The Caenorhabditis elegans gene lin-44 controls the polarity of asymmetric cell divisions

The generation of polarity during metazoan development requires the formation of asymmetric cells and multicellular structures and the orientation of these cells and structures with respect to the rest of the organism. Much of what is known about the generation and orientation of polar structures comes from the study of single cells, the budding yeast Saccharomyces cerevisiae and the syncytial embryo of the fruit fly Drosophila melanogaster.

In budding yeast, genes involved in bud formation function to establish or maintain cell asymmetry (Hartwell et al., 1973, 1974; Adams et al., 1990; Chant et al., 1991). The orientation of this asymmetry depends both upon cell interactions and upon a set of genes that act to define budding pattern. Specifically, during the process of mating, yeast cells orient to the neighboring cell producing the highest level of mating pheromone and generate a projection toward that cell (Jackson and Hartwell, 1990; Jackson et al., 1991). Genes that determine where a new bud is formed include those that seem to act to localize proteins required for bud formation to a specific site within the cell (Chant et al., 1991; Chant and Herskowitz, 1991; Snyder et al., 1991; Ruggieri et al., 1992).

During the development of the single-celled embryo of Drosophila melanogaster, anteroposterior polarity is generated by intracellular gradients of the bicoid and nanos morphogens, while dorsoventral polarity is established by a localized signal from outside the egg cell (reviewed by St. Johnston and Nusslein-Volhard, 1992). bicoid determines anterior pole identity, whereas nanos determines posterior pole identity. The protein product of staufen is required for the final localization of bicoid mRNA (St. Johnston et al., 1991) as well as for the localization of oskar mRNA, which in turn determines the localization of nanos mRNA (Ephrussi et al., 1991). The products of two additional genes, cappuccino and spire (Manseau and Schüpbach, 1989), are required for the localization of both oskar mRNA (Ephrussi et al., 1991) and staufen protein (St. Johnston et al., 1991). How cappuccino and spire cause such localization is unknown.

Little is known about how cell polarity becomes oriented in multicellular organisms. After the generation of the singlecelled zygote, metazoan development involves extensive cell proliferation that results in an adult organism with a large number of cells and a diversity of cell types. Many of the cell divisions involved in this process can be regarded as asymmetric, since they produce sister cells that have distinct fates (see review by Horvitz and Herskowitz, 1992). What are the mechanisms responsible for the asymmetry of such cell divisions, and how does the asymmetry become oriented? To address these questions, we have been studying the cellular and genetic control of the cell lineage of the nematode Caenorhabditis elegans. The complete cell lineage of C. elegans is known (Sulston and Horvitz, 1977; Kimble and Hirsh, 1979; Sulston et al., 1983). Of the 949 non-gonadal cell divisions that occur during the development of the C. elegans hermaphrodite, 807 are asymmetric, generating sister cells that differ in their fates (Horvitz and Herskowitz, 1992). We have discovered a mutation that does not disrupt the generation of asymmetry in this cell lineage but causes the orientation of some of these cell divisions to be reversed with respect to the body axis of the animal. This mutation defines a new gene, lin-44 (lin, cell lineage abnormal).

Nematode strains were cultured by standard techniques (Brenner, 1974; Sulston and Hodgkin, 1988). N2 is the wild-type parent of all strains used in this work (Brenner, 1974). Mutations used are described by Hodgkin et al. (1988) or are noted below. Alleles used were: LGI: lin-6(e1466), unc-11(e47), unc-73(e936), unc-89(e1460), let-353(h46), let-504(h448), let-505(h426), let-506(h300), let507(h439), let-503(h313) (Howell et al., 1987; Howell and Rose, 1990), unc-74(e883), unc-57(e406), dpy-5(e61), unc-13(e450), unc-54(e190); LGV: him-5(e1490), LGX: mec-2(e75), dpy-7 (e1364ts), sup-7(st5cs), unc-9(e101). Rearrangements used were: LGI: hDf6, hDf7 (Howell and Rose, 1990), sDp2, hDp2, hDp7 (McKim and Rose, 1990); LGII: mnC1 dpy-10(e128) unc-52(e444); LGIII, V: eT1(III, V); LGIV, V: nT1(IV, V). Male characteristics were studied using strains containing the him-5(e1490) mutation, which increases the frequency of spontaneous male progeny (Hodgkin et al., 1979).

him-5 hermaphrodites were mutagenized with ethyl methanesulfonate (EMS; Brenner, 1974), individual F₁ progeny were transferred to a Petri plate, and their progeney were screened. The mutation lin44(n1792) was identified on the basis of gross defects in the structure of the male tail (Fig. 1A). The second lin-44 allele, n2111, was isolated in a screen for mutations that failed to complement lin44(n1792) for the phasmid dye-filling defect in hermaphrodites (see below). unc-57 I; unc-9 X hermaphrodites were mutagenized with EMS and mated with unc-73 lin-44(n1792)/ + + males. About 20,000 F₁ hermaphrodite progeny were screened for defects in phasmid DiOfilling. Since lin-44(n1792)/hDf6 and lin-44(n1792)/hDf7 strains are viable and have defective phasmids (see Table 1), we could identify and recover complete loss-of-function alleles in this way.

