The term free-living amoebae comprises more than 15,000 species (Adl et al., 2007) forming a heterogeneous group of phylogenetically distantly related protists that are widespread in water and soil ecosystems and display similar ecological characters. These unicellular eukaryotes were recently classified into two main suprakingdom-level groups; (i) the Excavata notably comprising the Andalucia, Jakoba (Jakobids), Naegleria, Sawyeria, Vahlkampfia (Heterolobosea) as well as the parasites Trypanosoma, Leishmania (Euglenozoa) and (ii) the Amoebozoa comprising among others Acanthamoeba, Hartmannella, Vannella, Dictyostelium and the medically important parasite Entamoeba (Hampl et al., 2009; Pawlowski and Burki, 2009).

Most amoebae live under the form of a trophozoite that replicates by binary fission, but in unfavorable conditions they can differentiate into a dormant form, the cyst. This latter is resistant to harsh conditions such as high temperature, desiccation, pH, and saline stress as well as disinfection processes (Thomas et al., 2004). Phagocytic amoebae graze on various microorganisms free-living or established in biofilms, including algae, bacteria, yeasts, and viruses (Rodriguez-Zaragoza, 1994). Microorganisms are phagocytosed and normally follow the endocytic pathway to be degraded in acidic phagolysosomes by a number of hydrolases (Greub and Raoult, 2004). However, several giant viruses and bacteria have evolved strategies to escape degradation, hence their naming as ARMs. They live symbiotically within their host or replicate in vacuoles before lysing the amoeba.

Taking profit from these characteristics, Rowbotham (1983) used cultures of amoebal cells to grow Legionella species. Since then, amoebal co-culture has become a method of choice to retrieve new microorganisms able to resist and grow in these professional phagocytes. This method uses amoebae as a cell background to inoculate environmental or medical samples, in order to retrieve ARMs (Lienard and Greub, 2011). However, the almost uniform use of Acanthamoeba castellanii (Thomas et al., 2006; Corsaro et al., 2009) and A. polyphaga (Greub et al., 2004b; Pagnier et al., 2008) largely biases and underestimates the diversity of known ARMs. Diversifying the species of amoeba used in co-culture experiments is required to improve our understanding of the pool of amoebal symbionts and parasites.

Concisely, amoebae can act as a replicative niche and a reservoir of ARMs that are established in water and soil environments (Greub and Raoult, 2004). As shown in Figure 1, amoebae may hide several ARMs in their cytoplasm or more commonly in phagocytic vacuoles. The cyst may function as an armor to protect internalized microorganism from difficult external conditions as well as disinfection procedures. Moreover, the development of strategies to resist microbicidal effectors by ARMs may help selecting virulence traits enabling to survive in the macrophages, the first line of human defense, as it is the case for Mycobacteria, Legionella, Parachlamydia (Greub, 2009; Lamoth and Greub, 2009; Salah et al., 2009), and Mimivirus (Ghigo et al., 2008). More importantly, amoebae were lately suggested as a place that favor genetic exchanges by bringing in close vicinity ARMs (Moliner et al., 2010; Thomas and Greub, 2010; Merhej et al., 2011). The recent sequencing of some amoebal genomes such as Entamoeba histolytica, E. dispar, Dictyostelium discoideum, and A. castellanii provide the opportunity to highlight the first hints on the genetic exchanges between the amoebae and their intracellular microbes.

The search for amoeba-resisting viruses (ARVs), i.e., viruses able to replicate alone or in combination with others within amoeba, started with the discovery and the sequencing of Mimivirus in the early 2000s (La Scola et al., 2003; Raoult et al., 2004), which raised an extraordinary interest. Within less than a decade, the known complexity of ARVs was boosted by (i) the discovery of the small virophage Sputnik (La Scola et al., 2008), (ii) the description of Marseillevirus and Lausannevirus, two large viruses encoding histone-like proteins (Boyer et al., 2009; Thomas et al., 2011), as well as (iii) the publication of Megavirus chilensis (Fischer et al., 2010; Arslan et al., 2011), the largest identified virus harboring a 1.26 Mb genome (Table 1). According to new systematic searches, ARVs seem to be fairly common in the environment (La Scola et al., 2010).

