Differential expression of porcine microRNAs in African swine fever virus infected pigs: a proof-of-concept study

Tissue samples analyzed in this study were derived from an experimental ASFV infection included in the investigation of ASFV pathogenesis and vaccine development [10]. Two different viruses were used. First, the España- 75 (E75) virulent strain (accession number FN557520), a very infectious and virulent strain causing almost a 100% mortality classified within p72 genotype I. This virus was isolate in an outbreak in Lerida in 1975 [1, 22]. And secondly, the E75CV1, an attenuated strain derived from the E75 strain obtained by cellular passage in a stable line of monkey kidney fibroblasts, the kidney of an adult male African green monkey (Cercopithecus aethiops) [23].

From the in vivo infection, available samples of four 8-week-old Landrace x Large White pigs were used. Submandibular lymph node (SLN) and spleen samples from two pigs intramuscularly inoculated with 10⁴ HAU of the E75 virulent strain [10] and SLN and spleen samples from two pigs inoculated with same amount (10⁴ HAU) of the E75CV1, were used. One animal inoculated with the virulent strain and two with the attenuated strain were necropsied at 3 days post-infection (dpi). At 7 dpi, one animal inoculated with the virulent strain was necropsied. All animal experiments conducted by Lacasta et al. (2015) were performed at CReSA facilities under biosafety level 3 (BSL-3) and all procedures were carried out in strict accordance with the guidelines of the institutional animal ethics committee of Universitat Autònoma de Barcelona (UAB), with the permit number DMAH-5796 for this specific study, according to the Spanish and European animal experimentation ethics law. Animals were euthanized following the European Directive 2010/63/EU, using an anaesthetic overdose of 60–100 mg of pentobarbital per kilogram of weight administered via the vena cava.

SLN and spleen samples were collected and immediately frozen at −80 °C.

Tissue samples of approximately 100 mg were homogenized with a pestle in 1 mL of Trizol (Thermo Fisher Scientific, Massachusetts, United States) to lyse the tissues and deactivate virus. Total RNA and DNA were isolated following the manufacturer’s protocol and resuspended in 25 and 200 μL of RNAse-, DNAse- and proteinase free water (Sigma- Aldrich, St. Louis, United States) for RNA and DNA, respectively.

ASFV qPCR on both tissues was carried out in order to quantify viral load. SYBR Green qPCR was the selected procedure with pF1 primer (5′ CCTCGGCGAGCGCTTTATCAC 3′) as forward primer and pR2 (5′ GGAAACTCATTCACCAAATCCTT 3′) as the reverse one that were designed to target a highly conserved region within the p72 ORF as described by [24]. The mix for the qPCR contained 480 nM each primer, 12.5 μl Power SYBR® Green PCR Master Mix (Thermo Fisher Scientific) and 2.5 μl template. Nanopure autoclaved water was added to final volume of 25 μl. The conditions for the amplification were 10 min at 95 °C, 2 min at 50 °C and 40 cycles of 15 s at 95 °C and 1 min at 60 °C. Triplicates of each sample were used for the qPCR.

Total RNA was quantified using ND 1000 Nanodrop® Spectrophotometer (Thermo Fisher Scientific). RNA integrity and quality was analyzed using the 2100 Expert Bioanalizer (Agilent Technologies, Santa Clara, USA). Total and low size RNA measurements were carried out. Total RNA was calculated using the Eukaryote total RNA Nano Series II software and specific chips (Agilent Technologies, Santa Clara, USA), while low size RNA was measured using the Small RNA Nano software and specific chips (Agilent Technologies, Santa Clara, USA).

