The HIV-1 protein Rev controls a critical step in viral replication by mediating the nuclear export of unspliced and singly-spliced viral mRNAs. Multiple Rev subunits assemble on the Rev Response Element (RRE), a structured region present in these RNAs, and direct their export through the Crm1 pathway. Rev-RRE assembly occurs via several Rev oligomerization and RNA-binding steps, but how these steps are coordinated to form an export–competent complex is unclear. Here, we report the first crystal structure of a Rev dimer-RRE complex, revealing a dramatic rearrangement of the Rev-dimer upon RRE binding through re-packing of its hydrophobic protein–protein interface. Rev-RNA recognition relies on sequence-specific contacts at the well-characterized IIB site and local RNA architecture at the second site. The structure supports a model in which the RRE utilizes the inherent plasticity of Rev subunit interfaces to guide the formation of a functional complex.https://doi.org/10.7554/eLife.04120.001
To be able to multiply, viruses have to first infect a host cell and then hijack the host's molecular machinery to make viral proteins. One stage of this process takes place in the nucleus of the host cell and involves the viral DNA being transcribed to make RNA molecules. These RNA molecules must then be exported from the nucleus to the cytoplasm, where the viral proteins are made.
In the case of HIV-1, a protein called Rev has an important role in the export process. The Rev protein, which is supplied by the virus, binds to a region on the viral RNA molecules called the Rev Response Element. The Rev protein then binds to a group of host proteins called the Crm1 export complex to send the viral RNA molecules to the cytoplasm.
Jayaraman et al. now provide the first in-depth 3D structure of two Rev molecules bound to a fragment of the Rev Response Element. The Rev molecules change shape when they bind to the element, and specific regions of the element were found to be important for this. The experiments suggest that the Rev Response Element directs the positioning of the Rev proteins on itself to match the shape needed to bind to Crm1 export complex. In parallel work from the same laboratory, Booth et al. have produced a 3D structure of the whole complex.
Both structures shed new light on how the HIV-1 virus is able to multiply in its host, which may aid future efforts to develop new treatments for the disease.https://doi.org/10.7554/eLife.04120.002
Retroviruses such as HIV have small genomes and utilize multiple strategies such as overlapping reading frames and alternative splicing, to encode their repertoire of proteins. The ∼9 kb HIV-1 genome codes for 15 proteins expressed from a set of unspliced and singly-spliced mRNAs in addition to the fully spliced messages (Frankel and Young, 1998). These unspliced and singly-spliced species encode the viral structural and accessory proteins and genomic RNA needed to assemble new virions, but because they contain introns are typically retained in the nucleus for splicing. To export these RNAs to the cytoplasm, the viral protein Rev (Figure 1A), expressed from a fully spliced message, translocates into the nucleus, forms an oligomeric complex on a ∼350 nucleotide, highly structured intronic RNA element, the Rev Response Element (RRE) (Figure 1B), and directs their export through the Crm1 nuclear export pathway (Fornerod et al., 1997; Pollard and Malim, 1998).
Rev cooperatively assembles on the RNA using its oligomerization and RNA-binding domains to form a Rev hexamer on a shorter, but functional ∼240 nucleotide RRE (Malim and Cullen, 1991; Zapp et al., 1991; Daugherty et al., 2008, 2010a). This assembled RNP then presents Leu-rich nuclear export sequence (NES) peptides (Figure 1A) to interact with the host export receptor, Crm1, targeting these RRE-containing mRNAs for export (Fornerod et al., 1997). In this paper we address the structural principles that govern assembly of the Rev oligomer onto the RNA and its consequences for Rev function.
The binding of Rev to the RRE is nucleated at the stem IIB hairpin, where the α-helical arginine-rich motif (ARM) of Rev (Figure 1A,B) binds in the major groove of the RNA and makes an extensive set of sequence-specific and electrostatic contacts (Malim et al., 1990; Cook et al., 1991; Heaphy et al., 1991; Kjems et al., 1991; Battiste et al., 1996). The oligomeric complex proceeds to form by the sequential addition of Rev subunits (Pond et al., 2009). Crystal structures of Rev oligomerization surfaces depict a central hydrophobic core that mediates interactions between Rev subunits and positions the ARMs on one side of the Rev oligomer to bind RNA and the NESs on the other side to recruit Crm1 (DiMattia et al., 2010; Daugherty et al., 2010b), highlighting the modular nature of the Rev domains.
In addition to the modular architecture of Rev, the RNA scaffold also plays a key role in defining the organization of the complex. Biochemical studies have shown that the RRE controls the oligomeric state of Rev (Daugherty et al., 2010a), and a recent SAXS structure shows that the RRE adopts an ‘A’ shaped architecture poised to recruit Rev (Fang et al., 2013). It remains unknown how the RNA structure is matched to the arrangement of the Rev oligomer to form the export–active complex.
One particularly fascinating question is how the individual subunits of Rev recognize the RRE. An NMR structure of a single α-helical ARM bound to the IIB RNA hairpin has been known for quite some time (Battiste et al., 1996), and a second binding site in stem IA was identified that utilizes a different surface of the helical ARM from another subunit to recognize the RNA (Daugherty et al., 2008). Other binding sites in the RRE may use yet other recognition strategies that have not been defined. As a step towards understanding how individual binding sites are organized on the RRE, how their arrangement relates to the architecture of the Rev oligomer, and how the different sites are recognized by Rev, we solved the structure of a Rev dimer bound to an RRE fragment (IIB40) containing two Rev-binding sites (Daugherty et al., 2008) (Figure 1C). This first high-resolution structure of a Rev-RRE complex (Figure 1D) uncovers yet another RNA recognition mode for Rev and shows how the remarkable plasticity of the Rev dimer allows it to adapt to the RNA framework.
To better understand how Rev assembles an oligomeric complex on the RRE, we sought to determine the crystal structure of a Rev-RNA complex. Rev has a strong tendency to aggregate and fold incorrectly when not bound to RNA (Daugherty et al., 2010b). We circumvented these problems by making RNA complexes with Rev 1–70 harboring the oligomerization-disrupting mutations, Leu12Ser and Leu60Arg (Jain and Belasco, 2001), which forms a stable Rev dimer (Daugherty et al., 2010b), and a Glu47Ala surface-entropy reducing mutation (Goldschmidt et al., 2007), which aided in crystallization (Figure 1A, Figure 1—figure supplement 1).
