Maternal Phosphatase Inhibitor-2 Is Required for Proper Chromosome Segregation and Mitotic Synchrony During Drosophila Embryogenesis
Weiping Wang, Claire Cronmiller, David L. Brautigan


Protein phosphatase-1 (PP1) is a major Ser/Thr phosphatase conserved among all eukaryotes, present as the essential GLC7 gene in yeast. Inhibitor-2 (I-2) is an ancient PP1 regulator, named GLC8 in yeast, but its in vivo function is unknown. Unlike mammals with multiple I-2 genes, in Drosophila there is a single I-2 gene, and here we describe its maternally derived expression and required function during embryogenesis. During oogenesis, germline expression of I-2 results in the accumulation of RNA and abundant protein in unfertilized eggs; in embryos, the endogenous I-2 protein concentrates around condensed chromosomes during mitosis and also surrounds interphase nuclei. An I-2 loss-of-function genotype is associated with a maternal-effect phenotype that results in drastically reduced progeny viability, as measured by reduced embryonic hatch rates and larval lethality. Embryos derived from I-2 mutant mothers show faulty chromosome segregation and loss of mitotic synchrony in cleavage-stage embryos, patchy loss of nuclei in syncytial blastoderms, and cuticular pattern defects in late-stage embryos. Transgenic expression of wild-type I-2 in mutant mothers gives dose-dependent rescue of the maternal effect on embryo hatch rate. We propose that I-2 is required for proper chromosome segregation during Drosophila embryogenesis through the coordinated regulation of PP1 and Aurora B.

PROTEIN phosphatase-1 (PP-1) is a major protein Ser/Thr phosphatase that fulfills multiple cellular functions (Bollen and Stalmans 1992; Bollen 2001; Cohen 2002). PP1 is an essential gene necessary for anaphase entry because cells undergo metaphase arrest due to PP1 mutations or inhibition. PP1 is extraordinarily conserved among eukaryotes, and mammalian PP1 rescues mutants of the single GLC7 gene in yeast, demonstrating functional complementation across species. In Drosophila, there are four PP1 genes that are named according to their cytological locations and isotypes: PP1α13C, PP1α87B, PP1α96A, and PP1β9C (Dombradi et al. 1993). Amorphic mutations in PP1α87B are larval lethal, with mutant larvae showing delayed progress through mitosis, defective spindle organization, abnormal sister-chromatid segregation, hyperploidy, and excessive chromosome condensation (Axton et al. 1990; Baksa et al. 1993). Different PP1 isoforms display distinct tissue distribution and subcellular localization in various species. Intracellular distribution of PP1 involves binding to as many as 100 different regulatory subunits that target the PP1 holoenzymes to different locations, such as glycogen particles, microfilaments, centrosomes, and the nucleus (Cohen 2002; Ceulemans and Bollen 2004). PP1 regulatory subunits also affect catalytic activity and impart substrate specificity. Thus, there actually are dozens of PP1 holoenzymes in any cell to fulfill the multiple functions ascribed to PP1.

PP1 function also is regulated by the action of multiple heat-stable inhibitor proteins that show selectivity for different PP1 holoenzymes. For example, in vertebrates, myosin phosphatase (MYPT1-PP1) is selectively inhibited by the protein CPI-17, which does not inhibit other holoenzyme forms of PP1, such as glycogen-associated phosphatase (GM-PP1) (Eto et al. 2004). Phosphorylation of Thr38 in CPI-17 increases its inhibitory potency >1000-fold (Eto et al. 1999), and smooth muscle contracts in response to hormones that trigger CPI-17 phosphorylation in a process called calcium sensitization (Stevenson et al. 2004). Inhibitor-2 (I-2) is the most ancient of the PP1 inhibitor proteins and is conserved among all eukaryotes, from GLC8 in yeast to I-2 in Caenorhabditis elegans, Drosophila, Xenopus, and humans (Gruppuso et al. 1985; Roach et al. 1985; Tung et al. 1995; Li et al. 2007). Phosphorylation of I-2 by GSK3 was studied years ago for its effects on the PP1 catalytic subunit (see Ballou et al. 1985). Dephosphorylation of I-2 bound to PP1 produces phosphatase activation in biochemical assays, leading to the name “Mg-ATP-dependent phosphatase.” This suggests that under different conditions I-2 can either inhibit or activate PP1. This two-faced nature of I-2 resembles the family of RCN protein regulators of the phosphatase calcineurin (Hilioti and Cunningham 2003). I-2 has been found to preferentially associate with certain PP1 holoenzymes such as Nek2-PP1, Spinophilin-PP1, and KPI-2-PP1, but not myosin phosphatase (Eto et al. 2002; Terry-Lorenzo et al. 2002; Wang and Brautigan 2002). However, the physiological functions of I-2 that would require such conservation of structure and involve these specific PP1 holoenzymes remain unknown.

