Binding of Drosophila Orc Proteins to Anaphase Chromosomes Requires Cessation of Mitotic Cyclin-Dependent Kinase Activity

Abstract

The initial step in the acquisition of replication competence by eukaryotic chromosomes is the binding of the multisubunit origin recognition complex, ORC. We describe a transgenic Drosophila model which enables dynamic imaging of a green fluorescent protein (GFP)-tagged Drosophila melanogaster ORC subunit, DmOrc2-GFP. It is functional in genetic complementation, expressed at physiological levels, and participates quantitatively in complex formation. This fusion protein is therefore able to depict both the holocomplex DmOrc1-6 and the core complex DmOrc2-6 formed by the Drosophila initiator proteins. Its localization can be monitored in vivo along the cell cycle and development. DmOrc2-GFP is not detected on metaphase chromosomes but binds rapidly to anaphase chromatin in Drosophila embryos. Expression of either stable cyclin A, B, or B3 prevents this reassociation, suggesting that cessation of mitotic cyclin-dependent kinase activity is essential for binding of the DmOrc proteins to chromosomes.

Initiation of DNA replication on the chromosomes of eukaryotic cells requires the prior assembly of the pre-replicative complex (pre-RC), establishing the licensed chromatin state. Of the multiple proteins participating in the stepwise formation of the pre-RC, the initiator origin recognition complex (ORC) makes the first chromatin contacts. In Saccharomyces cerevisiae, the heterohexameric ORC binds to distinct sequence elements in the context of small, modular origins of DNA replication. These are occupied by ORC throughout the cell cycle. Thus, while the initial ORC-DNA interaction can define the position of origins on the chromosomes, it is an unlikely candidate to determine the timing of pre-RC formation, which is confined to the M and G₁ phases of the cell cycle. Instead, this process is mainly controlled by changes in cyclin-dependent kinase (CDK) activity during these cell cycle phases and is part of the safeguard mechanisms against unscheduled DNA synthesis (4, 6, 15, 16, 28).

In contrast to yeast, origins in metazoans are less well defined and their distribution along chromosomes is subject to tissue-specific and developmental control. No conserved DNA sequence motifs that would indicate a sequence-specific DNA binding by initiator proteins have been identified within replication initiation regions. For these reasons, alternative modes of origin selection are discussed (13, 18, 30).

It is nonetheless assumed that the position of metazoan origins is determined by nonrandom ORC-chromatin interactions. To what extent these interactions persist after origin firing has not been extensively studied. The same holds true for the issue of reoccupation of the additional potential ORC binding site emerging immediately after DNA replication initiation. It is clear, though, that ORC has to be chromatin bound at the latest toward the end of mitosis for the subsequent pre-RC assembly steps to occur. When exactly ORC associates with chromatin and, especially, whether ORC is bound to metaphase chromosomes have not been firmly established. Experiments addressing these questions so far have not resulted in a uniform picture. This is most likely due to differences between the various organisms and cell types analyzed and possibly also has been influenced by the experimental approaches used to determine protein localization. The fruit fly Drosophila melanogaster is an excellent experimental system to address these open questions, an answer to which would also advance our understanding of the mechanisms contributing to origin specification. Using properly engineered transgenes, dynamic nuclear processes in live embryos can be readily visualized without perturbation of the homeostasis of early development. Furthermore, the detailed knowledge of many aspects of the Drosophila cell cycle allows integration of results on DNA replication control and the chromosome cycle in the context of proliferative processes in general (25).

The initial goal of our study was to help settle the debate of whether metazoan ORC dissociates from chromatin in mitosis or not. This controversy in the replication field is partly fueled by the use of different experimental systems that generally do not permit an organismal view of this process under physiological conditions. To this end, we analyzed ORC in vivo by creating transgenic Drosophila strains in which a fluorescent protein tag was fused to the D. melanogaster Orc2 (DmOrc2) subunit. DmOrc2 is part of both the hexameric holocomplex as well as the pentameric core complex. The DmOrc2-green fluorescent protein (GFP) fusion integrated quantitatively in these complexes, and the resulting modified initiator protein assemblies (DmORC-GFP) are functional in vivo as shown by genetic complementation of lethal DmOrc2 (k43) mutants. This experimental setting allowed us to follow DmORC-GFP during embryonic cell cycles. Here we show its binding to chromosomes in late anaphase. This interaction of DmORC-GFP with chromosomes is subject to mitotic kinase control as it requires the cessation of CDK signaling. To our knowledge, this constitutes the first model of a metazoan organism in which the intracellular and cell cycle dynamics of a functional DNA replication initiator protein have been traced in vivo.

MATERIALS AND METHODS

Generation of transgenic Drosophila.

