From Many Hosts, One Accidental Pathogen: The Diverse Protozoan Hosts of Legionella

1In the environment, bacteria are targets of predation by grazing protozoa (Hahn and Höfle,; Molmeret et al., ). In response to predation, many bacteria have developed strategies to either avoid predation or survive, and in some cases, replicate within protozoa. As bacteria are destined to encounter a large number of protozoa species in nature, their fitness will be determined by the breadth and diversity of protozoa within which they are able to grow. Though many types of bacteria are able to replicate within protozoa (Greub and Raoult, ), this behavior is best characterized in the bacterial pathogen Legionella, in particular Legionella pneumophila, which will be the major focus of this review.

2L. pneumophila is ubiquitous in nature (Fliermans,; van Heijnsbergen et al., ). While various species of Legionella have been isolated from soil and marine environments, freshwater systems serve as the major reservoirs of L. pneumophila (Fliermans,; van Heijnsbergen et al., ). L. pneumophila can exist in a planktonic form however, it is more often found within mixed community biofilms (Mampel et al., ). L. pneumophila intercalates into existing biofilms (Lau and Ashbolt,; Stewart et al., ) where it acquires nutrients by forming synergistic relationships with other members of the biofilm (Tison et al.,; Pope et al.,; Bohach and Snyder,; Wadowsky and Yee,; Stout et al.,; Stewart et al.,; Koide et al., ). L. pneumophila is also capable of surviving in nutrient-poor conditions by necrotrophic growth on dead cell masses (Temmerman et al., ). Although, its interactions with other bacteria promote L. pneumophila survival in oligotrophic environments, intracellular growth within protozoa is likely the predominant mechanism of L. pneumophila proliferation in its natural habitat (Rowbotham, ).

3Protozoa function as natural reservoirs of L. pneumophila and promote disease in humans. The intracellular environment of the host cell protects L. pneumophila from harsh environmental conditions while providing a nutrient rich replicative niche (Greub and Raoult,; Abdel-Nour et al., ). The ability of L. pneumophila to survive within amoebae also protects the bacteria from killing by water disinfection procedures (Plouffe et al.,; King et al.,; Kilvington and Price,; Biurrun et al.,; Storey et al.,; Bouyer et al.,; García et al.,; Cervero-Aragó et al.,, ), a reciprocal relationship that also enhances survival of the host (García et al., ). As a consequence, L. pneumophila are commonly found in man-made potable water supply and distribution systems (Ikedo and Yabuuchi,; Breiman et al.,; Yamamoto et al.,; Fields et al.,; Lasheras et al.,; Brousseau et al.,; Thomas et al., ). Although, there is one reported case of probable human-to-human transmission of Legionella (Correia et al., ), the vast majority of evidence suggests a non-communicable disease. Instead, human exposure predominantly occurs through the inhalation of contaminated water aerosols (Fields, ), which can lead to pneumonic respiratory disease. L. pneumophila passaged through amoebae are more virulent in animal models of infection compared to bacteria grown in broth culture (Cirillo et al.,,; Barker et al.,; Brieland et al.,; Garduño et al., ). The earliest description of L. pneumophila's interaction with amoebae even proposed that an important route of human infection may be the inhalation of the pathogen in an amoebal-encapsulated state (Rowbotham, ). Thus, the interaction of L. pneumophila with protozoa is a critical determinant in both the persistence of Legionella in environmental and man-made reservoirs, and the incidence and severity of disease.

4Many bacterial pathogens become highly specialized for growth in one or a small subset of hosts but few are able to grow in multiple hosts. Host jumping has been observed for some pathogens but often comes at a price, the inability to grow in the previous host (Ma et al., ). In contrast, L. pneumophila exhibits an extensive host range replicating within a diverse array of protozoan hosts that span multiple phyla, from Amoebozoa (amoebae) to Percolozoa (excavates) to Ciliophora (ciliated protozoa) (Rowbotham,; Fields, ). The ability to maintain such a broad host range is due to the assembly of a large cohort of genes that allow L. pneumophila to adapt to variations between hosts (O'Connor et al., ). Moreover, the ability to continually evolve and alter the composition of its virulence gene repertoire allows L. pneumophila to adapt to shifts in protozoan populations in their natural habitats (O'Connor et al., ). Since the discovery that L. pneumophila can survive and replicate within free-living amoeba (Rowbotham, ), the relationship between L. pneumophila and its protozoa hosts has garnered significant attention, largely due to the important role of protozoa in the epidemiology of this pathogen. In this review, we expand on the early works of Rowbotham and Fields (Rowbotham,,; Fields, ) to summarize the current knowledge of the host range of L. pneumophila in environmental reservoirs and the factors that impact the outcome of Legionella-protozoa interactions.

