Free-living amoebae are protozoa that feed by internalizing energy-rich particles, mainly bacteria [1]. However, some bacteria have adapted to become resistant to amoebal phagocytosis [2]. Some of them have been proposed to be amoebal endosymbionts, including Amoebophilus asiaticus [3] and Candidatus "Protochlamydia amoebophila" (Candidatus "P. amoebophila") [4]. The pathophysiological basis of this specific association is as yet unknown. Candidatus. "P. amoebophila" is the only obligate intra-amoebal bacterium whose genomic sequence has been released to date [5].

Recently, we described Legionella drancourtii (L. drancourtii) within an Acanthamoeba sp. amoeba. First named Legionella-like amoebal pathogen 12, it was initially thought to be strictly intra-amoebal [6]. To investigate the genetic features associated with its association with amoebae, we sequenced the genome of L. drancourtii. In order to identify the genes associated with amoebal parasitism, we herein compared the genome sequences of L. drancourtii and Candidatus "P. amoebophila" to detect the genes common to both species but absent from other prokaryotes.

By comparing the genomes of the intra-amoebal bacteria Candidatus "P. amoebophila" and L. drancourtii with those of closely related species, we identified a sterol delta-7 reductase, likely acquired from viridiplantae by horizontal gene transfer (HGT). The L. drancourtii genome, with a size of 4.2 Mb, appears to be larger than that of sequenced Legionella pneumophila strains, with an average size of 3.5 Mb [13, 14]. In contrast with other intracellular bacteria in which the intracellular lifestyle is associated with genome reduction [15, 16], amoebal endosymbionts appear to have larger genomes than their relatives. The largest difference is in Candidatus "P. amoebophila", whose 2.4 Mb genome is about twice as large as the genomes of pathogenic Chlamydia species [5]. Another example can be found in Rickettsia bellii, able to survive in amoebae, whose genome is the largest among Rickettsia species and contains an abundance of amoebal parasite genes, thus showing possible HGT within amoebae [17]. Among viruses, Acanthamoeba polyphaga mimivirus, an amoeba virus, also possesses the largest genome (1.18 Mb) of all known viruses [18]. In contrast with obligate intra-cellular bacteria from other eukaryotic cells, which are isolated in their hosts, and thus have a limited ability to exchange genetic material, we hypothesize that amoebae, which feed on bacteria, constitute a favourable place for genetic exchange between intra-amoebal prokaryotes, which may thus have larger genomes that those of other intra-cellular bacteria.

Among orthologous proteins of Candidatus "P. amoebophila" and L. drancourtii, we discovered a protein similar to enzymes of the sterol reductase family and more particularly similar to sterol delta-7 reductase, otherwise found in most eukaryotes but in only few bacteria. A genomic search confirmed that L. drancourtii is the only Legionella species to have this protein. Phylogenetically, the L. drancourtii protein clustered with the sterol delta-7 reductases of Candidatus "P. amoebophila" and C. burnetii, also amoeba-resistant, and this group was closely related to sterol delta-7 reductases of viridiplantae supported by a high bootstrap value. The most parsimonious explanation for the presence of this gene in these three amoeba-resistant bacteria is that it was transferred from eukaryotes, more precisely from viridiplantae. An alternative but less likely explanation would be that the gene was lost by all other bacteria.

Inter-kingdom gene transfers have already been reported, notably from bacteria to humans [19] and from bacteria to insects [20]. However, HGT from eukaryotic cells to bacteria is a rare event of particular interest because of the possible influence of such acquired genes on bacterial fitness [21]. In addition, it was shown that some bacterial genomes contain genes from viridiplantae, which may be the case in this study. The most clear-cut example is the eukaryotic ATP/ADP translocase, an enzyme of intracellular symbiont mitochondria and chloroplasts found in Chlamydiales and Rickettsiales, [22, 23]. On the basis of a phylogenetic analysis clustering these proteins with plant homologs, two plausible hypotheses were inferred: i) the ATP/ADP translocase was acquired from a plant by the Chlamydia ancestor that might have been a plant parasite and was transferred to rickettsiae [21] or ii) a nuleotide transport protein was invented by the Chlamydia ancestor to support its intracellular lifestyle, duplicated, evolved into an ATP/ADP translocase gene and transferred to the Rickettsia ancestor and to plants [22, 23]. Evidence for gene exchanges between intracellular bacteria of amoebae was shown in the Rickettsia belli genome which possesses amoeba-associated bacteria genes [17]. Thus, this transfer from chlamydiae to rickettsiae may have occured within amoebae. We believe that the first hypothesis may apply to the sterol delta-7 reductase-encoding gene, leading to two scenarios (Figure 4): i) the Chlamydia ancestor acquired this gene from plants, after a genomic reduction, it was conserved only by Candidatus "P. amoebophila" and subsequently transferred to other bacteria sharing the same intra-amoebal biotope or ii) the gene was directly transferred from the Chlamydia ancestor to the Legionella ancestor within amoebae and after genome reductions, it was conserved only by Candidatus "P. amoebophila" and L. drancourtii. However, another hypothesis may place the HGT between an as yet unidentified aquatic plant living in the natural habitat of amoebae and intra-amoebal bacteria.

In eukaryotes, the sterol delta-7 reductase enzyme catalyses the final step of cholesterol biosynthesis reducing a double bond at carbon 7 of 7-dehydrocholesterol to produce cholesterol (Figure 5), lipid playing important roles in the physiology of eukaryotic organisms [24, 25]. This enzyme is a member of the sterol reductase family including the C14 sterol reductase and the C24 sterol reductase, which exhibit high sequence similarities [26] and whose the phylogenetic relationships study showed an organization into three distinct clusters (Figure 1).

Although the role of the sterol delta-7 reductase is not clearly identified in prokaryotes, there is evidence that some intracellular bacteria can interfere with the host cholesterol metabolism. C. burnetii upregulates host genes involved in both cholesterol uptake and biosynthesis, inducing cholesterol accumulation necessary for its replication [27]. Moreover, the C. burnetii genome has genes predicted to encode enzymes involved in the first part of the sterol biosynthesis (Figure 5) [28]. Comparison of the presence of these genes in the three genomes showed that Candidatus "P. amoebophila" has only the sterol delta-7 reductase-encoding gene, while L. drancourtii has almost all genes. Some bacteria, like some Myxococcales and Methylococcales, contain only a portion of enzymatic pathway involved in sterol biosynthesis and as a consequence can synthesize sterols [29, 30], however, only these three pathogens have the last enzyme.

We demonstrated that the sterol delta-7 reductase-encoding gene was transcribed in L. drancourtii. This result, together with the involvement of this gene in the cholesterol metabolism of eukaryotes, and its presence in C. burnetii which is also amoeba-resistant and is cholesterol-dependent, raise the hypotheses that the sterol delta-7 reductase might be functional and play a role in host cholesterol parasitism in L. drancourtii.

In conclusion, by comparing the genomes of two intra-amoebal bacteria with those of their close relatives, we identified a sterol reductase-encoding gene likely acquired from viridiplantae. The role of this gene in bacteria is as yet unknown but it could be involved in host cholesterol parasitism and appear to be linked to intra-amoebal host fitness.