PP1 phosphatases control PAR-2 localization and polarity establishment in C. elegans embryos

Cell polarity is a fundamental property of cells required for many aspects of cell and animal biology. In migrating cells, for example, a front–rear polarity regulates migration in response to chemokines or antigens (Llense and Etienne-Manneville, 2015); in stem cells, cell polarity is a prerequisite for asymmetric cell division (Santoro et al., 2016); and in epithelial cells, the apical–basal polarity axis is required to establish the barrier function of the epithelium (Riga et al., 2020; Rodriguez-Boulan and Macara, 2014; Roignot et al., 2013).

In many different cells, polarity is regulated by the conserved partitioning defective (PAR) proteins, which have been identified in Caenorhabditis elegans (Goldstein and Macara, 2007; Rose and Gönczy, 2014).

The C. elegans zygote is a powerful model system for investigating the mechanism of cell polarity establishment and maintenance. The one-cell C. elegans embryo is polarized along the anterior–posterior (A-P) axis, with the anterior PAR proteins (the PDZ proteins PAR-3 and PAR-6; the atypical protein kinase C PKC-3; and the small GTPase CDC-42, from now on referred to as anterior PARs) enriched at the cortex in the anterior half of the embryo, and the posterior PAR proteins (the ring finger protein PAR-2; the kinase PAR-1; the lethal giant larvae ortholog, LGL-1; and the CDC-42 GAP CHIN-1, referred to as posterior PARs) enriched at the posterior cortex (reviewed in Goehring, 2014; Lang and Munro, 2017). This polarization results in a first asymmetric cell division, giving origin to two cells, AB and P1, with different sizes and fates. The zygote polarizes in two distinct phases, establishment and maintenance (Cuenca et al., 2003). Just after fertilization, PAR-3, PAR-6, and PKC-3 are uniformly distributed at the cortex, whereas PAR-1 and PAR-2 are in the cytoplasm (Fig. 1 A). Protein kinase C (PKC-3) phosphorylates PAR-2 and PAR-1, inhibiting their cortical localization and polarity establishment (Folkmann and Seydoux, 2019; Hao et al., 2006; Motegi et al., 2011). Polarity establishment relies on two redundant pathways, both dependent on the centrosomes. Shortly after the fertilization, a gradient of the mitotic kinase Aurora A (AIR-1) from the centrosomes of the paternal pronucleus triggers a cortical flow away from the newly defined posterior pole (Kapoor and Kotak, 2019; Klinkert et al., 2019; Reich et al., 2019; Zhao et al., 2019). This initiates the segregation of the anterior PARs to the anterior side of the embryo and liberates the posterior pole, allowing the localization of PAR-2 (and PAR-1; Munro et al., 2004). A second, redundant pathway relies on centrosomes-emanating microtubules, which promote loading of PAR-2 in the posterior (Fig. 1 A; Motegi et al., 2011). PAR-2 recruits PAR-1, thereby excluding the anterior PARs via PAR-1–dependent phosphorylation of PAR-3 (Motegi et al., 2011). Once polarity is established, mutual antagonism between the anterior and posterior PAR proteins ensures polarity maintenance (reviewed in Gubieda et al., 2020; Rose and Gönczy, 2014).

PKC-3 phosphorylates PAR-2 and inhibits its membrane association (Hao et al., 2006). Despite the fact that centrosomal microtubules can protect PAR-2 from PKC-3 phosphorylation and therefore promote PAR-2 membrane association, mutations of the PAR-2 microtubule binding sites delay but do not abolish posterior cortical loading of PAR-2 in presence of normal cortical flows (Motegi et al., 2011). On the contrary, when the PKC-3 phosphorylation sites in PAR-2 are mutated to mimic phosphorylation, PAR-2 localization at the posterior cortex is abrogated, resulting in a defect in polarity establishment (Hao et al., 2006). This suggests that PAR-2 phosphorylation by PKC-3 must be relieved to ensure PAR-2 posterior cortical localization and hence proper polarity establishment.

Here, we show that the PP1 phosphatase GSP-2 plays an important role in polarity establishment. GSP-2 depletion suppresses the lethality (as previously shown Fievet et al., 2013) and the polarity defects of a temperature-sensitive pkc-3 mutant, suggesting that GSP-2 antagonizes PKC-3 function. GSP-2 depletion also results in defects in PAR-2 cortical localization that are exacerbated by the depletion of the PP1 phosphatase GSP-1. Consistent with a role of GSP-2 and GSP-1 in PAR-2 dephosphorylation, PAR-2 contains a PP1 binding motif, which we show is required for the interaction with GSP-1 and GSP-2 in two-hybrid assays. Mutations of this site known to abolish PP1 binding abrogate PAR-2 cortical localization and polarity establishment in vivo. On the contrary, mutations that optimize the PP1 binding motif extend PAR-2 cortical localization.

Our work identifies the PP1 phosphatases, GSP-2 and GSP-1, as critical regulators of PAR-2 cortical localization and polarity establishment in the C. elegans embryo.

