Cohort characteristics
Totally, 103 unrelated patients with microcephaly of unknown etiology were enrolled. The female to male ratio was 55 to 48. The median age at latest investigation was 1 year 3 months old, ranging from 1 months old to 9 year-olds. PM and SM were determined in 53 (51.5%) and 36 (35.0%) patients, respectively (Table 1). In the other 14 (13.6%) patients, the onset of microcephaly could not be determined. Apart from microcephaly, varying degrees of different neurological signs were reported, among which developmental delay (GDD, ID) and cerebral MRI abnormalities represented the most common associated features (Table 1). There were additional features, such as dystonia or movement disorder, seizures, short stature/growth delay, and heart /urogenital malformations. An overview of the phenotypic spectrum of this cohort are summarized in Table 1.
Table 1
Summary of the main clinical characteristics in our cohort of 103 patients with microcephaly
Features | Number of cases |
Sex | 55 females, 48 males |
Age | Median 1.25 years old |
Microcephaly | 103/103, 100% |
Primary | 53, 51.5% |
Secondary | 36, 35.0% |
Unknown onset | 14, 13.6% |
GDD | 92, 89.3% |
ID (≥ 5y)a | 14/23, 13.5% |
ASD | 6, 5.8% |
ADHD | 12, 11.7% |
Dystonia or movement disorder | 21, 20.4% |
Epilepsy/seizures | 10, 9.7% |
Abnormal cerebral MRI | 32, 31% |
Hearing problems | 4, 3.9% |
Facial dysmorphism | 57, 55.3% |
Abnormality of the eye | 14, 13.6% |
Short stature or growth delay | 32, 31.1% |
Heart defect/Urogenital anomalies | 33, 32.0% |
Abbreviations: ASD, autism spectrum disorder; ADHD, attention deficit and hyperactivity disorder; GDD, global developmental delay; ID, intellectual disability. |
a23/103 patients were above the age of 5 years at last investigation and were evaluated for severity of ID using Wechsler intelligence scale. |
Diagnostic Yield
In 71 families (68.9%), causative SVs and clinically relevant CNVs were identified (Fig. 1). Of these, a causative SVs was detected in 55 patients (55/103,53.4%) owing to 44 genes with established neurodevelopmental phenotypes in humans (Fig. 2A). In particular, 10 of them were diagnosed with autosomal recessive, 38 with autosomal dominant, and 7 with X-linked forms (Fig. 2B). Noteworthy, segregation confirmation revealed a high percentage (72.7%) of de novo variants in our cohort (Fig. 2B). By incorporating coverage analysis to WES, a pathogenic or likely pathogenic CNV was detected in 15 families (16/103, 15.5%). In another eight families (8/103, 7.8%), eight convincing candidate genes that were not previously reported in regard to neurodevelopmental disorders or microcephaly were identified (Fig. 2A).
Expression, Annotation, And Functional Network Of Mutated Genes
We identified likely deleterious variants affecting eight different high-level candidate genes in eight (7.8%) patients without P/LP variants or VUS in established disease genes (Table 2). Four of them (PWP2, DOCK9, RHOF, KCNH3) were affected by biallelic variants, and four (CCND2, IRF2BP1, PPP1R9B, ELAVL3) by a de novo variant (Table 2and supplementary Figure S2). We first evaluated the 44 mutated monogenic known genes in our cohort for distinct temporal-spatial single cell mRNA expression pattern, by searching published single-cell transcriptomic data of the developing mouse neocortex and human brain organoids[13, 14]. The eight novel candidate genes showed expression levels similar to or higher than the expression levels of the 44 mutated known genes in mammalian developing cerebral cortex (Fig. 3A and Supplementary Figure S1).