For the RNA, we utilized variants of RRE IIB40, which has the high-affinity IIB site positioned adjacent to a second site (referred to as the junction site) that mimics the stem II junction (Daugherty et al., 2008) (Figure 1C, Figure 1—figure supplement 1), and variants of a 68-nucleotide fragment comprising stems IIA, IIB and IIC. Previous biochemical studies have shown that Rev assembly initiates at stem IIB and proceeds through stem II along the length of stem I with the second Rev molecule binding at the stem II junction (Mann et al., 1994; Charpentier et al., 1997). The junction site was designed based on those studies (Daugherty et al., 2008) and also on the observation that Rev binding melts a A-U base-pair in stem IIA at the junction (Charpentier et al., 1997), a finding recently confirmed by SHAPE-seq experiments (Bai et al., 2014). Gel shift assays show that Rev binds to RRE stem II and to IIB40 with comparable affinity and cooperativity (Figure 1—figure supplement 1), suggesting that IIB40 recapitulates Rev assembly at the entire stem II junction.
We attempted co-crystallizations using ∼35–40 RNA variants with altered stem lengths and tetraloop sequences and obtained well-diffracting crystals with only a single Rev dimer-IIB40 complex. We solved its structure using single-wavelength anomalous dispersion to 5.5 Å to generate models for molecular replacement at 3.2 Å, resulting in the first high-resolution view of a Rev-RRE complex (Figure 1D, Figure 1—figure supplement 2, Table 1).
The Rev dimer-RRE structure depicts essential early steps in the assembly of the larger oligomeric complex. The overall architecture of the complex shows the RNA held between two Rev molecules arranged slightly asymmetrically in a V-shaped topology, with the two Rev ARMs buried in distorted major grooves positioned on opposite sides of the RNA (Figure 1D). The structures of the individual Rev monomers are nearly identical to each other and to the RNA-free form (Figure 1—figure supplement 3). Comparison of the structures of the free and RNA-bound Rev dimers, described below, illustrates how the homo-oligomer adapts to bind RNA cooperatively.
The Rev dimer in its RNA-free state also displays a V-shaped topology (DiMattia et al., 2010; Daugherty et al., 2010b) but its conformation changes dramatically upon RNA binding, with the two RNA-binding helices coming much closer together in the RNA-bound complex (Figure 2A). The Rev subunit bound to the junction site forms an extensive surface with the RNA that causes the dimer interface to pivot around a single residue, Ile55 (Figure 2—figure supplement 1), known to be critical for dimer integrity and function (Jain and Belasco, 2001; Edgcomb et al., 2008). This altered conformer results in complete repacking of hydrophobic residues at the dimer interface (Figure 2B), formed by residues Leu18 and Leu22 from the first helix and Ile55 and Ile59 from the second helix. Phe21, which formed a significant part of the RNA-free interface, is largely excluded.
Interestingly, the bound conformation buries less surface area at the dimer interface than the free form (1000 Å2 vs 1500 Å2) and also shows looser hydrophobic packing (the two Rev subunits are ∼3.2 Å further apart) (Figure 2—figure supplement 1), suggesting that this bound dimer is energetically less favorable but is compensated by the energy of RNA binding of the second Rev subunit. The weaker interface may be further compensated by hydrogen bonding between the well-conserved Gln51 residues symmetrically located between the two subunits and observed only in the bound form (Figure 2C). Thus, mutation of Gln51 to alanine results in accumulation of Rev monomer-RNA complexes and ∼30-fold reduced affinity (Figure 2D). In the context of the full-length hexameric Rev-RRE complex, the effect of the Gln51Ala mutation was modest but reproducible, showing a twofold reduction in affinity in gel shift assays using a vast excess of protein over RNA (Figure 2—figure supplement 1) and reduced accumulation of dimer species in assays with stoichiometric amounts of Rev and RRE (Figure 2E,F). The modest effect of the mutation in the full-length context suggests that the Gln51 interaction may be less crucial during later stages of Rev-RRE assembly and the energetic contributions to cooperative binding may shift to other interactions, possibly coupled to further rearrangements of the Rev subunit interfaces. Alternatively, there may be multiple pathways towards cooperative assembly such that if one pathway is blocked, another may be used.
The pliability of the Rev dimer interface contributes to RNA-binding specificity by allowing the Rev subunits to orient properly to multiple binding sites arrayed on the RRE, with binding at the primary IIB RNA site nucleating assembly. Comparison of our Rev dimer-IIB40 structure to the NMR structure of a Rev ARM peptide-IIB complex (Battiste et al., 1996) shows that both the Rev arginine-rich-helix and IIB RNA site are remarkably similar in both the structures (Figure 3A), although with subtle changes to some Rev-IIB contacts (Figure 3B, Figure 3—figure supplement 1). For example, in the NMR structure, Asn40 hydrogen bonds to a G47-A73 base pair in IIB RNA, but is in a different plane and contacts G47 and G71 from adjacent base pairs in the dimer-RNA complex (Figure 3B). Interestingly, the Asn40-IIB interaction observed in the crystal structure is similar to that seen in an NMR structure of a Rev-ARM peptide-aptamer complex (Ye et al., 1996), suggesting that optimal Rev-RNA contacts can be selected from several favorable conformers. The subtle differences between the NMR and crystal structures likely reflect a lack of precision of the early NMR data, with no violations observed to the current structure. The similarity in Rev-IIB recognition between the NMR structure, which depicts the first step in Rev-RRE assembly, and the crystal structure, which captures the progression into cooperative assembly, reiterates the importance of preserving this highly sequence-specific interface to support a specific and productive assembly.
The junction site in the RNA forms a structural scaffold that helps orient the Rev subunits. It is formed by an A44-G76 base-pair, identified earlier by in vitro selection studies (Bartel et al., 1991), and a A42-U43 bulge that widens an otherwise narrow A-form major groove to accommodate the α-helix of the second Rev subunit (Figure 3C, Figure 3—figure supplement 2). Both A42 and U43 are unpaired and A42 is co-axially stacked on stem IIB, consistent with biochemical mapping experiments showing increased accessibility of these nucleotides upon Rev binding (Charpentier et al., 1997; Bai et al., 2014). Neither base is directly contacted by Rev, suggesting that their primary role is to create the junction site architecture. Consistently, deleting the bulge results in exclusively monomeric binding (Figure 3—figure supplement 2), and previous studies showed that simply introducing bulges with variable sequences juxtaposed to IIB is sufficient to mediate co-operative Rev binding (Zemmel et al., 1996). Most contacts at the junction site are to the phosphate backbone with the exception of Arg 43 and Arg44 (Figure 3—figure supplement 1). Notably, Arg44 from both Rev subunits contacts the two binding sites on the RNA, in one case using water-mediated interactions not previously observed for RRE recognition (Figure 3—figure supplement 1, 2).
