1 INTRODUCTION

The YWHAG gene (tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein gamma, Mendelian Inheritance in Man [MIM] 605356) encodes a member of the highly conserved 14-3-3 protein family, which is involved in multiple intracellular signaling pathways. Through binding to their targets, 14-3-3 proteins contribute to the regulation of a large spectrum of biological processes including signal transduction, cell cycle, transcription, and apoptosis.¹

Given their expression in the developing brain, especially the cortex, and their crucial role in neuronal development, 14-3-3 proteins are involved in several human diseases, mainly neurodevelopmental, neurodegenerative, and neuropsychiatric disorders.^2-4

The isoform gamma, named YWHAG, is highly expressed in the brain, skeletal muscle, and heart. Animal models indicate that alterations in this gene have consequences on neurologic development and cardiac function. Knockdown of Ywhag1 in zebrafish results in reduced brain size and enlarged diameter of the heart tube with increased risk for cardiac arrhythmia.⁵ In utero ablation in Ywhag^−/− mice, with consequent alteration of Ywhag levels, leads to delayed neuronal migration in the developing brain.⁶

Komoike et al.⁵ studied Ywhag in zebrafish and suggested that haploinsufficiency for the human orthologue YWHAG, which maps in the 7q11.23 interval that is deleted in Williams–Beuren syndrome (MIM 194050), was related to infantile seizures and cardiomyopathy in these patients. Guella et al.⁷ described heterozygous YWHAG single nucleotide variants in seven patients with early onset epilepsy. Overall, 13 missense, one nonsense, and one frameshift variant have been reported in patients with epilepsy or developmental and epileptic encephalopathy (DEE).

We characterized clinical manifestations, electroclinical features, epilepsy course, and outcome of 24 novel patients with pathogenic or likely pathogenic YWHAG variant alleles. Based on observations of the newly reported patients described herein and those previously reported, we expand the overall spectrum of phenotypes associated with YWHAG variants.

2 MATERIALS AND METHODS

We retrospectively collected clinical and molecular data from 24 patients with pathogenic/likely pathogenic YWHAG variants. Patients were diagnosed at Meyer Children's Hospital Istituto di Ricovero e Cura a Carattere Scientifico (IRCCS) (n = 2) or recruited through GeneMatcher (n = 8),⁸ collaborations with national and international epilepsy centers (n = 8), the Epi25 consortium (n = 5),⁹ and Decipher (n = 1).¹⁰ Three patients had been previously published (Patients 6, 10, and 11)^11-13 and are rephenotyped in this study. Referring physicians completed a questionnaire with comprehensive genetic and clinical information, including electroencephalographic (EEG) and neuroimaging data. Because patients were followed at different centers and had variable ages at the time of study, cognitive level was evaluated using different methods, including the Griffiths Mental Development Scale,¹⁴ the Wechsler Intelligence Scale for Children,¹⁵ and the Wechsler Adult Intelligence Scale–Revised,¹⁶ as well as adaptive behavioral criteria. Autism was diagnosed using Diagnostic and Statistical Manual of Mental Disorders, 5th edition.¹⁷

3 RESULTS

3.1 Genetic analysis

Genetic and clinical information on 24 individuals (21 new; three previously published) harboring pathogenic/likely pathogenic YWHAG variants are summarized in Table 1 and presented in more detail in Table S1. These 24 patients harbored 12 different YWHAG variants, which occurred de novo in 19 of them, were inherited from an unaffected mother in three (Patients 1, 8, 9), and remained of undetermined origin in two, whose parents were unavailable for segregation analysis (Patients 5, 11). Among the 12 pathogenic/likely pathogenic variants, nine were missense (9/12, 75%), two frameshift (2/12, 17%), and one nonsense (1/12, 8%). The three patients with a variant inherited from an unaffected parent carried a frameshift variant. Arginine 57 and arginine 132 represent two hotspot amino acid residues affected by multiple missense changes (p.Arg57Cys, p.Arg57Gly, p.Arg57His, p.Arg132His, p.Arg132Cys, and p.Arg132Gly), recurring in six and in 11 patients, respectively. Except for the p.Arg132Gly, all these changes had been previously described. One patient (Patient 10, also included in Epi4K 2013) carried the c.387C>G p.(Asp129Glu) missense substitution and another patient (Patient 23) the c.619G>A p.(Glu207Lys), a variant previously found in a different individual (Patient 48).²¹

TABLE 1. Genetic and phenotypic data of patients carrying YWHAG pathogenic/likely pathogenic variants (24 new patients, 24 patients from the literature, and the total cohort of 48 YWHAG patients).

