Scientist Stories: Anastasia Khvorova, RNAi Therapeutics

So thank you so much everyone for joining us today. What I would like to talk about is to give you an update or overview what was happening in the field of RNAi therapeutics in very quick summary in the last 12 to 18 years and how pathology progression is relevant for the demonstrated clinical success. So oligonucleotide therapeutics, or siRNA therapeutics in particular, are informational drugs, meaning that as long as the chemistry of the backbone is optimized, able to robustly treat a particular tissue, to a particular tissue, I'm using telephone. This is what I do. Can you hear back better this way? Okay, fantastic. So as long as the chemical architecture of oligo is modified and it enables robust delivery to tissue of interest, then the same chemical architecture can be reprogrammed to silence genes on demand. So as soon as the chemistry is optimized and delivery is confirmed, non-toxic delivery is confirmed, you easily can apply the same architecture to deliver, to reprogram, to silence any gene you want in the same tissue. And this concept is absolutely fascinating, and the power of RNAi was early recognized. But technology in general followed the typical technology hype curve, where original discovery of RNAi generated a lot of hype. We are going to make drug quickly. We now know the targets. We can just deliver siRNAs to the tissues and everything is going to work. There was a lot of hype, including Craig Meller getting the Nobel Prize, majority of the big pharma entering the field. But unfortunately, early clinical attempts and large number of early clinical attempts did not produce either inefficacy or significant efficacy, which resulted in the whole field crashing somewhere around 2010, big pharma leaving, and nobody wants to even use the word RNAi. And in ophthalmology field, even now people are saying sRNAs do not work. But fortunately, there is a lot of great science was going on both in R&D, both in academia and mainly in industry, Bionilum and other companies, which resulted in further technology development, which is now, now produce the first approved siRNA drug, which is called patisiran, or their proprietary name is Onpatria. What you see on the left here is data from New England Journal of Medicine article reporting Phase 3 clinical trial data for Patisiran in the data is amazing. It's not only that the target is continuously downregulated, it results in robust improvement on several clinical endpoints. And what was amazing about this drug, it's not only slowing down the progression, which was the main aim, but actually seeing some reversal of their phenotypes. As a result of it, this drug was approved this August, and it's a really disease-modifying treatment for TTR amyloidosis. As of now, apetizeron is administered every 3 weeks with IV infusion, and it's partially modified, actually, very minimally modified conventional siRNA, so it has its 19 base pair duplex with traditional deoxy oligodT overhangs, which has total of 11 methyl modifications. So there's no 2' fluoromodifications, there's no phosphatide modifications. And this siRNA is formulated in, really fancy lipid nanoparticle which using a combination of several lipids which really gives you this robust and non-toxic and highly potent delivery to liver. So while the first approved drug is partially modified or minimally modified lipid-formulated siRNA, the belief right now that the clinically most promising technology platform for liver delivery is not LNP formulation. It's a multivalent GalNAc, multivalent GalNAc conjugate to the fully chemically stabilized siRNA. So in this more advanced version of technology now used for liver delivery, the fully chemically stabilized siRNAs, there's not a single ribose left. Conjugated to the multivalent GolNac, which has been specifically recognized by azilglycoside receptors, which are overexpressed on the surface of hepatocytes. And single sub-q injection of this drug gives you up to 6 to 12 months clinical efficacy. And this is once again the data from a clinical trial reported by a medicine company in collaboration with Allnylon. This is unbelievable because we are first time talking about oligonucleotide drugs which duration of effect can be close to a year. So how is this achieved? What really matters? So as I mentioned here, the compound consists from two parts. It's multivalent GalNAc and fully chemically stabilized tesseramase. So is the structure of the GalNAc matter? And there's many different ways how this can be synthesized. And what I show here is this tribal and gulnac, which is being used in Inklisiran, which is a nylon linker. They're using pyrrolidine-based linker. So this is a structure which is used called pyrrolidine. You can engineer compounds which will have very similar, very similar overall geometry where you use the combination of tags and short carbon linkers. And this is the same triple Galnac, but conjugated through the standard TAG linker. Alternatively, Alnylam, Dicerna, and several others have utilized a strategy where you just make a fosfamide, a regular fosfamide, where the monomeric Galnac entity is connected to the linker. And in this particular case, we just connected Galnac through the C7 linker, so it's as simple as it gets. And in this case, using this monomeric phosphamide dite, you can attach as many GalNACs to the single siRNA molecule as you want. So what we did, we decided to compare what is going to be the impact of different architectures on in vivo efficacy. And if you look in vivo in liver at 10 micrograms and 2.5 mcg/kg dose, they— this cluster-based GalNAC look very similar. And when you have multiple monomeric GalNACs, you need to have at least 4. 