Research

Overview

The Rader lab leverages human genetics of a wide range of cardiometabolic diseases and traits as well as Alzheimer’s disease to identify new variants, genes, and pathways that underlie these phenotypes.  The research team employs a variety of experimental systems including mice, cell lines, human induced pluripotent stem cells, and humans to elucidate the molecular mechanisms by which these genes and pathways influence these phenotypes. In addition to elucidating fundamental mechanisms by which the protein influences relevant biology, the influence of specific mutations on protein structure and function are being explored. The lab makes use of the Penn Medicine Biobank, a large academic biobank with large-scale whole exome data linked to EHR phenotype data, for genetic discovery, biomarker studies, and ‘recall-by-genotype’ deep phenotyping studies. In addition to generating new insights in human biology and disease, the lab also works on translational therapeutics targeting these genes and pathways for cardiometabolic disease.

Currently Active Research Projects

Structure-function and physiology of APOA5, a natural activator of lipoprotein lipase

APOA5 is a liver-secreted protein with strong genetic association to plasma triglycerides and coronary artery disease.  This project integrates protein structure-mapping, biophysical studies, natural human genetic variants, and functional studies in vivo and in vitro to inform structure-function and mechanism of ApoA5.

Molecular mechanisms by which ABCA7 and ABCA1 activity influence Alzheimer's disease

ABCA1 and ABCA7 are related members of the ATP-binding cassette transporter subfamily A (ABCA) that exports lipid species out of cells.  Human genetics has firmly associated genetic variants that reduce their activity with increased risk of Alzheimer’s disease.   This project seeks to specifically elucidate the structure and function of ABCA7, the lipids it transports and in which cells, and to clarify its role in contributing to AD pathogenesis.  ABCA1 will be studied in parallel.  The primary model system is human iPSCs engineered to be deleted in either ABCA7 or ABCA1 and then differentiated to microglia, astrocytes and neurons to establish the effects on AD-relevant functions in these cells. Additional studies are aimed at a) identifying ABCA7 protein domains that distinguish it with regard to specificity of lipid translocase activities; b) in vitro characterization of the functional effects of naturally-occurring ABCA7 coding variants that are significantly associated with AD.

Functional interrogation of genes at GWAS loci associated with both plasma lipids and cardiovascular diseases

A large number of GWAS loci associated with plasma lipid traits are also associated with one or more major cardiovascular diseases, both atherosclerotic and non-atherosclerotic. The lab employs a variety of computational approaches to select and prioritize specific candidate causal genes at these loci, most of which do know harbor ‘known’ lipid genes.  Candidate genes are overexpressed and silenced in the human liver cell line Huh7 as an additional screen and then prioritized genes are overexpressed and knocked down in mouse liver to assess lipid phenotypes and mechanisms.  In addition, human induced pluripotent stem (iPSC) cells are CRISPR-engineered to delete genes of interest and differentiated to hepatocytes for phenotyping. In selected cases, multi-omics and deep phenotyping of somatic gene-targeted mice and stored plasma from PMBB participants carrying genetic variants of high interest are used to gain further insight into biology and mechanisms. In some cases, strategically-selected naturally-occurring protein-coding variants in genes of interest are studied compared to wild-type protein for insight into structure-function.  

‘Genome-first’ deep phenotyping of human subjects with genetic variants of high interest

The lab leverages the extensive genomic data in the PMBB to perform ‘deep dives’ into genes of high interest.  Participants with rare putative loss-of-function (pLOF) and computationally predicted deleterious missense variants are identified and compared with non-carrier controls through phenome-wide association studies (PheWAS) and other phenotypes are generated from the electronic health record through natural language processing and machine learning approaches.  Stored plasma samples are utilized for multi-omics studies.  In some cases, participants with rare variants of interest are invited to participate in additional ‘deep phenotyping’ studies that are customized to the gene and hypothesis.  Given our particular interest in lipoprotein metabolism, we often perform ‘oral fat tolerance testing’ and even detailed assessment of lipoprotein kinetics using stable isotopically-labeled amino acids incorporated endogenously into newly-synthesized proteins.

Generation of imaging-derived phenotypes using machine learning for integration with genomics

Participants in the PMBB have had many imaging studies performed clinically and working with colleagues in radiology we have developed machine learning approaches to quantitating a wide range of imaging-derived phenotypes (IDPs) for use in discovery research, such as integration with genomic data.  One example of this work focused on the quantitation of hepatic fat from CT scans, and we are applying this approach to a wide range of IDPs.

Research Studies for which we are recruiting study subjects

No active recruitment at this time. Please check again soon!