Genetic and clinical feature	New cohort: Patients/patients with available data, n (%)	Literature: Patients/patients with available data, n (%)	Total: Patients/patients with available data, n (%)
Variant type
Missense	20/24 (83)	17/24 (71)	37/48 (77)
Frameshift	3/24 (13)	6/24 (25)	7/488 (15)
Nonsense	1/24 (4)	1/24 (4)	4/48 (8)
De novo variant	19/22 (79)	16/21 (67)	35/43 (81)
Sex
M	14/24 (58)	12/22 (55)	24/46 (52)
F	10/24 (42)	10/22 (45)	22/46 (48)
Age, median (range)	17 years (3 years 6 months–67 years)	13 years 11 months (3 years–40 months)	15 years (3 years–67 years)
Primary diagnosis
DEE	20/24 (83)	2/15 (13)	22/39 (56)
GE	4/24 (17)	12/15 (80)	16/39 (41)
Focal epilepsy		1/15 (7)	1/39 (3)
Seizure onset, median (range)	10 months (1 months–2 years)	2 years (2 months–16 years)	1 years 5 months (1 month–16 years)
Predominant seizure type at follow-up	Myoclonic: 15/23 (65)	Myoclonic: 3/9 (33)	Myoclonic: 18/32 (56)
Seizure triggers	14/23 (58)	9/9 (100)	22/32 (69)
Fever	11/14 (79)	8/9 (89)	19/22 (86)
Abnormal EEG at onset	25/21 (71)	3/3 (100)	18/24 (75)
Abnormal EEG at last evaluation	18/22 (82)	5/10 (50)	24/32 (75)
ASM
Polytherapy	20/24 (83)	7/19 (37)	27/43 (63)
VPA monotherapy	4/24 (17)	8/19 (42)	12/43 (28)
No therapy		4/19 (21)	4/43 (9)
Response to treatment	SF: 13/24 (54)	SF: 15/19 (79)	SF: 28/43 (65)
DD/ID	23/24 (96)	13/21 (62)	36/45 (80)
Mild/moderate	13/23 (57)	13/13(100)	26/36 (72)
Severe	10/23 (43)	0	10/36 (28)
Language impairment	21/24 (88)	7/12 (58)	30/36 (86)
Neurological findings	13/23 (54)	5/5 (100)	18/28 (64)
Myoclonic jerks during sleep, nonepileptic	11/22 (50)	1/1 (100)	12/23 (52)
Sleep disorders	9/23 (39)	NA	9/23 (39)
Behavioral abnormalities/ASD	16/24 (67)	4/6 (67)	20/30 (67)
Abnormal brain MRI	4/24 (17)	4/15 (27)	8/39 (21)
Dysmorphic features	6/24 (25)	9/13 (69)	15/37 (41)

• Abbreviations: ASD, autism spectrum disorder; ASM, antiseizure medication; DD/ID, developmental delay/intellectual disability; DEE, developmental and epileptic encephalopathy–myoclonic atonic epilepsy; EEG, electroencephalogram; F, female; GE, generalized epilepsy; M, male; MRI, magnetic resonance imaging; NA, not available; SF, seizure-free; VPA, valproic acid.

In the five remaining patients, we identified four novel variants (c.578C>T p.[Ala193Val], c.698G>A p.[Trp233*], c.89dup p.[Thr31Aspfs*5], and c.187_188insC p.[Ile63Thrfs*3]).

None of the variants was present in the gnomAD database (v2.1.1, accessed May 2023), and all occurred at highly conserved amino acid residues. Arginine 57 and 132 were located inside a hydrophobic groove involved in ligand binding. p.Asp129Glu was predicted to be involved in interactions with ligands through a new hydrogen bond with phosphopeptides.²² Alanine 193 and glutamate 207 mapped in a protein region that could be involved in the interaction between two α-helical structures, thus contributing to stabilizing the three-dimensional conformation.

The p.(Thr31Aspfs*5), p.(Ile63Thrfs*3), and p.(Trp233*) frameshift variants introduce a premature stop codon, leading to early protein synthesis termination. The role of YWHAG truncating variants is controversial. YWHAG is classified as a loss of function (LoF) intolerant gene (probability of being LoF intolerant = .96), but it is also predicted to escape nonsense-mediated decay (NMD⁻ gene),²³ and it is unclear whether these truncating variants act via an LoF effect. All known YWHAG variants, including those identified in this and previous reports, are depicted in Figure 1.

3.4 Analysis of all patients

Combining our findings with previous reports, 48 patients with a pathogenic or likely pathogenic variant in YWHAG can now be identified. Genetic, clinical, EEG, and MRI data of all the 48 patients are summarized in Table 1 and Figure 2.

