Brief Overview of Ongoing Projects
High-Throughput Molecular Profiling
We are using several established or emerging high-throughput, genome-wide techniques such as RNA-seq, HiChIP and ATAC-seq to study expression patterns, transcription factor binding and chromatin interactions and accessibility. Applying these methods in cells of high relevance to insulin resistance, such as adipocytes, hepatocytes and skeletal myocytes, can improve our ability to disentangle genetics of insulin resistance and related traits.
In some projects, we use colocalization methods to systematically prioritize the most likely causal gene in loci associated with insulin resistance and related traits (such as fasting insulin, glucose, waist-hip ratio, HDL cholesterol and triglycerides). These methods seek to determine whether the same variants in a locus are responsible for signals from different layers of high-throughput molecular data, often GWAS and eQTL signals. We can also integrate this with other layers of high-throughput molecular data, such as ATAC-seq - to study chromatin accessibility - and HiChIP - a novel method for sensitive and efficient analysis of protein-centric chromosome conformation.
Pooled CRISPRi Screens
We have identified a large number of loci associated with insulin resistance and related traits using human genetics. By use of pooled genetic screens based on CRISPR, which enable hundreds to thousands of programmed perturbations per experiment, we can prioritize the most likely candidate genes within these loci using unbiased experimental approaches.
We are currently working with two different strategies for phenotypic readouts on the pools of genome-engineered cells. In one project, we apply single-cell RNA sequencing using the recently updated CROP-seq protocol as the readout after CRISPRi pooled gene perturbation. Another approach is based on reporter gene expression changes following pooled sgRNA delivery for CRISPR-mediated perturbations. The pooled screening approach allows us to evaluate many genes in parallel in a relevant biological context, and the single cell RNA sequencing helps us dissecting the mechanisms underlying changes in insulin resistance and related phenotypes.
In Vitro Studies in Adipocytes and Other Cell Types
To functionally characterize genes putatively implicated in development of insulin resistance, we work with CRISPR gene perturbations in adipocytes, hepatocytes and skeletal myocytes followed by functional assays relevant to insulin resistance. We primarily work with human cell lines including SGBS preadipocytes, HepG2 hepatocytes and HMCL-7304 skeletal myocytes; but for some applications, we also use murine cell lines, such as 3T3-L1 or OP9 adipocytes, AML12 hepatocytes and C2C12 myocytes, as well as primary cells.
We employ CRISPRi and CRISPRa approaches perturbing target gene expression to investigate cellular phenotypes. In addition, CRISPRi can be used to effectively, specifically, and homogeneously silence expression of up to 2-3 genes simultaneously, which also allows us to test for epistasis among target genes. After specific target gene perturbation, we assess the effect of gene regulation on a series of phenotypes affected by insulin resistance; specifically, basal and insulin-stimulated glucose uptake, lipolysis, insulin signaling, adipogenesis (adipocytes only), mitochondrial function, fatty acid oxidation, metabolomics and RNA.
In Vivo Studies
For selected genes showing insulin resistance phenotypes in human genetics and in vitro experiments, we proceed to in vivo models to study physiological effects after gene perturbation. Over the past years, we have participated in the development of zebrafish models to allow quantification of body size, lipid accumulation in liver and body fat, pancreatic ß-cell number, liver size and vascular accumulation of lipids. More recently, we are focusing more on mouse models which have proven to be critical in the assessment of insulin resistance, recapitulating expected phenotypes for Mendelian genetic forms of insulin resistance.
Our mouse experiments include feeding on standard chow or high-fat diet followed by tracking of body weight, assessing fasting lipids and glucose, as well as performing insulin and glucose tolerance testing. In selected projects, we measure whole-body insulin sensitivity by euglycemic clamps, assess basal metabolic rate using metabolic cages, and VO2 max by treadmill exercise testing. The effect of knockouts are also assessed using metabolomic and lipidomic profiles from plasma and liver, as well as RNA sequencing of white adipose tissue (subcutanous and visceral), brown adipose tissue, skeletal muscle and liver.
