Single-cell analyses of tissue regeneration in two true jellyfish

Abstract

The phylum Cnidaria is the outgroup of Bilateria and includes sea anemones, corals, hydroids, and jellyfish. Cnidarians play crucial ecological roles in marine ecosystems, including the formation of highly diverse and productive coral reefs, and acting as important predator and prey species. They are also well known for their remarkable regeneration capacities. Here, we report single-cell RNA sequencing of bell tissue remodeling/regeneration after amputation in two species of scyphozoans or “true jellyfish,” the Asian moon jelly, Aurelia coerulea, and the flame jellyfish, Rhopilema esculentum. We delineated 12 cell populations in Aurelia and Rhopilema and revealed their respective marker genes and enriched gene pathways. During this process, conserved transcription factor Otx, TFAP2A, Erg, NFIA, and Wnt/β-catenin signaling pathway genes were identified. Additionally, we discovered two conserved, sequentially activated patterns, with putative proliferative cells, gastrodermal cells, neural cells, and secretory gland cells modulated in the first phase, followed by cnidocytes in the second phase. Further comparison among cnidarian genomes identified a suite of lineage-specific scyphozoan genes, a subset of which were frequently significantly expressed in cnidocytes in both jellyfish species. Using powerful single-cell RNA sequencing approaches, this study elucidates the evolution of lineage-specific genetic networks and biological processes in true jellyfish, which remain comparatively poorly studied, and in particular provides key insights into the molecular pathways underlying their remarkable regenerative capacity.

Introduction

Animals display a fascinating diversity of body plans and cellular mechanisms that facilitate adaptation to the environment. The bilaterians comprise most extant animal species, and to understand how bilaterians evolved, it is essential to consider their immediate outgroup—the phylum Cnidaria. The Cnidaria contains ∼11,000 described species, including sea anemones, corals, hydroids, and jellyfish, that play important ecological roles in marine, and less commonly, freshwater environments. Macroevolutionary trends in gene repertoire evolution have been analyzed across many cnidarians over the last decade, with representative genomes of key cnidarian groups assembled, including anthozoans or “sea anemones and corals” (Putnam et al. 2007; Shinzato et al. 2011; Baumgarten et al. 2015; Shum et al. 2022; Yu et al. 2022), hydrozoans or “hydroids” (Chapman et al. 2010; Leclère et al. 2019), myxozoans (Chang et al. 2015), and scyphozoans or “true jellyfish” (Khalturin et al. 2019; Gold et al. 2019; Nong et al. 2020) and cubozoans or “box jellyfish” (Liegertová et al. 2015; Khalturin et al. 2019; Ohdera et al. 2019). These data have provided important insights into the evolutionary pathways of both bilaterians and cnidarians.

Regeneration refers to the ability to restore a lost body part and represents an important process in many animals. It likely contributes to evolutionary success, given the importance associated with replacing lost body parts (Gurtner et al. 2008; Bideau et al. 2021). Nevertheless, regenerative capacities vary across lineages in both cnidarians and bilaterians. For instance, while bilaterians such as amphibians, earthworms, reptiles, and zebrafish can heal wounds and regenerate certain lost organs (Liu et al. 2015; Nowoshilow et al. 2018; Shao et al. 2020), planarian flatworms can regenerate almost the whole body, similarly to some cnidarians (i.e. hydrozoans and anthozoans; DuBuc et al. 2014; Layden et al. 2016; Vogg et al. 2019; Sinigaglia et al. 2020; Vila-Farré et al. 2023). Yet, compared to many other cnidarians lineages, such as anthozoans and hydrozoans, the study of regeneration in scyphozoans has been relatively neglected.

Scyphozoans play important ecological roles throughout the oceans in different parts of the world and are well known to humans due to their often-painful stings, the occurrence of largescale jellyfish blooms, and their use in aquaculture as food in some countries. Scyphozoans have both asexually sessile polyps and dispersive and sexually reproductive medusae in their life cycles. Regenerative capacities have been demonstrated in certain species of scyphozoans at different life stages (Zeleny 1907; Steinberg 1963; Curtis and Cowden 1974; Stierwald et al. 2004; Abrams et al. 2015; Gamero-Mora et al. 2019; Abrams et al. 2021). However, the underlying cellular and genetic components contributing to these processes in scyphozoans remain poorly elucidated.

Here, we present the first single-cell transcriptomics analysis of bell remodeling/regeneration after amputation in two scyphozoan species—the Asian moon jellyfish, Aurelia coerulea (Semaeostomeae), and the edible flame jellyfish, Rhopilema esculentum (Rhizostomeae). Our findings provide insights into the evolutionary pathways underlying regeneration/remodeling in scyphozoans.

