Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

The term symbiosis refers to the close ecological relation between two (or more) species, able to report benefits to all (mutualism) or some of the implied organisms, with or without harm of one of the involved species (parasitism or commensalism, respectively). Symbiosis is an important source of evolutionary innovation, with examples in the entire Biosphere, being even at the origin of the eukaryotic cell (Margulis 1993). Since then, stable symbioses have evolved independently many times in diverse groups of eukaryotes (Moya et al. 2008). Most symbioses have a demonstrated biochemical basis: in some cases one of the partners benefits from organic compounds produced by the other; in others, its waste products (mainly nitrogen compounds) are recycled by the other. In mutualistic symbioses, matter and energy flow in both directions, so that both partners benefit from the association.

Numerous eukaryotic groups maintain a mutualistic relationship with prokaryotic cells, especially because many eukaryotic lineages present limited metabolic capabilities. Animal metabolism, in particular, is relatively narrow, and essential molecules (such as amino acids, vitamins, or fatty acids) must be retrieved from the environment for survival. Animals with specialized feeding behaviors tend to establish symbiotic associations with microorganisms, which provide the nutrients that are deficient in their diets. In fact, most intracellular mutualistic symbioses between bacteria and animals that have been analyzed at the genomic level (involving insects, nematodes and deep-sea animals, Table 1) are related to nutrient provision. Regarding insects, the most studied and diverse invertebrate group on earth, the presence of such associations throughout most of their evolutionary history suggests that symbiosis has been a driving force in the diversification of the group.

Table 1 Remarkable data of the mutualistic animal endosymbiont genomes sequenced as of November 2009
Full size table

A high proportion of mutualistic symbiotic relationships established by insects imply the participation of bacteria. Frequently, the association is so tight that it is called endosymbiosis, when the bacteria (endosymbiont) obligatorily live inside specialized eukaryotic cells (bacterocytes), which can even form a specialized organ (the bacteriome), located inside the abdominal cavity of the insect. It has been estimated that up to a 15% of all insects species carry bacterial endosymbionts (Baumann 2005), attributing to them the great adaptive success of the Insecta class, by making possible the colonization of new ecological niches and allowing them to feed on restricted diets (such as plant sap, cereals or blood), poor in some essential nutrients that are provided by the endosymbionts. The elimination of these bacteria, consequently, critically diminishes the biological fitness of the host, affecting to its growth, fertility or longevity.

The first to notice the link between a restricted diet and the presence of endosymbiotic bacteria in insects was Paul Buchner (1965), who coined the terms of primary (P-) endosymbiont and facultative or secondary (S-) symbiont, based on its morphologic characterization and its presence among the individuals of a certain taxonomic group. This classification has later been validated by means of molecular genetics techniques and the complete sequencing of genomes of an increasing number of endosymbiotic bacteria (Baumann 2005). Buchner classified as P-endosymbionts those bacteria of a unique morphological type that are present in all the insects of a defined taxonomic group, confined inside specialized insect cells located in the abdominal cavity. Such P-endosymbionts are essential for its host fitness and survival. On the other side, the S-symbionts were identified as morphologically diverse bacteria, without a defined spatial distribution in the host body, and whose sporadic presence in some individuals of a defined taxon suggested that they were not essential for host survival. In fact, S-symbionts vary in number and distribution among species and among individuals of the same species, and can live outside of the eukaryotic cells. The congruence between the phylogenetic trees based on host and their corresponding P-endosymbionts sequences, respectively, indicate that each endosymbiont derives from a single infection of the ancestor of the host by the ancestors of its P-endosymbiont, and follows a path of vertical evolution, promoted by their exclusively maternal transmission between insect generations (Munson et al. 1992). On the contrary, the topological incongruence between the phylogenetic trees based on sequences of the S-symbionts and their hosts and the polyphyletic character of such bacteria suggest the existence of multiple events of infection and/or the horizontal transfer of these bacteria among insects (Russell et al. 2003).

Buchner and the early researchers of prokaryote–eukaryote symbioses did not differentiate between the two prokaryotic domains, since the existence of archaea was not recognized until the nineties of the past century (Woese et al. 1990). Therefore, certain early symbioses described as involving “bacteria” were, in fact, involving archaea (Hackstein et al. 2006). This was the case for many methanogenic symbionts hosted by protists that were described in termite and cockroach guts in the 1980s. However, symbioses between arthropods and methanogenic archaea do not seem to have a nutritional foundation. The archaea are always restricted to the hindgut, where they can appear free in the gut lumen, attached to digesta or to the hindgut wall, or as endosymbionts of anaerobic ciliated protozoa that occupy the same gut compartment. Little is known about the function of methanogenic archaea in the guts of arthropods, besides their role in lowering H2 partial pressure by producing methane, while the archaea uses the hydrogen as a source for methane formation, which indicates that the relationship is mutualistic. Phylogenetic studies have been performed on anaerobic heterotrichous ciliates that keep an endosymbiotic association with methanogenic bacteria (van Hoek et al. 2000). This is an interesting study group because they live in the most divergent niches, such as marine and freshwater sediments and the intestinal tract of animals. The topology of the phylogenetic trees indicates that the coevolution of host and endosymbiont can only be demonstrated in a few analyzed cases, which most likely means that, although probably hydrogenosome-bearing ciliates acquired methanogenic endosymbionts at the very beginning of their evolution towards anaerobiosis, prior to the anaerobic heterotrichous ciliates radiation, endosymbiont replacements must have accompanied the evolution of these protists. In addition to its role in hydrogen transfer, it has also been proposed that artropode intestinal methanogens can contribute to nitrogen–carbon balance in the hindgut by the fixation of atmospheric nitrogen, since these archaea posses a complete gene repertoire needed for nitrogen fixation (Raymond et al. 2004). Whole genome studies will help to identify other possible benefits of these methanogenic archaea to their hosts.

