Marine Invertebrates One Cell at A Time: Insights from Single-Cell Analysis

Synopsis

Over the past decade, single-cell RNA-sequencing (scRNA-seq) has made it possible to study the cellular diversity of a broad range of organisms. Technological advances in single-cell isolation and sequencing have expanded rapidly, allowing the transcriptomic profile of individual cells to be captured. As a result, there has been an explosion of cell type atlases created for many different marine invertebrate species from across the tree of life. Our focus in this review is to synthesize current literature on marine invertebrate scRNA-seq. Specifically, we provide perspectives on key insights from scRNA-seq studies, including descriptive studies of cell type composition, how cells respond in dynamic processes such as development and regeneration, and the evolution of new cell types. Despite these tremendous advances, there also lie several challenges ahead. We discuss the important considerations that are essential when making comparisons between experiments, or between datasets from different species. Finally, we address the future of single-cell analyses in marine invertebrates, including combining scRNA-seq data with other ‘omics methods to get a fuller understanding of cellular complexities. The full diversity of cell types across marine invertebrates remains unknown and understanding this diversity and evolution will provide rich areas for future study.

Introduction

The basic unit of all animal life is the cell, a “small membrane-bound compartment filled with concentrated aqueous solutions of chemicals” that forms living organisms (Alberts et al. 1989). Ever since Robert Hooke coined the term “cells” (Hooke 1665), immense progress has been made in understanding the biology of cells and the cellular complexity that makes up the diversity of animal life. To better understand how this organismal diversity arose, the field of comparative evolutionary developmental biology has grown alongside technical advances in molecular biology and greater taxonomic sampling across the tree of life (Bonner et al. 2012).

What is a cell type?

This definition varies greatly depending on the method by which cell properties are evaluated (Valentine 2003). Historically, cell type definitions had been used to name cell types: most commonly morphological features such as size, shape, and pigmentation defined a cell type, but other functional characteristics (such as mucus-producing mucosal cells), or even the name of the researcher (as with Schwann cells) were employed instead (Schwann 1839; Valentine 2003). There are also evolutionary definitions, which describe a cell type as a set of cells in an organism that change in evolution together and are more closely related evolutionarily to each other than other cells (Arendt et al. 2016). Other cell type definitions have centered on differential gene expression to profile and describe cell types; recent studies have used higher throughput sequencing to perform these methods transcriptome-wide (Shapiro et al. 2013; Kotliar et al. 2019). Advances in single-cell isolation and DNA sequencing have made it possible to sequence the transcriptome of cells at a single-cell level (Tang et al. 2009; Islam et al. 2011; Jaitin et al. 2014; Klein et al. 2015; Trapnell 2015; Cao et al. 2017; Tanay and Regev 2017; Svensson et al. 2018). These species-agnostic technologies have allowed scientists to explore the molecular diversity of cell types and gain a deeper understanding of the cell as a fundamental unit across a range of marine invertebrates (Tanay and Sebé-Pedrós 2021; Li et al. 2022). Given the distribution of marine invertebrate species across animal phyla, these groups provide an opportunity to study the diversity and evolution of cell types that underlie the amazing diversity of marine invertebrate body plans and life history strategies.

A brief introduction to single-cell RNA sequencing

In single-cell RNA-sequencing (scRNA-seq), tissues are dissociated with physical or enzymatic digestion into single-cell suspensions, and then isolated individually by physical separation, such as flow cytometry, or in microfluidic droplets (Klein et al. 2015; Macosko et al. 2015). Additional techniques to isolate single cells for sequencing include nanowells and split-pool ligation barcoding (Rosenberg et al. 2018). After single cells have been isolated and barcoded to track transcripts back to their cell of origin, cells are then subjected to RNA sequencing. Once this sequence data are collected, it is mapped back to a species-specific reference genome or transcriptome, and cells can be plotted in a multidimensional gene expression space. Principal component analysis is used to group cells into clusters that represent various transcriptomic cell types in this multidimensional gene expression space. Then, dimensional reduction techniques such as Uniform Manifold Approximation and Projection for Dimension Reduction (UMAP) or t-distributed stochastic neighbor embedding (t-SNE) are used to visualize these data (Maaten and Hinton 2008; Becht et al. 2019). UMAP and t-SNE methods allow for visualization of high-dimensional data, but it is important to note that these methods are non-linear, meaning that distance between cells does not necessarily reflect similarity in high dimensional space (Chari and Pachter 2021; Wang et al. 2021). These approaches allow researchers to generate cell type atlases where cells are plotted based on gene expression and groupings are made of transcriptionally similar cells. So far, the transcriptomic cell types identified with these techniques have recapitulated known cellular diversity within organisms, but new (sometimes rare) cell types have also been predicted. Overall, scRNA-seq provides expression profiles of individual cells that can be used in many analyses downstream.

