Taysha Gene Therapies, Inc. (TSHA) Earnings Call Transcript & Summary

Good morning, and welcome to day 1 of Taysha Gene Therapies First R&D Day. [Operator Instructions] Today, the company will present new data and updates for many of its pipeline candidates. Joining the call today is R.A. Session II, Taysha's President, Founder and Chief Executive Officer; Dr. Suyash Prasad, Chief Medical Officer and Head of R&D; and Dr. Steve Gray, Chief Scientific Adviser and Associate Professor in the Department of Pediatrics at UT Southwestern. Before we begin, please note that this presentation will include forward-looking statements made pursuant to the safe harbor provisions of the Private Securities Litigation Reform Act of 1995. Please see Slide 2 of the accompanying presentation and Taysha's SEC filings for important risk factors that could cause the company's actual performance and results to differ materially from those expressed or implied in these forward-looking statements. Taysha undertakes no obligation to revise or update any forward-looking statements to reflect events or circumstances after the date of this conference call except as may be required by the applicable securities laws. I will now turn the call over to Taysha's President, Founder and Chief Executive Officer, R.A. Session II. Please go ahead, sir.

Thank you, [ Sarah ]. Good morning, and welcome to day 1 of our first R&D Day. We're extremely excited to present new data and updates for many of our programs. Today, you will hear from Dr. Suyash Prasad and Dr. Steven Gray on our giant axonal neuropathy, GM2 gangliosidosis, CLN1 disease, Rett syndrome programs. Tomorrow, we will be joined by Dr. Steven Gray, Dr. Berge Minassian, Dr. Rachel Bailey and Dr. Kim Goodspeed to discuss our SURF1-associated Leigh syndrome, APBD or adult polyglucosan body disorder program, Lafora disease, SLC13A5, SLC6A1 tauopathies and Angelman syndrome programs. Next slide. So let's jump right in. At Taysha, we are focused on discovering, developing and commercializing gene therapies for the treatment of monogenic diseases of the central nervous system, or CNS, in both rare and large patient populations. We were founded in partnership with UT Southwestern Medical Center and together have created a powerful engine to develop transformative therapies with the potential to significantly improve patient lives. We currently have a broad portfolio of 26 monogenic CNS programs across 3 distinct therapeutic franchises. The first is neurodegenerative diseases, which refer to diseases that are characterized by the progressive degeneration of the structures and functions of the CNS. The second is neurodevelopmental disorders, which is a group of conditions with onset during the time when the brain is developing and reflect disabilities associated primarily with the function of the neurological system and brain. The third is genetic epilepsies, which refer to disorders with recurrent seizures associated with abnormal brain development. We are frequently asked, how can you do it all? How do you prioritize programs? Simply put, we let biology lead the way and move at the speed of science. Our partnership with UT Southwestern is designed as such that each party can leverage the strength of the other. UT Southwestern is responsible for target identification, translational research and the production of GMP material suitable for toxicology studies and early-stage clinical research. Taysha is responsible for commercial-scale GMP manufacturing, regulatory activities, clinical development, patient advocacy, and ultimately the commercialization of our product candidates. Jointly with UT Southwestern, we select the most efficient pathway to pursue IND/CTA-enabling studies, manufacture clinical material, and design and conduct natural history studies. We believe the seamless integration conferred by our partnership provides us with a unique, durable, competitive advantage. In addition, our strategy to leverage clinically and commercially validated manufacturing and delivery methods as well as the established safety and efficacy profile of the AAV9 vector should continue to provide synergies across our portfolio. Next slide. So from our robust pipeline, our therapies have the potential to treat over 0.5 million patients just in U.S. and Europe alone, which ultimately allows us to create a sustainable business model. Next slide. Our scientific approach is focused on the use of validated gene therapy technology coupled with novel targeted payload design. We utilize AAV9 or its ability to effectively transduce cell types within the CNS, along with its well-validated safety and efficacy profile as demonstrated in multiple preclinical and clinical studies. Our HEK293 triple plasmid transfection suspension process is highly scalable and offers excellent yields in comparison to other methods currently utilized for recombinant AAV manufacturing. We believe this provides the right balance between scalability, yield and risk. In addition, it has been proven safe and effective across multiple serotypes in the clinic today. Importantly, we are confident that our manufacturing process is not associated with some of the inherent toxicities that have been related to other nonmammalian manufacturing systems where there is always the risk of host cell DNA ending up in the final presentation of the drug, having an effect on decreased transduction rates and an increase in inflammation, which could ultimately affect efficacy and durability. The third part of what we consider validated technology is our use of intrathecal delivery as our chosen route of administration. Physicians have been administering intrathecal medicine for decades in an outpatient setting across multiple modalities. Intrathecal delivery allows us to target the CNS broadly. It allows us to evade neutralizing antibodies by starting on the right side of the blood-brain barrier. And of course, the proof is in the data. It's been proven safe and effective when used in combination with AAV9. Intrathecal dosing has been studied extensively across 4 clinical programs, including in SMA with Zolgensma and the STRONG trial, CLN6 Batten disease, CLN3 Batten disease and in our very own giant axonal neuropathy study with TSHA-120. So by controlling for these key aspects, we believe that we increase the probability of success and reduce overall portfolio risk. Next slide. Where we take a very targeted approach is around payload design. So what does that mean for us? First, sometimes you need to express 2 genes at a 1:1 ratio in order to address the underlying cause of the disease. That's the case in our GM2 program where we are expressing both HEXA and HEXB to fully form a functioning heterodimer, which is the HEXA enzyme. So what we have done is pioneer the first bicistronic vector in gene therapy history, which includes 2 genes in a single [ conjugate ]. In some cases, you need to cap gene expression at wild-type levels on a cell-by-cell basis to guard against overexpression-associated toxicity. This is what we have accomplished in our Rett syndrome program by pioneering our miRARE platform, which is essentially a self-regulatory feedback loop built into our transgene that capture gene expression on a cell-by-cell basis essentially to guard against overexpression-associated toxicity. That's acting as a safety mechanism. In some cases, the gene is too big to fit inside a self-complementary AAV9. Therefore, we have -- what we have done is to vectorize an RNA approach to target a silencing mechanism of a silent allele. This is what we're doing in our Angelman program, where we're using a vectorized, short-hairpin RNA approach to target the silencing mechanism for the paternal UBE3A allele to restore wild-type expression and to guard against overexpression. And in some cases, we need to knock down the production of a toxic protein. This is what we're doing in our tauopathies program, where we have some very encouraging early data that target MAPT-associated tauopathies by knocking down [ tau ] by targeting the MAPT gene. In short, we have been very thoughtful as it pertains to payload design but also leveraging all the approaches that we know work extremely well in gene therapy. Next slide. An important part of our story is the novel platform technology that powers our research engine. From our giant axonal neuropathy program, we have key data that supports the potential to facilitate redosing the vagus nerve. Redosing gene therapies is a significant hurdle for patients, and we have compelling data demonstrating improvement in the autonomic nervous system function in humans. Lastly, we are using machine learning, capsid shuffling and directed evolution to allow for the rapid identification of next-generation capsids and improve targeted delivery. Next slide. Our strategic partnership with UT Southwestern allows access to a world-class team of scientists, which is led by Dr. Steven Gray and Berge Minassian. And it also allows us to access innovative technology in order to execute on our ambitious research strategy. Next slide. Manufacturing plays a significant role in our strategy to allow for the flexibility and scalability to support our broad pipeline. With such a deep portfolio, we never want to be capacity-constrained. Our 3-pillar approach to manufacturing includes our dedicated capacity at UT Southwestern, which currently runs a HEK293 suspension platform at 500-liter scale, soon to be 700 liters by the end of the year. This capacity takes care of our early IND-enabling tox material as well as early clinical trial material particularly for investigator-initiated studies. We couple that scale or add to it through our collaboration with Paragon, a subsidiary of Catalent, who essentially pioneered gene therapy manufacturing through its early collaboration with AveXis and now at [ Edtage ]. We have significantly dedicated -- we have significant dedicated capacity at Paragon, focused on early and late clinical phase material with the ability to augment our commercial scale once we get to that stage. Catalent is the only CMO that is licensed to manufacture commercial AAV9. That takes us to our third pillar, which is our very own internal manufacturing facility, which we are building in Durham, North Carolina. We recently broke ground on this 187,000-square-foot facility, which will be 2,000 liters of scale with enough space to double that capacity to 4,000-liter scale. We are very excited that it will be a multiproduct facility, meaning multiple production suites. But it will also have full analytical capabilities, manufacturing process development as well as product characterization and potency assay capabilities as well to release product from the facility. We will be starting a mini-series of Investor Days focused on specific topics and are pleased to announce that our first will be on our manufacturing platform on July 27. We will get into greater detail on our manufacturing strategy at that time, details to follow shortly. Now that I've given you an overview of the company, I will now turn the call over to Suyash to introduce our giant axonal neuropathy program. Suyash?

Well, thank you very much, R.A. And it's a real pleasure for me to be here with you on our very first R&D Day event for Taysha. This is a remote event, obviously. So it's split over 2 days this time around. And of course, we'll hopefully all meet in person at the next one. But I'm very excited to be here, very excited to be telling you all about our programs, the first of which is giant axonal neuropathy. Next slide, please. As we've talked about before, giant axonal neuropathy, or GAN, is a rare, inherited, severe neuromuscular disease. It's due to a mutation in the protein, gigaxonin, which is involved in breaking down old, degraded, waste protein material. In the absence of gigaxonin, you get a buildup of this protein, which causes impairment of electrochemical conduction down the nerve and ultimately the clinical phenotype of GAN. There's no disease-modifying treatment available for this particular disease. And there is actually an early-onset phenotype and a more recently has been described a later-onset phenotype, which is often miscategorized our Charcot-Marie-Tooth Type 2. The prevalence of the disease is about 2,400 patients in the U.S. and the EU. Next slide, please. So in terms of clinical progression, these children look fine at birth. And if you look at this particular slide, the top half of the slide looks at the early-onset form of GAN, the bottom half looks at the later-onset form of GAN. So with the early-onset form of GAN, children will be found at birth, around the age of about -- maybe 3 years of age, 3 or 4 years of age, the parent would notice something is not quite right. What they often notice is the child is a little imbalanced, a little bit unsteady in their gait and they have frequent falls. They also have very high steps, high-stepping gait. And this is because the children are losing the ability to feel the ground beneath their feet, and so they're a little unsteady on their feet due to the sensory impairment in the soles of their feet, which is one of the initial presenting features. As time progresses, most would just develop bulbar dysfunction in the speech and language area, cerebellar symptoms such as ataxia. These children end up in a wheelchair usually by age of 10, on a ventilator due to respiratory impairment around the age of 15, and they often pass away by the age of -- towards the end of teenage years or the early 20 years, so really a relentlessly progressive, awful neuromuscular disease without any treatment. As I say, there is a slightly milder later-onset phenotype, which still results in quite significant impairments of life, perhaps quality of life issues of ambulation, cognitive dysfunction, usually not particularly life-limiting though. And many of these children grow to adults who end up in the Charcot-Marie-Tooth Type 2 clinics. Next slide, please. Here is some nice pictures of children with GAN. One of the unusual feature of this disease is very tightly curled, coarse hair. These children develop scoliosis in line with many other neuromuscular conditions. There are pathological features on the nerve biopsy. And that's previously how GAN used to be diagnosed by performing nerve biopsy, looking at these swollen axons under the microscope. And nowadays, of course, a genetic testing is the more rapid route to diagnosis. There are white matter abnormalities on the MRI and of course, spinal cord abnormalities. And I'll touch on that with a couple of pictures a bit later on this presentation. Next slide, please. In fact, here we have the MRI scans. And it's an interesting point. We often get asked the question. Clearly, this is a peripheral nervous system disease with signs in the peripheral nervous system. Is there also central nervous system component of the disease? And absolutely, the answer is yes. You're seeing -- at the age of 3, you're seeing some signal abnormalities in the area surrounding the cerebellar nuclei. And then as time progresses in the same patient -- this is the image on the bottom, you actually see a lot of hyperintense white matter abnormalities in all parts of the brain. So not only is this a peripheral nervous system disease affecting the spinal cord and the peripheral nerves, it is also a central nervous system disease, which is evidenced by the clinical features of the disease. Next slide, please. In addition to the cerebellar findings and the neuromuscular findings, there's significant respiratory impairment. So in the recently published natural history study -- you can see the reference at the bottom. Bharucha-Goebel is the lead author. Carsten Bonnemann is the senior author of this. He's our partner at the NIH. This was published in Brain about a month or 6 weeks ago, and it really gives a very nice outline of the natural history data from this particular program. And part of that natural history data was an exploration of the respiratory function. And you can see that respiratory function as measured by pulmonary function test, specifically forced vital capacity, is significantly impaired in this particular condition. But also importantly, it corresponds quite nicely with some of the other key motor function and sensory function end points in the study. So for example, the MFM32, which is the primary end point of the study; and the neuropathy impairment score; and the FARS, which [ relates to ] sensory issues; and then ambulatory status, which relates specifically to functional compromise. The respiratory muscle impairment actually corresponds very nicely with all these aspects of disease. And of course, the children have quite significant sleep apnea, which causes issues with carbon dioxide retention at night, issues with hypoxia at night and very poor sleep and ultimately significant impairment to quality of life. Next slide, please. Another important feature that's often not addressed in neuromuscular diseases as a whole but which is addressed in some detail in this natural history paper and in this natural history study that Carsten Bonnemann has run is the issue around autonomic nerve system impairments. So you have a peripheral nerve system, you have a central nervous system, but you also have an autonomic nervous system, which manages things like respiratory rhythm and cardiac rhythm and gastrointestinal motility. And it's often a forgotten area of neurological disease that is not really looked at in much detail. We, of course, have looked in detail in this natural history study using particular scale numbers of COMPASS 31, which is a separated or parental rate of scale. But also, we've looked at things like the production of sweat, which is regulated by the autonomic nervous system, and the production of tears, so lacrimal function. But what you see here is that close to 80% of individuals in the study have significant gastrointestinal dysfunction. So this is severe constipation, abdominal pain, issues with digestion. And once again, this is not specifically severe from a functional perspective, but it is very severe from a quality-of-life perspective. When we speak to patients, they found this -- often, it's the autonomic nervous system features of disease that cause them the most concern. Also importantly, and I'll get to this a bit later, the fact there's deficit here really speaks towards our efforts of developing a system for vagal nerve dosing of drug where we can ameliorate the signs and symptoms and also allows us the opportunity to redose the patient, but I'll get into that a bit later. Next slide, please. Now the other important part of the GAN story and this natural history study is the fact that a really concerted effort was made to look at neurophysiology in giant axonal neuropathy, so not just muscle function, muscle strength but also how the nerves function. And the way this is done and the way to think about it most sensibly is that there's nerves that supply the muscles and there's nerves that supply the nerve, so nerves to do with movement and strength and power and also nerves to do with feeling. And what you can see on the top half of the slide where we look at CMAP amplitudes, or compound motor action potential, is when you look at the specific muscle, look at all the action potentials being generated in a specific region of the muscle and look at the actual health of physiology of the actual -- the electrochemical signals for that particular muscle, and you plot it out and look at some parameters, amplitude, duration, latency, et cetera. And then on the sensor side of things, you stimulate a particular nerve and you look at distal sensory response, so specifically looking at the median, the ulnar and the sural nerves in this particular study. But what you find is that there is a compromise in compound motor action potential. So motor physiology is affected, and there's also significantly decreased sensory nerve action potentials. And once again, these potentials correlate quite nicely with functional and motor capability. And the intent will be in a clinical trial to see stabilization of this neurophysiological compromise and hopefully even some improvement. Next slide, please. So I'm now going to hand over to our friend and colleague, Steve Gray, who's going to talk a little bit now about the preclinical data and science underlying giant axonal neuropathy. So Steve, let me hand over to you.

