Reversal of cell fate determination in Caenorhabditis elegans vulval development

Multicellular development involves numerous regulatory events whereby multipotent cells adopt specific fates. The precise mechanisms by which cells are endowed with a defined set of potential fates, and the regulatory steps by which those cells are directed toward a particular differentiated fate, are not well understood. A multipotent cell that responds to intercellular signals can become irreversibly determined by those signals, such that its fate is unaltered by subsequent experimental manipulation of its environment (Slack, 1983). However, since it is not always possible to exert complete control of a cell’s environment, cells that appear determined with respect to a particular developmental signal could harbor cryptic potentiality that would be revealed if the cell were placed in a different temporal and/or spatial context. Understanding the mechanisms of cell fate determination in multipotent cells requires characterization of all the regulatory steps leading from a multipotent, undetermined state, through the transduction and implementation of developmental signals, and finally to the differentiation of specific cell types. The identification of reversible steps in the process of cell fate restriction would reveal intermediate regulatory steps between multipotency and terminal differentiation.

Vulval development in C. elegans is an excellent system for the analysis of signal transduction, cell fate determination and differentiation. The six vulval precursor cells (VPCs) are born in the first larval stage (L1) and divide approximately 20 hours later at a precise time in the L3 stage (Fig. 1A). All six VPCs have the potential to express any one of three different fates, defined by the number and type of progeny that they eventually produce. The fate of a VPC depends on intercellular signals, including a gonadal inductive signal (Sulston and White, 1980; Kimble, 1981), an inhibitory signal from the surrounding hypodermis (Herman and Hedgecock, 1990) and a lateral signal from neighboring VPCs (Sternberg, 1988; Sternberg and Horvitz, 1989). The VPCs implement these signals and adopt a specific vulval fate at approximately the L2 molt and early L3 (Sternberg and Horvitz, 1986; Greenwald et al., 1983; Ferguson et al., 1987). The daughters of the VPCs appear to be irreversibly determined with respect to the positional signals that influenced their parents (Sternberg and Horvitz, 1986; Horvitz and Sternberg, 1991; Herman and Hedgecock, 1990; Hill and Sternberg, 1992; Greenwald et al., 1983).

The heterochronic gene pathway controls the time of the VPC divisions by regulating the timing of their G₁/S transition (Euling and Ambros, 1996). In loss-of-function lin-14 or lin- 28 mutants, VPCs divide precociously, in the L2 stage, resulting in early but otherwise essentially normal patterns of vulval cell division (Ambros and Horvitz, 1984; Euling and Ambros, 1996). Heterochronic mutations do not significantly affect the spatial patterning of vulval fates, but alter the temporal expression of those fates. Thus, it appears that the heterochronic gene pathway acts independently of the spatial patterning signals to control when VPCs are competent to select a fate, divide and produce specialized vulval cells (Euling and Ambros, 1996).

The fact that mutations in heterochronic genes cause precocious vulval cell division allowed us to examine the behavior of otherwise determined VPC progeny cells in a novel temporal context, the dauer larva and post-dauer developmental pathway. The dauer larva is an alternative, developmentally arrested third larval stage formed under poor growth conditions (Golden and Riddle, 1984). Since wild-type dauer larvae arrest development at the end of the L2, all of their VPCs are undivided. However, in lin-28 mutants, precocious vulval development is underway precisely when dauer larva developmental arrest occurs and, therefore, the mutant dauer larvae contain partially developed vulvae. The presence of VPC progeny cells in the lin-28 dauer larvae offered the opportunity to explore the effects of temporary developmental arrest on the process of vulval cell fate determination.

The effects of developmental arrest and post-dauer development on the fates of precociously formed vulval cells is of interest for two principal reasons: First, a post-dauer reprogramming of vulval cells would suggest a reversibility of steps in cell fate commitment in response to the intercellular vulval patterning signals. In this paper, we demonstrate that when VPC daughter and granddaughter cells develop through the dauer pathway, they revert to a state of tripotency and signalsensitivity indistinguishable from that of normal VPCs. Second, development through the dauer larva stage corrects the temporal mis-specification of lateral hypodermal cells in heterochronic mutants (Liu and Ambros, 1991). A similar reprogramming of vulval (ventral hypodermal) cell fates would suggest that there is a temporal reprogramming activity for the entire hypodermis of dauer larvae, and perhaps also for other cell lineages. Our results suggest that (1) the regulatory changes that distinguish multipotent VPCs from their daughters and granddaughters are reversible by dauer larva developmental arrest and that (2) a regulatory mechanism associated with the dauer larva sets vulval cells to the proper temporal state for post-dauer development.

