Modifier Genes in Microcephaly: A Report on WDR62, CEP63, RAD50 and PCNT Variants Exacerbating Disease Caused by Biallelic Mutations of ASPM and CENPJ

Congenital microcephaly is the clinical presentation of significantly reduced head circumference at birth. It manifests as both non-syndromic—microcephaly primary hereditary (MCPH)—and syndromic forms and shows considerable inter- and intrafamilial variability. It has been hypothesized that additional genetic variants may be responsible for this variability, but data are sparse. We have conducted deep phenotyping and genotyping of five Pakistani multiplex families with either MCPH (n = 3) or Seckel syndrome (n = 2). In addition to homozygous causal variants in ASPM or CENPJ, we discovered additional heterozygous modifier variants in WDR62, CEP63, RAD50 and PCNT—genes already known to be associated with neurological disorders. MCPH patients carrying an additional heterozygous modifier variant showed more severe phenotypic features. Likewise, the phenotype of Seckel syndrome caused by a novel CENPJ variant was aggravated to microcephalic osteodysplastic primordial dwarfism type II (MOPDII) in conjunction with an additional PCNT variant. We show that the CENPJ missense variant impairs splicing and decreases protein expression. We also observed centrosome amplification errors in patient cells, which were twofold higher in MOPDII as compared to Seckel cells. Taken together, these observations advocate for consideration of additional variants in related genes for their role in modifying the expressivity of the phenotype and need to be considered in genetic counseling and risk assessment.


Introduction
Microcephaly is characterized by a reduced occipitofrontal circumference (OFC) of <−3 standard deviation (SD) based on the age-, gender-and ethnicity-matched mean. It affects 2-3% of the population globally [1] and manifests in cognitive or neurological dysfunction that requires constant medical care. Microcephaly may be congenital (primary) or acquired postnatally (secondary). Primary microcephaly occurs as isolated (without any further features), non-syndromic (with neurological or psychiatric features but without further major morphologic or functional anomalies) and syndromic (in combination with brain malformations and/or other major morphological or functional abnormalities), such as Cohen syndrome (MIM #216550), Cornelia de Lange syndrome (MIM #122470), Nijmegen breakage syndrome (MIM #251260), Smith-Lemli-Opitz syndrome and many more [2].
Primary non-syndromic microcephaly makes a distinct subclass, termed as microcephaly primary hereditary (MCPH (MIM #251200)) or autosomal recessive primary microcephaly. It is known to be associated with 28 different genes, the latest being RRP7A, which encodes a novel component of ribosome biogenesis and was identified in a large Pakistani family [3][4][5]. MCPH has been clinically defined as a disorder of prenatal onset, non-progressive intellectual disability (ID), lack of major brain malformations or major birth defect in a non-central nervous system organ [6]. On the contrary, Seckel syndrome (MIM #210600 for SCKL1) is an allelic disorder of MCPH (shared genetic architecture) that features a short stature and characteristic facies with a prominent nose in addition to congenital microcephaly and ID. Ten genes have been reported for Seckel syndrome so far, and four of them (CENPJ, CEP152, CDK5RAP2 and CEP63) are also involved in MCPH [7]. Microcephalic osteodysplastic primordial dwarfism (MOPD) II manifests features overlapping with Seckel syndrome but these patients feature a characteristic skeletal dysplasia, too. MOPD is known to be caused by homozygous loss-of-function mutations in a particular gene, PCNT [8]. For classical MCPH and Seckel syndrome, both intra-and interfamilial clinical variability (even for the same variant) have been frequently observed [9,10], which makes it difficult to infer a genotype-phenotype correlation. Although other genomic loci had been considered to contribute to the variability, they were less tangible through traditional genetic approaches.
With the latest trends and down pricing of next-generation sequencing technologies, it is now possible to identify variants in other genes that are not pathogenic on their own but nevertheless influence the phenotypic outcome of the primary causal variants, called genetic modifiers [11]. Genetic modifiers contribute towards phenotypic variability and penetrance and are thus important in precise diagnostic, prognostic, therapeutic and overall patient management strategies [12]. Such variants or genetic modifiers are being increasingly observed in neurodevelopmental disorders [13][14][15][16] but they have been rarely reported for microcephaly and associated syndromes [17]. Interestingly, they have a pronounced appreciation in mouse models of microcephaly [18,19].
The application of massively parallel sequencing of multiplex families offers a unique opportunity to identify the modifying variants in a similar genetic background. Here, we propose that heterozygous variants of WDR62, CEP63, RAD50 and PCNT contribute to additional neurological and extra-neurological abnormalities in affected siblings of the families manifesting MCPH or Seckel syndrome.

