Antisense to Cyclin D1 Inhibits VEGF-Stimulated Growth of Vascular Endothelial Cells: Implication of Tumor Vascularization

Cyclin D1 is a breast cancer oncogene initially identified as a breakpoint rearrangement in parathyroid adenoma and cloned by three independent groups (10–12). The cyclin D1 gene encodes the regulatory subunit of the holoenzyme known to phosphorylate and inactivate the retinoblastoma protein. The identification of clonal DNA alterations in primary human tumors reflects in vivo selection, providing evidence for the importance of that gene product in neoplasia. The 11q13 chromosome rearrangement and/or amplification are frequently observed in multiple types of human cancer. At the breakpoint of the characteristic t(11;13)(q13;q32) clonal translocation in mantle cell lymphoma and in a subset of multiple myeloma, the cyclin D1 gene was proven to encode the BCL1 oncogene. The 11q13.4 to 11q13.5 is amplified in ∼20% of breast cancers within the 11q13 region. At this time, the cyclin D1 gene has been clearly defined as a collaborative oncogene. The abundance of cyclin D1 is strongly overexpressed when amplified and targeted expression of cyclin D1 to the mammary gland in transgenic mice was sufficient for the induction of mammary tumors with a long latency.

Overexpression of cyclin D1 has been observed in up to 50% of human breast cancers, including the early phases of ductal carcinoma in situ, suggesting that cyclin D1 may be involved in the early stages of mammary tumorigenesis. Curiously, however, cyclin D1 overexpression in tumors is not associated with markers of cellular proliferation, which is expected, as cyclin D1 is capable of functioning as a component of the holoenzyme that phosphorylates the retinoblastoma protein, promoting DNA synthesis in cultured cells. Additional studies have examined cyclin-dependent kinase–independent functions of cyclin D1. A body of evidence for additional functions for cyclin D1 includes the effects of cyclin D1 on genomic instability, DNA replication and repair in UV-treated fibroblasts, the physical association between cyclin D1 and proliferating cell nuclear antigen, and the interaction between cyclin D1 and a variety of transcription factors. Curiously, a subset of cyclin D1 transformation properties did not always require interaction with the retinoblastoma proteins.

Deletion of cyclin D1 in mice results in animals with modest developmental abnormalities, including failed terminal alveolar breast bud development and retinal apoptosis. Cyclin D1^−/− mouse embryonic fibroblasts have reduced DNA synthesis and increased apoptosis in response to multiple stimuli, including UV irradiation (13, 14). The blood vessels of cyclin D1^−/− mice are defective in their invasion of Matrigel (9), and widespread abnormalities of lipid metabolism have been identified (7). Like the blood vessels, the macrophages of cyclin D1^−/− mice are defective in migration (15), suggesting that, in addition to playing a role in differentiation of mammary epithelial cell precursors and cellular survival, cyclin D1 plays a role in diverse biological processes regulating normal physiology. In examining the molecular mechanisms by which cyclin D1 regulates normal physiologic processes or conducts its transforming function, microarray analysis has been conducted. Studies conducted of tumors overexpressing cyclin D1 identified a molecular signature in which nonretinoblastoma targets were regulated (16). These studies identified CEBP, a protein involved in fat cell differentiation, as a key target of cyclin D1. Due to the complexity of molecular changes occurring during tumor onset and progression, other studies examined the more direct transcriptional targets of cyclin D1 by reintroducing cyclin D1 into cyclin D1^−/− cells (8). Intriguingly, these studies identified a subset of genes involved in cellular differentiation and antiangiogenesis as molecular targets repressed by cyclin D1. In this regard the antiangiogenic factor TSP-1 and STAT signaling were repressed by cyclin D1 (Fig. 1).

Cyclin D1 regulation of angiogenic factors. A variety of studies have shown that the cyclin D1 gene itself is a downstream target of both antiangiogenic and angiogenic factors. VEGF and fibroblast growth factor induce cyclin D1 abundance, and endostatin inhibits cyclin D1 expression inducing a G₁ cell cycle arrest (17). Adiponectin (ACRP30) is a potent inhibitor of angiogenesis. Cyclin D1 is also known to inhibit secretion of ACRP30 (6, 7).

The first naturally occurring inhibitor of angiogenesis to be identified was TSP-1. Several growth factors have been identified as positive regulators of angiogenesis, including VEGF (2) and basic fibroblast growth factor (18). Negative regulators of angiogenesis include TSP-1 (5), endostatin (19), adiponectin (ACRP30; ref. 6), and angiostatin (20). The role of ACRP30 remains controversial (21). TSP-1 inhibits the activity of matrix metalloproteinase-9 (MMP-9), the extracellular metalloprotease that releases VEGF sequestered in the extracellular matrix. MMP-9 is induced by a variety of molecules and signaling pathways, including estrogen, c-Jun, nuclear factor-κB, and growth factors, which in turn induce mitogen-activated protein kinase. TSP-1 binds to the CD36 receptor present on the endothelial cell surfaces to further inhibit angiogenesis. TSP-1 itself is induced by a variety of growth factors, including serum and platelet-derived growth factor. However, oncogenic signals, such as Ras and Myc, inhibit TSP-1 via ROCK. PTEN is also capable of increasing TSP-1 expression (22). Several observations suggest that TSP-1 repression by Ras involves phosphatidylinositol 3-kinase signaling pathway via Rho and ROCK (23). The findings that TSP-1 was inhibited by cyclin D1 (8) and that cyclin D1 induces VEGF expression (Fig. 1; ref. 24) provide further evidence for a model in which cyclin D1 may indirectly promote angiogenesis.

