Latent Regeneration Abilities Persist Following Recent Evolutionary Loss in Asexual Annelids

Latent regeneration abilities persist following recent evolutionary loss in asexual annelids

Alexandra E. Bely1 and James M. Sikes2

Biology Department, University of Maryland, College Park, MD 20742

Edited by Douglas Futuyma, Department of Ecology and Evolution, State University of New York, Stony Brook, NY, and approved November 5, 2009 (received for review July 15, 2009) Regeneration abilities have been repeatedly lost in many animal Naidine annelids represent a promising model for investigat- phyla. However, because regeneration research has focused almost ing regeneration loss. Naidines sensu lato (Naidinae, Pristininae, exclusively on highly regenerative taxa or on comparisons between and close relatives) are a group of small aquatic oligochaetes, regenerating and nonregenerating taxa that are deeply diverged, many of which reproduce asexually by fission (13–16). The most virtually nothing is known about how regeneration loss occurs. Here, common mode of fission in this group is paratomic fission, in we show that, following a recent evolutionary loss of regeneration, which a new head and tail are intercalated in the middle of the regenerative abilities can remain latent and still be elicited. Using body within a region referred to as the fission zone, thus forming comparative regeneration experiments and a molecular phylogeny, transiently linked individuals (Fig. 1, A and B). Regenerative we show that ancestral head regeneration abilities have been lost abilities have previously been investigated in only a few species, three times among naidine annelids, a group of small aquatic worms but available data reveal important variation. Several naidine that typically reproduce asexually by fission. In all three lineages species possess excellent regenerative abilities, being capable of incapable of head regeneration, worms consistently seal the wound forming a new head and tail from just a small fragment of the butfailtoprogresstothefirst stage of tissue replacement. However, original animal (17, 18), and it is likely that regenerative ability is despite this coarse-level convergence in regeneration loss, further ancestral for naidines, as it is for annelids more broadly (3). We investigation of two of these lineages reveals marked differences in previously identified one species, however, that is incapable of how much of the regeneration machinery has been abolished. Most regenerating any anterior segments (19), suggesting at least one notably, in a species representingoneofthesetwo lineages, butnotin loss within the group. a representative of the other, amputation within a narrow prolifer- We investigated the pattern and process of regeneration fi ative region that forms during ssion can still elicit regeneration of an evolution among closely related species of naidine annelids. essentially normal head. Thus, the presence at the wound site of Comparative regeneration experiments and a molecular phy- elements characteristic of actively growing tissues, such as activated logeny together indicate multiple losses of head regeneration stem cells or growth factors, may permit blocks to regeneration to be within this group. To investigate the process of regeneration loss, circumvented, allowing latent regeneration abilities to be manifested. we assessed the developmental capabilities of two species representing independent losses of head regeneration, focusing on regeneration loss | Annelida | asexual reproduction | evolution the ability to regenerate segmental and asegmental tissue, to initiate cell proliferation after amputation, and to regenerate egeneration is a process by which many animals can replace following amputation within the fission zone. Our investigations Rlost body parts. Although basal animal lineages and some reveal a phenomenon, fission-zone regeneration, by which an bilaterians are excellent regenerators, numerous animal lineages, otherwise nonregenerating species can still regenerate, a finding including representatives of all three major bilaterian clades, have that has important implications for understanding how regen- incurred severe reductions in the ability to regenerate body parts eration abilities are lost. This study demonstrates that com- (1). Large groups such as nematodes, birds, mammals, and leeches, parative developmental investigations of regeneration among for example, are largely or entirely incapable of regenerating any close relatives represent a powerful approach for providing body part, indicating relatively old losses of regeneration. Many insight into the evolution of regeneration. other, more recent losses of regeneration have also occurred, for example, among annelids, arthropods, planarians, fishes, and liz- Results – ards (2 6). Despite longstanding interest in the process of animal We performed comparative head and tail regeneration experi- regeneration (7, 8), over a century of speculation on the root ments on 19 naidine species, representing 13 genera spanning causes of variation in this feature (1, 4, 9), and recent advances in the group, as well as two outgroups (see Table S1 for species understanding its developmental and molecular basis (10, 11), we acquisition information). Although all naidines in our trials can still know little about the evolutionary and developmental pro- reproduce by fission, we found that nearly one-third of these cesses involved in regeneration loss. species (6/19) are incapable of head regeneration (Fig. 1). For Understanding the pattern and process of regeneration loss these trials, we removed the cephalic segments, a set of anterior- requires a comparative approach that focuses on species that have most segments that are morphologically distinct [typically lacking recently lost regeneration and their regenerating close relatives. dorsal chaetae (bristles) and pigmented gut cells] and that in However, with few exceptions (e.g., 4, 6, 12), regeneration research has focused almost exclusively on regenerating species or on broad

