In which we hear how researchers came full circle to realise that some mutant versions of p53 do indeed behave like oncogenes, actively driving delinquent cells towards cancer, rather than simply losing their ability to act as tumour suppressors.
***
You say you have to know all these facts – well, clearly the facts, some of them, that you learn are wrong, so if you take them too seriously you won’t discover the truth. You could say that if you become too imbued in the ideas, and talk about them too long, maybe your capacity for ever believing they’re false would be burned out.
James Watson
When, 10 years after it was discovered, normal p53 was found to be a tumour suppressor not an oncogene (or tumour ‘driver’), many people lost interest in the mutants they had been inadvertently studying for so long. Instead, they began to focus their full attention on the wild-type protein, which was a much more exciting prospect. In doing so they chose to ignore what they had observed with the mutants. The few who said hang on a minute, the mutants may indeed be doing something to drive these tumours – something more than simply losing their ability to stop the cells running amok – became lone voices talking to empty rooms. ‘It was kind of a reaction to the fact that these mutant p53 clones had misled the field and caused us to draw the wrong conclusion,’ commented Moshe Oren. ‘They were a sore point in our history, and many people just wanted to forget about them.’
Not Varda Rotter. You will remember meeting her back in Chapter 7 in relation to the revolutionary discoveries that led to the recognition of p53 as a tumour suppressor. In 1979–80, her experiments with the Abelson cancer virus had led to malignant blood cells that had no p53 protein at all – a completely different result from that of most of her colleagues, who were finding an over-abundance of p53 protein in their tumour cells. The virus, Rotter discovered, was disabling the gene by inserting a bit of its own genetic material into p53 so that it could not produce any protein.
Curious to see what effect the loss of p53 had on these malignant blood cells, she injected them into laboratory mice, where she found that they caused small tumours to develop that soon regressed. Next she took some of these same malignant blood cells and, using some technical wizardry, replaced the crippled p53 with a functional copy of a mutant p53 gene – that is, a mutant that was able to produce its protein. She then injected these engineered cells into her mice, and this time she found they produced extremely aggressive tumours that were eventually fatal. She published her findings in Cell in 1984.
This was dramatic stuff that caught her imagination and when, five years later, normal p53 was revealed as a tumour suppressor, not a tumour driver, she was not about to dismiss the aggressive behaviour of the mutant as no longer important. Rotter did not follow the herd nor change the focus of her research and she became the standard bearer of what is known as mutant ‘gain of function’, often shortened to GOF (remember the analogy in Chapter 7 of the car with the jammed accelerator pedal or the failed brakes? GOF is the jammed accelerator). ‘What convinced me was this,’ said Rotter, rummaging through her computer images to show me an iconic slide of a thin slice of tumour tissue in which p53 protein showed up as bright red. Clearly the cells were stuffed with it. ‘When you take almost any tumour from a human and you stain it for p53, this is what you get . . . Did you ever see anything so covered in protein?’ she asked rhetorically. ‘I felt it can’t just be lying there for nothing.’
That is what prompted her to do the experiments with the mice and the engineered malignant blood cells – she had to prove that her hunch was right and that the abundant protein produced by the mutant p53 gene was actively doing something in tumour cells. This makes p53 unusual among tumour suppressors, almost all of which are simply knocked out by mutation, she explained. ‘Unlike other tumour suppressors, p53 has a schizophrenic personality. You have the wild type which is very important: the guardian of the genome that takes care of DNA repair, takes care of genomic fidelity, takes care of everything. But once this is mutated then it becomes a monster.’ It was this characterisation of p53 that her granddaughter sought to portray in her picture of a devil and an angel that Rotter has framed on her office wall.
Among a field of sceptics, one person who also remained curious about the mutants was Rotter’s colleague at the Weizmann Institute, Moshe Oren. It was while he was investigating their activity in dishes in his lab that he stumbled across the temperature-sensitive mutant when the thermostat in one of his incubators started to play up. This, you will recall, led to the discovery that p53 can trigger apoptosis or cell suicide. But it also contributed to another equally critical discovery about the nature of p53 – that the normal, wild-type protein can change its shape, and thereby its behaviour in cells, from a suppressor of growth to a promoter of growth, under certain conditions. Though it’s now thought likely to be a common feature of proteins that control many others in a cell, such flexibility of shape and behaviour in a single protein was virtually unknown at the time it was discovered in p53 and, as we’ll see, it shed light on mysteries way beyond the field of cancer biology.
