At Ending Age-Related Diseases 2021, Daniel Ives of Shift Bioscience discussed how genes themselves, rather than methylation, might provide effective clocks and therapeutic targets.
The title of my talk is Targeting Safe Cellular Rejuvenation. There’s a very powerful cellular rejuvenation paradigm that’s been touched upon earlier, but it has a neoplastic risk, or cancer risk. Increasingly, there seems to be a scope to remove this risk, perhaps partially, but perhaps entirely, which is the excitement. I’ll talk to you a little bit about how we’re going about doing that.
Just to start with this paradigm, basically, it’s called cellular reprogramming, or reprogramming with Yamanaka factors. Shinya Yamanaka found these factors with respect to pluripotency. He found that a combination of OSKM could turn any adult differentiated cell into a pluripotent stem cell. He put up this data, and that just sat there for about eight years.
Then, Tamir Chandra at University of Edinburgh came along and used a new technology called the epigenetic aging clock to show that when you express these reprogramming factors, you can rejuvenate from biological age 60, down to age zero in just 17 days. It takes 60 years to build up that aging in the system, just 17 days to jettison that aging out of the system. Up until 15 days, the cells still retain a degree of identity.
There isn’t really anything else like this, as far as rate of rejuvenation is concerned. It’s a really powerful paradigm. Later on, David Sinclair showed that these factors didn’t just affect an aging biomarker; they could do dramatic, functional things. He showed in an optic nerve crush assay that Yamanaka factors could restore remarkable ability only available to newborn mice. If you crush the optic nerve of a newborn mouse, the optic nerve can regenerate, but in an adult mouse, the regeneration no longer happens. However, if you express the Yamanaka factors in that adult mouse and crush the optic nerve, it can regenerate just like a newborn mouse.
They’re taking an adult mouse and basically restoring functions that it previously had access to. In 2016, Juan Carlos Belmonte also showed that transiently expressing these factors in progeroid mice could increase their lifespan substantially. It was a Goldilocks effect. Too little of the factors, no effect on lifespan, too much of the factors, and you’ve got these scary-looking pluripotent teratomas. These are cancers where you’ve got teeth growing out and hair and all sorts of things.
If you’ve got the amount just right, you’ve got this nice increase in lifespan. Since this study, there’s been some really interesting inroads. First of all, it’s been shown that there are alternative paradigms for rejuvenation or reprogramming paradigms. Jake Campbell from Calico earlier showed that a reduced set of factors can have the same rejuvenative effects with less effect on cell identity. He also showed the warning side of this, which is, when you express even transiently these factors, you do get pluripotent cells when you analyze at the single cell level.
There’s also some really promising results from Vadim Gladyshev’s lab, where he’s basically found a natural rejuvenation event in early mouse embryogenesis. That’s a natural rejuvenation event, we don’t know exactly the overlap with Yamanaka reprogramming, all those pathways, but the fact there’s a natural event suggests this is relatively safe. There may be reasons why early embryogenesis is more permissive or a neoplastic phenomenon, but the fact that your cells will activate this naturally is a cause for optimism.
Just to talk about what we’re doing, I wasn’t originally involved in cellular reprogramming or aging clocks. By training, I’m a mitochondrial biologist, and I got interested in mitochondria because of Aubrey. I’d read his book, he talks about mitochondrial DNA mutations. I bought into that. I was convinced that if we got rid of these mutations, that would be a fantastic therapeutic lever for slowing aging and rejuvenation.
In 2017, we set up Shift Bioscience, basically, to pursue that hypothesis. Very early on, we wanted to show that, out of all the different approaches, this was the best lever, and it was actually my investor, Jonathan Milner, that picked me and put me in touch with the epigenetic aging clock. I wasn’t aware of the clock; mitochondria was my world. I was perhaps a little bit too blinkered. The moment we discovered this clock, it presented a fantastic opportunity, which was to audit our hypothesis.
Now, we could ask the question, if we turn up mutations to the max, what does the clock do? If we affect mutations, does it affect the clock? We found the the nearest lab, and it happened to be the only lab that had a mouse epigenetic aging block. Wolf Reik at the Babraham Institute in Cambridge had defined the first multi-tissue mouse epigenetic aging clock.
We analyzed some samples from a special mouse called the mitochondrial DNA mutated mouse. It has 2500 times the level of mitochondrial DNA mutations. It gets a very comprehensive premature aging phenotype. We’ve shown that with our drugs that basically reduce the level of mutations and reduce these phenotypes, but we wanted to know what was going on with the clock. We actually got surprising results, which was that the mouse didn’t show acceleration of the clock at all, despite acceleration of the aging phenotype. There was a disconnect between the mutations and the phenotype.
The second result was that we did, in fact, slow the clock by 50% in these mice, but it was model-specific, didn’t happen in wild-type mice. There’s an interesting academic story to that, but that’s all I’ll say at this point. The result for us was the beginning of a pivot. The idea of using the clock was, let’s see if this is the best lever for therapeutic intervention in aging and rejuvenation, but the clock told us, at this time, that it’s not the best lever.
