PAR proteins (partitioning defective) are major regulators of cell polarity and asymmetric cell division. One of the par genes, par-1, encodes a Ser/Thr kinase that is conserved from yeast to mammals. In Caenorhabditis elegans, par-1 governs asymmetric cell division by ensuring the polar distribution of cell fate determinants. However the precise mechanisms by which PAR-1 regulates asymmetric cell division in C. elegans remain to be elucidated. We performed a genomewide RNAi screen and identified six genes that specifically suppress the embryonic lethal phenotype associated with mutations in par-1. One of these suppressors is mpk-1, the C. elegans homolog of the conserved mitogen activated protein (MAP) kinase ERK. Loss of function of mpk-1 restored embryonic viability, asynchronous cell divisions, the asymmetric distribution of cell fate specification markers, and the distribution of PAR-1 protein in par-1 mutant embryos, indicating that this genetic interaction is functionally relevant for embryonic development. Furthermore, disrupting the function of other components of the MAPK signaling pathway resulted in suppression of par-1 embryonic lethality. Our data therefore indicates that MAP kinase signaling antagonizes PAR-1 signaling during early C. elegans embryonic polarization.
ASYMMETRIC cell division, a process in which a mother cell divides in two different daughter cells, is a fundamental mechanism to achieve cell diversity during development. We use the early embryo of Caenorhabditis elegans as a model system to study asymmetric cell division. The C. elegans one-cell embryo divides asymmetrically along its anteroposterior axis, generating two cells of different sizes and fates: the larger anterior daughter cell will generate somatic tissues while the smaller posterior daughter cell will generate the germline (Sulston et al. 1983).
A group of proteins called PAR proteins (partitioning defective) is required for asymmetric cell division in C. elegans (Kemphues et al. 1988). Depletion of any of the seven par genes (par-1 to -6 and pkc-3) leads to defects in asymmetric cell division and embryonic lethality (Kemphues et al. 1988; Kirby et al. 1990; Tabuse et al. 1998; Hung and Kemphues 1999; Hao et al. 2006). PAR-3 and PAR-6 are conserved proteins that contain PDZ-domains and form a complex with PKC-3 (Etemad-Moghadam et al. 1995; Izumi et al. 1998; Tabuse et al. 1998; Hung and Kemphues 1999). This complex becomes restricted to the anterior cortex of the embryo in response to spatially defined actomyosin contractions occurring in the embryo upon fertilization (Goldstein and Hird 1996; Munro et al. 2004). The posterior cortex of the embryo that becomes devoid of the anterior PAR proteins is occupied by the RING protein PAR-2 and the Ser/Thr kinase PAR-1 (Guo and Kemphues 1995; Boyd et al. 1996; Cuenca et al. 2003). Once polarized, the anterior and posterior PAR proteins mutually exclude each other from their respective cortices (Etemad-Moghadam et al. 1995; Boyd et al. 1996; Cuenca et al. 2003; Hao et al. 2006). Loss of function of the gene par-1, as opposed to loss of most other par genes, results in embryos that exhibit only subtle effects on the polarized cortical domains occupied by the other PAR proteins (Cuenca et al. 2003). However defects in this gene are associated with a more symmetric division in size, an aberrant distribution of cell fate specification markers, altered cell fates of the daughter cells of the embryo, and ultimately embryonic lethality (Kemphues et al. 1988; Guo and Kemphues 1995).
PAR-1 controls asymmetric cell division and cell fate specification by regulating the localization of the two highly similar CCCH-type zinc-finger proteins MEX-5 and MEX-6 (referred to as MEX-5/6). MEX-5 and MEX-6 are 70% identical in their amino acid sequence and fulfill partially redundant functions in the embryo (Schubert et al. 2000). In wild-type animals, endogenous MEX-5 and GFP fusions of MEX-6 localize primarily to the anterior of the embryo while both proteins are evenly distributed in par-1 mutant embryos (Schubert et al. 2000; Cuenca et al. 2003). This suggests that in wild-type animals, PAR-1 acts in part by restricting MEX-5 and MEX-6 to the anterior of the embryo. The precise mechanism of this regulation is not known, but an elegant study performed for MEX-5 indicates that differential protein mobility in the anterior and posterior cytoplasm of the one-cell embryo contributes to this asymmetry (Tenlen et al. 2008). While increased mobility in the posterior of the one-cell embryo correlates with a par-1- and par-4-dependent phosphorylation on MEX-5, the kinase directly phosphorylating MEX-5 remains to be identified (Tenlen et al. 2008).
Some of the phenotypes associated with loss of par-1 function are dependent on the function of mex-5 and mex-6. First, loss of function of par-1 leads to a decreased stability and aberrant localization of the posterior cell fate specification marker PIE-1, a protein that is usually inherited by the posterior daughter cell in wild-type animals and ensures the correct specification of the germline (Mello et al. 1996; Seydoux et al. 1996). This decreased stability is dependent on mex-5/6 function as PIE-1 levels are restored, albeit with symmetrical distribution, in mex-6(RNAi); mex-5(RNAi); par-1(b274) embryos (Schubert et al. 2000; Cuenca et al. 2003; Derenzo et al. 2003). Second, embryos lacking par-1 function exhibit decreased amounts of P granules in the one-cell embryo, while these markers are present in mex-6(pk440); mex-5(zu199); par-1(RNAi) embryos of comparable age (Cheeks et al. 2004). Third, in par-1(RNAi) one-cell embryos the posterior cortical domain occupied by the polarity protein PAR-2 is extended anteriorly, when compared to wild-type embryos (Cuenca et al. 2003). This anterior extension is rescued in embryos deficient for both par-1 and mex-5/6 (Cuenca et al. 2003). Taken together, these results indicate that par-1 acts in the embryo—at least in part—by regulating the localization and/or activity of the proteins MEX-5 and MEX-6. However, it remains unclear whether other proteins can modulate PAR-1 function to affect MEX-5/6 activity.
