Regeneration mechanisms in Syllidae (Annelida)

Abstract Syllidae is one of the most species‐rich groups within Annelida, with a wide variety of reproductive modes and different regenerative processes. Syllids have striking ability to regenerate their body anteriorly and posteriorly, which in many species is redeployed during sexual (schizogamy) and asexual (fission) reproduction. This review summarizes the available data on regeneration in syllids, covering descriptions of regenerative mechanisms in different species as well as regeneration in relation to reproductive modes. Our survey shows that posterior regeneration is widely distributed in syllids, whereas anterior regeneration is limited in most of the species, excepting those reproducing by fission. The latter reproductive mode is well known for a few species belonging to Autolytinae, Eusyllinae, and Syllinae. Patterns of fission areas have been studied in these animals. Deviations of the regular regeneration pattern or aberrant forms such as bifurcated animals or individuals with multiple heads have been reported for several species. Some of these aberrations show a deviation of the bilateral symmetry and antero‐posterior axis, which, interestingly, can also be observed in the regular branching body pattern of some species of syllids.

F I G U R E 1 Schematic drawing of the syllid body plan. Illustration of the main features in the morphology of a syllid: prostomium, segments, pygidium and digestive tube components, pharynx, pharyngeal tooth, proventricle, and intestine Several studies in annelids have compared juvenile development with regeneration, since both processes (may) involve the establishment of a proliferation zone, also known as a segmental addition zone (SAZ) (Balavoine, 2014;. Additionally, developmental pathways of key transcriptional factors can be triggered by regenerative processes . However, during early development the body axes are established de novo, while during regeneration body axes are defined by the remaining old tissue (Boilly, Boilly-Marer, & Bely, 2017). Thus, a different mechanism might be used for establishment of body axes in regenerating individuals (Boilly, Boilly-Marer et al., 2017).
The annelid body consists of a prostomium (the anterior cap or head), a set of segments which are often equipped with chaetaebearing parapodia (body wall projections used for locomotion), and a pygidium (the posterior cap or tail) (Figure 1). In annelids, posterior regeneration is found widely distributed across taxa, whereas anterior regeneration is restricted to fewer groups (Hyman, 1940). It has been supposed that anterior regeneration is ancestral for annelids, and subsequently this ability has been lost independently across the annelid tree (Bely, Zattara, & Sikes, 2014;. The addition of segments by the action of a SAZ is clearly detectable during posterior regeneration (Balavoine, 2014). Observations of anterior regeneration show an antero-posterior developmental gradient of newly forming segments, in which the anterior cap (prostomium) forms first, and the SAZ is not clearly detectable or transient.
Syllidae is an interesting group for studying regeneration. It is one of the largest groups within Annelida in terms of biodiversity, with a wide variety of reproductive modes and different regenerative processes (Aguado, Nygren, & Siddall, 2007;Franke, 1999;Okada, 1929).
Studies about syllid regeneration are limited and inhomogeneous in terms of investigated species, processes, and applied methods. Nevertheless, the available information represents a treasure trove for future investigations. The purpose of this study is to summarize available data on regeneration in this group of animals. We focus on the regenerative abilities and reproductive modes of syllids, differences between anterior and posterior regeneration, cells and tissues involved, and aberrant forms as a result of anomalous regenerative processes.
