Comparative genetic, proteomic and phosphoproteomic analysis of C. elegans embryos with a focus on ham-1/STOX and pig-1/MELK in dopaminergic neuron development

Asymmetric cell divisions are required for cellular diversity and defects can lead to altered daughter cell fates and numbers. In a genetic screen for C. elegans mutants with defects in dopaminergic head neuron specification or differentiation, we isolated a new allele of the transcription factor HAM-1 [HSN (Hermaphrodite-Specific Neurons) Abnormal Migration]. Loss of both HAM-1 and its target, the kinase PIG-1 [PAR-1(I)-like Gene], leads to abnormal dopaminergic head neuron numbers. We identified discrete genetic relationships between ham-1, pig-1 and apoptosis pathway genes in dopaminergic head neurons. We used an unbiased, quantitative mass spectrometry-based proteomics approach to characterise direct and indirect protein targets and pathways that mediate the effects of PIG-1 kinase loss in C. elegans embryos. Proteins showing changes in either abundance, or phosphorylation levels, between wild-type and pig-1 mutant embryos are predominantly connected with processes including cell cycle, asymmetric cell division, apoptosis and actomyosin-regulation. Several of these proteins play important roles in C. elegans development. Our data provide an in-depth characterisation of the C. elegans wild-type embryo proteome and phosphoproteome and can be explored via the Encyclopedia of Proteome Dynamics (EPD) – an open access, searchable online database.


ham-1 mutation causes distinct defects in dopaminergic head neurons.
To find genes that affect dopaminergic neuron development, a mutagenised population of the BY200 strain that allows for visualisation of C. elegans dopaminergic neurons (Pdat-1::gfp) 35 was screened for defective dopaminergic head neurons. One of the isolated mutants, gt1984, exhibited distinct phenotypes in dopaminergic head neurons: while the number of ventral cephalic sensilla (CEPV) neurons was unchanged, the number of dorsal cephalic sensilla (CEPD) neurons was decreased and the number of anterior deirids (ADE) neurons was increased (Fig. 1a,c). Based on the largely invariant cell lineages of C. elegans 36,37 , it can be derived that the respective 'sister' cells of CEPVL and CEPVR neurons die by apoptosis (Fig. 1d − left panels). In contrast, in the lineages of left and right CEPD and ADEL neurons it is an 'aunt' cell that becomes apoptotic (Fig. 1d − middle and right panels).
We mapped 38,39 gt1984 to the ham-1 (HSN Abnormal Migration) locus and found that the mutation comprises a deletion and a splice site mutation (Fig. 1b). Thus, gt1984 is expected to result in a loss-of-function of HAM-1, an orthologue of the Storkhead box protein 1 (STOX1) transcription factor 6, 9 containing a winged helix DNA binding motif (Fig. 1b). We verified that the same dopaminergic head neuron defects occur in a second ham-1 mutant 4 (Fig. 1c). While ham-1 mutants were found to exhibit dopaminergic neuron defects 5,40 , genetic interactions in these lineages have not been described. We conclude that ham-1(gt1984) constitutes a new loss-of-function allele that differentially affects cell fate in dopaminergic head neuron lineages.
Apoptosis in C. elegans development is controlled by the cell death abnormality genes ced-3 and ced-4, which encode an executioner caspase and an Apaf1 (Apoptotic Peptidase-Activating Factor 1)-like molecule, respectively 41 . We tested if dopaminergic head neuron specification is affected by mutation of ced-3 and ced-4 alone. Blinded studies showed that the number of CEPV neurons -which possess an apoptotic sister cell -is increased in the ced-3 and ced-4 mutants; however, the number of CEPD and ADE neurons -which possess an apoptotic aunt cell -is unchanged (Fig. 1e). This observation is in line with studies showing that surviving cells in either the ced-3 or ced-4 mutants can adapt sister cell fates, but do not divide further to produce differentiated cells 5,41,42 . Moreover, we found that mutation of ham-1 in the apoptosis-deficient background did not result in a further increase in CEPV neuron numbers ( Fig. 1e -left panel). Therefore, the prevention of cell death appears to be sufficient to transform the normally apoptotic sisters into additional CEPV neurons.
