Although Legionella was the first pathogen demonstrated to multiply and persist in amoebae (18), several other fastidious intracellular bacterial pathogens, including Chlamydia pneumoniae (19), Mycobacterium avium (20), Listeria monocytogenes (21), and an Ehrlichia-like organism (22), may infect free-living amoebae. Extensive study of the ecology of Legionella pneumophila has confirmed empirical observations of its predilection for growth in hot water tanks and its localization in sediment (23). Rowbotham described the ability of L. pneumophila to multiply intracellularly within protozoa (18) and suggested that free-living amoebae could be a reservoir for Legionella species (24). As amoebae are common inhabitants of natural aquatic environments and water systems (25,26) and are resistant to extreme temperatures, pH, and osmolarity conditions while encysted (27), the Legionella reservoir is important. Growth of free-living amoebae at high temperatures (44°C to 53°C) was observed more frequently for strains isolated from hot-water tanks (mainly Hartmanella vermiformis) than for those isolated from moist sanitary areas (mainly Acanthamoeba, Naegleria, and Valkhampfia species) (26). This great tolerance of cysts and species-dependent thermotolerance of trophozoites could account for the difficulty in eliminating Legionellae from water systems (28). The resistance of Acanthamoeba spp. cysts to various disinfecting solutions (29–31) complicates the eradication of free-living amoebae. Moreover, a wide variety of Enterobacteriaceae have increased resistance to chlorination when ingested by Tetrahymena pyriformis (32). Thus, free-living amoebae could readily act as Trojan horses for bacterial endosymbionts (33,34).

The relationship between Legionellaceae and free-living amoebae, which serves as a model for other endosymbionts such as Parachlamydiaceae, is not restricted to the role of reservoir. Indeed, Acanthamoeba strains were found to produce Legionella-containing vesicles, which may be agents of transmission of legionellosis. The risk of transmission may be underestimated by plate count methods (35). In addition, Legionellae grown inside amoebae were more virulent (36,37), more motile (24), and more resistant to biocides (38) than are bacteria cultured in axenic media. The entry of Legionellae into monocytes was found to be enhanced by the intra-amoebal growth environment (39). In addition, intra-amoebal growth of L. pneumophila was shown to induce an antibiotic-resistant phenotype, while Legionellae cultured in broth did not (40). Similarly, M. avium living within Acanthamoeba had greater resistance to rifabutin, clarithromycin, and azithromycin than did strains living in macrophages (41). This finding could result from decreased uptake of antibiotics into the amoebae, an inactivation of the compound within amoebae, or a change in the bacterial phenotype. Replication of bacteria in amoebae was found not only to affect the bacterial host (through increased potential for spread, resistance to biocides and antibiotics, and acquisition of virulence traits) but also to enhance the pathogenicity of the free-living amoebae (42).

These Chlamydia-like endosymbionts are small Gimenez-stained (43) coccoid bacteria (Figure 1) that naturally infect amoebae and are inconsistently stained with Gram stain. Electron micrographs of Acanthamoeba demonstrate the presence of bacteria at different developmental stages typical of the Chlamydiales, such as elementary and reticulate bodies (Figure 2). A new Parachlamydiaceae family was proposed (44) that forms a sister taxon to the Chlamydiaceae, as it has a Chlamydia-like cycle of replication and 80% to 90% homology of ribosomal RNA genes. This family comprises two genera, of which the type strains are Parachlamydia acanthamoebae (17) and Neochlamydia hartmanellae (45). Members of the Parachlamydia were proposed to have at least 95% homology of the 16S or 23S rRNA genes with P. acanthamoebae (44). However, comparison of the 16S rRNA gene sequences of four additional Parachlamydia with P. acanthamoebae showed substantial phylogenetic diversity within this genus (Figure 3), with 91.2% to 93.1% 16S rRNA gene sequence homology with P. acanthamoebae (46). The ecologic loci and prevalence of the Parachlamydiaceae are unknown, but the latter could be underestimated, as this fastidious gram-negative bacteria was recovered only by amoebal cocultures, a procedure not performed routinely on clinical samples. Moreover, these Chlamydia-like organisms have potential for widespread dissemination, as they are mostly endosymbionts of Acanthamoeba, a free-living amoeba with worldwide distribution (27).

Nine strains of Parachlamydia have been described (Table). The first, P. acanthamoebae, was identified within Acanthamoeba BN9, an amoeba recovered from the nasal mucosa of a female volunteer (17). Its 16S rRNA sequence had 88.2% homology with Simkania negevensis and 87% homology with Chlamydophila pneumoniae (17). The second, Berg17 endosymbiont, also isolated from the nasal mucosa of a female volunteer, seems to have an rRNA signature similar to that of the Bn9 endosymbiont, as demonstrated by the binding of the Bn9₆₅₈ hybridization probe designed for in situ identification of P. acanthamoebae (17). The third, Hall’s coccus, was found in an Acanthamoeba isolated from water taken from a humidifier in a case of humidifier-associated fever in Vermont (16). Its 16S rRNA gene sequence had >99% similarity with that of Bn9 endosymbiont and 86% to 87% with those of the four recognized Chlamydia species (16). Two additional Parachlamydiaceae, UWE1 and UWE25, were also found to infect Acanthamoeba. Both amoeba strains were recovered from soil samples from Washington State (46). A sixth strain, UWC22 endosymbiont, infected an Acanthamoeba recovered from infected corneal tissues (46). TUME1 endosymbiont was found in an amoeba recovered from municipal sewage sludge in Germany (46). The eighth strain, Neochlamydia hartmannellae, is the only strain of Parachlamydiaceae isolated from Hartmanella vermiformis. It did not grow on Acanthamoeba sp. or Naegleria, and its 16S rRNA gene sequence had only 92% homology with that of P. acanthamoeba and varied from 91.6% to 97.1% with the four latter endosymbionts of Acanthamoeba (45). The last one, CorvenA4, could not be isolated. Only its 16S rRNA sequence was retrieved from a respiratory sample (47).

The role of Parachlamydia sp. as an emerging pathogen needs to be confirmed. In view of the genetic diversity of the Parachlamydiaceae (46), their phylogeny needs to be elucidated, as the various species could be associated with species-specific pathogenicity. Search for additional Parachlamydia strains in hospital water systems could help define potential nosocomial exposures. Because the Parachlamydiaceae are difficult to culture, simpler approaches are being developed, including serologic and molecular tests. These methods could be performed on a large number of samples from both healthy and ill persons. Patients with community-acquired pneumonia, nosocomial pneumonia, Kawasaki disease, and arteriosclerosis should be tested. Increased resistance to antimicrobial drugs, which may be associated with intra-amoebal growth, is another promising area for future study.