What are tracers?

The measurement of protein-ligand interactions is a commonly used technique during drug development or biophysical protein characterisation campaigns. The strength of the interaction between a protein and its ligand (peptide, small molecule, etc.) can be determined using direct biophysical binding assays such as isothermal titration calorimetry (ITC) or surface plasmon resonance (SPR). These direct binding assays provide precise protein-ligand binding affinities characterised by a dissociation constant (K_D) and are state-of-the-art methods for such determinations. In addition to highly precise K_D determination, ITC measurements provide information about the stoichiometry and thermodynamic parameters of a binding event whereas SPR allows for the measurement of association (k_on) and dissociation (k_off) kinetics.

While these methods provide high quality affinity measurements, they are limited by the need for large amounts of purified protein (ITC) or the need for protein immobilisation (SPR) with limited overall throughput. To compensate for this, indirect binding assays can be used to rapidly determine the affinity of intracellular and purified proteins. This makes them a versatile and cost-effective alternative for (high-throughput) ligand screening and evaluation.

Indirect binding assays require a reporter molecule that has measurable binding and displacement. Therefore, these reporter molecules contain a ligand to the target (which may be a peptide, DNA, small molecule), frequently a linker moiety, and a detection reagent, which enables the measurement of such reporters. In case of classical tracers, this detection is carried out through the measurement of a dye which is linked to the ligand. These tracers can be used to determine the affinity of unlabelled ligands in assays such as fluorescence polarization (FP, in vitro), NanoBRET (in vitro and in cellulo) and Time-Resolved Fluorescence Energy Transfer (TR-FRET, in vitro and in cellulo), which are discussed in detail in the tracerDB methods section.

For the establishment of such assays, a tracer, binding to the target of interest is required. Titration of the tracer binding to the protein yields the tracer K_D (K_D in vitro, K_D,app in cellulo). This type of experiment is often referred to as tracer titration, while in case of FP, the protein needs to be titrated to a fixed tracer concentration¹. Using these measurements, the determination of affinities of the dye coupled ligand (tracer) is possible, but caution is advised for transferring this affinity to the unlabelled ligand. Due to the modification of the ligand itself (through the linkage of a dye) and the possibility of unspecific binding to the target, the determined affinity may not necessarily correlate with the affinity of the unlabelled ligand.

To determine the affinity of the unlabelled ligand and to ensure a working assay system, displacement assays are carried out. For displacement assays, tracers are used at a fixed concentration and are displaced from their respective targets by titration of unlabelled ligands that bind to the same binding site. In such an experiment, the dissociation of the tracer (caused by binding of a ligand) is measured, optimally yielding a dose-dependent IC₅₀ (in vitro) or EC₅₀ (in cellulo). Based on the determined tracer K_D and a ligand IC₅₀/EC₅₀, the inhibitory constant (K_i) of the unlabelled ligand can be calculated using the Cheng-Prusoff equation² (or the Nikolovska-Coleska³ equation for fluorescence polarization assays).

Due to the necessity of both measurements and for the exclusion of unspecific binding events, both measurements (tracer titration and titration of unlabelled ligand to a fixed tracer concentration) are required for every assay within the database and are displayed and downloadable for the respective tracer-target pair.

Tracer criteria:

Since results from displacement assays heavily rely on the quality of the utilized tracers, thorough characterization of tracer molecules is required. To ensure high reproducibility, we compiled criteria for tracers. Generally, two parameters can be used to judge the quality of assay data obtained using the tracer molecule. The assay window describes the fold-change of tracer-bound signal (present at the recommended concentration) compared to the concentration of the free protein. If (in case of a NanoBRET assay) the bioluminescence resonance energy transfer (BRET) value is three times higher compared to a control without tracer, the assay window equals 3. Generally, an assay window of ≥ 2 is desired to achieve a high quality. We therefore recommend this for a robust assay system. However, lower assay windows (1.6-2) can be used, allowing for targets that don't have high quality binders. We therefore include these tracer molecules in the database but classify them as “expert assays”. Using these assays requires precise pipetting or liquid handling and they can provide usable results if performed by trained scientific staff.

As second parameter, the Z’ value should be determined⁴. The Z’ value takes the means and standard deviations (SD) of the 100 % tracer-bound (upper limit) and 0 % tracer bound controls (lower limit) into account. It is a sample statistic that quantifies how well the populations of both controls are separated. Hence, a good Z’ value is obtained if the SD are low and the mean values differ significantly. Generally, Z’ values of 0.5–1 are considered as excellent. As for the assay window, Z’ < 0.5 may result in usable binding data but assays may require a larger number of replicates for high quality data and may not be suitable for high throughput. Therefore, these assays are marked as “expert assay”. Since the plate format and therefore the assay volume may influence the quality of the assay (especially for downscaling), icons indicate the plate type, in which the presented assay was validated. If the desired assay is run in a different plate format, it is recommended to re-evaluate the quality criteria.