Cell fate transformations caused by mutation of ham-1 can be masked by cell death [5][6][7] . If this is true for the dopaminergic head neuron lineages, the lower number of CEPD neurons in the ham-1 mutant is expected to be reversed in an apoptosis-deficient mutant background. Indeed, we found that in the ced-4;ham-1 double mutant CEPD neuron numbers are either restored to wild-type levels, or are even higher (Fig. 1e -middle panel). However, the already increased number of ADE neurons in the ham-1 mutant was not altered by additional mutation of ced-4 ( Fig. 1e -right panel). Hence, ham-1 leads to cell fate transformations in both CEPD and ADE neurons, however in CEPD neurons the transformed cells appear to die via apoptosis. pig-1 genetically interacts with ham-1 to specify dopaminergic neurons. If PIG-1 is a general HAM-1 effector, pig-1 and ham-1 mutants are expected to show similar dopaminergic neuron defects. To determine pig-1 mutant phenotypes, we used pig-1(gm344), which likely represents a null allele (Fig. 2a). We found that both pig-1(gm344) single mutants and pig-1;ham-1 double mutants exhibited a wild-type number of CEPV neurons. Thus, CEPV neurons are not affected by loss of ham-1 and pig-1 (Fig. 2b -left panel). Furthermore, pig-1 mutants show a higher number of ADE neurons, similar to ham-1 mutants and pig-1;ham-1 double mutants ( Fig. 2b -right panel). Hence, pig-1 and ham-1 appear to repress ADE cell fate via a similar mechanism. However, in contrast with ham-1 mutants, pig-1 mutants did not display an decreased number of CEPD neurons (Fig. 2b -middle panel and ref. 8) and mutation of pig-1 rescued the reduced number of CEPD neurons in the ham-1 mutant (Fig. 2b -middle panel). Therefore, pig-1 acts as a suppressor of neuronal fate in both CEPD and ADE lineages and is epistatic to ham-1 in CEPD neurons.
We next tested genetic interactions between pig-1 and the ced-4 apoptosis gene in dopaminergic neuron lineages. We found that CEPD neuron numbers were higher in the ced-4;pig-1 double mutant, even though ced-4 and pig-1 single mutants did not exhibit a phenotype ( Fig. 2c -middle panel). This synergism is likely caused by the unmasking of a cell fate conversion in the apoptosis-deficient background, as is the case for ham-1 and ced-4 in CEPD neurons (Fig. 1e). In contrast, the higher number of CEPV neurons in the ced-4 mutants was not further increased by mutation of pig-1 (Fig. 2c -left panel). Also, the higher number of ADE neurons in the pig-1 mutants was not further increased by mutation of ced-4 ( Fig. 2c -right panel). These results show that only one of the three lineages leading to dopaminergic head neurons shows a synergistic genetic relationship between pig-1 and the apoptosis pathway as reported for other C. elegans lineages 8 . We further examined the phenotype of the ced-4;pig-1;ham-1 triple mutant and found that in the ADE neuron lineage -in contrast to the CEPV and CEPD neuron lineages -mutation of ced-4 did not lead to higher dopaminergic neuron numbers than in the ham-1;pig-1 mutant background (Fig. 2d). In summary, we report varying genetic interactions between ham-1, pig-1 and ced-4 in the distinct dopaminergic head neuron lineages.
Changes in protein abundance and signalling pathways in the pig-1 mutant. To identify proteins and pathways that are affected by loss of the PIG-1 kinase, we turned to an unbiased, mass spectrometry-based proteomics approach (Fig. 3a). We aimed to determine PIG-1 targets in 'late' C. elegans embryos, corresponding to the developmental time point at which the last cell divisions of the dopaminergic head neuron lineage take place. Thus, we compared the proteomes and phosphoproteomes of wild-type and pig-1 mutant embryos that were left to develop for 4 hours at 20 °C after isolation from a synchronised population of adult C. elegans hermaphrodites. Since PIG-1, which is an AMP-activated protein kinase-related protein, is likely to form part of a kinase cascade, we anticipate that both direct and indirect PIG-1 targets will be differentially phosphorylated in the pig-1 mutant. In addition, proteins participating in pathways affected by pig-1 mutation might also exhibit altered protein abundances. To facilitate interactive exploration of these large, complex datasets, the resulting protein abundance and protein phosphorylation data from the analysis of wild-type and pig-1 mutant embryos described below, were incorporated into the Encyclopedia of Proteome Dynamics (EPD) database (Fig. 3b).
