Two unrelated cases were recruited between March 2020 and April 2021 at the Prenatal Diagnosis Center, Shijiazhuang Obstetrics and Gynecology Hospital. A comprehensive prenatal ultrasonic examination was conducted on the patients. We carried out a thorough clinical survey. Subsequently, the peripheral blood samples of trio family members in the two pedigrees were collected for the following genetic detection. The studies involving human participants were reviewed and approved by The Ethics Committee of Shijiazhuang Obstetrics and Gynecology Hospital.

Amniocentesis was performed to obtain the fetal cell samples, along with 3 ml of peripheral blood collected from the parents using BD Vacutainer™ tubes (BD Biosciences, New Jersey, United States). Genomic DNA was extracted with the QIAamp DNA Blood Mini-Kit (Qiagen Sciences, New York, United States), and the DNA quality was validated by 1% agarose gels and the Qubit^® 2.0 Flurometer (Life Technologies, CA, United States).

WES was performed by MyGenomics, Inc. (Changping, Beijing, China) as described in our previous study (Zhang et al., 2021). Briefly, the enrichment of the exonic region sequences was conducted by the Sure Select Human Exon Sequence Capture Kit (Agilent, United States). The sequencing library was quantified using the Illumina DNA Standards and Primer Premix Kit (Kapa Biosystems, United States), and was massively parallel-sequenced using the Illumina Novaseq6000 platform. After sequencing and filtering out the low-quality reads, the high-quality data (with general quality level Q30 reads >89%) was aligned to the human genome reference sequence [hg19] using Burrows-Wheeler Aligner tool. The third-party software GATK (https://softw are.broad institute.org/gatk/) and the Verita Trekker^® Variants Detection system (Berry Genomics, China) were employed for variant calling. Variants with lower quality (read depth<10x, allele fraction<30%) were eliminated. The variations were identified by sequence alignment with the NCBI Reference Sequence (NG_016423.2, NP_001073932.1, NM_001080463.2) using Chromas v2.33. The pathogenicity of the identified variants was then assessed according to the common guidelines issued by the American Association of Medical Genetics and Genomics (ACMG) (Richards et al., 2015) referring to multiple databases (1000g2015aug_eas, https://www.internationalgenome.org/; ExAC_EAS, http://exac.broadinstitute.org; gnomAD_exome_EAS, http://gnomad.broadinstitute.org/; HGMD^®: Human Gene Mutation Database Professional v.2021.10) with the Enliven^® Variants Annotation Interpretation (Berry Genomics, China) system.

The suspected diagnostic variants were validated by Sanger sequencing using ABI 3730 Automated Sequencer (Applied Biosystems, United States) according to the manufacturer’s protocol. Fluorescent quantitative PCR (qPCR) was also carried out to verify the suspected deletion variant identified in Case 2.

The evolutionary conservatism of amino acid (AA) residues affected by specific missense variants was analyzed using UGENE (http://ugene.net/) with default parameters.

Referring to the crystal structure of 6rla. 1. A (Toropova et al., 2019) protein, the DYNC2H1 protein structures at 2241–2520 and 2611–2880 regions were constructed by Swiss-model program. The Swiss-Pdb Viewer program was referred to modeling the wild-type (WT) and DYNC2H1: p. Ile2430Thr and p. Leu2730Phe mutant models of DYNC2H1 protein segments. The molecular dynamics (MD) prediction analysis was generated by GROMACS (version 2020.6) (Rakhshani et al., 2019). We carried out 60 ns MD simulations on the DYNC2H1-WT, DYNC2H1-Ile2430Thr, and DYNC2H1-Leu2730Phe models. The CHARMM36 force field was applied to add hydrogen atoms and N-terminal and C-terminal patches to the models (Soteras Gutiérrez et al., 2016). The wild type or the mutant structure of the protein was immersed in cubic boxes which contains water and placed at least 1.0 nm from the box edge. Na⁺ and Cl⁻ ions were used for neutralization. The MD simulations were performed at a temperature of 300K for 60 ns after energy minimization, equilibration. The following GROMACS distribution programs were used in MD trajectories: gmxrms, gmxrmsf, gmx gyrate, gmxsasa, and gmxhbond. These MD analyses generated parameters values for root-mean-square deviation (RMSD), root-mean-square fluctuation (RMSF), radius of gyration, solvent accessible surface area (SASA), and number of h-bonds.

Case 1. A 32-year-old woman was referred to our center in March 2021 for previous multiple adverse gestations, when she was at the 17th⁺gestational week. Her husband was 31 years old, with no consanguineous relation to her. Based on the medical record and her personal dictation, we combed through the couple’s complete medical history and illustrated the pedigree in a diagram (Figure 1A). They had 3 previous pregnancies: in October 2016 and July 2017, they went through one spontaneous abortion and one stillbirth, respectively; in November 2018, the prenatal sonography revealed a 22-week fetus with short limbs and a narrow chest, which was diagnosed as thanatophoric dwarfism. That pregnancy was subsequently terminated.

