Most overviews of aging suffer from multiple problems:

  • They dump a bunch of findings with no high-level picture.
  • Many of the claims they make are outdated, purely theoretical, and sometimes even outright disproven by existing work.
  • They are usually written by working academics, who are shy about telling us when their peers’ work is completely wrong.
  • They are shy about making strong claims, since this would also implicitly mean denying some claims of the authors’ peers.

This post is a high-level brain-dump of my current best models of the core pathways of aging, as I currently understand them. I have no particular reason to avoid calling out claims I think are wrong/irrelevant, and I’m going to present high-level models without pages and pages of disclaimers and discussions about results which maybe disagree with them (but are probably just wrong/irrelevant).

Epistemic status: I would be surprised if none of it turned out to be wrong, but there are multiple lines of evidence supporting most claims. It is not highly polished, and references are included only when I have them readily on hand. My ideal version of this piece would have more detailed references, more double-checking behind the claims, and more direct presentation of the data which backs up each claim. Unfortunately, that would take enough time and effort that I’m unlikely to actually get to it soon. So… here’s what I could produce in a reasonable amount of time. Hopefully it will be wrong/unhelpful in ways orthogonal to how most overviews are wrong/unhelpful.

Foundations

First, let’s recap a couple foundational principles. I’ll go through these pretty quickly; see the linked posts for more info.

Homeostasis and “Root Causes” in Aging: the vast majority of proteins, cells, etc, in the human body turn over on a timescale from days to months. At any given time, their level (e.g. protein concentration, cell count, etc) is in equilibrium on the turnover timescale - i.e. the rate of creation approximately equals the rate of removal. For any X with turnover much faster than aging (i.e. decades), if we see the level of X increase/decrease on the timescale of a human lifetime, then that is not due to permanent “accumulation of X” or “depletion of X”; it is due to increase/decrease in the rate of creation/removal of X. For instance:

  • DNA damage is typically repaired on a timescale of hours or faster, depending on the type. If DNA damage levels increase with age, that is due to an increase in rate of damage or decrease in rate of repair, not permanent accumulation.
  • Typical senescent cells turn over on a timescale of days to weeks. If the number of senescent cells increases with age, that is due to an increase in rate of senescent cell production or decrease in rate of removal, not permanent accumulation.
  • Elastin is believed to not turn over at all in humans. So if we see elastin deposits increasing with age (e.g. in wrinkles), then that could be permanent accumulation.

Furthermore: suppose we have a positive feedback cycle. Increasing A decreases the rate of production of B, so B decreases. But decreasing B decreases the rate of removal of A, so A increases. If both A and B individually turn over on a timescale of hours or faster then this feedback loop as a whole will also typically operate on a timescale of hours or faster - i.e. count/concentration of A will explode upward on roughly that timescale. More generally, a feedback loop will usually operate on the timescale of its slowest component, exactly like the rate-limiting step of a chemical reaction.

Main upshot of all this: since aging involves changes on a timescale of decades, there must be some component which is out-of-equilibrium on a timescale of decades or longer (i.e. does not turn over significantly across a full human lifespan). These are the components which we’ll call “root causes”. Everything else which changes with age, changes only in response to the root causes. Reset the root causes to young-organism levels, and everything else will equilibrate to young-organism levels in response. Furthermore, a reset of the root causes only needs to happen once every few decades for humans - it fully resets the human to a youthful state, so ongoing treatment is not needed on short timescales.

The Lens, Progerias and Polycausality: The lens of the human eye consists of fiber deposits which do not turn over significantly over the course of a human lifetime. New fiber layers are added over time, so the lens grows from around 3.5mm in infancy to 5.5mm in old age. The main clinical result is the near-universal need for reading glasses in old age.

This is a well-understood root cause of one symptom of old age. Furthermore, it is very likely independent of most other age related diseases - lens thickening is unlikely to cause cancer or heart disease, for instance. So, it’s an existence proof: there is more than one root cause of aging.

That said, there’s a fair bit of evidence that most symptoms of aging - including the major age-related diseases - share a common root cause, or at least a common core pathway. Some kinds of evidence of this:

  • Most symptoms/diseases of aging are correlated - someone who has one early is likely to have others. Conceptually, if you do a factor analysis on aging symptoms, there’s one big factor for a bunch of diseases, even after controlling for the number of years one has lived. (“Aging clock” is a relevant piece of jargon here.)
  • At the cellular level, a lot of diseases of aging “look similar”, and involve similar pieces. There’s a decrease in cell count, increase in damaged proteins/DNA/fats, and inflammation. We see roughly this pattern in Alzheimers, atherosclerosis, muscle loss, and many others.
  • Certain simple interventions reliably produce many diseases of aging - for instance, progerias are single mutations which produce a whole “early aging” phenotype
  • Conversely, certain simple interventions reliably delay many diseases of aging - e.g. calorie restricted diets.

… so these all point to shared underlying causes. This post will be about the “core pathways” most likely involved.

Major Diseases

These subsections will talk about various specific age-related diseases, mainly highlighting how they connect to the cellular processes we’ll talk about later. Two main themes to watch: reactive oxygen species and senescent cells.

Reactive oxygen species (ROS) aka free radicals are produced in greater numbers in old age. These are short-lived, highly reactive molecules. They react with all sorts of things, oxidatively “damaging” whatever they hit, including proteins, fats, and DNA.

Senescent cells are cells which have partially shut down in a programmatic way, triggered by some sort of “stress” on the cell (e.g. DNA damage, exposure to harsh chemicals or radiation, etc). They pump out inflammatory signals (called the senescence-associated secretory phenotype, or SASP). Eventually, they’re removed by the immune system.

In later sections, we’ll see that these two are tightly coupled. For now, we’ll talk about how they seem to underlie a variety of age-related diseases.

Atherosclerosis

If you dissect young and old mammals, one of the most obvious internal differences is in the blood vessels:

Aortas from a 32 year old and a 24 year old on top, a 55 year old and a 65 year old on bottom. (source)

Mammals of any age have “fatty streaks” along the walls of the vasculature, which are exactly what they sound like. (These are the slightly lighter patches in the pictures above - more obvious in the older aortas, but faintly visible in the young pair as well.) In older mammals, the fatty streaks tend to be larger, until in old age they necrotize (aka die) in the middle and turn into thick “atherosclerotic plaques” (the dark patches in the lower right picture above). These can block blood circulation, and sometimes a chunk of the plaque can break off and block circulation in smaller vessels; either of these can cause e.g. heart attack or stroke. 

At any age, a lower-fat diet is associated with smaller fatty streaks and lower chance of atherosclerosis, though the streaks universally grow with age holding diet constant.

The big breakthrough in atherosclerosis came in the 80’s/90’s. It was known that a certain type of cell (called a macrophage) would hoover up fat from the bloodstream, then adhere to the cell wall when full, forming the fatty streaks. The missing puzzle piece was that the macrophages don’t just hoover up any random fats (there’s rather a lot of fat in the bloodstream and not that many macrophages; it would be like sweeping sand off the beach). They specifically hoover up partially oxidized fats. Steinberg has a great review of various experiments feeding into this discovery.

The upshot: streaks and plaques grow in old age primarily because the concentration of (partially) oxidized fats in the bloodstream increases in old age. These probably don’t have a very slow turnover time, so either the rate of (partial) oxidation of fats increases in old age, or the rate of removal decreases. Which is it, and what causes it?

The main candidate here is ROS: increasing ROS levels in old age increase the rate of oxidative damage to fats. (One interesting question: how does the relative increase in oxidative damage to fats compare to DNA/proteins? Do the numbers actually line up for these to share a source? I haven’t seen a solid Fermi estimate on this, I’d be interested to see one.)

Vascular Stiffening

Even aside from atherosclerosis, the walls of blood vessels change in old age. In particular they become stiffer.

In normal operation, the heart pumps out blood in discrete chunks - each heartbeat pumps out some amount of blood into the arteries. The arteries expand a bit to accumulate this blood, much like a balloon. And, like an inflated balloon, the arteries are at slightly higher pressure from this extra blood, until it’s pushed out through the capillaries and the cycle starts again with the next heartbeat. It’s sort of like a capacitor: the heart sends out blood at high pressure in sudden waves, and the arteries expand and contract to store the extra blood and smooth out the pressure variance.

If the walls become stiffer, then the arteries are less able to smooth out this pressure variance. From the heart’s perspective, it’s trying to pump blood into a hard container, which works about as poorly as you’d expect; this is one of the major causes of heart failure (alongside atherosclerosis). The vessels also become more likely to burst (i.e. aneurysm). 

What causes the loss of elasticity?

