Scientist Stories: Phillip Zamore, RNAi and Argonautes

It's a great honour for me to introduce today's speaker for the Walsh lecture, Dr. Phil Seymour. Phil is a pioneer in the field of RNA interference. And his lab has continued over the years to identify novel small RNA pathways, reveal their function, and elucidate their mechanisms. Currently, Phil is the Gretchen Stone Cooke Professor of Biomedical Sciences at UMass Medical School, and he's also a professor and chair at the RNA Therapeutics Institute, in addition to being a Howard Hughes Investigator. Phil is originally from Long Island, but then really started and did all of his education in Boston, training at Harvard and MIT. He worked out mechanisms of splicing, and then went to gene regulation by RNA-binding proteins. So he was really always interested in RNA-guided and RNA-binding proteins and their regulation.. And he was very successful at what he was doing. Already just before he got independent, he transitioned into something new and really explored a completely new field that we now call RNA interference or RNAi. And his early success was awarded a Pew Scholar Award, and he was also a scholar of the W.M. Keck Foundation. And if I get this right, together already while he started his independent group at UMass, then he already founded his first company. So, and since then he has been involved in academic research as well as in multiple biotechnology companies. So practically, whenever you design or purchase or use your siRNA to knock down your favorite gene, you're harnessing the intellectual property of multi Seymour patents, both for design and for the application of RNAi in use, especially in mammalian cells. And so it might not come as a surprise that Phil was on the list of the most influential scientific minds in 2014 by the Thomson-Reuters, and he's also a fellow of the National Academy of Inventors. I think one thing that is— there's so many things, but one thing that is so beautiful about Phil's work is that this is really a prime example that basic research has always been driven, always been driving biotechnology, and has led to major advances in innovation in medicine. So the companies that Phil founded and also that he's involved in now really brought the first RNA therapeutics on the market. And I think RNA therapeutics, just— we've learned this this year— are extremely promising. Phil has a lot of awards. I want to mention, of course, the NIH Merit Award from NIGMS. And in addition to a lot of scientific awards, he also got mentoring awards. And I would like to stress this point too, that Phil is an exceptional mentor. And you will find his previous postdocs in important faculty positions all around the world. And like for me as a younger one, younger generation in that field, he has always been also an amazing role model. Because he hasn't only supported his postdoc and students into great positions, he also showed us that it's possible to be super successful, have a super successful wife, and two amazing kids. And I think this is also something really special. So without much further ado, I'm a tenure-track investigator at the NIDDK, and my name is Astrid Haase, and I'm excited to hand over to Phil to tell us his new adventures with the Argonauts.

