1 Introduction

Photosymbiosis is a symbiotic relationship between a heterotrophic and a phototrophic organism (Cowen 1988). While this symbiosis is best known for corals that harbour Symbiodiniaceae Fensome, Taylor et al. 1993 (Muscatine and Porter 1977) photosymbiosis is quite prevalent among metazoans (Melo Clavijo et al. 2018). Porifera, for instance, can establish symbiotic relationships with photoautotrophic cyanobacteria, chlorophyte algae, or Symbiodiniaceae (Reisser 1984; Diaz et al. 2007; Hill et al. 2011). Tropical colonial ascidians establish photosymbiosis with cyanobacteria of the genus Prochloron (Hirose 2015) or Synechocystis trididemni (Hirose et al. 2009) and even several molluscs, including bivalves and heterobranch sea slugs, are photosymbiotic. While bivalves are in symbiosis with Symbiodiniaceae, in Heterobranchia Burmeister, 1837 two types of photosymbiosis can be distinguished. Sacoglossa Bergh, 1867 can incorporate photosynthetically active chloroplasts in their cytosol from their food algae (Händeler et al. 2009; Rauch et al. 2018), a phenomenon known as functional kleptoplasty (Händeler et al. 2009). Among Nudibranchia Cuvier, 1817 members of the suborder Cladobranchia are known to incorporate symbionts of the family Symbiodiniaceae. These slugs must acquire new symbionts (photobionts) in each generation by feeding on photosymbiotic cnidarians and, in a special case, by filtering water (Burghardt and Wägele 2014). The photobionts are then incorporated into epithelial cells of the digestive gland system and remain photosynthetic (Burghardt et al. 2005, 2008; Rola et al. 2022).

Photosymbiosis is considered to be highly advantageous for the animal host. Yet, aside from corals, the photobionts’ contribution to the hosts’ physiology is poorly understood. Previous studies used starvation experiments under photosynthetic and non-photosynthetic conditions to investigate the stability and benefits of the photosymbiosis between Heterobranchia sea slugs and their photobionts (Kempf 1991; Teugels et al. 2008; Melo Clavijo et al. 2018; Monteiro et al. 2019; Silva et al. 2021). When starved in the light, the slugs lost weight less rapidly compared to those starved in the dark (Hinde and Smith 1972, 1975; Hawes and Cobb 1980). This indicates a beneficial effect of light on the slugs probably due to the nutritional support by the photobionts. Further, degraded cells of Symbiodiniaceae in the feces of starving photosymbiotic sea slug species, suggest that slugs can also obtain nutrients by digesting the algae (Kempf 1984). Aside from starvation, light stress is also often used that leads to the photodynamic production of reactive oxygen species (ROS) by the photobiont, causing additional stress for the host (Halliwell and Gutteridge 1985). This formation of ROS has also been observed for Symbiodiniaceae which have been kept in the dark (Tolleter et al. 2013). Melo Clavijo et al. (2022) showed that starvation stress before feeding enhances ROS quenching in the sea slug Berghia stephanieae, notably inducing glutathione peroxidase (Melo Clavijo et al. 2022). These studies, however, neglect that the slugs not only harbour algal symbionts but also host a variety of bacteria phyla. Consequently, light and starvation stress may also affect these symbiotic bacteria, which could have consequences for the entire holobiont. Such a holobiont is a biological system composed of a eukaryotic host organism and cohabiting species from different phyla that perform diverse metabolic, immune, and environmental adaptation-related tasks (Bordenstein and Theis 2008; Baedke et al. 2020). For example, the composition of the hosts’ symbiotic bacteria is important for the host and the photobiont, leading towards an increased photochemical performance if certain bacteria are present (Matthews et al. 2023). Hence, it is crucial to identify all organisms involved in this system to better understand the physiology of the holobiont and its photosymbiosis.

The concept of holobiont has already been implemented in coral research, but it’s only been in the past ten years that the microbiome’s importance for corals has become apparent (Rosenberg et al. 2007; Ritchie 2011; Hernandez-Agreda et al. 2017; Mohamed et al. 2023). The bacterial communities appear to be highly dynamic and seem to contribute significantly to photosymbiosis, providing the host with carbon, nitrogen, and sulphur (Rohwer et al. 2002; Ritchie 2011; Röthig et al. 2016; Voolstra and Ziegler 2020; Meunier et al. 2021). Despite the importance of the microbiome for the coral holobiont and its photosymbiosis, only a few studies investigated the bacterial communities of other photosymbiotic animals (Prazeres et al. 2017; Britstein et al. 2020; Luter et al. 2020; Mahadevan and Middlebrooks 2020; Röthig et al. 2021; Posadas et al. 2022; Ng et al. 2022).

Sea slugs and their symbiotic bacteria received increased attention in past years due to antitumor and antibiotic potential of specific heterobranch-associated bacteria (Gerwick and Fenner 2013; Böhringer et al. 2017; Kristiana et al. 2019; Abdelrahman et al. 2021; Džunková et al. 2023; Elfeky et al. 2023). The detection of typical bacterial fatty acids, as well as ultrastructural analyses, revealed a high abundance of bacteria, especially in the notum, indicating the involvement of bacteria in nutrition and defence against predators (Martinez-Pita et al. 2005; Doepke et al. 2012; Leal et al. 2012; Zhukova and Eliseikina 2012). Heterobranchs probably obtain some of their symbiotic bacteria horizontally from their food source (Schuett and Doepke 2013; Mahadevan and Middlebrooks 2020; Davis et al. 2013), yet, the diversity of symbiotic bacteria, as well as their contribution to the slug holobiont and photosymbiosis, are still not well understood.

