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

Abstract

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.

Fig 1. Presence of spirochetes in microbiota of H. oligactis correlates with tumor formation.

(A) Phenotype of healthy H. oligactis polyps is defined by a tube-like body column and a specific microbiome dominated by Pseudomonadales; scale bar: 0.5 cm. Inset: Relative abundance of bacteria identified by 16S rDNA sequencing and resolved at the phylum level, n = 6. (B) Rod-shaped Pseudomonas bacteria colonize the ECM of Hydra. Bacteria are visualized on isolated mesoglea with SYBR-gold staining (scale bar: 5 μm) and on TEM-sections (inset, scale bar: 0.5 μm). (C) Tumor-bearing polyps of H. oligactis are characterized by tissue bulges and greatly altered microbiome with dominated by Leptospirales (scale bar: 0.5 cm). Inset: Relative abundance of bacteria identified by 16S rDNA sequencing and presented on the phylum level, n = 6. (D) Helical bacteria colonize the ECM of tumor-bearing H. oligactis, visualized on isolated mesoglea with SYBR-gold staining (scale bar: 5 μm) and on TEM-sections (inset, scale bar: 0.5 μm). (E) Taxonomic cladogram presenting OTUs differentially represented in the microbiota of control and tumor polyps generated using LEfSe analysis of 16S rDNA sequencing data; red—taxa enriched in tumor polyps; green–taxa enriched in healthy polyps. (F) The enrichment of certain taxa in tumor polyps is statistically supported by high positive LDA score values (red), and the taxa strongly enriched in control polyps are supported by negative score values (green); n = 6 (G) Phylogenetic tree of Pseudomonas species clusters the isolate OTU750018 from healthy H. oligactis together with other Hydra-associated Pseudomonas isolates close to the P. aeruginosa species. Neighbour-joining phylogram with numbers at nodes representing bootstrap support values calculated by 1000 iterations. (H) Phylogenetic tree of Leptospiraceae family clusters the spirochetes OTU4017244 isolated from tumor-bearing H. oligactis with an unculturable clone CN-20 and the reference T. parva strain DSM21527. For sequence accession numbers see S1 Table. Neighbour-joining phylogram with numbers at nodes representing bootstrap support values calculated by 1000 iterations.

Fig 2. Spirochetes are necessary for tumorigenesis in H. oligactis.

(A) Injection of a pure T. parva culture into healthy polyps (scheme, left) results in acquisition of tumorous phenotype (right) within 4 weeks post injection. Injected polyps show similar tissue outgrowth and Leptospirales-dominated microbiome as natural tumor polyps; scale bar: 0.5 cm. Inset: Relative abundance of bacteria identified by 16S rDNA sequencing and presented on the phylum level, n = 6. (B) T. parva densely colonizes the mesoglea of the injected polyps, revealed by SYBR-gold staining (scale bar: 5 μm). (C) Removal of the spirochetes from the tumorous polyps by antibiotic treatment (scheme, left) results in recovering of normal phenotype within 2 weeks. The antibiotic-treated polyps (right) show normal body shape and absence of Leptospirales in the microbiome. Inset: Relative abundance of bacteria identified by 16S rDNA sequencing and presented on the phylum level, n = 6. (D) The mesoglea of antibiotic-treated polyps shows no presence of spirochetes revealed by SYBR-gold staining (scale bar: 5 μm). (E) Relative abundance plots of the microbial composition of healthy (control) and antibiotic-treated (antibiotic+feeding) polyps, naturally tumorous polyps (tumor) and tumor-bearing hydras resulted from T. parva injection (injected) on the bacteria class level. (F) Taxonomic cladogram presenting OTUs differentially present in the microbiota of control and T. parva-injected polyps generated using LEfSe analysis; red—taxa enriched in injected polyps; green–taxa enriched in intact control polyps. (G) The enrichment of certain taxa in T. parva-injected polyps is statistically supported by high positive LDA score values (red), and the taxa strongly enriched in intact control polyps are supported by negative score values (green); n = 6.

Fig 3. Spirochetes cause developmental alterations and fitness loss in Hydra.

