| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 282, Issue 14, 10678-10689, April 6, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
11
2
From the
Architecture et Fonction des Macromolécules Biologiques, CNRS, and Universités d'Aix-Marseille I et II, UMR 6098, ESIL Case 925, 13288 Marseille, France, the Department of Microbiology, Monash University, Clayton, Victoria 3168, Australia, ¶Global Phasing Ltd., Sheraton House, Cambridge CB3 0AX, United Kingdom, the ||School of Molecular and Microbial Sciences, University of Queensland, Brisbane, Queensland 4072, Australia, and the **Department of Cellular and Molecular Medicine, University of Bristol, Bristol BS8 1TD, United Kingdom
Received for publication, August 1, 2006 , and in revised form, January 29, 2007.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTS AND DISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTS AND DISCUSSIONREFERENCES |
|---|
The Flavivirus positive sense RNA genome contains a single open reading frame encoding a polyprotein that is processed into three structural and seven non-structural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5). Signature-sequence analysis suggests that the non-structural protein NS5 is comprised of two domains. The DENV NS5 30-kDa N-terminal domain has been shown to possess RNA cap (nucleoside-2'-O) methyltransferase (MTase) activity, and its crystal structure has been determined (6). That of WNV has been shown to mediate both (guanine-N7) and (nucleoside-2'-O) methylation of the 5'-cap structure (7). The C-terminal domain of about 70 kDa harbors the RdRp (RNA-dependent RNA polymerase), as initially identified by the presence of signature-sequence motifs A to F (8, 9), and subsequently confirmed in RdRp activity assays (10-12).
The Flaviviridae RdRp is essential to viral replication. The RdRp duplicates the single-stranded RNA genome during a single, continuous polymerization event. The RdRp enters at the 3'-end of the genome and is able to copy the whole RNA molecule in a primer-independent fashion (12-17) referred to as de novo RNA synthesis. Crystal structures of RdRps from 
9 virus species have been reported, among them those of Hepaci- and Pestivirus RdRps (18-22), but none from a Flavivirus species.
Very little is known about the mechanism of flaviviral RNA capping, or its co-ordination with RNA synthesis. In eukaryotic cells, mRNA capping is a nuclear event, whereas Flavivirus RNA synthesis is thought to occur in the cytoplasm of the infected cell. The viral NS3 and NS5 proteins have been demonstrated to possess RNA-triphosphatase and 2'-O-/N7-MTase activities, respectively (6, 7, 23-26), however, the guanylyltransferase activity required for a complete capping reaction is still uncharacterized.
Interestingly, during infection, a proportion of the NS5 and, more recently, of the NS3 protein of a number of flaviviruses has been detected in the nucleus (DENV) or at least the perinuclear region (WNV and Japanese encephalitis virus) of infected cells (27-29). Trafficking of the DENV NS5 to the nucleus was proposed to rely on a nuclear import pathway based on the identification of functional nuclear localization sequences (NLSs) in NS5 and their interaction in vitro with distinct members of the karyopherin family of intracellular transport proteins (karyopherin-
/
, also known as importin-/ or Imp/) (30, 31). NLSs are small motifs composed of one or more clusters of basic amino acids that do not conform to a specific sequence consensus. Two adjacent regions encompassing distinct NLSs have been defined in DENV NS5 termed the bNLS (aa 320 to 368) and the a/bNLS (aa 369 to 405) (30, 31). The bNLS and a/bNLS bind with high affinity to Imp and an Imp/ heterodimer in vitro, respectively, and both are capable of targeting -galactosidase to the nucleus in the context of a fusion protein (30, 31). However, whether the bNLS and a/bNLS are required for nuclear import of the intact NS5 protein during viral infection is not known. The bNLS of DENV NS5 has also been shown to interact with NS3 and competition between Imp and NS3 binding has been confirmed using pulldown assays (30).