Results

Resequencing of a new moon jellyfish genome

Genomic resources for moon jellyfish Aurelia sp. and the edible flame jellyfish R. esculentum were generated previously (Fig. 1a and b) (Gold et al. 2019a, 2019b; Nong et al. 2020). However, genomes assembled from different populations of the cosmopolitan moon jellyfish show considerable variation (Khalturin et al. 2019; Dong et al. 2024). Correspondingly, sequence alignments between single-cell sequencing reads generated here and published genomes revealed sequence similarity of 96% to 98% for Rhopilema, but a staggering breadth of 39% to 95% for Aurelia (Tables S2, S6). The Aurelia draft genome originates from a specimen in California (Gold et al. 2019a, 2019b) and features an unusually high number of BUSCO duplicates and relatively short scaffold sizes (given today's standards). Thus, given these considerations, and the low rate of mapping observed for our Aurelia reads, we decided to generate a new assembly for a male moon jellyfish, obtained from Hong Kong (Figure S1). The new genome, Aurelia (CUHK_M), features an assembly size of 449.7 Mb (Fig. 1b) and is comparable to the published Aurelia genome size and duplicated BUSCO numbers but features considerably improved sequence continuity (contig N50 = 8.9 Mb; scaffold N50 = 22.4 Mb; Fig. 1c; Figure S1; Tables S1 to 3). We also achieved mapping rates of 90% to 92% for single-cell sequencing reads, considerably improving on efforts based on the other Aurelia draft genome (Tables S1 to 4, S6).

Figure 1.

a) Moon jellyfish Aurelia. b) Edible jellyfish Rhopilema. c) Genome assembly quality. d) Bell regeneration of Aurelia (upper) and Rhopilema (lower). The arrowheads depict the amputated regions. Scale bar = 1 cm. e and f) UMAP visualization of 12 cell clusters of Aurelia (e) and Rhopilema (f). Abbreviations: epidermal muscle (EM), gastrodermis (GAS), potential gastrodermis (PGAS), neural cells (NC), potential neural cells (PNC), cnidocytes (CC), secretory gland cells (SGC), putative proliferative cells (PPC) and unknown (UN). g) Cross species comparison between Aurelia and Rhopilema.

Aurelia and Rhopilema cell type atlas

Single-cell suspensions were generated by regenerating bell tissues of Aurelia and Rhopilema at five timepoints (0, 6, 12, 18, 24 h) for single-cell RNA sequencing (scRNA-seq; Fig. 1d; Tables S5, S8; Figure S2). The resultant scRNA-seq libraries contained an average of 53,280 reads/cell and 745 genes/cell for Aurelia, and an average of 35,597 reads/cell and 709 genes/cell for Rhopilema, which is comparable to other published cnidarian scRNA-seq studies (Sebé-Pedrós et al. 2018; Siebert et al. 2019; Hu et al. 2020; Chari et al. 2021; Levy et al. 2021; Dong et al. 2024). Using uniform manifold approximation and projection (UMAP) analyses on ∼27,000 and ∼31,000 single-cell profiles, 12 cell clusters with unique gene expression profiles were identified in both Aurelia and Rhopilema (Fig. 1e and f).

To generate the cell atlas, a combination of representative marker genes, KEGG/KOG/GO enrichment, and comparison with previously characterized cell types in other cnidarians were used to define cell clusters. For marker genes, these include myosin chain genes, supervillin (SVIL), tropomyosin genes, and calcium signaling regulators for epidermal muscle cells (Sebé-Pedrós et al. 2018; Tanaka et al. 2018; Nizhnichenko et al. 2022; Link et al. 2025); ELAV, advillin, synaptotagmin, tubulin chain, neuropeptide, and voltage-gated ion channels VGSC for neural cells (Nakanishi et al. 2012; Takahashi and Takeda 2015; Chuang et al. 2018; Geffeney et al. 2019; Dong et al. 2024; Li et al. 2024); Apolipophorin and Cathepsin L for gastrodermal cells (Sebé-Pedrós et al. 2018; Li et al. 2024); MUC2, CREB3L1, and zinc metalloendopeptidase ADAMTS for secretory gland cells (Fox et al. 2010; Sebé-Pedrós et al. 2018; Mayorova et al. 2019; Levy et al. 2021; Steger et al. 2022); and minicollagens and nematogalectin for nematocysts/cnidocytes (David et al. 2008; Hwang et al. 2010; Sebé-Pedrós et al. 2018; Hu et al. 2020). For Aurelia cluster 4 and Rhopilema cluster 11, cell cycle-related marker genes including centromere proteins, DNA topoisomerases, KI67, myc, myb, and SMC family proteins were expressed. Given stem cells (or I-cells) have only been identified in hydrozoans (Gold and Jacobs 2013), we have named these clusters as putative proliferative cells (Hartl et al. 2010; Dong et al. 2024; Fig. 2a to c; Figures S4 and 5; Supplementary Data S1 to 3).

