Transcription Profiling of Cultured Acropora digitifera Adult Cells Reveals the Existence of Ancestral Genome Regulatory Modules Underlying Pluripotency and Cell Differentiation in Cnidaria

Abstract Due to their pluripotent nature and unlimited cell renewal, stem cells have been proposed as an ideal material for establishing long-term cnidarian cell cultures. However, the lack of unifying principles associated with “stemness” across the phylum complicates stem cells’ identification and isolation. Here, we for the first time report gene expression profiles for cultured coral cells, focusing on regulatory gene networks underlying pluripotency and differentiation. Cultures were initiated from Acropora digitifera tip fragments, the fastest growing tissue in Acropora. Overall, in vitro transcription resembled early larvae, overexpressing orthologs of premetazoan and Hydra stem cell markers, and transcripts with roles in cell division, migration, and differentiation. Our results suggest the presence of pluripotent cell types in cultures and indicate the existence of ancestral genome regulatory modules underlying pluripotency and cell differentiation in cnidaria. Cultured cells appear to be synthesizing protein, differentiating, and proliferating.


Introduction
Cell-cell interactions are fundamental for body plan establishment and function as they integrate cell type-specific genome regulation during animal development (Peter and Davidson 2011). Coordination and assembly of cell-specific transcription profiles generate interconnected transcriptional networks that in cnidaria generate organismal complexity by deploying morphogenetic borders along an oral-aboral axis (Hayward et al. 2015). In cnidaria, tissue morphogenesis and homeostasis mechanisms are very diverse. The phylum

Significance
The inability to establish permanent cnidarian cell lines has directed attention to tissues with a high abundance of stem cells as an ideal material for establishing long-term cultures. Despite this, the lack of unifying principles associated with "stemness" across cnidaria complicates stem cells' identification and isolation. Here, we report gene expression profiles for cultured coral cells. Our results revealed pluripotent cell types in cultures, identifying coral orthologs of stem cell markers that could be used for further isolation and characterization. exhibits complex life cycles and diverse developmental mechanisms, spanning planula (P), polyp, and medusa morphologies (Daly et al. 2007). Anthozoans, in particular, display complex polyp morphologies and less regenerative potential, requiring signaling from an organizer region to reconstruct their tissues (Hayward et al. 2015).
At the cellular level, although hydrozoans maintain body structure via the integration of three distinct stem cell lineages established during gastrulation-two epithelial and one interstitial (Hemmrich et al. 2012)-nonhydrozoans cnidarians present two epithelial stem cell lineages lacking the interstitial cell type (i-cells) (review in Gold and Jacobs 2013). Despite this, it is accepted that transdifferentiation of epithelial cells and dedifferentiation of "committed" cell types predates the evolution of cnidarian stem cell systems (Gold and Jacobs 2013). In animals, precursor cells in vivo maintain "stemness" and control differentiation via complex cell-cell interactions within a tissue grade microenvironment known as "stem-cell niche" (Fuchs et al. 2004;Gattazzo et al. 2014). Within the niche, differential deployment of core networks regulates cellular identity by promoting the expression of "pluripotency" genes while repressing developmental signaling pathways (Orkin and Hochedlinger 2011).
In metazoa, stem cells have been reported as early as porifera, with archeocytes and choanocytes identified as the oldest animal stem cell system (Funayama 2010). In cnidaria, undifferentiated stem cells have been reported in a variety of adult (Mydlarz et al. 2008) and larval (Martin and Chia 1982) tissues, with transcriptional profiles of precursor cell types characterized for Hydra (Siebert et al. 2019) and Nematostella (Seb e-Pedr os et al. 2018). Due to the high abundance of stem cells in adult stages, adult tissues have been suggested as an ideal material for the establishment of longterm cnidarian cultures (Rinkevich 2011). However, the lack of unifying principles associated with stemness across the phylum complicates identification and isolation of cnidarian stem cells for culture.
In cnidaria, attempts to establish permanent cnidarian cell lines have failed due to decreasing in vitro viability and proliferation (review in Rinkevich 2011). Despite this, short-term primary cultures have been used to understand the cellular mechanisms underlying fundamental cnidarian processes, such as symbiosis (Barnay-Verdier et al. 2013), calcification (Domart-Coulon et al. 2001Mass et al. 2012), thermal stress (Nesa and Hidaka 2009), and regeneration (Schmid and Alder 1984). These studies reported inconsistent results regarding survival, proliferation, and viability, reflecting the lack of standardized protocols (reviewed by Rinkevich 2005Rinkevich , 2011. Nonetheless, as primary cell cultures are taken directly from in vivo tissues, they represent a powerful tool to study cellular and physiological processes unattainable using whole organisms. This paper, reports for the first time gene expression profiles for cultured coral cells, focusing on regulatory gene networks underlying pluripotency and differentiation. We initiated primary cell cultures from Acropora digitifera tip fragments, the fastest growing tissue in Acropora corals (Lirman et al. 2014). A. digitifera has become an emerging model to study coral responses to environmental change (Shinzato et al. 2011) and due to its basal phylogenetic position in cnidaria (Bridge et al. 1992), the species is also an ideal model to study the evolution of animal developmental mechanisms.
Overall, after 4 weeks of culture, we conducted quantitative RNA-seq analysis and compared the in vitro transcriptome with a previously published A. digitifera developmental time series (Reyes-Bermudez et al. 2016). In vitro transcription closely resembled gene expression used in early larvae during the establishment of larval cellular phenotypes. Likewise, we identified upregulation in vitro of orthologs of premetazoan and Hydra stem cell markers (HM) and transcripts with roles in DNA replication, cell division, migration, and differentiation. Our results suggest the existence of 1) pluripotent cell types in cultures and 2) ancestral genome regulatory modules underlying pluripotency and cell differentiation in cnidaria.

