no poles encodes a predicted E3 ubiquitin ligase required for early embryonic development of Drosophila

To ensure faithful transmission of the genome upon cell division,eukaryotic cells have developed checkpoints, regulatory pathways that delay cell-cycle progression until completion of prior events. The DNA damage/replication checkpoint plays a crucial role in preserving genomic integrity (Branzei and Foiani,2008). Upon detection of DNA defects, the kinases ATM (ataxia telangiectasia mutated) and ATR (ATM-Rad3-related) are recruited to sites of damage and activated. ATM and ATR substrates include checkpoint kinases CHK1 and CHK2, which phosphorylate proteins that mediate cell-cycle arrest. The ensuing delay, resulting from engagement of this checkpoint, presumably allows cells time to correct defects.

Research over the past decade has highlighted major roles for protein ubiquitination in regulating cellular responses to DNA damage(Harper and Elledge, 2007). This post-translational modification, which involves covalent linkage of one or more ubiquitin molecules to another protein, regulates many fundamental cellular processes (Pickart,2001). Ubiquitination may alter the fate of a protein in numerous ways, such as targeting it for destruction by the 26S proteasome, changing its subcellular location, or changing its protein-protein interactions.

Ubiquitination is a highly dynamic, multi-step process that requires three components: ubiquitin-activating enzyme (E1), ubiquitin-conjugating enzyme (E2 or Ubc) and ubiquitin ligase (E3). E3s can be divided into two main classes:HECT and RING domain-containing proteins. RING-type E3 ubiquitin ligases(Freemont, 2000; Jackson et al., 2000) contain a specialized motif of 40 to 60 residues that binds two zinc atoms. Many RING-type E3s bind to partnering E2 conjugating enzymes via their RING domains(Passmore and Barford, 2004). Database searches of the Drosophila genome predict that it contains one E1, 36 E2s and ∼130 E3s, which represents ∼40% of the ubiquitination machinery in humans (Ditzel and Meier, 2005).

Significant insights into the roles of many cell-cycle regulators have come from studying their functions in Drosophila. Drosophila is well suited for studying cell-cycle regulation during the formation of a multicellular organism, in large part because of its developmental use of cell cycles that differ in structure from canonical G1-S-G2-M cycles and the availability of genetic tools (Garcia et al., 2007; Lee and Orr-Weaver,2003). The first thirteen cell cycles of Drosophilaembryogenesis involve nearly synchronous nuclear divisions driven by stockpiles of maternally expressed mRNA and protein(Foe et al., 1993). These rapid cycles (∼10 minutes in length) consist of oscillating S-M (DNA replication-mitosis) phases without intervening gap phases or cytokinesis. Minimal gene transcription occurs during this developmental stage, so cell cycles are regulated by post-transcriptional mechanisms. At cycle 14, the embryo cellularizes and initiates zygotic transcription at the midblastula transition (MBT).

We report here the identification and characterization of a Drosophila maternal-effect lethal mutant that we have named `no poles' (nopo). Embryos from nopo females undergo mitotic arrest with acentrosomal, barrel-shaped spindles during syncytial divisions. Our results indicate that this arrest is secondary to the activation of a CHK2-mediated DNA checkpoint in early embryos. We show that NOPO, a predicted E3 ubiquitin ligase, interacts with an E2 component, BEN. ben females are sterile, producing embryos with nopo-like defects. We propose that BEN-UEV1A and NOPO function together as an E2-E3 complex required for genomic integrity during Drosophilaembryogenesis.

Flies were maintained at 25°C using standard techniques. y wwas used as wild type unless otherwise indicated. cn Z2-1447 bw/CyOwas a gift from Charles Zuker (UC San Diego); ben¹ and mnk⁶⁰⁰⁶ stocks were from Mark Tanouye (UC Berkeley) and Bill Theurkauf (UMass Worcester), respectively; and the EYG5845 stock was from GenExel (Seoul, Korea). Other fly stocks were from the Bloomington or Szeged stock centers.

Five newly eclosed females of the indicated genotype and five wild-type males were incubated in yeast-pasted vials for two days and transferred to egg-collection chambers at 25°C. Eggs were collected daily over five days and scored for hatching ∼40 hours post-collection (>500 eggs per genotype). Hatch rate is the ratio of hatched to total eggs expressed as a percentage.

We screened a second chromosome deficiency collection for non-complementation of female sterility of nopo^Z1447. Females carrying nopo^Z1447 in trans to any of several overlapping deficiencies (Df(2R)Pcl-11B, Df(2R)Pcl-XM82, Df(2R)Pcl-7Bor Df(2R)PC4) were sterile, placing nopo in the 55A1-C1 interval.