3 didn't work very well. And while the overall, the trend of potency is less, it's not statistically significant, but it still shows you very nice and potent modulation of gene expression in liver. So there's many different ways how the GalNAC architecture can be assemble to give you potent lipid delivery. So this concept of conjugating, putting fully chemically stabilized siRNAs connecting by the linker to the multivalent GalNAc is now utilized by many different companies. So the names of the products is different. I think it's called TriM by Arrowhead or Galaxy by Dicerna and Silence Therapeutics and Several others have their own variants of that, but conceptually it's the same thing. It's a fully chemically stabilized siRNA scaffold connected by the linker to the multivalent Golgi neck. And with that, there is now more than 40 drugs in a pipeline, and I think this number 40 drugs is old, it's a couple of months old, it's probably more like 45, 50 by this point. So this is the first time in the history of siRNA therapeutics where we see this realization of the concept that when the chemical architecture which gives you robust delivery to liver is optimized, you can reprogram it to silence multiple genes on demand. You can reprogram it to silence multiple genes on demand. So we talk about this GalNAc structure, yeah, how about this siRNA? What really matters, and what matters is has to be fully chemically stabilized. So if you look at the evolution of the siRNA chemical scaffolds, we went from non-modified siRNAs or very minimally modified siRNAs early in the clinic, then to partially modified siRNAs, and what I'm showing here is an antigen pattern, but it's conceptually similar to what is being used in patisiran, to then fully chemically stabilized their size. And this fully chemically stabilized siRNAs, this is what is absolutely essential for conjugate-mediated delivery. So there are, you know, there's so much interesting and beautiful conjugates developed over the years in the field of antisense. And you can modify the backbone. You can modify the 2' hydroxyl of the ribose. You can also modify the 5' phosphate. And all those different backbone sugar and phosphate modifications theoretically can be used for modifications of siRNA. There's one problem, and the problem is that if you— there's different classes of oligonucleotide therapeutics, and we start with the simplest one, which is steric blocker antisense. And now there is an approved drug called nusimersen from this class. And the job of steric blocker antisense is just to bind and sterically block certain position on mRNA. This is how the splicing, alteration of splicing is being induced. So the number of, within a cell, interactions of this class of compound is much lower than in a context of RNAi, where it has to be nuclease resistant, it has to get to the cytoplasm, it has to get to the right component and load into WISC complex, and most importantly, maintain compatibility with the Abitur assembly. As a result of it, vast majority of the chemical modifications and specifically backbone modifications backbone modifications which are working beautifully for steric blockers, they're not acceptable for RNAi because they interfere with the ability of those molecules to be recognized by RNA-induced silencing machinery and silence the genes. As of today, there's only 3 modifications which is being used in siRNA in clinic. And two of them is modifying the 2' hydroxyl of the ribose, which is 2' fluoromodification and 2' methyl modification. And another one is phosphatidate, which is modification of non-bridging oxygen of phosphodiester backbone. There is also new compounds which are just entering the clinic, which on top of it using different variants of the 5'-phosphate modification, which I will be talking about later on. So why, uh, patisiran is approved drug, and it has only 25% of bases modified. So it's only 25% modified, and it works beautifully in clinic. But patisiran is delivered through formulation. And if you're attempting to deliver oligonucleotides, deliver it to oligonucleotides, we are— by conjugation, the oligonucleotide content has to be fully chemically stabilized. So what I'm showing here is two very similar molecules. The molecule on top has 60% of bases modified, so there is a lot of you have this white circle still, which is non-modified RNA. And then the structure at the bottom is fully