Variants Around miR-451 Can Affect the Risk of Large-artery Atherosclerosis Stroke in Chinese

Background: Ischemic stroke has high incidence rate, mortality rate and disability rate. Genetic factors have a signicant impact on stroke risk. MicroRNAs are a class of small non-coding RNA. Intergenic variants can affect the regulation of microRNAs and modulate large-artery atherosclerosis stroke susceptibility. The low expression of miR-451 aggravated ischemic injury signicantly. Methods: Functional intergenic variants near hsa-mir-451 were identied by bioinformatics analysis. We conducted a case-control study to explore the associations of selected variants with large-artery atherosclerosis stroke risk in Chinese. Results: The rs901975 (G>A) near hsa-mir-451 was identied as a functional SNP for stroke susceptibility. The protective effect of A allele was signicant in codominant model (OR = 0.62, 95% CI = 0.44-0.89, P = 0.005), recessive model (OR = 0.65, 95% CI = 0.47-0.89, P = 0.006) and log-additive model (OR = 0.82, 95% CI = 0.71-0.96, P = 0.012). We also found the signicant effect in participants over 65 years old, male, hypertensive and diabetic people. Moreover, hypertensive people genotyped as GG+GA had 2.84 - fold increased risk compared with those genotyped as AA without hypertension (P interaction = 0.036). In the MEGASTROKE Consortium, rs901975 A allele also had protective effect on LAA stroke in Europeans (OR = 0.937, 95% CI: 0.882-0.996, P = 0.036). Combined analysis of our study and MEGASTROKE showed consistent trend (OR = 0.920, 95% CI: 0.869-0.973, P = 0.004). Conclusions: Our study suggested that rs901975 near hsa-mir-451 might affect large-artery atherosclerosis stroke susceptibility in Chinese population.


Introduction
Stroke is going to be an epidemic worldwide. The high disability rate and mortality of stroke is a challenge for health (Hankey 2017). About 80% of stroke are ischemic stroke (IS) (Benjamin et al. 2019). IS was classi ed to be ve etiological subtypes according to the TOAST criteria (Adams et al. 1993).
Among those subtypes, large-artery atherosclerosis (LAA) stroke is more prevalent in Chinese population (Liu et al. 2011). Clinical trials have shown that hypertension, hypercholesterolemia, and carotid stenosis are causal risk factors for LAA stroke (Hankey 2017). But these factors cannot explain all the causes of stroke. Studies in family found the importance of genetic risk factors involving in the progress of stroke (Fan et al. 2019). The role of single nucleotide polymorphisms (SNPs) in interleukin 6 (Akinyemi et al. 2017), paraoxonase 1 (Wei et al. 2017), C-reactive protein (Ye et al. 2018) and so on has been con rmed to be responsible for stroke susceptibility. The discovery of ANRIL prove the function of non-coding RNA (ncRNA) on stroke (Pasmant et al. 2011).
MicroRNAs (miRNAs) are a class of ncRNA with ~ 23 nucleotides. They play a negative regulatory mechanism by post transformation modi cation regulation. It mainly combines with 3' untranslated region (UTR) of mRNA to induce mRNA degradation. When only part of sequences is paired, it can inhibit mRNA translation (Bartel 2009). Atherosclerosis is mainly caused by lipid metabolism disorder and leads to functional changes of endothelial cells, smooth muscle cells, and leukocyte recruitment (Weber and Noels 2011). A series of studies revealed the importance of miRNAs such as miR-181b, miR-146a, miR-143/145 and so on in molecular signaling pathways and lipid homeostasis involved in atherosclerosis (Feinberg and Moore 2016). The miR-451 may suppress the in ammatory response mediated by microglia activation, targeting TLR4 (Toll-like receptor 4), but the overexpression of miR-451 offsets the inhibitory effect . miR-451 has been detected as biomarker and potential regulation target in tumors (Bai and Wu 2019), rheumatoid arthritis (Churov et al. 2015) and amyotrophic lateral sclerosis (Dolz et al. 2017). Moreover, the association between stroke and miR-451 is also being investigated in vivo and in vitro. In a Chinese group, miR-451 in circulation showed a lower level in patients compared with the controls and in a mouse middle cerebral artery occlusion (MCAO) model, upregulation of miR-451 can reduce ischemia-reperfusion injury (Fu et al. 2019).
Many disease-associated SNPs (daSNPs) have been indicated by Genome-wide association studies (GWAS). SNPs located in mature miRNAs or pre-miRNAs could cause complex in uence in miRNAs processing, expression, and/or binding to target mRNA (Hu et al. 2008). However, only parts of SNPs are in the functional genes. The daSNPs in intergenic regions (IGR) are also remarkable in disease phenotypes. The IGR-daSNPs can regulate target genes by disrupt the function of nearby host regulatory elements (HREs), causing abnormal expression of disease related genes, and consequently leading to disease. (Chen and Tian 2016).
A series of studies have reported SNPs in miRNAs can affect stroke susceptibility. The relationship of SNPs in IGR of pre-miRNAs on LAA stroke is still unclear. Thus, we tried to identify the functional SNPs ground IGR of the hsa-mir-451 and evaluated their effect on the LAA stroke risk in Chinese.