Figure 2.

a and b) Marker genes of Aurelia cell clusters (a) and R. esculentum cell clusters (b). c) Expression of cell type-specific marker genes in different cell types of Aurelia (left) and Rhopilema (right). d) Expression of conserved cnidocyte toxin genes in Aurelia (left) and Rhopilema (right). e) Enriched KEGG pathways of genes in the same cell types between Aurelia and Rhopilema.

In addition, gene pathway enrichment analysis was conducted to explore if any functional gene ontology terms, including those from the GO, KEGG, and KOG databases were enriched in the different cell types (Fig. 2e; Figure S14; Tables S11 and 12; Supplementary Data S4 to 7). Conserved gene pathways were identified in most assigned cell types, including cnidocysts/nematocysts, epidermal muscle, secretory gland cells, neural cells, and putative proliferative cells, between Aurelia and Rhopilema. For instances, epidermal muscle cells have enriched KEGG pathways related to cellular respiration (Aurelia cell clusters 0 and 11, and Rhopilema cell clusters 3 and 5) (Fig. 2e), while putative proliferative cells (Aurelia cell cluster 4 and Rhopilema cell cluster 11) have enriched KEGG pathways in cell cycle (Table S11; Supplementary Data S4 and 5); enriched KOG pathways in “Cell cycle control, cell division, chromosome partitioning,” “RNA processing and modification,” and “nucleotide transport and metabolism” (Figure S14); and enriched GO pathways such as mitotic nuclear division, chromosome segregation, spindle assembly, organellar fission, and meiotic cell cycle (cell cycle and proliferation) (Supplementary Data S6 and 7; Table S12).

Furthermore, our cell cluster assignments are also supported by single-cell RNA data comparisons to other cnidarians, including hydrozoans Clytia hemisphaerica and Hydra vulgaris, sea anemone Nematostella vectensis, and corals Xenia sp. and Stylophora pistillata (Siebert et al. 2019; Hu et al. 2020; Chari et al. 2021; Levy et al. 2021) (Figures S7 to 11; Table S10).

These analyses allow us to confidently identify a total of eight cell types, including cnidocytes/nematocytes (CC; Aurelia clusters 3, 8, 9, Rhopilema cluster 6), epidermal muscle (EM; Aurelia cluster 0, 11, Rhopilema cluster 3, 5), gastrodermis (GDC; Aurelia cluster 1, Rhopilema cluster 2), potential gastrodermis (PGDC), neural cells (NC; Aurelia clusters 5, 6, Rhopilema clusters 7, 8), potential neural cells (PNC), secretory gland cells (SGC; Aurelia cluster 10, Rhopilema clusters 9, 10), and putative proliferative cells (PPC; Aurelia cluster 4, Rhopilema cluster 11). Additionally, we performed mass spectrometry for cnidocysts, revealing conserved expression of four potential toxins in the assigned cnidocytes/nematocytes of both Aurelia and Rhopilema (Fig. 2d; Figure S12; Supplementary Data S8 and 9). Thus, a combination of signature marker gene expression and mass spectrometry allowed us to solidly assign cell identities.

Sesquiterpenoid hormone and its biosynthetic pathway genes were previously described in jellyfish (Nong et al. 2020). Utilizing the cell atlas generated here, we found that genes involved in the sesquiterpenoid hormone biosynthetic pathway, including AACT, HMGCS, HMGCR, MVK, MDC, FNTB, FNTA, ZMPSTE24, and IMCT, are expressed at different levels among cell types (Figure S13). Among these genes, AACT and ICMT have the highest expression level in most cell types in both species.

