Identification of candidate variants in a consanguineous family with DCM
A consanguineous family with 3 DCM-affected members (Family 1) was recruited (Fig. 1A). Conventional and dynamic electrocardiograms suggested occasional first-degree atrioventricular block, paroxysmal sinus tachycardia and complete left bundle branch block (Fig. 1B). The proband was treated with cardiac resynchronization therapy to provide simultaneous electrical activation. The proband’s brother was also DCM patient, and he received a heart transplant. However, the surgery did not substantially reduce his symptom severity, and he died one year later. The proband’s sister was also diagnosed with DCM and she showed frequent atrial premature contractions, and partial of the premature electrical impulse not transmitted downstream to the ventricle. All three DCM patients in the consanguineous family did not show any skeletal muscle abnormalities and their CK serum level was within the normal range. None of the parents or other antecedents showed cardiac dysfunction.
WES was performed on consanguineous family members with DCM but revealed no known DCM-causing gene mutations. We therefore sought new candidate pathogenic variants. WES data was analysed as illustrated (Additional file 1: Fig. S1). In total 14,208 candidate rare variants were obtained. Considering the possible autosomal recessive inheritance pattern, we obtained 56 candidate variants. Finally, 51 variants were excluded based on the American College of Medical Genetics and Genomics 2015 guidelines. The remaining 5 variants in 5 separate genes, BICD2, SERINC1, BVES, ADCY1 and PAPPA, were considered for further evaluation (Additional file 1: Table S1).
BICD2 is a dynein-activating adaptor protein that functions in minus-end-directed microtubule-based transport by docking dynein motor proteins to appropriate cargos. Reduced LMNA protein expression and MKL-1/SRF activity were observed in Bicd2-deficient mice [19]. LMNA is a known causative gene for DCM, accounting for 4–8% of patients with DCM [25,26,27]. In addition, some individuals with spinal muscular atrophy caused by the BICD2 ‘hot spot’ mutation, c.302C > T:p.Ser107Leu, presented with unexpected heart failure-associated symptoms and exertional and supine dyspnoea [17]. Considering the possible interaction between LMNA and BICD2 and the heart failure symptoms of BICD2 mutant patients, we hypothesized that the homozygous variant (NM_001003800.1:c.2429G > A) in the C-terminal region of BICD2 (Fig. 1C) would be a candidate DCM-causative variant in this family.
The missense variant of BICD2 (Fig. 1C), segregated with the disease phenotype. The Sorting Intolerant From Tolerant [28] algorithm prediction score for the variant was 0.003 (Additional file 1: Fig. S2), indicating a damaging role. Sanger sequencing confirmed that both parents were heterozygous carriers and that the two surviving DCM patients were homozygous at this locus (Fig. 1D). Furthermore, the amino acids affected by this variant are highly evolutionarily conserved in multiple species (Fig. 1E). Together, these findings indicate that the BICD2 missense variant could be harmful to protein function.
Three BICD2 variants in 210 sporadic DCM cases
We further performed Sanger sequencing of BICD2 in 210 DCM cases and found one missense variant and two synonymous variants (Additional file 1: Fig. S3). The missense variant, located in exon 2 (NM_001003800.1:c.421C > A) (Additional file 1: Fig. S3A), contributed to an amino acid substitution from arginine to serine. Bioinformatic prediction suggested that it was a deleterious variant (Additional file 1: Table S2). The other two variants of BICD2 were synonymous (Additional file 1: Table S2). These results support the hypothesis that BICD2 may play a significant role in DCM pathogenesis.
BICD2 expression in heart tissue
We explored the expression of BICD2 in heart tissues from humans as well as mice and zebrafish. Real-time PCR results showed that Bicd2 mRNA expressed in the hearts of mice and that its expression level increased with age (Fig. 2A). A similar expression pattern was also observed in the zebrafish heart, in which the expression of bicd2 was much higher at 7 months than at 5 months (Fig. 2B).
In addition, immunohistochemical staining of BICD2 in normal left atrial appendage supported its high expression in human (Fig. 2C). Moreover, RNA-seq of BICD2 from the GTEx database confirmed its relatively high expression in the human heart (Fig. 2D), and scRNA-seq data of BICD2 suggested that it was mainly expressed in myocytes, endothelial cells, and fibroblasts (Additional file 1: Fig. S4). Thus, we demonstrated that BICD2 expressed in the hearts of human, mice and zebrafish at both the mRNA and protein levels.