5While L. pneumophila has an extensive host range, the fate of the bacterium once it enters the host cell can vary greatly. Several protozoa are able to efficiently deliver L. pneumophila to the lysosome for degradation, resulting in the death of the bacterium (Amaro et al., ). L. pneumophila predation by protozoa does not seem to be restricted to one particular group. While members of the Cercozoa phylum seem to be especially adept at digesting L. pneumophila (Amaro et al., ), distantly related members of the Amoebozoa phylum (Cashia limacoides, Vannella platypodia, and Vexillifera bacillipedes) are also efficient at killing L. pneumophila (Rowbotham, ). In contrast, many protozoa serve as hosts for L. pneumophila replication. In these cases, the Legionella-protozoa interaction is detrimental to the host: the bacteria multiply to high numbers and then kill the host as they exit the cell (Rowbotham, ). Alternatively, L. pneumophila can be toxic to the host in the absence of replication, a protist version of food-poisoning (Amaro et al., ). L. pneumophila within amoebae has been shown to inhibit both amoebae proliferation (Mengue et al., ) and chemotactic motility (Simon et al., ). The fates of the two organisms are not solely defined by this “it's you or me” relationship, as a number of intermediate outcomes have been observed. In response to extreme stress, amoebae undergo encystation, transforming into a dormant, highly resistant cyst form. While encystation restricts bacterial replication (Rowbotham,; Ohno et al., ), L. pneumophila is able to survive the encystation process until more favorable conditions arise (Kilvington and Price,; Greub and Raoult, ). Similarly, for some Legionella-protozoa pairs, L. pneumophila is resistant to grazing by the protozoan and thus survives within the host cell but fails to replicate (Smith-Somerville et al., ). Alternatively, L. pneumophila can be packaged into multi-membrane vesicles that are distinct from the replication vacuole and expelled into the extracellular environment (Rowbotham,; Berk et al.,; Hojo et al.,; Amaro et al., ). The release of Legionella-containing pellets has been observed in both the ciliated protozoa Tetrahymena spp. (Faulkner et al.,; Hojo et al., ) and the amoebal hosts Acanthamoeba castellanii and Acanthamoeba astronyxis (Bouyer et al.,; Amaro et al., ), and does not appear to coincide with bacterial replication. Whether this process is driven by the bacterium or the host is still unclear. The pellet compartment can protect L. pneumophila from environmental stress (Bouyer et al.,; Koubar et al., ) which would be beneficial during its transition between host cells and thus a potential mechanism to ensure its survival. Consistent with this idea, a functional Type IVb secretion system, a major L. pneumophila virulence factor required for lysosome avoidance and intracellular replication, appears to be important for the release of L. pneumophila in pellets (Berk et al., ). Alternatively, the inability to digest the bacteria may simply trigger a host response that involves bacterial expulsion, as a similar phenomenon is observed with non-pathogenic Escherichia coli, Bacillus subtilis, and Mycobacterium luteus (Hojo et al.,; Denoncourt et al., ). Whether L. pneumophila resists predation or is expelled in pellets, the host is considered to be only partially restrictive due to the survival of L. pneumophila and its potential to transition to other host cells. Indeed, one might speculate that such intermediate host-bacterial interactions (resistance to protozoan predation in the absence of replication) might resemble the first evolutionary step toward becoming an intracellular pathogen.

