We identified emb-8 as a Pod locus by screening through published osmosensitive mutants for those that also divide symmetrically at the first cell division. The emb-8 allele, hc69, was initially isolated as a conditional embryonic lethal mutation required maternally for embryo viability 25. Initial characterization of hc69 showed mutant embryos were osmotically sensitive 26, and we verified that hc69 mutant embryos placed in hypotonic medium became swollen and arrested similar to other Pod mutants. No polarity defect was reported in these earlier works.

Analysis of emb-8(hc69) mutant embryos in isotonic medium reveals emb-8 mutant embryos have lost polarity in the one-cell embryo. In the wild-type embryo, the first cleavage produces a characteristic asymmetric two-cell embryo with a larger anterior and a smaller posterior blastomere (Figure 1A). These two cells subsequently divide asynchronously and in orthogonal orientations indicative of their dissimilar identities (only the posterior cell will give rise to germline, for instance). In contrast, half of emb-8(hc69) mutant embryos divide symmetrically, producing two blastomeres of roughly equal size (Figure 1E, Table 1). In embryos where the first division is symmetric, the second cell cycles are synchronous and the two blastomeres divide with roughly parallel orientations, transverse to the longitudinal axis (data not shown). A second emb-8 mutant, t1462, was isolated in a chromosome III-wide genetic screen for embryonic lethal mutations 27. Although emb-8(t1462) mutant embryos have penetrant osmotic defects, they divide asymmetrically (Table 1). Nonetheless, the symmetric first division in emb-8(hc69) mutants coupled with their osmosensitivity (both are also seen in emb-8(RNAi) embryos, see below), places emb-8 in the Pod group of polarity genes. That emb-8(t1462) is fully penetrant for osmosensitivity but has no polarity defect indicates that the two functions are separable, as has been demonstrated for other pod genes to date. Mitotic timing of the first cell cycle in emb-8 mutant embryos is not significantly different from wild-type (data not shown).

To determine if the symmetric division in emb-8(hc69) results from loss of one-cell polarity or from inability to position the first spindle, we examined the distribution of molecular markers of polarity in emb-8(hc69) mutant embryos. In wild-type one-cell embryos, PAR-3 occupies the anterior cortex and PAR-2 adopts a reciprocal location at the posterior cortex (Figure 1B and 1C, 1112). In response to this polarized state, germline granules are segregated to the posterior cytoplasm of the one-cell embryo (Figure 1D, 6). The asymmetric localizations of these markers are disrupted in emb-8(hc69) mutant embryos. Reduction of emb-8 function leads to either more uniform cortical localization of PAR-3 (Figure 1F, 5/13 embryos) or enrichment of PAR-3 along a cortical lateral edge that is sometimes reduced (Figure 1I, 5/13 embryos). The remaining three embryos had wild-type PAR-3 localization. Although PAR-2 is sometimes present at the posterior in emb-8 mutant embryos, it is usually reduced to a small patch at the cortex (Figure 1G, 5/15) or mislocalized to a lateral edge (Figure 1J, 6/15). The remaining four embryos had wild-type PAR-2 localization. In half of emb-8(hc69) mutant embryos, germline granules are mislocalized, remaining more dispersed about the center of the embryo (Figure 1H, 6/22) or moving to a posterior-lateral position (Figure 1K, 5/22). Thus, emb-8 functions in establishing the overall polarity of the one-cell embryo as judged by physical and molecular characteristics.

To gain insight into the role of emb-8 in establishing one-cell polarity, we determined the molecular identity of the emb-8 locus. Previous genetic mapping placed emb-8 in a central region of chromosome III 25. We narrowed the region through three-factor mapping crosses which placed the emb-8 locus between dpy-17 and unc-32 (Figure 2A); 25 of 28 Uncs and none of 20 Dpy F2 hermaphrodites from dpy-17+unc-32/+emb-8(hc69)+ segregated F3 progeny that produced osmosensitive embryos. Furthermore, the rarity with which we found Unc animals that did not carry the emb-8(hc69) allele suggested that emb-8 was located very close to dpy-17.

