Addressing the Delivery Challenges of Genetic Medicines

While genetic medicines promise to transform the way rare diseases are treated, one of the greatest challenges to realizing the full potential of these new therapies is the delivery of them to the cells within the body where they must go to be effective. James Dahlman, associate professor in the Department of Biomedical Engineering at Georgia Tech and Emory University, has been working to address this issue through the development of nanoparticles that could serve as vectors. We spoke to Dahlman about the delivery challenges of genetic medicines, how nanoparticles compare to viral vectors, and what it takes to develop new vectors that can deliver genetic medicines to where they need to go. This episode is part of our ongoing Platforms of Hope series that explores advances in gene therapy and gene editing.

Daniel Levine:

James. Thanks for joining us.

James Dahlman:

Thank you very much. It’s a pleasure to be here Daniel.

Daniel Levine:

We’re going to talk about the challenges of delivering genetic medicines, the case for looking beyond viral vectors, and how we’re able to engineer vectors for delivering these therapies where they’re needed to go. Let’s begin with the challenges. How much of a problem is the delivery of gene therapies to the tissue and cells that they need to go, and how much of a barrier is this to realizing the full potential for genetic medicines?

James Dahlman:

Yeah. Daniel, I think that’s a great question. You can make a compelling argument that this is the most important problem that needs to be solved. When you look at the liver, for example, we can deliver RNA, drugs and genetic medicines to the liver. As a result, you have many FDA approvals in many more clinical data sets that have been generated and look like they might lead to approvals for liver-RNA drugs. However, if you look at the lungs, the heart, the bone marrow, the kidneys, and other tissues where we know there are genetic diseases and we have the RNA drugs that can treat those genetic diseases, but we can’t deliver those drugs right now. You don’t see any FDA approvals. It really is a yes, no, type of situation. If you can deliver into the tissue, you can make a huge difference in the lives of many patients. If you can’t deliver into the tissue, you cannot meet your clinical end points and you will not get an FDA approved drug.

Daniel Levine:

I’m thinking of genetic medicines broadly, whether it’s gene replacement therapy, gene editing, or RNA therapies, do they share the same delivery challenges or are they unique to the particular modalities?

James Dahlman:

The delivery challenges have some traits that are shared across antisense oligonucleotides, siRNAs, mRNAs that encode genes for protein replacement, and mRNAs that encode nucleuses like CAS9, which I would say are the four classes of common drugs. There are others but those are four common ones. Some of these issues are shared. For example, for ASO, siRNAs, and mRNAs, it looks like we can deliver to tissues like the liver. Outside the liver it’s much harder. There are other ways that these differ. One, with siRNA and ASO’s you can actually make the drugs nucleotide by nucleotide. What I mean by that is for siRNA’s you can say, I want the fourth RNA nucleotide in my siRNA drug to be a normal RNA nucleotide, but I want the fifth one to be chemically modified in some way, and I want the sixth one to be chemically modified in a different way. So, you can actually control nucleotide by nucleotide, the types of chemical modifications, the structures, and so on. It’s an actual defined drug product. When you’re dealing with any kind of mRNA, we don’t have that nucleotide by nucleotide control yet. So, we’re making averages of products. There are some differences between the small drugs like ASO’s and siRNA’s and the larger drugs like mRNA, but people are working really hard to make the mRNA drugs better and better. Especially now that the COVID mRNA vaccines have shown such promise,

Daniel Levine:

Genetic medicines need vectors to deliver them. For the most part, therapeutic developers have focused on viral vectors like AAV and Lentiviral vectors. What are the limitations of those?

