Mutations of RNF213 are responsible for sporadic cerebral cavernous malformation and lead to a mulberry-like cluster in zebrafish

Abstract

Although familial forms of cerebral cavernous malformation are mainly attributed to three CCM genes (KRIT1, CCM2 and PDCD10), no mutation is identified in sporadic cerebral cavernous malformation cases with a unique lesion, indicating additional genes for sporadic cerebral cavernous malformation. To screen the candidate genes, we conducted whole exome sequencing in 31 sporadic cerebral cavernous malformation patients and 32 healthy controls, and identified 5 affected individuals carrying 6 heterozygous deleterious mutations in RNF213 but no RNF213 mutation in healthy individuals. To further confirm RNF213 was associated with cerebral cavernous malformation, we generated rnf213a homozygous knockout zebrafish and found mutation of rnf213a in zebrafish led to a mulberry-like cluster of disordered-flow vascular channels which was reminiscent of human cerebral cavernous malformation. In addition, we revealed kbtbd7 and anxa6 were significantly downregulated due to rnf213a mutation through transcriptomic sequencing and RT-qPCR analysis. Based on the mulberry-like phenotype partly rescued by mRNA of kbtbd7 as well as anxa6, we suggested that rnf213a promoted mulberry-like cluster via downregulation of kbtbd7 and anxa6. Altogether, we firstly demonstrate RNF213is a novel candidate gene for sporadic cerebral cavernous malformation and the mutation of rnf213a is responsible for the mulberry-like cluster in zebrafish.

Introduction

Cerebral cavernous malformation consists of a mulberry- or raspberry-like cluster of low-flow vascular channels and occurs in venous-capillary vascular bed.¹ The primary symptoms of cerebral cavernous malformation include epilepsy, headache, focal neurological deficits and hemorrhage.² Both sporadic (80% of cases) and familial forms (20% of cases) of cerebral cavernous malformation have been identified. Sporadic cases often have a single lesion, are not inherited, and do not carry a mutation of CCM genes (KRIT1, CCM2 and PDCD10). In contrast, familial cases usually present multiple lesions, have affected relative, and harbor a germline mutation of CCM genes.³ Mutation detection rates in CCM genes had been extremely high for familial cases, ranging from 87% to 94%.^4,5 However, no mutation of CCM genes was detected in sporadic cerebral cavernous malformation cases with a unique lesion, indicating there may be additional genes associated with sporadic cerebral cavernous malformation.^6,7

Human ring finger protein 213 gene (RNF213) is located on chromosome 17 and encodes a ring finger protein of 5207 amino acids that includes two important functional domains: a RING finger domain and an AAA+ATPase domain.⁸ In recent years, RNF213 was identified as a susceptibility gene for moyamoya disease (MMD) among East Asian populations.^9,10 Kamada et al.⁹ found that RNF213 p.R4810K was detected in 95% of familial MMD and in 79% of sporadic MMD patients. More interestingly, further studies demonstrated that RNF213 was not only related to MMD but also associated with intracranial major artery stenosis/occlusion (ICASO) and intracranial aneurysm (IA).^11–13 Thus, RNF213 may be an important gene for many cerebrovascular diseases. Moreover, the underlying molecular mechanisms of RNF213 that plays in cerebrovascular diseases remain poorly understood.

To gain insight into the underlying genes for sporadic cerebral cavernous malformation, we conducted whole exome sequencing for 31 sporadic cerebral cavernous malformation patients and 32 control individuals, and identified 6 RNF213 mutations in sporadic cerebral cavernous malformation patients with a unique lesion. Although it is hard to do pathological study in zebrafish brain vessels, zebrafish has emerged as a powerful model to study genetically related cerebrovascular diseases attribute to its transparency and genetic tractability. Thus, we further knocked out rnf213a (zebrafish ortholog of human RNF213) in zebrafish to explore the related phenotype and molecular mechanisms.

Material and methods

Patients’ selection and samples’ collection

We recruited 31 sporadic cerebral cavernous malformation patients and 32 healthy controls at the First Affiliated Hospital of Sun Yat-sen University, from January 2016 to August 2018. The diagnosis of cerebral cavernous malformation was made based on the findings of magnetic resonance imaging (MRI). In addition, further pathological examination would be carried out to confirm the diagnosis for the patients who underwent the operation of cerebral cavernous malformation. Patients with single lesion and negative or unavailable family history were considered sporadic cerebral cavernous malformation. The control individuals were subjects undergoing MRI and MRA for other reasons, such as dizziness, venous sinus thrombosis or headache. The MRI and MRA of healthy controls revealed no cerebral cavernous malformation, MMD, ICASO and IA. All subjects were Han Chinese. Blood samples were collected from the samples mentioned above after written informed consent had been obtained. The subjects’ consent was obtained according to the Declaration of Helsinki and the Ethics Committee of the First Affiliated Hospital of Sun Yat-sen University approved the study.