Together with Brendan Swain, we set to work thinking about, what’s the logical extrapolation of aging clocks? We’re using clocks in a piecemeal way. We’ve got this hypothesis. What does the clock say? There’s alternative hypotheses; you can see what the clock says. For us, the ultimate utility of a clock is to do a systematic screen or CRISPR screen of the genome.
Knock out each gene one by one, and see what the clock says, and that way, you can just basically scan the whole genome and let the genes do the talking, let the data do the talking. That might reinforce what we know, but it might tell us some really exciting new things. Perhaps there’s a new aging biology that outranks the current biology.
We identified a technical solution that needed to be solved, and this was a single-cell transcriptomics aging clock. The reason this was a technical test, the problem to be solved was that 10X Genomics had launched a machine called the 10X Genomics Chromium. It allowed you to do sort of relatively affordably, a single-cell CRISPR knockout and get a simultaneous transcriptome. The transcriptome’s a very rich signature that you can build clocks in.
We realized that if we could just find or create a single-cell transcriptomics aging clock, we could do a CRISPR screen on a 10X Genomics Chromium. We could basically survey all of biology and see what its link to aging was. We actually cracked this problem. We were ready to basically go ahead and do the CRISPR screen, but then we realized a few things.
The first thing was that we would only be doing single-gene knockouts, so we wouldn’t have access to combinatorial space. When you knock out a single gene, that can be compensation from different isoforms. It might not be enough to affect a pathway. Really, you want to go with lots of combinations and see what affects the biology. The question is what’s linked to aging, not just what’s a single gene that’s linked to aging.
Then, the second thing was, we looked at our clock, and it was a clock made out of genes. Methylation clocks are relatively well talked about, but this was a gene-based clock. When we looked at the genes, these genes were very interesting, because they contained or validated aging biology, things that had been linked to lifespan being linked to aging phenotypes.
We realized this might be a little bit more than just a clock: it might be a window into aging biology itself. We basically applied this clock to an aging time course, we’ve applied it to other time courses. Now we have a list of candidate genes for rejuvenation, and these aren’t very potent genes.
I’ll just unwrap that a little bit now. This is really the technical breakthrough and the amount of detail I’m allowed to describe. What we do is, we take a gene expression matrix, transcriptome data. We do something called dimensionality reduction, which is we basically look at the pathways. Instead of looking at the trees, we look at the whole forest, what’s going on. Then, what we do is we apply machine learning to the pathways, at a high level, but the type of machine learning we use allows us to basically have a connection; we’re able to basically show the priority of individual chains.
We got a ranked list of genes and their contribution to this machine learning model, which colloquially or in the academic field are called clocks, aging clocks. When we apply this to an aging time course, we basically generate a high-accuracy clock, so it has a single-cell accuracy of .94, Pearson’s correlation coefficient, and you can see we can predict the ages of individual cells. It’s a distribution of age at each time point.
This is a publicly available and single-cell time course for aging human fibroblasts. A bit polarized towards the extremes, but you can see there’s individual dots, which are the ages of individual cells. You can see that the older samples that seem to age, there are two individuals that seem to have aged at slightly different rates, all the cells seem to age, but you don’t get these two populations, non-aging and aging. It’s interesting to see, but the more interesting bit is the composition of the clock.
When we look at this clock, we see mitochondrial genes, mitochondria have got a long-standing relationship with aging, we see ribosomal genes, there’s a ribosomal DNA methylation clock. That piece of biology seems to be very linked to aging. We actually see a gene in our clock that’s sufficient to accelerate aging just on its own, just a single gene. This was really exciting because we were looking at this, and we were like, wait a minute, this might be more than just a clock, it might be driving the aging process or might be causing at least part of the aging process.
Then we realized, well, we’ve got this methodology, we’ve applied it to aging just because that’s our focus, but why not apply it to rejuvenation, which is not exactly the opposite of aging, but it’s something we’d really like to have a look at. You can take publicly available time courses, Shinya Yamanaka has posted many of these, and you can train a clock, and we call it a rejuvenation clock. We got predicted rejuvenation on one axis, and the actual rejuvenation means DNA methylation-based rejuvenation, so that’s how we qualify rejuvenation. We can train a clock with an accuracy of .95, so very high.
When we look at the genes that contribute to this clock, we’re just very excited, because some of these genes are known to aging biology; some of these genes are novel and linked to some quite left-field findings that are coming out of the literature. Most importantly, they don’t look like pluripotent genes at all. What we have now is this list of genes which have been given to us by this clock methodology. We want to throw them into a cell and see if the genes are as exciting as we think they are.
Just to hit home the advantages of using gene-based clocks as a window into aging biology. Firstly, we seem to enrich more causal biology than methylation-based clocks. That doesn’t necessarily mean we’re a better clock, it’s just we seem to have more overlap with what’s been defined historically.
