As in other clitellate annelids, embryonic development in the oligochaete Tubifex is characterized by the generation of five bilateral pairs of teloblasts (designated M, N, O, P and Q), which serve as embryonic stem cells to produce germ bands on either side of the embryo. A large part of the tissues comprising body segments has been assigned to the progenies of the teloblasts; however, the developmental fate of each teloblast has been inferred only from its initial position in the embryo. In the present study, the fate of the progenies of each teloblast was followed by means of intracellular injection of a tracer enzyme, horseradish peroxidase. Cell fate maps for teloblasts in the Tubifex embryo were constructed. M teloblasts gave rise to nearly all of the mesodermal tissues, which included circular and longitudinal muscles, coelomic walls, nephridia (in segments VII and VIII) and primordial germ cells (in segments X and XI). Although few in number, M teloblasts also contributed cells to the ventral ganglion. Similarly, each of the ectoteloblasts, N, O, P and Q, made a topographically characteristic contribution to the ectodermal tissues such as the nervous system (i.e. ganglionic cells and peripheral neurones) and epidermis, all of which exhibited a segmentally repeated distribution pattern. The P and Q teloblasts uniquely gave rise to additional ectodermal tissues, namely ventral and dorsal setal sacs, respectively. Furthermore, O teloblasts made a contribution to the nephridiopores in segments VII and VIII as well. These results confirm the previously held view that ectoteloblasts and mesoteloblasts are the main source of ectodermal and mesodermal segmental tissues, respectively, but also suggest that all of the teloblasts produce more types of tissue than has previously been thought.
One of the characteristic features of embryogenesis in clitellate annelids (i.e. oligochaetes and leeches) is the generation of five bilateral pairs of embryonic stem cells called teloblasts early in development ( Anderson 1973; Devries 1973a; Fernandez & Olea 1982; Shimizu 1982; Irvine & Martindale 1996). Teloblasts are derived from the second and fourth micromeres of the D quadrant and possess the ability to undergo extremely unequal divisions repeatedly to give rise to smaller daughter cells (referred to as primary blast cells), which are arranged into a coherent column (bandlet). Four of the five bandlets on each side of the embryo join together to form an ectodermal germ band, while the remaining bandlet becomes a mesodermal germ band ( Fig. 1G). From previous descriptive and cell ablation studies ( Whitman 1878; Wilson 1889; Penners 1924, 1926; Mori 1932; Devries 1973a, b), 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 progenies.
In spite of their developmental importance, the fates of individual teloblasts in clitellate annelids, except for those in leeches, have not been elucidated. Earlier investigators inferred the developmental fates of teloblasts on the basis of their initial position within an embryo (Whitman 1878; Wilson 1889; Penners 1924; Devries 1973a). However, recent extensive cell lineage studies on leech embryos, in which horseradish peroxidase (HRP) and fluorescent dextrans have been used as lineage tracers ( Weisblat et al. 1980 , 1984; Torrence & Stuart 1986), have shown that progenies of teloblasts undergo extensive rearrangement in their positions within the germ band and that all of the teloblasts generate more types of tissue than has previously been thought. This suggests that traditionally held views on fate maps for teloblasts in other clitellate annelids should be re-examined with reliable methods.
In the oligochaete Eisenia, Storey (1989) labeled some teloblasts with injected HRP to follow their progenies. However, Storey confined observation to the early division pattern of blast cells and did not extend it to more advanced developmental stages. Nevertheless, that suggested that intracellular injection of HRP to label blastomeres may be applicable to oligochaete embryos as well. This led us to re-examine cell fate maps for teloblasts in another oligochaete, Tubifex, using this labeling method. Here, we report the distribution patterns of progenies of HRP-injected teloblasts in the Tubifex embryo.
Materials and Methods
Embryos of the freshwater oligochaete Tubifex hattai were obtained according to Shimizu (1982) and cultured at 22°C. For the experiments, embryos were all freed from cocoons in the culture medium ( Shimizu 1982). Unless otherwise stated, all experiments were carried out at room temperature (20–22°C).
Microinjection of tracer enzyme HRP
Injection micropipettes were prepared by pulling thin-walled capillaries using a microelectrode puller. Horseradish peroxidase (Sigma Chemical Co., St Louis, MO, USA; type VI-A) was dissolved at 5% in 0.2 M KCl containing 0.5% fast green, and was stored at – 20°C ( Weisblat et al. 1984 ). This dye inclusion allowed us to monitor the progress of the injection. To sterilize their surface, cocoons were treated with 0.02% chloramine T (Wako Pure Chemical Co., Osaka, Japan) for 3 min and washed thoroughly in three changes of the culture medium.