Gene modulation during tissue remodeling/regeneration in Aurelia and Rhopilema

To assess the interplay of cellular and underlying molecular processes contributing to early remodeling/regeneration response in Aurelia and Rhopilema, we compared scRNA-seq data generated at five distinct timepoints (Figure S15). Using 0 h of putative proliferative cells (Aurelia cell cluster 4 and Rhopilema cell cluster 11) as the initial cell population, pseudo-time analyses revealed both conserved and divergent gene expression patterns in different cell clusters in Aurelia and Rhopilema (Figure S16). For example, in both species, the trajectory initiated with the proliferative cells, followed by the emergence of gastrodermal cells, secretory gland cells, and neural cells in the early phase of the process, while cnidocytes are specified in the second phase. Epidermal muscle cells differentiated concurrently with gastrodermis, neural cells, and secretory gland cells in early phase of such process in Aurelia, and yet, these cell types developed in the later phase in Rhopilema. To understand the genetic regulatory mechanisms underlying this process in Aurelia and Rhopilema, gene enrichment analyses were performed on the differentially expressed genes in the respective species (Supplementary Data S10 to 14; Figure S17). KEGG enrichment revealed that “focal adhesion” and “leukocyte transendothelial migration” gene pathways were upregulated in epidermal muscle of both species at 6 h post-amputation (Fig. 3a; Figure S18). Given that cnidarians do not possess a lymphatic system, we explored genes contributing to the term “leukocyte transendothelial migration”. We rationalize this, since (i) the molecular innovations necessary for leukocyte transendothelial migration presumably predate the lymphatic system itself, in the form of ancient adhesion and signaling molecules (eg integrins, selectins); (ii) jellyfish also possess genes for phagocytosis, diapedesis, and actin-based motility, which are essential for migration cells associated with host defense toward sites of defense relevance; (iii) jellyfish possess specialized phagocytic cells in the form of amebocytes; and (iv) jellyfish possess a complex epithelium, where immune-like cells migrate for host defense. In Aurelia, catenin alpha-2-like, vinculin-like, and POTE ankyrin/actin-related proteins were differentially expressed, while radixin-like, myosin regulatory light chain 9, afadin, paxillin-like, and various integrins were differentially expressed in Rhopilema (Supplementary Data S15). Thus, the modulation of these genes probably reflects cell motility and tissue reorganization, rather than leukocyte-specific processes. Regarding the “Focal adhesion,” filamin-A is commonly up-regulated genes in both species, and Aurelia and Rhopilema have different DEGs in the pathway (laminins, POTE ankyrin/actin-related proteins, and vinculin-like in Aurelia and integrins, paxillin-like, and myosin regulatory light polypeptide in Rhopilema) (Supplementary Data S11 and 12, S15 and 16).

Figure 3.

a) Enriched KEGG pathways of up-regulated genes in epidermal muscle. b) Expression of conserved differentially expressed genes (DEGs) in different cell types during bell remodeling in Aurelia (left) and Rhopilema (right). Asterisk “*” indicates DEGs with low expression level in the dataset.

To further elucidate the genes involved in these processes, we compared differentially expressed genes (DEGs) between Aurelia (343) and Rhopilema (357) (Supplementary Data S10; Figure S17). Among these, 52 Aurelia DEGs have orthologous relationships with 51 Rhopilema DEGs and exhibit cell type-specific expression in respective species (Supplementary Data S18 and 19). As shown in Figure S20, several conserved gene pathways were enriched in epidermal muscle of both species (Supplementary Data S19). In addition, we have also investigated genes that have well-known functions in regeneration/tissue remodeling from other studies (Supplementary Data S20 to 23), and transcription factors (Erg, TFAP2A, OtxD, and NFIA) and extracellular matrix genes (eg collagen α-1(I) chain-like, laminin, matrix metalloproteinase-2-like enzymes, dystroglycan adhesion complexes, cytoskeletal regulators filamin-A-like, myosin-10, tropomyosin-2) are differentially expressed in both Aurelia and Rhopilema.

Scyphozoan-specific genes expression during the tissue remodeling/regeneration process

To understand patterns in gene gains in scyphozoans, we compared orthologous genes across the genome assemblies of four scyphozoans, ten additional cnidarians, and a placozoan outgroup (Fig. 4a). A total of 435 scyphozoan-specific orthologues were revealed. Of these, 193 were shared only between Aurelia and Rhopilema (Fig. 4a). Focusing on the scyphozoan-specific orthologues that were expressed in the scRNA-seq dataset generated here (592 in Aurelia and 603 in Rhopilema), 24 scyphozoan-specific genes were found to be expressed in the same cell type in both Aurelia and Rhopilema (Fig. 4b; Figures S21 to 26; Supplementary Data S24 and 25). Half of these scyphozoan-specific genes were expressed specifically in the cnidocytes (n = 12), while the others were expressed in epidermal muscle (n = 7), neural cells (n = 3), and secretory gland cells (n = 2) (Fig. 4b, Figures S21 to 26). Notably, the three scyphozoan lineage-specific minicollagens (as determined by genome searches and gene trees) were expressed specifically in the cnidocytes, including OG0020136 as previously reported (Gold et al. 2019a, 2019b). Meanwhile, two scyphozoan-specific genes that were also found in the proteomes were expressed specifically in cnidocytes (n = 1) and neural cells (n = 1) in Aurelia (Fig. 4c; Supplementary Data S26), but they were expressed in multiple cell types in Rhopilema (Figures S27 to 29).

Figure 4.

a) Summary of genes gained in each investigated metazoan lineage. b and c) Scyphozoan-specific orthogroups (b) and toxins (c) with conserved cell-type expression in Aurelia and Rhopilema. d) Dot plot showing scyphozoan lineage-specific genes during bell regeneration/remodeling in Aurelia (upper) and Rhopilema (lower). OG0024159 with conserved cnidocyte cell expression is colored in green.

