Detection and full genome characterization of two beta CoV viruses related to Middle East respiratory syndrome from bats in Italy

Fresh carcasses of spontaneously dead bats were obtained from a wildlife rehabilitation centre in the context of a virological survey implemented in Northern Italy since 2010. The identification of bat species was made according to morphologic characteristics reported in the illustrated identification key to the bats of Europe [16].

Pools of viscera (lung, heart, spleen and liver) and intestine were homogenized in minimal essential medium (MEM, 1 g/10 ml) containing antibiotics and clarified by centrifugation at 3000×g for 15 min. Viral RNA was extracted from 100 μl of sample using the NucleoMag 96 Virus kit (Macherey-Nagel, Düren, Germany). The RNA was eluted in 100 μl of MV6 elution buffer and stored at −80 °C. CoV screening was performed by a pan-coronavirus one-step RT-PCR method based on degenerate primers that amplified a fragment (180 bp) of the RNA-dependent RNA polymerase (RdRp) gene [12].

Virus isolation was attempted by inoculation with tissue samples of different cell lines such as VERO cells (African green monkey kidney), MARC-145 (foetal monkey kidney), HRT-18 (human colorectal adenocarcinoma), FRhK 4 (foetal rhesus kidney), LLC-Mk2 (rhesus monkey kidney) and TB1 LU (lung, Mexican free-tailed bat, “Tadarida brasiliensis mexicana”). Confluent monolayers of cell lines were inoculated with samples, incubated at 37 °C with 5% CO₂ and observed daily for seven days for the development of cytopathic effects.

For CoV isolation, cell cultures were used with growth media (Eagle’s minimum essential medium (EMEM)) supplemented with 10% fetal bovine plus antibiotics (100 units/ml of penicillin and 100 μg/ml of streptomycin), 0.3% tryptose phosphate broth (Sigma, USA), 0.02% yeast extract (Sigma, USA) and 10 μg/ml trypsin. Twenty-four well tissue culture plates were inoculated with 0.2 ml per well of the clarified pathological material. After adsorption for 1 h at 37 °C, maintenance medium EMEM supplemented with 1% fetal bovine serum and antibiotics (0.8 ml per well) was added without removing the viral inoculum, and the cultures were incubated at 37 °C.

Libraries were prepared following sequence independent single primer amplification (SISPA) with several variations as described by Djikeng et al. [17].

Nine microlitres of extracted RNA was used for reverse transcription reaction using a combination of random (FR26RV-N) and poly T (FR40RV-T) primers tagged with the sequence 5’-GCC GGA GCT CTG CAG ATA TC-3′, using SuperScript III Reverse Transcriptase (Invitrogen, Monza, Italy) following the manufacturer’s instructions.

The second strand of cDNA was synthesized by DNA Polymerase I Large (Klenow) Fragment (Promega, Milan, Italy) using 20 μl of cDNA. Twenty microlitres of Klenow product was amplified by the Expand High Fidelity PCR System (Sigma Aldrich S.R.L., Milan, Italy) using FR20RV-T primer complementary to the sequence tag. Five microlitres of the PCR product was analysed on a 1% agarose gel. The PCR amplicons were purified using the QIAquick PCR Purification Kit (Qiagen, Milan, Italy) following the manufacturer’s instructions and eluted in 40 μl of nuclease-free water.

The purified DNA was quantified in the Qubit 2.0 Fluorometer (Invitrogen) using the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Rodano, MI, Italy) following the manufacturer’s instructions. Five hundred nanograms of purified DNA were digested with EcoRV enzyme (New England BioLabs, Pero, MI, Italy) to remove the tag sequences. Digested DNA was cleaned up adding a 1.8× volume of Agencourt AMPure XP beads (Beckman, Milan, Italy). DNA was quantified by Quibit 2.0 Fluorometer and the library prepared by Ion Xpress Plus gDNA Fragment Library Preparation (Thermo Fisher Scientific) following the standard protocol for 100 ng of DNA. Emulsion PCR was performed using the Ion PGM Template OT2 200 Kit and the sequencing run performed ac-cording to the instructions of the manufacturer (Ion PGM Sequencing 200 Kit v2) (Thermo Fisher Scientific) by Ion Personal Genome Machine (PGM) in Ion 316 Chip v2.

