Chapter 13:  III Establishment of Polarity in the One-Cell Embryo

A. Cytoplasmic Streaming

The period of cytoplasmic reorganization, which comprises only about 15 minutes of the first 100-minute cell cycle (at 16°C), converts the initially symmetrical egg into a highly polarized embryo. The cytoplasmic streaming observed during this period appears to have a crucial role in generating asymmetry. The streaming was first reported by Nigon et al. (1960) and later carefully documented by Hird and White (1993) using time-lapse video microscopy to follow individual cytoplasmic granules observed with Nomarski microscopy. Internal granules flow posteriorly toward the sperm pronucleus, whereas cortical granules flow anteriorly away from the sperm pronucleus. This “fountainhead” pattern of cytoplasmic streaming may be directed by the sperm pronucleus, the associated centrosomes and nascent microtubules, or a localized cortical change induced by sperm entry, since embryos in which the sperm entered laterally display cytoplasmic streaming that is laterally directed toward the sperm pronucleus (Goldstein and Hird 1996). The direction of streaming becomes shifted toward the nearest (future posterior) pole as the sperm pronucleus, associated centrosomes, and perhaps overlying membrane shift toward that pole.

Cytoplasmic streaming appears to have a major role in localizing germ-line-specific P granules to the posterior cortex of the one-cell embryo. This insight has come from monitoring the behavior of fluorescently tagged P granules in living embryos (Hird et al. 1996). The granules are distributed throughout the cytoplasm of newly fertilized embryos. The majority of granules flow toward the posterior pole at the same time and with the same speed as general cytoplasmic streaming occurs. In addition, as observed with general cytoplasmic streaming, P granules at the posterior cortex show some anterior movement away from the sperm pronucleus. Differential stability of P granules in different regions of the cytoplasm also appears to contribute to P-granule partitioning in the one-cell embryo; P granules in the anterior-most region of the embryo do not move posteriorly and instead disappear from view (Hird et al. 1996). This is thought to reflect disassembly or degradation of P granules in cytoplasm that is destined for the somatic daughter cell. This would suggest that, in addition to asymmetric localization of P granules, there is asymmetric localization of the ability to maintain P granules. The consequence of partitioning P granules to the posterior cortex is that the granules are inherited exclusively by P₁. During subsequent divisions, they are passed to P₂, P₃, and then P₄, the primordial germ cell.

B. The Crucial Role of Microfilaments

The actin cytoskeleton is required to generate asymmetry in the one-cell embryo. Treatment of embryos with the microfilament inhibitor cytochalasin D disrupts the microfilament cytoskeleton (Strome 1986b) and prevents cytoplasmic streaming (Hird and White 1993), P-granule segregation to the posterior, and the development of spindle asymmetry (Strome and Wood 1983). In cytochalasin-treated embryos, the pronuclei meet centrally and P granules coalesce in the center of the embryo, both asters of the spindle behave similarly, and the spindle remains symmetrically located (Fig. 5b). Cytokinesis does not occur (Strome and Wood 1983). On the basis of analysis of embryos treated with cytochalasin D for only brief intervals in the first cell cycle (Fig. 5c–e), the crucial time interval for microfilament function is the interval of cytoplasmic reorganization described above. Drug treatment of embryos during this interval, but not before or after, prevents P-granule partitioning and spindle asymmetry, leading to symmetrical division and distribution of P granules to both daughters (Hill and Strome 1988). Analysis of the resulting two-cell embryos further reveals that microfilament disruption during the one-cell stage also leads to missegregation of the potential for future unequal divisions and unequal partitioning of P granules (Hill and Strome 1990; see below). Thus, microfilaments appear to be involved in many or all aspects of asymmetry in the one-cell embryo.

Where are microfilaments located when they perform their crucial role(s) in the one-cell embryo? Staining of embryos with rhodamine phalloidin revealed that microfilaments exist as a meshwork of fine fibers just below the cell surface and as cortical dots or foci (Strome 1986b). The fine fibers remain around the entire periphery throughout the first cell cycle. However, the foci become concentrated in the anterior cortex during the period of cytoplasmic reorganization. Although this striking asymmetry in microfilament foci correlates roughly with the time of cytochalasin sensitivity, it is not essential for establishing polarity. Embryos from nop-1 (for no pseudocleavage) mutant mothers lack the asymmetrically distributed microfilament foci, yet are viable and undergo cytoplasmic streaming and P-granule localization (Rose et al. 1995). Thus, the anteriorly concentrated microfilament foci observed in wild-type embryos apparently are not responsible for generating asymmetry, but instead are another manifestation of asymmetry in the one-cell embryo. The component of the actin cytoskeleton that appears to be critical for the establishment of asymmetry is the uniform cortical meshwork. This meshwork may serve as a scaffold upon which other factors become asymmetrically localized.

