So what did they make of this extraordinary picture? What this means, explains Evan, is that if you don’t have p53 around when the DNA damage is occurring, most cells will be repaired. In real life this is a rather hit-and-miss affair, a patch-up process that leads to the kind of ‘mistakes’ that are the driving force behind evolution. But if you then restore p53 after this repair process has had time to work, the gene will only be activated in those cells that have sustained mistakes, or mutations, that make them dangerous (for example, ones that activate oncogenes) and therefore send out alarm signals to abort. ‘This basically means that you can separate out the DNA damage response from the tumour suppressor response,’ he comments.
This has enormous implications for treatment of cancer, because it implies that we could devise ways to prevent most of the dreadful side effects of chemo- and radiotherapy – the hair loss, nausea, exhaustion, immune suppression – that are the direct consequence of hitting all the body’s fast-dividing cells, and clear out just those cells that go on to become cancerous and that therefore continue to send out alarm signals that activate p53. But how did Evan’s colleagues react to his findings and his theories?
‘You know ideas like this take a long time to percolate through. I mean getting it published . . . I remember one of the reviewers just said, “I refuse to accept that the DNA response is not the major tumour suppressor pathway.” But this is not a faith-based thing; we’re not a religion! These are the data. And I wasn’t saying, “You’re all wrong.” I was saying, “These are the data. This is our explanation. This is our hypothesis.” The whole point about publishing in the literature is that you publish the data; you publish the hypothesis so that it can be tested by the community.
‘That experiment was a very intriguing experiment, and I think a very informative one. And I think the conclusions of it still stand. But the point is that cancers arise in many different tissues and many different ways, and the issue for me is much less about are these data wrong or are these data right than about getting to understand which set of rules apply in which case.’
A FINE BALANCE BETWEEN LIFE AND DEATH
Evan had met even stronger resistance to his ideas some years earlier when his research into oncogenes suggested to him that all our cells contain both growth and suicide programmes that are in constant, hair-trigger competition. Which course of action a cell takes is essentially controlled by its environment and the signals it receives from its neighbours: is it in the right place and at the right time? Is it behaving normally? If so, it will receive ‘stay alive’ signals; if not, it will be instructed to abort. This is an inbuilt defence mechanism and one of the reasons cancer is so rare, believes Evan.
Since this story takes us back among the Petri dishes in the lab, it may seem like a diversion from the topic in hand, animal models. But besides describing another pivotal moment in cancer research, it helps show why it is so important that experiments are performed in living organisms as well as in cell and tissue cultures. It begins in the late 1980s, when Evan, newly recruited to the ICRF in London, was doing some experiments with the powerful oncogene Myc, looking at how it drives cell proliferation. ‘I made this bizarre observation that when you expressed Myc at high levels in cells they did indeed proliferate – but when you looked a couple of days later, there were fewer cells than before,’ he explains. ‘I’m a great believer in personal observation – observing things, you ask questions of what you can actually see. So we took these cells and we put them under a microscope and we used time-lapse video to take one frame every three minutes. Then you speed it up and watch what happens over three or four days in just two or three minutes. And there we saw this amazing phenomenon . . . the cells were replicating, but also they were dying by apoptosis.’
This was exciting, but it didn’t make sense. Then a thought struck Evan that drew on his early training in immunology, where a common theory was that the immune system plays a part in protecting us from cancer by eliminating rogue cells that it recognises as foreign. ‘I thought what if, instead of the immune system acting as a police service to find aberrantly proliferating cells, there is, hard-wired into the very warp and weft of how cells proliferate, an abort programme? Every time you pick up the machinery to proliferate, you also pick up the machinery to kill yourself?’
If that were the case, he reasoned, there must be something that tells the cells whether to live or die, and here he found a clue in the growth medium he was using for his experiments. Most of the time, he used serum – the colourless liquid the body produces at the site of a wound that makes it ‘weep’ – because serum contains substances that promote clotting of blood, and survival, growth and proliferation of cells to help in the recovery and regeneration of injured tissue. Myc was killing cells when Evan removed the serum with all its life-enhancing properties and put the cells in a medium that was more like what they would find under normal conditions in the body.
