BIOLOGICAL OVERVIEW

Following completion of meiosis, DNA replication must be repressed until fertilization. In Drosophila, this replication block requires the products of the pan gu (png), plutonium (plu) and giant nuclei (gnu) genes. (For information on the Pan Gu legend, see Pan Gu Creates the World). These genes also ensure that S phase oscillates with mitosis in the early division cycles of the embryo. png encodes a Ser/Thr protein kinase expressed only in ovaries and early embryos; the predicted extent of kinase activity in png mutants inversely correlates with the severity of the mutant phenotypes. The Plu and Png proteins form a complex that has Png-dependent kinase activity, and this activity is necessary for normal levels of mitotic cyclins. Cyclin B is a key Png target. Mutations in cyclin B dominantly enhance png, whereas png is suppressed by cyclin B overexpression. Suppression occurs via restoration of Cyclin B protein levels that are decreased in png mutants. plu and gnu phenotypes are also suppressed by cyclin B overexpression. These studies demonstrate that a crucial function of PNG in controlling the cell cycle is to permit the accumulation of adequate levels of Cyclin B protein. These results reveal a novel protein kinase complex that controls S phase at the onset of development apparently by stabilizing mitotic cyclins (Fenger, 2000 and Lee, 2001).

In all species, fertilization of the egg by the sperm triggers two crucial cell cycle changes: an arrest point is relieved and the cell cycle restarts with DNA replication. The cell cycle stage at which the mature egg arrests while awaiting fertilization varies among different species: it can be the G2 phase preceding meiosis, metaphase I or metaphase II of meiosis, or arrest after the completion of meiosis. In Drosophila, the egg is activated to complete meiosis as it passes through the uterus, and fertilization is required for the onset of DNA replication and the cell cycle. Despite the range of possible arrest points, in all organisms fertilization must lead to re-entry into S phase and mitotic division (Fenger, 2000).

The understanding of the molecular changes triggered by fertilization and the potential cell cycle targets of fertilization is still rudimentary. In a number of marine invertebrates and vertebrates, fertilization causes a flux of Ca²⁺ in the egg. This appears to result from activation of phospholipase Cgamma in the egg, leading to an increase in PI3 levels and the cytoplasmic release of intracellular Ca²⁺ pools. It is not known whether such a Ca²⁺ flux also occurs at fertilization in Drosophila or whether phospholipase Cgamma is activated in response to fertilization. It is also not known which cell cycle regulators are affected by fertilization. The control of meiotic arrest and its release by fertilization are best understood in Xenopus, in which cytostatic factor (CSF), an activity dependent on the Mos kinase, maintains the egg in metaphase II by stabilizing high maturation promoting factor (MPF: Cdc2/cyclin B) activity. Fertilization inactivates CSF, although the molecular details of how Ca²⁺ causes this are not known, MPF activity drops and meiosis is completed (Fenger, 2000 and references therein).

Other organisms must also possess regulators that repress cell cycle progression prior to fertilization. As suggested by the fact that Mos has not been found in invertebrates, these regulators may not be universal among organisms. In addition to regulators that may repress the cell cycle only in the mature egg, some organisms may require specific cell cycle regulators during the early embryonic divisions. In organisms with rapid embryonic development such as marine invertebrates, amphibians and insects, the early embryonic divisions occur via a simplified cell cycle in which S phase alternates with mitosis. In these embryonic cycles, growth and transcription do not occur, and the S-M oscillations are driven by maternal components stockpiled into the egg during oogenesis. The post-transcriptional control of these early cell cycles might be through unique regulators of DNA replication and nuclear division. More likely, unique molecules might impose altered control of conserved cell cycle regulators that also function during the archetypal G1-S-G2-M cell cycle (Fenger, 2000).

During the first seven cell cycles in Drosophila, the mitotic cyclins A and B, and CDC2 activity are present at high levels with no detectable fluctuation in overall levels during the cell cycle. Localized degradation of Cyclin B has been detected, though, and it is thought that localization of cyclins A and B may be important in coordinating the nuclear cycles with the microtubules. Furthermore, inhibition of the cyclin degradation machinery with a destruction box peptide or injection of non-destructible Cyclin B causes mitotic arrest in the early embryo. All these results suggest that cyclins play a crucial role in coordinating S phase and mitosis in the early embryo. How the cyclins are controlled is unclear (Fenger, 2000).

In Drosophila there are three maternal effect genes, plutonium, pan gu and giant nuclei, that are needed both to inhibit DNA replication in the unfertilized egg and to control the S-M cycles of embryogenesis. Females mutant for any one of these genes produce eggs that complete meiosis but do not arrest, in contrast to unfertilized eggs from wild-type females. Instead, all four meiotic products inappropriately undergo DNA replication to become polyploid, often fusing into a single nucleus as they increase in ploidy. If the eggs from plu, png or gnu mutant mothers are fertilized, they undergo defective S-M cycles: DNA replication takes place, but nuclear division does not occur, resulting in giant polyploid nuclei similar to the mutant unfertilized eggs. Intriguingly, the centrosomes dissociate from the nuclei and continue to replicate and divide in mutant embryos, indicating that nuclear regulation has been uncoupled from the centrosome cycle (Fenger, 2000).

There appears to be an active requirement for the function of these genes during the S-M cycles, because weak alleles of png permit transient S-M cycling before nuclear division ceases (Shamanski, 1991). These weak mutations yield embryos with up to 16 giant polyploid nuclei. The three genes control the same biological process, because mutations in plu or gnu dominantly enhance the phenotype of weak png mutations, eliminating the transient S-M cycling (Shamanski, 1991). Thus the phenotypic analysis of mutations in plu, png and gnu indicates that these genes are necessary to inhibit S phase prior to fertilization, and they are needed to coordinate S and M during early embryogenesis (Fenger, 2000).

The plu gene has been shown to encode a 19 kDa ankyrin-repeat protein (Axton, 1994). png encodes a Ser/Thr protein kinase that exists in vivo in a complex with Plu. This protein kinase complex is present specifically in the egg and early embryo, and thus it appears to control S phase uniquely in response to fertilization at the onset of development. Analysis of cyclins A and B in png mutants shows that Png affects the levels and post-translational modification of mitotic cyclins, as well as the extent of histone H3 phosphorylation and CDC2 kinase activity. Therefore Png controls chromosome condensation and mitotic progression in the early embryo (Fenger, 2000).

