First isolation and genetic characterization of Puumala orthohantavirus strains from France

Puumala orthohantavirus (PUUV) causes a mild form of hemorrhagic fever with renal syndrome (HFRS) named nephropathia epidemica (NE), regularly diagnosed in Europe. France represents the Western frontier of NE expansion in Europe with two distinct areas: the endemic area (Northeast) where PUUV circulates in rodent populations and where many cases of NE are detected in humans and non-endemic area (Southwest) where the virus is not detected and only a few human cases have been reported. The country is a pertinent target to study factors that influence the evolution of PUUV distribution. In this study, we describe for the first time the isolation of two PUUV strains from two distinct French geographical areas: Ardennes (endemic area) and Loiret (non-endemic area). To isolate PUUV efficiently, we selected wild rodents (Myodes glareolus, the specific reservoir of PUUV) from these areas that were seronegative for anti-PUUV IgG (ELISA) but associated with viral RNA load in lung (qRT-PCR). With this design, we are able to cultivate and maintain these two strains in VeroE6 cells but also to propagate efficiently and rapidly both strains in a bank vole colony. Complete coding sequences of S and M segments were determined by Sanger sequencing of RNA extracted from positive bank voles (naturally and experimentally infected) and from supernatant of Vero E6. For the M segment, nucleotidic sequences were 100% identical for both strains. For the S segment, the amino acid sequences from each strain revealed one mismatch between sequences obtained from tissue and from supernatant, revealing a “bank vole” and a “cell” signature. High throughput sequencing confirmed Sanger results, and provided a better assessment of the impact of isolation methods on intra-host viral diversity.

positive bank voles (naturally and experimentally infected) and from supernatant of Vero E6.
For the M segment, nucleotidic sequences were 100% identical for both strains. For the S segment, the amino acid sequences from each strain revealed one mismatch between sequences obtained from tissue and from supernatant, revealing a "bank vole" and a "cell" signature. High throughput sequencing confirmed Sanger results, and provided a better assessment of the impact of isolation methods on intra-host viral diversity.

