The Role of the Replication Licensing System in Cell Proliferation and Cancer

Abstract

The precise duplication of chromosomal DNA during each cell cycle is essential for the maintenance of genetic stability. Failure to correctly regulate chromosomal DNA replication could lead to losses or duplication of chromosome segments. The precise duplication of chromosomes is normally achieved by correct regulation of the replication licensing system. Here we review our current knowledge of the licensing system and how this might be defective in cancer cells. We also review how detection of licensing components can be used for the diagnosis and prognosis of cancer. Finally we discuss the potential of the replication licensing system as a novel anti-cancer target.

The Replication Licensing System

The precise duplication of chromosomal DNA is ensured by dividing the process of chromosome replication into two discrete steps (Fig 1). In the first stage, which occurs during late mitosis and early G1, replication origins are licensed for replication by loading complexes of the Mcm2-7 proteins onto them to form a “pre-replicative complex” (pre-RC). At the onset of S phase, cyclin-dependent kinases (CDKs) are activated and they initiate replication forks only at these licensed origins. As initiation occurs, Mcm2-7 are displaced from the origins, probably moving along with the replication forks, so that the origin cannot fire again in the current cell cycle. Replicated origins are prevented from becoming re-replicated by inhibiting further licensing during late G1 through to early M phase. This is achieved by the combined activity of CDKs and an inhibitory protein called geminin.

Figure 1.

Replication licensing and cyclin-dependent kinase (CDK) activity through the cell cycle. A small segment of chromosomal DNA is depicted at different stages of the cell cycle, with bound or unbound Mcm2-7 (hexagons). The activities of the replication licensing system (RLS, dark) and CDKs (light) at the different stages are shown in the central circle.

For origins to load Mcm2-7 and become licensed at least three other proteins are required (Fig 2, top) (1-4). The Origin Recognition Complex (ORC), a complex of six proteins Orc1-6, first binds to each replication origin and then recruits two other proteins, Cdc6 and Cdt1. These three proteins then act in concert to load multiple Mcm2-7 complexes and to form a functionally licensed origin. This reaction has recently been reconstituted with purified proteins (2). ORC, Cdc6 and Cdt1 are required only for the loading, not the maintenance, of functional Mcm2-7 at origins. One important consequence of this is that re-licensing of replicated DNA can be prevented by inhibition or removal of ORC, Cdc6 or Cdt1 once S phase has started, without displacing functional Mcm2-7 at licensed origins.

Figure 2.

Pathways for the inhibition of origin licensing in higher eukaryotes. At the top, a replication origin is shown with bound ORC, Cdc6, Cdt1 and several Mcm2-7 complexes. Below are shown the different mechanisms by which CDKs and geminin can inhibit licensing activity. Cdt1 is bound and inhibited by geminin. Orc1 and Cdc6 may be degraded, whilst Cdc6 may also be exported from the nucleus into the cytoplasm.

Activation of CDKs in late G1 then appears to be the major trigger for blocking further origin licensing. There is evidence from a range of different organisms that all four components of the licensing system can potentially be inhibited by CDKs. The situation is clearest in S. cerevisiae, where CDKs cause the nuclear export of Cdt1 and unbound Mcm2-7, whilst ORC is inhibited and Cdc6 levels are down-regulated (1,3-5). In metazoans, inhibition also involves a number of different pathways, although these differ from those in yeast. These pathways are shown schematically in the bottom part of Figure 2.

ORC binds with a high affinity to chromatin during early G1 but the binding then weakens following origin licensing (6-8). In mammals, the Orc1 subunit is released from DNA during S phase and mitosis. A recent report has also shown that the released Orc1 undergoes ubiquitin- mediated proteolysis (9). Although a significant amount of chromatin-bound Cdc6 is observed throughout the cell cycle, some Cdc6 degradation during S phase has also been reported (10,11). The non-chromatin bound form is translocated from the nucleus to the cytoplasm during G1 when CDKs are activated (12-14). Like Cdc6, Cdt1 is also subject to proteolysis at the beginning of S phase (15). It is also inhibited by binding to geminin, a small protein whose levels and activity are under cell cycle control and peak in S, G2 and M phases (16-19).

