Comparative transcriptomic analysis reveals gene regulation mediated by caspase activity in a chordate organism

Background Apoptosis is a caspase regulated cell death present in all metazoans defined by a conserved set of morphological features. A well-described function of apoptosis is the removal of excessive cells during development and homeostasis. Recent studies have shown an unexpected signalling property of apoptotic cells, affecting cell fate and/or behaviour of neighbouring cells. In contrast to the apoptotic function of cell elimination, this new role of apoptosis is not well understood but seems caspase-dependent. To deepen our understanding of apoptotic functions, it is necessary to work on a biological model with a predictable apoptosis pattern affecting cell fate and/or behaviour. The tunicate Ciona intestinalis has a bi-phasic life cycle with swimming larvae which undergo metamorphosis after settlement. Previously, we have shown that the tail regression step during metamorphosis, characterized by a predictable polarized apoptotic wave, ensures elimination of most tail cells and controls primordial germ cells survival and migration. Results We performed differential transcriptomic analysis between control metamorphosing larvae and larvae treated with the pan-caspase inhibitor Z-VAD-fmk in order to explore the transcriptional control of apoptotic cells on neighbouring cells that survive and migrate. When caspase activity was impaired, genes known to be involved in metamorphosis were downregulated along with other implicated in cell migration and survival molecular pathways. Conclusion We propose these results as a confirmation that apoptotic cells can control surrounding cells fate and as a reference database to explore novel apoptotic functions in animals, including those related to migration and differentiation. Supplementary Information The online version contains supplementary material available at 10.1186/s12860-021-00388-0.


Background
Apoptosis is a regulated cell death defined by morphological features and depending on caspases [1][2][3]. Being ubiquitous in metazoans [4][5][6][7], the main function of apoptosis is the elimination of unwanted cells [8], such as the interdigital tissues during embryogenesis of tetrapods [9] or larval tissues during metamorphosis of many animals [10]. Apoptosis is also implicated in homoeostasis by fine-tuning the balance between cell death and proliferation like in gonad size control in rodents [11].
While the function of cell elimination by apoptosis is highly documented, apoptosis seems to have a broader morphogenetic function that goes beyond cells elimination. Accumulating evidence in many metazoans suggest that apoptotic cells can emit caspase-dependent signals to their neighbours, modulating their fate (survival/death, differentiation or proliferation) or behaviour (migration) [12,13]. Apoptotic-induced proliferation was reported in several organisms, such as in the cnidarian Hydra, where apoptosis promotes proliferation of adjacent cells through emission of Wnt3 signalling in a caspasedependant manner during head regeneration [14]. Similar observations have been made during rodent liver regeneration, where cleavage of iPLA2 by caspases in apoptotic cells led to proliferation of adjacent cells [15]. Apoptotic cells can also induce differentiation, as was reported during the metamorphosis of the cnidarian Hydractinia echinata [16]. Finally, apoptotic cells modulate death or survival balance, as was observed in the Drosophila imaginal disc, by producing the Tie ligand Pvf1, that induces death resistance of neighbouring cells [17]. Apoptoticdependent migration is well described in mammals, in which apoptotic cells secrete directional signals such as Lysophophatidylcholine or Sphingosine-1-Phosphate, promoting migration of leukocytes [18,19]. Furthermore, in Xenopus laevis tadpole, an apoptotic-dependent axon guidance was suggested during tail regeneration after injury [20].
While caspase-dependent signals emitted by apoptotic cells start to be characterised, almost no target genes are identified in receptor cells. Evaluating the capacity of caspases to affect gene expression and influence fate and/or behaviour of cells adjacent to apoptosis is the first fundamental step to better characterise this new function in animals.
In this report, we took advantage of the predictable spatiotemporal apoptotic profile of ascidians [21,22], in particular during the Ciona intestinalis metamorphosis, which is concomitant with cell survival and cell migration [21,23,24]. Ciona belongs to the Urochordata, the sister group of vertebrates [25], and has a biphasic life cycle composed of a pelagic larva and a benthic adult. Embryogenesis generates a tadpole-like swimming larva consisting of an anterior trunk and a posterior tail which, after few hours of swimming, settles and metamorphoses into a sessile juvenile. A hallmark of metamorphosis is the apoptotic-dependent tail regression [21] during which apoptosis starts at the tip of the tail and propagates toward the trunk, making it easily anticipated; a unique feature among chordates. Using TUNEL labelling and electron microscopy, we previously described in detail this apoptotic profile. Apoptosis starts in epidermal cells at the most posterior tail tip, then, it affects the notochord cells and the striated muscle cells in a postero-anterior wave. The tunic cells are also randomly affected. On the opposite, two tissues survive and migrate during, or prior to, tail regression; the ventral endodermal strand and the eight primordial germ cells (PGC) [23,24,26]. PGC, located at the postero-ventral side of the tail tip, escape from cell death by moving toward the trunk, and are consequently surrounded by apoptotic cells during the tail regression. Interestingly, we previously demonstrated that the pan-caspase inhibitor Z-VAD-Fmk blocks both apoptotic dependent-tail regression and PGC migration, suggesting that PGC migration is regulated by apoptotic cells [23].
To identify genes controlled by caspases activity during tail regression, we performed a comparative transcriptomic analysis at the beginning and at mid-tail regression between control metamorphosing larvae or larvae exposed to the pan-caspase inhibitor Z-VAD-fmk, already successfully used in Ciona [21] (Fig. S1). Based on a de novo transcriptome, we identified 61 genes differentially expressed between the beginning and mid-tail regression in control larvae. Next, we found 65 genes differentially expressed between control and treated larvae (21 and 44 at the beginning and at mid-tail regression, respectively). Interestingly, misregulation of some genes upon caspase inhibition was reported during ascidian metamorphosis, but also reported in cell survival and cell migration in several animals [27][28][29][30]. Our results offer a first database of genes transcriptionally modulated by apoptosis during a dynamic morphogenetic process in a chordate, allowing exploration of novel apoptotic functions in animals.