Text
Orthohantaviruses represent an increasing threat to humans due to their worldwilde distribution, the increase of infection number and the emergence or re-emergence of new viruses (1).
Puumala virus (PUUV) is the main orthohantavirus circulating in France. This tri-segmented enveloped RNA virus, hosted by the bank vole (Myodes glaerolus), is responsible for a mild form of hemorrhagic fever with renal syndrome (HFRS) named nephropathia epidemica (NE) (2). About 100 cases of NE are reported annually in the Northeast part of France (3).
In France, sequences of PUUV have been studied so far from bank voles samples (4) and recently from human samples (5) but isolated strains have never been cultured or maintained until now. Considering the adaptation capacity of PUUV to cell culture and changes in infectivity induced by cell culture passages (6), having a "wild-type" virus is a crucial step in the thorough study of the biology of this virus. Unfortunately, because of its slow growth in cell culture, the isolation of PUUV is often very difficult.
In this study, we tried to determine a method for selecting bank voles containing live virus in order to isolate PUUV strains. Seto et al (7) were able to isolate PUUV from bank voles that harbored orthohantavirus nucleocapsid protein (NP) in their lung and had no antibody against PUUV. This combination probably corresponds to the active viremic phase. We have adapted this approach by selecting, in our collections of bank vole samples captured in different French forests during the 10 last years (4) (8), seronegative animals with high amounts of viral RNA in lungs. Serum samples were screened using IgG ELISA as already described (8)  More specifically, we carried out these methods on bank voles trapped in two distinct French From these two animals, we tried to isolate live viruses from lung homogenates (5% weight/volume). We combined many conditions such as centrifugation (with or without), filtration (0.22µM) and dilution (10 -1 to 10 -4 ) for homogenate preparation, cell confluence and numbers of cell passages before infection (16 or 43) and different infection methods between the three passages before isolation (supernatant, transposition or co-culture). The conditions that led to the in vitro isolation of the two viral strains consisted in the use of the lung homogenate centrifuged (300 rpm 10 minutes), not filtrated, diluted at 10 -1 and applied on lowpassaged Vero E6 cells (16 passages) in DMEM Dulbecco′s Modified Eagle′s -Medium) 5% SVF (Gibco) and cultured at hyper-confluence (1x10 6 cells for one cupule of P6). It was necessary to proceed to three successive passages to reach a sufficient quantity of live viruses. These passages were realized using "co-culture": a quarter of the initially infected cells were transferred in a culture of non-infected cells (2x10 5 cell / cupule P6) accompanied by 500 µL of the supernatant from the infected culture added to 2 mL of fresh media.
The viral load measured in supernatant of veroE6 cells increased notably: -15.2 CP and -12.8 CP for Ardennes -Hargnies and Loiret -Vouzon respectively ( Figure 1A). Moreover, at the end of cell isolation, viral titer was 1.6x10 4 PFU/mL and 1.5x10 4 PFU/mL for Ardennes -Hargnies and Loiret -Vouzon respectively. It should be noted that their focus size when titrated were different from PUUV control the Sotkamo strain serially passaged in VeroE6 cell ( Figure   1B). The focus size of cultivated viruses was smaller (in particular for Loiret -Vouzon) than Sotkamo focuses. Phenotypical variations have already been described for PUUV titration assay (9). Altogether, these results showed that we were able to select viral isolates from natural host tissue and to cultivate them in cell culture.
Furthermore, we carried out an in vivo isolation process that consists of biological assays on bank voles that were maintained at ISS institute. We used the same lung homogenates from trapped animals already used for in vitro isolation process. We performed a subcutaneous injection of lung homogenate (5% weight/volume) on two animals for each strain. None of the inoculated animals showed clinical evidence of viral infection. Seven days after infection, we collected blood samples through retro-orbital sinus puncture and carried out PUUV serological detection and viral RNA analyses. We did not detect any anti-PUUV antibody ( Table 1) but we found viral RNA in all sera analyzed. Viral loads were similar whatever the PUUV strain considered (Ardennes -Hargnies versus Loiret -Vouzon). This indicated that our protocol lead to an effective transmission of PUUV in these animals. Two days later, at nine days post infection, we euthanized the four bank voles by cervical dislocation. We found that all animals had seroconverted with a higher antibody titer observed for Loiret -Vouzon strain (1/800) than for Ardennes-Hargnies strain (animal 1: 1/200 and animal 2: 1/100). The levels of RNA viral loads in sera had already decreased in all four animals analyzed but they were still positive. We also analyzed RNA viral loads in lung and liver, i.e. organs where PUUV antigen was previously detected (10, 11). We found high levels of viral RNA loads ( Table 2), what confirmed the effective infection of all animals by PUUV.
Molecular analyses were carried out as already described (12)  were analysed with the Data pre-processing tool of FROGS (Genotoul) (17) and the chimeric variants were remove using isBimeraDenovo function of Dada2 R package (18). Each sample was analysed independently (qRT-PCRs, PCR amplifications and sequencing) using at least three PCR replicates to distinguish the true genetic variants and the artefactual mutations due to polymerase or sequencing errors. Mutations below a threshold of 0.32% were removed (see supporting information S3). Validated variants were aligned and analysed using SeaView 5.0 (13). Two measures were used to compare the viral diversity between the two isolation protocols and the two strains: the number of polymorphic sites (19,20) and the mean of percent complexity (the number of unique sequence reads/total reads × 100) calculated on the 10 amplicons (21). Kruskall-Wallis test followed by Dunn multiple comparison test were conducted to compare the mean of percent complexity between strains origin (in natura, in vitro or in vivo).
In natura, the viral diversity between Ardennes -Hargnies and Loiret -Vouzon strains were similar. We found no difference in the number of polymorphic sites or in the mean of percent complexity between areas. After the in vitro isolation process, one variant of Ardennes -Hargnies strain (R63), mostly found in natura, and was still present at high frequency (about 23%) instead of a switch in major variant as Q63. The viral diversity decreased after the cell culture passages, whatever the diversity index considered. For Loiret -Vouzon strain, as observed with Sanger method we can see a switch of major variant to from A28 to S28. But HTS results were contradictory: the number of polymorphic sites increase but the mean of percent complexity stay stable. The in vitro isolation process lead to a significant difference of viral diversity between PUUV strains but not in the same way for both strains ( Figure 3A).
After the in vivo isolation process, we detected a lower number of polymorphic sites compared to natural strains. However we found no significant difference of mean of percent complexity when comparing PUUV strains or in natura and in vitro conditions ( Figure 3B). The sequence of the majority variant did not differ between these two conditions, whatever the isolation protocol considered. Results were similar between bank voles infected with the same strain.

Conclusion
In this study, we were able to cultivate and maintain in cell culture, for the first time, two PUUV isolates from two distinct French areas. Molecular analyses of the S and M segments of PUUV originating from natural isolates, bank voles experimentally infected and cell cultures revealed only one amino-acid mismatch for Loiret -Vouzon and Ardennes -Hargnies strains. Both mismatch were identified in cell culture. High throughpout sequencing confirmed this result, enabling to define a "bank vole" and a "cell culture" signature. Having these two modes of PUUV wild strain culture is an important asset. Cell adapted strains allow to dispose of wellcharacterized viruses that can be used for antiviral candidate studies or cell experimentations assessing the role of apoptosis, PUUV propagation or control (22). Besides, wild strains maintained in their natural host will better contribute to improve our knowledge of PUUV ecology and evolution (23). In in vivo experiment, bank vole used came from "neutral" origin: not from Ardennes or Loiret. For the moment, bank vole-PUUV interactions were poorly documented but could be a key point in viral diversity.
This study also provides new arguments enabling to understand the regional epidemiological differences concerning the circulation of these PUUV strains in humans and in bank voles (4).
It has previously been shown that there is a significant inter-regional viral genetic diversity (4,5,24), that could explain, at least partly, the regional differences in NE epidemiological status in France. Here, we showed that viral diversity of PUUV circulating in the NE endemic area     Serological results of bank vole experimentally infected with lung homogenates from animals trapped from Hargnies and Vouzon.
OD : Optical Density -ND : Not Defined