GSP-2 has been identified as a suppressor of the embryonic lethality caused by a temperature-sensitive mutant allele of pkc-3 (pkc-3[ne4246]; Fievet et al., 2013). PKC-3 phosphorylates PAR-2 and inhibits its localization at the anterior cortex (Hao et al., 2006; Motegi et al., 2011). In the pkc-3(ne4246) strain, at 25°C, PAR-2 occupies the entire cortex (Rodriguez et al., 2017), resulting in polarity defects and embryonic lethality. We, therefore, asked whether depletion of GSP-2 was able to rescue the aberrant localization of PAR-2 observed in the pkc-3(ne4246) embryos. Embryonic lethality of the pkc-3(ne4246); gfp::par-2; ctrl(RNAi) strain at 24°C (the semirestrictive temperature at which PKC-3 is still partially active; see Materials and methods) was about 93.5% (Fig. 1 B). Consistent with the high embryonic lethality, embryos from pkc-3(ne4246); gfp::par-2; ctrl(RNAi) mutant strain had impaired A-P polarity, with PAR-2 not being restricted anymore to the posterior, but distributed all around the cortex in one-cell stage embryos and partitioned symmetrically into the daughter AB and P1 cells (Fig. 1 C and Video 1). Depletion of GSP-2 in gfp::par-2 worms did not result in embryonic lethality, and PAR-2 was localized at the posterior cortex (Fig. 1, B and C). Depletion of GSP-2 in the pkc-3(ne4246); gfp::par-2 mutant rescued the embryonic lethality (Fig. 1 B). We found that PAR-2 localization was restored at the posterior cortex in one-cell stage embryos and was restricted to the P1 blastomere in two-cell stage embryos (Fig. 1 C and Video 1). Therefore, depletion of GSP-2 in the pkc-3(ne4246); gfp::par-2 rescued embryonic lethality and PAR-2 localization defects of early embryos.

Polarity controls the posterior positioning of the mitotic spindle leading to a more posterior cleavage and to two-cell embryos with a bigger anterior cell (AB) and a smaller posterior cell (P1). Depletion of GSP-2 in the gfp::par-2 control strain resulted in a higher AB/P1 ratio compared to ctrl(RNAi) (Fig. S1 A). In the pkc-3(ne4246); gfp::par-2; ctrl(RNAi), the cleavage furrow was shifted more toward the anterior, resulting in a reduced size asymmetry of the daughter cells. This phenotype was also rescued by depletion of GSP-2 (Fig. S1 A).

The two-cell embryos of C. elegans divide asynchronously, with the anterior AB cell dividing about 2 min before P1. This asynchrony is lost in the pkc-3 mutant. We, therefore, tested whether GSP-2 depletion was able to rescue this phenotype. AB divided roughly 2 min before P1 both in ctrl(RNAi) and in gsp-2(RNAi) embryos, whereas in the pkc-3(ne4246) mutant allele, the division time of AB and P1 was not significantly different (Fig. 1 D and Video 2). Depletion of GSP-2 in the pkc-3(ne4246) strain was able to restore the asynchrony between AB and P1 (Fig. 1 D and Video 2).

One mechanism behind the regulation of asynchrony is the polarity-dependent localization of the mitotic kinase Polo-Like Kinase 1 (PLK-1). In the one-cell embryo, PLK-1 becomes enriched in the anterior cytoplasm and, at division, it is preferentially segregated in AB. This enrichment in AB triggers the earlier division of this blastomere (Budirahardja and Gönczy, 2008; Nishi et al., 2008; Rivers et al., 2008). We, therefore, investigated the localization of PLK-1. In gsp-2(RNAi) embryos, PLK-1 was enriched in the anterior AB cell as in ctrl(RNAi) embryos (Fig. S1 B). In the pkc-3(ne4246) embryos, the AB/P1 ratio of PLK-1 levels was close to 1, consistent with what has been previously reported in absence of the anterior PARs (Budirahardja and Gönczy, 2008; Nishi et al., 2008; Rivers et al., 2008). In pkc-3(ne4246); gsp-2(RNAi) embryos, PLK-1 anterior enrichment was restored (Fig. S1 B).

Taken together, these results show that depletion of GSP-2 suppresses the embryonic lethality and the polarity defects of the pkc-3(ne4246) mutant allele.

The results that GSP-2 depletion rescues the PAR-2 localization defects of pkc-3(ne4246) mutant embryos suggest that GSP-2 and PKC-3 antagonize each other in the regulation of PAR-2 cortical localization. PKC-3 phosphorylates PAR-2, thereby inhibiting its membrane localization (Hao et al., 2006). Since GSP-2 is a phosphatase, one possibility is that GSP-2 dephosphorylates PAR-2 allowing its posterior cortical localization. If this was the case, depletion of GSP-2 in control embryos should result in a reduction of cortical PAR-2, as PAR-2 would not be dephosphorylated. When GSP-2 was depleted in gfp::par-2 embryos, PAR-2 was localized at the posterior cortex (Fig. 1 C and Fig. 2 A). However, the size of the PAR-2 domain was smaller (Fig. 2 A), indicating that GSP-2 contributes to the formation of a PAR-2 domain of the correct size.

C. elegans embryos express a second PP1 catalytic subunit, GSP-1, which is 85% identical to GSP-2 in the amino acid sequence (Sassa et al., 2003). GSP-1 and GSP-2 localize uniformly throughout the cytoplasm, in the nucleus, and on kinetochores in embryos (Fig. S2, A and B; Hattersley et al., 2016; Kim et al., 2017; Mangal et al., 2018), and they partially overlap in function (Peel et al., 2017). We, therefore, investigated whether GSP-1 can compensate for the function of GSP-2 in promoting PAR-2 cortical localization in one-cell embryos. We first asked whether GSP-1 depletion rescued the embryonic lethality of pkc-3(ne4246); gfp::par-2 worms. Depletion of GSP-1 in gfp::par-2 worms did not result in embryonic lethality, and its depletion in the pkc-3(ne4246); gfp::par-2 mutant did not rescue the embryonic lethality of this strain (Fig. 2 B). Consistent with this, PAR-2 was detected uniformly at the cortex in both pkc-3(ne4246); gfp::par-2; ctrl(RNAi) and pkc-3(ne4246); gfp::par-2, gsp-1(RNAi; Fig. S3 A and Video 3). In addition, embryos depleted of GSP-1 did not show a smaller PAR-2 domain compared to ctrl(RNAi) embryos (Fig. S2 B). Depletion of GSP-1 was efficient and specific (Fig. S4, A and B). Therefore, in contrast to GSP-2 depletion, GSP-1 depletion did not suppress the embryonic lethality and polarity defects observed in the pkc-3(ne4246) mutant allele and did not impair the size of the PAR-2 domain.