We used the osmosensitive phenotype of emb-8 mutants to identify a candidate gene near dpy-17. Gönczy, et al. recently documented the phenotypic analysis of individually depleting 2,232 of the 2,315 predicted open reading frames on chromosome III by RNA-interference (RNAi, 28). Of 45 open reading frames in the putative emb-8 region, only four showed embryonic lethal phenotypes. RNAi depletion of one of these, K10D2.6, produced embryos that "lose structural integrity upon dissection" and were placed in a phenotypic category that included the pod-1 locus, an osmosensitive mutant described previously 2128. To determine if the K10D2.6 locus contained molecular variations associated with the emb-8 mutants, we sequenced the hc69 and t1462 alleles and identified missense mutations that result in glycine to glutamate changes in residues 608 and 543, respectively, of the predicted protein (Figure 2A).

To confirm the K10D2.6 open reading frame encoded emb-8, we phenocopied the emb-8 mutant phenotypes by RNA-mediated interference (RNAi) of K10D2.6. Reduction of maternal K10D2.6 function by RNAi made embryos osmosensitive and permeable to their external environment – embryonic cells shriveled in hypertonic medium and absorbed the lipophilic dye, Nile Blue, when depleted of K10D2.6 (Table 1). RNAi of K10D2.6 also reproduced the polarity defects of emb-8(hc69) mutants, causing up to 72% of one-cell embryos to divide symmetrically (Table 1, Figure 2B). Furthermore, RNAi of K10D2.6 reproduced the mislocalization of PAR-3 and PAR-2 seen in emb-8(hc69) mutants (Figure 2C and 2D). Together with the sequencing of alleles, these data indicate that the open reading frame K10D2.6 represents the emb-8 gene and that osmotic sensitivity and symmetric cleavage phenotypes are caused by reduction of EMB-8 function.

K10D2.6 encodes the C. elegans ortholog of the NADPH-cytochrome P450 reductase (NCPR). The K10D2.6 encoded protein shows high sequence similarity to NADPH-cytochrome P450 reductase enzymes from a wide range of species; K10D2.6 is 46% identical and 63–64% similar to NCPR proteins from human, mouse, and Drosophila. This similarity extends along the length of the protein and includes the recognizable FMN-, FAD-, and NADPH-binding domains that are involved in transferring electrons from NADPH to cytochrome P450. The only region lacking significant similarity is the first 60 residues which comprises a hydrophobic membrane anchor in NCPR 2930. Metazoan species typically possess only one form of NCPR and this appears true in nematodes as well (data not shown). Given the high degree of sequence similarity to NCPR enzymes and the recognizable functional domains of the protein, we conclude that emb-8, encodes the C. elegans NCPR ortholog. Studies in other systems have shown that NCPR catalyzes the transfer of electrons from NADPH to cytochrome P450 enzymes. One of the physiological roles of NCPR-dependent enzymes is to catalyze the modification of fatty acids. This role has interesting implications given the pod-2 gene, identified below.

In the course of our RNAi experiments, we found that only feeding mid-L4 staged larvae reliably produced these polarity defects (Figure 2E). dsRNA-feeding of late-L3 to early L4-stages produced less penetrant polarity defects although osmosensitive embryos were obtained (data not shown). To date, we have not been able to demonstrate any essential larval functions for emb-8 since dsRNA feeding from the L1 stage does not prevent growth to adulthood.

The previously characterized Pod-class locus, pod-2, directly implicates fatty acids in the generation of polarity. Using single-nucleotide polymorphisms, we mapped the pod-2 gene to the region around cosmid W09B6 (data not shown). pod-2 was further shown to reside on cosmid W09B6 using transformation rescue. To determine which gene on W09B6 encodes POD-2, we performed RNAi of all open reading frames (ORFs) present on the cosmid. RNAi of a single ORF, W09B6.1, phenocopied the pod-2 mutant loss of asymmetric first cleavage and osmotic sensitivity (Table 1). To confirm that W09B6.1 encodes pod-2, cDNA from the W09B6.1 open reading frame was amplified from three pod-2 alleles and sequenced. For each, a single missense mutation was identified providing important evidence that pod-2 corresponds to W09B6.1 (Figure 3A). W09B6.1 encodes the sole C. elegans homolog of acetyl-CoA carboxylase (ACC) and contains all the domains conserved in ACC from other organisms (Figure 3A).

