Defensive symbiosis: a microbial perspective

The enemy of my enemy is my friend (Ancient Proverb) Defensive symbioses are indirect interactions that involve at least three species (host, symbiont and enemy) where the net benefits of symbiosis are contingent on the presence of enemies (Fig. 1, Clay, Holah & Rudgers 2005; Lively et al. 2005). The performance of host and non-host populations in the presence and the absence of natural enemies distinguishes the direct benefits of symbiosis (such as nutrient provisioning) from indirect benefits arising from protection from natural enemies. Defensive symbiosis requires that the fitness of hosts is proportionally higher than non-hosts in the presence of natural enemies relative to enemy-free conditions. In the absence of enemies, the defensive symbiont may decline in frequency or be lost from the host population (Janzen 1973; Lively et al. 2005; Palmer et al. 2008). Protective or defensive mutualisms have long been recognized. Thomas Belt (1874), in The Naturalist in Nicaragua, first described the defence of acacia trees from herbivores by ants (see also Boucher, James & Keeler 1982; Janzen 1985; Palmer et al. 2008). We now know that extrafloral nectaries, Beltian bodies and other traits that attract ants for defence against herbivores are widespread in plants (Bentley 1977; Beattie 1985; Rudgers & Strauss 2004; Rico-Grey & Oliveira 2007). Outside of ant–plant protective interactions, which are based on physical aggression, defensive mutualisms involving microbial symbionts often involve the production of toxic secondary metabolites. One well-understood example is grasses infected by endophytic fungi (family Clavicipitaceae) that grow systemically in above-ground plant tissues, are vertically transmitted through seeds and produce a variety of alkaloid compounds deterrent to herbivores (Clay 1988; Clay & Schardl 2002). The endophytes are under strong selection for alkaloid diversification which may improve their protective function (Schardl et al. 2013). Defensive symbioses between insects and bacteria are also widespread. For example, Currie and colleagues (Currie, Mueller & Malloch 1999a; Currie et al. 1999b, 2006) reported that leaf cutter ants (Atta spp.) harbour antibiotic-producing actinobacteria (Pseudocardinia spp.) that inhibit Escovopsis spp. pathogens of their fungal gardens and help maintain the mutualism between ants and their fungal gardens. Many sap-sucking insects like aphids also harbour defensive symbionts which provide protection against a range of natural enemies (Oliver et al. 2003, 2008; Scarborough, Ferrari & Godfray 2005; Oliver & Moran 2009; Tsuchida et al. 2010; Lukasik et al. 2013). Microbial interactions with plants and animals are typically invisible to the naked eye, but their impacts on hosts and host communities can be very large. Higher organisms host diverse microbial communities (Arnold et al. 2000; Costello et al. 2009; Rodriguez et al. 2009; Hawlena et al. 2013), and microbial dependency on the host will favour traits that help protect that resource and ensure their transmission (Lukasik et al. 2013). Microbes have high rates of evolutionary change, and horizontal transfer of adaptive genes or genomes is common (McCutcheon, McDonald & Moran 2009; Oliver et al. 2010; Werren et al. 2010; Smillie et al. 2011; Zhu et al. 2013). New approaches and technological advances are providing novel insights into plant and animal microbiomes, and more demonstrated and hypothesized examples of defensive symbiosis (Turnbaugh et al. 2009; White & Torres 2009; Zhu et al. 2011; Lundberg et al. 2012). Identifying the key microbial players and the underlying mechanisms of protection will improve our understanding of factors affecting the dynamics of ecological communities and provide applications for agriculture and human health (Wicklow et al. 2005; Mazmanian, Round & Kasper 2008; Mao-Jones et al. 2010). While a few systems have been well studied, our knowledge of the diversity, distribution, mechanisms and ecological consequences of defensive symbioses is limited. This is despite increasing scientific interest and technological innovations enabling rapid discovery and novel research directions in symbiotic systems. Most macro-organisms support diverse microbial communities, but we have limited understanding of how microbes interact with each other within hosts and with enemies of the host. It is perhaps not surprising that the best understood defensive symbioses occur in hosts with just one (e.g. grasses) or a few (e.g. aphids) dominant microbial symbionts, which are easier to evaluate. A major challenge is to identify potential defensive contributions of particular microbes in hosts with diverse microbiomes or determine whether protection arises through interactions within microbial consortia. To help begin to address these and other gaps, several research directions are recommended. Are defensive symbioses common and widespread, or do they occur only in a few specific systems with particular organisms and habitats? Based on the rate at which new examples of defensive symbiosis are being described, we may be seeing only the tip of the iceberg. However, in complex communities of macro- and micro-organisms, it is non-trivial challenge to obtain direct evidence of defensive symbiosis. Comparing performance of host and non-host populations