COL5A1 Variants Cause Aortic Dissection by Activating TGF‐β‐Signaling Pathway

Abstract

Background

Aortic dissection (AD) is one of the most life‐threatening cardiovascular diseases that exhibit high genetic heterogeneity. However, it is unclear whether variants within the COL5A1 gene can cause AD. Therefore, we intend to determine whether COL5A1 is a causative gene of AD.

Methods and Results

We performed targeted sequencing in 702 patients with unrelated sporadic AD and 163 matched healthy controls using a predesigned panel with 152 vessel matrix‐related genes. As a result, we identified that 11 variants in COL5A1 caused AD in 11 out of the 702 patients with AD. Furthermore, Col5a1 knockout (Col5a1^+/− ) rats were generated through the CRISPR/Cas9 system. Although there was no spontaneous AD, electron microscopy revealed a fracture of elastic fibers and disarray of collagenous fibers in 6‐week‐old Col5a1^+/− rats, but not in WT rats (93.3% versus 0.0%, P<0.001). Three‐week‐old rats were used to induce the AD phenotype with β‐aminopropionitrile monofumarate for 4 weeks followed by angiotensin II for 72 hours. The β‐aminopropionitrile monofumarate and angiotensin II‐treated rat model confirmed that Col5a1^+/− rats had considerably higher AD incidence than WT rats. Subsequent mechanism analyses demonstrated that the transforming growth factor‐β‐signaling pathway was significantly activated in Col5a1^+/− rats.

Conclusions

Our findings, for the first time, revealed a relationship between variants in COL5A1 and AD via targeted sequencing in 1.57% patients with sporadic aortic dissection. The Col5a1 knockout rats exhibited AD after an intervention, indicating that COL5A1 is a causative gene of AD. Activation of the transforming growth factor‐β‐signaling pathway may be implicated in the pathogenesis of this kind of AD.

Nonstandard Abbreviations and Acronyms

AD

aortic dissection

AngII

angiotensin II

BAPN

β‐aminopropionitrile monofumarate

COL5A1

collagen type V alpha 1 chain

EDS

Ehlers‐Danlos syndrome

MMP9

matrix metallopeptidase 9

α‐SMA

alpha‐smooth muscle actin

SMAD2

SMAD family member 2

SMC

smooth muscle cell

TGF‐β1

transforming growth factor beta 1

WT

wild type

Clinical Perspective

What Is New?

• This study demonstrates an association between the COL5A1 gene and aortic dissection by next‐generation sequencing in a large sample of subjects with sporadic aortic dissection and Col5a1 knockout rats for the first time.

What Are the Clinical Implications?

• Activation of the transforming growth factor beta 1‐signaling pathway is likely to contribute to the pathogenicity of this kind of aortic dissection, and these results provide further insight into the genetic origin and potential therapeutic targets of aortic dissection.

Aortic dissection (AD) is one of the most common deadly aortic diseases¹, ² and has ≈1/10 000 annual incidence in Europe.³, ⁴ Although AD is a catastrophic sudden cardiovascular disease with high mortality, it does not exhibit apparent manifestation until dissection emerges.⁵, ⁶ Since the first AD case in 1760, many studies have investigated the risk factors for AD including smoking and genetic disorders.⁴, ⁷, ⁸

The large‐scale genome‐wide association studies have been unsuccessfully applied in identifying a specific causative gene for AD. However, high throughput next‐generation sequencing has shown promising breakthrough in identifying pathogenic genes in patients with AD.⁹, ¹⁰, ¹¹ It has been revealed that >25% of patients with AD exhibit the underlying pathogenic variants of AD causative gene before its onset.¹² Among the patients with AD, syndromic diseases is one of the major causes that accounts for >10% of cases.¹³ To date, at least 13 causative genes (TGFB2, TGFB3, TGFBR1, TGFBR2, SMAD3, SLC2A10, PLOD1, NOTCH1, MYLK, MYH11, FBN1, COL3A1, and ACTA2) have been identified to cause AD in syndromic disease. These syndromic diseases include Marfan syndrome, Loeys‐Dietz syndrome, vascular Ehlers‐Danlos syndrome (EDS), bicuspid aortic valve, and familial thoracic aortic aneurysm and dissection syndrome.¹⁴, ¹⁵, ¹⁶

The classic type of EDS is a rare autosomal dominant disease caused by a mutation in COL5A1 and COL5A2. The disease is characterized by fragile, hyperextensible skin, hypermobile joints, and poor wound healing. Different from the vascular form of EDS, vascular lesions including aneurysms and dissections do not usually manifest in classic EDS.¹⁷ In 2018, a study reported that the knockout of Col5a2 in mice resulted in aortic aneurysms and dissections.¹⁸ Furthermore, recent studies have reported some instances where patients with iliac or renal artery dissection carry COL5A1 variants.¹⁹, ²⁰, ²¹ However, the AD phenotype has not been observed in patients harboring COL5A1 variants.