To gain insight into the mechanisms of par-1 function in the embryo, we sought to identify genes that act together with par-1 during embryonic development. We performed an RNAi-based screen for genetic interactors of the temperature-sensitive allele par-1(zu310), using the embryonic lethal phenotype of this mutant as a readout. This method has proven successful in previous screens to identify genes involved in early embryonic processes (Labbé et al. 2006; O'Rourke et al. 2007). We were able to identify six genes that, upon disruption of their function, suppress the embryonic lethal phenotype of par-1 mutant embryos. One of these genes is mpk-1, the C. elegans homolog of the highly conserved MAP kinase ERK. Closer analysis subsequently showed that reduction of function of mpk-1 not only increases viability of par-1 mutant embryos, but also reverts several polarity phenotypes associated with loss of function of par-1. Our data indicate that mpk-1 antagonizes par-1 activity to regulate polarization and asymmetric cell divisions in the early embryo.
MATERIALS AND METHODS
Strains were maintained at 15° unless noted otherwise and cultured according to Brenner (1974). N2 (Bristol) was used as a wild-type strain together with the following mutant alleles: LGI, dhc-1(or195); air-2(or207); LGIII, unc-79(e1068); mpk-1(ga111); LGV, sqt-3(sc8); par-1(it90); par-1(zu310). All statistical analyses were carried out using Student's t-test.
Sequencing of par-1(zu310):
To sequence the coding region of par-1(zu310) animals were lysed for 1 hr at 60° in buffer (50 mm KCl; 2.5 mm MgCl2; 10 mm Tris HCl pH 8.3; 0.45% NP40; 0.45% Tween 20; 0.01% gelatin) supplemented with 0.4 μg/μl of proteinase K, followed by inactivation of the enzyme for 15 min at 95°. The exons of the gene par-1 (WormBase 2008) were amplified using Taq polymerase according to standard procedures. Two independent lysates were sequenced for every amplified stretch. The deviant region in par-1(zu310) was amplified from three independent wild-type and par-1(zu310) mutant lysates using the following primers: 5′-GTCGGAAATACGAAACAGCC (forward) and 5′-CCATCAAATGGTAGACTTCCG (reverse).
For time-lapse microscopy, embryos were dissected from gravid hermaphrodites and mounted on a 2% agarose pad in M9 buffer. A PE94 temperature-controlled stage (Linkam Scientific) was used to ensure a constant temperature of 22° during acquisition. Images were acquired with a Hamamatsu Orca ER digital camera mounted on a Leica DM6000B microscope (Leica Microsystems), and the acquisition was controlled by Openlab software (Improvision). Images were acquired at 10-sec intervals using Plan Apochromat 63X/1.4 NA objectives. Image analysis was performed using ImageJ software (National Institutes of Health, NIH).
Genomewide RNAi-based screen:
The RNAi-based screen was performed in liquid medium using a Biomek FX robot (Beckman Coulter, courtesy of M. Peter) following a methodology described previously (Labbé et al. 2006). The embryonic lethality of par-1(zu310) animals was assessed after incubation at 22°. The RNAi screen was performed in duplicate for each clone and yielded 1639 and 1389 candidates, respectively. Of these candidates, 286 were identified in both assays and were kept for further analysis. All 286 bacterial clones were tested by feeding bacterial clones on NGM plates containing 1 mm IPTG to par-1(zu310) animals from L1 stage on. Plates were incubated at 15° for 2.5 days (until L3/L4 stage) and subsequently shifted to 22°. Adults were removed the next day and the progeny were examined for viability the day after removal. Using this method all 286 clones were tested in duplicate by feeding bacterial clones to L1 larvae on agar plates. This assay yielded 60 candidates that were estimated to have more hatching progeny when compared to par-1(zu310) animals fed with vector alone. These 60 candidates were subsequently retested in two independent assays (two plates per assay) and the embryonic lethal phenotype was determined by counting hatched vs. unhatched progeny and compared with control plates (data not shown). All candidates that resulted in a viability that was significantly higher than that of the empty vector control (P < 0.05) were kept for further analysis. These remaining clones were then tested in three more assays (three plates each). This resulted in the confirmation of eight clones that significantly suppressed par-1(zu310) (Table 1), each of which was confirmed in its molecular identity by sequencing. Viability in other strains was assessed using the same approach. However, the assays performed with the alleles dhc-1(or195) and air-2(or207) were carried out at 20° and 18°, respectively.
Generation of double mutants:
To generate the mutant strains unc-79(e1068) mpk-1(ga111); par-1(zu310) and unc-79(e1068); par-1(zu310), males of the allele par-1(zu310) were mated with unc-79(e1068) mpk-1(ga111) or unc-79(e1068) hermaphrodites at permissive temperature. Three animals of the resulting F1 progeny were singled and allowed to lay eggs. Subsequently 30 F2 progeny were singled and allowed to lay eggs at permissive temperature before being transferred to 25°. par-1(zu310) homozygous F2 animals threw dead embryos at this restrictive temperature (escaping larvae developed into sterile adults) as determined under a dissecting scope. unc-79(e1068) mpk-1(ga111); par-1(zu310) or unc-79(e1068); par-1(zu310) homozygous animals were identified among the F3 progeny on the basis of their sluggish, uncoordinated movement. The presence of a homozygous par-1(zu310) allele was reconfirmed for every double mutant background by sequencing (see above). To test the viability of the double mutants, staged L1 larvae of the indicated genotypes were put to 22°. After 2.5–3 days, three to six adults were cloned on new plates and allowed to lay eggs for 12 hr before removal. The viability of the progeny was counted 24 hr after removal of the adults.
To assess embryonic viability at 25°, L3/L4 animals were shifted from 15° to 25° for 24 hr (to circumvent the sterility conferred by the mpk-1ts allele at elevated temperature), transferred to a new plate, and allowed to lay eggs for ∼12 hr before removal of the mothers. Embryonic viability was counted ≥18 hr after removal of the mothers.
For phenotypic analysis, staged L1 animals were shifted to 22° for 2.5 days. Upon reaching adulthood, the animals were sacrificed to examine their progeny. For examination of sterility, progeny were allowed to develop until adulthood and thereafter examined by DIC microscopy for the presence or absence of oocytes and embryos. For the phenotypic analysis, wild-type animals were compared with animals of the following genotypes: par-1(zu310) single mutants, unc-79(e1068) mpk-1(ga111) double mutants, or unc-79(e1068) mpk-1(ga111); par-1(zu310) triple mutants. unc-79 encodes a protein that localizes—according to GFP reporter constructs—to the nervous system and the ventral nerve cord (WormBase 2008). Moreover, unc-79 is required for locomotion of the animals and for sensitivity to specific volatile anesthetics (WormBase 2008). As unc-79(e1068) did not influence embryonic lethality in the par-1(zu310) background (as shown in Table 2), we conclude that unc-79 is not influencing par-1 function.
