Chimeras display higher survival in natura

To investigate potential fitness differences between chimeric and non-chimeric Stylophora pistillata colonies, we assessed survival rates of experimentally produced chimeras and non-chimeras acclimated in natura at a depth of 10 m. Survival rates at 10-m depth were higher in chimeric colonies after both 6 and 12 months (Fig. 1B; χ2 = 6.53, p < 0.05; χ2 = 4.27, p value < 0.05, respectively), indicating that chimeras have a fitness advantage.

To investigate whether the higher survival of chimeras is due to better resistance to environmental stress, we translocated nine chimeras (from which five were validated by microsatellite, see below) and five non-chimeras from 10- to 2-m depth for 48 h, exposing them to sudden and higher temperature and light fluctuations (Fig. 1C; Additional file 1, Tables S1 and S2). At 2-m depth, the temperature ranged between 25.03 and 26.78°C while the temperature at 10-m depth ranged between 24.84 and 25.90°C. Differences between the two depths were even stronger for light intensities with an average sun radiation fourfold higher at 2 m (38,273 lux) than at 10 m (8262 lux). During diurnal light peak (10–16h), the maximum illumination level was sevenfold higher at 2m (85,422 lux) than at 10 m (12,400 lux). No coloration changes (e.g., bleaching or darkening) were observed for any of the corals during the experiment (coral reef watch color chart) which confirm that strong differences in Symbiodiniacaea density would not bias gene expression analyses.

Metabarcoding and dual RNA-seq were performed to check whether the above stressful environmental changes or chimeric status fostered a change in the coral colony microbiome or in the holobiont transcriptome. Before addressing this question, we validated the chimeric/non-chimeric status of the 27 colonies used for the translocation experiments using previously developed microsatellites (Additional file 1: Table S3). Among the 18 chimeras used, 11 were validated and displayed at least one locus with three or more alleles showing at least two co-occurring genomes in one colony. All seemingly non-chimeras (displaying just mono- or biallelic loci) but that could still present the undetectable state of microchimerism were removed from the metabarcoding and dual RNA-seq analyses. At the end, six and five validated chimeras and four and five non-chimeras were exposed to the 10-m or 2-m depth treatments, respectively (Fig. 1C). This higher allelic number per locus in chimeras was further confirmed using multiple nucleotide polymorphisms (MNPs) derived from the RNA-seq data (see below). This last approach showed that chimeras presented a significantly higher relative quantity of multiallelic MNPs (n allele > 3) than non-chimeric colonies (Mann–Whitney U test; U = 18, p value < 0.05; Additional file 1: Table S3).

No effect of chimerism or translocation on the microbial community

To check whether such stressful environmental change or chimeric status fostered a potential change in symbiont communities, we performed ITS2 amplicon sequencing for chimeras and non-chimeras exposed at 10- and 2-m depth. After clustering and filtering for operational taxonomic units (OTUs) containing less than 1% of sequence tags, a single ITS2 OTU, representing 97.7% of the dataset, was revealed (Additional file 1: Table S4). Taxonomic affiliation using blast comparisons to the NCBI nr/nt database enabled the identification of Symbiodinium sp. (former Symbiodinium microadriaticum clade A1 S. pistillata isolate Eilat; GenBank MH211592.1) with 100% identity over the whole amplicon sequence. Thus, we found no major differences between the symbiont communities hosted in chimeric or non-chimeric corals at both depths.

We then investigated bacterial microbiome diversity using 16S metabarcoding. We obtained 2,073,020 informative clusters (range: 28,000–263,000 per sample) from the 20 constructed libraries. After clustering and singleton filtering, 8016 OTUs were subjected to affiliation by blast against the SILVA SSU database [22] (Additional file 1: Table S5). Alpha diversity (species richness, Chao and Shannon) and beta diversity (Bray–Curtis dissimilarity) were calculated and plotted for chimeras vs. non-chimeras, with and without depth effect (Additional file 2: Fig. S1A and B). The sole significant difference observed was a lower Shannon diversity index for the non-chimeric corals at 2-m depth (MANOVA p = 0.00758; Additional file 2: Fig. S1A3). This difference was mainly driven by a single sample displaying low diversity compared to the other samples. The mean alpha diversity was higher in chimeras than in non-chimeric corals at both depths, but the difference was non-significant. Regarding the beta diversity, neither the depth nor the chimeric status had a significant effect (Bray–Curtis dissimilarity index; Additional file 2: Fig. S1B). Taxonomic composition at the family level was consistent in all the conditions and confirmed the lack of association between bacterial community composition and chimeric status or depth (Additional file 2: Fig. S1C).

