Dynamic interactions within the host-associated microbiota cause tumor formation in the basal metazoan Hydra

Summary

The extent to which disturbances in the resident microbiota can compromise an animal’s health is poorly understood. Hydra is one of the evolutionary oldest animals with naturally occurring tumors. Here, we found a causal relationship between an environmental spirochete (Turneriella spec.) and tumorigenesis in Hydra. Unexpectedly, virulence of this pathogen requires the presence of Pseudomonas spec., a member of Hydra´s beneficial microbiome indicating that dynamic interactions between a resident bacterium and a pathogen cause tumor formation. The observation points to the crucial role of commensal bacteria in maintaining tissue homeostasis and adds support to the view that microbial community interactions are essential for disease. These findings in an organism that shares deep evolutionary connections with all animals have implications for our understanding of cancer.

Keywords:

Genome analysis – Bacteria – Pseudomonas – Pathogen motility – Microbiome – carcinogenesis – Hydra – Spirochetes

Introduction

Recent studies provided evidence for the deep evolutionary origin of tumor formation [1,2] and the emergence of tumor-related genes at the base of multicellularity [3]. From the beginning of evolution, animals were multiorganismal associations of a multicellular host and microbial community[4]. Microbial symbionts influence virtually all aspects of eukaryote biology [4,5], and their impact on host fitness ranges from detrimental to beneficial, occasionally shifting along this continuum [6]. There is evidence that the presence of certain commensal microbes may turn harmful to the host and cause intoxication [7,8] and inflammation [9]; that some bacteria species including Helicobacter pylori [10,11] and Bacteroides fragilis [12] may promote tumor formation in the host; and that normally harmless commensal bacteria can become pathogenic under stress conditions [9,13]. A previous study has indicated [14] that the natural microbiota provides resistance against mutagen-induced tumorigenesis in mice. Despite the importance of a resident microbiome, the impact of its individual members on tumor development has yet to be rigorously tested.

The early emerging metazoan Hydra is colonized by a stable microbiome [15] and is one of the evolutionary oldest animals with naturally occurring tumors [1] characterized by a differentiation arrest and uncontrolled accumulation of female germline precursor cells. Earlier studies in Hydra largely focused on the innate immune system and identified a crucial role of the transcription factor FoxO [16,17] and antimicrobial peptides [15] for maintaining a specific microbiome. Previous work also suggested that interactions among commensal bacteria are essential to prevent pathogen infection [18] underlining the importance of microbiota diversity as a protective factor against disease. However, nothing is known in Hydra about the role of the resident microbiome in tumor formation. We have discovered that dynamic interactions between a resident bacterium Pseudomonas and an environmental spirochete microbe Turneriella induce tumorigenesis and gravely affect tissue homeostasis and fitness of the host. The findings provide evidence that bacteria-driven tumorigenesis has deep evolutionary roots.

Results

Presence of a spirochete correlates with tumor formation

In previous studies we have shown that Hydra polyps are colonized by a specific microbiota [15,19] and that tissue homeostasis and microbiota composition may depend on each other [17,20]. We therefore hypothesized that the previously characterized bona fide tumor formation in Hydra oligactis [1] could be accompanied by an altered microbiome. To test this hypothesis, we compared the microbiome of tumorous and healthy polyps (further referred to as control) by 16S rDNA sequencing, confocal and electron microscopy. Our analysis revealed that the microbiota of control polyps was dominated by a single bacterium (OTU750018) that can be assigned to the Pseudomonas genus (Fig 1A). These findings were supported by microscopic analysis, where a highly abundant rod-shaped bacterium was consistently detected in the mesoglea (extracellular matrix) of healthy polyps (Fig 1B, S1A Fig). In tumor bearing polyps, the abundance of Pseudomonas (OTU750018) was greatly reduced (Fig 1C) and the rod-shaped microbes were found only rarely. Instead, the microbiota of tumor bearing polyps was enriched with a Turneriella bacterium (OTU4017244) belonging to the spirochete phylum (Fig 1C). Consistent with this, numerous helically coiled bacteria were detected in the mesoglea (Fig 1D, S1B Fig). These findings were strongly supported by a LEfSe [21] analysis which uses the relative abundance of microbes to discover biomarkers that explain the differences between two or more microbial communities (see Methods for details). The Pseudomonas OTU750018 and its higher rank taxonomic categories up to the Proteobacteria phylum were under-represented in the tumorous polyps compared to controls (Fig 1E and 1F), while Turneriella OTU4017244 and all the ranks up to spirochetes phylum were strongly enriched in the tumorous polyps (Fig 1E and 1F). Additionally, two microbes belonging to the Comamonadaceae family (represented by OTU89333 and OTU59) were also enriched in the tumor-bearing polyps compared to control, yet with lower statistical support (Fig 1E and 1F). Since we have no microscopic evidence for the presence of Comamonadaceae in the mesoglea, and since other observations suggested that bacteria from this group are only transiently colonizing Hydra glycocalyx [18], we disregarded these changes in the further analysis.

Phylogenetic analysis of the 16S rDNA provided further insights into the identity of the bacteria colonizing the mesoglea in Hydra oligactis (Fig 1G and 1H). The OTU750018 isolated from control polyps clustered close with the other members of the Pseudomonas species earlier isolated from H. oligactis and H. magnipapillata (Fig 1G, S1 Table). The closest well-characterized relative of this group is P. aeruginosa. For simplicity, we refer further to OTU750018 isolated from control H. oligactis as Pseudomonas. We were able to isolate this bacterium in pure culture for further analysis. The spirochete OTU4017244 identified in the tumor-bearing H. oligactis polyps clustered close to the unculturable bacterium CN-20 (Fig 1H). Not surprisingly, all our efforts to cultivate the OTU4017244 failed. For further molecular and functional analysis we therefore used the most closely related and well-characterized strain Turneriella parva DSM21527 [22] (Fig 1H). Taken together, these results revealed a strong correlation between the presence of the spirochete Turneriella and the tumor formation in H. oligactis.