antoni . This species was also used for investigating muscular and nervous system regeneration Weidhase et al., 2017).  (Franke, 1999). (C) Gemmiparous individual during sequential gemmiparity based on Myrianida pachycera (Okada, 1933b). When the stolons are completely developed, they are detached for spawning. The stock animal regenerates the posterior body after the detachment of stolons. (D) Individual during posterior regeneration. (E) Male and female stolons perform spawning for exchange of gametes; scheme based on stolons of Syllis spp. (F) After spawning, stolons die. (G) Larval development in non-brooding species; larvae disperse in ocean. (H) Alternatively, autolytine female stolons live during external brooding and development of the embryos; mature female carries an egg sack until the release of juveniles; ventral brooding; based on Myrianida pachycera (Okada, 1933b (Okada, 1929). From an isolated segment, Amblyosyllis formosa was observed regenerating posteriorly the pygidium and three segments (Abeloos, 1952). Autolytines are able to regenerate anteriorly and posteriorly including short fragments with only two, three, and four segments (Allen, 1923(Allen, , 1927a(Allen, , 1927b. Fragments with two segments (chaetiger 12−13) of Procerastea halleziana can regenerate anteriorly up to seven segments and prostomium, and posteriorly the pygidium plus four segments (Allen, 1923). In contrast, most of the sylline species, such as Syllis amica (Boilly, 1967d), T. antoni  and Odontosyllis fulgurans (Okada, 1929), revealed a limited anterior regeneration demonstrated, for example, by the lack of a new proventricle, pharynx, or pharyngeal tooth (Figures 3, 4A, Table 1). Regeneration in syllids occurs by an epimorphic process, but morphallaxis is also observed (see Section 6) (Boilly, 1969;Delye, 1962;Weidhase et al., 2017;. After amputation, at first an invagination is formed at the edge of the wound; after about 1-3 days the wound heals, and then dedifferentiation or blastema formation is initiated (Boilly, 1962b(Boilly, , 1967a(Boilly, , 1967cWeidhase et al. 2017). The next steps are blastema growth, re-differentiation, tissue remodeling, and resegmentation (Boilly, 1968a(Boilly, , 1968bWissocq, 1970aWissocq, , 1970b) (see Section 5). The total regeneration time is variable among species, and there is a lack of detailed information about how long the process takes (Table 1).
In comparison with other species the development of the prostomium and brain of T. antoni is relatively fast (Weidhase et al., 2017) (Table 1).

REGENERATION AND SEXUAL REPRODUCTION IN SYLLIDAE
Syllids can reproduce sexually by epitoky, a process in which adult individuals undergo a metamorphosis before spawning Franke, 1999). Epitoky can be subdivided into two different modes: epigamy and schizogamy Franke, 1999;Nygren & Sundberg, 2003). In epigamy, the sexually mature organisms modify the body by developing a hypertrophied cephalic sensory apparatus and swimming notochaeta in the midbody-posterior segments.
After spawning the adult organisms usually die Schroeder & Hermans, 1975). In schizogamy, the modified segments Additionally, there are two types of schizogamic reproduction named gemmiparity and scissiparity (Figures 2 and 5). In gemmiparity several successive stolons are produced simultaneously, while in scissiparity only one stolon is produced at a time (Nygren & Sundberg, 2003).
In scissiparity, the stolons result from the modification of existing segments, whereas in gemmiparity the stolons are developed de novo, i.e., new segments are generated to form the stolons (Franke, 1999;Schroeder & Hermans, 1975). Scissiparity is also found, although with some differences, in other groups of annelids such as Eunicidae (Rouse & Pleijel, 2006); in contrast, gemmiparity is unique to syllids (Aguado, Glasby et al., 2015).
The experiments consisted of bisecting the animals in segments 13 or 14, where usually the stolon head is formed (Durchon & Wissocq, 1964). When this bisection is made in early stages of stolonization, in which a differentiation of a brain, eyes and appendages occurs, the posterior piece regenerates an adult head at the level of the stolon head (Okada, 1929) while the anterior piece regenerates a new tail, and the previous stolon head is completely regressed (Durchon & Wissocq, 1962, 1964. This regression depends on the section region. Dissections closer to the stolon head promote the regression, whereas dissections closer to the posterior end do not interfere in stolon formation (Durchon & Wissocq, 1962). These results point out that regeneration inhibits stolonization depending on the distance of the bisection from the stolon head (Durchon & Wissocq, 1962, 1964. Similar findings were also described in P. leidyi, a clitellate annelid that reproduces by paratomic fission, where amputation after the fission zone could provoke its resorption (Zattara & Bely, 2013).

