Microcephalin coordinates mitosis in the syncytial Drosophila embryo

Mutation in microcephalin (MCPH1) causes primary microcephaly (MIM#251200), an autosomal recessive disorder of human brain size in which brain volume is reduced to a third of normal, a size comparable with that of early hominids (McCreary et al., 1996; Wood and Collard, 1999; Woods et al., 2005). Comparative genomic sequencing has established that significant adaptive evolutionary changes have occurred in microcephalin in primates, in the lineage from the last common simian ancestor to man (Evans et al., 2004; Wang and Su, 2004). It has thus been suggested that changes in this gene could have contributed to the evolution of the human brain.

Microcephalin (MCPH1) is the first of four disease genes to be identified for this condition (Jackson et al., 2002) and encodes an 835 amino acid protein containing 3 BRCA1 C-terminal (BRCT) domains (Huyton et al., 2000). Such domains can bind other proteins in a phosphorylation-dependent manner (Yu et al., 2003), explaining their presence in many proteins involved in DNA repair and regulation of cell cycle (Huyton et al., 2000), processes dependent on phosphorylation signals. Subsequently microcephalin has been found to have multiple roles in cell cycle and DNA repair. Microcephalin has been implicated in the timing of cell cycle events and specifically in the regulation of chromosome condensation (Neitzel et al., 2002; Trimborn et al., 2004) and Cdk1 phosphorylation (Alderton et al., 2006). Both RNAi depletion and mutation of microcephalin result in a cellular phenotype of premature chromosome condensation (Neitzel et al., 2002; Trimborn et al., 2004). This is the consequence of premature onset of chromosome condensation, mediated by condensin II in G2 and delayed decondensation after mitosis (Trimborn et al., 2004; Trimborn et al., 2006), and correlates with the loss of inhibitory Cdk1 phosphorylation in S and G2 in microcephalin-deficient cells (Alderton et al., 2006). In the context of DNA repair, microcephalin has been found to be a chromatin-associated protein that also colocalizes with DNA repair proteins at ionizing radiation-induced DNA repair foci (Lin et al., 2005; Rai et al., 2006; Xu et al., 2004). It is required for intra-S and G2-M checkpoints (Lin et al., 2005; Xu et al., 2004), and acts downstream of Chk1 in the ATR signalling cascade regulating Cdc25A stability, and consequently G2-M transition (Alderton et al., 2006).

Three other primary microcephaly genes have been identified, ASPM, CENPJ and CDK5RAP2 (Bond et al., 2002; Bond et al., 2005). All their proteins have a centrosomal localization, and Drosophila orthologues that are required for centrosomal function: Abnormal spindle (Asp), Sas-4 and Centrosomin (Cnn), respectively (Basto et al., 2006; Gonzalez et al., 1990; Megraw et al., 1999; Riparbelli et al., 2002; Vaizel-Ohayon and Schejter, 1999). Asp is required for the integrity of centrosomes as microtubule-organizing centres, focusing the spindle poles at mitosis (do Carmo Avides and Glover, 1999). Cnn is essential for the localisation of other centrosomal components, gamma tubulin, CP190 and CP60 (Megraw et al., 1999; Vaizel-Ohayon and Schejter, 1999), whereas Sas-4 is necessary for centriole duplication, with mutant flies consequently lacking centrosomes, cilia and flagella (Basto et al., 2006). Despite these important cellular functions, all three mutants can generate morphologically normal adults, though sas-4 mutants die shortly after birth as they lack sensory neurons. asp and cnn have maternal-effect lethal alleles in which mitotic arrest occurs in embryonic cell cycles prior to cellularisation (Gonzalez et al., 1990; Megraw et al., 1999). Defects in localisation of asymmetric determinants have also been observed in cnn and sas-4 neuroblasts (Basto et al., 2006; Megraw et al., 2001).

Overall, there is a unifying theme for primary microcephaly of genes functioning in processes of cell division. It has therefore been proposed that reduced brain size is the result of reduced `neurogenic' mitosis, perhaps as a consequence of perturbed neural progenitor asymmetric cell division (Cox et al., 2006). However, microcephalin's function has appeared to be distinct from the other genes, and so we have generated a Drosophila model for microcephalin, to characterise its developmental role and to examine its mitotic functions. We report here the identification of Drosophila microcephalin (mcph1), establish that mcph1 mutants have a maternal effect lethal phenotype and determine that a short isoform of microcephalin is required for the co-ordination of syncytial centrosomal and nuclear division cycles.