with and without natural enemies is the most direct way to identify defensive symbioses. Cases where resistance to enemies can be “cured” by antibiotic or fungicide applications, or where resistance is strictly maternally inherited, also represent strong candidates for defensive symbiosis. More typically, evidence for defensive symbiosis is indirect. Microbial production of secondary metabolites with known toxic functions is often taken as prima facie evidence of defensive symbiosis. For example, Nakabachi et al. (2013) recently described a bacterial symbiont (Candidatus Profftella armature) of the Asian citrus psyllid, Diaphorina citri, where 15% of its highly reduced genome is devoted to two biosynthetic gene clusters that encode a polyketide toxin similar in structure to pederin. In Paederus rove beetles, pederin is synthesized by a Pseudomonas symbiont and accumulates in the body fluid of the beetle, where it serves to deter predators (Kellner & Dettner 1996). Likewise, antibiotic production by insect-associated bacteria also suggests that these symbionts play a protective role (Scott et al. 2008; Um et al. 2013). The increasing availability of molecular and genomic technologies offers new opportunities to identify and characterize defensive symbioses. 16S rRNA tag sequencing using a high-throughput pyrosequencing can efficiently search for and identify particular groups of bacterial symbionts (Hawlena et al. 2013), and metagenomic, whole-genome shot-gun sequencing can be used to identify alkaloid or antibiotic production genes diagnostic for defensive symbiosis. Fluorescent in-situ hybridization can be used to visualize and localize microbial symbionts within hosts (Klyachko et al. 2007), and q-PCR can be used to assess symbiont density in relation to host age, gender, habitat, etc. More generally, a combination of molecular and microscopic approaches can serve to identify and localize microbial symbionts and identify potential genes with defensive functions. The mechanism of defence is not known in many systems, even where defensive symbiosis has been experimentally documented (e.g. Scarborough, Ferrari & Godfray 2005; Jaenike et al. 2010; Xie, Vilchez & Mateos 2010; Busby et al. 2013). Identification of those mechanisms will enhance our understanding of defensive symbiosis and point to potential applications for pest control and improved plant and animal health. Where the mechanism of defence is clear, most defensive symbioses appear to exhibit one of two basic patterns but others may exist. First, the production (or detoxification) of bioactive secondary compounds by microbial symbionts is common to many defensive symbioses. Related chemistry can often be found in free-living relatives, suggesting that pre-existing microbial pathways have been co-opted for host defence. For example, toxin-producing bacterial and fungal endosymbionts infect poisonous plants such as locoweeds (Fabaceae, Braun et al. 2003; Panaccione, Beaulieu & Cook 2014) and some species of grasses (Clay & Schardl 2002), morning glories (Convoluvulaceae, Beaulieu et al. 2013; Cook et al. 2013) and Rubiaceae (Verstraete et al. 2011), as well as marine organisms (Lopanik, Lindquist & Targett 2004; Simmons et al. 2008). Similarly, antibiotics are produced by free-living soil bacteria but are also produced by bacterial symbionts of insects, where they provide resistance to pathogenic micro-organisms (Currie, Mueller & Malloch 1999a; Scott et al. 2008; Um et al. 2013). In the case of pea aphids and their Hamiltonella endosymbiont, host protection against parasitoids imparted by Hamiltonella depends on symbiont being infected with a toxin-producing bacteriophage (Moran et al. 2005; Weldon, Strand & Oliver 2013). Parasites and herbivores can also turn the table and use microbial symbionts to detoxify or resist their target organism's defensive chemistry. For example, conifers respond to beetle attack by producing toxic terpenes, but bacterial symbionts associated with mountain pine beetles (Dendroctonus ponderosae) can reduce terpene concentrations in artificial media (Boone et al. 2013), suggesting that beetle–bacteria symbiosis reduces the efficacy of host plant defence responses. Similarly, secretion of symbiotic bacteria by the Colorado potato beetle during feeding elicits salicylic acid-regulated defences, which inhibits plant activation of jasmonate-mediated resistance (Chung et al. 2013), and gut microbiota of woodrats that detoxify secondary metabolites produced by their creosote bush food plant (Kohl & Dearing 2012). A second general mechanism of microbially mediated defence is priming of the host's immune system with infection by the symbiont, increasing resistance to subsequent parasites or pathogens (Little & Kraaijeveld 2004). This mechanism is not limited to higher animals but occurs also in invertebrates and plants (Moreira et al. 2009; Jung et al. 2012; Hussa & Goodrich-Blair 2013). For example, colonization of plant roots by mycorrhizal fungi can induce resistance to plant pathogens by the priming of jasmonic acid-dependent defences (Cameron et al. 2013), and leaf endophytes of trees reduce infection by plant pathogens (Arnold et al. 2003; Busby et al. 2013). Microbial symbionts can therefore serve as a vaccine to enhance immunological defence. The role of microbial symbionts in immunological priming also raises the