Staging of early phases of regeneration based on morphological features
The development of a regenerating structure is a dynamic process that proceeds in a step-wise fashion to reconstitute a fully functional structure. To facilitate experimental investigation of arm regeneration in A. filiformis we identified five stages, which rely on observable morphological changes. Based on observations of >100 individuals (three of which are shown in Additional file 1: Figure S1) we propose a refined staging system (Fig. 1), which is relevant to the phases of regeneration when cell specification, differentiation and initial arm patterning begin to occur. The presence of clear morphological landmarks is supported by histological analysis carried out on paraffin sections stained with Milligan’s trichrome technique [43], which detects collagen (cyan) and individual cells (pink/magenta) (Figs. 2 and 3). The average duration and standard deviation (S.D.) of time it takes for an arm to regenerate up to a given morphological stage is reported in Fig. 1A, as well as Additional file 1: Figure S1A and B. The timing of regeneration shows a certain degree of variation in agreement to what has previously been reported [31].
All of the stages and corresponding average timings have been compiled based on arms amputated 1cm from the central disc. We categorize stage 1 of regeneration as wound healing and re-epithelialization. This phase occurs between 1 and 4 days post-amputation (dpa) (Fig. 1A; Additional file 1: Figure S1) and involves changes which mediate the closure of the wound and re-epithelialization and remodeling of the existing tissue; however, from whole mount DIC (differential-interference contrast) observations little or no changes are evident at the amputation plane. On the contrary, at this stage histological sections show (Fig. 2) the aboral coelomic cavity (ACC), the ectoneural sinus (ES) and the radial water canal (RWC) are sealed off and the wound is completely re-epithelialized by epidermal cells, already covered by a thin and faint cuticle indicated by the asterisk in Fig. 2A’. The intervertebral muscles adjacent to the amputation site acquire a disorganized pattern and show morphological signs of histolysis: myocytes lose their elongated shape (Fig. 2A”, B). The duration of time between amputation and stage 2 shows the highest variability (Additional file 1: Figure S1B). To assess whether the type of traumatic amputation might have an effect on the rate of the wound healing stage and the appearance of a regenerative bud, we amputated two arms from the same animal: one clean, blunt amputation similar to natural autotomy, and the other skewed (amputation at an angle; Additional file 1: Figure S1C, D). We used 5 animals of the same size to minimize the individual variability and we documented the process of regeneration of each arm for 10 days. These data reported in Additional file 1: Figure S1E reveal a general inter-individual variability in time to reach stage 2, which is not related to the type of amputation.
Once re-epithelialization is accomplished, a small regenerative bud protrudes from the distal end of the stump, in the oral region (Fig. 1A, B; Additional file 1: Figure S1A). We characterize this as stage 2, which occurs on average after 5.3 dpa ±0.91 S.D. (Additional file 1: Figure S1A and B). From whole mount DIC observations the regenerate appears optically homogeneous (Fig. 1B), however histological sections indicate already a certain degree of organization, in which the first outgrowths of the regenerating radial nerve, radial water canal and the aboral coelomic canal are visible (Fig. 2B). A thin layer of connective tissue is present below the wound epidermis where mesenchymal cells are embedded (Fig. 2B’, B”).
Stage 3, which occurs on average 6.3 dpa ±0.48 S.D. (Fig. 1A; Additional file 1: Figure S1A and B), differs from stage 2 in that the regenerate acquires a more organized and complex inner architecture with loss of the external optical homogeneity; at this stage the radial water canal and the sub-epidermal mesenchymal cells become clearly visible in whole mount DIC images (Fig. 1C, C’). Histological sections show the regenerate mainly contains projections of the radial nerve, radial water canal and aboral coelom, which protrude from the amputation plane (Fig. 3A, A”). Mesenchymal cells embedded in collagenous tissues can be detected throughout the dermal layer (Fig. 3A’, A”).
We define stage 4 (on average 7.8 dpa ±0.63 S.D., Fig. 1A; Additional file 1: Figure S1A and B) by the appearance of the first metameric unit, which is formed at the most proximal region of the developing regenerate. A small bulging of the epidermis reveals the early segregation of the regenerating segments (dotted lines, Fig. 1D). Outpocketing of the radial water canal system, which will eventually form the podia, can start to be distinguished in histological sections (Fig. 3B, B’).
Finally, stage 5 occurs on average after 8.8 dpa ±0.63 S.D. (Fig. 1A; Additional file 1: Figure S1A and B). This stage is characterized by several small repetitive units, which can be observed bulging at the proximal side (dotted line, Fig. 1E) with several podia precursors beginning to form along the new regenerate (Fig. 3C, C’).
As the regeneration process progresses metameric units form following a proximal-distal gradient. The late regeneration stages that we use in this work corresponds to the 50 % (2–3 weeks post-amputation) or 95 % (4–5 weeks post-amputation) regeneration stages described before [31, 37]. These arms have a 50 or 95 % differentiation index (DI), which is the ratio between the length of the arm that contains differentiated structures (like spines and podia) and the total regenerate length. The duration of regeneration time is highly variable between stage 5 and the 50 % stage depending on many aspects including length of arm lost, functional requirements, animal size or environmental factors [31].