chemically stabilized, so there is not a single ribose left. So if you can engineer those chemical scaffolds where there is not a single ribose left, but the ability to enter into WISC complex, it's as good as non-modified siRNA or even better. And in terms of WISC cleavage, you completely got rid of the ribose, but you still maintain the same rate of the loaded RISC cleavage. So you can fully chemically stabilize siRNA without any interference with its RISC ability to modify, to interact with the RISC complex. But when you inject those partially modified and fully modified siRNAs systemically, you can see that full chemical stabilization results in dramatic enhancement enhancement of overall level of tissue retention and delivery. And this enhancement in tissue retention and delivery is really not affected by the nature of the conjugate, and this is cholesterol, DHA, or the GULNAC, or the route of administration. This is IV delivery and this is subcu delivery. So in all cases, partially modified SIs gives you almost no systemic exposure, while a fully chemically stabilized siRNA enhances systemic exposure orders of magnitude. So what about, what about phosphate? This 5' phosphate of the guide strand is necessary for loading of the siRNAs into the RISC complex.. And when non-modified siRNAs are introduced into the cells, they're quickly being phosphorylated, and thus a fraction of them can enter into WISC complex. When you fully chemically stabilize a backbone, this type of molecules are poor substrates for intracellular kinases, and as a result of it, its ability to enter into WISC complex is reduced. If you chemically phosphorylate it, this chemical phosphate is cleaved off within no time within a tissue. So this is why it was assumed that chemical stabilization of this 5' phosphate might be beneficial for activity. And there are a lot of ways how this can be done chemically. So you can take this bridging oxygen and substitute it with a carbon, introducing a non-inducing stereo fixation. You can just put the phosphatidyl group Triad on the 5' end, which gives you almost the same level of stabilization. And this concept was first described by, um, Ionis in a context of a single-stranded siRNA, and then it was adapted later to double-stranded siRNAs as well. And what is important is that if you take fully chemically stabilized siRNAs and on top of it the stabilization of the 5' phosphate, you further enhance tissue retention and you further enhance duration of effect, which can be specifically important for delivery for extrahepatic tissues. Okay, so we talk about the gul neck structure, we talk about 5' phosphate stabilization. How about the is the modification pattern itself. Does it matter what combination or in what order you modify 2' fluros and you put 2' fluros and 2' methyl modifications? And Alnilam has published a beautiful paper last year, I highly recommend for people to read it in Molecular Therapy, where they screen a large number of chemical modification patterns in a context of one sequence. So this is their original chemical modification pattern they use, where the black is O-methyl and the green is 2' fluorine. And you can visually see that this compound has approximately 50/50 2' fluorine, 2' O-methyl modifications. And then they systematically substitute 2' fluorine with 2' O-methyls and came up with another configuration where 80% of modifications are 2' O-methyl, and there is still around only 20% of modifications to prime flora. And while the potency of this compound in the V-scan 3 is slightly affected, but it's still a pretty efficient oligo. But what they have seen that when they look at the efficacy in vivo in mouse on the left or in non-human primates on the right, the second compound which has 80%, only 20% of fluoromodification left, is significantly significantly more potent. It gives you higher level of silencing, and the silencing lasts longer. So this reduction in 2'-fluor content and increase in 2'-omethyl content really does enhance both potency and duration of effect of oligonucleotide drugs. It's most likely due to the fact that chemically, 2'-omethyl is much more stable than 2'-fluor. So when you're reducing the 2'-fluor content, and you're also enhancing overall stability of the compound. So another interesting flavor to this full chemical stabilization is how would you design this type of compounds. And what was interesting, we all know that for many, many, for more than a decade, that the siRNA activity can be predicted bioinformatically with a very high degree of confidence. And there is multiple publicly available algorithms which are defining or help you to predict efficacy of non-modified tetrarenies. Majority of them is really dominated by several predictive factors, and one of them is terminamic bias. It's just ensuring that the guide strand preferentially enter into this complex. In most of the advanced chemical configurations, the asymmetry is defined chemically. It's just the sense strand is modified in the way that it cannot physically enter into this complex. And what was