C. elegans strains were grown and maintained as described by Brenner (1974) and Wood (1988). All experiments were performed at 20°C unless otherwise noted.

The wild-type strain used was C. elegans var. Bristol strain N2. Mutant alleles are as described (Brenner, 1974; Wood, 1988; Ambros and Horvitz, 1984; Ambros and Horvitz, 1987) unless otherwise noted. Strains used are: MT1035 lin-12(n137sd n460ts), MT1397 lin14(n179ts), MT2015 lin-28(n947), MT3232 lin-10(n1390) (Kim and Horvitz, 1990), MT3639 unc-13(e1091am) lin-11(n566), MT5788 nIs2 (lin-11::lac-Z), MT5875 lin-28(n947); nIs2 (lin-11::lacZ), VT297 lin-28(n719), VT592 lin-10(n1390) lin-28(n719), VT757 lin-28(n719); lin-12(n137sd n460ts).

Dauer larva formation was induced as described (Golden and Riddle, 1984), and dauer larvae were selected by detergent treatment (Golden and Riddle, 1984) and/or morphological criteria (Singh and Sulston, 1978). Dauer pheromone was prepared as described (Golden and Riddle, 1984) with modifications from C. Bargmann (personal communication) and H. Tissenbaum (personal communication).

Vulval cell lineages were followed using Nomarski optics as described (Sulston and Horvitz, 1977). Cell lineage nomenclature is as described (Sulston and Horvitz, 1977; Sternberg and Horvitz, 1986). Pn.px denotes a daughter of a Pn.p cell, where ‘x’ refers to either the posterior or anterior daughter of P3.p-P8.p; similarly, Pn.pxx and Pn.pxxx are granddaughter and great-granddaughters, respectively, of P3.p-P8.p (Fig. 2). Lineal designations (e.g., Pn.px) do not necessarily imply other aspects of cell identity, such as cell type or state of determination or differentiation. For technical reasons, cell lineages were not followed directly as animals entered the dauer larva stage, so the pre-dauer vulval cell lineage history was inferred by anatomical analysis. Individual dauer larvae were observed using Nomarski optics. The likely cell lineage history of each vulval cell nucleus was determined from its size and position relative to the gonad, other ventral hypodermal nuclei and ventral nerve cord nuclei. Undivided VPCs were recognized by their relatively large and wellseparated nuclei. Cells related by division of a common ancestor were identified by the relatively small size and more closely grouped placement of their nuclei. One or more precocious VPC divisions were observed to have occurred prior to dauer larva arrest in approximately 45% (n=96) of lin-28(lf) dauer larvae. Occasionally, the lineage history of some cells was ambiguous, particularly when a group of three similarly sized nuclei were judged to be descendants of one VPC, but their precise lineage could not be deduced.

lin-28(n719); lin-12(n137sd n460ts) hermaphrodites were grown at 20°C, and eggs were collected, placed at 15°C or 25°C until the L1 molt or until a defined time thereafter, picked individually and examined by Nomarski optics at 20°C to confirm the animal’s developmental stage and to determine if VPC cells had divided. Each animal was then immediately placed on culture plates at the alternative temperature. Developmental stage was determined by examining animals for molting behavior and the extent of gonad development. These manipulations at 20°C took less than 5 minutes. Each temperature-shifted animal (and control animals maintained on the original stock plates) were examined at the L3 and/or L4 stages using Nomarski optics; the multivulva (muv) phenotype was scored based on the number of VPCs whose progeny exhibited vulval morphogenesis (see legend to Table 2).