To date, whether variants in COL5A1 and COL5A2 can cause AD is still unclear. Here, we investigate whether variants in the COL5A1 gene are the genetic cause of AD in 702 patients with sporadic AD and Col5a1 knockout rats.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Participants Enrollment

We enrolled 702 consecutive unrelated patients with sporadic AD and 163 ethnically and age‐matched controls with Han Chinese ancestry. Study of patients was conducted at the Tongji Hospital in Wuhan during a time period spanning the month of May 2008 and the month of May 2015. The study was approved by the Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology. An informed consent was obtained from the participants. All recruited patients with AD were diagnosed by computed tomography or magnetic resonance imaging. All the healthy controls were randomly recruited from healthy individuals undergoing health examinations in Tongji Hospital. All healthy controls were evaluated to be free of AD by medical history, physical examinations, and aorta imaging examinations (computed tomography or magnetic resonance imaging).

Panel Construction and Targeted Sequencing

A next‐generation sequencing panel, including 529.73 kb full coding regions of 152 vessel matrix‐related genes, was designed to identify potential genetic causes of AD as previously described (Table S1).²² Targeted sequencing was performed for all patients with AD and healthy controls according to the Ampliseq semiconductor next‐generation sequencing platform (Ion Torrent, Thermo Fisher, Carlsbad, CA) as per our previous description.²³

Bioinformatics Analysis

The raw data of the next‐generation sequencing were first processed using Ion Torrent platform‐specific software, Torrent Suite v5.0. The generated reads were then aligned to the hg19/GRCh37 human reference genome for call variants and coverage status analysis. Thereafter, all the variants were comprehensively annotated using the Ion Reporter software 5.0 (Thermo Fisher). Subsequently, the Human Gene Mutation Database and the ClinVar database were searched to identify pathogenic variants responsible for AD. The focus was mainly on rare potential pathogenic variants rather than common alleles with modest disease liability. Therefore, common variants (minor allele frequency ≥0.01) reported in the 1000‐Genomes‐Project, the Genome Aggregation Database, the Exome Sequencing Project, and Exome Aggregation Consortium database were excluded. To predict the pathogenesis of nonsynonymous variants, all nonsynonymous variants were scored using Polymorphism Phenotyping v2, Sorting Intolerant from Tolerant, and Mutation Taster. A potential pathogenic variant was predicted to be damaging if reported by at least 2 prediction softwares. The degrees of evolutionary conservation across multiple species of the missense variants were estimated with the phyloP score. All putative pathogenic variants were presumed to be absent in all 163 healthy controls and were validated by Sanger sequencing. The optimal sequence kernel association test was used to evaluate the association between COL5A1 and AD risk.²⁴

Sanger Sequencing Validation

Polymerase chain reaction amplification was performed using the Taq Hot Start version (TaKaRa, Japan) to eliminate false‐positive variants detected by next‐generation sequencing. Sanger sequencing was then performed for all putative pathogenic variants using the Big Dye v.1.1 terminator cycle sequencing kit and Applied Biosystems 3500xl capillary sequencer (Applied Biosystems, Foster City).

Animals

All animal care and investigation conformed to the Guide for the Care and Use of Laboratory Animals by the US National Institutes of Health (Publication No. 85‐23, revised 1996). This study was reviewed and approved by the Institutional Animal Research Committee of Tongji Medical College. The Col5a1‐flox (Col5a1^fl/fl ) and transgenic cre SD rats were established via the CRISPR/Cas9 system in the Beijing Laboratory Animal Research Center of the Chinese Academy of Medical Sciences. Col5a1^fl/fl rats were mated with the transgenic cre rats to create heterozygous Col5a1 knockout (Col5a1^+/− ) rats. All the experimental rats were housed at the animal care room at 25°C with 12/12 hour light/dark cycles. The rats were fed with a normal diet and adequate water during the investigation. The weight of each rat was recorded weekly. The genotypes of rats were identified using validation primers (Table S2).