The phasmids and amphids are tail and head sensory structures, respectively, that are exposed to the external environment (Ward et al., 1975; Ware et al., 1975; Hall and Russell, 1991). When wild-type hermaphrodites are soaked in fluorescent dyes, such as fluorescein isothiocyanate (FITC) or 3,3′-dioctadecyloxacarbocyanine (DiO), the two neurons of each phasmid as well as six neurons of each amphid fill with dye and can be visualized in living animals (Hedgecock et al., 1985 and personal communication). Briefly, adult animals were soaked in a 10 μg/ml solution of DiO in M9 salts for 2 hours, and L1 larvae were placed on agar plates containing 0.1 mg/ml FITC for 4 hours (Hedgecock et al., 1985). For the non-complementation screen and strain constructions (see below), dye-filled animals were examined on Petri plates using a Wild M420 macro-dissecting microscope equipped with epifluorescence. For quantitation and microphotography, dye-filled animals were mounted on a slide in a small drop of M9 salts (Brenner, 1974) on a 5% agar pad containing 10 mM sodium azide as an anesthetic and examined on a Zeiss Standard or Axioplan microscope equipped with epifluorescence. Generally, the amphids and phasmids of N2 animals filled with dye; by contrast, in lin-44 animals the amphids filled but the phasmids often did not. Occasionally, both the amphids and phasmids of an N2 or lin-44 animal, usually a young hermaphrodite, failed to fill with dye. For this reason, to score the phasmid defect of lin-44 animals, we counted as phasmid-defective only those individuals that showed amphid filling but no phasmid filling.

lin-44(n1792) was mapped to LGI (data not shown) (Fig. 1B). Threefactor crosses further defined the map position of lin-44. From heterozygotes of genotype unc-73 + dpy-5/ + lin-44 +, 16/18 Dpy nonUnc and 0/13 Unc non-Dpy recombinants segregated lin-44. From heterozygotes of genotype + unc-74 dpy-5/lin-44 + +, 11/11 Dpy non-Unc recombinants and 0/17 Unc non-Dpy recombinants segregated lin-44. From heterozygotes of genotype unc-89 + dpy-5/ + lin44 +, 0/15 Unc non-Dpy and 13/13 Dpy non-Unc recombinants segregated lin-44. From heterozygotes of genotype + lin-44 dpy-5/unc-89 + + 5/5 Dpy non-Lin and 0/22 Lin non-Dpy recombinants segregated unc-89. Thus, lin-44 maps very close to unc-89 I (Fig. 1B).

The deficiencies hDf6 and hDf7 failed to complement lin44(n1792) and lin-44(n2111) (see Fig. 1B, Table 1). The duplications hDp2 and hDp7 were tested for complementation of lin-44 by constructing strains of general genotype lin-44 dpy-5; hDp and scoring hermaphrodite phasmid dye-filling. Both of the duplications hDp2 and hDp7 failed to complement lin-44 (data not shown) (Fig. 1B). hDp2 covers let-503 and let-507 and hDp7 covers let-503 (see below) but neither covers unc-89 (McKim and Rose, 1990) Since hDf7 uncovers let-503 and let-507 but not unc-89 (Howell and Rose, 1990), lin-44 maps to the right of unc-89 (Fig. 1B).

lin-44(n1792) dpy-5/+ + males were mated with hermaphrodites of general genotype let dpy-5(e61) unc-13(e450); sDp2, and the Dpy cross progeny were scored for the phasmid dye-filling defect. Mutations in all known let genes in the hDf7 interval (Howell and Rose, 1990) (Fig. 1B) complemented lin-44(n1792) (data not shown).

To subdivide the hDf7 interval, we tested the let genes for complementation with the duplications hDp2 and hDp7 (data not shown). These experiments revealed that hDp7 complemented only let-503, and hDp2 complemented only let-503 and let-507. Thus, let-353, let504, let-505, let-506, and lin-44 map to the interval of hDf7 not covered by hDp2 (Fig. 1B).

Both lin-44 alleles were tested for suppression by the amber suppressor sup-7(st5cs) X (Waterston, 1981). The sup-7(st5cs) mutant is inviable at 15°C, and this mutation causes stronger suppression at 20°C than at 25°C. The dpy-7(e1364ts) X mutation, not an amber, is heat-sensitive: at 25°C animals are dumpy (Dpy) and at 20°C only slightly Dpy. To test for single-copy suppression of lin-44(n1792), we constructed a lin-44(n1792); + dpy-7 sup-7/mec-2 + + strain. We tested for suppression by scoring the non-Mec animals for suppression of lin-44 at 20°C. Each animal scored was picked to ascertain its genotype based on the progeny it segregated; those animals that segregated Mec progeny were counted as being + dpy-7 sup-7/mec-2 + + (Table 1). From hermaphrodites of genotype lin-44(n1792)/+; dpy7/+, 6/12 Lin progeny segregated Dpy animals and 12/12 Dpy progeny segregated Lin animals, indicating that the suppression was not caused by the presence of the dpy-7(e1364) allele. The penetrances of the phasmid defect for suppressed and non-suppressed strains are shown in Table 1.

The lin-44(n2111) allele was suppressed by neither one nor two copies of sup-7(st5cs). From hermaphrodites of genotype lin44(n2111); dpy-7 sup-7/+ +, 8/12 Lin progeny segregated Dpy animals and 8/10 Dpy progeny segregated Lin animals. A lin44(n2111); dpy-7(e1364ts) sup-7(st5cs) strain was constructed, and animals of this genotype were also Lin (Table 1).

