Introduction
Every year since 1948 the American Society of Human Genetics (ASHG) has hosted the world’s largest conference dedicated to showcasing cutting edge findings, developments, and future trends in human genetics and genomics research. The conference draws an impressive attendance of over 8,000 participants from around the globe which include leading scientists, clinicians, and industry experts. This year a group of Metabolon scientists attended the meeting to best understand how we can help drive groundbreaking advancements in the field.
During our talks with key scientists and opinion leaders we were surprised to learn that even though genomics is ubiquitously used to study human diseases, the field recognizes that genomics cannot explain everything about disease phenotypes. Here, we highlight key takeaways from our interactions with conference attendees and discuss the untapped contributions that metabolomics can make to advancing the field of genetics.
What can metabolomics reveal that genomics can’t?
The cellular and molecular events that contribute to a phenotype are mediated by genes, transcripts, proteins, and metabolites, 4 sets of molecules that represent different, but interrelated levels of cellular function. Gene expression is the first step in the central dogma of biology and as such, the functional readout of genomics does not always align with a physiological or pathophysiological endpoint. Furthermore, factors other than genetics contribute to an organism’s phenotype, which are not always reflected in expression of genes. By contrast, metabolites typically represent the last step of a cellular process, enabling them to not only manifest the combined inputs from the genome, transcriptome, and proteome, but also to reflect cellular processes that are influenced by external factors including the environment, lifestyle choices, and therapeutic interventions. The convergence of these internal and external inputs makes metabolites the closest cellular readout of an organism’s phenotype.
In the classic example of Mendel’s pea plant experiments we plainly see how genetics contribute to the phenotype. However, those experiments did not account for other factors including the type of soil, use of fertilizer, amount of sunlight and water the plant receives, and how the plant reacts to those things, will all contribute to the phenotype too. In humans, contributing factors that affect the phenotype include race/ethnicity, diet, exercise levels, and geographical location. Thus, studying metabolites in conjunction with other types of molecules can help provide a complete picture of the phenotype.
How can metabolomics differentiate the influence of external factors from genetic inputs to better explain the phenotype?
Several studies presented at the ASHG conference reported using metabolomics alongside genomics to better understand mechanisms that directly influence phenotypes. Here, we summarize some key poster presentations that demonstrate how much further understanding can go when metabolomics is included in phenotype characterization studies.
R. Lelievre et.al. Abstract 4064T. Gene-environment and metabolite-environment interaction effects of intraocular pressure in the Canadian Longitudinal Study on Aging.
Many genetic variants have been linked to increased intraocular pressure (IOP), a high-risk factor for hereditary glaucoma. Some studies show that making healthy lifestyle choices lowers the risk of developing IOP in persons with a genetic variant, yet the mechanisms that underly this outcome are poorly understood. To better understand the relationship between external factors and genetic variants in the onset of IOP this study investigated the influence of vitamin C-related metabolite levels, frequency of alcohol consumption, smoking status, and fruit and vegetable consumption with six genetic variants associated with intraocular pressure (IOP). Genomics and metabolomics data from 8060 participants from the Canadian Longitudinal Study on Aging was analyzed and linear and linear mixed models adjusted for relevant covariates were used to identify associations between the factors tested. After multiple testing corrections statistically significant associations (p < 0.05) were found between two of the tested variants (rs12414734 and rs8050812) and alcohol consumption, and between a third variant (rs61484427) and smoking status. Here, metabolomics helped reveal external factors that influence the onset of disease. Understanding how genetic variants and metabolites interact may reveal pathways that mediate the onset of IOP, leading to improved preventative interventions.
N. Wang et.al. Genetic architecture of circulating metabolites and its impact on health from trans-omics for precision medicine (TOPMed) program.
Circulating metabolite levels can be affected by genetics. Although hundreds of loci associated with circulating metabolites have been identified, little information is available on associations between metabolites and rare and structural variants found in multi-ancestry populations. In this study, metabolomics data from 25,058 samples across several large cohorts was pooled and harmonized. From this cumulative dataset, 231 novel locus-metabolite pairs were discovered. Follow up phenome-wide association studies (PheWAS) highlighted multiple metabolite common loci-trait associations associated with various diseases. The authors further harmonized an additional 2000 metabolites from large cohorts comprised of individuals from African, East Asian, European, and Latino ancestry. This study identified 99 novel locus-metabolite pairs, which will be further analyzed in a future study. Altogether, these data demonstrate the influence that race/ethnicity can have on metabolite-gene associations that contribute to the phenotype.
