Genotype–phenotype correlation in contactin-associated protein-like 2 (CNTNAP-2) developmental disorder

Contactin-associated protein-like 2 (CNTNAP2) gene encodes for CASPR2, a presynaptic type 1 transmembrane protein, involved in cell–cell adhesion and synaptic interactions. Biallelic CNTNAP2 loss has been associated with “Pitt-Hopkins-like syndrome-1” (MIM#610042), while the pathogenic role of heterozygous variants remains controversial. We report 22 novel patients harboring mono- (n = 2) and bi-allelic (n = 20) CNTNAP2 variants and carried out a literature review to characterize the genotype–phenotype correlation. Patients (M:F 14:8) were aged between 3 and 19 years and affected by global developmental delay (GDD) (n = 21), moderate to profound intellectual disability (n = 17) and epilepsy (n = 21). Seizures mainly started in the first two years of life (median 22.5 months). Antiseizure medications were successful in controlling the seizures in about two-thirds of the patients. Autism spectrum disorder (ASD) and/or other neuropsychiatric comorbidities were present in nine patients (40.9%). Nonspecific midline brain anomalies were noted in most patients while focal signal abnormalities in the temporal lobes were noted in three subjects. Genotype–phenotype correlation was performed by also including 50 previously published patients (15 mono- and 35 bi-allelic variants). Overall, GDD (p < 0.0001), epilepsy (p < 0.0001), hyporeflexia (p = 0.012), ASD (p = 0.009), language impairment (p = 0.020) and severe cognitive impairment (p = 0.031) were significantly associated with the presence of biallelic versus monoallelic variants. We have defined the main features associated with biallelic CNTNAP2 variants, as severe cognitive impairment, epilepsy and behavioral abnormalities. We propose CASPR2-deficiency neurodevelopmental disorder as an exclusively recessive disease while the contribution of heterozygous variants is less likely to follow an autosomal dominant inheritance pattern. Supplementary Information The online version contains supplementary material available at 10.1007/s00439-023-02552-2.


Introduction
Contactin-associated protein-like 2 (CNTNAP2) is one of the largest genes in the human genome located on chromosome 7q35-36.1 (Nakabayashi and Scherer 2001). It encodes for CASPR2, a member of the neurexin superfamily of cell adhesion proteins (Poliak et al. 1999). CASPR2 is a presynaptic type 1 transmembrane protein, with a large extracellular and smaller intracellular portion that participates in cell-cell adhesion and synaptic interactions. CNTNAP2 is expressed throughout the developing and adult central nervous system (CNS) (Peñagarikano 2011). Mouse studies have uncovered a role for CASPR2 in neuronal migration and postmitotic neuronal development Fernandes et al. 2019). Experimental studies on knock-out mice and in human cell lines support the hypothesis that CASPR2 is involved in neuronal migration, myelination, and neuronal Gianluca D'Onofrio and Andrea Accogli contributed equally to the present work.
Extended author information available on the last page of the article transmission with a reduction in both inhibitory GABAergic neuronal numbers and excitatory neurotransmission (Peñagarikano 2011).
A homozygous 1-bp deletion (c.3709delG) of CNTNAP2 was initially detected in an Old Order Amish kindred, whose nine affected children exhibited mild motor delay until the onset of intractable seizures during infancy, which were followed by deterioration in learning and language abilities, and social behavior (Strauss et al. 2006). Three subjects showed unilateral cortical dysplasia of the anterior temporal lobe, and neuronal migration defects from brain specimen biopsies. Altogether, this neurological disorder was named cortical dysplasia-focal epilepsy (CDFE) syndrome (Strauss et al. 2006). Subsequently, homozygous or compound heterozygous variants and/or intragenic deletions within CNTNAP2 were associated with Pitt-Hopkins like syndrome 1 (PTHSL1, MIM#610042), with variable features that included intellectual disability (ID), early seizure onset, regression of language ability, and hyper-breathing patterns (Strauss et al. 2006;Zweier et al. 2009;Smogavec et al. 2016). Given the lack of typical Pitt-Hopkins craniofacial features and hyper-breathing patterns in most patients, it has recently been proposed that biallelic loss of CNTNAP2 results in a disorder called "CASPR2-deficiency neurodevelopmental disorder (NDD)", which includes severe ID, early infantile seizures, language regression, variable presence of autistic features, hyporeflexia and ataxia (Rodenas-Cuadrado et al. 2016).
A growing body of literature over the last two decades underscored a possible role of heterozygous chromosomal translocations and deletions, single nucleotide polymorphisms (SNPs), and rare heterozygous variants of CNT-NAP2. These were found in a wide array of neuropsychiatric disorders, such as autism spectrum disorder (ASD), schizophrenia, obsessive-compulsive disorder, Gilles de la Tourette syndrome, attention deficit hyperactivity disorder (ADHD), dyslexia, specific language impairment and stuttering (Verkerk et al. 2003;Arking et al. 2008;Friedman et al. 2008;Mikhail et al. 2011;Newbury et al. 2011;Ji et al. 2013;Centanni et al. 2015). However, heterozygous CNTNAP2 variations are also present in the healthy population including healthy parents of children with either mono-or biallelic variants. Thus, the evidence for the role of heterozygous variants in CNTNAP2 in neuropsychiatric disorders has yet to be clarified (Toma et al. 2018). The identification and description of new patients with CNTNAP2 variants may further define the criteria of the syndrome and better characterize its genotype-phenotype correlation (Rodenas-Cuadrado et al. 2016). We report 22 patients harboring mono-or biallelic variants in CNTNAP2 and show genotype-phenotype correlations by including a further 50 previously reported patients.

