Intracranial aneurysm (IA) is a major cause of hemorrhagic stroke. Acute cerebrovascular IA events pose a notable threat to younger individuals, leading to a decrease in productive years and a tremendous burden on society (Johnston, Selvin & Gress, 1998; Korja & Kaprio, 2016; Lawton & Vates, 2017; Rivero-Arias, Gray & Wolstenholme, 2010). Systematic reviews and meta-analyses indicate that the overall worldwide prevalence of unruptured IA is 3.2% (95% CI [1.9–5.2]) (Vlak et al., 2011). Individuals with a family history of aneurysmal subarachnoid hemorrhage are at a higher risk of unruptured IA and aneurysmal subarachnoid hemorrhage (Bor et al., 2008; Bor et al., 2014; Vlak et al., 2011), indicating that genetics may play a contributing factor in IA. To date, the underlying mechanisms by which genetic factors contribute to IA remain poorly understood (Nakaoka et al., 2014).
A previous genome-wide association study identified a network of disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS) associated with IA (Arning et al., 2012). Members of the ADAMTS protein family have multiple biological functions related to the structure and remodeling of the brain’s arterial wall, including extracellular matrix degradation (Binder et al., 2017), modulation of endothelial cell angiogenesis (Tang et al., 2017), vascular smooth muscle cell migration (Wang et al., 2009), and inflammation (Lemarchant et al., 2016; Zhang, Lin & Wei, 2015). ADAMTS-like genes (ADAMTSLs) produce proteins similar to those encoded by ADAMTS genes, only without catalytic domains, and may be responsible for regulating activities of ADAMTS (Apte, 2009; Dubail & Apte, 2015). ADAMTSL1 was first identified in 1997 and was called punctin (Blobel, 1997; Hirohata et al., 2002). ADAMTSL1 is distinct from other genes in the ADAMTS family, and binds to the extracellular matrix in a spatially-specific manner (Hirohata et al., 2002). Because IA development is caused by changes such as the degeneration of the extracellular matrix (Sawyer et al., 2016), it is possible that ADAMTS protein dysfunction may contribute to the development of IA, but the exact mechanisms are unclear.
In the current study, we aimed to determine the relationship between ADAMTS genes and IA using multiple approaches, including sequence polymorphism, expression, and methylation of genes. Our results provided several testable hypotheses to guide future research.
Material and Methods
All experimental protocols were in compliance with the Declaration of Helsinki and were approved by the Institutional Review Boards and Ethics Committees of Tianjin Medical University General Hospital (IRB2019-KY-134) and Fuzhou Second Hospital Affiliated to Xiamen University (SQ2018-004). All subjects or their legal guardians gave written informed consent.
Determination of deleterious variants in ADAMTS genes
This study included 20 Han Chinese IA patients from 11 families with two or more affected members in each family. The cerebral aneurysm was confirmed using digital subtraction angiography or computed tomography angiography. Whole-genome sequencing (WGS) and whole-exome sequencing (WES) were both conducted in 10 patients, respectively (Fig. 1, Data S1). Sequencing data that met quality criteria were analyzed for deleterious variants. The potential harmfulness of variants within exonic or splicing regions was predicted using SIFT (Vaser et al., 2016), Polyphen (Adzhubei, Jordan & Sunyaev, 2013), MutationTaster (Schwarz et al., 2010) and CADD (Rentzsch et al., 2018). Variants were considered deleterious if they were deemed harmful by any of the algorithms. ADAMTS genes were screened for deleterious variants using data from non-IA controls in the Genome Aggregation Database (gnomAD, http://gnomad.broadinstitute.org/) (Zlotogora, Patrinos & Meiner, 2018). Fisher‘s exact test was used to determine significant differences in the frequencies of deleterious ADAMTS variants between our IA patients and the gnomAD control data.
Identification of differentially expressed ADAMTS genes using microarray data from the GEO database
We collected IA tissue mRNA expression data from 15 patients and control cerebral artery tissues from 15 individuals in the GEO database (http://www.ncbi.nlm.nih.gov/geo/GSE75436; Table S1). Differences in mRNA levels between the two groups were compared using the online analysis tool GEO 2R (http://www.ncbi.nlm.nih.gov/geo/geo2r/?acc=GSE75436) (Davis & Meltzer, 2007). Differentially expressed genes were defined as P < 0.05 and —logFC— ≥0.5, where FC is the fold change in gene expression level. All differentially expressed genes in the ADAMTS family were screened and presented in a heat map.
