Introduction

Bone morphogenetic proteins (BMPs) are multifunctional regulators of vertebrate and invertebrate development. The BMPs comprise the largest family of proteins within the transforming growth factor-β (TGF-β) superfamily and consist of several subgroups including the BMP2/4s and BMP5/8s. BMPs are involved in early patterning of the vertebrate embryo. They act as morphogens, specifying epidermal cell fate in the ectoderm in a concentration dependent manner. Antagonists such as chordin can block BMP signalling, leading to ectodermal cells adopting a neural fate (for recent review see Stern 2005). BMPs are also involved in the patterning of the vertebrate neural tube, where they act as morphogens to determine neural cell specification, also in a concentration-dependent manner (Wilson and Maden 2005).

BMPs have been isolated from numerous animal phyla, and it is apparent that these molecules perform conserved roles in dorsoventral embryonic patterning (De Robertis and Sasai 1996). Until recently, representative transcripts from the BMP5/8 subfamily (Ponce et al. 1999), the BMP2/4 subfamily (Angerer et al. 2000) and a novel BMP, univin (Stenzel et al. 1994), had been isolated from echinoderms. However, a recently conducted search of the sea urchin (Strongylocentrotus purpuratus) genome has now revealed the presence of 14 TGF-β genes from 11 subfamilies (Lapraz et al. 2006).

Functional evidence indicates a role for echinoderm BMP2/4s in embryonic ectodermal cell-fate specification (Angerer et al. 2000). Recently, it has been suggested that BMPs also play important roles in adult echinoderm regeneration. Distinct spatial and temporal patterns of BMP expression have been reported for both regenerating crinoid (Patruno et al. 2003) and ophiuroid (Bannister et al. 2005) arms. In addition, evidence from RNA in situ hybridisation experiments conducted in these studies indicate that, in both species, the cells expressing these BMP mRNAs are of coelomic origin.

In this study, we have isolated a novel member of the BMP2/4 subgroup from the brittle star, Amphiura filiformis, and have investigated the expression of the afBMP2/4 messenger RNA (mRNA) in regenerating and non-regenerating arms.

Materials and methods

Animal collection, RT-PCR and RACE

A degenerate primer reverse transcriptase-polymerase chain reaction (RT-PCR) approach was employed to search for novel bone morphogenetic proteins in regenerating A. filiformis arms. Animal collection, maintenance, induction of arm autotomy, RNA isolation and first-strand cDNA synthesis were performed as described in Bannister et al. (2005). First-round PCR was performed using the degenerate primers 5′-ACTCTAGATGGATYRTNGCNCC-3′ (sense) and 5′-ACGAATTCTTANCGRCANCCNCA-3′ (antisense) with an annealing temperature of 55°C and 45 amplification cycles. Nested PCR was performed using degenerate primers 5′-ACGAATTGCTTAYTWYTGYSANGG-3′ (sense) and 5′-ACGAATTCGTNGGNACRCARCA-3′ (antisense) with an annealing temperature of 50°C and 45 amplification cycles. First-round and nested primer sequences and cycling conditions were adapted from those originally used by Stenzel et al. (1994). PCR product was purified from a 1.5% agarose/Tris-acetate-EDTA (TAE) gel using the QIAquick gel extraction kit (QIAgen) and cloned into the pCR 2.1-TOPO vector (Invitrogen). Plasmids were transformed into TOP10 chemically competent Escherichia coli cells (Invitrogen). Selected colonies were sequenced at MWG-biotech AG laboratories. Extended 3′ sequence, including a stop codon, was obtained using a gene specific sense primer (5′-AAACTCAGTCAATCCGCAGCT-3′) with the degenerate antisense primer used in first-round PCR at an annealing temperature of 60.8°C for 35 cycles. 5′ Rapid amplification of cDNA ends (RACE) was performed using the Marathon cDNA amplification kit (Clontech). Touchdown PCR (annealing temperature of 70°C for 5 cycles followed by 68°C for 5 cycles and then 66°C for 25 cycles) was employed utilising a gene-specific antisense primer (5′-GCTTTGGGTACCAGCTGCGGATTGACTG-3′) and an adapter primer, AP1 (Clontech). Touchdown PCR products were resolved on a 1.2% agarose/TAE gel, bands extracted using the QIAquick gel extraction kit (QIAgen), cloned into the pGEM-T easy vector system (Promega), transformed into XL-1 Blue supercompetent E. coli cells (Stratagene) and sequenced at MWG-biotech AG laboratories.

