Investigation into the cellular origins of posterior regeneration in the annelid Capitella teleta

Abstract Many animals can regenerate, although there is great diversity in regenerative capabilities. A major question in regenerative biology is determining the cellular source of newly formed tissue. The polychaete annelid, Capitella teleta, can regenerate posterior segments following transverse amputation. However, the source, behavior and molecular characteristics of the cells that form new tissue during regeneration are largely unknown. Using an indirect cell tracking method involving 5′‐ethynyl‐2′‐deoxyuridine (EdU) incorporation, we show that cell migration occurs during C. teleta posterior regeneration. Expression of the multipotency/germ line marker CapI‐vasa led us to hypothesize that stem cells originate from a multipotent progenitor cell (MPC) cluster, migrate through the coelomic cavity, and contribute to regeneration of tissue. We show that the capacity for posterior regeneration and segment formation is greater with than without the MPC cluster. Finally, we propose a working model of posterior regeneration in C. teleta. This work is the first in C. teleta that addresses the potential source of cells contributing to posterior regeneration, and may provide clues as to why some animals are highly successful regenerators.


INTRODUCTION
Many animals can regenerate, although there is great diversity in relative regenerative capabilities. For example, some animals are able to regenerate their complete body from only a few thousand cells (e.g., some planarians), while others can regenerate only a single cell or tissue type (e.g., cardiac muscle in zebrafish) (Bely & Nyberg, 2010).
The polychaete annelid Capitella teleta displays robust posterior (but not anterior) regeneration following transverse amputation of body segments. Within 1 day of amputation, wound healing has occurred, and by 2 days a regeneration blastema is formed. The blastema is filled with actively proliferating cells and axonal extensions from the ventral nerve cord. During the following days, new tissue types appear, including ectodermal epithelia, gut, circular and longitudinal muscles, and neurons. By 12 days post-amputation (dpa), there are typically between three and 13 newly formed segments (de Jong & Seaver, 2016). The source of cells that generate the blastema is currently unknown, be it via proliferation of stem cells, cell migration, dedifferentiation, or transdifferentiation. Determining the source of cells recruited to form new tissue during regeneration will allow us to gain a greater understanding of the regeneration process, such as how C. teleta can regenerate multiple tissue types.
In this study, we investigated the cellular source of regenerating tissue in C. teleta. We hypothesized that a source of cells distant to the wound site makes a contribution to the regenerate. To test whether there is cell migration during C. teleta posterior regeneration, we used an indirect method involving incorporation of 5 ′ -ethynyl-2 ′ -deoxyuridine (EdU) and 5 ′ -bromo-2 ′ -deoxyuridine (BrdU) to label and track cells in living animals. We used the expression of CapI-vasa to mark putative stem cells present in the coelomic cavity, and analyzed their morphology and spatial distribution following transverse amputation. CapI-vasa is also expressed by a second population of cells that was previously identified as a putative primordial germ cell (PGC) cluster (Dill & Seaver, 2008). Whether the putative PGC cluster has a restricted role in formation of the germ line is not known. Expression of the orthologs of vasa, nanos and piwi are not restricted to the PGC cluster, but are also expressed in somatic stem cells of the posterior growth zone in larvae, juveniles, and adults in C. teleta (this study; Dill & Seaver, 2008), as well as in tissues outside the germ line in other annelids including Platynereis, Enchytraeus, and Tubifex (Oyama & Shimizu, 2007;Ozpolat & Bely, 2016;Rebscher, Zelada-González, Banisch, Raible, & Arendt, 2007;Sugio et al., 2008). Therefore, we investigated the possibility that the putative PGC cluster is actually a heterogeneous population of multipotent stem cells, and serves as a source of somatic stem cells during posterior regeneration. Henceforth we refer to the previously named PGC cluster as the multipotent progenitor cell (MPC) cluster. We assessed the relative capacity for posterior regeneration in juveniles with and without the putative MPC cluster, using axon outgrowth, cell proliferation, and number of regenerated segments as markers of regenerative capability.