In addition, 14 scyphozoan-specific genes were differentially expressed in either Aurelia or Rhopilema between the investigated timepoints (Fig. 4d; Figures S30 to 32; Supplementary Data S24 and S27). Seven of these were upregulated, including a putative surface protein (OG0015763), an uncharacterized protein with Src homology 3 domain (SH3_1, Pfam00018) (OG0015694), and five uncharacterized proteins (OG0011099, OG0015585, OG0017689, OG0017689, and OG0019916), while four of them were downregulated including a minicollagen (OG0024159) and three other uncharacterized proteins (OG0009415, OG0015616, and OG0024205). Among these 11 scyphozoan-specific DEGs, 4 of them were also found to be expressed in the same cell types in both jellyfish species, including OG0015694 and OG0019916, which were significantly upregulated in epidermal muscle; putative surface protein OG0015763, which was upregulated in cnidocytes; and minicollagen OG0024159, which was downregulated in cnidocytes (Fig. 4d).

Discussion

Here, comparative analyses considering the results of analyses of two newly generated single-cell transcriptome libraries, alongside previously reported findings for other cnidarians, reveal several novel insights into the cell types, conserved gene regulation, and lineage-specific genes involved in cnidarian regeneration/remodeling processes.

Genomes provide a crucial foundation for developing improved understanding of the conserved and divergent genomic characters present in Cnidaria, and Bilateria more generally (Khalturin et al. 2019; Gold et al. 2019a, 2019b; Nong et al. 2020). To gain a better understanding of the genomic sequences contributing at the individual cell level, scRNA-seq has proven a powerful tool to study the hallmarks of metazoan evolution (eg Sebé-Pedrós et al. 2018; Shao et al. 2020; Chari et al. 2021; Ghaddar et al. 2023; Cole et al. 2024; Dong et al. 2024; Mah and Dunn 2024; Baranyk et al. 2025). Here, we first show that genome quality and aspects related to genome quality, as well as the gene model prediction algorithms adopted, contribute to considerable variation in the mapping rate of scRNA-seq reads in nonmodel organisms. Hence, we provide a new genome assembly for a male Asian moon jellyfish Aurelia from Hong Kong and test various gene prediction models to establish a more accurate foundation for performing subsequent analyses.

Regeneration in metazoans has been examined anatomically in diverse lineages for centuries, and the processes in vertebrates and cnidarians are remarkedly different (Bideau et al. 2021). Among cnidarians, tissue remodeling/regeneration in scyphozoans is relatively understudied compared to other lineages, such as anthozoans (eg Exaiptasia and Nematostella) and hydrozoans (eg Clytia and Hydra). By comparing the results of scRNA-seq in scyphozoans generated in this study to those for other cnidarians, we show that regenerating/remodeling of tissues involves similar cell types and marker genes to other scyphozoans (Dong et al. 2024; Li et al. 2024; Link et al. 2025; Supplementary Data S28). These include Cathepsin L (CATL), consistently identified as gastrodermal cell marker across datasets; myosin light chain kinase (MLCK), Myosin heavy chain (MYS), Myosin essential light chain (MELC), Tropomyosin (TPM1/TPM2), Calmodulin, and Myosin regulatory light chain (RLC) in epidermal muscle cells; minicollagen, DKK3, LMAN1, nematogalectin, and nematogalectin-related genes in cnidocytes; and ELAVL, TUBA1A, CAB32, and SYT16 in neural cells, KI67 in proliferative cells, and MUC2 in secretory gland cells. Nevertheless, we also note that certain cell populations identified in the Dong et al. (2024) whole organism study, such as immune cells (marked by ITLN1, ITLN2, IRF2, MUC2, FKBP12, LAR, and SYT15), mesogleal cells (Mesoglein, SPON1, SC6A1, SSPOP, PRSS27, and CSMD1), and hair cells (LOXHD1, LRP5, CFAP141, TRPC4, CACNA1E, KIF13B, and DNAH1) did not form distinct clusters in our analyses.

Hormone biosynthesis is a crucial component in animal development and evolution, including metamorphosis. In a previous study, sesquiterpenoid hormone production, previously considered to be typical of arthropods only, was discovered in jellyfish and other cnidarians (Nong et al. 2020). Nevertheless, knowledge of sesquiterpenoid hormones in cnidarians remains inadequate. In the Aurelia and Rhopilema cell type atlas generated here, sesquiterpenoid biosynthetic pathway genes were found to be expressed in various cell types. In particular, AACT (Acetyl-CoA acetyltransferase) in the mevalonate pathway and ICMT (isoprenylcysteine carboxyl methyltransferase) downstream to mevalonate pathway displayed strong expression levels. Given none of these genes were differentially expressed throughout the regeneration/remodeling process in both jellyfish, we suggest that sesquiterpenoid hormones in jellyfish may be involved in life stage transformation and/or reproduction (as in other invertebrates), rather than in playing a key role in scyphozoan remodeling/regeneration.

