Think about the pill you take when you’re sick or in pain; you just swallow it hoping it doesn’t get stuck in your throat, but there is something amazing about just taking a pill or a shot and that is how the drug finds its target within the molecular jungle of our bodies. Finding the target as well as keeping attached to it for a sufficient period of time to initiate the desired effect, are characteristics of a drug in which affinity plays a vital role.
Binding affinity is the strength of the binding interaction between a molecule and its ligand/binding partner (e.g. drug or inhibitor). It is reported by the famous KD or equilibrium dissociation constant: the smaller the KD value, the greater the binding affinity of the ligand for its target. This biophysical parameter is so important that it forms the foundation of molecular recognition and communication, but it also offers a mechanism for exogenous control of biological systems, for example by drugs.
In the drug development process, ‘affinity’ can be found from beginning to end. In early-stage discovery it is used as the key selection parameter during candidate molecule screening; it is further optimized during the hit-to-lead phase; and it is also something that must be taken into account in terms of money, since it also influences the cost of a biologic: if it binds tighter to its target, less of it potentially needs to be manufactured. But how can affinities be measured? Is there a ‘best practice’?
Choosing a biophysical method
A measure of affinity can be obtained through several biophysical methods that are applied concurrently during hit validation. In general, biophysical assays can give information on binding specificity and stoichiometry, quantitative data on binding thermodynamics and kinetics, and even elucidate atomic details of the protein-ligand interactions.
Choices on what method to use depend on many things, which can go from the assay conditions used to sensitivity and even ease of use. For primary fragment-based screening, the method choice is frequently dictated by logistical and practical considerations such as: equipment and protein availability. Sometimes the strategy is progressing all hits from all the applied biophysical methods; in other occasions the strategy is overlapping hits validated by more than one technique. So it is very common to see more than one method used throughout a screening. Here’s a brief description of the main methods currently used:
Isothermal titration calorimetry (ITC) involves monitoring the heat produced (for an exothermic binding event) or absorbed (for an endothermic binding event) during the binding reaction by titrating a ligand, from an injection syringe, on to a protein solution (or the reverse). It is the most direct method used to measure thermodynamic parameters of binding (KD, ∆H and ∆S), and with variation of temperature,∆Cp. Its main drawbacks are that it requires lots of protein: 6-60 nmol of protein per titration (for example, 0.2-2 mg with a MM of ~30 kDa), has a very low throughput (only 10 samples per day) and it requires high solubility of the titrated component. ITC has an affinity range of 1nM-100mM.
Thermal shift analysis (TSA), also known as differential scanning fluorimetry (DSF) or thermofluor. Its biophysical basis is that proteins are thermodynamically stabilized by ligand binding. In this assay, the temperature of a solution composed of a protein and a dye (such as Sypro® Orange) is incrementally raised, causing proteins to unfold. The dye then binds selectively to hydrophobic patches, which get exposed on the protein during the unfolding process. Thus, during the temperature ramp the fluorescence signal increases proportionally with the degree of unfolding as the dye binds. This change in fluorescence intensity is easily measured in plate-based instruments, such as a qPCR, making it very accessible and possible to use both 96- and 384-well plate formats. However, the degree of thermal stabilization upon fragment binding to a protein may be too small to be measured and contaminants or counter ions present in samples can be sufficient to cause a thermal shift. This means many false-positive and false-negative hits. It requires 80 pmol protein per analysis (40 ml at 2 mM) and has an affinity range of 1 nM-100 mM. It is worth knowing that currently there is a new technology named nanoDSF, that can also determine thermal stability but in a label-free manner, using protein intrinsic fluorescence, thus protein solutions can be analyzed independent of buffer compositions and with a maximal protein concentration range from more than 250 mg/ml down to 5 µg/ml. This allows for the analysis of detergent-solubilized membrane proteins as well as highly concentrated antibody formulations.
