ATP/ADP transferases move ATP across membranes, in exchange for ADP. There are two kinds, ATP exporters and ATP importers. There is no sequence homology between the two; they must have originated independently. ATP exporters are present in all mitochondria but have never been found in bacteria. The origin of the mitochondrial ATP/ADP transferase is unknown (Winkler and Neuhaus, 1999. Kurland and Andersson, 2000).

ATP/ADP transferases were first demonstrated in R. prowazekii (Winkler, 1976) and then in C. psittaci (Hatch et al., 1982). ATP/ADP transferases are present in all species of Chlamydiaceae that have been sequenced [see: Genome comparison for references], in R. prowazekii (Andersson et al., 2000), in R. conorii (Ogata et al., 2001) and in the chloroplasts of Arabidopsis thaliana and other plants (Mohlman et al., 1998). ATP/ADP transferases are also appear in the genome of the microsporidian Encephalitozoon cuniculi (Katinka et al., 2001), a eukaryotic obligate intracellular parasite of mammals (probably of fungal lineage), and in that of Xylella fastidiosa (Simpson et al., 2001; Koonin et al., 2001), a gamma-proteobacterium that is an extracellular pathogen of orange trees. Wherever found, these genes show a remarkably high degree of sequence homology, about 65% by deduced amino acid sequences. The value of ATP importers to chloroplasts and intracellular parasites is obvious. Arabidopsis chlorplasts depend on cytosolic ATP for synthesis of carbohydrates and fatty acids (Mohlman et al., 1998). Chlamydiae (McClarty, 1999), rickettsiae (Winkler, 1990; Anderson, 1998). and microsporidia (Katinka et al., 2001) not only generate ATP on their own but also use ATP/ADP transferases to tap the abundant ATP of their hosts' cytoplasm. However, X. fastidiosa is not an intracellular parasite, and the role of ATP/ADP transferases in its energy metbolism is unknown.

Although of extant Bacteria, Rickettsiales most closely resembles mitochondria (Andersson and Kurland, 1999; Emelyanov, 2001); even the most primitive of mitochondria (Lang et al., 1997) have no ATP importer gene. Ancient cyanobacteria are thought to be the progenitors of chloroplasts (Gray, 1992), but the genome of Synechocystis, the only cyanobacterium sequenced so far (Kaneka and Tabata, 1997) is also without the gene. ATP importers exchange the cytoplasmic ATP of eukaryotic cells for the ADP of chloroplasts or intracellular bacteria. Since there would have been no ATP in ancient extracellular habitats, the gene must have appeared after the appropriate endosymbiotic event. It would certainly not have been selected for in a micro-organism growing extracellularly. Winkler and Neuhaus (1999) have suggested that ATP importer genes lay unused in the nucleus of eukaryotic hosts until they were appropriated for their own purposes by primitive endobionts.

To go from one unanswered question to another, how is the phylogenetically discontinuous distribution of the ATP / ADP transferase genes among plants, microsporidia, rickettsiae, chlamydiae, and other bacteria to be explained? Their high degree of homology brings horizontal transfer immediately to mind. Wolf et al., (1999) believe that the ATP importers first appeared in plants from which they were transferred horizontally into chlamydiae and thence by the same mechanism into rickettsiae. The unusually large number of plant - related genes in the Chlamydiaceae genomes (Wolf et al., 1999) suggests that the Chlamydiales lineage passed through plant - like hosts, perhaps ancient unicellular eukaryotes, before the plant – animal - fungi divergence (Figure 1). This idea is consistent with the early entry of the lineage into intracellular life postulated in Part I, but a plant to chlamydia to rickettsia order of transfer is by no means established, and the presence of ATP/ADP transferases in E. cuniculi and X. fastidiosa must be explained. Another reason for suspecting that chlamydiae and rickettsiae have exchanged genes has recently emerged (Ogata et al., 2001). The dihydro-pteroate synthetase genes [see: Folate metabolism] of chlamydiae and R. conorii share an unusual bifunctional domain organization so far found only in chlamydiae, rickettsiae, plants and fungi. Effective contact between extant chlamydiae and rickettsiae growing in their modern hosts would be most difficult, but if transfer occurred a long time ago, when hosts were not metazoan but unicellular, the opportunities for transfer would have been much better. Returning to the free-living amoebae as the best available example of what early eukaryotic hosts might have been like, these organisms are hosts for many representatives of both Rickettsiales and Chlamydiales (Fritsche et al., 1999; Amman et al., 1999). Multiple infections have been observed.

An alternate explanation for the occurrence of similar genes in plants, fungi, and chlamydiae is that the ancestors of these genes were present in the shadowy life forms existing before the Bacteria-Eukaryota divergence (Figure 1). Perhaps the genes were lost in other bacteria and conserved in Chlamydiales because of its long and isolated intracellular existence, But it may not be that simple. Remember that rickettsiae, which are almost certainly much newer players in the intracellular game, also have the genes for ATP/ADP transferase and bifunctional dihydropteroate synthetase.
[JWM]