The Peter Attia Drive - #281 ‒ Longevity drugs, aging biomarkers, and updated findings from the Interventions Testing Program (ITP)

00:00:00.000 Hey, everyone. Welcome to the drive podcast. I'm your host, Peter Atiyah. This podcast,

00:00:16.580 my website, and my weekly newsletter all focus on the goal of translating the science of longevity

00:00:21.580 into something accessible for everyone. Our goal is to provide the best content in health and

00:00:26.780 wellness, and we've established a great team of analysts to make this happen. It is extremely

00:00:31.720 important to me to provide all of this content without relying on paid ads to do this. Our work

00:00:37.000 is made entirely possible by our members. And in return, we offer exclusive member only content

00:00:42.760 and benefits above and beyond what is available for free. If you want to take your knowledge of

00:00:47.980 this space to the next level, it's our goal to ensure members get back much more than the price

00:00:53.260 of the subscription. If you want to learn more about the benefits of our premium membership,

00:00:58.080 head over to peteratiyahmd.com forward slash subscribe. My guest this week is Dr. Rich Miller.

00:01:06.920 Rich was a previous guest from all the way back in February of 2021. And that was such a remarkable

00:01:13.220 episode that I knew at that time we were going to have to do another one. And I can tell you now,

00:01:17.500 we're going to do a part three at some point as well. Rich is a professor of pathology at the

00:01:22.380 University of Michigan, where he is also the director of the university's Paul F. Glenn

00:01:27.500 Center for Biology of Aging Research. He is also one of the principal architects of the

00:01:33.480 Interventions Testing Program, or ITP, created to evaluate potential life-extending interventions

00:01:40.200 in mice. Rich received a bachelor's degree at Haverford College and then went on to earn an

00:01:46.020 MD and PhD at Yale, followed by postdoctoral training at Harvard and Memorial Sloan-Kettering.

00:01:51.640 Now, you have no doubt heard me talk about the ITP, not only in the first episode with Rich,

00:01:57.480 but it seems to come up all the time when we talk about gyroprotective molecules. Again,

00:02:03.100 what does gyroprotective mean? Gyroprotective means molecules that extend lifespan, but not

00:02:10.220 through targeting a very specific disease process, but rather by targeting the hallmarks of aging.

00:02:16.320 So in this episode, we talk about the ITP, not in as much detail as we did in the first episode,

00:02:21.800 because if you really want to understand that, you can go back and listen to it. But for those

00:02:25.520 who don't remember or haven't listened to the first episode, we certainly cover enough here so you can

00:02:29.960 understand it, what its purpose is, how its mouse model is significantly different and demonstrably

00:02:36.280 better than all the other mouse models used out there, how the studies are conducted,

00:02:40.460 what the metrics of interest are, how drugs are dosed, delivered, and more. We also talk about

00:02:46.480 how the ITP looks at healthspan, not only lifespan. We cover notable successes from the ITP,

00:02:53.340 including rapamycin, 17-alpha-estradiol, acarbose, canagaflozin, and a few others of late.

00:03:01.160 We talk about some of the most recent successes, including one that absolutely blew my mind,

00:03:07.400 meclizine, which is an over-the-counter drug used to treat seasickness. Additionally,

00:03:12.460 we do a deep dive into the idea of biomarkers of aging and what we know about various aging rate

00:03:19.040 indicators. I actually found this to be the most important and interesting part of the discussion

00:03:23.380 for me personally, because I'm quite steeped already in some of the drug stuff. We end the

00:03:28.260 discussion speaking about some of the most notable failures, including nicotinamide,

00:03:32.380 riboside, metformin, and resveratrol. So without further delay, please enjoy my conversation with

00:03:38.480 Dr. Rich Miller. Hey, Rich. Great to sit down with you again. I don't remember exactly when we did this

00:03:49.120 before, but I know I enjoyed it thoroughly. And it's actually one of the few podcasts I've gone back

00:03:53.880 and listened to. I don't often, for obvious reasons, go back and listen to podcasts that I've

00:03:57.940 recorded because I already heard them. But there was so much in that one that I've at times gone

00:04:02.480 back and listened to parts of it. So excited to sit down and chat with you again. But assuming that

00:04:07.800 maybe people who listened to us in the past, or maybe even aren't familiar with the ITP,

00:04:12.120 I think it's always great to start with sort of an overview of what the interventions testing

00:04:15.900 program is. I certainly refer to it a lot in both podcasts and even in things I write.

00:04:21.380 Sure. It was developed by the National Aging Institute under the leadership of Huber Warner

00:04:28.120 about 20 years ago. We are just now finishing our 20th year. We've sent in applications. So if the

00:04:35.260 peer reviewers like it, we may get five more years of funding. It represents work being done by three

00:04:41.520 different research laboratories. Mine at the University of Michigan, Randy Strong's at the University of Texas

00:04:46.960 Health Science Center at San Antonio, and a program at the Jackson Labs where David Harrison

00:04:52.020 was in charge. David will be stepping down next April, and he will be replaced by Ron Corstanja

00:04:58.620 as the first new appointment for the ITP leadership at the Jackson Labs. What we try to do is quite

00:05:06.460 simple. We try to find drugs that will slow aging and extend mouse lifespan. We have a national

00:05:13.680 announcement every year, international. Anyone who wants to suggest a drug sends us an application and

00:05:19.300 they tell us why they think we should test their drug and why they think it will be good and not hurt the

00:05:24.020 mice. We have a committee that evaluates those and then we pick five or six or seven each year to see if

00:05:31.700 indeed giving them to the mice will give them a lifespan extension. We've had four published

00:05:38.440 significant hits and another two or three that are significant, really small, and another two that

00:05:43.900 are in press that should be, I hope, accepted soon. So this gives us a range of successful drugs, and we

00:05:51.400 can then, and we do, try again. We give them to mice at varying doses to see if it's dose sensitive.

00:05:57.400 We look at the pathology. We make tissues that we can give away to other investigators for collaborative

00:06:03.020 studies, and we try to reason about mechanisms of aging and control points for aging based upon which

00:06:10.980 drugs work and which drugs don't work. A lot of people see this, and I can understand that. I agree

00:06:16.640 with this as a stalking horse for the important goal of finding drugs that would extend lifespan by slowing

00:06:23.140 aging in people. That is an important element, but there are many steps between a mouse drug and a human

00:06:28.940 drug. The other major things that our program does is it really gives us a lot of insight into the

00:06:33.540 biology of aging, which in the moderate term should give us many clues as to what to look at that may

00:06:39.960 be successful. There's a lot in there that I want to talk about, Rich. I think I'll start with just a

00:06:45.220 couple of simple questions. First, in a given year, how many candidates do you typically get nominated?

00:06:50.880 It varies a lot from year to year. This year was our winner. We had 28 suggestions, and we only have

00:06:56.820 enough money to do five or six or seven of them. In a typical year, it's 10 to 15, from which we pick

00:07:02.480 about six annually. Sometimes we fill up those slots with things that we want to do. For instance,

00:07:08.940 we found a couple years ago a drug Captopril, which is FDA-approved in people for blood pressure,

00:07:15.100 to my surprise, gave a really small increase in lifespan in mice. Mice don't die of hypertension.

00:07:21.980 They don't get strokes, etc. So I was betting against that, but it was a really small effect.

00:07:26.520 So we decided in the current year, try Captopril again, but at a higher dose. Maybe it'll work,

00:07:33.140 give us a nice big effect if we use a higher dose. Some of those slots each year then are

00:07:37.100 taken up with other doses, other dosage forms. Often if a drug works when you give it to young

00:07:43.480 adults, we say, okay, great. Now let's test it in middle-aged mice. Everyone would like to know,

00:07:49.200 as we would, of course, whether a drug would only work if you give it to young adults. We would love to

00:07:54.040 find drugs that work in middle age as well. So that's often fills up one of those available slots.

00:08:00.060 We'll definitely talk about a couple of those drugs. Tell me what the budget is. What does

00:08:03.640 the NIA provide to the three laboratories? They give $1 million a year to each of the three sites

00:08:09.600 in direct costs. The actual cost to the taxpayer is probably about 50% more than that because each

00:08:16.400 university will also receive indirect costs to pay for the building and the heat and the police force

00:08:22.740 and the library and the president and all of that stuff. But it's basically sub $5 million a year.

00:08:28.400 About $4.5 million total per year, of which $3 million actually goes $1 million per year

00:08:35.020 direct cost to each lab. Yeah. Yeah. So again, a relatively paltry sum of money when you consider

00:08:41.760 the insights that I think come out of the ITP. So maybe that's my way of lobbying for the budget

00:08:47.040 being increased. I agree. I do not think we get too little money. The first 10 years,

00:08:52.800 we had half a million dollars and the NIA thought of us, I'm pleased to say, as something that was

00:08:58.360 really working well and doing good stuff. So 10 years ago, they doubled our budget, which I think

00:09:03.920 is a sign of their endorsement and their ability to recognize good stuff. And it certainly made us feel

00:09:08.740 good. And there are 17 kind of divisions within NIH, of which NIA is one. What is the NIA's annual

00:09:15.960 intramural and extramural budget? Yeah, I don't know. I'd have to look that up. It's actually one

00:09:21.640 of the larger institutes now. It didn't used to be. That's misleading because more than half of their

00:09:27.280 budget goes to Alzheimer's disease. They have, through a variety of negotiations, been designated

00:09:33.560 sort of the lead agency for Alzheimer's, there are good reasons for wanting to spend money on

00:09:38.480 Alzheimer's research. And all of that goes through the National Institute on Aging. So their budget is

00:09:43.680 big, but their budget for biology all put together is only one sixth of the NIA budget. And for the

00:09:52.020 kind of biology I care about, it's much, much less than that. A lot of the good biology is what happens

00:09:58.360 to bone aging, what happens to eye aging, what happens to aging of the immune system. That's interesting

00:10:03.780 research. But of course, the kind of stuff I care most about is aging as a global phenomenon. What can you

00:10:09.700 do to slow aging? And how is it that aging increases your risk, basically, of almost everything that you

00:10:16.820 don't want to happen to you? That part of the NIA budget is small.

00:10:22.680 One last question on budgeting. Is there an opportunity for philanthropic giving to plus

00:10:30.560 up the NIA contributions? In other words, the $3 million in direct costs or the million to each

00:10:36.760 site, is there an opportunity for those numbers to go up with donations?

00:10:42.060 I don't believe that NIA accepts donations. However, a philanthropist, should he or she be listening to

00:10:50.320 this podcast, can certainly set up independent arrangements. For instance, if they wish to have

00:10:56.260 support for all three sites in the ITP, one can imagine a situation in which a foundation makes

00:11:02.520 awards. The universities do have the flexibility to take gifts and target them to specific research

00:11:08.380 groups or specific research projects, either independently or as a consortium.

00:11:12.560 Okay, let's talk a little bit about mice and men. Let's just talk about mice, actually.

00:11:19.460 So one of the hallmarks of the ITP is the mouse model that is used and how it differs from some of

00:11:28.120 the more typical mouse models that, shall we say, run rampant in biomedical research. Maybe tell us a

00:11:35.700 little bit about what the standard off-the-shelf mouse model is, where it came from, and maybe some of

00:11:42.540 the problems or limitations associated with that. 97%, the last time I checked, of requests for

00:11:48.900 aged mice to the National Aging Institute were for the same kind of inbred mouse. Its formal name is

00:11:55.580 C57 black 6, and everybody calls it the B6 mouse. So these are the standard mouse, and it's a really bad

00:12:03.760 thing for science, not just aging science, but science in generally that relies on an inbred mouse.

00:12:09.060 There are several problems. One is that it's a single genotype, and it has been shown many times

00:12:15.160 now that if you have a drug that works in black 6 mice, it might work in another kind of mouse. It

00:12:20.920 might not work. It might have the opposite effect in another kind of mouse. There are good, strong

00:12:25.800 papers on those issues. So people study the black 6 mice in the mistaken belief that it's sort of like

00:12:32.660 mice in general, despite the now really quite convincing evidence that it isn't. So the ITP from the word

00:12:40.580 go made a decision. It was controversial, but in retrospect was a really good decision instead to use a

00:12:48.020 genetically heterogeneous mouse. The particular kind of mice we use is called UMHET3. UMHET3, UMHET3, that's where

00:12:55.120 it was first derived, and HET is for heterogeneous. These are mice essentially which have the same

00:13:01.680 set of grandparents. So any two mice in our population share half of their genes, just like

00:13:07.760 you would share half of your genes with a brother or sister, but it's a random half. So if we have two

00:13:14.000 mice, we don't know which genes they'll share, though we know it'll be half of them, and half of the genes

00:13:18.960 will be different. The advantage of the system is you can make as many of these mice as you want

00:13:25.040 anywhere in the world at any time. Year after year after year, you'll get the same population

00:13:31.120 characteristics. No two mice are identical, but all populations of UMHET3 genetically are identical with

00:13:39.040 one another. So it's a form of reproducible heterogeneity. And this way, if we had by chance

00:13:45.760 tested a drug that worked in Black 6 and only tested in Black 6, we really wouldn't know whether

00:13:51.440 it would work in any other stock. And if we had tested a drug that failed to work in Black 6,

00:13:56.160 we would have fooled ourselves into thinking that it was a loser drug. Since there are thousands,

00:14:02.160 tens of thousands, hundreds of thousands of genotypes available in the UMHET3 population,

00:14:08.400 it's really unlikely that one weird genotype would either trick us into believing something to be true

00:14:14.080 when it really isn't, or trick us by missing a good response. The other ancillary benefit is you can

00:14:20.720 map genes. There's a set of collaborators, including Rob Williams at Tennessee and Johann Auercks in

00:14:26.800 Switzerland, which have taken these mice. We've given them at this point something like 12,000 tails,

00:14:32.880 12,000 DNA samples from mice that have a known lifespan. And they have already published a paper. It came out

00:14:40.480 last year in science, and there's another one in the pipeline now that says, oh look, here's a gene

00:14:45.520 that tells you how long the females will live. Here's a gene that tells you how long males and

00:14:49.440 females will live. Here's a gene that tells you how long you'll live, but it only counts if you've

00:14:54.400 made it past the midpoint. It only works on the oldest half of the mice. All that is very cool science.

00:15:00.800 There are hints to human genetics lying within that, and it gives you new tools for thinking about and

00:15:06.000 then working out ideas about the ways in which your inheritance modifies your aging and maybe even

00:15:12.320 your response to drugs. Rich, I want to make sure that listeners who maybe aren't as familiar with

00:15:17.680 genetics understand the significance of the UMHET3 mouse relative to the black six. So let's again talk

00:15:26.320 about what it means when you have a black six model. They are all identical, correct?

00:15:31.680 Absolutely correct, but it's even worse than that. Not only are they identical, they are homozygous.

00:15:39.360 I know, the gene from the mom and the gene from the father are the same. So it's like an

00:15:43.360 inbred form of homozygosity. We don't even have a human phenotype that is that inbred.

00:15:50.480 Right. People avoid inbreeding because it turns out that when you inbreed people, you get

00:15:56.320 very sick people, a lot of deaths, a lot of deformities, a lot of mental disabilities. And

00:16:02.240 that's true of inbred mice as well. Inbred mice almost always have something terribly wrong with

00:16:07.760 them. Nearly every kind of mouse that's used in aging research is fully deaf by one year of age.

00:16:13.920 Many of them are blind. Many of them get a single disease, which is not representative

00:16:19.680 of mice in general. It's almost like a thought experiment where you take a small population

00:16:24.480 of people and you make them breed and breed and breed and breed and breed until they all become

00:16:29.360 one person. And then there's two issues. One is the probability that that person is healthy is zero.

00:16:36.560 And then secondly, even if you accept that fact and do all of your testing,

00:16:41.440 what is the likelihood that what you learn is relevant to people who are not inbred?

00:16:46.880 Yeah. You can see it in the form of a clinical trial. Let's say you have a drug. You want to

00:16:51.120 test it to see if it prevents cancer in people or something. And you decide your test population will

00:16:56.000 be a set of identical triplets, Jim, Josh, and John. They're identical triplets and you've decided

00:17:01.760 you're going to test it in them. People would laugh at you. That's not a good design. If you want

00:17:07.680 to see if your drug works, you sort of have to test it on people who are not identical genetically

00:17:12.880 to one another. Yet that sort of thing, which is so obvious in human analogies is ignored by

00:17:18.960 nearly all mouse scientists. I hate asking people to sort of

00:17:22.480 speculate on the motivations of others, but why does the black six model still exist? Why is

00:17:27.920 biomedical research being done in this model? If we want to have any interest in some translational

00:17:34.080 insight? You're a scientist. You're setting up your own lab. Your mentor in her lab, she used black six.

00:17:40.080 And so all your preliminary data is in black six. And so you do black six. If you are aware of these

00:17:45.840 controversies, you just might say, oh, I want to test it in some other kinds of mice. But then you

00:17:50.800 say, no, no, no, my money is limited. I can only afford one kind of mouse. I'm going to take the kind

00:17:55.840 I'm familiar with. It's like lemmings. You follow the lemming in front of you because that's just how

00:18:01.120 lemmings do it. They don't look at a roadmap or think about the optimal path to take. They just follow

00:18:07.200 the person who trained them, who's following the person who trained them, et cetera, et cetera.

