Growth of non-transformed cells in vitro is regulated by density-dependent mechanisms via cell-cell contacts, leading to arrest in late G1-phase at con¯uency (contact-inhibition of growth). In the present study it is shown that this results from p16 INK4 -mediated dissociation of the complex cdk4-cyclin D1, which is responsible for the inactivation of the gate keeper of G1-S transition, the retinoblastoma protein pRb. As a consequence of the inactivation of cdk4, downstream the activation of cdk2 and hyperphosphorylation and thus inactivation of pRb was impaired. Direct evidence for the central role of p16 INK4 in growth control comes from the observation that a competitive inhibitor of p16 INK4 repressed contact inhibition of growth. These ®ndings provide an explanation for the high incidence of mutation or loss of INK4 in human tumours.Keywords: contact-inhibition; cdk4; p16 INK4 The cell cycle is regulated by an ordered cascade of phosphorylation reactions by a speci®c family of serine-threonine kinases, the cyclin-dependent kinases (cdks) (Lees, 1995;Sherr, 1996), and speci®c check points ensure that the cell's proper cycle is tightly controlled. In order to determine at which point of the G1-phase cells are arrested by contact-dependent inhibition of growth, time course studies were performed. In FH109 cells, human embryonal diploid lung ®broblasts (Wieser et al., 1985), contactdependent inhibition of growth is exclusively mediated by the interaction of two cell membrane proteins on adjacent cells, i.e. by the glycoprotein, contactinhibin (Wieser et al., 1990), which binds to its receptor referred to as contactinhibin receptor (Gradl et al., 1995).Cell-cell contacts were imitated by the addition of glutaraldehyde-®xed cells (in which contactinhibin remains active, as its growth inhibitory activity is mediated exclusively by its N-glycans not a ected by the ®xation process) to sparsely seeded ®broblasts in the presence of serum. In previous studies we have demonstrated that this system mimics physiological conditions concerning cell proliferation and differentiation which occur in con¯uent cultures (Wieser et al., 1985;Wieser and Oesch, 1986), thus enabling the study of early e ects of cell-cell contact-dependent signals. FH109 cells were sparsely seeded and cultured for 24 h in DMEM/0.5% FCS to achieve partial synchronization. Control cells were stimulated with 10% FCS in DMEM whereas contact-inhibited cells were generated by the addition of ®xed cells in DMEM/10% FCS prepared from con¯uent cultures (Wieser et al., 1985). These experiments revealed that when ®xed cells were added within the ®rst 8 h after serum-stimulation, proliferation was inhibited by 70 ± 80% as determined by [ 3 H]-thymidine incorporation, which correlates with the reduction of growth rate achieved by seeding ®broblasts to con¯uency (Figure 1). The addition of ®xed cells 10 h after serum stimulation no longer resulted in any growth inhibition, indicating that contact-dependent inhibition of growth is limited to a window in the cell ...