Proliferating cells migrate into the blastema during posterior regeneration
To investigate a possible distant source of cells that contribute to new tissue formed during regeneration of posterior segments, we performed a set of experiments we termed "EdU pulse−amputate−wait−BrdU pulse" (Fig. 1). In these experiments, uncut juveniles were incubated in EdU, which was incorporated into cells undergoing the S-phase of cell cycle during a 1 h exposure time. We used this as an indirect method to label and track cells in living animals over time. Following EdU incorporation, juveniles were amputated and allowed to regenerate for up to 3 days. Immediately before fixation, juveniles were exposed to a second DNA synthesis marker, BrdU, for 1 h. The location of both EdU-positive and BrdUpositive cells was subsequently visualized. EdU-positive nuclei in the regenerating tissue show cells that originated in pre-existing segments and are now present in the new tissue. BrdU-positive nuclei indicate cells that were actively dividing in the animal immediately before fixation. A nucleus labeled with both EdU and BrdU in regenerated tissue indicates a cell that was previously dividing in segments anterior to the wound that migrated to new tissue, and subsequently divided a second time. Ventral nerve cord architecture was also visualized by anti-acetylated -tubulin reactivity, which allowed for monitoring changes in the nervous system during blastema formation. In addition, because amputation results in an abrupt termination of the nerves of the ventral nerve cord, this labeling provides precise determination of the amputation site.
EdU-positive and BrdU-positive cells are scattered throughout the body in animals fixed at 0−1 h post-amputation (hpa), which is typical for actively growing 2-week-old juveniles ( Fig. 1A-A ″ ). At 1 dpa, wound healing has occurred but a regeneration blastema has not formed, and specimens resemble individuals at 0−1 hpa ( Fig. 1B-B ″ ). At 2 dpa, a small regeneration blastema has formed, and fine axonal projections extend into the new tissue (Fig. 1C, pale blue arrows). Several EdU-positive cells are present in this new tissue, along with a multitude of actively dividing cells, evidenced by BrdU incorporation ( Fig. 1C-C ″ ). At 3 dpa, a large regeneration blastema has formed, and it contains many EdU-positive cells, indicating migration of cells into the blastema (Fig. 1D). In addition, the blastema contains a large number of actively dividing BrdU-positive cells (Fig. 1D ′ ). This pattern is typical for a regeneration blastema of this stage (de Jong & Seaver, 2016), where proliferating cells are detectable in all three germ layers of the forming blastema (Fig. S1). In three independent EdU pulse−amputate−wait−BrdU pulse experiments, almost all blastemas at 2 dpa and 3 dpa contain EdU-positive cells (n = 40/42). In animals with EdU-positive cells, the number in the blastema ranged from only a few cells (two to three) to many cells (over 50). In general, at 2 dpa most animals possessed fewer than 25 EdU-positive cells in the blastema, while at 3 dpa most animals possessed more than 25 EdU-positive cells in the blastema. In addition, we occasionally see cells in the regeneration blastema at 2 dpa and 3 dpa that are both EdU-positive and BrdUpositive ( Fig. 1D ″ , open arrowhead). In order to ascertain if the initial incubation in EdU negatively affected subsequent posterior regeneration, we compared regeneration of animals that had been exposed to EdU with those that had not. One set of juveniles were amputated at the boundary of segments 10 and 11 following a 1 h EdU incubation. A second set of juveniles were treated similarly but were not exposed to EdU. Both sets of animals were allowed to regenerate for 14 days, and the number of segments that had regenerated were counted using the presence of segmentally repeated peripheral nerves through anti-acetylated -tubulin labeling as a marker of mature ganglia ( Fig. 2A). There is not a significant difference in the number of segments regenerated comparing animals that had been exposed to EdU and those that had not been exposed to EdU. In addition, EdU-positive cells are visible in the new tissue, where they had become incorporated into new segments. This indicates that cells that had incorporated EdU 14 days earlier remain healthy, and contribute to differentiated tissues in the  We monitored CapI-vasa expression following amputation in cells in F I G U R E 2 Robust regeneration of posterior segments following EdU exposure. (A) The average number of segments regenerated 14 days following amputation in animals either exposed to EdU or not exposed to EdU before amputation. Numbers at the top of the columns indicate the number of juveniles sampled. Error bars show the standard deviation from the mean. (B)−(E ′ ) Confocal stacks of posterior ends of two different amputated juvenile individuals in ventral view, with anterior to the left. The panels in each row are from a single individual. Each individual was exposed to EdU for 1 h, amputated at the boundary of segments 10 and 11, and was then allowed to regenerate for 14 dpa. Although it was technically impractical to count the number of CapIvasa-expressing cells within the coelomic cavity due to their small size and three-dimensional arrangement within the body cavity, the percentage of juveniles that contained CapI-vasa-expressing coelomic cavity cells was calculated (Fig. 3J). One set of animals was fixed immediately following amputation between segments 10 and 11 (0 hpa), and in these animals 41% of individuals have CapI-vasa-expressing cells within their coelomic cavity (n = 26/63). A similar proportion is observed at 1 dpa (49%, n = 34/69). In contrast, there was a substantial increase in the proportion of juveniles containing CapI-vasa-expressing cells in the coelomic cavity at 2 dpa (81%, n = 51/63) and at 3 dpa (77%, n = 49/64).
The proportion of juveniles that contain CapI-vasa-expressing cells in their coelomic cavities at 2 dpa and 3 dpa are significantly different from the proportion containing CapI-vasa-expressing cells at 0 hpa (P < 0.01 in both cases).
The CapI-vasa-expressing cells do not appear to have a specific location within each segment, other than being present in the coelomic cavity. However, when their relative axial position is taken into account, it is clear that these cells are restricted to particular segments. For each individual, the segment number where CapI-vasa-expressing cells was seen at four time points following amputation was recorded (0 hpa, 1 dpa, 2 dpa and 3 dpa). At all time points, the vast majority of individuals (96%, n = 25/26 at 0 hpa; 97%, n = 33/34 at 1 dpa; 98%, n = 57/58 at 2 dpa; and 98%, n = 48/49 at 3 dpa) contain these cells in segments 6−10, while only a small fraction of the cells are located in segments 1−5 ( Fig. 3K; 4% at 0 hpa, 3% at 1 dpa, 2% at 2 dpa and 2% at 3 dpa). When present, CapI-vasaexpressing cells are most often present in more than one segment ( Fig. S2). In cases where they were observed in only one segment, The proportion of juveniles with CapI-vasa-positive cells in the coelomic cavity immediately following amputation, or 1, 2, or 3 days following amputation. Numbers at the top of the columns indicate the number of juveniles sampled. Statistical significance compared with 0 h controls with a P value <0.01 is denoted by ***, while n.s. indicates that the difference in proportions was not statistically significant. (K) The axial distribution of CapI-vasa-positive cells in the coelomic cavity immediately following transverse amputation at segment 10 (0 hpa), or 1, 2, or 3 dpa. The percentage of juveniles with CapI-vasa-positive coelomic cavity cells in segments 1−5 or 6−10 is indicated with black bars. Numbers to the right of rows indicate the number of juveniles sampled at each time point. Scale bars in (A)−(E) represent 100 m. Scale bars in (F)−(I) represent 10 m the number ranged from only a few (approximately three to six cells) to many (too numerous to count). The MPC cluster is consistently located in segment 5 of juveniles, and also usually extends to include part of adjacent segments 4 and/or 6 (see below). Thus, almost all CapI-vasa-expressing coelomic cavity cells are positioned between the MPC and the wound site, which led us to hypothesize that cells of the MPC cluster migrate through the coelomic cavity and contribute to the blastema during posterior regeneration.