interesting to see is that if you look at the algorithms which have been developed for predicting of non-modified siRNAs and see whether they can predict efficacy of fully chemically stabilized siRNAs, they fail miserably. Have no predictive value. So effectively as random. All through that, if you do large number of screens in large number of fully chemically modified compounds, you can identify algorithm criteria which does give you very, very good predictive power. And what is also very interesting to see is that those predictive power of those algorithms is really affected by the chemical modification pattern used. I don't have time to talk into detail, talk about it in detail, but the take-home lesson is that if you want to develop fully chemically stabilized siRNAs, the screen needs to be done in a context of fully chemically stabilized siRNAs, and bioinformatically non-modified siRNA-based algorithms has very little predictive power for predicting efficacy of fully modified scaffolds. Okay, so this is all positive news and there is a lot of great data published, most of it coming from Alnylam, and overall standard non-clinical safety looks pretty good. The therapeutic index is order of magnitude or maybe even more for advanced GalNAc-modified siRNAs. So what are the potential outstanding issues, which I wouldn't even call it an issue, it's more a question, or what we need to better understand which might be causing long-term some safety concerns. So one of them is off-target activity, and we all We all do know that siRNAs can target the silenced target genes, but also partial homology to microRNAs, partial homology to the siRNA seed and its interaction with many different targets in 3' UTR can result in downregulation of many other targets on top of a target. And there's a nice paper from Janice et al. from Alnylam, which really reports that first, that indeed at very high concentrations, this off-target activity can cause toxicity. And second is that by modifying the sequence of the seed as well as the chemical modification in the seed, this off-target activity can be manipulated and off-target activity-based toxicity can be reduced. So this is just a consideration which needs to be taken into account, but it's— at this point, it's believed that this can be managed. There are a lot of discussions about the potential that siRNAs can bind to immune cells in blood or some other blood cells, and those complexes can potentially induce adaptive immune response. This has been seen as a very rare event with antisense compounds., and it so far have not been reported for siRNAs. But when GalNAc siRNAs have minimal affinity to immune cells, but when different type of conjugates are developed, it's possible that this is something which need to be watched for. There is a concern or discussion that now when compounds are fully chemically stabilized and we are talking about extensive period of time when siRNAs can continuously load to the WISC complex by creating this intracellular depour of the stable siRNAs, and when we observe up to 12 months efficacy with a single injection, there is a concern that if you dose too much, this can potentially overload long-term risk complex or risk machinery and interfere with microRNA homeostasis. So far, this is poorly a theoretical consideration. This has been demonstrated very early on by Marc Kay and then confirmed by many others for constantly overexpressed shRNA from AAV vectors, but so far this have not been been demonstrated for chemically synthesized siRNAs. The assumption right now is that even with a relatively high dose, less than 5% of total available RISC complex is loaded, and this should be fine. There's also ongoing discussion about potential for high, very high doses of 2'-fluoro and continuous exposure exposure to very high doses of 2'-fluorobene toxic. And we will talk, which I will talk a little bit more later on. And the last one is ionists have seen antisense compounds accumulating in kidney and certain neurons forming inclusions, and now the similar type of inclusions has been demonstrated in liver by the GalNAc sRNAs. So far they seem to be benign, so they're not causing toxicity, but this is something which we need to watch for. So to the best of my knowledge, this is a list of, I wouldn't say concerns, it's just the questions the field is continuing to watch for in terms of potential for some unexpected issues coming late on, but overall right now clinically the compounds looked very safe. Okay, so let's talk about this Revusiran versus Inclisiran controversy. So Revusiran was a Phase 3 fully chemically stabilized siRNA conjugated to the GalNAc, which Alnylam was evaluating in a Phase 3 clinical trial. And 2 or 3 years ago, this trial had to be stopped due to disbalance in mortality between drug-treated patients and placebo. And this disbalance in mortality was real, yeah. And unfortunately, in spite of a lot of investigations, at this point the etiology behind this mortality, observed mortality, is unknown. And we do not know whether it was specific to the sequence of