|Home | About | Journals | Submit | Contact Us | Français|
Ligand-responsive RNA mechanical switches represent a new class of simple switching modules that adopt well-defined ligand-free and bound conformational states, distinguishing them from metabolite-sensing riboswitches. Initially discovered in the internal ribosome entry site (IRES) of hepatitis C virus (HCV), these RNA switch motifs were found in the genome of diverse other viruses. Although large variations are seen in sequence and local secondary structure of the switches, their function in viral translation initiation that requires selective ligand recognition is conserved. We recently determined the crystal structure of an RNA switch from Seneca Valley virus (SVV) which is able to functionally replace the switch of HCV. The switches from both viruses recognize identical cognate ligands despite their sequence dissimilarity. Here, we describe the discovery of 7 new switches in addition to the previously established 5 examples. We highlight structural and functional features unique to this class of ligand-responsive RNA mechanical switches and discuss implications for therapeutic development and the construction of RNA nanostructures.
Over the course of the last decades, structural and functional studies have expanded our understanding of the numerous roles that RNA plays in addition to transferring genetic information from DNA into proteins, including its ability to regulate diverse cellular processes.1 Many of these roles involve a switching transformation between 2 different structural states of an RNA each of which exerts a distinct function. The most common of these switches are known as riboswitches which act as genetic regulatory on-off switches in mRNA by directly binding small molecule metabolites. The binding of a ligand to a riboswitch sensor domain induces secondary structure changes within a second domain, the expression platform, which regulates downstream transcription and translation of the mRNA (Fig. 1A). Exhaustive rearrangement of base pairs in the switching sequence results in different secondary structures of riboswitches in the presence or absence of a binding ligand.2-4
Unlike riboswitches, the RNA switches described here are unique in their small size (Fig. 1B) and mechanical responsiveness to ligand binding which gives rise to 2 distinct and stable conformational states implicated with a switching function during viral translation initiation5 (Fig. 1C). The viral RNA switches adopt an L-shaped conformation stabilized by magnesium ions (Fig. 2A, Fig. 3A) or cross-over base pairing (Fig. 3B), and are captured in a straightened conformation by ligands binding in a deep pocket5–12 (Fig. 2B,Fig. 4). The individual conformational states interconvert as a result of the rearrangement of magnesium ions and unpaired bases in the internal loop RNA, and without breaking or new formation of any base pairs in the secondary structure as occurs in riboswitches. The requirement for 2 stable switch conformations where each internal loop base has specific functional interactions in both the ligand-bound and unbound structural state leads to the high level of sequence conservation seen in virus clinical isolates.8,9 Conformational dynamics of the switches is required for biological function as capture in one of the individual states by synthetic inhibitor binding inactivates the switch.9 These switches do not undergo secondary structure changes but rather a simple mechanical switching between 2 distinct and stable conformations that are required as functional states during translation initiation, similar to a light switch whose moving parts translocate and rotate but do not undergo reassembly during actuation.
