Non‐invasive genomics of respiratory pathogens infecting wild great apes using hybridisation capture

Abstract Human respiratory pathogens have repeatedly caused lethal outbreaks in wild great apes across Africa, leading to population declines. Nonetheless, our knowledge of potential genomic changes associated with pathogen introduction and spread at the human‐great ape interface remains sparse. Here, we made use of target enrichment coupled with next generation sequencing to non‐invasively investigate five outbreaks of human‐introduced respiratory disease in wild chimpanzees living in Taï National Park, Ivory Coast. By retrieving 34 complete viral genomes and three distinct constellations of pneumococcal virulence factors, we provide genomic insights into these spillover events and describe a framework for non‐invasive genomic surveillance in wildlife.


| MAIN
Spillover of common human respiratory pathogens to wild great apes habituated to human presence for research or tourism has been repeatedly documented. [1][2][3] The development of long-term health monitoring programmes within conservation initiatives, 4 bringing together behavioural observations and non-invasive sampling (e.g. faeces, urine and performing necropsies on dead wildlife), has led to the establishment of a framework for studying disease epidemiology in these endangered populations. Viruses of different families have been identified as primary causative agents, with members of the family Pneumoviridae being frequently reported. Bacterial coinfections, most often caused by Streptococcus pneumoniae, 5,6 have contributed to disease severity, ultimately leading to mortality.
Emergence of a pathogen in a new host population raises many questions, such as whether the pathogen will spread efficiently, whether mutations will arise and be fixed whilst spreading and whether it will remain endemic in the population or not. With the exception of a few recent reports, 2,3,7 earlier molecular characterisations of the human pathogens causing disease in wild great apes have been limited to a few, partial genes. These investigations were conducted using PCR-based screening approaches aimed at detecting and genotyping common respiratory agents infecting humans. 6,8 To better understand pathogen introduction, more comprehensive genomic analyses, ideally performed on samples collected at different stages of an outbreak and from different individuals, are required. Unbiased metagenomic/metatranscriptomic analyses may allow to assemble complete genomes, 2,3 but these are likely to require relatively deep sequencing when applied to samples with a low ratio  Table S1). On-target reads represented from 0.1% to 99.08% of all reads and six complete viral genomes (three HMPV, two HRSVB and one HRSVA) could be reconstructed. To investigate whether we could broaden our dataset, we applied the method to 32 pneumovirus-positive faecal samples collected during the different outbreaks and tested in previous studies. 6,8 Our sample selection aimed at maximising the number of individuals and the time frame tested. On-target reads represented from 69.3% to 99.74% of all reads, and 21 complete viral genomes were reconstructed (28 at 2X coverage), highlighting a good performance also on non-invasive samples (Table 1 and supporting information Table S1). As opposed to  Whole-genome maximum-likelihood phylogenetic analyses confirmed previous genotyping, with the 2004 HMPV strain falling within the diversity of the B2 lineage (supporting information Figure S1), the HRSVB from 2005 and 2006 within the GB3 genotype (supporting information Figure S2) and the HRSVA from 2009 within the GA2 (supporting information Figure S3). We acknowledge that given the paucity of genomic information available for viral strains circulating in humans in these remote areas, we could not fit these data to seasonal local patterns.
In four out of six outbreaks, several deaths occurred among the chimpanzees and were attributed to co-infections with S. pneumoniae variability of these proteins are known to vary substantially across serotypes identified in humans. 13 To which extent the same occurs in pneumococcal strains circulating in wildlife remains largely unknown.
To investigate the diversity of virulence factors identified in the pneumococci infecting the Taï chimpanzee population and to simultaneously design a tool that would allow for the differentiation of S. pneumoniae from other commensal streptococci (e.g. Streptococcus. mitis and Streptococcus oralis), we designed a bait set targeting nine (entire or partial) virulence genes thus far only reported in pneumococci (supporting information Table S4). We generated DNA baits by using sheared long-range PCR products to which biotinylated adapters were subsequently attached.
Following hybrid capture on libraries generated from lung samples, 1.01% to 50.19% of the total reads were on target (supporting information PspA, HysA and PcpA) of the nine virulence genes tested were detected (supporting information Figure S4). When compared, consensus sequences for these genes were identical, suggesting the same strain was involved in both outbreaks. The same five virulence genes,