Here, we used Berghia stephanieae (Á. Valdés, 2005) to shed light on the origin, abundance, and impact of stressors on the microbiome of photosymbiotic cladobranchs. B. stephanieae feeds exclusively on the sea anemone Exaiptasia diaphana (Rapp, 1892), and incorporates the photobionts of its food source into cells of its digestive gland system where they remain active for up to a week before being digested or excreted (Rola et al. 2022). Both species are well-studied model organisms in the realm of mollusc-dinoflagellate ecology (Weis et al. 2008; Monteiro et al. 2019; Rola et al. 2022; Silva et al. 2023). E. diaphana has already been studied intensively regarding its microbiome. Various studies showed strong overlaps in the bacterial phyla and core microbiome (the microbial taxa shared by two or more samples from the same host or environment) of different clonal lines of E. diaphana. These results favoured a horizontally obtained microbiome in the anemones (Herrera Sarrias et al. 2017; Maire et al. 2021; Hartman et al. 2020). Contrarily, Curtis et al. (2023) could not replicate these findings and were unable to identify a shared core microbiome for different clonal lines of E. diaphana kept under identical rearing conditions. Hence, the microbiome in the anemones seems to be genetically shaped (vertical transmission) (Curtis et al. 2023). Furthermore, aposymbiotic and symbiotic E. diaphana show different bacterial microbiomes and core microbiomes, which seem to significantly impact the cnidarian holobiont and indicates a horizontal microbiome acquisition. These ramifications support the hypothesis that the entire holobiont adapts to the symbiotic state and that the microbiome likely plays a significant part in photosymbiosis (Röthig et al. 2016; Xiang et al. 2022; Curtis et al. 2023).

In our present study, we investigated the bacterial community of the sea slug Berghia stephanieae (Á. Valdés, 2005) and its food source Exaiptasia diaphana (Rapp, 1829) as well as the aquarium water to determine the origin of the slug’s microbiome, hypothesizing that the slug’s bacterial composition may originate from its food source, similar to its photobiont. Concurrently, we explored how starvation-induced stress, combined with different light conditions (regular, high, and darkness), may impact the microbiome over a week-long period. By analyzing bacterial abundance and metabolic pathways, we aim to understand the impact of applied light and starvation stress on symbiotic bacteria in B. stephanieae and their association with the photobiont. While previous studies have examined photosymbiosis under stress, none have delved into the symbiotic bacteria possibly shaping the photosymbiosis. We anticipate stress to influence bacterial composition and pathways, potentially leading to a reduction in bacterial abundance during starvation and a shift in composition under light stress. This investigation sheds light on the intricate interplay between the slug and its symbiotic bacteria within the context of photosymbiosis.

2 Methods

2.1 Exaiptasia diaphana culture

Several specimens of E. diaphana were purchased from a local provider (Seepferdchen24 Meeresaquaristik GmbH, Germany) in January 2022 and maintained in a 55 L tank (60 cm x 30 cm x 30 cm) filled with circulating artificial seawater (ASW) (AB Reef Salt, Aqua Medic, Germany) at a light intensity of 30 \({\mu }\)mol photons m\(^{-2}\) s\(^{-1}\) (Daylight Sunrise 520, Sera), on a 12h light/ 12h dark cycle. The salinity was kept at 33 Practical Salinity Unit (PSU), the temperature at 21 \(^{\circ }\)C, and two BactoBalls (Fauna Marin GmbH, Germany) were added and replaced every two weeks. Further, Artemia nauplii were cultivated using the Artemio® Set (JBL, Germany) and 16 g of Ready-to-use Artemia (brine shrimp) mixture (JBL, Germany). Anemones were fed with freshly hatched Artemia nauplii two times per week.

2.2 Berghia stephanieae culture

Six breeding pairs of B. stephanieae were purchased from a local provider (Seepferdchen24 Meeresaquaristik GmbH, Germany) in February 2019. Each pair was kept in a 75 mm diameter plastic container with a lid (FAUST, Germany) in 35 ml ASW at 21\(^{\circ }\)C, a light intensity of 30 \({\mu }\)mol photons m\(^{-2}\) s\(^{-1}\) (Daylight Sunrise 520, Sera), and a day/night cycle of 12 h / 12 h. Water was changed three times per week, using freshly prepared ASW with a salinity of 33 PSU, a pH of 8.0, and a temperature of 21 \(^{\circ }\)C. Slugs were fed with small E. diaphana anemones (7 mm foot and 4 mm oral disk or 4 mm foot and 3 mm oral disk) three times per week. Once the breeding pairs spawned, the egg masses were collected and placed individually in 55 mm diameter plastic containers with lids (FAUST, Germany) in 20 ml ASW and maintained as stated above. The larvae were observed and fed with tentacles of E. diaphana three times a week immediately after hatching. When the juvenile stage was reached, the slugs were fed with small anemones three times a week.

2.3 Sampling and sample processing

Twenty adult cladobranchs were randomly selected from the culture and separated into four groups (n=5). Each group was placed in one 75 mm diameter container with 35 ml ASW, and water was changed three times a week. The first group was kept under standard husbandry conditions as described above (Fed; BS 1-5) and were sampled directly after feeding. The second group was starved under regular light conditions of 30 \({\mu }\)mol photons m\(^{-2}\) s\(^{-1}\) (Starved L; BS 11-15). The third group was starved under high light with a light intensity of 90 \({\mu }\)mol photons m\(^{-2}\) s\(^{-1}\) (Starved HL; BS 16-20), and the last group was starved under complete darkness (Starved D; BS 21-25). After a starvation period of seven days, all slugs were snap-frozen at -80 \(^{\circ }\)C until further preparation. Five whole E. diaphana specimens were sampled directly from the tank after feeding and snap-frozen at -80 \(^{\circ }\)C until further preparation. Water samples (W1-W5) were sampled from the aquarium and directly processed.

