We interviewed Dr. Peter de Keizer, a researcher who is engaged in studying senescent cells and designing therapies to destroy them in order to delay and prevent age-related diseases. Cleara Biotech is currently using his research to develop a senolytic peptide that accomplishes this via interfering with the interactions between p53 and FOXO4.
Senescent cells and aging
As your body ages, increasing amounts of your cells enter into a state of senescence. Senescent cells do not divide or support the tissues of which they are a part; instead, they emit a range of potentially harmful chemical signals that encourage other nearby cells to also enter the same senescent state.
Their presence causes many problems: they degrade tissue function, increase levels of chronic inflammation, and can even eventually raise the risk of cancer. Today, we will talk about what senescent cells are, how they contribute to age-related diseases, and perhaps most importantly, what science is hoping to do about the problem.
Senescent cells normally destroy themselves via a programmed process called apoptosis, and they are also removed by the immune system; however, the immune system weakens with age, and increasing numbers of these senescent cells escape this process and build up.
By the time people reach old age, significant numbers of these senescent cells have accumulated in the body and caused inflammation and damage to surrounding cells and tissue. These senescent cells are one of the hallmarks of aging and are thought to be a key process in the progression of aging.
In 2017, a research team led by Dr. de Keizer published a quite significant study on senescent cells . The team discovered that the molecule FOXO4-DRI was able to cause these senescent cells to destroy themselves and was a candidate senolytic, a therapy designed to remove these problem cells.
We had the opportunity to speak with Dr. de Keizer and find out more about his research.
What impact did your publication in Cell have?
The most exciting aspect was that aging research as a whole gained a lot of attention. Even now, I notice that there is a lot of enthusiasm when discussing the various mini-revolutions in the lab. Many mainstream media outlets picked it up, and I think that this will be a boost for the field and our joint mission to raise awareness about targeting the negative aspects of aging.
It doesn’t seem like as many people in Europe talk about aging as in the U.S. Is being in Europe instead of the U.S. better or worse for your research?
As usual, the U.S. innovates, China imitates, and Europe hesitates. I returned to Europe for personal reasons, but I have been talking to American investors who want to explore Europe a bit more, and there are possibilities. People here do acknowledge aging as a problem, and the Undoing Aging in Berlin was a success. The downside with Silicon Valley is that there are big budgets and a great spirit, but we also need a style of research, which is, in every city, a little bit different.
In Europe, the focus is very much molecular. I would like to combine the great vision and budget of Silicon Valley with European quality and maybe a bit of skepticism. We never publish anything unless we are really convinced. In that sense, I like Europe because people are interested in aging here; you just have to talk to the right people, and many people are skeptical. When I talk to scientists about what we do, they also get excited.
Are the regulations regarding trials more stringent than in the U.S.?
Yes, that’s true, especially for animal work. There’s a lot of societal pressure not to do animal work. We have to deal with these hurdles, but there’s good money in science here, certainly in Western Europe, and we can make do quite well. This is generalizing, but we tend to talk less and do more.
Some researchers are worried about excessive lysis of cells, which might cause organ failure, particularly in the kidney, which contains more of these cells than other organs. How do you plan on addressing this issue?
This was found in fast-aging mice, which we used in our studies; they proved to have a lot of senescence in the kidneys. Natural aging, of course, is in virtually every organ, but the kidney is definitely susceptible.
Would you say that a potential side effect of the drug, if not used at the proper dose, could be excessive lysis?
The honest answer is “We don’t know.” I’ve seen in mice that you can go too far; if you look at the cell data, it’s tenfold more potent against senescent cells. That sounds like a lot, but if you want to treat relatively healthy people with this, if one in ten cells [that will be destroyed] is basically a healthy cell, I find it very risky. So, you need to have a perfect dose or a perfect range.
With mice, we could scale it and we could say if it’s too much or not, but for humans, it’s more difficult. What we’re doing now is trying to optimize this to make it tenfold more selective. This is version 4, and the published paper is on version 3; the first two were generated in the US in 2012, and they were not so effective. The first step was very short and had a very poor solubility, the second step lasted much longer, and the third peptide we made in D-amino acid is the one we published now.
