Computational Aging

I listened to a number of talks at the Mechanisms of Aging conference at Cold Spring Harbor Laboratory (CSHL) this week [1].

There were four back-to-back talks today that were especially promising for epigenetic reprogramming:

1. PRC2 clock—A universal epigenetic biomarker of aging and rejuvenation by Vittorio Sebastiano (Stanford University). Vittorio shared data supporting a new epigenetic biomarker of age, which is the DNA methylation of PRC2 in lowly methylated regions [2]. They found that this single measure could account for 90% of the DNA methylation age from other clocks pulling on many regions. The benefit of this clock is that is interpretable (measured directly) and without model bias (no training). It also suggests alignment with an underlying epigenetic mechanism, as PRC2 catalyzes histone methylation. Interestingly, Vittorio found in mice that rapomycin did not lower the PRC2 clock "age", while caloric restriction did.

2. Reprogramming to recover youthful epigenetic information and restore vision in age-related macular degeneration (sic) by Yuancheng Lu (Sinclair Lab, Ksander Lab, Margerete Karg; Harvard University). Yuancheng was the first author on the incredible Nature 2020 paper that showed youthful DNA methylation patterns, axon regeneration and vision restoration in mice with optic nerve crush and mice with a model of glaucoma [3]. This talk reiterated that data and then shared new data on similar results in a mouse model of macular degeneration. OSK were again the reprogramming factors used. No evidence of terratoma/cancer were found after a period of observation, though this bears more study. Most encouraging is that epigenetic reprogramming 1) works in a new use case (macular degeneration) and 2) did not lead to cancer. A paper is forthcoming.

3. Epigenetic reprogramming reverses neuronal aging and improves cognitive performance by Xiao Tian (Sinclair Lab, Harvard University). Xiao built on Yuancheng's work with a focus on the brain. There have been concerns that the blood-brain-barrier would make it difficult to deliver epigenetic reprogramming to the brain. Xiao & team created a new AAV to address this. They showed powerful benefits in the brain for memory and vs. old mice and mice with a model of Alzheimer's. Once again, this provides a powerful new use case. A paper is forthcoming.

4. Transcriptomic reprogramming screen using functional RNA clock for cellular rejuvenation by Alex Plesa (Church Lab, Harvard University). Alex pointed out that reprogramming been focused on the OKSM Yamanaka factors for over a decade now. They wondered if there might be other factors that work as well, that might be safer or have other complementary effects. Using a functional RNA clock (vs. a DNA methylation clock), they identified several candidates. SRSF1, a protein splicer and regulator of transcription, was the most promising in their data. When will we see SRSF1 tested in experimental models of aging? A paper is forthcoming.

Together, these build support for the promise of epigenetic reprogramming in treating aging. One of the most important things for long-term aging science is to identify the right paradigm for the core drivers of aging. Epigenetic reprogramming seems to address aging by rewiring gene expression to a more youthful state. This seems to align with a paradigm where epigenetic unwinding leads to loss of protein homeostasis, which then leads to cellular and tissue dysfunction. The talks above shared new data points in new situations (macular degeneration, Alzheimer's, etc), which shows that reprogramming might be fundamental across cells and tissues. It also points out that there may be other or complementary reprogramming factors, like SRSF1, that could overcome obstacles with OSK (or OSKM).

What comes next? We need to see if epigenetic reprogramming can work in vivo across tissues and lead to better health and longer life. This will require continued work, like Xiao's, to find ways to get the factors safely to their target. It would be especially convincing to see this across multiple organisms (worms, flies, mice, rats, primates?), to prove it is a conserved feature. How long will it be before we have a mouse living for 5+ years with epigenetic reprogramming?

References:

[1] Full list of abstracts for the 2022 Mechanisms of Aging conference at CSHL: https://meetings.cshl.edu/abstracts.aspx?meet=AGING&year=22

[2] Moqri et al. PRC2 clock: a universal epigenetic biomarker of aging and rejuvenation. bioRxiv (2022). doi: https://doi.org/10.1101/2022.06.03.494609

[3] Lu, Y., Brommer, B., Tian, X. et al. Reprogramming to recover youthful epigenetic information and restore vision. Nature 588, 124–129 (2020). https://doi.org/10.1038/s41586-020-2975-4

Let's start with the big goal that humanity should work to achieve significant lifespan and healthspan extension by 2063. That gives us 41 years, with no time to spare. How can we make sure we're on the right track? How do we know we're working on the right things?