All procedures were carried out using the miRCat™ microRNA Cloning Kit (IDT, Iowa, USA) protocol as basis, with the modifications described in [25]. For the enrichment of the small RNA fraction, slices were cut from a 12% denaturing (7 M Urea) polyacrylamide gel, using a miSPIKE™ (IDT, Iowa, USA), as internal size marker. Small RNA fractions were purified using Performa DTR gel filtration cartridges (EdgeBio, Gaithersburg, USA). A double ligation was carried out using 3′ and 5′ linkers from miRCatTM kit (IDT, Coralville, USA) in two steps. The 3′ preadenylated linker was coupled to the small RNA fraction in a reaction in which the mix contained 1X Reaction Buffer for T4 Ligase (Thermo Fisher Scientific) without ATP, 0.1 mg BSA (Thermo Fisher Scientific), 1 U T4 RNA ligase (Thermo Fisher Scientific) and 2.5 μM 3′ linker. 6.5 μL small RNA fraction was added and incubated at 37 °C for 2 h. The 5′ linker was coupled in a mix with 1X ligation buffer (Thermo Fisher Scientific), 1 U T4 RNA ligase (Thermo Fisher Scientific) and 50 μM 5′ linker in presence of ATP. 7.5 μL 3′linked RNA was added to a final volume of 10 μL. The mixture was incubated for 2 h at 37 °C.

RT was carried out with Super Script III Reverse Transcriptase (Thermo Fisher Scientific). Initial pre incubation at 65 °C for 5 min with 0.5 mM of each dNTP, 0.5 μM Primer mir5 and 11 μL RNA were carried out. In a second step a mix was prepared with 1X First-Strand Buffer, 5 mM DTT, 0.2 U RNaseOUT and 10 U SuperScript III RT in a final volume of 20 μL, the mix was incubated at 55 °C for 90 min, and 70 °C for 15 min.

PCR amplifications were performed using Expand High Fidelity Plus PCR System (Roche Diagnostics, Basel, Switzerland). Briefly, 0.05 U Expand High Fidelity Plus Enzyme Blend, 1X Expand High Fidelity Plus Reaction Buffer with 1.5 mM MgCl₂, 0.2 μM Primers A5 and B3 (included multiplex identifiers at the 5′ end), 0.2 μM of each dNTP, 4 μL cDNA and RNAse free water to a final volume of 50 μL. PCR conditions were 94 °C for 3 min, 20–25 cycles at 94 °C for 30 s, 57 °C for 45 s and 72 °C for 1 min followed by 71 °C for 7 min. The number of cycles was optimized for each library to avoid saturation. The resulting PCR was purified with the QIAquick PCR purification kit (Quiagen, Hilden, Germany).

Cloning in the pGEM-T Easy Vector System II Kit (Promega Corporation, Wisconsin, United States) and posterior Sanger sequencing with the ABI-PRISM 3700 (Thermo Fisher Scientific) were carried out to check the libraries construction prior to the HTS.

PCR amplifications from each library were quantified using Qubit™ fluorometer, Quant-IT™ (Invitrogen™, Carlsbad, USA) and prepared to a 10¹¹ DNA molecules/μL. As they included multiplex identifiers, they were equimolecular pooled to continue in one reaction. Ion Torrent adapters were ligated to 30 ng of pooled DNA and libraries were then amplified with Ion Torrent primers for 8 cycles and size selected (2% E-Gel Size Select, Invitrogen). Libraries were sequenced following the manufacturer’s protocol with Ion PGM™ sequencer (Thermo Fisher Scientific, Massachussetts, USA) at DNA sequencing facilities at CRAG (Bellaterra, Spain). Software version for base calling was Torrent-Suite v2.0.1, (Thermo Fisher Scientific, Massachussetts USA). Sequencing data were deposited at European Nucleotide Archive (ENA, http://​www.​ebi.​ac.​uk/​ena/​) [26] with the accession number PRJEB17083.

Primer sequences were trimmed and only those insert sequences between 15 and 29 nucleotides and with total number of sequences ≥3 were kept for further analysis.

For porcine miRNA profiling, sequences were compared to all available miRNA sequences (miRBase v21) using local Blast. Parameters were set to 100% identity and up to 4 mismatches allowed at the end of the sequences to solve the miRNA variability on 3′ and 5′ ends [27].