Following the manufacturer’s instructions, RNA from each sample was extracted using the TRItidy GTM (PanReac AppliChem, Spain) to get an understanding of the functional microbiome. RNA concentrations of samples were quantified on a QubitTM 4 Fluorometer (Invitrogen, USA). To generate a cDNA template, 50 ng of each RNA sample or 1 \({\mu }\)L of aquarium water was used with the LunaScript® RT SuperMix Kit (NewEngland Biolabs, UK) following the manufacturer’s instructions. For DNA sequencing, bacterial DNA was amplified by PCR using primers with Illumina adapters (underlined) targeting the V1-V2 regions of the 16S rRNA gene: V1V2forward (5’-TCTTTCCCTACACGACGCTCTTCCGATCTAGAGTTTGATCCTGGCTCAG-3‘), V1V2rev (5’-GTGACTGGAGT TCAGACGTGTGCTCTTCCGATCTTGCTGCCTCCCGTAGGAGT-3’) (Weisburg et al. 1991). Adapter PCR was performed in a total volume of 20 \({\mu }\)L: 5 \({\mu }\)L template cDNA, 10 \({\mu }\)L DreamTaq Green PCR Master Mix (2X) (Thermo Fisher, USA), and 2 \({\mu }\)L of 10 \({\mu }\)M of each primer. Each sample was adjusted to 20 \({\mu }\)L with distilled and sterilised water. The PCR amplification was performed with 30 cycles of denaturation at 94\(^{\circ }\)C for 1 min, annealing at 54\(^{\circ }\)C for 1 min, and extension at 72\(^{\circ }\)C for 1 min. The initial denaturation and final elongation were 95\(^{\circ }\)C for 15 min and 72\(^{\circ }\)C for 10 min, respectively. The PCR product for each sample was cleaned with the Agencourt AMPure XP magnetic bead system (Beckman Coulter, USA). Subsequently, an indexing PCR was performed to add Illumina sequencing adapters containing TrueSeq indices. Index PCR was conducted in 20 \({\mu }\)L volume using 5 \({\mu }\)L of the previous purified adapter PCR product, 10 \({\mu }\)L Q5® High-Fidelity 2X Master Mix (NewEngland Biolabs, UK), 1 \({\mu }\)L of 10 \({\mu }\)M of the respective forward and reverse index primer, and 3 \({\mu }\)L distilled and autoclaved water. The PCR amplification was performed following the manufacturer’s instructions. Indexed PCR products were cleaned using the Agencourt AMPure XP magnetic bead system. Final DNA concentrations of samples were quantified on a QubitTM 4 Fluorometer and samples were then pooled in equimolarly and sent to the University of Saarbrucken for sequencing on an Illumina MiSeq platform. Sequencing was conducted with a 2x250bp paired-end approach according to the manufacturer’s instructions aiming at 30,000 reads per sample.

2.4 Sequencing data workflow and diversity analyses

Raw, demultiplexed MiSeq reads were joined in QIIME2 v.2021.4 (Bolyen et al. 2019a), and denoised using the DADA2 plugin (Callahan et al. 2016) with a maximum error rate of 2. Taxonomy was assigned in QIIME2 against a SILVA database (v132) (Bolyen et al. 2019a) and amplicon sequence variants (ASVs) identified as eukaryotes, mitochondria, or chloroplasts were removed (Additional File 1; Table 1). Relative abundance of bacterial families, alpha and beta diversity calculations, as well as statistical analyses, were performed in QIIME2 v2023.9 (Bolyen et al. 2019b) and R v4.0.3 (R Core Team 2021) with differences considered significant at \(p < 0.05\). Shannon index (Shannon 1948) was used to describe and compare alpha diversity across samples. Differences in Shannon diversity between samples were evaluated by one-way analysis of variance (ANOVA, Girden 1992). Differences in beta diversity were analysed using Bray-Curtis dissimilarity matrices and tested via permutational multivariate analysis of variance (PERMANOVA, Anderson 2001). Principal Coordinate Analysis (PCoA) plot showing the variation in community composition among samples paired with the Shannon index was compiled in R v4.0.3 using the packages devtools (Wickham et al. 2022), ggplot2 (Wickham 2016), tidyverse (Wickham et al. 2019) and qiime2R (Bisanz 2018). Differences in pathway abundance between groups were evaluated by ANOVA or if data deviated from normality and/or homogeneity of variance by Kruskal-Wallis-Test (Kruskal and Wallis 1952).

2.5 Core microbiome

The core microbiome was analysed in QIIME2 v.2021.4. Using the command qiime feature-table core-features, a list of ASVs, present in at least 80% of the samples analysed, was created for each sample group. Venn diagrams and Pie charts were created in R v4.0.3 using the packages ggplot2 and RColorBrewer (Neuwirth 2022). ASVs that were present less than 1% were clustered under “Others”.

2.6 Pathway abundance

ASV IDs were stripped and ASVs with the same taxonomic assignment were combined using the web server METAGENassist (Arndt et al. 2012). Further, unassigned and unmapped reads were excluded, and 7 taxa were filtered out based on interquartile range (IQR) (Hackstadt and Hess 2009). The remaining 146 taxa were normalized across samples by sum and across taxa by Pareto scaling. Data were analysed by phenotype “metabolism” using the Spearman distance measure to cluster the most abundant metabolic processes. Differences in pathway abundance between the conditions were evaluated using METAGENassist by one-way analysis of variance (ANOVA) and Tukey HSD post hoc comparison of means. Statistical analyses were considered significant when \({p} < 0.05\). A Heatmap displaying putative functional differences based on the bacterial community composition of each sample was created in METAGENasisst.

