Eight fetuses from five unrelated families diagnosed with ciliopathies were collected. Informed consents for research investigations were obtained from the relatives of the fetuses. The research protocol was approved by the local ethics committee of Jiangsu Huai’an Maternity and Child Healthcare Hospital (2019036).

The whole exomes were captured by using Agilent’s SureSelect Whole Exome Gene Detection Kit. High-throughput sequencing was performed by using the Novoseq sequencer from Illumina. The obtained sequence was aligned on the human genome GRCh37/hg19 reference sequence by BWA (Burrows-Wheeler Aligner) software. A BAM (binary sequence alignment map format) file was produced via Picard software. GATK4 (Genome Analysis Toolkit) Realigner Target Creator software and Haplotype Caller software were used to adjust the sequence, extract variants, and generate VCF (Variant Call Format) files. The Annovar software was used to filter and annotate the variant.

All nonsense, frameshift, and canonical splice site variants were considered to be deleterious. The pathogenic potential of missense variants was predicted by PolyPhen2, SIFT, PROVEAN, and Mutation Taster. The frequency of putative variants was obtained from the Human Gene Mutation Database (HGMD), Genome Aggregation Database (gnomAD), and the 1000 Genomes (1000G) database. Conservation of mutated amino acid residues in different species was compared by UCSC. SpliceAI was used to evaluate a destroyed splice site. The deleteriousness of variants was assessed according to American College of Medical Genetics (ACMG) standards and guidelines.

Sanger sequencing was performed to confirm suspected variant segregation within probands’ family, and the authenticity of variants identified by WES. Primer 5 was used for primer designs. Target DNA of the fetus and its parents was amplified by PCR. Sanger sequencing results were compared with standard sequence in GenBank by SeqMan software.

Family 1 was a consanguineous marriage with a healthy girl and five adverse obstetric outcomes. In addition to two early miscarriages, the third time was due to a widened posterior fossa and vermis hypoplasia at the 27th week of pregnancy without genetic detection. Case 1 was the fifth pregnancy, whose prenatal ultrasound in 18⁺⁴th week of pregnancy showed absence of cerebellar vermis, dandy-walker malformation, hydrocephalus, dextrocardia, situs inversus, ventricular septal defect (0.3 cm), and double outlet of fetal right ventricle with pulmonary artery stenosis (Table 1). Pregnancy was terminated at the 23rd week and no obvious abnormality in appearance was observed (Figure 1A). The sixth pregnancy (case 2) showed recurrence of malformations by prenatal ultrasound at the 20th week of fetal development including occipital encephalocele (1.3 × 0.8 cm) (Figure 2A) and intra-uterine growth restriction (Table 1) (Figures 1B,C).

WES revealed case 1 with a novel homozygous frame shift variant of c.7939delC (p.H2647IfsTer51) in CPLANE1 (NM_023073.3). Sanger sequencing confirmed homozygosity in the proband and the heterozygosity in each parent (Table 1). Case 2 was also confirmed with same variant in case 1. The c.7939delC was not present in HGMD, gnomAD, or 1000G databases.

Case 3 and case 4 were affected siblings conceived of unrelated healthy parents with two early unexplained miscarriages. Prenatal ultrasound in 23⁺⁴rd week of case 3 showed separation of left lateral ventricles (0.9 cm), abnormality of the cerebellum, low amniotic fluid volume (amniotic fluid index 1.5 cm, maximum depth 1.0 cm), pericardial effusion (0.3 cm), and enlarged echogenic kidneys (left kidney 5.7 × 2.8 cm, right kidney 5.4 × 2.8 cm) with renal cysts (Table 1). Case 4 was found with congenital balkes pouch, cystic kidneys, polydactyly, and occipital encephalocele (Table 1). The pregnancy was terminated at the 24th week of gestation. Because prenatal detection and labor induction were performed in another hospital, we only obtained DNA of the two cases without ultrasound images and postnatal fetus examination.

