The precise mechanism of chromosome condensation and decondensation remains a mystery, despite progress over the last 20 years aimed at identifying components essential to the mitotic compaction of the genome. In this study, we analyse the localization and role of the CAP-D2 non-SMC condensin subunit and its effect on the stability of the condensin complex. We demonstrate that a condensin complex exists in Drosophila embryos, containing CAP-D2, the anticipated SMC2 and SMC4 proteins, the CAP-H/Barren and CAP-G (non-SMC) subunits. We show that CAP-D2 is a nuclear protein throughout interphase, increasing in level during S phase, present on chromosome axes in mitosis, and still present on chromosomes as they start to decondense late in mitosis. We analysed the consequences of CAP-D2 loss after dsRNA-mediated interference, and discovered that the protein is essential for chromosome arm and centromere resolution. The loss of CAP-D2 after RNAi has additional downstream consequences on the stability of CAP-H, the localization of DNA topoisomerase II and other condensin subunits, and chromosome segregation. Finally, we discovered that even after interfering with two components important for chromosome architecture (DNA topoisomerase II and condensin), chromosomes were still able to compact, paving the way for the identification of further components or activities required for this essential process.
Introduction
Chromosomal DNA, if stretched end to end, measures about 2 metres in any one cell of the human body and therefore must be highly folded to fit into a nucleus of only 5 μm diameter. DNA has to condense even further prior to cell division in order to form the highly compact metaphase chromosome comprising two sister chromatids, capable of withstanding anaphase segregation (Heck, 1997; McHugh, 2003; Swedlow and Hirano, 2003). Thus, mitotic chromosome condensation involves a process in which interphase chromatin is (1) compacted, and (2) resolved into two distinct rod-shaped sister chromatids. Recent studies have contributed to the unravelling of this process. A five subunit `condensin' complex was initially identified biochemically in Xenopus (Hirano et al., 1997; Hirano and Mitchison, 1994), and comprises two SMC (structural maintenance of chromosomes) proteins (SMC2 and SMC4) and three non-SMC proteins [CAP-D2, CAP-G or CAP-H (Barren in Drosophila)]. All five subunits have been identified in the eukaryotes examined to date (Cobbe and Heck, 2000). The SMC proteins belong to a superfamily of highly conserved and ubiquitous chromosomal ATPases (Cobbe and Heck, 2004). The non-SMC subunits have been proposed to have dual roles in the regulation of condensin function: one is to activate the SMC ATPases to perform ATP-dependent supercoiling activity, and the other is to allow the holocomplex to associate with chromatin in a mitosis-specific manner (Kimura and Hirano, 2000). Each of the condensin subunits is essential and required for proper organisation and segregation of mitotic chromosomes, but exactly how the complex contributes to these processes is still unclear. Recently a second condensin complex has been identified, with a different complement of non-SMC subunits (Ono et al., 2003).
Studies of condensin function have resulted in conflicting hypotheses as to the role of this complex. Cells of Schizosaccharomyces pombe mutant in Cut3 and Cut14 (the SMC4 and SMC2 subunits, respectively) exhibited chromosome condensation defects and displayed a `cut' phenotype where septation (cell division) occurred with incompletely separated chromosomes (Saka et al., 1994). Fluorescence in situ hybridization in these mutants demonstrated that the length of mitotic chromosome arms increased while the centromeric DNA could separate and move to the opposite poles normally. In S. pombe and Saccharomyces cerevisiae, studies have shown that all members of the condensin complex are required for proper chromosome condensation and segregation (Lavoie et al., 2000; Ouspenski et al., 2000; Strunnikov et al., 1995; Sutani et al., 1999). In Xenopus, when condensin was depleted from mitotic extracts before or after condensation, unreplicated chromatin failed to assemble into individual chromatids or became completely disorganised, suggesting that the complex was required both for assembly and maintenance of mitotic chromosomes (Hirano et al., 1997; Hirano and Mitchison, 1994). In contrast, genetic studies in Drosophila melanogaster and Caenorhabditis elegans showed that resolution between the sister chromatids rather than condensation of the chromosomes was compromised when the condensin complex was disrupted (Bhat et al., 1996; Hagstrom et al., 2002; Steffensen et al., 2001). In both organisms, a high degree of compaction relative to interphase and a bipolar metaphase plate formed. However, severe chromosome segregation defects and chromosome breakage were observed in anaphase and telophase. DmSMC4 RNAi in Drosophila cultured cells confirmed the requirement of this protein for chromosome organisation and segregation (Coelho et al., 2003). In both flies and worms, the absence of condensin subunits results in the formation of chromosome bridges because of incomplete resolution and ensuing failure of sister chromatids to separate completely during anaphase (Hagstrom and Meyer, 2003). Particular insight was gleaned when the robustness of chromosome architecture was assessed after SMC2 knock-out in chicken DT40 cells: while chromosomes still reached normal levels of condensation, clear defects in association of other non-histone chromosomal proteins was observed (Hudson et al., 2003).
Recently, a second condensin complex, called condensin II, was identified in vertebrates (Ono et al., 2003). This complex shares the same two SMC subunits as the original complex, now named condensin I, but appears to contain different non-SMC subunits (named CAP-D3, CAP-G2 and CAP-H2). The two complexes appear to alternate along the axis of metaphase chromatids but are both present at the centromere, albeit in distinct regions (Ono et al., 2004). Depletion of these complexes in Xenopus and HeLa cells produced distinct defects in chromosome morphology, suggesting that the complexes may contribute differently to mitotic chromosome architecture (Ono et al., 2003). Furthermore, the centromere/kinetochore regions were structurally disorganised from the depletion of either condensin I or II subunits, suggesting a specific role for both complexes in centromere organisation (Ono et al., 2004).
