原文地址

Variety is the key to survival in a changeable world – and evolution may have come up with an extraordinary way of generating more variety

A man walks into a bar. “I have a new way of looking at evolution,” he announces. “Do you have something I could write it down on?” The barman produces a piece of paper and a pen without so much as a smile. But then, the man wasn’t joking.

The man in question is Andrew Feinberg, a leading geneticist at Johns Hopkins University in Baltimore; the bar is The Hung, Drawn and Quartered, a pub within the shadow of the Tower of London; and what’s written on the piece of paper could fundamentally alter the way we think about epigenetics, evolution and common diseases.

Before setting foot in the pub, Feinberg had taken a turn on the London Eye, climbed Big Ben and wandered into Westminster Abbey. There, as you might expect, he sought out the resting place of Isaac Newton and Charles Darwin. He was struck by the contrast between the lavish marble sculpture of a youthful Newton, reclining regally beneath a gold-leafed globe, and Darwin’s minimalist floor stone.

As he looked round, Feinberg’s eyes came to rest on a nearby plaque commemorating physicist Paul Dirac. This set him thinking about quantum theory and evolution, which led him to the idea that epigenetic changes – heritable changes that don’t involve modifications to DNA sequences – might inject a Heisenberg-like uncertainty into the expression of genes, which would boost the chances of species surviving. That, more or less, is what he wrote on the piece of paper.

Put simply, Feinberg’s idea is that life has a kind of built-in randomness generator which allows it to hedge its bets. For example, a characteristic such as piling on the fat could be very successful when famine is frequent, but a drawback in times of plenty. If the good times last for many generations, however, natural selection could eliminate the gene variant for piling on fat from a population. Then, when famine does eventually come, the population could be wiped out.

Life’s built-in randomness generator allows evolution to hedge its bets

But if there is some uncertainty about the effect of genes, some individuals might still pile on the fat, even though they have the same genes as everyone else. Such individuals might die young in good times, but if famine strikes they might be the only ones to survive. In an uncertain world, uncertainty could be crucial for the long-term survival of populations.

The implications of this idea are profound. We already know there is a genetic lottery – every fertilised human egg contains hundreds of new mutations. Most of these have no effect whatsoever, but a few can be beneficial or harmful. If Feinberg is right, there is also an epigenetic lottery: some people are more (or less) likely to develop cancer, drop dead of a heart attack or suffer from mental health problems than others with exactly the same DNA.

To grasp the significance of Feinberg’s idea, we have to briefly rewind to the early 19th century, when the French zoologist Jean-Baptiste Lamarck articulated the idea - already commonly held – that “acquired characteristics” can be passed from parent to offspring. If a giraffe kept trying to stretch to reach leaves, he believed, its neck would get longer, and its offspring would inherit longer necks.

Darwin the Lamarckist

Contrary to what many texts claim, Darwin believed something similar, that the conditions an organism experiences can lead to modifications that are inherited. According to Darwin’s hypothesis of pangenesis, these acquired changes could be harmful as well as beneficial – such as sons getting gout because their fathers drank too much. Natural selection would favour the beneficial and weed out the harmful. In fact, Darwin believed acquired changes provided the variation essential for evolution by natural selection.

Pangenesis was never accepted, not even during Darwin’s lifetime. In the 20th century it became clear that DNA is the basis of inheritance, and that mutations that alter DNA sequences are the source of the variation on which natural selection acts. Environmental factors such as radiation can cause mutations that are passed down to offspring, but their effect is random. Biologists rejected the idea that adaptations acquired during the life of an organism can be passed down.

Even during the last century, though, examples kept cropping up of traits passed down in a way that did not fit with the idea that inheritance was all about DNA. When pregnant rats are injected with the fungicide vinclozolin, for instance, the fertility of their male descendants is lowered for at least two generations, even though the fungicide does not alter the males’ DNA.