For RNAi experiments targeting different components of the MAP kinase pathway by feeding, staged animals were allowed to develop until L3/L4 stage for 2.5 days at 15° on standard NGM plates seeded with the bacterial strain OP50. They were subsequently washed several times in M9 to remove bacteria and transferred on 1-mm IPTG plates seeded with HT115(DE3) bacteria transformed with the RNAi constructs. Worms were fed on these plates for 1–1.5 days until progeny were laid and subsequently removed from the plates. Survival of the progeny was scored 24 hr after removal of the parents. Constructs for RNA-mediated interference of components of the MAP kinase pathway were taken from the genomewide collection of feeding constructs (Kamath et al. 2003), unless noted otherwise. For RNAi of mpk-1, the feeding construct targeting both the 45-kDa and the 55-kDa isoform of mpk-1 has been used (Kamath et al. 2003). mek-2, lin-45, and ksr-2 targeting feeding constructs were generated by cloning full-length PCR fragments generated on cDNA libraries (for mek-2 and lin-45) or on cloned cDNA (clone yk343d6, kindly provided by Y. Kohara, encoding for ksr-2a) into a gateway-adapted L4440 vector.
Confocal imaging was performed using a SP2 confocal microscope (Leica Microsystems) equipped with Leica software. Antibodies recognizing PAR-1 (kindly provided by J. Ahringer) and PAR-3 (Labbé et al. 2006) were generated as previously described (Etemad-Moghadam et al. 1995; Guo and Kemphues 1995) and affinity purified from serum on a nitrocellulose strip containing bacterially expressed antigen. The mouse anti-PIE-1 (Tenenhaus et al. 1998) and mouse anti-MEX-5 (Schubert et al. 2000) antibodies were generously shared by J. Priess. Oic1d4, a mouse antibody recognizing P granules, and P4A1, a mouse antibody recognizing PAR-3, were obtained from Developmental Studies Hybridoma Bank (University of Iowa). For rabbit anti-PAR-1, rabbit anti-PAR-3, mouse P4A1, and Oic1D4 staining, samples were snap frozen in liquid nitrogen and subsequently fixed for 20 min in MeOH. For anti-PIE-1 and anti-MEX-5 stainings in contrast, samples were squashed for 20 min in 3% PFA solution in a humid chamber before freeze-crack and MeOH incubation.
After MeOH fixation, slides were washed three times in PBS, supplemented with 0.2% Tween 20 (later on referred to as PBST) before incubation with primary antibodies overnight at 4°. After three washes in PBST, slides were then incubated with secondary antibody solution (including DAPI) for 45 min at 37°. Slides were subsequently washed three times before being mounted in Mowiol (6 g glycerol, 2.4 g Mowiol (Calbiochem 475904), 6 ml H20, 12 ml 0.2 m Tris pH 8.5, 0.1% DABCO). For each staining, a second antibody was used on the same sample to validate antibody accessibility of the imaged embryos. The secondary antibodies used were Alexa488 or Alexa568-coupled anti-rabbit or anti-mouse antibodies (Molecular Probes).
Quantifications of stainings were performed using ImageJ software (NIH). Quantifications of MEX-5 and PIE-1 fluorescence intensities were performed on two-cell embryos before polarization of P1, measuring half-moon shaped areas (sparing the nucleus) either in the dorsal or ventral cytoplasm of AB and P1. Background areas of comparable size were measured outside the embryo and subtracted from the average fluorescence intensity. From these corrected values, ratios of either AB:P1 (MEX-5) or P1:AB (PIE-1) fluorescence intensities were calculated. The amount and distribution of P granules was judged by eye. PAR-1 domain size and PAR-3 domain size in individual stainings were determined after imaging the middle plane of the embryo and measuring the respective length of the domain in the embryo, compared to the total embryo length. The domain borders were determined by eye. PAR-1 and PAR-3 domain size were quantified in separate experiments as both antibodies were produced in rabbit. To visualize the domain overlap, costainings with rabbit anti-PAR-1 and mouse anti-PAR-3 (P4A1) on the same embryos were performed and imaged on the upper cortex of the embryos.
PAR-1 asymmetry at four-cell stage was assessed by measuring PAR-1 average fluorescence intensity along a line on the cortex between EMS/P2 and along the cortex between ABp/P2. Values were background corrected using comparable lines in the cytoplasm of ABp and subtracted from the cortex values. The gradient was determined by calculating the ratio between the two cortices.
Germline phenotypes of the posterior gonad arm were observed by DIC microscopy on F1 progeny that were sedated in M9 buffer, supplemented with 10 mm NaN3. To generate F1 progeny, mother animals were kept at 22° from L1 stage on until egg laying. The resulting progeny were kept at 22° until adulthood (1 day post L4 stage). The representative images shown have been acquired with a 63× objective on a DM6000B Leica microscope (Leica Microsystems, Heerbrugg, Switzerland) microscope using Openlab software (Improvision/Perkin Elmer, Coventry, U.K.) and have been subsequently assembled manually using Photoshop.
Characterization of the allele par-1(zu310):
To identify genes that act together with par-1, we sought to perform an RNAi-based screen for suppressors of the embryonic lethality caused by loss of par-1 function. par-1 gene function is provided maternally, so that par-1 homozygous mutant worms are themselves viable, but produce only dead progeny. Therefore, we chose the temperature-sensitive allele par-1(zu310) (Kao et al. 2004), as the use of conditional alleles of embryonic lethal genes facilitates growth in large scale. Since the embryonic phenotypes of par-1(zu310) have not been characterized, we first performed a phenotypic analysis of this allele. We found that par-1(zu310) mutant animals exhibit phenotypes closely resembling those reported for other, previously characterized par-1 alleles (Kemphues et al. 1988; Guo and Kemphues 1995; Schubert et al. 2000). Embryos produced by par-1(zu310) mutant mothers (hereafter referred to as par-1 embryos) are inviable at 25° (supporting information, Table S2) and nearly inviable at 22° when compared to wild-type embryos (Table 1). The semi-restrictive temperature of 22° was chosen for all subsequent analyses as screening at this temperature would allow us to identify weak par-1 suppressors. Like previously published alleles of par-1, par-1(zu310) mutant embryos exhibit a more symmetric division of the one-cell embryo (Figure 1, A and B; Figure 3A), more synchronous cell divisions at the two-cell stage (Figure 1, C and D; Figure 3B) and aberrant localization of different cell fate specification markers (Figure 1, E and F; Figure 4). Shifting adult mothers and embryos from permissive to restrictive temperature (or vice versa) allowed us to determine that the temperature-sensitive period of the allele par-1(zu310) starts prior to fertilization and does not end before the 28-cell stage (Figure S1). This indicates that the allele par-1(zu310) is a suitable tool to study the function of par-1 during early embryogenesis.