Reads obtained by Ion Torrent sequencer were checked by quality control, cleaned up and trimmed using CLC Workbench version 5.5.1 (www.​clcbio.​com). A de novo assembly was performed using the default parameters and excluding contigs shorter than 1000 bases. Reads were mapped against the full genome using an online tool (Bowtie2, Galaxy Aries) and visualized by IGV software. MEGA7 was used to edit, align nucleotide and amino acid sequences and to calculate the pairwise identities of the genomes and all ORFs that were predicted using the online tool ORF Finder (NCBI, http://​www.​ncbi.​nlm.​nih.​gov/​gorf/​gorf.​html). Complete genome sequences of MERS-CoV and apha- and beta CoVs from bats, human and camels (n. 131) were obtained from the NIAID Virus Pathogen Database and Analysis Resource (ViPR) [18] through the web site at http://​www.​viprbrc.​org/​. Multiple sequence alignment was calculated using the MUSCLE algorithm. The maximum likelihood phylogenetic tree was performed using IQtree software [19] and Model finder to determine the best model according to BIC [20]. Genetic relationships between Italian and SA bats and MERS-CoV were confirmed by comparison of the sequence distances of MERS-CoV and bat-BCoV 2c (SA bat and IT bats) using SSE v1.2 [21]. The complete S protein and the S1 domain responsible for DPP4 recognition (located in a C-terminal 240 residue RBS of IT-bat CoVs) were compared with those of MERS-CoV, HKU4, HKU5 and MERS-related bat CoVs from China and Africa. To better investigate the relationship between MERS-CoV and related bat sequences, a maximum likelihood phylogenetic tree based on the S1 protein was constructed using the IQtree software including only beta CoV sequences. Similarity plots of CoV clade c genomes were generated by using SSE v1.2 using a sliding window of 600 and a step size of 100 nucleotides (nt).

The three dimensional macromolecular structure (3D–MMS) of the DPP4 binding domain in the S protein was predicted using the sequence-homology method that is based on sequences and structures released by the protein data bank (PDB) and visualized by Cn3D v4.3 software [22]. The secondary structure elements are defined based on an ESPript (http://​espript.​ibcp.​fr) algorithm [23] and are labelled in a previous report on the MERS-RBD structure [24].

Two complete genome sequences of Bat-CoV/Hypsugo savii/206645–40/2011 (BatCoV-Ita1) and Bat-CoV/Pipistrellus khulii/Italy/206645–63/2011 (BatCoV-Ita2) were obtained from total RNA extracted from portions of intestine. Comparison of the RdRp sequences of the two samples obtained by Sanger method by Lelli et al. [12] showed 99% nucleotide identity between them; Initial BLAST analysis revealed they were highly similar to the beta CoVs clade 2csequences. The two full genome sizes were 30,048 nt for BatCoV-Ita1 and 30,039 for BatCoV-Ita2, with a G + C content of 39% each.

The genome organization of BatCoV-Ita1 and BatCoV-Ita2 (IT bat CoVs) is identical to that of MERS CoV species encompassing the 10 open reading frames (ORFs) in the order of ORF1ab-spike-ORF3-ORF4ab-ORF5-envelope (E)-membrane (M)-nucleocapsid (N)-ORF8b and the common non-translated sequences identified in CoV genomes at the 5′ and 3′ genomic termini and between ORF5 and the E gene (Table 1).

In ORF1ab the predicted slippery sequence “UUUAAAC” has been observed fitting the consensus motif X_XXY_YYZ (where XXX normally represents any three identical nucleotides; YYY represents strictly AAA or UUU; and Z represents A, C, or U) of nidoviruses involved in synthesis of the replicase pp1ab polyprotein by ribosomal frameshift.