6Protozoan hosts of Legionella are defined by two main techniques: co-culture and co-isolation. When combined with microscopy, co-culture techniques allow for the direct visualization of Legionella within host cells, and by analyzing infected cells over time, bacterial replication within a particular host provides direct experimental evidence of Legionella survival and replication. When combined with plating assays to monitor bacterial numbers, co-culture methods allow bacterial growth rates, maximum growth and the impact of bacterial dose and various external conditions on the interaction to be analyzed. However, while Legionella may be able to replicate in a given host under specific laboratory conditions, the experimental system may not reflect conditions encountered in the environment and thus, biologically relevant interactions that commonly occur in nature. Co-isolation studies attempt to address this issue by examining the co-existence of protozoa and Legionella in environmental samples. In rare cases, protozoa harboring Legionella have been isolated from environmental samples providing direct evidence of their interaction in the environment (Thomas et al.,; Hsu et al.,; Kao et al., ). More commonly, Legionella are identified by 16S sequencing of DNA extracts from bacteria isolated by Legionella-selective culture methods on bacteriological medium (Salloum et al.,; Sheehan et al., ) or enrichment through co-culture of environmental samples with amoebae (Pagnier et al., ). Protozoa may be identified microscopically by fluorescence in situ hybridization (FISH) or the morphological appearance of trophozoites (Jacquier et al.,; Muchesa et al., ), or by 18S sequencing of DNA extracts following an amoebal enrichment step in which individual isolates are cultured on lawns of bacteria permissive to amoebal grazing (Greub and Raoult,; Delafont et al.,; Muchesa et al., ). Thus, while most co-isolation studies do not provide direct evidence of Legionella growth within the protozoa identified, they can be used to predict environmentally relevant interactions, to substantiate experimental findings from co-culture techniques and are likely to implicate new protozoan species as potential hosts of Legionella.

7The initial discovery that L. pneumophila is capable of surviving and replicating in protozoa fostered a number of independent investigations to examine the host range of this bacterium (Table 1). Co-culture methods in combination with various microscopy techniques demonstrated growth of L. pneumophila in diverse protozoan hosts encompassing several species of Acanthamoeba (A. castellanii, Acanthamoeba polyphaga, and Acanthamoeba palestinensis), Hartmannella (Vermamoeba vermiformis, formerly Hartmannella vermiformis and Hartmannella cantabridiensis) and Naegleria (Naegleria gruberi, Naegleria lovaniensis, and Naegleria jadini) as well as Tetrahymena pyrofomis, Echinamoeba exudans, and Tetramitus jugosus (formerly Vahlkampfia jugosus) (Rowbotham,,; Tyndall and Domingue,; Anand et al.,; Barbaree et al., ). While the list of hosts was dominated by three particular genera (Acanthamoeba, Hartmannella, and Naegleria), collectively it represented three different phyla Amoebozoa, Ciliophora, and Percolozoa and amongst them, four distantly related classes of protozoa, Discosea (Acanthamoebae), Tubulinea (Echinamoeba and Hartmannella), Heterolobosea (Naegleria and Tetramitus), and Oligohymenophorea (Tetrahymena) (Figure 1).

8Subsequent studies to investigate L. pneumophila pathogenesis have progressively expanded the list of protozoan hosts of this bacterium (Table 1 and Figure 1), including additional species of Acanthamoeba (Acanthamoeba lenticulata and Acanthamoeba royreba) and Naegleria (Naegleria fowleri) as well as more distantly related genera from their respective phyla such as Dictyostelium discoideum (Hägele et al.,; Solomon et al., ) and Balamuthia mandrillaris (Amoebozoa) (Shadrach et al., ) and Willertia magna (Percolozoa) (Dey et al.,; Tyson et al., ). Similarly, a number of additional ciliated protozoa were identified that were permissive for L. pneumophila survival, including Tetrahymena spp. (Tetrahymena tropicalis and Tetrahymena vorax), Oxytricha bifaria, Stylonychia mytilus, Paramecium caudatum and a member of the Ciliophrya genus, and in one case L. pneumophila replication (Tetrahymena thermophila), greatly expanding representation from this group (Kikuhara et al.,; Rasch et al.,; Watanabe et al., ). The beneficial interaction of L. pneumophila with these organisms appears to be specific as members from each of the representative phyla were also identified that were highly restrictive to L. pneumophila survival (Figure 1): T. vorax (Ciliophora), A. astronyxis, and Cashia limocoides (Amoebozoa) and Solumitrus palustris (Percolozoa) (Rowbotham,; Smith-Somerville et al.,; Amaro et al., ). In addition, L. pneumophila was unable to grow in V. platypodia and V. bacillipedes (Rowbotham, ), which form a distantly related clade of the Amoebozoa phyla (Figure 1). Similarly, of the members of the Cercozoa phylum examined so far, Cercomonas sp., Euglypha sp., and Paracercomonas sp., all three are restrictive for L. pneumophila growth (Amaro et al.,; Rasch et al., ), suggesting that distinct orders and families within this class may be more restrictive than others. Thus, while the host range of L. pneumophila is vast, it does appear to have its limitations.