Genetic and biochemical evidence have suggested the involvement of PP1 and I-2 in cell cycle regulation, especially during mitosis. In Aspergillus, mutations in the PP1 gene produce cell cycle arrest at metaphase (Doonan and Morris 1989), implying that PP1 activity is low at metaphase, but required for the onset of anaphase. Phosphorylation of PP1 in a TPPR motif by CDK causes inhibition of the phosphatase activity during mitosis, first described for dis2 of Schizosaccharomyces pombe (Yamano et al. 1994). In budding yeast, mutations of the single PP1 gene GLC7 cause mitotic defects (Hisamoto et al. 1994). Yeast Ipl1 (Aurora B) suffer severe chromosome mis-segregation, which can be rescued by a specific mutation in Glc7 (Francisco et al. 1994). Alternatively, the Ipl1 phenotype can be rescued by overexpression of GLC8, the yeast homolog of I-2 (Tung et al. 1995). In mammalian cells, the expression level of I-2 fluctuates during the cell cycle, peaking at mitosis (Brautigan et al. 1990, 1991), and I-2 is phosphorylated by CDK-cyclinB1 at a conserved PXTP site during mitosis (Leach et al. 2003; Li et al. 2006, 2007). Finally, in Xenopus oocytes, I-2 and Aurora A together set the threshold for cyclin-B-dependent entry into mitosis (Satinover et al. 2006). Nonetheless, there is no genetic loss-of-function evidence to indicate any specific requirement for I-2 during mitosis.

Drosophila embryogenesis provides a good system for studying the role of I-2 during mitosis. The first 13 mitoses are synchronous cleavage divisions, and the division, migration, and spacing of nuclei in syncytial embryos are under the control of maternally provided protein and mRNA (Palter et al. 1979; Edgar and Schubiger 1986; Arbeitman et al. 2002). Importantly, there is a single I-2 homolog in the Drosophila genome. Dm-I-2 exhibits biochemical properties similar to mammalian I-2, including potent inhibition and activation of PP1 (Bennett et al. 1999; Helps and Cohen 1999). In some genetic backgrounds, overexpression of Dm-I-2 in flies can mimic PP1 loss-of-function phenotypes (Bennett et al. 2003). We describe here the first in vivo loss-of-function phenotype for I-2 in Drosophila. We show that maternal expression of the gene is required for embryonic development and that the consequences of reduced maternal I-2 can be traced to defects in chromosome segregation during mitosis in the early embryo.


Drosophila stocks and culture:

Flies were grown on molasses–cornmeal–yeast medium at 25°. I-2C363, Df(3L)AC1, and Df(3L)BSC46 stocks were obtained from the Bloomington Drosophila Stock Center. For generation of I-2 transgenic lines, the CaSpeR4 vector was a gift from Robert J. Duronio, and embryo injections were performed by Genetic Services. Ten independent I-2+ transgene insertions were generated in a w1118 background; the inserted chromosome and level of transgene expression were determined for each. The I-2 levels were unchanged or only slightly increased in 8 of 10 of the transgenic flies.

Antibody production:

Full-length recombinant Drosophila I-2 protein was used to generate anti-Drosophila I-2 polyclonal antibody in rabbits that was affinity purified against Drosophila I-2 protein coupled to Affigel-15.

