We recruited MMD patients (≥18 years old and ≤70 years old, male: female = 1:1) without previous medical history. Diagnostic criteria were based on the Japanese Research Committee on moyamoya disease of the Ministry of Health, Welfare and Labour, Japan (RCMJ) criteria (Research Committee on the Pathology and Treatment of Spontaneous Occlusion of the Circle of Willis and Health Labour Sciences Research Grant for Research on Measures for Infractable Diseases, 2012). Their clinical diagnosis was confirmed by imaging with transcranial computed tomography (CT), magnetic resonance imaging (MRI), or digital subtraction angiography (DSA) along with various clinical judgments. Fasting samples of venous blood were collected from all patients and healthy control subjects separately during the same period. All subjects signed the consent form prior to entering the trial.

Wild-type and mutant templates were designed with reference to the known mutation detection loci for the specific gene. Amplimers and crRNAs were then designed for the known mutation regions and oligonucleotides (crDNA) were synthesized. crDNA and cr-T7-F were mixed and boiled for 10 min, then the double-stranded transcription template being formed by natural cooling. The transcription template was then incubated for 16 h at 37°C under enzymatic-free conditions using the HiScribe T7 Quick High Yield RNA Synthesis Kit (NEB, Ipswich, United States). After the completion of the reaction, 2 μL of DNase 1 (TianGen, Beijing, China) was added to eliminate unreacted template before purifying the crRNA. Wild-type and mutant template sequences, amplimers, and crDNAs, were synthesized by Tianyi Huiyuan Biotechnology Co., Ltd. (Supplementary Table S1).

Fncas12a uses 5′-KYTV-3′ 999 as protospacer adjacent motif (PAM). It was chosen as the detection protein for providing more target sequence options compared with Ascas12a and Lbcas12a (Tu et al., 2017). In brief, 50 ng of template DNA was added into the detection reagent mixture containing 0.75 μM crRNA, 1.5 μM Fncas12a, 50pM of fluorescent probe, and 3 μL of NEBuffer 3.1 (NEB, Ipswich, United States). Reactions (50 μL) were then incubated at 37°C for 1h prior to fluorescence quantification. All reactions were carried out at 37°C.

Following plasma separation, DNA was extracted from venous blood samples. Thereafter, polymerase chain reactions (PCR) were performed using 50 ng of DNA as the template with specific primers (Supplmentary Table S1) at the following cycle conditions: 95°C for 5 min; 30 cycles of 95°C for 3 min; 56°C for 10 s, and 72°C for 20 s; followed by 72°C for 5 min. PCR products were then visualized by agarose gel electrophoresis and were sequenced using the Sanger method. In parallel, 1–5 μl of amplified product was used for CRISPR-Cas12a fluorescence detection.

We collected five samples from the familiy with clinical manifestations of the MMD phenotype from Liaocheng People’s Hospital Center from June 2020 to December 2020 (Table 1). All five family members were subjected to WES as depicted in the flow chart in Supplementary Figure S1. This study was approved by the ethics committee of Liaocheng People’s Hospital, Shandong Province. Informed consent for DNA analysis was obtained from patients in line with local Institutional Review Board (IRB) requirements at the time of collection.

DNA extracted from peripheral blood was fragmented to an average size of 180–280 bp and subjected to DNA library creation using established Illumina paired-end protocols. The Agilent SureSelect Human All ExonV6 Kit (Agilent Technologies, Santa Clara, CA, United States) was used for exome capture according to the manufacturer’s instructions. The Illumina NovaSeq 6,000 platform (Illumina Inc., San Diego, CA, United States) was utilized for genomic DNA sequencing in Novogene Bioinformatics Technology Co., Ltd. (Beijing, China) to generate 150-bp paired-end reads with a minimum coverage of 10× for 99% of the genome (mean coverage of 100×).

