Structural changes of envelope proteins during alphavirus fusion

Journal name:
Nature
Volume:
468,
Pages:
705–708
Date published:
DOI:
doi:10.1038/nature09546
Received
Accepted
Published online

Alphaviruses are enveloped RNA viruses that have a diameter of about 700Å and can be lethal human pathogens1. Entry of virus into host cells by endocytosis is controlled by two envelope glycoproteins, E1 and E2. The E2–E1 heterodimers form 80 trimeric spikes on the icosahedral virus surface1, 2, 60 with quasi-three-fold symmetry and 20 coincident with the icosahedral three-fold axes arranged with T = 4 quasi-symmetry. The E1 glycoprotein has a hydrophobic fusion loop at one end and is responsible for membrane fusion3, 4. The E2 protein is responsible for receptor binding5, 6 and protects the fusion loop at neutral pH. The lower pH in the endosome induces the virions to undergo an irreversible conformational change in which E2 and E1 dissociate and E1 forms homotrimers, triggering fusion of the viral membrane with the endosomal membrane and then releasing the viral genome into the cytoplasm3, 4. Here we report the structure of an alphavirus spike, crystallized at low pH, representing an intermediate in the fusion process and clarifying the maturation process. The trimer of E2–E1 in the crystal structure is similar to the spikes in the neutral pH virus except that the E2 middle region is disordered, exposing the fusion loop. The amino- and carboxy-terminal domains of E2 each form immunoglobulin-like folds, consistent with the receptor attachment properties of E2.

At a glance

Figures

  1. The structural proteins of an alphavirus.
    Figure 1: The structural proteins of an alphavirus.

    a, The cryo-EM density of Sindbis virus showing T = 4 symmetry. The four E2 molecules in one asymmetric unit (outlined in black) are coloured red, green, blue and yellow. These give rise to one trimeric spike on each icosahedral three-fold axis and one generally positioned spike. The E1 molecules are coloured grey. b, Threading of the Sindbis virus structural polyprotein through an endoplasmic reticulum membrane showing the position of the capsid, E3, E2, 6K and E1 proteins.

  2. Stereo diagrams showing the trimeric spike structure.
    Figure 2: Stereo diagrams showing the trimeric spike structure.

    a, The E1 molecule in a Sindbis virus spike (blue) compared with the E1 molecules in the crystal structure (red). b, Linear representation of polypeptides showing domains A (DA; cyan), B (DB; green), C (DC; pink) and the β-ribbon connector (purple) in E2; as well as DI (red), DII (yellow) and DIII (blue) in E1. c, Crystal structure of the trimeric spike at low pH. Domain B is disordered.

  3. The E2-E1 heterodimer.
    Figure 3: The E2–E1 heterodimer.

    a, The crystal structure (left) colour coded as in Fig. 2b. b, Comparison of the earlier E2 mapping8 with the E2 crystal structure. Amino acid sequence numbers are given in strategic positions8. The lipid envelope is shown diagrammatically. Modified from ref. 8 with permission.

  4. Cartoon showing maturation and fusion.
    Figure 4: Cartoon showing maturation and fusion.

    ad, Colour coded as in Fig. 2b. a, The viral spike while being transported through the Golgi and trans-Golgi network (TGN). Domain B is attached to DII because E3 (grey) binds the β-ribbon connector to domain A and holds E2 in place at low pH. b, The viral spike on the virus after E3 has been proteolytically cleaved and, in Sindbis virus, has been jettisoned. c, After the virus has recognized a host cell and entered into a low pH endosomal vesicle, domain B becomes unattached from DII, exposing the fusion loop (marked with an asterisk). The dashed line indicates the disordered β-ribbon connector. d, The trimeric spike disassembles, allowing escape of E2 and formation of E1 homotrimers. ER, endoplasmic reticulum.

Main

The X-ray crystal structure of the ectodomain of the E1 protein (residues 1–383) from Semliki Forest virus (SFV) is homologous to the flavivirus E glycoprotein and consists of three β-barrel domains (domains I, II and III; notated as DI, DII and DIII) with the fusion loop at the distal end of DII7. The structure of the E1 ectodomain had been fitted into the 9-Å-resolution cryo-electron microscopy (cryo-EM) reconstruction of Sindbis virus (Fig. 1a), generating a partial structure of the virus8, 9. After subtraction of the density representing E1, the E2 density was found to be a long, thin volume that covers the top of each E1 molecule including the fusion loop8. However, the crystal structure of E2 has remained unknown until now.

