Good evening everybody. It's a real pleasure to be here at this meeting; I really thank Aubrey de Grey and Michael Greve for the invitation to speak. I couldn't really have asked for a better sequence of speakers, as the prior presenter did a fantastic job of covering why senescent cells are important, and also why they are such a pain in the ass to work with scientifically. I'm a career academic scientist working in oncology and I'm relatively new to the senescence field, although I have a lot of experience in killing cells, and that's what I'd like to tell you about today: how Oisin Biotechnologies is working to develop very selective therapies to kill senescent cells and cancer.
I don't really have to give much of a background on what a senescent cell is, or what they do. These are cells that arise through programming in the body, a reaction to outside stresses - oxidative stress, genotoxic stress, and they basically prevent us from developing cancers. It is really important to note as well that as cells become senescent in the body, in response to these stresses, they also can send out factors that can spread. I'll reiterate the fact that we don't have great markers for identifying senescent cells. There are some common features we can use, scientifically, to identify them in this case, because of an accumulation of active enzymes in lysosomes. We can stain for active β-galactosidase. But really these are very heterogeneous populations, and they arise from a variety of different pathways. I don't want to go into detail about signaling, I just want to highlight the fact that the induction of this cell cycle arrest is mediated through a number of factors: p16, p53, p21. We've thought about this, and we've asked what are the common pathways that we could potentially target in order to create a selective therapy that can start to ablate these cells. Obviously, I don't need to go into the fact that while senescent cells have been implicated in aging and age-related processes, there are also very specific diseases that are consequences of aging and other phenotypes that an anti-senescence therapy could address clinically.
It was a really salient point in the last talk that p16 cells may not be the whole story. But you have to look at the data that has been shown in the past few years - and this is what really convinced me - that, sure, all senescent cells may not express p16, but it is very clear in a mouse model that if you engineer it such that you can selectively ablate all of the p16 expressing cells, you get dramatic changes in phenotype. For me, that was very impactful. In this case, Jan van Deursen's work, if you genetically engineer mice to express a suicide gene driven by a p16 promoter, using what is called INK-ATTAC, an inducable caspase-9 system, that allows them to then give a dimerizer that activates apoptosis in cells that have activated p16, that are putatively senescent, and these mice showed very dramatic changes in their phenotype. Significant improvements in healthspan, 25% median increase in lifespan, although some heterogeneity, 50% less cancer, and also functional phenotypes as well: delayed cataract formation, decreased frailty, decreased loss of hair.
I think that as a prelude to the next talk, some data that came out recently that really solidified for me that this was something worth going after as a therapy is Peter de Keizer's work, published last year, using a p53- and FOXO4-dependent mechanism. He was able to use an accelerated aging model, mice that are losing their hair, that are becoming frail, and show that treatment with an anti-senescence approach after these phenotypes have already manifested can reverse these phenotypes. This to me really solidifed the fact that this was a worthwhile development route to take for a therapy.
So that is the basis for Oisin's technology. As a team we thought that if we're going to develop a therapy that we can use for disease, then we were also thinking about the general anti-aging community and where this might be used some time in the future after it is proven clinically. So we wanted to utilize a strategy that is similar, leveraging successful animal models developed to date. Obvious we wanted to develop something that has a low toxicity profile, something that is well tolerated, and something that can be repeat dosed again and again. Something that is non-immunogenic, ideally didn't have overly off-target effects. Obviously many senescence phenotypes are tissue-specific, and the ability to target a therapy to different tissues would be a strength.
What I'm going to tell you about today is Oisin's technology that we developed. It is called the SENSOlytic platform. It is a lipid nanoparticle (LNP) platform that contains a non-integrating DNA plasmid. It is functionalized to be activated by a chemical inducer of dimerization that induces a very rapid and irreversible apoptotic response. Probably the most important part of this is the drug delivery system - the ability to deliver plasmid DNA systemically to many different tissues without significant toxicity. Oisin has developed a platform plasmid-based technology, and contrary to RNAi or delivery of messenger RNA, plasmids can be exquisitely engineered to only be activated in situations where specific pathways are activated, such as p16, p21, or p53. But they can also be engineered with enhancers or repressors and really tuned to specific tissues and diseases. We've created a system and a library of constructs that are active in various circumstances. The two I'm going to show you data on today are a version of the p16 promoter driving a suicide gene and a version of the p53 promoter driving a suicide gene.