The licensing system is also down-regulated as cells reversibly or irreversibly withdraw from the cell cycle (Fig 1) (4). Evidence from a range of different organisms and cell types suggests that both G0 and permanently arrested cells no longer have Mcm2–7 bound to chromatin and are thus functionally unlicensed. It has been proposed that the lack of chromatin-bound Mcm2-7 provides a meaningful definition of withdrawal from G1 and the proliferative state (4). Not only is the Mcm2–7 hexamer removed from DNA as cells pass into G0, but the unbound protein is also lost from the cells. A similar reduction in Cdc6 protein is seen in G0 and permanently arrested cells, though ORC and Cdt1 proteins may persist (4,20).

Re-replication of chromosomal DNA in cancer cells

Aberrations to mechanisms that ensure precise replication of the genomic DNA during each cell cycle can cause chromosomal instability and thus lead to the development of cancer (21). Aneuploidy (aberrations in chromosome number) is commonly observed in human cancer (22,23). There is compelling evidence to show that aneuploidy contributes to malignant transformation and progression of cancer. Firstly, specific chromosome aneusomies correlate with distinct tumour phenotypes. Secondly, aneuploid tumour cell lines have been reported to show an elevated rate of chromosome instability, and a number of mitotic genes regulating chromosome segregation have been found mutated in cancer cells implicating such mutations in induction of aneuploidy in tumours.

Chromosomal instability involving gains and losses of whole chromosomes or parts is likely to occur in most human malignancies, conferring a selective growth advantage to the cancer cell over its normal counterpart. Whole chromosomes are likely to be gained or lost by failure of correct sister chromatid segregation during anaphase. However, loss or duplication of chromosome segments might be driven by failures in regulating chromosomal DNA replication. Under-replication of even small segments of chromosomes will lead to chromosome breakage at these segments as the sister chromatids are pulled apart at mitosis (Fig 3, top panel). Over-replication of chromosomal segments represents an irreversible genetic change, which might be resolved to form tandem duplications (Fig 3, bottom panel). These sorts of chromosomal defects are commonly seen in cancer cells, though whether they are generated by replication defects such as these is currently unclear. This section will discuss how re-replication or endoreplication may play a role in this process.

Figure 3.

The consequence of origin misfiring. A small segment of chromosomal DNA replicated from three origins is shown during different stages of the cell cycle. In the top panel DNA is under-replicated as a result of the failure of one of the origins to fire. As sister chromatids are separated during anaphase, the chromosome is likely to be broken near the unreplicated section. The middle panel shows the successful duplication of chromosomal DNA. In the bottom panel chromosomal DNA is over-replicated as a result of one of the origins firing a second time in S phase. The local duplication of DNA in the vicinity of the over-firing origin is likely to represent an irreversible genetic change and might be resolved to form a tandem duplication.

There are two general mechanisms by which cells can increase their chromosomal DNA content, involving “endocycles” or “origin re-firing” (4), as outlined in Figure 4. During endoreplication cell cycles (endocycles), complete rounds of chromosomal DNA replication occur without intervening cell divisions (24). In endocycles there are alternating periods of high and low CDK activity so that essentially normal S phases occur. Re-licensing of origins occurs when the CDK activity present in G2 or mitotic cells is abolished before cytokinesis has occurred, thus leading to a further round of chromosome replication in a single cell to double the DNA content. In contrast, origin re-firing results in re-licensing and re-firing of certain origins in the continued presence of high CDK levels. This sort of localised re-replication is depicted in the lower panel of Figure 3.

Figure 4.

Different routes by which cellular DNA content can increase. A small segment of chromosomal DNA is depicted at different stages of the cell cycle. The hexagons represent Mcm2-7 bound to origins in G1 and S phases. The black arrows show the normal sequence of cell-cycle events. The grey arrows show routes by which cellular DNA content is increased (endocycles and origin re-firing). Cyclin dependent kinase (CDK) activity in S, G2 and M phase is denoted by the outer black circle. Redrawn from (4).