Most ARVs possess pseudo-icosahedral capsids hiding intricate large dsDNA genomes that pushed forward the recognized limits of viral genome size. Therefore, they were called (i) “giant viruses” (La Scola et al., 2003; Raoult et al., 2004), a concept previously used for large algal DNA viruses (Van Etten and Meints, 1999), or (ii) “giruses” (Legendre et al., 2012). Another interesting girus is Cafeteria roenbergensis virus (CroV) that infects a marine phagocytic flagellate belonging to the Chromalveolata (Fischer et al., 2010). CroV was not demonstrated to replicate in amoebae, but it is worth discussing as it shares many similarities with other ARVs. At the opposite, Sputnik is a 18 kb virophage able to replicate in A. castellanii only when Mimivirus, or its close relative Mamavirus, co-infects the amoeba (La Scola et al., 2008). Sequences homologous to Sputnik were detected in the environmental dataset of the Global Ocean Survey, suggesting that it may represents a new virus family, but its classification is currently unclear.

Mimivirus, Megavirus, CroV, Marseillevirus, and Lausannevirus belong to the monophyletic class of nucleo-cytoplasmic large DNA virus (NCLDV) (Iyer et al., 2006). They were classified into two main families of NCLDV, the Mimiviridae and the Marseilleviridae (Table 1). NCLDVs only share a core genome of 30–47 genes (Iyer et al., 2006; Yutin et al., 2009). Thus, core genes only represent a minor fraction of the gene repertoire, whereas ORFans, i.e., genes that do not have homologs in other organisms, and dispensable genes, i.e., genes present in two or more NCLDVs, are the major constituent of these viral genomes.

The reconstruction of deep phylogenetic relationship from viral sequences is controversial because of the rapid evolutionary rate of viruses and the presence of numerous horizontal transfers (Moreira and Brochier-Armanet, 2008), in particular from host genomes. Nevertheless, several evidences suggest that these giruses have evolved from a common ancestor. The two families of distant giruses, Marseilleviridae and Mimiviridae, harbor an unusual genomic repertoire that includes genes for protein translation, a hallmark of cellular organisms (Raoult et al., 2004; Boyer et al., 2009; Arslan et al., 2011). Moreover, the presence of tRNA synthetases in some viruses and their absence in others supports the idea that all Mimiviridae evolved by reductive evolution from a common ancestor, potentially a cellular ancestor: four tRNA synthetases homologs have been found in Mimivirus and Megavirus, an additional one in both CroV and Megavirus, and two additional ones are present in Megavirus only (Ghigo et al., 2008). A further example of lineage-specific deletion is given by the DNA photolyase: CroV possesses two intact copies, Mimivirus harbors fragmented ortholog remnant of one of them, and finally Megavirus encodes one intact ortholog and one ortholog split in two parts by a transposase (Ghigo et al., 2008).

The amazing diversity in genome size and gene repertoire among these phylogenetically related viruses questions the respective importance of both LGTs and vertical inheritance in evolution. The large differences observed in genome size (0.018–1.3 Mb) would imply either an extensive genome growth via LGTs or a divergent reductive evolution in the different phyla. An increased propensity to acquire genes of foreign origin surely accounts for such differences in genome size (Monier et al., 2007) and some authors even consider viruses as “bags of genes” (Hendrix et al., 2000; Moreira and Lopez-Garcia, 2005). However, it is questionable whereas LGTs are sufficient to explain such a large variation. Moreover, core genes seem to have originated from different kingdom, including eukaryotes, bacteria, and bacteriophages (Koonin and Yutin, 2010). These observations are in agreement with the scenario of a bacteriophagic origin of NCLDV (Koonin and Yutin, 2010). At this stage, a mix of genes from very different eukaryotic and bacterial organisms were acquired concurrently to the loss of phage genes except those essential for genome replication and virion formation (Koonin and Yutin, 2010). The following section attempts to provide an overview of the extent of genetic exchanges documented in ARVs.