The Drosophila melanogaster homologue of microcephalin was identified by TBLASTN (Altschul et al., 1990) searches of the Drosophila genomic and EST databases using the human protein sequence (CAC34661). We predicted a transcript, mcph1(L), corresponding to the human microcephalin protein that combines two adjacent annotated genes (CG8981 and CG30038) on chromosome 2R at 48C5 (Fig. 1). This transcript was confirmed by RT-PCR and sequencing of adult testis mRNA (Fig. S1, supplementary material). The nine exon mcph1 gene encodes a 1028 amino acid protein sharing 18% amino acid identity with human microcephalin (Fig. S2 supplementary material). Like its human counterpart, Drosophila MCPH1 has a unique pattern of BRCT domains, with a single N-terminal and tandem C-terminal BRCT domains, and is predicted to be a nuclear protein (PSORT II) (Nakai and Horton, 1999).

mcph1 has two sites of alternative splicing at intron 1 and 7, potentially resulting in four processed transcripts (Fig. 1A). The fully spliced transcript encodes the longest protein isoform, MCPH1(L), that contains all three BRCT domains (Fig. 1B,C). This transcript was experimentally validated as described above (Fig. S1 supplementary material). A second transcript, mcph1(S), encodes a short (S) variant of MCPH1 containing only one functional, N-terminal BRCT domain (Fig. 1B,C). This transcript contains an unspliced intron 7, resulting in a premature stop codon, and corresponds to the annotated CG8981-RB gene (FlyBase). We confirmed this transcript through re-sequencing of a full-length cDNA clone from the Drosophila Genomic Collection (LD43341, AY122187).

EST data also suggests the presence of other mcph1 transcripts with an unspliced intron 1 and an alternative translational start site (Fig. 1A) that results in a different, shorter N-terminal coding sequence. In combination with alternative splicing of intron 7, these would encode two further proteins. These proteins would have domain structures similar to MCPH1(S) and MCPH1(L), but be shorter by 47 amino acids at their N termini (one of these proteins is annotated on FlyBase as CG8981-RA). We experimentally confirmed the existence of unspliced intron 1 transcripts, which appear to be present in all developmental stages (data not shown). Unlike the intron 7 alternative splicing, the intron 1 splicing is not predicted to alter domain structure; we therefore focused our subsequent analysis on the `full length' MCPH1(L) and MCPH1(S) proteins.

RT-PCR experiments established that mcph1 transcripts are present at all developmental stages from embryo to adults (data not shown). Whole-mount in situ hybridisation of wild-type embryos was performed using a riboprobe common to both L and S transcripts. This detected mcph1 mRNA at its highest level in syncytial stages, distributed throughout the embryo. By the cellular blastoderm stage minimal mcph1 transcript was apparent, consistent with mcph1 being maternally derived until cellularisation and midblastula transition (Fig. 1D). During gastrulation, mcph1 transcript is again detectable, presumably as a result of zygotic expression of the gene.

To characterise MCPH1 protein expression, polyclonal rabbit antibodies were raised against an N-terminal fragment of the protein (Fig. 1C, antibody symbol). The affinity purified antibody detected a single specific 115 kDa band in immunoblots performed on extracts of syncytial embryos that is not present in embryos derived from mothers homozygous for a mcph1 deletion (mcph1^d2/d2), hereafter referred to as mcph1^d2/d2 embryos (Fig. 1E). This protein was also present in ovaries and at reduced levels in larval brain (Fig. 1F, and data not shown). This band corresponds to the predicted molecular mass of the L isoform; however, human microcephalin protein migrates significantly slower than predicted, so the 115 kDa band could represent a shorter isoform.

To determine which isoform is present in syncytial embryos, we cloned the MCPH1 S and L isoforms of the gene into pPMW (from Terence Murphy), an epitope-tagged (6xMyc) pUASP vector, using the Gateway system (Invitrogen). These transgenes were then expressed in embryos using the matαTub67::Gal4VP16 maternal driver and the protein mobility compared with endogenous protein (Fig. 1G). Correcting for the additional molecular mass of the epitope tag, the transgenic MCPH1(S) protein exhibits similar mobility to the endogenous embryonic protein, whereas the MCPH1(L) transgenic protein is substantially larger than endogenous protein. We also examined mcph1 transcripts by RT-PCR, and established that both spliced and unspliced intron 7 transcripts are present in embryos, whereas in testis only the fully spliced isoform is detectable (Fig. 1H).

Taken together, these data indicate that MCPH1(S), the shorter isoform of microcephalin is present in the syncytial embryo and ovaries, consistent with it encoding a maternally contributed protein present during syncytial embryogenesis. Though RT-PCR suggests that transcript for the MCPH1(L) form is also present in the embryo, the protein is not detected by antibody staining (Fig. 1E), suggesting that its expression may be subject either to post-transcriptional regulation (Maines and Wasserman, 1999) or rapid protein degradation.

Human microcephalin has been shown to associate with chromatin, and localise to DNA damage foci (Lin et al., 2005; Rai et al., 2006). As the Drosophila embryo presents an ideal opportunity to examine the localisation of MCPH1 during the cell cycle in vivo, we performed immunostaining on syncytial embryos with anti-MCPH1 affinity-purified polyclonal antibodies (Fig. 2A). Uniform nuclear staining was observed in interphase embryos, but was not detectable on the DNA and mitotic chromosomes from prometaphase to telophase. Nuclear staining was also present in yolk nuclei and pole cells (data not shown). Western blotting of staged cycle 8-10 embryos (Fig. 2B) revealed microcephalin to be present at a constant level throughout the cell cycle, suggesting that it was being redistributed from the nucleus rather than degraded. A small upward mobility shift was also noted at metaphase. Further investigation to identify what modifications, if any, are present at mitosis, will be useful to confirm this observation, and be of significant value in defining how microcephalin localisation at mitosis is regulated.