Laser microsurgery was performed as described (Sulston and White, 1980; Avery and Horvitz, 1987), and the operated and mock-operated animals were transferred to food to allow further development. Since it was difficult to identify the anchor cell in dauer larvae, approximately four cells in the mid-region of the somatic gonad were killed and successful destruction of the anchor cell was confirmed later using Nomarski optics. In lin-28(n719) hatchlings, the entire 4-cell gonadal primordium was ablated. The operated and mock-operated hatchlings were grown in microtiter wells with dauer pheromone and an excess of unoperated, phenotypically distinguishable, unc-13 lin-11 animals at 25°C in order to promote efficient dauer larvae formation.

Dauer larvae were selected as described above, placed in a drop of M9 on a microscope slide, covered with a coverslip and frozen on dry ice for 10 minutes. The coverslip was then removed and the animals were fixed in acetone and incubated overnight with 0.1% Xgal (Fire et al., 1990), and 1 μg/ml DAPI (Wood, 1988). Animals were observed using Nomarski optics with white light to detect lacZ expression and by epifluorescence to observe DAPI nuclear staining. Pharyngeal expression of lin-11::lacZ provided a positive control for fixation and Xgal staining (Freyd, 1991).

Cell lineage analysis showed that the L2-specific events of the lateral and ventral hypodermal cell lineages are deleted in lin28(lf) animals, and the animals express all subsequent hypodermal developmental events, including vulval development, one stage earlier than normal (Ambros and Horvitz, 1984; Euling and Ambros, 1996; Fig. 1B). Although vulval development is precocious in lin-28 mutant animals by approximately one full larval stage, it is otherwise essentially normal with respect to the number of rounds of cell divisions and spatial patterning and execution of the three vulval cell fates (Fig. 1B; Euling and Ambros, 1996). In continuously developing lin-28 mutants, vulva morphogenesis begins normally, and then appears to fail in the final phases, resulting in a nonfunctional protruding pseudovulva.

When C. elegans larvae encounter unfavorable growth conditions, they can arrest development at the end of the L2 as dauer larvae (Riddle, 1988). When dauer larvae are placed in conditions favorable for growth, they resume development and execute L3 and L4 developmental events. The two wild-type post-dauer larval stages, PDL3 and PDL4, consist of the same developmental events as the L3 and L4 stages of continuous development (Riddle, 1988; Liu and Ambros, 1991). Dauer larva developmental arrest, followed by development through the post-dauer larval stages, has a dramatic effect on the vulval phenotype of lin-28 hermaphrodites. Whereas lin-28(lf) adult hermaphrodites that develop continuously never lay eggs (n=118 for n179; n=84 for n947), 71% (n=816) of lin-28(n719) adults, and 69% (n=881) of lin-28(n947) adults that develop through the dauer larva stage exhibit normal vulva morphogenesis and can lay eggs. This dramatically improved vulval morphology and egg-laying ability of post-dauer lin-28 hermaphrodites suggests that the process of post-dauer development suppresses the effects of precocious vulval development.

To determine how development through the dauer larva stage corrects the egg-laying defect of lin-28 animals, we examined the anatomy of lin-28(lf) dauer larvae to identify the lineal origin of their vulval cells (see Materials and Methods; Figs 2, 3B), and then followed their post-dauer vulva cell lineages. In lin-28(n719) dauer larvae, precocious vulval development was variably arrested at any of the three vulval cell division cycles (Fig. 2). In the case of VPCs that had not divided prior to dauer larva arrest, the timing of their post-dauer vulval cell divisions was normal and they expressed one of the three usual vulval fates, depending on their position (for example, P7.p and P8.p in Fig. 3B). Thus, in lin-28 animals, VPCs that did not divide prior to the onset of post-dauer development retained their VPC identity throughout dauer larva developmental arrest and post-dauer development.

