I have promoted the idea that aging is programmed and that the program is epigenetic. Hence epigenetic age is fundamental. But what is it that imprints epigenetic age on the chromosomes and keeps it updated? Is the “methylation clock” responding to a higher authority, a separate clock which coordinates epigenetic age throughout the body? Do epigenetic clocks in different tissues talk to each other? Such questions are important not just for theoretical understanding, but also because they have two practical consequences. (1) Can we rejuvenate the body with system-wide signaling, or do we have to de-age cell-by-cell? (2) Can we be confident that if we set back the body’s methylation age the person will feel younger and live longer?
I have been reading and thinking about these questions for several weeks, and I can report no clear answers.
Is aging a cell-by-cell deterioration, or is it orchestrated at the level of the whole body and managed through signal molecules in the blood? If pressed, I think everyone would have to admit there is some of each, and differences within the community of aging biologists are about the relative importance of the former vs the latter.
One thing I think we ought to be able to agree on is that the system level, including signal molecules in the blood, makes a vastly more accessible target for anti-aging interventions. Repairing the body cell-by-cell is a daunting proposition; whereas modifying levels of signal molecules in the blood is a piece of cake, once we identify those molecules and determine their optimal youthful levels. The words “low-hanging fruit” come to mind, as well as “Pascal’s wager”.
If there are multiple, independent aging clocks, it is probable that the one that registers the oldest age is the one that can kill us, independent of the others. To make the big leaps in life extension that we are looking for, we probably will need to reset all the clocks. How much do the cell-level and system-level clocks talk to each other? How much progress can we expect to make by working at the (more accessible) system level without addressing the (more challenging) cell-level aging?
|Why is the preponderance of research devoted to aging at the cellular level? A small part of the explanation comes from scientific inertia; aging was understood in terms of increased cellular entropy for many years, whereas the paradigm of central control remained in a Russian backwater until publication of the Stanford parabiosis experiments in 2005. A larger part of the explanation has been the infusion of venture capital into aging research in the last decade. You can’t patent hormones and you can’t make money from rebalancing blood levels of the body’s native signal molecules. I believe that the profit motive has deeply corrupted aging research, as it corrupted medical research through the previous century.|
I am passionate about these issues, but I leave them aside to talk about questions of fundamental interest: Differential gene expression—epigenetics, and methylation in particular— seem able to change the body’s age state. It seems clear that gene expression is the primary way in which the age state of the body is transmitted and coordinated system-wide. But is gene expression the end of the line, the ultimate upstream aging clock? Or is there a “higher authority” that keeps track of time and programs the body’s methylation, etc accordingly? Does the epigenetic state of the body constitute an autonomous time-keeping mechanism, or is there a time-keeping reference clock, perhaps in the hypothalamus, which dictates the body’s age through secretions, and distant cells respond to these secretions by adjusting their methylation patterns?
And, if the answer is that methylation constitutes an independent clock, does that clock advance cell-by-cell independently, or does gene expression at the cell level export proteins that coordinate methylation age across the body?
I don’t have answers, but several experiments bear on these questions, and offer a nuanced outlook.
The practical question
We need measures of biological age in order to efficiently tell us when we are on the right track with an anti-aging intervention. Methylation clocks are presently the best technology we have for measuring biological age. So, can we be confident that if an intervention sets back the methylation clock that the intervention really is making a person (or animal) younger?
Reasons to think yes:
- Methylation clocks track chronological age better than any other biomarker
- Some of the difference between methylation age and chronological age is meaningful. It correlates with mortality. In other words, each of the major methylation clocks is a better predictor of life expectancy than chronological age. Remarkably, this is true even for the clock algorithms that were trained only on chronological age.
- There are theoretical reasons for believing that epigenetics is the primary driver of aging, so that methylation changes may actually be close to the causal nexus of biological age. (This conclusion is especially cogent for theorists like me who believe that aging is an evolutionary program; however, there are also prominent scientists in the field who reject programmed aging but embrace epigenetics as a primary driver of aging.)
Two things that could go wrong:
- There could be a higher authority, a centralized clock that sets up the methylation state. If this is the case, then setting back the body’s methylation age may be temporary, and the methylation state will revert to the age programmed by a central clock. (Cavadas and Cai have adduced evidence that aging signals are transmitted from the hypothalamus.)
- In the worst case, methylation changes with age could be an adaptive response when the body senses the accumulation of damage. In other words, the body changes its gene expression when damaged because it is working overtime to repair that damage. In this case, resetting methylation state to a younger age just makes the body less able to cope with the consequences of aging and actually shortens lifespan.