Surface plasmon resonance (SPR) is an optical assay. Optical biosensors typically work by generating a measurable change in some characteristic property of light, that is coupled to a process, in this case binding between molecules. SPR is a phenomenon that occurs when plane-polarized light hits a metal film, which is located on the flat surface of a semicircular prism. This light is totally internally reflected and it is under this condition that photons create an electric field on the film that can penetrate and interact with the free electrons in the metal surface. When this happens, the incident light photons are converted (absorbed) into surface plasmons (Quantum theory says that light and electrons have both wave and particle properties; ‘plasmon’ is the particle name for the electron density waves, just as ‘photon’ is the particle name for light waves). At a certain incident angle, known as the resonance angle, plasmon and photon momentum equals, or if you prefer a wave-like description, it is a ‘wave vector coupling’.
For a binding analysis experiment, a ligand is immobilized onto the surface and the sample is injected. Because surface plasmons are sensitive to the properties of the medium, when the analyte concentration at the sensor surface changes (as in the case of a binding event), the momentum of the surface plasmons is affected and the incident light angle at which resonance occurs shifts, which can be monitored.
Because it is a microfluidic system, with continuous detection of the signal, it can provide additional data on binding thermodynamics and kinetics (association rate constant, kon and dissociation rate constant, koff). The main challenges for SPR are retaining full protein functionality after immobilization as well as expertise to set up a high-quality assay and analysis protocol. Signals can also be affected by solvent effects. It requires 15 nmol protein (for example ~0.5 mg with a MM of 30 kDa) for assay development and screening of 2,000 compounds. With an affinity range between 1 nM-500 mM, it can analyze 100 samples per day.
Microscale thermophoresis (MST), monitors fluorescence in an infrared-laser-heated spot. It is an equilibrium-based method that can detect ligand-binding-induced changes in thermophoretic mobility (the motion of protein molecules along a microscopic temperature gradient). This technique is very fast, has the widest affinity range among the existing technologies, 1 pM-1 mM, and requires the least amount of protein. Depending on the instrument, proteins analyzed can be fluorescently labelled or, if they have detectable intrinsic fluorescence, label-free. MST can track molecules in complex solutions such as cell lysates or plasma and it is applicable to solubilized membrane proteins as well.
Nuclear magnetic resonance (NMR) is a spectroscopic method that can also be used to solve structures of protein-ligand complexes for targets with a molecular mass of less than 40 kDa. It is the most robust technique for identifying where a compound binds, in the absence of an X-ray structure of the ligand complex. Its main limitation is the amount of protein required- typically tens of milligrams for screening a 1000-member fragment family. It can cover a range of affinity of 100 nM to 1 or 10 mM, respectively, depending on the type of NMR (protein-observed NMR or ligand-observed NMR, respectively).
X-ray crystallography is the most powerful, robust and routine method to obtain a detailed atomic picture of a compound binding to its target, though it does not provide quantitative affinity information. It is considered the ultimate validation of the binding event. Its main limitations are the requirements for large infrastructures (synchrotrons) and the need for quality crystals.
All in all, do not expect different techniques to result in exactly the same hits, which doesn’t mean there is something wrong with some of them. Of course several differences exist in assay protocols, sensitivity to physical properties such as solubility, aggregation potential, stability and ability to interfere with the assay, and also differences in the conditions used, such as the effect of buffer conditions. Finally, as Jean-Paul Renaud et al. pointed out, it is important to keep in mind that, “… success in fragment binding screening depends strongly on users’ experience and expertise, specifically the ability to establish robust assays, stringent data interpretation protocols and well-considered screening cascade”.
This post was based in two recent and very interesting articles in the field. I strongly recommend these reads for more detailed information.
Biophysics in drug discovery: impact, challenges and opportunities. Jean-Paul Renaud et al. http://www.nature.com/nrd/journal/v15/n10/abs/nrd.2016.123.html
How to study protein-protein interactions. Marjetka Podobnik et al. https://journals.matheo.si/index.php/ACSi/article/view/2419