Endoreplication

Genetic and biochemical evidence from yeast show that an oscillation in CDK activity is sufficient to induce endoreplication. In the yeast Schizosaccharomyces pombe, CDK activity determines the dependency of S phase on M phase (25). Deletion of p56^cdc13, the major B-type cyclin in S. pombe, causes cells to undergo repeated rounds of S phase without progression into M phase. As a result the ploidy can be increased to >32C (26). Likewise, over-expression of Rum1, a CDK inhibitor, leads to increased DNA content and nuclear enlargement (27,28).

Studies in mammalian cells are also consistent with the above model, suggesting that endoreplication depends on transient loss of CDK activity to permit replication origins to become re-licensed for an additional cycle of replication while preventing the completion of mitotic events (29). The most detailed description to date of how endoreplication occurs in mammals comes from studies of trophoblast giant cells in rodent placenta. Giant cells undergo endoreplication during terminal differentiation, reaching ploidies of more than 1000C (30). During the transition from a mitotic cell cycle to endoreplication, trophoblast cells reduce expression of cyclin B (29). Coincidentally, there is an increase in cyclin A and cyclin E- associated CDKs, the levels of which oscillate prior to and during S phase. Modulation of G1 and S phase CDK activity in endoreduplicating trophoblast cells appears to involve the cyclic synthesis and destruction of the CDK inhibitor Kip2 (31). This would create two distinct phases: a G1-like period when Kip2 levels are high and there is low CDK activity when licensing can occur, and an S phase-like period when Kip2 levels are low and CDK activity is high (31). However, there is as yet no clear evidence to suggest that this type of endocycle occurs in cancer cells.

Another less well understood route for endoreplication with more potential relevance for cancer occurs in cells with defects in the retinoblastoma protein pRb. When the CDK inhibitor p21^Cip1/Waf1 was expressed in cells possessing functional pRb, the cell cycle was arrested in G1 (32). However, when p21Cip1/Waf1 was expressed in cells lacking pRb activity, cells arrested in G2 and underwent cycles of endoreplication. The observation that p21^Cip1/Waf1 can promote endoreduplication in pRb-negative contexts has potential practical implications for cancer therapy. For tumours that are pRb negative but p53 positive and therefore capable of inducing p21^Cip1/Waf1 in response to DNA damage, endoreduplication may be a consequence of radiation therapy, chemotherapy or any other genotoxic stress. This may lead, in some cells, to enhanced malignancy. An example of such a situation is familial retinoblastoma, where one germline copy of the Rb gene is mutated. Rb-negative cells occur at a relatively high frequency due to loss of heterozygosity at the RB locus and result in neoplasia, most notably in the retina (33). However, such patients are at a greatly increased risk of developing secondary tumours in surrounding mesenchymal tissue exposed to ionizing radiation therapy (34-36). It is possible that induction of p21^Cip1/Waf1 in pRb-negative cells within such tissues leads to endoreduplication, genetic instability, and ultimately, secondary malignancy.

Origin re-firing

Under certain circumstances, pre-RC proteins (ORC, Cdc6, Cdt1 or Mcm2-7) may not be able to respond to the inhibitory signals that normally act in S phase as depicted in Figure 2. Instead, origins are re-licensed and re-initiated in the continuing presence of high CDK level. One of the best studied examples of this occurs within ovarian follicle cells during Drosophila oogenesis, where certain genes are amplified to ensure sufficient production of chorion proteins that make up the eggshell. This occurs by the repeated initiation of replication origins close to the amplified genes in the presence of high CDK levels (24,37). In S. pombe, over-expression of wild type or a phosphorylation defective cdc6 mutant (SpCDC18) results in extensive over-replication (38,39). Similar results from Arabidopsis show that endoreduplicating cells up-regulate Cdc6, and ectopic expression of Cdc6 can induce endoreplication (40).

