Highlighting Cavities in Proteins by NMR Using Sulfur Hexafluoride as

Feb 17, 2010 - which act as reporter, such as a spy. The interest of sulfur hexafluoride (SF6) for probing hydrophobic cavities by solution-state NMR ...
2 downloads 0 Views 1MB Size
3398

J. Phys. Chem. B 2010, 114, 3398–3403

Highlighting Cavities in Proteins by NMR Using Sulfur Hexafluoride as a Spy Molecule Luca Fusaro,†,‡,§ Emanuela Locci,†,⊥ Adolfo Lai,†,# and Michel Luhmer*,‡ Dipartimento di Scienze Chimiche, UniVersita` di Cagliari, Cittadella UniVersitaria di Monserrato, 09042 Monserrato (CA), Italy, and Laboratoire de RMN haute re´solution CP 160/08, UniVersite´ Libre de Bruxelles, 50 AV. F.-D. RooseVelt, 1050 Brussels, Belgium ReceiVed: January 5, 2010; ReVised Manuscript ReceiVed: February 4, 2010

Cavities in proteins can be studied experimentally by using some detectable atoms, such as xenon, or molecules which act as reporter, such as a spy. The interest of sulfur hexafluoride (SF6) for probing hydrophobic cavities by solution-state NMR is investigated. The wheat nonspecific lipid transfer protein (LTP) was selected as a model system for this purpose. The binding of SF6 is straightforwardly detected by the 19F chemical shift, line width, or longitudinal relaxation time measurements, which can be carried out at low SF6 concentration without interference from resonances of the protein. Most interestingly, the binding of SF6 gives rise to selective intermolecular 1H{19F} heteronuclear Overhauser effects (HOEs). Molecular dynamics simulation and NMR spectrum modeling show that the experimental HOESY spectra are consistent with 1H{19F} HOEs arising from SF6 in the cavity of LTP. SF6 is found to be an advantageous alternative to hyperpolarized 129Xe and small organic compounds for probing cavities in proteins by solution-state NMR. 129

Introduction Cavities are nearly always present in proteins containing more than 100 residues and are commonly found in the hydrophobic protein core.1-4 The overall cavity volume increases with protein size, but it remains only a small fraction of the total protein volume, typically less than 2%. Even though cavities may be viewed as structural defects, they can also play an important role in protein function. Indeed, internal cavities may be involved in ligand binding and protein dynamics, for instance. Cavities in proteins are generally studied by using some detectable spy molecules (also referred to as reporter molecules). Noble gases have been widely used in X-ray diffraction studies.5-13 In solution-state NMR, cavities can be detected via chemical shift mapping, although ligand binding may induce conformational changes and indirectly affect the chemical shift of remote protein nuclei.14,15 Intermolecular nuclear Overhauser effects (NOEs) with water can be used to characterize solventaccessible cavities, but possible chemical exchange with labile protons of the protein must be taken into account and complicate the interpretation.16,17 As an alternative, small organic gas or solvent molecules have been used in 1H NMR to directly detect the hydrogen nuclei lining the protein cavity via homonuclear dipole-dipole cross relaxation.17,18 However, intense artifacts arising from the spy molecule 1H signal(s) may significantly affect the quality of the spectra and, more unfortunately, mask wanted NOE signals. Several solution-state 129Xe NMR studies have also been reported.19-27 It has been shown that 129Xe chemical shift measurements and, with paramagnetic proteins, longitudinal relaxation rate measurements as a function of the xenon concentration can be used for the detection and characterization of cavities in proteins. In addition, hyperpolarized * To whom correspondence should be addressed. Phone +32 2 650 6637. Fax +32 2 650 6642. E-mail [email protected]. † Cittadella Universitaria di Monserrato. ‡ Universite´ Libre de Bruxelles. § E-mail: [email protected]. ⊥ E-mail: [email protected]. # E-mail: [email protected].