Yes. Thank you very much. It's a pleasure to be here today and go -- be able to go through this with you. So I want to briefly touch on some of the molecular underpinnings and pathophysiology of the disease. And it starts off with this slide where, as Suyash mentioned, giant axonal neuropathy gets its name because if you do an electron micrograph, let's say a peripheral nerve biopsy, you see these large, distended axons that are densely filled with neurofilaments. And so if you go to the next slide, we can kind of see what this might look like in a cross-section, where you can have intermediate filament accumulations in the neuron cell body, but then it can also be these sort of "beads on a string" axonal swellings that can extend along the length of an axon. So you can imagine that with GAN, it's a situation where you first have neuronal dysfunction because these swellings cause dysfunction in the axons, problems with transporting materials and signals along the axons. So it starts off where the cells themselves are still alive but dysfunctional. And then over time, you can lose external dysfunction and eventually lead to neuronal death and degeneration. So if you go to the next slide, the function of gigaxonin itself is that it serves as an adapter protein for an E3 ubiquitin ligase. So Suyash and R.A. touched on briefly before, gigaxonin basically targets other proteins for ubiquitination and degradation. And interestingly, there's some evidence that gigaxonin can actually target itself for ubiquitination, which provides theoretically some level of self-regulation to prevent any kind of overexpression issues. All of the molecular details of this are honestly not completely worked out, but we know that if you lose gigaxonin, the downstream effect of that is that you have accumulation and disorganization of a wide variety of intermediate filament proteins. So like I said, these can include neurofilaments. It can include intermediate filaments in astrocytes, in skin cells, in muscle cells. So basically, there's many different cell types throughout the body that could have these types of accumulations, but neurons are particularly sensitive. So if we go to the next slide, we -- the gigaxonin gene, since this is an autosomal recessive disorder, we reason that you could have gigaxonin gene transfer as a therapeutic approach. We were able to package the full-length human gigaxonin transgene within a self-complementary AAV9 capsid utilizing this synthetic JeT promoter that was described about 20 years ago. This is a relatively weak but ubiquitous promoter and it expresses enough gigaxonin, and we've shown that even at these low levels. So if we go to the next slide, there was a wealth of data that we have generated since 2008, when we started working on this, numerous publications that would demonstrate basic proof of concept in vitro, in cultured human fibroblast, that we could deliver the gene and, within days, resolve these intermediate filament accumulations, also proof-of-concept studies in the GAN mouse model. And this was really the pioneer program for everything that we did where we first -- where Mylan first started investigating methods to deliver a gene-wide spread throughout the nervous system. So this is where -- this was really the motivation of why we did our initial studies in pigs and nonhuman primates with AAV9. So going into the preclinical data specifically around GAN, if we go to the next slide, we dosed GAN -- we have GAN knockout mice that we dosed intrathecally with AAV9-GAN vector. And then if you look at the sural nerve -- or, sorry, the sciatic nerves of these mice at 24 months old, you can see on the left an untreated knockout where without being a neuropathologist, I think you can appreciate a degenerative state that has substantially improved after gene transfer. And then if we look at the next slide, GAN has significant pathology in the dorsal root ganglia. These are one of the main pockets of sensory neurons. And if you can see where these bright yellow arrows are pointing, these sort of pink balls within the neuron bodies, those are the intermediate filament accumulations. And after treatment with AAV9-GAN, we see a significant reduction in those accumulations in the dorsal root ganglia. So if we go to the next slide, the -- we also developed a GAN knockout rat, which has ended up being a -- quite a nice model, a little bit better than the mouse model. And this is just where we've dosed GAN rats at 16 months old and then evaluated them on rotarod at 18 months old. And so we see, in 2 months post injection, basically a complete restoration or preservation of motor coordination in these rats. And I want to just touch on -- before I pass it back to Suyash. We had a fairly robust regulatory package that we sent to the FDA to approve this IND. It included toxicology studies, GLP toxicology studies in mice, in rats and in nonhuman primates that overall had a very favorable safety profile. So with that, I'll pass it back over to Dr. Prasad to go over the natural history data and some of the clinical trial data.

Great. Thank you, Steve. And yes, I think -- I mean it's really heartening to see that really robust preclinical package and all the scientific publications that come out from your work. They really are informative to the field. It's not just this program. It's really informing the notion of intrathecal dosing for neuromuscular disease. So congratulations on all that wonderful work. Anyway, let me move forward onto the clinical trial aspects of our program here. And it's split into 2 buckets. We have a natural history study that's been led by Carsten Bonnemann and his team at the NIH, and we have the interventional trial also led by Carsten at NIH. The natural history started in 2013, and we have about 45 patients in that study now. Patients are observed for that natural history study where the data are prospectively collected and then they roll over onto the interventional trial, which started in 2015 about 2 years later. So we do have quite extensive long-term data on the drug in interventional trial. Now the key end point in the study on natural history study and interventional trial is this MFM32. This is a well-known, well-recognized neuromuscular scale. It's 32 items. You can score 3 points for each item, scoring close to a total of 96, which is often converted to a percentage. It's well recognized by key opinion leaders and clinicians. It's well accepted by regulators. It's been used in many of the diseases. It's validated scale. It's been used in Duchenne, SMA, cerebral palsy and a number of neuromuscular diseases. And what we know for a fact is a 4-point change on this scale is a clinically meaningful change, and it's worth to remember that as we go through the natural history data and the interventional data over the next few slides. Next slide, please. And then one of the other important points to this natural history data set, once again that was recently published in Brain, is not only were many -- not only was MFM32 the key end point but Carsten and his team actually looked at the whole breadth of end points, looking at different aspects of disease, whether it's sensory function, motor function neurophysiology, imaging and even biopsies. And one of the things that Carsten did in this study was he performed that kind of evaluation to look at how well these measures correlate against each other. They are demonstrated visibly on this heat map on the right side of the slide, but you can see essentially its very, very strong correlation. What you're looking for is dark reds and dark blues in this particular map and you see lots of that. Anything that's not significantly correlated is crossed out. You can see actually more of the outcomes actually are correlated with each other than not. But the scale which has the highest correlation is the MFM32. And that's why that's been selected as the primary end point for the interventional trial. Next slide, please. Now this is a natural history data. As I say, the study was started in 2013. 45 patients of data have been shared here. There are actually a few additional patients in the study now. And what you can see in the graph on the right-hand side, there is a decline in the MFM32 score of approximately 8 points every year for each of the patients. This occurrence is quite predictable. This occurs whether a child is 5 years of age and they actually start at the higher level of MFM32 or whether they're a 10 year old and they're starting at a lower level of MFM32. They drop by about 8 points a year. And don't forget a 4-point change is deemed to be a clinically meaningful change. And it makes perfect sense if you think about the pathology of this disease, as Steve outlined. You get this buildup of abnormal, denatured proteins in the axons. They accumulate over time. So as time progresses, you get more severe disease. So one of the important things to do with this particular condition will be to try detect patients early and to treat early. Next slide, please. This is a brief table of the demography of this natural history data set. So you can see we have about 35 patients with the early-onset form of the disease and then about 10 patients with the late-onset form of the disease. And you can see there are slight differences in how -- in the ages of presentation and symptomatology. One important feature functionally that is important to note is the ambulation status, which is in the bottom half of this table. So what you can see is that over time, you move from independent walking to assisted walking to non-ambulant, meaning you need a wheelchair. And in the early-onset patients, of course, you have more patients who are non-ambulant in comparison to late-onset patients. But as time goes on, this worsens. The other important point to this -- the important part of this study is actually there's an in-depth analysis of genetic mutations of all the patients, and genotype-phenotype correlations although not completely understood are currently being defined. And you can see the pattern of pathogenic variants in the left-hand side of this particular slide. Next slide, please. Now the other important point that's made in this paper that's recently been published is the fact that the MFM32 score, as we've already mentioned, correlates very nicely with other end points. But specifically, you can see how it correlates very nicely with walking status. So the higher the score, which is the Y-axis on the graph, you see that there are more green dots higher up, meaning these are the children that are walking independently, a bit lower down. So as your MFM32 score drops, you can see in the orange you're needing assisted walking, meaning a mobility device such as a walker or walking stick. Then as the MFM32 score drops further, so when you're thinking about 50 points or even from 30 or below, you know you are going to be non-ambulant and need a wheelchair. And so it's quite a nice to be able to see very clearly how MFM32 score is very clearly functionally correlated with your ability to walk. So you can predict that as a child -- if a child starts with a score 90 and they're dropping 8 points a year, by the time they hit around about 50, they'll end up in a wheelchair. It's really quite nice because that's predictable. Next slide, please. Now that's the natural history data. I'll urge you to look at the paper that was published in Brain, the very esteemed neuroscience journal, recently. Now moving on to the interventional trial. And now don't forget the interventional trial started in 2015. So we actually have several years of data on several of these patients. There are 14 patients who we dosed in the study. There are 4 dose cohorts, which you can see on the right-hand side of the slide: a low dose of 3.53 x 10^13 total vg; a medium-low and a medium-high, 1.2 x 10^14, 1.8 x 10^14; then a high dose 3.5 x 10^14 total vg. We're presenting data on the first 3 cohorts. We have the high-dose data yet to come a bit later on this year. And don't forget the drug has been given intrathecally. R.A. mentioned earlier that our approach is AAV9, HEK293 and intrathecal dosing. And you can see that the modality of intrathecal dosage is at the bottom of this slide, where you insert a needle to the base of the spine. You drain off a few ml of CSF, attach a tubing, slowly inject using an infusion pump to inject the drug into intrathecal space. You lay the patient on their back and on their side. You can see the Trendelenburg position on the right-bottom part of this slide, which allows CSF to pull a bit closer towards the brain and really ensure a very nice transduction. And we have [indiscernible] now and intrathecally administered gene therapies, not just with GAN. But of course, GAN has the longest history. It was the first ever intrathecally dosed gene therapy. The other important point to note is that we dose intrathecal gene therapies on a total vg perspective rather than a vg-per-kilo perspective. So systemic drugs are given vg per kilo. So if a child weighs 20 kilos and is 6 years old, you multiply the number that you see on the dose by 20. But for intrathecal dosing, we give it as a total vg. So the viral load is actually reduced, i.e., it's the high dose being given, targeting the brain and spinal cord. It's relatively low dose being given in comparison to systemically administered gene therapies, which means that you'd likely get the efficacy but with less off-target, systemic side effects. And if we indeed saw side effects, the predominant side effects in gene therapy trials are immunological in nature. We're actually covering with prednisolone and sirolimus, 6 months of prednisolone or 1 year's worth of sirolimus, also known as rapamycin, as a T cell modulator. And given that immunosuppression regimen coupled with the very in-depth tox work that Steve mentioned, we actually have a very, very tolerable treatment and minimal adverse events in this clinical trial. Okay. Let's move on to the data itself. Next slide, please. Actually, the assessment -- let me touch on the assessment. We've already spent some time touching on assessments, but there are disease-specific, global assessments and looking specifically at neuromuscular function, neurophysiology. We touched on imaging already. Pathology is an important part of this, the superficial -- radial superficial nerve, what we biopsy, preimposed treatment, to look for change in the biopsies. Steve already mentioned that the diagnosis previously was always made on biopsied nerve. And we'll also be looking at certain biomarkers in that nerve tissue, DNA, RNA, protein and neurofilament, of course. And then vision is a really important functional aspect of disease, and a very in-depth look at the vision is being performed in this particular study. And I'll touch on some of these efficacy data on the vision a bit later on. Next slide, please. So what you see here is the primary efficacy end point, the MFM32. You can see that actually, the patients start dropping at least 8 points per year. Don't forget, a 4-point change is clinically meaningful. There's a clinically relevant deterioration of function on an annual basis. And I think that it's important to understand that when you see the treatment, the medium-low dose, the blue, the 1.2 e^14; and the medium-high dose, the green, the 1.8 e^14, you actually see real arrest in disease progression. So this 8-point decline changes to essentially an arrest, a stopping of ongoing deterioration. And so this translates to an 8-point improvement on an annual basis, every year. And once again, don't forget a 4-point change is deemed to be clinically meaningful. So there's a clinically meaningful benefit on an annual basis with the medium-low and medium-high dose. And don't forget, we still have the higher dose, the 3.5 x 10^14 total vg yet to go. Next slide, please. Now this is actually looking at individual patient plots for the medium-high dose, which is in the green, the 1.8 x 10^14 total vg; and the medium-low dose, which is in the blue, the 1.2 x 10^14 total vg. And what you can see is that pretreatment -- so the treatment time point is dashed line. So these patients have been observed in the natural history study for an ongoing basis, several -- up to periods of several years, and then they're rolled over to interventional trial. And you can see every patient essentially has a response to treatment, where you see this flattening of the deterioration. So if you think about it from a regulatory perspective, we're seeing really compelling data here. We have this really very nice natural history study of 45 patients with prospectively acquired data. And as you know, the FDA for rare disease is very open to using natural history studies as comparator arms. You see a nice dose response. So as you go up, you not only see a better change in the efficacy. You actually see stabilization of disease at the medium-high dose. We have yet a high dose to come. And you see long-term safety, long-term efficacy and importantly long-term durability. You can see there are patients here over -- with over 3 years' worth of data and no obvious diminution of effects as time progresses. So really, it checks all the boxes from a regulatory perspective, and we're looking forward to speaking with regulators a bit later on this year. Next slide, please. Now the other important thing I think that the NIH did very presciently really is they took the data. And in addition to applying standard frequentist methodology where you end up with a p-value, they applied what's known as Bayesian methodology, which is another way of looking at statistics that the FDA and the EMA are increasingly open to especially for these small patient disease populations are first-in-man studies. So there's 3 things to note about the Bayesian analysis. The first is, well, Bayesian methodology has been around for longer than frequentist methodology. It needs a lot of computing, horsepower. So it hasn't really been in vogue until the last 10 or 15 years when computers caught up with being able to manage the huge amounts of modeling data required. The second important thing to note is that it's an evolving analysis. And what I mean by that is that with a standard approach, you look at your prior information, set up the study, run the study and look at the results at the end. With a Bayesian approach, you look at prior information, set up your study. Then as the study is progressing, what you're learning from running the study is actually woven into the methodological analysis approach. So you end with a more -- in many respects, more accurate, more reflective answer at the end of the study using the Bayesian approach. And then the third thing to note with the Bayesian approach is that you end up with a -- you don't end up with a p-value. You end up with a probability statement, i.e., what is the probability of X or Y happening? In this particular case was the probability of a clinically meaningful effect at a particular dose. Now the first thing to note on this particular slide is actually, there are many similarities between the Bayesian analysis and frequentist analysis, i.e., the fact that you're seeing a close to 8-point change in the medium-high dose for both sets of analyses, once again an [ improving ] to this information. Next slide, please. Here are some of the complexities of the Bayesian analysis. But basically, there is a halting of treatment decline with the 1.8 e^14 dose and a clinically meaningful slowing of disease progression at the 1.2 e^14 dose. And both tests, of course, showed superior results compared to the natural decline of GAN patients. And this one -- it mirrors the frequentist analysis. Next slide, please. Now the important thing on this slide is the probability statements at the bottom of the -- the bottom right-hand part of this slide, where you ask the question -- if you look at the bottom line, what is the probability of a clinically meaningful slowing of disease at a 1.8 e^14 dose? It is 98.3%. So virtually everybody will have clinically meaningful slowing at a 1.8 e^14 dose, the medium-high dose. And even the 1.2 e^14, the low dose, there's still a very, very high chance of clinically meaningful improvement. And don't forget, we actually have the high dose yet to come. So my expectation is the high dose will be at least as good as if not a little better. Next slide, please. Now a new piece of efficacy data that I wanted to share with you today really looks at some of the -- one of the most troublesome features when you listen to patients and families, and that is the loss of vision over time. Because as children with neuromuscular diseases, all these degenerative diseases progress, it's really very sad because they lose their ability to communicate verbally with loss of bulbar function and also their ability to communicate nonverbally as the motor function dissipates but also they lose ability to see receptive communication because their visual function deteriorates over time. So it's one of the features that's hardest to treat in any of these diseases. These are mostly very upsetting for patients and families. And what you see here is the -- in 11 patients in the interventional trial, we looked at visual acuity and used the logMAR assessment tool. So this is logarithm of minimum angle of resolution. Now the standard way of assessing visual acuity, as I'm sure you all know, is you go to the optician and you sit in a chair and you read the letters of the Snellen eye chart. That's one way of doing it, but the more accurate, precise way of doing it is using this logMAR assessment. So this is what's generally used in research settings. But to give you some correlation with what you might understand is that a normal level of vision, so 20/20 vision, you score about -- you score as 0. A 0.3 on the logMAR scale will be about 20/40 vision, so mildly impaired vision or visual acuity. And then a score of 1.3 will equate to a 20/200 vision, which is essentially the registered blind. So 1.3, I'll just assume as meaning [indiscernible]. Now what you see here is that there's an ongoing progression, apart from one patient, on the 1.8 dose, where they seem to be moving opposite to others. For the most part, the logMAR scale increases over time before treatment, i.e., the visual acuity is decreasing. It doesn't actually reach the 1.3, which is where you're legally blind, but it's reaching a 0.5, 0.6, 0.7, 0.8 in certain patients. And of course, if you treat earlier the visual acuity -- if you catch early, the visual acuity has not quite deteriorated. What you do see is a stabilization of visual acuity decline after treatment, so where the dashed lines are. You see the trend is upwards and then it flattens out. And you're stabilizing this loss of vision, which is really very important from a patient and family perspective. There's more to come from the ophthalmology biomarkers. We all look at some other parameters, such as retinal nerve fiber or things -- and other things, but I look forward to sharing some more of this important information as time goes on. Next slide, please. So to summarize the clinical findings, the disease -- the treatment was very well tolerated clinically from a safety perspective, not many drug-specific adverse events, no drug-specific SAEs. There was some evidence of asymptomatic elevations of white cell count in some of the earlier dosed patients. But once this immunosuppressive regimen of rapamycin and steroid was instigated, there was none of that. And frankly, there's a little bit of white cells in the CSF, but there were no symptoms. So clinically, not sure how serious that particular adverse event was. There's no dose-limiting toxicity, no transaminitis, no issues on neuroimaging and no sudden sensory changes or issues of subacute inflammation or any type of thrombocytopenia. All of these are symptoms have seen on genotype studies but from our perspective -- and reflected by the very robust and well-completed toxicology studies, there doesn't seem to be much in the way of any major safety issues. Next slide, please. So to summarize, we've given you a nice review of giant axonal neuropathy. We've looked at the disease as a whole. Steve gave a very nice overview of the signs and the preclinical data, and I've touched on some of the clinical data, as I say, 14 patients dosed. We have more than 6 patients with more than 3 years of long-term data. Updated data on a high-dose cohort, an additional long-term data will be available later this year. We're in the process of completing our data transfer from the NIH initiating manufacturing of commercial-grade GMP material and are planning to talk to regulators specifically with the FDA, the EMA and the MHRA towards the end of the year. We will, of course, be looking to initiate new clinical sites as well, certainly one in Europe and likely one in the U.S. as well. And I look forward to updating on some of those discussions as time progresses. Once again, we've got some very nice, very strong clinical data. And importantly, I think this program really validates Steve's technology and has tremendous reach to a wider portfolio, as you'll hear over the next day, today and tomorrow. Let me stop with that, and I'll hand over to Kim for any questions.