Development of AD Rat Models

Col5a1^+/− and Col5a1^fl/fl SD rats aged 3 weeks (>8 rats in each group) were fed on water with low‐dose β‐aminopropionitrile monofumarate (BAPN) (Sigma‐Aldrich, St. Louis, MO) added (0.1 g/kg per day) for 4 weeks.²⁵ Osmotic minipumps (Model 1003D; Alzet, Cupertino, CA) filled with 1 μg/kg per minute angiotensin II (AngII) (Sigma‐Aldrich) were implanted subcutaneously into 7‐week‐old rats. The rats in the control subgroup were fed on a normal diet and water, followed by using osmotic minipumps filled with normal saline. Blood pressure measurements were taken before and after implantation using the tail‐cuff method. The rats were euthanized 72 hours after the implantation. The aorta of each rat was isolated and photographed.

Histology and Immunofluorescence Analysis

The thoracic aorta and abdominal aorta tissues were harvested and cut into 4‐µm‐thick sections. The sections were fixed in 4% paraformaldehyde and embedded in paraffin. Hematoxylin and eosin staining, Masson's trichrome staining, and elastic Van Gieson staining were subsequently performed. In addition, the sections were immunostained with rabbit polyclonal antibody against alpha‐smooth muscle actin (α‐SMA) and mouse monoclonal antibody against elastin. The nucleus was stained using 2‐(4‐amidinophenyl)‐6‐indolecarbamidine dihydrochloride. The immunofluorescence images were observed via a laser confocal microscope (Olympus, FV500‐IX71, Tokyo, Japan).

Electron Microscopy

Small parts of the aorta tissue samples were fixed using 2% glutaraldehyde in 0.1 mmol/L cacodylate buffer. After dehydrating and embedding, ultrathin sections were stained. Collagen and elastin were observed using the Tecnai G20 TWIN electron microscope in 10 randomly selected areas, and ultrastructure analyses were performed in 3 rats in each group.

Western Blot Analysis

Western blots were performed as previously described.²⁶ Briefly, the aorta tissues were homogenized in ice‐cold lysis buffer, and centrifuged at 4°C 12 000g for 20 minutes. The lysates (25 μg protein per lane) were resolved by 10% SDS‐PAGE and then transferred to polyvinylidine difluoride membranes (0.45 μm). The membranes were blocked with 3% BSA and 5% nonfat milk for 2 hours and then incubated overnight with primary antibodies at 4°C. Afterward, peroxidase‐conjugated secondary antibodies were incubated with membranes for 2 hours. The Western blots were visualized with enhanced chemiluminescence.

Antibodies

Antibodies against collagen type V alpha 1 chain (COL5A1) (A1515), SMAD family member 2 (SMAD2) (A11498), phosphorylation SMAD family member 2 (AP0269), and elastin (A2723) were acquired from ABclonal (Wuhan, China); antibodies against matrix metallopeptidase 9 (MMP9) (10375‐2‐AP), transforming growth factor‐beta 1 (TGF‐β1) (21898‐1‐AP), GAPDH (10494‐1‐AP), and α‐SMA (55135‐1‐AP) were obtained from Proteintech (Wuhan, China); elastin (sc‐166543) was obtained from Santa Cruz Biotechnology (CA); secondary antibodies were acquired from Pierce Biotechnology (IL).

Statistical Analysis

Data are presented as mean±SD. The analyses of differences among the groups were performed using Student t test and 1‐way ANOVA, which were corrected by Bonferroni correction. In small‐sample group, permutation tests were used for correction. Statistical analysis was performed using SPSS software 20.0 and R (version 3.6.1, Vienna, Austria). The differences were considered to be statistically significant if 2‐tailed P<0.05.

Results

Baseline Characteristics

We studied the baseline characteristics of 702 patients with AD and 163 healthy controls. There were no significant differences between the 2 groups in terms of age (53.2±11.9 versus 54.5±11.4), proportion of men (78.1% versus 78.5%), proportion of hypertension (54.1% versus 54.8%), and proportion of smokers (45.6% versus 46.1%). These outcomes were consistent with our previous study results.²²

Next‐Generation Sequencing Quality

In the next‐generation sequencing, the mean base coverage depth for 702 patients with AD and 163 health controls was more than 700‐fold per sample. This was similar to our previous study.²²

Targeted Next‐Generation Sequencing Revealed That COL5A1 was Associated With AD

Targeted next‐generation sequencing was performed for 702 unrelated Han Chinese patients with AD and 163 matched healthy controls. A predefined AD panel including COL5A1 was used to identify the pathogenic and likely pathogenic variants. As a result, 11 likely pathogenic variants in COL5A1 were identified in 11 patients with AD (Table 1). All the 11 variants were novel missense variants that had not been reported in the genetic variant database. These variants in COL5A1 accounted for 1.57% (11/702) of the AD cases. There were no potential pathogenic variants that were found in the healthy control group (0/163). The sequence kernel association test performed on the COL5A1 gene between the 2 groups revealed that the likely pathogenic variants in the COL5A1 gene were significantly enriched in patients with AD (P=0.0037), thus suggesting that variants in COL5A1 were associated with AD.