Living animals were observed using Nomarski differential interference contrast optics; cell nomenclature and cell lineage analysis were according to Sulston and Horvitz (1977). Nuclei were destroyed by a laser microbeam, as described by Avery and Horvitz (1987). For kills of AB.prappppap in the embryo, cells were identified using cell deaths and other landmarks provided in the diagrams of Sulston et al. (1983). Successful embryonic kills were verified by the absence of the descendants of the cell(s) killed in the L1 animal. Only those animals that showed no other morphological defects were studied further.

The polarity of the T cell division can be assessed in L1 animals after the time when all four T cell granddaughters are present by the nuclear morphologies of T.aa and T.ap and by the nuclear morphologies of T.pa and T.pp or by the presence of the neural cells they generate. In wild-type animals, T.aa has the nuclear morphology of a hypodermal syncytial cell (a large oval-shaped nucleus with a flattened appearance and a large nucleolus), T.ap has the nuclear morphology of a hypodermal blast cell (a large round nucleus with large nucleolus and smooth nucleoplasm), T.pa and T.pp prior to their divisions each have the nuclear morphology of a neuroblast (a large round nucleus with a smaller nucleolus than a hypodermal cell and granular nucleoplasm), and the descendants of T.pa and T.pp all have the nuclear morphologies of neural cells (small nuclei with granular nucleoplasm and most have no visible nucleolus) (Fig. 6A). T.pppp undergoes programmed cell death and has a characteristic refractile and flattened appearance. In hermaphrodites the cell T.apap is a seam cell, a specialized hypodermal cell with a nuclear morphology similar to a hypodermal blast cell but that displays an activity at the end of each intermolt period characterized by the appearance of refractile blobs in the surrounding cytoplasm. In both males and hermaphrodites the neuron PVN (T.appp) has the nuclear morphology of a neural cell, but has a small nucleolus.

Animals homozygous for hDf7 hatch and arrest development as L1 larvae (Howell and Rose, 1990), suggesting that the complete loss of the zygotic functions of genes in this interval, including lin-44, does not cause embryonic lethality. hDf7 dpy-5(e61) unc-13(e450) animals were generated by mating N2 males with hDf7 dpy-5 unc-13; sDp2 hermaphrodites, picking semi-Dpy non-Unc progeny of presumptive genotype hDf7 dpy-5 unc-13/+ + + (dpy-5(e61)/+ animals are semiDpy, and animals of genotype dpy-5/+/+, which carry sDp2, are nonDpy). From these semi-Dpy non-Unc heterozygotes, we picked Dpy Unc arrested larvae (genotype hDf7 dpy-5 unc-13). 57% (n=453) of hDf7 homozygotes arrested as L1 larvae without any further division of T or, based upon examination of five of these arrested larvae, of any other postembryonic blast cells. The remaining hDf7 homozygotes arrested after varying numbers of cell divisions; based upon the number of cells and cell types produced, no animals progressed beyond the L1 stage (our unpublished observations; also Howell and Rose, 1990). The time between cell divisions in hDf7 homozygotes was much longer than that in wild-type or lin-44 mutant animals.

The T granddaughters are generated approximately 7.5 hours after hatching, and their differences in nuclear morphologies become apparent about 0.5 hours later. In wild-type animals, the nucleus of the cell P6 descends into the ventral cord about 8 hours after hatching, and we used the presence of P6 in the ventral cord as a marker for this time point. Thus, for wild-type and lin-44 animals scored in Table 2, we scored the morphologies of the T granddaughter nuclei only if the P6 nucleus was in the ventral cord. However, in hDf7 homozygotes the entry of P6 into the ventral cord was not always a reliable indicator of the developmental stage of the T cell lineage so that in some animals in which P6 had migrated the T cell granddaughters might not have been old enough for differences in nuclear morphologies to have become apparent. For this reason, the hDf7/hDf7 animals in Table 2 were transferred to a separate plate 24 hours prior to being scored, ensuring that each hDf7 homozygote examined was at least 24 hours old. In hDf7/hDf7 animals in which the T cell lineage was followed, the differences in the nuclear morphologies of the T cell granddaughters, if generated, were apparent by 24 hours.

To identify mutants abnormal in the development of the C. elegans male tail, we screened the F₂ progeny of individual him-5 hermaphrodites mutagenized with ethyl methanesulfonate; the him-5 mutation was used because it results in the generation of a high frequency of male XO self-progeny from XX hermaphrodites (Hodgkin et al., 1979). We isolated a mutation, n1792, that caused gross defects in the male tail (Fig. 1A). This mutation defined the gene lin-44 (see Materials and Methods). As described below, lin-44(n1792) hermaphrodites have defective phasmids. Because this defect can be scored in living hermaphrodites and is highly penetrant (Table 1), we used it for genetic analysis and for the isolation of a second lin-44 allele, n2111 (see Materials and Methods). The tail defect of lin-44 males is much more difficult to use for two reasons: first, all analyses must be done in a him mutant background (Hodgkin et al., 1979) to generate males; and second, lin-44 males cannot mate, so that to recover and maintain new mutations and genetic combinations identified in males, hermaphrodite siblings must be picked and their male progeny scored. Like lin-44(n1792) males, lin-44(n2111) males have defective tails. We mapped lin-44 to a small interval on linkage group I (see Materials and Methods) (Fig. 1B).