In the 5% of each peptide fraction that was used to measure protein abundance, we detected 51,414 unique peptides for wild-type and 48,690 unique peptides for the pig-1 mutant, across the three biological replicates (Supplementary Table 1a). Peptides that were detected in at least two out of the three biological replicates were mapped to a total of 6,449 proteins for wild-type and 6,104 proteins for the pig-1 mutant, respectively. Of these proteins, 184 were found to be >2-fold upregulated and 113 > 2-fold downregulated in pig-1 mutant embryos, as compared with wild-type embryos, in at least two out of the three replicates (p-value < 0.05; false discovery rate (FDR) < 5% for both peptide and protein identification) (boxed in Fig. 4a) (Supplementary Table 2).
To investigate whether pig-1 mutation alters the abundance of proteins associated with specific biological processes, we analysed the Gene ontology (GO) term annotations associated with both the down-and upregulated proteins detected (Supplementary Table 5a,b). GO terms associated with downregulated proteins encompass myosin-regulation ('striated muscle myosin thick filament assembly' , 'muscle contraction'), locomotion ('locomotion' , 'inductive cell migration') and metabolism ('protein catabolic process' , 'growth' , 'cellular amino acid biosynthetic process' and 'metabolic process') ( Fig. 4b). An increase in abundance was observed for proteins linked to cell cycle regulation ('DNA replication' , 'cell cycle' , 'cell division' , 'mitotic spindle organisation'), embryo development, apoptosis (Fig. 4c) and microtubule regulation (Supplementary Table 5b). We conclude that pig-1 mutation leads to an overall decrease of proteins related to actomyosin processes, locomotion and metabolism and to an overall increase of proteins connected to the cell cycle, embryo development, apoptosis and microtubules.
We next examined selected proteins within these highlighted biological processes showing altered protein abundances in pig-1 mutant embryos. The selected proteins connected to either cell cycle regulation (Fig. 5a), or microtubule organisation (Fig. 5b), were more abundant in the pig-1 mutant than in wild-type embryos. In    Next, we examined protein abundance changes in pathways previously connected to pig-1 phenotypes. We found that proteins associated with asymmetric cell division, such as the centralspindlin complex-forming ZEN-4 (Zygotic epidermal ENclosure defective) and CYK-4 (CY to Kinesis defect) (reviewed in ref. 44), were more abundant in pig-1 mutant embryos (Fig. 5d). Further examples include the >3.5-fold higher abundances detected for GPR-2 (G Protein Regulator) and LIN-5 (abnormal cell LINeage), the orthologues of Drosophila PINS (Partner of INScuteable) and NuMA (Nuclear Mitotic Apparatus protein), respectively. GPR-2 and LIN-5 mediate the formation of differently sized daughter cells via microtubule-cortex interactions in the C. elegans one-cell embryo (for review see ref. 1). We also observed that Wnt signalling components, such as the Wnt receptor MOM-5 (MOre of MS) and DSH-2 (DiSHevelled related), showed higher abundance in pig-1 mutant embryos (Fig. 5d). Furthermore, as pig-1 genetically interacts with the apoptosis pathway (Fig. 2c, refs 8, 14, 15) and as pig-1 was suggested to promote the digestion of apoptotic cells 14 , we examined proteins related to cell death and corpse engulfment. We found that the apoptosis proteins CED-3 and CED-4 and the corpse engulfment proteins   e e e e e n bi n g in g i i i i g ding g end e e e ding nd g d d ding nd d g g g g e g g g g g g g g g g g g in rth tching g gg ha g at Legend: (click the squares for more information)   CED-2 and CED-12, all show increased abundance in pig-1 mutant embryos (Fig. 5e). We also detected altered abundances of several cell adhesion proteins in pig-1 mutant embryos (Fig. 