Morphological characteristics of the MPC cluster and its response to transverse amputation and posterior regeneration
Previous studies in C. teleta identified a cluster of cells in larvae, juveniles, and adults that express the germ and stem cell markers vasa, nanos, and piwi (Dill & Seaver, 2008). This group of cells is localized in the coelomic cavity and suspended by mesentaries, sandwiched between the ventral nerve cord and gut, and has been hypothesized to play a role in formation of the germ line. Although vasa, nanos, and piwi show indistinguishable expression patterns in the MPC cluster, expression analysis of a fourth marker, Ct-myc, indicates molecular heterogeneity within the MPC cluster (see below). In addition, vasa, nanos, and piwi are all also expressed in undifferentiated cells of the posterior growth zone. Therefore, we now consider the possibility that this cluster may be a heterogeneous population of multipotent stem cells that can contribute to somatic tissues during regeneration, in addition to formation (or replacement) of the germ line.
To further characterize the MPC cluster, the axial position, morphology, cell division characteristics, and response to transverse amputation of the MPC in 2-week post-metamorphic juveniles were investigated. We used CapI-vasa as a marker of the putative MPC cluster

Posterior regeneration in the presence and absence of the MPC cluster
To determine whether the MPC cluster is involved in posterior regeneration, we performed amputations that removed the MPC cluster. In a first set of experiments, a single amputation was made either anterior to the MPC cluster (between segments 3 and 4) or posterior to the MPC cluster (between segments 6 and 7). However, amputation anterior to the MPC cluster resulted in a high rate of mortality of animals (de Jong and Seaver, unpublished observations). Therefore, in a second set of experiments, two simultaneous anterior and posterior amputations were made. For the first group of animals, one transverse amputation was conducted between the third and fourth segments, and a second amputation between the thirteenth and fourteenth segments.
In this group, the MPC cluster, which resides in segment 5, was retained