Whole exome sequencing and data analysis

DNA samples were extracted from peripheral blood samples using QIAGEN DNA Blood Mini Kit (Cat#51106, QIAGEN Co., Ltd) following the Handbook. OD260, OD280 value was determined using NanoDrop® instrument and the OD260/OD280 ratio should be around 1.8–2.0 to be good enough for the following sequencing process. Whole exome sequencing was done on the illumina NextSeq 500 platform.

The sequencing data were analyzed using the ANNOVAR software, and several databases including the 1000 g, ESP6500, dbSNP, ClinVar, and HGMD were used for the screening and annotation of the variants following the ACMG guidelines.^14,15 The sequencing reads were mapped to the hg19 reference using bwa, variant calling was done using samtools (0.1.19), picard (1.123), and GATK (3.3-0-g37228af). Potential effects of variants were predicted by PolyPhen-2, SIFT, LRT, MutationTaster, MutationAssessor, FATHMM, GERP, PhyloP, SiPhy, NetGene2 Server and AUGUSTUS. Sanger sequencing on the ABI 3500 Dx platform was applied to validate the illumina Next 500 sequencing results. qPCR was applied to validate a large deletion which was suspected in the illumina Next 500 sequencing results. The extraction of DNA, whole exome sequencing and data analysis for patients was operated by the investigators blinded to patients’ identities.

Zebrafish lines and generation of rnf213a homozygous zebrafish

Embryos and adult fish were maintained, raised and staged as described previously.¹⁶ Zebrafish (Danio rerio) of Tg (flk1: eGFP/gata1: DsRed) transgenic line was kindly provided by Guangdong Medical University, Zhanjiang, China. The rnf213a knockout zebrafish (F0) generated by transcription activator-like effector nuclease (TALEN) technique was performed as described.¹⁷In order to further obtain homozygous mutant and purify the genetic background, we outcrossed the heterozygous mutant to the wild type for three generations and then heterozygotes intercrossed to generate homozygotes. All procedures involving experimental animals were approved by the Institutional Animal Care and Use Committee of the Sun Yat-sen University. Reporting of these experiments complies with the ARRIVE (Animal Research: Reporting in Vivo Experiments) guidelines.

Lightsheet and confocal microscope

For lightsheet microscopy, live embryos randomly selected were immobilized in 1% low melting agarose and maintained at 28°C. The embryos were imaged to observe the whole cranial vessels using a Zeiss Lightsheet Z.1 microscope at 72hpf, 96hpf and 120hpf with optimal stacks captured at each time point. The overall wiring pattern of major cranial vessels is largely unaltered after 60hpf and some small vessels continue to form in the head through stages, thus we selected the time points (72hpf, 96hpf and 120hpf) to observe the phenotype.¹⁸ Confocal movies of blood flow in embryos with the background of flk1: eGFP/gata1: DsRed were generated using a Leica TCS SP5 II confocal microscope. Phenotype of zebrafish was analysed by the investigators who did not know whether the randomly selected embryos were mutants or wild type.

High-throughput transcriptomic sequencing

Total RNA was extracted from randomly selected whole-mount embryos at 72hpf and 120hpf using the TriZol reagent (Invitrogen) and purified by RNeasy Mini Kit (QIAGEN). Total RNA of each sample was quantified and qualified by Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA), NanoDrop (Thermo Fisher Scientific Inc.); 1 µg total RNA with RIN value above 7 was used for following library preparation. Next generation sequencing library preparations were constructed according to the manufacturer’s protocol (NEBNext® Ultra™ RNA Library Prep Kit for Illumina®). Then libraries with different indices were multiplexed and loaded on an Illumina HiSeq 2000 platform for sequencing according to manufacturer’s instructions (Illumina, San Diego, CA, USA). Sequencing was carried out using a 2 × 150bp paired-end (PE) configuration; image analysis and base calling were conducted by the HiSeq Control Software (HCS) + OLB + GAPipeline-1.6 (Illumina) on the HiSeq instrument.

In order to remove technical sequences, including adapters, polymerase chain reaction (PCR) primers, or fragments thereof, and quality of bases lower than 20, pass filter data of fastq format were processed by Trimmomatic (v0.30) to be high quality clean data. Clean data were then aligned to zebrafish reference genome via software Hisat2 (v2.0.1). Differential expression analysis used the DESeq Bioconductor package, a model based on the negative binomial distribution. After adjusted by Benjamini and Hochberg’s approach for controlling the false discovery rate, significantly differentially expressed genes relative to the control group were identified with P value < 0.05 and |log2 (fold change)| >1.

