T umour-suppressor genes are needed to keep cells under control. Just as a car’s brakes regulate its speed, properly func- tioning tumour-suppressor genes act as brakes to the cycle of cell growth, DNA repli- cation and division into two new cells. When these genes fail to function properly, uncon- trolled growth — a defining feature of cancer cells — ensues. The p53 gene, first described in 1979, was the first tumour-suppressor gene to be iden- tified. It was originally believed to be an oncogene — a cell-cycle accelerator (Box 1) — but genetic and functional data obtained ten years after its discovery showed it to be a tumour suppressor. Moreover, it was found that the p53 protein does not function cor- rectly in most human cancers (Fig. 1). In about half of these tumours, p53 is inactivat- ed directly as a result of mutations in the p53 gene. In many others, it is inactivated indi- rectly through binding to viral proteins, or as a result of alterations in genes whose prod- ucts interact with p53 or transmit informa- tion to or from p53. The realization that p53 is a common denominator in human cancer has stimulat- ed an avalanche of research since 1989. Dur- ing that time there have been over 17,000 publications centred on p53 — 3,300 in the past year alone — and over 10,000 tumour- associated mutations in p53 have been dis- covered, in organisms ranging from humans to clams 1,2 . As might be expected, this work has led not only to considerable insights into tumour development, but also to consider- NATURE | VOL 408 | 16 NOVEMBER 2000 | www.nature.com 307 news and views feature Surfing the p53 network Bert Vogelstein, David Lane and Arnold J. Levine The p53 tumour-suppressor gene integrates numerous signals that control cell life and death. As when a highly connected node in the Internet breaks down, the disruption of p53 has severe consequences. able confusion and controversy. Here we sug- gest that signalling pathways involving p53 — like cellular signalling pathways in general — cannot be understood by looking at isolated components. Instead, it is essential to consid- er the tangled networks into which these sig- nalling components are integrated. Activating the p53 network The p53 network is normally ‘off’. It is activat- ed only when cells are stressed or damaged. Such cells pose a threat to the organism: they are more likely than undamaged cells to con- tain mutations and exhibit abnormal cell- cycle control, and present a greater risk of becoming cancerous. The p53 protein shuts down the multiplication of stressed cells, inhibiting progress through the cell cycle. In many cases it even causes the programmed death (apoptosis) of the cells in a desperate attempt to contain the damage and protect the organism. The p53 protein therefore pro- vides a critical brake on tumour develop- ment, explaining why it is so often mutated (and thereby inactivated) in cancers. What sort of stresses, then, activate the p53 network? Early work focused on DNA damage as the ‘on’ switch. A single break in a double-stranded DNA molecule may be suf- ficient to trigger a rise in levels of p53 protein. This remarkable sensitivity to DNA damage confounded subsequent studies that sought to establish whether the p53 response could be triggered by other signals. It was difficult to show that these other signals did not cause at least a few breaks in double-stranded Mechanism of inactivating p53 Amino-acid-changing mutation in the DNA- binding domain Deletion of the carboxy- terminal domain Multiplication of the MDM2 gene in the genome Viral infection Deletion of the p14 ARF gene Mislocalization of p53 to the cytoplasm, outside the nucleus Typical tumours Colon, breast, lung, bladder, brain, pancreas, stomach, oesophagus and many others Effect of inactivation Prevents p53 from binding to specific DNA sequences and activating the adjacent genes Occasional tumours at many different sites Sarcomas, brain Breast, brain, lung and others, expecially when p53 itself is not mutated Cervix, liver, lymphomas Prevents the formation of tetramers of p53 Extra MDM2 stimulates the degradation of p53 Products of viral oncogenes bind to and inactivate p53 in the cell, in some cases stimulating p53 degradation Failure to inhibit MDM2 and keep p53 degradation under control Breast, neuroblastomas Lack of p53 function (p53 functions only in the nucleus) Figure 1 The many ways in which p53 may malfunction in human cancers. Oncogenes. These are analogous to the accelerators in a car. Oncogenes stimulate appropriate cell growth under normal conditions, as required for the continued turnover and replenishment of the skin, gastrointestinal tract and blood, for example. A mutation in an oncogene is tantamount to having a stuck accelerator: even when the driver releases his foot from the accelerator pedal, the car continues to move. Likewise, cells with mutant oncogenes continue to grow (or refuse to die) even when they are receiving no growth signals. Examples are Ras, activated in pancreatic and colon cancers, and Bcl-2, activated in lymphoid tumours. Tumour-suppressor genes. When the accelerator is stuck to the floor, the driver can still stop the car by using the brakes. Cells have brakes, too, called tumour-suppressor genes. These keep cell numbers down, either by inhibiting progress through the cell cycle and thereby preventing cell birth, or by promoting programmed cell death (also called apoptosis). Just as a car has many brakes (the foot pedal, handbrake and ignition key), so too does each cell. When several of these brakes are rendered non-functional through mutation, the cell becomes malignant. Examples are the gene encoding the retinoblastoma protein, inactivated in retinoblastomas, p53 (Fig. 1), and p16 INK4a , which inhibits cyclin-dependent kinases and is inactivated in many different tumours. Repair genes. Unlike oncogenes and tumour-suppressor genes, repair genes do not control cell birth or death directly. They simply control the rate of mutation of all genes. When repair genes are mutated, cells acquire mutations in oncogenes and tumour-suppressor genes at an accelerated rate, driving the initiation and progression of tumours. In the car analogy, a defective repair gene is much like having a bad mechanic. Examples are nucleotide-excision- repair genes and mismatch-repair genes, whose inactivation leads to susceptibility to skin and colon tumours, respectively. Box 1 The genes that cause cancer © 2000 Macmillan Magazines Ltd