 So thank you very much for the opportunity to present my work in this conference that I've really enjoyed so far. So I'm going to show you what we've been doing with system CRISPR in bacteria. I know we're using them both to develop tools to study and to fight pathogenic bacteria. So I first want to give you a bit of history about these CRISPR systems. So they basically were first discovered but not understood in 1987 by this Japanese group who was just sequencing this IAP gene and just downstream of the genes. They found this interesting sequence where you had these repeats that were interspaced by this sequence that looked sort of random. And the size of the repeats and of these variable regions was very well conserved. At the time of course they didn't understand at all what these sequences were. And you should look at the citation report for that paper. Basically it was published in 1987 but it got almost no citations until 2007. So in 2015 it hasn't been updated but it's probably up there now. So what happened in 2007 is basically this paper. So that's a landmark paper in the field of CRISPR. It's basically the first experimental evidence that this CRISPR loci actually provides acquired immunity against bacteriophages. And what they did in that paper is they worked in this strain of streptococcus samophilus. The interesting thing also to note is that this paper didn't come out from an academic group but rather from the industry and from a company called Danisco that was later bought by Dupont that was developing ferments for lactic fermentation, for yogurt production, cheese production. And these people as other speakers before have noted have big problems with viral contamination that can ruin the batch production. So they always try to select strains that are resistant to bacteriophages. And when doing so basically they noticed that the CRISPR locus was changing and that it was capturing new sequences coming from the phages that were using the challenge. And so when CRISPR captures a specific spacer so the sequence coming from a phage it becomes resistant to the phage that matches this sequence. So that was a very interesting discovery. So since 2007 the field has moved at a quite incredible pace and we now understand a lot about this CRISPR systems. And this is basically the picture of how they work. So first I should also say that CRISPR itself stands for clustered regularly interspaced short palindromic repeats so that's very complex acronym and that was coined actually by bioinformaticians. So that explains very well the structure of this low-cyber, you have these clusters of repeats that are regularly interspaced and repeat themselves frequently contained short palindromes. So hence this long CRISPR name. And the idea is that when you have a bacteriophage injecting its DNA in the cell at some frequencies a CRISPR system thanks to some proteins known as CAS for CRISPR associated is able to capture a piece of DNA from the phage and integrate it along with a new repeat in the CRISPR locus. Once it's captured this information in this adaptation phase it's then able to use this information to fight infection by phages, similar phages. And the way it does that in this immunity phase is by transcribing the CRISPR locus in the first precursor RNA that's then processed into some small CRISPR RNA and those small CRISPR RNA are actually in complex with CAS proteins that are going to be able to cut and degrade sequences homologous to the CRISPR RNA. So basically the CRISPR is building a memory of past infection and then using that memory to fight infection by similar phages. I want to go just a bit more in details about one specific CRISPR system. So that's a Type II C CRISPR system from Streptococcus pyogenes and there's a good reason for that is this system is the one that is the most used for biotechnological application of CRISPR. And you see that this CRISPR system contains actually four CAS genes, CAS9, CAS1, CAS2, CSN2. It has a CRISPR RNA with actually six spacer and repeats. And the interesting part here is that it also has another small RNA involved in this process that's called a tracer RNA that stands for transactivating CRISPR RNA. And these three CAS genes here, CAS1, CAS2, CSN2 are actually involved in the capture of new sequences for the CRISPR system. And the CAS9 protein itself is able to carry out the whole immunity step from these systems. So the role of this tracer RNA is actually also very important in this system. What it does is that it's complementary to the sequence of the repeat. So when you transcribe the primary transcript here from the CRISPR array, the tracer RNA can then hybridize making duplex RNA with each of the repeats. And that duplex RNA is then recognized by CAS9 and by the host RNA3 and that processes the primary transcript into the smaller CRISPR RNA. And after this process, you basically end up with this complex as CAS9 and these two RNAs, the CRISPR RNA and the tracer RNA and all the elements of this complex are absolutely essential for the function of the system. So once you have generated this complex, this nucleoprotein complex, what happens is that it's going to basically look for possible targets. So it acts as a surveillance complex that's constantly scanning the DNA in the cell looking for homologies with the CRISPR RNA. But first what the complex is actually scanning for, is it's looking for a small motif, in this case the motif is NGG and that's known as the PAM motif for protospacer adjacent motif. And once CAS9 finds this GG motif, then it starts to unwind the double helix and pair it with the CRISPR RNA. If there is some