Now, we’re making the fourth one because we know where the critical amino acids and the non-important and important ones are in the interaction domain of the two proteins. We plan on giving number 4 to a team of drug development experts to get it to a hundredfold selectivity, and then it should be much safer for use.
Is there a good biomarker right now for senescent cells?
There are some aging biomarkers, such as the epigenetic clock, but that’s not necessarily a marker for senescence. For senescence, the honest answer is that we’ve tried. We’ve looked for interleukins, and there’s very poor correlation because you don’t know which organ has been affected. We have to look for tissue-specific biomarkers to know what’s going on; but it’s very difficult to do.
Maybe we don’t entirely need to; if we can get a rough aging score from the blood based on the methylome, phenotypical markers, and a frailty index, perhaps then we can suggest a treatment. This is something we have to think about as a field. I would, of course, love to have biomarkers. This is also part of our plan for part two; the first part is optimizing the drugs, and the second part is finding markers for senescent cells so that we can actually do antigen studies.
How long would you say it’s going to be for this safe version four to be optimal?
That’s the fun part. It took me ten years to come to this third version because in academia, we always have 20 other things that are also interesting. Now, we actually teamed up with a company, Cleara, that we founded just recently. The team has 20 people, with 10 structural experts, and they’re going crazy on this. Every week, we have a meeting at which they have made some more progress, and it is super fast. We gave ourselves four months for a library screen on the first version, and then it’s another ten rounds of optimizations.
Once we have a lead candidate, we will start doing all the things that academia never wants to look at, like a liver update and all the stuff that scientists aren’t interested in but is important to have. I want to do ten rounds of that, and it’s three weeks per round, then we’ll know roughly where the weak spots are in our current version, and we can go back and add heavy metal toxicity, etc. We gave ourselves a year for optimization, but I hope sooner.
Once the peptide is optimized, that’s when the real work begins. Which biomarker would you use to assess its efficacy?
We’ve known for a while now that some therapy-resistant cancer cells, such as melanoma and glioblastoma, also upregulate FOXO-p53 complex. It also goes up in therapy-surviving cancer cells. So, our lead indication would be to go for a lethal form of cancer. The benefit is that we’d only need to do Phase 2 trials, not Phase 3, before we get approval to progress to the market.
If we go with Phase 2 for cancer, we can go for an age-related disease, or for aging as a whole, much easier. The problem is that aging is not accepted as an endpoint. However it is now with the FDA, in Europe, we’re fighting to get, first, frailty as an endpoint. There’s currently no trial possible for simply treating aging.
With cancer, it’s slightly easier; because you can just take biopsies from the tumor to actually show p21 or some interleukin marker; usually, it’s a combination of things, and those are usually high in a target tumor. Then, after a month, with whatever’s left, you can check for a reduction of those markers. You can do a blood base, or you can test the DNA. For tumors, it’s relatively straightforward; you can just take biopsies every now and then. That’s actually why we also go for cancer. If it’s proven, you know how it works in humans, then it’s much easier to go for osteoarthritis. Sarcopenia would also be something I’d be interested in seeing and then maybe aging as a whole.
How often do you think people would need senolytic treatments, will they be for older or younger people?
In mice, over a year; we did it once a month. It seemed to be enough, and I think we can actually reduce that frequency. But, I still have to do the experiment. If we do it once in a while, once every three months, once every half a year in mice, it might actually be sufficient. I don’t think they accumulate that fast. Maybe later in life, you’ll do it a bit faster. Early in life, there’s really no reason to do it so often. It’s like a car. If it’s only a couple of years old, you don’t go to the mechanic as often.
Do you expect to see the benefits of senolytics in people who are already showing signs of aging, or as more of a preventive treatment?
I think the first one. This was what we showed in the mouse study; we took mice that were already in bad shape and gave them senolytics. I think it could be like if you take a rusty car, and it’s in very bad shape, and remove the rust, you may be removing too much at once. You should do it very carefully and very gently, in gradual stages. If you have a reasonably okay car, I think it’s much safer to treat.