Figure 1: The structural proteins of an alphavirus.
The structural proteins of an alphavirus.

a, The cryo-EM density of Sindbis virus showing T = 4 symmetry. The four E2 molecules in one asymmetric unit (outlined in black) are coloured red, green, blue and yellow. These give rise to one trimeric spike on each icosahedral three-fold axis and one generally positioned spike. The E1 molecules are coloured grey. b, Threading of the Sindbis virus structural polyprotein through an endoplasmic reticulum membrane showing the position of the capsid, E3, E2, 6K and E1 proteins.

An E2–E1 recombinant protein of Sindbis virus, in which the ectodomains of E2 and E1 were connected by a flexible Strep-tag linker10 (Fig. 1b and Supplementary Fig. 1), was expressed in Drosophila Schneider 2 (S2) cells. Size-exclusion chromatography showed that the purified protein existed in solution as trimers of the E2–E1 heterodimer over a pH range from 5.5 to 9.5. The protein was crystallized at pH 5.6, which is lower than the pH 6.0 fusion threshold for alphaviruses4, 11, 12. The resultant crystal structure consisted of trimers of E2–E1 heterodimers that were remarkably similar to the trimeric spikes in the virus (Fig. 2 and Supplementary Table 1), demonstrating the biological significance of the crystallized recombinant E2–E1 protein.

Figure 2: Stereo diagrams showing the trimeric spike structure.
Stereo diagrams showing the trimeric spike structure.

a, The E1 molecule in a Sindbis virus spike (blue) compared with the E1 molecules in the crystal structure (red). b, Linear representation of polypeptides showing domains A (DA; cyan), B (DB; green), C (DC; pink) and the β-ribbon connector (purple) in E2; as well as DI (red), DII (yellow) and DIII (blue) in E1. c, Crystal structure of the trimeric spike at low pH. Domain B is disordered.

The Cα backbone of E2 corresponded well with an earlier tracing obtained by connecting known markers such as glycolsylation and antibody-binding sites8 (Fig. 3). The structure of E2 consists of the N-terminal domain A (residues 1–132), the middle domain B and the C-terminal domain C (residues 264–343). The ~88 residues of domain B are mostly disordered and are connected to domains A and C by long connecting linker peptides (the ‘β-ribbon connector’). The connecting peptide from domain A to domain B starts at residue 133 and could be traced to residue 166. The connecting peptide from domain B to domain C picks up at residue 255 and continues to residue 263 where it enters domain C (Supplementary Fig. 2). The three domains of E2 are stretched out along the length of E1 in the order C, A, B, with C being closest to the viral membrane and mostly hidden from the viral exterior. Domain B, had it not been disordered, would correspond to the tip of the cryo-EM envelope (Fig. 3b). The glue between the three E1 molecules that constitute a spike is formed by E2 domain C, which binds to DII in adjacent E1 molecules within the trimeric spike (Fig. 2c and Supplementary Fig. 3a). The residues in the contact area are primarily hydrophilic making a number of potential hydrogen bonds (Supplementary Table 2.1). In contrast to the low pH, partially disordered structure described here, the fully ordered structure of E2 has been determined at neutral pH for Chikungunya virus in the accompanying paper13.

Figure 3: The E2–E1 heterodimer.
The E2-E1 heterodimer.

a, The crystal structure (left) colour coded as in Fig. 2b. b, Comparison of the earlier E2 mapping8 with the E2 crystal structure. Amino acid sequence numbers are given in strategic positions8. The lipid envelope is shown diagrammatically. Modified from ref. 8 with permission.

Both domain A and domain C have the topology of an immunoglobulin fold (Fig. 3). This is consistent with E2 functioning as a cell-receptor-binding protein. Furthermore, residues that had been identified in E2 as being associated with altered receptor binding and tropism are now seen to be in domain A of E2 (Supplementary Table 3 and Supplementary Fig. 4). However, other residues that were associated with cell recognition6, 14, 15 are in the disordered domain B showing that there could be several sites on the virus surface that associate with various cell surface molecules involved in virus attachment and entry.

