A multi-omics glimpse into the biology of arterial stiffness

It has long been recognized that the structure of arteries throughout the vascular tree is not uniform. Notably, the media of large proximal (central) vessels contains relatively much greater amounts of elastin and elastic lamellae than smaller, more distal (peripheral) arteries; the converse is true of vascular smooth muscle cells. Under physiological conditions, the greater elasticity of central arteries compared with more muscular peripheral arteries allows conversion of the pulsatile nature of ventricular ejection into a relatively steady flow of blood at the distal end of the arterial system, conferring protection from pulsatile energy [1,2]. Furthermore, these differences in impedance can generate partial wave reflections, which arrive in the aorta during diastole, boosting diastolic blood pressure and augmenting coronary perfusion pressure [3]. Arterial stiffness is a broad term used to describe loss of arterial compliance and changes in vessel wall properties [1]. Stiffening of central arteries impairs the above protective mechanisms and can in turn lead to damage of the microvascular beds within end organs [1,2] – the brain and kidneys are particularly susceptible as both are torrential circulations with minimal vascular resistance [4]. Furthermore, this decreased compliance causes reflected waves to arrive in the aorta prematurely, during systole, raising systolic blood pressure and perhaps more importantly, reducing diastolic blood pressure with a concomitant reduction in coronary perfusion pressures [3]. Although arterial stiffness can be assessed using a variety of techniques, carotid–femoral pulse wave velocity (cf-PWV) is the current ‘gold standard’ [5]. Measured noninvasively by transcutaneously recording pulse waves over the common carotid and femoral arteries, cf-PWV is assessed by measuring the pulse transit time and the distance travelled between the two recording sites [PWV (m/s) = travel distance (m)/transit time (s)]. Consequently, cf-PWV is easy to obtain with relatively inexpensive equipment and minimal expertise. It has been well established that arterial stiffness increases with both age and hypertension [6,7]. In recent years, arterial stiffness, as measured by cf-PWV, has also become an established predictor of cardiovascular risk in ‘apparently healthy’ individuals [8,9], as well as those with hypertension [10], chronic kidney disease [11], and diabetes mellitus [12]. In addition, cf-PWV is associated with all-cause and cardiovascular-specific mortality [13], coronary artery calcification [7], carotid atherosclerotic plaque burden [7], carotid intraplaque haemorrhage [14], peripheral vascular disease [15], increased cerebral white matter lesions [16,17], reduced cognitive performance [18], as well as progression of chronic kidney disease [19,20]. The predictive value of cf-PWV for coronary artery disease, cardiovascular events, and stroke appears independent of traditional risk factors, such as sex, hypertension, diabetes mellitus, chronic kidney disease, and smoking status [21]. Unfortunately, despite this ever-growing body of evidence linking arterial stiffness to adverse cardiovascular outcomes, the molecular factors that influence the initiation and progression of arterial stiffness remain poorly understood [22]. The decreasing costs of genetic analysis, increasing availability of publicly available resources with data from next-generation sequencing studies, and awareness of such resources within the broad scientific community has brought a change to the way genomics, transcriptomics, and epigenetics are applied to biomedical research. It is becoming increasingly common to use ‘omics’ not only as powerful tools at the discovery stage but also in combination to decipher the complex network of molecular interactions underlying the susceptibility to common cardiovascular disorders. In the present issue of the Journal, Mangino et al. have applied this multi-omics approach to gain insight into the biological mechanisms behind arterial stiffness. A genome-wide association analysis found a number of single nucleotide variants associated with cf-PWV (and therefore arterial stiffness) in just one gene: calcium and integrin-binding family member 2 (CIB2) [23]. The functions of the CIB2 protein are not fully understood, although structural similarity to calcineurin 2 and calmodulin may imply an intercellular signalling activity modulated by calcium binding [24]. It is tempting to speculate that expression of CIB2 could be linked to arterial stiffness through metabolism of calcium. The strongest association between CIB2 and arterial stiffness mapped to a common intronic single nucleotide variant – (rs7164338). To further explore how rs7164338 might modulate CIB2 gene expression, the authors used several public gene expression resources: MUltiple Tissue Human Expression Resource [25], GENe Expression VARiation (GENEVAR) (website: https://www.sanger.ac.uk/resources/software/genevar, data from Stranger et al.