5f). These data support the reported function of PIG-1 in cell detachment ('shedding') by interfering with the cell-surface expression of cell adhesion proteins in C. elegans caspase-deficient embryos 14 . PIG-1-mediated cell shedding was suggested to involve endocytosis 14 . Consistent with this proposal, we found that the vesicle trafficking protein ARF-1.2 (ADP-Ribosylation Factor-related), which is required for cell shedding 14 , was downregulated in pig-1 mutants, while the abundance of other vesicle trafficking proteins was upregulated (Fig. 5g). Mutation of the PIG-1 kinase leads to phosphorylation changes in key development pathways. We hypothesised that both direct and indirect targets of the PIG-1 kinase may exhibit altered protein phosphorylation levels in pig-1 mutant embryos. To test this hypothesis, we compared phosphopeptide abundances in pig-1 mutant and wild-type embryos. Using a phosphopeptide enrichment step prior to liquid chromatography-mass spectroscopy analysis (Fig. 3a), we detected 7,247 phosphosites in wild-type embryos and 7,913 phosphosites in pig-1 mutant embryos across the three biological replicates (FDR < 5%, localisation probability ≥75%) (Fig. 6) (Supplementary Table 1b). To facilitate the convenient exploration of this complex phosphoproteomics dataset, we created an interactive, searchable version of the data summarised in Fig. 6, which is available online via the EPD (http://www.peptracker.com/epd/). The plot features a custom visualisation tool that allows every phosphorylation site we identified to be selected and traced through a set of cognate metadata.
We provide information about the abundance of each selected protein and its phosphosites, the amino acid position of each phosphosite in the protein, statistical evaluation of biological replicates, as well as linking to the raw mass spectrometry files. For wild-type embryos, these phosphosites correspond to 1,925 phosphorylated proteins (localisation probability ≥75%) (Supplementary Table 1b, Supplementary Table 3). To the best of our knowledge, this is the first dataset describing global protein phosphorylation in C. elegans embryos.
To help evaluate the significance of differences in the phosphoproteomes of pig-1 mutant and wild-type embryos, we aimed to determine the stoichiometry of phosphorylation in relation to protein abundance. For this, we compared pig-1/wild-type phosphopeptide abundances to the corresponding pig-1/wild-type protein abundances (Fig. 7a). This showed that most proteins are located within two standard deviations of the diagonal of the plot, indicating that changes in phosphosite abundance tally with changes in protein abundance. Thus, these proteins are not significantly changing their stoichiometry of phosphorylation. However, proteins located more than two standard deviations from the mean were considered strong candidates for being differentially phosphorylated in the pig-1 mutant, as compared with wild-type, embryos (Supplementary Table 4a). In addition to the differentially phosphorylated proteins, we identified 220 proteins that were exclusively phosphorylated in wild-type embryos and 288 proteins that were only phosphorylated in pig-1 mutant embryos (Fig. 7b) (Supplementary  Table 4b,c). The GO terms associated with proteins harbouring the respective up-and downregulated phosphosites were similar and connected to embryo development, the cell cycle and apoptosis (Fig. 7c,d). We note that this similarity in the biological processes connected with either increased, or decreased, levels of phosphorylation contrasts with our earlier finding that distinct processes were associated with the proteins showing either increased, or decreased abundance (Fig. 4b,c). In summary, we find that pig-1 mutation leads to differential phosphorylation of multiple proteins associated with key developmental processes.