We further conducted in situ hybridization of bicd2 in zebrafish at embryonic stages(Fig. 2E). At 48 hpf, bicd2 was expressed in the retina, forebrain, midbrain and hindbrain regions, with strong expression in the pectoral fin bud base (Fig. 2E). Taken together, these data indicate that BICD2/Bicd2/bicd2 are expressed in the heart in multiple species, implying their involvement in heart development and function.
Knockout of bicd2 leads to partial embryonic lethality and altered cardiac function in homozygotes
We designed a bicd2 knockdown assay in zebrafish to unravel the associated phenotypic changes; the experimental workflow is shown in Fig. 3A. We injected Cas9/sgRNA into embryos of F0 generation zebrafish and further screened for bicd2-deficient zebrafish in F0 adult zebrafish (Fig. 3A) (CRISPR target sequence in Additional file 1: Table S3). Three fish were obtained and mated with each other to produce bicd2 heterozygous F1 offspring (Fig. 3A) (PCR primers for F1 mutant identification in Additional file 1: Table S4). Details for the generation of bicd2-KO zebrafish are provided in Additional file 1: Fig. S5. We then allowed the F1 generation fish to self-cross to obtain F2 generation fish with three segregating genotypes (Fig. 3A). We measured the survival rate of embryos with different genotypes at three embryonic time points (Fig. 3A). At the adult stage, we performed echocardiographic measurements and transcriptome sequencing to delineate possible regulatory mechanisms(Fig. 3A).
We firstly examined bicd2 immunostaining to determine whether bicd2 was efficiently knocked out. Immunostaining images showed lower bicd2 expression in the bicd2–fish (Fig. 3B). Viable embryos’ number was much lower in bicd2–than in the other two groups at all three developmental stages (Fig. 3C, Additional file 1: Fig. S6). At 50 hpf, only one bicd2–embryo survived (Additional file 1: Fig. S6) (Additional file 1: Table S5), while the number of surviving bicd2+−and bicd2++embryos were twenty-two and thirteen (Additional file 1: Fig. S6) (Additional file 1: Table S5), respectively. Thus, the proportion of viable homozygous embryos was 2.78%, far lower than the theoretically predicted ratio of 25%. A similar trend was observed at both 76 hpf and 120 hpf (Fig. 3C, Additional file 1: Fig. S6) (Additional file 1: Table S5). The results suggests that bicd2 plays a vital role in zebrafish embryogenesis and supports the conclusion that bicd2 knockout leads to partial embryonic lethality. The heart rate of bicd2–, bicd2+−and bicd2++fish at three embryonic stages showed no significant difference (Additional file 1: Fig. S7) (Additional file 1: Table S6). In addition, no obvious phenotype was observed during the embryo development period (images not shown).
Electrocardiography was performed on bicd2– and bicd2++fish at seven months. Ventricular chamber size was assessed based on volume and area. We therefore measured the two-dimensional end-diastolic area (VAd) and end-systolic area (VAs) and three-dimensional end-diastolic volume (EDV) and end-systolic volume (ESV) in both bicd2–and bicd2++zebrafish. Compared with bicd2++zebrafish, bicd2–fish showed slightly larger values in both VAd and VAs (Fig. 3D, Additional file 1: Table S7). The EF was much lower in bicd2–than in bicd2++fish (Wilcox test, P value = 0.063) (Fig. 3D, Additional file 1: Table S7). EF < 45% is diagnostic for DCM [29]. In our studies, most bicd2–fish presented EF < 45%, supporting that bicd2 is crucial for cardiac function. Moreover, cardiac output was significantly lower in the bicd2– fish than bicd2++fish (Wilcox test, P value < 0.05) (Fig. 3D, Additional file 1: Table S7). Echocardiographic images representing zebrafish hearts with two different genotypes were extracted from echocardiograms and were compared (Fig. 3E). In summary, these results indicate that bicd2 knockout may cause abnormal contraction of the heart.