comparisons between distantly related regenerating and non- Author contributions: A.E.B. designed research; A.E.B. and J.M.S. performed research; regenerating groups (e.g., amphibians vs. mammals) too deeply di- A.E.B. analyzed data; A.E.B. wrote the paper. verged to reveal much about the mechanism of regeneration loss. The authors declare no conﬂict of interest. What types of changes are correlated with regeneration loss? Are This article is a PNAS Direct Submission. some aspects of regeneration loss predictable? Can regenerative 1To whom correspondence should be addressed. E-mail: [email protected]. abilities re-evolve after being lost? Important questions such as 2Present address: Department of Cell and Developmental Biology, University of Illinois at these remain largely unanswered, in part because the currently Urbana-Champaign, Urbana, IL 61801. available animal regeneration models are largely inadequate for This article contains supporting information online at www.pnas.org/cgi/content/full/ addressing them. 0907931107/DCSupplemental.

1464–1469 | PNAS | January 26, 2010 | vol. 107 | no. 4 www.pnas.org/cgi/doi/10.1073/pnas.0907931107 Downloaded by guest on September 26, 2021 Fig. 1. Fission and comparative regeneration experiments in naidine annelids. In this and subsequent figures, anterior is left; dark-green and light green mark new head and tail tissue, respectively, formed by fission; dark-gray bars mark the original body region remaining after amputation; light-gray bars mark amputated tissue; orange bars mark regenerated tissue; dashed lines indicate plane of amputation. (A) Naidine paratomic fission occurs by intercalation of new head and tail tissue in the middle of the body (top: prefission; middle: fission; bottom: post-fission). (B) Pa. litoralis individual with a late-stage fission zone, showing the characteristically clear tissue of this zone. Original head is at top, pointing left. (Scale bar, 250 μm.) (C–F) Pr. leidyi can regenerate anteriorly and posteriorly (C and D: 4 days dpa), whereas Pa. litoralis can regenerate posteriorly but not anteriorly (E and F: 7 dpa). (G) Results from comparative regeneration experiments (19 naidine species, top; 2 outgroups, bottom). See Table S2 for amputation locations for each species. Color codes are based on the stage of the most advanced three individuals, scored at the end of that time period, with emergence of chaetae from new segments marking regeneration completion. Absence of regeneration scoring indicates that no individuals survived to that time point.

naidines correspond to those anterior segments that form during posterior regeneration could be induced. For the remainder of the fission (20). For brevity, we refer to these segments as “head study, we focused specifically on anterior regeneration ability. segments” and refer collectively to the cephalic segments plus To place the comparative regeneration experiments in a phy- the anterior asegmental tip of the body as the “head.” Following logenetic context, we reconstructed a molecular phylogeny for the amputation of the head (two to eight segments, depending on naidines. We infer that head regeneration ability is ancestral for the species; Table S2), most species regenerated fully within 2–6 naidines and that species incapable of head regeneration stem days (Fig. 1, C and G; Table S2; Table S3), proceeding through from three independent losses (Fig. 2). Bayesian analysis of a five- stages typical of annelid regeneration: wound healing occurred gene data set (Table S4) produced a well-resolved and strongly EVOLUTION by contraction of the severed edges of the body wall and for- supported tree in which Paranais, Chaetogaster, and Amphichaeta mation of an intact epithelium covering the wound; the severed are not each other’s closest relatives. [Previous analyses, based on gut edges sealed into a blind tube; a regeneration blastema (an more limited data sets subsumed within ours, have sometimes unpigmented mass of undifferentiated cells) formed at the recovered a clade composed of Chaetogaster and Amphichaeta, but wound site; and this blastema grew in size and ultimately dif- never one composed of Paranais, Chaetogaster, and Amphichaeta ferentiated to replace the structures that had been lost, including (14, 22).] Our inference that each of these genera represents an new segments. However, two species of Paranais (Pa. litoralis and independent loss of head regeneration is based on the assumption Pa. frici), two species of Chaetogaster (C. diastrophus and C. di- that regeneration losses are more likely than gains. This is a strong aphanus), and two species of Amphichaeta (A. “raptisae” B and expectation, given that numerous regeneration losses are evident A. “raptisae” C) sealed the wound but did not form a blastema or among metazoans, including among annelids, but no clear gains of replace any missing structures (Fig. 1, E and G; Table S2; Table regeneration have yet been documented (1, 3). S3). Follow-up experiments performed under a broad array of To determine whether species that independently lost head conditions all confirmed the inability of these species to regenerate regeneration have converged in other measures of regenerative the head. Some anteriorly amputated individuals from all three potential, we further investigated Pa. litoralis and C. diaphanus, genera continued to undergo fission during our trials, demon- representing two of the three lineages incapable of head re- strating that under the experimental conditions these species generation, and Pristina leidyi, a fully regenerating outgroup. We could form the head by fission, even though they could not re- performed two minimal anterior amputations, removing either generate this region. Nearly all species, including four of the six just the asegmental tip (prostomium and peristomium) or this tip species incapable of head regeneration, could regenerate poste- plus one anterior segment (Fig. 3). We found that neither Pa. riorly and did so in a time frame similar to that of anterior re- litoralis nor C. diaphanus can regenerate following the removal of generation (Fig. 1, D, F, and G; Table S2; Table S3). Posterior even a single head segment: as when all head segments are re- regeneration failure in the two remaining species was not further moved, all individuals wound healed (10/10 for each species), but explored in this study, and the possibility remains that under dif- none formed a blastema (0/10 for each species) (Fig. 3, G and ferent conditions (e.g., with food provided after amputation) H). However, Pa. litoralis, although not C. diaphanus, can still