00:18:11.920 Inbred mice are good for two things. They're almost always sick. And if you want to study

00:18:17.600 some kind of sickness, bingo, you've got it. If you want to study lymphoma, you've got some inbred

00:18:23.360 mice that get lymphoma or get blind or get hereditary deafness or something. Studying inbred mice is great

00:18:29.440 for that. The other thing that they're critical for is for transplantation. There are a lot of

00:18:33.920 experimental designs where you have to take cells from one kind of mouse, stick it into another kind

00:18:38.640 of mouse. For that to work, both mice have to have the same genotype. Inbreds are still bad for that.

00:18:43.840 The right ones you want to do are the children of two different kinds of inbred mice. That's called

00:18:48.640 an F1 mouse. They're better because they live longer. They're less sick. But that's what inbred mice

00:18:53.920 are good for. You can use them to construct real mice like they had three mice. They're good building

00:18:58.400 blocks like Lego blocks. But to do science on them is almost always a mistake. All right. Let's talk

00:19:04.000 about the way in which a study is conducted. So let's not necessarily even focus on a given molecule,

00:19:09.120 but there's a candidate molecule that's been nominated. The board has reviewed it and decided

00:19:13.600 there's enough biologic plausibility that we're going to test this. Let's talk about the metrics

00:19:18.800 of interest. Let's talk about median lifespan, maximal lifespan. What do those things mean?

00:19:23.440 Are those always the primary outcomes? What are some of the other outcomes you consider?

00:19:27.680 The primary outcomes, the things that we do our statistics on, are the proportional hazard,

00:19:33.120 that is the risk of death over the whole lifespan, which the closest easily understood term is the

00:19:39.040 median lifespan. It's not quite correct statistically, but nearly always it's a good shortcut to say,

00:19:46.480 see, this extended the median lifespan by 20% or 5% or whatever. The median lifespan is the age at which

00:19:53.600 half the mice have died and half are still alive. So if half of the mice in the normal group died by

00:20:00.240 800 days and in the drug treated group, half of the mice were still alive on day 80, that's 80 days later,

00:20:07.120 then that's a 10% increase in lifespan. That's a nice big jump. We also always calculate some

00:20:14.800 measure of maximum lifespan. The actual maximum lifespan, the age at death of the last mouse to

00:20:21.280 die is statistically not very useful, not valid. It varies so much depending on the population size and

00:20:28.400 just the luck of the draw. What we do is a better way, which was worked out by David Allison about 20

00:20:34.720 years ago. The test statistic we use is we wait until 90% of the mice are dead in both populations,

00:20:41.280 the control and the treated population. And then on the date when the 90th percent mouse dies,

00:20:46.560 we say what fraction of the mice are in the treated group of the ones that are alive,

00:20:50.400 and what fraction of the mice are in the control group. If we're lucky, if we have a good drug,

00:20:55.920 we might have of the mice in that pool population. If 80% of them are treated and only 20% are control,

00:21:02.960 we know that we have a drug that will give you a much better chance of being alive when you're

00:21:07.280 really, really old. That's the closest we can come to a statistical test for maximum lifespan.

00:21:12.720 Sorry, let's make sure I understand that again. So you have to wait until in both groups,

00:21:18.880 there's only 10% or less remaining? Because that's obviously going to occur at a different time.

00:21:24.000 Yes. What you do is you make your list of all the ages of death in both groups.

00:21:29.600 Then you figure out what age is the age at which 10% are still alive in the pool together and 90% have

00:21:37.840 died in both groups. In the total pool. I'm sorry.

00:21:41.280 Smushed together. Yeah. So there's only one age where 10% of the pool is still alive. Then you look at

00:21:48.560 the pool and say, okay, well, what fraction of those were in the treated group? If it's a 50-50 mix,

00:21:53.440 you've lost. If it's 80% treated and 20% untreated, you've got a winner.

00:21:59.520 Was there ever a test that looked at the 95th percentile for each group and compared those

00:22:04.800 time horizons as a surrogate or proxy for maximal lifespan? Am I making that up? It's possible I am.

00:22:10.400 You can do that. You can do the 99th percentile. There are a couple of concerns,

00:22:14.560 and there are technical ones. One is that the further you push it up, 95, 99, 99.9, the fewer

00:22:20.400 mice you have to work with. After a while, the statistical power drops. You'll miss a lot of

00:22:26.720 stuff if you only have one or two mice alive at that age. We've picked the 90th percentile because

00:22:32.960 there'll still be a few dozen mice alive at that age at the 90th percentile. The other thing you want

00:22:38.960 to be really careful of is defining your test and then picking the test that looks best. We could

00:22:44.480 have some groups actually do this, though it's sort of unethical. They look at the 80th percentile,

00:22:49.680 the 85th, the 90th, the 95th, and then they say, oh, it looks best at the 90th. Let's pick the 90th.

00:22:57.200 That's not a good thing. You will fool yourself. You'll get false positives. And so we define

00:23:02.320 arbitrarily 90th. We're going to do it at the 90th percentile all the time.

00:23:06.320 Yeah. So in other words, it's sort of pre-specified.

00:23:09.440 Oh, yeah, that's right. It's pre-specified. It's written down before we do any test. In fact,

00:23:13.280 it was written down 20 years ago. Do you come up with a new power analysis

00:23:18.480 for each experiment or do you generally power them the same and therefore have the same number of mice

00:23:24.640 in each ITP? Yeah, we use the same number of mice in the control group and the same number of mice

00:23:31.040 in treated group each and every year. Before we start, of course, we don't know whether it'll be a

00:23:36.160 good year or a bad year. Some years, the mice live five percent longer or shorter than in previous

00:23:40.880 years. And we also don't know which drugs will work. And the ones that do work, we don't know

00:23:44.880 how big an effect they'll have. So what we've done is, with the assistance of two professional

00:23:49.680 statisticians, and again, this was 21 years ago, we worked out the number of mice we want to have in

00:23:54.960 each group. And our criterion is we wanted to have at least an 80 percent chance of picking up a

00:24:00.880 significant result, even if one of the three sites had a disaster like an air conditioning failure or a

00:24:06.880 viral infection. So far, we've only had something like that happen once. But we said, let's be ultra

00:24:12.560 conservative. We want to have an 80 percent chance of picking up an effect that's at least a 12 percent

00:24:18.800 effect up or down a two-sided test. Even if only two sites survive, we use that amount to determine

00:24:26.720 how many mice we're using. Now, in practice, almost always, we can combine all three sites

00:24:32.480 together. We almost never have a site-specific disaster. And that means in practice that we have

00:24:38.560 80 percent power to detect drugs will give us like an 8 percent increase or a 10 percent increase.

00:24:44.720 12 percent was kind of conservative. And what does that amount to in total mice usage,

00:24:50.720 typically, for a given drug? At each site, we are using 100 male controls and 100 female controls.

00:24:59.280 So pooling, that's 300 males and 300 females in the control group each year. And for each drug,

00:25:07.360 we use 50 males and 50 females. That is half as many as the controls, again, at each site. So for any one

00:25:14.080 drug, 50 times 3 is 150. We have 150 male mice distributed across three sites and 150 female

00:25:20.880 mice distributed across three sites. We oversample the controls as we double the size of the control

00:25:26.800 group because we're going to compare controls to drug A, controls to drug B, controls to drug C. So

00:25:32.880 in a sense, we're reusing. We're getting our best mileage out of the controls since we can increase

00:25:39.280 statistical power for every one of those pairwise comparisons by adding extra mice to the control

00:25:44.880 group. So we do it that way. In other words, if you're doing a five-drug trial, there will be a

00:25:51.200 total of 300 male, 300 female control, but there will be 750 total male, 750 total female on the combination

00:26:02.000 of drugs. Yeah. Seven drugs times 50 is 350 mice getting some drug. Yeah. Male mice getting some drug.

00:26:10.960 Oh, I thought it was 150 male and 150 female per drug across the three strikes. Oh yeah, you're right.

00:26:16.480 I was talking about one site at three sites. You're right. It's 150 times the number of drugs.

00:26:20.800 Okay. And then how much, you've done this for now nearly 20 years. What is the range or variability

00:26:28.800 in lifespan, both median and maximal for the controls? We got something we didn't expect to see,

00:26:36.000 but we've seen it almost every year for the last 20 years. We were hoping that the three survival

00:26:41.920 curves, that is one at each site, would always be superimposable. Of course, we're all really good

00:26:46.480 at taking care of mice. And for females, that was correct. Each and every year, the Michigan, Texas,

00:26:52.880 and Jackson Lab's female survival controls are almost always superimposable. There's been one year,

00:26:59.440 2017, where that was not the case, but most of the time it's the same. For males, however,

00:27:05.600 there is a persistent and consistent difference, which we really do not understand. Males at Michigan

00:27:11.840 always live five to 10% longer than males at the other two sites. That's unexpected and problematic,

00:27:19.840 and we really don't understand it. Maybe it's the water tastes funny, or there's some smell that the

00:27:26.480 mice are obsessed about that we don't know about, or there's some contamination in the sonic environment

00:27:34.320 that is site-specific. We've tried to do everything we can to eliminate it, and we have failed. We also

00:27:40.560 know that there are site-specific differences that are read out as weight, and here it works in both

00:27:45.600 males and females. The Michigan mice controls are about 10% lighter, both in males and females,

00:27:52.560 than mice at the other two sites. So despite our best efforts, we get the food from the same place,

00:27:58.000 the bedding from the same place, the mice from the same place. We breed them in the same months,

00:28:04.000 trying desperately to get them to come out the same way in all three sites. And we've gotten close,

00:28:10.640 better than many people would do, but not perfect. And particularly for the survival curves in males,

00:28:15.760 there's been a consistent site-specific difference.

00:28:18.080 So it's interesting that both the males and females at Michigan are 10% lighter. Presumably,

00:28:23.360 they're eating less or moving more, yet it only translates to a survival benefit in the males and

00:28:29.600 not the females. You have the facts right. It's the causality that's uncertain. We don't know whether

00:28:35.520 the difference in survival in the males has anything to do with what's causing the weight difference.

00:28:41.360 We see them both, just as you said. We don't know whether the changes in weight contribute to the

00:28:47.680 changes in survival or whether it's just two independent facts. The observation that the females

00:28:53.440 are lightweight too, despite the fact that the female survival curves are superimposable,

00:28:57.840 suggests that it's not something simple like that.

00:29:00.320 Now, does that allow you to still pool the results for the males?

00:29:06.160 Yes, but we do it with a very standard, not esoteric, statistical trick. We cite Stratify.

00:29:12.400 That is, the survival curve for the Michigan males controls is compared to the survival curve for the

00:29:18.320 Michigan drug-treated mice, and the survival curve for Texas controls to the Texas drug-treated mice,

00:29:24.480 et cetera. And then what the statistical program does is it pools those three sets of results to

00:29:31.120 come up with one overall statistic. In every paper, we also report separately. Here's what we saw at

00:29:36.800 Michigan. Here's what we saw at Texas. Here's what we saw at the Jackson labs. But our primary hypothesis,

00:29:42.640 the thing that allows us to put the name of the drug with the word winner in the title of the paper,

00:29:47.840 is the pooled result in which the results have been stratified. And we stratify the test for maximum

00:29:54.240 lifespan, the Wang-Allison test. We adjusted that to make it stratified also.

00:29:59.040 How much complexity is there year after year, drug after drug to create a good formulation of the drug

00:30:07.120 to give the mice? How are you thinking about the dosing frequency, the dose itself, and how to ensure

00:30:14.880 that the animals indeed get the amount that you want? We've put a lot of effort both in planning

00:30:20.240 it and then executing it. The University of Texas is led by Randy Strong, who's a pharmacologist.

00:30:26.800 And he has a colleague, it used to be Marty Javors, and now Brett Ginsburg has taken over that role,

00:30:33.040 devotes a lot of his time to asking exactly that sort of question. So for example, before we give any

00:30:39.920 food to mice, we make a batch of food with drug in it, and Brett's lab takes it and measures the

00:30:45.760 amount of drug in the food. And if it's 5% of what we thought it would be, something has gone wrong.

00:30:52.480 And so we try to work out, and this again, Brett is an expert at this. Should we dissolve it first in

00:30:57.680 alcohol? Should we incorporate it in the form of a corn oil suspension? What's the best way to get it

00:31:04.080 into the food? So once Brett, with the involvement of Peter Reifsnyder, who runs the lab at the Jackson

00:31:11.120 Labs, once Brett and Peter have shown that they can make food with the stuff in it and is at the

00:31:15.920 right dose, we then give it to mice for eight weeks. And tissues from these mice then go back to Brett.

00:31:21.760 So Brett can say, yep, the tissue of the drug-treated mice has this amount of drug in the liver and this

00:31:27.680 amount of drug in the plasma. It's only at that point that we actually press the button

00:31:33.920 to say, yes, we are going to use this drug in a lifespan experiment. Those pilot mice, my lab,

00:31:39.520 gets their liver and we look at a batch of mRNAs because we want to at least satisfy ourselves that

00:31:44.880 the drug did something. We've picked RNAs in the liver that we know are almost always sensitive to

00:31:49.680 drugs. And if the drug was given to the mice and nothing happened in the liver, we start to get

00:31:54.880 worried. Maybe the drug was excreted quickly and didn't have any biological effect whatsoever.

00:32:00.800 And we've run into problems. I mean, the famous one, the early one was rapamycin.

00:32:05.280 About 90% of the rapamycin that was given to mice in the food never made it into the mouse because

00:32:11.280 it's digested in the stomach and the acid conditions of the stomach, it's degraded. And so Randy,

00:32:17.040 with colleagues of his, worked out a way to encapsulate the rapamycin in a capsule, a plastic capsule that

00:32:23.360 makes it through the stomach and dissolves in the more alkaline conditions of the small intestine.

00:32:29.360 So that was a way of tricking the body into getting the rapamycin to the portion of the

00:32:34.240 GI tract where it could be absorbed. And it's not uncommon for Brett to have to say,

00:32:40.640 well, you know, this is not going to work unless we dissolve it in a little bit of alcohol before

00:32:44.640 we mix it into the food. So we do test for that. These are all important issues.

00:32:48.800 The other thing that the data on blood concentrations, they can alert us to potential problems.

00:32:53.680 For instance, many of our papers, we've published, I think, 12 papers now on rapamycin.

00:32:58.240 And almost always, rapamycin has been giving a larger percentage increase in females than in males.

00:33:04.000 Well, the Texas group showed that the blood concentrations are threefold higher in females

00:33:09.600 than in males. We haven't proven that that's why the females live longer or have a greater

00:33:14.880 percentage increase, but it's certainly very plausible. We've stumbled onto a similar

00:33:19.760 situation with canagliflozin. This was published a few years ago. Canagliflozin, we found lovely

00:33:26.000 lifespan increase, but it was in males only, not in females, which is really weird because it's a

00:33:30.560 great drug for diabetes in men and women without sex differentiation. We thought it would work great

00:33:36.320 in both sexes. It turns out that the blood concentration in female mice is three times higher

00:33:42.000 than in males. And as they get older and older and older, the blood concentration in the old females is 10

00:33:48.400 times. The concentration of young males, and it's probably toxic. We now are going back to say,

00:33:55.360 okay, well, let's give it to females at a much lower dose. Maybe that'll work. Or let's give it

00:34:00.320 to females, but stop when they're middle-aged so it won't ever reach the really high toxic concentrations.

00:34:07.440 All of these assessments of drug levels allow us to be on the alert for problems of that sort and to

00:34:13.680 come up with ideas for how we might solve them. Rich, I assume it's the case that every candidate

00:34:19.760 drug must be administered in the food. You're not testing intravenous drugs into muscular drugs?

00:34:25.360 Yes, this is correct. There's a footnote, which I'll get to. We could give a drug in water if there

00:34:30.400 was some reason in which it wouldn't get into the food, but you could give it in the water. This would

00:34:34.400 be compatible with our protocol. We can't give it to animals intravenously or intramuscularly because

00:34:39.760 it's so much work to give 150 males, 150 females giving 300 shots a day or a week or a month would

00:34:47.680 be a mess. In addition to which we'd have to have a separate control group that got a saline shot

00:34:52.560 and we couldn't use regular old mice as our control. We'd have to have a separate control

00:34:58.400 group to do that. You can imagine a specific situation if someone said, I've got a an antibody

00:35:05.520 or something. All you really have to do is give it to them once and it will flip their immune system

00:35:10.640 forever into a good configuration. Please give my drug, has to be given by injection, but only one time.

00:35:17.600 We could consider that. We might need a separate once only sham group also, but I think that's the only

00:35:25.120 exception to the rule that we would think seriously about.

00:35:29.280 Okay. Final question on that. Just so folks understand, how are you regulating the actual

00:35:34.000 dose? In other words, how are you monitoring the amount of chow that's consumed since the

00:35:38.800 amount of chow that's consumed is proportional to the drug that's consumed?

00:35:42.080 We have no idea how much food any one mouse eats and therefore we have no measurement on a

00:35:47.040 mouse-by-mouse basis on how much of the drug they've consumed. We know, of course, that smaller mice

00:35:53.200 eat less food than larger mice. So they will get less drug per mouse, but probably about the same

00:36:00.560 amount of drug per gram of lean body mass or something like that. There's no way to control

00:36:06.160 that other than by putting individual mice in individual cages, which has its own major problems.

00:36:11.920 Mice are very social creatures and they don't like isolation cages anywhere that people do. In addition,

00:36:18.240 monitoring the actual amount of food a mouse eats is a fiction. No one can really do it.

