|Home | About | Journals | Submit | Contact Us | Français|
In the Drosophila early embryo the centrosome coordinates assembly of cleavage furrows [1–3]. Currently, the molecular pathway that links the centrosome and the cortical microfilaments is unknown. In centrosomin (cnn) mutants, where the centriole forms but the centrosome pericentriolar material (PCM) fails to assemble [4, 5], actin microfilaments are not organized into furrows at the syncytial cortex . While CNN is required for centrosome assembly and function [4, 6, 7], little is known of its molecular activities. Here, we show the novel protein Centrocortin (CEN), which associates with centrosomes and also with cleavage furrows in early embryos, is required for cleavage furrow assembly. CEN binds to CNN within CNN Motif 2 (CM2), a conserved 60 amino-acid domain at CNN’s C-terminus. The cnnB4 allele, which contains a missense mutation at a highly conserved residue within CM2, blocks the binding of CEN and disrupts cleavage furrow assembly. Together, these findings show that the C-terminus of CNN coordinates cleavage furrow formation through binding to CEN, providing a molecular link between the centrosome and cleavage furrow assembly.
The mutant protein encoded by cnnB4 (Figures 1A and S1A) was reported to localize to centrosomes yet the mutation was maternal effect lethal and embryos were deficient in cleavage furrow assembly . This finding indicated that the conserved domain at the carboxyl-terminus of CNN is necessary for a critical function of centrosomes: the organization of actin into cleavage furrows. We examined the centrosomes in cnnB4 embryos and neuroblasts to assess the microtubule-organizing center (MTOC) activity, and also to re-examine CNN localization and actin furrow formation in this point mutant. We found that CNNB4 was expressed in embryos at levels similar to wild type (Figure 1B) and localized to centrosomes as previously reported , but had a looser centrosome association compared to wild type CNN (Figures 1C and E, and S3–S5). Regardless of CNNB4 mutant protein localization to centrosomes, cnnB4 embryos exhibit the linked spindles phenotype characteristic of cnnhk21 (null) embryos (Figure 1D and E), a further indication of defective cleavage furrows. Furthermore, in cnnB4 embryos some PCM markers had a looser association with the centrosome in comparison to wild type (Figures S3-S5). Despite this PCM defect, cnnB4 centrosomes were very efficient MTOCs, producing robust astral microtubules similar to wild type (Figures 1E, S2 and S4).
To assess the kinetics of microtubule assembly from cnnB4 centrosomes, we performed a microtubule regrowth assay in larval neuroblasts (Figure S2). The cnnB4 centrosomes assembled astral microtubules at prophase rapidly, similar to wild type. In contrast, no astral microtubules assembled in cnnhk21 mutant neuroblasts, where the MTOC activity of the centrosome is severely deficient . Moreover, the microtubule nucleation and assembly factors γ-Tubulin, D-TACC and Msps were localized to cnnB4 centrosomes (Figures S3–S5). However, CNNB4 localization is more disperse than wild type, and the level of CNN signal at these centrosomes was about 28.8% higher compared to wild type (Figure S3). The dispersed CNNB4 PCM may expose more surface area for antigen sites upon antibody staining, making the level of CNN at centrosomes appear artifactually higher in cnnB4 embryos. Nevertheless, it appears that the level of CNNB4 is not less and may even be greater than wild type CNN at centrosomes. In contrast, the level of γ-Tubulin at cnnB4 centrosomes was ~30.5% less compared to wild type (Figure S3).