Table 2
Cases with variants in candidate genes, along with available evidence from the KO phenotype and literature to support candidacy
Family ID | Gene | Genotype | Clinical synopsis | Gene Description | Constraint scoresa | KO phenotypeb | Gene function |
NJ1050 | PWP2 | CompHet: c.1457G > A;p.Trp486*/ c.1979G > A;p.Arg660Gln | PM, brain atrophy, GDD, short stature | Small subunit processome component; | Z = 0.32 pLI = 0 | A zebrafish mutant with a recessive lethal variant in pwp2h, exhibiting smaller eyes, a smaller, misshapen head | Ribosome biogenesis and cell cycle regulation |
NJ3099 | CCND2 | De novo: c.505C > T;p.Gln169* | PM, growth retardation, hypothyroidism, dysmorphic features | G1/S-specific cyclin-D2 | Z = 2.13 pLI = 0.99 | Mutants also show decreased cerebellar granule cell and stellate neuron populations | Regulation of the cell-cycle during G(1)/S transition |
NJ233 | DOCK9 | CompHet: c.5887C > T;p.Arg1963Trp/ c.4849T > C;p.Trp1617Arg | Microcephaly, simplified gyral pattern, epilepsy, GDD, esotropia, dysmorphic features | Dedicator of cytokinesis protein 9 | Z = 3.10 pLI = 1.00 | ND | Regulation of dendrite growth in hippocampal neurons through activation of CDC42 |
NJ2639 | RHOF | CompHet: c.142T > C;p.Tyr48His/ c.343C > A;p.Pro115Thr | Microcephaly, GDD, hypotonia | Rho-related GTP-binding protein RhoF | Z = 0.87 pLI = 0 | ND | Functions cooperatively with CDC42 and Rac to generate additional structures, increasing the diversity of actin- based morphology |
NJ463 | ELAVL3 | De novo: c.889G > A;p.Glu297Lys | PM, severe cortical dysplasia, malnutrition, feeding difficulties | ELAV-like protein 3 | Z = 3.00 pLI = 0.77 | Mice heterozygous for the allele exhibit abnormal brain wave pattern and spike wave discharge | Neuronal differentiation and maintenance |
NJ2544 | PPP1R9B | De novo: c.1610C > T;p.Ala537Val | Microcephaly, GDD, erythema, atrial septal defect with pseudoventricular aneurysm, dysmorphic features | Neurabin-2 | Z = 3.02 pLI = 1.00 | Homozygotes mutants exhibited abnormal glutamatergic synaptic transmission and increased dendritic spine density | Modulates excitatory synaptic transmission and dendritic spine morphology |
NJ316 | IRF2BP1 | De novo: c.136G > T;p.Glu46* | Neonatal-onset microcephaly, epilepsy, hypotonia and GDD | Interferon regulatory factor 2-binding protein 1 | Z = 3.34 pLI = 0.99 | ND | Transcription corepressor activity and ubiquitin protein ligase activity |
NJ3479 | KCNH3 | CompHet: c.961C > T ;p.His321Tyr/ c.2812_c.2813delCT;p.Leu938fs*73 | GDD, PM, motor developmental delay, growth retardation, dysmorphic features | Potassium Voltage-Gated Channel Subfamily H Member 3 | Z = 3.85 pLI = 1.00 | Mice homozygous for a knock-out allele exhibit neuron hyperexcitability and epilepsy | A voltage-gated potassium channel alpha subunit predominantly expressed in the forebrain; FOXG1-target gene |
a Constraint scores utilizing a larger dataset of ∼141,000 exomes and genomes from the Genome Aggregation Database (gnomAD); Z, missense Z scores; pLI, probability of being loss-of-function intolerant. |
bNeurological phenotype data from Mouse Genome Informatics or published zebrafish model; KO, knockout; ND, not described; GDD, global developmental delay; PM, primary microcephaly. |
Biological functional annotation revealed that the novel and known mutated genes in our cohort converge on transcription and transcription regulation, host-virus interaction, cell cycle and division, chromosome partition, biological rhythms and neurogenesis (Supplementary Figure S1). Network analysis using GeneMania revealed that the eight candidate genes interact with each other and with the mutated known genes in our cohort by means of co-expression (49.62%), physical interactions (28.08%), co-localization (13.79%), shared protein domains (4.05%) and genetic interactions (1.79%) (Fig. 3B).
