MicroRNA and cellular targets profiling reveal miR-217 and miR-576-3p as proviral factors during Oropouche infection

Oropouche Virus is the etiological agent of an arbovirus febrile disease that affects thousands of people and is widespread throughout Central and South American countries. Although isolated in 1950’s, still there is scarce information regarding the virus biology and its prevalence is likely underestimated. In order to identify and elucidate interactions with host cells factors and increase the understanding about the Oropouche Virus biology, we performed microRNA (miRNA) and target genes screening in human hepatocarcinoma cell line HuH-7. Cellular miRNAs are short non-coding RNAs that regulates gene expression post-transcriptionally and play key roles in several steps of viral infections. The large scale RT-qPCR based screening found 13 differentially expressed miRNAs in Oropouche infected cells. Further validation confirmed that miR-217 and miR-576-3p were 5.5 fold up-regulated at early stages of virus infection (6 hours post-infection). Using bioinformatics and pathway enrichment analysis, we predicted the cellular targets genes for miR-217 and miR-576-3p. Differential expression analysis of RNA from 95 selected targets revealed genes involved in innate immunity modulation, viral release and neurological disorder outcomes. Further analysis revealed the gene of decapping protein 2 (DCP2), a previous known restriction factor for bunyaviruses transcription, as a miR-217 candidate target that is progressively down-regulated during Oropouche infection. Our analysis also showed that activators genes involved in innate immune response through IFN-β pathway, as STING (Stimulator of Interferon Genes) and TRAF3 (TNF-Receptor Associated Factor 3), were down-regulated as the infection progress. Inhibition of miR-217 or miR-576-3p restricts OROV replication, decreasing viral RNA (up to 8.3 fold) and virus titer (3 fold). Finally, we showed that virus escape IFN-β mediated immune response increasing the levels of cellular miR-576-3p resulting in a decreasing of its partners STING and TRAF3. We concluded stating that the present study, the first for a Peribunyaviridae member, gives insights in its prospective pathways that could help to understand virus biology, interactions with host cells and pathogenesis, suggesting that the virus escapes the antiviral cellular pathways increasing the expression of cognates miRNAs.