Next, we examined the phosphorylation of selected proteins groups associated with changes in either protein abundance, or protein phosphorylation levels, in pig-1 mutant embryos, as determined by GO term analysis. As seen in the phosphosite GO term analysis (Fig. 7e,f), pig-1 mutation differentially affects phosphorylation of proteins mediating similar processes (Fig. 8). For example, phosphorylation of S80 in the cell division cycle protein CDC-25.2 was only detected in wild-type embryos, while phosphorylation of CDC-5L T377 was only detected in pig-1 mutants (Fig. 8a). Furthermore, phosphorylation of S58 and T530 in the DNA replication protein MCM-10 [yeast MCM (licensing factor) related] is only detected in pig-1 mutant embryos, while phosphorylation of ICP-1 (INCENP (inner centromere protein) homolog) is decreased at least 10-fold at 4 different sites (Fig. 8a) (note logarithmic scale in Fig. 8).
PIG-1 was suggested to affect the expression of cell adhesion proteins 14 and we found that AJM-1 (Apical Junction Molecule) S766 and DLG-1 (Drosophila Discs LarGe homolog) S161 and S471 exhibit increased phosphorylation in pig-1 mutant embryos (Fig. 8f). DLG-1 is a MAGUK (membrane-associated guanylate kinase) protein that mediates a microtubule-dependent pathway important for the segregation of cell fate determinants in Drosophila (reviewed in ref. 1). Furthermore, the inappropriate cell-surface expression of cell-adhesion molecules in pig-1 mutants was proposed to be based on impaired endocytosis 14 , a process that appears to play an important role in asymmetric cell division 1 . We found increased phosphorylation for several endocytosis-associated proteins in pig-1 mutant embryos (Fig. 8g).
Lastly, we found that pig-1 mutation also affects the phosphorylation status of proteins connected to asymmetric cell division. PAR-3 (abnormal embryonic PARtitioning of cytoplasm) for example, the C. elegans orthologue of Drosophila Bazooka, exhibits decreased phosphorylation of S1032 (Fig. 8d). pig-1 mutation also leads to altered phosphorylation of MEX-5 (Muscle EXcess), MEX-6 and PLK-1 (POLO Kinase) (Fig. 8d), proteins that segregate into the anterior daughter cell and help to establish the correct division of the C. elegans one-cell embryo (reviewed refs 1, 45). Furthermore, we found that phosphorylation of DSH-1 S904 is only detected in pig-1 mutants (Fig. 8d). Dishevelled proteins are Wnt pathway components and normally enriched at the posterior part of an asymmetrically dividing cell 3 .
To provide the research community with convenient access to the results from this study, all the proteomics data describing protein and phosphopeptide abundance changes in the pig-1 mutant embryos, as well as phosphopeptide detection in C. elegans wild-type embryos, have been incorporated in the Encyclopaedia of Proteome Dynamics (EPD) − http://www.peptracker.com/epd/. The EPD currently includes the largest collection of annotated, searchable mass spectrometry-based proteomics data from studies in C. elegans, linked with other extensive   hing g n hatch hat hatc chi g g g g ha g g g g g g ha h gg g g gg g g r egg g gg g g gg gg gg gg g g g gg g g or e or eg th g g g g g g g g o  Figure 7. Global phosphoproteome changes in pig-1 mutant embryos. (a) Phosphorylation site abundance ratio (pig-1/wild-type) compared to protein abundance ratio (pig-1/wild-type). The orange dotted line indicates where the phosphorylation site ratio is equal to the protein ratio. Points within the orange and red areas are within one and two standard deviations from equality, respectively. An interactive version of this plot is accessible online via the EPD. In this illustration, detected phosphorylation sites for three example proteins were marked and a tooltip containing further information is shown for one of these phosphorylation sites. (b) Number of phosphorylated proteins detected either in wild-type (red) or in pig-1 mutant embryos (blue) or in both genetic backgrounds (purple overlap)(localisation probability >75%). (c and d) Gene ontology (GO) term enrichment (biological process) of pig-1 variant phosphosites. x-and y-axis indicate semantic space used to group GO terms of related biological processes. Bubble sizes indicate the frequency of the GO term in the underlying C. elegans protein database (larger bubbles reflect more common terms) and bubble colour indicates statistical significance (the greener, the lower the p-value). (c) GO term analysis of downregulated phosphosites in the pig-1 mutant (related to Supplementary Table 5c). Phosphosites downregulated in pig-1 (outside of the 2 standard deviation range in (a)) were pooled with phosphosites that were only detected in wild-type embryos (red in (b)). (d) GO term analysis of phosphosites with upregulated abundance in the pig-1 mutant (Supplementary Table 5d). Phosphosites upregulated in pig-1 mutant embryos (outside of the 2 standard deviation range in (a)) were pooled with the phosphosites that were only detected in pig-1 mutant embryos (blue in b)).