The archetype of these switches was discovered in the internal ribosome entry site (IRES) of hepatitis C virus (HCV). The 5′ untranslated region of some eukaryotic and viral mRNAs contains a structured IRES element that initiates translation by assembling functional ribosomes directly at the start codon and bypassing the need for 5′ cap recognition, ribosome scanning, as well as the requirement for most initiation factors.13-16 Within the HCV IRES, which is composed of several domains (Fig. 2C), domain II is highly conserved in thousands of clinical isolates7-9 and contains an internal loop in its lower stem, subdomain IIa (Fig. 2C), which is the target for viral translation inhibitors.17 The three dimensional structure of subdomain IIa was determined by x-ray crystallography6 (Fig. 2A, Fig. 3A), which revealed the RNA adopting a 90° bent architecture, in agreement with cryo-EM and NMR studies.18-20 FRET experiments demonstrated that the subdomain IIa undergoes a conformational switch in the presence of benzimidazole viral translation inhibitors.7,17 The co-crystal structure of subdomain IIa in complex with benzimidazole inhibitor 1 showed capture of the RNA in an extended conformation8 (Fig. 2B, Fig. 4). We recently demonstrated that the ligand pocket in subdomain IIa selectively recognizes guanine and serves as a fortuitous binding site for structurally similar benzimidazoles. Based on this observation in conjunction with other circumstantial evidence,21 we hypothesized that a guanosine within an RNA sequence of the viral genome or rRNA acts as a trigger of the IIa conformational switch.5
A number of other viruses in the flavivirus and picornavirus families also contain HCV-like IRES elements with varying degrees of structural and functional similarity but only limited sequence conservation.5,22 Despite the differences, most IRES elements are organized into similar domains and contain analogous domain II elements, 12 of which will be discussed here. The subdomain IIb hairpin loop and internal loop E motifs (Fig. 2C) are conserved in HCV, CSFV, BVDV, AEV, BDV, NPHV, and GPV while SVV, SPV, DPV, GBV, and DHV maintain sequence conservation of their loop E motif but not the IIb hairpin loop. All twelve of these viruses contain a subdomain IIa internal loop switch structure in the 5′- or 3′-proximal strand of the lower domain II stem. The switch motifs range in size from 9–12 bases, 3 to 6 of which are unpaired, and display little to no sequence conservation5 (Fig. 5).
Sequence and structure analyses of HCV-like IRES elements have allowed us to develop consensus models of domain II secondary structure and subdomain IIa switch motifs. X-ray crystallography has been used to determine the 3-dimensional structure of the HCV and SVV switches in the ligand-free state5,6 and the HCV switch in complex with a viral translation inhibitor ligand. The CSFV switch has been characterized by NMR spectroscopy.23
While HCV and SVV IIa switches have quite different sequences and predicted secondary structures (Fig. 1A), their overall 3 dimensional structures, as determined by x-ray crystallography, are almost identical5,7 (Fig. 3A, B). Both switches have a C-G pair (blue,Fig. 3) closing the internal loop at the upper flanking helix. For SVV, this C-G pair is formed by G425 from the 3′-proximal side of the IIa stem, crossing over and forming a Watson-Crick base pair with C375, which diverges from the predicted secondary structure (Fig. 1B) and inverts local strand directionality between G423 and G426 (Fig. 3B). The IIa switches from both HCV and SVV share an A base (A57 in HCV, A374 in SVV) stacking beneath the C (C58 in HCV, C375 in SVV) of the closing base pair. In SVV, 2 C residues (C373, C372) stack beneath A374, while in HCV, a single U base (U56) fills this same space and packs against the ribose of A57. This upper part of the switch, including half of the IIa internal loop and the flanking upper helix, is oriented at a 90° angle relative to the lower helix which connects through perpendicularly arranged bases of the lower IIa loop. The G52-C111 Watson-Crick pair closes the bottom of the internal loop in HCV, and corresponds to the C368-G426 pair in SVV (green,Fig. 3). Stacking above the C368-G426 pair in SVV, a reverse Hoogsteen pair U369○A424 forms in which the Watson-Crick edge of U369 pairs with the Hoogsteen edge of A424 (mauve,Fig. 3B). In HCV, this space is occupied by A53 which interacts with a magnesium ion (mauve, Fig. 3A) and stacks above the G52-C111 closing pair. Two bases (A54, C55) stack above A53 in HCV, which correspond to A370 and C371 stacking above U369 in SVV. All unpaired bases in the IIa motifs from both viruses stack continuously on either of the flanking helices with the exception of U56 in HCV whose stacking interaction may be interrupted by crystal packing, and is seen stacking in some solution conformations observed by NMR.20 The folds of both the SVV and HCV switch motifs are unprecedented in other RNA architectures. Part of the SVV motif (C368-G426, G425, and U369○A424) resembles a UA-handle which is found as a recurring minimal building block in numerous other RNA architectures.24-25 Systematic mutational analyses of residues in the HCV subdomain IIa reveal that some structural sites tolerate change while others do not, generally in agreement with interactions observed for these residues in the crystal structure (Fig. 3C). Consequently, certain combinations of multiple mutations lead to functionally fit switches while the constituting individual base changes may cause reduction in translation activity (Fig. 5).
