The role of dorsoventral interaction in the onset of planarian regeneration

The planarian is remarkable in its ability to regenerate. Even a small fragment cut from the body can give rise to an intact animal. This process includes at least two events: onset of regeneration (blastema formation) and pattern formation. Immediately after amputation, the wound surface is covered with epidermis stretching from the circumference of the wound (Chandebois, 1980). In adult planarians, totipotent stem cells called ‘neoblasts’ are the only cells that can proliferate (Baguñà et al., 1989a; Baguñà, 1998). Extensive cell proliferation is observed in the stump, and the proliferated cells may differentiate during migration to the wound surface (Saló and Baguñà, 1984; Hori, 1992). Subsequently, an unpigmented tissue called the ‘blastema’ is formed, which regenerates anterior structures when formed on the anterior end of the amputated fragment and posterior structures when formed on the posterior end of the amputated fragment. In this study, we focused on the onset of regeneration, in an attempt to find the trigger for the regeneration events.

Urodeles are also known for their ability to regenerate appendages. Several studies have suggested that matrix metalloproteases (MMPs) are critical factors for the onset of this regeneration. After amputation of the appendages of urodeles, epidermal cells cover the wound surface and dedifferentiated cells migrate to take part in blastemas (Gardiner et al., 1986; Brockes, 1997). MMPs expressed at the stump may degrade the extracellular matrix and release cells from the matrix to initiate this cell locomotion (Yang and Bryant, 1994). In the case of planarian regeneration, it has been suggested that epithelial-mesenchymal interaction, which is initiated by degrading the basement membrane beneath the wound epithelium, may stimulate cell proliferation and migration (Chandebois, 1980; Baguñà et al., 1994). Intestines are distributed throughout the planarian body and may contain a variety of proteases, including MMPs. It is thought that proteases released from injured intestine may stimulate regeneration; however, whether MMPs are involved in planarian regeneration is still unknown.

It is well known that nerve tissue has a crucial role in regeneration, since denervated limbs of urodele cannot regenerate (Singer, 1974; Filoni et al., 1995). Recently, Mullen et al. (1996) have shown that nerve provides growth factors such as FGFs and stimulates apical ectodermal cap formation. For planarians, there are many studies suggesting an important role for nerve tissue in regeneration (Kido, 1952; Kishida and Kurabuchi, 1978). It has been demonstrated that some neuropeptides stimulate cell proliferation during regeneration (Baguñà et al., 1989b; Hori, 1997). However, it is still unclear whether nerve is essential for the onset of planarian regeneration.

Chandebois (1979) reported interesting observations suggesting that contact of dorsal and ventral epidermis and suture stretching by wound closure may stimulate regeneration events. Concerning this possibility, several interesting experiments have already been reported. In Dugesia gonocephala, when a piece of anterior region was grafted in DV-reversed orientation to the prepharyngeal region, outgrowth occurred on either the dorsal or the ventral side, with a depression of the corresponding region on the opposite site (Okada and Sugino, 1937). In similar experiments using Dugesia gonocephala (Santos, 1964), outgrowth was initiated on the suture of the graft on both the dorsal and ventral sides. Schilt (1970) observed that outgrowth appeared at both head and tail when a rectangular fragment was grafted to the original position in DV-reversed orientation in Dugesia lugubris. Interpretation is difficult because these results conflict with each other, and also because these experiments involved variation along the anteroposterior (AP) axis, but they do suggest that ectopic DV interaction induces outgrowth. However, it is still unclear what events are induced by DV interaction.

Thus, to investigate more precisely the function of the DV interaction during regeneration, we sucked a small piece of planarian body into a Pasteur pipette and then grafted the piece to the original position in DV-reversed orientation, and analyzed subsequently formed projections using not only molecular markers but also chimeric analysis (Agata et al., 1998, Umesono et al., 1998).

Two clonal strains of the planarian Dugesia japonica were used, namely GI and HI, which were established in our laboratory. GI and HI originated from the Irima river, Gifu Prefecture, and Iwayadani park, Hyogo Prefecture, respectively. GI was mainly used in this study and HI was used for chimeric grafting with GI.