00:36:23.440 They can put a number down and get into the paper, but it's a fictitious number. And the reason is that

00:36:27.760 mice chew their food and leave a lot on the cage, on the floor of the cage. So you don't know how much

00:36:34.000 food the mouse has actually gotten into itself because you haven't measured little crumbs on the cage

00:36:40.720 floor.

00:36:41.040 In total, how many drugs have been run through the protocol in the last 20 years?

00:36:46.880 I would have to go to our recent review article and then add some more drugs to it. It's about 100.

00:36:52.240 And of course, many of those, you've already alluded to this, have been run through multiple times.

00:36:56.560 So there are many more experiments than 100.

00:36:59.120 Yes, that's right.

00:36:59.860 To the point you made at the outset, there really are some notable successes. I want to talk about those.

00:37:05.720 I also would like to talk about some notable failures. Let's start with the successes. So what was the

00:37:10.480 first exogenous molecule that proved a lifespan extension success for the ITP?

00:37:17.360 Yeah, well, rapamycin in our 2009 paper had a really big effect. We picked a dose that seemed

00:37:24.560 like it might work, and it did. It's not the optimal dose. It's less than the optimal dose.

00:37:29.360 The dose we chose, both males and females, had a significant lifespan extension. To put this into

00:37:35.120 perspective, these drugs are giving, at the middle dose, 15% to 20% increase in median lifespan.

00:37:41.280 To give a sense of what that means, if you had a cure for cancer in people, no one over the age of

00:37:45.920 50 ever got cancer again. Median lifespan of humans would go up by 3%. The same is true if you had a

00:37:52.320 drug that abolished heart attacks. No one over the age of 50 ever got a heart attack again. Median lifespan

00:37:58.320 for people would go up by less than 3%. That's work done by Jay Olshansky and Bruce Carnes and published

00:38:04.480 in Science in 1990. So the drugs that we consider, we have four of these now, that give more than a

00:38:11.600 10% increase in lifespan in terms of proportional change of healthy lifespan are doing about three

00:38:18.560 times better than some hypothetical drug that abolished cancer in people or abolished heart attacks in

00:38:24.480 people. That's a really significant chunk of additional healthy lifespan. Rapamycin in that

00:38:32.000 first paper was also the first drug I believe where anyone had showed, we found, that it works quite

00:38:37.440 well even if you start in really old mice. Some of the mice that were exposed in that paper did start

00:38:44.080 until 20 months of age, where the median survival is about 24 for males and 26 for females. It took me

00:38:51.760 very much by surprise. We thought if a drug was going to slow aging, you really do have to start it when

00:38:56.720 you're young because a lot of aging is what happens between the ages of 20 and 60 or something like

00:39:01.760 that, as everyone knows. So it was stunning that a drug could start as late as that and still have

00:39:07.200 a full lifespan benefit. That's really was news scientifically and I think that's one of the

00:39:12.160 reasons why the editors of Nature were interested in it. But that turns out not to be a fluke.

00:39:17.280 17 alfesterdiol, which is male specific, works just great if you start it at 16 or 20 months of age

00:39:24.320 in the males. Acarbose, which is significant in males and females, though better for males,

00:39:29.760 if you start it in middle age, it still works. It's only about half as good. Starting early is smart for

00:39:35.040 acarbose, but even if you start it at 16 to 20 months of age, it still works just fine. And

00:39:41.360 conagliflozin, our data in that group haven't been published yet. For males, it's still terrific.

00:39:49.040 For females, as I mentioned, it actually isn't good, but we suspect it's because the drug

00:39:54.800 concentrations in the blood of females may be toxic. So we really want to do that again,

00:39:59.040 but with lower concentrations of the drug. Do you have a sense of why rapamycin and

00:40:03.280 canagliflozin are being more concentrated in females? And are you seeing that with any of the

00:40:07.280 other successful candidate drugs, such as acarbose, for example, or 17-alpha estradiol?

00:40:13.360 We don't know the answers for any one of those drugs. It wouldn't be too hard to find out.

00:40:18.240 A pharmacologist could look at how quickly it's absorbed, how quickly is it conjugated,

00:40:22.400 how quickly is it excreted, does it go out in the urine, does it go out in the feces, all of that.

00:40:26.560 Very standard 50-year-old methods for answering that question, I think would be important to address.

00:40:31.520 There's a generic answer which is really quite firmly established. The enzymes that the liver uses

00:40:38.640 to deal with foreign drugs, these are called enzymes of xenobiotic metabolism, xenobiotic

00:40:43.360 metabolizing enzymes, are radically different between men and women and between male and female mice.

00:40:49.600 Most of them, not all, but most of them are a lot higher in females, but some are a lot higher in

00:40:55.520 males. And this is also true for people. So, the pace at which drugs are conjugated,

00:41:02.000 put into the bile, or put into the urine, or excreted in the feces, or excreted in the urine,

00:41:07.440 very often are sex-specific. It doesn't surprise anyone to find that the blood concentrations may

00:41:13.440 be different in men and women, or different between male and female mice. The details on a drug-by-drug

00:41:19.440 basis, we haven't looked at yet. The one thing I would want to add here as a footnote is for acarbose,

00:41:25.040 it has nothing to do with that. Acarbose, nearly all of it stays in the gut. It doesn't get absorbed

00:41:29.920 into the body, so excretion is not the key issue. Why the acarbose has such a big effect in males and

00:41:36.080 a small, significant, but small effect in females is unknown. It presumably has to do with males being

00:41:42.560 more sensitive to high glucose levels. Acarbose probably works by limiting very high glucose

00:41:47.520 levels, maybe for unknown reasons that triggers something horrible in the males and not so much

00:41:53.040 in females. Rich, when you talk about the difference in the pharmacokinetics between

00:41:58.080 male and female mice, we can only extrapolate and say that our cytochrome P450 system as humans must

00:42:06.160 have sex differences. But what I can't tell you for the life of me, Rich, is one drug that I'm aware

00:42:13.280 of that we really differentially dose in males and females beyond a weight difference. In other words,

00:42:20.880 we don't seem to take into account that difference when we give a person an antibiotic or a statin

00:42:30.400 or a chemotherapy. They're all based on either weight or nothing at all. I guess what I'm saying is,

00:42:36.160 this is kind of remarkable to me that we don't have a better set of the pharmacokinetics of these

00:42:42.320 drugs and their differences in human sex and how that maybe should factor into how we think about

00:42:47.680 dosing them. I would love to see you have on your show a real pharmacologist who knows the answers

00:42:54.400 to that question. I would love to talk to a real pharmacologist and ask him or her, hey, aren't there

00:42:59.440 differences between men and women in the rate at which drugs are excreted? And why isn't that informing

00:43:04.960 our recommendations for drug doses in people? I think it's a really good question, but I don't

00:43:09.120 know the answer. That's something we're going to have to dig into a little bit because it seems like

00:43:12.880 an enormous missed opportunity to get that level of granularity. You've rattled off a bunch of the

00:43:18.860 successes. I think it's worth a bit of a double click on each. We can take them chronologically. So

00:43:23.180 again, people who listen to this podcast are very familiar with rapamycin, recently had David

00:43:28.540 Sabatini and Matt Caberlin on together. So we did a pretty good deep dive on RAPA. But again,

00:43:34.180 just on the off chance that someone's coming to this as their first exposure to rapamycin.

00:43:38.760 Do you remember who made that nomination for the 2009 paper? Dave Sharp. Okay. So it came in-house

00:43:43.980 almost. And obviously the logic at the time was what? I mean, 10 years earlier, the drug had been

00:43:50.940 approved by the FDA for solid organ transplants. So it was a known immune suppressant. What was the

00:43:58.900 logical step that took it to, but we also think it could boost longevity?

00:44:03.880 People had just begun to make invertebrates, that is worms and flies, with the genetic modulation of

00:44:10.480 the TOR pathway. TOR is the target of rapamycin, the enzyme that rapamycin inhibits. And you know,

00:44:17.060 they were long-lived. And in fact, when you studied some of the yeast stuff that Matt Caberlin was in

00:44:22.940 on this early on with Brian Kennedy, many of the things that seemed to influence yeast lifespan were

00:44:29.140 also related to the TOR pathway. So Dave said, I wonder whether if you inhibited TOR in mice,

00:44:36.480 they would also live longer. It really didn't have to do with the notion of immune suppression.

00:44:42.300 It had more to do with control of growth and following up on the genetic data in the invertebrates.

00:44:49.880 Now, in retrospect, 20-some years later, we understand that the notion of rapamycin as an

00:44:55.860 immune suppressant is the sort of tip of the iceberg. There are some immune functions which

00:45:01.360 it increases. There are some that it decreases. It's dose-dependent and context-dependent.

00:45:06.940 By an interesting coincidence, the same year that our lifespan paper appeared, there was another paper

00:45:12.160 also from Michigan, though not my lab at all, in which they gave rapamycin to some old mice

00:45:17.240 and shown that their influenza vaccine response was terrific if they'd had rapamycin. What rapamycin

00:45:23.740 appears to do in their model is it increases the production of B cells from the bone marrow.

00:45:28.740 So the mice would respond to influenza vaccine, and then they were exposed to live virus and they

00:45:34.240 survived, whereas untreated controls did not survive. So in some circumstances, at least,

00:45:39.500 it's actually immunoboosting. And that was also demonstrated by Joan Manik and Lloyd Clickstein

00:45:45.120 six years later using Everolimus. That's right. In humans, it also improves vaccine responses on some

00:45:52.400 circumstances to influenza and vaccination in humans. So you've already alluded to this, but I think it's

00:45:57.780 again just worth making sure folks understand this. You had a little bit of a challenge getting the

00:46:02.800 rapamycin formulated. Why did you decide not to abort the study? When you finally got the food

00:46:08.060 formulation done, you've got these geriatric mice, you finally get to start feeding them rapa. I mean,

00:46:14.520 obviously, chance favors the prepared mind, according to someone famous. You elected to take a chance.

00:46:20.920 Do you remember the decision? I mean, the odds were long, right?

00:46:24.420 There are two decisions. Step one was Randy Strong saying, I'll bet we can get this into mice. I've got

00:46:30.780 buddies who do encapsulation. Let's encapsulate it and see if that stuff works. So Randy's

00:46:36.720 initiative and creativity were a first important step. Now we have a batch of mice. They're 20 months

00:46:43.980 of age. They were originally going to get rapamycin when they were four months of age. We could either

00:46:49.360 A, throw them out, or B, give them the rapamycin. We were sure it wouldn't work. Now we were wrong.

00:46:55.420 So when Randy finally, with his colleagues, figured out how to make the protected version,

00:47:01.100 the encapsulated version of rapamycin, we actually used it twice, the same batches. Some of it went to

00:47:06.520 the mice that were already 20 months of age, so we wouldn't have to throw them out. And then we

00:47:11.440 executed it. We gave the rest to the young mice that had been produced in the following year,

00:47:17.520 expecting that the old mice that would fail, the young mice that might work. And as you know,

00:47:22.020 it worked well in both ages. What was the difference between the male-females in the 20-month onset versus

00:47:28.600 the four-month onset, both by sex? Scott Pletcher actually did that analysis. When you compare

00:47:34.360 within sex, so you're comparing older females to younger females, there's no difference, surprisingly.

00:47:42.480 The benefit you get starting in old age and the benefit you get starting in young mice

00:47:47.660 is statistically insignificant in females, and it's statistically insignificant in males.

00:47:54.540 In both sexes, starting as late as 20 months of age does not diminish the ability of the drug to

00:48:02.000 extend lifespan. This is just really remarkable. Let's try to philosophize a little bit about what

00:48:06.940 that tells us about the biology of aging. Or does it tell us more about aging, or does it tell us more

00:48:11.160 about the drug? I think it tells you something about both, about the drug and its interaction with the

00:48:16.500 aging process. The inevitable conclusion, which I would have bet a lot of money against, is that by

00:48:23.060 the time you're 20 months of age in a mouse, which is sort of like 55 or 60 years old in a person,

00:48:28.760 something like that, not yet at the median survival, but getting pretty close, damage will have occurred

00:48:35.240 that's irreversible. Collagen cross-linking and death of some brain cells and clogging of the arteries

00:48:41.740 or whatever's going to get you started already by the time you're in your 50s and 60s, or for a mouse,

00:48:47.580 by the time you're in your 20th or 21st month. But apparently there's still some further stages

00:48:53.420 of that process that occur afterwards, after we started to administer the drug at 20 months of age,

00:48:59.860 which are dependent on aging and the drug inhibits. So that's the news that we would not have known

00:49:06.480 if we hadn't done an experiment starting in late middle age. Now what those processes are, whether they are

00:49:14.060 same processes as affected by many different drugs, that's unknown. If we have time at some point to talk

00:49:21.600 about these new aging rate indicators that has come out of the ITP program, we can begin to ask questions

00:49:27.400 about whether the aging rate indicators are effective in the old mice or the middle-aged mice or the young mice.

00:49:32.080 Clearly, we've got a lot to learn now about what is happening that is drug-sensitive, even in middle-aged mice.

00:49:40.160 The fact that acarbos works half as well in middle-aged mice, canagliflozin works at least in males quite well

00:49:47.500 in middle-aged mice, which is unpublished, and that 17-alpha-estradiol works great even in middle-aged male mice

00:49:55.400 suggests that it's a general phenomenon. It's not rapamycin only, but it applies to a lot of phenomena.

00:50:03.360 Mike Garrett, when he was in my lab, he's now set up his own lab in New Zealand, but he took a lot of mice

00:50:08.880 and he put them on 17-alpha-estradiol or acarbos in middle age, and he found, and this is all published,

00:50:15.860 that their grip strength is great, much better than untreated old mice,

00:50:21.040 and their ability to stay on a rotating rod, which is a complex phenomenon having to do with balance

00:50:26.480 and muscle strength and motivation. That is much better, even if they started the 17-alpha-estradiol

00:50:33.200 or acarbos in middle age. It's not just a matter of the things that are going to kill you.

00:50:38.580 These mice don't die of dizziness or loss of grip strength or something, but a whole batch of stuff

00:50:44.100 that is age-sensitive is slowed by these mice. The next frontier that I really want our lab and other

00:50:51.540 labs in the ITP to dive into is cognition. Michigan has just recruited an absolutely top-notch

00:50:57.540 mouse neurobiologist, a woman named Catherine Kazurowski, and we already have several studies

00:51:03.900 planned or underway in which we're going to be treating mice with these drugs, and in addition

00:51:09.680 to looking at their lifespan, test them for cognition. The obvious hypothesis is that the

00:51:14.960 drugs that extend lifespan will also postpone loss of cognitive function. For complicated reasons,

00:51:22.860 Catherine thinks that will be untrue, and I think it is true, so we'll see which of us is correct.

00:51:28.140 What other measures of health span are you capturing? Obviously, cognition is a piece of

00:51:32.340 health span, but you already mentioned grip strength and then some complex motor tasks and some

00:51:36.780 complex stamina tasks. Do you look at, for example, muscle mass at the time of demise in these mice,

00:51:42.940 though it's not a direct driver of health span, it's a highly correlated driver of health span. I mean,

00:51:48.120 what else can you say about these mice and their health beyond just the elongation of life?

00:51:53.920 We have stage one and stage two studies. In a stage one study, which we do for every new drug,

00:51:59.680 lifespan is the only thing we measure in addition to body weight at four ages. We could throw in other

00:52:05.900 tests, but it's really expensive to do that and to standardize it so that all three labs get the

00:52:11.040 same numbers proved to be really tricky. If we did devote lots of our efforts to looking at these

00:52:17.480 secondary measures of health in addition to lifespan, we would test only three drugs a year or maybe four

00:52:23.680 drugs a year. But once you've got candidates, I mean...

00:52:26.060 Once we've got candidates, yes, then we go into stage two. Every drug makes it to stage two. We develop a

00:52:32.540 protocol that takes advantage of the strengths and weaknesses and interests of each site. So Dave

00:52:38.320 Harrison, for instance, had many tests of visual acuity, hearing acuity, strength, body temperature

00:52:44.280 regulation. So the stage two experiments at the Jackson Labs incorporated many of those tests.

00:52:50.980 Randy Strong and his colleagues were interested in glucose control and glucose homeostasis.

00:52:54.980 So the stage two stuff that was done at Texas always had some taste of that. My lab is interested

00:53:01.760 in pathology. So we would take a lot of these stage two mice, euthanize them at 22 months of age,

00:53:07.780 send them to a veterinary pathologist and come back with a long list of, here's what's happening in the

00:53:13.240 liver. But look, it didn't happen in the drug treated mice. Here's what's happening in the gut. But look,

00:53:18.460 it didn't happen in the drug treated mice. So our other hope, of course, is that once we have a winner,

00:53:24.200 the laboratories that have a specific interest in aging of the aorta and aging of the heart and aging

00:53:30.400 of the lung and who know what they're doing, will just ask us for tissues. They can buy the drug and

00:53:37.260 treat their own mice, hopefully not Black 6, but they can treat some mice, like Cat 3 mice, with the drug and

00:53:43.860 test their organ-specific functional or pathological outcome tests. Or if we have tissues in our freezer,

00:53:52.000 they can ask us for tissues. Since 2015, every drug, whether it's a winner or a loser, we've been

00:53:58.040 putting aside 20 or 30 or 40 mice, euthanized at age 22 months and frozen. This is changing next year.