AN EXTREMELY FLEXIBLE PROTEIN
The person credited with the discovery of the protein’s flexibility is Jo Milner, who developed what is known as ‘the conformational hypothesis’ of p53. Milner, whom Oren describes as a ‘very clever and original’ scientist, traces her fascination with biology to a childhood spent in Bridlington, a seaside town on the north-east coast of England, where she spent long happy days walking the beach, guddling in rock pools and coming home with starfish in her pockets. It was her mother who, bringing up her children alone after World War II, nurtured a sense of curiosity and freedom of spirit in Milner and her siblings. Times were tough, she remembered, ‘but there was never any sense of hardship. We grew up in a tiny home brimming with friends who all seemed to adopt Mum as their own.
‘One of my abiding memories is of looking out of the window from an early-morning train and seeing, across a field, a white sheet being waved from the upper floor of a large house: the train was carrying me to London for an interview at the university; the large house was where my mother worked as a housekeeper; and the sheet was her fond farewell and good luck.’
After gaining a degree in zoology in London, Milner studied for her PhD at Cambridge University. Since then her career has taken her to Harvard, back to England for another 20 years at Cambridge and finally to the University of York, where she was, until very recently, director of the p53 Research Unit in the Department of Biology. I took the train south from my home in Scotland to visit her as she was busy packing up her lab on the brink of retirement and looking forward to following her still-lively curiosity as a scientist without the pressure any longer of leading a team. As we sat in the sun room of her elegant stone house in an old village nestled in farmland outside York, the scent from an ornamental lime tree in a pot hanging delicately in the air, she talked of the steps that led to her discovery of p53’s extraordinary flexibility – one of the rare and thrilling eureka moments in a scientist’s life.
A question that intrigued Milner about biology in general was how cells switch from a dormant, or quiescent, state to a dividing state as tissues grow or repair themselves in the normal course of life. She was investigating this question in healthy white blood cells, which remain happily quiescent in their nutrient-rich culture medium in the lab until stimulated to enter the cycle of division. For her experiments she was using antibiotic drugs as biochemical ‘tools’. One of these antibiotics was a toxin derived from the death-cap mushroom that works by inhibiting the synthesis of essential proteins and cutting off the metabolism in cells so that they grind to a halt and die.
Milner found, however, that if she exposed the cells in her culture only briefly to the toxin, the effect was reversible. She developed a method that enabled her to put the brakes on in her cells just at the moment they were about to enter division and then, by removing the toxin, to release the brakes and let the cells carry on through the cycle. Using this method, she found clear evidence of a gene expressing a protein that appeared briefly in her cells and stimulated them to switch from quiescence to division, before it was degraded and disappeared. ‘Obviously the next question was what that gene might be, and what was the protein involved,’ she commented.
p53 had recently been discovered and, after reading the papers by Lane and Levine, Milner thought it would be interesting, just on the off chance, to check this gene out. She managed to get hold of two different antibodies that had been tailor-made to recognise and flag up p53 when it was present in cells. Using these antibodies she found that p53 was indeed expressed, and that one of the antibodies recognised the protein in the quiescent cell and the other recognised it as the cell began to divide. Neither antibody recognised the protein in both states. Here, it seemed, was a good candidate for the genetic switch. But why was the protein different – so different that it was undetectable by some tailor-made antibodies to p53 – during each stage of the cell cycle? ‘The only thing I could think of,’ said Milner, ‘was that here was a protein that was changing conformation (or structure). So you have one epitope (the face of a molecule to which an antibody attaches) exposed and another hidden away in one conformation.’ She clenched her fist to demonstrate a folded protein showing its face to the antibody. ‘And then when you stimulate it, the conformation changes and the other epitope is exposed.’ She opened her fist slightly to show another face as the protein refolded itself.
This was a revolutionary idea, and her results raised the question of whether this was one protein changing shape or two slightly different versions of the protein produced by the same gene in order to flip the switch in the cell. It’s here that Moshe Oren’s temperature-sensitive mutant comes back into the picture. Just as Milner and her team were asking themselves about the nature of the p53 protein that was throwing the switch in dividing cells, she and her lab technician attended a p53 conference at which they heard Oren speak of his serendipitous findings, and a light went on in Milner’s head. ‘I thought: fantastic! We can check his mutant to see if our conformation idea is right,’ she recalled.
Her idea was to look at how a single blob of protein produced by the mutant p53 folded itself at the different temperatures, to see if this was what dictated its changes in behaviour from a growth suppressor to a growth promoter and back. When she suggested such an experiment to Oren he was doubtful it would work, fearing that the techniques involved, which included a period of incubation on ice, would upset the mutant’s temperature sensitivity. He had not considered the protein’s conformation, and anyway he had other research ideas in mind for his mutant.