4 DISCUSSION

The initial reports of individuals with YWHAG variants described an early onset epileptic encephalopathy with prevalently myoclonic seizures, intellectual disability, developmental delay, and facial dysmorphisms.^{7, 11} More recent reports included milder phenotypes of myoclonic epilepsy and febrile seizures.^{22, 24} By adding a consistent number of new patients, we expand the characterization of phenotypes associated with YWAHG pathogenic/likely pathogenic variants and illustrate emerging genotype–phenotype correlations for this gene.

Our data substantiate specific features observed in previously described patients and highlight additional aspects. Our series confirms a YWHAG-associated phenotype encompassing a wide spectrum of neurodevelopmental disorders, from mild developmental delay and epilepsy to severe DEE (Figure 2), and contributes novel insights about EEG features, clinical epilepsy, and its outcome. As previously reported, epilepsy is the most frequent clinical presentation, with onset in the first 2 years of life. Seizures are mainly generalized tonic–clonic, triggered by fever in 46% of our patients. Over time, myoclonic seizures become prominent (65%). EEG data in our cohort show, in most patients, slow background activity and a combination of focal and multifocal abnormalities and generalized polyspikes or spike and wave discharges. Seizure control is achieved in 54% of patients, with VPA monotherapy (23%) or a combination of drugs (77%), even when patients present as DEE at onset. These observations reinforce previous findings wherein pharmacological control of seizures was reported in several patients.^{7, 21, 22}

Developmental delay and intellectual disability of variable severity were present in most of our patients (96%), but were reported in only 60% of previously described series. This discrepancy may arise from the inclusion in the literature dataset of several patients carrying truncating variants and exhibiting mild manifestations, most of whom belonged to a single large family.²² Additional observations including a more mixed population are needed to clarify the spectrum of cognitive impairment associated with YWHAG variants. Language impairment is a common feature. Comorbidities, not systematically explored in previous reports, include behavioral disorders (67%), mainly autistic features, ADHD, and aggressive behavior. Nonepileptic myoclonic jerks during sleep were reported in 50% of our patients and sleep disorders (fragmented sleep, difficulty in settling at night, and nocturnal awakenings) in 39%. Fifty-four percent of patients exhibited neurological signs such as ataxia, tremors, clumsiness, and hypotonia. Variable brain MRI alterations, not suggesting a recurrent pattern of abnormality, were identified in 17% of our patients (4/24, 17%) and in 27% (4/15, 27%) of those previously reported, for whom, however, this information was not available in 38%. Lack of information does not allow discerning whether MRI was not performed or unrevealing, leaving open the possibility that a systematic analysis in a large cohort, ideally by applying morphometric methods, might reveal a common pattern of abnormality, despite nonspecific qualitative findings.^{25, 26}

Dysmorphisms, described in 69% of previously published patients, were observed in only 25% of those included in this report, and no recognizable recurrent pattern could be identified in this and previous studies. However, mild dysmorphisms in a rare condition are easily underreported, as they may escape recognition if patients are not seen by expert clinical geneticists and are reported via an international collaboration in which each center contributes single or very few observations, making interindividual comparison impossible.

Twelve distinct YWHAG variants are present in our series, totaling 19 when including those reported previously (Tables S1 and S2), most being missense variants. Two mutational hotspots exist: Arg132 and Arg57. These highly conserved amino acids, along with Tyr133, previously identified in two unrelated patients, form a positively charged pocket within a hydrophobic groove (BG), which is functionally important for the binding of the 14-3-3γ protein with different ligands. 14-3-3γ, encoded by YWHAG, like the other 14-3-3 proteins, functions almost exclusively in its dimeric form and exerts its function through binding with phosphopetides.²⁷ In the group of newly and previously described patients with variants affecting one of the three hotspot variants, the BG-group, the prevalence of DEEs is statistically significant (90% in BG-group vs. 19% in non-BG-group, p < .001), and associates with developmental delay and intellectual disability (100% in BG-group vs. 53% in non-BG-group, p < .001) and language impairment (96% in BG-group vs. 58% in non-BG-group, p = .006). These findings suggest that patients carrying an alteration in the key domain responsible for ligand binding are more severely affected. It is likely that the three hotspot alterations have a dominant negative effect that impairs phosphopetides binding and disrupts the regulatory role of 14-3-3γ during neuronal development, thus resulting in more severe neurological impairment. Conversely, an alteration at a different site could destabilize the dimeric structure without impeding its full formation and function. All variants falling into the BG-group are de novo and missense.

When comparing patients' phenotypes associated with missense (37 patients) versus truncating variants (11 patients), it becomes obvious that none of the patients with truncating variants received a diagnosis of DEE. In addition, the proportion of those with developmental delay/intellectual disability (94% of missense vs. 36% of truncating, p < .001), language impairment (93% of missense vs. 20% of truncating, p < .001), and neurological comorbidities (71% of missense vs. 0 truncating, p = .03) is less represented in the group with truncating variants. This milder phenotype could be due to haploinsufficiency, which leaves intact the functionality of the residual encoded protein, as opposed to the putative dominant negative effect exerted by missense variants. These data need to be confirmed in a wider series and corroborated by functional studies probing the effect of these variants on protein quantity and function.

