Conserved gene reactivity revealed in Cnidarian innate immunity during zoonotic infection

Cnidarians - the most primitive tissue-forming animals - can get infected with pathogens, defend themselves and heal. Recent sequencing projects on cnidarians have unveiled a rich innate immunity gene repertoire but little is known about their involvement in the host’s response against live pathogens over time. ​ Vibrio parahaemolyticus ​ (Vp) is a zoonotic threat originating from the ocean, which causes pandemics, and economic struggle to society. Here, we advance our understanding of the innate immunity from the response of a basal marine metazoan to infection with Vp. The dynamic of the response was assessed in a time series experiment following the transcriptomic activity of the anemone ​ Aiptasia pallida strain CC7 infected or not with the Vp clinical strain O3:K6 and compared to the gene expression profile under LPS treatment. RNA-sequencing, gene expression and functional analyses detected hundreds of genes responsive to the Vp infection after 1, 3, 6 and 12 hours, including very few shared with the response to LPS. Activated pathways involving TNFRs likely lead to apoptosis which was enzymatically confirmed and suggested as one of the mechanisms in ​ Aiptasia’s ​ defense against bacterial infection. A TLR-independent TLR-like pathway represented by MyD88, TRAF6, NF-κB and AP-1 is actively regulated in response to the infection but not by LPS, suggesting an alternative PAMP-PRR trigger to this pathway. In absence of TLRs, recognition of bacterial pathogens was possibly performed instead, by lineage-specific transmembrane immunoglobulin-containing ficolins unknown from vertebrates and/or diverse NOD-like receptor homologs.


