Various miRNAs are involved in efficient HCV replication

One of the determinants for tissue tropism of hepatitis C virus (HCV) is miR-122, a liver-specific microRNA. Recently, it has been reported that interaction of miR-122 to HCV RNA induces a conformational change of the 5’UTR internal ribosome entry site (IRES) structure to form stem-loop II structure (SLII) and hijack of translating 80S ribosome through the binding of SLIII to 40S subunit, which leads to efficient translation. On the other hand, low levels of HCV-RNA replication have also been detected in some non-hepatic cells; however, the details of extrahepatic replication remain unknown. These observations suggest the possibility that miRNAs other than miR-122 can support efficient replication of HCV-RNA in non-hepatic cells. Here, we identified a number of such miRNAs and show that they could be divided into two groups: those that bind HCV-RNA at two locations (miR-122 binding sites I and II), in a manner similar to miR-122 (miR-122-like), and those that target a single site that bridges sites I and II and masking both G28 and C29 in the 5’UTR (non-miR-122-like). Although the enhancing activity of these non-hepatic miRNAs were lower than those of miR-122, substantial expression was detected in various normal tissues. Furthermore, structural modeling indicated that both miR-122-like and non-miR-122-like miRNAs not only can facilitate the formation of an HCV IRES SLII but also can stabilize IRES 3D structure in order to facilitate binding of SLIII to the ribosome. Together, these results suggest that HCV facilitates miR-122-independent replication in non-hepatic cells through recruitment of miRNAs other than miR-122. And our findings can provide a more detailed mechanism of miR-122-dependent enhancement of HCV-RNA translation by focusing on IRES tertiary structure. Author summary One of the determinants for tissue tropism of hepatitis C virus (HCV) is miR-122, a liver-specific microRNA, which is required for efficient propagation. Recently, it has been reported that interaction of miR-122 with the 5’UTR of HCV contributes to the folding of a functional IRES structure that is required for efficient translation of viral RNA. In this study, we examined the minimum motifs in the seed region of miRNAs required for the enhancement of HCV replication. As a result, we found two groups of non-hepatic miRNAs: “miR-122-like miRNAs” that can bind HCV-RNA at two locations in a manner similar to miR-122, and “non-miR-122-like miRNAs” that target a single site that masking both G28 and C29 in the 5’UTR. The interaction of these non-hepatic miRNAs with the 5’UTR can facilitate not only the folding of active HCV IRES but also the stabilization of IRES 3D structure in order to facilitate binding to the ribosome. These results suggest the possibility of replication of HCV in non-hepatic cells through interaction with miRNAs other than miR-122 and provide insight into the establishment of persistent infection of HCV in non-hepatic tissues that lead to the development of extrahepatic manifestations.

been detected in some non-hepatic cells; however, the details of extrahepatic replication remain 23 unknown. These observations suggest the possibility that miRNAs other than miR-122 can support 24 efficient replication of HCV-RNA in non-hepatic cells. Here, we identified a number of such miRNAs 25 and show that they could be divided into two groups: those that bind HCV-RNA at two locations 26 (miR-122 binding sites I and II), in a manner similar to miR-122 (miR-122-like), and those that target a 27 single site that bridges sites I and II and masking both G28 and C29 in the 5'UTR (non-miR-122-like). 28 Although the enhancing activity of these non-hepatic miRNAs were lower than those of miR-122, 29 substantial expression was detected in various normal tissues. Furthermore, structural modeling 30 indicated that both miR-122-like and non-miR-122-like miRNAs not only can facilitate the formation of 31 an HCV IRES SLII but also can stabilize IRES 3D structure in order to facilitate binding of SLIII to the 32 ribosome. Together, these results suggest that HCV facilitates miR-122-independent replication in 33 non-hepatic cells through recruitment of miRNAs other than miR-122. And our findings can provide a 34 more detailed mechanism of miR-122-dependent enhancement of HCV-RNA translation by focusing 35 on IRES tertiary structure. 