Bipinnaria regeneration involves wound healing, body re-proportioning, cell proliferation and cell death
To make an informed comparison to other regenerative models, we first characterized the stages of larval regeneration in P. miniata. Bipinnaria larvae (7 days post-fertilization [dpf]) were bisected midway along the transverse anterior-posterior (AP) axis (Fig. 1b). Both resulting larval fragments were completely regenerative, restoring all lost tissues and organs over the course of 2 weeks. These findings are consistent with previous reports of larval sea star regeneration [42, 43]. Although we focus on the regeneration of the posterior fragments, a similar regenerative response is apparent within the anterior fragment (Additional file 1: Figure S1).
We observe that the initial wound is mostly closed by 3 h post-bisection (hpb; Fig. 2a, b, arrowheads). This also coincides with the appearance of several types of mesenchymal blastocoelar cells proximal to the wound epithelium. After this rapid wound healing response, larvae re-proportion their remaining tissues over the first several days post-bisection (dpb). This is evident when analyzing the position of the post-oral (lower) ciliary band (Fig. 2c). Prior to bisection, these ciliary bands are located in the middle of the larva; on average, the distance from the posterior end of the larva to the ciliary band is 47% of the total length of the larva (Fig. 2c). Immediately after bisection, this ratio increases to 80% as the anterior region has been removed (Additional file 1: Figure S2). However, over the subsequent 5 days, the larval proportions return to pre-bisection ratios (at 5 dpb, the ciliary band to larval length ratio is 57%). Importantly, this reallocation of tissues is not due to an increase in the total length of the larval fragments, as we show that the overall length of the bisected larva does not change during this time (Additional file 1: Figure S2). Although we did not quantify the change, we note a similar re-proportioning of the larval midgut between 1 and 5 dpb and also observed that the shape and position of the larval mouth changes. During bisection, the foregut is cut in half such that the anterior portion forms a new oral opening oriented along the anterior-posterior axis. However, by 3 dpb, the oral opening is reoriented ventrally and tissues are apparently anterior to this opening. Finally, by 6 dpb, we observe the return of most morphological features, including the anterior ciliary band, the oral field, and oral lobe. Together, these findings indicate that regeneration in larval sea stars occurs in at least three stages: healing at the wound site, re-proportioning the remaining tissues, and restoration of lost tissue. Similar patterns are evident in regenerating anterior fragments (Additional file 1: Figure S2).
We next analyzed the pattern of cellular proliferation during regeneration. Larvae were exposed to EdU (6 h pulses) to mark proliferating cells in normal (uncut) and over the course of larval regeneration (Fig. 3). In uncut larvae, EdU+ cells are widely distributed (Fig. 3a). We infer from this result that larvae are actively growing. However, upon bisection, the numbers of EdU+ cells steadily decrease (Fig. 3b; Mann-Whitney P < 2 × 10−4). This decrease in EdU+ cell number is accompanied by a change in the localization of proliferating cells. EdU+ cells localize proximally to the wound sites (3 dpb in posterior fragments and 6 dpb in anterior fragments), and fewer EdU+ cells are located in more distal tissues distal (Fig. 3c; Mann-Whitney P < 0.05). Moreover, the proliferating cells that localize to the wound site are distinct from cells that proliferate early. Cells proliferating at 1 dpb were labeled with pulse of BrdU followed by a wash-out. Cells proliferating during the later phases were then labeled with a pulse of EdU and processed for imaging. We find very little overlap of BrdU+ cells that are also EdU+ (Fig. 3d). This indicates that cells proliferating during early regeneration do not to continue to divide during the later, wound-proximal proliferation phase of regeneration. In non-bisected, stage equivalent control larvae, by contrast, there is extensive overlap between BrdU+ and EdU+ cells (Fig. 3d). This suggest that under normal conditions, cells that are proliferating normally continue to divide, but following bisection, different populations of cells now enter proliferation. Thus, during the regenerative response, typical, system-wide larval growth is inhibited, and regeneration-specific cell proliferation is concentrated at the regenerating edge where tissues later form.