We further mapped nopo^Z1447 by P-element-induced male recombination(Chen et al., 1998) relative to several insertions: lolal^EP2169, Dgp-1^BG00396,CG5721^EY03388, fj^KG03419 and EP(2)1081. Multiple independent recombinant chromosomes were recovered for each P-element tested. We narrowed nopo to five candidates in the 55B11-12 region (Dgp-1, CG10916, CG5726, CG5140 and CG5721)distal to Dgp-1^BG00396 and proximal to CG5721^EY03388, as annotated on FlyBase(Grumbling and Strelets,2006).

For each candidate gene, coding regions were sequenced as described(Rickmyre et al., 2007). nopo^Z1447 is a missense mutation in CG5140causing a glutamic acid to lysine change at residue 11 of the predicted protein. Df(2R)Exel7153, which deletes 15 genes in this region, was subsequently found to uncover nopo. Putative nopo homologs were identified using HomoloGene (release 56), and the RING domain of NOPO was identified using ScanProsite.

A nopo-null allele was generated by imprecise excision of P-element EYG5845. The 771-bp deletion nopo^Exc142 lacks part of the 5′-UTR and exons encoding residues 1-181.

cDNA encoding NOPO, BEN and UEV1A (GH03577, LD24448 and LD28904,respectively) were from the Drosophila Gene Collection. Human TRIP cDNA (ID 2821007) was from Open Biosystems.

A 3.8-kb genomic fragment containing CG5140 and flanking regions(Fig. 2A) was PCR-amplified from BAC clone BACR15G20 (Drosophila Genomics Resource Center) and subcloned into pCaSpeR4. A transgenic line carrying pCaSpeR4-CG5140was generated by P-element-mediated transformation via embryo injection (Rubin and Spradling,1982).

Methods for fixation, staining and fluorescence microscopy of embryos(1.5-2.5 hours unless otherwise indicated) and live-image analysis were previously described (Rickmyre et al.,2007). P-values for live-image data were obtained using a two-tailed, unpaired Student's t-test.

A fusion consisting of an N-terminal MBP tag and C-terminal NOPO was used to generate anti-NOPO antibodies. DNA encoding C-terminal NOPO (residues 224 to 435) was PCR amplified and subcloned into pMAL (New England Biolabs). MBP-C-NOPO was produced in bacteria, purified using amylose resin, and injected into guinea pigs (Covance).

Protein extracts were made by homogenizing embryos (1-2 hours) or dissected tissues in urea sample buffer (Tang et al., 1998). Proteins were transferred to nitrocellulose for immunoblotting using standard techniques. Antibodies used were as follows:guinea pig anti-NOPO (1:1000), mouse anti-GAPDH (1:1000, Abcam), mouse anti-α-tubulin (DM1a, 1:5000, Sigma), mouse anti-Cyclin B (F2F4, 1:200,Developmental Studies Hybridoma Bank), and rabbit anti-pY15-CDK1 (1:1000,Upstate).

HeLa cells were maintained in Dulbecco's modified Eagle Medium (DMEM)containing 10% fetal bovine serum. Plasmids encoding N-terminally tagged (eGFP or mCherry) versions of NOPO, TRIP and BEN generated by subcloning into pCS2 were transfected into cells using Lipofectamine 2000 (Invitrogen) according to the manufacturer's directions.

Cells were plated on fibronectin-coated coverslips 21 hours post-transfection and fixed three hours later. For direct fluorescence and centromere staining, cells were fixed for 20 minutes with 4% formaldehyde in CBS [10 mM MES (pH 6.1), 138 mM KCl, 3 mM MgCl₂, 2 mM EGTA, 0.32 M sucrose]. For PCNA staining, cells were fixed for 5 minutes in 70%methanol/30% acetone. For Cyclin A staining, cells were fixed for 20 minutes in 3% paraformaldehyde/20% sucrose in phosphate-buffered saline. Cells were permeabilized for 10 minutes with 0.5% Triton X-100 in Tris-buffered saline. Primary antibodies used were as follows: human autoimmune (CREST) serum(1:1000, ImmunoVision), Cyclin A (H-432, 1:100, Santa Cruz Biotechnology), and PCNA (PC10, 1:200 Santa Cruz Biotechnology). To visualize actin, cells were stained for one hour with fluorescently conjugated phalloidin (1:1000,Invitrogen). Fluorescently conjugated secondary antibodies were used at a dilution of 1:5000. Slides were mounted in Vectashield with DAPI (Vector Laboratories). Images were acquired using a Nikon Eclipse 80i microscope equipped with a CoolSNAP ES camera (Photometrics) and Plan-Apo 60×objective. For experiments involving quantification, at least 400 cells per condition were scored.

Yeast two-hybrid assays were performed as described(James et al., 1996). Plasmids expressing wild-type and mutant versions of NOPO, BEN and UEV1A fused to the Gal4 DNA-binding domain (`bait' vector pGBD-C) or Gal4-activation domain(`prey' vector pGAD-C) were transformed into Saccharomyces cerevisiaestrain PJ69-4A. Cells containing both bait and prey plasmids were selected by growth on synthetic complete (SC) plates lacking tryptophan and leucine, and spotted onto SC plates lacking tryptophan, leucine and histidine; growth on the latter (scored after two days at 30°C) indicates physical interaction between the fusion proteins tested.

