&p.1:Abstract Crinoid echinoderms can provide a valuableexperimental model for studying all aspects of regenera-tive processes from molecular to macroscopic level. Re-cently we carried out a detailed study into the overallprocess of arm regeneration in the crinoid Antedon me-diterranea and provided an interpretation of its basicmechanisms. However, the problem of the subsequentfate of the amputated arm segment (explant) once isolat-ed from the animal body and of its possible regenerativepotential have never been investigated before. The armexplant in fact represents a simplified and controlled re-generating system which may be very useful in regenera-tion experiments by providing a valuable test of our hy-potheses in terms of mechanisms and processes. In thepresent study we carried out a comprehensive analysis ofdouble-amputated arm explants (i.e. explants reamputat-ed at their distal end immediately after the first proximalamputation) subjected to the same experimental condi-tions as the regenerating donor animals. Our resultsshowed that the explants undergo similar regenerativeprocesses but with some significant differences to thosemechanisms described for normal regenerating arms. Forexample, whilst the proximal-distal axis of arm growth ismaintained, there are differences in terms of the recruit-ment of cells which contribute to the regenerating tissue.As with normal regenerating arms, the present work fo-cuses on (1) timing and modality of regeneration in theexplant; (2) proliferation, migration and contribution ofundifferentiated and/or dedifferentiated/transdifferentiat-ed cells; (3) putative role of neural growth factors. Theseproblems were addressed by employing a combination ofconventional microscopy and immunocytochemistry.Comparison between arm explants and regenerating

arms of normal donor adults indicates an extraordinarypotential and regenerative autonomy of crinoid tissuesand the cellular plasticity of the phenomenon. &bdy:

Introduction

Regeneration is widespread in the animal kingdom but inspite of the infinite choice of models, this phenomenonhas been explored in detail only in a few taxa. Even inthe many groups well known for their regenerative capa-bilities there is a substantial gap in our understanding notonly concerning regenerative mechanisms at the cellularand molecular level but also more fundamental aspectssuch as temporal and spatial conditions. This is certainlythe case for echinoderms whose spectacular regenerativephenomena often related to asexual reproduction(Mladenov 1996; Mladenov and Burke 1994) have at-tracted developmental biologists in the past but, save fora few notable exceptions, have been disregarded in morerecent times.

Among echinoderms, crinoids are well known fortheir striking regenerative potential (Perrier 1873; Reich-ensperger 1912; Amemiya and Oji 1992). In the past fewyears we have studied in detail the overall process of armregeneration in the crinoid Antedon mediterranea by em-ploying experimentally-induced regeneration of differentstages (Candia Carnevali et al. 1989, 1993, 1995, 1996,1997, 1998; Bonasoro et al. 1995, 1997, 1998; CandiaCarnevali and Bonasoro 1994, 1995). The majority ofour data so far highlight the fundamental aspects of thephenomena and suggest that this still largely unexploredechinoderm model could benefit from a more moderndevelopmental biology approach.

The process of arm regeneration covers a period ofabout 4 weeks and can be divided schematically intothree main phases: an initial repair phase, including thefirst 24 h post-amputation period, an early regenerativephase, including the 24 h72 h post-amputation period,and an advanced regenerative phase, covering the 72 h4weeks post-amputation period. These phases have been

Edited by D. Tautz

M.D. Candia Carnevali ()) F. BonasoroDipartimento di Biologia, Via Celoria 26, I-20133 Milano,ItalyM. Patruno M.C. ThorndykeDepartment of Biology, Royal Holloway University of London,Egham, Surrey TW20 OEX, UK &/fn-block:

Dev Genes Evol (1998) 208:421430 Springer-Verlag 1998

O R I G I N A L A RT I C L E

&roles:M.D. Candia Carnevali F. Bonasoro M. PatrunoM.C. Thorndyke

Cellular and molecular mechanisms of arm regenerationin crinoid echinoderms: the potential of arm explants

&misc:Received: 9 March 1998 / Accepted: 5 June 1998

formulated by reference to a standard model of regenera-tion resulting from self-induced amputation (CandiaCarnevali et al. 1989, 1993). In all the phases a key role isperformed by the brachial nerve and the coelomic canals,all of which are involved in cell proliferation and migra-tion. Different populations of migratory undifferentiated(amoebocytes and coelomocytes) and differentiated ele-ments (phagocytes and granule cells) can be distin-guished and are employed extensively in both the repairand regenerative processes. Our findings have shown armregeneration in Antedon to be a typical blastemal phe-nomenon, that is an epimorphic process comprising twoimportant components: (1) source and proliferation sitesof new cells, and (2) intervention of putative growth fac-tors. Both these points have been explored at differentlevels and significant results have been recently obtained(Holland 1994; Candia Carnevali et al. 1995, 1996, 1997,1998; Bonasoro et al. 1995, 1998).