Discussion
We report that the transcription factor HAM-1 and the kinase PIG-1 have a role in the specification of C. elegans dopaminergic head neurons. ham-1 and pig-1 appear to be required for asymmetric cell divisions that result in an apoptotic daughter cell and either a neuron, or a neuronal precursor 5, 6, 8, 9 . In the case of dopaminergic head neurons, the last cell division in the CEPVL/R neuron lineages and the second last division for the CEPDL/R and ADEL/R neuron lineages, are the first ones to directly produce such an apoptotic daughter cell (Fig. 1d, http:// wormweb.org/celllineage). We therefore consider it likely that ham-1 and pig-1 affect the asymmetric division of these neuroblasts. Genetic interactions in the CEPD lineage support that mutation of ham-1 and pig-1 might lead to sister-sister cell fate conversions, thereby turning a normally apoptotic aunt cell into a dopaminergic neuron precursor cell. When blocking cell death by mutation of the apoptosis pathway gene ced-4, these conversions are unmasked and result in higher neuron numbers. In addition, our experiments support that the genetic interaction between ham-1 and pig-1 depends on the exact neuronal lineage analysed. pig-1 was implicated as a transcriptional target and HAM-1 effector in the Q.a neuroblast 9 . Consistent with this, we find an increased number of ADE neurons in ham-1 and pig-1 single and double mutants. In contrast, the reduced number of CEPD neurons observed in ham-1 mutants is supressed by pig-1.
A key feature of this study was the use of a combination of quantitative, mass spectrometry-based proteomic and phosphoproteomic analyses to provide an in-depth, unbiased characterisation of the global changes that occur after genetic disruption of the PIG-1 kinase in C. elegans embryos. Direct targets of the PIG-1 kinase are expected to exhibit lower abundance of phosphorylated peptides in pig-1 mutants, as compared with wild-type embryos. However, by analysing overall protein phosphopeptide and protein abundance changes, we were also able to measure indirect effects of pig-1 inactivation. The extent of differential phosphorylation between wild-type and pig-1 mutant embryos suggests that the majority of detected phosphorylation sites affected may not depend on direct phosphorylation by PIG-1 kinase. We note that this is the first study to analyse the effects of a kinase mutation on the global phosphoproteome in C. elegans.
In addition to the comparison of wild-type and pig-1 mutant embryos described above, this study represents the first global phosphoproteome analysis reported for C. elegans embryos. We identify 5,051 phosphosites, corresponding to 1,639 phosphorylated proteins, in C. elegans wild-type embryos, which are detected in at least two out of the three biological replicates (FDR < 5%, localisation probability >75%) (Supplementary Table 1b). In addition, we detect 288 proteins that are only phosphorylated in pig-1 mutant embryos (Supplementary Table 4c). There are only a limited number of phosphoproteome studies in C. elegans, reporting the identification of ~6,800 46 , ~6,500 47 and ~3,500 48 phosphorylation sites, respectively, in adult animals. However, our dataset is the first to identify phosphosites in the embryo and thus represents a valuable resource for the investigation of C. elegans development. Comparison with the C. elegans dataset from PhosphoPep 48 , a protein phosphorylation database for model organisms, showed that ~50% of the phosphosites we detect in embryos are novel (Supplementary Table 6).