The use of too high protein/tracer concentrations can lead to apparently weaker affinities due to tracer competition. To avoid this, one has to ensure that the assay was measured in the “binding regime” according to Jarmoskaite et al⁵. The “binding regime” requires an assay where the constant component is well below the K_D (<< K_D). This is the case for the NanoBRET system with the low concentration of NanoLuciferase-labelled protein expressed by the cells, the low protein concentrations in TR-FRET or the low tracer concentrations in FP assays. The use of concentrations that are too high will result in an assay being in the “titration regime”, a system that is likely to produce artefacts. Additionally, it is recommended to ensure an equilibrated system to avoid similar artefacts. Therefore, before developing tracers and setting up binding assays, we recommend that the system is tested to be in the “binding regime” and equilibrated as described by the authors⁵.

As a final criterion, the assay can be tested to be in agreement with the Cheng-Prusoff equation. Generally, the equation describes the relationship between the mean inhibitory concentration (IC₅₀) or the mean effective concentration (EC₅₀) to give the half-maximal response and the K_i. This is important because the IC₅₀/EC₅₀ value can vary at different tracer concentrations used in an assay, whereas the K_i value marks a constant and therefore does not change. To agree with the Cheng-Prusoff equation the absence of cooperativity is crucial. Specifically, cooperativity is the effect in which binding of a first ligand favours binding of a second ligand. Therefore, if an assay does not agree with the Cheng-Prusoff equation, IC₅₀/EC₅₀ values cannot be reliably converted to K_i values.

Further information can be found in the tracerDB publication⁶.

Criteria for a robust assay:

• Assay window ≥ 2
• Z’ value 0.5–1
• Agreement/disagreement with Cheng-Prusoff equation
• Competition data available

 

Criteria for an expert assay:

• Assay window ≥ 1.6
• Z’ value 0.5–1
• Agreement/disagreement with Cheng-Prusoff equation
• Competition data available

 

Promiscuous vs. specific tracers for target engagement studies:

Some of the reported tracers contain highly promiscuous target ligands as opposed to a protein-specific binder. This approach is possible since either isolated systems (FP and TR-FRET) or proximity-based methods (NanoBRET and TR-FRET) are used (discussed in the respective methods section). The latter only allows for the detection of tracer in proximity to the NLuc- or Lanthanide-tagged protein and therefore disregards the unbound fraction which might interact with additional, untagged, targets in case of cellular systems. For cell-based measurements and therefore a system with a number of additional binding partners, the ratio of culture medium to cell volume is ~ 10⁴ and together with the given cell permeability, we can assume a system with [tracer]_total ≈ [tracer]_free^5,7. To test this hypothesis, a comparison of a highly promiscuous tracer (T000008) and a notably more selective tracer based on Palbociclib are compared below in a NanoBRET assay for CDK4. Ensuring an assay which is measured in binding regime and in agreement with the Cheng-Prusoff equation, the selective tracer displayed approx. 6-fold higher affinity towards CDK4 and was ultimately used in 8-fold higher concentration. Nevertheless, both displacement curves using Palbociclib were found to be in high agreement, proving the promiscuous tracer approach suitable for tracer development.

 

Further use of the database:

Even if tracers by themselves are not of interest, the tracerDB indirectly provides important information for the development of other bivalent small molecules including PROteolysis TArgeting Chimeras (PROTACs). The generation of bivalent compounds always harbours the risk of losing affinity towards the target, either through steric hindrance of the tracer attachment point or influences by the linker. All tracers in the tracerDB are validated binders towards their targets and their affinity is provided. Thus, the attachment point (exit vector) and linker tolerance are already validated together with membrane penetration ability in case of NanoBRET assays for intracellular proteins. Therefore, tracers can be directly converted into other bivalent small molecules like PROTACs by exchange of the detection reagent moiety. Such an approach was demonstrated by Ichikawa et al. by functionalizing Palbociclib-FITC (T000036) in analogy to degraders⁸.

References:

1. ¹ Owens, Dominic DG, et al. “A chemical probe to modulate human GID4 Pro/N-degron interactions.” bioRxiv (2023): 2023-01.

2. ² Cheng, Prusoff. “Relationship between the inhibition constant (KI) and the concentration of inhibitor which causes 50 percent inhibition (IC50) on an enzymatic reaction.” Biochem. Pharmacol 22 (1973): 3099-3108.

3. ³ Nikolovska-Coleska, Zaneta, et al. “Development and optimization of a binding assay for the XIAP BIR3 domain using fluorescence polarization.” Analytical biochemistry 332.2 (2004): 261-273.

4. ⁴ Zhang, Ji-Hu, Thomas DY Chung, and Kevin R. Oldenburg. “A simple statistical parameter for use in evaluation and validation of high throughput screening assays.” Journal of biomolecular screening 4.2 (1999): 67-73.

5. ⁵ Jarmoskaite, Inga, et al. “How to measure and evaluate binding affinities.” Elife 9 (2020): e57264.

6. ⁶ Dopfer, J., Vasta, J.D., Müller, S., Knapp, S., Robers, M.B., Schwalm, M.P. (2024). “tracerDB: A crowdsourced fluorescent tracer database for target engagement analysis.” available at Research Square. https://doi.org/10.21203/rs.3.rs-3967452/v1.

7. ⁷ Knight, Zachary A., and Kevan M. Shokat. “Features of selective kinase inhibitors.” Chemistry & biology 12.6 (2005): 621-637.

8. ⁸ Ichikawa, Saki, et al. “The Cyclimids: Degron-inspired cereblon binders for targeted protein degradation.” Cell Chemical Biology (2023).