Review

Regulation of cyclin D1 is central to both steroid and growth factor-induced mitogenesis in mammary epithelial cells and increased cyclin D1 expression has been implicated in mammary carcinogenesis [1]. Cyclin D1 activates two cyclin-dependent kinases (CDKs), Cdk4 and Cdk6, that regulate progress from G₁ phase into S phase. Increased cyclin D1-Cdk4/6 levels can also indirectly activate another G₁ CDK, Cdk2, via sequestration of the CDK-inhibitory proteins p21^WAF1/Cip1 and p27^Kip1. Phosphorylation of pRB, the product of the retinoblastoma susceptibility gene, by cyclin D1-Cdk4/6 and cyclin E-Cdk2 results in derepression of E2F-mediated transcription, and promotes cell cycle progression.

Cyclin D1's role in the control of cell cycle progression and the location of the cyclin D1 gene at chromosome 11q13, a locus that is frequently amplified in breast and other cancers, led to early investigations of its role as a potential mammary oncogene. These showed that cyclin D1 overexpression occurs in approximately 45% of breast cancers and that overexpression of cyclin D1 in the mammary gland of transgenic mice led to hyperplasia and eventual development of carcinoma [1]. A prevailing view has been that the ability of cyclin D1 to promote proliferation by activating CDKs likely accounts for its ability to promote tumour development. However, cyclin D1 levels do not correlate with Cdk4 activity in breast cancer cell lines [2], nor do they correlate with other markers of proliferation in human breast cancer tissue (reviewed in [3]), suggesting that this simple model may not be able to fully account for cyclin D1's activity as an oncogene. As well as Cdk4/6, cyclin D1 also interacts with a number of transcriptional regulators including the oestrogen receptor [4] and its ability to act as a transcriptional cofactor provides an additional potentially oncogenic function that is independent of Cdk4 binding and hence separate from its role in cell cycle progression. Expression of a point mutant of cyclin D1 that is unable to activate Cdk4 induced a set of genes that were also frequently co-expressed with cyclin D1 in primary human cancers, implicating transcriptional regulation as an alternative mechanism for cyclin D1-mediated oncogenesis [5]. Nelsen et al. [6] now provide evidence for a further mechanism of cyclin D1 action that may impact on its oncogenic capacity: induction of centrosome abnormalities.

Centrosomes are the microtubule-organising centres of the cell. Increases in centrosome number can cause the formation of multipolar spindles and consequent chromosome mis-segregation during mitosis, consistent with the hypothesis that the frequent abnormalities in centrosome number observed in human cancers, including breast cancer, may be linked with aneuploidy [7,8]. In their recent report Nelsen et al. [6] have transiently transfected hepatocytes in vivo using a cyclin D1-expressing adenovirus. Within 6 days of cyclin D1 adenovirus injection they observed abnormal mitoses with multipolar spindles, and cultures of explanted hepatocytes displayed increased centrosome numbers [6]. Flow cytometry of these cultures revealed a majority of cells with DNA content of ≥4 N, compared with the minor population of such cells in control cultures [6]. Analysis performed 2 days after adenoviral infection, when spindle morphology was still normal but the centrosome number had increased, showed a similar increase in the proportion of cells with DNA content of ≥4 N [6]. This was not simply due to induction of tetraploidy since cyclin D1 overexpression was associated with both numerical and structural chromosome abnormalities. Not unexpectedly, these abnormalities led to increased apoptosis, but nonetheless cells with abnormal numbers of centrosomes were apparent 4 months after adenoviral expression of cyclin D1, a time-point when cyclin D1 expression was no longer detectable[6]. Collectively these data indicate that cyclin D1 overexpression can cause centrosomal and chromosomal abnormalities that are not necessarily incompatible with ongoing proliferation. Importantly, Nelsen et al. go on to show that infection of cultured human breast epithelial cells with cyclin D1 adenoviruses also induces centrosome abnormalities [6]. Abnormalities in centrosomal size and/or number are correlated with chromosomal instability and are common in both in situ and invasive breast cancers [8], and altered cyclin D1 expression is an early event in breast cancer [1], so these new data raise the possibility that one way in which cyclin D1 overexpression may contribute to the development or progression of breast cancer is by inducing centrosomal or chromosomal abnormalities, although clearly further experimentation is necessary to substantiate this hypothesis.

One issue that impacts on the relevance of these data to breast and other cyclin D1-overexpressing cancers is whether the level of overexpression achieved in these experiments is comparable with the relatively modest cyclin D1 overexpression reached in these cancers. In an earlier publication, Nelsen et al. show that the maximal cyclin D1 level, achieved 2 days after adenovirus injection, is several fold greater than that after partial hepatectomy [9] although likely still within the range expected in cyclin D1-overexpressing cancers. Approximately half of the adenovirally-expressed cyclin D1 was present in large molecular mass complexes that did not appear to contain Cdk4 or Cdk6 [9]. These complexes were barely detectable after partial hepatectomy and thus their formation may largely be driven by supra-physiological levels of cyclin D1.

The observation that a molecule better known for its role in cell cycle control may also be involved in centrosomal replication is not unprecedented. To ensure that the centrosome is replicated once and only once in each cell cycle, there is close co-ordination between the control of centrosomal duplication and DNA replication. There is a clear role for aberrations in the p53 pathway in centrosomal amplification, probably through failure of the G₁/S checkpoint, and studies with viral oncoproteins have implicated the RB pathway in restricting centrosome duplication to once per cell cycle [7,8]. Centrosome duplication requires Cdk2 activation, and a proportion of cyclin E localises to the centrosome [7,10]. However, although cyclin E overexpression in fibroblasts or breast epithelial cells caused chromosome instability and polyploidy, this was not associated with abnormal centrosome numbers [11]. Similarly, Nelsen et al. [6] found no effect of cyclin E overexpression on centrosome numbers or spindle morphology in hepatocytes, in contrast to the effects of cyclin D1.

The intriguing observations of Nelsen et al. raise a number of questions about potential mechanisms whereby cyclin D1 might affect centrosomal replication. Is this effect mediated by a direct effect at the centrosome, or indirectly, perhaps by uncoupling the centrosomal duplication and DNA replication cycles? The authors have not examined the subcellular localisation of the overexpressed cyclin D1, so it is unclear whether a direct action is likely. Given the evidence for both CDK-dependent and -independent functions of cyclin D1, do its effects on centrosomal replication require CDK activation? In this context it will be particularly interesting to define the composition and potential functions of the high molecular weight cyclin D1 complexes apparently lacking Cdk4 and Cdk6 that form after cyclin D1 overexpression [9]. The effects of pRB deletion in hepatocytes include polyploidy but not centrosome amplification [12], indicating that loss of this essential target of cyclin D-dependent CDKs is not equivalent to cyclin D1 overexpression and thus suggesting that the effects of cyclin D1 on centrosome number may not be wholly mediated by the RB pathway. Whatever the mechanism, the observations of Nelsen et al. suggest new avenues for investigation that may be relevant to understanding the role of cyclin D1 in mammary carcinogenesis.