The A domains of the three E2 molecules within one trimeric spike are situated in the centre of the triangular cavity formed by the three E1 molecules in the spike and make extensive interactions with each other (Supplementary Table 2.2). The presence of histidines, arginines and lysines in the interface, although not conserved among alphaviruses, shows that the interactions will become weak or repulsive as the pH drops below the pK of the histidines. Furthermore, the extensive positive charge on these surfaces indicates that the A domains might be easily separated and then alter their orientation to present their arginines and lysines to the phospholipid headgroups in the membrane.

If the envelope of the E2 density derived from the 9-Å-resolution cryo-EM map8 is overlaid onto the E2–E1 trimeric low pH crystal structure by superimposing the E1 molecules, then a large proportion of the tip of the E2 cryo-EM envelope, presumably corresponding to the B domain, overlaps and clashes with a neighbouring trimeric spike assembly in the crystal structure (Supplementary Fig. 5a). This is shown even more markedly by superimposing the neutral pH crystal structure of the Chikungunya virus E2–E1 heterodimer as found in the accompanying paper13 (Supplementary Fig. 5b) onto the low pH crystal structure reported in this paper. (The root mean squared deviation (r.m.s.d.) between Cα atoms in the ordered part of E2 in these two structures is 4.5Å, a reasonable result considering that only 36% of the residues are identical in the compared region.) Thus, domain B as found in the virus cannot be in the same position in the crystal structure, partly consistent with previously reported findings16. The β-ribbon-connector peptides that extend from domains A and C follow the edge of domain A and are within the E2 envelope observed in the cryo-EM map. Thus, the ‘disorder’ of domain B (which is stabilized by two disulphide bonds13) might be because the long β-ribbon connector is flexible and allows domain B to detach itself from the tip of E2. The β-ribbon connector is flanked by two completely conserved histidines at Sindbis positions 169 and 256. These might be the acid-sensitive triggers that perhaps control the interaction with E3 (see below) and cause the β-ribbon connector to dissociate itself from the A domain. In summary, whereas the position of domain B in the neutral pH virus covers the fusion loop on E1, protecting the virus from premature fusion with other cellular membranes, acidification of the virus causes domain B to move away from its neutral pH position, thereby exposing the fusion loop and making it accessible to lipid membranes (Fig. 4). Antibodies that bind to epitopes on domain A or B are strong neutralizers17, perhaps because these antibodies might prevent not only receptor attachment but also prevent disordering of domain B and, hence, inhibit fusion.

Figure 4: Cartoon showing maturation and fusion.
Cartoon showing maturation and fusion.

ad, Colour coded as in Fig. 2b. a, The viral spike while being transported through the Golgi and trans-Golgi network (TGN). Domain B is attached to DII because E3 (grey) binds the β-ribbon connector to domain A and holds E2 in place at low pH. b, The viral spike on the virus after E3 has been proteolytically cleaved and, in Sindbis virus, has been jettisoned. c, After the virus has recognized a host cell and entered into a low pH endosomal vesicle, domain B becomes unattached from DII, exposing the fusion loop (marked with an asterisk). The dashed line indicates the disordered β-ribbon connector. d, The trimeric spike disassembles, allowing escape of E2 and formation of E1 homotrimers. ER, endoplasmic reticulum.

During maturation, E1 and PE2 (the precursor to the E3–E2 proteins before cleavage of E3, also referred to as p62) are assembled as heterodimers in the endoplasmic reticulum1, 18 and processed through the Golgi and trans-Golgi network, where E3 protects the E2–E1 heterodimer from premature fusion with cellular membranes. The resultant E2–E1 heterodimers, or E3–E2–E1 in some viruses19, are then transported to the plasma membrane20 where budding occurs following association of the glycoproteins with preformed nucleocapsids.