the compound used, or it was specific to the chemistry of the compound used, or it's something which is specific for a type of patients. Enrolled in this trial. But this was an early version of the technology, and what you see here is a chemical modification pattern used in Revusiran. And while all the riboses are modified, it's lacked any 5' backbone stabilization. And in a context of this lack of 5' backbone stabilization and 50% of 2' fluoromodification, the doses at which these compounds need to be administered were very high. So it was administered at— after the loading dose at 200 milligrams per week for the total dose of 25 grams per year. So— and this is what was toxic. Currently, the compounds like inclisiran, the more advanced compounds, they first have the reduction in 2'-fluoro content. So we went from 50% to 25%. And this increased reduction in 2' fluorine and increase in 2' methyl gives you better stabilization. On top of it, Alnylam included paired phosphatidyl modification on the 5' end of the guide and passenger strand, which gives you significant protection from XRN1 and other 5' nucleases, and the combination of those factors resulted in the fact that now those compounds can be administered every 6 months, or there is even a discussion that maybe it's every 9 months, or maybe even once a year, and overall tissue exposure or patient exposure is less than a gram. So we went from 25 grams a year to 1 gram a year, and from 50% 2'-fluoride to 25% 2'-fluoride. So in all the current current drugs, the overall tissue exposure, patient exposure to the chemically modified nucleotides is going to be dramatically lower. It's between 25 to 100-fold. And I think Alnylam position is, which is reasonably justified, that now when the doses of those oligos which is being used clinically is so much lower, the chances of seeing this type of toxicity again is is low. One can argue that, you know, you have seen it in 2 years with very high doses, and it might take 10 years to see it with new modification patterns, advanced modification patterns, but the remains to be seen. There are 2 publications from Ionis which report that single-strand single-stranded phosphatidated oligonucleotides, which are enriched in 2'-fluor, can be toxic in vitro and in vivo. And while these papers are specific only to single-stranded oligonucleotides, this has not been demonstrated with double-stranded siRNAs, and it seems to be that single-natured Nature of those compounds really matters, but it introduces concept that potentially 2'-fluorine can be toxic. So, um, I think it's enough. It's quite possible that more advanced chemical configuration in the long run will allow us to get rid of 2'-fluorine modifications completely, but at this point, the new advanced chemical modification patterns used in the clinic using reduced 2'-fluoroc content and clinically looks pretty safe. So what is next? I will say liver delivery looks super promising. We are talking about 9 to 12 months duration of effect with single injection and a wide therapeutic index. How about extrahepatic delivery? And there are several things in works. There are very interesting data from AstraZeneca-Ionis collaboration on using the GLP-1 variant of the glucagon-like peptide conjugates to antisense for delivery to pancreatic beta cells. I think it's a promising approach. It remains— the clinical utility of it remains to be seen, but preliminary data looks beautiful. And if you attend the next OTS meeting, I think there will be more data presented on this project. There are new type of small protein scaffold called sinterins, which is being developed by ARA Biotherapeutics, which is smaller version. It's not an antibody, but it's selectable scaffold which potentially give you an ability to select very high affinity ligands. Different surface chemistries, and this approach is being advanced towards clinical development by iroBio Therapeutics. There is ongoing work with using antibodies as delivery agents, and it's— AVDT and NanoMedicine is working on it, and they have some really interesting novel data and specifically talking on impact of the linker and chemical modification patterns. And of course, there is long story of using aptamers for, um, aptamers for targeted delivery. And there's more work need to be done, but there is several very interesting publications indicating that it might be foundationally an interesting approach. So alternative to true targeted delivery is chemical engineering of lipophilic conjugates, which gives you not a targeted delivery, but a functional delivery in delivery to CNS. So just very briefly, one of the things we have done, we have engineered a large number of different hydrophobic conjugates, and we demonstrated that changing the nature of the conjugate really changes hydrophobicity profile. So they are More hydrophobic conjugate like alpha-tocopherol will be eluted very late in reverse-phase columns, while certain things like phosphodiolcholine actually makes it less hydrophobic or more charged than the non-conjugated siRNA. So when you, when you