number of genes enables molecular diagnosis of atypical phenotypes, broadening phenotype-genotype correlations.

Using whole exome sequencing (WES) in combination with targeted massive parallel sequencing, we identified novel causative genes for EDD as well as novel genotype/phenotype correlations. In particular, we: 1) contributed to demonstrate that mutations in MRM2 causes a MELAS-like phenotype, and suggested the genetic screening of this gene in patients with MELAS-like presentation who are mutation negative to m.3243A>G screening; 2) proved that EML1 and NEDD4L are novel causative genes for giant and periventricular nodular heterotopia respectively and that mutations in KIF2A cause pachygyria and microcephaly; 3) demonstrated that both constitutional and mosaic mutations in PIK3R2 cause a spectrum of developmental brain disorders ranging from isolated bilateral perisylvian polymicrogyria with normal head size to bilateral perisylvian polymicrogyria with large head size, causing the megalencephaly-polymicrogyria-polydactyly-hydrocephalus syndrome (MPPH); 4) contributed to demonstrate that MTOR gene mutations are associated with a spectrum of brain overgrowth phenotypes extending from FCD type 2a to diffuse megalencephaly and that these phenotypes can be distinguished by different mutations and levels of mosaicism; 5) substantiated the link between mosaic mutations of MTOR and pigmentary mosaicism in the skin; 6) proved that PIK3CA-related overgrowth comprises a discontinuous spectrum of disorders that correlate with the severity and distribution of mutations. Through WES, we also showed that de novo heterozygous ATP6V1A mutations cause developmental encephalopathy with epilepsy and that homozygous or compound heterozygous DMXL2 mutations cause developmental encephalopathy with Ohtahara syndrome, two conditions that had never been associated with mutations in these genes.

Performing functional studies in cellular and animal models (i.e. in utero electroporation and overexpression of mutations identified in patients), we characterized expression profile of the novel causative genes identified through WES and investigated neuronal migration defects induced by their mutation. We used the same approach to increase knowledge on some known EDD causative genes.
Characterizing the Eml1 model, we found that this gene has a cell cycle-dependent localization enriched along spindle microtubules. In agreement with these observations, we demonstrated that the brain of HeCo mouse (a spontaneous model of Eml1 inactivation) shows changes in spindle orientation and significantly increased spindle length, which may contribute to progenitor cells detachment. Eml1 mutant cells are also significantly larger than control cells and exhibit perturbed microtubule dynamics that may contribute to the spindle length abnormalities. Eml1 heterotopia models also exhibit defects in the ventricular surface.
Silencing of DLGAP4, a candidate gene for heterotopia, polymicrogyria, and partial agenesis of the corpus callosum, we demonstrated that this gene is associated with cell disorganization, suggesting a possible role in maintaining cell polarity at the ventricular lining.
Using cellular and animal models, we also proved that alterations in NEDD4L cause disruption of the interplay between ubiquitination mediated by NEDD4L itself and mTOR signalling pathway. These results revealed a potential novel disease-causing molecular mechanism in which missense mutations might lead to a constitutive active state and loss of the mutant protein, but with functional consequences that are different from constitutive haploinsufficiency.
Performing studies in cell lines, patients’ fibroblasts and primary rat neuronal cultures, we proved that de novo heterozygous ATP6V1A mutations perturb lysosomal homeostasis and neuronal connectivity and cause defects in neurite elongation as well as loss of excitatory inputs. Using the same approach, we demonstrated that homozygous or compound heterozygous in DMXL2 significantly affect outgrowth and branching of neuronal processes during development.
Functional studies we carried out in cellular and animal models overexpressing RPS6 and MTOR mutations we found in hemimegalencephalic (HME) uncovered a novel two-hit based pathogenic mechanism. This mechanism is based on two independent activating somatic mutations affecting a single allele of different positive actors in the mTOR pathway. These studies also identified RPS6 as a novel HME related gene.
Analysing brains of FlnAEmx1-Cre mice, we identified a clear enlargement of the lateral ventricles together with morphological alterations of the hippocampal shape, and a discrete dense cellular mass suggestive of a rather local perturbation of neuronal migration reminiscent of the nodular heterotopia reported in FLNA patients. We did not find gross alterations in cortical structure as well as cortical layering. Neurons in brains of FlnAEmx1-Cre mice exhibited a slight but significant increase in their dendrites’ outgrowth when compared with control neurons. This increase involved specifically the arborisation of the basal dendrites component. We also proved that FlnA plays an important role in dendritogenesis via ARHGAP24/Rac1 signalling pathway. In rodents, we also proved that PIK3R2 overexpression causes an accumulation of ectopic neurons that are differentiated as upper-layer neurons in the white matter and deep cortical layers of the brain.

Within WP1, we also set up novel protocols to study genes of interest in unconventional animal models (i.e. the ferret). The studies we performed in ferrets showed that the Outer Subventricular Zone (OSVZ) is a key element of cortical development in large and gyrencephalic cortices. Studying the mechanisms involved in the formation of the OSVZ during embryonic development in the ferret, we also demonstrated that cells in this proliferative zone follow a lineage completely independent from the other germinal layers, a completely unexpected finding, different from what is seen in rodents. In addition, performing in utero electroporation studies in this animal model, we determined that gain of FlnA function severely impairs the integrity of the apical junction belt (which holds together radial glia cells). As a result, most of these cells either delaminate prematurely or die. We also demonstrated that overexpressing mutant and wild type PIK3R2 in ferret brain led to a significant increase in radial glia progenitor cells and intermediate progenitor cells. This increase is concomitant with changes in cell cycle parameters as well as accumulation of ectopic neurons in the white matter and deep cortical layers in the juvenile cortex, which is in part maintained in the mature cortex. Studies we performed in the ferret also demonstrated that the neuronal migration defect caused by overexpression of PIK3R2-WT is consolidated postnatally and it results in neuronal heterotopias and extra gyri.

To study epigenetic alterations in epileptogenic MCD, we generated high-density autosomal maps of CpG methylation sites and established high-density autosomal methylation quantitative trait loci (mQTL) and expression quantitative trait loci (eQTLs) maps in hippocampal brain tissue dissected from patients with pharmacoresistant temporal lobe epilepsy (TLE), identifying novel imprinted loci. In particular, we identified 194 CpG islands (CGI) displaying differences in methylation residuals between the dysplastic and non-dysplastic tissue blocks and a six-fold increase of meQTLs in bioptic dysplastic brain specimens compared to previously studied post-mortem brain tissue. We also showed methylation signatures to be highly distinct in seizure (e.g. FCD) versus non-seizure control samples and non-FCD seizure groups (i.e. TLE-HS). In particular, we identified a significant anti-correlation specifically at promoters. Hypermethylation was associated with reduced gene expression whereas hypomethylation was primarily present in upregulated genes. We also found that hypermethylated genes downregulated in FCD type IIb patients shared specific binding motifs. Among them, H3K27me3 signature primarily help to distinguish FCD type IIb samples from FCD type IIa, non-seizure controls and TLE-HS. We demonstrated significant (P = 6.8 x 10-12) differences between FCD Type IIa and FCD Type IIb for cg04221606, which is located in a CGI shore flanking the transcription start site of AKAP12, an effector of the rapid epithelial barrier reconstruction after injury, and for cg09762612 (P = 7.5 x 10-8), which is located in a CGI in exon two of the LT3 Interacting Zinc Finger 1 (FIZ1) on 19q13.42. We also demonstrated that the majority of the meQTLs in brain tissues are also detectable in whole blood cells, providing a valuable key to select accessible epigenetic biomarkers for brain disorders in such tissue. Globally, our data demonstrated that FCD type I, IIa, and IIb can be classified based on DNA methylation signatures which are independent from genetic features and that epigenetic modifications specifically affect different pathways in different malformations, providing a pathology-specific signature. We also identified several candidate genes which may exert epigenetic effects on neurodysplastic and epileptogenic processes underlying pharmacoresistant epilepsies in FCD Type-II patients.

