Two nested gonadal inductions of the vulva in nematodes

During development, cell fates are specified by multiple intercellular signals as well as by lineage history. How does this network of cell interactions vary during biological evolution? To address this question, nematode development offers reproducibility at the cellular level and the ability to compare development at a single cell level from species to species (Ambros and Fixsen, 1987; Sternberg and Horvitz, 1981, 1982). Also, cell interactions can be demonstrated after laser ablation of specific cells in different nematode species (Félix and Sternberg, 1996; Sommer and Sternberg, 1994, 1995; Sternberg and Horvitz, 1982). The cell interactions patterning the fates of the precursor cells of the vulva are well known in Caenorhabditis elegans. We thus study vulva patterning mechanisms in other nematode species to understand how developmental mechanisms vary in evolution. Such comparison can also shed light on the evolutionary origin of the mechanisms specific to the model organism C. elegans.

The precursors of the nematode vulva are the Pn.p cells (posterior descendants of the Pn cells). Of the twelve Pn.p cells aligned in the ventral cord in early C. elegans larvae (n=1 to 12 along the anteroposterior axis), six (P3.p to P8.p) are competent to form the vulva. During normal development, three of them (P5.p to P7.p) form the vulva (reviewed by Horvitz and Sternberg, 1991). The central cell, P6.p, acquires a specific, inner fate, characterized in particular by a TTTT lineage (transverse division of its four granddaughters; Fig. 1A), and the ability of its progeny to connect to the anchor cell (AC), a specialized uterine cell that links the gonad to the vulva and is located immediately dorsal to P6.p. The symmetric patterning of the Pn.p fates occurs via multiple cell interactions: a graded inductive signal from the anchor cell of the gonad (AC) to the vulva precursor cells and lateral signaling between the vulva precursor cells. Lateral signaling has been proposed to operate in either an inductive mode (only acting from P6.p to its neighbors P5.p and P7.p), or a comparative mode (acting more strongly from P6.p to its neighbors than vice-versa) (Katz et al., 1995; Koga and Ohshima, 1995; Simske and Kim, 1995; Sternberg and Horvitz, 1986). Induction by the AC and vulval precursor cell fate specification occur in C. elegans in early L3, before P(3-8).p divide (Kimble, 1981; M. Wang and P. W. S., unpublished observations).

We find a striking evolutionary change in how the symmetric centered pattern of vulval cell fates is determined. In other nematode species, the centered symmetry of the pattern is achieved by two successive, nested gonadal inductions of the vulva precursor cells and their daughters. This allows a spatial shift of centering of the pattern in the family Panagrolaimidae.

Strains are designated by their strain number in the Caltech collection kept by L. Carta. The hermaphroditic Panagrolaimus sp. cf. hygrophilusBassen, 1940 (strain PS1732) was collected by J. DeModena near Iceberg Lake, California, in July 1994 and is kept at 20-25°C. Oscheius sp. (PS1131; closely related to O. tipulaeLam, 1971 (Sudhaus, 1993); genus is subject to change, L. Carta, K. Thomas, and P. W. S., unpublished data) was collected in Tokyo, Japan, in July 1991 by W. Wood. It is hermaphroditic. Rhabditella axei (DF5006) Cobbold, 1884 was isolated by W. Sudhaus and given to us by D. Fitch. It is gonochoristic. Both species are kept at 20°C.

For strain culture, cell lineage and laser ablation, we used standard techniques described for C. elegans in Wood (1988). The vulva lineages were determined by continuous observation during the L3 molt and early L4.

In all nematodes examined so far, vulval lineages exhibit centered symmetry (Ambros and Fixsen, 1987; Sommer and Sternberg, 1994, 1995, 1996; Sternberg and Horvitz, 1981, 1982). However, a major variation occurs in the family Pana-grolaimidae (Sternberg and Horvitz, 1982): the vulva pattern is centered between P6.p and P7.p (not on P6.p like in C. elegans) (Figs 1C, 2). The anchor cell is positioned between P6.p and P7.p, and the inner vulval fate is shared between their two central daughters, P6.pp and P7.pa. The progeny of P8.p also participate in the vulva, which is thus formed from the progeny of four (instead of three) Pn.p cells. We find centering of the vulva between two precursors in several other families including the Cephalobidae and the Strongyloidae (R. J. Sommer, M.-A. Félix, P. W. Sternberg, unpublished observations). This centering also appears to be the case in the Tylenchidae (Hirschmann and Triantaphyllou, 1967; Roman and Hirschmann, 1969).

