Top Advances in Functional Genomics and Translational Biology for 2015
During the past year, the field of cardiovascular genomics has witnessed publication of an unprecedented number of outstanding articles. In addition to large-scale investigations of original hypotheses (eg, pleiotropic effects of height loci on coronary artery disease [CAD]), our community has welcomed seminal functional genomics studies using gene editing, induced pluripotent stem cells (iPSC), and total RNA sequencing. Taken together, these studies have produced valuable insights into mechanisms underlying the observed associations with cardiovascular disease (CVD).
The mission of the Functional Genomics and Translational Biology (FGTB) Council of the American Heart Association (http://www.my.americanheart.org/fgtbcouncil) is to advance new discoveries in the fields of genetics, omics-based approaches, and translational biology, as well as to facilitate their application in cardiovascular health and disease. By creating a multidisciplinary collaborative environment, this Council integrates scientific knowledge from molecules to populations and contributes to the global goal of building healthier lives, free of CVDs and stroke.
The Early Career Committee of the FGTB Council collectively selected studies that we believe to be the major advances published in 2015. In this Special Report, we summarize 10 articles that cover a variety of disciplines represented on the FGTB and highlight their particular significance to the cardiovascular field.
Through interactions with the host genes and environmental exposures, gut bacteria have previously been shown to influence metabolic and inflammatory processes although the specific contributions to disease risk had remained elusive because of the lack of large-scale human studies. In the past year, Fu et al1 began to peel back the layers of uncertainty surrounding the role of the gut microbes in CVD. Using fecal samples from the LifeLines DEEP cohort, they have profiled the gut microbiome of 893 participants by 16s ribosomal RNA sequencing and identified novel cross-sectional associations between several bacterial taxa and cardiometabolic phenotypes such as the body mass index and blood lipid levels. Gut microbiota composition accounted for ≈4.5% of the interindividual variance in body mass index, 6% in triglycerides, and 4% in high-density lipoproteins; the models containing gut microbiome variation significantly (P<4.1×10−3) outperformed those containing only age, sex, and genetic factors. Interestingly, the relationships between lipids and the gut microbiota were not modified by host genetics, operationalized as a risk score comprised known lipid and obesity loci. This finding highlights the importance of environmental factors in establishing gut microbial communities and the independence of microbial effects on lipid metabolism.
For the third year in a row, our list of top discoveries includes micro-RNAs (mi-RNAs), short regulatory elements that have emerged as key players in cardiometabolic disease pathogenesis. Two recent studies have reported notable insights into the role of mi-RNAs in controlling lipid homeostasis. Using a novel genome-wide screening assay, Goedeke et al2 identified miR-148a as a direct repressor of low-density lipoprotein receptor activity in human hepatocytes. Subsequent experiments conducted both in vitro and in vivo in APOBTg;Ldlr−/+ mice (which display an high-density lipoprotein [HDL]– and low-density lipoprotein [LDL]–dominant lipoprotein profile) showed miR-148a to be transcriptionally activated by Sterol regulatory element binding protein-1c, a sterol regulatory element-binding protein that is in turn activated by liver X receptor. Consistent with prior human studies, miR-148 inhibition increased LDL receptor expression and decreased LDL-cholesterol (LDL-C) levels in vivo. Furthermore, antagonism of miR-148 increased HDL-cholesterol (HDL-C) levels by increasing ABCA1 transcription and cholesterol efflux to APOA1. Although the murine metabolic peculiarities temper the translational excitement of these findings and warrant follow-up investigations in nonhuman primates, the observed simultaneous changes in both LDL-C and HDL-C concentrations make miR-148a an attractive therapeutic target for dyslipidemias.
The mi-RNA-148a finding was echoed in the study by Wagschal et al,3 who examined the 69 mi-RNAs located in physical proximity to lipid-related polymorphisms in genome-wide association data on more than 188 000 individuals. Of those, pathway analysis linked 4 mi-RNAs (miR-128-1, miR-148a, miR-130b, and miR-301b: all located in intergenic spaces) to cardiometabolic disorders, prioritizing them for validation. Wagschal et al3 subsequently pursued several functional experiments, which shed light on the causal mechanisms underlying the observed associations. The 4 mi-RNAs repressed LDLR and ABCA1 expression in human hepatocytes and mouse macrophages, thus directly impacting cholesterol efflux. In vivo, short-term antisense inhibition of miR-148a and miR-128-1 in C57BL/6J mice triggered an increase in both HDL-C levels and LDL-C clearance. The effects of miR-148a inhibition on HDL-C but not on LDL-C persisted >16 weeks in a follow-up study of Apoe−/− mice put on a Western-type diet. On the contrary, antagonizing miR-128-1 in those mice resulted in decreased very-low-density lipoprotein and LDL-C levels, as well as decreased triglycerides, improved glucose homeostasis, and reduced hepatic steatosis. In contrast to the miR-148a findings by Goedeke et al,2 the beneficial effects of miR-128-1 observed by Wagschal et al3 were not mediated by hepatic Ldlr and Abca1 levels, necessitating future investigations into the exact mechanism of association. Both studies have significantly advanced current understanding of mi-RNA involvement in metabolic processes, generating hypotheses for future studies and informing development of new treatments.