James Dahlman:

I’d start by saying there are limitations to both non-viral systems, like lipid nanoparticles or other nanoparticles, and there are different limitations to viral vectors. Oftentimes these limitations select for the disease. For example, there are diseases like hemophilia, where you need one protein produced for a very long time at pretty constant levels. That’s going to be better solved by an AAV than a lipid nanoparticle. In other cases like when you’re expressing a CAS9 nuclease, you don’t want that CAS9 nuclease around for a very long time. You want it to cut the genome and then you want it to go away. You don’t want something cutting the genome and then floating around and cutting more genome for many years. In that case, you want a non-viral system. I want to start off by saying that there are advantages to non-viral systems, and there are other advantages to viral systems. Oftentimes, those advantages mean it’s very obvious which solution you need for your particular disease. Having said that, the limitations of viral vectors are the inability to re-dose and the size of the gene that you can put in the viral system at least for AAV’s. So it’s very hard to administer a viral drug more than once. Even if you can administer the drug, it’s hard to put in large genes into AAV’s. Non-viral systems do not have these same limitations.

Daniel Levine:

You work with constructing nanoparticles as delivery mechanisms. What exactly are nanoparticles and how do you construct them?

James Dahlman:

Nanoparticles are nanoscale spheres. So, I’m about six foot two. That means I’m about 1 billion, 830 million nanometers tall to give you a scale. The nanoparticles that we’re talking about are, let’s say, 70 nanometers in diameter. What’s amazing is that we can actually design these nanoparticles to have different sizes. If we want a 30 nanometer particle versus a 130 nanometer particle, we can do that. We can design them to have certain charges. Sometimes you want them to have a positive charge, other times you want them to have no charge, and other times we want them to have a negative charge. We can do all this pretty simply by using fundamental chemistry. Nanoparticles can be composed of different ingredients. Common ingredients include peptides and proteins or separately polymers, imagine little balls of plastic, or as we’re now familiar with from the COVID vaccines, lipids. Lipid nanoparticles are LMPs are little balls of lipid. In all of these cases, peptides, nanoparticles, polymeric nanoparticles, or lipid nanoparticles, the nanoparticles are composed of components that are called amphiphiles. If you going back to some of your chemistry or biology textbooks, amphiphiles are structures that have both a hydrophilic component, something that loves water, and a hydrophobic component, something that hates water. So, that’s what these things are comprised of.

Daniel Levine:

You mentioned that there are advantages and disadvantages in general. What are the advantages and disadvantages of nanoparticles as vectors?

James Dahlman:

So, one of the advantages of nanoparticles as a delivery vector is the ability to re-dose. There’s a nanoparticle formulation, called ONPATTRO, that’s used in an FDA approved drug called patisiran. This lipid nanoparticle carries an siRNA, the siRNA turns off a disease and effectively cures this disease called TTR amyloidosis. This drug has been approved for a few years now. Patients that are treated with this drug get injected with it once every few weeks and have had that happen for several years. So, they’ve been injected and re-injected many, many times over several years. In the drug world we call this the ability to dose to effect. What we mean by that is, you can inject somebody often enough to keep the effect going and you don’t have to inject them more often or less often than that. We can dose as frequently or infrequently as we want until we achieve the effect that we want. You can do this with non-viral systems such as nanoparticles, but you cannot do this with viruses.

Daniel Levine:

I should note you’re in the department of biomedical engineering. Do you see the challenges of delivering genetic medicines is an engineering problem?

James Dahlman:

Yes. The challenge of delivering engineering medicines is definitely an engineering problem. As an example, in my laboratory, we have medicinal chemists, RNA virologists, structural RNA scientists, chemical engineers, and mechanical engineers. It’s a multidisciplinary problem, but it ends up being an engineering problem. You need great chemistry expertise, you need an understanding of the biology of delivery, you need to understand the biology of the disease that you’re trying to treat by delivering the drug, and combining all of that stuff together you need to engineer a chemical system that works within your own natural biological system. At the end of the day, it is an engineering problem because it’s a design problem. You need to understand the chemistry that goes into the design and you need to understand the biology that goes into the design, but it’s a design problem and therefore an engineering problem.

Daniel Levine:

What distinguishes one nanoparticle from another? Is it just a matter of size or do they have other properties that can be engineered to make them more targeted?