We then asked if codepletion of GSP-1 and GSP-2 in the gfp::par-2 embryos impaired PAR-2 localization. Consistent with the previous results, we could observe a small but significant reduction in the size of the PAR-2 domain in the ctrl(RNAi); gsp-2(RNAi) embryos, while the PAR-2 domain was not reduced in the ctrl(RNAi); gsp-1(RNAi) embryos (Fig. 2, C and D). The intensity of the PAR-2 domain did not change in the single depletion (Fig. 2 E). In the GSP-1/-2 depleted embryos, we observed two phenotypes: in the majority of the embryos PAR-2 was mostly cytoplasmic (class I, 78.6%), as shown by an almost flat intensity profile, whereas a smaller percentage of embryos (class II, 21.4%) showed a weak localization of PAR-2 at the posterior cortex (Fig. 2, C–E; and Video 4).

To conclude, codepletion of GSP-1 and GSP-2 results in a defect in PAR-2 posterior cortical localization. In addition, our data suggest that GSP-2 has a leading role in the regulation of PAR-2 localization, but GSP-1 can compensate in the absence of GSP-2.

GSP-2 and a PAR-2 N-terminal fragment (Fig. 3 A) were identified as interactors in a large-scale two-hybrid screen (Koorman et al., 2016). The genetic and physical interactions suggest that GSP-2 (with GSP-1) is the phosphatase that counteracts PKC-3 phosphorylation of PAR-2. We, therefore, tested whether GSP-2 and GSP-1 interacted with PAR-2. As previously shown (Koorman et al., 2016), we confirmed that GSP-2 interacted with PAR-2 (Fig. 3 B). Consistent with the redundancy in regulating PAR-2 localization, GSP-1 also interacted with PAR-2 (Fig. 3 B).

Analysis of the PAR-2 amino acid sequence revealed the presence of a degenerate PP1 docking motif (164 RLFF 167) in the N-terminus of PAR-2, conserved in closely related nematode species (Fig. 3 A, upper panel, and Fig. 3 C). This suggests that PAR-2 may physically interact with PP1 through this motif (Egloff et al., 1997; Hendrickx et al., 2009; Wakula et al., 2003; Zhao and Lee, 1997). To assess if the PP1 docking motif present in PAR-2 is required for the interaction with GSP-1 and GSP-2, we mutated two critical amino acid residues into alanine (referred to as PAR-2 [RAFA], Fig. 3 A, lower panel). These substitutions have previously been shown to interfere with the binding between PP1 phosphatases and their substrates (Meiselbach et al., 2006; Moreira et al., 2019). Interestingly, neither GSP-1 nor GSP-2 could interact with PAR-2 in the yeast two-hybrid system when the PP1 docking motif in PAR-2 was mutated (Fig. 3 B).

These data show that PAR-2 interacts with both GSP-2 and GSP-1 in the two-hybrid assay, and this interaction depends on a PP1 docking motif in PAR-2.

We set out to assess whether cell polarity was impaired if the PP1 docking motif in PAR-2 was mutated. We generated a strain in which the PP1 RLFF motif was mutated to RAFA in the endogenous gfp::par-2 (referred as gfp::par-2[RAFA]). The homozygote mutant worms were viable but exhibited high embryonic lethality (95.4% ± 4.5 SEM), and the surviving progeny was sterile (Fig. S5 A and Materials and methods). Our control, referred to as gfp::par-2°, was a mixture of homozygous worms expressing wild-type gfp::par-2 from both alleles and heterozygous worms expressing wild-type gfp::par-2 from one allele and mutant gfp::par-2(RAFA) from the other allele. We could not detect any difference in cortical PAR-2 domain size and intensity in the gfp::par-2° population compared to the gfp::par-2 (Fig. S5, B and C; and Materials and methods).

The gfp::par-2(RAFA) mutant embryos displayed phenotypes consistent with the impairment of polarity establishment. PAR-2 remained mostly cytoplasmic (Fig. 4, A and B; and Video 5), consistent with the phenotype of embryos codepleted of both PP1 catalytic subunits and similar to what has been shown with the phosphomimetic mutant of PAR-2, in which seven PKC-3 phosphorylation sites have been mutated to glutamic acid (Hao et al., 2006). The position of the cleavage furrow during the first cell division was variable and shifted toward the anterior, resulting in a more symmetric first cell division with AB and P1 of equal size (Fig. S5 D). The levels of GFP::PAR-2(RAFA) were not significantly different from GFP::PAR-2 (Fig. S5, E and F, and figure legend), indicating that this phenotype is not the result of reduced PAR-2.

In control embryos, the P1 spindle rotates to be oriented along the A-P axis, whereas the AB spindle is orthogonal to the A-P axis (Fig. 4 C). Mutations that interfere with the establishment of polarity can alter spindle orientation at the second division (Cheng et al., 1995; Kemphues et al., 1988). We found that in the gfp::par-2(RAFA) mutant embryos, the P1 spindle failed to rotate, resulting in an irregular arrangement of cells at the four-cell stage (Fig. 4 C). Furthermore, the AB and P1 blastomeres divided synchronously (Fig. 4 D and Video 6). Therefore, mutation of the PP1 docking motif in the endogenous PAR-2 strongly reduced PAR-2 cortical localization and led to polarity defects similar to the ones observed in par-2 loss of function embryos (Cheng et al., 1995; Kemphues et al., 1988).