The phenotypes of plu, png and gnu mutants suggest that these genes normally function to inhibit initiation of DNA replication during early embryogenesis. Activating mitosis is one means by which this could be accomplished. If mitotic functions were not activated in the mutant embryos then repeated rounds of DNA replication could occur. This phenomenon would be similar to the repeated rounds of DNA replication that result from loss-of-function of cdc2 and cyclin B in S. pombe. Mitotic cyclin/Cdk activity is needed to inactivate replication origins, and chromosome condensation in mitosis may also serve to block DNA replication. To test if png, plu and gnu affect regulation of mitosis, the protein levels and forms of CDC2 and the mitotic cyclins A and B were examined in mutant embryos. Interestingly, the levels of mitotic cyclins are decreased in mutant embryonic extracts, and the effect is allele specific with respect to png. Three forms of Cyclin A were detected on immunoblots, and the fastest migrating form was predominantly decreased in the mutants: barely detectable levels of this fast migrating form were present in weak png mutant embryos, and none was detected in strong png, plu or gnu mutants. The slower two forms of Cyclin A were less affected, although they also showed allele-specific reduction: the greatest decrease was seen in png¹⁰⁵⁸, png¹⁷², png¹⁹²⁰ , plu and gnu, and the highest levels of Cyclin A were seen in the three weak png alleles. This experiment was repeated twice, and the same decreases in Cyclin A levels were observed. It was also found that levels of Cyclin A were reduced in unfertilized eggs from png¹⁰⁵⁸ mutant mothers (Fenger, 2000).

To determine if the missing fast-migrating Cyclin A form was phosphorylated, and thus a potential Png substrate, wild-type embryonic extracts were treated with lambda-phosphatase. Compared with the 'no phosphatase' control and to wild-type extracts homogenized directly in urea sample buffer, phosphatase-treated extracts showed a loss of the slowest migrating form and an accumulation of the fastest migrating form, indicating that the fast form was actually unphosphorylated. If Cyclin A were a substrate of Png, the phosphorylated form of Cyclin A would be the most decreased of the forms. Since the unphosphorylated form appeared to be preferentially lost in png mutants, it is unlikely that Cyclin A is a Png substrate (Fenger, 2000).

An allele-dependent decrease in Cyclin B levels of mutant embryos was also detected. Again, all of the alleles of png, gnu and the plu null mutation showed decreased cyclin levels compared with wild-type embryos, and of the mutants, the highest levels were seen in the three weak png alleles. The plu null and gnu mutants had the lowest levels, and of the png alleles, the lowest levels were seen in png¹⁷², png¹⁹²⁰, png¹⁵⁸ and png¹⁰⁵⁸. It is interesting that png¹⁷² and png¹⁹²⁰ (which were both predicted to have the lowest kinase activity based on the nature of the mutations) also have the lowest levels of cyclins A and B. Cyclin B levels also were decreased in unfertilized eggs from png¹⁰⁵⁸ mutant mothers. Consistent with the decrease in levels of mitotic cyclins, it was found that CDC2 kinase activity was also decreased (Fenger, 2000).

Examination of the levels and forms of CDC2 showed no difference between wild-type and mutant embryos. It has been observed that Histone H3 phosphorylation correlates with CDC2 kinase activity in early Drosophila embryos, so anti-phospho-H3 antibodies were used to examine levels of H3 phosphorylation in mutant embryos. Phospho-H3 levels were decreased in all the mutants, and again the degree of decrease correlated with allele strength. The lowest levels of phospho-H3 were seen in plu, gnu, png¹⁷², png¹⁹²⁰ and png¹⁰⁵⁸. The highest levels were observed in the three weak png alleles and png¹⁵⁸. MPM2 antibodies also recognize phospho-epitopes present in mitotic cells and dependent on CDC2 activity. A decrease in at least one MPM2 epitope was observed in png¹⁰⁵⁸ and plu⁶ embryonic extracts compared with wild. To assay CDC2 kinase levels directly, CDC2 was immunoprecipitated and levels of histone H1 kinase activity were tested in the pellets. The proteins in the pellet were immunoblotted and the levels of phosphorylated histone H1 were compared with the amount of CDC2 immunoprecipitated, as measured by probing the membrane with antibodies against CDC2. In the mutants the levels of CDC2 kinase activity were decreased about twofold. In conclusion, these results show that mitotic cyclins and CDC2 activity are decreased in png, plu and gnu mutants, and the decrease is allele specific and consistent with predicted levels of Png kinase activity (Fenger, 2000).

There may be a requirement for an S phase inhibitor in the mature egg because the cytoplasm is stockpiled with maternal replication components: the meiotic products may be especially susceptible to initiation of DNA replication when the chromosomes decondense and go through an interphase-like state after completing meiosis. It is also possible that a homolog or functional analog is expressed zygotically, and therefore expression of Png is not required during later development. The phenotype of png mutants suggests that the Png kinase pathway must be regulated temporally and spatially: temporally, in order to inhibit replication during M phase but not S phase, and spatially, to inhibit S phase in the polar bodies, but not the zygotic nuclei. This regulation may be at the level of Png localization and activity, although it has not been possible to localize Plu and Png within the embryo by antibody staining. Alternatively, Plu and Png activity may be upstream of spatially and temporally controlled cell cycle regulators (Fenger, 2000).

As a new protein kinase, orthologs of Png have not yet been identified in other species. It seems likely that other organisms in which the embryonic divisions proceed via a rapid S-M cycle may contain orthologs of Png that control these early cycles. It is also possible that early embryonic cell cycle regulators are more divergent than other cell cycle regulators, because the coordination of meiosis, fertilization, and the first mitosis is divergent in different species. Although not a sequence homolog, in many ways the Png kinase parallels Mos, a Ser/Thr kinase unique to vertebrates. Similar to Png, Mos inhibits DNA replication until fertilization by promoting MPF activity, perhaps primarily by stabilizing mitotic cyclins. In contrast to Png, however, Mos is also required for entrance into meiosis and inhibition of DNA replication between the meiotic divisions, and it is degraded during the first mitosis (Fenger, 2000 and references therein).

