Translation of Polarity Cues into Asymmetric Spindle Positioning in Caenorhabditis elegans Embryos

See allHide authors and affiliations

Science  20 Jun 2003:
Vol. 300, Issue 5627, pp. 1957-1961
DOI: 10.1126/science.1084146


Asymmetric divisions are crucial for generating cell diversity; they rely on coupling between polarity cues and spindle positioning, but how this coupling is achieved is poorly understood. In one-cell stage Caenorhabditis elegans embryos, polarity cues set by the PAR proteins mediate asymmetric spindle positioning by governing an imbalance of net pulling forces acting on spindle poles. We found that the GoLoco-containing proteins GPR-1 and GPR-2, as well as the Gα subunits GOA-1 and GPA-16, were essential for generation of proper pulling forces. GPR-1/2 interacted with guanosine diphosphate-bound GOA-1 and were enriched on the posterior cortex in a par-3– and par-2–dependent manner. Thus, the extent of net pulling forces may depend on cortical Gα activity, which is regulated by anterior-posterior polarity cues through GPR-1/2.

The mechanisms that establish cell polarity are increasingly well understood, but relatively little is known about how polarity cues are translated into appropriate spindle positioning (1, 2). The PAR proteins, which are essential for cell polarity across metazoan evolution (3), were originally identified in the nematode C. elegans (4), where they establish polarity along the anterior-posterior (AP) axis after fertilization. During mitosis in one-cell stage C. elegans embryos, PAR proteins govern an imbalance of forces acting along astral microtubules and pulling on spindle poles (5). As a larger net force pulls on the posterior spindle pole, the spindle elongates asymmetrically and the first division is unequal. The components required for the generation of pulling forces have not yet been established, nor has the mechanism by which AP polarity is translated into differential net forces exerted on the two spindle poles.

Simultaneous inactivation of the genes encoding Gα subunits GOA-1 and GPA-16 in one-cell stage C. elegans embryos causes an equal first division with no apparent polarity defects (6), which indicates that Gα signaling is required downstream of the PAR proteins or in a parallel pathway to mediate proper spindle positioning. Such a role may be evolutionarily conserved. While Drosophilai and the interacting GoLoco (Gαi/o-Loco)–containing protein Pins play a role in maintaining polarity cues (79), they also function in conjunction with the PAR-3 ortholog Bazooka and associated components to direct spindle positioning (10). The nature of the forces governing asymmetric spindle positioning has not been investigated in Drosophila, precluding analysis of the consequences of Gα inactivation at the biophysical level.

We performed an RNA interference (RNAi)–based functional genomic screen to identify cell division components in C. elegans (11) and uncovered two genes, gpr-1 and gpr-2 (G protein regulator), whose inactivation caused a striking spindle-positioning defect in one-cell stage embryos. Because gpr-1 and gpr-2 are >97% identical at the nucleotide level (12), they are likely both silenced when either gene is targeted by RNAi, as is the case for other gene pairs so closely related (13). gpr-1 and gpr-2 encode essentially identical 525–amino acid proteins harboring a coiled-coil domain (residues 186 to 213) and a GoLoco motif (residues 425 to 445).

In wild-type one-cell stage embryos (Fig. 1, A to C) (movie S1), the spindle elongates asymmetrically toward the posterior during anaphase, accompanied by vigorous transverse movements of the posterior spindle pole, which flattens at telophase. As a result, the first division is unequal, generating a larger anterior blastomere and a smaller posterior one. In gpr-1/2(RNAi) one-cell stage embryos (Fig. 1, D to F) (movie S2), the spindle did not elongate asymmetrically during anaphase in 25 of 30 cases. As a result, the first division was equal, generating two blastomeres of identical size. In the other five gpr-1/2(RNAi) embryos, the posterior spindle pole underwent weak posterior displacement, which, although clearly distinguishable from the wild type, resulted in unequal first division (14). In all gpr-1/2(RNAi) embryos, transverse movements of the posterior spindle pole and its flattening at telophase did not take place. Other defects were apparent when gpr-1 and gpr-2 were inactivated by RNAi, including aberrant centrosome and spindle positioning in two-cell stage embryos as well as occasional chromosome segregation defects (movie S2).

Fig. 1.