Results
Genes differentially express between the onset and the mid-tail regression Globally, the robustness of our analysis is shown by a heat map allowed visualization of result congruence between replicates (Fig. 1). To confirm relevance of our data, we randomly chose 6 genes differentially expressed in our transcriptomic analysis to performed real-time PCR ( Fig. 2A). Five of them exhibited an expression profile that was similar for both experiments.
One of them, KH.C7.314, is less expressed at the midtail regression in comparison to the beginning of the process in control larvae in our transcriptomic analysis. Similar profile was observed with real-time PCR, but the difference is less pronounced and not statistically significant. Despite this, control by real-time PCR globally validated our transcriptomic data.
First, we have focused on genes differentially expressed in control metamorphosing larvae between the beginning and the mid-tail regression. Twenty-five genes (41%) were up-regulated from the beginning to the midtail regression with the strongest increase recorded for KH.C9.385 (smyd3) ( Table 1). Conversely, 36 genes (59%) show decreased expression, with a subset of genes whose expression was no longer detectable such as KH.14.261 (lysozyme-like) ( Table 2). Among these identified genes, several were already reported during a previous cDNA chip analysis of Ciona metamorphosis, such as KH.C7.314 (MED30), KH.C11.274 (p-selectin, known as Ci-sushi) which is expressed in almost all the tail, or KH.L132.4 (matrilin-2) expressed in epidermal cells at the tip of the tail in premetamorphic larva (Fig. 2B) [29].
In addition, some of the genes identified by our screen had already been detected by substantial RNA sequencing data coupled with in situ hybridization data available on ANISEED database (http://www.aniseed.cnrs.fr). For example, KH.L132.16 (fibrillin-3) starts to be expressed at mid-neurulae with a maximum expression in larva, in which it is detected in all the epidermis including the tail. Furhermore, KH.C10.250 (fox-1, RNAbinding) is expressed in tail muscle at mid-tailbud stage to become ubiquitous in larva [31]. We also observed that some among these were also impacted by Z-VAD-fmk treatment (Tables 3, 4 [27][28][29][30]. The detection of genes already known to be expressed in the tail and involved in tail regression (i.e. Ci-sushi, Ci-meta2) is consistent with previous studies on ascidians metamorphosis, validates our experimental procedures and supports the relevance of our database.

