Association of Common Variants in OLA1 Gene with Preclinical Atherosclerosis

Reactive oxygen species impair the blood vessels, leading to the initiation of atherosclerosis, and migration and proliferation of vascular smooth muscle cells and neovascularization by endothelial cells of vasa vasorum are essential for atherosclerosis development. Obg-like ATPase 1 (OLA1), a negative regulator in cellular responses to oxidative stress, binds to breast cancer susceptibility gene 1 (BRCA1), which protects vascular endothelial and smooth muscle cells against reactive oxygen species. However, it is not known whether OLA1 is genetically correlated with atherosclerosis. Here, we conducted two independent population-based case–control studies to explore the effects of variants in OLA1 genes on preclinical atherosclerosis. A total of 564 and 746 subjects who had thicker and normal carotid intima–media thickness (cIMT), respectively, were enrolled. Among 55 screened SNPs, rs35145102, rs201641962, rs12466587, rs4131583, and rs16862482 in OLA1 showed significant associations with cIMT. SNP rs35145102 is a 3′-utr variant and correlates with the differential expression of OLA1 in immune cells. These five genetic markers form a single closely linked block and H1-ATTGT and H2-GCCTC were the top two most prevalent 5-locus haplotypes. The H1 + H1 genotype negatively and H1 + H2 genotype positively correlated with thicker cIMT. The five identified SNPs in the OLA1 gene showed significant correlations with cIMT. Furthermore, we found that OLA1 was required for migration and proliferation of human aortic endothelial and smooth muscle cells and regulated vascular tube formation by human aortic endothelial cells. Therefore, these genetic variants in the OLA1 gene may serve as markers for risk prediction of atherosclerotic diseases.

The intima-medial thickness (IMT), which is the distance between the lumen-intima and adventitia-media borders of blood vessels, is an early preclinical phenotype of atherosclerosis. Carotid IMT (cIMT) can be measured easily and noninvasively by high-resolution Further, OLA1 has been shown to bind to the amino-terminal region of BRCA1 and the carboxy-terminal region of BARD1 for centrosome regulation [37]. OLA1 plays a role in cell migration and cell proliferation and regulates the cellular responses to oxidative and heat shock stresses [25][26][27][28][29][30][31]. These observations raise the possibility that defects in or altered expression of OLA1 is involved in the development of atherosclerosis and prompted us to determine whether polymorphisms of common genetic variants in the BRCA1, BARD1, and OLA1 genes are correlated with cIMT thickening and could be used as indicators of cardiovascular risk.

