Good afternoon, everyone, and thank you for joining me at the second day of the Needham Healthcare Conference. My name is Gil Blum, and I am a senior biotech analyst here at Needham & Company, and I cover the immunology and gene therapy subsectors. It is my pleasure to have with me today Dr. Geoff McDonough, the President and CEO of Generation Bio. Now, just as a reminder for viewers who are watching through our conference portal, you can submit a question at any time through the box window. So Geoff, maybe as an introduction for those who are new to this story, can you provide some background into technology and your two main platforms? What is the problem you're really trying to solve for? Thanks, Gil, and thank you for the invitation to be here. I expect this chat to be as fun as every other chat we've had in this format, so appreciate you asking me to be with you guys. Generation Bio is about building a suite of truly drug-like, redosable, in vivo genetic medicines that allow us to pursue a suite of current cell and gene therapy targets, but to do so in a much more clinically appealing and globally accessible format. To your point, we've developed two breakthrough platforms, each of them solving for a very specific unmet need in the field, and each of which leading to independent opportunity spaces, and in combination, allowing us to go after this very large goal of getting to truly drug-like, redosable, in vivo genetic medicines. Just one word on each of these platforms. The first is a delivery platform called cell-targeted LNP. As the name suggests, it is a ligand-driven, highly potent, highly selective delivery system that goes to unique receptors that are expressed on the cell target of choice. We're aiming to go after T cells and hematopoietic stem cells with that platform. What does it solve for? It solves for the reality that 99% of traditional LNPs get sunk into the liver and the spleen, with almost nothing available to be targeted to other tissues. By contrast, our ctLNP platform delivers less than 1% by mass to the liver and the spleen, leaving available a very large pool of targeted LNP to get to new cell targets that have not traditionally been reachable by LNPs. That's a unique opportunity for all cargos, for mRNA, siRNA, and for our own unique cargo, iqDNA. So that's, that's the ctLNP platform. The second platform is immune-quiet DNA. This is a novel DNA format we've invented, which entirely evades, in both mice and in non-human primates, the very intense innate immune receptors or, or pathways that recognize DNA in the cytoplasm, and this has been holding the field back from using DNA as a genetic medicine cargo. So iq DNA is all about making DNA, which is long-lasting, active in the nucleus, it can be episomal or inserted to the, to the genome, making DNA as immune-quiet as modified RNA is, and, and that's really the, the unique opportunity there. And interestingly, it makes it possible to use a wide variety of LNP delivery systems with iq DNA because it's immune-quiet. So that's why I say these platforms are independent. They solve for, for different things. They're breakthrough in different dimensions. Now, of course, what's really interesting is when they go together, and you get that opportunity to build years-long expression in the nucleus from DNA as a cargo in a wide variety of cell types. So that's really the way we think about each of those platforms independently and together here at Generation Bio. Before we dive into each individual platform, maybe just from, you know, high-level perspective, what, what are your milestones and expected accomplishments this year, in 2024? Yeah. Last year, when we announced this unique potency and selectivity of ctLNP, we announced it with a reporter gene in T cells to begin with. And so this year, we should expect to demonstrate the ability to achieve similar selectivity and potency, but when delivering a CAR, so an actual therapeutic cargo, and we intend to do that in humanized mice, which is the best place to show this platform potential. We are also right behind that, building a ctLNP system to access HSCs, because we see HSCs as not only a key cell type to address sickle cell and beta thal, but also as one that can lead to a whole variety of indications underneath that. And so we hope to show targeting proof of concept also in humanized mice for that targeting system for HSCs, with the same level of potency and selectivity that we've demonstrated for T cells. And what will follow behind that, likely at sometime next year, is the ability to show the delivery of the kind of editing efficiency required to change the course of SCD and beta thal. And then, for iq DNA, last year we announced the immune profile of iq DNA and some early factory expression in mice. We wanna show that we can amplify the potency of that expression. It's quite a new cargo, so this year it's important to show that we can elevate that level of potency in mice and ultimately later in the year, next year, to show the translation to NHP. Then lastly, we think it's gonna be an important year for partnering. We formed a very good partnership with Moderna this time last year. Feels like this year is a good year to potentially add a second partnership to the portfolio here. Excellent. I do wanna start with the ctLNP platform. There's