Thanks, Suyash. Now we're going to begin our Q&A session. The first question comes from Gil Blum from Needham. Considering this a disease of degeneration, how much benefit can be gained in adult patients, i.e., the late-onset GAN patients?

Yes. Maybe -- this is Steve Gray. Maybe I can jump in here. So we know we have a lower understanding of the adult disease except that's -- it's less severe. It's much lower progressing. So -- but GAN in general is a disease of neuron dysfunction before there's degeneration. So I think it's a reasonable assumption that in these adult patients, they're going to have a wide period of neuron dysfunction and symptoms before they have frank degeneration. So I would anticipate that there's a wide window of intervention.

Thanks, Steve. Next question comes from Laura Chico of Wedbush Securities. Could you speak to the natural history trends in MFM32 decline and COMPASS 31 changes over time? Do these trends -- do these trend in a similar manner?

So the [ specific ] question around the COMPASS 31, is that correct?

Right.

Okay. So we don't have long-term data on the COMPASS 31 as yet. But in speaking with the clinical investigator, it is clear that there is an age -- there's -- as time progresses, autonomic nervous system features, in general, do deteriorate. And we're seeing the MFM32 does deteriorate over time. We haven't done the correlation between the 2 yet. But my guess is yes, they will correlate.

Great. Thank you. The next question comes from Joon Lee of Truist. What was the identity of the GAN patient organization from whom you acquired the GAN program? And did they have any agreement from FDA that the trial is of registrational quality?

Yes. I can make a comment. So the patient organization -- and I'm glad you bring them up because it's a group called the Hannah's Hope Fund, led by a lady known as Lori Sames. And she was instrumental in getting this program going and, in fact, encouraging Steve's interest in this particular project. Steve and Lori told a nice story about how they first met. And if it wasn't for Lori, I think we wouldn't be where we are today with the pushing all the way through. I think in my -- our conversations, myself and Steve and R.A., a number of conversations with Lori in recent months and -- I think they've realized that they pushed as far as they could go and really want to be an industry partner who are patient-focused and committed to getting us over the line. I think they realized that they had good data. I don't think they realized it's fully registrational quality necessarily. Of course, we have to clean up the data and sign those discussions with the FDA. But we are confident. As I've already mentioned, we've got dose response. Clearly, we've got clear stabilization of disease. We've got a long-term safety, long-term efficacy and long-term durability. And so I think that -- we're looking forward for those discussions with regulators and continuing to partner with Hannah's Hope Fund on a long-term basis. So...

Okay. Thank you. The next question comes from Laura Chico of Wedbush Securities. Based on natural history data, could you discuss how you think about the most appropriate time window for therapeutic intervention? Said differently, will there be an optimal MFM32 at which treatment must be initiated?

I can make a general comment. Steve, you're welcome to make a comment as well. But I think as with all these diseases, the earlier you treat and the earlier you diagnose, the better. You could see that there's a time progression towards a decline in the MFM32 and you can see that you're stabilizing disease. So it stands to reason that if you stabilize disease to high MFM32 score, you're going to have better functioning. We've seen clearly from the data I've shared that the MFM32 is very closely related to key functional outcomes such as your ability to walk and your ability to breathe. So if you hit an MFM32 scale before you've declined to the point where you need a ventilator or before you declined to the point where you need a wheelchair, that is important. I think to Steve's earlier point though, I think that this is a disease of neuronal dysfunction before it becomes neuronal degeneration, which is more of a later-stage aspect of this disease. So I think that there may be some possibility of reversibility if you treat with a high dose and early on. I couldn't quantify what the best number of MFM32 is to treat, but what I will say is that the higher, the better. Let me stop there. Steve, if you've got another comment, feel free to add something.

No, no. Actually, that's great. I've nothing to add.

Great. The next question comes from Joon Lee of Truist. How are you getting to 2,400 addressable patients in the U.S. and European Union? And about what percentage would qualify for TSHA-120?

So I'm happy to address that question. So the way that we get to the 2,400 patients in U.S. and Europe, it's a bit conservative to be quite frank. That's taken into account both the early-onset and the late-onset patients. There's about a, let's call it, 80%-20% or 75%-25% split between the late-onset patient and early-onset patients, being that late onset is the bigger population of the 2. And the way that we back into that calculation is that most of these patients that are late onset are miscategorized as Charcot-Marie-Tooth Type 2 patients. So -- and the way that, that epidemiology breaks down is about 1 in 2,000 live births are Charcot-Marie-Tooth patients. About 17% of that is the Type 2 population, and around 6% or so have a confirmed GAN mutation. The number that we actually use in our epidemiology calculations is around 3%. We wanted to be a little bit more conservative. I think if we use the full 6% number, that number would be close to about 3,000 or north of 3,000 just in the U.S. and Europe. But that's essentially the way that we backed into it. What's nice is this Charcot-Marie-Tooth population are actually patients that are identified. Charcot-Marie-Tooth Type 2 particularly is a clinical diagnosis and not necessarily a genetic diagnosis. So we know what the patients are. The Charcot-Marie-Tooth Foundation has a very extensive network of clinics both here in the U.S. and in Europe, where most of these patients are actually clinically diagnosed with Charcot-Marie-Tooth. So the goal is to go into these clinics and fully genotype all the Type 2 patients in order to diagnose the giant axonal neuropathy patients within that population.

Thank you. And our time is up for the GAN section. Thank you very much. I'll turn the call now over to Dr. Suyash Prasad to discuss TSHA-101 for GM2 gangliosidosis.