How is this distinct centering of the vulva pattern obtained? By ablating the gonad or the anchor cell in Panagrolaimus sp. PS1732 (Family Panagrolaimidae) at different times, we found two temporally distinct signals acting on the Pn.p cells (Table 1B,C). The first signal is produced by the gonad in the early second larval stage (L2) and induces the P(5-8).p cells to divide twice in the late L3 stage and form vulval tissue. The gonad in the early L2 stage in this species comprises the two non-vulval fate outer vulval fate inner vulval fate abnormal vulval fate somatic precursor cells, Z1 and Z4 (before they divide to give rise to the AC), and the two germ-line precursors, Z2 and Z3. The result of the ablation in early L1 is the same whether the two somatic cells are ablated alone or with the two germ-line precursors (data not shown). Either Z1 or Z4 is sufficient to induce the two first rounds of vulva precursor cell divisions (Félix and Sternberg, 1996). The second inductive signal originates from the AC in late L3 and induces the central daughters of the Pn.p cells (P6.pp and P7.pa) to divide a third round (‘TT’ inner fate vs. ‘UU’ outer fate). This delayed induction of the vulval inner fate on the daughters of the Pn.p cells allows the change in centering of the pattern in between two Pn.p cells.

Which cells are competent to respond to each inductive signal? In this species (PS1732), as in Panagrolaimus sp. PS1159 (Sommer and Sternberg, 1996), P(1-4).p die in the L1 stage (P4.p survives in about 1/3 of the animals in PS1732). After ablation of P(5-8).p, P(9-11).p did not form vulval tissue (12/12 animals) and P4.p did not in the 4 animals in which it survived. Therefore only P(5-8).p are competent to respond to the first induction. Every one of their daughters appears competent to respond to the second induction (Table 1D) (although P8.pp appears less likely to do it). P6.pa and P6.pp can both adopt the ‘TT’ fate when isolated at the time of, or after, the division of P6.p, indicating that their fate has not been irreversibly determined earlier in P6.p (Table 1C,E). Specification of the inner versus outer fate thus does not occur in the Pn.p cells like in C. elegans, but in their daughters. When a single Pn.p cell is near the AC, it can generate a lineage like that generated by C. elegans P6.p (Table 1D,E).

In the genus Oscheius (Family Rhabditidae, like C. elegans), the vulva is centered, as in C. elegans, on P6.p. The lineages of P5.p and P7.p are simple: they undergo only two rounds of mitosis and, in contrast to P4.p and P8.p, their progeny participate in the vulval invagination (Sommer and Sternberg, 1995). As in Panagrolaimus, we observe two successive inductions in Oscheius sp. PS1131.In Oscheius, however, both are by the anchor cell. The first induction occurs in early L3 and restricts the vulval fates to P(5-7).p. The second induction occurs in late L3 and induces the inner fate in P6.pa and P6.pp (Table 2A, Fig. 2I). P(4-8).p are each competent to respond to the first induction (Sommer and Sternberg, 1995). All daughters of P5.p and/or P7.p are competent to respond to the second induction after ablation of P6.p in the mid-L3 stage (Table 2H).

In Rhabditella axei (Family Rhabditidae, phylogenetically closer to Oscheius thanto Caenorhabditis; Sudhaus, 1976), after AC ablation in mid-L3, P5.p and P7.p adopt their correct fate and lineage, whereas P6.p exhibits abnormal vulval fates (Table 3). We tested whether vulva precursor cells were more sensitive in this species than in C. elegans to the ablation conditions by ablating a ventral uterine cell close to the AC. This control ablation had no effect on the vulva lineages (4/4 animals). We interpret these P6.p lineages as outer cell lineages with defective anteroposterior polarity, or as incompletely specified inner fates. Induction of P(5-7).p by the AC is sufficient to specify the outer fates, whereas later induction of P6.pa and P6.pp by the AC is required for the correct specification or execution of the inner fate. Thus, in contrast to C. elegans, the AC is still necessary for correct vulval cell lineages after the divisions of the Pn.p cells.

A major variation in developmental mechanism was previously found in nematode species with a posterior vulva, such as Mesorhabditis sp. PS1179, which use mechanisms other than inductive signaling from the anchor cell to specify vulval cell fates (Sommer and Sternberg, 1994). However, the source of patterning information in these species is unknown. Here we demonstrate a unique set of cell interactions used to pattern vulval fates in other nematode species (Fig. 1). Whereas patterning occurs through multiple interactions acting on the Pn.p cells in C. elegans, it is achieved in Panagrolaimus sp. PS1732 by two successive nested gonadal inductions (i) on the Pn.p cells (specifying vulval versus non-vulval fates) and (ii) on their daughters (specifying inner versus outer vulval fates). In C. elegans, after the Pn.p cell is specified, the 1° inner vulval lineage is predominantly specified autonomously. There thus has been a transition between autonomously and non-autonomously specified lineages for the inner vulval cells. In Oscheius sp. PS1131 and Rhabditella axei (same family as C. elegans), the lineage of P6.p is not specified autonomously: it also requires a later induction by the AC.

