Clinical features of affected individuals

The main clinical features of the affected individuals from the three unrelated families are summarized in Table 1. Affected individuals presented with dysmorphism, ID, developmental delay, and behavioral abnormalities.

Table 1 Clinical features observed in individuals with NAV3-related neurodevelopmental disorderFamily 1

Family 1 is a large consanguineous family originating from southern Pakistan and segregates NDD with an autosomal recessive mode of inheritance (Fig. 1A-i). In this family, NDD was present in 7 affected individuals (6 males and 1 female) in three nuclear families, including one deceased individual (VI:10) due to unknown reasons (Fig. 1A-i). Affected individuals presented with dysmorphism, microcephaly, ID, and developmental delay (Fig. 1B-i).

At the time of physical assessment, detailed clinical features were obtained for three affected individuals: VI:1 age 33 years, VI:2 age 22 years, and VI:3 age 20 years (Fig. 1B-i). Their occipitofrontal circumferences (OFC) measurements were 53 cm (1.42 SD), 52 cm (−2.09 SD), and 52 cm (−2.17 SD), respectively, consistent with mild microcephaly (Supplementary Table 1). These three individuals had mild to moderate ID and developmental delay. Dysmorphic features included a prominent midface, apparently large prominent ears, and a large nose (Fig. 1B-i). They had normal height, absent speech, no cleft lip or cleft palate, no seizures, and no single central maxillary incisor was observed. Additionally, they exhibited behavioral abnormalities such as hyperactivity, aggressiveness, and attention deficit. They showed neither a social smile nor cooing sounds (Table 1 and Supplementary Table 1). Echocardiography was performed for VI:2, which revealed an apparently normal heart.

Family 2

The family 2 with complex neurological and developmental features was ascertained from northern Pakistan (Fig. 1A-ii). The three affected brothers (IV:2, IV:5, and IV:6) and a deceased girl (IV:1) presented with recessive NDD (Table 1 and Supplementary Table 3). The clinical features common among the three patients are dysmorphism, mild to moderate ID, developmental delay, and early-onset nystagmus.

The dysmorphic features include a prominent midface, apparently large prominent ears, and a large nose which was more prominent in IV:2 (Fig. 1B-iv). Remarkably, IV:2 had microcephaly with an occipitofrontal circumference of 49 cm (−3.5 SD). They had social smile, normal cooing sounds, normal height, normal speech, no cleft lip or cleft palate, and no seizures.

Family 3

Proband IV:1 in family 3 from Manipal, India was ascertained at five years of age (Fig. 1A-iii). She was born to a consanguineously married couple at term. Her antenatal history was uneventful, and her birth weight was 2.5 kg (−1.76 SD). Behavioral abnormalities noted were aggressiveness, hyperactivity, lack of eye contact, and non-verbal communication. She was noted to have delays in motor milestones. Neck holding was attained at one year, rolling over at two years, crawling after two years, sitting without support at three years, standing after three years, and walking at five years of age. Language milestones were also grossly delayed; she could only speak bi-syllables by the age of five years.

On examination at five years of age, her height was 116 cm (−1.2 SD) and her head circumference was 48 cm (−1.33 SD). She was noted to have dysmorphic features such as thick and arched eyebrows, up-slanted and wide palpebral fissures, broad bridge of the nose, thick vermilion, wide mouth, and mild retrognathia (Fig. 1B-ii). The ophthalmologic evaluation carried out was unremarkable. Computed tomography of the brain showed cerebellar vermis hypoplasia (Fig. 1B-iii).

Exome sequencing identifies homozygous nonsense and splice site variants in NAV3

To identify the pathogenic variant, two affected individuals (IV:2 and VI:5), one from each of the two branches of the pedigree of family 1, were selected for exome sequencing (Fig. 1A-i). As the disease segregating in family 1 is rare and transmits in an autosomal recessive mode of inheritance, we applied a filtration strategy as described in methods to detect rare homozygous variants. Further ranking of the filtered variants based on HPO term for intellectual disability (HP:0010864) and CADD score > 20, identified a homozygous nonsense variant c.6325C > T; p.(Gln2109Ter) in the NAV3 (NM_001024383.2) gene. This variant was found in a large homozygous region of 23.4 Mb (Chr12:76,386,271–99,825,429) shared by both the affected individuals (VI:2 and VI:5), as determined by AutoMap from their exome data (Supplementary Table 2 and Supplementary Fig. 2A-i). NAV3 appeared to be the best candidate after excluding the benign or likely benign variants in other genes.