To elucidate mechanisms underlying epileptogenesis, we also implemented the rat neonatal freeze lesion model of FCD and set-up intrinsic optical signal (IOS) recording in rat and human brain slices. In addition, we performed preliminary studies of neuronal and metabolic activity coupling in hippocampal slices from adult rats. Through IOS recordings in the freeze lesion model, we demonstrated that Levetiracetam has no effect on the seizure like events (SLEs) even at a saturating concentration. Investigating the role of lactate for ionic homeostasis maintenance and synaptic transmission, we found that it supports recovery of activity dependent ion concentration changes and synaptic transmission in rat CA3 hippocampus. In the freeze lesion model, we also proved that seizures start in somatosensory cortex, preferentially close to the lesion, and that Lacosamide is effective, adding new experimental evidence complementing clinically based that indicate the effectiveness of this molecule as an anti-epileptic drug. Measuring IOS in human brain slices, we observed a Ca-driven enhanced NADH production that accounts for the overshoot phase related to NAD(P)H autofluorescence during stimulation induced activity changes in the hippocampus. These measurements demonstrated that, in human brain slices, seizure onset site is the same in consecutive SLE. In addition, exploring the effects of different anti-epileptic drugs on tissue resected from patients with temporal lobe epilepsy, we found that zonisamide and bumetanide exert different, patient specific, effects while acetazolamide is effective in blocking SLEs already at low concentrations.

WP2
The study of pathological mechanisms in Epileptic encephalopathies (EE), intended as gene-environment(s) interactions, has been the main objective of the WP2. EE are severe disorders of infancy and childhood that present with often multiform and intractable seizures, with major persistent cognitive, behavioural, and neurological deficits and, sometimes, early death. They represent the most severe expression of epileptogenic developmental disorders. Epileptic activity at times discontinues in adolescence, but in general serious neurocognitive deficits persist. The definition of EE is based on the hypothesis that epileptic activity itself may contribute to progressive clinical and neuropsychological deterioration. WP2 was focused on Dravet’s syndrome (DS) and on epilepsy-aphasia spectrum (EAS) disorder (in particular Epilepsy with Continuous Spike-Wave during sleep, CSWSS, and Landau-Kleffner syndrome, LKS), which are the two extremes of the EE spectrum. DS shows a relatively consistent epilepsy phenotype and is caused by a well-defined molecular defect: >90% of patients carry loss of function mutations of the SCN1A gene, the NaV1.1 voltage-gated Na+ channel. EAS can occur in association with different types of epilepsies, with or without underlying pathology and are thought to impair cognition and language through the continuous epileptic activity that disorganizes sleep, thus interfering with the memory consolidation process. Less than 20% of the patients carry loss of function mutations of the GRIN2A gene, a subunit of NMDA receptors. The results of clinical studies were compared with those obtained from reliable experimental models. For DS, gene targeted animal models that closely recapitulate genetic mutations and clinical features were available. For EAS, we employed a phenotypic mouse model presenting with sleep activated spike and waves discharges (SWD) and a genetic one (GRIN2A knock-out) with an incompletely characterized phenotype.

The nine clinical partners from five European countries involved in WP2 recorded in a dedicated DS database data of 196 patients recruited for a longitudinal study with homogenous criteria. Clinical information analysis showed that age at onset, seizure types and frequency are correlated with clinical progression and cognitive deterioration, suggesting that an early and intensive therapeutic intervention might favourably affect the prognosis of this devastating disorder. Particularly interesting for the longitudinal evaluation was the information drawn from 46 patients enrolled in the prospective study. All these patients fulfilled the inclusion criteria: 1) Febrile or prolonged afebrile seizure’s onset in the first year of life; 2) Normal development before seizure onset; 3) No risk factors for symptomatic epilepsy 4) Age < 18 months at first evaluation. All patients underwent standardized neuropsychological assessment and EEG at seizure onset and at the following time points: 6, 12, 18, 24, 36, 48, 60 months of age. During follow up, in most cases (83%) seizures became polymorphic, and the percentage of patients with myoclonic seizures increased from 10% (at 12 months of age) to 50% (at 48 months of age). At disease onset, all patients had normal neurological examination, the sequential assessments revealed that neurological signs appeared from the 12th month. At 24 months, half of the patients had neurological signs, mainly ataxia and myoclonus and, at 48 months of age, only six patients had a normal neurological examination. The first neuropsychological assessment revealed a normal psychomotor development in all patients, the differential general quotient obtained between the first and the last evaluation of each patient showed, in all, progressive psychomotor deterioration with individual differences.

A second group of patients was included in the protocol of historical study if the first observation was antecedent the beginning of DESIRE, provided that they: 1) Had been followed up by one of the DESIRE partners since the earliest phases of the disease 2) Had a first neurocognitive evaluation before the age of 24 months 3) Had at least one examination after 24 months (and any other subsequent control should have been recorded). Patients fulfilling the criteria for the historical study (150) have been (and are being) evaluated separately from the prospective group, but the information obtained by both studies will eventually concur in drawing the neurocognitive outcome of DS in a significant number of patients. Available results of the ongoing analysis point out that: 1) the disease onset after the 6th month of age represents a positive prognostic factor and in general, the later the onset the better the outcome; 2) myoclonic seizures represent a negative prognostic factor; 3) the recurrence of status epilepticus does not have a clear relation with the cognitive outcome.

For genotype-phenotype correlations, we analysed only patients with known SCN1A mutations. Thus, the analysis included only 167 patients, 44 males and 123 females, whose mean age at last follow-up was 13.5 years (median 11.5 years; range from 1 to 50 years). Information about family history of seizures was available in 118 probands and was positive in 51 (43%). Age at seizure onset for all 167 patients ranged from 1 to 15 months with a mean and median of 5 months. Globally, patients included in the cohort carried 89 mutations (53.3%) with a truncating effect (including splice site mutations and deletions); 70 missense mutations (42%) and 8 mutations (4.7%) with unknown effect on the protein. Inheritance was known for 136 of the 167 probands. In 124 of them (91%), mutations were found to be de novo, 6 (5%) were inherited and in one proband there was a suggestion of parental germinal mosaicism due to the occurrence of affected brothers, yet it could not be confirmed. For 31 patients, inheritance could not be determined due to the absence or not availability of parental DNA. We spilt patients into 2 groups according to whether they carried a 1) truncating or a 2) missense mutation. There were significant differences between the two groups and we could only remark the following observations: 1) there was a general predominance of female patients versus males (74% vs 26%); 2) there was a trend for a younger age at onset in patients with truncating versus missense mutations (5 m vs 6.5 m); 3) tonic clonic seizures were the most frequent seizure type at onset with a slight predominance in patients carrying truncating mutations (61.8% vs 52.8%); whereas focal seizures seemed to be more frequently observed at onset in patients with missense mutations (18.5 vs 7.8); 4) myoclonic seizures are infrequent seizure type at onset in both group of patients; 5) At last follow up, which occurred at a mean age of 13 years, most patients had ongoing seizures; 6) cognitive functions are impaired in both patients with a slight predominance of moderate-severe ID in patients carrying truncating mutations (54% vs 45%); 7) we observed recurrent missense and truncating mutations. The interesting point of inheritance remains unexplained. About 40% of probands had a family history of seizures yet over 90% of the mutations appeared to be de novo. Low-level parental mosaicism may have not been detected with the employed technique. We thus cannot exclude that some mutations are indeed inherited from an unaffected relative carrying low a level of mosaic mutation.