The tail defect of lin-44 males led us to examine the cell lineages of the male-specific blast cells in the tail (Fig. 1C). In the wild-type male the F cell undergoes a left-right symmetric division during the mid-L2 stage, generating two equivalent sister cells that subsequently undergo identical lineage patterns. During the L3 stage, the left and right F daughters F.l and F.r divide asymmetrically, and each produces four descendants (Fig. 2). We studied the F cell lineages of nine lin-44 males. In two animals we observed division patterns that were mirror symmetric to those of the wild type (Fig. 2). One plausible explanation is that the polarity of the first asymmetric cell division was reversed, causing the polarities of all subsequent asymmetric cell divisions also to be reversed. One other animal appeared to have a single polarity reversal of the F.l division, and two other animals had lineages with multiple apparent polarity reversals (Fig. 2). We also observed one animal with a loss of asymmetry in the F cell lineage, in that both F.l and F.r underwent two rounds of dorsal-ventral divisions generating a total of eight neuronal cells, and three animals with wild-type F cell division patterns.

In the wild-type male, the B cell divides to generate neurons and structural cells that constitute a large portion of the tail (Sulston and Horvitz, 1977; Sulston et al., 1980). B divides asymmetrically during the late L1 stage to generate a large anterodorsal cell B.a and a smaller posteroventral cell B.p. By the end of the L2 stage B.a has undergone three rounds of division and B.p only one, generating a total of ten cells that lie in characteristic positions (Fig. 2). During the mid-L3 stage these ten cells begin to divide, eventually generating a total of 42 descendants.

We followed the B cell lineage in eleven lin-44 males through the end of the L2 stage. In eight lin-44 males B.a was small and underwent one round of division, and B.p was large and underwent three rounds (Fig. 2). This observation suggests that in these lin-44 males either the fates of B.a and B.p were reciprocally transformed or the polarity of the asymmetric division of B in the L1 stage was reversed. Based on our observations of the F cell lineages of lin-44 males, we suggest that the B cell defect also results from a reversal in the polarity of an asymmetric cell division. This reversal in polarity is evidenced by a change in cell size at the time of cell division, suggesting that lin-44 acts prior to or at the time of cell division (also see below). In a few lin-44 males the B cell divided symmetrically, so that B.a and B.p were of equal size and underwent similar patterns of division. In one of eleven lin-44 males the B cell divided according to the wild-type pattern.

U is another male-specific blast cell in the tail. In wild-type males U undergoes a left-right symmetric division during the mid-L2 stage. During the L3 stage U.l and U.r divide, and either one or both of the anterior daughters, U.la or U.ra, divide again (Sulston et al., 1980) (Fig. 2). This pattern gives the U lineage a variably asymmetric division. We observed that in three lin-44 males U.lp or U.rp divided, indicating that a polarity reversal might have occurred (Fig. 2). In five other lin44 males neither U.la nor U.ra divided, indicating a possible loss of asymmetry (Fig. 2). In four lin-44 males U divided according to the wild-type pattern.

Our interpretations of the F cell, B cell and other cell lineage defects seen in lin-44 animals are summarized in Fig. 3. The elliptical arrows indicate divisions with apparent polarity reversals. The horizontal arrows indicate possible reciprocal transformations. Fig. 3A shows a polarity reversal of a single asymmetric cell division. If ‘2’ and ‘3’ are each symmetric in their fates, such a polarity reversal cannot be distinguished from a reciprocal transformation. If ‘2’ and/or ‘3’ are dividing cells, the behavior of their daughters can distinguish whether ‘2’ and ‘3’ are reversed in their polarities or have maintained their polarities and are reciprocally transformed in their fates (Fig. 3B). The mutant cell lineages we observe in lin-44 males could in principle be explained either by heritable polarity reversals or by reciprocal transformations. Since far fewer polarity reversal events than reciprocal transformation events could account for the mutant lineages (e.g. one versus four for the first lin-44 F cell lineage shown in Fig. 2), we prefer the hypothesis that in lin-44 mutants cells undergo polarity reversals. Fig. 3C shows the effects of two sequential polarity reversals in a lineage. Examples of lineages with multiple sequential polarity reversals can be seen in the lin-44 T cell lineages shown below in Fig. 4.

In our interpretations of mutant cell division patterns we assume a polarity reversal to be a heritable event such that all divisions after the reversal are also reversed. Although each asymmetric cell division within an affected lineage could have been affected independently, this interpretation would require more primary events than does the interpretation that polarity reversals are heritable. Our usage of ‘heritable’ includes the possibility that once polarity is reversed, it still can be reversed again in subsequent divisions.

In general, we cannot directly observe whether a cell is ‘2’ or ‘reverse-2’ in its polarity or even whether it is ‘2’ or ‘3’ in its type. In such cases, we must define both cell type and the polarity of a cell division based only upon whether the daughter cells divide and if they do divide, according to what pattern. For example, the patterns of cell division in the lin-44 F lineages in Fig. 2 and model lineages in Fig. 3D are indicative of apparent polarity reversals. For simplicity we interpret such lineages by assuming that the minimal possible number of polarity reversals has occurred, e.g. one rather than two in the case of the example in Fig. 3D.