If the phosphatase PP1 antagonizes the kinase activity of PKC-3 on PAR-2, depletion of PKC-3 in the gfp::par-2(RAFA) mutant strain should result in PAR-2 localizing around the entire cortex. pkc-3(RNAi); gfp::par-2(RAFA) embryos displayed uniform PAR-2 cortical localization in all the embryos analyzed (Fig. 4, E and F).

These results suggest that the interaction between PAR-2 and GSP-1/-2 has a crucial role in the regulation of PAR-2 cortical localization and establishment of polarity.

One of the hallmarks of the PP1 docking motif is the high degeneracy at key positions of the motif (Davey et al., 2015). The optimal PP1 docking motif is the RVxF sequence (Wakula et al., 2003), and the valine was shown in vitro to contribute to a stronger interaction between PP1 and its substrates (Meiselbach et al., 2006). The PP1 docking motif in PAR-2 presents a leucine at position 165 instead of the optimal valine (Fig. 3 A). We, therefore, asked if optimizing the PP1 docking motif by mutating leucine to valine in vivo would alter the cortical localization of PAR-2. We hypothesized that if this mutation improves the binding of PAR-2 to PP1 phosphatases, the PAR-2 cortical domain may be extended.

We generated a strain where leucine 165 in the PP1 docking motif was mutated to valine in the endogenous gfp::par-2 (referred as gfp::par-2(L165V)). The mutant strain did not show embryonic lethality compared to gfp::par-2 worms (0.31% ± 0.25 SEM vs. 0.17% ± 0.06 SEM respectively), and the progeny was fertile. All gfp::par-2(L165V) zygotes, analyzed before symmetry breaking, displayed PAR-2 localization uniformly around the entire cortex, a phenotype that was not observed in control embryos (Fig. 5 A). During pronuclei migration, we detected both an anterior and a posterior PAR-2 domain in all the zygotes analyzed, indicating that in this strain restriction of PAR-2 to the posterior cortex is impaired. The anterior PAR-2 domain remained in 32% of embryos in mitosis and 16% of two-cell embryos (Fig. 5 A and Video 7). gfp::par-2(L165V) embryos are viable (see above), and the resulting worms are not sterile and do not show any detectable phenotype, suggesting that either the anterior PAR-2 domain observed in 16% of embryos is removed later or it has no deleterious effect on embryonic development. GFP::PAR-2(L165V) levels were not increased compared to GFP::PAR-2, indicating that the increase in cortical PAR-2 is not due to increased PAR-2 levels (Fig. 5, B and C, and figure legend).

If the rate of PP1-dependent dephosphorylation was increased in the gfp::par-2(L165V) mutant strain, depletion of GSP-2 should rescue the aberrant localization of PAR-2 in this mutant. PAR-2 showed normal localization at the posterior pole in 75% of the gfp::par-2(L165V); gsp-2(RNAi) early embryos and in 87.5% of the embryos analyzed at the onset of pronuclear migration; from pronuclear migration to later stage all the embryos analyzed showed only one PAR-2 domain at the posterior cortex (Fig. 5 D).

Collectively these results show that optimizing in vivo the PP1 binding motif of PAR-2 results in aberrant cortical localization of PAR-2 before and during polarity establishment. This localization is corrected in most but not all embryos at later stages, suggesting that other yet unknown mechanisms are involved in the restriction of PAR-2 at the posterior cortex.

An anterior PAR-2 domain, similar to the one observed in the gfp::par-2(L165V) mutant, has been reported in embryos in meiosis (Wallenfang and Seydoux, 2000) and in embryos where the activity of the mitotic kinases PLK-1 and AIR-1 has been reduced (Kapoor and Kotak, 2019; Klinkert et al., 2019; Noatynska et al., 2010; Reich et al., 2019; Schumacher et al., 1998). We asked whether the anterior PAR-2 domain observed in the plk-1(or683) embryos might depend on GSP-2. To test this, we used the temperature-sensitive mutant allele plk-1(or683). We stained plk-1(or683); ctrl(RNAi) embryos with PAR-2 antibodies and found that PAR-2 localized at both the anterior and posterior domains, whereas in the wild type; ctrl(RNAi) embryos only one PAR-2 domain was detected (Fig. 6). Depletion of GSP-2 in the plk-1(or683) embryos resulted in a significant reduction of embryos with the anterior PAR-2 domain (Fig. 6). In addition, the length of the PAR-2 domain in the gsp-2(RNAi); plk-1(or683) embryos was comparable to the PAR-2 domain in control embryos, indicating that reduction of PLK-1 activity can rescue the smaller PAR-2 domain caused by GSP-2 depletion (Fig. 6 B).

Therefore, the anterior PAR-2 domain in the plk-1(or683) mutant allele depends, directly or indirectly, on the GSP-2 phosphatase, and the shorter PAR-2 domain caused by GSP-2 depletion depends on PLK-1, suggesting a genetic interaction in polarity regulation between PLK-1 and GSP-2.

Establishment of the A-P axis is an essential process to ensure asymmetric cell division and proper development of the C. elegans embryo. Here, we show that in the one-cell embryo, this process is regulated by PP1 phosphatases. We find that the PP1 phosphatases GSP-2 and GSP-1 are required for the loading of PAR-2 at the cortex. Mutations in the PP1 binding motif of PAR-2, which abrogate the interaction with GSP-1 and GSP-2 in yeast two-hybrid assays, result, in vivo, in the inability of PAR-2 to properly localize at the cortex. Our data suggest a model in which GSP-2 and GSP-1 counterbalance the activity of PKC-3 in the early embryo, allowing PAR-2 posterior cortical localization (Fig. 7).