Concerning the study of biological and clinical factors in LKS/CSWSS, we modified the inclusion criteria to include, besides patients with Continuous Spikes and Waves during Sleep (CSWS), also patients within the “ESES spectrum”, such as Benign Epilepsy with Centrotemporal Spikes (BECTS), to be compared with patients with Childhood Absence Epilepsy (CAE), and to those with refractory focal epilepsy who are candidates for epilepsy surgery. The protocol included a comprehensive neuropsychological evaluation, assessment of declarative memory consolidation related to sleep and awake states, and high-density EEG in resting state conditions. Patients were to be assessed at baseline (T0), 1-3 months (T1), and 6 months (T2) after introduction of an anti-epileptic treatment (classical AED or steroids) aimed at decreasing the amount of interictal spikes during sleep, or absences (in patients with CAE), or after a resective surgical procedure. Seventy-one subjects (31 patients and 40 controls) were included. We assessed memory consolidation at baseline in 37 controls, 8 BECTS, 5 CSWS, 4 surgical cases, 5 CAE, and at T1 in 5 BECTS, 3 CSWS, 1 surgical case, and 4 CAE. Data showed that epilepsy syndromes with epileptiform activity during sleep, as seen in BECTS and CSWS, may suffer from the loss of the beneficial effect of sleep that healthy children instead showed for verbal learning. In our small sample, lack of sleep related benefit persisted after treatment. High density EEG data were available at T0 in 7 BECTS, 7 CSWS, 5 surgical cases, and 4 CAE, and at T1/T2 in 6 BECTS, 4 CSWS, 2 surgical cases, and 4 CAE. An evaluation of the available evidence failed to disclose any quantitative single EEG measure strongly correlated with cognitive outcome in the ESES spectrum. Thus a combination of measures or new quantitative methods were required. Accordingly, we used an innovative approach measuring the time-frequency entropy on the EEG signals in time windows without epileptiform discharges. This analysis revealed significant changes of the considered parameters between T0 and T1 for BECTS and CSWS patients, suggesting a possible relationship with treatment and evolution of the disease.

Studying biomarkers of disease outcome, we identified a variant in the 3’UTR of PAX6 (rs662702 T allele) known to disrupt microRNA-328 binding in BECTS. We investigated its association with CSWS and compared its incidence with that in Myoclonic Astatic Epilepsy (MAE) and Childhood and Juvenile Absence Epilepsy (CAE/JAE). The rs662702 T allele frequency was elevated (Odds ratio 3.31 p=8.3 x 10-4) in BECTS sample, whereas for all other epilepsies we found no significant differences. Then, using statistical and machine learning methods, we analysed the rs662702 SNP in a prediction model to discriminate between BECTS and other epilepsy types, incorporating sex and age of first seizure onset into the models being evaluated. We also searched for potential epigenetic modifications associated with LKS/CSWS in a monozygous twin pair discordant for LKS. Intra-twin differentially methylated probes (DMPs) >15% were identified and tested for reversal of DMP direction in the recovery sample of the affected twin. Fourteen differential methylation probes (DMPs) showed reversal of methylation change and were localised in brain expressed genes. In particular, among the most significant probes, one was localised in a synaptic gene; one in a gene involved in autophagy; one in a paralog of CDKL5; and one was enriched in inhibitory interneuron progenitor cells. Studying genetic cause of Atypical Benign Partial Epilepsy/ESES, we found that 3/18 (17%) patients carried pathogenic GRIN2A variants (all inherited); one carried a de novo pathogenic variant in ZYMND11, an established intellectual disability gene that is also a candidate for autism associated with seizures. We did not find SLC6A1 mutations.

Data obtained studying DS animal models (gene targeted mice), showing that the interaction between the epileptic history and the causative mutation can set the severity of the phenotype, confirmed the above results. In these models, also short (<1min long) repeated seizures induced by hyperthermia (a natural seizure trigger in DS patients and mice) can transform a very mild phenotype (GEFS+ knock-in mice show no spontaneous convulsive seizures and no cognitive/behavioural deficits) in a DS-like one (with frequent spontaneous convulsive seizures and severe cognitive/behavioural deficits, including hyperactivity, sociability and spatial and working memory deficiencies). Pathological remodelling in mice that experienced seizures was evidenced by exacerbated dentate gyrus excitability. Furthermore, when we induced similar short seizures with the convulsant flurothyl in both WT and mutant mice, only mutant mice developed an identical severe phenotype. In summary, both hyperthermic and flurothyl induced short seizures transform the mild/asymptomatic ‘GEFS+-like’ phenotype of Scn1a mutant mice into a severe DS-like phenotype, but the Scn1a mutation is necessary to make the brain susceptible to seizure-dependent long-term phenotypic alterations. These data demonstrate that severe phenotypes in this Scn1a mouse model result from the interaction between the mutation and seizures, thus requiring a two-hit context. In light of the current debate, we therefore provided evidence that DS should not be segregated either as an EE or a channelopathy. These results confer an important breakthrough as they encourage clinicians to keep trying to block childhood seizures in DS with innovative treatments to prevent long-term co-morbidities. Our work will also be instrumental to study other epilepsies in which SCN1A variants are risk factors and in which mutations and epilepsy interact to create pleiotropy.

In WP2, we also disclosed homeostatic responses and other factors that can modulate the phenotype in DS mice and the effects of NaV1.1 mutations in vitro. In particular, although loss of function NaV1.1 mutations cause hypoexcitability of GABAergic neurons, we found that in some of these neurons NaV1.1 loss of function can induce a remodelling that leads to their hyperexcitability, implementing a homeostatic response. These homeostatic responses are consistent with the unaltered network activity and firing of single GABAergic neurons we observed during spontaneous cortical dynamics in vivo in DS mouse models. We also identified molecular factors that can modulate the functional effects of NaV1.1 missense mutations in cellular models, in particular rescuing folding/trafficking defects of the protein. Studies in transfected cells have also evidenced a novel pathological mechanism for some DS mutations: negative dominance (i.e. the inhibition of the functions of co-expressed WT channel, which can increase phenotype severity). We have observed this effect for some missense NaV1.1 mutations and we are currently studying the mechanism of negative dominance, which can be exploited for stratifying patients and as a target of therapeutic approaches.