The site of E3 has been mapped by various cryo-EM studies21, 22, 23 that can now be identified as corresponding to a site where domain A borders onto domain B (Fig. 4 and Supplementary Fig. 3b). This indicates that the function of E3 is to stabilize the β-ribbon connector. As long as the distal end of the β-ribbon connector remains associated with domain A and DII, domain B will continue to protect the fusion loop. This is supported by mutational studies of H169 in Sindbis virus and its equivalent residue in SFV24, 25.

Fusion of the viral membrane with the cellular plasma membrane in alphaviruses requires the dissociation of the E1–E2 heterodimer and formation of E1 homotrimers at low pH26 in the presence of a lipid membrane. In the E1 homotrimeric structure the E1 molecules are arranged with their long direction running roughly parallel to a common three-fold axis of rotation, exposing the fusion loops of each monomer at one end of the trimeric complex27. Residues of E1 that were in contact with E2 in the pre-fusion trimeric spike (in part determined from the structure given in the accompanying paper13) are located on the surface of the post-fusion E1 homotrimer, facing away from the spike axes, and are predominantly in surface loops8 (Supplementary Fig. 6). The conformational changes required to form a post-fusion trimer will require the removal of the E2 molecules from the centre of the trimeric spike to the outside and rotation of the E1 molecules by about 180° about their long axes to face each other across what was the three-fold axis of the pre-fusion spike. E2 might escape contact with the surrounding E1 molecules by detaching the C domains and slipping to the outside via the gap left by them. Alternatively, two E1 molecules on neighbouring spikes might combine with a third E1 molecule recruited from another unmatched E1 molecule elsewhere on the virus to make the post-fusion trimer (Fig. 4). Similar, huge conformational changes occur in the maturation and fusion of flaviviruses28, 29; however, it is still unclear what the pathway is by which these events happen.

Methods

Protein production

The coding sequences of E3 (residues 1–64), E2 (residues 1–344) and E1 (residues 1–384) of Sindbis virus (Toto64) were amplified by PCR31. The DNA sequence of a peptide linker, STREP, that contains the Strep-tag II10 (GGGSWSHPQFEKGGGG) was inserted between E2 and E1 by overlapping PCR, thereby omitting the transmembrane and 6K region between E2 and E1. The recombinant gene that codes for E3–E2 (residues 1–344) STREP-E1 (residues 1–384) was inserted between the BglII and XhoI restriction sites in the pMT/BiP/V5-His A vector (Invitrogen). A (His)6 tag and a stop codon was introduced right after the 3′ end of the E1 gene by the 3′ PCR primer, so the V5 and His tags from the original vector were omitted from the expression product. Ten micrograms of the resultant plasmid were co-transfected with 0.5μg of the plasmid pCoHygro (Invitrogen) into 3ml of Drosophila S2 cells at 2–4×106 cellsml−1 by using 30μl of cellfectin (Invitrogen) in Schneider’s Drosophila medium (no serum) (Invitrogen). The stable cell line expressing the recombinant protein was selected out after one month in the presence of 300μgml−1 hygromycin B (EMD Chemicals) in Schneider’s Drosophila medium supplemented with 10% heat inactivated fetal bovine serum. Stably transfected cells were gradually scaled up and adapted into EX-CELL 420 serum-free medium (Sigma-Aldrich). Protein expression was induced by the addition of 500μM CuSO4 after 2l of the cells, maintained in two 3-l spinner flasks, had grown to a density of 6–10×106cellsml−1. The recombinant protein was secreted into the medium as the E2–E1 complex because E3 was cleaved by furin and the furin-like proteases during secretion in the S2 cells. The medium was harvested 3–5 days after induction. The cells were spun down and the supernatant medium was collected and clarified by passing through 0.22-μm cutoff membrane filters (Millipore). The medium was then loaded onto a column packed with 18ml IMAC sepharose 6 ff resin (GE Healthcare), pre-charged with Ni2+ (ref. 32). The protein was eluted with 100mM and then 500mM immidazole using a step-elution protocol. The partially purified protein was dialyzed against a phosphate buffer (20mM NaPO4, 50mM NaCl, pH 6.9) at 4°C overnight. The dialyzed protein solution was loaded onto a 1-ml Hitrap Q and then onto a 1-ml Hitrap SP column (GE Healthcare). The protein was eluted from the SP column with a 50–500mM NaCl gradient. Solubility of the protein was 1–2mgml−1 in neutral and low pH buffers. However, the protein could be concentrated to 5–10mgml−1 when the pH of the buffer was raised to 9.5 or higher. Thus, the protein solution was dialyzed against a CHES (N-cyclohexyl-2-aminoethanesulphonic acid) buffer (20mM CHES, 200mM NaCl, pH 9.5) and the protein was purified by passing through a gel filtration column (Superdex 200, GE Healthcare). The purified protein was concentrated to 4–5mgml−1 in the CHES buffer for crystallization.