GPR-1/2 are required for asymmetric spindle positioning. (A and D) Time-lapse DIC microscopy of wild-type (A) and gpr-1/2(RNAi) (D) one-cell stage embryos. In this and other figures, anterior (0% egg length) is to the left; arrowheads point to spindle poles. Time elapsed is shown in minutes and seconds. Panels in (A) and (D) are at same magnification; scale bar, 10 μm. (B and E) Tracings of anterior and posterior spindle pole position of wild-type (B) and gpr-1/2(RNAi) (E) embryo displayed in (A) and (D), from spindle assembly until 100 s thereafter. Vertical dashed lines indicate starting positions of spindle poles. (C and F) Position of anterior and posterior spindle poles at the end of anaphase in 9 wild-type embryos (C) and 10 gpr-1/2(RNAi) embryos (F).

To determine whether the lack of posteri- or spindle displacement in one-cell stage gpr-1/2(RNAi) embryos resulted from defective AP polarity cues, we examined the distribution of PAR-6, PAR-3, PAR-2, and PAR-1, which are asymmetrically distributed along the AP axis in the wild type (1518). We found that all four PAR proteins were correctly distributed in gpr-1/2(RNAi) embryos (fig. S1). Moreover, the distribution of P granule ribonucleoproteins and of the germline protein PIE-1, which are restricted to the posterior of wild-type one-cell stage embryos in response to polarity cues (19, 20), was not altered (fig. S1) (14). Thus, lack of posterior displacement in gpr-1/2(RNAi) embryos was not due to defective AP polarity, and GPR-1/2 act downstream of the PAR proteins or in a parallel pathway to mediate proper spindle positioning.

Because the phenotype of gpr-1/2(RNAi) embryos is indistinguishable from that of goa-1/gpa-16(RNAi) embryos (6), and because other GoLoco-containing proteins regulate Gαi or Gαo activity (2123), we reasoned that GPR-1/2 may be required for Gα signaling in C. elegans embryos. Consistent with this view, simultaneous depletion of gpr-1, gpr-2, goa-1, and gpa-16 did not result in more severe phenotypic manifestations (14). Furthermore, as for goa-1/gpa-16(RNAi) embryos (6), inactivation of the Gα subunit gpb-1 or of the Gγ subunit gpc-2 did not rescue the phenotype of one-cell stage gpr-1/2(RNAi) embryos (14), which demonstrates that the spindle-positioning defect is not caused by excess Gα/Gγ activity.

We investigated whether GPR-1/2 physically interact with GOA-1 and/or GPA-16. Using a yeast two-hybrid assay, we detected an interaction with GOA-1 but not with GPA-16 (Fig. 2A). Similarly, glutathione S-transferase (GST)–GPR-1 bound to GOA-1 in a GST pull-down assay (Fig. 2B) but did not exhibit interaction above background with GPA-16 (14). The lack of detectable interaction between GPR-1 and GPA-16 in these assays may reflect weaker binding affinity. We used the GST pull-down assay to map the domain of GPR-1 that mediates interaction with GOA-1, and found it to reside within the C-terminal–most 185 amino acids (Fig. 2B). The GoLoco motif contained in this fragment, which is identical in GPR-1 and GPR-2, was sufficient for interaction (Fig. 2B).

Fig. 2.

GPR-1/2 interact with GOA-1–GDP through the GoLoco motif. (A) Two-hybrid experiment using a histidine reporter to test interaction between full-length GPR-1 and GOA-1 or GPA-16; plate contains 50 mM 3-aminotriazol. (B) GST pull-down experiment with in vitro translated [35S]GOA-1 (arrowhead) and GST fused to full-length GPR-1 (GST–GPR-1/FL) or fragments thereof (GST–GPR-1/NT, residues 1 to 340; GST–GPR-1/CT, residues 341 to 525; GST–GPR-1/GoLoco, residues 425 to 445). Quantification of two experiments (including the one shown) indicates the following average increases in radioactive material pulled down over GST alone: GST–GPR-1/FL, 18.6 (SD = 2.6); GST–GPR-1/NT, 1.8 (SD = 1.1); GST–GPR-1/CT, 14.0 (SD = 4.0); GST–GPR-1/GoLoco, 7.0 (SD = 1.9). GST–GPR-1/FL pulled down ∼10% of input material (14). (C) SPR binding assay testing nucleotide dependence of interaction between GOA-1 and GST–GPR-1/GoLoco; the y axis indicates specific binding (in relative resonance units) as subtracted from background binding to GST alone.