Even though new data are accumulating on archaeal symbionts of animals, most analyses concentrate on nutritional and physiological aspects. At the beginning of the genomics era, research on prokaryote endosymbionts of eukaryotic cells focused on a limited group of arthropods, mostly sap-sucking insects (Hemiptera: Sternorrhyncha), and those have been for quite a while the main models used to define the evolutionary and molecular aspects of prokaryote-animal symbioses. Therefore, we will focus mostly on bacterial endosymbionts of insects to detangle the molecular aspects of these symbioses from a genomics perspective, paying special attention to the genomic changes experienced by the bacterium in their adaptation to an endosymbiotic lifestyle.

The advent of genomics allowed the complete sequencing of genomes and the development of metagenomic methods, making possible the study of environmental samples and non-cultivable microorganisms, thus offering new opportunities for symbiosis research. The availability of many genomes of bacterial endosymbionts, opens the door to comparative analyses among them, unveiling common molecular aspects regarding the establishment and maintenance of symbiotic associations. In order to completely understand the different stages of genomic evolution of bacterial endosymbionts, it became necessary to analyze and compare genome sequences from endosymbionts in different stages of their symbiotic integration. These comparative analyses allowed researchers in the field to define a plausible scenario for the process of symbiotic integration, from a free-living bacterium to an obligate mutualistic lifestyle (Moya et al. 2009) (Fig. 1). The first step towards the establishment of an obligate intracellular mutualistic symbiosis takes place when a free-living bacterium infects an eukaryotic host. From this point, both organisms will co-evolve to adapt to the new situation. The host develops specialized cells to harbor the bacterium, which in turn provides benefits to the host that end up being essential. From an evolutionary point of view, this new stable situation triggers a cascade of changes that model the shape and content of the bacterial genome. In the course of this chapter, we will see how genomics and metagenomics studies helped researchers on the field to detangle the physiological and evolutionary changes that bacteria experience in their way towards an obligatory mutualistic intracellular symbiosis with eukaryotic hosts.

Fig. 1
figure 1

Genetic and population factors involved in the genome reduction syndrome experienced by mutualistic endosymbionts. At the beginning of endosymbiosis, the new rich, protected and stable intracellular niche provided by the host makes superfluous some gene functions, that become redundant (since they can be contributed by the host) or unnecessary in a stable and protected environment, but forces the preservation of genes required for the maintenance and viability of the partnership. The decreased efficiency of the purifying selection causes a fast accumulation of slightly deleterious mutations on non-essential genes, increasing the rates of genomic evolution. In addition, the drastic reduction of the bacterial effective population size between successive insect generations increases the relative influence of random genetic drift. Furthermore, the obligatory intracellular life-style prevents the entrance of genetic material by horizontal gene transfer, making the losses irreversible

Full size image

2 Survival, Replication and Transmission, the Three Biological Processes Involved In the Establishment of a Permanent Symbiotic Association

Mutualism and parasitism are two sides of the same coin. At the very beginning, it is not possible to determine if the relationship that would be established will be parasitic or mutualistic, since this distinction is based on the effect of the bacterium in the eukaryotic host but, from the bacterial point of view, the biological processes needed to successfully infect hosts are largely the same for both types of microorganisms (Gil et al. 2004a). In both cases, it will be necessary to overcome the physical, cellular, and molecular barriers presented by the host, to achieve internalization, survival, and proficient replication of the prokaryote inside the eukaryotic host cell. No matter if the interaction is harmful, neutral, or beneficial to the host, natural selection will favor the bacteria that achieve this goal (Ochman and Moran 2001). Most evolutionary transitions leading towards symbiotic lifestyles involve gene loss and horizontal gene transfer (HGT) of virulence genes within bacterial lineages. Genomic analyses indicate that, in many cases, the same molecular factors are involved both in pathogenic and mutualistic relationships although, in the case of mutualism, traditionally considered parasitic traits, at some point, became beneficial for both partners. In facultative symbionts, toxins that are known or suspected to target eukaryotic cells are involved in protecting the host against natural enemies (Oliver et al. 2009). Such toxins are encoded by genes present in lysogenic bacteriophages that participate in mutualistic functions but also act as hot spots for non-homologous recombination events that allow gene exchange of virulence cassettes among heritable symbionts (Degnan and Moran 2008). But even endosymbiotic bacteria with a long-time established relationship with their hosts, which have suffered a dramatic genome size reduction (as will be discussed below), maintain genes that encode essential endosymbiotic factors that are proposed to be virulence associated in bacterial pathogens, such as type III secretion systems and urease (Gil et al. 2003; Goebel and Gross 2001; Shigenobu et al. 2000). In many free-living bacteria, genes encoding the type III secretion system are located within pathogenicity islands that have been acquired by HGT. This system is present in many insect endosymbiotic bacteria where it has been proposed to be essential to invade the host cells, thus playing an essential role in the establishment of the symbiosis (Dale et al. 2001, 2002).