In this study we have explored the problem of the fateand possible regenerative potential of the amputated armsegments isolated from the animal body. In other echino-derms amputated arm segments can display striking re-generative capabilities (a famous case is represented bythe comet forms of the starfish Linckia guildingi;Mladenov and Burke 1994): this point has never beforebeen taken into account in crinoids and seems to be acrucial test of the regenerative potential of these animals,in general, and the regenerative autonomy of their armsystem, in particular. Taking advantage of the exception-al amenability of feather stars from an experimentalpoint of view, we have carried out a parallel analysis onboth the regenerating arms and the respective amputatedarm segments (explants). The isolated explants can beeasily maintained in living condition for a relatively longtime (more than 2 weeks) and unexpectedly undergo re-generative processes comparable to those already de-scribed for donor regenerating arms which remain in-situ. This paper focuses in particular on double-amputat-ed arm explants (i.e. arm segments reamputated at theirdistal end immediately after the first proximal amputa-tion) and on their regenerative mechanisms. Uniquely,the explant represents a simplified, controlled and tract-able experimental model system which can provide in-despensable correlative information as well as confirma-tion of the regenerative mechanisms and autonomy ofthis system in terms of its cellular and molecular poten-tial. In particular, those specific aspects already exploredin standard arm regeneration (point 1 and 2, above) havenow been specifically addressed in explants by employ-ing a combination of conventional microscopy (light,transmission electron and confocal) and immunocyto-chemistry. The crucial problem of identification of celllineages responsible for both repair and regenerative pro-cesses has been approached by employing specific meth-ods for monitoring cell proliferation which we estab-lished in standard regenerating samples (Candia Carnev-ali et al. 1995, 1997; Bonasoro et al. 1998). In parallel,we have also carried out a series of tests focusing on thepresence and pattern distribution of regulative molecules

in the regenerating arm explants. It is relevant to pointout that our previous studies have revealed the presenceand respective pattern distribution of three differentclasses of regulatory molecules in standard regeneratingarms: monoamine neurotransmitters (Bonasoro et al.1995; Candia Carnevali et al. 1996), neuropeptides (Bo-nasoro et al. 1995; Candia Carnevali et al. 1998) and, fi-nally, growth factors (Candia Carnevali et al. 1998). Thisproblem has only been approached in a preliminary fash-ion in the regenerating explants by employing a series oftests selected on the basis of the quality of the resultspreviously obtained in standard regenerating arms. Inparticular the present experiments have been carried outon neuropeptides (S1 and S2 SALMFamides; Elphick etal. 1991a, b) and a growth factor (TGF-); this choicewas determined by the wide and well-defined distribu-tion of these substances seen in both normal non-regen-erating and regenerating arms.

Our results highlight some significant differences interms of basic mechanisms and the elements involvedcompared to those described for the normal regeneratingadult.

Materials and methods

Experimental animals and explant culture

Specimens of Antedon mediterranea collected by scuba diversfrom the southern coast of Italy (Gulf of Taranto) were maintainedin aquaria with a closed artificial seawater system at 14C. Experi-mental regeneration of arms was induced by mimicking the condi-tions of natural autotomy (see Candia Carnevali et al. 1993). Theamputated arm segments (explants) were immediately reamputat-ed at their distal end and maintained in little compartments of theaquaria together with the respective donor animals and under theexperimental conditions described above. The artificial seawateremployed was not sterilized. The isolated explants could be easilymaintained in a living condition for 2 weeks or more and under-went regenerative processes in parallel with their donor regenerat-ing arms. The mortality of the explants was very low but slightlyincreased according to the culture time. Regeneration was moni-tored at fixed times [24, 48, 72 h, 7 and 15 days post amputation(p.a.)] in both the explants and the donor arms. These sampleswere prepared according to the procedures outlined below.

Light microscopy (LM) and electron microscopy (TEM)Explants and regenerating donor arm were prefixed with 2% glu-taraldehyde in 0.1 M cacodylate buffer for 45 h, then, after anovernight washing in the same buffer, postfixed with 1% osmic ac-id in the same buffer. After standard dehydration in an ethanol se-ries, the samples were embedded in Epon-Araldite 812. The semi-thin and thin sections, cut with a Reichert Ultracut E (diamondknife), were stained by conventional methods (crystal violet-basicfuchsin for LM, uranyl acetate and lead citrate for TEM) and thenobserved in a Jenaval light microscope and Jeol 100 SX electronmicroscope respectively.

Immunocytochemistry (ICC)The samples were fixed in paraformaldehyde 4% - glutaraldehyde0.1% in 0.1 M phosphate buffer for 2 h. Following an overnightwash in the same buffer, the samples were dehydrated and embed-ded in Epon-Araldite (see above). This fixation and embedding

422

protocol maintains good tissue integrity and provides good preser-vation of antigenicity (see Candia Carnevali et al. 1995, 1997). Italso allows preparation of both semi-thin sections for LM and ul-tra-thin sections for TEM. Semi-thin sagittal sections cut with anLKB Ultratome V and Reichert Ultracut E were processed for im-munocytochemistry (see below).