The main answer seems to be oxidative damage to proteins in the wall of the blood vessels. The key experimental evidence backing this up is the effect of aminoguanidine (AG), a powerful scavenger of certain ROS: AG administration reverses age-related stiffening of the arteries. This is only temporary - i.e. the arteries go back to their stiff state shortly after AG administration stops - so it isn’t a root cause, but it is a link in the causal chain. Once again, oxidative damage likely caused by increased ROS seems to be the key causal intermediate.

(Please don’t go running out to take aminoguanidine. The side effects are serious, and the benefits are short-term - you’ll go right back to where you would have been as soon as you stop taking it.)

An interesting side note: many sources claim that a shift in collagen:elastin ratio is involved in stiffening of the vasculature. This seems to be based on a purely theoretical paper from decades ago, and the actual data doesn’t back it up, yet review articles and textbooks continue to mention it. 

Alzheimer’s

The first and most important thing to know about Alzheimer’s (aka dementia, aka old folks losing their memory) is that it is not caused by accumulation of amyloid beta.

Decades ago, people noticed that if you look at the brains of old people with dementia, they usually have lots of plaques, and these plaques are made of a particular protein fragment called amyloid beta. Therefore clearly amyloid beta causes dementia. Pretty soon people were using amyloid beta plaques to diagnose dementia, which made it really easy to show that the plaques cause dementia: when the plaques are how we diagnose “dementia”, then by golly removing the plaques makes the “dementia” (as diagnosed by plaques) go away.

As far as I can tell, there has never at any point in time been compelling evidence that amyloid beta plaques cause age-related memory problems. Conversely, I have seen at least a few studies suggesting the plaques are not causal.

Meanwhile, according to wikipedia, 244 Alzheimer’s drugs were tested in clinical trials from 2002-2012, mostly targeting the amyloid plaques. Of those, only 1 drug made it through. Bottom line: they don’t work.

So what does cause Alzheimer’s? I don’t know; it’s not a disease I’ve studied in depth. I know plenty of studies find the usual culprits involved - inflammation, damaged proteins, etc.

I do know of one particularly interesting cluster of studies which found that the brain “opens up” during sleep, increasing flow of cerebral fluid to clear out whatever junk accumulated during the day - including amyloid beta. The flow takes place in “paravascular” spaces, i.e. spaces around the blood vessels, which widen during sleep. Given the age-related changes in blood vessels (e.g. thickening & stiffening of walls), it would make sense for this paravascular space to not open up as much in older people, and indeed that seems to be the case. Whether this has anything to do with dementia, I don’t know, but it is my current best guess for the main cause of the amyloid plaques. If true, this would mean that the plaques share their main root causes with changes in the vasculature.

Sarcopenia

Sarcopenia is age-related loss of muscle mass - i.e. old people becoming physically weak. Overviews often say it can be (at least partially) “reversed” via exercise, which seems to be fairly obvious bullshit aimed at making people exercise more. Bottom line: for any fixed amount of exercise, muscular strength will fall with age. (Obviously more exercise can still increase muscle strength at any age.)

Many sources will claim that lots of research has shown sarcopenia to be caused by loss of muscle innervation - i.e. problems with the neuromuscular junctions (NMJs). As with amyloids and Alzheimers, as far as I can tell there was never at any point in time any compelling evidence that NMJ changes are actually causal for muscle loss, and my current best guess is that they are not causal. However, research on age-related changes in NMJ structure did produce very shiny images, which I think is probably the main reason they received so much attention in the first place.

Neuromuscular junctions from young (2-month) and old (25-month) mice. (source)

So what does cause sarcopenia? I don’t yet have a full picture here, but I can sketch out my current best guess.

Muscle cells are not normal cells; they’re mega-cells, a hundred times longer than a normal cell, with hundreds of nuclei all within the same cell. They need so many nuclei because molecules which would diffuse quickly from one end to the other of a normal cell would take far too long to diffuse across a muscle cell (diffusion time increases faster-than-linearly with distance); so, things need to be produced locally.

This setup also makes it difficult for an entire muscle cell to turn over. Instead, the nuclei turn over (along with all the usual protein/membrane/etc turnover mechanisms); they’re regularly replaced from “satellite cells”, a type of stem cell nestled in next to the muscle whose job is basically to crank out new nuclei from time to time.

In sarcopenia, one cross-section of the long muscle cell will fail first - a “ragged red” section - and then failure gradually spreads along the length. The failing section involves the usual culprits: ROS, mitochondrial deficiency (related to ROS; more on that later), inflammation signals, etc.

My best guess is that this is basically just how cellular senescence manifests in a muscle cell. One particular satellite cell is “close to senescence”, and produces nuclei which rapidly “senesce”. But since muscle cells are really a bunch of relatively-isolated components along their length (due to long diffusion time), this only results in one section of the cell failing. Eventually, the “senescence” spreads to adjacent sections via the usual mechanisms of senescence-induced-senescence (more on that later). Key characteristics of “ragged red” sections of the muscle cell - like high ROS, mitochondrial deficiency, or inflammation - are the usual characteristics of senescent cells, and sarcopenia probably shares its root cause(s) with cellular senescence more generally.

Other Diseases

This section briefly mentions a few other age-diseases which I’ve read a little bit about, but haven’t studied in as much depth. I’ll just give a few comments on how they tie in to the cellular processes discussed later.

Arthritis: arthritis is basically inflammation in the joints. I’ve heard plenty of claims that it’s caused by increasing numbers of senescent cells. This would make sense; senescent cell counts are firmly established to increase with age, and senescent cells secrete inflammatory factors, so put 2 and 2 together. On the other hand, I also heard a high-profile clinical trial based on this hypothesis recently failed; I don’t consider that especially strong evidence, since that could easily be due to something specific to the trial. I don’t know enough to have a strong belief on whether senescent cells are the main factor here, but it’s my most likely current model.

Osteoporosis: calcium regulation goes completely bonkers with age. I’ve looked into this only briefly, and my main conclusion is that it’s a confusing mess and I have no idea what’s going on. The most promising direction I’ve stumbled on was in a physiology class, unrelated to aging, when the professor mentioned that osteoblasts (bone-making cells and a major calcium regulator) are derived from immune cells and respond to inflammatory signals. Given that aging in general tends to involve low-grade chronic inflammation (most likely due to increasing numbers of senescent cells), “calcium regulation goes completely bonkers” seems like the sort of thing which would result.

Cancer: the key requirement for cancer is cells with oncogenic mutations. As usual, increasing cancer rates could be caused in two main ways: increasing production rate of mutations, or decreasing removal rate of mutant cells. There are plausible age-related mechanisms for both of these. On the production side, we’ll later discuss how genomic instability relates to ROS and senescence, and in particular the role of transposons. On the removal side, the immune system weakens with age, likely for the same reasons as everything else weakens with age. Also, in this case it’s plausible that an increase in the production rate of precancerous cells and/or senescent cells would slow the removal rate as well, simply because the immune system has limited capacity.

Cataracts: as with hardening of the vasculature, cataracts can be reversed by aminoguanidine, so the same considerations apply.

Core Intermediates

Now we get to the meaty part. At the microscopic level, there’s a handful of pieces which pop up again and again in age-related diseases:

  • Oxidative damage to DNA, proteins, fats, etc.
  • ROS
  • Senescent cells
  • Inflammation
  • Mitochondrial dysfunction

Some of these have obvious connections. For instance, more ROS presumably lead to more oxidative damage to DNA/proteins/fats/etc. Some key questions here:

  • Where are the ROS produced? Mitochondria are the top candidate - there’s a known mechanism for ROS production by mitochondria, as well as experimental evidence that mitochondrion-targeted antioxidants specifically reduce ROS-induced damage.
  • How do the ROS and/or damaged molecules move between compartments, e.g. nucleus/cytoplasm/extracellular? I have seen very little on this, and consider it a major blindspot. I’m not sure if it’s a blindspot for the field or if I just haven’t found the right cluster of papers.
  • Are the quantitative changes in DNA/protein/fat damage compatible with a single underlying cause? Do they match plausible estimates of ROS from dysfunctional mitochondria? Again, I haven’t seen Fermi estimates here, but I’d like to.

We do have evidence (from aminoguanidine and similar drugs; good overview here for protein damage) that ROS are causal for various types of damage.

Another connection: we’ve already mentioned that senescent cells release inflammatory factors, the so-called “senescence-associated secretory phenotype” (SASP). The one question here for which I haven’t seen a clear answer is whether increasing numbers of senescent cells quantitatively match age-related increases in inflammation. Drugs to remove senescent cells are a hot area right now, and should provide more evidence on whether senescent cells are causal for age-related inflammation.

So we have two clusters of probably-causally-connected processes. One of these involves dysfunctional mitochondria producing excess ROS which damages DNA/proteins/fats, and the other involves senescent cells inducing inflammation.