attach those compounds or those conjugates to the siRNA,, we were interested to see how this is going to impact its ability to deliver things extrahepatically. And we looked at distribution, cystine tissue distribution at day 2, and we also looked at the efficacy at day 7. And what we have seen is that in general there is a clear trend where the nature of conjugation conjugate gives you a change in kidney versus liver distribution. So when you start with unconjugated compound, it's over here, so it's unconjugated, it does not deliver to liver almost at all. Very little liver retention. And majority of it is being cleared from the blood bloodstream. So more than 90% is actually being peed out on a first passage through kidney, but small fraction of it is being retained. A small fraction of it is retained and can give you delivery to the tubulin cells. If you conjugate it to things which are slightly more hydrophobic, you can completely shift this ratio and get compounds which deliver deliver it to liver very well, but do not deliver it to kidney almost at all. And simultaneously, when you engineer conjugates, you can see things which gives you really super high level of kidney delivery. And actually, some of those conjugates start to deliver not only to the tubule structures, but to middle as well. What I wanted to just make one point over here is that the nature of the conjugate not only defines what tissues it goes to, but it also defines what subset of cells it goes to. And what I'm showing here, it's a very related example. So it's EPA and phosphatidylcholine DHA— phosphatidylcholine DHA. Dihylin EPA amide. So it has this extra modality. So it's a very chemically very similar structure, but you can see that EPA gives you a nice delivery to the glomerulus of the kidney, while when you put an extra phosphodilhylin modality, you completely lose it. So the point which I'm trying to make is that engineering the conjugates not only affects overall tissue distribution profile, but it gives you its effect what cell type types within a tissue the compound goes to. So last but not least, from the screen of 15 different conjugates, we were able to identify ones which gives you really nice functional silencing in heart. So non-targeting control and PBS doesn't work, and you get a very nice knockdown, reproducible knockdown in the heart. You get compounds which gives you nice delivery to the lung and silencing in the lung., and you can get functional knockdown in muscle and functional knockdown in fat. But the beauty of this type of platform is, of course, not only to be able to deliver to certain cell types extrahepatically, but this type of conjugate, some of them, gives you beautiful distribution in CNS, where the single injection can get complete elimination of the protein expression throughout the brain, which can last for several months. So what is the overall outlook for 2018? So we have the first drug approved, which is due to the heroic efforts of Alnylam, hundreds of people of Alnylam over the last decade and a half diligently working on a technology, both from a technology platform development as well as clinical development, and this is disease-modifying treatment for really devastating genetically defined disorder. And I think it's a huge event for Oligonucleotide Therapeutic Society in general and RNAi subset of Oligonucleotide Therapeutics and The second one that with the GalNA conjugates to fully chemically modified siRNAs, we now have a promise of up to 12 months exposure, tissue exposure and clinical efficacy following single injection. And there are more than 40 compounds in clinic using this conceptual, conceptual approach by many different companies, once again led by Alnylam. I do believe that the next issue for siRNA is going to be CNS, which is so far is a unique domain for antisense, and I think we will be talking a lot about it at next OTS meeting in Seattle, and I hope some of you will come and join us. There are some ongoing safety considerations, but none of them are serious at this point, and this is at least what we have discussed today. It seems to be that most of them are manageable, and there are no clinical signs of toxicity. I'm just listing the things which need to be understood and watched very, very carefully. Carefully. We had a very successful symposium here at RNA Therapeutic Institute in UMass Medical School beginning of the summer, and as part of the symposium, there was a panel which discussed what really took the field to get towards approval. And I I think what was uniform opinion is that you need to have two things. You need to have a working delivery, but on top of working delivery, you really need to have an intimate understanding of intracellular pathway interactions because those informational drugs, specifically RNAi, work with different pathways within a body and within a cell, and understanding how they interact act really is essential for ensuring productive and non-toxic silencing and developing additional disease-modifying treatments based on this Nobel Prize-winning discovery.