Great. Thank you, Kim. Let's move on to GM2 gangliosidosis, which, as R.A. mentioned earlier, we have an open-IND-equivalent CTA up in Canada for and looking forward to showing biomarker from that initial study later this year. Let's go to the next slide. So GM2 gangliosidosis is a severe lysosomal storage disorder. And what happens in this particular disease is that there's a deficiency in the enzyme, beta-hexosaminidase A, or HEXA as it's shortened to. The HEXA is a lysosomal enzyme. So in the absence of HEXA, you get a buildup of GM2 ganglioside, which is the fatty substrate that accumulates and which is broken down usually by HEXA. You get an accumulation of this ganglioside in the lysosomes. The lysosomes swell. That causes lysosomal dysfunction. The lysosome then rupture. They leak out their acidic, enzymatic content and causes neuronal damage, neuronal deterioration, neuronal dysfunction and as time goes on, relatively rapidly neuronal death. So this is a true neurodegenerative disease where it's an ongoing progressive loss of function and loss of neurons. Now the HEXA enzyme is comprised of 2 subunits, an alpha subunit and a beta subunit. It's important to remember this, the mutation in the alpha-subunit gives you the disease known as Tay-Sachs disease. A very well-known disease. It's been on about for many, many years. And the disease -- mutation the beta subunit gives you Sandhoff disease. Essentially, they are the same disease because they both result in this lack of functioning beta-hexosaminidase A, or HEXA, which then goes on and causes the clinical phenotype of GM2 ganglioside. In terms of patient numbers, it's a relatively rare disease. We feel the prevalence in the U.S. and the EU is about 500 patients. Next slide, please. Now with regard to the clinical progression and the clinical phenotype, you can see they are summarized on this slide on the right-hand side, there's an infantile onset form of the disease, which comprises the vast majority, 80% to 90% of patients. Maybe 10% have a juvenile form, maybe 5% to 10% have the adulthood onset form. Although we actually feel there's a lot of adulthood onset that's undiagnosed currently. So with the infantile form, there is a very early on, around the age of 6 weeks of age, there's a clear hyperacusis, so a real sensitivity to noise. These children have a really exaggerated startup to severe noise. Also as noted, the hypertonia, so weakness or floppiness, the inability to head control or reach out for objects. The patient ends up getting to the doctor. As part of the examination, an eye examination takes place. And there is, often, this cherry-red macula seen on the back of the retina. Cherry-red macula is a -- pattern them on it, of this particular disease. As time progresses, the child will have seizures, they could carefully they'll lose their vision and usually die between about the age of 3 and 5. The juvenile form is slightly less rapidly progressive, still awful, visual loss, seizures, definitely the mid-teens, and the adulthood form, there are a number of cognitive features and some weakness and some psychosis. Often, these patients are misdiagnosed with ALS or sometimes in schizophrenics clinic. But it tends not to be life-limiting in the same way as the infantile or juvenile onset forms. Now one important thing to notice is on the X-axis of this graph you see the residual HEXA activity. And the enzyme levels are very highly correlated to phenotypes. So the infants have less than 0.1%; juveniles, 0.5%; adults, 2% to 4%. And the other important thing you'll notice is that you only need a little bit of enzyme to make a big difference in terms of lifespan. So as with all lysosomal storage disorders, a little bit of enzyme goes a long way. Don't forget these are secreted enzymes, meaning the HEXA can break down substrates, GM2 ganglioside in one cell, can exit the cell, enter another cell through the Mannose phosphate receptor, breakdown substrate there and then do the same thing again and again. So when an enzyme is secreted, you only need a little bit. You only need a small number of the cells to be transduced to have a really dramatic clinical response. Next slide, please. Now another question we often get asked when we're sharing data from GM2 is, how well is the disease known? What is known about the natural history? And the disease has been known about for a long, long time. So the disease was first described in the late 1800s by a British ophthalmologist know as William Tay. That's why it's called Tay-Sachs disease. Here wrote the first descriptions of the disease. And then the actual enzyme, the HEXA enzyme, was described in 1969. So actually, we've known about this enzyme for many, many years. And due to the preponderance of the disease in the Ashkenazi Jewish population, which has been screened for many, many years, we have a good understanding of enzyme levels and natural history and genotype and phenotype. So a lot is known about this disease. I'm going to share a little bit of data from 2 key publications, which I'd encourage you all to look at. The first is a 2011 [indiscernible] by [ Christine Oren Bley ], Boston Children's Hospital, with [indiscernible] Minneapolis. And what we see here is this is a retrospective study of Tay-Sachs and Sandhoff patients. This was in partnership with the NTSAD, with the patient organization that we work very closely with. And what you can see very nice in the graph on the right, you see this progression of elevated noise sensitivity, exaggerated startle, hypotonia, loss of sitting. So really quite nicely delineated. And you can see specifically the numbers on the left-hand side. You get loss of head control by 9.7 months, loss of ability to vocalize by 14 months. And then early mortality, median around is 47 months. So there's a lot of very granular data on milestone acquisition, regression and loss in this particular population. Next slide, please. This is the second natural history study. It was published in 2017. And essentially mirrored a lot of what was reflected in the earlier [ Christine Bley ] publication. So motor developmental delays seen in the first few months of life, documented specific delays by 12 months. Hypotonia within 6 months of life, seizures and mid-scale vision loss. All generally lost by 2 years of age. And median survival in this particular group of patients, 43.3 months. Next slide, please. Now one of the interesting parts of this second study is that it showed this very nice progression of a child gaining a milestone a little bit later than usual, having the milestone for a period of time and then losing it. So this is very nicely illustrating this time point, this loss of an acquired milestone, which is something we'll need to follow very carefully in the clinical trial. So if you look at the first 2 lines on this graph, this specifically looked at independent head control, which is a key developmental milestone for a baby. Of the 14 patients tested, every patient acquired it, so 100% acquired this milestone. 79% between 0 and 6 months of age, 7% between 7 and 12 months of age. So a little later than normal. But then if you look at the next line down, 93%, so virtually all of the patients then went on and lost this particular milestone. And you see the major loss is at 7 to 12 months and then 13 to 18 months of age. So the natural progression in this particular disease is to actually gain the milestone of head control within the first 6 months of life and then losing it either in the next 6 months or the next 12 months of life. And so this time course of development of progression is really important, is very well described and is actually going to help us very much in terms of running the international trial. Because all these functional changes, these functional gains and losses are very nicely documented already in 2 previously published data sets. Next slide, please. Okay. Going to the construct now itself, we already talked about the fact there's an alpha and a beta subunit to HEXA. And this is the first time ever a bicistronic vector has been used in a clinical setting. So what I mean by bicistronic is that there's 2 genes in 1 capsid. So you have the gene, the alpha subunit or HEXA gene and a gene for the beta subunit or HEXB gene enjoined by a P2A peptide linker, which has clinical precedent, both are driven off the CAG promoter, which is a relatively strong ubiquitous promoter. Now the important thing here is that you're going to get equivalent levels in a 1:1 ratio of the alpha and beta subunit within each cell. So you get lots of alphas and lots of betas in 1 cell. They can combine and form functioning beta-hexosaminidase A. Now an alternative approach will be to package the alphas in 1 capsid, betas in another capsid and co-administer. The problem there is that you'll get lots of cells with alphas, they can't produce functioning HEXA enzyme; lots of cells with betas, they also can't produce functioning HEXA enzyme; and some cells with alphas and betas, which can do. If you're having both the alpha and beta subunit within 1 capsid driven with the same promoter, you get these equivalent levels in a 1:1 ratio. So it's the most efficient way of actually causing enzyme production from transducing a cell. Okay. I have some slides now of preclinical data, 3 specific slides, we can move to the next slide, please. A survival slide, a functional outcome slide and a biomarker slide. This is the survival slide. This is in mice that were dosed at about 5 or 6 weeks of age. And what you see here is a really beautiful dose response. So the mice that are not treated in the blue, they die -- this is the survival curve, they die around about 17 or 18 weeks of age. As you go up in dose with intrathecal dosing of drug, you can see the black line is the low dose, green is the mid, yellows the high. I see a very nice improvement in survival as time progresses. Now you don't see normalization of survival. The reason you've don't is that we're treating these mice quite late, in fact. They're being dosed at 6 weeks of age after the disease has had a really good chance to take a foothold and that's meant significant neuronal loss. One of the earlier studies we ran actually did dose mice the newborn period, and actually saw a very flat survival observed in that. But what we see is a really nice dose response improvement from our seminal mass pharmacology study upon which the CTA was approved up in Canada. Next slide, please. You can see here functional outcomes. This is looking at the rotarod assessment. So this is how far a mouse can travel on one of these rotating wheels. And once again, you see that the knockout/vehicle-treated mice in the blue. They don't do so well, they drop off quite early. And as you go up the back line, green line, yellow line, you actually see a nice improvement in functional capability on the rotarod. In fact, the high dose essentially mirrors the heterozygote/vehicle-treated mouse. So showing close to normalization of neuromuscular outcomes in this particular test. And don't forget, once again, these mouse are those that are 6 weeks of age after their birth. Next slide, please. And what we see here in this slide is a nice drop in accumulated GM2 ganglioside. Now there's 2 important biomarkers in these studies. The first is the amount of enzyme that's present, the HEXA levels in this particular case. Now of course, in a mouse, a mouse only have 35 microliters of CSF. So actually take the sample of CSF in measuring that same level of CSF possible, you can't do it. So we look at the next best thing, which is looking at reduction of accumulated substrate. Don't forget, this is the substrate that the HEXA is due to be breaking down. So we see here the -- once again, the dose-dependent amount, we see a nice drop in the accumulated GM2 substrate for black. For -- in yellow, as you go from there, see a nice drop in accumulated substrate, which correlates very nicely with the neuromuscular functional outcomes and also the survival curves. Next slide, please. In terms of clinical trial design, as I say, we have an open IND equivalent up in Canada, and we'll be looking forward to showing data from that later this year. We have a single-dose cohort for that study 5 x 10^14 total vg. Once again, the very high dose targeted to the brain and spinal cord but a relatively low dosing comparison systemically administered gene therapies because you've given the drug as a total vg rather from vg per kilo. We can, of course, go higher than that if we need to. We have plenty of tox coverage. So we could go up to 1 x 10^15 and even higher if need be. We've got several multiple fold coverage from a tox perspective. But for this initial study, we are planning on just 1 dose cohort. As when we expand into the U.S., which is later on this year, we'll introduced additional dose cohorts. Once again, the drug is to be given intrathecally, covered with prednisolone and rapamycinin. And the patient will be held in the Trendelenburg position for about 1 hour after the infusion. Next slide, please. In terms of -- in terms of end points, you can see really the 2 key areas are the global assessments, disease-specific assessments and also the biomarker activities. In addition, we're looking at a whole host of other endpoints such as quality of life, imaging, echocardiography, MRI. We'll look at hearing function, visual function, communication capability and, of course, seizures, which are an important part of this, and many other diseases that we try and treat. And specifically, we look at EEG, and the neurophysiology. We also look at things like how frequent the seizures, how severe are they, how they last for what medications, improve them. And the intent is to try and reduce seizure severity and duration and frequency as time goes on. Now in terms of cadence of events, what I anticipate happening is that we'll dose patient intrathecally. We'll probably get maximal transgene expression around about being maybe 3 weeks after dosing, and then levels of HEXA should be rising. So I think at the 1-month time point after dose, we may see some elevation of HEXA in the CSF. Likely, it's going to be higher by the 3-month time point. But I think that with the clinical assessments, the key assessments are things like head control scale, CHOP INTEND the Bayley. And just a clinical assessment of hypertonia. I think we should see -- I'd like to see some signs of stabilization by about the 3 months' time point, but there may be a little bit -- a little way after that before we see a significant clinical improvement. Let's go to the next slide, please. And so as I've already mentioned, we've got this study, the open CTA IND equivalent up in Canada. The studies are currently enrolling patients. We are going to be utilizing material from a commercial process to the U.S. study. And there's in general, the pivotal part of any study, we'd like to be doing that with commercial products. We'll be submitting the IND for the U.S. in the second half of this year and initiating the U.S. arm of the study later this year. And as I say, biomarker data for the Queen's study towards the end of this year. Let me stop there and hand over to Kim for questions.

Thanks, Suyash. Now we'll begin the Q&A section for GM2 ganglion-cytosis. The first question comes from Raju Prasad of William Blair. What level of enzyme expression do you believe will provide therapeutic benefit? Is this the same across patients with early, mid or late onset of disease?

Sure. Let me jump in. So I think the key piece of information here was the slide where I looked -- where I showed enzyme levels correlation with clinical outcome. And I'll also reemphasize the fact that a lot is known about this disease from an enzyme relating to clinical outcome perspective. So the infants are less than 0.1%, the juveniles 0.5%, the adults 2% to 4%. My guess is that if you have 5% levels of enzyme, given the enzyme is secreted, you will be seeing quite a dramatic improvement in clinical phenotype. I'd had to get even more than that, but I think 5% will be our minimal threshold. I think even with an adult patient, it's quite clear, it's 2%, 3%, 4%. So getting over 5% will improve their symptomatology. I do think there's one other piece though to this that we shouldn't forget. And in terms of outcome and expectations, it's not just about enzyme level. What it's more about, I think, is treating early. This is a disease where you do get significant progressive loss of neurons. And once you've lost a neuron, it's not coming back. So it's important, I think, to -- more important than enzyme levels, I think, is to actually diagnose patients early on in life and treat them before the symptoms had a chance to really progress. So this is why it's important to spend time focused on new infant screening programs and -- which we are doing in partnerships with Invitae and GeneDX, et cetera. So I think that time to treatment is going to be critical. But from an enzyme level perspective, I think anything over 5% will result in a considerable improvement.

Thanks, Suyash. The next question comes from Yun Zhong of BTIG. What do you think will be the most important developmental milestone or milestones for efficacy analysis? Or is it going to be the totality of the data?

The -- I think that -- I'll answer both of those questions. First of all, yes, it is always the totality of data. The FDA and the regulators really like to see the totality of data. It's not just around one milestone or one end point. They want to see several milestone improve in the same direction. They want to see biomarkers change in the same direction. They want to see the child's ability to communicate improve. They want to see the MRI findings improve. But there is also a hierarchy to that. So I think the totality of data is important. You can see that in the breadth of endpoints we're looking at. If I had to pick one specific milestone, I'd actually probably pick 2. I think they are head control and ability to sit upright. So head control, because you saw very clearly in that natural history data that that is gained and then that is lost. And that's a really important part. That's a really important functional milestone. The child's ability to control their head movements or to raise that is a really important function for a whole host of reasons. The other endpoint, I think, that's really important is the ability to sit upright. I think that just translates to so many other functional capabilities. That's also an endpoint, we'll be looking at from a milestone perspective. But I think more importantly, as the question has suggested, the totality of data is what will really persuade the FDA.

Great. Thank you. And a follow-up question from Yun Zhong of BTIG. If you don't see symptom stabilization at 3 months, what other information we'll be looking at to gain comfort that there will be a treatment benefit.

I think the main thing is enzyme levels. So I think if we are seeing nice improvement in levels of enzyme and the patient was treated not too late, i.e., it's not a patient who is so far affected that the possibility for recoverability are low. I think we see nice improvements in the enzyme, and I think that it will be unusual to see nice improvements in enzyme and no change in disease progression at 3 months. It could just be that you need a little more time to see that clinical effect. But my guess is absolutely, if we treat the patients early enough and we're enrolling the younger patients before the neuron losses had a chance to take hold, if we treat early enough in the disease, we're seeing elevated levels of enzyme, it seems hard to me that you wouldn't see -- it seems unlikely that you wouldn't see an improvement, at least stabilization and disease progression.

Great. Thank you. The next question comes from Joon Lee of Truist. There was a recent data shared from a sheep model of TSD from UMass. Is that a relevant model for GM2? And can you discuss some similarities and differences in the trial conducted and read through to your Phase I data?

Sure. So it wasn't bicistronic vector they used. And we haven't really used that sheep model in our own experiments. There are certain complexities to using a large animal model. And I think that there is definite read-through from that. There were some nice signs in that data package. And I think that probably does actually read through to our approach as well. There are many similarities. But I might ask Steve to see if he is aware of that data and any comment he'd like to make in addition.

I think that there's some extrapolations that you can make, but they have to be very general because it's not the same product. The design is different. I think that the design of that bicistronic vector is with the promoter in the middle, and then it drives expression of 2 different open reading frames for HEXA and HEXB whereas our design is to express the A and B in one open reading frame as a single polypeptide of the self-cleaving linker. So ours is designed for high level of expression at kind of this perfect 1:1 ratio. And it's -- there's just not data that I'm aware of to know how comparable those 2 designs are. So I think I'm encouraged by the benefits that were presented in the sheet at the American Society of Cell & Gene Therapy meeting, but it's not apples and apples. Maybe I'll leave it off there.

Thank you. And there's a follow-up question from Joon Lee of Truist on that same note. Is there any reason why you put HEXB in front of HEXA? And a question on the viral P2A leaves 1 amino acid at the end of HEXB and a few at the end terminus of HEXA. What are the consequences of these additional amino acids?

Yes. So maybe I'll address that, too. In terms of the order of HEXA and HEXB in the expression cassette, it really doesn't matter. Because they're both expressed, it doesn't matter which one is first and which one a second. In terms of the additional amino acids in the linker, there could be a small chance of that representing a unique immune epitope. But since it's really just a couple of amino acids different, we can't always say that that won't happen, but it's not very like. But functionally, I think that Dr. Walia and us with our collaborative studies showed that you do get enzyme activity, hexosaminidase activity with that. So those extra amino acids don't compromise the hexosaminidase activity at all.

Okay. Thank you. And a question from Yun Zhong at BTIG. Have you been able to confirm the 1:1 ratio in the 2 subunit expression in mouse models?

Yes, I have to apologize here. I'd have to go back and look closely at some of the preclinical data that Dr. Walia generated. I can't answer right now.

Thank you. The next question comes from Laura Chico of Wedbush Securities. How do you envision the competitive landscape evolving in GM2? You mentioned stabilization will be important to see in preliminary data. Are there other outcomes of notes?

So Kim, maybe I'll start, and then I can pass it over to Suyash from a clinical perspective. I think when we look at GM2 currently, there is no therapeutic alternative for these patients. And this is a progressive -- severely progressive fatal disease in infants. The 2 approaches, I think, that are currently in clinical development of ours where we have the bicistronic approach which is essentially providing both HEXA, HEXB at the endogenous 1:1 ratio within a single capsid running off of a single promoter. And then you have the competitive approach that it's essentially packaging HEXA in a separate AAV8 capsid, HEXB in a single AAV8 capsid and trying to infuse those simultaneously. Ultimately, unfortunately, through that approach, you're not able to guarantee or kind of force this endogenous 1:1 ratio or even to get close to this endogenous 1:1 ratio because we can't control transduction. I think other important aspects of the current programs that are in the clinic is the fact that we're using intrathecal delivery to target the CNS broadly. And the fact that they're using kind of a combo route of administration to essentially deliver their co-administered capsids. I would say we feel pretty strongly that the intrathecal route of delivery offers us the opportunity to target -- again, to give us broad expression throughout the CNS but also to abate neutralizing antibodies. I would say probably the next key aspect from the 2 programs would be the fact that we're using AAV9 to deliver directly to the CNS to deliver both the bicistronic payload to the CNS, and the competitive program is using AAV8. We know AAV9 has natural tropism for the CNS delivered. From an ID perspective, it crosses the blood-brain barrier, but we know it has really high transduction when delivered directly to the CNS. And a lot of this has been proven out in multiple clinical trials, including the Zolgensma trial, the giant axonal neuropathy trial, CLN6, CLN7, FTD and a number of others, MPS. And I think the competitive program is using AAV8, which is a capsid probably better suited for muscle delivery or IV delivery to muscle or cardiac tissue or to the liver. So we know our construct, and we're fortunate to have Dr. Gray and Dr. Walia leading the way here. But our construct just naturally gives us a competitive advantage, not only for the fact that we're delivering both HEXA and HEXB in the endogenous 1:1 ratio in a single capsid but the fact that we're also evading neutralizing antibodies by delivering -- by choosing the inertial route of administration, but also packaging this in a natural serotype that has tropism for the CNS and AAV9. So we feel pretty comfortable with our competitive advantage.

