Viruses rule over adaptation in conserved human proteins

Abstract

12Adaptive evolution often involves fast-evolving proteins, and the fastest-evolving 13 proteins in primates include antiviral proteins engaged in an arms race with viruses 14 1-3 . Even though fast-evolving antiviral proteins are the most studied cases of 15 primate host adaptation against viruses, viruses predominantly interact with host 16 proteins that are broadly conserved between distant species in order to complete 17 their replication cycle 4 . Broadly conserved proteins are generally viewed as playing 18 a negligible role in adaptive evolution. Here, we used a dataset of ~4,500 human 19 proteins known to physically interact with viruses (VIPs for Virus-Interacting 20Proteins), to test the involvement of broadly conserved proteins in adaptive 21 evolution against viruses. We found that VIPs conserved between animals and fungi 22 have experienced not only high rates of adaption, but also strong adaptive events. 23 Broadly conserved proteins that do not interact with viruses experienced very little 24adaptation. As a result, the arms race with viruses explains more than 75% of 25 adaptation in the most phylogenetically conserved subset of the human proteome. 26Our results imply that broadly conserved proteins have played a significant role in 27 adaptation, and that viruses were likely one of very few selective pressures that 28 were able to force the conserved, central pillars of host cellular functions to adapt 29 during evolution. 30 31 driving force behind these patterns (Extended Data Fig. 2). Again, the results obtained with 129 ABC-MK were broadly concordant with polyDFE ( Figs. 2 and 3). 130 131 Figure 2. Adaptation in VIPs and control non-VIPs with increasing breadth of phylogenetic 132 conservation 133 The red areas highlight the 95% confidence intervals for control non-VIPs, estimated from 1,000 iterations 134 during the bootstrap test (Methods). Blue dots indicate a for VIPs. The legend in the figure provides the scale 135 for how significant the bootstrap test is. A. Results for polyDFE. B. Results for ABC-MK. 136 137 138 139

Ancient RNA virus epidemics through the lens of recent adaptation in human genomes

“…If gene expression, protein–protein interactions [ 36 ] or other factors affect the prevalence of recent sweeps on their own, independently of interactions with viruses, they might confound the comparison of VIPs and non-VIPs by creating differences between the former and the latter that have nothing to do with interactions with viruses. Thus, we build random sets of control non-VIPs that match VIPs for multiple potential confounding factors using a previously described bootstrap test [ 9 ] (see Methods). Furthermore, the control sets of non-VIPs exclude all non-VIPs that are too close to VIPs and may thus be found in the same large sweeps extending over multiple genes.…”

“…VIPs harbour remarkably high levels of past protein adaptation, with rates of adaptive amino acid changes several times higher than human proteins that are not known to interact with viruses (non-VIPs) [ 8 , 10 ]. Furthermore, adaptation was not only more frequent at VIPs compared to non-VIPs; it was also stronger, more intense adaptation [ 9 , 10 ], suggesting that viruses repeatedly imposed strong selective pressures on their human hosts during evolution. Thus, frequent new zoonoses and abundant adaptation at VIPs together suggest that viruses drove many epidemics in past human evolution.…”

Comparison of the Full Distribution of Fitness Effects of New Amino Acid Mutations Across Great Apes

“…Hence, there is no need for very unlikely back mutations to restore the fitness losses incurred by previous mutations (Charlesworth and Eyre-Walker 2007). An interesting implication of such mode of evolution is that rates of adaptive substitutions may not be driven only by external conditions (such as viruses, see Enard et al 2016;Castellano et al 2019;Uricchio et al 2019) but also by the amount of deleterious mutations already present in the genome as this mutation load conditions the current level of adaptation in a population. This mechanism is not often invoked to explain Darwinian adaptation (due to environmental changes), yet a small pool of compensatory mutations will contribute to the amino acid differences between species in the long term (Hartl and Taubes 1996).…”

Ancient RNA virus epidemics through the lens of recent adaptation in human genomes

“…The bootstrap test we used to match VIPs with control non-VIPs that match for multiple 476 confounding factors has already been described extensively in a previous manuscript that the 477 reader can refer to for more ample details (Castellano, 2019). An implementation of the 478 bootstrap test is available at https://github.com/DavidPierreEnard, as part of a larger pipeline 479 that also estimates the whole enrichment curve p-value (see below).…”