Epigenetics and the Direction of Anti-Aging Science

Dear Readers –

It’s been a deeply gratifying year for me.  Twenty years ago, I first started writing that aging is something the body does to itself, a body function, rather than deterioration or loss of function.  Journals would not even send my submission out for peer review.  Journal of Theoretical Biology sent me the considered opinion of their anonymous reviewer, “JTB shouldn’t touch this topic with a ten-foot pole.”  The conflict with prevailling evolutionary theory was just too deep.

But in the interim, the evidence has continued to pile up, and many medical researchers have taken the message to heart in a practical way, setting aside the evolutionary question and just pursuing approaches that seem to work.  The most promising developments in anti-aging medicine involve changing the signaling environment rather than trying to “fix what goes wrong” with the body.

My popular book exploring the evolutionary origins of aging (and implications for medical science) came out in June, and an academic version of the same content came out in October.  Gandhi taught me,

“First they ignore you…
then they laugh at you…
then they fight you…
then you win.”

The paradigm of programmed aging passed this year from stage 2 to stage 3, with prominent articles arguing against the possibility of programmed aging [Kirkwood, de Grey, Vijg & Kennedy].  Current Aging Sciences devoted a full issue to the question.  I welcome the discussion.  This is a debate that colleagues and I have sought to initiate for many years.  There are powerful theoretical arguments on one side, and diverse empirical observations on the other.  The scientific community will eventually opt for empiricism, but not until theory digs in its heels and fights to the death.  A basic principle of evolution is at stake, and many theorists will rise to defend the basis of their life work; but a re-evaluation of basic evolutionary theory is long overdue.

The idea that fitness consists in reproducing as fast as possible is no longer tenable.  For plants, this may be approximately true.  But animal populations cannot afford to reproduce at a pace faster than the base of their food chain can support.  Animals that exploit their food supply unsustainably will starve their own children, and there is no evolutionary future in that.  This is a principle that links together entire ecologies, and the foundation of evolutionary theory will have to be rewritten to take it into account.


The Programmed Aging Paradigm: Is it a Cause for Optimism in the Quest for Extended Human Life?

For many years, I put forward the argument that programmed aging means there are genes that serve no other purpose than to hasten our death, and that medical research should be targeting the products of those genes.  (Once a pathway has been identified, pharmaceutical science knows well how to block it with engineered drugs, like beta blockers and MAO inhibitors and serotonin re-uptake inhibitors.)

But in recent years, epigenetics has eclipsed genetics as the major theme in molecular biology.  Everything that happens in the body is determined by which genes are expressed where and when.  The vast majority of our DNA is devoted not to coding of proteins but to promoter and repressor regions that control gene expression with exquisite subtlety.

There has been a growing recognition of aging as an epigenetic program.  As we get older, genes that protect us are dialed down, and genes for inflammation and apoptosis are dialed up so high that healthy tissue is being destroyed.  Many epigenetic scientists have discovered this, and they find it natural to see aging as a programmed phenomenon.  (Most have never been to graduate courses in evolutionary science, where they would have been indoctrinated into the perspective of the selfish gene.)

At first blush, it seems that an epigenetic program is just as amenable to pharmaceutical intervention as a genetic program.  A few years ago, I wrote about transcription factors as the key to aging.  Transcription factors bind to DNA and turn whole suites of genes on and off in a coordinated way.  If we can restore a youthful transcription environment in an older animal or human, the body knows how to repair damage and re-create a younger self.  The model I had in mind was based on a small number of powerful transcription factors controlling expression of a much larger number of genes, controlling a yet larger population of metabolites.

I no longer believe in this model.

This summer, I had a chance to work in a worm genetics lab and consult closely with people who know the experimental details.  I learned that there is no clear line between functional proteins and transcription factors, that many proteins have multiple functions, and that metabolites feed back to control gene expression.  Instead of a pyramid, I should think of a web of causes and effects.

The entire system is primed for homeostasis, meaning that it responds to any intervention, seeking to move the system back toward its previous state.  Of course, it is this capacity for recovery that makes life robust; but it also means that there is no “command central” which can be tweaked to move the whole system at once toward a desired state.

