Transcriptional response of the calcification and stress response toolkits in an octocoral under heat and pH stress

Up to one‐third of all described marine species inhabit coral reefs, but the future of these hyperdiverse ecosystems is insecure due to local and global threats, such as overfishing, eutrophication, ocean warming and acidification. Although these impacts are expected to have a net detrimental effect on reefs, it has been shown that some organisms such as octocorals may remain unaffected, or benefit from, anthropogenically induced environmental change, and may replace stony corals in future reefs. Despite their potential importance in future shallow‐water coastal environments, the molecular mechanisms leading to the resilience to anthropogenically induced stress observed in octocorals remain unknown. Here, we use manipulative experiments, proteomics and transcriptomics to show that the molecular toolkit used by Pinnigorgia flava, a common Indo‐Pacific gorgonian octocoral, to deposit its calcium carbonate skeleton is resilient to heat and seawater acidification stress. Sublethal heat stress triggered a stress response in P. flava but did not affect the expression of 27 transcripts encoding skeletal organic matrix (SOM) proteins. Exposure to seawater acidification did not cause a stress response but triggered the downregulation of many transcripts, including an osteonidogen homologue present in the SOM. The observed transcriptional decoupling of the skeletogenic and stress‐response toolkits provides insights into the mechanisms of resilience to anthropogenically driven environmental change observed in octocorals.

of diverse groups, including algae, sponges and other cnidarians such as corallimorpharians and octocorals, in these ecosystems (see Norström et al., 2009 for a review).
Octocorals (e.g., soft-corals, gorgonians) are common and important community members in many marine ecosystems. They increase the spatial complexity of the habitats they inhabit (Quattrini et al., 2014) and provide refuge to numerous invertebrate species (Buhl-Mortensen & Mortensen, 2005;Cúrdia et al., 2015). In shallowwater ecosystems, such as coral reefs, octocorals can outgrow stony corals and become dominant after extreme climatic events, such as anomalously strong El Niño events (Ruzicka et al., 2013), or under extreme environmental conditions, such as those prevailing in volcanic-seep acidified waters (Inoue et al., 2013). For instance, in Florida, octocorals increased in abundance in the 11 years following the 1997/98 El Niño event, becoming the most abundant taxon in some localities and dominating shallow fore-reefs (Ruzicka et al., 2013). Similarly, the soft coral genus Rhytisma became dominant after a significant coral bleaching event in the Aldabra Atoll (Indian ocean) in 1998 (Spencer et al., 2005), and the "blue coral," Heliopora coerulea, a species of the reef-building zooxanthellate octocoral clade Helioporacea, increased its abundance to >50% over a decade in the Bolinao Reef Complex in the northern Philippines (Atrigenio et al., 2017). Like scleractinians, helioporacean octocorals produce a rigid skeleton made of aragonite (a polymorph of calcium carbonate) (Colgan, 1984), but their calcification machinery appears to be more resilient to environmental stressors than that of most scleractinians (Kayanne et al., 2002;Shaish et al., 2010). Thus, octocorals may be possible winners in future oceans due to global climate change, potentially replacing scleractinians as the main reef framework-building organisms (Inoue et al., 2013).
Studies on scleractinian corals indicate that these organisms' response to anthropogenically driven climate change depends on the stress source and differs among species and ontogenetic stages (Davies et al., 2016;Moya et al., 2012;Thomas et al., 2018). In Acropora hyacinthus, heat stress triggers a large and dynamic transcriptomic response characterized by the modulation of different metabolic and cell cycle processes in the early stress response and of spliceosome activity, RNA-and DNA-related metabolic processes, cell stress, and cell metabolism before bleaching onset (Seneca & Palumbi, 2015). A meta-analysis of the transcriptomic response to different stress sources in 10 Acropora species identified an "environmental stress response" toolkit consisting of genes involved in the response to reactive oxygen species, protein folding and degradation, NF-κB signalling, immune response, and cell death (Dixon et al., 2020) that corals exposed to environmental stress consistently modulated. In contrast, ocean acidification does not appear to trigger a pronounced transcriptional response in adult coral colonies (Davies et al., 2016;González-Pech et al., 2017) but elicit the modulation of several calcification-related genes in the primary polyps of Acropora millepora (Moya et al., 2012).
Despite the observed higher tolerance of octocorals to climate change-induced stress, their physiological response to these stimuli remains poorly studied. Wiens et al. (2000) and Shimpi et al. (2016) observed heat-shock protein overexpression in thermally stressed Dendronephthya klunzingeri and Sinularia cf. cruciata colonies, respectively. Similarly, the sea pen Veretillum cynomorium displayed elevated heat-shock protein concentrations upon exposure to thermal stress, but this species did not show increased antioxidant or lipid peroxidation activities (Lopes et al., 2018). To date, aside from these limited data on the octocoral response to environmental stress, no assessment of the transcriptomic response of octocorals has been made, and the resilience mechanisms used by these organisms to tolerate and, eventually, outgrow and outcompete less resilient organisms, such as stony corals, remain largely unknown.
However, since growth in octocorals requires the deposition of new calcium carbonate skeletal elements (calcite sclerites, or aragonite fibres in case of the blue corals) to support the colony structurally (Lewis & Vonwallis, 1991), resilience to climate change in this group must be linked to their ability to sustain calcification under challenging environmental conditions (Gabay et al., 2013(Gabay et al., , 2014Gómez et al., 2015;Inoue et al., 2013). Indeed, except for the Helioporacea, the octocoral tissues effectively isolate sclerocytes from the surrounding seawater (Gabay et al., 2014), putatively allowing these cells to sustain the expression of transcripts involved in calcification under situations of environmental stress, and allowing the colonies to outgrow and outcompete less resilient organisms. To test this hypothesis, we characterized the skeletal proteome of the common Indo-Pacific gorgonian octocoral Pinnigorgia flava and used manipulative experiments and transcriptomics to assess the effect of heat and pH stress on the expression of transcripts involved in calcification in this species. Together, our data provide a first assessment of the transcriptional response of an octocoral to environmental stress factors induced by anthropogenically driven climate change and give insights into the octocoral resilience mechanisms.