As a corollary to understanding cell division during larval regeneration, we examined the patterns of cell death using TUNEL assays. In normal larvae, TUNEL+ cells are distributed organism wide (Fig. 4a). Following bisection, the number and distribution of apoptotic cells remains largely unchanged for several days (Fig. 4b–d and Additional file 1: Figure S3). However, at 6 dpb, there is a significant increase in the total number of TUNEL+ cells in both anterior and posterior regenerating fragments (Mann-Whitney P < 4 × 10−5). Unlike cell proliferation, these cells are not preferentially located with respect to the wound epithelium (Additional file 1: Figure S3B). Together, these results indicate that regeneration induces a global decrease in cell proliferation, followed by a rapid increase in cells cycling near the wound site. In contrast, the rate of cell death is consistent and increases across the larva coincident with the onset of wound-localized cellular proliferation.
These cellular and tissue changes during larval sea star regeneration define landmark features of the regenerative process including wound healing, re-proportioning of larval tissues, and onset of wound-proximal proliferation along with a coincident increase in apoptotic cell death. These broad characterizations mirror regenerative processes described in other organisms and suggest a shared toolkit of regenerative responses.
Transcriptome analyses of larval regeneration explain the genetic basis underlying observed cellular and morphological phenomena
To characterize the molecular events that operate during larval sea star regeneration and to establish a dataset amenable to inter-species comparison, we surveyed gene expression changes across a time course of larval regeneration. Pools of regenerating posterior fragments, anterior fragments, and non-bisected sibling control larvae were collected at three points following bisection: one early time point (approximately 3 hpb), one intermediate time point (3 days post-bisection, dpb), and one time point at the initiation of wound-localized cell proliferation (6 dpb). By separately sampling RNA from each pool of regenerating fragments, we were able to identify changes in gene expression changes that occur in both the anterior and posterior fragments as well as those that are specific to regeneration in each context. The inclusion of non-bisected, age-matched, sibling larvae control for transcriptional changes due to continuing larval development as well as genetic differences among cultures. For each time point, transcript levels were compared between each pool of regenerating fragments and the control larvae (i.e., anterior vs. uncut and posterior vs. uncut). In total, 9211 differentially expressed genes (DEG) were identified from these comparisons (Additional file 2: Table S1).
We implemented a hierarchical clustering approach to distinguish fragment-specific expression patterns from expression changes that are shared in both regenerating fragments (Fig. 5a and Additional file 1: Figure S4). In total, five expression clusters were identified: (I) genes upregulated early in both anterior and posterior fragments, (II) genes downregulated early in both fragments, (III) genes up in the anterior and down in the posterior, (IV) genes up in the posterior and down in the anterior, and (V) genes upregulated later (i.e., by 6 dpb) in both fragments (Fig. 5a). Thus, we have identified three subsets of DEGs that exhibit similar expression profiles during regeneration in both fragments (i.e., clusters I, II, and V) and two subsets that are strongly fragment-specific (i.e., clusters III and IV). To validate the RNA-Seq measurements, we analyzed the same samples using a custom Nanostring nCounter codeset. In total, 69 of the 74 genes (92.3%) tested by our Nanostring experiments exhibited either a similar trend and significance status or just a similar trend to the measurements made by RNA-Seq (Additional file 1: Figure S5).