Insights from single-cell RNA-seq in marine invertebrates

scRNA-seq in marine invertebrates has provided three major insights: capturing the molecular diversity of cell types, clarifying the cellular contribution to dynamic processes, and generating new hypotheses on the evolution of cell types. Many marine invertebrate species are well suited for this approach as whole organisms can be dissociated into single-cell suspensions and then subsequently sequenced as individual cells. A key insight that scRNA-seq has provided in marine invertebrates is the transcriptional state of cells, not only in static snapshots, but over broader dynamic processes such as development and regeneration. In these studies, scRNA-seq allows researchers to generate hypotheses of cellular progressions with a class of tools that use the concept of “pseudotime” in which cells are ordered according to expression similarity (Haghverdi et al. 2016). Additionally, phylogenetic methods have been applied to scRNA-seq data to identify the possible branch points that may have led to the evolution of new cell types. In the following sections, we describe the various scRNA-seq efforts that have been performed in marine invertebrates across the major animal lineages and contextualize the insights of these findings (Fig. 1). Finally, we highlight some of the current challenges in scRNA-seq and suggest possible future avenues of investigation.

Fig. 1.

Distribution of scRNA-seq studies across a range of marine invertebrate phyla. The dashed line indicates uncertainty in the placement of Xenacoelomorpha. Blue or yellow circles denote whether whole animals or specific tissues were used as input in scRNA-seq libraries. For the life history stage being sampled green circles denotes embryonic sampling, orange squares denote a larval stage, pink polygons denote a juvenile stage, and cyan ovals denote adult tissue. Stars denote a perturbation that was performed as part of the scRNA-seq study, starvation in the cnidarian C. hemisphaerica and morpholino-knockdowns in the echinoderm S. purpuratus and tunicate C. intestinalis. Public domain silhouettes courtesy of PhyloPic.

Ctenophora

Ctenophora are a group of marine invertebrates that include the pelagic comb jellies and are characterized by rows of combs (ctenes) used for locomotion. In a single-cell study in Mnemiopsis leidyi, many of the cell clusters could not be assigned to known functions/cell types as many were strongly associated with unannotated, often Ctenophore-specific proteins (Sebé-Pedrós et al. 2018a). Here researchers took advantage of cell type functional inference, recovering a population of cells which express known photoproteins and opsins, and determining they are photocyte cells responsible for ctenophore bioluminescence.

Porifera

One of the earliest branching animal groups, sponges are filter-feeders that live both in marine and freshwater environments and morphologically appear quite simple (Hyman 1940). Sponges are made of three main cell types: choanocytes, pinacocytes, and archaeocytes. These three cell types have previously been well characterized, and scRNA-seq in the marine sponge Amphimedon queenslandica not only confirmed these cell types, but also revealed that they are made of multiple distinct cellular subpopulations (Sebé-Pedrós et al. 2018a). By sampling two life history stages, it was also clear that most A. queenslandica larval cell types have a distinct transcriptional signature compared to the adult; only larval archaeocytes show strong similarity to a cell type in the adult. In additional work in freshwater sponge, Spongilla lacustris, a greater number of cells were sequenced and new cell types were named to reflect the role they play within the organism (Musser et al. 2021). From the expression profiles of distinct cell types, a phylogenetic approach of neighbor-joining tree reconstruction was used to build a cell type tree. This approach of building an evolutionary lineage of cell types is distinct from a developmental trajectory of cell types and allowed the authors to propose the evolutionary relatedness of versus homoplasy of sponge cell types.