The Plu protein controls the same cell cycle processes as Png, and biochemical observations support the idea that the two proteins act together. Plu and Png are present in a complex in mature eggs and embryos. Furthermore, functional Png is needed for it to be complexed with Plu. This may be because Png and Plu interact through an adapter protein that is a target of Png. Png and Plu did not interact in the yeast two-hybrid system, supporting this hypothesis. The levels of Plu protein are depressed in strong png mutants (Elfring, 1997), and this appears to result from a modified, unstable form of Plu present in png mutants. The functional relationship between Plu and Png remains to be determined. One possibility is that Plu is a downstream effector of Png activity, but it is also possible that Plu activates the Png kinase (Fenger, 2000).

The primary defect in png, plu and gnu mutants is likely caused by a decrease in the levels of mitotic cyclins A and B and associated CDK activity. The over-replication phenotype might result because mitotic cyclins are required to block re-replication in the unfertilized egg and early embryo, because the embryo requires chromosome condensation to block re-replication, or a combination of both. Work in Schizosaccharomyces pombe has shown that Cyclin B can function to inhibit DNA replication, since strains deleted for cdc13, which encodes Cyclin B, undergo repeated rounds of S phase without mitoses. Moreover, it has been shown that CDC2 and Cyclin A inhibit endoreplication in diploid cells during Drosophila larval development. The replication of centrosomes that are unattached to nuclei, seen in these mutants, is observed also in embryos lacking Cyclin B, so this mutant cytoskeletal defect may be due to decreased Cyclin B levels (Fenger, 2000).

It is unlikely that Png interacts directly with cyclins because Png and Plu do not co-immunoprecipitate with Cyclin A or Cyclin B, and the unphosphorylated form of Cyclin A appears to be predominantly lost in png mutants. The differential decrease of the different forms of Cyclin A suggests that it is unstable in the mutants, particularly the faster migrating form, and that the decrease is not due to decreased cyclin translation. One possibility is that Png controls cyclin stability by acting through the protein degradation machinery of the APC/cyclosome. It is known that yeast Cdh1, an activator of the APC and subsequent Clb degradation, is inactivated by phosphorylation, so Fizzy-related (Fzr), the Drosophila Cdh1 homolog, could be a substrate of Png. The identification of the Plu-Png protein kinase complex opens the way to identify its regulators and substrate targets in controlling the S-M cell cycles. In a survey of the genome to identify genes that genetically interact to suppress or enhance png mutations, several intervals and loci have been identified. Recovery of Png interactors will permit the unraveling of the mechanism by which Png stabilizes mitotic cyclins and inhibits S phase at this time in development (Fenger, 2000).

Control of PNG kinase, a key regulator of mRNA translation, is coupled to meiosis completion at egg activation

The oocyte-to-embryo transition involves extensive changes in mRNA translation, regulated in Drosophila by the PNG kinase complex whose activity is shown in this study to be under precise developmental control. Despite presence of the catalytic PNG subunit and the PLU and GNU activating subunits in the mature oocyte, GNU is phosphorylated at Cyclin B/CDK1 sites and unable to bind PNG and PLU. In vitro phosphorylation of GNU by CyclinB/CDK1 blocks activation of PNG. Meiotic completion promotes GNU dephosphorylation and PNG kinase activation to regulate translation. The critical regulatory effect of phosphorylation is shown by replacement in the oocyte with a phosphorylation-resistant form of GNU, which promotes PNG-GNU complex formation, elevation of Cyclin B, and meiotic defects consistent with premature PNG activation. After PNG activation GNU is destabilized, thus inactivating PNG. This short-lived burst in kinase activity links development with maternal mRNA translation and ensures irreversibility of the oocyte-to-embryo transition (Hara, 2017).

The massive changes in mRNA translation accompanying egg activation occur in a matter of minutes and must be linked to completion of meiosis in the oocyte. This study found that in Drosophila the solution to this developmental challenge is the regulation of PNG kinase activity. The results show that GNU is phosphorylated at CycB/CDK1 sites in mature oocytes, and in vitro CycB/CDK1 can directly phosphorylate GNU and thereby inhibit its ability to activate PNG kinase via inhibition of formation of the complex. In mature oocytes that are arrested at metaphase I, GNU is phosphorylated at CDK1 consensus sites and prevented from interaction with the PNG-PLU sub-complex. Following egg activation, as CycB protein and H1 kinase activity decline, GNU is dephosphorylated. This corresponds to the completion of meiosis and activation of the PNG kinase complex. Consistent with dephosphorylation of GNU being the crucial event for activation of PNG, substitution of a phosphorylation-resistant form of GNU into the oocyte results in premature elevation of CycB protein, implying PNG activation. Thus it is proposed that control of PNG kinase activity via GNU phosphorylation by CycB/CDK1 links meiotic completion and translational control of maternal mRNA to coordinate their timing precisely during egg activation. Active PNG leads to decreased GNU protein levels. This makes a negative feedback to shut down PNG kinase activity, thereby ensuring PNG kinase activity is constrained to the short developmental window of the oocyte-to-embryo transition (Hara, 2017).

These findings highlight the linchpin role the GNU subunit plays in the developmental control of PNG kinase activity. This regulation is exerted both at the levels of GNU protein and via its phosphorylation state. Although the presence of all three PNG kinase complex proteins is limited to late oogenesis through the early embryo, GNU is present over a narrower time window. GNU protein is undetectable in stage 10 oocytes and is rapidly accumulated during oocyte maturation. A previous genome-wide study showed that GNU protein accumulation during oocytes maturation relies on translational activation of its mRNA. Although the regulatory mechanisms for gnu translational activation remain to be defined, it is clearly dependent on CDK1, but not MOS, activity. Thus CDK1 promotes the appearance of GNU, permitting all PNG subunits to be present in the mature oocyte and poised for activation, while it simultaneously prevents activation by phosphorylation of GNU (Hara, 2017).

Several previous experimental observations now have clear significance in support of the role of CDK1 phosphorylation of GNU in inhibiting PNG complex formation and kinase activation. Ectopic GNU protein expression in early stage oocytes, which are in prophase I, causes png-dependent premature CycB protein expression and actin disorganization. Importantly, these defects result from expression of GNU at a developmental stage when although PNG and PLU are present at low levels, CycB/CDK1 is not active, and thus would not be able to block formation of the PNG complex. Analysis of the phosphorylation state of GNU in mutants for the calcipressin Sarah (sra) or the meiosis-specific APC/C activator Cortex (cort) showed that GNU remains hyperphosphorylated in eggs laid from both types of mutant mothers. Interestingly, both of these mutants fail to complete meiosis, with sra mutant eggs arrested in anaphase I, and cort mutants arrested in metaphase II. The failure of GNU to be dephosphorylated in both of these mutants agrees with the demonstration that GNU becomes hypophosphorylated after the completion of meiosis (Hara, 2017).

