MicroRNA-134 regulates poliovirus replication by IRES targeting

Global poliovirus eradication efforts include high vaccination coverage with live oral polio vaccine (OPV), surveillance for acute flaccid paralysis, and OPV “mop-up” campaigns. An important objective involves host-directed strategies to reduce PV replication to diminish viral shedding in OPV recipients. In this study, we show that microRNA-134-5p (miR-134) can regulate Sabin-1 replication but not Sabin-2 or Sabin-3 via direct interaction with the PV 5′UTR. Hypochromicity data showed miR-134 binding to Sabin-1 and 3 but not Sabin-2 IRES. Transfection of a miR-134 mimic repressed translation of Sabin-1 5′UTR driven luciferase validating the mechanism of miR-134-mediated repression of Sabin-1. Further, site directed mutagenesis of the miR-134 binding site in Sabin-1 IRES relieved miR-134-mediated repression indicating that these regulatory molecules have an important role in regulating the host gene response to PV. Binding of miR-134 to Sabin-1 IRES caused degradation of the IRES transcript in a miR-134 and sequence specific manner. The miR-134 binding site was found to be highly conserved in wild type PV-1 as well as EV71 strains indicating that miR-134 may regulate function of these IRES sequences in circulation.

screens to determine pro-PV or anti-PV host genes have enabled development of enhanced vaccine cell lines 22 to increase vaccine production and aid eradication. Similarly, studies have shown that replication of viruses carrying microRNA response elements (MREs) are abrogated in tissues expressing the cognate miRNA and may aid vaccine development [23][24][25][26][27] . We previously showed that PV replication is modulated by alterations in miR-555 activity 28 . Increasing miR-555 activity inhibits a host factor critical for PV replication, heterogeneous nuclear ribonucleoprotein C1/C2 (hnRNP C) 29 , which inhibits PV replication 28 .
Notably, we previously showed that miR-134 regulated host cell pathways involving Ras-related nuclear protein (RAN) which can control PV replication in a strain and cell type-independent manner 30 . However, post-infection upregulation of miR-134 function inhibited only Sabin-1, but not Sabin-2 replication. In this study, we show that miR-134 can directly target PV replication by targeting IRES mediated translation in Sabin-1, but not Sabin-2. We demonstrate that this strain-specific inhibition is mediated by complementarity between the miR-134 seed site, and the domain 5 of Sabin-1 IRES which inhibits initiation of viral translation. Additionally, we show that miR-134 binding to Sabin-1 UTR causes degradation of the IRES region. Further, sequence analysis of Sabin-1 and EV71 sequences suggests that the miR-134 binding site is highly conserved in circulating PV-1 and EV71 strains but is absent in PV-2 (Sabin-2) strains. These findings are important as they provide a better understanding of virus-host interaction and potentially a solution for therapeutic outcomes.

Results
MiR-134 represses Sabin-1 but not Sabin-2 or Sabin-3 replication. We recently showed that upregulation of miR-134 during PV infection causes a reduction in replication for both Sabin-1, Sabin-2 and Sabin-3, as well as Enterovirus 71 (EV71) 30,31 . We demonstrated that miR-134 downregulation of Ras-related nuclear protein (RAN) inhibits both PV and EV71 replication. EV71 repressed miR-197 causes RAN upregulation leading to increased nuclear transport of viral protein 3D/3CD and hnRNPK supporting increased viral replication 30,32 ; increasing miR-134 led to reduced RAN and repressed PV replication. Upregulation of miR-134 using a miRNA mimic inhibited RAN and siRNA-mediated RAN silencing reduced Sabin-2 replication. In contrast replication of all three PV viruses in miR-134 inhibitor transfected cells was comparable to mimic negative control (MNC1) supporting our hypothesis that miR-134 upregulation abrogated viral replication. While expecting opposite phenotypes with mimic vs inhibitor transfections