Role of microorganisms in the evolution of animals and plants: the hologenome theory of evolution
Editor: Ramón Díaz Orejas
Correspondence: Eugene Rosenberg, Department of Molecular Microbiology and Biotechnology, Tel Aviv University, Ramat Aviv, Israel 69978. Tel.: +972 3 640 9838; fax: +972 3 642 9377; e-mail: email@example.com
We present here the hologenome theory of evolution, which considers the holobiont (the animal or plant with all of its associated microorganisms) as a unit of selection in evolution. The hologenome is defined as the sum of the genetic information of the host and its microbiota. The theory is based on four generalizations: (1) All animals and plants establish symbiotic relationships with microorganisms. (2) Symbiotic microorganisms are transmitted between generations. (3) The association between host and symbionts affects the fitness of the holobiont within its environment. (4) Variation in the hologenome can be brought about by changes in either the host or the microbiota genomes; under environmental stress, the symbiotic microbial community can change rapidly. These points taken together suggest that the genetic wealth of diverse microbial symbionts can play an important role both in adaptation and in evolution of higher organisms. During periods of rapid changes in the environment, the diverse microbial symbiont community can aid the holobiont in surviving, multiplying and buying the time necessary for the host genome to evolve. The distinguishing feature of the hologenome theory is that it considers all of the diverse microbiota associated with the animal or the plant as part of the evolving holobiont. Thus, the hologenome theory fits within the framework of the ‘superorganism’ proposed by Wilson and Sober.
More than a hundred years of biological research has demonstrated the importance of microorganisms in the health and disease of higher organisms. As a result of the recent development of culture-free molecular techniques, it is now accepted that in many cases the number of symbiotic microorganisms and their combined genetic information far exceed that of their hosts. The diverse types of symbioses between microorganisms and eukaryotes have received growing attention with regard to many different features of their complex interactions, such as the diversity and abundance of the symbionts, the type of advantage or harm the partners experience, how the interaction is initiated and, in recent years, also the genes responsible for establishing the symbioses (Hentschel et al., 2000). Probably the most well-studied symbiotic system is man/woman and their microbiota, where the well-acknowledged impact of microorganisms on the human body has led O'Hara & Shanahan (2006) to coin the gut flora as ‘a forgotten organ’ and Relman & Falkow (2001) to suggest a second human genome project directed at the endogenous microbiota as its target. The latter is now being carried out as part of the US NIH Roadmap Initiatives called the ‘Human Microbiome Project’ (http://nihroadmap.nih.gov/hmp/).
If microbial symbionts play such an important role in the lives of their eukaryotic hosts, why should they not also play a role in the evolution of these higher organisms? In the present paper, we shall address this question by developing and discussing the hologenome theory of evolution, which was briefly mentioned in a previous review on coral microbiology (Rosenberg et al., 2007). The hologenome is defined as the sum of the genetic information of the host and its microbiota. In the hologenome theory of evolution, we suggest that the holobiont (Margulis, 1993; Rohwer et al., 2002) (the host and its symbiotic microbiota) with its hologenome, acting in consortium, should be considered a unit of selection in evolution, and that relatively rapid variation in the diverse microbial symbionts can have an important role in the adaptation and evolution of the holobiont.
In essence, the hologenome theory of evolution focuses on the holobiont as a single dynamic entity in which a vast amount of the genetic information and variability is contributed by the microorganisms. Evolution of the holobiont can occur by changes in the host genome and/or in any of the associated microbial genomes, and relies on cooperation between the genomes within the holobiont, as much as on competition with other holobionts. Although much of the important research on symbiosis has been carried out with a small number of model systems involving a single major symbiont, the hologenome theory places importance not only on these major symbionts but also on the enormously diverse associated microbiota, which have only been uncovered in recent years using molecular techniques.
A large body of empirical data provides the foundation for the hologenome theory of evolution. We choose to discuss this information within the following framework: (1) all animals and plants establish symbiotic relationships with diverse microorganisms. (2) Symbiotic microorganisms can be transmitted between generations with fidelity. (3) The association between host organism and its microbial community affects the fitness of the holobiont within its environment. (4) Genetic variation in holobionts can be enhanced by incorporating different symbiont populations and can change under environmental demand more rapidly and by more processes than the genetic information encoded by the host organism alone.
Before elaborating on the above four points and examining the hologenome theory in the light of other evolutionary concepts, we would like to mention briefly how we came upon this theory. In 1996, we discovered bacterial bleaching of corals (Kushmaro et al., 1996). After 6 years of studying the mechanisms of infection (Rosenberg & Falkovitz, 2004), we observed that the coral had become resistant to infection by the specific pathogen, Vibrio shiloi. Because corals possess a restricted adaptive immune system and do not produce antibodies, we presented the coral probiotic hypothesis to explain the development of resistance to infection by V. shiloi (Reshef et al., 2006). The hypothesis posits that a dynamic relationship exists between symbiotic microorganisms and corals under different environmental conditions that selects for the most advantageous coral holobiont in the context of the prevailing conditions. The hologenome theory is a generalization of this concept.
