The Native Microbiome is Crucial for Offspring Generation and Fitness of Aurelia aurita

The generation of daughter polyps was monitored every 48 h over 14 days. Native as well as recolonized polyps showed a similar budding rate of 0.4 daughter polyps per week (0.38 ± 0.02 and 0.41 ± 0.01, respectively), whereas sterile polyps generated 0.3 ± 0.02 daughter polyps/week, representing a statistically significant decrease of 21% (P value of 0.0003) (Fig. 2; see also Fig. S1 and Table S1C in the supplemental material). Environmentally challenged polyps (treatments 3 and 4) had further decreases in budding rates, with decreases for sterile polyps in native environment by 45% and sterile polyps incubated with unmanipulated food by 87% (Fig. 2 and S1; Table S1D) (treatment 3: t = 4.0, P value < 0.0002; treatment 4: t = 6.0, P value < 0.0001). Budding rates also decreased remarkably when polyps were infected with potential pathogens, resulting in almost completely impaired generation of daughter polyps, in particular, when challenged with P. espejiana (Fig. 2 and S1; Table S1E) (F = 71.3, P value for permutational multivariate analysis of variance [P_PERMANOVA] < 0.0001).

In a parallel experimental setup, strobilation of polyps was induced by adding the synthetic strobilation inducer 5 μM 5-methoxy-2-methyl indole (Fig. 1) (58). Strobilation induction was monitored each day, and strobila phenotypes as well as number of segments were monitored and detected beginning on day 5, when native polyps initiated segmentation (early strobila) (Fig. 3A; Table S2). Native polyps showed full segmentation and shaping as late strobila on day 9 (Fig. 3A; Table S2). Specifically, 51% of native and 58% of recolonized polyps showed initiation of segmentation on day 5 after induction (Fig. 3B; Table S2). Even 64% of native polyps in the sterile environment formed strobilae, which showed typical segmentation (Fig. 3B). However, the formation of strobilae was delayed, with early strobilae appearing after 8 days and late strobilae after 17 days. In contrast, 45% of sterile polyps showed segmentation (Fig. 3B; Table S1C) (t = 10.7, P value < 0.0001) (Table S2), but remarkably, phenotypically abnormal strobilae were formed (Fig. 4A). Phenotypical abnormalities were manifested by colorless, minimized, and thickened strobilae without absorbed tentacles (see examples in Fig. 4A). Polyps, which were environmentally challenged (treatments 3 and 4) showed massively reduced formation of strobilae, and the few ones observed were phenotypically abnormal (colorless, minimized, and thickened strobilae with inflated calyx) (Fig. 4). Likewise, infected polyps (treatments 6 to 11) showed almost no formation of strobilae (Fig. 3B; Table S1E) (F = 121.7, P_PERMANOVA < 0.0001), resulting in almost completely halted segmentation (Fig. 4A). Furthermore, subsequent ephyra release was detected for all treatments, beginning on day 12 after induction of strobilation, when native strobilae began to release ephyrae (Fig. 3A and 4A; Table S2). Native and recolonized strobilae formed approximately eight segments and finally released seven ephyrae with a size of 1.5 ± 0.1 mm in diameter after 12 to 14 days after induction (Fig. 4B; Table S2). Native polyps in a sterile environment formed nine segments but deferred release of only three ephyrae (Fig. 4B; Table S2). Sterile polyps were only slightly affected in induced segmentation (45% strobilae with a mean of six segments); however, an abnormal strobila phenotype was continuously detected, resulting in massively impaired ephyra release (Fig. 4; Table S1C) (F = 171.9, P_PERMANOVA < 0.0001). Moreover, only 15% of the formed sterile strobilae were able to release only a single ephyra of increased size (2.4 ± 0.2 mm, diameter increase of 70%) (Fig. 4C). Ultimately, infected polyps showed almost no segmentation; consequently, no ephyrae were released (Fig. 4; Table S1C) (F = 10.2, P_PERMANOVA < 0.0001). These findings strongly argue that strobilation and consequently ephyra release are massively interrupted in the absence of the native polyp microbiota as well as when infected with potential pathogens, ultimately resulting in halted generation of offspring.