more general question of how hosts permit infection by beneficial symbionts while simultaneously discriminating against damaging pathogens. Other potential mechanisms of defensive symbiosis may exist. Formation of biofilms or epiphytic layers can provide a physical or chemical barrier between host and pathogen. For example, seaweeds are often colonized by epiphytic bacteria with antifouling properties that protect chemically undefended seaweeds from secondary colonization by detrimental epiphytic growth (Egan et al. 2013). Similarly, skin bacteria of amphibians can reduce susceptibility to chytridiomycosis (Daskin & Alford 2012). Microbial symbionts may also directly interfere with the growth or replication of the pathogenic agent. For example, some Wolbachia strains protect Drosophila hosts against RNA virus infection, suggesting that Wolbachia may interfere with viral replication (Hedges et al. 2008; Teixeira, Ferreira & Ashburner 2008; Osborne et al. 2009). Other potential mechanisms of defence may include competitive exclusion of pathogens by symbionts (Koch & Schmid-Hempel 2011) and use of protective viruses by parasitoid wasps that inject symbiotic polydnaviruses during oviposition to suppress the host's immune response (Strand & Burke 2012). Macro-organisms possess a variety of physical, chemical and immune defence mechanisms. Under what circumstances are symbiont-based defence favoured over inherent defence? Presumably, microbial symbionts provide services that the host is not capable of, or they provide services more cheaply than can the host themselves. For example, the fungal endophyte-derived alkaloids found in many grasses occur in a plant family that does not typically produce the diverse secondary compounds found in many other plant families (Clay & Schardl 2002). Likewise, the capacity for antibiotic production is limited to bacteria and fungi but has been exploited by various insect groups via symbiosis with antibiotic-producing partners (Currie, Mueller & Malloch 1999a; Currie et al. 1999b; Kaltenpoth et al. 2005). The relative costs and benefits of microbial symbiosis vs. innate defensive mechanisms may vary within a host species or among groups of closely related species as evidenced by trade-offs between defensive strategies. For example, fungus-growing Trachymyrmex ants exhibit a trade-off where species with bacterial symbionts do not exhibit behaviours to reduce Escovopsis infection and species with strong behavioural responses do not possess antibiotic-producing symbionts (Fernández-Marín et al. 2013). Detailed analysis of the spatial and temporal patterns of defensive symbiosis and inherent host defences will shed light on the costs and benefits of each strategy. A key question is whether defensive symbionts are fixed in host populations or whether they are dynamic in space and time, contingent on variable pest pressure? There are well-documented cases of rapid changes in symbiont distribution and prevalence (Oliver et al. 2008; Jaenike et al. 2010), but the underlying causes may vary. In cases where symbionts cause reproductive incompatibility or induced parthenogenesis, symbionts can sweep rapidly through host populations (Perlman, Kelly & Hunter 2008; Himler et al. 2011). Relatively few experiments have directly manipulated defensive symbionts and natural enemies and measured host population responses, but they demonstrate that symbiosis is favoured in the presence of natural enemies and that symbiont prevalence increases over time. For example, endophyte-infected tall fescue grass (Lolium arundinaceum) exhibits increased resistance to herbivores due to the production of alkaloids by the symbiont (Bush, Wilkinson & Schardl 1997; Rudgers & Clay 2008). In a factorial experiment where insect and mammalian herbivory were manipulated independently, infection frequency increased significantly faster in populations with the greatest level of herbivory (Clay, Holah & Rudgers 2005). The endophyte spreads only through vertical transmission, so increasing infection frequency reflects the greater relative fitness of infected individuals under herbivore pressure (see also Koh & Hik 2007). Similarly, in population cage experiments with the pea aphid (Acythrosiphon pisum), the frequency of the protective endosymbiont Hamiltonella defensa increased dramatically after exposure to parasitoid wasps but decreased in the absence of parasitoids (Oliver et al. 2008). Evaluation of symbiont dynamics over time can also provide insights into factors leading to intermediate or fluctuating infection frequencies and frequency-dependent selection. Understanding the mechanisms of symbiont transmission will provide additional insights into their dynamics over space and time. Most cases of defensive symbiosis involve vertically transmitted symbionts (through eggs, seeds, maternal environment, etc.), ensuring continuity of symbiosis across generations. A general threat to mutualism is the exploitation of host resources by symbionts that do not provide any benefits (i.e. cheaters; Bronstein 2001; Sachs et al. 2004; Orona-Tamayo & Heil 2013). Hereditary symbioses provide strong sanctions against cheaters because vertically transmitted pathogens and their hosts will go extinct (Ewald 1987; Lipsitch, Siller & Nowak 1996; Haine 2008). Given the strong correlation between mutualism and vertical transmission, vertically