actually a decent amount of companies, mostly private names these days, that are spending a lot of time thinking about delivery. Some of these, not to be named, have pretty impressive, you know, pre-market valuations. How are you guys different? What, what is your, like, what is your secret sauce? Yeah, I alluded to it earlier, Gil. I, I think from the perspective of creating a broad therapeutic platform, potency and specificity, in general, they always win. They're so important for creating the kind of therapeutic index that's required to get broad adoption for a, a new modality. And so thinking about ctLNP, the key to getting that large circulating pool of targeted LNPs to live in the systemic circulation is avoiding delivering any, if possible, and certainly not the majority of your drug product, to the liver and the spleen. Not only because you want to create and preserve that targeted pool for reaching other tissues, but because you don't want off-target delivery to the liver and the spleen as it relates to the activity of your cargo. So the first answer to how we're different is that we don't deliver to the liver and the spleen. And if you look at the consequences of that, that can allow you to look at the benchmark comparisons. If you don't deliver to the liver and spleen, you can deliver a very low dose, to my point about potency. We're showing very high levels of T cell uptake and transduction at doses as low as 0.005 mgs per kg. That's the consequence of not having a big sink in the liver and the spleen. The other thing we're showing is that in addition to engaging almost 100% of targeted T cells, we're delivering no ligand-driven off-target delivery to other immune cells, B cells and macrophages, both in circulation and in the spleen. So, you know, if you want to really examine how we're different, look at the input doses, look at the measured off-target delivery to liver and spleen, and then look at the measured off-target delivery to other immune cells. And that, the secret in it is both the avoidance of that sink and the specificity of the biological ligands we've developed to go after T cells, and as we've talked about in HSCs. How broad is your portfolio? I mean, we're going to spend time talking about HSCs, don't worry, and T cells, but I mean, the company has specialized in delivering to all sorts of tissues. You know, what else do you have in your, you know. Yeah, it's such a, it's such a good question, Gil, because, you know, if you just zoom out far enough, you can say, aside from the dimensionality of having ctLNPs for multiple tissues and multiple different kinds of cargos, just, just in the world of, of these tissues, T cells alone, of course, lead to all kinds of oncology and, and autoimmune indications. HSCs, likewise, beta thal, sickle cell, and a whole set of haploinsufficiency states. HSCs are also very powerful cells to employ as biofactories for lysosomal enzymes. They, they lead to a whole series of things. But, you know, just if you collapse those again and you say: What would it be like if you could get to the blood-brain barrier or to skeletal muscle or to kidney cells or to vascular endothelium? And what we're learning after having spent four years developing this ctLNP platform is that that rotation of identifying and attaching very specific ligands is really rapid. So it took about seven months to do it for T cells. We're about two months into the work for HSCs, and we're seeing that leverage. So you're asking a really good question. We don't feel limited to T cells and HSCs and hepatocytes. We see a whole range of other targets in our scope over time. So let's speak to the HSCs a little bit, and then I have a couple of things here. So initially, kind of the obvious target, you know, your sickle cell diseases, your thalassemias, maybe pyruvate kinase deficiency, right? That's kind of the bread and butter here. What would you say is the key difference from your existing genetic therapy that is approved and out there? Right. So first of all, I think what's been approved for these indications, particularly beta thal and sickle cell, the efficacy is undoubted. They really work well. The problem is they're lengthy. You have to give conditioning, which is not easy to tolerate. It's quite toxic. They are- Mortality associated with it. Right. I mean, it's just not an easy thing to do, and they're also very expensive, for good reason. They're not overpriced. They're just very expensive for the technology, because of the time and all that's required to produce these therapies. So my intuition, I hope I'm wrong, my intuition is they will benefit a very small minority of the patients who could and should benefit from a therapy like that. By contrast, rather than sort of harvesting and transducing and amplifying and reintroducing after toxic conditioning over a six to eight week period, it's not a guaranteed outcome, what we're intending to do is say, "Let's go very specifically and very potently to engage these cells in vivo where they are, the full diversity of cells in vivo where they are, and deliver the edit in vivo." Now, that sounds good when you say it fast because it, it means it's point of care. It's not specific for each patient; it's universal for every patient. It can be redosed. There's no conditioning required. Maybe