Structural and functional analysis of the roles of the HCV 5′ NCR miR122-dependent long-range association and SLVI in genome translation and replication

The hepatitis C virus RNA genome possesses a variety of conserved structural elements, in both coding and non-coding regions, that are important for viral replication. These elements are known or predicted to modulate key life cycle events, such as translation and genome replication, some involving conformational changes induced by long-range RNA–RNA interactions. One such element is SLVI, a stem-loop (SL) structure located towards the 5′ end of the core protein-coding region. This element forms an alternative RNA–RNA interaction with complementary sequences in the 5′ untranslated regions that are independently involved in the binding of the cellular microRNA 122 (miR122). The switch between ‘open’ and ‘closed’ structures involving SLVI has previously been proposed to modulate translation, with lower translation efficiency associated with the ‘closed’ conformation. In the current study, we have used selective 2′-hydroxyl acylation analysed by primer extension to validate this RNA–RNA interaction in the absence and presence of miR122. We show that the long-range association (LRA) only forms in the absence of miR122, or otherwise requires the blocking of miR122 binding combined with substantial disruption of SLVI. Using site-directed mutations introduced to promote open or closed conformations of the LRA we demonstrate no correlation between the conformation and the translation phenotype. In addition, we observed no influence on virus replication compared to unmodified genomes. The presence of SLVI is well-documented to suppress translation, but these studies demonstrate that this is not due to its contribution to the LRA. We conclude that, although there are roles for SLVI in translation, the LRA is not a riboswitch regulating the translation and replication phenotypes of the virus.


273
First, using mfold structure prediction (Zuker 2003), we identified two sites within SLVI 274 at which synonymous substitutions could be introduced that should disrupt formation of the 275 structure (Fig. 1A). Substitutions C 436 G and A 439 U in the 5' stem of SLVI (designated 'L' 276 mutants) and U 496 A and G 499  Manuscript to be reviewed 279 prevent the LRA due to disruption of the complementarity with the miR122 seed site 1 (S1).
280 Conversely, 'R' mutants would free the 5' sequences forming the basal stem of SLVI to 281 contribute solely to formation of the 'closed' structure. However, since formation of the 'closed' 282 structure would also be dependent on the S1 site being unoccupied by miR122, we also 283 introduced substitutions into the latter (at positions U 25 A and G 28 C) that were predicted to 284 prevent miR122 binding and at the same time would restore complementarity with the 'L' 285 mutations in SLVI (Fig. 1C, Table 1).

286
To verify that binding of miR122 to S1 was abrogated in the S1 mutants we conducted 287 electrophoretic mobility shift assays (EMSAs) using synthetic miR122 and an RNA 288 oligonucleotide corresponding to the first 50 nucleotides of JFH-1 (JFH1 1-50 ). With the addition 289 of miR122 to unmodified JFH1 1-50 we observed the expected two complexes with reduced 290 mobility, representative of binding of miR122 to both S1 and S2 seed sites. Saturation of both 291 seed sites was achieved upon addition of a 3:1 molar ratio of miR122:JFH1 1-50 ( Fig. 2A), while 292 addition of an antisense miR122 RNA (miR122-Comp) showed no change in mobility (Fig. 2B).
293 In contrast to unmodified JFH1 1-50 , S1-mutated JFH1 1-50 only formed the faster migrating single 294 complex, even at a 5:1 molar ratio, indicating that miR122 remained bound to S2 alone (Fig. 295 2C). Restoration of both mobility-shifted complexes was achieved upon addition of a 50-50 mix 296 of unmodified and S1-modified miR122 (S1-miR122), the latter containing mutations 297 complementary to those introduced in S1 mutated JFH1 1-50 (Fig. 2D). These studies confirmed 298 that substitutions introduced to the S1 site were sufficient to disrupt miR122 binding to the S1 299 seed site, but that binding to the S2 seed site was unaffected, in agreement with similar mutation 300 analysis of miR122 binding (Mortimer & Doudna 2013).

301
To investigate the influence on the conformation of the 5' end of the HCV RNA the L, R 302 and S1 mutations predicted to influence the 'open' or 'closed' conformation were introduced 303 individually, or in combination, into the core-extended translation and replicon reporters, JFH1-304 CEtrans and JFH1-CErep respectively, and individual templates validated by sequence analysis 305 (Table 1).

307
The LRA is detected in the absence, but not presence, of miR122 308 309 We have previously used SHAPE mapping to demonstrate a long-range interaction 310 between the 3'UTR of the HCV genome and distal sequences located within the polyprotein- 318 2012), the 5' end of the S1 miR122 binding site (nts 1-20) proved difficult to accurately map due 319 to excessive background signal. Scrutiny of the predicted pattern of base pairing between the 320 miR122 binding site and miR122 also shows that it is highly similar to that between the miR122 321 binding site and the 5' base stem of SLVI. Together, these issues meant that the S1 region was

333
We first compared the structural conformations of parental JFH1-CEtrans in the absence 334 of miR122 during the RNA folding reaction, or with a 3:1 molar excess of miR122 to saturate 335 binding to S1 and S2, as determined from EMSAs ( Fig. 2A). In the presence of miR122 the basal 336 stem of SLVI was predominantly NMIA-unreactive, indicating that the pairing through this 337 region was in agreement with the structure predicted bioinformatically (Fig. 3A). Indeed, the 352 Under these conditions we observed gross changes to the structure of the basal region of SLVI.
353 The G 434 G 435 motif -predicted to be a key interaction with the S1 site -are highly unreactive,  Having investigated the pairing of the basal stem of SLVI and the occurrence of the LRA 373 in unmodified templates, we went on to study the influence of mutations introduced to prevent 374 these interactions, or that we had previously shown prevent miR122 binding. All subsequent 375 analyses were carried out in the presence of miR122.