00:54:05.560 I'll get to that in a minute. But anyone who wants those tissues just writes us a note saying,

00:54:09.800 here's what I want. This is the tissue I need. Here's what I'm going to do with it. Here's my

00:54:14.240 power analysis. And we just send in the tissue. It's clearly a national or international resource.

00:54:20.200 We've sent tissues overseas to other labs to make use of tissues that have tons of useful information

00:54:26.420 in them. But we can't study everything. We can hopefully attract the interest of people who can

00:54:32.420 study everything. This program, which we call CIP, the Collaborative Interactions Program,

00:54:37.720 unfortunately, I think, is going away next year. The National Aging Institute has decided that

00:54:44.220 instead they want to replace it with a new program, which they're calling Interventional

00:54:49.600 Biogerontology Repository, in which the tissues will not be requested from us. They will be requested

00:54:55.060 directly from the National Aging Institute. And the National Aging Institute will make all decisions

00:55:00.360 as to who gets the tissues, who doesn't get the tissues, how much tissue they get.

00:55:04.740 I'm not so sure that's a grand idea. But the worst thing is that they will have only once a year

00:55:11.040 call. When people send us a note, we generally can make a decision within two or three weeks. And they

00:55:17.440 can generally get the tissue within a month or two. Adding an extra year of time for the NIA

00:55:23.080 staff to sort of figure out who gets the tissue or not is not going to speed things up. So I view that

00:55:29.680 as a step backwards, but I'm not in charge of it. Rich, based on these tissue blocks,

00:55:35.180 what have you learned, if anything, through collaborations with people looking at epigenetic

00:55:40.400 changes in the treated versus untreated mice? I mean, we have to imagine that there's some

00:55:45.700 difference in gene expression, and that would be at least one way to look at it, correct?

00:55:50.980 Yes, that is one way to look at it. And we have indeed. There are some labs. Steve Horvath,

00:55:55.560 I think, is well-established and does lovely work. Steve has asked for tissues, and we have sent him

00:56:00.900 tissues. And we've explained to anybody else who is working on some aspect of either global or

00:56:07.220 localized tissue-specific epigenetic change, we'd be delighted to send them tissues.

00:56:12.540 Has that been done?

00:56:13.540 I know Steve has done some of that work. I think some of our drug-treated mice have gone to him.

00:56:19.540 Vadim Gladyshev at Harvard has gotten lots of tissues from us and has published

00:56:22.840 metabolomic assays, for instance. I don't think any of his epigenetic stuff has come out yet,

00:56:28.440 though I'm not certain of that. The paper that Johan Auwerks just published with Rob Williams and

00:56:34.560 Maroon Boo Slayman in Science had some epigenetic materials in it. The problem is that no one at this

00:56:42.420 point knows enough to know what tissue to look at. So it may be that a particular drug is working

00:56:48.300 because it sensitizes the liver to high glucose levels or something. So you'd want to then look at

00:56:56.280 liver tissue or pancreas tissue or islet tissue or fat tissue or, quite plausibly, tissue of some

00:57:02.960 obscure cell type in the hypothalamus, which regulates hormones that make you hungry or not hungry.

00:57:09.480 I'm going to guess the liver is important. I'm going to look at all the epigenetic changes in the

00:57:13.300 liver is a very crude way of addressing that. In my view, the progress will come when someone says,

00:57:19.420 hey, look, this drug works by hitting this enzyme in this set of hypothalamic neurons.

00:57:26.020 Now let's look at the epigenetic change in those neurons. We're not there yet.

00:57:31.800 You've already alluded to the aging rate indicators. Say a little bit more about that and how that's

00:57:36.980 factoring into the work.

00:57:37.940 Everybody is familiar with the idea of concept of biomarkers of aging. And by analogy, it's sort

00:57:44.660 of like the odometer in your car. Your odometer tells you how far the car has been driven and a

00:57:50.600 biomarker of aging in a crude sense tells you how far your body has been driven.

00:57:55.740 What would you say they are, Rich? I mean, what biomarkers do you think we really have of aging?

00:58:00.760 What are the odometers?

00:58:02.160 Yeah, that's a whole separate complicated issue. I personally don't think we have very many at all.

00:58:06.280 I would agree with that. I don't even think we have the odometer.

00:58:09.420 I agree with you. But conceptually, you can imagine if someone comes into your test facility

00:58:14.760 and they've got a lot of high affinity antibody and their vision is terrific and they've got no

00:58:20.540 cataracts and they're great at hearing and they can do 100 push-ups and they're great at running up

00:58:25.540 and down a hill and their skin is smooth, et cetera, et cetera. All 10 or 20 domains, they sort of look like

00:58:32.260 they're 40. You can say, okay, they are biologically young.

00:58:36.400 I put that in the realm of function. So I agree. We have lots of functional tests that give you

00:58:42.440 odometer-like insights. But when you think about a biomarker that is assay-based, I think we're both

00:58:51.820 saying the same thing. There's nothing there.

00:58:54.080 Yeah. I wasn't attempting to praise biomarkers of aging. I wanted to set that up as a familiar

00:59:00.800 concept because I mostly wanted to say that aging rate indicators are not that. Aging rate

00:59:06.080 indicators by this analogy are the speedometer. They tell you how quickly you're aging and not

00:59:11.200 how far you've gone. So what we were looking for, and in a moment I'll tell you the evidence

00:59:16.340 that we think suggests this is a good idea. We were looking for things that always change

00:59:21.120 in the same direction in every kind of slow-aging mouse. They would distinguish not how old they

00:59:26.500 were, but how rapidly they were going to be aging. So we have nine published and one unpublished

00:59:33.800 slow-aging mice. We have four genetic mutants, the Snell, the growth hormone receptor knockout,

00:59:39.720 the Ames dwarf, and the PAP-A. We have a famous diet, calorie restriction, and we have at least

00:59:44.700 four well-vetted drugs, acarbose, canagliflozin, 17-alpha estradiol, and rapamycin. We said,

00:59:50.580 let's find something that is changed in the same direction in all nine kinds of mice. And this is

00:59:57.640 true for a tenth kind, P10 overexpressors, though we've just submitted that for publication. So the

01:00:03.380 next time we talk, we'll have at least 10 mice for which these things are true. And the pleasant

01:00:07.720 surprise is that we now have 13 things that always change in the same direction in all 10

01:00:14.020 kinds of slow-aging mice, even when they are young adults. And that's the crucial thing.

01:00:20.580 The biomarkers are useless until the animal or the person gets old. They've got to have a certain

01:00:25.240 amount of aging behind them to see if they are young-like in comparison to control untreated

01:00:32.640 people or people with a different gene. You have to wait. The aging rate indicators, because they

01:00:38.060 are measures of speed, you can look even when the animal is young, hypothetically, when a human is

01:00:43.800 a young adult. So most of the work that we've done on the mutant mice was done on animals that were

01:00:50.940 four to six months of age. And the work on the drug-treated mice was on animals that were only

01:00:55.740 12 months of age. There are changes that are in famous molecules, molecules whose connection disease

01:01:01.800 is not just fanciful. One of these is UCP1, uncoupling protein 1. It's a mitochondrial protein that allows your

01:01:10.660 mitochondria to burn fat without doing a lot of work. It just turns the fat into heat. It's involved

01:01:16.160 in thermogenesis. And it's long been known that having a lot of UCP1 suddenly happens when you do

01:01:23.320 exercise. Exercise increases UCP1. And mice that have a lot of UCP1 live a long time. So it's thought

01:01:30.420 to play a major role in protecting you from obesity, from diabetes, from metabolic syndrome,

01:01:35.520 from sorts of inflammation. Well, every one of our slow-aging mice has a lot of UCP1 in the white fat

01:01:43.220 under skin, in the white fat in the abdomen, and also in the brown fat between the scapulae. All three

01:01:52.040 of these fat depots have elevated UCP1. The exception to that rule is very informative. You remember two of

01:01:58.900 those drugs, can agliflozin and 17-alpha-estradiol, extend lifespan in males only. And UCP1 goes up in

01:02:06.300 males only for those two drugs, which is a strong indication that whatever process is making the

01:02:14.560 drugs slow the mortality rate and increase longevity is the same process, at least in its sex specificity,

01:02:21.660 as the UCP1 story. And tell me, UCP1, you are measuring that how? We take the fat and we look

01:02:30.920 at the protein by Western Bloc. Okay. How much biologic noise do you think exists in that on a

01:02:39.160 day-to-day basis? So let's now talk about a normal mouse, a control mouse on the control diet. If you

01:02:46.420 sample his fat every day for a month, and some days you put him on a treadmill wheel, some days you

01:02:53.920 don't, some days he doesn't eat that much, some days he does. In other words, he replicates for

01:02:59.400 short bursts of time some of the activities that might either be associated with a longer life and

01:03:06.060 or associated with a behavior that increases the protein of interest, it would be really troublesome

01:03:12.160 even if everything you said were true, it would be really troublesome if UCP1 spiked on those days

01:03:18.140 because then it wouldn't really be a useful odometer or rather speedometer. So you brought up

01:03:23.680 three separate interesting issues and they need to be treated separately. The first, in a normal mouse

01:03:30.480 where you're not making exercise and you're not feeding them foie gras, if hypothetically you get a

01:03:35.940 little bit of fat from that same mouse every single day or every hour over the day, would there be much

01:03:41.800 change in UCP1? That experiment, of course, can't be done, but it doesn't invalidate our findings

01:03:47.160 because even if that would introduce some noise, the consistent difference between the mutant mice

01:03:52.480 or drug-treated mice and the controls, the noise is already built in, budgeted into that. If there was a

01:03:58.200 lot of noise, too much noise, we wouldn't see a significant effect of drug or the diet or the genetic

01:04:04.020 intervention. The second point you raised is, could one perturb this by getting an animal to exercise?

01:04:10.520 And people have done that. UCP1 is changed by chronic exercise. It's one of the reasons why

01:04:18.040 it's thought to be amongst the mediators of the health benefits that attribute in people and in

01:04:24.360 mice to exercise. Sorry, I was asking a slightly different question on that, Rich, although that's

01:04:29.300 good to know. The question I was asking was, if you took an otherwise sedentary mouse and exercised the

01:04:36.260 hell out of him for a day and checked it, then would you be fooled by an elevated level?

01:04:42.900 Yeah, I don't know.

01:04:43.480 Could somebody cram for the test?

01:04:45.920 I don't know. There may be some people out there who've done that, but I'm not one of them and I

01:04:50.380 don't know that literature very well. It's an empirical question, not hard to address. It may

01:04:54.840 already have been done. And the related question is about time of day. We try to normalize it. That is,

01:05:00.280 our mice are always euthanized between nine in the morning and 10 in the morning. So it's not the case

01:05:06.140 that some of the mice are morning mice, some of them are afternoon mice, some of them are evening

01:05:09.540 mice, and we feed them ad lib. So that does introduce some variation. Some may have had a

01:05:15.320 early morning snack and some may have had their last meal four hours ago before the lights go on.

01:05:22.380 That also is going to introduce some noise into the measure.

01:05:26.200 How are the animals euthanized?

01:05:27.360 We use a method that makes them go unconscious within five seconds. That is, we put them into

01:05:33.760 a bag, a plastic bag, and then we fill the bag quickly with carbon dioxide gas. They take a few

01:05:40.380 breaths and within five seconds, they're unconscious. And within 10 seconds, they stop breathing.

01:05:46.440 So in other words, it kind of minimizes, to some extent, the hormonal stress at the end of life,

01:05:51.840 given that carbon dioxide is highly sedating.

01:05:54.520 That is exactly why we do it. The American Veterinary Medical Association recently, that is

01:06:01.280 five years ago, made the method we use a suspect method. We still have permission to use it and

01:06:06.920 we've talked to the vets about it so we're not violating any of the rules. But they recommend

01:06:12.220 now, I think it's a rotten decision, that mice be placed in a cage and that carbon dioxide gas be added

01:06:19.460 gradually. We time this on a batch of mice to see what it would take. It takes seven minutes for them

01:06:25.440 to stop breathing or to lose consciousness. And over that seven minutes, their adrenaline level

01:06:31.720 goes up, as you can imagine. Their glucose doubles. Their blood becomes acidic. The pH drops.

01:06:39.180 So we would never want to do that because who knows what is that doing to all the protein kinases

01:06:45.380 and the metabolites. Anything that is glucose or hormone sensitive is going haywire there.

01:06:50.660 So we asked our animal care committee for that reason. Could we do it the way that it had been

01:06:55.960 approved for the previous hundred years? Because doing it the new way would spoil all of our

01:07:00.260 experiments. They said, yeah, you're right. Which I think is both more humane and also better science.

01:07:04.540 Yeah, I agree. There's an interesting parallel here with the way animals are harvested for human

01:07:10.540 consumption, which is unfortunate that the way most animals are harvested is very stressful on the

01:07:15.120 animal. And therefore, it actually erodes the quality of the food that you're about to consume

01:07:20.780 based on the stressful environment of the animal and the way it ties. Are there any other candidates

01:07:25.860 besides UCP-1 that serve as a fantastic indicator of a speedometer? Yeah.

01:07:31.160 A few of them are really interesting and exciting stuff. The same woman that did the UCP-1 study,

01:07:36.980 her name is Jinna Lee, which is L-I, also looked at macrophages in the fat. There are two kinds.

01:07:43.040 One makes a lot of inflammation and all of these slow aging mice, they go down. And the other prevents

01:07:49.560 inflammation, the M2 cells. All these mice, that goes up. So at least in the fat, all of these influences,

01:07:56.840 genes, diet, and drugs make the fat much less inflammatory. That's likely to be important

01:08:04.540 because you are fully aware of all of those studies suggesting that a lot of diseases involve

01:08:10.060 inflammation. High inflammation is a sign of stress. It's really bad for you. So it may be that

01:08:15.360 these drugs and diets and genes are working in part by reducing inflammatory tone. Now, in addition,

01:08:21.460 we have looked at proteins in the brain. The two we looked at in the brain, one is called BDNF,

01:08:28.660 brain-derived neurotrophic factor, and it's thought to protect brain cells from stress.

01:08:33.860 The other is double-cortin, DCX, which is a sign that the brain cells are making new brain cells.

01:08:39.740 It's a sign of neurogenesis. These go up in the brain in all of the slow aging mice. Again,

01:08:46.600 the exception being the two drugs that are sex-specific. And here, the BDNF and double-cortin

01:08:51.920 changes are also sex-specific and seen in males only. The macrophages and M2 were in the fat cell.

01:08:59.480 Was UCP1 also... Not in the fat cells. In the macrophages, in the fat depot.

01:09:04.100 Okay. Where was the UCP1? Was that in the liver? No, we looked at it in the brown fat,

01:09:08.920 in the inguinal white fat, which is subcutaneous, and in the perigonatal white fat, which is an

01:09:16.120 abdominal. Three different fat depots. And the macrophage changes are seen in each of those

01:09:21.860 depots. And the UCP1, which is in the adipocytes, is also seen in all three of those depots.

01:09:28.500 Got it. And I'm sorry I interrupted you. You were about to say one more.

01:09:31.040 We have many others now, but the last one that's really, I think, thrilling is a protein called

01:09:35.200 GPLD1. GPLD1, it's made by the liver. It's also made by the fat. What it does is there are lots of

01:09:42.500 proteins that are sticking out on the outside of a cell, and the linkage is a specific sugar bridge,

01:09:50.360 glucose phosphoinositide. That GPI bridge, GPLD1, cleaves that, and so it releases lots of different

01:09:58.260 kinds of proteins from cellular surfaces. The reason we thought it was interesting was another lab,

01:10:03.960 Horowitz, had shown, this is just two or three years ago, that if you exercise, GPLD1 goes up.

01:10:11.420 It's true for mice. It's true for people. And more exciting even than that, if you have GPLD1 go up,

01:10:18.460 cognition goes up. So Horowitz and his colleagues have argued that one of the reasons exercise is so

01:10:24.460 good for your cognitive powers is that it tricks the liver, maybe the fat. They said liver, but we found

01:10:30.940 it from fat also into making GPLD1, which in an unknown way improves cognition. So what we found

01:10:38.520 now in our lab was that all nine kinds of slow aging mice, 10 now, also have elevated GPLD1 production

01:10:46.580 in the liver and amount of it in the plasma of the mice. We can't prove that that's why BDNF goes up in

01:10:55.740 the brain. Double cortin goes up in the brain. Some of these mice are known to have great cognition. We

01:10:59.700 can't prove that it's due to GPLD1, but obviously we are hoping that that is the case. The reason that

01:11:06.260 the GPLD1 result was, to me, so extremely exciting was that in another part of the forest, another part

01:11:12.600 of the lab, a guy named Gonzalo Garcia was looking on differential mRNA translation, and he had discovered

01:11:19.000 that slow aging mice, turns out all of them, have a lot of CAP-independent translation. That is, they

01:11:25.520 can pick a subset of the messenger RNAs and translate them in a special way and cause proteins to be made

01:11:32.140 in ways that are independent of the amount of RNA for that protein. Well, we proved and published now

01:11:38.280 that GPLD1 is one of those proteins. It is controlled not by changes in the transcription of the DNA into the

01:11:45.820 RNA, but by the differential translation of the RNA into protein in a CAP-independent translation

01:11:53.540 module. So this is our first serious link between the molecular biology of protein translation,

01:12:01.180 Gonzalo's stuff, and the physiological effects like cognition and BDNF, which was Jina's domain.