The establishment of a permanent intracellular association necessarily implies the development of efficient mechanisms for bacterial survival and replication inside the host cell. The bacteria must adapt their replication, so that their growth rates are coordinated with the development of their hosts in a way that depends on their location inside the host cell. In Buchnera aphidicola, which lives confined in vacuole-like organelles inside the aphid bacteriocytes, there is a tight coupling of bacterial cell number and aphid growth, with the bacteria showing a doubling time of approximately 2 days, much longer than the maximum exhibited by many free-living bacteria (Baumann and Baumann 1994). Blochmannia floridanus and Wigglesworthia glossinidia, which live free in the cytosol of bacteriocytes of their hosts (carpenter ants and tsetse flies, repectively), lack dnaA, the gene that encodes the essential DNA replication initiation protein in bacteria. Other alternative mechanisms reported so far for DNA replication initiation are also absent in B. floridanus. It has been suggested that this could imply the existence of a more direct control of DNA replication of the symbionts by the host (Gil et al. 2003).

An efficient transmission of the bacteria to the offspring must also be guaranteed. The acquisition of mechanisms ensuring maternal transmission to the host progeny allows the association to be heritable, resulting in the emergence of a new composite organism host-endosymbiont. The fine-tuning of this process detected in long-established obligate mutualistic symbioses suggests a long history of selection favoring host adaptations that help to maintain the association (Moran and Telang 1998).

3 Early Stages In the Symbiotic Relationship

The genomic era has allowed the sequencing of whole genomes of many bacteria living in symbiosis with eukaryotic hosts, allowing the comparison among the different evolutionary innovations carried out by these bacteria on their way from free-living to varied stages of integration with their respective hosts. To detangle the changes involved in each stage, over the next paragraphs we will follow the path from facultative symbiosis to early obligate endosymbiosis, as it has been revealed by molecular studies and comparative genomics over the past years.

3.1 Facultative Symbionts

Many different types of facultative or S-symbionts have been described in arthropodes, and have been extensively studied in several lineages of aphid, psyllids, whiteflies, leafhoppers, tsetse flies, fruit flies and mosquitoes (Table 1). They can be maternally transmitted between host generations but, unlike P-endosymbionts, they can also be horizontally transferred among host individuals and species and, therefore they do not share long evolutionary histories with their hosts. S-symbionts do not reside exclusively in specialized cells and organs, and can also be found in gut tissues, glands or body fluids, and when a P-endosymbiont is also present, they can occupy cells surrounding the P-bacteriocytes, or even invade them. Phylogenetic studies indicate that facultative symbionts have established relatively recent associations with their hosts (Dale and Moran 2006). Thus, their genomes may resemble those in the early stages of a transition from a free-living lifestyle to an obligate mutualism.

Their uneven presence among species and individuals of the same species indicates that S-symbionts are not necessary for host survival, but their influence on host fitness is variable. A range of effects, from negative to beneficial, have been described. Some described S-symbionts have negative effects on growth and reproduction to the host or may establish neutral or parasitic associations. Heritable S-symbionts can spread among lineages by manipulating host reproduction to enhance matrilineal transmission through parthenogenesis, male killing and feminization of genetic males or cytoplasmic incompatibility. This is the case of Wolbachia infecting arthropods, where it undergoes transfer among host lineages (McGraw and O’Neill 2004). Remarkably, Wolbachia appears as a typical P-endosymbiont in filarial nematodes, where it is required for normal development. The complete genomes of four different Wolbachia strains are already available, allowing unraveling the molecular basis of their interaction with their respective hosts by comparative genomics. Three of them are reproductive parasites of arthropods, Wolbachia pipientis wMel strain, found in Drosophila melanogaster (Wu et al. 2004); wRi strain, from Drosophila simulans (Klasson et al. 2009), and wPip strain, from the mosquitoes of the Culex pipiens group (Klasson et al. 2008); the last one, belongs to the wBm strain, the obligate mutualist of the nematode Brugia malayi (Foster et al. 2005). When the genomes of the parasitic strains were compared, a high degree of rearrangements was observed, revealing the most highly recombining obligate intracellular bacterial community examined to date (Klasson et al. 2009). The presence of abundant copies of transposable elements and prophages, that provide numerous sites for homologous recombination, can explain that. Most of the genome size differences are due to the presence of repeated elements, especially to the amplification of the WO prophage. Furthermore, the WO elements can experience intragenic recombination (Bordenstein and Wernegreen 2004). They present a conserved core of structural genes plus a variable fraction of genes that encode for ankyrin repeats, which correlate with the effects of the bacterial strain as reproductive parasite.

The first completely sequenced genome of a S-symbiont with no clear negative or positive effect corresponded to Sodalis glossinidius (Toh et al. 2006), the S-symbiot of the tsetse fly. It has been proposed to play a role in the acquisition of trypanosome infections (Welburn and Maudlin 1999). Its genome size (4.2 Mb) is close to that of free-living bacteria, but its coding capacity is highly diminished by the presence of a big amount of pseudogenes, only similar to what has been observed in some parasites such as Mycobacterium leprae (Cole et al. 2001; Gomez-Valero et al. 2007). The genome also contains certain amounts of repetitive and mobile DNA, such as transposable elements and bacteriophages, which could promote recombination. Therefore, it appears that this bacterium is at the early stages in the reductive process affecting symbiont genomes. S. glossinidius coexists in the gut lumen of tsetse flies with the P-endosymbiont, W. glossinidia, but occupying different portions of the insect gut, and it can be found both intra- and extracellularly (Toh et al. 2006). Moreover, it can be cultured in vitro (Dale and Maudlin 1999), an indication that the association with its host is not yet irreversible.

