Pharmacogenomics of GPCR Drug Targets — Data-Mining Experiment, rather than a Genome-Wide Association Study

COMMENT: Yes, Dan –– This is a very key point in precision medicine.

Many of the examples that you mention are likely to be modulated by the immune system, with completely unanticipated immune reactions against the whole drug molecule. If it is a large molecular-weight drug (such as abacavir), it can bind directly to specific HLA types [Human leukocyte antigen genes (HLA) encoding the major histocompatibility complex (MHC) proteins in humans; these cell-surface proteins are responsible for regulation of the immune system) –– located on the surface of antigen-presenting cells for T-cell activation. Alternatively, if it is a small-molecular weight drug, the antibody that is produced can be directed against a hapten (a small molecule that, when combined with a larger carrier, e.g. a protein, can elicit the production of antibodies that bind specifically to it), which means the drug binds covalently with a peptide.

At the present time, there are about 40 different HLA alleles that have been associated with increased risk of ADRs caused by different drugs. To a smaller degree, I think adverse drug reactions can be determined by rare genetic variants [abstract of recent article (Hum Genomics 2o18; 12: 26) pasted below].

Integrating rare genetic variants into pharmacogenetic drug response predictions

Ingelman-Sundberg M, Mkrtchian S, Zhou Y, Lauschke VM.

BACKGROUND: Variability in genes implicated in drug pharmacokinetics or drug response can modulate treatment efficacy or predispose to adverse drug reactions. Besides common genetic polymorphisms, recent sequencing projects revealed a plethora of rare genetic variants in genes encoding proteins involved in drug metabolism, transport, and response.

RESULTS: To understand the global importance of rare pharmacogenetic gene variants, we mapped the variability in 208 pharmacogenes by analyzing exome sequencing data from 60,706 unrelated individuals and estimated the importance of rare and common genetic variants using a computational prediction framework optimized for pharmacogenetic assessments. Our analyses reveal that rare pharmacogenetic variants were strongly enriched in mutations predicted to cause functional alterations. For more than half of the pharmacogenes, rare variants account for the entire genetic variability. Each individual harbored on average a total of 40.6 putatively functional variants, rare variants accounting for 10.8% of these. Overall, the contribution of rare variants was found to be highly gene- and drug-specific. Using warfarin, simvastatin, voriconazole, olanzapine, and irinotecan as examples, we conclude that rare genetic variants likely account for a substantial part of the unexplained inter-individual differences in drug metabolism phenotypes.

CONCLUSIONS: Combined, our data reveal high gene and drug specificity in the contributions of rare variants. We provide a proof-of-concept on how this information can be utilized to pinpoint genes for which sequencing-based genotyping can add important information to predict drug response, which provides useful information for the design of clinical trials in drug development and the personalization of pharmacological treatment.

COMMENT This topic falls precisely under the heading of “gene-environment (G x E) interactions.” Or better yet, in the case of this paper, “drug-genome interactions.” This is among the most intriguing mysteries in all of clinical pharmacology: a drug is given to patient A and it works as expected (efficacy), but given to patient B, the drug causes an adverse drug reaction (ADR), and given to patient C, there is no beneficial or toxic effect (therapeutic failure). How does a small-molecular-weight drug — given to some patients, but not the majority of patients in any population — cause an ADR that is often indistinguishable from a complex disease?

For example, sitagliptin is approved by the FDA to treat type-2 diabetes; yet, a small subset taking the recommended prescribed dose develops acute pancreatitis. Hydroxychloroquine — given to treat malaria, lupus erythematosus, or rheumatoid arthritis — can also lead to acute pancreatitis in some patients. In a subpopulation of patients receiving many psychotropic drugs (e.g. valproic acid), undesirable weight gain can occur as a dose-independent ADR; in another small subset, hepatic steatosis (fatty liver) has been found. In a small subpopulation of patients taking bisphosphonates for osteoporosis, increased risk of esophageal and gastric cancer has been reported in a number of studies; however, a large meta-analysis of this association has found no significantly increased risk [Wright et al., BMJ Open 2015; 5: e007133].


COMMENT: Hi Dan, The most remarkable finding in this Cell paper, I think, is the shift of the mu-opioid receptor, by a rare mutation, to respond to naloxone as an agonist instead of antagonist. However, more wet-lab experiments are needed to verify some of these key findings.


The [attached] article is a bombshell report and, to our knowledge, represents the first study of its kind. Rather than a genome-wide association study (GWAS), authors performed an avante gard data-mining in silico approach — to search for DNA variants in or near each of the 108 G-protein-coupled receptor genes (GPCRs) known to exist in the human genome. In the field of pharmacology and drug response, these 108 genes are the known targets of 475 prescription drugs that have been approved by the U.S. Food and Drug Administration (FDA). These 475 drugs, which comprise ~34% of all prescription drugs, account for a global sales volume of >US$180 billion annually..!!

Each of the genomes of almost 68,500 individuals was separately investigated for missense variants in and near each of the GPCR genes. Then the authors searched the literature for the clinical associations with altered drug response in these individuals. To estimate the de novo missense mutation rate within these GPCR genes, authors in addition identified novel mutations from >1,700 control trios (having no reported pathological conditions) –– which were compiled from ten different studies registered in the “denovo-database,” an intriguing collection of germline de novo variants (

To demonstrate proof-of-principle, authors then experimentally showed that certain variants of the mu-opioid and cholecystokinin receptors resulted in altered drug responses and/or idiosyncratic dose-independent adverse drug reactions. These amazing results — on just two of the 108 GPCR genes — underscore the need to characterize DNA variants among all 108 of the GPCR genes. Authors suggest that “the ultimate results of this novel type of in silico study might enhance prescription precision, improve patients’ quality-of-life, and remove some of the economic and societal burden caused by variability in drug response.”

We anticipate that such “dry-lab” data-mining studies, i.e. just sitting in front of a computer and searching databases online — such as this landmark publication [attached] — are likely to become a major new way to approach pharmacogenomics research in the near future..!! J


Cell Jan 2o18; 172: 41–54

This entry was posted in Center for Environmental Genetics. Bookmark the permalink.