Evidence for contribution of migrating cells to posterior regeneration in the polychaete annelid Capitella teleta
In this study, we present the first evidence of cell migration during posterior regeneration in C. teleta. Using a technique we term "EdU pulse−amputate−wait−BrdU pulse" to track a subset of cells in living animals, we show that, following transverse amputation, proliferating cells from pre-existing segments migrate into the regeneration blastema. These observations are striking considering that a short EdU pulse, which labels only a subset of dividing cells, allowed us to visualize cell migration into the blastema. Furthermore, a proportion of these cells divide after migration. The fact that these migratory cells are dividing and at least some can continue to divide suggests that they are undifferentiated stem cells. These migratory cells contribute to differentiated cells and tissues in the regenerate. Our data add to previous results indicating migration of cells into the wound site during regeneration of various annelid species. However, to our knowledge, our results are the first demonstration through analysis of living tissue for a cellular contribution from migrating cells in any polychaete.
The origin of cells that form the blastema has long been the subject of intense research and speculation in annelids (Bely, 2014;Berrill, 1952;Clark & Clark, 1962;Cornec et al., 1987;Cresp, 1964;Faulkner, 1929Faulkner, , 1932Herlant-Meewis, 1964;Hill, 1970;Krecker, 1923;Zattara & Bely, 2011). Cell migration during annelid regeneration has been documented in multiple species, and appears to be a general feature of annelids (see Bely, 2014). Migration can initiate minutes after  Krecker, 1923;Myohara, 2012;Randolph, 1891Randolph, , 1892Tadokoro et al., 2006). Morphologically, they have been described as either spindle shaped or round cells with a large nuclear to cytoplasmic, that reside on the intersegmental septa on the dorsal surface of the ventral nerve cord within the coelomic cavity. In the oligochaete Enchytraeus, cells with these features express vasa, and appear to migrate to the wound site following injury (Sugio, Yoshida-Noro, Ozawa, & Tochinai, 2012;Sugio et al., 2008). The cells we observe to migrate into the regenerating tissue in C. teleta have some characteristics of neoblasts. For example, the initial EdU exposure shows that they were actively dividing before migration began, and co-labeling with BrdU in some of these cells shows that they continue to divide. We hypothesize that a proportion of these dividing, migrating cells also express vasa, and therefore have a large nuclear to cytoplasmic ratio. However, rather than definitively identify the migratory cell population in C. teleta as neoblasts, we instead acknowledge that migration is occurring during regeneration and that these cells have a stem cell character. While some of the proliferating cells that migrate into the blastema in C. teleta are later incorporated into new segments, there may also be unlabeled cells that migrate into the newly forming tissue. For example, EdU incorporation only labels cells that are undergoing DNA synthesis at the time of exposure, and thus the 1 h pulses used in these experiments will probably only label a subset of a larger population of actively dividing cells. In addition, it is also possible that non-dividing cells (e.g., differentiated cell types) also migrate, as reported in previous studies (see Bely, 2014), including in a live imaging study in Pristina leidyi (Zattara et al., 2016). These non-proliferating migrating cells could then undergo a dedifferentiation or transdifferentiation step when they reach their final destination, or be part of an early immune response to the wound.
Evidence from fixed tissue studies in multiple annelids suggests that, in general, the formation of each germ layer in the new tissue is derived from the corresponding germ layer in the pre-existing tissue (Bely, 2014;Berrill, 1952;Boilly, 1967Boilly, , 1968Clark & Clark, 1962;Cornec et al., 1987;Cresp, 1964;Hill, 1970). That is, ectoderm derives from ectoderm, mesoderm from mesoderm, and endoderm from endoderm. For example, in the sabellid Sabella melanostigma, all three germ layers appear to be involved in forming the blastema; ectodermal cells originate from ectoderm, endodermal cells contribute directly to outgrowth of endodermal epithelia, and mesodermal tissues arise from mesodermal coleomic cavity cells and from fragmentation of pre-existing muscle, which probably dedifferentiates before dividing and redifferentiating (Hill, 1970). Migration of differentiated endodermal cells and subsequent incorporation into regenerating digestive tissue has also been documented in the oligochaete Lumbriculus (Tweeten & Reiner, 2012). Although our experiments do not specifically address the cellular origins of the different germ layers in the regenerate, our EdU pulse−wait−BrdU pulse experiments show that mesodermal structures in the regenerate are formed, at least in part, by EdU-positive cells in the coelomic cavity (mesodermal layer) that have migrated from segments anterior to the wound. In addition, proliferating cells are present in the blastema of C. teleta in all three germ layers within 3 dpa (Fig. S1). These proliferating cells are likely to be incorporated into newly differentiated tissue during regeneration, and may contribute to the regenerate from outgrowth of existing tissue.
As additional cell types that contribute to the regenerated tissue are revealed, e.g., by time-lapse imaging of fluorescently labeled cells in live animals, we will be better able to determine the origins of these cells, and their contributions to the formation of each germ layer in the new tissue.
Although our results show the migration of EdU-positive cells into the blastema, it should be noted that we are not able to determine from where these cells originate. For example, it is possible that the cells originate in several segments anterior to the wound, and then migrate posteriorly across multiple segments following an amputation event. Alternatively, migration may occur on a small scale. For example, cells could originate within the segment closest to the wound site, and migrate only within this segment. It is also possible that cells originate directly adjacent to the wound, so that migration need only occur over very small distances. A previous study in C. teleta quantified the number of dividing cells following transverse amputation in the three segments closest to the wound site, and compared the number of these cells to the number in the corresponding segments in uncut controls (de Jong & Seaver, 2016). In the first 24 hpa, there is a decrease in the number of dividing cells in the three segments closest to the wound site. However, by 2 dpa, cell proliferation recovers to levels comparable to uncut controls, and then increases over the following days. The initial observed decrease in the number of dividing cells might suggest that the origin of the proliferating, migrating cells are within the three segments proximal to the wound site. The migration of these cells would deplete the reserves available in these segments, resulting in the observed decrease noted in the previous study. The recovery and then subsequent increase in proportion of dividing cells in these segments could be due to the birth of new cells within these segments, or continued migration from cells further anterior. A second explanation to account for this decrease in proliferating cells proximal to the amputation site is that long-range signals originating from the wound cause a shutdown of cell division, as has been proposed in the annelid Pristina leidyi (Zattara & Bely, 2013). That is, a cell within the cluster may divide, and the daughter cell migrates from the cluster to the wound site, while the mother cell remains in its original position. Such a self-renewing strategy is characteristic for stem cells in multiple species and contexts (see Shenghui, Nakada, & Morrison, 2009;Tanaka & Reddien, 2011).