All animals and plants establish symbiotic relationships with microorganisms
Eukaryotes presumably arose from prokaryotes (Margulis, 1993) and have remained in close relationship with them ever since (Hickman, 2005). It is therefore not surprising that the surfaces of animals and plants contain a great abundance and variety of microorganisms. In addition, some microorganisms are able to grow inside animal or plant cells, i.e., endosymbionts. As mentioned above, the number of microbial cells and their combined genetic information often far exceed that of their hosts. Taking the Homo sapien as an example, it has been reported that the number of bacterial species colonizing the human body, based on 16S rRNA gene analysis, is c. 2000 (Henderson, 2005). Because the average number of genes in each bacterial species is around 2500, it follows that each person contains an addition of 5 million bacterial genes to his genome. A conservative estimate that 5% of these bacterial genes are not shared among the bacteria yields 250 000 unique bacterial genes. By comparison, the human host genome contains about 20 500 genes (She et al., 2004). This large number of microbial genes considerably increases the potential for change in the hologenome.
At this point, it is useful to provide some common definitions. The term ‘symbiosis’ was first coined by Anton de Bary in the mid 19th century as ‘the living together of different species’. This broad definition is generally accepted and easily comes to terms with the hologenome theory. The symbiotic system is usually constructed from a large partner termed the ‘host’ and smaller partners called ‘symbionts’. This arbitrary division by dimension between host and symbiont may not fit such systems as the holobiont described above because size can also be measured by cell number or by genome size, and in the case of many holobionts, the microbiota outnumber their host. In spite of these limitations, we found it appropriate to continue using the classical terms. Endo- and exosymbionts refer to those living inside or outside host cells, respectively.
Because the vast majority of microorganisms that have been observed on or in animal and plant tissues cannot be cultured, current research on the diversity of microorganisms associated with a particular species relies primarily on culture-free DNA-based technology (Hugenholtz et al., 1998). Although censuses of microorganisms associated with different animal and plant species are only in an early stage, certain interesting generalizations have emerged: (1) The diversity of microbial species associated with a particular animal or plant species is high (Table 1). (2) The host-associated microbial community is very different from the community in the surrounding environment (Chelius & Triplett, 2001; Frias-Lopez et al., 2002; Rohwer et al., 2002; Sharp et al., 2007). (3) In some cases it has been shown that similar, but not identical, microbial populations are found on the same species that are geographically separated, while different populations are found on different species at the same location (Rohwer et al., 2002; Lambais et al., 2006; Fraune & Bosch, 2007). (4) Different microbial communities often dominate different tissues of the same organism (Tannock, 1995; Koren & Rosenberg, 2006; Dethlefsen et al., 2007). (5) In several cases where a large diversity of associated bacterial species exists, certain bacterial groups dominate. For example, the human gut has been reported to contain more than 1000 bacterial species (Rajilić-Stojanovićet al., 2007), but only two divisions, Bacteroidetes and Firmicutes, make up 99% of the total bacterial population (Ley et al., 2006a).
It is important to note that the numbers of reported microbial species associated with specific organisms are probably underestimations, mainly because of the failure to score rare species. In support of this argument, a recent paper by Frank & Pace (2008), using the powerful technique of metagenomics, estimated the number of bacterial species in the human gut to be more than 30 times higher than the estimate mentioned above.
The association of microorganisms with hosts can take many different forms. Some may be transitory and have little effect on adaptation or evolution of the holobiont. At the other extreme, there are several examples of well-studied long-lasting interactions (e.g. the rumen system) between host and microorganism, that can lead to total dependence of one on the other. Between these two extremes lies a gradient of interactions of varying strengths, including pathogenesis. It should be noted that the study of the interactions between hosts and their associated microorganisms is complicated by the fact that most associated microorganisms have not been cultured and that most of the interactions involve more than one microorganism with the host, e.g., the human gut microbiota (Ley et al., 2006a) and the coaggregating bacteria in the human mouth (Foster & Kolenbrander, 2004).
Let us consider some factors that determine the diversity of microorganisms associated with the holobiont. We shall first consider those characteristics that would result in high diversity. Many microorganisms are specialists. Given that hosts provide a variety of different niches that can change with the developmental stage of the host, the diet and other environmental factors, a diverse microbial community is established, with different microbial strains filling the different niches. This microbial diversity, and therefore its versatility, may allow the holobiont as a whole to function more optimally and adapt more rapidly to changing conditions. The idea that microbial diversity can play a critical role under conditions of fluctuating environments has been referred to as the insurance policy hypothesis (Yachi & Loreau, 1999).