The sensitivity of nopo larvae to hydroxyurea or irradiation was tested as described (Rickmyre et al.,2007).

To assess the visually mediated jump response, white-eyed control(w¹¹¹⁸) and mutant flies (two days old) were dark adapted,transferred without anesthesia to a Petri dish covered in vellum, and exposed to a `lights off' stimulus using an LED light apparatus as described(Fayyazuddin et al., 2006). Ten males per genotype were each tested in 10 trials separated by 30 seconds. The climbing ability of adult males was assessed as described(Silva et al., 2004), with three replicates per genotype. P-values were obtained using two-tailed, unpaired Student's t-tests. To visualize TDT muscle attachment sites, adult males (30 per genotype) were ventrally trans-illuminated with a dissecting microscope lamp as described(Edgecomb et al., 1993).

Adult males (5- to 7-days old) were injected using a Drummond Nanoject with∼50 nl of an overnight culture of Escherichia coli resuspended in phosphate-buffered saline. Six hours later, RNA was isolated by homogenizing flies in STAT-60 buffer according to the manufacturer's directions (Isotex Diagnostics). Following DNase treatment, cDNA was prepared by reverse transcription using Superscript II (Invitrogen). A diptericin-specific LUX primer (Invitrogen) was used to perform quantitative real-time PCR with the 7300 Real-Time PCR System (Applied Biosystems). diptericin levels were normalized to Rp49levels as an endogenous control. Results from three independent experiments were averaged and further normalized against buffer-injected Canton S flies. P-values were obtained using a two-tailed, unpaired Student's t-test.

We previously screened the maternal-effect lethal subset of the Zuker collection to identify genes that regulate S-M cycles of Drosophilaearly embryogenesis (Koundakjian et al.,2004; Lee et al.,2003; Rickmyre et al.,2007). We identified an allele (Z1447) of a gene that we have named `no poles' (nopo), based on the phenotype of acentrosomal mitotic spindles in mutant-derived embryos (see below). nopo^Z1447 females are completely sterile(Table 1). DNA staining of the egg chambers of nopo^Z1447 females revealed no obvious oogenesis defects, and the presence of polar bodies in their unfertilized eggs indicated that meiosis was completed (data not shown).

We found that nopo^Z1447-derived embryos undergo mitotic arrest during syncytial divisions with none developing to cellularization or gastrulation (Tables 1, 2; data not shown). Nuclei were unevenly spaced (compare Fig. 1B with 1A), and centrosome duplication before telophase was occasionally evident(Fig. 1F), both of which are consistent with failed mitotic divisions. nopo spindles are barrel-shaped, lack tubulin foci, and have misaligned chromosomes(Fig. 1D-F; Table 2); the lack of tubulin foci correlates with the loss of centrosomes at the poles, as revealed by staining for Centrosomin, a core component(Fig. 1I,J)(Li and Kaufman, 1996). Approximately 10% of nopo spindles are tripolar(Fig. 1K; Table 2). Bipolar nopospindles often appear to be wider and to contain more than the wild-type complement of chromosomes (compare Fig. 1E with 1C). Similar results were obtained for nopo^Z1447 in trans to Df(2R)Exel7153, which deletes nopo(Table 1).

We identified CG5140 as the nopo gene, using a combination of genetic mapping and molecular biology approaches(Fig. 2A; see Materials and methods for details). A wild-type copy of CG5140 carried as a transgene fully restored fertility to nopo females(Table 1), confirming that CG5140 is, indeed, the nopo gene. nopo encodes a predicted protein of 435 amino acids containing an N-terminal RING domain(Fig. 2B)(Saurin et al., 1996). The putative mammalian homolog of NOPO was named `TRAF-interacting protein' (TRIP)based on its ability to bind tumor necrosis factor (TNF) receptor-associated factors (TRAFs) (Lee et al.,1997). Mammalian TRIP was recently demonstrated to have RING-dependent E3 ubiquitin ligase activity in an auto-ubiquitination assay(Besse et al., 2007). Drosophila NOPO and human TRIP are 20% identical and 34% similar overall, with 47% identity and 65% similarity in their RING domains. Importantly, nopo^Z1447 causes a glutamic acid to lysine change in the RING domain at position 11 of the predicted protein, a residue that is invariantly negatively charged across species(Fig. 2C).