We found that mutation of the PIG-1 kinase preferentially affected proteins associated with the cell cycle and asymmetric cell division, supporting the idea that PIG-1 might regulate the timing of cell cycle events in the neuroblast and thereby influence neuroblast polarity 8 . We note that cell division, cell polarity and differentiation are all tightly coordinated during animal development (reviewed in refs 49, 50). In addition, our results indicate that PIG-1 influences proteins regulating myosin distribution and spindle positioning. These processes are known to influence asymmetric cell divisions (for review see refs 1, 11) and might thus provide clues as to why pig-1 mutants exhibit daughter cells of approximately equal size. Finally, microtubule-associated proteins are more abundant and differentially phosphorylated in pig-1 mutants. During asymmetric cell divisions, microtubules maintain the established cell polarity by mediating interactions between centrosomes and the cell cortex 1,11 . Therefore, the affected candidate proteins offer a promising starting point for further research.
The mammalian PIG-1 homologue, MELK, has also been implicated in cell cycle regulation 21,[27][28][29][30] . For example, MELK was shown to phosphorylate the mitosis-promoting phosphatase CDC25B that promotes transition to mitosis 21 . Since MELK overexpression blocks the transition to mitosis 21,29,51 , CDC25B phosphorylation by MELK is assumed to be inhibitory and mutation of the MELK kinase is therefore expected to stimulate cell division. Moreover, MELK overexpression has been associated with tumour growth and aggressiveness (for review see ref. 20).
In summary, the data presented here provide important new insights into the regulatory processes and signalling events involving PIG-1 that control asymmetric cell division during neuronal differentiation in C. elegans. In addition, this first detailed phosphoproteomic analysis of C. elegans embryos, together with the new interactive data visualisation tools we have developed, provide a valuable resource for developmental biologists studying both C. elegans and other species.

Methods
C. elegans strains and maintenance. TG2435 Bristol, a backcrossed version of BY200 35 , was used as the wild-type strain. TG4200 (ham-1(gt1984) IV;vtIs1 V) was isolated following ethyl methanesulfonate (EMS)mutagenesis and backcrossed six times. Worms were maintained at 20 °C on nematode growth medium (NGM) plates seeded with E. coli strain OP50. Microscopy. For live imaging of neurons, L4 stage and young adult C. elegans hermaphrodites were placed in a drop of 20 mM levamisole on a slide containing an agar pad. A 1.5 mm cover slip was then placed on each slide and animals were imaged with a DeltaVision microscope. Quantification was performed blind with a Zeiss Axio Scope microscope.
Mass spectrometry. Protein extraction and digest. For protein extraction, twelve 9 cm plates containing a synchronised population of C. elegans gravid hermaphrodites (ca. 40,000 animals) were washed to remove already laid eggs. The unlaid eggs were collected by bleaching and let develop for 4 hours at 20 °C before adding 500 μL of lysis buffer (8 M urea in 100 mM triethyl ammonium bicarbonate (TEAB) pH 8.5). The C. elegans embryos were then flash-frozen in liquid nitrogen and stored at −20 °C. The samples were ultrasonicated using a Bioruptor (30 cycles: 30 sec on, 30 sec off), reduced using tris(2-carboxyethyl)phosphine (TCEP) (25 mM) for 30 minutes at room temperature, then alkylated in the dark for 30 minutes using 2-iodoacetamide (50 mM). Total protein was quantified using the EZQ assay (Life Technologies). The lysates were diluted with 100 mM TEAB 4-fold for the first digestion with endoprotease Lys-C (Wako), then further diluted 2.5-fold before a second digestion with trypsin. Lys-C and trypsin were used at an enzyme to substrate ratio of 1:50 (w/w). The digests were carried out overnight at 37 °C, then stopped by acidification with trifluoroacetic acid (TFA) to a final concentration of 1% (v/v).
Peptide desalting and solid phase extraction. Prior to fractionation, the protein digests were desalted using C18 Sep-Pak Cartridges (Waters). Cartridges were first activated with Acetonitrile (ACN) and equilibrated with 50% ACN in water according to the manufacturer's protocol. The samples were loaded and washed 4 times with 500 μL water containing 0.1% TFA. The peptides were eluted into a fresh Eppendorf tube with 800 μL 50% ACN and then dried in vacuo.