Quantitative real-time PCR

RT-qPCR assays were used to verify significant differential transcripts selected in transcriptomic profiles. Total RNA was extracted from 40 whole embryos per sample for each condition using the TriZol reagent (Invitrogen) followed by purification using RNeasy Mini Kit (QIAGEN). Each group contained three samples and therefore represented three biological replicates. After RNA quality control by gel electrophoresis and biophotometer, RNA was reverse transcribed into cDNA using TransScript® One-Step gDNA Removal and cDNA Synthesis SuperMix (Transgene). RT-qPCR was carried out using TransScript® Tip Green qPCR SuperMix (Transgene). Eukaryotic translation elongation factor 1 alpha 1, like 1 (elfa) was used as the reference gene. Primers sequences of the selected genes used for RT-qPCR were described in the supplement data (Supplementary Table 1).

mRNA rescue studies

Full-length cDNA of the genes including kbtbd7 and anxa6 was amplified from cDNA and cloned into pCS2+. Primers sequences used for high-fidelity PCR were described in the supplement data (Supplementary Table 2). Capped mRNA was synthesized from ApaI-digested constructs using mMESSAGE mMACHINE with SP6 RNA polymerase (Ambion) followed by RNA purification using MEGAclear™ Kit (Ambion). Gradient doses of capped mRNA (50 pg, 100 pg and 200 pg) were injected into one-cell stage embryos from rnf213a^−/− intercrosses.

Total RNA isolation, transcriptomic sequencing experiments, RT-qPCR and mRNA rescue studies were operated by the investigators blinded to identities of the sample.

Statistical analysis

Gene mRNA levels from the RT-qPCR were analyzed by t-test using IBM SPSS Version 20.0 for Windows (SPSS Inc., Chicago, IL, USA). Significance was predefined at P < 0.05.

Results

Identification of RNF213 mutations in sporadic cerebral cavernous malformation patients

To screen the candidate genes for sporadic cerebral cavernous malformation, whole exome sequencing was performed in 31 sporadic cerebral cavernous malformation patients and 32 healthy controls. We selected candidate genes that met the following criteria: (i) minor allele frequency (MAF) was <0.003; that were (ii) located in coding regions or predicted to affect mRNA splicing; (iii) potential deleterious variants predicted by PolyPhen-2, SIFT, LRT, MutationTaster, MutationAssessor, FATHMM, GERP, PhyloP, SiPhy, NetGene2 Server and AUGUSTUS; (iv) shared by two or more sporadic cerebral cavernous malformation patients but not identified by healthy individuals. Under the strategy above, five sporadic affected cases with a unique lesion were carrying six deleterious mutations in RNF213 (Figure 1(a) and Table 1). However, there was no RNF213 mutation in control individuals (Supplementary Table 3).

Figure 1.

Sanger sequencing and schematic diagram of mutations, brain MRI and histology in sporadic cerebral cavernous malformation patients with RNF213 mutations. (a) Six mutations of RNF213 in affected individuals were detected by Sanger sequencing. Red arrows indicated the position of mutations. (b) Schematic diagram of the six mutations at RNF213 locus. Genomic structure of RNF213 gene included 68 exons (red rectangles) and RNF213 protein contained 2 domains (marked in blue). Three mutations led to a truncated protein, respectively (position of the translation termination codon was marked by arrows). Horizontal dotted line indicated the exons not shown. (c) Brain MRI revealed the lesions of cerebral cavernous malformation (red arrowheads) in affected patients carrying RNF213 mutations and HE staining for lesions showed hyalinized caverns of varying size (black arrowheads) in patient 3. Bar = 500 µm.

Table 1.

Main basic information and genetic features of mutations screened in patients with cerebral cavernous malformation.

Patient identification					Gene variation
Patient	No of CCM lesions	Age-onset	Gene	Exon/intron	Nucleotide change	Amino acid change	References
P1	Single	37	RNF213	Exon18	c.3086T>C	p.(Leu1029Ser)	Present work
P2	Single	10	RNF213	Exon60	c.14430A>C	p.(Arg4810Ser)	Present work
P3	Single	34	RNF213	Exon28 Exon42	c.6281delT c.11759G>A	p.(Leu2094fs) p.(Trp3920*)	Present work
P4	Single	41	RNF213	Exon52	c.13322delG	p.(Gly4441fs)	Present work
P5	Single	35	RNF213	Exon6	c.1066_1068delAAG	p.(Lys356del)	Present work
P6	Single	41	NOTCH3	Exon3	c.224G>A	p.(Arg75Gln)	Liu X et al. (2015)38
P7	Single	30	HTRA1	Exon2	c.518C>T	p.(Ala173Val)	Present work
P8	Single	46	F5	Exon7	c.1000A>G	p.(Arg334Gly)	Dirven RJ et al. (2010)39
P9	Single	48	IFIH1	Exon5	c.937A>G	p.(Met313Val)	Present work
P10	Single	27	SMAD4	Exon8	c.911T>C	p.(Val304Ala)	Present work

Note: Reference sequences: RNF213 transcript (NM_001256071.2), NOTCH3 transcript (NM_000435.2), HTRA1 transcript (NM_002775.4), F5 transcript (NM_000130.4), IFIH1 transcript (NM_022168.3) and SMAD4 transcript (NM_005359.5).