Although PNG and PLU levels in embryos are gradually decreased 2-4 hr after laying GNU seems to disappear after the completion of meiosis but before the initiation of embryogenesis. This is evidenced by GNU protein levels being decreased in unfertilized in vivo activated eggs, which complete meiosis II but do not enter into the embryonic cycles. This rapid GNU disappearance requires PNG kinase activity, revealing the existence of a negative feedback to shut off PNG activity shortly after its activation. It remains to be determined how GNU protein levels decline. PNG phosphorylates GNU on sites other than the CDK1 sites, and this phosphorylation may target GNU for degradation mediated by an E3 ubiquitin ligase of the SCF or the APC/C families. It is noted that GNU levels fail to decline in cort mutants, which lack one form of the APC/C. This may be an indirect effect of failure of dephosphorylation of the CDK1 sites in GNU and thus absence of PNG activation, but it is also possible that GNU is targeted directly to the APC/Ccort after completion of meiosis (Hara, 2017).

A key question raised by the demonstration of the regulatory role of GNU phosphorylation is whether developmental control of phosphatase activity is required. One possibility is that the phosphatase responsible for dephosphorylation of the CDK1 sites in GNU is constitutively active. CycB/CDK1 activity is high in the metaphase I-arrested mature oocyte, but egg activation triggers meiotic resumption and CDK1 inactivation. Reduction of CycB/CDK1 activity in the presence of active phosphatase may be sufficient for hypophosphorylated GNU. Alternatively, egg activation could lead to activation of a phosphatase. The identity of the GNU phosphatase for the CDK1 sites awaits elucidation. Although it has been shown that PP1 is capable of dephosphorylating GNU in mature oocytes, PP2A also is a possible GNU phosphatase during the process, particularly given it is the major phosphatase that removes phosphates on CDK1 substrates in mitotic exit (Hara, 2017).

The partial activity of the GNU 9A protein form provides insights into the mechanism by which GNU activates the PNG kinase. This mutant protein binds to the PNG-PLU sub-complex even in high CDK activity, however, it does not activate PNG kinase to the full extent that wild-type GNU does. This implies that there is an additional step to PNG kinase activation after GNU and the PNG-PLU sub-complex interaction. It is likely that the kinase activation following complex formation involves the N-terminus of GNU. A point mutation in the GNU N-terminal region (changing Pro 17 to Leu) retains ability to bind the PNG-PLU sub-complex but not does not activate PNG kinase activity in vitro (Hara, 2017).

The PNG kinase is significant for the understanding of how a kinase can rapidly control translation of hundreds of mRNAs. In addition to these insights into mRNA translation, identifying and defining the role of regulators involved in triggering the profound changes accompanying the oocyte-to-embryo transition is crucial for understanding of the onset of development, with implications for human fertility. This study has shown two forms of regulation of PNG kinase activity: one being regulation of protein expression of PNG kinase complex components and another being regulation of its activity. Strikingly, a cell cycle regulator, CDK1, controls both. This implies that CDK1 precisely regulates PNG kinase activity, a translational regulator, thus coordinating cell cycle progression and the translational landscape change during the oocyte-to-embryo transition. There are interesting parallels between the current findings and those in C. elegans. In C. elegans another kinase, the MBK-2 member of the conserved DYRK family of dual specificity tyrosine kinases, like PNG is crucial for the oocyte-to-embryo transition. MBK-2 activation is linked to the meiotic cell cycle by being downstream of the APC/C. MBK-2 controls proteolysis of oocyte proteins through the SCF E3 ubiquitin ligase, and it also affects RNA granule dynamics and thus likely impacts translation. Although PNG and MBK-2 are distinct kinases, these distantly related invertebrates utilize parallel approaches of coupling to cell cycle regulators to limit kinase activity to the oocyte-to-embryo transition. In both organisms, this links meiotic progression to gene expression changes after egg activation. This conservation suggests that this strategy may be employed for the onset of mammalian development (Hara, 2017).

Identification of PNG kinase substrates uncovers interactions with the translational repressor TRAL in the oocyte-to-embryo transition

The Drosophila Pan Gu (PNG) kinase complex regulates hundreds of maternal mRNAs that become translationally repressed or activated as the oocyte transitions to an embryo. A previous paper (Hara, 2017), demonstrated PNG activity is under tight developmental control and restricted to this transition. This study's examination of PNG specificity showed it to be a Thr-kinase yet lacking a clear phosphorylation site consensus sequence. An unbiased biochemical screen for PNG substrates identified the conserved translational repressor Trailer Hitch (TRAL). Phosphomimetic mutation of the PNG phospho-sites in TRAL reduced its ability to inhibit translation in vitro. In vivo, mutation of tral dominantly suppressed png mutants and restored Cyclin B protein levels. The repressor Pumilio (PUM) has the same relationship with PNG, and PUM was shown to be a PNG substrate. Furthermore, PNG can phosphorylate BICC and ME31B, repressors that bind TRAL in cytoplasmic RNPs. Therefore, PNG likely promotes translation at the oocyte-to-embryo transition by phosphorylating and inactivating translational repressors (Hara, 2018).

As an initial approach to identify substrates for the PNG kinase, predicted to be a Ser/Thr kinase, attempts were made to determine whether PNG phosphorylation occurs at consensus sequences. A positional scanning peptide library was treated with active PNG kinase complex or a complex with catalytically inactive PNG (KD: kinase dead) purified from Sf9 cells. Peptides were robustly phosphorylated by the active PNG kinase complex in contrast to the kinase-dead control. PNG exhibited a strong preference to phosphorylate threonine, because peptides whose phospho-acceptor site was fixed with threonine were strongly phosphorylated, whereas serine peptides were phosphorylated at reduced levels. Although no strong consensus sequence was identified, PNG was most strongly selective for hydrophobic amino acids at -3 relative to the phosphorylated residue, and it had some preferences for aromatic residues at position -2 and for arginine at position +2. Increased phosphorylation of peptides with threonine present outside of the intended phospho-acceptor position was likely an artifact resulting from the presence of two potential phosphorylation sites (Hara, 2018).