9Protozoa in both natural and man-made environments can alter the composition of microbial communities by eliminating bacteria through predation or augmenting populations of bacteria that are capable of replicating within these organisms (Yamamoto et al., ). Co-isolation techniques have been used to describe the composition of these communities within natural fresh water systems such as hot springs, thermal spas, lakes, ponds, streams, and anthropogenic reservoirs, such as cooling towers, industrial and private water networks and compost facilities. L. pneumophila is capable of surviving an array of physical conditions including temperatures ranging from 6 to 63°C (Fliermans et al., ). Thermal springs have been of particular interest as they boast characteristically high water temperatures, providing optimal conditions for L. pneumophila growth (Hsu et al.,; Ji et al.,; Rasch et al., ). Artificial aquatic reservoirs are of considerable epidemiological significance and typically support higher numbers of bacteria compared to natural water systems (Yamamoto et al., ), likely due to higher average water temperatures (Ikedo and Yabuuchi,; Fields et al.,; Lasheras et al., ). The results of these population level analyses have validated many of the co-culture defined hosts of L. pneumophila while identifying several additional potential hosts (Table 2).

10There is tremendous concordance between co-culture-confirmed Legionella-protozoa interactions and the results of co-isolation studies (Tables 1, 2). With the exception of Balamuthia and Dictyostelium, all protozoan genera shown to support intracellular growth in laboratory co-culture studies reside with L. pneumophila in the environment (Table 2). While this is not surprising for Acanthamoeba, Hartmannella, and Naegleria, as these are some of the most abundant protozoa in nature, in many cases co-isolation studies identified the same species of these genera. In particular, three of the protozoa identified, A. palestinensis, N. lovaniensis, and V. vermiformis that had been shown to support L. pneumophila replication in co-culture experiments (Anand et al.,; Rowbotham,; Declerck et al.,; Thomas et al., ) were isolated from water samples harboring L. pneumophila (Kao et al., ). Similarly, amoebal enrichment assays resulted in the isolation of Acanthamoeba jacobsi harboring L. pneumophila directly from a thermal spring water sample (Hsu et al., ). These results identify A. jacobsi as a new host of L. pneumophila and provide direct evidence of an interaction between L. pneumophila and these four protozoan hosts in the environment. The lack of co-isolation of L. pneumophila with either Balamuthia or Dictyostelium species is likely because these protozoa are typically found in soil and the majority of samples analyzed were isolated from aquatic environments (Dunnebacke et al.,; Vadell and Cavender, ). The high degree of correlation between the co-culture and co-isolation studies supports the role of these organisms as natural hosts of L. pneumophila in environmental reservoirs.

11Co-isolation studies predict a number of additional phyla and classes of protozoa may support L. pneumophila survival or growth (Table 2). In addition to the Amoebozoa, Ciliophora, and Percolozoa phyla, protozoa from Apusozoa (Diphylleia rotans), Cercozoa (Euglypha sp.), Euglenozoa (Bodonidae sp.), and Opsithokonta (Rhinosporidium sp.) were identified. Two additional classes of protozoa from previously identified phyla are also represented, Variosea (Flamella balnearia) and Oligohymenophorea with representatives encompassing four different families spanning three orders within this group. For those classes of protozoa already identified as hosts by co-culture experiments, three additional orders, Thecamoebida (Stenamoeba limacina), Arcellinida (Centropyxis sp.), and Sporadotricina (Aspidiscidae family) and five genera (Comandonia operculata, C. limacoides, Paravahlkampfia ustiana, Learamoeba waccamawenis, and Singhamoeba horticola) were identified. Finally, of the known hosts of L. pneumophila from co-culture experiments, additional species of Acanthamoeba (A. jacobsi), Naegleria (Naegleria pagei and Naegleria australiensis), Tetramitus (Tetramius enterica), and Vahlkampfia (Valkampfia avara) were also isolated. Combined, co-isolation and co-culture experiments represent 7 of the 8 phyla of the protozoa kingdom, 12 of the 41 classes within these phyla and 21 of the 82 defined orders, demonstrating the tremendous diversity amongst L. pneumophila hosts.