James Dahlman:

Nanoparticle design is really interesting and multifaceted. If you just look at lipid nanoparticles, so you exclude polymeric nanoparticles, peptide nanoparticles, and other types, and you just focus on LMPs. The LMPs that have been EUA approved for the vaccines and separately the LMPs that have been FDA approved for siRNA drugs. These LMP’s tend to have four components, an ionizable lipid, what’s called a PEG-lipid, cholesterol, and then you have a helper lipid, which is often a phospholipid or something that looks like a phospholipid. You have four ingredients. Each one of those ingredients can have a different structure. If you look at ionizable lipids, one of the ingredients, there are thousands and thousands of different ionizable lipids that we can design. Similarly, there are a few hundred PEG-lipids that we could design, a few hundred cholesterol variants that you could buy or design, and so on. You have four different variables, each variable has between hundreds and thousands of different structures that you could try. We know that the structure for each one of those variables matters. If you change the ionizable lipid slightly, but keep the other three variables the same, that particle will behave very differently. Similarly, if you change the PEG-lipid, but you keep everything else the same, the particle will behave differently. You have a four dimensional chemical problem where each dimension as hundreds to thousands of combinations. So, it really is a combinatorial chemical trait that gives the particle its identity and targeting. This is a little bit different than just thinking about the particle size or the particle charge. It really is the combination of all four chemical traits that give the particle its function and its biological identity.

Daniel Levine:

I think of engineers using a design, build, test, and learn approach. Is this just a matter of trial and error or are there ways to calculate the properties of a particle to get it to the tissue you want to penetrate?

James Dahlman:

Eventually we will get to a point where we can rationally design nanoparticles. However, we are not there yet and we are not anywhere close to being there. So, the biggest challenge for the field right now is developing assays that give you meaningful delivery data. Imagine for a second you’re a nanoparticle scientist and you want something to deliver to the lungs, it can be any tissue but for this example we’ll say the lungs. You can make thousands and thousands of different nanoparticles. You make a big one, a small one, a charged one, a neutral one, whatever you’d like. Let’s say you make 2000. Well, you’re not sure which one’s going to work to target the lungs. So, what are you going to do? You can either perform a 6, 8, or 10 thousand mouse experiment, which is unethical and impractical, or you can test them in a cell culture plate by taking some lung cells, putting them on a dish and then testing all 2000 in the dish. Then you can whittle that down from 2000 to three or four top nanoparticles, and then put those in the mouse. That’s what the field’s done for awhile. Make a lot of stuff, test it in cell culture, and then hope that the best particles in cell culture work in the animal. Because again, it’s unethical to do thousands and thousands of mice. The problem is that the delivery in the cell culture plate does not predict the delivery in the animal. This is a fundamental issue for the field. So, one of the things that my lab has done has developed ways to test hundreds and hundreds of particles in one animal. We do this using DNA barcoding. The barcoding essays basically allow you to put these very, very large and chemically diverse libraries into one animal instead of having to do it in thousands of animals.

Daniel Levine:

You actually spun off a company that’s since been acquired by Beam therapeutics around the barcoding technology. Can you explain what DNA barcoding isn’t how it actually works?

James Dahlman:

Yeah. The company was called Guide therapeutics. It was spun out of Georgia tech and, as you mentioned, was acquired by Beam in 2021. Guide was based on barcoding technologies that we developed in the lab. The barcoding technologies allow you to run a lot of experiments at once. Let me walk you through how this works a little bit. It’s actually quite interesting. Let’s do a thought experiment. Let’s say you wanted to test two nanoparticles and you only had one mouse in front of you. How can you run two experiments at once? Well, you might think I’ll use color. I’ll make one nanoparticle red and I’ll make another one purple. You can then inject both nanoparticles into the mouse and isolate the lungs, if you want to target the lungs, and see how much red color got there and how much purple color got there. If a lot more purple color got there you can say, the purple nanoparticles is better than the red nanoparticle. The issue with that process is that you run out of colors. If you do multiple experiments in one mouse, it tops off around five or six different colors. So, the question becomes, how do you run that same experiment but use hundreds of particles instead of five? You can’t use colors anymore. Instead of encoding the nanoparticle with the color saying, red equals nanoparticle one, purple equals in a particle two, you encode the nanoparticle with a DNA sequence, the DNA barcode. Nanoparticle one, i’m going to make that nanoparticle carry DNA sequence number one, and then nanoparticle 150, i’m going to make the nanoparticle carry DNA sequence 150. I can then inject all 150 nanoparticles into the mouse, isolate the lungs, and then use DNA sequencing, essentially the 23andme machine, to ask, how much barcode one is here, how much barcode two is here, and how much barcode 150 is here? If the sequencing results come back and they show a lot of a barcode 150. Then you can say, nanoparticle 150 is better than nanoparticles 1, 2, 3, all the way through 149. By coding nanoparticle one with barcode one and encoding nanoparticle XYZ with barcode XYZ, and then just sequencing the barcodes using the 23andMe type machines, you can access an infinite number of particles all at once. That’s how we figured out how to run hundreds of experiments in one animal.

Daniel Levine:

How quickly can particles be constructed and tested?

James Dahlman:

Pretty quickly. it takes about a day to make and characterize, let’s say, 200 particles in my lab, and then can run the assay that day and get your readouts within the week. I think it’s feasible to do 200 particles and readout how they’re being delivered into 30 different cell types out of the same animal. If you’re doing 200 particles in 30 different cell types, you’re running 6,000 experiments in one animal. You can do that within a week. Now that the barcoding is becoming more mainstream, you can do a year or two years worth of work in an afternoon. It really is an exciting time to be designing nanoparticles because now you can run all sorts of experiments that five years ago, you wouldn’t even have a prayer of completing within a few years.

Daniel Levine:

What are the other limitations of viral vectors is the size of the payloads they can carry? Is that a limitation with nanoparticles or can they be built to suit any size needed?

James Dahlman:

So, nanoparticles do not have the same size constraints as virus. As you increase the size of the payload, if you go from a small mRNA, for example, to a really big mRNA, of course the delivery becomes harder. Nanoparticles do not have any sort of hard physical limitations that I’m aware of. I find it very likely that nanoparticles will be able to deliver very large payloads, including things like base editors, which is what Beam develops or even newer constructs like prime editors, or these other CRISPR CAS constructs that are pretty large. I think nanoparticles are going to be able to carry those things.

Daniel Levine:

The earliest genetic medicines we’ve seen have targeted the liver, the eye, and blood. What tissues and cell types can we target today with nanoparticles and how rapidly do you see that expanding?

James Dahlman:

In humans, we can deliver siRNA and separately mRNA plus coding CAS9 or a variant of CAS9 plus sgRNA into hepatocytes. We can do that in humans. As we’ve seen with the vaccines, you can deliver mRNA locally through intramuscular administration, into cells that eventually get traffic to the immune system, that then confer immunity to SARS CoV-2 and its variants. Moving forward, I think the next tissues that might be targeted will be immune cells. I could envision delivery to T-cells, macrophages, B-cells, and other cells in the immune system. They’re going to be some tissues that are going to be really hard to hit. For instance, it will be very hard to administer an a nanoparticle in the blood and then have a crossover into the brain. Yet, there are plenty of good diseases you could go after within the immune system. I think you’re going to see deliveries into endothelial cells, which are the cells that line your blood vessels. You might see delivery in tissues, such as muscle. I think those are the tissues that are going to be next. How quickly that’s going to come, I don’t know. Non-liver delivery is much harder than liver delivery. We’ve now had success in the liver, but that doesn’t mean that we’re immediately going to have success in these other tissues. This is why I’m so grateful that both investors and funding agencies like NIH has started to allocate resources and support for substantial efforts to look at delivery outside the liver because without those resources it’s very unlikely to work.

