Virus discovery in small ruminants with non-suppurative encephalitis

Background

Of all emerging viral infectious diseases, nearly half are characterized by encephalitis or severe neurological disease and the majority of them are zoonotic, i.e., they pose a risk to animals as well as human health [1, 2]. In small ruminants, which in Switzerland are sheep and goats, there is a knowledge gap on causes of encephalitis associated neurological diseases. In a comprehensive neuropathological survey of the Swiss small ruminant population, 93.5% of non-suppurative encephalitis cases, suggestive of viral infection, remained etiologically unresolved [3]. A very similar situation is evident in other livestock species, such as cattle, but also in humans in which many encephalitis cases remain etiologically unresolved as well; in a study with over 5’000 patients with encephalitis, a cause was identified only in approximately one third of the cases, despite intensive conventional molecular testing for a panel of pathogens [4]. Thus, the identification of novel or unexpected pathogens involved in encephalitis, especially viruses, is essential.

In recent years, novel techniques for virus discovery using unbiased next-generation sequencing and bioinformatics, so called viral metagenomics, became available and allowed the identification of novel emerging viruses, such as Schmallenberg virus and Variegated Squirrel 1 Bornavirus [5, 6]. In our division, previous work discovered novel encephalitis associated astroviruses in cattle [7, 8]. This situation encouraged me to plan a PhD project on virus discovery in neurologically diseased small ruminants by viral metagenomics. We conducted a small-scale pilot study with brain tissues of 3 sheep with severe histopathological brain lesions and a non-suppurative inflammatory pattern and subjected RNA and DNA extracts of these samples to NGS and a bioinformatics pipeline similar to the one we established for cattle [9]. Strikingly, in one sheep yet another novel astrovirus (Ovine Astrovirus-CH16, OvAstV-CH16) was detected, which was almost identical to one of the bovine astrovirus encephalitis strains (Bovine Astrovirus-CH15, BoAstV-CH15). However, a major limitation of this approach was, that the sample pretreatment and the bioinformatics pipeline were not optimal for the purpose of virus discovery in small ruminant brain tissues, which results in very high sequencing costs as well as reduced sensitivity. Still, our preliminary data clearly show that a metagenomics approach in small ruminants is feasible, but needs optimization, and will very likely generate additional virus candidates that will require being further investigated for their association with the disease.

Hypothesis & Aims

Hypothesis: Non-suppurative encephalitis in small ruminants is caused by unknown viruses or known viruses that have not been associated with encephalitis previously.

Aims:

The main goal of this project is to elucidate the etiology of non-suppurative encephalitis cases in sheep and goats. In order to achieve this, I will:

1establish a NGS-based metagenomics pipeline for the discovery of novel viruses using brain material of sheep and goats

2analyse brain tissues of sheep and goats with non-suppurative encephalitis (cases) and without brain lesions (controls) by viral metagenomics

3assess the possible causative relationship between virus infection and disease by in-situ virus detection and in-vitro virus propagation

Methods & Research Plan

Tissue pretreatment

The sensitivity of viral sequence detection by NGS relies on efficient removal of undesired host derived nucleic acids (RNA and DNA). Different pretreatment methods have been used in metagenomics studies in the past. These methods may be grouped in the following categories: low-speed centrifugation, filtration, ultracentrifugation and nuclease treatment [10]. Protocols will be modelled via literature research and evaluated with respect to the detection of viruses and the alteration of the ratio of viral to non-viral nucleic acids in favour of the viral nucleic acids [11]. To this end, I will create a mock virome in order to account for different virus properties (DNA vs RNA, enveloped vs non-enveloped, virion size, genome size). Viruses are available either from positive brains from our lab (Borna disease virus and Bovine Astrovirus-CH15, both in sheep brains and Bovine Herpesvirus-6 in a bovine brain) or ordered from the American Type Culture Collection (Bovine Rotavirus-A, Bovine Parvovirus-1, Ovine Adenovirus-5, derived from bovine or sheep). The viruses will be spiked into a brain homogenate of a healthy sheep and a goat and subjected to different combinations of tissue pretreatment procedures. The abundance of the host genome and transcriptome as well as the relative amount of virus genomes, i.e. the signal-to-noise ratio will be determined by 12S ribosomal DNA and RNA (RT)-qPCR and virus specific (RT)-qPCR protocols, respectively. In addition, different methods for NGS library preparation will be evaluated and compared for their efficiency to detect the different types of viruses: (i) RNA (cDNA) libraries alone, (ii) RNA (cDNA) and DNA libraries separately and (iii) combined RNA (cDNA) and DNA libraries.