you need to mobilize to, to bring the HSCs into circulation- Higher level, yeah. We'll see. But it, it's also going to be very inexpensive, compared to the kind of technology hurdle that ex vivo presents, which means it can be much more sustainably, accessed on a global basis. Now, when I say it, "It sounds good when you say it fast," it does sound good. The art, of course, is making sure that you can get the editing system to the key, HSCs at a level of potency that can deliver the edit and do that durably. So those are our hurdles and gates that we're preparing to move through over the next couple of years, and they're not trivial, but the outcome, of course, what you're aiming for, is really valuable and important clinically. So maybe as to data cadence, we spoke about this earlier, but how should we think about the data as it comes from this HSC program? Yeah, I think broadly speaking, it's targeting this year, editing next year. That's kind of how we see the build of the program, and that, of course, sets up the chance next year to be able to be clear about the timing and the milestones to the INDs or the first INDs for that technology. So maybe something that's a little bit out there. I'm sure you're familiar with lysosomal storage disorders? I've heard of them. Pompe, Pompe fabric, et cetera. A big component, big issue that a lot of ERT had, still has, is no brain penetrance. So, one nifty trick, modifying HSCs is they, they become glial cells. Have you guys, looked into this? I think it's such an interesting area, Gil. It's why I said earlier that I think HSC is, you know, the sickle cell and beta thal is just the tip of the pyramid here. I think AVROBIO demonstrated amply that if you transduce HSCs with lysosomal enzymes, they do get into the brain, and they do, of course, also correct somatic disease. Bone marrow transplantation in young patients in the MPS field has shown something quite similar. And so I think the precedent is out there, and it's one of the reasons I think HSCs are so interesting, not only to correct, you know, daughter lineage deficiencies as you would do in sickle cell or beta thal, but to use the HSCs themselves as a biofactory for elaborating therapeutic enzymes. I think LSDs, for a whole bunch of reasons, at least many of them, would be well served by being produced in HSCs. The CLNs are a good example there. Mm-hmm. Yeah. Some of the challenges of effectively using LNPs in vivo, for specifically in gene editing, became apparent again a week ago with a pretty significant setback for Verve's lead program. What do you think makes gBio's LNPs less likely to cause this kind of liver tox, a very specific liver tox that we're talking about here? Yeah, I mean, I think it's easy to lose sight of the fact that LNPs, for all their virtues, they do have some tox profile. I should say at the outset, we don't fully understand exactly what the nature of the event was that was disclosed by Verve, not because it's understood by someone and they're not saying. I think it's genuinely like they're on the front edge of learning what the specific combination of an LNP and, in that case, a base editing cargo can do for the liver. But nevertheless, I think it's generally true, and it's likely to be specifically true for LNPs, that the lower the input dose you can give, the better off you are. I think, in published data, several others who've been in clinical trials have shown this kind of ceiling of, like, 0.6 mgs per kilo, 0.7 mgs per kilo, as a place where the innate tox profile of LNPs becomes an issue. So this is where I was going back to this point, that if you can really get to, an LNP that does not drive a sink into the liver and the spleen, then you can dramatically reduce your input doses, and, and that's where I think the game will be won and lost on LNPs. It's specificity for the, the target cell on the one hand and a very low input dose on the other. I think this is—I mean, it's not new. This is true for small molecules, it's true for biologics. I think it's gonna be true for genetic medicines that employ LNPs as well. If I remember correctly, this in the first generation, first iteration of LNPs, we did see ALT elevations as well, and it took- Absolutely ... generations before. Absolutely. This is why Alnylam, I think, is such a pioneer in this space. They lived through all of the first-generation, non-biodegradable LNP tox. They worked it through, and I think subsequent companies have improved in the current second generations of LNPs because they're now biodegradable, and they can be dosed episodically. You know, I'd like to see us take that yet to a further improvement. We spoke of collaborations. I wanna spend a minute here on your Moderna collab. Can you remind us a bit about the structure? Sure. So it's, it's really two transactions, if you will, or two relationships. The first relationship is a true collaboration, where we're working together to create that specific targeting moiety that goes on our core ctLNP platform to reach T-cells. So that work is paid for by Moderna. It's based on our ctLNP background IP, and the collaboration is specifically on the targeting. And on the back of that work, we each have space to operate, to build programs, Moderna primarily with messenger RNA and us primarily with iq DNA. And where in that