We explore Argonaute proteins' binding in human nuclear RNA and its effects on alternative splicing. Our study reveals a novel RNAi mechanism controlling gene expression through splicing, highlighting miRNAs' importance in physiology and disease. Leveraging a state-of-the-art bioinformatics platform, we show how democratizing bioinformatics with reproducible computation, interpretability, and user-friendliness accelerates discovery.
Speaker
David Corey, PhD
Professor, Pharmacology & Biochemistry
UT Southwestern Medical Center
David Corey is a Professor at UT Southwestern Medical Center. He received his undergraduate training in chemistry at Harvard University and his PhD in chemistry at the University of California at Berkeley. He did postdoctoral work at the University of California at San Francisco before joining UT Southwestern in 1992. Dr. Corey's area of interest is modulation of gene expression by synthetic nucleic acids.
Webinar Transcript
David Corey:
Okay, thank you very much. And thank you all for spending your lunchtime with me hearing about the science that my lab is doing in collaboration with Almaden Genomics. Okay. So first, we're going to go through a little bit of introduction about big data and the challenges that it faces. So, I don't think I really need to emphasize too much in this audience, the enormous power of the acquisition of large datasets, whether it be RNA sequencing, or mass spec, or almost everything else, and how much of an impact that that's having on how people like me experimentalist go about approaching our day-to-day projects. You know, from my perspective, these techniques are providing amounts of data that we could not have imagined 10 or 20 years ago, but a bit landing in the hands of experimenters who are much more comfortable designing benchtop experiments, and who have little or no ability to write code or deal with complex programs, or even analyze these large datasets. So, we have a sort of paradox where the scientific opportunities provided by these powerful techniques are enormous. But there's also enormous dangers for wasted effort, misleading interpretations and irreproducibility. So, what's a way of thinking about that, that what lab researchers right now go, and we do experiments, we submit the results for analysis, usually, to some kind of core facility, that it all falls into this black box, and then we get some kind of spreadsheets or something back, and then we do something to somehow analyze this data and get our results. And I think this is the problem, right here is that we really don't understand what's going on. Okay, so specifically, what are these? What are these challenges for a person like me? Well, one thing is that the datasets are very large, they're complex, and then overlay that, that the biological data can be messy, that becomes very difficult to work with.
There's also a problem with documentation. We don't have any clear step by step guides that can aid in understanding what we're doing or repeating previous studies. There's method uncertainty. steps can be vague or missing, that make it hard to repeat analysis. So that applies to the data that we get in my lab, we don't exactly know how it was generated, or what went into it. And it's even worse when we look at published data. We don't really have any clue about what on what went on there. Okay, finally, you can have different software versions that can lead to different results, and that complicates studying repetition. Okay, so how do biologists address these challenges? Well, I think there's a certain subset of biologists that just ignore them, or maybe I should say, are completely ignorant of them. They just go on, they take what they have at face value, and they cherry pick their data, they publish their paper, and that's success for them. It's also possible to collaborate with a computational biologist. And in theory, this is ideal. The problem with this is that the computational biologist who's highly talented and whose skills are sought after could make your project their 10th priority, not put a lot of effort into thinking about it. And that leads to time, and it means that your data doesn't get as much attention as you would want. Finally, you can always analyze the data yourself. But that means growing the expertise in your own lab, and that's a pretty tall order to expect an experimental biologist to end up training as a computational biologist. I think that's probably ideal, someone who's expert at both of those, but I don't think there's too many of those people walking around. Okay, so what are these practical challenges? Well, specifically, the biologists lack experience with coding and statistics. managing data is messy. We always have a big signal to noise problem, for example. staying up to date with the latest bioinformatics tools can be challenging or perhaps even impossible if you're an experimentalist. And then there's these collaboration hurdles. So, if we want to talk to a computational biologist, that person will know the computation really well. They will have skills that I cannot even imagine, and they may not actually understand what the biological processes are, or even have the vocabulary to talk to us about the vital biological processes without difficulty. And then finally, we can be limited by the availability of high-performance computing. Okay, so what are the consequences? So, you know, to give you my background, I've been a professor at the University of Texas Southwestern Medical Center for 30 years, I'm present to the oligonucleotide therapeutics, society, a society that contributed most of the science behind the COVID vaccine. And it's producing several approved drugs every year. I'm also an editor for nucleic acid research, I handle about 300 manuscripts a year. And I've been doing that for about the last 11 or 12 years. And one of the things that we see is a lot of misuse of large datasets, a lot of cherry picking, which, with the number of manuscripts that I've been able to see over the years, I can pretty much editorially reject, out of hand. But which actually, the more cherry picking there is, the more the tendency is that they actually go to high profile journals. And it's created a great deal of distrust within the experimental biology community of the kinds of things that you guys do. Because of the misuse by experimentalists of the tools of computational biology, so that's a real problem that I believe needs to be fixed.