Clinical Manifestations
We performed deep phenotyping and genotyping of five Pakistani multiplex families presenting 28 patients segregating MCPH (three families) or Seckel syndrome (two families) (Table 1, Figure 1A,B). We recruited these families from different areas of Punjab, Pakistan, with prior written consents from the parents of all the families.

Next-Generation Sequencing
To reveal the causative variants, we selected the affected siblings showing the most severe phenotypic abnormalities for targeted sequencing of a microcephaly gene panel (family 2: V-1, family 3: IV-1, family 4: V-6 and VI-2) or whole-exome sequencing (family 1: V-9 and family 5: IV-1). The complete list of genes included into the panel and the process of targeted sequencing has been published elsewhere [10]. For whole-exome sequencing (WES), we used the Agilent (Santa Clara, CA, USA) version 6 enrichment kit and Illumina HiSeq 4000 sequencing system (paired end reads, 2 × 75bp). The detailed procedure for WES has been described elsewhere [10]. Variant calling and interpretation was performed with the help of our in-house VARBANK database and analysis platform (http://varbank.ccg.uni-koeln.de (accessed on 9 May 2021)).

Linkage Analysis
For family 4, we genotyped the DNA of the four affected members (V-6, V-8, VI-2 and VI-3), indicated by asterisks in Figure S1B, using the HumanCoreExome 24 v.1.1 BeadArray (Illumina; San Diego, CA, USA) according to the manufacturer's instructions. Linkage analysis was performed assuming autosomal recessive inheritance, full penetrance, consanguinity and an allelic frequency of 0.0001. The procedure of data handling and evaluation has been described earlier [20]. The detailed procedure of the statistical analysis is given in Supplementary Materials. To fine-map the PCNT locus on chromosome 21, we genotyped six neighboring microsatellite markers with the DNA of ten family members (indicated by asterisks in Figure S1C). HaploPainter v.1.043 was used to construct the haplotypes [21].

Sanger Sequencing
Co-segregation studies were performed by Sanger sequencing, including all family members for which a DNA sample was available.

Clinical Manifestations
We performed deep phenotyping and genotyping of five Pakistani multiplex families presenting 28 patients segregating MCPH (three families) or Seckel syndrome (two families) (Table 1, Figure 1A,B). We recruited these families from different areas of Punjab, Pakistan, with prior written consents from the parents of all the families.

Next-Generation Sequencing
To reveal the causative variants, we selected the affected siblings showing t severe phenotypic abnormalities for targeted sequencing of a microcephaly gen (family 2: V-1, family 3: IV-1, family 4: V-6 and VI-2) or whole-exome sequencing 1: V-9 and family 5: IV-1). The complete list of genes included into the panel and cess of targeted sequencing has been published elsewhere [10]. For whole-exome s ing (WES), we used the Agilent (Santa Clara, CA, USA) version 6 enrichment kit mina HiSeq 4000 sequencing system (paired end reads, 2 x75bp). The detailed pr for WES has been described elsewhere [10]. Variant calling and interpretation w formed with the help of our in-house VARBANK database and analysis p (http://varbank.ccg.uni-koeln.de (accessed on 9 May 2021)).

Linkage Analysis
For family 4, we genotyped the DNA of the four affected members (V-6, V and VI-3), indicated by asterisks in Figure S1B, using the HumanCoreExome BeadArray (Illumina; San Diego, CA, USA) according to the manufacturer's instr Linkage analysis was performed assuming autosomal recessive inheritance, fu trance, consanguinity and an allelic frequency of 0.0001. The procedure of data h and evaluation has been described earlier [20]. The detailed procedure of the s analysis is given in Supplementary Materials. To fine-map the PCNT locus on some 21, we genotyped six neighboring microsatellite markers with the DNA of te members (indicated by asterisks in Figure S1C). HaploPainter v.1.043 was used struct the haplotypes [21].