01:12:10.300 It's really very pleasant to see these two different lines of experimentation sort of get tied together

01:12:15.100 here through the same protein, through the GPLD1 protein.

01:12:18.940 Tying this back to what we said earlier, this is an example of where, for example,

01:12:23.320 the epigenome might not matter as much because that presumably would have a greater impact

01:12:28.580 on transcription. And here you're saying, actually, this seems kind of independent of transcription.

01:12:33.880 This is a purely translational phenomenon.

01:12:35.960 Let me agree with that about 90%. So the 90% I agree with, many, many labs just look at RNA levels.

01:12:42.300 Transcriptone biology is relatively easy. And now you can follow it up with epigenetic

01:12:47.380 exploration of what controlled most of the published omics information is lists of RNAs that go up or

01:12:54.260 down. However, RNA is very poorly correlated with protein, and it's the protein that counts.

01:13:00.940 There's a lovely pair of studies from the Jackson labs. Ron Costanza was involved in one of these.

01:13:06.380 They measured, a lot of this is Gary Churchill's work, they took mice of four different age groups,

01:13:12.720 black six, but what can you do? And they measured the changes in proteins. They made a long list of

01:13:18.620 proteins that changed with age. Good. They did it in two tissues. Then they looked at the same mice,

01:13:22.980 a long list of RNAs that changed with age in the same mice in the same tissues. It turns out that the

01:13:28.760 correlation between the RNA and the protein was 30%. That is, only 30% of the age effect

01:13:36.720 on protein level could be blamed on, attributed to, changes in the underlying transcription data.

01:13:43.540 So if all you've got is the transcription data, which is what most people have, you have sort of

01:13:49.660 blinded yourself to the 70% of what is controlling protein levels. And it's the proteins that actually

01:13:56.060 do stuff in the cell. I think until people come to grips with that discontinuity, they won't really

01:14:04.280 be motivated to look at the proteins, which are harder to study, but doable. And I think it is the

01:14:10.980 proteomic data collection that will be valuable. Our lab has shown that the proteins can be modified.

01:14:18.540 We're not the first to show this, both by differential RNA translation, this cap-independent

01:14:23.300 translation. And Joe Endicott has found that there's also differential degradation, subsets of

01:14:28.420 proteins that are degraded by the lysosomes through chaperone-mediated autophagy. They mold the proteome

01:14:35.180 in ways that are completely independent of the mRNA for the underlying proteins. I think that's a big

01:14:41.560 part of the story, which people are just gradually waking up to. Yeah. I think this is so important.

01:14:46.520 I'm going to just slow you down. And I want to make sure we go through this again in some detail.

01:14:51.000 I understand what you're saying. I want to make sure everybody does, because I hadn't actually

01:14:55.900 known that fact about the poor correlation between protein translation and mRNA transcription. That's a

01:15:04.080 very big deal. I would have guessed that to be a much higher number. So can you just go back and give

01:15:09.580 everybody the sort of bio 101 explanation for how we turn DNA into mRNA into tRNA and protein,

01:15:18.280 where that's occurring in the cell? I want people to understand the point that you just made. I wrote

01:15:23.140 down, it's so vital, which is without the proteomic assessment, the story is incomplete. That's, to me,

01:15:30.100 the takeaway from what you just said. It's not only incomplete, but mostly wrong. So yeah, sure. I mean,

01:15:37.080 you learn nowadays, I guess, in high school that DNA can be transcribed into messenger RNA. It has the

01:15:43.320 sequence more or less and codes proteins. That's where epigenetic control comes in. Proteins made

01:15:48.800 only by the liver are in large part because the liver has turned some transcripts on and others

01:15:53.620 off. The same is true for the eye and the brain, et cetera. So now each cell has its own complement of

01:15:58.400 RNAs. Let's make sure people see that point again, right? The liver and the eye and the muscle have the

01:16:04.080 same DNA. Why does the hepatocyte make a protein that the liver needs, whereas the neuron makes a

01:16:12.560 protein that it needs? This is where turning on and off the gene, the epigenome matters. Okay.

01:16:17.860 It certainly does matter. People used to think it was the only thing that mattered.

01:16:21.260 And I think that's probably wrong, but it's certainly an important thing. Many of the differences

01:16:26.280 between the neurons and the skin cells and the blood cells and the liver cells are because they express

01:16:31.980 different messenger RNAs from the same DNA template. It's just like if you have a library of books and

01:16:37.300 someone decides to read the trollop and somebody else wants to read Emily Dickinson. They have

01:16:42.940 different experiences, even though they have the same library to work with. So now you've got a batch

01:16:47.140 of RNAs that you've transcribed selectively, depending on the cell. Now you have to make them into

01:16:52.960 proteins. So the ribosome will generally bind to them and churn out protein. And the sequence of the

01:16:58.300 protein will be based upon the sequence of the bases in the messenger RNA. Most ribosomes start this by

01:17:06.620 binding to the very end of the message at a place called the cap, the five prime cap. That's not where

01:17:12.900 the translation starts. That's the sort of start here signal. Then the ribosome bumbles its way down

01:17:17.700 to the place where it's going to start, and then it starts making proteins. So most translation is

01:17:22.700 cap dependent. The ribosome can only find and get working on a messenger RNA by binding to the cap,

01:17:29.880 bumbling down to the start site, and then making the protein. So the default presumption, which turns

01:17:36.660 out to be wrong, is that once you've got those RNAs out there because of transcription into the mRNA,

01:17:44.940 the rest is automated. They just churn out proteins based upon the RNA that they've got.

01:17:50.520 So there are now many studies, and I quoted my favorite ones because they come from friends of

01:17:55.640 mine, and because it's related to aging. There are now lots of studies that say the idea that the

01:18:01.560 set of proteins depends only on what mRNAs you've got is really a poor approximation. What Gary

01:18:10.420 Churchill and Rod Costanza and their buddies accomplished was, they actually looked at this

01:18:16.540 in the context of aging in a systematic way. They looked at a tissue like kidney, which is Ron's

01:18:22.160 special favorite tissue. They looked at kidneys of 6, 12, 18, 24-month-old mice, and they made a long

01:18:29.200 list of which genes change at the RNA level as aging progressed. And now they did the same thing,

01:18:36.380 but at the protein level, same tissue, same age, same genetic stock, the same mice as far as I know.

01:18:42.820 So now they have two parallel lists. And the old-fashioned default assumption would be

01:18:47.780 that if you are on the winner list for age sensitivity for the mRNA, you're going to be

01:18:53.460 on the winner list for the proteins encoded by that messenger RNA. And that was right 30% of the time,

01:19:01.400 not 100%, but 30%. There were big differences with age in the kidney and one other tissue. I don't

01:19:08.240 remember what other tissue they looked at. There were big differences in proteins as aging went on,

01:19:13.740 but only 30% of those changes were corresponded to the same change, same amount, same direction

01:19:22.040 in the messenger RNA. The rest came in somewhere else.

01:19:26.280 Is it post-translational?

01:19:27.840 That's what we don't know. There are a batch of possibilities. I've pointed to differential

01:19:33.460 translation. This is the Garcia's work on cap-independent translation, where the ribosome

01:19:39.560 doesn't care about the cap. It can bind to something else on some of the mRNAs and do those

01:19:45.640 instead, even if the cap is no longer working. So that could be selective RNA translation.

01:19:51.300 Now, there's also selective RNA sequestration. The RNA can be hidden and not made available.

01:19:57.380 There's selective RNA degradation. This plays a role. Once the protein has been made,

01:20:03.200 it can be degraded, chopped up into amino acids in many different ways. The proteasome can do that.

01:20:10.720 And Joe Endicott's specialty, there are lysosomes that can take in some, but not all proteins.

01:20:18.220 A fairly small percentage of the proteins get degraded by the lysosome. So there are changes in

01:20:24.580 RNA stability, changes in RNA location, RNA translation, protein degradation of many different

01:20:31.820 flavors that will, in important ways, modify the amount of the protein independent of any underlying

01:20:39.960 changes in the messenger RNA. I think more people learn about the non-transcriptional pathways that

01:20:48.120 mold the proteome, in our case, of course, in the context of aging and anti-aging drugs. But these same

01:20:54.300 principles apply to any disease-specific process, any health process, any drug response. I'm just talking

01:21:01.580 about it in aging terms. Once people buckle down and sort of learn this, they will not devote their

01:21:08.400 entire labs to analysis of transcripts. They'll pay more attention to proteins and also to the subtleties

01:21:16.240 of what happens between transcription of the DNA to the RNA stage, and then eventually the stable,

01:21:24.020 steady-state level of the protein.

01:21:26.400 Well, there's at least five. I lost count. At least five plausible explanations, and none of them are

01:21:33.060 mutually exclusive. So you could obviously have hybrids of these as to all the places where you can

01:21:38.660 quote-unquote go wrong between transcription and the final protein output.

01:21:43.500 Or go right. You mold it. Whether it's bad for you or good for you is interesting.

01:21:48.380 Let's go back to the speedometer, the aging speedometer. So you used these 10 known cases

01:21:55.340 of slower aging. So four genetic mutations that result in slower aging, the tried and true

01:22:02.580 caloric restriction, and then four drugs. So you've got these 10 slowly aging phenotypes,

01:22:09.020 and now you've identified consistent, very consistent, even down to sex-specific

01:22:15.800 differences. Is there a dose effect that you're seeing? Because presumably these 9 or 10 phenotypes

01:22:22.680 technically have slightly different lifespans and therefore are slightly differentially aging.

01:22:28.840 I'll address that question in the context of a battery of things we don't know, want to know,

01:22:33.840 and could find out in the next few years if we're given the opportunity. One of the things we would

01:22:38.820 really like to know is, what about the next three drugs? Do they do the same things to the same

01:22:44.360 aging rate indicators? That's a test of our hypothesis. If the answer is yes, yes, and yes,

01:22:50.020 that is great. If the answer is no, no, and no, something has gone wrong, and we need to reconsider

01:22:54.920 the whole foundational idea. Another important area, which is getting closer to what you were just asking,

01:23:00.720 is if you give a drug to a mouse, how long does it take for the aging rate indicators to switch?

01:23:07.660 If it takes a few months, that is terrific, because that means we can take 100 drugs and test all 100

01:23:15.220 of them, not for lifespan, which is really expensive, but test all 100 of them for the ability to switch

01:23:21.540 aging rate indicators. If we test 100 drugs off the shelf or that some colleague suggests to us,

01:23:27.080 and five of them switch all of the aging rate indicators, those are the five that are most

01:23:33.480 likely, we think, to be winners for the lifespan experiment. So we could use these as a screen

01:23:41.000 to try to identify drugs that are more likely than not to work in the context of a lifespan experiment.

01:23:49.120 We also don't know how long they stay switched. So let's say we give a mouse rapamycin or 17-ounce

01:23:56.400 estradiol or mystery drugs A, B, and C, and yes, all the aging rate indicators, eight weeks later,

01:24:02.520 they're at the slow aging level. Okay, we remove the drugs, the aging rate indicators, will they stay

01:24:08.820 where they are? We don't know that for drugs. We do know it, however, because of a great experiment

01:24:15.420 that we did with Andrzej Bartke. He was the inventor of the Amesdorf mouse. He's the first person who

01:24:20.760 showed they were long-lived. Andrzej had found, you take these Amesdorf mice, they're mice that have

01:24:26.300 very low growth hormone, very low IGF-1, and they live 40% longer. He has found, he published this 10

01:24:32.540 years ago, if you give them growth hormone shots when they're a little baby, just starting at two weeks

01:24:37.820 of age, and only for six weeks, so you stop giving them shots when they're eight weeks old.

01:24:42.640 That's enough to turn off the whole anti-aging program. They are no longer long-lived. We found

01:24:50.320 that they no longer have stress-resistant cells, and we found they no longer have low inflammation

01:24:56.140 in the brain. They go back to normal. So what we did was we got from Andrzej Bartke some 20-month-old

01:25:03.220 mice that were treated, but only when they were babies. So any epigenetic change that happened to

01:25:09.460 them, they'd have to do it in that growth hormone treatment period and remember it for 20 more

01:25:15.560 months. And the answer is that the changes in all the aging rate indicators easily seen in the 20-month-old

01:25:23.140 mutant mice, they all went away in the mice that had gotten the growth hormone shots in the juvenile

01:25:29.340 period. So the exposure to growth hormone shots by injection for a brief period of time in youth

01:25:37.260 was sufficient to lead to lifelong reversion of the aging rate indicators to the normal,

01:25:45.360 that is, away from the slow aging position. What we want to do now is the inverse of that. We want to

01:25:51.740 give them something good for them and see if we turn them on, that is, to slow aging, and then it stays

01:25:57.300 up forever. That's what we're hoping to see, of course. We can also use these for people. That's the sort of

01:26:03.180 next frontier. If you give these drugs to people, do the people change the aging rate indicators?

01:26:10.020 If the answer there is yes, that opens up, of course, a massively productive frontier for aging

01:26:16.380 research in people. One quick question about the Ames mice. What happens with the reverse experiment

01:26:21.900 when you wait until they are 20, call it even 12 months of age, and give them growth hormone once

01:26:30.300 they've fully matured? Do you shorten their life or revert their life back to normal?

01:26:35.660 No, we tried that. My lab failed. That is, we started our shots at four weeks of age. We did it

01:26:41.280 in Snelldorf mice, which are more or less the same. We started at four weeks, and then three years later,

01:26:46.540 we had found that it had no effect whatever. It took three years for the mice to age. It had no effect

01:26:51.180 on lifespan. So Bartke then did it starting at four weeks of age, and he failed too. That made me feel

01:26:57.280 great. We had just messed up. I would have given up. I had given up at that point. Bartke did not give

01:27:02.800 up. He did it over again, but this time starting at two weeks. When you start at two weeks, it works.

01:27:07.680 When you start at four weeks, it doesn't. Okay. Going back to the biomarkers, BDNF, DCX,

01:27:15.440 you're measuring those directly in sections of the brain. You're measuring those in CSF. Where are you getting that?

01:27:21.440 We take bits and pieces of the hippocampus, make a suspension of the proteins, and do a Western blot.

01:27:28.560 It's just measuring the amount of protein right in the brain. As you think about the application

01:27:33.360 of bringing this to humans, what would it look like to bridge that gap? In other words,

01:27:39.680 if you wanted to know if this type of exercise routine versus that type of exercise routine,

01:27:45.600 this type of diet versus that type of diet, or your home brew of rapamycin versus not is having

01:27:54.160 a benefit at some level, we will need to get this out of plasma. It will be very difficult to do this

01:28:00.480 out of CSF or even fat biopsies. How difficult to bridge is that?

01:28:05.440 I'll tell you about two steps we're taking. One of them is a collaboration with Stephen Cummings and

01:28:11.520 Theresa Mao and their colleagues. They have a project at UCSF called SOMMA, S-O-M-M-A. They

01:28:17.840 have a collection of several hundred human volunteers, all in their 70s, in good health. These people took

01:28:23.920 a lot of functional tests. How good are they at thinking? How fast are they? How strong are they?

01:28:28.720 And then they allowed tiny, tiny muscle biopsies and tiny fat biopsies. So we have requested,

01:28:36.320 I think, are likely to receive tiny bits of muscle and fat from these brave volunteers. And we will test,

01:28:43.520 do those that have a lot of the muscle-specific change, which is a protein called FNDC5,

01:28:49.600 and the fat-specific change, like UCP1, for instance. Our prediction is that amongst the 70-year-olds,

01:28:55.600 the really fit ones will be the ones that look as though they have always had youthful aging rate

01:29:02.080 indicators. So that will be one way in humans, and we'll have plasma from these same people as well.

01:29:07.680 So that will be one way in humans of beginning to test internal tissues to compare with plasma.

01:29:14.960 But that's impractical for clinical use. So what we really need now is ways of extending our results

01:29:20.560 to plasma. Two of the things that we can measure, we have measured and published in plasma. It turns out

01:29:26.640 that iresin, which is the product of FNDC5, the thing that goes from muscle to fat, that's in plasma.

01:29:33.600 And in fact, it goes up in all of our slow-aging mice. And the other is GPLD1, which I was talking about

01:29:39.680 previously. GPLD1 is also in plasma, and it also goes up in all of our slow-aging mice. So we'll be able to

01:29:47.280 evaluate that in human plasma samples. But we really want more than that. So one of the studies that we'll be

01:29:53.440 doing in the next few years with Katherine Kaczorowski and with a colleague named Costas Lysiotis at

01:29:59.680 Michigan, he's a metabolomics expert. We will take a batch of mice, healthy, young, UMHET3 mice,

01:30:06.640 the same kind that the ITP uses. We'll take blood samples from them. We will measure the aging rate

01:30:13.920 indicators in the blood, but also in the muscle, fat, liver, and brain of all these mice. And Costas

01:30:21.360 will measure several hundred metabolites in the blood of the same mice. And our goal will be to ask

01:30:28.720 which 2 or 10 or 20 or 100 metabolites, plasma metabolites, correlate with the plasma and the

01:30:36.080 internal tissue, ARIs. If we can derive from this exploratory exercise a list of 5 or 10 things you can

01:30:44.960 measure in mouse and human plasma that tell you where the ARIs would be internally, that's great.

01:30:53.360 We think it can be done in principle. There's a terrific younger scholar named Hamilton O working

01:30:59.120 at Stanford with Tony Weiss Coray. Hamilton has been able to deconvolute plasma signals by saying,

01:31:05.920 these represent changes in the pancreas. These represent changes in the liver. These represent

01:31:11.520 changes in the brain. He's not working in mice yet. I'm trying to twist his arm to get him to do it.