Daniel Levine:

You mentioned NIH. Is there a particular role you think federal funding of research can play here that the private sector can’t do or won’t invest in?

James Dahlman:

Yes. If you do not have federal funding in drug delivery, you will not have drugs that reach the clinic in new tissues period. The private investors for the companies that are developing these drugs are not set up to perform the fundamental, really high risk. and, really early stage science that’s needed to deliver outside of the liver. As a result, that work has to be done via federal funding. If that work isn’t federally funded, it’s just too risky for private investors to take on and it just won’t happen. As stark as that sounds, I do think it’s an existential thing. If you don’t have federal funding for this, it doesn’t happen. And if it doesn’t happen, you’re going to see all these drugs that could be cures for horrible genetic diseases just not get developed. I do think it’s really, really important.

Daniel Levine:

You had mentioned you don’t foresee using lipid nanoparticles injected into the blood to deliver therapies to the brain, but we’ve seen viral vectors used through intrathecal delivery. Would that be something that would be desirable for a lipid nanoparticle?

James Dahlman:

Yes, I do think it is feasible to inject a lipid nanoparticle or other nanoparticle intrathecally and have it go to the brain. The physical barriers between where you’re injecting the particle and where that particle has to go to treat the disease, aren’t as significant if you do intrathecal delivery. By contrast, if you inject in the blood, there are a lot of barriers, including a really tough physical barrier between the blood and the brain, and it’s really hard for particles to get across that barrier. I think this is a general lesson. I tell my lab members not to believe in magic, sort of jokingly. What I mean by this is, if you have to inject an nanoparticle in location X, and that nanoparticle needs to deliver a drug to location Y, and in between X and Y there are a ton of physical barriers, that’s just going to be awfully hard and the nanoparticle is not magic. It is hard for a particle to skip through tough barriers. I think your intrathecal administration idea is on point, because it reduces the barriers between the nanoparticle administration and where the nanoparticle has to go in and work.

Daniel Levine:

Do you see the field eventually moving away from viral vectors or is there always going to be a place for that?

James Dahlman:

There will always be a place for viral vectors. The traits of a viral vector, the advantages of the viral vector, the disadvantages of the viral vector, as compared to a nanoparticle, are so different from one another. I think there’s always going to be a disease that makes sense to treat with a viral vector and makes no sense at all to treat with a nanoparticle. Just like they’re going to be other diseases where it makes no sense at all to treat with a virus, but it does make sense to treat with the nanoparticle.

Daniel Levine:

Ultimately, what do you see as the potential here to use this delivery technology, to advance the field of genetic medicines?

James Dahlman:

I look to the liver. If you look at the drug development in the liver, where we can deliver RNA drugs today, it’s amazing. Alnylam pharmaceuticals has, I think, 13 or 14 different clinical programs ongoing to treat different liver diseases. You’re talking about diseases that until these drugs came along were debilitating or, in some cases, uniformly fatal.Now you can just get a shot every once in a while and the disease is halted. If we can deliver to any other tissue, the RNA drugs themselves and the genetic targets, the diseases are there. If we can deliver to one more tissue, you could see a dozen or two dozen diseases within that tissue cured eventually. If we can deliver it to two tissues, then you’re into the dozens. The way I think about it is for every tissue we solve delivery to, you’re looking at a dozen diseases that could be cured within that tissue. Again, I think it’s borderline existential. I think it’s really important to work on non-liver delivery because the potential impact on patients is drastic, it’s night and day. I think that the potential is just so high if we could just get something outside the liver, it could really save a lot of lives.

Daniel Levine:

James Dahlman, associate professor in the department of biomedical engineering at Georgia tech and Emory University. James, thanks so much for your time today.

James Dahlman:

Hey, thanks Daniel. I appreciate it. Stay safe out there and if you’re listening, get vaccinated.

Thanks to Pfizer, Inc., Bluebird, and Novartis Gene Therapies for their support of this podcast, part of our Platforms of Hope: Advances in Gene Therapy and Gene Editing series.