| Experimental model and subject details
We used clonal explants of a colony of Pinnigorgia flava (Nutting, 1910) kept at 25°C and a 12:12-h light cycle in a 642-L marine aquarium system at the Department of Earth-and Environmental Sciences,

Paleontology & Geobiology, Ludwig-Maximilians Universität
München (see below). P. flava is a colonial, zooxanthellate soft coral (Octocorallia) largely endemic to the Great Barrier Reef but sporadically found in the coral triangle in SE Asia. The studied colony has been cultured in our research aquaria for more than a decade, but its exact geographical origin is unknown. It forms nonanastomosing, pinnated light purple colonies with brownish polyps. The sclerites are white to yellow bent spindles, capstans and rods. Currently, P. flava belongs to the family Gorgoniidae.

| Determination of calcification hotspots along the body axis of P. flava
We used calcein, a calcium-binding fluorescent dye that permanently incorporates into the newly formed skeleton, to investigate the distribution of calcification sites along the body axis of P.
flava (Holcomb, Cohen, & McCorkle, 2013/2). We incubated three colonies of P. flava for 72 h in a glass container with 500 ml of a 50 µg ml -1 calcein disodium salt (Sigma-Aldrich) in 0.2 µm filtered artificial seawater ( Figure S1). We exchanged the seawater with fresh seawater+calcein every 24 h. After staining, we fixed the colonies in 80% EtOH and stored them at ~5°C until further processing.
To assess whether calcification preferentially occurs on the tip of the colonies or, by contrast, the calcification hotspots occur along the colony body axis, we cut the colonies in top and bottom sections using a sterile scalpel and placed each piece in 1.5 ml microcentrifuge tubes containing 1 ml of sodium hypochlorite (NaOCl 10%; Fluka).
After a 3-h bleach incubation, we rinsed the sedimented sclerites six times with distilled water and stored them in 80% ethanol. We then placed a sample of sclerites onto glass slides and embedded them in Eukitt quick-hardening mounting medium (Fluka Analytical) before covering the sample with a glass coverslip.
We observed the sclerites under epifluorescence (excitation band-pass filter 420-490 nm, barrier long-pass filter 515 nm) on a Leica DMLB microscope coupled to a Leica DFC 480 camera and an I3 filter set. We exposed the stained sclerites for 10 s and acquired pictures using Leica Application Software "las version 4.5." To determine the number of stained sclerites per colony region, we sampled the top and bottom fragments of the colonies and counted stained and total sclerites per visual field at 100× magnification along one horizontal transect crossing the slide from left to right.

