Commensal and free-living diplomonads have been vastly understudied compared to their pathogenic relatives, despite providing a fascinating framework to study the evolution of these distinct lifestyles. Previously, Xu et al. [30] sequenced the transcriptome of Trepomonas sp. PC1 to explore the origin of a free-living lifestyle in diplomonads. They concluded that Trepomonas likely evolved from parasitic ancestors and re-acquired lost metabolic functions through horizontal gene transfers. We re-examined the hypothesis that the free-living lifestyle is secondarily acquired in diplomonads by inferring a robust phylogenomic framework from the transcriptomes of 14 newly sequenced diplomonad isolates from four genera and one novel lineage. We mapped the life histories of these newly sequenced diplomonad taxa onto this phylogenomic framework and have identified potential signals of parasitic ancestry in the transcriptomes of free-living taxa.

Endobiotic and free-living status of studied diplomonad species and isolates

The endobiotic or free-living status of most diplomonads included in our study is reliable. Giardia spp. and Spironucleus spp. are endobionts, with most being well-studied and characterized as pathogens. Gyromonas ambulans, the novel diplomonad lineage GhostHex, and all Trepomonas spp. (with one possible exception) have been isolated from either freshwater or marine sediments and are free-living. Because the Trepomonas agilis PIG and MIS2C cultures were initiated from a manure pile on a pig farm and a waste water cleaning factory (respectively), it is conceivable that they could be endobionts. However, this seems unlikely, as to our knowledge Trepomonas has never been observed or isolated from any mammalian host. Thus, we are confident that T. agilis is free-living and that these isolates originated from contaminations from the environment.

The normal living environment of Trimitus spp. and Hexamita inflata-like isolates is uncertain because they have been observed and cultured from both gut contents and from sediments. Trimitus sp. FISH was isolated directly from the gut of an unidentified catfish and is clearly endobiotic, as are all but one of the published Trimitus sp. isolates [6]. Trimitus sp. IT1 [6] and numerous other newly isolated Trimitus spp. that possess almost identical SSU rRNA gene sequences have been cultivated from anoxic sediments and are presumed to be free-living (Fig. 3). Hexamita inflata and numerous other Hexamita inflata-like cultures were all initiated from anoxic freshwater sediments and trophozoites can be directly observed in many of these sediments. This taxon is widely considered free-living [42]. However, all H. inflata-like isolates have an almost identical SSU rRNA gene sequence to that of Hexamita sp. observed within and cultured directly from the gut of a horse leech, Haemopsis sanguisuga (described as Hexamita gigas, [43]). In fact, the Hexamita sp. PARU isolate was cultured from sediments collected from the same small pond where the infected horse leech was living. Morphology and SSU rRNA gene sequences are consistent with all of these Hexamita inflata-like isolates belonging to the same taxon. These findings indicate that Trimitus spp. and Hexamita inflata may be capable of existing as facultative endobionts. We are further exploring this possibility (see below for extended discussion).

Fig. 3

Maximum likelihood tree of Diplomonads based on the SSU rRNA gene. The tree was constructed in RAxML using GTRGAMMA as a model and 1000 non-parametric bootstrap replicates (values above the branches, dots denote 100% bootstrap support, all bootstrap values below 50% were removed). The color of the taxa corresponds to the lifestyle (as in the other figures): blue for free-living; orange for host-associated; green for potentially amphizoic. The origin of the isolates is denoted by the icons on the right side of the taxa

A well-resolved phylogeny of diplomonads

Several previous studies have placed free-living diplomonads deep within their parasitic relatives; yet, all of these studies are based on either a single locus, or at best a few genes, and show low resolution and a severe lack of support for the relationships within diplomonads [5, 6, 8]. Only three previous studies have included the published transcriptome of Trepomonas sp. PC1 in their phylogenomic analyses; however, those studies were all focused on other parts of the metamonad phylogenetic tree (Anaeramoeba and Carpediemonas-like organisms [35, 36, 44]). Here, we have vastly increased the amount of available genome-scale data from free-living and commensal diplomonads, and our phylogenomic analyses (Fig. 1A) provide a solid evolutionary framework supporting these previous studies. The closest relatives to diplomonads are endobiotic vertebrate-inhabiting retortamonads. Likewise, the earliest diverging diplomonads are Giardia and Spironucleus spp., which are all endobionts. These data support previous assertions of a host-associated ancestral state for the entirety of Diplomonadida. Consistent with a reversion of habitat and life histories, all free-living taxa, including the monophyletic Trepomonas, are derived members of the crown group of diplomonads. A few endobionts and taxa with ambiguous native habitats are scattered among these free-living taxa in our phylogenetic and phylogenomic trees, rendering it difficult to ascertain the number of reversions to a secondarily free-living lifestyle that have occurred in the evolving history of diplomonads. Notably absent from our phylogenomic analyses are several known endobiotic diplomonad lineages, including Enteromonas spp. and several Spironucleus spp. Inclusion of these taxa in 4-gene and SSU phylogenetic analyses (Fig. 3) produces a topology, whose well-supported nodes are consistent with the phylogenomic tree. However, the unsupported nodes show that future phylogenomic analyses that include these missing taxa might produce topologies suggesting more switches to free-living or even secondarily derived endobiotic lifestyles (Fig. 1A, B).