Astrid, thank you for that incredibly kind introduction. My lab has worked on Argonauts now for 22 years, and mainly we have worked in plants, a little bit— I'm sorry, plants, animals, a little bit in fungi, But until recently, we never worked on Eubacteria. And on Monday of this week, I spoke at the NCI about piRNAs, which is an Argonaut-driven pathway found in all animals, including the animals depicted here. But today I want to talk about this little fellow, Thermus thermophilus. Which also has an Argonaut protein. Argonaut proteins are found in all clades of life, including Archaea and Eubacteria. The Argonaut proteins I spoke about on Monday are in the PeeWee clade. The Argonaut proteins that enable RNAi drugs to be possible are in the Ago clade. And then the Argonauts in Eubacteria and Archaea are called P. agos, because they're found in prokaryotes. So if we look at the siRNA pathway, it was present in the last common ancestor of all eukaryotes. It has antiviral functions, antitransposon functions, and it controls the production of heterochromatin in many different eukaryotes. MicroRNAs are found only in plants and animals, and they evolved separately, uh, independently from the RNAi pathway, and they regulate mRNA stability and translational, uh, rates of protein production. And then piRNAs are found only in animals, uh, where they defend the germline from transposons. But eubacteria and archaea have Argonaut proteins, and we don't actually know why. All of the experiments that have been feasible to date have involved Argonaut proteins from organisms that are difficult to culture in the lab, and then moving them into E. coli to study their biochemistry and, to the extent possible, their biology. And what I'd like to tell you about today is our lab's attempts to learn something about ubacterial Argonaut proteins using Thermus thermophilus as a model system. So typically eukaryotic binding protein Argonaut proteins bind double-stranded RNA, for example siRNAs. But what does TTAGO, the Thermus thermophilus Argonaut, bind? So to answer this question, my grad student Sam Jolly, with help from some talented visiting scientists in the lab, set out to figure out the affinity purified PTAGO for different kinds of RNA and DNA structures. So in many ways, Thermophilus argonaut has the exact opposite binding specificities of mammalian Argonaut proteins. It binds very poorly to double-stranded RNA. It binds weakly to single-stranded RNA. It binds weakly to RNA-DNA duplexes, and it binds strongly to double-stranded DNA and very tightly to single-stranded DNA. So this is pretty close to the opposite order of most eukaryotic argonauts. Thermus thermophilus is a gram-negative bacterium that grows at 65 degrees Celsius, and it was found originally in a thermal vent in Izu, Japan, by Tara Oshima and Kazutomo Imahori. It now can be cultured in the lab by substituting hot spring for an incubator, and that enables us to grow the bacterium, to perform mutagenesis and homologous recombination to change genes and make site-directed mutants. We can measure its growth, we can examine it, and so we have the tools now to study the actual biology of this organism that unlike E. coli has an Argonaut protein. There are two lab strains of Thermus. I'll tell you results today largely from HB27, but all of these experiments were replicated in parallel in HB8. Now, Thermus thermophilus has a large circular chromosome of about 2.1 megabases. It also, depending on the strain, either has one or two megaplasmids. These are near megabase size plasmids that are present at many copies in the bacterium. And its chromosome itself is hexaploid. Okay, so both the TTAGO protein and the CRISPR-Cas system are located on the megaplasmid, but they're quite distant from each other. So even though they might, theoretically both have defense functions, we can't infer anything about the function of Argonaut from the neighborhood on the chromosome where it lives. So for many years, the Schwartz Labs and— yeah, has studied the role of TTAGO in vivo using both thermus and E. coli as models. And what they found was that small interfering DNAs, that is single-stranded DNA guides, bind to TTA GO and cleave complementary DNA strands, typically in AU-rich regions that at 65 degrees tend to open periodically. And the main function of Argonaut that was learned using this approach was that it acts as a barrier for the uptake and propagation of foreign DNA. In other words, if you attempt to transform Thermus thermophilus with a plasmid, the efficiency of transformation goes up tenfold if you mutate the Argonaut protein. So Sam Jolly, an MD-PhD student in my lab, along with Han Zhang, who at the time was a visiting scholar from China and is now a graduate student in my colleague Eric Sonnheimer's lab, as well as my postdoc Ildar Genetinov, all set out to try to understand the biological function of TTAgO in vivo at 65 degrees. And Sam asked 3 fundamental questions. What does it use for guides? Does it interact with specific proteins in vivo? And does loss of TTAGO have a phenotype that informs its biological function? So to tackle the first question, Sam raised an antibody against full-length TTAGO and used it to immunoprecipitate the protein from logarithmically growing bacteria. He then obtained the RNA that was bound to the— I mean, sorry, the DNA that was bound to the Argonaut protein and ran it on a gel after using the exchange reaction to label its 5' end with radioactive phosphate. And you can see here that he treated the DNA bound to Argonaut with 3 different kinds of conditions. Treating it with buffer alone as the control, and he found that the Argonaut protein was associated with small DNAs, 14 to 19 nucleotides long. When he digested these with DNase, these were destroyed, establishing that they're made of DNA. When he digested with RNase, they were not changed, so that we concluded that the guides are strictly from DNA, and we only recover them in strains that have the Argonaut protein. So we know now where— we know that the TTA ago protein binds DNA guides, but what Sam wanted to know was where those guides come from. And so to do this, he performed high-throughput sequencing of the