Immunoblot analysis:

Western blotting was performed as described previously (Stefansson and Brautigan 2006) with the following primary antibodies (dilutions): rabbit polyclonal antiDM-I-2 (1:5000), chicken anti-pan PP1 (1:20,000), and mouse anti-β-tubulin (Developmental Studies Hybridoma Bank, University of Iowa) (1:1000). Goat anti-rabbit Alexa Fluor 680 was used at 1:8000 dilution. Goat anti-mouse IRDye 800 and anti-chicken IRDye 800 antibodies were purchased from Rockland Immunochemicals and used at 1:8000 dilution. Immunoblots were developed with a Li-COR Odyssey Infrared Imaging System (Li-COR Biotechnology).


Immunostaining and DAPI staining of ovaries or embryos were carried out as described previously (Cronmiller and Cummings 1993; Cummings and Cronmiller 1994). Anti-I-2 antibody was used at 1:1000 dilution. Anti-p(PXTP)I-2 antibody (Invitrogen) was used at 1:1000 dilution. FITC-conjugated anti-α-tubulin antibody (Sigma) was used at 1:200 dilution. Rhodamine-conjugated secondary antibody (Jackson ImmunoResearch Laboratories) was used at 1:2000 dilution. Wide-field images were obtained using a Nikon Eclipse E800 microscope equipped with a Hamamatsu 3580 camera using OpenLab software 3.0. Confocal images were obtained using an Olympus FluoView FV 1000 system.


We affinity purified an antibody specific for Drosophila I-2. Specificity was demonstrated by Western blotting extracts of Drosophila S2 cells, ovaries, early embryos, and whole adult female flies, which were compared to HeLa cells (Figure 1A). A protein of the expected size was the predominant band in all of the Drosophila samples; this band was absent from the HeLa cell extract, which has high levels of human I-2. The minor band with reduced mobility was attributed to phosphorylated I-2. We have previously shown that phosphorylated I-2 has reduced mobility in SDS–PAGE (Leach et al. 2003).

Figure 1.—

Endogenous I-2 in Drosophila ovaries and early embryos. (A) Western blot with affinity-purified antibody shows endogenous DmI-2 protein in extracts of Drosophila S2 cells, ovaries, 0- to 4-hr embryos, and whole adult females, compared to HeLa cells that have high levels of human I-2 that does not cross-react. Immunoblotting for β-tubulin (bottom) was used as a loading control. Lanes from a single immunoblot were cut and arranged together. (B) Dissected ovaries were fixed and stained with anti-I-2 antibody plus fluorescent secondary antibody (left, top and bottom) and DAPI to stain DNA (right, top and bottom) and images were acquired as described. (Bottom, left and right) The oocyte (“O”) is indicated with a line. Endogenous I-2 and DNA were stained in stage 1 early embryos (C); stage 2 embryos in interphase (D) and in mitosis (E) (inset box shown enlarged to the right); stage 3 embryos (syncytial blastoderm) in mitosis (F, top) and in interphase (F, bottom). (G) Confocal images showing endogenous I-2 (red) and α-tubulin (green) in the blastoderm embryo, with center Z sections (left) and peripherial Z sections (right). Bars, 50 μm.

This specific I-2 antibody was used to stain ovaries and early embryos (Figure 1). Both somatic and germline cells in the ovaries contained I-2, with the highest levels in germline cells. The protein was predominantly in the cytoplasm of the nurse cells in the ovary and excluded from the nuclei (Figure 1B). In more mature follicles, the appearance of a plume of staining in the central region of the oocyte was consistent with transfer of I-2 from the intensely stained nurse cells into the oocyte. The presence of I-2 in the germline was the first indication that this protein may have a maternal function. In early cleavage stage embryos (Figure 1C), there was uniform distribution of the endogenous I-2 protein. Later, in syncytial embryos, the level of I-2 protein increased and was uniformly distributed in the cytoplasm during interphase (Figure 1D), but was concentrated in dense clouds around condensed chromosomes during mitosis (Figure 1E). As in the earlier stages, I-2 was excluded from blastoderm nuclei during interphase and concentrated around their chromosomes during mitosis (Figure 1F). At the blastoderm stage, when the nuclei migrated and organized along the surface of the embryo, I-2 also concentrated with the nuclei at the embryo perimeter (Figure 1G). Consistent with the second band seen on the Western blot analysis, I-2 was phosphorylated in vivo in the conserved PxTP site, as detected with a phospho-site-specific antibody that cross-reacts with the conserved phospho-I-2 from many organisms (Figure 2). In blastoderm embryos, phospho-I-2 was detected as brightly stained clouds around mitotic chromosomes (Figure 2A), whereas staining of interphase nuclei was much weaker, consistent with phosphorylation of the protein primarily during mitosis. Indeed, in gastrulating embryos, elevated phospho-I-2 staining corresponded with the mitotic centers of the ectoderm (Foe and Alberts 1983; Hartenstein and Campos-Ortega 1985; Figure 2B). This visualization of phospho-I-2 showed in vivo phosphorylation of endogenous I-2 in mitotic cells.