would be ideal, it is not always observed for multiple reasons (miRNA expression levels, cell cycle state, stability of the miRNA:inhibitor duplex etc). miRNAs can regulate various co-expressed genes due to the complementarity with their target MREs 17 . Commercial miRNA mimics are designed to resist intra-cellular degradation and are stable for long periods. Thus, typical transfections with miRNA mimics can significantly knock down expression of multiple target gene(s)/pathways over the typical 48 h transfection cycle and exhibit off-target effects making it difficult to dissect miRNA regulated pathways that alter viral replication. MiRNAs can also directly inhibit viral replication [33][34][35][36][37][38][39][40][41][42][43] . Inhibition of host protein synthesis following PV infection is well-established 13 . To determine if a miR-134 mimic could target the PV viral genome, cells were infected with Sabin-1, Sabin-2 or Sabin-3 viruses (MOI = 0.01) for 2 h to allow attachment and entry, followed by miR-134 mimic transfection (25 nM). This short transfection period allows examination of the direct effects of miR-134 mimic on Sabin-1 vs Sabin-2/3 infection and is not expected to significantly alter the pathways that would be modulated otherwise. Viral titers were determined at 24 h pi by plaque assay 31 . Transfection with the miR-134 mimic significantly (p < 0.01) modified Sabin-1 replication but had no detectable impact on Sabin-2 or Sabin-3 replication (Fig. 1). These data suggest that miR-134 regulates replication of Sabin-1, but not Sabin-2 or Sabin-3, and suggests that miR-134 regulates host machinery necessary for Sabin-1 replication to proceed.
Since rapid inhibition of viral protein synthesis is needed for controlling PV replication, we hypothesized that miR-134 was differentially inhibiting internal ribosomal entry site (IRES)-mediated translation of Sabin-1 during early stages of PV replication. Analysis of genomic vRNA sequences of Sabin-1, Sabin-2 and Sabin-3 using ViTA (a method for prediction of host miRNA targets on viruses) 44 identified multiple potential miR-134 binding sites in all the three PV strains with the most in Sabin-1 (Table 1). A majority of these binding sites were in the coding regions which are highly conserved across PV strains ( Supplementary Fig. S1). Among all sites, binding in the PV IRES (nt 473-493) was predicted to be most stable (ΔG = −21.8) (Table 1) for Sabin-1. This 5′UTR region containing the IRES has been shown to be crucial for initial PV translation of the polyprotein 45,46 . Comparative studies between attenuated PV vaccine strains and PV progenitor strains that are neurovirulent have identified three major determinants of attenuation at nt 472 47 , 480 48 and 481 49 in this IRES 49 . Additionally, reversion of nt 472 to wild type in an attenuated Sabin-3 strain was shown to restore virulence 50 .
Domain V in the PV IRES (nt 472 onwards) showed distinct differences between Sabin-1, Sabin-2, and Sabin-3. Sabin-1 miR-134 binding prediction involved a contiguous stretch of six Watson-Crick (WC) base-pairs in the miRNA seed site followed by two wobble G:U pairs and was predicted to be stable (ΔG = −11.96) per the DINAmelt server 51 . In contrast, miR-134 binding for Sabin-2 and Sabin-3 was predicted to be less stable (ΔG = −7.3 for Sabin-2 and ΔG = −7.95 for Sabin-3) owing to lesser WC and non-WC base pairing in the seed site ( Fig. 2a-c). Thus, Sabin-1:miR-134 sequences are predicted to rapidly anneal as shown by recent studies demonstrating that a 7 bp complementarity is critical for rapid RNA/DNA duplex formation 52 , and may be key to siRNA/miRNA "seed" based gene silencing in the canonical model. This 7-mer complementarity is also considered core to miRNA recognition of their target mRNAs 53 , thus these data suggest that the Sabin-1, but not Sabin-2 or Sabin-3 IRES sequences, are target for miR-134 regulation. Thus, we hypothesized that miR-134 would be able to form duplexes with Sabin-1, but not Sabin-2 and Sabin-3 IRES.
