Dynamics of DNA replication: an ultrastructural study

Abstract

DNA replication in cells takes place in domains scattered throughout the nucleoplasm. We have characterized the dynamics of DNA synthesis in synchronized mid-S-phase HeLa cells. Saponin-permeabilized cells were allowed to elongate nascent DNA chains in presence of biotin-dUTP for 5, 15, and 30 min (a pulse experiment), or for 5 min followed by an incubation with unlabeled precursors for 10 or 25 min (a pulse-and-chase experiment). The replication foci were then identified in ultrathin sections using immunogold labeling of the incorporated biotin. Total number of particles per nucleus, total scanned area of the nucleus, size, shape, and gold particle number of each labeled cluster, and the density of clusters per nucleus were evaluated. We have demonstrated that as replication proceeds, the labeled sites increase in size up to 240 nm (30 min incorporation) while maintaining a broadly round shape. In pulse-and-chase experiments the labeled DNA was shown to spread to occupy DNA foci of ∼400 nm in diameter. These results demonstrate that DNA replication is compartmentalized within cell nuclei at the level of DNA foci and support the view that the synthetic centers are spatially constrained while the chromatin loops are dynamic during DNA synthesis.

Introduction

It is accepted that the mammalian nucleus, despite an absence of intranuclear membranes, is organized into functional domains or foci, where nuclear processes like DNA replication, transcription, RNA processing and ribosome biogenesis take place (Hendzel et al., 2001, Jackson, 2003, Scheer and Weisenberger, 1994, Spector, 2001, Stein et al., 2003, van Driel and Fransz, 2004). For example, domains of active DNA synthesis that are scattered throughout the nucleoplasm during S-phase contain replication proteins such as DNA polymerases and proliferating cell nuclear antigen—PCNA (Hozák et al., 1993, Sporbert et al., 2002) as well as proteins that regulate the cell cycle—such as cyclin A and cdk2 (Cardoso et al., 1993)—and chromatin structure—such as uracil-DNA glycosylase (Otterlei et al., 1999) and DNA methyltransferase (Leonhardt et al., 1992). The existence of localized replication domains appears to be a common architectural theme as these structures are found in species ranging from mammals (Hozák et al., 1993, Nakamura et al., 1986, Nakayasu and Berezney, 1989) to plants (Sparvoli et al., 1994). Perhaps surprisingly, the theme may even apply in prokaryotes, as the active Bacillus subtilis DNA polymerase complexes were found to be localized in fixed intracellular positions (Lemon and Grossman, 1998). Also, no fundamental differences were found in the spatio-temporal organization of replication patterns between primary, immortal or transformed mammalian cells (Dimitrova and Berezney, 2002).

Replication domains in animal cells were first described in the pioneering experiments of Nakamura et al. (1986). In this study, rodent fibroblast were shown to have about 100 sites of active DNA synthesis, which grew in size for about an hour before new active sites were activated at adjacent nuclear positions. More recent reports have used improved imaging techniques to estimate that both mouse 3T3 cells (Ma et al., 1998) and HeLa cells (Jackson and Pombo, 1998) contain ∼1000 replication sites in early S-phase. As estimated by light microscopy, typical DNA foci are ∼500 nm in size (Ma et al., 1998) and contain about 1 Mb DNA. During the early stage of S-phase, average replicons are ∼150 kb in size. This is consistent with the view that each replication site must contain 5–6 replicons in order to complete synthesis in the observed time frame of 45–60 min given that replication proceeds at a rate of about 1.5 kb/min (Jackson and Pombo, 1998, Ma et al., 1998). Importantly, these groups of replicons appear to remain stably associated within DNA foci over at least 10 cell generation (Jackson and Pombo, 1998, Ma et al., 1998, Zink et al., 1998) suggesting that they may be essential sub-units of chromosome structure. The kinetics of replication sites was also studied in cells of Hydra (Alexandrova et al., 2003). Following subsequent incorporations of IdU and CldU (chase 2 h), the resulting replication sites did not coincide; nevertheless, the replication pattern was similar.

This observation raises the possibility that DNA foci might be fundamental units of both chromosome structure and function. This notion is supported by the fact that distinct groups of DNA foci are replicated reliably at specific times of S-phase during subsequent cell cycles (Jackson and Pombo, 1998, Ma et al., 1998). Furthermore, the organization of DNA foci might also provide a mechanism that allows DNA synthesis to spread efficiently throughout the genome by virtue of a ‘domino’ effect whereby the completion of replication at a particular site will activate replication at spatially adjacent sites (Ma et al., 1998, Manders et al., 1996, Sporbert et al., 2002). In this way, the organization of DNA foci will define the replication program. In addition, many lines of evidence suggest that the active replication complexes of mammalian cells are tethered to an underlying framework called the nuclear matrix or nucleoskeleton (Hozák et al., 1993, Tubo and Berezney, 1987). If this is true, during DNA synthesis the DNA polymerase complexes must remained fixed and chromatin must move so that DNA is reeled through the active synthetic centers (Hozák and Cook, 1994, Jackson, 1990).

In fact, surprisingly little is known about the dynamic properties of chromatin during DNA synthesis even though this must be an essential feature of the replication process. Specialized approaches that allow large chromosomal domains to be tagged and visualized in living cells suggest that large-scale chromatin movements might occur during S-phase (Li et al., 1998). However, within natural chromatin domains, chromatin dynamics appear to be locally constrained (Chubb et al., 2002); though this analysis does not rule out the possibility that chromatin might be highly dynamic at the level of DNA foci. In principle, it should be possible to establish the extent of chromatin movement in response to DNA synthesis from the distribution of nascent DNA with replication foci. Ma et al. (1998) used a spot-based segmentation analysis to investigate the architecture of replication foci using confocal microscopy. They demonstrated that the dimensions of individual replication sites changed little over labeling periods of 2–30 min, suggesting that the replication sites correspond to discrete domains of DNA whose replication occurs uniformly over the individual replication site.

In a recent study (Jaunin et al., 2000), classical immuno-electron microscopy techniques were used to show that nascent sites—detected using a 2 min pulse-label with BrdU—were localized to the diffuse chromatin of the perichromatin regions, often in the vicinity of perichromatin fibrils. During prolonged pulses and pulse-chase experiments the regions of dense chromatin adjacent to these perichromatin regions became labeled. When a pulse-chase-pulse approach was used to label the DNA with IdU and then CldU, the nascent label was found in the perichromatin zone, supporting the view that replication occurred within a defined nuclear zone and that the replicated DNA then moved into the adjacent chromatin-rich area. Earlier studies using a specialized technique that allowed cell structure to be preserved while removing almost all chromatin showed that early S-phase replication occurs within discrete replication ‘factories’ of 50–100 nm, and in larger ‘factories’ in mid- and late-S-phase (Hozák et al., 1993, Hozák et al., 1994). Although these reports cannot be directly compared as they use so different techniques, they both support the hypothesis that replication occurs within a defined inter-chromatin compartment so that during replication DNA from the adjacent chromatin-rich compartment must be translocated to the active site at the border of the chromatin domain.

In the current report, we develop the argument that DNA movement during replication must be an essential part of the replication process. To explore this issue, we have undertaken a detailed analysis of the spread of DNA from the nascent replication sites. Nascent DNA was labeled with biotin-dUTP in permeabilized cells using pulse and pulse-chase strategies. This approach was taken for two specific reasons. First, when biotin-dUTP is used as the replication precursor the incorporated biotin can be detected in fixed cells using antibodies or streptavidin. BrdU, in contrast, can only be detected when the DNA is denatured, compromising both morphology and labeling efficiency. Second, in permeabilized cells the concentrations of the precursor pools can be regulated so that the levels of modified precursor incorporated and overall elongation rates can be manipulated at will. Using immunogold procedure on ultrathin sections we show that the clusters of labeled nascent DNA grow gradually during replication while their shape remains essentially circular in cross-section. Our data support the view that replication occurs within discrete active sites and is not in agreement with the view that active replication complexes scan along the DNA during synthesis. Instead, replication seems to involve a cycle of DNA movement, where chromatin first moves from the chromatin-rich regions into the active site where replication occurs. Following synthesis and chromatin maturation, it is necessary that the daughter DNA molecules re-fold and move out of the active site to regenerate the chromatin-rich nuclear compartment.