Measurements and characterization of binding interactions between proteins and low-molecular weight ligands are fundamental for drug discovery. Among the most recognized challenges in characterizing binding interactions are (1) the need to accurately assess a wide span of binding affinities (KD) and (2) accurately rank and characterize low-molecular weight (LMW) ligands based on affinity, mechanism of action, and energetics of interaction.
Measurements and characterization of binding interactions between proteins and low-molecular weight ligands are fundamental for drug discovery. Among the most recognized challenges in characterizing binding interactions are (1) the need to accurately assess a wide span of binding affinities (KD) and (2) accurately rank and characterize low-molecular weight (LMW) ligands based on affinity, mechanism of action, and energetics of interaction.
Reliable interpretation of binding data can be complicated by the presence of inactive protein fraction or inaccurate assessment of protein concentration. Assessment of this data can be further hampered by inherent uncertainty in concentration of compound stocks. This uncertainty results from inaccurate measurement, limited solubility, or potential chemical heterogeneity of the compounds such as presence of enantiomers and isomers.
Isothermal titration calorimetry (ITC) directly measures heat released or absorbed in a binding event, providing means for studying protein-small molecule interactions in solution without the need for labeling or immobilization. A highly sensitive ITC instrument and properly designed experimental conditions make it possible to account for inaccurately assessed protein concentration or inactive protein fraction, and to account for imprecise concentration of compound solutions. From quality control (QC) to assay development and lead optimization, ITC has a role in improving our understanding of biochemical data. Normalizing for protein or compound concentrations and accounting for inactive populations of protein can improve decision-making processes, one example being a more careful assessment of changes in enthalpic contributions to binding, which are critical in many best-in-class drug studies.
This white paper highlights the advantages of the new ITC instruments—MicroCal PEAQ-ITC and MicroCal PEAQ-ITC Automated systems and MicroCal PEAQ-ITC Analysis Software. The software allows analysis of binding interactions complicated by the presence of inactive protein material, incorrect protein concentration, and/or inaccurate concentration or heterogeneity of compound solutions.
ITC is utilized throughout early stage drug discovery, from assay optimization work and secondary screening to late stages involving lead optimization. In addition, ITC can be used in the protein purification QC step to assess batch variations and freeze-thaw stability. In the earlier stage applications, one would typically test a set of compounds including a well-established positive control. The periodic use of a positive control ligand provides means for monitoring protein quality and establishing and confirming active protein concentration throughout a series of ongoing ITC experiments. The MicroCal PEAQ-ITC Analysis Software generates scatter plots of binding parameters for a series of consecutive titrations. Such plots facilitate the detection of possible trends and help to identify potential issues with assay setup and reagent quality.
Figure 1 presents a scatter plot of N values obtained for a series of consecutive ITC titrations. It shows a trend in apparent number of binding sites on the target protein. The gradual decrease in N value over time most probably reflects stability issues of the target protein. Limited protein stability, batch-to-batch variation, and limited freeze-thaw stability of a target protein are common challenges associated with studies of protein-ligand interactions. ITC data is useful to determine whether one lot of protein differs in apparent activity (binding) relative to previous lots and provides potential normalization criteria, which are useful across many assays run throughout the screening/characterization process.
With the concentration of the active protein established and used in the ITC data analysis, any remaining errors in the apparent stoichiometry (i.e. non integer values 1,2 …etc) can be assigned to incorrect concentration determination of the ligand.
For many reasons, it is not uncommon in drug discovery that the concentration of compound solutions is given with significant errors. This inaccuracy directly affects the quality of the binding parameters returned by fitting binding data; most notably in the correct determination of the enthalpic and entropic contributions to binding. In the new MicroCal PEAQ-ITC Analysis Software, it is possible to identify and account for the error in the concentration of the compound to minimize any errors in the thermodynamic measurements.
Figure 2 presents the scatter in the stoichiometry (N) determined by ITC for an interaction of a target protein with a series of LMW hits in a drug discovery and development project. This was conducted by a client using the MicroCal Auto iTC200. The N values ranged from 0.2 to 1.8 and were not correlated with a fraction of inactive protein which was already established using the method described above. In addition, 1:1 binding stoichiometry was expected for the compounds based on X-ray structures of the complexes. It was clear therefore that differences in the apparent stoichiometry were due to incorrect determination of the ligand concentrations.
These errors directly impact on the enthalpy data generated by the ITC making it very difficult to interpret the differences in the thermodynamics of interaction of each ligand to the target protein.
The new MicroCal PEAQ-ITC Analysis Software can account for this by allowing the concentration of the LMW ligand in the syringe to vary while setting stoichiometry value to 1 in the fitting process (this can be performed automatically on many data sets simultaneously). Figure 3 shows the enthalpy data returned with and without this correction factor.
It is clear that the enthalpy data returned is dramatically different when the errors in ligand concentration are factored into the analysis and clearly will have a large impact on the interpretation of the SAR (structure activity relationships) of this lead optimization study.