So, kind of falls into this. And one of the consequences is don't try this at home, that these tools are put in the hands of people who either don't understand them, or willfully cherry pick the data from them to suit their hypotheses. And that leads to misuse of these powerful computational approaches. The other one is just trust me that they will produce data, they will produce nice pie charts or heat maps or volcano plots, but they don't really explain how they got there. And if you're not explaining how, you got there, and if you're not doing some experimental validation, it really is, just trust me, I'm telling you the truth. But I think by about 2023, I think we've all learned that that isn't enough. And finally, that all leads to a lack of reproducibility. You know, reproducing published experiment is a thankless task, and it becomes impossible when protocols are garbled and computational probe. Pipelines are not available or poorly defined. I mean, simple test, you're doing the student journal club, if that journal club, with students who are extremely bright. If they cannot delve into the data and critically assess it, then there's probably something wrong with the way the data and is being explained and presented. Okay, so now, we're going to back up a little bit and tell you specifically what I do, and then how we've been working with Almaden Genomics to solve some of the problems that that we face. So, this is going to be about RNAi, which I think even after over 20 years since the discovery of RNAi in mammals is still an unmet opportunity. Okay, so there are two general ways in which we approach gene therapy, gene silencing, if we want to silence a therapeutic gene to make a drug. The first is to use antisense oligonucleotides to bind a messenger RNA that can change splicing or inhibit gene expression. The second approach would be to take a double stranded RNA, it binds to Argonaut protein. A guide strand, then forms an Argonaut guide strand complex, and that in theory can recognize just about any RNA sequence in the genome. So, you can tell between these two approaches, we have the ability to pretty much bind to any RNA and then possibly affect the course of almost any, any disease. So tremendous potential there. And it's certainly, you know, as I said, coming to the point where success in the clinic is routine, it's just a question of how efficiently we can go through the drug discovery process, and how we can imagine what the disease targets are going to going to be. Okay, so that's led to a large number. Let's see, I guess 15 and counting, approved oligonucleotide and double stranded RNA drugs. And this is a remarkable achievement from five years ago when I believe there was only one or two.