Defective inhibition of Cdt1 by geminin could also lead to origin re-firing. Experiments in Xenopus egg extracts have shown that geminin is the major inhibitor of origin licensing in S, G2 and M phases (18,19). A recent study carried out in Drosophila cells shows that silencing of geminin leads to cessation of mitosis and asynchronous over-replication of the genome, producing cells containing single giant nuclei and partial ploidy between 4N and 8N DNA content (41). Although it therefore represents a potential tumour suppressor gene, no defects in geminin expression have been reported to date in cancer cells.

Does origin over-firing contribute to the chromosome instability of cancer cells? Amplification of subchromosomal portions of the genome is frequently seen in many human cancers (42). Amplification of specific genes may be advantageous to cancer cells by increasing the copy number of oncogenes or genes that confer resistance to anti-cancer drugs (43). Amplified regions of the genome occur either as homogeneously staining regions or small acentric autonomously replicating double minutes (44-47). Several mechanisms for initiating DNA sequence amplification in mammalian cells have been proposed, which fall either into the over-replication or non-disjunction groups. Amplification of genes in the non-disjunction group occurs as a consequence of defects in chromosome segregation during mitosis (reviewed in(48,49)), and is thus outside the scope of this review.

For over-replication to generate tandem duplications, it is proposed that multiple strands of over-replicated DNA are ligated together and subsequently recombined to generate expanded chromosomal regions containing amplified genes. Alternatively, the over-replicated strands might circularise to form extrachromosomal elements. Consistent with these ideas, several studies have shown that amplification frequencies can be increased by treatments that transiently inhibit DNA synthesis, resulting in chromosomal breakage (50). This suggests that the substrate for breakage might be a replication bubble encompassing the locus that will be amplified. Acentric amplicons can be generated by chromosome breakage within a segment of over-replicated DNA (51) or within a stalled replication bubble (50) encompassing a particular gene and one or more replication origins. If over-replication is really an important route to gene amplification in cancers, then mutations might be expected in pre-RC components or geminin, to relieve the repression of re-replication.

The licensing system in cancer diagnosis and prognosis

The development of cell proliferation markers has revolutionised the early detection of malignant cells. One of the most widely used proliferation markers is Ki67, a protein of unknown function which is present in cycling cells. Ki67 shows a good correlation with bromodeoxyuridine labelling, thus validating its reliability as a marker of proliferation (52). It is unclear, however, whether the Ki67 labelling index provides additional prognostic information and what levels of Ki67 labelling distinguish between good and poor prognosis for tumours. Moreover, Ki67 may not be expressed in cells entering G1 from G0 (53) and its expression may also be down-regulated in proliferating cells by nutritional deprivation (54). Proliferating cell nuclear antigen (PCNA), the auxiliary factor for DNA polymerase δ, has also found use as a marker for the assessment of proliferation (55). PCNA gives good immunostaining in nuclei during S phase, but because of its role in DNA repair synthesis as well as chromosome replication, can also give a signal when repair is occuring (56,57), thus eroding the quantitative relationship between PCNA expression and proliferation (58).

One limitation of PCNA and Ki67 is that they effectively label cells actively engaged in later cell cycle events, but do not label cells in G1. Since proliferating cells may spend most of their time in G1, this significantly reduces the sensitivity of PCNA and Ki67 as proliferation markers. Mcm2-7 and Cdc6 are present in G1 as well as S phase, but are lost as cells withdraw from the cell cycle (Fig 1) and are therefore potentially more sensitive markers of proliferative capacity. We have recently proposed that the presence of Mcm2-7 can be used as a functional definition of cells that are in G1 phase (4). Therefore Mcm2-7 provide markers for cells that have high proliferative potential, even if they are dividing only very slowly.