Evidence from parabiosis
In parabiosis, a young mouse is surgically joined to an older mouse of the same genotype. Tissues of the old mouse respond by becoming functionally younger. Since this 19th-century finding was brought to the modern scientific community, the search has been on for chemical factors in the blood that either promote aging or promote youth. [read more].
The parabiosis phenomenon and related findings in rejuvenation through blood plasma transfusions has led to a paradigm that says aging is coordinated throughout the body by signals in the blood. To the extent that single cells age, this is happening under central control, and the process can even be reversed if the cell is exposed to the right signals.
Evidence from bone marrow transplants
Bone marrow transplants are the most powerful available treatment for leukemia, and are also applied for some rarer diseases. The bone marrow comes with the epigenetic age of the donor, and thus the (white) blood cells subsequently generated by the transplanted bone marrow also carry age information. Several different studies [ref, ref, ref, ref] have found consistently that the white blood cells (and presumably the bone marrow from whence they came) retain the age signature of the donor. The donor may be younger or older than the patient. In either case, the methylation age of the patient’s white blood cells—post-op and for years afterward—remains keyed to the donor and does not correlate significantly with the patient’s age.
The lesson of parabiosis experiments was supposed to be that cell aging is not cell-autonomous, but rather a response to signals in the blood that instruct the cells what age to be. Young somatic cells could be aged rapidly in an old blood plasma environment, and — more impressively — old somatic cells could be made younger in a young environment.
Now we have a series of bone marrow transplant studies where the methylation age of the donor is the determining factor, not the patient into which the marrow was transplanted. Bone marrow contains the stem cells from which blood cells grow. Blood cells turn over every few months and they represent an accessible tissue sample which reflects the age state of the bone marrow in approximately real time.
“We found that the DNAm age of the reconstituted blood was not influenced by the recipient’s age, even 17 years after HSCT, in individuals without relapse of their hematologic disorder.” Soraas et al (2019)
This seems on its face to contradict our paradigm from parabiosis that says cell age is not cell-autonomous, but is programmed by the environment. How can we interpret the two results together? Some possibilities…
- Maybe only differentiated somatic cells are susceptible to age programming by plasma proteins, and not stem cells.
- Maybe these stem cells are providing the biochemical environment in the plasma. Maybe the stem cells and the white blood cells that they generate are the agents that secrete the plasma proteins responsible for sending age signals.
- Please think creatively about other possibilities.
Another result from these bone marrow studies
Consistently, the blood cells get older after transplant, whether they are transplanted from young-to-old or from old-to-young. This says two things. First, the point of comparison is the donor age, i.e., the age of the cells pre-transplant, and not the age of the patient who is associated with the systemic environment. Second, the cells seem to age rapidly after transplant, as measured by the methylation age. From there, the age of the cells may (in some studies) revert slowly to their original age trajectory over a period of several years.
Why the rapid methylation aging? It seems like a good guess that the rapid aging initially comes from high rates of reproduction in these transplanted cells that are generating a whole new source of much-needed blood. Could this be a link between telomeres and methylation age (which previously were found to be inversely correlated? Or is there another mechanism by which stem cells keep track of the number of times they have divided asymmetrically?
Am I the only one asking these questions?
Already a decade ago I was thinking about the question How Does the Body Know How Old It Is? Questions about time-keeping mechanisms and coordination of age information through the body go hand-in-hand with conceptions of aging as a programmed phenomenon, and perhaps the prejudice against programmed aging helps to explain the fact that few aging researchers are thinking in this way. A welcome exception is this article by Argentine gerontologists, which I was delighted to discover just yesterday. Lehmann et al: Hierarchical Model for the Control of Epigenetic Aging.
Although there is evidence suggesting that the cellular epigenetic clock possesses an intrinsic ticking rate [ref, ref, ref] multiple observations at organismal level in humans and other mammals lead to the inference that in vivo, the ticking rate of the clock in tissues is synchronized by a master pacemaker.
Lehmann cites as prima facie evidence for this
For a given chronological age, it was found that in DNA samples taken from whole blood, peripheral blood mononuclear cells, buccal epithelium, colon, adipose, liver, lung, saliva, and uterine cervix, Horvath’s algorithm read essentially the same epigenetic age, the only exceptions being some brain regions and very few other organs.
In addition, she cites Katcher’s success in rejuvenating rats (and their diverse organs) using only a set of intravenous signals. The article goes on to propose a model in which there are four time-keepers in the body, coordinated by signals in the circulatory systems. The four are:
- Light-sensing and neural processing
- Neuroendocrine signaling (esp the suprachiasmatic Nucleus of the Hypothalamus)
- The Immune system, including thymic involution
Curiously, she does not include the replication counter implicit in telomere shortening, which Fossel, Blasco and other luminaries have adduced as the primary source of aging. I also would add that the hypothalamus is the best candidate we have, not just for one aging clock among several, but as a central, coordinating organ.