36 37 Author summary 38 One of the determinants for tissue tropism of hepatitis C virus (HCV) is miR-122, a liver-specific 39 microRNA, which is required for efficient propagation. Recently, it has been reported that interaction of 40 miR-122 with the 5'UTR of HCV contributes to the folding of a functional IRES structure that is 41 required for efficient translation of viral RNA. In this study, we examined the minimum motifs in the 42 seed region of miRNAs required for the enhancement of HCV replication. As a result, we found two 43 groups of non-hepatic miRNAs: "miR-122-like miRNAs" that can bind HCV-RNA at two locations in 44 a manner similar to miR-122, and "non-miR-122-like miRNAs" that target a single site that masking Introduction 53 Hepatitis C virus (HCV) infects over 71 million people worldwide and is a major cause of chronic 54 hepatitis, liver cirrhosis and hepatocellular carcinoma [1]. One of the most important host factors for 55 HCV infection is a liver-specific microRNA (miRNA), miR-122 [2]. On the other hand, chronic 56 infection with HCV is often associated with extrahepatic manifestations such as mixed 57 cryoglobulinemia, B-cell lymphoma, thyroiditis, and diabetes mellitus [3]. Supported by clinical 58 observations, low levels of replication of HCV-RNA were detected in PBMCs and neuronal tissues in 59 chronic hepatitis C patients [4,5] and the treatment of chronic hepatitis C patients who developed B cell 60 lymphoma by direct-acting antivirals for HCV resulted in the clearance of HCV and lymphoma [6]. 61 These observations suggest that the replication of HCV in non-hepatic cells can be established in 62 miR-122-deficient condition. 63 In general, miRNAs negatively regulate translation of target mRNA through interaction with the 3'UTR 64 in a sequence-specific manner. In this way, miR-122 regulates the expression of genes involved in the 65 maintenance of liver homeostasis, including lipid metabolism, iron metabolism, and carcinogenesis [7, 66 8]. In contrast, miR-122 has also been shown to stabilize HCV-RNA [9] and enhance internal ribosome 67 entry site (IRES)-mediated translation [2, 10, 11] and replication [12] of HCV-RNA through direct 68 interaction with the 5'UTR of HCV [13,14]. The HCV 5'UTR has two binding sites (sites I and II), 69 which are highly conserved among HCV genotypes [15], to which the miR-122 seed sequence can bind 70 [2,13]. In addition, the overhanging regions in miR-122 have been reported to interact with HCV-RNA 71 and are important for HCV-RNA abundance [14]. Importantly, activation of translation via IRES is 72 promoted by the Argonaut-containing miRNA-induced silencing complex (miRISC) [11]. The 73 miR-122-miRISC complex prevents degradation of HCV-RNA by the cellular 5'-3'exonucleases Xrn1 74 4 and Xrn2, and stabilizes the HCV-RNA interaction [16][17][18]. Upon HCV infection, the interaction 75 between miR-122-miRISC and HCV-RNA further results in the sequestration of miR-122 from host 76 mRNA targets, a phenomenon known as the "sponge effect", which may be responsible for the 77 long-term oncogenic potential of HCV infection [19]. Recently, it has been reported that miR-122 has 78 an RNA chaperone-like function that induces the folding of the HCV IRES such that in it can readily 79 associate with the 80S ribosome for efficient translation [20,21] folding in manner that is miR-122-independent but structurally similar to that of miR-122-mediated 90 folding [20,21]. Interestingly, the G28A mutation does not occur under miR-122-abundant conditions. 91 Moreover, in clinical samples obtained from patients infected with HCV genotype 2, the rate of G28A 92 mutation was higher in miR-122-deficient peripheral blood mononuclear cells (PBMCs) than in sera 93 that mainly included hepatocyte-derived HCV [23]. Therefore, in this study, we examined the 94 possibility of HCV-RNA replication in miR-122-deficient cells, such as non-hepatic cells, and of 95 participation of miRNAs other than miR-122 to support replication of HCV-RNA. 96 97 By mutagenesis analysis, we observed miRNA binding in a manner distinct from that of miR-122 that 98 resulted in enhanced HCV-RNA replication. While miRNA-122 binds HCV-RNA at sites I and II 99 using 6 matching nucleotides containing a GAGUG motif in the seed sequence, we observed a pattern 100 wherein one miRNA molecule was bound to a position intermediate to sites I and II using 7 nucleotides. 101 Although the ability of such miRNAs to promote HCV-RNA replication in miR-122 knockout cells 102 was lower than that of miR-122, the expression of the miRNAs significantly enhanced viral RNA

111
Interactions between miR-122 and 5'UTR required for enhancement of HCV-RNA replication 112 The direct interaction of two molecules of miR-122 with the 5'UTR is known to be essential for 113 efficient HCV-RNA replication in hepatocytes ( Fig 1A). One miR-122 molecule binds the HCV-RNA    ). These results suggest that miRNAs binding to a single site masking both G28 and C29 in the 222 5'UTR with a 7-nucleotide match are able to enhance HCV-RNA replication. We denote this binding 223 mode as "non-miR-122 like".