In parallel with asexual reproduction (budding), further important fitness-correlated traits, in particular, survival and growth of A. aurita polyps, were monitored every 48 h over 14 days. Based on the overall phenotypical appearance of polyps and their tentacles, three health conditions were differentiated, namely, “healthy” (normal polyp phenotype, contracting tentacles), “harmed” (polyp phenotype impaired, tentacles degenerated), and “dead” (degenerated or no polyp) (see “Host fitness experiment” for detailed definition on phenotype characteristics), which are exemplified in Fig. 5A. Recolonized polyps showed similar survival rates as native polyps without an impaired phenotype (99%, corresponding to 89 ± 1 polyps, and 97%, corresponding to 87 ± 3 polyps, respectively), whereas 37% of sterile polyps were affected in health after 14 days (t_end) (Fig. 5B; Table S1C) (t = 3.8, P value < 0.0001). Native polyps were able to maintain their associated native microbiota (see below and Fig. 5) when kept in a sterile environment at least over the short period during the experiment (code xooo, only 5% polyps were affected) (Fig. 5B). However, sterile polyps were affected in survival or at least showed malformation when kept under native environmental conditions (39% affected) (Table S1D) (t = 2.8, P value < 0.0008) and were affected even more when fed with A. salina harboring their native microbiota (78% affected) (Fig. 5B; Table S1D) (t = 5.6, P value < 0.0001). Environmentally challenged polyps (treatments 3 and 4) (Fig. 1) showed significantly increased malformation and mortality already at the onset of the experiment, which worsened over time. Remarkable health loss, with up to 76% impaired polyps, was also detected when challenging with potentially pathogenic bacteria which were previously isolated from A. aurita and ambient water (65) (Fig. 5B; Table S1C) (P value < 0.0001). The survival rate of polyps was significantly affected by each bacterial challenge (Table S1E) (F = 15.9, P_PERMANOVA < 0.0001), with the exception of R. mobilis infection (Table S1E) (pairwise tests, P value was not significant). In more detail, we observed that unmanipulated native polyps were affected by infection with potentially pathogenic bacterial isolates during the course of the experiment, whereas sterile polyps were affected early on. Furthermore, R. mobilis and V. anguillarum (belonging to A. aurita native microbiota) affected the health/survival of native polyps only slightly (28%, of which 16% were dead, and 29%, of which 14% were dead, respectively) (Table S1E) (P value < 0.0001) compared to the effect of P. espejiana (56%, of which 19% were dead) (Table S1E) (P value < 0.0001) (Fig. 5B). A significant interaction of the type of ambient water (unmanipulated or sterile) as well as Artemia supplied as food (unmanipulated or sterile) was further identified (Table S1E) (F = 7.6, P_PERMANOVA < 0.0005 and F = 34.5, P_PERMANOVA < 0.0001, respectively), indicating that specific bacteria differed in their challenge depending on both water and food quality.

Healthy and harmed polyps were also analyzed with respect to growth and feeding rates. Native polyps had a mean polyp size (length plus width) of 3.85 ± 0.91 mm, whereas sterile polyps showed reduced start sizes of 2.3 ± 0.43 mm (Fig. 6). Native and recolonized polyps showed similar initial sizes and further similar growth rates (Fig. 6). Analyses indicated that growth was primarily affected in sterile artificial seawater and when infected with Vibrio anguillarum and Ruegeria mobilis (Fig. 6; Table S1). Feeding rates of healthy and harmed polyps were assessed in week three of the experiment on the five following days (Fig. 7 and S2). Native polyps had a mean clearance rate of 93.8% ± 8.6%, thus, similar to recolonized polyps (92.4% ± 9.5%), whereas sterile polyps showed a 77.2% ± 12.5% clearance rate (Fig. 7 and S2; Table S1C) (t = 12.5, P value <0.0001). Feeding rates were significantly reduced when treatments affected the overall health of polyps (Fig. S2). Analyses further disclosed a significant effect on the presence of the associated polyp microbiota as well as an effect of bacterial infection on feeding rates (Fig. 7 and S2). Native polyps were further affected in effective feeding when kept in sterile artificial seawater (Table S1) (P values < 0.0001).