Hexamita diversity

Within the Hexamitinae, several morphological criteria differentiate members of the genera Hexamita from Trepomonas. While the members of each genera possess two nuclei with an associated quadri-flagellated kinetid, the arrangement of the flagella are dramatically different. Each flagellar apparatus of Trepomonas possesses an anterior flagellum and three trailing flagella that lie in a lateral groove along each side of the cell body [45]. Hexamita does not possess lateral grooves and one posterior flagellum from each kinetid traverses through the length of the cell body in a tube to emerge from the posterior end. Mazancová et al. [8] have shown that organisms commonly identified by light microscopy as Hexamita likely do not form a monophyletic group.

In phylogenomic analyses, our isolate NDL GhostHex branches sister to the Hexamita-Gyromonas-Trimitus clade, while in the four-gene phylogeny it branches sister to the Trepomonas-Hexamita-Gyromonas-Trimitus clade (Fig. 1A, B) and it likely represents a novel diplomonad lineage. The SSU rRNA gene phylogeny, containing the known breadth of diplomonad diversity, shows that NDL GhostHex is a close relative of, or even belongs to, Hexamita clade I as defined by Mazancová et al. [8]. Hexamita clade I branches away from Hexamita inflata, further confirming that NDL GhostHex belongs to a previously unidentified phylogenetic lineage of diplomonads. Indeed, we temporarily had in culture two isolates with a Hexamita phenotype that branch within this environmental clade (Hexamita sp. PC and PC004; Fig. 3). Although the backbone resolution of the SSU rRNA gene tree is poor, organisms with a Hexamita gross morphology branch in several distinctively different positions [8] (e.g., clades with NDL GhostHex, Hexamita sp. SM, and Hexamita nelsoni, Fig. 3). A formal description of NDL GhostHex, including morphology and ultrastructure studied by light and transmission electron microscopy, along with the inclusion of representatives of other Hexamita clades [8] into phylogenomic analyses will likely result in splitting the genus Hexamita into several separate genera.

Evidence of parasitic ancestry of diplomonads

Genomes of parasites contain virulence factors, which are genes that are known or expected to facilitate diverse functions in host-parasite interactions; for example, avoiding the host immune system, attaching to host tissues, or modifying the host environment. If our phylogenetic analyses are correct and the ancestral diplomonad was an endobiont, it is reasonable to expect that some free-living taxa will still retain a fingerprint of the parasitic lifestyle within their genomes. Therefore, we selected 12 known virulence factors from Giardia intestinalis and attempted to identify these genes in all diplomonad datasets. In addition, Giardia spp. possess an adhesive disc that is used to attach to host epithelium. While this disc structure has not been observed in other diplomonads, we have attempted to identify homologs of the ± 200 proteins known to be localized to the disc in Giardia.

Gene presence is easy to interpret; however, it can be problematic trying to infer gene loss from transcriptomic datasets, as the condition of the organisms during RNA isolation might lead to specific gene expression profiles lacking certain genes. Yet, the studied organisms do not have complex life cycles or variable life stages that would lead to massive shifts in gene expression [35, 46]. Nevertheless, we conservatively define a gene as not present when it is missing from the majority of representatives of a particular lineage.

VSPs and CRMPs

Cysteine-rich proteins are very common in diplomonad genomes and there are several categories defined based on their properties: Variant-specific surface proteins (VSPs), Cysteine-rich membrane proteins (CRMPs), Cysteine-rich secreted proteins (CRSPs), etc. Variant-specific surface proteins (VSPs), which are massively duplicated in Giardia and Spironucleus salmonicida, have been experimentally shown to aid in avoiding host immune responses [4, 47, 48]. Their typical structure is a long extracellular cysteine-rich domain, followed by a transmembrane domain and a short intracellular N-terminus with a highly conserved motif. This motif differs from lineage to lineage: CRGKA in Giardia intestinalis, CGRKG in Giardia muris, and [RK][RK]X[RK][RK] in Spironucleus salmonicida. The CRMPs are similar in structure to VSPs, but without the highly conserved motif on the N-terminus [49].