WES was selected by Family 2 and the result showed that case 4 had compound heterozygous variants in TMEM67 gene (NM_153704.6): a missense variant at c.1175C > G (p.P392R) and a synonymous variant at c.2439G > T (p.A813A), which were inherited from the parents, respectively (Table 1). The c.1175C > G, not reported previously, was predicted as damaging by PolyPhen2, SIFT, PROVEAN, and Mutation Taster. Total population frequency of this variant is 3.23185e-05. The c. 2439G > T variant is located at the last base of exon 23 and was classified as a disease-causing variant in HGMD. It might abolish the donor splice site following in silico analysis via SpliceAI. The predicted score of a donor site loss via SpliceAI was 0.64. After this, the DNA of case 3 was subjected to molecular analysis by Sanger sequencing. The result revealed the c.1175C > G and c.2439G > T variants in the TMEM67 gene in case 3.

Case 5 was the first pregnancy of unrelated healthy parents. Prenatal diagnosis was performed in the 20th week of pregnancy due to a 12 Mb deletion in chromosome 1 as indicated by noninvasive prenatal testing (NIPT). The Affymetrix CytoScan 750K SNP array was used for chromosomal microarray analysis (CMA) and the result showed a region of homozygosity in 1p36.33-p36.13 (arr[hg19] 1p36.33p36.13 (888,658_18,337,268) x2 hmz) involving no established imprinted genes. In 26⁺⁴th week, missing nasal bone, delayed growth and development, and increased echogenicity in both renal cortexes were indicated by ultrasound (Figure 2B and Table 1). Prenatal trio-WES was chosen for further detection. A homozygous variant of c.3730C > T (p.Q1244X) (chr1:5,925,248) in NPHP4 (NM_001184.3) (Table 1) was detected. This variant, occurring in the region of homozygosity, was inherited from the father. These results indicated that a segmental uniparental disomy at 1p36.33-p36.13 inherited from the father caused the autosomal recessive NPH.

Case 6 and case 7 were affected sibling fetuses from a non-consanguineous family with a healthy 6-year-old daughter. Ultrasound imaging of case 6 revealed abdominal circumference (10.3 cm), a femur length of 1.3 cm (<5th percentile), a humerus length of 1.0 cm (<5th percentile), and polydactyly. The pregnancy was subsequently terminated at the 18th week of gestation (Figure 1D). Fetal X-ray (Figure 2C) showed bilateral shortened curved femora and disproportionately shortened tibiae. Metaphyseal flaring of femora, tibiae, and fibulae were indicated. Ultrasound imaging of case 7 at the 28th week of gestation from this family exhibited similar sonographic features including small short femur (1.2 cm), narrow chest, light band separation in double kidney collection system (0.8 and 0.9 cm), peritoneal effusion (deepest 0.5 cm), and polydactyly. We only obtained DNA of case 7 but failed to perform postnatal fetus observation.

WES of case 6 revealed the compound heterozygous variants of DYNC2H1 (NM_001377.3): c.152T > G (p.L51R) in exon 1 and c.988C > T (p.R330C) in exon 6. Sanger sequencing confirmed that the c.152T > G was present in the mother, and the c.988C > T in the father (Table 1). Case 7 was also confirmed with variants as case 6 through Sanger sequencing (Table 1). The c.988C > T has been reported as a pathogenic variant (El Hokayem et al., 2012; Schmidts et al., 2013). The c.152T > G was not present in HGMD, gnomAD, or 1000G databases, and was predicted as deleterious following in silico analysis. In addition, the c.152T > G is localized to the region of nucleotide-binding site of the DYNC2H1 gene, which is highly conserved in mouse, rat, and dog orthologs of DYNC2H1 gene.