Patient recruitment
We recruited 22 previously unreported patients from 17 unrelated families carrying mono or biallelic variants in CNTNAP2. Patients were followed up at 16 centers worldwide for developmental and epileptic encephalopathy (DEE) and/or neurodevelopmental disorders. Genetic analyses were performed either in a diagnostic or research setting. Subsequently, they were enrolled using the international platform GeneMatcher (Sobreira et al. 2015).
Firstly, the respective referring clinicians were asked to fill in a spreadsheet with all clinical and genetic information for each patient (Online Resource). Secondly, all available clinical and genetic data, electroencephalography (EEG) and neuroradiological images were reviewed by expert pediatric neurologists, neuroradiologist and geneticists. Written informed consent was obtained from parents or guardians.

Genetic testing
Most CNTNAP2 variants were detected by epilepsy Next Generation Sequencing (NGS) panel (n = 11) or autism/ID NGS panel (n = 1). Exome sequencing (ES) (singleton n = 3; trios n = 7) was performed in the respective collaborating centers using different analysis platforms according to the BWA/GATK's based pipelines. Targeted Sanger sequencing using standard methods was also performed either for verification of identified variants or segregation analysis. Sequencing methods and additional genetic analyses performed per individual are summarized in Online Resource. All variants were classified according to the ACMG/AMP criteria (Richards et al. 2015). CNTNAP2 variants are listed according to the transcript NM_014141.6 and copy number variants (CNV) refer to the hg19/GRCh37 assembly.

Literature review
We performed a literature review on MEDLINE (accessed by PubMed, updated to December 2022) with the search term "CNTNAP2" and "CASPR2", including articles with reported pathogenic or likely pathogenic variants or variants of uncertain significance (VUS) in CNTNAP2 that were suspected to contribute to the phenotype of patients. Patients with copy number variation (CNV) that encompassed other genes that were likely to contribute to the phenotype and/or reports without available clinical information were excluded. We also excluded reports of subjects with limited clinical information.

Statistical analysis
We used descriptive analysis to characterize our cohort and previously published CNTNAP2 patients. Based on both datasets, we compared the phenotypes of patients harboring a heterozygosity variant versus (vs.) patients with biallelic variants using chi-squared test or a Fisher's exact test.

Neurodevelopment
Global developmental delay (GDD), of variable severity, is reported in almost all patients (21/22, 95.5%). Moreover, individual-1 (Ind-1) and Ind-3 had early normal development before the onset of epilepsy, leading to major irreversible regression, while Ind-17 experienced a partial recovery of her cognitive and motor skills after seizure control. Intellectual disability (ID) has been assessed as mild in 4 patients, moderate in 9, severe in 7 and profound in 1 subject, whereas Ind-22 had a borderline intelligence quotient.

Other comorbidities
Extra-neurological comorbidities occurred in nine individuals (9/22, 40.9%), including recurrent respiratory Table 1 Summary of clinical and neuroradiological features in our cohort of patients with CNTNAP2 pathogenic variants  Table 1 (continued) Family 8 Family 9 Family 10 Family 11 Family 12 Family 13 Family 14 Family 15 Family 16 Family 17 Language impairment infections, haematological disorders (pancytopenia, haemolytic anaemia) and rectal prolapse (2/22, 9%). Precocious puberty, asthma, hypogammaglobulinemia, gastroesophageal reflux and osteopenia were reported once (1/22, 4.5%). None of our patients presented with congenital abnormalities of any extra-CNS organ. Two patients in our cohort deceased: the first at the age of 13 (Ind-1) due to cachexia in the context of feeding difficulties and severe GDD, while the other (Ind-20) at 7 years for unknown reasons. No statistically significant differences were observed when comparing patients with a history of consanguinity and non-consanguinity.

