A Dead Theory Still Walks

The way evolution works makes it impossible for us to possess genes that are specifically designed to cause physiological decline with age or to control how long we live.”  —from a Scientific American article by Jay Oshansky, Bruce Carnes, and Leonard Hayflick (2004)

Most biologists still think this way, even among people who study aging, even those working on anti-aging medicine.  If you believe this as a matter of bedrock theory, then what do you say when a gene is discovered that cuts life short, but still manages to dominate the gene pool?  You say that the gene has benefits that outweigh its costs.  It is a fertility gene, but it has side effects that kill you slowly.  Or it has survival benefit in the wild that are difficult to study in the laboratory.  This is called the theory of antagonistic pleiotropy.  “Pleiotropy” is the biological term describing a situation where one gene has two or more effects on the phenotype.  In 1910 when the term was invented, this was thought to be a special situation, requiring a special name.  We now know that almost all genes have multiple effects.

In theories of aging, antagonistic pleiotropy (in different variants), is considered the unassailable king of the roost.  It is not questioned.  There is no such thing as an aging gene, so as more and more aging genes are discovered, they carve out more and more excuses and exceptions to preserve their bedrock evolutionary theory.  Just this week, there are two new examples, in worms and in people.  

The First Aging Genes

In the 1980s, Tom Johnson, working at UC-Irvine, was studying aging in the lab worm C. elegans.  Johnson grew worms with a defective gene, which he named age-1 after he discovered that worms without it lived half again as long as normal worms. No one had ever imagined that a single gene could have such an effect on life span. In fact, the best experts in evolution had theorized that “everything ought to wear out at once,” so that no single gene could have any noticeable effect. Johnson’s discovery was the more remarkable because longer life required nothing new but rather the deletion of an existing gene. This implied that the effect of the age-1 gene was to cut the worm’s life short. What was it doing in the genome? How did it get there? And why did natural selection put up with it?

Johnson had a ready explanation. He believed (and still believes, I believe) in antagonistic pleiotropy. The worms without age-1 laid only a quarter as many eggs as other worms. It was easy to see how they had been losers in Darwin’s struggle. In fact, Johnson’s finding looked like a dramatic confirmation of the theory that aging was a side effect of genes for greater fertility, greater individual fitness. Aging had not evolved directly, selected for its own sake, but as a cost of greater fertility, a real-life example of antagonistic pleiotropy.

But a few years later, this story unraveled, and what had been confirmation of theory became a direct contradiction. Johnson discovered that his mutant worms actually had two genes that were different. In addition to age-1, there was another, unrelated gene defect (fer-15) on a separate chromosome. By crossbreeding, he was able to separate the two. Worms with the fer-15 mutation had impaired fertility without extended life spans. Worms with the age-1 mutation had extended life spans with unimpaired fertility. This was a full- fledged Darwinian paradox: the age-1 gene found in nature was the one that gave the worm a short life span. It was the “defective” gene that caused the worm to live longer. Age-1 looked not like a selfish gene but an aging gene. It was just the kind of gene that natural selection ought to eliminate handily. How had this gene survived, and what was it doing in the worm genome?

Age-1 was only the first case of an aging gene in worms.  There are now hundreds of genes known that lengthen life span when they are deleted. In other words, these genes, when present, have the effect of shortening life span. Some of them tend to improve fertility; some don’t. Some have other beneficial side effects, but about half the known life-shortening genes offer nothing in return, or at least nothing that has yet been identified.

Still, the pleiotropic theory is rarely questioned.


Fertility in male worms

A recent Nature paper from the Shanghai laboratory of Shi-Qing Cai identifies a pair of C. elegans genes that affect the span of fertility in males.  The group collected worms from many different locations around the world.  They found that in some worms, the males remain fertile almost their entire lives, while other males undergo rapid reproductive senescence.  With some excellent detective work, using database searches and genetic manipulation that would have been impossible a few years ago, they identified the genes rgba-1 and npr-28.  Each exist in two versions in wild populations, even though they have powerful effects on reproductive fitness.  Evolutionary theory tells us that genes with a close relationship to fitness should be subject to strong selection, so that the high-fitness version should promptly wipe out the low-fitness version.  In accord with theory, the authors cite statistical evidence that the high-fitness version of npr-28 has recently displaced the low-fitness version.  But, paradoxically, the low-fitness version of rgba-1 has displaced the high-fitness version.