The study of human macrophages is crucial to understand several biological processes and genetic defaults resulting in the maladaptative immune response to cholesterol accumulation in atherosclerotic plaques. The iPSCs are an interesting alternative to genetically rearranged tumor-induced cell lines and human peripheral blood mononuclear cells. In this study,4 Zhang et al4 aimed to understand the morphological, functional, and transcriptional characteristics of iPSC-derived macrophages (IPSDMs) using a high-throughput procedure and compare human peripheral blood mononuclear cell–derived macrophage counterparts. First, the authors developed an efficient protocol to differentiate human iPSCs to a large amount of highly pure macrophages (≈2×107 of CD45+/CD18+ cells per 6-cell plate). These IPSDMs shared morphological and phenotypic characteristics with native macrophages, as demonstrated by the expression of a large list of specific cellular markers. Human peripheral blood mononuclear cell–derived macrophages and IPSDMs also shared several functional characteristics, including polarization to proinflammatory M1 with comparable phagocytosis capacities, and had ability to tissue repair and metabolic homeostasis as M2 macrophages. Using total RNA sequencing data, authors showed that human peripheral blood mononuclear cell–derived macrophage and IPSDM also share whole gene expression characteristics with 98% overlap of expressed genes and 89% of differentially expressed genes. Transcriptomic changes were also comparable during M1 and M2 polarization. One important finding is the IPSDMs capacity to reproduce macrophage cholesterol defects in a cell model for Tangier disease, a rare Mendelian disorder characterized by loss-of-function mutations of the ABCA1 transporter. The authors showed that IPSDM and human peripheral blood mononuclear cell–derived macrophage of patients with Tangier disease carrying these mutations had similarly eliminated cholesterol efflux to apoA-I and impaired efflux to HDL, which validate a specific use of IPSDM to assess the functionality of genetic mutations in lipid metabolism.
To date, 48 common genomic loci have been implicated with risk of developing CAD through genome-wide association studies (GWAS). By contrast, the effects of rare and low-frequency variants on common disease have not been tested as extensively, partly because of limited study power. Recently, Nikpay et al5 performed a comprehensive GWAS meta-analysis of ≈185 000 CAD cases and controls and utilizing the expanded coverage of the phased 1000 Genomes reference panel for improved imputation, which enabled the authors to test ≈10 million genetic variants with CAD6,7 (and 2.7 million common and rare variants, respectively). Aside from confirming the majority of known common variant associations with CAD, the authors also identified 10 new common loci associated at a genome-wide level of significance; 8 of the 10 new common loci conferred an additive effect on CAD risk (odds ratio range, 1.04–1.27). As a novelty of this GWAS meta-analysis, the effect of 2 loci was recessive. Notably, several of the 10 novel loci were in close vicinity of genes implicated in arterial vessel wall pathophysiology. No rare or low-frequency genetic variants (including insertions and deletions) were associated with CAD on a genome-wide level although joint association analysis to identify possible synthetic associations based on linkage disequilibrium structure among rare variants (P<5×10-5) yielded 15 such variants. However, these 15 rare variants only explained ≈2% of overall CAD heritability. Collectively, this study demonstrates the utility of improved reference panels for GWAS discovery, reinforces the common disease-common variant hypothesis, and underscores the increasingly importance of genetic modifiers of vessel wall integrity in CAD.