Transcription factors are key components of genome function, which have provided fruitful clues into our understanding of metazoan regeneration. Here, we revealed that the transcription factors OtxD, TFAP2A, Erg, and NFIA in epidermal muscle, and NFIA in secretory gland cells, were upregulated in both species of true jellyfish. Whether this finding represents the case that these transcription factors are shared across cnidarians to generate specific cell types or indicate a potentially shared genetic regulatory network remain an exciting area to be differentiated and tested functionally. However, at present, we find limited evidence to support a conserved scyphozoan-specific regeneration regulatory network.

Lineage-specific genes provide genetic novelty and are frequently identified in metazoan genomes via comparative genomic analysis. Most of our understanding of lineage-specific genes relates to their evolutionary origins, via processes such as gene duplication or horizontal gene transfer (eg So et al. 2022). The extent to which novel genes contribute to lineage-specific biological processes such as regeneration remains an exciting and largely unexplored area. Here, we first show that scyphozoan-specific genes are expressed with a conserved pattern in certain cell types, with about half specific to cnidocytes. This explosive cell type is well known to have a distinctive structure and be specific to the Cnidaria for purposes of delivering a painful sting to prey and predators, via delivery of a fast-acting venom stored within the nematocyst, triggered by mechanical stimulation. In addition, we also identified scyphozoan-specific genes that are differentially expressed during regeneration/remodeling in the two jellyfish species considered here. However, we only identified four such genes that were also expressed in the same cell type between the two species (two in epidermal muscle and two in cnidocytes). Overall, these findings imply that certain lineage-specific genes are under selective pressure to contribute distinct novel morphological structures and biological processes that characterize the diverse lineages that have evolved in the metazoan tree of life.

Material and methods

Animal collection and husbandry

Individuals of Aurelia and Rhopilema were purchased from a local supplier in Hong Kong. Medusae of both species were maintained and cultured in 30 ppt circulating seawater at 25 °C at the Chinese University of Hong Kong. Animals were fed three times weekly with newly hatched brine shrimp Artemia.

DNA extraction of male Aurelia and sequencing

The bell and gonad tissues of a male Aurelia sp. individual were dissected and snap-frozen in liquid nitrogen. The gonad tissue was used for sex identification via histological examinations with modifications that the tissue was immersed in 4% formalin for 24 h fixation (García-Rodríguez et al. 2018). The hematoxylin-eosin stained sections were examined under a light microscope (Leica DM500), and the bell tissue was sent to Dovetail Genomics (United States) for high molecular weight DNA isolation. The fragment size and quantity of the isolated HMW DNA was inspected in a 5200 Fragment Analyzer System (Agilent Technologies). A PacBio SMRTbell library was prepared and sequenced on the PacBio Sequel IIe sequencing platform with two SMRT cells to obtain ∼33 Gb data. The Omni-C library and sequencing data was generated by Dovetail Genomics LLC. Briefly, the chromatin was fixed with formaldehyde and then extracted from the nucleus, followed by digestion with DNAse I. Chromatin end repair was performed and a biotinylated bridge adapter was used for proximity ligation of adapter containing ends. Subsequently, crosslink reversal was carried out, followed by DNA purification and removal of biotins that were not ligated to DNA fragments. The library was generated with the use of NEBNext Ultra enzymes and Illumina-compatible adapters. The biotin-containing fragments were retained using streptavidin beads, followed by a PCR enrichment step. The final library was sequenced on an Illumina HiSeqX platform to produce Omni-C data of ∼48× sequencing coverage.

Genome assembly

De novo genome assembly was performed using Hifiasm (version 0.19.8). Haplotypic duplications were identified and removed using purge_dups (version 1.2.5). The primary genome assembly was checked for contamination using BlobTools (version 1.1.1). Proximity ligation data from the Omni-C library was used to scaffold the genome assembly using YaHS (version 1.2a.2). The final assembled genome was soft-masked using redmask (version 0.0.2), and gene models were predicted using funannotate (version 1.8.15) with the parameters “–protein_evidence uniprot_sprot.fasta –cpus 100 –genemark_mode ET –optimize_augustus –busco_db metazoa –organism other –max_intronlen 350000” as previously described (Nong et al. 2020; So et al 2022).