Bely and Sikes PNAS | January 26, 2010 | vol. 107 | no. 4 | 1465 Downloaded by guest on September 26, 2021 persists long after the epithelium appears intact, which usually occurs within one day after amputation as in other naidines, and wound healing in C. diaphanus shows no comparable cell proliferation. Amputated C. diaphanus continue to proliferate cells at the fission zone and posterior growth zone (as do amputated Pa. litoralis), demonstrating that the failure to proliferate at the wound site in this species is not simply a consequence of energetic limitations (Fig. 3, K and N). Together, these findings suggest that the block to regeneration may occur before blastema initiation in C. diaphanus but after blastema initiation in Pa. litoralis. We investigated the possibility that head regeneration might be elicited in non-anteriorly regenerating species by amputating within the new tissue of the fission zone (Fig. 4, A and A’). We found that, following careful bisection of a developing fission zone head (removing the asegmental tip and one to two head segments), Pa. litoralis, but not C. diaphanus, can indeed regenerate a largely normal head through a process we call fission-zone-regeneration (FZ-regeneration) (Fig. 4, B and C). After such cuts, ∼73% (16/22) of Pa. litoralis developed a blastema that actively proliferated and grew in size, and nearly half of these individuals (7/16) differentiated most or all of the structures that had been amputated, including the asegmental tip and one to two new segments (Fig. 4, D–H; Table S5). Although FZ-regeneration in Pa. litoralis progressed at a slower and more variable rate than is typical for regular naidine regeneration (first sign of blastema 3–7 dpa; first segmental chaetae 4–11 dpa) and sometimes produced a head with some abnormal or missing elements (Fig.4H), FZ-regeneration clearly resulted in the replacement of structures that had been amputated. Regeneration and fission are similar processes, but FZ-regeneration is not merely a continuation of fission. The contexts of FZ-regeneration and fission differ markedly because FZ-regeneration (and regular regeneration) begins with a wound, involves terminal addition of tissue, and necessitates fi Fig. 2. Phylogenetic distribution of head regeneration ability in naidines. Open perforation of a new mouth, whereas ssion begins without a wound, circles mark species that can regenerate a head; solid circles mark species that involves intercalation of tissue, and involves modification of the cannot. Head regeneration appears to have been lost three times within one of the original gut to form the mouth. Furthermore, new head tissue two naidine clades that undergo fission (green ovals). Regeneration information is formed by FZ-regeneration (e.g., prostomium, segment 1, segment based on our comparative regeneration experiments and previously published 2) is added anterior to the remaining fission zone segment (i.e., seg- data for Ripistes parasita (21). Tree topology is from a Bayesian analysis. Nodes with ment 3), whereas, during fission, development always proceeds from a posterior probability of 0.98 or greater are black; those with a posterior proba- anterior to posterior (i.e., prostomium and segments 1 and 2 always bility less than 0.98 are gray. All posterior probabilities are 1.0 unless denoted by a form before segment 3). Pa. litoralis that were amputated at the base letter (a: 0.99; b: 0.98; c: 0.95; d: 0.89; e: 0.86; f: 0.82; g: 0.56). Maximum likelihood fi tree has an identical topology and nearly identical branch lengths, although weak of the ssion zone head, rather than within it, consistently wound bootstrap support. healed but failed to regenerate (10/10), suggesting that the nature of the wound stump (the original tissue immediately adjacent to the wound site), rather than merely close proximity to a fission zone, is regenerate the anterior asegmental tip if only this region is re- the critical factor allowing FZ-regeneration to proceed in this spe- moved (6/7 Pa. litoralis regenerated; 0/7 C. diaphanus regen- cies. C. diaphanus wound healed but failed to regenerate whether erated; Fig. 3, D and E). The regenerating species Pr. leidyi could amputation occurred within or adjacent to the fission zone (12/12 per regenerate following either type of minimal amputation (10/10 amputation type), whereas Pr. leidyi regenerated fully following ei- for each amputation treatment; Fig. 3, F and I). For brevity, in ther of these cuts (12/12 per amputation type). the remainder of this article we refer to both Pa. litoralis and To determine whether FZ-regeneration recovers gene expres- C. diaphanus as “non-anteriorly regenerating” because neither sion characteristic of regular regeneration, we investigated the species can regenerate anterior segments, even though Pa. litor- expression of a homolog of nanos, a gene expressed in somatic stem alis can still regenerate the anterior asegmental tip. cells and the posterior growth zone in annelids (23, 24). We found We also asked whether the block to regeneration in Pa. litor- that nanos in Pa. litoralis is undetectable at the (nonregenerating) anterior wound site of head-amputated individuals (Fig. 4J), yet is alis and C. diaphanus occurs before or after the initiation of post- expressed in the FZ-regeneration blastema (Fig. 4, K and L)ina amputation cell proliferation, as assayed by BrdU incorporation pattern very similar to that in the regular regeneration blastema of Pr. during DNA replication. In the fully regenerating species Pr. leidyi (Fig. 4I). In both species, nanos is expressed broadly leidyi (and other regenerating naidines), cell proliferation in- within the internal mesenchymal cells of the blastema as well as creases markedly in the epidermis near the wound site after faintly in the overlying epidermis (although in Pa. litoralis ex- – amputation, peaking 2 3 days post amputation (dpa) (Fig. 3, L pression appears less strong and does not extend to the distal tip). and O). Although neither Pa. litoralis nor C. diaphanus forms a Importantly, during blastema stages of FZ-regeneration, Pa. li- visible blastema following head amputation, we found that Pa. toralis nanos expression is weak or undetectable in the remaining litoralis, but not C. diaphanus, consistently exhibits persistent fission-zone segment that comprises the stump (Fig. 4, K and L), low-level cell proliferation at the wound site for at least 3 days contrasting with the higher expression within the blastema. Al- after amputation (Fig. 3, J, K, M, and N). Some early cell pro- though nanos is expressed in early- to midstage fission zones (in a liferation at the wound site could conceivably result from wound- pattern similar to that seen in the FZ-regeneration blastema), in- healing processes. However, cell proliferation in Pa. litoralis cluding in this future stump segment, by the time the FZ-