Human Genetics and Other -Omics
Over the past ten years, we have led and participated in many genome-wide association studies (GWAS) of complex traits. At this point, these analyses comprise a standard methodology that we routinely apply as part of larger projects, often as a starting point for functional studies in the wetlab or as a part of a Mendelian randomization study.
We have also led a large number of studies applying other -omics methods, primarily epigenomics (mostly DNA methylation), transcriptomics (using RNA sequencing), proteomics (mostly affinity-based using proximity extension assays), and metabolomics (mostly mass spectroscopy-based). In some of the cohort studies that we work with, we have produced all of these layers of data, providing a very powerful resource for integrative -omics studies.
Phenome-Wide Association Studies (PheWAS)
We have a series of projects investigating the phenome-wide characteristics of individuals carrying gene-disrupting alleles; hence, characterizing phenotypic effects as a function of number of functional copies of specific genes. Using such "human knockout models", we can simulate and predict what the result would be if blocking the corresponding protein.
Increased knowledge about downstream effects of naturally occurring variation in druggable genes provides important insights into disease mechanisms, predict potential for repurposing drugs and unknown side effects. This can help initial characterization of the functional impact of gene variation while bypassing the inherent translational uncertainty of model systems. We work with a range of projects, including collaborations with pharmaceutical companies, aiming to predict efficacy and adverse effects of novel drugs for insulin resistance, non-alcoholic fatty liver disease and other cardiometabolic diseases.
We use Mendelian randomization (MR) extensively to study causal relationships of risk factors with health outcomes. This method has become hugely popular, and we are amongst the world leaders using this approach to address questions of clinical important for cardiometabolic disease. We are working with various risk factors, such as salt, alcohol and coffee intake, migraine, lung disease and periodontitis in relation to coronary heart disease, stroke and type 2 diabetes.
Also, we are working with several projects addressing the causal role of biomarkers representing different biological systems in cardiovascular disease and type 2 diabetes. Such biomarkers include vitamin D, IGF-1, Apo-AI, urate, SHBG, estrogen and testosterone – all debated as to whether they are causally related to disease, and hence if their perturbation should be part of a preventive strategy or not - but also novel biomarkers from our proteomics and metabolomics efforts.
We have performed a large number of precision medicine studies using big data approaches over the past ten years. We have developed novel methods for risk prediction metrics, reported new biomarkers and their role in prediction of cardiovascular disease, and were among the first to address the role of genetics in risk prediction. More recently, we are using machine learning techniques as an unbiased way of discovering novel risk predictors among a very large set of potential variables.
For many of our projects in this space, we use the UK Biobank due to its very large sample size and rich data. We have several recent publications applying these methods in this large cohort study, but we were also the first to utilize the UK Biobank in a high-profile paper. Our UbbLE project was published in Lancet in 2015, reporting risk predictors for five year mortality, and we built a home page that allows calculation of five-year mortality risk, UbbLE age and exploring of mortality predictors in an interactive fashion.
UK Biobank and Other Cohort Studies
In 2006-2010, the UK Biobank recruited 502,650 participants aged 37-73 years to undergo physical measurements, detailed assessments about risk factors and future disease events, and sampling of blood, urine and saliva. Genome-wide genotyping has been done in all participants, along with extensive phenotyping including incident outcomes. We are working on a wide range of projects using this excellent cohort using the methodologies described above to address questions of high importance for public health and clinical medicine.
In addition to UK Biobank, we are also working with several other datasets. We have led a range of –omics projects in several Swedish cohorts – ULSAM, PIVUS, TwinGene and EpiHealth - and are currently also working with several Stanford-based study samples. Such -omics efforts include genomics, transcriptomics, epigenomics, proteomics and metabolomics, often used in combination - aiming at increasing the biological knowledge of obesity, insulin resistance and CVD, and to identify new biomarkers for risk prediction and novel drug targets.