A comprehensive map of molecular drug targets

Abstract

The success of mechanism-based drug discovery depends on the definition of the drug target. This definition becomes even more important as we try to link drug response to genetic variation, understand stratified clinical efficacy and safety, rationalize the differences between drugs in the same therapeutic class and predict drug utility in patient subgroups. However, drug targets are often poorly defined in the literature, both for launched drugs and for potential therapeutic agents in discovery and development. Here, we present an updated comprehensive map of molecular targets of approved drugs. We curate a total of 893 human and pathogen-derived biomolecules through which 1,578 US FDA-approved drugs act. These biomolecules include 667 human-genome-derived proteins targeted by drugs for human disease. Analysis of these drug targets indicates the continued dominance of privileged target families across disease areas, but also the growth of novel first-in-class mechanisms, particularly in oncology. We explore the relationships between bioactivity class and clinical success, as well as the presence of orthologues between human and animal models and between pathogen and human genomes. Through the collaboration of three independent teams, we highlight some of the ongoing challenges in accurately defining the targets of molecular therapeutics and present conventions for deconvoluting the complexities of molecular pharmacology and drug efficacy.

Figure 1

Major protein families as drug targets. (a) Distribution of human drug targets by gene family. (b) distribution by the fraction of drugs targeting those families; the historical dominance of four families is clear. (c) Clinical success of privileged protein family classes. Distribution of non-approved compounds in ChEMBL 20 (extracted from the medicinal chemistry literature, with bioactivity tested against human protein targets) per family class, and distribution of approved drugs (small molecules and biologics) per human protein family class. 7TM, seven transmembrane family; GPCR, G protein-coupled receptor; LGIC, ligand-gated ion channel; NTPase, nucleoside triphosphatase; VGIC, voltage-gated ion channel.

Figure 2

Innovation patterns in therapeutic areas. Each node in the inner ring corresponds to a drug represented by its ATC code(s). The inner ring corresponds to the level 1 of the ATC code (see Table 2) scaled to the number of drugs in that category. The outer ring represents the level 3 of the ATC code. Each of the subsequent histograms illustrates the number of drugs (small molecules and biologics) distributed per year of first approval per level 3 of the ATC code. Max histogram scale: 100. The approval year refers to the first known worldwide approval date, if available, otherwise the first FDA approval date.

Figure 3

Innovation patterns in privileged protein classes. Histogram depicting the number of drugs (small molecules and biologics) that modulate the four privileged families, distributed per year of first approval. On top of each bar, the total number of approved drugs is shown, together with the number and percentage of drugs approved since 2011 in respect to the total drugs modulating these four families. A spreadsheet view of this data is provided in supplementary information S6 (table). 7TM1: G-protein coupled receptor 1 family; Ion channel: Voltage-gated ion channel and Ligand-gated ion channel. Drugs without an ATC code (U – Unclassified) were excluded from this analysis.

Figure 4

Promiscuity of privileged protein family classes. Each node in the outer ring corresponds to a drug represented by its ATC code(s). The outer ring corresponds to the level 1 of the ATC code (see Table 2) scaled to the number of drugs in that category. The inner ring represents the level 2 of the ATC code. A node is connected to another when two drugs have an efficacy target that belongs to the same target class. (a) Footprint of privileged family classes modulated by organic small-molecule drugs across disease. (b) Footprint of privileged family classes modulated by biologic drugs across disease. G-protein coupled receptor 1 family (blue); Voltage-gated ion channel (orange); Ligand-gated ion channel (green); Kinase (black). Drugs without an ATC code (U – Unclassified) were excluded from this analysis.

Figure 5

Protein efficacy targets availability across several model organisms. (a) Each node in the outer ring corresponds to a drug represented by its ATC code(s). The outer ring corresponds to the level 1 of the ATC code (see Table 2), scaled to the number of drugs in that category. The next ring represents the level 4 of the ATC code. Each of the subsequent rings represents a different species, as indicated in the legend, and each section of the ring is coloured according with the presence or absence of orthologues of the efficacy targets of the drugs in that ATC level 4 category. The dark blue sections indicate the species of the protein efficacy targets. (b) An expanded portion of section J of the chart.

Figure 6

Overlap of cancer drug targets with cancer drivers. We grouped the cancer drugs into the three categories: broadly cytotoxic agents such as platinum complexes and DNA intercalating agents; cytotoxic agents that act through a protein, such as tubulin inhibitors, that do not have biological selectivity; and targeted agents that act through clear protein function-modulating mechanisms such as kinase inhibitors and nuclear hormone receptor antagonists. When we compared the targets of agents in the third group to a consensus reference list on cancer driver genes we observe only a small overlap between cancer drivers and current cancer drug targets.