Xe (also referred to as laser-polarized 129Xe) enables longitudinal relaxation rate measurements in diamagnetic systems and, most interestingly, the observation of selective magnetization transfers to protons of the biomolecule through heteronuclear dipole-dipole cross relaxation. Indeed, SPINOE (spinpolarization-induced nuclear Overhauser effect) experiments have been successfully used for mapping hydrophobic cavities in the lipid transfer protein from wheat and tobacco, in T4 lysozyme, and in β-cryptogein.28-32 Laser-polarized 129Xe NMR studies require optical pumping equipment. The transient nature of the 129Xe hyperpolarization imposes additional experimental constraints and may restrict the range of feasible NMR experiments, notably, advanced two-dimensional NMR experiments. Repeated shaking of the NMR tube or a specific experimental setup is needed to bring fresh hyperpolarized gas into the sample. This may be at the origin of temperature and magnetic field inhomogeneities that significantly affect line widths and line shapes and consequently deteriorate resolution and signal-tonoise ratio. 19 F NMR is finding increasing applications in biological systems, notably, for structural analysis of macromolecules, binding studies, and metabolic studies.33 Various original spectroscopic or imaging approaches have been reported. For instance, intermolecular 1H{19F} NOE measurements can be used to probe solvent-peptide interactions, solvent exposure in proteins, or ligand binding,34-36 and competition experiments exploiting the 19F transverse relaxation of a fluorinated spy molecule allow the rapid screening of drug candidates,37 19F chemical shift imaging of fluorinated reporter molecules enables gene expression assay,38,39 and 19F imaging of inhaled sulfur hexafluoride (SF6) allows quantitative studies of lung function on the basis of signal intensity or longitudinal relaxation time measurements.40,41 Recently, we have reported on the binding of SF6 into the hydrophobic cavity of organic host molecules such as cucurbit[6]uril and R-cyclodextrin (R-CD) in aqueous solutions.42,43 SF6 is chemically inert at usual temperatures. It is a nontoxic, colorless, and odorless hydrophobic gas. The SF6 molecule has an octahedral geometry and is somewhat larger

10.1021/jp100098u  2010 American Chemical Society Published on Web 02/17/2010

Using SF6 for Probing Protein Cavities by NMR

Figure 1. Representation of the 1:1 complex between LTP and prostaglandin B2 (PDB code 1CZ2).47 Orthogonal views showing the protein backbone as a red tube and the prostaglandin B2 atoms as spacefilling spheres.

than a xenon atom (diameters of 5.3 and 4.2 Å, respectively).44 The detection limit of dissolved SF6 by 19F NMR is in the µM range for measurement times of a few minutes with standard equipment. In addition to chemical shift measurements, solutionstate 19F NMR of SF6 allows quantitative integral measurements, relaxation time measurements, and demanding experiments, such as HOESY (heteronuclear Overhauser effect spectroscopy).42,43 Selective magnetization transfers through heteronuclear 19F-1H dipole-dipole cross relaxation have been successfully used to highlight SF6 binding and characterize the geometry of the SF6/ R-CD complex.43 In this paper, we report on the use of SF6 for probing hydrophobic cavities in proteins by solution-state NMR and, especially, via 1H{19F} HOESY experiments. The studied system is the wheat nonspecific lipid transfer protein (LTP). This 90 residue protein contains a large hydrophobic cavity that can accommodate lipid molecules (Figure 1). The solution-state structure was determined by NMR,45,46 and the LTP was previously selected as a model system for protein cavity characterization by intermolecular 1H-1H NOE measurements using small organic spy molecules18 as well as by 129Xe-1H SPINOE measurements.25,28,29 The present experimental study is supported by a 50 ns unrestrained molecular dynamics simulation, which was used to model and assign the experimental HOESY spectra. Experimental Methods SF6 gas (99.97%) was purchased from Air Liquide. D2O (99.98%), benzoylated dialysis tubing (D7884), and the color marker ultralow range for SDS-PAGE (C6210) were purchased from Sigma-Aldrich. Vivaflow filtration system membranes (5.000 and 10.000 MWCO PES) were purchased from Sartorius. The extraction and purification of the wheat nonspecific lipid transfer protein are described in the Supporting Information. The LTP solutions were prepared by dissolving the protein in D2O and were transferred into 5 mm NMR tubes equipped with a valve (Wilmad 507-PV-7). The samples were degassed at room temperature by three cycles of evacuation. SF6 gas, about 10 atm, was pressurized at room temperature, and the samples were gently shaken for a couple of minutes. The samples were stored at 4 °C when not used. They were left at the desired temperature in the magnet for at least 30 min before starting the NMR measurements. 1 H and 19F NMR spectra were recorded at 9.4 T (399.917 MHz for 1H and 376.318 MHz for 19F) with temperature