Transcriptome sequencing and gene model prediction

Male Aurelia gonad was pretreated with a CTAB solution (1 M Tris-HCl, 0.2 M EDTA, 5 M NaCl, CTAB, 1% PVP, and 1% beta-mercaptoethanol) prior to extraction of RNA using TRIzol solution (Invitrogen). Quality control was performed using a NanoDrop spectrophotometer (Thermo Scientific), gel electrophoresis, and an Agilent 2100 Bioanalyzer (Agilent RNA 6000 Nano Kit). Transcriptome sequencing was carried out on Illumina platform (Table S1), and data were first processed using trimmomatic (version 0.39) and Kraken (version 2.0.8 with database k2_standard_20210517) and assembled as previously described (Nong et al. 2020). Other transcriptomic data of Aurelia were also downloaded from NCBI database and mapped to the soft-masked genome using HISAT2 (version 2.2.1). The data generated in this study, as well as sample SRR8040389, have sequence alignment rates of 84.38% and 84.56%, respectively, and were further used for genome annotation using funannotate (version 1.8.15) and BRAKER (v3.0.8).

Orthologous genes assignment

The longest transcript of gene sets from 15 taxa including 4 scyphozoans (Aurelia sp., R. esculentum, Chrysaora quinquecirrha, and Sanderia malayensis), 1 cubozoan (Morbakka virulenta), 3 hydrozoans (C. hemisphaerica, Hydractinia symbiolongicarpus, and H. vulgaris), 6 anthozoans (Actinernus sp., N. vectensis, S. pistillata, Acropora millepora, Xenia sp., and Dendronephthya gigantea), and an outgroup from placozoan (Trichoplax adhaerens) were used to infer the gene orthology using OrthoFinder v2.5.5 (Emms and Kelly 2019). DupGen_finder (Qiao et al. 2019) and MCScanX (Wang et al. 2012) were used to identify syntenic gene pairs between the four analyzed scyphozoan genomes. Potential scyphozoan-specific genes were searched against NCBI nr database using blastp to check if only scyphozoan hits were returned. In addition, the scyphozoan-specific genes were searched for conserved domains in NCBI Conserved Domain Database (Yang et al. 2020; Wang et al. 2023), proteomes of investigated taxa using HMMER (version 3.3.1; cut-off E-value <10⁻⁵) (Eddy 2011), and NCBI nr database. Amino acid sequences were aligned using MAFFT v7.455 (Katoh and Standley 2013), followed by gene tree constuction with FastTree (Price et al. 2010). Sequence alignment, gene tree, and gene synteny were visualized in Jalview v2.11.4.1 (Waterhouse et al. 2009), evolview v3 (Subramanian et al. 2019), and R package gggenomes (Hackl et al. 2021), respectively.

Jellyfish single-cell dissociation and transcriptome sequencing

Bells of individual jellyfish were excised with a stainless-steel trapezium mold (80 mm top, 150 mm bottom, 70 mm height). Then, 2 mm of tissue was cut along the amputated region at 0, 6, 12, 18, and 24 h post-amputations. Excised tissues were put in trypsin dissociation solution (10× TrypLE Expression solution, ThermoFisher) and incubated for 30 min at 37 °C and 20 min at 25 °C for Rhopilema and Aurelia, respectively. Cells were collected by centrifugation at 600 × g for 5 min and resuspended in phosphate-buffered saline (PBS) solution containing 0.04% bovine serum albumin (BSA). Aggregated cells were filtered with Flowmi cell strainer (Bel-Art H-B Instrument). Cells were stained with trypan blue and were counted using hemocytometer to estimate their viability and concentrations. Only cell suspension with cell viability >90% would be further proceeded to barcoding and library construction using the Chromium Next GEM technology and quantified with High Sensitivity D5000 and D1000 DNA ScreenTape assays (Agilent). Sequencing was carried out using Illumina NovaSeq6000 platform (PE150) with a median depth of 100,000 read pairs per cell.

Single-cell RNA (scRNA) transcriptome data clustering and cell type atlas identification

ScRNA sequencing data were mapped to the respective reference genomes of Aurelia and Rhopilema with 10× Genomics Cell Ranger (v7.1.0). Cell-UMI count tables were uploaded to Seurat (v4.4.0), with dead cells, low-quality cells, empty droplets, and cell doublets were filtered (Hao et al. 2021). For identification of common cell types, data from all five timepoints of each jellyfish species were merged and integrated using FindIntegrationAnchors and IntegrateData. ElbowPlot was used to identify the optimal dimensionality, while Clustree_0.4.4 was applied to detect the optimal resolution of clustering (Zappia and Oshlack 2018). The clustering results were visualized with uniform manifold approximation and projection (UMAP). Published cnidarian single-cell transcriptome data, including H. vulgaris, C. hemisphaerica, Xenia sp., S. pistillata, and N. vectensis (Sebé-Pedrós et al. 2018; Siebert et al. 2019; Hu et al. 2020; Chari et al. 2021; Levy et al. 2021) were used to define cell types using OrthoFinder (version 2.5.4) with the parameters “-t 138 -a 138 -M msa -S diamond”. Kullback–Leibler divergence (KLD) was calculated for the cell types of Aurelia, Rhopilema, and other cnidarians (Levy et al. 2021). Conserved cell markers for different cell populations were identified using Seurat anchors FindAllMarkers (logfc.threshold = 0.25). Representative genes for each cluster were further manually annotated and checked. Marker genes of each cluster were also tested for KEGG, GO, and KOG enrichment analyses (Wu et al. 2021; So et al. 2022).