REGENERATION AND ASEXUAL REPRODUCTION IN SYLLIDAE
In annelids, asexual reproduction is always accomplished by subdivision of the body and regeneration of missing parts (Schroeder & Hermans, 1975). This process is widely known as fission and can occur in two different ways: fission or architomy, in which physical separation precedes the development of new tissues; and paratomy, in which new tissues are developed prior to physical separation . Phylogenetic analyses in annelids suggest that fission has repeatedly evolved by co-option of regenerative abilities and the presence of anterior regeneration is necessary to evolve fission .
True asexual reproduction seems to be extremely rare in syllids (Franke, 1999). Some few species with high regenerative abilities may reproduce asexually by fission, such as P. halleziana ( Figure 6) (Allen, 1923(Allen, , 1927a, see below) and S. gracilis  ( Figure 3). This reproductive mode implies the development of new individuals from fragments of the stock organism that are capable of regenerating new anterior and/or posterior ends (Allen, 1923(Allen, , 1927a(Allen, , 1927bBoilly & Thibaut, 1974;Okada, 1929) (Figures 6, 7).  Table 2). As a result, the pattern of fission areas in P. halleziana could be described mathematically by a "fragmentation formula": H7 + 2 + 2 + 2 + 3 + 3 + 3 + 4 + 4 + 4 + 4(+4) + 3 + 3 + 3 + 3 + n(+3) + xP, where the numbers indicate the segments of each fragmented piece (H, head; P, pygidium) (Allen, 1923) ( Figure 7A, Table 2). The possible existence of specific fragmentation formulae stimulated a series of experiments in different groups of syllids, especially in Autolytinae (Allen, 1927a(Allen, , 1927bOkada, 1929). Okada (1929) found that the addition of a KCl solution induced this fragmentation as well. The contact with the solution provoked an unusually strong contraction of the longitudinal muscles of specific segments (Okada, 1929). Both methods (distilled/salted water; KCL solution) revealed that the animals fragmented in specific areas of the body (Figure 7, Table 2). However, the fragmentation formula of each species may vary depending on the method used, either with distilled water or KCl solution (Okada, 1929).
According to these studies, the first fragment (with prostomium) usually breaks in an odd number of segments and it is usually longer than the rest ( Figure 7A-C, Table 2). In autolytines and eusyllines, the number of segments in the anterior fragments is usually 7, 9, 11 or 13 (Allen, 1927a(Allen, , 1927bOkada, 1929). Generally, posterior fragments F I G U R E 7 Fragmentation series in species of three subfamilies of Syllidae. (A) Autolytinae, Procerastea halleziana, after Allen (1923Allen ( , 1927aAllen ( , 1927b. (B) Eusyllinae, Synmerosyllis lamelligera (Allen, 1927b). (C) Syllinae, Trypanosyllis zebra, after Okada (1929). (D) Autolytinae, Procerea picta, collected from Ferrol, Galicia, Spain; the method of intercalated addition of distilled water and seawater was performed with this specimen, following Allen (1923). Regions of constrictions are marked indicating their number of segments. The series obtained with the constriction regions is comparable to the fragmentation series described for this species, following the color pattern (Allen, 1927a;Okada, 1929). Formula characters: H, head; n, variable number of fragments; x, variable number of segments; P, pygidium. Scale bar 500 m vary in number of segments and in a stolonizing individual the break occurs at the level where the stolon head is developed (Allen, 1927a;Okada, 1929). Furthermore, some species exhibited high variation in the formula, depending on the presence or absence of stolon (Allen, 1927a). Interestingly, the fragmentation formulae of Proceraea picta and Epigamia macrophtalma coincide with the color pattern found in the body surface of these species (Allen, 1927a). We have repeated the experiment of fragmentation induction with distilled water in a specimen of P. picta. Although the animal did not break, constrictions appeared in the same fragmenting areas previously described coinciding with its color pattern ( Figure 7D).