X-ray crystal structure analysis of the HCV IIa switch in complex with a viral translation inhibitor ligand revealed an elongated conformation of the RNA which maintains all base pairs seen in the bent conformation.8 A deep ligand binding pocket is formed by the rearrangement of magnesium ions and unpaired internal loop bases to form additional base triple interactions. The roof of the binding pocket is formed by base triple U59-A109○A53 with the Watson-Crick edge of A53 pairing to the Hoogsteen edge of A109. The N7 atom of A53 is coordinated to a magnesium ion, unchanged from the ligand-free state of the RNA (mauve, Fig. 4). The U59-A109 pair remains stacked above the C58-G110 pair as it does in the ligand-free state (not shown in Fig. 3B). A53 stacks above the binding ligand (yellow, Fig. 4) which engages in a pseudo-base triple interaction with C58-G110 (blue, Fig. 4). The Hoogsteen edge of G110 recognizes the guanidinium cation-like moiety of the amino-benzimidazole ligand. The floor of the binding pocket is formed by base triple A57○C111-G52, with G52 stacking beneath the ligand, and the sugar edge of C111 (green, Fig. 4) pairing with the Watson-Crick edge of A57. U56, C55, and A54 are all flipped out of the helix with U56 and C55 stacking onto each other. The phosphate group of U56 is rotated into the back of the ligand-binding pocket, where it coordinates 2 magnesium ions (cyan, Fig. 4). One of the metal ions interacts with U59 in the roof of the binding pocket while the other is coordinated to A57 in the floor. An additional hydrogen bonding interaction occurs between the protonated dimethylamino propyl side chain of the ligand and the phosphate group of A109.
IIa RNA switch motifs have been shown to facilitate viral translation initiation by assuming bent and elongated conformations, and directing the IIb hairpin to the E-site at the ribosomal subunit interface.5,18,19 We recently demonstrated that switching between bent and elongated conformations occurs in IIa switches from diverse viruses including HCV, BVDV, CSFV, SVV, and AEV by FRET.5,7 Additionally, IIa switch functionality in initiating HCV IRES translation has been demonstrated using an in vitro translation (IVT) assay and replicon-transfected human cells.5 HCV chimeric IVT reporter constructs obtained by replacing HCV whole domain II with analogous domains from BVDV, CSFV, or AEV yield IRESs proficient in initiating translation, with BVDV domain II displaying near wild-type activity5 (blue, Fig. 5B). Replacement of the HCV subdomain IIa switch only with the analogous motif from SVV, guided by x-ray crystal structures, furnished a fully functional IRES both in an in vitro translation and HCV replicon assay, despite only marginal sequence similarity shared between the viral IIa motifs5 (Figs. 1B and 5A).5A). These experiments established the distinct viral RNA IIa motifs as minimal architectural switch modules participating in a conserved biological function. Inspired by these findings, we set out to search for similar switch motifs in other viruses. Guided by a consensus secondary structure model for HCV domain II-like fragments, we identified 7 new distinct switches in addition to those from HCV, SVV, BVDV, CSFV, and AEV. The new switches were found in the viral IRES elements of GBV, DPV, DHV, NPHV, SPV, GPV, and BDV. All switches, in addition to those from HCV and SVV, demonstrated proficiency in translation initiation when tested in the context of HCV IRES chimeras in the IVT assay (Fig. 5). IIa switches with an internal loop occurring in the 5′ proximal strand of the lower domain II stem were deemed to be HCV-like and were used to replace the HCV subdomain IIa in the chimeric IVT reporter construct as described above for SVV (Fig. 5A). IIa switches with an internal loop in the 3′ proximal strand of the lower domain II stem were deemed to be BVDV-like and were used to replace the BVDV subdomain IIa in the HCV/BVDV domain II chimeric IVT reporter construct as described above (Fig. 5B). IIa switch sequences used to replace respective HCV or BVDV switches are shown in Figure 5. The modular nature of these distinct switches suggests that they have adapted in their different viral contexts to perform corresponding biological functions. It has been suggested that IRES architectures may have been exchanged between viral genomes by horizontal transfer and multiple recombination events.22 Transfer of IRES elements has been observed within viral families and may also occur between families and through capture of cellular RNA elements.22 Widely divergent switches, such as those involving variation of the internal loop motif occurring in the 3′ versus 5′ proximal strand of the lower domain II stem, in distinct viral families, may also have originated from repeated adaptation of independently evolving motifs that eventually converged to provide similar functions.