To investigate the nucleotide dependence of this interaction, we used a surface plasmon resonance (SPR) binding assay to test the ability of GOA-1 to bind to GST–GPR-1. Before injection over SPR surfaces, recombinant GOA-1 was first incubated with GTP-γ-S [guanosine 5′-O-(3′-thiotriphosphate)], with guanosine diphosphate (GDP)–AlF4 to mimic the transition state of GTP hydrolysis, or with GDP alone. The GoLoco motif of GPR-1 bound robustly to GOA-1–GDP (binding affinity Kd = 0.31 μM, χ2 = 44.3) but exhibited no binding to GTP-γ-S–bound or AlF4-activated GOA-1 (Fig. 2C) (fig. S2). Similar results were obtained with the C-terminal–most 185 amino acids of GPR-1 (fig. S2). By analogy with other GoLoco-containing proteins that specifically interact with GDP-bound Gα subunits (2123), GPR-1/2 likely function as guanine nucleotide dissociation inhibitors. Because the loss of GPR-1/2 caused a phenotype indistinguishable from that of GOA-1 and GPA-16, it appears that the GDP-bound form of Gα subunits, rather than the GTP-bound form, mediates spindle positioning in one-cell stage C. elegans embryos.

We next used in vivo spindle-severing experiments to investigate the extent of astral pulling forces in the absence of gpr-1 and gpr-2 or of goa-1 and gpa-16. We cut the spindle with a localized ultraviolet laser microbeam and monitored the behavior of spindle poles with the use of time-lapse differential interference contrast (DIC) microscopy (24). After severing in the wild type (Fig. 3, A, D, and E) (movie S3), the peak velocity of the posterior spindle pole was ∼40% greater than that of the anterior one, reflecting a larger net pulling force (5). Moreover, the posterior spindle pole traveled farther and underwent more extensive transverse oscillations (5). After severing in gpr-1/2(RNAi) embryos (Fig. 3, B, D, and E) (movie S4) (25), both spindle poles achieved the same peak velocity after severing, which was much less than that of even the anterior spindle pole in severed wild-type embryos. Moreover, the liberated spindle poles hardly traveled and did not undergo transverse oscillations. We found essentially identical results in goa-1/gpa-16(RNAi) embryos (Fig. 3, C to E) (movie S5) (25). Impaired pulling forces are unlikely to result from defective astral microtubules, because the microtubule network in fixed one-cell stage gpr-1/2(RNAi) or goa-1/gpa-16(RNAi) embryos was indistinguishable from that in the wild type (Fig. 4J) (14). Thus, Gα signaling is required to generate substantial pulling force on spindle poles during mitosis.

Fig. 3.

GPR-1/2, as well as GOA-1 and GPA-16, are required for generation of astral pulling forces. (A to C) Time-lapse DIC microscopy sequence of spindle-severing experiments in wild-type (A) (anterior is at the lower left), gpr-1/2(RNAi) (B), or goa-1/gpa-16(RNAi) (C) one-cell stage embryos. The first frame in each sequence corresponds to the time of the last laser shot (white bar indicates location of cut), the second frame is 7.5 s later, the third frame 7.5 s thereafter. All panels are at same magnification; scale bar, 10 μm. (D) Tracings of spindle pole position corresponding to sequences shown in (A) to (C). Tracings start at first laser shot (20 s after beginning of movies S3 to S5); arrowheads indicate time points corresponding to frames in (A) to (C). (E) Average peak velocities achieved by anterior and posterior spindle poles after severing [wild-type, n = 34, values from (5); values for gpr-1/2 (RNAi) and goa-1/gpa-16(RNAi) in (25)]. Error bars show SEM at the 0.95 confidence interval.

Fig. 4.

Cortical distribution of GPR-1/2 is asymmetric and controlled by polarity cues. Fixed embryos from wild-type [(A and B) early prophase one-cell stage; (C and D) metaphase one-cell stage; (E and F) early two-cell stage; (G and H) early four-cell stage], gpr-1/2(RNAi) [(I and J) telophase one-cell stage], and early two-cell stage GFP–GPR-2 (K and L), goa-1/gpa-16(RNAi) (M and N), par-3(it71) (O and P), and par-2(RNAi) (Q and R) stained with antibodies to GPR-1/2 (A to J, M to R) or GFP (K and L) and to α-tubulin (panels in right columns show GPR-1/2 or GFP in red, α-tubulin in green). All panels are at approximately same magnification; scale bar, 10 μm.