INTRODUCTION
The immune system is an ancient and complex process that works to detect and defend against pathogens, as well as regulate the interactions between microbes and hosts. In vertebrates, the immune system consists of a two-fold mechanism: the innate immunity which provides protection to the host, and an adaptive immune system that mount a specific attack against foreign bodies (e.g. bacteria) and display memory. The current consensus is that invertebrates lack the components of the well characterized vertebrate adaptive immune system. However, with the recent increase in sequencing projects on basal metazoans, the ancestral immune repertoire is clearly appearing more complex than previously suspected. As the sister group of Bilateria, Cnidaria together with Ctenophora form the coelenterates -the most basal animals with true tissue layers and a gastro-vascular cavity also known as the coelenteron. This game-changing internal feature is especially relevant to the evolution of innate immunity, as it marks the apparition of an internal location for interactions between host cells and microorganisms. Therefore, questions regarding the evolution of key bilaterian traits such as pathogen and symbiont recognition, can be beneficially explored using cnidarians. Indeed, among cnidarians, anthozoans including corals and anemones have co-evolved since the Paleozoic era, 570-245 million years ago, with a multitude of unicellular algae, bacteria and viruses, and possess an innate immune system with an extensive list of genes homologous to those of the innate immunity of vertebrates Brennan and Gilmore 2018;Mansfield and Gilmore 2019).
Although studies of traditional model systems such as the mouse, fruit fly and roundworm have provided a wealth of information on the molecular basis of immunity in humans (Kuo et al. 2018), our knowledge is currently improving dramatically thanks to studies conducted on less conventional species from other taxa such as Amphioxus (Zhang et al. 2018), and Hydra (Bosch 2013). Corals, for example, can have up to 10% of their genes matching human's but not those of the fruit fly or roundworm models (Kortschak et al. 2003).
Moreover, corals show intriguing richness in certain immune gene families such as the TNF receptors (TNFR) -a family counting more members in corals than in humans and other vertebrates (Quistad et al. 2014;Quistad and Traylor-Knowles 2016). Factor-(TNF) and Death-domains (DD) containing proteins Shinzato et al. 2011;Hamada et al. 2013) across those species. This diversity in the immune repertoire of cnidarian remains unexplained and warrants further work to determine which portion of the cnidarian immunome is functional against various pathogens, and help understand the origin of certain innate immunity processes. Thus, cnidarian models and their transcriptomes/genomes provide the material for testing hypotheses about the nature of ancestral eumetazoan (i.e. Cnidaria and Bilateria) pathways and interactions of innate immune defenses relevant to human health. Of particular interest are the mechanisms of pathogen recognition, containment, and protection among immune cells, as well as the biological processes to protect the rest of the cells against invading pathogens, shared in both cnidarians and bilaterians.
Aiptasia pallida 's transcriptome/genome (Sunagawa et al. 2009;Baumgarten et al. 2015, respectively) show an active repertoire of immune-related genes involved in the establishment of host-algae symbiosis , and the disruption of this symbiosis also known as bleaching (Detournay et al . 2012), which warrant the question whether the same is true against pathogen invasion. To challenge the innate immune system of Aiptasia , the Gram-negative halophilic bacteria Vibrio parahaemolyticus was chosen for two reasons.
Firstly, marine zoonotic pathogens such as Vp strain O3:K6 can have disastrous consequences for human populations. Indeed, Vp is the leading cause of marine-borne illnesses worldwide (Livny et al. 2014) causing gastroenteritis via the consumption of contaminated seafood (Czachor 1992;Lipp and Rose 1997). And secondly, environmental strains naturally occur in the mucus of coral species (Chimetto et al. 2008;Nithyanand and Pandian 2009), which make Vp a member of the rich bacterial community associated with cnidarians. Using Vp to infect Aiptasia makes an ideal -marine zoonotic pathogen versus host -model system that holds great potential for studying ancestral innate immunity, and especially the evolution of those conserved pathways shared with vertebrates including humans.
Brennan and Gilmore (2018) have recently reviewed the information accumulated on the most well-studied pathogen-associated recognition receptors (PRRs) -the Toll-like receptors (TLRs) -in basal phyla including Porifera, Cnidaria, Annelida, Mollusca, and Nematoda. Based on TLR sequence analyses in these phyla, the authors hypothesized that the prototypical TLR originated within the cnidarians. Nevertheless, not all cnidarians possess prototypical TLRs. Studies focusing on the innate immunity of cnidarians have identified various prototypical TLRs as well as TLR-like genes. Surprisingly, the numbers and types of TLR-like genes and TIR-containing proteins vary greatly between Acropora corals, Nematostella vectensis (Nv), Hydra vulgaris (Hv) and Aiptasia pallida (Ap). For example, corals can have one or several multiple cysteine cluster TLRs (mccTLRs), one or no single ccTLR (sccTLR) and many TIR-only proteins (Shinzato et al. 2011;Poole and Weis 2014;Palmer and Traylor-Knowles 2012;Miller et al. 2007). In contrast, Nv only has a single mccTLR and no sccTLR or TIR-only proteins (Brennan ad Gilmore 2018), while Hv possesses separate transmembrane LRR-and TIR-containing proteins (Bosch et al. 2009).
When comparing the conserved components of the TLR to NF-κB pathway, Brennan and Gilmore (2018) show that the pathway's structure in Ap diverges slightly from that of Nv, Hv, or the three coral species under consideration, as no TLR, NEMO or IκB homologs could be identified. Moreover, recent studies have characterized and provided functional evidences that TLR-like and the key mediator MyD88 proteins can interact to activate the transcription factor NF-κB and certain aspects of the innate immune response in the coral Orbicella faveolata (Williams et al. 2018), Nc (Brennan et al. 2017), and Hv (Bosch et al. 2009). Most recently, a study brought further support for the role of NF-κB in the symbiosis of Ap with Symbiodinium algae and its survival to bacterial infection .
Here, the symbiotic Ap strain CC7 was used to identify the molecular innate immune response of the host to bacterial infection, and compare the responsive genes and represented pathways to the functional data derived from other cnidarian models. This study focuses on the gene expression profiles of immune-related genes in response to infection after 1, 3, 6, and 12 hours with the pathogen, and exposure to the endotoxins lipopolysaccharides (LPS) for 3, 6 and 12 hours. We confirm that no prototypical TLR, nor LRR-containing TLR-like genes are regulated in the innate immune response of Ap, however we identify a TIR-and Immunoglobulin (Ig) -containing transcript responsive to the infection, which we hypothesize could be the TIR-containing protein interacting with the Ap_MyD88 and inducing signaling down to the activation of Ap_NF-κB. We also discuss the potential roles of homologs belonging to the NOD-like receptor, lectin-complement, and tumor necrosis factor receptor (TNFR) families as well as conserved genes of the extrinsic and intrinsic apoptotic pathways.
To our knowledge, this study is pioneering in following the dynamic transcriptomic activity of a basal marine animal under infection with a zoonotic bacterial pathogen in a time series experiment. This work improves our understanding of innate immunity through the measurement of gene activity in response to infection in one of the most primitive tissue-forming animals and highlights several fascinating gene targets for future functional studies.