Table 1.

Potential Pathogenic Variants in COL5A1 Gene Identified in 702 Patients With AD

No. of Related Patients	Coding Change	Protein Change	SIFT (Score)	Polyphen‐2 (Score)	Mutation Taster (Score)	PhyloP
1	c.487G>A	p. Gly163Ser	Damaging (0.002)	Probably damaging (1.000)	Disease causing (0.99999)	1.048
1	c.1372C>T	p. Pro458Ser	Damaging (0.032)	Probably damaging (1.000)	Disease causing (0.99999)	0.852
1	c.2768C>T	p. Pro923Leu	Tolerated (0.097)	Probably damaging (0.998)	Disease causing (0.99999)	0.786
1	c.2842C>T	p. Arg948Trp	Damaging (0.003)	Probably damaging (1.000)	Disease causing (0.99985)	0.883
1	c.3292G>A	p. Ala1098Thr	Damaging (0.022)	Probably damaging (1.000)	Disease causing (0.99999)	1.038
1	c.3398G>A	p. Arg1133Gln	Damaging (0.041)	Probably damaging (0.997)	Disease causing (0.99999)	0.943
1	c.3431G>C	p. Gly1144Ala	Damaging (0.000)	Probably damaging (0.999)	Disease causing (0.99999)	0.943
1	c.3445G>A	p. Val1149Met	Tolerated (1.000)	Probably damaging (0.978)	Disease causing (0.96904)	0.952
1	c.3752C>T	p. Pro1251Leu	Tolerated (0.130)	Probably damaging (1.000)	Disease causing (0.99999)	0.843
1	c.4241G>C	p. Gly1414Ala	Damaging (0.000)	Probably damaging (0.954)	Disease causing (0.99999)	1.026
1	c.5263G>A	p. Ala1755Thr	Tolerated (0.565)	Probably damaging (0.910)	Disease causing (0.75559)	0.953

RefSeq: NM_000093.4.

AD indicates aortic dissection​; PhyloP, phylogenetic P‐values.

Correlation Between Genotype and Phenotype

In patients with AD carrying variants in COL5A1, the mean age of AD onset was 59.9 years old, significantly higher than the mean onset age of all patients with AD (53.2 years). Meanwhile, the percentage of smokers (63.6% versus 45.6%, P=0.612), hypertensives (72.7% versus 54.1%, P=0.219), and DeBakey Type 3 AD (mild type of AD) (81.8% versus 75.2%, P=0.614) in COL5A1 carried patients were higher compared with the total AD cohort, though there were no statistically significant differences (Table 2).

Table 2.

Clinical Characteristics and Genotypes of the Patients Carrying Potential Pathogenic Variants in COL5A1 Gene

Gene	Genotype	Age, y	Sex	History of Smoking	Medical History of Hypertension	DeBakey Type of AD
COL5A1	c.487G>A	46	Female	No	Yes	3
COL5A1	c.1372C>T	68	Male	No	Yes	3
COL5A1	c.2768C>T	54	Male	Yes	No	3
COL5A1	c.2842C>T	36	Male	No	Yes	3
COL5A1	c.3292G>A	65	Male	Yes	No	3
COL5A1	c.3398G>A	60	Male	Yes	No	1
COL5A1	c.3431G>C	71	Male	Yes	Yes	3
COL5A1	c.3445G>A	59	Male	No	Yes	1
COL5A1	c.3752C>T	86	Male	Yes	Yes	3
COL5A1	c.4241G>C	58	Male	Yes	Yes	3
COL5A1	c.5263G>A	56	Male	Yes	Yes	3

RefSeq: NM_000093.4.

AD indicates aortic dissection.

