Dynamics of the Genome during Early Xenopus laevis Development: Karyomeres As Independent Units of Replication

Embryos were grown in 0.1× Barth's medium as described (28). When necessary, DNA replication was blocked by a 45-min incubation in 0.1× Barth's medium containing 150 μg/ml aphidicolin. G2/prophase-synchronized embryos were obtained by 45 min incubation in 0.1×Barth's medium containing 150 μg/ml cycloheximide.

The interphase egg extract was prepared as described (7). Metaphase extracts refer to cytostatic factor (CSF) extracts (38), except that the CSF extract was prepared at 4° instead of 16°C.

Embryo extracts containing synchronized G2/prophase nuclei were prepared from 1,024-cell stage embryos. Embryos were dejellied with 2% cysteine HCl, pH 7.9, at the 256-cell stage and then incubated at the 512 cell-stage in 0.1× Barth's medium containing 150 μg/ml cycloheximide during 45 min. After several washes in 0.1× Barth's medium, embryos were homogenized through a 1-ml Gilson tip (Gilson, France) and centrifuged at 4°C for 10 min at 10,000 g. The supernatant was collected, avoiding the top lipid layer. In this condition, the endogenous embryonic nuclei remain in the supernatant. The extract was further supplemented with 8% glycerol, aliquoted, and then stored at −80°C. Progression of G2/prophase nuclei to mitosis was obtained by mixing 2 vol of the embryo extract with 8 vol of metaphase-arrested egg extract. Under these conditions, the nuclear envelope breaks down after a 30-min incubation at 23°C. Metaphase arrest was checked by microscopic examination of aliquots stained with Hoechst 33258 (Sigma Chemical Co., St. Louis, MO). Completion of mitosis and entry into S phase was induced by addition of 0.6 mM CaCl₂ at that stage.

Extracts containing nuclei were diluted 10-fold with fixation buffer (FB)¹ containing 50 mM KCL, 2% sucrose, 5 mM MgCl₂, 0.5 mM EDTA, 4% formaldehyde, 20 mM Hepes, pH 7.6. After a 30-min incubation at room temperature, the nuclei were overlaid on a 0.6-M sucrose cushion and then centrifuged onto glass coverslips. Samples were then treated for immunochemistry analysis. Alternatively, samples were placed on slides and directly fixed with an equal volume of 2× FB containing 1 μg/ml Hoechst 33258. Immunocytochemistry on embryos was carried out by removing dejellied embryos at different stages after fertilization that were then placed on coverslips. The medium was carefully removed and 5 μl of FB without sucrose was added to cover the embryo. After 10 min of fixation at room temperature, a second coverslip was overlaid onto the embryo which was consequently squeezed. Samples were frozen on dry ice and both coverslips were separated with a razor blade before a 3-min postfixation in −20°C methanol. Rehydration was done in PBS at room temperature. Embryos were subsequently treated for immunocytochemistry analysis as for extracts.

DNA synthesis in embryos was analyzed by 10 min of BrdU pulses, each for 5 min within a single cycle. Embryos were incubated for 10 min in 0.1× Barth's medium containing 4 mM BrdU and 0.5% DMSO. Embryos were taken between the 8- and 16-cell stages and between the 512- and 1,024-cell stages. They were first treated for 45 min in 0.5× PBS, 2 M HCl, and then further postfixed in methanol at −20°C.

Embryos or isolated nuclei on coverslips were incubated for 1 h at room temperature in PBS 2% BSA, 0.1% Tween 20 to block nonspecific interaction of the antibodies. The incubation with specific antibodies was overnight at 4°C in PBS, 2% BSA. After several washes in PBS, 0.1% Tween 20, the second FITC-conjugated or Texas red-conjugated streptavidin was added, following instructions from the manufacturers. To reveal biotin-dUTP, FITC or Texas red-conjugated streptavidin was mixed with the second antibody at the appropriate dilution. DNA was stained with either a 10-min incubation in 1 μg/ml Hoechst 33258 in PBS, or a 5-min incubation in 10⁻⁶ M propidium iodide in PBS. After several washes in PBS containing 0.1% Tween 20, slides were mounted in Citifluor antifading reagent (Canterbury, UK).

Samples were examined using a BDS enhanced video microscope (Biological Detection Systems Inc., Pittsburg, PA). The system is based on a Zeiss Axiovert (Photometrics, Ltd., Tuscon, AZ) coupled to a charged-coupled device camera with low light level admission. Confocal microscopy was carried out using the TCS confocal imaging system (Leica Instruments, Heidelberg, Germany), equipped with a 63× objective (plan apo; NA = 1.4). For FITC and propidium iodide excitation, an argon–krypton ion laser adjusted to 488 and 568 nm was used. For each optical section, double fluorescence images were acquired in sequential mode (i.e., FITC first and propidium iodide second) The signal was treated with line averaging to integrate the signal collected over eight lines in order to reduce noise. The confocal pinhole was adjusted to allow a minimum depth of field. A focal series was collected for each specimen. The focus step between these sections was generally 0.5 μm. Focal series were then processed to produce single composite images (extended focus), combining high spatial resolution and high depth of field.

When Xenopus embryos are analyzed during the early cleavage stage, specific nuclear structures are observed during each mitosis. During anaphase, each individual chromosome becomes surrounded by an envelope forming micronuclei-like structures distinctly observed at the beginning of telophase (Fig. 1,A, a and b). These individual vesicles grow and fuse during telophase (Fig. 1,A, b and c) to lead to more typical interphasic nuclei (Fig. 1, A d). The formation of the nuclear membrane around karyomeres and during the reconstruction of the nucleus has also been demonstrated by the membrane lipids dye DiOC6 (Fig. 1,B), indicating that during most of the cell cycle chromosomes behave as relatively independent micronuclei structures. Karyomeres are first detected at the two-cell-stage embryo, when the division furrow forms and represent the initial step of nuclear reconstruction during the entire segmentation period. At 8 h after fertilization, these structures disappear and nuclear reconstruction typical of somatic cells is observed (Fig. 1,A, e and f and Fig. 2). We conclude that karyomere formation is strictly specific to early development. This specific pathway was not observed later in embryonic development, nor in Xenopus somatic cells in culture. The abrupt disappearance of these structures after cycle 13 (Fig. 2) defines a novel transition occurring at blastula stage, which might reflect changes previously observed both in replication and transcriptional programs during development (21, 42).

These specific features led us to consider to what extent the formation of karyomeres might explain the accelerated rate of DNA replication during early development. The formation of a nuclear envelope is essential to the initiation of DNA replication (4, 29, 40, 44) and also to prevent illegitimate reinitiation within the same cell cycle (4). In its absence, prereplication foci can form but DNA replication is not initiated (1). One essential component of the nuclear envelope that is required for the initiation of DNA replication is the lamina (16, 35, 44) and its presence or absence in karyomeres was not known. We used an antibody directed against the embryonic lamins to ask if they were incorporated into the karyomeres during the early cell cycles. Distinct small lamin spots were detected on mitotic chromosomes during anaphase (Fig. 3,A). The formation of the nuclear lamina proceeds asynchronously around each chromosome, with both punctated and continuous layers of lamina on karyomeres migrating towards both poles of the spindle (Fig. 3,B). Individual karyomeric vesicles grow and continuous lamina layers are observed around every karyomere (Fig. 3, C and D). Then the fusion of karyomeres occurs but internal folds remain detectable until late stages of nuclear reconstitution (Fig. 3, E and F ). This pathway for lamina establishment was entirely specific for pre-midblastula transition (MBT) mitotic stages.

The observation of individual chromosomal vesicles surrounded by a lamina suggested that DNA replication might be underway before reconstruction of the nucleus and indeed before completion of mitosis. We used antibodies against three markers specific for different steps in DNA replication. The minichromosome maintenance protein XMCM3, a component of Xenopus licensing activity (6, 27, 34), and replication protein A (RPA) associate with sperm chromatin in egg extract (1, 49). XMCM3 was detected on migrating chromosomes during early anaphase (Fig. 4,A, a and b) and during nuclear reconstruction (data not shown). A similar staining was observed for other members of the minichromosome maintenance (MCM) family of proteins (data not shown), identifying an early step in replication origin formation. RPA associates with prereplication centers, before the assembly of the nuclear envelope and the initiation of DNA replication (1, 49). In vivo, RPA was first detected during anaphase as a punctated staining on chromosomes (Fig. 4,B, a and b). When the nuclear lamina is formed around karyomeres, an intense and uniform RPA staining characteristic of ongoing DNA replication (1, 49) was observed (Fig. 4,B, c and d). Therefore, initiation of DNA replication starts in vivo at anaphase, and elongation begins in early telophase when chromosomes still behave as independent units. Synthesis of DNA was directly evaluated by incubating embryos with BrdU for 10-min pulses, and was detected in karyomeres during early telophase (Fig. 5,A). No incorporation of BrdU was observed in anaphase, confirming that DNA synthesis observed in karyomeres was not due to BrdU incorporation during the preceding cell cycle. Finally, proliferative cell nuclear antigen (PCNA), a processivity factor for DNA polymerase-δ was used as a specific marker for elongation of DNA synthesis (2, 5, 18), at a stage when the karyomeres are still distinct vesicles. PCNA was detected as a bright and non-punctated staining of karyomeres during early telophase, indicating that DNA synthesis was already occurring in karyomeres as the division furrow is just starting (Fig. 5,B, a and b). At this stage the mitotic spindle has not totally regressed (Fig. 5 B, c and d) and the blastomere has not yet divided. We conclude that DNA replication occurs in karyomeres while both mitosis and nuclear reconstruction are not yet completed.

Although karyomeres are observed as the initial step of nuclear reconstruction during the entire segmentation period, they were not observed in cell-free systems derived from Xenopus eggs, except in rare cases when a second cycle was observed (20). Nevertheless, Xenopus egg extracts are very efficient in DNA replication of demembranated sperm nuclei added in vitro (3, 19, 32). One hypothesis to explain this discrepancy could be that passage of prophase nuclei through mitosis is a crucial step in the formation of karyomeres during development.

To address this question, we used a novel Xenopus in vitro system prepared from early embryos at the 1,024-cell stage that were synchronized at the G2 phase by addition of cycloheximide. Such embryonic cells exhibit two centrosomes composed of two centrioles and pericentriolar material (14). The preparation of the extracts ensured that they already contained endogenous G2 nuclei. The addition of a mitotic extract containing MPF activity to this extract allowed the nuclei to progress synchronously into mitosis in vitro (Fig. 6,A, a and b). Nuclei arrested again at this stage, with a disassembled nuclear envelope and condensed chromosomes arranged in a metaphase plate with a typical bipolar spindle (Fig. 6,A, b). Completion of mitosis and progression to the following cell cycle then can be induced by addition of Ca²⁺. Chromosomes segregate (Fig. 6,A, c), form karyomeres (Fig. 6,A, d), and then fuse together to produce a G2-like nucleus 45–60 min after Ca²⁺ addition (Fig. 6,A, e and f). The extent of DNA replication, as measured by ³²PdATP incorporation with TCA precipitation, was consistent with one round of DNA synthesis and hence the completion of S phase (data not shown). We conclude that this acellular system faithfully mimics the successive steps of the organization of the nucleus observed in vivo during the early embryonic cell cycle, and that it is particularly suitable for the analysis of the transition between mitosis and the onset of the S phase. When biotin-dUTP was added to the extract, DNA replication was first detected during telophase, starting asynchronously in karyomeric vesicles before nuclear reconstruction (Fig. 6,B), consistent with the asynchrony in the establishment of the lamina around individual karyomeres (refer to Fig. 3,B). Initiation of DNA replication starts on karyomeres whereas chromatin is not fully decondensed and presents a punctated RPA staining (Fig. 6,B). Other karyomeres display decondensed chromatin and homogenous biotin-dUTP and RPA staining characteristic of later stages of replication (Fig. 6 B). Karyomeres can be at different stages of replication when they fuse, indicating that during early development the unit of replication is not the nucleus but the chromosome.

When aphidicolin is added to the extract, DNA synthesis is blocked but nuclear reconstruction from segregated karyomeres nevertheless occurs (Fig. 7, C and D). The same result was observed in vivo in early embryos in the presence of aphidicolin (data not shown). We conclude that during early development, karyomere formation and reconstitution of the nucleus from karyomeres are not directly linked to the DNA replication process.

Nocadazole added to the extract prevents spindle formation and chromosome segregation after dissolution of the nuclear membrane. In these conditions, karyomeres were not formed but a nuclear envelope develops around the chromosomes (Fig. 7, E and F), as during telophase in post-MBT embryos or in somatic cells. We conclude that karyomere formation requires the segregation of chromosomes at each mitosis. This explain the observation that sperm nuclei do not form karyomeres in an interphasic extract, as chromosomes do not segregate in this context.

The formation of karyomeres during early development is probably permitted by the maternal accumulation of proteins and membrane components required to assemble functional nuclear envelopes for ∼4,000 nuclei (31, 32, 43, 48). The mitotic spindle is also essential for the formation of karyomeres. First, it permits chromosomes to behave separately; if the formation of the mitotic spindle is prevented, the nucleus reforms around the entire mass of chromosomes. Second, the size of the egg (1.2 mm) imposes long mitotic spindles during the first cleavages of the embryo, and only by the blastula stage does the size of the cells approach that of a somatic cell. Assuming that the average rate of chromosome movement is constant (39), the regroupment of the chromosomes at the poles will be more rapid at the end of the cleavage stage. Moreover, membrane components also become limiting at the blastula stage. Therefore, the disappearance of karyomeres at this stage may be a consequence of both the depletion of maternal components for nuclei construction and the uncoupling between growth and division during early embryogenesis. Although karyomere formation is a physiological strategy developed by the embryo, the structure of karyomeres is reminiscent of micronuclei. Micronuclei formation in normal somatic cells can occur as a consequence of a variety of physical or chemical damaging agents (8–10, 30, 45). Most physical or chemical reagents that damage the mitotic spindle or inhibit its formation produce micronuclei due to failure to properly assemble chromosomes at the mitotic poles. Disruption of the function of NuMA, a protein involved in the architecture of the mitotic spindle, also results in the formation of micronuclei (24, 25). The size of the cells in the early embryo that imposes a long transit time to the spindle poles may parallel the situation in somatic cells when the mitotic spindle is damaged, leading to a disturbed migration of chromosomes and micronuclei formation.

Thus, we think that the organization in karyomeres during early development occurs as a consequence of the timing for segregation and reassembly of chromosomes at the spindle poles relative to the acceleration of other cell cycle events permitted by previously stored maternal components.

Our observations both in vivo and in vitro reveal an organization for DNA replication during early development which is relatively simpler than in somatic cells (Fig. 8). DNA replication is initiated and proceeds within karyomeres before nuclear reconstruction. Prereplicating complexes, as detected by RPA immunolocalization, form on condensed chromosomes in anaphase and DNA synthesis initiates in karyomeres at the beginning of telophase. This behavior is in accordance with the licensing-loading factor hypothesis which postulates that licensing DNA for replication occurs during mitosis before nuclear envelope formation (4). Formation of the nuclear lamina that is essential to the initiation of DNA replication (16, 23, 36) occurs within karyomeres during anaphase–telophase.

As opposed to the context in somatic cells, where some factors specific for S phase are produced only in G1, all components necessary for DNA replication are already present in the unfertilized egg. Therefore, the isolation of chromosomes into independent karyomeric vesicles allows both the formation of prereplication centers on condensed chromosomes during anaphase and the initiation of DNA synthesis before nuclear formation. A quite unusual situation results, in which a 2n phase nucleus is virtually absent during the entire early embryonic development (Fig. 8). This phenomenon might also account for a peculiarity of the genus Xenopus, which is that all but one of its 20 species are of ancient polyploid origin. Moreover, they form a polyploid series in the proportion 2:4:8:12 (26). The partial overlap between S and M phase together with the early formation of a nuclear envelope during the first embryonic divisions might have contributed to a relative facility to polyploidization.

This specific organization, where the unit of replication is not the nucleus but the chromosome, might also favor a higher frequency of initiation of DNA replication. Prereplication complexes studied by confocal microscopy of RPA foci in early embryonic nuclei were smaller and two- to threefold more numerous when nuclei were formed from karyomeres (unpublished observations). Moreover, the increase in the number of prereplication centers also correlates with a decreased size of the DNA loop domains. Using the maximum fluorescent halo technique (47), we measured a mean DNA loop size in karyomeres of 14.6 kb (data not shown) similar to the mean replicon size during early Xenopus development (22). This structure may provide a structural framework for an accelerated rate of replication during the early developmental period.

The abrupt disappearance of karyomeres after the twelfth cell cycle is a new transition occurring at the mid-blastula stage, which might affect not only replication but also transcriptional programs during development. During early development, a cell cycle clock, driven by the cycling MPF activity, imposes cell division every 30 min whether or not DNA is replicated (17, 41). The organization of the genome in karyomeres appears to be a developmental strategy to permit both a reduced replication period and an early initiation, so enabling replication to keep pace with the cytoplasmic clock. The formation of a nuclear envelope around isolated chromosomes also provides an in vivo validation of its role in regulating initiation of DNA replication (4). Indeed, as soon as a nuclear envelope isolates chromosomes, they can replicate once and only once, even if initiation of the whole set of chromosomes is permitted before the end of the cell cycle.

We are grateful to U.K. Laemmli (University of Geneva, Science II, Geneva, Switzerland) for the RPA antibody, R.A. Laskey (Welcome CRC Institute, Cambridge, UK) for the anti-MCM3 antibody, and R. Stick (University of Göttingen, Göttingen, Germany) for the lamin antibody. We also thank P. Brooks (Genoscope, Evry, France), J.C. Courvalin, and T. Grange (both from Institut Jacques Monod, Paris, France) for critical reading of this manuscript, and the photographic department of the Institut Jacques Monod.

FB

fixation buffer

RPA

replication protein A