To gain insight into the molecular lesion underlying the par-1(zu310) mutation, we sequenced the coding region of the par-1 gene in this mutant and identified a single nucleotide change in the region encoding the kinase domain. This change of T→A in position 917 (with respect to cDNA prediction for par-1a) causes an isoleucine-to-asparagine change in the amino acid sequence, affecting all four predicted splice variants (Figure 1G) (WormBase 2008). Consistent with previous reports for other alleles of par-1 that carry point mutations in the kinase domain (Guo and Kemphues 1995), the PAR-1 protein is still detectable by Western blot analysis on embryonic extracts of par-1(zu310) animals at both elevated and completely restrictive temperature (Figure S2 and data not shown). These results are consistent with par-1(zu310) having residual activity of PAR-1 at semi-restrictive temperature.
Identification of genetic interactors of par-1:
We performed an RNAi-based screen for genes that suppress the embryonic lethality caused by the temperature-sensitive allele par-1(zu310), testing 93% of the clones in an RNAi library of 16,757 bacterial strains each expressing a dsRNA corresponding to a C. elegans gene (Kamath et al. 2003). Of 15,671 clones analyzed in duplicate, 286 candidates suppressed the embryonic lethal phenotype of the par-1(zu310) mutant animals in both assays (see Table S1).
To verify which of these candidates are reliable suppressors of the embryonic lethal phenotype of par-1(zu310), we tested them in multiple assays by feeding bacterial clones on agar plates to par-1(zu310) mutant L1 larvae (see materials and methods). This led to the identification of eight bacterial clones that can result in a significant (P < 0.05) increase in the viability of par-1(zu310) mutant animals (Table 1).
The suppression assays were performed at a semi-restrictive temperature for par-1(zu310). To examine if the identified clones can suppress a strong loss of par-1 function, we tested whether the suppressors can restore viability in animals mutant for par-1(it90), which is a presumptive null allele of par-1 (Guo and Kemphues 1995). We found that no clone could restore viability of par-1(it90) animals in these conditions (Table 1). These results suggest that the suppressors that we identified cannot bypass the requirement for par-1.
To investigate if the identified suppressors are specific for a par-1 conditional mutant background, we assessed whether targeting these candidates by RNAi also alters the embryonic lethal phenotype of two other temperature-sensitive mutant alleles that affect embryonic development, air-2(or207) and dhc-1(or195) (Severson et al. 2000; Hamill et al. 2002). air-2 encodes the C. elegans homolog of Aurora B kinase and dhc-1 encodes a C. elegans dynein heavy chain, and loss of function of either gene affects early embryonic development without any obvious consequence on polarity and cell fate specification. Of the eight suppressors identified in the screen, we found that RNAi disruption of none of them could suppress the lethality of air-2(or207) mutants, while RNAi disruption of two of them, asd-2 and F23H11.3, partially restored embryonic viability of dhc-1(or195) mutants (Table 1). We therefore conclude that the effect of asd-2 and F23H11.3 on par-1(zu310) embryonic viability might be unspecific. This is consistent with previously published data indicating that asd-2(RNAi) is able to act as a genetic suppressor in a variety of mutants alleles, including dhc-1(or195), lit-1(or131) and spn-4(or191) (O'Rourke et al. 2007).
The genes that specifically suppress the embryonic lethal phenotype of par-1(zu310) mutant, except for K03B4.4, have previously been reported to exhibit embryonic lethality upon their depletion by RNAi in wild-type or RNAi-sensitive mutant backgrounds (Fraser et al. 2000; Kamath et al. 2003; Ceron et al. 2007). This is consistent with the notion that these genes are, like par-1, required for maturation of germ cells or during embryonic development. Under the conditions used for the suppression however, knockdown of the function of these genes leads, if so, only to subtle embryonic lethality in wild-type animals (data not shown).
The suppressors form a heterogeneous group of genes as judged from the information available on their molecular nature (Table 1). For two genes, F36A2.7 and K03B4.4, no clear homolog has been identified yet in other species and no molecular function has been proposed. D2030.4 and C34B2.8 encode subunits of the NADH:ubiquinone oxidoreductase, a component of the mitochondrial electron transport chain, and are therefore key players in cellular respiration. ntl-2 encodes a protein similar to NOT2, a component of the transcriptional regulatory complex CCR4/NOT. Reduction of function of ntl-2 is synthetic lethal with a mutation in ksr-1, a gene in the MAP kinase signaling pathway, indicating that this gene may impinge on MAP kinase signaling (Rocheleau et al. 2008). Finally, mpk-1, the gene encoding the central kinase of the highly conserved MAP kinase pathway (Figure 2A), also suppressed embryonic lethality in a par-1 mutant background.
mpk-1, a suppressor of par-1(zu310) embryonic lethality:
We next focused our efforts on defining the relationships between par-1 and one of the suppressors identified, mpk-1. The MAP kinase pathway is essential for the regulation of a variety of different cellular processes from yeast to mammals (Schlessinger 2000) and aberrant regulation of this pathway has been largely implicated in the development and progression of cancer (Dhillon et al. 2007). In C. elegans, MAP kinase signaling has been best characterized for its role in cell fate specification during the formation of the vulva (Lackner et al. 1994; Wu and Han 1994). Interestingly, abrogating the function of genes that either positively or negatively regulate MAP kinase signaling impairs embryonic development (Page et al. 2001; Hajnal and Berset 2002; Kao et al. 2004; Sonnichsen et al. 2005). Moreover, the gene par-1 was reported to act antagonistically to MAP kinase signaling during vulva development (Kao et al. 2004; Yoder et al. 2004) and a PAR-1-related kinase interacts with the MAP kinase scaffold Ksr in mammalian cells (Muller et al. 2001), suggesting that PAR-1 negatively regulates MAP kinase signaling in these contexts. This is consistent with our results that reduction of mpk-1 function leads to suppression of the embryonic lethality exhibited by par-1(zu310) mutants and suggests that the genetic interaction of par-1 and mpk-1 in embryonic development is a conserved mechanism. We therefore decided to address whether mpk-1 acts together with par-1 during embryonic development in C. elegans using both genetic and cell biological approaches.