J. Phys. Chem. B, Vol. 114, No. 9, 2010 3399 regulation and an automated triple broad-band probe simultaneously tuned to 1H and 19F. 1H NMR spectra were recorded using a 90° pulse (12.1 µs), 2.0 s acquisition time, 8.0 s relaxation delay, and a spectral width of 16.0 ppm centered at 5.0 ppm. The signal of HDO was used for 1H chemical shift referencing.48 19F NMR spectra were recorded using a 90° pulse (10.3 µs), 2.0 s acquisition time, 5.0 s relaxation delay, a spectral width of 32 ppm centered at about +58 ppm with respect to CFCl3, and, typically, 32 scans. A sample of SF6 (1 atm) dissolved in pure D2O was used as an external 19F chemical shift reference. 19F spin-lattice relaxation times (T1) were measured using the inversion-recovery pulse sequence, 17 recovery delays ranging between 0.1 and 5 s, a 5.5 s repetition time, and 4 scans. Monodimensional (1D) 1H{19F} HOESY spectra were recorded using the NMR pulse sequence previously reported by Gerig,49 several values of the mixing time ranging between 0.1 and 1.2 s, a 0.5 s acquisition time, a 4.5 s relaxation delay, and 10000 scans. The processing of both the 1H and 19F NMR measurements included exponential multiplication of the free induction decay (FID) and zero-filling prior to Fourier transform. T1 data were determined using integrated intensities and a three parameter nonlinear fit; reported confidence intervals correspond to twice the fitting errors. Molecular dynamics simulation details are described in the Supporting Information. Results and Discussion NMR Spectroscopy. 1H NMR spectra of 0.86 and 4.9 mM LTP aqueous solutions were recorded at 278 and 293 K in the absence of SF6 (degassed samples) and in the presence of about 10 atm of SF6 gas. Considering the solubility of SF6 in pure D2O, 2.6 × 10-4 M at 298 K and 1 atm,43 the concentration of SF6 free in solution is estimated to be in the mM range, that is, of the same order of magnitude as the LTP concentration. The 1 H NMR signals of LTP are somewhat broader at 278 K than those at 293 K (see Supporting Information), but no effect of SF6 was detected in 1D 1H NMR spectra. In contrast, the presence of LTP affects the 19F NMR spectrum of dissolved SF6. The data indicate fast exchange of SF6 between at least two environments. Indeed, a single signal was detected for 32SF6, but increasing low-field shifts as well as signal broadening were observed for increasing LTP concentrations (Figure 2).50 Under the present experimental conditions, dissolved SF6 is not expected to form aggregates. In contrast to the 129Xe chemical shift, the 19F chemical shift of SF6 is not expected to be sensitive to transient nonspecific interactions with the protein surface.43 Therefore, the observed low-field shifts and signal broadening are ascribable to SF6 binding, most probably in the hydrophobic cavity of LTP. Both the chemical shift and line width variations are linear in the investigated LTP concentration range (Figure 2), and this suggests weak binding of SF6 to LTP. In the presence of LTP, the line width of the 19F NMR signal of 32SF6 is significantly larger at 278 K, and this could be due to stronger binding, enhanced transverse relaxation as a consequence of longer correlation time for bound SF6, and/or slower exchange between free and bound SF6. The 19F longitudinal relaxation time (T1) of SF6 is also affected and was found to decrease for increasing LTP concentrations (Table 1). This is similar to the trend previously observed for the complexation of SF6 by R-CD43 and indicates enhanced longitudinal relaxation as a consequence of binding to LTP. In simple solvents, the longitudinal relaxation of 32SF6 is dominated by the spin-rotation mechanism, as indicated by shorter T1 values at higher temperature.43 The T1 values measured in the presence of LTP are

3400

J. Phys. Chem. B, Vol. 114, No. 9, 2010

Fusaro et al.

Figure 2. 19F chemical shift and line width results. (a) Chemical shift, δ, and (b) full line width at half-height, ∆ν1/2, of the 19F NMR signal of 32 SF6 dissolved in pure D2O (∼1 atm of SF6) and in aqueous solutions of LTP (∼10 atm of SF6) measured at 278 (empty symbols, solid line) and 293 K (filled symbols, dashed line). The chemical shift measured in D2O at 278 K was set to 0. Line width data comprise a contribution due to exponential apodization of the FID (lb ) 0.3 Hz).