3 Results

DNA sequencing produced 646,391 reads across anemone samples (n = 5) and slug samples (n = 20). After merging, denoising, and chimera filtering, 372,783 reads remained. After removing contaminants, 346,225 ASVs were kept. Rarefaction curves for bacterial sequences plateaued in all samples, suggesting sufficient sequencing depth to capture bacterial species diversity (Additional File 2: Fig. 1).

Fig. 1
figure 1

Relative frequency of the most abundant bacterial families. The ten most abundant bacterial families across all anemone and all slug samples are represented by a different colour. Less abundant families are grouped under “Others”

Full size image
Fig. 2
figure 2

Principal Coordinate Analysis (PCoA) plot of alpha- and beta-diversity. PCoA plot based on Bray-Curtis dissimilarity. Shape of circle represents taxon (anemone or slug), size of circle indicates Shannon diversity index, colour of circle represents husbandry condition (fed, starved in darkness (starvedD), high light (starvedHL), or regular light(starvedL)

Full size image

3.1 B. stephanieae’s distinct bacterial composition shifts during starvation

Sequence identification at the family level revealed noticeable differences between the most abundant bacterial communities associated with E. diaphana and B. stephanieae (Fig. 1; Additional file 1: Table 2). For instance, Endozoicomonadaceae, Clostridiaceae, Pleomorphomonadaceae, and Rhodocyclaceae, were the most common bacterial families in E. diaphana, accounting for more than 50% of all identified bacteria (Fig. 1). In the slugs, Mycoplasmataceae, Pseudoalteromonadaceae, and Rhodobacteraceae were the most dominant and accounted for more than 50% of all bacteria, irrespective of the experimental condition. Mycoplasmataceae were absent in anemones, but Comamonadaceae (clustering under others for B. stephanieae fed, Fig. 1), Flavobacteriaceae, Propionibacteriaceae, and Beijerinckiaceae (all clustering under others for E. diaphana, Fig. 1) were present in almost all anemones and slugs, albeit on a low abundance. To further identify patterns of differential abundance of ASVs, we determined the core microbiome for each condition group (all ASVs present in at least 80% of all samples of the respective condition; Additional File 1: Table 3 and 4). The core microbiome of the anemones contained 41 ASVs, with 40 ASVs exclusive to the anemones (Figs. 3 and 4A). On the ASV level of the core microbiome, several bacteria were highly abundant in the anemones, including Clostridium sensu stricto 11 sp. 2 (25.51 %), Rhizobiales sp. (18.42 %), Azirospira sp. (11.11 %), and Endozoicomonas sp. (9.83 %) that in combination account for approximately 65 % of total reads of the anemones’ core microbiome. The fed slugs’ core microbiome contained 28 ASVs, with Mycoplasma sp. as the most abundant bacteria, accounting for 43.39 %. Besides Mycoplasma sp., an unclassified Rhodobacteraceae (7.36 %), Cytophaga sp. (6.98 %), and Kordiimonas sp. (5.9 %) dominated the core microbiome of fed slugs (Fig. 3). Only Endozoicomonas sp. was present in both the anemones’ and the fed slugs’ core microbiome (Figs. 3 and 4A).

Fig. 3
figure 3

Core microbiome of anemone and slugs kept under different husbandry conditions. All ASVs found in at least 80% of all samples of the corresponding condition were classified as “E. diaphana core microbiome”, “B. stephanieae fed core microbiome”, “B. stephanieae starved L core microbiome”, “B. stephanieae starved HL core microbiome”, and “B. stephanieae starved D core microbiome”. Each colour denotes a unique ASV represented by more than 1%; rare ASVs have been condensed in the group “Others” and are shown in grey

Full size image
Fig. 4
figure 4

Venn diagrams comparing core microbiome composition of anemone and slug. Shared ASVs of the core microbiome of E. diaphana and fed B. stephanieae (A) and all B. stephanieae samples (fed, starved in light (L), high light (HL) and darkness (D)) (B). Numbers indicate total abundance of either unique or shared ASVs

Full size image

From the 216 ASV detected in the water samples, 9 were shared with the slugs and 4 with the anemones. None of these ASV could be taxonomically assigned differently than “uncultured bacteria”(see Additional file 1: Table 5 and Additional File 2A).

In the slugs, starvation resulted in a microbiome composition shift, including an increase in Mycoplasmataceae. For example, in the fed sample BS2 (B. stephanieae fed), Mycoplasmataceae represented approximately 42% of the total reads, whereas in the dark-starved sample BS 25 (B. stephanieae starved in darkness), Mycoplasmataceae represented 76% of the total reads (Fig. 1). Starvation further induced the depletion of Pseudoalteromonadaceae and Cryomorphaceae in most samples and an increase in Moraxellaceae. Moreover, Alteromonadaceae were found only in starved slugs (Figs. 1 and 2).

