Specification of ectodermal teloblast lineages in embryos of the oligochaete annelid Tubifex: involvement of novel cell-cell interactions

Embryogenesis in clitellate annelids (i.e. oligochaetes and leeches) is characterized by the generation of five bilateral pairs of embryonic stem cells called teloblasts early in development (Anderson, 1973; Devries, 1973a; Fernandez and Olea, 1982; Shimizu, 1982; Irvine and Martindale, 1996). Teloblasts, which are derived from micromeres of the D quadrant, repeatedly undergo extremely unequal divisions to produce a coherent column (bandlet) of smaller daughter cells (referred to as primary blast cells). Four of the five bandlets on each side of the embryo join together to form an ectodermal germ band (GB), while the remaining bandlet becomes a mesodermal GB. From previous descriptive and cell ablation studies (Whitman, 1878; Penners, 1924; Penners, 1926; Devries, 1973a; Devries, 1973b), it has been suggested that teloblasts (and their progenies) play a pivotal role in clitellate annelid development. In fact, teloblasts are the only source of ectodermal and mesodermal segmental tissues; none of the non-teloblastic cells can replace missing teloblasts in this respect. Furthermore, morphogenetic events such as body elongation and segmentation depend solely on the presence of teloblasts and their progeny (Blair, 1982; Wedeen and Shankland, 1997; Goto et al., 1999a; Shain et al., 2000; Kitamura and Shimizu, 2000; Nakamoto et al., 2000).

Ectodermal teloblasts (ectoteloblasts N, O, P and Q) on either side of the embryo are produced through an invariable sequence of cell division of a proteloblast, NOPQ, that is derived from the second micromere 2d; a bilateral pair of mesodermal teloblasts (mesoteloblasts M) results directly from equal division of the fourth micromere 4d (Fig. 1A,B). (Note that the precursors of the M and NOPQ in leech embryos have been designated as DM and DNOPQ, respectively; see Stent et al., 1982.) Recent cell lineage analyses of teloblasts have shown that developmental fates of the four ectoteloblasts are not only different from those of the mesoteloblast but also distinct among themselves (Weisblat et al., 1980; Weisblat et al., 1984; Storey, 1989; Goto et al., 1999b). At present, it is not clear how and when these teloblasts (and their progeny) acquire distinct developmental fates. As has been well documented, teloblasts N-Q emerge at different positions along the embryonic axes and at different times (Fernandez and Olea, 1982; Shimizu, 1982; Sandig and Dohle, 1988); however, nothing is known about the causal relationship between developmental fates of teloblasts and spatiotemporal aspects of their emergence. The only thing that is known about specification of ectoteloblast lineages in clitellate annelids is that in the leech Helobdella, bandlets derived from the O and P teloblasts are initially equipotent and can be differentiated from each other according to their position within the GB (Weisblat and Blair, 1984; Zackson, 1984). Recently, it has also been suggested that the cell fate determination in this equivalence group occurs through inductive signals from another teloblast lineage and the micromere-derived epithelium (Ho and Weisblat, 1987; Huang and Weisblat, 1996). It remains to be determined whether this network of cell interactions is widespread in clitellate annelids.

The present study was undertaken to gain an insight into the mechanisms that underlie specification of ectoteloblast lineages in the oligochaete annelid Tubifex. The objectives of this study were to determine the timing of specification of ectoteloblast lineages and to determine whether specification of each lineage depends on external cues. For this purpose, we used embryological techniques such as cell ablation in combination with labeling of specific blastomeres with lineage tracers. Our results show that teloblasts N, P and Q are specified to express the N, P and Q fates, respectively, as early as their birth, and that the O teloblast and its progeny are initially pluripotent and their fate becomes restricted through inductive signals emanating from its sister P lineage. On the basis of these findings, we suggest that it is unlikely that sister teloblasts O and P in Tubifex embryos constitute an equivalence group such as that seen in the leech embryo.

Embryos of the freshwater oligochaete Tubifex hattai were obtained as previously described (Shimizu, 1982) and cultured at 22°C. For the experiments, embryos were all freed from cocoons in the culture medium (Shimizu, 1982). To sterilize their surface, cocoons were treated with 0.02% chloramine T (Wako Pure Chemicals, Osaka, Japan) for 3 minutes and washed thoroughly in three changes of the culture medium. The culture medium used in cell-ablation experiments was autoclaved and, shortly before use, antibiotics (penicillin G and streptomycin, 20 units/ml each) were added. Unless otherwise stated, all experiments were carried out at room temperature (20-22°C).