In 2020, they had the 4th gestation, and the serological screening and NIPT (non-invasive prenatal testing) results were normal. At the 17th week, ultrasonic examination and MRI revealed that the fetus had extremely short limbs and small, bell-shaped chest, short ribs, rhizomelic shortening in all extremities (Figures 1B–E). Prenatal ultrasound did not identify abnormalities in the brain, heart, kidneys, or liver. No exposure history to tobacco smoke, alcohol, ir-radiation, or infectious diseases during the pregnancy were admitted. Pregnancy termination was performed at 22 gestational weeks with informed consent. Afterwards, X-ray imaging result demonstrated that the fetus had a narrow thorax and short limbs, but no polydactyly (Figure 1F). And whole-exome sequencing (WES) was introduced afterwards.

Case 2. A 31-year-old pregnant woman was referred to our center in April 2021. Her husband and her were non-consanguineous. The pedigree diagram was depicted in Figure 2A: in November 2018, a fetus was diagnosed with constricted thoracic cage, extremely shortened tubular bones and bowing of long bones, and then aborted at 12th week; in April 2021, ultrasonography revealed that their second fetus had extremely short limbs at the 13th week (Figures 2B, C). No brain, heart, kidney, or liver abnormality was found by prenatal ultrasonography. The woman denied to be exposed to tobacco smoke, alcohol, radiation or infectious diseases during pregnancy. After informed consent by the couple, genetic analysis including WES was also performed after induction.

Karyotyping by G-bandeding showed that the results of the fetuses in both Case1 and 2 were normal, and array-CGH analysis did not reveal any genomic abnormality associated with known microdeletion or microduplication syndromes, either. On the other hand, according to the WES results, the two fetuses were recognized as positive with compound heterozygous variation in the DYNC2H1 gene. To be specific, Proband 1 (II-4 in Case 1) carried two heterozygous missense variants, namely c.2386C>T (p.Arg796Trp) and c.7289T>C (p.Ile2430Thr) (Figures 1G, H); while Proband 2 (II-2 in Case 2) carried an exonic deletion, exon (64–83)del, and a missense variant, c.8190G>T (p.Leu2730Phe) (Figure 2G). qPCR validation confirmed the one copy loss of 64–83 exons in DYNC2H1 of Proband 2 (Figure 2D). Validation with Sanger sequencing and qPCR demonstrated that the variants these probands carried were inherited from their asymptomatic heterozygous carrier parents, respectively (Figures 1G, H; Figures 2D–G). The location of each variant was illuminated in the gene and peptide diagrammatic sketches (Figure 3A).

In this study, three missense variants were detected, which were DYNC2H1: c.2386C>T (p.Arg796Trp), c.7289T>C (p.Ile2430Thr) and c.8190G>T (p.Leu2730Phe). We analyzed the evolutionary conservatism of AA residues they affected. Results indicated that all three AAs maintained highly conserved among species (Figure 3B).

To investigate the intramolecular effect of these missense variants, we performed structural analysis and MD simulation. Due to the structural inaccuracy of the region where p. Arg796Trp was located, we only analyzed the DYNC2H1 protein structures at 2241–2520 region (violet part in Figure 4A) and 2611–2880 region (orange part in Figure 4A), where p. Ile2430Thr and p. Leu2730Phe were located.

The results demonstrated that the Ile2430Thr variant affected the hydrogen bonding between amino acids inside the protein. Particularly, in the wild-type, Ile2430 formed hydrogen bonds with Val2290 residues, with the hydrogen bond length of 2.9Å; while the Ile2430Thr mutant formed a hydrogen bond with Val2290 with a hydrogen bond length of 1.9Å (Figures 4B–E; 4C and 4E are local amplifications of 4B and 4D, respectively). As for the MD results, first, the Thr2430 mutant formed more hydrogen bonds with other amino acids in the protein than the wild-type residue Ile2430 (Figure 4F); for the total number of hydrogen bonds inside the models within 60ns, the amount of Ile2430Thr mutant was more than that of the wild-type (Figure 4G). Besides, the Ile2430Thr variant resulted in a change in secondary structure around 2350th and 2400th residue (Figures 4H, I). Specifically, in wild-type, the secondary structure at 2350 position was dominated by BEND; yet in the mutant, the secondary structure alternated between BEND and TURN and was dominated by TURN. In wild-type, the secondary structure at 2400 position was dominated by BEND; yet in the mutant, the secondary structure alternated between BEND and TURN and was dominated by TURN. Finally, the Ile2430Thr variant resulted in decreased changes in protein structure (Figure 4J), decreased flexibility of amino acids in protein (Figure 4K), increased protein compactness (Figure 4L), and decreased surface area accessible to protein solvent (Figure 4M).

Structural result demonstrated that Leu2730Phe variant affected the hydrogen bonding between amino acids inside the protein. Particularly, Leu2730 and Phe2730 both formed hydrogen bonds with Ile2726 and Lys2802 residues, yet in the Leu2730Phe mutant the hydrogen bond length is shorter than that in the wild-type (Figures 5A–D; 5B and 5D are local amplifications of 5A and 5C, respectively). As for the MD results, there was no difference between Leu2730 and Phe2730 in the number of hydrogen bonds formed with other amino acids in the protein (Figure 5E); for the number of total hydrogen bonds inside the models within 60ns, the amount of Leu2730Phe mutant was more than that of the wild-type (Figure 5F). In wild-type, the secondary structure at 2690 position was dominated by H-helix; yet in the mutant, the secondary structure showed a high rate of BEND and TURN (Figures 5G, H). The Leu2730Phe mutation has little influence on RMSD, RMSF, Gyrate, and SASA (Figures 5I–L).