Together, these data show that chimeric S. pistillata colonies have higher survival than non-chimeric counterparts and that this difference is not likely due to differences in symbiont communities or bacterial microbiome.

Chimerism induces transcriptomic changes in S. pistillata and Symbiodinium sp.

We next investigated potential changes in gene transcription in chimeras and non-chimeras at the two different depths at the whole transcriptome scale using a RNA-seq approach. This yielded an average of 25.25 ± 0.8 million paired-end reads per sample, of which ~1.2% raw sequences were discarded after the preprocessing steps (trimming, quality filtering, and adaptor removal). Most (68 ± 4.0%) of these filtered reads were uniquely mapped and properly paired on the reference genome of Stylophora pistillata [23] or Symbiodinium microadriaticum [24]. No significant differences between samples in the proportion of reads mapped to S. microadriaticum were observed which confirms the absence of bleaching (10m vs. 2m Mann–Whitney U test, U = 47.5, p value = 0.88; chimeras vs. non-chimera, U = 44.5, p value = 0.73; Additional file 1: Table S6). Hierarchical clustering analysis performed on the distance matrix based on the whole transcriptome of each samples controlled for the mother colony of origin showed that for both, S. pistillata and S. microadriaticum, samples clustered first in function of their chimeric status and then by the treatment (Additional file 2: Fig. S2A & B). This last result is confirmed by principal component analysis (PCA) showing that the first axis mostly explains the chimeric status while the second axis explains the depth (Additional file 2: Fig. S3A & B). These first results showed the strong transcriptomic effects associated to the chimeric status and then to the translocation.

When acclimated for 1 year at 10-m depth, chimeras vs. non-chimeras differentially expressed 596 genes (DEGs; FDR < 0.05; 373 over-expressed; 223 under-expressed in chimeras; Fig. 2A1; Additional file 1: Table S7). When exposed to a sudden environmental stress, 48h at 2 m, only 27 genes were differentially expressed (14 over-expressed; 13 under-expressed; Fig. 2A3). Only one gene was differentially expressed between chimeras and non-chimeras at both 10-m and 2-m depth (Fig. 2A2).

Gene ontology (GO) term enrichment analysis (GO_MWU; p value < 0.05) that compared the transcriptomes of the acclimated chimeras and non-chimeras at 10-m depth has revealed that 14 biological process were significantly enriched with 6 GO terms over-represented among genes over-expressed and 8 among genes under-expressed (Fig. 2C). The over-represented GO terms highlighted that chimeras are characterized by a higher level of processes associated to DNA and RNA replication, recombination, and integration, suggesting a high activity of transposable elements. The GO terms over-represented among genes under-expressed (Fig. 2C) showed a lower metabolic activity in chimera (GO terms: “organic acidmetabolic process,” “lipid metabolic process,” “carbohydrate metabolic process,” etc.). Individual screening of each DEGs confirmed the trends revealed at the biological process level, with chimeras displaying over-expressions of eight transposable elements (TEs) with log2 fold change (log2FD) between 9.53–2.2 (Additional file 1: Table S7). Surprisingly, it also revealed that approximately 9% of the DEGs were genes encoding proteins involved in the extracellular matrix, cell proliferation, wound healing, and tissue remodeling (Additional file 1: Table S7). These genes displayed a stochastic expression pattern composed by a mix of over- and under-expressed genes. Activation of immune pathway was also identified with the over-expression in chimera of eight different TNF receptor-associated factors (TRAF), myeloid differentiation primary response 88 (MYD88), stimulator of interferon gene (SIG), and interferon regulatory factor (IRF). Inflammation was interestingly characterized by the under-expression of two tyrosinase (melanization pathway) as well as the under-expression of the complement factors C3 and C1q (Suppl. Table 7).

