560 Direct measurement of protein binding energetics by isothermal titration calorimetry Stephanie Leavitt and Ernesto Freire* Of all the techniques that are currently available to measure The heat after each injection is therefore obtained by binding, isothermal titration calorimetry is the only one calculating the area under each peak. Because the capable of measuring not only the magnitude of the binding amount of uncomplexed protein available progressively affinity but also the magnitude of the two thermodynamic decreases after each successive injection, the magnitude terms that define the binding affinity: the enthalpy (∆H) of the peaks becomes progressively smaller until com- and entropy (∆S) changes. Recent advances in plete saturation is achieved. Once this situation is instrumentation have facilitated the development of reached, subsequent injections produce similar peaks experimental designs that permit the direct measurement of corresponding to dilution or mechanical effects that need arbitrarily high binding affinities, the coupling of binding to to be subtracted from all the injection peaks before protonation/deprotonation processes and the analysis of analysis. The corrected area under each peak is given by binding thermodynamics in terms of structural parameters. Equation 1, which is used to analyze the data. The quan- ∆ Because isothermal titration calorimetry has the capability to tity Li is the difference between the concentration of measure different energetic contributions to the binding bound ligand in the ith and (i–1)th injections, and its affinity, it provides a unique bridge between computational functional form depends on the specific binding model. and experimental analysis. As such, it is increasingly For the simplest case, in which the protein has one becoming an essential tool in molecular design. binding site, Equation 1 becomes: K [L] K [L] − (2) Addresses q = v × ∆H ×[P]× a i − a i 1 i + + Department of Biology and Biocalorimetry Center, 1 Ka[L]i 1 Ka[L]i−1 The Johns Hopkins University, Baltimore, MD 21218, USA *e-mail: [email protected] where Ka is the binding constant and [L] is the concentra- Current Opinion in Structural Biology 2001, 11:560–566 tion of free ligand. As the known experimental quantity is the total ligand concentration, rather than the free ligand 0959-440X/01/$ — see front matter © 2001 Elsevier Science Ltd. All rights reserved. concentration, Equation 2 needs to be rewritten in terms of the total ligand concentration. The solution to this and Abbreviations other more complicated binding models in terms of the ITC isothermal titration calorimetry PTP-1B total ligand concentration has been published in [1,2]. protein tyrosine phosphatase 1B ∆ ∆ Analysis of the data yields H and G=–RTlnKa. The entropy change is obtained by using the standard thermo- Introduction dynamic expression ∆G=∆H–T∆S. By repeating a Isothermal titration calorimetry (ITC) measures directly titration at different temperatures, it is also possible to ∆ the energy associated with a chemical reaction triggered determine the change in heat capacity ( Cp) associated by the mixing of two components. A typical ITC experi- with the binding reaction ment is carried out by the stepwise addition of one of the µ ∂∆H reactants (~10 L per injection) into the reaction cell ∆C = p ∂ (~1mL) containing the other reactant. A typical experi- T . ment is shown in Figure 1. The chemical reaction created by each injection either releases or absorbs a cer- During the past few years, significant advances in ITC tain amount of heat (qi) proportional to the amount of instrumentation, data analysis and the structural interpre- ligand that binds to the protein in a particular injection tation of binding thermodynamic data have taken place. ×∆ ∆ (v Li) and the characteristic binding enthalpy ( H) Together, these developments have permitted the imple- for the reaction: mentation of accurate experimental protocols aimed at measuring the binding energetics of protein–ligand and = × ∆ × ∆ qi v H Li (1) protein–protein interactions, and at dissecting ∆G into the fundamental thermodynamic components: ∆H, ∆S ∆ ∆ where v is the volume of the reaction cell and Li is the and Cp. Consequently, the range of application of ITC increase in the concentration of bound ligand after the ith has been extended considerably, including systems that injection. As modern ITC instruments operate on the could not be studied by ITC before. In this review, we heat compensation principle, the instrumental response will discuss the application of ITC to high-affinity binding (measured signal) is the amount of power (microcalories processes, processes coupled to protonation/deprotonation per second) necessary to maintain constant the temperature reactions and the development of quantitative structure/ difference between the reaction and reference cells. thermodynamic correlations. Isothermal titration calorimetry Leavitt and Freire 561 Figure 1 A typical ITC experiment. The experiment shown corresponds