Ten species are known to show a fragmentation formula, including those that do not reproduce by fission, i.e., species in which the fragments are unable to regenerate completely (Table 2). Fission can be an asexual reproductive mode easily confounded with non-reproductive autotomy and post-injury regeneration, especially from field observations . The fragmentation formulae observed experimentally in syllids (Allen, 1923(Allen, , 1927aOkada, 1929) might indeed be the result of a stress induction and not a truly asexual reproductive process (Franke, 1999;Langhammer, 1928). The addition of distilled water could be the factor that provokes autotomy as a defensive strategy based on discard of part of the body.

ANTERIOR VERSUS POSTERIOR REGENERATION
Most of the studied species regenerate anterior as well as posterior ends of the body, with the exception of A. formosa and E.
macrophtalma that are not able to regenerate anteriorly ( Figure   3, Table 1). However, fragments of both these species can survive for some weeks without a regenerated head (Abeloos, 1952;Allen, 1927a). Interestingly, in A. formosa, a species with a fixed number of segments, the regeneration of the posterior end never
Anterior resegmentation and posterior resegmentation involve different processes that vary among species (Figure 8). Most authors observed that posterior regeneration involves the establishment of a SAZ, immediately before the pygidium (Allen, 1923; (Figure 8B, C). Concerning anterior resegmentation, two patterns were observed in different species of syllids. In most syllids the blastema grows and differentiates into a prostomium (at first) and regenerated segments, and a SAZ is transient or not detectable ( Figure 8B) (Aguado, Helm et al., 2015;Boilly, 1961Boilly, , 1965a, Weidhase et al., 2017. However, in some cases the anterior regeneration seems to bear a SAZ, because segments are regenerated by sequential addition from the original segment of the stock ( Figure 8C) (Aguado, Helm et al., 2015;Allen, 1923).
Whereas the first pattern is widely recognized in annelid regeneration (Balavoine, 2014), the second is unusual, reported only for the syllids P.
halleziana and (in some instances) T. antoni (Aguado, Helm et al., 2015;Allen, 1923). In this case, fragments of the midbody of P. halleziana can establish two SAZs (anterior and posterior), in which segment addition occurs following the posterior−anterior direction (Allen, 1923) ( Figure 8C). As is well known for nereidids, posterior growth during juvenile development involves similar genetic pathways (as evidenced by similar gene expression patterns) to those occurring in the SAZ during posterior regeneration (Balavoine, 2014;Prud'homme et al., 2003). Unfortunately, these animals do not regenerate the anterior body. Whether anterior growth could also share similar mechanisms to juvenile growth is a topic that still needs to be clarified. Boilly (1961) and Okada (1929) showed that the extent of regeneration depends on the region of dissection. Specimens of Eusyllis blomstrandi failed to regenerate anteriorly or died when dissected in the most posterior part of the body (Boilly, 1961). A similar limitation was observed in the posterior regeneration of Myrianida edwarsi, in which the anterior end needed a certain number of segments to regenerate a new tail (Okada, 1929). In general, posterior regeneration usually produces more new segments than anterior regeneration, with the exception of S. gracilis that is able to regenerate anteriorly up to 18 segments and posteriorly up to eight segments (Boilly & Thibaut, 1974;Mesnil, 1901) ( Table 1).

CELLS, TISSUES AND ORGANS INVOLVED
Early stages of regeneration including cell dynamics, histogenesis, and renewal of organs were the focus of some investigations in syllids.