01:31:17.760 But in principle, it can be done. You will be able, we hope, to detect plasma molecules which correlate

01:31:24.960 with tissue-specific levels of ARIs. That will be the bridge to human studies.

01:31:30.000 Is he doing that by looking at cell-free DNA?

01:31:32.800 I can't tell you for two reasons. A, I don't understand it. And B, it was at a Gordon conference

01:31:38.880 and I'm not allowed to talk about it, but you could call Hamilton, his last name is spelled O-H,

01:31:43.920 or Tony Weiss Coray who runs the lab and maybe they'll tell you.

01:31:46.960 Going back to irisin or irisin, that is the product from what protein?

01:31:52.880 It is a cleavage product of a muscle protein called FNDC5. So we do two things. We measure FNDC5

01:32:01.840 as a protein in muscle and we measure irisin as a peptide or protein in the plasma. And they

01:32:10.640 always in our hands go up and down together. All the slow aging mice have more of the protein in the

01:32:15.520 muscle and more of the irisin in their blood. There was a study in Nature, I want to say it was about

01:32:21.600 2011, maybe 2012, that made all this promise that basically I think identifying that irisin

01:32:29.520 concentrations were high in people post-exercise. And the promise of the paper was we now have the

01:32:36.240 exercise drug. It was, we're going to just give people irisin, you're giving them an exercise pill.

01:32:42.800 The temptation I suspect for any of these things is the same, right? Presumably,

01:32:47.520 GPLD1 in a pill is something that someone's going to think, hey, that's got to be a good thing.

01:32:54.240 Do you think that's a good thing? And if not, why not? By good thing, I mean, do you think it's going

01:32:58.640 to have efficacy? Let's take a moral judgment out of that and just talk about sort of clinical efficacy.

01:33:03.040 Iresin has had a checkered past history. The original papers that said what you just said

01:33:11.360 turned out to be using an assay for iresin that was highly inaccurate in the sense that they

01:33:18.480 overestimated the actual concentration of iresin by a factor of about a hundred.

01:33:23.200 That's a problem.

01:33:24.080 Not subtle. Okay. And people who were skeptical of the original results, they actually worked out

01:33:29.520 a mass spec-based assay for iresin, gold standard, and proved that the original antibodies were not

01:33:37.120 specific enough to be useful to actually measure iresin levels. And so many people who had felt

01:33:43.520 those original papers were highly promising, as indeed they were, discounted the whole system.

01:33:48.800 It now looks as though they were throwing out the proverbial baby with the bathwater.

01:33:53.440 Now there are good antibodies that are sensitive enough to detect iresin at actual levels.

01:33:59.520 When Jin Ali brought me her iresin data, I said, oh, okay, I sort of believe it, but

01:34:05.680 iresin antibodies have a terrible reputation. Let's look at the precursor protein, FNDC5 in the muscle.

01:34:12.400 We decided not to publish it until we had both the iresin plasma and the FNDC5 by Western blot in muscle.

01:34:20.720 They were paralleling one another so very well, I believe both of them now, because they always

01:34:25.120 came out the same.

01:34:26.000 And why not just do the mass spec on the iresin in that situation?

01:34:29.600 Oh, we're no good in mass spec and we know how to do Western blots.

01:34:32.960 So back to the point, do we think that these molecules are merely biomarkers of all of the

01:34:39.440 myriad good things that these behaviors, drugs, or exercises do?

01:34:43.440 I will bet you, I don't have any secret inside dope. I will bet you that pharmaceutical companies,

01:34:50.080 thrilled with what Ozempic and its competitors are doing, have devoted tons of money to figuring

01:34:56.560 out whether they can get something like iresin into you in a way that doesn't hurt you and does you

01:35:01.360 some good. Neither you nor I is the first person to have thought of this idea. I'll bet the farm

01:35:06.320 companies have devoted tons of money into looking at that. I think it's a highly promising area of

01:35:12.880 research, although I imagine a lot of it currently is proprietary. GPLD1 is a protein and swallowing

01:35:21.120 a pill might not work because it might be digested in the stomach, just like every bit of meat you eat

01:35:26.000 is digested to amino acids.

01:35:27.680 The same is true with iresin as well. It's also a peptide, isn't it?

01:35:30.240 Yeah, yeah.

01:35:30.960 So these are going to need to be injectables, presumably.

01:35:33.200 Yes, unless you can come up with a small factor that turns on FNDC5 in your muscles.

01:35:38.880 We have those. They're called anti-aging drugs. Rapamycin does that.

01:35:44.240 That's sort of the part here that I'm really trying to wrap my head around. Philosophically

01:35:47.920 is the wrong word, but metaphysically, I suppose, because on the one hand, we have molecules

01:35:54.960 that are now doing things that are impacting aging at a fundamental level. Again, this is counterintuitive.

01:36:03.200 It's not counterintuitive to me. Exercise or calorie restriction induce a longer life in the

01:36:10.480 right model. It's a little counterintuitive to me that rapamycin does, to be completely honest.

01:36:15.680 I don't dispute it for a moment because I've seen it now over and over and over again, just as you

01:36:20.160 have. But it's still remarkable to me that a molecule is able to act at a fundamental level

01:36:26.640 of aging as opposed to way, way, way downstream in the way that a lipid lowering drug works,

01:36:34.080 where it works on one disease pretty much, and it works through one path. And the proof of that is

01:36:40.720 indeed your aging accelerator. That is, in fact, the proof of giroprotection. It almost becomes the

01:36:47.760 sin-quan-on of a molecule being giroprotective versus simply targeting a disease.

01:36:54.160 Let me phrase that in another way. If you have a drug that extends mouse lifespan, I think that's a

01:36:59.760 critically important step towards making a case that it's slowing aging, but it's not the last step.

01:37:04.720 I would not fully endorse that hypothesis unless someone has shown that the mice treated with that

01:37:10.800 drug, in addition to living a long time, they also retain lots of youthful function.

01:37:17.440 Their muscles are great. Their hearing is great. Their cognition is improved. Their bones are better.

01:37:22.400 We've gone through all those steps for the Snell dwarf mice, the Amesdwarf mice,

01:37:26.160 for the calorie-restricted diet, for the growth hormone receptor knockout mice.

01:37:30.480 We're beginning to make that kind of a story for acarbose. Our first acarbose paper had

01:37:36.720 grip strength and blood glucose control. We're beginning to make that case for 17-alph estradiol

01:37:42.400 as well and acarbose. But building those cases brick by brick by brick is really necessary to say it's not

01:37:48.880 merely an anti-cancer drug, something that was a broad-spectrum anti-cancer drug. I'll bet people are

01:37:54.720 interested in that, but it's not necessarily an anti-aging drug. To make me happy, it has to be

01:38:00.640 an anti-aging drug, and the evidence has to be effects on many different age-sensitive properties.

01:38:06.560 But we have great evidence now for at least three mutants, and the calorie restriction diet, and

01:38:13.040 the methionine restriction diet. And we're getting there for rapamycin and several of the other drugs

01:38:18.960 that came along five years, eight years after that. So I think there will be a very strong case that

01:38:23.600 these drugs are acting by slowing the aging process and delaying maybe not quite all,

01:38:29.360 but maybe all of the aspects of aging that make people unhappy about getting older.

01:38:35.440 And I agree with you. It's a fundamental reorientation of instinct. That's what this

01:38:40.320 experimentation is designed to do. It's designed to reset one's instinct on these points.

01:38:46.080 We've already talked a lot about why the black six mouse has a lot of problems. You've also alluded to

01:38:50.400 the fact that they're basically genetically programmed to die of cancer. What is the

01:38:55.440 natural history of your mice? And how does the natural death of the mice in the control group

01:39:02.080 by cause, we've already obviously talked about length of life, but what is the cause of death

01:39:07.280 in the controls typically versus the treated in the success cases?

01:39:11.280 The context that's necessary here is that nearly all of the mice that are available throughout the

01:39:16.800 world for medical experimentation come from the Jackson laboratory, which for 30 or 40 years

01:39:23.200 was mostly interested in cancer. So whenever a mouse got cancer, they kept it. So most of the strains-

01:39:29.120 They were positively selected for cancer.

01:39:31.040 They were selected for getting a lot of cancer. UM had three mice, had four different grandparents,

01:39:36.880 and cancer is the cause of death in about 80% of our mice, but it's varied. Some sort of lymphoid or

01:39:44.000 leukemia cancer, maybe 30 or 35% of the deaths. In the males, pulmonary cancer, liver cancer are both

01:39:51.760 prominent. In the female, it's not pulmonary so much, but breast cancer, liver cancer again, hemangiosarcoma,

01:39:59.680 and then maybe 30% of the mice, if both sexes, die of 1% will die of this kind of cancer, 2% will

01:40:06.400 die of this kind of cancer, 3% of that kind of cancer. 80% of the time, it's some sort of neoplasia

01:40:12.640 that is the lethal injury. And that's why one could make initially a case, all these drugs are doing

01:40:20.240 is slowing down every single kind of cancer. That's why we have to look at a lot of things that are not

01:40:24.480 cancer. The second part of your question is, does the proportion of different kinds of cancers

01:40:29.520 or causes of death change in these different mice? That's a hard question to answer because

01:40:36.160 when we do a necropsy series, at the end of life, you have to do it at the end of life

01:40:41.600 to see what they died of. We usually have only about 60 mice in the treated group and about 60 mice in

01:40:47.520 the control group. So if a particular kind of cancer, let's say liver cancer kills 10% of the mice,

01:40:54.160 you only have six cases. If that goes up to nine or goes down to three, that would answer your

01:40:59.760 question with a yes. But statistically, we just don't have enough cases to be confident. Only

01:41:05.440 once did we come across a statistically significant alteration and could have been a fluke. We had a

01:41:11.280 drug that was extending lifespan in males and females, but it did not increase the age at death of the

01:41:18.480 females dying of breast cancer. So one could have made a case that breast cancer is caused by

01:41:24.720 something that is not related to this drug's anti-aging mode. That's something we're always

01:41:30.800 on the lookout for, but the number of autopsy cases is almost always too small to really have a good

01:41:38.320 grip on it. Well, that's why I think it makes it very interesting that we're going to see your

01:41:44.160 colleague coming on board who specializes in the neuroscience and the neurology of the mice, because

01:41:53.120 I think to make this truly exciting, we want to continue to see functional improvements. And

01:41:59.920 functional improvements in both strength and cognition would go a long way. And by the way,

01:42:07.360 it also is worth, I guess, asking, I haven't asked you this, but do you know if there are cases of drugs

01:42:14.800 that do not improve lifespan, but do improve, say grip strength and treadmill time? In other words,

01:42:21.840 are there drugs that are improving health span without lifespan in the ITP that you've documented?

01:42:26.080 This is a chestnut that always comes up, and the Aging Institute in particular is passionately

01:42:33.040 interested in the notion that maybe a drug will make you healthy and then you'll drop dead right

01:42:37.280 on schedule. The answer to your question is no one has looked for it, and that's because in our phase

01:42:41.040 one assays, we don't do it. We only do those detailed studies on things that extended lifespan.

01:42:46.560 The reason that things extend lifespan is basically it postpones all the bad stuff that lead to death.

01:42:52.640 So we could screen a lot of drugs for age-sensitive variables and the hopes that we would find one

01:42:59.040 that made age-sensitive variables go away but didn't have any effect on lifespan. I'm not so sure that we

01:43:05.440 would find any. It's trivially easy to do that. If you teach a mouse to do push-ups, you will postpone

01:43:12.400 age-associated changes in muscle mass and muscle strength. And if you teach them they won't get

01:43:18.720 food until they solve a maze, they're going to get pretty darn good at solving that maze. So

01:43:24.240 system-specific postponement of age-sensitive outcomes is not too hard to achieve. It's not

01:43:32.240 really relevant, I think, to the issue of what you can do to postpone all the aspects of aging together.

01:43:39.120 Let's pivot and talk about one of the successes of the ITP that still somewhat perplexes me. You've

01:43:44.560 already alluded to it several times, which is 17-alpha-estradiol. So I'm pretty sure I asked

01:43:49.600 you this question last time, but I will ask you again. Remind me the difference between 17-alpha-estradiol

01:43:55.200 and 17-beta-estradiol, which is the estradiol that is the dominant form in both males and females.

01:44:01.840 We're rolling around with lots of 17-beta. Traditionally called estrogen. So the 17-alpha-estradiol

01:44:08.960 is just the same chemically as the 17-beta, except for one of the bonds, instead of pointing up out

01:44:14.640 of the plane, points down in the opposite direction. So it's a stereoisomer. Same chemical formula,

01:44:20.640 all the atoms are attached in the same place, it's just that two of them are pointing up instead of

01:44:24.960 pointing down. And because of that manipulation, it doesn't bind very well to the traditional,

01:44:30.960 famous estrogen receptors. So it's doing something, it's got to be binding to something.

01:44:37.680 But it probably is not the traditional estrogen receptors, or it might be that plus something else,

01:44:44.480 to get an effect on estrogen-sensitive tissues. You can do it with 17-alpha-estradiol, you just have

01:44:50.320 to use a lot more. I think tenfold more is what Jim Nelson found when he did that titration.

01:44:55.440 And what was the rationale when this was proposed to the ITP? What was the scientific

01:44:59.920 rationale for why this would be a JIRA protective? Jim Simpkins recommended it. He's a steroid

01:45:05.360 physiologist, neuroendocrinologist. He reasoned, and a lot of this was wrong. This is the rationale.

01:45:12.960 Estrogens are good for you. That's why females live longer than males. Let's find an estrogen we can

01:45:18.000 give to males. We don't want to give them 17-beta-estradiol because they'll turn into girls and they won't

01:45:23.920 like that. No one wants to be a girl. So let's use 17-alpha-estradiol because they won't turn into

01:45:30.800 girls. It does not turn on secondary sexual characteristics. Maybe it will do all the

01:45:34.640 good stuff that estrogen 17-beta actually does. That was Jim's argument. Very plausible. Now,

01:45:42.160 it turns out that if you give 17-alpha-estradiol to male mice, it pushes their lifespan way beyond

01:45:48.640 females. It's not merely mimicking the good stuff, if there is good stuff, that estrogen 17-beta does

01:45:56.880 in females. If so, it wouldn't go much further than females are and it goes well beyond, significantly

01:46:03.200 beyond normal females or drug-treated females because the drug doesn't affect female longevity

01:46:09.280 at all. What it binds to in which cells, in which tissues, what it's turning on biochemically,

01:46:15.760 is at this point quite obscure. There are at least two labs that I know of, Mike Stouts. I think Mike

01:46:21.360 is now in Oklahoma. And my former student, Mariana Sadogursky. Mariana is at Wayne State. They've

01:46:28.800 published some really nice papers getting at the issue of what is 17-alpha-estradiol actually doing

01:46:34.880 physiologically. What does it bind to? Mariana, her last three papers and her just-awarded grant are

01:46:41.200 focused on what 17-alpha-estradiol does in the brain, what it does to estrogen-sensitive and

01:46:46.320 estrogen-insensitive parts of the brain. So there are a small number of labs, I wish there were more,

01:46:51.520 that are diving into that question. What is the target? What is the receptor? What is the

01:46:56.240 physiological effect? The more people that work on that, I think the happier we will be.

01:47:01.040 And again, 17-alpha-estradiol is as potent in males as rapamycin?

01:47:07.280 I'd have to double check. I think it's about a 19% increase. Don't quote me on this. I need to look

01:47:14.760 it up. Yeah, we'll have it in the show notes, but yeah.

01:47:17.120 Yeah, something around there in the optimal dose in males. The original dose of rapamycin is there

01:47:22.920 or slightly below that. We now have a better result with rapamycin when we combine it with

01:47:28.160 a carbose. We can kick the male lifespan up to 29% increase.

01:47:32.520 Unbelievable.

01:47:33.560 That's our winner.

01:47:34.940 That's simply unbelievable.

01:47:36.880 It's the largest percent increase we've ever gotten. And also it's the first time we've gotten

01:47:40.840 an increase by combining two drugs together. So that's the best we've been able to do. And so

01:47:46.720 far we can't get 17-alpha-estradiol up to 29%, but I think we can get it up to 19%.

01:47:51.600 Have you combined 17-alpha-estradiol with rapa yet?

01:47:54.780 What a good idea. We're testing that now.

01:47:56.940 Are you? Okay, nice.

01:47:58.000 Yeah. I mean, several other groups are testing it also because it is a good idea. I'm not making

01:48:02.820 fun of you. It is such a good idea that we're trying it. And I know of at least one startup

01:48:07.240 company that's trying that as well.

01:48:09.220 There's something about this 17-alpha-estradiol story that is so fascinating to me.

01:48:13.280 One, there's the total lack of clarity around the mechanism of action. And then there's this

01:48:18.520 sex difference, which can't be attributed to any mimicking of 17-beta-estradiol given the two

01:48:26.820 facts you mentioned, that the males leapfrog the females and the females accrue no benefit.

01:48:32.580 Yeah, I guess I'm a little bit disappointed to hear there are, I guess, only Mariana and Mike

01:48:36.100 working on this problem.

01:48:37.560 There may be others. I mean, those are just friends of mine that I know of. There may be

01:48:40.680 others as well. I don't know the literature well enough to tell you. I can give you an insight into

01:48:45.460 what we thought was going on with the 17-alpha-estradiol and how badly wrong we were.