The T blast cells seem to be reversed in polarity in lin-44 hermaphrodites and males. In wild-type males and hermaphrodites, the bilaterally symmetric T cells divide during the early L1 stage to produce a hypodermal cells that joins the large hyp7 hypodermal syncytium (T.aa), a hypodermal blast cell that divides later in a sex-specific manner (T.ap), and a group of neural cells (derived from T.p) (Fig. 4). In wild-type males each T.ap blast cell divides during the L3 stage to produce the cells of three of the sensory rays of the male tail. Many of the cells produced by the T cell have distinct nuclear morphologies, so that for the T cell lineages we can assign cell fate based not only upon subsequent division pattern but also upon nuclear morphology (see Materials and Methods). We determined the complete T cell lineages of two lin-44 males (Fig. 4). We interpret these T cell lineages to have polarity reversals at many divisions. The T cell lineage of one of the two lin-44 males studied also had other defects: the cell T.paaaa appeared to have divided symmetrically (an extra ray cell group was formed by the cell T.paaaaa), and the cell T.papp underwent an extra division. We also followed four complete T cell lineages in lin-44 hermaphrodites. Both lin-44(n1792) hermaphrodites studied showed multiple T cell lineage polarity reversals (Fig. 4). In one lin-44(n2111) hermaphrodite three divisions within the T cell lineage showed polarity reversals, while in the other the T cell underwent two rounds of symmetric divisions generating a syncytial hypodermal cell and three cells that appeared to have nuclear morphologies with characteristics of both a seam cell and a blast cell (see Materials and Methods) but did not divide further (data not shown).

The polarities of five of the seven L1 divisions in the T cell lineage can be determined using Nomarski optics (see Materials and Methods). We scored these five divisions by direct lineage analysis of a total of 12 lin-44 males and five lin-44 hermaphrodites. We observed polarity reversals for all five divisions (data not shown). The polarity of the first division of the T cell was reversed in 16 animals (the 17th had a symmetric division). The polarities of the other divisions were affected at different frequencies, e.g. the polarity of T.a was reversed in only two animals. The T cell lineages of these 17 lin-44 animals can be explained by a total of 67 reciprocal transformations (or nonheritable polarity reversals) or by a total of 58 heritable polarity reversals. Although these numbers are not very different, based upon our observations of the F lineages in lin-44 males we interpret these lin-44 T cell lineages also to have heritable polarity reversals.

The abnormalities in the T cell lineage suggested that lin-44 animals might have defective phasmids. The phasmids are sensory structures in the tail consisting of neurons and supporting cells (Ward et al., 1975; Ware et al., 1975; Hall and Russell, 1991). The phasmids are open to the environment, so that when animals are soaked in the fluorescent dyes FITC or DiO the two neurons of each phasmid as well as six of the eight amphidial neurons in the head fill with these dyes and can be visualized in living animals (Perkins et al., 1986; Hall and Russell, 1991). The cells T.paa and T.pap are the phasmid socket cells PHso1 and PHso2 (Fig. 4), which function to provide channels for the phasmid neurons PHA and PHB to the outside (Sulston et al., 1980; White et al., 1986).

The phasmids of lin-44 animals failed to fill with fluorescent dyes (Fig. 5; Table 1). This defect appears to be caused by the abnormalities in the T cell lineage. During the early L1 stage, the T cells function as the phasmid socket cells (Sulston et al., 1980). We observed that 92% (n=100) of the phasmids in young L1 lin-44(n1792) animals fill with FITC, suggesting the phasmid neurons are not affected by lin-44 mutations and that the phasmid defect of lin-44 animals arises after the T cells normally divide to generate the phasmid socket cells of the older animal. In wild-type larvae and adults, the four phasmid socket cells label with antisera generated against the Lin-26 protein (M. Labouesse, personal communication). In lin-44 mutant late-L1 and L2 larvae, Lin-26 antisera do not label cells in the normal positions of the four phasmid socket cells but do label four cells near the anus, suggesting that the phasmid socket cells are anteriorly displaced (M. Labouesse, personal communication). We have directly observed the anterior displacement of the presumptive phasmid socket cells in lin-44 animals in our studies of T cell lineages (our unpublished observations). Thus, phasmid socket cells seem to be produced in lin-44 animals and are normal at least with respect to Lin-26 expression, but are displaced and do not function normally.

While characterizing the Lin-44 mutant phenotype, we observed one lin-44(n1792) male in which P12 appeared to be transformed to express the fate of P11. In wild-type animals the posterior daughter of P12, P12.p, divides to generate a small hypodermal cell, P12.pa, and a cell that undergoes programmed cell death, P12.pp. In this lin-44(n1792) male P12.p failed to divide during the L1 stage and instead assumed a hypodermal blast cell appearance like that of P11.p. During the L3 stage P12.p in this animal divided with a lineage pattern normally produced by P11.p, and P12.aap divided (P12.aap does not divide in wild-type males) like P11.aap normally does. Since in this lin-44(n1792) male both branches of the P12 lineages were like that of P11, we suspect that P12 was transformed to express the fate of P11. Subsequently, we identified another lin-44(n1792) male in which P12.p had failed to divide during the L1 stage and then observed the rest of the P12 lineage. In this male also P12 assumed the fate of P11. We examined the nuclear morphology of the cell in the position of P12.pa in 35 other lin-44(n1792) L3 males and found that in six of these animals a cell that looked like P11.p, presumably P12.p, was present instead. These observations indicate that in lin-44 animals the cell that normally expresses the P12 fate can instead express the P11 fate.