Another factor that contributes to bacterial diversity is bacteriophages. It has been established that high concentrations of bacteriophages are present in animal and plant tissues (Breitbart et al., 2003). If any microorganism becomes too abundant, it may be lysed by bacteriophages. This concept, referred to as the ‘kill the winners’ hypothesis (Thingstad & Lignell, 1997), is supported by mathematical models of the bacteria: bacteriophage dynamics (Weitz et al., 2005).
On the other hand, there exist opposing forces that limit the number of strains that can survive and become established in the holobiont, notably the innate and adaptive immune systems. The innate or nonspecific immune system is the first line of defense and includes physical barriers, antimicrobial molecules, enzymes, specific binding proteins for microbial attachment (e.g. the peptidoglycan-binding protein and lectin complement system), production of reactive oxygen species and phagocytes (Iwanaga & Lee, 2005). Interestingly, resident symbiotic bacteria are also part of the innate immune system – by occupying potential adhesion sites and by producing antibiotics (Ritchie, 2006). The adaptive or the specific immune system in vertebrates includes specific recognition of ‘foreign’ microorganisms, generation of responses to eliminate these microorganisms and development of immunological memory to hasten the response to subsequent infections with the same microbe. In essence, the immune system of the host is responsible for both limiting the types of microorganisms that can survive within the host and recognizing and accommodating the normal microbiota, thereby regulating the kinds of microorganisms that can reside in the holobiont. Also, it is important to note that plants have evolved myriad phytochemicals, whose purpose is to prevent infection by harmful microorganisms (Wallace, 2004) and enable coexistence with beneficial ones (Smith et al., 1999; Stougaard, 2000; Wilkinson, 2001).
The human gut is an example in which, although one finds a substantial diversity of bacterial strains, they belong by and large to only 30–40 species, which themselves belong to two main bacterial divisions (out of 70 divisions identified), the Firmicutes and the Bacteroidetes, while the Archaea are represented mainly by only one strain, Methanobrevibacter smithii (Ley et al., 2006a). The gut, in spite of providing a diversity of niches for microbial colonization, imposes strict requirements for their survival, the important ones being adaptation to digestive enzymes, evading the potent innate and adaptive immune systems, escaping washout from the gut and the ability to live anaerobically. These strict requirements will force a narrowing of the variability of microorganisms, leaving only those that are able to create a viable and well-adapted holobiont with their host.
Transmission of symbionts between holobiont generations
The hologenome theory of evolution relies on ensuring the continuity of partnerships between holobiont generations. Accordingly, both host and symbiont genomes must be transmitted with accuracy from one generation to the next. The precise modes of vertical transmission of host genomes are well understood and need not be discussed here. However, in recent years, it has become clear that microbial symbionts can also be transmitted from parent to offspring hosts by a variety of methods. McFall-Ngai (2002), in an insightful review on the influence of bacteria on animal development, divided the modes for maintaining symbionts faithfully between generations into two categories, trans-ovarian and environmental transmissions, while correctly acknowledging that there are numerous intermediate cases. We would like to take this approach one step further by suggesting that the numerous intermediate cases, in fact, best describe the large variety in modes of transmission, which are known at present to reconstitute plant and animal holobionts. It is this continuum of modes of transmission from direct to indirect that makes it impractical to place them in any specific categories.
Table 2 presents some of the diverse modes of transmission of symbionts in animals and plants. The first case listed in the table is the mitochondria and chloroplasts, which can be considered (extreme) symbionts that are transmitted by the most direct mode, namely, cytoplasmic inheritance. Direct transmission from parent to offspring also occurs with other symbionts where the microorganisms are in or on the reproductive cells. For example, in the aphid–Buchnera symbiosis, bacteria are intracellularly situated in bacteriocytes and are transferred to and transmitted via the eggs (Baumann et al., 1995). Direct contact is another slightly less direct mode of transmission demonstrated in mammals in which many of the symbionts are derived during passage through the birth canal or subsequently by close physical contact with parent or family and community members. In humans, for example, a greater similarity was observed within-family members when compared with between families (Zoetendal et al., 2001) and within the same European population as compared with between different European populations (Mueller et al., 2006). The similarity of microbiota within family or group can evolve as a result of genetic relatedness and/or because of early and similar physical contact and transmission from the parent.