1466 | www.pnas.org/cgi/doi/10.1073/pnas.0907931107 Bely and Sikes Downloaded by guest on September 26, 2021 Fig. 3. Minimal amputation experiments and post-amputation cell proliferation. Body region indicators are as in Fig. 1; arrows mark the prostomium; asterisks mark the mouth. (A–C) Head morphology of uncut worms. (D–I) Following amputation of the asegmental tip, Pa. litoralis (D, 4 dpa) and Pr. leidyi (F,3 dpa) can regenerate this region but C. diaphanus merely wound heals (E, 7 dpa, showing the characteristic square anterior margin following within-head amputation in this species). Following amputation of one head segment and the asegmental tip, Pa. litoralis (G, 7 dpa) and C. diaphanus (H, 7dpa) only wound heal, but Pr. leidyi regenerates this region (I, 3 dpa; caret marks ventral chaetae of regenerated segment). (J–O) BrdU labeling on days 2 and 3 following head amputation (removing three segments in Pa. litoralis, four in C. diaphanus, and six in Pr. leidyi) indicates proliferation at the wound site in Pa. litoralis (J and M: solid arrowheads) even though no blastema will form; no proliferation at all at the C. diaphanus wound site (K and N: open arrowheads); and extensive proliferation associated with the developing blastema of Pr. leidyi (L and O: solid arrowheads). Right panels in K and N are the BrdU-labeled tails (undergoing normal growth) of the anteriorly amputated individuals pictured in the associated left panels.