As RNA interference can result in off-target effects (Rual et al. 2007), we first confirmed the genetic interaction observed between mpk-1(RNAi) and par-1(zu310) by generating a double mutant strain carrying the allele par-1(zu310) together with the temperature-sensitive loss-of-function allele mpk-1(ga111) (Lackner and Kim 1998). The allele mpk-1(ga111) carries a point mutation near the predicted MEK binding site and the encoded MPK-1 protein was proposed to be less prone to activation than its wild-type counterpart (Lackner and Kim 1998). We found that while only 4 ± 4% of the unc-79(e1068); par-1(zu310) mutant embryos escaped embryonic lethality at semi-restrictive temperature, 38 ± 9% of the unc-79(e1068) mpk-1(ga111); par-1(zu310) mutant embryos were viable (P = 3 × 10−8) (Table 2). Consistent with our RNAi results, we found that the allele mpk-1(ga111) cannot suppress the lethality of par-1(zu310) animals at fully restrictive temperature (Table S2), and phenotypic analysis demonstrated that the par-1-dependent embryonic phenotypes were not suppressed in mpk-1(ga111); par-1(RNAi) animals at this temperature (Figure S3). This again suggests that residual PAR-1 activity is required for mpk-1 to efficiently suppress par-1(lf) mutants. These results indicate that functional interaction between par-1 and mpk-1 is important for early embryonic development of C. elegans.
Several components of the MAP kinase pathway genetically interact with par-1:
Because MPK-1 functions downstream of a conserved signaling pathway, we sought to determine whether disrupting the function of other components in the MAP kinase pathway, besides mpk-1, could increase embryonic viability of par-1(zu310) animals. To test this, par-1(zu310) mutant animals were treated with dsRNA targeting several components of the MAP kinase pathway. Disrupting the function of the MAP kinase pathway core components lin-45, mek-2, and mpk-1 by RNAi resulted in a robust increase in the viability of par-1(zu310) mutant embryos (Figure 2B). We also observed modest but significant (P < 0.05; n = 3) increase in viability when we disrupted the function of lin-3, let-60, and ksr-2 by RNAi (Figure 2B). In contrast, RNAi directed against several other components such as let-23, sem-5, sos-1, or ksr-1 did not alter viability of par-1(zu310) mutant embryos to a significant degree in the same assays. While these genes may not impinge on par-1 function in the early embryo, it is also possible that, in these latter cases, RNAi directed against these genes did not efficiently abrogate their function. We conclude that knockdown of several genes in the MAP kinase pathway suppresses the embryonic lethal phenotype of the allele par-1(zu310), indicating that par-1(zu310) functionally interacts with the MAP kinase signaling pathway during early embryonic development.
Loss of MPK-1 function restores cell division asynchrony in par-1(zu310) animals:
Reduction of function of the gene mpk-1 or other genes of the MAP kinase pathway leads to a significant increase in viability in a par-1(zu310) background. To address whether this is due to a function for mpk-1 in early embryogenesis and asymmetric cell division, we examined whether reduction of function of the gene mpk-1 can restore polarity-dependent processes in par-1(zu310) mutant animals.
During the first division of the C. elegans embryo, the mitotic spindle is displaced toward the posterior of the embryo, leading to the generation of two daughter cells that differ in size. These two cells subsequently divide asynchronously, with the anterior AB cell dividing ∼2 min before the posterior P1 cell. Mutations in par-1 result in defects in both of these processes, and blastomeres at the two-cell stage are of more equal size and divide synchronously (Kemphues et al. 1988). We first determined whether par-1(zu310) animals exhibit similar early embryonic defects as other par-1 mutants when grown at the semi-restrictive temperature of 22°. Quantitative analysis revealed that the anterior cell is significantly smaller in par-1(zu310) embryos (56 ± 2%, n = 22) when compared to wild type (57 ± 1%, n = 24, P = 0.04) (Figure 3A). This phenotype was not suppressed in mpk-1(ga111); par-1(zu310) double mutant embryos (56 ± 2%, n = 28) when compared to par-1(zu310) single mutants (P = 0.96). This suggests either that loss of function of mpk-1 cannot restore the cell size asymmetry in par-1 mutants or, given the expected suppression frequency of 30–40%, that the amount of data acquired was not sufficient to reveal the suppression. However, quantitative analysis of cell division timing at the two-cell stage indicated that mpk-1(ga111) can suppress the synchronous division of par-1(zu310) mutant embryos at semi-restrictive temperature. In wild-type embryos, nuclear envelope breakdown in the two blastomeres occurs 7:27 min (AB) and 9:45 min (P1) after the end of the first cytokinesis while both daughter cells divide more synchronously in par-1(zu310) embryos (8:25 min and 8:28 min), with AB being slightly slower and P1 being faster than in wild type (Figure 3B). Interestingly, this phenotype in P1 was restored in mpk-1(ga111); par-1(zu310) in which the AB nucleus broke down at 8:30 min, while P1 nuclear envelope breakdown occurred only after 12 min, making the division more asynchronous than in wild-type embryos. In this assay, mpk-1(ga111) mutants alone behaved more asynchronous than wild type with division timings of 7:48 min (AB) and 10:48 min (P1), consistent with previously published results, suggesting that reduction of mpk-1 function by RNAi leads to more asynchronous division (Sonnichsen et al. 2005). However, this effect of mpk-1 on cell cycle timing is unlikely to be the primary cause of par-1(zu310) suppression, as we did not find any correlation between cell cycle timing and par-1 suppression in the other suppressors identified (data not shown), and slowing down the cell cycle in P1 by disrupting the function of two unrelated genes (F33H2.5 and div-1, Figure S4A) did not suppress par-1(zu310) phenotypes (Figure S4, B and C; see below). These results indicate that mpk-1 can restore the cell cycle timing asynchrony defect of par-1 and therefore link mpk-1 function to one of the PAR protein-dependent features of asymmetric cell division.