To confirm the finding of altered localisation in cell cycle, we cloned the L and S isoforms of mcph1 into the pPGW vector and generated UASP::GFP-mcph1 transgenic fly lines. Using a maternal αTub67::Gal4VP16 driver we then performed live imaging of (overexpressed) GFP-MCPH1 in wild-type syncytial embryos to confirm that MCPH1 varies in its localisation during cell cycle (Fig. 3A,B and see Movies 1 and 2 in supplementary material). Strikingly, though the two isoforms are both nuclear during interphase, they have different localisations during mitosis. The GFP-MCPH1(L) protein is rapidly lost at the onset of mitosis, reappearing as several foci, which then expand and adopt the shape of decondensing anaphase chromosomes (Fig. 3B). By contrast, the GFP-MCPH1(S) isoform appears to localise to the centrosomes and mitotic spindle, just prior to nuclear envelope breakdown, remaining centrosomally localised until chromosome decondensation at telophase (Fig. 3A).

Notably, endogenous, as well as overexpressed chicken Mcph1 has recently been reported to localise to the centrosome, the N-terminal BRCT domain being implicated in this localisation (Jeffers et al., 2007). A centrosomal localisation has also previously been suggested for human microcephalin (Zhong et al., 2006). Such a localisation is reminiscent of BRCA1, another BRCT domain containing protein (Hsu and White, 1998) as well as human CHK1, a direct interactant of microcephalin (Alderton et al., 2006), which negatively regulates CDK1–cyclin-B complex activity (Kramer et al., 2004).

Immunostaining of paraformaldehyde-fixed GFP-MCPH1(S) embryos confirmed that this protein is redistributed at mitosis (Fig. 3C,D,E). GFP-MCPH1(S) colocalises with DAPI staining during interphase, and is lost at the onset of chromosome condensation just after the appearance of phosphorylated histone H3 (P-H3). GFP-MCPH1(S) nuclear staining reappears at chromosome decondensation during telophase (Fig. 3E), closely correlating with loss of P-H3 staining. Localisation to the centrosome and mitotic spindle was not confirmed and could be due to limitations in detecting low levels of GFP-MCPH1 protein using anti-GFP immunostaining, or epitope masking in fixed preparations as previously described (Martinez-Campos et al., 2004; Raff et al., 2002). Further work will therefore be necessary to confirm the centrosomal localisation of MCPH1(S), which will be important in establishing its cellular functions.

We created deletions in the mcph1 gene by imprecise excision of a P-element located 50 base pairs upstream of the first mcph1 translational start site (P{GawB}NP6229, Kyoto). Two independent deletion mutants were identified by PCR screening of genomic DNA, and confirmed by sequencing the deletion breakpoints. mcph1^d1 has a 1.2 kb deletion involving exons 1-3 of mcph1, whereas mcph1^d2 is deleted for a 1.6 kb region containing exons 1-4 (Fig. 1A).

Both mcph1^d1/d1 and mcph1^d2/d2 were viable, with no morphological or behavioural phenotype evident in adult flies. In particular, we did not detect any alteration in size of the adult brain on histological examination (data not shown). While homozygous males were fertile, their female counterparts were infertile, even when crossed with wild-type males. Likewise, mcph1^d1/d2 and mcph1^d1/Df(2R) BSC39 were also viable with a maternal effect lethal phenotype.

Phase-contrast microscopy of eggs laid by mcph1^d1/d1 and mcph1^d2/d2 females only identified embryos at early developmental stages, with at most a narrow zone clearing of cytoplasm around the egg periphery without pole cell formation or cellularisation. This phenotype suggested an early defect in syncytial embryogenesis, resulting from loss of maternally contributed gene product.

mcph1^d1/d1 were crossed with UASP::myc-mcph1(S) and myc-mcph1(L) transgenic fly lines that express protein in embryos at comparable levels to endogenous protein when driven by αTub67::Gal4VP16 (see Fig. 1G). Gal4 driven UASP::myc-mcph1(S) rescued the female sterile phenotype in 21.8% of embryos, whereas very few embryos (2.6%) were rescued with the myc-mcph1(L) transgene (Table 1). In conjunction with the expression data, we therefore infer that loss of embryonic MCPH1(S) isoform is the cause of the mcph1 female sterile phenotype.