Remarkably, Pn.px and Pn.pxx cells that had been generated by division of VPCs prior to dauer larva arrest also behaved like tripotent vulval precursor cells during post-dauer development. In fact, Pn.p, Pn.px, and/or Pn.pxx cells in lin-28 dauer larvae all subsequently behaved equivalently, like tripotent VPCs (Fig. 3B). Post-dauer Pn.p, Pn.px and Pn.pxx cells always expressed cell lineage patterns characteristic of the 1°, 2°, or 3° cell fates expressed by normal P3.p-P8.p cells (VPCs). In 26 lin-28(n719) animals in which post-dauer cell lineages were followed, 28/28 Pn.px cells and 3/3 Pn.pxx cells (all of which had been formed prior to dauer larva arrest) expressed post-dauer cell lineage patterns like those normally generated by VPCs. These post-dauer vulval cell lineage patterns closely resembled the usual patterns generated by VPCs by all discernible criteria, including the number and orientation of cell divisions, and the nuclear morphology and morphogenetic behavior of their progeny. Further, a lin-11::lacZ reporter gene that is specific for certain cells of 2° lineages (Freyd, 1991; G. Acton and H. R. Horvitz, personal communication) was expressed during post-dauer development of lin-28; lin11::lacZ animals in a pattern consistent with the generation of normal 2° cell fates (data not shown).

We interpret this behavior of precocious Pn.px and Pn.pxx cells during post-dauer development to reflect a dauer larvarelated process that reprograms the developmental state of these cells to that of the tripotent VPCs. It is noteworthy that the timing of post-dauer vulval development in lin-28 animals appears to be normal (Fig. 3), indicating that these reprogrammed vulval cells obey temporal signals during post-dauer development that are independent of lin-28, which controls the timing of VPC division during continuous development (Euling and Ambros, 1996).

Significantly, the specific fate expressed by a Pn.p, Pn.px or Pn.pxx cell during lin-28 post-dauer development appears to depend on its position relative to the gonadal anchor cell and not on its pre-dauer lineage history. In each post-dauer lin-28 animal observed, exactly three vulval cells generated 1° or 2° vulval lineages, and the 1° and 2° lineages were always centered under the anchor cell, as in wild-type vulval development. For example, in the lin-28(n719) dauer larva shown in Fig. 3B, a Pn.pxx cell (P6.paa) that was positioned closest to the gonadal anchor cell expressed a 1° cell lineage pattern during post-dauer development. The cells flanking P6.paa in that animal, a Pn.px cell (P5.pp) and a Pn.pxx cell (P6.pap), generated cell division patterns characteristic of the 2° fate. Vulval cells more distant from the anchor cell, P5.pa, P6.pp, P7.p and P8.p, expressed the 3° lineage pattern, which is the uninduced fate typical of a VPC located further from the anchor cell. Thus, these reprogrammed vulval cells appear to acquire the potential to express any of the three vulval fates during post-dauer development and they display the normal sensitivity to spatial cues in their cell fate specification.

During continuous development, Pn.px and Pn.pxx cells seem to be restricted to specific programs of vulval cell division and differentiation since they do not alter their fates when intercellular signals are experimentally manipulated (Sternberg and Horvitz, 1986). This determined behavior of Pn.px and Pn.pxx cells contrasts with the state of VPCs, which are tripotent cells whose fates are signal-sensitive. To further test whether the post-dauer-reprogrammed Pn.px and Pn.pxx cells are equivalent to normal VPCs, we examined whether they required an inductive signal from the gonadal anchor cell for the expression of 1° and 2° fates. During wild-type postdauer development, all VPCs expressed 3° fates in the absence of an anchor cell (Table 1), (as is the case during continuous development in the wild type (Kimble, 1981) or lin-28 animals (Euling and Ambros, 1996)). To test whether post-dauer reprogrammed vulva cells in lin-28(n719) animals exhibit a similar anchor cell dependence, we ablated the anchor cell in lin28(n719) dauer larvae and followed their post-dauer vulval cell lineages. As expected for VPCs, Pn.px and Pn.pxx cells in anchor cell ablated lin-28(n719) dauer larvae expressed 3° fates during post-dauer development (Table 1; Fig. 4). In contrast, complete induction of 1° and 2° fates during postdauer development was observed in all lin-28(n719) mockoperated dauer larvae. Thus, Pn.px and Pn.pxx cells that were generated prior to dauer larva arrest behaved like VPCs during post-dauer development by the criterion of anchor cell dependence of their 1° and 2° fates.