Embryos were freed from the cocoon and removed from vitelline membranes on a 2% agar bed. For injection, the embryos were placed in shallow holes made in the agar layer and carefully orientated with target cells upward. Target cells were impaled with micropipettes that had been backfilled with HRP solution, and a small volume (~4% of the cell volume) of the solution was forced into the cells by pressure.
Horseradish peroxidase-injected embryos were transferred to Petri dishes with a 2% agar bed and the culture medium containing antibiotics (penicillin G and streptomycin, 20 units/mL each) and allowed to develop at 22°C. The culture medium containing antibiotics in the Petri dishes was renewed daily.
Detection of HRP-containing cells
Embryos were fixed with 1% glutaraldehyde in phosphate buffer (40.5 m M Na2HPO4, 9.5 m M NaH2PO4.2H2O) for 1 h and washed with phosphate buffer containing 0.5% Triton X-100. The embryos were then incubated for 30 min in phosphate buffer containing 0.025% diaminobenzidine. Color development was carried out in phosphate buffer containing 0.025% diaminobenzidine and 0.01% hydrogen peroxide for 5–10 min. Horseradish peroxidase-containing cells became brown colored by this treatment.
Stained embryos were dehydrated in ethanol and cleared in a mixture of one part benzyl alcohol and two parts benzyl benzoate, and then mounted in the mixture for observation. In order to closely examine the distribution of the progeny of HRP-injected cells, some embryos were embedded in epoxy resin and serially sectioned with glass knives. Both whole-mount preparations and thick epoxy-resin sections were observed with Nomarski differential interference contrast optics.
General features of HRP-injected Tubifex embryos
Blastomeres that had been injected with 5% HRP continued cell divisions according to the normal pattern. When HRP-injected embryos were fixed at various times following HRP injection, the brown color was found to be confined to the injected cell itself or its descendants, not to their adjacent cells, suggesting that HRP does not pass through gap junctions between cells of the Tubifex embryo. In all stages examined, cell nuclei were labeled more intensely than any other portion of the cells; nevertheless, development proceeded normally in tracer-injected embryos. Similar intense nuclear staining of HRP-containing embryonic cells has also been reported in other animals ( Nishida & Satoh 1983; Weisblat et al. 1984 ).
Judging from the intensity of nuclear staining, injected HRP appeared to remain active within embryonic cells for at least 7 days following injection; color development of HRP activity was sufficiently detectable at 7 days after injection. It is unlikely that injected HRP is digested, inactivated or diluted so much in developing Tubifex embryos. However, it should be mentioned that in embryos at 8 days or later (after oviposition), endogenous peroxidase activity becomes detectable in developing blood vessels and setae. Unlike injected HRP, however, this endogenous activity never exhibits nuclear localization but appears as diffuse staining. In the present study, we used such stained setae as landmarks for the center of each body segment.
Summary of teloblastogenesis and germ band formation during Tubifex development
A brief review of early development in Tubifex is presented here as a background for the observations described later (for details, see Shimizu 1982; Goto et al. 1999 ). Precursors of teloblasts are traced back to the second (2d) and fourth (4d) micromeres of the D quadrant. At the 22-cell stage, 2d11 (resulting from the unequal divisions of 2d), 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;Fig. 1B). 2d111 (Resulting from the unequal division of 2d11) divides into a bilateral pair of ectoteloblast precursors, NOPQl and NOPQr ( Fig. 1C). Ectoteloblasts arise from an invariable sequence of divisions of cell NOPQ on both sides of the embryo ( Fig. 1K; A. Arai & T. Shimizu, unpubl. data, 1998). NOPQ undergoes unequal divisions twice after its birth and then divides into the smaller N teloblast and the larger cell OPQ ( Fig. 1D). Similarly, after producing small cells twice, OPQ divides into the smaller Q teloblast and the larger cell OP ( Fig. 1E). Finally, OP undergoes unequal division four times after its birth and then cleaves almost equally, yielding the ectoteloblasts O and P ( Fig. 1F); then teloblastogenesis reaches completion.
After their birth, each of the teloblasts thus produced divides repeatedly to give rise to small cells called primary blast cells, which are arranged into a coherent column (i.e. a bandlet). Bandlets from N, O, P and Q teloblasts on each side of the embryo join together to form an ectodermal germ band, while the bandlet from the M teloblast becomes a mesodermal germ band that underlies the ectodermal germ band ( Fig. 1G; Goto et al. 1999 ). It should be noted that the anteriormost portion of the ectodermal germ band includes cells that have been produced, during teloblastogenesis, from ectoteloblast precursors such as NOPQ, OPQ and OP ( Fig. 1K).
