Special Reviews

Anterior-Posterior Polarity in C. elegans and Drosophila--PARallels and Differences

See allHide authors and affiliations

Science  06 Dec 2002:
Vol. 298, Issue 5600, pp. 1946-1950
DOI: 10.1126/science.1072162


The eggs of Caenorhabditis elegans andDrosophila bear little similarity to each other, yet both depend on the par genes for control of anterior-posterior polarity. Here we explore possible common roles for the pargenes (pars) in converting transient asymmetries into stably polarized axes. Although clear mechanistic parallels remain to be established, par-dependent regulation of microtubule dynamics and protein stability emerge as common themes.

The key to every biological problem must finally be sought in the cell, for every living organism is, or at some time has been, a cell. E. B. Wilson, 1925 [(1), p. 1]

A major challenge in developmental biology is to understand how asymmetries are elaborated along the main body axes. How are heads made different from tails and everything in between? Remarkably, in many organisms these morphological differences can be traced back to the one-cell stage, where axis determinants localize to opposite ends of the egg. For many biologists, this realization has meant that to understand axis formation, one must first understand how asymmetries arise within a single cell: the egg.

Genetic screens in Drosophila and C. eleganshave identified several regulators of egg polarity. These two models were long thought to bear little resemblance to one another. InDrosophila, polarization of the egg begins during oogenesis and requires microtubules. In contrast, in C. eleganspolarization begins after fertilization and requires the actin cytoskeleton. The discovery of a group of genes essential for polarization of the C. elegans embryo (“par” genes) proved to be a turning point in the field (2).par-3 and par-6 encode two PDZ domain proteins, which together with the atypical protein kinase C PKC-3, form a complex in the anterior half of the C. elegans zygote (3–5). The serine threonine kinase PAR-1 and the ring finger protein PAR-2 occupy the posterior half (6,7). Two other genes, par-4 andpar-5, encode proteins that are uniformly distributed (8, 9). Mutations in any one of these genes disrupt polarization of the zygote. Homologs of the pargenes were soon discovered in mammals, where they regulate the polarization of epithelial cells, and in Drosophila, where they regulate epithelial and neuronal polarity [reviewed in (10, 11)]. These observations prompted several groups to investigate whether the par genes might also regulate polarity in the Drosophila egg. Thus far, results indicate that this is the case. In fact, except for par-2, homologs of all the par genes have now been identified inDrosophila and are required for egg polarity (12–20). This remarkable conservation raises an apparent paradox: how can the same group of genes regulate polarity in such dissimilar cells (Fig. 1)? Here, we explore this issue by focusing on the role of the par genes in regulating anterior-posterior (A/P) polarity. We refer the reader to (21) and (22) for comprehensive reviews of axis formation in Drosophila and C. elegans.

Figure 1

Asymmetric localization of germline proteins inDrosophila and C. elegans eggs. (A) Stage 10 Drosophila egg chamber with oskar mRNA in blue and oocyte outlined. [Reprinted from (14), with permission from Elsevier Science] (B) Nomarski (top) and ultraviolet (UV) (bottom) exposures of a C. eleganszygote expressing PIE-1:GFP. Eggs are approximately to scale (Drosophila oocyte ≈ 100 μm; C. elegansembryo ≈ 50 μm), with anterior to the left and posterior to the right.

Establishment of A/P Polarity in C. elegans

In C. elegans, polarization of the egg begins after fertilization and is initiated by the sperm asters, which marks the future posterior end of the embryo (23–26). The first sign of polarity is seen when contractions of the egg cortex suddenly cease in a small area near the sperm pronucleus and internal cytoplasm begins to flow toward that area (27). The cue that initiates these rearrangements is not known but appears linked to the nucleation of microtubules by the sperm-derived centrosomes (24–26). The actin cytoskeleton is also involved: cytochalasin treatment and depletion of the nonmuscle myosin NMY-2 block polarization (28–30). Close contact between the sperm asters and the cortex has been correlated with the onset of polarity (31), suggesting that intimate interactions between the sperm asters and the actin-rich cortex are what initiate the polarization process.