Differential methylation of ADAMTS genes in IA and cerebral artery tissues
High-throughput methylation data from IA and cerebral artery tissues were obtained from the GEO database (GSE75434) (Yu et al., 2017). This set of data included nine IA tissues and nine matched cerebral artery tissues from different patients. GEO 2R was used to identify differentially methylated sites between IA and cerebral artery tissues. P < 0.05 was regarded as statistically significant.
Predicting ADAMTS protein-protein interactions in IA
We used the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING, version 11.0; https://string-db.org/) to predict protein interactions. We analyzed the subset of ADAMTS genes that contained deleterious variants, that were differentially expressed, and were methylated in IA.
Analysis of ADAMTSL1 expression in IA and cerebral artery tissues
IA specimens were donated by three patients who underwent clipping of IA at Fuzhou Second Hospital Affiliated to Xiamen University. Control cerebral artery samples were taken from autopsies performed in the Department of Pathology at Fuzhou Second Hospital Affiliated to Xiamen University. All specimen donors provided written informed consent. The study protocol was approved by the hospital’s Ethics Committee (SQ2018-004).
Tissue samples were homogenized and 20 µl was fractionated using SDS-PAGE (5% stacking gel, 10% separating gel). Broad-range pre-stained protein markers were also separated, and electrophoresis was conducted for 1.5–2 h in a buffer containing Tris-aminomethane, glycine, and 0.1% SDS. After electrophoresis, proteins were transferred onto polyvinylidene difluoride membranes as previously described. Block Ace (orb436742; Biorbyt, Cambridge, UK) was included to determine non-specific binding. The transferred membranes were incubated overnight at 4 °C with a primary rabbit antibody against human ADAMTSL1 (1:1,000; ab155597, Abcam). The membranes were washed extensively, then incubated for 1 h at room temperature with a universal biotinylated streptavidin-HRP secondary antibody against rabbit IgG (ab97051, Abcam). Bands were detected and quantitated using the ChemiDoc™ MP System (170-8280, Bio-Rad, USA).
Determining differing ADAMTS SNVs between sporadic IA patients and control subjects in a Han Chinese population
Clinical information and blood samples from 595 IA patients were obtained from CMAD (http://database.cmadtj.com/). IA was confirmed using subtraction or computed tomography angiography. A total of 600 control subjects were selected from a database in the Physical Examination Center of Tianjin Medical University General Hospital. The control subjects’ medical records were checked to ensure that they had no indication of IA.
Genomic DNA was extracted from the blood samples and stored at −80 °C. The SNV rs11750568 in ADAMTS2 and the SNVs rs2301612 and rs2285489 in ADAMTS13, previously reported (Arning et al., 2016), were genotyped using the Sequenom MassArray system (BioMiao Biological Technology, Beijing, China). Data were analyzed using MassArray Typer 4.0 (Agena Bioscience, San Diego, USA). The association between ADAMTS2 and ADAMTS13 variants and IA was examined using Plink 2.0 (Supplemental Information) (Purcell et al., 2007).
Differences in mRNA levels and methylation sites were compared between the two groups using GEO 2R, and P < 0.05 was considered potentially differentially expressed or methylated. The Benjamini & Hochberg multiple correction method (false discovery rate) was used (Madar & Batista, 2016). Protein levels, determined by Western blot, were compared using Student’s t-test in SPSS 22.0 (64-bit edition; IBM, Chicago, IL, USA). Chi-squared test and multiple-test correction for analyses of rs11750568, rs2301612, and rs2285489 were performed in Plink 2.0. P < 0.05 was regarded as statistically significant.