The really big discovery of the past twenty years was the connection between cellular senescence and ROS/mitochondrial dysfunction. Turns out, ROS, DNA damage, and mitochondrial dysfunction are all tied together in one bistable feedback loop, and cellular senescence is basically a state-change in that feedback loop.

The DNA Damage <-> Mitochondrial ROS Feedback Loop Underlying Senescence

The two key papers here are “Feedback between p21 and reactive oxygen production is necessary for cell senescence“ and “Mitochondrial Dysfunction Accounts for the Stochastic Heterogeneity in Telomere-Dependent Senescence”. Another group found basically the same phenomenon, but they found some weird differences compared to normal senescence which I think were probably artifacts of how they blocked mitochondrial function, so they’re less directly relevant to cellular senescence in the wild.

Here’s how the feedback loop works:

  • A cell’s DNA is damaged, inducing a damage response.  (p21 is the main signalling molecule which activates the damage-response genes.)
  • As part of this damage response, mitochondria are shifted into a lower-efficiency state, producing less energy and more ROS.
  • The ROS then further damage DNA.

At low levels of activation, the ROS do not produce enough DNA damage to make this loop self-sustaining. The cell stays in a low-ROS, low-damage state - the “normal” state. But once it passes some threshold, the positive feedback takes off, and the cell transitions into a high-ROS, high-damage state - the “senescent” state.

Notably, after a few days, the cell changes in some other (not yet fully understood) manner, locking in the senescent state. After this point, even if ROS are suppressed, the cell will remain in senescence rather than switching back to the normal-cell state. Transposons are one plausible candidate for this lock-in; more on that shortly.

The experimental evidence for this process is beautiful (I definitely recommend those papers, especially the second). And it neatly unifies all of the pieces we listed above: cellular senescence is a positive feedback loop between ROS, damage and mitochondrial dysfunction, and the SASP connects it all to inflammation.

There’s still one big question: why is this positive feedback loop active more often, for more cells, in old age? The whole feedback loop is fast, senescent cells are removed on a timescale of days to weeks, so there must be some upstream change either increasing the rate of triggering senescence (presumably by somehow damaging DNA, possibly via ROS, or inducing mitochondrial dysfunction) or decreasing the rate of removal of senescent cells. In fact, both of these do occur, although personally I think the increase in rate of triggering senescence is much more likely to be causal.

Further Up The Causal Chain

There are lots of stories about how various plausible root causes could, perhaps, trigger the senescence feedback loop. I think transposons are the most likely candidate, though mitochondrial mutations are a plausible mechanism as well. We’ll talk about both of those in the next two subsections. Most other proposed root causes can, I think, be ruled out at this point - we’ll talk about some of those in the “Other Root Causes” subsection below.

Transposons

A transposon is a gene whose main function is to copy itself. The LINE-1 family of transposons (most common active transposons in humans) consist of a protein which snips the DNA, and another protein which reverse-transcribes the transposon’s mRNA into DNA attached to the snipped end. (I’m glossing over some details here.)

These things are extremely common in the genome - more than half of human DNA consists of dead transposons. (“Dead” here means that the transposon mutated at some point in evolutionary history, so it’s no longer functional.) Fortunately, the number of non-dead transposons in the human genome is much smaller  - even the highest estimates I’ve seen put the number typically below 100.

We do have mechanisms to repress transposon activity, most notably epigenetic mechanisms. Most DNA is usually tightly coiled up around little cylindrical proteins (called histones), where it can’t be easily transcribed. “Epigenetics” typically refers to modifications of the DNA and/or histones which make the coils tighter or looser, making the DNA difficult or easy to access. Most transposons are epigenetically tagged so that they’re kept tightly coiled most of the time. Indeed, an argument could be made that this is the primary role of epigenetics - it’s certainly what most epigenetic modifications are doing most of the time, since transposons fill so much of the genome.

There’s an obvious story by which transposons could be a root cause of aging. Most of the time they’re repressed, but every once in a while, one of them manages to copy itself. Once it’s copied, there’s no undoing it - the transposon count will only go up over time. Eventually, there’s enough active transposons that the repression mechanisms aren’t so reliable. At that point, the transposon protein which snips the DNA will be expressed quite a bit, resulting in lots of DNA damage - those snips are exactly the sort of damage which can trigger senescence. (Note: it’s a lot easier to snip the DNA than to reverse-transcribe, so I generally expect there to be a lot more snips than successful transposon-copy events.)

Setting aside this root-cause story for a moment, there’s also evidence that cellular senescence causes derepression of transposons. We’ll talk more about the details of that later, but the key idea here is that there’s a trade-off at the cellular level between repairing damage and repressing the transposons. When DNA damage is high, the cell temporarily shifts resources to repair that damage, deregulating the transposons. In senescence, the level of DNA damage is constantly high, so transposons go wild.

Of course, transposons themselves cause DNA damage (i.e. the snips), so eventually this can lock in senescence. That’s my current best understanding of why senescence gets “locked in” after a few days: with transposons driving the DNA damage, the cell will remain senescent even if ROS are suppressed. It’s a second senescence feedback loop, slower to start up, but permanent - once the transposons are active enough, the cell cannot leave senescence.

Unfortunately, this makes it difficult to distinguish the chicken from the egg. We know that senescence can cause transposon activity, but we also suspect that transposon activity comes first and causes senescence. We can’t test that hypothesis just by looking at transposon activity in senescent cells. In principle, we could test it by looking for an age-related increase in transposon count in non-senescent cells, but that turns out to be actually-pretty-difficult in practice. (Modern DNA sequencing involves breaking the DNA into little pieces, sequencing those, then computationally reconstructing which pieces overlap with each other. That’s a lot more difficult when the pieces you’re interested in have millions of near-copies filling most of the genome. Also, the copy-events we’re interested in will vary from cell to cell.)

One more thing to note about this model: suppose that a stem cell has a high transposon count, but not high enough to undergo senescence itself. That stem cell will pump out new cells, as stem cells typically do, and those cells will themselves be close to senescence. We should therefore see little clusters of senescent cells, each derived from one stem cell. This is where I expect most senescent cells come from in aged tissue. The “ragged red” sections of aging muscle cells are a good example - one satellite cell is near senescence, so it pumps out nuclei which rapidly senesce, and the section of the muscle cell near that satellite ends up senescing.

This picture also offers an obvious story for cancer - transposons are a major driver of mutations.

Mitochondrial Mutations

Mitochondrial mutations are the center of the “mitochondrial free radical theory of aging” (MIFRA); Aubrey de Gray’s MIFRA book provides an excellent (though out-of-date) summary of the model. Some parts of the model have been pretty well nailed down by evidence at this point - e.g. the idea that most core diseases of aging are mainly driven by ROS, which in turn are produced by mitochondria. The positive feedback loop underlying senescence even further connects symptoms of aging to mitochondrial ROS.

The main piece of the MIFRA model which still seems up-in-the-air is the idea that the root cause is mitochondrial mutations.

Background: mitochondria have their own separate DNA. There’s only a handful of genes on it; most necessary mRNA/protein sequence is supplied by the nuclear DNA. But the mitochondria’s little DNA is particularly prone to mutation - it doesn’t have the nucleus keeping it safely separated from the highly-energetic processes of the rest of the cell. The flip side is that mitochondria turn over frequently, separate from turnover of the cell itself, and they have quality-control mechanisms - e.g. if a mitochondrion isn’t producing energy (as indicated by low transmembrane potential) then it’s broken down.

A key idea hypothesis for the mitochondrial mutation model is that mutant mitochondria which are only partially defective aren’t broken down. In fact, under this hypothesis, such mitochondria have a replicative advantage, and can expand to take over the whole cell. This cell then becomes a “hotspot”, pumping out lots of ROS.

Today, we would expect that such hotspots are senescent cells - this degree of mitochondrial dysfunction should certainly be enough to induce senescence.

There is some evidence for this - for instance, mitochondrial mutants tend to take over whole cells, rather than being spread evenly across cells. However, there just aren’t very many of them - senescent cells outnumber mutant-mitochondria-dominant cells by a wide margin in old age (order of magnitude: think 10% vs 1%). Also, as we mentioned earlier, senescent cells don’t reproduce and do turn over, so even if mutant mitochondria took over one cell, they’d need to somehow expand to others.

This still doesn’t rule out the model entirely. Some possibilities:

  • Maybe a small fraction of senescent cells are long-lived, and these produce enough ROS to induce senescence in a much larger population of cells.
  • Maybe mutant mitochondria spread before the cell is cleared (there is evidence for exchange of mitochondria between neighboring cells).
  • Maybe stem cells with some mutant mitochondria produce new cells which go on to rapidly senesce, much like we hypothesized for transposons.