In response to the sperm aster signal, the PAR proteins, which start out uniformly distributed, rearrange dramatically. PAR-1 and PAR-2 localize to the posterior half of the cortex (6,7), whereas the PAR-3/PAR-6/PKC-3 complex localizes to a complementary region in the anterior (3–6). Localization of the PARs is interdependent: PAR-3/PAR-6/PKC-3 are required to localize PAR-2 to the posterior (7), and PAR-2 is required to localize PAR-3/PAR-6/PKC-3 to the anterior (3–5). All are required to localize PAR-1, but PAR-1 is not required for the initial localization of other PARs (3, 7). Thus, antagonistic interactions between PAR-3/PAR-6/PKC-3 in the anterior and PAR-2 in the posterior create distinct cortical domains, which in turn regulate the placement of PAR-1. Time-lapse observations of green fluorescent protein (GFP)–tagged PAR-2 and PAR-6 (GFP:PAR-2 and GFP:PAR-6) were used recently to clarify the role of the sperm asters in this process (30). GFP:PAR-2 localizes to the posterior of wild-type embryos coincident with sperm aster formation, but fails to do so when expression of the anterior PARs is reduced by RNA-mediated interference (RNAi). In contrast, GFP:PAR-6 localizes to the anterior during sperm aster formation even in the absence of PAR-2. These observations indicate that the sperm aster signal functions primarily by excluding PAR-6 and the other anterior PARs from the posterior. Although GFP:PAR-6 initially localizes normally in the absence of PAR-2, it eventually reenters the posterior domain as the zygote prepares to divide. During that time, the microtubules form the mitotic spindle and are no longer restricted to the posterior. Thus, polarization of the cortex proceeds in two phases: an initial “establishment” phase regulated by the sperm asters, and a later “maintenance” phase regulated by PAR-2 (Fig. 2).

Figure 2

The PAR hierarchy in C. elegans. (A) Anterior PARs, red; posterior PARs, orange. In the cytoplasm, PIE-1 and P granules, blue; MEX-5 and MEX-6, yellow. During meiosis (top), all are uniformly distributed throughout the zygote. Circles, pronuclei; black lines, microtubules. (B) Sequential repression model [modified from (42)]. Lines with bars, antagonistic interactions; lines with arrows, positive interactions.

The PARs are required for all subsequent asymmetries, including the asymmetric segregation of P granules and the asymmetric placement of the first mitotic spindle. Thus, the establishment of distinct anterior and posterior PAR domains converts a transient cue (eccentrically placed sperm asters) into stable positional coordinates that are used to localize cytoplasmic factors. The mechanisms that restrict PAR proteins to specific cortical domains, however, remain unknown. Localization of PAR-2 and PAR-6 requires only coding sequences (24, 30), suggesting that the mechanisms involved act on the proteins and not the RNAs. Ring finger domains such as the one in PAR-2 (7) have been found in E3 ubiquitin ligase subunits, raising the possibility that PAR-2 excludes anterior PARs by triggering their degradation in the posterior. Proteins involved in membrane trafficking also have been implicated in the establishment of PAR domains. POD-1 is a coronin-related protein required for secretion of the egg shell and for PAR asymmetry in the zygote (32). OOC-3 and the torsin-related protein OOC-5 are two endoplasmic reticulum (ER) proteins required for oocyte growth and PAR asymmetry in the two-cell stage (33, 34). The PARs are not secreted proteins, and thus are unlikely direct targets for these factors. Targeted secretion of membrane proteins, however, could be used to localize anchorage sites for the PARs. An important task for the future will be to delineate domains in PAR proteins critical for localization and to identify factors that interact with these domains.