Thank you. The next question comes from Gill Blum of Needham & Company. Any comments on market research around the approximate 500 EU and U.S. prevalence of GM2? Are the incidence numbers more important in such a severe disease? Are most of those adult and juvenile?

It's a great question. So you're absolutely right. Incidence is extremely important. It all depends, and Suyash mentioned this earlier in his discussion. It's when you're able to diagnose and intervene, you always want to be able to diagnose and intervene quickly. The 500 patients in U.S. and Europe is going to be predominantly focused on the infantile form of the disease, which represents somewhere between 80% to 85% of the patients. We feel pretty strongly that there's probably a large cohort of patients that are actually living in the adult onset aspect or the adult onset phenotype of the disease that are just essentially going undiagnosed. But again, as our program continues to progress, I think we'll learn more and more about the number of patients that may be living in care centers or just simply living undiagnosed. And so I think we'll learn more over time. But that 500 patients in U.S. and Europe is essentially based off of the current prevalent population primarily geared towards the infantile.

Great. Thank you, R.A. And our final question comes from Salveen Richter of Goldman Sachs. How many patient follow-ups will we see in the data in the second half of this year?

Suyash, maybe you want to take that?

Sure. So in terms of patient follow-ups the second half of this year, well, I think the intent is to just share biomarker data by the second half of this year. And then I think the -- as I mentioned earlier, the cadence will likely be biomarker from the initial finding followed by clinical data to later time points. We'll be sharing biomarker data towards the end of this year.

Great. Thank you. And -- Suyash, and this concludes our discussion on GM2. I'd now like to turn the call back over to Suyash to discuss TSHA-118 for CLN1 disease.

Great. Thank you, Kim. So now we're going to move on to CLN1 disease, one of the forms of Batten disease. And we will spend some time just talking a bit about the disease, the natural history of the disease. Steve will talk about preclinical data and some of the science behind the disease, and I'll touch on the clinical trial plans. So next slide, please. So this is a severe progressive neurodegenerative lysosomal storage disorder, also known as one of the forms of Batten disease. There are many, many forms of Batten, probably about 15 described types of Batten disease now. Is caused by mutations in CLN1 gene, and similar to GM2 is a lysosomal storage disorder. So what that means is that in the absence of the enzyme PPT1 you get a buildup of palmitoylated substrate within the lysosome of the cell. The lysosome swell, they rupture, leak out their acidic, their enzymatic content, causing neuronal damage, neuronal cell dysfunction, ultimately, neuronal cell death, which results in the clinical phenotype of CLN1. Now it's similar to GM2, actually, in how it progresses and how it presents, but there's a couple of differences. First of all, it's generally a little less rapid in progression. So disease onset is usually after about 6 months of age. There is ongoing cognitive decline, loss of fine and gross motor skills, seizures. Death usually occurs by 7 years of age in the early infantile onset form of this disease. And it's also a more heterogeneous disease than GM2. For GM2, we have very clear infantile onset, which is the most patients, the juvenile group and an adult-onset group. But the disease is a little more heterogeneous. And I'll take you through some of the natural history data so you can see. Prevalence is estimated to be 900 patients in the U.S. and EU. And you can see the construct design on the right-hand side and is a simple gene replacement therapy approach. So we include the full-length human CLN1 gene, wrapped up -- associated with CBh promoter, which is ubiquitous and strong, wrapped up in an AAV9 sub-complementary package. Next slide, please. So typically, the onset, as I say, is between 22 to 24 months of age for the early infantile onset; late infantile onset, 2 to 4 years; juvenile onset, between 4 to 10 years. The symptoms across the different subtypes are developmental regression and decreased muscle tone and also you see development of ataxia, muscle issues, epilepsy, spasticity, loss of brain tissue, which you can see quite clearly on an MRI scan and, of course, severe feeding difficulties requiring a feed tube. And many of these children are also end up having severe respiratory compromise. Next slide, please. Now importantly, this is a fully new paper that came out, probably, maybe a month or 2 ago. And the lead also is Erika Augustine is a well-known and noted CLN1 or Batten disease experts. And these are some consensus guidelines that were developed with a key group of experts internationally in association with the patient's organization, Taylor's Tale, and some patients and families who contributed to develop these guidelines. And it's really seminal for CLN1 because it really gave a very nice overview of the diagnosis of this disease, how quickly it's diagnosed, how patients are detected, what the form of diagnosis is now, which is generally, of course, mutation testing and enzyme testing. Also, it gave a lot of nice insight on the progression of the disease across the different subtypes. And in addition to that how the disease is treated. Of course, there are no treatments apart from supportive, but the page gave some very nice indication on how to treat such patients from a supportive perspective to try and prolong life as much as possible. They also did a very detailed look at the different phenotypes of the disease. So you can see diagrammatically represented. And this is a specific picture we lifted from the paper. You can see in the infantile onset, the late infantile, juvenile and adult, and the progression of the various clinical features over time. So if you look at the infants, for example, motor dysfunction shortly after birth is noted, cognitive dysfunction thereafter, seizures fairly soon after that, followed by visual loss and then death in the -- approaching age of 10 or sometimes a little bit after the age of 10. And you can see a slightly slow progression in the late infantile, juveniles and the adults. Next slide, please. This also gives more detail of the different phenotypes. For our clinical trial, we're actually including the breadth of phenotype, and this particular study is the respiratory study, and we're looking in particular at biomarkers as well as a clinical outcome, specifically PPT1 activity and reduction of palmitoylated substrates. Next slide, please. And this just gives us a summary of that specific guidance that I've already talked about. So 15 disease experts, 39 caregivers contributed. There's no disease-modifying therapy for CLN1 disease, although clinical trials are being planned. So the current management is based on symptom relief, palliative care. And, of course, early diagnosis is key. And they've touched on some of the important features of many of these severe multisystemic diseases where not much is known. The fact that clinicians lack some experience in management and treating all kind of lead to delays in diagnosis. That is less the case nowadays because some of these diseases are now on these GeneDX or Invitae screening panels so a clinician would see a patient with signs of a suspected metabolic disease and will send them for blood test and CLN1 will be part of that blood test. So actually diagnosis is getting quicker and is getting earlier on with these mutation analysis tests. And of course, this disease requires individualized multidiscipline care. There is a huge burden to the patient caregiver, and society as a whole. It's very expensive. There's lots of permissions, lots of expensive equipments, lots visits to lots of different doctors. So it's important to really understand the health outcome piece and the burden, the economics of illness this disease often have on society. Next slide, please. There are another 2 natural history studies that are ongoing currently. One is based in Hamburg, Germany under Angela Schulz, who's one of the world's expert in lysosomal storage disease in general and in particular, CLNs in general. And also at the University of Rochester. We already mentioned Dr. Erika Augustine and Professor Jon Mink who are overseeing this. And the Rochester study is both a retrospective and a prospective approach. The Hamburg study is an ongoing prospective observational study. And we're working closely with both groups to try and obtain this natural history data and act as comparator and learn what we can in informing international trial. Next slide, please. Now from the Rochester study, you can see the different categories of disease, infantile late infantile, juvenile, with increasing age first symptom depending on the form of the disease. And also in the graph on the right-hand side, you see that the symptomatology, how closely that's related to underlying phenotype. So in the blue, you've got the infantile form, this is the early diagnosed most rapidly progressive form; in the green, the late infantile form; and in the yellow, the juvenile form. So if you look at something like seizures, for example, you see that there's lots of blues early on. So lots of infants are having seizures early on. The late infantile, develop seizures at a later time point and the juveniles develop seizures at an even later time point still. So once again, confirm what we know about the pathophysiology for this disease, where there's an ongoing progressive loss of neurons resulting in ongoing progressive phenotype dependent on the form of the phenotype disease. Next slide, please. Now this particular slide looks at a scale that we are going to be using as one of the key endpoints in the integrational trial. There are 2 specific disease -- sorry, 2 disease-specific end points. One is the UBDRS, Unified Batten Disease Rating Scale, and the other is the Hamburg CLN scale. And you can see here that the UBDRS scales closely correlate quite nicely with the type of disease, whether it's infantile, latent infantile or juvenile. And you can see there's only a progressive in the score of the UBDRS as time goes on. Next slide, please. So that's the brief overview of the natural history data and the construct of the disease as a whole. I'm now going to hand over to Steve. Now Steve led the development of the preclinical package that resulted in the open IND in which we have for this particular study. So I'll hand over to Steve and let him talk about the preclinical package, Steve?

All right. Thanks a lot, Suyash. So this is -- there's obviously a lot of information on this slide, but the main point of this is to convey that there was a wide variety of very large, very detailed studies that were done primarily in my laboratory at University of North Carolina when I was there. And so the first was a proof-of-concept study, testing basically an age and dose response, treating mice at different ages via an intrathecal route of administration. And then we also did a kind of a second study looking at injecting newborn mice via an IV route to essentially try to get in as much of the vector as we possibly could as early as we possibly could and then to take those mice out for 21 months to look for efficacy as well as any signs of adverse effects. The study eventually evolved where we looked at the possibility of doing combination dosing, like an intrathecal injection and an IV injection, and there were different studies done, one at mice that were 4 weeks old and one that were early symptomatic at 20 weeks old. And finally, this culminated in some studies, primary toxicology studies in rats that were done. So if we move to the next slide, this is kind of the overall highlight of what you can see for TSHA-118, where mice that were given an intrathecal injection at 4 weeks old, which is the blue line, or were 12 weeks old, which is the orange line, had significantly prolonged life spans compared to untreated knockout mice, which typically die about 8 months old. I'm going to get a little bit more granular with this in the next slide to really show you that we tested a wide variety of variables. This was a very large study. And I'll highlight here on the left the data -- or before I showed you a 4-week high-dose data, which is the dark orange line, but if we look at treating mice even younger at 1 week old, then that dark blue line is the intrathecal high dose at 1 week old. And you can see that those mice basically had normalized life spans. And when we tested a variety of kind of where we saw a dose response, and you can see that's the 4-week old, the orange lines, where the dotted line is the low dose and then you give them a little bit more and they live longer and you give them a little bit more and then they get close to a normalized life span. So that's all essentially treating the mice before they have symptoms. They have some disease pathology at these ages, but any behavioral phenotypes would be very subtle. If we look at treating mice after the onset of diseases, and these would be like gross phenotypes like motor dysfunction, for example. Then we can see that this is where we really -- it justifies the higher doses that we're moving to and potentially the combination treatments. And you can see that when we dose a 20-week old mice where the -- doing an intrathecal injection alone, which I can highlight the dark blue line, if you can see that, but then we do get a significant extension in life span. But then if we essentially double the dose by giving the mice half intrathecally and then -- well, the full high intrathecal dose plus the same vector dose IV, then we get a further prolongation of their lifespan. And I should say that this combination dosing was really -- it showed us a substantial benefit when we were treating mice after the symptom, after kind of overt symptoms had started. But when we went to the presymptomatic age, the early intervention, the combination dose really didn't provide much benefit over intrathecal alone. So if we move to the next slide, you can look at survival, but we also did a wide cadre of behavioral assessments, and all of these mice went to the UNC Behavioral Phenotyping Core. All these assessments were done blinded longitudinally. And this is a really busy slide, but I think you'll see the same trends here that we saw with the survival where if we treated the mice early, particularly with the high dose and maybe I can highlight the left-hand graph, the dark blue boxes, this would be treating mice at 4 weeks old at the high dose. Then over time, they basically maintained function equivalent to -- or very close to unaffected control mice, which are the gray circles. And then, of course, the untreated knockout mice die by about 8 weeks old. And then on the right, it's sort of the same story. When we're treating the mice, they already have some behavioral deficits, but then the mice that survive retain their motor function basically through their lifespan. So it's not just extending survival in their debilitated state, the mice that survive long term retain, for lack of a better word, a good quality of life. So if we go to the next slide, we did investigate some of the biochemical measures. And in the mice, if we looked at the PPT1 activity, this is the missing enzyme that we're replacing. Interestingly enough, even the mice that got intrathecal injections had super physiological levels of PPT1 circulating in the blood for the life of the animals. And so we can see that it really didn't matter if we treat them pre-symptomatically or post symptomatically, we still achieve these high levels of enzyme activity, and they remain stable. And in particular, the mice that got the neonatal injections, high-dose IV injections at newborn mice, we carried these out for 21 months post injection. They had these high levels of enzyme for their whole lives. And we did a pretty comprehensive histopathological assessment on these mice and looked at blood chemistry, and there were no adverse consequences of this high level of enzyme overexpression that we could find. And we did look pretty hard. So with that, I can -- that will wrap up the preclinical data that's supporting this. I've tried to cover the highlights of this for the sake of time. And I'll pass it back over to Suyash Prasad to go over the clinical trial design for this open IND.

Steve, thank you very much. And yes, it was a really comprehensive preclinical package that I think serves us well in order to translate through to human study. So let's move to the next slide, and I'll tell you a little bit about our study plans. So we already touched on the construct, which you can see here in the middle part of this slide. And specifically, on the dosing, we're going to be starting off with a 5 x 10^14 total vg dose and a cohort with a 1 x 10^15 total vg dose as well. And we'll obviously expand doses as and when necessary as we learn with the clinical trial progressing. As Steve mentioned and as you'll reflect on from the previous slide, there is a very, very wide therapeutic window for this particular condition and indeed, for GM2. So don't forget on the lower end, a little bit of enzyme goes a long way. So you don't need much to have a clinical effect. On the higher end, you can overexpress the enzyme quite dramatically, I see no toxic side effects. So it's very helpful from my perspective as a clinician because a very wide therapeutic window. So we have the 5 x 10^14 total vg as the first dose cohort, 1 x 10^15 total vg, and we can go higher than that if, indeed, we need to. Once again, the drugs will be given intrathecally. Patients will be held in the Trendelenburg position for a period of time thereafter and covered with this immunosuppressive regime of 6 months of prednisolone and 1 year of sirolimus or rapamycin. Next slide, please. In terms of the assessments, we've touched on some of these already based on our early discussion of natural history. But the biomarkers are going to be key. So specifically PPT1 enzyme activity in the CSF and in the serum. Also, we don't have it on the slide, actually, but accumulation of substrate, the palmitoylated substrate that the PPT1 is meant to be breaking down. We'll be looking at both of those biomarkers in some detail. On the clinical assessment side of things, we are fortunate that for this particular disease we do have 2 disease-specific rating scales. One is the Unified Batten Disease Rating Scale, or UBDRS. The other is the Hamburg Scale, obviously, developed from the team in Hamburg, Germany, and both of which have very nice natural history data. We will be looking at several other clinical assessments such as the CHOP INTEND, we'll be looking at seizures, Vineland III adaptive behaviors, Bayley scales. And then you'll have heard earlier in this discussion, this phrase, the totality of data. As a general principle, in all our studies, we do try and capture a breadth of assessments, which if everything is moving in the right direction, even if some of the assessments, the secondary exploratory assessments are not statistically significant, if they're moving in the same direction, it contributes to this notion of totality of data that can be quite persuasive to the regulators. So we're looking at communication skills. We're looking at quality of life, specifically looking at sleep parameters, parenting stress and the PedsQL, which is a very well accepted quality of life measure. We'll be looking at imaging, so EEGs from a neurophysiology perspective and, of course, MRIs looking at the brain. And because of vision, it's one of those things that is affected very early on in this disease and is very troubling to patients, as we talked about when we talked about GAN, we'll be looking at ERG, electroretinogram; OCT, optical coherence tomography to the retinal nerve fiber, layer fitness and, of course, visual acuity using the LogMAR approach. So what I anticipate happening is that we'll dose the patient intrathecally. We'll get maximal transgene expression 2, 3, 4 weeks after dosing. At the 1-month time point, we will take a CSF sample. We hope to see an elevation in the PPT1 enzyme activity. We may not see a reduction in clinical-related assumptions at that point, but I think we'll see -- I think we'll see an increase in the PPT1 enzyme activity at that time point. And hopefully, by the 3-month time point, we'll start to see some signs of clinical stabilization. Next slide, please. So we do have an open IND already with this particular study. So there's 2 things we'll need to do before dose patients. The first is make a few minor tweaks to the protocol, and the second is manufacture some commercial brain material, but we anticipate initiating the study dosing patients later on this year. We are also in the process of talking to European regulators with the intent of opening up a site -- at least 1 site, maybe 2 sites in Europe. And that we're working hard on patient-finding activity in collaboration with our partners at UTSW, Rochester, Hamburg and of course, with the patient organizations and with our screening panel organizations such as Invitae to make sure CLN1 is on all the appropriate panels. Let me stop there and hand over to Kim to see if we have any questions.