On the other hand, it’s also plausible (I think more plausible) that defective mitochondria are negatively selected in healthy cells, and they only expand to take over in already-senescent cells, where all the mitochondria are already pumping out less energy and more ROS anyway.

Other Root Causes?

Telomeres

DNA has repetitive regions called telomeres at the ends . Each time a cell divides, the copying starts a little ways in from the end, so the telomeres get a bit shorter. Eventually, the telomere runs out, and the bare DNA end is interpreted as damage, inducing senescence. This has long been known to cause cellular senescence in vitro, and was hypothesized as a root cause of aging.

Indeed, telomeres are known to shorten with age. On the other hand, upregulating telomerase seems to do approximately nothing to prevent cellular senescence or aging more generally. What’s going on here?

Stem cells produce a protein (telomerase) which extends their telomeres. Since most cells are replaced regularly by stem cells, that should be enough to prevent telomere-induced senescence. But then why do we observe shorter telomeres in old age? The key result here is that DNA damage, and ROS in particular, shorten telomeres. In vivo, telomere length is mainly a measure of ROS damage, not a measure of age directly. So, while telomere loss can cause senescence, the main thing which causes telomere loss is not gradual shortening as cells divide, but rather ROS-induced damage.

So, telomere loss is likely involved as an intermediate cause (as a type of DNA damage induced by ROS), but not a root cause.

AGEs

Advanced glycation end-products (AGEs) are proteins which have somewhat-randomly reacted with a sugar in a Maillard reaction - the same type of reaction which browns foods and gives them flavor when cooking. These products are hypothesized to accumulate long-term in the body, since they can’t be broken down.

I don’t know whether there’s any evidence that these molecules actually accumulate long-term. (Just because they’re not broken down doesn’t mean they’re not simply excreted.) I haven’t seen direct evidence, but I haven’t searched very carefully either, and I haven’t seen direct evidence against.

The main argument I’ve seen in favor of AGEs as a root cause of (some) diseases of aging was from aminoguanidine. It seems like people thought it prevented AGE formation before they realized that it interrupted oxidative damage more broadly, and thus interpreted results from aminoguanidine as indicative of a causal role for AGEs. In hindsight, even if this had shown a causal role for AGEs, it would have ruled them out as a root cause: the effects of AG rapidly wear off once administration of the drug ceases, so it’s definitely not blocking any root cause.

Senescent Cells as Root Cause

For a while, people hypothesized that senescent cells accumulate with age without turning over, acting as a root cause. As mentioned earlier, the actual evidence suggests that senescent cells turn over on a timescale of days to weeks, which would mean this theory is wrong - senescent cell accumulation is not a root cause.

However, there is a saving throw: maybe a small subset of senescent cells are longer-lived, and the experiments measuring senescent cell turnover time just weren’t capturing the long-lived subset in particular. Results from senolytics (drugs which kill senescent cells) suggest this is also wrong: the effects of senolytics rapidly wear off once the drug stops being administered, whereas reversing a root cause should set an organism back to a youthful state longer-term.

Protein Damage, DNA Damage, Etc as Root Cause

Sometimes people suggest protein damage, DNA damage, etc, as root causes. These generally turn over on fast timescales, so... no. I expect that most of these people do not have any thought-out notion of what “root” means in “root cause”, and are just using it as a synonym for “really important”.

Other Pieces of The Pathway

This last section will briefly dive into a couple other aspects of the core pathways of aging which I expect people might be interested in, and talk about how they fit into the picture.

Sirtuins and NAD

David Sinclair published a popular book on aging a couple years ago, mainly talking about his own research areas. The sirtuins are one of the main key pieces of that research.

We mentioned earlier that, when damage is detected, the cell redirects resources from repressing transposons to repairing damage. Sirtuins are one such resource. They directly trade off genomic stability (including transposon repression) for repair capacity.

Notably, sirtuins consume the energy carrier NAD as part of their repair role. Lots of things use NAD as an energy carrier, switching it between a high-energy and low-energy state, but sirtuins actually consume it - the whole molecule is incorporated into a new structure. Usually the cell has rather a lot of NAD, but if the damage load is high and the sirtuins are doing lots of repairs, then it can be depleted. That leaves less of it for various cellular processes which use NAD as an energy carrier - including mitochondrial energy production.

This all fits neatly into our main model: high ROS-induced damage draws sirtuins away from transposon repression, so they become active. Meanwhile, the sirtuins’ consumption of NAD can interfere with mitochondrial function, resulting in more ROS production.

This also adds in one more piece: at younger ages, certain kinds of cellular stress (like radiation exposure or chemical damage from an inflammatory response to an infection) can also damage DNA, temporarily reducing repression of transposons. This probably won’t result in immediate senescence in most cells, but a few transposons may copy before everything goes back to normal. The aging clock ticks forward a little faster than usual, due to these events.

Damaged Proteins Connect To Everything

We’ve mentioned oxidatively damaged proteins many times now. What we didn’t mention was the numbers: in young organisms, perhaps 10% of protein is damaged. In old organisms, it’s more like 20-30%.

That is a percentage of all proteins. And almost everything our cells do is done by proteins.

Natural conclusion: efficiency of a very wide variety of processes will be thrown off in aging.

In most cases, this shouldn’t be too noticeable - we’re only talking about a 10-20% change, and random noise does that for most protein species anyway. And the body has lots of feedback loops in place to handle exactly this sort of thing. By and large, biological networks evolve to be robust to 10-20% changes in protein concentrations. But, it does make things difficult for science: if there’s a 10-20% change in everything, then there are always going to be statistically significant age-related changes in everything, even though most of them aren’t actually all that relevant.

So, be warned: there’s lots of 10-20% age-related changes all over the place which mostly aren’t that relevant.

Recap

Here’s the core positive feedback loop again:

  • A cell’s DNA is damaged, inducing a damage response.
  • As part of this damage response, mitochondria are shifted into a lower-efficiency state, producing less energy and more ROS.
  • The ROS then further damage DNA.

Since this is a positive feedback loop, it has two stable states: one state with low damage and ROS (the “normal” cell state), and one with high damage and ROS (the “senescent” cell state).

A few days after senescence is triggered, the cell’s transposons become active, copying themselves and further damaging the DNA. At this point, the transposon activity alone is enough to maintain the senescent state, and senescence is locked in. The cell will remain senescent until it’s cleared out by the immune system, a few days to weeks later.

What causes more senescent cells as we age? The transposon model says that transposon count increases with age - they are the root cause which permanently accumulates over time. Once the transposon count in a cell is high enough, it produces enough damage to trigger the senescence feedback loop. More specifically, we end up with stem cells with enough transposons to be just below the senescence trigger, and these stem cells produce new cells which rapidly senesce.

Senescent cells release inflammatory factors (the SASP) as well as ROS. These cause the bulk of age-related diseases. ROS damage to fats causes buildup of fatty streaks and eventually plaques in the arteries - i.e. atherosclerosis - eventually leading to blockage and strokes. Damage to proteins hardens the blood vessels, leading to heart failure and aneurysm. Chronic inflammation underlies arthritis and possibly osteoporosis. Senescence itself also leads to loss of cells, including muscle loss.

Finally, in very old age, the whole process can accelerate: ROS produced by senescent cells can cause damage in adjacent cells, inching them closer to senescence as well. The more cells senesce, the more damage they deal to healthy cells. Eventually the whole thing goes supercritical (though on a slow timescale), leading to an exponential acceleration of disease progression in old age.

108

54 comments, sorted by Highlighting new comments since Today at 6:32 AM
New Comment

(I made an account to post this)

We do have mechanisms to repress transposon activity, most notably epigenetic mechanisms. Most DNA is usually tightly coiled up around little cylindrical proteins (called histones), where it can’t be easily transcribed. “Epigenetics” typically refers to modifications of the DNA and/or histones which make the coils tighter or looser, making the DNA difficult or easy to access. Most transposons are epigenetically tagged so that they’re kept tightly coiled most of the time.

These mechanisms are small interfering RNAs (siRNAs) and Piwi-interacting RNAs (piRNAs), that increase methylation of the histones associated with transposons, making them 'tighter', or harder to access. According to the Wikipedia page for transposon silencing, these siRNAs and piRNAs are most active in the gonads. This makes sense, as it would avoid germline transmission of active transposons, allowing offspring to be born with a lower active transposon count than their parents.

After reading that, I wondered why on earth we don't have these transposon-suppressing RNAs coursing through our bloodstream in the same concentration as we do in our gonads. According to this paper, suppressing transposons also has the effect of suppressing neighboring genes, leading to a possible reduction in the organism's fitness. The same paper claims that having transposons could have beneficial effects on genome evolution, as transposons create regions of suppressed recombination around them, although I don't fully understand the reasoning behind this being good for organism fitness. Also, if suppressing transposons does have negative effects on the genome, that doesn't at all explain why it happens more in the gonads. Perhaps aging just wasn't selected against enough in the ancestral environment.