To gain further insights into the functions of nopo, we obtained additional alleles. 5-SZ-3004 (abbreviated as SZ3004) and EYG5845 are P-element insertions in the 5′-UTR of nopo (Fig. 2A). nopo^SZ3004 females have decreased embryonic hatch rates that are completely restored by precise P-element excision(Table 1; data not shown). nopo^SZ3004 is weaker than nopo^Z1447,based on the percentage of mutant-derived embryos that develop to gastrulation and embryonic hatching, and its phenotype is strongly temperature dependent. We generated a null allele of nopo (Exc142) via imprecise excision of EYG5845 (Fig. 2A; see Materials and methods). nopo^Exc142adults are viable and appear normal, except that females are sterile,producing embryos with the nopo phenotype(Fig. 1G); a CG5140transgene fully restored fertility (Table 1). nopo^Exc142 is stronger than nopo^Z1447 with 15% versus 72%, respectively, of their embryos reaching cortical divisions, suggesting that nopo^Z1447 has residual function.

We assessed NOPO levels in mutant embryonic extracts by immunoblotting using anti-NOPO antibodies that we generated(Fig. 2D). Consistent with the predicted size of NOPO (435 residues), these antibodies recognize an ∼48 kDa band in wild-type embryos that is absent in nopo^Exc142-derived (null) embryos. NOPO was not detected in nopo^SZ3004-derived embryos, although we occasionally observed trace amounts (data not shown). Wild-type levels of NOPO were found in nopo^Z1447-derived embryos, suggesting that the E11K mutation alters the function of NOPO, but not its stability.

We assessed NOPO levels throughout Drosophila development(Fig. 2E). NOPO is abundant in ovaries and early (0- to 3-hour) embryos; trace amounts are present in older(3- to 24-hour) embryos. We did not detect NOPO in larval brains, imaginal discs, testes or adult carcasses lacking germline tissues. Subsequent experiments, however, revealed roles for NOPO outside of early embryonic development, suggesting that our antibodies might not be sufficiently sensitive to detect its expression during other stages (see below). Using the UAS/Gal4 system, we found NOPO overexpression in the female germline causes severely reduced egg-laying and hatch rates, whereas broad overexpression of NOPO in somatic cells causes lethality, which suggests that NOPO levels must be tightly regulated (data not shown).

In Drosophila syncytial embryos, mitotic entry with incompletely replicated or damaged DNA triggers a CHK2-mediated protective mechanism known as centrosomal inactivation (Sibon et al.,2000; Takada et al.,2003). This damage-control system senses DNA defects and elicits localized changes in spindle structure that block mitotic progression,presumably to prevent the propagation of defective DNA. We previously reported that embryos from microcephalin (mcph1) females arrest in mitosis with acentrosomal, barrel-shaped spindles similar to those that we now observe in nopo-derived embryos(Rickmyre et al., 2007). We demonstrated that these mcph1 defects were suppressed by mutation of maternal nuclear kinase (mnk), also known as loki,which encodes Drosophila CHK2(Abdu et al., 2002; Brodsky et al., 2004; Masrouha et al., 2003; Xu et al., 2001). mnknulls exhibit increased sensitivity to ionizing radiation, but are viable and fertile. Suppression of mcph1 by mnk revealed that centrosomal inactivation significantly contributes to the mcph1phenotype.

To determine whether the mitotic defects of nopo-derived embryos are, like those of mcph1, due to CHK2-mediated centrosomal inactivation, we created lines doubly mutant for nopo and mnk. The nopo phenotype of acentrosomal, barrel-shaped mitotic spindles was strongly suppressed by mnk, as evidenced by the restoration of normal spindles with attached centrosomes(Fig. 3A-C; Table 2). DNA defects, however,were common in embryos from mnk nopo females, particularly during cortical divisions. We frequently observed abnormal DNA aggregates, some of very large size, shared by more than one spindle(Fig. 3D,E).

Like mcph1, we found that the developmental arrest of nopo mutants is suppressed by mnk(Fig. 3F,G; Table 2). In contrast to nopo-derived embryos, which arrest in syncytial divisions, most embryos from mnk nopo females complete syncytial divisions,cellularize, and arrest with aberrant morphology upon the initiation of gastrulation. Cellularized embryos from mnk nopo females contain unusually large DNA masses within irregularly sized cells compared with wild type. Thus, mnk suppresses the spindle/centrosomal defects and the developmental arrest of nopo mutants, but DNA defects appear to accumulate.

Based on these findings, we propose that nopo is required for the preservation of genomic integrity during syncytial embryogenesis. Lack of nopo activity leads to the occurrence of DNA defects, which then trigger CHK2-mediated centrosomal inactivation, thereby causing widespread mitotic arrest and the blockade of embryonic development. Mutation of mnk (Chk2) allows further nuclear divisions and developmental progression in nopo-derived embryos, despite the accumulation of extensive DNA defects that eventually lead to their arrest at the onset of gastrulation.