Phosphopeptide enrichment. Phosphopeptide enrichment was carried out on the HILIC fractions with titanium IMAC (immobilised metal affinity chromatography) beads (MagReSyn) using the original protocol by Larsen et al. 43 on an automated bead processing robot (KingFisher, Thermo Fisher Scientific) as described in Tape et al. 55 . The enriched phosphopeptide fractions were dried, redissolved in 5% formic acid and analysed in data-dependent mode on Q Exactive Plus Orbitrap mass spectrometer (Thermo Fisher Scientific).

Liquid chromatography electrospray tandem mass spectrometry analysis (LC-ES-MS/MS
). 5% of the material was analysed using a Q Exactive Plus Orbitrap mass spectrometer (Thermo Fisher Scientific) equipped with a Dionex ultra high-pressure liquid chromatography system (nano RSLC) without enrichment in order to quantify changes in protein expression levels. RP-LC was performed using a Dionex RSLC nano HPLC (Thermo Scientific). Peptides were injected onto a 75 μm × 2 cm PepMap-C18 pre-column and resolved on a 75 μM × 50 cm reverse phase-C18 EASY-Spray temperature-controlled integrated column-emitter (Thermo Fisher Scientific) using a four hour multistep gradient from 5% B to 35% B with a constant flow of 200 nL/min. The mobile phases were: 2% ACN incorporating 0.1% FA (Solvent A) and 80% ACN incorporating 0.1% FA (Solvent B). The spray was initiated by applying 2.5 kV to the EASY-Spray emitter and the data were acquired under the control of Xcalibur software in a data-dependent mode.
95% of the material was subject to phosphopeptide enrichment and analysed on a Q Exactive Plus using a two hour gradient.
Database searching and protein quantification. The raw mass spectrometry data were processed using the MaxQuant software suite (version 1.5.2.8). Proteins and peptides were identified using the UniProt C. elegans reference proteome database (Swiss-Prot and Trembl separately), using the Andromeda search engine 56, 57 with the following search parameters: carbamidomethylation of cysteines was defined as a fixed modification, while methionine oxidation, acetylation of N-termini of proteins, conversion of glutamine to pyro-glutamate and phosphorylation on STY were set as variable modifications. The false discovery rate was set to 5% for proteins and peptides.
Mass spectrometry data analysis. The volcano plot depicting protein abundances was generated based on the label-free quantification intensity of proteins present in at least two of the three biological replicates in wild-type and mutant embryos. Statistical significance was determined with a two-tailed Student's t-test with a randomisation setting fixed at 250 on the Perseus software platform 58 . For phosphorylation site analysis, data analysis was performed in Python using the scipy statistical package. Only phosphosites with a localisation site probability higher than 75% and a score higher than 40 were considered. Elements present in both wild-type and pig-1 mutant embryos were added to the phosphosite comparison analysis, while elements present in only one of the genotypes were separated into two genotype-specific output files and visualisations. In the comparison analysis, the mutant over wild-type phosphosite abundance and protein abundance ratios were calculated and converted into log 2 format for visualisation purposes. In the genotype-specific analysis, a log 10 transformation was applied to the protein intensity, phosphorylation site intensity and the position of the post-translational modification within the protein to render the visualisation more user-friendly. The visualisation is presented in D3.js and the data is stored in Cassandra.
Data depository. Data generated in this study is accessible via the Encyclopaedia of Proteome Dynamics (EPD) 34 (http://www.peptracker.com/epd/) for convenient exploring and display. In addition, the raw mass spectrometry files as well as the MaxQuant output have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE 59 partner repository with the dataset identifier PXD005752.
Gene ontology term analysis. Gene ontology (GO)-term enrichment analysis was performed with the DAVID functional annotation tool (https://david.ncifcrf.gov/) 60,61 . The complete list of C. elegans proteins detected in mass spectrometry-based proteomics was used as background and the GO term subcategories 'GOTERM_BP_ DIRECT' 'GOTERM_CC_DIRECT' and 'GOTERM_MF_DIRECT' selected for analysis. The results were analysed with REVIGO 62 and plotted using the provided R script to remove GO term redundancy. In REVIGO, p-values were provided and the C. elegans database and the SimRel semantic similarity measure were used for analysis. A 'medium' REVIGO list size was selected.