Morphological response to infection
In order to study Ap's innate immune response to Vp infection, we first monitored the response of anemones at the morphological level over 48 hours. Interestingly, animals removed from the treatment medium after 12 or 24 hours, and subsequently left in filtered seawater (FSW) for another 36 or 24 hours, respectively, recovered from such temporary assault suggesting an efficient early innate immune response (Supplementary Figure 1).
Morphologically, Ap's response to infection consists of condensed tentacles, increased mucus production, shriveling and cell detachment ( Figure 1). Under the microscope, Vp in contact with the host can be seen swarming on the surface of the anemones including the foot, column and tentacles, which show signs of tissue damage and invasion at 12 and 24 hours post-inoculation. Treatments consisting of 24 hours in the infectious medium started to cause mortality among Ap specimen. To investigate the early innate immune response that allows Ap to defend itself against Vp's attack, we focused our gene expression analyses on the first 12 hours of infection, when all anemone specimens remain alive.

General transcriptomic response
The general trends primarily revealed from the gene expression analyses between control and infected samples over the four time points show that: 1) more genes are up-regulated than down-regulated throughout the host response to infection, 2) the number of regulated genes increases with time, and 3) the regulatory activity in response to the infection over 12 hours ranges from less than 1% to 3% of the whole transcriptome (Table 1).  Figure 2). These analyses show that: 1) there are more genes unique to each time point than genes detected twice or several times, 2) there are very few genes continuously regulated during 12 hours of infection, and 3) the further along the host response, the more genes are shared between time points. These trends suggest a progression in the host response even though anemones were kept in the same infectious medium over 12 hours. As Vp cells settle down and anemones concentrate them from the medium, host-pathogen physical interaction likely diverges with time, while signal transduction from pathogen recognition to effector activation progresses at the same time.     Table 2 for statistical values). As the immune response progresses, the 'regulation of apoptotic process' becomes evident after 3 hours, and will be so through the rest of the experiment. Among the genes indicative of apoptotic activity, some belong to the 'regulation of: NF-κB transcription factor, interferon-beta production, and MDA-5 signaling pathway' GO_BP -as well as the GO_MF: 'caspase binding' highlighted at 3, 6, and 12 hours post-infection. The role of the TNF pathway in the immune response of Aiptasia gathers even more support at 3 and 6 hours, with the enrichment for 'TNF-mediated signaling'. In addition, the 'binding to AP-1 adaptor complex' is highlighted at these two time points as well. At 6 hours, 'activity of endopeptidase' and 'endopeptidase inhibitor' was detected for the first time and was supported again at 12 hours. Finally, after 12 hours of infection, while activity within TNF receptor, apoptosis, and proteolysis pathways is still strong, additional peptidase activities join in the host response. In summary, these results reveal a noticeable chronology in the immune activity of Aiptasia against the pathogen, starting with the involvement of TNFRs, then the activation of apoptosis followed by proteolysis. The genes supporting these functional categories, as well as DECs without GO annotations but containing immunity-related conserved domains such as DD, Ig, lectin (Lect), leucine rich-repeat (LRR), NACHT, scavenger receptor (SR), TIR, and TNF represent an interesting source of novel innate immunity candidates, which we partly discussed below.
Apoptosis is a form of cell death regulated by caspase proteases. The caspase-3 (CASP3) like enzymatic activity was measured in Ap during infection using human caspase-3 activity detection kit applied to protein samples. The Ap CASP3-like enzymatic activity was found on average 2.9 fold higher in infected Ap at 12 hours, but showed similar levels to controls at 3 and 6 hours ( Figure 4A). In addition, the CASP3 gene expression measured via sequencing detected up-regulation at 3, 6 and 12 hours with a large peak at 6 hours (Supplementary Table 3). This CASP3 gene activity was confirmed by qRT-PCR with specific primers ( Figure 4B). These results show that the effector CASP3 gene is positively expressed all along the infection and that CASP3-like enzymatic activity is detectable after 12 hours at the organismal level in response to infection.