Establishment of A/P Polarity in Drosophila

In contrast to C. elegans, polarization of theDrosophila egg begins during oogenesis and is completed before fertilization. Asymmetries along the A/P axis are established during two distinct stages of oocyte development. The first polarization phase occurs early in oogenesis, during the process of oocyte specification. Later in oogenesis, the egg is repolarized, and it is during this second polarization that A/P determinants assume their final positions along the A/P axis [reviewed in (21)]. Here, we focus on the first polarization event and its dependence on the par genes.

Drosophila oogenesis begins with the division of a germline stem cell to form a 16-cell cyst, where sisters are interconnected by cytoplasmic bridges called ring canals. One of the 16 cells becomes the oocyte and the others become polyploid nurse cells. Which cell adopts the oocyte fate may depend on the asymmetric distribution of a membranous organelle called the fusome. The fusome is partitioned asymmetrically during cystoblast divisions and it has been hypothesized that the cell that inherits the most fusome material is the one that will become the oocyte (35,36). This hypothesis, however, has been difficult to test directly because the fusome disappears before the oocyte can be identified unambiguously. Differentiation of the oocyte is a gradual process that depends on the accumulation of several determinants, such as Orb (37), which travel to the oocyte on a polarized microtubule network that forms in a fusome-dependent manner throughout the cyst. The determinants initially accumulate at the anterior side of the oocyte and later shift to the posterior, coincident with relocalization of the oocyte microtubule organizing center (MTOC) to the posterior. This redistribution marks the first polarization phase of the oocyte and appears to be coupled to stable establishment of oocyte fate [reviewed in (38)].

Drosophila PAR-1 localizes to the fusome (15,17) and becomes restricted to the future oocyte in a microtubule dependent–fashion (18). In par-1null mutants, oocyte selection is delayed and transient (15) or does not occur at all (17), suggesting that PAR-1 may be one of the fusome-associated factors that participates in oocyte selection. PAR-1 is also required for polarization of the new oocyte: in the absence of PAR-1, the MTOC and Orb never relocalize to the posterior, Orb eventually disappears, and the oocyte reverts back to the nurse cell fate (15). Thus, as in C. elegans, PAR-1 appears to transform a transient asymmetry (fusome) into a stably polarized axis that is used to localize determinants, in this case required for oocyte fate.

Remarkably, the Drosophila homologs ofpar-3 (bazooka, baz), par-4(lkb1), par-5, par-6, andpkc-3 (DaPKC) are also required for anterior-to-posterior translocation of oocyte markers and for maintenance of oocyte fate (13, 16,19, 20). At first glance, the regulatory hierarchy that links these genes in C. elegans does not appear fully conserved in Drosophila. In C. elegans, localization of PAR-1 is dependent on all the other PARs. In contrast, in Drosophila localization of PAR-1 to the fusome is independent of Baz, PAR-6, and PAR-5 (13,16, 19). In the oocyte, PAR-1 initially accumulates in the anterior and relocates to the posterior during polarization (18). In baz mutants, PAR-1 is present early in the anterior but is lost after polarization and never appears in the posterior, suggesting that BAZ may regulate PAR-1 localization at this stage (18). The reverse, however, is also true: BAZ disappears from oocytes in par-1 mutants (18). It is possible that these effects do not reflect true regulatory interactions. but rather are secondary to the failure to maintain an oocyte in these mutants. The inter-dependence of oocyte fate, oocyte polarization, and PAR asymmetry will certainly complicate the unraveling of any potential regulatory hierarchy. In established oocytes, BAZ and PAR-1 occupy complementary cortical domains (18) as they do in C. elegans, but the importance of this localization, which is observed after oocyte polarization, is not yet known. As in C. elegans, par genes inDrosophila function together to create a polarized axis, but their relative contributions remain to be determined.