The DNA-replication checkpoint mediated by MEI-41 and Grapes, orthologs of ATR and CHK1, respectively, is developmentally activated in late syncytial embryos of Drosophila (Sibon et al., 1999; Sibon et al.,1997). Checkpoint activation, which may be triggered by titration of a maternal replication factor, leads to inhibitory phosphorylation of CDK1 and a gradual slowing of mitotic entry, presumably to allow sufficient time to complete DNA replication. At MBT (cycle 14), the first G2 gap phase is introduced. Embryos from mei-41 or grapes (grp)mutant females fail to lengthen the interphases of late syncytial cycles and are thought to enter mitosis without completing DNA replication; a secondary damage-control system, CHK2-mediated centrosomal inactivation, then becomes operational (Sibon et al.,2000; Takada et al.,2003).

Mitotic entry with incompletely replicated DNA can cause CHK2-mediated centrosomal inactivation in syncytial embryos(Sibon et al., 2000; Takada et al., 2003). Control mechanisms to ensure completion of DNA replication prior to mitosis may be particularly critical during rapid S-M cycles. Oscillating CDK1-Cyclin B activity plays a key role in coordinating these cycles(Edgar et al., 1994; Su et al., 1998). S-M transitions appear to be controlled by Cyclin B levels prior to cycle 10, and by both Cyclin B levels and a DNA-replication checkpoint in cycles 10-13(Ji et al., 2004; Sibon et al., 1997).

Using a previously described approach, we monitored the timing of nuclear envelope breakdown and reformation in cycles 11-13 by differential interference contrast (DIC) microscopy to test whether nopo mutants,like mei-41 and grp mutants, fail to lengthen interphase(see Fig. S1 in the supplementary material)(Rickmyre et al., 2007; Takada et al., 2007). Live imaging of nopo-derived embryos during cortical divisions was not feasible because the majority arrest either before or during these cycles(Table 1; data not shown); yolk proteins obscure the interior nuclei of precortical embryos, making imaging of these earlier divisions technically difficult. We analyzed the timing of cortical cell cycles in embryos from mnk nopo females (lacking a CHK2-mediated checkpoint) because they develop further than nopo-derived embryos; we reasoned that primary defects in cell-cycle kinetics due to nopo mutation would still be apparent. In support of this line of reasoning, embryos from mnk grp females have been shown to retain the cell-cycle timing defects of grp-derived embryos(Takada et al., 2007). Embryos from mnk nopo^Z1447 females exhibited significantly shorter interphases during cycle 11 (mean of 3.3 minutes) compared to wild type with mnk (4.8 or 5.4 minutes, respectively; Fig. 4A,B). Essentially identical results were obtained for mnk nopo^Z1447/Df(2R)Exel7153 and mnk nopo^Z1447/nopo^Exc142 females. Importantly,interphase 11 length was restored by transgenic rescue. A shortened interphase 11 (2.7 minutes) was observed in grp-derived embryos, as expected. Unlike grp, however, mnk nopo-derived embryos exhibited normal cycle 12 and 13 interphase lengths.

Based on our observations of shorter cycle 11 interphases in mnk nopo-derived embryos, we infer that interphases of earlier (precortical)syncytial cycles may be relatively short in nopo-derived embryos. We hypothesize that DNA replication is not completed during these truncated interphases, resulting in mitotic entry with unreplicated DNA, the triggering of CHK2-mediated centrosomal inactivation, and mitotic arrest with failure of further embryonic development.

Most embryos from nopo-null females arrest prior to the onset at cycle 11 of a detectable DNA-replication checkpoint effect(Table 1)(Crest et al., 2007). Thus, we reasoned that nopo is unlikely to regulate interphase length via mei-41/grp. Nonetheless, we tested the intactness of the MEI-41/GRP-mediated DNA-replication checkpoint by assessing levels of CDK1 inhibitory phosphorylation in mnk nopo-derived embryos and found them to be comparable to wild type (Fig. 4C). We also observed an intact DNA damage response by nopo larvae treated with hydroxyurea, a DNA replication inhibitor, or irradiation (see Table S1 in the supplementary material). We observed no genetic interactions between nopo and mei-41 (data not shown). We detected wild-type levels of Cyclin B and Cyclin A in mnk nopo-derived embryos and observed no genetic interactions between nopo and cyclin B (Fig. 4D; data not shown). These results suggest that noporegulates the S-M transition independent of the MEI-41/GRP-dependent checkpoint and mitotic cyclin levels.

To determine the subcellular localization of NOPO, we used transfected mammalian cells because our anti-NOPO antibodies did not work for immunofluorescence, and epitope-tagged forms of NOPO expressed via transgenesis were not stable in Drosophila embryos (data not shown). We transfected HeLa cells with fluorescently tagged versions of Drosophila NOPO and human TRIP (candidate homolog of NOPO) and assessed their localization by immunofluorescence microscopy. Whereas eGFP(control) was homogeneously distributed, eGFP-NOPO localized to nuclear puncta in the majority of interphase cells (compare Fig. 5A with 5B); a similar pattern was observed for Myc-tagged NOPO (data not shown). mCherry-TRIP also exhibited a punctate distribution in nuclei (Fig. 5D). Co-expression of eGFP-NOPO and mCherry-TRIP in HeLa cells confirmed their essentially identical localization patterns, underscoring the likelihood that NOPO and TRIP are functional homologs(Fig. 5C-E).