Many other S-symbionts described in aphids confer beneficial effects on the survival and reproduction rates of their hosts. They can rescue the host from heat damage (Chen et al. 2000; Montllor et al. 2002), provide resistance to natural enemies (Ferrari et al. 2004; Guay et al. 2009; Oliver et al. 2005; Scarborough et al. 2005) or stress (Russell and Moran 2006), are involved in host plant specialization and reproduction (Simon et al. 2003; Ferrari et al. 2004; Tsuchida et al. 2004), and even compensate the loss of the essential endosymbiont, as it was experimentally proven (Koga et al. 2003). Recently, the genome of one strain of Candidatus Hamiltonella defensa (from now on H. defensa), S-symbiont of the pea aphid Acyrtosiphon pisum, also became available (Degnan et al. 2009b). H. defensa can be found in aphids and other sap-feeding insects, where it has been proposed to play a beneficial role by protecting its host from attack by parasitoid wasps. Genes that encode for toxins, effector proteins, and two type-III secretion systems have been identified in the sequenced genome and seem to be involved in this function. The 2.1-Mb sequenced genome has undergone significant reduction in size relative to its closest free-living relatives, and important gene losses have been detected (it relies on the the P-endosymbiont B. aphidicola for the synthesis of 8 of the 10 essential amino acids), which indicates that the reductive process affecting endosymbiont genomes is already advanced. Nevertheless, the genome contains considerable amounts of genes devoted to regulatory functions involved in regulation of virulence factors and quorum-sensing genes, which indicates that it still retains at least a partial ability to deal with changing environments and invasion of new host species. This genome also contains important amounts of repetitive DNA (21% of the genome), including insertion sequences, group II introns, prophages and plasmids. APSE, a lysogenic phage that infects many H. defensa populations, has been involved in the protective role of this bacterium against parasitic wasps, since the different variants of APSE identified all encode toxins that target eukaryotic tissues (Oliver et al. 2009). Therefore, the beneficial role of the phage toxins for the insect host fitness is contributing to the spread and maintenance of H. defensa in host populations. This is another evidence of the direct implication of virulence factors on the basis of a mutualistic symbiosis. Furthermore, the APSE lysis region is a hot spot for non-homologous recombination of novel virulence cassettes, allowing gene exchange among S-symbionts by horizontal transmission (Degnan and Moran 2008).

Candidatus Regiella insecticola (from now, R. insecticola) is another common facultative symbiont in aphids. Similar to H. defensa, it is not only involved in resistance to parasitoid wasps but also to fungal pathogens (Scarborough et al. 2005). Most of its genome (about 2.07 Mb) has been sequenced and compared with the close relative H. defensa (Degnan et al. 2009a). The complete genome assembly was not performed, because it was hampered by the presence of high amounts of repetitive DNA, mostly insertions sequence (IS) elements, representing up to 14% of the genes and pseudogenes. Similar to what has been found in the parasitic W. pipientis strains, the genomic architecture of these two genomes is highly divergent, as a consequence of recombination and gene inactivation facilitated by the presence of mobile DNA. In contrast, core genes reveal clonal evolution in H. defensa and R. insecticola, and the nucleotide divergence in this case is similar to what has been found in obligate mutualists. No intact prophages have been found in the already sequenced part of the genome of R. insecticola.

The genomes of two Serratia symbiotica strains, from the cedar and tuja aphids, are also being sequenced. Although some strains of S. symbiotica appear as typical facultative symbionts, this is not the case of the SCc strain, which has become essential for its host, the cedar aphid Cinara cedri (Gosalbes et al. 2008). Preliminary results of its genome project indicate that it has established a permanent and stable cooperative consortium with the host and the P-endosymbiont, B. aphidicola BCc, thus becoming essential for the maintenance of the fitness of all three partners (see Sect. 6).

3.2 Insertion Sequences, Shaping the First Steps Towards an Obligate Endosymbiosis

It has been postulated that soon after the establishment of obligate symbiosis, a massive gene loss must occur, probably by means of large deletion events that cause the elimination of series of contiguous genes (Moran and Mira 2001). Later on, as it has been shown by comparative genomics, genome shrinkage proceeds through a process of gradual pseudogenization and gene loss scattered throughout the genome (Gomez-Valero et al. 2004; Silva et al. 2001). However, the mechanism involved in the large deletion events was unknown at that time. The identification of the genome changes that occur in these initial stages of the adaptation towards endosymbiosis requires the genome analysis of clades of bacteria that have recently established such associations. For this purpose, our group selected SOPE, the P-endosymbiont of the rice weevil Sitophilus oryzae. With an estimated 3.0-Mb genome (Charles et al. 1997), within the range of many free-living bacteria, this γ-proteobacterium maintains a typical obligate mutualistic endosymbiosis with its host. The bacteria live inside bacteriocytes organized in an organ called bacteriome surrounding the midgut of the insect and near the female ovaries. The bacterium cannot be cultured outside the host, and it provides at least amino acids and vitamins to the insect, which has recognizable effects on fertility, development and the flying ability of adult insects (Heddi et al. 1999). SOPE is closely related with S. glossinidius (Dale and Welburn 2001; Heddi et al. 1998), which is still able to grow in laboratory culture conditions. Although SOPE and S. glossinidius are respectively P- and S-symbionts of hosts belonging to different insect orders (Coleoptera and Diptera), which feed on very different diets (storage grain and blood, respectively), their close phylogenetical position indicates a relative recent divergence. Therefore, the analysis of the similarities and differences between these two genomes will help to achieve a better understanding of the differences between the primary and secondary forms of endosymbiosis and what molecular events are implied in the establishment of an obligatory endosymbiosis.