In affected patient 3, a heterozygous 1 bp deletion (c.6281delT) and a heterozygous nonsense mutation (c.11759G>A) in RNF213 leading to premature stop codon (p.Leu2094fs*45 and p.Trp3920*) were identified (Figure 1(b) and Table 1). Minor allele frequency (MAF) of the two RNF213 mutations including c.6281delT and c.11759G>A in the GnomAD database was 0.00006 (16/246272) and 0.00047 (115/246238), respectively. Another frame-shift mutation (c.13322delG) of RNF213 which was not annotated in the 1000 Genomes or dbSNP database also caused early termination of protein translation (p.Gly4441fs*3) in patient 4 (Figure 1(b) and Table 1). Two rare missense mutations (c.3086T>C and c.14430A>C) and 3 bp deletion mutation (c.1066_1068delAAG) in RNF213 were identified in the other three sporadic cerebral cavernous malformation patients, and the MAF was 0.00005 (7/137444), 0.00000 (1/246232) and 0.00001 (2/249878), respectively (Figure 1(b) and Table 1). Although these three mutations should not lead to a truncated protein, they were predicted to have the potential to cause disease by PolyPhen-2, SIFT, LRT, MutationTaster, MutationAssessor, FATHMM, GERP, PhyloP, SiPhy, NetGene2 Server and AUGUSTUS.

In addition to RNF213, mutations of other genes were identified in sporadic cerebral cavernous malformation individuals including NOTCH3, HTRA1, F5, IFIH1 and SMAD4 (Table 1). However, these positive genes above could not be shared by another affected individual. Collectively, these results suggested that RNF213 was the candidate gene associated with sporadic cerebral cavernous malformation with a unique lesion.

Clinical data in sporadic cerebral cavernous malformation patients with RNF213 mutations

The main basic information and genetic features of cerebral cavernous malformation patients with positive mutations are summarized in Table 1. The cerebral cavernous malformation patients with RNF213 mutations had only one cerebral cavernous malformation lesion. Brain MRI demonstrated that the lesions of affected patients with RNF213 mutations were all in supratentorial except for patient 1 whose lesion was located in infratentorial (Figure 1(c)). The five patients with RNF213 mutations showed a highly variable age of onset (10 to 41 years) and diverse symptoms. Patient 2 and patient 5 presented with epilepsy. In addition to headache, patient 1 complained of limb weakness. The primary symptoms of patient 4 included epilepsy and headache. Patient 3 with both RNF213 mutations which led to a truncated protein had a past history of recurrent epilepsy. He was referred to the neurosurgeon at the age of 36 years due to poor drugs control and subsequent histology showed hyalinized caverns of varying size in the lesion (Figure 1(c)).

Comparison of RNF213 protein sequence and generation of rnf213a homozygous zebrafish

To further confirm whether RNF213 was associated with cerebral cavernous malformation, we aimed to use zebrafish to explore if rnf213 mutant zebrafish could mimic the characteristics of human cerebral cavernous malformation. Firstly, we did homologous analysis for RNF213 protein between zebrafish and human. There were two zebrafish rnf213 genes (rnf213a and rnf213b) and one human RNF213 gene in genebank. Multi-alignment analyses indicated a highly conserved AAA+ATPases domain and RING finger domain between zebrafish and human (Figure 2(a)). However, the sequences alignment showed that zebrafish rnf213a was more similar to human RNF213 compared with zebrafish rnf213b (RNF213 VS rnf213a: max score: 4400; total score: 4400; query cover: 99%; identity: 45.44%. RNF213 VS rnf213b: max score: 1901; total score: 2336; query cover: 97%; identity: 36.90%). Thus, we aimed to knock out rnf213a in zebrafish to explore the related phenotype. In our previous study, we formed rnf213a mutant chimera and found that chimeric embryos developed abnormal angiogenesis mimicking the characteristics of MMD.¹⁷ To further investigate the role of rnf213a in cranial vessels and reduce the potentially off-target effects, we successfully generated two rnf213a^−/−mutant lines including 2-bp insertion and 7-bp deletion based on Tg (flk1: eGFP/gata1: DsRed) background (Figure 2(b)).

Figure 2.

Highly conserved functional domains of RNF213 and the generation of rnf213a homozygous zebrafish. (a) Multi-alignment analysis of AAA+ATPases domain and RING finger domain sequences of human RNF213 gene, zebrafish rnf213a gene and zebrafish rnf213b gene. (b) Sanger sequencing results of +2bp mutant and −7bp mutant rnf213a homozygous zebrafish.