I still believe that there are one or more aging clocks that inform the body of an age-appropriate metabolic state, and synchronize the aging of different systems.  Telomere length is one such clock.  If we can reset an aging clock, the body will repair and clean itself up.  If we can reset several clocks, the body may be able to restore itself to a younger state.  But I recognize the possibility that the clock is diffused through the detailed epigenetic status of a trillion cells, and may be beyond the reach of foreseeable technology.

A few years ago, Steve Horvath of UCLA gave us the most explicit picture of an aging clock.  He found it in one particular epigenetic marker: the pattern of DNA methylation.  The Horvath clock is distributed over 353 sites.  This is a clock that can be read for any given cell, but can it be changed or adjusted in each of the body’s trillions of cells?  The prospect is daunting, and we do not yet know how to go about it.  2016 saw the first use of CRISPR technology for therapy in a living human.  CRISPR could, in principle, be used to alter gene expression, one gene at a time.  To re-regulate the expression of 353 genes in one treatment would probably require new concepts and new discoveries.

Short of resetting the aging clock, there are several technologies just over the horizon that should offer substantial life extension benefit.  I believe the best prospects are senolytics (ridding the body of senescent cells), telomerase activators (rejuvenating old stem cells), and adjusting blood levels of key hormones and cytokines that increase or decrease with age.


The Bottom Line

Programmed aging is the right model to keep in mind as we search for interventions that slow the aging process and lessen the incidence of heart disease, cancer and Alzheimer’s all at once.  Prospects are good for real breakthroughs, perhaps as soon as 2017, but I am no longer as optimistic as I was just a few years ago that complete rejuvenation is on the horizon.

Wishing you health and vitality for the coming year,




– Josh Mitteldorf

From Santa Diego, a Jolla Xmas Gift

From the Salk Institute in La Jolla, CA came an announcement last week that the four factors previously identified to turn ordinary cells into stem cells (in cell cultures) was successfully used as a rejuvenation procedure in live mice.  The results provide important new evidence for the hypothesis that aging is under epigenetic control, and a proof of principle that we might slow aging by modifying the chromosome markers and attachments that determine gene expression.  But the way in which this was done involved genetic engineering before birth, and there is no obvious way to translate the results quickly into an anti-aging teatment for living humans.

Ten years ago, Shinya Yamanaka’s Kyoto University laboratory announced that just four proteins could turn an ordinary, differentiated cell back into the stem cell from whence it came.  The four were transcription factors, high level switches that turn whole systems of genes on and off with one signal, and the “Yamanaka factors” became known by their initials, OSKM.

Last week, Izpisua Delmonte’s laboratory at the Salk Institute announced a success in rejuvenating whole animals, live mice, using the same OSKM.

Whether this is the germ of a potent new rejuvenation treatment remains to be seen; but the immediate message is a dramatic affirmation of the new paradigm in anti-aging medicine: aging can be reversed by signaling, without artificially-engineered repair of damage.  A bold form of this paradigm is the epigenetic hypothesis—now just 4 years old—which says that aging is controlled by gene expression.  It is the set of genes that are turned on and off, and the genes’ levels of expression that determine the state of the body’s age.  (This idea does not deny that tissues and biomolecules suffer damage with age; but the hypothesis says that the body knew how to repair this damage at one time, and is capable of repairing the damage again, when the signal molecules engage repair mechanisms appropriately for a young individual.)

Gene expression  is controlled, in turn, by markers on the chromosome and on the histone spool around which the DNA is wrapped like thread on a spool.  These markers are known to change with age in a characteristic pattern.  Shifting epigenetic markers program all the stages of development, and (according to the hypothesis), the program continues, and causes the body to pass through the stages of aging.

But in nature this clock never goes backward.  The epigenetic clock is reset to zero as the genome is wiped clean and reprogrammed.  This happens twice: once in the creation of germ cells, the sperm and egg; and a second time after sperm and egg join to make a zygote.  (This is a simplification; some traits are epigenetically inherited, implying that some genome markers are retained and pass across generations.)