Five SNPs in OLA1 Gene Are Significantly Correlated with cIMT
To present our approaches to determining whether OLA1, BRCA1, and BARD1 are genetically correlated with cIMT, a schematic illustration of this study is shown in Figure 1. The anthropometric and clinical characteristics of the study subjects in the discovery and validation stages have been described previously [38]. In the discovery stage, the mean (SD) ages at enrollment for cases and controls were 58.9 (8.8) and 52.6 (8.7) years, respectively. The percentages of male participants were 55.7% and 45.9% in the case and control groups, respectively. In the validation study, the mean (SD) ages at enrollment for cases and controls were 58.5 (8.5) and 58.0 (8.5) years, respectively. The percentage of male participants in both groups was 50.7%. The detailed anthropometric and clinical profiles of the study subjects are showed in Supplementary Table S1. The results of multivariate logistic regression analyses showed that older age, male sex, higher SBP, higher BMI, higher LDL-C/HDL-C ratio, and cigarette smoking were significantly correlated with higher likelihoods of having thicker cIMT [38].
Among 59 screened SNPs, one SNP (BARD1 rs10498020) with a p-value in the HWE test <0.001 and another three SNPs (OLA1 rs13019401 and rs16862465 and BARD1 rs34066909) with call rates <95% were excluded (Supplementary Table S2). There were 25, 21, and 9 eligible SNPs in OLA1, BARD1, and BRCA1, respectively, for association analyses. The pre-set critical values of the candidate genetic markers were 0.0040, 0.0048, and 0.011, respectively. A total of five SNPs in OLA1 were eligible for the validation study. None of SNPs in BRCA1 or BRAD1 had a p-value less than the pre-set critical values of candidacy. The primers for PCR reaction and annealing of these five candidate OLA1 SNPs are shown in Supplementary Table S3. In the discovery case-control study, the minor allele frequencies of these five candidate OLA1 SNPs in the controls ranged from 0.198 to 0.268 and were 0.227 to 0.310 in the cases ( Table 1). The minor alleles of all candidate SNPs were significantly more prevalent in the cases. After adjustment for all significant anthropometric and clinical characteristics, heterozygous genotypes were correlated with significantly higher ORs of having thicker cIMT. The multivariate-adjusted ORs ranged from 1.59 to 1.86. Similar results were observed in the validation study.
The results of the Breslow-Day test showed that the estimated ORs for all candidate SNPs in the discovery case-control study were not statistically different from those in the validation study (all p-values > 0.2). Accordingly, it was appropriate to obtain the pooled ORs for each candidate SNPs. The pooled ORs of thicker cIMT ranged from 1.55 to 1.72. Among 59 screened SNPs, one SNP (BARD1 rs10498020) with a p-value in the HWE test <0.001 and another three SNPs (OLA1 rs13019401 and rs16862465 and BARD1 rs34066909) with call rates <95% were excluded (Supplementary Table S2). There were 25, 21, and 9 eligible SNPs in OLA1, BARD1, and BRCA1, respectively, for association analyses. The pre-set critical values of the candidate genetic markers were 0.0040, 0.0048, and 0.011, respectively. A total of five SNPs in OLA1 were eligible for the validation study. None of SNPs in BRCA1 or BRAD1 had a p-value less than the pre-set critical values of candidacy. The primers for PCR reaction and annealing of these five candidate OLA1 SNPs are shown in Supplementary Table S3.
In the discovery case-control study, the minor allele frequencies of these five candidate OLA1 SNPs in the controls ranged from 0.198 to 0.268 and were 0.227 to 0.310 in the cases ( Table 1). The minor alleles of all candidate SNPs were significantly more prevalent in the cases. After adjustment for all significant anthropometric and clinical characteristics, heterozygous genotypes were correlated with significantly higher ORs of having thicker cIMT. The multivariate-adjusted ORs ranged from 1.59 to 1.86. Similar results were observed in the validation study.   Linkage analyses of these five OLA1 SNPs in the controls of the discovery case-control study showed that they formed a single LD block ( Figure 2). However, SNP rs12466587 was nearly completely linked with rs16862482 (r 2 = 0.97) and SNP rs35145102 showed stronger linear trends with rs201641962 and rs4131583. The most prevalent rs35145102-rs201641962-rs12466587-rs4131583-rs16862482 haplotype was H1-ATTGT, followed by haplotypes H2-GCCTC and H3-GCTTT ( Table 2). The estimated ORs for these prevalent haplotypes in the discovery case-control study were not statistically different from those in the validation study (all p-values > 0.2). The pooled OR for H1 haplotype was significantly decreased (OR = 0.72, 95% CI: 0.59-0.86) and for the H2 haplotype it was significantly increased (OR = 1.37, 95% CI: 1.12-1.69).
The results of the Breslow-Day test showed that the estimated ORs for all candidate SNPs in the discovery case-control study were not statistically different from those in the validation study (all p-values > 0.2). Accordingly, it was appropriate to obtain the pooled ORs for each candidate SNPs. The pooled ORs of thicker cIMT ranged from 1.55 to 1.72.

