Integrated analysis identies hypertension-GWAS-loci regulated, differentially expressed genes in aorta and blood of aortic dissection

Background: Aortic dissection is a life-threatening condition caused by a tear in the intimal layer of the aorta. Hypertension is the most common risk factor of aortic dissection. However, only a small proportion of subjects with hypertension will develop aortic dissection. Genetic basis and effect genes determining the development and progression of aortic dissection remained to be identied. In this study, we attempted to recognize the underlying effect genes from reliable large-scale genome-wide association studies (GWASs) of hypertension. Methods: As most GWAS locus functions through its biological role in gene expression regulation, we initially converted the GWAS signals to transcriptomic proles in aorta using the Functional Summary-based Imputation (FUSION) algorithm. The FUSION derived genes were then checked whether they were differentially expressed in aorta of subjects with and without aortic dissection. Results: We found 23 genes that were regulated by hypertension GWAS loci and were altered in aorta of dissection patients. In particular, the DCAF16 gene could be detected in blood, providing a possibility of non-invasive early detection or prediction of hypertensive individuals at risk of aortic dissection. Conclusions: Our analyses identied effect genes in aorta and provided a possibility of non-invasive early detection of aortic dissection.

Integrated analysis identi es hypertension-GWAS-loci regulated, differentially expressed genes in aorta and blood of aortic dissection The aorta is the largest blood vessel in the body, originating from the left ventricle of the heart [1][2][3][4]. It is made up of the intima, media, and adventitia layers, consisting of endothelial cells and connective and muscle tissues. A tear in the intimal layer of the aorta that results in the separation of the layers of the aortic wall will lead to a life-threatening condition called aortic dissection [1][2][3][4]. Aortic dissection is a lethal but rare cardiovascular disease [5][6][7][8][9]. Clinically, the symptoms of aortic dissection are sometimes atypical, and the onset are always acute. It can be suddenly fatal and 40% of people die immediately once the aorta dissects [4][5][6]. Delays in recognition, diagnosis, and treatment are associated with rapid increases in mortality. Therefore, patients with related symptoms or at risk should be thoroughly investigated despite its rare nature. Previous observational studies have analyzed clinical variables for rapid estimation of the individual risk of dissection on arrival in the emergency department [10]. Noninvasive, early prediction or detection of dissection in earlier stage might bene t to improved survival and effective treatment. This relies on a better understanding of the underlying molecular and genetic mechanisms.
While it is not always clear why a tear may occur, quite often it involves uncontrolled hypertension, aging, heart surgery, and genetic disorders that involve the connective tissue, such as Marfan's syndrome. More than 70% of individuals with aortic dissection have a previous history of hypertension. About 18% of individuals with an acute aortic dissection have a history of heart surgery. Aging is also an important risk factor, as there is reduced resistance of arterial walls with age. Indeed, aortic dissection is more prevalent in the elderly. Of these risk factors, hypertension is the most signi cant and the only modi able factor.
There is no doubt that a subgroup of individuals with hypertension will develop aortic dissection. Identi cation of this atrisk subgroup, or assessment of risk of artic dissection in the large hypertension population, might help for potential early intervention. Taking advantage of large scale and reliable genome-wide association study (GWAS) and high throughput transcriptome data, we herein attempted to identify molecular predictor of dissection for at-risk individuals with hypertension.

