Identification of a newly described OsHV-1 µvar from the North Adriatic Sea (Italy)

The surveillance activities for abnormal bivalve mortality events in Italy include the diagnosis of ostreid herpesvirus type 1 (OsHV-1) in symptomatic oysters. OsHV-1-positive oysters (Crassostrea gigas) were used as a source for in vivo virus propagation and a virus-rich sample was selected to perform shotgun sequencing based on Illumina technology. Starting from this unpurified supernatant sample from gills and mantle, we generated 3.5 million reads (2×300 bp) and de novo assembled the whole genome of an Italian OsHV-1 microvariant (OsHV-1-PT). The OsHV-1-PT genome encodes 125 putative ORFs, 7 of which had not previously been predicted in other sequenced Malacoherpesviridae. Overall, OsHV-1-PT displays typical microvariant OsHV-1 genome features, while few polymorphisms (0.08 %) determine its uniqueness. As little is known about the genetic determinants of OsHV-1 virulence, comparing complete OsHV-1 genomes supports a better understanding of the virus pathogenicity and provides new insights into virus–host interactions.

Italy is the third European producer of marine bivalves, with more than 100 000 tons estimated in 2015 [21]. The bivalve farming industry is economically relevant for the regions bordering the Adriatic Sea and, even if the production of the Pacific oyster is at its onset, the global diffusion of infectious OsHV-1 microvariants raises significant concern in the national authorities and farmers.
In this study, we investigated the identity and infectivity of an OsHV-1 virus detected in oysters, diploid C. gigas, produced and farmed in the North Adriatic Sea. Basically, a first supernatant from pooled virus-positive oysters allowed us to propagate the virus in nine subsequent in vivo infection trials, from which we selected a homogenate of pooled gills and mantle to purify total DNA and sequence the whole genome of a new OsHV-1 microvariant, applying next-generation sequencing and de novo assembly protocols (Fig. S1, available in the online version of this article).
In April 2016, initial signs of mortality were observed following an event of reduced water salinity (25 psu; 18-19 C) in oysters no more than 4-5 months of age, farmed in the Porto Tolle area (Po Delta basin, North Adriatic Sea, Italy). No massive mortality appeared, but some moribund and asymptomatic individuals were found to be positive for the presence of OsHV-1 DNA and were used as a source for the subsequent inocula (Table 1). Details on the DNA extraction and quantitative real-time (qPCR) protocols are reported in the Supplementary Materials and methods. Following experimental infection models based on the intramuscular injection of OsHV-1 preparations [22][23][24][25], we set up an infection protocol that aimed to produce and maintain a suitable quantity of virus in vivo, in the absence of mollusk cell lines, and to characterize the Porto Tolle OsHV-1, hereafter referred to as OsHV-1-PT. A batch of about 300 native C. gigas of about 4 cm in shell length and 4-5 months of age, obtained from the Porto Tolle area (Consorzio Cooperative Pescatori del Polesine, Scardovari), was preliminarily demonstrated to be OsHV-1-negative through the testing of 30 individuals with the standard qPCR protocol that was then used to measure the OsHV-1 DNA in the injected oysters. As detailed in the Methods section (see also the Supplementary Materials and methods), up to 13 oysters per trial were tentatively infected (145 in total), while the negative controls (10 per trial) were injected with the same volume of sterile seawater. At the end of each trial, the gills and mantle [25 mg wet weight (w.w.) tissue] were sampled from individual oysters to assess the presence of OsHV-1 DNA (ORF100 region) by qPCR. Starting from the naturally infected oysters, all inocula were freshly prepared by homogenization of the pooled gills and mantle fragments of oysters showing viral titres above 10 6 OsHV-1 copies µl À1 . Native OsHV-1-free oysters were experimentally injected with a minimum viral load of 10 7 DNA copies.
Using the Kaplan-Meier method (reported in detail in the Supplementary Materials and methods) we estimated the oyster survival probability over time, which was found to be 97.3 % on the first day and 75 % on the sixth day post-injection, while no mortality was observed in the control animals ( Fig. 1).
Amongst the samples generated during the nine infection trials, we selected one sample that was rich in OsHV-1-PT (1.8Â10 8 copies µl À1 ) for confirmatory transmission electron microscopy (TEM). Tissues were prepared according to standard procedures, negatively stained with 2 % sodium phosphotungstate solution and finally observed via TEM (Philips 208S). Virions that were compatible with herpesvirus particles in terms of both size and shape were detected (Fig. S2).