The junction site in the full-length RRE is formed by three stems (IIA, IIB, and IIC), with only IIB and IIC present in the crystallized IIB40 RNA. Stem IIA likely forms an A-form helix that merges into the junction to determine the relative placement of the remaining RRE. The second Rev subunit is positioned in the crystal structure to use its higher order oligomerization surface to recruit a third Rev subunit with its ARM oriented towards stem IIA for sequential Rev assembly. Presumably, the overall ‘A’-shaped RRE topology (Fang et al., 2013) and arrangement of binding sites dictates the positioning of newly recruited subunits (Figure 3—figure supplement 3). The specific model for Rev-RRE assembly proposed by Fang et al. (2013) portrays an initial Rev dimer bridging the RRE IIB and IA sites, but our structure and other biochemical studies of Rev-RRE assembly (Charpentier et al., 1997; Bai et al., 2014) suggest that the first Rev dimer binds to stem II before additional subunits are added. Rev dimers formed later during assembly are likely positioned to bridge the two arms of the ‘A’-shaped RRE, presumably involving the IA site.
The diversity of RNA recognition by the Rev ARM is quite striking, now with examples of how Rev binds to three sites in the RRE, and yet other binding modes observed with in vitro selected nucleic acids (Xu and Ellington, 1996; Ye et al., 1999; Bayer et al., 2005). The current crystal structure shows that each of the two ARMs of the Rev dimer deeply insert into a distorted RNA major groove, each burying ∼1500 Å2 of surface area, and that the same face of the ARM is used to contact the RNA in both cases (Figure 3D). However their helical registers differ and recognition of IIB is highly sequence-specific compared to the junction site. Furthermore, at another characterized Rev-binding site in the RRE, stem IA, yet a different binding mode seems to be utilized, with mutational studies indicating a different face of the ARM used to bind the RNA (Figure 3D) (Daugherty et al., 2008).
The Rev-RNA interface at the junction site illustrates that protein contacts at individual RRE sites can be largely sequence-nonspecific if the ARM is optimally presented to the RNA, especially from the context of additional bound Rev subunits. While binding to some sites such as stem IIB or IA can show substantial specificity even in their isolated contexts, this does not seem to be the case for the remaining Rev binding sites (Daugherty et al., 2008). Binding at these sites might require a more extensive RRE framework and neighboring Rev subunits and may differ from the configuration seen at the junction site, potentially adding further diversity to the recognition modes utilized by the ARM.
To test the functional implications of the observed interactions in the crystal structure, we monitored the effects of key mutations in Rev on RRE binding, viral RNA export, and viral replication. A previous study identified Asn40 as the most critical residue for IIB recognition, followed by Arg44, Arg38 and Arg39 (Tan et al., 1993). In the dimer-RNA complex however, alanine substitution of Arg44, which contacts both RNA sites, was most deleterious (250-fold reduced affinity) followed by Asn40 (100-fold), Arg39 (30–50-fold), and Arg38 (10-fold) (Figure 4A, Figure 4—figure supplement 1). These mutations, as well as mutation of Gln51 which hydrogen bonds across the dimer interface, were all defective in viral RNA export and translation in Rev-RRE-dependent Gag-Pol reporter assays (Figure 4B). When engineered into HIV-1, all mutations, except Gln51, also showed attenuated viral replication in tissue culture (Figure 4C, Figure 4—figure supplement 2). As noted above, Gln51 is important at the level of dimer formation but may be less critical in the full Rev-RRE context. RNA fluorescence in situ hybridization (FISH) of infected cells indicated that these Rev mutations caused retention of unspliced HIV-1 RNAs in the nucleus (Figure 4D), correlating with reduced Gag protein production (Figure 4—figure supplement 3). Thus, key interactions observed in the Rev-RNA crystal structure have important functional roles in HIV RNA export and viral replication.
Previous crystal structures of Rev dimers and the NMR structure of a Rev ARM-IIB complex defined key building blocks of Rev-RRE assembly, but it has been unclear how they are employed to guide cooperative RNA binding of the oligomer to assemble a fully functional complex. Our structure of the Rev-RRE complex illustrates a direct physical coupling between RNA binding and protein oligomerization, where RNA binding drastically alters the conformational state of the Rev dimer. The architecture of the RRE and positioning of individual Rev-binding sites configures the predominantly hydrophobic protein–protein interfaces to form a cooperative complex. This is reminiscent of the DNA-induced hetero-dimerization of nuclear receptors, where DNA sequence and spacing between protein-binding half-sites configures the dimer interface and dictates the choice of dimerization partners (Meijsing et al., 2009; Rastinejad et al., 2013).
The use of non-polar residues at protein–protein interfaces has been postulated to be an attractive choice for evolutionary change as it can readily accommodate structural perturbations (Lesk and Chothia, 1980). Structural studies of the HIV-1 capsid protein clearly illustrate the role of plasticity at interfaces to assemble complex structures such as fullerene cones (Byeon et al., 2009; Pornillos et al., 2009, 2011). The simple modular structure of Rev, including its pliable protein–protein interface and diverse RNA-binding domain, highlights how a virus with limited coding capacity can build a large, asymmetric RNP using a small, homo-oligomeric protein to achieve remarkable structural and functional complexity.
Our structure expands the repertoire of known Rev dimer conformers to three, each with different crossing angles between the RNA-binding helices. Given the hydrophobic nature of the interfaces, other conformers are possible, especially when placed in the context of host–protein complexes. These conformers can be combined in multiple arrangements to generate oligomers with diverse architectures (Figure 5A). Since Rev likely interacts with other host proteins to transport and package viral RNAs during the virus life cycle (Jager et al., 2012; Naji et al., 2012), different quaternary structures of Rev may generate conformational states adapted to different host complexes at various stages of viral replication. This is highly reminiscent of the Ebola virus protein VP40, whose dimeric, hexameric, and octameric forms each have different roles in the viral life cycle (Bornholdt et al., 2013), again highlighting the value of adaptable homo-oligomers in small RNA viruses. Recent crystal structures of HIV Tat (Tahirov et al., 2010) and Vif (Guo et al., 2014) with host protein complexes illustrate the importance of scaffolding structures to enable viral proteins to adopt ordered, functional states. It is remarkable that for Rev, it is not the folding of a single polypeptide chain that is dictated by its scaffolding partner(s) but rather its oligomeric architecture.