Placozoa

Placozoans are tiny, benthic, marine animals with ciliary-based locomotion and a morphologically simple body plan composed of two cell layers (Schierwater 2005). Placozoans are the simplest (nonparasitic) multicellular animals and have no apparent body axis or tissue-level organization. While only six cell types had been reported according to earlier ultrastructural studies (Smith et al. 2014; Schierwater and DeSalle 2018), scRNA-seq in Trichoplax adhaerens revealed 11 cell types, including six which produce unique cell-type-specific regulatory peptides (Sebé-Pedrós et al. 2018a). This is significant as placozoans do not have a nervous system, but one population of cells recovered in the study expresses the regulatory peptide TaELP, which regulates T. adhaerens locomotion through control of ciliary beating. The five other peptidergic cell types do not show coexpression of genes involved in synaptic and neuronal functions indicating the absence of a synaptic scaffold or any other neuronal gene module.

Cnidaria

Cnidarians, the group composed of anemones, corals, and sea jellies, are united by a stinging cell type called the cnidocyte. While cnidarians were thought to have a limited number of cell types, a whole-body cell atlas generated for the cnidarian Nematostella vectensis has revealed far more complexity, including homologous relationships between some neuron types in metazoans (Sebé-Pedrós et al. 2018b). Furthermore, sampling eleven developmental timepoints uncovered the shared developmental origin of neurons, stinging cells, and gland cells from a common multipotent progenitor population (Steger et al. 2022). Additionally in N. vectensis, single-cell data are beginning to be integrated with known spatial information; a recent profile of endomesoderm with scRNA-seq was used to construct a three dimensional spatial gene expression atlas (He et al. 2023). Researchers uncovered a molecular code for segment polarity in N. vectensis and suggest polarized structures existed in the Cnidarian-Bilaterian common ancestor.

The cnidarian Hydra has continual cellular turnover, thus a single timepoint of sampling captured a broad set of differentiation trajectories (Siebert et al. 2019). In one of the first whole organism atlases, the researchers identified key molecular actors which specify i-cells, a population of multipotent progenitors, which differentiate into the many additional cell types in Hydra. Additionally, the authors were able to identify similarities between neurogenesis, gland cell differentiation, and nematocyte formation, which they suggest may point to a shared or similar progenitor state (Siebert et al. 2019). Since then, additional scRNA-seq data has been generated and mapped to a new genome reference assembly for this specific strain of Hydra (Cazet et al. 2022).

The relationship between neurons and nematocytes has been furthered by studies in the medusozoan Clytia hemisphaerica, where scRNA-seq suggested there were clear connections between i-cell, neuronal, and nematocyte populations, but not any connection to gland cells (Chari et al. 2021). Additionally, Chari et al. (2021) was a proof-of-principle perturbation scRNA-seq study in a nontraditional model organism; the authors generated cell type atlases for fed and starved jellyfish and found that oocyte and digestive cell types displayed distinct transcriptional responses.

Finally, scRNA-seq has been used in cnidarians to understand the cellular interactions of animal hosts and their symbiotic algal partners. In the soft coral Xenia sp. both scRNA-seq and bulk RNA-seq on FACS sorted alga-containing and alga-free cells were used to identify an endosymbiotic cell type. This may help to understand how coral take up or lose their endosymbionts, a key relationship in the health of the coral (Hu et al. 2020). Additionally in the stony coral, Stylophora pistillata, multiple stages, including larvae, primary polyps, and adults were sampled with scRNA-seq and uncovered a population of alga-hosting coral cells (Levy et al. 2021). scRNA-seq studies in cnidarians have thus been used to broaden our understanding of cell type diversity in these organisms, uncover novel information about developmental trajectories of cell types, and generate hypotheses about key aspects of their life histories.

Xenacoelomorpha

The group Xenacoelomorpha, composed of Xenoturbellida and Acoelomorpha, are relatively simple worm-like creatures that lack a through-gut and are proposed to be either sister to bilaterians or ambulacrarians (Cannon et al. 2016; Kapli et al. 2021). In xenoturbellids, the cell type repertoire of Xenoturbella bocki suggested that cells of the nerve net share regulatory features with cnidarians and protostomes (Robertson et al. 2022). In the acoel Isodiametra pulchra, Duruz et al. focused on cell type composition and identified ten major cell types that correspond to well-characterized bilaterian cell types, but also noted a number of clade-specific marker genes (Duruz et al. 2021). In another acoel, Hofstenia miamia, various timepoints were sampled to consider the underlying regulatory pathways in tissue formation during regeneration (Hulett et al. 2022). By identifying cellular level responses, analysis showed that both differentiated and stem cells exhibit changes during regeneration, which suggests a complex response to injury. In H. miamia development, scRNA-seq sampling of various timepoints was combined with lineage inference tools to predict putative differentiation paths of major cell populations in H. miamia (Kimura et al. 2022).