Thanks, Suyash, for that presentation. Yes, we do have questions. The first one comes from of Eun Yang of Jefferies. Early intervention is key here, is CLN1 included in newborn screening tests? If not, what efforts are being made to have it included?

Sure, I can make an initial comment and then R.A., Steve, please feel free to jump in. The technology clearly exists to have CLN1 on the newborn screening panel. We understand the enzyme so well. We know many, many, many of the mutations, many of which are actually described in that consensus guideline page that I mentioned earlier on. Unfortunately, from a policy perspective, no one will add a disease to newborn screening peel until there is a treated disease. Personally, I disagree with that approach, but from a policy perspective, that's what happens all around the world. Although there are some moves to change that. So what we're doing is we have our head of government affairs and patient advocacy, Emily McGinnis, who is working very closely on moving that policy forward, not just with the CLN1, but with many of our other programs. She was instrumental in getting SMA on newborn screening panels and we're looking to have some success there. Let me stop there. R.A., did you want to add anything else?

No, I'll let Steve chime in and happy to add after so.

Yes. No, I was just going to add that the technology definitely exists to do this in dried blood spots to add to other tests that are being done with newborn screening. So the technology is there. It can be implemented. It's just a question of execution of the policy.

Yes. The only thing that I would add, and I think Suyash mentioned this, is that there are a number of pilot programs, particularly one in New York state around newborn screening and for a number of diseases where the technology actually exists, particularly these lysosomal storage disorders where the enzyme can be measured relatively accurately in the serum and tested and the assays have been available to accurately test these enzyme levels and to make that information available to patients if and when they present with disease. And so we've been in active discussions around a number of programs within our portfolio that could potentially fall in that bucket around getting them into some of these particular pilot programs. And so we'll continue to do that. I think Suyash mentioned, our Head of Patient Advocacy and Government Affairs, Emily McGinnis, we're fortunate to have her on -- within the company, leading the effort here. She led this similar effort around newborn screening in the SMA community, particularly for Zolgensma. As Zolgensma was wrapping up its clinical development and embarking on registration, she worked extremely closely with the patient advocacy groups around newborn screening. The primary issue here is really around the availability of a therapeutic alternative, and that's kind of the main gating item in order to get newborn screening on both the federal panel and the state panel. And the second key issue, as I just mentioned, is the fact that it needs to be both at a federal level and the state level is something to complicate it. We're fortunate to have walked this path, particularly a few years ago in the time leading up to a Zolgensma approval, and we'll be taking a similar path, particularly in the early days on our giant axonal neuropathy program, our CLN1 program and our GM2 program.

Great. Thank you, R.A. Just a clarifying question from Eun Yang of Jefferies. Is PPT an enzyme-secreted protein that can be easily measured in the CSF?

Yes. Yes, it's a secreted enzyme that is taken up by the mannose 6-phosphate, which is this cross-correction benefit that we have for this lysosomal storage disease. So it's measurable in blood and CSF and in tissue samples.

And I think, Kim, just to add on what Steve said. I think the really interesting thing here, particularly about CLN1 and GM2, the fact that these are both secreted enzymes, the goal really with the construct, and that's why we've gone with pretty strong ubiquitous promoters, is to essentially turn the cells that are transduced within the CNS in the bio factors so that you're trying to get a really high level of expression to be taken up, not only for the cell that has been transduced, but for the enzyme to be able to travel and exert it's symptomatic activity from cell to cell. So this is extremely important, i.e., cross correction. So this is an extremely important aspect when you start to think about some of these diseases where the enzyme is secreted, where you're just not bound from an efficacy perspective to the cell that you transduce, and a lot of -- like a lot of diseases where the protein itself is membrane-balanced. So this is a really key attribute, which ultimately gives you a little bit of comfort from an activity and efficacy standpoint and really one that you're able to take advantage of cross correction.

Great. Thanks, R.A. And Eun Yang also has a -- from Jefferies, has a follow-up question. What ages would be included in the Phase I/II study?

Suyash, do you want to take that?

Sure, I can do that. So different to GM2, we're actually including a fairly broad group of children in this study. We're including the early infantile group, and you'll reflect on the natural history data I shared earlier that they're usually diagnosed before or at the age of 2. We're looking -- we're also including the early -- sort of the late infantile onset group that are diagnosed usually between about the age of 2 and 4 or 5 and then also the juvenile group, diagnosed usually between the age of about 5 or 6 and 12. So we're actually including a broader group than the GM2 population. We are going to ensure that we do have some parity between the different cohorts so we have some equivalents of juveniles and early infantiles and infantiles within each cohort. But it's something of an exploratory study, and it's not as rapidly progressing a disease as GM2. And the heterogeneity is quite broad. There's quite a lot -- quite a big mix of patients. So we wanted for this particular study to go a bit broader than necessarily narrower.

Thanks, Suyash. The next question comes from Raju Prasad of William Blair. What learnings were you able to leverage from other Batten disease programs in clinical development when planning for your clinical studies?

Suyash, do you want to take that and maybe I can provide a comment.

Sure. Let me start and then please jump in, R.A. It's a really valuable having that CLN2 program, frankly. I was actually at BioMarin when that program was running. It wasn't my program, but I spent a lot of time talking to my colleagues there at that time. And several of the endpoints are actually going to be similar. There are quite a few similarities between CLN2 and CLN1. CLN2 is the enzyme replacement therapy for Batten. Our program is CLN1. So for example, the UBDRS, the Unified Batten Disease Rating Scale and also the Hamburg scale. And actually, the -- we're using one of the same centers, Angela Schulz in Hamburg as a clinical trial site. So there's actually a lot of learning just in how to observe a patient prospectively, initially roll them over onto the drug, continue to observe them over time and think about wider assessments to capture as well as just the UBDRS and the Hamburg, which we are doing. I think importantly, from a regulatory perspective, that was -- that model is similar to what we hope to do for several of our programs, i.e., the model where you do a study which encompasses first-in-man but also as time progresses, includes pivotal data with commercial-grade material that can result in an approval based on that one study. So that's what happened with the CLN2 program at BioMarin. And that's what we hope to do not with all our programs but with GM2 and CLN1 certainly, where the disease is severe, rapidly progressive, and there's no other treatments available. Let me start with our -- I'll see if R.A. wants to add anything.

Yes. The only thing that I would add is I had the fortunate position of being able to transact the CLN3 and the CLN6 Batten disease program from a gene therapy perspective from Celenex to Amicus many years ago, that CLN3 program being the lead program there that was developed in the lab of -- at Nationwide Children's. And I think the clinical data that's been shown from that program has shown that using intrathecal delivery, AAV9 gene replacement, you're able to demonstrate a clear halting of disease progression. And I think similar to a number of the programs that we've talked about today, early treatment and early intervention seems to be the overarching factor in order to get the best outcome. Obviously, dose plays into that as well. But I think what's even more important, which is a key difference between our CLN1 program where the enzyme is secreted and the CLN3 program that's currently being developed now by Amicus, is the fact that CLN3 is a membrane-bound protein. And the fact of the matter is that using intrathecal delivery and the AAV9 construct, they were able to transduce enough cells in order to see a really nice halting of disease progression and a pretty progressive population. And I think that gives us a lot of comfort in the fact of the matter, is that intrathecal delivery and AAV9 was able to really ameliorate kind of the progressive nature of CLN3. And when you compare that to our program, again, where we're going in with a relatively high dose from an intrathecal perspective, low dose when you think systemically, but really somewhat of a real dose and one that should be able to convey benefits. The bar for us because our enzyme here is secreted is a lot lower. And so really, the goal is to go -- is to transduce as many cells as possible and secrete out as much enzyme into the CSF as possible. But really having that data set from CLN3, I think, gives us a lot of comfort and really, I think, improves the risk profile, increasing the probability of success in this particular program.

Thanks, R.A. And actually, you hit a great point that one of our analysts from Cantor, Kristen Kluska, was going to ask. So you've touched on the read-throughs that you see from the CLN3 and 6 programs. And could you talk a little more about the rationale to use this higher dose versus what's been used in these trials? Is this based on tox coverage established?

Suyash, maybe do you want to chime in, and Steve?

Sure. I can jump in and then Steve can also make a comment. So we're starting in the CLN1 studies at a 5 x 10^14 dose, which certainly is a high dose being given intrathecally. There aren't many studies that have gone up to that high level. But we feel very confident with that dose because we have such good tops coverage. But we have even broader tops coverage. So with these diseases, we're also happy to adopt the 1 x 10^15 total vg, which we can deliver intrathecally. And for these diseases that are severe, rapidly progressive, that caused death early on, you've got to give the children every best hope of success. So if you can dose high based on preclinical findings and one's level of comfort, I think you should feel free to dose highly because you're going to want to give the kids every chance of success. And don't forget, even though we're giving a dose -- even though there's a high dose that can be given intrathecally, there's actually a relatively low dose in comparison to the systemically administered drugs. And let me give you a specific example. If a systemically administered drug for Duchenne, if you're giving 3 vg per kilo, if you have a 6-year-old child who weighs 20 kilos, you multiply that 3 x 10^14 by 20, so you're close to a 1 x 10^6 total vg alone. So a 5 x 10^14 total vg being given intrathecally is 1/20 of that. Even the 1 x 10^15 dose being given intrathecal is 1/10 of that dose, which means the systemic exposure is dramatically reduced in comparison to the systemic gene therapies and the risk of a side effect are low. And don't forget, in addition to that, we're covering with steroid and rapamycin. So if I can comment on the immunology -- immunosuppressive regime for gene therapy. A few years ago, they weren't really used. And as time went on, some patients developed this Batten inflammation that was thought, okay, maybe this is responsive to steroids because it looks like an autoimmune hepatitis. Steroids are tried reactively to these -- to the inflammatory process that was going on and the ALT, AST inflammation. As time went on, it was decided they should be given the begin a short course of steroids prior to dose with gene therapy. As time progressed, those short course of steroids actually went longer and longer to the point where they then have 6 months, which is what we're using. And on top of that, we've added in rapamycin. That was really based on the learnings that Carsten and Steve had from the giant axonal neuropathy payer program in the first couple of patients. And on that regime of immunosuppressive medication, we're actually really seeing minimal side effects. So all those things gives us confidence to have the dose [ climb ] aggressively for this particular disease and for GM2 where there is this very wide therapeutic window and minimal risk of toxicity and where we have to give the patient every best chance of success.

Great. Thanks, Suyash.

Did you want to...

Yes, maybe I'll just make one last comment. I mean we can look at other clinical programs and have encouragement and gain knowledge from that. But ultimately, as we're moving forward with this program, we have to rely on our own preclinical data. And also considering CLN1 is -- it is more aggressive than any of the other forms of Batten that does deteriorate faster. So those doses really came -- they were extrapolated out from the preclinical studies in the mice. And so that's best information that we have.

Thanks, Steve. The next question comes from Gil Blum of Needham & Company. Could you discuss the relative benefit of using self-complementary AAV versus single-stranded AAV?

Steve, do you want to take that question?

Yes. I was just thinking I can jump in on that. So the single-stranded AAV versus self-complementary AAV, this is something where the self-complementary, I think people understand that you'll get expression a little bit faster. But in fact, if you look at -- the self-complementary has an advantage that when the vector genome enters the cell and uncoats the nucleus, it has to be converted to a stable episome for you to get sustained transgene expression long term. For a single-stranded AAV or just AAV in general, that's actually a very -- a fairly inefficient process. But the self-complementary design of the genome actually facilitates that process better. So in other words, the self-complementary vector genome design is much more likely to form a stable episome conferring long-term transduction to a targeted cell. So what we had published in 2011 and other groups have published is from kind of a systemic administration where you're not saturating the target, then the self-complementary design actually leads to about 10x more cells being transduced long term. So that's the advantage of self-complementary. And any time that we can fit the packaged genome into that compact self-complementary design, we do that.

Okay. Thank you, Steve. That's all the time we have now for the CLN1 section. So happy to pass it back to Suyash to discuss Rett syndrome.