12Protozoa more commonly found associated with L. pneumophila in environmental reservoirs may indicate that they are more likely to be true hosts of the bacterium. While the Acanthamoeba spp., Naegleria spp., Vahlkampfia spp., and Hartmannella spp. (including Vermamoeba vermifomis) are commonly found in multiple sources (Table 2), particular protozoa appear to co-reside with L. pneumophila in more than one environmental sample (Table 2). A. hatchetti, A. polyphaga, H. cantabrigensis, N. fowleri, N. lovaniensis, Neoparamoeabe sp., and Willertia sp. have been isolated from both natural and man-made water sources (Table 2), suggesting that these protozoa may function as hosts of L. pneumophila in both natural reservoirs and potable water. Both E. exudans and Echinamoeba thermarum have been identified in more than one potable water sample (Table 2), suggesting these amoebae may play more prominent roles in the epidemiology of L. pneumophila. A higher incidence of specific protozoa with L. pneumophila may indicate a stronger likelihood that these protozoa are responsible for the persistence of L. pneumophila in environmental reservoirs.

13Not all protozoa species isolated from the same environmental source are hosts of L. pneumophila. Of several species of free-living amoeba collected from a cooling tower, only A. polyphaga supported intracellular growth of L. pneumophila whereas L. pneumophila failed to replicate within C. limacoides, V. platypodia, and V. bacillipedes (Rowbotham, ). Similarly, of several ciliated protozoa species in biofilm samples isolated from a thermal spa, L. pneumophila was able to infect Ciliophrya sp., O. bifaria, and S. mytilus, but no intracellular bacteria were detected within Euglypha sp. or Centropyxis sp. (Rasch et al., ). Thus, L. pneumophila is able to persist in environments comprised of both L. pneumophila-restrictive and permissive protozoan hosts. The relative abundancy of L. pneumophila in different environmental niches may reflect mixed populations of these two types of protozoa. Alternatively, in some circumstances L. pneumophila may deplete entire populations of permissive hosts, enriching for resistant species of protozoa that remain. Thus, the absence of certain types of protozoa may not necessarily rule them out as contributors to L. pneumophila growth and persistence in the environment.

14The distribution of protozoa between the types of water sources examined (natural water reservoirs, cooling towers, potable water distribution system, and compost sites; Table 2) was relatively uniform with a few notable exceptions. Amoebozoa and Percolozoa, making up the majority of the protozoa identified, were found in all water sources. Amoebozoa were more predominant in cooling towers and potable water systems. The lower abundance of Percolozoa in cooling towers coincided with a higher abundance of Ciliophora (ciliated protozoa) whereas in potable water, an enrichment in organisms from the Tubulinea class of Amoebozoa, in particular Echinamoeba was observed. In contrast, fewer members of the Discosea class were reported and in particular, no members of the Centramoebida order despite their presence in all other sites. The perseverance of L. pneumophila within various water environments despite variation in the protozoa composition demonstrates the highly adaptive nature of this bacterium to fluctuations in host population dynamics.

15Although co-isolation studies provide valuable insights into the microbial communities that support L. pneumophila, these methods cannot adequately define the full diversity of these communities (Kunin et al., ). While enrichment steps are often necessary to identify low abundance organisms, they create experimental bottlenecks and biases by selecting against protozoa that cannot be cultured using standard protocols (Hugenholtz and Tyson,; Gomez-Alvarez et al., ) and Legionella isolates with host specificities that do not overlap with amoebal species commonly used in these techniques (Evstigneeva et al., ). Metagenome-based analyses may circumvent the limitations inherent to culture-based approaches and provide a more comprehensive, unbiased profile of these communities (Hugenholtz and Tyson,; Gomez-Alvarez et al., ). For example, metagenomic studies of samples from three separate watersheds showed both a high level of diversity in the population of Legionella (encompassing 15 different species) and a correlation between the levels of Amoebozoa present in the water and the abundance of Legionella isolates (Peabody et al., ). Monitoring the abundance of Legionella, Hartmannella, and Naegleria from two environmental water sources over the course of a standard water purification procedure suggested a correlation between the abundance of Legionella and Naegleria, but not Hartmannella (Lin et al., ). In general however, metagenomics studies have been somewhat difficult to interpret. Often individual sites are dominated by one or a few amoebal species and the relative abundance of L. pneumophila is extremely low compared to other bacteria (Liu et al.,; Delafont et al., ): these features make it difficult to correlate the presence of L. pneumophila with specific protozoa. As the sensitivity and depth of metagenomics analysis improves, metagenomics will most certainly be a source of tremendous insight into the full repertoire of protozoan hosts of L. pneumophila.