The first signs of re-establishment of tissues in the regeneration of S. amica are an invagination and retraction of the intestine (Boilly, 1967a). Both anterior and posterior fragments exhibit wound healing, followed by blastema formation by bulging of the scarring epidermis in 2-3 dad (Boilly, 1962b(Boilly, , 1967a. As a result of several studies, three phases of cell dynamics during regeneration were observed: (1) cellular dedifferentiation, loss of the elements that configure the characteristics of each cellular type; (2) activation and transformation of the regenerative cells related to intense synthesis of RNA; and (3) differentiation, characterized by an acquisition of a new structure and specific proteins (Boilly, 1962a(Boilly, , 1962b(Boilly, , 1965a(Boilly, , 1965b(Boilly, , 1967a(Boilly, , 1967b(Boilly, , 1967c(Boilly, , 1967d(Boilly, , 1968a(Boilly, , 1968b. Blastema formation is the main characteristic of the dedifferentiation stage. The cells involved in blastema formation and renewing tissues derive from the last segment preceding the section site (Boilly, 1962a(Boilly, , 1962b(Boilly, , 1965. In contrast to clitellates, migratory regenerative cells have so far not been found in Syllidae (Boilly, 1967a). A quantitative antero-posterior gradient has been observed in blastemal tissues of anterior regeneration in M. edwarsi (Okada, 1929). The more posterior the dissection region, the smaller blastema, and the total quantity of tissue derived from blastema is reduced (Berrill, 1952;Okada, 1929). The activation stage is supposed to be dependent on nervous fibers close to the wound region, and it begins once the signals of activation appear in the cells near the nerve chain (Boilly, 1967a(Boilly, , 1967b. This process does not occur immediately after section, due to the time required to heal the axon that elongates towards the microenvironment of the blastema and stimulates regeneration (Boilly, 1967a;Boilly, Faulkner et al., 2017). Neuropeptides, neurotransmitters, growth factors, and morphogens were suggested as mediators in this activation process (Boilly, Boilly-Marer et al., 2017). Finally, during the differentiation stage, the regenerative cells experience changes in cytoplasmic elements to acquire specialized functions, e.g., muscular and epidermal cells (Boilly, 1967a(Boilly, , 1968b. During this process, ribosomes are abundant and proteins associated with cell proliferation and differentiation are clearly observed (Boilly, 1968b).
Regeneration of the nervous system occurs by growth and elongation of the remaining nerve fibers (Weidhase et al., 2017). The nervous system of T. antoni begins to regenerate anteriorly during the blastema stage (2 dad) with an ingrowth of neurites originating in the remaining ventral nerve cord, and then neurites form loops in 4 dad together with the nerve cord elongation and development of a new brain (Weidhase et al., 2017). In S. amica, the ventral nerve cord was also observed in an elongation process from 3 dad and the prostomium was completely differentiated in 7 dad (Boilly, 1967a). The nervous system at the posterior end of T. antoni regenerates in a similar way, but the blastema is formed in 3 days, and a day after a new pygidium with cirri is present (Weidhase et al., 2017). This process is comparable to the posterior regeneration of S. amica, differing only in the appearance of the pygidium at 9 dad (Boilly, 1967a).

THE ROLE OF THE PROVENTRICLE DURING REGENERATION
Several studies in schizogamous species focused on the triggers of the stolonization process. Usually, experiments with removal of the proventricle indicated a potential role of this structure during reproduction (Durchon, 1957(Durchon, , 1959Franke, 1980Franke, , 1983Franke & Pfannenstiel, 1984;. Posterior ends of fragments without proventricle seem to increase the production of male stolons, suggesting that the proventricle inhibits the stolonization and is involved in sexual determination (Durchon, 1957(Durchon, , 1959Durchon & Wissocq, 1964;Franke & Pfannenstiel, 1984;Heacox & Schroeder, 1982;Wissocq, 1966). Experiments in which the proventricle is removed in S. prolifera have shown that stolonization stops when the structure is re-implanted, indicating that the proventricle may inhibit stolonization while promoting regeneration (Franke & Pfannenstiel, 1984). However, no glandular structure was found in the proventricle of T. antoni, and a suggestion is that structures associated with this organ (e.g., nervous system) might be responsible for the regulation of reproductive and regenerative processes .
In effect, stolonization is favored over regeneration in an individual unable to regenerate the proventricle and vice versa. Similar results were described regarding the interactions between paratomic fission and regeneration in other annelids and flatworms (e.g., Child, 1903;Zattara & Bely, 2013). This tendency of mutual inhibition between regeneration and stolonization or fission operates probably by a comparable machinery.