The genomic region containing the k43 (DmOrc2) gene was retrieved from the BAC clone “BACR25B05” (BACPAC Resources, Oakland, CA). The clone was digested with BamHI, and the resulting 20-kb fragment containing the gene was integrated into the pGem3Zf(+) vector (Promega). Further digestion with restriction enzymes StuI and XbaI followed by a partial EcoRI digest yielded the 5.7-kb fragment used for the fusion constructs and was inserted into pBluescript KS(+) (Stratagene). The fragment was transferred back into pGem3Zf(+) via a digest with XbaI/SalI. The enhanced green fluorescent protein (EGFP) gene was PCR amplified from pEGFP-C1 (Clontech) using primers creating flanking ApaI and SacII sites (N-terminal fusion) or XhoI sites, respectively. Both the DmOrc2 gene and the EGFP-containing PCR fragment were digested by the respective restriction enzymes and ligated in frame as outlined in Fig. 1A (ApaI [A], XhoI [X], SacII [S], EcoRI [E]). To generate an insertion vector, we exchanged the UAS promoter of pUAST (8) for the modified k43 genomic sequences covering the EcoRI-flanked region outlined in Fig. 1A. The resulting vectors were used for P-element-mediated transposition. Cloning boundaries and PCR-amplified DNA fragments were verified by DNA sequence analysis. Transgenic flies were generated according to standard protocols using a transposon helper plasmid and the white¹¹¹⁸ fly strain. Identification of the integration chromosome and maintenance of transgenic flies was carried out according to standard procedures.

FIG. 1.

Complementation of k43^−/− Drosophila by DmOrc2 transgenes. (A) The genomic region of the k43 gene coding for DmOrc2 was engineered by in-frame insertions of GFP coding sequences as indicated. Promoter (P), intron structure within the DmOrc2 coding region, and selected restriction sites are indicated (see Materials and Methods for details). (B) Crossing scheme for DmOrc2-GFP transgene complementation of heteroallelic k43^−/− flies. (C) Immunoblot analysis of DmOrc2 proteins in Drosophila of the indicated genetic constellation: wild-type (wt) embryos, rescue (r) embryos with no endogenous DmOrc2, and transgenic (tg) embryos with both endogenous and transgenically encoded DmOrc2. DmOrc5 levels are shown as loading controls. C-term. and N-term., C terminal and N terminal, respectively.

Rescue crosses.

The principle scheme of the rescue crosses is outlined in Fig. 1B. Besides the Drosophila lines with a DmOrc2-GFP transgene inserted on the second chromosome, other fly stocks used for the rescue were k43¹/TM6B, k43^γ4/TM6B, yw; Elp/CyO; Ki/TM6y⁺, yw; Pin^88k/CyO-fts-lacZ, and yw; TM3Sb/TM6Tb. For some rescue crosses, the autosomal transgene was initially balanced by a CyO-fts-lacZ chromosome instead of the CyO balancer chromosome depicted. Rescue lines harboring a second chromosome balancer as well as the two k43 mutant third chromosomes showed a slightly delayed development. Adult rescue flies started to eclose at day 11 after egg deposition (AED), as compared to day 10 for w¹¹¹⁸ flies. When rescue crosses shown in Fig. 1B were quantified, the frequency of viable flies with genotypes indicating genetic complementation by the DmOrc2-GFP transgene was according to Mendelian expectations. Only heteroallelic combinations of k43 chromosomes were viable.

Biochemical analyses of DmORC.

Crude embryo nuclear extracts (0 to 12 h) were loaded without any additional fractionation steps on a Sephacryl S-300 column (Amersham) calibrated with thyroglobulin (669 kDa), catalase (232 kDa), and bovine serum albumin (67 kDa) as molecular mass standards. The protocols for nuclear extraction and chromatography were described previously (19). For the salt elution experiment, nuclear pellets were extracted at the indicated salt concentration for 30 min each. Eluted proteins were fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes. Immunoblot analysis was done by using antibodies specific for DmOrc2 and DmOrc5, horseradish peroxidase-coupled goat anti-mouse antibody as a secondary reagent (Pierce) and chemiluminescent detection methods. For coimmunoprecipitation experiments, extracts were precleared with protein A-Sepharose and subsequently incubated with DmOrc5-specific antibodies for 1 h in the presence of a proteinase inhibitor cocktail (Complete; Roche). Protein A-Sepharose was added for another hour. Bound complexes were recovered by sequential washes in phosphate-buffered saline-0.1% NP-40. The immunoprecipitation experiment was performed at 4°C, using siliconized tubes throughout.

Imaging.

