Mass Die-Off of Saiga Antelopes, Kazakhstan, 2015

In 2015, a mass die-off of ≈200,000 saiga antelopes in central Kazakhstan was caused by hemorrhagic septicemia attributable to the bacterium Pasteurella multocida serotype B. Previous analyses have indicated that environmental triggers associated with weather conditions, specifically air moisture and temperature in the region of the saiga antelope calving during the 10-day period running up to the event, were critical to the proliferation of latent bacteria and were comparable to conditions accompanying historically similar die-offs in the same areas. We investigated whether additional viral or bacterial pathogens could be detected in samples from affected animals using 3 different high-throughput sequencing approaches. We did not identify pathogens associated with commensal bacterial opportunisms in blood, kidney, or lung samples and thus concluded that P. multocida serotype B was the primary cause of the disease.

sequencing libraries using a Kapa Hyper prep kit (KapaBiosciences). Constructed libraries were purified and quantified using a Kapa Illumina library quantification kit (Kapa Biosciences).
Libraries were diluted to 15pM and loaded onto a MiSeq 2x300 version 3 reagent cartridge and run on an Illumina MiSeq with a 5% PhiX spike-in control. Files were run as fastq only and transferred onto a high-performance computing cluster for further analysis.
Two separate analysis protocols were employed to interrogate the data; a k-mer based approach and a de-novo approach (7,8). Before both protocols adaptors were removed, and reads quality trimmed (using a q-score threshold of 30), with trim-galore (version 0.4.3) (9). For the kmer based protocol, any reads less than 100 bases in length were removed. All remaining reads were then analyzed and classified taxonomically using the Kraken mapping software (version 0-15-β) using a local database for identification of all viruses, bacteria and archaeal genomes extracted from the NCBI refseq database (downloaded 15/04/2017). Results were visualized using Kronatools (10). In the case of the de-novo protocol, the sequencing reads from all samples were initially pooled together and then assembled using SPAdes (11). Only contigs with length 200 nt or more were kept, to filter out possible false positives due to short sequences. The resulting contigs were scanned against several BLAST (12) pre-generated databases downloaded from NCBI (nt, tsa_nt, ref_prok_rep_genomes, ref_viruses_rep_genomes, and vectordownloaded 15/04/2017). The results were filtered keeping only the best hits and discarding hits having <80% similarity with the query, and length <50% than the length of the query. Hits were subsequently accumulated by species. As most BLAST databases containing genomes of organisms assembled with Illumina technologies also contain the phiX genome due to it being present as positive sequencing control, we manually fished out the phiX contigs before running BLAST. The contigs for which no hit from any of the 5 BLAST databases was recorded were classified as unknown. The unknown contigs were subsequently scanned against the same BLAST databases using TBLASTX (13), to find similarities at protein level. Only hits with evalue <10 5 , and only the best hit per contig, were kept, to reduce the number of spurious matches. Separately, the unknown contigs were also translated using all possible open reading frames and subsequently processed with SUPERFAMILY version 1.75 (13) to identify potential homologies with known proteins. The raw paired fastq reads generated by this protocol and subsequent assemblies (identifiable and unknown) were submitted to the European nucleotide archive (ENA) archive under accession no. (to be confirmed). of 11 A local contamination by an unrelated laboratory adapted avian coronavirus (Infectious Bronchitis Virus IBV M41-CK) was detected by both methods. To exclude the possibility that IBV was present in the original RNA samples, we tested remaining stocks of original RNA using an IBV-specific 5UTR RT-qPCR (14), which was negative in 6/6 samples tested. IBV reads/contigs were then excluded from subsequent analysis and conclusions, as they were not considered to have any impact on the investigation.

RNA Sequencing Protocol
This protocol was applied by FLI, and RNA was extracted from blood spots on FTA cards. To this end, before extraction, from each FTA card representing one individual, five 5mm punches were ground in a 2ml tube with a 5mm stainless steel bead in 1ml Trizol (Invitrogen) using a TissueLyser (Qiagen) set at 20 Hz for 3 min. Thereafter, the tubes were spun in a standard table-top centrifuge at 13,000 rpm and the supernatant transferred to a fresh tube.
Subsequently, the published protocol for the extraction of RNA was applied (15). In brief, the aqueous phase was mixed with ethanol (40% v/v) and this mixture transferred to a Qiagen RNeasy spin column and all further steps, including the optional on-column DNase treatment, carried out according to the manufacturer's instructions. The extracted RNA was quantified using a Nanodrop ND1000 instrument (Peqlab, Erlangen, Germany), and 500 ng were used for cDNA synthesis and library preparation as described (16). Briefly, after the addition of random hexamer primers, the RNA was denatured at 95°C for 2 min, immediately followed by snapfreezing. This RNA-primer mix was used as input for reverse transcription and second strand synthesis with the cDNA synthesis system kit (Roche, Mannheim, Germany). The obtained double-stranded cDNA was fragmented to a peak size of approx. 500 bp using the M220  (17). In addition, the datasets were mapped along the available P. multocida genome sequence (NC_002663.1) using the Roche/454 software suite (v3.0; Roche) and the generated contigs analyzed using BLASTX (v2.2.26+) (12). The raw reads generated by this protocol were submitted to the ENA archive under accession no. PRJEB28164.

16S Metagenomic Sequencing Protocol
This protocol was applied by IMV, and microbial DNA was extracted from lung and kidney tissues separately using Trizol Reagent (Thermo Fisher Scientific, USA) according to manufacturer's recommendations.