And I'm happy to answer questions. Yes, if anyone has questions, can you please type them into the chat box. Chat box and then Anastasia can read them. Does anyone else have any questions yet? Anastasia, let me just say that while everyone's typing the questions, it was a very informative webinar.

Thank you very much for doing this. It's my pleasure. Could you comment on the tested conjugation chemistry? I, um, could you comment on the tested conjugation chemistry? I'm not exactly sure, um, we're talking about the architecture of the GalNAc or the different type of conjugates. Maybe if you can retype your question and I will go to the next one. Could you comment on IVDT-RC in DAR-1 for efficacy? I think it's beautiful. It's a great— I have seen their recent publication, recent public presentation, and it's very interesting data. They're clearly getting some functional delivery to the tumor, which is huge. It's a very small percentage of injected dose which actually goes to the tumor, but as long as it's gives you preferential delivery and tox profile is reasonable, why not? And anything which is at this point extrahepatic is worth pursuing. Since GalNAc is specific to liver hepatocyte cells, what type of delivery will be required for CNS? GalNAcs do not work in CNS, so you will— the compound The compounds which do work in CNS require capillarative uptake to the neurons and glial cells, and it's a completely different chemistry. And we actually will be talking— there will be certain presentations at OTS this year, so if you're interested in CNS delivery, just attend the OTS 2018. How much is RNA delivered to spleen in general if IV injected? Injected. So the percentage of spleen delivery, and I think I might have a slide here. So what you see over here is liver, kidney, spleen, placenta. So if you look at the cholesterol conjugate, which is a very hydrophobic conjugate, you get around 15 3 nanograms per milligram. And so you only get 3, 4 times less than the liver. Spleen is small, so percent of injected dose is going to be a couple of percent, but you will get a very robust delivery. If you talk about the gallbladder, it's much less, yeah? So we are talking in the spleen is going to be around 2 nanograms per milligram, and it's more than 100-fold less than in a liver. So the percent percentage of spleen delivery is really affected by the nature of conjugate. How did you connect ligand and RNAi? So the nature of the linker really matters. So as I was discussing earlier on, And this is just an example for the GalNAc, but this is a uniform approach which can be used for conjugation compounds. There's two ways. One, you engineer the support, yeah? So you engineer the support where you have a trivalent, trifunctional linker, and you build your compound on a support. Support. So we usually— this is our preferred way to conjugate compounds to siRNAs because specifically when you're working with complex structures or slightly, even slightly hydrophobic structures, doing it post-synthetically is really, really hard because your yields are low and you are never sure that you get full conjugation. Attaching it to support ensures that 100% of your oligo will have the conjugate. Alternatively, what one can do, if chemistry allows you and the nature of the conjugate you use allows you, you can build a phosphamide and apply it during the buildup process. This usually has limited utility because in this case all the protection groups should bear orthogonal to solid-based synthesis of the oligo, which might be complex. And the third way, which is my unfavorite way to go, is to do it post-synthetically. And it's— there's many different published conjugation chemistries which can be used for that, but this is a less reliable way to do it. Do PK effects seen with different conjugation chemistries correlate with PD? This is a beautiful question. Thank you, Eric. Actually, no. So, and it was a surprising effect for us to see. So generally, yes, if you do not deliver to tissue, you are not going to see silencing. But the minimal concentration of oligo accumulated in the tissue, which is necessary for silencing of gene silencing is very different. So for example, for kidneys you need to have minimally 100, 150 micrograms per gram to see the silencing efficacy, while in muscle or in fat, 1 microgram per gram is fully efficacious. And I think it tells you that in filtering tissues like kidney, much larger fraction of the oligo is being trapped nonproductively while in tissues like muscle, neurons, and brain, muscle, fat, and lung to some extent, very low accumulation of the oligo can