Spirochetes are necessary for tumorigenesis

To test whether there is a causal relationship between spirochetes and tumor formation in Hydra, we injected a pure culture of T. parva into control H. oligactis polyps (Fig 2A). Four weeks after injection, over 20% of polyps (22 out of 96) had developed tumors (Fig 2A, S2 Fig) which were very similar to the ones observed before (Fig 1C). 16S rDNA analysis and confocal microscopy confirmed that the injected polyps were successfully colonized by the helically coiled T. parva (Fig 2A and 2B). Notably, the abundance of Pseudomonas was greatly reduced in the injected polyps (Fig 2A) suggesting that T. parva may displace the commensal Pseudomonas from its niche in the mesoglea. For the remaining cases (74/96), we could detect neither phenotypic changes nor helically coiled bacteria in the mesoglea nor 16S rDNA sequences corresponding to T. parva (S3 Fig).

To further determine a functional role of microbes in H. oligactis tumor formation, we treated the naturally arising tumor-bearing polyps for two weeks with a cocktail of antibiotics (ampicillin, rifampicin, streptomycin, spectinomycin and neomycin, 50 μg/ml each) and thereby eliminated the spirochetes from the mesoglea (Fig 2C and 2D, S4A Fig). Although feeding with Artemia nauplii introduced some bacteria into the antibiotics-treated polyps (Fig 2C, S4B Fig), no spirochetes could be detected even weeks later (Fig 2C, 2D and 2E). Enrichment of spirochetes and depletion of Pseudomonas in the T. parva injected polyps was statistically supported by LEfSeq analysis (Fig 2F and 2G). Strikingly, all (55/55) tumor polyps recovered a healthy phenotype within two weeks after antibiotic treatment (Fig 2C, S4 Fig) and phenotypically were indistinguishable from control polyps. In a similar manner, antibiotic treatment of tumorous polyps generated by injection of T. parva eradicated the tumors (S4C Fig). These results are consistent with the view that colonization with spirochetes is indispensable to elicit tumor formation in H. oligactis.

Spirochetes cause developmental alterations and fitness loss

The presence of Turneriella OTU4017244 in the naturally arising tumorous polyps caused multiple morphological alterations and developmental disturbances. While in control polyps the ectodermal muscular fibers are highly organized and strictly oriented along the oral-aboral body axis [23] (Fig 3A), in tumorous polyps these actin filaments are greatly disorganized and lay even perpendicular to the main body axis (Fig 3C). Furthermore, consistent with our previous observations [1], numerous female-committed germline precursor cells positive for the periculin marker [24] were present in the tumorous H. oligactis polyps, but absent from control polyps (Fig 3B and 3D). Additionally, in tumor-bearing polyps the thickness of the mesoglea between the ectoderm and endoderm had doubled (S5A Fig) and the total number of epithelial cells per polyp had increased two-fold compared to controls (S5B Fig). Tumor-bearing polyps also had more tentacles (on average 12 per polyp) compared to the control polyps (6 per polyp, Fig 3I and 3J). Consistent with our previous findings [1], asexual growth by budding was retarded in tumor bearing polyps (Fig 3K), most likely due to an increased time of bud detachment (Fig 3L). Strikingly, all morphological and developmental changes associated with the presence of spirochetes were reversed by the antibiotic treatment of tumorous polyps (Fig 2C): actin filaments became properly organized (Fig 3E), periculin-positive germline cells almost disappeared (Fig 3F), number of tentacles (Fig 3J) and growth dynamics (Fig 3K and 3L) returned to values characteristic for control polyps.

To better understand the impact of spirochetes on these morphological changes, we next injected a pure culture of T. parva into healthy control polyps (Fig 3G, 3H and 3J–3L). Remarkably, the addition of this single bacterium was sufficient to trigger all the tumor-specific phenotypic alterations (Fig 2A): actin filaments became disorganized (Fig 3G), germline precursor cells accumulated (Fig 3H), tentacle number increased (Fig 3J), asexual reproduction declined even more than in the naturally tumorous polyps (Fig 3K), and the bud detachment time increased compared to the healthy control (Fig 3K and 3L). Since budding rate and bud detachment time are essential parameters for asexual reproduction, the fitness costs for H. oligactis to carry spirochetes are substantial. The complete restitution of the tumorous phenotype by solely the introduction of T. parva into healthy polyps provides an additional evidence for the essential role of spirochetes in tumor formation.

We also note a consistent increase in the interindividual variation associated with tumor-formation. The clonal culture of healthy H. oligactis demonstrates a very low interindividual variation, evident in minimal deviations in the number of tentacles (Fig 3J), bud detachment time (Fig 3L), mesoglea thickness (S5A Fig) and number of epithelial cells (S5B Fig). This is not surprising, since the number of tentacles and pattern of their emergence as well as the number of cells per polyp are species-specific traits and are genetically controlled [25–27]. In tumorous polyps, interindividual variation is markedly higher compared to the healthy polyps. It is evident in a broad range of tentacle number (Fig 3J), bud detachment time (Fig 3L), mesoglea thickness (S5A Fig) and epithelial cell number (S5B Fig) variance. This variation clearly indicates a loss of a developmental control in tumorous polyps and suggests that tumorigenesis greatly affects morphogenesis. This interindividual variation among tumorous polyps is most likely due to a gradual development of the tumorous phenotype. Supporting this view, the progressive acquisition of the tumorous phenotype with conspicuous tissue bulges and increased tentacle numbers takes several weeks (S2 Fig, S6A Fig). These gradual phenotype changes correlate with a progressive increase in density of spirochetes in the tissue (S6B Fig). Remarkably, the microbiota composition in tumorous polyps is also more variable compared to the healthy polyps (S7 Fig). Particularly, the relative abundance of Gammaproteobacteria, to which the Pseudomonas colonizer belongs, substantially differs between polyps. Taken together, our observations uncover a high interindividual variation associated with the tumor phenotype and strongly suggest that the dynamic interactions within the microbiota are causing this variability.

A commensal bacterium promotes tumorigenesis

The above studied effects indicate that the presence of Turneriella spirochetes is necessary for tumor growth (Figs 2 and 3). To test whether this spirochete bacterium by itself is sufficient in the absence of other microbes to elicit tumorigenesis, we performed two experiments. First, we injected a pure culture of T. parva into germ-free control polyps. Unexpectedly, none of the 158 injected polyps developed tumors (Fig 4A). PCR amplification of T. parva 16S rDNA confirmed that the spirochetes were absent in these polyps (S8 Fig) suggesting that in the absence of other bacteria the spirochetes cannot colonize the hydra tissue. Noteworthy, while Pseudomonas was totally absent in these polyps, some other microbes including Burkholderiales and Sphingobacteriales bacteria (Fig 2C) were present in the injected polyps; they most likely were introduced by feeding the polyps with non-sterile Artemia nauplii. The specific absence of Pseudomonas led us speculate that the presence of this bacterium in the mesoglea of Hydra is necessary for T. parva to settle and subsequently to promote tumor development. To test this hypothesis directly, we first made attempts to colonize the germ-free Hydra polyps with Pseudomonas and subsequently to introduce the spirochetes. However, all efforts to achieve a stable mono-association with Pseudomonas failed for not yet clear reasons.

In a second experiment, we used the tumor-free polyps that spontaneously appeared in the culture of tumorous H. oligactis polyps as a result of budding (Fig 4B). While most of the buds produced by tumor-bearing polyps develop tumors [1], very rarely some buds emerge and detach from tumor-bearing polyps that remain healthy and do not develop tumorous phenotypes over months (Fig 4B). Analysis of the microbiota composition of these tumor-free buds revealed, unexpectedly, that Pseudomonas OTU750018 was virtually absent, with an abundance below 0.5% (Fig 4B), suggesting a spontaneous loss of this bacterium in the developing buds. This is consistent with a high intraindividual variation in Pseudomonas abundance (S7 Fig). Since Turneriella OTU4017244 was even more abundant in these tumor-free polyps compared to the parental tumor-bearing polyps (Fig 4B), the presence of the Turneriella spirochete alone is not sufficient to elicit tumorigenesis, and a substantial amount of Pseudomonas bacteria may be needed for tumor formation. Because the relative abundance of Turneriella OTU4017244 increased significantly in these Pseudomonas-depleted buds compared to the parental culture (Fig 4B), the spirochete seems to occupy niches normally populated by Pseudomonas.

To further elucidate the interplay between Pseudomonas and Turneriella, next we performed transplantation experiments (Fig 4C and 4D). Transplanting tissue with natural microbiota dominated by Pseudomonas and devoid of spirochetes from control polyps into tumor-free polyps densely populated by spirochetes and lacking Pseudomonas (Fig 4C) caused in most cases (78.8%) tumor formation (Fig 4D). PCR amplification of Tureneriella and Pseudomonas 16S rDNA confirmed the presence of the both bacteria in the newly formed tumor-bearing polyps (S9 Fig). Polyps that did not develop tumors were always lacking Pseudomonas bacteria. Grafting tissue fragments of polyps lacking Pseudomonas into tumor-free polyps densely populated by spirochetes only rarely resulted in tumor outgrowth (Fig 4C and 4D). This was demonstrated, first, by transplanting tissue from polyps with a greatly altered microbiota due to antibiotics treatment followed by non-sterile feeding (Fig 4A) into tumor-free polyps densely populated by Turneriella spirochetes (Fig 4C), and second, by transplanting Pseudomonas-free but spirochete enriched tissue from buds into tumor-free polyps densely populated by spirochetes (Fig 4C). Both transplantation approaches resulted in a very low level of tumor formation compared to transplanting Pseudomonas-containing tissue (7.7% and 12.8% compared to 78.8%, Fig 4D). This suggests that the common resident bacterium Pseudomonas plays a crucial and synergistic role in tumorigenesis and somehow interacts with the spirochete Turneriella.

Tumor induction by a phylogenetically distant spirochete, Leptospira sp. from H. circumcincta

Our experiments indicated that the presence of spirochete bacteria is necessary for tumor formation, yet not sufficient. Only if a strain of Pseudomonas colonizes the same niche–the mesoglea of Hydra, the tumor formation is triggered. To provide an additional evidence that the presence of both bacteria in the ECM of Hydra is sufficient for tumorigenesis, we made use of another Hydra species—H. circumcincta. According to our previous studies [15], polyps of this species have spirochetes in their microbiota (S10A Fig). Microscopic analysis confirmed the localization of numerous spirochete cells in the mesoglea (S10B Fig). The phylogenetic analysis of 16S rRNA gene sequence identified this spirochete as a member of Leptospiraceae family, however phylogenetically distant from Turneriella OTU4017244 of H. oligactis and T. parva DSM21527 (S10C Fig). Notably, no Pseudomonas bacteria could be detected in the mesoglea of H. circumcincta host, and tumors have never been reported in this Hydra species. In order to test the tumorigenic potential of the spirochete from H. circumcincta, we introduced this bacterium into the mesoglea of healthy H. oligactis polyps that normally harbor Pseudomonas in their mesoglea (S10D Fig). In 20% of the cases, this “xeno”-transplantation resulted not only in a successful colonization of the foreign spirochetes but also in subsequent tumor formation (S10D Fig). Taken together, these observations provide strong evidence that colonization of the extracellular matrix (mesoglea) of Hydra by both, Pseudomonas strain OTU750018 and spirochete bacteria such as Turneriella OTU4071244, T. parva DSM21527 and Leptospira from H. circumcincta) is sufficient to elicit tumorigenesis. Evidently, in the absence of Pseudomonas in the mesoglea, the tumorigenic potential of the spirochetes cannot be expressed.

Insights from the Pseudomonas and Turneriella genomes

Our results suggest that an interaction between the environmental spirochete bacterium Turneriella and the commensal Pseudomonas drives the tumorigenesis in H. oligactis (Fig 5A–5E). In order to gain insights into the molecular mechanisms of the interactions between the two bacteria, we first sequenced the genome of the resident bacterium Pseudomonas. For this, Pseudomonas OTU750018 was isolated from the mesoglea of H. oligactis control polyps, cultured in vitro, and its genomic DNA was isolated and sequenced using the MiSeq platform (Illumina). We annotated the gene content in the Pseudomonas genome using multiple databases (see Methods, S1 Data, S2 Table). Analysis of the Pseudomonas genome revealed the presence of, among others, genes coding for the entire flagellum assembly (Fig 5E, S11A Fig), type II and VI secretion systems, and the Sec-SRP complex (S12 Fig) as well as multiple ATP-binding cassette (ABC) transporters (S13 Fig), bacteriocins, and toxin/antitoxin systems (S3 Table). These factors are of particular interest since they may be involved in the interaction of Pseudomonas with both the spirochetes and the Hydra host. The flagellum and pili [28–34], diverse secretion systems [35–37] as well as ABC-transporters [38–40] have been previously reported as virulence factors crucial for colonization, persistence and pathogenesis of diverse Gram-negative bacteria, including the closely related P. aeruginosa species. These factors promote bacterial virulence not only by enhancing attachment to eukaryotic cells, but also by mediating the delivery of secreted effector proteins (toxins) from the cytosol of the bacteria into host cells. They may also activate the innate immune response of the host and mediate interactions between bacteria. The discovery of a complete prophage integrated into the Pseudomonas genome (Fig 5E, S14 Fig, S4 Table, S1 Data) points to additional tools to interact with the spirochetes. Furthermore, several genes known to be essential for the formation of outer membrane vesicles (OMVs) on the bacterial surface were found in the Pseudomonas genome (Fig 5E). Consistently with that, high-resolution electron microscopy analysis uncovered conspicuous OMVs on the surface of Pseudomonas bacteria colonizing the extracellular matrix (S15 Fig). Substances released by means of OMVs are known to digest the bacteria environment or even affect the eukaryotic host cells by releasing toxins [41–43] Finally, multiple genes encoding secreted enzymes, such as collagenases and hydrolases, were detected (S2 Table). These findings are consistent with the observation (S1 Fig) that Pseudomonas cells are always located in the mesoglea within lacune-like electron transparent areas free from fibrillary components. Hence, Pseudomonas appears to actively digest the extracellular matrix of Hydra. Taken together, these findings demonstrate that Pseudomonas possesses a rich repertoire of factors that may play a critical role in the interaction with both, the spirochetes and the Hydra host (Fig 5E).

To perform a similar study in Turneriella spirochetes, we analyzed the genome of T. parva DSM21527 that has been previously sequenced [22]. Importantly, this strain is phylogenetically very close to the native Turneriella strain found in H. oligactis tumors (Fig 1H) and is also capable of eliciting the tumorous phenotype in the same manner as the native strain (Figs 2 and 3 and S2 Fig). In contrast to Pseudomonas, the T. parva genome contains relatively few putative virulence factors (Fig 5E). Our analysis uncovered only the genes coding for flagellum assembly (S11 Fig, S5 Table, S2 Data), few bacteriocin-like proteins (S6 Table) and several ABC-transporters (S13 Fig) as well as putative collagenase and hydrolase enzymes. The genes coding for components of the type II secretion system and Sec-SRP complex were only partially represented in the spirochete genome (S12 Fig).

In sum, Pseudomonas and T. parva both possess a variety of putative virulence factors that could mediate the interactions between them and the Hydra host resulting in impaired tissue homeostasis and tumorigenesis (Fig 5E).

Evidence for interaction between Pseudomonas and Turneriella

In order to provide a direct evidence that Pseudomonas and Turneriella do interact, we performed several experiments. First, we estimated the motility of both, Pseudomonas OTU750018 isolate and T. parva DSM21527 strain in a swarming assay (Fig 5F–5H, S16 and S17 Figs). Both bacteria demonstrated a prominent motility on semi-solid agar and, if plated alone, spread as perfectly round colonies with sharp borders (S16A and S16B Fig). Remarkably, if both bacteria were inoculated on the same spot into the agar, the motility of both was significantly increased (S16C and S16D Fig) with the diameter of the T. parva colony being over 60% larger in the presence of Pseudomonas compared to a T. parva colony grown alone (S16C and S16D Fig).

Second, when Pseudomonas OTU750018 and T. parva DSM21527 were spotted onto the same plate at a distance 10 mm between each other the spreading of T. parva colony became strongly biased towards the Pseudomonas culture (Fig 5F, S17A–S17C Fig). The conspicuous growth protrusion directed towards the Pseudomonas colony may indicate a substantially increased motility of the spirochete cells induced by Pseudomonas. Importantly, this phenomenon was not observed in the presence of a sterile-filtered supernatant from Pseudomonas culture (Fig 5H, S17D Fig), indicating that only living Pseudomonas cells can alter the spirochetes’ motility. In the presence of another T. parva culture spotted onto the same culture dish, the motility of T. parva was not affected either (S17E Fig) and both colonies spread as completely circular until they merged in less than 72 hours (S17E and S17F Fig). This clearly indicates that an interaction between two bacteria accompanied by increased motility takes place in vitro.

Last, we monitored the motility of individual Pseudomonas OTU750018 and T. parva DSM21527 cells using phase-contrast microscopy (S18 and S19 Figs, S1–S6 Movies). T. parva demonstrates a relatively slow and undirected swimming motility (S18A Fig, S1 Movie) similar to that described in closely related Leptospira biflexa species [44]. In contrast, Pseudomonas cells display a rather fast directed swimming behavior (S18B Fig, S2 Movie), likely assisted by its flagellum and clearly different from the previously described twitching motility characteristic for P. aeruginosa [45]. Interestingly, in the presence of a Pseudomonas cell, the motility of T. parva changes to a directional migration towards Pseudomonas cells (S19A and S19B Fig, S3 and S4 Movies) resulting in a physical contact between the both bacteria. These interactions were either transient (S19A Fig, S3 Movie), recurrent (S19B Fig, S4 Movie), or permanent, whereby the two bacteria became firmly attached to each other and floated jointly in the same direction (S19C and S19D Fig, S5 and S6 Movies). Taken together, these data provide direct evidence that two bacteria do interact and also alter their behavior upon physical contact.

In order to gain insights into the molecular mechanisms behind these interactions that ultimately may lead to tumor formation, we analyzed the dynamics of gene expression in the both bacteria in the tumor context. First, we performed a metatranscriptome analysis of the tumorous Hydra tissue colonized by both, Turneriella OTU4017244 and Pseudomonas OTU750018, and also of healthy control tissue that harbors only Pseudomonas. Among the top 50 genes differentially expressed by both bacteria in the tumor context, we discovered transcripts for multiple genes coding for flagellar proteins, secretion system assembly machinery, porin-like toxins and enzymes (S7 Table). Interestingly, some other genes present in the genomes of both Pseudomonas and Turneriella (S2–S6 Tables), were not present in the transcriptome and thus are unlikely involved in tumorigenesis. To validate the differential expression of these candidate genes in the tumor context, we performed qRT-PCR experiments (Fig 6). Consistently with the RNA-seq analysis, multiple genes coding for putative virulence factors were up-regulated in the tumorous tissue (Fig 6) including a flagellin gene which was transcribed in Pseudomonas in the presence of Turneriella (Fig 6A). These expression data strongly support our in vivo motility observations (Fig 5F, S16C and S16D Fig, S17 Fig). Additionally, Pseudomonas genes coding for a hydrolase, lipase, a chemotaxis protein as well as a porin were strongly up-regulated in the tumorous polyps (i.e. in the presence of spirochetes) compared to controls (i.e. without spirochetes; Fig 6A). This indicates that these virulence genes are specifically activated in the tumor context and in the presence of spirochetes. Similarly, several transcripts coding for putative virulence factors of Turneriella, such as flagellin, porin, TonB and SecA protein, were specifically enriched in the tumor context, (Fig 6C). Taken together, these observations provide strong support for the view that interaction between Pseudomonas and spirochetes causes a change in behavior and also in gene expression of putative virulence factors.

Stress facilitates spirochete infection

Because tumor formation in H. oligactis is a rather rare event [1] and Turneriella bacteria are normally not found in association with H. oligactis species (Fig 1A) [15] but are reported to be present in tap water [46,47], the normal Hydra epithelium with its resident microbiome appears to efficiently protect against spirochete–induced tumorigenesis. To determine if this natural protection gets lost under environmental stress, we challenged control Hydra polyps with high temperature (22°C for 3 days, Fig 7A). Interestingly, this had a strong impact on the microbiota and resulted in a significantly reduced relative abundance as well as absolute abundance (density) of Pseudomonas in temperature-stressed polyps (Fig 7A and 7B). Subsequent injection of T. parva into the temperature-stressed polyps resulted in much higher colonization rates and tumor outcomes (60%, Fig 7A) compared to injection into intact polyps (20%, Fig 7A) indicating that a reduced number of Pseudomonas cells may open up a niche for the spirochetes to settle. These data also could provide a cue for how the tumor might have originated in the mass culture of H. oligactis over 10 years ago [1]. We assume that an undefined environmental stress may have reduced the density of Pseudomonas in an otherwise healthy and well protected H. oligactis polyps and allowed the ubiquitously present spirochetes to colonize the founder polyp.

Discussion

This study emphasizes how little we know about the role of the resident microbiome in maintaining tissue homeostasis and pathogen defense, a fundamental process that is likely to take place in every tissue of every animal species. Our data have unveiled that Turneriella colonization is essential for triggering the complex tumorous phenotype in Hydra (Figs 1–3). Surprisingly, Turneriella is not sufficient to cause tumorigenesis and the simultaneous presence of resident Pseudomonas bacteria in the same niche, the mesoglea (ECM), is essential for the outbreak of the disease (Fig 4C). Although the genomes of both T. parva and Pseudomonas contain a number of virulence factors (Fig 5E), both bacteria appear to be innoxious if they colonize the H. oligactis mesoglea alone. The pathogenic potential is only realized when both bacteria are present at the same time. Hence, the pathogenicity of Pseudomonas, a natural member of the H. oligactis microbiota, is strictly context-dependent; it is strongly dependent on the presence of the Turneriella spirochete. A similar context-dependent activation of virulence and shift from host-commensal to host-pathogen interaction has been reported previously [48–50] for P. aeruginosa which in the pathogenic state may act on the actin cytoskeleton of the host cells [51,52].

Our observations clearly indicate that Pseudomonas and the spirochetes elicit tumorigenesis synergistically, suggesting a molecular interplay to take place between these two colonizers. Our in vitro motility assays and gene expression analysis provide direct evidence for this interaction to take place. Although the precise mechanism of this interaction remains to be uncovered, secreted molecules (as suggested by the agar motility assay, S17 Fig) and a physical interaction (S19 Fig, S3–S6 Movies) are likely to be involved. Since expression of hydrolases and collagenases is induced when both bacteria are present in the mesoglea (Fig 6, S7 Table), and a local degradation of the extracellular matrix may affect its stiffness and thus the cytoskeletal organization of eukaryotic cells [53–55], one plausible scenario could be that active degradation of the ECM by Pseudomonas and/or Turneriella (S1 Fig) elicits actin fiber disorganization in Hydra’s epithelial cells ultimately resulting in loss of tissue homeostasis and multiple developmental defects (Fig 5E). It also is conceivable that some virulence factors or toxins are released by OMVs (S15 Fig) or directly delivered into the host cells through a secretion system [56]. In sum, here we show that in the holobiont Hydra [57] interactions between two bacteria, Pseudomonas and Turneriella, have profound effects on tissue homeostasis and fitness. Considering that Turneriella is an environmental bacterium present in pond water, these findings highlight the role of environment as a source (reservoir) of opportunistic pathogenic microbes, that, upon favorable conditions, may colonize a novel habitat, i.e. the multicellular host, and cause adverse effects on its development, physiology and fitness.

A causal connection between microbiota and cancerogenesis is being increasingly appreciated [58,59]. Several tumors in humans are strongly associated with pathogenicity of a single bacterium, such as Helicobacter pylori [60], or are likely caused by interactions between two commensal bacteria, like Bacteroides fragilis and Escherichia coli [12]. Additionally, there is an accumulating evidence that commensal bacteria may support tissue homeostasis by, for instance, stimulating neurogenesis [61,62]. These observations point to the evolutionary conserved role of the host-associated bacteria in maintaining host tissue homeostasis. The basal phylogenetic position of Hydra allows considering the tumorigenic potential of bacteria in an evolutionary context. Better understanding of the protective function of the resident microbiome in Hydra may thus help refine current concepts of tumorigenesis, with important implications for early diagnosis and therapy. Last, our findings may also stimulate thoughts about the link between microbiota, antibiotics and cancer. Antibiotics not only have substantial direct and indirect effects onto microbiota [63,64], they may also represent a novel strategy for prophylaxis and/or treatment of certain cancer types.

Materials and methods

Animals and culture conditions

Experiments were carried out using H. oligactis strain St. Petersburg. Animals were maintained under constant environmental conditions, including culture medium (HM), temperature (18°C) and food according to standard procedures [65] (twice a week). The propagation of the polyps occurs by only asexual reproduction resulting in clonal cultures. Natural occurrence of tumor polyps has been already described [1]. Tumor buds have been separated long time ago from the original cultures and been kept in tumor only cultures. The fitness loss in tumor polyps is still recognizable but constant feeding results in slow and ongoing asexual reproduction. The majority of the buds attain the tumor phenotype after around 4 weeks. In rare cases, buds do not exhibit the tumor phenotype and remain healthy. These specific buds were separated and kept apart from the parental cultures. In this study, work has only been performed with the H. oligactis species. Pelmatohydra robusta tumor cultures are intact but have not been of interest in this work.