Prenatal ultrasound of case 8 (the first fetus from non-relative parents) in the 12th week of pregnancy showed abnormal NT (4.4 mm). Recheck in the 17th week showed multiple malformations with large abdominal circumference (12.1 cm), ventricular septal defect (0.08 cm), left ventricular dysplasia, valve thickening, and transposition of the great arteries (Table 1). Besides, increased bowel and kidney echo, and bone dysplasia (femur = 1.3 cm, humerus = 0.9 cm, narrow chest) (Figure 2D) were detected. Examination of the aborted fetus showed low-set ears, polydactyly, a swollen abdomen, and short limbs (Figures 1E,F). Compound heterozygous variants of c.4267C > T (p.R1423C) and c.7858C > T (p.R2620X) in DYNC2H1 (NM_001377.3) were identified. We confirmed a heterozygosity in each parent by Sanger sequencing (Table 1). The c.4267C > T was a pathogenic variant in short-rib polydactyly syndromes (SRPS) (Zhang et al., 2018). The c.7858C > T was a novel nonsense variant causing a truncated protein.

Since CPLANE1 and TMEM67 variants show significant clinical overlap, genotype-phenotype relationship analysis between them could help discriminate related disorders.

CPLANE1 was identified as causative of OFD6 (orofaciodigital syndrome VI) and JBS. More than 60 pathogenic variants in the CPLANE1 gene have been identified (Srour et al., 2012; Lopez et al., 2014; Vilboux et al., 2017; Bonnard et al., 2018) and 66% of cases carried biallelic truncating variants (Bonnard et al., 2018). Penetrance of common organ systems involving brain, skeleton, and kidney diseases in CPLANE1 patients were compared (Table 2) (Lopez et al., 2014; Vilboux et al., 2017; Bonnard et al., 2018). MTS was nearly 100% detected in both JBS and OFD6. Polydactyly was with higher penetrance (100%) in OFD6 than JBS. The penetrance of occipital encephalocele in both JBS and OFD6 was estimated to be about 20%. However, cystic kidneys, short limbs, and congenital heart disease (CHD) were rarely observed.

TMEM67 was identified as causative of JBS and MKS. A genotype-phenotype correlation analysis based on a literature review shows that combination of two truncating variants in TMEM67 gene is more common in lethal MKS than milder JBS (Iannicelli et al., 2010; Szymanska et al., 2012; Bachmann-Gagescu et al., 2015; Vilboux et al., 2017), which suggests that severity of variant changes disease outcomes. Thus, it is reasonable to presume that missense variant of TMEM67 identified in MKS might bring more severe damage to protein function. There were about 58 (likely) pathogenic missense variants in TMEM67 identified thus far including 37 in JBS, 15 in MKS, and 6 in both (Baala et al., 2007; Consugar et al., 2007; Khaddour et al., 2007; Brancati et al., 2009; Doherty et al., 2010; Iannicelli et al., 2010; Suzuki et al., 2016; Vilboux et al., 2017; Radhakrishnan et al., 2019) (Supplemental Table S1). Analysis of the topology of human TMEM67 suggests the presence of a signal peptide (1–36 aa), a cysteine-rich region (50–187 aa), an extracellular area with unknown structure (188–526 aa), seven transmembrane regions (527–967 aa), and a short cytoplasmic tail (968–995 aa) (Figure 3) (Smith et al., 2006). Pathogenic missense variants of TMEM67 mainly fall in the extracellular cysteine rich region, extracellular area with unknown structure, and the seven transmembrane regions (Figure 3). No pathogenic missense variant in the signal peptide and cytoplasmic tail has been reported (Figure 3). The comparison of their distribution along the gene in MKS versus JBS showed no obvious difference.

Two studies with larger TMEM67 gene mutated cases were reviewed to analyze the characteristic of clinical phenotypes (Table 2) (Brancati et al., 2009; Szymanska et al., 2012; Vilboux et al., 2017). Incidence of liver and kidney diseases caused by TMEM67 variants was high in both JBS and MKS. The prevalence of occipital encephalocele in MKS was higher than in JBS.