TBC1D24 genotype–phenotype correlation

Objective: To evaluate the phenotypic spectrum associated with mutations in TBC1D24. Methods: We acquired new clinical, EEG, and neuroimaging data of 11 previously unreported and 37 published patients. TBC1D24 mutations, identified through various sequencing methods, can be found online (http://lovd.nl/TBC1D24). Results: Forty-eight patients were included (28 men, 20 women, average age 21 years) from 30 independent families. Eighteen patients (38%) had myoclonic epilepsies. The other patients carried diagnoses of focal (25%), multifocal (2%), generalized (4%), and unclassified epilepsy (6%), and early-onset epileptic encephalopathy (25%). Most patients had drug-resistant epilepsy. We detail EEG, neuroimaging, developmental, and cognitive features, treatment responsiveness, and physical examination. In silico evaluation revealed 7 different highly conserved motifs, with the most common pathogenic mutation located in the first. Neuronal outgrowth assays showed that some TBC1D24 mutations, associated with the most severe TBC1D24-associated disorders, are not necessarily the most disruptive to this gene function. Conclusions: TBC1D24-related epilepsy syndromes show marked phenotypic pleiotropy, with multisystem involvement and severity spectrum ranging from isolated deafness (not studied here), benign myoclonic epilepsy restricted to childhood with complete seizure control and normal intellect, to early-onset epileptic encephalopathy with severe developmental delay and early death. There is no distinct correlation with mutation type or location yet, but patterns are emerging. Given the phenotypic breadth observed, TBC1D24 mutation screening is indicated in a wide variety of epilepsies. A TBC1D24 consortium was formed to develop further research on this gene and its associated phenotypes.

reported. We evaluated the types of epilepsy seen with a wide mutational spectrum in TBC1D24, using new data from 11 previously unreported and 37 published patients. METHODS Standard protocol approvals, registrations, and patient consents. This study was approved by the relevant institutional ethics committees or review boards. Parental (or legal guardian) written informed consent was obtained for all affected children and adults with intellectual disability, or existing published data were collated. Authorization has been obtained for disclosure of videos (videos 1-5 on the Neurology ® Web site at Neurology.org).
Participants. We collected data from a cohort of new patients and contacted physicians to seek additional information about published patients by using a standardized questionnaire. Available neuroimaging and EEG recordings were evaluated. Individuals were included if they had a confirmed TBC1D24 mutation and had epileptic seizures within their phenotype. We report 2 additional patients (31,32) where a clear association of phenotype with changes in TBC1D24 could not be established: therefore, they were not included in the final analyses. In patient 31, 2 single nucleotide polymorphisms in the maternal allele were found through a next-generation sequencing panel; in patient 32, array comparative genomic hybridization showed a 16p13.3 duplication including TBC1D24.
Procedures. TBC1D24 variants were identified through various methods, detailed in supplemental data S1. Conserved motifs were detected in MEME suite 21 through discriminative motif discovery. Bioinformatics and in vitro modeling methods are described in supplemental data S1.

members of a large
Arab-Israeli family with multiple consanguineous unions, 6 8 members of a large Italian family, 1,7 and 3 members of a large Turkish family (all patients born to consanguineous parents). 9,10 Six other sporadic patients, including one previously described, 23 have consanguineous parents; 15 patients (31%) are isolated, from nonconsanguineous parents.
Seizures and treatment responsiveness. The types of seizures and epilepsies were diverse. Seizure types included infantile spasms and febrile convulsive, myoclonic, clonic, tonic, absence, tonic-clonic with or without apparent focal onset, and focal seizures with retained or impaired awareness. Myoclonic or clonic seizures were the most frequent seizure types (29/48, 60%), often unresponsive to medication. Myoclonic seizures were segmental (often involving eyelids, perioral region, or other facial parts) or generalized, with initially no loss of consciousness, but sometimes evolving into tonic-clonic seizures. Various other features of myoclonic seizures were described; they could be unilateral or bilateral, migrating, alternating, rhythmic, or pseudorhythmic, occurring both at rest and on maintaining posture. They often occurred in clusters, which could be very prolonged, lasting several days. In some patients, they were triggered by fatigue, drowsiness, intense and persistent stimulation (acoustic stimuli or variations in light intensity), repetitive movements, feeding, febrile episodes, constipation, or delayed medication. Eighteen patients had myoclonic epilepsy (including infantile myoclonic and progressive myoclonic epilepsies). Semiologic features of 5 patients, 4 with myoclonic epilepsy (4, 23a, 23b, 24) and 1 with familial epilepsy of infancy with migrating focal seizures (EIMFS) (6b), are shown in videos 1-5. The other patients had focal, multifocal, generalized, or unclassified epilepsy, or early-onset epileptic encephalopathy (including EIMFS). There was no marked variability of epilepsy phenotype in affected siblings.