The peptide arrays did not yield a consensus sequence for PNG of sufficient specificity to be used to identify putative substrates. Because of the limitations of this approach, an unbiased biochemical screen was carried out. Purified recombinant PNG kinase was used to thio-phosphorylate substrates in embryonic extracts, identifying them by recovery of thio-phosphorylated peptides by mass spectrometry. A pilot screen was done with wild-type PNG kinase and 45 proteins were phosphorylated. A second screen was done in which extracts were treated in parallel with wild-type and kinase-dead PNG. In this second screen, the total representation of peptides in the extract was quantified by doing mass spec analysis of the peptides that did not bind to iodoacetyl agarose. In the second experiment, 36 proteins had at least two independent peptides phosphorylated by wild-type but not kinase-dead PNG. These included 27 of the proteins identified in the pilot experiment (Hara, 2018).

A high representation of phosphopeptides was recovered for the translational repressor Trailer Hitch (TRAL) with wild-type but not kinase-dead PNG. Other phosphorylated proteins were ribosomal proteins and translation factors, as well as the PLU activating subunit of the PNG complex. Out of 36 substrates identified, 19 were proteins known to be involved in mRNA translation (Hara, 2018).

79% of the identified unique peptides had threonine as the phospho-acceptor residue. The identified peptides showed an enrichment of hydrophobic residues at -3 position as in the peptide library, confirming that PNG tends to phosphorylate threonine three residues downstream of a hydrophobic amino acid. The threonine preference was also highly significant in the context of the Drosophila proteome. The correspondence with the peptide sequence preference of PNG is further confirmation that the observed phosphopeptides likely reflect direct phosphorylation by PNG. Although the substrates don't reveal a strong PNG consensus sequence, it is possible that interaction between substrates and the PLU or GNU activating subunits may provide specificity beyond that at the phosphorylation site (Hara, 2018).

Focus was placed on TRAL, because although there were many more abundant proteins in the extracts, a high number of PNG-phosphorylated peptides for TRAL were recovered. TRAL is a member of the (L)Sm protein family composed of RAP55 in vertebrates, CAR1 in C. elegans, and Sdc6 in yeast. Tests were performed to see whether PNG can phosphorylate TRAL in vitro. A powerful aspect of the thio-phosphate substrate screen is that the MS analysis identifies the phosphorylated amino acids. 15 amino acids (13 of them threonine), clustered in the C-terminal half of the protein, were phosphorylated by PNG in embryonic extracts. MBP fusions of purified full length TRAL, or the N- and C- terminal fragments were incubated with purified PNG and [γ 32P]-ATP and analyzed by autoradiography. The full-length protein and the C-terminal half, but not the N-terminal half, were phosphorylated by PNG in vitro. To determine whether PNG-dependent phosphorylation required the amino acids identified in the substrate screens, all 15 were changed to alanine. For both the full-length protein and the C-terminal half, the level of phosphorylation by PNG was reduced with the alanine-substituted forms. Residual phosphorylation of the alanine-substituted form of TRAL raises the possibility that there are other potential PNG phosphorylation sites in the C-terminus of TRAL that were not detected in the screen (Hara, 2018).

Whether phosphorylation of TRAL by PNG inhibits its activity was tested. RAP55 from Xenopus and Sdc6 from yeast are able to inhibit translation in vitro in yeast apparently by blocking the function of the eIF4G subunit of the eIF4F initiation factor. This study examined translation of an mRNA encoding Myc-tagged GFP in reticulocyte lysates and found that as for other family members, addition of Drosophila TRAL inhibited translation. Because in the in vitro reaction purified PNG does not phosphorylate TRAL to full stoichiometry, the effect of PNG phosphorylation was evaluated by generating a phosphomimetic form of TRAL in which aspartic acid was substituted for the fifteen PNG phosphorylation sites. Strikingly, the phosphomimetic mutations suppressed the translational repression by TRAL. The potential existence of additional PNG phosphorylation sites in the C-terminus of TRAL could account for why suppression of translational repression by the phosphomimetic form of TRAL was not complete. In contrast, TRAL in which these residues were replaced by alanine still inhibited translation of the reporter mRNA in the extracts. These results are consistent with phosphorylation of TRAL by PNG relieving its ability to repress translation. Together these results support the conclusion that TRAL is a PNG substrate, but they reveal that TRAL phosphorylation is developmentally dynamic and involves several kinases (Hara, 2018).

Therefore, genetic interactions were sought between png and tral mutants. The png gene was identified because mutant females produce eggs that complete meiosis but subsequently fail to initiate mitotic divisions. Nevertheless, DNA replication continues, resulting in embryos with giant, polyploid nuclei. In strong alleles of png there is no mitosis, whereas weaker alleles permit a few mitotic divisions but these nuclei ultimately also become polyploid. The absence of mitosis in png mutants is due to a failure to promote cyclin B mRNA translation at egg activation. Removal of one copy of some genes (such as the translational repressor pum,) can suppress the giant-nuclei png phenotype, resulting in embryos that undergo more mitotic divisions and thus have more nuclei. If the gene acts downstream of png, this suppression is consistent with png acting negatively on the gene. In contrast, removal of one copy of a gene such as cyclin B enhances the png phenotype, consistent with png having a positive effect on this gene (Hara, 2018).

A comparison was made of embryos laid by females with png¹⁰⁵⁸/png³³¹⁸ with one copy of tral mutated to sibling controls solely mutant for png. Reducing the dosage of tral (a heterozygous tral1 mutation, which has a P element insertion) suppressed the png phenotype, permitting additional mitoses and increased numbers of nuclei. This suppression is even more pronounced with a deletion that completely removes the tral gene. These genetic epistasis results complement the in vitro translation results with the phosphomimetic TRAL form. They are consistent with TRAL being a target of PNG and phosphorylation negatively affecting TRAL (Hara, 2018).

To test whether the genetic interactions between tral and png affect cyclin B mRNA translation, protein levels were examined by immunoblotting of extracts from the mutant and control embryos. Strikingly, Cyclin B protein levels were increased in the png transheterozygous embryos when the dosage of tral was reduced. Consistent with the suppression phenotypes, the amount of Cyclin B was restored more with the deletion than with the tral1 allele. Cyclin A, another PNG translational target, also was increased with reduced TRAL (Hara, 2018).