01:48:52.140 Mike Garrett, whom I've mentioned earlier, he collaborated with a guy named Mo Jane to look at,

01:48:57.280 among other things, steroids in the tissues of mice treated with 17-alpha-estradiol. And he noticed

01:49:03.560 something really interesting. He found two steroids. They were members of the estriol family, not

01:49:09.460 estradiol, but estriol, that were elevated at least 20-fold in males that got the drug. And they

01:49:17.420 were not elevated in females at all. It was a male-specific production of estriol when 17-alpha-estradiol

01:49:27.600 was given to the males. And we knew it was sex-specific because if he castrated the males

01:49:32.660 before the drug, you didn't see the estriol production, the conversion from estradiol to

01:49:38.480 estriol depended upon testosterone or some other testicular hormone. So we said, okay, great. We've

01:49:44.500 got a winner, estriol. That's going to be the one that is going to work in both males and females.

01:49:49.740 That's the active ingredient. And the data set that's 90% complete, and we'll probably start writing

01:49:56.900 it up in a month or two when we have 90% of the mice dead, but we have 50% of the mice dead,

01:50:02.660 we've presented at meetings and I'm allowed to talk about it, says that that guess was partially

01:50:07.840 right and partially wrong. The hydroxy version of estriol is great for males. It's actually at least

01:50:15.560 as good as 17-alpha-estradiol. We won't know until we have the last few deaths, but it's terrific.

01:50:21.240 That was a good guess. You don't need 17-alpha-estradiol because the estriol works terrific.

01:50:26.420 However, we thought it would work in females, and it is the first drug we've found that diminishes

01:50:32.140 lifespaned females. So the idea that it would work to benefit females was wrong. It is for

01:50:38.600 mysterious reasons harmful in females. So we really don't know yet.

01:50:43.820 And to be clear, this is just straight estriol, E3.

01:50:47.740 It's 16-hydroxy estriol.

01:50:49.500 Okay. Why did you pick 16-hydroxy as opposed to 4-hydroxy, 2-hydroxy, or just pure estriol?

01:50:54.800 Was that because that's the only one that went up with the administration of 17-alpha-estradiol

01:51:00.140 in the males?

01:51:01.420 You've stumped me. I don't know the answer. Mike Garrett, who told us which estriol to buy,

01:51:06.600 does know the answer. And we could ask him. I don't know what prompted Mike. Maybe it was

01:51:12.280 commercial availability. Maybe it was prior studies of toxicity in mice. Mike has his reasons. And I

01:51:19.040 read the application three years ago, and I don't remember what specifically led him to suggest

01:51:23.820 this compound. It wasn't just stab at the sigma catalog.

01:51:28.620 The female mice that are dying at an accelerated rate, anything specific about the manner of death?

01:51:34.580 We don't know yet. We haven't done any necropsies.

01:51:36.600 Okay. Let's talk about a couple notable failures in the ITP, i.e. drugs that everybody thought were

01:51:42.920 home runs that didn't pan out. I guess the three that come to my mind are resveratrol,

01:51:47.840 metformin, nicotinamide riboside. Any others that should be on that list?

01:51:52.460 Most people who suggest drug think their drug is going to work. But I think the three you've pointed

01:51:56.620 out are the ones that have gotten the largest numbers of notices in AARP bulletins and on social

01:52:03.460 media and at the conventions where people want to mingle with snake oil salesmen. So they are certainly

01:52:11.280 the most famous. And I think there's a different level of enthusiasm for each one of them. Metformin,

01:52:17.540 I think, has been very sensibly proposed as a potential anti-aging drug in people. I don't know

01:52:24.780 enough about its benefits and side effects. I know that you yourself have 10 times more information

01:52:29.520 about this than I do. But a case has been made because it's so very safe in people that it could

01:52:35.180 be used in people to postpone aspects of aging. I can see reasons not to believe that. But at

01:52:41.780 least you can make a case for that. It doesn't seem to work in mice. The ITP showed that it didn't

01:52:48.020 work in mice. And now several other groups have confirmed that result. Rafa de Cabo at one point

01:52:53.360 claimed that it worked in mice. But he used a very weird statistical test. And I suspect that if he had

01:53:00.140 used the standard statistical test, it would have failed in his lab as well. I haven't seen the data,

01:53:06.060 so I'm not sure of that. But that's my guess. Resveratrol was hyped for many years. People often

01:53:12.940 with commercial interests or who had a grant or who wanted to get a lot of money for a clinical trial

01:53:17.480 would start their talk with a beautiful bottle of red wine and then say resveratrol is in red wine

01:53:22.340 and sirtuins are important and resveratrol influences sirtuins. And just take some of my resveratrol

01:53:28.820 or sirtuin activating agent and you'll live forever. None of that was right. I mean, it's

01:53:33.820 been shown very clearly now that the amount of resveratrol in red wine, to get enough of it,

01:53:39.100 you need to drink 30 bottles a day. Its status as a sirtuin activator has been questioned by very

01:53:46.440 serious and skilled biochemists. The original data on worms has been disconfirmed by a couple of very

01:53:52.580 good labs. So it was mostly hype. People made a lot of money by selling companies that

01:53:58.820 had an interest in sirtuin activators. We tested it because the director of the National Aging

01:54:05.500 Institute, Richard Hodes, for the first and last time said, you will test resveratrol or you will

01:54:11.560 not get any money this year. We said, yes, sir. Yes, sir. So we tested it. We checked with David

01:54:16.180 Sinclair and asked David, what is the concentration we ought to use? He said, use this concentration and

01:54:21.140 this concentration. We said, sure, we'll do it your way. Let's find out. And it didn't work. And

01:54:25.660 subsequently, many groups now, including groups that Dr. Sinclair is associated with, have shown

01:54:31.280 that it doesn't work to extend lifespan of regular mice. Famous paper was one in which the mice were

01:54:36.920 poisoned with a 60% coconut oil diet. And they weren't dying of aging. They were dying because

01:54:42.740 their liver swelled up to the point that it crushed their lungs and they couldn't inhale. They couldn't

01:54:48.720 breathe. This is not, I believe, a pretty good model for the aging response. So I think the evidence

01:54:54.900 that resveratrol by itself should have been tested was quite weak. And the fact, the evidence that it

01:55:01.320 works is very bad. It almost certainly doesn't do anything, at least in mice, in terms of aging.

01:55:08.240 Does it surprise you how ubiquitous resveratrol supplements are still on the internet?

01:55:13.260 Sorry to be cynical. People are very easy to fool. It's easy to come up with eight or 10 things that

01:55:18.880 people believe because they read them on the internet or they watched them on Fox News or whatever.

01:55:23.180 And they're just wrong about this. But people are very, very gullible. The anecdote about

01:55:27.420 resveratrol that I think gives you a sense of what that time was like, I had a friend,

01:55:31.940 a neurologist at Michigan, who had been given a huge grant to give resveratrol to Alzheimer's

01:55:37.560 patients at the early stage to see if it would slow Alzheimer's. Tons of money. And he came around

01:55:43.760 to a meeting of resveratrol biologists. I was attending it. And he asked people,

01:55:48.520 what dose to use? And the range of suggested doses as milligrams of drug per person per day

01:55:56.800 ranged over a million fold. That is among the experts, the world's experts on resveratrol,

01:56:04.520 the consensus ranged from one to a million as to what dose was the most logical was. So this is a

01:56:10.920 sign of a field that is making it up as it goes along. NR is the last of the three drugs that you

01:56:16.880 mentioned. I had high hopes. Probably more hype with NR than resveratrol, truthfully,

01:56:21.500 on some level. Somebody's making a lot of money.

01:56:24.100 Yeah, many people are. So tell us about the findings and what it tells us or doesn't tell us.

01:56:30.000 Well, we tested it and it didn't work. That is, it didn't extend mouse lifespan.

01:56:34.340 Some people said, oh, well, you have to use NMN, a metabolite. That would have a

01:56:39.780 more bioavailability and better profiles. And this is reasonable. And if someone makes a good

01:56:45.300 case that we should test NMN and we could afford it, the commercial sources are expensive.

01:56:51.260 We would probably test that as well. I think.

01:56:53.880 I would have to imagine that some commercial party would donate the drug at this point, right?

01:56:58.880 We are in negotiations with one such company. Yes, I'm hoping you're right.

01:57:02.720 I'm hoping we could accomplish that. I'm not a non-believer.

01:57:06.220 I am a non-believer of resveratrol. For NM and the whole nicotinamide modulating family,

01:57:13.420 I think the book is still open. And there's a reasonable chance that some really good stuff

01:57:18.440 could come out of that. It might be that you'd have to give an enzyme that modified an inhibitor

01:57:24.420 of one of the metabolizing enzymes or a different form. I have a colleague who has suggested,

01:57:30.920 I don't know if this is public yet, but there's someone who suggests that NR may work in combination

01:57:35.400 with another drug. His ideas are good ones and they've been accepted by the ITP. We're going to

01:57:40.600 try the NR plus something else that this colleague has recommended to us. Although NR by itself did

01:57:48.120 not extend mouse lifespan, it could be that some other trick will lead to physiologically important

01:57:55.420 modulation of NAD availability in some cell of interest. It could be that what counts is changing

01:58:02.180 availability in a cell in the hypothalamus or in the pancreatic beta cell or in the lymph nodes or

01:58:08.400 something. Finding a dose that is appropriately good for the cells that count but doesn't produce

01:58:14.280 side effects in other cells may be tricky, but it might work. I'd love to test other things in that

01:58:21.340 general nature. And of course, I mean, it goes without saying, I should have said this earlier,

01:58:25.960 the fact that something fails in mice doesn't mean it's going to fail in people. Testing it in people is

01:58:30.720 going to be much harder. It's easier to sell stuff that's untested. But in principle, one could actually

01:58:36.940 test it in people and see if it does anything good. I think with those three failures, if I'm going to

01:58:43.380 summarize your point of view, I think the failure of resveratrol in the ITP, I think you view as

01:58:50.600 dispositive that that drug never worked in any circumstance anywhere. It doesn't work in humans.

01:58:55.900 It doesn't work in mice. And you just demonstrated that more.

01:59:00.460 Our evidence is not dispositive. That is, it could be true that it doesn't work in mice,

01:59:05.040 but it works great in people. Or it could be that it would work in mice at a 20 times higher dose.

01:59:09.460 Yeah, but I think your point is the plausibility for that is very low, given that it fails over and

01:59:16.480 over and over again.

01:59:17.660 It's one brick in building the case. Resveratrol has been overhyped. If you look in detail

01:59:22.880 at the evidence suggesting it has health benefits, most of those studies

01:59:26.860 are unconvincing. And many of the ones that are convincing were submitted by people who are trying

01:59:32.720 to sell something.

01:59:34.400 And by the way, even putting that aside, the only study that's really convincing is the one you

01:59:38.440 described with the force feeding of coconut oil. So that's sort of problematic. What I hear you saying

01:59:43.960 with NR and metformin is even if we repeat these studies over and over again in ITPs, which you won't

01:59:52.160 directly, but you will potentially in other combinations, you have more faith in the

01:59:57.380 possibility that those could still be viable in humans.

02:00:00.820 Yeah. I mean, the theoretical case that metformin might be good for you, that's plausible. It's not

02:00:07.440 completely proven, but it's sensible. And the same is true for things that are attempting to rescue

02:00:12.800 age-associated changes in NAD. I'm no expert in either of those fields, but the little bit I know is

02:00:18.980 consistent with what these sponsors are saying. There's a good, plausible case. We get, as I said,

02:00:24.800 you know, in Goodyear, 20 or 25 applications for eight or 10 of them. There's a good, plausible case

02:00:29.940 to be made that this drug deserves testing. And most of those good, plausible cases yield negative

02:00:35.880 results. And that's expected. That's one of the nice things about aging rate indicators. If they work,

02:00:41.700 if they are flipped by drugs in a short period of time, then we hope our hit rate will, right now,

02:00:47.000 it's about 10% give us big effects and a total of 15% give us significant effects. If we can get

02:00:53.160 that up to 50% by pre-screening with aging rate indicators so that half of the drugs we throw into

02:00:59.860 lifespan studies actually give a lifespan benefit, that would be nice.

02:01:04.740 So Rich, I don't know if you know this about me, but 15 to 20 years ago, I used to spend an awful

02:01:09.940 lot of time on boats. These weren't big boats. These are typically small boats. So these are kind of

02:01:16.200 30 foot boats where you're out in really, really, really rough water. So hours and hours. I don't

02:01:23.200 know how many hours of my life I've spent 30, 40 miles off the coast of California getting thrown

02:01:29.060 around like crazy. And I was pretty lucky. I've only been seasick once in my life, which when you

02:01:37.220 consider the amount of time I've spent out there, I consider pretty fortunate. But most of the people

02:01:41.840 I spent time with out there got seasick a lot more. And one of the drugs that people used to

02:01:48.680 take care of their seasickness was an over-the-counter drug called meclizine. So I would

02:01:56.120 occasionally take it. I didn't take it that often because again, I almost never got seasick. But

02:02:00.760 sometimes just to be safe, I would say, you know what? The water looks unbelievably rough today.

02:02:06.100 I'm going to take my meclizine. Over-the-counter. Bonine, I think, is the brand name.

02:02:10.820 B-O-N-I-N-E. Bonine. Yep.

02:02:13.100 Bonine. Yeah. Why am I telling you this story, Rick? I'm sure the listener is wondering. Why

02:02:17.480 is Peter bringing this up?

02:02:19.500 I'm spending two weeks in December in a small boat, actually, off the coast of Chile. We're going to go

02:02:24.800 up and down the archipelago doing photography. And the seas there can be very choppy. And we've been

02:02:30.480 told to take meclizine or some other anti-sea sickness remedy so that we'll be able to photograph

02:02:36.880 to our heart's content without feeling ill. And meclizine is also included in the latest

02:02:42.640 paper for the ITP, which also makes it of interest. The paper that we've just submitted,

02:02:47.760 it hasn't been accepted yet. But again, because it's complete, I'm allowed to present it at meetings

02:02:53.060 and in talks like this. There are two drugs. One is meclizine. One is astaxanthine,

02:02:58.000 which in males led to a significant increase, about 10% in the lifespan of the males. Meclizine

02:03:07.080 was suggested to us by Geno Cortopassi. He knew that rapamycin was good as an anti-aging drug and

02:03:15.580 it was a TOR inhibitor. So he took several thousand FDA-approved drugs. And in a tissue culture

02:03:22.820 he said, which of these inhibit TOR? Maybe a safe drug that inhibits TOR could find a place as an

02:03:31.480 anti-aging remedy. And at the top of his list, to everyone's surprise, certainly his was meclizine.

02:03:37.340 It was famous as an antihistamine and also has mysterious CNS effects, which is why it's so good

02:03:43.060 for seasickness. It wasn't, I believe, known as a TOR inhibitor, but Geno found that it was.

02:03:48.480 And so he suggested that we test it. And it does indeed lead to, at the dose we used,

02:03:54.280 a significant increase, about 10% in lifespan of the male mice. It did not affect females. So we're

02:04:01.760 going to try it again at higher concentrations and see if we can get that to go. We haven't proven

02:04:06.700 that the effect is mediated by TOR inhibition. It has a lot of CNS effects and maybe the good stuff

02:04:13.640 that it's doing to the male mice is unrelated to TOR. Maybe it has to do with changes in serotonin

02:04:19.880 or histamine production in some critical nucleus in the brain. We need to now look at that.

02:04:25.320 That's the cool news. The other thing, that same paper has the two points of interest. One is

02:04:30.700 astaxanthine. Astaxanthine is also available over the counter. You can buy it in your local drugstore.

02:04:36.620 It is alleged to have health benefits, but they're all over the map and I can't assess the strength of

02:04:43.680 the evidence for health benefits. It's also alleged to have many different physiological effects like

02:04:49.500 an antihistamine or an anti-inflammatory or an antioxidant. It's been alleged to do a little

02:04:55.860 bit of everything. And it's a food dye, isn't it? We eat it all the time. It's famous because it's

02:05:00.640 what's turned salmon pink. If you want to make your farm-grown salmon pink, you dump tons. I mean tons.

02:05:06.560 Dump truck loads of astaxanthine into the water and the salmon turn pink. It's not that you dip the

02:05:14.020 salmon into astaxanthine. Natural salmon eat a lot of crustaceans, which have this stuff in their

02:05:19.840 shells. And you can mimic that effect in farm-grown salmon by giving them synthetic or naturally

02:05:26.260 derived astaxanthine. In any case, there's a company in Hawaii that believes it might have health

02:05:31.200 benefits. And they asked us to test it and we did test it. And that also has a significant

02:05:36.040 effect and also it's males only. So this paper hopefully will be accepted soon and out for

02:05:42.880 people to judge the strengths of the evidence. Neither of these led to a change in our measurement

02:05:49.280 of maximum lifespan, this test of percentage of live mice at the 90th percentile. Possible that at a higher

02:05:55.940 or maybe even a lower dose, it might have done that. We're going to need to go back and do a more

02:06:01.120 complete dose response curves. The reason this paper is, I think, likely to be of particular interest

02:06:07.960 to the general public is that it is the first time we've gotten winners that you can buy without a

02:06:14.360 prescription, over the counter. They're not as strong in terms of lifespan benefit as the four drugs we

02:06:19.680 were talking about earlier on in our discussion. And of course, we don't know if they will work in

02:06:24.660 humans at all. But the position of the FDA that you can't test a drug claiming it has an anti-aging

02:06:32.060 effect will be of less relevance if some of the over-the-counter non-prescription medicines

02:06:38.720 actually slow aging, first in mice and then maybe also eventually in people. I think that's an

02:06:46.560 interesting political and legal question in terms of the science. We now need to figure out what is

02:06:52.540 astaxanthine doing and is meclizine acting through TOR or through some other target. As any new drug

02:06:59.040 does, it opens up new possibilities for mechanistic exploration. You can bet we're going to be looking

02:07:04.900 at the agey grade indicators in tissues from meclizine-treated and astaxanthine-treated mice.