Study rationale and design
As it is hard to perform a reliable GWAS of aortic dissection, which requires a large sample size, recognizing at-risk subgroups in hypertensive population before the dissection occurs is a feasible strategy. Hypertension has a heritability around 50% [11][12][13]. Numerous large-scale GWASs have identi ed genetic loci that are associated with blood pressure or hypertension [14][15][16][17][18][19][20]. As hypertension is the most signi cant risk factor for aortic dissection, it is reasonable to speculate that there are aortic dissection risk genes among those hypertension-associated loci. Since most GWAS locus functions through its biological role in gene expression regulation, we converted the hypertension GWAS signals to transcriptomic pro les in aorta using the Functional Summary-based Imputation (FUSION) algorithm [21]. The FUSION derived genes represent genetically driven gene expression in the target organ. Those hypertension GWAS loci regulated genes in aorta were thus de ned as effect genes underlying the genetics factors for at-risk hypertension individuals. These effect genes were then checked whether they were differentially expressed in aorta and blood of patients, to explore the possibility of non-invasive detection. The rationale and work ow were shown in Figure 1.
Transcriptome pro les of aorta and blood of patients with aortic dissection The cross-validated, hypertension-GWAS-based, FUSION-predicted genes were investigated in expression dataset of aorta biopsies from individuals with and without aortic dissection. By searching the term "aortic dissection" through the NCBI GEO (Gene Expression Omnibus) browser (https://www.ncbi.nlm.nih.gov/geo/browse/), we got the dataset GSE52093. In GSE52093, gene expression pro les of ascending aorta were measured by the Illumina HumanHT-12 V4.0 expression beadchip. Differentially expressed genes were identi ed by comparing the gene expression pro ling of dissected ascending aorta (n = 7) with that of control (n = 5) using the limma R package. Those differentially expressed FUSIONderived genes were genetically regulated effect genes for aortic dissection in hypertension population.
To explore the clinical potential of the identi ed effect genes in aorta, we investigated the gene expression of these genes in expression dataset of blood from aortic dissection patients (GSE9106) [25]. GSE9106 contains gene expression pro les of 61 peripheral blood RNA samples (training set) collected from 25 controls and 36 patients with thoracic aortic aneurysm [25], which usually leads to aortic dissection [26,27,28,29]. The testing set of GSE9106 that contains 22 TAA samples and 11 controls was also used for validating the expression change of DCAF16 [25].The transcriptomes were analyzed by the Applied Biosystems Human Genome Survey Microarrays that include 29,098 human genes [25]. Detection of those effect aortic genes in blood of patients might provide a way to identify individuals at risk for aortic dissection.

Suggestively signi cant genes cross-validated by four hypertension-related GWASs
In addition to the genome-wide signi cant hits, genes showing suggestively signi cant TWAS P-values in all four GWASs might also be relevant in aorta. We then extended the gene list with a nominally signi cant threshold of TWAS-P-value < 0.01 (Supplementary Table 1). There were 113 such cross validated genes with a TWAS P-value < 0.01 in all the four GWASs "Hypertension (Self-reported)" (n = 385), "High blood pressure" (n = 381), "Systolic blood pressure, automated reading" (n = 402), and "Diastolic blood pressure, automated reading" (n = 423) ( Figure 2). These 113 suggestively signi cant genes might be hypertension-risk-loci regulated genes that contribute to aortic dissection and were subjected to differential expression analysis. 19 genes were differentially (P-value < 0.05) expressed in aorta of patients in GSE52093, including four genome-wide signi cant genes mentioned above. This added 15 additional targets (ALDH2, AGTRAP, SF3B3, CTSW, SPNS1, CCNT2, PRADC1, ACADVL, TUFM, SLC26A1, TPD52L2, MRAS, SMOC1, OIP5-AS1, and TOP3A) in the effect gene list (Table 1). These 23 genes were genetically-regulated and were differentially expressed in aorta of patients, thus acting as the underlying effect genes for aortic dissection in at-risk individuals with hypertension.

Detection of genetically dysregulated aortic gene DCAF16 in blood of dissection patients
We then asked whether the dysregulated genes in aorta could be observed in blood of patients with thoracic aortic aneurysm [25], which usually leads to aortic dissection [26][27][28][29]. In the gene expression dataset GSE9106 which contains 36 cases and 25 controls [25], we found that three (OPRL1, HAUS8, DCAF16) out of the 23 effect genes were signi cantly altered in blood of patients compared with controls. However, the directions of differential expression of OPRL1 and HAUS8 were opposite between aorta (GSE52093) and blood (GSE9106). DCAF16 was the only hit that shows consistent decrease in both aorta and blood (P = 0.00948 in the training set, 36 cases and 25 controls) of patients ( Figure 3). Notably, expression change of DCAF16 was well-validated by the testing set (P = 0.0002 in 22 cases and 11 controls) of GSE9106.