Genes pointing to active innate immunity pathways and apoptosis
The results shown in Table 2 follow the organization chosen by Zhang and colleagues (2018) in a study extremely relevant to this one, analyzing the transcriptomic response of the amphioxus -Branchiostoma belcheri -to Vp. Here, a majority of the DECs reported in the tables 2 and 3 also represent the significantly enriched GO classes mentioned above. The tables of DECs shown below are subsets of exhaustive lists given in Supplementary Tables 3   and 4. For clarity purpose, Table 2 and 3 show DECs with fold changes > 5 or < -5, and homolog peptide sequence similarity values > 70%. Table 2. Innate immunity-associated DECs -Immune response related DECs are organized as in Zhang et al. (2018) into the following four broad categories: complement and coagulation cascades, PRRs, cytokines and regulators, and adaptors and signal transducers.
Vertically, DECs appear: 1) in chronological order of their time of detection, 2) from highly to lowly regulated, and 3) per gene family. While horizontally, the time of detection, the fold change in expression level and the false discovery rate (FDR) corrected p-value from the Robust Exact Test (edgeR package) represent gene expression analysis results. In addition, the contig sequence length in base pairs (bp), the score and similarity percentage of the alignment (e-value and similarity %, respectively) represent the level of confidence in the best gene identification (i.e. gene name) derived from the BLASTx to nr (from NCBI database) alignment analysis. The DECs highlighted here were filtered for a fold change greater or equal to 5 and lower or equal to -5, as well as a percentage of similarity greater or equal to 70%. The entire list of DECs is provided in Supplemental Table 3.
Among the most regulated and well supported (i.e. showing high similarity) DECs (  Table 3). Another DEC annotated IRF2-like isoform X1 is also the most up-regulated sequence at 3 hours post-infection (FC = 220.6, FDR = 3.62E -2 ), and isoforms were also up-regulated at 3, 6 and 12 hours (Supplementary Table 3). At the same time point, the TNF receptor-associated factor 5 (TRAF5) -like DEC is down-regulated 230.6 times (Robust Exact Test, FDR = 1.79E -2 ), but isoforms are up-regulated at 3, 6 and 12 hours (Supplementary Table 3 Table 3). Based on comparisons between the Vp and LPS results (Supplementary Table 3 versus Table 1,  Similarly, to the innate immunity genes illustrated in Table 2, highly regulated genes associated with apoptosis are shown in Table 3. DECs likely involved in the apoptotic process were already detected after 1 hour post-infection and further progression of the apoptotic signal was observed at each following time point. The DECs annotated 'acidic leucine-rich nuclear phosphoprotein 32 family member A' are the most regulated sequences at 1 hour post-infection (  Table 4). Finally, after 12 hours, the DEC -calcium/calmodulin-dependent protein kinase type 1D -is the most up-regulated with a fold change of 157.2 in (FRD = 4.56E -2 ), while the DEC described as T-box transcription factor (TBX1) is down-regulated 11.9 times (FDR = 1.05E -3 ). Briefly, the genes in Table 3 suggest that both the extrinsic and intrinsic apoptotic pathways are playing a role in the anemone's response. The former likely represented by caspase 8 (CASP8), Ced3 (CASP9) and CASP3, and the latter by mitochondria-associated genes such as B-cell lymphoma 2 (Bcl-2), Bcl-2-associated X protein (Bax), and Bcl-2 antagonist/killer (Bak), among others. Table 3. Apoptosis-associated DECs -Apoptotic response related DECs. The DECs represented here were filtered and organized as in Table 2. The entire list of DECs is provided in Supplemental Table 4.