Introduction
Oropouche Virus (OROV) is the etiological agent of Oropouche fever, an arthropod-borne viral disease characterized by symptoms like fever, headache, myalgia, arthralgia, malaise, photophobia, nausea, vomiting, dizziness, skin rash, and in few cases encephalitis and meningitis [1][2][3][4][5][6][7]. It was first described in Trinidad in 1955 [8] and the first Brazilian strain was isolated from a dead pale-throated three-toed sloth (Bradypus tridactylus) near a highway construction campsite in Belém, Pará state, northern Brazil [9]. It is estimate that more than 500,000 people were infected in at least 30 outbreaks in South and Central America between 1961 and 2009 [8,10,11,12], placing Oropouche fever as one of the most prevalent arboviral disease in some states of Brazil, after Dengue, Chikungunya and Yellow Fever. However, the virus pathogenesis is still obscure, and Oropouche fever is still considered a neglected disease. During urban outbreaks, the virus is mainly transmitted by its major transmission vector, the midge Culicoides paraensis [3,9,13]. Other insect species, like mosquitoes of the genus Aedes and Culex, might also be potential vectors [9]. OROV is classified in the order Bunyavirales, Peribunyaviridae family, Orthobunyavirus genus, as Bunyamwera Virus, La Crosse Virus and the recently discovered Schmallenberg Virus [14]. The order Bunyavirales is the largest virus order, containing several viruses implicated in the etiology of relevant human diseases, such as La Crosse Virus (LACV) and Oropouche Virus (Orthobunyavirus), Rift Valley Fever Virus (RVFV) (Phlebovirus), Crimean-Congo Fever Virus (CCFV) (Orthonairovirus) and the rodent-borne Hantaan Virus (HTNV), Andes Virus (ANDV) and Sin Nombre Virus (SNV) (Orthohantavirus). OROV has a tri-segmented negative strand RNA genome with a small segment (S) that encodes the nucleocapsid protein N and a non-structural protein NSs; a medium (M) segment that encodes the glycoproteins Gc and Gn and another non-structural protein, NSm, and a large (L) segment that encodes the viral RNA-dependent RNA polymerase (RdRP) [15]. Despite its relevance as a human pathogen and its high prevalence in South America, little is known about OROV replicative cycle, pathogenesis and virus-host interactions. A recent study demonstrated that the OROV entry in HeLa cells is dependent on clathrin-coated pits [16]. Another report showed the relevance of MAVS, IRF-3 and IRF-7, components of the innate immune response, in restricting OROV infection in knockout mice models and nonmyeloid cells [17]. Despite that, the virus pathogenesis and the cellular pathways regulated by OROV infection are not known in detail. Gene expression and post-transcriptional regulation is mediated by short non-coding RNAs (microRNAs, miRNAs or miR) that plays important roles during virus replication.
MicroRNAs span between 19-22 nucleotides in length and their first description was made in nematodes [18,19], though now they have been identified in several phyla of plants and animals [20], and even in viral genomes [21]. In mammals, they can be generated from intronic and exonic regions of protein-coding genes or intergenic regions [22]. They can be found as single miRNA genes or in clusters that encodes long precursor molecules, the pri-miRNA, ranging from hundred to thousand nucleotides in length [23,24]. Pri-miRNAs begins to be edited in the nucleus by the enzyme Drosha into pre-miRNAs, shorter 70 nucleotides long molecules with hairpin structures [25,26]. Those pre-miRNAs are exported from the nucleus into the cytoplasm by proteins such as exportin 5 and RAN-GTP [27], and are further processed by Dicer into a 22 nucleotides long double-stranded RNA (commonly referred as miRNA:miRNA Ã ) [28,29]. The double-stranded RNA is loaded into an Argonaute-driven RNA induced silencing complex (RISC), which selects one strand and binds to a target mRNA (commonly in the 3'-untranslated region, or 3'-UTR region) [30,31] by base complementarity. The miRNA interaction with its target mRNA induces gene silencing by degradation (when full complementarity between the miRNA and the target sequence occurs) [32], or translational inhibition (in case of partial complementarity) [33,34]. Since the seed sequence (the minimal complementarity site between miRNA and mRNA) is usually 7-8 nucleotides long, a single miRNA could regulates expression of several genes, as well as a single gene could be regulated by many miRNAs [35,36].
MiRNAs have already been described influencing disease progression, pathogenicity and replicative cycle of several viruses, being either inhibitory or stimulatory of the infection [37,38]. The liver-specific miRNA-122 stimulates HCV translation, stabilizing and protecting the 5'-UTR of viral RNAs from degradation, leading to an accumulation of the same in the cytoplasm [39][40][41][42]. In resting CD4+ T lymphocytes, HIV-1 viral production is impaired by cellular miRNAs that contribute to establish the viral latency [43]. Another miRNA, miR-29a, targets HIV-1 RNA to accumulate in RNA processing bodies (P-bodies), inhibiting virus infection through translation suppression [44]. Even different strains of the same virus can elicit different miRNA regulation responses, as demonstrated for the highly-pathogenic avian-derived Influenza A H7N7 strain and the low-pathogenic swine-derived Influenza A H1N1 strain [45], suggesting that miRNA signature profiles could raise clues about pathogenicity variation.
Concerning miRNA regulation by bunyaviruses, a study with pathogenic and non-pathogenic strains of hantaviruses demonstrated the variation on miRNA profile among the different specie-specific viruses and cell types [46]. Another study with the Hantavirus Respiratory Syndrome (HPS)-causing agent, Andes Virus (ANDV), identified down-regulation of miR-126 expression, a miRNA that acts as regulator of SPRED1 [47]. Increased expression of SPRED1 was suggested to be one of the mechanisms that augment endothelial cells permeability, leading to HPS. A recent study with PBMC of patients presenting acute hemorrhagic fever caused by the Crimean-Congo Hemorrhagic Fever Virus (CCHFV) showed the deregulation of several miRNAs, some of them associated with innate immunity and viral escape mechanisms [48]. The only study with phleboviruses described the association between miR-142-3p and the endocytic vesicle protein VAMP3, suggesting a control mechanism for the intracellular trafficking of Uukuniemi Virus (UUKV) [49].
Due to the scarcity of information regarding the regulation of bunyaviruses by miRNA and the increasing necessity of better understanding of virus-host interactions of relevant emerging pathogens, we aimed to evaluate and identify the cellular miRNA profile and target genes induced by OROV infection in vitro. We demonstrated that miRNAs miR-217 and miR-576-3p, differentially expressed during infection, could be regulating crucial pathways, like innate immunity response, mainly in upstream proteins of interferon-β induction pathway (adaptor and kinase proteins, as well as transcription factors), protein shutoff and apoptosis.

Virus titration
Virus titration was performed by plaque assay in Vero cells plated at 3 x 10 5 cells/well in 12 well plates 1 day prior to infection. After 1 h incubation with the virus, cells were replenished by DMEM supplemented with 1% v/v FBS, 1% v/v antibiotics and 1% v/v carboxymethyl cellulose (CMC) (Sigma-Aldrich), and incubated at 37˚C and 5% CO 2 during 4 days. Cells were fixed with 4% formaldehyde for 20 min at room temperature, washed in Phosphate Buffered Saline (PBS) (Gibco) and stained with 20% v/v ethanol-violet crystal solution for 15 min.

THP-1 PMA treatment
In order to induce monocyte-to-macrophage differentiation, THP-1 cells were stimulated with 100 nM phorbol 12-myristate 13-acetate (PMA) (Sigma-Aldrich) in standard RPMI medium for 24 h or 3 days followed by 5 days incubation at RPMI medium without PMA. Fresh RPMI medium was provided to cells after treatment and before infections. THP-1 derived macrophages cells were infected as described above, at 24 h or 8 days post PMA treatment.

Flow cytometry and immunofluorescence staining
Cells (10 5 cells/sample) were fixed with 4% paraformaldehyde for 20 min and permeabilized in 1% v/v Triton X-100 PBS solution. Blocking was performed in 5% v/v Donkey Serum (Sigma-Aldrich) PBS solution for 1h at 37˚C. OROV infected and uninfected cells were incubated with mouse polyclonal anti-OROV antibody at 1:300 dilution in blocking solution at 37˚C for 30 min. Cells were then washed thrice in PBS and incubated with 2 μg/ml Donkey anti-mouse AlexaFluor 488 secondary antibody (Thermo Fisher Scientific) at 37˚C for 30 min. After incubation with the secondary antibody, cells were washed and resuspended in PBS. Flow cytometry miR-217 and miR-576-3p are proviral factors for Oropouche replication was performed in Accuri C6 cytometer (BD Biosciences). At least 10,000 gated events were counted per experimental replica at FITC channel.