regeneration blastema is developing, the remaining fission zone annelids (this study; ref. 3), malformed regenerates in fish fins (4), segment has differentiated and nanos expression within it dis- and hypomeric, nonfunctional outgrowths in amphibians (12). appears (Fig. 4, K and L). Thus, nanos expression provides evi- However, we found that species with superficially similar amputa- dence that the process of FZ-regeneration is comparable to that of tion outcomes can vary dramatically in the extent to which re- regular regeneration and that the FZ-regeneration blastema rep- generation has been abrogated. Naidines incapable of regenerating resents a developmental field distinct from that of the stump tissue. head segments all converge on a common post-amputation phe- notype, wound healing without forming a detectable blastema, yet Discussion whereas one lineage (represented by Pa. litoralis) retains the capa- Multiple losses of head regeneration have occurred among naidines, bility for anterior asegmental regeneration, cell proliferation ini- making this group a useful model in which to investigate regener- tiation, and FZ-regeneration, a second lineage (represented by C. ation loss. The recent losses revealed within this group challenge diaphanus) has lost all of these capabilities. These differences may

several widespread views regarding regeneration evolution. For be due to the latter lineage having accumulated either a greater EVOLUTION example, naidines provide the only known natural examples of ani- number of blocks or more severe blocks to regeneration. mals capable of agametic asexual reproduction (e.g., fission, bud- A major conclusion from this study is that regenerative abil- ding) in which regenerative and asexual capabilities are decoupled, ities can remain latent for a considerable period of evolutionary countering the common assumption that all agametically repro- time following their effective loss. Both Paranais species that we ducing animals can also regenerate. Although many animals capable studied have lost head regeneration ability, suggesting that this of agametic reproduction have extraordinary regenerative abilities loss dates back at least to the time of their common ancestor. (21), and asexual and regenerative developmental processes can However, we demonstrate that in Pa. litoralis, at least, replace- overlap substantially (1, 11, 25), naidines clearly demonstrate that ment of head structures can still occur by FZ-regeneration. Why regeneration is far from universal among agametically reproducing has FZ-regeneration ability been retained when regular head species, even in species, such as naidines, in which agametic re- regeneration has been effectively lost? FZ-regeneration is almost production is derived from regeneration (26). Furthermore, nai- surely ecologically irrelevant. Successful amputation within the dines join the small but growing list of examples showing that loss of fission zone head is difficult to achieve, both because this region regenerative abilities is not necessarily accompanied by large-scale is a very narrow amputation target and because stress near a changes in morphological complexity (2, 4, 5, 12). Naidines all share fission zone (especially at mid- to late-fission stages) tends to a similar body plan, and species that cannot regenerate do not differ cause a premature detachment of the posterior individual (by in any consistent way from fully regenerating species. Although in- muscular contraction at the fission plane) without accompanying creases in morphological complexity may well entrain a decreased tissue loss for either individual. Thus, it is implausible that fission regenerative ability (although even this remains speculative), re- zone amputation is more common than regular amputation. generation losses can clearly occur without any such changes. Instead, we propose that amputation within the fission zone This study demonstrates that the superficial similarity of post- yields a more permissive context for regeneration that allows the amputation phenotypes can mask important differences in regen- evolved blocks in this lineage to be circumvented. In some ver- erative potential. An emerging trend regarding regeneration loss, tebrates, regenerative abilities are greater in younger individuals and one further supported by our data, is that there is a strong than in older ones (21, 27), and FZ-regeneration in Pa. litoralis phylogenetic component to how regeneration tends to fail. Evolved represents a new example of developmental youth conferring blocks to normal regeneration tend to result in blastema failure in greater regenerative potential. These findings suggest the in-