The spindle-severing experiments raised the possibility that GPR-1/2, as well as GOA-1 and GPA-16, also acted to ensure an imbalance of pulling forces in response to polarity cues, because residual forces in gpr-1/2(RNAi) or goa-1/gpa-16(RNAi) embryos were equal on both spindle poles. We investigated whether the distribution of GPR-1/2 may offer an explanation for this observation. Antibodies to a peptide identical in GPR-1 and GPR-2 labeled the cell cortex and the cytoplasm, as well as the vicinity of centrosomes to a lesser extent (Fig. 4, A to H) (fig. S3). All aspects of the staining pattern were essentially abolished in gpr-1/2(RNAi) embryos (Fig. 4I). Although the distribution of GPA-16 has not been reported, GOA-1 is present at the cell cortex and the cytoplasm as well as in the vicinity of centrosomes (6, 26), which supports the notion that GPR-1/2 function with GOA-1 in vivo.

We analyzed the distribution of GPR-1/2 in detail throughout the cell cycle. Cortical GPR-1/2 were uniform in 67% (n = 60) of prophase one-cell stage embryos (14) but slightly asymmetric in 33% of them, with more protein at the posterior cortex (Fig. 4, A and B) (fig. S3). Between the end of prophase and early anaphase, 96% of one-cell stage embryos (n = 24) exhibited an enriched distribution at the posterior cortex (Fig. 4, C and D) (fig. S3). A similar enrichment was also apparent early in the cell cycle in the posterior blastomeres P1 and P2 at the two- and four-cell stages, respectively (Fig. 4, E to H). We obtained an identical distribution with antibodies from another rabbit immunized against the same peptide (14). To independently assess the distribution of GPR-1 and GPR-2, we generated transgenic animals expressing GPR-2 fused to GFP, which exhibited a distribution similar to that seen with antibodies to GPR-1/2 (Fig. 4, K and L).

Both GPR-1 and GPR-2 lack discernible protein motifs that may target them to the plasma membrane. In contrast, both GOA-1 and GPA-16 harbor N-myristoylation sites, raising the possibility that the presence of GPR-1/2 at the cell cortex occurs through interaction with membrane-tethered Gα subunits. Consistent with this view, the cortical distribution of GPR-1/2 was severely diminished in goa-1/gpa-16(RNAi) embryos [Fig. 4, M and N; average cortical/cytoplasmic signal of 0.92 (SD = 0.30, n = 8) versus 1.75 in the wild type (SD = 0.26, n = 9)].

We used the cortical asymmetry of GPR-1/2 to address whether they act downstream of the PAR proteins or in a parallel pathway. We focused on early two-cell stage embryos in which the enrichment of GPR-1/2 at the posterior cortex was easiest to score and apparent in 81% of the wild-type embryos (n = 57; Fig. 4E). The cortical distribution of GPR-1/2 was uniform in 100% of par-3 mutant and par-2(RNAi) early two-cell stage embryos (n = 24 and 17, respectively; Fig. 4, O to R). Moreover, levels at the cell cortex were stronger in par-3 mutant embryos than in par-2(RNAi) embryos, as expected because both daughter cells exhibit posterior-like features in the absence of par-3 function and anterior-like features in the absence of par-2 function (3). Analogous conclusions were reached by examining one-cell stage embryos (14). Furthermore, spindle-severing experiments in embryos simultaneously lacking par-3 function and gpr-1/2 function revealed that gpr-1/2 are epistatic to par-3 for force generation (25). Thus, gpr-1/2 act downstream of AP polarity cues to mediate proper spindle positioning in one-cell stage embryos. par-2 and par-3 also control the distribution of the DEP domain–containing protein LET-99 in a posterior cortical belt in one-cell stage embryos (27). However, the first cleavage is unequal in let-99 mutant embryos (28), which suggests that let-99 is not essential for differential force generation during anaphase.

We propose a working model in which AP polarity cues set by the PAR proteins translate into distinct pulling forces on the two spindle poles via differential activation of Gα signaling at the cell cortex (fig. S4). Because there is more GPR-1/2 on the posterior cortex, there is also more Gα signaling, and a larger net pulling force is exerted on the posterior spindle pole. A fluctuation analysis predicts that an imbalance of pulling forces stems from a difference in the number of cortical force generators pulling on astral microtubules, with ∼50% more being present on the posterior cortex (29). The posterior enrichment of GPR-1/2 is of a similar extent (fig. S3), which suggests that Gα signaling at the cell cortex is a rate-limiting step for force generator function. Given that cortically localized Gαi and Pins also dictate spindle positioning in Drosophila (10), we propose that modulation of cortical Gα signaling to generate defined pulling forces on astral microtubules is a conserved mechanism to translate polarity cues into appropriate spindle positioning.

Supporting Online Material

Materials and Methods

Figs. S1 to S4


Movies S1 to S7

References and Notes

View Abstract

Navigate This Article