Clinical WP2 partners also analysed several features extracted from EEG signal analysis of Epilepsy-aphasia spectrum (EAS) patients, which can be used to monitor the effect of treatments aimed at preventing the deleterious effect of sleep disorganization on cognition and learning. In particular, we found novel time-frequency entropy analyses to objectively quantify the improvements of patients with continuous spike-waves during sleep after medication. Experimentally, WP2 partners used an approach similar to that used for DS models (integrating in vivo, ex vivo and in vitro experimental models), showing that both phenotypic and genetic mouse models exhibited hyperexcitability of cortical circuits, although phenotype features can be different. Moreover, we identified pro-pathologic responses in these mouse models as well as specific effects of mutations on NMDA receptors function and novel properties of NMDA calcium signals ex vivo and in vitro. Mutations of the GRIN2A gene, encoding the GluN2A subunit of NMDA receptors, have been identified in patients with EAS disorders, but little is known about the pathological mechanisms induced by the dysfunctions of the GluN2A subunit and implicated in disease outcome. We have shown that Grin2a KO mice replicate several anomalies found in patients with EAS disorders and identified specific pathological mechanisms. Indeed, we observed transient brain anomalies with longitudinal MR-DTI analysis in Grin2a KO mice; most brain microstructural abnormalities were detected in 1 month-old Grin2a KO mice, but not at earlier or later ages. Grin2a KO mice also showed altered slow-wave EEG activity induced by low dose isoflurane and reduced proportion of slow-wave sleep and sleep quality at one month of age, but these features were no longer altered in the same mice at two months of age. GluN2A KO mouse pups also exhibited altered vocal communication and increased neocortical bursting activity. Furthermore, Grin2a KO mice aged one/two months presented spontaneous spike-wave discharges, which occurred nearly exclusively during slow-wave sleep. These data indicate that the subset of GluN2A-containing NMDARs is involved in slow-wave activity at the early postnatal stages studied here, and that the period of postnatal brain development (P30), when several structural and functional anomalies peaked, might be critical for GluN2A-dependent, sleep-related physiological and pathological processes. Our data also suggest that transient structural alterations may be a biomarker of disease outcome in EAS disorders and that Grin2a KO mice may be used to develop new therapeutic strategies.

Overall, the WP2 study had two main strengths: the multicentre design of the clinical arm and the combined clinical and experimental approach. The ongoing analysis of the significant population of EE patients collected with strictly homogeneous criteria and studied longitudinally since the onset of the disease already shows that, in addition to the type of genetic mutation, the age at onset and the type and frequency of seizures contribute in determining the EE outcome. The identification of initial causes (e.g. genetic mutations) and of homeostatic and pro-pathologic responses induced by interactions with the internal and the external environment, within a framework in which the holistic understanding of detailed pathological mechanisms obtained comparing results of clinical and experimental studies is crucial, will be essential for developing precision medicine approaches, including the identification of predictive biomarkers for disease course and of targets for transformative therapies leading to disease modification and antiepileptogenesis. The combined clinical/experimental approach exploited in WP2 will provide an effective platform for future studies of pathological mechanisms and gene-environment interactions in other epilepsies.


WP3
High frequency oscillations (HFO) are oscillatory events between 80 and 500 hz, closely linked to epileptogenicity. Even if HFOs are promising biomarkers of epileptogenic tissue, their analysis is not a routine procedure due to methodological challenges. In the framework of WP3, the SMEI Micromed developed an advanced Hardware and Software prototype for HFO analysis, which received CE mark. In particular, the hardware system allows EEG recordings at very high sampling frequency, making it possible to analyse ripples and fast ripples. In addition, in order to implement data analysis, SMEI Micromed included an HFO manual marking feature and an automatic detection tool in the new software platform.
Using the Micromed prototype, we first developed a new protocol for scalp and intracranial HFO analysis. Then, we analysed scalp and intracranial HFO analysis in patients with drug-resistant seizures and malformations of cortical development or tumours, finding a correlation between scalp and intracranial ripples, seizure onset zone (SOZ) and postoperative outcome. Specifically, scalp ripple rate was significantly higher in irritative and SOZ compared to non-spiking and non-seizure onset areas. In addition, scalp ripple rates decreased after surgery in relation to outcome, irrespective from aetiology. Concerning intracranial HFOs, we found ripple rates to be statistically higher in the SOZ than in non-SOZ areas (p<0.05) in patients with FCD-related epilepsy. The incomplete removal of the ripple-containing areas correlated with an unsatisfying outcome. As a whole, our findings suggest that HFOs are a biomarker of epileptogenicity.

Concerning Ultra-High Field (UHF) 7T MR data analysis, we detected structural lesions in six out of 21 patients (29%) with intractable focal epilepsy, exhibiting clinical and EEG features clearly indicative of a unique seizure onset zone but with unrevealing conventional MR imaging. Four of the six patients with abnormal 7T underwent epilepsy surgery and histopathology revealed focal cortical dysplasia (FCD) in all. In 12 patients with MRI-visible FCD, UHF imaging confirmed the radiological signs of cortical dysplasia highlighted by 1.5/3T MR and revealed additional subtle findings in patients with FCD with balloon cells. We demonstrated that UHF MR at 7T can be a useful tool for revealing hidden epileptogenic lesions and has the potential to improve the assessment of FCD compared to lower fields MRI.

WP4
WP4 addressed four major objectives: 1) Translate epigenetic FCD signatures from surgical brain tissue to blood samples of the same individuals and develop molecular biomarker assays; 2) Integrate advanced electro-magnetic measures and new strategies for its analysis to improve non-invasive diagnostic tools and methods for the presurgical evaluation of epileptogenic networks in FCD patients; 3) Determine electro-magnetic and morphological measures for cognitive impairment in FCD patients; 4) Histopathologically validate molecular and morpho-functional biomarkers obtained, and make available high quality tissue biospecimens for translational research (European Epilepsy Brain Bank)

Histopathological analysis of surgically resected human tissue samples remains the gold standard for good medical practice. It also paves the way towards personalized medicine by helping to stratify disease-related patient cohorts for research of disease- and tissue-specific molecular targets. In the arena of epilepsy surgery, microscopic review will not only explain any suspected structural brain lesion but also validates the clinical hypothesis of the epileptogenic focus on which the patient was consented for surgical treatment. The anatomo-pathological spectrum of brain lesions in epilepsy surgery is, however, very broad and not yet fully described or classified. As a consequence, prediction of postsurgical outcome from currently available clinico-pathologic classification systems with respect to complete seizure freedom and tapering of antiepileptic drugs awaits clarification.

Within WP4, we developed international guidelines for the histopathologic work-up of epilepsy surgery tissues in cooperation with the Task Force of Neuropathology of the International League against Epilepsy (ILAE). The consensus group agreed on standardized operational procedures for inspection, distribution and processing of epileptogenic brain tissue in order to reduce sampling errors for microscopy review, ensure the best possible histological assessment, and support research activities and brain banking initiatives.

We analysed our unique data collection of more than 10.000 patients obtained from 36 European epilepsy centres across 12 European countries, i.e. the European Epilepsy Brain Bank. A total of 9523 patients met the inclusion criteria with completion of a minimal dataset. Histopathological diagnoses were obtained from microscopic review in local hospitals (41%) or at the German neuropathology reference centre for epilepsy surgery at P4 UKER (59%) in 6900 adults and 2623 children (< 18 years of age at surgery). A minimal dataset was collected for all patients, which included side and location of the lesion, sex, age at epilepsy onset and surgery. Postsurgical seizure outcome was available after 12 months in 7286 patients. We collected 36 histopathological diagnosis, classifiable into 7 major histopathological disease categories. The 10 most common histopathological diagnoses explained 86.6 % of the entire series. 75.9% of patients had seizure onset in childhood, whereas 72.5% were operated as adults. Whereas 44.5% of our adult cohort suffered from temporal lobe epilepsy with hippocampal sclerosis (HS), malformations of cortical development (MCD) were most common in children (39.3%), mainly FCD type II in the frontal lobe (17%). The mean duration of epilepsy was 16 years, and did not change during the observation period of 25 years. One year after surgery, complete seizure freedom (Engel 1A/ILAE 1) was achieved in 66.4% of children and 58.6% of adults. Multivariate analysis identified histopathological diagnosis, age at surgery and disease duration as significant predictors of outcome (Blumcke et al., 2017).