No one now doubts that environmental factors can produce changes in the offspring of animals even when there is no change in DNA. Many different epigenetic mechanisms have been discovered, from the addition of temporary “tags” to DNA or the proteins around which DNA is wrapped, to the presence of certain molecules in sperm or eggs.

What provokes fierce argument is the role that epigenetic changes play in evolution. A few biologists, most prominently Eva Jablonka of Tel Aviv University in Israel, think that inherited epigenetic changes triggered by the environment are adaptations. They describe these changes as “neo-Lamarckian”, and some even claim that such processes necessitate a major rethink of evolutionary theory.

While such views have received a lot of attention, most biologists are far from convinced. They say the trouble with the idea that adaptive changes in parents can be passed down to offspring via epigenetic mechanisms is that, like genetic mutations, most inherited epigenetic changes acquired as a result of environmental factors have random and often harmful effects.

At most, the inheritance of acquired changes could be seen as a source of variation that is then acted on by natural selection - a view much closer to Darwin’s idea of pangenesis than Lamarck’s claim that the intent of an animal could shape the bodies of its offspring. But even this idea is problematic, because it is very rare for acquired changes to last longer than a generation (Annual Review of Genomics and Human Genetics, vol 9, p 233).

While epigenetic changes can be passed down from cell to cell during the lifetime of an organism, they do not normally get passed down to the next generation. “The process of producing germ cells usually wipes out epigenetic marks,” says Feinberg. “You get a clean slate epigenetically.” And if epigenetic marks do not usually last long, it’s hard to see how they can have a significant role in evolution – unless it is not their stability but their instability that counts.

Rather than being another way to code for specific characteristics, as biologists like Jablonka believe, Feinberg’s “new way of looking at evolution” sees epigenetic marks as introducing a degree of randomness into patterns of gene expression. In fluctuating environments, he suggests, lineages able to generate offspring with variable patterns of gene expression are most likely to last the evolutionary course.

Is this “uncertainty hypothesis” right? There is evidence that epigenetic changes, as opposed to genetic mutations or environmental factors, are responsible for a lot of variation in the characteristics of organisms. The marbled crayfish, for instance, shows a surprising variation in coloration, growth, lifespan, behaviour and other traits even when genetically identical animals are reared in identical conditions. And a study last year found substantial epigenetic differences between genetically identical human twins. On the basis of their findings, the researchers speculated that random epigenetic variations are actually “much more important” than environmental factors when it comes to explaining the differences between twins (Nature Genetics, vol 41, p 240).

More evidence comes from the work of Feinberg and his colleague Rafael Irizarry, a biostatistician at the Johns Hopkins Bloomberg School of Public Health in Baltimore, Maryland. One of the main epigenetic mechanisms is the addition of methyl groups (with the chemical formula CH₃) to DNA, and Feinberg and Irizarry have been studying patterns of DNA methylation in mice. “The mice were from the same parents, from the same litter, eating the same food and water and living in the same cage,” Feinberg says.

Stunning finding

Despite this, he and Irizarry were able to identify hundreds of sites across the genome where the methylation patterns within a given tissue differed hugely from one individual to the next. Interestingly, these variable regions appear to be present in humans too (Proceedings of the National Academy of Sciences, vol 107, p 1757). “Methylation can vary across individuals, across cell types, across cells within the same cell type and across time within the same cell,” says Irizarry.

It fell to Irizarry to produce a list of genes associated with each region that could, in theory at least, be affected by the variation in methylation. What he found blew him away. The genes that show a high degree of epigenetic plasticity are very much those that regulate basic development and body plan formation. “It’s a counter-intuitive and stunning thing because you would not expect there to be that kind of variation in these very important patterning genes,” says Feinberg.

The results back the idea that epigenetic changes to DNA might blur the relationship between genotype (an organism’s genetic make-up) and phenotype (its form and behaviour). “It could help explain why there is so much variation in gene expression during development,” says Günter Wagner, an evolutionary biologist at Yale University. But that does not necessarily mean epigenetic changes are adaptive, he says. “There has not been enough work on specifying the conditions under which this kind of mechanism might evolve.”