If nothing else, these siRNAs and piRNAs seem to be effective at making babies have fewer active transposons than their parents. If someone has injected old mice with a bunch of copies of these RNAs (probably wrapped in viruses first) and observed the results, then I can't find their paper published anywhere. On the off chance that the transposon model is correct, and that the cure for transposon proliferation really is as simple as an RNA injection, this is one experiment we can't afford not to do.

Nice comment!

A couple minor expansions on this (you might know these already, but I want to make sure it's clear to everyone else):

  • siRNAs and piRNAs don't quite make babies have fewer transposons than their parents. The babies have the same number of transposons as the parents' egg/sperm. The piRNA/siRNA activity in the egg/sperm is just higher than in other (somatic) cells to make extra sure that the transposons don't copy before the genome is passed on.
  • There is a little more to it than just injecting RNAs. The RNAs would have to get into at least the cells and possibly the nucleus somehow, and also RNAs turn over quickly so the effect would not last very long. Conceptually, though, the idea is basically viable, modulo some technical hurdles.
  • piRNAs/siRNAs repress transposon activity, but they don't remove existing transposons outright. So this would effectively put aging on pause, and clear up symptoms of aging, but not reverse aging. Once the administration stopped, things would bounce right back to normal.

There are some similar approaches. For instance, we could try to upregulate expression of the repression mechanisms, rather than add more RNA directly. Another approach is to directly interfere with the reverse transcription via antiretroviral drugs, e.g. lamivudine. (Retroviruses are viruses which reverse-transcribe themselves into the genome - HIV is the most notable example). This paper tested lamivudine in aged mice, and indeed found that it improved a bunch of age-related problems. (Though note that we would not expect this to remove existing transposons or even prevent damage from the endonuclease snipping the DNA if transposon count is already high; it just interferes with the reverse transcription step.)

I'm still confused. My biology knowledge is probably lacking, so maybe that's why, but I had a similar thought to dkirmani after reading this: "Why are children born young?" Given that sperm cells are active cells (which should give transposons opportunity to divide), why do they not produce children with larger transposon counts? I would expect whatever sperm divide from to have the same accumulation of transposons that causes problems in the divisions off stem cells. 

Unless piRNA and siRNA are 100% at their jobs, and nothing is explicitly removing transposons in sperm/eggs better than in the rest of the body, then surely there should be at least a small amount of accumulation of transposons across generations. Is this something we see?

I vaguely remember that women are born with all the egg cells they'll have, so, if that's true, then maybe that offers a partial explanation (only half the child genome should be as infected with transposons?). I'm not sure it holds water, because since egg cells are still alive, even if they aren't dividing more, they should present opportunities for transposons to multiply.

Another possible explanation I thought of was that, in order to be as close to 100% as possible, piRNA and siRNA work more than normal in the gonads, which does hurt the efficacy of sperm, but because you only need 1 to work, that's ok. Still, unless it is actually 100%, there should be that generational accumulation.

This isn't even just about transposons. It feels like any theory of aging would have to contend with why sperm and eggs aren't old when they make a child, so I'm not sure what I'm missing.

My understanding is that transposon repression mechanisms (like piRNAs) are dramatically upregulated in the germ line. They are already very close to 100% effective in most cells under normal conditions, and even more so in the germ line, so that most children do not have any more transposons than their parents.

(More generally, my understanding is that germ line cells have special stuff going to make sure that the genome is passed on with minimal errors. Non-germ cells are less "paranoid" about mutations.)

Once the rate is low enough, it's handled by natural selection, same as any other mutations.

Thanks! I changed "transposons" to "active transposons" to be more accurate. Much of my knowledge in this domain comes from a genetics course I took in the 10th grade, so it's not super comprehensive.

piRNAs/siRNAs repress transposon activity, but they don't remove existing transposons outright. So this would effectively put aging on pause, and clear up symptoms of aging, but not reverse aging. Once the administration stopped, things would bounce right back to normal.

My understanding was that methylated DNA stayed methylated (silenced), and methyltransferases made sure that copies of the methylated DNA sequences were also themselves methylated. If all transposons in a cell were methylated by piRNAs and siRNAs, wouldn't all descendants of the cell also have methylated transposons, making those transposons effectively removed? (Of course, that assumes that methyltransferases and  transposon-suppressing RNAs have 100% success rates, which I'm sure they don't. This would explain why babies have a few active transposons, but not nearly as many as their parents.)

This paper asserts that piRNAs both methylate transposons, and also cleave the RNA transcripts of transposons in a cell's cytoplasm, and that doing so guards the germline against transposons. Cleaving the transcripts of transposons would repress transposon replication in the short term, but, as I understand it, methylation of transposons would silence them in the long term, including in daughter cells. Therefore, even if there's a one-time transposon-methylating event (as opposed to a permanent epigenetic upregulation in transposon-suppression mechanisms, which seems to be a promising idea as well), the number of active transposons in the genome should still be reduced, pushing the growth trajectory of transposons backward.

So, DNA methylation. This is another area where the things-people-typically-say seem to be completely wrong. I had also heard that methylation was long-lived (making it a natural candidate for a root cause of aging), but at one point I looked for experimental evidence on the turnover time of epigenetic methyl groups. And it turns out that most methyl groups turn over on a timescale of ~weeks. The mechanism is enzymatic - i.e. there are enzymes constantly removing and replacing epigenetic methyl groups, so they're in equilibrium.

I'm glad this came up, in hindsight I probably should have mentioned it in the post.

Wow, I had no idea that methylation was that impermanent, thank you for the belief update. I guess that leaves upregulation (via acetylation?) of transposon-suppressing RNA, extending lifespan by varying expression of other genes that alter chromatin structure to be more transposon-hostile, or as this comment says, using Crispr/CAS9 to incapacitate transposons. I wonder if anyone has done/will soon do an experiment like this in mammals.

 The first steps are (probably) to come up with an estimate of both material and labor costs for all 3 of the above options. The labor costs might be mostly nullified if you can find altruistic biologists, or biologists that are status-seeking and have sufficient confidence that the transposon hypothesis is true. Or if a motivated person who isn't a biologist takes a crack at it.

UMichigan offers a transgenic mouse service for $5,800. From the item description: 

The Transgenic Core guarantees that at least 300 fertilized mouse eggs will be microinjected with CRISPR/Cas9 reagents. Microinjected eggs will be transferred to pseudoopregnant female mice. Tail tip biopsies will be provided to the Investigator's laboratory for genotyping. Mouse pups will be transferred to the investigator at weaning.

This is the service for C57BL/6 (C57 black 6) mice, the mice most commonly used as disease models, and the best-selling mice from mouse-breeding laboratories. For another $1,100, UMich will also "build CRISPR/Cas9 reagents to target a specific location in the mouse or rat genome". So, for $7,000, one can get a founder population of transgenic mice, targeting any genome location the customer desires. Another transgenic mouse service from UMich, also for $5,800, guarantees at least 3 transgenic founder mice will be produced. 'Founder' implies that the actual experimental subjects will be the children of the transgenic mice you get, so you'd need to head down to PetStop and buy a few dozen hamster cages, some rodent chow, and a mouse-breeding manual.

Jackson Labs, the primary provider of experimental mice, sells C57BL/6 mice for $90 per mouse at 25 weeks old, and at $430 per mouse at 90 weeks old, with cost per mouse growing roughly linearly in between. At 25 weeks old, that's $2,700 for 30 mice (enough for a pilot study's control group). 

There's also the cost of shipping and handling live mice, which will vary depending on where the experiment is conducted. There are probably a bunch of auxiliary costs I haven't considered yet as well. My main point is, as far as the Crispr route goes at least, I don't anticipate material costs over $50,000, meaning an unofficial pilot study is probably quite doable by a small group of motivated individuals / one crazy person in a shed / crowdfunding.

In terms of experimental endpoints, would this mainly just be an experiment to see how long the mice live? If so, that does seem like a high-upside experiment which even someone with relatively little domain knowledge could just go do. The main investment would be time - it would take at least a couple years of mouse-care, and hopefully longer.

If the project were undertaken by someone with more domain expertise, the main value-add (relative to the bare-minimum version of the experiment) would probably be in checking more endpoints, especially as a debugging tool. For instance, since the CRISPR/CAS targets would presumably have very high copy number, it might be hard to get it to actually remove all the live transposons and not be saturated by dead copies which share a lot of the sequence. Checking that it actually worked would require sequencing, and special tools are needed to get accurate transposon counts from sequencing data. Also, it might require some nontrivial design to find CRISPR/CAS targets which actually work. Then there's the possibility that CRISPR/CAS9 themselves trigger transposon derepression (they involve snipping then repairing DNA, after all), which probably wouldn't be a game-breaker but could throw some general weirdness into things. There's also the question of which transposons to target, since there's a few major families and presumably a long tail of minor families...