However, Milner still believed the experiment was worth a try and, together with her technician, went ahead as soon as she arrived back in Cambridge. ‘It was just beautiful!’ she said with a big smile at the memory. ‘That was a real high point, because for a long, long time we’d been trying to get a handle on conformation, juggling with different conditions to see if we could induce any change. And here we had this mutant that did it perfectly. It was one of the special moments in life . . .’ Its great significance was the revelation that p53 can switch behaviour – from a suppressor to a promoter of growth and back again – without the need for mutation. Both roles, it seems, are part of the protein’s normal repertoire, and this flexibility of form is what makes it possible. Its flexibility is also an explanation for how p53 is able to play such a varied, subtle and central role in cells, both normal and diseased. It has turned out that the concept is much broader than just p53, said Oren. ‘But p53, in my mind, set the paradigm for this duality of function.’
THE ‘DOMINANT-NEGATIVE’ EFFECT
The discovery of p53’s intrinsic shape-shifting nature opened the door to all kinds of experiments designed to help us understand the activity of the gene in more detail. Milner’s next step was to investigate a phenomenon known as the ‘dominant-negative’ effect, in which the behaviour of cells that have a wild type and a mutant copy of the same gene, both active, is dominated by the mutant. This had never been seen with a tumour-suppressor gene: all those discovered thus far conformed to Knudson’s model in which a cell that still has one wild-type allele will function as normal until that allele is knocked out by some event. In other words, the wild type dominates over a mutant.
However, many of the people who remained sceptical of the gain-of-function theory – that mutant p53 produces a protein with new and abnormal functions – suggested that the dominant-negative effect might well be what people had observed with p53 and mistaken for gain of function. After all, they were used to this tumour suppressor breaking the rules. But if this were the case, it was not true gain of function people were seeing at all, but loss of function by an unusual route – by the wild-type p53 being, as it were, overwhelmed by the mutant which crippled its brakes.
No one knew how this might occur, however, and the object of Milner’s next experiments was to try to understand the relationship between different versions of p53 operating in the same cell. Do they work as separate agents, or stick together to form a co-operative unit? To make things clear, she used combinations of mouse and human p53 in her test tube because, being slightly different-sized molecules, the two proteins could be easily followed in her experiments. As with two different-coloured but similar blocks of Lego, you could see how they fitted together, if they did, rather than being confronted with an amorphous blob.
Milner already knew that p53 protein molecules can clump together in groups of two or four to form co-operative units. Now she showed that this assembly was restricted to proteins of the same conformation. Thus she found that mixing the pre-formed proteins in her test tubes gave either suppressor-suppressor complexes or promoter-promoter complexes, but not suppressor-promoter complexes. Clearly, the affinity between the p53 building blocks was determined by their shape.
However, in real life the dominant-negative effect occurs when wild-type and mutant proteins are co-expressed – that is, produced together and simultaneously in the same cell by the two different alleles, or copies, of a single gene. So Milner and her colleague simulated this scenario in the lab, co-expressing the suppressor (wild-type) and promoter (mutant) forms of p53 side by side and simultaneously in the same mixture. This was the acid test, and the results were spectacular: not only did the two co-expressed proteins form a complex, but the only antibody that recognised the new unit was the one tailored to the promoter (mutant) form. The dominant-negative effect people had speculated about was a real possibility, and here was a novel mechanism to explain it – a clear demonstration that one misfolded p53 protein in a co-operative unit of proteins can force the others to change shape in domino-like fashion.
Milner’s research, conducted under artificial conditions in the lab, was proof of principle; no one knew if this is what happens in real life. But her findings, published in Cell in 1991, quickly caught the attention of Stanley Prusiner, a scientist with an equally original mind working in a very different field – that of the so-called ‘spongiform encephalopathies’ that include mad cow disease and its human equivalent, Creutzfeldt-Jakob disease or CJD, as well as scrapie in sheep. His ideas about how these diseases might arise had been widely scorned as heretical and he was looking for just such a mechanism as Milner described to strengthen his case.
In 1972, Prusiner, then working as a neurologist at the University of California, San Francisco, had admitted a female patient to his ward suffering from CJD, which kills nerve cells in the brain, leaving holes that give it the characteristic sponge-like texture. His patient was progressively losing her memory and her ability to perform routine tasks, and Prusiner was told she was dying of a ‘slow virus’ infection. However, in years of research, no one had been able to pin down this slow virus, so-called because of the long incubation period between exposure to the agent and the appearance of symptoms.