The fates of P11 and P12 are known to be specified by cell interactions. If P12 is killed using a laser microbeam, P11 will express the fate normally expressed by P12, indicating that P12 or its descendants normally act to prevent P11 from becoming P12-like (Sulston and White, 1980) In addition, like lin-44 mutations, mutations in the gene let-23 can cause P12 to express the P11 fate (Fixsen et al., 1985; Aroian and Sternberg, 1991). let-23 encodes a receptor tyrosine kinase that is probably the receptor for the inductive signal that functions during vulval induction (Aroian et al., 1990; reviewed by Horvitz and Sternberg, 1991). It seems likely that let-23 functions in controlling P11 and P12 fates as it does in vulval induction. These considerations suggest that the transformation of P12 into a P11-like cell in lin-44 mutant animals might result either from a primary defect in intercellular signaling or more indirectly, for example from a defect in the development of a cell that interacts with the presumptive P12 cell to specify the P12 fate.

The cells affected by lin-44 mutations are related by position (Fig. 1C) and not by lineage history. This observation raises the possibility that cell signaling may control the polarities of some or all of the affected cells. If so, the polarities of affected cells might be coordinately altered within given animals. The B, F, and U cell divisions were examined together in seven animals. In four animals the B, F and U lineages all showed defects, while in the other three animals the B, F, and U lineages were wild-type in polarity (data not shown). This observation suggests that the abnormalities in these three cell lineages are not caused independently.

The abnormalities of B and P12 were examined together in 26 animals: the fate of P12 was normal in 16/18 animals in which B was abnormal and in 5/8 animals in which B was wildtype in polarity. The abnormalities in B, P12 and T were examined together in 20 animals: T cell polarity was abnormal in all 20 animals, P12 was normal in 17/20 animals and B was wild-type in polarity in 6/20 animals. These results suggest that the abnormalities in the B, P12 and T lineages occur independently.

F and U polarity reversals have been observed in wild-type males in which the B cell was killed with a laser microbeam (Chisholm and Hodgkin, 1989). We have confirmed the finding of Chisholm and Hodgkin (1989) by killing the B cell in three wild-type L1 males (data not shown). In addition, we killed the mother of the B cell (AB.prppppap) in a single him5 male embryo. (We used the him-5 mutation to generate males in part because we wanted to compare effects of these kills to kills in lin-44; him-5 mutant males; see below.) Again, polarity reversals were seen in the F and U lineages (data not shown). Thus, in the absence of the B cell, polarity reversals occur in the F and U lineages, indicating that B or its descendants normally act to control polarity within these lineages. This observation is consistent with our finding described above that the B, F, and U lineages are affected coordinately in lin-44 animals. Since lin-44 mutants display F and U polarity reversals like those seen in the absence of a B cell, it seems likely that in lin-44 animals these reversals are a secondary consequence of the abnormalities in the B cell lineage.

By contrast, the B cell does not control polarity in the T cell lineage. We killed the mother of the B cell (AB.prppppap) in two him-5 male embryos and observed that polarity in the T lineages was normal, based on observations of the complete T cell lineages of one animal and on the morphologies of the T cell granddaughters in the other.

Could an abnormal B cell be responsible for the polarity reversals seen in the T cell lineages of lin-44 males? To test this possibility, we killed the mother of the B cell in four lin44; him-5 male embryos and one hermaphrodite embryo. The morphologies of the T cell granddaughters of both TL and TR cells were examined in each animal. Both T cells in each male were reversed, as was TR in the hermaphrodite; TL in the hermaphrodite generated four hypodermal nuclei that did not divide. Thus, T cell polarity in the absence of the B cell was indistinguishable from that seen in unoperated lin-44 animals.

These studies establish that the B cell does not control polarity in the T cell lineages of either wild-type or lin-44 animals.

We have observed several other defects in lin-44 animals: 59% (n=231) of lin-44(n1792) and 67% (n=245) of lin-44(n2111) hermaphrodites are egg-laying defective (Egl), and 10% (n=596) of lin-44(n1792) hermaphrodites and 15% (n=323) of lin-44(n2111) hermaphrodites have a protruding vulva. We determined the vulval cell lineages in 14 lin-44(n1792) hermaphrodites. All were normal, even in four animals that became Egl and developed a protruding vulva. The hermaphrodite sex muscles, which open the vulva and squeeze the uterus during egg laying (Horvitz et al., 1982; Trent et al., 1983; White et al., 1986), appear normal in lin-44 animals as visualized by polarized light microscopy (M. Stern, personal communication). The HSN neurons, which innervate the sex muscles and drive egg-laying (Desai et al., 1988), are normal in staining and morphology as visualized with fluorescent antibodies to serotonin (G. Garriga, personal communication). Thus, the basis of the lin-44 Egl defect is unclear.