Anyway, point is, there's a lot of potential failure points which could be addressed with some effort and expertise. The bare-minimum version of the experiment would be huge if it worked, but if it failed, it would be hard to tell whether the theory was wrong or the experiment was flawed in some way. That said, the chance of success and the potential upside are high enough that it seems worthwhile even for the bare-minimum version.

I could imagine Constantin being interested in this - it's not exactly the thing she set up LRI for, but it's not a huge number of steps removed, and she'd probably at least have useful advice on how to make it happen and what to watch out for in terms of execution.

I'm also curious if anyone knows of existing groups already running this kind of experiment; I would not be surprised if it were already underway and we saw results published in another year or two (since the mice take a while to age). (More generally, do people have advice on searching for projects which have started but not published yet? I frequently stumble on them on the "projects" pages of the websites for particular labs, but I don't know a good way to search for them.)

It does sound like this research is already planned or underway.

But new experiments are planned. For example, the team will purposely encourage expression of transposable elements to see if that undermines health and lifespan. Another approach could be to use the powerful CRISPR gene editing technique to specifically disable the ability of transposable elements to mobilize within the genome. If that intervention affected lifespan, it would be telling as well, Helfand said.

The wording is a little ambiguous as to whether the CRISPR approach is merely being contemplated, or whether they're just floating the idea. Working with flies first makes sense, since it gives you a faster feedback loop on whether transposon elimination affects lifespan.

Stephen Helfand, the researcher quoted in the article, seems not to have published a new article since 2016, when the report I linked was published appears not to have updated his publication page since 2016, but you can find his later works on Google Scholar by searching his name (SL Helfand).

I've emailed him to ask whether this idea has been acted upon. I'll post back here if I hear from him. In the meantime, I'm going to investigate the work of the followup project and the leaders associated with it.

It does look like this cluster of researchers is making progress.

Treatment of aged mice with the nucleoside reverse transcriptase inhibitor lamivudine downregulated IFN-I activation and age-associated inflammation (inflammaging) in several tissues.

Lamivudine is also called 3TC, and it's already approved for use against HIV. A clinical trial on its efficacy against Alzheimer's is underway and scheduled to be complete in June 2022.

Have you thought of https://www.vium.com/ to reduce labor costs?

https://www.twitch.tv/videos/962157156 -> laura deming recently ran a cool twitch on methylation - MAKE SURE TO SAVE THE VIDEOS BEFORE THEY GET DELETED BY TWITCH IN 14 DAYS

Wouldn't the role of transposons be easy enough to investigate by incapacitating functional transposons with Crispr/CAS9? Has something like that been done in mice? 

I would love to see a study like that.

Minimal cell experiments (making cells with as small a genome as possible) have already been done successfully. This presumably removes transposons, and I have not heard that such cells had abnormally long lifespans.

One possibility is that there are at least two aging pathways-the effect of transposons, which evolution wasn’t able to eliminate, and an evolved aging pathway intended to eliminate older organisms so they don’t compete with their progeny (doing so while suffering ill health from transposon build-up would be less fit than dying and delegating reproduction to one’s less transposon-heavy offspring).

There is significant evidence that most organisms have evolved to eventually deliberately die, independent of problems like transposons that aren’t intentional on the level of the organism. Yamanaka factors can reverse some symptoms of aging, and appear to do so by activating a rejuvenation pathway. This makes perfect sense if the body deliberately ordinarily reserves that pathway for gamete production, while letting itself deteriorate. It is extremely confusing if aging is purely damage, however. Yamanaka factors don’t provide new information (other than the order to rejuvenate) or resources; a body that is doing its best to avoid aging wouldn’t seem to benefit from them, and could presumably evolve to produce them if evolution found this desirable. Other examples include the beneficial effects of removing old blood plasma (this appears to trick the body into thinking it is younger, which should work on a deliberately aging organism but not one that aged purely through damage), the fact that rat brain cells deteriorate as the perceive the brain to gradually stiffen with age, but rejuvenate if their ability to detect stiffness is removed, and the fact that some species of octopuses commit suicide after reproducing, and refrain from doing so if a particular gland is removed.

If both transposons and a deliberate aging pathway contribute to aging, it would be very interesting to see what happens in an organism with both transposon inactivation and Yamanaka factor treatment. Neither appears to create massive life extension on its own, but together they might do so, or at least point out worthwhile directions for further inquiry.

Minimal cell experiments (making cells with as small a genome as possible) have already been done successfully. This presumably removes transposons, and I have not heard that such cells had abnormally long lifespans.

The minimal cell experiments were done with mycoplasma, which (as far as I know) does not age. More generally, as I understand it, most bacteria don't age, at least not in any sense similar to animals.

Also, I expect wild-type mycoplasma already had no transposons in its genome, since the organism evolved under very heavy evolutionary pressure for a small genome. (That's why it was chosen for the minimal cell experiments.)

An initial search doesn’t confirm whether or not mycoplasma age. Bacteria do age though; even seemingly-symmetrical divisions yield one “parent” bacterium that ages and dies.

If mycoplasma genuinely don’t, that would be fascinating and potentially yield valuable clues on the aging mechanism.

Bacteria do age though; even seemingly-symmetrical divisions yield one “parent” bacterium that ages and dies.

Do you have a reference on that? I'm familiar with how it works with budding yeast, but I've never heard of anything like that in a prokaryote.

https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.0030058

This is the source I found. It’s fairly old, so if you’ve found something that supersedes it I’d be interested.

Oh wow, that's really neat. I doubt that it has any relevance to the aging mechanisms of multicellular organisms, but very cool in its own right. And definitely not transposon-mediated.

Naked mole rats don't age. Other mammals do. Therefore, whatever causes ageing must be hard but not impossible for evolution to stop. Here is one plausible hypothesis. 

The environment of naked mole rats provides unusually strong evolutionary pressure against ageing. So transposon-killing RNAs are unusually prevalent. Every time a mutation breaks a transposon, that provides an advantage, the fewer transposons you start with, the slower you age. This selection is balanced by the fact that transposons occasionally manage to replicate, even in the gonads.  In naked mole rats, that selection was unusually strong, and/or the transposons unusually unable to replicate. So evolution managed to drive the number of functioning transposons down to 0. 

If naked mole rats have no functioning transposons, and animals that age do contain transposons, that would be strong evidence for transposon based ageing. 

Of course, even if ageing is transposon based, evolution could have taken another route in mole rats. Maybe they have some really effective transposon suppressor of some kind.

I don't know how hard this would be to test. Can you just download the mole rat DNA and put it into a pre-made transposon finder?

My guess would be transposon suppression rather than evolving away all of the transposons - upregulating existing repression mechanisms would be easier than removing every single active transposon copy. Though I'd still be interested in the test - even if suppression is the main mechanism, I'd still be interested to see how the number of transposons in naked mole rats compare to other rodents. (It is a nontrivial test to run, though - DNA sequencing is particularly unreliable when it comes to transposons, because there are so many near-copies.)

Alternatively, it's also plausible that the naked mole rat's defenses are elsewhere in the chain. In particular, they live in a low-oxygen environment, so it's often speculated that they have very low ROS levels. If that's the case, we'd still expect transposons to copy occasionally, but cellular-stress-events in which DNA is damaged by ROS, and transposons are temporarily less-repressed while the damage is repaired, would be much more rare.

I've seen this claim about naked mole rats thrown around a bunch but it's left me with the question of what naked mole rats do die of? If their mortality likelihood truly doesn't increase, we'd expect there to be some very long-lived naked mole rats. Is the issue just we haven't held them in captivity for long enough to see them die of natural causes? I vaguely remember reading somewhere that eventually they stop eating or die in other ways but can't seem to find the reference now.

From https://www.livescience.com/61568-naked-mole-rats-no-aging.html:

In the lab, the cause of death is usually hard to find; the main issue that shows up in necropsies, Buffenstein said, are mouth sores, indicating the animals weren't eating, drinking or producing saliva well in their last few days and infection set in.

"We really don't know what's killing them at this point," Buffenstein said. 