Fig. 2

Chimerism induces transcriptomic changes in S. pistillata and Symbiodinium sp. (A, B) Number of genes differentially expressed, between non-chimeras and chimeras (A1–A3) or between Symbiodinium sp. hosted in non-chimeras or chimeras (B1–B3): (A1, B1) When acclimated for 1 year at 10-m depth; (A3, B3) after 48h of exposure to a sudden and stressful environmental change (translocation to 2-m depth) and (A2, B2) shared between the two environmental treatments. (C) Biological process (GO terms) significantly enriched (Mann–Whitney U test) in S. pistillata at 10m. Over-expressed/represented genes/GO terms are in red and under-expressed/represented genes/GO terms in blue (italic: p value < 0.05; normal: p value < 0.01; bold: p value < 0.001). x/y reflects the number of genes with a |log2FC| > 2 in the GO category/the total number of genes in the GO category

GO term enrichment analysis using the GO_MWU script revealed that in chimeras exposed at 2 m, no GO terms were significantly enriched. Among the 27 DEGs (Additional file 1: Table S8), a gene encoding an acid ceramidase involved in the sphingolipid biosynthesis pathway (a regulatory pathway involved in many cellular processes including innate immunity, acute inflammatory responses, and activation of immune cells and apoptosis [25]) and two genes encoding proteins of the extracellular matrix were under-expressed.

The stochastic expression patterns of genes involved in cell–cell interactions and tissue organization probably highlight physical interactions between the genotypes co-occurring within the chimera. The over-expression of TEs in chimeras suggests that these physical interactions between cells with different genetic backgrounds may lead to a stress at the genome scale. Therefore, chimerism appears to induce transcriptomic changes characteristic of internal conflict that may occur between the different co-occurring genotypes.

When acclimated for 1 year at 10-m depth, differential gene expression analysis showed that 1111 genes were differentially expressed (628 over-expressed; 483 under-expressed; FDR < 0.05; Fig. 2B1, Additional file 1: Table S9) in symbiotic algae hosted in chimeras vs. those hosted in non-chimeras. By contrast, only a single gene was over-expressed in chimeras after 48 h at 2 m (Fig. 2B3, Additional file 1: Table S10). No genes were differentially expressed both at 10 and 2 m (Fig. 2B2).

GO_MWU analysis showed that no biological processes were significantly enriched at 10-m or 2-m depth. At 10 m, many DEGs were involved in bicarbonate transport or conversion (e.g., electrogenic sodium bicarbonate cotransporter, carbonic anhydrase), nitrate transport (e.g., high-affinity nitrate transporter), or ion transport (e.g., sodium channel protein type 8, potassium voltage-gated channel protein etc.). This trend suggests an effect of chimerism on the organic/inorganic matter uptake/excretion function of the symbiont (Additional file 1: Table S9).

Gene expression of the Symbiodinium sp. is surprisingly affected by the coral chimerism while generally acknowledged as relatively stable. The regulation of biological functions linked to ion transport and nutrient uptake may highlight a change in the trophic interactions between the cnidarian host and its symbiont.

Chimera frontload stress-responsive genes

In response to the translocation from 10- to 2-m depth, non-chimeric colonies differentially expressed 327 genes (105 over-expressed and 222 under-expressed; Fig. 3A1; Additional file 1: Table S11). By contrast, chimeras differentially expressed 131 genes (34 over-expressed, 97 under-expressed; Fig. 3A3; Additional file 1: Table S12). Comparisons of these two responses showed that only 10 genes were differentially expressed in both responses (Fig. 3A2).

Enrichment analyses revealed that 38 GO biological processes were significantly enriched (p value < 0.05; 11 were over-represented among genes over-expressed and 27 among genes under-expressed) in non-chimeric colonies (Fig. 3B). In chimeras, 42 GO categories were enriched (35 over-represented, 7 under-represented; Fig. 3C). Comparisons between each enrichment results revealed that the overall pattern of responses was close for both entities (Fig. 3D).

Fig. 3

Transcriptomic modifications induced by the translocation in chimeras and non-chimeras of S. pistillata. (A) Number of genes differentially expressed in response to the translocation in non-chimeras of S. pistillata (A1) and in chimeras of S. pistillata (A3) or shared between entities (A2). (B, C) Biological processes (GO terms) significantly enriched (Mann–Whitney U test): (B) in non-chimeras of S. pistillata; (C) in chimeras of S. pistillata. Over-expressed/represented genes/GO terms are in red and under-expressed/represented genes/GO terms are in blue. p values under the 0.05, 0.01, and 0.001 thresholds are indicated in italic, in normal font, and in bold, respectively. x/y reflects the number of genes with a |log2FC| > 2 in the GO category/the total number of genes in the GO category. (D) Delta-rank comparison between the responses of non-chimeras and chimeras