The function of Rev in RNA export is well defined, and the arrangement of Crm1 export complexes provides one clear example in which RNA organizes the Rev oligomer into a functionally important state. The ‘jellyfish’ model depicts Rev hexamers to be associated with the RNA on one side of the protein and NESs that recruit Crm1 export receptors on the other side (Daugherty et al., 2010b). Recent structural studies show that the Rev-RRE complex binds a Crm1 dimer utilizing two NES-binding pockets positioned for Rev recognition (Booth et al., 2014). This study also shows that the RRE plays an important role in assembly of the export complex by enhancing the affinity of the Rev oligomer for Crm1, probably by using the RNA scaffold to organize the Rev subunits and increase the local concentration of NES peptides for Crm1 binding. Additionally, SAXS studies of the RRE (Fang et al., 2013) and SHAPE-seq analysis of Rev-RRE assembly (Bai et al., 2014) reveal that the RNA is compact and pre-ordered to bind Rev. Consequently, it may be inferred that changes to the RRE scaffold can affect the overall arrangement of Rev subunits and redistribute the spatial availability of NESs for Crm1 recruitment and nuclear export (Figure 5B). Multiple studies support this hypothesis: (1) Resistance to a dominant negative mutant of Rev with a defective NES arose through two mutations in the RRE that changed the RNA structure but not Rev multimerization (Legiewicz et al., 2008), consistent with an RNA-guided reorganization of Rev oligomers. (2) Conversely, Rev complexes formed with simple repeats of the IIB hairpin displayed high RNA-binding affinity but did not recapitulate full export activity (Symensma et al., 1999), indicating that the specific RRE architecture properly arranges the Rev oligomer for function. (3) A study of cognate Rev-RRE pairs from HIV-infected patients found that mutations in the RRE were responsible for changes in activity as the pairs evolved during the course of infection (Sloan et al., 2013), suggesting that small changes in RRE sequence can alter the architecture of the Rev-RRE complex to tune its functional output. Previous studies likened the RRE to a molecular rheostat, where it is sensitive to the intracellular Rev concentrations and export activity adjusts accordingly (Mann et al., 1994). It appears now that RRE can also perform as a structural/evolutionary rheostat, where changes to RRE structure during the course of infection can rearrange the Rev oligomer and tune the levels of nuclear export. It will be interesting to determine if interactions with other host proteins exploit the Rev-RRE oligomer in other ways.
We used a Rev 1–70 dimer construct harboring the higher-order oligomerization-disrupting mutations, Leu12Ser and Leu60Arg (Daugherty et al., 2010b), and a Glu47Ala surface-entropy reducing mutant (Goldschmidt et al., 2007) for initial crystallization screens. We note that position 47 in most HIV-1 Rev isolates already is alanine, unlike the Glu47 found in HIV strain HXB3. Rev protein was expressed in Escherichia coli BL21 (DE3) with an N-terminal hexa-histidine tag and GB1 solubility-enhancing domain and purified using Ni affinity chromatography essentially as described (Daugherty et al., 2010a). However, instead of using RNAses to eliminate contaminating endogenous E. coli RNA that co-purifies with Rev, we added NaCl to 2 M and urea to 1 M to the cleared lysate (Marenchino et al., 2009) and incubated for 1–2 hr at 4°C while binding in batch to the Ni-NTA resin. Washes and elutions were performed as described (Daugherty et al., 2010a). Fractions of pure protein were pooled, concentrated to ∼0.2 mM and purified by size-exclusion using a Superdex 75 column equilibrated in 40 mM Tris pH 8.0, 0.2 M NaCl, 0.1 M Na2SO4. To remove the His-GB1 tag, protein fractions after size-exclusion were pooled and cleaved with Tev protease after adding ammonium sulfate to 0.4 M. The pure, tag-free Rev obtained was concentrated to 40–50 μM and stored at 4°C prior to complex formation with RNA.
RNAs were synthesized by T7 in vitro run-off transcription from synthetic DNA templates as described (Daugherty et al., 2008). Purified and annealed RNAs were lyophilized and stored at −20°C.
Lyophilized RNAs were resuspended in water at 0.4–0.5 mM. Protein and RNA were incubated together at 1.9:1 ratio at room temperature for 30 min, concentrated to 5–6 mg/ml using an Amicon ultrafree 4 3 kDa cut-off concentrator (EMD Millipore, Billerica, MA), exchanged into crystallization buffer (10 mM HEPES pH 6.5, 50 mM KCl, 2 mM MgCl2) using zeba desalting columns (Thermo Fisher Scientific, Rockford, IL) and concentrated to 10–12 mg/ml. Initial screens were performed using a Mosquito robot with ProComplex and Nucleix suites (Qiagen Inc, Valencia, CA). After screening ∼35–40 Rev dimer-RNA complexes, one complex comprising the Rev dimer Glu47Ala mutant and an RNA derived from IIB40, containing a UUCG tetraloop, formed cubic crystals in several conditions with PEG 400, PEG 4000 and PEG 8000. Optimized crystals were grown using hanging drop vapor diffusion by mixing the complex at 1:1 with reservoir solution containing 50 mM MES 6.0, 50 mM KCl 1–4% PEG 4000 (for native diffraction data collection) or 50 mM sodium cacodylate pH 6.0, 5 mM MgCl2, 8–14% PEG 400 (for heavy atom soaking). Crystals appeared in 1–2 days and grew to their full size of 100–200 microns in 3–4 days. After 1 week, native crystals were transferred into a drop containing well solution supplemented with 5% PEG 4000. PEG 400 and PEG 4000 were added directly to the drop in small increments to a final concentration of 25% PEG 400 and 14% PEG 4000 after which the crystals were flash frozen in liquid nitrogen. Crystals grown in MES buffer failed to survive heavy atom soaks but crystals grown in 50 mM cacodylate pH 6.0, 5 mM MgCl2, 8–14% PEG 400 were more tolerant to heavy atom soaks although they did not diffract beyond 4.5 Å. Crystals for heavy atom soaking were transferred to a fresh drop of well solution supplemented with 5% PEG 4000 and PEG 400–25% was added to the drop in small increments. For anomalous data collection, crystals were soaked in 7.1 mM ammonium tetrathio tungstate (Hampton Research, Aliso Viejo, CA, Heavy Atom Screen M2) for 15 min before flash freezing in liquid nitrogen.