TABLE 1: 19F Longitudinal Relaxation Time (T1) of 32SF6 Dissolved in Pure D2O (∼1 atm of SF6) and in Aqueous Solutions of LTP (∼10 atm of SF6) [LTP] (mM)

T1 (s) at 278 K

T1 (s) at 293 K

0.00 0.86 4.9

1.18 ( 0.03 1.09 ( 0.01 1.00 ( 0.01

1.06 ( 0.02 0.95 ( 0.01

shorter at 293 K than at those at 278 K (Table 1). However, this does not indicate that the longitudinal relaxation of bound SF6 is dominated by the spin-rotation mechanism. Indeed, for fast exchange and weak binding, it is likely that the temperature dependence of the observed T1 is dictated by the temperature dependence of the longitudinal relaxation of free SF6. 1D 1H{19F} HOESY spectra recorded for increasing mixing times clearly reveal the binding of SF6 (Figure 3 and Supporting Information). Indeed, these spectra show signals at about 1 ppm arising from selective intermolecular magnetization transfers. The 1H enhancements are negative, indicating that the average correlation time of the 19F-1H dipole-dipole interactions and, consequently, the average residence time of SF6 in its binding site(s) are longer than ∼0.5 ns.51 The 1D HOESY spectra obtained with SF6 have a better overall signal-to-noise (S/N) ratio than the 1D SPINOE spectra obtained with hyperpolarized 129Xe, which were recorded for xenon concentrations in the mM range as is the case for SF6 in the present study.29 This is probably mainly due to the limited number of scans that can be recorded before the 129Xe hyperpolarization vanishes. In contrast, extensive signal averaging is possible with SF6 since the measurements rely on the equilibrium 19F polarization, which is recovered in a few seconds (see Table 1). With small organic molecules, Liepinsh et al.18 obtained better S/N ratios, but their experiments involved much higher concentrations of the spy molecule in solution. For instance, they used 5% v/v acetonitrile and, with methane, pressures up to 190 bar, corresponding to concentrations in solution up to about 0.4 M. On the whole, the spectra obtained with SF6 at 9.4 T (400 MHz for 1H) are in agreement with the spectra recorded at 14.1 T (600 MHz for 1H) by using hyperpolarized 129Xe or small organic molecules such as acetonitrile, methane, ethane, or cyclopropane; they mainly show signals in the spectral region between 0.6 and 1.1 ppm.18,29 It is worth noting that the NOE spectra obtained with these small organic molecules are not identical; weak signals observed with cyclopropane are not detected with acetonitrile, for instance. In these spectra, there are two signals of significant intensity at

Figure 3. 1H{19F} HOESY results. (a-f) Region of the experimental spectra recorded at 9.4 T (400 MHz for 1H) and 293 K for a 4.9 mM solution of LTP in D2O and ∼10 atm of SF6; (g) modeled HOESY spectrum based on the MD results. (a) Region of the equilibrium 1H spectrum of LTP for which HOESY signals were observed (full spectra are shown in the Supporting Information); (b) control HOESY spectrum obtained without 19F excitation (no power on the 19F channel, mixing time of 400 ms); (c-f) HOESY spectra recorded for increasing mixing time (100, 400, 800, and 1200 ms, respectively). The line broadening factor (lb) used for exponential apodization is 3 Hz.

about 4.2 and 0.6 ppm which are not visible in the spectra obtained with SF6. The signal at 4.2 ppm was not observed with hyperpolarized 129Xe either. The assignment of protein signals arising from intermolecular Overhauser effects is not trivial. Only two residues were unambiguously identified by Liepinsh et al.18 in the spectra obtained with acetonitrile, methane, ethane, or cyclopropane; two weak signals at about 6.5 ppm were assigned to Tyr79, and the signal at about 0.6 ppm was assigned to Leu77. For assignment, Landon et al.29 used 1H chemical shift variations observed at a higher xenon concentration and performed a solvation simulation using static solution-state structures of LTP. They identified 16 hydrophobic residues giving rise to SPINOE signals, which mainly correspond to methyl groups. The signal at about 0.6 ppm was assigned to Val75 and not to Leu77.