CREST staining of HeLa cells expressing eGFP-NOPO revealed that NOPO/TRIP localizes to nuclear regions distinct from the centromeres. To assess whether eGFP-NOPO localizes to nuclear puncta in a cell cycle-dependent manner, we immunostained transfected HeLa cells for PCNA or Cyclin A. We found that>99% of cells positive for eGFP-NOPO puncta were negative for insoluble PCNA foci in the nucleus, a marker of S-phase(Somanathan et al., 2001). By contrast, ∼97% of cells positive for eGFP-NOPO puncta were positive for nuclear Cyclin A, a marker of both S and G2 phases(Girard et al., 1991). Taken together, these data indicate that eGFP-NOPO specifically localizes to nuclear puncta in transfected HeLa cells during G2 phase.

The presence of a RING domain in NOPO suggested that it might function as an E3 ubiquitin ligase (Lorick et al.,1999). In a high-throughput yeast two-hybrid screen(Giot et al., 2003), NOPO interacted with an E2 ubiquitin-conjugating enzyme, Bendless (BEN), the Drosophila homolog of Ubc13(Muralidhar and Thomas, 1993; Oh et al., 1994; Zhou et al., 2005).

To confirm and extend these observations, we tested for interactions between combinations of wild-type and mutant NOPO and BEN proteins in a yeast two-hybrid assay (Fig. 6A). We used mutant NOPO and BEN forms encoded by nopo^Z1447 (E11K in the RING domain; Fig. 2A-C)and ben¹ (proline to serine change at position 97)(Muralidhar and Thomas, 1993). We found that both wild-type and mutant NOPO self-interact in this assay. When used as bait, wild-type BEN strongly interacted with wild-type NOPO; we occasionally observed weak interaction in the reverse direction (data not shown). By contrast, wild-type BEN and mutant NOPO do not interact, and mutant BEN interacts only marginally with wild-type or mutant NOPO.

We detected comparable levels of mutant and wild-type fusion proteins (both NOPO and BEN) in transformed yeast. Furthermore, nopo^Z1447-derived embryos have wild-type NOPO levels, and ben¹ ovaries have wild-type BEN levels(Fig. 2D; data not shown). Thus, the lack of two-hybrid interactions observed for mutant NOPO and BEN is likely to reflect changes in protein-protein interactions rather than decreased stability.

To obtain further evidence that NOPO and BEN interact, we compared the localization patterns of fluorescently tagged versions of these proteins in transfected HeLa cells. Both eGFP-NOPO and mCherry-TRIP accumulate in nuclear puncta (Fig. 5E, Fig. 6C). When transfected alone, eGFP-BEN distributes throughout cells with a perinuclear concentration but no obvious nuclear puncta (Fig. 6B). When eGFP-BEN and mCherry-TRIP were co-transfected, however,eGFP-BEN localized to nuclear puncta in the majority (56%) of cells positive for mCherry-TRIP nuclear puncta; furthermore, all puncta positive for eGFP-BEN were positive for mCherry-TRIP (Fig. 6D-F). These data suggest that NOPO/TRIP can recruit BEN/Ubc13 to chromatin, and provide further evidence for in vivo interactions between these proteins.

The E2 activity of Ubc13 has been shown in other systems to require heterodimerization with a UEV (ubiquitin-conjugating E2 enzyme variant) family member (Pickart, 2001). UEV proteins (Mms2p in budding yeast; Uev1A and Mms2 in mammals) resemble E2s but lack an active site cysteine (Broomfield et al., 1998; Sancho et al.,1998). Our BLAST searches revealed a single UEV homolog in Drosophila encoded by Uev1A. In our two-hybrid assay, UEV1A interacted strongly with wild-type and mutant BEN, but not with NOPO(Fig. 6A). Our attempts to detect BEN-NOPO complexes in Drosophila embryos were unsuccessful,however, possibly due to transience of this interaction (data not shown). Our yeast two-hybrid data suggest that the RING domain of NOPO interacts directly with BEN to promote the formation of a UEV1A-BEN-NOPO (E2-E3) complex.

ben was identified in a screen for Drosophila mutants with neuronal connectivity defects (Thomas and Wyman, 1982). Its yeast two-hybrid interaction with NOPO suggested that BEN might regulate embryonic development. We found that embryos from ben¹ homozygotes or hemizygotes fail to develop,revealing a new function for ben(Table 3).