Data sets were collected at the Advanced Light Source beamline 8.3.1 at 100 K. Crystals belonged to either a tetragonal (P41212) spacegroup or a cubic spacegroup (P4132). Diffraction data was processed and scaled using XDS (Kabsch, 2010). Attempts with molecular replacement using the unbound Rev dimer structure or models of Rev-IIB RNA complexes as starting models with native diffraction datasets (collected at 1.116 Å) were unsuccessful. Using single wavelength anomalous dispersion (SAD) data to 5.5 Å on a W-soaked crystal (collected at 1.2146 Å), tungstate ions were located using ShelxD, phased and density-modified using ShelxE (Sheldrick, 2010). The crystal belonged to P41212 space group with three copies of the Rev dimer-RNA complex and one W site per asymmetric unit. Density for the Rev helices and RNA was very clear and Rev molecules from the Rev-dimer structure (Daugherty et al., 2010b) and an RNA fragment from IIB RNA NMR structure (Battiste et al., 1996) were fit into the protein and RNA densities, respectively, using rigid-body fit in Coot (Emsley et al., 2010). The resulting model was used for molecular replacement into the 3.2 Å native data in the cubic space group (P4132) in Phaser (McCoy et al., 2007). Initial solutions scored poorly but the appearance of positive density near the RNA fragment at the second Rev-binding site suggested that the solution was correct. Following iterative rounds of model building of the RNA using RCrane (Keating and Pyle, 2010) in Coot and refinement in Phenix (Adams et al., 2010), a second round of molecular replacement was carried out using the newly built RNA alone as the search model. This time, the solution was clear with positive density for both Rev subunits. Rev subunits were then built into the density with iterative model building and refinement. The final model had good stereochemistry and showed no Ramchandran outliers (Chen et al., 2010). While density for most water molecules or ions was not apparent at this resolution, we could place one water molecule bridging a Rev-RNA contact. Solvent-accessible surface areas were calculated using PISA (Krissinel and Henrick, 2007) and figures were generated using PyMOL (Schrodinger, 2010).
Electrophoretic mobility shift assays were performed in gel shift buffer (10 mM HEPES pH 7.5, 0.1 M KCl, 1 mM MgCl2, 0.5 mM EDTA, 2 mM DTT, 10% glycerol, 50 μg/ml yeast tRNA and 0.2 mg/ml BSA). Rev protein stock for RNA binding was prepared in gel shift buffer supplemented with fivefold molar excess of yeast tRNA. Rev was diluted in gel shift buffer and mixed with an equal volume of <25 pM 32P-RNA. Reactions were incubated at room temperature for 15 min and loaded onto continuously running 6% (for 234 nucleotide RRE) or 10% polyacrylamide gels (0.5× TBE). Gels were run at room temperature for 60–90 min, dried and exposed to a phosphorimaging screen for >12 hr. Bands were quantified using ImageJ (Schneider et al., 2012) and data were fit using MS Excel.
For gel shift assays using RRE and detection with SYBR Green II staining (Invitrogen), gel shift buffer contained 10 mM HEPES pH 7.5, 0.3 M KCl, 1 mM MgCl2, 0.5 mM EDTA, 2 mM DTT, 10% glycerol and 0.2 mg/ml BSA (Fang et al., 2013). Rev was serially diluted and mixed with an equal volume of RRE at 125 nM, incubated for 15 min and loaded onto continuously running 6% polyacrylamide gels (0.5× TBE). Gels were run at room temperature for 1–2 hr, stained with SYBR Green II and visualized under UV light. Bands were quantified using ImageJ (Schneider et al., 2012).
The pCMV-GagPol-RRE reporter construct (Srinivasakumar et al., 1997) was modified to contain the HIV-1 SF2 RRE sequence (used in biochemical assays) and was complemented in trans with pcDNA4TO-Rev-3xFlag (Invitrogen mammalian expression vector) in 293T cells by transient transfection. The optimal ratio of reporter to Rev expression plasmids was first determined to establish maximal signal under non-saturating conditions for the assay. Subsequently, 50,000 HEK 293T cells, each in 96-well microtiter plate wells, were transfected with 90 ng of reporter plasmid, 0.3 ng of Rev expression plasmid, 5 ng of an mCherry fluorochrome expression vector (to visualize transfection efficiency) and 4.7 ng of pBS carrier DNA. 48 hr after transfection, the culture supernatant was removed, the cells were lysed and intracellular p24 was quantified by ELISA. mCherry expression was consistent across all transfections at the time of cell lysis. The expression of Rev-3xFlag and GAPDH was monitored by western blot analyses using mouse α-Flag (Sigma-Aldrich Corp. St. Louis, MO) and mouse α-GAPDH (Abcam, Cambridge, MA) antibodies, respectively, followed by goat α-mouse IgG-HRP (Santa Cruz Biotechnology, Dallas, TX). HRP-bound antigens were then detected by chemi-luminescence and imaged using a BioRad ChemiDoc MP imager (Bio-Rad Laboratories, Inc. Hercules, CA). Expression levels for Rev mutants were within twofold of wild-type Rev, when normalized to GAPDH loading control, and were used to normalize the quantified p24 levels (Figure 4B).
The rev cDNA was relocated to the nef region (nef is dispensable for ex vivo HIV replication in SupT1 cells) (Figure 4—figure supplement 2) to facilitate mutagenesis of rev without affecting the overlapping tat or env reading frames.
The infectious molecular clone, HIVNL4-3, was acquired from the NIH AIDS Reagent Program. HEK 293T cells were co-transfected with proviral plasmids and mammalian expression vectors containing HIVNL4-3 Rev and Tat at a ratio of 1:50 (wt/wt, mammalian expression vector:proviral plasmid) to promote robust production of viruses containing mutant Rev. Virus-laden supernatants were collected 48 hr post transfection, centrifuged at 500×g for 5 min, 0.45 μM-filtered, and stored in aliquots at −80°C. Virus stocks were titered using ELISA to quantify p24 viral capsid protein. Mouse monoclonal and rabbit polyclonal α-p24 antibodies used in the ELISA were obtained from the NIH AIDS Reagent Program.