We built a library of plasmids that are basically a specific selective promoter tied to an iCas9 inducable suicide gene that is then induced, or dimerized, through a chemical inducer of dimerization. Many of you may have seen this before, but iCas9 is a modified caspase, so it is truncated, the recruitment domain has been chopped off and replace with an FKBP dimerization domain. These domains interact very strongly with this chemical inducer of dimerization, AP20187, or its clinical analog, AP1903, that has already been shown to be safe in phase II clinical trials. What's really nice about this system is that you transiently express using a plasmid in the target cell, it is only expressed in cells with that pathway active, so p16 or p53 in our case. Then basically nothing happens until you add the dimerizer. The small molecule dimerizer is very well tolerated, goes systemic in a matter of minutes, and induces an irreversible apoptotic response. The iCas9 will then dimerize under these conditions, self-cleave, go to the apoptosome, and carry out a very rapid cell death with two to three hours. It is very hard from cells to escape from this. They can't evolve or otherwise get away from it. It is definitely final.
Some of our in vitro proof of concept experiments utilized placental lung myofibroblast cell line IMR-90. In this case we were inducing senescence using 10 grays of radiation and transfecting cells with a p16-driven iCas9. iCas9 is a little smaller than caspase-9 and can be detected with caspase-9 antibodies. In cells that haven't been treated with radiation, we don't see any expression of the iCas9. In cases where we cells are becoming senescent, expressing p16, we see induction of iCas9, and when we add a little bit of dimerizer to these cells, it is gone. It is very rapidly clearing from these cells. Then when we look at the ability for this to actually kill these cells, in viability assays, we see that every cell successfully transfected with the plasmid dies. We've shown through a number of other experiments, I'm just showing one example here, if we do flow cytometry, to look at the pathway of death, we confirm that we are inducing apoptosis in these cells.
So we have a plasmid that is very selective for p16-expressing cells. We can kill them very rapidly upon adding of the dimerizer. The question is how do we make this into a drug that works in people. It really is the delivery mechanism that is critical to making this both effective and safe. We opted to use a lipid nanoparticle platform. Lipid nanoparticles have been used for years and I'd say that mostly there's been a lot of promise and a lot of investment and very few successes. Alnylam Pharmaceuticals in Boston has just had a recent phase III successful trial with an RNAi drug, and the issue is that lipid nanoparticles tend to accumulate in the liver preferentially, and their mechanism of delivering nucleic acids into cells is a positive charge. It is a sort of a very simple technology. They've created lipids that have a positive charge. If you use a constitutive positive charge they punch holes in membranes very easily, so they associate and punch holes, disrupt endosomes, disrupt plasma membranes, so you can get stuff into cells very effectively, except they are really toxic. So there is a very low maximum tolerated dose in humans.
In response to this, several companies have developed what is called a conditionally cationic lipid. This is a lipid that is generally neutral in the bloodstream, gets taken up into endosomes, and becomes cationic in that acidic environment. These are the subject of the current clinical programs that are making their way through clinical trials for lipid nanoparticles. They work, but they are still quite toxic. The ideal delivery system is one that can use neutral lipids that are non-toxic, but use an alternative mechanism for cellular delivery of nucleic acids. I'm going to give you a tiny bit of background as to how we got to this this point. If you have a lipid nanoparticle and it has to get inside a cell, it has to get past an intact plasma membrane with all of its defenses. Viruses have evolved over millions of years to be able to solve this problem, and have evolved a variety of fusion proteins. Unfortunately, these fusion proteins are beautiful and gigantic and elegant and the way that they bring membranes together and create pores and mix lipids is really fantastic, but to attach this to a lipid nanoparticle is insane, because they are multi-protein, multi-subunit, they have gigantic active domains that are highly immunogenic.