Here, we report the crystal structures of enzymatically active and inactive WNV RdRp domains. They reveal a classic RdRp fold bearing palm, thumb, and finger domains decorated with Flavivirus-specific attributes such as a priming loop adopting a different fold. Structural elements in the WNV RdRp structure corresponding to putative DENV NLSs were found to be integral parts of the RdRp domain. Interestingly, despite sequence conservation in the NLSs of DENV and WNV NS5 the latter was not found to localize to the nucleus. To gain insight into the overall spatial organization of the full-length NS5 protein, a reverse genetic approach was used to identify amino acids in the DENV MTase and RdRp domains, which potentially interact and could be used to position one domain relative to the other. The structure of the WNV MTase domain was modeled using DENV MTase atomic coordinates. Both WNV MTase and RdRp domains were assembled in silico using docking procedures, generating a model of full-length WNV NS5 protein.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTS AND DISCUSSIONREFERENCES |
|---|
-D-thiogalactopyranoside, and further incubated 16-18 h at 17 °C. Cells were harvested by centrifugation. The cell pellet was resuspended in 50 mM Tris buffer, pH 8.0, containing 150 mM NaCl, 10 mM imidazole, DNase I (2 µg/ml), a protease inhibitor tablet (Sigma), and sonicated on ice. The sample was centrifuged, the supernatant collected and filtered through a 0.22-µm filter. The sample was applied to a 5-ml bed volume HiTrap nickel immobilized metal ion affinity chromatography column (Amersham Biosciences) connected to a FPLC system (Amersham Biosciences). The protein was eluted with 50 mM Tris buffer, pH 8.0, containing 150 mM NaCl and 500 mM imidazole. Protein-containing fractions were then applied onto a preparative Superdex 200 gel filtration column pre-equilibrated in 10 mM Tris buffer, pH 9.0, with 300 mM NaCl and 5% glycerol. Protein was concentrated to 6 (POL1) and 9 mg/ml (POL2 and POL3) using a Vivaspin 30-kDa molecular mass cut-off centrifugal concentrator (Vivascience). A selenomethionine-substituted protein was used for structure determination. It was expressed according to standard conditions of the methionine-biosynthesis pathway inhibition (32) and purified following the same procedure as that for the native protein.
Crystallization—Initial crystallization trials were set up with a nano-drop dispenser in 96-well sitting drop plates using commercial crystallization kits and purified protein at 6 mg/ml for POL1 and 9 mg/ml for POL2. Initial hits were further optimized by the hanging drop vapor diffusion method in Linbro plates by mixing 1 µl of protein solution with 1 µl of reservoir solution. Crystals of POL1 were grown from 10% PEG 1000, 0.2 M sodium cacodylate, 0.3 M MgCl2, pH 7.0. POL2 was crystallized using two conditions (i) 5% PEG 10,000, 0.2 M imidazole malate, pH 8.0, and (ii) 10% PEG 8000, 0.1 M imidazole pH 7.0, and 0.2 M calcium acetate.
Crystals were briefly soaked in a cryo-protectant solution composed of mother liquor supplemented with 36% glycerol and flash frozen in liquid nitrogen. WNV POL1 and POL2 diffraction intensities were recorded on different beamlines (Table 1) at the European Synchrotron Radiation Facility (Grenoble, France). Integration, scaling, and merging of the intensities were carried out using programs from the CCP4 suite (33) and statistics are provided in Table 1.
|
RdRp Activity Tests—RdRp activity tests were performed in 50-µl reactions containing 50 mM HEPES, pH 8.0, 5 mM dithiothreitol, 10 mM KCl, 5 mM MnCl2, 5 mM MgCl2, 10 µM GTP (0.01 µCi of [3H]GTP per µl, 6.1 Ci/mmol, Amersham Biosciences) using 1 µM poly(rC) (Amersham Biosciences) and WNV RdRp (264 nM POL1, 760 nM POL2, and 800 nM POL3). Reactions were started with a premix of MnCl2, MgCl2, and GTP and incubated at 30 °C. Samples of 12 µl were taken after 5, 10, and/or 15 min and the reaction stopped by adding 30 µl of 50 mM EDTA in 96-well plates. Samples were then transferred to glass fiber filter mats with DEAE active groups (DEAE filter mat, Wallac) using a Filtermat Harvester (Packard Instruments). Filtermats were washed three times with 0.3 M ammonium formate, pH 8.0, twice with water, once with ethanol, dried, and the filter transferred into sample bags. Liquid scin-fluid fluid was added and incorporation was measured in counts per minute using a Wallac MicroBeta TriLux Liquid Scintillation Counter.