02:07:11.580 The last thing in the paper that many people will be interested to know is we've tested

02:07:16.600 phycetin. This was suggested to us by Paul Robbins and Jim Kirkland and Tamara Chikonia and their

02:07:23.120 colleagues. Phycetin is undergoing a lot of human trials because of claims that it is a senolytic

02:07:30.760 drug. There are some people, though I am not in that group, who think that there's such a thing as a

02:07:35.760 senescent cell and that you get a lot of them when you get old and that they're bad for you and that a

02:07:41.740 drug that removes senescent cells therefore will be good for you. I'm not convinced, but this is a

02:07:47.060 very popular line of thought. So phycetin was given to us as a drug to test the hypothesis that if you

02:07:54.340 remove senescent cells from mice by giving them phycetin, they would live longer. I thought it

02:07:59.300 wouldn't work, but it was a very reasonable and important thing to do. So we gave phycetin two

02:08:04.520 different dose regimes suggested by Dr. Kirkland. He had found in his lab with his kind of mice that they

02:08:11.000 did work at this dose, so that was good news. We thought we were trying to replicate his stuff in

02:08:16.220 our mice at much larger scale and the take-home messages were two. First, it had no effect

02:08:21.800 whatsoever on lifespan of male or female mice using either of the dosage regimes that Dr. Kirkland

02:08:28.960 recommended. However, that is not dispositive because it turns out it didn't remove any senescent

02:08:33.920 cells either. What we can do is there are quite a number of surface markers like P16 that the

02:08:40.780 senescent cell gurus say, this will tell you whether you've got senescent cells. They do go

02:08:46.680 up with age. They go up with age, and we looked at three different tissues. That is Paul Robbins' lab

02:08:51.760 and Jim Kirkland's lab. We sent them tissue. It was blind, so they wouldn't know which ones were

02:08:56.860 controls and which ones were treated. They sent us back this number of beta-galactosidase positive cells,

02:09:03.000 this number of P16 positive cells, this number, this amount of P21. And when we unblinded it,

02:09:09.440 there was no effect of phycetin on brain, on liver, on muscle, on kidney, at either lab for any marker

02:09:17.100 that we looked at. So we thought we were testing the notion that removing senescent cells would be

02:09:23.220 good for you, and it turned out we were testing the notion that phycetin removes senescent cells.

02:09:29.180 Now, maybe it works in people. This is not my area, but our colleagues were disappointed,

02:09:33.620 and I can understand why they were disappointed in both results.

02:09:37.760 Rich, say more about your lack of belief around the role of senescent cells. You made it sound like

02:09:43.680 you don't believe in senescent cells. Was that actually what you meant, or did you mean you don't

02:09:47.800 believe that senescent cells drive aging, or that you don't believe removal of senescent cells slows or

02:09:54.080 reverses an aging phenotype? Say a bit more on that whole thing.

02:09:57.280 I'm in a slightly awkward position because to tell you the details, all that is true,

02:10:03.100 and the details require...

02:10:05.200 Just don't say you'll have to kill me if you tell me. That's the only thing I don't want to hear.

02:10:08.380 No, no, no, no. I don't do that. It's a long, long story, but let me give you an analogy.

02:10:14.220 If someone says, do you believe in stress? The answer is, sure, I believe in stress. But that

02:10:20.360 doesn't end the conversation because there's the stress of being about to undergo a tense podcast

02:10:25.740 discussion, or the tension of having to give a talk before the National Aging Institute's committee,

02:10:32.140 or the stress of dental work. I hate dental work. Or the chronic stress of being locked into a marriage

02:10:38.120 or a job that you really hate, or getting a diagnosis of cancer, or dropping a bit of poison

02:10:45.420 into your GI tract. All of those produce stress, but they're radically different things. They produce

02:10:51.960 different physiological effects. And to say, do you believe in stress, or is this caused by stress,

02:10:57.500 is a way of blinding yourself accidentally to the critically important distinctions.

02:11:03.620 So, do I believe that there is such a thing as a senescent cell? Yeah. If you take cells in human cells

02:11:09.140 in culture, they stop dividing, and those have been called senescent cells. They exist. No question about it.

02:11:15.420 And they're caused by telomere shortening. Now, if you take another kind of cell and you zap it with

02:11:20.980 x-rays, you get a different kind of cell. They make different proteins. They don't have telomere

02:11:26.640 problems. People have referred to those as senescent cells. If you have cells that when you get older,

02:11:32.400 they have P16 on them. People have referred to those as senescent cells. And by saying that this drug

02:11:39.160 removes senescent cells, they are hoping you won't begin to think about what the vague definition

02:11:47.740 of what a senescent cell is. In laboratory A, these proteins are called the senescent secreted proteins.

02:11:56.140 In lab B, it's two of those plus seven more. In lab C, well, they don't see those, but they do see changes

02:12:01.880 in the nucleolus, which they view as senescent. So I believe that there certainly are cells that

02:12:08.580 accumulate in mice and in people when you get old that do stuff that's bad for you. Some of them

02:12:15.000 might make this set of cytokines. Some of them, maybe they can't divide anymore, and that's bad

02:12:20.340 for you. Some of them may even have two of these problems together. Maybe some of that has changed

02:12:26.500 because RAS has been mutated or DNA has been damaged or something. And exploring what causes

02:12:33.040 those things, how they become bad for you, whether removing that cell type is good. I think that's

02:12:39.280 wonderful. Now, if you say, these are senescent cells, I've got a drug that removes senescent cells,

02:12:45.800 you are skipping all the interesting stuff. That's my view.

02:12:49.320 Got it. It's as much a lack of nuance and a semantics issue as it is potentially, or even

02:12:56.940 more so than a biology problem. And when people talk about senolytics, you're saying, hey, we can't

02:13:02.920 really talk about senolytics without understanding which senescent cell you're targeting or which

02:13:07.700 secretory product of senescent cells you're targeting.

02:13:11.320 That's a part of it, although the problem is much deeper than that. Let me tell you a story. There's

02:13:16.800 this famous paper. Judy Campisi was the key author.

02:13:20.100 Of course.

02:13:20.860 This is the paper that introduced beta-galactosidase as the way to count senescent cells. And she looked

02:13:26.400 at the skin of a lot of people, young and old people. And she counted the number of senescent

02:13:31.260 cells and proved that it went up a lot with age. That was a very influential paper.

02:13:36.540 She's at the Buck Institute, correct, if I'm not mistaken?

02:13:39.420 Dr. Campisi, yeah, Judy is, I believe, at the Buck Institute. So in this original paper,

02:13:44.220 which was many years ago, the scores were 1 plus, 2 plus, 3 plus, and 4 plus. Not actually

02:13:49.660 a percentage of cells that were beta-gal positive in the skin sections. And only one person was

02:13:54.840 4 positive. It was a 90-some-year-old grandmother. The actual cell counting was done by a friend

02:14:02.060 of mine, Monica Peacock, who is a dermatopathologist. We were all at BU at that time. So I called Monica

02:14:07.880 and I said, okay, Monica, so 4 plus, how many cells do you have to get to be 4 plus? And she said,

02:14:13.800 yeah, that's 10 to the minus 4th. The skin section that had the highest number of beta-gal

02:14:19.580 positive cells had only one positive cell in 10,000. One in 10 to the fourth cells. Everybody else,

02:14:29.700 all the 70s, 60s, 50s, all of those people had fewer than one cell in 10,000. So the statement

02:14:39.420 is literally true. Senescent cells go up with age. These people wouldn't make up the data,

02:14:44.260 but they did emphasize the fact that even in the very oldest sections, the oldest people,

02:14:49.900 the number of actual senescent cells was really quite small.

02:14:54.100 By the way, what was the difference between the 1 plus, 2 plus, 3 plus? Clearly those numbers don't

02:14:58.540 refer to log differences. I don't know. That would be worth knowing. If there's a log full

02:15:03.520 jump between every group, you could argue maybe there's something interesting, but...

02:15:07.700 Yeah, that was the high limit. I mean, the take-home message here is that senescent cells,

02:15:12.940 at least as indicated by that one marker, beta-galactosidase, are so rare that they're

02:15:18.180 virtually absent from the skin of people of any age. That is, virtually absent, I would consider 10 to the

02:15:24.620 minus 4th and lower. But it's been a very influential, I believe, much overly emphasized,

02:15:31.880 over-influential concept. So I was hoping we would show that phycetin would remove senescent cells from

02:15:37.400 mice and they would have no lifespan effect. That was what I was betting on. But in fact,

02:15:42.680 that didn't seem to actually remove P16 or P21 positive cells from any of the tissues that we

02:15:48.020 evaluated. So we're back at square one.

02:15:49.960 What we have learned is that phycetin doesn't do jack. What we haven't learned is if removing

02:15:55.520 senescent cells is or is not beneficial.

02:15:58.780 It might work in people. Jim Kirkland believes, and he could be right. You can get it to work even

02:16:03.140 in mice if you administer it as a bolus dose, a huge dose once a day rather than gradually in the food.

02:16:10.100 It's reasonable. He might be right about that.

02:16:11.740 So there's another very popular anti-senescent drug out there. It's a chemotherapy, actually.

02:16:20.000 Yeah, there's a combination. People often give quercetin plus dosatinib.

02:16:24.000 Yeah. Has that been proposed yet as part of the ITP?

02:16:27.400 Yes. And I can't discuss it because we're not allowed to discuss anything

02:16:31.040 that comes in except the things we accept. It was not accepted.

02:16:34.320 So it sounds like we have a lot to catch up on in a few years. We're going to need to

02:16:40.060 do a rundown on the repeat studies of 17-alpha-estradiol plus rapamycin. We're going to need to

02:16:48.800 understand the tier two studies of meclizine and...

02:16:55.240 Astaxanthin?

02:16:56.360 Correct. And we're going to also have to see how they validated against the aging indicators as well,

02:17:03.520 presumably we'll also have a little bit more insight into the crossbridge or the link

02:17:10.520 between the aging indicators and plasma biomarkers that may start to bridge that gap towards

02:17:17.600 actually assessing interventions in humans. I guess there's no shortage of stuff we'll have to

02:17:22.760 catch up on.

02:17:23.940 I like that whole list and I would add one point to it that we touched upon a little earlier. I'd love

02:17:27.700 to know whether these drugs slow cognitive failure, of course.

02:17:30.020 Yeah, that's right. I think the addition of your colleague now coming on board for all tier two

02:17:35.120 studies to have a cognitive component I think is incredibly exciting. Again, I think this point

02:17:39.780 about really understanding healthspan fully can't be overstated. Some might argue even more important

02:17:45.940 than lifespan.

02:17:47.020 One of the nice things about this program is that one of the things it's designed to do is to

02:17:51.460 stimulate work in other labs. There are a lot of labs that are really good at cognition or heart

02:17:55.640 failure or bone failure. We send these people tissues all the time. Papers have begun to appear

02:18:00.620 showing that the drug does this or it fails to do this or whatever. So the hope is that for every

02:18:06.180 paper we publish with a new drug that works, this will trigger work in a couple dozen labs using that

02:18:13.780 drug with our tissues or with their own tissues and that some of those will come up with disease-specific

02:18:21.240 indications, disease-specific functional benefits. The ITP can't do everything by itself, but we're

02:18:27.900 really hoping that publicization of our results will trigger others into doing good work. One of the

02:18:36.260 reasons I was so pleased to be invited back to speak with you was that the last time I was invited to

02:18:41.100 speak with you was the most productive interaction I had had, not just in the sense of how enjoyable it

02:18:46.740 was to chat with you. But also for several weeks after that, I got a lot of people writing to me

02:18:52.300 saying, hey, here's a good idea. And often it was a good idea or let's collaborate on this or can you

02:18:57.640 send me this tissue? This particular podcast is listened to by a lot of smart people who pay attention

02:19:04.400 to what is going on. And that's a major resource.

02:19:07.500 I would echo that, Rich. We certainly don't have the largest audience on the planet, but I would argue we

02:19:11.860 have the most intelligent and the most curious and also those who participate a lot. That's the

02:19:17.860 other reason I asked the question about other mechanisms for funding. Again, I think it's a

02:19:22.540 remarkably paltry sum that is spent on this when you consider the utility that can come of this,

02:19:27.600 especially with some of the other, I don't want to use the word ancillary because that's almost

02:19:31.460 disparaging in the sense, but these other sort of accretive tools that are being bundled onto it,

02:19:37.080 such as the neurocognitive assessment and the aging biomarkers. But I really see this as a very

02:19:42.440 important program that even when you include the indirect cost is a four and a half million dollar

02:19:48.040 program per year. There are lots of philanthropists out there who would happily put their dollars to

02:19:54.240 work if they could double the throughput of these molecules and the biomarkers and the insights.

02:20:01.040 So I know that there are going to be people listening to this who are probably going to say, look,

02:20:04.440 I'm kind of interested in doubling down on some of that funding. It's a higher ROI than giving a

02:20:10.060 couple million dollars to a university to put an endowed chair in place. And so anyway, I'm hopeful

02:20:15.440 a lot of good comes in this. Good. That would be nice. Well, Rich, thank you again. Always enjoyable

02:20:21.320 to speak and enjoy your trip tomorrow. And not to mention your trip to Chile later on. I'll send you a

02:20:27.360 picture from the boat. I know you will. I can't wait. Thanks. Bye-bye. Thank you.

02:20:31.900 Thank you for listening to this week's episode of The Drive. It's extremely important to me to

02:20:37.320 provide all of this content without relying on paid ads. To do this, our work is made entirely

02:20:42.320 possible by our members. And in return, we offer exclusive member only content and benefits above

02:20:48.720 and beyond what is available for free. So if you want to take your knowledge of this space to the

02:20:53.080 next level, it's our goal to ensure members get back much more than the price of the subscription.

02:20:57.780 Premium membership includes several benefits. First, comprehensive podcast show notes that detail

02:21:04.540 every topic, paper, person, and thing that we discuss in each episode. And the word on the street

02:21:10.160 is nobody's show notes rival ours. Second, monthly ask me anything or AMA episodes. These episodes are

02:21:18.260 comprised of detailed responses to subscriber questions, typically focused on a single topic

02:21:23.120 and are designed to offer a great deal of clarity and detail on topics of special interest to our

02:21:28.480 members. You'll also get access to the show notes for these episodes, of course. Third, delivery of

02:21:34.220 our premium newsletter, which is put together by our dedicated team of research analysts. This newsletter

02:21:39.760 covers a wide range of topics related to longevity and provides much more detail than our free weekly

02:21:46.000 newsletter. Fourth, access to our private podcast feed that provides you with access to every episode,

02:21:52.900 including AMA's sans the spiel you're listening to now and in your regular podcast feed. Fifth,

02:21:59.800 the qualies, an additional member only podcast we put together that serves as a highlight reel featuring

02:22:06.000 the best excerpts from previous episodes of the drive. This is a great way to catch up on previous

02:22:11.180 episodes without having to go back and listen to each one of them. And finally, other benefits that

02:22:16.180 are added along the way. If you want to learn more and access these member only benefits, you can head

02:22:21.580 over to peteratiamd.com forward slash subscribe. You can also find me on YouTube, Instagram, and Twitter,

02:22:29.040 all with the handle peteratiamd. You can also leave us review on Apple podcasts or whatever podcast

02:22:35.620 player you use. This podcast is for general informational purposes only and does not

02:22:40.820 constitute the practice of medicine, nursing, or other professional healthcare services, including

02:22:45.200 the giving of medical advice. No doctor patient relationship is formed. The use of this information

02:22:51.320 and the materials linked to this podcast is at the user's own risk. The content on this podcast is not

02:22:57.700 intended to be a substitute for professional medical advice, diagnosis, or treatment. Users should not

02:23:03.260 disregard or delay in obtaining medical advice from any medical condition they have, and they should

02:23:08.000 seek the assistance of their healthcare professionals for any such conditions. Finally, I take all conflicts

02:23:14.340 of interest very seriously. For all of my disclosures and the companies I invest in or advise, please visit

02:23:20.680 peteratiamd.com forward slash about where I keep an up-to-date and active list of all disclosures.

02:23:33.260 Peteratiamy.com forward slash about where I keep an up-to-date and active list of all disclosures.

02:23:49.340 Please visit peteratiamd.com forward slash about where I keep an up-to-date and active list of all disclosures.

02:23:54.180 Thank you.

The Peter Attia Drive - December 04, 2023

#281 ‒ Longevity drugs, aging biomarkers, and updated findings from the Interventions Testing Program (ITP) ｜ Rich Miller, M.D., Ph.D.

Episode Stats

Length

Words per Minute

Word Count

Sentence Count

Misogynist Sentences

Hate Speech Sentences

Summary

Transcript