Caspase inhibition affects gene expression during the tail regression
In total, 65 genes were affected by the inhibition of caspases during Ciona tail regression ( Fig. 1). At the beginning of tail regression 12 genes were upregulated by the Z-VAD-fmk treatment (so negatively regulated by caspase activity in physiological conditions) with the greatest expression increase observed for KH.C11.506 (girdin) and KH.S1380.1 (mucin-5 AC) ( Table 3), and 9 genes were downregulated (so positively regulated by caspase activity) with extreme effects on KH.C1.66, which is no longer expressed (Table 4).

Transcription is necessary for the PGCs migration
We previously demonstrated that Z-VAD-fmk blocked both tail regression and PGC migration, and that apoptotic wave propagation and PGC migration speed were correlated [23]. Here, we show that transcription is necessary for PGC survival and migration (Fig. 3). We exposed metamorphosing larvae to the transcription inhibitor actinomycin-D; in the control (devoid of actinomycin-D), a classical apoptotic profile is observed with PGC movement, as already described [23], and in actinomycin-D-treated larvae PGC migration is blocked and PGC nuclei become TUNEL-positive, indicating (See figure on previous page.) Fig. 1 A, Counting of genes differentially expressed between conditions. B, Heatmaps showing genes differentially expressed in control between the beginning of tail regression (TR) and the mid-regression. C, Heatmaps showing genes diffenrentially expressed at the beginning of tail regression between control and Z-VAD-fmk treated larvae. D, Heatmaps showing genes diffenrentially expressed at mid-regression between control and Z-VAD-fmk treated larvae destruction as the apoptotic wave progresses (Fig. 3). This result, and our previous studies, argue that survival and migration of PGC depend on transcription that could be partially controlled by caspase activity from apoptotic cells (Fig. 3). The comparative transcriptomic analysis between control and Z-VAD-fmk treated larvae is fundamental to understand this apoptotic dependent migration.

Discussion
Caspase activity modulates expression of genes implicated in cell death/survival Interestingly, the 2 genes most inhibited by caspase activity at the beginning of tail regression, girdin and mucin-5 AC, were both reported to promote survival in mammalian cells [32,33]. Among the other genes negatively impacted by caspase activity, KH.C1.1084, the thioredoxin homologue in mammals, is able to control the activation of the executioner caspase 3, the initiator caspase 9, and the release of the cytochrome c, all characterising the intrinsic apoptosis [34][35][36][37]. In addition, KH.C3.45 (kinase C alpha) homologue is known to promote survival and increases anti-apoptotic Bcl-2 efficiency in cultured cells [38]. Taken together, our database highlights that cells expressed a variety of genes regulating cell death/survival fate in a caspase dependent manner, suggesting that apoptotic cells actively participate in apoptosis propagation during tail regression.

Caspase activity modulates expression of genes implicated in cell migration
We identified numerous genes modulated by caspase activity, with a majority of them upregulated, previously observed in cell migration regulation. It is the case of KH.C1.216 (myosin light chain 3) and KH.C9.494 (myosin light chain kinase) for which homologues are key players in mesenchymal migration [32,39] and KH.C13.153 (uromodulin-like) which provides guidance cues for neutrophile migration [40]. Furthermore,  integrin subunits and components of the Extra Cellular Matrix, crucial for mesenchymal migration, are well represented in our transcriptomic approach. In addition, KH.C1.10 (lasp1) mammalian homolog is necessary for migration of NIH 3 T3 cells [41,42]. Furthermore, KH.C12.516 (ceramidase) could be involved in tissue migration regulation, as it has been described as participating in the sphingolipid pathways. Interestingly, sphingosine 1-phosphate signalling leads to tissue migration [43], particularly in germline cells (due to to a sphingosine 1-phosphate receptor) in the tunicate Botryllus [44], making this pathway a strong candidate for PGC migration in Ciona. Given that these genes and PGC migration are both controlled by caspases [23], PGC migration can be regulated by apoptotic cells.

