Well-characterised, selective small molecules -‘chemical probes’ - are essential tools for target validation during drug development and in basic biological research.1 Criteria for small molecule modulators to qualify as chemical probes have been established by chemical biologists and are widely accepted in the community.2 These include target-related criteria for potency, selectivity, and proof for target engagement in addition to the suitability of the chemical matter itself.1 By creating these quality criteria, chemical probes became important and generally recognized tools aiding the scientific community and accelerating drug discovery. Inspired by this approach, our goal is the standardization of quality criteria within the drug candidate evaluation process. Biochemical and cellular assays often rely on displacement assays using fluorescence-labelled molecules, called tracers (sometimes referred to as fluorescent probes - not to be confused with chemical probes themselves or medical radiotracers).3-5 Tracers comprise a chemical moiety that binds to the protein of interest (POI), such as a small molecule, DNA, RNA, and peptides, a chemical linker, and a reporter label, typically a fluorescent dye.6,7 To avoid interference of the linker with the binding of the molecule to the POI, the choice of the right exit vector, a solvent exposed attachment point of the linker to the molecule, is important (Figure 1 a).
Tracers are used in cellular target engagement assays (in cellulo) such as time-resolved Förster resonance energy transfer (TR-FRET)6 or bioluminescence resonance energy transfer (BRET)4 assays or biochemical in vitro studies, which can be BRET, TR-FRET- based or comprise fluorescence polarization (FP)8. In particular, NanoBRET, a method frequently applied in kinase live-cell target engagement assays, critically relies on the use of suitable tracer molecules. This method validates the binding of a small molecule such as an inhibitor to its cognate target in the cell. It is also suitable for assessing cellular selectivity using a single tracer.9 Owing to the stringent distance constraints of BRET and the localization-specificity of the BRET donor, tracers do not need to be specific for the protein of interest. On the contrary, unselective BRET tracers are ideal as they allow assay development for multiple targets. Using this principle, we successfully included 206 (as of Feb. 2024) validated kinase BRET assays using tracer K10 (T000008).
Due to the importance of quantifying protein-ligand interactions, a large number of tracers are reported in the literature. However, scientists face several problems in establishing displacement assays for their respective target: I) finding established tracers in the literature using search engines is difficult, as much of the required information is buried in supplemental methods; II) reproducibility of the reported assays is often problematic due to insufficient validation of the tracer or unfavourable assay parameters; III) the availability of the tracer is often unknown. We created a database for fluorescent tracer molecules named tracerDB to address these problems. It has been developed and standardized to provide design and application guidance based on strict performance criteria. For each tracer-based assay, chemical structure or commercial availability is provided, as well as the assay parameters and a reference. tracerDB allows to search for the protein of interest or the tracer enabling fast assessment of available assay options for a specific target. Within the first four months (as of Feb. 2024), 38 tracers, targeting 308 different proteins in 464 experimentally validated assays were reviewed and uploaded.
Scientists worldwide can submit their tracer data for review and subsequent inclusion in the tracerDB. The submission of tracer data must include all necessary information (no physical molecules) required to judge the quality and reproducibility of a tracer-based assay. First, general information such as the molecular structure (SMILES, fluorophore characteristics and trivial name) are required for the creation of a tracer page (Figure 1 b). All target proteins bound by the tracer need to be listed in UniProt10. Experimental data from the tracer validation process against a certain target have to be uploaded, consisting of information on a recommended concentration, Z’ value of the assay, and the observed assay window. Additionally, the measured data for tracer titration and compound displacement is required. To facilitate the upload and review process, a submission file is available for download at the footer of the website (https://www.tracerdb.org/). After insertion of all required information, the sheet is sent to [email protected], for final approval and upload. Additionally, tracer IDs can be assigned prior publication, allowing a direct link to the database (in analogy to PDB).
The interaction network between tracer molecules and their respective targets can be modelled as a many-to-many relationship. As a result, the underlying database structure consists of three entity sets: the tracer, the protein, and their interaction (Figure 1b). To ensure a user-friendly submission of data and standardize the presentation, all molecular representations and calculations are created and executed on the server side. We chose Django11 as a python-based web framework together with a MySQL database to enable high-frequency read operations.
In addition to the information on the crowdsourced tracers, we have also included general information on tracer molecules and illustrations of different assay systems on the 'about' page (https://www.tracerdb.org/about/). Here, we describe the quality control criteria and how to calculate the respective values. In order to further increase the reproducibility of the described assays, each assay is classified according to its parameters into robust, expert and unsuitable assays with exemplary data for clarification (Figure 2). These assay levels are represented by a traffic light icon for each registered assay. In addition, we have included a methods section describing the different assays used to collect the submitted data (https://www.tracerdb.org/methods/). This is supported by an illustration and key references.
As an additional purpose, the chimeric structure of tracer molecules provides an excellent basis for the development of degraders including proteolysis targeting chimeras (PROTACs). It allows for the replacement of the fluorescent dye with a binder to an E3 ubiquitin ligase. A PROTAC leads to the formation of a ternary complex (POI, E3 ligase and PROTAC), ubiquitination and subsequent proteasomal degradation of the target protein.12 The validated exit vector in the tracer, assessed tolerance of the linker, and -in case of NanoBRET- limited information about cellular permeability of the parent ligand, providing an ideal starting point for PROTAC synthesis. Furthermore, a functional NanoBRET assay indicates which protein terminus is suitable for fusion proteins.
TracerDB therefore marks the first resource for drug-screening scientists as well as the chemical biology community, that gathers detailed, reviewed and high-quality information on tracer-based assays and their applications.
