New insight into flavivirus maturation from structure/function studies of the yellow fever virus envelope protein complex.

IMPORTANCE: All enveloped viruses enter cells by fusing their envelope with a target cell membrane while avoiding premature fusion with membranes of the producer cell-the latter being particularly important for viruses that bud at internal membranes. Flaviviruses bud in the endoplasmic reticulum, are transported through the TGN to reach the external milieu, and enter other cells via receptor-mediated endocytosis. The trigger for membrane fusion is the acidic environment of early endosomes, which has a similar pH to the TGN of the producer cell. The viral particles therefore become activated to react to mildly acidic pH only after their release into the neutral pH extracellular environment. Our study shows that for yellow fever virus (YFV), the mechanism of activation involves actively knocking out the fusion brake (protein pr) through a localized conformational change of the envelope protein upon exposure to the neutral pH external environment. Our study has important implications for understanding the molecular mechanism of flavivirus fusion activation in general and points to an alternative way of interfering with this process as an antiviral treatment.

Molecular cloning, expression and purification of protein 2A of hepatitis A virus.

Expression of the protein 2A of Hepatitis A virus (HAV), spanning amino acids 764 through 981 of the viral polyprotein results in a strong inhibition of cap-dependent translation (Maltese et al., 2000). However, the molecular mechanism responsible has remained unclear, in part because the HAV 2A protein was not available in amounts large enough to allow biological or structural studies. To address this issue, a cDNA representation of the sequences encoding HAV 2A was generated by PCR, using primers that introduced an AUG triplet, and a sequence coding for 6 histidine residues at the 5'- and 3'-termini of the genomic sequence, respectively. The cDNA fragment was introduced by cassette exchange in the inducible expression vector pQE-60, and the construct was propagated in bacteria E. coli M15 which constitutively expresses the lac repressor. Upon induction with IPTG (1 mM), HAV 2A was visualized by SDS-PAGE of bacterial lysates as a prominent band M(r) = 21 kDa. The identity of the polypeptide was confirmed by both MALDI-TOF peptide mapping and direct amino acid sequencing. The His-tagged HAV 2A was extracted from bacterial pellets under totally denaturing conditions (6 M urea), subjected to Ni(++)-Sepharose affinity chromatography, allowed to refold while still attached to the matrix, and eluted with 250 mM Imidazole. Contaminant material was partly removed by differential ammonium sulfate precipitation. The protein was further concentrated (Vivaspin centrifugal concentrator), the insoluble material (if present) was discarded, and the homogeneity of the dispersion was ascertained by light scattering. SDS-PAGE revealed that in addition to the main protein (Mr = 21 kDa), a second one of apparent Mr = 14 kDa was always present in variable amounts. The proportion of the latter tended to increase with aging of the preparation. Edman degradation analysis proved that the 14 kDa protein resulted from the cleavage of HAV 2A at a so far undetected scissile bond Gly856/Val857 of the viral polyprotein. A first attempt to crystallize the protein by the hanging drop procedure yielded only small crystals containing exclusively the 14 kDa derivative of HAV 2A. Western blot analysis of HeLa cell extracts that had been incubated with the His-tagged HAV 2A so purified failed to reveal any change in the electrophoretic mobility of the eukaryotic initiation factor (eIF) 4G I.

X-ray structure and catalytic mechanism of lobster enolase.

Enolase prepared from lobster tail muscle yielded trigonal crystals with one 47 kDa subunit per asymmetric unit. X-ray data were collected on the apoenzyme at 2.4 A resolution and on a complex with Mn2+ and the inhibitor phosphoglycolate at 2.2 A resolution. The corresponding cDNA was amplified from a library of lobster muscle cDNA, and a sequence corresponding to residues 27-398 was determined. It is highly homologous to other enolases, including yeast enolase for which an X-ray structure is available. Yeast enolase was used as a starting point for crystallographic refinement, which led to models of lobster enolase having R-factors below 22% and good stereochemistry. These models are very similar to yeast enolase; they have the same fold with a beta 3 alpha 4 N-terminal domain followed by an atypical alpha/beta barrel. Lobster apoenolase and the ternary complex differ only in the position of three mobile loops. In the complex, a single Mn2+ ion is seen ligated to three carboxylates and three water molecules. Phosphoglycolate binds near, but not directly to, the metal. His 157, which belongs to one of the mobile loops, is in contact with the C2 atom of the ligand. A water molecule hydrogen-bonds to the carboxylate of the ligand and to those of Glu 166 and Glu 209. We suggest that His 157 is the base that abstracts the C2H proton, whereas the water molecule is part of a proton relay system keeping the substrate in the carboxylic acid form where the pKa of the C2H group is low enough for proton transfer to His 157. The resulting catalytic mechanism is different from those proposed on the basis of the yeast enzyme X-ray structures, but it fits with earlier biochemical and spectroscopic data.

Phasing with mercury at 1 A wavelength.

Synchrotron sources provide a continuously tunable X-ray beam which makes it possible to optimize the anomalous contribution to phase determination using heavy-atom replacement. This method was used to solve two protein structures, those of Dictyostelium discoideum nucleoside diphosphate kinase and of lobster enolase. The first had 17 kDa of protein in the asymmetric unit, the second, 47 kDa. In both cases, a single mercury derivative yielded single isomorphous replacement with anomalous-scattering phases from which an interpretable electron-density map was derived by solvent flattening. The efficient solution of the X-ray structure was largely due to the large anomalous scattering of mercury at a wavelength shorter than the L(III) absorption edge.

Lobster enolase crystallized by serendipity.

An unknown protein crystallized from a lobster muscle preparation in which arginine kinase was the majority component. It was identified as enolase by peptide sequencing and activity testing, and a SIRAS electron density map showed its three-dimensional structure to be very similar to that of yeast enolase.

Protein-protein recognition analyzed by docking simulation.

Antibody-lysozyme and protease-inhibitor complexes are reconstituted by docking lysozyme as a rigid body onto the combining site of the antibodies and the inhibitors onto the active site of the proteases. Simplified protein models with one sphere per residue are subjected to simulated annealing using a crude energy function where the attractive component is proportional to the interface area. The procedure finds clusters of orientations in which a steric fit between the two protein components is achieved over a large contact surface. With five out of six complexes, the native structure of the complexes determined by X-ray crystallography is among those retained. Docked complexes are then subjected to conformational energy refinement with full atomic detail. With Fab HyHEL 5 and lysozyme, a native-like complex has the lowest refined energy. It can also be retrieved when starting with the X-ray structure of free lysozyme. However, some non-native complexes cannot be rejected: they form large interfaces, have a large number of H-bonds, and few unpaired polar groups. While these are necessary features of protein-protein recognition, they are not sufficient in determining specificity.

Protein-protein interaction: an analysis by computer simulation.

A survey of protein-protein interactions in structures derived by X-ray crystallography of protease-inhibitor and antigen-antibody complexes shows that they form close-packed interfaces from which water is excluded. The interfaces are of almost constant size, and they contain about ten hydrogen bonds. These features account for the stability of the complexes. To test whether they also account for specificity, we designed a computer simulation that searches for complementary surfaces on two protein molecules. In all cases tested, the simulation finds a number of complexes having interfaces and hydrogen bonds equivalent to those of the native complexes. These artificial complexes might represent secondary specificities, which can be detected when normal association is prevented by mutation or other means.