After sequencing, base-call file conversions and demultiplexing were performed with bcl2fastq software (Illumina). The resulting fastq data were submitted to in-house quality control software to remove low quality reads; and these were then aligned to the reference human genome (hs37d5) using the Burrows-Wheeler Aligner (bwa) (Li and Durbin, 2009). Duplicate reads were marked using sambamba tools (Tarasov et al., 2015). Single nucleotide variants (SNVs) and indels were identified by samtools to generate Genome VCF (gVCF) (Li et al., 2009). Raw calls for the SNVs and INDELs were further filtered with the following inclusion thresholds: 1) a read depth > 4; 2) a root-mean-square mapping quality of covering reads that was > 30; and 3) a variant quality score > 20. Copy number variants (CNVs) were detected with CoNIFER software (Version 0.2.2) (Krumm et al., 2012). Annotation was performed using ANNOVAR (2017) (Wang et al., 2010). Annotations included minor allele frequencies from public control data sets as well as deleteriousness and conservation scores, thus enabling further filtering and assessment of the likely pathogenic variants.

Filtering for rare variants was performed as follows. First, variants with a MAF < 0.01 in 1000 genomic data (1000g_all) (Auton et al., 2015), esp6500siv2_all, and gnomAD data (gnomAD_ALL and gnomAD_EAS); (Kim, 2016) only SNVs occurring in exons or splice sites (splicing junction 10 bp) were further analyzed since we were targeting amino acid changes; (Deng et al., 2021) synonymous single nucleotide variants (SNVs) which were not relevant to the amino acid changes predicted by dbscSNV were discarded; the small fragment non-frameshift (<10bp) indel in the repeat region defined by RepeatMasker was discarded; and (Zhao et al., 2018) variations were screened according to SIFT scores (Kumar et al., 2009), PolyPhen (Adzhubei et al., 2010), MutationTaster (Schwarz et al., 2010) and CADD (Kircher et al., 2014) software packages. Potentially deleterious variations were reserved if the scores from more than half of the four software packages identified the variations as harmful (Muona et al., 2015). Sites (>2bp) that did not affect alternative splicing were also removed. To better predict the harmfulness of each variation, we applied the classification system put forward by the American College of Medical Genetics and Genomics (ACMG). The variations were classified as pathogenic, likely to be pathogenic, of uncertain significance, likely to be benign, or benign (Richards et al., 2015). Depending upon various considerations (pedigree, homozygous, and compound heterozygous), variants were considered to be candidate causal variations. The relationship between the proband and the parents was estimated using the pairwise identity-by-descent (IBD) calculation in PLINK (Purcell et al., 2007). The share of IBD between the proband and parents for all trios ranged from 45 to 55%.

SPSS 17.0 software was used for statistical analysis. The qualitative data and the number of cases described in percentage, and the quantitative data were compared by independent sample t-test or analysis of variance. p < 0.05 indicates a significant difference.

Literature searches identified RNF213 as a susceptibility gene for MMD. In addition, two SNP loci of RNF213, rs112735431 and rs148731719 have been confirmed closely related to MMD (Liu et al., 2011; Zhang et al., 2017; Wang Y. et al., 2020). crRNA was designed to detect these two SNP point mutation loci in RNF213. The cleavage efficiency of the crRNAs was then verified against wild-type and mutant-target DNA (Supplementary Figure S3). The fluorescence levels derived from the mutant were significantly higher than the wild type (p < 0.05), indicating that the CRISPR-Cas12a system constructed with the indicated crRNAs could successfully detect whether there was a mutation at this locus in clinical samples.

We collected 34 samples of patients who had been clinically diagnosed with MMD and 37 healthy control samples from Liaocheng People’s Hospital. DNA was extracted from these samples and the RNF213 gene of samples was tested using the CRISPR along with Sanger sequencing (Supplementary Table S2). The coincidence rate of the CRISPR-Cas12a system and Sanger sequencing for detecting mutation samples was 100%, indicating that the CRISPR-Cas12a detection is accurate and highly sensitive.