Phylogenetic analyses

Additional sequences used in phylogenetic analyses were obtained from the National Centre for Biotechnology Information web resource (http://www.ncbi.nlm.nih.gov/). Theoretically translated amino acid sequences were aligned using ClustalW (Chenna et al. 2003) on the European Bioinformatics Institute (EBI) website (http://www.ebi.ac.uk/clustalw/) and presented using BOXSHADE (http://www.ch.embnet.org/software/BOX_form.html). Phylogenetic tree construction was performed by Maximum Likelihood with PhyML (Guindon et al. 2005) using the JTT substitution model and 500 bootstrap analyses.

Whole-mount in-situ RNA hybridisation

Antisense and sense RNA probes were synthesised using a 735-base fragment (bases 34–768) cloned into the pGEM-T easy vector (Promega) as a template. The identified region was isolated by PCR with the gene specific primers 5′- ATGCCTGACGAGGAAGTACTC-3′ (sense) and 5′-TTTGGGTACCAGCTGCGGATT-3′ (antisense) using an annealing temperature of 60°C and 35 amplification cycles. Plasmid DNA was linearised using the restriction enzymes SacII (antisense) and SacI (sense), and digoxygenin (Roche) labelled RNA probes were transcribed in-vitro using Sp6 and T7 RNA polymerases (Promega), respectively.

Whole-mount in situ hybridisation was performed on whole non-regenerating and regenerating (2 days and 1, 2, 3 and 5 weeks post ablation at 12°C) A. filiformis arms as described in Bannister et al. (2005).

Results and discussion

Isolation of afBMP2/4

Nested degenerate primer RT-PCR was used to identify a 132-base fragment of a novel TGF-β gene from A. filiformis regenerating arm tissue. Use of further RT-PCR and RACE led to the identification of an 885-base fragment of the coding region and stop-codon of this novel TGF-β (Genbank accession number DQ182545). The sequence contains several features characteristic of the BMPs, including a carboxy-terminal region containing seven cysteine residues (Fig. 1), and a putative arg-X-X-arg (where X is any amino acid) proteolytic cleavage site upstream of the C-terminal region (amino acids 172–175 of GenBank entry). The theoretically translated amino acid sequence of the conserved carboxy-terminal region of this molecule has high sequence identity to numerous members of the BMP2/4 subfamily (Fig. 1). This new ophiuroid TGF-β was thus named afBMP2/4. Carboxy-terminal sequence similarity to BMP2/4s from other echinoderms is high, for example, there is 89% identity with sea urchin (S. purpuratus) BMP2/4 (Angerer et al. 2000). Furthermore, this novel echinoderm BMP shows greater similarity to other vertebrate and invertebrate BMP2/4s than to members of other TGF-β groups found in echinoderms. For example, identities of the afBMP2/4 C-terminal region with those of univin (67%, Stenzel et al. 1994) and afuni, a univin-like molecule isolated from A. filiformis (69%; Bannister et al. 2005), are somewhat lower than to other BMP2/4 subfamily members, as is identity to members of the BMP5/8 subclass, including the echinoderm molecule spBMP5–7 (58%; Ponce et al. 1999). A phylogenetic tree constructed using Maximum Likelihood supports the inclusion of afBMP2/4 within the BMP2/4 subfamily of the TGF-β superfamily (Fig. 2). In this study, afBMP2/4 lies within the BMP2/4-decapentaplegic clade (supported by a bootstrap value of 98%) and is distinct from other BMP subfamilies including the univins and BMP5/8s.