Discussion
Aortic dissection is a life-threatening vascular disease caused by a tear in the intimal layer of the aorta [4]. It is most prevalent in the elderly with an incidence of 35 cases per 100,000 old people per year [4,5,6,22]. Since most cases happen with an acute onset and bad prognosis, development of methods regarding risk prediction and early diagnosis will be the key for better management of the condition [30]. Because of the di culty of sample collection of aortic dissection, patient-based studies, such as genetic study and longitudinal study, is limited in the eld [31]. Hypertension is the most common risk factor for the dissection. Recognition of patients at substantial risk for aortic dissection in the large hypertensive population before the dissection occurs is essential for early intervention. In the current study, we performed a data-mining aiming at identify effect genes and potential biomarkers based on genomic and transcriptomic data.
Note that there is currently few large-scale genetic studies of aortic dissection and the number of risk genes was limited [26,[32][33][34][35]. By converting hypertension GWAS signal to transcriptome in aorta, we obtained 113 genes regulated by hypertension-related genetic variants. These 113 genes might function in the aorta, underlying the contribution of hypertension to aortic dissection. Expression of 23 out of the 113 genes indeed altered in aorta of patients, providing a list for further functional mechanistic studies. Compared with classical genetic studies, it seemed more fruitful by taking advantage of recent advances on GWAS of hypertension related traits and bioinformatic approaches. Note that expression of FBN1, one of the recognized top hits for aortic dissection [26,[32][33][34][35], is also altered in aorta of patients (P = 0.0074, logFold_Change = 0.865 in GSE52093), supporting the reliability of the current strategy.
In addition to the implications for mechanistic studies, we identi ed a gene DCAF16 altered in both aorta and blood of patients. Our results suggest that decreased DCAF16 expression in blood of hypertensive individuals might predict the risk of thoracic aortic aneurysm and aortic dissection. Clinically, people with chest pain or back pain were subjected to history taking and physical examination on arrival in the emergency department [1,2,3,5,10]. Clinical prediction of aortic dissection mainly relies on clinical variables such as back pain and blood pressure. However, in most cases the dissection is acute. Noninvasive and rapid prediction or diagnosis of dissection in early stage using biomarkers might bene t to improved survival and effective treatment. For those people assessed as at high risk, early intervention such as adrenergic blocking and surgical avoiding could be suggested. Knowledge of molecular sign of dissection might help for early prediction of dissection.
There are several limitations of the current study. Firstly, it is not sure whether we can use this predictor to separate acute dissection from other aortic disorders such as penetrating aortic ulcer. Secondly, datasets used for aorta and blood were from different individuals, longitudinal study is needed con rm the results. Also, there is gaps between the list of effect genes and the mechanism of the disease. Further functional investigation of these genes might bene t from the current study.

Conclusions
Taken together, the present study identi ed that 23 genes were regulated by hypertension GWAS loci and were altered in aorta of dissection patients. Among the 23 genes, the expression alteration of DCAF16 could be detected in blood of human patients. Therefore, this potential functional gene is expected to be a biomarker for aortic dissection. Future studies will be necessary to demonstrate whether the DCAF16 gene has the early diagnostic potential in hypertensive patients at risk of aortic dissection.

Declarations
Funding This work was supported by the National Natural Science Foundation of China (81760059) and Yunnan health training project of high level talents (H-2017018) and Special Joint Program of Yunnan Province (2018FE001-181).

Availability of data and materials
All data generated or analyzed during this study are included in this published article [and its supplementary information les].
Ethics approval and consent to participate Not applicable.

Consent for publication
All the authors have consented for the publication.

Competing interests
The authors declare that they have no competing interests.  Table   Table 1. Genes implicated by hypertension-GWAS based TWAS analysis Figure 1 Work ow of the current study. Numerous large-scale GWASs have identi ed genetic loci that are associated with blood pressure or hypertension [14,15,16,17,18,19,20]. As hypertension is the most signi cant risk factor for aortic dissection, it is reasonable to speculate that there are aortic dissection risk genes among those hypertension-associated loci. Since most GWAS locus functions through its biological role in gene expression regulation, we converted the hypertension GWAS signals to transcriptomic pro les in aorta using the Functional Summary-based Imputation (FUSION) algorithm [21]. The FUSION derived genes represent genetically driven gene expression in the target organ. Those hypertension GWAS loci regulated genes in aorta were thus de ned as effect genes underlying the genetics factors for at-risk hypertension individuals. These effect genes were then checked whether they were differentially expressed in aorta and blood of patients, to explore the possibility of non-invasive detection. Figure 2 113 cross-validated genes were regulated by hypertension-related loci in aorta. Gene with a nominally signi cant threshold of TWAS-P-value < 0.01 from the four GWASs were used for cross-validation (Supplementary Table S1-S4).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.