Taken together, the in vitro and in vivo phosphorylation results and the genetic interaction data indicate that phosphorylation of TRAL by PNG blocks its repressive effects on translation, permitting translation of cyclin B at egg activation to permit embryonic mitoses. This could be due to PNG phosphorylation directly repressing TRAL function or via an effect of phosphorylation on the localization of TRAL. TRAL is present in large cytoplasmic RNP granules in mature oocytes in both Drosophila and C. elegans, and these disperse on egg activation. Thus, one model for the effect of PNG on TRAL is that phosphorylation could affect the localization of TRAL to RNP granules. These large visible granules were examined using a GFP-Tral FlyTrap line with or without png mutations and following TRAL localization during in vitro egg activation. Early in activation, by about 10 min, TRAL granules became diminished. In png mutant eggs, the TRAL granules also disappeared with normal timing. It is concluded that PNG does not appear to be involved in this reorganization of TRAL granules. Indeed, dispersal of TRAL from granules occurs prior to when PNG becomes active at 30 min after egg activation. PNG phosphorylation may more directly affect the ability of TRAL to inhibit translation initiation, as indicated by the effect of the phosphomimetic form on translation in reticulocyte lysates (Hara, 2018).

Given the hundreds of mRNAs whose regulation at egg activation is dependent on PNG, it seemed probable that PNG affects translation through multiple mechanisms and may have multiple substrate targets. Previous work showed that the translational repressor pumilio (pum) dominantly suppresses png; a heterozygous mutation of pum restores both Cyclin B protein levels and mitosis in png mutant embryos. Even PUM nonphosphorylated peptides were not recovered in the substrate screen, therefore, the possibility of PUM being a PNG substrate could not be evaluated. Consequently, a direct interaction between png and pum was tested by asking whether PNG can phosphorylate PUM in vitro. A GST-PUM fusion protein is phosphorylated by purified wild-type PNG kinase but not by the kinase-dead form (Hara, 2018).

The ME31B RNA helicase acts as a translational repressor and is a binding partner to TRAL. The helicase was not recovered above the cut off in the substrate screen, although one ME31B phosphopeptide was present in the wild-type but not kinase-dead PNG sample. Given its interaction with TRAL, ME31B was directly tested in vitro, and PNG was able to phosphorylate it. Thus, PNG phosphorylation may affect both of these conserved proteins and their role as a complex in controlling translation (Hara, 2018).

Another translational regulator that is a potential PNG substrate is BICC. BICC binds to the GNU subunit of the PNG complex directly through its SAM domain, and BICC also is known to physically interact with TRAL. BICC was not, however, recovered from the substrate screen. Despite this, PNG readily phosphorylates BICC in vitro (Hara, 2018).

These results raise the possibility that PNG acts on a number of translational repressors. The two PNG substrate screens likely were not saturating to identify all potential translational repressor targets. The translational repressors Cup and Caprin were recovered in the first substrate screen but not by this study's criteria in the second. The dominant genetic suppression of png observed with mutation of tral or pum generates the hypothesis that PNG may inactivate multiple translational repressors by phosphorylation to promote translation of different sets of mRNAs at egg activation. It is also possible that PNG's effect on multiple repressors may target a single set of mRNAs localized to RNP granules. For example, ME31B is bound to TRAL. BicC genetically interacts with tral, the protein appears to localize to the RNP granules in which TRAL and ME31B reside, and it binds to GNU. From these observations, PNG might phosphorylate multiple targets on RNP granules to de-repress translational inhibition of maternal mRNAs at egg activation (Hara, 2018).

In addition to its effects at egg activation, PNG may indirectly affect translational repressors later in embryogenesis, at a developmental time when PNG appears to be inactivated. In the embryo the TRAL, ME31B, and Cup proteins form an inhibitory complex that represses the translation of maternal mRNAs. These proteins have been shown to be degraded during the maternal-to-zygotic transition, and functional PNG is a prerequisite for this degradation (Hara, 2018).

Previous work has shown that the PNG kinase is activated by a signal downstream of egg activation and thus controls massive changes in maternal mRNA translation. This study has now found that TRAL is a PNG substrate using a biochemical screen. Phosphorylation by PNG suppressed TRAL's ability to repress mRNA translation. This antagonism also was supported by genetic interaction between png and tral in fertilized embryos, suggesting that TRAL phosphorylation by PNG during the oocyte-to-embryo transition is a key to remodel maternal mRNAs' translation activity (Hara, 2018).

The PNG kinase functions as a signal transducer for the external egg activation signal to mRNA translation in the cytoplasm in the activated eggs. Similar strategies can be used in oocyte maturation, during which a hormonal signal leads to phosphorylation of translational regulators to control mRNA translation. In neurons, stimuli cause translocation of mRNA followed by translational activation. Understanding signaling pathways that transmit extracellular signals to translational controls thus is likely to provide insight into molecular mechanisms in fertility as well as synaptic plasticity and memory (Hara, 2018).

The E2 Marie Kondo and the CTLH E3 ligase clear deposited RNA binding proteins during the maternal-to-zygotic transition

The maternal-to-zygotic transition (MZT) is a conserved step in animal development, where control is passed from the maternal to the zygotic genome. Although the MZT is typically considered from its impact on the transcriptome, previous work found that three maternally deposited Drosophila RNA-binding proteins (ME31B, Trailer Hitch [TRAL], and Cup) are also cleared during the MZT by unknown mechanisms. This study shows that these proteins are degraded by the ubiquitin-proteasome system. Marie Kondo, an E2 conjugating enzyme (FlyBase: Ubiquitin conjugating enzyme E2H), and the E3 CTLH ligase are required for the destruction of ME31B, TRAL, and Cup. Structure modeling of the Drosophila CTLH complex suggests that substrate recognition is different than orthologous complexes. Despite occurring hours earlier, egg activation mediates clearance of these proteins through the Pan Gu kinase, which stimulates translation of Kdo mRNA. Clearance of the maternal protein dowry thus appears to be a coordinated, but as-yet underappreciated, aspect of the MZT (Zavortink, 2020).