A large variety of amoeba-resisting bacteria (ARBs) have been isolated using amoebal co-culture or directly retrieved from their host by amoebal enrichment (Greub et al., 2004b; Horn and Wagner, 2004; Lienard and Greub, 2011). In addition, many microorganisms have been shown to survive in vitro in amoebae such as different Burkholderia, Coxiella burnetii, a strain of E. coli, Francisella tularensis, Helicobacter pylori, Listeria monocytogenes, Porphyromonas gingivalis, and Vibrio cholerae (Greub and Raoult, 2004; Wagner et al., 2006). As shown in Figure 2, these encompass various clades scattered through the prokaryotic phylogeny, including members of the Actinobacteria, Bacteroidetes, Chlamydiales, Firmicutes, and different subdivisions of Proteobacteria (α, β, γ, ε). Although ARBs are found in most major taxonomic phyla, only few major groups of bacteria have been studied more extensively, including Mycobacteria, Chlamydia-related bacteria, Rickettsia and Legionellae and will be the focus of this review (Table 2).

Actinobacteria: The genus Mycobacteria comprises many bacteria such as M. tuberculosis and M. leprae that are major threat to human health and 19 different species have been sequenced to date. Most Mycobacteria, including members of the nontuberculous and the tuberculous complex groups, have been shown to survive and grow in various amoebae such as Acanthamoeba sp., Dictyostelium discoideum, and Tetrahymena pyriformis (Thomas and McDonnell, 2007; Mba Medie et al., 2011). It was suggested that adaptations for the intra-amoebal survival and multiplication may have facilitated the virulence toward mammalian cells (Molmeret et al., 2005).

Chlamydiales: The Chlamydiales order includes important human and animal pathogens such as Chlamydia trachomatis. A large number of amoebal symbionts called “Chlamydia-related bacteria” have been retrieved and shown to infect various organisms, including free-living amoebae, arthropods, insects as well as vertebrates (Corsaro and Greub, 2006; Horn, 2008). Four of them have been fully sequenced (Table 2), two additional strains have been published as draft genomes and new representative of the Parachlamydiaceae and Criblamydiaceae families are currently being sequenced (Horn et al., 2004; Greub et al., 2009; Bertelli et al., 2010; Collingro et al., 2011).

Alpha-proteobacteria: Highly pathogenic representatives of Rickettsia are typically transmitted through arthropods (Merhej and Raoult, 2011) but some Rickettsia-like endosymbionts were observed in Acanthamoeba (Fritsche et al., 1999). Rickettsia bellii, a species that diverged early in evolution, was shown to survive for three weeks in A. polyphaga (Ogata et al., 2006). Odyssella thessalonicensis is a strict intracellular bacteria isolated from an air conditioning system in Greece (Doolittle and Bapteste, 2007). An unfinished genome sequence has been released, but unfortunately, little information is known on the genetic characteristics of this bacterium.

Gamma-proteobacteria: Legionellae are distributed worldwide and commonly found in water environments where they are major components of the biofilms (Rogers et al., 1994). Since the first culture of L. pneumophila, the causative agent of the Legionnaire's disease, within amoeba (Rowbotham, 1980), more than fifty strains of Legionella have been discovered. Although they are sometimes considered as all growing in amoebae, only a few have been shown to grow within amoebae (Rowbotham, 1980; Neumeister et al., 1997; La Scola et al., 2004) and the genomes of three species have been sequenced.