On the \(\text {ASV}\) level, starvation led to a depletion of 17 bacterial species identified in the core microbiome of fed slugs. Fed and starved slugs shared only two bacterial species, Mycoplasma sp. and Cutibacterium sp. in their core microbiome (Fig. 4B). In starved L slugs, Mycoplasma sp. accounted for 93.19 % of all 27 identified ASVs, Cutibacterium sp. for 2.59 %, and the remaining 25 ASVs accounted for 4.22% (Fig. 3). Only in starved L slugs, Enhydrobacter sp. (0.9 %, included in “Others”) could be identified (Fig. 4B). The core microbiome of starved HL slugs contained 43 ASVs, out of which 80.96 % of the reads were Mycoplasma sp., 3.49 % Methylobacterium sp., 2.58 % Cutibacterium sp., and the remaining ASVs with an abundance lower than 2 % (Fig. 3). Among the starved slugs, HL slugs showed the highest number of unique ASVs (17; Fig. 4B) and shared the most ASVs with a total of 15 bacterial phyla with slugs starved in regular light (L; Fig. 4B). Lawsonella sp. was only present in the core microbiome of starved L and HL slugs. In starved D slugs, the core microbiome contained 94.70 % Mycoplasma sp., 2.84 % Alteromonas sp., and 14 ASVs that accounted for 2.46 % in total (Fig. 3). Paracoccus marinus, Coxiella sp., and unclassified Terasakiellaceae were exclusive to the core microbiome of starved D slugs (Fig. 4B).

3.2 The diversity of the microbiome declines during starvation

Anemones showed the highest alpha-diversity metric (Shannon indices), while one starved L slug (BS14) showed the lowest Shannon indices. A one-way ANOVA revealed a statistically significant difference in Alpha-diversity metrics (Shannon index) between at least two groups F\(_{(4, 20)}\) = 20.93, \({p} < 0.001\); Fig. 2). Subsequent Tukey’s HSD Test for multiple comparisons indicated that the starved groups (L, HL, and D) did not differ significantly from each other (Additional File 1:Table 4). In contrast, fed slugs and the anemones differed significantly compared to all starved groups but not to each other (Additional File 1: Table 6). PCoA visualization of beta-diversity using the Bray-Curtis dissimilarity index revealed a coherent grouping of anemones, fed slugs, and starved slugs, respectively (Fig. 2; Additional File 1: Table 7). PERMANOVA testing (999 permutations) indicated that bacterial community structure varied significantly depending on the husbandry condition (pseudo-F\(_{4}\) = 24.01,p = 0.001). Pairwise comparisons between all conditions did yield significant differences between all groups except between Starved L and HL (Additional File 1: Table 8). Analysis of the water samples revealed significant differences in alpha and beta diversity between water samples and fed slugs (statistical results and see Additional Material File 1: Table 9).

Fig. 5
figure 5

Taxonomy-based functional profiling of bacterial communities. Heatmap displaying changes in putative functional differences between the five groups (E. diaphana(anemone), B. stephanieae fed and starved in light (StarvedL), high light (StarvedHL), and darkness (StarvedD) on a relative scale, with enrichment indicated in red and depletion in blue

Full size image

3.3 The change in bacterial communities causes significant differences in the regulation of metabolic pathways

The observed differences in the relative abundance of bacterial taxa in E. diaphana and B. stephanieae could indicate a functional change in the bacterial composition. Therefore, we determined the metabolic pathway abundances of each sample and compared them regarding the condition (Additional File 1: Table 10). Anemones and fed B. stephanieae clustered together, indicating homogeneity in the enrichment and depletion of metabolic pathways. However, starved L, HL, and D slugs did not cluster and differed in the enrichment and depletion of metabolic pathways (Fig. 5). Regarding the regulation of metabolic pathways, “Dehalogenation” and “Ammonia oxidizer” were enriched for all anemones and B. stephanieae fed samples, while “Nitrogen fixation” was upregulated only in starved B. stephanieae samples. “Chitin degradation”, “Sulfur metabolizing” and “Xylan degrader” were only enriched in anemone samples (Fig. 5). Testing for differences (ANOVA or Kruskal-Wallis Test; see Additional File 1: Table 11 for further information) revealed a statistically significant difference in the regulation of the metabolic pathways between at least two groups for all pathways except “Sulfate reducer” (p \(< 0.05\)). Post hoc testing was conducted using either Tukey’s HSD or Dunn’s test with Bonferroni correction (Additional File 1: Table 11 and Additional File 4: Figure 1). These revealed significant differences in the regulation of 4 metabolic pathways between E. diaphana and B. stephanieae fed “Chitin degradation”, “Xylan degradation”, “Sulfur metabolizing”, and “Nitrite reducer”) with the first 3 being depleted in the slug and the last one in the anemone. Between fed slugs and all starved slugs, significant differences in the regulation of 1 metabolic pathway was detected (“Nitrogen fixation”) being the only pathway significantly enriched in starved slugs. Compared to fed slugs, in HL starved slugs, an additional 5 bacterial metabolic pathways were enriched (“Xylan degrader”, “Degrades aromatic hydrocarbon”, “Atrazine metabolism”, “Sulfur oxidizer”, and “Carbon fixation”), and in LL starved slugs 4 (“Ammonia oxidizer”, Dehalogenation”, “Sulfide oxidizer” and “Atrazine metabolism”). Furthermore, in HL slugs, the pathway “Chlorophenol degrading” was significantly enriched compared to dark-starved slugs. LL starved slugs showed no significant changes in metabolic pathways compared to the other starved slugs.