DNA guides using a strategy originally devised by Alexei Aravetz, lab. He had quite a surprise. What he found was the guides predominantly came from a region of the plasmid directly across from the origin of replication. And using a series of experiments that I won't discuss today, Sam was able to demonstrate that the region that they come from is in fact the flanking DNA that surrounds the terminus of replication, the place where replication forks proceeding counterclockwise and clockwise around the circular chromosome collide and end replication. He found that the same thing was true for the megaplasmid, that the guides on the megaplasmid largely mapped to the terminus of replication, uh, across from the origin. And These guides, just like the radioactive guides, had a modal length of 16 nucleotides. So Sam was very rapidly able to answer the first question in vivo: what does TTECO use for guides? And the answer was it uses roughly 16 nucleotide long single-stranded DNA guides from the terminus of replication. Next, he asked does it interact with specific proteins? To test this, he again performed the immunoprecipitation, but instead of sequencing RNA, he performed mass spectrometry on the associated DNA proteins. What he found was that in a wild-type strain, the immunoprecipitated proteins that were enriched in the IP relative to the deletion mutant all had known functions in replication, in the resolution of cell division, in chromosome segregation, and in DNA repair. So these proteins were really quite well-known proteins, well studied in E. coli, such as single-stranded binding protein that protects single-stranded DNA, the AddAB system, which is involved in recombination in Thermus thermophilus and is the equivalent of the RecBCD system in E. coli, PriA, a primase that restarts stalled forks, gyrase A and gyrase B, the two subunits of gyrase that decatenate the circular chromosomes at the end of replication, and similarly topoisomerase and a histone-like protein called Hu that helps package the DNA in the bacterium. Some of these proteins were associated directly with the Argonaut, as evidenced by the fact that if you digest the immunoprecipitate with DNase, they remain bound to the Argonaut protein. And the proteins that bind Argonaut directly include AdAB, the gyrase, and the PriA, which restarts stalled forks. If you looked at a mutant Argonaut that was catalytically inactivated by two amino acid changes at its endonuclease active site, he found that a different subset of proteins were able to interact with the Argonaut protein. But again, many of these were shared with the other IPs. So for example, he found the two subunits of gyrase, he found PriA, he found SSB. And lastly, if he treated that with DNase, he found a subset of the proteins were directly associated with the Argonaut protein. And And so this subset also included the gyrase A protein, PryA, topoisomerase, RepA, which initiates replication, and so on. So again, he came to a very similar answer. The guides were from the region of the chromosome where replication finishes, and the specific proteins associated with the Argonaut were also associated with the final steps required to replicate the circular chromosome. So Sam next asked, does TTAgo have a phenotype if you delete it? So Sam built 3 isogenic strains that all contained the same antibiotic resistance marker that was used to generate the strains, a wild-type strain that simply had an insertion of the— I'm sorry, wild-type strain that simply had an insertion of the resistance gene, the catalytically inactive mutant that again had the same resistance marker, and the deletion that was missing only Argonaut and contained the resistance marker. So because the bacteria have relatively small genomes, we were able to completely sequence all of the three strains and demonstrate that Except for what I've diagrammed here, they're identical to the single nucleotide. So Sam had his first major disappointment when he set out to try to understand what the phenotype of Argonaut deletion was. What he found was that no matter what assay he used, the Argonaut protein seemed to have no required function. The cells grew at a normal rate, they looked normal under the microscope, the cells grew at the same density, the colonies were the same size, and so on. And so acting on the advice of one of my colleagues, Nick Rind, who works in Schizosaccharomyces pombe, Stan asked if they had different sensitivity to replication inhibitors and other DNA-damaging agents. And he explored agents such as hydroxyurea, uh, 4-nitroquinolone N-oxide, cisplatin, mitomycin C, methylmethanesulfonate, ciprofloxacin, which is an inhibitor of the A subunit of gyrase, and novobiacin, which is an inhibitor of the B subunit of gyrase. Of all of these, only ciprofloxacin had an effect. So what he found was that the strains that lacked Argonaut or that had catalytically inactive Argonaut were more sensitive to ciprofloxacin than wild type. So here I'll show you an example of Sam's data. This top row is wild type Thermus thermophilus with its fully intact argonaut gene, and in both the presence and the absence of ciprofloxacin, Sam observed the same number of colonies of the same size, suggesting that ciprofloxacin had no effect. In other words, the bacteria are naturally ciprofloxacin resistant. When you remove the argonaut, as I said before, he could detect no difference between wild type and Argonaut deletion under normal conditions. But when he deleted the Argonaut and then asked whether it could grow on ciprofloxacin, he had quite a surprise. What Sam found was that ciprofloxacin inhibited growth. It wasn't bactericidal. The bacteria still formed colonies at the same number as the minus ciprofloxacin control, but the colonies were tiny, suggesting that growth was slowed dramatically in the presence of ciprofloxacin and the absence of arginine. So when Sam began to explore why the colonies grew so slowly, he turned to electron microscopy, in this case SEM, to examine the bacterial appearance to see if there was a phenotypic difference between wild type and mutant strains. So here on the left as a control is wild type growing under normal conditions. Each