DNA topoisomerase II (topo II) has long been implicated in chromosome structure and dynamics, since its discovery as a chromosome `scaffold' protein and its localization at the base of chromatin loops (Earnshaw et al., 1985; Earnshaw and Heck, 1985; Gasser et al., 1986). SMC2 has also been shown to be a component of the chromosome `scaffold' fraction (Saitoh et al., 1994). Yeast mutants in topo II exhibit segregation defects similar to those observed in condensin mutants (Holm et al., 1985). The decatenation activity of topo II facilitates the resolution of sister chromatids and the complete separation of chromosomes in anaphase. Topo II has also been shown to be required for chromosome condensation but its role in this process has been controversial since disruption of its function resulted in different chromosome morphologies (Swedlow and Hirano, 2003; Warburton and Earnshaw, 1997). Recently, an unexpected role for topo II in chromosome arm congression emerged, after RNAi depletion in Drosophila cells (Chang et al., 2003). Localisation studies revealed that Topo II was axially distributed in the chromosome arms with some concentration at the centromeres, a distribution that was also observed for condensin subunits (Maeshima and Laemmli, 2003; Steffensen et al., 2001). The localisation pattern and the phenotype after the depletion of either the condensin complex or topo II suggested that these two components of the chromosome scaffold may collaborate to bring about chromosome compaction and organisation. Biochemical and localisation studies in Drosophila showed that CAP-H/Barren associated with topo II throughout mitosis, while SMC4 appeared to be responsible for the axial localisation of topo II on the chromosomes and for its decatenation activity (Bhat et al., 1996; Coelho et al., 2003). However, in S. pombe, the localisation but not the activity of topo II was affected in condensin mutants, while topo II localization to well-defined axial structures in Xenopus was independent of condensin (Bhalla et al., 2002; Cuvier and Hirano, 2003). Given that chromosomes experience different constraints depending on the cell, organism, or time in development, it is perhaps not surprising that a `unified' hypothesis regarding these two components has not emerged.
In this study, we demonstrate that a condensin I complex exists in Drosophila. We subsequently examined the cell cycle localization and role of the Drosophila CAP-D2 subunit in chromosome dynamics during mitosis. We show that depletion of CAP-D2 by RNAi in Drosophila cultured cells results in a compromised mitotic chromosome architecture with defective sister chromatid resolution and subsequent chromosome bridges in anaphase. We demonstrate here for the first time, that the fate of other members of the condensin complex and other chromosomal proteins essential for chromosome dynamics during mitosis is altered when a condensin subunit is disrupted either by depletion or mutation. We additionally examine Drosophila mutations of condensin subunits to assess the stability of the complex in the context of the organism. We also show that mitotic coordination with regard to chromosome passengers is compromised when condensin function is affected. Finally we demonstrate that when both condensin and topo II are disrupted by RNAi, chromosomes are surprisingly still able to compact, intimating the existence of additional, currently unknown, factors essential to the assembly of mitotic chromosomes.
Materials and Methods
Identification of the condensin complex
Extract preparation
Drosophila embryos (0-5 hours) were homogenised in one volume of buffer A (15 mM NaCl, 60 mM KCl, 2 mM EDTA, 0.34 M sucrose, 15 mM Hepes pH 7.9) and complete protease inhibitors (Roche) using a loose pestle. The homogenate was centrifuged for 5 minutes at 5000 g at 4°C in a benchtop centrifuge. The middle layer was collected, divided into aliquots and flash frozen in liquid nitrogen. Dimethyl pimelimidate cross-linking of CAP-D2 antibody to protein A-Sepharose was performed according to the method of Harlow and Lane (Harlow and Lane, 1999). For immunoprecipitation from crude extracts, antibody beads were incubated with the extract for 1 hour at 4°C. The beads were washed 10 times with 20 volumes of PBS, and then eluted and washed with PBS containing 250 and then 500 mM NaCl. Protein interactions that were resistant to 500 mM NaCl were eluted using 2% SDS in PBS.
Generation of antigens and antibodies
Polyhistidine (6His)-tagged fragments of DmSMC2 and DmCAP-D2 were produced using the residues 24-309 and 951-1162 respectively, by cloning a BamHI-HindIII fragment of the cDNA into a pQE30 (QIAexpress system, Qiagen) expression vector. The constructs were expressed in E. coli M15[pREP4] cells (Qiagen) and the proteins were purified on Ni-agarose columns (Qiagen) and subjected to SDS-PAGE. Individual gel bands were excised from the gel, ground under liquid N2, and injected into rabbits (Scottish Antibody Production Unit, Pentlands Science Park, Edinburgh, UK). The same procedure was followed for the generation of topo II antibody. Residues 1079-1333 were used and a BamHI-HindIII fragment of the genomic DNA was cloned into a pET23a (Novagen) expression vector. The construct was transformed into E. coli BL21 cells (Novagen) and the protein was purified as described above. For CAP-D3 antibody, a peptide corresponding to the terminal 15 amino acids of the protein (residues 1253-1267, CDMLVTELIFQPPDW) was synthesised and injected into rabbits (Genosphere, Paris, France).
Cell culture and RNAi
Drosophila embryonic S2 cells were maintained in serum-free Schneider's insect medium supplemented with 10% insect qualified fetal bovine serum (FBS). Cells were grown at 27°C in a humidified atmosphere. RNAi was performed in Drosophila S2 cells as previously described (Clemens et al., 2000; Vass et al., 2003). Two fragments from the 5′ end of CAP-D2 [forward primer at 251-270 bp of the open reading frame (ORF) and reverse primer at 873-892 bp of ORF] were fused to T7 RNA polymerase promoter and were used as PCR primers. The Drosophila EST clone LD40412 was used as a template for the PCR reaction. The PCR product obtained (∼642 bp) was used as template for RNA synthesis using the Megascript kit (Ambion). A random human intronic sequence was used to generate control dsRNA, which was synthesized in the same way. Topo II dsRNA was obtained as described previously (Chang et al., 2003). For the CAP-D3 dsRNA the EST clone RE18364 was used as a template for the PCR reactions. Two PCR products were obtained from two different regions of the ORF. The first one was obtained with a primer pair that includes the start codon (r1; ∼639 bp) and the second one with a primer pair further inside the message (r2; ∼723 bp). The two products were used in two independent experiments. For the RNAi experiment, 1 ×106 cells/ml were diluted in serum-free Schneider's insect medium. 20-25 μg/ml of dsRNA was added to the cells (30 μg/ml for the CAP-D3 RNAi). Aliquots of 1 ml of these cells was placed into each well of a 35 mm 6-well plate and incubated for 60 to 90 minutes. After the incubation 2 ml of Schneider's medium supplemented with 10% FBS was added to each well. The cells were placed in the 27°C humidified incubator and samples were processed for immunoblotting or for immunofluorescence at certain time points. For some experiments, cells were grown on sterilized concanavalin A-treated coverslips in a 35 mm 6-well plate (Rogers, 2002), but only for the last 24 hours before being processed for immunofluorescence. In some cases, S2 tissue culture cells (RNAi-treated or control cells) were exposed to 30 μM colchicine for 7 hours, 67 hours after the RNAi treatment. Cells were then processed for immunoblotting or immunofluorescence.