Inter-/intra-molecular base pairing in DNA/RNA molecules can appreciably reduce UV light absorbance leading to a hypochromic shift 54 . Conversely, non-complementary DNA and RNA strands increase absorbance of the solution and can lead to a hyperchromic UV shift due to additive absorption 55 . To determine if miR-134 interacted and hybridized to the IRES of Sabin-1, oligonucleotides containing these regions from Sabin-1, Sabin-2 and Sabin-3 were synthesized. We assumed that miR-134 and Sabin-1 would cause a "hypochromic" 54 shift due to base-pairing. Heat-denatured solutions containing only Sabin-1, Sabin-2, and Sabin-3 or miR-134 oligonucleotides showed characteristic UV absorbance peaking at ~260 nm as expected (Fig. 2a-c). However, the addition of heat-dissociated miR-134 to Sabin-1, Sabin-2, or Sabin-3 solutions followed by renaturation at physiological temperatures produced a hypochromic shift in Sabin-1 (Fig. 2a) and Sabin-3 (Fig. 2c), but not for Sabin-2, suggesting that miR-134 can base pair with Sabin-1 and Sabin-3 but not Sabin-2 ( Fig. 2a and b). Lack of binding between the Sabin-2 oligonucleotides and miR-134 produced a hyperchromic shift (Fig. 2b) as expected. This is also supported by the higher numbers of Watson-Crick base pairs between Sabin-1 and miR-134 relative to Sabin-2 and Sabin-3 (Table 1).
To validate the specificity of inhibition of Sabin-1, but not Sabin-2 UTR driven luciferase expression by miR-134, we used directed mutagenesis of Sabin-1-pMLR and Sabin-2-pMLR plasmids. Mutant Sabin-1-pMLR plasmids contained a miR-134 complementarity sequence identical to wild type Sabin-2 (S1mut) while mutant Sabin-2-pMLR plasmids contain a miR-134 complementarity sequence identical to Sabin-1 (Fig. 4a). Thus mutant Sabin-1-pMLR constructs should be able to escape miR-134 inhibition while mutant Sabin-2 pMLR plasmids should be inhibited by miR-134. Equal amounts of sequence verified endotoxin-free wild type or mutant plasmids were co-transfected with MNC1/miR-134 mimics into HEp-2 cells as described. Plasmids encoding SEAP were transfected as a control in parallel. Data were normalized to control (MNC1) transfections and expressed as percentage activity relative to control. Expression of SEAP was equivalent across all transfections validating equivalent transfection efficiency. Transfection of the miR-134 mimic substantially inhibited wild type Sabin-1 driven luciferase activity. However, mutating the miR-134 binding site in Sabin-1-pMLR led to a noteworthy increase in luciferase activity (Fig. 4b). Conversely, the miR-134 transfection had little effect on Sabin-2 wild type IRES driven Luciferase activity. However, mutating the Sabin-2 miR-134 region to resemble Sabin-1 allowed for gain of miR-134 binding ability in Sabin-2 and repressed luciferase expression. These data establish that miR-134 can bind to and regulate Sabin-1 but not Sabin-2 IRES and regulate PV gene translation.
Mir-134 binding degrades Sabin-1 IRES. We hypothesized two independent mechanisms of miR-134 inhibition of Sabin-1 UTR driven luciferase expression. In the first mechanism, binding of miR-134 does not alter stability of the IRES transcript but instead prevents IRES mediated translation of the PV-1 polyprotein. Alternatively, miR-134 could alter the stability of the PV-1 IRES and curtail IRES mediated translation of the PV-1 polyprotein completely. To determine if stability of the IRES transcript is altered by miR-134 binding, cells were co-transfected with Sabin-1 wild type/mutant plasmid and miR-134 mimics or control miRNAs as stated previously for 24 h. Total RNA was isolated and treated with DNAse I and quantified with RNA specific fluorescence based sensitive Qubit assay. Equal amounts of total RNA from all treatments were reverse transcribed using a validated Sabin-1 IRES specific reverse oligo. First strand cDNA was amplified with Sabin-1 specific forward and reverse oligos using a high fidelity polymerase to prevent mispriming. Electrophoresis of the PCR products showed a lack of IRES amplification in cells co-transfected with Sabin-1 wild type plasmid and miR-134 mimic but not in case of the mutant plasmid or control miRNA transfected cells (Fig. 5). These data clearly support mechanism 1 where binding of miR-134 at the binding site causes degradation of the Sabin-1 IRES and thus ablates IRES driven translation. Additionally, sequence analysis indicated that miR-134 could also potentially bind and regulate PTBP1 and eIF4B but does not seem likely given the short period of miR-134 mimic exposure.