Fortunately, there is a Canadian researcher who has been studying all his life these fusogenic orthoreoviruses and what he discovered in this particular class of viruses was that they don't use the fusion protein to enter cells, but once they enter cells in their reptilian or bird hosts, they cause all of the cells around them to rapidly fuse together. He spent his career characterizing this class of fusion-associated transmembrane proteins that are two orders of magnitude smaller than the smallest fusion protein produced by another virus, but are sufficient to induce cell-to-cell fusion, and most importantly, lipid nanoparticle to cell fusion.
While incorporating these proteins into a neutral lipid nanoparticle platform, you will find that neutral lipids by themselves are extremely poor at delivering things. In this example we're delivering an mCherry plasmid to cancer cells, and so without the fusogenic protein there is no delivery, with it we get fantastic delivery. So it increases delivery of a neutral lipid formulation by 80-350 times, and these are well tolerated in vivo. So this is an example, we're delivering an mRNA expressing luciferase, injecting into the tail vein. We get systemic expression of luciferase throughout the body. You get some accumulation in lungs and liver, but we get broad expression in many tissues including skin and soft tissues throughout the body.
This platform is what we are using to deliver the anti-senescence payload. The platform is called Fusogenix. It uses a neutral lipid formulation that is non-toxic and well tolerated. It uses these fusogenic proteins to deliver intracellularly. I'm not going to go into all the data. It actually took us three years to create an antibody against these proteins, they are really not immunogenic whatsoever. The reason for this is because most of it is a transmembrane domain. They are lipophilic, so they pack lipids around them, and they have a low profile to the immune system. We spent a lot of time working on these, engineering these fusogenic proteins to make them better. I'm not going to get into it all, but we're at the point now where we have a manufacturing platform to create these at scale, lypholize them even, and ship them around for use.
Let me show you data now taking the p16-activated caspase-9 and putting it in vivo in mice. In this case we've done an experiment now with 16 mice, an aged mouse cohort 80 weeks old. We've divided them into three groups, we're giving them a control LNP, we actually not giving them a dimerizer, or two doses, 5 and 10 mg/kg - and 10 mg/kg, if you know this field, is quite a huge dose. We treated these animals a single time by tail vein injection. We waited 96 hours, and then we gave them a single dose of dimerizer, also intravenously. Then we waited two more days, and we collected tissues, blood, and in this case we're doing a sensitive RT-PCR and controlling it with some housekeeping genes. We get a convincing dose-dependent reduction in p16 expression in a variety of tissues.
I'm going to show a couple of images where we spend a lot of time optimizing β-gal staining in mice. These are the prettiest images we've got, but we saw in multiple tissues a dose-dependent reduction in the expression of β-galactosidase. So very, very encouraging data. Obviously, creating data in the lab is great, but if we're going to translate this into humans, there's a lot of things that must be figured out. Toxicology is extremely important. It is important for a drug that you are going to deliver more than once to make sure that you don't create any neutralizing antibodies. So we've done a ton of studies looking at repeat dosing, and we don't produce any anti-drug antibodies whatsoever, so we can give this in repeated doses over time without any reduction in efficacy. CARPA assays are up there: CARPA is something that I learned about recently, complement activation-related pseudoallergy, an immune reactivity response that many patients who receive nanoparticle therapies like doxil can have. We've run all the assays for this, and it has a lower profile than doxil, so it is very well tolerated that way.
I'm happy to say that we've done some pilot non-human primate studies, giving ten times the maximum estimated human dose, and it was extremely well tolerated. We are just going through that information. Those monkeys actually got the treatment, both the p53 and p16 alone and in combination, and the dimerizer, so we will look at that and get some rich data. We're in the process of working through that now.
I'll keep coming abck to this point: the tolerability of these formulations is really important. I show this slide because it shows all of the efforts put into clinical trials of lipid nanoparticles and why they failed. If you look at the first three programs, these were really promising ten years ago, using cationic liposomes and lipid nanoparticles. If you can see by their maximum tolerated dose, way below 1 mg/kg, and these programs all failed due to liver-related toxicity. The second generation in the middle, conditionally cationic lipids, these were tolerated to a more or less better extent, and some of those programs have been successful and will result in approved drugs, but all of the targets are liver. Because the lipid nanoparticles preferentially accumulate in the liver, you're going to see dose-limiting toxicity if you don't use a neutral lipid formulation. Then you can see work using a neutral lipid formulation similar to ours, and they were not able to find a maximum tolerated dose in the one study. Based on our non-human primate studies, we expect our formulation to be equally as well tolerated.