Cyclin D1 and STAT3 in angiogenesis. STATs are latent transcription factors that are activated by cytokines and growth factors. The activation of STATs is dependent on protein phosphorylation and acetylation, which induces dimerization. On activation, STATs translocate to the nucleus, inducing binding to key cis elements in known target genes. Typically, STAT tyrosine phosphorylation is transient; however, in primary tumors, persistent tyrosine phosphorylation of STAT proteins is frequently observed. Constitutively activated STAT3 transforms fibroblasts and mammary epithelial cells (25, 26). STAT3 is known to induce genes involved in proliferation and apoptosis, including cyclin D1, c-myc, bcl-xl, and VEGF. Like cyclin D1, STAT3 is required for skin tumorigenesis induced by oncogenic Ras (17). STAT3 and STAT5 both activate cyclin D1 expression and transcription (25, 27). Furthermore, cyclin D1 expression is inversely correlated with caveolin-1, which in turn has been shown to regulate STAT signaling (28, 29). Transformation by STAT3 of mammary epithelial cells requires the activity of MMP-9 (26). STAT3 itself is a direct inducer of the MMP-9 promoter, and the correlation between activated STAT3 and MMP-9 expression in breast tumors provides additional evidence for an important role of STAT3 regulation of MMP-9 in mammary tumor progression. In studies by Yasui et al., cyclin D1 antisense was shown to decrease VEGF protein production. Although only a modest reduction in VEGF promoter activity was observed through reducing cyclin D1 abundance, additional factors may be involved in reducing VEGF abundance. The finding that cyclin D1 is induced by STATs and that cyclin D1 itself inhibits STAT signaling (8) suggests the presence of tightly controlled feedback loops within the cell that normally attenuate the activity of STATs.

Additional factors regulated by cyclin D1 include Dickkopf homologue (Dkk)-1 (8). Cyclin D1 inhibits Dkk expression and transcription (8). Furthermore, cyclin D1 inhibited p300-dependent induction of Dkk. Wnt signaling, which plays an important in cell fate determination, tissue patterning, tumorigenesis, and angiogenesis, is antagonized by several secreted proteins, including Dkk, Cereberus, the Frizzled-related proteins (30), and Wnt inhibitory factor. Frizzled-related proteins, Cereberus, and Wnt inhibitory factor bind and sequester Wnt. Dkk inhibits Wnt signaling by interaction with LRP5/LRP6 (31). Dkk-1 down-regulates β-catenin and increases expression of epithelial differentiation markers (32). Dkk-1 binds to the Wnt coreceptor LRP6, causing its endocytosis through the formation of a ternary complex with the transmembrane protein. As LRP6 activates cellular Wnt/β-catenin signaling, Dkk is a specific antagonist of Wnt signaling (32). The finding that cyclin D1 down-regulates Dkk expression provides a mechanism by which cyclin D1 may modulate and fine-tune patterns of Wnt activity and thereby its downstream functional activities.

Collectively, these observations provide compelling evidence that cyclin D1 regulates extracellular signals that either fine-tune the recruitment of blood vessels and their growth, such as TSP-1 and VEGF, or regulate intracellular signaling pathways that in turn regulate angiogenic factors, such as Dkk, ACRP30, and STAT3 signaling (8).

How might cyclin D1 coordinate such diverse genetic changes leading to the production of proangiogenic factors and heterotypic signals? One mechanism may involve transcriptional regulation. Cyclin D1 is known to regulate >30 different transcription factors (33). The mechanisms by which cyclin D1 inhibits transcription factor activity include the recruitment of histone deacetylases (HDAC; refs. 34, 35). In this regard, cyclin D1 recruits HDAC1 and HDAC3 together with SUV39 and HP1 to the promoters of target genes regulating cellular differentiation (34). As several transcription factors repressed by cyclin D1 were also activated by the coactivator p300, a formal dissection of the molecular mechanisms by which cyclin D1 regulates p300 activity had been conducted (8). These studies showed that cyclin D1 inhibits transcription factor activity in part through transcriptional repression of the p300 coactivator. Cyclin D1 bound p300 and inhibited the histone acetyl transferase activity of p300. A molecular signature of genes induced by p300 was repressed by cyclin D1, consistent with a model in which cyclin D1 antagonized p300 function (8).

An alternate mechanism by which cyclin D1 may regulate expression of diverse genes may involve the recently discovered ability of cyclin D1 to regulate chromatin remodeling proteins and thereby multiple different genes. Cyclin D1 has been identified within the local chromatin structure at transcription factor binding sites (34, 35). Moreover, cyclin D1 is capable of inhibiting HDAC activity in vitro. The association of cyclin D1 with HDACs may coordinate transcriptional repression or activation function (34). These interactions with HDACs may in turn regulate epigenetic changes and thereby coordinate altered expression of multiple genes. Thirdly, cyclin D1 may regulate the expression of multiple genes governing angiogenesis through the small monomeric GTPases. p27^KIP1 regulates Rho GTPase activity (36). Cyclin D1 physically interacts with p27^KIP1, and cyclin D1^−/− mice when mated to p27^−/− animals provide partial rescue of their mammary epithelial cell and retinal developmental abnormalities. These findings are consistent with the notion that cyclin D1 and p27^KIP1 may function in trans for a subset of developmental roles. Together, these studies raise the possibility that cyclin D1 may interact with p27^KIP1 and thereby regulate Rho/Rac signaling. Understanding the mechanisms by which cyclin D1 regulates angiogenesis may provide alternate avenues for improving therapeutic outcome for our patients affected by cancer.