During the entire funding period of DESIRE, we organized a total of nine 4-day summer schools for neuropathology and epilepsy surgery (INES) to disseminate knowledge about the most frequent causes of difficult-to-treat focal epilepsies, consensus classification systems and protocols for the histopathological work-up, and to assure use of standardized reporting. We invited distinguished tutors to train participants in small groups. All information was made available also by a special course booklet summarizing case presentation, protocols for neuropathological work-up in epilepsy surgery, as well as several histopathological review articles introducing all major entities of epileptogenic brain lesions. We also offered morning lectures by distinguished clinical colleagues covering the broad spectrum of inter-disciplinary epileptology, i.e. neuroimaging, epilepsy surgery in adults and children, principles in pharmacological treatment, human genetics and tissue-related molecular research. More than 400 participants from more than 30 nations from all over the world participated in this unique training program.

Essentially, we need better strategies for clinical diagnosis and treatment decision in patients with difficult-to-treat focal epilepsies, in particular for those with a negative or non-informative MRI. Such measures will support the clinician in: 1) the identification of the epileptogenic zone (EZ), 2) planning the stereo-EEG (SEEG) strategy, namely in guiding electrode placement to compensate for the limited spatial coverage of invasive recordings, 3) defining/modifying patient management plans (e.g. addition of neuroimaging investigations, exclusion from invasive investigation or from surgery). To address this challenge, we integrated advanced electro-magnetic measures with currently available presurgical monitoring to improve non-invasive diagnostic tools. The task was carried out along two lines. The first consisted in the localization of the possible EZ through Non-Invasive Functional Neuro-imaging (NIFN: MEG, HDEEG, EEG-fMRI, PET) with a focus on the integration of the results and their evaluation in terms of concordance with SEEG and surgical outcome. The second approach was based on connectivity and network analysis applied to SEEG data. To the best of our knowledge, this attempt is the first to combine four different non-invasive procedures to be compared with invasive recordings for the identification of the EZ in candidates to epilepsy surgery.

Among the different NIFN, MEG appeared to be the most sensitive, reaching the value of 100% for both sensitivity and negative predictive value (NPV) in the multilobar (ML) subgroup of 20 patients, while EEG-fMRI appeared to be most specific. In the subgroup of patients with multifocal epilepsy who became seizure-free after surgery (100% sensitivity), MEG results were always concordant with SEEG data. It means that, within the total group of patients admitted to surgery, MEG was able to identify the “TRUE positive” subjects in Engel Class I. This implies that in our ML subgroup, a discordance between MEG and SEEG always predicted a negative prognosis after surgery (100% NPV). Conversely, EEG-fMRI results largely discorded with those of SEEG in the subgroup of patients in whom seizures persisted after surgery (83.3% specificity), meaning that EEG-fMRI correctly identified most of the “TRUE negative patients” in Engel Class II-IV after surgery. In conclusion, our data showed that convergent concordance among electro-magnetic, metabolic and hemodynamic data with SEEG invariably predict a positive outcome after surgery.

The cellular origin of abnormal neurophysiology markers remains, however, yet to be clarified. In WP4, we aimed therefore to also develop a semi-automated algorithm to co-register histology with invasive EEG recordings in patients submitted to epilepsy surgery. We included into the study eight patients with FCD Type II and identified a total of 39 intracerebral electrode contacts after alignment of anatomically well preserved surgical tissue samples with CT and MRI scans. We immunohistochemically quantified key cell populations (e.g. dysmorphic neurons, balloon cells, as well as all neurons and astrocytes) in areas of 1-5 mm diameter from the identified electrode contact. Neurophysiology markers included seizure onset, spike distribution, and oscillatory activity in delta, theta, gamma and high-frequency oscillation bands, as well as theta-gamma phase-amplitude coupling. Correlations between histopathology measures and neurophysiology markers as well as principle component analysis provided compelling evidence for a contribution of dysmorphic neurons to seizure onset, interictal spikes, high-frequency oscillations, fast gamma activity, and phase-amplitude coupling. In contrast, areas with increased balloon cell densities were electrically less active across all frequency bands. Our protocol provides a new powerful tool to address the cellular source of abnormal neurophysiology signals and leverage current and novel biomarkers for the localization of epileptic activity in the human brain.

We also developed dynamical EEG source imaging algorithms based on state-space modelling and Kalman filtering for accurate localization of the epileptogenic zone (EZ). Firstly, we studied the influence of spike averaging on the performance of the spatio-temporal Kalman filter (STKF). We used 128-electrode EEG data that were simulated by the "Epileptogenic Systems: Signals and Models" group at the University of Rennes in France using neuronal population models. We used this kind of physiological modelling for epileptic activity typically generated by FCD. In the pre-processing stage, we used the 128-electrode EEG data to create derivative datasets using standard electrode montages with 64, 32, 19, and 9 electrodes. Additionally, we obtained an averaged spike by averaging 55 spikes from each dataset and visually chose a single spike for the purpose of comparison. EEG source imaging via STKF showed that spike averaging improved the accuracy of STKF. The decrease in the number of electrodes made the sources less focal and shifted their locations away from the target areas. Spike averaging, however, made the STKF more robust to the aforementioned influence of the decrease in the number of electrodes. We concluded that averaging of a moderately large number of spikes (tens of spikes) does not suppress the interesting dynamics in the data and may still be used for EEG source imaging via STKF.

Secondly, we further developed and tested new generalized versions of the STKF that adapt the parameters in time or in space. We analysed 256-electrode EEG recordings from an adult female epilepsy patient with hippocampal sclerosis, who became seizure free after epilepsy surgery in the right temporal lobe. We used STKF and a regional variant of the STKF, the RSTKF, which allows for region specific model parameters to model regional brain dynamics. The EEG data were made available by the Epilepsy Center in Freiburg. We then applied STKF with state-space generalized autoregressive modelling with conditional heteroscedasticity (STKF-ssGARCH). This is a variant of the STKF that allows for temporal adaptation of model parameters. We applied, for the first time, STKF-ssGARCHon presurgical EEG data of epilepsy patients. Additionally, we implemented the array square-root implementation of STKF-ssGARCH (ASR-STKF-ssGARCH) for improved numerical stability. With respect to accuracy and spatial resolution, the regional STKF and the array square root STKF-ssGARCH outperformed STKF and STKF-ssGARCH. In addition to that, ASR-STKF-ssGARCH improved the accuracy of STKF ssGARCH as well as the reconstruction of the time courses of current dipoles in the brain. We concluded that RSTKF and ASR-STKF-ssGARCH are promising and powerful source imaging algorithms for presurgical evaluation of epilepsy patients, especially when the epileptogenic zone is represented by a deep and small lesion.

Clinical diagnosis and clinico-pathological classification of FCD subtypes still remain challenging issues in daily practice. We attempted to discover molecular signatures that can be translated into clinically useful classification tools. One major target was whole genome DNA methylation profiling from surgically resected brain tissue and matched blood samples. DNA methylomes were generated from massive parallel sequencing in 15 surgical FCD specimens following microscopic review according to ILAE guidelines (see above), matched with 5 epilepsy and 6 non-epilepsy controls. Unsupervised hierarchical cluster analysis discriminated three FCD subtypes in our cohort with distinguishing methylation signatures in promoter regions and gene bodies. Review of clinical histories and microscopic features assigned molecular subtypes to previously proposed FCD phenotypes, e.g. FCD Ia, IIa and IIb. Deep gene panel sequencing analysis detected 2 mutations in genes of the mTOR pathway in only the FCD IIa subtype. Our studies extend the evidence for disease-specific methylation signatures towards focal epilepsies, and endorse an integrated clinico-pathological and genetic classification system of FCD subtypes.