Microsurgery was performed on intact planarians (13-18 mm in length) treated with 0.2% Chloretone on ice. A small piece of the head, prepharyngeal or tail region of the planarian body was sucked out using a Pasteur pipette and grafted into the original position in the dorsoventrally (DV) reversed orientation (Fig. 1A,Ba). As control experiments, the pieces were grafted to the original position in the anteroposteriorly (AP) reversed or original orientation (Fig. 1Bb,c). The planarians were kept wet after surgery with modified Holtfleter’s solution (Betchaku, 1970) saturated with casein at 10°C for 24 hours and were then kept in autoclaved tap water at 14°C for 24 hours. Subsequently, the animals were maintained at 18°C and, in most cases, fixed at 30 days after operation. To make GI/HI chimeras, a small piece of the head or tail region of GI was grafted into the equivalent region of HI, or vice versa. The chimeric planarians were maintained as above, starved for a week prior to fixation, and fixed 7 days, 14 days or 30 days after the grafting.

For fixation, a modified relaxant solution (Dawar et al., 1973), 1% HNO₃/2.25% formalin/50 μM MgSO₄, in modified Holtfleter’s solution, was used. Fixed samples were embedded in paraffin and serially sectioned at 4 μm. Some sections were stained with cresyl violet and others were used for in situ hybridization according to Kobayashi et al. (1998). Double fluorescent in situ hybridization was performed using a TSA-indirect kit (NEN Life Science Products). Biotin-labeled probes were detected by staining with Texas Red and then sections were treated with 2% glutaraldehyde or modified relaxant solution at room temperature for 15 minutes. Subsequently, DIG-labeled probe was detected by staining with FITC. Cell nuclei were labeled with Hoechst No. 33342 (Sigma). Whole-mount in situ hybridization was performed as described previously (Umesono et al., 1997; Agata et al., 1998). The probes used in this study were PH04 (planarian pro-hormone convertase 2 homologue: Agata et al., 1998), DjotxB (planarian Otd/otx homologue: Umesono et al., 1998), Plox5-Dj (planarian Hox/HOM-C gene, Orii et al., 1995, H. O., K. K., Y. Umesono, T. Sakurai, K. A. and K. W., unpublished data), DjvlgA (planarian vasa-related gene, Shibata et al., 1999), PN8 (the mucous core protein gene, K. A., unpublished data), IFb (intermediate filament; A. Tazaki, H. O., K. A. and K. W., unpublished data), PN5 (the mucous component gene, Y. Umesono, K. K., K. W. and K. A., unpublished data), and PH20 (planarian retrotransposon, K. A., K. K. and C. Kobayashi, unpublished data).

We grafted a small piece of the anterior (A), middle (M) or posterior (P) region to the original position in the DV-reversed orientation (Fig. 1A,Ba). All of the grafts with complete union (259/308) led to cup-shaped projections on both the dorsal and ventral sides of the host (Fig. 2A,B; Table 1). Grafts with incomplete union (49/308) induced nothing, or sometimes a small bulge (Table 1). As control experiments, a small piece was grafted to the original position in the AP-reversed or original orientation (Fig. 1Bb,c). These grafts with complete union (117/130) never caused such projections (Fig. 2F; Table 1). A few cases with incomplete union (13/130) caused a small bulge.

After 4 days of DV-reversed grafting, a white region was formed at the margin of the graft (Fig. 2C). This region resembled the blastema, which is unpigmented tissue formed at the stump during ordinary regeneration (Fig. 2D). The white region increased in size and subsequently formed a cup-shaped projection around 1 week after grafting. By 3 to 4 weeks after grafting, outgrowth of the projection ceased. In contrast, control grafts never formed such white regions and the scar faded away (Fig. 2E,F). Thus, only DV-reversed grafts caused the formation of blastema-like regions and subsequent projection formation.

DV-reversed grafts that had developed for 30 days were sectioned and stained with cresyl violet to analyze the DV polarity of the cup-shaped projections. In normal planarians, a pair of eyes, thin body-wall muscle layer and pigment cells are dorsal characteristics, whereas a thick body-wall muscle layer and fewer pigment cells are ventral characteristics (Fig. 3A). The cephalic ganglion (brain) in the anterior ventral region, and the ventral nerve cord (VNC) throughout the length of the body can easily be observed due to their strong staining with cresyl violet (Fig. 3A). The inner side of the projections formed on the dorsal side of the host (dorsal projections) exhibited ventral characteristics (Fig. 3B,C, white arrows), while the outer side exhibited dorsal characteristics (Fig. 3B,C, black arrowheads). In projections formed on the ventral side of the host (ventral projections), the inner side showed dorsal characteristics (Fig. 3B,D, black arrows), while the outer side showed ventral characteristics (Fig. 3B,D, white arrowheads). Remarkably, ectopic eyes were formed in the inner side of the ventral projections of the anterior region (Fig. 3B). It seems that the DV polarity of the grafts is kept in reversed orientation relative to the host and that the projections also have the DV polarity.