As a thought experiment mostly for testing my own understanding, suppose we could do a bulk culling of transposons in all of an elderly human's stem cells (or all cells). If I understand correctly, this post's main hypothesis (DNA damage <-> ROS feedback loop) would imply the following should happen:

  • Senescent cell fraction quickly (within days or months) starts reverting to its healthy level.
  • Atherosclerosis heals on its own because ROS production reduces to its healthy level meaning the plaque equilibrium returns to the young level.
  • Similarly, vascular stiffening reverses for the same reason AG works temporarily.
  • Alzheimer's remains unclear without further understanding but we can guess this might help.
  • Sarcopenia same story as atherosclerosis and stiffening.
  • Lens and elastin fibers continue to build up, so we'll all be blind and wrinkly but otherwise healthy...

The one thing I'm less clear on is where immune system aging fits into this. I feel pretty confident that a treatment like this wouldn't cause the thymus to spontaneously grow for example but am more uncertain about some of the other aged immune system phenotypes. It seems plausible that reducing the load on the immune system would allow it to regain some of its ability to deal with infectious diseases for example.

Does this fit with your understanding?

Yup, exactly right. This would be the most direct possible test of the hypothesis.

Re:thymus, this study found that a mitochondrially-targetted antioxidant prevented thymic involution, so there is at least some evidence that thymic involution is caused by the same core pathways. Though the timing of thymic involution is pretty suspicious, when compared to the other core-pathway diseases.

As a funny aside, a few months ago, I had the thought "removing all transposons would be a nice somewhat pointless but impressive demonstration of a civilization's synthetic biology mastery." I guess the "pointless" part may have been very wrong!

This suggests an interesting way to test the theory. JCVI had their "minimal cell" a few years back: they took a bacteria with an already-pretty-small genome, stripped out everything they could while still maintaining viability, then synthesized a plasmid with all the genes and promoters but with the "junk" DNA between them either removed or randomized (to make sure there was no functionality hiding in there which they didn't know about), and grew the bacteria with the synthesized plasmid. More recently, they have a project to do something similar with yeast.

Once this sort of project scales up to mammals, I expect they'll try it with mice/rats, and removing transposons is an obvious step. One prediction from the transposon theory of aging is that, when they do so, they'll find that their mice are far longer-lived and have near-zero rates of cancer, heart disease, etc.

Good point, this also suggests that Genome Project-Write is an important project.

One thing I wonder about here is whether or not having a certain amount of "garbage" in the DNA is not actually a good thing. My understanding is that material transfers due to chromosomal overlaps as well. As that would be a purely random process there's no guarantees that transfers occur at the beginning and end of the used/functional gene segment. Having some amount of meaningless sections seems like it would reduce the probability of the legs of the chromosomes overlapping at dangerous locations.

A private message (from someone who may name themselves if they wish) asked about the claim that "the effects of senolytics rapidly wear off once the drug stops being administered".

This is a lower-confidence claim than most in this piece; I do not have a study on hand directly proving it. The vast majority of papers on senolytics either use regular administration (frequency ~1 per 2 weeks or faster) or do a short regimen and then measure results within ~2 weeks; the ubiquity of those practices is itself a significant piece of evidence here. If a single dose does nothing long-term, then I mostly expect to see very few papers published using a single dose on long-term outcomes (because people don't publish negative results), and that is indeed what I see.

Of course, lots of people still claim that long-lived senescent cells are a thing, but it's always either uncited or cites someone else who also has no actual data directly supporting the claim. I have yet to see any compelling data backing it up, and this is a case where I expect absence of evidence to be evidence of absence. It feels very similar to the claims about amyloid beta as a cause of Alzheimers, or collagen:elastin ratio as a cause of vascular stiffening: lots of people claim it, but following the citation chains shows no compelling evidence.

What I'd really like to see is an experiment which gives one (or a few) doses of a senolytic to reasonably-old mice (maybe ~20 mo), measures outcomes ~ 1 week later, then waits at least a couple months before evaluating outcomes again. If persistent senescent cells are both real and relevant, then we'd expect to see relatively little regression in outcomes over time. If persistent senescent cells are not real or not relevant, then we'd expect to see the mice mostly regress to the pre-treatment state after a couple months. (Though they would probably be slightly better off than control mice, since they'd have effectively taken a couple weeks with the core aging feedback loop on pause.)

This is interesting work, and I like seeing how your other writing (on gears-level models, constraints, etc) seems to lie behind your approach to this research.

That said, there’s a fair bit of evidence that most symptoms of aging - including the major age-related diseases - share a common root cause, or at least a common core pathway. Some kinds of evidence of this:

  • Most symptoms/diseases of aging are correlated - someone who has one early is likely to have others. Conceptually, if you do a factor analysis on aging symptoms, there’s one big factor for a bunch of diseases, even after controlling for the number of years one has lived. (“Aging clock” is a relevant piece of jargon here.)
  • At the cellular level, a lot of diseases of aging “look similar”, and involve similar pieces. There’s a decrease in cell count, increase in damaged proteins/DNA/fats, and inflammation. We see roughly this pattern in Alzheimers, atherosclerosis, muscle loss, and many others.
  • Certain simple interventions reliably produce many diseases of aging - for instance, progerias are single mutations which produce a whole “early aging” phenotype
  • Conversely, certain simple interventions reliably delay many diseases of aging - e.g. calorie restricted diets.

One interpretation of the idea of a "common core pathway" is an argument that there's some single aging factor (SAF), which triggers various forms of damage, which in turn cause the major diseases of aging. Hence, fighting those intermediate forms of damage would be ultimately futile, while fixing the SAF would be a panacea, at least if followed by other treatments that fix lingering damage.

An alternative idea is that there are multiple aging factors (MAFs), each of which triggers a particular form of damage. The damage it triggers goes on to cause a range of other forms of damage, which cause still others, resulting in a damage chain reaction that causes the major diseases of aging. Under this model, there are many MAFs that can trigger this chain reaction, and it's just the first one that's responsible for kicking off the aging process.

The difference is that preventing or curing one MAF wouldn't halt or prevent aging. Preventing one MAF still leaves you open to others. And since it's not the MAF, but a self-reinforcing chain reaction of damage, that drives aging, curing the MAF wouldn't reverse or halt the damage.

As an analogy, in the SAF model, aging is like a flood. It causes a range of problems: washing away buildings and roads, drowning deaths, erosion. Try to repair the buildings while the flood's going on, and you'll make little progress. But if you can stop the flooding, then you can repair all the damage it caused and be "good as new."

By contrast, in the MAFs model, aging is like a collapsing building. It could be caused by many underlying problems: termites, an earthquake, a bomb, demolition, fire, rot, a flood. But once the process is underway, the building enters a feedback loop. The flood introduces mold, which leads to rot, which opens holes in the house to allow the entry of vermin, which cause support beams to sag, and increases the risk of a short-circuit that could cause a fire. At a certain point, the original cause (the flood) can vanish entirely, but the accumulated damage is just perpetuating itself, and you can only save the house by eliminating all the damage sources.

The MAFs model can also explain many forms of evidence you cite. Since any given MAF can kick off the damage chain reaction, you'd expect the diseases caused by that damage to be correlated, though perhaps with some "leading" and the others "following" in any given case. The types of damage would be common, so you'd see common damage symptoms. You'd expect to see that certain interventions would cause the whole phenotype (an early MAF causes an early damage chain reaction).

The harder one to explain is the interventions that delay diseases of aging. Why should one intervention avoid many MAFs, unless all those MAFs had an underlying SAF that caused them all? So the crux of the MAF vs. SAF models seems to be whether calorie restricted diets (or other similar interventions) really work.

Great comments!

In dynamical system terms, I'd call the MAF scenario a single bistable feedback loop with many redundant components. ("Redundant" in the sense that many component subsets are sufficient to support the bistable feedback loop.) The senescence feedback loop is an example of this: there's multiple components, and only a subset are needed to support the state change. For instance, either mitochondrial dysfunction or transposon activation would be sufficient to trigger the state change, and either one will cause the other once the state change is triggered.

So just because there's one core feedback loop underlying all these core diseases, does not mean that there's one root cause which activates that feedback loop.

There are reasons to believe there's one root cause (or at least very few), but as you point out, these do not fully overlap with the reasons to believe there's one core feedback loop. The main reason I expect one/few root cause(s) is that things which are out-of-equilibrium on the timescale of human aging are really rare in biological systems. The vast majority turn over much faster. So there just aren't that many things which could be root causes.

Calorie restriction and the like do provide additional evidence, though I actually don't think these provide particularly strong evidence for one/few root cause - not because it's unclear whether they actually work (CR is one of the best-replicated findings in the field), but because they might just slow down the core feedback loop without directly touching the root cause(s). It is very likely that the core feedback loop itself accelerates the root cause(s), based on the exponential-acceleration pattern of the major age-related diseases, so slowing down the core feedback loop should slow down the root cause(s) too (though not reverse them or slow them down below the rate at which they progress in youth).

I like seeing how your other writing (on gears-level models, constraints, etc) seems to lie behind your approach to this research.