Central spindle and astral microtubules are important for furrow assembly at cytokinesis [8, 9]. Therefore, the role of the centrosome in actin organization at the embryonic cortical membrane may rely on the MTOC activity at centrosomes. However, despite the proficient MTOC activity at cnnB4 mutant centrosomes, furrow assembly was severely deficient (compare Figure 1F and G). Thus, MTOC activity per se is insufficient for the proper assembly of actin into furrows at mitosis, a finding consistent with previous experiments indicating that MTOC and actin-organizing activities may be partly separate activities for the centrosome . However, since microtubules are required at anaphase for cleavage furrow assembly in the following prophase , we cannot exclude the possibility that qualitative features of the microtubules or alterations in their dynamics are altered at cnnB4 centrosomes. Furthermore, in contrast to cnnhk21 null embryos, which lack any perceptible actin organization at the cortex [6, 12], cnnB4 centrosomes were partially competent to organize actin into cleavage furrows, attesting to the hypomorphic nature of this mutation.
The cnnB4 point mutation resides in CM2, the conserved C-terminal 60 amino-acid domain of CNN (Figures 1A and S1A). Thus, CM2 is essential for CNN function in vivo, specifically for cleavage furrow assembly. How centrosomes coordinate furrow assembly with the cleavage cycle at the cortex is unknown, but we reasoned that proteins that associate with CM2 may mediate this function. Therefore, we screened for interacting partners of CM2 in order to discern the molecular signal conveyed from the centrosome to the cortex. By using CM2 (the C-terminal 67 amino acids of CNN) as bait in a yeast two-hybrid screen, we identified one interacting partner encoded by a cDNA for the CG1962 gene, which we hereafter refer to as centrocortin (cen) because of the association of CEN protein with the centrosome and the cortex (see below). In the two-hybrid assay, the CEN-CNN interaction was reduced when the CNNB4 protein was tested (Figure 2A,B), suggesting a functional link between CEN-CNN association and CM2 function.
cen is an uncharacterized gene with orthologs in mammals including two in humans: cerebellar degeneration related-2 (Cdr2) and Cdr2-like (Cdr2L)/Cdr3 (Figure 2C and S1B). Cdr2 is an autoimmune antigen targeted by “anti-Yo” antibodies associated with paraneoplastic cerebellar degeneration . CEN is a 790 amino acid protein that shares a highly conserved 110 amino acid domain near its amino-terminus that is similar to Cdr2L and Cdr2 (Figures 2C and S1B) and to orthologs found throughout metazoans (not shown). Cdr2 is reported to bind to the transcription factor Myc, sequestering it in the cytoplasm, through the association of the helix-leucine zipper motifs of Cdr2 and Myc . CEN, however, does not contain a predicted leucine zipper in the conserved domain, appearing more similar to Cdr2L in this regard, and therefore may not share the Myc-binding property of Cdr2. Other than Myc inhibition, the function of Cdr2 remains largely unknown, and no activity has been attributed to Cdr2L.
To further verify the interaction between CNN and CEN we performed co-immunoprecipitation from embryo and S2 cell lysates. We detected a weak association between CNN and CEN with this assay, but this was inefficient regardless of varied attempts and buffer conditions (Figure S6). We suspect that the major association between these partners occurs at embryonic centrosomes (see below), which complicates co-IP analysis since centrosomes, which are large organelles, typically pellet at low g forces during the preparation of lysates. Thus, we were unable to detect a strong association between CNN and CEN by IP.