To label ectoteloblasts and their progeny cells, ectoteloblasts or their precursors (cells NOPQ, OPQ and OP; see Fig. 1B-D) were pressure-injected with DiI (1,1´-dihexadecyl-3,3,3´,3´-tetramethylindocarbocyanine perchlorate; Molecular Probes). DiI was dissolved in ethanol at 100 mg/ml and stored at room temperature. Before use, an aliquot of this solution was diluted 20 times in safflower oil (Kitamura and Shimizu, 2000). Ectoteloblasts or their precursors were injected with oil droplets containing DiI by means of micropipettes. DiI-injected embryos were kept in darkness.

Embryos without vitelline membranes were placed on 2% agar in the culture medium. Blastomeres were killed by making a wound on their surface with fine forceps. Within a minute, the yolk mass of these cells began to coagulate. The coagulating cells were removed by pulling them away from the remainder of the embryo. The operated embryos were allowed to develop in culture medium containing antibiotics, which was renewed daily.

DiI-labeled embryos were fixed with 3.5% formaldehyde in phosphate buffer (40.5 mM Na2HPO4, 9.5 mM NaH2PO4, pH 7.4) for 1 hour and mounted in phosphate buffer for observation. Images were collected on a Molecular Dynamics Sarastro-2000 confocal laser-scanning microscope. Some specimens were viewed under a Zeiss Axioskop epifluorescence microscope.

A brief review of the early development in Tubifex is presented here as a background for the observations described below (for details, see Shimizu, 1982; Goto et al., 1999a; Goto et al., 1999b). Precursors of teloblasts are traced back to the second (2d) and fourth (4d) micromeres of the D quadrant. At the 25-cell stage, 2d¹¹¹, 4d and 4D (sister cell of 4d) all come to lie in the future midline of the embryo (Fig. 1A). 4d divides equally to yield the left and right mesoteloblasts (Ml and Mr); 2d¹¹¹ divides into a bilateral pair of ectoteloblast precursors, NOPQl and NOPQr; and 4D divides equally yielding endodermal precursors E^D (Fig. 1B). Ectoteloblasts N, O, P and Q arise from an invariable sequence of divisions of cell NOPQ on both sides of the embryo (Fig. 1J; Goto et al., 1999b). NOPQ on either side of the embryo undergoes unequal divisions twice after its birth and then divides into a smaller N teloblast and a larger cell, OPQ (Fig. 1C). Similarly, after producing small cells twice, OPQ divides into a smaller Q teloblast and a larger cell, OP (Fig. 1D). Finally OP undergoes unequal division four times after its birth and then cleaves almost equally, yielding the third-born ectoteloblasts O and P (Fig. 1E), at which point teloblastogenesis is complete.

After their birth, each of the teloblasts thus produced divides repeatedly, at 2.5 hour intervals (at 22°C), to give rise to small cells called primary blast cells, which are arranged into a coherent column (i.e. a bandlet; Fig. 1F). Within each bandlet, primary blast cells and their descendants are arranged in the order of their birth. Bandlets from N, O, P and Q teloblasts on each side of the embryo join together to form an ectodermal GB, while the bandlet from the M teloblast becomes a mesodermal GB that underlies the ectodermal GB (Fig. 1F; Goto et al., 1999a). The GBs are initially located on the dorsal side of the embryo (Fig. 1G). Along with their elongation, they gradually curve round toward the ventral midline and finally coalesce with each other along the ventral midline (Fig. 1H). The coalescence is soon followed by dorsalwards expansion of GBs. The edges of the expanding GBs on both sides of the embryo finally meet along the dorsal midline to enclose the yolky endodermal tube (Fig. 1I; Goto et al., 1999a; Goto et al., 1999b).