225
Natural non-miR-122-like miRNAs enhance HCV-RNA replication. 226 To find natural non-miR-122-like miRNAs, we screened human miRNAs masking G28 and C29 227 between stem loop I and II of HCV RNA with at least a 7-nucleotide match in the seed region. In this 228 way, two non-miR-122-like candidates (miR-25-5p and miR-4730), with 7 and an 8-nucleotide 229 matches, respectively, were identified ( Fig 3C and S3A Fig). In addition, we used miR-652-3p, with a 230 6-nucleotide match, as a negative control (Fig 3C and S3A Fig) miR-652-3p, which showed no effect on viral replication (Fig 3D). These results suggest that 239 non-miR-122-like miRNAs binding to single site on HCV-RNA between stem loop I and II via a 7-or 240 8-nucleotide match have the ability to enhance HCV-RNA replication.

302
In this study, we have shown that not only miR-122-like miRNAs but also non-miR-122-like miRNAs 303 can enhance HCV-RNA translation (S5B Fig) and replication. (Fig 3D). Recent studies have revealed 304 that binding of an miR-122:Ago2 complex or introduction of nucleotide substitutions, such as the G28A 305 mutation, facilitate efficient translation by making an HCV IRES stem-loop (SLII, "long arm") 306 12 energetically favorable [20,21]. More recently, HCV IRES has been reported to hijack the translating 307 80S ribosome by interaction between the HCV IRES body to the 40S subunit [22]. We hypothesized 308 that the binding of one non-miR-122-like miRNA with a 7-nucleotide match (miR-25-5p) can similarly 309 alter the IRES structure. In contrast, binding of a negative control (miR-652-3p), with only a 6 310 nucleotide match, is predicted not to alter the IRES structure favorably (Fig 3D). In order to test these 311 hypotheses, we docked representative miRNAs to HCV-RNA, and carried out Replica-Exchange All pairwise RMSDs among 5 representative models (Fig 5A) of nt118-333 regions were determined. 336 The bottom and top boundaries of the box correspond to Q1 (25 th percentile) and Q3 (75 th Percentile) 337 quartiles, respectively. The lower and upper whiskers correspond to Q1 -1.5 * IQR and Q3 + 1.5 * IQR, 338 respectively, where IQR = Q3 -Q1. Median value (Q2, 50 th percentile) is indicated by an orange line 339 within the box, and mean value is indicated by a triangle symbol in green. Asterisks indicate significant 340 differences (**P < 0.01, n.s.: not significant). qRT-PCR. Although expression of miR-25-5p, miR-504-3p, miR-574-5p and miR-1236-5p were 354 significantly lower than that of miR-122 in liver, these miRNAs were detected in various normal tissues 355 including brain, thyroid, lung, stomach, small intestine, colon, kidney, liver and bone marrow (S9 Fig). 356 Interestingly, miR-504-3p is expressed mainly in non-hepatic tissues. These observations suggest that 357 not only liver-specific miR-122 but also other miRNAs, such as miR-25-5p, miR-504-3p, miR-574-5p 358 and miR-1236-5p, expressed in various tissues, can facilitate HCV-RNA replication in non-hepatic 359 tissues.