The germ bands are initially located at the dorsal side of the embryo ( Fig. 1G). Along with their elongation, they gradually curve round, first toward the embryo’s side and then toward the ventral midline ( Fig. 1H). Thereafter, the germ bands on both sides of the embryo move more ventrally and finally coalesce with each other along the ventral midline ( Fig. 1I). The coalescence first occurs at the anteriormost part of the embryo, and it progresses in an anterior-to-posterior fashion.
Results and Discussion
In order to construct cell fate maps for the teloblasts M, N, O, P and Q in the Tubifex embryo, we followed the fate of progenies of each teloblast by means of intracellular injection of HRP. In the current study, teloblasts were injected with HRP shortly after their birth (see Fig. 1D–F), and more than 20 labeled embryos were examined for each teloblast.
M teloblasts: Mesodermal segmental tissues
Following their coalescence along the ventral midline ( Fig. 1I), the germ bands generated by the M teloblasts expanded toward the dorsal midline on either side of the embryo ( Fig. 1J; Goto et al. 1999 ). To examine whether cells derived from the germ bands migrate across the dorsal or ventral midline, M teloblasts on the left side of embryos were injected with HRP and cultured for 4–6 days before fixation. As Fig. 2 shows, HRP-labeled cells were confined to the left side of the embryos, and the opposite right half of the embryos remained unstained. This suggests that in normal Tubifex embryos, cells derived from the germ band do not cross either the ventral or dorsal midline.
Figure 3(A) shows a representative embryo at 6 days after HRP injection of the 4d cell, the mother cell of the pair of M teloblasts. Labeled cells were distributed throughout the body wall of the embryo. In cross- sections, it is evident that the labeled cells were confined to the mesoderm ( Fig. 3B); there was no trace of their contribution to either the ectoderm or endoderm. Conversely, unlabeled cells, which were organized in segmentally iterated clumps, were present in the mesodermal region ( Fig. 3B). As described later, these clumps correspond to setal sacs and ventral ganglia, both of which originate from the ectodermal germ bands.
The labeled cells of the body wall comprised circular and longitudinal muscle fibers and coelomic walls bounding a coelomic cavity of each segment ( Fig. 3C,D,G). By focusing through whole-mount specimens, it was confirmed that among the three layers of labeled cells, the circular muscle layer was located most externally, the coelomic walls most internally, and the longitudinal muscle layer in between. As was expected from their location in the coelomic cavity, nephridia in segments VII and VIII and primordial germ cells in segments X and XI were all found to be labeled with HRP ( Fig. 3F,G).
The M teloblast-derived cells also contributed to setal sacs and ventral ganglia ( Fig. 3A). As Fig. 3(C) shows, labeled cells in setal sacs appeared as centrally located tiny dots; these dots are probably the points to which muscles converge. Labeled cells seen in each ganglion were located in its center and appeared as a wall that divided it into anterior and posterior halves ( Fig. 3A,E). In favorable transverse sections of ganglia, these labeled cells were found to extend thin processes toward the dorsal aspect of the ganglionic midline (not shown). This may suggest that the M-derived, intraganglionic cells include neurones.
In summary, we suggest that nearly all of the mesodermal tissues in the Tubifex embryo are derived from M teloblasts. It should be noted, however, that these tissues are not produced directly from M teloblasts but arise from a spatiotemporally stereotyped sequence of divisions in primary blast cells. As demonstrated previously, each primary blast cell serves as a founder cell of each mesodermal segment ( Goto et al. 1999 ), suggesting that mesodermal tissues in each segment originate from a single primary blast cell. Apparently, primary blast cells are still pluripotent, and tissue restrictions occur during proliferation of blast cells.
General distribution pattern of ectoteloblast progenies
As the mesodermal germ bands expand toward the dorsal midline, the overlying germ bands generated by the ectoteloblasts (N–Q) also expanded, in a similar fashion, into the dorsal region and constituted a part of the body wall ( Fig. 1H–J). As Fig. 4 shows, progeny cells of the left ectoteloblasts were confined to the left side of the embryo; they did not cross the ventral midline, although some cells, especially those of P and Q lineages, appeared to cross the dorsal midline. The distribution pattern of labeled cells along the dorsoventral axis (i.e. circumferentially from the ventral to the dorsal midline) was distinct among the lineages but was reflected, to some extent, by the position of a bandlet in the germ band. Progenies of the ventralmost N bandlet were largely confined to the ventral region of each segment ( Fig. 4A). In contrast, Q-derived cells were largely distributed in the dorsal region ( Fig. 4G). The O and P lineages contributed to the lateral region as well as the dorsal and ventral regions of each segment ( Fig. 4C,E). It should be noted, however, that as described later, N-derived peripheral neurones were seen near the dorsal midline (arrowheads in Fig. 4A) and Q-derived cells in the ventral ganglion ( Fig. 4G,H). This suggests that a fraction of cells in any of the lineages undergoes circumferential migration to change their position relative to other lineages during morphogenesis of the germ band.
