The network structure and eco-evolutionary dynamics of CRISPR-induced immune diversification

Abstract

As a heritable sequence-specific adaptive immune system, CRISPR-Cas is a powerful force shaping strain diversity in host-virus systems. While the diversity of CRISPR alleles has been explored, the associated structure and dynamics of host-virus interactions has not. We develop theory on the role of CRISPR immunity in mediating the interplay between host-virus interaction structure and eco-evolutionary dynamics in a computational model and three natural systems. We show that the structures of networks describing who infects whom and who is protected from whom are modular and weighted-nested, respectively. The dynamic interplay between these networks influences transitions between dynamical regimes of virus diversification and host control,leading to eventual virus extinction. The three empirical systems exhibit weighted nestedness, a pattern our theory shows is indicative of viral control by hosts. Previously missing from studies of microbial host-pathogen systems, the protection network plays a key role in the coevolutionary dynamics. IntroductionThe structure of complex ecological communities is clearly non-random. It is generally argued that interaction structure affects the stability of communities to perturbations [1-3] and arises from coevolutionary dynamics between interacting partners [4][5][6][7], yet the structure-dynamics nexus remains a long-standing central question in community and network ecology. Host-parasite infection networks, in which interaction structure represents 'who infects whom', have been typically described as modular, nested, or both [8,9]. Modularity concerns patterns of specificity, in which the network is partitioned into modules of hosts and parasites that interact densely with each other but sparsely with those outside the group. Nestedness concerns instead patterns of specialization, and describes a network structure in which specialist hosts are infected by subsets of parasites that in turn also infect the more generalist hosts [8,9].Common to all host-parasite network studies is a dominant focus on patterns of infection. Infection structure should critically depend however on the genetic basis of resistance of hosts to pathogens [10]. The emergence of network structure from immune selection has been shown in pathogens of humans such as Plasmodium falciparum [11, 12]. In particular, strain theory developed for pathogens with multilocus encoding of antigens posits that negative frequency dependent selection, mediated by competition for hosts through cross-immunity, can structure pathogen populations into clusters of strains with limited overlap of antigenic repertoires [11][12][13][14][15]. Parasites with 2 . CC-BY-NC-ND 4.0 International license is made available under a resistance by surface mutation can arise quickly to take over CRISPR or restriction-modification systems in providing immunity. This experimental and theoretical evidence for advantages of surface modification may entail strong fitness costs under natural conditions because mutations in surface receptors such a...

Competition and coevolution drive the evolution and the diversification of CRISPR immunity

“…We stress that our short-term experiment focuses on a very specific scenarios where (i) the initial diversity in the host population was manipulated artificially with equal frequency among different strains and no multiresistance to the phage and (ii) the initial diversity of the phage population was also manipulated experimentally (treatment B versus C). Yet, the distributions of CRISPR immunity and phage diversity are expected to buildup naturally after a phage epidemic and the network structure of strain diversity may be very different from the one used in our experiment [40]. Our work should be viewed as a first attempt to monitor coevolutionary dynamics experimentally and the relevance of the "royal family model" remains to be investigated in a more natural setting.…”

Killer Archaea: Virus-Mediated Antagonism to CRISPR-Immune Populations Results in Emergent Virus-Host Mutualism

“…We have shown that highly active and diverse CRISPR-Cas immunity can create conditions of low-susceptibility host density in which lytic viruses can go extinct (7). These conditions are defined as distributed immunity, where CRISPR-immune hosts have a diversity of spacers targeting a virus, thus leading to a decreased probability of a single escape mutation by the virus leading to the emergence of a viral epidemic of the population (8,9). Under these conditions, nonproductive infection of CRISPRimmune hosts leads to degradation of viral genomes and therefore comes at a significant cost to the virus's fitness (7,(10)(11)(12)(13).…”

“…This results in very few susceptible hosts and a probability of evolutionary emergence of a successful viral escape variant on the scale of 10 Ϫ4 (see Fig. S1 in the supplemental material) (8,9,14,15). Here, we test the virus-host interaction and competitive host fitness when the virus competes against cells that are CRISPR immune and susceptible.…”

Identifying flow modules in ecological networks using Infomap

“…Modularity for example can result from, and therefore be a signature of, ecological and/or coevolutionary dynamics (Beckett and Williams, 2013). For example, in host-pathogen systems, a modular structure has been shown to be a signature of negative-frequency dependent selection that leads to the formation of niches (hosts as resources for pathogens), with consequences for epidemiology and evolutionary dynamics of both the hosts and parasites (He et al, 2018;Pilosof et al, 2019Pilosof et al, , 2020. In mutualistic systems, it can indicate coevolutionary associations between plants and their mutualistic patterns resulting from trait matching (Olesen et al, 2007;Dormann et al, 2016).…”

“…Infomap minimises an objective function known as the 'map equation', which measures how much a modular partition of a network can compress a description of flows on the network (Rosvall and Bergstrom, 2008). Infomap has been thoroughly described mathematically and computationally (Rosvall and Bergstrom, 2008;Rosvall et al, 2010;Rosvall and Bergstrom, 2011;Rosvall et al, 2014) and is widely used in non-ecological disciplines, but only in a handful of papers in ecology (Pilosof et al, 2019;Bernardo-Madrid et al, 2019;Pilosof et al, 2020). Instead of comparing or benchmarking Infomap against other methods, which has already been done (Lancichinetti and Fortunato, 2009;Aldecoa and Marín, 2013), we explain how Infomap works on ecological networks and provide several hands-on examples using empirical data.…”