Proper embryogenesis is critical for animal development. Many of the earliest events occur prior to the onset of zygotic transcription, and they are instead directed by maternally deposited proteins and messenger RNAs (mRNAs). During the maternal-to-zygotic transition (MZT), genetic control of developmental events changes from these maternally loaded gene products to newly made zygotic ones. Thus, the MZT requires both the activation of zygotic transcription and clearance of maternal transcripts. Failure to mediate either of these processes is lethal for the embryo (Zavortink, 2020).

In contrast to understanding of the transcriptome during the MZT, much less is known about changes in the proteome. Despite the fact that the maternal dowry of proteins plays key roles in embryogenesis, there are only a handful of examples of cleared maternal proteins. Recently, three RNA-binding proteins (ME31B, Trailer Hitch [TRAL], and Cup) were found to be rapidly degraded during the MZT in Drosophila melanogaster, at a time point coinciding with the major wave of zygotic transcription. ME31B, TRAL, and Cup form a complex that blocks translation initiation. All three proteins are required for oogenesis, and they appear to bind and repress thousands of deposited maternal mRNAs. The degradation of ME31B, TRAL, and Cup coincides with many of the hallmarks of the MZT, but explorations into this issue have been hindered by a lack of understanding of how their destruction is controlled (Zavortink, 2020).

An intriguing observation has been made that genetically linked the clearance of ME31B, TRAL, and Cup, to the Pan Gu (PNG) kinase (Wang, 2017). Composed of three subunits (PNG, Giant Nuclei [GNU], and Plutonium [PLU]), the PNG kinase is central to the oocyte-to-egg transition and mediates key aspects of embryogenesis, including resumption of the cell cycle, zygotic transcription, and maternal mRNA clearance. Unlike many animals, the oocyte-to-egg transition in Drosophila does not require fertilization but is instead triggered by egg activation. The PNG kinase is activated by mechanical stress as the oocyte passes through the oviduct, and then phosphorylation and degradation of the GNU subunit quickly inactivates the kinase, restricting its activity to the first half hour after egg activation (Hara, 2017). One way that PNG mediates the oocyte-to-embryo transition is by rewiring post-transcriptional gene regulation. Possibly by phosphorylating key RNA-binding proteins such as Pumilio, PNG activity leads to changes in the poly(A)-tail length and translation of thousands of transcripts during egg activation (Hara, 2018). Importantly, two targets induced by PNG activity are the pioneer transcription factor Zelda, which is responsible for initial zygotic transcription, and the RNA-binding protein Smaug, which is responsible for clearance of many maternal transcripts. The PNG kinase also phosphorylates ME31B, Cup, and TRAL (Hara, 2018), but it is unclear what effect phosphorylation has on these proteins. One possibility has been that PNG phosphorylation could lead to the degradation of ME31B, TRAL, and Cup, but this model has been thus far unexplored (Zavortink, 2020).

The ubiquitin-proteasome system is a major protein degradation pathway. A series of ubiquitin activating enzymes, conjugating enzymes, and ligases (E1, E2, and E3, respectively) lead to the post-translational addition of a polyubiquitin chain on a target protein, which then serves as a molecular beacon for degradation by the proteasome. E3 ligases are typically thought to recognize target proteins, while E2 conjugating enzymes provide the activated ubiquitin and in turn recognize the E3 ligase. There are hundreds of different E3 ligases and 29 annotated E2 conjugating enzymes in Drosophila, but most of the client substrates are unknown, and few have been implicated in the MZT (Zavortink, 2020).

Given the key roles of ME31B, Cup, and TRAL in oogenesis and embryogenesis, it was of interest to understand the mechanisms controlling their degradation. In particular, this study sought to answer how PNG activity at egg activation leads to the degradation of these three RNA-binding proteins several hours later, and how their degradation is coordinated with other elements of the MZT, including zygotic transcription and maternal mRNA clearance. To answer these questions, a selective RNAi screen was performed in Drosophila, and the E2 conjugating enzyme was identified as UBC-E2H/Marie Kondo (Kdo) and the E3 ligase as the CTLH complex. Interestingly, structural models based on the S. cerevisiae complex (Qiao, 2020) suggest that the Drosophila version is organized differently than its orthologous complexes. The CTLH complex recognized and bound ME31B and Cup even in the absence of PNG activity, strongly suggesting that phosphorylation is not required for the destruction of these proteins. In contrast, Kdo mRNA is translationally upregulated by more than 20-fold upon egg activation in a PNG-dependent manner. Thus, egg activation through PNG mediates translation upregulation of Kdo and so leads to ME31B, Cup, and TRAL destruction (Zavortink, 2020).

Kdo is conserved from yeast to humans and is known to work through the CTLH E3 ligase, a multicomponent complex. (Note that the S. cerevisiae complex is called the Gid complex.) Using BLAST for the human CTLH components, it was easy to identify putative D. melanogaster homologs: RanBPM (homologous to Hs RanBP9), Muskelin, CG3295 (homologous to Hs RMND5A/GID2), CG7611 (homologous to Hs WDR26), CG6617 (homologous to Hs TWA1/GID8), and CG31357 (homologous to Hs MAEA). Putative homologs for Hs GID5/ARMC8 or Hs GID4 were not found. Notably, none of these genes were annotated as putative E3 components in FlyBase, and thus none were included in an original screen (Zavortink, 2020).

ME31B, Cup, and TRAL are RNA-binding proteins that are degraded during the MZT. Despite occurring several hours after egg laying, degradation of these proteins is triggered by egg activation through the activity of the PNG kinase and appears to be mediated by the ubiquitin-proteasome system. Through a medium-scale RNAi screen, the E2 conjugating enzyme Kdo was identified as being required for the clearance of ME31B, TRAL, and Cup. Kdo is conserved from yeast to humans and, as in those systems, appears to work with the CTLH complex, which acts as the E3 ligase. Components of the CTLH complex physically interact with ME31B and Cup, and the CTLH complex is also required for the degradation of ME31B, TRAL, and Cup during early embryogenesis. Structure-based homology suggests that, despite its conservation from yeast to humans, the Drosophila CTLH complex has an unusual architecture, and it remains unclear how it recognizes its substrate. The association of CTLH with ME31B occurs in the absence of PNG activity, suggesting that, although ME31B (as well as TRAL and Cup) are phosphorylated by the kinase, phosphorylation may not be required for their destruction. Instead, translation of Kdo appears to be suppressed during oogenesis by its short poly(A) tail length and binding of ME31B. Its translation is dramatically upregulated at the oocyte-to-embryo transition, in a process that depends on PNG activity. Together, these data suggest a model that egg activation via the PNG kinase leads to translational activation and production of Kdo, which then allows the CTLH complex to ubiquitinate ME31B, TRAL, and Cup and ultimately leads to their destruction. Interestingly, based on RNA-seq data from FlyBase, Muskelin shows exquisite tissue-specificity and is only strongly expressed in the ovaries. This observation, together with the translational control of Kdo, may partly explain how ME31B, a ubiquitous protein, is specifically destabilized in the early embryo (Zavortink, 2020).