Several mechanisms allow genetic exchanges among organisms from the various domains of life (Paz and Espinosa, 2010). Briefly, virus-mediated transfers occur via transduction from phages (prokaryotes) or transposon (prokaryotes and eukaryotes). Prokaryotes use transformation, i.e., DNA uptake from the environment, or conjugation, i.e., DNA exchange through pilus-like systems, to transfer gene fragments or plasmids. Prokaryote to eukaryote transfers may arise as the result of the ingestion of cells, a process called phagotrophism, or symbiogenesis, i.e., the establishment of a permanent association.

Such events were notably involved in the formation of the animal lineage through the primary symbiosis forming the primary eukaryotic cell. Some authors hypothesized that Archaebacteria and an Actinobacteria interacted in the primary symbiotic event leading to the creation of mitochondria, a hypothesis that was soon rejected (Cavalier-Smith, 2002). Others suggested that an archaebacteria was invaded by a second prokaryote related to α-proteobacteria that had a bacteriovory ability similar to Bdellovibrio (Davidov et al., 2006; Cox et al., 2008). Georgiades et al. (2011) proposed that mitochondria are more related to Rickettsiales and the bacterium Pelibacter ubique. Recently, it was also noticed that mitochondria may be sister to Rhizobiales, Rhodobacterales, or Rickettsiales suggesting an eventual chimeric origin of mitochondria (Atteia et al., 2009; Abhishek et al., 2011; Georgiades and Raoult, 2011). Similarly, the symbiosis of a cyanobacterium with this primary eukaryotic cell was at the basis of the establishment of the chloroplast 1.2 billion years ago (Dyall et al., 2004). However, some lateral transfer events or chimeric events may have occurred with an ancestral Chlamydiales, as suggested by the high proportion of plant-like genes in chlamydiae (Brinkman et al., 2002).

Lateral transfers occur with varying frequency, magnitude and resulting fitness, which modulates the establishment of the transferred domain in the population. LGTs are frequent in prokaryotes and transferred genes can become integrated rapidly in the population thanks to short generation times (Paz and Espinosa, 2010). On the contrary, although increasingly reported, gene transfers involving eukaryotic species are still currently underrepresented in the litterature (Keeling and Palmer, 2008; Andersson, 2009). This is in part due to the large number of available prokaryotic genome sequence facilitating such analyses, compared to the currently restricted number of published sequence of unicellular eukaryotes.

Mobile genetic elements (MGEs) are essential actors and markers of non vertical genome evolution (Frost et al., 2005). Indeed, they often encode mechanisms to spread in new host and form sites for the preferential acquisition of exogenic sequences. Insertions sequences (IS) and transposases, mobile endonucleases of the HNH family and DNA methylases that may form restriction/modification systems, as well as inteins are typical examples of MGEs commonly found in bacteria and archaea. As such, they were found in several, if not all, ARBs and we only provide a few examples below. Interestingly, they were also identified in several ARVs and we report them here in more details.

The genome sequence of M. tuberculosis harbor many IS (n = 54) that are preferentially integrated in intergenic regions, close to tRNAs (Cole et al., 1998), whereas M. smegmatis and M. marinum present fewer IS (Stinear et al., 2008). Similarly, Chlamydia-related bacteria all encode numerous entire or remnants of transposases, whereas their relatives Chlamydia have little if no trace of invasion by such mobile elements (Bertelli et al., 2010). Among ARBs, A. asiaticus is an interesting case as it exhibits a massive proliferation of MGEs (up to 24% of the genome), including abundant IS, but its genome seems relatively stable (Schmitz-Esser et al., 2010). An interesting example of composite transposon is found in A. asiaticus where the cluster of gene for lasso peptide synthesis is flanked by two IS, suggesting its ability to be mobilized for genetic transfer (Schmitz-Esser et al., 2010). The genomes of Legionellae exhibited a large plasticity indicated by the presence of numerous mobile elements and three plasmids in L. pneumophila (Chien et al., 2004; Gomez-Valero et al., 2009; Cazalet et al., 2010).