To produce the selenomethionine (SeMet)-substituted protein, the medium was replaced with ESF 921 serum-free methionine-free medium (Expression Systems) after 2l of cells had reached a concentration of 10–12×106cellsml−1. The cells were starved for 4–6h before 400mgl−1 L-SeMet (Acros Organics) and 500μM CuSO4 were added. The medium was harvested 2–3 days after induction. The purification was similar to that of the native protein. Amino acid analysis showed that 90% of native methionine in the protein had been substituted by SeMet.

Crystallization and diffraction data collection

Crystals of the native E2–E1 protein grew at 20°C by using the hanging-drop method. 1.5μl of the protein solution and 1.5μl of the mother solution were mixed and hung over 500μl of the mother solution containing 11% polyethylene glycol (PEG) 8000, 200–275mM Na/K tartrate, 0.1M Na/KPO4 pH 5.6. Crystals of the SeMet protein grew with the mother solution containing 9–10% PEG 8000, 300–375mM Na/K tartrate, 0.1M Na/KPO4 pH 5.6. The crystals appeared in 3 days and grew to full size in 2–4 weeks. The crystals were soaked in the mother liquid plus 20% PEG 400 as a cryo-solvent, followed by flash-freezing in liquid nitrogen. Hundreds of crystals were screened at the Advanced Photon Source (APS, Argonne National Laboratory). The limits of the diffraction were gradually improved from 8-Å to 4-Å resolution. In general, the SeMet protein crystals diffracted better than the native protein crystals. The first complete data set of a native crystal extended to only 4-Å resolution. The best data sets were obtained from the SeMet protein crystals pre-treated with the mother liquid plus 20% PEG 400 and 1% H2O2 for 60s before freezing33. A multi-wavelength anomalous diffraction (MAD) data set with a resolution limit of 3.3Å was collected at the APS beamline 23 ID-B (Supplementary Table 1).

Structure determination

The data were processed using the HKL2000 program34. The output intensity files were converted to structure factor files for the CCP4 program package35. For the first native data with the resolution of 4Å, the space group was determined to be P321 with one E2–E1 heterodimer in an asymmetric unit. The pre- and post-fusion structures of the SFV E1 protein (Protein Data Bank accession numbers 2ALA and 1RER) were used to generate search models for molecular replacement using the program Phaser36. A solution was found when domains I and II (DI and DII) from the pre-fusion E1 structure were used as a search model. However, the quality of the resultant electron density map was not good enough to interpret the E2 molecule. A single wavelength anomalous dispersion (SAD) data set of a SeMet crystal was collected with a resolution of 3.7Å and space group P321. Ten out of the eleven selenium (Se) sites in the protein were found by using the program SHELXD37. Later, a three-wavelength MAD data set was collected with the peak wavelength data having a resolution of 3.3Å. The space group was determined to be P1 with six E2–E1 heterodimers in an asymmetric unit and cell dimensions similar to the P321 space group. The crystallographic 3, 2 and 21 axes present in the P321 space group were now only approximate non-crystallographic symmetry (NCS) axes indicating that the overall molecular packing remained similar to that of the P321 space group. The initial Se sites were generated by using all the Se sites in one unit cell of the SAD data collected for the space group P321. The sites were refined in the program SHARP38. The quality of the resultant map was greatly improved by six-fold NCS averaging and other density improvement methods implemented in the programs DM39 and RESOLVE40. DI and DII of E1 and domains A and C of E2 could be traced in the density map. A barrel-shape density was evident for each of the six independent DIII domains of E1, although the individual chains could not be traced. Therefore, the DIII model of SFV was placed in the density using a real space search procedure with the program ESSENS41, followed by rigid body refinement with the program REFMAC542. The E2–E1 model was built by using the program COOT43 with the guidance from the (2FoFc), (FoFc), omit, and B-factor sharpened (2FoFc) maps. Structure refinement was performed using the programs PHENIX44 and REFMAC542 with tight NCS restraints and TLS refinement45 (Supplementary Fig. 7 and Supplementary Table 1). The electron density for DIII was recognizable, but the mean B factor of the main chain atoms of DIII was 270Å2 as opposed to 120Å2 for the main chain atoms in DI and DII. The structure of E2, lying roughly parallel to E1, was as expected. Each spike structure consisted of a three-start right-handed helix with a maximum diameter of about 78Å.