We established a low input miRNA sequencing protocol that made it possible to detect miRNA expression patterns of microvesicles, platelets, other blood compartments, neuronal and non-neuronal cells. We conducted principal component analysis to evaluate similarity of detected miRNA expression patterns. There were significant differences in miRNA profiles between brain and blood. Also peripheral mononuclear blood cells seem to have a different miRNA profile from platelets and platelet-derived microvesicles. However, pairwise comparisons between all sample groups identified distinguishing miRNAs for all sample types and cell fractions is still under study. Data analysis is further ongoing to explore commonalities between brain and blood miRNA profiles to identify potential peripheral molecular biomarkers of epilepsy and associated pathologies. The low input protocols for miRNA sequencing were of particular relevance also for the SME CeGaT and has been readily implemented in their product portfolio.
Integration of genotype-phenotype analysis will inform the current ILAE consensus classification system from 2011. A revision of this classification system is currently in preparation. Any new genetic information, in particular a genomic DNA methylation classifier, will directly translate into such a new, integrated clinic-pathologic and molecular classification system.
WP5
WP5 was targeted at improving epilepsy treatment in children with FCD ILAE Type II. This involved a clinical trial assessing whether pre-treatment of children with epilepsy undergoing surgical resection of FCD improved seizure free outcome. We finalised a protocol, registered the trial on the NCT no 02261753, and achieved ethical approval in the UK. We identified twenty sites eligible to participate and planned registration and initiation of sites in remaining countries. We designed, developed, tested and validated in OpenClinica the online database through which secure electronic data entry was to take place at sites. In addition, we created an array of training material ranging from user guides to demonstration DVD to assist the on-site personnel and prevent potential mistakes. A face-to-face training session took place in order to pilot the training material. The Trial Management Group (TMG), responsible for the day-to-day running and management of the trial, oversaw and approved the development and amendments of all trial documentation and processes to date. By February 2016, we finalised the development of the monitoring plan. The appropriate trial documentation underwent a translation process in four languages including Italian, German, French and Czech. We undertook this necessary step to allow for separate submissions to Research Ethics Committees (RECs) in participating countries/regions outside the UK. Due to delays and complications imposed by the translation process and subsequent changes to trial documentation for each participating country/region outside the UK, as recommended by local RECs, we did not achieve local approval for all sites. Out of the 14 participating sites, 5 opened to recruitment and an additional 4 received on-site training. The first patient was recruited to the EDIBLE trial on 10th August 2016 by Birmingham Children’s Hospital in the UK. Monitoring of activities continued regarding set-up progress, disease prevalence and recruitment rates informing adjustment to the trial strategy and corrective actions. Although we undertook work to open further sites across Europe, there were continued delays with contracting and approval processes. In opened sites it was also clear that changes in process (reduced waiting times for surgery) and a smaller pool of patients (reduced numbers of FCD II in sites, children already trialled on a ketogenic diet) meant numbers required for the primary outcome would not be achieved. Therefore, the Data Safety Monitoring Committee decided that the trial should be closed. We completed closure of all sites on December 2017. We made a plan to achieve neuropathology results, by determining changes induced by the ketogenic diet by utilising samples from children undergoing surgery across sites, with documentation of whether they have been trialled at any time on the ketogenic diet in the past. In total, we have been able to submit 4 samples for methylation studies to P4 UKER. A manuscript has been drafted with regard to the study protocol, and lessons learnt, and submitted for publication.

WP6
Objectives of WP6 were to reduce or prevent epileptogenesis by genetically engineering neurons to express innovative molecular tools or by reinforcing inhibition by implanting inhibitory neurons derived from skin fibroblasts. We expect that the results of these endeavours will allow, in the future, mitigating seizure severity and help preventing epileptogenesis or delaying seizure onset. The main results we obtained over the project duration include: 1) development of engineered nanoparticles that safely reach the brain to deliver therapeutic molecules to the epileptic tissue; 2) modulating of the activity of epileptogenic networks by ‘switching off’ hyperactive neurons through a variety of molecular approaches; 3) the selective elimination of malfunctioning neurons, and the prevention of the onset of epileptic manifestations by inhibiting the activity of specific neurons in models of neurodevelopmental epilepsies; 4) the control of gene transcription in the epileptic environment through optogenetics and pharmacological approaches; 5) the development of novel cell therapy approaches through reprogramming of mouse and human fibroblasts into functional inhibitory neurons, and their genetic modification; 6) implementation of more efficient and safe viral vector for gene delivery into the brain.

To develop targeted drug delivery tools using nanoparticles coupled with engineered antibody fragments, we focused on the Tuberous Sclerosis Complex (TSC). TSC is a genetic disease characterized by the development of benign tumours in many organs, including multiple brain regions, which causes epilepsy, intellectual disability and autism. One peculiar feature of TSC is the presence, in the brain tissue, of giant neurons characterized by an abnormally high expression of a specific protein, named GluR4, which plays an important role in controlling neuronal function. Currently, several obstacles are encountered when trying to develop treatments for TSC. Firstly, it is very difficult to access the affected areas of the brain. Secondly, and more generally, the blood-brain-barrier (BBB), which separates the circulating blood from the brain, represents an insurmountable obstacle to the passage of drugs and other substances such as therapeutic or diagnostic tools. In recent decades, the use of nanoparticles (NPs) has proven effective in transporting drugs, including peptides and macromolecules, through the BBB and into the brain. P18 DI.V.A.L. has manufactured novel polymeric NPs conjugated to a specific peptide (g7), which cross the BBB and are selectively retained in the brain of healthy mice. This type of polymer is highly biodegradable and is approved by the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) for human applications. Further, we developed a monoclonal antibody that selectively targets dysplastic cells in specimens from TSC patients by recognizing an overexpressed membrane glutamate receptor (GluR4) and coupled NPs coated with the g7-peptide with this antibody. Such ‘double functionalized’ NPs are currently under evaluation in a TSC mouse model, and represent a promising first step toward the implementation of a novel, non-invasive therapeutic and diagnostic (theranostic) tool for TSC.

Brain activity in epilepsy patients with developmental brain disorders is often uncontrolled. Thus, a possible approach to ameliorate seizures is to engineer molecular tools that allow controlling neuronal activity. We have tackled this issue following two lines of research. First, we engineered and tested specific molecules that, once delivered to specific neuronal populations, ‘switch off’ by RNA interference the expression of the main cellular components (voltage-dependent sodium channels) causing neuronal hyperexcitability. We demonstrated the efficacy of such probes both in primary neurons and upon delivery to specific regions of the brain of healthy rodents. This therefore represents a suitable approach to dampen neuronal hyperactivity that is associated with epileptic seizures. As a second approach, we exploited the fact that epileptic seizures are associated with intra/extracellular acidification that, in turn, exacerbates the epileptic phenotype. To break this vicious circle, we developed a novel optogenetic/chemogenetic probe made of a molecule able to sense the pH drop coupled with an effector protein (an inhibitory opsin), which by hyperpolarizing the neurons effectively counteracts network hyperactivity. We successfully tested both probes in vitro, demonstrating that they represent a valid tool to be implemented for pre-clinical and clinical studies.