Although not statistically significant, data on treatment response suggest that the percentage of patients who respond to ASMs seems to correlate with the type and location of genetic variants, and good seizure control is often achieved if variants fall outside the hydrophobic BG (p = .09) or are truncating (p = .02). This observation, if confirmed, might help future precision medicine approaches.

The five YWHAG variants classified as variants of uncertain significance (VUS) were associated with heterogeneous epilepsy phenotypes in four, including absence seizures (Patient 49), generalized tonic–clonic seizures triggered by fever (Patients 51, 52), and developmental delay without epilepsy (Patient 50). Except for Patient 50, who has global developmental delay and facial dysmorphisms, the remaining four patients exhibited normal development and language skills. These phenotypic characteristics seem to overlap with the milder side of the spectrum of YWHAG-related disease. Further observations in a wider cohort of patients with VUS might support the hypothesis that hypomorphic variants exist in this gene.

5 CONCLUSIONS

This cohort highlights the wide range of clinical presentations and severity in YWHAG-related disorders, which vary from mild developmental delay and epilepsy to DEE. Epilepsy onset is within the first 2 years, developmental delay/intellectual disability occurs in most patients, and language impairment is common. Other comorbidities, such as behavioral abnormalities and dysmorphic features, are variably manifested, making this syndrome difficult to recognize on clinical grounds only. A genotype–phenotype correlation begins to emerge, which reflects distinct molecular mechanisms and may prove helpful for clinical management, prognostication, genetic counseling, and precision medicine.

AUTHOR CONTRIBUTIONS

Study concept and design: Valentina Cetica, Tiziana Pisano, and Renzo Guerrini. Acquisition, analysis, and interpretation of data: Valentina Cetica, Tiziana Pisano, Renzo Guerrini, Gaetan Lesca, Dana Marafi, Laura Licchetta, Florence Riccardi, Davide Mei, Hon-yin B. Chung, Allan Bayat, Meena Balasubramanian, Daniel H. Lowenstein, Milda Endzinienė, Maha Alotaibi, Nathalie Villeneuve, Julia Jacobs, Bertrand Isidor, Roberta Solazzi, Nicolette S. den Hollander, Dragan Marjanovic, Christelle Rougeot-Jung, Julien Jung, Marion Lesieur-Sebellin, Andrea Accogli, Vincenzo Salpietro, Nebal W. Saadi, Eleni Panagiotakaki, Thomas Foiadelli, Sylvia Redon, Meng-Han Tsai, Francesca Bisulli, Trine B. Hammer, James R. Lupski, and Elena Parrini. All authors critically reviewed the manuscript and approved the final version for publication.

CONFLICT OF INTEREST STATEMENT

J.R.L. has stock ownership in 23andMe, is a paid consultant for Genome International, and is a coinventor on multiple US and European patents related to molecular diagnostics for inherited neuropathies, eye diseases, genomic disorders, and bacterial genomic fingerprinting. The Department of Molecular and Human Genetics at Baylor College of Medicine receives revenue from clinical genetic testing conducted at Baylor Genetics Laboratories. The other authors have no potential conflicts to disclose.

FUNDING INFORMATION

R.G. is supported by Tuscany Region Call for Health 2018 (grant DECODE-EE) and the Optical Brain Mapping project by Fondazione Cassa di Risparmio di Firenze. Meyer Children's Hospital IRCCS is funded by Current Research Funding (Italian Ministry of Health). M.B. is funded by Medical Research Council (MR/V037307/1) academic salary support. The Genome Sequencing Program efforts for Epi25 were supported by National Human Genome Research Institute grant 5U01HG009088. A supplemental grant for Epi25 phenotyping was supported by Epi25 Clinical Phenotyping R03, National Institutes of Health (R03NS108145). This study was supported in part by the US National Human Genome Research Institute (NHGRI) and National Heart, Lung, and Blood Institute to the Baylor-Hopkins Center for Mendelian Genomics (UM1 HG006542) and Baylor College of Medicine Genomic Research to Elucidate the Genetics of Rare Diseases (U01 HG011758), US National Institute of Neurological Disorders and Stroke (R35NS105078 to J.R.L.), and Muscular Dystrophy Association (512 848 to J.R.L.). Da.M. was supported by a Medical Genetics Research Fellowship Program through the US National Institutes of Health (T32 GM007526-42).

ETHICS STATEMENT

This study was approved by the Pediatric Ethics Committee of the Tuscany Region, and informed consent was obtained by patients, parents, or legal guardians.