Bely and Sikes PNAS | January 26, 2010 | vol. 107 | no. 4 | 1467 Downloaded by guest on September 26, 2021 Fig. 4. FZ-regeneration following fission-zone head bisection. Body region indicators are as in Fig. 1; dark-green and light-green bars mark new head and tail tissue, respectively, formed in the fission zone; paired vertical dashes mark the boundary between original segments and fission zone tissue; asterisks mark chaetae of remaining fission zone segment (segment 3 in Pa. litoralis). Labeling of structures: mouth (m), prostomium (pr), cerebral ganglion (cg), pharynx (ph), segment 1 chaetae (ch1), ventral ganglion of segment 1 (g1). (A–C) Fission-zone head bisection, diagrammed in A (A', Pa. litoralis late-stage fission zone), leads to replacement of head structures in Pa. litoralis (B, 8 dpa) but only wound healing in C. diaphanus (C, 6 dpa). Insets in B show the chaetae of segment 1 (left) and of the remaining fission zone segment (right) from more superficial focal planes. (D–F) FZ-regeneration in Pa. litoralis occurs by the formation of a blastema that shows extensive cell proliferation and actively grows. BrdU labeling (D, 4 dpa) is widespread throughout the blastema epidermis (inset in D is high-magnification view of chaeta from fission zone segment). Live imaging of the same FZ-regenerating Pa. litoralis individual at 7 dpa (E) and 15 dpa (F) shows blastema growth. (G and H) FZ-regeneration in Pa. litoralis (H, 10 dpa) restores internal and external structures present in a normal head (G). Specimens are labeled for acetylated α-tubulin (green, labeling peripheral nerves and cilia of the foregut, prostomium, and body wall), for serotonin (red, labeling serotonin+ nerve cells and tracts), and with phalloidin (blue, labeling the body-wall muscles to show the body contour). Note that the ventral mouth and buccal cavity are unciliated and are thus unlabeled. Segmental chaetae are autofluorescent. Inset in H shows segment 1 chaetae. Abnormalities in the FZ- regenerate in H include a misshapen pharynx, supernumerary serotonin+ ganglion cells, a laterally rotated cerebral ganglion, and absence of segment 2 (or at least its chaetae). (I) Pr. leidyi nanos is expressed throughout the anterior blastema during regular regeneration (2 dpa). (J–L) Pa. litoralis nanos is not expressed during failed regeneration after removal of two adult head segments (J, caret marks segment 3 chaetae), but is expressed in the FZ-regeneration blastema following amputation of these same segments from within the fission zone (K, 4 dpa; L, 5 dpa).

triguing possibility that there may be an ontogenetic pattern to regeneration employs developmental processes shared with regeneration loss, with regenerative ability becoming restricted to other phenomena, such as fission and budding, post-embryonic the early stages of development before being lost completely. growth, and embryogenesis (1, 11). Pleiotropies are also ex- That FZ-regeneration can occur in Pa. litoralis, which is oth- pected between regenerative processes that restore different erwise incapable of head regeneration, demonstrates that the parts of the same animal, such as head regeneration and tail morphogenetic process of head regeneration has not itself been regeneration. In naidines, regeneration and fission are clearly abolished. Instead, the block to regular regeneration appears to closely related processes (18, 26), and posterior regeneration can reside in the potential of the stump, a region that in other re- persist in non-anteriorly regenerating taxa. Thus, the head regenerating animals is known to be an important source of signals generation machinery in Paranais may remain largely functional and undifferentiated cells critical to blastema formation (11). In because much of it is shared with posterior regeneration and Pa. litoralis, regeneration does not proceed when the stump is anterior development by fission. composed of fully developed tissue, but can be elicited when the A fundamental question regarding the evolution of animal distal part of the stump includes even just a single developing regeneration is whether regenerative abilities that have been lost segment of the fission zone. Factors that may be required for can later re-evolve. Although no examples of this have yet been regeneration, such as activated stem cells and growth factors, are documented, there is accumulating evidence that lost morpho- likely to be already present in developing fission tissue but not logical traits can subsequently re-evolve, presumably by the re- within fully grown heads. Thus, we suggest that the block to re- expression of latent potential for producing those traits (28, 29). generation in Pa. litoralis lies in the activation of stem cells and/ Our study provides evidence that the potential for re-evolution or growth factors in the stump and that when these are provided of regenerative abilities is real: regenerative abilities can remain at the wound site (i.e., by amputating within the fission zone), latent after being effectively lost and can be re-elicited through regeneration can still proceed. entirely natural means. Regeneration has been at least partly What can explain the persistence of latent regenerative abil- rescued in Hydra and vertebrates through experimental manip- ities in a lineage? We propose that even after the effective loss of ulations, for example, by activating Wnt signaling, ectopically anterior regeneration from a lineage, much of the anterior expressing an ion pump, or suppressing the immune response regeneration process remains intact because of pleiotropies with (25, 30–32). Our study complements these experimental rescues other developmental phenomena. There is growing evidence that by demonstrating the possibility for a “natural rescue”: by con-