In non-chimeras, we identified biological processes classically enriched in coral response to light and temperature stress, including processes associated to reactive oxygen species (ROS) detoxification (e.g., oxidation-reduction process; over-represented), immunity (e.g., regulation of I-kappa B kinase/NF-kappa B signaling, under-represented), and stress response (e.g., regulation of response to stress, under-represented). In chimeras, GO terms such as cellular response to stress and protein folding (over-represented) were also found. At the single gene level, a gene-by-gene screening for stress-responsive genes among the DEGs expressed during the response of both entities revealed more significant over-expressions in non-chimeras of (i) three members of the HSP family (e.g., HSP70 and HSF) in the non-chimeras; (ii) 10 (e.g., thioredoxin and peroxiredoxin) and three (e.g., cytochrome P450) ROS scavengers in non-chimeras and chimeras, respectively; (iii) 10 immune-related genes in non-chimeras (e.g., complement C3 and TRAF3); and (iv) four genes involved in cell death pathways (e.g., p53) in non-chimeras (Suppl. Tables 11 and 12). This screening showed that the stress response induced by the translocation in non-chimeras is stronger than in chimeras.

To better understand the differences in gene expression patterns between non-chimeric colonies and chimeras, we focused on the 327 genes responding to the translocation of the non-chimeric colonies and looked how these genes were expressed in the chimeras acclimated to the environment and in response to the translocation to 2-m depth (Fig. 4A). These differences were highlighted accordingly to their transcriptomic plasticity and using the following categorization: (i) frontloaded genes (i.e., higher basal expression level in chimera), (ii) higher plasticity (i.e., genes with identical or lower basal expression but over- or under-expressed higher in chimeras in response to the translocation), and (iii) frontloaded and higher plasticity (higher basal expression and higher over- or under-expression in chimeras in response to the translocation). The analyses revealed that 215 genes followed one of these patterns: 111 were frontloaded, 56 displayed higher transcriptomic plasticity, and 48 were frontloaded and displayed higher transcriptomic plasticity (Fig. 4A). GO term enrichment analyses were then performed on these different sets of genes.

Fig. 4

Chimeras frontload stress-responsive genes (A, C). Expression pattern differences between non-chimeras and chimeras for the genes responding to the translocation in the non-chimeras: A for the host and C for the symbiont. B, D GO categories from the biological process roots that are significantly enriched (Fisher exact test; FDR < 0.05), in the frontloaded gene set for the host (B) and the symbiont (D). p values under the 0.05, 0.01, and 0.001 thresholds are indicated in italic, in normal font, and in bold, respectively; x/y reflects the number of DEGs in the GO category/the total number of genes in the GO category

Significant enrichments were obtained for the frontloaded set only (p value < 0.05; Fig. 4B). In particular, 8 GO terms mostly illustrating a higher activity linked to the energetic metabolisms at the basal expression level (e.g., glucan metabolic process, energy derivation by oxidation of organic compounds, etc.) were enriched. Interestingly the GO term oxidation-reduction process was also composed by genes involved in ROS detoxification, a key coral stress response.

At the single gene level, a screening among all the genes displaying a putative adaptive expression pattern in chimera revealed that 5 ROS detoxifiers (e.g., peroxidasin, peroxiredoxin, etc.) were frontloaded and one (a thioredoxin) displayed a higher plasticity. Among the HSP family, the HSF and HSP16 genes were frontloaded and the HSP70 gene was frontloaded and higher expressed in response to the translocation. In addition, seven genes involved in apoptosis and cell death pathways (e.g., FAS-associated factor 1 and p53) and 10 immune-related genes (e.g., complement C3 and galectin) were also frontloaded.

In non-chimeric colonies, the stress response was more potent, but most of the essential genes needed to display an efficient stress response were already expressed constitutively at higher levels in acclimated chimeras and/or were regulated higher. These different forms of transcriptomic plasticity may increase the robustness of chimeras.

Symbiodinium sp. hosted in chimeras and non-chimeras respond differently

After translocation of the S. pistillata colonies to 2-m depth, the endosymbiotic algae in the non-chimeric colonies differentially expressed 157 genes (122 over-expressed and 35 under-expressed; Fig. 5A1; Suppl. Table 12) compared to 175 genes in the algae within the chimeras (42 over-expressed and 133 under-expressed; Fig. 5A3; Suppl. Table 13). Only seven DEGs were commonly found in Symbiodinium sp. from chimeric and non-chimeric colonies in response to the translocation (Fig. 5A2). In non-chimeric colonies, we identified five significantly (p value < 0.05) enriched GO categories (Fig. 5B). The three GO terms over-represented among over-expressed gene were associated to energy production and photosynthesis (e.g., generation of precursor metabolite and energy, photosynthesis). In chimeras, no GO terms were significantly enriched at the 5% error level. Comparison between each enrichment results revealed that the five GO terms enriched in the Symbiodinium sp. hosted in non-chimera displayed close delta-rank value to those hosted in chimera which illustrate that only the most significant process responds in the same way (Fig. 5C).