Bacteria culture conditions and generation of germ-free Hydra

To obtain germ-free (gf) H. oligactis, polyps were kept for 2 weeks in an antibiotic solution containing 50 μg/ml each of ampicillin, rifampicin, streptomycin, spectinomycin and neomycin with exchange of the solution every second day. Post treatment polyps were transferred into sterile-filtered HM for one week which was changed every second day. The absence of bacteria was verified by a 30-cycle PCR using the universal 16S bacteria Primer EUB_27F and EUB_1492R [66], whereas the positive control of none-treated (DMSO) polyps showed a signal. Because T. parva was uncultivable, the closest relative T. parva [46] DSM 21527 was ordered and kept in Leptospira semisolid medium which was produced and handled based on the manufactures protocol (Difco Leptospira Medium Base EMJH & Difco Leptospira Enrichment EMJH; Thermo Fisher Scientific, USA). T. parva was inoculated twice a week to new media to keep the culture alive.

Isolation of the mesoglea

Isolation of the Hydra mesoglea was performed as previously described [67]. Adapted from the protocol IGEPAL CA-630 was used (1:1000 in ddH₂O, Sigma-Aldrich, Germany) to separate the mesoglea from the Hydra cells. Isolated mesoglea was used for the isolation of Pseudomonas and SYBR-gold stainings.

Isolation of T. parva

We attempted to isolate a pure culture of Turneriella OTU4017244 from tumorous H. oligactis polyps. Although we tested a number of conditions, including the temperature, oxygen concentration, medium composition and addition of selective antibiotics, we never succeeded in obtaining a pure culture. Therefore, we used a closely related culturable strain T. parva DSM21527 [68] for functional analyses. The bacteria were cultivated in the semisolid Leptospira Base and Enrichment Medium (Difco, purchased from Thermo Fisher Scientific, USA) according to the manufacturer’s instruction. Optimal growth was achieved at 29ºC. To maintain an active T. parva culture, the bacteria were re-inoculated into a fresh medium every week.

Isolation of Pseudomonas

Pseudomonas OTU750018 was isolated from the ECM of Hydra oligactis control polyps. First, H. oligactis polyps were treated with IGEPAL CA-630 [67] to eliminate all bacteria present on the outer surface of the polyps, referred to as glycocalyx, and thus to facilitate the isolation of a clean Pseudomonas strain. Further, the mesoglea was inoculated onto R2A-Agar (Carl Roth; Karlsruhe) plates. Finally, Pseudomonas OTU750018 bacteria were cloned by repeated plating of single colonies onto R2A-Agar and identified by 16S rDNA Sanger sequencing (IKMB, Kiel).

Injection of T. parva in H. oligactis

Polyps of H. oligactis were injected with a pure culture of T. parva DSM21527. The concentration of T. parva cells was adjusted to 20,000 bacteria per 1μl using a C-Chip Disposable Hemocytometer DHC-N01 (NanoEnTek, USA) and methylene blue cell staining. We injected 5μl of T. parva suspension containing 100,000 cells into hydra tissue with a fine needle pulled from a glass capillary and connected to a rubber tube with a sterile 0.2 μm filter (Sarstedt, Germany). Polyps were kept in the bacterial medium for 24h and subsequently moved into new dishes with fresh HM. Phenotypes were observed after 4–6 weeks post injection.

DNA extraction and sequencing of 16S rRNA gene

For total DNA extraction, single polyps (six replicates for each treatment) were washed three times in sterile Hydra-medium and subjected to the DNeasy Blood & Tissue Kit (Qiagen, Germany). Extraction was performed following the manufacturer’s protocol, except that DNA was eluted in 30 μl and kept at -20°C until sequencing. Variable regions 1 and 2 (V1V2) of the bacterial 16S rRNA genes were amplified using the primers 27F and 338R [69]. Bacterial 16S rRNA profiling was performed in the Illumina MiSeq platform with paired-end sequencing of 2x 300bp. The 16S RNA sequencing raw data are deposited at the SRA and are available under the project PRJNA602941. Sequence analysis was executed by using QIIME 1.9.0 package (RRID:SCR_008249) [70]. SeqPrep was used to assemble Paired-end reads (RRID:SCR_013004), chimeric sequences were screened with ChimeraSlayer (RRID:SCR_013283) [69] and manually verified prior to removal from the data set. Sequences were kept in the analysis if they have been present at least in two independent samples.

Bacterial community analysis

Operational taxonomic units (OTUs) picking was performed using the pick_open_reference_otus.py script with at least 97% identity per OTU. Annotation was conducted using UCLUST algorithm (RRID:SCR_011921) [71] was used to operate the annotation against the GreenGenes database v13.8 (RRID:SCR_002830) [72] implemented in QIIME. To avoid false positive OTUs originated from sequencing errors OTUs with <50 reads were removed from the data set [73]. Read number was normalized to the lowest number of reads in the dataset (8,000). The tables of bacterial abundance were further processed using Linear discriminant analysis effect size (LEfSe) analysis (RRID:SCR_014609) [21] to identify bacterial taxa that account for major differences between microbial communities in tumorous and healthy polyps. To detect bacterial taxa with significant differential abundance, Kruskal–Wallis and Wilcoxon signed-rank sum tests were implemented, and the bacteria were ranked by effect size obtained from linear discriminant analysis (LDA score). P-values were corrected for multiple hypotheses testing by Benjamini and Hochberg’s false-discovery rate correction (q-value). A q-value of 0.001 and an effect size threshold of 4.0 (on a log₁₀ scale) were used for all comparisons discussed in this study. The results of LEfSe analysis were visualized by plotting the phylogenetic distribution of the differentially abundant bacterial taxa on the Ribosomal Database Project (RDP) bacterial taxonomy.

Statistics

Statistical analyses for the 16s rRNA data were performed using two-tailed Student’s t-test or Mann-Whitney U-test if applicable. If multiple testing was performed, p-values were adjusted using the Benjamini-Hochberg correction [74]. Statistics for the fitness measurements were performed using one-way-ANOVA (Kruskal-Wallis test) and an additional Dunnett’s multiple comparison posttest comparing all values to the control.