1468 | www.pnas.org/cgi/doi/10.1073/pnas.0907931107 Bely and Sikes Downloaded by guest on September 26, 2021 trolling only the position and developmental timing of amputa- Sequencing and Phylogenetic Analysis. We obtained a 5228-bp nucleotide tion, we achieved an essentially complete recovery of normal data set composed of fragments of five genes, two mitochondrial ribosomal regeneration, a process likely not seen by selection for millions of genes (12S, 16S), one mitochondrial protein-coding gene (cytochrome oxi- years. Thus, a simple morphological change in the Paranais lin- dase I), and two nuclear ribosomal genes (18S, 28S). The data set included all eage, such as an increase in the relative length of the head region taxa included in our comparative regeneration experiments and several fi that develops within a fission zone, could be all that is required to additional taxa. DNA extraction, PCR ampli cation, and sequencing were make FZ-regeneration sufficiently prevalent in the wild that performed using standard methods and primers (available from the authors upon request). GenBank accession numbers for all sequences included in our head regeneration could once again be favored by selection. The data matrix are listed in Table S4, with new sequences deposited under re-expression of retained latent regenerative abilities provides a GQ355365–GQ355466. Sequences were aligned and analyzed using both viable mechanism by which regenerative abilities that have been Bayesian and maximum likelihood approaches with methods described in effectively lost from a lineage might subsequently re-evolve. SI Methods.

Materials and Methods BrdU Incorporation, Tissue Labeling, and nanos Expression. BrdU labeling (0.1 fi Animal Acquisition. Species were obtained through eld collections or other mg/mL; 18-h pulse) and acetylated α-tubulin, serotonin, and phalloidin la- sources (Table S1). Most species were cultured in the lab to generate ma- beling were performed using standard methods (SI Methods). nanos was fi terial for experiments (SI Methods); eld-collected individuals were used isolated from Pa. litoralis and Pr. leidyi using degenerate PCR, 5′ RACE, and directly in a few cases. 3′ RACE (GenBank accession nos. GQ369728 and GQ369729). Phylogenetic analysis was performed to confirm the identity of the recovered sequences Regeneration Experiments. Comparative regeneration experiments consisted as nanos-class genes (SI Methods; Fig. S1). Expression was investigated of an anterior amputation treatment, a posterior amputation treatment, and through in situ hybridization using a previously published protocol (26). an uncut control (n ≥ 10 individuals/treatment for most species; Table S3). Amputation positions were tailored for each naidine species to represent a Imaging. Samples were imaged with a Zeiss Axioplan 2 epifluorescence comparable minimum amputation challenge, specifically removing body regions formed through fission in each species (Table S2, SI Methods). For microscope or a Zeiss LSM-510 confocal microscope. Confocal image capture minimal anterior amputation experiments, we removed either just one head and processing are described in SI Methods. segment and the asegmental tip or only the asegmental tip. For fission-zone amputations, either the developing head of the fission zone was bisected, ACKNOWLEDGMENTS. We thank K. Avila, S. Desiraju, D. Freund, L. Krass, removing the asegmental tip and one or more developing head segments M. Hertz, E. Milne, L. Potts, B. Walker, and C. Ward for assistance with com- (leaving a stump composed of at least one segment from the fission zone), parative regeneration experiments; L. Shapiro for assistance with phyloge- or the cut was made at the base of the head, removing the asegmental tip netic analyses; E. Zattara for tubulin–serotonin–phalloidin labeling and and all developing head segments (leaving a stump consisting entirely of old imaging; D. Sanger and F. Holland for Monopylephorus samples; and L. Sha- segments of the original worm). For these amputations, we used animals at piro, E. Abouheif, K. Nyberg, E. Zattara, and two anonymous reviewers for mid- to late-stage fission, when segments have formed (as evidenced by the comments on the manuscript. This work was funded by National Science Foun- emergence of segmental chaetae) but the head is still actively growing. See dation grant IOB-0520389 (A.E.B.) and a University of Maryland General Re- SI Methods for details of amputation procedures. search Board Award (A.E.B).