Preparation of cell extracts
Cells were counted under the microscope with a hematocytometer, centrifuged at 2000 rpm for 2 minutes and washed with 0.5 ml of PBS (phosphate-buffered saline: 137 mM NaCl, 2.7 mM KCl, 6.46 mM Na2HPO4, 1.47 mM KH2PO4 pH 7.4). The pellet was resuspended in PBS and warm sample buffer [3 × sample buffer: 2% SDS, 10% glycerol, 0.01% Bromophenol Blue, 2 mM EDTA, 50 mM trizma base pH 6.8] with 10 mM 1,4-dithioerythritol (DTT) (1/10 of final volume). The cells were lysed by sonication and the extract was boiled at 100°C for 5-10 minutes. The volume of PBS and sample buffer was calculated so that the final cell concentration was 5 ×104 cells/μl.
Preparation of embryo extracts
To identify embryos homozygous for barrL305 and glu17C the mutant chromosomes were balanced over CyO, P{w+mc=GAL4-Kr.C}DC3, P{w+mc=UAS-GFP.S65T}DsC7 as described previously (Casso et al., 2000). One hour collections of embryos were aged for 12 hours, dechorionated with 50% bleach for 1 minute and then sorted under a fluorescence dissecting microscope. Homozygous embryos were homogenised with a motorised pestle in EBR lysis buffer (13 mM NaCl, 0.47 mM KCl, 0.19 mM CaCl2, 1 mM Hepes pH 6.9) plus 10 mM EDTA, 10 mM DTT, 1:100 dilution of CLAP (chymostatin, leupeptin, antipain, pepstatin), phenylmethanesulfonyl fluoride (PMSF) and trasylol, and warm sample buffer with DTT (1/10) was added immediately. The embryo extract was boiled at 100°C for 10 minutes.
Immunoblotting
Protein extracts were separated by SDS-PAGE and transferred onto nitrocellulose membranes in a Trans-Blot apparatus. Membranes were blocked in PBS+0.1% Tween 20 (PBSTw) and 5% semi-skimmed milk for 1 hour at room temperature (RT) and then incubated for 1-1.5 hours with the primary antibody in PBSTw. After washing three times for 5, 15 and 10 minutes with PBSTw, the membranes were incubated in a horseradish peroxidase-linked (HRP) secondary antibody for 1 hour in PBSTw at RT. Finally the membranes were washed as above in PBSTw and immunocomplexes were detected by enhanced chemiluminescence (ECL; Amersham Biosciences). For the RNAi experiment, 5 ×105 cells were loaded per lane while for the analysis of embryos, five embryos were loaded per lane for both the wild type (Canton S) and mutant material.
Phosphorimager analysis
Samples were processed as described for immunoblotting but instead of using an HRP secondary, a Cy5-conjugated secondary antibody (Jackson ImmunoResearch Laboratories) was used at a dilution of 1:200. The intensity of the signal was measured by using the STORM 860 scanning phosphorimager and calculated using the ImageQuant program (Amersham). The intensity was finally normalised by using α-tubulin as a loading control. Values were corrected by subtracting the background of the corresponding lane from each band.
Immunofluorescence of Drosophila cultured cells
2 ×105 cells diluted in EBR were centrifuged onto poly-L-lysine-coated slides for 2 minutes at 2000 rpm in a cytospin column. The cells were immediately fixed for 4 minutes in 4% paraformaldehyde (PFA) in PBS at RT, washed for 5 minutes in PBS+0.1% Triton-X (PBSTx) and permeabilized in PBS +0.5% Triton X-100 for 5 minutes. After the permeabilization, cells were blocked for 1 hour at RT in 0.5% bovine serum albumin (BSA; Sigma). Cells were washed in PBSTx for 5 minutes and incubated for 1 hour in the primary antibody in PBSTx+0.5% BSA. After three 5-minute washes in PBSTx, cells were incubated in the secondary antibody (conjugated fluorochromes; Molecular Probes or Jackson ImmunoResearch Laboratories) in PBSTx+0.5% BSA and in 1 μg/ml DAPI for DNA staining. Cells were washed four times, for 5 minutes each, in PBSTx and coverslips were mounted onto the slides with Mowiol (Calbiochem). The cells were viewed under the fluorescence microscope. For some antibodies (SMC2, SMC4, guinea pig topo II and CAP-D3), 0.5% Triton-X was included in the fixative and a 10-minute incubation was used. A permeabilisation step was performed as well, as described above. For hypotonic treatment, cells were diluted in 0.25 × EBR at a concentration of 2 ×105 for 2 minutes at RT and then centrifuged onto poly-L-lysine slides and processed as described above. Cells that were grown on concanavalin A-coated coverslips for 24 hours in a 35 mm six-well plate were then fixed and processed as described above.
BrdU incorporation in Drosophila cultured cells
Cells growing in flasks were incubated for 30 minutes in medium containing 6 μg/ml BrdU (bromodeoxyuridine). Approx. 2 ×105 cells were centrifuged onto poly-L-lysine slides and processed as described for immunofluorescence except that DAPI was omitted. The primary antibody in this case was anti-CAP-D2 and the secondary was Alexa Fluor 594 goat anti-rabbit conjugate. After the anti-CAP-D2 immunofluorescence was complete, the cells were post-fixed immediately in 4% PFA in PBS for 1 hour at RT, washed for 5 minutes in PBSTx and then incubated for 15 minutes in freshly prepared 2 N HCl. After a 5-minute wash in PBSTx, cells were blocked for 30 minutes in PBS+5% BSA and washed for 5 minutes in PBSTx and then incubated with the rat anti-BrdU antibody (1:2 dilution) in PBSTx+0.5% BSA at 4°C for 16 hours. After three 5-minute washes, the cells were incubated with Alexa Fluor 488 goat anti-rat conjugate (1:500 dilution) in PBSTx+0.5% BSA for 1 hour at RT. 1 μg/ml DAPI was added during the last incubation. Finally cells were washed four times for 5 minutes each with PBSTx and coverslips were mounted onto the slides with Mowiol.