| Proteomic analysis of the skeletal organic matrix of P. flava's sclerites
To determine the skeletal proteome of P. flava, we sampled four colonies of about 4 cm in length and incubated them in sodium hypochlorite (5%; Fluka) for 72 h under moderate shaking (30 rpm.; IKA Rocker 3D digital). We then rinsed the sedimented sclerites six times with Milli-Q water and dried them at 37°C for 24 h. This procedure yielded approximately 0.75 g of dry sclerites, which we ground with a mortar and pestle before incubating again in sodium hypochlorite (2%) for 4 h under moderate mixing (30 rpm). After bleaching, we rinsed the powder six times with Milli-Q water and dried it overnight at 37°C. To dissolve the calcitic mineral, we incubated the dry powder in 10% acetic acid overnight under moderate mixing (20 rpm).
We centrifuged the resulting solution at 13,500 rpm for 30 min to separate the acetic-insoluble matrix (AIM) from the acetic-soluble matrix (ASM). To isolate the ASM, we centrifuged (4600 rpm for 70 min at 16°C) the supernatant through 15-ml Amicon ultrafiltration devices with a 3-kDa cutoff membrane and added four volumes of methanol, one volume of chloroform and three volumes of Milli-Q water to one volume of desalted solution to precipitate the proteins by centrifugation at 5500 rpm for 15 min (Wessel & Flügge, 1984).
After discarding the upper phase, adding three volumes of methanol and centrifuging the sample at 5500 rpm for 15 min, we airdried the resulting ASM pellet and resuspended both the ASM and AIM fractions in 95% Laemmli buffer +5% β-mercaptoethanol. We used a one-dimensional sodium dodecyl sulphate-polyacrylamide (SDS-PAGE) gel (Mini-PROTEAN Tetra System; Bio-Rad) to separate electrophoretically the skeletal organic matrix proteins before mass spectrometry. To visualize the extracted skeletal organic matrix (SOM) protein fractions, we ran an SDS-PAGE for 90 min at 80 V, increasing the voltage to 100 V after the gel front passed the boundary between the stacking and the resolving gel. We used the Precision Plus Protein Dual Xtra Standard (Bio-Rad, 12 band marker, 2-250 kDa) as a size standard and stained the gel after fixing for 20 min in a fixation solution (50% ethanol, 40% Milli-Q and 10% acetic acid), washing in 30% ethanol for 10 min, and in Milli-Q water for 10 min, with silver nitrate using the Proteo Silver Plus Silver Stain Kit (Sigma-Aldrich). For this, we incubated the gel in sensitizer solution for 10 min, washed it as described above, and equilibrated it for 10 min in the silver solution. Before developing, we washed the stained gel for 1 min in Milli-Q water and submerged it in a developing solution for 5 min. After stopping the development reaction, we washed the gel for 15 min in Milli-Q water. We used an orbital shaker at 60 rpm for all steps described above. For mass spectrometry, we ran the SDS-PAGE with the six technical replicates of the extracted SOM protein fractions for 40 min at 80 V until the protein extracts passed the boundary between the stacking and the resolving gel, and manually excised the bands with a sterile scalpel. We then subjected the isolated proteins to alkylation, reduction and tryptic digestion (0.1 μg μl -1 trypsin at 37°C, overnight). We used an LTQ Orbitrap mass spectrometer (Thermo Fisher Scientific) coupled with a Rheos Allegro liquid chromatograph (Flux Instruments) to analyse 3 μl of the digested sample after separation using a selfmade column of 75 µm diameter, 15 cm length, C18 particles of 2 µm diameter and 100 Å pore size (Dr. Maisch). We prepared the mass spectrometry (MS)-grade mobile phases as follows: (A) water containing 10% acetonitrile (ACN), (B) ACN containing 10% water, each combined with 0.1% formic acid. We used the following gradient: 40 min (0%-23% B), 40 min (23%-85% B), 5 min (85%-100% B), 25 min (100% B), 3 min (100%-0% B) and 20 min (0% B) for reequilibration, and a constant 40 µl min -1 flow at room temperature (22°C). We used the programs xcalibur 2.0 (Thermo Fisher Scientific) and maxquant version 1.5.2.8 (Cox and Mann 2008, https://doi. org/10.1038/nbt.1511) to acquire and analyse the MS/MS data, respectively. We estimated protein abundances as relative iBAQ values (Schwanhäusser et al., 2011) of the detected SOM proteins using the MaxLFQ algorithm (Cox et al., 2014). We filtered the LC-MS/ MS results to remove hits from known conventional contamination sources using the common Repository Adventitious Proteins (cRAP) database before mapping the peptides against a transcriptome reference of P. flava (Conci et al., 2019). We translated the transcripts matching proteins present in the SOM fractions and annotated them against the UniProtKB database using blastp (https://blast.ncbi.nlm. nih.gov/) and signalp 4.0 (Petersen et al., 2011) to predict the presence of signal peptides, transmembrane regions and GPI anchors within the predicted protein sequences. Additionally, to assess the distribution of the detected P. flava SOM proteins among cnidarians, we screened a database composed of 120 transcriptomes from representatives of this phylum using P. flava's SOM proteins as queries for the search (for details see Eitel et al., 2018).