BrdU labelling

Cell proliferation was monitored using in vivo incorporation of thesubstituted nucleotide, 5-bromodeoxyuridine (BrdU), then later re-vealed by a monoclonal antibody against BrdU (Amersham: Cellproliferation kit). Animals were immersed in BrdU dissolved inartificial seawater at a final concentration of 0.05% for the final2 h of the prefixed regeneration periods. This incubation protocolallowed detection of cells actively proliferating immediately be-fore fixation (Gratzner 1982; Candia Carnevali et al. 1995, 1997;Bonasoro et al. 1998). Control samples were obtained from non-regenerating arms of the same animals. For use with semi-thinEpon-Araldite sections, the standard BrdU-immunocytochemistryprotocol for paraffin sections was modified as described in detailelsewhere (Candia Carnevali et al. 1995, 1997). After a brief treat-ment (2 min) with a resin-remover mixture (methanol, propyleneoxide and KOH), the sections were rinsed with methanol, thenwith phosphate-buffered saline (PBS) and incubated overnight at4C with anti-BrdU serum diluted 1:100 with nuclease (Amers-ham: Cell proliferation kit). A pre-treatment of 20 min with 0.3%H2O2 in PBS was performed to exclude the potential activity ofendogenous peroxidases. After several washings in PBS the speci-mens were incubated (3 h) with peroxidase anti-mouse IgG(Amersham: Cell proliferation kit) at room temperature and, aftera further washing in PBS, incubated (5 min) with 0.05% 3,3-di-aminobenzidine and 0.03% H2O2 in PBS, and then washed in dis-tilled water. To amplify the peroxidase reaction product, the exper-iments included use of the cobalt and nickel intensifier suppliedwith the kit. Control reactions were carried out by omitting theprimary antiserum.

The same fixation/embedding protocol described for LM wasemployed for TEM. Ultra-thin sections were cut with a ReichertUltracut E and collected on gold grids. After rapid etching withNaOH 0.15% in absolute methanol (1 min), sections were careful-ly soaked on drops of methanol (30 min). The grids were washedrepeatedly with distilled water (30 min) and soaked on drops ofTRIS-buffered saline (TBS) containing 1% bovine serum albumin(BSA) (pH 7.2, 30 min). Incubation with the anti-BrdU murine an-tibody diluted 1:100 with nuclease (Amersham: Cell proliferation

kit) was performed for 1 h at room temperature. After repeatedwashing with TBS (pH 7.2, 30 min), with TBS/BSA 0.2%(pH 7.2, 1 min) and TBS/BSA 1% (pH 8.2, 5 min) sections wereincubated with goat anti-mouse IgG conjugated to 10-nm colloidalgold particles (Sigma; 1:30) in TBS/BSA 1% (pH 8.2) for 45 min,at room temperature. After repeated washing with TBS/BSA 0.2%(pH 7.2), TBS (pH 7.2) and distilled water the specimens werepost-fixed with glutaraldehyde 2.5% in distilled water for a fewminutes and then washed carefully in distilled water. Finally thesections were stained with aqueous uranyl acetate and observedwith a Jeol 100 SX electron microscope.

Peptide and TGF- labellingICC tests were carried out on resin sections according to the spe-cific immunoperoxidase ABC system (Vector) or immunofluores-cence methods. For neuropeptide ICC-specific polyclonal antibod-ies anti-S1 (BLV) and anti-S2 (BGII; Elphick et al. 1991a, b) wereused. For growth factors ICC commercial mouse anti-TGF- (Se-rotec) was used. For single labelling, standard ICC methods usingeither fluorescein or Texas red-conjugated secondary antibodywere employed (Candia Carnevali et al. 1995; Moss et al. 1998).

For double labelling both BrdU and S1- or S2- SALMFamideneuropeptide, the ABC avidin-biotin fluorescence system wasused with some modification. Processing for the first antiserumwas as normal, using fluorescein-conjugated avidin to visualizethe reaction. The second antibody was diluted in PBS and 0.1%Tween to block further any crossreactivity between the antisera,followed by visualization with Texas red-conjugated avidin D. An-ti-BrdU was applied after anti-S1 to avoid any possibility of theDNase reagent affecting peptide antigenicity. The sections wereobserved in a Leica TCS NT confocal microscope.

ICC controls were: (1) preabsorption with the appropriate anti-gen, (2) replacing the primary antiserum with non-immune sera ofthe animal in which the primary was raised, or (3) omitting the pri-mary antibody and incubating in PBS and 1% normal goat serum.