Syncytial embryos between 1 and 2 h AED were collected, and embryos corresponding to Bownes' stage 4 were used for imaging. Cellularized embryos up to 6 h AED were collected, and embryos corresponding to Bownes' stages 6 to 8 were used for imaging. After manual dechorionation, embryos were mounted on a Petriperm50 culture dish (Vivascience) in Voltalef 10s oil (Altofina). A coverslip supplied with double-sided tape as a spacer was placed on top. Imaging was carried out on a Leica TCS SP confocal microscope using a 40× (1.25 oil) or a 63× (1.32 oil) objective. For detection of GFP, the 488-nm laser was used and the emission signal was detected at 500 to 530 nm. For detection of monomeric red fluorescent protein (mRFP), a 543-nm laser was used and emission signal was detected at 590 to 700 nm. Images were aquired using the time-lapse feature of the Leica confocal software. Movies were assembled either directly using this software or by using the Advanced Batch Converter (Gold-Software) and Quicktime.

Expression of stable cyclins.

DmORC's association with mitotic chromosomes with dependence on stabilized cyclin expression was analyzed in the following genetic background: heat shock-inducible Drosophila Δcyclin A, Δcyclin B (both 3rd chromosome), and Δcyclin B3 (2nd chromosome) were described previously (44, 45). Embryos analyzed for imaging were from DmOrc2-GFP/His2AvD-mRFP; hsΔCycA flies, the respective hsΔCycB flies, and hsΔCycB3; DmOrc2-GFP/His2AvD-mRFP flies. For efficient cell cycle arrest, flies had to be homozygous for hsΔCyc transgenes.

Heat shock induction was carried out essentially as described previously (38) with the following modifications: Embryos were collected for 30 min at 24°C and aged for 150 to 165 min. After a 30-min heat shock, the embryos were manually dechorionated and mounted as described above. Thus, embryos were imaged at least 30 min after the heat shock. Control experiments revealed that the conditions of the heat shock regimen did not lead to alterations in the cell cycle dynamics of DmORC (data not shown).

RESULTS

Fluorescent protein-tagged Drosophila Orc2 is functional in genetic complementation.

We generated a fusion of Drosophila Orc2 (DmOrc2) with a fluorescent protein tag to monitor its temporal and spatial distribution in vivo. A genomic copy of k43, the DmOrc2 gene (24), was fused either at the 5′ or 3′ end of its open reading frame to the GFP coding sequence (Fig. 1A). These fusion genes with the authentic DmOrc2 promoter region and 3′ untranscribed region were used in P-element-mediated transgenesis of Drosophila. Each of these DmOrc2-GFP rescue constructs could complement the otherwise lethal combination of k43-null alleles, when introduced in that genetic background (see Fig. 1B for the crossing scheme). Therefore, by creating N-terminal as well as C-terminal fusions, which we initially pursued as a precautionary measure against potential steric constraints arising from the fusion moiety, we obtained two independent, functional DmOrc2-GFP transgenes. The rescue strategy which we introduced here for a wild-type allele should also be applicable to the functional analysis of mutant DmOrc2 alleles, facilitating detailed analyses of initiator protein functions in vivo.

Characterization of DmOrc2-GFP transgenic rescue lines.

Next we determined DmOrc2 protein levels in the transgenic lines by immunoblotting of embryo extracts. In non-rescue DmOrc2-GFP flies, the transgenes were expressed at a level below that of endogenous DmOrc2. Upon crossing of these transgenes in a k43-null background, the expression levels of the fusion genes were similar to that of the DmOrc2 gene in wild-type flies (Fig. 1C). Consistent with these results, we found no free, noncomplexed GFP fusion proteins when unfractionated nuclear extracts of early rescue embryos were analyzed by size exclusion chromatography (Fig. 2A). We conclude that DmOrc2-GFP participates quantitatively in complex formation in the rescue animals analyzed, where DmORC-GFP is essential for viability. One could argue that the situation differs in DmOrc2-GFP transgenic animals: i.e., in the presence of endogenous DmORC. Here, DmOrc2-GFP might compete poorly with DmOrc2 for complex integration, which would question our basic assumption of the DmOrc2-GFP-derived fluorescence signal being tantamount to the intracellular localization of DmORC-GFP. To rule out this possibility, we analyzed nuclear extract from transgenic flies in two further experiments. First, when isolated embryonic nuclei were subjected to differential salt extraction, both DmOrc2 and the DmOrc2-GFP fusion protein eluted from chromatin under the same ionic conditions, with the majority of the protein extracted between 220 mM and 320 mM salt (Fig. 2B). Second, efficient complex integration of the fusion protein could also be monitored by coimmunoprecipitation experiments using an antibody against DmOrc5 (Fig. 2C). In both experiments, the ratio between DmOrc2 and DmOrc2-GFP did not change between total (i.e., input) and recovered, complexed proteins, strongly arguing for equally efficient complex integration of these two proteins. In a further experiment, we followed the previously described decline in the expression of DmOrc genes during embryogenesis (5, 9, 19), which was also evident from the abundance of maternally deposited DmOrc2-GFP (Fig. 2D). Confocal fluorescence microscopy of live rescue embryos corroborated this finding, showing a steep drop of the overall GFP signal from early to late embryonic stages. Expression levels of DmOrc2-GFP are therefore sufficient to follow DmORC-GFP in early embryos and also in DNA replication-active tissues in Drosophila larvae and imagos (data not shown). With DmOrc2 as the fusion partner, the DmORC-GFP signal defines the localization of both the holocomplex and the core complex.

FIG. 2.

Characterization of DmORC-GFP. (A) Unfractionated nuclear extracts (NE) of Drosophila embryos were analyzed by S300 size exclusion chromatography, with subsequent immunoblot detection of DmOrc proteins. As for the endogenous DmOrc2 in wild-type embryo extracts (upper panel), the transgenically encoded DmOrc2-GFP of rescue embryos is exclusively present in the high-molecular-mass DmORC fractions. Molecular mass standards (MM) in kDa for SDS-PAGE and the peak elution fractions of marker proteins for the sizing column are indicated. (B) Differential salt extraction of nuclear proteins from transgenic DmOrc2-GFP flies. Cytoplasmic extract (cyto), wash of nuclei (w), and a nuclear pellet fraction after extraction (np) are shown as controls, as well as the total protein counterstain of the blot membrane. Immunoblot analyis of equal fractions of the nuclear extract (ne) at the indicated salt conditions showed that DmOrc2-GFP elutes from chromatin under the same salt conditions as the endogenous DmORC proteins. (C) Coimmunoprecipitations of both DmOrc2 and DmOrc2-GFP with DmOrc5 are equally efficient, using a DmOrc5-specific antibody (α-DmOrc5). Results are shown for 10% of the input nuclear extract and 50% each of the immunoprecipitates. The control shows a parallel experiment using an unrelated antibody. To suppress the signal of the immunoglobulin heavy chain (Ig-hc) close to the position of DmOrc5 on the blot membrane, a protein A-horseradish peroxidase conjugate was used as the secondary reagent in this immunoblot analysis (23). (D) Levels of maternally deposited DmOrc proteins decrease during embryonic development. Shown is an immunoblot analysis of w¹¹¹⁸ wild-type (wt) and rescue embryos collected in the indicated time windows AED. For the rescue embryos, two live, dechorionated specimens from the indicated time windows were imaged by confocal fluorescence microscopy for GFP signal detection. A single Z-scan is shown (scale bar, 50 μm).

Dynamic changes of the intracellular distribution of DmORC-GFP during the cell cycle.

DmORC-GFP localization was analyzed by time-lapse video microscopy of cellularized rescue embryos. The protein complex was strongly enriched in nuclei, which increase in size as cells pass through interphase, resulting in the gradual “dilution” of the GFP signal intensity (see Movie S1 in the supplemental material). DmORC-GFP was also dispersed during early mitosis (compare Fig. 3B to A), with no obvious subcellular enrichment. A chromosomal GFP signal appeared on anaphase chromosomes (Fig. 3C). As cells move toward telophase, the dispersed DmORC-GFP is completely chromatin bound again (Fig. 3D). Both N- and C-terminal fusion proteins showed identical oscillatory changes in these localization studies. In the absence of any other visual marker in these GFP-only transgenic animals, it was difficult to precisely align DmOrc2-GFP loading to the course of mitotic progression. Therefore, recording more details of this process required costaining of the chromosomes. To achieve this without additional experimental interventions before and during imaging, we used a second transgenically encoded marker protein, a Drosophila histone 2A variant fused to mRFP (His2AvD-mRFP) (43). His2AvD is chromatin bound throughout the cell cycle (11), providing an omnipresent chromatin marker. In doubly transgenic flies, the subcellular distribution and the dynamics of the GFP signal was indistinguishable from that observed in the DmOrc2-GFP rescue lines. Figure 4A depicts the localization of DmORC-GFP relative to chromatin at the major cell cycle stages. Imaging of cells progressing through mitosis illustrates that DmORC-GFP disperses in early mitosis (Fig. 4B) and subsequently reaccumulates on the chromosomes as the cell cycle moves toward telophase (Fig. 4C). Pictures shown in Fig. 4 were taken from Movie S2 in the supplemental material. Next we asked if the dynamic shuttling of DmORC changes during embryonic development, given the major changes in cell cycle control and origin determination between the syncytium and older blastula embryos (42). Figure 5A shows that DmORC-GFP is nuclear throughout interphase of syncytial cell cycles (see also Movie S3 in the supplemental material). The increase in nuclear volume during interphase is paralled by a decrease in nuclear GFP signal intensity. Preliminary quantitative evaluation of this dilution process suggests that the GFP signal, when integrated over the expanding nucleus, stays approximately constant during interphase (data not shown). As expected, throughout mitosis the histone signal intensifies in more condensed chromatin. In contrast, the DmOrc2 signal is diffusely distributed in the nuclear area during early mitotic stages, with no sign of chromatin enrichment. It should be noted that in this stage of development the nuclear membrane becomes permeable to large molecules already very early in mitosis (35). By late anaphase, a reassociation of DmORC-GFP with chromatin is clearly visible (Fig. 5A). It appears that the loading process is not synchronous along chromosomes but initiates at the centromeric part and proceeds along the axis (Fig. 5B). Whatever the biological relevance, with respect to its cell cycle timing and subchromosomal localization this process is strikingly similar to the disappearance of histone H3 phosphorylation in precellular blastoderm stages of Drosophila embryos (47). This histone modification also reflects an increase in chromatin compaction, and its reversal toward the end of mitosis could also contribute to DmORC-GFP chromosomal rebinding. It is unclear if these findings also relate to the preferential association of Orc2 to centromeric heterochromatin found in cell lines at earlier mitotic stages (36, 39), as we did not see any distinct DmORC-GFP signal on metaphase chromosomes in live embryos.

FIG. 3.

Dynamic changes in the intracellular localization of DmORC-GFP. A cellularized DmOrc2-GFP rescue embryo was imaged to monitor changes in the DmORC signal over time. (A) Highlighted by an arrow is the interphase cell to be followed along the cell cycle. A nuclear DmORC signal is visible. Bar, 10 μm. (B) About 5 min later, DmORC is dispersed (with the arrow pointing at the outer border of the DmOrc2-GFP-positive area), typical for early mitotic cells. This cell cycle staging can be made as 1 min later an anaphase chromosome figure with an intense fluorescent signal becomes visible (arrowheads), with part of the DmORC-GFP still dispersed (C). As the cell moves toward telophase, virtually all of the detectable DmORC-GFP is recruited to the chromosomes (D). Images in this figure are video stills taken from Movie S1 in the supplemental material.

FIG. 4.

Intracellular distribution of DmORC-GFP in cellularized embryos. (A) Optical section of an embryo with asynchronously dividing cells. Nuclei in different stages of the cell cycle are marked: interphase (black arrow), metaphase (white arrow), late anaphase (white arrowheads), and telophase (black arrowheads). The scale bar is 20 μm. (B) Enlarged picture of a nucleus moving into metaphase. (C) Enlarged picture of a nucleus moving out of metaphase. For panels B and C, the scale bar is 5 μm and the elapsed time from the first picture is indicated. The color coding of the fluorescent proteins is as in panel A. Images in this figure are video stills taken from Movie S2 in the supplemental material.

FIG. 5.

DmORC-GFP in syncytial embryos. (A) Cell cycle distribution of DmORC (DmOrc2-GFP [shown in green]) and chromatin (His2AvD-mRFP [shown in red]) starting at interphase of cycle 13 (scale bar, 10 μm). Images in this figure are video stills taken from Movie S3 in the supplemental material. (B) Merging of the DmOrc2-GFP and His2AvD-mRFP signals. DmORC-GFP is initially visible at the centromeric region of anaphase chromosomes. Quantification of relative fluorescence intensity along the indicated white lines is shown on top of the pictures (Leica confocal software, Kernel 7).

DmORC-GFP binding to late mitotic chromosomes depends on the cessation of CDK activity.

The observed binding of DmORC-GFP to late mitotic chromosomes raises the question of possible control mechanisms triggering the changes in protein localization in preparation for pre-RC assembly. In Drosophila, the ordered progression through mitosis requires the stepwise cessation of different CDK1/cyclin activities. This process is controlled by the anaphase-promoting complex/cyclosome (APC/C)-mediated ubiquitination and subsequent proteasomal degradation of the regulatory subunits of CDK1, the mitotic cyclins A, B, and B3 (for review, see references 25 and 48). The cell cycle can be arrested by expression of stable mitotic cyclin mutants (Δcyclins), resulting in sustained CDK activity (38, 44). We asked if loading of DmORC-GFP on chromosomes was prevented under these conditions or if the initiator protein complex followed its dynamic relocalization pattern independent of mitotic CDK activity. To this end, we crossed Δcyclin transgenes into a DmOrc2-GFP/His2AvD-mRFP background. Expression of the proteolysis-resistant cyclins was under the control of a heat shock promoter. Without heat shock induction, none of the embryos transgenic for Δcyclins showed a discernible cell cycle phenotype (Fig. 6A and see Movie S4 in the supplemental material). Upon temperature shift in cellularized embryos, expression of the truncated cyclins prevented the progression of the cell division cycle out of metaphase (Δcyclin A), early anaphase (Δcyclin B), or late anaphase (Δcyclin B3), either by an arrest or at least a very substantial cell cycle delay. Mitotic progression could be followed by the His2AvD-mRFP signal. Throughout the embryo, arrest in mitosis was far from complete. This is likely due to cell-to-cell variations in transgene expressivity or penetrance, in our hands precluding a conclusive analysis of potential direct changes in the phosphorylation status of DmORC subunits. In cells not affected by a mitotic cell cycle block and also in cells escaping from a transient cell cycle arrest, the DmORC-GFP signal reoccurred on the segregating late mitotic chromosomes (see Movies S5 to S7 in the supplemental material), as shown before for DmOrc2-GFP transgenic flies with endogenous cyclin control. This demonstrated that in this genetic background, control over DmORC-GFP localization was not disturbed whenever a given cell was able to progress beyond its expected mitotic cell cycle arrest point.

FIG. 6.

CDK control over DmORC-GFP binding to mitotic chromosomes. Transgenes coding for the indicated stable cyclins under heat shock control were crossed into a doubly transgenic DmOrc2-GFP/His2AvD-mRFP background. (A) The GFP signal and cell cycle distribution of cellularized non-heat-shocked (−hs) embryos (as shown by a Δcyclin B transgenic embryo with video stills taken from Movie S4 in the supplemental material) were indistinguishable from those of embryos without the stable cyclin transgene (see Fig. 4A for comparison). The larger left panel corresponds to the first frame of Movie S4 in the supplemental material. The smaller right panel rows are follow-ups of an individual nucleus (highlighted in the overview by a white arrowhead) as it moves out of metaphase. The upper row shows merged GFP-RFP channels, and the lower rows shows the GFP channel only, corresponding to the DmORC-GFP signal. (B) Mitotic arrest figures of the indicated stable cyclins after heat shock induction (larger left panels). Aside from the metaphase-arrested Δcyclin A embryo, the first picture (0′) corresponds to a time-lapse frame of the depicted nucleus about 1 min before exit from metaphase. Video stills are taken from Movies S5 to S7 in the supplemental material. The scale bars are 10 μm throughout.

Expression of each of the stable mitotic cyclins prevented the chromosome binding of DmORC-GFP (Fig. 6B). In the case of Δcyclins A and B, the cell cycle arrest points clearly preceded the microscopically determined mitotic stage in which DmORC-GFP binds to chromosomes. For Δcyclin B3, though, the arrest point roughly coincides with the timing of DmORC-GFP binding in unperturbed cell cycles. Nevertheless, even after prolonged arrest these anaphase chromosomes did not show any GFP signal. We conclude that both the general oscillation in DmORC-GFP localization and the precise timing of chromatin binding require the cell-cycle-controlled changes in CDK activity during mitosis.

DISCUSSION

The elucidation of the mechanisms responsible for the cell-cycle-dependent changes in the intracellular localization of ORC will be essential for a comprehensive understanding of the cascade of events leading to DNA replication initiation. Such analyses could also be critical for developing new concepts of origin specification. We decided to use Drosophila as an experimental model to follow early events of pre-RC formation, expressing a transgene coding for one of its ORC subunits, DmOrc2, fused to a fluorescent protein tag. DmOrc2 was chosen for two reasons. First, it constitutes an essential part of the ORC core (for review, see reference 14). As such, we expect it to better reflect the localization of the complex in its origin-defining function compared to peripheral ORC subunits. Second, several null alleles of k43, the DmOrc2 gene, have been identified (24, 27), allowing us to pursue a genetic complementation strategy to verify transgene functionality. Rescue transgenes consisted of genomic copies of the DmOrc2 gene in which GFP was inserted in frame, coding for either an N- or C-terminal fusion of the ORC subunit. Using complementation as the most conclusive genetic criterion, both fusions proved functional. This indicates DmORC's flexibility to accommodate substantial heterologous protein moieties, exemplified here by GFP attached in two independent positions within the complex structure. In particular for a protein like DmOrc2, functioning as an integral part of a multiprotein complex, it was essential to avoid an overexpression situation, as in the absence of authentic binding partners its fluorescence signal is expected to mislocalize. Thus, fly lines with physiological DmOrc2 expression levels and the quantitative participation of the fusion protein in the holo- or the core complex were a prerequisite to address the cell cycle events governing DmORC's interaction with chromosomes by a biologically meaningful experimental approach. Both criteria were met in the DmOrc2-GFP transgenic as well as the rescue lines. Thus, for all parameters tested, the Drosophila model reflected faithfully the behavior of endogenous DmOrc2, allowing the visual tracing of DmORC-GFP. The use of the term “DmORC-GFP” throughout this article therefore refers to both the DmORC core and the holocomplex formed upon DmOrc1 association, which cannot be distinguished by the imaging approach chosen by us.

Aside from ensuring the congruence between transgenic and endogenous ORC subunits, our experimental strategy of tracking ORC during the cell cycle also avoids ambiguities sometimes associated with the fixation or physiological stressing of cells and organisms. Notwithstanding such methodological issues, the observed dynamics of ORC-chromatin interactions reported previously suggested a significant divergence between different biological systems (see reference 14 and references therein). In yeast, ORC binds chromatin throughout the cell cycle, including metaphase, as has also been reported for embryonic Drosophila (36) and mammalian ORC core subunits (22, 26, 29, 33). Other studies came to the conclusion that members of the mammalian core complex are mostly excluded from metaphase chromatin (1, 32, 39), similar to Xenopus (12, 41) and also Drosophila larval neuroblasts (27). Based on immunolocalization studies, the latter analysis came also to the conclusion that DmOrc2 accumulates on late anaphase/telophase chromosomes, similar to our findings. Differences in the reported localization patterns of ORC can possibly be explained by cell-type-dependent or, in particular, by interspecies variations in the control mechanism of the cell division cycle. In this context, the modest conservation of ORC subunits between species is interesting to note (7). Aside from that, conclusions about ORC localization from these studies have to take the strengths and limitations of the applied methodologies into account. For example, while biochemical fractionation protocols can provide information about the affinity of protein-DNA interactions, they cannot address the issue of intrachromosomal protein distribution. For in vivo imaging, this situation is reversed, with the additional advantage of precisely capturing cell cycle changes of protein localization. For immunolocalization studies, some of the observed variations might also be attributed to the precise protocol applied to analyze ORC localization on chromosomes. We were never able to control fixation conditions such that we could come to unequivocal conclusions about DmORC subunit localization in whole Drosophila embryos, occasionally observing a signal on fully condensed metaphase chromatin. The same phenomenon was observed for live embryos under anoxic conditions (data not shown). Fixative-sensitive intracellular distribution patterns of facultative DNA binding proteins have been reported before (21, 34, 36), and even experimental manipulations like hypotonic swelling of live cells can trigger a relocalization of such proteins (10). These examples emphasize the advantage of in vivo imaging under physiological conditions, allowing us to avoid such experimental ambiguities to the greatest possible extent.

From the imaging analysis in our in vivo model, we conclude that the majority of DmORC-GFP is displaced from the chromosomes in early mitosis and diffusely distributed throughout the cell without any recognizable localization pattern. Therefore, current models of the embryonic Drosophila ORC cycle should be scrutinized when they place the core DmORC on mitotic chromosomes. Toward the end of mitosis, DmORC-GFP is chromatin bound again, and this relocalization seems to be quantitative within the detection limits of the methodology employed. We did not observe any principal differences in this dynamic behavior of DmORC-GFP between syncytial and cellularized stages of embryonal development.

Proteolytic control of ORC core subunits has not been reported so far. In line with this lack of evidence, our study does not indicate that DmORC-GFP levels are subject to mitosis-specific protein degradation (as are other regulators of cell cycle progression [see below]), with the fluorescence signal of DmORC-GFP clearly visible in early mitosis, before gradually refocusing on late mitotic chromosomes. This entire process might be completely attributed to control over intracellular localization of DmORC-GFP during the cell cycle. However, while a substantial resynthesis of DmORC core subunits appears unlikely given the observed timing of this process, in particular with the additional requirements for complex assembly and chromophore maturation, we cannot rule out a partial destruction of core DmORC subunits, followed by chromosomal recruitment of DmORC from cytoplasmic pools at the onset of a new round of pre-RC formation.

Origin specification and pre-RC assembly in eukaryotes start with the chromatin binding of ORC. We showed the cell-cycle-dependent changes of DmORC-GFP localization in embryos. Its rapid accumulation on chromosomes is detectable by late anaphase when CDK activity drops to the low levels observed in the late M and early G₁ phases. The dependence of DmORC-GFP chromosome binding on low CDK activity was established by following the fluorescence signal upon cell cycle arrest in response to the expression of stable mitotic cyclins A, B, and B3, which are not subject to proteasomal degradation. Their presence prevented chromatin binding of DmORC-GFP. Previous reports describing the reloading of ORC to late mitotic chromatin in various cellular systems of metazoan origin have implicated mitotic CDKs in this process, supported by corresponding biochemical analyses (see below). In Drosophila, it is known that the expression of individual stable cyclins does not interfere with the cell-cycle-controlled degradation of the endogenous cyclins (44). Thus, our in vivo analysis allows us to extend the general assumption of a role for mitotic CDK involvement in triggering the start of pre-RC assembly to specifically conclude that all mitotic CDK/cyclin activities have to cease for DmORC-GFP to become chromatin bound.

How can this dynamic behavior of DmOrc2 be interpreted in the light of previous observations regarding the APC-dependent degradation of DmOrc1 in late mitosis, only to reemerge in late G₁ (2, 3)? Even when considering that metazoan Orc1 often shows expression, localization, and turnover patterns independent of other ORC subunits, reflecting temporal events in the control over ORC activity (14), the almost converse mitotic shuttling patterns of DmORC subunits are somewhat surprising. It should be noted, however, that DmOrc1-GFP could also be detected on telophase chromosomes before being degraded (2). Most studies of metazoan ORC concur that Orc1 is essential to establish initial DNA binding of ORC and subsequent steps of pre-RC formation (see reference 15 and references therein), supported by the recent finding that elevated Orc1 levels can actually promote binding of endogenous Orc proteins during late mitosis (32). It is conceivable that in Drosophila this process takes place during a brief time window in late mitosis and could be sufficient to trigger the recruitment of other pre-RC proteins, which according to most analyses occurs prior to late G₁. Alternatively, the remaining chromatin-bound DmORC core might be sufficient to promote completion of the pre-RC. From these lines of reasoning, it is already obvious that further experiments, in directly comparable settings for both the experimental protocols followed as well as for the cell types and developmental stages analyzed, will be required to resolve this issue. This will be facilitated by the availability of Drosophila orc1^−/− lines as recently described (37).

After this initial step in pre-RC assembly, other replication initiation proteins have to be loaded on chromosomes for them to become licensed for replication. Among these factors is the heteromultimeric minichromosome maintenance (MCM) complex, associated with a DNA helicase activity (31). Previous immunolocalization studies of the association/dissociation cycles of Drosophila MCM demonstrated their binding to mitotic chromatin upon cell cycle arrest by expression of stable cyclin B, corresponding to early anaphase stages (46). Assuming an unconditional requirement for prebound ORC for MCM chromatin binding, our data would predict MCM binding at later cell cycle stages, after cessation of mitotic CDK activity. At first glance, these results on the timing of MCM-chromatin association might not be easy to reconcile with our findings but can be explained by (i) the influence of the imaging methodology as outlined above for DmORC localization, (ii) different sensitivity thresholds of the detection systems, or (iii) potential uncharacterized effects of the stable cyclin-CDK complexes used in both studies (for discussion, see reference 44). In any case, we do not see a real discrepancy, as chromatin loading of MCM proteins in unperturbed cell cycles was only evident in late anaphase/telophase (46), fully compatible with our results for DmORC in both perturbed (i.e., stalled by stable cyclin expression) and unperturbed cell cycles.

It will be interesting to determine if this binding is partly responsive to potential changes in chromosome structure occurring as mitotic chromosomes pass toward telophase or whether DmORC responds directly or indirectly to changes in the kinase environment of late mitotic cells. The latter possibility would argue that a decrease in DmORC's phosphorylation state results in its increased affinity to chromatin. This scenario appears attractive as Remus et al. (40) demonstrated that in vitro binding of DmORC to DNA is strongly diminished whenever it is phosphorylated by various CDK/cyclin activities. Combined with our cytological studies, these findings make mitotic CDKs attractive candidate kinases for actively suppressing DmORC binding to chromatin. This view is also in line with the localized cyclin destruction in syncytial cell cycles of Drosophila (17, 20, 47). The resulting abrogation of CDK1 activity in the vicinity of the mitotic spindle can be monitored by the distribution of phospho-histones (47), akin to the observed gradual rebinding of DmORC, starting from the centromeric regions of anaphase chromosomes.

In summary, we report the spatial and temporal dynamics of the initiator protein ORC in a live metazoan organism. Along with the cell cycle, ORC periodically associates with and dissociates from chromatin. The initial interaction in preparation for the next chromosome cycle occurs in late anaphase. This binding of ORC to chromatin depends critically on the cessation of mitotic cyclin activity, linking this first step of replication licensing to the CDK-driven control pathways of cell cycle progression. Finally, it is obvious that different mechanisms evolved between species controlling the activities of ORC. While all of them are compatible with the general requirements for origin definition, pre-RC assembly, and the prevention of rereplication, it cautions against the extrapolation of findings from one experimental system to another. This underscores the value of multipurpose in vivo models like the one described here, allowing a comprehensive approach for probing ORC functions. Its use should not be restricted to further exploring ORC in DNA replication initiation, but it should also be useful to study ORC's role in proliferation and in the development of an organism.