DISCUSSION
Overall, studies of innate immunity in cnidarians have focused on its evolutionary history Miller et al. 2007), its role in bacteria sensing and aging (Franzenburg et al. 2012;Bosch 2013), self-/non-self-recognition ( cnidarian classes, families and species (Mansfield and Gilmore 2018). This provides a set of models to study the evolution of innate immunity as it is known in vertebrates. In particular, studies focusing on the characterization of the TLR to NF-κB like pathway and protein activity of its components in Hydra (Bosch et al. 2009;Franzenburg et al. 2012), Nematostella (Wolenski et al. 2011;Brennan et al. 2017) and Orbicella faveolata (Williams et al. 2018), reveal some functional conservation in response to immune challenges (discussed below). With this in mind, this study aimed to contribute information about the active molecular component of innate immunity towards bacterial pathogens in Aiptasia, in order to identify genes and pathways relevant to vertebrate and human health. To do so, the transcriptomic activity of the cnidarian model -Aiptasia pallida strain CC7 -was analyzed while challenged with the human pathogen Vp strain O3:K6 over 12 hours.
Anemones are suspension-feeders and therefore concentrate particles from the seawater within their coelenteron over time, via the mechanical action of cilia covering their surfaces. This feeding mechanism is relevant to the pathway from which humans get affected by Vp. Indeed, a majority of the vector species of vibriosis are filter-feeders and constitute "seafood" which can accumulate plankton-associated pathogens if present in their environment (Baffone et al. 2006). When in contact with the high concentration of Vp used here, anemones can die in 24 hours, however if relieved from bathing too long in the highly infectious medium, anemones are capable of recovery, denoting an effective immune system. Similarly, to the cross-pathogenicity observed here, another study using the gram-negative human pathogen -Serratia marcescens -shows that Ap exposed to 10 8 cfu/mL has a survival rate of 70% after 24 hours (Krediet et al. 2014), which supports the present observations using Vp at the same concentration. S. marcescens does not originate from the marine environment unlike Vp , but a couple of strains (PLD100 and PDR60) were isolated from lesions on the Elkhorn coral Acropora palmata , and are therefore also relevant to cnidarian etiology. Here, we show that the clinical Vp strain O3:K6 responsible for gastroenteritis in humans remains pathogenic in saltwater at 27 degrees celsius and can kill Aiptasia anemones over 24 hours.

Potential cytoplasmic PRRs
One of the essential aspects of innate immunity this study sheds light on, is the identification of active genes with PAMPs recognition potential in Aiptasia. Within the cells, intracellular PRRs are critical to respond to microbial components in the cytosol and detect pathogens. The NLRs are cytoplasmic proteins, which recognize bacterial peptidoglycans and trigger proinflammatory and antimicrobial immune response. Among the NLR family, the NLR family CARD domain containing protein 4 (NLRC4) is the best-known mammalian member capable of triggering the inflammasome upon detection of microbial ligands (Duncan and Canna 2018). Moreover, inherited NLRC4 mutations are responsible for autoinflammatory diseases in humans (Kitamura et al. 2014 Figure 5). This is in contrast with human's cytoplasmic NLRs lacking TMDs. However, these infection-induced NLRC contigs in Aiptasia may encode for intracellular proteins playing a role in the recognition of Vp molecules within the cells via LRRs. Additional genes were identified as potential cytoplasmic PRRs, interferon-induced helicase C domain-containing protein 1 (IFIH1) and RIG-I-like receptor 3 (RLR-3), which contains the RIG domain of the retinoic acid-inducible gene-I-like receptors (RLRs), play a role in the intracellular recognition of viruses. In total eight IFIH1 sequences and three RLR-3 sequences were detected as up-regulated at 3, 6 and 12 hours (Supplementary Table 1). As the pathogen Vp acquired several new genomic islands through horizontal transfer, which contain phage-encoded virulence factors (Ceccarelli et al. 2013), it is possible that the Ap genes mentioned above are induced by viral-like genetic material brought into the host cells by invading Vp.