If you can take a human at 60 who’s starting to get senescent cells, it’ll work nicely. If you take a 90 year old who’s in very bad shape, I think you have to be very careful how much you take at once. But, the idea is, indeed, that you can do this. You might take a very old person and give him a few shots of an anti-senescence drug, and you would revitalize him.
My theory is that senescent cells, because they secrete all these inflammatory factors, they stop the neighboring cells from differentiating. You see a correlation of senescent cells and stem cells right next to them, at least in the kidney. This reprogramming is permanent, so if there’s a need for rejuvenation, it malfunctions because the stem cells are unable to differentiate. This is called senescent stem lock theory. But, if you remove the senescent cells, it means the stem cells can differentiate again, supporting tissue rejuvenation.
So, just by removing the senescent cells, you could unlock the stem cells’ potential?
Yes, that’s the idea. I think that the senescent cells are a brake on rejuvenation, so they prevent proper rejuvenation from taking place. They do two things. First of all, they are annoying for the environment, and second of all, if you have lots of them and you get injured, we know that rejuvenation doesn’t take place very well. But, if you remove senescent cells in mice, not just using my method but the genetic method used by Darren Baker, for instance, you see that the mice do better. There is a rejuvenation increase.
Arguably, you could do this with a very old person, that’s my theory. In mice, it works. If you take a mouse that’s 130 weeks old, which is really like 80, 90 years old in human years, you can restore the kidney function; that’s what we showed.
There’s a lot of heterogeneity in old mice, so it’s not easy to study a lot of things, because there is a lot of variation between individuals. This is why we use fast-aging mice, but I think it’s probably true for many organs, and that’s why if we have a drug that’s tenfold more potent, then I think we can treat for longer and higher doses.
The variation was always in the untreated group. The old mice always showed a lot of variation in aging. Some mice aged in better health and some mice were very sick. In the treated group, we could reduce the signs of aging, so this heterogeneity was not in the treatment group but in the control group.
To make a long story short, I think that if you have an old group of people and you don’t know how sick they are, you take biomarkers to know who’s in bad shape and who’s in good shape, and then you could create a personalized therapy. Some people could get a lot, and some people could get less. In theory, it’s all possible.
Can you tell me more about using FOXO4-DRI for people who have had chemotherapy?
Yes. There are many off-target effects of chemotherapy. What we have done a lot of work on, and you can look it up in publications, are chemotherapy-surviving cells. Breast, melanoma, and glioblastoma are usually the ones that we look at. They upregulate FOXO and active p53. That’s essentially what we find in senescence, but when they become senescent cells, they just keep on dividing.
The idea is that you use chemotherapy or radiotherapy and kill 90 percent of the cells, whatever is left becomes senescent-like, then you give anti-senescence treatment, such as the FOXO4 drug, in order to kill the other 10 percent, or maybe they go back to the original state, and you go back and forth. Chemotherapy, senescence therapy, repeat. That’s essentially what we are aiming for.
What would stem cells be doing in that situation? Would that be positive for cancer patients?
When you have chemotherapy-surviving cells, they become more stem-like. They have more pluripotency markers, and they become more migratory. They get epithelial mesenchymal transition. This is seen in patients, and it means that whatever you do not kill is more aggressive than the original and harder to treat. However, we now know their weak spot; we now know that they become senescent-like. This is why I am saying that if we do a combination therapy, with the chemo and anti-senescence, you may solve the problem to a large extent. This is, of course, still in the test phase. As soon as we have more information, I’ll let you know, but in the lab, it worked.
So, you’ve tested this sequence of chemo and senescence therapy?
Yes, we have done it in melanoma, for instance. However, we have only done it for one cycle, not ten. We do the chemo first, and if we have cells that are growing continuously, the very resistant survivors, we give an anti-senescence drug. The parental line is not sensitive to this drug, but the chemo survivors are. We create our own target, basically. We want to go for cancer first. If that works, we can measure the biomarkers; p21, for instance, is a good biomarker that should be up after chemotherapy and radiotherapy, and it goes down with senescence treatment. That’s the next biomarker assay we’d use, and it can be a test for age-related diseases.
If you wanted to change the way clinical trials are run, would that take years and require dealing with politicians?