The crystallographic 32 symmetry in the trigonal space group (or quasi-32 symmetry in the triclinic space group) produced two two-fold-related trimers of E2–E1 heterodimers. In the triclinic cell the crystallographic three- and two-fold axes were only approximate because of a slight rotation and displacement of each of the two spikes in the unit cell. Superposition of the crystal structure onto the cryo-EM structure8 gave an r.m.s.d. of 3.4Å between equivalent Cα atoms in D1 and D2.

Fitting of the crystal structure into the cryo-EM density

The crystal structure of the E2–E1 heterodimer was fitted into the 9-Å resolution cryo-EM map of Sindbis virus8 (Electron Microscopy Data Bank accession number 1121) using the program EMfit30, and gave a fit that was at least as good as the independent fitting of three SFV structures of E1 into the Sindbis virus cryo-EM density8. DI, DII and DIII of E1, domain A, the β-ribbon connector and domain C of E2 were fitted as individual rigid bodies for each of the four quasi-equivalent positions (Supplementary Table 4).

The DI–DII component of E1 had behaved as a rigid body in previous structural analyses of alphavirus particles8, 9, but the angle between DIII and DI–DII was found to be variable. In the crystal structure reported here the angle between DI–DII and DIII has changed by about 9° relative to the angles observed in the neutral pH cryo-EM structure of Sindbis virus or 18° relative to the crystal structure of SFV E1.

Domain A was found to be rotated (~10°) and translated (~5Å) slightly relative to the crystal structure in order to obtain the best fit into the cryo-EM density. This small conformational change of the A domain within the trimeric spike may be the result of crystal packing where a carbohydrate moiety associated with Asn318 in E2 interacts with domain A in a neighbouring spike. As the structure of the B domain has two disulphide bonds13, it is likely to be a rigid body. Attempts were made using the computer program EMfit30 to determine alternative positions of the B domain in the uninterpreted regions of density tentatively interpreted as solvent. A number of likely positions were found that had the N and C termini of domain B within 20Å of the C and N discontinued ends of the β-ribbon connecter, respectively. Figures displaying the structure were generated with the program Chimera46.

