I’m really happy when a plan comes together, which means when my general plan of “talk about science journalism and cut through some of the hype” converged with a new year’s resolution plan of “summarize papers regularly on MAL”, I was thrilled. There’s an article recently published in Cell that’s actually quite cool, and it’s being picked up in the popular press with only the usual dose of hype, although the first few articles I read certainly made me raise an eyebrow. Which means it’s time for a science lesson! Woo!
How We Understand Aging
For a few decades starting in the 1970s, the best analogy for the reason that animals age the way they do was, basically, rusting. This is called the Mitochondrial Free Radical Theory of Aging, and it goes like this: Oxygen is a very reactive substance. Mitochondria (often called the power plants of the cell, responsible for most metabolic processes) use lots of oxygen to power the cell. But you can’t really control oxygen 100%. Some reactive oxygen and oxygen-like compounds will float around unused, and these can react with whatever they bump into, and break (rust) that component. A rusted-through component of a cell doesn’t work, and enough of these reactions can make the whole cell — and by extension the whole organism — sort of rust over. Like a tractor.
This theory worked great at first, and it fits common sense fairly well. Because it’s easy to see how a tractor ages: it rusts. And that’s not something that can be easily undone. It makes sense that maybe that’s what causes us to age as well. Plus, it jived with a few classic observations about how animals age, namely:
(1) Mitochondria die first. This makes sense if being bombarded by oxygen is what’s causing a cell to age, because there are lots of oxygen compounds in the mitochondria, and there are fewer elsewhere.
(2) Eating less means you live longer (so long as you don’t starve). This is called “caloric restriction” and it’s one of the most counter-intuitive and yet reproducible observations in the study of aging. But it makes a kind of sense under the rust analogy, because eating less means having less to digest and metabolize, which means a slower metabolism and less active oxygen bumping around in mitochondria.
But the problem is, that theory just isn’t true. We’ve basically disproved it. In a bunch of ways:
(1) Having fewer reactive oxygen species doesn’t make you live longer. This is the “antioxidant” idea I talked about a while ago. It just doesn’t track. Often, instead of having any effect on aging as such, too many antioxidants can act as a poison.
(2) Having more reactive oxygen species doesn’t shorten lifespan. The one exception is when they are obviously poisonous — but in those cases there’s an obvious cause of death, an obvious organ failure, rather than a general “accelerating aging” kind of deal. Sometimes having more reactive oxygen species can even make an animal live longer, which just does not jive with this rust analogy at all.
(3) Long lived organisms don’t generally have fewer reactive oxygen species. When you look at, say, a naked mole rat (which I recommend you do, because who doesn’t want to do that?), which lives about ten times longer than other mammals of similar size, you would perhaps expect under the rust analogy that there would be less oxygen in their mitochondria. But that’s not true, or it’s not reliably true in most instances.
So there was an open question: if we’re not rusting, why do mitochondria die first? Why does eating less make animals live longer? What’s the link between the two? And how do scientists actually start figuring something like that out?
If It Looks Like A Duck
A huge portion of biology (and especially genetics) is determined by carefully breaking things. One of my genetics professors once likened it to understanding how a computer works if you had a thousand computers and one tiny hammer. You can start by asking “I wonder what that does” — pick a component, break it, and see what happens. That’s called reverse genetics. Or you can start by asking “I wonder what’s involved in this action” — break things, at random, until you get the error you’re interested in, and then work to narrow down which thing you broke actually threw the error. That’s called forward genetics.
But in either case, we’re essentially saying “Gene X is involved in trait Y because when gene X is broken, it looks like Y.”
This works great when you have a really specific trait that you’re studying that is independent of other traits: for instance, hair color. If you mutate a gene and get a mouse with white hair, it’s likely that that gene was required for darkening mouse hair. If you mutate two genes and they each independently turn the mouse white, they’re probably both required for hair pigmentation, and they’re probably — therefore — related.
That’s a big part of the beginnings of studying a genetic pathway. But all we’ve really shown is that the fail conditions of each of these genes are the same. This doesn’t actually necessarily mean that they act together. And if your trait is something that can have side-effects, or is too vague (like “dead”, there are lots of ways to die), sometimes it’s really hard to see this equivalence and therefore really hard to figure out what’s actually in the same pathway and what’s just coincidentally similar.
So I think of this as the “looks like a duck” principle of biology: a big hint as to what you study has to be based on appearances, hints, and likenesses.
In this study, the basis of the research was definitely based on a “looks like a duck” hypothesis: they saw a gene, SIRT1, and when that gene was broken, mice seemed to age faster. More importantly, when mice had too much SIRT1, they showed a list of benefits in common with caloric restriction. So it stood to reason that SIRT1 had something to do with aging, and caloric restriction, since the fail conditions looked so similar.
This principle comes up again and again within the paper, too; it’s how the authors choose most of their candidate genes and molecules to study: something which usually changes over the course of a mouse lifespan, and something which when messed up causes changes in mouse aging. And in order for this reasoning to be strong, the authors had to use a very specific phenotype (not just “old”) — in particular, they looked for the presence of actively growing mitochondria, which are one of the first things to disappear as muscles age.
A Pathway is an Assembly Line
The idea that to understand something you break each component separately and in combination, and observe, has a couple complexities that make it particularly useful when components are actually related. That’s because genetic pathways are often kind of like molecular assembly lines: changing one thing will mess up everyone else, in a different way.