‘Myc turned out to be, I think, the first example of what we now know as a generic feature of how growth control is orchestrated within our cells – which is that everything that makes a cell proliferate (and is potentially therefore a cancer risk if it gets mutated and stuck in the “on” position) comes with something that also suppresses the expansion and growth of those cells.’
Similar experiments with other oncogenes showed that they too shut down growth programmes one way or another after a short spurt of proliferation. Ras, for example, does it by permanently arresting, but not killing, the cell – putting it in a state known as ‘replicative senescence’, where it stops dividing but stays alive and active. But this raised a number of further questions. Oncogenes like Myc and Ras, when not mutated, have regular work to do in promoting growth in cells, but if they also serve to shut down or kill cells after a while, how is new tissue ever produced? ‘The answer seems to be that if a cell switches on Myc in response to a growth signal and starts to replicate, if that cell is in the right place in the body, and it stays in its little niche and doesn’t spill out like a cancer, then it will get all the goodies that tell it not to commit suicide, okay?
‘So cell replication is an obligatorily social enterprise. Cells are not autonomous. By taking them out and putting them in a bottle and adding all the things that would stop them dying, we just completely ignored this fundamental piece of biology. It had always been ignored! Now, the notion that things that drive cell growth also drive cell death and growth arrest is, I think, completely embedded in the understanding of molecular biology; it’s just generally accepted that this is how things work. But at the time, people literally walked out of my talks!’
In fact, Scott Lowe, then doing his PhD at MIT, and his supervisor Earl Ruley, had observed the same extraordinary phenomenon – oncogenes killing cells or condemning them to replicative senescence. And they too had had a tough time getting people to listen. ‘If you’d walk down the hall at MIT Cancer Center and say, “I have an oncogene and it kills cells”, they’d think you were crazy. Because that’s not what oncogenes do; they make cells grow better,’ laughs Lowe.
The insights he and Evan gained in this work also helped to explain a long-standing mystery: why oncogenes are only able to generate tumours in co-operation with one another. Evan believes that when, for example, you put Myc and Ras together, Myc overcomes the replicative senescence programme of Ras, and Ras overcomes the apoptosis programme of Myc. Thus singly, the growth spurt fuelled by either oncogene soon fizzles out; together, all hell breaks loose. In time, he and Lowe would discover that the effects they had both witnessed independently and wondered about – death among their oncogene-driven cells – were caused by the oncogenes switching on tumour suppressors, frequently p53.
The multiple experiments with mouse models – knocking out p53 altogether, or else toggling the gene back and forth between active and passive – made it very clear that this is an extremely powerful protein. As an arbiter of life-and-death decisions within our cells it must be under strong control. So how does it work
?
CHAPTER THIRTEEN
The Guardian’s Gatekeeper
In which we learn: a) that the enormously powerful p53, which can arrest or kill cells, is kept on a tight leash by a protein called Mdm2, which sticks to the p53 protein in cells and marks it up for degradation; and b) that Mdm2 releases p53 from its deadly embrace only when the tumour suppressor is needed to respond to stress signals.
***
The reason that cancer research is such a compelling area to be in is that, in order to understand how things go wrong in cancer, you first have to understand how things go right almost all the rest of the time.
Gerard Evan
‘Guardian of the genome’, the epithet David Lane gave to p53 in 1992, caught the popular imagination and has helped give this extraordinary gene with the eminently forgettable name a public profile in the media. But in the 20 years since that phrase was coined, the list of stresses to which p53 responds has expanded way beyond simple damage to the DNA. Arnie Levine suggests it might be more appropriate to think of the gene as a ‘fidelity factor’ – something that ensures faithful copying of the DNA during cell division. We now know that the gene responds to heat shock and cold shock (when a cell is subjected to temperatures above or below the ideal body temperature), to lack of oxygen or glucose, to certain poisons, to natural ageing and to oncogene activity – all things that threaten the fidelity of the DNA, without actually breaking it, as the cells divide.