16The outcome of the interaction between L. pneumophila and protozoa can be influenced by a number of factors; the identity of the host cell, variations in the predatory behavior or feeding preferences of the host, the strain or species of the bacterium, the relative abundance of the two organisms, the external environment, and other microorganisms.

17The identity of the host cell can greatly impact the outcome of the infection. While some hosts are permissive for L. pneumophila replication, others are restrictive, either impeding bacterial growth or in extreme cases, survival (Amaro et al., ). The maximum amount and rate of L. pneumophila growth between hosts can vary significantly (Declerck et al., ). For example, L. pneumophila can achieve up to 10,000-fold growth in A. castellanii but only 10-fold growth in N. lovaniensis over the same time period (Declerck et al., ). Similarly, L. pneumophila strain Paris grows robustly in A. castellanii and V. vermiformis but is defective for growth in W. magna (Dey et al., ). Moreover, the differential growth of L. pneumophila Paris varies between different strains of W. magna, with robust growth in strain T5[S]44 (Tyson et al., ) but failure to grow in strains c2c Maky or Z502 (Dey et al., ). Thus, some hosts are more optimal than others for L. pneumophila survival and replication.

18The predatory behavior and feeding preferences of the host can also influence Legionella-protozoa interactions. For example, the L. pneumophila auto-inducer LAI-1 disrupts chemotactic migration of D. discoideum (Simon et al., ) and promotes L. pneumophila uptake in both D. discoideum and A. castellanii (Tiaden et al., ). By restricting amoebal movement, L. pneumophila may localize feeding to the site of the bacteria—such modulation may also enrich for specific types of amoebae that support L. pneumophila replication. The LAI-1 biosynthesis genes are not conserved in all Legionella species (Burstein et al., ) suggesting that individual species may differentially promote their interaction with amoebae or do so via different mechanisms. Consistent with this idea, the host cell receptors that mediate L. pneumophila adhesion to V. vermiformis, A. castellanii, A. polyphaga, and N. lovaniensis and the underlying mechanisms governing bacterial uptake vary between these amoebal hosts (Venkataraman et al.,; Harb et al.,; Declerck et al.,, ). As a consequence, bacterial uptake can vary between protozoa. Indeed, A. castellanii has been shown to ingest L. pneumophila with much greater efficiency than N. lovaniensis (Declerck et al., ). Variations in sensing, targeting, adhesion and phagocytosis of bacteria can influence the affinity, specificity, frequency and duration with which L. pneumophila interacts with specific protozoa and thus, the impact of their cohabitation on the persistence of L. pneumophila in environmental reservoirs.

19The genetic composition of the bacterium can greatly impact its fate within the host cell, as the survival and replication of different strains and species of Legionella can vary dramatically. Despite the growth defect of L. pneumophila Paris in Willertia magna, both the L. pneumophila Philadelphia-1, Lens and 130b strains are able to replicate in this amoebal host (Dey et al.,; Tyson et al., ). Similarly, comparisons between clinical and environmental isolates of L. pneumophila showed that while one clinical isolate was highly adept at growing in A. lenticulata another was severely defective and the relative amounts of replication of the environmental isolates in this host were somewhere in between (Molmeret et al., ). Similar differences are observed between species of Legionella. While L. pneumophila, Legionella steelei, Legionella dumoffii, and Legionella norrlandica are able to grow within A. castellanii, several other species including Legionella longbeachae, Legionella jordanis, and Legionella anisa are unable to do so (Neumeister et al.,; Edelstein et al.,; Rizzardi et al., ). Thus, the fate of both the bacterium and the host cell is greatly determined by the inherent properties of each organism.