Okay, but now let's flip it. Okay, so those are synthetic RNAs or antisense oligonucleotides to target messenger RNA. What about natural micro RNAs. So micro RNAs are synthetic or naturally expressed synthetic RNAs that bind to RNA and regulate gene expression. So, the idea is that this is a natural gene silencing mechanism. Argonaut protects the RNA and facilitates recognition potential to control almost any gene. And therefore, if we could learn how to bind synthetic molecules to this natural micro-RNA, we'd be able to interfere with this process. And again, have powerful tools to control gene expression. So, this idea of using synthetic oligonucleotides to bind a microRNAs control gene expression, that's nothing new. People have been looking at that since the 2000s. And a few 100 million dollars has been invested in companies that tried to make drugs that would target micron as well, how have those fared? Well, that's the number of approved therapeutics to target micro RNAs. There haven't been any yet. So, you know, we have to ask ourselves, these micro RNAs supposed to be incredibly powerful. Why have we not been able to get therapeutics for these when we've been so successful with synthetic double stranded RNAs to target gene silencing? Okay, so, clinical success demonstrates the power of RNAi. In vivo, all you have to do is go across the river to Kendall Square visit on the island. And you'll see how powerful this approach approaches. We know that Micronase have demonstrated roles in plants and animals that's undoubted. And what's more, if you were just looking at PubMed, you would find that there are 1000s of papers, 1000s of them with the keyword micro-RNA in them every year. So, you would think that this is a well-known field, where everything is settled, and we should be able to go straight to clinical targets and have success. Okay, and this just makes that point.
This is the number of papers that have been published that cite micro RNAs per year, for example, 6700, site cancer, and micro-RNA. So, if you were just look at this number of papers, you would think that this field is, is well settled. And there's plenty of information for a productive Drug Discovery Program, the kind of thing that is going to be very profitable for a lot of companies. All right, unfortunately, first impressions, deceive. In this case, when you look at these papers closely, what you find is that most of them lack proper controls, I would say almost all of them, almost all of them lack insight into mechanism. And what's more, there's an insufficient appreciation, that micro-RNA function is complex. Now, as someone who appreciates complexity, I would say it is beautifully complex. This is something that can fill an entire career for the next 30 years figuring out how this works. But if you do not appreciate that complexity, appreciate that it is their drug discovery program is not going to advance the complexity must be treated with respect. Okay, so let me just give you a little bit of a taste of this complexity. A lot of investigators in the micro-RNA field proceed from the assumption that if you can calculate a small match of a micro-RNA to a gene, that micro–RNA is controlling the gene, something like that. However, that almost certainly is not what's happening. Probably what's happening is that you have multiple micro RNAs binding next to one another on the gene, forming cooperative interactions that produce an effect that is strong enough to actually control the gene expression. But what's worse is that these are not the same micro RNAs, these are probably different micro RNAs binding beside each other. So, you have to separately make accurate calculations about both. And then what happens is there's another layer of proteins on top of all of that, in case this case, the protein TNR C six, which binds to Argonaut protein and actually promotes cooperative interaction. So not only do you have multiple micro-RNA species that are different micro RNAs binding near one another on an RNA, but then you have an extra level of cooperativity on top of that, making the system even more complex to model. Okay, and then underneath that, we have the problem that this is like typical data, showing which potential micro-RNA Argonaut complexes bind at a gene. And what you can see is even when you restrict the micro RNAs to only the ones that are the most highly expressed in By the cell own, don't even get me started with the fact that almost no studies actually appreciate that more highly expressed Micronase are much more likely to be active than the bulk of micro RNAs, which are expressed at extremely low levels inside the cell, that it's hard to predict which of these micro RNAs is actually going to be the lung. That's going to have a function.
So, I think you see how there's complexity here. And that accurate analysis is going to be incredibly important for figuring out what are going to be the experiments that actually have to be done in the laboratory with the limited amount of time we have available. Okay, so this is what we need to figure out, we need to figure out which micro RNAs modulate a critical function, we'd figure out which genes are modulated? And how does controlling them lead to therapy? And finally, that's all going to lead us to which when does a micro-RNA play a critical role in health and disease? If we can answer all of these questions, then we have a chance realistically of having a drug discovery program that might be able to succeed at targeting a micro-RNA.
Okay, so now I'm going to tell you a specific story in my lab, about one of my special interests is thinking about how RNAi in the nucleus of cells might be able to affect splicing. Alright, so for the last, oh, wow, almost about 1617 years now. My lab has been interested in the topic about whether RNAi, which is dogmatically thought to work in the cytoplasm, might also work in the in the nucleus. Just as an aside, this idea met with a tremendous amount of pushback for many years. But now it's becoming widely accepted that it does happen. And if you think about RNAi, in the nucleus, you can now start thinking about could it affect transcription? Could it affect enhancer elements? Could it affect splicing, a whole new world of gene targets?