Do they raise a flag in their article and protest that the theory is all wrong?  No, they are almost apologetic, and don’t dare to suggest that there’s anything wrong with the theory.  Such stark contradictions between empirical findings and the evolutionary theory of aging have become so commonplace that most everyone has become inured to them.  They shrug their shoulders and say, “there must be some hidden benefit associated with the wild-type gene that we have not yet identified.”  Part of the reason that they do this again and again is that this is happening in many different labs.  Perhaps each researcher in experimental genetics has only discovered one or two anomalies—they may be unaware that their finding is part of a larger pattern. 


Fertility in male mice

In August, a very similar discovery was made by a research group (Xiao-dong Wang’s) at the National Institute of Biological Sciences, Beijing, where I have been resident the last two summers.  Wang published a groundbreaking study demonstrating programmed reproductive senescence in male mice.  The RIPK1-RIPK3-MLKL signaling pathway in wild-type mice was identified as causing a kind of necrosis in male reproductive organs.  Inhibiting this pathway caused the males to retain fertility longer.   

In their Discussion, they say right off the bat, “The above presented data indicated that the previously unknown physiological function of necroptosis is to promote the aging of male reproductive organs.”  But they don’t challenge the pleiotropic theory.  Instead—quite typically for experimentalists—they speculate on a possible loophole that will save the theory:  Mice sired by older males are less healthy than those sired by younger males.  Aha—maybe this is completely unavoidable, and evolution has had to do what it could to prevent these less healthy pups from coming into the world.  “We therefore propose that necroptosis in testis is a physiological response to yet-to-be-identified, age-related, TNF family of cytokine(s) that transduces necroptosis signal through the canonical RIPK1-RIPK3-MLKL pathway.”  One thing they omit is that cutting off fertility to prevent the births of offspring that are (statistically) less healthy is no more consistent with the orthodox evolutionary theory (based on selfish genes) than are the theories that say aging is an adaptation.  Both require group selection, about which orthodox theory is in denial.


An Amish family lacking a death gene

Just this week, Douglas Vaughan’s group at Northwestern University reports identification of a rare genetic “defect” that gives some Amish families longer, healthier lives.  The gene called SERPINE1, encoding PAI-1, is mutated and non-functional in these families.  The result is longer telomeres, better insulin sensitivity, protection from cardiovascular disease, and longer life expectancy.  Conversely, the SERPINE1 must be regarded as an aging gene, having no purpose (we know of) except to hasten the demise of its owner.

What do the authors say about the evolutionary implications of their finding?  Exactly nothing.

In Japan, the life-shortening effects of PAI-1 have been known for several years, and there is already a drug in development that blocks its effect.  The drug is called TM5441, and a quick Google search located two lab houses [one, two] that sell it for the same exorbitant price.

Gericault – the Raft of Medusa

In Defense of Pleiotropy

To be fair, I should point out that these genes that have no other purpose than to cause early death really are the exception.  Almost all genes are pleiotropic in one way or another.  Much more common than pure aging genes like SERPINE1 is the situation where genes are dialed up or dialed down late in life in a way that is detrimental (or fatal).  The canonical example is mTOR, the target of rapamycin gene.  This gene plays an essential role programming the development of a young animal.  But when it is turned on late in life, it promotes aging and shortens lifespan.

My position is that this doesn’t let the theory of antagonistic pleiotropy off the hook.  Epigenetic programming is every bit as much under the control of evolution as gene sequences.  Many genes are turned on and off as needed, and this is a matter of course.  A matter of life and death, in fact.  If mTOR is turned on late in life, I presume that natural selection has deemed it so.