The same spatial signaling systems that govern vulval patterning during continuous development also appear to act during post-dauer development. A lin-3::lacZ transgene is expressed in the gonad of wild-type dauer larvae and postdauer animals (data not shown), indicating that the gonadal signal affecting post-dauer vulval cell lineages is likely the lin3 gene product, as in continuous development. lin-15, lin-3, lin-10 and lin-12 mutations all affect the patterning of wildtype post-dauer vulval fates, although the penetrance of the lin3 and lin-10 defects are reduced compared with continuous development (Ferguson and Horvitz, 1985; data not shown). Furthermore, lin-28 mutations do not prevent the expression of lin-3, lin-15, lin-10 and lin-12 vulval phenotypes during continuous development (Euling and Ambros, 1996), or during post-dauer development (data not shown). Thus, it is likely that the pattern of fates expressed by reprogrammed vulval cells of lin-28 post-dauer animals is governed by the same intercellular signals as in the wild type.

In contrast to the Pn.px and Pn.pxx cells, Pn.pxxx cells generated prior to dauer larva arrest failed to divide during post-dauer development. In 26 lin-28(n719) dauer larvae, four Pn.pxxx cells were observed to have been formed prior to dauer larva arrest. None of these four Pn.pxxx cells divided during post-dauer development. In one other lin-28(n719) dauer larva, Pn.pxxx cells were evident that had begun to undergo vulval morphogenesis. During post-dauer development these cells continued vulval morphogenesis in the PDL3 and no further cell divisions were observed (data not shown). (This animal was egg-laying defective, suggesting that the predauer occurrence of differentiated Pn.pxxx cells may account for some of the egg-laying defective lin-28 post-dauer animals). Thus, unlike Pn.px and Pn.pxx cells, Pn.pxxx cells in lin-28 animals appear to be irreversibly determined under the conditions of post-dauer development.

To test for whether the reprogramming of precocious Pn.px cells in lin-28 animals reflects the reversal of their normal state of commitment, and whether reprogramming occurs before or after dauer larva developmental arrest, we performed two experiments: First, we confirmed that Pn.px cells in lin-28 hermaphrodites are committed with respect to the lateral signal controlled by lin-12, as in the wild type. Second, we tested whether precocious Pn.px cells formed prior to dauer larvae arrest in lin-28 animals exhibit a normal response to the presence or absence of the gonadal inductive signal.

To determine whether precocious Pn.px cells of lin-28 animals are committed with respect to lin-12 activity, we performed temperature-shift experiments with lin-28(n719); lin-12(n137sd n460ts) animals and scored their multivulval (muv) phenotype. n460ts is a temperature-dependent suppressor of the lin-12(n137sd) muv phenotype (Greenwald et al., 1983; Table 2). Previous temperature-shift experiments with lin-12(n137sd n460ts) animals showed that vulval fates are affected by lin-12 activity during the early L3, only prior to VPC cell division, and that VPC progeny (Pn.px) cells appear to be irreversibly committed with respect to lin-12 activity (Greenwald et al, 1983). Similarly, we found that temperature shifts with lin-28(n719); lin-12(n137sd n460ts) animals indicated that vulval cell fates were altered by temperature shifts performed before the time of VPC cell division but were unaffected by temperature shifts performed after VPC cell division (Table 2). These results indicate that in precocious vulva development, as in normally timed vulval development, VPC progeny are committed such that they are not sensitive to lin-12 activity. Therefore, the post-dauer reprogramming of Pn.px cells in lin-28 animals likely reflects a reversal of their normal state of determination.

To determine when, with respect to dauer larva developmental arrest, vulval cells become reprogrammed, we used laser microsurgery of the gonad to assess the developmental state of precociously formed Pn.px cells in lin-28(n719) dauer larvae. VPC cell division is gonad-independent and Pn.px cell division is gonad-dependent. Thus, if the precocious Pn.px cells of lin-28 animals are reprogrammed to VPCs after dauer larva developmental arrest, and not before, then their response to the presence or absence of the gonadal inductive signal prior to dauer arrest should be like that of normal Pn.px cells. Specifically, in animals lacking a gonadal anchor cell or lin-10 activity, all VPCs, including those that normally express 1° and 2° fates, will express the 3° fate (Kim and Horvitz, 1990), which is reflected by division of the VPC to produce two nondividing Pn.px cells. Thus, if Pn.px cells of lin-28 animals retain their identity until after developmental arrest, then they are expected to be undivided in dauer larvae lacking an anchor cell or lin-10 activity. Alternatively, if Pn.px cells become reprogrammed to the VPC fate prior to dauer larva developmental arrest, then they are expected to act like multipotent VPCs, and thus could divide prior to arrest in dauer larvae lacking the gonad or lin-10 activity.