We're currently evaluating a variety of constructs to see which one is the best to bring into humans, and - obviously it has been talked about at this conference - the creation of biomarkers that are viable endpoints for clinical trials, and also viable in animal models to look at efficacy. We're keen to talk to anybody who has a great biomarker. We have cohorts of mice in which we are looking at the life span and health span of these mice. We are thinking about the transition to the clinical stage where we're getting GMP manufacture going and doing our GLP toxicity analysis.
So I'm going to switch to cancer for a second because this is our route to the clinic. My day job is as a prostate cancer researcher. The one thing that really intrigued me about the crossover between senescence and cancer is the activation of the p53 pathway. p53 is the most mutated gene in cancer, and there are a lot of cancers that have a high burden of p53 mutation. I put prostate up there because it is actually relatively low, an average might be 10%, the vast majority of prostate cancers are low-grade. Once you get to metastatic disease, that mutation rate is well over 50%. So this is a viable target for cancers. While the p53 protein itself hasn't been a great target for oncology therapies, I think the pathway is great. If you think about p53, you can get two kinds of mutations. With all the stress cells have through replication and mutation burden, they will activate p53, to either resolve the damage or go through apoptosis. So cells either mutate or get rid of p53 to get around this. As a result the actual activation pathways that are driving this are highly upregulated. So can we exploit this activation to kill cancer cells?
I'm skipping over all the in vitro data as I only have three slides left, and I want to show you some in vivo data, as that's really important. In this case, we are using a very similar formulation to the one I showed before. There is an engineered p53 promoter driving this iCas9 suicide gene, wrapped in a neutral lipid nanoparticle. In this case we're growing gigantic prostate cancer tumors in an immunocompromised mouse. These are NOD/SCID mice. We're growing them up to over 500 m^3, so big tumors. We're doing a single intratumoral injection of the nanoparticle, waiting three days, and then doing a systemic injection of the dimerizer. We saw most of the tumors reduced 90-95% in 48 hours - and this is not amazing for an intratumoral injection, but I was very pleased because this means we're successfully transfecting the plasmid into the majority of tumor cells, which I thought was very exciting.
The real proof is to be able to do with a systemic injection, and we've done those studies. This is just an example of four mice in that cohort. We've grown these same very large tumors, growing them to a size of 500 m^3. In this case we're giving four daily tail vein injections of the LNP and on the fifth day we're giving them a single dose of the dimerizer, systemically. Again, we saw remarkable results, between 50-98% tumor reduction in just two days. This resulted in a significant prolongation of survival with a single dose in these animals. On average, six mice per group, almost a 70% reduction in tumor volume.
One thing that is really important as well: primary tumors don't kill patients in most situations. It is the metastatic disease, so we're really interested in seeing whether we can hit metastatic cancers. Obviously this will be the population we go into for the first clinical trials as well. So we have done a number of models looking at the ability of these LNPs to control metastatic disease, in this case we've got a prostate cancer systemic metastasis model, and an immunocompetent melanoma model. In both cases with a multiple dosing regimen, we were able to control this disease effectively.
We're on the path. Obviously, the long term goal of Oisin is to develop senescence-clearing therapies for aging and age-related diseases. But I think in the short term it is really important, not only for the nanoparticle technology, but also for this platform technology, to prove it in the clinic - safety and efficacy. So we've already got a phase I/phase IIb cancer trial designed, and we're right now gearing up for GLP toxicology that will be enabling for those studies. We're hoping to dose our first person in early 2019. We're excited about accelerating the translation of this technology. One thing I'll mention as well, there are many cancers in which this can work in. In Canada, we can actually do a phase I trial with all types of cancer, basically, so colorectal, prostate, lung, etc. We'll be looking for the biggest signal and most important cancer to be able to expand that cohort and then do the phase II.