Notably, NAV3 is predicted to be intolerant to loss of function variants, with a pLI score of 1 (gnomAD v4.0 project; Karczewski et al. 2019). The identified nonsense variant p.(Gln2109Ter) is predicted to create a premature stop codon and a truncated NAV3 protein with 2109 amino acids compared to wild-type protein with 2385 amino acids. The variant is ultra-rare with MAF: 1/ 1,432,136 in gnomAD v4.0. The variant is predicted to be pathogenic by several bioinformatics tools, including CADD, with a CADD score of 46 (Table 2). Finally, we confirmed by Sanger sequencing that the variant segregated in the extended family (Fig. 1A-i and Supplementary Fig. 1).

Table 2 Variants identified in NAV3 in families with recessive NDD

In family 2, ES performed on individual IV:2 (Fig. 1A-ii) identified a rare (MAF: 2/1,597,914 in gnomAD v4.0) homozygous variant c.243 + 1G > T in the NAV3 gene. The variant was predicted to be pathogenic (CADD score 35) and to alter the splice donor site of exon 1, resulting in aberrant splicing (Splice AI: 0.99) (Table 2). Validation of the variant c.243 + 1G > T followed by segregation analysis by Sanger sequencing revealed that the variant was homozygous in the affected individual (IV:2) and heterozygous in his unaffected parents. However, the variant was also heterozygous in both the affected siblings IV:5 and IV:6 (Fig. 1A-ii and Supplementary Fig. 1). Since the variant is ultra rare and predicted to have a strong effect, resampling and repetition of segregation analysis were performed to exclude the sample mix-up or any technical error. However, the results confirmed the earlier segregation analysis, revealing that both affected siblings IV:5 and IV:6 were indeed heterozygous for the NAV3 variant. ES performed for IV:5 (Fig. 1A-ii) heterozygous for the NAV3 variant, revealed a missense variant c.9410C > T; p.(Ala3137Val) in the HSPG2 gene (NM_005529) (Table 2). The missense variant is rare in gnomAD v4.0 (MAF: 3/1,613,864) and segregates in the extended family (Fig. 1A-ii). Bioinformatics analysis using CADD predicts this variant to be of uncertain significance (VUS). Altogether, it can be concluded that both variants in NAV3 and HSPG2 segregate independently in family 2 (Fig. 1A-ii). This finding is consistent with the autozygosity analysis by the AutoMap tool. Specifically, the NAV3 variant in patient IV:2 is within a 18.6 Mb RoH absent in patient IV:5, who is homozygous for the HSPG2 variant. Additionally, the HSPG2 variant is within a 6.8 Mb RoH shared by patients IV:2 and IV:5, both of whom are homozygous for HSPG2 and heterozygous for the NAV3 variant (Supplementary Fig. 2A-ii and B).

In family 3, ES analysis of IV:1 (Fig. 1A-iii), identified a homozygous variant, c.6577C > T; p.Arg2193Ter in exon 37 of NAV3 gene (NM_001024383.2). Sanger validation and segregation of the variant confirmed the carrier status in her parents (Supplementary Fig. 1). This variant is rare with a MAF of 2/1,610,416 in the gnomAD v4.1.0 database. This variant is also absent in the in-house database of 3200 local exome datasets. The variant p.(Arg2193Ter) was predicted to be pathogenic (CADD score: 44) and to create a premature stop codon, resulting in a truncated NAV3 protein with 2192 amino acids compared to wild-type protein with 2385 amino acids. Consistent with family 1, the variant was found in a large homozygous region of 64.7 Mb (chr12:52,451,817–117,222,662) as identified by AutoMap (Supplementary Fig. 2A-iii).

Overall, these results confirm that protein truncating variants in NAV3 are likely a rare cause of autosomal recessive NDD.

Expression of NAV3 in the embryonic and adult human brain

NAV3, among other tissues, is strongly expressed in the brain in mice and humans. However, its specific expression pattern in the developing and adult human brain is unclear. We re-analyzed published single-cell transcriptomic datasets of the human embryonic (GW14 to GW25) whole brain and young adult (4 to 20 years) human cortex (Fig. 3A–D). The analysis revealed that NAV3 is expressed across various cell types in embryonic and adult human brains. In general, the percentage of cells that express NAV3 is higher in young adult brains than in embryonic brains across all cell types, with a higher expression in early and late-born excitatory neurons, caudal and medial ganglionic eminences (CGE/MGE) derived inhibitory neurons, and microglia.

In both datasets, the expression is the highest in several neuronal cell types as well as microglia. In the embryonic brain, more than 12% of both excitatory and inhibitory neurons express NAV3 (Supplementary Fig. 3B). In the cortical tissues of young adult brains, more than 87% of the neurons of the cortical layers express NAV3. In both these datasets, the NAV3 expression is generally reduced in non-neuronal cells, except for microglia (Fig. 3A-D). This finding further supports the role of NAV3 in axonal guidance and cell migration during early neurodevelopment.