Results and Discussion
Cultures Consist of Multicellular Aggregates Actively Dividing and Differentiating After 4 weeks, cultures consisted of cellular aggregates formed by cells displaying a previously reported unique and nonspecific small round morphology (Reyes-Bermudez and Miller 2009). Cells did not attach to the substrate, and signs of CaCO 3 precipitation were not observed. Contrasting with previous results (Lecointe et al. 2013), cultures did not show evidence of decreased proliferation. The fact that one round of subculturing was necessary for 2 weeks after initiation, suggests that cells were actively dividing. Although zooxanthellae were present abundantly at initiation, after 4 weeks, chlorophyll fluorescence was absent. Our results support the idea that the capacity to maintain or reconstruct cell signaling between epitheliums in vitro is critical for the establishment of "long-term" successful primary cultures (reviewed by Rinkevich 2005Rinkevich , 2011. Moreover, upregulation in vitro of genes with roles in protein synthesis, proliferation, and differentiation indicate that, at least after 4 weeks, a subset of cells was proliferating and differentiating in cultures (table 1). The fact that cultures were initiated from branch tips, the fastest growing tissue in Acropora corals (Lirman et al. 2014), suggest that undifferentiated cell populations in founder tissues might be responsible for in vitro enrichment of corals cells that is a crucial factor for culture viability (reviewed by Rinkevich 2005Rinkevich , 2011. Reduced proliferation in primary cnidarian cultures has been linked to a decrease in the proportion of animal cells as result of culture contamination (Frank et al. 1994).  . 1B). Moreover, enrichment in the subset of differentially expressed genes (DEGs) upregulated in vitro with molecules involved in diverse morphogenetic and differentiation processes (table 1) indicates uncoordinated overexpression of genome regulatory programs following the loss of morphogenetic borders (Beloussov 2015). Results revealed a transcriptionally heterogenous C cell population that differed transcriptionally from the in vivo system. Significant transcriptional changes that reflect the emergence of more active and proliferative subgroups have been reported in cell cultures over time (Januszyk et al. 2015).