4 Discussion

4.1 The slug microbiome differs from the anemone

The horizontal acquisition of the microbiome is usual for marine organisms, and we anticipated that the slugs obtain at least part of their bacteria simultaneously with their photobionts from the anemones. The slugs are maintained in plastic containers devoid of any substrates (i.e. stones) capable of transmitting microbes through physical contact, thereby minimizing potential microbial transmission. The anemones provided for the slugs’ feeding represent the sole organisms introduced to the container harbouring potential symbiotic microbes. Moreover, the slugs not only consume these anemones but also manoeuvre in their proximity, facilitating physical contact and, consequently, exposure to the microbes inhabiting the anemone’s surface. Thus, alongside the bacteria hypothesized to inhabit the slug’s gut owing to dietary intake, an anticipated microbial overlap between the anemones and the slugs due to physical interaction was expected. Nonetheless, such overlap was not evident in our investigation. Additionally, we analysed the bacterial composition of the water housing both the slugs and the anemones to ascertain whether the slugs acquire symbiotic bacteria from this source. Our slugs exhibited an overlap of 9 out of 28 bacteria species in their core microbiome with the water. Nevertheless, all corresponding bacteria elude species-level identification and are classified as uncultured. Analysis of alpha and beta diversity further unveiled significant disparities between the water and slug samples. According to our data, B. stephanieae receives the majority of its microbiome vertically, not by feeding on its food anemone E. diaphana or the water column. Vertical transmission from parent to offspring is essential in ensuring the accuracy of the microbiome for the next generation (Rosenberg and Zilber-Rosenberg 2021). For instance, Porifera are known for their “sponge-specific” microbiome, inherited from the parental lineage (de Oliveira et al. 2020; Funkhouser and Bordenstein 2013). In sea slugs, little is known about the microbiome and its origin. While the microbiome has been studied in a handful of different sea slugs, its origin has not been explored (Abdelrahman et al. 2021; Davis et al. 2013; Mahadevan and Middlebrooks 2020; Ng et al. 2022). Further microbiome studies in sea slugs will enhance our understanding of their biology and also will shed light on their adaptations to changing environments and their potential roles in marine ecosystem health and resilience.

The use of RNA and subsequently cDNA for microbiome analyses comes with a few biases as well as challenges related to transcript abundance leading to an overrepresentation of certain taxa compared to others, differential expression, and temporal dynamics. However, we have decided to use RNA for our analysis because we prioritize gaining functional insights into microbial communities, understanding metabolic activities, and tracking responses to environmental stimuli. Despite potential issues such as temporal bias and RNA stability concerns, RNA sequencing allows us to identify active microbial taxa and differentiate functionally redundant species. Thus, we can delve deeper into microbial community dynamics and uncover their functional roles, ultimately enhancing our understanding of complex ecosystems (Stewart 2013; Laroche et al. 2017; Brown et al. 2022; Giroux et al. 2022).

In E. diaphana, Clostridium sensu stricto 11 sp. 2 and Rhizobiales sp. were the most abundant bacteria of the core microbiome. Clostridium are ambient pathogens, especially in soils and as normal flora in the intestines of higher organisms (Wiegel et al. 2006). Rhizobiales are diazotrophs belonging to the Alphaproteobacteria and reside in the tissue, skeleton, and mucus of several coral species (Herrera Sarrias et al. 2017; Lesser et al. 2007; Rohwer et al. 2002; Shashar et al. 1994). In addition to the photobiont, they are known to be involved in the production of bacterial dimethylsulfopropionate (DMSP) (Curson et al. 2017; Kuek et al. 2022). DMSP is a stable and soluble sulfate source and an important molecule in the marine sulfur cycle involved in several cellular processes, such as antioxidant, osmolytic functions, or cryptoprotection (Kiene et al. 2000). In our E. diaphana, the microbiome was up-regulated in “Sulfur metabolizing” by bacteria metabolizing sulfur to DMSP, while “Sulfate reducer” was mostly down-regulated. One possibility might be that the absorbed sulfate gets transformed into DMSP by Rhizobiales in the anemones providing excessive sulfate to other bacterial groups as substrate. Further, we identified two species of Endozoicomonas, a genus that is generally the highest abundant bacterial group in the coral microbiome (Bayer et al. 2013; Hernandez-Agreda et al. 2017; Meyer et al. 2014). According to Neave et al. (2016), the main functions of Endozoicomonas for the coral holobiont are nutrient metabolism, microbiome structuring, and antimicrobial activity (Neave et al. 2016).

In the slugs, Mycoplasma sp. is the most abundant bacterial genus, followed by Methylobacterium cerastii and Cytophaga sp.. Mycoplasmas are obligate parasitic, intracellular, and extracellular bacteria that can be the cause of numerous diseases in humans and vertebrates. They live on and in epithelial cells from which they obtain essential growth factors such as cholesterol, nucleotides, amino acids, and fatty acids (Johansson and Pettersson 2002; Razin et al. 1998). Still, they seem to be a key phylum of the microbiome of many marine animals such as fish but also sea slugs (Rasmussen et al. 2021, 2023; Abdelrahman et al. 2021; Davis et al. 2013; Mahadevan and Middlebrooks 2020; Ng et al. 2022). Moreover, in B. stephanieae “Sulfate metabolizing” is down-regulated, while “Sulfate reducer” is up-regulated, the opposite to its food anemone.