This 2-step induction mechanism would work without the reinforcement provided by lateral signaling between the vulva precursor cells, especially in species such as Panagrolaimus sp. PS1732 in which all the cells competent to respond to the first inductive signal actually do respond: there is no need for spatially precise signal production. Yet we have no evidence against a role for lateral signaling and interactions between the Pn.p cells may contribute patterning information, particularly in Rhabditella in which P5.p and P7.p adopt a fate distinct from P6.p when the AC is ablated in mid-L3.

We speculate that the 2-step patterning mechanism is ancestral to the ‘1-step/many-interactions’ mechanism found in C. elegans, and that centering of the vulva on P6.p in Oscheius and Rhabditella is an intermediate. A weak argument in favor of this hypothesis is that this mechanism has been found in distinct nematode families. In this view, the graded action of the anchor cell signal in C. elegans could derive from the temporal inductive patterning mechanism that we have described here. Specifically, the first induction might correspond to the intermediate level of inductive signal that promotes 2° fates in C. elegans; the second induction would then correspond to the high level of inductive signal that promotes the 1° fate in C. elegans.

Irrespective of the direction of evolution, these results demonstrate an extreme variation of developmental mechanism at the level of the spatiotemporal connectivity of cell interactions, that nonetheless results in the same centered pattern of cell fates. This evolutionary variation has however two consequences.

Firstly, heterochrony in the time of induction of vulval (vs. non-vulval) fates allows the change in number of precursor cells participating to vulva formation (four or three). In Panagrolaimus sp. PS1732, the first induction occurs when P(5-8).p are positioned symmetrically anterior and posterior to the gonad primordium (Fig. 2A); consequently, an even number of cells are induced (four). In C. elegans, mechanisms operate to lock one of the Pn.p cells (normally P6.p) in a position directly ventral to the AC at the time of the induction (K. Tietze and P. W. S., unpublished observations), and an uneven number of cells (three) are induced. The timing and source of the signal can thus specify the number of precursor cells (four or three) forming the vulva.

Secondly, because the inner (vs. outer) vulval fates are specified in the Pn.p daughters, the vulva pattern can be centered in between P6.p and P7.p in the Panagrolaimidae. The sharing of the inner fate between two Pn.p cells in the Panagrolaimidae occurs because the inner ‘TT’ fate is specified in their daughters, when the AC elongates and closely apposes two of them (normally P6.pp and P7.pa; Fig. 2D). Apparent tight contact of the AC to the vulva precursor cells occurs at the 2-cell stage in the three species considered here, but only later, at the 4-cell stage in C. elegans (K. Tietze and P. W. S., unpublished observations). Reproducible centering of the second induction between P6.pp and P7.pa might involve a specific mechanism of alignment of the AC in between two non-sister cells. Altogether, this allows a slight anteroposterior shift of the pattern between the two families Panagrolaimidae and Rhabditidae and contributes to changes in vulval position.

At the molecular level, the two successive inductions might be mediated by distinct signaling pathways, or by two phases of action of one signaling pathway. In C. elegans (reviewed in Aroian and Sternberg, 1993; Horvitz and Sternberg, 1991), the AC induction is mediated by the LIN-3 ligand produced by the AC and the LET-23 receptor on the vulva precursor cells. LIN-3 is encoded as a transmembrane protein that is presumably processed and cleaved off the plasma membrane of the anchor cell (Hill and Sternberg, 1992). Lateral signaling between the vulva precursor cells is mediated by the LIN-12 receptor, the known ligands of which are membrane-bound. In species with a 2-step induction mechanism, the first induction presumably can act without cell-cell contact between the anchor cell and the induced vulva precursor cells: the ligand, as LIN-3 in C. elegans, presumably traverses the extracellular matrix. This first inductive signal would thus likely be via a non-membranebound LIN-12 ligand, or a non-membrane bound form of a LIN-3 homolog. The second induction occurs when the anchor cell is closely apposed to the induced cells; this signal could be membrane-bound, consistent with either a LIN-12 ligand or a membrane-bound form of LIN-3.

This transition between 1-step and 2-step vulval induction now provides a model to analyze at a molecular level how the connectivity and temporal execution of intercellular signaling evolves. Genetics provides one way to begin a mechanistic analysis: as for Pristionchus pacificus (Sommer and Sternberg, 1996), we have found that Oscheius sp. is amenable to genetic studies (M.-A. F. and P. W. S., unpublished data).

We thank L. Carta, J. DeModena, D. Fitch, W. Sudhaus and W. Wood for worm strains, R. Sommer, K. Tietze and M. Wang for sharing unpublished observations, and T. Clandinin, D. Dimster-Denk, M. Friedrich, J. Liu and R. Sommer for comments on the manuscript. M.-A. F. is supported by an HFSP fellowship. P. W. S. is an investigator with the Howard Hughes Medical Institute.