The association of insects of the genus Sitophilus and their present endosymbionts is not antique. Some data indicate a recent endosymbiont replacement of an ancestral endosymbiont in the family Dryophtoridae to which the rice and maize weevils belong (Lefevre et al. 2004). During the first stages of the SOPE genome sequencing project (in progress), big amounts of repetitive DNA, mainly IS elements, were identified (Gil et al. 2008). It has been estimated that IS elements occupy about one third of its genome, and a similar situation has been observed in its close relative SZPE, the P-endosymbiont of the maize weevil (Plague et al. 2008). This impressive amount of repetitive DNA was not expected in an obligate mutualistic endosymbiont. Repetitive DNA is common in free-living bacteria, and its presence increases in bacteria that have recently evolved as specialized pathogens (e.g., the enteric bacteria Shigella and Salmonella enterica Typhi) (Jin et al. 2002; Wei et al. 2003), intracellular parasites (e.g. W. pipientis strains, reproductive parasite of arthropods) (Klasson et al. 2008, 2009; Wu et al. 2004), or facultative insect symbionts (e.g. H. defensa, Candidatus Arsenophonus arthropodicus and R. insecticola) (Dale and Moran 2006; Degnan et al. 2009a,b). Thus, the increase in transposable elements is a common trait among bacteria that have recently established mutualistic relationships with their hosts, and must have subsequent effects on the outcome of the symbiotic process (Bordenstein and Reznikoff 2005; Moran and Plague 2004). However, it was assumed that after the establishment of an obligate endosymbiont lifestyle, repetitive DNA tends to diminish until its total disappearance. Several observations support this conclusion, from total absence of phages or transposable elements in bacterial endosymbionts with a long-established obligatory relationship with their hosts, to the identification of only 5.4% of repetitive DNA, mostly composed of inactivated IS, in the mutualistic W. pipientis wBm (Foster et al. 2005).

The IS are the most abundant and simplest transposable elements in nature (Touchon and Rocha 2007). Habitually they only include the elements needed in its own mobilization: short terminal inverted repetitive sequences (IR) define the ends of the IS and flank the ORF(s) that encodes the transposase activity that mediates the transposition events after the recognition and processing of the IR sequences. The IS are able to move between replicons of a certain genome and can also be transferred between genomes of different organisms by horizontal gene transfer. Its persistence is usually explained by an intense ability for intergenomic mobilization and to its more or less efficient infecting capacity. Four IS types have been identified in SOPE (Gil et al. 2008). At least two of them (ISsope1 and ISsope2) are present in large copy numbers in SZPE (Plague et al. 2008), and ISsope1 has also been identified in S. glossinidius, but representing just 2.5% of the total genome (Toh et al. 2006), is an indication that this element must have been present in a common ancestor of these bacteria.

The massive presence of IS must be related with some of the syndromes that appear at the beginning of the intracellular life (Fig. 2). IS elements are widespread in free-living bacteria, but their transposition is tightly controlled, so that only a few copies of a limited number of categories appear in each genome. The dramatic increase of these elements in intracellular bacteria must reflect an enhanced replicative transposition of elements that were already present at the onset of symbiosis, and can then act as a source of gene inactivation and chromosomal rearrangements. After the establishment of the symbiosis, the decrease in the selective pressure caused by functional conditions and population dynamics in the new environment, can favor the uncontrolled proliferation of such elements, which could be involved in the inactivation of non-essential genes. The high abundance of very similar (or even identical) repetitive elements in direct orientation can then serve as a substrate for unequal recombination, which would lead to a loss of the region between two elements, thus promoting genome size reduction in early stages. Additionally, the presence of these elements in opposite orientation, will lead to genome rearrangements. Comparative genomics analyses between several B. aphidicola strains form different aphids, B. floridanus, and close free-living relatives indicate that the massive gene loss that took place in the process towards the last common symbiotic ancestor (LCSA) of both species was accompanied by many chromosomal rearrangements. The former presence of repetitive elements, already disappeared in the present genomes, might explain such genome reorganizations, while the current lack of repetitive sequences, with a great potential as recombination sites, as well as the loss of loci needed for the catalysis of such recombination events in later stages of the symbiosis (see next section), appears to be in the origin of the high genomic-architecture stability levels in old endosymbionts, quite unusual among the prokaryotes (Silva et al. 2003). This is an indication that most of genomic modeling, including chromosomal rearrangements and the loss of many functionally dispensable genes, must take place at an early stage of the process of genomic adaptation to intracellular life (Dougherty and Plague 2008; Touchon and Rocha 2007). Genes needed for DNA repair and recombination are also among the first losses detected, thus contributing to genomic stasis in further steps in the endosymbiotic evolutionary path. The loss of the genes coding for the enzymes RecA and RecF in SOPE, SZPE, and S. glossinidius (Dale et al. 2003) supports this idea.

Fig. 2
figure 2

An evolutionary scenario for the implication of IS in gene inactivation, genome reduction and chromosomal rearrangements. (a) Free living cell. (b) Beginning endosymbiosis: Many genes become superfluous or redundant. Massive transposition. (c) IS can be a source of genomic recombination. (d) Interrupted genes and IS degenerate by mutation

Full size image

4 Long-Established P-Endosymbioses

Most bacterial insect P-endosymbionts that have been analyzed belong to the γ-proteobacteria (Table 1). However, more recently, the genomes of several endosymbionts belonging to other groups of proteobacteria and to the phylum Bacteroidetes have also been analyzed (McCutcheon and Moran 2007; Lopez-Sanchez et al. 2008; Tokuda et al. 2008; Sabree et al. 2009), revealing convergent evolution among endosymbionts belonging to different phyla (López-Sánchez et al. 2009). In general, endosymbionts with a long-established relationship with their hosts have genomes eight to ten times smaller than those of their free-living relatives. In bacteria, whose genomes are highly compact, gene content correlates quite well with genome size (Casjens 1998). Therefore, the reduced size of endosymbiont genomes reflects the presence of a smaller number of genes than those of free-living bacteria. Several additional characteristic genome features have traditionally been associated with the degenerative syndrome affecting endosymbiotic bacteria. These include almost total absence of recombination, increased rate of nucleotide substitution, high A + T content (although as it will be discussed later, this no longer can be considered a general trait), accumulation of deleterious mutations by random genetic drift, loss of codon bias towards A or T, and accelerated sequence evolution (Andersson and Kurland 1998; Clark et al. 1999; Moya et al. 2002; Wernegreen 2005). Most of these characteristics are linked with the above mentioned informational and demographic factors affecting bacteria that live in close association with eukaryotic cells, although the accommodation to symbiotic life varies according to the age of the association, the host lifestyle, and the way of living within the host.