Introduction of NS5 Mutations into a Genomic Length DENV cDNA Clone—Mutations encoding changes in NS5 amino acids were introduced into the genomic length DENV serotype 2 strain New Guinea C (DENV-2) cDNA clone pDVWS601, which yields the virus v601 (43, 44). Mutations were initially introduced into subgenomic PCR fragments by overlap extension PCR using mutagenic primers (details available from the authors upon request) and then transferred into pDVWS601 as follows. The mutation K46A/R47A/E49A (each NS5 amino acid targeted for mutagenesis is numbered followed by the substituted amino acid) was engineered into a 1001-bp overlap extension PCR fragment (DENV-2 nucleotides 7165-8165), which was HpaI7406/StuI7874 digested and introduced into the corresponding sites of pDVWS601 to produce pDVWS601-NS5K46A,R47A,E49A. The NS5 mutation L512V was engineered into a 2077-bp fragment (DENV-2 nucleotides 8085-10161), which was AatII8570/MluI9732 digested and introduced into the corresponding sites of pDVWS601 and pDVWS601-NS5K46A,R47A,E49A to produce pDVWS601-NS5L512V and pDVWS601-NS5K46A,R47A,E49A,L512V, respectively.
Recombinant DENV Recovery, Growth, and Sequencing—Procedures for the recovery of infectious DENV-2 from pDVWS601 including transcription of RNA, electroporation of BHK-21 cells, and passaging and plaque assay of virus in Aedes albopictus C6/36 cells have been described previously (43, 44). Confirmation of the mutations in the recovered viruses and the identification of second site mutations were performed by sequencing the entire DENV-2 genome. Viral RNA was extracted from the culture supernatants of infected cells using a QIAamp Viral RNA Extraction kit (Qiagen) and used for the production of six overlapping 2-kilobase pair RT-PCR products spanning the DENV-2 genome using DENV-2-specific primers and the SuperScript One-Step RT-PCR with Platinum Taq System (Invitrogen). Each RT-PCR product was then purified using a Qiagen PCR Purification Kit and used for automated DNA sequencing using DENV-2-specific primers.
Molecular Modeling and Docking—Alignment of the MTase sequences of DENV-2 (PDB code 1L9K [PDB] ) and WNV was performed using SPDBV (45) (sequences share 59% identity and 71% similarity). The homology model of WNV MTase was built using the SWISS-MODEL server (45) and stereochemistry was verified using WHATCHECK (46). The docking of the MTase and the POL1 domain was carried out with the protein-protein docking algorithm FTDOCK (47), based on shape complementarities (using a grid spacing of 1 Å) and electrostatic treatment. A search over the complete binding space for both molecules was performed. Results were filtered using the distance restraints between putatively interacting residues as determined by reverse genetic experiments (6 Å filter) and according to the relative positions of the MTase C terminus and POL1 N terminus (45 Å filter for the 13 missing residues), which led to a single solution.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTS AND DISCUSSIONREFERENCES |
|---|
1to 9 from aa 274 to 498, and 12 to 3 from aa 542 to 609), palm (10 and 11 from aa 499 to 541, and 14 to 17 from aa 610 to 717), and thumb (6to 25 from aa 718 to 905) domains (Figs. 1A, 2B, and supplemental Fig. 1S). The most closely related structures are those of RdRps from members of the flaviviridae family, the hepacivirus hepatitis C virus (HCV) RdRp (18, 19, 21), and the pestivirus bovine viral diarrhea virus (BVDV) RdRp (20), which have structural identities as low as 11-21%, depending on the domain considered (Table 2).
|
|
|
The Palm Domain—The architecture of the palm domain is a highly conserved feature among all RdRps structurally characterized so far. The WNV POL is no exception, with a r.m.s. deviation (on topologically equivalent C atoms of the palm domain) of 2.14 and 1.97 Å relative to the BVDV (PDB code 1S49) (20) and HCV RdRps (Fig. 1B, PDB code 1NB6) (50), respectively. The palm consists of a central anti-parallel -sheet (4 and 5) flanked at either side by one (11) and five -helices (10 and 14 to 17) (Fig. 1A).