Using SF6 for Probing Protein Cavities by NMR

J. Phys. Chem. B, Vol. 114, No. 9, 2010 3401

Figure 4. Representation of the 1:1 SF6-LTP complex. Orthogonal views showing the protein backbone as a red tube (50 ns MD time average), the spatial density function of SF6 as a green volume, and the protein nonexchangeable hydrogen atoms giving rise to 1H-19F dipole-dipole interactions (apparent average H-F distance shorter than 5.5 Å) as white spheres.

Molecular Dynamics Simulation. The 1:1 SF6-LTP complex dissolved in water was studied by molecular dynamics (MD) simulation using an all-atom description. A single SF6 molecule was initially located in the cavity of LTP, and the external surface of the protein was solvated by 4437 water molecules (see the Supporting Information). After equilibration, a 50 ns unrestrained MD simulation was run, saving the trajectories for subsequent analysis. No important conformational change of the protein occurred during the course of the simulation, as indicated by the backbone root-mean-square deviation (see the Supporting Information). SF6 did not escape from the LTP cavity; the calculated SF6 spatial density function suggests a single preferential binding site deep inside of the cavity of LTP (Figure 4). As expected, the rotational motion of SF6 was found to be fast compared to the reorientation of the protein. The experimentally observed Overhauser effects were not used as input data in the analysis of the MD trajectories.52 LTP is comprised of 90 residues and 646 hydrogen atoms in total, which corresponds to 3876 H-F pairs of atoms. The number of H-F pairs to analyze was reduced by discarding those residues which are too far from SF6 to give rise to significant 1 H-19F dipole-dipole interactions. Sixteen residues were found to have, on average, at least one atom less than 7 Å from the center of the SF6 molecule (Figure 5). It is worth noting that Val75, considered by Landon et al.29 as giving rise to SPINOE signals, does not belong to the selected set of residues, while Leu77 does. These 16 residues are comprised of 133 hydrogen atoms, and for each of them, the time average of the inverse sixth power of their distance to the fluorine atoms of SF6, that is, the average proximity factor of the 1H-19F dipole-dipole interactions, was computed according to

rH-6i-F

6

)

∑ rH-6-F j)1

i

j

(1)

where rHi-Fj is the distance between one hydrogen atom, Hi, of the selected set of 16 residues and one of the six fluorine atoms, Fj, of SF6. The results of these calculations are shown in Figure 6, where the contribution of chemically equivalent 1H, such as the hydrogen atoms of the methyl groups, is summed up (see also the Supporting Information). Twelve LTP residues are found to give rise to 1H-19F dipole-dipole interactions corre-

Figure 5. Time average of the shortest interatomic distance between the LTP residues and the sulfur atom of SF6. It is shorter than 7 Å for 16 residues (Leu9, Val10, Cys13, Val17, Cys27, Val31, Leu34, Leu51, Ala66, Ile69, Pro70, Cys73, Leu77, Tyr79, Thr80, and Ile81).

Figure 6. Relative average proximity factor of the 1H-19F dipole-dipole interactions between SF6 and the 16 nearest LTP residues. The horizontal dashed line indicates the corresponding value for a 1H-19F distance of 5.5 Å. The data are tabulated in the Supporting Information.

sponding to an apparent distance shorter than 5.5 Å, which is the NOE cutoff distance commonly used in protein NMR structure refinements.52,53 The major effects are due to aliphatic 1 H, mainly from methyl groups, of eight residues, Val10, Val31, Leu34, Leu51, Ile69, Leu77, Thr80, and Ile81. The comparison of the MD and experimental results can be achieved by spectrum modeling. The adopted procedure relies on the chemical shift assignment of Simorre et al.,45 which is unfortunately not complete (about 90% of the 1H resonances were assigned), ignores scalar coupling, and uses Lorenztian lines of identical width for all of the signals. Though basic, this approach provides rather good results for the equilibrium 1D 1H NMR spectrum of LTP (see the Supporting Information). The 1D 1H{19F} HOESY spectrum was modeled by scaling the signal intensities according to the calculated proximity factor of the 1H-19F dipole-dipole interactions. In this scaling procedure, possible spin-diffusion effects are ignored, and it is implicitly assumed that the correlation time characterizing the fluctuations of the 1H-19F dipole-dipole interactions as well as the autorelaxation rate are identical for all of the 1H. A few HOESY signals predicted by the MD simulation are not observed experimentally, among which are the signals at about