Microscopy
All preparations were examined with an Olympus Provis microscope, equipped with epifluorescence optics. Images were captured with an Orca II CCD camera (Hamamatsu) and Smart Capture 2 software (Digital Scientific). All images were processed with Adobe Photoshop software. For examining the localisation and the presence of several proteins on the chromosomes in the RNAi experiment, all images were obtained with the same camera settings and the same exposure time was maintained between the controls and the CAP-D2 RNAi experiment.
Antibodies
The primary antibodies were used for immunofluorescence as follows: α-tubulin (mouse monoclonal antibody B5-1-2 used at 1:500; Sigma), γ-tubulin (mouse monoclonal T 6557 used at 1:50; Sigma), phosphohistone H3 (P-H3; mouse monoclonal used at 1:500, 9706L; New England Biolabs), BrdU (rat antibody used at 1:2; Harlan Sera Lab, Loughborough, UK), centromere identifier (CID) [chicken polyclonal used at 1:200, (Blower and Karpen, 2001)], topo II (rabbit polyclonal used at 1:500, and guinea pig polyclonal used at 1:500), Barren [rabbit polyclonal used at 1:500 (Bhat et al., 1996)], SMC2 (rabbit polyclonal used at 1;500), SMC4 [rabbit and sheep polyclonals used at 1:500 (Steffensen et al., 2001)], SCC1 [rabbit polyclonal used at 1:500 (Warren et al., 2000)], DSA1 [rabbit polyclonal used at 1:500 (Valdeolmillos, 2004)], BubR1 (rabbit polyclonal used at 1:1000). Sheep Cyclin B, INCENP [rabbit polyclonal used at 1:500 (Adams et al., 2001b)], CAP-D2 (rabbit polyclonal used at 1:1000 for immunofluorescence and 1:10,000 for immunoblotting), CAP-D3 (rabbit polyclonal used at 1:200 for immunofluorescence and 1:500 for immunoblotting). All fluorescently conjugated secondary antibodies (Molecular Probes, Jackson ImmunoResearch Laboratories and Amersham Biosciences) were used according to the manufacturer's instructions.
Results
Biochemical identification of a condensin complex in Drosophila
In this study we demonstrate for the first time that SMC2, SMC4 and the CAP-D2, CAP-G and CAP-H/Barren proteins exist as a complex in Drosophila. Immunoprecipitations with an antibody generated to DmCAP-D2 (Fig. S1A in supplementary material) were performed using 0- to 5-hour embryo extracts. Bound proteins were eluted sequentially with 0.5 M NaCl and 2% SDS in PBS. The fractions were analysed by immunoblotting with SMC2, SMC4, CAP-D2 and CAP-H/Barren antibodies. All four proteins were present suggesting that they participate in the same complex. SMC2 and SMC4 proteins were eluted with both NaCl and SDS solutions, while CAP-D2 and CAP-H were only eluted with the SDS (Fig. 1A). This result suggests two possibilities. First, the CAP-D2 and CAP-H non-SMC subunits may be more tightly associated with each other than they are with the SMC subunits of the condensin complex [we made a similar observation for the cohesin complex (Vass et al., 2003)]. Second, the SMCs that elute with NaCl may be participating in another complex, such as condensin II. As no antibody was available to the CAP-G protein, the presence of this protein in the complex could not be assessed by immunoblotting. However, we were able to identify CAP-G by mass spectrometric analysis of proteins interacting with CAP-D2 (data not shown).
Characterisation of CAP-D2 during the cell cycle
To determine the subcellular localization of the CAP-D2 protein during the cell cycle, S2 cultured cells were centrifuged onto poly-L-lysine slides and processed for indirect immunofluorescence using the CAP-D2 antibody. Immunofluorescence revealed that CAP-D2 was mainly nuclear with fluctuating levels during interphase. Several markers were utilized to further characterise these levels. An antibody to γ-tubulin, a centrosomal component, was used to identify cells that were in G1 phase prior to centrosome duplication. In G1 cells the level of CAP-D2 in the cytoplasm was nearly the same as in the nucleus (Fig. 1B, top). Cells with two centrosomes (in S or G2 phase) (Loupart et al., 2000) had much higher levels of nuclear CAP-D2 (Fig. 1B, bottom). S phase cells were identified following BrdU (a thymidine analogue) incorporation. All S phase cells showed nuclear staining of CAP-D2 (Fig. 1C). While the intensity of the CAP-D2 signal varied among S phase cells, analysis of replication patterns with two thymidine analogues, CldU and IdU (Manders et al., 1992), demonstrated that the nuclear level of CAP-D2 increased as cells progressed through S phase (data not shown). Cyclin B, a cofactor for the mitotic CDC2 kinase, moves from the cytoplasm into the nucleus in late G2/early prophase (Huang, 1999). Cells with nuclear Cyclin B showed an intense nuclear staining for CAP-D2 (Fig. 1D). These results suggest that Drosophila CAP-D2 is nuclear during interphase, with low levels in G1 that increase during S and are highest in late G2/early prophase.
The localisation of CAP-D2 during mitosis was also examined. In prometaphase, the protein localised along the condensed chromosome arms with more intense labelling of the centromeric regions (Fig. 2A,B). The accumulation of CAP-D2 at centromeres can be clearly observed when cells are immunostained for both CAP-D2 and CID, the Drosophila homologue of CENP-A, a histone H3 isoform (Henikoff, 2000) (Fig. 2B). In metaphase and early anaphase, when chromosomes were fully condensed, the CAP-D2 signal was most intense throughout the core of the chromatids. In telophase, even as histone H3 becomes dephosphorylated and chromosomes start to decondense, CAP-D2 was still tightly associated with chromatin (Fig. 2A). The CAP-D2 staining was still intense during late telophase and cytokinesis (Fig. 2C). Our results suggest that CAP-D2 associates with chromosomes in prophase, distributes axially on the chromosome arms with enrichment at the centromeres during prometaphase, is present throughout anaphase and telophase, and even remains tightly associated with chromosomes as decondensation ensues.