These data also support previous findings in the field. Establishment of cell polarity in the one-cell C. elegans embryos relies on cortical flows that displace anterior PARs, including PKC-3, from the posterior, allowing loading of PAR-2 and PAR-1 (Munro et al., 2004; Shelton et al., 1999). In addition to this pathway, binding of PAR-2 to astral microtubules protects PAR-2 from PKC-3 phosphorylation and promotes its posterior cortical localization, therefore contributing to polarity establishment (Motegi et al., 2011). However, abolishing binding of PAR-2 to microtubules does not abrogate polarity establishment in the presence of cortical flows while a PAR-2 mutant that mimics PKC-3 phosphorylation is unable to polarize the embryo (Hao et al., 2006; Motegi et al., 2011). This suggests that the major polarization pathway involving the flows needs the action of phosphatases to dephosphorylate PAR-2, as shown by our data.

The PP1 phosphatase GSP-2 was previously shown to be a suppressor of the embryonic lethality of the pkc-3(ne4246) mutant allele (Fievet et al., 2013). We find that depletion of GSP-2 in the pkc-3(ne4246) allele also rescues the polarity-related defects observed in the pkc-3(ne4246) mutant alone. Depletion of GSP-2 in otherwise control strains does not result in major defects in polarity establishment, as it would have been expected if GSP-2 was the only phosphatase targeting PAR-2. Codepletion of GSP-1 does result in polarity establishment defects. Although the polarity phenotype is enhanced in absence of both GSP-1 and GSP-2, only GSP-2 depletion can rescue pkc-3(ne4246) embryonic lethality and polarity defects and result in a smaller size of the PAR-2 domain in one-cell embryos, suggesting that GSP-2 is the phosphatase that has a more important role in polarity regulation. This is reminiscent of previous work showing that GSP-1 and GSP-2 have a partially overlapping role in the regulation of centriole amplification. Interestingly, in this process, GSP-1 plays a more important role (Peel et al., 2017).

We find that PAR-2 contains a PP1 docking motif, a common motif used by PP1 phosphatases to physically interact with their substrates (Egloff et al., 1997; Hendrickx et al., 2009; Wakula et al., 2003; Zhao and Lee, 1997). This motif is also present in PAR-2 of closely related nematode species, suggesting that dephosphorylation of PAR-2 by PP1 is a conserved process in nematodes. In this context, it is interesting to note that in C. elegans embryos the ortholog of the lethal giant larvae protein, LGL-1, can partially compensate for the loss of PAR-2 (Beatty et al., 2010; Hoege et al., 2010). In flies, Lgl is also an important regulator of polarity (Su et al., 2012). In epithelial cells, Lgl is removed from the cortex during mitosis in an aPKC and Aurora A phosphorylation-dependent manner. Restoration of cortical localization of Lgl is essential to maintain cell polarity and tissue architecture and is regulated by a PP1 phosphatase (Moreira et al., 2019), similar to what we observe for PAR-2 in C. elegans zygotes.

Consistent with the role of the PP1 docking motif for the interaction with PAR-2 and the phenotype observed with the codepletion of both GSP-1 and GSP-2, mutations of the two important amino acids in the PP1 docking motif (leucine165 to alanine and phenilalanine167 to alanine, gfp::par-2[RAFA]) abrogate the interaction with PAR-2 in the yeast two-hybrid system; more importantly, in vivo, these mutations impair the localization of PAR-2 at the posterior cortex and embryo viability, resembling the par-2 loss-of-function phenotype, consistent with the fact that dephosphorylation of PAR-2 is required for polarity establishment (Hao et al., 2006; Fig. 4 and Fig. S5). The gfp::par-2(RAFA) mutant showed two interesting features. First, embryos from gfp::par-2° mothers did not show any difference in PAR-2 cortical levels and PAR-2 domain size compared to embryos from homozygote wild-type worms. We speculate that thanks to the property of PAR-2 to form oligomers (Arata et al., 2016), mutant PAR-2 can form oligomers with wild-type PAR-2 and localize properly to the cortex. Second, we also noticed that the gfp:par-2(RAFA) mutant shows a weak PAR-2 localization at the posterior cortex (Video 5). One possibility is that in this mutant, the microtubule redundant pathway protects some PAR-2 from phosphorylation by PKC-3, but this alone is not sufficient to ensure the proper establishment of polarity.

Based on previous in vitro studies (Meiselbach et al., 2006), we have also optimized the PP1 binding motif in vivo. The leucine 165 to valine mutant (gfp::par-2(L165V)) showed an aberrant PAR-2 localization around the entire cortex prior to polarity establishment. During polarity establishment and at later stages, some of the embryos displayed an anterior and a posterior PAR-2 domain. This phenotype is reminiscent of the phenotype observed in embryos where PLK-1 and AIR-1 have been depleted (Kapoor and Kotak, 2019; Klinkert et al., 2019; Noatynska et al., 2010; Reich et al., 2019; Schumacher et al., 1998). Aurora A can be an activator of Polo-like kinase (Macurek et al., 2008; Seki et al., 2008; Tavernier et al., 2015). PLK-1 in C. elegans embryos is enriched in the anterior cytoplasm (Budirahardja and Gönczy, 2008; Nishi et al., 2008; Rivers et al., 2008). One possibility is that Aurora A–activated PLK-1 contributes to keeping the GSP-2 levels/activity low at the anterior so that PAR-2 remains cytoplasmic (Fig. 7 A). This is consistent with previous genetic data showing that PLK-1 depletion can rescue PAR-2 cortical localization in par-2 temperature-sensitive mutants (Noatynska et al., 2010) and with our current data that GSP-2 depletion in plk-1(or683) reduces the number of embryos with an anterior PAR-2 domain. However, consistent with published data, we did not observe any asymmetry in the localization/levels of GSP-2 and GSP-1 (Fig. S2). PLK-1 may phosphorylate and inhibit GSP-2 activity directly (Fig. 7 A). Our efforts to mutagenize predicted PLK-1 phosphorylation sites of GSP-2 into nonphosphorylatable ones did not show any phenotype consistent with an upregulation of GSP-2 activity. GSP-1 and GSP-2 are catalytic subunits of the protein phosphatase PP1. Catalytic subunits bind to regulatory subunits and these, in turn, regulate substrate specificity and the activity of the holoenzyme (Aggen et al., 2000). Therefore, PLK-1 might act on the yet-to-be-identified regulatory subunits to control PP1 activity.