The IRES subdomain IIa switch has been exploited as a therapeutic target to selectively suppress viral translation with synthetic inhibitors that capture specific conformations of the RNA.5,7,9-12 High levels of conservation in HCV clinical isolates and the presence of a deep solvent-excluded ligand binding pocket render the IIa RNA switch a target for the development of antiviral drugs. In the context of viral IRES function, the IIa switches are involved in the dynamic interplay of ribosomal subunit recruitment and the transitioning from translation initiation to elongation,18-19,23,26-31 which are facilitated by transient interaction with a G residue that captures a conformational change in the IIa RNA.5
It is conceivable that similar ligand-responsive switches occur in other noncoding RNAs to regulate interactions with nucleic acid or protein partners. Mechanical switches that do not undergo global changes in secondary structure, involving the kinetically slow disruption and formation of base pairs as is common in riboswitches, may be preferred for transient regulatory or pausing events that are quickly reversible. While the dynamics of IRES binding and ribosome release during initiation is not yet fully understood, it has been observed that ribosomes stall on the IRES in the presence of IIa-targeting inhibitor ligands or when the IIa switch is deleted.9,23,29 Ligand-responsive mechanical switches provide a simple mechanism to regulate binding and release events of RNA-RNA and RNA-protein interactions contingent on the presence of a trigger ligand.
The IIa RNA switch motifs have also been exploited as rectangular building blocks to rationally design and construct nanostructures with a variety of potential applications. We used the HCV IIa switch to build a self-assembling RNA square that adopts a well-defined structure as revealed by X-ray structure analysis.32 Recently, the same RNA switch has been incorporated into a self-assembling nanoprism.33 Other RNA switches, including the previously discovered SVV subdomain IIa5 and the IIa-like viral motifs described here, provide a growing tool box of bent RNA motifs to use as building blocks for complex RNA nanostructures that self-assemble from short oligonucleotides and have the potential to change conformation in response to selective ligand binding.
In conclusion, we have described what are currently the simplest known ligand-dependent mechanical switching motifs found in noncoding RNA architectures. The HCV IIa switch has been characterized by x-ray crystallography for both the ligand-free and ligand-bound conformation. The crystal structure of a functionally analogous IIa switch from SVV in its ligand-free conformation was described, revealing a structural and functional interchangeability with the archetypical HCV switch despite a lack of sequence conservation. In addition to HCV and SVV, IIa-like RNA switches were discovered in 10 other viruses and were able to functionally replace the HCV IIa switch to create translationally competent IRES elements. These switches contain a recognition site for a guanosine residue and are fortuitous targets for synthetic viral translation inhibitors. Unlike metabolite-sensing riboswitches, these ligand-responsive mechanical switches represent a new class of simple RNA modules that are structurally well-defined in both ligand-free and bound states despite their small size. The viral subdomain IIa-like modules may represent the simplest form of ligand-responsive mechanical switches in nucleic acids. While functioning in a dynamic fashion during viral translation initiation and elongation, these RNA motifs represent a unique therapeutic target as well as well-defined ligand-responsive building blocks for the construction of RNA nanostructures.
No potential conflicts of interest were disclosed.
M. A. B. was supported by a GAANN fellowship from the US Department of Education. We are grateful for support by the National Institutes of Health, grant R01 AI72012, and the UC San Diego Academic Senate, grant No. RM069B. The Biomolecule Crystallography Facility at UC San Diego was supported by the National Institutes of Health, grant OD011957.