Sup-T1 cells (NIH AIDS Reagent Program) were cultured in RPMI media supplemented with 25 mM HEPES pH 7.4, 10% heat-inactivated fetal calf serum (Hyclone, Waltham MA), and 1% penicillin/streptomycin. 106 Sup-T1 T-cells were inoculated with 2.5 ng p24 in 250 μl media containing 1 μg/ml PEI and 8 μg/ml polybrene in 96-well flat bottom polystyrene cell culture microplates and then centrifuged for 2 hr at 1200×g at 37°C. Each virus was assayed in biological triplicate. Input virus was then removed via washing cells twice with 250 μl phosphate-buffered saline (PBS). Cells were resuspended in 250 μl of media and incubated at 37°C, 5% CO2. Under these centrifugal inoculation conditions, the multiplicity of inoculation (MOI) is approximately 0.05–0.025 as determined by acute infection of Sup-T1 T-cells with serial dilutions of HIVNL4-3 expressing green fluorescent protein (GFP) followed by flow cytometry to quantify the GFP-expressing population. The infections were monitored every 48 hr following inoculation by immunofluorescence assay to detect cellular HIV antigen synthesis and by HIV p24 ELISA on the culture supernatant to quantify virus release. The infections were monitored until robust HIV cytopathic effect (cell necrosis and syncytia formation) was observed in the wild-type virus control.
HeLa cells (150,000) cultured on 12 mm poly-L-lysine coated glass coverslips were acutely infected with 100 ng p24 of HIV pseudotyped with VsV-G envelope glycoprotein and containing 3x-flag epitope-tagged Rev in the nef locus in DMEM media containing 8 μg/ml polybrene. These inoculation conditions equate to an effective MOI of 1.0–0.5 as determined previously via TCID50 determination in HeLa cells.
24 hr post inoculation, the cells were washed twice with PBS, fixed in 4% paraformaldehyde/PBS for 10 min at RT, washed in 0.15 M glycine/PBS, permeabilized with 0.5% Triton X-100/PBS, and then blocked with 1% bovine serum albumin (BSA)/PBS/0.1% Tween-20 at 37°C for 1 hr. Antigens were then probed with a 1:250 dilution (vol/vol) of rabbit polyclonal α-HIV p24 (NIH AIDS Reagent Progam) and a 1:500 dilution of mouse α-flag (Sigma) in 1% BSA/PBS/0.1% Tween-20 at 37°C for 1 hr. The cells were washed twice with PBS/0.1% Tween-20 and then probed with a 1:250 dilution of goat α-rabbit IgG-rhodamine (MP Biomedicals, LLC, Santa Ana, CA), a 1:500 dilution of goat α-mouse IgG-FITC (MP Biomedicals) and 100 ng/ml DAPI nuclear stain. The cells were washed three times with PBS/0.1% Tween-20 and the coverslips were mounted to glass slides using Vectashield hardset mounting media for fluorescence. Cells were imaged using a 63×, 1.3 NA Leica objective and a Hamamatsu C4742-12 cooled CCD camera on a Leica DMI6000B stand. 25 z-stack images were taken using exposures normalized to the brightest sample, background-subtracted, and then merged to a single image using the brightest point projection method in the Volocity imaging software suite (Perkin Elmer, Waltham, MA).
For RNA subcellular localization by FISH, acutely infected HeLa cells were fixed with methanol for 10 min at RT and rehydrated in 2× SCC/10% formamide at 37°C for 10 min. They were then probed using Stellaris probe sets (Biosearch) targeting HIV-1 gag nt790–2292 conjugated to Quasar570 dye and HIV-1 nef nt8787–9407 conjugated to Quasar670 dye at 125 nM in 2× SCC/10% formamide/100 mg/ml dextran sulfate for 2 hr at 37°C. The total HIV-1 RNA probe set included 39 unique DNA oligonucleotides and the HIV-1 gag RNA probe set included 48 unique DNA oligonucleotides each computationally predicted to bind specifically to their target sequence (Biosearch Stellaris FISH Probe Designer web-based program). Following incubation, 2 ml of 2× SCC/10% formamide was added and incubated for 30 min at 37°C to dissociate nonspecifically bound probe. The wash was aspirated, and 1 ml of 2× SCC/10% formamide containing 100 ng DAPI nuclear stain was added and incubated at 37°C for 30 min. The cells were then washed twice with 2× SCC, coverslips mounted to glass slides, and cells were imaged with brightness and contrast adjusted using the same linear parameters for all images.
Atomic coordinates and structure factors for the Rev dimer-RRE crystal structure have been deposited in the Protein Data Bank under the accession code 4PMI.
PHENIX: a comprehensive Python-based system for macromolecular structure solutionActa Crystallographica. Section D, Biological Crystallography 66:213–221.https://doi.org/10.1107/S0907444909052925
The export receptor Crm1 forms a dimer to promote nuclear export of HIV RNAeLife, 10.7554/eLife.04121.