The size and genomic localization of the nonstructural protein (NSP 1–16) encoded by ORF1ab were predicted by sequence comparison with other beta CoV species. Table 2 shows the 15 expected cleavage sites, 11 recognized by the “main protease” 3C–like protease (3CLpro, NSP4–10, NSP12–16), 3 by papain-like protease (PL2pro, NSP1–3) as well as the autocatalytic site (NSP11). The IT bat CoV cleavage sites, recognized by viral proteases, were identical to those of a BatCoV isolated in China (BtVs BetaCoV/SC2013) and differed from the MERS-CoV by one amino acid in the cleavage sites between NSP1/2 and NSP6/7 (Table 3). A predicted leader transcription regulatory sequence (TRS-L), and seven putative transcription regulatory sequences body TRS-B, representing signals for the discontinuous transcription of subgenomic mRNAs (sgmRNAs), have been identified. The two IT bat isolates shared the same TRS-L, the seven TRS-B (Table 1) as well as 98.8% nucleotide identity. Across the whole genome, the percentage of overall nucleotide identity among other beta CoVs was 80% to MERS CoV, 82% to SC2013, 82% to NeoCoV, 81% to PREDICT, 72.4% to bat CoVs HKU4 and 72,5% to HKU5. The genomic sequence identity between IT bat CoVs, MERS and other MERS-related Bat CoVs is reported in Fig. 1; in particular the lowest identity of the IT bat CoVs with the other beta CoV strains was evidenced in the genomic regions encoding Pl2pro (NSP3) within ORF1ab, Spike, ORF4ab and ORF5.

A phylogenetic tree of the complete genomes showed a monophyletic group placed in the Beta2c clade formed by MERS-CoV sequences originating from humans and camels and bat-related sequences. The closest bat sequences are those originating from Africa, Italy and China (Fig. 2).

Comparison of the predicted protein sequences of the IT bat CoVs and the other beta CoVs showed the highest amino acid sequence identities for E (69.5–91.5%), M (81.7–83.7) and N (71.8–83.7%) proteins and the lowest in ORF3 (28.6–53.5%) and ORF 4b (29.9–56%)(Table 4). Comparison of MERS-CoV protein sequences to EriCoV, HKU4 or HKU5 displayed the highest amino acid sequence identities in the E, M and N proteins whereas the lowest were observed in the S and ORF4b.. In particular, the S protein of MERS-CoV showed above 50% identity to the related bat S proteins: 68.8% to IT bat CoVs, 68.6% to SC2013, 67.7% to HKU4, 64.9% to HKU5 and 64% to NeoCoV. However, if we analyse only the RBD region spanning amino acids 367–606 within the S1 subunit, which is the DPP4-interacting region, the percentage of identity to MERS changed to being slightly higher (54.1–55.0%) for HKU4 with respect to other CoVs (HKU5, IT-batCoVs) (52.5–51.7%). The RBD of Neo CoV showed the lowest percentage of identity to MERS (33.5%).

Moreover, the phylogenetic tree based on the S1 protein (Fig. 3) shows four clearly different clades: 2a, 2b, 2c and 2d corresponding to the sequences of human CoVs OC43, SARS-like, MERS-like and HKU9 CoV, respectively. Clade 2c reflects the RBD percentage of identity and was further differentiated into three groups: MERS sequences and the highly related HKU4 sequences form one group; HKU5 and IT-batCoVs sequences were placed in the second group; the third one, which is the most distant, includes African bats and hedgehog sequences.

The secondary and three dimensional structures of the RBD domain of IT bat CoVs were analysed in comparison with the MERS, HKU4 and HKU5 sequences. Two aa deletions located in a region corresponding to the external subdomain of MERS-RBD were found in IT bat CoV-RBDs in the same positions as in HKU5-RBD: three and eight aa in the HKU5-RBD and two and six in IT bat CoVs. The two deletions are located in two regions corresponding to scaffold strands β7 and β8 in the MERS/HKU4-RDB structure (Fig.4a, b). These two β strands together with β6 and 9 form the external subdomain characterized by four anti-parallel β strands that expose a flat sheet-face for receptor engagement [8].