To correlate the impaired asexual reproduction, survival rates, growth, and feeding with changes in the microbial community of the native A. aurita microbiota, 16S V1-V2 amplicon sequencing was performed during the course of the experiment (see Table S3). Antibiotic-treated polyps and strobilae, developed from sterile polyps, were indeed shown to be sterile (Fig. 8 and S3, S4, and S5). In line with our previous report (29), the present analysis further showed restructuring of the microbial community during polyp-jellyfish transition (Fig. S3, S4, and S5). Native and recolonized polyps showed similar bacterial community patterns due to successful reconstitution (Fig. 8A and S4; Table S3). However, relative abundances of community members were different, resulting also in identification of specific indicator operational taxonomic units (iOTUs) explaining the microbial communities (Fig. S4; Table 1). Moreover, changes in the native microbial community composition induced by the experimental treatments were identified (Fig. 8 and S3). Notable are the higher abundances of unclassified (uncl.) Gammaproteobacteria, uncl. Cyanobacteria, and Melitea as well as the lower abundance of Arcobacter in the native microbiota of the animals (Fig. S3). Redundancy analysis revealed four distinct microbial community clusters resulting from the respective manipulation (native, environmental challenged, native plus challenged, and sterile plus challenged) for polyps as well as strobilae (Fig. 8). Indicator OTU analyses further revealed bacterial taxa, which significantly correlated either positively or negatively with fitness traits (Tables 1 and 2). For instance, the presence of OTU0142 (uncl. Prevotella) was linked to essentially impaired fitness traits, whereas OTU0015 (uncl. Neptuniibacter) and OTU0003 (uncl. Alteromonas) might be beneficial bacteria (Tables S4 and S5). Moreover, OTU0057, OTU0178, and OTU0336 (Olleya, Rothia, and Gemella, respectively) were exclusively beneficial for strobilation or, in contrast, the presence of OTU0110 and OTU0144 (Corynebacterium and Balneola, respectively) resulted in halted transverse constriction (Table S5). Finally, the multivariate ordination analysis visually summarizes the observed fitness effects (Fig. 9), which mirror the identified microbial community clusters from redundancy analysis (RDA) (Fig. 8), ultimately correlating the observed fitness effects with microbial community shifts.

Husbandry and generation of sterile polyps are described in detail by Weiland-Bräuer et al. (29). Briefly, polyps of subpopulation North Atlantic (Roscoff, France) were kept more than 10 years in the lab in 2-liter plastic tanks in 3% artificial seawater (ASW) (tropical sea salts; Tropic Marin). Polyps were fed twice a week with freshly hatched Artemia salina (HOBBY, Grafschaft-Gelsdorf, Germany). Generation of sterile polyps was performed with antibiotic mixture added to sterile (filtered through 0.22-μm Durapore membrane filters; Merck Millipore, Darmstadt) artificial seawater (Provasoli’s antibiotic mixture with final concentrations of 360,000 U/liter penicillin G, 1.5 mg/liter chloramphenicol, 1.8 mg/liter neomycin, and 9,000 U/liter polymyxin B; all components from Carl Roth, Karlsruhe, Germany). Generation of sterile Artemia salina was similarly conducted. Eggs were hatched in sterile ASW with Provasoli’s antibiotic mixture (complemented with 3.5 mg/liter nystatin and 3.5 mg/liter amphotericin B). The absence of bacteria was confirmed by plating homogenized animals on marine bouillon agar plates (Carl Roth). After incubation at 19°C for 5 days, the CFU were determined. Absence of CFU indicated successful antibiotic treatment. Additionally, the full-length bacterial 16S rRNA gene was amplified using GoTaq polymerase (Promega, Madison, WI, USA) and primer set 27F (5′-AGAGTTTGATCCTGGCTCAG-3′)/1492R (5′-GGTTACCTTGTTACGACTT-3′) with genomic DNA isolated from putative sterile polyps and A. salina using a Wizard genomic DNA purification kit according to the manufacturer’s instructions (Promega). Samples without successful amplification were graded as sterile.