This is the kind of genetic theory that often messes up students in graduate level genetics classes. So I’m going to use a couple examples from the paper, and hopefully it’ll make more sense afterwards.
Using basically the descriptive studies I described above, researchers figured that nuclear NAD+, SIRT1, VHL, c-Myc, HIF-1a, PGC-1a, and TFAM were all in a pathway together. Already they knew a few things about the function of each of these: NAD+ is a reducing agent that SIRT1 uses to alter protein structures, so it’s a required cofactor of SIRT1. VHL tags other proteins for degradation. HIF-1a and PGC-1a are transcription factors: they push transcription of other genes. And they both mess up metabolism when they’re broken. So is c-Myc, although when it gets messed up cells mostly just turn into cancer. And TFAM is a master regulator of mitochondria, so it’s probably at the end of the line, since we’re looking for things that are responsible for mitochondrial health.
So we know that NAD+ and SIRT1 are at the beginning of the assembly line, and we know that TFAM is at the end of the assembly line. And somehow, VHL, c-Myc, HIF-1a, and PGC-1a fall in the middle.
The first test is to verify that all of these guys are acting in the same pathway (assembly line). Which would mean, if one of them is broken, then no matter what we do with SIRT1 (add in more, take it out, whatever), we won’t get TFAM and we won’t get healthy mitochondria.
This is true for everyone except PGC-1a, so PGC-1a is probably part of a different assembly line (even though it does respond to SIRT1 and it does activate mitochondria).
Obviously, you can use the same idea to tease out where each factor is in the pathway, since the presence of each protein and the phenotype in the cell give you a read-out. So, the researchers have observations like this: In the absence of SIRT1, HIF-1a levels go up. In the absence of VHL, HIF-1a levels go up. If you have lots of SIRT1, but no VHL, HIF-1a levels still go up; if you have no SIRT1 but lots of VHL, HIF-1a levels go down. So SIRT1 acts on VHL which acts on HIF-1a.
Using lines of reasoning like that, the researchers pieced together this entire pathway from NAD+ levels in the nucleus to transcription in the mitochondrion:
In addition, the researchers used information about direction of effect to determine whether these proteins facilitated eachother’s action or repressed it. If they increased SIRT1, or if they increased VHL, they got similar phenotypes. But when they increased HIF-1a, they got an opposing phenotype (older-looking muscle). So SIRT1 and VHL repress HIF-1a. You can follow this up on a molecular level — increasing SIRT1 or VHL lowered levels of HIF-1a. Adding that information into the flowchart above gives you something like this:
That’s the big result of the paper. And it leads to the even bigger result of the paper — if all of this works, then adding in NAD+ could rejuvenate muscle. Which it did, and which lots of people got really excited about. But it kind of skips over all the detailed and interesting biology, which made me a bit sad.
What Does it Mean?
First, this suggests a link between caloric restriction and mitochondrial function that isn’t the “rusting” hypothesis I talked about at the top. Caloric restriction buffers levels of these proteins, in particular SIRT1, which helps mitochondria stay healthy over time. That’s because SIRT1 is responsive to insulin, among other things.
Second, the SIRT1/HIF-1a/c-Myc link is a very cool piece of the puzzle of why cancer becomes so much more likely over time. If mutation rate was absolutely purely random, we would expect the risk of cancer to grow more slowly than it does. But HIF-1a is mostly known for its association with cancer. c-Myc is an oncogene: something which, when messed up, causes cancer. If aging causes an accumulation of HIF-1a and a misregulation of c-Myc, this would make it much more likely for cancer to develop. (In particular, HIF-1a pushes cells to not use the oxygen that they have, which in turn makes it more likely for that oxygen to react with important cellular components and, perhaps, cause mutations which can lead to cancer.)
Of course, like I said above, most people are probably most interested by the idea of giving grandpa a vitamin pill that would make him as hale and hearty as his own kids (maybe not his grandkids). There’s some reason to be hopeful: the effect in mice was truly remarkable. And certain traits that are known to accelerate the weakening and degeneration of muscles (things like a high fat diet and diabetes) also recapitulate the things the researchers saw in mice. HIF-1a can mess up mitochondria in a lot of tissues. But there are still a lot of pieces we don’t know. SIRT1 was really only active in mouse muscle. Which doesn’t necessarily mean that NAD+ wouldn’t work — maybe there’s another SIRT that’s controlling this same pathway in the liver, or the brain, or any other tissue. But it means that there’s a lot more we’re not certain about, even in mice. And something working in mice really rarely means it’ll work in humans, much less without major side effects.
Especially since NAD+ is used in a huge number of pathways. Like I said in the title and at least once before in this article; it’s something that you first hear about around the same time that you hear about ATP. Adding NAD+ into cells artificially by feeding them NMN may work great, but in reality the effects of NMN as an ingested drug haven’t been studied (as it says in big bold letters on the MSDS). NAD+ is two steps away from vitamin B3, though. So that suggests possible side effects to this “NAD+ for better strength” idea (vitamin B3, in high doses, causes uncomfortable rashes, flushing, and occasionally gastrointestinal problems). And it also, perhaps, gives you a hint towards long-term muscle health? If you want to go with the hype of the latest bleeding edge research, that is.