The relative amounts of OsHV-1 DNA and C. gigas DNA in such a virus-rich sample were assessed by qPCR with the same set of primers used for the OsHV-1 DNA quantification and with a primer set designed for elongation factor (EF1a), a single-copy oyster gene (see details in Supplementary Materials and methods). The resulting ratio of 25 : 1 copy number between OsHV-1 and C. gigas made us confident in applying a direct next-generation sequencing (NGS) approach to the total DNA, purified from the above- Following library preparation and Illumina sequencing, we generated 3 436 820 paired-end reads (2Â300 bp), which allowed the recovery of 87 582 high-quality reads truly belonging to the order Herpesvirales (2.6 % OsHV-1 DNA to exogenous DNA ratio). The OsHV-1 reads represented a 200-fold base pair sequence coverage and a 279-fold physical coverage of the OsHV-1 genome. The genome was assembled by applying a de novo approach tailored with a scaffolding step on the OsHV-1 reference genome (AY509253), which allowed us to produce five large contigs ranging in length from 2684 to 164 511 bp and to merge them into a continuous sequence with three short 'N' stretches (64 'N' bases in total), two of which were subsequently resolved by Sanger sequencing. The remaining 'N' stretch could not be resolved and its length was estimated solely on the basis of the scaffolding step. The final assembly was 203 983 bp long, with 1 'N' stretch of 26 nucleotides (details are provided in the Supplementary Material).
The OsHV-1-PT genome displayed the same structure as OsHV-1 µVar [13], with an organization that can be represented as TR L -U L -IR L -X-IR S -U S -TR S -X¢ or X¢-TR L -U L -IR L -X-IR S -U S -TR S , due to the impossibility of placing the X region exactly. OsHV-1-PT is characterized by the five large deletions and the large insertion discriminating the OsHV-1 µVar from the OsHV-1 reference [13]. The latter insertion had previously been detected in both the acute viral necrosis virus (ANVN) and Scapharca broughtonii ostreid herpesvirus-1 (OsHV-1-SB) genomes [19,20]. The 86 bp insertion, found in the OsHV-1 µVar genome when compared to the OsHV-1 reference, was missing in OsHV-1-PT genome; instead, in the inverted repeat IR S /TR S we found two additional deletions of 115 (starting at nucleotide 191 861 of IR S and at nucleotide 199 578 of TR S ) and 235 bp (starting at nucleotide 192 125 of IR S and at nucleotide 200 107 of TR S ). The assembled OsHV-1-PT shared 122 indels with the OsHV-1 µVar genomes, accounting for 1363 nucleotides, the majority of which (82.8 %) were short in length (<10 bp). The localization of most indels (82 %) in repeated sequence motifs (TR/IR) is not surprising as the performance of the de novo approach is well known to be difficult to apply on repeat-containing regions.
The open reading frame (ORF) prediction resulted in 125 different putative OsHV-1-PT proteins, including 111 unique ORFs and other 14 ORFs that were repeated twice in the genome because located in the IR regions (Table 2). We compared the ORFs of OsHV-1-PT (ORF PT ) with those already described for OsHV-1 µVar (ORF V ).    All putative ORFs were functionally annotated using the NCBI NR protein database, Gene Ontology (GO) [26,27] and the Kyoto Encyclopedia of Genes and Genomes (Table 2). It was possible to assign definitions [28] to 119 of the ORFs (95.2 %) by homology search. We also assigned GO terms to 45 ORFs (36 %), Enzyme Commission numbers and InterPro GO terms to 7 and 8 ORFs, respectively. As a result, it was possible to assign a definition to two of the seven new predicted ORFs (PT1-PT7) and to assign GO terms to PT1, revealing its putative function as an integral membrane protein.
To better investigate the genotype of the Italian OsHV-1-PT, a phylogenetic analysis based on the C region, currently regarded as the most variable region, was performed according to previously published studies [4,12]. The OsHV-1-PT sequence of the C region, including ORFs 4/5, was compared with all OsHV-1 sequences retrieved from GenBank and representing different geographical areas. As expected, all the OsHV-1 microvariant sequences and OsHV-1-PT clustered together, although with a bootstrap value lower than 70 (data not shown). The progressive whole-genome sequencing of new Malacoherpesviridae viruses should produce a more refined phylogenetic classification and provide support the functional characterization of the OsHV-1 variants currently affecting bivalve hosts.
In conclusion, we demonstrated that the next generation sequencing and subsequent de novo assembly approach represent a valid strategy for reconstructing the genome of a dsDNA virus, such as OsHV-1, with high-confidence, even in case of non-enriched, unpurified samples at relatively low sequencing depth. The availability of Malacoherpesviridae genomes can lead to a real understanding of functional virus features, i.e. the identification of virulence factors in OsHV-1 variants, as well as phylogenetic relationships and the evolutionary origin of mollusk viruses. Owing to the reported sequence features, we propose the Porto Tolle OsHV-1 virus as a new microvariant. Needless to say, additional studies that relate the pathogenic occurrence of OsHV-1 to developmental stages and environmental conditions are needed to fully characterize the pathogenicity of the Italian OsHV-1-PT virus.

Funding information
This study was funded by the H2020 project VIVALDI (Scientific basis and tools for preventing and mitigating farmed mollusc diseases; grant agreement 678589) of the European Commission. We thank the Italian Ministry of Health for supporting the work of G. Z. through RC IZS VE 05/14.

Conflicts of interest
The authors declare that there are no conflicts of interest.