The diagnostic potential of Mcm2-7 detection is revealed by analysis of cervical epithelium low- and high-grade squamous intraepithelial lesions (LSILs and HSILs), both of which are viewed as representing a potential precursor of malignancy (59). As shown in Fig 5, antibodies to both PCNA and Mcm5 stained the basal proliferating layers of normal cervical squamous epithelium but did not stain the superficial differentiating cells. This shows that terminal differentiation of the superficial cells is accompanied by down-regulation of Mcm5. In LSIL and HSIL there are increasing numbers of PCNA- and Mcm5-positive cells seen among the superficial cells, showing that these cells are failing to withdraw from the cell cycle as they should. It is also clear that a much higher proportion of cells stain for Mcm5 than stain for PCNA, as expected from their cell cycle behaviour. A similar increase in the Mcm-staining cells is also seen in premalignant lesions of the lung (60) and in kidney tumours (61), whilst increased Cdc6 staining was detected in brain tumours (62).

Figure 5.

Diagnostic potential of Mcm2-7 detection. Frozen sections of normal cervix, LSILs, and HSILs were immunostained for Mcm5 or PCNA by the immunoperoxidase method. In normal cervix, surface cells likely to be sampled by cervical smear examination are negative with both antibodies. In LSIL, anti-PCNA antibodies stain basal and some parabasal nuclei but surface cells are negative; in contrast, anti-Mcm5 antibodies stain nuclei in superficial as well as basal epithelial layers. In HSIL, anti-PCNA antibodies show focal staining of 10% of nuclei in the surface layers, whereas anti-Mcm5 antibodies stain virtually all nuclei. Reproduced from (59).

Mcm2-7 staining may also have prognostic value in cancers. In non-small-cell lung cancers, an increased Mcm2 immuno-reactivity corresponded to poorer survival, whilst no such correlation was seen with Ki67 (63). A similar correlation between high Mcm2 expression and poor prognosis was reported in prostate cancers (64) and in oligodendrogliomas (65). These correlations presumably reflect the increased proliferative capacity of licensed cells. It would be of considerable interest to determine whether Mcm2-7 expression also provides a marker for responsiveness to particular treatment regimes.

An exception to the normal correlation between the absence of Mcm2-7 and withdrawal from the cell cycle proliferation occurs in glandular epithelial cells of the breast. Even though the proliferative rate was very low in the breast cells of non-pregnant and non-lactating women, a large percentage (47-65%) stained positive for Mcm2-7 (66). As the lack of origin licensing in non-proliferating cells may be an important barrier to their inadvertent re-entry into the cell cycle, these licensed but slowly proliferating breast cells might therefore be at a higher risk of undergoing malignant transformation.

The licensing system as a potential anti-cancer target

Not only are the components of the licensing system useful in early cancer detection, they can potentially be exploited for the design of novel anti-cancer drugs which are highly target specific. This potential arises from the fact that most normal proliferating cells in the human body are likely to have the option of temporarily withdrawing into the G0 (unlicensed) state. Many cancer cells may be unable to take this option, and instead be locked into a state of continuous proliferation. If the licensing system was inhibited by a small molecule, it is plausible that normal cells finding their origins unlicensed would be able to respond to this by activating a checkpoint and undergoing temporary arrest in a G0-like state. When the drug was removed, re-licensing of origins and re-entry into S phase should then be possible. When treated with the same drug, however, a cancer cell may be unable to arrest in a G0-like state and instead would be forced to try to progress into S phase with unlicensed origins. Since origin licensing is essential for DNA replication, the cell would become blocked in an S phase state due to its inability to complete replication of its DNA. This arrest would be irreversible, since even on withdrawal of the drug, such cells would be unable to re-license origins due to the combined activity of endogenous CDKs and geminin that are active in S phase. If these suppositions are correct, transient inhibition of the licensing system would cause a transient cell cycle arrest of normal cells whilst killing cancer cells.

At present, this idea is hard to test due to the lack of inhibitors of the licensing system. The only reported licensing inhibitor is geminin, which has a ~10 kDa domain with inhibitory activity (16). Given the difficulty in delivering a molecule of this size into cells, a cell-permeant small molecule that inhibits replication licensing is required to fully test these ideas.