Lineage-specific hypothetical PRRs and their putative pathways
In search for potential genes involved in Symbiodinium recognition, Baumgarten and colleagues (2015) Table 1). These results support the potential role of TMD/Ig-containing CniFLs in the recognition of pathogens and strongly suggest that both the lectin and alternative pathways play an active role in the defense of Aiptasia against Vp. However, it remains to be demonstrated whether the CniFLs bind to certain PAMPs and can trigger the lectin-complement pathway in Aiptasia .
The scavenger receptor lectin-like family is rich and diverse in Anthozoans (Neubauer et al. 2016) and plays an important role in host-symbiont interactions during onset of symbiosis (Davy et al. 2012). Moreover, a mannose-binding lectin is capable of binding pathogens in the coral Acropora millepora (Kvennefors et al. 2008). In Aiptasia , several lectins were detected during the response to Vp infection. For example, the L-rhamnose-binding lectins (RBLs) interacts with various types of bacteria in fish and invertebrates to induce proinflammatory cytokines such as IL-1β, TNF-α and IL-8, and enhanced macrophage's phagocytosis (Watanabe et al. 2009). Here, Ap_RBL contigs are up-regulated at 3, 6 and 12 hours post-infection (Table 2 and Supplementary Table 1 These results highlight several interesting lectins candidates for the recognition, potential activation of phagocytosis, and proinflammatory signaling upon contact with a pathogen in Aiptasia . HyLRR-2 reciprocally reduced the expression of the other gene, and the production of antimicrobial peptides (Bosch et al. 2009). In the coral O. faveolata too, the mccTLR protein can interact with human MyD88 (Williams et al. 2018), while MyD88-deficient Hydra were more susceptible to infection with the human pathogen Pseudomonas aeruginosa (Franzenburg et al. 2012). Taken together, these reports support an ancestral role of TLRs and TLR-like to NF-κB signaling in response to bacterial stimuli in cnidarians.

Active components of TLR-like pathway
Here, with the exception of two DECs with low similarity to TLR-6 and -13, neither homologs of vertebrate TLRs, nor orthologs of the Hydra's HyTRR or HyLRR genes were  Table   1). The TIR-domain containing MyD88 homolog in Aiptasia (Ap_MyD88), is 254 times more regulated in infected anemones at 3 hours, and remains up-regulated by a fold change of 2 after 12 hours. Ap_TRAF6, an important signal mediator likely acting downstream of Ap_MyD88 is up-regulated 254.7 times at 6 hours and continues at a lesser degree after 12 hours. This suggests that in Ap, the recruitment of MyD88 and downstream signaling to activation of NF-κB is induced by a PAMP other than LPS and must be triggered by a PRR other than a prototypical TLR or HyLRR-2 homolog. When searching further for genes potentially interacting with MyD88 via their TIR or Death domains, a DEC up-regulated at 6 and 12 hours annotated substance-P receptor-like (SPR) containing a TIR-2 and Ig domains actually matches a TLR-2 type-1-like gene (GenBank: KXJ09394) in the Ap genome ( Figure 6). Here, we hypothesize that this Ap TLR-2 type-1-like gene may be the cytosolic TIR-containing protein interacting with the Ap_MyD88. We suggest that similarly to the results in Hydra , Ap_MyD88 could be the central mediator of the pathogen-recognition signal to transcription factors or regulators such as NF-κB, AP-1, and IRFs, which are all up-regulated during the infection and likely control the production of cytokines.

