Solid-state 31P NMR spectroscopy of bacteriophage M13 and tobacco mosaic virus

In this thesis, the results of various ³¹P NMR experiments observed for intact virus particles of bacteriophage M13 and Tobacco Mosaic Virus (TMV), are presented. To explain the results in a consistent way, models are developed and tested. ³¹P nuclei in M13 and TMV are only present in the phosphodiesters of the encapsulated nucleic acid molecule. Therefore, ³¹P NMR spectroscopy reveals structural and dynamic properties of the nucleic acid backbone selectively without isotope labeling, even though the virus particles largely consist of coat proteins. In the Introduction (Chapter 1), it is discussed that the ³¹P chemical shift is sensitive to local nucleic acid backbone geometry and that the ³¹P NMR relaxation is dependent on the isolated and collective backbone motions. As shown in Chapter 3, high-power ¹H-decoupled one-dimensional ³¹P NMR spectra observed for nonspinning samples of M13 and TMV contain a single, broad line dominated by the ³¹P chemical shift anisotropy (CSA), which masks any structural inequivalence among the encapsulated phosphodiesters. However, these spectra do contain interesting mobility information. On the one hand, they show that the nucleic acid molecule in each of the viruses is strongly immobilized in comparison to free nucleic acids in solution, as a result of interactions with the protein coat. On the other hand, the ³¹P resonance lineshapes; show clear signs of motional narrowing, which is indicative for (restricted) motion with frequencies in the order of the static linewidth or larger (≥10 ⁴Hz). In contrast, the nonspinning ³¹P transversal relaxation measured for M13 indicates motion in the slow or intermediate frequency region as compared to the static linewidth (≤10 ⁴Hz), because T _2e becomes shorter as the viscosity of the gel decreases

To analyze the results in a more quantitative manner, three different rotational diffusion models for the phosphorus motion are developed in Chapter 2. These models are first tested at a theoretical level to get a feeling for their accuracy and to check their correspondence with standard theories under appropriate limiting conditions. Simulations show that a clear distinction between the effect of motional amplitude and frequency cannot be made within experimental error on the basis of one-dimensional spectra or transversal relaxation alone. However, these parameters can be extracted from the combined data. For fast motions, the transversal ³¹P NMR relaxation predicted by our models is consistent with standard Redfield relaxation theory. The relaxation effects caused by ultraslow rotational diffusion closely resemble the effects of translational diffusion of water protons in an inhomogeneous magnetic-field gradient. It is discussed in Chapter 3, that simple models, like isotropic and rigid-rod diffusion cannot reproduce the experimental data. Instead, a consistent description is offered by a combined diffusion model, in which the ³¹P NMR lineshape is dominated by fast internal DNA or RNA motions, and transversal relaxation reflects slow overall rotation of the rod-shaped virions about their length axis.

To obtain more specific structural information, magic angle spinning (MAS) NMR spectroscopy is employed, which breaks up the broad ³¹P NMR lineshape into a sharp centerband at the isotropic chemical shift position flanked by rotational sidebands (Chapter 4). MAS ³¹P NMR spectra of TMV show two resolved sideband patterns with an overall intensity ratio of approximately 2, which are assigned to the three types of phosphodiesters in TMV on the basis of RO-P-OR' bondangles and supposed arginine bonding effects. In contrast, MAS ³¹P NMR spectra of M13, only contain a single, relatively broad centerband flanked by sidebands, indicating that a continuous distribution of phosphodiester conformations, rather than a few distinguishable, exists within the phage. The observed decrease of inhomogeneous linewidth at increasing temperature and hydration could perhaps be caused by some sort of "conformational averaging" as a consequence of nucleic acid backbone motion. This is illustrated by use of a simple model, which shows the lineshape effects caused by fast restricted fluctuation of the dihedral angles between the POC and the OCH planes on both sides of the ³¹P nucleus in the nucleic acid backbone. The presence of internal phosphodiester motions with frequencies ≥10 ⁵Hz, as concluded from the motional narrowing of nonspinning ³¹P NMR lineshapes in Chapter 3, is confirmed by the deviation of sideband intensities in MAS ³¹P NMR spectra of dilute M13 gels from the theoretical values for solid powders. No dramatic broadening of the sidebands is observed, indicating that motions with frequencies in the order of the spinning rates applied (10 ³Hz) are absent. Backbone motions also seem to be the main cause of transversal relaxation measured at spinning rates of 4 kHz or higher. At spinning rates below 2 kHz, transversal relaxation is significantly faster. This dependence of T _2e on the spinning rate is assigned to slow, overall rotation of the rod-shaped M13 phage about its length axis.

Both nonspinning and MAS ³¹P NMR spectra are analyzed in Chapter 2 and 3, respectively, to study possible mobility differences among the phosphodiesters in M13 and TMV. The nonspinning lineshape of 30% TMV is best simulated, if it is assumed that one of the three binding sites is more mobile than the other two. It is shown that this is compatible with the reduced CSA reflected by the major sideband pattern in MAS spectra of TMV as compared to the minor one. A large mobility of one of the three binding sites would agree with structural models based on x-ray diffraction data, in which two of the binding sites are interacting with arginine residues, whereas no arginine is close to the third one. Two-component analysis of the nonspinning ³¹P NMR data of 30% M13 suggests that the encapsulated DNA molecule perhaps contains 83% immobile and 17% mobile phosphodiesters. This would shed new light on the nonintegral ratio 2.4:1 between the number of nucleotides and protein coat subunits in the phage: if 83% of the viral DNA is less mobile, the binding of the DNA molecule to the protein coat would actually occur at the integral ratio of two nucleotides per protein subunit. However, MAS NMR spectra provide no additional evidence for such a two-component model.

Finally, in Chapter 5, the slow overall motion of M13 and TMV is investigated using 2D-exchange ³¹P NMR spectroscopy. 2D-exchange ³¹P NMR spectra recorded for TMV with mixing times t _m  ≤1 sec do not show any offdiagonal broadening indicating that the value of 3 Hz for the overall motion of TMV determined in Chapter 3 from nonspinning transversal relaxation, is an overestimation. For 30% M13, a log-Gaussian distribution around 25 Hz of coefficients mainly spread between 1 and 10 ³Hz must be introduced to reproduce the 2D-exchange spectra recorded at various mixing times in a consistent way. Motional inhomogeneity in gels of M13 is probably caused by the tendency of the bacteriophages in solutions to form variously sized aggregates. Taking the same coefficient distribution and a minor relaxation contribution caused by fast backbone motion into account, nonspinning transversal relaxation can even be better simulated for inhomogeneous overall motion, than it was done for homogeneous motion in Chapter 3. The shrinking of the σ ₂₂ -discontinuity on the diagonal with respect to the lineshape as a whole for t _m  ≥0.1 sec, cannot be explained by slow overall motion, but seems to be caused by restricted spindiffusion between 31P nuclei with chemical shifts that differ less than 1 ppm.