Sequencing and annotation of bacterial genomes

For sequencing the Pseudomonas OTU750018 genome, a pure culture of this isolate was grown overnight in R2A medium, washed in sterile PBS and subjected to the DNeasy Blood & Tissue Kit (Qiagen, Germany). Nextera XT kit (Illumina) was used for library preparation and paired-end sequencing was conducted on the MiSeq platform (Illumina) at Centre for molecular biology in Kiel. Genome was assembled with Spades 3.12 using default settings. Further, we annotated this genome of Pseudomonas OTU750018 as well as the previously sequenced and assembled genome of T. parva DSM21527 (PRJNA50821) [22]. Several complementary approaches were used to annotate the assembled genome sequences. First annotation was performed with Dfast_core (v.1.0.8). The genes were annotated by aligning the genome sequence with the data previously deposited in diverse protein databases including the National Center for Biotechnology Information (NCBI; https://www.ncbi.nlm.nih.gov/) non-redundant protein (Nr) database, UniProt/Swiss-Prot (https://www.uniprot.org/) and Kyoto Encyclopedia of Genes and Genomes (KEGG; https://www.kegg.jp/ [75]). Additional annotation was carried out using the following databases: Virulence Factors of Pathogenic Bacteria (VFDB; http://www.mgc.ac.cn/VFs/main.htm [76]) and antibiotics and Secondary Metabolite Analysis Shell (anti-smash; https://antismash.secondarymetabolites.org/ [77]). Prophages were identified using the PHAge Search Tool (http://phast.wishartlab.com/, [78]). Nr and GO annotation was carried out using Blast2GO. An E-value of 1e−5 was used as the cut-off for all basic local alignment search tool.

Metatranscriptome sequencing and analysis

To identify genes of Pseudomonas and Turneriella transcribed in the H. oligactis tumor, total RNA was isolated from whole polyps as previously described [1]. After a ribosomal RNA depletion, 15 cDNA libraries (5 for each control, female and tumor animals) were generated and sequenced on the Illumina HiSeq2500 v4 platform, with 125 bp paired-end sequencing of 12 libraries per lane. This resulted in 13–24 million reads per sample after quality control. Quality and adapter trimming was performed using Trimmomatic 0.36 [79]. The RNA-Seq raw data are deposited at the Sequence Read Archive (SRA) and are available under the project ID PRJNA602941. Bowtie2 2.3.5 [80] was used to map the RNA-reads against the gene models of an in-house assembly of the Pseudomonas OTU750018 and the publicly published T. parva DSM21527 genome (PRJNA50821) [22]. The analysis was performed by mapping reads against a combined reference comprising gene models of both Pseudomonas and Turneriella genomes. Mapping rates ranged around 3% in all samples, resembling the fact that most of the sequenced reads belong to the Hydra transcriptome. All downstream analyses were conducted in ‘R’ software [81]. Differentially expressed contigs were identified with the package DESeq2 1.16.1 [82].

Quantitative real-time PCR analysis (qRT-PCR)

In order to validate the differential expression of putative virulence genes of Pseudomonas and T. parva in the tumor, we performed quantitative real-time PCR analysis. Total RNA was extracted from H. oligactis polyps and converted into cDNA using First Strand cDNA Synthesis Kit (ThermoFisher Scientific) according to manufacturer’s instruction. Amplification was performed as previously described [83] using GoTaq qPCR Master Mix (Promega, Madison, USA) and specific oligonucleotide primers (S8 Table). Three biological replicates of healthy control H. oligactis, tumorous polyps and polyps grown from buds without tumor were analyzed with two technical replications. The data were collected using ABI 7300 Real-Time PCR System (Applied Biosystems, Foster City, USA) and analyzed by the conventional ΔΔCt method. To estimate the stress-induced changes in abundance of Pseudomonas OTU750018, genomic DNA was isolated from 6 biological replicates of H. oligactis incubated at 18°C or stressed at 22°C for 3 days, and from tumor H. oligactis polyps. Real-time amplification was performed as previously described using GoTaq qPCR Master Mix (Promega, Madison, USA) and oligonucleotide primers specific for 16S rDNA or eubacteria or strictly Pseudomonas OTU750018 (S8 Table). Samples were equilibrated to the actin of H. oligactis and to the overall bacterial load of each polyp by Eub-primers.

Phylogenetic analysis

To identify the phylogenetic relations of Pseudomonas and spirochetes bacteria, we compared their 16S rRNA sequences to previously published sets [22,84]. Sequence alignment was generated using Clustal_x [85] and the phylogenetic tree was inferred using MEGA 6 software [86]. A model-test was used to estimate the best fit substitution models for phylogenetic analyses. For the maximum-likelihood analyses, genes were tested using the General Time Reversible (GTR + I) model. A bootstrap test with 1000 replicated for maximum likelihood and random seed was conducted.

Bacteria motility assay

In order to test the motility of T. parva DSM21527 and Pseudomonas OTU750018, we inoculated 40,000 bacterial cells in 2 μl culture medium into semi-solid agar plates (3.5% Agar; Difco Leptospira Medium) and incubated at 29°C for 72 h. Bacteria cell number was counted using a Neubauer chamber and adjusted to 20,000 bacteria per μl. In order to test the interaction between T. parva DSM21527 and Pseudomonas OTU750018, the bacteria were inoculated either onto the same spot or 1 cm apart into semi-solid agar plates and incubated as described above.

Transmission electron microscopy

To visualize bacteria in the mesoglea of Hydra at high resolution, we used transmission electron microscopy (TEM) analyssis of thin sections of healthy and tumorous H. oligactis polyps. The sections were prepared following the previously published protocol [87]. TEM was done using Tecnai G² Spirit/BioTWIN (FEI Company, Thermo Fisher Scientific, USA).

Immunohistochemistry

Immunohistochemical detection of female germline marker, the periculin protein, in whole mount Hydra preparations was performed as described [24] using a polyclonal mouse antiserum against periculin 1a protein (1:500 diluted produced by Bosch lab) and Alexa488 conjugated donkey–anti-mouse secondary antibodies (2 μg/ml; Invitrogen, USA). Phalloidin staining was conducted as described previously [88]. Detection of Pseudomonas and T. parva on the mesoglea was performed using SYBR-gold staining. Isolated mesoglea fragments were rinsed in sterile ddH₂O and stained with SYBR-gold (1:20,000 in ddH₂O; Thermo Fisher Scientific) for 3 min in the dark at room temperature. After a brief washing, in ddH₂O, the mesoglea was embedded in Moviol-Dabco [89]. Confocal microscopy images were taken using TCS SP1 laser-scanning confocal microscope (Leica, Germany).