| In vivo experiments
We used a 360-L marine aquarium under a 12:12 h light cycle controlled by GHL Mitras LX 6200-HV LED lights that yielded 10 kLux (photosynthetically active radiation = 250 μmol m -2 s -1 ) at the water surface. Based on hourly measurements over one year (2017), the system's average water temperature and pH were 24.92 ± 0.24°C and 8.30 ± 0.14, respectively. Based on weekly measurements over one year (2017), the average PO 4 −3 , NO 2 − and NO 3 − concentrations in the water were 0.092 ± 0.071, 0.014 ± 0.072 and 2.681 ± 3.882 mg L − ; the concentration of NH 3 /NH 4 + in the water was consistently below detection (i.e. <0.05 mg L − ). Weekly or daily monitoring in subsequent years revealed that these values were stable and can be taken as a baseline for the system. We measured the , NO 3 − and NH 3 /NH 4 + concentrations every other day, and monitored water pH, density and temperature (manually twice per day) every day (Tables S1-S3) during the experiments. We also measured the nutrient profile in the 10-L tanks at the end of the experiment. The measured concentrations did not differ from the baseline concentrations for the tank.
To assess the effect of heat stress on P. flava, we randomly assigned nubbins (n = 18) to six aquaria (10 L) filled with ~6 L of artificial seawater and partially immersed in the 360-L aquarium described above. Water evaporation in the 10-L tanks was compensated every day with water filtered by reverse osmosis. We used a submersible water pump (300 L h -1 ; Eheim) to provide adequate water mixing in each tank. For the temperature experiment, after an acclimation period of 4 days, we randomly selected three tanks and gradually increased the water temperature to 29-30°C for 5 days (~1°C per day) using 50-W water heaters (Eheim). We then kept the P. flava colonies at 29-30°C for 3 days. The bleaching threshold of P. flava is not known, but the colony used for our experiments has grown >10 years at 25°C. We used this temperature baseline to selected a final water temperature (i.e., 29-30°C) lying >4°C over the control temperature conditions under which this coral grows in the aquarium. Thus, the increase in temperature from 25°C to 29-30°C is expected to cause a heat stress response in P. flava without necessarily causing its death. In scleractinian corals, for example, bleaching can be induced by heat stress only about 1°C above the average summer maximum temperature (Glynn & D'Croz z, 1990). We visually monitored the colonies, which had their polyps everted and did not show any visible signs of bleaching during the experiment. In this regard, the bleaching threshold is not known for P. flava. At the end of the experiment, we cut octocorals in two (i.e., lower and upper) sections using sterile scissors and flash-froze them in liquid nitrogen before storing them at −80°C until further processing.
We assessed the effect of seawater acidification on 12 colonies of P. flava distributed in two 30-L tanks (control and treatment) connected to a 320-L salt-water "mother" tank. To lower (0.1 pH units drop per day) the seawater pH, we pumped CO 2 into the treatment tank (see details in González-Pech et al., 2017) to achieve a pH of 7.8 (from a starting value of 8.2). We kept the pH stable for 3 days and lowered it again to 7.6 over 3 days. After reaching a pH of 7.6, we allowed the octocorals to acclimate for 3 days before decreasing the pH to its final value of 7.3. Gomez et al. (2015) showed that Eunicea fusca grows under a broad range of seawater pH values but possibly experiences some level of decalcification at pH 7.1. Therefore, we used a pH of 7.3 to evaluate P. flava's response to seawater acidification as this value represents the lowest reported pH at which octocorals continue growing without sclerite decalcification. We maintained this pH for 2 months using an automatic pH computer and monitored it throughout the experiment using a PCE-PHD 1 datalogger (PCE Instruments). The pH of the "mother" and control tank was 8.2 throughout the experiment. The temperature and nutrient levels in the "mother" tank were as detailed above. During the entire experimental phase, the coral colonies showed everted polyps and no signs of stress (e.g., bleaching). At the end of the experiment, we cut the octocorals at the base using sterile scalpels and then flash-froze them in liquid nitrogen before storing them at −80°C until further processing.