Bacteria (Vibrio anguillarum, Pseudoalteromonas espejiana, and Ruegeria mobilis; GenBank accession numbers MK967055, MK967174, and MK967091, respectively) for microbial challenge during host fitness experiments were isolated from A. aurita polyps of subpopulations from the Baltic Sea and North Atlantic as well as 3% ASW (30 practical salinity units [PSU]), respectively, as described in reference 65. Strains were grown in marine bouillon (MB; Carl Roth, Karlsruhe, Germany) at 30°C and 120 rpm to a turbidity optical density at 600 nm (OD₆₀₀) of 0.8. Bacterial cell numbers were detected using a Neubauer count chamber (91), and 10⁵ cells/ml were complemented with 5% dimethyl sulfoxide (DMSO) (1-ml aliquots) to store at −80°C before using 10 μl of 10⁵ cells/ml in 1 ml ASW per well during host fitness experiments. Concisely, Vibrio anguillarum isolated from A. aurita Baltic Sea polyps is widely distributed in marine and estuarine environments around the world (92–94). It is a global causative agent of vibriosis in marine fish and shellfish species, because it can efficiently grow and proliferate under environmental stress (95). Pseudoalteromonas espejiana isolated from artificial seawater is a ubiquitous bacterium abundantly found in marine environments and often associated with marine multicellular organisms (96). P. espejiana is able to degrade polymers and induces metamorphosis of Hydractinia echinata hydroid larvae (20, 97). Ruegeria mobilis was exclusively isolated from the polyp life stage (65). R. mobilis has primarily been isolated from marine aquaculture, where the tropodithietic acid (TDA)-producing strains have probiotic potential due to the inhibition of fish pathogens (98). V. anguillarum and R. mobilis belong to A. aurita’s native microbiota but are known as a potential pathogen and beneficial bacterium, respectively. P. espejiana is a nonnative bacterium.

Sterile polyps were used for recolonization with generated native A. aurita polyp microbiota. For generation of native microbiota, native polyps were transferred from 2-liter husbandry tanks into 1.5 ml reaction tubes (polyps were not fed for at least 3 days before transposition) and washed three times with 1 ml sterile ASW. Polyps (10 per reaction tube) were homogenized with a motorized pestle (KONTES; DWK Life Sciences, Wertheim, Germany) and resuspended in 1 ml sterile ASW. The homogenate was subsequently filtered through 3.1-μm filters (Lab Logistic Group, Meckenheim, Germany) to remove eukaryotic cells. Prior to conducting the host fitness experiment, 10 μl of the filtrate (approximately 2.5 × 10⁷ cells/ml) was added to a single polyp in 1 ml ASW (48-well plate) and incubated for 48 h at 19°C. Polyps were washed three times with 1 ml sterile ASW to remove noncolonized bacteria. Additionally, prepared native microbiota (filtrate) and recolonized polyps (after 48 h of incubation) were used for bacterial 16S rRNA amplicon sequencing.