(A) In the control polyps (ctrl), actin fibers in the ectoderm are organized parallel to the polyp body axis, revealed by phalloidin-rhodamine staining, scale bar: 10 μm. (B) No germline precursor cells can be detected in control polyps using anti-periculin immunostaining. (C) In the naturally-occurring tumorous polyps (tum), actin fibers are disorganized, scale bar: 10 μm. (D) Periculin-positive germline precursor cells accumulate in the gastric region of tumor polyps, consistent with the previous observations (Domazet-Loso et al., 2014). (E) Antibiotics treatment (antib) of tumorous polyps reverts the cytoskeleton structure to normal. (F) In antibiotics-treated polyps, the density of periculin-positive germline cells declines. (G) In the polyps injected with T. parva (inj), actin cytoskeleton is disorganized similar to naturally-occurring tumors. (H) Numerous periculin-positive germline cells appear after injection of T. parva into control polyps. (I) Tumorous H. oligactis polyps have higher number of tentacles compared to control animals. (J) Tentacle number is significantly higher in tumorous polyps (tum) compared to controls (ctrl). Antibiotics treatment reduces the number back to the normal level, while T. parva injection increases the tentacle number almost to the tumor polyp level. ***—p<0.001, *—p<0.05. (K) Population growth rate measured as number of buds generated in 25 days is reduced in naturally tumorous polyps and T. parva-injected polyps compared to the healthy controls and antibiotic-treated polyps. (L) Bud detachment time is increased in the tumorous polyps (tum) compared to healthy controls (ctrl). Antibiotics treatment of tumorous polyps (antib) brings detachment time to the normal level, and T. parva injection increases it significantly. ***—p<0.001, *—p<0.05.

Fig 4. Spirochetes are not sufficient to cause tumorigenesis.

(A) Presence of some bacteria on Hydra is necessary for T. parva to colonize the host. T. parva injected into antibiotic-treated polyps free of bacteria were not able to settle, and neither of injected polyps (0/158) developed tumorous phenotype. (B) In rare cases, tumor-bearing polyps generate buds that result in healthy tumor-free polyps. These polyps harbor a microbiota dominated by spirochetes, but devoid of Gammaproteobacteria, indicating the loss of Pseudomonas. Insets: Abundance plots of the microbial composition of parental tumorous polyps (tumor) and spontaneously recovered polyps (bud without tumor) on the bacteria class level. (C) Setup of the transplantation experiment to prove that both bacteria, T. parva and Pseudomonas, are necessary and sufficient to elicit tumorous phenotype in H. oligactis. A tissue fragment from a healthy donor polyp (control) provides a source of Pseudomonas. This graft, or the grafts from an antibiotic-treated polyp devoid of Pseudomonas and spirochetes (antib.+feeding), or a tissue fragment from a spontaneously recovered polyp enriched in spirochetes but free from Pseudomonas (tum. bud without tumor) are transplanted onto tumor-free recipients devoid of Pseudomonas but harboring spirochetes. The tumor formation in resulting polyps is evaluated. (D) While only a small fraction of the recipients grafted with only T. parva (red), or no bacteria (white) on the ECM develop tumors, the recipients with grafts from tissue with Pseudomonas (green) have the highest rate of tumor formation.

Fig 5. Interplay between the commensal Pseudomonas and environmental spirochetes within Hydra mesoglea induces tumor formation.