Skeletogenesis during early and late regeneration stages
To define when and where skeletogenesis occurs during early stages of regeneration in the brittle star arm we combined light transmission microscopy observations of whole regenerates with calcein staining to detect the newly forming mineralized skeleton. Calcein is commonly used to visualize calcium carbonate deposition, which is shown as green fluorescence [44]. Single spots of green fluorescence, corresponding to the forming skeletal primordia, are consistently first observed at stage 3 indicating the earliest stage at which differentiated skeletogenic cells are present (Fig. 4a, n = 10). No fluorescence can be detected at stages 1 and 2 (Additional file 1: Figure S2). At stage 4 more elaborate tri-radiated or tetra-radiated skeletal elements called spicules can be observed in the regenerate (Fig. 4a, n = 10). At stage 5, the spicules present in the distal end of the regenerate have no obvious pattern of distribution. In contrast, skeletal elements in the newly forming metameric units, at the proximal end of the regenerate, are formed with a typical bilateral distribution that corresponds to where the future lateral shields will be (Fig. 4a , n = 10). High magnification of the regenerates at different stages reveals that the single spicules form just below the well-developed epidermis in the dermal layer (Fig. 4b), corresponding to the position of the mesenchymal cells embedded in the collagenous matrix observed in histological sections (Figs. 2 and 3).
Later during regeneration, when the arm reaches the 50 % regeneration stage (50 % DI), the whole maturation of developing skeleton can be observed (Fig. 4c) simply by analysis of subsequent metameric units from proximal (most mature skeleton) to distal (new elements) and using light microscopy. This is due to the birefringent properties of calcified structures. At the distalmost end the distal cap becomes highly calcified (terminal ossicle). The first observable distal metameric unit already contains a tiny tri-radiated lateral spicule (Fig. 4c, first lateral spicule inset). Several metameric units (eight in Fig. 4c) separate the appearance of the first lateral spicules from the appearance of the first vertebral spicules, forming more proximally in the inner layers of the regenerating arm. Contrary to skeletal elements developing during early regeneration and at lateral positions during late regeneration, which always form multi-branched spicules, the vertebrae first appear as two long, non-radiated skeletal rods (Fig. 4c, first vertebral spicule inset). Later, as the skeletal elements mature, they also begin to branch out and the two parallel vertebral elements fuse together at the midline between distal and proximal ends of the metameric unit (Fig. 4c, advanced vertebrae inset). Oral and aboral shields begin to form approximately at the same level as the developing vertebrae (Fig. 4c, first aboral shield inset). In a distal to proximal gradient the segments contain increasingly complex spicule structures that will eventually form the stereom of skeletal ossicles, as is typical for echinoderms (Fig. 4c, advanced shield formation inset). The fully formed structures found in an adult non-regenerating arm are shown in Fig. 4d. A schematic diagram of the organization of individual skeletal shields in the adult non-regenerating arm is shown in Additional file 1: Figure S3.
Expression of the skeletogenic marker genes in the early and late stages of regeneration
To better characterize the cells involved in skeleton formation during regeneration in the brittle star we assessed the spatial expression pattern of the skeletogenic cell differentiation genes. We first analyzed the closest matching gene, identified in an A. filiformis embryonic transcriptome, to the sea urchin C-lectin. This spicule matrix protein containing a C-type lectin domain is expressed specifically in the skeletogenic mesodermal cells [45]. Proteomic studies identified the C-lectin protein as an integral part of biomineralized structures both in sea urchin larval spicules and adult test plates and spines [16, 17], and in ophiuroid adult arm skeletal elements of the species Ophiocoma wendtii [46]. The expression of the A. filiformis c-lectin gene (Additional file 1: Table S1) has been first examined in the larval stages with extended skeletal elements (i.e., ophiopluteus). Indeed, only a group of mesenchymal cells arranged in a pattern coinciding with the skeletal elements express Afi-c-lectin (Additional file 1: Figure S4). We have therefore used it as a marker for skeletogenic cells in adult regenerating arms (Fig. 5). It is first detected by in situ hybridization (ISH) at stage 2 of early regeneration in a broad sub-epidermal domain (Fig. 5b). It then becomes much more restricted to the dermal layer only by stage 4 (Fig. 5c, d). This is consistent with the developmental changes that occur between these two stages, when the regenerate acquires a more heterogeneous structure containing morphologically differentiated tissue types. Strong colorimetric staining has the tendency to diffuse in the tissue, rendering the precise boundary of expression not clearly distinguishable. To discriminate whether the expression of Afi-c-lectin is restricted specifically to the sub-epidermal mesenchymal cell layer, where spicules form, or extends to other domains of the regenerate, we additionally performed a fluorescent ISH on regenerating arm sections. We found that the expression of this gene is indeed specifically localized to a single cell layer just underneath the epidermis (Fig. 5e, f).