Identification of candidate migratory stem cells
At all time points examined (0 hpa, 1 dpa, 2 dpa and 3 dpa), a small proportion of juveniles do not have detectable CapI-vasa-expressing coelomic cavity cells. It is possible that the expression of CapI-vasa is dynamic within these cells, and therefore fixation at a specific time point results in a subset of individuals that do not express CapI-vasa.
Alternatively, the absence of these cells might reflect variability in regeneration capability among individuals. For example, the number of segments regenerated after 14 days varies widely (Fig. 2A).
The absence of CapI-vasa-expressing cells in the coelomic cavity may account for some of this individual variation; individuals that do not have CapI-vasa-expressing coelomic cavity cells regenerate fewer segments than those that do. We cannot distinguish between these possibilities directly, as we do not currently have a way to label these cells in living animals.
Molecular heterogeneity was identified within the MPC cluster.

Specifically, Ct-myc is expressed in a subset of the cells that express
CapI-vasa  in the MPC cluster of C. teleta. This molecular heterogeneity within the MPC cluster could indicate different fates among cells. One annelid example in which differences in stem cell fates are reflected by molecular differences comes from the oligochaete Enchytraeus. Here, germline stem cells and germ cells express piwi and both vasa paralogs (vlg-1 and vlg-2), while neoblasts (somatic stem cells) express only vlg-2 (Sugio et al., 2008). Similarly, in the planarians Dugesia and Schmidtea, all neoblasts (note that planarian neoblasts are distinct from annelid neoblasts) express piwi, but a subpopulation destined to form the germline expresses both piwi and nanos (Handberg-Thorsager & Saló, 2007;Sato et al., 2006;Shibata, Rouhana, & Agata, 2010). In C. teleta, there may be further subpopulations of cells in the MPC cluster, coelomic cavity, or posterior growth zone that express unique combinations of stem cell markers. Unique combinations of markers might reflect differing potency within these populations. As more molecular markers are identified, it will allow us to further investigate the molecular heterogeneity and potential differences in developmental and regenerative potency of these cells.

Assessing the contribution of the MPC cluster to posterior regeneration
As a functional test of the importance of the MPC cluster to posterior regeneration, we compared the regenerative capability between animals with and without an MPC cluster. Due to the design of these experiments, the regeneration response was not as robust as in animals undergoing posterior regeneration with anterior segments intact. With intact anterior ends, there is substantial cell division in the blastema by 3 dpa (de Jong & Seaver, 2016). However, in our experiments, most animals in both groups did not consistently form a proliferative regeneration blastema until 5 dpa, suggesting that there is a delay in the onset of proliferation in the blastema. All experimental animals with the MPC cluster and a majority without the MPC cluster possessed EdU-positive cells in the blastema at 5 dpa. At 12 dpa, most animals with the MPC cluster had regenerated one to three segments (85%, n = 12/14, Fig. 6), while most animals without the MPC cluster had not regenerated any segments (89%, n = 8/9; a single individual regenerated one segment). This is in stark contrast to the situation in animals with intact anterior ends, which regenerate between three and 13 segments after 12 days. In addition, a new posterior growth zone forms, which is indistinguishable from the growth zone seen in intact, actively growing animals (de Jong & Seaver, 2016). In our experiments, we did not observe the formation of a posterior growth zone in any animal from either group, regardless of the number of segments formed. The lack of a posterior growth zone in these animals would then, in turn, probably limit the formation of new segments. The limited number of regenerated segments and the lack of a posterior growth zone at 12 dpa is probably due to the inability of the animals to feed. A previous study noted that starvation of both intact and regenerating animals results in decreased cell proliferation and gene expression in C. teleta (de Jong & Seaver, 2016). A direct link between cell division and nutritional status has also been observed in the oligochaete Pristina leidyi (Zattara & Bely, 2013). Our results suggest that posterior regeneration in both the presence and absence of the MPC cluster can be initiated, and new segments form in the animals with an MPC cluster. However, the compromised nutritional status of the animals in this experiment reduces the regeneration response in both groups. Our results demonstrate that the presence of the MPC cluster confers an advantage to regenerating animals, allowing them to regenerate a small number of new segments.
The results of the experiments we conducted directly demonstrate a role for cells residing within segments 4−6 in conferring regeneration ability, and provide support that the identity of these cells is within the MPC cluster. Other strategies were attempted to directly remove the MPC cluster, but proved problematic. For example, the precursor cell of the MPC cluster (3D) was deleted in cleavage stage embryos. While most larvae resulting from these deletions lack an MPC cluster, juveniles compensate for loss of the embryonic precursor cell (Dannenberg and Seaver, in preparation). Direct laser ablation (e.g., Pernet, Amiel, & Seaver, 2012;Yamaguchi & Seaver, 2013) of the MPC cluster in juveniles is not currently possible because the MPC cluster is not visible without the aid of molecular markers, prohibiting its direct observation in living animals. Clearly, further studies aided by alternative techniques are needed before the contribution of the MPC cluster to regeneration can be more thoroughly understood.

Importance of the nervous system in regeneration
One unexpected observation from our analyses of regeneration in individuals with and without the MPC cluster is that posterior regeneration in C. teleta can occur in the absence of a brain, albeit with fewer segments compared with animals that have an intact anterior end.
In many annelids, the presence of a brain (supraesophageal ganglion) is required for regeneration of posterior segments (Casanova, 1955;Clark & Bonney, 1960;Clark & Clark, 1959;Durchon, 1956;Golding, 1967a;Hausenchild, 1960). The brain in these animals is proposed to secrete hormones that promote regeneration (Clark, 1959;Golding, 1967b;Herlant-Meewis, 1964;Hofmann, 1966). However, there are also examples in which posterior regeneration can occur in the absence of the brain, such as in the two annelids Branchiomma and Chaetopterus (Hill, 1972). Notably, in annelids that can regenerate a head, the presence of brain tissue is not required for successful regeneration (see Bely, 2006). These examples highlight the substantial variation across annelids in the requirement of a brain for regeneration. In C. teleta, decreased posterior regeneration capability in the absence of anterior structures might simply be due to the inability of the animal to feed after amputation. However, it may also be that, although the brain is not absolutely required, it plays a role in posterior regeneration, and regeneration is impeded if the brain is removed.
Our analyses of regeneration in individuals with and without the MPC cluster suggest a functional link between the presence of dividing cells in the blastema and the extension of axons from pre-existing segments. In specimens lacking neurite extension into the blastema, EdU-positive cells are not detectable in the blastema. This suggests a temporal or functional link between extension of axons and the onset of cell proliferation. For example, during the initial stages of blastema formation, it is possible that a first step in the process is axon extension, and a second step is an increase in proliferating cells.
Two theories explaining this phenomenon have been proposed: nerves secrete factors that activate cell proliferation and differentiation in the blastema (Herlant-Meewis, 1964;Müller et al., 2003;Varhalmi et al., 2008), or axons act as highways to facilitate the migration of cells to the wound site (Cornec et al., 1987;Stephan-Dubois, 1954). Based on current data, both of these alternatives are possible in C. teleta. Future experiments aimed at assessing the involvement of the nerve cord during C. teleta posterior regeneration will allow us to gain a greater understanding of the mechanisms driving a successful regenerative event.

Model of posterior regeneration in C. teleta
From this study and from data in other annelids, we propose a working model of posterior regeneration in C. teleta, which includes both local and distant sources of cells that contribute to the regenerating tissue (Fig. 7). Following injury, a signal originates at the wound site and stimulates a cellular response. This response includes the initiation of cell migration to the wound site. The origin of these migrating cells is both close to the wound site (e.g., less than one segment away) and distant from the wound site (e.g., more than one segment away). A subset of the migrating cell population are proliferating stem cells that express the stem cell marker vasa. A second, simultaneous response to the initial injury involves the extension of axons from the ventral nerve cord into the forming blastemal tissue. Following migration of cells to the wound site, there is at least one further round of cell division before these cells differentiate and become incorporated into a subset of tissues in the regenerated posterior segments. The remaining tissues in the regenerate are derived from proliferation of cells from local sources. In some cases, these cells may be restricted to contribute only to the germ layer from which they originate. Future studies investigating the origin and fate of cells that contribute to regenerating tissue, and the mechanistic involvement of neuronal projections, will allow us to better understand the processes that are essential for successful regeneration. This in turn may give insight into why C. teleta is capable of posterior, but not anterior, regeneration. Broader implications of this study include identifying aspects of annelid regeneration shared with other metazoans, and those that are unique. This will provide clues as to why some animals are highly successful regenerators, while others are not.