There is an established epidemiological evidence supporting an association between short stature and increased risk for CAD and several cardiovascular risk factors (eg, hypertension and diabetes mellitus). Facilitated by GWAS, important progress has been made in understanding the genetics of height and CAD. In this study, Nelson et al6 had the interesting idea of using the genetic information available in large samples (≈200 000 individuals) combing individual cohorts to test the hypothesis of an increased cardiovascular risk in individuals, as determined by their genetic make-up according to 112 height-associated single nucleotide polymorphisms. By using Mendelian randomization, Nelson et al6 demonstrated a relative increase of 13.5% in the risk for CAD per 1 SD decrease in height. Of note, there was a lack of association between genetically determined height and CAD in women after sex stratification, which implies that the CAD association linked with genetically determined height is only manifested in men. The authors also generated a genetic risk score based on the effects size of 112 single nucleotide polymorphisms associated with height and found that the use of quartiles of genetic score confirmed the Mendelian randomization finding with quartile 4 (group of individuals with most increasing height alleles) versus quartile 1 having a significant protection from CAD (odds ratio, 0.74; 95% CI, 0.68–0.80), P<0.001). Looking at CVD risk factors, the authors found that LDL-C and triglycerides were also higher in genetically determined short individuals. In subgroup analyses, no association with systolic and diastolic blood pressure was identified, but could be because of lack of power as smaller samples were available for GWAS for these traits. However, this reasoning is not applicable to body mass index and type 2 diabetes mellitus analyzed in larger samples with sufficient power, which suggests the absence of a link between genetically determined height and these diseases. Finally, pathway-enrichment analysis indicates bone morphogenetic protein- and transforming growth factor-β signaling, axon guidance and signal transducer and activator of transcription 3 and insulin-like growth factor 1 as potential overlapping pathways between genes near-height and CAD loci. This approach presents the main advantage of reducing the influence of demographic, lifestyle, socioeconomic, or behavioral confounders that have an independent effect on height and the risk of CAD. However, the authors acknowledge that this method does not completely rule out these confounders as shorter persons could adopt a higher CVD risk behavior.
As illustrated by the analyses of shared genetics of height and CAD, the study of genetic pleiotropy is important as it provides useful insights into the underlying function of the involved genes in physiology and disease. In this second study on genetic pleitropy, Homsy et al7 tested whether protein damaging denovo mutations in 1213 parent–offspring trios (probands and their unaffected parents) with congenital heart disease (CHD) was also associated with extra cardiac congenital anomalies (CAs) and/or neurodevelopmental disabilities (NDDs). Using exome-sequencing data, the authors found an overall 1.4-fold enrichment of protein damaging mutations among CHD probands when compared with healthy controls. The assessment of the top quartile of genes highly expressed in the heart identified a significant 2.4-fold enrichment of damaging mutations among CHD probands, whereas no enrichment was found among controls. Notably, a significant dose–response association with protein damaging mutation enrichment was also observed in highly expressed in the heart genes: 2-fold for CHD probands with NDDs, 2.9-fold for CHD probands with CAs, and 4.7-fold for CHD probands with both CAs and NDDs. Damaging mutations in highly expressed in the heart genes accounted for a 20% of CHD probands with both CAs and NDDs, but only 2% of patients with isolated CHD. Notably, post hoc expression analysis of the damaging mutations showed a high-expression overlap between developing heart and brain, suggesting that the damaging mutations in CHD may also contribute to nonsyndromic NDDs/CAs. To explore this further, 7 cohorts ascertained for NDDs were examined (excluding CHD). Here, a significant enrichment of CHD identified damaging de novo mutations, particularly among highly expressed in the heart genes (≤4.4-fold). This study provides support to the notion that a single mutation can have pleiotropic effects and in this case contribute to both CHD and neurodevelopmental abnormalities.
Truncating mutations in titin (TTN) have been identified as the most common cause for familial or sporadic dilated cardiomyopathy, accounting for ≈20% of all cases. However, despite recognizing the importance of this gene to the risk of dilated cardiomyopathy, establishing the specific TTN mutation as the culprit has been made difficult by the size of titin (which is the biggest protein in humans), its heterogeneity, as well as the lack of appropriate disease models to determine pathogenicity. Utilizing cellular reprogramming techniques to generate cardiac microtissues engineered from human iPSCs, Hinson et al8 sought to determine the pathogenicity of mutations identified in affected and unaffected dilated cardiomyopathy patients with TTN mutations. Generating human cardiac microtissue models of TTN variants, the authors were able to demonstrate that some missense TTN variants were associated with impaired contractile force, rendering them pathogenic. Hinson et al8 also tested whether the location of mutations within the I-band or A-band of the sarcomere affected their possible pathogenicity. Using functional analyses including cardiac microtissue models and CRISPR-cas9 editing, the authors found that A-band truncating TTN variants produced a markedly reduced contractile function, and thus were more pathogenic compared with I-band truncating TTN variants. Additional experiments using total RNA sequencing technology suggest that the mechanism for differences in pathogenicity between A- and I-band mutations was caused by alternative exon splicing, making I-band TTN variants less pathogenic. Considering that TTN is also important for sensing and responding to myocardial stress, the authors tested whether TTN mutations were associated with an abnormal stress response. By mimicking β-adrenergic response using isoproterenol, cardiac microtissue models with TTN variants were shown to have an impaired inotropic and chronotropic response, which suggests a reduced cardiac adaptation ability in the setting of increased mechanical load and β-adrenergic signaling. The study is noteworthy as it couples a range of recent advances in stem cell reprogramming, gene editing, and tissue engineering to provide functional and genomic insights that ultimately enable the determination of possible pathogenicity for mutations previously classified as being of uncertain clinical significance.