Along the anteroposterior axis of the embryo, progenies of a blast cell bandlet of any of the lineages exhibited a segmentally iterated distribution pattern ( Fig. 4). In all lineages, labeled cells in each segment showed almost identical distribution patterns, except for the frontmost labeled segments (viz., segments II, VI, VI and III in the lineages N, O, P and Q, respectively), in which labeled cells were confined to their posterior portion and comprised about a half of the segmental complement of progeny cells (see later).
Fates of ectoteloblast progenies: Ectodermal segmental tissues
The following descriptions of segmental distribution of labeled cells are based on examination of segments VII–X of embryos at 6 days after HRP injection of one of the left ectoteloblasts. At this stage, positions of progenies of the ectodermal germ band were almost fixed in these segments ( Fig. 5), although cells in more posterior segments were still undergoing dorsalward expansion. The results described in the following are diagrammatically summarized in Fig. 6.
N lineage A characteristic feature of the N lineage is its contribution to the ventral ganglion. Cells derived from the left N teloblast were distributed throughout the left half of each ganglion except for its anterior edge and mid region ( Fig. 4B). The N-derived extraganglionic cells in each segment comprised three peripheral neurones (which are designated n1–n3, respectively; see Fig. 6N) and approximately 10 epidermal cells. The peripheral neurones were confined to the posterior part of each segment ( Fig. 5A). One of the three neurones (n2) had migrated as far away as the dorsal midline (arrowheads in Fig. 4A; also see Fig. 5A); the remaining two neurones (n1 and n3), which were located more ventrally, both extended their axons into the posterior portion of each ganglion ( Fig. 5A). Epidermal cells were localized near the ventral midline ( Figs 4A,5A).
In segments VII and VIII, in addition to the cellular components described earlier, a small cluster of N-derived cells was found in the anterior ventrolateral region ( Fig. 4A). On the basis of their morphology and position, these clusters appeared to be nephridiopores, which are termini of nephridia that are known to occur specifically in segments VII and VIII during embryogenesis ( Meyer 1929).
O lineage In each segment, cells derived from the left O teloblast comprised ~20 ganglionic cells, five peripheral neurones (designated o1–o5, respectively; see Fig. 6O) and ~14 epidermal cells. Many of the O-lineage ganglionic cells were distributed in the anterior half of a ganglion and were organized in a single cluster; those located in the posterior half were distributed at its periphery ( Fig. 4D). In contrast, four (o2–o5) of the five peripheral neurones were distributed in the posterior portion of each segment; only one peripheral neurone (o1) was located in the anterior portion ( Fig. 5B). Two peripheral neurones, o1 and o5, were characterized by their association with three O-lineage cells whose cell type was unknown ( Fig. 5B). Epidermal cells were organized in a relatively large cluster, which was elongated circumferentially in the posterior half of each segment ( Figs 4C,5B).
P lineage The P-derived cells in each segment consisted of approximately seven ganglionic cells, seven peripheral neurones (designated p1–p7, respectively; see Fig. 6P), ~20 epidermal cells, and a cluster of ‘deep’ cells (see later). Ganglionic cells were organized in a single cluster, which was located in the central region of each ganglion ( Fig. 4F). Three of the seven peripheral neurones (p1–p3) were present at the anterior edge of each segment, three (p4–p6) in the middle region and one (p7) at the posterior edge ( Fig. 5C). Epidermal cells were usually organized in three to four separate clusters.
The P lineage was characterized by a relatively large cluster of ‘deep’ cells located under the epidermis in the ventral region ( Figs 4F,5C). Judging from the fact that ventral setae were associated with this cluster, it appeared to be a ventral setal sac.
Q lineage The Q-derived cells in each segment comprised nine ganglionic cells, seven peripheral neurones (designated q1–q7, respectively; see Fig. 6Q), epidermal cells, and a cluster of ‘deep’ cells (see later). Ganglionic cells in the anterior half of each ganglion were organized in a chain running along the dorsoventral axis, while those in the posterior half were solitary ( Fig. 4H). Peripheral neurones were located in the dorsal region. Among these neurones, two cells (q4 and q5) were connected by an axon to the ventral ganglion. Epidermal cells were also localized in the dorsal region. Under this epidermis was a cluster of ‘deep’ cells, which appeared to be a dorsal setal sac based on its position.