Disrupting mpk-1 function restores cell fate specification in par-1(zu310) mutants:
PAR-1 is required for the segregation of a number of cell fate determinants along the anteroposterior axis of the one-cell embryo by regulating the anterior localization of MEX-5/6 (Schubert et al. 2000; Cuenca et al. 2003; Tenlen et al. 2008). Consistent with this, we find that MEX-5 localizes to both daughter cells in par-1(zu310) embryos at semi-restrictive temperature (Figure 4, A and B). We measured the average ratio of cytoplasmic MEX-5 fluorescence intensity in both blastomeres at the two-cell stage of wild-type embryos and found an enrichment of 2.2 ± 0.5 (n = 35) in the anterior blastomere compared to the posterior one. This ratio was reduced to 1.5 ± 0.3 (n = 35) in par-1(zu310) embryos, a value significantly different from wild type (P = 10−7) (Figure 4M). Consistent with previously published data (Tenlen et al. 2008), MEX-5 distribution in mpk-1(ga111) mutants did not differ significantly from wild-type animals (Figure 4, C and M), even at a fully restrictive temperature for the allele ga111 (Figure S5), but the MEX-5 fluorescence ratio measured in mpk-1; par-1 double mutant embryos (1.8 ± 0.4; n = 35) was significantly higher than in par-1 embryos (P = 0.004) (Figure 4, B, D, and M). However, this fluorescence ratio was still significantly lower than the ratio measured in wild-type embryos (P = 0.001). This indicates that disruption of mpk-1 can partially restore the localization of MEX-5 in par-1(zu310) animals and that MPK-1 acts upstream of a mechanism contributing to MEX-5 localization at this stage.
MEX-5 is required to restrict the germline determinant PIE-1 to the posterior of the one-cell embryo and subsequently to the P1 blastomere at the two-cell stage, leading to a reciprocal pattern of PIE-1 and MEX-5 localization in wild-type embryos (Figures 4, A and E) (Schubert et al. 2000). Consistent with the localization pattern of MEX-5, we find that the posterior enrichment of PIE-1 is not as pronounced in par-1(zu310) embryos (1.8 ± 3, n = 39) as in wild-type embryos (6.1 ± 1.4, n = 39, P = 7 × 10−31) (Figure 4, E, F, and N). We observed that posterior enrichment of PIE-1 was partially restored in mpk-1; par-1 double mutants (2.5 ± 0.9) (n > 38, P = 4 × 10−5) (Figure 4, F, H, and N). Therefore, reduction of mpk-1 can partially restore the asymmetric localization of PIE-1 in par-1 mutant embryos, which is consistent with the restored localization of its upstream localizing factor MEX-5.
We then assessed the localization of P granules, which consist of ribonucleoprotein particles that, as PIE-1, localize to the posterior of the embryo and segregate to germline blastomeres (Seydoux and Strome 1999; Schisa et al. 2001). P granules were proposed to be destabilized in the anterior of the embryo by MEX-5/6 activity (Schubert et al. 2000) and PAR-1 was suggested to mediate their posterior stabilization by restricting MEX-5/6 at the anterior (Cheeks et al. 2004). Because P granule distribution is highly dynamic during the first mitosis, we scored their localization during prophase, at the time of centration of the nucleo-centrosomal complex. We found that par-1(zu310) one-cell embryos are largely devoid of P granules at semi-restrictive temperature, which is consistent with patterns previously reported for other par-1 alleles: we detected P granules in 24% (n = 32) of par-1(zu310) mutant embryos while we observed them in 90% (n = 38) of the wild-type controls (Figure 4, I and J). Interestingly, we found that 69% (n = 34) of mpk-1; par-1 embryos contained P granules, but that these particles were not restricted to the posterior in most (20/24) of the double mutant one-cell embryos (Figure 4L). This indicates that while mpk-1 can restore P granule stability in par-1 mutant one-cell embryos, P granules exhibit an apolar distribution in a mpk-1(ga111); par-1(zu310) mutant background, a phenotype that was not observed in either mpk-1(ga111) or par-1(zu310) single mutant embryos (Figure 4, J, K, and L). This latter phenotype is reminiscent of one-cell embryos lacking the function of par-1; mex-5; mex-6 (Cheeks et al. 2004), which also show ectopically stabilized, yet symmetrically localized P granules. Interestingly, we found that P granules were restricted to one or two cells in later stage mpk-1; par-1 double mutant embryos, indicating that mpk-1 can act to asymmetrically segregate P granules at a later step during embryonic development (data not shown). Taken together, our data show that reduction of function in mpk-1 restores MEX-5 localization as well as several mex-5/6-dependent phenotypes, indicating that mpk-1 could act through a regulation of mex-5 and/or mex-6 in par-1(zu310) mutant embryos (see below).
The localization of P granules and of PIE-1 to posterior cell at one-cell stage is important for the generation of the germline and therefore the fertility of the adult animal. Accordingly, we found that 81% (n = 136) of the par-1(zu310) animals that escaped embryonic lethality at semi-restrictive temperature developed into sterile adults, having either no visible germline (21%) or exhibiting some cellular material in the body cavity that resembled neither oocytes nor germ cells (60%) and may correspond to the somatic gonad (Figure S6). This data is consistent with the agametic phenotype that was previously described for other alleles of par-1 (Guo and Kemphues 1995). As cell fate specification was partially restored in mpk-1(ga111); par-1(zu310) embryos when compared to par-1(zu310), we asked whether fertility was likewise restored. In contrast to par-1(zu310) animals, we found that only 31% of the mpk-1(ga111); par-1(zu310) double mutant animals that escaped embryonic lethality were sterile (n = 132), and most of the animals of this genotype contained both oocytes and embryos in their germline (Figure S6). These data suggest that restoring the asymmetric localization of PIE-1 in early embryos and the enrichment of P granules in germ cell precursors of late mpk-1; par-1 double mutant embryos is physiologically relevant, as it correlates with increased fertility in the progeny of these double mutants when compared to the progeny of par-1 animals.