SMC4 and Barren proteins are loaded onto chromosomes in early prophase but they dissociate from chromosomes late in anaphase/telophase when decondensation begins (Steffensen et al., 2001). Since CAP-D2 appears to persist on chromosomes late in mitosis, we performed co-localisation studies with SMC4 in order to observe if the two subunits dissociate independently from chromatin late in mitosis. During anaphase, SMC4 began to dissociate from chromosomes as they started to decondense. Like phosphohistone H3 (P-H3), the SMC4 signal was fainter near the centromeric regions but was quite strong at the end of chromosome arms, while CAP-D2 was still present throughout the chromatids (see Fig. S1C in supplementary material). In telophase, while CAP-D2 was still associated with chromatin, SMC4 was diffuse and the chromosomal staining reduced (see Fig. S1C in supplementary material).
CAP-D2 dsRNA-mediated interference in Drosophila cultured cells results in abnormal chromosome morphology and behaviour
As no CAP-D2 mutants were available, dsRNA-mediated interference (RNAi) in S2 cells was used to assess CAP-D2 function. S2 cultured cells were incubated with a 642 bp dsRNA fragment of CAP-D2 and samples were subsequently collected at 24-hour intervals. Two negative controls were performed at the same time: cells were incubated with dsRNA synthesized from a human intron and cells with no dsRNA. Protein extracts from each time point were analysed by immunoblotting (Fig. 3A). At 48 hours CAP-D2 had decreased, by 72 hours it was greatly depleted and by 96 hours it was nearly undetectable (the final depletion was estimated to be 94%, based on titration of signal from known cell numbers). The level of the protein was unaffected in control cells, and chromosomes appeared normal throughout mitosis (Fig. 3B). Immunofluorescence of the CAP-D2 RNAi cells confirmed that these cells had much diminished CAP-D2 chromatin staining. 48 hours after RNAi, CAP-D2 signal had decreased (data not shown) and by 72 hours CAP-D2 protein was hardly detectable on the chromosomes of cells exhibiting abnormal chromosome morphology (Fig. 3C). Chromosomes were condensed relative to interphase but appeared fuzzy, unresolved and hypercondensed (a possible consequence of prometaphase delay). Dramatic failures in segregation resulted in chromosome bridges forming bi-nucleate or even tri-nucleate cells (Fig. 3C4). We also frequently observed daughter nuclei of unequal size, suggesting that chromosomes had not been equally distributed during anaphase.
In order to better examine chromosome morphology, we treated cells with a hypotonic buffer that results in better chromosome spreading and improved observation of individual chromatids. Control cells appeared to have normal chromosome resolution and individual chromatid arms could be easily distinguished (Fig. 3D). Individual chromatids were not visible in hypotonically treated CAP-D2 RNAi cells, and chromosomes appeared as condensed balls of chromatin distinct from one another, with usual levels of P-H3 (Fig. 3E).
The quantification of numerous mitotic parameters is presented in Fig. S2 in supplementary material. CAP-D2 RNAi cells grew more slowly than control cells and cell numbers plateaued 72 hours after treatment. Different time points were assessed for mitotic index, phenotype and the distribution of mitotic phases. The percentage of mitotic cells (those positive for P-H3) after CAP-D2 RNAi increased threefold between 36 hours and 72 hours (Fig. S2B in supplementary material). 25% of the CAP-D2 RNAi mitotic cells were abnormal at 24 hours, while 95-100% were abnormal after 48 hours (Fig. S2C in supplementary material). After staining for Cyclin B, it appeared that CAP-D2 RNAi cells were delaying in prometaphase (Cyclin B positive) – this was corroborated by scoring the anaphase index (Cyclin B negative; Fig. S2D,E in supplementary material). While the dsRNAi experiment was repeated several times, quantification of the phenotypes was performed three times with insignificant deviation between the results. Most of the phenotypic analysis was therefore carried out at 65 or 72 hours after treatment with dsRNA.
As shown in Fig. 3, dramatic chromosome defects were observed after treatment with CAP-D2 dsRNA – apparent failures in chromatid resolution, alignment and segregation. We therefore asked whether centromere resolution was compromised after staining for two centromeric proteins – CID and Scc3/DSA1, a subunit of the cohesin complex. The levels of DSA1 (and CID, not shown) appeared largely unaffected in the CAP-D2 RNAi cells (Fig. 4A). DSA1 was highly concentrated at centromeres during prometaphase and metaphase, and localized between CID spots (Fig. 4B). DSA1 persisted at centromeres until sister chromatid separation at the onset of anaphase, while CID stained centromeres constitutively (Fig. 4B, see also Fig. 3B). DSA1 also stained the spindle poles and microtubules, especially during anaphase (Valdeolmillos, 2004). Centromeres were often stretched in the CAP-D2 RNAi cells (Fig. 4C2-5, see also Fig. 3C), while distinct double dots or cleanly separated sets of dots were observed in control metaphases and anaphases, respectively. Daughter nuclei connected by chromosome bridges still had centromeres extending between them (Fig. 4C5 and Fig. 3C4).
In order to quantify these results, cells were categorised into different groups. 95% of the cells that displayed CID stretching between sister centromeres had lower levels of centromeric DSA1 and thus were considered to be in anaphase (Fig. 4C3-5). 60% of these anaphase figures were in early anaphase, as deduced from the short distance between the centromeres of sister chromatids. The remaining 5% of cells displaying CID stretching were positive for DSA1 and considered to be in prometaphase/metaphase (Fig. 4C2). The CID stretching in these cells was not as extreme as in anaphase cells. A prometaphase/metaphase cell with little or no CID stretching between sister centromeres is shown in Fig. 4C1. We observed that the distance between paired CID spots (measured in prometaphase/metaphase cells) was more than twofold higher in the CAP-D2 RNAi cells than in control cells (1.625 μm versus 0.75 μm, Fig. 4D). This suggests that tension applied upon bipolar attachment accentuates the defect in resolution, and therefore in anaphase the most extreme stretching of CID-stained regions was detected (Fig. 4C5). These results suggest that CAP-D2, and probably the entire condensin complex, is primarily required for resolution of sister chromatids (arms and centromeres) rather than chromosome compaction in Drosophila.