So, we published a series of papers over the last three years looking at this issue. And we relied on a variety of cell lines that were knocked out for various RNAi factors so that we could do extremely well controlled studies. And we found that in normally growing mammalian cells, RNA, I do not do that much. And that it's very hard to find really good targets for AI focused experiments. So, what we hypothesized was that maybe these cells need to be pushed by environmental stimuli to go into a state where RNAi becomes more important. So, the idea here is that if you stress the cells, maybe that's where this extra level of RNA regulation on top of normal transcription factors, and everything else, the cells you use, becomes an adaptive response that the cells need. So, we asked the question, very simple question of we use HCT 116 cells, those are the cell line that we've made all of our knockout cells, what happens when they're grown to overconfidence? Very simple type of cell stress. Okay, so this project is done by a brilliant student in the lab on Crystal Johnson. She looked at cells and to normal 2d cell culture in 3d organized current culture and in primary tumors. And so, what HCT 116 cells will do is they will not only grow until they completely fill a cell culture dish, but they will also start growing on top of one another until they stack about 4d. So, I'm going to be telling you about the results of cells that are different days of confluence. So, they might be at 80% Confluence on day three 200% On day five, and maybe 400%. On day seven, that's our simple experimental model. So, crystal did this. And what did she find? She found that we saw a shift of Organon from mainly being a cytoplasmic protein with some in the nucleus to almost being entirely nuclear, in just regular 2d culture. And this is not only true of HCT 116 it's also true of other cell lines where there's no contact inhibition, and they can grow on top of each other.
Okay, then we took a step up towards biological relevance by working with colleagues and Rolf Perkins lab who do 3d organoid culture. So, they did the organoid culture with the HCT 116 cells. And here the effect is even more striking. Everything's in the nucleus now. So, this is exactly the opposite of what the people thought who had been rejecting my papers and my grant applications by saying that Argonaut was always in the cytoplasm. In fact, here's a case where it's entirely in the nucleus. So, let's take another step up in biological relevance. We obtained tumor samples from our tissue core. And this just is what happens in benign tissue and it's mostly in the, in the cytoplasm. In a tissue that was actually, the pathologist told us was premalignant colon, tubular adenoma mainly in the cytoplasm. Okay, but then in five out of the 21 tissues we analyzed, we saw a shift to not only the nuclear fraction, but even when we isolated the chromatin, Argonaut was associated with the chromatin. So, you know, we now have a system that not at all colon cancer tumors, but in a fair number of them RNAi factors like Argonaut had been sucked out of the cytoplasm and moved relocated into the nucleus. So, then the question becomes, you know, how is this changing? What's going on? Okay, so let's just remind you of what we're what we're dealing with here. What you the basics of what you have to understand is that this is a ribonucleoprotein complex that is programmable. You have an RNA domain, which is a micro-RNA. And you'd have Argonaut, which is a protein domain, those bind together to form the ribonucleoprotein.