Differences in host miRNA expression were assessed. The total number of sequences obtained for each porcine miRNA was normalized by library size (in counts per thousand) and, then, averaged by group. Fold changes (FC) between groups were calculated using normalized data. Only miRNAs up or down-regulated for all individuals per group were taking into account. miRNAs were considered differentially expressed (DE) with a FC > 5, and highly expressed with a copy number > 80, according with a previous study from our group [28].

For the in silico study of the potential targets, the DIANA-microT-CDS v5.0 web server with an adjusted threshold of 0.7 was used [29, 30]. There were no porcine genes in the databases, so the study was done using the human genome assuming sequence conservation [31]. In order to identify potential targets in the viral genome, the miRanda program was used with the following parameters: -sc 140 –en 20. Potential targets were then studied employing Web-based Gene Set Analysis Toolkit (WebGestal) [32, 33]. The analysis was done using the Kyoto Encyclopedia of Genes and Genomes Database (KEGG) to check the biological pathways in which miRNAs are involved. The selected parameters for the study were the multiple test adjustment by Benjamini and Hochberg [34] and the significance level set at 0.05.

Finally, Cytoscape 3.2.1 software [35] was used in order to build the gene regulatory networks formed by the DE miRNAs and their cellular and viral target genes.

Pigs inoculated with the E75CV1 attenuated strain developed no clinical signs while pigs inoculated with the E75 virulent strain developed typical African swine fever clinical signs [10]. Samples from three animals necropsied at 3 dpi as early times of infection and from one pig inoculated with the virulent strain euthanized at 7 dpi because of the development of severe clinical signs were analyzed. These time points reflect early times of infection (3dpi) where low viral load and mild clinical/ histopathological signs were observed, and, on the other hand late times of infection (7 dpi) where very high viral load was detected with severe and lethal clinical and histopathological signs. At 3 dpi animals presented no fever but a reduction of their activity and mild awkwardness in their movement with few hemorrhages or cyanotic areas and low dyspnea. Histopathology showed mild bleeding, splenic infarcts, hyperplasia, necrosis, lymphocytolysis or vascular damage in spleen. At 7 dpi animals presented high fever with mild movement or even immobile, high percentage of cyanotic areas, cutaneous necrosis and hemorrhages, bloody faeces and dyspnea. Histopathology showed bleeding, lymphoid depletion, splenic infarcts, necrosis and massive hemorrhages in spleen). Ct values in tissue samples determined by qPCR are indicated in Table 1. DNA from attenuated virus was not detected at 3 dpi by qPCR. Animals included in the study of Lacasta et al. (2015), inoculated with the attenuated strain and maintained until 31 dpi showing neither viremia nor clinical signs of acute ASF except fever episodes at 7 dpi that reached undetectable levels by 28 dpi.

RNAs from SLN and spleen were used to prepare small RNA libraries from animals euthanized at 3 and 7 dpi. A total of 8 small RNA libraries from both tissues were constructed and sequenced with an Ion Torrent PGM sequencer. From the 301,802 total reads obtained, after trimming the adaptor sequences, the number of inserts comprising 15 to 29 nt were 105,156. Sequences were aligned to miRBase database (v.21) and a total of 66,599 hits were obtained, corresponding to 366 entries in miRBase.

The normalized count of sequencing reads (reads/total sequencing tags in the library) could be used to quantify miRNA expression levels among early (3 dpi) and late (7 dpi) in both analyzed tissues. miRNAs were considered differentially expressed (DE) with a FC > 5, and highly expressed with a copy number > 80, according with a previous study from our group [28].

From the 14 miRNAs expressed at higher levels (>80 reads) in spleen, 9 miRNAs (64.3%) were DE, two were up-regulated at 7 dpi when compared with 3 dpi (miR-451 and miR-145-5p) and 6 down-regulated (Table 2). Interestingly, miR-451 and miR-145-5p were the most represented miRNAs among all miRNAs, with more than two and one thousand counts found at day 7 for E75-infected pigs, respectively. Among the down regulated miRNAs, miR-126-5p presented the highest FC differences with 225 FC. In SLN, from the 25 miRNAs more expressed (>80), six miRNAs (24%) were DE, two up-regulated at 7 dpi when compared with 3 dpi (miR-181a and miR-145-5p) and four down-regulated (Table 3). As has been shown in spleen, miR-126-5p was the miRNA with the highest FC (41.84) in SLN. A more conserved miRNA expression pattern was observed in SLN than in spleen.