Cell viability assay
HuH-7 were plated at 2 x 10 4 cells/well density on 96-well plate and incubated at 37˚C and 5% CO 2 for 12 h. After that, cells were infected as described above. Cell viability was evaluated by CellTiter-Blue (Promega) according to manufacturer's instructions. The fluorescence was measured at SpectraMax Paradigm Multi-Mode Detection Platform (Molecular Devices).

RNA isolation and quality assessment
Total cellular RNA for microarray and target mRNA RT-qPCR analysis was isolated using Mir-Vana kit (Thermo Fisher Scientific) according to manufacturer's instructions. RNA quantification and quality was assessed by 2100 Bioanalyzer using RNA 6000 Nano kit (Agilent Technologies). Only samples with a RNA Integrity Number (RIN) ! 9.0 were used for microarray. Extraction of RNA for miRNA validation with specific primers was performed using PureLink RNA Mini Kit (Thermo Fisher Scientific) and quantification and integrity were assessed in NanoVue Spectrophotometer (GE Life Sciences). All RNAs were treated with DNase (TURBO DNA-free Kit, Thermo Fisher Scientific) before RT-qPCR experiments to avoid DNA contamination.

MiRNA screening
In order to evaluate the expression profile of miRNAs, an array using Taqman chemistry was performed as follows: 12 h after infection, six independent replicas of mock-infected or Oropouche infected (4 x 10 6 cells/replica at MOI 1) HuH-7 cells were trypsinized (Trypsin 0.25%, Gibco) and the total cellular RNA was extracted and quantified as described above. cDNA was generated using TaqMan MicroRNA Reverse Transcription Kit (Thermo Fisher Scientific) with 100 ng of RNA per sample according to manufacturer's instructions. The cDNA was preamplified using Megaplex PreAmp Primers (Thermo Fisher Scientific) and Taqman PreAmp Master Mix (Thermo Fisher Scientific) as instructed by manufacturer. The qPCR reaction was performed using Taqman OpenArray Human MicroRNA Panels (Thermo Fisher Scientific), Taqman Open-Array Real-Time Master Mix and the OpenArray Accufill system OpenArray real-time robotics (Thermo Fisher Scientific). This platform is able to quantify 754 human inventoried miRNAs.

MiRNA array statistical analysis
R statistical language [50] was used for background correction and data exploratory analysis (Rn intensity cumulative curve and High Resolution Melting-HRM graphs) for each RT-qPCR reaction. For relative expression quantification, a four parameters sigmoidal curve adjustment was done using the qpcR functions in R language [51]. Quantification cycle (Cq) was determined as the relative cycle to second derivative maximum point of adjusted sigmoidal curve (cpD2). The amplification efficiency was determined at the exponential amplification region, at the mean point between relative cycles to the first derivative maximum point and second derivative maximum point of adjusted sigmoidal curve [expR = cpD2-(cpD1-cpD2)], and calculated as the ratio between the expR corresponding cycle fluorescence and the prior cycle fluorescence. For each miRNA, the amplification efficiency was determined as the mean of efficiencies calculated for the corresponding miRNA. Endogenous small-nucleolar RNAs RNU 44, RNU 48 and U6 RNA were candidates for normalization controls selected by the geNorm method [52]. As an alternative normalization method, the normalization factor was calculated by the geometric mean of all miRNA expressed in each sample [53]. For normalized expression comparison miR-217 and miR-576-3p are proviral factors for Oropouche replication between two sample groups, we performed a non-parametric T-test with 1,000 permutations [54]. For three or more groups comparison we used a one-way non-parametric ANOVA with unrestricted permutation (n = 1,000) followed by a non-parametric pairwise T-test mean comparison with permutation (n = 1,000) followed by Bonferroni correction [54]. Results were presented as mean ± S.E.M (standard error mean). Two-tailed p-values in sample groups' comparison lower or equal to 0.01, 0.05 or 0.1 were considered as highly significant, significant and suggestive, respectively. The relationship between sample profiles was investigated by Bayesian Infinite Mixtures Model cluster analysis [55] and represented by 2D heatmap with dendrograms (bi-cluster). For the purpose of display in the heatmap, k-nearest neighbors method (k = 5) was performed to predict the missing values in uninfected cells for miR-217, miR-26a-2-3p and miR-92a-5p. After imputation of the missing values, a scaled (Z-score) normalization was performed (subtracted miRNA mean divided by miRNA standard deviation).

Validation of miRNA through primer designed RT-qPCR
Reverse transcription was performed using miRNA 1st-Strand cDNA Synthesis Kit (Agilent Technologies) and qPCR reactions were made with High-Specificity miRNA QPCR Core Kit (Agilent Technologies) and forward specific primers for each miRNA investigated. Human U6 RNA forward primer (Agilent Technologies) was used as normalization control. All the experiments were done in four independent replicas for each time point and sample group. The qPCR reaction was performed in 7500 Real-Time PCR System (Applied Biosystems). The cycling parameters were set for standard SYBR Green method according to manufacturer's instructions as follow: 95˚C-10 min and 95˚C-10 sec, 60˚C-15 sec, 72˚C-20 sec for 40 cycles. The miRNA forward primers sequences are depicted in S1 Table. Statistical analysis was performed using non-parametrical Mann-Whitney tests.