Figure 2.—

Phosphorylation of endogenous I-2 at PXTP site in early Drosophila embryos. Early embryos were fixed and double stained with DAPI (blue) and anti-p(PXTP) I-2 antibody plus secondary antibody (orange). (A) Endogenous I-2 was phosphorylated around the mitotic spindles in blastoderms (bottom), but relatively unphosphorylated during interphase (top). (B) Embryos starting gastrulation have distinctive patterns of both mitotic and interphase cells on the surface, with the mitotic cells selectively stained due to phosphorylation of I-2 at the PXTP site. (Middle and bottom) Enlarged images of the area of the embryo marked by the red square in the top panel. Bar, 50 μm.

To identify an in vivo function for DmI-2, we characterized a loss-of-function allele. The Drosophila gene for I-2 (CG10574) is on chromosome 3, and we identified the mutant, I-2C362, with a 5500-bp PiggyBac transposable element inserted 81 nt upstream of the putative transcription start site for I-2. Genomic DNA from wild-type and I-2C362 mutant strains was used in PCR with a series of nested primers to confirm the location of the PiggyBac insertion (Figure 3, A and B). In addition, RT–PCR of mRNA isolated from the wild-type strain confirmed the unique transcription start site for the I-2 gene (data not shown). The I-2 cDNA from the I-2C362 mutant was sequenced, and no mutation was found in the entire coding region, so whatever protein was produced from this locus was wild type. These results confirmed that the PiggyBac transposon was inserted into the proximal promoter region of the I-2 gene, and we expected this would interfere with the expression of I-2 mRNA. To test this hypothesis, we performed real-time RT–PCR for I-2 and the nearest neighbor gene, cdk8, as a control, using cDNA prepared from wild-type and I-2C362 homozygous adults. There was no difference in the mRNA levels for either I-2 or cdk8 between wild-type and I-2C362 males (Figure 3C). However, there was a reduction in the levels of I-2 mRNA in I-2C362 females, compared to wild type (Figure 3C). We observed some increase in control cdk8 expression in I-2C362 females relative to wild type, showing that the PiggyBac at least did not reduce expression of cdk8. The decrease in mRNA for I-2 in the I-2C362 mutant relative to wild type was gender specific, observed only in females. We used immunoblotting to examine I-2 protein in ovaries dissected from wild-type and I-2C362 females and found that the I-2 level in mutant ovaries was reduced by 34%, relative to the wild type (Figure 3D). However, immunoblots showed that the whole-body levels of I-2 protein were not lower in either I-2C362 females or I-2C362 males, relative to wild type (Figure 3D). Thus, this PiggyBac insertional mutation has produced a weak hypomorphic I-2 allele that shows reduced expression primarily in the female germline.

Figure 3.—

Analysis of Drosophila I-2 gene and transposable element insertion. (A) Diagram mapping the location and structure of DmI-2 (CG10574) and neighboring genes on chromosome 3, left arm. The cytological map location of DmI-2 is 67C10. The black arrows indicate the coding regions and direction of the genes. The seven-digit numbers are gene sequence locations. In the mutant I-2C362, a 5.5-kb PiggyBac transposable element was inserted in the core promoter region of I-2, 81 bp upstream of the transcription start site. The bottom line shows the exon composition of I-2, with the putative transcription start site marked. F1–F6 and R1–R2 are forward and reverse primers used for PCR to identify the transcription start site and to confirm the insertion site of the transposon insertion mutational lesion of I-2C362. (B) Genomic DNA isolated from wild-type and I-2C362 mutant flies was used as PCR template to confirm the transposon insertion site with primers shown in Figure 3A. Lanes 1 and 2 used wild-type DNA, lanes 3 and 4 used I-2C362 mutant DNA with primer pairs F5 + R2 in lanes 1 and 3 and F4 + R2 in lanes 2 and 4. The rectangle on the gel marks the PCR products. (C) Real-time PCR for I-2 and the neighboring gene, cdk8, was performed using cDNA prepared from wild-type and I-2C362 homozygous adults. GAPDH2 transcript was used as a loading control to normalize transcript levels. Data are averages from two independent experiments. (D) Western blots showing endogenous DmI-2 protein in ovaries, adult female, and adult male, comparing wild-type and I-2C362 homozygous mutants with β-tubulin as the loading control.