Disrupting specifically MEX-6 function suppresses par-1(zu310) lethality:
Our finding that mpk-1 restores several mex-5/6-dependent phenotypes in par-1(zu310) embryos prompted us to test whether disrupting mex-5 or mex-6 by RNAi could suppress par-1(zu310) mutants. Because mex-5 and mex-6 are partially redundant and share much sequence identity, we specifically disrupted their function using RNAi by targeting unique regions in the genes, as previously described (Schubert et al. 2000). Interestingly, we found that while disrupting mex-5 did not suppress par-1(zu310) phenotypes, disruption of mex-6 suppressed the embryonic lethality of par-1(zu310) mutant embryos at the semi-restrictive temperature of 22° (Table 1). It is unclear why this effect is specific to mex-6. Since these two genes have partially redundant functions, one possibility is that disrupting mex-6 can partially compromise, yet not abrogate MEX-5/6 activity, and that this remaining activity is sufficient to ensure normal embryogenesis downstream of PAR-1. Alternatively, we cannot exclude that this may uncover a new, specific function for MEX-6 that would be independent of MEX-5. The finding that RNAi disruption of both mpk-1 and mex-6 results in similar effects on par-1(zu310) animals is consistent with these genes functioning in the same direction to regulate par-1-dependent processes.
mpk-1 regulates the size of cortical PAR protein domains in the zygote:
Previous studies have reported that while par-1 is not required for the asymmetric localization of anterior or posterior PAR proteins in the one-cell stage embryo, disrupting its function leads to an expansion of the PAR-2 cortical domain toward the anterior during embryonic polarization (Cuenca et al. 2003). This phenotype depends on the activity of MEX-5/6, suggesting a feedback loop that functions during the polarization of the embryo (Cuenca et al. 2003). We therefore determined the localization of PAR proteins during the first asymmetric divisions of the C. elegans embryo by quantifying the length of cortical PAR protein domains in the middle focal plane of embryos fixed at the anaphase or telophase stages (Figure 5, A and B). In agreement with previous observations on the localization of PAR-2 (Cuenca et al. 2003), we found that PAR-1 extends toward the anterior in par-1(zu310) mutants: the size of the PAR-1 domain was on average 44 ± 7% (n = 22) of total embryo length in par-1(zu310) mutant embryos compared to 32 ± 4% in wild-type embryos (n = 22) (P = 3 × 10−8) (Figure 5, A and B). Disrupting mpk-1 function could partially suppress this phenotype as the size of the PAR-1 domain in mpk-1(ga111); par-1(zu310) embryos was 39 ± 5% (n = 23), which is significantly smaller than in par-1(zu310) embryos (P = 0.01) (Figure 5A). Interestingly, we also found that PAR-3 extends toward the posterior in par-1(zu310) embryos: the size of the PAR-3 domain was 58 ± 5% (n = 22) of total embryo length in par-1(zu310) compared to 52 ± 7% (n = 39) in wild-type animals (P = 0.004) (Figure 5, A and B). However, contrary to what was observed with PAR-1, the localization of PAR-3 at the anterior cortex was not restored in mpk-1(ga111); par-1(zu310) double mutants when compared to par-1(zu310) embryos (57 ± 5%; n = 19; P = 0.9) (Figure 5A). In mpk-1(ga111) mutant embryos, the domain size of PAR-1 (35 ± 8%, n = 13) or PAR-3 (52 ± 7%, n = 19) did not significantly differ from wild type (P > 0.5) (Figure 5A). Our results therefore indicate that the localization of both PAR-1 and PAR-3 is perturbed in par-1(zu310) mutant embryos and that the anterior extension of PAR-1, but not the posterior extension of PAR-3, depends on mpk-1 activity, suggesting that mpk-1 functions upstream of par-1 to regulate the localization of posterior PAR proteins.
Following the first division, PAR proteins are required for the asymmetric division of germline blastomeres and adopt polarized distributions in these cells (Guo and Kemphues 1995). However, PAR polarity might be differentially established in the one-cell embryo compared to later cells of the P lineage, as genes have been reported to be required for PAR polarity in the two-cell stage, but not in one-cell stage embryos (Basham and Rose 1999). We therefore tested whether mpk-1 suppresses par-1 phenotypes at later embryonic stages by determining whether it can restore proper PAR-1 localization in the P2 blastomere of four-cell stage par-1(zu310) mutants. This was assessed by quantifying the fluorescence intensity of PAR-1 along the cortices that P2 shares with its neighboring blastomeres (Figure 5G, top). We found that the levels of PAR-1 in wild-type four-cell stage embryos are 3.1 ± 0.9 (n = 42) higher at the cortex between P2 and EMS than at the cortex between P2 and ABp (Figure 5, C and G). This difference was lost in par-1(zu310) animals, which displayed a ratio of 1.1 ± 0.4 (n = 42; P = 3 × 10−10) (Figure 5, D and G). Interestingly, the asymmetric enrichment of PAR-1 at the P2/EMS boundary was not restored in mpk-1(ga111); par-1(zu310) double mutant embryos, which exhibited a fluorescence ratio of 1.1 ± 0.5 (n = 38; P = 0.9) and mpk-1(ga111) mutants embryos did not differ significantly from wild-type embryos in these assays (2.8 ± 1.8, n = 35, P = 0.5) (Figure 5, E–G). We therefore conclude that while disruption of mpk-1 can restore PAR-1 protein localization in one-cell stage par-1(zu310) embryos, it does not modulate PAR-1 protein distribution in later stage embryos.
We performed a genomewide screen for genetic interactors of a par-1 mutant allele to better define the function of PAR-1 in embryonic polarity. By this approach, we identified six genes that partially but specifically suppress a conditional mutant allele of the gene par-1, suggesting that they might act together with par-1 in fulfilling its functions. According to their molecular nature, the suppressors do not cluster in one prominent pathway, suggesting that they might act via different mechanisms to suppress the par-1 mutant. One of the genetic interactors of par-1 identified here is mpk-1, a gene encoding MAP kinase. We show that par-1 is genetically linked to the MAP kinase signaling pathway and that the interaction between mpk-1 and par-1 impinges on several aspects of asymmetric cell division. Taken together, our results provide a novel link between PAR-1 and MAP kinase signaling in the regulation of asymmetric division of the C. elegans embryo.