‘The amazing properties of the presumed causative “slow virus” captivated my imagination, and I began to think that defining the molecular structure of this elusive agent might be a wonderful research project,’ Prusiner wrote some years later in an autobiographical sketch. His efforts gradually convinced him that he was dealing not with a virus, nor with any other known infectious agent such as a bacterium or a fungus, but with a misfolded protein – and he named his novel pathogen a ‘prion’. But how could a substance with no DNA to carry the instructions of replication transmit a disease? This was the heresy that caused the firestorm when Prusiner published his prion hypothesis in 1982. ‘Virologists were generally incredulous and some investigators working on scrapie and CJD were irate,’ he wrote. ‘The term prion, derived from “protein” and “infectious”, provided a challenge to find the nucleic acid of the putative “scrapie virus”. Should such a nucleic acid be found, then the word prion would disappear!’
Of course no DNA was ever found, and as evidence mounted for his novel theory of infection, the sometimes vicious personal attacks from his critics gradually died down. But still Prusiner needed an explanation of how a misfolded protein might corrupt the normal protein we all have in our brains. ‘Our paper was the very first evidence that such a thing could happen,’ commented Milner. ‘Prusiner was visiting a colleague in Germany at the time and contacted me to arrange a meeting. We met in my office in Cambridge and talked for three hours before I drove him back to the station. It was so exciting to exchange ideas – just lovely!’ Six years later, in 1997, Prusiner was awarded the Nobel Prize for Medicine for his prion hypothesis which, though it still has its critics, is now widely accepted as the explanation for the deadly spongiform brain diseases.
A LIFE OF ITS OWN?
Not all p53 mutations produce so-called ‘conformational’ (or ‘structural’) mutants that behave this way. ‘Contact’ mutants, which produce a protein unable to attach to DNA and switch on other genes, are the type most commonly found in human tumours; in these cases the wild-type protein will win the day, keeping the mutant in check until that good copy of the gene is lost in the course of living. As the unusual variability of p53 mutants became apparent, the debate about ‘loss of function’ versus ‘gain of function’ became ever more intense and, in the late 1990s and early 2000s, a number of research groups created transgenic mice to try to resolve it and to find out what happens in real life.
Guillermina (‘Gigi’) Lozano, whom we met working with mouse models at MD Anderson in Houston in Chapter 13, headed one such group. Lozano’s family had immigrated to the US from Mexico in search of a better life, and Gigi was the first among them to go to college. In 1986 she earned a doctorate in biochemistry from Rutgers University in New Jersey, but as a postdoc she chose to join Arnie Levine’s molecular biology lab at Princeton, attracted by the fact that he was working on one of the very first mouse-tumour models. ‘When I realised you can manipulate the mouse genome to mimic the kinds of tumours you find in human cancer, I was fascinated. There was no going back for me,’ she told me with a grin when I met her at a mutant p53 conference in Toronto in 2013.
Trained by Levine, Lozano got a job in molecular genetics at MD Anderson, where she is now Professor and Head of the Department of Cancer Genetics. Much of her research involves mouse models and in the early 2000s she set about creating one that mimics the human Li-Fraumeni syndrome, in which the p53 gene has one wild-type allele and one allele with a ‘point’ mutation, meaning that just a single letter in its code is changed. People had become adept at creating knock-out mice, with a whole gene or one of the two alleles ‘deleted’ from the DNA, but a knock-in mouse – one with a point mutation – was an altogether trickier proposition that took time, skill and patience. Lozano and her group chose a mutation that corresponds to the R175H mutation in human cancers ‘because it’s the worst mutant you can possibly have,’ she explained. ‘And if you’re going to generate a mouse for the first time you don’t want a mutant that’s kind of wimpy.’
Meanwhile, at MIT in Boston, Tyler Jacks – renowned for creating one of the two first p53 knock-out mice in 1992 – was on the same track. His lab was busy generating two different mouse models that mimicked LFS – one with the same point mutation as Lozano’s mice, corresponding to human R175H, and another corresponding to R273H. The two groups published their findings in the same edition of Cell in December 2004. What distinguished their mouse models from others designed to test the activity of mutant p53 was that here the gene was being switched on naturally in response to signals from the cell’s environment. In most other models, the gene was switched on artificially by the researchers – and herein lay the big sticking point. The sceptics argued that in all experiments that appeared to show gain of function, the gene had been over-stimulated by the researchers, leading to an over-abundance of the protein. The artificial manipulation by researchers was bound to upset the delicate machinery of the cell, they said, and they were not convinced this pooling of protein – and hence gain of function – is what happens in real life.