We raised antibodies against the amino and carboxyl ends of CEN in order to examine its expression pattern and localization in embryos. CEN is expressed at high levels in ovaries and early embryos and then drops to very low levels in late embryos and in larval stages (Figure S7A). We also examined the localization of CEN during early embryogenesis to determine if it colocalizes with CNN at centrosomes. To accomplish this, we used affinity-purified anti-CEN antibodies, and also examined the localization of green fluorescent protein (GFP)-tagged CEN fusion proteins expressed from transgenes. Immunofluorescent staining revealed that CEN is localized to centrosomes in a unique and dynamic pattern. CEN localized to a structure adjacent to centrosomes, with some overlap with CNN at centrosomes (Figures 2D and E). At late syncytial cleavage cycles, beginning at cycle 10, CEN localized to a discrete particle between centrosome pairs at interphase or early prophase (Figure 2D). Upon centrosome separation at mitosis, the major CEN puncta segregated asymmetrically with one of the two centrosomes. At mitosis, the CEN dot associated with one centrosome per mitotic spindle. The signal for CEN at this pericentrosomal particle was dispersed at telophase (Figure 2D). The majority of mitotic spindles showed this asymmetric pattern of CEN localization, however, some figures showed localization to both centrosomes, while, in others, neither harbored CEN puncta or only a weak signal (Figure 2D, S8A). CEN did not localize to centrosomes or other subcellular structures in early embryos prior to cycle 10. CEN localized at centrosomes from cycle 10 (Figure S8A) until cycle 13. By mitosis of cycle 14 and thereafter, no signal for CEN was detected in the embryo except in the germ cells (Figure S7C). The asymmetric localization of CEN on spindles during blastoderm divisions was surprising since no asymmetry has previously been defined for these divisions. Whether CEN is partitioning between centrosomes randomly, or favoring the older or younger of the two is unclear. Moreover, the pericentrosomal structure that CEN localizes to has not been previously defined.
In addition to the localization of CEN at the novel pericentrosomal particle described above, CEN localized at cleavage furrows (Figures 2E, S8A). Localization of CEN at centrosomes and cleavage furrows was detected with antibody staining directed at CEN and also with the CEN-CFP and CEN-Venus fusion proteins (Figures 2D, E and S8A). Overexpression of CEN-CFP or CEN-Venus resulted in symmetrical localization of CEN to both centrosomes in a pattern that completely surrounds the centrosome (Figure 2F). Overexpression of CEN-Venus fusion protein frequently resulted in assembly of smaller furrows at sites where large CEN aggregates formed, and also caused nuclear fall-out (not shown), similar to the furrow defects seen in cen mutant embryos (Figure S9B). Interestingly, the cen mRNA was reported recently to also localize near or at centrosomes , prompting the compelling possibility that the translation of cen mRNA is regulated at this site [16–18]. Thus, CEN localizes to a novel structure that sits adjacent to centrosomes, with the CEN and CNN signals co-localizing at the point of contact. Unlike CNN however, CEN also localized at the cleavage furrows (Figures 2E, S8A). These sites of localization implicate CEN as a conveyor of the signal transmitted from the centrosome to the furrow.
To address the role of CEN in early embryos, we examined the phenotype of a cen mutant. A piggybac transposon insertion mutation within the coding sequence of cen at amino acid position 290 was available for this analysis (Figure 2C) . Maternal effect cen mutant embryos, collected from hemizygous cenf04787 mutant mothers (cenf04787 heterozygous with a deficiency chromosome, Df(2L)Fs(2)KetRX32, which deletes the cen locus), contained no detectable CEN protein by Western blotting using antibodies directed against either the C- or N-terminus of CEN (Figure 3A). A truncated protein product, predicted by the site of insertion of the transposon to be at least 33 kD, was also not detected with the antibody directed against the amino-terminus of CEN, which was raised against a polypeptide included within this truncation (Figure 2C). Moreover, no CEN signal was detected at centrosomes or furrows in cenf04787 embryos upon immunostaining with CEN antibodies (data not shown).
Homozygous and hemizygous cenf04787 mutants were viable and fertile. However, hemizygous cenf04787 females laid eggs that failed to hatch at a significantly higher rate of 12.10 +/− 1.06% (mean +/− SEM) compared to wild type (5.35 +/− 0.68%) (Figure 3F). A cen transgene that expresses a cen cDNA (see Figure 3A for expression), including the entire coding sequence with the 5′ and 3′ UTRs, rescued this hatch rate deficiency to the levels seen with wild type (5.67% +/− 1.44%). Together, these data show that cenf04787 is a mutation in cen that reduces its expression to an undetectable level and affects embryonic development maternally.