Genetic results
A total of 18 distinct CNTNAP2 variants were identified, seven of which were novel (Online Resource). Except for two heterozygous variants, all other individuals were found to harbor biallelic variants; either homozygous (n = 16) or compound heterozygous (n = 4) variants. Variants included ten likely gene-disrupting (LGD) variants, four intragenic deletions (identified either by microarray or an epilepsy NSG gene panel) and three missense variants. Sanger sequencing confirmed variants segregation with the phenotype within these families. All variants were absent or extremely rare in human population variant databases (allele frequency ranging from 0 to 0.0001557 in the gnomAD database). None of the variants were reported in a homozygous state in healthy individuals.
LGD variants were scattered Mild white matter volume reduction with consequent ventricular enlargement is noted in Ind-3, Ind-13 and Ind-16 (asterisks). Cerebellar dentate nuclei T2 hyperintensity is visible in Ind-6 and Ind-12 (arrowheads). In Ind-2, Ind-8 and Ind-12 there are additional uni-or bilateral T2 hyperintensities at the level of the anterior temporal lobes (thick arrows) in keeping with focal cortical dysplasias     NA not available, -not present throughout CNTNAP2 and included four different frameshift and four nonsense changes and two splice site variants. All frameshift and nonsense variants were predicted to result in premature termination codon and, therefore, likely be degraded through nonsense-mediated mRNA decay (NMD). As such these were classified as pathogenic/likely pathogenic. Of note, the frameshift variant c.1361_1362del p.(Asn454ArgfsTer24) was recurrent in eight subjects of families 1-4 of Croatian Roman ancestry and the nonsense variant c.3262C > Tp.(Arg1088Ter) was found in three subjects of two nonrelated Egyptian families suggesting these variants are likely to be founder mutations in these populations. Two individuals carried homozygous splicing variants as follows: the variant c.1777+2T > C (Ind-11) affects the consensus GT-splice donor site of intron 11 and was computationally predicted to cause a loss of a splice donor site disrupting the reading frame and resulting in NMD (Splice AI score 0.98). Thus, it was classified as likely pathogenic according to the ACMG criteria. The homozygous variant c.550+5G > T (Ind-15) predicts a loss of a splice donor site (Splice AI score 0.66), yet it remains a VUS according to the current ACMG guidelines. CGH-array revealed two intragenic CNTNAP2 deletions in Ind-17: a 31,949 bp deletion in 7q35 (147,651,683,766) encompassing exon 15 and inherited by her mother and a paternally inherited deletion of 9317 bp in 7q36.1 (148,071,080,632), encompassing exon 22. These deletions were confirmed by multiplex ligation-dependent probe amplification. Epilepsy gene panel showed two compound heterozygous deletions in Ind-10, namely the c.98-?_402+? that encompasses exons 2-3 of CNTNAP2 and the heterozygous deletion c.98-?_1348+?, encompassing exons 2-8. Both deletions were confirmed by CGH-array. Individual-13 harbored the compound heterozygous missense variant c.400T > G p.(Trp134Gly) and c.2449G > A p.(Gly817Arg) that were classified as VUS. We included subjects harboring these biallelic VUS given supporting criteria of pathogenicity and consistent phenotype. The heterozygous missense variant c.3814A > T p.(Ile1272Phe) was found to be de novo in individual 21 while an ID/ASD panel identified in Ind-22 the frameshift variant c.1628del p.(Ser543Ilefs*13) that was maternally inherited. Both variants were classified as VUS.
Overall, we ascertained a diagnosis with biallelic CNT-NAP2 pathogenic/likely pathogenic variants in 18 out of 22 subjects included in this study. No other pathogenic/ likely pathogenic variants were identified in the currently known NDD-related genes in the ES data in these families. Additional VUS detected in our cohort either by ES or microarray are listed in Online Resource. Fig. 2 EEG features. a. Ind-1, 10 years old. Sleep recording. High voltage bilateral anterior delta waves and focal spikes over the frontal regions of both hemispheres. b Ind-2, 2 years 9 months. Awake recording. Synchronous and asynchronous spikes on bilateral fron-tal-temporal regions. c Ind-10, 3 years old. Awake recording. Right central-temporal medium voltage sharp waves. d Ind-10, 3 years old. Sleep recording. Nearly sub-continuous trend of right central-temporal sharp waves in the N2 phase, with a tendency to spread

