Arginine has a unique role among the 20 standard proteinogenic amino acids. In contrast to other charged amino acid side-chains the charge of arginine side-chains is relatively unaffected by the surrounding environment.1 As a consequence, arginine provides nature with a reliable means of placing a positive charge at any required position in a biomolecule, such as in active sites or at protein–ligand interfaces. Few methods are currently available to probe charged side-chains and their dynamics under physiological conditions, although such methods are pivotal in order to elucidate protein–ligand interactions and important to relate molecular motions to biomolecular functions.
NMR spectroscopy is a powerful technique to probe protein environments and characterize dynamics of proteins at atomic resolution.2 For amino acids with large side-chains, such as arginine and isoleucine, the terminal moieties are nearly uncoupled from the backbone.3 Methods that probe large methyl-bearing side-chains have therefore been successfully applied to characterize the function of enzymes and macromolecular machines.4 Techniques that probe charged side-chains are currently limited,5 despite their importance for enzyme catalysis and protein–ligand interactions. With regard to probing arginine and lysine side-chains, conventional methods are often impeded by rapid exchange of the detected protons with the bulk solvent, which can lead to extensive line-broadening and effectively undetectable signals at and above physiological pH. To alleviate this problem, Mulder and co-workers6 introduced an elegant approach relying on detection of aliphatic protons. Whilst this method circumvents the rapid exchange and allows measurements of the 15Nε chemical shift of arginine, it is less beneficial for nuclear spin relaxation measurements and for studying larger proteins, where the two bound protons enhance the transverse relaxation of the aliphatic carbons.
Herein we describe an NMR pulse scheme for probing arginine side-chains and their dynamics at neutral-to-high pH by carbon-detected 13Cζ–15Nε correlated spectra (Figure 1). Magnetization is transfered from 13Cζ to 15Nε via the scalar coupling, while transfers from 13Cζ to 15Nη are avoided by applying a selective 15Nε inversion pulse in the INEPT transfer steps. After the initial INEPT transfer, the magnetization is proportional to 2Cζ,zNε,z, which allows spin relaxation measurements and quantification of the squared order parameter, S2, that reports on the motions of the arginine side-chain.7 Here, S2≈0 indicates that the motion of the side-chain is completely uncoupled from the overall molecular tumbling, whereas S2≈1 shows that the side-chain is rigid with respect to the overall molecular frame.
As a first application we probed the arginine side-chains of T4 lysozyme L99A (T4L L99A; 18 kDa) shown in Figure 2 a. T4L L99A is stable at pH 5.512 and therefore provides an opportunity for a comparison of the carbon-detected methodology proposed here with the conventional proton-detected methods. The 13Cζ–15Nε spectrum of T4L L99A, Figure 2 b, shows many well-dispersed peaks and a chemical shift dispersion similar to the corresponding 1Hε–15Nε spectrum (Supporting Information). As expected, the signal/noise ratios of the disperse peaks of the 13Cζ–15Nε spectrum are approximately a factor of 30 less than the corresponding peaks of the 1Hε–15Nε spectrum obtained with the same recording time. The proton-detected experiments therefore remain preferable at pH below ca. 6, whereas at higher pH the disadvantage of the longer recording time of the 13Cζ–15Nε spectrum is often outweighed by the fact that the 1Hε–15Nε spectrum provides very limited information.
Spin relaxation rates were obtained from a series of 13Cζ–15Nε spectra and order parameters were subsequently calculated for the arginine side-chains of T4L L99A. Figure 2 d shows a good agreement when these order parameters are compared to the corresponding order parameters, S2HN, obtained using proton-detected experiments.16 For the seven isolated peaks (R8, R14, R52, R95, R96, R148, R154), we obtain RMSD(S2CN, S2NH)=0.05. Including the peaks of the more crowded region (green in Figure 2 d) gives the same general picture, despite a higher uncertainty of these parameters.
The human histone deacetylase 8 (HDAC8) catalyzes the de-acetylation of lysine side-chains in cells and helps to balance the acetylation state of proteins.17 HDAC8 is a 42 kDa metalloenzyme with 11 arginines, at least one of which is crucial for activity.18 Activity assays and purifications have been established at pH≈8 and the unliganded form of HDAC8 appears to be unstable at pH lower than ca. 7, based on 2D NMR. Thus, the 1Hε–15Nε spectrum is not applicable to probe the arginine side-chains of HDAC8, as shown in Figure 3 a, and these side-chains must be probed independently of the 1Hε spin. Figure 3 a and 3 b show a 1Hε–15Nε and a 13Cζ–15Nε spectrum, respectively, of HDAC8 at pH 8.2 and clearly demonstrate that under these conditions the 13Cζ–15Nε spectrum reveals many more features than the corresponding 1Hε–15Nε spectrum.
Potassium binding has recently been shown to regulate the activity of HDAC8.19 Crystal structures of HDAC820 show that the Cζ of R223 is ca. 12 Å from one of the two potassium binding sites, which allows us to investigate the wider consequences on the side-chain packing of K+ binding. The peak of R223 was assigned by mutation of R223 to lysine, which caused the disappearance of an isolated peak as seen in Figure 3 c. Other minor perturbations are observed in the spectrum as expected for this allosteric enzyme and also observed for the assignment of other side-chain chemical shifts.4 At low concentrations of K+ the R223 peak is hardly visible, whereas its intensity increases as the concentration of K+ is increased, Figure 3 d. On the contrary, addition of Na+ does not lead to the same increase in intensity, thus attesting that the effect observed is due to specific K+ binding rather than electrostatic stabilization of the protein. No extra isolated peak appears at low concentrations of K+, suggesting that the R223 side-chain of the potassium-free form is either disordered or undergoing chemical exchange, such that the corresponding peak is located in the random-coil region or broadened beyond detection, respectively. Although the four titration points obtained here are not sufficient to determine an accurate dissociation constant for K+, an initial estimation of KD≈40 mM is in agreement with a previously determined KD from activity assays.19 Overall, our arginine data show that binding of K+, which activates HDAC8, affects not only the backbone binding site, but also changes the side-chain packing beyond the binding site.
In general, the side-chains of amino acids probe a different environment from that of the backbone, and charged side-chains probe a different environment from that of hydrophobic side-chains. It has now become clear that probing methyl-bearing side-chains provides very important information about protein function and dynamics.21 The methodology presented here extends the utility of side-chains as probes of structure and dynamics to include the charged arginine side-chain, and in the context of proteins at physiological pH provides an avenue for characterizing arginine side-chain interactions at a level of detail that has largely been, until now, reserved for applications to methyl-bearing side-chains.