Gene set enrichment analysis (GSEA)
Ontology enrichment analysis [56] was performed for the predicted targets of the differentially expressed miR-217 and miR-576-3p. The ontologies were enriched mainly to biological processes, molecular function, cellular components, and gene interaction/regulation pathways. Only genes predicted in at least 3 out of 6 databases were considered candidate targets. Gene Entrez id for the predict ontologies were used in the Gene Ontology Database (GO, available at http://www.geneontology.org/), KEGG (available at http://www.genome.jp/kegg/) and REAC-TOME (available at http://www.reactome.org/PathwayBrowser) for this purpose. Only genes over represented in hypergeometric tests with p-value 0.001 were considered.

MicroRNA inhibitors transfection
HuH-7 cells were seeded (10 5 cells/replica) in triplicate into 24 wells plate overnight. Negative control inhibitor, miR-217 inhibitor and miR-576-3p inhibitor (Integrated DNA Technologies) were transfected at a final concentration of 75 nM using 2 μl of Lipofectamine 2000 (Thermo Fisher Scientific) per replica. Green fluorescent short RNA siGLO (Dharmacon, GE Life Sciences) was used to assess transfection efficiency and establish the miRNA inhibitor concentration for inhibition experiments. 3 h post-transfection, cells were infected with OROV at MOI 1 and RNA were extracted for miRNA quantification (6 h post-infection) or target gene and OROV RNA quantification (18 h post-infection) by RT-qPCR. OROV segment S RNA was quantified using primers and probe [57] with Taqman 2x Universal PCR Master Mix (Thermo Fisher Scientific) and normalized by GAPDH using PrimeTime primers and probe mix (Integrated DNA Technologies).

Validation of target mRNA through RT-qPCR
Cells were seeded (10 6 cells/sample) and infected with OROV at MOI 1. The RNA was extracted at 12 h post infection and reverse transcription was performed using High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific) and 1 μg of RNA. Quantitative PCR was done in six replica per condition using 50 ng/well of cDNA on Custom Taqman Array Fast plates (96 well) (Thermo Fisher Scientific) using specific primers and probes and Taqman Fast Universal PCR Master Mix (Thermo Fisher Scientific) according to manufacturer's instructions on 7500 Fast Real-Time PCR System (Thermo Fisher Scientific). Statistical analysis was done as described for microarray using endogenous 18S, GAPDH, HPRT1 and GUSB as normalization genes. For target kinetics SYBR Green PCR Master Mix (Applied Biosystems) and pre-designed PrimeTime primers (Integrated DNA Technologies) were used according to manufacturer's instruction (for primers sequences see S1 Table).

OROV infects hepatocytes and blood cell lines in vitro
In order to expand the knowledge on the range of OROV-permissive cells, blood and hepatocyte cell lines were used to evaluate in vitro infection (Fig 1A). T CD4 + lymphocytes (Jurkat), monocytes (THP-1) and hepatocytes (HuH-7) cell lineages were infected with OROV at MOI 1 and, at 12 h post infection, infectivity was assessed by immunofluorescence using specific antibodies against OROV proteins and virus-positive cells were counted by flow cytometry. At indicated time points, 21% of Jurkat cells were infected, while THP-1 presented no susceptibility to the OROV infection. THP-1 cells can be induced to differentiate into macrophage by PMA treatment, becoming permissive to some viral infections, as described elsewhere [58][59][60][61]. In order to assess if THP-1 cells differentiated into macrophages were permissive to OROV infection, THP-1 cells were treated with PMA for 24 h or for 3 days, followed by incubation in medium without PMA for 5 more days. Differentiation of THP-1 into macrophage-like phenotype was accompanied by microscopy and attachment. At 12 h post infection, 31% and 50% of THP-1 treated with PMA for 24 h or 8 days, respectively, were infected with OROV, suggesting an increasing permissiveness to OROV infection as the cells shift from monocyte to macrophage-like phenotypes (Fig 1A). At the same MOI, HuH-7 cells showed to be more permissive to OROV infection, presenting 90% infected cells at 12 h post-infection ( Fig 1A). Based on this result with HuH-7 cells, and considering previous demonstrations that the liver is an important replication site during experimental OROV infection in hamster [62,63] and mouse [64], we chose the hepatocyte cell line HuH-7 as our in vitro model for further experiments.
To assess the most suitable conditions to ensure that most cells would be infected at indicated time points, HuH-7 cells were infected with different MOIs and the infectivity was measured by flow cytometry (Fig 1B). We reached 30% of infectivity at MOI 0.1 with a plateau of 90% in higher concentrations of virus (MOIs 1, 5 and 10), with no further increase of miR-217 and miR-576-3p are proviral factors for Oropouche replication infectivity levels ( Fig 1B). To assure that cells were still viable for further experiments, we assessed the cytopathic effect at 6, 12, 18 and 24 h post infection with MOI of 1 using Cell Titer-Blue ( Fig 1C). We did not detect cell death associated to the OROV infection at least 18 h post infection. However, only 44% of cells were viable at 24 h post infection. We also quantified the virus titer generated in those cells by plaque assay and at 6 h post-infection, the titer in the supernatant was 6 x 10 3 PFU/ml (Fig 1C, gray line). As the infection progressed, a peak of 8.6 x 10 5 PFU/ml could be detected at 18 h post infection, reaching a plateau with no further increase in viral titer at 24 h post infection (Fig 1C, gray line). Based on these data, we proceeded using HuH-7 cells infected with MOI 1 in subsequent experiments to evaluate the virus-host interactions.