So, they could bind to a three prime UTR, and a cytoplasm and do the canonical silence gene expression that you find in textbooks. But could it also be possible that they bind near splice sites to effect splicing and could have an effect on splice and be one of these effects that we're seeing in these tumor cells. And this would not be unprecedented. Anyone who's followed the antisense oligonucleotide world knows that the most successful antisense oligonucleotide drugs, the ones that are actually saving people's lives. Right now, helping kids to walk instead of dying when they're three are drugs that affect splicing by binding their splice sites. So, we tried making some of those same drugs as double stranded RNAs. And this just shows one result that was just in a simple model system showing how these double stranded RNAs can affect splicing and do just as well as an antisense oligonucleotide. So, proof of principle we know that a synthetic double stranded RNA delivered by Argonaut can efficiently affect splicing. So, the question we face is this very robust result? Can it also occur with natural micro RNAs? Okay, so what do what do we need as experimentalists? Okay. This is something that is very difficult. And we've learned that even cytoplasmic RNA is the thing that people thought was all settled and well understood, it's very, very difficult to find an actual be able to say that this micro-RNA controls this gene, or this gene is controlled by micro RNAs at all, now, we're going to have to try to find out what's going on with splicing something that's really new. So, we need an efficient workflow to analyze the effect of increasing cell density on gene expression, we need to know how that increase in cell density affects splicing. And it's essential that we prioritize genes accurately before time consuming experimental bench science. Because I want my graduate students to graduate and no more than five years, I do not want them spending eight years in the lab trying to chase down results, we need to be highly efficient. Okay, and that's were engaging with them, but then Almaden Genomics has come in.
So, this is very, very recent. I was in Boston for a conference only a couple months ago on antisense on therapeutics, and that's when a representative Almaden Genomics which is also located in Dallas, met with me to talk about they're their platform and suggest that maybe we wanted to, to work together since we were in the same the same city. So, everything that I'm going to be telling you about has advanced in the last two months and actually primarily only in the last three weeks. Okay, so what does my lab need? If we cannot work, hand in glove with computational biologist all the time, if we're going to do this, and understand what's going on ourselves. There can't be coding we just don't, we don't code. Things should be drag and drop. It everything should be transparent, intuitive and above anything, a pipeline needs to enable independent use by us. Okay, so if you have questions, I'm going to let all those and explain this a little bit more. But what I understand from this is that this is kind of this is an essentially like a Boston subway map, you may not understand every stop on it. But this provides a transparent map for the investigator about how the workflow is put together. And well, I may not be able to understand it very well, because I can barely, you know, push the on button on my computer, this is something my students can understand. And this allows them to know what is going into this black box that ends up processing their data, okay, but there's more is what they really like is the fact that through a just a drag and drop mechanism, they can drag their different datasets over here. And then they can choose different outputs for how they want the data to actually appear. And I mean, this is just marvelous compared to the way we've been doing it. Before it gives them a great deal more power to fully utilize the data sets that we spent so much time and money acquiring. Okay, so basically, here's the experimental setup, we grow cells to different cell densities, 52 5370, they're day three, day five, day seven, we prep the RNA, we get it sequenced. And then we use their genome pipeline to analyze it. Okay, now remember, all of this has been done in the last just two to three weeks. There are various kinds of RNA splicing events. So, first thing they did was to look at various RNA splicing events, and just put the different splicing events that are observed into the various piles. Okay, next, we decided to try to narrow down which of those are going to be significant. And it's reasonable to expect that something is only going to be biologically significant, if it occurs is both a five and a seven.
That has to be consistent. It may miss a few things, but at least it cuts down the database a little bit. So, there you go, we go from a couple 1000 events to about 750 By doing that, okay, that's the first step. Okay, the second step is we have data from a technique called Eclip, I won't get into it in detail. But basically, Eclip is a sequencing technique that allows you to find out where Argonaut binds to RNA. The places where Argonaut binds to RNA are great candidates for where micro RNAs are going to be binding. So, this was really a very productive way of narrowing down your candidate targets. More than that, we only also demand that those have sites for highly expressed micro RNAs. So that's two levels of stringency that we're imposing there. Okay, so now when we demand that those two things are happening, we've gone down to about 250 different events. So that's 250 candidates so far. Okay, then what do we do after that? Well, they've also are omitted, and colleagues have also been able to look at where are those Argonaut binding sites, relative to an intron exon junction, relative to a place where the splicing apparatus might bind, and where you might expect that there'll be a disruption. And what we observe is that most of these are occurring within a couple 100 basis of these intron exon junctions. So again, these are probably better, better candidates. Now what we haven't had time to do now that Almaden's pipeline has brought us down to a couple 100 candidates is what we would do is we would individually curate each one, look at the data and just by eye and figure out is this something reasonable? And that's something we routinely do. My lab always looks at the individual pieces of data.