[26]), and Genotype-Tissue Expression project [27] that all contain high-resolution transcriptomic profiles for many different tissue and cell types, as well as genetic information. These extremely high-resolution transcriptomic data can be used in many different ways to study the associations with a phenotype or effect of a genetic variant on gene expression (Fig. 1). Through exploration of these resources, the authors provided evidence for the potential regulatory effect of rs7164338 on CIB2 mRNA expression in lymphoblastoid cell lines, skin, and adipose tissue. The direction and the significance of the association between the genotype and CIB2 mRNA expression were consistent across each of the examined tissues. This suggests that the identified common intronic variant (or its proxies) may be biologically active at the transcriptional level and alter the bioavailability of CIB2 mRNA in a genotype-dependent manner in multiple tissues.FIGURE 1: Central relevance of next-generation RNA-sequencing data to investigation in the post-genome-wide association study era.Further analysis of DNA methylation data provided additional insights into how CIB2 may affect arterial stiffness. Indeed, a highly significant inverse correlation between rs7164338 genotype and the degree of DNA methylation in whole blood was identified using methylation microarrays. This may suggest that the variant directly influences the cellular machinery responsible for DNA methylation. Furthermore, CIB2 mRNA levels and methylation were also significantly associated demonstrating the interplay between genotype, gene expression, and epigenetic factors. Taken together, these findings from different levels of molecular biology, appear to point to a genotype-controlled process of altered DNA methylation leading to changes in CIB2 expression that in turn translate into arterial stiffness. Putting all these independent associations together into a rational hypothesis is complicated by the noncausal nature of each association (correlation does not imply causation). Therefore, in a further Mendelian randomization analysis [28], the authors combined the data collected at the genomic, transcriptomic, and epigenetic levels to investigate if genotype does indeed have a causal effect on CIB2 mRNA levels. A Mendelian randomization analysis uses sample genotype as a randomizing variable to replicate the properties of a randomized control trial where participants are grouped, not by treatment group, but according to genotype [28]. The theoretical basis behind Mendelian randomization is that if two variables are causally linked (i.e. promoter methylation and gene expression) then a genetic variant that is found to influence the first variable is also expected to influence the second (if they are causally related). However, along with the power of Mendelian randomization comes strict assumptions and restrictions and no small amount of controversy in the literature (see [29] and [30] for informal argument and counter argument). A Mendelian randomization analysis must employ a genotype marker that is not directly associated with the outcome of interest (in this case CIB2 expression) or any other confounding variable [31] (e.g. binding of regulatory factors). The authors’ Mendelian randomization analysis found a significant causal relationship between genotype, DNA methylation, and CIB2 gene expression. The literature is supportive of such findings; often when the variant in question lies in a transcription factor binding site [32] it can lead to significantly modified DNA methylation, which in turn translates into altered gene expression [33] in the local genomic region [34]. Beyond the primary interest of the biological mechanisms behind cf-PWV, the work of Mangino et al. provides potential solutions for two further theoretical issues common to current biomedical research: (1) what is the likely impact of intergenic and intronic variants identified by genome-wide association studies (GWAS)? (2) how can we rationalize multiple independently derived findings into a cohesive hypothesis? Furthermore, this work is a powerful example of how publically available data can be used to augment and generalize a purely phenotype-driven analysis. More work in this area is still necessary; the level of replication with previous GWAS of arterial stiffness was limited and the associated variants identified in this study were not reported before. Furthermore, a transcriptome profile more closely related to arterial tissue would be preferable over tissues used in this study, but obtaining suitable human tissue is difficult. Further successes of omics-driven strategies in deciphering the biological mechanisms of genetic associations are critically dependent on the availability of resources with appropriate human tissues. Our recent data from the TRANScriptome of renaL humAn TissuE (TRANSLATE) study showed how application of next-generation RNA sequencing to human kidney tissue may help to identify renal mechanisms driving the previously reported association between a common intronic variant in the fibroblast growth factor 1 gene and blood pressure [35]. We anticipate that such applications of transcriptomics and epigenomics will become increasingly popular in the post-GWAS era bringing us closer to unravelling the secrets of many associations between genes, genetic variants, and complex cardiovascular phenotypes. ACKNOWLEDGEMENTS Requests for Reprints: Professor Maciej Tomaszewski. Conflicts of interest There are no conflicts of interest.