The Causes and Consequences of Changes in Virulence following Pathogen Host Shifts

Emerging infectious diseases are often the result of a host shift, where the pathogen originates from a different host species. Virulence—the harm a pathogen does to its host—can be extremely high following a host shift (for example Ebola, HIV, and SARs), while other host shifts may go undetected as they cause few symptoms in the new host. Here we examine how virulence varies across host species by carrying out a large cross infection experiment using 48 species of Drosophilidae and an RNA virus. Host shifts resulted in dramatic variation in virulence, with benign infections in some species and rapid death in others. The change in virulence was highly predictable from the host phylogeny, with hosts clustering together in distinct clades displaying high or low virulence. High levels of virulence are associated with high viral loads, and this may determine the transmission rate of the virus.

infection. These flies were then left for 3 days at 22°C and the infected fly was removed. Any vials where the infected fly had died were excluded. The sentinel flies were tipped onto a fresh vial of food and left for 3 further days at 25°C to allow any acquired virus to replicate, before being frozen and undergoing RNA extractions for qRT--PCR, as described above. Out of the 38 species; 32 had 3 biological replicates, 5 had 2 biological replicates and 1 had 1 biological replicate. For each sample four technical replicates of qRT--PCR were carried out with the DCV primers, but we found a number of samples contained low amounts of viral RNA and were on the limit of detection for the qRT--PCR assay. We therefore took a conservative approach by only classifying the sample as infected if all 4 replicates detected viral RNA. We used these data to define a quantitative and categorical measure of infection. The quantitative measure was simply the average viral load relative to the endogenous control across the replicates with samples where all 4 technical replicates did not detect virus being censored.

S2
The original vial of medium (in which the infected and sentinel flies and been co-housed) was incubated at 25°C, until a second generation of adult sentinels eclosed. The second generation of sentinel flies were frozen and RNA extractions were carried out for qRT--PCR, as described above. From pilot data, we knew transmission via this route of transmission was rare, so we limited ourselves to assaying 8 species (from four major host clades) with high viral loads (these species sit in the top 33% highest viral loads).

Dead adult--larvae transmission
We also measured transmission from the corpses of flies that had died following DCV infection to embryos/larvae. As the chorion of eggs means that embryos are unlikely to be infected we will refer to this as adult--larval transmission, although we cannot exclude infections of the developing embryo. 4--7 day old males from 15 species were infected as described and placed at 22°C. Flies that died within 24 hours of inoculation were presumed to have died from the inoculation procedure and so were not used. Flies were tipped onto fresh medium every 3 days and dead flies were collected daily. Individual dead flies were placed in a fresh vial of medium and 3μl of eggs collected from a population cage of the isogenic D. melanogaster DrosDel w 1118 line were placed into the same vial. Vials were then placed at 25°C for the development of the eggs of the isogenic sentinel flies. After eclosion, 1--2 day old sentinel flies were frozen and their RNA extracted for qRT--PCR, as described above. Out of the 15 species, 14 had 3 biological replicates and 1 had 2 biological replicates.
In order to assess how viral load on day 2 post infection (from the previous experiment) correlated with viral load on the day a fly died, 3--5 day old flies from 15 species were infected as described above, and dead flies were collected twice daily and frozen at --80°C. Flies that died within 24 hours of inoculation were presumed to have died from the inoculation procedure and so were not used. We carried out 3 biological replicates per species, pooling the dead flies within each replicate. Each replicate consisted of 6 dead flies on average (range S3 of means per replicate=2--10). To calculate the change in viral load from day 0 to death, we measured the change in viral load for the three biological replicates from a single day 0 sample, for each species.

Statistical analyses of transmission data
The statistical analysis of the transmission data follows the same form as those presented for the main analysis above but using a bivariate formulation, where the two mortality response variables were replaced with a single response variable. In the first set of models the response variable was the categorical measure of infection from sentinel flies and was treated as binary. In the second set of models the response variable was the quantitative measure of infection from sentinel flies and was treated as censored Gaussian. All analyses gave similar results so we present the categorical data only. In the model comparing day 2 and day of death viral loads the response was the viral load and was treated as Gaussian. Due to the number of species sampled in these experiments being small (between 15 and 38 species) there was little power for decomposing inter--specific variation into phylogenetic and non--phylogenetic components.
Consequently, we fitted two models to each set of data; a phylogenetic model and a non--phylogenetic model. As we detected no adult+parent--offspring transmission, models were not fitted for these data. Data is included as datasets S2 and S3.

Transmission
To understand how virulence might be linked with transmission after a host shift, we compared viral loads in different species with transmission rates. The  Combine and bring to a boil for 5mins, cool to 70°C before adding Nipagin