A major difference between the WNV POL palm domain and those of other RdRps is that the -sheet consists of only 2 -strands instead of 3, as the five residues preceding motif A are not organized in a typical -strand. Moreover, these 2 -strands are very short (10 Å) compared with those of HCV and BVDV (20 Å). In addition, for WNV POL, an extra loop/-helix (15) (aa 629 to aa 663) is found in WNV POL in an unusually long insertion connecting motif B to C (Fig. 1A and supplemental Fig. 1S). This insertion is either absent in the HCV RdRp or quite variable in sequence, length, and conformation in all other RdRp structures. In WNV POL, it packs against an -helix (5) located at the base of the fingers (Fig. 1A).
|
-phosphate. The second promotes a suitable geometry for the catalytic reaction. In structurally characterized RdRps, aspartic acids corresponding to Asp536 and Asp668 are known to bind two metal ions involved in catalysis. Our search for crystallization conditions suitable for x-ray analysis led to three different buffers. The first contained no divalent ion (native data set), the second 0.3 M MgCl2 ("magnesium" data set), and the third 0.2 M calcium acetate (calcium data set). The crystal structure of the WNV POL is devoid of metal ions in catalytic positions. Nevertheless, in the last two conditions only, a metal ion (magnesium and calcium, respectively) could clearly be identified in the same position in the electron density maps. This position is 5.0 and 7.1 Å away from the expected positions of the two catalytic metal ions. Due to the distance from the active site, it is unlikely that it participates in the catalytic reaction (Fig. 3). It is coordinated by Asp536 and Asp669; and consequently, Asp536 points away from the catalytic position. A similar geometry for a non-catalytic ion and the aspartate equivalent to Asp536, with concomitant lack of catalytic metal ions, has been described for other RdRps (51-54). The role of this non-catalytic ion is still elusive. The non-catalytic ion was proposed to be present in an inactive open conformation of rabbit hemorrhagic disease virus RdRp, which affects the active site conformation (52). Nevertheless, WNV POL corresponds to the closed, active conformation of rabbit hemorrhagic disease virus RdRp, which does not have an ion in the non-catalytic position (52) as judged by superimposition of both structures (not shown). Additionally, in Phi6 RdRp (PDB code 1HI0
[PDB]
), which adopts also a closed, active conformation, the presence of a non-catalytic ion was shown to be compatible with the binding of ions in the catalytic positions and the formation of an initiation complex. In this case the residue corresponding to Asp536 rotates toward the catalytic ion position and the residue equivalent to Asp669 coordinates the non-catalytic ion. It was proposed that the non-catalytic ion plays a role in the de novo initiation mechanism facilitating the movement of the nascent dsRNA after formation of the first dinucleotide out of the active site (48). Further studies have to be undertaken to elucidate the role of the non-catalytic ion for WNV RdRp but we suppose that for the conformation of the WNV POL1 active site observed here, a simple change of the Asp536 side chain orientation might allow binding of catalytic ions, building of the ternary complex, and RNA synthesis.
The Fingers Domain—Comparison of the fingers domain of WNV POL1 (aa 274 to 498 and 542 to 609) to the corresponding domains of the HCV and BVDV RdRps revealed major differences. First, there is an extra N-terminal stretch (aa 274 to 318), which is comprised of an -helix (1) and a -strand (1) (Fig. 1A in yellow).
Second, the loop connecting -helices 6 and 7 (aa 407 to 420) is longer and adopts a conformation distinct from that of all RdRps structurally characterized so far (Fig. 1A, in orange). In primer-dependent RdRps, this loop corresponds to the conserved motif G located at the entrance of the RNA templatebinding groove (55), and establishes contacts with the template (53). In HCV and BVDV RdRps, the position of this loop is relatively conserved although the motif G is not. In WNV POL, motif G is not conserved either but, in contrast, the corresponding loop takes an entirely different direction protruding toward the active site and the thumb domain priming loop (see below). Thus it seems to contribute to closure of the active site. Interestingly, its tip is partially disordered (aa 410-415 could not be modeled due to lack of electron density), which suggests conformational flexibility. Its location on the structure seems to coincide with the position of the C termini of the BVDV and HCV RdRps. The C terminus of HCV RdRp has been shown to play a regulatory role for RdRp activity (56, 57). It prevents RNA binding and possibly also the binding of incoming NTPs. The loop connecting -helices 6 and 7 might play a similar regulatory role in Flavivirus RdRps.