Single native and sterile polyps of the subpopulation from the North Atlantic were transferred from 2-liter husbandry tanks to 48-well plates with 1 ml 3% ASW per well. Polyps settled to the bottom of the wells prior to the experiment start. For each treatment, 96 replicates were used. Twelve different treatments comprising combinations of polyp, food, and ambient seawater in terms of microbial composition (native, polyps maintained in ASW in the lab; sterile, antibiotic [AB] treated, challenged) were conducted to elucidate the impact of microbes on the fitness of the host (Fig. 1). Briefly, polyps either harbored their native microbiota (lab conditions) or were made sterile using an antibiotic mixture (see above). Moreover, one of the 12 treatments comprised sterile polyps, which were recolonized with generated native A. aurita polyp microbiota. Native or sterile artificial seawater (30 PSU, filtered through 0.22-μm filters) was used as ambient water, and Artemia salina with its native microbiota or sterile brine shrimps were applied as food. Native as well as sterile polyps were additionally infected with selected bacteria Vibrio anguillarum, Pseudoalteromonas espejiana, and Ruegeria mobilis. Six different fitness traits (survival, growth, feeding, budding, strobilation, and ephyra release) were analyzed using the stereomicroscope Novex Binokulares RZB-PL Zoom-Mikroskop 65.500 (Novex, Arnhem, Netherlands) with a high-definition multimedia interface (HDMI)/HD camera as follows. The survival/malformation of polyps was assessed every 48 h for 14 days based on overall phenotypical appearance of polyps and presence of tentacles and categorized as healthy, harmed, and dead (see Fig. 5). Healthy polyps showed typical morphology comprising an attached stalk and a stretched calyx with fully developed and extended tentacles. Polyps were rated as harmed when they lost at least two of those criteria, mostly showing a shrunken calyx and contracted or even absorbed tentacles. Dead polyps were of a shrunken roundish morphology or even dispersed. The health of polyps was compared between the first monitoring after 48 h (t₁) and at the end of experiment after 14 days (t_end). The growth of polyps was documented every 48 h for 14 days by measuring the length and width of the polyps, which were still alive (healthy and harmed). Mean start (t₁) sizes (length plus width) and end sizes after 14 days (t_end) of polyps were compared per treatment, and growth rates were calculated in percentages after 14 days for the remaining replicates. Moreover, budding as one form of asexual reproduction of A. aurita polyps, where daughter polyps are generated, was monitored for 14 days (in parallel to health and growth). Numbers of daughter polyps were counted for all alive (healthy and harmed) polyps, and the corresponding budding rate was calculated as daughter polyp generation per week. In addition, on days 15 to 19, feeding of alive polyps (healthy and harmed) was monitored each day. Therefore, freshly hatched Artemia salina were harvested from 500-ml brood containers in a harvesting container with an Artemia sieve (0.15-mm mesh). A. salina were washed three times with sterile ASW on the mesh and concentrated to 50 ml. A single polyp was fed with 20 A. salina (in 50 μl of sterile ASW) and incubated for 1 h before counting remaining A. salina. A clearance rate of A. salina was calculated per treatment in percentages for each day (day 1 [t₁] up to day 5 [t_end]). Finally, a new set of polyps (96 replicates per treatment) was prepared as mentioned above in 48-well plates, with a single polyp in 1 ml ASW corresponding to treatments shown in Fig. 1. Strobilation of polyps was induced by adding 5 μM 5-methoxy-2-methyl indole to the ambient water on the three following days (involving daily washing). Afterwards, polyps were washed on day 4 with the respective ambient water. Strobilation induction was monitored each day, and strobila phenotypes as well as number of segments were detected beginning on day 5, when native polyps began segmentation (early strobila). Native polyps showed full segmentation and shaping as late strobila on day 9. Recently, 5-methoxy-2-methyl indole was demonstrated to act as the temperature-dependent synthetic activator of strobilation in A. aurita (58). The selected procedure of chemical induction enables the reduction of the complexity of the strobilation process to a minimum, particularly, for investigating the impact of the microbiome and not the environmental triggers during reproduction. This procedure is more time efficient than the natural induction by lowering the temperature (approximately 1 month at 8°C). Ephyra release was monitored each day after first appearance, and numbers of released ephyrae were detected beginning on day 12, when native strobilae began to release ephyrae. Ephyra release was monitored for the next 4 weeks.

In addition to the monitoring of fitness traits, 16S rRNA amplicon sequencing was performed to analyze potential changes in the microbial community composition based on the treatments, ultimately correlating observed fitness traits with changes in microbial community patterns. Therefore, six polyps were removed from the 48-well plates before the experimental start (t₀) for DNA isolation and subsequent 16S rRNA amplicon sequencing (90 replicates remaining for each treatment at t₀). After 14 days (t_end) of experiment, further six polyps were removed for 16S rRNA analysis. Furthermore, six polyps were also removed from the second set of polyps before induction of strobilation (t₀). Likewise, six late strobila (removed on day 10 after strobilation induction) per treatment were removed for sequence analysis.

Factor effects were assessed for each fitness variable (i.e., counts of harmed and dead polyps, growth rate in percent, clearance rate in percent, budding rate in percent, strobilae count, and ephyra production in percent) using univariate permutational analysis of variance (99). All variables were tested under three test designs as outlined in Table S1. PERMANOVAs were performed on Euclidian resemblance matrices calculated from log₂ (x + 1)-transformed data (100) and were based on 9,999 permutations of residuals under a reduced model and type III partial sums of squares using Primer-E V6. (100). Within each significant factor, pairwise post hoc tests followed, providing insight into differences between the treatment groups (Table S1).