The average age at seizure onset was 7 months (range from within 1 hour after birth to 8 years; SD 15 months). Thirty-eight (79%) individuals had had status epilepticus, either convulsive or nonconvulsive, or prolonged seizures (.5 minutes). In 19 patients, seizures or status episodes were precipitated by fever or infections.
In 30 patients, epilepsy was drug-resistant 24 ; 18 patients responded well to treatment. The Italian family with familial infantile myoclonic epilepsy was drug-responsive, with 5 members (1a-1e) free from tonic-clonic seizures and with rare myoclonic seizures, on valproate or phenobarbital, while the remaining affected individuals (1f-1h) were not on any antiepileptic medication, and experienced mild myoclonic seizures triggered by repetitive movements or fatigue and rare tonic-clonic seizures (every 2-3 years). One patient (25) with generalized epilepsy was seizure-free on phenytoin and clobazam. Two siblings (12a and 12b) with focal seizures had significant improvement of seizure control after introduction of zonisamide; 1 patient (11) with infantile myoclonic epilepsy had a good response to topiramate. Patient  Supplemental data at Neurology.org months, on a combination of oxcarbazepine and sulthiame, possibly due to these drugs or less frequent febrile episodes. Another 5 patients (5c, 5d, 20a, 20b, 22) had adequate seizure control on 2 or 3 antiepileptic drugs. Details of epilepsy phenotypes are presented in table e-1.
EEG and neuroimaging. Results are reported in table e-2. Thirteen patients had a normal interictal EEG record. Various features, including slow background activity and multifocal paroxysmal abnormalities, were described in 35 patients. Only 2 patients had a photoparoxysmal EEG response (1b, 1h). There was no evidence of clinical photosensitivity in any patient. Eye closure sensitivity 25 was reported in one patient with DOORS with generalized epilepsy (29).
Developmental course. Thirty-nine individuals had intellectual disability or developmental delay, from mild to profound. Only the 8 affected individuals of the Italian family with familial infantile myoclonic epilepsy (1a-1h) had normal psychomotor development and no signs of cognitive deterioration over an average follow-up of 52 years. Patient 2 had normal development to the most recent follow-up (aged 7 months).
Genotype-phenotype correlation. The TBC1D24 mutations in the 48 individuals are presented in tables e-1 and e-4. All patients had biallelic mutations, but we also include in this analysis patient 24, in whom only one TBC1D24 mutation could be identified, but who has a typical DOORS phenotype. Figure 1 shows the different types of mutations and the related phenotypes. We noted an unfavorable outcome associated with frameshift, nonsense, or splice-site mutations, indicating that loss of function produces more severe disease. At least one such mutation occurred in 17 patients. Of these, 15 had drug-resistant epilepsy and 8 of them died by the age of 7 years. Only the 2 siblings (12a, 12b) with focal seizures and a frameshift mutation (in the TLDc domain) in trans with a missense (p.Ala39Pro) mutation had adequate epilepsy control, following the introduction of zonisamide. The diagnosis of DOORS per se was not associated with a specific epilepsy type or outcome. No specific pattern of outcome or severity emerged in association with missense mutations, except for missense mutations associated with death: these mutations occurred in or before the TBC domain. Frameshift, nonsense, and splice-site mutations led to drug-resistant epilepsy or death, except when occurring in the last exon (figure 1). In supplemental data S3, we present genotype-phenotype correlation by mutation, rather than by patient, crossreferencing data from tables.
To identify conserved protein regions in silico, we compared human TBC1D24 and Drosophila Sky protein sequences with 3 vertebrate and 7 insect homologues (figure 2A). To find motifs specific for this TBC domain protein, we defined a negative set (human and mouse TBC1D1 and TBC1D7). Discriminative motif discovery yielded 7 highly conserved motifs (E , 10 2100 ; p , 10 220 ) of 21-22 amino acids, completely conserved among Drosophila species and in key amino acids across species, signifying the importance of these domains ( figure 2B). Interestingly, the second motif (figure 2B) contains the site of the most frequent mutation (p.Arg242Cys) in patients with DOORS, and the central arginine residue is conserved in every species tested. The residue altered in the p.Arg242Cys mutation is flanked by 2 lysines and pairs of valines and leucines in all sequences analyzed, suggesting that this motif's positive charge is highly conserved. In addition, motif 1 encompasses the p.Pro93Ser mutation identified in patient 9 (table e-1). Mutations also lay in motif 3, located in the region between the region between the TBC and TLDc domains, and in motifs 4 and 7, located in the TLDc domain (figure 2C), though no clear correlation with the phenotype emerges.
