Comparative analyses of co-evolving host-parasite associations reveal unique gene expression patterns underlying slavemaker raiding and host defensive phenotypes

Abstract

The transition to parasitism is a drastic shift in lifestyle, involving rapid changes in gene structure, function, and expression. After the establishment of antagonistic relationships, parasites and hosts co-evolve through reciprocal adaptations, often resulting in evolutionary arms-races. Repeated evolution of social parasitism and slavery among Temnothorax ants allows us to examine those gene expression patterns that characterize slavemaker raiding and reciprocal host defensive phenotypes. Previous behavioural studies have established that raiding strategies between Temnothorax slavemakers diverge, while host defense portfolios shift similarly under parasite pressure. We are the first to confirm this at the molecular level, revealing that slavemaking species exhibit a wider variety of genes with species-specific patterns of expression within their raiding phenotypes, whereas expression similarity is commonly found during the non-raiding phenotype. Host species response to slavemaker aggression, however, is indicated by strong changes in the expression of a relatively few number genes. Additionally, the expression of individual genes such as Acyl-CoA-Delta(11) desaturase and Trypsin-7 is strongly associated with the raiding phenotype of all three slavemaking species. Here, we provide novel insight into the gene expression patterns associated with raiding and nest defense behavior in Temnothorax ants, suggesting lineage-specific evolutionary patterns among both slavemakers and hosts.

Figure 1

Maximum Likelihood phylogenetic relationship between the six Temnothorax species herein examined, with Acromyrmex echinatior as outgroup (used with permission). Produced using RaxML based on 5199 orthologous gene clusters (ML boostrap values given at each node). Slavemaker T. americanus preferentially parasitizes T. longispinosus, T. duloticus preferentially parasitizes T. curvispinosus, and T. pilagens preferentially parasitizes T. ambiguus.

Figure 2

Number of genes found to be differentially-expressed within six Temnothorax species, up-regulated during either slavemaker or host raiding or non-raiding behavioral states. Upper row: Slavemakers; Bottom row: Hosts. Species abbreviations are as follows: T. ambi: T. ambiguus, T. curvi: T. curvispinosus, T. longi: T. longispinosus, T. amer: T. americanus, T. dul: T. duloticus, T. pila: T. pilagens.

Figure 3

Dendrograms resulting from WGCNA of orthologous clusters. (a) WGCNA of host-specific clusters shows that samples group primarily according to species. Phenotype does not appear to have a strong influence on host grouping. (b) WGCNA of slavemaker-specific clusters yields patterns of grouping driven first by species and then secondarily by phenotype. Unlike hosts, phenotype does appear to be associated with more similar gene expression patterns within slavemakers.

Figure 4

Average log fold change of differentially-expressed genes between behavioural phenotypes of hosts and slavemakers herein examined. While non-raiding phenotypes only show a trend towards differentiation between slavemaker and host (p < 0.059), the log fold change in expression between slavemaker and host during the raiding/defensive phenotypes (p < 0.028) does differ significantly. Log fold change within lifestyle groups between raiding and non-raiding phenotypes differ significantly for both slavemakers (p < 0.013), as well as hosts (p < 0.028).

Figure 5

Venn diagrams displaying the number of significantly differentially-expressed contigs that are either private to a specific species, or shared by multiple species. Homology determined by cluster analysis. Un-bracketed numbers indicate total number of clusters, bracketed numbers indicate subset of clusters with accompanying functional annotation. For species abbreviations, see Fig. 1.

Figure 6

Module-Trait relationships within all three slavemaking species (a) and all three host species (b). Contigs within Slavemaker Module 1 are strongly associated with the non-raiding phenotype, where contigs in slavemaker Module 7 are strongly associated with the raiding phenotype. Host contigs within Host Module 9 are positively associated with the defensive phenotype. Numbers to the right of module identifiers indicates the number of contigs within that specific module. Slavemaker Module 10 and Host Module 12 are “leftover” modules, containing random contigs that did not fall into any other module.

Figure 7

Venn diagrams displaying the number of KEGG pathways that are either private to a specific species, or shared by multiple species. Pathways generated by obtaining KO (KEGG Pathway) terms for phenotype-specific, significantly-differentially-expressed genes. For species abbreviations, see Fig. 1.