Eukaryotic as well as bacterial DNA from single A. aurita polyps was isolated using a Wizard genomic DNA purification kit (Promega, Madison, WI, USA) according to the manufacturer’s protocol.

Bacterial DNA was isolated as described above and used for the generation of PCR amplicon libraries using uniquely barcoded primers flanking the V1-V2 hypervariable regions (27F–338R; V1_A_Pyro_27F [5′-CGTATCGCCTCCCTCGCGCCATCAGTCAGAGTTTGATCCTGGCTCAG-3′] and V2_B_Pyro_27F [5′-CTATGCGCCTTGCCAGCCCGCTCAGTCAGAGTTTGATCCTGGCTCAG-3′]) with fused MiSeq adapters and heterogeneity spacers in a 20-μl PCR using Phusion high-fidelity DNA polymerase (New England Biolabs) (101). Reactions included the following components: 4 μl of HF buffer, 0.4 μl deoxynucleoside triphosphates (dNTPs; 200 μM each), 0.8 μl each of forward and reverse primers (2 μM), 0.2 μl Phusion Hot Start II high-fidelity DNA polymerase, and 2 μl DNA (20 ng). PCRs were conducted with the following cycling conditions (98°C, 30 s; 30 × [98°C, 9 s; 55°C, 60 s; 72°C, 90 s]; 72°C, 10 min; 10°C, infinity). Amplicons were size checked and purified using a MinElute gel extraction kit (Qiagen). Purified amplicons were quantified using a Quant-iT PicoGreen kit (Invitrogen), and equal amounts of the purified PCR products were pooled for subsequent Illumina sequencing as indicated by band intensity and measured with Quant-iT PicoGreen kit. Amplicon sequencing was performed on the Illumina MiSeq platform with v3 chemistry (2 × 300 cycle kit) according to the manufacturer’s instructions at Max-Planck Institute for Evolutionary Biology in cooperation with S. Künzel.

All steps of sequence processing were conducted with the program mothur v1.39.5 according to the MiSeq standard operating procedure (SOP). In brief, 16S rRNA amplicon reads were aligned against the bacterial SILVA database, which was previously trimmed to the 16S V1-V2 rRNA gene regions to improve accuracy of the alignment (align.seqs). Subsequently, a random subset of 1,000 sequences per sample was generated to eliminate bias due to unequal sampling effort (sub.sample). The aligned and subsampled sequences were then classified using the Greengenes database v13_8_99 (classify.seqs) and reads assigned to the lineages Archaea, Chloroplast, Mitochondria, and unknown, were removed. The remaining sequences were used to compute a distance matrix (dist.seqs) for binning sequences into operational taxonomic units (OTUs) by average neighbor clustering (cluster.split). Five hundred eleven OTUs at a 97% similarity threshold (roughly corresponding to species level) distinction were considered. All downstream computations were performed in R v2.15.1 (102) as described in reference 29. Indicator OTU analysis was also performed in R. Mothur-shared OTU tables were imported into R using the phyloseq package (103). Subsequently, the multipatt function of the indicspecies package (104) was used to identify OTUs that were significantly associated with the replicate groups. For this analysis, only the original replicate groups, as provided in the group file, were considered, and 500 permutations were conducted. Significantly associated OTUs were summarized for each group, and their taxonomic classification was assigned. Furthermore, relationships of bacterial OTU abundance with host fitness variables in the treatment groups were evaluated using distance based linear models (DistLM) as implemented in Primer-E V6. Prior to analysis, host fitness data were averaged across the treatment groups, normalized, and aligned with square root-transformed OTU abundance data (100). The DistLM routine was performed using Euclidian distances and implementing a forward selection procedure with 999 permutations and adjusted R² criterion. Additionally, Spearman rank correlation coefficients were obtained for the predictors.

Sequence data were deposited under the NCBI BioProject PRJNA633008 comprising locus tag prefixes SAMN14930335 to SAMN14930451.