of these is one single bacterium. Here's a pair replicating. They've just separated from each other. Then in the presence of ciprofloxacin, you can see that most of the bacteria are fine and are single individual bacteria. There's a slight increase in the number that appear to be very close to each other, or perhaps not completely separated, but all in all they look perfectly fine growing at 12.5 micromolar, uh, ciprofloxacin. Sam had quite a surprise when he looked at the deletion mutant of Argonaut. However, in the presence of ciprofloxacin What we found were essentially long filaments representing many individual bacteria that remained fused at the end of cell division, unable to complete cytokinesis. So these, these long filaments, or as we like to call them in the lab, sausages, in the Argonaut mutant were not observed at all in the wild type. But they were also found in the point mutant that had an intact Argonaut gene except for 2 amino acid substitutions that made it unable to cut DNA. So here you can see the quantification of those experiments, and you can see that the ciprofloxacin bacteria in wild type are actually slightly longer, suggesting that they're filamenting a bit. And then in the mutant that can't cleave DNA, they're longer still. And then without the Argonaut, they're extremely long. So something like 4 or 5 times their normal length. And this is probably a significant underestimate of the length of these filaments, because in order to examine them by SEM, Sam had to spin them down onto the grids, which likely shears very long filaments of bacteria. So our model was that the Argonaut protein was required in the presence of ciprofloxacin to complete cell division, and that predicts that there ought to be DNA linking the individual bacteria in the filament. And to do this Sam turned to super-resolution microscopy called STED and used a dye called PicoGreen that only fluoresces when it's bound to double-stranded DNA. So without Argonaut, in the presence of ciprofloxacin, you see again these long filaments, and you can see if we zoom in here on the PicoGreen data that the DNA is actually extending continuously, forming a bridge between the two individual bacteria on either side of the septal junction that normally would close completely and separate the bacteria, creating two bacteria at the end of cell division. Instead, we have two linked bacteria with DNA running through the septum and linking them together. You see the exact same thing in the point mutant Argonaut that can't cleave DNA, but the septal junction does not contain DNA, even in the bacterial examples we could find in wild type where two bacteria lie very close to each other in space, suggesting they had just finished replication. You see in fact that the DNA is pulled well back from the septal junction linking the two. Sam also used transmission electron microscopy to examine the nucleoid, which is essentially like the chromatin in eukaryotes. So the nucleoid is a DNA-protein complex that runs the length of the bacteria. It's where the chromosomes are packaged, just as we might package our DNA in the nucleus in chromosomes with histones on them to allow compaction. The bacteria package them in these nucleoids. And in the wild-type case, the nucleoid actually does not extend all the way to the extremities of the bacterium. So there's a space between the nucleoid and the membrane, and there's no DNA running in between them. And you can see even here where we caught the very end of cell division, there's— the nucleoid is well pulled back, well separated from the septal junction. In contrast, in the Argonaut mutants, you can see that the nucleoid is actually threading through the septal junction. And as we saw before in the pico green staining, DNA is running through the septal junction. And so our interpretation is that we actually have a protein package DNA fiber or fibers running through the septal junction, preventing cytokinesis from completing. And causing filamentation. And we saw the same phenotype in the point mutant that's unable to cleave DNA. Sam asked, as I mentioned before, does loss of TTA-CoA have a phenotype? Well, in the presence of ciprofloxacin, the answer he got was yes, probably because they can't decatenate their circular genomes. Ciprofloxacin is an inhibitor of gyrase. What gyrase does is it takes the two linked circles that result from DNA replication, it cuts one of them on both strands, passes them through, and rejoins them to allow them to separate. Somehow Argonaut is allowing that process to continue in the presence of the gyrase inhibitor ciprofloxacin. Without the Argonaut protein, the, the ciprofloxacin completely blocks cell division. So our conclusion was the biological function of TTiGO in vivo at 65 degrees was to participate in the resolution of the catenated circular genomes produced by DNA replication in the presence of ciprofloxacin. And our model was the normal separation of circular chromosomes carried out by the gyrase A subunit of the heterodimer, which is blocked by ciprofloxacin, was complemented by the Argonaut protein somehow recruiting the recombinational and repair machinery, perhaps by catalyzing two repair— two recombination and repair events to decatenate the chromosomes. This type of mechanism has been demonstrated not with Argonaut, but using recombination in place of gyrase in E. coli. Now, I have to tell you that for a long time we thought about this whole process backwards. So, until truthfully, until I visited the NCI on Monday, I always thought of this as the period in our lab when we went around saying, woe is me, my protein is normally dispensable, which at least in the mouse and fly community is kind of an unfortunate occurrence where people pity you. But on Monday, it occurred to me talking to people at the NIH that in fact what we have found is that Argonaut confers ciprofloxacin resistance on Thermus thermophilus, which would be the first example of an argonaute protein that's an antibiotic resistance gene. So I'll be— my lab is very interested in the future in trying to understand if there are other examples, perhaps in pathogenic bacteria, where argonautes are involved in antibiotic