To provide further insight into the functions of genes that were assigned to each cluster, we identified enriched Gene Ontology (GO) terms (Fig. 5b and Additional file 1: Figure S6). Genes in clusters I and II (i.e., genes that are up- or downregulated early in both regenerating fragments) are enriched for GO terms associated with a robust wound response. Upregulated genes (cluster I) are enriched for terms that include cell signaling pathways (e.g., “MAPK cascade” and “calcium channel activity”), “response to wounding,” and “immune system process” (Fig. 5b and Additional file 1: Figure S6). This cluster is also enriched for terms that indicate an early involvement of innervation and ciliogenesis (e.g., “neuron projection development” and “motile cilium”) which are common in other regeneration models [44,45,46,47]. The downregulated genes (cluster II) are enriched for terms that point to a shut-down of anabolic processes (“ribosome biogenesis” and “gene expression”) as well as primary metabolism (e.g., “mitochondrion” and “metabolic process”). Together, these clusters of early-regulated genes are consistent with a rapid response to the bisection insult that involves downregulation of highly energetic cellular processes and upregulation of functions that are specific to the injury response.
Clusters III and IV are composed of genes whose profiles are highly fragment-specific; these genes are differentially regulated in each fragment relative to control larvae. Many of these genes are expressed asymmetrically along the AP axis. Thus, bisection results in the loss of posterior-specific gene expression from anterior fragments and vice versa. For example, cluster III is enriched for genes annotated with functions specific to anterior larval fragments, such as “head development” [48], whereas cluster IV is enriched for genes associated with posterior fates in embryonic sea stars, such as “Wnt signalling pathway” [49].
Finally, although cluster V is comprised of relatively few genes, it is the most functionally coherent cluster. That is, the GO term enrichment analyses are the most statistically significant and reproducible across the three sources of functional annotations tested, i.e., de novo annotations and annotations based on orthology to Strongylocentrotus purpuratus and Mus musculus (Fig. 5b and Additional file 1: Figure S6). Genes assigned to cluster V are enriched for terms related to the cell cycle, DNA replication, and extracellular matrix (ECM) remodeling. The cluster V genes, which are upregulated late (by 6 dpb) in both fragments, likely reflect the onset of localized cellular proliferation that occurs at this time (Fig. 3a). Importantly, these genes are upregulated in regenerating fragments although the total number of proliferating cells has decreased compared to controls (Fig. 3a). This suggests that the cluster V genes represent a regeneration-specific increase in expression of proliferation-associated genes that is distinct from the normal, growth-associated proliferation.
Comparative transcriptome analyses reveal homologous genes with shared expression profiles among distantly related animals
Having identified the overall morphological progression of larval sea star regeneration (i.e., wound response, axis re-proportioning, and cell proliferation), we sought to determine if orthologous genes with similar temporal expression exist in other models of WBR. Such homology could indicate not only a shared overall progression, but that the genes involved are also in common. To address this question, we used published transcriptome data from regenerating planaria (S. mediterranea) [4] and hydra (H. magnipapillata) [5] for comparison. The Kao et al. dataset [4] was selected because it consolidated several planarian transcriptome assemblies, resulting in a more complete gene set, and also independently sampled both regenerating anterior and posterior worms, which is analogous to our own study design. Furthermore, the time points sampled range from 0 h post-amputation (hpa) to 72 hpa, at which point planarian blastemal proliferation reaches its peak [9]. This time frame roughly corresponds to the phases of regeneration considered in our study of larval sea stars. Regeneration has been less well characterized from a molecular standpoint in hydroids; the Petersen et al. dataset [5] is the only available transcriptome study from regenerating hydra. Here, RNA was sampled only from the distal tip of regenerating aboral tissues during the 48 h it takes to achieve complete head regeneration. As blastemal proliferation is not a feature of hydra regeneration, this characteristic cannot be used to synchronize the regenerative phases in this study to the other datasets. Nonetheless, these published datasets provide the best available basis for comparison to our sea star dataset.