The rationale for selectively inducing the degeneration of epileptogenic/malformed neurons is to kill malfunctioning neurons in the epileptic brain, while leaving healthy neurons unaffected. To accomplish this, it was first necessary to identify cellular markers, or ‘tags’, which are unambiguously associated to epileptic neurons. We generated and characterized a new rat model of ‘bilateral double cortex’ (Dcx), characterized by neurodevelopmental defects of the brain cortex and early onset spontaneous epilepsy. Importantly, epileptic manifestations in this experimental model are, as in human disease, age-dependent, thus providing a time window for testing potential interventions before the epilepsy onset. The structural and electrographic features of Dcx rats were fully described and the model will help future therapeutic investigations. Using this model, we identified a specific protein (Suppression of tumorigenicity 18, St18) that was mostly associated with malfunctioning cells and used it as a genetic tool to specifically express, in these neurons, the receptor for the lethal Diphteria toxin. We subsequently set up an experimental system to selectively kill St18-expressing cells by injecting the toxin into the brain.

Using the experimental model described above (Dcx rats), we demonstrated that the epileptogenic network involves not only the malfunctioning neurons, but also the surrounding and apparently healthy brain area. Moreover, we demonstrated that neurons located into the seemingly healthy tissue display altered activity patterns. We thoroughly investigated the functionality of the affected brain areas, and described the alterations of cortical circuits that underlie epileptogenesis, and the region (‘epileptogenic focus’) that initiates the pathogenic activity. Finally, to assess the therapeutic potential of our studies, we demonstrated that selective silencing of malfunctioning neurons (e.g. by overexpressing the potassium channel Kir2.1) was not sufficient to reduce seizures, while the silencing of both malfunctioning and healthy neurons significantly reduced epileptogenesis. These data highlight the contribution of apparently healthy tissue to the pathology process and provide new vistas for preventing/cure epileptogenesis in cortical malformation disorders.

Epilepsy is known to cause, in the long term, profound alterations of neuronal physiology, including changes in gene transcription. Thus, controlling the activity of specific transcription factors could counteract the long-term changes in gene expression that take place in a pathological environment. We tackled this problem following two independent approaches.
We engineered synthetic probes based on a light-sensitive protein (LOV) derived from the oat plant, to interfere with the activity of the transcriptional repressor REST, whose expression is abnormally increased in the epileptic brain. In primary neuronal cultures, the activation of the LOV-based probe by light could efficiently modulate neuronal transcription, indicating that it was effective in inhibiting REST activity. In addition, the optogenetic suppression of REST was associated with changes in neural activity. To evaluate the impact of REST modulation on epileptogenesis in vivo, we bilaterally injected the probe into the brain of healthy mice. Remarkably, the expression of REST target genes was increased in the brain of injected mice. When compared to control animals, mice infected with the probe also showed a reduced overall seizure propensity.

Through genetic and functional studies, we identified three genes (SCN1A, STXBP1, and SLC2A1) whose mutations cause the complete loss of the physiological protein function, leading to epileptic encephalopathy (EE). We subsequently screened molecules from libraries of pharmacologically active compounds for their ability to enhance the expression of the normal alleles of the above genes through an optimized, cell-based high-throughput screening assay. We obtained the best results for the SLC2A1 gene. We started testing 774 FDA-approved drugs and, through successive refinements, we selected 3 activators and one inhibitor of SLC2A1 gene expression, which showed a dose-response effect at concentrations that are compatible with a pharmacological treatment in patients. These active compounds are currently being evaluated through further functional assays, and their efficacy tested in human iPSC-derived neurons. These molecules may represent the basis for future development of novel pharmacological treatments for epileptic encephalopathies.

A highly innovative approach named ‘genetic reprogramming’ has been recently devised to generate functional neurons directly from primary mouse and human skin cells. We exploited this technology for the development of novel cell and gene therapy approaches for the treatment of Dravet Syndrome (DS) and potentially other epileptic syndromes in which inhibitory neurons degenerate or are not functional. Our strategy is based on the manipulation of the activity of inhibitory GABAergic interneurons, which behave under normal conditions as a ‘brake’ for the excitability of the network, and whose activity is altered in epilepsy syndromes. To achieve our aim, we followed two parallel approaches. First, we optimized the experimental procedures required to obtain functional inhibitory neurons from mouse and human skin cells. Such reprogrammed neurons were active in vitro and, when implanted into the brain of adult mice, displayed the ability to migrate out of the injection site and functionally integrate in the complex brain circuits. In parallel, we developed a gene therapy approach that exploits a synthetic protein called ‘CRISPR/Cas9’, which - when appropriately tuned - is able to increase the expression of specific target genes in inhibitory neurons. We optimized this system to increase the expression of one particular gene (Scn1a), whose expression is reduced in DS. Our engineered CRISPR/Cas9 probe could efficiently increase gene expression in cells and neurons in vitro. To test the therapeutic potential of this approach, we delivered the probe to the brain of transgenic mice that are a model for DS, and remarkably, febrile seizures were significantly attenuated in animals that were administered the active probes, compared to control animals. These results thus provide great expectations for circumventing the impact of gene mutations on neuronal networks activity and associated epileptic syndromes.

All above studies benefit of the implementation of novel, more efficient viral vectors to deliver the various probes to the chosen brain areas. One of the most promising viruses for therapeutic gene transfer is called SV40 for being non-immunogenic so that vector-transduced cells are not recognized and cleared by the immune system. Unfortunately, we showed that SV40 viruses are poorly neurotropic and do not transduce neural tissue with high efficiency. For this reason, in order to carry the vectors engineered within the WP, we implemented throughout the project various types of adeno-associated viral vectors that were endowed with different promoter regions for targeting specific neuronal subpopulations.