To determine the extent to which specification of ectoteloblast lineages depends on external cues, we followed the development of ectoteloblasts that had been forced to be ‘solitary’. In this study, we assessed fates of operated ectoteloblast lineages according to compositions and spatial distribution of terminally differentiated cells descending from these ectoteloblasts. In intact embryos, each ectoteloblast makes a topographically characteristic contribution to the ectodermal tissues, which exhibit a segmentally repeated distribution pattern (Fig. 2; Goto et al., 1999b).

Our previous study has shown that a bandlet derived from a ‘solitary’ O teloblast (resulting from removal of all of its ipsilateral sister teloblasts) exhibits early morphogenetic features (e.g., shape of bandlets) characteristic to the P lineage rather than the O lineage, while bandlets derived form ‘solitary’ N, P and Q teloblasts are very similar to the respective bandlets in intact embryos (Nakamoto et al., 2000). In this study, we have extended this observation to more advanced developmental stages when cells are terminally differentiated. To do this, we labeled one of the four ectoteloblasts with DiI and ablated the other three ipsilateral ectoteloblasts (or their precursors) simultaneously. After 5 days culture, we examined the composition and distribution of labeled cells descending from ‘solitary’ teloblasts. In nearly all of the operated embryos, bandlets derived from ‘solitary’ teloblasts were found to have elongated along the anteroposterior axis in a normal fashion; the dorsalward migration of blast cell progeny also occurred to the same extent as that in intact embryos. This allowed us to assess fates of ‘solitary’ teloblasts on the basis of the distribution pattern of differentiated cells. More than 15 embryos were examined for each lineage.

Fig. 3A,B shows the organization of labeled cells derived from an intact and a ‘solitary’ N teloblast, respectively. These two cases are indistinguishable from each other in that nearly all of the labeled cells were located in the ventral region of the embryo and occupied each hemiganglion. Similarly, labeled cells derived from ‘solitary’ P and Q teloblasts are organized in a pattern comparable with that in intact P and Q lineages, respectively (Fig. 3E-H).

In contrast, organization and composition of cells derived from ‘solitary’ O teloblasts are distinct from those in normal o bandlets. As Fig. 3D shows, ‘solitary’ o bandlets apparently exhibited a P pattern rather than an O pattern of progeny cells (also see Fig. 3C). This result suggests that ‘solitary’ o bandlets adopt the P fate rather than the O fate.

In the foregoing experiments, only the left GB was operated on and the contralateral (right) GB remained intact in each embryo. This raises the possibility that the fate of the ‘solitary’ bandlets on the left side could be affected by the intact right GB when they aligned themselves along the ventral midline (Fig. 1H). To examine this possibility, we followed the fate of ‘solitary’ left bandlets in embryos that had been subjected to bilateral ablations of ectoteloblasts. To do this, we ablated right NOPQ (i.e. exclusive source of the right ectodermal GB; see Fig. 1B) and three of the four left ectoteloblasts, leaving a single (DiI-labeled) ectoteloblast in each embryo. After 5 days culture, the operated embryos (five to seven for each lineage) were examined for the composition and distribution of labeled cells.

We found that even after bilateral ablations of ectoteloblasts, ‘solitary’ n, p and q bandlets exhibited N, P and Q patterns of distribution of progeny cells, respectively (Fig. 4A,C,D) and that o bandlets followed the P fate rather than the O fate (Fig. 4B). These results are apparently the same as those obtained in the unilateral ablation experiments. Thus, we suggest that the presence of contralateral GBs does not influence the fate decision of ‘solitary’ bandlets. In the following experiments, we used embryos in which right GBs were intact.

The aforementioned results suggest the possibility that in intact GB, o bandlets are induced to assume O fate by interactions with other bandlets. To test this possibility and to find out which bandlet acts as such an inducer, we ablated teloblasts in various combinations, leaving an O teloblast plus one or two other teloblasts in each embryo, and followed the fates of the progenies of O teloblasts.

The results are summarized in Table 1. It was only when p bandlets survived that o bandlets assumed the O fate (Fig. 5B). Neither n nor q bandlets were effective at all in this respect (Fig. 5A); even when both n and q bandlets coexisted with o bandlets, they failed to induce o bandlets to assume the O fate (Fig. 5C). It is unlikely that this failure resulted from separation of o bandlets from n and/or q bandlets in operated embryos, as bandlets in a GB from which one bandlet had been deleted were found to be aligned tightly with each other (not shown). These results suggest that p bandlets exclusively serve as an inducer of O fate in o bandlets.