Cellular miRNA signature in HuH-7 cells infected with OROV
MiRNAs can be informative of cellular targets modulated by virus infection. In order to identify candidate cellular pathways differentially expressed in OROV infected cells, we performed an exploratory screening of 754 human miRNAs through probe-based RT-qPCR. MiRNAs expression was evaluated in four uninfected (control) and five OROV infected biological replicas at 12 h post infection. We found thirteen miRNAs differentially expressed upon OROV infection in HuH-7 cells with statistical significance: twelve up-regulated after infection and only one down-regulated (miR-450b-5p) (Fig 2 and Table 1). The reproducibility of effects in miRNAs was indicated by small variance noted among biological replicas, as depicted in the heat map hierarchical dendrogram (Fig 2).
In order to validate the miRNAs that were significantly deregulated in the array (miR-26a-1-3p and miR-576-3p) and to verify the expression of the miRNAs only detected in infected cells in the expression profile array (miR-217, miR-26a-2-3p and miR-92a-1-5p), we designed specific primers for each miRNA and checked its expression by RT-qPCR (Fig 3). Our validation experiments showed the same tendency of the miRNAs panel with an increasing expression of both miR-217 ( Fig 3A) and miR-576-3p ( Fig 3B)  To confirm the robustness of our analysis, we further validated the expression of three other less stable star miRNAs: the highly significant miRNA miR-26a-1-3p and two miRNAs detected only upon infection, miR-26a-2-3p and miR-92a-1-5p (Fig 3C). Those three miRNAs were up-regulated 5.3, 4.5 and 6.3 fold, respectively, at 12 h post-infection in comparison with uninfected cells (p 0.01). Altogether, these results with specific primers to each miRNA corroborate with our large-scale panel data, identifying miRNAs that are modulated during OROV infection showing the same tendency with different approaches. Star miRNA nomenclature corresponds to passenger strands less favorable to processing by RISC with lower likelihood to regulate gene expression [65,66]. As most of those miRNAs are previously annotated as star miRNAs (as example of miR-26a-1-3p previously annotated as miR-26a-1 Ã ) and some prediction database algorithms use proved interaction as criteria for prediction, we only selected miR-217 and miR-576-3p, both mature strand miRNAs, for further target prediction analysis (one detected only in infected cells and the other one up-regulated significantly upon infection in the array, respectively, and both validated).

Cellular target genes regulated by miR-217 and miR-576-3p
To investigate possible pathways regulated by miR-217 and miR-576-3p during OROV infection, we performed target prediction using TargetScan, miRTarget2, PicTar, miRBase, TarBase and miRanda databases. Target genes predicted by at least 3 out of 6 of those databases were miR-217 and miR-576-3p are proviral factors for Oropouche replication considered candidates. We predicted 195 cellular genes to interact with miR-217, miR-576-3p, or both, using that criterion (S2 Table). We used enrichment analysis with GO, KEGG and REACTOME to identify cellular pathways affected by the predicted targets identified with our selection criteria. Our analysis showed the enrichment of cellular pathways related to regulation of cell metabolic processes, cell cycle and differentiation, chromatin stability and RNA metabolism and expression, suggesting that OROV infection possibly affects cell basic processes and RNA-related regulation processes, as expected for a RNA virus (Fig 4). This can be confirmed by the increasing numbers of observed genes (gray columns) compared with the expected numbers (black columns) for each cellular pathway analyzed (Fig 4). All the analysis showed very significant statistical levels with p values < 0,0001.