Generation of Col5a1 Knockout Rat Model

Col5a1 knockout rats were established to validate the hypothesis of a relationship between the Col5a1 gene and AD. The Col5a1^fl/fl rats were generated using the CRISPR/Cas9 system by inserting a loxP sequence into intron 1 and intron 2. After mating the Col5a1^fl/fl and transgenic cre rats, 214 bp bases including exon 2 were deleted, generating a new termination codon in exon 3 (Figure S1). Consequently, all the critical domains of the Col5a1 gene were truncated (Figure 1A), and Col5a1 knockout (Col5a1^+/− ) rats were obtained. The genotypes of the rats were identified using polymerase chain reaction amplification and Sanger sequencing (Figure 1B, Figure S2). Meanwhile, the COL5A1 protein levels in different genotype rats were examined using Western blotting. The results revealed that protein levels of COL5A1 in Col5a1^+/− rats were significantly lower than in Col5a1^fl/fl rats (Figure 1C). These results indicate that the Col5a1 knockout rats were successfully generated.

Figure 1. Col5a1 KO rat revealed fracture in elastic fibers and disarray in collagenous fibers.

A, The mode of Col5a1 KO compared with that of the WT. B, Identification of littermate control rats via Sanger sequencing. C, Western blot demonstrated that protein levels of COL5A1 decreased in Col5a1^+/− rats, indicating that the Col5a1 KO model was successfully established. D, The percentage of survival in 48 wks showed that there was no significant difference between WT (n=25) and Col5a1^+/− rats (n=20). E, The Col5a1^+/− group rats exhibited scoliosis, but was not found in the WT group. Scale bar: 1 cm. F, Aortic dissection phenotype was found in 6‐wk‐old rats in either WT group or Col5a1^+/− group. Scale bar: 1 cm. G, No significant structural abnormality or fibril disarray was observed from H&E, Masson, and EVG staining. Scale bar: 100 μm. H, Electron microscopy showed fracture of elastic fibers in the aorta of 6‐wk‐old Col5a1^+/− rats and this was more severe in 48‐week‐old Col5a1^+/− rats. The white arrows indicate elastic fibers fracture. Scale bar: 1 μm. I, Electron microscopy revealed a disarray of collagenous fibers in the aorta of 6‐wk‐old Col5a1^+/− rats, and the disarray was more severe in 48‐wk‐old Col5a1^+/− rats. Scale bar: 1 μm. Col5a1 indicates collagen type V alpha 1 chain; EVG, elastic Van Gieson; H&E, hematoxylin and eosin; KO, knockout; and WT, wild type.

Col5a1 Knockout Rats Exhibited Collagen and Elastin Disarray Ultrastructure

Littermate controls were generated by mating Col5a1^+/− rats. The birth sex ratio of the wild type (WT) to Col5a1^+/− rats was ≈1:2. However, no Col5a1^−/− rats were obtained during the entire study, indicating homozygous knockout of Col5a1 can cause embryonic death (Table S3). This is consistent with our previous research.²⁷ The body weight of Col5a1^+/− rats from 4 to 9 weeks postbirth was significantly lower than WT rats, indicating that heterozygous knockout of Col5a1 affects the growth of Col5a1^+/− rats (Table S3, Figure S3). There were no significant differences found in the noninvasive systolic and diastolic blood pressure between the 2 groups (Table S3). Thus, knockout of Col5a1 does not affect the baseline blood pressure of the rats. There was no significant difference in the percentage of survival in 48 weeks between the 2 groups (Figure 1D). Rats in the Col5a1^+/− group exhibited scoliosis, while this phenotype was not found in the WT group (Figure 1E). Notably, no AD phenotype or aortic aneurysm was found in 6‐week‐old rats in either the WT group or the Col5a1^+/− group (Figure 1F). Moreover, no significant structural abnormality or fibrils disarray was found in hematoxylin and eosin, Masson, and elastic Van Gieson staining (Figure 1G). In elderly rats (48 week‐old) in the WT or Col5a1^+/− groups, either significant aortic lesion or structure abnormality was not found. The ultrastructural structure of aorta was observed using electron microscopy, and 10 random microscopic fields were selected in each sample. The fracture of elastic fibers and disarray of collagenous fibers was observed in 6‐week‐old rats in the Col5a1^+/− group, but not in the WT group (93.3% versus 0.0%, P<0.001). The structural damages in elastic and collagenous fibers were more severe in 48‐week‐old rats in the Col5a1^+/− group (Figure 1H and 1I). Although Col5a1^+/− rats did not exhibit the phenotype of AD or aortic aneurysm, the results still revealed that knockout of the Col5a1 gene caused fibrils fracture in the ultrastructural structure.