We examined whether there were phenotypes associated with reduced levels of I-2. Homozygous I-2C362 flies were viable, with no apparent phenotype, and even with reduced levels of I-2 in the ovaries, female fertility appeared unaffected. To reduce I-2 expression levels further, we used a chromosomal deletion, Df(3L)AC1, and examined the I-2 protein in I-2C362/Df(3L)AC1 hypomorph hemizygotes. There were drastically reduced levels of I-2 in this genotype (Figure 4). These I-2C362/Df(3L)AC1 flies were viable and showed no defects other than a Minute phenotype that could be attributed to the unrelated Minute locus [M(3)i] uncovered by Df(3L)AC1 (Leicht and Bonner 1988; data not shown). The reduction of I-2 mRNA and protein in ovaries, together with the lack of any zygotic phenotype, led us to examine the maternal effect of reduced I-2 expression. A maternal function for I-2 was revealed by reduced viability of progeny from mutant females, primarily observed by a statistically significant reduction in hatching of embryos (Table 1). Homozygous I-2C362 had only slightly reduced embryonic hatch rates of progeny relative to sibling controls. However, the hatch rate from I-2C362/Df(3L)AC1 mothers was severely reduced (27%), relative to sibling controls (94%; P < 0.001). Furthermore, only 54% of hatched larvae from I-2C362/Df(3L)AC1 mothers survived to adulthood, another significant decrease (P < 0.001). The results indicate deleterious effects due to a reduction in germline levels of I-2. As a control, the same analysis was done with a second overlapping deletion, Df(3L0BSC46), which does not uncover I-2. The hemizygotes of this genotype did not show reduced embryo survival (Table 1).

Figure 4.—

I-2 levels in mutant ovaries, unfertilized eggs, and early embryos. Western blot shows reduced levels of endogenous I-2 protein in ovaries, unfertilized eggs, and 0- to 3-hr embryos, dissected from or produced by I-2C362/Df(3L)AC1 females in comparison with their I-2C362/+ female siblings. β-Tubulin was used as loading control.

View this table:

Effects of altered maternal I-2 expression on Drosophila embryonic survival

Embryos derived from I-2C362/Df(3L)AC1 mothers (hereafter referred to as mutant embryos) were fixed, stained with DAPI, and studied by fluorescent microscopy (Figures 5 and 6). A striking phenotype in early embryos was the abundance of interphase nuclei connected by DNA bridges (Figure 5A, left; note the dumbbell-shaped double nuclei), which were found in 13 of 39 syncytial interphase mutant embryos, but in only 1 of 36 control embryos. These were interpreted as cells with lagging chromosomes and incomplete anaphase. The nuclei in mutant embryos entered mitosis at different times, in contrast with the mitotic synchrony of wild-type embryos; as a result, mutant embryos had patches with nuclei at different stages of mitosis, from metaphase to anaphase (Figure 5A, right). Lagging chromosomes were visualized in several embryos. Comparison of blastoderm stage embryos from I-2 mutant and I-2C362/+ control sibling genotypes revealed patches of missing nuclei (Figure 5B, red circles). This phenotype was observed in 18 of 30 blastoderm embryos and not observed in any of 30 control embryos. In one extreme case, most of the nuclei were lost from the blastoderm, and only a couple of patches of nuclei remained (Figure 5B). When we examined the unhatched embryos from I-2 hypomorph hemizygous mothers, we found extensive, random loss of cuticular pattern (data not shown), consistent with this earlier nuclear loss.