OLA1-Depleted Human Aortic SMCs Are Defective in Proliferation and Migration
Vascular ECs and SMCs play the key roles in the initiation and progression of athersclerosis and the stability of vascular plaque (see reviews in references [1][2][3][4]7]). Endothelium forms a barrier between blood and tissues, and its lesion increases its permeability to and accumulation in the subendothelial matrix of apoprotein-B-containing lipoproteins (apoB LPs), which in turn leads to fibrinolysis and migration and proliferation of vascular SMCs (see reviews in references [1][2][3][4]). Based upon our observations that the common genetic variations in the OLA1 gene we identified were statistically significantly associated with cIMT (Tables 1-3), we further investigated the physiological roles of OLA1 in human vascular SMCs and ECs, as described in this section and the subsequent section, respectively.
It is known that vascular SMCs and SMC-derived extracellular matrix accumulate in fibrous plaques (see reviews in references [1,2,4]), indicating that vascular SMCs migrate from media to intima and, thereafter, proliferate at the intima where fibrous plaques form. Furthermore, OLA1 plays a role in promoting cell migration and proliferation through regulating expression of E-cadherin, FAK, and/or p21 in nonvascular cells [28][29][30][31]. Thus, we applied the RNAi technique to determine whether OLA1 is crucial for proliferation and migration of human aortic SMCs. The OLA1 siRNA duplexes we used did efficiently deplete OLA1 in human aortic SMCs ( Figure 3A,B). Using the nuclear BrdU incorporation assay (see the Materials and Methods section (Section 4.8)), we found that, compared to non-silencing treatment, depletion of OLA1 caused a reduced proliferation of human aortic SMCs ( Figure 3C). Using the trans-well assay (see the Material and Methods section (Section 4.8)), we found that, compared to non-silenced human aortic SMCs, OLA1-depleted human aortic SMCs had reduced migration ability ( Figure 3D). These observations indicate that OLA1 is required for migration and proliferation of human aortic SMCs.
deplete OLA1 in human aortic SMCs ( Figure 3A,B). Using the nuclear BrdU incorporation assay (see the Materials and Methods section (Section 4.8)), we found that, compared to non-silencing treatment, depletion of OLA1 caused a reduced proliferation of human aortic SMCs ( Figure 3C). Using the trans-well assay (see the Material and Methods section (Section 4.8)), we found that, compared to non-silenced human aortic SMCs, OLA1-depleted human aortic SMCs had reduced migration ability ( Figure 3D). These observations indicate that OLA1 is required for migration and proliferation of human aortic SMCs.  After human aortic SMCs were transfected with non-silencing (NS) or OLA1 siRNA duplexes for 42 h, cells were collected and lysed for immunoblot analysis with antibodies against OLA1 and GAPDH (loading control). Immunoblot images of both OLA1 and GAPDH are shown in (A). The band intensities of OLA1 and GAPDH from (A) were determined using LS software from VisionWorks (UVP, LLC). The band intensity of OLA1 was normalized to that of GAPDH, and the relative protein levels of OLA1 normalized to the nonsilencing treatment in the graph shown in (B) are the means ± s.d. (error bars) from three independent experiments. ***, p < 5 × 10 −7 . (C) Depletion of OLA1 causes a reduced proliferation of human aortic SMCs. After human aortic SMCs were transfected with the indicated siRNA duplexes for 42 h, 1.5 × 10 4 cells were incubated with the BrdU-containing medium for 8 h. The BrdU-incorporated cells were stained using 3,3 -diaminobenzidine tetrahydrochloride for analysis (see the Materials and Methods section (Section 4.8)). Each independent experiment was quadruplicated. In the graphs, values are the relative percentages of the amounts of incorporated BrdU in cells normalized to the non-silencing treatment and the means ± s.d. (error bars) from three independent experiments. ***, p < 5 × 10 −4 (Student's t-test). (D) Depletion of OLA1 causes reduced migration ability in human aortic SMCs. After human aortic SMCs were transfected with the indicated siRNA duplexes for 42 h, 3 × 10 4 cells were resuspended in 100 µL of 0.5% FBS medium for the trans-well assay (see the Materials and Methods section (Section 4.8)). The cells stuck in the pores of the filter were imaged and quantitated. Each independent experiment was triplicated. In the graphs, values are the relative percentages of migrating cells normalized to the non-silencing treatment and the means ± s.d. (error bars) from three independent experiments. **, p < 5 × 10 −5 (Student's t-test).

OLA1-Depleted Human Aortic ECs Are Defective in Proliferation, Migration, and Tube Formation
ECs of vasa vasorum migrate to and proliferate to form the angiogenic buds in intimal hyperplasia and atherosclerotic lesions and, after bud expansion, the angiogenic buds merge and develop new vessels, which is required for plaque growth, plaque destabilization, and thromboembolic events (see review in reference [7]). Thus, we applied the RNAi technique to determine whether OLA1 is crucial for proliferation and migration of human aortic ECs. The OLA1 siRNA duplexes we used did efficiently deplete OLA1 in human aortic ECs ( Figure 4A,B). Using the nuclear BrdU incorporation assay, we found that, compared to non-silenced human aortic ECs, OLA1-depleted human aortic ECs were defective in proliferation ( Figure 4C). Using the trans-well assay, we found that, compared to non-silenced human aortic ECs, OLA1-depleted human aortic ECs had reduced migration ability ( Figure 4D). These observations indicate that OLA1 is required for migration and proliferation of human aortic ECs. After human aortic ECs were transfected with non-silencing (NS) or OLA1 siRNA duplexes for 42 h, cells were collected and lysed for immunoblot analysis with antibodies against OLA1 and GAPDH (loading control). Immunoblot images of both OLA1 and GAPDH are shown in (A). The band intensities of OLA1 and GAPDH from (A) were determined using LS software from VisionWorks (UVP, LLC). The band intensity of OLA1 was normalized to that of GAPDH, and the relative protein levels of OLA1 normalized to the nonsilencing treatment in the graph show in (B) are the means ± s.d. (error bars) from four independent experiments. **, p < 5 × 10 −5 . (C) Depletion of OLA1 causes a reduced proliferation of human aortic ECs. After human aortic ECs were transfected with the indicated siRNA duplexes for 42 h, 1.5 × 10 4 cells were incubated with the BrdU-containing medium for 8 h. The BrdU-incorporated cells were stained using 3,3′-diaminobenzidine tetrahydrochloride for analysis. Each independent experiment was duplicated. In the graph, values are the relative percentages of the amounts of incorporated BrdU in cells normalized to the non-silencing treatment and the means ± s.d. (error bars) from three independent experiments. *, p ˂ 5 × 10 −4 (Student's t-test). (D) Depletion of OLA1 causes reduced migration ability in human aortic ECs. After human aortic ECs were transfected with the indicated siRNA duplexes for 42 h, 3 × 10 4 cells were resuspended in 100 μL of 0.5% FBS medium for the transwell assay. The cells stuck in the pores of the filter were imaged and quantitated. Each independent experiment was triplicated. In the graphs, values are the relative percentages of migrating cells normalized to the non-silencing treatment and the means ± s.d. (error bars) from three independent experiments. **, p ˂ 5 × 10 −5 (Student's t-test). (E,F) Depletion of OLA1 causes reduced capillary tube formation by human aortic ECs. After human aortic ECs were transfected with the indicated siRNA duplexes for 42 h, 5 × 10 4 ECs were used for Matrigel tube formation assay (see the Materials and Methods section (Section 4.8)). The formed capillary tubes were imaged as shown in (E) and quantitated as shown in (F). Each independent experiment was triplicated. In the graph shown in (F), values are the relative percentages of the formed capillary tubes normalized to the non-silencing treatment and the means ± s.d. (error bars) from three independent experiments. ***, p ˂ 1 × 10 −6 (Student's t-test).