The FDA is not known to be very progressive. The only thing we can argue for is that it should be easier for people to enroll. There are multiple compassionate care programs for terminal patients, but for volunteers, we don’t have anything. For aging, it should be possible for people to participate in a trial after a good amount of preclinical work. Right now, it’s not possible. I’m talking to a lot of politicians and am trying to convince them to do this, but it’s going to take a long time. It doesn’t mean we shouldn’t do it. It helps a lot that we have a lot of media attention.
Would you say that being out there, showing your face, and talking to journalists have played a large role in your work recently?
The irony is that universities are a bit reluctant to share their science. It’s a shame, because when I talk to people about aging, they’re always excited, and they always have the wrong ideas too. When we start, they all think we are working on immortality, and if the discussion progresses, it’s always about cryopreservation, crazy apps where you can exchange life hints, or unrealistic sci-fi. Then, I have to separate realistic science approaches from sci-fi.
Then, after doing that, the discussion is “Yeah, but if we are immortal, how would we deal with that? Are we psychologically not equipped for immortality?” It’s not the discussion we should have, we should instead teach the masses about what’s going on and what’s realistic: for instance, the nicotinamide work, rapamycin, the stem cell replacement work, the Yamanaka factors from the Belmonte group, and then the senolytic drugs. Those are all realistic revolutions in the lab in the last two years.
I would like to promote the message everywhere that if we do a good job now, in twenty years, we can have something that’s effective in humans. However, if we don’t do anything, we’ll only have old and frail humans. Society will be stuck with a lot of old people in bad shape, so we have to do something now.
Is anyone working on an immunotherapy approach to senolytics?
That’s what everybody’s been looking for. But, immunotherapy works because of mutation load; there’s a lot of mutation in melanoma, and it’s one of the most mutating type of cancer there is. Senescent cells are relatively genetically stable, so I think immunotherapy will be very difficult on senescent cells.
The T cells look for new antigens, and that’s why for melanoma and other cancers, it’s brilliant, and it’s why I don’t think classical immunotherapy using CAR-T will work. You would have to use natural killer cells or something like that.
What about the suggestion that the immune system is the best tool for finding senescent cells?
It’s a paradox, right? The natural killer cells in vitro can do this very well. However, moles, for example, are all senescent. We’re born with them, and, sometimes, we die with them 80 years later. We do not yet understand why.
So, it’s not clear how the immune system works with senescent cells.
I think it’s too easy to say that the natural killer cells destroy senescent cells. It’s true, but there are so many exceptions to that. I think it may be the secretion profile. As a field, we need to identify subtypes of senescent cells, because there is no such thing as just a senescent cell; there are different kinds.
Our drug, the FOXO drug, works very well against IL-6 cells, but it doesn’t work very well against MNP3-high cells. It definitely doesn’t work very well against low-inflamed senescent cells. So, we need to identify homogeneity in subtypes of senescent cells, and maybe each subtype has a different immune response. Let’s say that there are five or ten different senescent cell types.
In our lab, and you can look this up on my academic webpage, we’ve followed this goal for the last ten years and are now able to track individual cells in vivo. We track certain senescent cells and sort them, and then we see how they are different from other senescent cells at the same location in the body via single-cell sequencing. Then, you can quantify subgroups, which correspond differently to diseases. For cancer, IL-6 is a problem, but for aneurysms, MNP3 is a problem. So, different subtypes have different diseases and maybe different treatments.
Our peptide will not work against MNP3-high cells or IL-6 low cells, and it works well against IL-6 high cells. For immunotherapy, IL-6 is a decoy for the immune system. So, maybe if you have IL-6 senescent cells, they are not targeted by T cells, but the other ones are. It’s a very new concept.
Do cells become senescent differently due to their tissues of origin?
That’s probably true, because the kidney carries a lot of waste products, and there is a lot of oxidative stress in the kidney and the lung. The skin, liver, and kidney are heavily exposed to stress, although other organs are more shielded from oxidation. The skin can have damage-induced senescence, as UV rays can cause it. Let’s identify the tissue subgroups and then do tissue studies.