Mesoglea thickness measurement

To measure the thickness of the mesoglea in healthy and tumorous polyps, individual hydras were relaxed in 2% urethane and fixed with 4% formaldehyde, dehydrated in ethanol and embedded into LR-White resin (Ted Pella, Redding, CA, USA) according to the manufacturer’s instruction. Semi-thin sections (0.5 μm thick) were cut using Ultracut S ultratome, mounted on slides and stained with methylene blue/azur II as described previously [1]. Average thickness of mesoglea was measured on 10 random locations of each section, and twelve sections for each condition (control and tumor) were evaluated.

Cell number quantification using flow cytometry

To estimate the number of epithelial cells per polyp, we disintegrated healthy control and tumorous hydras and subjected to flow cytometry analysis following previously established protocol [27]. For each replicate, individual polyps were digested in 100 μl of 50 U/ml Pronase E (Serva) in an isotonic culture medium for 4 h at 18 °C. Living cells were counted on BD FACSCalibur with CellQuestPro v5.2 (Becton–Dickinson) using forward scatter and side scatter parameters. Gating and further analyses were performed with FCSalyzer 0.9.13-alpha (https://sourceforge.net/projects/fcsalyzer/).

Fitness assay

To assess the growth rate of healthy and tumorous H. oligactis polyps, single polyps served as founder for a clonal population. Progeny of this polyp and all other following offspring’s were placed individually into wells of 12-well plate until and fed daily ad libitum. The number of polyps per well was recorded daily until at least 100 polyps were generated. For population growth rate calculation the experiment was split in sub experiments at the second generation. Since Hydra grows clonally and the genotype is the same for one line, the resulting sub-experiments were treated as independent replicates. For growth rate calculation a regression of the log²-transformed polyp number per time unit was performed and the slope of the curve as descriptor for growth rate was used. For data analysis a MySQL-database as well as custom written python and R functions were utilized.

Transplantation experiments

In order to introduce Pseudomonas into tumor-free H. oligactis polyps we performed lateral transplantation experiment. Fragments of the middle body column were excised from healthy H. oligactis polyps, antibiotic-treated polyps devoid of Pseudomonas, or from tumor-free buds. The tissue fragments were transplanted onto tumor-free H. oligactis polyps as previously described [1]. In contrast to the original protocol, the grafted tissue fragment was removed 12 h after the transplantation in order to avoid migration of the cells from the transplant into the recipient tissue. In a similar way, spirochetes from H. circumcincta were introduced into healthy H. oligactis polyps.

Ethics statement

No ethical approval is required to perform experiments on the pre-bilaterian invertebrate Hydra oligactis.

Supporting information

S1 Fig [a]
Presence of and on the mesoglea of .

S2 Fig [tif]
Injection of results in tumor formation.

S3 Fig [tif]
Low presence of in healthy polyps injected with spirochetes.

S4 Fig [a]
Antibiotic treatment results in complete elimination of spirochetes from tumorous polyps and obliterates the tumor.

S5 Fig [a]
Tumor growth is accompanied by morphological and developmental alterations.

S6 Fig [a]
Gradual acquisition of the tumor-specific phenotype in polyps injected with .

S7 Fig [tif]
Intraindividual variation in microbiota composition among healthy control and tumorous polyps.

S8 Fig [tum]
is not able to settle onto the antibiotic-treated polyps.

S9 Fig [t]
Introduction of into the mesoglea densely populated by via tissue grafting consistently induces tumors.

S10 Fig [a]
harbors spirochetes in the mesoglea that are capable to induce tumors in .

S11 Fig [a]
The flagellum assembly system encoded in the genomes of OTU750018 and DSM21527.

S12 Fig [a]
The bacterial secretion systems encoded in the genomes of OTU750018 and DSM21527.

S13 Fig [a]
Repertoire of ABC-transporters encoded in the genomes of OTU750018 and DSM21527.

S14 Fig [tif]
The genome of OTU750018 carries a complete prophage sequence.

S15 Fig [omvs]
OTU750018 produces outer membrane vesicles.

S16 Fig [a]
Motility of OTU750018 and DSM21527.

S17 Fig [a]
Motility of DSM21527 increases in the presence of living OTU750018 cells.

S18 Fig [a]
DSM21527 and OTU750018 demonstrate different motility patterns.

S19 Fig [a]
Interaction between DSM21527 and OTU750018.

S1 Table [xlsx]
Accession numbers of 16S rDNA sequences used for phylogenetic analysis of and spirochetes.

S2 Table [xlsx]
Annotation of genes in the genome of OTU750018 isolate.

S3 Table [xls]
Putative secondary metabolites produced by deduced by ANTI-SMASH analysis of Pseudomonas OTU750018 genome.

S4 Table [xlsx]
Annotated phage genes detected in the genome sequence of OTU750018 and identified using PHAST tool.

S5 Table [xlsx]
Annotation of genes in the DSM21527 genome.

S6 Table [xlsx]
Putative secondary metabolites produced by the spirochetes deduced by ANTI-SMASH analysis of DSM21527 genome.

S7 Table [xlsx]
Top50 genes of and differentially expressed in the tumor. Genes validated in qRT-PCR are highlighted green.

S8 Table [xlsx]
Oligonicleotide primers used to amplify bacterial 16S rDNA genes and virulence gene in qRT-PCR.

S1 Data [zip]
Full genome sequence of OTU750018 isolate.

S2 Data [zip]
Full genome sequence of DSM21527.

S1 Movie [avi]
Motility of DSM21527.

S2 Movie [avi]
Motility of OTU750018 isolate.

S3 Movie [avi]
Interaction of DSM21527 and OTU750018: Directed swimming of towards Pseudomonas and transient contact.

S4 Movie [avi]
Interaction of DSM21527 and OTU750018: Circular swimming of around and recurrent contact.

S5 Movie [avi]
Interaction of DSM21527 and OTU750018: Stable contact and aggregated swimming of around .

S6 Movie [avi]
Interaction of DSM21527 and OTU750018: Stable contact and aggregated swimming of around .