Experiments and modeling have shown that a spatially segregated kinase and a uniform phosphatase activity can generate a protein gradient (Griffin et al., 2011). Therefore, the asymmetric localization and activity of PKC-3 in the one-cell embryo could be sufficient to explain PAR-2 posterior cortical enrichment, even in the absence of a GSP-2 activity gradient (Fig. 7 B). Further studies are needed to understand whether the genetic interaction that we observe between PLK-1 and GSP-2 reflects a regulation of PP1 activity, direct or indirect, by PLK-1 that contributes to polarity establishment and maintenance.

Altogether, our data suggest a model in which the proper balance between the activity of PP1 phosphatases and PKC-3 is crucial to properly establish cortical polarity in one-cell embryos.

The C. elegans strains used in this work are listed in Table S1. Worms were maintained on Nematode Growth Medium (NGM) plates seeded with OP50 bacteria, using standard methods (Brenner, 1974). Thermosensitive strains, such as pkc-3(ne4246), pkc-3(ne4246); gfp::par-2, and plk-1(or683), were maintained at 15°C. All the other strains were maintained at 22°C.

Mutant strains were generated using CRISPR/Cas9 technology as described in Arribere et al. (2014). Single guide RNAs, repair templates to generate the mutant, PCR primers used to detect and sequence the mutation, and enzymes used for the screening are listed in Tables S2, S3, and S4. The ZU316 (gfp::par-2(L165V)) was backcrossed three times. For the ZU297 (gfp::par-2[RAFA]), two independent isolates were analyzed. The gfp::par-2(RAFA) strain was kept as a heterozygote (gfp::par-2/gfp::par-2[RAFA]). gfp::par-2(RAFA) adult worms are viable but produce 100% dead progeny, indicating that this mutation is maternal effect lethal. We balanced the mutation using the sC1(s2023)(dpy[s2170]umnls41) III balancer strain (CGC51). The sC1(s2023)(dpy[s2170]umnls41) hermaphrodites were first crossed with N2 males. Heterozygote males for the balancer strain were next crossed with the heterozygotes gfp::par-2(RAFA). Heterozygote worms express the myo-2p::mKate tag and look phenotypically wild type; homozygote wild-type worms are dumpies, whereas homozygote mutant worms do not express the myo-2p::mKate tag. The balanced strain was used in Fig. S5, E and F.

Clones from the Ahringer feeding library (Ahringer, 2006; Kamath et al., 2003) were used when available (see Table S5). For the control, in the injection experiments, we used the clone C06A6.2 previously found in the laboratory to have no effect on the early embryonic cell division and to be 100% viable (Bondaz et al., 2019). For GSP-2, a DNA fragment was amplified from cDNA using Gateway-compatible oligonucleotide primers for Gateway-based cloning into the final pDEST-L4440 vector (forward primer, 5′-GGG​GAC​AAG​TTT​GTA​CAA​AAA​AGC​AGG​CT-GTG​ACG​TGC​ACG​GAC​AAT​AC-3′, reverse primer, 3′-GGG​GAC​CAC​TTT​GTA​CAA​GAA​AGC​TGG​GT-CTG​GTG​AGC​TCT​GCA​AAT​C-5′). To produce double-strand RNA (dsRNA) for injections, the Promega Ribomax RNA production system was used. The RNAi constructs for GSP-1 and GSP-2 are specific, as shown in Fig. S4.

dsRNA was injected in L4/young adults, which were incubated at 20°C; embryos from injected hermaphrodites were analyzed after 24–28 h (Fig. 5 D) and 18–20 h (Fig. 2, C–E; and Fig. 4, E and F) after injection.

For the depletion of PKC-3 in the gfp::par-2(RAFA) mutant strain, which is embryonically lethal, L4/young adult worms (homozygote wild-type, heterozygote, or homozygote par-2[RAFA] mutant) were singled on OP50 seeded-NGM plates, let to lay a few eggs, and subsequently injected with pkc-3 dsRNA. Injected worms were transferred to a new OP50-seeded NGM plate. Homozygous mutant gfp::par-2(RAFA) worms were recognized by looking at the progeny in the original plates of noninjected worms, which did not hatch, as homozygote gfp::par-2(RAFA) worms only lay dead eggs. For the codepletion of GSP-1 and GSP-2, a mixture of 1:1 dsRNAs was injected (Fig. 2, C–E).

RNA interference by feeding was performed using plates with 1 mM IPTG for the pkc-3(ne4246), gfp::gsp-1 and mNG::gsp-2 experiments and 3 mM IPTG for the plk-1(or683) experiments. The empty L4440 vector was used as control. In Fig. 1, B–D, Fig. 2 B, and Figs. S1 and S3 (feeding of the pkc-3[ne4246] strains and controls), worms were incubated at 24°C, which is the semi-restrictive temperature as complete loss of PKC-3 results in larval development arrest (Castiglioni et al., 2020). L4 larvae were added to RNAi feeding plates and incubated for 24 h. For the experiment in Fig. 6 (feeding of the plk-1[or683] strain and N2), L1 worms were incubated at 15°C until the adult stage.