Pleiotropy is real.  Most genes have several functions.  But for the pleiotropic theory of aging to be valid, it must be true that tradeoffs are unavoidable.  In fact, when the theory was put forth by George Williams [1957], epigenetics had not yet been discovered, and there was yet no notion of turning genes on and off.  We now know that this process of gene regulation is an essential part of life in all eukaryotes, and that the timing of gene expression is exquisitely regulated.  It makes no sense to imagine (as Williams did) that once you have a gene you’re stuck with it, even if it kills you.  In fact, there are many genes that are turned on in youth and turned off in old age, and the effect is almost always to pro-aging.  In other words, aging is programmed for the most part not through aging genes like SERPINE1, and certainly not through pleiotropy, but rather through epigenetics.  Essential body systems like inflammation and apoptosis are re-purposed later in life as a means of self-destruction.

This opens onto a larger story, the subject of my book.

Aging in the news this week

In the press this week

  1. High-profile, misleading “proof” that aging is inevitable
  2. Disappointing results from Alkahest trials
  3. NewYorker article on exercise in a pill
  4. Splicing factors rescue senescent cells

  1. Mathematical proof that aging is inevitable

The headlines in the secondary scientific press said

Humans living forever is ‘impossible’ according to science

It’s mathematically impossible to beat aging, scientists say

Aging is Inevitable: Math shows Humans can never be Immortal

Mathematical models of aging are my specialty, but I’m not foolish enough to believe in the models.  I’m skilled and experienced at modeling so that I can adjust the assumptions to make a model do anything I want it to do.  I’ve seen time and again how tiny parameter changes can lead to opposite conclusions.  

Mathematical models can prove something is possible.  “Nature might arrange things in this way…”  But math models can never prove something is impossible.  Nature always has the option of arranging things in a way that’s different from the assumptions in your model.

In fact, the paper purports to be a general proof that aging is inevitable in all multicelled life.  But there are a few animals and many plants that don’t age.  Long periods of negative actuarial senescence (during which the probability of death goes down and down for years at a time) are common in trees, molluscs, and sea animals that keep growing without a characteristic, limiting size.  Turritopsis and Silphidae are capable of regressing to larval stage when starved and beginning life anew with a full life expectancy in front of them.  Annette Baudisch has made a career studying and documenting “negative senescence”.  So the idea that aging is some kind of mathematical certainty has about as much credibility as the authoritative declaration in Scientific American that flight by a heavier-than-air craft was impossible (1904 – more than a year after the Wright Brothers’ first flight).

The paper that appeared last week in PNAS is based on the premise that there is a kind of Darwinian competition among cells in the body.  Cells reproduce and mutate within the life of an organism.  In their model, somatic evolution–genetic change over time among cells in the same body–must navigate a course between Scylla and Charibdis.  The result is that mutations must accumulate, leading either to dysfunctional cells, too weak to do their job, or to cancer cells that have lost their allegience to the body and go on

They call this “aging,” but in fact somatic mutations do not contribute significantly to aging [ref].  Rather, in humans, the causes of aging include runaway inflammation, loss of insulin sensitivity, and thymic involution.  (In my view, most of these changes are driven in turn by programmatic epigenetic changes in gene expression.)  They redefine the term “senescent cells” to mean “cells that lose vigor due to cellular damage”, and then look for somatic mutations that cause the loss of vigor; but in general usage the term usually applies to cells that have critically short telomeres, or have otherwise entered a senescent state through epigenetic changes.  

The bottom line is that Masel and Nelson demonstrate a process that theoretically must kill us in the end, but their proof is silent about how long “in the end” might be, and they offer no evidence that the process they describe has to do with aging as humans (or other animals or plants) experience it.  Whatever “in the end” might mean, it must certainly be longer than 80,000 years, because that is the age of the Pando Grove which, last time I checked, qualifies as a multicelled life form.

