Genome-wide association studies (GWAS) with intermediate phenotypes, like changes in metabolite and protein levels, provide functional evidence to map disease associations and translate them into clinical applications. However, although hundreds of genetic variants have been associated with complex disorders, the underlying molecular pathways often remain elusive. Associations with intermediate traits are key in establishing functional links between GWAS-identified risk-variants and disease end points. Here we describe a GWAS using a highly multiplexed aptamer-based affinity proteomics platform. We quantify 539 associations between protein levels and gene variants (pQTLs) in a German cohort and replicate over half of them in an Arab and Asian cohort. Fifty-five of the replicated pQTLs are located in trans. Our associations overlap with 57 genetic risk loci for 42 unique disease end points. We integrate this information into a genome-proteome network and provide an interactive web-tool for interrogations. Our results provide a basis for novel approaches to pharmaceutical and diagnostic applications.

Author contributions In an academic-industry partnership, SomaLogic, Inc. and the academic collaborators worked together on study design, interpretation of the data and preparation of the manuscript. S.A.W., P.G. and N.W. were responsible for designing, writing and final editing of the manuscript and responses to reviewer comments. In addition to all authors being generally involved in the program, specific contributions were as follows: M.K. and M.J.S. were accountable for the data from the Whitehall II study and advised on the study design for the CV and diabetes models. C.L. and N.W. were accountable for the data from the Fenland study and advising on diabetes risk and behavioral models. C.B. and M.A.S. were accountable for the data from the Heritage Family study. C.J. was accountable for the data from the HUNT3 study. R.O. was accountable for the data from the Covance study.

SignificanceDuchenne muscular dystrophy (DMD) is a rare and devastating muscle disease caused by mutations in the X-linked DMD gene (which encodes the dystrophin protein). Serum biomarkers hold significant potential as objective phenotypic measures of DMD disease state, as well as potential measures of pharmacological effects of and response to therapeutic interventions. Here we describe a proteomics approach to determine serum levels of 1,125 proteins in 93 DMD patients and 45 controls. The study identified 44 biomarkers that differed significantly between patients and controls. These data are being made available to DMD researchers and clinicians to accelerate the search for new diagnostic, prognostic, and therapeutic approaches.

The vitamin D receptor (VDR) heterodimerizes with retinoid X receptors (RXR) on many vitamin D-responsive promoter elements, suggesting that this complex is the active factor in vitamin D-mediated transcription. However, the mechanism of transcriptional regulation following VDR-RXR binding to DNA is not well characterized. Using a yeast two-hybrid protein interaction assay, we demonstrate that VDR forms specific protein: protein contacts with the basal transcription factor TFIIB. Deletion analysis indicated that the carboxyl-terminal ligand binding domain of VDR interacted with a 43-residue amino-terminal domain in TFIIB. The interaction with TFIIB showed selectivity for the ligand binding domain of VDR as similar regions of RXR alpha or of retinoic acid receptor alpha did not couple with TFIIB. Binding assays with purified proteins showed a direct interaction between VDR and TFIIB in vitro. These data suggest a mechanism for VDR-dependent transcription in which protein contacts between VDR and TFIIB may impart regulatory information to the transcription preinitiation complex.

Gemfibrozil-1-O-beta-glucuronide (GEM-1-O-gluc), a major metabolite of the antihyperlipidemic drug gemfibrozil, is a mechanism-based inhibitor of P450 2C8 in vitro, and this irreversible inactivation may lead to clinical drug-drug interactions between gemfibrozil and other P450 2C8 substrates. In light of this in vitro finding and the observation that the glucuronide conjugate does not contain any obvious structural alerts, the current study was conducted to determine the potential site of GEM-1-O-gluc bioactivation and the subsequent mechanism of P450 2C8 inhibition (i.e., modification of apoprotein or heme). LC/MS analysis of a reaction mixture containing recombinant P450 2C8 and GEM-1-O-gluc revealed that the substrate was covalently linked to the heme prosthetic heme group during catalysis. A combination of mass spectrometry and deuterium isotope effects revealed that a benzylic carbon on the 2',5'-dimethylphenoxy group of GEM-1-O-gluc was covalently bound to the heme of P450 2C8. The regiospecificity of substrate addition to the heme group was not confirmed experimentally, but computational modeling experiments indicated that the gamma-meso position was the most likely site of modification. The metabolite profile, which consisted of two benzyl alcohol metabolites and a 4'-hydroxy-GEM-1-O-gluc metabolite, indicated that oxidation of GEM-1-O-gluc was limited to the 2',5'-dimethylphenoxy group. These results are consistent with an inactivation mechanism wherein GEM-1-O-gluc is oxidized to a benzyl radical intermediate, which evades oxygen rebound, and adds to the gamma-meso position of heme. Mechanism-based inhibition of P450 2C8 can be rationalized by the formation of the GEM-1-O-gluc-heme adduct and the consequential restriction of additional substrate access to the catalytic iron center.

A hallmark of Alzheimer's disease is the brain deposition of amyloid beta (Aβ), a peptide of 36-43 amino acids that is likely a primary driver of neurodegeneration. Aβ is produced by the sequential cleavage of APP by BACE1 and γ-secretase; therefore, inhibition of BACE1 represents an attractive therapeutic target to slow or prevent Alzheimer's disease. Herein we describe BACE1 inhibitors with limited molecular flexibility and molecular weight that decrease CSF Aβ in vivo, despite efflux. Starting with spirocycle 1a, we explore structure-activity relationships of core changes, P3 moieties, and Asp binding functional groups in order to optimize BACE1 affinity, cathepsin D selectivity, and blood-brain barrier (BBB) penetration. Using wild type guinea pig and rat, we demonstrate a PK/PD relationship between free drug concentrations in the brain and CSF Aβ lowering. Optimization of brain exposure led to the discovery of (R)-50 which reduced CSF Aβ in rodents and in monkey.

Decision trees have been used extensively in cheminformatics for modeling various biochemical endpoints including receptor-ligand binding, ADME properties, environmental impact, and toxicity. The traditional approach to inducing decision trees based upon a given training set of data involves recursive partitioning which selects partitioning variables and their values in a greedy manner to optimize a given measure of purity. This methodology has numerous benefits including classifier interpretability and the capability of modeling nonlinear relationships. The greedy nature of induction, however, may fail to elucidate underlying relationships between the data and endpoints. Using evolutionary programming, decision trees are induced which are significantly more accurate than trees induced by recursive partitioning. Furthermore, when assessed on previously unseen data in a 10-fold cross-validated manner, evolutionary programming induced trees exhibit a significantly higher accuracy on previously unseen data. This methodology is compared to single-tree and multiple-tree recursive partitioning in two domains (aerobic biodegradability and hepatotoxicity) and shown to produce less complex classifiers with average increases in predictive accuracy of 5-10% over the traditional method.

In recent years the trend in combinatorial library design has shifted to include target class focusing along with diversity and drug-likeness criteria. In this manuscript we review the computational tools available for target class library design and highlight the areas where they have proven useful in our work. The protein kinase family is used to illustrated structure-based target class focused library design, and the G-protein coupled receptor (GPCR) family is used to illustrate ligand-based target class focused library design. Most of the tools discussed are those designed for libraries targeted to a single protein and are simply applied "brute-force" to a large number of targets within the family. The tools that have proven to be the most useful in our work are those that can extract trends from the computational data such as docking and clustering or data mining large amounts of structure activity or high throughput screening data. Finally, areas where improvements are needed in the computational tools available for target class focusing are highlighted. These areas include tools to extract the relevant patterns from all available information for a family of targets, tools to efficiently apply models for all targets in the family rather than just a small subset, mining tools to extract the relevant information from the computational absorption, distribution, metabolism, excretion and toxicity (ADMET) and targeting data, and tools to allow interactive exploration of the virtual space around a library to facilitate the selection of the library that best suits the needs of the design team.