Here are some questions that could help guide a long-term aging research program:

1. What are examples of research programs that made major progress on an intractable disease, e.g. multiple sclerosis? Why did it work? What hasn't worked? What can we recreate for aging science?

2. What does a multi-decade research program look like to significantly extend human lifespan and healthspan? What are some examples of this done well in aging? Done poorly?

3. What is the right "paradigm" for aging science? Thomas Kuhn's The Structure of Scientific Revolutions argues that identifying the right "paradigm" or high-level lens for a problem is critical. Until you get the right high-level paradigm for a problem, researchers can be focused on the wrong things. For aging, there is not consensus on what causes it. We have 9 "hallmarks" of aging, but what are the drivers?

One tantalizing theory is the "information theory of aging." It goes something like this:

1. DNA damage => epigenetic unwinding
2. Epigenetic unwinding => loss of transcriptional control
3. Loss of transcriptional control => protein noise
4. Protein noise => cellular dysfunction
5. Cellular dysfunction => tissue dysfunction
6. Tissue dysfunction => increased risk of disease and death

Is this right, or at least close enough? What would be true if this theory were right? For example, do organisms with slower aging have slower DNA damage and less epigenetic unwinding? Does each item in the chain hold, e.g. if you reduce DNA damage, do you get less epigenetic unwinding, etc.

If we can settle on a good paradigm, we could focus our efforts.

4. How would we know we're on track? Can you even predict what the benchmarks might be? Science can feel like you're wandering in the woods, and it's hard to know if you're making real progress or not.

5. Pre-mortem: How might we fail? In other words, what are things that could have gone wrong such that we don't make major progress in aging science over the next 40 years? One interesting example of this is the case of Alzheimer's disease research. Karl Herrup's How Not To Study a Disease: The Story of Alzheimer's argues that an excessive focus on the amyloid beta hypothesis came at the expense of other inquiry and led to a herd mentality among funders and researchers. Are there other similar examples?

6. How do we answer questions 1-5? I expect it will take a while to form a good perspective. I'll start by reading a lot of scientific histories. There are good books written about many of the major scientific discoveries (E.g. James Watson's The Double Helix). I'm working down a list, but please share any recommendations you might have. Reading the seminal papers on each field (e.g. aging, Alzheimer's, Parkinsons, spatial transcriptomics, deep learning, etc) help give clues to how intellectual progress happens. I'm also asking these questions of people who might know. UCSF and the SF Bay Area is a great place for this, and I've learned a lot from the coffee chats and lunches with top researchers.

Maybe science doesn't proceed in this systematic of a manner? Still, if we answer these questions I believe we will find better approaches to aging science.

While there are tailwinds for aging science, there are also many unanswered questions, barriers and headwinds.

Here are a few issues:

1. Lack of understanding of root cause of aging. We have 9+ hallmarks of aging, but what drives aging? Until we have a clear answer on the root causes, we will be shooting in the dark.
2. Decades of effort on aging science with no "life extension" drugs on the market. The Mediterranean diet might be the closest thing we have so far. 
3. Decades of effort on Alzheimer's with little to show for it. Alzheimer's is narrower than aging and therefore seems like it should be easier to make progress on. Yet, Alzheimer's patients have few options.
4. Slow progress in model organism life extension. There are examples, mostly of marginal life extension in lower organisms, but we aren't seeing mice live for 2x their typical lifespan. (Encouragingly, a June 2022 study in Nature Communications Biology showed Drosophila melanogaster living to >200 days, a 120% increase vs. average lifespan, with a combination of interventions [1].) 
   
5. Not a national/global priority. Only a tiny sliver of our population works on aging science.
6. How do you run clinical trials? Healthy humans live to be 80+. How do you prove an intervention significantly extends lifespan in a reasonable amount of time?
   

These issues raise concerns about our ability to make progress on aging science, especially if we want significant lifespan and healthspan extension in the next 40 years (which is my goal). Most importantly in my view, we need to get clear on the drivers of aging at the molecular, cellular, tissue and organism level. Once we know the drivers / root causes, it will make it easier to create therapies.

[1] Shaposhnikov, M.V., Guvatova, Z.G., Zemskaya, N.V. et al. Molecular mechanisms of exceptional lifespan increase of Drosophila melanogaster with different genotypes after combinations of pro-longevity interventions. Commun Biol 5, 566 (2022). https://doi.org/10.1038/s42003-022-03524-4