There is a new technique called Cytof, and it can do up to 20, and maybe even 50 in the future, stainings at once on a tissue. With Cytof, we shoot a small pixel out of the tissue, then incubate it with heavy metal-labelled antibodies. Fifty metals mean fifty antibodies. For each pixel, we get a range of values with different proteins. If we have a big tissue sample, we can get a whole grid. For the whole tissue, we can get 50 markers at once. This is what we want to do for aging, and we’re one of the first to do that.
We’re talking five or ten years from now. We’ll be able to know whether a particular tissue has 5% senescent cell subtype A and 20% senescent cell subtype B. That’s what I want to do: make an atlas of senescent subtypes.
Can the heterogeneity of cancer theory be applied to senescent cells?
Absolutely, except cancer goes faster because cancer mutates rapidly. This is not a genetic thing per se. We know that if we have a dish of senescent cells, they’re completely homogenous, so it’s a genomically stable cell line. If we synchronize the cell cycle and irradiate them so that they all get an equal amount of damage, there’s still a massive amount of heterogeneity. Only 20% of the cells are IL-6 positive, so 80% is something else.
The field still considers “senescent cells” as if they are one thing, like cancer. There is not one cancer, and there is not one senescence. This puts us on the wrong track. It’s something that I think more and more people realize, but now we actually have to identify the subgroups.
Would making senolytic drugs successful require finding the niche, focusing on the one type of senescent cells you’re going to target?
Yes. There is no magic bullet. Our drug only targets IL-6 high, but Navitoclax targets the BCL pathway, and there are quercetin and dasatinib; no one is sure if or how they really work, but maybe they target another subtype, as they seem to work on adipocytes, which we have not tested. It may be cell type-specific and also molecular background dependent.
FOXO p53 is only active in a subset of senescent cells, not in everything. So, we will end up with a bunch of senolytics. I don’t think anyone is going to say, “My senolytic is better than your senolytic.” It’ll be niche driven.
How well does this treatment compare to other treatment options, such as fasting?
With fasting, you don’t kill, you just delay the secretions from senescent cells. It’s like rapamycin and aspirin; it just blocks the secretion profile. Fasting offers a transient benefit for sure, but a week later, you eat again, and they’re just there again. It’s just making them dormant. We have not seen evidence that senescent cells are removed by fasting, in mice or in cells.
Fasting works well before exposure to damage. It increases chemosensitivity in patients if you fast the patient 24 hours in advance; it works really well because you prevent the build-up of senescence, that is my theory at least. If you do it right after, it’s completely toxic. They’ve done it in mice as well. If you fast 24 hours in advance, great, every mouse survives; if you do it 24 hours after transplantation, all the mice die. With senescence, it’s the same; we should not do it at the time of damage. There’s a tangent benefit of senescence, of wound healing. If you do it after the huge stress is gone, it’s fine.
Have you looked at other senolytics?
We tried a lot, and the BCL inhibitors look the most promising. We used navitoclax and ABT drugs 263 and 727. They identified the subgroups, and we could reproduce that in vitro, at least, not in humans. What we saw when comparing them to FOXO4 is that they are toxic at low levels and should not be given to healthy people. That’s the downside of these drugs.
However, in vitro, if you do low-level navitoclax on healthy cells, you get 10, 20 percent cell death. That’s relatively stable. That’s a decrease in viability because you’re affecting some cells that are apparently sensitive to BCL inhibition. We did not see that with FOXO4, and that’s what’s reported in our paper. You usually get a low level of toxicity with low-level navitoclax, so it’s questionable if you want to do this in a healthy 80-year-old.
As for quercetin and dasatinib, I’m absolutely not a fan of those. We’ve tried a couple of experiments; we’ve never seen a good result. Campisi’s lab has tried it extensively, and they said that quercetin actually prevents senescence. It upregulates stress responses through antioxidants and promoting DNA repair. But it’s’ still not clear, there is no known mechanism for quercetin.
Now there are Hsp90 inhibitors too, but we have not tested them.
What about gene therapies; could they be more powerful than senolytic drugs?