Animal husbandry and transverse amputations
A colony of C. teleta was maintained in the laboratory at 16−19 • C, according to published culture methods (Grassle & Grassle, 1976).
Larvae were allowed to emerge naturally from the maternal brood tube, and 40 individuals were induced to metamorphosed into single glass fingerbowls containing filtered seawater (FSW) and previously frozen and sieved estuarine mud as a food source. Regeneration experiments were performed on juveniles at 2 weeks post-metamorphosis.
To isolate enough juveniles of the appropriate age for experiments, animals were initially removed from the mud and placed in 0.5% cornmeal agar:FSW plates (Sigma 42347, Darmstadt, Germany) supplemented with 60 g/mL penicillin (Sigma P3032), plus 50 g/mL streptomycin (Sigma S6501). Typically animals remained in agar plates for 1−4 h before amputation and fixation, which also aided in removal of mud particles from the exterior of their bodies. Prior to amputation, animals were immobilized in 0.5% cornmeal agar supplemented with MgCl 2 (1:1 FSW:0.37 mol/L MgCl 2 ), for between 15 and 30 min. Immediately before amputation, animals were placed in a drop of 0.37 mol/L MgCl 2 :FSW (1:1) on a platform of black dissecting wax (American Educational Products, Fort Collins, CO, USA) and amputated at the posterior edge of the target segment using a microsurgery scalpel (Feather; 15 degree blade, Carlsbad, CA, USA).
Chaetal morphology and morphological differences between thoracic and abdominal segments allowed accurate targeting of a specific amputation position. Following amputation, animals were returned to 0.5% cornmeal agar:FSW dishes for up to 24 h. For analysis of longer periods of regeneration, animals were placed into bowls of FSW and sieved estuarine mud for the desired length of time.

Detection of cell proliferation by incorporation of EdU
To detect cells in S-phase of the cell cycle, EdU (Life Technologies C10337, Carlsbad, CA, USA) was added to a dish of FSW containing whole or regenerating juveniles at a final concentration of 3 mol/L for 1 h. Animals were then placed in 1:1 0.37 mol/L MgCl 2 :FSW for F I G U R E 7 Working model of posterior regeneration in C. teleta. Schematic representation of the steps involved in C. teleta posterior regeneration. All panels show a representation of the posterior ends of amputated juveniles in ventral view, with anterior to the left. Black vertical lines indicate the amputation site in each panel. The ectoderm is shown as a white outermost layer, the mesoderm is shaded in grey, and the endoderm is shown in light brown. The coelomic cavity is represented by the space between the endoderm and mesoderm. Axons of the ventral nerve cord are shown in red. (1) Following injury, a signal originates at the wound site and stimulates a cellular response. (2) This response involves both migration of cells to the wound site and extension of axons from the pre-existing segments into the forming blastemal tissue. A subset of cells that migrate are dividing (green), and some of these also express CapI-vasa (dotted green). (3) Migrating cells reach the wound site and divide. Proliferation of cells in pre-existing tissue close to the wound site also occurs. Both local and distant sources of cells contribute to the growing blastema. (4) Incorporation and differentiation of previously proliferating cells into the regenerated tissue 15 min before fixation overnight at 4 • C in 3.7% paraformaldehyde (PFA):FSW. Animals were then rinsed several times with phosphatebuffered saline (PBS) and exposed to PBS + 0.5% Triton X-100 before the EdU detection reaction was performed, following the manufacturer's recommendations. For some experiments, in situ hybridization or immunohistochemistry was performed on samples prior to detection of EdU (see appropriate sections below).

Cloning of C. teleta Ct-myc gene
A single C. teleta myc homolog was identified with a tblastn search against the C. teleta genome (JGI; ELT88315). A recent study confirmed the presence of a single myc homolog in C. teleta, and phylogenetic analyses placed it within the Myc protein family (Bao, Xu, & Shimeld, 2017).
Fragments of coding sequence were amplified by polymerase chain reaction from mixed larval stage cDNA, using gene-specific primers

Whole-mount in situ hybridization
Following fixation in 3.7% PFA:FSW overnight at 4 • C, juveniles were washed in PBS, dehydrated through a methanol series to 100% methanol, and stored at −20 • C for up to 4 weeks. A DIG-labeled riboprobe for CapI-vasa (Dill & Seaver, 2008) (BK006523) was generated with the SP6 MEGAscript kit (Ambion Inc., Austin, TX, USA) and DIG-11-UTP (Sigma 11209256910). The CapI-vasa probe length was 1122 bp, and was diluted to a final concentration of 1 ng/ L. A DIG-labeled riboprobe for Ct-myc (MF693912) was generated with the T7 MEGAscript kit (Ambion) and DIG-11-UTP (Sigma). The Ct-myc probe length was 964 bp, and was diluted to a final concentration of 1 ng/ L. Whole-mount in situ hybridization was performed following published protocols (Seaver & Kaneshige, 2006). Following hybridization at 65 • C for 48-72 h, the probe was detected using nitroblue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) color substrate. Typically, the reaction was allowed to develop for 4−5 h. Extended development of the substrate reaction did not result in different expression domains for either gene. If a combination of EdU incorporation and in situ hybridization was to be performed, the in situ hybridization procedure was completed before the EdU detection reaction.