To characterise the phenotype further we performed a time course analysis of mitotic nuclei in early embryos visualised by anti-phospho-histone H3 antibody staining (Fig. 4A). mcph1^d1/d1 embryos showed marked slowing in nuclear division compared with wild-type embryos (Fig. 4A,B). Morphologically, they also exhibited an abnormal spatial distribution of nuclei compared with wild-type embryos (Fig. 4A). Mean terminal nuclear number in mutant embryos was 11.3 (±7.7; s.d.), and nuclei failed to migrate to the periphery to form the syncytial blastoderm. The terminal nuclei are in a metaphase-like state, have condensed fragmented DNA, broad mitotic spindles, and frequent separation of centrosomes from spindle poles (Fig. 4C,D). Numerous detached centrosomes were seen spread throughout the embryos indicating continued centrosomal duplication independent of nuclear division, a common feature of mitotic defects in Drosophila (Freeman et al., 1986; Raff and Glover, 1988). Monopolar (Fig. 4E) and multipolar spindles (Fig. 4F) as well as acentrosomal broad-based spindles (Fig. 4G) could be observed in the terminally arrested embryos. In view of the phenotype we therefore concluded that MCPH1(S) is essential for mitosis during early embryogenesis.

Multiple perturbations, such as blocked DNA replication (Raff and Glover, 1988), loss of centrosomal components (Megraw et al., 1999; Vaizel-Ohayon and Schejter, 1999), extensive DNA damage, and failure to degrade cyclins (Swan et al., 2005) can result in a metaphase-like state with detached centrosomes in the syncytial embryo. To understand further the underlying mechanism of the phenotype, we analysed mitosis by immunostaining of embryos 0-30 minutes after egg deposition (AED). We found morphological abnormalities apparent from cycle 1 in mcph1 mutant embryos (Fig. 5B) compared to control embryos (Fig. 5A).

The most striking abnormalities involved the centrosomes. Centrosomin signals were duplicated as early as prophase (Fig. 5C) suggesting the occurrence of premature separation of centrosomes (i.e. of centrioles and pericentriolar material). The occasional presence of additional centrosomin foci (prophase, Fig. 5B), also raised the possibility of the alternative explanation of centrosome overduplication.

Additionally, detachment of centrosomes from the spindle pole(s) was also frequently seen at metaphase 1 (26%, n=19) and onwards (Fig. 5B, anaphase). Increased α-tubulin localisation was also noted at the centrosomes, and astral microtubules, though fainter, were present. In cycle 1, tripolar and unipolar metaphase figures were also seen, presumably the consequence of centrosome separation and detachment. Similar abnormalities were seen in later cycles (Fig. 4E,F,G), and, from cycle 2 onwards, asynchronous mitotic divisions could also be observed.

Other, non-centrosomal abnormalities were also present. At pronuclear fusion and prophase in mutant embryos, chromosomes may be less compact than in wild-type embryos, relative to progress made in centrosome positioning and spindle formation in the respective embryos (Fig. 5A,B). However, at metaphase, mutant chromosomes were condensed to a similar extent as in wild-type embryos. Mitotic spindles at metaphase were irregular and less well formed in mcph1 mutants compared with simultaneously fixed wild-type embryos. During anaphase, there was occasional appearance of chromosome lagging, perhaps due to the less robust mitotic spindles. At telophase there was a less well formed midbody.

To obtain an accurate description of mitotic progression in mcph1 embryos, we quantified the relative frequency of mitotic phases in 0-30 minute AED embryos (Fig. 5D). We found that there was a disproportionate accumulation of mcph1 embryos in cycle 1 (29.3% mcph1, 12.0% Oregon, χ²-test P<0.001, n=601, n=656, respectively). In particular many more mcph1 embryos were in prophase 1 (9.3% mcph1, 4.6% Oregon, χ²-test P<0.001) and metaphase 1 (12.1% mcph1, 3.5% Oregon, χ²-test P<0.001).

Given that the short isoform of Microcephalin may localise to the centrosome during mitosis, we also investigated whether it is required for assembly of known components of the mitotic spindle poles and centrosomes. Asp, Cnn, and γ-tubulin all localised appropriately and at comparable levels to those in wild-type embryos (Fig. 4D,H,I). Thus we found no evidence to suggest a role for MCPH1 in the assembly of structural elements of the centrosome and/or spindle poles.

To investigate whether perturbation of key regulatory cell cycle proteins could explain the observed slowing of the nuclear cycle, we performed immunoblotting on 0- to 1-hour AED mcph1 embryos (Fig. 6). There was no accumulation of cyclins as has previously been reported as a cause of mitotic arrest in remnant mutant embryos (Swan et al., 2005), neither were there any abnormalities in total Cdk1 or Chk1 (also known as Grapes; Grp) protein levels. Reduced inhibitory phosphorylation on tyrosine 15 of Cdk1, similar to that seen in human patient cell lines (Alderton et al., 2006) was observed.