Validation of miRNAs predicted target genes modulated by OROV infection
We selected 95 target genes, which were either in the group of 195 predicted target genes (92 genes) and/or either were already published as target genes for those miRNAs, to evaluate their expression through RT-PCR in OROV infected hepatocyte cells. The selection criterion was based on the function described in the literature or involvement in relevant biological pathways related to RNA viruses such as: intracellular trafficking, apoptosis, innate immunity, gene expression regulation, antiviral restriction factor, protein synthesis regulation and intracellular signaling. The predicted selected targets and their association with miRNAs are depicted in the Fig 5 interaction network (see also S3 Table for a brief description of targets function).
From the 95 selected genes tested by RT-qPCR analysis we showed only the 18 genes that were differentially expressed 12 h post infection in comparison with uninfected cells (Fig 6A). The majority (16 genes) were down-regulated, corroborating with the opposite up-regulation trend of the related miRNAs during infection. The gene expression of membrane anchor protein ADAM9, the component of SCF (SKP1-CUL1-F-box protein) E3 ubiquitin-protein ligase complex F-box Only protein 11 (FBXO11), the TNF Receptor Associated Factor 3 (TRAF3), the Mitogen-Activated Protein Kinase 1 (MAPK1) and the Mitochondrial Antiviral-Signaling protein (MAVS), all had a trending of (0.05 p 0.1) down-regulation. They were 3.9, 3.9, In order to evaluate if target regulation could present a higher effect in a later point of the infection, we selected two predicted and published targets for miR-217 and three for miR-576-3p to assess their expression 24 h post infection ( Fig 6B). As it was demonstrated that apoptosis is regulated by OROV replication [67], we selected the Mitogen-Activated Protein Kinase 1 miR-217 and miR-576-3p are proviral factors for Oropouche replication (MAPK1) for being a known miR-217 target that regulates apoptosis [68]. Although MAPK1 was not significantly deregulated at 12 h post infection it showed a 2.23 fold down-regulation at 24 h post infection (p 0.05). The three selected and unpredicted miR-576-3p targets, MAVS, TRAF3 and STING, are known to be important genes in the regulation of IFN-β response in viral infected cells [69]. MAVS and TRAF3 did not presented a significant downregulation at 12 h post infection ( Fig 6A); however, both presented significantly down-regulation at 24 h post infection (2 and 7.4 fold, respectively). STING, the only one of the three selected targets of miR-576-3p that already demonstrated a significant down-regulation at 12 h post infection, showed an even higher down-regulation at 24 h post infection (39 fold downregulation at 24 h post infection compared to 7.16 at 12 h post infection). The Silent Information Regulator 1 (SIRT1), a histone deacetylase known to be involved in stress-responsive pathways as inflammation [70,71,72], was a miR-217 target that did not show significant Gene set enrichment analysis (GSEA) was performed for all predicted targets using gene ontology available in GO, KEGG and REACTOME databases. Over-represented pathways are depicted in the y-axis. Quantity of genes associated to a specific pathway is represented in the x-axis. Black columns represent the quantity of genes expected to be associated to a determined pathway and gray columns represent the observed quantity of genes associated to the same pathway, respectively. Significance of overlapping is represented by the negative log of P-value for each pathway.
https://doi.org/10.1371/journal.pntd.0006508.g004 miR-217 and miR-576-3p are proviral factors for Oropouche replication differential expression relative to uninfected cells 12 h post infection but presented a significant down-regulation at 24 h post infection (3.45 fold down-regulated), what reinforces that different target genes of the same miRNA have different dynamics of regulation. Although not proved as a miRNA target yet, we included DCP2, the only selected miR-217 target already significantly down-regulated at 12 h post infection, because of its relevance as a restriction factor for other bunyavirus [73]. In our model, DCP2 kept a decreasing expression in infected cells, being 20 fold significantly down-regulated at 24 h post infection (p 0.001).
Overall, we confirmed the modulation of target genes transcription in the opposite direction of its cognate miRNA, showing that miRNA screening is very informative to predict cellular host genes modulated by virus infection.

IFN-β induction response is attenuated at later stages of infection
The type I interferon response is an important canonical innate immunity response mechanism to viral infection. As STING, MAVS and TRAF3 were demonstrated to be key factors in regulation of that response [69], we aimed to quantify the variation in IFN-β transcripts in response to the infection. The IFN-β mRNA levels increased until 12 h post-infection, when it began to drop abruptly, reaching lower levels at 24 post infection (Fig 7A). Those results are consistent with interferon immune response being triggered at early stages of virus replicative cycle. Virus RNA secondary structures are recognized by RIG-I-like receptors (RLR) or toll like receptors members at early stages of virus replication. However, as the infection proceed, the virus induces the miR-576-3p expression promoting the down regulation of its target

Fig 5. Interaction network between miRNAs and predicted target genes modulated by OROV infection.
Network depicting the relation among miR-217, miR-576-3p and 92 selected target mRNAs found in at least 3 out 6 databases. Green rectangles displays the two selected up-regulated miRNAs; red ellipses represent predicted down-regulated mRNAs; gray ellipse represents mRNA predicted to be regulated by both miRNAs.

Inhibition of miR-217 and miR-576-3p impairs OROV replication
In order to assess if miR-217 and miR-576-3p were playing a role in OROV infection, we aimed to evaluate their impact in OROV replication using specific anti-miRNAs (Fig 8).
To accomplish that, HuH-7 cells were transfected with non-human negative control miRNA inhibitor, miR-217 inhibitor, miR-576-3p inhibitor or both miRNA inhibitors and infected with OROV 3h post-transfection. At least 70% of cells were efficiently transfected with negligible cytotoxicity at the concentration tested (S1 Fig). Both miRNA presented a 3-fold decrease in the presence of its respective inhibitor in comparison with negative inhibitor control (Fig 8A). Nonetheless, the predicted target genes for miR-217 and miR-576-3p, DCP2 and STING, respectively, recovered to similar levels to non-infected cells in the presence of miRNA inhibitors 18 h post-infection (Fig 8B). DCP2 RNA levels were slightly above of those in non-infected cells (up to 0.5 fold) whereas STING mRNA levels did not recovered completely but presented a lower decrease compared to the positive control (1.9 and 7.4 fold decrease, respectively).
To further confirm if the miRNA inhibition would influence OROV replication, we measured the intracellular viral RNA levels 18 h post-infection in the presence of miRNA inhibitors (Fig 8C). Inhibition of miR-217 led to a 2.3 fold decrease in viral RNA replication, while inhibition of miR-576-3p led to a 7.7 fold decrease. The highest reduction was observed with inhibition of both miRNAs (8.3 fold), but was not significantly lower than miR-576-3p miR-217 and miR-576-3p are proviral factors for Oropouche replication inhibition alone (Fig 8C). Finally, reduced viral titers confirmed the diminished replication, as a 3-fold decrease was observed in the same time point using miR-217 and miR-576-3p inhibitors ( Fig 8D). Altogether, those data demonstrate that inhibition of miR-217 and miR-576-3p is a prospective approach to restrict OROV replication in HuH-7 cells.