Involvement of TNFR pathways
The next important aspect revealed by this study is the activation of cytokine-dependent signaling pathways taking part in the innate immune response to Vp.
Two of these pathways could lead to apoptosis and an inflammatory response. The conservation of the TNF-induced apoptotic response was demonstrated in the coral Acropora digitifera (Quistad et al. 2014). Here, the regulation of the 'TNF-mediated signaling pathway' was enriched at 3, 6 and 12 h post infection. When focusing on the genes representing this GO_BP, key players of two pathways stand out as highly regulated and contain some of the conserved domains in support of functional conservation in Aiptasia 's innate immunity ( Figure 7).  Table 4). Both pathways could engage the apoptotic executor -CASP3, which is up-regulated all along the infection, with a peak of 1 250 in fold change at 6 hours and enzymatically active at 12 hours.

Apoptosis as defense mechanism
An essential immune defense mechanism in eukaryotes is to sacrifice infected cells in order to protect healthy cells. The major cell death pathways, including apoptotic cell death, is a crucial barrier against microbial infection (Bergsbaken et al. 2009;Lamkanfi and Dixit 2010;Zitvogel et al. 2010). For instance, in response to bacterial infection, apoptosis or programmed cell death is used in the host innate immune response to: 1) eliminate pathogens at the early stage of infection without emitting alarm signals, and 2) induce dendritic cells (DCs) to engulf apoptotic bodies containing infected microbes (Elliott and Ravichandran 2010). In Aiptasia's response to infection, a functional enrichment for 'regulation of apoptotic process' could be detected at 3 hours. Among the DECs supporting the enrichment, many key players of apoptosis as well as genes with potential connections to the apoptotic pathway were identified at each time point, while an increased CASP3-like enzymatic activity was observed at 12 hours. More specifically, 216 contigs annotated with a potential role in apoptosis took part in Ap's response to infection. Besides the hallmarks of apoptosis mentioned above, other genes connected to apoptosis were highly regulated. For example, isoforms of the acidic leucine-rich nuclear phosphoprotein 32 (ANP32) gene were 635 and 265 times up-and down-regulated, respectively at 1 hour, and up-regulated 174 times at 3 hours. Several studies showed that ANP32 proteins allowed apoptosome activation at physiological levels of dATP and also promoted CASP3 activation directly (Reilly et al. 2014). The transcription factor v-ets erythroblastosis virus E26 oncogene homolog 1 (Ets-1) is involved in the regulation of apoptosis (Teruyama et al. 2001), by regulating genes encoding Bax, Bcl-2, Caspase-1 and Fas ligand (Nagarajan et al. 2009). Here, the Aiptasia 's Ets-1 homolog is up-regulated at 3, 6, and 12 hours post infection including a peak expression of 249 fold change at 3 hours. It is noteworthy that the FasLG, Bax and Bcl-2 homologs are among the DECs detected in this study. These results strongly suggest that apoptosis plays an important role in the innate immune response of Aiptasia against bacterial pathogens. As Aiptasia is capable of recovering from a punctual exposure to Vp , it is probable that apoptosis plays a part in the removal of invading bacteria and the survival of the host. Further work is envisaged to study if and how Aiptasia is capable of containing bacterial invasion via programmed cell death.

Novel genes responsive to infection
Several of the genes discussed above as potential  Sequencing data processing Raw sequences in fastq files were processed to remove Illumina adapters, low quality bases and short reads using the Trimmomatic-0.36 tool (Bolger et al. 2014). New reference transcriptomes were assembled de novo using all of the Aptasia pallida libraries produced here and either guided by the Ap genome produced by the Pringle laboratory (version 1.1) or genome-free using Trinity v2.6.5 release (Grabherr et al. 2011). Contiguous sequences or contigs were used to retrieve gene descriptions using blatsx against the non-redundant NCBI database with the following filter: E-value < 10 -3 , and only Blast description annotation excluding the terms 'unknown', 'hypothetical', or 'uncharacterized'. In addition, coding regions were predicted using TransDecoder and the predicted peptide sequences were used to retrieve further gene annotations from several different databases using blastp against Salzberg 2012) packages. The pairwise differential expression analysis comparing the transcript abundance in control anemones to anemones in contact of the pathogen was performed using edgeR (Robinson et al. 2010) on the Blast2GO platform (Götz et al. 2008).
The exact test based on the quantile-adjusted conditional maximum likelihood method for single-factor experimental design was applied here to detect differentially expressed contigs  Table 2. The complete list of DECs containing sequence IDs, hits information, folds changes and statistical results is accessible in the Supplementary   Table 3.