If we buy the epigenetic hypothesis, then the holy grail of anti-aging medicine would be to reset all the epigenetic markers, say from age 60 to age 20, but not all the way back to zero.  This must be “possible” in some sense of the word; but if it depends on us to read (for example) the methylation of a 20-year-old’s chromosomes and write the results onto the chromosomes of a 60-year-old, in every cell of the body, without otherwise disrupting the living organism, then the task is daunting.  And methylation is just one of about 100 known epigenetic modifications of the chromatin.

Nature knows how to reset the epigenetic clock all the way to age 0, but there is no precedent for a partial reset.  All tissue differentiation is lost, and the growth rate is pedal-to-the-metal high.  It should be no surprise that previous attempts to rejuvenate the living mouse by resetting the clock with OSKM have led to cancer disasters [ref, ref].  In this new report, the Salk researchers used short, intermittent exposure to OSKM to “partially reset” the epigenome.  If this really works, and if the epigenetic hypothesis continues to pan out, then this is indeed a week to celebrate.


The experiment

The procedure was first tested in human cell cultures, demonstrating “partial de-differentiation”, where functional cells were rejuvenated without returning them completely to stem cell status.  This was an important proof of concept.  The authors cite previous experiments suggesting that OSKM re-setting is a multi-step process, so it is possible in principle to halt the process after partial re-programming, and hope for a somewhat younger state.

Next, the procedure was tested on mice with genetically short life spans, living an average of  just 18 weeks.  Mice with the OSKM treatment lived 24 weeks.  Impressively, the treatment was tuned so that it did not increase cancer rates.  The particular life-shortening gene was defective LMNA (Lamin A).  Lamin A is important for structure and function in the cell nucleus (details hazy), and there is no way to distinguish from this experiment whether the OSKM treatment merely counteracts the deficiency of Lamin A or whether it slows aging generally. Finally, the procedure was tested in normal, aged mice.  They showed signs of improved healing, nerve regrowth, and mitochondrial chemistry typical of younger mice.  But the mice were sacrificed to determine these things, so there was no demonstration of increased lifespan for genetically normal mice.  Sometimes “publish or perish” pressure leads researchers to kill the goose that lays the golden egg…

But the big asterisk on this new result is that the delivery system was through a gene added to the mice at the egg stage.  It is easy to modify genes at the egg stage, when there is just one cell, and then the modification is copied in every cell of the adult; but to do gene therapy for adult humans is still at an early experimental stage.  Every cell in their bodies had an extra copy of each of the 4 genes for the OSKM factors, and these genes were so configured that they could be turned on and off with a drug (doxycycline).  Administration of the drug was arranged to be just right to reprogram the cells, but not all the way.  Optimum was found to be a low dose, just two days a week.


Theoretical hedging

Bench scientists have learned to adopt the epigenetic hypothesis in practice, but some are still bound to the obsolete theoretical ideas, inherited from sclerotic thinking about “selfish genes”.  Belmonte was quoted as saying the epigenome becomes “damaged” late in life “At the end of life there are many marks and it is difficult for the cell to read them.”

But studies of aging epigenomics show that in addition to random changes in the epigenetic state, there are definite, programmed changes—enough to make an accurate epigenetic clock—and that some of these changes turn down cell repair functions and turn up inflammation.  Aging involves a loss of order (“damage”), but it also entails a set of programmed changes.  It is the latter that we may hope to address through a streamlined, signaling approach to anti-aging medicine, and, if we are lucky, the body may take up the ball from here and undo part or all of the damage.


The bottom line

This is an important new confirmation of the epigenetic hypothesis.  Previous confirmations were

  • in parabiosis experiments, but the experiment could not be continued long enough to be sure that lifespan was extended
  • in “methylation clock” measurements, but there has been no way to distinguish whether epigenetic changes were a cause or a result of aging.

This new experiment shows that epigenetic changes can extend lifespan.  But the experiment offers no clear extrapolation to life extension for humans.  The treatment depends on four large molecules that need to be delivered to every cell nucleus in the body.  These molecules cannot be taken orally because they will not survive digestion, and even intravenous delivery will not get the molecules to cell nuclei where they are needed.