Although the CTLH complex is conserved, it has not yet been studied in Drosophila. The data point to this complex being composed of multiple components (Muskelin, RanBPM, Houki, Souji, and Katazuke), as in other organisms. However, due to a lack of available reagents, the stoichiometry of these components is unknown, and it remains possible that there are additional, Drosophila-specific components. Nonetheless, so far, the CTLH complex in Drosophila appears different than the human and yeast complexes. Although Gid7 and WDR26 are important in the yeast and human versions, respectively, and a Drosophila ortholog (CG7611) was identified, no evidence was found of its association with ME31B or requirement for ME31B degradation; the role of CG7611 in the Drosophila CTLH complex warrants further investigation. Orthologs of Gid4 and Gid5, which are critical for substrate recognition in S. cerevisiae, were not found. Intriguingly, the residues and domains important for the Gid1-Gid4 and Gid8-Gid5 interactions in budding yeast appear absent to be in RanBPM and Hou, raising the fundamental question of how the Drosophila CTLH complex recognizes and positions its substrate proteins. Answering this question will require future investigation and may shed light on other proteins targeted by the Drosophila CTLH complex and the extent to which ME31B, a ubiquitously expressed protein, is targeted outside of the MZT (Zavortink, 2020).

One unexpected result is the role of PNG in mediating the destruction of ME31B, TRAL, and Cup. PNG phosphorylates all three proteins, and so the initial hypothesis was that this modification also stimulated their destruction. However, contrary to expectations, ME31B and Cup interacted with the CTLH complex even in png50 embryos, demonstrating that phosphorylation by PNG was not required for binding of ME31B and Cup by the E3 ligase. An unresolved question, then, is how PNG phosphorylation affects the activities of ME31B, TRAL, and Cup. Intriguing observations from the Orr-Weaver lab suggest that the modification can impact the ability of these proteins to repress gene expression (Hara, 2018). It is tempting to speculate that phosphorylation may then contribute to the MZT by modulating the activities of ME31B, TRAL, and Cup, rather than their stability (Zavortink, 2020).

The link between PNG and the destruction of ME31B, TRAL, and Cup instead appears to be mediated through the translational upregulation of Kdo. Although PNG may contribute through other, as-yet undiscovered, mechanisms as well (such as phosphorylating unknown CTLH adaptor proteins), this link is sufficient to explain the PNG requirement for ME31B degradation: in the absence of Kdo, ME31B is stable during the MZT, and in the absence of PNG, Kdo is not detectably expressed. An important question for the future will be to understand what elements in the Kdo mRNA are responsible for its translational repression during oogenesis. One hint may be that the 3'UTR of Kdo contains several putative Pumilio-binding sites, and translation of Kdo is upregulated in ovaries where Pumilio has been knocked down. Pumilio is also a target of PNG (Hara, 2018), and so a possible model is that translational repressors, such as Pumilio, are phosphorylated and inactivated at egg activation, leading to the production of Kdo (Zavortink, 2020).

PNG also mediates the translational upregulation of key MZT effectors: Zelda, the pioneer transcription factor, and Smaug, an RNA-binding protein that targets nearly two-thirds of the maternal transcriptome for degradation. Together with the current results, a picture is emerging that egg activation stimulates the production of multiple key factors that are important for clearing the maternal RNA and protein dowry and for producing zygotic gene products (Zavortink, 2020).

Although the MZT has typically been considered from the perspective of RNA, a role for maternal protein clearance is becoming clearer. Over the past few years, the list of proteins degraded during the Drosophila MZT has grown and now includes GNU, Matrimony, Cort, Smaug, ME31B, TRAL, and Cup. Unbiased mass spectrometry experiments also suggest that Wispy and Dhd are also robustly degraded. As this list of proteins in Drosophila and other developmental systems increases, a new question is emerging: how many maternally deposited proteins are degraded during the MZT? Understanding the mechanisms controlling protein degradation during the MZT as well as the impact of removing the maternal protein dowry will be key issues to explore in the future (Zavortink, 2020).

GENE STRUCTURE

cDNA clone length - 1126

Bases in 5' UTR - 80

Exons - 1

Bases in 3' UTR - 170

PROTEIN STRUCTURE

Amino Acids - 291

Structural Domains

To determine the biochemical function of the png gene product, the png gene was cloned by positional cloning. Complementation tests between png mutations and deficiencies, and duplications delineate png to a 130 kb region. Using genomic DNA clones from the region in germline transformation rescue experiments, a 1.7 kb fragment capable of rescuing the maternal effect lethality of png minus homozygotes was isolated. Sequencing of the fragment reveals an open reading frame (ORF) that spans most of the 1.7 kb and is contained entirely within it, suggesting that it is the png gene. To identify the png transcript, Drosophila ovary cDNA clones were isolated that hybridized to the rescuing fragment. The longest cDNA is 1126 bp, not including the poly-A tail, and the 5' end of the transcript starts 80 bp upstream of the ATG. The sequence 5' to the ATG contains an in frame stop codon and no other ATG codons in any frame. Alignment of the 1.7 kb genomic sequence with the eight cDNAs shows no introns, indicating that the png transcript is a single exon. Conceptual translation of the longest cDNA reveals that the Png protein has strong homology to Ser/Thr kinases. The protein shows highest homology to members of the Snf1/AMP kinase family, with 27% amino acid identity to Snf1 over 265 amino acids. Phylogenetic analysis, however, shows that Png is more distantly related to Snf1 family members than they are to each other, and probably represents a new family of Ser/Thr kinases. png is small, encoding a 291 amino acid protein with a predicted molecular weight of 33 kDa, that contains only a catalytic domain. The absence of a regulatory domain suggests that separately encoded regulatory subunits may associate with Png (Fenger, 2000).

date revised: 5 August 2021

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.