When he began exploring the idea with Feinberg, Irizarry constructed a computer simulation to help him get his head round it. At first, he modelled what would happen in a fixed environment where being tall is an advantage. “The taller people survive more often, have more children and eventually everyone’s tall,” he says.

Then, he modelled what would happen in a changeable environment where, at different times, it is advantageous to be tall or short. “If you are a tall person that only has tall kids, then your family is going to go extinct.” In the long run, the only winners in this kind of scenario are those that produce offspring of variable height.

This result is not controversial. “We know from theory that goes some way back that mechanisms that induce ‘random’ phenotypic variation may be selected over those that produce a single phenotype,” says Tobias Uller, a developmental biologist at the University of Oxford. But showing that something is theoretically plausible is a long way from showing that the variability in methylation evolved because it boosts survival.

Jerry Coyne, an evolutionary geneticist at the University of Chicago, is blunter. “There is not a shred of evidence that variation in methylation is adaptive, either within or between species,” he says. “I know epigenetics is an interesting phenomenon, but it has been extended willy-nilly to evolution. We’re nowhere near getting to grips with what epigenetics is all about. This might be a part of it, but if it is it’s going to be a small part.”

To Susan Lindquist of the Massachusetts Institute of Technology, however, it is an exciting idea that makes perfect sense. “It’s not just that epigenetics influences traits, but that epigenetics creates greater variance in the traits and that creates greater phenotypic diversity,” she says. And greater phenotypic diversity means a population has a better chance of surviving whatever life throws at it.

Lindquist studies prions, proteins that can not only flip between two states but pass on their state to other prions. While they are best known for causing diseases such as Creutzfeldt-Jakob disease, Lindquist thinks they provide another epigenetic mechanism for evolutionary “bet-hedging”. Take Sup35, a protein involved in the protein-making machinery of cells. In yeast, Sup35 has a tendency to flip into a state in which it clumps together, spontaneously or in response to environmental stress, which in turn can alter the proteins that cells make, Lindquist says. Some of these changes will be harmful, but she and her colleagues have shown that they can allow yeast cells to survive conditions that would normally mean death.

While Jablonka remains convinced that epigenetic marks play an important role in evolution through “neo-Lamarckian” inheritance, she welcomes Feinberg and Irizarry’s work. “It would be worth homing in on species that live in highly changeable environments,” she suggests. “You would expect more methylation, more variability, and inheritance of variability from one generation to the next.”

As surprising as Feinberg’s idea is, it does not challenge the mainstream view of evolution. “It’s straight population genetics,” says Coyne. Favourable mutations will still win out, even if there is a bit of fuzziness in their expression. And if Feinberg is right, what evolution has selected for is not epigenetic traits, but a genetically encoded mechanism for producing epigenetic variation. This might produce variation completely randomly or in response to environmental factors, or both.

Feinberg predicts that if the epigenetic variation produced by this mechanism is involved in disease, it will be most likely found in conditions like obesity and diabetes, where lineages with a mechanism for surviving environmental fluctuation would win out in the evolutionary long run. He, Irizarry and other colleagues recently studied DNA methylation in white blood cells collected in 1991 and 2002 from the same individuals in Iceland. From this, they were able to identify more than 200 variably methylated regions.

To see if these variable regions have something to do with human disease, they looked for a link between methylation density and body mass index. There was a correlation at four of these sites, each of them located either within or near genes known to regulate body mass or diabetes. Feinberg sees this in a positive light. If random epigenetic variation does play a significant role in determining people’s risk of getting common diseases, he says, untangling the causes may simpler than we thought. The key is to combine genetic analyses with epigenetic measurements.

Feinberg is the first to admit that his idea could be wrong. But he’s excited enough to put it to the test. Perhaps, he suggests, it could be the missing link in understanding the relationship between evolution, development and common disease. “It could turn out to be really quite important.”

Henry Nicholls is a science writer based in London. His latest book is The Way of the Panda (Profile, 2010)