To explore our data for concrete evidence of parasitic ancestry in free-living lineages, we looked for the retention of VSPs across diplomonads and their relatives—Carpediemonas-like organisms, Retortamonas spp., Chilomastix spp., and Barthelona sp. (Fig. 2, Table 1, Additional file 1: Table S1). In our newly sequenced diplomonads, we did not identify any representatives of Dicer or Argonaute proteins, which in Giardia, are assumed to regulate VSP expression. This suggests a different manner of gene regulation, as seems to be the case in S. salmonicida. While all diplomonads possess a sizable number of cysteine-rich proteins, their relatives possess close to none, with the exception of their closest relatives. Specifically, the two Retortamonas strains and Dysnectes brevis possess a small number of CRMPs and one putative VSP protein each. Interestingly, typical VSP proteins appear to be specific to parasitic diplomonads, Giardia intestinalis, G. muris, and Spironucleus salmonicida, which were previously hypothesized to have independently evolved their VSPs, or at least the conserved motifs [18]. The apparent absence of VSPs in the other two Spironucleus species is likely due to a paucity of data, rather than an actual lack of these proteins (S. vortens and S. barkhanus are both represented by highly incomplete datasets). Although we were unable to identify a motif specific to any of the free-living diplomonads, we did find 28 putative VSPs in the transcriptome of Hexamita sp. SM with the [RK][RK]X[RK][RK] motif identical to that of S. salmonicida. Additionally, when we relaxed the motif criteria to include only three [RK] amino acid pairs, the number of putative VSPs in Hexamita sp. SM grew to 99. Comparatively, only between 1 and 3 (3–34 with relaxed motif) putative VSP sequences with this motif were found in the free-living NDL GhostHex, Gyromonas, Trimitus, and Trepomonas species (Table 1). Based on currently available data, it seems that the [RK][RK]X[RK][RK] motif might be ancestral to Hexamitinae diplomonads.

Interestingly, when the numbers of putative VSPs are plotted along the diplomonad phylogeny, the pattern of differential VSP retention and loss appears to reflect the evolution of life strategies (Fig. 2A). The pathogenic Giardia spp. and S. salmonicida possess a large number of VSPs. Hexamita sp. SM, which was isolated from a pond as free-living, but has a nearly identical endobiotic congener ( [43], Fig. 3), retains at least 28 putative VSP sequences. In contrast, members of the free-living Trepomonas clade, Gyromonas ambulans, NDL GhostHex, and Trimitus appear to have lost the majority of their VSPs; the small number of remaining putative VSPs is intriguing evidence of their past presence. Notably, we did find a large number of cysteine-rich membrane proteins (CRMPs) in all diplomonad lineages. The distribution of CRMPs across the tree shows that the massive duplication and presence of cysteine-rich proteins is exclusive to diplomonads. Cysteine-rich proteins may have existed in their ancestors, but in much smaller numbers as compared to diplomonads. Taken together, CRMPs exist in multiple copies in all diplomonads, while the VSPs appear to be more specific to endobiotic and pathogenic species. This pattern suggests that CRMPs likely play a biologically important role across the diplomonads, perhaps in sensing or protection from the outer environment. The lack of expansion or loss of VSP genes in the endobiotic Trimitus spp. suggests, there is a different nature of its host interactions compared to that of Giardia and Spironucleus. Given the conservation of a specific motif in S. salmonicida and Hexamita sp. SM, it seems that VSPs were present in the genome of the Hexamitinae ancestor, but are not being maintained in free-living species, supporting the hypothesis that these lineages are secondarily free-living. Alternatively, the motif existed in the Hexamitinae ancestor functioning in any lifestyle (i.e., more akin to a CRMP protein), which would then have been independently derived in Hexamita sp. SM and S. salmonicida.