Interestingly, as an error of the fragmentation process, P. halleziana exhibited a specimen with three regenerating regions, suggesting the activity of three different SAZs (Allen, 1923) ( Figure 5D). Bifurcated or branching forms occurring in non-gemmiparous species suggest that errors can arise during genetic control of regeneration and reproduction of syllids. It might be speculated that these kinds of "errors" may have led to the evolution of gemmiparity and branching species in Syllidae (Aguado, Glasby et al., 2015).

CONCLUSIONS AND FUTURE DIRECTIONS
Regenerative ability in syllids is linked to both sexual and asexual reproductive modes. During sexual reproduction by schizogamy, syllids regenerate the posterior body after release of a stolon, thereby being able to undergo another stolonization cycle. Asexual reproduction by architomic fission is known for species that can restore a complete body from fragments of the original individual. Stolonization in syllids and paratomic fission in annelids seem to share a regenerative mechanism in which mutual inhibition occurs, implying that the reproductive modes are comparable.
Studies in regeneration and fission facilitated another finding in syllid biology. Some species were able to split in specific areas during experiments of stress induction, including species that do not reproduce asexually. These events of fragmentation can be interpreted as an autotomic response rather than an architomic fission.
Considering whole-body regeneration, syllids are able to regenerate anterior and posterior ends but often are limited in restoration of certain anterior organs (e.g., proventricle). Although this limited anterior regeneration is not complete (see , from a morphological point of view, the regenerating individuals are able to reproduce since the stolonization process is induced by proventricle removal. This organ seems to play an important role in reproduction and regeneration in syllids.
Some patterns regarding the speed and ability to regenerate segments and organs can be recognized in syllids. For instance, posterior regeneration is usually faster than anterior regeneration. Restoration of a new anterior end becomes easier when the bisection plane is close to the prostomium. Additionally, some species can renew the pharyngeal armature, and also show better overall anterior regenerative ability. Implications for ensuing studies can be a comparison of the genetic mechanisms during regeneration among syllids that reproduce by different modes.
The origin of regenerative cells in syllids has never been elucidated.
Knowledge is limited to histological visualizations in different stages of regeneration, restricted to a single species. Studies of cell lineages during regeneration using fine-scale methods and techniques of labeling and histology are missing.
Heteromorphic regeneration points out errors in the regenerative process that may have occurred and been established during the evolution of branching body patterns in syllids, such as those of R. multicaudata and S. ramosa. The occurrence of aberrant forms in stages of regeneration exemplifies an intriguing possibility of investigating the underlying genetic control processes. Therefore, syllid regeneration should be considered as a good model to study the evolution and development of body axes. At present, studies about gene expression patterns during regeneration or development in syllids are still missing. Emerging genomic methods will be useful to understand this process. By sequencing genomes and transcriptomes, genes involved in regeneration can be identified (Bhambri et al., 2017;Myohara et al., 2006;Nyberg et al., 2012), thereby opening up the possibility of a functional characterization, e.g., by CRISPR or RNAi methods (Boettcher & McManus, 2015). Different studies of annelid regeneration identified the expression of genes such as engrailed, wingless (Wnt), orthodenticle homologs, markers of germline cells (e.g. piwi, vasa and nanos), and Hox genes during regenerative processes Bely & Wray, 2001;Novikova et al., 2013;Özpolat & Bely, 2015;Prud'homme et al., 2003). Furthermore, morphogens (e.g., bone morphogenetic protein) and fibroblast growth factors might be involved in nervous activation of blastema formation (Boilly, Faulkner et al., 2017;Satoh, Makanae, Nishimoto, & Mitogawa, 2016;Takeo et al., 2013).
These approaches are opportune to investigate the triggers of the regenerative response, the pathways that control growth and patterning of the new structures, and the re-establishment of antero-posterior polarity in syllids and other bilateral animals.

ACKNOWLEDGMENTS
We We acknowledge support by the German Research Foundation and the Open Access Publication Funds of the Göttingen University.