resistance. Now we didn't stop there because we really did want to know whether or not Argonaut was important in the absence of antibiotics. And so Sam performed an experiment that was designed to simulate bacterial evolution on a more practicable timescale. So what Sam did was he took bacteria and seeded them into a large culture flask, and then every 12 hours, serially transferred them into fresh media, and he performed this at 65 degrees for 12 days, and he performed it on the wild-type strain that has an Argonaut protein and the mutant strain that doesn't. But there were no selective agents here, no ciprofloxacin, this was enriched media, everything was optimal to the best of our ability. And as a readout, Sam used high-throughput sequencing of the entire genome, which gives us a coverage map of the Argonaut gene and then gives us the entire rest of the chromosome as a control. And so you can see here that if you use wild type and we zoom in on the region of the genome that contains Argonaut at day 0, the sequencing experiment gives us even coverage across this region of the chromosome, demonstrating that in fact we're growing Argonaut-containing bacteria. When Sam does the same experiment sampling at the beginning of the experiment in a strain that's deleted for Argonaut, you can see that there are no reads mapping to the Argonaut gene, but the flanking reads are essentially identical everywhere else in the sequence of the chromosome. So this absence here is diagnostic of the mutant, and the presence of sequencing reads here is diagnostic of wild type. So Sam mixed wild type and mutant bacteria in roughly equal proportions on the first day of the experiment and sequenced them. And you can see that the coverage of the Argonaut gene is now intermediate between no coverage in the mutant, which is about 50% of the mix, and complete coverage in wild type. But at the end of 12 days, what Sam found was that the coverage of the Argonaut gene, but not the flanking sequences, had increased significantly. So to show you this, Graphically, for 3 independent trials of the same experiment, let me show you how we quantified this first, and then I'll show you the data. So Sam took the coverage over the Argonaut protein relative to the coverage over the resistance gene that's shared in both bacteria and that was used originally to make the deletion. And you can see here that there's a significant increase in all 3 trials between the amount of wild type relative to the amount of mutant at the beginning of the experiment and the end. So by the end of the experiment, wild type had outcompeted the mutant strain with no selective pressure. So this helped us understand that Argonaut is maintained in Thermus thermophilus because it actually confers a selective advantage under normal growth conditions. Now, we don't formally know that it's assisting in decatenation of chromosomes in the absence of ciprofloxacin, but I think right now that's the simplest explanation. Is that it participates with gyrase in completing cell division. Now I showed you that the catalytically inactive mutant also filamented and had DNA and nucleoid running through the septal junction. Sam wanted to understand whether or not this endonuclease activity was required for the production of guides or if it was only required for some downstream function like cleaving DNA. And so Sam, using again high-throughput sequencing, measured the number of molecules of DNA guides per cell in both wild-type and catalytically inactive Argonaut strains. What he found was guides were dramatically reduced, 20-fold fewer guides in the mutant strain, suggesting that the catalytic endonuclease activity of the Argonaut protein is required for the production of guides. This wasn't because the protein is unstable, since there are identical numbers of molecules of protein per individual cell in both genotypes. The guides are also the same length. They both have a modal length of 16 nucleotides, but in the mutant, they're just rarer. So if we zoom in, you can see that again, modal length is 16. The guides also have the characteristic first nucleotide C that has been observed first in the Schwartz lab and now in Sam's data, that the majority of guides begin with C in both wild type and also in the catalytic mutant. And finally, they come from the same place in the chromosome. Although the guides are more rare, they still come from the terminus of replication. So this allowed us to construct a model for how Thermus thermophilus acquires its guides from a pre-existing pool of short single-stranded DNAs. So we imagine that TTAGO loaded with guides remove this. I'm sorry, so we examined whether TTA Go loaded with guides by grabbing a pre-existing pool of short DNAs. So you can imagine that an easy way for TTA Go to obtain its guides is if the chromosome is occasionally fragmented as part of, for example, the collision of replication forks. Works. So to test whether or not there were free guides separate from TTAGO, Sam quantitatively depleted a wild-type Thermus thermophilus extract of TTAGO. You can see here that in the wild-type strain, the IP contains all of the protein available. So the soup has no extra protein in it, but the soup also has no guides, just random noisy bits of DNA. So the IP actually contains all of the 16-nucleotide guides characteristic of the DNAs that guide Argonaut. Similarly, in the catalytic mutant, and in the deletion mutant, there are no guides in the deletion mutant, and obviously none are recovered in the IP since there's no Argonaut, and the noise in the supernatant is identical to what you see in wild type depleted of Argonaut. You can see a slight peak here at 16 nucleotides representing that low level of guides I showed you before that accumulate in the catalytically inactive mutant, And these guides come down with the IP, but cannot be found in the supernatant, ruling out the hypothesis that TTAGO acquires its guides from a pre-existing pool of short single-stranded DNAs. And in both cases, in the wild type and in the mutant, the guides that were associated with the total immunoprecipitated Argonaut protein begun with C. So in order to try to understand where the guides come from and