However, both Lozano’s and Jacks’ mice showed that indeed the mutant protein does pool in real life. They furnished compelling evidence, too, that gain of function – by some p53 mutants at least – is a real phenomenon and not an artefact produced by force. The two groups used different strains of mice in their experiments, which meant that the mutant genes were operating against a variety of background environments, thus adding weight to their findings. As controls for their experiments, both groups used a mouse with one wild-type allele and the other p53 allele missing altogether; these control mice, too, were prone to cancer, but mutation would be no part of the picture, allowing the researchers to see, by comparison with their LFS-like mice, what effect, if any, mutation had on the types of tumours that developed.
So what did they find? As one would expect with living creatures, the different mice met different fates. They all – those with one mutant p53 allele (the LFS-like mice) and those with one missing allele (the controls) – developed tumours. In all Jacks’ mice with mutant p53 the range of tumours that developed was different from that seen in the control mice. In Lozano’s R175H mutants, on the other hand, the tumours that developed were similar to those of their controls, but they were much more aggressive: they spread readily to the lymph nodes, lungs, liver and brains of the mice, while the tumours of the controls did not metastasise.
‘To me that was the most convincing experiment,’ Lozano commented. ‘When you compare those two mice and one has a tendency to produce tumours that metastasise and the other one doesn’t, how can you argue against gain of function? I mean, you can’t!’ What finally clinched the argument for most of the p53 community – including some die-hard sceptics of gain of function – was that both Lozano’s and Jacks’ mice developed some novel tumours that are never seen in knock-out mice with no p53. This could mean only one thing: that the mutant was doing more than simply hobble the wild-type allele and shut down its protective functions – clearly it had a life of its own.
These and other mouse models have allowed researchers gradually to build a picture of how the mutants work and how they interact with wild-type p53. Context, it seems, is all-important. Not only do the mutants differ from one another in their actions, but they behave differently in one cell type, tissue or organ from another, and in one strain of mouse (and presumably one human being) from another. Timing, too, is critical: in some cancers p53 mutation is an early event; in others it occurs when the tumour is already advanced, and may (as in the case of colon cancer) mark the turning point between a benign growth and malignancy.
As for their mechanism of action, it seems that p53 mutants sometimes co-operate with other oncogenes, such as Ras, to drive the growth of tumours. Sometimes they achieve the same effect through interaction with another protein in the cell – notably one or other of p53’s closest relatives, p63 or p73, which share some of its tumour-suppressor characteristics and can be hobbled by the mutant. Some mutants can, like wild-type p53, switch on and orchestrate the activity of other genes. However, this is a travesty of healthy behaviour: the genes switched on by mutant p53 are not the same genes that are controlled by the wild-type tumour suppressor, and can have the opposite effect. And a distinctive characteristic of many of the mutants is that they make cells extremely resistant to self-destruct signals. This not only encourages the growth of tumours, but makes them very difficult to treat, since most anti-cancer therapies are designed to trigger the apoptosis response.
An unexpected finding, made by Gigi Lozano and her group, is that the over-expression of mutant p53 protein is not an intrinsic property of the mutant gene, as had been assumed. The pooling of the protein occurs only in tumour cells, while in the normal cells surrounding the tumour and beyond, the protein is at barely detectable levels. This implies that, like the wild-type protein, the mutant is being regularly produced and degraded in the normal course of events until something occurs to release it from the loop. Though theories abound, no one yet knows why or how this happens – only that the mutant protein has to be over-expressed to be able to act as a growth promoter.
VINDICATION
As Varda Rotter’s steadfast insistence on the importance of mutant p53 has been vindicated and the spotlight has swung back in this direction, enormous effort is being made to understand its biology. ‘Over the last five years alone,’ wrote Carol Prives and William Freed-Pastor in a review for Cold Spring Harbor Laboratory Press in 2012, ‘p53 mutants have been found to actively contribute to tumor proliferation, survival, limitless replication, somatic cell reprogramming (i.e. turning differentiated body cells back towards stem cells), genomic instability, inflammation, disruption of tissue architecture, migration, invasion, angiogenesis (development of a blood supply to a tumour), and metastasis.’ They concluded that this confirms mutant p53’s central role in the development of malignant tumours, with an impact on nearly all of the ‘hallmarks of cancer’– the list of 10 defining characteristics of all cancers – proposed by Bob Weinberg and Doug Hanahan in 2000.
It also makes the aberrant protein a prime target for therapy, as scientists and Big Pharma alike look for new, more effective ways to treat people with cancer that do not do such devastating damage to the body’s normal, fast-dividing cells at the same time.