The analysis of gene order in the first completely sequenced endosymbiont genomes lead to interesting observations regarding the evolution of these genomes. The availability of complete genome sequences from four different strains of B. aphidicola clonally evolving in their aphid hosts revealed that, after a short period of large genome rearrangements at the beginning of the symbiotic process, there were large periods of evolutionary stasis. All these strains present a nearly perfect gene-order conservation (Perez-Brocal et al. 2006; Shigenobu et al. 2000; Tamas et al. 2002; van Ham et al. 2003), which suggests that B. aphidicola can be considered as a “gene-order fossil”, and that the onset of genomic stasis coincided with the establishment of the obligate symbiosis with aphids, 80–150 MY ago (von Dohlen and Moran 2000). As mentioned in the previous section, this astonishing genome stasis can be explained by the total absence of repetitive DNA in these genomes, as well as the loss of genes involved in DNA repair and recombination in early stages of the symbiotic integration. Repetitive DNA is quite abundant at the beginning of the obligate endosymbiosis, but these elements tend to disappear in the later stages of the relationship and are absolutely absent in endosymbionts that share long evolutionary histories with their hosts (Fig. 2). The progressive loss of transposable elements might have been favored by the energetic benefit of decreasing transposase activity and avoiding the increase in genome size derived from the proliferation of these elements or by the need to control the mutagenizing effect of its mobilization. It is presumable that, at some point, IS elements expansion will be deleterious and these elements would be also affected by the process of genome degradation that these genomes suffer. The sexual isolation of P-endosymbionts and the loss of recombination genes must also have participated in the process, since horizontal gene transfer is the way of entrance of these elements in prokaryotic genomes (Touchon and Rocha 2007). The reduced genomes of endosymbiotic bacteria and some pathogens have lost most (if not all) genes involved in recombination processes and, consequently, the genome size cannot be increased by acquisition of foreign DNA (Silva et al. 2003). Nevertheless, some recombination events can still take place in these reduced genomes, probably involving the RecBCD system, which in the absence of RecA might serve as a general exonuclease repair enzyme (Sabater-Munoz et al. 2004), as revealed by the great plasticity of the plasmids involved in the biosynthesis of leucine in different lineages of B. aphidicola, showing that several events of insertion from a plasmid to the main chromosome have occurred since the divergence of these strains (Latorre et al. 2005).

In general, smaller genomes correlate with longer obligate associations. The differences in host lifestyle are also introducing changes in this degenerative process among strains of the same endosymbiont species. The small genomes of B. aphidicola are still suffering this reductive process, as evidenced by the fact that B. aphidicola strains from several aphid subfamilies showed differences up to 200 Kb (Gil et al. 2002), and the presence of pseudogenes in the B. aphidicola genomes that have been sequenced (Perez-Brocal et al. 2006; Shigenobu et al. 2000; Tamas et al. 2002; van Ham et al. 2003). In addition, the degenerative process is randomly affecting different genes in each genome, conditioning the essentiality of the rest of the genes that are present in these reduced genomes. Therefore, although we can hypothesize that the LCSA of B. aphidicola suffered a drastic genome reduction at the beginning of the symbiotic integration, since then, the different strains of the bacteria have undergone a reductive process in a way that correlates with their hosts.

In addition to changes in genome size, obligate and facultative endosymbionts of different insect hosts also differ in nucleotide composition. P-endosymbionts with an old association with their hosts have in general small genomes, and an A + T content higher than 70%, while P-endosymbionts with a younger association and S-symbionts have a genome size and an A + T percentage intermediate with respect to older P-endosymbionts and free-living relatives (Dale and Maudlin 1999; Heddi et al. 1998; McCutcheon and Moran 2007; Moya et al. 2002; Nakabachi et al. 2006). The loss of the bias in codon usage in these obligate intracellular bacteria, highly mitigated in P-endosymbionts with larger genomes and in S-symbionts and almost absent in B. aphidicola, is considered to be a consequence of this base composition bias (Moya et al. 2002; Rispe et al. 2004). This notable enrichment in A + T has been related to the loss of DNA repair enzymes, since the most common chemical changes in DNA (cytosine deamination and guanosine oxidation) led to changes in GC pairs leading to AT. However, several cases that do not follow this nucleotide composition rule have been described. The partial genome sequences available from Candidatus Tremblaya princeps, the P-endosymbiont of the mealybug Planococcus citri, indicated that this genome has a 57% G + C content, much higher than expected for an endosymbiont (Baumann et al. 2002). Recently, a remarkable small genome with a high G + C content has also been reported (McCutcheon et al. 2009) (see next section). Candidatus Hogkinia cicadicola (from now on H. cicadicola), P-endosymbiont of the cicada Dieroprocta semicincta, presents a 144-Kb genome with a 58.4% G + C content. Therefore, it has been proposed that, while gene loss associated with genome reduction is a critical step in endosymbiont genome evolution, mutational pressure favoring A + T is not.