Ascidians as crucial biological models to studies novel apoptotic function
This study, coupled with our previous work suggested that apoptosis regulates the migration of primordial germ cells [23], confirm Ciona as a perfect biological model to understand apoptotic-induced migration. In addition, a regulation of cell proliferation by apoptosis was reported in this species [45]. Ascidians are well known to have an impressive regenerative capacity for a chordate, in particular in the context of the reconstruction of the distal body parts after amputation, thanks to stem cell requisition from the branchial sac [46][47][48]. Apoptosis was reported to occur as a transient event at the site of injury, and precedes the regeneration of the neural complex, the oral siphon, or the atrial siphon [49]. In the case of siphon amputation, it has been shown by caspases inhibition experiment that an apoptotic-dependant Wnt signalling allows the cell proliferation, leading to the successful regeneration [45]. In addition, apoptosis appears crucial to maintain the growth of branchial sac and its inhibition lead to a hazardous organization of this tissue, suggesting a fundamental function in homeostasis maintaining. Importantly, stem cells requisition and migration from the branchial sac toward the wounding site seem also dependant on caspases, making it an other potential   apoptotic-dependant migration [45]. Taken together, all of these studies make Ciona one of the rare biological models where several cell fates and behaviours, controlled by apoptotic cells in a caspases dependent manner, has been put in evidence or suggested in adults as well as in larvae.
In the colonial ascidian Botryllus schlosseri, apoptosis is known to play on budlets elimination [50][51][52]. In this species, blastogenetic cycle (asexual reproduction of adults) give rise to a bud, itself given next four budlets (reviewed in [53,54]). Usually, two of them growth despite the two other undergo apoptosis and are reabsorbed. In addition to this function in the blastogenetic cycle, apoptosis was also reported in tunic, epidermis, branchial leaflet, vessel epithelium, circulatory system or gonads in adults. With so many tissues presenting apoptosis in physiological condition, it seems clear that apoptosis should have crucial functions in Botryllus and likely could participate to regulation of other cell fates and behaviours as it was reported in Ciona. All of these tissues are as many opportunities to deepen or understanding of apoptosis functions in a same species.

Conclusions
Here, we offer a potential roadmap in order to design future functional studies on Ciona and other animals, including mammals, where apoptosis takes place together with other cell fates and behaviours. Identification of molecular actors needing cleavage for their activation, cell receptors receiving signals from apoptotic cells, or  The last decade has brought evidence that the cleavage capacity of caspases is a prerequisite to activate molecular actors promoting migration or differentiation in a non-apoptotic context [55][56][57][58][59]. Consequently, Z-VADfmk blocks all processes led by caspases, making it crucial to distinguish the level of implication of apoptotic and non-dying cells for future research.
Finally, our study provides an additional and novel chapter in the research of apoptotic functions beyond their role as a simple cell elimination mechanism, opening opportunities in the field of medical science [60].

Sample collection
Adult Ciona intestinalis were collected in the field in Brittany (France) by the Centre National Marine Biological Resources, EMBRC-France, Roscoff (Station Biologique de Roscoff, Sorbonne Université, Brittany, France). Individuals were maintained in 35‰ salinity artificial seawater (ASW) at 18°C in the UMR7138 Evolution Paris Seine laboratory (Sorbonne University, Paris, France). Oocytes and sperm were obtained by dissection and cross-fertilizations were performed in plastic Petri dishes. Embryos were cultured at 18°C in 0.2 μm filtered ASW with 0.1 M Hepes (#H4034, Sigma-Aldrich; Merck KGaA). Embryos of each fertilisation were divided into two subsets before the first cell division to obtain one control and one treatment condition. Six fertilisations were performed, divided into twelve petri dishes (6 control with their respective 6 treatment conditions). After the treatment, larvae were collected at two time points (see below), and compacted by manual centrifugation before RNA extraction.