Accession codes

Primary accessions

Protein Data Bank

References

  1. Strauss, J. H. & Strauss, E. G. The alphaviruses: gene expression, replication, and evolution. Microbiol. Rev. 58, 491562 (1994)
  2. von Bonsdorff, C. H. & Harrison, S. C. Sindbis virus glycoproteins form a regular icosahedral surface lattice. J. Virol. 16, 141145 (1975)
  3. Kielian, M. Membrane fusion and the alphavirus life cycle. Adv. Virus Res. 45, 113151 (1995)
  4. White, J. & Helenius, A. pH-dependent fusion between the Semliki Forest virus membrane and liposomes. Proc. Natl Acad. Sci. USA 77, 32733277 (1980)
  5. Dubuisson, J. & Rice, C. M. Sindbis virus attachment: isolation and characterization of mutants with impaired binding to vertebrate cells. J. Virol. 67, 33633374 (1993)
  6. Strauss, E. G., Stec, D. S., Schmaljohn, A. L. & Strauss, J. H. Identification of antigenically important domains in the glycoproteins of Sindbis virus by analysis of antibody escape variants. J. Virol. 65, 46544664 (1991)
  7. Lescar, J. et al. The fusion glycoprotein shell of Semliki Forest Virus: an icosahedral assembly primed for fusogenic activation at endosomal pH. Cell 105, 137148 (2001)
  8. Mukhopadhyay, S. et al. Mapping the structure and function of the E1 and E2 glycoproteins in alphaviruses. Structure 14, 6373 (2006)
  9. Zhang, W. et al. Placement of the structural proteins in Sindbis virus. J. Virol. 76, 1164511658 (2002)
  10. Schmidt, T. G. & Skerra, A. The Strep-tag system for one-step purification and high-affinity detection or capturing of proteins. Nature Protocols 2, 15281535 (2007)
  11. Smit, J. M., Bittman, R. & Wilschut, J. Low-pH-dependent fusion of Sindbis virus with receptor-free cholesterol- and sphingolipid-containing liposomes. J. Virol. 73, 84768484 (1999)
  12. White, J., Kartenbeck, J. & Helenius, A. Fusion of Semliki Forest virus with the plasma membrane can be induced by low pH. J. Cell Biol. 87, 264272 (1980)
  13. Voss, J. E. et al. Glycoprotein organization of Chikungunya virus particles revealed by X-ray crystallography. Nature doi:10.1038/nature09555 (this issue)
  14. Meyer, W. J. & Johnston, R. E. Structural rearrangement of infecting Sindbis virions at the cell surface: mapping of newly accessible epitopes. J. Virol. 67, 51175125 (1993)
  15. Zhang, W., Heil, M., Kuhn, R. J. & Baker, T. S. Heparin binding sites on Ross River virus revealed by electron cryo-microscopy. Virology 332, 511518 (2005)
  16. Wu, S.-R. et al. The dynamic envelope of a fusion class II virus. Prefusion stages of Semliki Forest virus revealed by electron cryomicroscopy. J. Biol. Chem. 282, 67526762 (2007)
  17. Griffin, D. E. Roles and reactivities of antibodies to alphaviruses. Semin. Virol. 6, 249255 (1995)
  18. Wahlberg, J. M., Boere, W. A. & Garoff, H. The heterodimeric association between the membrane proteins of Semliki Forest virus changes its sensitivity to low pH during virus maturation. J. Virol. 63, 49914997 (1989)
  19. Ziemiecki, A. & Garoff, H. Subunit composition of the membrane glycoprotein complex of Semliki Forest virus. J. Mol. Biol. 122, 259269 (1978)
  20. Ziemiecki, A., Garoff, H. & Simons, K. Formation of the Semliki Forest virus membrane glycoprotein complexes in the infected cell. J. Gen. Virol. 50, 111123 (1980)
  21. Ferlenghi, I. et al. The first step: activation of the Semliki Forest virus spike protein precursor causes a localized conformational change in the trimeric spike. J. Mol. Biol. 283, 7181 (1998)
  22. Paredes, A. M. et al. Structural localization of the E3 glycoprotein in attenuated Sindbis virus mutants. J. Virol. 72, 15341541 (1998)
  23. Wu, S. R., Haag, L., Sjoberg, M., Garoff, H. & Hammar, L. The dynamic envelope of a fusion class II virus. E3 domain of glycoprotein E2 precursor in Semliki Forest virus provides a unique contact with the fusion protein E1. J. Biol. Chem. 283, 2645226460 (2008)
  24. Zhang, X. & Kielian, M. Mutations that promote furin-independent growth of Semliki Forest virus affect p62–E1 interactions and membrane fusion. Virology 327, 287296 (2004)
  25. Heidner, H. W., McKnight, K. L., Davis, N. L. & Johnston, R. E. Lethality of PE2 incorporation into Sindbis virus can be suppressed by second-site mutations in E3 and E2. J. Virol. 68, 26832692 (1994)
  26. Wahlberg, J. M. & Garoff, H. Membrane fusion process of Semliki Forest virus. I. Low pH-induced rearrangement in spike protein quaternary structure precedes virus penetration into cells. J. Cell Biol. 116, 339348 (1992)
  27. Gibbons, D. L. et al. Conformational change and protein–protein interactions of the fusion protein of Semliki Forest virus. Nature 427, 320325 (2004)
  28. Li, L. et al. The flavivirus precursor membrane-envelope protein complex: structure and maturation. Science 319, 18301834 (2008)
  29. Yu, I. M. et al. Structure of the immature dengue virus at low pH primes proteolytic maturation. Science 319, 18341837 (2008)
  30. Rossmann, M. G., Bernal, R. & Pletnev, S. V. Combining electron microscopic with X-ray crystallographic structures. J. Struct. Biol. 136, 190200 (2001)
  31. Owen, K. E. & Kuhn, R. J. Identification of a region in the Sindbis virus nucleocapsid protein that is involved in specificity of RNA encapsidation. J. Virol. 70, 27572763 (1996)
  32. Lehr, R. V., Elefante, L. C., Kikly, K. K., O'Brien, S. P. & Kirkpatrick, R. B. A modified metal-ion affinity chromatography procedure for the purification of histidine-tagged recombinant proteins expressed in Drosophila S2 cells. Protein Expr. Purif. 19, 362368 (2000)
  33. Sharff, A. J., Koronakis, E., Luisi, B. & Koronakis, V. Oxidation of selenomethionine: some MADness in the method!. Acta Crystallogr. D 56, 785788 (2000)
  34. Otwinowski, Z. & Minor, W. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307326 (1997)
  35. Collaborative Computational Project Number 4 The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D 50, 760763 (1994)
  36. McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Cryst. 40, 658674 (2007)
  37. Sheldrick, G. M. A short history of SHELX. Acta Crystallogr. A 64, 112122 (2008)
  38. de La Fortelle, E. & Bricogne, G. Maximum-likelihood heavy-atom parameter refinement for multiple isomorphous replacement and multiwavelength anomalous diffraction methods. Methods Enzymol. 276, 472494 (1997)
  39. Cowtan, K. D. ‘dm’: an automated procedure for phase improvement by density modification. Joint CCP4 ESF-EACBM Newsl. Protein Crystallogr. 31, 3438 (1994)
  40. Terwilliger, T. C. Maximum-likelihood density modification. Acta Crystallogr. D 56, 965972 (2000)
  41. Kleywegt, G. J. & Jones, T. A. Template convolution to enhance or detect structural features in macromolecular electron-density maps. Acta Crystallogr. D 53, 179185 (1997)
  42. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D 53, 240255 (1997)
  43. Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D 60, 21262132 (2004)
  44. Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D 66, 213221 (2010)
  45. Winn, M. D., Isupov, M. N. & Murshudov, G. N. Use of TLS parameters to model anisotropic displacements in macromolecular refinement. Acta Crystallogr. D 57, 122133 (2001)
  46. Pettersen, E. F. et al. UCSF Chimera—a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 16051612 (2004)