1. Bely AE, Nyberg KG (2009) Evolution of animal regeneration: re-emergence of a field. 18. Berrill NJ (1952) Regeneration and budding in worms. Biol Rev Camb Philos Soc 27: Trends Ecol Evol (doi:10.1016/j.tree.2009.08.005). 401–438. 2. Arnold EN (1984) Evolutionary aspects of tail shedding in lizards and their relatives. 19. Bely AE (1999) Decoupling of fission and regenerative capabilities in an asexual J Nat Hist 18:127–169. oligochaete. Hydrobiologia 406:243–251. 3. Bely AE (2006) Distribution of segment regeneration ability in the Annelida. Integr 20. Hyman LH (1940) Aspects of regeneration in annelids. Am Nat 74:513–527. Comp Biol 46:508–518. 21. Vorontsova MA, Liosner LD (1960) Asexual Propagation and Regeneration (Pergamon 4. Wagner GP, Misof BY (1992) Evolutionary modification of regenerative capability in Press, London). vertebrates: a comparative study on teleost pectoral fin regeneration. J Exp Zool 261: 22. Bely AE, Wray GA (2004) Molecular phylogeny of naidid worms (Annelida: Clitellata) 62–78. based on cytochrome oxidase I. Mol Phylogenet Evol 30:50–63. 5. Vollrath F (1990) Leg regeneration in web spiders and its implications for orb weaver 23. Dill KK, Seaver EC (2008) Vasa and nanos are coexpressed in somatic and germ line phylogeny. Bull Br Arachnol Soc 8:177–184. tissue from early embryonic cleavage stages through adulthood in the polychaete EVOLUTION 6. Kolmayer S, Stéphan-Dubois F (1960) Néoblastes et limitation du pouvoir de régénération Capitella sp. I. Dev Genes Evol 218:453–463. céphalique chez la planaire Dendrocoelum lacteum. J Embryol Exp Morphol 8:376–386. 24. Rebscher N, Zelada-González F, Banisch TU, Raible F, Arendt D (2007) Vasa unveils a 7. Morgan TH (1901) Regeneration (Macmillan, London). common origin of germ cells and of somatic stem cells from the posterior growth 8. Dinsmore CE, ed (1991) A History of Regeneration Research: Milestones in the zone in the polychaete Platynereis dumerilii. Dev Biol 306:599–611. Evolution of a Science (Cambridge University Press, Cambridge, UK). 25. Lengfeld T, et al. (2009) Multiple Wnts are involved in Hydra organizer formation and 9. Goss RJ (1992) The evolution of regeneration: Adaptive or inherent? J Theor Biol 159: regeneration. Dev Biol 330:186–199. 241–260. 26. Bely AE, Wray GA (2001) Evolution of regeneration and fission in annelids: insights from 10. Sánchez Alvarado A, Tsonis PA (2006) Bridging the regeneration gap: Genetic insights engrailed- and orthodenticle-class gene expression. Development 128:2781–2791. from diverse animal models. Nat Rev Genet 7:873–884. 27. Wallace H (1981) Vertebrate Limb Regeneration (John Wiley & Sons, New York). 11. Brockes JP, Kumar A (2008) Comparative aspects of animal regeneration. Annu Rev 28. Kohlsdorf T, Wagner GP (2006) Evidence for the reversibility of digit loss: a Cell Dev Biol 24:525–549. phylogenetic study of limb evolution in Bachia (Gymnophthalmidae: Squamata). 12. Scadding SR (1981) Limb regeneration in adult amphibia. Can J Zool 59:34–46. Evolution 60:1896–1912. 13. Erséus C (2002) 18S rDNA phylogeny of the Tubificidae (Clitellata) and its constituent 29. Collin R, Miglietta MP (2008) Reversing opinions on Dollo's Law. Trends Ecol Evol 23: taxa: Dismissal of the Naididae. Mol Phylogenet Evol 22:414–422. 602–609. 14. Envall I, Källersjö M, Erséus C (2006) Molecular evidence for the non-monophyletic 30. Kawakami A, et al. (2006) Wnt/beta-catenin signaling regulates vertebrate limb status of Naidinae (Annelida, Clitellata, Tubificidae). Mol Phylogenet Evol 40:570–584. regeneration. Genes Dev 20:3232–3237. 15. Erséus C, Wetzel MJ, Gustavsson L (2008) ICZN rules: a farewell to Tubificidae 31. Adams DS, Masi A, Levin M (2007) H+ pump-dependent changes in membrane (Annelida, Clitellata). Zootaxa 1744:66–68. voltage are an early mechanism necessary and sufficient to induce Xenopus tail 16. Brinkhurst RO, Jamieson BGM (1971) Aquatic Oligochaeta of the World (Oliver & regeneration. Development 134:1323–1335. Boyd, Edinburgh). 32. Fukazawa T, Naora Y, Kunieda T, Kubo T (2009) Suppression of the immune response 17. Hyman LH (1916) An analysis of the process of regeneration in certain microdrilous potentiates tadpole tail regeneration during the refractory period. Development 136: oligochaetes. J Exp Zool 20:99–163. 2323–2327.

Bely and Sikes PNAS | January 26, 2010 | vol. 107 | no. 4 | 1469 Downloaded by guest on September 26, 2021