Further, we investigated if this mechanism is restricted to the type strains used or is conserved among wild type clinical isolates. To determine this, wild type clinical sequences of PV-1, PV-2 and EV71, a related enterovirus were analyzed. Sequences were analyzed using CLUSTAL W. Sequence conservation at the putative miR-134 binding site across all sequences was plotted with WebLogos 3.0 using sequence logos (Fig. 6) where height of each residue is proportional to its conservation. Analysis clearly showed that miR-134 binding site was highly conserved in PV-1 but not in PV-2 sequences. Analysis of EV71 sequences identified three potential miR-134   binding sites in the EV71 IRES all of which were also highly conserved in wild type EV71 strains. These analyses clearly suggest that miR-134 may be able to bind and regulate the translation of circulating PV-1 and EV71 strains.

Discussion
PV is nearly eliminated from the developed world but incomplete vaccine coverage and vaccine-derived PV hinder eradication. OPV has also increased risks of VDPV in the immune compromised individuals necessitating effective anti-PV therapeutics. Strategies that target viral components are effective but carry the risks of revertant or mutant viruses. Thus, host directed therapeutics that reduce viral burden can reduce the reemergence of wild type variants. Among the multiple therapeutic options under investigation, insights into the role of non-coding miRNAs at the host-PV interface can help identify pathways critical for viral replication and novel intervention strategies.
In previous studies, we showed that overexpression of six miRNAs markedly reduced PV replication in HEp-2 cells of which miR-555 was validated to regulate heteronuclear ribonucleoprotein C 28 . We also identified miR-134 as a key anti-PV miRNA and showed that it altered replication of Sabin-1, Sabin-2, and Sabin-3 post-transfection 28,30 . In these studies, we showed that a 48 h miR-134 mimic transfection inhibited replication of PV and EV71 in a virus and strain-independent manner by knockdown of RAN. Among other miR-134 regulated pathways, we showed regulation of RAN to be a major pathway affecting PV replication in a strain-independent manner. However, inducing miR-134 activity following infection repressed only Sabin-1, but not Sabin-2 or Sabin-3 replication (Fig. 1), suggesting an alternative mechanism of regulation. Changes in viral replication upon miRNA modulation were modest but consequential as has been shown with several other studies 35,36,[57][58][59][60] . In this study, we show that miR-134 interacts with nts 472-492 in the domain V of Sabin-1 5′UTR. Sequence analysis identified perfect Watson-Crick complementarity between the miR-134 seed site and Sabin-1 5′-UTR. In case of Sabin-2 and Sabin-3, the miR-134 seed site complementarity contained both Watson-Crick and wobble base pairs reducing the predicted stability of the IRES: miRNA duplex.