In cell cultures with the OSKM factors applied externally, stem cell yields are still just a few percent, even after ten years of experience.

The Salk researchers got around this by inserting the OSKM genes into the cell, but for already-living humans this is not a possibility.  The best we know how to do is to modify some of the body’s cells with gene therapy; CRISPR in living humans is itself a new technique in its experimental phase.

So for the foreseeable future, I see a two-pronged approach to cell-level rejuvenation.  One is to remove senescent cells (senolytics); and the other is stem cell removal, rejuvenation, multiplication in vitro, and return to the body.  OSKM may be useful in this second step, rejuvenation of stem cells in vivo.

Telomeres—too much of a good thing?

One of the major themes in aging science of the last 15 years has been that there is natural variation in telomere length, and individuals with longer telomeres have lower disease risk and longer life expectancy than those with shorter telomeres.  A paper last week in Nature Structural and Molecular Biology found that stem cell telomeres are actively maintained at a target length, not just by elongation (with telomerase or ALT) when they get too short but by active trimming when they get “too long”.  We know what “too short” means: short telomeres lead to cellular senescence.  Cells with short telomeres are not just falling down on the job; they are toxic.  But what does it mean for telomeres to be “too long”?

The headline in MedicalXpress says “Scientists find that for stem cells to be healthy, telomere length has to be just right”.  The story underneath includes the claim that “really long telomeres caused telomeric fragility, which can lead to initiation of cancer”.  But now that I’ve read the research article on which it is based and some of the references in that article, I see that the part about cancer was tacked onto a (new and interesting) research finding.  I’m convinced that the meme “long telomeres lead to cancer” has been resounding in an academic echo chamber for 25 years; that it never has had an experimental foundation, and its theoretical foundation is just wrong.

What is new and interesting is this:  researchers at Salk Institute have discovered a mechanism for trimming telomeres.  In our previous understanding, telomeres lose length every time a chromosome is copied (every time a new cell is created).  Telomeres are partially rebuilt by the enzyme telomerase, or by a less direct mechanism called ALT.  Telomeres are fully rebuilt only when new life is created, in a germ cell or a fertilized egg.  In the previous understanding, shortening of telomeres is passive, while lengthening is active.  The new study documents an active mechanism for shortening telomeres.

Part of each telomere is unpaired, a single strand of DNA extending past the end of the chromosome, and folded back over the main (double-stranded) part.  Single-stranded DNA normally means a problem, and the cell nucleus has multiple means to repair or degrade it.  To protect the telomere from being attacked (to prevent fixing of what ain’t broke) the telomere is chaperoned by various protective proteins, most famously shelterin.

We have known that stem cells can express telomerase to counteract telomere shortening, though (in humans) there is not enough to keep telomeres from shortening progressively through a lifetime.  The new finding is that when a stem cell detects that telomeres are “too long”, there is a way to trim them back.  A strand of DNA is manufactured that is complementary to the telomere’s repeated sequence TTAGGG.  The complementary strand (that would be AATCCC, repeated)  has an affinity for the telomere repeats, and it finds and binds to a segment of telomere, then circles, “bites its tail”, and breaks off a ringlet of double-stranded telomere-stuff (called a T-circle), effectively shortening the telomere.

What’s wrong with extra-long telomeres?

The obvious question: why is the cell doing this?  What is the danger of telomeres that are too long?  One natural place to look is in the telomere position effect (TPE).  Telomeres fold back over the chromosome in such a way as to silence genes near the ends.  We might expect that the right genes must be silenced at the right times, and that silencing too many genes with an extra-long telomere would cause problems.  My own best guess is that this is the right answer.

Another hypothesis is that extra-long telomeres are inherently unstable and unmanageable.  But the present studies were done with human cell cultures, where telomeres are ~10,000 BP in length; mice commonly have telomeres ten times that long without causing problems.