Figure 5.—

Chromosome defects in I-2 mutant embryos. Early embryos were collected, fixed, stained with DAPI, and studied by fluorescent microscopy. (A, left) DNA bridges (arrowheads) were found in 13 of 39 syncytial I-2 interphase embryos, but in only 1 of 36 control embryos. (Right) During mitosis there was a loss of synchrony and presence of lagging chromosomes. Chromosomes of nuclei at prometaphase (bottom, square), early anaphase (bottom, hexagon), and late anaphase (bottom, circle) were found in the same embryo. Lagging chromosomes (white arrowheads) were also evident in the same region. (B) Blastoderm embryos from I-2C362/Df(3L)AC1 mothers displayed patches of missing nuclei (circles) (18 of 30 blastoderm embryos, bottom); embryos produced by their I-2C362/+ siblings did not (0 of 30 embryos, top). Bar, 50 μm.

Figure 6.—

Confocal images of I-2 mutant syncytial embryos show loss of mitotic synchrony. Mutant embryos were double stained for α-tubulin (green) and DAPI (magenta). (A) Loss of synchronization was revealed across the whole embryo. Representative regions across the anterior–posterior axis (squares) are enlarged to show metaphase (B) and early anaphase with a misaligned chromosome (C, arrowhead) and late anaphase with microtubules in the midbody region between separated chromosomes (D and E).

Double staining of mutant embryos for α-tubulin and DAPI highlighted regions of localized mitotic synchrony, even though there was an overall loss of synchrony across the embryo (Figure 6A). Metaphase spindles were seen in the anterior region (Figure 6B), in contrast to anaphase nuclei in the posterior region of the same embryo (Figure 6, D and E). Mis-segregated chromosomes were visualized in early anaphase nuclei (Figure 6C). The loss of synchrony was found in 6 of 50 syncytial mutant embryos, but in none of 54 control embryos. These results show that reduced levels of maternally expressed I-2 caused defects in both chromosome segregation and mitotic synchrony during the cleavage stages of Drosophila development.

We used transformation rescue to confirm that reduction of maternal I-2 was specifically responsible for the embryonic phenotypes. The genomic DNA construct for genetic transformation included the I-2 gene, plus 750 bp of the surrounding 5′ and 3′ sequences (Figure 3A). Several independent I-2 transgenic lines were created and assayed for I-2 protein levels in the flies and for rescue of the mutant maternal-effect phenotype. Line WW5 had the highest levels of whole-body expression of I-2 protein, whereas line WW4 had whole-body levels of about the same as the endogenous I-2 in wild type (data not shown). In the mutant background, introduction of a single copy of the WW5 transgene showed a partial rescue of the embryo hatch-rate mutant phenotype, whereas the WW4 transgene did not (Table 1, D and E). We examined whether introducing a second WW5 transgene would further increase the hatch rate, and it did (Table 1, D). Although the whole-body levels of I-2 were elevated relative to wild type by one or two copies of the WW5 transgene, the I-2 expression level in ovaries was below wild type. There was indeed a stepwise increase in the level of the I-2 protein in ovaries of I-2 hypomorph hemizygous females with one copy vs. two copies of the WW5 transgene insertion (Figure 7). However, even with two copies of this transgene in the I-2 mutant background, the I-2 protein levels in the ovaries were still well below wild-type levels (Figure 7). This partial restoration of the levels of ovarian I-2 was consistent with partial rescue of the maternal-effect phenotype on embryonic hatch rate. These results establish a functional dose-dependent rescue of the mutant phenotype by transgenic I-2.

Figure 7.—

Expression of transgenic I-2 to rescue the maternal-effect phenotype. Female flies of 10 independent I-2+ transgenic lines were assayed for I-2 protein levels by immunoblotting, using β-tubulin and PP1 as loading controls. The transgenic line WW5 showed the highest levels of I-2 protein among the 10 transgenic lines, and the WW5 transgene (P{I-2+}) was bred into the I-2C362/Df(3L)AC1 hypormorphic background. Ovaries from I-2C362/Df(3L)AC1 sibling females with no, one, and two copies of the WW5 I-2+ transgene were dissected and subjected to immunoblotting to compare I-2 protein levels, with β-tubulin as the loading control. The blot shows increased I-2 levels that correlate with the number of transgene copies; however, even in the mutant supplemented with two I-2+ transgenes, the I-2 protein levels were still lower than that in wild type on the basis of quantitation normalized by β-tubulin levels. The relevant genotypes of the samples were (from left to right): I-2C362/Df (3L)AC1 (mutant + 0 transgene), P{I-2+}/+; I-2C362/Df(3L)AC1 (mutant + one transgene), P{I-2+}/P{I-2+}; I-2C362/Df(3L)AC1 (mutant + two transgenes), and wild type.