P. Jain et.al. Abstract 1134W. Genetic architecture of plasma metabolites in the Southeast Asian population.
Genome wide association studies (GWAS) have identified many genetic loci associated with metabolic variation but most of those studies were done in European populations. This group conducted the first large scale GWAS of plasma metabolite levels in 8200 multi-ethnic, Asian individuals in the Health for Life in Singapore (HELIOS) study. Their findings identified 2617 independent genetic variants across 502 genomic loci associated with 852 metabolites at genome-wide significance. Multiple loci found here were unreported in previous metabolite GWAS studies. These findings also highlighted associations observed in only one population, which included 63 independent variants associated with 82 metabolites in Indians and 41 variants associated with 42 metabolites in Malay that were not observed in the Chinese sub-population. Here, metabolomics helped demonstrate the relationship between ethnicity, genetic variants, and their bearing on metabolite levels, which all contribute to the phenotype and may have implications as to what normal health looks like in different ethnic populations.
E. Alhathli et.al. Abstract 5091W. Hypothesis-free mendelian randomization identifies New Metabolites and Athletes Metabolites linked to risk of ALS.
Most cases of amyotrophic lateral sclerosis (ALS) are sporadically caused by a combination of genetic and environmental factors, and very little is known about the influences these variables have on the onset of disease or the mechanisms that regulate these interactions. Here, Mendelian Randomization (MR), a machine learning algorithm, was used to infer statistically causal relationships between certain exposures and disease risk. GWAS of 575 plasma-based metabolites in a normal population were compared to those from a GWAS of ALS patients. Analysis causally associated five metabolites with increased risk for ALS. Two of these hits were in the carnitine synthesis pathway, which has been linked to ALS severity. Since strenuous physical exercise has also been previously linked to the onset of ALS, this group compared GWAS of 751 metabolites from 490 elite athletes to the same ALS GWAS and analyzed the shared metabolites. Acetylcarnitine was significantly associated with reduced risk of ALS in both datasets. Altogether, these findings show how metabolomics can identify functional contributing factors to a phenotype and lead to better understanding of disease pathophysiology.
B. Jumentier et.al. Abstract 5119W. Mendelian randomization reveals causal links between circulating metabolites and type 1 diabetes.
Type I diabetes (T1D) is thought to develop from a combination of genetic and environmental factors, including diet, but the relationship between these factors and how they work together to arrive at the onset of disease remains a topic of intense investigation. The goal of this study was to better characterize hypothetical cause-and-effect relationships between certain metabolites and genetic variants associated with T1D. To this end, the study authors tested whether genetically predicted levels of circulating metabolites are causally linked to T1D risk. 1679 metabolite-T1D associations were investigated. 46 of these metabolites showed a significant effect on T1D. Colocalization analysis showed six of these metabolites shared a causal variant with T1D while two presented significant genetic correlation with T1D. Other metabolites showed a protective effect on development of T1D. In this study metabolomics identified causal circulating metabolites for T1D, while also revealing other metabolites that have a significant effect on T1D but were not significantly associated with a causal variant, thus generating new hypotheses regarding the onset of disease.
How are leading scientists using metabolomics to test their hypotheses?
While learning about studies that have reached a stage of completion, we spoke with several scientists about their work that is still in progress. We were excited to hear that many investigators are turning to metabolomics to test their hypothesis regarding phenotypic traits.
One investigator at the University of Utah is using metabolomics to better characterize the off-target effects of Ibuprofen in people with rare congenital disorders of glycosylation. Ibuprofen is a standard of care for this patient population, yet it affects other aspects of metabolism, which could have implications for kidney and liver function. Metabolomics will allow key pathways to be analyzed and off-target effects to be characterized.
Another investigator at Vanderbilt is studying Sickle Cell Anemia (SCA). SCA is a bleeding disorder associated with many comorbidities including stroke, cardiac events, and kidney dysfunction. This investigator is using metabolomics in longitudinal studies to try and identify and differentiate SCA patients likely to develop one or more of these comorbidities with the end goal of being able to anticipate clinical outcomes sooner and thus start preventative treatments earlier.
We also spoke to another investigator who is currently using all 4 omics technologies to study Chronic Obstructive Pulmonary Disorder (COPD). He correlated his preliminary findings to clinical measurements of COPD and shared with us that proteomics and metabolomics provided significantly better diagnostic markers than transcriptomics and genomics, and that metabolomics provided the best correlative markers for several measurements, perhaps due to them being the closest molecules to the phenotype.
Conclusions
Our interactions with key scientists and opinion leaders at the ASHG conference showed how using metabolomics to generate phenotypic context around genomics data advanced scientific discovery and understanding. For over 20 years Metabolon has been helping scientists use, understand, and integrate metabolomics into their studies to drive discovery and innovation. To see how Metabolon’s industry-leading metabolomics platform and software, and scientific experts can help you take advantage of metabolomics in your studies, speak with one of our experts here.