Subjects
The diagnosis of LAA stroke is mainly based on the medical history, focal neurological de cit and imaging ndings of intracranial and extracranial vascular stenosis over 50%. All individuals in this study have been described previously (Yang et al. 2018). In short, 746 patients with LAA stroke and 1,101 controls were nally included in this study. All patients were initially diagnosed as LAA stroke according to TOSAT by experienced clinician at enrollment. Those who are younger than 18, have cancer, autoimmune diseases or organ failure were excluded. The 1,101 controls were selected from healthy individuals without a history of cardiovascular or cerebrovascular disease during the same period. This study was approved by an ethical review in Jinling Hospital, Nanjing, China and each participant gave a written informed consent.

SNP selection and genotyping
Based on the UCSC Genome Browser database (http://genome.ucsc.edu/, 2020-03-05), we browsed 5 kb upstream and 5 kb downstream IGR of hsa-mir-451. The 310bp 3' of hsa-mir-451 is the coding region of ERAL1. Thus, we only chose the 5kb 5' anking region of hsa-mir-451 and 13 SNPs was found in Chinese Han South population (Supplementary Table 1). After removing the SNPs with more than two alleles, 4 common SNPs with minor allele frequencies (MAF) ≥ 5% were selected. Then we used rSNPBase (http://rsnp.psych.ac.cn/, 2020-03-05) to nd potentially functional SNPs that might be an eQTL of genes or have association with regulatory elements. Finally, rs901975 was included in our study.
The genotyping methods has been described before . Finally, 715 patients with LAA stroke and 1,088 controls were genotyped successfully. The call rate was 97.6% (1,803/1,847).

Statistical Analysis
We used Student's t test and chi-squared test to compare continuous and categorical variables respectively. Hardy-Weinberg equilibrium (HWE) of allele frequencies was tested by c 2 test in control group. Logistics regression was used to test the association of rs901975 with LAA stroke susceptibility in log-additive, codominant, codominant, and recessive models. Crude and adjusted odds ratios (ORs) and 95% con dence intervals (CIs) were calculated to evaluate the effect of genetic model. The results of strati ed analysis were shown as forest plot. We tried to nd the interaction between SNP and other factors. Two-tailed P < 0.05 (α = 0.05) was considered signi cant. R software (version 3.5.1) was used for all the statistical analyses.

Individual characteristics
Characteristics of individuals have been described in Table 1. There was no signi cant difference in age (P = 0.450) and sex (P = 0.167) distribution between cases and controls. The patients have a higher proportion of hypertension and diabetes compared with the controls (P < 0.001).