| Analysis of differential gene expression
We extracted total RNA from the upper section of the octocorals using the Direct-zol RNA MiniPrep kit (Zymo Research) following the manufacturer's protocol. We assessed the purity and integrity of the RNA extracts using a Nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific) and a Bioanalyzer 2100 (Agilent  (Patro et al., 2017). We truncated the resulting (holobiont) pseudocounts matrix and analysed it using deseq2 (Love et al., 2014) to determine differentially expressed genes (DEGs) in (1) (Yu et al., 2010) to calculated the pairwise semantic similarity (Wang) between GO-terms and clustered (Ward) the enriched GO-terms into groups of semantically related GO-terms. To assign a representative GO-term to each cluster of semantically related, significantly enriched GO-terms in the different stress treatments, we took advantage of the hierarchical structure of the gene ontology and retrieved subgraphs including the GOterms in each cluster. These subgraphs show how semantically similar GO-terms relate to each other in the more broad gene ontology directed acyclic graph. We used these subgraphs to find GO-terms from which other GO-terms in a cluster derived (i.e., "ancestor" GOterms) and used these more general GO-terms as cluster descriptors.

| Calcification in P. flava occurs along its entire body axis
Visual Despite the increased number of stained sclerites at the colony's top, we could not detect any evident calcification hotspots along the colony axis.