To determine whether the position of the amino acid change in TBC1D24 or the corresponding severity of the disease correlate with perturbation of induction of neurite overgrowth, we studied a panel of 4 human TBC1D24 mutants, representing the phenotypic breadth from this study: Arg40Leu, 3 Arg242Cys, 3 Arg270His, 23 and Arg360Leu. 12 In primary mouse cortical cells expressing wild-type TBC1D24, we observed a large (;10-fold) increase in neurite outgrowth compared to those expressing a control empty vector, as expected ( figure 3). All 4 TBC1D24 mutants tested were also able to promote outgrowth; however, whereas expression of either the Arg242His or Arg360Leu mutant leads to a significant reduction in neurite length compared to wildtype, the Arg40Leu and Arg270His mutants did not significantly affect outgrowth ( figure 3). Together, these data indicate for the first time that mutations causing the most severe TBC1D24-associated disorders do not necessarily give rise to the most disruptive functional effects on neurite outgrowth in cultured neurons. DISCUSSION The reported broad spectrum of epileptic and developmental phenotypes associated with TBC1D24 mutation is unusual, and seen only with a few epilepsy-related genes, mostly with dominant causal mutations (e.g., SCN1A, SCN2A, SCN1B, KCNQ2, KCNT1, PRRT2, DEPDC5, TSC1, and TSC2). TBC1D24 is associated with an even broader phenotypic spectrum, including a number of conditions other than epilepsy alone (i.e., DOORS syndrome and nonsyndromic deafness) and involving multiple organs other than the brain (see supplemental data S2). Furthermore, TBC1D24 epilepsy syndromes occur with both compound heterozygous and homozygous recessive mutations. The additional diversity of TBC1D24 phenotypes might be due to its broader expression pattern; TBC1D24 is expressed in several human tissues, with the highest expression in brain, in multiple cerebral areas, including all layers of cerebral cortex and hippocampus. 1, 10,28 Although early-onset myoclonic epilepsy, with onset in the first year of life and myoclonic seizures often occurring in prolonged clusters, and drug resistance are the most common TBC1D24 epilepsy phenotypes, many dissimilarities have emerged in our cohort. We delineate a spectrum ranging from a benign pattern, restricted to infancy, with complete seizure control and without intellectual disability, to early-onset epileptic encephalopathy with drug resistance, severe developmental delay, intellectual disability, and early death (figure 4). Seizure types can be diverse, often triggered by fever or infections. Episodes of status epilepticus or prolonged seizures are common. Interictal EEG and neuroimaging results are variable. There was no marked intrafamilial phenotypic variability in affected patients with the same mutations, unlike that observed in families with SCN1A or DEPDC5 mutations. While we noted phenotypic pleiotropy for the same mutation in unrelated patients, in most patients these were compound heterozygous with a different second mutation, confounding detailed comparison.
Mutations leading to DOORS syndrome and those causing primarily epilepsy syndromes have mostly been mutually exclusive to date. This phenotypic pleiotropy is unexplained by current knowledge, and few genotype-phenotype correlations emerged from our study. More severe phenotypes were associated with nonsense, frameshift, or splice-site mutations, as expected, except when these occurred in the last exon. The latter might lead to milder phenotypes because RNA escapes from nonsense-mediated decay when truncating mutations occur in the ultimate exon and the end of the penultimate exon. A more variable outcome was found with missense mutations; these were associated with death only when they occurred in or before the TBC domain ( figure 1). Otherwise, the occurrence of mutations in the TBC domain vs the TLDc domain did not seem to lead to different clinical phenotypes. A TLDc domain is also found in the protein encoded by the human gene OXR1. Mice lacking Oxr1 display cerebellar neurodegeneration: neurons are less susceptible to oxidative stress-induced neurodegeneration when Oxr1 is overexpressed both in vitro and in vivo. 29,30 Not all patients with mutations in the TBC1D24 TLDc domain in our study show clinical or neuroradiologic evidence of cerebellar neurodegeneration. The involvement of the TLDc domain in oxidative stress resistance, 20 and thereby potentially in inflammatory mechanisms, might underlie the role of fever or infections in precipitating seizures or status in some patients. Conservation analysis showed a highly conserved arginine residue underlying the recurrent pathogenic chr16:2546873C . T transition, resulting in an arginine to cysteine (p.Arg242Cys) substitution, in the TBC domain. In the TLDc domain, 4 conserved motifs were identified. However, the analysis of evolutionarily conserved motifs also does not provide a complete answer to the diversity observed.