We report here that Drosophila MCPH1 localisation is mitotically regulated and that mcph1 mutant flies have a maternal effect lethal phenotype, as a result of mitotic arrest occurring in early syncytial cell cycles. We find that a short isoform of MCPH1 lacking C-terminal BRCT domains rescues this phenotype, indicating that this protein has an essential role in these embryonic cell cycles. GFP-MCPH1(S) localises to the centrosomes during mitosis, suggesting that it might have centrosomally related functions, relevant to the primary microcephaly phenotype. No centrosomal or spindle abnormalities during neuroblast mitosis have been observed in the larval brains of mcph1 mutants and the asymmetric determinants Inscuteable and Miranda also appeared to be appropriately localised (B.V. and A.P.J., unpublished data). Therefore, the centrosomal abnormalities appear to be specific to the early embryonic cell cycles. These data, along with the slowed rate of nuclear division in embryos (Fig. 4A), present from cycle 1 (Fig. 5D), lead us to conclude that the embryonic phenotype is the result of nuclear division cycle being slowed relative to the centrosome cycle. Additional centrosomes are consequently generated, perturbing subsequent nuclear divisions, leading to nuclear fragmentation and terminal mitotic arrest or catastrophe. In the syncytial embryo even a small reduction in nuclear division rate may have major effects, as cell cycle is very rapid, comprising rapid synchronous S and M phases without intervening G1 and G2 phases (Edgar et al., 1994). A subtle effect on cell division timing in all cells could therefore be sufficient to explain the essential requirement for MCPH1 in the syncytial embryos, but not in other cell cycles. Alternatively, MCPH1, and in particular the S isoform may have a specialised function, which is important for these cycles.

What causes the nuclear cycle to slow? The nuclear cycle is retarded prior to metaphase in cycle 1 (Fig. 5D). This might be because of prolonged mitotic entry, similar to the perturbed cell cycle regulation observed in human cells where there is an extended prophase-like state of “premature chromosome condensation” (Alderton et al., 2006; Trimborn et al., 2004). In this view, the loss of MCPH1 could affect the coordination of chromosome condensation with the rest of cell cycle entry. The larger, less compacted nuclei seen at pronuclear fusion and prophase could suggest that chromosome compaction or condensation might be delayed relative to other events at mitotic entry. However, live imaging of mutant embryos would be required to confirm this, but has not been technically possible in such early embryos. Immunofluorescence and live imaging studies of GFP-tagged and endogenous MCPH1 protein in wild-type embryos revealed an inverse correlation of MCPH1 localisation with mitotically condensed chromosomes, consistent with a role for this chromatin-associated protein (Lin et al., 2005) in maintaining decondensed chromosomes (Fig. 3C,D,E). Reduced Cdk1 inhibitory phosphorylation (P-Cdk1^tyr15) has been observed in S and G2 phases of human MCPH1 cell lines (Alderton et al., 2006), and suggested to underlie the early onset of chromosome condensation in human cells relative to other mitotic events. P-Cdk1^tyr15 is also reduced in mcph1^d2/d2 embryos (Fig. 6), which could explain the asynchrony of centrosomal and nuclear events observed. However, we cannot rule out that this is a secondary consequence of mitotic arrest in these embryos.

A second possibility is that S phase is prolonged in mcph1 embryos. Embryos in which DNA synthesis is blocked by aphidicolin injection, have similarly slowed cell cycles, with particularly slowed (paradoxically) M phase (Raff and Glover, 1988). Such embryos still undergo cycles of chromosome condensation and decondensation and exhibit centrosome overduplication. In Drosophila the ATR/Chk1 pathway prevents premature mitotic entry while S phase is still ongoing, only becoming developmentally required at mid-blastula transition when both ATR and Chk1 become essential (Sibon et al., 1999; Su et al., 1999). The mcph1 phenotype, however, is manifest much earlier than either of these mutants, inconsistent with its being initiated by defective ATR/Chk1 signalling. Additionally, though RNAi experiments have suggested that human microcephalin might transcriptionally regulate CHK1 (Lin et al., 2005; Xu et al., 2004), Chk1 protein levels are unaltered in mcph1 embryos (Fig. 6). Rather it seems more likely, that there is a separate function for Drosophila MCPH1 in cell cycle regulation, independent of its downstream role in ATR/Chk1 signalling, as has been previously concluded from analysis of human MCPH1 patient cell lines (Alderton et al., 2006).

Nuclear cell cycle can also be slowed if cyclins fail to degrade, and is the cause of mitotic arrest in the Cks30A mutant (Swan et al., 2005). A global accumulation of Cyclin A, B or B3 does not, however, occur in mcph1^d2/d2 embryos. (Fig. 6; K.B. and A.P.J., unpublished). In the embryo, cell cycle progression is regulated by localised Cyclin B degradation on the mitotic spindle (Huang and Raff, 1999). Degradation occurs from spindle poles inwards, and appears to be mediated by the centrosome (Wakefield et al., 2000). Therefore, both the poorly formed mitotic spindles and detachment of centrosomes observed in mcph1 embryos could also delay progression in the nuclear cycle through inhibition of local Cyclin B degradation. Consistent with this, Cyclin B immunostaining was observed on terminally arrested mutant nuclei (data not shown).

In summary, any of these abnormalities – localised failure in cyclin degradation, perturbed regulation of chromosome condensation, or prolonged DNA synthesis – could account for the mcph1 maternal effect lethal phenotype. Each would delay mitotic entry through slowing the nuclear cycle while the centrosomal cycle would continue unchecked, resulting in premature separation and detachment of centrosomes. Lost connections of centrosomes with spindles would then delay mitosis further, through impaired Cyclin B degradation, and insubstantial mitotic spindle formation. Further centrosome-nuclear incoordination subsequently would result in mitotic catastrophe and arrested embryo development.

