We adopted a novel approach to understanding the co-occurrence of AD and CV disease and traits based on several multi-trait GWAS to characterise the shared genetic architecture of AD with CV traits. Convergent evidence from colocalisation between AD, CV traits and eQTLs prioritised two genetic regions that each included a single candidate causal variant (rs11786896 expressed via PLEC and rs7529220 expressed via C1QA, C1QB, and C1QC) shared between AD and AF. Single-cell RNA-sequence data, co-expression network and protein-protein interaction analyses together were consistent in showing that PLEC is upregulated in left ventricular endothelium and cardiomyocytes with HF and in brain astrocytes with AD. By contrast, while C1Q genes are predicted to be upregulated with greater disease risk in cardiac macrophages for HF and in brain microglia for AD, we found opposite directions of difference with disease for both. We explored differences in the direction of changes from early disease (low beta-amyloid pathology load) to late (high beta-amyloid pathology load) for microglia and found the congruence with directions predicted for disease risk in early disease that was lost with in later disease progression. Our findings provide new insights into genetic pleiotropic effects and potential shared mechanisms causally related to both AD and CVD.
Out of the several CV traits and diseases examined with AD, AF showed the largest number of pleiotropic signals with AD. Numerous observational studies, provide growing evidence that AF is associated with cognitive impairment, risk of AD and other dementias16. However, it has been unclear whether the diseases have a shared pathophysiology or whether the relationship arises as downstream consequences of AD (e.g., stroke). Here we provide evidence defining common genetic determinants for the two diseases. The colocalised intronic variant rs11786896 within the plectin gene (PLEC) was associated with lower expression of PLEC in the cardiac left ventricle (and skeletal muscle) and increased the risk of both AD and AF. PLEC is a member of a protein family, named plakins, with a crucial structural role in the cytoskeleton including cell architecture and tissue integrity and a partially functional role in the assembly, positioning, and regulation of signalling complexes17,18. Plectin is expressed as various tissue-dependent protein isoforms in several tissues with each tissue to be characterised by different proportion and composition of plectin isoforms and each distinct isoform to be characterised by specific functions19–23. However, the precise role of plectin in macrophages is unknown and thus, further study is needed to determine its specific functions and interactions in macrophages.
Previous studies of human tissues or preclinical models provide independent evidence for an association of plectin with diseases including AD and AF24,25. Our new data and analyses provide evidence that risk in AD is affected via functions of plectin in astrocytes26. Astrocytes play multiple roles, central to the pathology of AD, including metabolic support for neurons, modulation of brain microvascular function and, through activities associated with those of microglia, inflammatory responses26,27. We hypothesise that these functional roles are mediated in part by interactions of plectin with intermediate filaments (IFs), microtubules and actin filaments26. IFs are important structural components of the cytoskeleton with crucial roles in synaptic activity, neurogenesis and repair after brain injury28. Differences in expression of plectins modulate neuronal function and vesicular trafficking generally and interactions with tau suggest potential roles specific to AD29–31. PLEC may play related roles in cardiomyocytes for assembling and mobilizing the intermediate filaments and their networks, effects that both modulate contractile function in cardiomyocytes and inflammatory responses in macrophages32.
Another colocalised variant between AD and AF, the intergenic rs7529220, which is located 19k upstream from Heparan Sulfate Proteoglycan 2 (HSPG2) and 21k downstream from Chymotrypsin Like Elastase 3B (CELA3B), was associated with increased risk of AD and AF and higher expression of three genes of the Complement Component 1, Q Subcomponent (C1Q) family (C1QA, C1QB, C1QC) in breast mammary tissue (and, by inference, in brain vasculature). The variant is located 680kb downstream of C1Q genes. The complement system plays a central role in synaptic remodelling in the brain and in cellular damage response more generally in the body33,34. We hypothesise that greater expression of C1Q may lead to higher activity of the complement system which in turn may potentiate synapse loss in early AD35. Similarly, C1Q has roles in the genesis of atherosclerotic plaques36 and in the regulation of early stages of inflammatory responses to the cardiomyocyte injury associated with a range of cardiac traits37.
Our study, which applied the MTAG approach in a novel way across diseases, had several strengths. First, we secured high statistical power for our study by including GWAS with substantial sample sizes ranging from 185,000 to 1,000,000 participants and we boosted the power even higher by performing suitable multivariate methods. Second, we combined advanced methods of genetic epidemiology and basic sciences and sought to provide supporting evidence from a variety of data. However, a number of limitations also must be acknowledged. We restricted our analyses to a population of European ancestry. The lack of genetic diversity may have hampered the possibility of detecting other relevant variants. Additionally, we did not investigate a considerable portion of the genetic predisposition coming from rare variants (MAF < 1%) as we excluded them from our analyses. However, including these variants might lead to false-positive findings and biased results. Moreover, we used statistical methods to detect pleiotropy, and therefore considered a genetic locus pleiotropic if it was statistically significantly associated with two or more phenotypes. However, this approach for identification of pleiotropic genes may not always highlight shared biological pathways, as the identified genes could affect the traits independently via different pathways (horizontal pleiotropy), or they could even be expressed in different tissues in response to different signals38,39. Furthermore, due to a limited number of cells for specific cell types, we had to combine single-cell data from multiple samples. We focused on tissue samples that were already enriched for cardiomyocytes, endothelium, and macrophages. Finally, the expression for some candidate genes in our data was limited and thus additional sequencing data and reads are needed to investigate them further. Additional RNA sequencing data of different AD and CV conditions would probably be even more informative.
In conclusion, we performed a multi-trait analysis on AD and CV traits and a subsequent colocalisation analysis detecting 19 shared genetic loci and further prioritizing two shared causal variants between the aforementioned traits. Our findings define shared mechanisms for AD and different cardiovascular diseases. The complement system has been explored as a target for novel preventive or disease-modifying therapies in cardiovascular disease40 and AD41. Our work suggests that plectin or members of its interactome could offer new and potentially promising targets for preventive and therapeutic medicines with benefits across these common comorbid disorders.