Fig. 1. Proposed organismal regulatory network in mammals. The diagram includes the autonomic nervous system (ANS, acting via neurotransmitters), the neuroendocrine system (NES, acting via blood-borne hormones), the immune system (acting via blood-borne cytokines and thymic hormones), the circadian clocks (acting via blood-borne hormones and neurotransmitters) and a putative pathway connecting the neuroendocrine network to the DNAm clock in organs and cells. All networks act on peripheral organs. Inset- Bidirectional interactions among all networks including (in red) the hypothetical DNAm network.
In addition to Lehmann, there is a 2021 review by Raj and Horvath, speculation from the horse’s mouth. They note that all the Horvath clocks are based on small differences in % of cells methylated at a given site (conventionally notated as β).
Increase in epigenetic age is contributed by changes of methylation profiles in a very small percent of cells in a population.
One way to interpret this fact (my speculation, not R&H) is that immune sensitivity, (anti-) oxidation, and inflammation are all under tight homeostasis in the body, because these are sensitive functions, balanced on a knife edge between insufficient protection and self-destruction. It is easy to tip the balance over toward self-destruction with small changes in the set point for a few signal molecules in blood plasma.
Another way to interpret this (again, my speculation) is that it is only a handful of cells at the tail end of the distribution that go over an edge into a state where they cause all the damage of aging. This hypothesis is consonant with the story about short telomeres, cell senescence, SASP, and the powerful benefit of senolytics. However, a big hole in the narrative is that it requires a set of CpG’s that would be capable of precipitously tipping the cell over into a toxic state. We know that critically short telomeres can do this, but there is no study yet of methylation-induced cell senescence. R&H speculate about such a mechanism connected with PCR=Polychrome Repressive Complex.
Raj and Horvath also stress the continuity between epigenetic changes that begin in utero, associated with development, and the changes that lead to senescence late in life. Blagosklonny as well has emphasized this point.
“Collectively, these five features of DNA methylation allow one to summarize with some degree of certainty that epigenetic ageing is a measure of change of epigenetic heterogeneity, contributed by a relatively small percentage of cells, seemingly in line with developmental processes that are conserved across species and begins very soon after conception. This seemingly inescapable deduction provides us with a reference point against which models and hypotheses can be measured.”
If I may carry the logic of these two experts one step further, I would emphasize the role that methylation has in determining what hormones and enzymes are secreted into the blood. Therein lies the possibility that intracell methylation clocks are coordinating, both with other cells and with other clocks, via signal molecules in blood plasma.
Other provocative findings that we might hope to integrate into a theory of aging
The methylomes of naked mole rats age at a normal rate, but the phenotypes of the rats themselves show no signs of age [ref]. Males and females age epigenetically in somewhat different ways [ref]. Methyl donor molecules in the diet can lead to a younger methylome, with benefits both for hyper- and hyomethylated regions (validated for MTHFR snps only) [ref]. When human fibroblasts are reprogrammed (with RNA) to turn them into neurons, they remember their Horvath age even after forgetting their identity [ref]. BMI is associated with accelerated methylation aging [ref]. Mice challenged with a high-fat diet can be brought back to normal weight with a normal diet, but accelerated methylation aging persists [ref]. Cessation of smoking decreases Hannum and Horvath DNAmAge [href]. The methylation shadow cast by years of smoking is a better predictor of subsequent morbidity and mortality than the smoking history itself [ref]. Methylation image of telomere length is a better prediction of age and mortality than is telomere length itself [ref]. Pregnancy increases Hannum Age, DNAmAge, and PhenoAge [ref].
(Apologies to Rafil Kroll-Zaiti)
Katcher has been conducting a longevity trial for rats treated with E5 (background story here). Partial results suggest that treated rats are living statistically longer than untreated, but not as much as you would expect if the greatly reduced methylation age indicated full rejuvenation. The results are preliminary, and I will publish a full analysis in this space as soon as I can get the detailed dataset.
The finding, if validated, suggests that multiple clocks in the body are not completely synchronized, and the “fastest clock wins”, meaning that it kills the animal no matter what the other clocks may say.
I am disappointed as you are in not being able to provide fundamental answers, but I hope that (together with Lehmann, Goya, Raj, and Horvath) we have provided a framework and a set of questions that can guide fundamental research. Very few other researchers are addressing these questions, and the answers will be crucial both for devising effective interventions and also for measuring the effectiveness of interventions that we already have.