structure, Moderna has the opportunity to license up to three targets using iq DNA, but only three indications. So that, that's one side of the relationship. They are all costs and all development for programs that Moderna decides to take forward, they pay for, and of course, every program that we take forward ourselves with iqDNA outside of their reserved target list, we pay for, and we lead, and, you know, we primarily benefit from. The second side of the agreement is much more straightforward. It's a straightforward licensing agreement for two targets in the liver using iq DNA with our delivery system, and they have a small pool of reserved targets that they can choose from over the next couple of years. So that's the structure of the arrangement. Yeah. Speaking of potential applications and the modifying T-cells and the fact there's a collaboration with Moderna, we're talking mRNA CAR-Ts here, or is this kind of open-ended? I think it's... Well, maybe separate from Moderna, just to say broadly, this idea of moving ex vivo CAR-T to in vivo is, I think, going to be a fascinating modality evolution because ex vivo, you're mobilizing T cells, and by definition, you're mobilizing a subset of T cells. And then those T cells, when you bring them ex vivo, a subset are ultimately transduced, and a subset still are expanded and activated to be reinfused, again, with a certain number of cells at that reinfusion moment. That's fundamentally different from redosably engaging the full diversity of resident T cells and transducing them to express a CAR or, over time, other moieties. And it's going to be, you know, really revealing to figure out what that looks like when you're dosing and redosing the in vivo population of T cells. Just to put that opportunity and challenge into context, you're reinfusing an activated subset of T cells, probably quite narrow from a diversity perspective that represents about- Probably a single, single clone, pretty much, given it's a single action. Right. Right. This is often probably what we're actually dealing with in practice. That number of T cells that you're reinfusing represents roughly 5% of the total native resident T cell population. You compare that to engaging maybe up to 100% of resident T cells within a redosable in vivo approach. It just becomes very interesting to understand: is that a benefit? Is it neutral? Are there downsides to it? We'll learn, you know, across the spectrum of those things. Now, when you do that, you can well imagine that redosable short-term expression of mRNA may be ideal for certain things. Redosable long-term expression, but expression that washes out with iq DNA might be ideal for other things. And then driving insertion of large DNA-based genes or multiple genes, for long-term, non-diluting modification of that full T cell diversity would be preferred. I think there's gonna be a spectrum of cargo options based on what you're aiming to do with T cells, and above that will be a whole set of learnings about what it means to engage the full diversity of in vivo T cells. Yeah. In the oncology application, it'll be particularly of interest to see kind of the role of lymphodepletion and bridging, for that matter. I'm sure you follow the hemonc space, too, and bridging has become a pretty important part of the therapeutic landscape. Actually, it's a little hard to define what exactly the role is, but it's very clearly an important role- Mm-hmm ... in driving, you know, outcomes. And I wonder if this entire picture kind of flips on its head when it's, you know, in vivo. Yeah, I think that's right. I think that's totally right. And, you know, of course, it's a natural evolution of the field. And, you know, you might have this instinct to say, well, you're aiming for more long-term and even permanent solutions for hem-oncology and autoimmunity, and maybe shorter-term, you know, solutions for autoimmunity. Let's see. I mean, you know, this population of autoreactive clones of B cells and autoimmunity aren't all reachable in one day or two, right? They're deep in the parenchyma and some tissues. So I think we're gonna learn a lot about both of these, both of these ends of the spectrum over time. As to your collaboration with Moderna again, so do you consider this a two-way street? I mean, do you expect to gain, you know, further LNP development insights from your work with Moderna? We don't. We specifically don't. We really want to make sure that the full estate of the work that we've done to invent ctLNP stays inside Generation Bio. So this is really narrowly focused on that ligand part of the system, which is really, you know, target cell specific. So we don't. We are not collaborating in that way on the LNPs per se. Okay. Shifting to iq DNA, and this is a little bit backwards-looking, so one of the big changes you had in 2023 is transitioning from ceDNA to iqDNA. Remind our viewers, what was the change, and what, what was the early data that kind of supports that decision? Yeah. So maybe just a general, orienting point, which is that DNA of any kind, plasmid, minicircle, siRNA, all of them have a very characteristic innate immune profile, which is most pronounced in immune cells and progressively less pronounced, but not absent in somatic cells, like adipocytes, for example. And all of those DNA formats have