To investigate CEN’s role in centrosome and cleavage furrow function, we immunostained hemizygous maternal cenf04787 mutant embryos, hereafter referred to as cenf04787 embryos, for CNN, γ-Tubulin and filamentous actin to examine centrosome and cytoskeleton organization during cleavage. cenf04787 embryos display defects in centrosome separation, with 9.62% (5/52) of embryos showing centrosome separation failure compared to 0% of wild type (0/41) (Figure 3C, right panel). In addition, mitotic spindles were frequently linked together (Figure 3C, left panel), a phenotype also characteristic of cnn mutant embryos, and an indication of furrow assembly failure. While the severity of linked spindles is variable among cenf04787embryos (Figure 3C shows a severe case), this phenotype is highly penetrant, with 30.77% (16/52) of metaphase cenf04787 embryos showing linked spindles, compared to wild type (2.44% (1/41)). Actin staining showed that furrows form aberrantly, are consistently less robust, and with decreased furrow depth in cenf04787 embryos (Figure 3E). Actin organization into pseudocleavage furrows displayed variable degrees of disruption in cenf04787 embryos, however, overall 27.85 % (22/79) of cenf04787 embryos displayed broken or weak furrows at prophase or metaphase compared to 0.91% (1/110) of wild type embryos. In addition, the distribution of actin density in furrows was frequently irregular in cenf04787 embryos (Figure S9A) and patches of small furrows were common (Figure S9B). However, no obvious effects on actin cap formation were observed at interphase (Figure S9A). Thus, cen mutant embryos are deficient in centrosome separation and in mitotic cleavage furrow assembly. Although the furrow defects and linked spindles of cenf04787 embryos is ~28–31%, the embryo hatch failure rate is ~12%, only 7% higher than wild type, attesting to the ability of embryos to cope with the furrow defects in cen mutant embryos, with the likely exception of those with very severe defects like the embryos shown in Figure 3C.
On the minority of wild type spindles where CEN appeared to split symmetrically between the two centrosomes at mitosis, no affect on furrow assembly was observed. This suggests that the asymmetry of CEN distribution at mitosis may not have any acute affect on furrow assembly. We propose that CEN localization at centrosomes may function early in mitosis to initiate a signaling process that is required as furrow assembly proceeds, impacting furrow actin assembly at mitotic furrows.
Recycling endosomes (RE) have a defined role in the trafficking of actin- and membrane-containing vesicles to organize cleavage furrows [20–22]. To investigate an impact on RE function in cenf04787 embryos we stained for the RE markers Nuf and Rab11, which localize to REs that are distributed in a pericentrosomal pattern. Localization of neither of these RE components nor the Dystrophin ortholog and furrow membrane protein Dah , which is very sensitive to perturbations of RE activity [21, 22, 24], was affected by loss of CEN function (Figure S10). Since the REs activate Rho1 at furrows through RhoGEF2 recruitment to promote actin assembly [21, 22], it therefore appears that CEN promotes actin assembly at furrows by a Rho-independent pathway.
The cleavage furrow defects seen in cenf04787 embryos are likely not due to microtubule assembly defects since the astral microtubules in cenf04787 embryos are comparable to wild type (Figures 3E and S11). Moreover, the pericentrosomal localization of Nuf and Rab11 are dependent upon microtubules , yet their localization appeared normal in cenf04787 embryos.
The phenotypes seen in cenf04787 embryos are consistent with CEN functioning in conjunction with CNN in the early embryo: mutations in cnn cause centrosome separation failure and defective furrow assembly [6, 7, 12]. However, the defects in these processes in cenf04787 embryos are not as severe as those that occur in cnn mutants, including the hypomorphic cnnB4 mutant. This suggests that CEN may not be the only factor involved in conveying the signal from the CM2 domain of CNN to cleavage furrows. Nevertheless, since CNNB4 localizes to centrosomes, albeit with an altered PCM pattern, we examined the ability of CEN to localize to centrosomes in cnnB4 embryos. Consistent with the yeast two-hybrid interaction assay, CEN failed to localize detectably at cnnB4 centrosomes, yet was localized variably at furrows during early cortical cycles (10 and 11) (Figure S8B), but not at the residual patches of furrows that form at later cycles (12 and 13) in cnnB4 embryos. Since the interaction of CEN and CNNB4 is reduced but not abolished (Figure 2A,B), it is possible that some CEN is localized to centrosomes and that this facilitates the inefficient localization to cnnB4 furrows. Alternatively, CEN association with CNN may be required for its activation prior to its action at furrows to promote actin assembly.