Table 2. Modes of transmission of symbionts and their contribution to the fitness of the holobiont
|• All eukaryotes: Mitochondria||Cytoplasmic inheritance||Respiration||Margulis (1993)|
|• Plants: Chloroplasts||Cytoplasmic inheritance||Photosynthesis||Margulis (1993)|
|• Aphids: Buchnera sp. (primary-endosymbiont)||Via intracellular bacteria in bacteriocytes; present in ova||Provision of specific required amino acids lacking in the plant sap diet||Baumann et al. (1995), Wernegreen (2002), Perez-Brocal (2006)|
|• Aphids: Secondary endosymbionts||Via intracellular bacteria in addition to environment||Growth at high temperature; resistance to parasites||Sandström et al. (2001), Russell et al. (2003)|
|• Termite: Microbiota in hind gut||Feces of adult termites fed to newly hatched juveniles||Utilizable energy and carbon; nitrogen metabolism; recognition signal from odor of bacterial metabolites||Abe et al. (2000), Minkley et al. (2006)|
|• Anthropods/nematodes: Wolbachia spp.||Intracellular transmission via egg cytoplasm||Fertility and sex determination||Veneti et al. (2005)|
|• Stinkbug midgut: Burkholderia||Specific transmission via environment||More efficient food utilization||Kikuchi et al. (2007)|
|• Squid nidamental gland: Microbiota||Via cover of eggs originating from the gland||Protection of eggs and embryos against pathogens||Kaufman et al. (1998), Barbieri et al. (2001)|
|• Squid light organ: Vibrio fischeri||Environmental from surrounding water||Camouflage against predators||McFall-Ngai (1999)|
|• Corals: Microbiota||From the environment and by vegetative reproduction||Photosynthesis (intracellular algae); nitrogen fixation; protection against pathogens||Rohwer et al. (2002), Buddemeier et al. (2004), Rosenberg et al. (2007)|
|• Sponges: Microbiota||Environmental in addition to possible transmission from parent||Breakdown of complex polymers; nitrogen cycling; protection against pathogens||Webster et al. (2001), Hickman (2005), Taylor et al. (2007)|
|• Cow rumen: Microbiota||Physical contact with parents and via food contaminated with feces and sputum||Provision of all nutritional needs from cellulose||Dehority (2003), Russell & Rychlik (2001)|
|• Whale forestomach: Microbiota||Physical contact with mother||Provision of nutritional needs from chitin and other complex organics||Herwig et al. (1984), Olsen et al. (1994), Olsen et al. (2000)|
|• Human gut and mouse model: Microbiota||Via physical contact and from environment||Protection against pathogens; stimulation of immune system; angiogenesis; vitamin synthesis; fiber breakdown; fat storage||Hooper et al. (2002), O'Hara & Shanahan (2006), Ley et al. (2006a), Xu et al. (2007)|
|• Land plants: Mycorrhiza fungi||Via seeds on ground and by vegetative reproduction||Supply of minerals from soil||Wilkinson (2001), Wang & Qui (2006)|
|• Nonphotosynthetic plants–fungi (some orchids)||Seeds falling on ground and by vegetative reproduction||Supply of minerals from soil and organics from other plants||Bidartondo (2005)|
|• Legume plants: Rhizobium||Environmental from surrounding||Nitrogen fixation||Stougaard (2000), Jones et al. (2007)|
|• Plant: Growth-promoting rhizobacteria||Environmental from surrounding soil||Protection against pathogens; nitrogen metabolism; acceleration of mineralization; carbon cycling; salt tolerance||Smith et al. (1999), Somers et al. (2004), Singh et al. (2004), Egamberdieva et al. (2008)|
|• Rice plants: Azoarcus sp.||From surrounding soil||Associative nitrogen fixation||Hurek & Reinhold-Hurek (2003)|
Another slightly less direct mode of transmission is used in the termite hindgut–microbiota symbiosis where feces of adult termites (containing abundant microorganisms) are fed to newly hatched juveniles by workers in the colony (Abe et al., 2000). Similarly, in the bovine rumen–microbiota symbiosis, the offspring acquire the microbiota by feeding on grass that is contaminated with feces and sputum from their parents, as well as by passage through the birth canal (Dehority, 2003).
A less direct, but precise, mode of transmission is exemplified in the squid light organ–Vibrio fischeri symbiosis where the high specificity of the light organ for V. fischeri has evolved together with the need to acquire the motile bacteria from the surrounding seawater. The adult squid releases large amounts of V. fischeri into the water at dawn every day, assuring that sufficient symbionts are available to colonize the hatchlings (McFall-Ngai, 1999). Furthermore, the squid provides a habitat in which only V. fischeri that emits light is able to maintain a stable association (McFall-Ngai, 1999; Visick et al., 2000). Thus, even in transfer via the environment (often referred to as horizontal transfer), the holobiont is reconstituted faithfully.