Enrichment analyses
To detect a significant enrichment for the immunity associated GO terms within the lists of DECs, two tests were applied: one quite permissive -the Fisher's exact test and the other more stringent -the gene set enrichment analysis method. Both tests were run using the Blast2GO platform (Götz et al. 2008) and the whole transcriptome annotated with GO terms as the reference gene set. For both tests, only contigs with false discovery rate corrected p-value lower than 0.05 were considered. The enrichment analysis using the Fisher's exact test was performed on the up-and down-regulated gensets produced via the gene expression analyses for the 1h, 3h, 6h and 12h time points. Results were then reduced to the most specific GO categories applying a lower than 0.05 filter on FDR values to produce Figure 3. Some limitations of the functional analysis (FET) need to be kept in mind and concerns the level of fragmentation of our transcriptome, meaning that several contigs can belong to the same transcript. This can introduce a bias when running enrichment analysis which compares the proportion of genes representing a gene ontology category in a test geneset, with the proportion of genes for that same GO category in a reference geneset. In a case where, a test geneset made of a large number of DECs representing a certain number of transcripts is compared to a different geneset representing the same number of transcripts but by less DECs.
The Fisher's exact test is more likely to find GO categories significantly represented in the former geneset than the latter. Therefore the enrichment is somewhat dependent on the level of fragmentation of the genes represented in the geneset. The enrichment for the GO categories discussed here were manually validated and are truly represented by many relevant genes provided in Supplementary Table 3 and 4.

Caspase-3-like activity assay
Caspase-mediated apoptosis in control and infected anemones was assessed by increased activity of caspase-3-like and other DEVD-specific proteases using an enzymatic assay (EnzCheck Caspase-3 assay kit, Molecular Probes: Cat.# E-13183) on freshly isolated protein extracts. Frozen anemone's tissues were disrupted and cells lysed using the cell lysis buffer of the kit (EnzChek Caspase-3 kit #1, Molecular Probes) and the precellys as for RNA extractions above. The fluorescence of the benzyloxycarbonyl group (Z-DEVD-AMC) upon proteolytic cleavage was continuously measured in triplicate samples using the microplate reader Synergy H1M (BioTek) at 441 nm for 1 hour. Confirmation of the caspase-3-like activity was obtained by using the Ac-DEVD-CHO inhibitor (negative control), successfully suppressing the signal detected in this experiment almost entirely. The positive control consisted of inducing apoptosis in anemones exposed to 1 μM staurosporine (Cat.# 9953S, Cell Signaling) and triggering a caspase-3-like activity up to 9.2 times higher than in controls.
The fluorescence signal was normalized to total protein quantity measured with the DC protein assay (Cat.# 5000116, Bio-Rad).
Quantitative real-time PCR RNAs were extracted as explained for the sequencing analyses. Complementary DNAs were then synthesized from 100ng of purified total RNAs using oligodT primers following the RevertAid first strand cDNA protocol (cat. K1622; ThermoScientific). Specific primers designed from the DEC sequences identified as caspase-3 in Ap's transcriptome were used in quantitative real-time PCR (qRT-PCR) reactions run on the StepOnePlus machine (Applied Biosystems). The GoTaq qPCR master mix (cat. A6002; Promega) was used in reactions containing: 1uL of each primer at 10pmol/uL, 10uL of SYBR, 0.1uL of CXR, 5.9uL of DEPC treated water, and 2uL of 10X diluted cDNA. Normalized expression levels for the gene of interest (GOI) were calculated using normalization factors (geometric mean) based on the stable expression levels across time points and treatments of three reference genes: 60S ribosomal protein L11, F-actin and adenosylhomocysteinase 2.

Ethics approval and consent to participate
Not applicable

Availability of data and materials
The datasets generated during and/or analysed during the current study WILL BE available in the NCBI BIOPROJECT repository

Consent for publication
Not applicable