Col5a1 Knockout Predisposed Rat AD Model to High AD Incidence

To further validate the association between COL5A1 and AD, we used BAPN and AngII to induce AD in rats. Noninvasive systolic and diastolic blood pressure increased significantly in both groups after using AngII, and there were no significant differences between the 2 groups (Figure S4). After the rats were fed on BAPN for 4 weeks, 37.5% of the rats revealed microaneurysm in the Col5a1^+/− group, but not in the WT group (Figure 2A, 2B, and Figure S5). However, no AD or microaneurysm was observed in both groups after AngII implantation for 3 days. After the administration of BAPN+AngII, 70.0% of the rats exhibited AD in the Col5a1^+/− group and only 8.3% of rats presented with AD in the WT group (Figure 2A and 2B). This indicates that Col5a1 knockout rats were predisposed to a high incidence of AD. Compared with the WT group, the intervention of BAPN+AngII was associated with high mortality in the Col5a1^+/− group in 72 hours (Figure 2C). Furthermore, after induced AD via BAPN and BAPN+AngII, histology analyses showed significant disarray of collagenous fibers, fracture of elastic fibers, and tears in the vessel wall in the Col5a1^+/− group but not in the WT group (Figure 2D through 2F). The immunofluorescence results revealed that there was disarray and reduced expression of elastic fibers and αSMA was observed in the Col5a1^+/− group compared with the WT group after administering BAPN and BAPN+AngII (Figure 2G). These results confirmed that Col5a1 knockout was associated with high AD morbidity in the stress model compared with WT rats.

Figure 2. Col5a1 knockout predisposed rat AD model to high AD incidence.

A, The Col5a1^+/− rats presented with microaneurysm in BAPN subgroup and AD in BAPN+AngII subgroup. The red arrow indicates microaneurysm; the white arrow indicates AD. B, A significantly high percentage of aortic lesion in the Col5a1^+/− group was found after the intervention. *P<0.05. C, Compared with the WT group, the intervention of BAPN+AngII was associated with high mortality in the Col5a1^+/− group after 72 h. *P<0.05. D, H&E staining disclosed minor tears in the vascular wall and AD in the BAPN subgroup and the BAPN+AngII subgroup of Col5a1^+/− rats, respectively. Scale bar: 200 μm. The red arrow indicates a fracture of elastic fibers; the white arrow indicates the false aorta lumen; the black arrow indicates the original aorta lumen. E, EVG staining indicated a fracture of the elastic fibers in Col5a1^+/− rats. Scale bar: 200 μm. F, Masson staining revealed a disarray of collagenous fibers in Col5a1^+/− rats. Scale bar: 200 μm. G, Immunofluorescence indicated disarray and reduced expression of elastic fibers and αSMA in Col5a1^+/− rats treated with BAPN alone as well as BAPN+AngII. Scale bar: 200 μm. AD indicates aortic dissection; AngII, angiotensin II; BAPN, β‐aminopropionitrile monofumarate; Col5a1, collagen type V alpha 1 chain; DAPI, 2‐(4‐amidinophenyl)‐6‐indolecarbamidine dihydrochloride; EVG, elastic Van Gieson; H&E, hematoxylin and eosin; αSMA, alpha smooth muscle actin; and WT, wild type.

Col5a1 Knockout Triggered AD by Activating the TGF‐β‐Signaling Pathway

The TGF‐β signaling pathway is an important signaling pathway associated with the pathogenesis of AD. A previous study reported that the TGF‐β‐signaling pathway was activated in patients with AD.²⁸, ²⁹ Western blot analysis of the aorta samples showed that the level of TGF‐β1 was higher in the Col5a1^+/− group than in the WT group (Figure 3A and 3B). Meanwhile, phosphorylation SMAD family member 2, total SMAD2, and the ratio of phosphorylation SMAD family member 2/SMAD2 increased significantly in the Col5a1^+/− group. Moreover, in the Col5a1^+/− group, the ratio of phosphorylation SMAD family member 2/SMAD2 was higher in AngII and BAPN+AngII subgroups than in the control subgroup (Figure 3A, 3C, 3D, and 3E). MMP is a protein that maintains the homeostasis of the vascular matrix.³⁰ Previous studies have reported that the expression of MMP9 increases once the TGF‐β‐signaling pathway is activated, subsequently affecting the smooth muscle cell (SMC) elastin‐contractile units composite of elastin and SMCs.³¹, ³² Accordingly, the expression levels of MMP9 were significantly elevated in the Col5a1^+/− group compared with the WT group. These levels were significantly higher in the intervention subgroups than in the control subgroup in the Col5a1^+/− rats (Figure 3A and 3F). In addition, the levels of α‐SMA and elastin were significantly reduced in the Col5a1^+/− group. In the Col5a1^+/− group, the α‐SMA and elastin protein levels were reduced in BAPN and BANP+AngII subgroups compared with the control subgroup (Figure 3A, 3G, and 3H). Overall, these results suggest that Col5a1 knockout partially causes AD by activating the TGF‐β‐signaling pathway.