Some animals and most plants can develop from cells other than gametes, namely, from somatic cells (Buss, 1987). The most striking example is vegetative reproduction in plants. When a fragment of a plant falls to the earth, it may root and grow into a fully developed plant. In such cases, it will clearly contain some of the symbionts of the original plant (direct transfer). In addition, it will most likely incorporate rhizosphere fungi (mycorrhiza) and other microorganisms from the soil adjacent to the parent. A recent publication on symbionts of the metazoan Hydra demonstrates both the specificity and the accuracy of transmission (Fraune & Bosch, 2007). The Hydra reproduces similarly to plants, namely, vegetatively (by budding) and sexually. The report showed, first, that two different species of Hydra were colonized by different communities of microorganisms and, second, in both cases the two species of Hydra were populated with similar microorganisms both in the laboratory and in nature, even after more than 30 years of maintaining the animals in the laboratory.
Summarizing this section, we suggest that regardless of the mechanism used, there is now growing evidence (Cary, 1994; Gros et al., 1996; Krueger et al., 1996; Kaufman et al., 1998; Mateos et al., 2006; Fraune & Bosch, 2007; Kikuchi et al., 2007; Sharp et al., 2007) that the microbial component of the holobiont is transferred from generation to generation. The large varieties in modes of transmission have an interesting implication: individuals can acquire and transfer symbionts throughout their lives, and not just during their reproductive phase. This means that the parents, grandparents, nannies, siblings, spouses or any organism that is in close contact with an offspring can transfer symbionts and thereby influence the holobiont of the next generation.
Cooperation between the host and the microbiota contributes to the fitness of the holobiont
Natural selection is the central concept of Darwinian theory – the fittest survive and spread their advantageous traits through populations, but ‘fittest’ is not an absolute property because it varies with environmental influences. Considering the holobiont as a unit of selection in evolution, we argue that the cooperation between the normal microbiota and the host generally leads to improved fitness. In addition, the genetic diversity of the microbiota can extend the range of environments in which the holobiont can compete successfully.
Table 2 displays various representative symbiotic systems, indicating some of the ways in which the microorganisms contribute to the fitness of the holobiont. In several well-studied cases, neither the host nor the primary symbiont can survive without the other (absolute mutualism). For example, in the aphid–Buchnera symbiosis, the primary endosymbiotic bacterium has lost many genes required for independent growth during evolution, whereas the aphid partner depends on essential amino acids lacking in its diet that are synthesized and furnished by the symbiont (Baumann et al., 1995). While the aphid–Buchnera primary endosymbiosis is an example of absolute dependency, most of the symbiosis systems, as indicated in Table 2, are not based on life or death interactions, but rather the microbial partners contribute in different degrees to the holobiont's well-being. One such example, also found in the aphids, is the secondary endosymbioses with bacteria belonging to a number of distinct lineages within the Gamma- and Alphaproteobacteria. These microorganisms are intracellular endosymbionts, which are not essential for growth and reproduction, but that contribute to the fitness of the holobiont (Russell et al., 2003). Regardless of the extent of dependency, a large body of data has been accumulated in recent years demonstrating that both endosymbionts and exosymbionts play numerous roles in metabolism, regulation, disease resistance and in sex and fertility determination (which may lead to species determination) of their hosts.
We would like to present here briefly what are probably the best-studied metabolic systems in vertebrates, namely, the bovine rumen and the human/mouse gut symbioses. These systems have provided a wealth of detailed information on how diverse symbionts contribute to the health of the holobiont. Interestingly, in both cases the microbiota are extracellular and are introduced into the sterile newborn hosts via the environment, namely, by close contact with the parents (mostly the mother) and from the surroundings, enabling accurate transmission, with possible variations between generations.
The bovine rumen symbiosis has been studied extensively (Dehority, 2003), largely because of its obvious commercial importance. The rumen acts as a temperature-controlled anaerobic fermentor which is mainly fed ground cellulose-containing material and saliva from the cow's mouth. In the rumen, bacterial enzymes convert the cellulose into its glucose subunits, which are then fermented by different groups of bacteria to produce short-chain fatty acids. These fatty acids are absorbed through the wall of the rumen into the bloodstream, and then circulated via the blood to the various tissues of the body where they are respired. The microbial population of the rumen grows rapidly, and some of the microbial cells pass out of the rumen with undigested plant material into the lower stomachs, where they are digested by secreted enzymes of the host, providing nitrogenous compounds and vitamins that are absorbed by the animal. The bovine holobiont benefits from the cooperation by being able to grow and reproduce on a simple diet of cellulose, water and inorganic salts, in spite of the fact that the host lacks the ability to synthesize cellulases and some vitamins and essential amino acids. Although the ruminant symbiosis may be a special case, the concept that symbiotic microorganisms benefit their hosts by allowing them to derive energy from complex compounds and by providing essential vitamins and amino acids is a general phenomenon in animals.