Sanger Sequencing
Co-segregation studies were performed by Sanger sequencing, including a

In Silico Analyses of Identified Variants
Pathogenicity of the identified variants were predicted by several in silico tools like the ACMG classification system, Mutation Taster To investigate the conservation status of the mutated sites, reference sequences were retrieved either from UniProtKB (https://www.uniprot.org/, (accessed on 9 May 2021)) or NCBI (https:// www.ncbi.nlm.nih.gov/ (accessed on 9 May 2021)) and were aligned by Clustal Omega (EMBL-EBI, Wellcome Trust Genome Campus, Hinxton, Cambridge, UK).

Reverse Transcription PCR
To explore the possible consequences of the CENPJ missense variant on splicing accuracy, RNA was extracted from the patient's blood using the PAXgene blood RNA system (QIAGEN, Hilden, Germany) and subsequently converted to cDNA as described previously [22]. In brief, RT-PCR was performed using forward 5 -AGCCACTTGAACCACTGAAC-3 and reverse 5 -CAGTCTGGTCAGGAAACGTG-3 primers to amplify the relevant portion of CENPJ. Amplicons were resolved on a 2% agarose gel along with a 1 kb plus DNA ladder (10787018, Thermo Fisher Scientific, Waltham, MA, USA) as the size marker. Each band was Sanger sequenced separately to determine the effects on splicing.

Immunocytochemistry
To investigate the effects of the CENPJ missense variant identified in families 4 and 5, primary dermal fibroblasts were cultured from skin biopsies obtained from the affected individuals (V-8 and VI-2) of family 4, as described elsewhere [23]. For immunofluorescence, primary fibroblasts were grown on 12 mm coverslips with a maximum confluency of 70%. The fixation of cells was performed using ice-cold methanol for 10 min at −20 • C. After 10 min, cells were incubated with PBS (1×) followed by the blocking buffer PBG (solution composition reported previously [20]) for 15 min. For immunofluorescence, primary antibodies-mouse anti-CENPJ, 1:25 [24], and rabbit anti PCNT, 1:1000 (Abcam, ab4448)-were applied to the cells and incubated at 4 • C overnight. Alexa Fluor 488 donkey anti-mouse IgG (Invitrogen, A21202) and Alexa Fluor 568 goat anti-rabbit IgG (Invitrogen, A11011) were used as secondary antibodies, each at a dilution of 1:1000, along with 4 ,6-diamidin-2 -phenylindol (DAPI) (MilliporeSigma, Saint Louis, MO, USA, D9564) for staining of DNA. The slides were observed under a confocal microscope (Leica Microsystems, Wetzlar Germany, LSM TCS SP5) to capture images.