To identify orthologs that share similar temporal dynamics during regeneration, the reported expression values from each dataset were clustered. For each comparative dataset, we assigned genes to three coarse clusters: those that were upregulated early in regeneration and downregulated later, those that were downregulated early in regeneration and upregulated in later regeneration, and those that exhibited some other temporal dynamic (Additional file 1: Figure S7 and S8). Finally, we identified genes in each of the five sea star expression clusters with orthologs in each of the planaria and hydra clusters. Using this approach, we find statistically significant overlaps between genes differentially expressed early in all three datasets as well as genes in the posterior-specific sea star cluster with clusters indicating fragment specificity in each of the other organisms. In the following sections, we describe how this allowed us to identify not only broad groupings of shared expression patterns but also specific orthologs similarly expressed across regeneration in these metazoans.
Early features of the regenerative response are highly similar
By analyzing the kinetics of orthologous gene activity in WBR, we find the strongest correlation among genes that are differentially expressed early in each dataset. That is, a significant number of orthologs are upregulated at early regenerative stages in both the sea star and planaria, as well as the sea star and hydra datasets (hypergeometric p = 4.5 × 10−3 and p = 8.8 × 10−9, respectively; Additional file 1: Figure S7 and S8). This set of genes is enriched for GO terms that include “cilium,” “calcium transport,” and “signaling.” Similarly, we also found a significant number of orthologs are downregulated in response to bisection in both sea star and planaria (hypergeometric p = 3.3 × 10−4). These orthologs are enriched for GO terms such as “ncRNA processing” and “ribosome,” suggesting that early repression of the energetically expensive process of ribosome biogenesis is a fundamental element of WBR.
Two intracellular signaling pathways, Ca2+ mobilization and MAPK signaling, have been broadly implicated in wound response [50,51,52,53,54] and are found to be upregulated early in bipinnaria regeneration. Recent proteomic data indicate that calcium signaling is involved in the anterior regeneration in planaria [55]. MAPK signaling, through both ERK and JNK pathways, is important in neoblast control and blastema differentiation in planaria [56, 57], and JNK signaling has been specifically linked with restoration of proper axial patterning in planaria by re-activation of appropriate WNT signaling [58]. Studies in hydra have similarly demonstrated that wound-responsive MAPK signaling is necessary for early specification of the head organizer, and thus functional regeneration. Early MAPK signaling may thus be shared feature of highly regenerative organisms [59].
The genes upregulated early in regeneration are also enriched for cilium-associated functions. The activation of these genes (e.g., Ccdc11, Rsph3, Iqcd, and Iqub; Fig. 6a) indicates that, in all three models, cilia play a central role in early regeneration. While this feature has not been reported in either planaria or hydra, a role for cilia in wound response and regeneration has been observed in mammals [45], zebrafish [47], and a related cnidarian (Nematostella vectensis) [46].
The set of similarly early activated genes also includes several key regulatory genes including orthologs of several tumor suppressor genes (i.e., Abl, Menin, Frk, Pten, Rbbp6L, Plk2, and Wee1; Fig. 6a). Several of these are also upregulated early in other regeneration models [60, 61]; these findings present an additional context in which the tumor suppressor genes show activity during regeneration. In regenerating sea star larvae, normal cell proliferation ceases prior to the emergence of the distinct wound-proximal proliferation (Fig. 3). The coincident activation of tumor suppressor genes and downregulation of ribosome biogenesis genes may be associated with this response. There is also an early signature of general cell cycle arrest in the hydra transcriptome [5]. While planarian neoblasts continue to proliferate at sites distal to the injury even during blastemal proliferation, inactivation of planarian PTEN gene homologs resulted in defective regeneration due to neoblast hyperproliferation [62]. These results indicate that a common early feature of WBR in these systems is the modulation of regulators of cell proliferation.