Another consequence of inaccurate ligand concentrations is that experiments may be difficult to analyze due to the incomplete nature of the binding isotherm and may need repeating. An example of this is shown in Figure 4.
The MicroCal PEAQ-ITC Analysis Software can fit these incomplete binding isotherms by using a constant offset, representing the control heats, in the fitting process. By combining the control offset and varying the ligand concentration in the fitting process, previously ‘unfittable’ or difficult data sets can be analyzed automatically in a non-subjective manner. In many cases this is sufficient for the application and, at least, can be used to establish the correct concentration of the compound stock. In this example, the ligand concentration was initially estimated to be 200 µM but was found to be 126 µM. This error had a large impact on the enthalpy (ΔH) and entropy (-TΔS) which changed by -2.8 kcal/mole and 2.28 kcal/mol, respectively, while the apparent KD changed from 241 nM to 122 nM. These differences highlight the value of the new features in the analysis software and the benefits it will bring to robust data interpretation.
In late-stage drug discovery, a significant number of compounds are synthesized as mixtures of enantiomers or isomers. In many of these cases, the compound mixture can be separated by high-performance liquid chromatography (HPLC) and one or more isomers can be tested individually. However in some cases, particularly those involving mixtures of enantiomers, separation can be difficult and time-consuming, and in these cases the mixtures must be tested directly. Biochemical assays primarily provide information on the tightest binding component in the mixture. In contrast, ITC can provide binding information about both the strong and the weaker binder simultaneously in a single experiment.
The biphasic binding isotherm in Figure 5 represents an example how the ITC data may look for the injection of an enantiomeric ligand mixture into a target protein in the cell. Similarly, biphasic isotherms are observed when high affinity ligands are intentionally mixed with weaker ligands as a form of competition experiment to determine KDs outside the range of direct measurements. This differs from the more typical approach of injecting the tight binding ligand into a mixture of the target and the weak inhibitor but has the advantage that both inhibitors can be resolved simultaneously and without prior knowledge of the interaction parameters of the weak interaction.
To demonstrate the utility of the MicroCal PEAQ-ITC Analysis Software for the analysis of complex binding isotherms we have performed experiments with mixed ligands (ethoxzolamide (EZA), and furosemide (FUR)) in the syringe injected into a target protein (Bovine carbonic anhydrase II, bCAII) in the cell. The data are shown in figure 6 and represent the type of biphasic isotherm data that is expected for an enantiomeric mixture or a ‘syringe’ competition experiment.
These experiments demonstrate the level of resolution that can be achieved using the MicroCal PEAQ-ITC and the potential for quantitative characterization of the binding of mixtures. The fitting of complex ITC data, and in particular the fitting of these competitive experiments, has been made simpler in the MicroCal PEAQ-ITC software than in analysis software previously available on earlier models. While the input of good initial guesses into the fitting software is always a good start, the current software contains more appropriate numerical boundaries increasing the chances of a successful fit and minimizing the risk of the fitting process getting trapped in a local minima. Fitting these data also benefited from combining the Simplex fitting and Marquardt-Levenberg fitting approaches available in the software. More specifically, using the Simplex approach at the start and finalizing using the Marquardt-Levenberg fitting algorithm. The thermodynamic parameters obtained using this approach with the ‘two-sites’ model and ‘ligand in cell’ setting are summarized in Table 1.
bCAII in cell, mM | N1 (sites) | KD1 (nM) | DH1 (kcal/mol) | N2 (sites) | KD2 (nM) | DH1 (kcal/mol) |
---|---|---|---|---|---|---|
5 | 0.5 (4) | 0.8 (100) | -15.1 (1) | 0.6 (7) | 700 (43) | -7 (13) |
10 | 0.49 (3) | 1 (93) | -15.2 (2) | 0.56 (3) | 630 (42) | -7.1 (5) |
The data compare well with the thermodynamic values obtained for these interactions when measured independently (see table 1 legend).
The new MicroCal PEAQ-ITC instrument along with improved signal stability, mixing and signal-to-noise characteristics has data analysis software well-suited for use in the biophysical assay laboratories involved in small-molecule drug discovery.
The analysis is completely automated, minimizing user subjectivity in assessing data quality and the analysis process. Data quality assessment and fitting is performed rapidly allowing for analysis of large data sets of 50 or more experiments in seconds.
Key additional features of the new MicroCal PEAQ-ITC Analysis Software package allow for the determination of the active concentration of the target protein or the ligand concentration or both when the appropriate controls are used. This provides for the more accurate determination of the affinity and thermodynamic allowing for rigorous structure-activity relationship in hit validation and lead optimization programs.
In addition, the new ITC data analysis software has tools that simplify the fitting of complex binding isotherms that may be observed when titrating with enantiomeric mixtures, some competition experiments or with targets with more than one binding sites.