Results
Family 1 is a very large consanguineous family with eight affected members having microcephaly of −10 to −15 SD accompanied by mild to severe ID ( Figure 1A,B and Table 1). Sequencing data analysis of family 1 revealed a nonsense mutation, NM_018136. 4:c. 9601C>T; p.(Gln3201*), of ASPM (Table 1). A CADD score of 41 was obtained for this variant and it was placed in the category of pathogenic variants (PVS1, PM2, PP3) by the ACMG classification system. We found it neither in gnomAD nor in dbSNP build 153 (Table 2). This variant was segregating, as expected, a recessive inheritance pattern ( Figure 1A and Figure S2). Additional phenotypic presentation of the contracted joints, drooling and short stature, which were observed only in a few affected members of this family, prompted us to look for additional potentially disease-causing variants. As a result, we found a heterozygous missense variant, NM_001083961.1:c.3316A>G;p.(Ser1106Gly), of WDR62 (Table 1). Interestingly, this variant is not catalogued in Iranome and the Greater Middle Eastern Variome. It is, however, present in gnomAD (with two heterozygous alleles) and dbSNP build 153 (rs1389367700) ( Table 2). Based on this ultralow allele frequency, we investigated this variant further. Sanger sequencing revealed that this variant was found only in patients showing additional features of joint contractures, short stature and drooling ( Figure 1A and Figure S2, Table 1). The variant p.(Ser1106Gly) does not reside in any of the 15 WD repeats of WDR62 yet shows its deleterious nature by PhD-SNP, SIFT, SNAP, Meta SNP and MuPro (Table 2). According to the ACMG classification system, it is considered as a variant of uncertain significance (VUS). This position is not conserved in mammals in general ( Figure S3A); however, it is strictly conserved among primates ( Figure S3B). Notably, this variant could be detrimental and impair the function of WDR62 because Ser1106 is reported to be phosphorylated during mitosis [26].   Family 2 is a five-generation consanguineous family with five affected members, two of them deceased for unknown reason, manifesting microcephaly of −11 to −15 SD, moderate ID and aggressive behavior ( Figure 1A,B and Table 1). Sequence data analysis of this family revealed a frameshift variant, NM_018136. 4:c.719_720delCT;p.(Ser240Cysfs*16), in ASPM. As presumed, all the tested patients were homozygous for this variant ( Figure  1A and Figure S2, Table 1). Because clinical manifestation of the proband subjected to gene panel sequencing was severe (Table 1), the possibility of additional variants was explored. We found a missense variant, NM_025180. 3:c.1261A>T;p.(Thr421Ser), of CEP63 that was subsequently confirmed in his younger sister (V-2), manifesting a similarly severe clinical presentation ( Figure 1A and Table 1). Their asymptomatic brother (V-3) was a heterozygous carrier of both the aforementioned variants in ASPM and CEP63. The CEP63 variant was inherited from their asymptomatic father (IV-6), whereas the mother was homozygous for the wild-type allele ( Figure 1A and Figure S2). Interestingly, investigation of a second loop of the same family showed absence of the CEP63 variant in affected (IV-1) as well as in phenotypically normal members (IV-3 and IV-4), even though their mother was heterozygous for this variant ( Figure 1A). The variant was predicted to be disease-causing, deleterious and probably damaging, and decreasing the protein stability by the in silico tools Mutation Taster (score = 58), Provean (score = −2.830), Polyphen-2 (score = 1.000) and MuPro (DDG = −0.54948253), respectively; however, it was classified as VUS (PM2, BP1) by the ACMG classification system ( Table 2). The pathogenic potential of this variant was further strengthened by its absence in databases of genomic variants like gnomAD, GME variome, Iranome and dbSNP build 153 (Table 1). Furthermore, multiple alignments of Cep63 peptides, spanning the mutant residue, revealed that threonine at position 421 of the human protein is highly conserved throughout the vertebrate lineage ( Figure S3C).