In addition to cell proliferation, these analyses suggest that cell death is tightly regulated early in regeneration. Genes associated with regulating cell death pathways are another example of similar differential expression early in these models. Notably, at least seven genes in the autophagy pathway are downregulated in regenerating sea star larvae, planaria, and hydra (i.e., Atg16L1, Atg12, Atg10, Atg14, and Uvrag; Fig. 6a). This is consistent with findings in hydra that suggest autophagic cell death is repressed during regeneration [63]. Conversely, as autophagy is downregulated in sea star larvae, genes that modulate apoptotic cell death are activated (e.g., Fadd, Birc6, and Ulk1). Apoptotic cell death is necessary for increased I-cell proliferation in hydra [18] and, in planarian regeneration, has been implicated in tissue remodeling and neoblast proliferation [64, 65]. Despite these early transcriptional changes, an increased number of TUNEL+ cells is not apparent until much later in bipinnaria regeneration (6 dpb; Fig. 4). Therefore, this modulation in cell death may be pathway specific (i.e., autophagy vs apoptosis) or otherwise undetected by our TUNEL assay. Alternatively, these transcriptional changes may be involved in establishing an appropriate balance between cell death and cell proliferation during this early phase.
Finally, we identified a suite of immediate early genes that are activated in all three animals. In regenerating sea star larvae, we find rapid, significant upregulation of Jnk, Elk, Egr, Klf2/4, Mcl, Creb3l3, Fra2, and FoxO (Fig. 6a). For example, Egr is one of the most strongly upregulated genes in both anterior and posterior regenerating sea stars (Fig. 6c), while in planarian regeneration EGR is one of the earliest and strongest wound-proximal genes induced during planarian regeneration [10]. The similar early downregulation of the Egr repressor Toe1 in both sea stars and planaria suggests these genes are parts of concerted early response in these contexts. Several of these early activation factors are also known to be regulated by MAPK signaling pathways in other systems [66]. For example, in the sea urchin Strongylocentrotus purpuratus, SpElk is a target of MAPK signaling (ERK) and regulates both SpRunt1 and SpEgr expression during embryogenesis [67]. In planaria, MAPK signaling (Jnk) activates Runt1 and Egr following wounding [65]. Jnk signaling in hydra has been shown to regulate FoxO expression [68], which is an important regulator of hydra I-cells [69].
These overlapping sets of genes differentially expressed early reflect a common response to the bisection insult. This suggests that these gene orthologs define key shared characteristics between highly regenerative species in a specific response to injury that permits the regeneration program.
Genes underlying shared early response are dramatically upregulated in the sea star wound site
We additionally chose a subset of these genes to examine their spatial localization during regeneration. Elk and Egr are both normally expressed in coelomic pouch epithelium (Fig. 6b″, c″), but by 3 hpb they are also strongly expressed in the sites of wound closure (Fig. 6b′, c′, Additional file 1: Figure S9 A, B). Fgf9 expression is also localized to wound sites during early regeneration (Additional file 1: Figure S9 F). Although neither Ets nor Erg were significantly differentially expressed by RNA-Seq or nanostring, we examined their expression given their known expression in sea star mesenchyme [70]. We find that both are localized to wound sites during early regeneration (Additional file 1: Figure S9 D, E), suggesting an early role for mesenchymal cells, although not necessarily due to a transcriptional change. Klf2/4 is normally expressed strongly in the mouth and foregut and after bisection is strongly upregulated in wound-proximal foregut (Fig. 6d′ and Additional file 1: Figure S9 C). Conversely, FoxO, Jnk, and Runt are expressed in the tip of the foregut proximal to the wound site, but not in the wound itself (Additional file 1: Figure S9 G–I). The tumor suppressor genes Abl and Pten are expressed broadly around the wound during early regeneration (Additional file 1: Figure S9 J, K). This spatial expression therefore shows that the set of gene homologs with early regenerative response among these deeply divergent animals are expressed in the early wound region of the sea star larva.
Axis respecification precedes wound-proximal proliferation
Restoration of normal gene expression levels along the bisected AP axis must be a central component of regeneration. Gene expression domains for components of the GRN that controls early axial patterning in sea star embryo have been well defined. The Wnt pathway, for example, has well-characterized functions in specifying the embryonic AP axis [49, 70]. Anterior ectoderm domains required for the development of the larval nervous system have also been delineated [71,72,73]. This enables us to analyze the expression of these genes during regeneration. And indeed, analysis of genes within the two expression clusters differentially expressed in regenerating anterior and posterior larval fragments (clusters III and IV; Fig. 5) demonstrates that embryonic axis patterning genes are expressed during AP axis restoration.