HNH and restriction-like endonucleases have been found in Marseillevirus (n = 10), Lausannevirus (n = 8), Megavirus (n = 6) and Mimivirus (n = 3) (Raoult et al., 2004; Boyer et al., 2009; Arslan et al., 2011; Thomas et al., 2011). In addition, a phage-type endonuclease discovered in Megavirus and two HNH flanking a prophage gene in Mimivirus might have been laterally acquired from prophage genomes.

Inteins are typical selfish elements that mediate protein splicing to trigger their excision from the protein precursor (Perler et al., 1994). They often contain a homing endonuclease (HE) that cuts double-stranded genomic DNA to integrate itself in the corresponding gene of non-infected organisms (Gogarten et al., 2002). Giruses are invaded by inteins of different types and at different locations (Table 3).

The eukaryotic-like DNA polymerase B of Mimivirus encodes an intein and its likely functional HE that is most closely related to extremophile archaea (Ogata et al., 2005a). Marseillevirus and Lausannevirus present two highly similar inteins and their HE in orthologous genes, suggesting that these MGEs were acquired by their ancestor and are now evolving differently toward degradation (Thomas et al., 2011). Megavirus and CroV also possess respectively two and four inteins that were until now not studied in detail. As it is often the case, all viral inteins are integrated in highly conserved genes related to replication, transcription, or DNA metabolism. These conserved genes constitutes preferential target as they are essential to the virus. These observations are in agreement with the hypothesis that viruses might play a central role in the transmission of inteins across species (Pietrokovski, 1998) and suggest that these mobile elements had or still have the ability to excise, spread an insert into new hosts. However, since the various ARVs present inteins in orthologous genes, further studies are required to investigate if these MGEs have spread in giruses sharing a sympatric lifestyle or if they were acquired by common ancestors.

The role and the exact origin of these MGEs in amoeba-infecting viruses are in most cases unclear. MGEs are often colocalized with genes of putative bacterial origin (Filee and Chandler, 2010). Like in bacteria, MGEs could thus be involved in promoting or facilitating lateral transfers, which may in turn provide some selective advantage to the ARVs by facilitating the acquisition of new advantageous genes.

The role of genetic exchanges has been a matter of intense debate since the publication of the Mimivirus (Raoult et al., 2004; Moreira and Lopez-Garcia, 2005; Ogata et al., 2005b). The first analyses suggested massive LGT from various origins, including members of the four domains of life. It was suggested that the genomes of giant viruses infecting protists are largely affected by LGTs and non-orthologous gene displacements (Filee et al., 2007; Fischer et al., 2010; Koonin and Yutin, 2010). Some authors even regarded these viruses as “bags of genes” thus suggesting that the amount of genes transferred from amoebae to viruses overweight the flux of genes in the opposite direction (Moreira and Lopez-Garcia, 2005; Moreira and Brochier-Armanet, 2008). It was shown that genes with an anomalous nucleotide composition may lead to the overestimation of LGTs (Monier et al., 2007), and family specific genes do not present an accelerated evolutionary rate and were not laterally exchanged (Ogata and Claverie, 2007). Several studies thus used BLAST-based searches and phylogenetic methods to investigate the occurrence of lateral transfers from host to virus, virus to host as well as between viruses and other microorganisms (Table 4). Unfortunately, the use of different cutoffs renders some results difficult to compare.

Intracellular bacteria are thought to evolve by genome reduction rather than by acquisition of new genes (Moran, 2002). Individual genome publications as well as more detailed phylogenetic studies explored the events of LGT between bacteria, between bacteria and eukaryotes as well as with other microorganisms. Again, the various methods and cutoffs used make it difficult to directly compare the extent of genetic exchanges in each bacterial phyla. However, they provide an essential knowledge to appreciate the importance of such transfer events and the large diversity in the potential couples of donor-acceptor as shown by some well studied examples in Table 5.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.