| Taxonomically widespread and restricted proteins form the skeletal proteome of P. flava
After filtering potential known contaminants (e.g., keratins, trypsin), we identified a total of 27 transcripts as the SOM proteome of P.
flava. Label-free quantification (LFQ) revealed 14 proteins with a higher abundance in the AIM, seven proteins exclusive to the ASM and five proteins shared by both fractions (Figure 2). Biochemically, seven of the detected SOM proteins are membrane-bound, con- abundances >5% in the skeletal proteome of P. flava. The two acidic proteins ranked first and third, osteonidogen and collagen were the second and fourth most abundant proteins, and galaxin and agrin ranked fifth and sixth in abundance. These six proteins account for 90% of the total iBAQ-derived skeletal protein abundance (Figure 2).
On average, the remaining proteins accounted for only 0.5% (±0.6%) of the total iBAQ-derived abundance.
In terms of their taxonomic distribution within Cnidaria (Figure 2), similarity searches revealed that the skeletal proteome of P. flava includes proteins with a widespread distribution in this group, such as all detected collagens, agrin, osteonidogen, galaxin and several enzymes, among others. Other proteins, such as hemicentin or the protein kinase Nell 1, displayed a patchier occupancy and were found mostly in other anthozoans. Finally, only a few SOM components, namely the two acidic proteins found, a serine protein-kinase receptor and a protein similar to laminin, had a restricted taxon occupancy with significantly similar proteins found almost exclusively in other octocorals (i.e., order Alcyonacea).

| Heat and pH stress does not modulate the calcification toolkit of P. flava
Sublethal, high-temperature (~29-30°C, ~5°C above long-term average in the aquaria) seawater caused a similar (global) transcriptional response in the exposed colonies ( Figure S2 Detailed information about the GO-terms included in each cluster,

Fisher's test values and the information content of each GO-term
in the clusters is provided in Tables S6-S12. We found seven clus- test values, and the information content of each GO-term in the clusters is provided in the Tables S13-S19.
In contrast to the response of the coral, heat stress resulted in the modulation of only 19 transcripts in P. flava's symbionts; nine of these transcripts had a log-fold change ≥1 and only one transcript had log-fold change ≤ −1 (Table S20).
In contrast to heat stress, pH stress did not trigger a stark stress response in P. flava, only causing the significant (p <.05) downregulation of a set of 70 host transcripts, 61 of which had a log 2 fold change ≤ −1 (Figure 6a; Figure S31 and Table S21). We found one calcification-related, uncharacterized skeletal matrix protein and two SOM-encoding transcripts, namely osteonidogen-the second most abundant SOM protein-and a prosaposin-homologue downregulated in pH-stressed corals (Figure 6b). We did not detect changes in the expression of stress-related proteins from the heat-shock protein family (Figure 6b). Our GO-term enrichment analysis resulted in nine clusters of semantically similar, significantly enriched BP GOterms (Figure 7; Figures S32-S49). These clusters were related to the "regulation of metalloendopeptidase activity" (GO:1904683), "protein localization to synapse" (GO:0035418), "regulation of bone mineralization" (GO:0030500), "regulation of biomineral tissue development" (GO:0070167), "terpenoid metabolic process" (GO:0006721), "interaction with host" (GO:0051701), "response to BMP" (GO:0071772), "transmembrane receptor protein serine/threonine kinase signaling pathway" (GO:0007178), "regulation of leucocyte proliferation" (GO:0070663) and "adaptive immune response" F I G U R E 3 (a) Expression (variancestabilized, z-scored transformed counts) of 76 transcripts modulated in Pinnigorgia flava colonies exposed to control and heat stress. Only transcripts with log 2 fold change ≥ |1| and p < .01 were included in the heat plot for visualization purposes. (b) Change in gene expression of 27 SOM proteins, calcification-related proteins sensu Conci et al. (2019), carbonic anhydrases, galaxins and heat-shock proteins in P. flava colonies exposed to heat stress. Asterisks (*) indicate transcripts significantly differentially expressed in control vs. treatment samples (GO:0002250). We found one cluster grouping largely unrelated GO terms difficult to summarize with a single representative GO-term.
Detailed information about the GO-terms included in each cluster,

Fisher's test values and the information content of each GO-term in
the clusters is provided in the Tables S22-S30. pH stress did not result in the upregulation of any transcript in P. flava or the modulation of any transcripts in P. flava's symbionts.