Together, these data indicate that the conserved domain at the C-terminus of CNN is critical for cleavage furrow formation and for recruitment of CEN to centrosomes. However, since furrow defects in the cen mutant are not as severe as cnnB4, CEN is unlikely the only factor involved in the regulation of furrow formation by CNN CM2. In summary, CNN CM2 instructs cleavage furrow formation, and appears to accomplish this in part through the recruitment of CEN to the centrosome and/or directing it to the cleavage furrow.
To test the cooperative functions of CNN and CEN, we examined the ability of mutations in cnn to influence cen. A single copy of a null cnn mutant, cnnhk21, enhanced the rate of embryo hatch failure from cen04787/Df(2L)Fs(2)KetRX32 females from 12.10 +/− 1.06% to 17.89 +/− 2.21% (mean +/− SEM). The control cnnhk21/+, Df(2L)Fs(2)KetRX32/+, cen04787 +/+ cnnhk21, and Df(2L)Fs(2)KetRX32 +/+ cnnhk21 females produced wild type embryo hatch rates (2–5%). In all experiments, females were mated to wild type males. This significant decrease in embryonic development shows that cnn and cen interact genetically, further evidence that CNN and CEN cooperate in a common pathway.
In conclusion, the CM2 domain of CNN is required for the signaling from the centrosome to instruct cleavage furrow assembly at the embryonic cortical membrane. CM2 accomplishes this through binding with CEN, which is required for efficient cleavage furrow formation. Thus, the CM2 domain of CNN and CEN represent a molecular link between centrosomes and the signals that regulate cleavage furrow assembly.
The cenf04787 mutant stock was obtained from the Bloomington Stock center. The Df(2L)Fs(2)KetRX32, and Df(2L)DS5 stocks were a gift from Paul Lasko (McGill University, Montreal). The cnnB4 mutant stock was a gift from Eyal Schejter (Weizmann Institute of Science, Rehovot, Israel; ). The cnnhk21 allele was recombined with cenf04787 for double mutant analysis. For construction of the untagged cen expression transgene, the cDNA clone LD41224 was cut with BstXI and XhoI, made blunt, and ligated into pBluescript SK vector cut with EcoRV. The cen cDNA sequence, including 5′ and 3′ UTRs, was then cut with KpnI and XbaI ligated into pUASp vector KpnI-XbaI sites. For expression of CEN-CFP and CEN-Venus, CEN ORF sequence was PCR amplified and cloned into pENTR/TOPO vector (Invitrogen) and recombined into Gateway destination vectors pPWC and pPWV, respectively (Terence Murphy, The Drosophila Gateway Vector Collection, Carnegie Institution of Washington, Baltimore, MD). Transgenic lines were generated by standard methods. nos-Gal4VP16 was used to express pUASp-cen transgenes in ovaries and embryos.
cen mutant (maternal effect) embryos were collected from cen hemizygous or homozygous mutant females. Df(2L)Fs(2)KetRX32 was used to generate hemizygous mutant flies. For hatch rate determination, embryos were collected for four hours, lined up on apple juice/agar plates (450–550 embryos/plate), and unhatched embryos were counted after two days incubation. At least three experiments were conducted for each genotype. Statistical tests were two-tailed Student’s t test, and were considered statistically significant when p<0.05.