The conventional hypothesis is that telomeres are trimmed to prevent cancer, and that is the spin put on the findings by the authors of the paper in their press release.  “We were surprised to find that forcing cells to generate really long telomeres caused telomeric fragility, which can lead to initiation of cancer.”  In the paper itself, they were more circumspect about this explanation, as is academically appropriate.  These people are masters at what they do (Since spending time at NIBS in Beijing, I have an expanded awe for the experimental virtuosi who are able to infer reliable data about the inner workings of cells.)  But they are not theorists and they trust the community of biological theorists to supply the theoretical framework for interpreting their result.

In this case, the trust is misplaced.  The idea that telomeres are kept short to prevent cancer was originally proposed by (Nobel laureate) Carol Greider (1990), and has been promoted most explicitly by Judith Campisi (1999) before she became convinced by experimental data that the situation is more complicated, that short telomeres are more likely to cause than to prevent cancer (2013).  Through strength in numbers the cancer/telomere hypothesis has achieved “echo chamber” status–many researchers cite each other’s secondary statements on the subject, until tracing the empirical support for the hypothesis becomes unnecessary.  It is common knowledge.

One of the authorities cited in the original paper is this article which is actually about deletion of a gene for a shelterin-related protein that binds to telomeres.  When this gene is deleted, telomeres become unstable and cancer rates rise.  But the article is not about telomeres that are “too long”.  Another authority for the hypothesis that is cited in the original paper is this book chapter.  The chapter is not about cancer, but it does peripherally cite this study from NCI, which finds that cancer is associated with short telomeres, but not long telomeres.

Contrary evidence: health benefits from extra-long telomeres

Just last spring, Maria Blasco’s group at the Spanish National Cancer Research Centre gave us this study, in which stem cells with hyper-long telomeres (up to 300,000 BP) were introduced into mice (in which telomeres normally are already 10 times as long as humans’).  “Mice with hyper-long telomeres…accumulate fewer cells with short telomeres and less DNA damage with age, and express lower levels of p53….We further show that wound-healing rates in the skin are increased in chimaeric mice.”  No life span data is reported, but cancer risk was lower in the mice with hyper-long telomeres.

This study from another research group at the same institution linked the extraordinary healing and regeneration capacity of very young mice to their extra-long telomeres.

In this study from UCSF, published just this fall, heart patients whose telomeres were lengthening over the four-year span of the study had 1/3 the mortality rate of matched patients whose telomeres were shortening over the same time span.

Are short telomeres a symptom or a cause of age-related disease?

This is a controversial question only because the causal hypothesis is in direct opposition to standard evolutionary theory.  So much the worse for standard evolutionary theory.

One causal mechanism which is incontrovertible is that cells become senescent when their telomeres shorten beyond a critical length.  Senescent cells are not just non-functional, they are toxic.  Removing senescent cells from the body has been shown to lengthen life span in mice, and senolytic agents are being developed for human use.  Many of us in the life extension movement regard senolytics as the #1 most promising strategy for major life extension in the near term.

The question of causality can be answered definitively by intervening to make telomeres longer or shorter “by hand”.  If short telomeres are a mere marker of past stress, then this should make little difference in the trajectory of aging; but if short telomeres are a cause of aging, then we expect that lengthening telomeres should lengthen life expectancy and lower the (age-adjusted) risk of disease.  In fact, this experimental model has been realized several times in mouse studies, two of which are referenced just a few paragraphs above [#1, #2].  The most dramatic success was in dePinho’s Harvard lab, but there are also impressive results from Blasco’s group in Madrid.

If lengthening telomeres is an effective life extension strategy for mice, it should be all the more so for humans, who have shorter telomeres, longer life spans, and less telomerase than mice.

In the face of this evidence, there are still some influential researchers and advocates in the anti-aging community who opine that “On the whole telomere length looks a lot like a marker of aging rather than the cause of problems: the groups that primarily seek to engineer longer telomeres in search of a way to slow aging are probably putting the cart before the horse.” [quoted today at FightAging.org]  Meanwhile, Michael Fossel has initiated a clinical trial of telomerase gene therapy to treat dementia.  Cancer scares from the echo chamber are spooking the venture capital that would be so welcome for startups that are seeking to bring telomerase therapy to the public.