Discussion
In this study, we first identified five candidate SNPs-OLA1 rs35145102, rs201641962, rs12466587, rs4131583, and rs16862482-of cIMT in a case-control study and then validated their effects in another independent frequency-matched case-control study. The relationships between these SNPs and cIMT remained statistically significant after adjustment for the effects of traditional cardio-metabolic risk factors. We also found that these five SNPs formed a single LD block and the homozygosity of the most prevalent haplotype was correlated with significantly decreased OR of thicker cIMT. Further, we found The BrdU-incorporated cells were stained using 3,3 -diaminobenzidine tetrahydrochloride for analysis. Each independent experiment was duplicated. In the graph, values are the relative percentages of the amounts of incorporated BrdU in cells normalized to the non-silencing treatment and the means ± s.d. (error bars) from three independent experiments. *, p < 5 × 10 −4 (Student's t-test). (D) Depletion of OLA1 causes reduced migration ability in human aortic ECs. After human aortic ECs were transfected with the indicated siRNA duplexes for 42 h, 3 × 10 4 cells were resuspended in 100 µL of 0.5% FBS medium for the trans-well assay. The cells stuck in the pores of the filter were imaged and quantitated. Each independent experiment was triplicated. In the graphs, values are the relative percentages of migrating cells normalized to the non-silencing treatment and the means ± s.d. (error bars) from three independent experiments. **, p < 5 × 10 −5 (Student's t-test). (E,F) Depletion of OLA1 causes reduced capillary tube formation by human aortic ECs. After human aortic ECs were transfected with the indicated siRNA duplexes for 42 h, 5 × 10 4 ECs were used for Matrigel tube formation assay (see the Materials and Methods section (Section 4.8)). The formed capillary tubes were imaged as shown in (E) and quantitated as shown in (F). Each independent experiment was triplicated. In the graph shown in (F), values are the relative percentages of the formed capillary tubes normalized to the non-silencing treatment and the means ± s.d. (error bars) from three independent experiments. ***, p < 1 × 10 −6 (Student's t-test).
Given that neovascularization by ECs of vasa vasorum in atherosclerotic lesions is critical for plaque growth, plaque destabilization, and thromboembolic events (see review in reference [7]), we determined whether depletion of OLA1 may interfere with the ability of human aortic ECs to form vascular tubes and found that, compared to non-silenced human aortic ECs, OLA1-depleted human aortic ECs were defective in tube formation ( Figure 4E,F). This indicates that OLA1 regulates vascular tube formation by human aortic ECs.