Thank you, Kim. And so we're going to spend a few minutes now talking about Rett syndrome. And as you'll all remember, the intent is to initiate the clinical trial and an open IND or CTA for this by the end of the year. So I'll spend a few moments talking about the disease and the -- some of the molecular biology underpinning the disease, and then Steve will delve deeper in the science and the preclinical data before I round off with some comments on the clinical trial. And next slide, please. So Rett syndrome. This is a really awful, difficult disease. And as a pediatrician, I actually saw a number of cases of this in my clinical practice some years ago. And it is a rare disease. It's officially designated a rare disease, but it's one of the common rare diseases. The prevalence is about 25,000 patients in the U.S. and EU. So a pediatric neurologist will see one of these a week in that clinic. So it really is a disease which is well recognized and well known, well understood. Sadly, there are no treatments for this disease. And once the diagnosis is made, the course is generally set. So it's a very, very devastating condition for the patient and family because for the first few months of life, these children are normal. They're born. They look beautiful. They start babbling at the right time. They start reaching objects at the right time. They sit up right at the right time. But at some point towards the end of the first year of life, the parents realize something is wrong. And often what's noticed is that the child loses their acquired scale of purposeful hand movement. So by the age of a year, roughly, a child will be able to reach out, grasp an object, pass an object from hand to hand, maybe have a slight [ improvement ] being able to pick up jellybean or something. But then these purposeful hand movements end up being replaced by this hand clapping or hand rimming type movements. And then thereafter, there's a fairly rapid deterioration. The children lose the abilities to make eye contact. They lose their, whatever, babbling or speech they've gained pretty quickly. They continue to lose that purposeful hand use. They start to develop seizures over time. They have a number of autonomic features that develop. So respiratory rhythm out of the mouth where they alternatively breathe very rapidly and then very shallowly. It causes tremendous amount of distress. They have gastrointestinal features and cardiac arrhythmias as well. About 20% of these children or into kind of the older children or young adults who develop the QT abnormality and the associated risk of sudden death. So there is this very rapid deterioration for the next couple of years, and then they stabilize out into a stable patient but in a very functionally compromised situation where they're not moving. They're not really able to communicate whatsoever. There may be some function going on their brain. It's very difficult to tell. And they stay in this stable state for many, many years and then enter the deterioration phase and then ultimately succumb to their illness. So there is a large prevalent population out there. Now the underlying defect is due to a mutation in the excellent MECP2 or MECP2 gene. And MECP2 is a protein that regulates the expression of many genes involved in normal brain functions. So in the [ absence ] of MECP2, you get this global developmental delay that results in the clinical phenotype that I've just described. Now there's a couple of important things to know about this disease. The first is that this -- it's felt to be highly reversible. And the field shifted its paradigm of thinking a few years ago. There was a publication that came out from Professor Sir Adrian Bird from Edinburgh that demonstrated reversibility in the animal model. Now I say it's reversible for 3 reasons. First of all, there's this decline, but it's not a decline towards death. It's a decline on the stabilization. And that suggests to me that there is some recoverable brain tissue if the children are stable and not declining in an ongoing way. That's the first point. The second point is that if you look at the autopsies of brains of children with Rett, they don't have a loss of neurons. They don't lose their neurons. What they lose are their somatic connections, the branches that come off each neuron. So if you're a healthy individual, every neuron you have in the brain has at least 1,000 dendrites coming off touching other neurons and what is built is this dense network of synaptic connectivity. In Rett syndrome, you have a paucity of these dendrites. It's what's called the lack of dendritic arborization, arborization meaning branching out. You have far few of these branches, far less synaptic connectivity, and that's what results in the clinical situation of Rett syndrome. So if you can somehow encourage the growth of those dendrites, you should be able to ameliorate many of the signs and syndromes of Rett. And [ the professor ] actually did that in the mouse model very nicely, which was published in 2007, has been replicated several times since then. And what he did was he took some mice, inserted transgenes cassette, turned off the MECP2 gene, thereby these mice exhibited features of Rett. He then turned back on the MECP2 gene by administration of an exogenous small molecule, which then allowed MECP2 to start expressing again. And then the clinical features, the functional features of Rett syndrome in these mice, dissipated. So he was able to show reversibility of Rett syndrome in this mouse model. And so for those reasons, it's felt the Rett syndrome is really likely to be highly reversible. Next slide, please. Now there's 2 important things to know about the molecular biology of Rett in order to understand that the solution that Steve and his colleague, Sarah Sinnett, has worked on is a really elegant approach to meeting the challenges of development of gene therapy of this disease. First of all, the MECP2 is toxic. So this is what we call a Goldilocks disease where you have too little MECP2 on one end and you have Rett syndrome, too much MECP2 on the other end and you have what's known as MECP2 duplication syndromes. There is a disease known as MECP2 duplication syndrome, where children have 2 copies of the MECP2 gene, double the amount of MECP2, and they have a very serious clinical phenotype developmental regression and many similarities to Rett but actually is probably a little more rapidly progressive than Rett. So you need to make sure that the MECP2 that's produced stays within these physiological parameters. Too little, you get Rett. Too much, you get MECP2 duplication or overexpression toxicity. Now this becomes very challenging because Rett syndrome is an X-linked dominant disease. And by why the process of lyonization or random X inactivation, half the cells in the body of a girl with Rett have normal levels of MECP2 because they have their functioning X chromosome. The other half have abnormal or mutated MECP2. So this 50-50 ratio of wild-type cells versus MECP2 -- so when you take MECP2 cells, which you can see in the diagram on the left side, this is Rett syndrome, you got 50-50 level of wild-type versus abnormal mutated MECP2 cells. That 50-50 ratio is what results in the clinical phenotype of Rett. So what you've got to do when you administer gene therapy is in the cells that have no MECP2, you bring up levels of MECP2 to the point which is functional. In the cells that already have MECP2, you have minimally expressed MECP2 or not expressed MECP2, or otherwise, it becomes toxic. So the amount of MECP2 expression has to be regulated on a cell-by-cell basis. This has posed many, many challenges from a gene replacement perspective, but as Steve said, there is really a very elegant way of self-regulating MECP2 on a cell-by-cell basis. And on that note, we'll go to the next slide, and I'll hand over to Steve to take us through some historical context and the construct and how this self-regulating construct works. Steve?

Thank you, Suyash. Yes. So this gene therapy approach for Rett syndrome is something that my lab has been working on since 2007. It's been, I'd say, the most -- or one of the most challenging diseases to try to attack with this level of dose sensitivity, and it really required what I'd term breakthrough technology in this miRARE regulatory element. And so if you look at the design of the vector genome, it's fit into a self-complementary genome, uses a fragment of the endogenous MECP2 promoter. And the key feature is really this miRARE element that I'll go into and explain how that works in a moment. And in order to package the appropriate regulatory elements within this, we utilized a miniature version of the MECP2 transgene that was published by Sir Adrian Bird. So if we go to the next slide. I'll try to walk you through how this regulation system works because this is really the key feature that I think enables this to function in a safe way -- safe and effective way. This miRARE element is denoted by a blue diamond in the vector genome. So if we think about that, the vector genome will get delivered to the nucleus, and then it will start expressing the transgene. So we have the messenger RNA here. You get expression of MECP2 going on to step 4. And something that is a feature that we really investigated and was important to developing miRARE is that whenever you overexpress MECP2 in a cell, there are certain microRNAs, and these are endogenous cellular microRNAs. There are certain cellular microRNAs that get upregulated when -- in response to MECP2 expression. So if MECP2 is very low, these microRNAs are low. If MECP2 gets high, these microRNAs get high. So the key feature of how this regulation works and how it's a true feedback loop is that if MECP2 levels get up, and then these microRNAs will also rise, and then they come back and they bind to this miRARE element, which is a series of microRNA binding sites. And then they inhibit expression of the transgene. So again, just to emphasize this, we're not expressing any microRNAs off of our vector. We have simply built in essentially a microRNA sensor, a series of microRNA binding sites that basically respond to the cell's response to overexpress MECP2. So if we go to the next slide. I'm not going to go through all of this, obviously, but it's really to emphasize that it was a fairly involved process to develop -- to identify the microRNA binding sites to include in miRARE and to really develop this kind of platform. There are 2 key features of how we did this. One was empirically, we deliberately overdosed mice with an unregulated MECP2 vector. And then we identified microRNAs that elevated in the brain in response to that MECP2 overexpression. So that was empirical. And then once we had that panel of candidate microRNAs, we also did a bioinformatics approach to look across the 3 prime UTRs of multiple developmental -- developmentally regulated genes. And in fact, we saw that a lot of those microRNAs that we identified empirically showed up at a higher -- relatively high prevalence in these developmentally regulated genes. So this is not just MECP2, but this is also other developmentally regulated genes. And so we took that empirical data, that bioinformatics data. And ultimately, we narrowed it down to a select panel of microRNAs by looking for microRNA binding sites that were 100% conserved between mice and humans and also these microRNAs that were at relatively high expression levels in the cell. So if you want more information on how we develop this miRARE element, if you go to the next slide, we just recently published this in Brain. So you can see all the details about how this was developed in the proof-of-concept studies, where we demonstrated how it works. And I'm going to go through a couple of highlights from that paper in the next few slides. So if you go to the next slide. I'm going to start by looking at the RNA level of -- so this is where knockout mice were injected with a vector expressing the miniMeCP2 transgene with the miRARE element in blue and without the miRARE element in green. So the left side is just looking at DNA copies in the brain and different levels of the spinal cord per vector genome -- or per host genome. And so it's basically the blue and the green lines in the left panel are the same. So we're delivering the transgene to all of these targets to the same extent. But if we look at the RNA, the mRNA in the right panel. Again, these vectors are exactly the same, except that the blue contains the miRARE element. You can see that it's approximately a tenfold reduction in the level of MECP2 RNA that's being expressed in the cells. So that's about the level of knockdown that we're getting with the miRARE element. So we want to look at this at the protein level because there was also a question about whether these microRNA binding sites are just providing a general knockdown of expression or if it is actually genotype dependent and responsive to MECP2 dose levels. So if you go to the next slide. We looked at MECP2 protein expression, and this was done by histology, staining for MECP2 in various regions of the brain and then quantifying the level of MECP2 protein expression. So if we look at a nonregulated transgene, and that's in the gray and the blue, if we dose wild-type mice or knockout mice, you can look across the different brain structures. And we get the same level of expression, whether it's a wild-type mouse that has MECP2 or knockout mouse that doesn't, the transgene expresses to the same extent. In contrast, if we look at the mice dosed with the miRARE regulated vector, and that's in green in wild-type mice and orange in knockout mice, then if I can get you to focus on the pons and the mid-brain, these would be areas important, particularly for autonomic function, breathing dysfunction, things like that. And then if you look in the wild-type mice, we get this repression of MECP2 protein expression. So these mice that already have endogenous MECP2, if the miRARE regulated vector goes in, then we suppress the expression. In contrast, the knockout mice within the miRARE regulated vector goes in to these null cells, then it permits expression of the MECP2 transgene. So I hope that you'll agree that this is strong evidence that we're getting actually genotype-specific regulation, which can solve -- is really -- or solve this issue of trying to deal with these heterozygous females that have some wild-type cells and some null cells that we're dosing at the same time. So if we go to the next slide, Slide 88. This is just really to show you visually what this looks like. And again, if you focus on the 2 images, the right with green, then if we go into a wild-type mouse, the green is showing MECP2 expression. So in the wild-type mice, we get -- we suppress the transgene, whereas in the knockout mouse, we're actually permitting transgene expression. So if you go to the next slide. I'm going to go through some of the functional kind of safety and efficacy data. And this really hits home the importance of why this regulation is important. We -- this was essentially a safety study initially in wild-type mice, where you can see, let's see, in blue and in light green -- well, this is a Kaplan-Meier survival plot. And again, these are wild-type mice so they shouldn't be dying. But in dark blue and in light green, you can see that if we give high doses of the unregulated MECP2 gene, either the full-length gene or the mini-gene, then in the end, we get about 40% mortality in wild-type mice versus if we do the miRARE regulated vector, which is in orange or red, then this is safe and well-tolerated wild-type mice. So this is survival. If we go to the next slide. We can actually quantify these adverse effects from a behavioral standpoint. This is an aggregate score of Rett-like behaviors. So if MECP2 gets overexpressed, you actually -- the mice will exhibit Rett-like behaviors as a duplication phenotype. So in black are our control mice. And these are all, again, wild-type mice. They shouldn't be developing significant Rett-like behaviors, but you do have some of this that shows up in control mice. So again, black is what we're comparing this to. If we have our unregulated vectors in blue and in green, you can see that they develop and then they persist to have these adverse Rett-like behaviors, whereas our regulated vector in orange and red tracks pretty similar to the vehicle-injected mice. So again, from a mortality perspective as well as an adverse behavioral perspective, the TSHA-102 is well tolerated in mice, whereas a nonregulated vector is not. So I'm going to end with one more slide. Next slide is just showing that we do get some efficacy with this vector. And this is a survival study in knockout mice, which is a severe mouse model of Rett syndrome in gray, is our untreated mice, and then in orange are our mice treated intrathecally at 4 to 5 weeks old with the TSHA-102, where we see a nice extension in survival. And this matches what we would get previously with a full-length MECP2 vector that we've published, which is in blue. So basically, it has the same survival benefit as a traditional full-length vector, but our regulated vector really provides a much greater safety, where, I'd say, a nonregulated vector from a safety perspective is going to have a hard time getting dosed properly in humans. So with that, I think that wraps up the preclinical data and the explanation of how miRARE works, and I'm happy to pass this back to Suyash Prasad to talk about what the next steps are for this program.

Thank you, Steve, for the very nice explanation. You can see that there is this challenge of trying to keep MECP2 within appropriate parameters and the fact that the MECP2 levels need to be regulated on a cell-by-cell basis. And Steve's approach does this very, very nicely, and that's evidenced by the preclinical data that Steve was able to show. And I encourage you all to look at the paper that's published very recently in Brain, probably about -- probably a couple of months ago now. Sarah Sinnett is the lead author of this. She is the senior author. And the paper really explains a lot of detail how all this works to complement what Steve has just mentioned. Okay. So in terms of what we're planning from a clinical development perspective, if we go to the next slide, we do have some thoughts on how the clinical trial is going to look. Of course, this depends a little bit on our discussions with regulators over the next few months. But this is at the height of what we're thinking we're planning to do. You can see the construct is in the middle of the graph, the miniMECP2 gene, the miRARE driven off an endogenous promoter. This study is going to be a little different to our other clinical development programs where rather than this attempt for one and done given the rapid progressive nature of some of the diseases such as GM2 and CLN1 and the rarity of the disease, we just feel like coming on in one study and aggressively dosing is the right approach for those conditions, which results in a high mortality. For Rett, it's a little different. We have a little more time. The disease is not as rapidly progressive or doesn't result in death and quite the same mobility. And we have this obvious concern about the risk of overexpression in contrast to some of our other programs where we have this very wide therapeutic window. That's not the case for Rett syndrome. So we'll go to be a little more conservative in our approach. So the intent is for the first study to look at adult females with Rett, demonstrate safety, get some preliminary sense of efficacy and then fairly soon afterwards, expand to the younger patients and think about doing a pivotal study in that population. The plan will be to have at least 2 dose cohorts, likely a 5 x 10^14 total vg cohort and the 1 x 10^15 total vg cohort. That may modify a little bit as we get a little more information from our pharmacology and toxicology studies with commercial product. But those will be the general ballpark doses we use. We will be including randomization and -- in an open-label manner within each cohort to try and bring a bit more additional rigor and robustness to the clinical trial. So one patient in each cohort will be delayed, will be treat -- will be randomized with the late treatment arm. So they won't be blinded and not treated with placebo. They'll be randomized and observed for a period of time before being dosed, and that adds -- that patient then acts as a comparator to the rest of the patients within the cohort. So from the FDA regulatory perspective, what you're doing in this situation is you're trying to reduce bias by bringing some randomization in those not blinding, which will not be applicable in this kind of study. Once again, we're giving the drug intrathecally, and we'll be keeping the patient in a head down position for about an hour after dosing to maximize pooling around the brain and spinal cord. Next slide, please. In terms of endpoints, we unfortunately don't have any biomarker endpoints for Rett, or at least no biomarkers that are very well associated with Rett and/or are highly clinical correlative. We do have a group of exploratory biomarkers we're looking at. We look at a whole host of different parameters such as neurofilament and a number of other metabolomic approach and assessment. But in general, there are no good biomarkers for Rett. So we're going to rely more on clinical assessments. So you can see how the clinical assessment is characterized on this slide. There are going to be some Rett-specific global assessments such as the Motor Behavior Assessment Scale, the Hand Apraxia Scale. We'll look at hand function, the behavior questionnaire and the functional mobility scale. And many of these scales have been used in the natural history data, in natural history context for Rett syndrome. And we do actually have some partnerships with some of the patient organizations here, the Rett Syndrome Research Trust and the Rettsyndrome.org, for example, where we have access or are in the process of obtaining access to the large natural history data sets where we could use them to help inform our clinical trial and act as comparatives for our clinical trial. So those are the Rett assessments. We'll be looking at some aspects of behavior, anxiety. We'll be looking at seizures. Seizures are a big part of Rett syndrome. So how frequent seizures come on? What triggers the seizures? What medication the patients are on? And if we can reduce the severity, duration, triggering, aspects of seizures and the medications, that will definitely be a win. We'll be looking at respiratory assessments. There is this respiratory rhythm disturbance associated with Rett, which is specifically related to some of the autonomic dysfunction, autonomic nervous system dysfunction, which we've already touched on when I talked about GAN. There will also be -- there is also sleep apnea, which forms a significant component of this disease. We'll be looking at communication and quality of life. And also increasingly, people are starting to use wearable devices as part of clinical trials. It's not entirely clear what the best wearable is for Rett yet, but we're doing a lot of work. And one that's emerging very nicely is this Hexoskin, which is a very tight vest that the children wear or the adults wear, and it picks up all kinds of assessments. It picks up cardiac activity, respiratory activity, abnormal movements, sleep activity. And it's a way of getting large data sets that they can then be computationally analyzed. As I said, it's something with our exploratory endpoint that a lot of people are pacing a lot of time and energy on these endpoints. So the intent is for this study to start towards the end of the year. And I would hope to be able to share clinical data from the study at some point during the course of next year. Next slide, please. So the INDs and CTAs are going in by the end of this year, and the plan is to initiate studies by the end of this year. We're in the process of completing our GMP manufacturing using commercial process. And we are partway through our pre-IND and CTA scientific advice meetings on Rett syndrome. So we're very excited to be showcasing Rett at this meeting. It's great to have Steve on board today to tell us a bit more about the really elegant, nuanced science in designing this construct. And I know Steve has been working on this for many, many years. So it's great to see it coming to fruition. So on that note, I'm going to end my part of this R&D Day. I think we're going to move over to Q&A before closing the meeting. So let me hand back to Kim for a Q&A on Rett.