Genotype-phenotype correlation
Altogether, GDD and epilepsy were significantly more present in patients harboring homozygous variants than in heterozygous patients (p < 0.0001) (Online Resource). Similarly, Fig. 3 Summary of the key clinical features of patients carrying mono-or bi-allelic pathogenic CNTNAP2 variants in our cohort and the literature. ID intellectual disability, NPsy neuropsychiatric findings. Statistical significance refers to patients with biallelic versus monoallelic variants Fig. 4 CNTNAP2 variants position in our cohort (in bold, # individual) and previously published patients. The arrow indicates a deletion, and the line a duplication ASD (p = 0.009), hyporeflexia (p = 0.012), language impairment (p = 0.020), as well as a moderate to severe degree of ID (p = 0.031) were more frequent in patients with biallelic variants.

Discussion
We reported a cohort of 22 new patients harboring either biallelic (20) or monoallelic variants (2) in CNTNAP2.
To the best of our knowledge, this is the largest cohort of patients with CNTNAP2 variants reported together to date. Our study corroborates previous literature, confirming that CNTNAP2 deficiency due to biallelic variants leads to a distinct neurodevelopmental disorder typically characterized by developmental delay, seizure onset within the first 2 years followed by developmental regression, moderate to severe ID and variable occurrence of ASD and behavioral abnormalities. Similarly to previous reports, hypotonia and hyporeflexia are frequent, whereas only a few patients display ataxia. Likewise, occipital frontal circumference is normal in the majority of patients in contrast to the initial reports of relative macrocephaly. Furthermore, our patients harboring biallelic variants do not display the typical craniofacial features and abnormal breathing patterns reported for PTHS. Together, this supports previous literature suggesting that the name PTHS1 should be replaced by CASPR2-deficiency NDD (Rodenas-Cuadrado et al. 2016). In addition, the occurrence of epilepsy in virtually all patients within the first 2 years with consequent regression of development and cognitive impairment would suggest a DEE. Epilepsy is indeed a cardinal feature in patients with biallelic CNTNAP2 variants. The onset of seizures typically occurs in the first two to three years of life. Seizures initially are very frequent and difficult to treat. However, most patients achieve good seizure control within a few years after onset. Seizures are most frequently focal motor, at times with secondary generalization This is in line with previous descriptions in the literature (Strauss et al. 2006;Rodenas-Cuadrado et al. 2016;Smogavec et al. 2016). Cortical areas most typically involved seem to be the frontal and temporal regions (Strauss et al. 2006).
CASPR2 is found in the inhibitory presynaptic compartment and, to a lesser extent, in the excitatory postsynaptic compartment where it is involved in several pivotal processes, such as neurite development and synapse maturation, stability, and function (Horresh et al. 2008). It also localizes to juxtaparanodes of myelinated axons, where it is involved in neuron-glia interactions, and mediates the clustering of potassium channels via interaction with contactin 2 (also known as TAG-1) (Horresh et al. 2008). Similar to humans, Cntnap2−/−mice display epilepsy in addition to ASD features and cortical developmental abnormalities (Peñagarikano et al. 2011). RNAi-mediated knock-down of Caspr2 produced a cell-autonomous decrease in dendritic arborization and spine development in pyramidal neurons, decreasing the number of excitatory and inhibitory synapse numbers, and impairing synaptic transmission (Anderson et al. 2012). Together, these observations suggest that a perturbation of synaptic homeostasis and function due to CASPR2 deficiency leads to an imbalance of excitatory and inhibitory post-synaptic currents in neural networks that may contribute to epilepsy phenotypes (Anderson et al. 2012). Strauss et al. (2006) described neuroimaging features of focal cortical dysplasia in three subjects that were consistent with findings of neuronal migration defects from brain biopsies. These results were in line with neuropathological and physiological studies in the Cntnap2−/−mice showing neuronal migration abnormalities, reduced number of interneurons and abnormal neuronal network activity (Peñagarikano et al. 2011). Subsequent to this, no further reports have described malformations of cortical development: however, cerebellar hypoplasia and nonspecific white matter abnormalities have been occasionally reported in subjects with biallelic CNTNAP2 variants (Zweier et al. 2009;Smogavec et al. 2016). Here, we describe the largest cohort of subjects for whom brain MRI was available, showing that three subjects had unilateral or bilateral anterior temporal lobe T2 hyperintensities consistent with focal cortical dysplasia, supporting the notion of malformation of cortical development due to CNTNAP2-deficiency. Interestingly, we also noted several nonspecific findings that have been described in subjects with PTHS, including callosal anomalies, white matter volume reduction, dentate nuclei signal alterations and other minor posterior fossa abnormalities.