‘Ever since its evolution in invertebrates, p53 has been a fidelity factor,’ says Levine. Starve a worm of glucose and it won’t produce eggs; radiate a fruit fly and it won’t produce sperm or eggs until progenitor cells with healthy genomes are restored. With the evolution of the vertebrates – organisms with much more complex bodies, including us – the principle of protecting the fidelity of the germ cells, the sperm and eggs that give rise to offspring, was also applied to all other cells of the body for the first time. ‘And that’s where p53 comes in,’ says Levine. ‘It responds to stress, and it kills. It enforces fidelity by death! So it has a very interesting evolutionary history.’
Very recently scientists have discovered another fascinating aspect of p53’s role in fidelity assurance. Imagine for a minute what would happen if, in the normal course of events, our biological clocks could go backwards in time; if our mature cells could revert to their original undifferentiated state as stem cells, complete with the potential to develop afresh into something new. It’s a nightmare scenario in which your liver cells might morph spontaneously into bone cells, gut into teeth, blood into kidney, and no bodies would be stable. It is p53’s job to ensure that such de-differentiation doesn’t happen; that biological time moves inexorably forward and that our bodily development cannot unravel (except, that is, in the deranged environment of cancer). Scientists creating what are known as ‘induced pluripotent stem cells’ (IPSs) – stem cells with the potential to become any kind of specialised cells that have been engineered in the lab from already differentiated body cells – are frustrating a fundamental law of nature, and they must overcome p53’s defences to do so.
Many people have been involved in uncovering the mechanism controlling this powerful gene, and once again Moshe Oren’s temperature-sensitive mutants provided vital insights. This was the early 1990s, before the creation of knock-out mice had become a cottage industry. Oren and his team were using his temperature-sensitive mutant p53 as a tool in cell cultures to ask a simple question: what is different in cells with active p53 compared with those in which the gene is inactive? They soon observed that a protein appeared hitched to the p53 protein whenever the tumour suppressor was behaving like wild-type p53, at 33°C (91°F), but never at the higher temperature, when it behaved like a mutant.
Levine had observed the same thing in a different set of experiments designed to explore the workings of wild-type p53, and he had identified the hitchhiking protein as Mdm2, already known to cancer researchers as a possible oncogene. Levine had also discovered that by attaching itself to p53, Mdm2 restrained the tumour suppressor – more like a policeman handcuffed to a criminal suspect than a hitchhiker. Tinkering with the temperature in their Petri dishes, Oren’s team discovered that Mdm2 protein appeared only in the cells with normal p53, and was absent altogether from the mutant cells. What did this mean?
Oren and company soon realised they had in their hands a crucial piece of the jigsaw that would reveal the control mechanism of p53. Timing was all. ‘This was 1993, and a year earlier we wouldn’t have known what to do except to say that this was very interesting,’ he commented. But just the year before, Carol Prives and Bert Vogelstein had shown that p53 is a ‘transcription factor’ whose job is to switch other genes on and off; Vogelstein had identified the first of its ‘downstream’ targets and it seemed reasonable to suggest that Mdm2 was another – that it depended on wild-type p53 to switch it on. Subsequent experiments proved their hypothesis right, as Oren explained: ‘Knowing what Mdm2 does to p53 in the way of acting as an inhibitor, which was Arnie’s discovery, and seeing that p53 itself is activating Mdm2, we discovered this “feedback loop” – which luckily turned out to be not just interesting but extremely important, because this is the main way in which p53 is regulated in the cells. This Mdm2 feedback loop is perhaps the heart of the p53 regulatory network.’