An alternative, more speculative, interpretation is that the mcph1 phenotype has a centrosomal aetiology. Firstly, there is a correlation between phenotypic rescue with MCPH1(S), and the apparent centrosomal localisation of this isoform during mitosis. Furthermore, the early separation of centrosomes observed could reflect a requirement for microcephalin in maintaining the two centrioles together during mitosis and attachment of centrosomes to the spindle poles. Early separation and centrosome detachment would result in failure of localised Cyclin B degradation (Wakefield et al., 2000) and a consequently delayed nuclear cycle, culminating in mitotic catastrophe and arrested embryo development.

The cerebral cortex is reduced to a third of normal volume in primary microcephaly. The marked reduction in neural cell number is presumed to be caused by inefficient neurogenic mitosis (Woods et al., 2005). Primary microcephaly brain size is similar to that of early hominids, and the Darwinian adaptive evolutionary changes observed in primate microcephalin suggest that changes in this gene could have contributed to the evolutionary expansion of the human brain (Evans et al., 2004). Given the evolutionary scaling of the brain (Northcutt and Kaas, 1995), it is unsurprising that the invertebrate mcph1 fly brain is not grossly reduced in size. Nevertheless, our finding here that mcph1 is required for mitosis in the Drosophila embryo is pertinent to understanding the disease phenotype. Although a complete elucidation of the cell cycle function of MCPH1 will be necessary for unravelling its role in brain development and evolution, our finding here that the short isoform, lacking C-terminal BRCT domains, functions in coordinating mitosis is likely to be relevant to the pathogenesis of primary microcephaly. Moreover, as BRCT domains are important in phosphoserine/threonine-dependent protein binding (Manke et al., 2003; Yu et al., 2003) we would predict that this protein lacking functional C-terminal tandem BRCT domains will have significantly different functions from the full-length protein. In this context, it is notable that a recent analysis of MCPH1 in DT40 cells has demonstrated that the C-terminal BRCT domains are required for microcephalin localisation at ionising radiation-induced DNA damage repair foci (Jeffers et al., 2007). The fly, therefore, provides an ideal opportunity for analysis of the role of microcephalin in DNA damage repair, having physiological isoforms of microcephalin, with and without these C-terminal BRCT domains.

Two of the other primary microcephaly homologue mutants (asp, cnn) also exhibit maternal effect lethal phenotypes with mitotic arrest and free centrosomes (Gonzalez et al., 1990; Megraw et al., 1999; Vaizel-Ohayon and Schejter, 1999) and are therefore also required for mitosis in the early fly embryo. Furthermore, the mitotic centrosomal localisation of GFP-MCPH1(S) is intriguing, as it suggests that Microcephalin could co-localise with the other primary microcephaly proteins during mitosis. In contrast to the nuclear localisation generally reported for Microcephalin (Lin et al., 2005; Xu et al., 2004), all other primary microcephaly proteins, ASPM, CDK5RAP2 and CENPJ, and their Drosophila homologues (Asp, Cnn and Sas-4, respectively) are centrosomally localised (Basto et al., 2006; Bond et al., 2002; Bond et al., 2005; do Carmo Avides and Glover, 1999; Megraw et al., 1999). Therefore Drosophila MCPH1 could still act in the same biochemical pathway as other primary microcephaly orthologs.

In summary, we have shown that MCPH1 is a mitotically regulated protein, which co-localises with decondensed chromosomes, with a short isoform, MCPH1(S), localising to the centrosome and spindle during mitosis. MCPH1(S) is essential for the early rapid syncytial nuclear divisions, where it is required to co-ordinate centrosome and nuclear division cycles, prior to the physiological requirement for ATR-Chk1 signalling in the blastoderm embryo. The mcph1 fly therefore provides a relevant model for further genetic and biochemical characterisation of microcephalin's role in cell cycle regulation during development. Given that MCPH1 may co-localise with other primary microcephaly proteins at the spindle and centrosome during mitosis, this model also provides the opportunity to examine whether they may act in a common developmental and cellular pathway underlying the human phenotype.

Stocks were maintained on standard cornmeal Drosophila medium at 25°C. Wild-type flies were of Oregon^R strain. mcph1 deletion alleles were generated by imprecise excision of the isogenised P-element insertion line P{GawB}NP6229 (Kyoto), and maintained on a CyO, P{ActGFP}JMR1 balancer.

A DNA fragment, encoding the first 130 amino acids of MCPH1 was cloned into pGEX-6P1 (Amersham) and used to generate a N-terminal GST-fusion protein. The mcph1(S) transcript was excised from the LD43341 clone (DGRC, pOT2 vector) using NcoI and XhoI restriction sites. The NcoI and XhoI fragment was then directionally cloned (after filling in of the NcoI site) into SmaI, XhoI sites in the pGEX-6P1 vector to create pGEX6P1-MCPH1(S). A BglII-NotI digest was then performed to remove the 3′ end of this transcript, and after infilling of overhangs, the vector was religated to generate the final pGEX6P1-MCPH1 (1-130aa) construct. The vector was transfected into E. coli Rosetta (Novagen) and soluble GST fusion protein generated using standard methods (GST Gene Fusion Manual; Amersham Biosciences). After cleavage of the GST tag using Precision Protease (Amersham), the MCPH1(1-130aa) polypeptide was then used to generate polyclonal rabbit antibodies (Eurogentec). Antibodies were affinity purified from sera using the antigen coupled to HiTrap NHS-activated HP columns (GE Healthcare) following the manufacturer's protocol.