In normal Tubifex embryos, differences between the O and P lineages are manifested as early as the time of the first division of primary blast cells. Primary o blast cells enter first mitosis at a distance of 7 cells from the parent teloblast and undergo equal division (Fig. 6A); in contrast, primary p blast cells undergo unequal division at a distance of 5 cells (Fig. 6C). These differences suggest the possibility that the induction of the O lineage by the P lineage occurs prior to first mitosis of primary o blast cells. This possibility was verified by our observation that primary blast cells derived from ‘solitary’ O teloblasts entered first mitosis at a distance of 5 cells from the parent teloblast and underwent unequal division (Fig. 6B). This division pattern is evidently a P pattern, not an O pattern. Thus, it is likely that in intact GB, inductive signals from the P lineage also determine the pattern of first division in primary o blast cells.

Unlike O teloblasts, fates of N, P and Q teloblasts do not appear to be affected by neighboring bandlets. This suggests that these three teloblasts are specified as early as their birth. Considering the fact that the O teloblast is the sister of the P teloblast (see Fig. 1J), however, it is also possible that, like the O teloblast, the P teloblast is pluripotent and can express fates other than the P fate in an appropriate environment. To distinguish these possibilities, we labeled P teloblasts with DiI shortly after birth and simultaneously ablated other teloblasts in various combinations, leaving a labeled P teloblast plus one or two other teloblasts in each embryo. After 5 days culture, we examined the distribution of labeled cells.

The results are summarized in Table 2. Irrespective of the presence of any other teloblasts, p bandlets assumed the P fate (Figs 4C, 5D). The p bandlets did not show any sign of N, O or Q pattern of progeny cells in any of the combinations with other teloblasts. These results suggest that p bandlets are unlikely to receive inductive or inhibitory signals from neighboring bandlets. Thus, it is more likely that P teloblasts are specified to assume the P fate at their birth.

During Tubifex embryogenesis, NOPQ on either side of the embryo generates four ectoteloblasts (N, O, P and Q) through a stereotyped sequence of cell division. These teloblasts are strictly regulated not only in the order of their emergence but also in their position along the embryonic axes. Furthermore, these four teloblasts assume distinct fates (Goto et al., 1999b). The present study shows that the fates of any of the N, P and Q teloblasts are not affected by ablation of the other three teloblasts at birth. This suggests that these teloblasts are committed to their respective fates at their birth.

At present, nothing is known about either the mechanisms for cell fate determination in teloblasts or the way in which different teloblasts acquire distinct developmental fates. One possibility is that fates of teloblasts are determined according to positional cues along the embryonic axes. Alternatively, fates of teloblasts could be determined by intrinsic factors that are asymmetrically segregated during teloblastogenesis.

Additionally, it is possible that the commitment of a teloblast is due to inductive signals coming from its sister cell at their birth. It is therefore interesting to note that left NOPQ that has been transplanted to the right side of another embryo (from which right M and NOPQ had been ablated) undergoes a sequence of cell division that is identical to that in left NOPQ of intact embryos; a first-born teloblast, which is located dorsalmost on the right side of such a reconstituted embryo, assumes the N fate, and a second-born teloblast, located ventralmost, expresses the Q fate (A. A., unpublished). In intact embryos, the N teloblast is born first and located ventralmost, and the Q teloblast, which is born next, is located dorsalmost (Fig. 1E). Thus, these results suggest it is unlikely that cell fates of teloblasts are determined according to the positional cues along the embryo’s dorsoventral axis. Rather, it seems more likely that NOPQ is already polarized at its birth and that cell fate decision occurs along this polarity during teloblastogenesis.

Of the four ectoteloblast lineages, the O lineage is the only one that is affected by ablation of the other lineages (i.e. bandlets). The present study showed that o bandlets assume the O fate in the presence of p bandlets; otherwise they express the P fate. Apparently, for the O lineage, the P fate is the primary fate and the O fate is the secondary fate. Thus, it is reasonable to assume that O teloblasts are pluripotent.

It appears that in intact GBs, pluripotent o blast cells are induced, by interactions with a p bandlet, to assume the O fate prior to their entry into first mitosis. In intact embryos, the O and P teloblasts lie next to each other, and primary o blast cells come to be in contact with primary p blast cells during their birth (Fig. 7A). It is conceivable that primary o blast cells are induced to assume the O fate as early as their birth.