Scylla and Charibdis


Blowing my stack (forgive me)

No one wants to think that death was handed to us with malice aforethought by evolution/nature/the gods.  In African myth, death was an accident caused by the laziness of a canine messenger of the gods.  In Judeo-Christian tradition, man would have been immortal if only Adam had not tasted the forbidden fruit.  William D Hamilton, one of the most insightful and best-grounded thinkers in evolutionary biology, proved that aging was an inevitable result of natural selection in 1966; forty years on, Baudisch and Vaupel used very similar reasoning to prove the exact opposite–that natural selection could never lead to aging [2004].  There are smart, famous people even today who argue that aging derives from the Second Law of Thermodynamics (Hayflick, of all people, is the man who discovered that cell lines run out of telomere).

We want to think that Nature is beneficient, that evolution has done her best by us and made us as strong and durable as possible.  If we get old and die, it must be because of something beyond evolution’s control.  But it’s just not true.  Natural selection first imposed aging on one-celled protozoans, and some of the same mechanisms that cause aging and programmed death in protozoans are active ingredients in human aging today (including telomere shortening and apoptosis).  Aging and programmed death have a long evolutionary history, and an ancient genetic basis.  We must conclude they exist for a purpose.

William Wordsworth asked, “Who shall regulate with truth the scale of intellectual ranks?”

Winston Churchill told us, “A lie gets halfway around the world before the truth has a chance to get its pants on.”

Arthur C. Clark said, “When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong.”  

Young Paul Nelson may be excused for getting carried away by his mathematics, but his mentor (and my former colleague) Joanna Masel ought to know that what they have done is irresponsible.  These memes have consequences.   Arguably, the small and under-funded community of anti-aging research is the most promising frontier of medical science today, offering a vision that may eclipse multi-billion dollar research programs in cardiovascular disease, cancer and Alzheimer’s disease.  Articles like theirs have power because the people who make funding decisions are not experts, they don’t like to be ridiculed, and they’re easily swayed by general sentiment in the research community = people who are already getting the funding.  

If we do not correct this impression, it is likely to discredit the most innovative and dynamic field of medical research today.


  1. Disappointing results from Stanford’s first trials of infusions of young blood

Alkahest is a for-profit spin-off from the Stanford lab of Tony Wyss-Coray, doing research with blood plasma from young animals infused into older animals.  I first wrote about the project two years ago.  The company leapt ahead of animal studies to try infusions of young plasma as a treatment for human Alzheimer’s patients.  Last week, Science Magazine reported on a pre-printed meeting abstract: no change in cognitive trajectory of patients who received the infusions.

The people I know best in the field of young plasma are Irina and Mike Conboy.  When I visited them last Spring, Irina told me she expected Wyss-Coray’s protocol couldn’t work.  The dosage is not sufficient, the duration of treatment is too short, and (according to the Conboys’ research) it is more important to remove pro-aging factors from old blood than it is to add the factors found in young blood.

Wyss-Coray took a chance, and I wouldn’t want to criticize his ambition.  But the research world being what it is, this high-profile failure is likely to set back funding for a promising research field.  Let’s do what we can to make sure that research by Wyss-Coray, the Conboys and Amy Wagers continues apace.


  1. New Yorker touts the Exercise Pill

An article in last week’s New Yorker began with a long encomium to the drug GW501516, developed by GlaxoSmithkline some 20 years ago, sold in the grey market as Cardarine or Endurobol.  Looking behind the headline led me to learn about  a family of transcription factors called PPAR.  They seem to be promising targets for life extension drugs that are just beginning to be explored.

“In mice, GW501516, either when combined with exercise or at higher doses by itself, induces some hallmarks of [exercise] adaptation such as mitochondrial biogenesis, fatty acid oxidation, an oxidative fiber-type switch and improved insulin sensitivity via AMP-activated protein kinase (AMPK)” [source]  

Sounds pretty good, doesn’t it?  But

“To its detriment however, tumorigenic effects of GW501516 have been reported and development was discontinued by Glaxo in Phase II clinical trials.”   