To confirm the DV polarity of the grafts, we performed whole-mount in situ hybridization using dorsal and ventral markers. PN8 was used as a dorsal or ventral marker, depending on the position along the AP axis. Normally, PN8 is expressed in both the dorsal and ventral sides as a band anterior to the prepharyngeal region. However, it is expressed only in the dorsal side of more anterior regions (Fig. 3E, arrowhead), and in the ventral side of more posterior regions (Fig. 3F, arrowheads). PH04, which is expressed in neuronal cells of the cephalic ganglion (brain) and ventral nerve cord (VNC), was used as a ventral marker (Fig. 3I). In the dorsal projections of the anterior region, PN8 was not expressed in the inner [ventral(V)-like] side, but was expressed in the outer [dorsal(D)-like] side (Fig. 3G). In the ventral projections, PN8 was expressed in the inner (D-like) side, but not in the outer (V-like) side (Fig. 3H). In addition, PH04 was expressed in dorsal projections (Fig. 3J). In the dorsal projections of the posterior region, expression of PN8 was detected in the inner (V-like) side, indicating ventral characteristics (Fig. 7B). PH04 was expressed in the inner (V-like) side of the dorsal projections, but not in the inner (D-like) side of the ventral projections (data not shown). These results indicate that the inner (V-like) side of the dorsal projections had ventral characteristics, while the inner (D-like) side of the ventral projections had dorsal characteristics. Thus, the DV polarity of the grafts was kept in reversed orientation relative to the host even when the projections had developed for 30 days.

We performed whole-mount in situ hybridization using DV boundary markers IFb and PN5. Both genes are expressed in the margin of the planarian body, but IFb is expressed in a narrower and more ventral region (Fig. 4A,E). In projections that had developed for 30 days, ectopic expression of both IFb and PN5 was observed in the edges of the projections (Fig. 4B,D,F,H). In contrast, no ectopic expression of these genes was observed in the control grafts that had developed for 30 days (Fig. 4C,G). The grafts had elevated levels of expression of both IFb and PN5, indicating that they formed the margin of the planarian body in the projections. Formation of an ectopic DV boundary structure in the projection demonstrates that a DV axis was newly established in the projection.

To address more precisely how the projections develop, we performed chimeric analysis. Chimeras were made between two clonal strains of Dugesia japonica, namely HI as host and GI as donor. PH20, which encodes a retrotransposon, is expressed more strongly in GI cells than in HI cells, which enables us to distinguish these cells by in situ hybridization (K. A., K. K. and C. Kobayashi, unpublished data). After 7 days of chimerism with HI as host and GI as donor, PH20 was expressed in the cells on the inside of the cup-shaped structures. It appears that the host and donor cells mixed a little, but the boundary between the host and the donor was still distinguishable at this stage (Fig. 5A). In particular, expression of PH20 on the epidermis clearly indicated that the host-donor boundary existed in the projection (Fig. 5B). Furthermore, double staining of PH20 and IFb revealed that the DV boundary was formed on the host-donor boundary (Fig. 5C). These results suggest that DV interaction induces the formation of the cup-shaped projections and establishment of the DV axis within the projections

We performed in situ hybridization using a marker for neoblasts and differentiating cells, DjvlgA. In the intact worm, DjvlgA-expressing cells are distributed in the mesenchymal space from head to tail. During regeneration, DjvlgA-expressing cells accumulate in the vicinity of the blastema (Fig. 6A). 4 days after DV reversed grafting, accumulation of the DjvlgA-expressing cells was observed in the developing projections (Fig. 6B). Differentiation of IFb-expressing cells in the edge of the projections could be detected starting 4 days after grafting (data not shown), indicating that cell differentiation occurred at the projections. These results suggest that the neoblasts and/or neoblast-derived cells participate in the formation of the projections. Further, no marked cell movement could be observed 7 days after grafting (Fig. 5), suggesting that neoblasts and/or neoblast-derived cells accumulating in the projections may have originated at ectopic DV boundary in the grafts. 30 days after grafting, extensive cell mixing between the host and donor was observed. In grafts of an HI fragment containing the brain into a GI host in the anterior region, the host cells seemed to migrate from the dorsal side and become positioned in the brain of the donor ventral side (Fig. 6C,D). This result indicates that the host cells and the donor cells gradually mix with each other during the development of the projection.