transmitted microbes should be highly represented in defensive symbioses. Nevertheless, non-hereditary symbionts such as mycorrhizal fungi or leaf endophytes can confer protection against pathogens (Arnold et al. 2003; Busby et al. 2013), possibly by priming the plant immune system, and socially transmitted gut microbiota can protect bumblebees against intestinal parasites (Koch & Schmid-Hempel 2011). Defensive toxins, and the genetic machinery to synthesize them, may represent highly specific constitutive traits that are best maintained by vertical transmission. By contrast, immune system-priming mechanisms may be better accomplished by symbionts that are not permanent residents of the host. One overarching question is how do defensive symbioses affect the structure and dynamics of communities and ecosystems? With defensive symbiosis, we should predict that the prevalence of the host and symbiont will increase while the prevalence of enemies will decrease. But we have limited understanding of how the density of natural enemies is affected by defensive symbiosis, and how those changes cascade through higher and lower trophic levels. In one example, manipulation of endophyte infection of tall fescue grass in field plots demonstrated that the structure of the plant community in plots with the symbiont became increasingly dominated by the host grass over time, while the abundance and diversity of insect herbivores in the same communities decreased (Clay, Holah & Rudgers 2005; Rudgers & Clay 2008). However, there were no changes in plant productivity (Clay & Holah 1999), and as a result, endophyte infection altered the relationship between diversity and ecosystem processes (Rudgers, Koslow & Clay 2004). There is a need for additional manipulative field studies as well as theoretical models exploring community dynamics of defensive symbioses analogous to models of nutritional or pollination symbioses (Schwartz & Hoeksema 1998; Bronstein 2001; Sachs et al. 2004; Holland & DeAngelis 2010). For example, if the density of enemies decreases with defensive symbiosis, then the advantage of symbiosis would also decrease, potentially leading to time-lagged, oscillatory dynamics in the numbers of symbiotic hosts and their enemies. This type of response may be less likely where interactions are non-specific or where there is little cost to symbiont infection. More research is required to evaluate the impacts of defensive symbioses on larger ecological processes. The goal of this Special Feature is to explore the diversity, mechanisms and consequences of defensive symbiosis mediated by micro-organisms to help organize and interpret the growing body of work and place it within a broader ecological and evolutionary context of mutualism and symbiosis. In the accompanying papers, leading researchers in the field synthesize their own and related research on defensive symbiosis and provide independent perspectives on the current state of the field and future directions. Kaltenpoth & Engl (2014) consider defensive symbiosis in the Hymenoptera. In addition to the well-known association of leaf cutter ants and wasps with antibiotic-producing symbiotic bacteria, there exist several other mechanisms of defence in this diverse and important insect group, including the use of symbiotic viruses to disable defensive responses of victims attacked by parasitoids. Lopanik (2014) explores the great diversity bioactive secondary chemistry in marine organisms and their role in defensive symbiosis. Many important marine groups such as corals and sponges represent complex symbiotic amalgamations & et al. and marine to more potential cases of defensive symbiosis. & (2014) the important of interactions of symbionts within the same host. Most plants host a great diversity of fungal endosymbionts that can exhibit variety of on their hosts from to The specific of these symbionts on the and of their hosts may be on their interactions with each other such that the of symbionts on hosts may be or & (2014) on bacterial symbionts of insects and explore their population dynamics and fluctuating infection frequencies in natural host and fluctuating infection frequencies of defensive symbionts that their relative costs and benefits vary in time and The rapid and populations of insects systems to explore dynamics of defensive symbiosis. Panaccione, Beaulieu & Cook (2014) explore the of fungal endosymbionts of plants that produce bioactive alkaloid While the role of fungal endophytes in protection of grass hosts from herbivores has been understood for several research is that the produced by many poisonous plants are microbial in and the microbes are and vertically transmitted through seeds, suggesting of a common defence strategy. common from this of including of based on the larger community of similar defensive by independent groups of organisms and the exploitation of defensive symbionts by by symbionts and by enemies for a range of ecological The on defensive symbiosis a that the and diversity of microbes a of and immunological mechanisms by which host defence can be and that research has been by the of and molecular The of research on defensive symbiosis have many for ecological communities and agriculture and human health. would like to and other of the for their support and help in this the for their contributions and the for their and