Discussion
In this study, we first identified five candidate SNPs-OLA1 rs35145102, rs201641962, rs12466587, rs4131583, and rs16862482-of cIMT in a case-control study and then validated their effects in another independent frequency-matched case-control study. The relationships between these SNPs and cIMT remained statistically significant after adjustment for the effects of traditional cardio-metabolic risk factors. We also found that these five SNPs formed a single LD block and the homozygosity of the most prevalent haplotype was correlated with significantly decreased OR of thicker cIMT. Further, we found that loss of OLA1 caused reduced ability in human aortic ECs and SMCs to migrate and proliferate and reduced ability in human aortic ECs to form vascular tubes in cell culture system (Figures 3 and 4).
OLA1 is ubiquitously expressed in brain, thyroid, esophagus, testis, and 23 other tissues (see information at https://www.ncbi.nlm.nih.gov/gene/29789#gene-expression) (accessed on 15 March 2021). OLA1 belongs to the YchF subfamily of P-loop GTPases, which regulate diverse cellular functions, including protein translation, intracellular transport, signal transduction, and cell proliferation [22,39]. However, to our knowledge, there are no reports on OLA1 association with preclinical traits of atherosclerosis or cardiovascular events. Further, there are no reports showing that OLA1 directly influences the functions of human immune cells and vascular ECs and SMCs. Based upon the previous observations and our observations in this study, several lines of evidence reveal that the genetic variants in OLA1 may correlate with the development of atherosclerosis.
The first line of evidence is that SNP rs35145102 is located at the 3 -untranslated region (3 -UTR) of the OLA1 mRNA, while the others are intron variants. Given that the 3 -UTR is often targeted by microRNA for regulation of mRNA stability [40,41], we searched for the possible microRNA targeting the 3 -UTR of OLA1 mRNA in the TargetScan database (release 7.1) at the website http://www.targetscan.org/ (accessed on 3 March 2022; see reference [42]) and found that rs35145102 resides in a possible but less conserved target sequence of hsa-miR-6770-5p at the 3 -UTR of OLA1 mRNA (Supplementary Figure S1). This raises the possibility that rs35145102 is correlated with the expression level of OLA1. Thus, we retrieved the expression data in human immune cells by using the Ensemble Genome Browser (see the website at http://asia.ensembl.org/Homo_sapiens/Info/Index) (accessed on 15 April 2021) and found that the expression levels of OLA1 in monocytes are significantly correlated with SNPs rs35145102 and rs4131583, while those of OLA1 in macrophages are significantly correlated with SNPs rs12466587 and rs16862482 [43]. Furthermore, knockdown of OLA1 elicited an increased resistance to oxidizing agents in HeLa cells, whereas overexpression of OLA1 increased cellular sensitivity to oxidizing agents [25]. Therefore, the significant associations between these five OLA1 SNPs and cIMT may result from the differential expression of OLA1 in immune cells, which confer the differential resistance to oxidative stresses.
The second line of evidence is that OLA1 interacts with several proteins involved in the development of atherosclerosis. Among those molecules that interact with OLA1, BRCA1 is the most promising candidate. Although none of the nine BRCA1 SNPs in this study were correlated with cIMT thickening (Supplementary Table S2), a previous study has shown that the cytosine methylation changes at specific CpG sites in BRCA1 are associated with cIMT [36]. It is known that DNA cytosine methylation in the CpG dinucleotide usually results in gene silencing [44]. Additionally, BRCA1 mRNA and protein levels in samples of plaque-containing segments of human carotid arteries were significantly lower than those in the control samples of adjunct plaque-free segments [20]. Similarly, BRCA1 can protect human aortic smooth muscle cells against inflammation-and toxin-induced reactive oxygen species [19]. As compared with ad-null-treated ApoE −/− mice, ad-BRCA1-treated ApoE −/− mice developed significantly fewer aortic plaque lesions, exhibited reduced macrophage infiltration, and generated fewer reactive oxygen species [20]. These observations raise the possibility that a reduced expression of BRCA1 contributes to development of atherosclerosis. Further, given that OLA1 and BRCA1 regulate the cellular responses to oxidative stresses [19,20,25], it can be speculated that OLA1 and BRCA1 work together in the development of atherosclerosis.
The third line of evidence is that vascular endothelium forms a barrier between blood and tissues, and its lesion leads to fibrinolysis and migration and proliferation of vascular SMCs (see reviews in references [1][2][3][4]). Further, migration and proliferation of ECs, and then their neovascularization of vasa vasorum, in atherosclerotic lesions are crucial for plaque growth, plaque destabilization, and thromboembolic events (see review in reference [7]). In this study, we found that loss of OLA1 caused reduced ability in human aortic ECs to migrate, proliferate, and form vascular tubes (Figure 4). Therefore, it is possible that a reduced expression of or a defective mutation in OLA1 may lead to a lesion in the vascular endothelium, which occurs at the early stages of the development of atherosclerosis, or that overexpression of OLA1 promotes neovascularization by ECs of vasa vasorum, leading to plaque growth, plaque destabilization, and thromboembolic events in atherosclerotic lesions.