CAP-H/Barren is unstable in cells depleted of CAP-D2
As CAP-D2 is one subunit of a complex, it is possible that interacting proteins may be affected by its depletion. For this reason, the stability and localisation of the other subunits of the complex were analysed upon depletion of CAP-D2. Immunoblotting RNAi cell extracts for CAP-H/Barren revealed that the level of this protein was decreased when CAP-D2 was depleted (Fig. 5A). CAP-H/Barren may be more tightly associated with CAP-D2 compared to the SMC subunits, as suggested earlier. As no sequence similarity exists between the two proteins, the depletion by RNAi should be specifically directed against CAP-D2. In contrast, no substantial change in SMC4 and SMC2 levels was observed (Fig. 5A and Fig. 6A). To examine the localization of residual CAP-H, cells were cytospun onto poly-L-lysine slides and processed for immunofluorescence. In control cells, CAP-H associated with the axial cores of sister chromatids (Fig. 5B). In the CAP-D2-depleted cells, residual CAP-H protein appeared to localise diffusely in the cytoplasm with no CAP-H detected on the chromosomes (Fig. 5C).
Immunolocalisation of SMC4, SMC2 and Topoisomerase II in CAP-D2-depleted cells
To examine the localisation of the SMC4 subunit after CAP-D2 RNAi, we performed double staining with CAP-D2 and SMC4 antibodies. In control cells, SMC4 was localised on mitotic chromosomes (Fig. 5D). In the CAP-D2-depleted cells, SMC4 was still localized to chromatin, even though chromosome morphology was abnormal (Fig. 5E). Distribution was, however, not restricted to an axial core, presumably because of aberrant chromosome organisation. In approximately half of the abnormal cells, the SMC4 signal was more severely reduced (Fig. 5E, top) than in other CAP-D2-depleted cells (Fig. 5E, bottom). This may be the result of uneven RNAi efficiency across the cell population.
We also examined SMC2 level and localisation. As observed for SMC4, the level of SMC2 appeared largely unaffected when CAP-D2 was depleted (Fig. 6A). Because both CAP-D2 and SMC2 antibodies were raised in rabbits, double immunolabelling was not possible. Therefore, we double-labelled cells for SMC2 and topo II, a marker for the chromosome `scaffold'. The overall level of topo II also appeared not to change (Fig. S2F in supplementary material). In control cells, both SMC2 and topo II associated with chromosomes in a striking axial pattern (Fig. 6B). After CAP-D2 depletion, SMC2 localized to chromatin masses, albeit in an unrestricted localization, similar to SMC4. Variation in the level of SMC2 was also observed in these cells: cells with more reduced levels of SMC2 also had lower levels of topo II (Fig. 6C, top versus bottom). Taken together, these results suggest that the axial localisation of SMC4, SMC2 and topo II correlates with proper chromosome resolution, which requires CAP-D2, and by inference CAP-H/Barren (as it is also depleted).
Reciprocal dependence on stability: CAP-D2 is unstable in a barren mutant
Depletion of the CAP-D2 protein by dsRNAi affected the stability of the CAP-H/Barren protein. To test if there was reciprocal stability (i.e. the level of CAP-D2 protein was dependent on the presence of CAP-H/Barren), a barren mutant (barrL305) was examined. barrL305 is a null mutation that causes lethality late in embryogenesis (Bhat et al., 1996). Analysis of barr homozygous embryo extract (∼12- to 13-hours old) showed that when Barren was decreased, CAP-D2 was greatly reduced as well (Fig. 6D). Quantitation of protein levels by phosphorimaging revealed that Barren was reduced eightfold compared to that in wild-type embryos while the level of CAP-D2 was sixfold lower. The fact that both proteins were decreased to nearly the same extent strongly suggests that the two proteins are very probably interdependent for stability. When the same membrane was reprobed for SMC2 and SMC4, it was apparent that their levels were reduced, albeit not to the same extent (Fig. 6D). SMC2 was reduced fourfold and SMC4 3.5-fold, consistent with a fraction of the two SMC proteins participating in a second complex. Quantification was repeated twice in independent experiments. The presence and stability of Scc1/Drad21, a member of the cohesin complex (Vass et al., 2003), and α-tubulin were analysed and found to be unaffected by mutation of CAP-H/barren (Fig. 6D).
The stability of the condensin subunits was also examined in an SMC4 mutant, gluon17C. This is an embryonic lethal null mutation of SMC4 that removes 31 kb including the SMC4 gene (accession AJ543652). Analysis of gluon17C homozygous embryo extract (∼12- to 13-hours old) showed that the SMC4 protein decreased 5.9-fold relative to wild-type embryos (Fig. 6E). Residual protein may suggest that the maternal component is quite stable (and may also explain the SMC4 protein in barr extracts). When the levels of other subunits was assessed, we observed that the SMC2 subunit was decreased to the same extent as SMC4 (5.6-fold), while CAP-D2 and CAP-H/Barren were reduced 4.7-fold (Fig. 6E). These results revealed that in the absence of one of the condensin subunits in Drosophila, the abundance and stability of the rest were affected. Furthermore, the strength of the interactions between the four examined members of the condensin complex supports the notion that the non-SMC subunits, CAP-D2 and CAP-H/Barren, show tight dependencies on one another for stability both in cultured cells and in the organism.