Results

The isolated explant, immediately reamputated at its dis-tal end and maintained in living conditions for 1 or 2weeks, underwent regenerative processes similar to thoseof its respective donor arm. However, since it was char-

423

Fig. 1 Schematic drawing il-lustrating the experimentalpreparation of the explant. Onthe left is shown the donor arm,on the right the respective armexplant. The explant, immedi-ately reamputated at its distalend, undergoes regenerativeprocesses in a distal directionin parallel with its donor arm &/fig.c:

424

acterized by two symmetrical amputation surfaces, prox-imal and distal (Fig. 1), these two amputation surfaceswere analysed in parallel at different stages in terms ofprocesses and cellular elements involved.

Distal amputation

In terms of regenerative stages and general histology theresults were comparable to those obtained for the stan-dard regenerating arm. The initial repair phase of2448 h p.a. was followed by an early regenerative phasecharacterized by the growth of the typical regenerativeblastema (72 h p.a.). At 1 week p.a. (Fig. 2b) the regen-erative bud was well developed although its growth wasslightly delayed when compared with a standard regener-ating arm at the same age p.a. In the bud (Fig. 2d) typicalhistological components, such as the ambulacral epitheli-um, skeletal elements, coelomic compartments (hydroco-elic and somatocoelic) and the brachial nerve, were de-veloping and differentiating according to the modalitiespreviously described for normal regenerating arms (Can-dia Carnevali et al. 1993). All these structures are contin-uous with those of the stump: in particular brachial nerveregrowth was characterized by its appreciable bendingtowards and inside the regenerative bud (Fig. 2b,d). Atthe same time the regenerating coelomic canals grew dis-tally and developed in the bud in the form of solid cordsof coelomocytes. The internal cavity was seen later fol-lowing the appearance of a split in the coelomocyte cord.The brachial nerve and coelomic canals are preferentialpathways for extensive cell migration in a distal direc-tion. This took place both in the distal regenerate itselfand the intermediate stump. The various cell types iden-tified (Fig. 3a,b,c,d) were the same as those already de-scribed for standard arm regeneration: amoebocytes, coe-lomocytes, phagocytes and granule cells (Smith 1981).The amoebocytes (Fig. 3a) are apparently undifferentiat-ed non-neural cells, normally located around the brachial

425

Fig. 3ad Microscopic anatomy of the arm explant at 1 week p.a.TEM details of the migrating cells involved in repair and regenera-tion. a amoebocyte (bar 2 m); b phagocyte (bar 2 m); c granulecells (bar 3 m); d coelomocytes (bar 4 m) &/fig.c:

Fig. 2af Microscopic anatomy of the arm explant at 1 week postamputation (p.a). a Light microscope (LM) semi-thin sagittal sec-tion at the level of the proximal amputation. The amputation sur-face is covered by a complete and thick cicatricial layer (arrows).There is no sign of a regrowing blastema (ae ambulacral epitheli-um, cc coelomic canals, m muscle, n brachial nerve, bar 200 m).b LM semi-thin sagittal section at the level of the distal amputa-tion. A prominent regenerating bud (rb) is recognizable on the am-putation surface. Its developmental stage corresponds to that of astandard regenerating arm of 45 days. The regenerative regrowthof the coelomic canals and the brachial nerve is outlined by amarked cellular flux towards the bud region (ae ambulacral epithe-lium, cc coelomic canals, m muscle, n brachial nerve, bar200 m). c LM view of the proximal amputation surface in semi-thin sagittal section. Detail of the cicatricial layer. The unusualthickness of this layer mainly is due to the overlapping of conflu-ent flows of migrating cells originating from the brachial nerve(arrows) and the coelomic canals (double arrowheads) respective-ly (bar 50 m). d LM view of the distal amputation. Detail of theregenerative bud. The regenerate largely comprises a blastema ofundifferentiated elements (rb) inside which some elements are dif-ferentiating. The coelomic system is developing as solid cords ofproliferating coelomocytes which split to produce the internal cav-ity (ae ambulacral epithelium, hc hydrocoelic canal, sc somatoco-elic canal, bar 50 m). e LM sagittal semi-thin section of the in-termediate stump. Detail of a muscle. All the muscle bundle is in-volved in a massive histological and cellular rearrangement. Its pe-ripheral regions, in particular, are characterized by the presence ofapparently dedifferentiating myocytes at different stages (arrows)intermingled with a number of coelomocytes and phagocyteswhich form areas of intense cell proliferation/migration (doublearrowheads) in the adjacent coelomic canal (cc) (bar 100 m).f Transmission electron microscope (TEM) section of the interme-diate stump at the level of the rearranging muscle. Detail of a pres-umptive dedifferentiating myocyte (dm). A phagocyte (ph) and acoelomocyte (c) are also shown (bar 2 m). &/fig.c:

s

nerve, whereas the coelomocytes (Fig. 3d) are also ap-parently undifferentiated elements, but move freely inthe coelomic canals (Endean 1966). These two types ofcell migrate towards the amputation site and are em-ployed extensively in both repair and regenerative pro-cesses. Phagocytes (Fig. 3b) and granule cells (Fig. 3c)are separate classes of well-differentiated migratorycells, the first characterized by a number of phagosomes,the second by large chromatophilic granules. These lattercells are associated with both the brachial nerve and thecoelom and are employed specifically during the repairphase.