Platyhelminthes

Within protostomes are platyhelminths, a phylum of unsegmented, soft-bodied free living and parasitic flatworms. While most scRNA-seq studies in platyhelminths have been performed in freshwater planarians (Fincher et al. 2018; Plass et al. 2018) a recent preprint considered the cell type composition of the Müller’s larva of the marine flatworm Prostheceraeus crozieri (Piovani et al. 2023). Researchers sampled pelagic larvae in two different phyla and identified homologous and clade-specific cell types based on scRNA-seq by finding orthologous genes in clusters such as the ciliary band.

Mollusca

The phylum Mollusca includes gastropods, cephalopods, and bivalves. As part of their cell type comparison study, Piovani et al. (2023) also considered the trochophore larva of the pacific oyster Crassostrea gigas, and recovered cell types, such as the shell gland, that express novel genes. In cephalopods, scRNA-seq was used in the optic lobe of Octopus bimaculoides to reveal six major neuronal cell classes (Songco-Casey et al. 2022), and in the visual and nervous system of Euprymna berryi to identify major cell types and describe the parallel evolution of camera-type eyes of cephalopods and vertebrates (Gavriouchkina et al. 2022). Additionally in cephalopods, scRNA-seq in the paralarval brain of Octopus vulgaris recovered a set of genes used in glial cells in octopus, fly, and mouse brains, suggesting an ancestral urbilaterian glial cell type (Styfhals et al. 2022). In the head of the squid, Loligo vulgaris, Duruz et al. (2023) characterized cell types present in other phyla such as neurons and muscles, as well as cephalopod specific cell types, such as chromatophores and sucker cells. Another mollusc with scRNA-seq data includes the scallop, where Sun et al. (2021) compared molecular profiles of muscle types finding a more diverse population of neurons in striated versus smooth muscle.

Annelida

Annelids are a segmented worm phylum that include Platynereis dumerilii and Capitella teleta, two marine species profiled with scRNA-seq. In the larval stage of P. dumerilii, scRNA-seq identified five distinct groups of cells with expression domains that the authors link to subdivisions of the annelid body (Achim et al. 2018). In another study, authors created a computational method for spatial transcriptomics by combining in situ based gene expression and scRNA-seq data in the larval brain of P. dumerilii to identify the spatial origin of each cell (Achim et al. 2015). In a scRNA-seq study of Capitella teleta development, the authors focused on neurogenesis and used pseudotime analysis tools to identify two potentially distinct differentiation trajectories for the neuronal specification (Sur and Meyer 2021).

Arthropoda

The arthropods are united by an exoskeleton, paired appendages, and a segmented body. In marine arthropods, scRNA-seq has been used to profile hemocytes in the shrimp Marsupenaeus japonicus to begin to understand the role of the immune system in this commercially important species (Koiwai et al. 2021). In the crustacean regeneration model Parhyale hawaiensis, single nucleus RNA sequencing was used to demonstrate that all expected cell types are recovered in regenerating legs (Almazán et al. 2022). While previous work comparing regeneration to development in Parhyale demonstrated differences in the timing of gene expression (Sinigaglia et al. 2022), the faithful repopulation of similar cell types in regeneration uncovered by scRNA-seq suggests that there are distinct regulatory programs that lead to conserved appendage formation.

Echinodermata

Echinoderms, an entirely marine phylum characterized by radial symmetry, include sea urchins, sea stars, and sea cucumbers. Within echinoderms specific methods for scRNA-seq have been published for ease and standardization (Oulhen et al. 2019).

In the sea star Patiria miniata scRNA-seq was used to generate a cell type atlas from six developmental stages, from  eight  hours post fertilization to mid-gastrula stage. By analyzing expression patterns of the germ cell markers nanos and vasa, Foster et al. (2022) found that members of the Nodal pathway, which signals to restrict the germ cell region, are never coexpressed with nanos/vasa-positive cells that give rise to primordial germ cells (Foster et al. 2022). In Patiria pectinifera, authors generated a scRNA-seq cell atlas and compared expression patterns to chordates to discuss hypotheses on deuterostome body plan evolution (Tominaga et al. 2023). As an expansion of Garstang’s 1928 theory of dipleurula ciliary band evolution (Garstang 1928), Tominaga et al (2023) suggest the presence of an oral ectodermal region in sea star larvae that corresponds to floor plate in chordates and that the ciliary band instead migrated to the oral side.