A condensin II complex has been recently identified in vertebrate cells, consisting of SMC2 and SMC4, and three additional non-SMC subunits: CAP-D3, CAP-H2 and CAP-G2 (Ono et al., 2003). CAP-D3 and CAP-H2, but not CAP-G2, have been found by sequence analysis in Drosophila. To assess the role of CAP-D3 in the experiments of this study, we generated an antibody to CAP-D3 (Fig. S1B in supplementary material) and examined localization of the protein during mitosis in S2 cultured cells. Immunostaining revealed that the protein localised at the centromeres, partially overlapping with CID at prometaphase, metaphase and anaphase. However, the signal was reduced by telophase and less restricted to the centromeres (Fig. S3A in supplementary material). We attempted to deplete the protein with two different dsRNAs targeting different regions of the mRNA, but it remained even after 120 hours, suggesting that CAP-D3 may be very stable. Furthermore, chromosome morphology and chromosome segregation appeared normal (data not shown). When the fractions immunoprecipitated from embryos using the CAP-D2 antibody were probed for CAP-D3, no CAP-D3 was detected (data not shown). In addition, the overall localisation of CAP-D3 was not affected in the CAP-D2-depleted cells. CAP-D3 was present at the centromeres in the CAP-D2-depleted cells (Fig. S3B in supplementary material) and when CID was stretched, CAP-D3 was also stretched (Fig. S3B, top panel, in supplementary material). These results suggest that functions of CAP-D2 and CAP-D3 are probably independent of one another. Interestingly, flies homozygous for insertion of a P-element in the third exon of the CAP-D3 gene are viable, but male sterile. This strongly suggests a meiotic role for the CAP-D3 protein, and possibly the condensin II complex in Drosophila.
Depletion of both CAP-D2 and topo II results in chromosome morphology no more severe than that observed in cells depleted of CAP-D2 alone
Since the resolution of sister chromatids is compromised in the CAP-D2-depleted cells and topo II is important for decatenating DNA strands, it is possible that condensin and topo II collaborate to organise mitotic chromosomes. Furthermore, the phenotype observed in this study after CAP-D2 RNAi, was similar to the phenotype obtained after topo II depletion (Chang et al., 2003). In order to determine if these proteins have similar or distinct roles in organising mitotic chromosomes, CAP-D2/topo II double RNAi was performed. Immunoblotting showed that both proteins were greatly depleted 72 hours after treatment (Fig. 7A). The topo II level was not significantly affected after CAP-D2 RNAi, nor was the level of CAP-D2 affected in the topo II-depleted cells [similar to previous observations for Barren (Chang et al., 2003)]. We observed only a twofold increase in mitotic index, using P-H3 as a mitotic marker. Immunofluorescence analysis at 96 hours revealed that cells depleted of both proteins exhibited abnormal chromosome morphology and behaviour no more severe than with either of the single depletions (Fig. 7C). Surprisingly, the ability of chromosomes to compact appeared as it did with either single RNAi. Chromosomes in cells treated with the double RNAi failed to resolve and formed bridges during anaphase. We also noted the unusual phenotype of chromosome arm congression defects observed previously after topo II depletion (Chang et al., 2003) (Fig. 7C, top row).
In order to better assess the chromosome phenotypes of the single and double depletions, metaphase chromosome spreads were performed after cells were arrested in metaphase with colchicine. dsRNA-treated and control cells were cytospun onto poly-L-lysine slides and stained for CAP-D2, topo II and DNA. Control cells displayed the typical `X' shape indicative of proper sister chromatid resolution. Both CAP-D2 and topo II localised axially along the length of the chromosomes and the two proteins partially co-localised (Fig. 8A1-2,B, control). In both single and the double depletions, chromosome morphology was highly abnormal. The structural integrity of these chromosomes was compromised and chromatin appeared torn and stretched. No axial staining could be observed in any of the experiments (Fig. 8A3-5). Individual chromosomes, while compacted, were not resolved and appeared fuzzy and diffuse (Fig. 8B, RNAi panels). Since no difference in phenotype was observed at this level between the CAP-D2, topo II and CAP-D2/topo II doubly depleted cells, we believe that topo II and the condensin complex participate in the same pathway to resolve sister chromatids. We additionally noted no difference in the level of CAP-D3 in either the single depletions or CAP-D2/topo II double depletion, suggesting that this protein was not affected (Fig. S3C in supplementary material). We therefore suggest that additional, as yet unidentified, factors are required to bring about appropriate chromosome condensation during mitosis.
Discussion
In this study we have shown that CAP-D2 is part of a complex containing other condensin subunits (SMC2, SMC4, CAP-H/Barren and CAP-G) in Drosophila. Interestingly, the non-SMC subunits, CAP-D2 and Barren, appear to be more tightly associated with each other than they are with SMC2 and SMC4. This finding is reminiscent of our studies of the cohesin complex in Drosophila where we showed that Scc1/Drad21 and Scc3/SA1 appeared to have a higher affinity for one another than for SMC1 and SMC3 (Vass et al., 2003).
The subcellular localisation of CAP-D2 revealed that the protein was nuclear throughout interphase with levels increasing during S phase, and peaking in late G2/early prophase. This was intriguing as Drosophila SMC4 and Barren are cytoplasmic during interphase and associate with chromatin only during prophase (Steffensen et al., 2001). In human cells, the majority of CAP-D2, SMC2 and SMC4 is cytoplasmic during interphase, finally becoming nuclear during prophase, suggesting that distribution of the holocomplex may be regulated coordinately in human cells (Ball et al., 2002; Schmiesing et al., 2000). Differential localisation of condensin subunits during interphase in Drosophila suggests that CAP-D2 may be sequestered away from SMCs to regulate activity or may be utilized for a function other than that required of other condensin subunits during interphase. It has recently been shown in Xenopus that CAP-D2 (pEg7) along with SMC2 (XCAP-E) localise in the granular compartment of the nucleolus, suggesting that these subunits may be involved in nucleolar organisation and/or ribosome biogenesis (Uzbekov et al., 2003). In S. cerevisiae, the CAP-D2 orthologue Ycs4 is nuclear during interphase, implicated in suppressing inter-repeat rDNA recombination and required for silencing at the mating-type loci (Bhalla et al., 2002). Thus, Drosophila CAP-D2 may be involved in a non-mitotic function – possibly during S phase when its level increases. Alternatively, expression during S phase may be important to `template' the protein appropriately for a mitotic role (Loupart et al., 2000; Shelby et al., 1997). The increase of nuclear CAP-D2 during S phase could be because protein expression is up-regulated or because the protein is preferentially imported into the nucleus. A potential nuclear bipartite targeting signal between amino acids 1347 and 1363 was identified in the Drosophila CAP-D2 sequence (this study). Human CAP-D2 (CNAP1) has a functional bipartite nuclear localisation signal (Ball et al., 2002). During mitosis, CAP-D2 localises to the chromatid cores and remains tightly associated with chromosomes until late telophase/cytokinesis, even as chromosomes are decondensing. In contrast, while DmSMC4 and Barren proteins are loaded onto chromosomes in early prophase, they dissociate from chromosomes late in anaphase/telophase when decondensation begins (Steffensen et al., 2001) (and this study). This distinct timing of condensin subunit dissociation is again suggestive of a role for CAP-D2 beyond that of condensin. In human cells it has been shown that the C-terminal end of CAP-D2 possesses a mitotic chromosome-targeting domain that does not require the other condensin subunits (Ball et al., 2002). Perhaps a similar regulation occurs in Drosophila, and CAP-D2 binds to chromosomes independently of the other subunits.