The gonad was ablated in lin-28 animals at hatching and the animals were allowed to form dauer larvae (Materials and Methods). Also, lin-10 lin-28 dauer larvae were selected. Both types of dauer larvae were then examined using Nomarski optics to determine whether P5.px-P7.px cells had divided prior to dauer larva arrest (Materials and Methods). In unoperated (AC+) control lin-28(n719) dauer larvae, we found that approximately 25% of P5.px-P7.px cells had divided prior to dauer larva arrest (Table 3). In contrast, a statistically significant decrease in the proportion of P5.px-P7.px cell divisions was observed in lin-10 lin-28 and anchor cell-ablated lin-28 animals (Table 3). Specifically, in lin-28(n719) dauer larvae lacking either lin-10 activity or a gonadal signal (combining experiments 3 and 4, Table 3), less than 4.5% of the Pn.px cells appeared to have divided prior to dauer larvae arrest. The significant decrease in Pn.px cell divisions observed upon removal of the gonadal signal or lin-10 activity indicates that the majority, and perhaps all (see Table 3 legend), of the precociously generated pre-dauer Pn.px cells in lin-28 dauer larvae exhibited a normal Pn.px cell identity prior to developmental arrest. This result is consistent with the conclusion that precocious Pn.px cells are determined towards their fates and that they become reprogrammed to VPCs after dauer larva arrest.

We examined lin-11::lacZ expression in lin-28 dauer larvae in order to assess the developmental state of certain Pn.pxx cells in dauer larvae. In wild-type animals carrying an integrated copy of the lin-11::lacZ reporter gene, β-galactosidase is expressed in particular Pn.pxx cells of 2° lineages (Freyd, 1991). Of 100 lin-28(n947); lin-11::lacZ dauer larvae that were examined, none exhibited β-galactosidase activity in vulval cells (data not shown). Based on the frequency of P5/7.pxx cells in lin-28 dauer larvae and the frequency of lin11::lacZ expression in P5/7.pxx cells of lin-28 animals (data not shown), approximately 10% of these 100 dauer larvae were expected to contain lacZ-expressing P5/7.pxx cells. This apparent repression of lin-11::lacZ expression in lin-28 dauer larvae may reflect an inhibition of vulval gene expression that may contribute to the reprogramming of vulval cells.

Previous studies of continuously developing wild-type larvae showed that VPCs are tripotent, while their Pn.px and Pn.pxx progeny are determined with respect to positional information (Sulston and White, 1980; Kimble, 1981; Greenwald et al., 1983; Sternberg and Horvitz, 1986). Two observations suggest that the precocious Pn.px and Pn.pxx cells of lin-28 animals are also positionally determined. First, the fates expressed by precocious vulval cells in lin-28 animals exhibit the same sensitivity to the gonadal inductive signal and the same involvement of genetically identified signalling molecules as does wild-type vulval development (Euling and Ambros, 1996). Second, here we show that the time of action of a temperaturesensitive lin-12(sd) mutation is prior to VPC cell division in lin-28 precocious animals. This indicates that, with respect to the lateral signal controlled by lin-12, the fates of precocious Pn.px and Pn.pxx cells are established prior to VPC cell division, as in the wild type.

We found that, if Pn.px and Pn.pxx cells are placed in the novel temporal context of the dauer larva and post-dauer development using a lin-28 mutation, they become reprogrammed to the tripotent, signal-dependent VPC state. This conclusion is based on three observations of the behavior of precociously formed vulval cells during post-dauer development of lin-28 mutants: (1) they divide during the first post-dauer stage (PDL3) at the same time as do wild-type VPCs; (2) they each express one of the three VPC-characteristic (1°, 2° or 3°) cell lineage fates and (3) they express those fates in response to intercellular signals in a manner indistinguishable from normal VPCs.