Zebrafish rnf213a mutations led to a mulberry-like cluster occurring in posterior cerebral vein

To uncover the role of rnf213a in zebrafish brain vessels, we did the imaging analysis of two rnf213a homozygous mutant lines (+2bp mutant and −7bp mutant). At upper middle layer, we observed that some abnormal sprouts emerged from the junction of posterior cerebral vein (PceV) and primordial hindbrain channel (PHBC) at 96hpf, and spread to connect basilar artery (BA) at 120hpf whether in +2bp mutant (49.4%, n= 42/85) or in −7bp mutant (46.4%, n= 39/84) (Figure 3(a)). More interestingly, at top layer, not only +2bp mutant (69.4%, n= 59/85) but −7bp mutant (71.4%, n= 60/84) exhibited a mulberry-like cluster in PceV at 72hpf and did not degenerate even by 120hpf (Figure 3(b)). To better understand dynamic circulation in mulberry-like cluster, we examined red cells flow labeled with red florescence using live confocal imaging and found that the mulberry-like cluster displayed disordered red cells flow in the two rnf213a mutant lines (Figure 3(c), Supplementary Movies 1 to 3). The major head vessels in rnf213a^−/− mutants presented no circulation defects (Figure 3(c), Supplementary Movie 4). No obvious vascular dilation including PceV, basal communicating artery (BCA), posterior communicating segment (PCS), BA, PHBC, dorsal aorta (DA) and posterior cardinal vein (PCV) were observed in rnf213a^−/− zebrafish (Figure 3(c) and (d)). However, subintestinal vessels in + 2bp mutant (42.7%, n= 35/82) or in −7bp mutant (39.5%, n= 34/86) displayed some disorganized sprouts but no expanded branching compared to wild-type embryos at 96hpf (Figure 3(d)).

Figure 3.

Phenotype analysis for cranial vessels and subintestinal vessels in rnf213a^−/− mutant and wild-type zebrafish. (a) Two mutant lines of the rnf213a^−/− embryos had abnormal brain vessel angiogenesis at upper middle layer from 72hpf to 120hpf. White boxes indicate abnormal sprouts emerged from the junction of posterior cerebral vein (PceV) and primordial hindbrain channel (PHBC), and spread to connect basilar artery (BA). (b) Both +2bp mutant and −7bp mutant exhibited a mulberry-like cluster in PceV and did not degenerate even by 120hpf. Red boxes indicate a mulberry-like cluster. PceV: posterior cerebral vein; DLV: dorsal longitudinal vein; MsV: mesencephalic vein. (c) Mulberry-like vessels in rnf213a^−/− mutants displayed disordered red cells flow but the major cranial vessels in rnf213a^−/− mutants had no circulation defects. Arrowheads indicate disordered red cells flow. BCA: basal communicating artery; PCS: posterior communicating segment; BA: basilar artery; PHBC: primordial hindbrain channel. (d) rnf213a^−/− embryos viewed laterally revealed some disorganized sprouts in subintestinal vessels by comparison with wild-type embryos. Arrows indicate disorganized sprouts. DLAV: dorsal longitudinal anastomotic vessel; Se: intersegmental vessel; DA: dorsal aorta; PCV: posterior cardinal vein; SIV: subintestinal vein.

Transcriptomic analysis

To explore the downstream molecules of rnf213a, we conducted transcriptome sequencing to detect differentially expressed genes (DEGs) between +2bp rnf213a^−/− mutant and wild-type zebrafish at 72hpf and 120hpf when the phenotype appeared. After removing technical sequences, clean data produced approximately 47,550,124 total reads from a wild-type group at 72hpf, 48,199,602 total reads from a +2bp mutant group at 72hpf, 60,562,166 total reads from a wild-type group at 120hpf, 65,579,532 total reads from a +2bp mutant group at 120hpf, respectively. The total mapping rates were 85.22%, 85.28%, 82.69%, and 82.92%, respectively. The RNA-Seq data generated in this study have been deposited to NCBI under the accession no.SRR10436022-10436029.

The differentially expressed genes between +2bp mutant and wild-type zebrafish at 72hpf and 120hpf were shown at volcano plots (Figure 4(a1) and (a2)). By comparing +2bp mutant with wild type at 72hpf, 277 transcripts were affected, of which 41 were up-regulated and 236 down-regulated. What’s more, 235 differentially expressed genes were identified between +2bp mutant and wild type at 120hpf. Of these DEGs, 91were upregulated and 144 were downregulated. Then we compared the DEGs between 72hpf and 120hpf, and identified 95 common DEGs.

Figure 4.