Using Drosophila, we provide the first evidence of an in vivo function for phosphatase I-2 in a metazoan organism. Although I-2 was discovered >30 years ago as an inhibitor of PP1 (Huang and Glinsmann 1976) and its biochemical properties as an inhibitor and component of MgATP-dependent phosphatase have been studied intensively, the physiological function of I-2 has remained unknown and the subject of speculation. Because the presence of multiple pseudogenes and analogs for I-2 in the mouse (and human) complicates knockout of the gene, we chose to use Drosophila to examine the I-2 function in vivo, producing a mutant with reduced expression of I-2. The severely reduced survival of embryos from this genotype indicates that I-2 is required for early development in Drosophila. Further, we show that I-2 is provided from the maternal genome for early embryonic cell cycles. Mutant syncytial embryos showed abundant DNA bridges during interphase and the presence of lagging chromosomes during mitosis, suggesting a role for I-2 in chromosome segregation. This could involve I-2 regulation of steps for chromosome alignment during metaphase and/or separation during anaphase. In addition, I-2 is necessary to maintain mitotic synchrony in syncytial embryos, and the loss of synchrony may arise from faulty chromosome segregation. Transgenic expression of I-2 revealed a dose-dependent partial rescue of the embryonic lethality, indicating that the phenotypes observed were due to, and even proportional to, the reduced levels of I-2 protein. These data indicate a physiological role for I-2 during mitosis that is consistent with the essential but undefined action of PP1 for metaphase-to-anaphase transition in all eukaroytes.

It is our hypothesis that the high levels of maternal I-2 expression depend on both proximal promoter sequences (that are in the transgenes) and other enhancer elements (not in the transgene constructs). This situation would account for a difference in the expression of endogenous I-2 in somatic and germline cells, but not for a difference in expression from the transgenes, leaving the ovaries with low levels of I-2 protein relative to wild type, even with two transgenes. On the other hand, this limited maternal expression of the transgene gave us a good chance to show a dose-dependent rescue of the mutant maternal effect on embryonic survival, which convincingly links the genotype to the phenotypes.

The phenotypes of I-2 mutant embryos resemble those associated with deletion or mutation of other Drosophila mitotic regulators. Mutant alleles of PP1α87B show aberrant mitosis, defective spindle organization, abnormal sister-chromatid segregation, hyperploidy, and excessive chromosome condensation, with germline transformation with wild-type PP1α87B restoring normal mitosis (Axton et al. 1990; Baksa et al. 1993). BubR1 is a checkpoint protein that is encoded by a maternal-effect gene in Drosophila. In striking similarity to the phenotype of the I-2 mutant embryos, a BubR1 hypomorph also exhibits loss of synchrony in syncytial division and defects in chromosome segregation (Perez-Mongiovi et al. 2005). Prat and Prat2, which are involved in de novo purine synthesis, produce maternal-effect phenotypes that include loss of mitotic synchrony and segmentation defects in embryos. These results reinforce the idea that chromosome segregation defects lead to loss of mitotic synchrony and further suggest that purines are limiting during embryonic development (Malmanche and Clark 2004). Finally, in Drosophila S2 cells, RNA interference knockdown of Aurora B produces chromosome mis-segregation and cytokinesis defects, resulting in polypoidy (Giet and Glover 2001). Given the biochemical role of I-2 in regulating PP1, we propose that unbalanced Ser/Thr phosphorylation during mitosis may be a common cause of these shared mutant phenotypes.