Family 3 is consanguineous, having three affected female siblings showing HC of -9 to −13 SD with phenotypic variability in the manifestation of seizures, ID and behavior ( Figure 1A,B and Table 1). Variant filtration identified a nonsense mutation, NM_018136.4:c.9492T>G;p.(Tyr3164*), of ASPM, homozygous in all affected members ( Figure 1A and Figure S2, Table 1). Two affected members (IV-1 and IV-2) of this family presented severe ID accompanied by aggressive behavior compared to their younger sister (IV-3), who showed mild ID and normal behavior (Table 1). Considering the heterogeneity of the clinical features, we extended the search for additional disease-causing variant(s) in other genes. To this end, we identified a heterozygous missense variant, NM_005732. 3:c.3643C>G;p.(Leu1215Val), of RAD50 inherited from their asymptomatic mother (III-2) ( Figure 1A and Figure S2). This variant is not listed in any of the aforementioned public databases of genomic variations and is predicted to be pathogenic by Mutation Taster, Polphen-2, PANTHER, PhD-SNP, SIFT, SNAP, Meta SNP, MuPro, MetaDome and CADD (score = 23.5), albeit likely benign (PM1, PM2, BP1, BP4) by the ACMG (Table 2). We considered it a potential modifier because this variant was only carried by affected members with severe phenotypic manifestations, IV-1 and IV-2, whereas the affected member IV-3 with a milder phenotype did not carry this variant ( Figure 1A,B). Furthermore, the mutated residue is highly conserved within a large group of organisms, including vertebrates, insects, plants and even yeast-only worms contain leucine at this position ( Figure S3D).
Family 4 is a multigenerational family with nine affected members born to two consanguineous couples and exhibiting major phenotypic differences among patients of both loops ( Figure 1A,B). Affected siblings in loop 1 (right side) showed typical features of Seckel syndrome whereas those of loop 2 (left side) showed more severe clinical features reminiscent of MOPDII ( Figure 1A,B and Table 1). Affected members of loop 1 showed moderate ID, HC of −7 to −8.5 SD, short stature and protruding noses. In addition, two of the three affected individuals of loop 1 showed hypopigmentation (Table 1 and Figure 1B, white arrowheads). Affected members of loop 2 showed microcephaly of HC −10 to −13 SD, short stature, mild ID, excessive drooling, speech impairment, clinodactyly of only toes and brachydactyly of both fingers and toes ( Figures 1B and 2 Asp1196Asn)) in both patients that were later confirmed in all affected siblings of both loops by Sanger sequencing (Figure 1A and Figure S2). The involvement of the missense variant of CENPJ in the disease etiology was further corroborated by linkage analysis that highlighted a region on chromosome 13 with a maximum LOD score of 4.3 ( Figure S1A). The homozygous segment of 2.63 Mb was flanked by the markers rs7981441 (24,075,006bp; GRCh38.p12) and rs9551309 (26,712,326bp; GRCh38.p12), thus harboring CENPJ ( Figure S1B).
Screening for additional variants revealed a previously reported heterozygous nonsense variant of PCNT (NM_006031.5;c.5767C>T;p.Arg1923*) in the patients of loop 2, suggesting a modifier role for PCNT ( Figure 1A and Figure S2, and Table 2). Sanger sequencing demonstrated that the PCNT variant was inherited from the mother (V-2) and carried by all the affected members of this loop, whereas this variant was absent from all individuals of loop 1 ( Figure 1A). Additionally, genotyping of microsatellite markers in selected family members revealed that the PCNT variant haplotype (colored in red, Figure S1C) was introduced by the unrelated male individual IV-2 and absent from all tested family members of loop 1. The PCNT variant obtained a very high CADD score of 42 and was predicted to be pathogenic (PVS1, PM2, PP3, PP5) by ACMG criteria and disease-causing by Mutation Taster (Table 2).