Novel Candidates Involved In Cell Cycle And Cell Division (Pwp2 And Ccnd2)
In patient NJ1050, by trio WES we identified compound heterozygous variants (c.1457G > A;p.Trp486* and c.1979G > A;p.Arg660Gln) in PWP2 in a 2 years old boy, who had presented with primary microcephaly (at birth 30.1cm,-3.27SD, at 1 years old 42.1cm, -4.52SD), short stature, global developmental delay, cerebral paralysis and cortical atrophy on MRI (Table 2, Fig. 4A-B). The nonsense p.Trp486* variant was absent from the control database gnomAD. The second allele variant p.Arg660Gln occurred 14 times heterozygously (0/14/282772) in the gnomAD database and yielded predominantly deleterious prediction scores by three algorithms (PolyPhen-2, MutationTaster and SIFT). As shown in Fig. 4C, the Arg660 residue in PWP2 is well conserved from Homo sapiens to D. melanogaster (Fig. 4E). We transient overexpressed N-terminally Flag-tagged cDNA constructs modeling the wild-type allele and two independent PWP2 variants in HEK293 cells. As shown in Fig. 4D, the p.Trp486* variant produced a lower mount band with decreased expression. Previous studies have shown that PWP2 is localized in the area of the nucleolus involving in ribosome biogenesis and cell cycle progression[15]. Consistent with this, wildtype PWP2 localized to discrete nuclear region, while PWP2 bearing the case associated variants (p.Arg660Gln and p.Trp486*) mis-localized diffusely to the cytoplasm (Fig. 4F), indicating loss of function effects.
Patient NJ3099 showed microcephaly (30cm, -3.27SD) and hypothyroidism at birth. Her brain MRI have shown no structural abnormalities except for overall small brain size. At her most recent follow-up at 2 years of age, she presented with short stature (height 74.0cm, <-4SD), progressive microcephaly (41.6cm, <−4SD), global developmental delay and hypothyroidism (Fig. 5A-B). Trio-based WES analysis identified a de novo missense variant in CCND2 (NM_001759: c.505C > T;p.Gln169*, Fig. 5D). The variant was likely to be disease causing because (1) the variant is predicted to cause loss of function, because the resulting mRNA transcript is likely to subjected nonsense-mediated decay (Fig. 5C); (2) it is exceedingly rare and absent from the the gnomAD database; (3) the CCND2 locus exhibits high loss-of-function (LoF) intolerance (gnomAD probability for loss-of-function intolerance pLI = 0.99; (4) Homozygotes for a targeted null ccnd2 variant in mice showed decreased cerebellar granule cell and stellate neuron populations; (5) The mutant CCND2 (p.Gln169*) was subjected to proteasomal degradation in vitro compared to wild-type CCND2(Fig. 5E).
Novel Candidates (Dock9, Rhof) Related To Cdc42/rac Signaling Related Actin Cytoskeletal Organization
DOCK9 encodes a guanine nucleotide-exchange factor that plays important roles in dendrite growth in hippocampal neurons through activation of the Rho GTPases Cdc42[16]. In family NJ233, we identified compound heterozygous variants (NM_015296: c.5887C > T;p.Arg1963Trp/c.4849T > C;p.Trp1617Arg) in DOCK9 (Table 2and supplementary Figure S2). The patient showed microcephaly, facial dysmorphism including hypertelorism and a flat nasal bridge, congenital esotropia, delayed development of speech and language, motor developmental delay and epilepsy. The brain MRI revealed dilated ventricles and simplified gyral pattern. The two variants were extremely rare in the gnomAD database and yielded predominantly deleterious prediction scores by three algorithms (PolyPhen-2, MutationTaster and SIFT). Two variants both located within the DHR2 domain of the protein and well conserved from Homo sapiens to D. rerio (Supplementary Figure S2).