match, then it can completely pair the full length of the CRISPR RNA and that triggers a conformational shift in the protein in the catalytic domains in contact with the double helix, introducing a double strength break in the target DNA and that can then lead of course to the degradation of the target DNA. So the way this system works has basically been very well described by the group of Emmanuel Charpentier and Jennifer Dodna and that got them to get this breakthrough price last year and mingle with some Hollywood stars like Cameron Diaz and the CEO of Twitter here. So it's good to know that if you do exciting science you can meet your movie stars. So what have we been doing with this system? So you said we're a microbiology group and we work on both understanding the ways this CRISPR system works in bacteria but also developing technologies and today I'm going to mostly talk about the technological aspects. And the first technology we developed using this is the strategy to edit the genome of bacteria. And the strategy here that we employed, so I'm showing here what we did in E. coli, we introduced a plasmid carrying Cas9 in the cell and then we can electroporate together another plasmid that is programmed with a CRISPR that will target here a gene we want to modify and at the same time we introduce a specific mutation we want by electroporating a single-stranded oligo that carries a mutation we want to introduce. And we don't do that just in any strain, we do that here in this HME63 strain from the lab of Donald Cort that expresses a lambda-red recombination machinery that allows to basically recombine this oligo at a high frequency and then the CRISPR system can basically select this mutation by killing all the cells that did not introduce this mutation. And that works relatively well. So at the same time that we were doing that work, there has been a lot of groups starting to work on CRISPR systems and people have called that the CRISPR craze and this is just a tiny, tiny sample of all the papers that have been published describing the development of CRISPR tools for genome editing in a very broad range of organisms. The global idea of all this tool work is, in all cases, you start with programming Cas9 to cut out a specific position in the genome that you want to modify. And once you generate this double-strand break, then the cell can deal with it in different ways. So either you can provide a template for homologous recombination and that's where you can control maybe the exact point mutation you want to produce, gene insertion, deletions, etc. If you don't provide a template for recombination, you can maybe rely on endogenous Ripper systems of the cell such as non-homologous and joining and HEJ. And in that case, the cell is able to just join together the broken ends, but it frequently makes mistakes doing so. And like this, you can introduce small indels at the target position and that's used a lot to introduce knockouts. And the last possible outcome is if the cell is not able to repair the break, then this will, of course, lead to the death of the cells. And when you look at the literature, the picture that we have is that this basically two pathways seem to work pretty well in eukaryotic systems, actually very well in eukaryotic systems to the point that the technology has spread super fast and now CRISPR editing has become really a standard technology to introduce mutation in eukaryotic systems. And in bacteria, it looks like CRISPR system is actually pretty good at killing the cells, and you can still use that for editing purpose but more as a selection tool than a way to trigger the introduction of a specific mutation. And the fact that you can actually use CRISPR to kill the cell, you can think that maybe you can even use that as an antimicrobial strategy. And like the very basic experiments you can do is if you have a plasmid carrying Cas9 and a CRISPR that you program, for instance, here to target the kinamycin-resistant gene, so that's work done in staphylococcus eras, that plasmid, you can transform it very well in the cell if the target is not in the chromosome. But if the target is in the chromosome, then you recover very little transformments and the idea is that the CRISPR is actually going to kill the cell. So like this, you can specifically kill bacteria that carry, for instance, antibiotic-resistant genes or virulent genes. But of course, you need to be able to deliver the system to an old population of cells. And the way we're doing that is by using phage as a vector. So phage are naturally able to package not only their own DNA but other DNA present in the cell, and that's called transduction. It turns out you can actually increase the frequency of transduction really a lot by simply cloning, for instance, a packaging signal present on the phage DNA. If you put that on a plasmid, then you can have the plasmid being packaged at a very high frequency in the phage capsid. And then you can use this system to inject a plasmid in the old population of cells. So we're using this strategy to inject a CRISPR system consisting of Cas9 and a CRISPR that we program to target an antibiotic-resistant gene present in the chromosome of Staphylococcus aureus cell. And the idea is that this should kill the bacteria. So this is just to show you the specificity of it. So what you see is loans of cell on a plate and it's actually either a strain that carries a kinamycin-resistant gene in the chromosome or that does not. And then we program the CRISPR either to target the kinamycin-resistant gene or with a spacer that doesn't target anything, and then we just put a drop of