In the projections of the anterior region (anterior projection), ectopic eyes were frequently formed and a brain-like structure was detected by whole-mount staining using a neuronal cell marker (Fig. 7A). In the grafts of the posterior region, the projections became tail-like structures (Figs 2B, 7B) and the expression pattern of PN8 showed three stripes of positive cells, which is characteristic of the tail region (Figs 3F, 7B).

To explore the mechanisms causing these phenomena, we investigated the expression of region-specific transcription factor genes. DjotxB and Plox5-Dj were used as anterior and posterior markers, respectively (Fig. 7C,D). In the anterior projection, expression of DjotxB was clearly observed, but that of Plox5-Dj was not (Fig. 7E,F). In contrast, the projections of the posterior region (posterior projection) expressed Plox5-Dj strongly, but did not express DjotxB (Fig. 7G,H). Unexpectedly, expression of Plox5-Dj was observed strongly and homogeneously in the posterior projections, although Plox5-Dj is expressed strongly in the posterior tip, and its expression decreases in the anterior region in intact planarians (Fig. 7D). The strong and widespread expression of Plox5-Dj in the posterior projection may be related to the observation that Plox5-Dj is expressed more strongly and in a little wider region at the posterior tip during regeneration than in the intact animal (H. O., K. K., Y. Umesono, T. Sakurai, K. A. and K. W., unpublished data). These results indicated that the anterior projections had only anterior characteristics, while the posterior projections had only posterior characteristics. Interestingly, expression of PH04 and DjotxB clearly demonstrated that the brain ectopically formed in the anterior projection has a symmetric branch structure that seems to be identical to the structure from the stump to the most anterior tip (Fig. 7A,E). The results obtained with DV-reversed grafts of the middle region were even more informative than those of the anterior and posterior regions. Three types of projections were observed (Table 2). A complete head structure was formed in one case, including two eyes (Fig. 8A). More than half of the grafts formed tail-like structures expressing Plox5-Dj, and may have contained a pharynx in the projection (Fig. 8B). We could not identify the characteristics of the rest of the grafts even though we stained them using several molecular markers (Fig. 8C). Although unidentified structures developed in grafts of the middle region, these results imply that the projections induced by the DV interaction may developed into structures from the grafted position to the most distal tip.

Fig. 9 compares the process of ordinary regeneration with that of projection formation in the DV-reversed graft. During regeneration, dorsal and ventral tissues adhere to each other as a result of wound closure immediately after amputation. Subsequently, blastema is formed, and AP and DV axes are established. In this study, we have shown that ectopic DV interaction induced formation of a blastema-like region, outgrowth, establishment of a DV axis and formation of a structure from the grafted position to the most distal tip (Fig. 9A). These events induced by ectopic DV interaction are strikingly similar to the regeneration process, suggesting that the DV interaction has a key role in the onset of regeneration (Fig. 9B).

It had been shown that neural factor(s) have a role in regeneration. Kido (1952) grafted the ventral tissue, including the VNC, to the dorsal side and observed outgrowth. He concluded that neurons included in ventral tissues may have an important role in the outgrowth. However, our observations suggest that this kind of outgrowth may be induced by DV interaction, rather than by neural function. Although the nerve was severely damaged in our control grafts (AP-reversed or correctly orientated grafts), these grafts could not induce blastema-like tissue or outgrowths. This suggests that injured VNC is not essential for the onset of the regeneration.

While intercalary regeneration along the AP axis has been described (Okada et al., 1937), it has remained still unclear whether DV interaction can cause regeneration in an intercalary manner. By histological and chimeric analysis using molecular markers, we have shown here that the donor pieces maintained their original DV polarity and that ectopic DV interaction established a DV axis in an intercalary manner. Such intercalary regeneration is also observed in appendages of higher animals. In larval cockroaches and amphibians, when an amputated leg is grafted to the stump of the contralateral counterpart with or without a rotation of 180°, two supernumerary legs develop. In these manipulations, the dorsal part of the graft is juxtaposed to the ventral part of the stump or the anterior part of the graft is juxtaposed to the posterior part of the stump and, subsequently, intercalary regeneration is initiated between these misaligned tissues (Iten and Bryant, 1975; French, 1976; Bryant and Iten, 1976). These results suggest that it is a general phenomenon that ectopic AP or DV interaction induces establishment of AP or DV axes, respectively, in an intercalary manner in regenerating organs.