We considered the missense mutation, NM_018451.4:c.3586G>A;p.(Asp1196Asn), of CENPJ causative of Seckel syndrome in family 4 because it was not recorded in any of the genomic variation databases and predicted to be pathogenic with a CADD score of 28.1, likely pathogenic (PM1, PM2, PP3, PP5) by the ACMG classification system and disease-causing by Mutation Taster, PROVEAN, Polphen-2, PhD-SNP, SIFT, SNAP, Meta SNP, MuPro and MetaDome (Table 2). Furthermore, the amino acid position 1196 of CENP-J is strictly conserved from fungi to vertebrates, advocating for an indispensable role of this aspartic acid residue ( Figure S3E). RT-PCR analysis of a fragment spanning the variant and neighboring exons using patient and control mRNA revealed that the presumable missense mutation c.3586G>A (p.(Asp1196Asn)) of CENPJ at least partially also results in aberrant splicing by activating a cryptic splice donor site located within exon 14. We noted two bands of reduced intensity as compared to the single band of expected size (372 bp) seen in the control ( Figure 3A). One band of the mutant sample was of the same size as that of the control while the second band was considerably smaller (~294 bp) ( Figure 3A). Sanger sequencing of the wild-type PCR product resulted in the expected transcript sequence ranging from exon 13 to exon 16 ( Figure 3B   These data intrigued us to follow the consequences of mutant CENP-J at the cellular level. Immunofluorescence analysis in primary fibroblasts of wild-type origin revealed bright staining of CENP-J at the centrosome, which is marked by pericentrin ( Figure 3D, white arrow). Contrarily, primary fibroblasts derived from two different patients of family 4-one carrying only the CENPJ mutation (V-8) and another carrying the CENPJ and the PCNT mutation (VI-2)-showed faint staining of CENP-J, marked by white arrow heads ( Figure 3D). These findings were corroborated by immunoblotting where protein lysates of the patient-derived primary fibroblasts showed faint immunoreactivity of CENP-J as compared to the wild type ( Figure 3E, upper panel, and Figure S4). Equal concentrations of lysates were loaded as evident by the immunoreactivity of α-tubulin ( Figure 3E, lower panel, and Figure S4). In addition to the compromised stability of CENP-J, we also observed a significant number of patient interphase cells with supernumerary centrosomes ( Figure 3D). On average, 7.33% of cells derived from the patient carrying only the CENPJ mutation showed supernumerary centrosomes, whereas more than twice as many (15.66%) of the cells derived from the patient carrying both mutations showed them ( Figure 3F). Notably, only 1.66% of the wild-type cells showed extra centrosomes ( Figure 3F). Based on this data, we speculate that the PCNT variant exacerbates cellular anomalies caused by CENP-J dysfunction, consequently worsening the clinical phenotype in VI-2.
Family 5 is a consanguineous family with two affected male siblings showing typical features of Seckel syndrome, such as a prominent nose, HC-11 to -12 SD and short stature ( Figure 1A,B and Table 1). Whole-exome sequencing, performed in one affected member, revealed the same CENPJ missense variant (NM_018451.4:c.3586G>A;p.(Asp1196Asn)) that had been identified in family 4 ( Figure 1A,B and Figure S2, Tables 1 and 2). The identification of the same CENPJ variant in a second unrelated family with Seckel syndrome provides strong evidence for its causal role in the disease etiology. Interestingly, exome data analysis did not show any variation of PCNT in family 5.