Transcriptomic analysis and confirmation by RT-qPCR. Volcano plots of the overall significant DEGs (differentially expressed genes) between +2bp mutant and wild type at 72hpf (a1) and 120hpf (a2). Blue dots represent down-regulated genes. Red dots represent up-regulated genes. RT-qPCR analysis on several genes of concern between wild-type and +2bp mutant embryos at 72hpf (b1) and 120hpf (b2). RT-qPCR analysis on several genes of concern between wild-type and −7bp mutant embryos at 72hpf (c1) and 120hpf (c2). Data are presented as mean ± SD, n = 3 repeats (40 embryos per repeat). *P < 0.05.

RT-qPCR validation

The transcriptomic data were validated by RT-qPCR analysis for several genes of concern, which were selected based on human ortholog and the function of zebrafish common DEGs. These selected genes included kbtbd7, mal, gstt2, anxa6, nub1 and nudt13, and the transcriptomic data of the genes above are shown in Table 2. Gene mRNA levels from the RT-qPCR assays (+2bp mutant vs. wild type) showed similar results to the transcriptomic results (Figure 4(b1) and (b2)), indicating the reliability of the high-throughput transcriptomic sequencing results. To select the most specific downstream genes, these genes of concern were further validated by RT-qPCR analysis between −7bp mutant and wild-type zebrafish (Figure 4(c1) and (c2)).After RT-qPCR analysis (−7bp mutant vs. wild type), some genes including kbtbd7, mal and anxa6 were confirmed to have expression patterns similar to the RT-qPCR results (+2bp mutant vs. wild type). Taken together, we showed that the two genes (kbtbd7 and anxa6) were significantly downregulated in rnf213a^−/− mutant compared to wild type, suggesting that kbtbd7 and anxa6 were the major downstream molecules of rnf213a.

Table 2.

Transcriptomic information of the selected genes at 72hpf and 120hpf (+2bp mutant vs. WT).

72hpf				120hpf
GeneSymbol	logFC	P value	FDR	GeneSymbol	logFC	P value	FDR
kbtbd7	−1.05E+01	1.01E-11	6.33E-08	kbtbd7	−1.02E+01	5.92E-11	1.26E-07
mal	−7.84E+00	2.82E-05	5.20E-03	mal	−7.95E+00	1.80E-05	3.57E-03
gstt2	−3.06E+00	6.02E-05	9.18E-03	gstt2	−3.51E+00	2.35E-06	7.07E-04
anxa6	−5.91E+00	5.30E-12	6.33E-08	anxa6	−5.10E+00	2.41E-10	3.43E-07
nub1	−9.37E+00	8.67E-09	1.03E-05	nub1	−9.06E+00	5.90E-08	3.52E-05
nudt13	−3.73E+00	4.04E-05	7.02E-03	nudt13	−4.16E+00	7.00E-06	1.81E-03

Zebrafish rnf213a promoted mulberry-like cluster via downregulation of kbtbd7 and anxa6

To identify whether the downstream molecules (kbtbd7 and anxa6) were associated with the mulberry-like phenotype in rnf213a^−/− zebrafish, we injected the mRNAs of kbtbd7 or anxa6 to rnf213a mutant embryos and found that injection of 100 pg or 200 pg mRNA (kbtbd7 or anxa6) caused all rnf213a mutant embryos dead and 50 pg mRNA resulted in nearly half the deaths (Figure 5(b) and (c)). Thus, we reduced the amounts of mRNA and revealed that 13% (12/92) or 23.4% (22/94) embryos rescued completely (one cavity) and 69.6% (64/92) or 64.9% (61/94) embryos rescued partly (two cavities) after 30 pg kbtbd7 mRNA or anxa6m RNA injection, respectively (Figure 5). However, 17.4% (16/92) or 11.7% (11/94) embryos still exhibited more than three cavities representing mulberry-like phenotype after kbtbd7 or anxa6 mRNA injection, respectively (Figure 5). These results suggested kbtbd7 and anxa6 partly rescued mulberry-like phenotype in rnf213a mutant zebrafish.

Figure 5.

Mulberry-like phenotype in rnf213a mutant partly rescued by kbtbd7 and anxa6. (a) Phenotypes after mRNAs injection were classified as ≥3 cavities (not rescued), 2 cavities (partly rescued) and 1 cavity (completely rescued). (b–c) Different amounts of kbtbd7 or anxa6 mRNA were injected into rnf213a mutant embryos. Arrow indicates 1 cavity. Red box indicates a mulberry-like cluster.

Discussion

In the present study, we identified five affected individuals carrying six heterozygous deleterious mutations in RNF213 including frameshift or nonsense or predicted pathogenic missense mutations in 31 sporadic cerebral cavernous malformation patients, and then selected RNF213 as a candidate gene for sporadic cerebral cavernous malformation. To further confirm if RNF213 was associated with cerebral cavernous malformation, we generated rnf213a homozygous knockout zebrafish and found a mulberry-like cluster of disordered-flow vascular channels in rnf213a^−/− embryos. Results from transcriptomic sequencing and RT-qPCR demonstrated kbtbd7 and anxa6 were the downstream molecules of rnf213a. Combined with mRNA rescue results, kbtbd7 and anxa6 were associated with the mulberry-like phenotype due to the mutation of rnf213a. These results suggest RNF213 is a candidate gene for sporadic cerebral cavernous malformation.