There is only one case of advanced symbiosis described in archaea: Nanoarchaeum equitans, a tiny coccus living attached to the outside of the cells of its host, the Crenarcheote Ignicoccus hospitalis. The study of this association, including the sequencing of the genome of both species (Waters et al. 2003; Podar et al. 2008), shows a highly specialized relationship, which so far cannot be assigned to any classical symbiosis type (mutualism, commensalism or parasitism). With a highly reduced genome (491 Kb), N. equitans was initially suggested to be a representative of a novel phylum within the domain archaea, Nanoarchaoeta. However, further genomic analyses indicate that it is likely to be a highly derived Euryarchaeon, possibly related to the Thermococcales that has evolved through a unique pathway of genome degradation (Brochier et al. 2005; Makarova and Koonin 2005). Features such as the extreme N. equitans genome reduction, bias in codon usage, and evolutionary acceleration, are common to those observed in bacterial endosymbionts, probably an indication of the generality of the reductive mechanisms among prokaryotes. Interestingly, this reductive process has affected simultaneously both genomes, since the I. hospitalis genome is only 1.3 Mb in length, one of the smallest among free-living organisms. Further analyses will be necessary to understand the implication of this dual reductive genome process (Forterre et al. 2009).

5 Final Stages in Endosymbiotic Relationships

As the endosymbiotic integration progresses, genes that are rendered unnecessary experience a random process of gradual pseudogenization and gene loss scattered throughout the genome (Gomez-Valero et al. 2004; Silva et al. 2001). The final step of this minimization process might, in theory, lead to the loss of all genes except those that are essential for keeping the host-bacterial interaction reproducing. Therefore, even the most reduced genome must retain those genes involved in the symbiotic relationship, as well as a reduced repertoire of genes necessary to maintain the three essential functions that define a living cell: maintenance, reproduction and evolution (Luisi et al. 2002). One of the most comprehensive efforts to define the minimal core of essential genes was that presented by Gil et al. (2004b). This study can be a good starting point to identify essential genes involved in informational processes that must be present in any living cell, while the essential genes devoted to the symbiotic association can be deduced by the knowledge of the host needs for survival and reproduction. However, most extremely reduced genomes that have been described have lost part of such essential functions. In most cases, as it will be discussed in the next section, genome degradation can proceed over the expected limit because of the implication of a second endosymbiont on the relationship. But there is an intriguing case: Candidatus Carsonella ruddii (from now on, C. ruddii), considered the P-endosymbiont of the psylid Pachpsylla venusta. Although a second bacterial symbiont has not been found in the psyllid, C. ruddii does not fulfil the conditions to be considered as a mutualistic endosymbiont, not even as a living organism. Its genome consists of a circular chromosome of 160 Kb, averaging 83.5% A + T content (Nakabachi et al. 2006). It also presents a high coding density (97%), and 182 described open reading frames, many of which overlap and present a reduced gene length. A detailed analysis of the coding capacities of C. ruddii, revealed that the extensive degradation of the genome is affecting vital and symbiotic functions (Tamames et al. 2007). Most genes for DNA replication, transcription and translation are completely absent, and gene shortening causes, in some cases, the loss of essential domains and functional residues needed to fulfil these and other vital functions. In addition to the essential functions that define life, as a mutualistic endosymbiont, C. ruddii should provide its host all essential complements to its nutritionally deficient diet, limited to phloem sap, rich in sugars but relatively poor in nitrogenated compounds, especially essential amino acids. However, the genomic analysis revealed that the pathways for the synthesis of three essential amino acids (i.e. histidine, phenylalanine and tryptophan) are lost. Since this strain of C. ruddii is not able to sustain the requirements of its host, neither to sustain its own vital functions, it can be viewed as a further step towards the degeneration of the former P-endosymbiont, and its transformation in a subcellular new entity between living cells and organelles, which might be taking advantage of mitochondrial functions encoded by the nucleus, especially for basic informational processes needed for maintenance and multiplication. It might even be possible that some C. ruddii genes have been transferred to the host nuclear DNA, as it has been proved for present organelles. If confirmed, this would be the first example of such a scenario in animal cells.

6 Replacement or Complementation, and the Establishment of Microbial Consortia

Eventually, after the establishment of a permanent symbiotic association between a bacterium and an animal host, a second bacterial species can join the association. Although initially this new association can be facultative (as seen in Sect. 3.1), if the second bacterium provides benefits to the organization, with time, it can become essential for host fitness. The involvement of two bacteria in the fitness of an insect host adds one extra element to the evolutionary scenario that explains the reductive evolution of endosymbiont genomes, but there is no need to invoke any supplementary reductive factor in addition to the informational and population dynamis factors already indicated. Subsequently, all three components of the association will co-evolve, and the evolutionary process of genome shrinkage will now affect both bacteria. New genes will become unnecessary due to redundancy, but which one of the two bacterial genomes loses them will be a matter of chance. Depending on which genome is affected by the loss of genes needed for the synthesis of essential molecules, either both bacteria will become indispensable to keep a healthy consortium (complementation) or one of them can enter an extreme degenerative process, which may end with its extinction (replacement), and the retained bacteria will continue the degenerative process alone (Moya et al. 2009). Replacement has already been reported, for example, in the Family Dryophthoridae, where a former endosymbiont Candidatus Nardonella was replaced by the ancestor of the Sitophilus P-endosymbionts (Lefevre et al. 2004). However, there are many more described cases in which both bacteria loose part of the gene complement necessary for their host fitness, so that both of them become indispensable and a stable consortium is established. Several symbiotic consortia have already been reported and sequenced, using metagenomics approaches, and many more will surely be available in the near future thanks to the use of new massive sequencing technologies.

