1.  INTRODUCTION

       
     The search for genetic alterations that give rise to mammary cancer, in animal models or humans, has revealed several growth factors, growth factor receptors, as well as cell cycle regulators, that can act as proto-oncogenes when they are over-expressed or constitutively activated through somatic mutation. In the mouse, where mouse mammary tumour virus has been shown to act as a powerful insertional mutagen, there is a preferential activation of dominant acting growth and differentiation factors, such as fibroblast growth factors (FGFs) or members of the wnt gene family (reviewed). In human breast cancer, the inappropriate activation of cell cycle genes such as c-myc and cyclin D1 often in conjunction with loss of tumour suppressor genes appears to be more common, although mutation or over-expression of the epidermal growth factor-tyrosine receptor gene family also plays a significant role (reviewed).
        
        During the last decade, there has been a massive expansion in understanding of the genes involved in normal mammalian development which has implicated many proto-oncogenes, as important signalling molecules in pattern formation during organogenesis. The mammary gland is unusual in that a large proportion of its development occurs after birth, through puberty and during pregnancy and lactation. A key question was therefore whether, when correctly regulated, the oncogenes implicated in mammary cancer are involved in normal mammary gland development. We have examined this question with regard to FGF-3 (originally called int-2 ) which is inappropriately expressed in many MMTV induced mouse mammary tumours, and cyclin D1 which is abnormally elevated in 40% of human breast cancers.

       
     Fibroblast growth factors are pleiotropic cell to cell signalling molecules that can act as broad spectrum mitogens, promote cell migration or modulate cellular differentiation depending on their context (reviewed). These secreted ligands signal by binding to high affinity cell surface receptors, inducing their dimerization and subsequent activation of their cytoplasmic tyrosine kinase (reviewed). FGF-receptors contain two or three immunoglobin-like loops in the extracellular domain, a transmembrane element as well as the cytoplasmic tyrosine kinase. There are four FGF receptor genes ( FGFR-1 to FGFR-4 ), in mammals, although alternative splicing of FGFR-1, FGFR-2 and FGFR-3, yields seven proto-type receptors with different FGF binding specificities and tissue distributions. The two membrane proximal Ig-like domains constitute the ligand binding site, and it is part of the third Ig-loop which is encoded by an alternative exon (designated IIIb or IIIc) that defines FGF binding specificity.
       
        As FGF-3 is not found in the normal mammary gland, it seemed likely that an FGF with a receptor binding specificity overlapping that of FGF-3 was important for gland development. To determine the role such an FGF might have, we sought to compromise FGF signalling in the mammary epithelium and assess the effect this had during puberty and pregnancy. 
        
         FGF-3 binds to the IIIb isoform of both FGFR- 1 and FGFR-2, and the latter receptor was selected for expression as a dominant negative receptor in the mammary epithelium of transgenic mice. The dominant negative receptor gene was constructed by removing the cytoplasmic tyrosine kinase domain from an appropriate mouse cDNA, leaving the extracellular and transmembrane coding domains intact. Hence, ligand mediated dimerization of the truncated receptor protein potentiates the sequestration of wild-type endogenous receptors as inactive heterodimers. Expression in the mammary epithelium was achieved using the MMTV promoter, which has a particularly high expression in this cell lineage, especially during pregnancy. Whole mount preparations and histological examination of the mammary glands from adult virgin females showed no discernible abnormalities associated with transgene expression. However, by mid-pregnancy, the mammary glands of transgenic mice showed a distinct reduction in lobuloalveolar development that was maintained into lactation, although those alveolar lobules that developed showed a normal histology, and were filled with milk (Fig. 1). These findings indicate that FGF signalling is an important facet of lobuloalveolar development during pregnancy. Moreover, it explains why inappropriate and deregulated FGF-3 expression in this tissue leads to mammary hyperplasia since this would institute autocrine stimulation of epithelial cell growth.

  3. CYCLIN D1 IS REQUIRED FOR MAMMARY GLAND GROWTH AND DIFFERENTIATION
  
      The FGF-3 locus maps to human chromosome 11 band q13 and is amplified in approximately 15% of breast tumour DNA samples, although the gene is very rarely expressed in these tumours. This suggested that amplification of FGF3 was fortuitous, and that another gene on the amplicon was driving tumour development. The cyclin D1 gene ( CCND1) was identified as a likely candidate oncogene since it was closely linked to and always co-amplified with FGF-3. Furthermore, CCND1 was involved in chromosomal translocations for a subset of lymphomas and parathyroid adenomas further supporting its potential as an oncogene. Moreover, immunohistochemical examination of breast cancer samples revealed that cyclin D1 is expressed at elevated levels in 40% of tumour samples, considerably more than the subset accounted for by gene amplification.