Last week, I attended the tail end of a Keystone conference on Epigenetic Regulation of Aging, followed by a one-day brainstorming session to kick off a project called GILGA-mesh, intended to take this bull by the horns. Though the subjects of the two days were virtually identical, the approach and attitudes of the scientists in attendance set very different tones. Both days featured smart, creative and careful scientists, but they saw the same material through different frameworks. Sometimes philosophy makes a difference.
For readers who know me less well, I should introduce my perspective: I believe that aging is an evolved epigenetic program. When we are young and growing, particular genes are turned on and off with exquisite timing to determine the growth and development of bones, muscles, and organs. When we are old, the program continues, more slowly and more diffusely, but inexorably nonetheless. Genes are turned on that destroy us with inflammation and cell senescence and auto-immunity and programmed cell death, while the systems that protect us from pathogens and from free radical damage are gradually shut down. Evolution has left nothing to chance.
[I first wrote an academic paper about this idea in 2013, excited by a paper by Adiv Johnson on methylation, but unaware that Tom Rando had written on the same lines the previous year. Jeff Bowles had hinted at similar ideas in a paper more than a decade earlier. Soon the field was broken wide open by the work of a bio-statistician.
Steve Horvath ran a computer analysis on thousands of genes as they are expressed in young and old humans, and produced an “epigenetic clock” that could accurately report how old a person using measurementis of methylation in 353 DNA sites in particular.]
Epigenetics is a new science in the 21st century. All the cells in one body have the same DNA (pretty much), but differernt genes are “expressed” (translated into proteins) in different tissues and at different times, and this is what controls the body’s metabolism. In fact, only 2% of our DNA is genes, and 98% determines how the DNA is folded and spooled, opened and closed at particular times and places, and this in turn controls gene expression. We are 2% genetic and 98% epigenetic.
There is a language called the “genetic code” which determines how genes are translated into proteins. It was decoded by Francis Crick and others in the 1950s. It is as simple as it can be, and is completely understood. There is another language, the “epigenetic code” that determines gene expression. It is anything-but-simple, with a convoluted and self-referential syntax that we are just beginning to understand. The epigenetic code starts with signals embedded in the DNA that serve as “start” and “stop” codons. The stretch in between comprises a piece of a gene, a kind of Gutenberg movable type that is transcribed from the chromosome and then spliced and combined to form functional RNAs and proteins. The complicated part of the epigenetic code is implemented as a pattern of methyl and acetyl groups. These are little chemical decorations that attach to the DNA and to the “histones” (spools around which DNA is wound up in the cell nucleus for safe storage). The methyl and acetyl groups are continually being attached and removed according to instructions that come from within the cell and other instructions that are passed through the blood. It is the methyl and acetyl groups that determine how the DNA is folded and spooled, which effectively turns particular genes on and off as needed.
The part of the epigenetic code on which we have the best handle at present is called “methylation of CpG islands”. Long stretches of DNA have CGCGCGCG… on one strand, complemented by GCGCGCGC… on the other. Often the C’s in this region get an extra methyl group, turning from cytosine to 5-methylcytosine. Then this stretch becomes a “repressor region,” a signal to NOT express the adjacent gene.
DNA methylation can be persistent, turning a gene off for decades at a time. When a cell divides and its DNA is copied, the methylation pattern can be copied with it. This accounts for some of the persistence of epigenetics, and the way gene expression can be inherited across generations.
DNA methylation has been appreciated for 30 years, but two recent developments make the subject attractive and accessible to research. (1) There is now a simple lab/computer technique for reading the methylation pattern from DNA. It relies on commercially available, automated machinery for PCR to sequence a full genome before and after chemical modification of the methylated C’s. (2) There is now a simple lab/computer technique for changing the methylation state of any chosen target site in the DNA. It is based on CRISPR technology that is taking genetics labs by storm the last two years.
Epigenetics and aging
Three years ago, Horvath demonstrated that there are specific patterns of methylation associated with particular ages of the body. It’s not just that the fresh, clear pattern of youthful gene expression becomes muddied and random with age—although there is some of that. But it’s also true that some genes that are active in youth become inactive as we get older and (especially) that other genes that were suppressed in youth become activated in old age. What Horvath’s paper says is, “show me methylation pattern of a person’s cells, and I can tell you how old s/he is.”
Is epigenetics a cause or effect of aging?
The correlation between aging and epigenetic status is established beyond dispute. But what does it mean? This is the big question. Most researchers think of the body as programmed by evolution to be as strong and healthy as possible. So, when different genes are expressed in old age, they find it natural to assume that the body is protecting itself in response to damage that it has suffered over the years. We express different genes when we are older because we need different genes when we are older. This was the predominant attitude at the first conference (where I was present just for the last day).