why we formulated the model we have, I want to just review some of the known properties of Argonaut, one of which, as I mentioned before, is that its guides tend to begin with C. Now, many Argonaut proteins have a first nucleotide preference. For example, Argonaut2 in flies prefers guides that begin with C, whereas in plants, there are several Argonauts that prefer guides that begin with A, and the Argonauts in mammalian cells, such as human Ago1, prefer guides that begin with U. So to test the idea of whether or not TTiago has a first nucleotide preference, Dan Schwartz and his collaborators asked if they could detect guides beginning with a specific nucleotide bound to the Argonaut protein using a fluorescence assay, a single molecule fluorescence assay. What they found was that there was no significant difference between guides beginning with A, C, G, or T. So it wasn't that the Argonaut preferred to bind single-stranded guides beginning with C. However, when the Argonaut protein was given a chance to bind double-stranded molecules that not only began with C, but the C was paired with G, what, what Dan found was that there was a slight preference for a target that began with G. And you can actually see in the crystal structure of the protein that there is a pocket for a G in which it can be read by an arginine and several other amino acids present in the pocket, reading the functional groups on the G and accommodating it in a way where it actually can't base pair with the C across from it. So it isn't that the Argonaut protein is recognizing this C. In fact, based on the crystal structure of Argonaut, we know that this C is in its own pocket, which even though it doesn't recognize a C, recognizes the 5' phosphate, causing it to be unpaired from the G. And the G was actually in its separate pocket being recognized by its functional groups. So any model about how TTiago acquires its guides has to take into account the fact that it prefers targets with a G but doesn't prefer guides that begin with a C. So our model begins like this. If a replication fork stalls or collides, it's known that it will often produce nicks in the DNA, that is, a single-stranded break ending in a 5' phosphate. When that happens in front of a CG base pair, this may serve as a substrate for Argonaut binding. And as I showed you before, Argonaut has a high affinity for double-stranded DNA that begins with a 5' monophosphate. In this way, TTiAGO can take advantage of its ability to recognize the G across from the C. If this C-containing strand ultimately becomes the guide, then it will appear as if TTA-Co prefers guides that begin with C, when in fact it was acquiring them for double-stranded RNA that contained a G across from the C. Next, we postulate that like all Argonaut proteins, the C is flipped out, uh, so that the 5' phosphate can be secured in a binding pocket that confers high stability to guide binding. This would allow Argonaut to then cut the DNA strand that contained that gene nucleotide, the target strand. If Argonaut makes that cut, then you might imagine that the piece of DNA that's secured by only 10 base pairs here, since Argonaut always cuts after the 10th base pair of its target, would have a low enough hybridization energy at 65 degrees to dissociate, leaving the Argonaut protein loaded with a precursor of its guide. And this model has the advantage in that it explains why guide acquisition is so inefficient in the catalytically inactive mutant Argonaut. Based on the proteins that associate with Argonaut, we hypothesize that a protein such as AddAB might make a downstream— might resect the DNA from a downstream double-stranded DNA break, chewing the DNA back until the footprint of the protein, which typically for Argonauts determines guide length, and then you would only have 6 nucleotides left to get rid of, which at 65 degrees would just fall off, revealing loaded guides. Now, I won't have time to show you data because I want to leave time for questions at this point, but a student in the lab, Shannon Bailey, has now shown that if you give purified TTA-Go a substrate that contains a single nick at a CG nucleotide that has a 5' monophosphate, that Argonaut will not only bind, but it will cut at exactly the predicted position on the target strand. If there's no monophosphate, this reaction is much more inefficient. And what she's now testing is whether or not there's an influence of this G across from the C in guide acquisition and whether or not she can detect departure of this cleaved product. Ultimately, her goal is to use in vitro extracts of Thermus thermophilus to find which host factors are involved in guide acquisition. Now, one of the things about our model for TTAGO function is that we found it was associated with many of the proteins involved in replication, decatenation of chromosomes, finishing replication and leading to cytokinesis and successful separation of two bacteria. Now, originally we thought maybe that since TTAgO is an endonuclease, that it might be decatenating chromosomes by cutting once on one strand and once nearby on the other strand. But our finding that TTAgo may acquire its guides actually through cutting one of the two DNA strands in the chromosome suggests a second possible explanation for how it functions. It may be that TTAgo simply is a landing platform that recognizes nicks in the genome and then is able to make a double-strand break across from the nick but is not involved in any way in the repair process, but instead these breaks are promoting recombination using the other proteins associated with TTAgO. So I'd like to stop there and thank the members of my lab who worked on this project. Karina, who performed single molecule experiments on TTAgO that helped us evaluate our model for its function and for how it acquires guides. Han, who I mentioned before, Ildar, who helped Sam with most of the bioinformatics that went into the paper, and Sam, who very diligently stuck with this project through many, many years, even though he knew he had to get back to medical school. Sam is now finishing up his medical training and looking forward to a residency in a post-COVID world. Thank you for your attention.