The last line of evidence is that it is known that one of the key steps for the development of atherosclerosis is that vascular smooth muscle cells migrate from media to intima and then proliferate in intima [2,45]. In this study, we found that OLA1 was required for migration and proliferation of human aortic SMCs ( Figure 3B,C), suggesting that OLA1 should play a role in the development of atherosclerosis. However, contrary to our observation, loss of OLA1 causes an increased proliferation of mouse pulmonary artery SMCs [27]. Indeed, OLA1 is required for migration and proliferation of mouse embryonic fibroblasts [28]. Therefore, it is possible that OLA1 exerts an effect on the proliferation of cells in either a species-or cell type-dependent manner. It has been shown that OLA1 binds to and inhibits GSK3 to negatively regulate E-cadherin expression and, thereby, promotes cell migration [30]. Further, OLA1 negatively regulates cell adhesion and spreading by attenuating FAK expression and increasing cofilin phosphorylation [31]. FAK controls proliferation and migration of rat aortic SMCs, and that phosphorylation of cofilin controls platelet-derived growth factor-induced migration of human aortic SMCs [46][47][48]. Furthermore, OLA1 negatively regulates translation of p21, a CDK inhibitor, and, thereby, maintains optimal cell proliferation for developmental progression in mice [28]. Additionally, p21 can inhibit proliferation of vascular ECs and migration of vascular SMCs [49,50]. Based upon these observations, it is possible that OLA1 regulates expression of E-cadherin, FAK, and/or p21 and, thereby, controls migration and proliferation of human aortic SMCs.
To further confirm that these five OLA1 SNPs are the major genetic determinants of cIMT, we searched for functional OLA1 SNPs with a minor allele frequency >3.0% in the Taiwan Biobank (see the website at https://taiwanview.twbiobank.org.tw/index) (accessed on 20 April 2021). A total of six functional SNPs were identified: rs35145102, rs3739153, rs3739152, rs6737629, rs1063613, and rs10209474 (see the website at https://taiwanview. twbiobank.org.tw/index) (accessed on 20 April 2021). Except for SNP rs10209474, a 5 -UTR variant, the others are 3 -UTR variants. The LD data in the 1000 Human Genome Project Phase 3-Southern Han Chinese (see reference [51]) showed that SNPs rs3739153 and rs3739152 were in complete LD (r 2 = 1.0), and these two SNPs were also in complete LD with two screened SNPs (rs12693034 and rs6741764). SNP rs6737629 was in complete LD with two screened SNPs (rs17239055 and rs2358443). SNP rs10209474 was closely linked with one screened SNP (rs12479030; r 2 = 0.802). Our discovery case-control study showed that SNPs rs12693034, rs6741764, rs17239055, rs2358443, and rs12479030 were not correlated with cIMT. Accordingly, rs3739153, rs3739152, rs6737629, and rs10209474 are unlikely to play critical roles in the development of atherosclerosis. Of note, SNP rs1063613 was not included in our screened SNPs and was not linked with any other SNPs located 50 Kb down-or up-stream of its location. Its physiological significance needs further exploration.
In the study, we conducted two independent case-control studies and used two different genotyping methods to explore the relationships between common genetic variants in OLA1, BRCA1, and BRAD1. Additionally, we found that the minor allele frequencies of the five OLA1 candidate SNPs in the controls were similar to the corresponding frequencies in the Taiwan Biobank, which enrolls subjects from the general population. However, there were potential limitations in the present study. The negative finding for the relationship between SNPs in BRCA1 and BARD1 genes and cIMT does not necessarily preclude the possibility that these two genes may play important roles in atherosclerotic diseases. In the study, only nine SNPs in BRCA1, in contract to 25 SNPs in OLA1 and 21 SNPs in BARD1, were screened for their association with cIMT. Accordingly, BRCA1 seems more likely to suffer from false negative bias. Additionally, to control the overall type I error and to reduce the likelihood of false negative findings at the same time, only SNPs with a p-value less than twofold the corrected significance level were included in the validation study. However, false-negative and false-positive findings may still exist.
There were two further potential limitations in this study. The first was that, before the start of cohort I, we conducted a prior study to evaluate the feasibility of obtaining data associated with pharmacological treatments. Most interviewees could provide the information related to the classes of medicines they took. However, many could not provide the names of medicines and the duration of medical treatment. Therefore, we did not collect detailed data associated with treatments and were unable to assess the effects of these treatments. However, based on the following reasons, our findings were more likely to be conservative. The reasons include: (a) genotypes of subjects were not influenced by treatment. (b) None of the subjects had ever received carotid ultrasonographic scans before. Therefore, it is very unlikely that individuals would take medicines to control carotid IMTs. Finally, (c) all technicians who operated the ultrasonographic systems and measured the carotid IMTs were blinded to examinees' clinical profiles. Variations in measurements of carotid IMTs tend to be non-differential. The second limitation is that, although our study revealed that OLA1 was required to promote proliferation and migration of human aortic ECs and SMCs and vascular tube formation by human aortic ECs in a cell culture system, we cannot clarify whether loss of OLA1 in vascular ECs or SMCs would cause development of atherosclerosis or whether overexpression of OLA1 in vascular ECs or SMCs would cause development of atherosclerosis. Therefore, vascular EC-or SMC-specific inducible OLA1 knockout mice are needed to verify whether loss of OLA1 in vascular ECs or SMCs may cause development of atherosclerosis in mice. Similarly, vascular EC-or SMC-specific inducible OLA1-overexpressing mice are also needed to verify whether overexpression of OLA1 in vascular ECs or SMCs may cause development of atherosclerosis in mice.
In conclusion, we identified five OLA1 SNPs showing significant associations with cIMT. These genetic markers formed a closely linked LD block. The homozygosity of the most prevalent multi-locus haplotype was correlated with significantly lower ORs of having thicker cIMT. Furthermore, we found that OLA1 was required for migration and proliferation of human aortic ECs and SMCs and regulated vascular tube formation by human aortic ECs (Figures 3 and 4). Given that vascular ECs and SMCs play the key roles in development of atherosclerosis [1][2][3][4]7], these genetic markers may be used for individual risk prediction of atherosclerotic diseases. These genetic markers may possess the potential for use in the development of therapeutic agents for the secondary prevention of atherosclerotic diseases.