Deleterious variants in the ADAMTS gene family
A total of 16 variants in the ADAMTS family were found in exons and other functional areas (Table 1). Further comparisons against gnomAD control data showed that these deleterious variants were enriched in the IA patients from 11 families in the present study (Table 2).
|ID||REF||ALT||Amino acid change||Gene||Function||Prediction tool|
|SNP_ID||Gene||Polymorphic locus||Our study||GnomAD||P (Fisher’s exact test)||OR||95% CI|
|risk allele/ normal allele||risk allele/ normal allele||Lower||Upper|
genome aggregation database
single-nucleotide polymorphism identification
ADAMTS genes differentially expressed between IA patients and controls
The online analysis tool GEO 2R was used to analyze data collected from GEO. Among the 16 potentially differentially expressed ADAMTS genes, ADAMTS9-AS1, ADAMTS8, ADAMTS9-AS2, ADAMTS1, ADAMTS9, ADAMTS15, ADAMTS4, ADAMTSL4, and ADAMTSL1 were upregulated in IA; and ADAMTS3, ADAMTS17, ADAMTS13, ADAMTSL3, ADAMTS7, ADAMTS19, and ADAMTS2 were downregulated. The changes remained significant after multiple testing corrections for ADAMTS9-AS1, ADAMTS8, ADAMTS9-AS2, ADAMTS1, ADAMTS13, ADAMTS7, ADAMTS19, and ADAMTS2 (Table 3). Differentially expressed ADAMTS genes are shown in Table 3 and Fig. 2A.
|Gene symbol||Adjusted P value||p||t||log FC|
|ADAMTS9-AS1||2.88 × 10−4||<0.001||5.433435||3.296323|
|ADAMTS8||2.63 × 10−3||<0.001||4.393505||2.777007|
|ADAMTS9-AS2||1.58 × 10−5||<0.001||6.82637||1.786802|
|ADAMTS1||5.44 × 10−3||<0.001||4.049965||1.650133|
|ADAMTS9||1.38 × 10−1||0.026||2.330043||1.404467|
|ADAMTS15||1.21 × 10−1||0.022||2.411667||1.096499|
|ADAMTS4||2.04 × 10−1||0.047||2.063526||0.751622|
|ADAMTSL4||1.58 × 10−1||0.032||2.241619||0.708539|
|ADAMTSL1||1.42 × 10−1||0.028||−2.311788||−0.678263|
|ADAMTS3||1.05 × 10−1||0.018||−2.49951||−0.76742|
|ADAMTS17||7.83 × 10−2||0.012||−2.67964||−0.77946|
|ADAMTS13||2.52 × 10−2||0.002||−3.29649||−1.24966|
|ADAMTSL3||5.51 × 10−2||0.007||−2.87834||−1.36329|
|ADAMTS7||3.74 × 10−2||0.004||−3.09323||−1.44272|
|ADAMTS19||2.39 × 10−2||0.002||−3.32586||−1.77612|
|ADAMTS2||8.23 × 10−9||2.86 × 10−12||−11.0591||−2.87598|
- log FC
log fold change
statistic from Student’s t test
The data sourced from GEO database (http://www.ncbi.nlm.nih.gov/geo/GSE75436). Differentially expressed genes identified with multiple correction are shown in bold.
Sites in ADAMTS genes differentially methylated between IA and cerebral artery tissues
A comparison between the IA samples and the cerebral artery revealed a total of 299 potentially differentially methylated sites in 24 ADAMTS genes. Among these sites, 193 were upregulated and 106 were downregulated (Table S2).
The subset of ADAMTS genes with deleterious variants, differential expression, and differential methylation in IA
Seven ADAMTS genes were found to contain deleterious genetic variants and were differentially expressed and methylated in IA: ADAMTS15, ADAMTS2, ADAMTS4, ADAMTS7, ADAMTS8, ADAMTSL1, and ADAMTSL3 (Fig. 2B). Protein-protein interaction prediction suggested that ADAMTSL1 was at the center of the network of all seven genes (Fig. 2C).
Lower expression of ADAMTSL1 in IA issue
Based on the prediction that ADAMTSL1 plays a key role in ADAMTS genes in IA, we investigated its expression in IA and cerebral artery tissue (Fig. 3). The results showed lower ADAMTSL1 levels in IA tissue than in a cerebral artery.
Determining specific SNPs in sporadic IA patients and control individuals from a Han Chinese population
The following ADAMTS gene variants were not associated with IA in a cohort of Han Chinese patients and controls even after multiple corrections (Table 4): allele A in rs11750568, OR 0.9696 (95% CI [0.7464–1.26], P = 0.82); allele T in rs2285489, OR 0.93 (95% [0.7309–1.187], P = 0.56); and allele G in rs2301612, OR 0.96 (95% [0.77–1.20], P = 0.71).