We have determined the lineages of many other cells throughout the body in lin-44(n1792) animals and have not observed any additional defects. The cells examined include: H1 (n=2), H2 (n=2) and V1-V6 (n=2) (lateral hypodermal syncytial and seam cells from the head to the posterior body region); V5 (n=3) in the male (postdeirid sensillum, one male sensory ray and lateral hypodermal syncytial and seam cells); QL (n=2) and QR (n=2) (neuroblasts that generate cells that migrate anteriorly and posteriorly); G2 (n=2) (neurons and the excretory pore cell in the head); K (n=2) (DVB neuron and a rectal epithelial cell in the tail); P3.p-P8.p in hermaphrodites (n=14) (vulva equivalence group), P9.p-P11.p in the male (n=10) (preanal equivalence group); all other Pn.p cells (14 hermaphrodites and 10 males) (hypodermal cells in the ventral cord); and the male Y cell (n=2) (cloacal sensillum near the tail).

The existing alleles of lin-44 appear to be null alleles based on the phenotype they cause when heterozygous to deficiencies. Table 1 presents data for the hermaphrodite phasmid defect: both lin-44 alleles result in recessive phenotypes, and neither is enhanced when placed in trans to a deficiency for lin-44; moreover, both lin-44 homozygotes look similar to the lin44(n1792)/lin-44(n2111) trans-heterozygote. Similarly, males from a variety of him-5-containing strains lin-44(n1792), lin44(n1792)/hDf6, lin-44(n1792)/hDf7, lin-44(n2111), lin44(n2111)/hDf7 and lin-44(n1792)/lin-44(n2111) were examined using the dissecting microscope, and the range of male tail defects appeared similar (unpublished observations). Thus, the n1792 and n2111 alleles behave in these experiments like deficiencies that eliminate lin-44 function. Furthermore, n1792 is an amber mutation and presumably results in a truncated lin-44 protein (see Materials and Methods) (Table 1). These results indicate that n1792 and n2111 reduce lin-44 function and could be null alleles.

hDf7 fails to complement lin-44 and is presumably null for lin-44 function. Molecular analysis has shown that a 7.5 kb genomic DNA fragment that rescues the lin-44 phenotype in microinjection experiments lies approximately 40 kb to the right of the left hDf7 endpoint and approximately 160 kb to the left of the right hDf7 endpoint (M.A.H., unpublished results), suggesting the lin-44 locus is completely removed by hDf7. Because 43% of hDf7 homozygotes hatch and initiate development (see Materials and Methods), we could determine whether polarity reversals occur in these animals. Analysis of these deficiency homozygotes indicated that complete loss of lin-44 function most often resulted in T cell polarity reversal. We scored the polarities of the T cell divisions by examining the nuclear morphologies of the descendants of the T cell present at late the L1 stage in wild-type, lin-44, and homozygous hDf7 animals (see Materials and Methods) (Table 2). The polarity of the T cell division was normal in wild-type animals (Fig. 6A). In most lin-44 animals and in most homozygous hDf7 animals the polarity of the T cell division was reversed (Fig. 6). Thus, a complete absence of lin-44 gene activity can reverse the polarity of the asymmetric T cell division. Despite the complete loss of lin-44 gene activity the polarity reversal phenotype was not completely penetrant: hDf7 homozygotes with T cell divisions of normal polarity or a loss of asymmetry (symmetric divisions) were observed. Furthermore, the frequencies of T cell divisions with reversed and normal polarities were similar in the lin-44 mutants and the hDf7 homozygotes (Table 2). The observation of divisions with normal polarity and loss of asymmetry in the hDf7 homozygotes demonstrates that these defects are not caused by residual lin44 gene activity.

To confirm that these reversals were defects in cell lineage, we directly followed the T cell lineages of eight hDf7 homozygotes. Based on the morphologies of the granddaughters of the T cells, the polarity of the T cell division was reversed in five hDf7 homozygotes. One of these T cell lineages is shown in Fig. 6D. In one other hDf7 animal the T cell division was of normal polarity, and in two hDf7 animals the polarity of the T cell division could not be assessed (data not shown). Thus, direct lineage analysis of these eight hDf7 homozygotes confirmed that a complete absence of lin-44 gene activity can reverse the polarity of the asymmetric T cell division.

Three of the cell divisions that can be reversed in polarity in lin-44 animals generate daughter cells with nuclei of different sizes, which allowed us to determine if the effects of lin-44 on the polarities of the divisions of B, T.pp and T.ppp could be observed at the time of cell division. In wild-type males, the nucleus of B.a is larger than that of B.p (Fig. 7A, Table 3). In lin-44 males, as noted above, B.p was larger than B.a (Fig. 7B, Table 3). In wild-type animals, T.pp divides to generate a smaller anterior nucleus, T.ppa, and a larger posterior nucleus, T.ppp. In 8/15 lin-44 animals, the polarity of the cell expressing the T.pp fate (identification based upon its subsequent divisions pattern) was reversed, and in all eight of these animals the reversal could be seen by the reversal in the nuclear sizes of the daughters immediately after division (data not shown). Similarly, in 11/15 lin-44 animals, the polarity of the cell expressing the T.ppp fate was reversed, and in all 11 of these animals the reversal could be seen by the reversal in the nuclear sizes of the daughters immediately after division (data not shown). Since the polarity reversals caused by lin-44 mutations can be seen at the time of cell divisions, lin-44 must act prior to or at the time of cell division.

lin-17 mutations cause the B and T cells to lose polarity and generate daughters of the same type (Sternberg and Horvitz, 1988). In lin-17 males, the nucleus of B.a is the same size as the nucleus of B.p (Fig. 7C, Table 3), which led Sternberg and Horvitz (1988) to propose that lin-17 might act at or before the time of the B cell division. We constructed a lin-17 lin-44 double mutant and observed that in this mutant the nuclei of B.a and B.p were similar in size (Fig. 7D, Table 3). Furthermore, we observed in 8/8 lin-17 lin-44 double mutant animals the same effect on the T cell lineage as Sternberg and Horvitz (1988) saw in the lin-17 single mutant: the T cell usually underwent two rounds of symmetric divisions generating four hypodermal cells. These observations suggest that lin-17 is needed to generate asymmetry, i.e. to make two daughter cells different, whereas lin-44 is needed to define the spatial orientation of that asymmetry.