A dynamic in vivo view of the HIV-I Rev-RRE interactionJournal of Molecular Biology 266:950–962.https://doi.org/10.1006/jmbi.1996.0858
MolProbity: all-atom structure validation for macromolecular crystallographyActa Crystallographica. Section D, Biological Crystallography 66:12–21.https://doi.org/10.1107/S0907444909042073
Characterization of HIV-1 REV protein: binding stoichiometry and minimal RNA substrateNucleic Acids Research 19:1577–1583.https://doi.org/10.1093/nar/19.7.1577
HIV Rev response element (RRE) directs assembly of the Rev homooligomer into discrete asymmetric complexesProceedings of the National Academy of Sciences of USA 107:12481–12486.https://doi.org/10.1073/pnas.1007022107
Structural basis for cooperative RNA binding and export complex assembly by HIV RevNature Structural & Molecular Biology 17:1337–1342.https://doi.org/10.1038/nsmb.1902
Implications of the HIV-1 Rev dimer structure at 3.2 A resolution for multimeric binding to the Rev response elementProceedings of the National Academy of Sciences of USA 107:5810–5814.https://doi.org/10.1073/pnas.0914946107
Human immunodeficiency virus type 1 regulator of virion expression, rev, forms nucleoprotein filaments after binding to a purine-rich “bubble” located within the rev-responsive region of viral mRNAsProceedings of the National Academy of Sciences of USA 88:7366–7370.https://doi.org/10.1073/pnas.88.16.7366
Semiautomated model building for RNA crystallography using a directed rotameric approachProceedings of the National Academy of Sciences of USA 107:8177–8182.https://doi.org/10.1073/pnas.0911888107
Structural analysis of the interaction between the human immunodeficiency virus Rev protein and the Rev response elementProceedings of the National Academy of Sciences of USA 88:683–687.https://doi.org/10.1073/pnas.88.3.683
Inference of macromolecular assemblies from crystalline stateJournal of Molecular Biology 372:774–797.https://doi.org/10.1016/j.jmb.2007.05.022
Resistance to RevM10 inhibition reflects a conformational switch in the HIV-1 Rev response elementProceedings of the National Academy of Sciences of USA 105:14365–14370.https://doi.org/10.1073/pnas.0804461105
NUCPLOT: a program to generate schematic diagrams of protein-nucleic acid interactionsNucleic Acids Research 25:4940–4945.https://doi.org/10.1093/nar/25.24.4940
Rapid and efficient purification of RNA-binding proteins: application to HIV-1 RevProtein Expression and Purification 63:112–119.https://doi.org/10.1016/j.pep.2008.09.010
Host cell interactome of HIV-1 Rev includes RNA helicases involved in multiple facets of virus productionMolecular & Cellular Proteomics 11:M111 015313.https://doi.org/10.1074/mcp.M111.015313
UCSF Chimera–a visualization system for exploratory research and analysisJournal of Computational Chemistry 25:1605–1612.https://doi.org/10.1002/jcc.20084
HIV-1 Rev protein assembles on viral RNA one molecule at a timeProceedings of the National Academy of Sciences of USA 106:1404–1408.https://doi.org/10.1073/pnas.0807388106
Understanding nuclear receptor form and function using structural biologyJournal of Molecular Endocrinology 51:T1–T21.https://doi.org/10.1530/JME-13-0173
The PyMOL molecular Graphics system, Version 1.3r1
Experimental phasing with SHELXC/D/E: combining chain tracing with density modificationActa Crystallographica. Section D, Biological Crystallography 66:479–485.https://doi.org/10.1107/S0907444909038360
The effect of viral regulatory protein expression on gene delivery by human immunodeficiency virus type 1 vectors produced in stable packaging cell linesJournal of Virology 71:5841–5848.
Polyvalent Rev decoys act as artificial Rev-responsive elementsJournal of Virology 73:4341–4349.
Anti-peptide aptamers recognize amino acid sequence and bind a protein epitopeProceedings of the National Academy of Sciences of USA 93:7475–7480.https://doi.org/10.1073/pnas.93.15.7475
Deep penetration of an alpha-helix into a widened RNA major groove in the HIV-1 rev peptide-RNA aptamer complexJournal of Applied Crystallography 3:1026–1033.https://doi.org/10.1038/nsb1296-1026
RNA architecture dictates the conformations of a bound peptideChemistry & Biology 6:657–669.https://doi.org/10.1016/S1074-5521(99)80117-3
Oligomerization and RNA binding domains of the type 1 human immunodeficiency virus Rev protein: a dual function for an arginine-rich binding motifProceedings of the National Academy of Sciences of USA 88:7734–7738.https://doi.org/10.1073/pnas.88.17.7734
Flexible regions of RNA structure facilitate co-operative Rev assembly on the Rev-response elementJournal of Molecular Biology 258:763–777.https://doi.org/10.1006/jmbi.1996.0285
Wesley I SundquistReviewing Editor; University of Utah, United States
eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.
Thank you for sending your work entitled “RNA-directed remodeling of the HIV–1 Rev oligomer orchestrates Rev-RRE assembly” for consideration at eLife. Your article has been favorably evaluated by John Kuriyan (Senior editor), Wesley Sundquist (Reviewing editor), and 2 reviewers, one of whom, Jamie Williamson, agreed to reveal his identity.
The Reviewing editor and the reviewers discussed their comments before we reached this decision, and the Reviewing editor has assembled the following comments to help you prepare a revised submission.
Jayaraman et al. present a 3.2 A resolution crystal structure of a Rev dimer bound to a fragment of RNA corresponding to the high-affinity binding site (Stem loop IIB) of the Rev-response element (RRE) plus the adjacent IIABC junction. In order to crystallize the complex, the authors used a Rev construct in which two residues were mutated at one of the hydrophobic oligomerization interfaces, promoting formation with only a Rev dimer. While dimerization of Rev occurs using the same hydrophobic interfaces as previously reported, the structure reveals that the Rev dimer binds the RNA with a crossing angle that is much more acute than was previously shown in the dimer in the absence of RNA (50 deg vs. 120-140 deg.). The authors suggest that this degree of freedom (different crossing angles) may be exploited in the formation of higher order complexes. Strong biochemical and molecular virological analyses are presented in support of the structure.
The crystal structure is the first example of a high resolution structure of a Rev dimer-RNA complex, despite considerable efforts over two decades, and is thus an important advance. Important new contributions include: 1) the discovery that the crossing angle and hydrophobic interface of the Rev dimer can change quite dramatically (∼70 degrees) in response to RNA binding, which allows the two ARM helices to bind in adjacent major grooves of the RNA, and 2) the definition of a new Rev-RNA binding mode (seen for the Rev-RNA Junction site), which may represent the second step in cooperative assembly of Rev with the RRE. The manuscript is clear and well written, and the descriptions of the structure are very lucid. The study is of high technical quality, reveals remarkable molecular flexibility in both the Rev protein configuration and its RNA interactions, and represents an important advance in our understanding of the molecular basis for HIV-1 RNA export. This and the accompanying manuscript describing the EM of a Crm1 dimer bound to Rev-RRE are the most significant advance in this important story for many years.
Major issues (which must be addressed by the authors):
This work provides the most advanced structural framework for thinking about how the Rev-RRE export complex is assembled. No additional experiments need to be performed, but the text should be modified to put the observations into better context and to address the specific issues highlighted below.