The great thing about genes is that you can test causality with mature technology. You can overexpress a gene to see how involved it is, you can knock it out with CRISPR, these are mature technologies. The great thing about genes is you can drug them precisely with the mRNA therapeutic paradigm, or you can target the gene products with small molecules.
This is where we are now. The next step, we want to weed out the unsafe genes from what we think are safe genes sitting on our list. This is where we are at the moment, which is rejuvenating genes which also have pluripotency; you can play around with the timing and the composition to try and get around it. The prospects for a greater safety window, it’s really where we want to go and just at least exhaust the possibility of greater safety.
We’re quite confident there is a list or a core set of safe rejuvenation factors, but we need to go through this experimental process now. This is actually a subject of our current fundraising. This is just where we are at the moment with a big red dot. We identified targets by developing this gene clock framework, and we’ve protected the framework, which makes it easier to talk about. What we want to do now is basically collide the bioinformatics with the real world, in a dish and see if this list is fit for purpose.
We expect that to be a partial overlap at the beginning; we don’t expect the bioinformatics to perfectly represent what’s going on in nature. We will be able to get a lot of information from that first collision, and see what the overlap is. Iterate, change the biometrics so we get a greater overlap, and then just slowly reach the best set of factors.
Once we do so, that’s basically the basis for drug development, so you can start thinking about how you’re going to target these genes. That can be mRNAs or small molecules. Then you can decide based on the identity of the genes and which indication you’re going to pursue.
I’ve just created this timeline just to highlight a few things. First of all, once we’ve got our list of genes, or list of safe rejuvenation genes, of course, we want to see how we can target them with drugs, but would also be fantastic to engineer a mouse embryonic stem cell that expresses the safe rejuvenation genes, basically grow that mouse up to adulthood, and then just start expressing these genes and just see what the mouse epigenetic clocks do, or the mouse aging biomarkers. Does it go down, takes ages to get back up, go down? How long can you keep that mouse going?
Then a little bit later, just see, are we moving beyond natural lifespans with this special genetic system? I think these would be really fantastic proof-of-concept results. Then when we get to Phase 1, we actually have the opportunity to measure clocks, aging clocks, human aging clocks in healthy human individuals, and see what these therapeutics are doing. Getting to market approval’s a very long road. Getting to Phase 1 is not a short road by any means, but it’s a shorter road; it’s not so far, you could say within this decade.
Some of the characters that have been roped into this projects. I’ve got a background in mitochondrial biology, Brendan Swain, really sort of self-taught on the bioinformatics. He came up with the methodology. We’ve got all these orthogonal clocks now, so we can triangulate biology from different bioinformatic methods. He’s a huge part of this.
Romina’s a mitochondrial biologist that works with me previously, came to join us. so we could basically take clock measurements in our mice studies, and Steve Ives is my dad, but he’s a serial entrepreneur. He brought a lot of business expertise, which meant we didn’t make stupid mistakes.
We are actively recruiting right now, so we want to bring in a couple of lab scientists and basically just to take this list of genes, put it into the dish as soon as possible, basically get to work with the wet lab stuff and a data scientist, because we’ll have these two core activities: basically the lab scientists testing if it works, throwng the ball back to the data scientist, we’re going to iterate the methodology, and back into the lab. That’s going to be a big cycle for us over the next few years.
Jonathan Milner’s supported us from the very beginning. In Cambridge, UK, he’s a fantastic life science angel, and he’s behind very many interesting companies. Ken Raj has been fantastic. He’s just provided us a view of the field and the frontlines over a period of time; it just means that we’re much closer to what we want to do. He is a fantastic guy, and he’s a co-author of many of the landmark epigenetic aging clock studies.
Wolf Reik allowed us to embed Romina in his lab. He’s done some fantastic things besides the multi-tissue mouse epigenetic clock. He’s creating these multi-omic methods, which allow you to create aging clocks across molecular layers, so you could go transcriptome, epigenome, many other omics layers I think are in the works, and basically create a list of the best biology, not just the best transcripts, or the best methylation sites, the best biology. It could be a mixture of all of these different things.
David’s got big pharma experience, and then just some supporters there. I’m in the Milner therapeutics institute, Jonathan’s institute, and it’s really just a very supportive environment. You just do the value add, and they take care of everything else, like taking out the rubbish, just sorting out all of the infrastructure.
We recently were accepted onto the Foresight Institute longevity accelerator, and that’s been fantastic. It provides an open door to, as far as I’m concerned, the best in the aging ecosystem. Creative Destruction Lab is like a Y Combinator, Canadian version that’s decentralized. I’ve had some fantastic contacts come out of that. Accelerate Babraham’s a local accelerator. We’ve been on a lot of accelerators, and I really think that has helped.
By all means, reach out to me if you’re interested in the science or you’re interested in supporting us during our next stage where we want to reach these safe rejuvenation genes. We really just want to get to work on these genes. It’d be such a shame if they were just sitting there on the list waiting to be revealed. This is something I’m spending an increasing amount of my time on, but it’s absolutely necessary. Thanks, everybody, for listening.