Basically, the raw data stage, but their pipeline also makes that easy. Okay, so what else do you need to know about it? Comprehensive logging their full records logs for every part of the analysis software versions and parameters that ensures that everything is well documented, the results are traceable, allows researchers to repeat their analysis bind offers a checkable record of the process. There's containerization. Every study is marked with a version keeps track of the software use, that's always a nice thing. They give a stable repeatable setup for the software and analysis. And importantly, all the images in study are free for everyone. So, when we go to publish our results, we're going to be able to put everything out there. Because obviously, we can't get it published if we don't give people the opportunity to reproduce it. Okay, finally, I just like to share what crystal set. So, Mark, who told me, and I were having dinner last night, when working on the final version of this talk, I opened up my computer and there was an email from Crystal, because it asked her what her experience had been using this. And I showed it to Mark without actually knowing what Crystal was going to say. So, crystal could have said, this is garbage. This is a complete waste of time, I guess that would have saved me from having to come here and give the talk. But in fact, this is what she said. And she didn't write this for this, this is straight from what she's, you know, very talented grandson said is very straightforward how to upload and incorporate the data. The splicing pipeline module that they built is transparent and easy to modify. The user interface is clear and colorful, lots of nice drop-down menus to zoom in, modify, and export useful data. So that's the experience of an outstanding graduate student with this pipeline. You know, my view is that this is a direction that we as experimental scientists need to be moving in. And both to increase our productivity to make sure that we're not turning out garbage. But also, to make sure that we get the most out of these datasets that we have spent so much time collecting. And that in the case of for example, data from patient tissue, is extraordinarily precious and difficult to reproduce them that we really need to squeeze every drop of utility out of. Finally, I just like to thank the people in my laboratory, in particular, Crystal Johnson, who's been working with the scientists at Almaden to get something in time for this conference. And I hope you understand that it's remarkable how quickly this has come together for a novel protocol. And in my experience, when protocols come together quickly, that's usually a good sign that they're going to be robust and useful. Finally, thank you all for taking your lunch hour to listen to me. I would be happy to answer any questions. Thank you.
Yeah?
Audience 1:
Have you looked at the response during complements when the sales are going up line elements? At all?
David Corey,
No, we haven't done anything like that. But what I can tell you is that right now, we're working out methods and having some success and being able to isolate chromatin. And those should be going in for sequencing very soon. And needless to say, I'm very excited to know what's going to happen with chromatin associated RNAs. So that if I was going to project out in the next couple of years near less to say, I'm really excited about moving into that area and adapting some of these techniques for chromatin associated RNAs. Yeah.
Helper(Dont know who it is 33:45)
Anybody has any more questions? If you want to go up to the microphone, in the middle there, that would be great.
Audiece 2
That was a great presentation, I had tons of questions, but we can talk about that later. So, when we're doing the experiment with the cell line, that is (not sure what that man said 34:12) Do you guys ever measure this in primary cells get completely covered, transformed and so on? But what I'm saying they normally get quiescent?
David Corey:
I, you know, oh, we haven't done that. I know (Don't know who this is 34:44) and he has been doing experiments like that and has been observing a shift into the nucleus and she thinks that's involved in transposon silencing. And I know she has a paper and review about that. I think experiments of that sort are extremely exciting. You know, we have a small lab. So, we're focused on what we can focus on. But I think that putting these cells into biologically significant states and then observing whether Argonaut is shifted in those states as an important thing for people who are versed in those systems to do.
Audience 2:
Yea I completely agree. And the other thing I was thinking, when you show your 3d spheres, you clearly see that, you know, the shifting, have you guys ever tried to put for example, the sales in vivo? And then try to see the same phenotype.