1) There are now three different arrangements of the Rev dimer that have been observed, and surely there is interesting, and perhaps confounding, structural plasticity of the Rev dimer interface. This plasticity is portrayed as being a “strategy” of the virus, but the possibility that one or more of the in vitro structures may not reflect functional complexes. The manuscript skirts around this issue, and does not really address how to build the present complex into a functional Rev RRE complex, or how all of the structures may or not be consistent with that functional complex.
2) The authors should also discuss how well the junction serves as a proxy for the junction in the viral Stem II. In particular, Stem IIA is almost certainly an A-form helix that must merge into the junction, which is only two nucleotides in the crystallized construct. It is possible that the native Stem II RNA has a different conformation, and that the plasticity of the Rev-RRE interface has allowed crystallization of a non-native conformation.
3) There is no conclusive evidence that a second molecule of Rev binds to stem IIC in the context of the larger RRE, and the data in Figure 2, panels (d) and (e) provide little support since a mutation that is absolutely required for dimerization in the Stem II40 RNA has little to no discernable effects in the context of the larger RNA. While the authors argue that, in the context of a larger RNA, this mutation's effects can be overcome is an interesting idea, there is no actual evidence. A simpler explanation is that the complex uses a different pathway for oligomerization, or even that multiple pathways exist, and if one pathway is blocked then another can be readily utilized. The authors should address this issue more directly or present data showing that the observed complex is truly a part of the oligomerization pathway.
4) Another issue that needs to be addressed more directly is how the present structure fits into the Rev hexamer structure and the previously described jellyfish model. Is it possible to accommodate the second Rev-Rev interface seen in the Rev-only crystal structure with the Rev dimer conformation seen in the present structure? Is it possible to envision a plausible hexamer structure in the context of RNA with the present dimer interface?
5) The Wang lab at NIH has proposed a specific conformation of a Rev-RRE complex based on SAXS data. How does the present structure fit into that model?
6) The final figure is problematic. It is certainly fine to speculate, but while it is formally plausible that the RRE may guide Rev into an array of dimers with multiple crossing angles upon protein-RNA contact, there is as yet no experimental basis for this and the hexamers modeled here are entirely speculative. It is equally plausible, for example, that crossing angles similarly acute to the one reported herein are ubiquitously used for RNP assembly, with larger angles used for interaction with other host-cell proteins (or other variations on the same theme). This figure should be deleted or, alternatively, replaced with a figure that explains the authors’ current general model, particularly in light of the accompanying Crm1 dimer structure.
Other issues for the authors' consideration:
1) Title. Change “Rev oligomer” to “Rev dimer”. Throughout the manuscript there should be careful usage of “Rev oligomer”, which, in at least some places here and in the accompanying manuscript, is alluding to (at least) a hexamer, and in other places to a Rev dimer. Say “dimer” when “dimer” is meant. Similarly, we suggest sticking to the term “subunit” when a single polypeptide chain is meant and not alternating with “molecule”. A dimer is also a molecule.
2) Final sentence of the Abstract: “our studies support an RRE-directed Rev oligomer assembly, where RRE utilizes the inherent plasticity of Rev interfaces and optimally positions Rev subunits to match the architecture of the Crm1 export complex”. Omit. There is nothing here about Crm1 nor about a Rev oligomer larger than a dimer.
3) The three mutations in the crystallized Rev construct should be indicated in Figure 1. The position of the oligomerization-inhibiting mutations should be indicated on the structure.
4) Figure 1–figure supplement 1 shows gel shift assays of Rev with RRE-Stem II and IIB40. It is clear the RNA below is a single species. However, the first observed shift shows a doublet band in the IIB40, and a smeared/fuzzy band in stem II. The authors should note this somehow as these are both evidence of conformational heterogeneity within their samples (the first being the presence of two stable conformations and the second a constant switching between two conformations). Further this doublet nature appears to be resolved upon the second Rev binding, indicating a resolving of this phenomenon.
5) The authors use a 240 nt RRE. They should state that the full RRE is ∼351 nt, and the smaller construct has been shown to be proficient, though not optimal, in nuclear export.
6) The Rev ARM-RNA IIB RRE interface generally agrees well with the analogous interface presented in the 1996 Battiste et al. NMR structure, but the crystal interface is described as being subtly different in detail. It seems equally likely, however, that the solution structure simply lacked precision, particularly given the state of NMR structural studies at that time. This should probably be acknowledged explicitly (or, if the authors wish to make a point of the differences, they should show that their current structure is incompatible with the earlier NMR data).
7) Figure 2C–figure supplement 1 is not useful in its current form. The intent seems to be to demonstrate more clearly than in Figure 2E a binding defects for Rev-Q51A. However, the effects are modest and it is nearly impossible for the reader to draw any conclusions from the data presented.
8) The authors should show the raw western blots that underlie the normalized data in Figure 4C. The results seem clear enough, but it is important to show that the Rev mutations do not destabilize the protein significantly, and this information is lost through the normalization procedure.
9) “… enables a virus with limited coding capacity to build a large, asymmetric RNP using a small, homooligomeric protein to achieve unprecedented structural and functional complexity.” This is really just hype, given that the structure of the large asymmetric RNP is not known nor is its structural and functional complexity.https://doi.org/10.7554/eLife.04120.019
- Alan D Frankel
- Bhargavi Jayaraman
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We are grateful to V Paavilainen, O Rosenberg, J Finer-Moore, R Stroud, L Beamer, J Fraser, C Waddling and G Meigs for assistance with crystallography and structure determination, B Adams for help with protein and RNA purification, J Fernandes for providing the Gagpol-SF2 RRE reporter construct and C Smith for help with reporter assays and viral replication assays. We thank D Booth, M Daugherty, J Fernandes and members of the Frankel lab, J Gross, L Beamer, J Fraser, Y Bai and J Doudna, for helpful discussions and comments on the manuscript. We also thank Yun-Xing Wang for providing us the coordinates for the RRE SAXS envelope. This work was supported by NIH grant P50GM082250 grant to ADF and a CHRP postdoctoral fellowship F09-SF-204 to BJ.
- Wesley I Sundquist, Reviewing Editor, University of Utah, United States
© 2014, Jayaraman et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Downloads (link to download the article as PDF)
Download citations (links to download the citations from this article in formats compatible with various reference manager tools)
Open citations (links to open the citations from this article in various online reference manager services)
Two new structures shed additional light on the nuclear transport of viral transcripts.