Whereas in the bovine rumen's symbiotic system most of the information acquired to date deals with the metabolic interaction within the holobiont, in the human gut–microbiota relationship (or its model system in the mouse), a substantial amount of recent information has been gathered regarding many other facets of this interaction (Xu & Gordon, 2003; O'Hara & Shanahan, 2006; Xu et al., 2007). The diverse interactions reported between the human gut and its microbiota include not only the cooperation in food breakdown [e.g. fiber into short-chain fatty acids (Hooper et al., 2002)] and biotransformation of certain molecules [e.g. bile acids (Hylemon & Harder, 1998)] but also participation in the development and normal function of the innate and adaptive immune systems in the gut (Hooper & Gordon, 2001; Braun-Fahrländer et al., 2002; O'Hara & Shanahan, 2006), participation in the structural buildup of blood vessels [angiogenesis (Stappenbeck et al., 2002)] and participation in the regulation of fat accumulation (Bäckhed et al., 2004).
Germ-free animals, born and grown under sterile conditions, are a useful tool for studying the relationship between host and its microbiota (Wostmann, 1981). Research on germ-free mice exhibits significant differences in gut development and function as compared with mice grown conventionally, i.e., possessing normal gut microbiota. The germ-free mice demonstrate enlarged caeca (Wostmann, 1981), a slow digested food transit time (Abrams & Bishop, 1967), altered kinetics of epithelia turn-over in the small intestine (Savage et al., 1981), an increased caloric intake (Wostmann et al., 1983) and a greater susceptibility to infection (Silva et al., 2004). The influence of microbiota on energy metabolism in germ-free conventionalized mice was observed within 2 weeks of the introduction of microbiota (Bäckhed et al., 2004). It included microbial fermentation of polysaccharides not digested by the host, absorption of the microbially produced short-chain fatty acids, more efficient absorption of the monosaccharides from the intestine, conversion of breakdown products in the liver to more complex lipids and microbial regulation of host genes that promote fat deposition in adipocytes. These events were accompanied by lower food intake and higher metabolic rate. The cross-talk between microbiota and its mouse host regarding energy metabolism involves at least two demonstrated mouse factors (Bäckhed et al., 2007). In addition, it has been shown in mice (Ley et al., 2005; Turnbaugh et al., 2006) and humans (Ley et al., 2006b) that obesity is correlated with a higher proportion of microorganisms from the Bacteroidetes division as compared with those from the Firmicutes division. Moreover, in obese humans a gradual transition was observed from the obese microbiota to the lean microbiota during the course of a restrictive energy intake (Ley et al., 2006b).
These examples, the bovine rumen and the human/mouse gut, demonstrate the very tight and complex interactions occurring within the holobiont between the host and its microbiota, justifying the reference to the microbiota as an ‘additional organ’.
From the beginning of evolutionary thought, natural selection and speciation in an organism have been associated, more than anything else, with morphological changes. It is thus interesting that there are several examples of biological structures that appear to have evolved to accommodate symbiotic microbiota. These include both primary endosymbionts, such as aphid symbiosomes containing bacteriocytes with their intracellular membrane vesicles that house the symbiotic Buchnera, and exosymbionts, such as the bovine rumen that accommodate a large and diverse anaerobic cellulose-degrading community of exosymbiotic microorganisms. Another example is the light organ of squid, which houses a pure culture of the exosymbiotic V. fischerii. Over 15 000 plant legume species form specific associations with nitrogen-fixing Rhizobia that involve the formation of root nodules. Studies on the legume–Rhizobia symbiosis have demonstrated that biochemical cross-talk is a prerequisite for nodule formation and that the hybrid molecule leghemoglobin (part plant, part bacterial) maintains the low oxygen tension required for nitrogen fixation (Jones et al., 2007).
Pathogenic microorganisms are traditionally characterized as those that harm their animal or plant host by cell or tissue damage and sometimes death. Although pathogens represent only a small minority of the microorganisms that are associated with higher organisms, they have been studied extensively and much of what we know about host–microbial interactions has emerged from research on infectious diseases. One of the most important recent discoveries in this area is that many of the genes necessary for pathogen–host and symbiont–host interactions are located on similar mobile genetic elements (Hentschel et al., 2000). As discussed in the next section, genetic variability brought about by these elements plays an important role in the interactions of pathogens and symbionts with their hosts.