Proximal amputation

Here, regeneration stopped at the first repair stage. At2448 h p.a. wound healing was completed and the am-putation surface was covered by a well-developed cica-tricial layer (Fig. 2a). The brachial nerve and the coelo-mic canals of the stump appeared to be involved in sub-stantial cell migration in a proximal direction. This re-sulted in progressive symmetrical cellular fluxes that inthe brachial nerve tended to diverge towards the ambul-acral region, whereas in the coelom their direction wastowards the brachial nerve (Fig. 2a). These phenomenabecame even more evident at more advanced stages(1 week, Fig. 2c) with particular reference to the severedends of the coelom which proliferated as prominentdense rods of cells curved towards the brachial nerve.This massive invasion of migrating cells produced a verythick and multistratified cicatrial layer (Fig. 2c) givingrise sometimes to local hyper-thickenings. The cells in-volved in these phenomena were the same migrating ele-ments described above. There was no blastema growingin the distal-proximal direction.

Although the general tissue histology and ultrastruc-ture was well preserved for the entire explant length, inthe intermediate tract of the stump some tissues showeda degree of rearrangement, particularly the connectivetissue, the ambulacral epithelium and the muscles(Fig. 2e). The latter appeared to undergo significant reor-ganization and perhaps dedifferentiation (Fig. 2f) whichinvolved all muscle bundles, particularly in their periph-eral regions, where the presence of elements at differentdegrees of rearrangement was evident at both histologi-cal and ultrastructural levels. These same regions werecharacterized by large numbers of migrating coelomocy-tes and phagocytes (Fig. 2e). Masses of coelomocyteswere also seen gathering and accumulating locally in thecoelomic canals of areas adjacent to the rearrangingmuscles (Fig. 2e).

In order to identify the source, proliferation sites andrecruitment times of the new cells employed in both re-pair and regenerative phases, cell proliferation was moni-tored by employing the well-established BrdU methodalready used successfully in normal regeneration stages(Candia Carnevali et al. 1995, 1997; Bonasoro et al.1998). At the proximal amputation site, in whichever

426

stage was analysed (from 24 h to 1 week p.a.) strong la-belling of nuclei was localized to the cicatricial layer(Fig. 4a) particularly in association with the proximalends of the brachial nerve and the coelom. As expected,a comparable distribution of labelling could be seen inthe distal amputation zone during the repair phase(2448 h p.a.). As in normal regenerating arms, duringthe following regenerative phases (72 h and 1 week p.a.)a particularly intense reaction could be found in the api-cal blastema and in the regrowing coelomic compart-ments (Fig. 4b). In advanced stages, a marked labellingwas also still detectable in the stump, even far from theamputation site, at the level of both the coelomic epithe-lium and the brachial nerve (Fig. 4d). It is relevant thatthe labelling involved migrating amoebocytes specifical-ly and only rarely phagocytes and granule cells. The em-ployment of specific methods of immunogold labellingfor BrdU in TEM is useful for identifying the variouscell types involved in proliferation. As expected, an in-tense BrdU reaction could be detected in the nuclei ofblastemal cells, migrating amoebocytes, coelomocytesand coelothelial cells (Fig. 4e). In contrast to normal re-generation, many strongly labelled nuclei were also not-ed at the level of the muscle bundles of the stump(Fig. 4c) corresponding to the areas of extensive cell re-arrangement. Labelled nuclei involved both migratingcoelomocytes and presumptive dedifferentiating myo-cytes (Fig. 4f).

Taking into account the primary role of the nervoussystem widely demonstrated in normal regeneratingarms, particular attention has been given to the presenceof neurally-derived factors in the explants. Preliminaryresults were obtained for the native echinoderm S1- andS2-SALMFamide peptides. These peptides are normallypresent in non-regenerating arms in various components

Fig. 4af Cell proliferation in the arm explant at 1week p.a. BrdUimmunocytochemistry. a LM semi-thin sagittal section at the levelof the proximal amputation. A marked labelling is detectable inthe cicatricial layer and at the level of the brachial nerve, the coe-lomic lining and the ambulacral epithelium. The muscle bundlesare also involved in the reaction. ABC immunocytochemistry (aeambulacral epithelium, cc coelomic canals, m muscle, n brachialnerve, bar 200 m). b LM semi-thin sagittal section at the level ofthe distal amputation. Detail of the regenerating blastema (rb). Astrong immunoreaction involves the blastemal cells and the epithe-lium of the regrowing coelomic canal (cc). ABC immunocyto-chemistry (bar 100 m). c LM semi-thin sagittal section at thelevel of the intermediate stump. Detail of the muscle. A marked la-belling can be seen in the entire bundle, with particular referenceto its peripheral regions. The epithelium of the adjacent coelomiccanal (cc) is also strongly reactive. ABC immunocytochemistry (nbrachial nerve, bar 50 m). d LM semi-thin sagittal section at thelevel of the intermediate stump. Detail of the brachial nerve. Thelabelling involves many cells of the cortex. ABC immunocyto-chemistry (bar 50 m). e TEM detail of the coelomic epitheliumof the stump showing the strong specific reaction for BrdU at thelevel of the nuclei. Immunogold labelling (cl cilium, bar 1 m).f TEM detail of a rearranging muscle in the stump showing theBrdU reaction in the nucleus of a presumptive dedifferentiatingmyocyte. The remains of the contractile apparatus are still recog-nizable (arrows). Immunogold labelling (bar 1 m) &/fig.c:

427

of the nervous system (brachial nerve and basiepithelialnerve plexuses of both the ambulacral epithelium and thecoelothelium) with only a few minor differences. Duringregeneration the pattern of reactivity of these peptides isre-established in parallel with the regeneration of thenervous system (Bonasoro et al. 1995; Candia Carnevaliet al. 1998). This also occurred in the explants where thereaction for both peptides was more or less comparableto that shown by the respective regenerating donor arms.The immunoreaction was particularly evident during theadvanced regenerative stages when the nerve began to re-grow in the distal regenerating bud. In the explants of 1week (Fig. 5a,b) labelling for both S1 and S2 was partic-ularly strong in the brachial nerve and was also apprecia-bly intense in the basiepithelial nerve plexuses of boththe ambulacral epithelium and the coelothelium, espe-cially in the regenerate. The employment of doublestaining techniques for S1 and BrdU (Fig. 5b) allowed us

to discriminate neural elements from proliferating cellsin the nervous system. Whichever method was em-ployed, there was no significant immunoreactivity in theblastema for peptides. The role of putative growth fac-tors was explored in a preliminary fashion with particu-lar reference to TGF-. According to our previous resultsTGF-, or at least an antigen which cross-reacts positive-ly with specific antisera against this factor, is not onlynormally present in the different components of the ner-vous system in normal non-regenerating arms, but is sig-nificantly involved in regeneration, with a markedly en-hanced and diffuse reaction in both cells and processesof the nervous system and at the level of the blastema it-self (Candia Carnevali et al. 1998). In the explants, al-though the reaction for TGF-, even in the advanced re-generative stages, was generally weaker and less diffusethan in normal regenerating arms, a significant labelling(Fig. 5c) could be observed at the level of the distal am-putation, especially in the advanced regenerative stages(2 weeks) and involved both nervous system components(nerve processes and granule cells) and the distal regen-erative blastema.

Discussion

Our present results show that the crinoid explant is po-tentially a valuable model for studying regenerativemechanisms at many levels. It is a system endowed withstriking autonomy and is able to manage and control ex-tensive repair and regenerative processes utilizing bidi-rectional phenomena of cell proliferation and migration.BrdU incorporation confirms that in explants the overallrepair/regeneration process is due to extensive cell pro-liferation at preferential sites. As in normal regeneratingarms, these are the terminal blastema and, even distant

428

Fig. 5ac Growth factors in arm explants at 12 weeks p.a. S1-and S2- SALMFamide-neuroptides and TGF- immunocytochem-istry. a Confocal fluorescence image of semi-thin sagittal sectionof the intermediate stump (1 week p.a.). Detail of the brachialnerve. A positive immunoreaction for S2-neuropeptide specificallyinvolves neural cells and processes. Secondary antibodies conju-gated with Texas red (bar 20 m). b Semi-thin sagittal section ofthe distal regenerating bud (1 week p.a.). Detail of the coelotheli-um (ce) and of the ambulacral epithelium (ae). Double staining forS1-neuropeptide (Texas red-conjugated secondary antibody) andBrdU (FITC-conjugated secondary antibody). The immunoreac-tion for the peptide involves scattered cells and processes of thebasiepithelial nerve plexuses (orange) of both the coelotheliumand the ambulacral epithelium. Many cells of the coelomic epithe-lium are also strongly reactive for BrdU (green) (bar 20 m). cLM semi-thin sagittal section at the level of the distal amputation(2 weeks p.a.). ABC immunocytochemistry for TGF-. The detailshows some cells positive for TGF- (arrows) at the level of theregenerating blastema (rb) and the brachial nerve (n, cc coelomiccanals, bar 50 m) &/fig.c:

from the amputation zone, the coelomic epithelium andbrachial nerve. Apart from the blastema, the primarysources of new cells are, therefore, the major continuousstructures along the arm. The two main cell componentswhich contribute to the regenerate seem to have a differ-ent derivation: the blastemal cells (and all blastemal-de-rived cells) from amoebocytes, whereas the coelomiccells arise from the migratory coelomocytes which intheir turn derive from proliferation of the coelomic epi-thelium. In contrast to normal regenerating arms, howev-er, the recruitment of new cells contributing to the regen-erating tissues also involves the rearrangement of differ-entiated tissues of the stump, in particular the muscles. Itis not clear at present if this represents direct dedifferen-tiation of myocytes to give new migrating coelomocytesor if it is a more complex phenomenon mediated byphagocytes, in which the myocytes are only passively in-volved as sources of raw materials for the production ofnew coelomocytes from the coelomic epithelium. What-ever mechanism is involved, it is clear that in explantsmorphallactic mechanisms play a significant role. Thus,blastemal regeneration of Antedon arm seems to be aphenomenon much more plastic than expected and in-volving a variety of cell recruitment mechanisms. It nor-mally invokes the intervention of presumptive stem cellsand trans-differentiation of differentiated elements fromthe coelomic epithelium, but, in extreme cases, it canalso involve a significant contribution of highly differen-tiated tissues via extensive cell rearrangement and dedif-ferentiation.

Utilizing a suitable combination of these fundamentalmechanisms the explant can undergo (1) complete repairat the proximal amputation site, without subsequent re-generative phenomena, that is without developing a re-generative blastema in the distal-proximal direction; (2)complete repair and subsequent regeneration at the distalamputation site, by developing a typical regenerativeblastema in the proximal-distal direction. This continuesto grow up to at least 2 weeks p.a. following the times,modalities and developmental processes comparable tothose described previously in regenerating donor arms.Therefore, although the basic mechanisms of cell prolif-eration/migration occur in both distal-proximal and prox-imal-distal directions, the normal developmental pro-cesses in terms of growth, morphogenesis and differen-tation appear to be strictly directional and, even in thedouble-amputated explants, maintains a strict proximal-distal axis. This significant difference in terms of regen-erative potential of the two symmetrical amputationsthus appears not to be due to differential capacities ofcell proliferation/migration, but must be genetically pro-grammed and is possibly regulated by the differential ex-pression of signal molecules and growth factors. Thestrictly directional blastemal regeneration of the explantis an important basis to exclude the possibility that crino-ids are spontaneously able to perform reconstructivemechanisms comparable to strategies of asexual repro-duction. On the other hand, it is relevant to point outthat, since crinoid arms normally undergo continuous

apical growth, developmental processes in the arms haveto be always maintained, though slowly, throughout life.Their acceleration during regeneration is possibly due tothe stimulatory action of specific factors and growth reg-ulators. As pointed out above, current work is focusingon the intervention of such presumptive growth factors inregenerative processes. The contribution of the brachialnerve and the coelom during regeneration potentially in-vokes not only a prompt cell supply but also the releaseof presumptive growth factors. With regard to neuropep-tides, our previous and present results suggest, in partic-ular, a basic physiological, neurotrophic and modulatoryrole for S1 and S2 peptides at the level of the nervoustissue perhaps without a significant involvement in re-generation itself. On the other hand, with respect toTGF-, we can confirm the hypothesis that also inechinoderms this factor can be considered not only as aconstitutive neurotrophic factor playing an importantfundamental role in development, repair, maintainanceand regulation of neuronal function, but possibly also asa broad-spectrum multifunctional regulator of cell prolif-eration and differentiation which plays a key role duringregenerative developmental processes (Logan et al.1994). The results obtained for the explants suggest, inparticular, that arm regeneration in crinoids is largely de-pendent on a remarkable functional autonomy of the armsystem not only in terms of cells and tissues involved,but also in terms of regulative mechanisms and mole-cules implied. In fact, apart from possible quantitativevariations which have still to be demonstrated, theseseem to be controlled by the same type of substances onthe basis of a comparable pattern of specific tissue distri-bution.

In conclusion, our studies of explants show clearlythat the regenerative potential of crinoid echinoderms ismuch wider than expected and suggests a remarkableflexibility of mechanisms. In the light of these results,epimorphic and morphallactic mechanisms, which aretraditionally considered contrasting strategies of regener-ation (Bonasoro et al. 1998), lose their defined bound-aries. In other words, the direct recruitment of undiffer-entiated stem cells, or the indirect recruitment of differ-entiated elements, previously dedifferentiated and/ortransdifferentiated, seem to be alternative mechanismswhich can be employed by the same organism and thesame structure according to local conditions. Both, how-ever, produce an identical result: the development of aregenerative blastema formed by undifferentiated cellswhich proliferate actively under the control of regulatorymolecules.

&p.2:Acknowledgements The present work has received financial sup-port from: (1) Consiglio Nazionale delle Ricerche, (CNR), Roma;(2) MURST Research Project 1997-98: Biologia ed Evoluzionedel riconoscimento e delle interazioni nelle cellule animali; (3)British-Italian Collaboration Grant for Research and Higher Edu-cation (The British Council/MURST) 1996-97.