Thank you very much, Jim. We have some questions already in the chat, and if you have any additional questions, there is a way to put your question in the interface, and we'll see this. The first question is from Ryan Stanton, who is asking, what would happen in bacterial species that perform rolling circle DNA replication instead? Would you then expect the Pago to actually reside— would you expect the Pago coding region would be adjacent to the origin of replication instead of across from it?

It's a great question. I don't know the answer. We've never tested it. It would be, it would be great to collaborate with someone who works on that type of bacterium since this is the first time we've ever studied a bacterium in the lab, and it's already been a big adventure, and I think I'd prefer to have someone more knowledgeable holding my hand if we were going to pursue that.

Susan Gottesman is asking, are there ADD proteins, so ADD, I hope I pronounced it correct, are they needed for ciprofloxacin resistance in the EGO+ or is there any additive effect in the EGO-?

I don't know the answer to that. We hope to begin making mutants in many of the proteins associated with Argonaut. To test that kind of hypothesis, but we haven't performed those experiments yet.

And Susan has a second question. If a NICRAS stall is made elsewhere, would the guides then accumulate in that location?

Well, that's definitely the prediction of the model. The number of guides per cell is low enough, though, that if there is a 10-to-1 preference for the terminus of replication, it becomes very hard to distinguish the guides from elsewhere in the chromosome from noise. But I will definitely say that although the guides are highly enriched in the terminus of replication, there are guides from other parts of the genome. Including the CRISPR loci, which of course have to be nicked in order to acquire a new spacer.

There's a question from Tom Snyder, who did the first sequence logo algorithm, and he said that he noticed the conservation of a C every 3 bases, so that this fading is usually associated with coding frames. And how does that fit with your model?

That was a really, really great observation. It turns out if you simply take the genome of Thermus thermophilus, there is an enrichment of C in every third position, even in regions that don't encode proteins. And I don't know why.

I got to get some questions. I know I get to get to ask some questions afterwards again. The first one, is, do you think there's actually space for another endonuclease? Like zucchini, one of the pioneer endonucleases famously has homologues in bacteria that are DNases, and the only one known is a double-stranded DNase. So do you think that could participate too? Does, does Thermococcus have a nucromologue?

I don't know the answer to that. But many of the proteins that are tightly bound to CTA-Go in our mass spec IPs are themselves DNases.

So you think there could be another— there could be a collaboration between multiple nucleases too? Absolutely. And perhaps one more. Now, given the whole universe of small RNA silencing pathways in eukaryotes, and like the double invention of microRNAs, do you expect that in prokaryotes too? That we will identify all these different classes that have different functions?