When examining these clusters, it should be noted that although genes in these clusters appear to be rapidly downregulated following bisection, because transcript levels were normalized to those in whole larvae, this phenomenon is actually a result of removing cells and tissues in the other half of the larva. For example, genes normally expressed in anterior larval domains (e.g., Frizz5/8 and FoxQ2) initially appear to be downregulated in posterior fragments relative to uncut larvae but are unaffected in anterior fragments (solid lines, Fig. 7; cluster III, Fig. 5). Correspondingly, genes that are typically expressed in the posterior domain (e.g., Frizz9/10, Wnt16, and Nk1) are absent in anterior fragments but unaffected in posterior fragments (dashed lines, Fig. 7; cluster IV, Fig. 5). For several genes in each of these clusters, expression levels recover to pre-bisection levels within 6 days. Notably, however, this process appears to be delayed within the regenerating anterior fragments relative to the posterior fragments (Fig. 7).
To characterize more fully the re-establishment of axial patterning during regeneration, we examined the spatial expression of two Wnt pathway receptor genes: Frizz5/8 (normally expressed in the anterior) and Frizz9/10 (localized in the posterior). In the anterior regenerating fragments, Frizz9/10 transcripts are undetectable following bisection (immediately after the posterior halves were removed). However, by 5 dpb Frizz9/10 transcripts are evident in the newly developed posterior domain (Fig. 7c). Additionally, we detect the re-expression of Frizz9/10 before the onset of wound-proximal proliferation. Likewise, Frizz5/8 is undetectable in regenerating posterior fragments until about 2 dpb when it is seen in the anterior aspect of these fragments (Fig. 7d), again before proliferating cells localize to this region. Appropriately localized expression of Frizz9/10 and Frizz5/8 persists in regenerating posterior and anterior fragments, respectively (Additional file 1: Figure S10 B, E). This finding extends to other genes with known roles in embryonic AP axial patterning that are identified in our clusters. For instance, we find similar recapitulation of embryonic expression patterns for, e.g., FoxQ2 (another anterior marker) and Wnt8 (an additional posterior marker; Additional file 1: Figure S10 F–J). Thus, embryonic patterning genes are used again during the restoration of the AP axis, and this precedes the initiation of blastemal proliferation.
This pattern mirrors planarian regeneration in which blastema formation, and regeneration cannot proceed when axis specification is perturbed [74,75,76]. Although hydra regeneration does not require blastemal proliferation, interstitial cells proliferate following wounding and this proliferation is initiated by a transient release of Wnt3, a protein implicated in head organizer function [18]. This comparison between animals positioned across the metazoa suggests the important finding that regeneration-associated proliferation requires a resetting of an axial positional program.
Common regulatory toolkit used for axial respecification
We sought to determine if any of the genes involved in sea star axis respecification during regeneration are conserved among animals. We examined the genes assigned to these fragment-specific clusters (clusters III and IV) to identify orthologous genes with similar expression trends in the other datasets. We find significant overlaps between the posterior-specific sea star genes (cluster IV) and asymmetrically expressed genes in both hydra (cluster 1, Additional file 1: Figure S8) and planarian (cluster 2, Additional file 1: Figure S7) datasets. The hydra oral-aboral axis corresponds to the posterior-anterior axes in bilaterians [77]. The RNA-Seq data from hydra were generated using oral regions of the regenerating aboral body stalk [5]. Thus, the signature of late stage upregulation reflects the recovery of transcripts typically expressed in the head (cluster 1, Additional file 1: Figure S8) and we expect that oral gene expression in hydra would correspond to posterior gene expression in sea stars. These nominally oral-specific genes in hydra in fact do exhibit a significant overlap with the posterior-specific sea star genes (hypergeometric p = 2.7 × 10−3). Likewise, genes asymmetrically expressed between anterior and posterior halves in the planaria dataset overlap the posterior-specific sea star genes (hypergeometric p = 1.4 × 10−2). In both cases, the overlapping genes include Wnt ligands and receptors (e.g., Wnt7, Wnt5, and Frizz9/10) and other regulatory genes associated with posterior fates (e.g., Bra, Hox11/13a, and Six1/2). The observed overlap in asymmetrically expressed genes among these datasets suggests that a common regulatory toolkit is deployed for axis respecification in each of these models that includes Wnt signaling. The absolute orientation of the axes is not conserved, but this likely reflects developmental usage.