Patterning of the A/P Axis in C. elegans

The next step after the creation of PAR domains on the cortex is to localize determinants in the cytoplasm. In C. elegans, the PAR most directly involved in this process appears to be PAR-1. PAR-1 is required for all cytoplasmic asymmetries but is not required for the initial localization of other PARs. Here we focus on PAR-1's role in localizing germ plasm components to the posterior end of the embryo. The germ plasm is a complex mixture of proteins (e.g., PIE-1) and RNA-rich organelles (P granules) essential for germline development. These factors start out uniformly distributed in oocytes and segregate to the posterior after fertilization coincident with the reorganization of the PAR proteins into two domains. In the absence of PAR-1, P granules and PIE-1 remain uniformly distributed and eventually disappear from all cells (2, 39, 40). PAR-1 does not act on P granules and PIE-1 directly, but instead functions through a pair of redundant intermediates. MEX-5 and MEX-6 are nearly identical cytoplasmic proteins, which, under the influence of PAR-1, localize to the anterior in a pattern opposite that of PIE-1 and P granules (41). In the absence of MEX-5 and MEX-6, P granules and PIE-1 remain uniformly distributed (41), suggesting a “sequential repression model” whereby PAR-1 excludes MEX-5 and MEX-6 from the posterior and MEX-5 and MEX-6, in turn, exclude P granules and PIE-1 from the anterior (42) (Fig. 2).

In principle, PAR-1 and MEX-5 and MEX-6 could affect protein localization by a number of mechanisms, including: (i) RNA localization, (ii) localized translation, (iii) protein transport, (iv) protein trapping, and (v) localized protein degradation. MEX-5 and MEX-6 belong to the CCCH finger family of RNA binding proteins, but their effect on PIE-1 localization is unlikely to be dependent on interactions with the pie-1 RNA, because PIE-1 asymmetry is regulated at the protein level (40). PIE-1 localization involves two complementary mechanisms: a first mechanism that causes PIE-1 to become enriched in the posterior at the time of MEX-5 and MEX-6 localization in the anterior, and a second mechanism that degrades any PIE-1 left over in the anterior after division (40). P granule asymmetry also involves two mechanisms: movement toward the posterior and degradation in the anterior (43). In par-1 mutants, P granules and PIE-1 are degraded in all cells (2, 39,40), suggesting that PAR-1 creates an environment in the posterior that protects these factors from degradation. In mex-5;mex-6 double mutants andmex-5;mex-6;par-1 triple mutants, PIE-1 is maintained in all cells (41), consistent with the idea that MEX-5 and MEX-6 act downstream of PAR-1 to activate PIE-1 degradation in the anterior.

These data indicate that local inhibition of protein degradation is one of the PAR-1–dependent mechanisms used to maintain determinants in the germline. Whether protein transport and/or trapping also contribute to the initial segregation of PIE-1 and MEX-5 to opposite poles in the zygote remains an open question. One point, however, is clear. Microtubule-based transport is unlikely to be involved, because asymmetries arise before microtubules span the axis and because microtubule-depolymerizing drugs do not affect P granule or PIE-1 asymmetry (28, 40).

Patterning of the A/P Axis in Drosophila

In contrast to C. elegans, segregation of determinants in the Drosophila oocyte depends on microtubules. Initial polarization of the oocyte during the oocyte-specification phase requires microtubules and correlates with the formation of a MTOC in the posterior of the cell. The posterior MTOC does not form in par-1 mutants (15,17), but the role of PAR-1 in this process remains unclear. PAR-1 functions again later in oogenesis during repolarization of the oocyte (12, 14). The asymmetry-generating mechanisms used in this second polarization are understood in more detail.

After oocyte specification, the 16-cell cyst becomes surrounded by somatic follicle cells. The oocyte is positioned at the posterior end of the cyst, in direct contact with follicle cells on all sides except at its anterior end, where it remains connected to the nurse cells via ring canals. This arrangement permits two essential processes. First, messenger RNAs (mRNAs) encoding determinants important for embryonic development (e.g., bicoid andoskar) are transcribed in the nurse cells and transported into the oocyte via the ring canals. Second, the oocyte and the follicle cells exchange signals that define both the A/P and dorsal/ventral (D/V) axes. In particular, follicle cells at the posterior end of the cyst send a signal that repolarizes the oocyte: the posterior MTOC is lost and replaced by a new microtubule network that is essential to localize bicoid and oskarRNAs to opposite poles of the oocyte [reviewed in (21)].