3402

J. Phys. Chem. B, Vol. 114, No. 9, 2010

4.2 and 0.6 ppm mentioned above (Figure 3 and Supporting Information). The signal predicted at about 8 ppm corresponds to the exchangeable NH of Ile81, which cannot be observed in D2O. Considering the experimental S/N ratio, the remaining HOESY signals predicted in the spectral region between 1.5 and 5 ppm are too weak to be detected. In agreement with the experimental spectra, the HOESY signals are mainly predicted in the spectral region between 0.6 and 1.1 ppm (see Figure 3). In this region, the resulting shape of the modeled spectrum is highly similar to the experimental spectra, showing three extrema at about 0.85, 1.00, and 1.10 ppm. In both the calculated spectrum and the experimental spectrum (d), recorded with a mixing time of 0.4 s, the intensity at 0.85 ppm is about twice as large as that at 1.10 ppm. The intensity predicted at 1.00 ppm is much larger, however. Experimentally, it gets larger at longer mixing times, but it must be stressed that spin-diffusion is likely to affect the shape of spectra recorded with mixing times longer than 0.4 s. It is worth emphasizing that the calculated spectrum is only comprised of intermolecular HOEs, and in contrast to intramolecular 1H-1H NOE calculations, overall qualitative agreement with the experimental spectrum is not expected if the structures used in the calculations do not properly model the experimental system.52

Fusaro et al. the MD simulations as well as Rita D’Orazio (Brussels) for technical support for the NMR. The authors thank the Executive Program for Belgium-Italy Scientific Cooperation (Project 2007-2008 05618-S) for financial support. M.L. thanks the “Fonds de la Recherche Scientifique” (FNRS-FRS), the “Bureau des relations internationales et de la coope´ration” (BRIC) of the Universite´ Libre de Bruxelles, and the “Regione Autonoma della Sardegna” for financing missions in Cagliari. Supporting Information Available: LTP extraction and purification, (Figure S1) SDS-PAGE electrophoresis, (Figure S2) experimental 1D 1H NMR spectra of LTP, computational details, (Figure S3) comparison of experimental and modeled spectra of LTP, significant contributions to the 1H-{19F} HOE spectrum as determined by MD simulation, and (Figure S4) rmsd of the CR atoms with respect to the original PDB structure. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Rashin, A. A.; Iofin, M.; Honig, B. Biochemistry 1986, 25, 3619– 25. (2) Hubbard, S. J.; Gross, K. H.; Argos, P. Protein Eng. 1994, 7, 613– 26.

Conclusions The wheat nonspecific lipid transfer protein (LTP) was selected as a model system to evaluate the interest of SF6 for detecting and characterizing hydrophobic cavities in proteins by solution-state NMR. The observation of the 19F NMR signal of SF6 dissolved in aqueous solutions is straightforward, even at low concentration, and does not suffer from interference with the protein signals. SF6 aggregation or SF6 adsorption at the protein surface is unlikely, and nonspecific interactions do not significantly affect the 19F chemical shift of SF6. Therefore, the occurrence of SF6 binding can be readily detected by chemical shift, line width, and/or longitudinal relaxation time measurements as a function of the protein concentration. Most interestingly, the binding of SF6 gives rise to selective intermolecular 1 H{19F} heteronuclear Overhauser effects (HOEs). Molecular dynamics simulation was used to calculate the average proximity factor of the 19F-1H dipole-dipole interactions between SF6 within the hydrophobic cavity of LTP and the surrounding protein hydrogen nuclei; an original spectrum modeling approach was developed for the comparison of calculated and experimental data. The results show that the experimental HOESY spectra are consistent with 1H{19F} HOEs arising from SF6 in the cavity of LTP. SF6 permits one to avoid the experimental constraints associated with hyperpolarized 129Xe and provides similar results. The 1H{19F} HOE signal intensities depend on the affinity of the protein for SF6, which happens to be weak in the case of LTP, and are expected to get stronger at higher SF6 concentrations. In contrast to small organic molecules or gases exploited for probing protein cavities via 1H{1H} NOE measurements, SF6 does not give rise to interfering resonances and could be used at higher concentrations. In conclusion, SF6 is a valuable alternative to hyperpolarized 129Xe and small organic compounds for probing cavities in proteins by solution-state NMR. Acknowledgment. The authors gratefully acknowledge Enrico Sanjust, Francesca Sollai, and Alessandra Olianas (Cagliari) for their contribution to the extraction and purification of LTP, Francesca Mocci and Giuseppe Saba for fruitful discussions on