Cerebral cavernous malformation can be sporadic or familial with the familial form linking to three CCM genes (KRIT1, CCM2 and PDCD10). Cerebral cavernous malformation is an autosomal dominant condition and heterozygous germline mutations in CCM genes were detected in approximate 90% familial cases.^4,5Some isolated cases with multiple lesions have been reported and 57% isolated multiple cases harbored a mutation of CCM genes.⁵ Moreover, a systematic analysis of 22 isolated cases with multiple lesions demonstrated that 75% of cases were in fact inherited through asymptomatic parents, indicating most isolated multiple cases were in fact familial.¹⁹However, sporadic cerebral cavernous malformation cases with single lesion had no mutation in CCM genes, indicating there may be additional genes for sporadic cerebral cavernous malformation.^6,7 To explore the additional genes in sporadic cerebral cavernous malformation, we performed whole exome sequencing in 31 sporadic cerebral cavernous malformation patients with single lesion and 32 healthy controls, and identified 6 heterozygous deleterious mutations of RNF213 in 5 affected individuals (16.1%). Moreover, there was no RNF213 mutation in control individuals. Thus, we demonstrated RNF213 was associated with sporadic cerebral cavernous malformation. A large genetic study in 255 Chinese MMD patients identified 27 rare mutations of RNF213 but all of them were missense mutations.²⁰ However, among the six rare mutations in our study, two frame-shift mutations and a nonsense mutation were predicted to generate a truncated protein and then resulted in RNF213 loss of function. Besides, two rnf213a^−/− mutant lines in zebrafish which caused rnf213a loss of function exhibited a phenotype reminiscent to human cerebral cavernous malformation. These results suggest RNF213 loss-of-function is responsible for sporadic cerebral cavernous malformation.

Cerebral cavernous malformation is vascular lesion characterized by a mulberry- or raspberry-like cluster of low-flow and enlarged cavities without intervening brain parenchyma.¹ Here, rnf213a^−/− zebrafish showed a mulberry-like cluster of disordered-flow vascular channels occurring in venous vascular bed, which was reminiscent of human cerebral cavernous malformation. In our previous study, we injected TALEN mRNAs into embryos forming rnf213a mutant chimera and found that chimeric embryos developed abnormal angiogenesis and circulatory disorder.¹⁷ Nevertheless, chimeric embryos in which cells with different genotype might cause unexpected phenotype. Thus, we further investigated the role of rnf213a in homozygous knockout zebrafish. In order to purify the potential off-target mutagenic effects, we outcrossed the heterozygous mutant to the wild type for three generations and then heterozygotes intercrossed to generate homozygotes. Moreover, we generated two rnf213a homozygous mutant lines (+2bp mutant and −7bp mutant) and found that +2bp mutant displayed identical phenotypes to −7bp mutant. Taken together, we identified the specificity of rnf213a and mulberry-like phenotype.

Classically, cerebral cavernous malformation lesions in humans exhibit a cluster of dilated blood vessels and the cluster gives the lesions an appearance likened to a mulberry or a raspberry.^21,22 Moreover, single dilated vessels, defined as capillary telangiectasias, are not cerebral cavernous malformation.²³ The major phenotype in ccm genes mutant or knockdown zebrafish was a dilated, thin-walled heart whose structure resembles cerebral cavernous malformation vessels.^24–26 Secondly, a thorough analysis of vasculature in ccm genes mutant zebrafish showed some dilated, thin-walled vessels.²⁷ In the current study, obvious vascular dilation was not observable in rnf213a mutant zebrafish. However, rnf213a^−/− zebrafish exhibited obvious mulberry-like cluster and occurred in venous vascular bed. Additionally, some disorganized sprouts indicating angiogenesis were observed in rnf213a^−/− subintestinal vessels, which was partly reminiscent of the phenotype in ccm genes knockout or knockdown zebrafish.^25,27 Thus, our results suggest rnf213a (zebrafish ortholog of human RNF213) is more relevant to angiogenesis phenotype of cerebral cavernous malformation compared to dilation phenotype.