UASP-Myc-mcph1^Long, UASP-Myc-mcph1^Short, UASP-mcph1^Long and UASP-EGFP-mcph1^Short constructs were generated using the Gateway system (Invitrogen). mcph1 full length and CG8981 coding sequences were cloned by PCR amplification and recombination of the PCR products into pENTRY 221 (Invitrogen) using the following primers:

Forward primer for both isoforms (5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCACCATGGACAAGTTTCTGTTGCGc-3′), Long isoform reverse primer (5′-GGGGACCACTTTGTACAAGAAAGCTGGGTCCTATTTGTCCAGCACAAACGGTGGA-3′), Short isoform reverse primer (5′-GGGGACCACTTTGTACAAGAAAGCTGGGTCTCACTCCGTCGTATTACCAACCTTG-3′). The plasmid clone LD43341 (DGRC) and pGEX6P1-mcph1(L) were used as the PCR template for mcph1(S) and mcph1(L).

(pGEX6P1-mcph1(L) was generated by utilising an internal NheI restriction site within the mcph1 sequence of the pGEX6P1-MCPH1(S) plasmid to insert the longer alternative 3′ transcript end: primers D8F, CATCGAGCTCGGTGCAGAAG and D8R, GGTACTCTGTGTGTCCTAGC were used to amplify the 3′ end of mcph1(L) transcript from testis cDNA, and were cloned into pGEM-Teasy. From this a NheI and SpeI restriction fragment was then subcloned into pGEX6P1-mcph1(S), that had been digested with XhoI (filled in) and NheI to remove the 3′ end of mcph1(S).)

The resulting mcph1(S) and mcph1(L) pDonor vectors were then recombined into UASP N-terminal Myc and N-terminal EGFP destination vectors (pPMW and pPGW; Terence Murphy, Carnegie Institute, USA). Transgenic lines were generated by standard germ line transformation. Transgenes were expressed in embryos using the alphaTub67C::Gal4-VP16 maternal driver (w[^*]; P{w[+mC]=matalpha4-GAL-VP16}V2H, Bloomington #7062). All strains, unless otherwise stated were obtained from the Bloomington stock center. Others genotypes were generated by standard Drosophila genetics.

Embryos were processed as described previously (Davis, 2000). Embryos were imaged by time-lapse confocal microscopy using a Zeiss LSM 510 Meta scanning confocal system mounted on a Zeiss Axiophot II microscope with a ×63 objective Plan-Apochromat (1.4 NA). Three focal planes spaced by 0.5 μm were acquired every 12-15 seconds. All images shown are maximum intensity projections and were processed with Zeiss LSM 510 Images Browser (Zeiss).

Embryos from timed collections were dechorionated in 50% sodium hypochlorite, fixed in a two phase mixture of (1) heptane and methanol (1:1) containing 10 mM EGTA for α-tubulin, γ-tubulin, Asp and Cnn immunostaining; (2) heptane and 37% formaldehyde (1:1) for 5 minutes at room temperature for the detection of phosphohistone H3, GFP and MCPH1. Embryos were devitellinised in methanol, rehydrated in PBT (phosphate-buffered saline plus 0.1% Triton X-100) at room temperature and blocked in PBTA (PBS, 1% BSA and 0.1% Triton X-100) for 1 hour at room temperature. Primary antibody incubations were carried out overnight at 4°C. Subsequently, embryos were washed for 1 hour at room temperature in PBTA, and incubated for 1 hour with Alexa Fluor 488, Alexa Fluor 568 and 594 secondary antibodies (Molecular Probes) at 2 μg/ml in PBTA, washed for 1 hour at room temperature in PBT then mounted in Vectashield (Vector Labs) with DAPI at 5 μg/ml. Primary antibodies, rat anti-α-tubulin (1:10; YL1/2, Serotec), mouse anti-γ-tubulin (1:100; GTU-88, Sigma), mouse anti-GFP (1/200 JL-8, Clontech), rabbit anti-phospho-histone H3 (1:500; Upstate Biotechnology), rabbit anti-Cnn (1:100; a gift from T. Kaufman), rabbit anti-MCPH1 (1/10, this study), rabbit anti-Asp (1/25; a gift from D. Glover, Cambridge University, UK).

Single images or Z-series were collected on an Axiovert 200 microscope (Zeiss) with a 63× Plan-APOCHROMAT (1.4 NA) objective using IPLab 4.0 image capture software. Deconvolution was performed using Autoquant X (MediaCybernetics, Bethesda, MA).