20The outcome of a Legionella-protozoa interaction is not only influenced by their respective identities but the relative abundance of each organism. For instance, when L. pneumophila is present at low levels they are digested for nutrients by Tetrahymena sp. but when the bacteria reach a threshold concentration, they are packaged into vesicles and secreted in pellets (Berk et al.,; Hojo et al., ). The greater the number of bacteria present, the greater the production and secretion of these bacterial pellets. Similar packaging and secretion of other types of bacteria (Denoncourt et al., ) suggests this may be a mechanism by which protozoa compensate for over-eating, or stock-pile food (Hojo et al., ).

21The external environment can have a profound effect on Legionella-protozoa interactions. For example, temperature can greatly impact the intracellular fate of L. pneumophila. Although, intracellular replication of L. pneumophila in A. castellanii occurs at a range of temperatures (Rowbotham, ), intracellular growth is significantly reduced at lower temperatures (Ohno et al., ). Within more restrictive hosts, such as A. polyphaga, intracellular replication only occurs at higher temperatures whereas below 25°C, L. pneumophila is readily consumed (Nagington and Smith, ). In contrast, in Tetrahymena spp. L. pneumophila exhibits robust intracellular growth at 35°C (Fields et al.,; Barbaree et al.,; Kikuhara et al., ) but at lower temperatures, L. pneumophila is packaged into vesicles and secreted into the environment (Faulkner et al.,; Koubar et al., ). The factors affecting intracellular growth of L. pneumophila are not mutually exclusive, as different combinations of the strain of L. pneumophila, the host cell type and temperature can significantly alter intracellular growth of the bacterium (Buse and Ashbolt, ).

22Much of the research examining Legionella-protozoa interactions has focused on specific bacterial-host pairings, which cannot address the impact of other organisms on these interactions. L. pneumophila naturally inhabits complex microbial communities, which could have both positive and negative impacts on L. pneumophila survival and population dynamics. For example, A. castellanii harboring the endosymbiont Neochlamydia S13 are unable to support L. pneumophila replication despite efficient uptake and lack of degradation in the lysosome (Ishida et al., ). The impact of Neochlamydia S13 on L. pneumophila replication is specific because L. pneumophila is able to replicate in A. castellanii infected with the endosymbiont Protochlamydia R18. Moreover, curing A. castellanii of Neochlamyida S13 restores intracellular growth of L. pneumophila, suggesting that the presence of the endosymbiont renders A. castellani resistant to L. pneumophila pathogenesis. In contrast, L. pneumophila has been shown to promote the intracellular growth of Brucella neotomae when the two pathogens share the same vacuole (Kang and Kirby, ). While sharing resources does not appear to affect L. pneumophila, it is conceivable that L. pneumophila may similarly benefit from the activities of other bacteria when it finds itself in more restrictive protozoan hosts.

23A critical challenge in understanding the molecular mechanisms of L. pneumophila pathogenesis, evolution and environmental persistence is the staggering diversity of the protozoan hosts that support L. pneumophila replication. Indeed, such diversity is thought to be responsible for shaping L. pneumophila into a generalist pathogen with a broad host range—a feature clearly important for pathogenesis in humans. Rather than having a single, defined “natural host,” L. pneumophila wanders from host to host and is constantly shaped by these disparate interactions. Such a lifestyle is a challenge for researchers studying these bacteria: (1) many protozoa remain poorly characterized, difficult to culture, and/or unsequenced; (2) the shear diversity of protozoa and complexity of natural interactions makes experimental analysis of phenotypes under “physiologically relevant” conditions extremely daunting (which hosts should be used and under what chemical and physical conditions should the interaction be studied?); and (3) how can non-binary interactions with mixed bacterial and host populations be examined in a reproducible and informative fashion? Given the importance of protozoa to L. pneumophila biology (and pathogen evolution in general), we strongly advocate efforts for the sequencing and detailed study of these organisms. While it is enticing to retreat to the comfort of studying Legionella-host interactions in mammalian macrophages and perhaps one or two model protozoa, an exciting, informative, frustrating, and messy reality remains largely unexplored. Perhaps once the diversity of bacterial-protozoan behaviors is better understood, a panel of model hosts could be chosen not based on ease of culture, but instead to capture the greatest breadth of this diversity.

24TO, DB, AE, and GZ wrote the manuscript. GZ and AE generated the phylogenetic tree.

25The 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.