Figure 3. Col5a1 knockout activated TGF‐β‐signaling pathway.

A, Western blotting of TGF‐β1, SMAD2, p‐SMAD2, MMP9, αSMA, and ELN in the aorta. B, TGF‐β1 levels were higher in Col5a1^+/− group than the WT group. C, SMAD2 was significantly elevated in the Col5a1^+/− group. D, p‐SMAD2 was significantly elevated in the Col5a1^+/− group. The expression levels of p‐SMAD2 in AngII and BAPN+AngII subgroups in Col5a1^+/− rats were higher than in the control subgroup. E, The ratio of p‐SMAD2/SMAD2 levels was significantly elevated in the Col5a1^+/− group. The ratio of p‐SMAD2/SMAD2 in AngII and BAPN+AngII subgroups in Col5a1^+/− rats was higher than in the control subgroup. F, Col5a1 knockout led to elevated expression of MMP9. The levels of MMP9 in Col5a1^+/− rats after intervention were significantly higher than in the control subgroup. G, The expression of αSMA decreased significantly in the Col5a1^+/− group. The expression of αSMA in AngII and BAPN+AngII subgroups in Col5a1^+/− rats decreased significantly. H, ELN expression was significantly decreased in the Col5a1^+/− group. The expression level of ELN in AngII and BAPN+AngII subgroups in Col5a1^+/− rats was significantly lower than in the control subgroup. *indicates that P<0.05, **indicates that P<0.01, and ***indicates that P<0.001. AngII indicates angiotensin II; BAPN, β‐aminopropionitrile monofumarate; Col5a1, collagen type V alpha 1 chain; ELN, elastin; MMP9, matrix metallopeptidase 9; p‐SMAD2, phosphorylation SMAD family member 2; α‐SMA, alpha smooth muscle actin; SMAD2, SMAD family member 2; TGF‐β1, transforming growth factor beta 1; and WT, wild type.

Discussion

The targeted sequencing completed in samples in 702 patients with sporadic AD in our previous study suggested that variants in COL5A1 might be related to AD.²² Accordingly, further analyses were performed in this study to identify these variants. A total of 11 likely pathogenic variants were identified in the COL5A1 gene from 11 patients. These variants accounted for 1.57% of the AD cases studied and indicated that COL5A1 is associated with AD. To further investigate the relationship between the COL5A1 gene and AD, a Col5a1 knockout rat model was established via the CRISPR/Cas9 system. Although no spontaneous AD or aortic aneurysm was found without intervention, fracture of elastic fibers and disarray of collagenous fibers were observed from electron microscopy in 6‐week‐old Col5a1^+/− rats. Furthermore, in the rat AD model, Col5a1 knockout rats were predisposed to higher AD incidence and mortality compared with WT rats. These results validated our hypothesis that COL5A1 causes AD. Finally, subsequent mechanism analysis revealed that Col5a1 knockout partially caused AD by activating the TGF‐β‐signaling pathway. In summary, these results suggest that variants in COL5A1 trigger AD in the stress model via TGF‐β‐signaling pathway activation.

The majority of heritable AD with syndromic disease features can be attributed to mutations in the currently known causative genes. However, only 30% of heritable AD without syndromic disease features have mutations in these genes; thus more AD causative genes are unidentified.¹⁴ The rare and highly penetrant genetic variants for AD are easy to recognize, but there are still a few penetrant variants that can increase the risk of AD only in combination with environmental risk factors or with a second low‐risk variant.¹⁴ Here, 11 patients with AD carrying potential pathogenic variants in COL5A1 were identified. All the patients had either a medical history of hypertension or a history of smoking besides carrying the variants. Compared with all patients with AD recruited in this study, these 11 patients had an older mean age of onset, a higher percentage of hypertension, smoking, and DeBakey Type 3 AD. This is an indication that COL5A1 is a modest AD causative gene with low penetrance. These results confirmed why the Col5a1 knockout rats exhibit AD only after drug intervention.