Download references

Acknowledgements

We wish to thank S. Sun, A. Aksyuk and T. Edwards for discussions. We are also grateful to S. Kelly for help in the preparation of the manuscript. We thank F. Rey for sharing the coordinates of Chikungunya virus E2–E1 to help interpret the Sindbis virus cryo-EM density of the E2 domain B. We would like to thank the staff at the Advanced Photon Source, Argonne National Laboratory, GM/CA sector for their help in data collection. The work was supported by NIH grant P01 AI055672 to R.J.K. and M.G.R.

Author information

Affiliations

  1. Department of Biological Sciences, Purdue University, 915 W. State Street, West Lafayette, Indiana 47907-2054, USA

    • Long Li,
    • Joyce Jose,
    • Ye Xiang,
    • Richard J. Kuhn &
    • Michael G. Rossmann

Contributions

L.L. designed the expression constructs, J.J. cloned the constructs, and L.L. and J.J. developed the expression system and purified the protein. L.L. crystallized the protein, collected X-ray diffraction data and, with Y.X., determined the structure. L.L., R.J.K. and M.G.R. discussed the results and wrote the paper.

Competing financial interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to:

The atomic coordinates of the E2–E1 heterodimer crystal structures have been deposited with the Protein Data Bank (accession number 3MUU). The fit of the E2–E1 heterodimer into the cryo-EM reconstruction of Sindbis virus has been deposited with the Protein Data Bank (accession number 3MUW).

Author details

Supplementary information

PDF files

  1. Supplementary Information (598K)

    The file contains Supplementary Tables 1-4, additional references and Supplementary Figures 1-7 with legends.

Additional data