CM2 bait included the C-terminal 67 amino acids of CNN cloned into pBTM116. A Drosophila cDNA library in pACT2 (a gift from Stephen Elledge) was used to screen for interactors. We screened approximately 3 × 106 colonies and recovered one clone, encoding amino acids 84–790 of cen.
Antibodies to CEN, TACC and Msps were raised in rabbits by Cocalico Biologicals, Inc. using 6XHis-tagged fusion proteins expressed from pRSETB (Invitrogen) in E. coli from amino acid regions 2–307 and 259–686 of CEN, 191–583 of TACC, and 1024–1270 of and purified by immobilized-metal affinity chromatography using Ni++Sepharose Fast Flow. The antibodies to TACC, Msps and the amino terminus of CEN were affinity purified over Affigel-10 (Bio-Rad) coupled protein columns.
Embryos fixed with heptane/methanol were used for antibody staining as described  except for phalloidin staining, where embryos were fixed with heptane/formaldehyde and the vitelline membrane were removed manually by rolling the embryos between the frosted slide and coverglass. Antibodies were diluted as follow: rabbit or guinea pig anti-CNN (1:1000) , rabbit anti-CEN C-terminus (1:1000), rabbit anti-CEN N-terminus (affinity purified 1mg/ml, 1:500), affinity purified rabbit anti-Msps (0.6mg/ml, 1:500), affinity purified rabbit anti-TACC (0.8mg/ml, 1:500), mouse anti-γ-tubulin (GTU88 from Sigma, 1:500), mouse anti-α-tubulin (DM1A from Sigma, 1:1000), rabbit anti-GFP (Invitrogen, 1:1000), rabbit anti-Dah (1:300, a gift from Tao-Shih Hsieh, Duke Univ., Durham, NC), rat anti-Rab11 (1:1000, a gift from Robert Cohen, Univ. of Kansas, Lawrence, KS), and rabbit anti-Nuf (1:500, a gift from William Sullivan, Univ. of California, Santa Cruz, Santa Cruz, CA). Phalloidin-Alexa546 (Invitrogen) was used at 1:160 dilution. DNA was stained with DRAQ5 (Axxora) at 1:2000 dilution. Fluorescent secondary antibodies were highly cross-absorbed goat conjugated Alexa 488 and 546 (Invitrogen) used at a 1:500 dilution. Images were captured on a Leica TCS SP2 confocal microscope (Deerfield, IL) or with a Zeiss Axioskop Mot2 microscope equipped with a 63x/NA1.4 oil immersion objective and a Coolsnap FX CCD camera using Metamorph software.
Third-instar larvae of wild type, cnnB4 and cnnhk21 were placed in ice-cold Schneider’s insect medium for 30 min. and the larval brains were dissected and placed on ice for one hour. Microtubules, disassembled after one hour cold-treatment, were followed by a time-course recovery at room temperature. The larval brains were then fixed and stained with anti-α-tubulin, anti-CNN and either Draq5 or anti-phospho-histone H3 (Upstate) .
Confocal images were taken at the same settings for wild type and cnnB4 mutant embryos. To quantify the fluorescence intensity of CNN or γ-Tubulin at centrosomes, identical circular regions were drawn around the centrosomes to measure the signal intensities. Control regions were also drawn nearby the centrosomes to subtract as background. The fluorescence intensity (arbitrary units) was calculated using Leica TCS SP2 software. Thirty to forty spots were counted per image field and three different embryos were counted for each genotype.
We thank Eyal Schejter for the cnnB4 mutant stock, Paul Lasko and Risa Shapiro for stocks of mutants in the 38D region, Steve Wasserman, Bill Sullivan, Bob Cohen, and Tao-shih Hsieh for antibodies, Terence Murphy for Gateway destination vectors, and Steve Elledge for the Drosophila pACT2 library. This work was supported by grants from the National Institutes of Health (NIH) (GM068756) and the Welch Foundation (I-1610) to T.L. Megraw.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.