In the sea urchin Lytechinus variegatus, Massri et al. (2021) densely sampled timepoints with scRNA-seq and computationally traced lineage trajectories finding support for an early lineage divergence in skeletogenic cells, which confirmed existing experimental data (Oliveri et al. 2002, 2003; Revilla-i-Domingo et al. 2007). Massri et al. (2021) also found transcriptional evidence that the endomesodermal lineage divergence is asynchronous, as some endomesodermal cells retain both cell fate markers for an extended period, while others express only one cell fate marker (Massri et al. 2021). To compare a lecithotrophic species, Heliocidaris erythrogramma, to the planktotroph L. variegatus, Davidson et al. (2022) used scRNA-seq in the early blastula of H. erythrogramma. Researchers found that the lecithotrophic, derived life history state has fewer cell types present and less localized expression of known marker genes at an equivalent stage, suggesting a delay in fate specification (Davidson et al. 2022).

In another sea urchin species, Strongylocentrotus purpuratus, Foster et al. (2019) used scRNA-seq to test cell type specificity with Wnt and Delta-Notch pathway inhibitors. Researchers observed distinct cell type shifts when using an inhibitor in the Wnt pathway versus the Delta-Notch pathway, demonstrating specificity in cell type lineages in these pathways (Foster et al. 2019). Researchers also created a resource of eight scRNA-seq timepoints, from 8-cell stage to late gastrulae, which demonstrated 22 major cell clusters (Foster et al. 2020). The same dataset was reanalyzed by Satoh et al, (2022) and confirmed that the Brachyury expressing cells coexpress ventral-organizer genes and that some of these cells invaginate to form the stomodeum (Satoh et al. 2022). Perillo et al. (2020) used scRNA-seq to determine the driver of pigment cell number in S. purpuratus by knocking down gcm and observing that the pigment cell cluster was reduced in cell number in the gcm knockdown (Perillo et al. 2020). Spurrell et al. (2023) investigated differences in pigment cell specific genes in S. purpuratus and P. miniata by comparing the available scRNA-seq atlases and finding an ancestral regulatory program for larval pigment cells in both species even though P. miniata does not have this cell population (Spurrell et al. 2023). A separate study used the existing S. purpuratus scRNA-seq data, generated new single nucleus RNA-seq data for P. miniata, and used an orthology approach to compare cell types between the two species (Meyer et al. 2022).

Another scRNA-seq study in S. purpuratus 3-day-old pluteus larvae found many neuronal cell states coexpressing two transcription factors, Pdx-1 and Brn1/2/4, which are also found in vertebrate pancreatic cells (Paganos et al. 2021). The authors suggest that due to this conserved gene regulatory network and coexpression in the scRNA-seq data, the neuronal and pancreatic cell lineages in sea urchins and vertebrates diverged from a common ancestral cell type present in stem deuterostomes (Paganos et al. 2021). In a further analysis of these data, researchers found that exocrine-like and endocrine-like pancreatic cells in sea urchin larvae use orthologs of transcription factors that are known pancreatic gene markers in mammals (Paganos et al. 2022a).

Finally, in the sea urchin Paracentrotus lividus, researchers manually dissected out the mature rudiment for scRNA-seq. Paganos et al. (2022b) specifically investigated cells expressing retinal gene orthologs required for photoreception and found two neural clusters that expressed opsins. These clusters were transcriptionally different, as they only shared thirty-seven marker genes, suggesting a divergent developmental path and potentially different functions of these cell types (Paganos et al. 2022b).