Immunostaining of ben¹-derived embryos revealed that they arrest early in syncytial development with a small number (1-8) of mitotic nuclei (Fig. 7; Table 3). Most (72%) ben¹-derived embryos contain one acentrosomal spindle. In Drosophila females, meiotic spindles lack centrosomes, which are provided by sperm (Foe et al.,1993). Several lines of evidence suggest that ben¹ acentrosomal spindles are mitotic rather than meiotic. Their presence requires fertilization, and they are positioned deep within the egg interior where the first mitotic spindle resides (meiotic spindles are positioned near the cortex); furthermore, the presence of polar bodies indicates completion of meiotic divisions, and centrosomes are occasionally seen near the spindle (Fig. 7A-E; data not shown).

The spindle defects of ben¹-derived embryos strikingly resemble those of nopo mutants (compare Fig. 7H,I with 7G). ben¹spindles are often acentrosomal, barrel-shaped, variable in width, and have misaligned chromosomes; similar phenotypes were observed in ben¹ hemizygotes (Fig. 7J,K; Table 3). We were unable to test whether CHK2-mediated centrosomal inactivation causes mitotic arrest in ben-derived embryos because doubly homozygous adults were not viable. Taken together, the yeast two-hybrid interactions,co-localization, and similar mutant phenotypes that we have observed suggest that BEN-UEV1A and NOPO function together as an E2-E3 complex required to preserve genomic integrity during early embryonic development in Drosophila.

Because a given E2 can act in concert with multiple E3 ubiquitin ligases(Pickart, 2001), we sought to determine which activities of BEN are mediated by NOPO. We assayed our nopo mutants for four additional biological functions previously ascribed to BEN. The Drosophila giant fiber system (GFS) is a simple neural circuit that mediates an escape response to visual stimuli(Allen et al., 2006). Because ben is required for proper synaptic connectivity in the GFS, ben mutant adults fail to elicit a normal jump response to a light-off stimulus (Thomas and Wyman,1982; Thomas and Wyman,1984). ben mutant adults have also been reported to exhibit abnormalities in thoracic musculature and impaired mobility, and a role in innate immunity has been ascribed to ben(Edgecomb et al., 1993; Zhou et al., 2005).

We assessed the intactness of the GFS of nopo flies by testing their jump response to visual stimuli and found them to be defective, like ben flies (see Fig. S2A in the supplementary material)(Oh et al., 1994; Thomas and Wyman, 1982). Unlike ben, however, nopo flies exhibited normal mobility in a climbing assay and had normal sites of attachment of tergal depressor of the trochanter (TDT) thoracic muscles to the scutellum (see Fig. S2B,C in the supplementary material). We tested the innate immune responses of nopo and ben males and found that both exhibited slightly reduced levels of diptericin induction after infection (see Fig. S2D in the supplementary material). These results suggest that BEN and NOPO act as an E2-E3 complex that regulates GFS synapse formation and innate immunity,whereas BEN regulates mobility and muscle attachment sites via a different E3 ligase; thus, NOPO mediates a subset of BEN's functions.

We propose a model in which NOPO, a RING domain-containing protein,interacts with the BEN-UEV1A heterodimer to form a functional E2-E3 ubiquitin ligase complex required during syncytial embryogenesis for genomic integrity,cell-cycle progression, and the continuation of development(Fig. 8). In the absence of NOPO, a lack of ubiquitination of, as yet unidentified, NOPO targets results in the truncation of S-phase and/or spontaneous DNA damage. Mitotic entry with unreplicated and/or damaged DNA triggers the activation of a CHK2-mediated checkpoint that leads to changes in spindle morphology, mitotic arrest and failure of nopo-derived embryos to develop to cellularization.

We favor a model in which NOPO regulates the timing of S-M transitions in syncytial embryos to ensure that S-phase is of sufficient length to allow the completion of DNA replication prior to mitotic entry. The inhibition of DNA replication in syncytial embryos (e.g. via aphidicolin injection) leads to chromatin bridging in subsequent mitoses and CHK2 activation, both of which occur in nopo-derived embryos, presumably because of mitotic entry with unreplicated chromosomes (Raff and Glover, 1988; Takada et al.,2003). The mechanism by which NOPO coordinates S-M transitions is unknown. Our data suggest that nopo may alter the timing of these transitions independently of CDK1-Cyclin B, although localized changes in the levels and/or activities of these regulators not detectable by immunoblotting of whole-embryo lysates could play a crucial role. It is unclear why the MEI-41/GRP-dependent checkpoint, which appears to be functional in nopo-derived embryos, is not sufficient to slow mitotic entry.

The punctate nuclear localization observed for NOPO and its human homolog,TRIP, expressed in HeLa cells may indicate a direct role for these proteins in the regulation of chromatin structure. Furthermore, the G2 phase-specific localization that we observe for NOPO/TRIP in transfected HeLa cells may be consistent with a role for NOPO in slowing S-M transitions in syncytial embryos; in the absence of nopo, embryos that enter mitosis prematurely would probably do so without finishing DNA replication because of a lack of gap phases.