RNA-seq of bicd2-deficient zebrafish revealed a transcriptomic shift in cardiomyocytes
We conducted transcriptome sequencing of heart tissues from the bicd2–, bicd2+−and bicd2++fish to examine the transcriptional differences. Consistent with expectations, bicd2 expression was significantly lower in the knockout group than in the bicd2++fish (Additional file 1: Table S8). Gene set enrichment analysis (GSEA) [30] based on ranking of all expressed genes can provide more informative biological evidence than enrichment analysis focused solely on differentially expressed genes (DEGs). Therefore, we first utilized GSEA to identify significantly enriched gene sets. Compared to the bicd2++fish, the bicd2–group showed enrichment of cardiopathy-related signalling pathways and metabolic pathways, for example, KEGG_CARDIAC_MUSCLE_CONTRACTION, KEGG_CALCIUM_SIGNALI-NG_PATHWAY,KEGG_OXIDATIVE_PHOSPHORYLATION,KEGG-GLYCOL
YSIS_GLUCONEOGENESIS, KEGG_CITRATE_CYCLE_TCA_CYCLE and KEGG_PENTOSE_PHOSPHATE_PATHWAY (Fig. 4A), as well as pathways associated with nervous system disease (Additional file 1: Table S9). These results are consistent with functional explorations of Bicd2 in mice, which proved a causal role of Bicd2 in neuronal disorders [19]. In addition, pathways related to mitochondrial energy metabolism, such as oxidative phosphorylation (OXPH) and the tricarboxylic acid cycle (TCA), are common characteristics of distinct heart failure, as demonstrated previously [31, 32]. Interestingly, 6 out of 14 pathways (Fig. 4B) exhibited increases in expression in both the bicd2–and bicd2+−fish, including KEGG_CALCIUM_SIGNALING_PATHWAY and KEGG_GLYCOLYSIS_GLU.
-CONEOGENESIS. Only 3 common pathways exhibited decreased expression in those two groups (Fig. 4C), implying the transcriptome differences between the bicd2–and bicd2+−fish.
We then extracted DEGs to further delineate biologically meaningful pathways involved in the bicd2 gene regulatory network. As expected, the bicd2–fish displayed the largest change among all three groups, with a total of 1708 DEGs (483 upregulated genes, 1225 downregulated genes, Fig. 4D, Additional file 2: Table S10). Surprisingly, the difference between the two groups of bicd2-KO genotypes was much larger than that between the bicd2+−fish and the bicd2++fish. In detail, there were 1645 DEGs (417 upregulated genes, 1228 downregulated genes, Fig. 4D, Additional file 3: Table S11) between the bicd2–fish and the bicd2+−fish. And 535 DEGs (334 upregulated genes, 201 downregulated genes, Fig. 4D, Additional file 4: Table S12) between the bicd2+−fish and the bicd2++fish. The top 15 enriched GO pathways based on 1708 DEGs discovered between the bicd2–fish and the bicd2++fish are displayed (Additional file 1: Fig. S8-9). Genes with increased expression in the bicd2++fish were mainly related to blood vessel development, extracellular matrix organization, cell–cell adhesion, carbohydrate metabolic process and circulatory system process (Additional file 1: Fig. S8). However, the genes with decreased expression were mainly related to the immune system, including regulation of immune system processes, response to biotic stimulus, and leukocyte activation (Additional file 1: Fig. S9). Notably, an increased expression of genes encoding extracellular matrix proteins in DCM patients was previously demonstrated by microarray [31]. In our study, extracellular matrix genes (Fig. 4E) (Additional file 1: Fig. S10), such as col4a2, col4a4, col5a1, col8a1b, col18a1b, and genes involved in circulatory system processes showed higher expression in the homozygous group than in the wild-type group (Fig. 4F) (Additional file 1: Fig. 11). Genes related to the regulation of immune system processes (Fig. 4G) (Additional file 1: Fig. 12), such as cd74b and cd79a, exhibited decreased expression in the two bicd2-KO groups, implying activation of the immune system after bicd2 knockout in zebrafish.
A set of 51 genes with evidence of association with susceptibility to inheritable DCM in humans was established in a previous study [4]. Orthologue retrieval in the zfin data report and Ensembl identified zebrafish homologues for 48 of these genes (Additional file 5: Table S13). Among these 48 genes, 28 had a single orthologue in zebrafish (Additional file 1: Fig. S13, Additional file 5: Table S13). 8 out of these 28 orthologues, including bag3, nebl, psen2, jph2, dtna, psen1, gatad1 and plekhm2, showed the lowest expression in the homozygous group (Additional file 1: Fig. S13). In contrast, 10 out of 28 orthologues, including mybpc3, lama4, nppa, tnni3k, cmlc1, nexn, lrrc10, csrp3, abcc9, and tcap, showed the highest expression in the homozygous group (Additional file 1: Fig. S13). Multiple zebrafish orthologues were identified for 20 human DCM candidate genes (Additional file 1: Fig. S14).