First, the results of the Sanger test for RNF213 showed that there was a C > T mutation at locus rs112735431 and a G > A mutation at locus rs148731719 in the RNF213 gene (Supplementary Figure S4A). T-tests showed that the p value for the rs112735431 locus mutation was < 0.05 when comparing between the case group and the healthy control group from the Liaocheng area. In contrast, there was no significant difference between the groups with respect to rs148731719 (p > 0.05) (Supplementary Table S3), indicating that the rs112735431 mutation within the RNF213 gene was significant (p < 0.05) and that the rs112735431 was a significant mutation locus for MMD in the RNF213 gene.

Then, we identified a base insertion mutation (rs3025058) in the MMP3 gene (Supplementary Figure S4B). This mutation was identified by Sanger sequencing and detected in 67.6% of the 34 patients with MMD in Shandong province, and 5.4% of the 37 controls, indicating statistical significance (p < 0.05). The 1171 (6A/6A) mutation in the MMP3 gene is associated with the risk of MMD. furthermore, the risk of the (6A/6A) genotype is higher than that of the (5A/6A) genotype (Supplementary Table S3.

The pedigrees of five samples and the results of the patient’s CT and CAT tests are shown in Figure 1. The average sequencing depth of the five samples exceeded 100×, and the coverage of regions > 10× exceeded 99%. The number of SNVs and Indels obtained from each sample after data analysis are shown in Supplementary Table S4.

Mutation loci were screened in accordance with the scores predicted by SIFT, PolyPhen, MutationTaster, and CADD. Candidate loci were further screened according to the process shown in Figure 2. The analysis identified multiple recessive pathogenic genes and notably, of these, loci mutation-related genes were within the TTN gene (rs771533925, rs559712998 and rs72677250) (Table 2).

The test results obtained by the CRISPR-Cas12a system for mutation loci in the TTN gene in family samples (Figure 3) were consistent with those obtained from WES sequencing (Table 3), thus verifying the presence of mutations in the samples.

Next, CRISPR-Cas12a system was used to test a total of 50 sporadic samples for gene mutations. No mutation was found at rs771533925, rs559712998 and rs72677250 of TTN gene in sporadic samples (Figure 4).

In addition, SIFT (Choi and Chan, 2015) PROVEAN (Vaser et al., 2016) and PolyPhen (Adzhubei et al., 2013) algorithms were used to predict the effects of amino acid substitutions on protein function (Table 4). All three databases showed that rs771533925 was potentially destructive. On the contrary, rs559712998 was considered tolerable according to these analyses. However, while rs72677250 was considered tolerable according to the SIFT database, it was considered to be potentially harmful according to the PROVEAN and PolyPhen databases.

We analyzed the risk alleles (rs72677250, rs559712998 and rs771533925) in accordance with the EXAC database. We identified significant differences in frequency across the global population. The highest frequency of rs72677250 in the South Asian population was 0.00003269, the highest frequency of rs559712998 in the East Asian population was 0.002574, the highest frequency of rs771533925 in the East Asian population was 0.00005568, and the total frequency of rs559712998 mutations was 0.000192; the latter being the highest frequency of all three mutation sites (Table 5). According to age analysis of these three loci within the global population, we found that the rs72677250 mutation site was predominant in subjects aged 50–55 years, the rs559712998 mutation site was predominant in subjects aged 30–80 years, and the rs771533925 mutation site was predominant in subjects aged 65–70 years (Figures 5A–C). GO analysis was then conducted using Cytoscape 3.8.2 software with the ClueGO (Bindea et al., 2009) plugin, showing that the mutation locus for TTN were involved in a range of important biological processes, including myosin thick filament assembly in skeletal muscle, positive regulation of protein transport, serine/threonine kinase activity, and cardiac muscle fiber development (Figure 5D).