Mutations in the gene lin-44 cause apparent polarity reversals in the B, F, U and T cell lineages. In each case of polarity reversal, a cell divides to produce sister cells that appear to be reversed with respect to both their normal positions and their own polarities. The polarity reversals of different cells occur along different axes. The affected cell divisions in the T and U cell lineages are anterior-posterior, that of the B cell is skewed and anterior/dorsal-posterior/ventral, and those of the F cell lineage are dorsal-ventral. Thus, lin-44 seems to act in a general process involved in orienting the polarities of asymmetric cell divisions rather than in one involved in specifying polarities along a single axis.

Two considerations indicate that the reversals in cell polarity caused by lin-44 mutations probably occur at the stage of the mother cell rather than after the two sisters have been generated (Fig. 8). First, a reversal in the polarity of the mother cell would explain the coordinate reversals of both the positions and the polarities of the sister cells. Second, certain cell divisions that can be reversed in polarity generate sister cells of different sizes, and relative cell size which is determined prior to the time of cell division is also reversed when a polarity reversal occurs during such a cell division. Since polarity reversals appear to be caused by a complete loss of lin-44 function, the lin-44 gene normally is needed to orient the polarities of the B, F, U and T cells and their descendants.

In the absence of lin-44 function, certain cells have polarities opposite to those expressed in the wild-type animal. Why should cells be generated with polarities that must be reversed by lin-44? One possibility is that the cell polarities seen in lin44 mutants reflect polarities that were expressed at an earlier stage of nematode evolution. In other words, perhaps the evolution of nematode cell lineage has involved the reversal of certain cell polarities. The suggestion that polarity reversals might function in the evolution of nematode cell lineage was made previously, based upon comparisons of homologous cell lineages both within C. elegans and between C. elegans and another free-living nematode, Panagrellus redivivus (Sternberg and Horvitz, 1981, 1982). In particular, it is interesting that a descendant of B, the cell B.ppa, appears to display opposite polarities in these two species.

How does lin-44 affect the B, F, U, T, and P12 cell lineages? These cells are all located in the tail and far posterior body region (Fig. 1C). One possibility is that the defects in these cell lineages are secondary consequences of defects in another cell or cells. For example, the polarity reversals that occurred in the F and U lineages of lin-44 animals were identical to those previously seen by Chisholm and Hodgkin (1989) in wild-type animals in which the B cell was killed with a laser microbeam. Since defective F and U lineages were seen only in lin-44 animals with defective B lineages, it seems likely that the F and U polarity reversals were consequences of the abnormalities in the B cell lineage. By analogy, the polarity reversals in the B and T lineages of lin-44 animals might also have been secondary consequences of abnormalities in cells that specify the polarities of these B and T cells. The effects of lin-44 on P12 could be similarly explained: since, as discussed above, the fate of P12 seems to be specified by cell interactions, lin44 could be altered in P12 cell fate as a secondary consequence of its effects on a cell that specifies this fate.

Alternatively, lin-44 might act more directly to specify cell polarities. Polarity may be determined by a polarizing signal from a neighboring cell. lin-44 mutations might disrupt this signal, for example if lin-44 encodes a protein that acts in a signal transduction pathway, such as a ligand, a receptor, or a Ras protein (Beitel et al., 1990; Han and Sternberg, 1990; Aroian and Sternberg, 1991; Hill and Sternberg, 1992). Another possibility is that lin-44 encodes a protein that responds to such a signal transduction pathway such as a transcription factor needed to specify both the fate of P12 and the polarities of cells in the B and T lineages.

A third possibility is that lin-44 encodes a protein that acts within polar cells to specify their orientations. An analogy is provided by the two distinct sets of genes involved in polarity in the budding yeast Saccharomyces cerevisiae: one set is need for the generation of polarity, while the second set is needed for its orientation (Hartwell et al., 1973; Hartwell et al., 1974; Adams et al., 1990; Chant et al., 1991; Chant and Herskowitz, 1991). The C. elegans gene lin-17 is needed for the generation of the polarity of the B and T cells (Sternberg and Horvitz, 1988), while lin-44 is needed for the normal orientation of these polar cells along the body axis of the animal. It is possible that these nematode genes act analogously to the yeast genes noted above to establish and orient cellular polarity in the developing animal.

We thank Andrew Chisholm, Erik Jorgensen and Gary Ruvkun for critical reading of the manuscript and Bob Herman for critical reading and extensive suggestions. We also thank members of the Horvitz laboratory for advice and stimulating discussions. This work was supported by NIH grant GM24663 to H. R. H. M. A. H. was supported by NIH training grant GM07287 and by NIH grant GM22387 to R. K. Herman. H. R. H. is an Investigator of the Howard Hughes Medical Institute.