Third, the motif F fingertip loop corresponding to a motif named F2 (58) between motifs F1 and F3 (see supplemental Fig. 1S) is partially disordered in the WNV POL (aa 459-470 are missing in the density map). Surprisingly, the ordered part (aa 455-458 and 471-476) does not adopt a -sheet structure as in all other RdRp structures, but F3 appears as an -helix. The structure POL1 clearly indicates that the structural elements corresponding to motif F run perpendicular to the motif F -sheet structure observed in other RdRps, thus parallel to the fingers -sheet (in WNV POL1 2 and 3 combined with 1of the N-terminal extension) and exposed to solvent (Fig. 2B). How this unusual conformation relates to the replication mechanism of Flavivirus RdRp remains to be elucidated. The basic residues of motif F, especially F2 and F3, have been shown to provide a binding site (named the interrogation site I) for incoming NTPs entering through the tunnel from the back of the molecule (18, 20, 21, 48, 51, 53, 59). The residues involved in NTP binding are conserved in motif F of WNV POL. Should they play the same role during Flavivirus RNA synthesis, a conformational change would be necessary to create the binding site in WNV POL. Thus it is expected that upon formation of the ternary initiation complex, a major conformational change of the structural elements constituting motif F of the Flavivirus RdRp will take place.
|
The Thumb Domain—The thumb domain encompasses the C-terminal 187 aa of WNV POL (aa 718-905). Structural motif E, which consists of two anti-parallel -strands (6 and 7, aa 714-736), forms the interface between the thumb and the palm domain (Fig. 1A). The rest of the thumb consists of eight -helices connected by large loops. Its overall topology is very different from that of the HCV RdRp thumb domain, and is more related to that of BVDV RdRp. The five N-terminal -helices of the WNV POL thumb domain can be superimposed on the corresponding -helices of BVDV RdRp and the connectivity between them is conserved. However, the length, conformation, and orientation of the structural elements connecting these -helices are very different. Moreover, the three C-terminal -helices of WNV POL cannot be superimposed onto the corresponding -helices of either the HCV or BVDV RdRps, as there is a 45 degree difference in their orientation. The overall structural diversity of the thumb domain is even more salient when the WNV POL is compared with primer-dependent RNA polymerases, like Picornavirus and Calicivirus RdRps, which have a much smaller thumb domain (52, 53, 59, 60). To date, two structural elements have been correlated with the difference between primer-dependent and de novo RNA synthesis by RdRps: (i) the presence of a characteristic loop (also called the priming loop, -hairpin in HCV, -thumb region in BVDV, and critical loop region in Phi6 RdRps) and (ii) the size, overall fold, and location of the C-terminal segment. For example, in the HCV RdRp, the 40 residues upstream of the hydrophobic C-terminal 21-amino acid stretch fold back toward the active site cavity and the RNA binding groove constituting a regulator for RNA synthesis (56, 57). Several residues within this stretch contact the -hairpin loop, which serves as a priming platform for initiation and prevents back-primed initiation (61). The situation is similar for BVDV NS5B, even though the structure of the C-terminal upstream stretch has not been fully modeled. In WNV POL, the C-terminal six amino acids are not defined in either of the two space groups in which the crystal structures of POL1 and POL2 have been solved. However, the last modeled amino acid, Leu899, is more than 40 Å away from the active site, clearly indicating that the C terminus of WNV POL cannot fold back into the active site. Moreover, there is no difference in RdRp activity between WNV POL1 (aa 273-905) and a 23-aa C-terminal truncated form (POL3, aa 273-882) (Fig. 4A), confirming that the C-terminal extremity of Flaviviruses does not play a regulatory role in the polymerization process, as it is the case for Pesti- and Hepaciviruses. Our structure proposes that the element that might overtake this function in Flavivirus RdRps may be the loop connecting -helices 6 and 7 (see above).