A total of 16 deleterious variants in 13 ADAMTS genes were found to be associated with IA (Table 2). Sixteen potentially differentially expressed genes were discovered from transcriptomics data and 299 potentially differentially methylated sites in 24 ADAMTS genes were identified from methylation sequencing data taken from IA and cerebral artery samples in the GEO database. An overlapping set of seven ADAMTS genes were found to contain deleterious variants and were differentially expressed and methylated in IA: ADAMTS15, ADAMTS2, ADAMTS4, ADAMTS7, ADAMTS8, ADAMTSL1, and ADAMTSL3. A genome-wide association study and other molecular biology studies have recognized ADAMTS15 as a candidate gene for IA in a Japanese population (Yan et al., 2015). Surprisingly, this gene did not show a significant association with IA in our cohort (P = 0.057; Table 2). This negative result is likely due to a lack of statistical power, and should be verified in larger cohorts. The ADAMTS2 variant rs11750568 has been previously associated with IA and pediatric stroke (Arning et al., 2012; Arning et al., 2016). ADAMTS4 protein and mRNA are expressed at higher levels in thoracic aortic aneurysm and dissection tissues than in control aortic tissues, and increased ADAMTS4 levels can degrade versican and facilitate macrophage invasion (Ren et al., 2013). Human aortic aneurysm induction is related to upregulated ADAMTS-7 and downregulated COMP in the ADAMTS7/COMP pathway. In patients with peripheral arterial occlusion, levels of ADAMTS8 and macrophages in the blood are lower if an aortic aneurysm is present (Lamblin et al., 2010). ADAMTSL3 is a candidate gene for diabetes, which is, in turn, a risk factor for IA (Jambaljav et al., 2018; Lindgren et al., 2013).
A previous study of ADAMTS gene polymorphisms and IA risk in a European population identified three risk alleles: allele A at rs11750568 in ADAMTS2 (OR 1.32, P = 0.006), allele T at rs2301612 (OR 1.26, P = 0.011), and allele G at rs2285489 in ADAMTS13 (OR 1.24, P = 0.02) (Arning et al., 2016). In our present study of a Han Chinese population, there were no significant differences between the risk alleles of the IA patients and controls after multiple testing corrections (Table 4). One possible explanation for this is that risk alleles for IA may differ among different populations. It is also possible that the current or previous studies’ designs were underpowered.
The ADAMTSL gene ADAMTSL1 was expressed at lower levels in IA tissue than in the cerebral artery in both our patients and in data from GEO (Table 3). ADAMTSL1 is well-positioned to have a substantial influence on IA development since it is predicted to be at the center of the protein-protein network (Fig. 2C). The resemblance of ADAMTSL proteins to ADAMTS proteases and their matrix binding properties indicate a potential function in ADAMTS regulation (Apte, 2009). Therefore, we speculate that ADAMTSL proteins, including ADAMTSL 1 through 6 and papilin, may act as upstream regulators of ADAMTS proteins (Apte, 2009; Dubail & Apte, 2015; Kelwick et al., 2015). Additional experiments are needed for further verification. ADAMTSL1’s binding to the extracellular matrix (Hirohata et al., 2002) may influence the degradation of extracellular matrix levels, which may contribute to IA development (Sawyer et al., 2016). Lower levels of ADAMTSL1 mRNA and protein in IA tissue may be associated with differential methylation (Yong, Hsu & Chen, 2016). Our results suggest that ADAMTSL1 may regulate the influence of ADAMTS genes in IA. However, this speculation must be tested directly in future studies.
There are some limitations to this study. First, although we present evidence that ADAMTS are novel candidate genes associated with IA, these findings should be verified and explained using additional mechanistic studies. Second, we did not conduct experiments to directly verify whether ADAMTSL1 influences the levels or activities of ADAMTs genes. Third, the associations between IA and ADAMTS variants should be explored in larger and more ethnically diverse samples.
IA development is associated with genetic variants, differential expression, and abnormal methylation of ADAMTS genes, specifically ADAMTSL1.