394
We next investigated the conformation of templates containing combinations of 395 mutations that were predicted to favour the LRA and the 'closed' conformation: JFH1-CEtrans-396 R, JFH1-CEtrans-S1/L and JFH1-CEtrans-S1/L/R (Fig. 5). Unexpectedly, both JFH1-CEtrans-R 397 and JFH1-CEtrans-S1/L/R failed to demonstrate the LRA, again as evidenced by the reactivity of 398 the G 434 G 435 motif, as well as the overall lack of reactivity in the 3' basal stem of SLVI that 399 would be expected (Fig. 5A, B). As with JFH1-CEtrans-L and JFH1-CEtrans-L/R, we observed These results demonstrate that, while miR122 binding has a profound effect on the

535
Although the strong stem-loop (SLI) in the 5'UTR confounded SHAPE interrogation of 536 sequences forming the S1 site of miR122 binding, those contributing to the basal stem of SLVI 537 were readily mapped. Having determined the influence on miR122 binding of mutations in the 538 S1 site (Fig. 2) we inferred the LRA and formation of the 'closed' structure from exposure or 539 otherwise of the basal stem of SLVI. The LRA was detectable only under very specific 540 conditions, including an in vitro assay in which miR122 was omitted. Similarly, mutagenesis of 541 the template within the S1 miR122 binding site (to prevent miR122 binding) and introduction of 542 complementary mutations to the 5' basal stem of SLVI allowed the LRA to be inferred. In 543 contrast, in the presence of miR122, and/or unmodified sequences at the basal stem of SLVI, we 544 were unable to detect the LRA and formation of the 'closed' structure. We propose that, under 545 conditions in which miR122 is present in significant amounts, the phylogenetically conserved 546 basal stem of SLVI is unlikely to separate to form a long-range association.

547
However, incorporation of 5' (L) and 3' (R) mutations to SLVI did lead to structural 548 changes within the stem-loop that were not predicted by mfold. With the exception of JFH1-549 CEtrans-S1/L, which clearly adopts the 'closed' conformation, all the tested substitutions to the 550 basal stem of SLVI increased the NMIA-reactivity of the structure (Fig. 4 and 5), indicating a 551 reduction in complementary pairing that was more extensive than the sites of modification. In 552 addition, when not paired with the 5' mutations, the 3' mutants (JFH1-CEtrans-R and JFH1-553 CEtrans-S1/R) showed further modification of the SLVI structure with the loss of reactivity of 554 nts 500-501 ( Fig. 5A and 4D). In these cases, it is clear that SLVI had undergone more extensive 555 alteration of base pairing and structure. Previous studies of SLVI ,independent of LRA 556 disruption, were shown to result in alteration to the translation phenotype (Vassilaki et al. 2008).
557 Without a greater understanding of the RNA structure in this region, for example by expanding

571
We investigated translation from a core-extended template transfected into Huh 7.5 cells.
572 As expected from previous studies ), extension of core-encoding sequences -to 573 encompass SLVI -reduced reporter gene expression by ~50%. When normalised to this lower 574 level of translation, the template engineered to adopt the 'closed' conformation (JFH1-CEtrans-575 S1/L) exhibited levels of translation essentially indistinguishable from the control. In contrast, 576 the template unable to recruit miR122 to the S1 seed site (JFH1-CEtrans-S1) and therefore solely . Despite inhibition of miR122 binding at the S1 site, 584 both templates were still able to recruit miR122 to the S2 site (Fig. 2), thus maintaining parental, 585 or near-parental levels of translation. Therefore, templates of known conformation with regard to 'S1/L'   RNA-RNA electrophoretic mobility gel shift assays of miR122 binding to JFH-1 5'UTR.
A synthetic RNA of nts 1-50 of JFH-1 (JFH1 1-50 ) was complexed with increasing molar ratios of (A) wild type or (B) antisense synthetic miR122 and separated by non-denaturing PAGE. JFH1 1-50 mutated at the S1 binding site was similarly complexed with (C) wild type or (D) wild type plus an S1-mutated miR122 and separated by non-denaturing PAGE. miR122 binding was denoted by 1 (+ miR122) or 2 (++ miR122) reductions in RNA mobility compared to    Manuscript to be reviewed

G A U C G U U G G C G G A G U A U A C U U A A G G A A A A C U U C G G A G C G G U C
JFH1-CEtrans-S1 P:   Manuscript to be reviewed  Phenotypic characterisation of JFH-1 reporter bearing S1 and SLVI mutations.