Next, we examined the expression of Afi-c-lectin in the 50 % DI stage of the regenerating arm (Fig. 6). In the distalmost-undifferentiated segments of the regenerate, Afi-c-lectin is expressed in the dermal layer similar to the early stages of regeneration (Fig. 6A). Towards the proximal side of the regenerating arm, its expression becomes more complex, corresponding to the formation and patterning of the different skeletal elements (Fig. 6A). The scattered mesenchymal cells expressing the marker gene are arranged in regular and repetitive patterns, which mirror the pattern of the stereom structure of the oral, aboral, lateral shields with spines and vertebrae (Fig. 6A’) in each of the differentiating segments of the arm. The expression pattern of Afi-c-lectin thus coincides specifically with the location and time of appearance of the different skeletal elements; therefore, confirming it as a reliable marker of skeletogenesis during regeneration as well as embryonic development. This is further supported by the expression pattern of Afi-c-lectin in the most advanced proximal segments of the 95 % DI regenerating arms (Fig. 6B–D). Two additional genes, identified in a recent embryonic study of A. filiformis larval skeleton development [36], were examined to confirm the molecular signature of the mesenchymal cells in the dermis and the cells in the differentiating skeletal elements of late stages of regeneration (Additional file 1: Figure S5). Both Afi-p19 and Afi-p58b have a similar expression pattern to Afi-c-lectin at the early stages of regeneration and at the distal end of the 50 % DI stage arms (expression in the dermis, Additional file 1: Figure S5). The genes are more restrictively localized to individual skeletal elements in differentiating proximal segments of the arm (Additional file 1: Figure S5). Afi-p58b is expressed in spines and vertebrae, whereas Afi-p19 is localized to the oral and lateral shields, and vertebrae.
Differentiated skeletogenic cells are not proliferative
To test when active cell proliferation is initiated post amputation and if mesenchymal cells - located in the dermal layer where the skeleton is formed - have the ability to proliferate during regeneration, we used the EdU assay to label cells in, or having gone through S-phase during the early stages of regeneration. In normal non-regenerating arms some proliferating cells, labeled by EdU, are identifiable in different tissue types including the epidermis, podia, radial water canal and in cells surrounding the vertebrae (Additional file 1: Figures S6 and S7; n = 3). On the contrary, no cells are labeled during the first hours (8–24) post-amputation (Additional file 1: Figure S7; n = 3) in the plane of injury. Only at the end of stage 1 (between two and three dpa) is a marked increase in EdU labeling visible at the wound site prior to bulging of the regenerative bud, mainly in correspondence with the position of the radial nerve cord and the radial water canal (Additional file 1: Figure S7; n = 3) in the oral half of the metameric unit. Stage 2, which marks the appearance of the regenerative bud, shows extensive cell proliferation in both the epidermis and the inner tissues, containing the outgrowths of the above-mentioned structures (Fig. 7A; n = 3).
Five individual arms were used for EdU labeling of stage 3 regenerates. Confocal maximum projections of whole z-stacks also show a large amount of proliferating cells at this stage (Fig. 7B). Notably, closer inspection of individual z-planes reveals that the dermal layer contains the only cells, that are not labeled with EdU, implying that they did not proliferate during the time course of labeling (Fig. 7B’, B”).
As the number of cells that express Afi-c-lectin clearly increases during the regeneration process, we investigated whether the cells have the ability to divide at these late stages. To answer this question we performed a fluorescent ISH using Afi-c-lectin on arms previously labeled with EdU (n = 2). If skeletogenic cells marked by the expression of Afi-c-lectin were proliferating, cells with a red nucleus (EdU DNA incorporation) and surrounding green cytoplasm (Afi-c-lectin RNA expression) would be observed. Extensive cell proliferation can be observed throughout the regenerating arm at this stage (Fig. 8A). However, closer inspection of individual confocal z-planes shows that none of the green-labeled Afi-c-lectin cells overlap with the red EdU+ cells in a manner indicating that skeletogenic cells have nuclei in the S-phase of mitosis (Fig. 8; Additional file 1: Table S2; Additional files 2 and 3). Specifically, none of the EdU labeled red nuclei are surrounded by green Afi-c-lectin labeled cytoplasm (Fig. 8B, D). This trend is also apparent in the whole 50 % stage arm as shown in the proximal segments (Fig. 8A, B) and the distalmost tip (Fig. 8C, D). Therefore, the Afi-c-lectin expressing cells do not proliferate throughout the whole arm regeneration process, in agreement with the role of c-lectin as a final differentiation gene in skeletogenesis. Notably, the distalmost part (terminal ossicle and podium) (Fig. 8C) of the regenerate contains very few proliferating cells. However, the area just proximal to it is a domain of major accumulation of EdU+ cells. This is consistent with previous observations [37], and the expression patterns of genes here reported, and thus we conclude that the regenerating arm proliferative growth zone is located just underneath (proximally) to the terminal ossicle. New metameric units are then added proximal to the growing tip of the regenerate.