Discussion
Here, we have provided evidence that heterozygous missense variants of WDR62, CEP63 and RAD50 aggravate the phenotype of MCPH and a PCNT nonsense variant exacerbates the severity of Seckel syndrome features. Multiple lines of evidence advocate for their potential to act as genetic modifiers; this includes absence of any record or report of the rare alleles in the literature or databases of genomic variations, high pathogenicity scores as calculated by several in silico tools, high conservation of the affected amino acid positions, and last but not least, the known involvement of these genes in the neurodevelopmental disorders, which are discussed below in a continuum of the respective modifiers.
The genes identified with modifier variants in this study have previously been found dysfunctional in non-syndromic primary microcephaly and its syndromic forms. Mutations in WDR62 were reported to cause primary microcephaly with or without severe brain malformations (lissencephaly and pachygyria) [27,28]. A nonsense mutation in CEP63 was reported to cause Seckel syndrome in a Pakistani family, with three affected members showing microcephaly and reduced height [29]. For this reason, short stature in our patients could be explained by the combined effect of this heterozygous modifying variation of CEP63 together with the homozygous 2 bp deletion in ASPM. Nonsense and frameshift variants of RAD50 were previously reported in families afflicted with Nijmegen breakage syndrome-like disorder (MIM #613078). Patients with this disorder feature microcephaly, ID, a characteristic 'bird-like' face and short stature [30]. Interestingly, mutations in PCNT were found to be associated with MOPDII [8,31]. Notably, many of the encoded proteins interact with each other and form a crucial complex necessary for centriole biogenesis; e.g., Cep63, Asp homolog, also known as ASPM, and WDR62 recruit CENP-J at the site of centriole biogenesis [7]. Therefore, it could be speculated that modifying variants may affect the structure and/or function of the complex, thereby resulting in a more severe phenotype. A critical role of the WDR62 mutation p.(Ser1106Gly), identified in family 1, is highly convincing as Ser1106 has been reported to be a key site of phosphorylation in mitosis [26]. Additionally, the conservation constraint on Ser1106 in primates could also implicate that serine at this position might be subject to positive selection of this gene attributing some primate-specific function.
hRAD50 is a well-known component of the hRAD50 MRN (Mre11, Rad50 and Nbs1) complex with a well-established role in double-strand break (DSB) repair [32]. Our patients, carrying a variation of RAD50, showed rare and irregular episodes of seizures, aggressive behavior and severe ID. None of these features were observed in the affected sibling (IV-3) who did not carry the RAD50 variation (Table 1). Furthermore, the variant p.(Leu1215Val) resides within an Ala/Asp-rich domain (DA-box) highly conserved across vertebrates.
CENP-J mutation p.(Asp1196Asn) is located in its T complex protein 10 (TCP) domain (residues 1159-1337), which interacts with STIL [33,34] and plays a crucial role during centrosome biogenesis by mediating the tethering of pericentriolar material [34]. Another mutation of CENPJ, NM_018451.4:c.3704A>T;p.(Glu1235Val), reported in a Pakistani MCPH patient [35], lies close to our mutation and has been shown to impair this interaction, thus inhibiting centriole biogenesis [33]. Therefore, we speculate that the mutation identified by us may also impair the interaction with STIL and compromise the tethering ability of CENP-J in general. Notably mutant CENP-J encoded by the second mutant transcript lacks 26 amino acids (p.(Val1181_Val1206del)), spanning a region of the TCP 10 domain and could also show detrimental effects on the interaction with STIL. Furthermore, aspartic acid at position 1196 is a highly conserved residue and, due to its negative charge, most likely participates in protein/protein interaction.
Interestingly, homozygous pericentrin mutation p.Arg1923* was originally described as a cause of MOPDII [8]. If the mutant protein would be formed by escaping from nonsense mediated mRNA decay, it would result in a loss of two helical strands, loss of nek2 interaction (residues 2983-3246) and loss of calmodulin-binding (residues 3195-3208), which should result in severe consequences for the protein function in centrosome arrangement. Therefore, we have seen severe centrosomal defects in cells carrying mutations in CENPJ and PCNT. On a background of biallelic CENPJ mutations, haploinsufficiency of PCNT may become overt and result in a more severe clinical picture, as seen in some of our patients of family 4 ( Figure 1B and Table 1).
To the best of our knowledge, there has been only one publication reporting the contribution of a genetic modifier to the phenotype of MCPH. In addition to a homozygous stop mutation in WDR62, NM_001083961.1:c.1605dupT;p.(Glu536*), originally reported as c.1605_1606insT, the MCPH patient also carried a duplication of the chromosomal segment 17q25-qter and a missense mutation, c.3361T>G;p.Phe1121Val, of TBCD on the non-duplicated allele. More severe phenotypic manifestations were observed in this patient as compared to one carrying only the homozygous WDR62 stop mutation [17], which was attributed to the modifying effects of TBCD. In case of Seckel syndrome, only a digenic inheritance pattern has been reported, with heterozygous mutations of CDK5RAP2 and CEP152 [36]. Another report showed apparently digenic triallelic inheritance in patients manifesting MCPH, with the following combinations of genes: WDR62/CDK5RAP2, ASPM/WDR62, and WDR62/CEP135. Unfortunately, the authors did not present any data for the presumptive modifying effects of the additional variants on the phenotype. Furthermore, they did not find any modifying effects of wdr62 or aspm ablations in zebrafish upon knocking out casc5/knl1; merely, the quadriallelic ablation of wdr62 and aspm showed primary microcephaly in zebrafish [37]. Hence, these data lay a foundation for future studies to investigate the modifying variants in MCPH cases. Importantly, the phenotypic severity, corroborated by the sub-cellular anomalies confirmed by our experiments, models an efficient strategy for future studies for validation of the modifying variants at the functional level.

Conclusions
The data presented in our study are unique for the fact that our multiplex pedigrees show intrafamilial phenotypic and genotypic differences, thus providing a solid basis to investigate the role of potentially modifying variants. We have provided compelling evidence of the pathogenic nature of these variants, yet further functional analyses will provide deeper insights into the pathomechanisms involving more than one mutated gene. Based on our data, we recommend clinicians and researchers to subject the DNA of the most severely affected members of larger families to comprehensive genomic analyses, which may help to identify supplementary disease-modifying gene variants in addition to the disease-causing ones. Institutional Review Board Statement: The study was designed following rules described in the Declaration of Helsinki and approved by the ethics committee of the National Institute for Biotechnology and Genetic Engineering (NIBGE) and Government College University, Faisalabad, Pakistan.
Informed Consent Statement: Informed written informed consents for publication was obtained from parents of all the families. Data Availability Statement: Data concerning this study is available in the article and Supplementary Materials.

Conflicts of Interest:
The authors declare no conflict of interest.