C. The Role of the par Genes

Maternal-effect lethal mutations have identified several genes with roles in establishing early embryonic polarity: mex-1 (Mello et al. 1992) and the six par genes (Kemphues et al. 1988b; Kemphues 1989; Kirby et al. 1990; Morton et al. 1992; Cheng et al. 1995; J. Watts et al., in prep.). mex-1 has a role in localization of SKN-1 and P granules (see Schnabel and Priess, this volume). Mutations in the six par genes (for partitioning-defective) lead to disruption of several aspects of A-P polarity in the zygote, including P-granule localization, pseudocleavage and cytoplasmic streaming, and asymmetric placement of the first cleavage spindle (Kemphues et al. 1988b; Kirby et al. 1990). The daughter cells that result do not exhibit typical polar behaviors. They divide synchronously and in many cases have altered spindle orientations. In addition, molecules that are normally restricted to either anterior or posterior blastomeres have abnormal distributions in par mutant embryos. SKN-1 or GLP-1 or both fail to localize in par-1, par-2, and par-3 mutant embryos (Bowerman et al. 1993; B. Bowerman; S. Crittenden and J. Kimble, both pers. comm.), and PAL-1 and PIE-1 are undetectable in par-1 embryos (C. Hunter and C. Kenyon; C. Schubert et al., both pers. comm.). Subsequent development is aberrant, with alterations in timing and spindle orientation in later cleavages and alterations in cell fates. Embryos arrest as amorphous masses of differentiated cells.

Although mutations in all of the par genes affect the same processes, the mutant phenotypes are gene-specific (see Table 1) (Kemphues et al. 1988b; Morton et al. 1992; Cheng et al. 1995). For example, par-1 and par-4 strongly affect P-granule localization but only weakly affect spindle positioning and orientation, whereas the remaining four par genes have a strong effect on spindle behavior and a weaker effect on P-granule localization. The differences are not simply quantitative: par-2 and par-5 affect spindle orientation in the P₁ cell, and par-3 and par-6 affect spindle orientation in the AB cell. Similarly, par-1, par-3, and par-6 mislocalize SKN-1 and par-2 does not.

The par genes appear to be exerting their primary effects during the first cell cycle. First, as described above, the earliest deviation from normal occurs in the zygote (Kirby et al. 1990). Second, temperature-sensitive mutations in par-2 and par-4 are insensitive to nonpermissive temperature after the one-cell stage (Morton et al. 1992; Cheng et al. 1995). Consistent with this view, the PAR proteins are present in the one-cell embryo (Fig. 6 and see below). However, the PAR proteins may also play a part in germ-line development. At least one allele at each locus is incompletely expressed; i.e., some embryos escape the maternal-effect lethality and grow to be adult worms. Most such worms are agametic (Kemphues et al. 1988b; Morton et al. 1992; Cheng et al. 1995; K.J. Kemphues, unpubl.). This could be a secondary consequence of a weak defect at the one-cell stage or could reflect a requirement for the par genes in later germ-line development. The protein distributions described below are consistent with the latter possibility.

D. Mode of Action of the par Genes

Although it is not yet known how the par genes work, information obtained by cloning of three of the par genes gives some important clues. PAR-1 is a 126-kD protein that includes near its amino terminus a serine/threonine protein kinase domain with strong similarity to a small subclass of widely conserved kinases. PAR-1 shares with these kinases an additional domain of unknown activity at the carboxyl terminus (Guo and Kemphues 1995). The kinase activity appears to be important for par-1 function, since two mutations affecting conserved amino acids of the kinase domain produce phenotypes indistinguishable from the putative null allele. PAR-2 is a novel protein with a predicted size of 72 kD and contains a zinc-binding domain of the “ring finger” class and a myosin-type ATP-binding site (Levitan et al. 1994). PAR-3 is a novel protein of 138 kD (Etemad-Moghadam et al. 1995).