WP7
An extensive and wide reaching number of dissemination activities has been performed by the Consortium, at local, national and international levels, in order to promote the visibility of DESIRE to the scientific and industrial community and to raise awareness among relevant lay and patients’ organizations, general public, public health organizations, health policy makers and opinion leaders on the socio-economic benefits that has been achieved by means of DESIRE.
For maximum visibility, a clear and precise identity, linked with a graphically coherent and consistent representation was created. The dissemination toolkit consisted of the project logo, the project website and a project leaflet / brochure.
The logo provided the DESIRE project with a “corporate” image, thus allowing for more visibility and homogeneity, especially when interacting with external parties. The logo has been used extensively and is now well recognized among the wider scientific and lay epilepsy community.
The project website was created and maintained by P17 UOM, and was regularly updated throughout the project duration. The structure and layout contains key and easily accessible information about the project, its objectives, and organization, consortium partners, news and events, dissemination activities and results such as publications, It is also linked to media and social networking channels. The website acknowledges FP7 support and incorporates the EU flag and FP7 logo. Statistics about registered user access to the website are monitored. The website is available at the following address: www.epilepsydesireproject.eu. The website is updated regularly with relevant news and posts. It also contains links to DESIRE related publications, presentations at conferences and also links to the Facebook page. A page ‘For patients’ has also been included to ensure easy to read information about DESIRE for persons with epilepsy and their relatives
Press releases were issued at the beginning of the project, to announce the awarding of the Grant Agreement soon after the kick off meeting and along the project duration, mostly in conjunction with the annual meetings.
A leaflet, a poster and a banner were created by P17 UOM. Hardcopies were printed for distribution at conferences and other events. Relevant patients‘ organisations and other international organizations also received hard and soft copies for internal dissemination.
A regular newsletter (or e-newsletter) was prepared and published in the website in order to communicate the main project developments, such as the achievement of a project results, the organization of a project meeting, etc. In total seven (7) newsletters were published in the lifetime of the project. A mailing list of patients‘ organisations and other stakeholders has been also compiled in order to ensure distribution.
A DESIRE Facebook® and Twitter ® accounts were also created by P17 UOM and linked to the website in order to ensure DESIRE visibility to a wide epilepsy community.
Dissemination of scientific results from the project was done mainly through the submission of articles for publication in high impact peer reviewed journals or in the proceedings of the national/international scientific meetings. To date more than 170 scientific articles were published acknowledging DESIRE.
Moreover, the DESIRE results have been showcased during 320 national and international meetings by the consortium members.
During the project duration, DESIRE participated to several important concertation initiatives:
- during 11th European Congress on Epileptology, held in Stockholm (Sweden) in 2014, a Forum entitled “DESIRE and euroEPINOMICS: Large scale EU research networks to unveil the genetic and pathogenetic mechanisms of epileptic encephalopathies” was organized.
- Moreover, during the same event, the “Epilepsy Advocacy Europe Symposium. A political agenda for supporting epilepsy care and research in Europe” was also organized, during which a summary of the epilepsy projects financed by FP7 was presented by Hannah Cock (UK).
- At the beginning of the third year, DESIRE participated in a joint meeting of epilepsy funded consortia. The meeting “Epilepsy Research in the EU: state of the art and opportunities for the future” has been organized in Ferrara (Italy) in 2015. The meeting was an opportunity to build a framework for epilepsy research in Europe, to identify opportunities for the future supporting interactions and cooperation between projects, and to approach specific, and yet unresolved, aspects of epilepsy.
- In 2018, DESIRE joined seven large EU-funded projects to pave the way for future epilepsy research on 23 May 2018, in Brussels, Belgium. This event called EpiXchange, was designed to gather a critical mass of epilepsy researchers and showcase the latest progress in research aiming to improve the way epilepsy is diagnosed and treated and thereby increase the quality of life of affected people. The other projects were EPISTOP, EpiTarget, EpiPGX, EpimRNA, Epixchange and a European Reference Network (ERN) funded by DG Sanco (EpiCare).

A smart phone/tablet app for children was developed. This app is educational and enhances the understanding of not only children with EED about their condition but can also be used by other children and also in educational settings for teachers and parents to better impart knowledge about EED to siblings, schools and other educational institutions in a playful yet pedagogical manner so as to facilitate inclusion, better acceptability, improve awareness and reduce stigma towards EED. EPIDES: An interactive app about epilepsy. EPIDES is a mobile app game aimed at the general public, especially children and children with epilepsy. It provides a fun way of learning more about epilepsy through as simple quiz and child user-friendly interface, with several colourful graphics and images which make it attractive to use. It can be downloaded free from the Google playstore and will remain available after the end of the project. It consists of simple questions to be answered on epilepsy, while to player throws a dice (metaphorically by shaking the phone) moving through the countries which were involved in DESIRE project.

Dissemination activities were also performed towards patients and general public: a special webpage about DESIRE was developed for patients http://epilepsydesireproject.eu/?page_id=66 with extensive information about the project and a special link to the clinical trial EDIBLE being carried out in WP5 http://epilepsydesireproject.eu/?p=1073. Furthermore, several lectures, presentations and radio interviews were undertaken towards patients and general public.
The Final DESIRE Conference took place in Florence (Italy) on 12-13 September 2018.
The PCC members decided to organise the first day as a high level neuroscience meeting, during which the most important results achieved by the project has been presented. Moreover, they decided to invite the Scientific Advisory Committee members (Prof. Jeffrey Noebels, Prof. Solomon Moshe and Prof. Emilio Merlo Pich) to organise an introductory session dealing with the translational aspects of epilepsy, to offer the DESIRE consortium (including young researchers) the possibility to benefit of their recognised international contribution to basic and clinical research in this field.

WP8
The DESIRE project was based on a coordination structure and decision making mechanisms set up to cope with the complexity of the project. The consortium placed particular attention in designing its management structure in order to have strategies for keeping control on the activities and on the potential problems that may arise.
P1 UNIFI has acted as project coordinator and was responsible for the day-to-day, administrative, financial and overall aspects of project management in line with Art. II.2.3 of the GA. The Project Coordination Committee (PCC), composed of the leaders of the different Workpackages, was in charge of the strategic direction of the project. The General Assembly (GA), consisting of one representative from each beneficiary, was the decision-making body of the consortium.
At the beginning of the project, two external boards have been also created - the Ethical Advisory Board and the Scientific Advisory Committee – respectively responsible for the monitoring of ethical issues and of the scientific activities. A non-disclosure agreement have been set up for external members involved in the management boards to protect the project foreground.
The DESIRE consortium organized a kick off meeting at the start of the project, four annual meetings at the end of each reporting period, as well as a Final Conference at the end of the project. The Scientific Advisory Committee members participated to all the annual meetings, as well as to the Final Conference, in order to evaluate the progress of the project and to provide scientific advice.
A mailing list have been created and maintained throughout the project, to ensure the communication among the beneficiaries.
Along the project duration, Workpackage 8 activities were also related to the overall financial and administrative management, including the management of the EC contribution, the distribution of the EC payments, the monitoring of the schedule (including the achievement of deliverables and milestones), and the set up of a framework for the preparation and submission for the periodic reports.
The project has faced three amendments to the Grant Agreement, mostly due to changes to the consortium (termination of beneficiaries and/or inclusion of new beneficiaries) and to revision of the project budget (transfer among the beneficiaries). Amendment no. 3 in particular, was requested to change the Grant Agreement after the notification of insolvency procedure of one of the beneficiary and due to the termination of the EDIBLE trial (WP5) by the Independent Safety Monitoring Board.
From an ethical perspective, WP activities included also the collection of the ethical committee approvals from the beneficiary.

Potential Impact:
This project has gathered a multidisciplinary team of European research groups bridging basic with clinical sciences, which are leading international players in their discipline.
Participating groups have established a strong track record in the multidisciplinary study of MCD, FCD and EE and of genetic developmental brain disorders in general. They have access to advanced experimental models, large patient cohorts, the European Epilepsy Brain Bank, as well as large DNA and cell line collections. They have extensive experience in participation in, and co-ordination of, European projects. This consortium has produced research outputs that are expected to have a major impact in diagnosis, genetic counselling, treatment and health care of children with EDD and their families.
The active involvement of patient’s and lay organizations through the dissemination activities is ensuring the rapid implementation of results and immediate benefit for families.
DESIRE scientific outputs will improve the understanding of the etiology and mechanisms of epilepsy and epileptiform disorders. By deciphering PM in EDD and developing mechanism-related, and advanced therapeutic strategies, DESIRE has discovered novel genes and related molecular pathways that are involved in epilepsy and other developmental brain disorders.
Identification of novel neurophysiological and neuropsychological biomarkers will lead to earlier diagnosis and treatment, making it possible to prevent or minimize epilepsy related cognitive decline.
With most advanced technologies, DESIRE has successfully proposed experimental strategies that may rapidly translate into clinical practice by using specific pathophysiological individualized approaches.
The participation of five SMEs operating in the areas of technology and molecular medicine is expected to translate soon the new technologies into larger scale diagnostic and therapeutic tools and to open new perspectives towards the participation of the industry in an integrated and rational drug discovery process heading towards more targeted treatments for epilepsy.

List of Websites:
http://www.epilepsydesireproject.eu/