Structure of sesbania mosaic virus at 3A resolution

Abstract

Viruses were recognized for the first time in 1898 as pathogens causing mosaic disease in tobacco plants. Since then, they have been studied by a variety of biophysical and biochemical techniques, as they are good model systems for studying cellular phenomena. Structural studies on viruses were carried out as early as the 1940s. However, the first three-dimensional structure of a virus, tomato bushy stunt virus (TBSV), was determined by Harrison and his colleagues only in 1978. This opened the floodgates for virus crystallography. Today, three-dimensional structures of several viruses are known. In this context, the thesis reports the three-dimensional structure of a plant virus, Sesbania mosaic virus (SMV), at 3 Å resolution.

Sesbania mosaic virus was first detected as the causal agent of the mosaic disease of sesbania plants in Tirupati, India. Most studies on viruses in India are concerned with epidemiology, damage they cause to flora and fauna, and methods for limiting their harmful effects. Structural studies on SMV were initiated to explore the possibility of gaining further insights into understanding aspects of viral architecture, assembly/disassembly, evolution, etc., as part of the program on structural studies on biological macromolecules at the Indian Institute of Science, Bangalore.

SMV belongs to the sobemovirus family of plant viruses, which are characterized by stability in vitro and mechanical transmissibility. SBMV is the most well-studied member of this group. SMV is serologically related to SBMV-cp (cowpea strain). The three-dimensional structure and the complete nucleotide sequence were known only for SBMV-cp when structural studies on SMV were initiated.

The virus culture was maintained on sesbania plants in a greenhouse. SMV crystallized predominantly in two crystal forms. The rarer, rhombohedral form was used for our studies. The other form could not be characterized. SMV crystals belong to space group R3 with cell dimensions of a = 291.46 Å and ? = 61.95°. The crystals diffract to better than 3 Å resolution. The unit cell contains one virus particle, with its threefold coincident with the crystal threefold. The unit cell volume of SMV is 25% less than that of SBMV.

High-resolution data collection on SMV crystals was primarily carried out by photographic methods. Nevertheless, three data sets of intermediate resolution were collected on a Nicolet/Siemens area detector. XENGEN was used for processing these data frames, which was not straightforward due to the pseudo-cubic nature of the R3 crystals. The merging R-factor for the data set was 15.82% for all measurements with I/?(I) > 1.0.

Kodak DEF films were used to record diffraction data to 3 Å by the method of screenless oscillation photography. For this, the crystal was mounted such that the crystal threefold was along the spindle. A total of 37.5° was covered using an oscillation of 0.5° for each pack consisting of two films. A total of 86 A/B pairs were recorded. The films were digitized on a Joyce Loebl film scanner and processed using the oscillation program suite from Purdue University. The final R-factor for the data set was 10.93% for all data with I/?(I) > 1.0. The film and area detector data were merged to form a combined data set. For this, the merging R-factor was 28% for all measurements with I/?(I) > 1.0. (The definition of R is such that it gives twice the value that is obtained using the conventional definition of the R-factor.)

The final data set consists of 347,196 unique measurements with I/?(I) > 1.0.

Rotation function studies had established the similarity between SBMV-cp and SMV even before the relationship between them was demonstrated by serology. A polyalanine model of SBMV was used for phasing SMV reflection data by molecular replacement (MR). Phases that were refined to 4.7 Å were the starting point for phase extension to 2.9 Å. The protocol for phase refinement and averaging was modified such that it avoided the double sorting required in Bricogne’s procedure. A total of 168 cycles of averaging and solvent flattening were performed. MR computations converged to give good statistics. The final averaging R-factor was 14.1%, and the correlation coefficient was 0.9267.

The electron density map, calculated for the icosahedral asymmetric unit comprising the chemically identical A, B, and C subunits, revealed continuous density for the polypeptide backbone beginning from residue 65 in the A and B subunits and residue 39 in the C subunit. A model was built into the map using FRODO. For this, SBMV coordinates for the A, B, and C subunits were used as the starting point. Side chains for 94% of the residues could be built. Strong electron density peaks at sites identified to contain calcium ions in SBMV were observed here also. Additionally, a strong peak on the quasi-threefold has been identified as a putative calcium.

Refinement of the model was performed using XPLOR. The R-factor for the initial model, which consisted of 4223 non-hydrogen atoms, was 43%. Only atoms of the icosahedral asymmetric unit were refined independently. The strict non-crystallographic symmetry (NCS) option was used to impose 20-fold NCS. The R-factor for the final model, which contained 4519 non-hydrogen atoms, was 22.7% for all measurements between 10-3 Å with I/?(I) > 1.0. The RMS deviation in bond lengths and angles from ideal values were 0.014 Å and 1.77°, respectively.

The final model revealed a polypeptide fold that is very similar to that of SBMV except in loops and regions close to the calcium on the quasi-threefold (not found in SBMV). The polypeptide consists of the R domain (residues 1-65) at the N-terminus and the S domain (residues 66-260), which is a canonical 8-stranded anti-parallel ?-barrel found in most viral coat proteins. Compared to SBMV, SMV has fewer positively charged residues in the R domain, and protein-RNA interactions seem to be mediated differently in the two viruses.

The structure of SMV is a basis for future studies on the assembly/disassembly of the virus, which will provide a better understanding of its biological properties. Refinement of the structure to a resolution better than 3 Å will provide details of the water structure in the virus. Preliminary experiments on calcium removal have shown unusual asymmetry in their removal. Suitable experiments can be designed to study the expanded state of the virus and the structure of the isolated protein subunits, which might shed light on the disassembly mechanism of the virus.

The work reported in this thesis has been published in:

Bhuvaneshwari M, Subramanya HS, Gopinath K, Savithri HS, Nayudu MV, Murthy MRN Structure of Sesbania Mosaic Virus at 3 Å Resolution. Structure (1995) Vol 3, No. 10: 1021-1030.

Gopinath K, Sundareshan S, Bhuvaneshwari M, Karande A, Murthy MRN, Nayudu MV, Savithri HS Primary Structure of Sesbania Mosaic Virus Coat Protein: Its Implications for the Assembly and Architecture of the Virus. Indian Journal of Biochemistry and Biophysics (1994) 31: 322-328