The observation that, although Pn.px and Pn.pxx cells are positionally determined during continuous development, they nevertheless become reprogrammed to VPCs after dauer larva arrest suggests that the steps in cell fate restriction of Pn.px and Pn.pxx cells are reversible in the wild type. By contrast, Pn.pxxx cells were not reprogrammed upon post-dauer development. These findings suggest that Pn.px and Pn.pxx cell fate determination involves steps in vulval cell fate restriction intermediate between the multipotency of VPCs and the terminal differentiation characteristic of Pn.pxxx cells. It is striking that vulval cells (specifically Pn.px and Pn.pxx cells) that are never found in wild-type dauer larvae are always reprogrammed to VPCs if they are ‘temporally transplanted’ to dauer larvae of precocious mutants. This suggests that Pn.px and Pn.pxx cells may normally possess cryptic VPC character that is unmasked when the vulval developmental program is interrupted in dauer larvae.

Reversal of cell determination has been observed in other systems. Myogenic differentiation appears to be reversible during vertebrate limb regeneration (Lo et al., 1993). Similarly, yeast cells can switch from the meiotic program back to the less differentiated state of mitotic growth in response to temporary cell cycle arrest (Honigberg and Esposito, 1994). While it is unclear what these other situations may have in common with the vulval cell reprogramming, perhaps in general the interruption of developmental and cell division programs, combined with specific novel developmental signals, can lead to the reversal of cell fate restriction. The post-dauer reprogramming of precocious vulval cells reported here, together with the postdauer reprogramming previously observed for lateral hypodermal cells of C. elegans heterochronic mutants (Liu and Ambros, 1991), suggest that a global activity sets cells of the worm to the proper L3-equivalent temporal identity during the PDL3.

The reprogramming exhibited by precocious vulval cells was observed only during post-dauer development of lin-28 animals, not during continuous development, or under conditions of starvation without dauer larva differentiation (data not shown). The reprogramming is not simply a result of the absence of lin-28 activity, since vulval cells of lin-14 mutants exhibit a similar post-dauer reprogramming (data not shown). However, because we observed vulval cell reprogramming only in dauer larvae of heterochronic mutants, we cannot rule out the possibility that the phenomenon reflects a role for the heterochronic gene pathway in maintenance of vulval cell developmental identity specifically in dauer larvae. We do not favor this latter interpretation because it requires that one assume that lin-28 and lin-14 activities are required to maintain a Pn.px or Pn.pxx identity exclusively during the dauer larva stage, and not during continuous development or during the execution of post-dauer development. We favor the simpler interpretation that reprogramming is unrelated to lin-28 or lin-14 per se, and instead reflects an activity present in wild-type animals that is revealed by the post-dauer behavior of precocious vulval cells in these mutants.

Three observations indicate that reprogramming to the multipotent VPC occurs after dauer larva developmental arrest. First, the patterns of pre-dauer cell divisions were inconsistent with reprogramming prior to dauer larva arrest; we never observed greater than three rounds of vulval cell divisions before dauer larva arrest in lin-28 animals. Second, analysis of lin-28 dauer larvae that lacked an anchor cell signal or lin-10 activity suggests that pre-dauer Pn.px cell division is gonad-dependent, as is the case for normal Pn.px cells, but unlike VPCs. These results indicate that Pn.px cells become reprogrammed to the VPC state only after dauer larva arrest – perhaps in the arrested dauer larvae itself, or during early post-dauer development. Third, we observed that all Pn.pxxx cells generated prior to dauer larva arrest exhibited the morphology and non-dividing behavior characteristic of terminally differentiated vulval cells. The observation that the Pn.pxxx cell identity can be established in lin-28 dauer larvae suggests that pre-dauer vulval lineages in these mutants proceed normally up to the point of developmental arrest and supports the view that reprogramming of Pn.px and Pn.pxx cells to VPC cells occurs after arrest.

Our finding that lin-11::lacZ expression was not detected in lin-28(n947); lin-11::lacZ dauer larvae is consistent with the notion that reprogramming of vulval cells, which would be expected to result in repression of lin-11 expression, occurs in the arrested dauer larva. However, since the dauer larva is characterized by a global decrease in transcription (Dalley and Golomb, 1992), the absence of lin-11::lacZ expression could reflect a general suppression of gene activity in dauer larvae that may or may not be related directly to the reprogramming of vulval cells in particular.