Adult hermaphrodites were dissected on a coverslip into a drop of Egg Buffer (118 mM NaCl, 48 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, and 25 mM Hepes, pH 7.5). Embryos were mounted on a 3% agarose pad. Time-lapse recordings (frames captured every 10 s) were performed using a Nikon ECLIPSE Ni-U microscope equipped with a Nikon DS-U3 Digital Camera and a 60×/1.25 NA objective. For Fig. 1 D, a Leica DM6000 microscope, equipped with a DFC 360 FX camera and a 63×/1.4 NA objective, was used. Images of embryos were taken every 10 s. Imaging was performed at 22°C.

For staining of embryos, 20 gravid hermaphrodites were dissected in a drop of M9 (86 mM NaCl, 42 mM Na₂HPO₄, 22 mM KH₂PO₄, and 1 mM MgSO₄) on an epoxy slide square (Thermo Fisher Scientific), previously coated with 0.1% poly-L-lysine. A 22 × 40-mm coverslip was added crosswise on the slide to squash the embryos. The slides were transferred on a metal block on dry ice for at least 10 min. Afterward, the coverslip was removed before fixing for 20 min in methanol. Immunostaining was performed as described in Spilker et al. (2009). The slides were transferred to a solution of PBS plus 0.2% Tween 20 (PBST) and 1% BSA for 20 min. The slides were incubated with primary antibodies diluted in PBST with 1% BSA overnight at 4°C (Rabbit anti PLK-1 [1:500; Tavernier et al., 2015], Rabbit anti-PAR-2 [1:200; Labbé et al., 2006], and mouse anti-tubulin [1:1,000; Sigma-Aldrich]). After two washes of 10 min each in PBST, slides were incubated for 45 min at 37°C with a solution containing secondary antibodies (4 μg/ml Alexa Fluor 488– and/or 568–coupled anti-rabbit or anti-mouse antibodies) and 1 μg/ml DAPI to visualize DNA. Slides were then washed two times for 10 min in PBST before mounting using Mowiol (Calbiochem, 475904; 0.2 M Tris, pH 8.5, and 2.5% 1,4-diazabicyclo[2.2.2]octane).

Images were acquired using a Nikon ECLIPSE Ni-U microscope, equipped with a Nikon DS-U3 Digital Camera, and using a 60×/1.25 NA objective. Imaging was performed at 22°C.

The interaction between PAR-2 and GSP-1/-2 was assessed using a GAL4-based system (Gateway, Invitrogen) using the MAV203 yeast strain. Full-length cDNAs of GSP-1 and GSP-2 were fused to the GAL4 DNA binding domain (Bait plasmid). A PAR-2 (1-335) fragment, both wild-type and mutant (RAFA), was fused to the GAL4 activation domain (Prey plasmid). The PAR-2 wild-type and mutant fragments and the GSP-1 and GSP-2 full length were first cloned into the pDONR201 and subsequently transferred to the pDEST22 vector (GAL4AD) and pDEST32 (GAL4DBD), respectively, using Gateway technology. Mutations were inserted by Pfu site-directed mutagenesis. A list of plasmids and primers used for the Y2H is provided in Tables S6 and S7, respectively. Transformants were selected on synthetic-defined medium (lacking leucine and tryptophan) plates. The interactions were tested by spotting single colonies containing the desired plasmids on a medium lacking leucine, tryptophan, and histidine and containing 50 mM of 3AT (3-amino-1,2,3-triazole; Sigma-Aldrich). Pictures of the plates were taken using the Fusion FX6 EDGE Imaging System (Vilber) equipped with an Evo-6 Scientific Grade CCD camera.

To count the embryonic lethality, young adult worms were singled onto NGM plates seeded with OP50 and incubated 24 h at 24°C (Fig. 1 B and Fig. 2 B) and 20°C (Fig. S3 A and for the gfp::par-2(L165V) mutant). After 24 h, the adult worms were removed and the plates were again incubated at 24 and 20°C, respectively for 24 h. The ratio between the unhatched embryos over the total F1 progeny (unhatched embryos and larvae) was used to calculate the percentage of embryonic lethality.

Embryos obtained by hypochlorite treatment from three medium plates of adult worms were resuspended in SDS sample buffer and denatured for 5 min at 95°C. An equal amount of embryos (∼7,000) were loaded onto a 10% SDS acrylamide gel, and Western blotting was performed according to standard procedures with ECL detection (Vilber, Fusion FX). Primary antibodies (anti-TUBULIN, 1/2,500, mouse [Sigma-Aldrich] and anti-GFP, 1/2,500, rabbit [Pines]) were incubated overnight at 4°C, and HRP-conjugated secondary antibodies appropriate for each primary antibody were incubated for 45 min at room temperature.

α-TUBULIN and α-GFP levels were measured by drawing a region of interest of equal size using Fiji ImageJ, and the mean intensity was used for quantification. The mean intensity of an equal region of interest, placed below the protein of interest, was used for background subtraction.

The line profile of cortical PAR-2 was measured in one-cell stage embryos at pronuclear meeting. A segment 5-pixel-wide and of constant length, centered at the posterior cortex of the embryo and positioned with an angle of 90°C to the cortex, was traced using ImageJ software. For each embryo, the average of three segments (upper, center, and lower posterior cortex) was used for quantification. The segment length was normalized to 1 and the line profile of each embryo was normalized to the average of the value in the cytoplasm at position 0 of the segment traced.

The size of cortical gfp::par-2 domain was determined by measuring the length of the PAR-2 domain normalized for the total perimeter of the embryo at the pronuclear meeting. The perimeter of the embryo and the length of the PAR-2 domain were traced manually by using ImageJ software. The length of the PAR-2 domain is represented as a percentage of the total perimeter of the embryo.

The mean intensity of cortical gfp::par-2 was determined by tracing a line 5 pixels wide along the posterior cortex at pronuclear meeting. The mean intensity of the posterior cytoplasm was obtained by making a square of a fixed area using ImageJ software. For Fig. 4 F, the ratio between the PAR-2 intensity at the cortex over the one in the cytoplasm is plotted, whereas for Fig. S5 B the mean intensity of the cytoplasm was subtracted from the mean intensity of cortical PAR-2.