RHOF encodes a plasma membrane-associated small GTPase, which works cooperatively with CDC42 and Rac to generate additional structures, increasing the diversity of actin-based morphology[17, 18]. In family NJ2639, we identified compound heterozygous variants (NM_019034: c.142T > C;p.Tyr48His/c.343C > A;p.Pro115Thr) in RHOF in a patient presenting with microcephaly, global developmental delay, motor developmental delay and hypotonia (Table 2and supplementary Figure S2). The two variants were extremely rare in the gnomAD database and yielded predominantly deleterious prediction scores by three algorithms (PolyPhen-2, MutationTaster and SIFT). The two variants both located within the RHO domain of the protein and well conserved from Homo sapiens to D. rerio (Supplementary Figure S2).
Novel Candidates (Elavl3, Ppp1r9b, Kcnh3) Involved In Neuronal Differentiation And Maintenance
Neuronal Elav-like proteins are RNA-binding proteins that regulate RNA stability and alternative splicing, promote the differentiation and maturation of neurons [19]. Elavl3 is highly expressed in the adult brain. A de novo missense variant (NM_001420: c.889G > A;p.Glu297Lys) in ELAVL3 was identified in patient NJ463 with severe cortical dysplasia, thin corpus callosum, dilated lateral ventricles, simplified gyral pattern and overlapping cranial sutures (Table 2and supplementary Figure S2).
PPP1R9B (protein phosphatase 1, regulatory (inhibitory) subunit 9B) encodes spinophilin, a protein located in the heads of neuron dendritic spines[20]. Spinophillin functions as a targeting and regulatory subunit of protein phosphatase 1, an enzyme involved in postsynaptic signal integration, and has an essential modulatory function for synaptic transmission and dendritic spine morphology in mouse[20]. A highly conserved de novo missense variant (NM_032595: c.1610C > T;p.Ala537Val) was identified in patient NJ2544 who presented with microcephaly, facial dysmorphism, global developmental delay, motor developmental delay and atrial septal defect with pseudoventricular aneurysm (Table 2and supplementary Figure S2). Notably, PPP1R9B was suggested as the driven gene for the neurologic phenotype of 17q21.33 microduplication[21, 22].
KCNH3 (Potassium Voltage-Gated Channel Subfamily H Member 3) encodes a voltage-gated potassium channel alpha subunit, which is predominantly expressed in the forebrain[23]. Mice homozygous for a knock-out allele exhibit neuron hyperexcitability and epilepsy[24]. In mature neurons, KCNH3 have been identified as a FOXG1-target gene and might involve in the FOXG1 syndrome pathology[25]. In patient NJ233 who presented with global developmental delay, microcephaly growth retardation and slightly dysmorphic features was found to have compound heterozygous variants (NM_012284: c.961C > T;p.His321Tyr/c.2812_c.2813delCT;p.Leu938fs*73) in KCNH3 (Table 2and supplementary Figure S2).
Additional Mutated Cases Of Candidate Genes From 5066 Families With Ndds
When querying our data repository for additional likely disease-causing variants in the eight candidate genes in 5066 families with NDDs, we identified two cases with variants in DOCK9 and PPP1R9B, respectively (Supplementary Table 3). In particular, Individual NJ4386 with dysmorphic features, speech and language delay, autism and epilepsy carried compound heterozygous missense variant (c.898G > T;p.Asp300Tyr and c.5288G > A;p.Arg1763Gln) in DOCK9. The second individual (NJ2637) presented with microcephaly (42cm, -2SD, at 1 years old), global developmental delay and growth delay. WES identified a heterozygous frameshift variant (c.1424_c.1425insA; p.Asp475Glufs*8) in PPP1R9B, parental samples were unavailable for segregation analysis (Supplementary Table S3).