Fig. 1
figure 1

Alignment of the carboxy-terminal region of afBMP2/4 with corresponding sequences of other TGF-β superfamily members. Amino acid identities to afBMP2/4 are boxed in black. Similar amino acids are shaded. Percentage amino acid identities to afBMP2/4 are shown. The position of conserved cysteine residues are indicated by asterisks. Sequences shown (and GenBank accession numbers) are: Amphiura filiformis (Af) BMP2/4, Homo sapiens (Hs) BMP4 (M22490), Hs BMP2 (M22489), Strongylocentrotus purpuratus (Sp) BMP2/4 (AF119713), Drosophila melanogaster (Dm) decapentaplegic (DPP; M30116), Halocynthia roretzi (Hr) BMPb (D85464), Af uni (AY954372), Sp univin (U10533), Hs BMP7 (X51801), Hs BMP5 (M60314), Sp BMP5–7 (Z48313), Mus musculus (Mm) GDF6 (U08338), Xenopus laevis (Xl) Vg1 (M18055), Dm 60A (M77012), Hr BMPa (D83183), Hs inhibin β a (M13436), Hs TGF-β1 (X02812)

Full size image
Fig. 2
figure 2

Phylogram of the relationships between selected TGF-β family members. Tree was constructed by Maximum Likelihood with the JTT substitution model using the conserved C-terminal domains of selected TGF-βs. The tree indicates that afBMP2/4 (indicated by asterisk) lies within the BMP2/4 subfamily of the TGFβ superfamily. Numbers associated with branches indicate ≥50% bootstrap values (500 replicates) supporting the topology shown. Sequences (and GenBank accession numbers) not given previously are: Drosophila melanogaster (Dm) Screw (U17573); Homo sapiens (Hs) BMP6 (M60315) and BMP8 (M97016); Mus musculus (Mm) GDF1 (M57639), GDF3 (L06443) and GDF5 (U08337)

Full size image

Expression of afBMP2/4 in regenerating and non-regenerating arms

Application of the afBMP2/4 antisense RNA probe to non-regenerating arms revealed an expression pattern running in a proximal-distal direction along the oral side of the arm (Fig. 3a). This staining pattern was not seen in arms treated with the afBMP2/4 sense probe (Fig. 3b). Expression is upregulated at segmental intervals, located equidistant between, and slightly proximal to, the bases of the podia pair associated with each segment. Sections (30 µm) taken of the colour developed non-regenerating arms indicate that this afBMP2/4 antisense probe staining is occurring in the radial water canal (RWC), a coelomic vessel that runs proximal-distal near the oral side of the ophiuroid arm with branches extending into each podium (Fig. 3c). This expression pattern was also observed in the non-regenerating regions of arms undergoing regeneration distally (Fig. 3d). In the non-regenerating part of an arm 1 week post-ablation, a similar pattern of expression to that observed in arms showing no evidence of recent regeneration can be seen. Also apparent here is the granular nature of the reaction (Fig. 3d), indicating that this staining within the established RWC is cellular. It should be noted that for technical reasons, despite several attempts, it was not possible to cut sections of non-regenerating arms embedded in medium hardness Agar 100 resin to investigate further the cellular location of afBMP2/4 due to the hard calcareous nature of the tissue.

Fig. 3
figure 3

AfBMP2/4 is expressed in non-regenerating and regenerating A. filiformis arms. a Whole-mount in situ RNA hybridisation shows high expression of afBMP2/4 along the proximal-distal axis of the oral side of the arm upregulated at segmental intervals (white arrows). Red arrows indicate intense regions of staining at the base of podia. b Sense probe control. c Staining is seen occurring proximal-distal along the oral side of the arm (black arrow) in a 30-µm longitudinal section taken of a non-regenerating arm. White arrow indicates podium. d The segmental pattern of intense regions of afBMP2/4 expression (arrows) can be seen on the oral side of the non-regenerating part of a regenerating arm 1-week post-ablation. Note the granular appearance of the segmental staining in this structure. e Expression of afBMP2/4 can be seen at the distal tip of a 2-week regenerate (black arrow). Red indicates the transition between the non-regenerating arm tissue and the regenerate f Two-micrometer resin section taken of the regenerating tip in e, showing specific cellular afBMP2/4 expression (arrows). All scale bars shown are 200 µm except c 5 µm. Proximal (p)-distal d orientation is shown in plates a–d

Full size image

In regenerating animals, afBMP2/4 could only be detected by in situ hybridisation in regenerating arm tissue 2 weeks post-ablation (Fig. 3e). Here, upregulation of this novel BMP RNA was seen at the tip of the regenerate. The cellular localisation of this RNA message is confirmed by 2-μm sections taken of the regenerating tip at this time (Fig. 3f). Expression was not observed at any other time-point studied. These results suggest that a peak level of afBMP2/4 in the regenerating region of arms is restricted to a short temporal window around 2 weeks post-ablation.