| DISCUSS ION
Sclerite growth in octocorals involves the synthesis by sclerocytes of SOM proteins and their transport to a sclerite-forming vacuole where primordia, mostly made of irregularly shaped CaCO 3 crystals, form (Kingsley, 1984). Sclerite primordia continue growing by the deposition of more regular crystals in an extracellular space created by multiple sclerocytes (Goldberg & Benayahu, 1987). In Pinnigorgia flava, our results indicate that the deposition of new sclerites occurs throughout the colony axis, increasing toward the colony tips.
This pattern of calcification is similar to that of Leptogorgia virgulata, one of the few other octocoral species where data on calcification dynamics exist (Kingsley & Watabe, 1989). It indicates that active sclerocytes intersperse along the colony axis and that under normal conditions, we should not expect spatial differences in the expres-  (Conci et al., 2020). This trend is particularly evident among the most abundant proteins in the SOM.
Among those, galaxin and the two acidic proteins detected had a very narrow taxonomic distribution. Galaxins mainly occur within Cnidaria (Conci et al., 2019), and acidic proteins appear to evolve rapidly; species-specific proteins are difficult to assign to cnidarian orthology groups using similarity searches (Conci et al., 2020).
Our results are consistent with that observation, as the two acidic proteins found do not show significant similarity to other proteins deposited in public databases such as SwissProt, and even within Cnidaria, only matched proteins found in other octocorals. Among the group of taxonomically widespread proteins, we found typical components of animal basement membranes (Erickson & Couchman, 2000) with a broad distribution within Cnidaria and generally within Metazoa. Indeed, three of the six most abundant proteins, namely agrin, collagen and osteonidogen, are glycoproteins involved in modulating cell-extracellular matrix interactions (Erickson & Couchman, 2000). Among vertebrates, agrin participates in synaptogenesis (Kröger & Schröder, 2002). Specifically, this protein coordinates the development of neuromuscular junctions, stabilizing and aligning the pre-and postsynaptic apparatuses of neurons and muscle fibres, respectively. It also triggers the differentiation of neuron growth cones into presynaptic terminals capable of calcium-dependent F I G U R E 4 Clusters of semantically similar, significantly enriched Biological Process GO-terms found in the set of overexpressed genes under heat stress. Characteristic GO-terms are annotated to the right of each cluster neurotransmission (Kröger & Schröder, 2002;Ruegg & Bixby, 1998).
Its presence in the SOM of sclerites may imply a similar role in calcification, aligning the sclerocytes around sclerite primordia and triggering membrane differentiation to allow for calcium secretion into the calcifying space. In line with this reasoning, in addition to biomineralization-related GO-terms, colonies exposed to low pH stress modulated transcripts enriched in GO-terms pointing to the active regulation of processes involving the localization of proteins to synapses. We did not observe any apparent effect of the pH treatment on colony behaviour that suggested colonies could not react to mechanical stimuli. Thus, we propose to interpret the localization of proteins to "synapses" in a biomineralization context as the localization of proteins to calcification sites, which, given the presence of presynaptic proteins, appear to be related to some extent to synapses. Although speculative at this point, the analysis of single-cell transcriptomes of octocorals exposed to low-pH seawater can help disentangle this stress' effect on different octocoral cell types (e.g., neurons vs. sclerocytes). In the case of osteonidogen, mammalian osteocytes and osteoclasts overexpress this protein (Bechtel et al., 2012), suggesting a direct role in calcification in that group and a possible involvement in calcification in octocorals.
To respond more rapidly to and survive episodes of environmental stress, resilient stony corals constitutively upregulate components of the coral cell death and immune pathways and genes involved in response to stress, such as heat-shock proteins (Barshis et al., 2013). The concomitant downregulation of genes involved in calcification observed during environmental stress in these organisms (Barshis et al., 2013;Ramos-Silva et al., 2013) suggests that the transcriptional frontloading of the stress response toolkit comes at the expense of the coral calcification machinery and could lead to its collapse. Accordingly, calcification in colonies of Siderastrea siderea exposed to ocean acidification and warming shows a parabolic response, driven mainly by the abrupt drop in calcification rates under more extreme environmental regimes (Castillo et al., 2014). Our results revealed a similar response of P. flava to heat stress, indicating a degree of commonality in the transcriptomic response of stony and soft corals to heat stress. Whether the octocorals' reaction to heat stress is generally similar to that of stony corals remains to be investigated in a larger sample of octocoral species to assess the universality of the pathways modulated during environmental stress. We used a P. flava colony reared in seawater aquaria and thus exposed to abnormally stable conditions over years. Ideally, future investigations on the octocoral response to stress should use specimens collected from natural populations to avoid possible biases caused by aquarium conditions.