(3) Vallone, B.; Brunori, M. Ital. J. Biochem 2004, 53, 46–52. (4) Dubey, V. K.; Jagannadham, M. V. Curr. Proteomics 2008, 5, 157– 160. (5) Schoenborn, B. P. Nature 1965, 208, 760–2. (6) Schoenborn, B. P.; Watson, H. C.; Kendrew, J. C. Nature 1965, 207, 28–30. (7) Schoenborn, B. P.; Nobbs, C. L. Mol. Pharmacol. 1966, 2, 495–8. (8) Tilton, R. F., Jr.; Kuntz, I. D., Jr.; Petsko, G. A. Biochemistry 1984, 23, 2849–57. (9) Schiltz, M.; Fourme, R.; Broutin, I.; Prange, T. Structure 1995, 3, 309–16. (10) Montet, Y.; Amara, P.; Volbeda, A.; Vernede, X.; Hatchikian, E. C.; Field, M. J.; Frey, M.; Fontecilla-Camps, J. C. Nat. Struct. Biol. 1997, 4, 523–6. (11) Prange, T.; Schiltz, M.; Pernot, L.; Colloc’h, N.; Longhi, S.; Bourguet, W.; Fourme, R. Proteins 1998, 30, 61–73. (12) Inoue, K.; Yamada, H.; Imoto, T.; Akasaka, K. J. Biomol. NMR 1998, 12, 535–541. (13) Collins, M. D.; Quillin, M. L.; Hummer, G.; Matthews, B. W.; Gruner, S. M. J. Mol. Biol. 2007, 367, 752–763. (14) Diercks, T.; Coles, M.; Kessler, H. Curr. Opin. Chem. Biol. 2001, 5, 285–291. (15) Pellecchia, M.; Sem, D. S.; Wuthrich, K. Nat. ReV. Drug DiscoVery 2002, 1, 211–219. (16) Otting, G.; Liepinsh, E.; Wuethrich, K. Science 1991, 254, 974– 80. (17) Otting, G. Prog. Nucl. Magn. Reson. Spectrosc. 1997, 31, 259– 285. (18) Liepinsh, E.; Sodano, P.; Tassin, S.; Marion, D.; Vovelle, F.; Otting, G. J. Biomol. NMR 1999, 15, 213–225. (19) Locci, E.; Dehouck, Y.; Casu, M.; Saba, G.; Lai, A.; Luhmer, M.; Reisse, J.; Bartik, K. J. Magn. Reson. 2001, 150, 167–174. (20) Rubin, S. M.; Spence, M. M.; Dimitrov, I. E.; Ruiz, E. J.; Pines, A.; Wemmer, D. E. J. Am. Chem. Soc. 2001, 123, 8616–7. (21) Rubin, S. M.; Spence, M. M.; Pines, A.; Wemmer, D. E. J. Magn. Reson. 2001, 152, 79–86. (22) Rubin, S. M.; Lee, S.-Y.; Ruiz, E. J.; Pines, A.; Wemmer, D. E. J. Mol. Biol. 2002, 322, 425–440. (23) Groger, C.; Moglich, A.; Pons, M.; Koch, B.; Hengstenberg, W.; Kalbitzer, H. R.; Brunner, E. J. Am. Chem. Soc. 2003, 125, 8726–8727. (24) Lowery, T. J.; Rubin, S. M.; Ruitz, E. J.; Pines, A.; Wemmer, D. E. Angew. Chem., Int. Ed. 2004, 43, 6320–6322. (25) Dubois, L.; Parres, S.; Huber, J. G.; Berthault, P.; Desvaux, H. J. Phys. Chem. B 2004, 108, 767–773. (26) Nisius, L.; Stadler, M.; Kalbitzer, H. R.; Brunner, E. J. Phys. Chem. B 2005, 109, 17795–17798. (27) Blobel, J.; Schmidl, S.; Vidal, D.; Nisius, L.; Bernado, P.; Millet, O.; Brunner, E.; Pons, M. J. Am. Chem. Soc. 2007, 129, 5946–5953. (28) Berthault, P.; Landon, C.; Vovelle, F.; Desvaux, H. C. R. Acad. Sci., Ser. IV: Phys. Astrophys. 2001, 2, 327–332. (29) Landon, C.; Berthault, P.; Vovelle, F.; Desvaux, H. Protein Sci. 2001, 10, 762–770.