The normalized count of sequencing reads (reads/total sequencing tags in the library) could be used to quantify miRNA expression levels among virulent and attenuated ASFV infected pigs at early times post-infection (3 dpi). miRNAs were considered DE with a FC > 5 and when the up- or down-regulation was conserved for both animals infected with the E75CV1 strain.

From the 22 more expressed miRNAs (>80 reads) in spleen, 7 miRNAs (31.8%) were DE, four up-regulated in virulent ASFV infected animal (miR-92a, miR-126-5p, miR-92c and miR-30e-5p) and 3 down-regulated (miR-125b, miR-451 and miR-125a) (Table 4). In SLN, from the 33 more expressed miRNAs (>80), three miRNAs (9.1%) were DE, two up-regulated in virulent infected animal (miR-30e-5p and miR-500a-5p) and one down-regulated (miR-125a) (Table 5). miR-303-5p and miR-125a presented the same regulation in both tissues analyzed, up and down regulated, respectively. In this case, the miRNA expression seems also to be more conserved in SLN than in spleen at that time post-infection.

Interestingly, only two miRNAs were equally regulated in spleen and SLN when comparing tissues from virulent and attenuated ASFV infected animals at 3 dpi. Thus, while the ssc-miR-125a showed a down-regulation in its expression, ssc-miR-30e-5p was overexpressed in tissues infected with virulent ASFV when compared to animals infected with the attenuated strain.

Target predictions were carried out for DE miRNAs. Ten miRNAs were selected for target prediction according to the highest representation by tissue and conditions: miR-23a, miR-30e-5p, miR-92a, miR-122, miR-125b, miR-126-5p, miR-145-5p, miR-125a, miR-451 and miR-126-3p.

From these ten miRNAs a total of 8774 target genes were identified (Additional file 1). Cellular target genes were functionally analyzed through the KEGG pathways database (Table 6). For 7 miRNAs: miR-23a, miR-30e-5p, miR-92a, miR-122, miR-125b, miR-126-5p and miR-125a, significant pathways were found related to immune response such as B and T cell receptor signaling pathway, natural killer cell mediated cytotoxicity or Fc gamma R-mediated phagocytosis and with some processes related to the pathogenesis and virus-host interaction, like desencapsidation, apoptosis inhibition, autophagy or host DNA damage response. For miR-451, miR-126-3p and miR-145-5p, no pathways were identified, although some of the target genes have been related to ASFV infection.

DE miRNAs were analyzed with the miRanda algorithm in order to identify the possible targets in the ASFV genome. For miRNA expression in animals infected with the virulent ASFV at different dpi, a total of 25 interactions were found for the seven DE miRNAs and the annotated 165 genes in ASFV genome (Fig. 1). Also, 42 interactions were identified with the porcine genes involved in virus-host interactions such as: apoptosis, chemokine receptors, endoplasmic reticulum (ER) stress, transcription factors, autophagy, ATM pathway, virus entry, immune response and replication. For miRNA expression in attenuated and virulent infected animals at early times post-infection, 16 interactions were found for the five DE miRNAs and the annotated genes in ASFV genome (Fig. 2) and 45 interactions were identified with the porcine genes involved in virus-host interaction, like: chemokine receptors, ER stress, apoptosis, transcription factors, autophagy, virus entry and P300 coactivator protein. In a global approach, considering all DE miRNAs by ASFV infection condition, 37 interactions were found between 12 DE miRNAs included and the annotated 165 genes in ASFV genome (Fig. 3). Also, 76 interactions were identified with the porcine genes involved in virus-host interaction as previously described for both conditions.