One of the first described consortia involves strains of B. aphidicola and S. symbiotica living inside the cedar aphid. S. symbiotica appears as a facultative symbiont in many aphid species. However, it was always found in cedar aphids, coexisting with B. aphidicola BCc in the insect bacteriome, so that S. symbiotica strain SCc cannot be considered as a facultative symbiont. Comparative, functional and evolutionary genomic analysis, plus microscopic observations, led Perez-Brocal et al. (2006) to conclude that S. symbiotica SCc might be replacing B. aphidicola BCc. Contrary to other sequenced B. aphidicola strains, BCc has partially lost its symbiotic role, as it cannot synthesize tryptophan. Genes involved in the biosynthesis of this essential molecule were found in the genome of S. symbiotica SCc (Gosalbes et al. 2008), but included an additional surprise: the pathway to synthesize tryptophan is distributed between both genomes: B. aphidicola BCc produces a metabolic intermediate that is then provided to S. symbiotica SCc to synthesize the final product. Therefore, coexistence of both bacteria is needed to keep a healthy consortium due to metabolic complementation, and both of them keep an intracellular obligatory mutualistic association with their host.

The establishment of an endosymbiotic bacterium consortium can be in the origin of big evolutionary changes in host lifestyle. This is the case of the consortium formed by Baumannia cicadellinicola and Sulcia muelleri, co-resident P-endosymbionts of the xylem-feeding sharpshooter Homalodisca vitripennis. Their whole genome analysis revealed that they have complementary sets of biosynthetic capabilities needed to provide to their host the nutrients that are lacking in the xylem sap (Wu et al. 2006). While B. cicadellinicola contains a large number of pathways for biosynthesis of vitamins, S. muelleri encodes the enzymes involved in the biosynthesis of most essential amino acids. Phylogenetic studies indicate that S. muelleri was ancestrally present in a host lineage that acquired B. cicadellinicola at the same approximate time that the host ancestor switched to a xylem-feeding lifestyle, consistent with the view that Baumannia’s nutrient-provisioning capabilities were a requirement for the acquisition of this new feeding behavior.

A newly described consortium also involving S. muelleri is on the basis of the dramatic genome reduction experienced by H. cicadicola, the P-endosymbiont of the cicada D. semicincta (McCutcheon et al. 2009). H. cicadicola is an α-proteobacteria with the smallest described genome to date (144 Kb), an unusually high G + C content (58.4%), and a coding reassignement of UGA stop codon to Trp. It has been found in other cidadas, thus suggesting that this symbiont infected an ancestor of the cidadas and, since then, has been maternally transmitted.

Our group is also involved in the metagenomic study of another exceptional symbiotic consortium: the one established among the mealybug P. citri and their two endosymbiotic bacteria: the P-endosymbiont T. princeps, a β-proteobacterium (Thao et al. 2002), which contains inside a γ-proteobacterium (von Dohlen et al. 2001), considered as an S-symbiont based on its polyphyletic origin (Thao et al. 2002). This is the first described case of a double-endosymbiosis, although the symbiotic relationship between the two bacteria (parasitic, commensal or mutualistic) has not been elucidated (Kono et al. 2008). As it has been mentioned in Sect. 4, T. princeps was the first endosimbiotic genome in which it was detected that the A + T bias rule does not apply (Baumann et al. 2002).

Some other consortia can involve more than two microorganisms. The marine oligochaete Olavius algarvensis, which lacks a digestive and excretory system, harbors four co-occurring symbionts essential for host survival (Woyke et al. 2006). The symbionts are located just below the worm cuticle, and they are essential to manage energy and waste needs of the host. The symbionts, γ1- and γ3-(sulphur-oxidizing chemolithoautotrophs), and δ1- and δ4-(sulphate reducer) proteobacteria, are engaged in an endosymbiotic sulphur cycle, fix CO2, provide nutrients to the host, and are also involved in host waste recycling. They can heterotrophycally feed the host by taking up dissolved organic carbon compounds from the environment, and can synthesize almost all amino acids and several vitamins. The host probably takes these nutrients by digesting the bacteria (Fiala-Médioni et al. 1994). This is another case in which, contrary to most cases of obligate host-associated bacteria, the available genomic sequences do not show A + T bias.

7 Concluding Remarks

Symbiosis between prokaryotes and eukaryotes is an expanding field, thanks to the advent of the metagenomics and high-throughput sequencing technologies. Systems biology approaches are also allowing the exploration of metabolic interdependences among the members of the symbiotic consortium. Now that endosymbiont genomes are accumulating, comparative analyses allow making predictions on the evolutionary paths followed by endosymbiotic bacteria in their adaptation to the intracellular environment provided by the host. Now, more clearly than ever before, the association and functional interaction of genomes from different species observed during symbiosis can be viewed as a power, like mutation, recombination and other genome rearrangements, able to generate genetic variation, acting as a fuel for evolution. The action of forces such as natural selection and/or random drift will be the responsible of transforming this variation in evolutionary novelties. However, as the number of the available genomes increases, new features are appearing and open new questions that need to be experimentally solved. We do not know what drives symbiotic associations to mutualism versus parasitism, since both types of associations derive from common mechanisms for symbiont-host interaction. We cannot anticipate when a facultative association will become essential for host fitness and, when two or more prokariotes are involved, we cannot determine which forces will lead it towards the establishment of a consortium or, alternatively, will end up in a replacement. More recently, an additional question was opened about the nucleotide content bias, most of the times towards an increase of A + T, but also possible towards an increase in G + C content… For sure, we will learn a lot more about the molecular mechanisms and evolutionary forces acting on these systems once eukaryotic host genomes become available.