The O and P teloblasts are sister blastomeres resulting from the equal division of an OP proteloblast. As discussed above, the O teloblast is pluripotent and its progeny cells are able to respond to inductive signals from the neighboring p bandlet. In contrast, P teloblasts do not appear to be affected by the O teloblast. Furthermore, P teloblasts assumed only the P fate under any of the experimental conditions that included ablation of teloblasts in all possible combinations. On the basis of these observations, we suggest that sister teloblasts O and P in the Tubifex embryo are not equivalent but are distinct from each other in their developmental potency. It appears that the OP proteloblast undergoes an asymmetric division giving rise to two equal-sized teloblasts.

The aforementioned situation of Tubifex O and P teloblasts is in sharp contrast to that of O/P teloblast pairs in the leech Helobdella, which are thought to constitute a so-called ‘equivalence group’ (Weisblat and Blair, 1984; Zackson, 1984). Leech O/P teloblasts are homologs of Tubifex O and P teloblasts, in that O/P teloblasts are third-born teloblasts that result from the equal division of an OP proteloblast (Fernandez and Olea, 1982; Sandig and Dohle, 1988). Unlike the O and P teloblasts in Tubifex, however, sister O/P teloblasts in Helobdella are both pluripotent and have the potential to follow either an O or P fate. Blast cells derived from either O/P teloblast assume the secondary (P) fate if they interact with a bandlet derived from the Q teloblast; otherwise o/p blast cells express the primary (O) fate (Huang and Weisblat, 1996; Fig. 7B). More importantly, an o/p bandlet is unable to induce another o/p bandlet of the same GB to assume the secondary fate (Huang and Weisblat, 1996). Furthermore, in Helobdella, it has also been suggested that the micromere-derived epithelium that overlies the GB plays a role in fate decision in o/p bandlets (Ho and Weisblat, 1987). This is not the case for Tubifex, however, as the ectodermal GB in embryos of this animal is not overlain by an epithelium during its migration toward the ventral midline (see Fig. 1G,H; A. A., unpublished). Thus, Tubifex and Helobdella involve distinct cell interaction networks in patterning the ectodermal GB. It would be of interest to investigate whether the molecular natures of inductive interactions (between o and p in Tubifex and o/p and q in Helobdella) are also different in these animals.

In relation to these differences in the mechanisms for specification of ectoteloblast lineages, it is noteworthy that Tubifex and Helobdella are also different in the mode of bandlet formation. In Tubifex, the four ectoteloblasts on either side of the embryo are arranged in a row running along the dorsoventral axis, and they are closely associated with each other (Fig. 7A). Blast cells produced from each teloblast are located on the surface of the embryo and they are integrated into a GB shortly after their birth. There is a strict correspondence in the ventrodorsal order of the ectoteloblasts and their descendant bandlets (Fig. 7A). In Helobdella, the four ectoteloblasts are arranged in an invariable pattern, but as a result of extensive changes in their relative positions during teloblastogenesis, their final positions do not necessarily reflect their birth order (Fig. 7B). The Q teloblast is always located between O/P teloblasts; the N teloblast is located at a distance from the other teloblasts (Fernandez and Stent, 1980). As a result of such an irregular location of teloblasts, bandlets are initially ‘solitary’; in about one third of embryos, two bandlets derived from an O/P pair cross each other before they are integrated in the GB (Fernandez and Stent, 1980; Zackson, 1984; Weisblat and Blair, 1984), suggesting the spontaneous transposition of o/p bandlets in Helobdella embryos.

In spite of these differences, however, the resulting (final) pattern of the ectodermal GB is strikingly similar in Tubifex and Helobdella (Weisblat and Shankland, 1985; Goto et al., 1999b). Thus, we suggest that during their evolutionary isolation, oligochaetes and leeches have preserved an ancestral pattern of the ectodermal GB despite the divergence of cell interaction networks through which this pattern is brought about. Similar evolutionary changes in cell interaction networks that can produce structures with similar morphological pattern have also been reported in vulval development of nematodes (Sommer and Sternberg, 1994; Sommer et al., 1994).