Pharmacological treatments and TUNEL labelling
The pan-caspase inhibitor Z-VAD-Fmk (#V116, Sigma-Aldrich; Merck KGaA) was stored in DMSO (#D8418, Sigma-Aldrich; Merck KGaA) at − 20°C and used at final concentration of 10 μM. After the settlement of more than 50% of larvae on the dish, and when some of them stopped moving (indicating that metamorphosis will begin soon), supernatant was discarded and replaced with filtered ASW containing DMSO as control, or Z-VAD-Fmk diluted in DSMO as treatment. Z-VAD-Fmk was renewed every two hours. The experiment was divided into two time points (with three biological replicates for each oneitself divided between control and treated larva) following control tail regression stages: first when the tail regression just began in the control (beginning), and the second at mid-progression (mid-tail regression) given three controls with three Z-VAD-Fmk equivalents for each point.
In a separate experiment, 1 μg/ml of actinomycin-D (#A1410, Sigma-Aldrich; Merck KGaA) was added to swimming larvae. After settlement and tail regression initiation, paraformaldehyde fixation, TUNEL labelling and Vasa immunostaining were performed as previously described [23].

RNA preparation and sequencing
RNA was extracted using the RNAqueous™-Micro Total RNA Isolation Kit (#AM1931 -ThermoFisher Scientific) according to the manufacturer's protocol. DNAse treatment was performed using the TURBO DNA-free™ Kit (#AM1907 -ThermoFisher Scientific). RNA was purified and concentrated using the RNeasy MinElute Cleanup Kit (#74204 -Qiagen) according to the manufacturer's protocol. RNA quantity was measured using a Nano-Drop and quality was assessed with the Experion RNA HighSens (#700-7155 -BioRad) from 1 μL (0.1 μL of RNA completed with 0.9 μL of Milli-Q water) following standard instructions. Transcriptomes were sequenced on Illumina Hi-Seq 2000 (150 bp reads) at the Sistemas Genomicos plate-form of the Ascires Biomedical Group (Spain).

De novo transcriptome assembly and differential expression
Raw reads of the twelve Ciona intestinalis samples were checked with FastQC and merged to assemble a de novo transcriptome with Oases [61] and Trinity [62]. Best assemblies from each one (best N50) were merged with Cap3 to obtain the de novo transcriptome. De novo transcriptome contains 85,506 contigs (N50 = 1314, minimum contig length = 102, maximum contig length = 13,998). Then, raw reads of each sample were mapped against the de novo transcriptome with Bowtie2 [63] with standard parameters. Low quality mapping reads (MQ > 20) were removed using Samtools [64] and Picard Tools (http://broadinstitute.github.io/picard). Expression quantification was performed using HTSeq [65] and differential expression evaluated with the DESeq2 package from R. We kept the genes with an expression ratio -(control 1 + control 2 + control 3) / (treatment 1 + treatment 2 + treatment 3)above 2 (upregulated) and lower than 0.5 (downregulated). Then, Gene Ontology (GO) enrichment was performed with Blast2GO 5.2 and sequences were blasted on the Ciona genome to identify their KyotoGrail KH gene model.

In situ hybridization combined with vasa immunnodetection
Larvae were fixed 2 h at 4°C in MEM-PFA (4% paraformaldehyde, 0.1 M MOPS pH 7.4, 0.1 M NaCl, 1 mM EGTA, and 2 mM MgSO4, 0.05% Tween-20). After fixation larvae were washed three times and dehydrated through a graded series of ethanol/PBS baths and stocked in 100% ethanol at − 20°C. Rehydratation was made by successive washes of ethanol/PBS solution to a full PBS final solution. Hybridization was made according to a previous protocol [67] except that methanol was replaced by ethanol. After hybridization, immunnodetection of VASA was conducted as we previously described [23].
Additional file 1: Supplementary Fig. 1. Experimental design. Six independent fertilisations were performed, and each of them was divided into two batches, in twelve petri dishes. After settlement, supernatant was discarded and replaced with filtered ASW containing DMSO (control) or Z-VAD-Fmk diluted in DSMO (treatment). RNA sampling was done at two time points according to the control: immediately after observation of tail regression beginning, and at mid-tail regression.