0- to 16-hour embryos were collected on apple juice agar plates, dechorionated in 100% household bleach, devitellinised and fixed as described previously (Lehmann and Tautz, 1994). Embryos were fixed in 3.7% paraformaldehyde in PBS. Digoxigenin-labeled antisense and sense RNA probes corresponding to the GDC.11-D11 (LD43341, BDGP) embryonic cDNA clone sequence were prepared by in vitro transcription using digoxigenin-11-UTP labelling kit (Roche) with T7 and SP6 RNA polymerase respectively. Hybridisation to the fixed embryos was performed as described previously (Lehmann and Tautz, 1994).

Embryos were collected and dechorionated in 50% sodium hypochlorite as above. Dechorionated embryos were washed twice in 0.7% saline, 0.1% Triton X-100 and subsequently three times in embryo sample buffer [50 mM Tris-HCl pH 8.0; 150 mM NaCl; 0.5% Triton X-100; 1 mM EGTA, and protease inhibitor cocktail (Roche)]. Embryos were homogenised in 1.5 ml tubes in 1 volume embryo sample buffer using a pellet pestle homogeniser (Anachem, Luton, UK). The homogenate was cleared by repeated centrifugation for 10 minutes at 4°C at 13,000 g and stored at –20°C. Extracts were separated by SDS-PAGE and transferred to Hybond ECL membrane (GE Healthcare, Chalfont St Giles, UK). Primary antibodies used: rabbit anti-MCPH1 (this study, 1:1000), mouse anti-myc (clone 9B11, 1:2500, Cell Signaling), mouse anti-α-tubulin (clone B-152, 1/10000, Sigma), rabbit anti-cyclin A (Rb270, 1/3000, a gift from D. Glover), rabbit anti-cyclin B (Rb271, 1/3000, a gift from D. Glover), mouse anti-actin antibody (clone AC-40, 1:1000, Sigma), rabbit anti-chk1 (grp,1:500, a gift from T. T. Su, University of Colorado, Boulder, CO), rabbit anti-phosphoTyr15-cdc2 (P-Cdk1^tyr15, dilution1:1000, Cell Signaling), mouse anti-cdk1 (PSTAIR, 1:5000, Sigma). Secondary antibodies (horseradish peroxidise labelled, anti-rabbit, Sigma; anti-mouse, DAKO) were used at a 1:2500 dilution. Chemiluminescence detection was performed using ECL western blotting substrate or ECL Advance detection system (GE Healthcare). For ECL Advance system all primary antibody concentrations were decreased tenfold and secondary antibodies were used at 1:200.000.

Extracts from defined mitotic stages were prepared as previously described (Su, 2000) from syncytial embryos at division cycles 8-10. DNA was visualised by DAPI staining and for each cell cycle stage 15 hand selected embryos were electrophoresed and blotted.

RNA was isolated using Tri reagent (Sigma) according to manufacturer's instructions. Reverse transcription was performed using random primers and AMV reverse transcriptase (Roche). PCR was carried out using control primers for RP49 (Gabler et al., 2005): rp-49-F, TGTCCTTCCAGCTTCAAGATGACCATC and rp-49-R, CTTGGGCTTGCGCCATTTGTG. mcph1 primers were, exon 6: F, GCCATTAACCACAATAAGCCATCGG, and exon 8: R, CTCATAGGGCTCTTCATTGATCC. Thirty-two cycles of PCR were performed with an annealing temperature of 57°C.

RT-PCR primers for the mcph1(L) transcript fragments (supplementary material Fig. S1) were respectively:

Fragment 1F/R: D4F, GCAATAAGGAGCGCCAGAGTG and D4R, ACGCCCTCAGAGCGATTGTC, Fragment 2F/R: D5F, ATCATCACGATGGAAGAACGCA and D5R, CATTCTCGTATCCGTTGCAGGA, Fragment 3F/R: D6F, GCCAGCCACCAAACTACTCAA and D6R, CCTGGCCCTCGGAGCTAAGTT, Fragment 4F/R: D7F, TGATGGACATAAGTTCTCCTCG and D7R, TTGGATCCAGCCGCATGCCG, Fragment 5F/R: D8F, CATCGAGCTCGGTGCAGAAG and D8R, GGTACTCTGTGTGTCCTAGC.

We thank Vicky Thomas-McArthur and Debbie Sutton for technical assistance, Margarete Heck, Helen Strutt, Carl Smythe, Ian Adams, Nick Hastie, Veronica van Heyningen and other members of the CDBG and HGU for advice and helpful discussions. Elwyn Isaac for hosting K.B. in his lab during this work. S. DasGupta, S. Waddell and Laura Lee for sharing unpublished data. David Glover, Thomas Kaufmann, Christian Lehner, Jordan Raff, Trudy Schupbach, Tin Tin Su, Bill Chia, Bloomington stock center, and Developmental Hybridoma Bank for antibodies and fly stocks; Douglas Stuart and Sandy Bruce for assistance with illustrations. Paul Perry for imaging expertise. This work was supported by an MRC Clinician Scientist Fellowship (A.P.J.) and MRC Centre Development Grant (P.W.I.). K.B. is funded by a White Rose Studentship.