To further confirm that the previously observed polarity phenotypes in the pod-2 mutants resulted from loss of ACC-function, we analyzed characteristics of anterior-posterior polarity in embryos depleted of W09B6.1/ACC by RNAi. Wild-type adult hermaphrodites fed control bacteria produce embryos with normal polarity that divide asymmetrically at the first cell division (not shown). In contrast, adult nematodes fed bacteria expressing double-stranded RNA (dsRNA) encoding the ACC gene produce embryos that divide symmetrically at the first cell division (Figure 3B, Table 1). Feeding of W09B6.1 dsRNA to pod-2(ye60) animals at 15° can increase the number of symmetric embryos to greater than 90% symmetric cleavage (data not shown). As expected, embryos depleted of POD-2/ACC by RNAi display 100% penetrant osmotic sensitivity (Table 1). RNAi of the other gene with similarity to pod-2 in the C. elegans genome (T28F3.5 that encodes carbamoyl-phosphate synthase) resulted in neither symmetric 2-cell nor osmotically sensitive embryos (data not shown).

POD-2/ACC's role in establishing the anterior-posterior (a-p) axis in the embryo appears to be specific. This conclusion is supported by the following observations. First, embryos depleted for POD-2/ACC are healthy and show no overt phenotypes (e.g., cytokinesis defects, loss of cell integrity) apart from loss of polarity and osmotic sensitivity. When placed pressure-free in osmotically balanced medium, these embryos are essentially indistinguishable from other polarity mutants (e.g., par's) and exhibit first cell cycle mitotic timings statistically equivalent to wild type (data not shown). Second, the cold-sensitive pod-2(ye60) allele is a strict maternal affect lethal, and temperature shift studies with this allele suggest that the embryonic defects can be traced to a POD-2 function ~3 hours prior to fertilization 22. These data suggest the pod-2/ACC gene can be mutated to uncover a specific role in a-p axis polarization. However POD-2/ACC must also have functions in addition to a-p axis formation since first stage larvae fed pod-2/ACC dsRNA grow slowly and often die at the L3 stage (6/9).

The above data suggest that fatty acids, which are synthesized by ACC and potentially modified downstream by EMB-8/NCPR-dependent reactions, play an important role in a-p axis formation. If so, then elimination of the next enzyme in the biosynthetic pathway, fatty acid synthase (FAS; Figure 3C), would also be predicted to lead to loss of a-p polarity. In C. elegans, a single open reading frame, F32H2.5, encodes FAS. Elimination of maternal FAS in embryos by dsRNA feeding results in a loss of a-p polarity in the progeny as judged by symmetric 2-cell embryos (Figure 3D, Table 1). The polarity phenotype became apparent at 28 hours after the initiation of feeding at 15° C (33%; n = 24) and reached a maximum (89%; n = 19) at 64 hours. FAS(RNAi) embryos also displayed completely penetrant osmotic sensitivity.

We also looked at molecular markers of a-p polarity. Out of 33 embryos depleted for FAS, 15 showed PAR-3 protein localization clustered in the cytoplasm (Figure 3E), ten showed lateral, cortical staining (Figure 3G), two had no cortical staining, one had staining all around the cortex, and five had wild-type staining. As for germline granules, 14/33 embryos had staining near the middle of the embryos (Figure 3F), two showed lateral, posterior staining (Figure 3H), one had staining all over the embryo, and 14 were wild-type in localization. Similar PAR-3 and P granule defects are seen in pod-2(ye60)/ACC mutant embryos 22. The loss of PAR-3 staining at the cortex observed in pod-2 mutants is intriguing and may be indicative of significant changes in membrane composition. Nonetheless, as indicated below, the primary polarity defect in pod-2 mutant embryos is caused by events that precede PAR-3 localization.