in common a broad spectrum of cytokine elevation and activation that results from the presence of DNA in the cytoplasm. And these are pathways such as cGAS and STING, and a few others that are specifically oriented towards detecting DNA. So the result of that has been to hold back DNA as a therapeutic cargo for five decades, because while you can get some degree of activity, it's never high enough within a zone that is well-tolerated. So you can, you can push the dose to get to, to therapeutic expression, but you're also carrying that, tolerability profile with you that elevates with dose. So that's the background of DNA in general. And in the evolution of our story for closed-ended DNA, we invented a fully cell-free synthetic process for making closed-ended DNA, and that had the effect of kind of widening the aperture of the kinds of structures and modifications that we could access because we had then nucleotide-by-nucleotide control. So we started to experiment with DNA formats that have never been possible to really make before in these kinds of levels of purity. That's when we, to your point about what was the early data—we invented iq DNA, we put it into mice, and we literally wondered if we didn't dose them. Because while we were seeing expression, we weren't seeing any. Across a whole spectrum of cytokines, we didn't see any elevation of cytokines, and of course, we went on to scale that and translate it to NHP and had similar findings. And I think once you're in a world where you say, "Wow, this was the seminal evolution of the messenger RNA field," when they came to chemical modifications and purity and zeroed out the very same cytoplasmic systems, tuned for RNA, not DNA, but still, that was the seminal moment for RNA. We looked at this data for DNA and said, "Maybe this is the same seminal moment for DNA therapeutics, when you have a format that evades this innate immune stimulation. So speaking of, you know, the kind of challenges you face for translation with iqD NA from mice to non-human primates, why you know, what gives you confidence that this is an achievable target? Well, I would say there are three things that we need to know and demonstrate, maybe four things, for iqDNA to get to this point, and we've eliminated three, so we've got one to go. The first is, can we make it, and can we make it with the right level of purity and scale to satisfy our ambitions? As a company that wants to work at global scale, the answer is: yes, we can do that. The second is: is the immune profile one that transcends species and constructs? Yes, it does. We've shown it with Luciferase and now with factor eight in mice and monkeys. And the third thing is: does it express? You know, can you get expression from this? 'Cause, you know, of course, you can get rid of the immune stimulation, but if you don't also express, it's not much help. Yes, it does express. So, the last thing is: does it formulate? And we've shown that we can formulate it reliably across five different formulations in monkeys. So it's that third one, it's the expression that we really need to tune, and I think the question we had on our minds is: is it an engineer-able problem? Are there handles, are there places with this DNA format that can allow us to tune potency? I can say early in the year now, yes, it is tunable. So one of the things, as we talked about right at the beginning, Gil, when you asked the question, what to pay attention to this year, that third thing I mentioned is, we're going to show, as we make progress in potency for iqDNA, we'll show that first in mice and then ultimately in non-human primates. So that's it. That's the last piece that we're focused on now. Okay, by solving for an engineering problem. Yeah. So some have brought up a, you know, potential concern, this is pretty hypothetical, that by... you traded immune stimulation for transduction efficiency. Is there any reason to think that? Well, I think it's straightforwardly true. I mean, you know, the reality is, with a new cargo, you're still working out the kinks of substance purity, both for drug substance and drug product. You're working out new analytical methods, and you're learning how the format of the cargo correlates with drivers of potency. And we've had six years of experience to work through that journey for closed-ended DNA. We're nine months into that journey for immune-quiet DNA. And exactly to your point a moment ago, that is an engineering journey. We've lived through it with our first cargoes, closed-ended DNA, and we're living through it with our iq DNA cargoes. So, you know, if you asked me as someone working to build a new modality, if I would trade a wide therapeutic index from which I could move up on potency, I would take that 100 times out of 100 over a world where you're saying you have very, very, very limited tolerability. You've got to try to raise that tolerability to get to potency. So I think we've made a good trade here with the invention, and, you know, time will tell. Another hypothetical, is there a way that the ctLNP platform could assist in driving additional expression here? When you say ctLNP, what are you, ctLNP, or? ctLNP, sorry. Ah. My mind is elsewhere. Not ctDNA. That's a different- You know, it's a super point because, let's talk about immune cells for a moment, right? You