Genetic variation in holobionts
Variation is the raw material for evolution. According to the hologenome theory of evolution, genetic variation can arise from changes in either the host or the symbiotic microbiota genomes. Variation in host genomes occurs during sexual reproduction, chromosome rearrangements and ultimately by mutation. These same processes occur in microorganisms with the noteworthy difference that in haploid bacteria recombination occurs, within the same species, by conjugation, transduction and DNA transformation. In addition to recombination and mutation, changes in the genome of the microbiota of holobionts can occur by three other processes that will be discussed below: microbial amplification, acquisition of novel strains and horizontal gene transfer between different species. These latter three processes can occur rapidly under environmental demand and may be important elements in the evolution of animals and plants (Dinsdale et al., 2008). Let us also bear in mind that all modes of bacterial variation can be driven stochastically by random events (i.e. genetic drift) and/or can occur as a result of deterministic effects such as change of nutrients, phage infection or temperature change (Dethlefsen et al., 2006).
Microbial amplification is the most rapid and easy to understand mode of variation in holobionts. It involves changes in the relative numbers of the diverse types of associated microorganisms that can occur as a result of changing temperatures (for plants and invertebrates), nutrient availability, exposure to antibiotics or other environmental factors. The holobiont is a dynamic entity with certain microorganisms multiplying and others decreasing in number as a function of local conditions. An increase in the number of a particular microbe is equivalent to gene amplification. Considering the large amount of genetic information encoded in the diverse microbial population of holobionts, microbial amplification is a powerful mechanism for adapting to changing conditions. In fact, changes of symbiont populations as a function of external factors are well documented (Weimer et al., 1999; Russell & Rychlik, 2001; de la Cruz & Davies, 2005; Koren & Rosenberg, 2006).
Another mechanism for introducing variation into holobionts is acquiring new symbionts from the environment. Animals and plants come in contact with billions of microorganisms during their lifetime. It is reasonable to assume that occasionally, as a random event, one of these microorganisms will find a niche and become established in the host. Under the appropriate conditions, the novel symbiont may become more abundant and affect the phenotype of the holobiont. Unlike microbial amplification, acquiring new symbionts can introduce entirely new genes into the holobiont.
Horizontal gene transfer (between different microbial species) is an additional potent mechanism for generating variability in symbionts. Most prokaryotes possess different classes of mobile genetic elements that allow for the acquisition, loss or rearrangement of sometimes large regions of the bacterial genome. Horizontal gene transfer is mediated by transposons, plasmids, bacteriophages and genomic islands, which can be either on the bacterial chromosome or on the plasmids. Interestingly, genomic islands encode many functions necessary for bacteria–host interactions and are found in pathogens, where they are referred to as pathogenicity islands (Hacker & Kaper, 2000), as well as in beneficial symbionts, where they are called symbiosis islands (Finan, 2002). The proximity and, in many cases, the high density of bacteria within a holobiont would accelerate the rate of horizontal gene transfer. The evolutionary significance of horizontal gene transfer is that a large block of DNA, e.g., a symbiosis island, can be transferred from one bacterium to another in a single event. This has resulted, for example, in rapid evolution of diverse strains of nitrogen-fixing mesorhizobia in legumes (Nandasena et al., 2007). Horizontal gene transfer may also be a mechanism by which genetic information can be exchanged between pathogens and symbionts (Hacker et al., 2005).
Based solely on the host genome, animals and plants would evolve slowly because of (1) their relatively long generation times, (2) the fact that only changes in the DNA of the germ line are transmitted to the next generation and (3) often a whole set of new genes is required to introduce a beneficial phenotypic change. If the environment changes relatively rapidly, the host genome alone may not evolve quickly enough and the organism may lose competitiveness and, in the worse case scenario, may become extinct. We argue that rapid changes in the symbiotic microbiota could allow the holobiont to adapt and survive under changing environmental conditions, thus providing the time necessary for the host genome to evolve if required. In some cases, the required new function remains within the symbiotic microbiota. For example, the animals that first evolved to feed on cellulose never developed the genes to degrade cellulose, but rather adapted to provide better conditions for microbial degradation of the polymer.
In his book Evolution by Association, Sapp (1994) reviews the vast literature on the role of symbiosis in the evolution of animals and plants that dates from the 19th century. One of the major turning points in the debate on whether or not symbiosis is important in evolution was the demonstration (using DNA technology) that mitochondria, chloroplasts and probably centromeres were derived from bacteria. The concept of coevolution of symbiont and host, ultimately leading to a stable organelle, is now widely accepted by biologists. There are also a number of well-studied, highly stable primary endosymbiont–host associations, such as the aphid–Buchnera symbiosis, that are best understood in terms of coevolution (Baumann et al., 1995). However, previous evolutionary concepts did not consider the effects of the diversity of the microbial symbionts in evolution of animals and plants.