Discussion
In this study, we aimed to identify miRNAs and the target genes regulated in OROV infected hepatocyte cell lines. We demonstrated that miR-217 and miR-576-3p were up-regulated during infection and that their cognate targets were down-regulated. Gene targets related to apoptosis, type I interferon-mediated response and antiviral restriction factors were associated with those miRNAs, suggesting a post transcriptional modulation of those pathways by OROV infection, giving new insights about virus-host interactions.
We initially investigate the susceptibility of human cell lines to the viral infection (Fig 1). Lymphocytes T CD4+ cells (Jurkat) demonstrated low permissiveness to OROV, though being miR-217 and miR-576-3p are proviral factors for Oropouche replication susceptible to infection in vitro, as denoted by the 20% of infected cells (Fig 1A). On the other hand, the monocyte cell line THP-1 was not infected in the same conditions. Activation with PMA leads to progressive differentiation of THP-1 into macrophage-like phenotype, as demonstrated elsewhere [58,59,60]. We observed an increase in THP-1 infected population under two different PMA treatment conditions, suggesting a higher susceptibility of those cells as they shift to macrophage phenotype. Indeed, a recent case report detected OROV in peripheral blood mononuclear cells of two patients [74], sustaining the possibility of blood cells playing a role in OROV pathogenesis in humans. In mouse models, however, macrophages only sustained viral replication in immune-compromised individual with deletions in IFN genes [17]. The human hepatocyte cell line HuH-7 was highly permissive to OROV infection, corroborating with previous data that suggest a sustainable liver tropism for OROV [17,62,63]. HuH-7 presented a 90% rate of infected cells with no associated cytopathic effect until 18 h post-infection at MOI 1 (Fig 1C), leading us to choose it as our cell model. We initially found 13 miRNAs differentially expressed in infected cells relative to uninfected cells (Fig 2, Table 1). Some of them were induced while others were modulated upon infection, in agreement with the complexity of miRNA regulation network. MiR-217 was already described in cancer cells involved in tumor migration suppression [75,76]. Expression kinetics showed a peak at 6 h post-infection for this miRNA (Fig 3A). Regarding the predicted target genes, ten of them were significantly down-regulated 12 h post-infection in RT-qPCR screening (Fig 6A). SLC1A2, also known as Excitatory Amino Acid Transporter 2 (EAAT2) is a glutamate transporter in astrocytes. Lower expression of this transporter was associated with neuropathogenesis outcomes in HIV-1 [77] and Human Herpesvirus 6 (HHV-6) infected cells [78]. In hepatocytes, an increased expression was associated to cholestasis outcome [79]. It is unclear how this transporter could affect OROV infection in hepatocytes, but considering the neurotropism of OROV infection in vivo, an investigation of its role in neuropathogenesis should be considered. Another possible cellular factor related to neuropathogenesis of OROV is NF1. The regulation of NF1 by other miRNAs was already demonstrated in neurons and other tissues [80], and is considered a mechanism of fine-tuning in neurological disorders such as neurofibromatosis. The transcription factor NFIA was recently demonstrated to be a novel factor that is negatively regulated by miR-373 [81]. As a consequence, IFN-β response is down-regulated, facilitating Porcine Reproductive and Respiratory Syndrome Virus (PRRSV) replication. The data suggest that both miR-217 and miR-576-3p could act synergistically to inhibit IFN-β antiviral response. CLCF1 is a member of Interleukin-6 (IL-6) family and play a dual role as pro-inflammatory and anti-inflammatory cytokine. FNDC3B has a role in cell migration and invasiveness in hepatocellular carcinoma [82] and glioblastoma cells [83]. In the second case, it was shown that FNDC3B could be down-regulated by miR-129-5p. Likely, FNDC3B could be one of many targets which down-regulation leads to an apoptotic state in OROV infection. The chaperone protein CCT6B as well as its relevance in OROV infection remains elusive. The assembly factor COX18 is a key component for cytochrome oxidase complex works properly [84]. Although its regulation showed an opposite trend, we speculate that this phenomenon could be a collateral effect of the apoptosis state, with cells trying to increase cytochrome oxidase efficiency due to a leaking of cytochrome c to cytoplasm.
We further assessed the expression in a later point of infection (24 h) of three targets of miR-217 during OROV infection: DCP2, MAPK1 and SIRT1 (Fig 6B). DCP2 is a decapping protein involved in mRNA decay. It was recently demonstrated that bunyaviruses compete for the same cellular capped mRNAs that DCP2 targets for degradation in a process known as "cap-snatching" [73]. Bunyaviruses need to snatch capped cellular mRNAs in order to replicate the virus RNA genome; therefore, DCP2 is a direct competitor for bunyaviruses replication. Our results suggested that the down-regulation of DCP2 by miR-217 could explain OROV sustained replication. The kinase MAPK1 and protein SIRT1 both presented a significant lower transcription only 24 h post infection (Fig 6B), when most living cells presumably are in apoptosis process.
The miR-576-3p expression peaked at 6 h post-infection and presented a kinetic similar to miR-217 (Fig 3B). The expression levels at 12 h post-infection were consistent in both quantitative assays (e.g. miRNA array and validation RT-qPCR), corroborating our findings (Table 1 and Fig 3B). At the same time point, two candidate targets, FGD4 and CAV2 were down-regulated, confirming the inverse trend of miR-576-3p (Fig 6A). FGD4 is a protein involved in regulation of actin cytoskeleton and cell migration. Another miRNA, miR-155, was associated to reduced FGD4 levels, resulting in impaired neutrophil migration in myelodysplastic syndromes [85].