All our patients suffered from severe speech impairment and one-third had ASD or other behavioral abnormalities including aggressive behavior and stereotypic movements. There is evidence that supports a role for CNTNAP2 in language development, including enriched expression during human brain development in frontotemporal-subcortical circuits known to be critical for human executive function . Despite some conflicting results (Sampath et al. 2013;Murdoch et al. 2015;Toma et al. 2018;Zhang et al. 2019), several studies have linked SNPs in CNTNAP2 variants with ASD and/or language-related disorders (Vernes et al. 2008;Li et al. 2010;Gregor et al. 2011;Uddin et al. 2021). Further, some SNPs (e.g. rs2710102 and rs7794745) have been associated with abnormal activation of the right inferior frontal gyrus (Broca's area homologue) and right lateral temporal cortex in subject with ASD and reduced volume of specific grey matter areas (Whalley et al. 2011). Together this evidence supports an impact of CNT-NAP2 variation on language related brain regions and phenotypes; however, it is not yet clear what role (if any) CASPR2 has in the development of language.
While the loss of function (LoF) mechanism due to biallelic CNTNAP2 variants is well understood, the impact of heterozygous CNTNAP2 variants is more controversial. It has been suggested that the phenotypic picture of each heterozygous variant may result from the combination of two mechanisms. On the one hand a dominant-negative effect on wild-type Caspr2 function might be due to endoplasmic reticulum (ER) retention mimicking the homozygous null phenotype . On the other hand a loss of function mechanism for adhesion-defective variant proteins, could enable the interaction with their extracellular partners . According to this model, the phenotype of our patient, Ind-21 (mild ID, epilepsy and behavioral abnormalities) harboring the de novo missense variant p.(Ile1272Phe) lying in the extracellular domains may be due to a LoF mechanism if the protein is secreted from the ER. However, the impact of the de novo frameshift variant in the patient (Ind-22) with isolated ASD remains controversial since it is predicted to undergo nonsense-mediated decay and thus it would unlikely exert a dominant negative effect. It is also noteworthy that CNT-NAP2 is not constrained for missense and Lof variants in the gnomAD databse (Z score − 0.29, pLI score 0) indicating that heterozygous missense and Lof variants of CNTNAP2 are not subject to negative selection (Lek et al. 2016). This is in line with the fact that carrier parents of CNTNAP2 variants are healthy. Furthermore, large scale studies on gene enriched for de novo variants in NDD have failed to highlight this gene with any meaningful significance (Kaplanis et al. 2020;Satterstrom et al. 2020) and several other studies did not identify a significant burden for CNTNAP2 rare variants in patients with ASD or schizophrenia comparing to controls (Murdoch et al. 2015;Toma et al. 2018;Zhang et al. 2019), suggesting that CNTNAP2 is not a a primary risk gene for psychiatric disorders. Although it might be possible that CNTNAP2 heterozygous variants contribute to ASD and related neuropsychiatric phenotypes with a polygenic inheritance pattern, it seems unlikely based on the above observations that they solely result in a neuropsychiatric phenotype following a classical autosomal dominant Mendelian inheritance. Taken together, we propose CNTNAP2-related NDD as an exclusively recessive disorder while the dominant version is becoming weaker with the increase body of evidence in the literature and in human population variant databases.
In conclusion, we report the largest cohort of patients with CNTNAP2 variants to date and define the core phenotype associated with biallelic CNTNAP2 variants. These data suggest that patients with biallelic variants are likely to develop severe cognitive impairment, epilepsy and variable behavioral abnormalities.
In most cases, patients have an unremarkable perinatal history and a normal psychomotor development or slightly delayed during the first year of life. Concomitant with the epilepsy onset, occurring more often during the second year of life, developmental stagnation or regression is observed. Epilepsy can be difficult to control at the beginning, with a "stormy" phase, while during childhood seizures are usually well-controlled with ASMs. Response to ASMs may be associated with a slight cognitive improvement in some cases, although most patients still suffer from moderate to profound ID throughout their lives. In more severe cases, feeding difficulties, failure to thrive with increased potentially fatal comorbidities may be observed.
The role of heterozygous variants remains to be fully elucidated. Future studies should address the functional impact of heterozygous CNTNAP2 variants and the related pathomechanisms with ultimately important implications for patient management and counselling. Consent to publish The authors affirm that human research participants provided informed consent for publication of the images in Figs. 1 and 2.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.