Fig. 5

Transcriptomic modifications induced by the translocation in Symbiodinium sp. hosted in chimeras and non-chimeras. (A) Number of genes differentially expressed in response to the translocation in non-chimeras (A1) and in chimeras (A3) or shared between entities (A2). (B) Biological processes (GO terms) significantly enriched (Mann–Whitney U test) in non-chimeras. Over-expressed/represented genes/GO terms are in red and under-expressed/represented genes/GO terms are in blue. p values under the 0.05, 0.01, and 0.001 thresholds are indicated in italic, in normal font, and in bold, respectively. x/y reflects the number of genes with a |log2FC| > 2 in the GO category/the total number of genes in the GO category. (C) Delta-rank comparison between the responses of non-chimeras and chimeras

For non-chimeras and in accordance with the results of the enrichment analysis, many genes related to photosynthesis and energy production (i.e., light harvesting complex, oxygen-evolving enhancer, ATP synthase) were differentially expressed. Interestingly, six genes involved in stress response were over-expressed, four cytochrome genes (C1, C6, B5, and B6), one HSP90, and one ferredoxin (Suppl. Table 12). In chimeras, genes involved in photosynthesis were also retrieved over-expressed (e.g., photosystem I reaction center subunit II and oxygen-evolving enhancer protein 1) with only two genes involved in stress response, a HSP70 and a cytochrome peroxidase (Suppl. Table 13).

Similarly to the host, transcriptomic plasticity differences between chimeras and non-chimeras were studied. This was done for the DEGs characterizing the responses of the Symbiodiniaceae hosted in non-chimeras by categorizing the following three patterns: (i) frontloaded genes (i.e., higher basal expression level in chimera), (ii) higher plasticity (i.e., genes with identical or lower basal expression but over- or under-expressed higher in chimeras in response to the translocation), and (iii) frontloaded and higher plasticity (higher basal expression and higher over- or under-expression in chimeras in response to the translocation). As for the host, frontloading was the predominant category (Fig. 4C) and the only one for which significant enrichment was obtained (Fig. 4D). Among the four biological process enriched, three were linked to photosynthesis and energy production (“photosynthesis,” generation of precursor metabolites and energy, and protein-chromophore linkage). At the gene level, gene encoding protein involved in photosynthesis and energy production was retrieved as two genes involved in thermal and light stress response, a flavodoxin and a ferredoxin.

In summary, Symbiodinium sp. hosted in non-chimera colonies showed a slight activation of genes and pathways to respond to light and thermal stress. This activation can also be detected in chimeras but it involved very few genes, suggesting a better control of the stress in the holobiont. Interestingly, Symbiodinium sp. hosted in chimeras displayed a higher level of expression for genes involved in photosynthetic activity that may results in a higher quantity of energy and energy reserves.

Chimerism reduces transcriptomic plasticity in S. pistillata

To assess the degree of transcriptomic plasticity of chimeras compared to non-chimeras, we first conducted a discriminant analysis of principal component (DAPC) based on the transcriptomes of acclimated (i.e., acclimated at 10-m depth) and stressed (i.e., translocated at 2-m depth) non-chimera colonies. We next predicted the coordinates of chimera based on their overall gene expression levels onto the DAPC first discriminant axis. In S. pistillata and according to the resulting DAPC, the overall gene expression profiles of the control and stressed chimera showed more overlap than the overall transcriptomic profiles of non-chimera under stress versus control conditions (Fig. 6A).

Fig. 6

Chimerism reduces transcriptomic plasticity in S. pistillata. Discriminant analysis of principal component (DAPC) illustrating the transcriptomic plasticity and the transcriptome diversity (X axis) expressed in (A) S. pistillata and (B) Symbiodinium sp. hosted in non-chimeras and in chimeras

The DAPC analyses performed from the overall gene expression datasets obtained from the Symbiodinium sp. associated with non-chimeric colonies vs. chimeras showed a similar, yet less pronounced, pattern (Fig. 6B). These DAPC analyses illustrate that chimerism reduces transcriptomic plasticity and increases transcriptomic diversity (e.g., more genes expressed over a wider range of levels) at the overall holobiont level but with a much stronger effect in the host.