We have shown that depletion of CAP-D2 by RNAi in Drosophila results in an abnormal mitotic phenotype. The major defect is the loss of sister chromatid (arm and centromere) resolution and the resulting failure of chromosomes to segregate normally during anaphase, a phenotype similar to that observed with barren, DmSMC4 and dCAP-G mutant alleles (Bhat et al., 1996; Dej et al., 2004; Steffensen et al., 2001) and DmSMC4 RNAi (Coelho et al., 2003). Chromosome compaction still occurs, as in cultured chicken cells (Hudson et al., 2003), consistent with the existence of additional factors essential for this process. Although loss of Barren in Drosophila does not affect cell cycle progression (Bhat et al., 1996), loss of SMC4 causes a metaphase delay in embryos (Steffensen et al., 2001) and even more pronounced prometaphase delay (data not shown). Similarly, loss of dCAP-G causes a prometaphase delay in embryos (Dej et al., 2004). Depletion of CAP-D2 in S2 cells results in a prometaphase delay, though cells do eventually undergo aberrant anaphase and cytokinesis. The delay is possibly a consequence of kinetochore checkpoint activation, since chromosomes exhibit elevated levels of BubR1 at centromeres persisting into anaphase and cytokinesis (data not shown). This may, in turn, be a result of altered centromere resolution, manifest as stretched centromeres in anaphase. Similar results have been observed after DmSMC4 RNAi where kinetochores associated with spindle microtubules were stretched polewards, sometimes well beyond the chromatin mass (Coelho et al., 2003).
In C. elegans, the condensin complex is required for the restricted orientation of centromeres towards spindle poles (Hagstrom et al., 2002), while in an S. cerevisiae BRN1 mutant, centromere function is defective and centromeres do not segregate normally (Ouspenski et al., 2000). Recently it has been shown that depletion of condensin I or condensin II in vertebrate cells results in structural distortions of the centromere/kinetochore region that affects the opposing orientation of sister kinetochores and interaction with the mitotic spindle (Ono et al., 2004). Taken together, these results suggest that the condensin complex may have a role in centromere organization and ensuing kinetochore-spindle interactions, which could impact the localization of chromosome passenger proteins. Indeed, we noted a temporal disruption of INCENP localisation following CAP-D2 depletion (date not shown). We observed similar defects in INCENP localization when the DRad21 cohesin subunit was depleted (Vass et al., 2003). These results argue that proper chromosome organisation is necessary for the correct temporal localisation of at least one chromosome passenger protein.
Behaviour of different condensin subunits varies following depletion of CAP-D2. Barren is unstable when CAP-D2 is depleted and any residual protein is cytoplasmic. Barren localisation was also shown to be dependent on SMC4 (Coelho et al., 2003). In a barren mutant, CAP-D2 was unstable, corroborating the interdependence of the two proteins on one another for stability. In contrast, levels of SMC2 and SMC4 were not greatly affected by CAP-D2 depletion and the two proteins still localised to chromatin, albeit in a non-axial fashion. This is probably a reflection of compromised structural integrity of mitotic chromosomes as observed in chicken DT40 cells depleted of SMC2 (Hudson et al., 2003). In S. cerevisiae and S. pombe, the condensin SMCs bind chromatin only in the presence of non-SMC subunits (Freeman et al., 2000; Lavoie et al., 2002). One intriguing question is whether the SMCs remaining on chromosomes are in any way functional? If a condensin II complex exists in Drosophila, then the chromosomal SMCs remaining after CAP-D2 depletion may be members of this complex. Consistent with this, we observed that SMC2 and SMC4 are similarly reduced in a barren mutant (but not as severely as Barren and CAP-D2). We were unable to deplete the CAP-D3 subunit of the Drosophila condensin II complex by RNAi in S2 cells (perhaps because of long half-life), but mutation by P-element insertion in the CAP-D3 gene caused male sterility (data not shown). It remains to be determined what precise role a putative condensin complex II has in meiosis in Drosophila.
In this study, we also performed the first disruption in Drosophila of two components known to be involved in mitotic chromosome dynamics: condensin and topo II. When a double depletion of CAP-D2/topo II was performed, abnormal chromosome appearance and behaviour similar to that observed in the single topo II and CAP-D2 RNAi was observed. Chromosomes were fuzzy and not resolved, and formed chromosome bridges during anaphase and cytokinesis. Surprisingly, the double depletion phenotype was no more severe than in either of the single depletions. This result suggests not only that topo II and the condensin complex participate in the same pathway required to organise mitotic chromosomes into well-defined structures, but also that factors other than these proteins still remain to be identified for their essential roles in chromosome condensation. These may include proteins involved in DNA replication (Dej et al., 2004; Loupart et al., 2000) or diverse factors such as kinases (Yu et al., 2004) and proteases (McHugh et al., 2004), demonstrating that much remains to be learnt about the full requirements for chromosome condensation.
Acknowledgements
We would like to thank the following people for generous gifts of antibodies: Hugo Bellen (Barren), Bill Earnshaw (INCENP), Gary Karpen (CID), Neil Osheroff (topoisomerase II), Jordan Raff (Cyclin B), and Claudio Sunkel (BubR1). We are grateful to Mar Carmena and Chih-Jui Chang for collaborating in the CAP-D2/topo II double RNAi. Ellada Savvidou has been supported by a PhD studentship from the Darwin Trust. Margarete Heck and the research in her laboratory is funded by a Wellcome Trust Senior Research Fellowship in the Biomedical Sciences.