MMD is characterized by the progressive stenosis of the internal carotid artery and an abnormal vascular network at the base of the brain.²⁸ Our results showed some abnormal sprouts emerged from the junction of PceV and PHBC, and spread to connect BA in rnf213a^−/− mutant zebrafish, which was similar to moyamoya vessels in MMD. The phenotype was firstly identified in two rnf213a homozygous mutant lines which diminished the off-target effects. Nevertheless, the major head vessels in rnf213a^−/− mutants presented no luminal stenosis, which did not resemble the stenosis of the carotid artery in MMD. The presentation of moyamoya vessels was known to be the collateral pathways due to progressive stenosis of the carotid artery terminations.²⁹ However, most atherosclerotic cerebrovascular disease patients with the same arteriostenosis as MMD have no moyamoya vessels in clinical work, indicating moyamoya vessels of MMD may not be attributed to compensation alone. In addition, two recent studies revealed that MMD strongly differed from the atherosclerotic cerebrovascular disease in terms of some pro- and anti-angiogenic factors.^30,31 Thus, our phenotype mimicking moyamoya vessels but no stenosis further suggest the moyamoya vessels are not just compensatory mechanism but may also involve some angiogenic factor which may be resulted from the mutation of RNF213. On the other hand, it was found that transient knockdown of rnf213 by morpholino in zebrafish caused abnormal intersegmental sprouting vessels.¹⁰What’s more, RNF213 was reported critical for survival of breast cancer⁶ and possibly other malignancies in the hypoxic tumour microenvironment.³² Our present study also revealed that mutation of rnf213a led to a mulberry-like angiogenesis and a moyamoya-like angiogenesis. Thus, RNF213 may be associated with angiogenic phenotype in many diseases.

In order to explore the downstream molecules of rnf213a, we firstly conducted transcriptomic sequencing for +2bp mutant and wild-type zebrafish and detected some significant DEGs. To identify the specificity of downstream molecules, the selected DEGs by transcriptomic sequencing were validated not only between +2bp mutant and wild type but also between –7bp mutant and wild type by RT-qPCR, and then confirmed kbtbd7 and anxa6 as the downstream molecules of rnf213a. Based on the mulberry-like phenotype partly rescued by mRNA of kbtbd7 as well as anxa6, we therefore reason that rnf213a promote mulberry-like angiogenesis via downregulation of kbtbd7 and anxa6.

There are some limitations in our present study. Firstly, we only recruited Han Chinese populations. Ethnic differences have been reported in cerebral cavernous malformation genetics.^4,33,34In addition, compared with Japanese and Korean MMD patients, the rate of RNF213 p.R4810K mutation in Chinese MMD patients is lower.^10,35 These suggest both RNF213 gene and cerebral cavernous malformation are ethnically diverse. Further studies should expand sample size and select other ethnical populations to identify that RNF213 is a candidate gene of sporadic cerebral cavernous malformation. The second limitation may be the accurate functional tests for RNF213 mutations identified in affected patients. Our further work would generate the mutant zebrafish corresponding to RNF213 mutations in patients to confirm if they would cause the mulberry-like phenotype. Thirdly, our analysis shod is not extended for adult zebrafish. Zebrafish emerging as a powerful model for some diseases are attributed to its transparency. However, it is hard for zebrafish to remain transparent to adult period even treated with 1-pheyl-2-thiourea.³⁶ Over the past few years, a new method (CLARITY) has emerged to increase tissue transparency,³⁷ which might be useful to adult zebrafish but the method requires complex procedures and excellent imaging techniques. In the future, we might try to overcome the difficulties of the method to extend analysis shod to adult zebrafish.

In summary, we showed herein that 16.1% of sporadic cerebral cavernous malformation cases with a single lesion were associated with deleterious heterozygous mutations of RNF213, and firstly identified RNF213 as a candidate gene for sporadic cerebral cavernous malformation. Together with a mulberry-like cluster which was reminiscent of human cerebral cavernous malformation in rnf213a mutant zebrafish, our findings suggest RNF213 is a novel candidate gene for sporadic cerebral cavernous malformation.

ORCID iDs

Jing Lin https://orcid.org/0000-0001-6272-1039

Jie Liang https://orcid.org/0000-0001-7606-9088

Jun Wen https://orcid.org/0000-0002-9037-3526

Man Luo https://orcid.org/0000-0002-0543-9763

Jiaoxing Li https://orcid.org/0000-0001-6351-8833

Xunsha Sun https://orcid.org/0000-0002-4018-4188

Xiaowei Xu https://orcid.org/0000-0001-8790-2450

Jianli Li https://orcid.org/0000-0003-3106-7227

Dongxian Wang https://orcid.org/0000-0003-4820-796X

Jie Wang https://orcid.org/0000-0001-8499-5866

Huimin Chen https://orcid.org/0000-0003-0329-8175

Rong Lai https://orcid.org/0000-0002-4025-6778

Fengyin Liang https://orcid.org/0000-0002-1667-3678

Chuan Li https://orcid.org/0000-0001-5147-6699

Fei Ye https://orcid.org/0000-0002-2359-1486

Jingjing Zhang https://orcid.org/0000-0002-8789-4638

Shulan Yang https://orcid.org/0000-0002-6864-2195

Wenli Sheng https://orcid.org/0000-0001-6867-371X