David Corey:
No, I mean, we haven't tried to do that, though. I mean, it's certainly something that we can do now that we're, we're currently writing this up, and that will close the first chapter. I think that's an obvious thing to do. Not for us to do, but for our collaborators who did the organoid culture to do because they do that all the time. It would be simple. Yeah. Okay. Great. Thank you. Are there any other any other questions? Okay, well, if not, I'm a bit. Oh, Leonard. We couldn't have one without you giving me a question. We have we have to make up for the years that we haven't seen each other.
Audience 3:
For the years of like, not asking questions from the front row to each other. Yes, exactly. It's wonderful seeing you here. And thank you for as always, for the stimulating and provocative talk. You know, this is great. I've never seen any of that. I mean, the, you know, the nuclear (36:40 Dont know what he said) 116. And the in the tumor samples, especially the correlation with the you know, benign versus malignant. This very impressive, I guess, I'm kind of wondering two things. Do you think that nuclear localization of Argonaut, even though it's real and functional, is basically it's a property of wacky over confluence or like stage four metastatic tumor cells that at that point, have such diverse genomes and morphologies that they don't look like a human cell anymore? Everything's fields already took up parameters that they shouldn't be in. And that's how they got new nuclear Arriba? That's basically one question.
David Corey:
I hear. I know it sounds concern, do a one at a time. Yeah, it's an important question. Okay. My speculation is that when cells are growing, normally, they're in cultural or in the body. RNAi doesn't do that. Much. I'm sure it does some important things like buffering expression and stuff, but not really anything outstanding, that we're going to have an easy time finding, I think probably, it's when you push them to more extremes. For example, in cancer, the things start to get regulated differently. So that would Be my guess, for example, in development, you might think that there needs to be a little bit more regulation to make sure everything happens, right. And that's when things might start changing. Also, I shouldn't say that one thing that we've also observed is when they move into the nucleus, you get D-repression of genes in the cytoplasm. So, it's also possible that the effect is nothing important going on in the nucleus. It's that you're reversing cytoplasmic regulation of gene expression. So those are my guesses right now. But I think other investigators have to look in other systems to really answer your question.
Audience 3:
Yeah, yeah, exactly. So like, what are the larger implications of this? Is every component of the RISC complex currently known to be nuclear or you know, aside from agrifoods to it's the still the jury's out on that?
David Corey:
Oh, that's a wonderful question. Okay. Again, based on our data, right now, T NRC, six, the critical scaffolding protein that promotes cooperativity, its expression at high Confluence seems to go way down. So that would suggest that it's not as able to do via scaffolding protein at high Confluence. So, what is the impact on that that's something we're going to have to look at. But it appears that it can't form the high molecular weight complex as much. But then also, I want to get into complexity of it. But there may be some differences also in the three GNRC sixes and how they bet. So yeah, a lot of interesting stuff there.
Audience 3:
As we discussed there, you know, earlier, there is a growing body of anecdotal reports and probably some literature about successful RNAi of allegedly nuclear RNAs, especially noncoding RNAs, which seems to indicate that unless there's some type of cytoplasmic, an undetected, which is unlikely, especially we'll be able to look for them with fish. There seems to be a way for Si-RNAs to successfully reach allegedly nuclear transcripts somehow, and this is fueling speculation in the field about nuclear risk.
David Corey:
Right, right. No, absolutely. Although one thing that I will say is I don't think there's any doubt that small RNAs can get loaded in the cytoplasm imported in the nucleus and then bind to targets in our experience, the actual cleavage of those targets is much, much less predictable. So right now, imagine them working like binding factors, not cleavage factors like it happens in the cytoplasm.
Audience 3:
Think that's great, very provocative.
David Corey:
Okay. Thank you. All right. Are there any other questions? All right, thank you very much. Thank you so much.