|
The Importance of the N-terminal 44 Amino Acids for Activity—Polymerase activity assays demonstrated that POL2, in contrast to POL1, is unable to incorporate [3H]GMP using the homopolymeric RNA template poly(rC) (Fig. 4A). The two structures are, however, very similar with a r.m.s. deviation on equivalent C atoms as low as 1.16 Å (even if POL1 is in slightly more closed conformation than POL2 (see Supplemental Material)). There are no differences in the orientation of the residues within the active site between POL2 and POL1. The major difference between the two structures is the presence of 44 additional amino acids at the N terminus of POL1. Interestingly, these 44 residues adopt a fold not found in other RdRp structures, except that of the foot-and-mouth disease virus RdRp (53). The additional residues induce the formation of a three-stranded -sheet (made of strands 1, 2, and 3, see Figs. 2 and 4B). The presence of this ordered three-stranded -sheet might stabilize the whole "back" subdomain of the fingers domain. Indeed, 70 amino acids of this subdomain, disordered in POL2, are clearly visible in the POL1 electron density map (aa 317-321, aa 339-361, aa 416-419, aa 453-458, aa 471-472, aa 577-602, and aa 748-751). This back subdomain closes the groove accommodating the incoming template RNA and so contributes to the formation of the RNA template tunnel, whereas in POL2, this groove might remain open due to partial disorder (Fig. 4B).
The Priming Loop—Several structural and biochemical studies on HCV, BVDV, and Phi6 RdRps have demonstrated that the priming loop constitutes a platform stabilizing the RNA synthesis initiation complex (22, 48, 53, 56, 57, 62). Conformational changes, not yet characterized by structural studies, are presumably required to shift to an RNA synthesis elongation mode. Comparative enzymatic characterization of WNV and HCV RdRps suggests that a similar conformational change occurs for WNV POL, even though some structural differences could account for the different processivity of the two RdRps (12). The WNV POL structure reveals that the priming loop consists of aa 796-809 (Fig. 1A in purple and Fig. 2 in yellow), which was difficult to infer from sequence alignment. It originates from the same part of the thumb domain as for the BVDV and HCV RdRps (it connects -helices 20 to 21) but is larger. Its overall size and positioning is most similar to that of Phi6 RdRp, whereas its folding is unique, bearing no secondary structural elements (Figs. 1A, 2B, and 5). It is stabilized by hydrogen bonds with side chains belonging to residues located in -helices 11 (Gln515), 14 (Tyr610), and 19 (Lys761, Tyr763, and Gln765), and by van der Waals interactions with side chains belonging to residues in -helices 18 (Arg742) and 19 (Arg754 and Met766). Most of these residues are conserved in the DENV, Japanese encephalitis virus, and yellow fever virus RdRps. In the priming loops of the Phi6, HCV and BVDV RdRps, a tyrosine side chain (Tyr630 in Phi6, Tyr448 in HCV, and Tyr581 in BVDV) is used to orient the base of the priming nucleotide via a stacking interaction. In WNV POL, there is no equivalent tyrosine. However, two conserved residues Trp800 and His803, which may perform similar function, are found in close proximity.
A Model for Initiation of RNA Synthesis—To characterize the initiation complex of WNV POL, many attempts of co-crystallization with short sequence-specific RNAs, oligonucleotides, and divalent ions were made. As these attempts were unsuccessful, we used the similarity with the HCV and Phi6 RdRps to propose a polymerization initiation model (Fig. 5B). rNTPs were positioned at both the priming (site P) and catalytic (site C) sites according to the HCV (PDB code 1GX5 [PDB] ) and Phi6 (PDB code 1HI0 [PDB] ) models. The interrogation site I, missing in WNV POL, is sufficiently far away from the active site, such that a meaningful initiation complex could be built. It became obvious that a simple change in the Trp800 side chain conformation resulted in the ideal positioning of its aromatic base to stack against the priming nucleotide, thereby playing a role similar to Phi6 Tyr630 and HCV Tyr448 (Fig. 5A). The side chain conformation of Asp536, the position of the two catalytic magnesium ions, and the position of a 4-mer RNA corresponding to the 3' end of the (-)-strand genomic RNA of WNV, was inferred from the initiation complex structure of Phi6 RdRp. No major steric clash between the RNA, the nucleotides modeled, and the solved structure was observed. The resulting model, after minimization, is shown in Fig. 5B. In conclusion, residue Trp800, which is strictly conserved in Flavivirus NS5 proteins, seems to be the key residue in the priming loop providing the initiation platform for de novo initiation of RNA synthesis by Flavivirus RdRps.