There were some conflicting results about DV-reversed grafts in previous studies, which seemed to be confused by a factor of the AP axis. We made an effort here to remove the factor of the AP axis by grafting a small piece to the original position. In our experiments, anterior and posterior projections had characteristics only of the grafted position. Interestingly, grafts of the middle region developed either head or tail characteristics, suggesting that these projections may consist of not only the structure of the grafted position. Our results suggest that the projection is a structure from the grafted position to the most distal (anterior or posterior) tip. In ordinary regeneration, the regenerating structure may be formed coordinately with the residual incomplete AP axis. In our graft experiments, it seemed that the structure of the projection was formed coordinately with the host AP axis. The unidentifiable structures in the middle region may be the result of confusion of AP information due to the residual intact host AP axis. Chandebois (1980) suggested that the manner of the wound closure plays a key role in determining regenerating structures; dorsal epidermis covering the stump results in induction of an anterior structure and ventral epidermis covering the stump results in induction of a posterior structure. Since it was difficult to judge when the host and the donor were united with each other in our graft experiments, we could not investigate whether the manner of epidermis covering at the union was different according to the grafting position. Chandebois also observed that the junction of the dorsal and ventral epidermis is maintained ventrally in the anterior tip, and dorsally in the posterior tip even in an intact animal. She concluded that the manner of wound closure reflected the manner of epidermis covering of mesenchyme in an intact animal. However, as shown in Fig. 5B, we only observed that dorsal and ventral epidermis adhered to each other at their boundary, suggesting that the manner in which the epidermis covers the stump is not involved in determination of the regenerating structure. It is still unclear how AP polarity is determined during projection formation and further study will be required to clarify this.

To understand how DV interaction induces the projection at the cellular level, it is indispensable to analyze cell proliferation and migration during projection formation. Unfortunately, we have failed to identify proliferating cells using BrdU. We then investigated the distribution of the neoblasts and differentiating cells during projection formation using DjvlgA as a marker. 4 days after DV reversed grafting, DjvlgA-expressing cells accumulated around the projections, as they accumulated around the blastema during regeneration. This result supports our morphological observation that DV interaction induced blastema-like regions. At the same time, ectopic expression of IFb was observed at the edge of the projection, indicating that new cell differentiation occurred at the projection. Although distinct projections were formed at 7 days after grafting, no marked cell movement could be observed by chimeric analysis. These results suggest that the projections are composed of neoblasts and/or neoblasts-derived cells, which may originate in the vicinity of the ectopic DV boundary. Using chimeric analysis with HI as the host and GI as the donor, we showed that 30 days after grafting, the brain (a ventral structure) in the graft was partially composed of the host cells, which seemed to migrate from the dorsal side of the host. This result suggests that neoblasts may not have positional identity and can differentiate into appropriate cells in response to surrounding positional cues that may be presented by differentiated cells. Reconstitution experiments using dissociated cells will be necessary to confirm this and to understand what kind of cells are responsible for induction of the projections.

However, we can suggest some likely molecular events induced by DV interaction, since molecules involved in DV interaction during early development of Drosophila imaginal disk and vertebrate limb are well understood. As mentioned above, when dorsal and ventral parts in larval cockroach or amphibian legs are ectopically adjoined, supernumerary legs develop. One model, the boundary model, explains the development of supernumerary legs (Meinhardt, 1983). It is proposed that boundaries of different cell populations producing molecule(s) controlling patterning act as organizing regions for establishment of the ectopic proximodistal (PD) axis (Campbell and Tomlinson, 1995). Although the planarian body cannot be compared simply with insect and amphibian legs, the formation of projections can be explained by the boundary model. Consistent with this model, these observations and our study in planarians suggest that cell-cell interactions induced by wound closure may be a widespread mechanism for the onset of regeneration. The boundary model does not contradict the molecular mechanism of PD axis formation in early development of Drosophila legs (Lecuit and Cohen, 1997) and of vertebrate limbs (Martin, 1995; Johnson and Tabin, 1997). It has been demonstrated that secreted molecules such as decapentaplegic/BMP, hedgehog/SHH and wingless/Wnt are involved in leg development of Drosophila melanogaster and vertebrates. Recently, we have reported that the BMP homolog of planarians is specifically expressed in the dorsal side of both intact and regenerating planarians (Orii et al., 1998). It is expected that homologous molecules may also be involved in planarian regeneration.

We thank Elizabeth Nakajima for critical reading of the manuscript. We also thank Yoshihiko Umesono, Chiyoko Kobayashi, Norito Shibata and Akira Tazaki for their helpful discussion. This work was supported by Special Coordination Funds for Promoting Science and Technology to K. A. and a Grant-in-Aid for Scientific Research on Priority Areas to K. A. and K. W.