Despite the observed similarities in the transcriptional response of the stress toolkit of stony corals and octocorals exposed to heat stress, the calcification machinery of these two groups appear to react differently to adverse environmental conditions. Contrary to stony corals, our results indicate that P. flava can sustain the production of all the molecules necessary for the formation of new sclerites during stress events. In this regard, we detected only a mild effect of F I G U R E 5 Clusters of semantically similar, significantly enriched Biological Process GO-terms found in the set of underexpressed genes under heat stress. Characteristic GO-terms are annotated to the right of each cluster environmental stress on the calcification toolkit of P. flava, with heat stress affecting the expression of a galaxin and a carbonic anhydrase and seawater pH stress modulating the expression of osteonidogen and a prosaposin homologue. Galaxins, carbonic anhydrases and osteonidogen are calcification-related genes in corals and other organisms, and mutations in human prosaposin cause Gaucher's disease, a disorder characterized by skeleton deterioration (Vaccaro et al., 2010). The reported linear decrease in octocoral calcification rates under reduced seawater pH (Gómez et al., 2015) is consistent with the hypothesis of a decoupling between the calcification and stressresponse toolkits of octocorals, as it indicates that the observed decrease in calcification is mainly driven by the environment, not by a transcriptional response of the octocoral calcification machinery to the adverse conditions. Under long-term adverse conditions, decoupling the physiological response to stress from calcification may give octocorals a competitive advantage over other species, such as stony corals, adapted to respond better to episodic stress and lead to the community shifts observed in many reef locations (Inoue et al., 2013a;Ruzicka et al., 2013). Characterizing the transcriptional response to environmental stress of a broader taxonomic sample of octocorals will determine whether the calcification toolkit of these organisms is transcriptionally resilient.
In summary, our results provide new insights into the octocoral stress response to environmental stress. They suggest that the calcification and the stress-response toolkits of soft and stony corals react differently to stress episodes. This difference probably F I G U R E 6 (a) Expression (variancestabilized, z-scored transformed counts) of 70 transcripts modulated (p < .05) in Pinnigorgia flava colonies exposed to control and low pH stress. (b) Change in gene expression of 27 SOM proteins, calcification-related proteins sensu Conci et al. (2019), carbonic anhydrases, galaxins and heat-shock proteins in P. flava colonies exposed to low pH stress. Asterisks (*) indicate transcripts significantly differentially expressed in control vs. treatment samples (a)

(b)
determines the somewhat contrasting response to stress observed in these groups. Anthropogenically induced global climate change will undoubtedly impact future marine communities in unprecedented ways (Hughes et al., 2019). Processes such as acclimatization and adaptation (Palumbi et al., 2014), acting at organismal and population levels, and phenomena affecting the community, such as ecological memory (Hughes et al., 2019), shape the response of coral reefs to these environmental pressures. Our results suggest that, compared to stony corals, octocorals use different gene regulation strategies to face climate change. Thus, understanding the diversity of molecular mechanisms involved in resilience and their regulation in different reef organisms is pivotal to predicting the future of the world's coral reefs.

ACK N OWLED G M ENTS
We thank Gabriele Büttner for support during laboratory work and Dr Peter Naumann for supporting the aquarium facilities of the we thank them. Open access funding enabled and organized by ProjektDEAL.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.

O PE N R E S E A RCH BA D G E S
This article has earned an Open Data Badge for making publicly available the digitally-shareable data necessary to reproduce the reported results. The data is available at https://data.ub.uni-muenchen.