Early studies led to a simple model for bicoid andoskar RNA localization. A β-galactosidase fusion of the plus-end–directed microtubule motor Kinesin localizes to the posterior pole of the oocyte, whereas a β-galactosidase fusion of the minus-end marker Nod localizes to the anterior pole of the oocyte (44, 45). These localizations suggested that plus-end–directed motors might transport oskar mRNA to the posterior, whereas minus-end–directed motors might transport bicoid mRNA to the anterior. Consistent with this model, oskar RNA was reported to localize to the anterior inkinesin I heavy chain (khc) mutants (46).

A new study challenges this simple model. Reexamination of oskar localization in khc mutants showed thatoskar mRNA is distributed over the entire cortex, not just the anterior (47). Depolymerization of microtubules leads to a similarly broad delocalization (47). Surprisingly, components of the γ-tubulin ring complex, which nucleates microtubules by associating with their minus ends, are uniformly distributed over the entire oocyte cortex; only the microtubule themselves are present at a lower density in the posterior (47). Together, these findings suggest an alternative explanation for why oskar becomes delocalized in the absence of Kinesin. According to the new model (47), Kinesin is not required to transportoskar to the posterior per se but rather to move it away from the microtubule-rich anterior and lateral cortices. The decreased density of microtubules at the posterior subsequently allowsoskar access to the cortical actin network specifically in this region, thereby promoting its posterior localization.

These new findings imply that reduction of microtubule density at the posterior pole is a critical step in the localization ofoskar to the posterior. How is this step regulated? Current evidence points to PAR-1. Partial loss-of-function mutants inpar-1 retain enough activity to polarize the oocyte early, but they fail to localize oskar during the late polarization phase (12, 14). In these mutants, microtubule density remains uniform (14), andoskar mRNA accumulates in the center of the oocyte rather than the posterior (12, 14), as predicted by the new model. These observations suggest that PAR-1 regulatesoskar localization indirectly by destabilizing microtubules in the posterior (Fig. 3).

Figure 3

Comparison of par-1 phenotypes inDrosophila and C. elegans eggs. (A) Wild-type Drosophila oocytes: Microtubules (green) are nucleated around the entire cortex but at reduced density in the posterior due to PAR-1 (orange). oskar mRNA (blue) is transported away from the cortex by Kinesin (black circles), except in the posterior where the low density of microtubules allows access to the cortex. Oskar protein is stabilized in the posterior by PAR-1–dependent phosphorylation. (B) In the absence of par-1, microtubule density is uniform andoskar cannot access the cortex. (C) Wild-typeC. elegans zygotes: PAR-1 (orange) in the posterior promotes enrichment of the germline determinants PIE-1 (blue) and P granules (blue dots) by a microtubule-independent mechanism. PAR-1 also contributes to the posterior displacement of the mitotic spindle (green) by an unknown mechanism. (D) In par-1mutant embryos, PIE-1 and P granules are uniformly distributed and spindle asymmetry is compromised.

Downstream of PAR-1

Mammalian PAR-1 homologs destabilize microtubules by phosphorylating microtubule-associated proteins (48), suggesting that regulation of microtubule dynamics may be a conserved aspect of PAR-1 function. If so, is there any evidence that PAR-1 also regulates microtubule dynamics in C. elegans? PAR-1's effect on the localization of germline proteins is unlikely to be mediated by microtubules, but PAR-1 does play a role in the asymmetric positioning of the first spindle. During anaphase, the spindle becomes displaced toward the posterior as a result of asymmetric forces generated by interactions between the astral microtubules and the cortex (49). In par-2 andpar-3 mutants, these forces are balanced and the spindle remains in the middle of the egg. In par-1 mutants, spindle displacement is also compromised, although not as severely as in par-2 and par-3 mutants, suggesting that spindle asymmetry involves both par-1–dependent andpar-1–independent processes. At first glance, these observations seem consistent with a possible role for PAR-1 in regulating microtubule dynamics. However, other observations suggest that PAR-1's effect on spindle placement may be secondary to its segregation functions. For example, mex-5;mex-6 double mutants occasionally also misplace their spindle (30). Future studies will be required to distinguish whether PAR-1 influences microtubule dynamics directly or indirectly, by localizing other regulators.