Study Subjects
We used two case-control studies to screen and validate the relationships of common genetic variants in BRCA1, BARD1 and OLA1 genes with cIMT thickening. The cases and controls of the case-control studies were selected from among the participants of our community-based studies and have been described previously [38]. In brief, the subjects in the discovery stage were selected from among the participants of our previous community study (cohort I), which recruited middle-aged adults and elderly subjects from September 2010 to May 2012. The case group was a random sample of 284 subjects whose mean common carotid artery (CCA) IMT was ≥ 0.70 mm, the 75th percentile of the distribution of the mean far-wall CCA IMT. The control group was a random sample of 464 subjects who had a mean CCA IMT < 0.70 mm (Figure 1).
All participants of the two community-based studies voluntarily provided informed consent. The studies complied with the 1975 Helsinki Declaration on ethics in medical research and were reviewed and approved by the Institution Review Board of Mackay Memorial Hospital (14MMHIS075).

Measurements of Anthropometric and Cardio-Metabolic Profiles
The measurements of anthropometric characteristics and cardio-metabolic profiles have been described previously [52]. The anthropometric characteristics included age at enrollment, sex, body weight and height, and circumferences of the waist and hip. Blood pressure was measured three times after 10 min of rest. The cardio-metabolic profiles included systolic blood pressure (SBP), diastolic blood pressure (DBP), total cholesterol, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), fasting triglycerides (FTG), and fasting plasma glucose (FPG). A structured questionnaire was used to collect personal health behaviors, such as cigarette smoking, alcohol drinking, and dietary pattern. Personal histories of common diseases, including hypertension, hyperlipidemia, hyperglycemia, diabetes mellitus (DM), and CVDs, were also collected by the structured questionnaire.

The cIMT Measurement
The measurement of cIMT was performed as described previously [52]. In short, two experienced technicians who were blinded to patients' anthropometric and clinical characteristics obtained CCA images by using high-resolution B-mode ultrasonography systems (GE Healthcare Vivid 7 and Vivid E9; General Electric Company, Milwaukee, WI, USA). A well-trained technician retrieved the digital images and measured the far-wall IMTs blindly by using automatic contouring software (GE Healthcare EchoPAC version 112.0.2; General Electric-Vingmed, Horten, Norway). The IMT was defined as the distance between the lumen-intima and media-adventitia interfaces and included plaques. The average, minimum, and maximum IMTs of the distal 1 to 2 cm of the left and right CCA were recorded. Mean cIMT was calculated as the mean of the left and right average IMTs and was used for correlation and regression analyses.

SNP Selection and Genotyping
In the discovery study, we used a plate (Axiom ® CHB 1 Array Plate; Affymetrix Ltd., Santa Clara, CA, USA) to screen the associations of common genetic variants in OLA1, BRCA1, and BARD1 with cIMT. In the plate, there were 59 SNPs within 25 Kb up-or downstream of the OLA1 (n = 27), BARD1 (n = 23), and BRCA1 (n = 9) genes. The eligibility of SNPs for association analyses were a call rate >95%, a p-value in the Hardy-Weinberg equilibrium (HWE) test in the controls that was >0.001, and a minor allele frequency >3%.
The significance level of the discovery study, designated α corrected , was calculated as 0.05 divided by the number of eligible SNP in the locus. To reduce the influence of over-adjustment, we used 2 × α corrected as the pre-set critical value for candidate genetic markers. Eligible SNPs with a p-value less than the pre-set critical value for candidacy were subject to a validation study using the Sequenom iPLEX MassARRAY system (Sequenom, San Diego, CA, USA). All genotyping was performed at the National Center for Genome Medicine, Academic Sinica, Taiwan.

siRNA Duplexes
The OLA1 siRNA duplexes used in this study were the same as previously described [29]. The sense strand sequences of non-silencing and OLA1 siRNA duplexes were 5 -r(UUCUCCGAACGUGUCACGU)dTdT-3 and 5 -r(GCUGCUGGAAAGUACAGAC) dTdT, respectively. The siRNA duplexes were synthesized by and purchased from GE Healthcare Dharmacon Inc., Lafayette, CO, USA.