There are numerous models for the kinds of regulatory changes that might underlie the observed post-dauer reprogramming of vulval cells. In general, it seems that a reversal of vulval cell determination would require a despecification of vulval cells and a reactivation of the VPC tripotent state (Fig. 5). This process could occur in two steps. First, all vulval cells that have not become terminally differentiated (Pn.p, Pn.px and Pn.pxx cells) lose their cell fate specification upon dauer larva arrest. Second, for example at the onset of post-dauer development, these despecified cells are susceptible to a post-dauer ‘L3-specification’ activity such that they acquire the tripotency and intercellular signal-dependence typical of L3-stage VPCs. In contrast, reprogramming could occur by a one-step process, where the despecification of a determined fate in the dauer larva results directly in the unmasking of the tripotent VPC state. Perhaps the tripotent VPC identity is a ground state for the vulval cell lineage that is expressed by Pn.p cells during continuous development when the lin-14 pathway is down-regulated (Euling and Ambros, 1996) and can be re-expressed by Pn.p progeny if the vulval differentiation program is repressed in dauer larvae.

It is likely that reprogramming would involve changes in vulval-specific gene expression. For example, in the wild type, lin-12::lacZ expression is relatively high in Pn.pxx cells of 2° lineages and low in P6.pxx cells (Wilkinson and Greenwald, 1995). This means that, if a precocious P6.pxx cell of a lin-28 dauer larva becomes reprogrammed to a VPC fated to express the 2° fate (for one such observed case, see Fig. 3B), lin-12 expression would need to become reactivated in that cell during post-dauer development. Similarly, if lin-11 is expressed in precocious 2° cell lineages prior to dauer larva arrest, it would likely become inactivated in those cells if they subsequently become reprogrammed to VPCs that express a non-2° cell fate.

The mechanism by which vulval cells are reprogrammed to VPCs may be related to the cell cycle arrest and/or suspended development of dauer larvae. It is likely that cells of the developing larva enter the resting phase (G₀) of the cell cycle upon dauer larva arrest, since no cell divisions occur in dauer larvae and they are metabolically depressed (reviewed in Riddle, 1988; Wadsworth and Riddle, 1988). Physiological change(s) related to G₀ cell cycle arrest may contribute to the reprogramming of Pn.px and Pn.pxx cells, perhaps by repressing steps in vulval differentiation. A detailed understanding of the reprogramming mechanism requires a genetic and molecular analysis of the reprogramming process and a precise characterization of the cell cycle state of vulval cells in dauer larvae.

We propose that the reprogramming mechanism acting on precocious vulval cells reflects a process that affects the developmental state of many or all cell types during dauer larva arrest and postdauer development (Liu and Ambros, 1991). A reprogramming mechanism may serve to coordinate the temporal patterns of development in the animal after an indefinite period of suspended development as a dauer larva. According to this view, the vulval cell reprogramming that we describe here for precocious vulval cells would reflect a normal process acting in the vulval lineage of wildtype dauer larvae. Specifically, in wild-type dauer larvae, the reprogramming activity may prevent VPCs from executing prematurely steps in vulval differentiation. Perhaps developmental arrest would otherwise predispose VPCs in wild-type dauer larvae to execute prematurely, in the absence of cell division, one or more steps toward vulval cell differentiation. Without a mechanism for reversing the state of such aberrantly differentiated cells to the proper VPC state, post-dauer vulval development would be abnormal. Perhaps precursor cells in other systems, such as other organ primordia, employ analogous regulatory mechanisms for actively reinforcing the proper developmental state during periods of developmental quiescence.

We thank Jim Dykes for advice with statistical analysis, Beth Sawin and Bob Horvitz for the use of a laser microbeam and microscope, Gwen Acton and Bob Horvitz for providing a lin-11::lacZ strain, Paul Sternberg for providing lin-3::lacZ strains, and Ann Rougvie, Min Ku and members of our laboratory for helpful discussions. This work was supported by an NIH predoctoral fellowship and Paul Mazur Fellowship (through Harvard to S. E.) and a US Public Health Service research grant GM34028 (V. A). Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR).