Various studies in the last couple of years have shown that the initiation of DNA synthesis in eukaryotes is a complex, multi-step process that requires the participation of a large number of proteins. Many of these components have been identified and characterized, and it has been found that the identity and the order of assembly of eukaryotic replication factors is highly conserved across all eukaryotic species from yeast to man.
However, due to the increasing size and complexity of the genomes in metazoa, the demands on DNA replication exceed those of simpler eukaryotes. Therefore it is not surprising that new mechanisms to control DNA replication not seen in yeast have been evolved. This includes additional factors, as for instance the negative regulator Geminin, as well as variations regarding the regulation of some of the key factors involved in replication initiation such as ORC, MCMs and Cdc6p. Thus, the search for particular features regarding the properties of specific proteins may reveal additional mechanisms involved in replication control. Elucidation of the precise localization and dynamics of the human ORC subunits should therefore help to uncover roles of the human ORC proteins in chromosomal DNA replication.
ORC IS LOCALIZED TO TWO DIFFERENT CHROMATIN COMPARTMENTS In a first approach, the distribution of the different subunits of ORC was investigated by chromatin fractionation (Rose and Garrard, 1984) revealing that two fractions of ORC proteins exist (FIG 4-2). One fraction is found to be located in a nuclease-resistant compartment of the nucleus containing the nuclear matrix and nuclease-resistant chromatin including actively transcribed gene sequences, whereas the second fraction resides on open chromatin characterized by its high accessibility to micrococcal nuclease. Interestingly, these two fractions of ORC proteins differ in the presence of hOrc1p, which is only found in the nuclease-resistant fraction. The described situation was demonstrated for the subunits hOrc1p and hOrc2p and the behavior of hOrc2p seems to be representative for the subunits hOrc3p, hOrc4p and hOrc5p (data not shown).
The presence of two different forms of ORC proteins was confirmed by cell fractionation experiments (FIG 4-3, FIG 5-1). All ORC subunits with the exception of hOrc6p are found exclusively in the nucleus. While the bulk of hOrc2-5p can be eluted with 250 mM NaCl, hOrc1p together with the remaining fractions of hOrc2-5p require 320-450 mM NaCl to be released from chromatin.
In synchronously growing HeLa cells the hOrc1p-containing fraction of ORC disappears during S-phase independent of the fractionation procedure (FIG 4-6). This indicates that the ORC fraction found in the nuclease-resistant chromatin fraction is identical to the ORC fraction released with 320-450 mM NaCl from chromatin.
Recently, Fujita and coworkers described a similar distribution of ORC applying an in vivo chemical cross-linking method, where the bulk of hOrc2p binds to nuclease-sensitive chromatin while a smaller fraction of hOrc2p is associated with non-chromatin nuclear structures in a physical interaction with hOrc1p (Fujita et al., 2002).
It has been suggested that DNA replication foci are attached to non-chromatin nuclear structures such as the nuclear matrix, since nascent DNA and many proteins involved in DNA synthesis have been found associated with these sites (Berezney et al., 1995;
Cook, 1999). Thus, our finding that hOrc1p is only present in a subset of ORC complexes in a chromatin fraction harboring the nuclear matrix suggest that only the hOrc1p-containing form of ORC is bound to replication sites. This hypothesis is in agreement with two recent reports that found both hOrc1p and hOrc2p to be cross-linked in vivo to chromatin fragments harboring origins of replication (Keller et al., 2002; Ladenburger et al., 2002) while hOrc2p was also cross-linked to heterochromatic satellite DNA in a region that did not include a replication start (Keller et al., 2002). To further support this data, it certainly would be of interest to assess whether hOrc1p and replication start sites do colocalize in vivo in general. Therefore we attempted to establish cell lines stably expressing GFP-hOrc1p in order to track the behavior of the overexpressed hOrc1p in combination with BrdU-incorporation (data not shown).
However, although GFP-hOrc1p was transiently expressed and localized to the nucleus,
cells did not survive extended expression of GFP-hOrc1p, in contrast to overexpression of GFP alone. Assuming that hOrc1p is a limiting factor in the initiation of DNA replication (see below), it appears plausible that overexpression of hOrc1p may lead to a malfunction of DNA replication, which could finally trigger apoptotic cell death. To circumvent such a possible mechanism, future experiments with a regulated expression of GFP-hOrc1p (e.g. using the Tet ON/OFF system) should clarify whether hOrc1p and replication start sites colocalize.
THE SUBUNIT COMPOSITION OF THE HUMAN ORC
In the yeast Saccharomyces cerevisae, all six subunits of ORC are essential for cell viability. They collectively bind to origin sequences in order to recruit additional replication factors. Although similar six-protein complexes have been discovered in Xenopus laevis, Drosophila melanogaster and Schizosaccharomyces pombe, the purification of a six-protein human ORC has not yet been successful. Therefore we tempted to demonstrate a physical interaction between the single subunits of both ORC fractions (see above) by a combination of co-immunoprecipitations and sucrose gradients (FIG 5-2). Complexes of hOrc2-5p in the 250 mM salt fraction or hOrc1-5p in the high salt fraction could readily be isolated by immunoprecipitation, but hOrc6p could not be detected as part of either of these complexes in vivo. In addition, hOrc6p could be precipitated quantitatively from extracts prepared with 100 mM NaCl (data not shown) without an apparent interaction with other subunits. Thus, the putative interaction of hOrc6p with the other subunits seems to be rather weak. This is in agreement with published data showing that in contrast to hOrc1-5p only small amounts of hOrc6p are present on isolated chromatin when ORC subunits were cross-linked to DNA (Keller et al., 2002).
Centrifugation of the high and low salt extracts through sucrose gradients revealed the presence of subcomplexes in both extracts, characterized by prominent hOrc4p and hOrc5p bands sedimenting with a lower S-value compared to the hOrc1-5p complex (FIG 5-2A). This was confirmed by immunoprecipitations, where Orc2p-specific anti-bodies depleted hOrc2p and hOrc3p efficiently from the extracts, while significant amounts of hOrc4p and hOrc5p remained in the supernatant (FIG 5-2B). Since Orc4p-specific antibodies depleted almost all of the hOrc4p present, while large amounts of hOrc2p, hOrc3p and hOrc5p were not co-immunoprecipitated, it is likely that hOrc4p is also rather loosely bound to the other subunits. Thus, the minor subcomplexes might result from disintegration of the ORC complex. Interestingly, the low salt extract lacking hOrc1p contained higher amounts of subcomplexes, suggesting that hOrc1p might stabilize the interactions between the other subunits as well as their association to chromatin. This is in line with the observation that hOrc1p-containing complexes require higher salt for dissociation from chromatin compared to the hOrc2-5p complexes (FIG 5-1). However, it remains to be elucidated whether the hOrc2-5p complex is also released from chromatin together with hOrc1p, or whether it is only less
tightly bound to chromatin following release of hOrc1p. The presence of hOrc2-5p on chromatin even in absence of hOrc1p implies, however, that a more limited complex is competent for DNA binding, although the specificity of this association remains unclear.
Recently published investigations addressing the assembly of ORC in human cells also identified a complex of hOrc1-5p lacking hOrc6p (Vashee et al., 2001) while another group only found a complex comprised of hOrc2-5p (Dhar et al., 2001a). These differences might be explained by the fact that more stringent conditions (e.g. RIPA buffer) during the immunoprecipitation procedure were used in the latter study, indicating that the interaction between hOrc1p and the other subunits is not resistant to high amounts of detergents (Dhar et al., 2001a; Thome et al., 2000).
By expressing hORC subunits in insect cells, however, both groups were able to identify a complex containing all six subunits, although hOrc1p and hOrc6p did not enter the complex in a stoichiometric ratio (Dhar et al., 2001a; Vashee et al., 2001).
Further analysis of the interactions between individual subunits indicated that hOrc5p is necessary to stabilize the association of hOrc4p with hOrc2p and hOrc3p (Dhar et al., 2001a), being in line with our observation that hOrc4p disintegrates easily from the complex. Collectively, these findings suggest that hOrc2-5p form a core complex with hOrc1p and hOrc6p joining the complex either at very restricted times or locations, or in a labile interaction that is easily disrupted upon cell lysis.
Recent findings in Drosophila and human cells contribute to the understanding of the function of Orc6p (Chesnokov et al., 2001; Prasanth et al., 2002). In both species, the localization pattern of Orc6p changes dramatically during the cell cycle. While Orc6p is localized in the nucleus during interphase, it localizes to kinetochores and to a reticulate-like structure around the cell periphery during mitosis. As chromosomes segregate during anaphase, the reticulate structures align along the plane of cell division and some Orc6p localizes to the midbody before cells separate. Silencing of Orc6p by small interfering RNA (siRNA) resulted in spindle defects, polyploidy and cytokinesis defects, as well as decreased BrdU-incorporation, thus providing evidence that Orc6p functions in multiple aspects of the cell division cycle, including DNA replication, chromosome segregation, and cytokinesis (Prasanth et al., 2002). Because Orc6p silencing leads to phenotypes similar to those of cell cycle checkpoint proteins, it has been suggested that Orc6p might play an important role as an integral component of the cell cycle checkpoint machinery involved in the temporal coordination of DNA replication and cell division. Furthermore, the observed changes of localization support the hypothesis that Orc6p joins the remaining ORC subunits only at very restricted periods during the cell cycle, thus providing an explanation why Orc6p could not be detected in complex with other ORC subunits in asynchronously growing cells.
Assuming that all six subunits of ORC are required for DNA replication, it would be also possible that only a small subset of intact hORCs is required for DNA replication.
The finding that a decrease of the hOrc2p-level to 90% does not appear to have any effect on the utilization of cellular origins of replication supports this hypothesis (Dhar et al., 2001b). Although ORC acts as a DNA replication initiator protein, it is known to function in other cellular processes such as transcriptional silencing and heterochromatin formation (reviewed in Bell and Dutta (2002)). However, considering that the hOrc1p-lacking subcomplex of ORC is located to the transcriptional active chromatin regions, a function of this subcomplex in heterochromatin formation becomes unlikely.
CHANGES IN THE COMPOSITION OF HUMAN ORC DURING THE CELL CYCLE
Our detailed analysis of the association of hOrc1p with nuclear structures in the course of the cell cycle revealed that hOrc1p is released from chromatin in a regulated manner during S-phase (FIG 4-5). The association of hOrc1p with chromatin during G1-phase, followed by its degradation during S-phase correlates remarkably well with the in vivo footprint analysis at the human lamin B2 origin. An extended footprint covering over 100 bp is observed in G1-phase, shrinks to 70 bp when cells enter S-phase and totally disappears in mitosis (Abdurashidova et al., 1998). Whether hOrc1p or other factors such as MCM proteins are responsible for these dynamic protein-DNA interactions, however, remains to be answered.
Changes in the affinity of one or more ORC subunits for chromatin during mitosis have been reported for other multicellular eukaryotes such as Xenopus, Drosophila and hamster. In Xenopus, both XlOrc1p and XlOrc2p are present on chromatin in cultured cells during interphase but not during metaphase (Romanowski et al., 1996). In Drosophila, DmOrc2p remains associated with chromosomes throughout the cell cycle, while DmOrc1p levels on chromatin fluctuate being greatest during late G1- and S-phase (Asano and Wharton, 1999). In hamster, CgOrc1p binds weakly to chromatin during mitosis and early G1-phase under conditions where CgOrc2p remained bound.
Therefore, it appears that higher eukaryotes regulate the initiation of DNA replication through cell cycle-dependent changes in ORC activity, an essential step in the assembly of pre-RCs. In human cells this is accomplished by selective release of hOrc1p in S-phase probably after pre-RCs have been activated. This would provide an additional mechanism to delay the assembly of new pre-RCs until both DNA replication and mitosis are completed.
However, in mammals there remains some controversy as to the behavior of Orc1p during the cell cycle. A recent report suggests that hamster Orc1p levels associate with chromatin during the whole cell cycle, including S-phase (Okuno et al., 2001). Besides the possibility that human and hamster Orc1p may not be regulated in the same way, the discrepancy could be explained by the different synchronization methods used. Okuno et al. (2001) synchronized a cell population using consecutive drug blocking steps,
therefore keeping the cells under constant influence of drugs that possibly affect the metabolism of the cells. In addition, every fraction collected contained significant amounts of cells in early and mid S-phase. Thus even if CgOrc1p was released from chromatin during S-phase, these cell populations could contain detectable levels of CgOrc1p in the chromatin fraction.
In order to rule out that the differences in the behavior of mammalian Orc1p arise from different extraction procedures, synchronized HeLa cells were prepared according to the protocol of Okuno et al. (2001) (data not shown). However, hOrc1p was again released from chromatin during S-phase, indicating that discrepancies indeed may occur due to different synchronization procedures. The release of Orc1p from chromatin during S-phase was subsequently confirmed by two other groups using both hamster CHO cells and human HeLa cells (Li and DePamphilis, 2002; Mendez et al., 2002). In addition, a recent study reports that hOrc1p can only be in vivo cross-linked to HeLa cells synchronized in G1-phase but not to cells synchronized in S-phase (Ladenburger et al., 2002). Taken together, this suggests that the regulated release of Orc1p from chromatin is conserved among mammals (Li and DePamphilis, 2002; Mendez et al., 2002).
THE RELEASE OF HUMAN ORC1 DEPENDS ON ONGOING REPLICATION Interestingly, the S-phase-dependent release of hOrc1p from chromatin depends on progression of replication forks, because it can be inhibited by addition of aphidicolin (FIG 5-4), a specific inhibitor of replicative DNA-polymerases (α, δ and ε) (Krek and DeCaprio, 1995), or by indirectly stalling DNA replication with proteasome inhibitors (FIG 5-7C). Therefore, the gradual release of hOrc1p during S-phase might correspond to early and late origins of replication.
Sun and coworkers, on the other hand, had reported that the programmed release of ORC from chromatin occurs prior to initiation of DNA replication and does not require the progression of replication forks but rather depends on the loading of MCM proteins at ORC/chromatin sites (Sun et al., 2002). This model is based on experiments using hamster metaphase chromatin incubated in Xenopus egg extract under conditions that initiate a single round of semi-conservative DNA replication. In this system DNA synthesis succeeds very rapidly, and the low resolution of the time course might be misleading, because upon release of ORC the proportion of S-phase nuclei had already reached 60-70%. Thus, S-phase was already well under way at that time. The authors could also demonstrate that depletion of MCMs from the Xenopus extract prevents the release of ORC as well as DNA synthesis, thus these results could also be interpreted as to release of ORC indeed requires DNA synthesis. A connection to the MCM proteins could therefore be also drawn in regard to their helicase activity. If MCMs would indeed represent the replicative helicase, the required loading of MCM proteins would be in line with the finding that ORC displacement from chromatin is coupled to the progression of replication forks.
It also has to be taken into consideration that the system used in these experiments derives from very early developmental stages with unusual cell cycles. Xenopus eggs are enriched in initiator protein factors for the very rapid S-phases that occur during early embryonic stages of life, thus one cannot rule out the possibility that there are fundamental differences in the regulation of Orc1p in this system compared to somatic cells. This may also contribute to the finding that not only Orc1p but also other subunits of ORC were released from the hamster nuclei used in this system, in contrast to somatic cells where only Orc1p seems to be released while the other subunits are only less stably bound (FIG 4-5, FIG 5-1).
However, the precise mechanism that triggers hOrc1p release remains unclear. One possibility is suggested by the observation in S. cerevisiae that ORC displays two different conformations dependent on whether it is bound to double-stranded or single-stranded DNA (Lee et al., 2000). Moreover, it has been shown that single-single-stranded DNA stimulates the ORC ATPase. Thus, it is possible that the formation of single-stranded DNA at the origin triggers ORC ATP hydrolysis and a change in ORC conformation, which could then lead to a destabilization of ORC and a release of hOrc1p. An alternative mechanism could involve subunit modification (e.g.
phosphorylation) followed by hOrc1p release. The two proposed mechanisms are not mutually exclusive. It remains possible that structural changes of ORC after initiation of replication may lead to exposure of modification sites and vice versa.
These models would be in line with some of our additional experiments not shown in this thesis. HeLa cell nuclei prepared from G1-phase cells as well as in vitro translated T7-hOrc1p was not degraded in cytoplasmic/nucleoplasmatic extract prepared from S-phase cells even when incubated at 37°C for 1h. Thus, the presence of soluble hOrc1p alone does not seem to be sufficient as a signal for degradation, suggesting that hOrc1p release from chromatin is accompanied either by a conformational change or by a modification directly connected to the progression of replication forks, such as phosphorylation.
HUMAN ORC1 IS PHOSPHORYLATED IN VITRO AND IN VIVO
Plausible candidates for a phosphorylation of ORC are cyclin-dependent kinases, which are involved in the regulation of the cell cycle. In this case kinase activity should somehow be restricted to replication forks in order to be able to distinguish between licensed and already replicated chromatin. In this context it is of interest that Furstenthal and coworkers demonstrated that a cyclin E-Cdc6p interaction localizes the cdk2 complex to pre-RCs in close vicinity of ORC (Furstenthal et al., 2001a). Moreover, kinase complexes containing cyclin A were found to be immunoprecipitated in a complex with the CDK inhibitor Cip1 and PCNA (proliferating cellular nuclear antigen), a cofactor of DNA-polymerases (Zhang et al., 1993).
Several pieces of evidence indicate that cyclin-dependent kinases are indeed involved in the regulation of hOrc1p. It has been shown that hOrc1p appears to be a good in vitro substrate of both recombinant cdk2/cyclin A and cdk2/cyclin E (Mendez et al., 2002).
In addition, hOrc1p is readily phosphorylated in extracts prepared from HeLa cells synchronized in late G1- and the middle of S-phase, but not from early G1-phase, consistent with the presence and absence of cyclin E and cyclin A, respectively (FIG 5-6). A similar correlation between hOrc1p phosphorylation and cyclin E and A expression was observed in vivo by metabolic labeling of HeLa cells with [32 P]-ortho-phosphate (FIG 5-5). This points to a regulation of hOrc1p by CDKs, which is in line with the presence of 15 potential acceptor sites with the minimal CDK consensus motif S/T.P (serine/ threonine followed by proline) in hOrc1p.
Interestingly, cyclins E and A have specialized roles during the transition from G0- to
Interestingly, cyclins E and A have specialized roles during the transition from G0- to