Immunoblot Assay
After cells were transfected with the indicated siRNA duplexes for 42 h, cells were lysed with RIPA buffer and 200 µg cell lysates were used for the immunoblot assay. To perform the immunoblot assay, OLA1 was detected using rabbit anti-OLA1 antibodies (Abcam Inc., Cambridge, UK) and GAPDH (internal loading control) using rabbit anti-GAPDH (Genetex Inc., Irvine, CA, USA), and they were visualized by chemiluminescence (Genetex Inc., Irvine, CA, USA) and imaged with a ChemiDoc-It ® 815 Imager (Analytik Jena, An Endress + Hauser Company, Jena, Germany) after incubation with HRP-conjugated secondary antibodies (Genetex Inc., Irvine, CA, USA).

Assays for Cell Proliferation, Cell Migration, and Tube Formation
Proliferation of human aortic endothelial and smooth muscle cells was determined with the nuclear bromodeoxyuridine (BrdU) incorporation assay using a BrdU immunochemistry kit (Merk Millipore Inc., Burlington, MA, USA). The assay was performed following the manufacturer's instruction (Millipore Inc.). In brief, cells (1.5 × 10 4 /well) were seeded on coverslips in a 24-well plate and incubated with 10 µM BrdU for 8 hours. After being washed with 1 × PBS buffer, cells were fixed with ice-cold 70% ethanol at 4 • C for 30 min. The BrdU-incorporated cells were then stained using 3,3 -diaminobenzidine tetrahydrochloride, and the stained intensity was determined by a microplate reader (Molecular Devices Inc., San Jose, CA, USA).
The migration assay for human aortic ECs and SMCs was performed using a 24-well trans-well plate with a pore size of 8 µm (Corning Inc., Corning, NY, USA). A total of 3 × 10 4 cells were resuspended in 100 µL of 0.5% FBS medium and plated in the upper chamber, and 600 µL of 2.5% FBS medium was added into the lower chamber. After incubation at 37 • C for 4 h, cells were fixed with −20 • C methanol for 5 min and stained with 18.7 µM bisbenzimide (Sigma-Aldrich Inc., Saint Louis, MO, USA) in the dark for 15 min. After the filter was washed with 1 × PBS three times, the cells on the upper surface of filter were completely wiped away with the cotton swab. The cells stuck in the pore of the filter were imaged under a fluorescence microscope (Leica Microsystems Inc., Wetzlar, Germany) using a 40 × objective lens and the image analyzed using the QWIN software (Leica Microsystems Inc., Wetzlar, German).
A Matrigel tube formation assay was performed following the manufacturer's instruction (BD Biosciences Inc., Mississauga, ON, Canada). The Matrigel (BD Biosciences Inc., Mississauga, ON, Canada) was dissolved at 4 • C overnight before use. Each well in the 48-well plate was coated with 150 µL Matrigel and then incubated at 37 • C for 30 min. A total of 5 × 10 4 ECs were seeded per well on the layer of polymerized Matrigel for 24 h. The formed capillary tubes were imaged under an inverted phase contrast microscope (Zeiss Inc., Oberkochen, Germany) using a 20 × objective lens and quantified by measuring the long axis of each tube in three randomly selected fields per well using MacBiophotonics Image J software.

Statistical Analyses
In the study, the chi-square test was used to compare whether there were significant differences in the frequency distributions of genotypes between cases and controls. The relationships with thicker cIMT for each eligible SNP were assessed by minor allele dominant, recessive, and additive and (minor and major allele) co-dominant genotypic models and adjusted for age, sex, and other significant cardio-metabolic factors. Assessment of the pairwise LD of candidate SNPs in the control group of the discovery study was performed in Haploview 4.2 software [53].
Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated with multivariate logistic regression models. The homogeneity of the estimated ORs between the discovery and validation studies was evaluated with the Breslow-Day test. The pooled ORs were obtained by summation of the study-specific ORs, weighting by the inverse of their variances when the estimated ORs were not statistically different between studies. Except for the assessment of pairwise LD, all statistical analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The datasets used and/or analyzed in the current study are available from the corresponding author (Li-Yu Wang) on reasonable request.