The distinguishing feature of the hologenome theory is that it considers all of the diverse microbiota associated with the animal or the plant as part of the evolving holobiont and that changing the microbial community by amplification, novel acquisition from the environment and horizontal gene transfer provide additional mechanisms for rapid evolution. To support this theory, we have discussed published results that exemplify the diversity of symbionts of higher organisms, their ability to be transmitted from one generation to the next with fidelity, their contributions to the fitness of the holobiont and their potential to change rapidly under environmental shifts. The consequences of considering all of the diverse symbiotic microbiota of the holobiont (endo- and exocellular) are far reaching. Most importantly, they can increase the genetic information of the holobiont several-fold. Although in many cases it is not presently known how this microbial information improves the fitness of the holobiont, we have presented some examples that demonstrate how microbially directed processes can regulate and extend the metabolic capability of holobionts. These include degradative and biosynthetic reactions in addition to regulatory mechanisms.
In terms of modern Darwinian theory, the unit of selection in evolution must contain two properties (Lloyd, 1994): the replicator or genome that is subject to variation, and the interactor or phenotype, that interacts with the environment in a way that creates differential reproduction. The data reviewed in this article clearly demonstrate that both properties are expressed by the holobiont. The usefulness of considering the holobiont, with its hologenome, as a unit of selection is that it makes adaptability to changing environments in a relatively short time frame more comprehensible and amenable to experimental tests.
Holobionts are complex systems. Thus, from a standpoint of evolutionary theory they include several functional levels: (1) the individual organisms (the host and its multitude of microorganisms), (2) cooperation between the individual organisms (Sachs et al., 2004; Foster & Wenseleers, 2006) and (3) different communities (Day et al., 2003; Leibold & Norberg, 2004) within the holobiont. Today the generally accepted driving force of evolution is the single organism as the unit of selection (Maynard Smith & Szathmary, 1995). Individual selection holds true for all participants of the holobiont, namely, for the host as well as for the individual microorganisms as claimed by West et al. (2006) describing a general evolution theory for microorganisms and by Dethlefsen et al. (2006) and Dethlefsen et al. (2007) reviewing the situations in the human gut and the human holobiont, respectively. Although individual selection is currently the accepted dogma, this paper offers the holobiont as a unit of selection in evolution. The major arguments that have been derived from the information presented here, favoring the holobiont over the individual animal or plant as a unit of selection, include: (1) there has never been any natural animal or plant free of microorganisms, (2) the holobiont has its own specific properties that are not necessarily the sum of those of the host+its microbiota, e.g., ‘Metabolomics of a superorganism’ (Goodacre, 2007), (3) holobionts have their specific structures, (4) each holobiont stands by itself, facing its environment, competing as a whole with other holobionts (whereas the smaller units of selection, namely, the separate genomes and individual genes, are selected for within the holobiont) and (5) the hologenome is transmitted from one generation to the next with reasonable accuracy. These features of the holobiont seem to fit best within the model of a ‘superorganism’ proposed by Wilson & Sober (1989), the superorganism being a multispecies community that functions as an organizational unit. In their paper, Wilson and Sober provide theoretical and empirical evidence for their levels-of-selection theory, which acknowledges the existence in evolution of hierarchical levels of selection from genes to multispecies communities, each of which is defined by being a unit of functional organization. Another model of a superorganism, through intergroup competition, suggested by Reeve & Hölldobler (2007) and based on individual selection, may also be considered when trying to understand the evolution of the holobiont.
Interestingly, the hologenome theory incorporates aspects of both Darwinism and Lamarckism. Individual organisms evolve by selection of random variants, whereas the holobiont can evolve by adaptive processes. Consider, for example, marine invertebrates living in an ocean of increasing temperatures. The hologenome will change as a result of the shift-up in temperature by amplification of those microbiota symbionts that grow better at the higher temperature and possibly by gaining more thermophylic microorganisms from the environment. The holobiont will not only be better adapted to its environment, but it will also have an increased probability of passing on the acquired genetic traits to the next generation, as discussed in previous sections. Thus, when considering a holobiont, rather than the individual organisms that comprise it, it is possible to have an inheritance of acquired characteristics. How important this is in long-term evolution is debatable. However, in the short term, inheritance of acquired characteristics by holobionts may help them survive, multiply and buy the time necessary for the host genome to evolve.
Evolution is an ongoing event. Currently, the rapid rise in global temperature is placing stress on many species and is threatening their survival (Harvell et al., 2002). The rate of adaptation and evolution is an important unknown in any prediction of climate impacts (Etterson & Shaw, 2001). Consideration of the role of symbiotic microorganisms in the evolution of animals and plants may modify predictions of their fate.
We thank R.J. Martinez, A. Marglin, H. Hartman, E.Z. Ron, E. Jablonka and D. Segal for helpful comments on the manuscript.