know, in a world where you have innate immune sensing, most highly expressed in immune cells, and you have a highly innate immune-triggering cargo, as we did with closed-ended DNA, as plasmid DNA is, as minicircle DNA is, there's no hope of bringing the kind of therapeutic benefits of DNA in the nucleus to immune cells. So now, the idea that we can deliver specifically and potently to immune cells with our ctLNP platform and add iqDNA as the cargo, just opens up a huge range, as we were talking about before, for target cells like T cells, eventually B cells, NK cells, HSCs, and so on. It's just very different to ask, what can I do either to knock down a gene or express a gene for eight hours, or a protein or a gene for eight hours and protein for some longer time with mRNA, versus what could I do if I could modify those cells for months per dose, or permanently insert a large gene or a group of genes? So I think it opens up a huge opportunity to combine ctLNP with iqDNA in these cell types. Right. And maybe spending a minute here on manufacturing. So you spoke about non-cell-based extracellular manufacturing areas. What have been the biggest advantages garnered from this transition? And- Yeah ... maybe scale this for us from a COGS perspective, compared to- Yeah ... some other modalities. ... So just to put in perspective the magnitude of the change, when we were manufacturing in Sf9 cells, an example, but not the only biological system for making structured DNA, it took us six to eight weeks to make material in any given drug substance batch, and we had hundreds, if not thousands, of species of DNA in what was being produced, and that would lead to a very lengthy and difficult purification process and correspondingly relatively high costs. Nowhere near the cost of making AAV gene therapy, but still nowhere near where you'd want to be, which is at biologics levels or lower. Now, with our rapid enzymatic synthesis process, we go from six to eight weeks to four hours. We go from hundreds to thousands of DNA species per batch to one DNA species per batch, because you can only make one the way that we do the reaction. And that means you really save all the time and cost in purification because what you're getting from the reaction is largely the species- Exactly that you want. That has enormous implications also from a regulatory perspective, because you're able to show exactly what your DS is in a genetic material like this before you've formulated it. And then last but not least, and most importantly, I think, to your point in the questioning, is the COGS here can get into really, like, the low end of the biologics range per dose, when if you're dosing iq DNA, you're talking about years-long or permanent correction, and that's what we need to get in vivo redosable drug-like genetic medicines to be global. It's just not gonna work if we have to extract cells or use viruses to get a one-time shot at things, or at least it won't work broadly, I don't think. Okay. And maybe another note on partnerships. When you're considering new ones, are we talking primarily for the delivery platform, maybe for the DNA platform, either, agnostic? I would say agnostic. The main thing is we're focused on, in areas that we're interested on, in, you know, pretty narrow, single-indication-type, partnerships, and in areas that we just aren't going to get to, we still want an opportunity to participate. We don't want to just, you know, have an entire area disappear. But that's a good fit for what most people are interested in, and of course, it won't surprise you that most partners would like access to both things. They want both the specific delivery, but also the full spectrum of cargo options. Okay. And kind of a last one that I would normally ask Matthew, but he's not here today. So, I don't know, cash management, cash runway, cost, that's- So we currently are guiding into the second half of 2027, so from where we are today, the better part of 3.5 years of cash on hand, and that feels like the right place to be building a portfolio of discrete programs after the first 5.5 years of platform invention. So we are - we're transitioning not to leave platform invention behind, but we're mature enough now to be building a portfolio, and that kind of cash envelope, I think, makes sense to allow us to do that. We expect to be fully self-funded and independent with that cash guidance. In other words, there's no assumption about partnership revenue or milestone payments or equity financing. So that is really the independent financing envelope that we have to work with. This should support into post-IND, into IND? Oh, certainly to IND, and you know, it's like if we're in a position where everything works and we're driving three or four programs to IND and beyond, you know, then some combination of partnering and equity financing will allow us to pursue all of them. If we're in a world where, you know, one of those is where we focus, then we're certainly self-finance, much closer all the way to the clinic. But these are the trade-offs we'll make as things develop over the next couple of years. It's not a near-term thing. Great. I think we've covered everything that we had today, Geoff. Always a pleasure, Gil. Thank you so much for the invitation, and great conversation. Thank you for having me.