For the second time in the FGTB council special reports, we highlight an additional episode of the fat mass and obesity (FTO) locus saga. The work from Claussnitzer et al9 is one of the most compelling functional explorations at a GWAS locus. To decipher the regulatory circuitry and mechanistic basis of weight balance in humans, the authors used an optimal combination of complementary functional genomics experiments. FTO has been previously limited to a 47 kb minimally associated region with obesity and supported by chromatin conformation capture sequencing to physically interact with the promoter of the Iroquiois-class homeodomain protein 3 (IRX3) in the brain of mouse. Here, the authors performed an important in silico investigation of public resources of epigenomic annotation and chromosome conformation to identify a long enhancer measuring 12.8 kb. They also showed that the causal variant lies within a long-range enhancer that regulated not only IRX3 but also IRX5 expressions in human adipocyte progenitors. Most importantly, the authors linked the effect of the FTO locus to the control of thermogenesis by the analysis of transcroptomics of perirenal and perithyroid brown adipose tissue, where they found that IRX3 and IRX5 expression is negatively correlated with genes involved in mitochondrial function and positively correlated with genes of lipid metabolism. These differences in expression were also regulated by genotypes in preadipocytes from patients. When compared with noncarriers, preadipocytes from FTO risk alleles carriers presented reduced mitochondrial oxidative capacity, reduced white adipocyte browning, and reduced thermogenesis; all these functions were restored after IRX3 and IRX5 siRNA knockdown in preadipocytes from patients carrying the risk allele and induced by the overexpression of IRX3 and IRX5 in preadipocytes of noncarriers. Finally, the identification of the causal variant was achieved by a genomic conservation analyses across species that pointed at rs1421085, predicted to alter the interaction with ARID5B, a gene expression repressor that controls the expression of IRX3 and IRX5. CRISPR-cas9 editing specifically at the rs1421085 position from TT (protective) to CC (risk) impeded the interaction with ARIDB5 and increased the expression of both genes, reduced the expression of genes of lipid metabolism, and increased lipid storage and lipolytic markers. In summary, this comprehensive mechanistic study of FTO functionality points at adipocyte thermogenesis regulation as a potential mechanism of obesity, with a gain of function (increased enhancer capacity) of 2 prolipid storage and antilipolysis genes to facilitate weight gain. Additional episodes from the FTO story will attempt to clarify which mechanism is pathogenic: the thermogenesis effect demonstrated here using human cell models and/or or appetite regulatory effect as suggested by previous data in the mouse.
Finally, in the year 2015, the efforts of the Roadmap Epigenomics Project culminated in the publication of a valuable resource of 111 reference human epigenomes, characterized on DNA methylation, DNA accessibility, RNA expression, and histone-modification patterns.10 Combining the new data with 16 previously published reference epigenomes profiled by the Encyclopedia of DNA Elements project, the Roadmap team conducted a uniquely comprehensive analysis of regulatory elements that control gene expression in 127 types of cells and tissues, including embryonic stem cells. Among their big picture findings are epigenomic enrichments of disease-related genetic variants in relevant tissues, distinct epigenomic signatures of large-scale regions that suggest separate chromosomal domains, increased evolutionary conservation of enhancer/promoter regions (≈5% of each epigenome), and relationships between histone mark combinations and gene expression that are not reflected in either DNA methylation or accessibility. On the methodological side, this project demonstrated the feasibility and validity of imputing epigenomic data at high resolution by exploiting correlations between marks and lineages and offered techniques for integrative analysis of multiple epigenomic layers. Much like HapMap and the 1000 Genomes Project have done in the past, the global reference maps of regulatory elements and analytic methods presented by the Roadmap Epigenomics Project empower the next generation of genome-wide studies of complex traits.
We thank early career members of the following Early Career Members of the Functional Genomics and Translational Biology Council for their assistance in selecting the articles: Pankaj Arora, Kent Arrell, Emelia J. Benjamin, Audrey Y. Chu, Eric Green, Jennifer E. Huffman, Jennie Lin, Almudena Martínez-Fernández, Kiran Musunuru, Anna Pilbrow, Sony Tuteja-Stevens, and Connie Wu.
- Received February 25, 2016.
- Accepted March 1, 2016.
- © 2016 American Heart Association, Inc.
- Fu J,
- Bonder MJ,
- Cenit MC,
- Tigchelaar EF,
- Maatman A,
- Dekens JA,
- et al
- Zhang H,
- Xue C,
- Shah R,
- Bermingham K,
- Hinkle CC,
- Li W,
- et al
- Homsy J,
- Zaidi S,
- Shen Y,
- Ware JS,
- Samocha KE,
- Karczewski KJ,
- et al
- Hinson JT,
- Chopra A,
- Nafissi N,
- Polacheck WJ,
- Benson CC,
- Swist S,
- et al