Interaction between hypertension and rs901975
Hypertension is also a manifestation of atherosclerosis. In people with hypertension, the LAA stroke risk caused by rs901975 were still signi cant. Interaction analysis found an additional increased risk of LAA stroke (adjusted by age, sex and diabetes, Fig. 2

Bioinformatics information
The allele frequencies of rs901975 for different races were explored on Emsembl website (http://asia.ensembl.org/, 2020-03-05, Supplementary Table 2). The minor allele of the SNP rs901975 was A in all races. Based on HaploReg v4.1 (http://pubs.broadinstitute.org/mammals/haploreg/haploreg.php/, 2020-03-16), SNP rs901975 and its nearby variants within a LD block are located in regulatory regions, and to be an eQTL for genes in brain and blood. The potential function of rs901975 and has-miR-451 were showed in Fig. 3. On miRWalk 3.0 web tool (http://mirwalk.umm.uni-heidelberg.de/, 2020-03-16), we browsed target genes of hsa-miR-451 and found that VANGL2, MCTS1 and POM121C are mentioned in both TargetScan and miRDB database in Human. The rs901975 is in promoter and enhancer sites. This SNP is an eQTL for DHRS13 in brain and cerebellar hemisphere (P = 6.65×10 − 6 ). These evidences indicate that rs901975 is related to the functional regulation of genes in brain. Analyses in the MEGASTROKE (Malik et al. 2018) yielded consistent trends for LAA stroke (OR = 0.937, 95% CI: 0.882-0.996). Combined analysis of our study and MEGASTROKE showed that rs901975 A allele was associated with decreased risk of LAA stroke (OR = 0.920, 95% CI: 0.869-0.973, P = 0.004).

Discussion
Numerous predisposing risk factors involved in LAA stroke. Genetic factors can explain the risk of stroke which cannot be covered by traditional risk factors. Heritability for LAA stroke was described as 40.3% (Bevan et al. 2012). The genetic risk factors play signi cant roles in the diagnosis and prevention of LAA stroke. MiRNAs are a group of ncRNA with about 23 nucleotide length. They can regulate gene expression by targeting the 3'UTR of the mRNA. Growing evidence has indicated circulating miRNAs can be unique biomarkers for stroke diagnosis, prognosis, and therapy (Mirzaei et al. 2018). In previous studies, our team found that microRNA related SNPs can affect LAA stroke susceptibility ).
The associations of SNPs in 5kb 5' anking region of hsa-mir-451 with the LAA stroke risk were explored in our study. Our ndings suggested that rs901975 (G > A) near hsa-mir-451 was associated with a decreased LAA stroke risk in Chinese population. The results of strati ed analysis based on age, sex, hypertension, and diabetes for codominant and recessive genetic model of rs901975 were also signi cant. The protective effect of rs901975 A allele was also con rmed in other ethnic groups in MEGASTROKE.
In a research for human myocardial infarction (MI), miR-451 levels were up-regulated in 7-day MI (Bostjancic et al. 2009). The miR-451 is a special miRNA because pri-miR-451 is too short to be recognized by Dicer and it requires Argonaute2 catalytic activity (Cifuentes et al. 2010). Researchers often pay attention to miR-451 when they study miRNA function. miR-451 targets CUGBP Elav-like family member 2, which protects neurons from apoptosis and oxidative stress induced by oxygen and sugar deprivation/reoxygenation (Liu et al. 2018). By targeting TLR4, miR-451 could inhibit microglia activationmediated in ammation, which is common in the injury and reparation of brain tissue in stroke . The miR-451 has protective effect on cerebral ischemia-reperfusion injury both in mice and human. It was speculated that higher miR-451 levels relate to a good prognosis for stroke patients. ( Therefore, we speculated that rs901975 is associated with high level of Ox-LDL, which can negatively regulate TLR4 level and thus increase the level of circulating miR-451. Although rs901975 cannot directly match miR-451 to regulate its circulating level, it can participate in miR-451 induced stroke through other molecular pathways. In conclusion, our study indicated that the functional intergenic variants near miR-451 were signi cantly associated with LAA stroke risk. These ndings need further functional studies and large populationbased studies to verify.  Forest plot of strati ed analysis by age, sex, hypertension, and diabetes for rs901975  The potential function of miR-451 and rs901975 (color should be used for this gure in print) a. target genes of hsa-miR-451 based on miRWalk 3.0 b. an LD plot of SNPs in 5kb downstream of hsa-mir-451 c. the gure represents chr17q11.2 within a 20-kb window around rs901975 on UCSC Genome web tool.

Declarations
(NCBI Human Genome GRCh38). Tracks (from top to bottom) in the picture are Genome Base Position,