Transfection of miR-134 mimic inhibited translation of Sabin-1 UTR driven Metridia luciferase in HEp-2 cells. In contrast, miR-134 did not inhibit translation of Sabin-2 or Sabin-3 UTR-driven reporters. We observed increased luciferase expression for Sabin-3 relative to Sabin-2 in miR-134 mimic transfected cells; the mechanisms involved are presently unknown and outside the scope of this study. These data show that miR-134 can bind to Sabin-1 UTR in a sequence-specific manner to inhibit PV translation. Further, substituting the Sabin-1 miR-134 binding site with Sabin-2 site relieved miR-134 suppression of Sabin-1 IRES driven luciferase. Conversely, Sabin-2 IRES driven luciferase expression was considerably reduced when the Sabin-1 miR-134 binding site was inserted into Sabin-2 plasmid. Analysis of other miR-134 binding sites in the PV genomes showed most to be conserved between strains (Supplementary Fig S1), and thus could not explain the selective inhibition of Sabin-1 by miR-134 observed previously (Fig. 1). Analysis of Sabin-1 IRES transcripts in cells transfected with wild type/ mutant plasmids in the presence of miR-134 mimic/control miRNAs showed that miR-134 caused degradation of the Sabin-1 IRES. The Sabin-1 binding site is highly conserved in circulating Sabin-1 strains as well as in the related enterovirus EV71. This is the first report of a miRNA directly targeting the PV IRES leading to degradation of the IRES transcript and repression of IRES mediated polyprotein translation. These findings implicate a potential prophylactic/therapeutic role for miR-134 in ablating or reducing PV-1 shedding in OPV vaccinees. miR-134 is expressed in multiple human tissues (pancreas, saliva and breast tissue, brain tissue 61 ), as well as in the small intestine, the target tissues for PV replication. miR-134 expression is regulated by neuronal activity in response to multiple cellular and environmental cues 62 . In conjunction with other miRNAs specifically miR-34c-5p, miR-124-3p, miR-132-3p, and miR-212-3p, it has been shown that miR-134 can regulate proteins that regulate transcription upon synaptic activity 62 . In neuronal cells, miR-134 expression is regulated by Sirtuin-1 (SIRT1) and miR-134 regulates synaptic plasticity and memory 63 . MiR-134 also regulates formation of dendrites from neuronal stem cells (NSCs) via regulation of pumilio homolog 2 (Pum2) 64,65 . In a normal physiological state, miR-134 regulates the spinal volume in hippocampal neurons via LIM domain kinase 1 (Limk1) 66 . These findings show that multiple pathways are affected by miR-134 and suggest there may be other ways miR-134 may regulate PV replication.
Translation of the PV polyprotein is dependent on residues in the IRES in the 5′ UTR. PV and other enterovirus IRES sequences form a highly structured 5′UTR organized into 7 distinct domains 67 . This structure is highly conserved across serotypes and strains suggesting that it has an important role in viral polyprotein translation. Mutations or deletions in the different domains of the PV IRES are lethal 67 . Additionally, it was shown that pyrophosphate modified oligomers (PPMOs) targeting the domain 5 region of PV are able to inhibit multiple picornavirus species either by disrupting the IRES secondary structure and preventing interaction with the 40 S ribosomal subunit or by preventing binding of multiple IRES trans-activating factors (ITAFs), or by modifying the poly(RC) binding protein 2 (PCBP2) 68,69 , splicing factor SRP20/SRSF3 70 , EIF4B 71 , ribosomal protein 6 (RPS6) 72 , ribosomal protein 25 (RPS25) 73 , or polypyrimidine tract binding protein1 (PTBP1) which have been shown to be essential for PV replication. Targetscan analysis of miR-134 targets identifies PTBP1, and eIF4B as targets 74 . PTBP1 has been shown to interact with the base of domain V in cooperation with EIF4G 75 and stimulates PV IRES activity 75 . Mutations at nt 472 which are associated with neuro-attenuation are also linked to decreased binding of PTB in neuroblastoma cells 75 . While it is possible that miR-134 overexpression reduces PTBP1 expression, and thus reduces IRES-mediated Luc translation for Sabin-1, it does not explain the lack of reduction for reporter expression for Sabin-2 and Sabin-3. It has also been shown that EIF4B binds to domain V in a PTB-independent but sequence-dependent manner and mutations in nt 482 and 529 of domain V reduce EIF4B binding and PV translation 71 . Since miR-134 interaction with domain V extends from nt 472-493, it is possible that it either reduces EIF4B levels, or prevents EIF4B binding leading to a loss in translation. These data suggest that miR-134 can inhibit PV genome replication in a strain-dependent manner via degradation of viral RNA by targeting the IRES.