|
Although WNV NS5 was not nuclear localized, the structure of WNV POL reveals the fold and localization of the regions corresponding to the DENV NLSs on the WNV POL surface (Fig. 6B). An equivalent structural scaffold can be found in both BVDV and HCV RdRps (aa 132-217 of BVDV and aa 20-94 of HCV), with a well conserved overall shape and secondary structure, but not charge. The DENV bNLS has been shown to interact with the 300 C-terminal amino acids of NS3 (30). No precise mapping of this interaction is available to date. However, assuming that the corresponding region of WNV NS5 (situated on the top of the back of the polymerase as shown in Figs. 6B and 7C) interacts with NS3, this NS5-NS3 interaction may facilitate localization of NS3 close to the entrance of the RNA template tunnel, consistent with the putative role of its helicase activity in dsRNA unwinding during RNA synthesis.
|
Using a DENV-2 infectious cDNA clone, we identified a genetic interaction between the MTase and RdRp domains of DENV-2 NS5. RNA transcribed from the infectious DENV-2 clone yields infectious virus (v601) when introduced into permissive cells. Based on the crystal structure of the DENV-2 MTase domain, a series of mutations designed to disrupt surface interactions involving charged residues was introduced into the ns5 gene. In particular, amino acids Lys46, Arg47, and Glu49 were chosen as they are charged and surface exposed on the MTase. Mutation of these three charged amino acids to alanine could induce a non-complementary charge between the POL and MTase, disrupting domain interactions, and inhibiting virus growth and/or infectivity. RNA transcripts encoding the K46A/R47A/E49A mutation in the DENV-2 MTase domain failed to routinely produce infectious virus when transfected into permissive cells. However, after repeated attempts, virus was twice recovered from cells transfected with the mutant transcripts. As RT-PCR and sequencing confirmed the presence of the K46A/R47A/E49A mutation in each of the recovered viruses, the entire viral genomes were sequenced to determine whether a second site mutation had arisen that compensated the effects of the K46A/R47A/E49A mutation. Both viruses contained a Leu to Val substitution at surface residue 512 (L512V). The L512V mutation was then tested for its ability to compensate for the effects of the initially introduced K46A/R47A/E49A mutation. Mutations encoding either L512V alone, or in combination with K46A/R47A/E49A, were introduced into the DENV-2 genome. As before, virus was not recovered from BHK-21 cells transfected with RNA transcripts containing the K46A/R47A/E49A mutation. By contrast, recombinant viruses were recovered from both the parental RNA transcripts and transcripts containing the mutations L512V (vNS5L512V) and K46A/R47A/E49A/L512V (vNS5K46A,R47A,E49A,L512V). The entire genome of all viruses was sequenced, confirming that only the expected mutations were present. Growth curve analysis of the viruses in Vero cells (Fig. 7A) demonstrated that vNS5K46A,R47A,E49A,L512V and vNS5L512V had similar replication kinetics and peak viral titers to v601. Overall, the presence of mutation L512V either alone, or in combination with mutation K46A/R47A/E49A appeared to maintain or even slightly increase virus replicative ability compared with the parental v601. These results suggest that L512V, in the RdRp domain, acts to compensate for the K46A/R47A/E49A mutation, either directly (i.e. V512 contacts the K46A/R47A/E49A cluster) or indirectly (V512 restores functionality by a long-range conformational change) in the MTase domain. Our model of full-length NS5 is based on the first hypothesis.