The expression of this gene in 2-week regenerates indicates that afBMP2/4 may be playing an important role in the growth and development of the regenerate at this time. AfBMP2/4 was observed in regenerates showing no clear sign of segmental definition or of the development of segment-associated spines or podia. In situ RNA hybridisation analyses of another echinoderm BMP2/4 revealed expression occurring during the early stages of arm regeneration by the crinoid Antedon bifida (Patruno et al. 2003). Here, AnBMP2/4 was detected in the coelomic canal immediately proximal to the regeneration site as early as 24 h post-ablation (p.a.). Expression continued until 1-week p.a. when it became focussed in the regenerating blastema amongst the mass of blastemal “stem cells” that accumulate there. By 2 weeks p.a., the signal was undetectable. This pattern implied a putative role for anBMP2/4 in early crinoid arm regeneration, as the spatial and temporal expression coincided with a region where new tissues are differentiating. The location of afBMP/4 in the coelom of non-regenerating arms proximal to the site of regeneration and at the tip of the ophiuroid regenerate before segmental and structural definition suggests a possible role for afBMP2/4 comparable to anBMP2/4. The later onset of detectable afBMP2/4 signal (by in situ hybridisation) in the regenerate with respect to anBMP2/4 may reflect temporal differences in the regeneration processes of these two species. Functional analyses are required to elucidate fully any possible class and/or species-specific role(s) of BMP2/4s in echinoderm arm regeneration.

Significance of the coelom in echinoderm regeneration

Recently, our laboratory reported the sequence of afuni, another BMP isolated from A. filiformis arms (Bannister et al. 2005). This mRNA is expressed in discrete proximal and distal sites during late regeneration (3–5 weeks post-ablation), implying multiple roles for afuni during regeneration. Interestingly, afuni was located in the RWC at both the proximal and distal regions of the regenerate and in the RWC of non-regenerating arms. Based on this evidence, it was suggested that the coelomic epithelium that lines the RWC may be a key source of cells used in the ophiuroid regeneration process. It should also be noted that anBMP2/4 was similarly upregulated in the coelomic canal epithelium of the crinoid arm during regeneration, both before and coincident with expression in the proximal regenerate (Patruno et al. 2003). The organogenetic potential of the echinoderm coelom has been recognised for some time now. Indeed, in their review, Candia Carnevali and Bonasoro (2001) suggested such a functional possibility for this structure. In the present study, afBMP2/4 was detected in the RWC of non-regenerating tissue of both regenerating arms and in arms showing no visible sign of regeneration and was also detected at the tip of 2-week regenerates. There is continuous afBMP2/4 expression observed in presumptive coelomocytes all along the RWC of the non-regenerating arm, and this is appreciably upregulated at regular segmental intervals along its length. This observed pattern hints at a possible role for this gene in the somatic growth and maintenance of the arm. It is also possible that the cells expressing this mRNA in the regenerating tip originated in the RWC. Therefore, one possible explanation for the regular segmental intensity of afBMP2/4 may be that these areas of high message levels represent segmental pools of cells set aside and ready for use in arm regeneration. Such a system may well be beneficial to this species due to the high incidence of epimorphic arm regeneration (i.e. undifferentiated cells migrate to the autotomy site and form a regenerative bud or ‘blastema’; Thorndyke et al. 2001) that occurs naturally in this species (Sköld and Rosenberg 1996). These segmental pools would provide a source of cellular material close to the autotomy planes, where breakage occurs along the length of the arm. This hypothesis further implies a key role for the coelomic RWC as a source of cellular material for use in ophiuroid arm regeneration.

The importance of the coelom as a source of cellular material used in the formation of the adult echinoderm body plan has been highlighted by Davidson et al. (1995). The sea urchin S. purpuratus undergoes maximal indirect development (i.e. the construction of the larval and adult body plans are mechanistically separated in time) through a planktotrophic larval phase. After embryogenesis, the larval cells only replicate two to three more times on average. The pentameric body plan of the adult S. purpuratus arises from pluripotent ‘set-aside’ cells derived from the coelom of the bilateral larva. Larval features remaining at metamorphosis are jettisoned and autolysed.

The present study adds to the increasing evidence for the importance of the echinoderm coelom as a source of set-aside cells for the regenerative development of new structures in adults, in addition to its role as a source of cells for the metamorphic development of members of this group.