In fact, the feedback loop – in which the inhibitory protein is switched on by the protein it then inhibits – did not quite make sense yet. There were still two vital pieces of the puzzle missing, and Oren’s lab was again at the forefront of the search. They had started to use Mdm2’s ability to restrain p53 as a tool much like the temperature-sensitive mutant: by adding or removing the gatekeeper from cells in which p53 was present, they could switch the protein between active and inactive states. ‘We were using it routinely to try to get insights into what p53 does, which is what doesn’t happen when Mdm2 inhibits it,’ explained Oren. However, the researchers soon started to notice something puzzling: that whenever they put the two proteins together, instead of Mdm2 simply attaching itself to p53 and stopping it in its tracks, as they had expected, the p53 protein seemed to disappear altogether. Initially the scientists didn’t trust their experiments, so they repeated them with slight modifications and extra care. But when they came up with the same results again and again, they knew they were real: Mdm2 was destroying p53. And it was doing so, they discovered later, by delivering the ‘kiss of death’ – attaching a little chemical tag to the p53 protein that marked it out for collection and degradation by the cell’s recycling machine, the proteasome.
So, the picture Oren’s team had built up thus far showed that p53 switches on Mdm2, which in turn marks p53 up for degradation in an endless cycle, lasting around 5–20 minutes, that keeps p53 protein in our cells at almost undetectable levels most of the time. (This incidentally helped to explain why there were such high levels of the protein in cells in those early experiments with the mutant clones: mutation typically severs the bond between p53 and its controller Mdm2). But how then does the tumour suppressor escape its gatekeeper in order to protect us from cancer in the normal course of events? This was the final missing jigsaw piece, and it was discovered tucked away in a corner of Carol Prives’ lab in Columbia, where a ‘fantastically talented’ young postdoc called Sheau-Yann Shieh from Taiwan was interested in a process called phosphorylation.
Phosphorylation is one of the most important mechanisms by which proteins are activated and silenced in cells, and it works by attaching small phosphate molecules, as ‘tags’, to some of the amino acids that make up the protein. Shieh was focusing on p53 and asking the question: does the protein get phosphorylated? If so, how does this affect its function? She found that p53 does indeed become phosphorylated; it changes shape and, incidentally – for she was not paying special attention to the Mdm2 story, which was still evolving – that this weakens the bond between the p53 and its minder.
‘But we were missing a critical link,’ explained Prives. The p
ressing questions that remained were: what is the trigger for p53 to become phosphorylated? Does this happen in real life? (So far they had seen it only in cell cultures in the lab.) And is it related to DNA damage and other stressful events in the cells? With the help of some fancy new tools for singling out phosphorylated proteins, they found the answer to all three questions. They discovered that phosphorylation of p53 does happen in real life. It is indeed related to DNA damage and the stress response. And it is the mechanism by which ATM – the gene that makes patients with ataxia telangiectasia so acutely sensitive to radiation – signals distress and triggers the response from p53.
This piece of the jigsaw completed the picture of the p53/Mdm2 feedback loop and suggested how it might work. In essence, this is how it looked: in the normal course of events, p53 protein is being made in our cells all the time so that it can give a hair-trigger response to danger signals; and it is being cleared away almost as fast as it is made by Mdm2 lest it lead to unnecessary death of cells. But danger or stress activates genes such as ATM, which phosphorylate p53, preventing Mdm2 from getting a proper hold and allowing the p53 protein to accumulate in the cells and perform its job of orchestrating the response – arresting the cell temporarily and sending in the repair team; condemning it to permanent arrest or senescence; or forcing it to commit suicide.
Another researcher, Gigi Lozano, working with mouse models at MD Anderson Cancer Center in Houston, confirmed just how important each protein is to the normal functioning of the other in real life when she created mice with Mdm2 knocked out, and discovered that this was biologically lethal: with no holds barred, p53 caused massive apoptosis. But when Lozano created mice with both p53 and Mdm2 knocked out, they could survive until, eventually, they developed cancer.
Exactly how the stress-response mechanism is turned off again when it is no longer needed, and how the tight regulation of p53 is restored, no one is completely sure as yet. But it could be that when the stress signals stop coming, any newly produced p53 will not be protected from degradation by phosphorylation, and the protein will resume its normal dance of death with its minder, Mdm2. This remains an important question since the feedback loop offers tantalising opportunities for tinkering with the regulation of p53 to create new cancer treatments. It’s a topic I will return to in a later chapter.
p53 Page 14