429

References

Amemiya S, Oji T (1992) Regeneration in sea lilies. Nature357:546547

Bonasoro F, Candia Carnevali MD, Thorndyke MC, Welsch U(1995) Neural factors in crinoid arm regeneration. In: EmsonR, Smith AB, Campbell AC (eds) Echinoderm research 1995.Balkema, Rotterdam, pp 237243

Bonasoro F, Candia Carnevali MD, Patruno M, Sala F (1997) Pot-enzialit rigenerative in espianti di braccia di crinoidei (Ante-don mediterranea). Proceedings of the 58th Congresso U.Z.I.,Cattolica, Italy 1997, p 67

Bonasoro F, Candia Carnevali MD, Moss C, Thorndyke MC (1998)Epimorphic versus morphallactic mechanisms in arm regenera-tion of crinoids and asteroids: pattern of cell proliferation/dif-ferentiation and cell lineage. In: Mooi R, Telford M (eds)Echinoderms: San Francisco. Balkema, Rotterdam, pp 1318

Candia Carnevali MD, Bonasoro F (1994) Mechanisms of arm re-generation in Antedon mediterranea (Echinodermata, Crino-idea). Anim Biol 3:8388

Candia Carnevali MD, Bonasoro F (1995) Arm regeneration andpattern formation in crinoids. In: Emson R, Smith AB, Camp-bell AC (eds) Echinoderm research 1995. Balkema, Rotter-dam, pp 245253

Candia Carnevali MD, Bruno L, Denis Donini S, Melone G (1989)Regeneration and morphogenesis in the feather star arm. In:Kiortsis V, Koussoulakos S, Wallace H (eds) Recent trends inregeneration research. (Nato Asi Series, vol 172) PlenumPress, New York London, pp 447460

Candia Carnevali MD, Lucca E, Bonasoro F (1993) Mechanismsof arm regeneration in the feather star Antedon mediterranea:healing of wound and early stages of development. J Exp Zool267:299317

Candia Carnevali MD, Bonasoro F, Lucca E, Thorndyke MC(1995) Pattern of cell proliferation in the feather star Antedonmediterranea. J Exp Zool 272:464474

Candia Carnevali MD, Bonasoro F, Invernizzi R, Lucca E, WelschU, Thorndyke MC (1996) Tissue distribution of monoamineneurotransmitters in normal and regenerating arms of the feath-er star Antedon mediterranea. Cell Tissue Res 285:341352

Candia Carnevali MD, Bonasoro F, Biale A (1997) Pattern of bro-modeoxyuridine incorporation in the advanced stages of arm

regeneration in the feather star Antedon mediterranea. CellTissue Res 289:363374

Candia Carnevali MD, Bonasoro F, Welsch U, Thorndyke MC(1998) Arm regeneration and growth factors in crinoids. In:Mooi R, Telford M (eds) Echinoderms: San Francisco. Balk-ema, Rotterdam, pp 145150

Elphick MR, Price DA, Lee TD, Thorndyke MC (1991a) TheSALMFamides: a new family of neuropeptides isolated froman echinoderm. Proc R Soc London ser B 243:121127

Elphick MR, Reeve JR Jr, Burke RD, Thorndyke MC (1991b) Iso-lation of the neuropeptide SALMFamide-1 from starfish usinga new antiserum. Peptides 12:455459

Endean R (1966) The coelomocytes and coelomic fluids. In: Boo-lootian RA (ed) Physiology of Echinodermata. John Wiley,London New York, pp 301328

Gratzner GH (1982) Monoclonal antibody to 5-bromo and 5-iodo-deoxyuridine: a new reagent for detection of DNA replication.Science 218:474475

Holland ND (1994) Cell cycle subdivisions in regenerating armblastema of a feather star (Antedon mediterranea). In: DavidB, Guille A, Feral JP (eds) Echinoderm through time. Balk-ema, Rotterdam, pp 217220

Logan A, Oliver JJ, Martin B (1994) Growth factors in CNS repairand regeneration. Prog Growth Factor Res 5:379405

Mladenov PV (1996) Environmental factors influencing asexualreproductive processes in echinoderms. Oceanologica Acta19:227235

Mladenov PV, Burke RD (1994) Echinodermata: asexual propaga-tion. In: Adiyodi KG, Adiyodi RG (eds). Asexual propagationand reproductive strategies. (Reproductive biology of Inverte-brates, vol VI B) Oxford and IBH (Put), New Delhi BombayCalcutta, pp 339383

Moss C, Hunter AJ, Thorndyke MC (1998) Patterns of bromode-oxyuridine incorporation and neuropeptide immunoreactivityin the regenerating arm of the starfish, Asterias rubens. PhilosTrans R Soc London ser B, 353:421436

Perrier E (1873) Lanatomie et la rgnration des bras de la co-matula. Arch Zool Exp Genet 2:2886

Reichensperger A (1912) Beitrge zur Histologie und zum Verlaufder Regeneration bei Crinoiden. Z Wiss Zool 101:169

Smith VJ (1981) The echinoderms In: Ratcliffe NA, Rowley AF(eds) Invertebrate blood cells, vol 2. Academic Press, Londonpp 513562

430