Leucine-rich repeat proteins

Hirt et al. [37] identified leucine-rich repeat proteins (BspA-like proteins) on the surface of the parasite Trichomonas vaginalis and suggested that they might be utilized to mediate oxidative stress. We searched our data for all eleven different leucine-rich repeat domains from the Pfam database and identified six of them in several of our datasets (Additional file 2: Table S2). This would be expected, given that these proteins are involved in many cellular functions (for example in Giardia, the cyst wall proteins contain leucine-rich repeat domains). Interestingly, the BspA-like proteins (i.e., containing the Pfam domain LRR_5) are present in high copy number in Trepomonas spp., and several have a recognizable transmembrane domain, which suggests their localization on the cell surface (Fig. 2A, Additional file 2: Table S2). Free-living Trepomonas spp. are likely exposed to oxygen more frequently than endobiotic diplomonads, as an absence of oxygen in their free-living environments is less stable than in a host gut habitat. Therefore, these proteins could play a role mediating oxidative stress, like in T. vaginalis.

Universal and specific virulence factors

We observed two distinct patterns in the retention of analyzed virulence factors across the diplomonads. First, some virulence factors are almost universally present across diplomonads, representing genes that are assumed to act as virulence factors in Giardia, but also possess some other basic metabolic function in free-living organisms. Indeed, these include relatively common housekeeping enzymes—for example, enolase, enzymes involved in the arginine deiminase pathway, and fructose-bisphosphate aldolase—and their presence in almost any eukaryotic cell would be unsurprising. Likewise, thioredoxin and NADP-glutamate dehydrogenase are enzymes involved in protection against reactive oxygen species. Since all diplomonads are anaerobes, these enzymes are expected to exist in free-living species as well. It is likely that most of these enzymes are moonlighting as virulence factors due to the experimental settings—for example, the differences between tissue cultures and a real gut environment—rather than being genuinely involved in pathogenesis.

Secondly, there are a few virulence factors that are present only in Giardia (Cystatin and BPI), or in Giardia spp., Spironucleus, and Hexamita sp. SM (tenascins). Cystatins are likely involved in host immune system modulation [50, 51] and BPI may serve as a regulator of bacterial growth [15, 52]. Cystatin and BPI seem to be specific to Giardia spp., while tenascins are also found in Spironucleus spp., Hexamita sp. SM and a few Trepomonas isolates (see Additional file 7: Fig. S4). Animal tenascins are vertebrate-specific and are involved in cell migration and adhesion [53]. Diplomonad tenascin-like proteins share the EGF (Epidermal Growth Factor) binding domain with vertebrate tenascins, yet they appear to have evolved independently from cysteine-rich proteins [53]. They are hypothesized, without strong cell biological evidence, to bind to EGF receptors of host epithelial cells to help maintain the separation of individual epithelial cells [25, 54]. If for argument’s sake, we consider this to be a specific function of diplomonad tenascins, then they would be differentiated from other putative virulence factors, as they would be directly involved in pathogenicity.

This is consistent with their retention in pathogens, but not in free-living diplomonads. The retention of several copies of tenascins in Hexamita SM might be suggestive of the nature of the organism’s relationship with its potential host. To better understand the distribution of these virulence factors and their impact on the interpretation of lifestyle evolution, it is important to obtain data from other endobiotic/parasitic species. H. nelsoni is especially key to this, as it is a known pathogen that, based on SSU data, branches among the free-living taxa.

As with other virulence factors, many of Giardia’s adhesive disc proteins are well studied and possess additional cellular functions [38, 40, 41]—like tubulins, kinases, and ankyrin repeat proteins; thus, their presence cannot be considered specifically associated with the disc structure. There are 52 such proteins present in all major diplomonad lineages, while 73 proteins localized to the disc appear to be unique to Giardia. Interestingly, there are several disc-localized proteins that are variably present in other diplomonads. Giardin proteins are an illustrative example: alpha-3 and alpha-6 giardin homologs were found in several Hexamitinae diplomonads, while alpha-11 giardin is present only in Hexamita sp. SM and alpha-17 is only in Giardia. From the distribution pattern, it appears that the Giardia disc is composed of both unique and specific proteins to Giardia and repurposed proteins of older evolutionary origins. Knowing the precise molecular/cellular function of some of these proteins in Hexamitinae diplomonads will likely provide further insight into the evolution of the Giardia disc.