Temporal dynamics of regeneration-induced cell proliferation differ among these animals
The patterns of cellular proliferation are one aspect in which the three models of WBR differ considerably. Sea star larvae and planaria exhibit concerted wound-proximal proliferation that coincides with the final time points sampled here: 6 dpb for sea star larvae and 3 dpb for planaria. Early in planarian regeneration, a global burst of neoblast proliferation is also observed (i.e., within 6 h post-amputation) [9]. No such early increase in proliferation is observed in sea star larvae (Fig. 3). While hydra do not rely on a proliferative blastema to resupply cells for regeneration, interstitial stem cells (I-cells) proliferate proximal to the wound within the first 2–4 h post-amputation [18]. This I-cell proliferation follows the early suppression of mitosis that is observed after wounding [5].
In sea star larvae, the genes upregulated later in regeneration in both the anterior and posterior fragments (cluster V; Fig. 5) are strongly associated with cell proliferation. It is important to note that while overall numbers of proliferation cells are decreasing, the timing of the upregulation of these genes correlates with the emergence of wound-localized proliferation. We compared these genes with orthologs that exhibit similar expression dynamics in the other datasets. None of the expression clusters from planaria or hydra are significantly enriched in orthologs of the sea star proliferation genes. Specifically, very few orthologs are apparent between the later upregulated sea star cluster (cluster V) and the corresponding gene clusters from planaria and hydra (i.e., planaria cluster 1 and hydra cluster 3; Additional file 1: Figure S7 and S8). Instead, there is a strong, though not statistically significant, overlap between the genes upregulated late in sea star and those upregulated early in planaria (e.g., cluster 3, Additional file 1: Figure S7) and hydra (e.g., cluster 1, Additional file 1: Figure S8). Many of these shared genes are associated with cycling cells (e.g., DNA polymerase subunits, MCM genes, structural maintenance of chromosomes [SMC] genes, Orc3, Rrm1, Plk, and Ttk). These data suggest the intriguing hypothesis that wound-proximal proliferation in sea star larvae is more similar to early bursts of cell proliferation than the later blastemal proliferation observed in planarian regeneration.
Regeneration induces coincident expression of normally tissue restricted proliferation-associated genes
We examined the expression patterns of proliferation-associated genes during regeneration (i.e., cluster V). Mcm2, Runt1, GliA, and Dach are all expressed in the anterior region of regenerating posterior fragments, coincident with the wound-proximal proliferation (Fig. 8b–e). Each gene is expressed in multiple distinct tissues, including the anterior foregut, anterior epithelium, coelomic epithelium, and gut (Fig. 8b–e). Notably, however, during embryonic and larval stages, these genes exhibit non-overlapping expression patterns. For example, Mcm2 is expressed in the ciliary band and foregut; Runt1 is expressed in the mouth, midgut, and hindgut; GliA is strongly associated with the developing coelomic epithelium; and Dach is expressed throughout the gut and in ciliary band epithelium (Additional file 1: Figure S11). These results indicate that a suite of genes that function in cell proliferation and are normally expressed in diverse tissues are re-deployed during regeneration and are co-expressed in the proliferating blastema.