Nitrogen fixation, meaning the conversion of elemental dinitrogen into ammonia, is a fundamental process for corals to increase the input of fixed nitrogen into the reef ecosystem (Rädecker et al. 2015). Here, the pathway “Nitrogen fixation” is down-regulated in both E. diaphana and B. stephanieae and thus not performed by specific bacteria. Nitrogen is a limiting nutrient in the coral holobiont, and the density of algal symbionts is controlled by nitrogen availability (Falkowski et al. 1993). Thus, in anemones, the photobiont is the primary nitrogen metaboliser, and the abundance of nitrogen-cycling bacteria in corals may be dependent on the presence of a photobiont (Röthig et al. 2016). The first essential step of nitrification is ammonia oxidation, the conversion of ammonia to nitrite and further to nitrate (Lehtovirta-Morley 2018). Although nitrification (“Ammonia oxidizer”) is up-regulated in E. diaphana, resulting in nitrite, denitrifying pathways are down-regulated or not present. The photobiont prefers dissolved inorganic nitrogen in the form of ammonium (D’elia et al. 1983; Taguchi and Kinzie Iii 2001). Nitrification may eventually limit the amount of nitrogen available for the photobiont’s growth (Rädecker et al. 2015). Further, denitrification is an energy-consuming process. Xiang et al. (2022) proposed that denitrifying bacteria may be regulated by the holobiont’s nutritional status and thus contribute to the maintenance of a nitrogen-limited state (Xiang et al. 2022). In B. stephanieae “Nitrogen fixation” is down-regulated, while “Ammonia oxidizer”, as well as “Nitrite reducer”, are up-regulated. This suggests that the photobiont is still the primary nitrogen fixator, with bacteria converting the resulting ammonia to nitrite. In contrast to E. diaphana, the formed nitrite is then probably further reduced by bacteria (Additional File 5: Figure 2 A and B).

Thus, there are distinct differences between the anemone and the slug regarding their roles as holobionts in nitrogen cycling. In E. diaphana, the photobiont is the primary nitrogen fixator, and its growth is regulated through controlled nitrogen limitation. In B. stephanieae, the photobiont seems to be involved in nitrogen fixation but does not contribute to the nitrogen cycle any further and is not controlled by nitrification and/or denitrification regulation.

The bacterial metabolic pathway abundance of B. stephanieae is similar in function to aposymbiotic E. diaphana. Aposymbiotic E. diaphana show downregulation in “Sulfur metabolizing” and “Sulfur oxidizer”, whereas “Sulfide oxidizer”, “Sulfate reducer” and “Nitrite reducer” are upregulated (Röthig et al. 2016). Unlike symbiotic E. diaphana, aposymbiotic anemones are unable to produce DMSP (Van Alstyne et al. 2009) and are restricted in their sulfur cycling. Further, aposymbiotic E. diaphana show an increased abundance in denitrifying bacteria as more nitrogen is available due to the missing photobiont. We observed a similar pathway abundance pattern for B. stephanieae in the presence of its photobiont Breviolum. This indicates that the core microbiome of this sea slug is not adapted for a photobiont and could help to explain why B. stephanieae is not suited for stable photosymbiosis and lacks core molecular adaptations (Melo Clavijo et al. 2018). Microbiome analyses of Cladobranchia with a stable photosymbiosis like Phyllodesmium briareum would give valuable insights into the relevance and interplay of bacteria with the photobionts and animal host.

Furthermore, we see differences in the metabolic pathways “Chitin degradation” and “Xylan degrader”, which can be explained because E. diaphana has a different mode of life than B. stephanieae. Both pathways are up-regulated in the anemones, which is consistent with the results of other studies (Röthig et al. 2016; Herrera Sarrias et al. 2017). The anemones in our study were fed with Artemia nauplii which contain chitin, thus, it is plausible that bacteria that utilize chitin are also more abundant in anemones than in slugs. Xylan is a common carbohydrate component of plant cell walls that is required for vascular tissue development and appropriate cell wall construction (Curry et al. 2023). “Xylan degrader” is up-regulated in the anemones indicating an involvement of xylan-degrading bacteria, such as Clostridium in the carbon cycling related functions (Zverlov et al. 2005).

4.2 Influence of starvation

Starvation is a great stressor for an organism triggering a cascade of reactions and greatly impacting the microbiome in the slugs. Particularly, there was a strong decline in the diversity and abundance of bacterial groups. However, the most striking was the increase in Mycoplasma sp., which accounts for over 80% of the total core microbiome of starved slugs. Because Mycoplasmatales were also found in high abundance in other sea slugs like Chormodoris quadricolor, Pteraeolidia semperi, Elysia rufescens, and Elysia crispata, Mycoplasma sp. seems to be a significant part of the core microbiome of. B. stephanieae (Abdelrahman et al. 2021; Davis et al. 2013; Mahadevan and Middlebrooks 2020; Ng et al. 2022) but could nevertheless increase the stress of the host during starvation. The exact role of these bacteria in the heterobranch holobiont is not yet clarified. Here, reducing the microbiome composition due to starvation seems to give the Mycoplasma enough space to spread and possibly become pathological, reducing the slug’s fitness.

Starvation caused a decrease in the diversity and abundance of bacterial phyla, which was followed by severe changes in the slug’s metabolic pathways. Thus, the pathways “Dehalogenation”, “Sulphide oxidizer”, and Ammonia oxidizer” showed significant differences between fed slugs and slugs starved in light. Whereas the pathways “Xylan degrader”, “Degrades aromatic hydrocarbonates”, “Sulfur oxidizer” and “Carbon fixation” were significantly different between fed and starved in high light slugs. Worthy to note is that out of the analysed pathways, “Nitrogen fixation” was the only pathway that was significantly up-regulated in all starved slugs regardless of the light condition. B. stephanieae has an unstable symbiosis with its photobiont, which means that, unlike other Cladobranchia, the slug cannot rely solely on its photobiont for nutrition. Further, the symbionts remain photosynthetically active for up to five days before they get digested or expelled (Monteiro et al. 2019; Melo Clavijo et al. 2022). If the slug is not provided with any prey containing Symbiodiniaceae, the slug loses its photobiont turning into an aposymbiotic state. Here, we starved B. stephanieae for one week, meaning that after this week, the slug lost almost all its photobionts that provide a range of functions, including nitrogen fixation, as shown in this study. Nitrogen is considered to be the limiting factor for the photobiont. As a result, excessive available nitrogen caused by the loss of the photobiont may encourage the growth of diazotrophic bacteria, which then take over nitrogen fixation in the slug. Further, nitrifying as well as denitrifying pathways are down-regulated, indicating an insufficient nitrogen cycle in starved slugs (Additional File 5: Figure 2C).