HGT candidates and a secondarily free-living lifestyle

In the newly sequenced transcriptomes, we investigated the HGTs that were hypothesized to enable the adaptation to a secondarily free-living lifestyle [30, 55] for their presence/absence, copy number, and likely ancestry. The first group of these genes, those that are likely involved in the digestion of bacterial cell walls, such as lysozyme, NlpC/P60, and murein hydrolase (Fig. 2A, Additional file 6: Fig. S3), appear to be present only in free-living (mostly in Trepomonas spp.) isolates and are present in a large number of copies per isolate, suggesting frequent gene duplication post HGT gain. Phylogenetic analyses of lysozyme and NlpC/P60 genes demonstrate their presence in other metamonads that branch separately from Trepomonas spp.; however, the resolution of the tree is too low to make any strong conclusions about common or independent HGTs (see Additional file 6: Fig. S3). These genes were not present in parasitic Giardia spp. or Spironucleus spp. This, coupled with their high copy number within free-living isolates likely indicates their importance to free-living diplomonads.

Genes encoding proteins involved in the metabolism of nucleotides—such as Ribonucleotide reductase (RNR), which is responsible for the reduction of ribonucleotides to deoxyribonucleotides [56]; Adenosine deaminase (ADA) that can shuttle inosine into purine metabolism making organisms less dependent on scavenging from the host [30]; Deoxyuridine triphosphatase (DUT) that synthesizes the conversion of dUMP into dTMP, the direct precursor of thymidine nucleotides [55]; Uracil phosphoribosyltransferase (UPRT), which creates UMP from uracil, making organisms more flexible in pyrimidine salvage [55]; and Purine nucleoside phosphorylase (PNP), an enzyme of the nucleotide salvage pathway that metabolizes inosine and guanosine to hypoxanthine and guanine, respectively [55] (see Additional file 6: Fig. S3)—tend to be less abundant and their presence/absence among free-living and endobiotic species is spotty. It was previously believed that the eukaryotic class of RNR was lost in the common ancestor of Diplomonadida, and then Hexamitinae regained anaerobic RNR, with subsequent loss in S. salmonicida [55]; however, after extending the taxon sampling, our data show that RNR was likely gained independently by the Trepomonas-Trimitus-Gyromonas-Hexamita clade, Spironucleus vortens, and other representatives of free-living diplomonads. ADA, which has previously been considered an independent gain by Trepomonas, appears to be an ancestral HGT to all Metamonada, followed by differential loss in Giardia and Spironucleus salmonicida. DUT is specifically present in Hexamitinae diplomonads, suggesting its origin by a single HGT event, as suggested previously by Jimenez-Gonzales and Andersson [55]. PNP, which is a key enzyme of the purine salvage pathway, is present across all eukaryotes including diplomonads, whereas UPRT, essential for pyrimidine salvage, is an ancestral HGT to Metamonads and has been lost in the free-living diplomonads.

Anaerobic energy generation

Diplomonads, like many other anaerobic eukaryotes, have adapted to their anoxic lifestyle by anaerobically generating ATP through metabolizing pyruvate to acetyl-CoA to acetate by substrate-level phosphorylation (among other possible pathways). Mitochondrial organelles have been largely reduced in diplomonads. Giardia intestinalis possesses mitosomes that are only known to function in Fe-S cluster synthesis and associated reactions [13, 57]. Giardia also generates ATP through the cytosolic arginine deaminase pathway, which in turn may serve to reduce host innate immunity by depleting arginine [58]. We have identified key enzymes of both anaerobic pyruvate metabolism and the arginine deaminase pathway in most diplomonads. Spironucleus salmonicida seems to possess a more complex mitochondrial organelle than Giardia, as it has been shown to be involved in ATP generation and to generate molecular hydrogen, characterizing it as a hydrogenosome [11]. Organellar ATP generation in S. salmonicida is possible due to the presence of a unique acetyl-CoA synthetase enzyme that is localized to the hydrogenosome. We found orthologs of this hydrogenosomal acetyl-CoA-synthetase in Trimitus, Hexamita sp. SM, and a few Trepomonas isolates (Fig. 2A), indicating its origin in the ancestor of Hexamitinae diplomonads and suggesting that some of them might contain hydrogenosome-like organelles (Additional file 5: Fig. S2).

Evolution of parasitic and free-living lifestyles within diplomonads

If we assume only one direction of lifestyle change within the clade of diplomonads, then the phylogenomic tree presents two possible scenarios for the evolution of parasitic and free-living lifestyles: (1) an endobiotic/parasitic ancestor of diplomonads, followed by three independent transitions to a free-living lifestyle; or (2) a free-living ancestor of diplomonads, followed by up to four independent transitions to an endobiotic lifestyle. A reversal from parasitism to a free-living lifestyle is believed to be an extraordinarily rare, if not impossible event, as the parasitic lifestyle is often considered a terminal evolutionary path [33, 59]. Host reliance commonly leads to genomic modifications that include the loss of essential metabolic pathways, which is thought to render parasites (and other symbionts) incapable of returning to a free-living lifestyle. However, diplomonads represent one of the few lineages where the transition from parasitic back to a free-living lifestyle has been suggested [30].