CAV2 is a protein component of caveolae structures. A recent study demonstrated that the caveolae and, therefore, its components could act as restriction factor for Tiger Frog Virus (TFV) release in late steps of viral cycle [86] in another hepatocyte cell line, HepG2. As OROV entry is mediated by clathrin-endocytosis [16], we speculate that caveolae could be a restriction site for viral budding/release; therefore down-regulation of a structural component could favors viral release.
MiR-576-3p was recently proposed as a key miRNA in feedback regulation of IFN-β pathway in response to viral infections [69]. Our results regarding down-regulation of STING and TRAF3 corroborated that hypothesis (Fig 6). Moreover, IFN-β transcription regulation correlated with miR-576-3p, STING and TRAF3 transcription dynamics (Fig 7), implying in a temporal feedback mechanism in response to OROV infection, as suggested for other viruses. Based on our data and in conclusions of other group [69], we proposed the following dynamics in antiviral response (depicted in Fig 7D): upon viral entry and uncoating, double-strand viral RNA triggers the IFN-β signaling pathway through STING, MAVS and TRAF3 action, leading Interferon Responsive Factor 3 (IRF3) to activate INF β transcription. Concomitantly, miR-576-3p transcription is also activated by the transcription factor IRF3 and the miRNA accumulation increases until peak 6 h post-infection (Fig 3B). When enough miR-576-3p accumulates in cytoplasm (6 h post-infection) the target mRNA levels, mainly for TRAF3 and STING, begin to fall progressively (Fig 7B and 7C)). At 12 h post-infection, as result of the decrease of STING and TRAF3 mRNA levels, the IFN-β response begins to be relieved, starting a feedback mechanism that leads to a halt in antiviral response and sustaining viral replication.
The miR-576-3p is a primate specific miRNA that was conserved along the evolution, presumably, to avoid tissue damage derived from an excessive inflammatory response due to an infection. Indeed, in mice, OROV infection can be successfully controlled by IFN pathway in immune competent individuals. On the other hand, immune compromised mice (e.g. deleted for genes of IFN pathway) have high mortality rates and fast disease progression with notable liver damage [17]. Our results suggested that, unlike mice, the presence of miR-576-3p in primates and repression of INF-ß rendered them more susceptible to OROV infection.
Furthermore, the inhibition of miR-217 and miR-576-3p partially restricted viral replication, as demonstrated by a decreasing in both viral RNA and titer in the presence of miRNA inhibitors (Fig 8). Those results are in accordance with previous data for miR-576-3p inhibition in other viral infections [69]. We speculate that the restriction is a consequence of a longer sustained innate immune response due to lower suppression of IFN-β pathway signaling cascade in a miRNA inhibition scenery, since both miRNAs might regulate target genes of that pathway.
Finally, as NSs protein has been demonstrated elsewhere to be a candidate viral protein that regulates host innate immune response in other bunyaviruses [87]. A recent study demonstrate that a mutant NSs-deleted OROV induces a strong IFN-α production in opposition to the virus with functional NSs [88]. However, sensitivity to IFN-α treatment was not related to the presence of NSs, as both viruses presented similar sensitivity. It was also demonstrated that OROV is more resistant to IFN-α in comparison to BUNV. Although NSs alone seems to be a candidate viral protein to modulate IFN pathway, we cannot exclude the role of other viral or cellular proteins, as well as viral secondary RNA structures in this conundrum.
We focused our analysis on miR-217 and miR-576-3p given the aforementioned reasons; nonetheless, we cannot exclude the possibility that the other miRNAs identified could be playing a role in OROV infection, as we could validated some of them ( Fig 3C). As most of them were star miRNAs and could not be properly investigated by our methodology, a different approach would be necessary to further evaluate if they regulate target genes. Recently, new methodologies to investigate miRNA-mRNAs interactions have been proposed [89,90] and could be an alternative for future studies. We chose the hepatocyte cell line HuH-7 as our model for an initial, representative study. However, it is also possible that at different time points or using different cell models we could identify different miRNA signatures. It would be interesting to compare the miRNA signature among other permissive cells and using different OROV strains to investigate unique and common miRNA responses to the viral infection. Although the targets validated by RT-qPCR are a good indicative of regulation, those assumptions must be considered with caution, as only RNA down-regulation not necessarily reflects a decrease in protein expression coded by the RNA. Protein quantification (e.g. western blot) would be necessary to assure that the final products of those genes are indeed being regulated. We limited the present study to identify miRNAs and their targets regulated during OROV infection, however, the mechanism by which that regulation occurs remains elusive. A further functional study with expression and knockout of viral proteins could shed a light on the role of viral proteins in this mechanism.
To our knowledge, this is the first study to identify candidate miRNAs that could modulate infection of a member of Orthobunyavirus genus, the most representative genus from Peribunyaviridae family. Taken together, the data obtained in this study hint at pathways that could impact OROV infection, replication and pathogenesis, and expand the knowledge of the complex interactions in bunyavirus infections.
Supporting information S1