How serious is the risk of cancer?  Are there ways to separate the benefits from the hazards, either by combing with other drugs or by chemical modifications to the structure of GW501516?  Is there anyone with a lab who is seeking answers to these questions?  


Personally, at age 68 the three main ways that I feel my age are (1) decreased flexibility in yoga postures, (2) decreased speed in running and swimming, and (3) I can’t remember what the third one is.  I have charted my steady progression.  Swimming and running times are 30-35% longer than when I was 40, and increasing year by year on an accelerating schedule.  Exercise is my personal biomarker for age.  For reasons of vanity and vitality as well, I eagerly seek pathways to improved performance.  I also think that the activities of GW501516 and other PPAR agonists suggest potential for life extension, though there seem to be no lifespan studies either in rodents or humans.
Much of my source for what follows comes from a new paper summarizing exercise-mimetic drug state of the art, and references therein.

PPAR stands for Peroxisome Proliferator-Activated Receptor.  Peroxisomes are organelles in every cell that specialize in breaking down fat into short chains that the mitochondria can burn.  Thirty years ago, PPARs were discovered in the context of making more peroxisomes, but we now know that their most important function is to increase insulin sensitivity and signal a switch from burning sugar to burning fat.

Stimulating PPAR-α lowers LDL cholesterol and blood triglycerides.

PPAR-γ is a transcription factor that controls creation of new mitochondria.  (Mitochondria are the source of cell energy, and as we age, we have fewer of them and they become less efficient, linked to all diseases of age. [from my blog last summer: Part 1, Part 2]  Stimulating PPAR-γ improves insulin sensitivity and atherosclerosis.  PGC-1α is a protein that turns on PPAR-γ, indirectly creating more mitochondria.  Activating PPAR-γ has been discussed as an anti-cancer strategy.

Stimulating PPAR-δ (the modus of GW501516) switches the body from a preference for burning sugar to burning fat.  Great for weight loss and also for endurance.  You can double the running endurance of mice with GW501516.  Presumably, it was rather effective in enhancing performance in human long-distance runners before it was banned in 2009.  In calorie-restricted mice and long-lived mutants,    PPAR-δ is overactive.  (I’ve seen PPAR-β referred to only as similar to PPAR-δ. Maybe they’re the same.)

Joe Cohen at Self-Hacked sings the praises of GW501516.  Comments on this blog claim that (1) the increased cancer risk in rats was at very high doses*, and (2) the mechanism in rats doesn’t apply to humans.  Other commenters also minimize the cancer risk, but don’t offer references, and they may well be trolls for the companies that profit from GW501516.

“Although peroxisome proliferators have carcinogenic consequences in the liver of rodents, epidemiological studies suggest that similar effects are unlikely to occur in humans.” [source, ref, ref, ref, ref, ref].  “A number of experimental observations suggest that there is a species difference between rodents and humans in the response to PPAR agonists.” [same source] The article goes on to say that PPAR agonists may be more likely to create cancers in rat livers than human livers because rat livers have 10 times the PPAR expression compared to humans. It may be that tumorogenesis comes from the function for which PPARs were named: multiplying the number of peroxisomes.  But we now know that PPARs promote new peroxisomes in rodents but not in humans.

Here’s what I’ve been able to find out about PPARs, GW501516 in particular, and cancer:

PPAR is upregulated in colon cancer cells.  This shows that cancer causes PPAR, but not that PPAR causes cancer. There are many articles like this one, comprising evidence that activation of PPAR-δ promotes growth of existing tumors of the colon. The evidence is indirect, and gives no suggestion of the magnitude of the risk in humans who have colorectal cancer, let alone whether it in implies a risk for people who don’t have colorectal cancer.

PPAR-δ increases expression of COX2, the opposite of what aspirin and NSAIDs do.  NSAIDs decrease risk of cancer, and this suggests both that PPAR-δ increases risk of cancer and that the effect may be offset with NSAIDs.