As loss of ACC or FAS function would be expected to impair fatty acid synthesis, dietary supplementation with exogenous fatty acids might be able complement the pod-2/ACC polarity defects. We found that adding the fatty acids palmitic acid, stearic acid, oleic acid, linoleic acid, arachidonic acid, or docosahexaenoic acid to the diet of pod-2(ye60) mothers rescues the a-p polarity defects in their progeny (Table 2). Rescue by palmitic acid was shown to be dose-dependent. Conversely, neither solvents nor glycerol 3-phosphate (required for making phosphatidic acid) rescue. The polarity defect appears to be specific for fatty acids or the fatty acid moieties of lipids as addition of diacylglycerol, the other major component of lipids, to the diet also did not rescue. Although all fatty acids tested rescued the pod-2(ye60) polarity defects, none rescued osmotic sensitivity, confirming that the function of pod-2/ACC required for osmotic protection (i.e., normal eggshell) is separable from its function in a-p polarity. That the addition of fatty acids rescues only a subset of ACC phenotypes has also been seen in yeast 31.

These data demonstrate that fatty acids play an important role in establishing C. elegans polarity. However, since acetyl-CoA is also used in cholesterol biosynthesis, it was possible that the pod-2/ACC polarity defect might be caused by cholesterol deficiencies. To test this, we used RNAi to deplete the function of hydroxylmethylglutanyl (HMG) CoA reductase, an enzyme that uses acetyl-CoA and catalyzes the first committed step in the cholesterol biosynthesis. HMG CoA reductase in C. elegans is encoded by the open reading frame F08F8.2. Injection of F08F8.2 dsRNA into wild-type hermaphrodites resulted in embryos that were osmotically sensitive (100% penetrance 32 hours after injection) but had normal two-cell asymmetry (Table 1). This result confirms previous demonstrations that osmotic sensitivity itself does not lead to polarity defects 172122. After this time, the mothers became sterile as previously described 32. Thus the pod-2 polarity defect appears to be specific for defects in fatty acid biosynthesis.

The fact that emb-8 and pod-2 have similar phenotypes and are either implicated in (emb-8) or directly involved in (pod-2) fatty acid biosynthesis, respectively, raises the issue as to whether they function in the same pathway. Although for temperature reasons, an emb-8; pod-2 double mutant is difficult to construct and score (one is temperature sensitive and the other is cold sensitive, including by feeding of dsRNA), we have previously shown that the pod-2 mutant functions genetically in the same pathway as a deletion allele of pod-1. We similarly tested whether emb-8 functions in the same pathway as pod-1 by constructing a pod-1(ye11) emb-8(hc69) double mutant and examining the penetrance of the polarity defect. We found no significant enhancement of the loss of a-p polarity in the double mutant as compared to the single mutants when two cell symmetry was scored (29% symmetric, n = 41 for pod-1(ye11); 42% symmetric, n = 24 for emb-8(hc69); and 29% symmetric, n = 31 for the double mutant). This result, together with the similar polarity phenotypes, suggests that pod-1, pod-2, and emb-8 all function in a similar pathway with regards to polarity.

This finding predicts that both emb-8 and pod-2 mutants should manifest similar defects in the events leading to the polarization of the one-cell embryo. We have previously reported that the APC Pod-class of mutants are defective in the interaction of the pronucleus centrosome complex (PNCC) with the posterior cortex and that our observations, as well as important observations from other laboratories, suggest interaction of the PNCC with the cortex was responsible for repelling PAR-3 from the posterior, thereby establishing anterior-posterior asymmetry. This interaction is characterized as movement of the paternal pronucleus (with associated centrosome) towards the posterior (Figure 4A) and subsequent tight juxtaposition with the posterior cortex (Figure 4B,4C). This association lasts for approximately five minutes and coincides temporally with the exclusion of PAR-3 from this region of the cortex 1233, after which the paternal pronucleus dissociates from the cortex to meet the maternal pronucleus in the posterior cytoplasm (Figure 4D).