Materials and Methods
Cell culture and viruses. HEp-2 cells were grown in high glucose DMEM (Hyclone) supplemented with 5% heat inactivated FBS (Hyclone). The PV strains were kindly provided by Dr. Steve Oberste (CDC, Atlanta) and viral titers were determined by plaque assays on HEp-2 cells and 50% cell culture infectious dose (CCID 50 ) 76 .
Transfection and cytotoxicity assays. RNA inference (RNAi) including miRNA mimics and inhibitors were reverse transfected into HEp-2 cells (8000/well) at a final concentration of 25 nM using Dharmafect-4 (DF4) per the manufacturer's recommendation. Plates were further incubated at 37 °C/5% CO 2 and 95% humidity before assaying for cytotoxicity, infection or RNA isolation. Cells were infected with PV strains, Sabin-1, Sabin-2 and Sabin-3, in 2% FBS for 24 h as described 77 . Cell supernatants were stored till further analysis. Cytotoxicity was measured 48 post-transfection using CytoxGlo assays (Promega) following manufacturer's recommendations and siTOX positive control.
Plaque assays. Plaque assays on HEp-2 cells were used to determine viral titers 78 . Briefly, 1.2 × 10 6 cells/well were plated in six-well plates overnight. Monolayers were washed twice with MEM containing 2% FBS followed by an hour long incubation with 10-fold dilutions of the stock virus/cell supernatants. Inoculum was removed and plates were overlaid with nutrient agarose gel (0.9% agarose + 2X MEM supplemented with 4% FBS in a 1:1 ratio). Assays were incubated at 37 °C/5% CO 2 and 95% humidity for 3 days followed by fixation in 10% formaldehyde and crystal violet counterstaining.

RNA Extraction and RT-PCR.
Total RNA was extracted using RNAzol RT (MRC gene) per the manufacturer protocol. Concentrations were determined using Nanodrop 1000 (ThermoFisher Scientific) spectrophotometry followed by first strand cDNA synthesis using 500 ng RNA across all samples and treatments using miRNA 1 st strand cDNA synthesis kit (Agilent Technologies). Primer sequences used for qPCR were as follows: miR-134-5p (5′-TGTGACTGGTTGACCA GAGGGG-3′ and proprietary adaptor reverse primer).
Analysis of IRES stability. Endotoxin free plasmid DNAs from Sabin-1 wild type (S1wt), miR-134 mutant (S1mut) were transfected into HEp2 cells along with miR-134 mimic/mimic negative control 1 (25 nM) for 24 h. Total RNA was extracted using RNAzol RT as per manufacturer's instructions, quantified and then treated with RNAse free DNAse I (Ambion, Invitrogen, USA) at 37 °C/30 minutes followed by inactivation of the DNAse I with 10 mM EGTA at 65 °C for 10 minutes. DNAse treated RNA was precipitated overnight with 2.5 M LiCl and 2.5 volumes absolute ethanol. DNA free RNA was quantified using RNA specific high sensitivity Qubit RNA assays as per manufacturer's protocol and equal amounts were used for reverse transcription with Protoscript II (New England Biolabs, USA) using oligo Sab-1UTR-Rev (5′-CTGCAGTATGATACAATTGTCTGATT3′). First strand cDNA was next amplified using Sab1-GC-UTRF (5′-GCGCGCGATCTTTAAAACAGCTCTGGGGTTG -3′) and Sab-1UTR Rev above in a 25 µL reaction containing 200 nM each oligo, 2X Q5 Hot Start master mix using the following conditions, initial denaturation 98 °C/30 seconds followed by 25 cycles of 98 °C/10 sec, annealing at 68 °C/30 second and amplification at 72 °C/30 seconds. No RT reactions were run in parallel to determine absence of DNA. PCR products were resolved on a 1% agarose gel in 1X TAE buffer and imaged on a ProteinSimple gel Imager.
Computational analysis of miR-134 binding sites in PV-1 and EV71 genomes. Whole genome sequences of PV-1, PV-2 and EV71 were mined from NCBI. Sequences were aligned using Clustal W on a local machine and alignments were then analyzed for miR-134 binding sites using WebLogo 3.0 79 . List of accession numbers is provided as Supplemental Material 2.