From the above-mentioned observations, it was apparent that teloblasts N, O, P and Q were specified as ectodermal precursors. As summarized in Fig. 6, however, segmental cellular compositions were different among these four ectoteloblast lineages. If, as in the M lineage, tissue restrictions occurred during proliferation of blast cells, these differences may arise from the differences in the division pattern of blast cells and/or the timing of tissue restrictions among the ectoteloblast lineages.
Comparison with previous studies on Tubifex embryogenesis
Present results confirm the previously held views that ectodermal and mesodermal tissues comprising body segments are mainly derived from ectoteloblasts and mesoteloblasts, respectively ( Penners 1922, 1924). However, it should be mentioned that there are several types of tissue whose embryonic origins have been incorrectly or insufficiently assumed. First, we have shown that circular muscles are derived from M teloblasts, not from ectoteloblasts. This is in contrast to Penners’ conclusion that circular muscles originate from O, P and Q teloblasts. Second, in contrast to the traditionally held view that N teloblasts are the exclusive source of ganglionic cells, the present study has shown that all of the five teloblasts (M–Q) make a topographically characteristic contribution to the central nervous system. Third, setal sacs, which were assumed to originate from N or O teloblasts, have been shown to be derived from P and Q teloblasts. Furthermore, the present study has also shown that ectoteloblasts contribute peripheral neurones to the body wall, which had not been mentioned in previous studies. Taken together, the present results suggest that all of the teloblasts in the Tubifex embryo generate more types of tissue than has been previously thought.
The present results also confirm the previous notion that the prostomium and the supraesophageal ganglion (brain) are both derived from cells other than progenies of teloblasts ( Penners 1924). Even when NOPQ or 2d111 (see Fig. 1B,C) was injected with HRP, no labeled progeny cells were found in either of these two regions (data not shown). Apparently, teloblasts and their precursors do not make any contribution to these anterior structures.
Comparison with other annelids
The fate maps of teloblasts in the Tubifex embryo are very similar to those in leech embryos ( Weisblat et al. 1980 , 1984; Torrence & Stuart 1986). In both of these animals, M teloblasts originating from the fourth micromere of the D quadrant give rise to nearly all of the mesodermal tissues and a few intraganglionic cells that appear to include neurones. The ectoteloblasts derived from the second micromere generate the nervous system (ganglionic cells and peripheral neurones) and epidermis. As mentioned earlier, patterns of cleavages for teloblastogenesis are also similar in these animals. Thus, it appears that processes leading to teloblastogenesis and specification of each teloblast are highly conserved in oligochaetes and leeches. In this connection, it is interesting to note that in leech embryos, each of the O and P teloblasts resulting from the division of an OP preteloblast (see Fig. 1E) is developmentally plastic, in that o/p blast cells assume the P fate only when they come into contact with cells of another lineage ( Huang & Weisblat 1996). Although it is presently unclear whether such ‘equivalence groups’ exist in other leeches as well, it will be of interest and importance, from a phylogenetic viewpoint, to investigate the mechanisms for restriction of developmental potency of teloblasts and their progeny in the Tubifex embryo.
While key elements of early developmental mechanisms have been conserved in clitellate annelids, there are also significant differences between oligochaetes and leeches in the details of fate maps of teloblasts. The P and Q teloblasts in the Tubifex embryo, but not in leech embryos, give rise to setal sacs. Nephridiopores are derived from N lineage in Tubifex but from O lineage in leeches ( Weisblat & Shankland 1985). During embryogenesis of Tubifex, nephridia and primordial germ cells (PGC), both of which are derived from M teloblasts, occur in segments VII–VIII and X–XI, respectively, whereas in leeches, nephridia are seen in segments II–XVIII (except for segments VI and VII) and PGC in segments VIII–XVIII ( Weisblat & Shankland 1985). In view of the notion that the developmental fate of each primary m blast cell is specified at its birth ( Gleizer & Stent 1993; K. Kitamura & T. Shimizu, unpubl. data, 1998), this may indicate that developmental programs, residing in M teloblasts, for specification of primary m blast cells are different in Tubifex and leeches. These above-mentioned differences seem to be only some of the differences existing between Tubifex and leeches, but they are sufficient to suggest that developmental fates of teloblasts have been modified during the evolutionary isolation of oligochaetes and leeches. The accumulation of such small, but distinct, modifications in teloblast lineages, together with changes in other cell lineages, may have given rise to diversity in the body plan of clitellate annelids.