Philip R. Dormitzer
Harvard Medical School and Children's Hospital, Boston
Abstract: (Note: references and images not provided)
Rotavirus is the
most important cause of dehydrating childhood gastroenteritis worldwide,
killing approximately 500,000 children each year. To enter cells, rotavirus
must translocate a 710 angstrom, non-enveloped particle across a cellular
membrane. The rotavirus spike protein, VP4, has a central role in this
membrane penetration event. Rotavirus is primed for entry when trypsin
cleaves VP4 into hemagglutinin (VP8*) and membrane interaction (VP5*)
regions. Crystallographic structure determination for the hemagglutinin
region was guided by NMR analysis of domain boundaries and ligand
specificity. Crystallization of the membrane interaction region required
protease triggering of a rigidifying conformational change. Structure
determination from crystals of the membrane interaction region was
complicated by perfect hemihedral twinning, masked by pseudocentering.
Combining high resolution structures with electron cryomicroscopy-based
molecular envelopes indicates that the part of VP4 that protrudes from the
virion undergoes a two-fold to three-fold reorganization during entry. This
reorganization is accompanied by a jack-knifing conformational change that
translocates a hydrophobic potential membrane interaction region from one
end of the protein to the other. This fold-back resembles the
rearrangements that mediate membrane fusion during enveloped virus entry.
Thus, structural studies combining a variety of techniques suggest common
elements in the entry mechanisms of enveloped and non-enveloped viruses.
Rotavirus. (Rotavirus biology and epidemiology are reviewed in reference [1].) Antibodies that neutralize the virus can protect from disease, provided that they are present in the gut lumen during infection. These protective antibodies bind the moving parts of the molecular apparatus that the virus uses to enter cells. Many neutralizing antibodies against rotavirus bind specific conformations of their targets. Therefore, a high resolution understanding of the structural basis for rotavirus entry could allow a structure-based approach to optimizing antigens for use in a second generation rotavirus vaccine.
Rotavirus entry is something of a puzzle. To initiate infection, the virus must translocate a large (approximately 710 Å diameter) subviral particle across a cell membrane. The delivered particle then acts as an mRNA factory, extruding messages through pores in its icosahedral surface. Some viruses (enveloped viruses) are surrounded by lipid bilayers. An enveloped virus enters cells by fusing its envelope with a cellular membrane. Proteins anchored in the viral envelope mediate this membrane fusion through a mechanism that involves a jack-knifing rearrangement. Rotavirus, however, does not have an envelope. It can not fuse with a cellular membrane, but instead must directly penetrate a host lipid bilayer to enter a cell. This process has yet to be understood in detail for any non-enveloped virus. Structural analysis, combining X-ray crystallography (laboratories of Philip Dormitzer and Stephen Harrison at Harvard Medical School), NMR spectroscopy (laboratory of Gerhard Wagner at Harvard Medical School), and electron cryomicroscopy (laboratories of Venkataram Prasad at Baylor College of Medicine and Mark Yeager at The Scripps Research Institute) has provided new insights into membrane penetration by rotavirus [2].
Biochemical and virologic studies show that the rotavirus spike protein, VP4, is a key component of the cell entry apparatus. Like some enveloped virus fusion proteins, VP4 is primed for activity when it is cleaved by a protease (Fig. 1). VP5*, the C-terminal fragment of VP4 produced by the priming trypsin cleavage contains a hydrophobic sequence (the "membrane interaction loop"), which resembles the "fusion loop" of an alphavirus fusion protein [3]. Fitting X-ray and NMR structures of the receptor binding domain of VP4 to electron cryomicroscopy image reconstructions of trypsin primed virions showed that, like some enveloped virus fusion proteins (such as influenza HA and HIV gp160), the rotavirus VP4 spike is tipped by receptor binding heads (Fig. 2, middle panel) [4]. Based on these similarities, we hypothesized that, like a fusion protein, rotavirus VP4 may undergo a jack-knifing rearrangement during membrane penetration.