Tunicata

Tunicates are marine filter feeders, made of a water-like sac, two siphons, and an outer covering or “tunic.” Sharma et al. (2019) profiled the larval brain in Ciona intestinalis using scRNA-seq and identified ten cell types (Sharma et al. 2019). Two of these cell types, the palp cells (Horie et al. 2018a) and the coronet cells (Horie et al. 2018b), were investigated in more detail with knockdown experiments prior to scRNA-seq. Morpholino knockdown of the most strongly expressed cell-specific transcription factor in coronet cells results in their loss, while coexpression of this transcription factor and another highly expressed gene cause the entire central nervous system to transform into coronet cells (Horie et al. 2018b). Using scRNA-seq, Ilsley et al. (2020) recovered and validated a new pattern of single-cell specific gene expression in the two most vegetal, posterior cells at the 16-cell stage (Ilsley et al. 2020). Treen et al. (2018) manually dissociated blastomeres in C. intestinalis to investigate regulators of the maternal to zygotic transition and found a dramatic decline in Cyclin B3 with the onset of zygotic genome activation. Wang et al. (2019) used FACS sorted cardiopharyngeal cells of C. intestinalis to reconstruct the developmental trajectories in two heart lineages with scRNA-seq. Cao et al. (2019) used scRNA-seq in C. intestinalis to reveal fourteen cell types and their developmental trajectories from the 110-cell to larva stage. Additionally, Cao et al. (2019) identified the palp sensory cells, a previously described but rare cell type, showing that scRNA-seq can help identify cryptic, molecularly distinct cell types (Cao et al. 2019).

In Ciona savignyi, Zhang et al. (2020)investigated how gene expression underlies lineage differentiation by analyzing the epidermal lineage. Researchers reported that the “a” and “b” line epidermal cells remain as one cell type at the 64-cell stage, but diverge into two distinct cell types by the 110-cell stage Zhang et al. (2020). In another tunicate, Phallusia mammillata, scRNA-seq data were integrated with light sheet imaging to combine physical position, lineage history, and RNA velocity to create a spatiotemporally resolved atlas of gene expression (Sladitschek et al. 2020).

Cephalochordata

Cephalochordates, commonly called lancelets or amphioxus, are small, segmented, fishlike marine invertebrate chordates with elongated bodies containing a notochord, dorsal nerve cord, endostyle, post anal tail, and pharyngeal slits. In Branchiostoma floridae, single nucleus RNA-seq was generated for nine embryological stages, plus the adult stage, and then integrated with single cell Assay for Transposase-Accessible Chromatin sequencing (ATAC-seq) data at six of these stages to create a cell lineage tree from blastula to larval stages (Ma et al. 2022). Satoh et al. (2021) generated a scRNA-seq atlas in six embryonic stages of Branchiostoma japonicum with a focus on distinguishing the expression profiles of the two brachyury genes in amphioxus. scRNA-seq data supports previous data (Tominaga et al. 2018; Yuan et al. 2020) that Bra2 is the ancestral Brachyury as it is expressed in the blastopore like other deuterostomes, and Bra1 is a duplicate copy (Satoh et al. 2021). To investigate the conservation of maternal patterning during chordate evolution through asymmetrically inherited maternal transcripts in B. floridae, Lin et al. (2020) separated the 2-cell embryo into blastomeres and the 8-cell embryo into animal and vegetal tiers and sequenced each. Maternal transcripts were classified either as vegetal tier/germ granule enriched or animal tier/anterior enriched.

Challenges and opportunities in drawing comparisons in scRNA-seq data

One of the outstanding challenges in the future of single-cell biology, and one i.e. not limited to marine invertebrates, is the comparison of data from different experiments. In general, scRNA-seq studies involving multiple samples may be confounded by batch effects resulting from multiple distinct library preparations for different timepoints in an experiment. Whether it be across different stages of development or regeneration, it will be important to ensure that the differences in cell composition are due to real biological variability. Computational techniques such as Seurat (Butler et al. 2018), Harmony (Korsunsky et al. 2019), LIGER (Welch et al. 2019), Scanorama (Hie et al. 2019), and mnnCorrect/fastMNN (Haghverdi et al. 2018), have been developed to solve this problem by integrating scRNA-seq datasets across experiments (Shafer 2019).

An additional challenge when using scRNA-seq datasets across species is how to use these cell atlases to interrogate evolutionary and developmental relationships. To make unbiased, rigorous comparisons between species, several authors have put forth universal standards and tutorials to make technically compatible cell atlases of similar resolution (Luecken and Theis 2019; Oulhen et al. 2019; Andrews et al. 2021; Kharchenko 2021). Hurdles here include normalizing for cell size and RNA quantity differences, differences in transcriptional activity between species, and sequencing depth. Another set of barriers in comparing cell atlases include high-quality reference genomes, high-quality gene annotations, and the ability to assign gene orthology. This can be difficult across multiple species or phyla, as it can be more difficult to infer orthologous genes with greater phylogenetic distance (Tanay and Sebé-Pedrós 2021). Another computational tool, SAMap was designed for cross-species manifold alignment by considering non-one-to-one orthology (Tarashansky et al. 2021).