Bioinformatics pipeline

The efficiency of selected pretreatment protocols will be further assessed by NGS. An in-silico depletion of host derivce genomic sequences will be conducted using the sheep and goat reference genomes, respectively. In addition, genome sequences of healthy sheep and goats that are available from public sources or from the Institute of Genetics at the Vetsuisse Faculty can be used to deplete host sequences by using e.g. a k-mer approach [12]. Ideally, the genome sequences of the host should be derived from an animal of the same breed. The bioinformatics pipeline will comprise two types of analyses. The first is mapping the non-host reads to virus database entries to identify sequences of known viruses. The second consists of the assembly of non-host reads (to contigs), in-silico translation of the contigs and the comparison of the deduced protein sequences to virus protein databases. A cut-off value for contig length, protein similarity, read depth and length and the elimination of non-viral nucleic acids (e.g. host-, bacterial and parasitical sequences) will be evaluated. Finally the best suited combination of sample pretreatment and bioinformatics analysis will be selected for further studies.

Virus identification and discovery

Whole brains of 56 small ruminants (47 sheep, 9 goats) with non-suppurative encephalitis and the same number of sheep and goats without brain lesions, diagnosed by three neuropathologists via histopathology, will be used in this project. The samples are representative of the sheep and goat population in Switzerland at the period of collection and were tested negative for Borna disease virus and Rabies virus, via immunohistochemistry [3]. Brain samples of the cases and the controls will be analysed by the metagenomics pipeline established above to identify virus candidates. Sequences of virus candidates will likely not cover the full virus sequence and gaps will be filled by another round of NGS or by conventional (RT)-PCR, rapid amplification of cDNA ends (RACE) and Sanger Sequencing.

Assessment of disease association

In order to assess a possible causative relation of virus candidates with the disease, I will develop tools to demonstrate the virus in-situ, either by immunohistochemistry or in-situ hybridization. In-situ hybridization uses labelled nucleic acids that bind to specific DNA or RNA sequences in tissue, whereas immunohistochemistry uses labelled antigens to detect proteins in tissue. The localization of the histopathological lesions will be compared to that of the candidate virus. The statistical association of virus detection and disease will be assessed in case control studies. Depending on the virus candidates, I will attempt to isolate the virus in cell culture, using different cell lines as well as ruminant organotypic brain-slice cultures as established by our lab [13].

Possible Pitfalls

If no additional virus candidates are found, I will focus my project on OvAstV-CH16. The sequence of this virus is very similar to that of BoAstV-CH15 isolated from a cow (accession KT956903.1). However, the bovine isolate shows a 9 nucleotide deletion in one of the non-structural proteins and lacks 160 bases at the 5’ end of the genome. In a peers’ project in our division an infectious cDNA clone of BoAstV-CH15 is currently being established. In this case, I will establish a similar clone for the ovine strain and address the question whether the observed differences between these strains are determinants of host specificity.

Significance

Encephalitis patients are difficult to treat as etiology has proven to be difficult to determine. More information about involved pathogens is needed in order to develop better suited diagnostic tools. A broad knowledge of the spectrum of viral pathogens involved in encephalitis will lead to better treatment options for practitioners in the field and thus results in an improvement of animal welfare. Given the close contact between small ruminants and humans and the fact that the majority of emerging infectious diseases is zoonotic, a fast response to emerging diseases is important. Therefore, a quick virus identification pipeline is crucial.