be productive. The second interesting effect we have seen is that you can have several conjugates which will deliver, for example, to kidney equally well. You achieve very high concentration, 150 micrograms per gram, which is huge. It's like 100x more than what you need to get silencing in liver. And some of them will be fully functional where you get 70-80% of target knockdown, and some of them would be completely inactive, meaning that the nature of the conjugate does impact the internalization process or fraction of the compound trapped nonproductively or potentially rate of chromosomal escape. So while the fact is there that there's no perfect correlation between tissue accumulation and silencing, mechanistically I think it needs to be further understood. Okay. Javanka, if you were to combine phosphodiylcholine with GalNAc modification, would tissue distribution to liver and kidney be roughly equal. Actually, we have done not exactly that, but we played with the gallbladder quite a bit. So if you put extra hydrophobic modifications on the gallbladder, you do enhance extrahepatic delivery and you start to enhance kidney delivery. You also enhance liver delivery, like the percentage of injected dose in liver will be higher, but you're losing activity. So with GalNAc right now, I will say modifying conjugating GalNAc to non-asymmetric, like fully duplex si's, gives you reasonable delivery. Like percentage of injected dose retained in liver is going to be relatively low, so you only will retain couple of percent of injected dose, but they will get in hepatocytes. Hepatocytes and hepatocytes only, there will be very little kidney exposure and you get true targeted knockdown. When you start to engineer GalNAc further, you can dramatically enhance liver retention, but you're losing specificity. And as you suggested, you will get significant percent of injected dose in the kidney. And I'm not sure whether this enhancement in liver delivery delivery, it's really productive because I think most of it will be trapped nonproductively. What is the duration of effect of siRNA in CNS? So the data which we have on hand, it really depends on your initial delivery to the region of the brain, but in some fractions of the brain, like hippocampus and thalamus, we still can silence target gene like Huntingtin below level of detection at 6 months post-injection. In some targets like cortex, it starts to be more variable, where at 4 months it's 80-90% silencing, but at 6 months it starts to disappear. Is there any decrease in efficacy with subsequent doses related to inactivation by innate adaptive hormone response. So I have no experience with that, but my understanding is that Alnylam have presented the data and they have done it extensively where they demonstrated that you can dose time over time over time and there is really no reduction or changes in either clearance kinetics or level of liver retention or duration of effect. So I will say the overall answer to your question is that with the chemistries and formulations which have been studied so far, there is really no indication of inactivation by innate adaptive immune system, and then they can be administered repetitively. With that said, with new chemically engineered conjugates, I think there's quite question will need to be answered, um, answered separately. And I think the last question, how hydrophobicity could change delivery? I think it could be related to ASO aggregation and self-assembled structure. What is your point of view? We actually do know what the mechanism is. We have a paper under review in Nucleic Acid Research which Hopefully the reviews were very favorable, so we hopefully will get it out soon. I think mainly what's happening by changing the hydrophobicity profile, you're changing serum binding profiles. So it's binding to the different proteins in the serum, which in part define distribution to different tissues.

Okay, I think we are out of time. Yes, thank you very much, Anastasia. If there are no more questions, I would just like to remind people that the next webinar is on the 9th of November— sorry, 11th of November on microRNA. Hopefully we might manage to get another webinar in, so keep watching the space on the OTS website. And thank you very much, Anastasia. I know you have to hurry up now.

That was a great presentation. My pleasure. Petra, thank you so much for organizing it. I really appreciate your help. Just a reminder, everybody come to OGS 2018.

Yes, and there will be a recording of this webinar made available in the next few days.

Thank you very much.

Thank you.

Bye-bye. Bye.