It was actually the other way around - I studied aging (among many other things) first, and those systems were what I drew on to understand gears, abstraction, constraints, etc.

There's one additional crucial phenomenon that I think needs to be better highlighted here, which is natural selection for cancer as an organism ages.

Cancer isn't best understood as a manifestation of genetic damage that the body is unable to remove or repair. That's what causes individual cancer cells to emerge and survive. But that alone does not guarantee that the damaging agent (the cancer cell) will replicate itself. Damage and failure of disposal alone only leads to local blockage.

Cancer goes beyond this to propagate itself throughout the organism. It evolves, developing drug resistance, the ability to colonize new "soils," and better evade the immune system with time. This is a feature that, as far as I know, is unique to cancer. And it depends on the mechanism of natural selection.

By contrast, lipofuscin, transposon-caused DNA damage, and protein plaques seem mainly to accumulate, causing local dysfunction. It's deadly, but it's ultimately a passive force. 

Hence, I think that "selection for cancer" deserves to be mentioned as a "further up the chain" core pathway of aging. This is an active, antagonistic force actively working to defeat our efforts to cure it.

Eh, yes and no. I'm definitely on board with selection pressure as a major piece in cancer. But for purposes of finding the root causes of aging, the key question is "why does cancer become more likely with aging?", and my current understanding is that selection pressures don't play a significant role there.

Once a cancer is already underway, selection pressure for drug resistance or differentiation or whatever is a big deal. But at a young age, the mutation rate is low and problematic cells are cleared before they have time to propagate anyway. Those defences have to break down somehow before cancer-selection-pressures can play any role at all, and it seems like the rate-limiting step (i.e. the step which accounts for the age-related increase in cancer) is the breakdown of the defences, not the time required for selection itself. Once cancer is underway, the selection process operates on a faster timescale.

it seems like the rate-limiting step (i.e. the step which accounts for the age-related increase in cancer) is the breakdown of the defences, not the time required for selection itself. Once cancer is underway, the selection process operates on a faster timescale.

Maybe there are multiple aging clocks.

  • Some run independently.
  • In other cases, clock A 'triggers' clock B. 

Cancer selection might be triggered by the 'primary aging clock' you've proposed. Once that happens, you now have two clocks running simultaneously. However, the 'cancer clock' could still be potentially triggered even in the absence of the 'primary aging clock.'

Then the empirical question becomes when in the life cycle the 'cancer clock' starts ticking. Is it always running in the background, but DNA damage from other sources makes it speed up over time? Or, in the absence of the 'primary aging clock,' would the 'cancer clock' stay off for most people, most of the time, barring exposure to some potent carcinogen?

There's also a possibility of hard-to-specify combinations of physical and mental damage that are self-perpetuating, and operate independently of DNA damage. Each year, perhaps people face a risk of trauma that triggers a cycle of progressively self-destructive behavior, a 'behavioral dysfunction clock.' That's speculation, just meant to illustrate the idea.

It still seems conceptually important to focus attention on the 'primary aging clock' as a hypothetical target. I agree with you that it's plausible the 'primary aging clock' is the bottleneck. So this is more a move from "there's one aging clock" to "there might be more than one aging clock, but one of them is a lot more responsible for diseases of aging than the others."

Main upshot of all this: since aging involves changes on a timescale of decades, there must be some component which is out-of-equilibrium on a timescale of decades or longer (i.e. does not turn over significantly across a full human lifespan). These are the components which we’ll call “root causes”. Everything else which changes with age, changes only in response to the root causes.

A quibble: Just because some component turns over frequently, doesn't mean that higher-level structures made from that component aren't degraded in the process. For example, if I accidentally cut off the tip of my finger, the relevant cells will all grow back, but the finger will not; the larger-scale pattern remains degraded for life.

In the case of my fingertip, obviously we would consider that an injury, not an aspect of aging. But it seems hard to be sure that there aren't any aspects of aging that work this way?

The key idea here is the difference between "local" vs "nonlocal" changes in a multistable system - moving around within one basin vs jumping to another one. The prototypical picture:

For your finger example, one basin would be with-finger, one basin without-finger. For small changes (including normal cell turnover) the system returns to its with-finger equilibrium state, without any permanent changes. In order to knock it into the other state, some large external "shock" has to push it - e.g. cutting off a finger. Once in the other state, it's there permanently (as long as there aren't more large shocks); the new state is stable.

In the absence of large external shocks, the system hangs around in a stable basin. In terms of information in the high-level structures, this means that any information is either (a) degraded quickly, on roughly the same timescale as component turnover, or (b) maintained indefinitely. Picture an actual ball being gently shaken around in a bowl: information about the ball's exact position at any given time will be lost quickly; its position a minute or two later won't tell us much about its position now. However, the fact that it's in the bowl will be maintained indefinitely (more technically, maintained for an exponentially long time, assuming the average shaking energy is substantially lower than needed for the ball to jump out). Any information which isn't lost quickly, will likely stick around for a very long time.

Very interesting post! you have radically changed my way of thinking about aging. Until now, I have always thought of aging as a very complex problem with most part of the body beginning to independently fail in different ways. The idea that all the symptoms of aging might ultimately be caused by a few 'root causes' is really very interesting.

Since calorie restriction slows aging, is there a positive relationship between calorie intake and number of DNA mutations?

Thanks for writing this up! We need more people thinking about how to defeat aging. 

Curious if you've read Nintil's posts? https://nintil.com/categories/aging/

In principle, we could test it by looking for an age-related increase in transposon count in non-senescent cells, but that turns out to be actually-pretty-difficult in practice. (Modern DNA sequencing involves breaking the DNA into little pieces, sequencing those, then computationally reconstructing which pieces overlap with each other. That’s a lot more difficult when the pieces you’re interested in have millions of near-copies filling most of the genome. Also, the copy-events we’re interested in will vary from cell to cell.)

I wonder if something like single cell ATAC-seq could help here? There's still the problem of aligning near-copies but it seems like there's already some work trying to deal with this problem. (I haven't read either of these papers in detail but the second specifically mentions transposons as a use-case.)

Yeah, I haven't read up on the topic in depth, but there's a few toolkits specifically intended for sequencing transposons. So it's probably not something which would require major breakthroughs at this point, but it does require specialized tools/knowledge, rather than just the standard sequencing toolkit.

If I follow the logic correctly, the root cause of aging is that stem cells can irreversibly and invisibly accumulate active transposones, which are then passed on to derived cells, which then become senescent much faster. Also, for some reason this process is supressed in gonads. So, I see these alternatives:

  1. Transposone activation is essentially blocked in gonades, or
  2. There is a barrier which prevents embryos with above-normal number of active transposones from developing, or
  3. Children born from parents of old age will age faster, or
  4. Active transposone accumulation is not a root cause of aging.

On the 'old parent' hypothesis : let me point out that if there is not a full 'reset' somewhere in the chain, well, we wouldn't exist, because some of our ancestors were old parents, and the effect would be cumulative, such that none of us would live past childhood due to aging.  (there is a disease like that, albeit with a different root cause)

If the suppression of transposons in the gonads is good enough, the reset could be the same as with any other harmful mutation - shuffling by sexual reproduction and natural selection. 

 

Which may suggest a reason why sexual reproduction exists in the first place.

?  So the hypothesis here from you is this:

           base age = (age of parents at conception) * <a small number>

           human bio age = calendar age + base age

With this formula, without anything to subtract base age, it will monotonically increase and eventually extinct the species.  Sexual reproduction doesn't solve the problem because it can only recombine traits that exist, and if all humans in the mating pool have high base age, it won't work.  

 Also, pre-human primates would probably increment the 'base age'.  

I am saying that probably there is another piece: 

                 On Human Embryonic development:

                              base age = 0


This would also suggest how to repair aging:

              trigger human embryonic development flags in cells taken from the patient, then trigger flags to differentiate the cells to the target stem cell, then reinject the stem cells where they go into the target tissue.  Example, bone marrow.

          Tissues that can't be repaired this way (the brain) you would have to slowly replace with artificial prosthetics, connected by neuro links.

It's a stochastic process, not a clock. One person gets an extra transposon copy at location A, another gets one at location B, sexual reproduction drops both 1/4 of the time.

That would work. Though why don't we observe lots of children suffering from aging.

It's possible that natural selection has historically kept the quantity of transposons down to small levels relative to the amount that one gains in non-gonad cells during aging. While this may change now that selection is relaxed, if the transposon suppression in gonads is good enough, it may take a long time. (and selection may not really be relaxed in our case, given our tendencies to late reproduction).

Solid reasoning.

Transposon activity is indeed believed to be repressed in the gonads to a much greater extent than elsewhere. I've also seen a few papers talking about health problems in the children of old parents, though I don't know as much about that.