All three proteins become localized to the cell periphery of the zygote in an asymmetric fashion (Fig. 6). PAR-1 and PAR-2 are restricted to the posterior 40–50% of the zygote (Guo and Kemphues 1995; L. Boyd and K. Kemphues, unpubl.), whereas PAR-3 is restricted to the anterior 50–60% of the zygote (Etemad-Moghadam et al. 1995). None of the protein sequences have features consistent with membrane localization, so it seems likely that the proteins are associated with the cortical cytoskeleton. It is possible that the effects of cytochalasin on polarity (Hill and Strome 1988, 1990) could be due in part to the mislocalization of the PAR proteins.

Analysis of PAR protein distributions in par mutant embryos has revealed relationships among the par genes with respect to control of their localization (for summary, see Fig. 7a) (Etemad-Moghadam et al. 1995; Watts et al. 1996; L. Boyd and K.J. Kemphues, unpubl.). The major conclusions are (1) PAR-2 and PAR-3 are mutually dependent for their localization, (2) PAR-3 restricts PAR-1 to the posterior, (3) par-6 acts to stabilize PAR-3 at the cortex, and (4) par-4 is not involved in localizing PAR-1, PAR-2, or PAR-3. The role of par-5 is unclear. The results suggest the following model (Fig. 7b): A graded distribution of PAR-3 along the future A-P axis is generated in response to the polarity cue provided by the sperm. This graded distribution is reinforced by the activity of PAR-2, which excludes PAR-3 from the posterior cortex. PAR-1 is excluded from the anterior cortex by PAR-3.

This model for localization can explain an unexpected genetic interaction. Reducing wild-type par-6 activity suppresses par-2 loss-of-function mutations (J. Watts et al., in prep.). Two observations of PAR-3 distribution suggest a likely explanation. First, in par-6 mutant embryos, PAR-3 is not maintained at the cortex. Second, in par-2 mutant embryos, PAR-3 is not restricted to the anterior but rather is present at the cell periphery in a gradient along the A-P axis. Suppression of the par-2 mutations by reducing par-6 could be the result of a uniform decrease in the amount of PAR-3 at the cortex. Because of the graded distribution of PAR-3 in par-2 mutants, an overall reduction of PAR-3 in the cortex would reduce the amount of PAR-3 at the posterior periphery to negligible levels, but would leave significant amounts of PAR-3 at the anterior, resulting in a more nearly normal distribution of PAR-3. This would, in turn, lead to a nearly normal distribution of PAR-1. Preliminary results from examining the distribution of PAR-3 in the suppressed embryos support this view (J. Watts and B. Etemad-Moghadam, pers. comm.). If this interpretation is correct, then the primary function of PAR-2 is to restrict PAR-3 protein to the anterior periphery.

The modes of action of PAR-3 and PAR-1 are less clear. PAR-3 has at least two functions. One, as described below, is control of spindle orientation, and the other is the localization of PAR-1. It is possible that localizing PAR-1 is the major way that PAR-3 influences intracellular polarity. Alternatively, PAR-3 could be mediating the localization of multiple cellular components including PAR-1. PAR-1 presumably acts via phosphorylation, but its substrates and site of action are unknown. Although its localization to the posterior periphery suggests that it has a role there, a detectable amount of PAR-1 protein is found in the cytoplasm (Guo and Kemphues 1995). Furthermore, the PAR-1 protein exhibits partial function when it is not localized, as occurs in par-2 mutant embryos (see Fig. 7a). In par-2 mutants, P granules become localized to the posterior of the zygote (Cheng 1991) despite the fact that PAR-1 is not localized to the cortex. Because par-1 activity is required for P-granule localization (Kemphues et al. 1988b; Guo and Kemphues 1995), these observations indicate that asymmetric peripheral localization of PAR-1 is not necessary for it to mediate the posterior localization of P granules. Why is PAR-1 localized? Perhaps events in the zygote other than localization of P granules require high concentrations of PAR-1 at the posterior periphery. Alternatively, asymmetric cortical localization may be required to assure proper amounts of PAR-1 in P₁, P₂, and P₃ (see next section).