An alternative explanation for CHK2 activation in nopo-derived embryos is that they might incur elevated levels of spontaneous DNA damage. Syncytial embryos are considered to be unusual in that they activate CHK2 but not CHK1 in response to DNA-damaging agents(Fogarty et al., 1997; Sibon et al., 2000; Takada et al., 2007). Thus,spontaneous DNA damage would not be predicted to elicit the MEI-41/GRP-mediated replication checkpoint but would cause CHK2-dependent centrosomal inactivation during mitosis. Such a model would be consistent with the apparent lack of activation of the MEI-41/GRP-dependent checkpoint in nopo-derived embryos, although it would not explain why interphase 11 is shortened.

We previously reported that syncytial embryos from microcephalin(mcph1) mutant females undergo mitotic arrest with a phenotype similar to that described herein for nopo(Rickmyre et al., 2007). Like nopo, CHK2-mediated centrosomal inactivation causes mitotic arrest in embryos lacking mcph1. nopo and mcph1 are unique among maternal-effect lethal mutants in which CHK2-mediated centrosomal inactivation has been reported (e.g. grp, mei-41, wee1) in that their phenotypes appear to be more severe: centrosomes typically detach from spindles, and mitotic arrest occurs earlier, during precortical syncytial divisions(Rickmyre et al., 2007; Sibon et al., 2000; Stumpff et al., 2004; Takada et al., 2003). The underlying defects in nopo and mcph1 mutants may be distinct, however, because mnk mcph1-derived embryos exhibit normal cycle 11 interphase length, which is truncated in mnk nopo-derived embryos (Rickmyre et al.,2007). Furthermore, we did not detect a genetic interaction between nopo and mcph1 (J.L.R., J.A.M. and L.A.L.,unpublished).

Mammalian TRIP was identified in a yeast two-hybrid screen for tumor necrosis factor (TNF) receptor-associated factor (TRAF) interactors(Lee et al., 1997). TRAFs transduce signals from members of the tumor necrosis factor (TNF)/tumor necrosis factor receptor (TNFR) superfamily, which elicit diverse cellular responses in the immune and inflammatory systems(Hehlgans and Pfeffer, 2005). TRIP has been reported to inhibit TRAF2-mediated NFκB activation; the RING domain of TRIP, however, was not required for inhibition(Lee et al., 1997). By contrast, our analysis of nopo^Z1447 indicates that this motif is essential for NOPO function in Drosophila embryogenesis,probably by mediating its interactions with E2 components, as has been shown for other E3 ligases (Passmore and Barford, 2004). Drosophila Eiger (TNF ligand) and Wengen(TNF receptor) play roles in dorsal closure, neuroblast divisions, and the response to fungal pathogens (Kauppila et al., 2003; Schneider et al.,2007; Wang et al.,2006). A role for TNF signaling in early Drosophilaembryogenesis has not been reported to our knowledge.

TRIP was recently reported to be an essential factor in mice(Park et al., 2007). TRIP-deficient mice die soon after implantation as a result of defects in early embryonic development. Compared with wild-type littermates,TRIP^-/- embryos are smaller in size with a reduced cell number. TRAF2 does not appear to be required until later in development, suggesting that TRIP has TRAF2-independent roles in early embryos(Nguyen et al., 1999). It will be interesting to see whether mammalian TRIP, by analogy to Drosophila NOPO, is required for genomic integrity during embryonic development.

Our data support a model in which NOPO ubiquitin ligase acts in concert with BEN-UEV1A heterodimers to regulate Drosophila syncytial embryogenesis. The yeast two-hybrid interaction and co-localization of NOPO and BEN led us to identify an unanticipated role for BEN in early embryogenesis and additional roles for NOPO in synapse formation and innate immunity. Although the spindle defects of ben-derived embryos are strikingly similar to those of nopo mutants, they typically arrest earlier in syncytial development, suggesting that another E3 ligase that requires BEN may function in parallel with NOPO. Although nopo egg chambers appear normal, we have not ruled out a possible requirement for BEN-UEV1A-NOPO complexes during oogenesis; some defects in nopo- and ben-derived embryos could be a secondary consequence of previous defects during oogenesis.

K63-linked ubiquitin chains are thought to act as non-proteolytic signals(e.g. affecting protein localization and/or interactions), whereas K48-linked ubiquitin chains have established roles in targeting proteins for proteasome-mediated degradation (Pickart and Fushman, 2004). BEN-UEV1A E2 homologs in budding yeast(Ubc13-Mms2p) mediate K63-linked polyubiquitination of PCNA during postreplicative repair (Andersen et al.,2005). In mammalian cells, the E2 heterodimer Ubc13-Mms2 mediates DNA damage repair, while Ubc13-Uev1A promotes NFκB activation; both E2 complexes regulate these processes by mediating K63 ubiquitin chain assembly on target proteins. We propose that BEN-UEV1A-NOPO (E2-E3) complexes mediate the assembly of K63-linked ubiquitin chains on proteins that preserve genomic integrity in early Drosophila embryogenesis.