The other possible interpretation is my own, and it has become common among those who are closest to the field of epigenetics. It is that epigenetic changes with age are means of self-destruction. The body is programmed to die, and its suicide plan is laid out in the form of transcribing an unhealthy combination of genes. This idea flies in the face of traditional evolutionary theory. (How could natural selection prefer a genome that destroys itself and cuts off its own reproduction?) Nevertheless, the evidence for this hypothesis is robust. The genes that are turned on don’t protect the body—quite the opposite. Genes for inflammation are dialed up. Genes for the body’s defense against free radicals are dialed down. Cell turnover is dialed down. DNA repair is dialed down. The mechanisms of programmed cell death (apoptosis) are strengthened in healthy cells, at the same time that they are perversely weakened in cells that are a threat to the body, like infected cells and cancer cells.
How will we determine who is right?
In my opinion, the existing evidence heavily favors the hypothesis that aging is caused by epigenetic changes, rather than the other way around. When we look at the kinds of changes that occur, they seem to be pouring fuel on the fire, not putting it out. Protective genes are turned off and inflammatory genes are turned up. I also think that
parabiosis experiments provide a strong clue. Three researcher groups (at Stanford Harvard, Berkeley) have shown that injecting blood plasma from a young mouse into an old mouse makes the old mouse healthier, and relieves some problems associated with age. The blood plasma contains no cells—only signal molecules that are the product of gene expression. This is powerful evidence that youthful gene expression is supporting a strong and youthful body, and (conversely) that the kind of gene expression that characterizes old age is not doing the body any good.
But the ultimate experiment will be to re-program gene expression in an old mouse and see if there is a rejuvenating effect.
As of now, the GILGA-Mesh project is dominated by numbers geeks (like me) who practice the “Google approach” to bioinformatics. Huge databases of gene expression are screened for epigenetic candidates that seem to be well-correlated with good outcomes. I think what we need is an infusion of biolochemists who understand something about the body’s signaling networks, and can orient us toward “upstream” and “downstream” molecules. Here’s my proposed program:
- Repeat Horvath’s (human) analysis for mice. In other words, identify several hundred places where methylation is different in young and old mice.
- Determine which genes are associated with these regions. (Map needed for this should already be available.)
- Look at the set of genes and identify transcription factors. These are likely to be “upstream”, in that they control other genes.
- Start with old mice. Use CRISPR to change the methylation status in a handful of promoter regions that control transcription factors, making them match the methylation status of young mice.
- Measure metabolic functions to see if the old mice are more healthy or less after these procedures. Look particularly for changes in inflammation, propensity for cancer, and especially life span.
If this experiment goes as I expect, we will be ready for rejuvenation experiments in humans.
How does the body know how old it is?
Even further upstream, is there a central master clock that dictates the body’s epigenetic expression, and thereby determines our biological age? Logically, it seems that the body would need an accurate clock to time the events of growth and development. Evolution likes to re-use the parts she has created, and it would not surprise me if the developmental clock morphs into an aging clock.
I have reasoned that there are two possibilities. It may be that there is a timekeeper, probably in the neuro-endocrine regions of the brain, that controls the processes of development and aging. This possibility is supported by works of Kasper Daniel Hansen and Claudia Cavadas. If this pans out, it would present the handiest target for true rejuvenation in humans. But it also may be that epigenetic expression itself is a kind of clock that is diffused through the body. Today’s gene expression includes transcription factors that control tomorrow’s gene expression, and so epigenetic state may be a feedback loop, or self-contained clock. This may also be a target for rejuvenation, but a little accessible, harder to address or to tinker with.
Random notes—other things I learned last week
I was tickled to find how many members of the GILGA-mesh team already support the
programmed aging perspective that I have advocated. I was particularly gratified to receive encouragement from Caleb Finch, a grand old man of the field who wrote the
encyclopedia of aging in 1990, and continues a very active research program today.
From Finch, I learned that infections in childhood and even in the womb can have a serious effect on diseases of old age, decades after the fact. He hypothesizes a lifelong burden of inflammation. Evidence includes an elevated incidence of heart disease for the cohort born just after the influenza epidemic of 1918.
I was chagrined to learn that air pollution, especially particulate matter, is associated with increased risk of dementia. This poses a personal dilemma for me, as I plan to spend the summer at the lab of Meng-qiu Dong in Beijing.
I learned that hospital errors are the third leading cause of death in the US, accounting for about 10% of all deaths, about the same number as smoking. Maybe you already read that in the New York Times.