In this context, two previous key studies explored the evolution of a free-living diplomonad. Xu et al. [30] sequenced and analyzed the transcriptome of Trepomonas sp. PC1, and identified several HGTs unique to Trepomonas sp. PC1 that appear to have helped it overcome the loss of essential functions required for a free-living lifestyle. More recently, Jimenez-Gonzales and Andersson [55] analyzed the gene content of diplomonad genomes and the transcriptome of Trepomonas sp. PC1, and identified several genomic signatures characteristic of parasitism that were present in the last common ancestor of diplomonads, including streamlined genomes and metabolic capabilities. Both studies support the parasitic ancestry of diplomonads. However, our extended taxonomic sampling of diplomonads complicates this story, as it suggests several independent switches from, or to, a parasitic lifestyle. Like the previous studies, our results could support a parasitic ancestry followed by reversals to a free-living lifestyle. Both VSPs and tenascins were likely present in the diplomonad common ancestor, as they are still retained in pathogens and were independently lost/reduced in the free-living taxa (Fig. 2A). Hexamita sp. SM also possesses both VSPs and tenascins and its SSU rRNA gene sequence is nearly identical to that of an endobiont of horse leeches, supporting its likely endobiotic nature. The branching of the VSP and tenascin-lacking endobiotic Trimitus sp. among the free-living taxa further complicates this scenario. Therefore, our results highlight the need for a more nuanced interpretation of transitions between parasitism and free-living lifestyles in diplomonads (Fig. 2B).

When discussing a reversal from parasitism to free-living lifestyle in diplomonads, we must not assume only two distinct states but take into account that the interactions of endobionts with, and the extent of dependence upon their hosts, likely form a continuum [60]. This can range from highly specialized pathogens with complex host interactions, through harmless commensals that simply occupy a unique niche, feeding on the host’s prokaryotic microbiome and having marginal direct interactions with the host, to mutualists that can help modulate the host microbiome or immunity [61]. Moreover, free-living diplomonads reside in anaerobic sediments, which are generally rich in prokaryotes and can experience swift environmental changes. In many respects, such habitats represent a similar ecological niche to the colon of an animal, especially from the perspective of a diplomonad [33]. Together with previous studies, our phylogenomic and comparative analyses, as well as the variability in lifestyles, hosts, and habitats observed across the diplomonads suggest that their ancestor was an endobiont—possibly parasitic, but likely not a highly specialized parasite (Fig. 2B). Giardia and Spironucleus then evolved separately to become highly specialized pathogens, while the ancestors of Trepomonas, Gyromonas, and NDL GhostHex transitioned towards a free-living lifestyle (Fig. 2B). Hexamita spp. and Trimitus spp. either remained endobiotic (and our isolates are artifacts of sampling contaminations), or they might represent amphizoic lineages capable of both free-living and endobiotic lifestyles. While the latter might seem unlikely, it is further supported by the pattern of specific environments from which these diplomonads have been isolated. Giardia and Spironucleus are not commonly detected in sediments, while the free-living Trepomonas spp. are not generally isolated from host environments (with the exception of two reported isolates from insects [8]). On the other hand, representatives from the genera Hexamita and Trimitus have been isolated from both free-living (anaerobic sediments) and host associated environments (Fig. 3). This would suggest that either Hexamita or Trimitus are facultative endobionts/free-living (i.e., can exist as endobionts and as free-living organisms), or that these genera contain species that are endobiotic and others that are free-living. Interestingly, Gyromonas ambulans, which has only been described as free-living, is closely related to Trimitus, suggesting a relatively recent switch in lifestyle preferences.

To better understand the evolution of transitions between parasitic and free-living lifestyles, it will be necessary to obtain complete genome sequences from additional free-living diplomonads to confirm our results regarding gene presence/absence. Since HGTs have been suggested to facilitate expansion of the metabolic potential of secondarily free-living diplomonads, large-scale analyses of horizontally transferred genes will also help resolve this complex evolutionary scenario. Lastly, culturing experiments of potentially amphizoic species in free-living and endobiotic settings along with differential expression analyses would clarify the ability of these species to survive both as free-living and as endobionts.