Immunohistochemistry
Following EdU incorporation (see Section 4.2), juveniles were washed several times in PBS + 0.1% Triton X-100 (PBT), before being treated with block solution (PBT + 10% normal goat serum, Sigma G9023) for 45-60 min at room temperature. Mouse anti-acetylated -tubulin antibody (Sigma T6743) was diluted 1:300 in block solution, and animals were incubated for 2−18 h at 4 • C. Animals were washed twice in PBT, followed by four PBT washes of 20-30 min each. Donkey anti-mouse-546 secondary antibody (Invitrogen A21203, Carlsbad, CA, USA) was diluted 1:250 in block solution, and animals were incubated for 2-18 h at room temperature. Following two rinses in PBT, four PBT washes of 20-30 min each were conducted. Animals were imaged and analyzed as described in Section 4.8.

EdU pulse−amputate−wait−BrdU pulse
Two-week-old juveniles were exposed to EdU at a final concentration of 3 mol/L in FSW for 1 h. Animals were then placed in 1:1 0.37 mol/L MgCl 2 :FSW for 15 min, and then amputated at the boundary between segments 10 and 11. Juveniles collected for the 0 hpa time point were then exposed to BrdU (Sigma 858811) at a final concentration of 0.1 mg/mL for 1 h, before being placed in 1:1 0.37 mol/L MgCl 2 :FSW solution for 15 min to immobilize them, and then fixed in 3.7% PFA:FSW overnight at 4 • C. Regenerating juveniles were allowed to regenerate for specific time periods following amputation (1 dpa, 2 dpa, or 3 dpa), and were then collected and exposed to BrdU for 1 h, immobilized, and fixed overnight as described above. Following fixation, animals were rinsed several times with PBS and digested with 0.01 mg/mL Proteinase K (Invitrogen 25530049) in PBS for 5 min. After this incubation, the Proteinase K solution was removed and animals were re-fixed in 3.7% PFA:PBS for 10 min. Animals were then washed three times in PBS. DNA was denatured by incubation in prewarmed 4 mol/L HCl for 15−30 min at 37 • C. The solution was then neutralized with at least five washes of 0.1 mol/L sodium borate over 15−30 min. Animals were then washed twice in PBS, followed by exposure to PBS + 0.5% Triton X-100, and the EdU detection reaction was performed according to the manufacturer's recommendations. Following detection of EdU, animals were placed in block solution (PBT + 10% normal goat serum; Sigma G9023) for 45 min to 1 h at room temperature. Animals were then incubated for at least 3 h at room temperature, or overnight at 4 • C, in mouse anti-BrdU antibody (1:250; Thermo

Statistical analyses
CapI-vasa-expressing cells in the MPC cluster were manually counted at 0 hpa, 1 dpa, 2 dpa, and 3 dpa using a compound microscope. Cell counts were taken from at least 55 individuals for each time point, and comparisons were made between the number of CapI-vasaexpressing cells at 0 hpa and either 1 dpa, 2 dpa, or 3 dpa. Statistical significance was determined using a Student's two-tailed t test. Differences in the number of CapI-vasa-expressing cells in the MPC cluster were considered statistically significant if the P value was <0.05.
The proportion of juveniles with CapI-vasa-expressing cells in their coelomic cavity at 0 hpa was compared with the proportion of juveniles that contained CapI-vasa-expressing cells in their coelomic cavity at 1 dpa, 2 dpa, or 3 dpa. Juveniles were counted as possessing CapI-vasaexpressing cells in their coelomic cavity if any cells were seen in any segment, regardless of the number of cells. At least 63 individuals for each time point were included. Statistical significance was determined using a two-tailed Z test. A P value <0.05 was considered statistically significant.

Microscopy and imaging
Before imaging, animals were placed in 80% glycerol:PBS plus 0.125 g/ L Hoechst 33342 (Life Technologies H3570) overnight.
Specimens analyzed by in situ hybridization alone were imaged using an Axioskop 2 motplus compound microscope (Zeiss, Gottingen, Germany), coupled with a SPOT FLEX digital camera (Diagnostic Instruments Inc., Sterling Heights, MI, USA). Images were captured using SPOT imaging software (version 5.2). All other specimens containing a combination of EdU, BrdU, and/or immunohistochemistry and in situ hybridization were imaged using a Zeiss LSM 710 confocal microscope (Zeiss, Gottingen, Germany). Z-stack projections were