It depends on what gene you want to target. Maybe it relates to the Oisin work, with the p16 promoter. However, as we’ve seen with FOXO4, if you do something permanently, it’s not a good idea. These genes are there for a reason. If you always block p53, you get cancer. If you always block p16, you get cancer. If you block FOXO4, the mice are apparently normal, but if we treat the mice with chemotherapy, they all die. We don’t see that with the peptides, but we see with the p16 knockouts that we get a lot of other effects.
That’s why I like peptides so much because peptides are very specific: you only break one interaction. However, kinase inhibitors block everything, and if you do a knockout, then the whole protein is absent. I don’t think gene therapy will work for senescence.
What about a transient gene therapy?
Yes, that should probably work, such as with a temporary expression of the p16 promoter. The Oisin work seems plausible, and they’re testing it now for prostate cancer.
Is it difficult to design studies in vivo to show proof of concept?
That’s why I like fast-aging mice, where we know the driver of the aging phenotype, such as a DNA repair defect that causes senescence. We can also test late-stage cancer that upregulates the markers that we’re interested in. We’re different from all the other approaches that we focus on, and this direction is very niche-specific.
It’s not like p16 or p53; we don’t inhibit all p53, as every cell has p53. This is the only criticism I have with Oisin’s approach: their p53 construct will be activated in all stressed cells. That’s a risk. If you’re in an acute state of damage, then you don’t want to have that construct.
What stage is FOXO4-DRI currently at?
We are now trying to optimize the FOXO4-DRI peptide further so it will become truly potent, and especially safe, enough for human translation. FOXO4-DRI is the third therapeutic version in an evolutionary process of about a decade in research on FOXO4-p53. Right now, the selectivity of FOXO4-DRI for senescent cells is good enough for proof-of-concept experiments in experimental laboratory settings, but to truly allow for usage in humans, the safety profile needs to be even higher. Especially since it will take a long time and serious investments, we want to make sure translation happens with a version that is optimized to allow for this.
I recently become an assistant professor at the University Medical Center in Utrecht, the Netherlands, where there is great expertise on the molecular regulation of FOXO4 and when I started this journey so long ago, to make this a reality. Excitingly, we also teamed up with a group of drug development experts in our start-up company Cleara to really be able to make translation steps to the clinic.
What challenges are you facing in its development?
For one, the time and investment needed for a broadly applicable drug to make it to the market causes people to take matters into their own hands and resort to self experimentation. Darren Moore is an example hereof. I think Darren’s vision of targeting aging is very exciting. I still advise against such approaches though, as, of course, it is dangerous and could be harmful to the real development program if unexpected side effects occur. Gene therapy is a hallmark example hereof; this field is only slowly recovering now after some bad cases in the early days.
Another issue is that “aging” itself is not a measurable end point that the FDA or EMA considers for clinical translation. This means that we need to focus on single (age-related) diseases first, even though there may be broad applicability.
Last, the system of clinical trials dates back to the pre-internet era when people did not have easy access to experimental compounds and tended to rely more on the expert opinion of clinicians and scientists. Nowadays, compounds are easier than ever to obtain through the internet, and people have become more convinced of their own opinion. When left ignored, this is a recipe for disaster, as we will see more self-experimentation with anti-aging drugs.
To overcome this, I argue for an update in the rules and regulations for clinical trials to allow easier access for (late-stage) patients and potentially even volunteers to enrol in trials with drugs that have been effectively tested in preclinical studies.
How do you think anti-aging research could provide proof of principle in humans without having to run lifetime studies?
A valid, but tough, question. For one, it would be best if we can study the effects in patients that are middle/late aged and not just measure a delay but an actual reversal of one or more features of aging, such as tissue function (using blood/urine samples or biopsies) or methylation status. Studies on the prevention/delay of aging will be much harder. While I agree that this needs to be done properly and with scientific rigor, it should be easier for people to enrol when preclinical tests are (reproducibly) positive and the treatment shows little negative side effects.
We would like to thank Dr. de Keizer for taking the time to do this interview.
 Baar, M. P., Brandt, R. M., Putavet, D. A., Klein, J. D., Derks, K. W., Bourgeois, B. R., … & van der Pluijm, I. (2017). Targeted apoptosis of senescent cells restores tissue homeostasis in response to chemotoxicity and aging. Cell, 169(1), 132-147.