4.3 Light and darkness

Slugs starved in high light showed the highest number of total and unique bacterial phyla compared to the other starved groups and showed the most significant differences in metabolic pathways compared to fed slugs. Additionally, B. stephanieae starved L and HL shared the most common bacterial phyla such as the cyanobacteria Microcoleus sp., Nostoc sp., and Arthrospira sp.. This is contrary to slugs starved in darkness which showed the lowest number of total bacterial phyla and a significant difference in beta-diversity to L and HL slugs. Studies that looked at the influence of light on the microbiome of photosymbiotic sponges observed similar results. The heterotrophic species Stylissa flabelliformis and Ianthella basta remained unaffected by light stress, while the phototrophic species Cliona orientalis and Carteriospongia foliascens underwent a significant change in microbiome composition and discoloured after a few days in the darkness caused by loss of Cyanobacteria (Pineda et al. 2016; Curdt et al. 2022). This alteration in the microbiome could imply an attempt at symbiont shuffling to a community that can operate better under low light conditions which has been observed for Rhopaloeides odorabile (Webster et al. 2011). However, we assume that the shift in the microbiome of slugs starved in darkness can be attributed to the lack of light. Due to the deprivation of light, the slug loses its symbiont Breviolum and phototrophic bacteria such as cyanobacteria, which gives more space to non-light-dependent bacteria such as Mycoplasmatales. Further, “Chlorophenol degrading” was the only pathway that differed significantly between HL and D and was upregulated in HL slugs. Chlorophenols are aromatic ring structures that have one or more hydroxyl and chlorine atoms attached to benzene rings and are highly toxic due to their carcinogenic, mutagenic, and cytotoxic properties. Bacteria that can use Chlorophenol as a source of carbon and energy through aerobic degradation include several species of the genus Pseudomonas, which were found to be abundant in HL slugs (Arora and Bae 2014). The significant differences in beta-diversity between slugs starved in the dark (D) and slugs starved in the light (L and HL) indicate that darkness as a stress factor has a greater impact on the microbiome composition than high light stress however with little effect on metabolic pathway abundance.

5 Conclusion

In coral research, the microbiome’s importance and the holobiont’s idea have already found wide acceptance. In terms of photosymbiotic molluscs, the cladobranch B. stephanieae is a well-studied model organism. Our results indicate a vertical transmission of the slug’s microbiome since B. stephanieae and its food source E. diaphana differed significantly in their alpha and beta diversity, as well as in their core microbiome. The microbiome of Berghia stephanieae is not accommodated to support photobionts which can explain why B. stephanieae cannot maintain its photobionts in the long term. In addition, our data show that light and starvation experiments, which are commonly used in photosymbiosis research, induce a shift in the slug’s microbiome towards an increased abundance of potential pathogens, with serious consequences for metabolic pathways such as sulphur and nitrogen cycling. These results emphasize that B. stephanieae is more than just a host to its photobiont, and it is time to apply the concept of the holobiont to other photosymbiotic animals like molluscs.

6 Supplementary information

6.1 Additional File 1

— Table 1. ASV table. Number of ASVs organised by sample and treatment, as well as after filtering and denoising. Biosample ID and Accession number are included for each sample. — Table 2. Absolute abundance of bacterial families across all samples. — Table 3. Absolute abundance of core bacteria species for all samples. — Table 4. Relative abundance of core bacteria species for all samples. — Table 5. ASVs shared with water samples. — Table 6. Results of ANOVA based on Shannon alpha diversity metric and pairwise t-tests between conditions. — Table 7. List of vectors PC1 and PC2 of PCoA plot and Shannon entropy for each sample. — Table 8. Results of PERMANOVA based on Bray-Curtis beta diversity metric and pairwise comparison of conditions. — Table 9. Shannon alpha diversity and Bray-Curtis beta-diversity between water samples and all groups. — Table 10. List of absolute pathway abundance for all samples. — Table 11. Results of the statistical analysis of the metabolic pathways between the different conditions.

6.2 Additional File 2

— Figure 2. Rarefaction curves illustrate ASV richness as a function of sequencing depth for subsampled dataset. The five groups shown correspond to E. diaphana (Symbiotic), B. stephanieae fed and starved in light (StarvedL), high light (StarvedHL), and darkness (StarvedD).

6.3 Additional File 3

— Figure 3 A - B. Venn diagrams comparing core microbiome composition of anemone, fed slugs and water. Numbers indicate total abundance of either unique or shared ASVs (A). Principal Coordinate Analysis (PCoA) plot of alpha- and beta-diversity. PCoA plot based on Bray-Curtis dissimilarity. Shape of circle represents taxon (anemone, slug or water), size of circle indicates Shannon diversity index, colour of circle represents husbandry condition fed, starved in darkness (starvedD), high light (starvedHL), regular light(starvedL), anemone or water.

6.4 Additional File 4

— Figure 4 A - M. Boxplots of post-hoc Tukey’s HSD pairwise test between conditions for each metabolic pathway with original and normalized abundance. The five groups are indicated by different colours: E. diaphana in pink, B. stephanieae fed in red, starved in light (StarvedL) in light blue, starved in high light (StarvedHL) in dark blue and starved in darkness (StarvedD) in green.

6.5 Additional File 5

— Figure 5 A - C. Differences in nitrogen cycle of E. diaphana (A), B. stephanieae fed (B) and B. stephanieae starved (C). Created with BioRender.com.