In this article, we continue addressing the analysis of ligand-protein interactions by NMR. Now, we will focus on one of the key experiments that monitor the recognition process from the ligand’s viewpoint: 'The Saturation Transfer Difference' method. The employment of this technique permits the existence of interaction and, in favourable cases, the identification of the ligand epitope to be deduced. (M. Meyer and B. Meyer, Angew. Chem. Int. Ed. Engl., 38, 1784-1788 (1999), B. Meyer & T. Peters, Angew. Chem. Int. Ed. Engl., 42, 864–890 (2003)).
Sample Preparation and Experimental Set-up of Saturation Transfer Difference (STD) Experiments
The protocol is fairly similar to that described before for the TRNOESY method (NMR Solutions #1). However, this experiment is more sensitive and less amount of protein is required. As standard, 0.5 mL of 0.01–0.2 mM of the receptor is employed in this experiment. The ligand is now present in 10–100-fold molar excess over the concentration of receptor binding sites. All the requirements regarding nature, preparation and tests for the sample, including the receptor, the ligand, and the spectrometer requirements are almost identical to those mentioned before in the preceding article and should be addressed as described therein. Independent estimations of binding affinities and dissociation constants are very valuable for the quantitative analysis of STD data. Again, the employment of different ligand/protein molar ratios (i.e., 10/1, 30/1, 70/1,…) is advisable.
The basic STD experiment relies on simple 1H NMR difference spectroscopy. The STD spectrum is obtained by subtracting one spectrum in which the protein resonances are saturated, by using a selective radiofrequency pulse train, from a second one, without protein saturation. In this manner, the difference spectrum (STD) contains only the signals of the ligand(s) that have experienced the transfer of saturation from the protein. As a key decision, the irradiation frequency for saturating the protein is set at a position where only resonance signals from the protein protons and no resonances from the ligand nuclei are present. Therefore, in the so-called 'on-resonance' experiment, the selective saturation of the signals of the protein nuclei is achieved. For practical reasons, the 'on-resonance' irradiation frequency is set at chemical shift values (δ) around -1 or -2 ppm. Most frequently, no ligand proton resonances will be found in this spectral region, unless they display fairly unusual chemical structure. However, the significant line width of the protein signals and the frequent presence of protein aliphatic signals close to aromatic rings, especially in globular proteins, warrant the presence of the protein signals in this spectral region. Therefore, the selective saturation of the protein signals is granted. Nevertheless, if the ligands show no resonance signals in the aromatic proton spectral region, the saturation frequency may also be placed there (at δ ca. 7 ppm).
From the technical perspective, to achieve the required selectivity, shaped pulses (Gaussian-like or similar) are employed for the saturation of the protein signals. While the protein signals are saturated, for those ligand protons that interact with the protein protons, a decrease in their signal intensity will be observed, due to the transfer of saturation from the protein signals. Obviously, this process will only take place if the ligand interacts with the protein. For any other putative ligand molecule present in the NMR tube that does not interact with the protein, its corresponding signal intensities will not be perturbed upon saturation of the protein signals. In particular, the degree of ligand saturation depends on the residence time of the ligand in the protein-binding pocket. Therefore, in principle, by comparing the intensity of the saturated experiment with a regular experiment without saturation, visual inspection could permit the existence of the decrease of intensity and, therefore, the presence of interaction to be determined. However, in the presence of other molecules such as impurities and other non-binding components, it is not usually possible to identify the existence attenuated signals of the ligand. In fact, the attenuation of the ligand signals may be as small as 1-2% of their regular intensities. Therefore, the experiment is performed in the difference mode. A second experiment, the so-called 'off-resonance' spectrum, is then recorded for the same sample. Now the irradiation frequency is set at a value that is far from any signal, ligand or protein resonance, for example, at δ 40-60 ppm. This spectrum is then recorded using identical parameters other than the position of the irradiation frequency and a normal NMR spectrum of the protein-ligand mixture is obtained with the 'natural' intensities. Subtraction of the 'on-resonance' from the 'off-resonance' spectra leads to the STD spectrum, in which only the signals of the protons that were attenuated by saturation transfer from the protein are visible. All molecules without binding activity are cancelled out, since they display the same intensity in the 'on-resonance' and in the 'off-resonance' datasets.
Saturation of the protein and the bound ligand is very fast. In fact, after a few hundreds of ms, the saturation of the protein signals is achieved. Therefore, taking into consideration the existence of an 'on-off' (or bound-free) chemical exchange process for the ligand, the existence of a fast off-rate for the exchange process allows the transfer of the information about saturation quickly into solution. Under these conditions, if a significant excess of the ligand is present, the receptor binding site can saturate many ligand molecules within a few seconds. In fact, the STD experiment (as the TRNOESY) is very suitable for molecular recognition processes whose dissociation rates are fast in the relaxation timescale (at least ca. 2 s-1). Ligands in solution lose the saturation information by regular T1 and T2 relaxation, which is in the second timescale. Therefore, the amount of 'saturated' ligands in solution continuously increases with the saturation time. In this manner, the information resulting from the saturated receptor is largely amplified. As consequence, only a relatively small amount of protein is required. On the other hand, for very tight binding, and concomitantly very slow off-rates (less than 1 s-1), the saturation transfer process is not efficient enough. Thus, the STD method is valuable for dissociation constant values of 100 nM, or larger, even reaching very weak binding processes with dissociation constant about 10 mM.
Apart from detecting binding, STD-NMR is an excellent technique for determining the binding epitope of the ligand: the region of the ligand that is in intimate contact with the protein binding site. This key information is very valuable for a rational drug design process. Indeed, as an outcome of the intermolecular saturation transfer process described above, the ligand saturation is much larger for those protons that are nearer to the receptor protons. This sort of information can be precisely analysed and the STD effect can be quantitatively calculated for a given protein-ligand model by accounting for the equations for the chemical exchange. This is the basis of the CORCEMA-ST program (Rama Krishna et al., J. Magn Reson 155, 106-118 (2002)).
Many applications of STD have been reported, including the derivation of the dissociation constants and allowing the follow-up of molecular recognition processes including living cells, viruses and other large entities. Moreover, the employment of STD methods associated to NMR-active nuclei other than protons has introduced a novel dimension in the study of ligand-receptor interactions.