An important factor in comparing cell types is the need for dense sampling across animal groups. The processes of isolating single cells with minimal disruption of the normal expression profile can be challenging when adapting commercial mammalian optimized instrumentation to osmotically different marine invertebrates. To this end, there are ongoing efforts to optimize isolation techniques that minimize the possible variance introduced by isolation, including recent advances such as a dissociation technique that simultaneously fixes cells and a microfluidics free method for encapsulation and barcoding (García-Castro et al. 2021; Clark et al. 2023). Additionally, certain systems can pose unique challenges, e.g. the tough cuticle of crustaceans defies many current methods used to produce the single cell suspensions required for input into scRNA-seq technologies, but using frozen nuclei as input into single nucleus sequencing is a viable alternative (Almazán et al. 2022).

Finally, there is the challenge of comparing equivalent life history stages. In comparing scRNA-seq data from C. hemisphaerica and Hydra, gland cells are thought to originate from i-cells in Hydra while there are not clear connections between i-cells and gland cells in C. hemisphaerica. Future data could compare the Hydra data to a more similar stage in C. hemisphaerica development like the polyp stage.

Beyond single-cell RNA sequencing

While scRNA-seq allows for a view of the transcriptomic identities and differences between cell types, combining these data with other ‘omics methods such as genomic, epigenomic, and proteomic data could allow scientists to describe the complexities of cells more deeply. One method, ATAC-seq (Buenrostro et al. 2015), has been used in some marine invertebrates to assay the chromatin landscape of whole organisms or specific cell types. For example, scRNA-seq studies in Hydra (Siebert et al. 2019), N. vectensis (Sebé-Pedrós et al. 2018b), H. erythrogramma (Davidson et al. 2022), and B. floridae (Ma et al. 2022) have highlighted the benefit of including ATAC-seq data. While, these studies collected ATAC-seq and scRNA-seq libraries separately, there are new technologies, which simultaneously collect these data such as 10X Genomics Multiome® (Hao et al. 2021), SHARE-seq (Ma et al. 2020), and HyDrop (De Rop et al. 2022). In 10× Multiome some challenges include preserving both high quality RNA while still achieving sufficient tagmentation for ATAC-seq while SHARE-seq requires very high numbers for initial input (Lee et al. 2020).

Furthermore, gene expression changes in individual cells in a spatial context are crucial to understanding cells within tissues and whole organisms. Techniques in spatial transcriptomics include targeted, microscopy based approaches such as multiplexed error-robust FISH (MERFISH) (Chen et al. 2015; Moffitt et al. 2016) and sequential FISH (seqFISH) (Lubeck et al. 2014), and untargeted, sequencing approaches such as spatial transcriptomics/10X Genomics Visium (Ståhl et al. 2016), Slide-seq (Rodriques et al. 2019), and Stereo-seq (Chen et al. 2022). These methods have been applied in model organisms and will need to be attempted and likely optimized for marine invertebrates.

Conclusion

scRNA-seq has provided unprecedented resolution in understanding cell type composition across animal groups, dynamics in processes such as regeneration and development, and generating new hypotheses about the evolution of cell types. Sampling individual transcriptomes in dynamic biological processes will allow scientists to not only uncover what genes are differentially expressed, but how trajectories themselves differ from each other. With these great advances, we are now able to compare cell types across samples and distantly related animal groups, but there are many obstacles ahead. We see these as areas of growth; whether technical or biological, and these problems represent intellectual hurdles to overcome that lead to new insights. Initial observations in marine invertebrates with scRNA-seq have provided compelling evidence that previously undescribed differences exist between and within cellular populations and illuminated numerous starting points for future fruitful investigations.

Notes

From the symposium “Genomics of Marine Larval Evolution and Development” presented at the annual meeting of the Society for Integrative and Comparative Biology, January 3–7, 2023 in Austin, Texas.

Funding

This work was supported by funding from National Institutes of Health [R35GM128817 for P.B.] and [T32GM007276 to L.L.]

Conflict of interest

There are no conflicts of interest.