There are no studies in humans.  There are many websites selling Cardarine, from which I guess that at least several thousands of people have taken taken it since 2005.  I have seen no sales numbers or estimates of the number of self-experiments, let alone cancer statistics. I have been unable to locate any anecdotes about cancer.

This 2004 review preceded GW501516, and reaches no conclusion.  It does, however, state baldly that PPAR-γ (not δ) is generally anti-cancer and that PPAR-α (not δ) causes cancer in rats but not in humans.

I have been unable to find published reports of the origina Smithkline-Glaxo experiment with rats that led to concern about cancer and abandonment of GW501516.

SR9009 is an unrelated mitochondria-growing drug sometimes mentioned in the same articles as GW501516.  There are no studies suggesting that it is carcinogenic, but that may be because it is much newer and there are so few studies altogether.

I don’t know whether Cardarine is too dangerous for human use, or whether similar drugs can be developed that target PPR-delta more safely.  But I’m outraged that the decision to abandon research on Cardarine has been made by investors in a board room who have no concern for public health and consider only the corporate bottom line.  This is an example of the worst kind of collision between capitalism and medicine–a collision which claims millions of casualties each year in the US alone.

I can’t blame the suits in the board room for doing their job, marching to the tune of those who paid the piper.  But this is emblematic of a gross failure of our regulation system, the FDA, and the reliance on for-profit drug companies to decide on our nation’s research priorities.  We now have (presumably) thousands of people taking a drug which may have large benefits and may have large dangers.  Most of them are motivated by wanting to be more buff or more sexy, and they are paying little heed to long-term consequences.  And because FDA has washed its hands of responsibility, there is no one even keeping records or collecting data to learn from the massive experiment about long-term health effects of GW501516.

Cardarine (GW501516) is available from LC Labs ($2240/g), from Monster Labs ($45/g), and from IRC Bio ($108/g, cheaper in quantity)


  1. Splicing Factors rescue senescent cells

I must admit that RNA splicing factors weren’t on my radar until this week, but I find this new experiment pretty convincing.  Eva LaTorre and colleagues from University of Exeter (UK) claim that splicing factors, more than sirtuins, are the pathway by which resveratrol (and analogs) extend life.

Sections of DNA (genes) are transcribed into messenger RNA, which finds its way to ribosomes, where the mRNA is translated into protein molecules.  But there is an in between step (in eukaryotes, but not in bacteria).  The DNA contains not whole (contiguous) genes but pieces of genes that need to be spliced together to assemble instructions for a whole protein.  Large sections of the DNA, called introns, are not intended for coding, and they need to be spliced out.  And, in fact, the pieces can generally be spliced together in different ways to make different useful proteins.  The work of splicing is performed by molecular complexes called splicing factors.  This is a process to which I had not given much thought until reading this article, but apparently it is a crucial step in epigenetics.  Epigenetics, the process of turning genes on and off, seems to get more complex with each passing year.

Resveratrol was identified about 15 years ago as a compound that extends lifespan in many species (but perhaps not in mammals).  Resveratrol has many effects, but the primary mode of action has been thought to be through SIR2 (or SIRT1) or related compounds called sirtuins that are selective gene silencers.  But the LaTorre group set out to show that the anti-aging benefit was through splicing factors rather than sirtuins.  They synthesized variations on the resveratrol molecule and tested them until they found one that promotes slicing factors but has no effect on sirtuins.  

Using this resveratrol analog, they were able to turn senescent cells back into fully functioning cells, with restored telomeres and other epigenetic changes.  They demonstrated that this was accomplished through splicing factors, and without sirtuins.

All this was done in (human) cell cultures, and it the horizons are now open to see what effect such rejuvenation has at the whole body level.


* Of course, there is no established dosage for GW501516, but pills come in 10mg and 20mg typically, corresponding to ~0.1 to 0.3 mg/Kg.  The highest doses I’ve seen discussed in  humans are ~2mg/Kg daily, nominally the same as the rat dosage.