to the titration of a phosphotyrosine 0.5 peptide (TEGQpYQPQPA) with the SH2 domain of Lck (S Leavitt, E Freire, unpublished data). The experiment was performed in 10 mM 0.0 Pipes, pH 7.5 at 15°C. [SH2] = 81 µM and [TEGQpYQPQPA] = 0.4 mM. Analysis of the data, as described in the text, yields a binding affinity of 5.8 × 106 M–1 and a ∆H of –0.5 –13.5 kcal/mol. The inset illustrates the configuration of an ITC reaction cell. The cell volume is 1.4 mL and is filled with the protein –1.0 solution (red). The injection syringe, which cal/sec also stirs the solution to assure proper mixing, µ is filled with the ligand solution (green). At specified time intervals, a small volume –1.5 (typically 10 µL) of the ligand solution is injected into the cell, giving rise to the characteristic titration heat effects. Once the protein is saturated, the residual heat effects –2.0 originate from dilution of the peptide and also from mechanical effects associated with the injection. These effects need to be subtracted –2.5 before thermodynamic analysis. In an ITC 0 20 40 60 80 100 120 140 160 180 200 220 experiment, the quantity measured and Time (min) displayed on the y-axis is the time dependence of the electric power (µcal/sec) Current Opinion in Structural Biology necessary to maintain constant the temperature difference between the reaction and reference cells after each injection of reactant. The area under each peak is the heat (microcalories) associated with the process. Binding reactions with arbitrarily high to a weaker inhibitor. Sigurskjold [3••] recently presented binding affinities a rigorous protocol for the analysis of ligand competition In the past, one of the most significant limitations of ITC experiments by displacement ITC. This approach was the absence of protocols aimed at characterizing the requires three titrations: first, a titration with the weak thermodynamics of binding reactions with very high inhibitor in order to characterize its binding thermody- association constants. This limitation is especially severe namics; second, a titration with the high-affinity in drug design, in which lead compounds, usually with inhibitor to measure its binding enthalpy; and, third the micromolar affinities, are optimized to nanomolar or even displacement titration. The titration with the weak higher affinities. During this optimization process, the inhibitor needs to be performed only once for a particular compounds usually exceed the upper affinity limit of set of conditions. The titration with the high-affinity ITC analysis. inhibitor alone is performed to improve the robustness of the analysis, as it provides a direct measurement of its In 1989, Wiseman et al. [1] showed that the product of the binding enthalpy. protein concentration in the calorimeter cell, [P], and the ≡ × •• binding constant, Ka, a parameter known as c Ka [P], Velazquez-Campoy et al. [4 ] have implemented the must be lower than 1000 for the reaction to be measured ITC competition assay in the analysis of inhibitors of directly by ITC. In practical terms, this restriction sets an HIV-1 protease. In their implementation, Velazquez- upper limit of 108–109 M–1 for the binding constant. Campoy and colleagues used a low-affinity inhibitor with Figure 2 shows the effects of increasingly higher binding a binding enthalpy of opposite sign to that of the high- affinities on the outcome of an ITC titration. As shown in affinity inhibitor, which has the added benefit of the figure, beyond a certain value, the titrations lose their amplifying the calorimetric signal in the displacement characteristic curvature, become indistinguishable from one titration. Figure 3 illustrates the implementation of this another and lack the information necessary to determine the approach to HIV-1 protease. binding constant. The same principle can also be used for low-affinity Recently, a solution to the problem has been obtained by ligands by monitoring the change in binding affinity of a the design of competition experiments in which the high-affinity ligand in the presence and absence of the high-affinity ligand is titrated into protein that is prebound low-affinity ligand under examination. Zhang and Zhang 562 Biophysical methods Figure 2 0.0 –0.2 –0.4 –0.6 –0.8 cal/sec –1.0 µ –1.2 –1 –1 –1 Ka = 1E6 M Ka = 1E8 M Ka = 1E12 M –1.4 c = 50 c = 5000 c = 5E7 –1.6 –1.8 0 20 40 60 80 100 120 0 20 40 60 80 100 120 0 20 40 60 80 100 120 Time (min) Current Opinion in Structural Biology Illustration of the effect of increasing binding affinity on the ability of are shown in the panels. For c > 1000, only a lower limit for the ITC to measure the binding constant. The ITC experiments were binding affinity can be obtained, even though the binding enthalpy can simulated using the following parameters: injection volume 10µL, be measured accurately. This situation is often encountered in drug [protein] = 0.05 mM, [ligand] = 0.6 mM, ∆Hb = –10 kcal/mol, design because the affinity of lead compounds is improved beyond the ≡ vcell = 1.4 mL.