Cell cycle length was measured from the onset of nuclear envelope breakdown (NEBD) in P0 to the onset of NEBD in AB and P1. NEBD was measured at the time of nuclear membrane disappearance.

AB and P1 length were measures at the time of cytokinesis furrow’s ingression, and then the ratio AB/P1 was calculated.

A value equal to 1 indicates symmetry, whereas a value >1 indicates wild-type asymmetry with AB bigger than P1.

The area of the AB and P1 cells was determined manually by using ImageJ software and the intensity of PLK-1 was measured. For the nucleus intensity, a circle around the nucleus was drawn. The intensity of the nucleus was subtracted from the intensity of PLK-1 of each cell. PLK-1 ratio was determined by dividing the mean of PLK-1 intensity in AB over the mean of PLK-1 intensity in P1. Value equal to 1 indicates symmetric localization of PLK-1 in AB and P1, whereas value >1 indicates wild-type asymmetry with PLK-1 more enriched in AB compared to P1.

The total length of the embryo and the length from the anterior pole to the cleavage furrow were measured manually using ImageJ software. The cleavage furrow position was determined as the ratio of the length from the anterior pole to the cleavage furrow over the total length of the embryo.

The mean intensity of cytoplasmic mNG::gsp-2 and gfp::gsp-1 was determined by tracing a line in the cytoplasm from the anterior (position 0) to the posterior (position 1) of the embryo using ImageJ software. The background was subtracted from each value.

The mean intensity of cytoplasmic mNG::gsp-2 and gfp::gsp-1 was obtained by drawing a fixed square in the cytoplasm using ImageJ software. The mean intensity of the background was subtracted from the mean intensity of the cytoplasm.

The area of the whole embryo was obtained by tracing an ellipsoid using the ImageJ software, and the mean intensity of PAR-2 was measured. The mean intensity of the background was subtracted from the mean intensity of PAR-2.

The PP1 docking motif was identified using the Eukaryotic Linear Motif resource for Functional Sites in Proteins (http://elm.eu.org).

Statistical analysis was performed using GraphPad Prism 9. Details regarding the statistical test, the sample size, the experiment number, and the meaning of error bars are provided for each experiment in the corresponding figure legend, in the results, and summarized in Tables S8 and S9. Data distribution was assumed to be normal, but this was not formally tested.

Significance was defined as: ns, P > 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Fig. S1 describes cell size asymmetry and PLK-1 localization in pkc-3 temperature-sensitive mutant and in GSP-2 depleted embryos. Fig. S2 describes mNG::gsp-2 and gfp::gsp-1 cytoplasmic localization. Fig. S3 describes the phenotype of gsp-1(RNAi) embryos. Fig. S4 shows the specificity and efficiency of gsp-1(RNAi) and gsp-2(RNAi). Fig. S5 compares the phenotypes (lethality, PAR-2 intensity, and domain and cell size) of gfp::par-2° and gfp::par-2(RAFA) embryos and shows the levels of PAR-2 in the gfp::par-2 and the gfp::par-2(RAFA) balanced strains. Videos 1 and 3 show the first division and gfp::par-2 localization of pkc-3(ne4246) and pkc-3(ne4246); gsp-2(RNAi) (Video 1) and pkc-3(ne4246); gsp-1(RNAi) (Video 3) embryos. Video 2 shows the first and second division of pkc-3(ne4246) and pkc-3(ne4246); gsp-2(RNAi) embryos. Video 4 shows the first division and gfp::par-2 localization of gsp-1/-2(RNAi) and control embryos. Video 5 shows the PAR-2 localization in gfp::par-2(RAFA) embryos and control. Video 6 shows the first and second division of gfp::par-2(RAFA) embryos and control. Video 7 shows the PAR-2 localization in gfp::par-2(L165V) embryos and control. Table S1 shows the genotypes of strains used in this study. Tables S2, S3, and S4 show all the reagents used for CRISPR. Table S5 summarizes the clones used for RNA interference. Tables S6 and S7 summarize the plasmids used for the two-hybrid experiment. Tables S8 and S9 summarize the statistical analyses used in this study.

All the strains and reagents generated in this study are available from the corresponding author upon request. All raw data associated with the experiments have been deposited in https://doi.org/10.26037/yareta:wrebkz2i2vc4ddxmpelewshg6m.

We would like to thank N. Goehring (Francis Crick Institute, London, UK) for reagents and discussions, and Dhanya Cheerambathur (Wellcome Centre for Cell Biology—University of Edinburgh, Edinburgh, UK) for reagents. We thank present and past members of the Gotta laboratory for discussions, suggestions, and feedback on the manuscript. A special thanks to Sofia Barbieri for helping with the statistical analysis. We thank Patrick Meraldi, Florian Steiner, and their laboratories for fruitful discussions and comments on the manuscript.

Some strains were provided by the C. elegans Genetic Center, which is funded by the National Institutes of Health office of research infrastructure program (P40OD010440). Work in the laboratory of M. Gotta is funded by the Schweizerischer Nationalfonds (grant number 31003A_175850) and by the University of Geneva.

The authors declare no competing financial interests.

Author contributions: Conceptualization: I. Calvi, M. Gotta. Investigation: I. Calvi, F. Schwager (Fig. 3 B, Fig. 5 C, and Fig. S5 F). Formal analysis: I. Calvi. Methodology: I. Calvi, F. Schwager. Funding acquisition: M. Gotta. Resources: I. Calvi, F. Schwager. Supervision: M. Gotta. Writing, review, and editing: I. Calvi, M. Gotta. Project administration: M. Gotta.