Previous study revealed that at least 25% of AD is caused by genetic factors, and have a potential pathogenic mutation in at least 1 specific gene.¹² However, up to 80% of patients with AD are nonsyndromic and sporadic, which makes it more difficult to assess the genetic risk factors in these populations.¹³ In this study, we revealed 1.57% of patients with AD carrying potential pathogenic variants in COL5A1 gene, and we have identified other pathogenic genes that might be associated with sporadic AD using targeted next‐generation sequencing in previous studies.²², ³³ However, targeted sequencing is limited in terms of understanding the full spectrum of AD; thus whole exome sequencing or whole genome sequencing should be performed in the future.

Previous studies have reported that Col5a1^−/− mice at the embryonic stage died because of the absence of collagen fibril formation.²⁷ In this study, homozygous embryonic death was found in Col5a1 knockout rats. Meanwhile, all the COL5A1 variants carried by the patients were found to be heterozygous, which is consistent with the phenomenon detected in rats.

COL5A1 is a causative gene of the classical EDS, and variants in COL5A1 have been linked with scoliosis in some reported cases.³⁴, ³⁵ However, the scoliosis phenotype was only found in Col5a1 knockout rats rather than patients with AD carrying COL5A1 variants. This may be because null variants of COL5A1 are the major type of mutation in classic EDS.³⁶ Thus, patients carrying missense variants do not exhibit syndromic features of classic EDS.

Earlier studies reported that activation of the TGF‐β‐signaling pathway is implicated in the production of vascular matrix proteins, leading to vascular remodeling.³⁷, ³⁸ In addition, the SMAD signaling pathway can be activated in the extracellular matrix mediated by TGF‐β.³⁹ The TGF‐β1/SMAD signaling pathway was found to be significantly activated in patients with AD,²⁸, ²⁹ consistent with findings of this study. Furthermore, TGF‐β1 highly enhances the secretion of MMP9. The MMP9 was reported to be implicated in the pathogenesis of AD in human and animal models.³², ⁴⁰, ⁴¹ MMP9 activates elastin degradation and affects the function of SMC elastin‐contractile units. The SMC elastin‐contractile units are important for maintaining the structural integrity of the aorta, and loss of these units is involved in the pathological process of AD.³¹, ⁴², ⁴³ In this study, the level of MMP9 was significantly increased in the Col5a1^+/− group, and disarray and reduced expression of elastin and α‐SMA were observed in the Col5a1^+/− group after intervention, thus indicating the functional SMC elastin‐contractile units were reduced by elevated MMP9.

Previous studies demonstrated that angiotensin‐converting enzyme (ACE) inhibitors are considered to be protective against AD or aortic aneurysm phenotype.⁴⁴, ⁴⁵ However, it is still unclear whether the ACE inhibitor could have benefits in patients with sporadic AD. In our study, a total of 11 patients were identified carrying potential pathogenic variants in COL5A1. According to 5‐year follow‐up data, 7 patients were using ACE inhibitors, and no death or recurrence of AD occurred in these patients, while recurrence of AD occurred in 1 patient without using an ACE inhibitor (0.0% versus 25%, P=0.417). More studies will be done to investigate whether ACE inhibitors could protect against AD in the future.

After targeted sequencing in 702 patients with AD and 163 controls, potentially pathogenic variants in COL5A1 were found significantly enriched in the patients with AD group. Since the control group did not include a high number of participants, more cases and controls are needed in further studies. In addition, although it was found that Col5a1 knockout can cause AD in the stress model partly by activating the TGF‐β‐signaling pathway, these data are still preliminary in delineating the causal relationship between TGF‐β‐signaling pathway and AD. More mechanism studies should be performed to investigate the role of other mechanisms in the pathogenesis of AD resulting from Col5a1 knockout.

In conclusion, we demonstrate that variants in COL5A1 are responsible for AD through targeted sequencing. Furthermore, it was confirmed that Col5a1 knockout rats predisposed the rat AD model to high AD incidence. Subsequently, mechanism analysis revealed that activation of the TGF‐β‐signaling pathway was partially implicated in AD pathogenesis. This study provides more insight into the genetic cause of AD.

Sources of Funding

This study was supported by National Natural Science Foundation of China projects (No. 91839302) and National Key Research and Development Program of China (No. 2017YFC0909401).

Disclosures

None.

Supporting information

Tables S1–S3

Figures S1–S5

(J Am Heart Assoc. 2021;10:e019276. DOI: 10.1161/JAHA.120.019276.)