Pseudotime and differential gene expression analyses

Cell differentiation trajectories were inferred using R package Monocle3 (Cao et al. 2019), with the putative proliferative cells of 0 h of each species being used as root cells for pseudotime analysis. For identification of differentially expressed genes (DEGs) in each cell type, data from 6, 12, 18, and 24 h post-amputation were compared to both 0 h as well as preceding timepoint of respective species. Functional term annotation for DEGs was performed using EggNOG-mapper v2.1.2 (Cantalapiedra et al. 2021), while enrichment analyses for KEGG pathways and Gene Ontology (GO) were carried out using “clusterProfiler” (Wu et al. 2021). Gene expression changes were visualized with either “ggplot2” package in R or Seurat anchor FeaturePlot or VlnPlot.

Liquid chromatography-tandem mass spectrometry (LC-MS/MS) on nematocysts

Nematocysts were enriched as previously described (Bloom et al. 1998) with the following modifications. Jellyfish samples were first placed in 1:10 (v:v) 35 g/L NaCl at 4 °C, allowing the tissues to autolysis for 4 d. The autolyzed mixture was sequentially filtered through 200 µm and 85 µm cell strainer to remove debris. Proteins were extracted from enriched nematocysts by dissolving in lysis buffer (6 M urea, 2 M thiourea, 1 mM dithiothreitol [DTT] in 250 mM Tris [pH 7.6]) supplemented with Pierce protease inhibitor (Thermo Fisher Scientific). Protein samples were alkylated with 5 mM of iodoacetamide for 30 min in the dark and further digested overnight at 37 °C with sequencing-grade trypsin (Promega) at a 1:20 ratio. The resulting peptides were desalted with Pierce C18 spin columns (Thermo Fisher Scientific) per the manufacturer's guidelines. LC-MS/MS analysis was performed using a Dionex UltiMate 3000 RSLC nano system interfaced with an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific). MS and MS/MS scans were acquired in the Orbitrap with a mass resolution of 60,000 and 15,000, respectively. The higher-energy collisional dissociation (HCD) mode was used as the fragmentation mode with 30% collision energy, while the precursor isolation windows were set to 1.6 m/z. Acquired mass spectra were analyzed by Proteome Discoverer version 2.4 with SEQUEST as a search engine. Data were searched against the translated protein sequences of Aurelia and Rhopilema transcriptomes. Putative toxins were first identified by searching protein sequences against the UniProt animal toxin and venom database using BLASTp (e-value of <1.0 × 10⁻⁵) (Jungo et al. 2012), with ToxPred2 being used to further exclude proteins with nontoxic physiological functions (Sharma et al. 2022).

Author contributions

Y.L., T.H.L., and J.T. carried out the single-cell isolation. Y.L., S.T.S.L., W.N., and Y.X. carried out the single-cell analyses. W.N. carried out the gene annotation. W.L.S., T.C.N.L., and S.M.N. carried out the toxin proteomic analyses. O.J. and J.Z. assembled the genome. H.Y.Y. ensured the sequencing project management and animal husbandry. A.P.Y.C., K.F.L., A.J., Z.K., W.G.B., A.H., Y.W., T.F.C., and S.M.N. contributed the discussion of project at different stages. J.H.L.H. designed and coordinated the project. Y.L., S.T.S.L., W.N., and J.H.L.H. wrote the initial manuscript; all authors revised and contributed to the final version of the text.

Supplementary material

Supplementary material is available at Molecular Biology and Evolution online.

Funding

This work was supported by the Hong Kong Research Grant Council General Research Fund (14103823; 14102224) and Collaborative Research Fund (C4015-20EF) and the TUYF Charitable Trust.

Data availability

The raw reads generated in this study have been deposited in the NCBI database under BioProject accession numbers PRJNA1229829 (PacBio HiFi and Omni-C reads of Aurelia), PRJNA1049062 (single-cell sequencing data of Aurelia), and PRJNA1045829 (single-cell data of Rhopilema). The genome and genome annotation files have been deposited in the Figshare dataset (https://figshare.com/s/f89fdf550cf68d97a541) and CUHK Research Data Repository (https://doi.org/10.48668/JVK8SW). The scripts used in this study can also be found at GitHub repository (https://github.com/Fanny-bio/Jellyfish.rg.scRNA/tree/main) and CUHK Research Data Repository.