To understand how loss of mpk-1 function suppresses par-1, we analyzed several par-1 phenotypes in embryos mutant for both par-1 and mpk-1. Importantly, a precise regulation of MAP kinase signaling is required for proper oocyte maturation and embryonic cell fate specification (Page et al. 2001; Kao et al. 2004). Loss of mpk-1 function in the embryo leads to embryonic lethality and defects in cell cycle timing (Kamath et al. 2003; Simmer et al. 2003; Sonnichsen et al. 2005) indicating that mpk-1 function is required for oogenesis and embryogenesis. How mpk-1 fulfills these functions in detail remains to be elucidated. Our analysis shows that in embryos mutant for both mpk-1 and par-1, several polarity phenotypes of par-1 were partially restored, indicating that mpk-1 antagonizes par-1-dependent signaling in embryonic polarity and thereby links mpk-1 function to polarity. PAR-1 levels were comparable to wild type in mpk-1(ga111); par-1(zu310) embryos, indicating that MPK-1 does not suppress par-1(zu310) by regulating its protein levels (Figure S2). Notably, mpk-1 mutant embryos alone did not exhibit apparent defects in cell fate specification with the exception of cell cycle asynchrony. This suggests that mpk-1 is not the only negative regulator of par-1 activity in polarity.
Our data indicate that mpk-1 impinges on PAR-1 localization in one-cell stage par-1(zu310) mutant embryos and that loss of mpk-1 function cannot bypass the requirement for PAR-1 activity, as asymmetric distribution of MEX-5 was not restored in mpk-1(ga111); par-1(zu310) embryos at fully restrictive temperature (data not shown). Interestingly, loss of function of mpk-1 at semi-restrictive temperature in par-1(zu310) animals not only partially restored the asymmetric distribution of MEX-5, but also suppressed several phenotypes that have been reported to depend on the genes mex-5 and mex-6, such as the distribution of PIE-1 or the precise definition of the cortical domains occupied by the posterior PAR proteins at one-cell stage. Moreover, mpk-1(ga111); par-1(zu310) mutant embryos exhibited ectopic stabilization and apolar distribution of P granules, a phenotype reminiscent of mex-6(pk440); mex-5(zu199); par-1(RNAi) embryos (Cheeks et al. 2004). Taken together, these results support a model in which mpk-1 could act upstream of par-1 to impinge on the distribution and/or activity of MEX-5/6 (Figure 6A). As loss of mpk-1 function does not suppress all of the phenotypes associated with a reduction of PAR-1 function, this model implies that some cellular processes are more sensitive to a reduction in PAR-1 activity. However, our results are also compatible with alternative models in which mpk-1 would act downstream or parallel of par-1 (Figure 6, B and C), which would be consistent with previous reports indicating that MPK-1/Erk activation depends on PAR-1/C-TAK1 activity (Muller et al. 2001; Yoder et al. 2004). As our results show that mpk-1 regulates PAR-1 localization, these models would entail that such regulation occurs indirectly via a negative feedback loop through MEX-5/6, which have been reported to control PAR protein localization (Schubert et al. 2000; Cuenca et al. 2003). Biochemical analysis will be required to resolve the details of this epistatic relationship.
Interestingly, embryos mutant for the E2F and DP transcription factor-related genes efl-1 and dpl-1, which lead to sustained activation of MPK-1 in the C. elegans germline, exhibit a Mex (Muscle excess) phenotype reminiscent of mutations in mex-5 or mex-6 (Page et al. 2001). The morphological defects of both efl-1/dpl-1 and mex-5 mutants can be partially suppressed by reduction of function of components of the MAP kinase pathway (Page et al. 2001). In contrast to our results, this indicates that an increase in MAP kinase activity in efl-1 phenocopies a loss of function of mex-5/6, suggesting that mpk-1 and mex-5/6 act antagonistically during embryonic development. One possible explanation for this apparent discrepancy between this and our data could be that both gain and loss of function of mpk-1 have the same phenotypic outcome. Alternatively, a recent study using gene expression profiling proposed that mutations in mpk-1 and efl-1/dpl-1 may actually control different target genes in the C. elegans germline (Leacock and Reinke 2006), suggesting that mpk-1 and efl-1 act via different mechanisms. Future identification of additional genes that regulate mpk-1- and/or mex-5/6-dependent phenotypes will help to address this. Taken together, this indicates that the interplay between MPK-1 activity and MEX-5/6 function is indeed complex and warrants further analysis.
par-1 was previously shown to function genetically antagonistic to MAP kinase signaling during the development of the C. elegans vulva (Kao et al. 2004; Yoder et al. 2004) and homolog kinases in other species have been reported to target MAP kinase pathway components (Muller et al. 2001; Benton and St Johnston 2003). This antagonistic relationship is consistent with our results that show that reduction of function of mpk-1 can compensate for the loss of function of par-1 during embryonic development and it might therefore indicate that the antagonistic behavior of par-1 and the MAP kinase pathway reflects a conserved mechanism. The core components transmitting MAP kinase signaling are conserved between different cell types and developmental processes in C. elegans. While this work highlights a contribution of MAP kinase signaling during asymmetric cell division, it will be of interest to probe the mechanistic relationships between MAP kinase signaling and other pathways controlling cell polarity.
We thank Ken Kemphues for sharing par-1 mutant strains; Jim Priess and Julie Ahringer for providing anti-MEX-5, anti-PIE-1, and anti-PAR-1 antibodies; Yuji Kohara for sharing the cDNA encoding ksr-2; and to Matthias Peter for generously providing access to the robot unit. We are grateful to Anne Pacquelet for sharing initial mex-5/6 suppression data and for insightful advice, to Alexander Schultze for technical help, as well as to Marc Therrien and Christian Rocheleau for comments on the manuscript. We also thank the laboratories of M. Gotta, J.-C. Labbé, A. Hajnal, M. Hengartner, and M. Peter for stimulating discussions. Some of the strains used in this study were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health. This work was supported by a Boehringer Ingelheim Foundation studentship to A.S., by a Cole Foundation studentship to A.R., by the Canada Research Chair in Cell Division and Differentiation and Canadian Institutes of Health Research grant no. 158715 to J.-C.L., and by funds from Swiss Federal Institute of Technology (ETH) and from the Swiss National Science Foundation to M.G. The Institute of Research in Immunology and Cancer is supported in part by the Canadian Center of Excellence in Commercialization and Research, the Canada Foundation for Innovation, and the Fonds de Recherche en Santé du Québec.
Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.109.106716/DC1.
↵1 Present address: Goodman Cancer Centre, McGill University, Montréal, Quebec H3A 1A3, Canada.
↵2 These authors contributed equally to this work.
Communicating editor: K. Kemphues
- Received June 26, 2009.
- Accepted August 26, 2009.
- Copyright © 2009 by the Genetics Society of America