In silico Prioritization of Transporter-Drug Relationships From Drug Sensitivity Screens

Abstract

The interplay between drugs and cell metabolism is a key factor in determining both compound potency and toxicity. In particular, how and to what extent transmembrane transporters affect drug uptake and disposition is currently only partially understood. Most transporter proteins belong to two protein families: the ATP-Binding Cassette (ABC) transporter family, whose members are often involved in xenobiotic efflux and drug resistance, and the large and heterogeneous family of solute carriers (SLCs). We recently argued that SLCs are collectively a rather neglected gene group, with most of its members still poorly characterized, and thus likely to include many yet-to-be-discovered associations with drugs. We searched publicly available resources and literature to define the currently known set of drugs transported by ABCs or SLCs, which involved ∼500 drugs and more than 100 transporters. In order to extend this set, we then mined the largest publicly available pharmacogenomics dataset, which involves approximately 1,000 molecularly annotated cancer cell lines and their response to 265 anti-cancer compounds, and used regularized linear regression models (Elastic Net, LASSO) to predict drug responses based on SLC and ABC data (expression levels, SNVs, CNVs). The most predictive models included both known and previously unidentified associations between drugs and transporters. To our knowledge, this represents the first application of regularized linear regression to this set of genes, providing an extensive prioritization of potentially pharmacologically interesting interactions.

Keywords: ABC transporters; drug sensitivity and resistance; drug transport; regularized linear regression; solute carriers.

FIGURE 1

(A) Network visualization of known SLC/ABC-mediated drug transport cases. Circular nodes represent SLC and ABC transporters, and squares represent chemical compounds. Drugs approved by the FDA (Food and Drug Administration) are displayed with thicker gray borders. Edges connect transporters to compounds and their thickness indicates the number of sources supporting each connection (see section “Materials and Methods”). Size indicates node degree (number of edges incident to the node). Relevant transporter families are color coded. (B) Transporter degree distribution. The inlet bar chart displays the transporters connected to at least 15 compounds. Bar colors correspond to transporter families in (A). (C) Same as (B) for drugs.

FIGURE 2

(A) Number of transporters (SLCs and ABCs) expressed across cell lines in GDSC dataset. A cut-off of 3.5 in RMA units is set to consider a gene as expressed (∼73% genes expressed). The red line indicates the median number of transporters expressed per cell line. The inlet lists the 11 cell lines expressing the highest number of transporters, indicated between parentheses. (B) Number of cell lines expressing each of the transporters. The color bars and inlets indicate sets of transporters showing more common or specific expression across the panel. (C) Median expression vs. maximum expression for each transporter across the cell line panel. Color indicates the tissue of origin of the cell line presenting the maximum expression for the transporter. (D) Transporter Z-scores of the average expression values within each tissue. Tissue names with number of cell lines between parenthesis are indicated on the x-axis. Transporters are ordered alphabetically by family and name.

FIGURE 3

(A) Comparison of Elastic Net regression performance (Concordance Index) using different input data: gene expression, genomics (CNVs and SNVs) and a combination of both. (B) CI value distribution using gene expression as input. Red bars indicate drugs with a median CI higher than 0.65, which were selected for subsequent analysis. (C) Elastic Net results for drugs with the highest CI values. The top five associations are shown for each compound. Purple indicates associations linked to sensitivity (higher expression value confers sensitivity to the compound), and orange indicates resistance. (E) Network representation of three transporters appearing as “hubs” (e.g., connected to several different compounds) in the results, including the well-known multidrug resistance protein ABCB1. (D) Same as (E) for MEK inhibitors, which show a similar association pattern.