VP4 is flexible prior to trypsin priming: uncleaved VP4 does not appear in averaged electron cryomicroscopy image reconstructions of rotavirus particles (Fig. 1, lower left) [5], and purified uncleaved VP4 has not crystallized. The priming trypsin cleavage rigidifies part of VP4, so that well-ordered spikes appear in the image reconstructions (Fig. 1, upper right) [5]. Proteolytic cleavage of recombinant VP4 triggers a rearrangement in its VP5* region, yielding a rigid (and crystallizable) oligomer [6]. Although the crystals of this cleavage fragment (VP5CT) diffract X-rays, structure determination proved unexpectedly challenging. Analysis was complicated by perfect hemihedral twinning, which was masked in intensity statistics by pseudocentering. A new statistic that allows the detection of twinning in the face of pseudocentering and anisotropy [7] allowed detection of the twinning and screening for crystals in which the difference between twin blocks remained insignificant to a usefully high resolution.
The VP5CT structure contained a surprise [2]. Although the protruding portion of the trypsin-primed VP4 spike on virions appears dimeric, VP5CT is a well-ordered, umbrella-shaped trimer (Fig. 2, left panel). It is held together by a triple-stranded coiled-coil, by a nine-stranded b-annulus, and by rings of hydrophobic contacts near its peak. Some enveloped virus fusion proteins, such as the dengue virus envelope glycoprotein, rearrange from dimers to trimers during cell entry [8]. In enveloped viruses, lateral diffusion of membrane anchored proteins allows for changes in oligomeric association. How can a dimer anchored on a non-enveloped virus particle rearrange to a trimer? Either the rearrangement occurs after VP4 is shed from the virion during entry, or three VP4 molecules must be anchored in each cluster in the outermost capsid of the virion. Neither possibility has been definitively demonstrated. However, electron cryomicroscopy image reconstructions of the VP4 spike show that, although the protruding part of the spike is an asymmetrical dimer, the part of the spike that anchors into the virion has three-fold (or perhaps even 6-fold) symmetry [9]. This observation suggests that three VP4 molecules may be anchored in each cluster on the virion. In this case, all three are flexible prior to trypsin priming; two of the three associate to form the rigid spike; and all three then join to form the umbrella-shaped trimer after an unknown trigger.
The globular domain that forms each shade of the umbrella-shaped trimer (excluding N- and C-terminal regions that make three-fold contacts; Fig. 2, right panel) neatly fits the electron cryomicroscopy envelope of the body of the VP4 spikes (excluding the volumes that form two-fold contacts; Fig. 2, middle panel) [2]. The orientation of the globular domain in the spike is confirmed by electron cryomicroscopy image reconstructions of rotavirus virions decorated with a VP4-specific FAb with a known binding site (not shown [10]). In the globular domain, the potential membrane interaction loop joins two other loops to form a hydrophobic apex [2]. In the dimeric spike, this hydrophobic prominence points away from the virion surface and is masked by the receptor binding heads. In the trimer, the hydrophobic prominence points towards the base of the umbrella (and towards the virion surface, if the trimer forms on the virion). Therefore, during the two-fold to three-fold rearrangement of the protruding part of the VP4 spikes, the molecule must jack-knife, with the globular domain folding back against the post of the umbrella. This domain rotation translocates the hydrophobic prominence by at least 55 Å towards the base.
Structural analysis of the rearrangements of enveloped virus fusion proteins is complemented by a rich set of functional data, which allow an increasingly detailed understanding of cell entry by enveloped viruses. Currently, our knowledge of structural rearrangements in the membrane penetration apparatus of rotavirus is more detailed than our functional understanding of rotavirus entry. The fold-back rearrangement inferred for the VP5* fragment of rotavirus VP4 is highly suggestive of a mechanism for disrupting a cellular membrane. We now aim to establish the links between these structural snapshots and discrete steps in rotavirus membrane penetration during cell entry.
2008 Run
Nov 19th - Dec 22nd