The phenotypic spectrum observed might partly be explained by the effect of individual TBC1D24 mutations that affect synthesis, stability, and activity of the protein. Alternately, variations in TBC1D24associated proteins and pathways could underlie the disease spectrum. The TBC1D24 protein interacts with ARF6, 1 a small GTP-binding protein implicated in exchange between the plasma membrane and the endocytic compartments. 31 The Drosophila homolog of TBC1D24, Sky, also interacts with, and activates, Rab35, a small GTPase involved in endosomal trafficking of synaptic vesicles, regulating neurotransmitter release. 2,32 Several genetic modifiers of this pathway were identified recently. 5 In the CNS, ARF6 participates in several aspects of neuronal development and plasticity. [33][34][35] Interestingly, some pathogenic mutations in TBC1D24 affect protein binding to ARF6 and result in severe impairment of neuronal development. 1,4,11 Furthermore, overexpression of TBC1D24 induces a marked increase in neurite length in vitro. 1,6 From these initial studies, it was postulated that each of the disease-causing mutations tested (Phe251Leu, Asp147His, and Ala515Val) were likely to be loss-of-function based on the observed reduction in outgrowth compared to wild-type. 1, 6 We investigated this specific feature of TBC1D24 function for a range of pathogenic mutations and showed that mutations that cause even the most severe seizure-related disorders with neurodegeneration are not necessarily detrimental to neurite outgrowth in the chosen assay. Thus, we found no obvious correlation between this particular function of TBC1D24 in neurons and the phenotypic spectrum we describe here. However, we provide functional data for the first time that compare pathologic human mutations in parallel across the spectrum of TBC1D24-associated disorders. Neuronal outgrowth has been used previously as an assay to assess TBC1D24 function, and our data extend these findings, showing that the relative position of the amino acid change in the protein, with respect to disease severity, does not correlate with abnormal outgrowth. As such, we have discovered that the association between TBC1D24 genotype and phenotype is complex and is likely to involve several aspects of the gene's function, potentially beyond the ARF6-mediated trafficking and signaling pathways described to date.
TBC1D24 activity also regulates synaptic function. The inhibition of excessive endosomal sorting of dysfunctional synaptic vesicle proteins and their subsequent breakdown at lysosomes, by partially inhibiting the ESCRT complexes or the HOPS complex, significantly suppresses the excessive neurotransmission in Sky mutants. 2,5 Variations in the aforementioned pathways may thus modify the neuronal pathway in which Sky/TBC1D24 functions, further adding to the phenotypic spectrum associated with its loss of function in patients. Though we used a standardized template for data collection, we acknowledge that one limitation of the study is the involvement of different clinicians in defining the clinical phenotype, with potential differences in interpreting clinical, EEG, and neuroradiologic patterns. In several patients, segmental myoclonic seizures were detected on video-EEG telemetry, but had not been reported by the parents. Involvement of other systems, e.g., hearing, might also have not been formally assessed and therefore the concomitant presence of DOORS syndrome might have been underreported. We collected a modest number of patients in absolute terms, but TBC1D24-related epilepsy is rare and we report a relatively large series compared with other rare genetic neurologic conditions with epilepsy. We only included patients with confirmed TBC1D24 mutations and epileptic seizures as part of their phenotype. This might represent ascertainment bias, and we may have missed individuals with neurologic involvement other than epilepsy.
Our findings support the need to test for TBC1D24 mutation if the phenotype is appropriate. Screening should be considered, for example, if a patient has myoclonic seizures, intellectual disability, and any other extra-CNS features (including facial, cranial, acral, or other organ abnormalities).
Anyone interested in joining the TBC1D24 consortium should contact the corresponding authors.  Phenotypic spectrum of patients with TBC1D24 mutation

AUTHOR AFFILIATIONS
The graph displays the age at epilepsy onset along the horizontal axis, the categorical outcome varying from seizure control to death along the vertical axis. The degree of intellectual disability is represented by the different fill of the circles, varying from white to black.