I-2 function is required in embryos for mitotic chromosome segregation. The mis-aligned and lagging chromosomes during anaphase and the DNA bridges during interphase that we found associated with loss of maternal I-2 appear to be due to failed chromosome segregation. Chromosome segregation involves assembly of kinetochores at centromeres during prometaphase, followed by microtubule attachment to the kinetochores. Initially, there are many improper merotelic and syntelic microtubule attachments that need to be resolved for accurate congression of the chromosomes to the metaphase plate (Walczak and Heald 2008). Faulty kinetochore–microtubule attachment or incomplete resolution of asymmetric attachments can result in chromosome mis-alignment and lagging chromosomes that lead to DNA bridges during interphase. Such chromosome segregation defects would activate cell cycle checkpoints that act only during later embryonic divisions, thereby giving rise to a loss of synchrony without complete cell cycle arrest at the syncytial blastoderm stage. In Drosophila embryos, severely defective nuclei ultimately undergo a process of removal from the cortical region of the embryo proper. It is likely that this process produced the barren patches devoid of nuclei that we observed in I-2 mutant embryos.

One clue to the specific biochemical function of I-2 during chromosome segregation comes from genetic studies of the I-2 homolog in yeast, Glc8. Glc8 mutants suppress mutations in ipl1, which encodes the Aurora kinase in yeast (Tung et al. 1995). In metazoans, there are two Aurora kinases, A and B, with Aurora B thought to function equivalently to yeast Ipl1. Aurora B is critical for mitosis because it phosphorylates proteins that are required for assembly of kinetochores and resolution of abnormal kinetochore–microtubule attachments (Vader et al. 2006). PP1 dephosphorylates and inactivates Aurora B kinase and also dephosphorylates Aurora B substrates in counterbalance to Aurora B (Sugiyama et al. 2002). By inhibiting PP1, I-2 would indirectly activate Aurora B, as well as enhance phosphorylation of Aurora B substrates. On the other hand, depletion of I-2 would tip the balance of Aurora B vs. PP1 in favor of the phosphatase, thus interfering with Aurora B function and leading to defects in chromosome segregation. In yeast, the Ipl1 kinase and GLC7 phosphatase phosphorylate and dephosphorylate a common substrate, Dam1, which is localized at kinetochores (Pinsky et al. 2006). Our hypothesis predicts that loss-of-function Aurora B would have the same phenotype as the I-2 mutant. Unfortunately, no Aurora B mutants have been reported in Drosophila. However, the results of both Aurora B knockdown in Drosophila S2 cells and chemical inhibition of Aurora B in human tissue culture cells support our hypothesis: Each causes chromosome mis-segregation and polyploidy (Giet and Glover 2001; Hauf et al. 2003). We have observed similar defects following small interfering-RNA-mediated knockdown of I-2 in human cells (our unpublished results).

Finally, our model in which I-2 enhances Aurora B by inhibition of PP1 can also account for the spatio-temporal changes in both Aurora B and PP1 during mitosis. Aurora B is localized at centromeres and becomes activated during prophase and inactivated during anaphase. Conversely, PP1γ localizes to kinetochores (Trinkle-Mulcahy et al. 2003), and PP1 activity is low until metaphase and then increases for entry into anaphase (PP1 mutants arrest at metaphase). Part of the temporal change in I-2 function is explained by the cell cycle fluctuation in the levels of I-2 protein, which peak during mitosis (Brautigan et al. 1990). In addition, phosphorylation of I-2 at Thr-72 in the PXTP motif by CDK2-cyclinB peaks early during mitosis (Li et al. 2006), and dephosphorylation of I-2 by PP1 triggers conversion of PP1 from an inactive to an active conformation (Li et al. 1985). This mechanism has been studied for decades with purified I-2 and PP1 and may account for the activation of PP1 at the metaphase–anaphase transition of mitosis. Thus, the duality of I-2 as both inhibitor and activator of PP1 may be essential for its physiological function in regulating chromosome segregation during mitosis.


We thank Robert J. Duronio for the CaSpeR transgenic vector and P. Todd Stukenberg for discussions. This research was supported by grant GM56362 (to D.L.B.) from the National Institute of General Medical Sciences, National Institutes of Health, United States Public Health Service.


  • Communicating editor: K. G. Golic

  • Received May 27, 2008.
  • Accepted June 3, 2008.


View Abstract