Cells Expressed Orthologs of Premetazoan and HM
Consistent with 500 Myr of independent cnidarian evolution (Steele et al. 2011), only a small fraction of DEGs were identified as orthologs of HM. We observed over expression of a higher number of HM in Blastula (PC) ( fig. 1D), which is consistent with the predominant pluripotent cellular phenotypes present at the stage. Interestingly, Acropora HM orthologs upregulated in vitro were enriched with markers overexpressed by Hydra's i-cells (nanos) ( fig. 2A), which, in a strict sense, is the only true characterized cnidarian stem cell population (Frank et al. 2009). These results may not indicate the presence in coral tissues of stem cell populations homologous to Hydra's i-cells but most likely indicate the utilization of conserved molecules underlying pluripotency and cell differentiation in Acropora.
Similarly, we observed enrichment in cultures of orthologs of endodermal markers such as Brachyury (Yasuoka et al. 2016) and Hedgehog (Matus et al. 2008) (supplementary material S1-S4, Supplementary Material online). Whether this reflects endodermal enrichment in vitro or utilization of conserved transcriptional networks by different lineages, is not clear. For example, although i-cells are thought to be a "recent" cnidarian innovation, their transcriptome is phylogenetically older than those of Hydra's epithelial lineages (Hemmrich et al. 2012), suggesting recruitment of ancestral regulatory networks during Hydra's i-cell evolution. Our results support this idea and suggest that regulatory networks associated with maintenance of pluripotency and differentiation are not fixed entities linked to specific cell types, but dynamic modules recruited and modified multiple times by natural selection during cnidarian diversification.
Likewise, identification of premetazoan and metazoan stem cell markers in the subset of DEGs exclusively upregulated in vitro (table 2) is consistent with the idea that animal stem cell systems were built upon ancestral regulatory gene networks present in the last common metazoan ancestor (Ali e et al. 2015). Upregulation in vitro of Acropora orthologs with roles in cell cycle, replication, chromosome maintenance, stress response, DNA repair, as well as diverse transcripts coding RNA-binding proteins, such as a Musashi-1 ortholog (table 2), imply that components of regulatory gene networks associated to stemness are being expressed in cultures. Transcripts (18,264) were assigned to 38 different gene modules that ranged from 38 to 3,413 transcripts and grouped in two main coexpression clusters (C1 and C2). Eigengenes were calculated for each module and although we were able to identify discrete gene expression patterns, in most cases. significant module-trait correlations were observed in a stage-specific fashion. *P value 0.05, **P value 0.01, ***P value 0.01. Blastula, PC; gastrula, G; early larvae, S; planula, P; adult, A; cultured cells, C.    2B). Coexpression units usually reflect common functionality and regulation (review in Peter and Davidson 2011). Interestingly, most in vitro upregulated DEG's were coexpressed in PC, S, and A but were significantly downregulated in G and P stages ( fig. 2B). Differential usage of enhancers between G and P stages have been reported for Nematostella (Schwaiger et al. 2014), suggesting the existence of distinct G-/P-specific genome regulatory programs in cnidaria. More research is necessary to test this idea as transcriptional networks underlying early morphogenetic transitions in metazoans are variable and, in some cases, taxa-specific (Erwin and Davidson 2009;Davidson 2010). Finally, transcriptome comparisons using coexpression networks and transcript composition showed slightly different results. Although the topology built using complete transcriptomes clustered S and C as a sister group to A, leaving P as the most dissimilar stage and PC and G as a separate group, the topology based on coexpression data, resolved P and A as a sister group to PC and G ( fig. 1D). In both cases, the similarity between C and S was clear, indicating the usage in the two stages of shared genome regulatory programs based on similar transcript composition. On the other hand, differences between topologies reveal that in vivo complexity is strongly dependent on network interactions and supports the idea that body plan morphogenesis and evolution is a "systemlevel problem" that cannot be understood by looking at developmental conserved genes in isolation (Peter and Davidson 2011).

Conclusion
Our study demonstrated that primary coral cultures are valuable tools for studying genome regulatory programs and revealed the existence of ancestral genome regulatory modules underlying pluripotency and cell differentiation in cnidaria. Caution must be taken to interpret in vitro experiments as C populations are heterogeneous cell types that drastically differ transcriptionally from the in vivo system.

Collection of Samples and Tissue Culture
Tip fragments ($3 cm) from six different colonies were kept in 50 ml falcon tubes containing 0.2 lm filtered seawater with antibiotics (FSWA) (1% Pen/Strep/L-Glu-solution, Sigma- Aldrich and 0.1% Fungizone, Invitrogen) prior the initiation of cultures. Samples were washed 3Â with FSWA and then incubated at 32 C for 4 h to induce bleaching (Desalvo et al. 2010). Fragments were washed (3Â) with FSWA and further incubated (2 h/gentle shaking) in calcium-free FSWA (Marshall and Clode 2004). Following tissue dissociation, naked skeletons were removed, and detached tissue centrifuged (1,500 rpm/10 min) and resuspended in 5 ml of FSWA. Cell pellets were gently washed 3Â with 5 ml of cell culture media (30% DMEM Gibco, 10% FBS Gibco, 1% Pen/Strep-solution, 0.1% Fungizone, 1% Glutamax, Gibco, 25 mM HEPES pH 8.0, and 55% 0.2 lm filtered seawater) and resuspended in 5 ml of fresh media. Founding cultures were kept individually in 6-well cultured plates and incubated at 23 C in the dark for 48 h. After that, 1 ml of the original cultures were used to inoculate 4 ml of fresh media in 6-well cultured plates and returned to incubation conditions. Cultures were monitored daily on a standard inverted light microscope fitted with a color digital camera. Media was changed when cultures reached 60% confluence. After 4 weeks, cells from three wells were harvested for RNA extractions.

Sequencing and Data Analysis
Library preparation and data analysis were conducted as reported in Reyes-Bermudez et al. (2016). To identify Acropora HM, we download the T-CDS data set from Hydra vulgaris strain AEP from http://www.compagen.org/datasets. html. Orthologs were determined using OrthoMCL v.1.4 with a BLASTp E value cut-off of 1e À5 , a minimum coverage of 70% and an inflation index of 1.5 (Li et al. 2003).

Supplementary Material
Supplementary data are available at Genome Biology and Evolution online.