Thanks, Suyash. So first question comes from Eun Yang of Jefferies. In treated mice, it slowed time to death, but it does not seem robust. Why do you think that is? And what can you do to enhance the effects?

Yes. So this is Steve. So the way that we designed this experiment was a couple of things. One, we were using male Rett knockout mice, which are quite severe model of Rett syndrome. It's more severe than a human patients would be. And we also treated them at 4 to 5 weeks old. So treating mice at that late in the disease progression, that type of extension in survival is about the best that anybody has seen with any treatment. But it's a way for us to screen the mice or screen therapeutics quickly. The -- if -- for example, other publications have treated mice as newborns and the efficacy results look much better as you might expect. So anyway, that -- the short answer is the mice were very severe model. They were treated late. And this was really just a way to get an initial read on therapeutic effect.

Yes. Just to add to what Steve said, I think this is an important point because I think this is consistent with the way that Steve and the group at UT Southwestern conduct a lot of our early translational work. It's essentially to try to mimic what's happening in a clinical study. And unless there's newborn screening, you're never going to be able to treat a patient at the first day of light. If you translate that to a mouse, that would be P0, P1. So it's a little bit of an artificial bar. And essentially, what you want to do is to take into account kind of the disease progressing to a level of symptoms that would kind of allow for a successful diagnosis. And I think that's where we always see, and you kind of can see this is consistent across our preclinical studies, us treating at 4 weeks of age, 5 weeks of age. And again, this is a much higher bar because they've been significant neuronal damage. In the case of neurodegenerative diseases, neurons dying off. In the case of neurodevelopmental disease, development loss. And so I think this is an extremely important point. Leaning back to what Steve mentioned, you can see that our extension in survival in the knockout model, which is much more severe than the human disease, is statistically significant. And when you compare that with the unregulated vector, it's pretty -- it is pretty comparable. And then if you try and take that translation back to what Steve was saying where other groups have basically taken almost an identical vector being the unregulated, self-complementary AAV9, unregulated MECP2, you get pretty good survival. So we feel pretty comfortable the earlier that we treat, the better that the outcome is going to be, but we wanted to make sure that we were checking a couple of boxes: one, that this would be truly translational when you get -- or as translational as you can get with a pretty severe mouse model once the mouse actually demonstrated some symptoms compared to when a patient would actually be diagnosed. That's an extremely important bar. And two was the safety benefit that we demonstrated in wild-type mouse consistent when we went over to the knockout efficacy model.

Yes. I would add one more comment to complement R.A. and Steve's thoughts. And that is just that there is a clinical correlate to that wild type -- sorry, that knockout mouse, which is the male Rett patient. So if you think about the females, the X-linked dominant, which means that the females have one functioning -- half of their cells have one functioning X chromosome and the other half of their cells have no functioning X, which results in this 50-50 ratio of wild type versus mutated or absent MECP2. In the males with Rett, they all have no MECP2 whatsoever. So it's equivalent to the knockout mouse in some respects. Now the majority of these mouse die in utero, is incredibly severe. And a handful of them survive to birth, and then they usually die within a few weeks after birth. There's a handful, less than 100 of these patients, who actually survive out even further. So it's maybe a year. So it's an incredibly severe form of Rett that's uniformly lethal. The vast majority of this is lethal in utero. And so it really is a high bar from a preclinical perspective to try and treat some of these knockout mice because their survival is extremely compromised. And as Steve has already said, the data we're getting is far, far superior to anything else that's been shown in that particular model.

Thanks, everyone. The next question comes from Yun Zhong of BTIG. What is the likelihood that the miRNA induced by MECP2 expression can affect the expression of other developmental genes like off-target effect?

Yes. So this is Steve. I'll jump in on that. So I'd say if we were not regulating MECP2 expression, then that would actually be a pretty strong concern. But in our case, I want to emphasize, we are not expressing any of the microRNAs off of the vector. All of those are coming from the cell, their endogenous microRNAs. And so as long as we cap the expression of MECP2 and we don't have an instance of MECP2 overexpression, then you're not going to have any resulting overelevation of those cellular microRNAs. So these things go hand in hand. And I think so far, we have not seen any instances of MECP2 overexpression after gene transfer with our miRARE regulated cassette.

Thanks, Steve. The next question comes from Gil Blum of Needham & Company. Any chance around the biological function of the MECP2-driven miRNA sequences?

Yes. I mean this is an interesting question. MECP2 regulates -- or not -- MECP2 affect the expression of literally hundreds of other genes. And so it's maybe not surprising that it affects the expression of the certain microRNAs as well. We don't know if this is kind of a bystander effect. I think it -- scientifically, it would be interesting for me to think that since we see an overabundance of these microRNA targets in a number of developmentally regulated genes, then there may be some biology behind that where this is a natural mechanism of the cell to kind of fine-tune level -- expression levels of these dose-sensitive genes. But clearly, that type of mechanism is not effective enough endogenously where we see we have things like MECP2 duplication syndrome. So in essence, this may be some level of an endogenous mechanism for fine-tuning gene expression for these certain genes. But what we've done with our synthetic design of this miRARE element is to essentially make that mechanism more potent. So it prevents overexpression and we've got multiple copies of transgene.

Thanks, Steve. The next question comes from Laura Chico of Wedbush. Could you discuss how you will prioritize candidates beyond confirmed MECP2 mutational status? What other inclusion/exclusion criteria are most important in terms of patient identification for the Phase I/II study?

Suyash, maybe you want to take that?

Sure, I can take that. I think the vast majority of patients with a Rett mutation will be eligible. There are certain groups that won't be. I've already mentioned one group. So for example, boys who have Rett will not be eligible. There are certain unusual rare mutations on different patterns of mosaicism that may also not be eligible. But the vast majority of patients with Rett mutations will be eligible certainly for the initial clinical study. As I've mentioned, it's going to be the older patients first. So it's going to be adults over the age of 18 years of age, and they will have to be in stable -- in this stable phase as opposed to a low, deteriorating, declining phase. And the reason for that is we want them to be stable so it's easier to then tease out any kind of preliminary improvement from an efficacy perspective. So this is primarily a safety study, but we are importantly looking for some signs of efficacy as well. And it's much harder to tease that out if a patient has a very baseline or on the declining path. So the vast majority of young women with Rett will be in a stable situation. So that's the group of patients we'll be including. If they're very severely afflicted or they are declining, they will not be eligible for the study either. That's the first study. Once we've dosed a few patients, got some of that safety data, reassured ourselves that we're getting MECP2 levels with the right level and hopefully seeing some signs of preliminary efficacy, we'll relatively quickly drop down into a younger patient age group. An intent with the younger patients will be to actually catch them either when they just stabilize that or as they rapidly -- as they go through the rapid deterioration phase. If you remember from the slide I presented earlier, there's 4 phases of development. One is the early period of clinical symptoms of diagnosis. then there's a period of rapid descent. But in general, even though we feel that this is quite reversible, in general, the sooner you treat the disease, the better. So once we've fished our adult study relatively quickly, and we've already started designing and thinking on the study, we're going to move to the children dosed at the younger age group and identify them earlier and treat them as rapidly as possible. That's our current thinking.

Thanks, Suyash. And are there any other patient subgroups that you would think to target? This comes from Yun Zhong of BTIG.

Yes, yes, there are. I mean one obvious one once again that I mentioned is the boys with Rett. There aren't many of these patients. It's really an ultra-orphan indication and likely will need a different dose. But the risk of overexpression toxicity in these boys is much reduced because they have no MECP2 whatsoever. They're like the knockout mice. So we won't be targeting that subgroup initially, but in the future time point, once we've got some safety and efficacy data, I think it's likely we'll target that group. It's not a large population, maybe 100 patients in the world. So that's something that would certainly be of service to the patient community.

Great. Thanks, Suyash. Now the next question comes from Yun Zhong of BTIG. Can neurotransmitters such as glutamate and GABA be used as biomarkers in Rett syndrome?

Steve, do you want to take that?

I'm not sure about that. It may be too -- well, I'm not even sure if you looked in the CSF if they're highly different, but I'm really not sure.

Suyash, do you want to talk about some of the exploratory things that are kind of going on within the Rett community?

I can do that, yes. So I think certainly, neurotransmitters like GABA, there's nothing really much known about its association with Rett. We spent a lot of time talking with some of the experts in Rett. We had about 2 Advisory Boards, and we attended a biomarker symposium that was run by the Rett -- I think it's the Rett Syndrome Research Trust. There's a really high-quality discussion on biomarkers in Rett. And unfortunately, we haven't got a good biomarker for Rett. We -- the whole field has been looking for one because if we can get a read on something that might work, that will be really, really helpful. There are a few glimmerings. There are some possibilities of things out there. And what we're actually doing is we're going to save serum. We're going to save CSF. And we're going to be performing profiles of metabolomic, proteomic and genomic profiles on CSF blood and saving serum. So if a biomarker does come up, or indeed, if we find a biomarker as part of our clinical trial, then we'd love to be able to use that. But at the moment, it's really hard from a specific biomarker perspective. What I will say though is if you put aside the CSF and blood-based biomarkers, there are arguments to use things on these wearable devices as something with biomarker -- an EEG, look at baseline EEG while children are not having a seizure as biomarkers because abnormalities in the basal EEGs -- and we'll be doing all that in the clinical trials. So most of the exploration work -- most of the work we're doing from a biomarker perspective are exploratory in nature in this study. But we're staying very close to the key opinion leader network and the science to see if there's something there. But as Steve says, at the moment, there's nothing that looks particularly promising.

Great. Thanks, Suyash. Our next question comes from Joon Lee of Truist. Does MECP2 promoter recapitulate expression for natural motor? And is it possible to use other well-characterized promoters in this sector?

Yes. So the MECP2 promoter, we published the first version of this, which is a [ 29 ] base paired fragment in 2011. And we and other groups have been using different iterations of this MECP2 promoter. We made it a little bit longer to include an inhibitory element, which when we published in, I think, 2017, we showed that, that had a benefit to make the vector safer than that shorter version. But all in all, this is -- the full-length MECP2 promoter has additional elements that's too big to fit into an AAV vector. So this was a truncated version that appears to work well. It's not going to amend the endogenous expression, endogenous regulation perfectly. But it is, I'd say, a reasonably well-characterized promoter that we've used that other groups have published similar versions of. And I -- personally, I'd rather stick with this MECP2 promoter than try and switch to some other like synapsin promoter or neuron-specific promoter. So that's kind of where we are right now.

Thanks, Steve. Final question comes from Gil Blum of Needham & Company. Do you believe the AveXis program suffered from the complexity of expressing MECP2?

So maybe, Kim, I'll take that question. And it's an interesting question, one that is probably not appropriate for us to comment on, not having kind of full information. Obviously, I was part of the management team at AveXis. And understanding some of the preclinical data and just some of the preclinical data that Steve and his group have published over the years, when they've looked at unregulated self-complementary endogenous promoter full-length MECP2, you do have this issue around overexpression because, again, you can't control transduction. This is not, not having any information on the AveXis program. I can tell you where it is. But I do think when we've looked at similar constructs, and some of the preclinical data was just shared here, particularly around when we dosed wild-type mice, you do see in an unregulated full-length MECP2, we did observe 40% toxicity or a loss of survival in the wild-type mouse model. And so we do know regulation is extremely important in this disease and kind of the Goldilocks nature of guarding against overexpression. But getting expression just right is additionally complex when you have patients that are mosaics and have half of their cells as wild type. So again, not knowing what's happening over there from a developmental point of view, we do feel having this miRARE self-regulatory feedback loop act as somewhat of a safety valve is extremely important in titrating the amount of expression of MECP2. So maybe we'll just kind of stop there.

Great. Thanks, R.A. That wraps up our time on the Rett section, and I'd like to now turn the call back over to R.A. for your closing remarks. R.A.?

Sure. So thanks, Kim. We've really enjoyed sharing with you guys greater detail of our clinical and near-clinical stage programs. Looking ahead, we will continue to focus on rapidly advancing our pipeline with many key milestones from the programs discussed to date over the next 18 months or so. We have made a significant transition into a pivotal-stage gene therapy company with our acquisition and development of TSHA-120 and expect to provide both the clinical and regulatory update on this program by the end of the year. Further, we remain on track to report first-in-human clinical data for TSHA-101 program in GM2 gangliosidosis as well as to initiate our Phase I/II trial in TSHA-118 in CLN1 disease that currently has an open IND. We expect to open INDs or CTAs for our Rett syndrome program, which we just went over and our SURF1-associated Leigh syndrome program that we'll review tomorrow by the end of the year. Lastly, I would like to give special thanks to Dr. Steven Gray for participating in our event today. He's been phenomenal. His lab at UT Southwestern has really supported, from a foundational perspective, the development of the company into where these programs are today. So we really appreciate his and UT Southwestern's strong support and collaboration. We appreciate you taking the time to provide a unique perspective and educating us all on the programs that you've pioneered over the last years. Our goal this morning was to provide with a greater, deeper understanding of our tech -- our goal this morning was to provide kind of the sector with a deeper understanding of our technology and our programs and the exciting potential of our product candidates and the promise they hold for patients. Thank you again for your interest, and we hope to see you again tomorrow for day 2. Have a great day. Thank you.