(A) In healthy H. oligactis polyps the mesoglea (ECM) is colonized only by Pseudomonas, actin cytoskeleton of the epithelial cells (red lines) is well organized and no developmental abnormalities can be observed. (B) In the polyps that spontaneously lost Pseudomonas, the mesoglea is colonized only by T. parva, and the phenotype is normal. (C) If the both bacteria, Pseudomonas and Turneriella, are present in the ECM of a polyp, an interaction between them occurs and likely causes actin fiber disorganization (red curved lines) and tumor outgrowth. (D) In the tumor-bearing polyps, both bacteria–rod-shaped Pseudomonas and helical-coiled T. parva, are found in close proximity in the ECM (revealed by SYBR-gold staining; scale bar: 5 μm), suggesting an direct physical interaction. (E) Analysis of genome sequences from the Pseudomonas OTU750018 isolated from H. oligactis and T. parva reference strain DSM21527 uncovers their repertoires of putative virulence factors. In Pseudomonas genome (left), genes coding for the entire flagellum assembly, type II and VI secretion systems (Type II & VI SS) and the Sec-SRP complex are detected along with multiple ABC-transporters (ABC-Tp), bacteriocins, and toxin/antitoxin systems. Additionally, a complete prophage (Phage) is integrated into the Pseudomonas genome. Electron microscopy analysis (S15 Fig) also suggests that Pseudomonas cells release outer membrane vesicles (OMVs). The genome of T. parva (right) harbors genes coding for flagellum assembly machinery, few bacteriocin-like proteins and several ABC-transporters (ABC-Tp), few genes coding for components of type II secretion system (Type II SS) and Sec-SRP complex. We speculate that the interplay between two bacteria (arrows) and some virulence factors produced by them affect the H. oligactis cells, their morphology (including the actin cytoskeleton depicted as red curved lines), disturb the tissue homeostasis and cause tumor formation. (F) Evidence for direct interaction between T. parva and Pseudomonas. In the presence of Pseudomonas cells, motility of T. parva increases, as evidenced by a prominent colony protrusion toward the Pseudomonas (red arrow). (G) The motility bias parameter Δ was used to quantify the asymmetric motility of T. parva. (H) T. parva demonstrates a positive motility bias only in the presence of living Pseusomonas cells (+Pseud.). Cell-free supernatant of Pseudomonas culture (+Pseud. supernat.) does not cause motility bias in T. parva. ***—p<0.00.1. See also S17 Fig.

Fig 6. Multiple putative virulence factors Pseudomonas and Turneriella are up-regulated in tumor context.

(A) Expression of six genes encoding putative virulence factors of Pseudomonas in tumor polyps compared to healthy control polyps. The relative expression values are normalized to Pseudomonas 16S rDNA expression level to account for a differential abundance of Pseudomonas in tumorous and control polyps. Consistently with RNA-seq analysis (S6 Table), these six virulence factors are enriched in tumorous tissue, i.e. their expression in Pseudomonas is activated by the presence of Turneriella. (B) The density of Pseudomonas in tumorous polyps is dramatically lower than in healthy control polyps, consistent with earlier observations (Fig 1C, S6 Fig). (C) Expression of four genes coding for putative virulence factors of Turneriella in tumorous polyps compared to polyps that spontaneously lost tumors (Fig 4B). The relative expression values are normalized to Turneriella 16S rDNA expression level to account for a differential abundance of spirochetes in tumorous and polyps free of tumors with dramatically reduced Pseudomonas density (Fig 4B). Consistently with RNA-seq analysis (S6 Table), the virulence factors are enriched in tumorous tissue, i.e. their transcription in Turneriella is activated in the presence of Pseudomonas. (D) The density of Turneriella in tumorous polyps is slightly lower than in healthy polyps that spontaneously lost Pseudomonas and tumors, consistent with earlier observations (Fig 4B).

Fig 7. Temperature stress facilitates colonization of H. oligactis by T. parva and tumor formation.

(A) T. parva injected into H. oligactis colonizes the polyps more successfully, reaches higher abundance and induces tumors more frequently if the host was pre-stressed with elevated temperature (22°C). The microbiome of healthy control H. oligactis changes substantially after 3-day treatment of polyps at 22°C. Relative abundance of bacteria identified by 16S rDNA sequencing and resolved at the phylum level, n = 6. A successful injection rate with tumorous outcome is higher (52 of 94 injected polyps) in pre-stressed animals (22°C) compared to untreated animals (22 of 96 injected polyps) (chi-square statistics: χ² = 20.9; df = 2; P<10–5). (B) Temperature stress results not only in a decrease of Pseudomonas OTU750018 relative abundance, but also in a significant decrease of density (absolute abundance). In the tumorous polyps, the abundance of Pseudomonas is even lower. The density of Pseudomonas colonization was estimated using qRT-PCR amplification of Pseudomonas OTU750018 16S rDNA gene and normalized to the values in the healthy control polyps at 18°C. gDNA was extracted on day 3 of temperature treatment, before injection of T. parva.