And yeah, I do, because if you look at the amount of evolutionary time in terms of generations, obviously bacteria have had much greater opportunity to explore novel, potentially beneficial functions for Argonaut proteins. And so about, say, a quarter of all archaea and a little bit less for the eubacteria have argonauts. And so it's not entirely clear if they gained the argonauts by horizontal transfer or by descent. I think the evolutionary experts who I've spoken to tend to prefer the descent model. And if that's true, then I think we should expect a large number of functions that we've never seen in eukaryotes, like this one where, you know, no one's ever seen such a role for an Argonaut protein in the eukaryote.

Is there any— I might be mistaken here, please correct me— is there any connection to the CRISPR system? Because some of them, I think, are in operons, aren't they, with the Agos?

Am I wrong there? Well, in this organism, no. I think there's 80,000 bases between Argonaut and the next defense gene, and it isn't a CRISPR locus. It's a different kind of defense gene. So I don't think there's likely to be a connection, but that's— that's sort of a weak piece of evidence. I would say we can't rule it out, but there's no reason to rule it in.

Have you seen— I'll just continue here. Have you seen multiple, like how many egos does a prokaryote usually have? And do you think they would have like an RNA-based thing and a DNA-based thing, or is it either or?

So far it mainly looks either or. So there are bacteria that have Argonauts that use RNA guides and appear to be DNA-cleaving enzymes. And I suspect there are bacteria that have RNA guides that are RNA-cleaving enzymes. It's actually really easy for Argonauts to evolve between these two. And part of the reason is that even when Argonauts are using DNA guides to cleave DNA, the double helix in the Argonaut protein is still A-form. So the DNA is actually accommodating the Argonaut rather than the Argonaut learning to bind B-form DNA. And I think that makes it easier for evolution to have constant movement between whether the guide is DNA or RNA, whether the substrate is DNA or RNA, And probably the limiting factor is simply the guide acquisition pathway.

So that is a whole— I think what we know now about the diversity in eukaryotes, so you really expect that they could be harnessed for anything, or really fast evolve to do different functions in prokaryotes.

I do, because if you don't have to acquire your guides through a very complicated pathway depending on having the right dicers and the right helicases in exactly the right complexes and compartments, I think if guide acquisition could be as simple as finding a nick in the chromosome, then it should be possible to repurpose Argonauts across evolution pretty rapidly. But I mean, we have to test this.

Do we already kind of know all that? Because there are at least these like short and long Argonauts, isn't it, in prokaryotes? Do we know the whole wealth? Like, did anyone look at all the Argonauts in all the prokaryotes where we know the genomes and see how much diversity we actually have in the proteins themselves?

Yeah, a number of people, including Alexei Aravan, have done that and classified them, as you said, into short Argonauts and long Argonauts, Argonauts that are likely to retain their catalytic activity and those that are likely to have lost it. But I think the truth is, you know, when we study Argonaut in a mouse and a fly, we're talking about organisms that are much closer together in evolutionary time than many of the bacteria whose genomes have been sequenced.

That's true. Chris just reminded me that before the thing went online, before your talk went online, you told us the story about argonauts, and he was wondering if you would mind repeating that for the audience.

That's a nice— so, so the name argonaut comes from the plant community. Where a mutation in the genetic screen produced an unusual growth phenotype that caused Arabidopsis to look very much like an octopus. That's named Argonaut. So you can imagine the tentacles were sticking out in all the directions you would expect for an octopus rather than growing up spindly the way Arabidopsis does. And so that name was concurred by a group in France, and Argonaut ending with an E, that's the spelling in French. And once it was figured out that the same protein family was active in the RNAi pathway, it was quickly, you know, decided to name everything Argonaut with a number. The only exceptions are the Kiwi proteins that have names that come out of the fly tradition of naming Argonaut protein, I'm sorry, Kiwi proteins with some variant of Kiwi, like Miwi for mice and Hiwi for human. So, which is a tradition that's unsustainable. And I think more and more organisms are ending up with PB proteins preceded by the first two letters of the genus and species of the animal. But, you know, the history of nomenclature in the RNA silencing world is pretty arcane.

But it is a very— it gives you a very nice story into the, into the ego world. And With that, Chris also asked me to repeat the code for your educational credits, which would be 31301. And Phil, if you don't mind, if there are no more questions, guys, I would like to thank Phil. And I can clap for everyone. Thank you so very much. And it would be lovely if you could join me on the other link afterwards.

Will do. Thank you, everyone.

Thank you so much for doing this. Thank you.