Deep Sequencing of MHC-Adapted Viral Lines Reveals Complex Recombinational Exchanges With Endogenous Retroviruses Leading to High-Frequency Variants

Abstract

Experimental evolution (serial passage) of Friend virus complex (FVC) in mice demonstrates phenotypic adaptation to specific host major histocompatibility complex (MHC) genotypes. These evolved viral lines show increased fitness and virulence in their host-genotype-of-passage, but display fitness and virulence tradeoffs when infecting unfamiliar host MHC genotypes. Here, we deep sequence these viral lines in an attempt to discover the genetic basis of FVC adaptation. The principal prediction for genotype-specific adaptation is that unique mutations would rise to high frequency in viral lines adapted to each host MHC genotype. This prediction was not supported by our sequencing data as most observed high-frequency variants were present in each of our independently evolved viral lines. However, using a multi-variate approach to measure divergence between viral populations, we show that populations of replicate evolved viral lines from the same MHC congenic mouse strain were more similar to one another than to lines derived from different MHC congenic mouse strains, suggesting that MHC genotype does predictably act on viral evolution in our model. Sequence analysis also revealed rampant recombination with endogenous murine leukemia virus sequences (EnMuLVs) that are encoded within the BALB/c mouse genome. The highest frequency variants in all six lines contained a 12 bp insertion from a recombinant EnMuLV source, suggesting such recombinants were either being favored by selection or were contained in a recombinational hotspot. Interestingly, they did not reach fixation, as if they are low fitness. The amount of background mutations linked to FVC/EnMuLV variable sites indicated that FVC/EnMuLV recombinants had not reached mutation selection equilibrium and thus, that EnMuLV sequences are likely continuously introgressing into the replicating viral population. These discoveries raise the question: is the expression of EnMuLV sequences in mouse splenocytes that permit recombination with exogenous FVC a pathogen or host adaptation?

The evolution of host specialization in an insect pathogen

“…Our results finding many candidate genes with lower selection coefficients are generally in line with previous evolve and re-sequence experiments that start with standing genetic variation and less-specific environmental stressors, though clonal interference may have been less prominent in our experiment due to the relatively smaller bottleneck sizes in vivo infection processes (Miller et al 2011;Tenaillon et al 2012;Lang et al 2013;Long et al 2015;Schlötterer et al 2015). In the context of virus adaptation to host genotype, Middlebrook et al (2021) also do not see parallel genetic evolution when FVC virus specializes on mice with different MHC genotypes from a clonal starting population, although they did see evidence that virus populations adapted to each MHC genotype are more similar to each other than to those adapted to foreign MHC types.…”

“…In the context of virus adaptation to host genotype, Middlebrook et al. (2021) also do not see parallel genetic evolution when FVC virus specializes on mice with different MHC genotypes from a clonal starting population, although they did see evidence that virus populations adapted to each MHC genotype are more similar to each other than to those adapted to foreign MHC types.…”

“…2020; Middlebrook et al. 2021). We then used constrained ordination analysis on Hellinger and Chi‐square pre‐transformed SNP frequencies in ‘vegan’ and performed a Monte Carlo permutation test to determine whether treatment significantly predicted SNP frequency variance amongst the virus populations (Oksanen et al.…”

“…To determine regions of variation, we plotted variant frequencies against genome position, identified genome regions with high genetic variation, and compared these genetic regions to the annotated PiGV reference genome by hand to identify potentially interesting nearby genes (Harrison et al 2016). To determine whether passage 9 variant community composition was predicted by treatment, we made multidimensional scaling plots in "vegan" using the "metaMDS" function with the 'Canberra' method, which deemphasizes zero values (Oksanen et al 2020;Middlebrook et al 2021). We then used constrained ordination analysis on Hellinger and Chi-square pre-transformed SNP frequencies in 'vegan' and performed a Monte Carlo permutation test to determine whether treatment significantly predicted SNP frequency variance amongst the virus populations (Oksanen et al 2020).…”