362
In this study, we examined the possibility of involvement of miRNAs other than liver-specific miR-122 363 in the enhancement of HCV-RNA replication and showed that HCV-RNA replication is enhanced by 364 14 the binding of non-hepatic miRNAs. We found that HCV-RNA can be targeted by miR-122-like 365 miRNAs with 6 matching nucleotides containing a GAGUG motif at two sites (sites I and II). Such 366 miRNAs include miR-504-3p, miR-574-5p, miR-1236-5p. In addition, HCV-RNA can also be targeted 367 by non-miR-122-like miRNAs with at least 7 matching nucleotides that can mask G28 and C29 and Although non-miR-122-like miR-652-3p can interact with the HCV 5'UTR via G28 and C29, which 383 promotes the formation of the SLII structure based on 2D predictions, it did not enhance HCV RNA 384 replication ( Fig 3D). We therefore focused on the 3D structural dynamics of the 385 HCV-RNA:miRNA:Ago2 complex. Our 3D structural modeling suggested the importance of 386 conformational stability of the IRES body structure for efficient translation (Fig 5A and 5B). The We also observed that, in the presence of either miR-122-like or non-miR-122-like miRNAs, the 399 emergence of the G28A mutation in HCV was suppressed (Fig 4). Our previous report showed that 400 wild type G28 virus was still detected in miR-122-deficient PBMCs from patients infected with 401 genotype 2a HCV and that the G28A mutant exhibited lower replication than wild type in the presence 402 of miR-122 [23]. These observations suggest that the G28A mutant interacting with an Ago2-miRNA 403 complex, mediated by not only miR-122 but also non-hepatic miRNAs, has disadvantages in stages of 404 the viral life cycle other than translation, such as replication, assembly or particle production in 405 miRNA-abundant conditions. Although the involvement of miRNAs in viral particle formation has not 406 been well-studied, it has been reported that the binding of miRNAs such as miR-155 and miR-92a to From an evolutionary perspective, among seven genotypes of HCV, gt2 is predicted to be the oldest Analysis of the 5'UTR sequence of HCV. For a rapid identification of 5'UTR sequence of HCV, 508 RNAs extracted from 100 µl of virus-containing supernatants or PBMCs were amplified by using a 509 PrimeScript ® RT reagent Kit (Perfect Real Time) (Takara Bio) and 5'RACE was performed by using a Docking simulation and modeling. First, we established the secondary structures of HCV-RNA 536 (residues 1-289), miRNA-122 (Fig 1), miR-504-3p (Fig 2), miR-1236-5p, miR-574-5p, miR-25-5p (Fig   537   3), miR-4730 and miR-652-3p, respectively. In order to predict the intact structures of HCV-miRNA 538 complex, we utilize the cryo-EM structure of HCV (PDB ID: 5A2Q) as restraints to perform iFoldRNA 539 simulation [29,30]. iFoldRNA is an automatic and accurate RNA 3D structure prediction tool, which is 540 developed on the basis of discrete dynamics simulation (DMD) [31,32]. Distances between C1' atoms 541 in 5A2Q are extracted from its 3D structure as restraints to facilitate the RNA modeling by iFoldRNA. 542 iFoldRNA first builds a ring structure composed of all residues in the complex. Then, it conducts a 543 10000-step initial molecular dynamics simulation to build a preliminary structure. Next, iFoldRNA 544 utilizes replica exchange molecular dynamics simulation (REMD) to thoroughly search through the 545 conformation space. The temperatures of the 8 threads dispatched in REMD are distributed from 0.2 to 546 0.4 kT. Subsequently, the structures extracted from all 8 trajectories generated by REMD are ranked 547 based on their respective free energies, and the 1% lowest free energy structures are clustered with a 548 specific RMSD cutoff, which is typically one-tenth the number of residues. Since iFoldRNA utilizes a 549 coarse-grained model to expedite the simulation process, the centroids of these clusters will be subject to 550 all-atom reconstruction processes to generate the final all-atom candidate structures of the 551 HCV-miRNA complex. 552 553 Statistical analysis. The data for statistical analyses are the average of three independent experiments. 554 Results were expressed as the means ± standard deviation. The significance of differences in the means 555 was determined by Student's t-test.