To determine if emb-8 and pod-2 were required for this process, we examined this polarizing interaction in embryos that had been depleted of emb-8 or pod-2 functions. In embryos depleted of emb-8 function by RNAi, the paternal pronucleus initiated posterior movements (Figure 4E) but close association with the cortex was often abbreviated or prevented entirely (Figure 4F,4G). Lack of interaction was followed by meeting of the pronuclei near the middle of the embryo (Figure 4H) and symmetric division at first cleavage. Loss of pod-2 function similarly impaired this cortical interaction (Figure 4I,4J,4K,4L).

The variability in the interaction of the pronucleus with the cortex in the emb-8 and pod-2 mutants allowed us to investigate whether the quality and duration of the interaction directly correlated with polarity defects. Indeed, in emb-8(RNAi) embryos, strong correlation was observed between impaired pronucleus-cortex interaction and loss of first division asymmetry (Figure 4M). Embryos that did not contact the cortex always divided symmetrically (relative area of the anterior blastomere less than 53%) and those that exhibited significant cortical interactions (longer than 2–3 minutes), with one exception, produced asymmetric divisions (Figure 4M). In emb-8-depleted embryos that subsequently divided symmetrically, posterior-directed cytoplasmic flows were also qualitatively absent (data not shown). Similar correlation between loss of polarity and lack of pronucleus-cortical interaction was also observed in pod-2 mutant embryos (Table 3). Importantly, dietary supplementation of pod-2(ye60) mutants with palmitic acid rescued both the cortical interaction and a-p axis polarization (Table 3). Thus, emb-8 and pod-2 are each required for the cortical interaction by the pronucleus/centrosome complex, and this interaction correlates strongly with the establishment of a-p polarity.

Culture of C. elegans strains and complementation tests were performed using standard techniques 47. Nematodes harboring the temperature sensitive emb-8(hc69) allele were maintained at 15°C and placed at non-permissive conditions (25°C) as L4-staged hermaphrodites overnight to produce mutant phenotypes. Nematodes harboring the cold-sensitive pod-2(ye60) allele were maintained at 25°C and were shifted to non-permissive conditions (15°C) for at least 24 hours to examine mutant embryos. Osmosensitive phenotypes were assayed using Nile Blue uptake and verified by placement of mutant embryos in hypertonic media 17. Alleles used in this study were obtained from the Caenorhabditis Genetic Center (CGC) and from H. Schnabel and included the chromosome III loci: emb-8(hc69), emb-8(t1462), dpy-17(e164), unc-32(e189), him-5(e1490). Genetic mapping of the emb-8 locus was performed by crossing emb-8(hc69) males into dpy-17(e164) unc-32(e189) hermaphrodites, isolation of Unc nonDpy and Dpy nonUnc F2-hermaphrodites, and monitoring the production of osmosensitive embryos by F3-hermaphrodites using NGM plates with Nile Blue 17. Complementation of the pod-2 mutation was done through cosmid rescue carried out by injecting cosmid W09B6 (20 μg/μl) with the dominant marker rol-6 DNA (50 μg/μl) into pod-2(ye60) and pod-2(ye162) mutant animals. Double mutant embryos for the emb-8(hc69) and pod-1(ye11) alleles were generated by first marking emb-8(hc69) chromosomes with unc-32(e189) and then crossing the pod-1(ye11) allele into homozygous emb-8 unc-32 animals at the permissive temperature. Progeny were immediately mated to unc-32(e189) / qC1 males to maintain emb-8(hc69) and pod-1(ye11) in the heterozygous state over the balancer chromosome qC1. Verification of this emb-8(hc69) pod-1(ye11) chromosome was done by test crosses to emb-8(hc69) males and by PCR verification of the pod-1(ye11) deletion allele. Homozygous double mutants were isolated as Unc hermaphrodites and their progeny tested for osmosensitivity and loss of two-cell asymmetry.