A model of the WNV MTase domain was generated by homology modeling based on the structure of the DENV MTase (PDB code 1L9K
[PDB]
, 59% identity). Fig. 7B shows the WNV MTase domain with bound co-product S-adenosylhomocysteine, the RNA substrate binding groove, and the site where the RNA cap is bound (with a bound GTP analogue) during methyltransfer at the 2'-O-position of the first nucleotide. Subsequently, the entire binding space for both the MTase and POL1 domains was screened. A 6-Å filter between the amino acids corresponding to DENV NS5 512 in the POL1 domain and DENV NS5 46, 47, and 49 in the MTase domain was applied. A second filter based on restraints on the distance between the C-terminal amino acid of the MTase domain and the N-terminal amino acid of the POL1 domain reduced the number of solutions from 6 to 1. In the corresponding model (Fig. 7C), the C terminus of the MTase domain is 36 Å away from the N terminus of the POL1 domain, implying that the 13-residue linker between both domains (from aa 264 of the MTase domain to aa 278 of the POL domain) is expected to adopt an extended, stretched conformation. The MTase "Lys46/Arg47/Glu49 loop" fits well into a large groove located between the thumb and the fingers domains (Fig. 7, C and D). The bottom of this groove is made of the short "Leu512 loop" of the palm domain, which connects -helices 10 to 11 and which neighbors the C and E motifs (Figs. 2B and 7D). This loop also delineates "the bottom" of the tunnel through which the newly synthesized dsRNA should come out from the active site (Fig. 2B).
From a structural perspective it is not possible to provide a precise explanation about the nature of the interaction between the K46A/R47A/E49A cluster and L512V in DENV NS5. It might be possible that Val512 protrudes less from the surface than Leu512, and Val512 allows the formation of a shallow hydrophobic surface region with the neighboring glycines. This region might then interact better with the three alanines of the MTase domain. Both the direct contact hypothesis and our model remain speculative. The possibility exists that L512V promotes an allosteric change compensating the alanine mutations. Thus, the model should be taken with caution as a preliminary guide in the current efforts to understand how RNA synthesis and RNA capping are coordinated. Nevertheless, it is interesting to note that the RNA-substrate binding groove of the MTase domain is positioned near the RdRp dsRNA exit tunnel defined above. The model seems to be structurally and mechanistically consistent with RNA cap methylation of the nascent genome occurring after the (+) RNA product leaves the RdRp domain.
Concluding Remarks—The Flaviviridae virus family comprises important and emerging human pathogens. The crystal structure of the WNV RdRp domain presented here reveals a number of novel Flavivirus-specific features in comparison to known structures of Hepacivirus and Pestivirus RdRps. It will constitute a basis for structure-based drug design studies against flaviviruses. Our model of full-length WNV NS5 represents a structural hypothesis for further investigation of the spatial organization of the Flavivirus replication complex.
|
FOOTNOTES |
|---|
* This work was supported by SPINE project Grant QLRT-2001-00988, VIZIER integrated project Grant LSHG-CT-2004-511960 of the European Union 6th Framework Programme (FP6), as well as the Conseil Régional de la Région Provence-Alpes-Côte d'Azur. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 
The on-line version of this article (available at http://www.jbc.org) contains supplemental material and Fig. 1S.
1 Both authors contributed equally to this work.
2 To whom correspondence should be addressed. Tel.: 33-491-82-86-44; Fax: 33-491-82-86-46; E-mail: bruno.canard{at}afmb.univ-mrs.fr.
3 The abbreviations used are: DENV, dengue virus; aa, amino acids; a/bNLS, nuclear localization sequence that recognize importin-; bNLS, nuclear localization sequence that recognize importin-; BVDV, bovine viral diarrhea virus; DENV-2, dengue virus type 2; ds, double-stranded; HCV, hepatitis C virus; Imp/, karyopherin-/ heterodimer, also known as importin-/; Imp, karyopherin-, also known as importin-; Imp, karyopherin-, also known as importin-; MTase, methyltransferase; NLS, nuclear localization sequence; NS3, non-structural protein 3; NS5, nonstructural protein 5; POL, West Nile virus polymerase; POL1, construct 273-905 of West Nile virus polymerase; POL2, construct 317-905 of West Nile virus polymerase (in complex with calcium ion); POL3, construct 273-882 of West Nile virus polymerase; Phi6, Pseudomonas phage Phi-6; RdRp, RNAdependent RNA polymerase; r.m.s., root mean square; RT, reverse transcriptase; WNV, West Nile virus.
4 B. Coutard, unpublished results.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTS AND DISCUSSIONREFERENCES |
|---|
), 760 nM POL2 (
), and 800 nM POL3 (
) were used. Absolute [3H]GMP incorporation in counts per minute for each enzyme was normalized to 100 nM enzyme concentration. B, top view (rotated by 90° when compared with 