Does this mean that C. elegans PAR-1 andDrosophila PAR-1 regulate egg polarity by completely different mechanisms? A new study reveals one possible area of overlap: control of protein stability (50). Oskar protein is a substrate for the PAR-1 kinase, and phosphorylation increases Oskar's stability in extracts (50). A moderate reduction in PAR-1 levels reduces Oskar protein levels in vivo, even when Oskar translation is uncoupled from localization. These observations suggest that, in addition to localizing oskar RNA, PAR-1 also protects Oskar protein from degradation. This function appears to be part of an amplifying feedback loop, because PAR-1 itself becomes restricted to the posterior in an Oskar-dependent manner (14). In C. elegans, PAR-1 stabilizes germ plasm proteins indirectly through regulation of other factors, whereas in Drosophila, PAR-1 stabilizes Oskar directly by phosphorylation. The two mechanisms, however, are not exclusive and could be used in concert for maximum efficiency.

The list of PAR-1 targets is unlikely to stop there. Rab11, a protein implicated in membrane trafficking, is required for efficient transport and anchoring of oskar mRNA to the posterior pole (51, 52). PAR-1 homologs in yeast and mammalian cells have been implicated in the regulation of the exocytic machinery (53). Thus, polarized secretion may be yet another mechanism used by PAR-1 to localize determinants.

An important task for the future will be to identify all direct targets of PAR-1. Recently, PAR-1 was found to bind to 14-3-3 proteins (19), a conserved family of proteins that recognize phosphorylated peptides (54). PAR-1 phosphorylation can create a binding site for 14-3-3, suggesting that PAR-1 substrates become bound to 14-3-3 after phosphorylation (19). 14-3-3 mutants in Drosophila have polarity defects identical to those seen in par-1 mutants, consistent with the idea that 14-3-3 binding is essential for PAR-1 signal transduction (19). Remarkably, one of the C. elegans 14-3-3 homologs is encoded by par-5 (9), and PAR-5 protein binds to PAR-1 in a yeast two-hybrid assay (19). PAR-5, however, is unlikely to function only with PAR-1 in C. elegans, because it is required for the initial establishment of PAR domains, a process that is independent of PAR-1 (9, 30). 14-3-3 proteins have been implicated in many cellular processes (54), including actin dynamics (55), and could potentially act multiple times in the PAR hierarchy. The identification of 14-3-3 proteins as potential mediators of PAR-1 function may facilitate the identification of PAR-1 substrates.


Three main themes emerge from a comparison of PAR functions in Drosophila and C. elegans eggs. First, PAR proteins act together to convert a transient polarity cue into a stably polarized axis. Second, of all the PARs, PAR-1 appears most directly involved in converting cortical polarity into cytoplasmic asymmetry. Last, PAR-1 orchestrates cytoplasmic asymmetries by impinging on diverse cellular functions, including microtubule dynamics, protein degradation, and, likely, many others. Thus, the secret to thepar genes' remarkable adaptation to different cell types may lie in their ability to regulate a number of basic cellular processes. Although much has been learned, a complete picture awaits the identification of the essential cell machineries that interact with the PARs. As E. B. Wilson predicted, the key to this problem also lies in the cell biology of the egg.

  • * To whom correspondence should be addressed. E-mail: gseydoux{at}jhmi.edu


View Abstract

Stay Connected to Science

Navigate This Article