Using SF6 for Probing Protein Cavities by NMR (30) Dubois, L.; Da Silva, P.; Landon, C.; Huber, J. G.; Ponchet, M.; Vovelle, F.; Berthault, P.; Desvaux, H. J. Am. Chem. Soc. 2004, 126, 15738– 15746. (31) Desvaux, H.; Dubois, L.; Huber, G.; Quillin, M. L.; Berthault, P.; Matthews, B. W. J. Am. Chem. Soc. 2005, 127, 11676–11683. (32) Berthault, P.; Huber, G.; Ha, P. T.; Dubois, L.; Desvaux, H.; Guittet, E. ChemBioChem 2006, 7, 59–64. (33) Cobb, S. L.; Murphy, C. D. J. Fluorine Chem. 2009, 130, 132– 143. (34) Chatterjee, C.; Hovagimyan, G.; Gerig, J. T. ACS Symp. Ser. 2007, 949, 379–392. (35) Kitevski-LeBlanc, J. L.; Evanics, F.; Prosser, R. S. J. Biomol. NMR 2009, 45, 255–264. (36) Keidel, D.; Bonaccio, M.; Ghaderi, N.; Niks, D.; Borchardt, D.; Dunn, M. F. ChemBioChem 2009, 10, 450–453. (37) Dalvit, C. Prog. Nucl. Magn. Reson. Spectrosc. 2007, 51, 243– 271. (38) Yu, J.-X.; Kodibagkar, V. D.; Cui, W.; Mason, R. P. Curr. Med. Chem. 2005, 12, 819–848. (39) Yu, J.-X.; Kodibagkar, V. D.; Liu, L.; Mason, R. P. NMR Biomed. 2008, 21, 704–712. (40) Adolphi, N. L.; Kuethe, D. O. Magn. Reson. Med. 2008, 59, 739– 46. (41) Scholz, A. W.; Wolf, U.; Fabel, M.; Weiler, N.; Heussel, C. P.; Eberle, B.; David, M.; Schreiber, W. G. Magn. Reson. Imaging 2009, 27, 549–556. (42) Fusaro, L.; Locci, E.; Lai, A.; Luhmer, M. J. Phys. Chem. B 2008, 112, 15014–15020.

J. Phys. Chem. B, Vol. 114, No. 9, 2010 3403 (43) Fusaro, L.; Locci, E.; Lai, A.; Luhmer, M. J. Phys. Chem. B 2009, 113, 7599–7605. (44) Kuethe, D. O.; Pietrass, T.; Behr, V. C. J. Magn. Reson. 2005, 177, 212–220. (45) Simorre, J.-P.; Caille, A.; Marion, D.; Marion, D.; Ptak, M. Biochemistry 1991, 30, 11600–11608. (46) Gincel, E.; Simorre, J.-P.; Caille, A.; Marion, D.; Ptak, M.; Vovelle, F. Eur. J. Biochem. 1994, 226, 413–22. (47) Tassin-Moindrot, S.; Caille, A.; Douliez, J.-P.; Marion, D.; Vovelle, F. Eur. J. Biochem. 2000, 267, 1117–1124. (48) Gottlieb, H. E.; Kotlyar, V.; Nudelman, A. J. Org. Chem. 1997, 62, 7512–7515. (49) Gerig, J. T. J. Org. Chem. 2003, 68, 5244–5248. (50) Sulfur possesses four stable natural isotopes, and consequently, various lines may be observed in the 19F NMR spectrum of SF6; see: Gillespie, R. J.; Quail, J. W. J. Chem. Phys. 1963, 39, 2555–2557. The singlet signal of 32SF6 (95.06% of natural abundance) dominates the spectrum. (51) The zero-crossing of the heteronuclear 1H{19F} Overhauser effect arises for ωHτc ∼1.16; at 9.4 T, the corresponding correlation time, τc, is 0.46 ns. (52) Zagrovic, B.; van Gunsteren, W. F. Proteins: Struct., Funct., Bioinf. 2006, 63, 210–218. (53) Schwalbe, H.; Grimshaw, S. B.; Spencer, A.; Buck, M.; Boyd, J.; Dobson, C. M.; Redfield, C.; Smith, L. J. Protein Sci. 2001, 10, 677–88.

JP100098U