Live analysis of embryos was performed by dissecting embryos from gravid hermaphrodites into isotonic medium (EBGM, 48) and placing the embryos on slides with coverslips supported by petroleum jelly to maintain pressure-free conditions. Embryos were observed using 100X DIC optics. Asymmetry of the two-cell embryo was determined by comparison of the cross-sectional areas of the anterior and posterior blastomeres as described previously 21. For determination of pronuclear behavior and timing, recently-fertilized embryos were monitored and scored for meiosis completion (reformation of the maternal pronuclear envelope) and juxtaposition of the paternal pronucleus with the posterior cortex. Immunolocalization of PAR-3, PAR-2, and germline granules (O1C1D4 epitope) were performed as described 1112 except embryos were released into isotonic EBGM buffer. Embryos were co-stained with antibodies to tubulin (DM1a, Sigma) and DAPI to allow unambiguous determination of cell cycle state. Immunofluorescence images were collected with an Olympus IX-70 microscope and deconvolution system (softWoRx, v2.5, Applied Precision, Inc.).

The 5' end of the K10D2.6 mRNA was deduced by amplification and sequencing of K10D2.6 clones from a cDNA library (RIKEN DNA Bank) using a K10D2.6-specific reverse primer and a primer anchored in the flanking vector sequences. It is possible that the 5' end extends beyond this clone, but we note that the predicted gene structure extends the entire length of other NCPR proteins. The W09B6.1 5' end was similarly deduced however two clones were isolated: one matched the predicted transcript end for W09B6.1 and the other contained an additional 333 coding nucleotides derived from genomic regions several kb upstream of the predicted start. Thus pod-2 may have two forms alternatively spliced at the 5' end.

For examination of mutant alleles, mRNA was isolated from a population of homozygous emb-8(hc69) or pod-2 mutant animals by TRIZOL (Gibco BRL)-extraction and reverse-transcription using gene-specific reverse primers and Superscript II reverse-transcriptase (Stratagene). Full length cDNA copies were generated using a 5' end-anchored primer and PCR amplified for sequencing. For sequencing of the emb-8(t1462) allele, genomic DNA was prepared from single Unc hermaphrodites homozygous for the t1462 mutation as described 49 and the full length genomic region amplified by PCR and cloned into pBluescript(SK+) (Stratagene) for sequencing. Our sequencing efforts have confirmed the amino acids and intron/exon boundaries as predicted by the Genome Center for both K10D2.6 and W09B6.1

RNAi was performed by plating L4-staged hermaphrodites onto NG + Ampicillin (100 μg/ml) plates containing 1 mM IPTG and seeded with E. coli HT115 expressing specific open reading frame dsRNA from plasmid L4440 5051. Open reading frames used included: K10D2.6 cDNA (emb-8), W09B6.1 cDNA (pod-2), F32H2.5 cDNA (FAS), T28F3.5 cDNA (CPSase), and F08F8.2 cDNA (HMG CoA Reductase). Embryos were dissected from gravid hermaphrodites in EBGM and visualized by DIC microscopy or processed for immunofluorescence with PAR-2 antibodies. For determination of the optimal feeding duration, synchronized wild-type hermaphrodites were plated as early L4-, mid L4-, late L4-, and young adult stages for varying durations of time at 20°C. Hermaphrodites were individually dissected in 100 μl EBGM and two-cell embryos scored for symmetry. A drop (approximately 25 μl) of 5 M NaCl was added to determine osmosensitivity.

Fatty acids and lipids (except for diacylglycerol; BIOMOL, PA) were purchased from SIGMA and resuspended typically at 100 mg/ml (notable exception is DAG at 10 mg/ml) in the solvent indicated in Table 2. 100 μl of each was spread over a 10 ml NG agar plate. 40 μl of E. coli OP50 were added as food source and ~25 pod-2(ye60) L4 hermaphrodites were plated, grown at 15° for 72 hours, and then dissected to release embryos. In control experiments, 100 μl of either ethanol, chloroform, or water was spread.