our CRISPR phage-med preparation on top of the bacterial loan and we see that it's clearing the loan only when the CRISPR is programmed to target the gene and the gene is present in the chromosome. So this is a very specific antimicrobial. So you might wonder why do we want to make antimicrobials that are so specific? What's the purpose of this? And the idea is that if you're able to specifically eliminate strains that carry antibiotic resistance, for instance, then you can take advantage of the competition with other strains that do not carry this resistance gene to more effectively eliminate the threat. And to demonstrate this idea, we did this very simple experiment where we just make a co-culture of two Staphylococci strain, one that is resistant to kinamycin and has this APH gene in the genome and another that's sensitive to kinamycin. And what we did, too, is that we put GFP-plasmid in the kinamycin-resistant so that we can easily follow both populations in a co-culture. And so this is what happened in the control experiments. So the cells are mixed one-to-one. We follow both the optical density and the dashed line is the fluorescence in the culture. And so in the control experiment, you see that about half of the population is the kinamycin-resistant cells. Now what would happen if you use, for instance, kinamycin as an antibiotic? So if you had an infection like that, it would be a very bad antibiotic choice. And here what you see is that, of course, you kill all the sensitive ones and you will only end up with a fluorescent Staphylococci. If you make a slightly better antibiotic choice, so, for instance, streptomycin, here what happens is that you basically will kill both populations of cells. But after a little while, you would select resistance in both populations and you would still end up with about half of the population being the bad guys. But now if you actually treat with a CRISPR system that specifically targets the kinamycin-resistant gene, so that the purple lines, you see that the fluorescent signal here is not recovered. In fact, we killed all the kinamycin-resistant cells, but what it means is that we killed enough of them and we let the other population grow in the culture. You see that by following the OD curve and these cells don't occupy the niche and prevent any possible survivors from coming again. So in this specific scenario, using a CRISPR antimicrobial can actually be more efficient than using antibiotic treatments to eliminate bad bugs. So we showed that we can not only do that in a laboratory strain of staphylococcus aureus, but also with some really pathogenic and problematic strains, like the metisline-resistant USA-300 strains that are a big problem in the US right now. And we showed that if we make a mixed population with this MRSA strain and some non-pathogenic staphylococci, we can specifically eliminate the MRSA strains by targeting the MAC-A-resistant gene. So I also want to mention that there is actually a very interesting side effect to speak of this strategy, which is when we treat a population of bacteria, we inject the CRISPR system not only in the bad bugs, but also in bacteria of the same species that might not carry the target. They're going to survive, and if our CRISPR is carried on a replicative plasmid, they're going to keep that CRISPR and know they are basically immune to horizontal gene transfer of the targeted genes. So in this case, we show, for instance, that we can immunize a population of staphylococci against the acquisition of the tricycline resistance by injecting this CRISPR system, and then we show that if we control that basically a non-working CRISPR system, we can do a transduction experiment to recover the tricycline resistance, but if we immunize, we cannot do it. So that's also an interesting possibility. So we went to do also some animal experiments with this ID, and this is a skin colonization model in the mice. What we do is we shave the back of the mice, and then we colonize with a mixture of resistant and non-resistant staphylococci, and then we treat with our phagemid preparations that target specifically the canine mice and resistant ones. The resistant ones are also here fluorescent, so we can nicely follow them after treatment, and here we show that we're able to specifically decolonize the antibiotic-resistant bugs. So it's not as efficient as in vitro, but it also works in a more complex environment like the skin. I also want to mention that this work was pursued by a startup company called Eligo Bioscience, of which I'm a co-founder, and that's hosted at the Institut Pasteur. So basically, we had this picture right now where Cas9 is very useful to make genome editing in your carrots. In bacteria, it tends to rather kill the bacteria, and you could ask the question, why, and is this really always the case? Will CRISPR really always kill the bacteria, or does it kill bacteria? And not your carrots. So some experiments that we've done recently to address this question is simply to do a very easy assay where we still have Cas9 in the cell and we transform a CRISPR that we program to target many different positions across the genome of E. coli. And then we see what happens. We basically just count how many colonies do we get when we do this transformation. And this is the result that we got that was very surprising at first, because we really expected the CRISPR to kill E. coli at any target position, but what we realized is that some target positions we don't recover colonies, meaning that the CRISPR killed the cells, but other positions we actually recover just as many colonies as a non-targeting CRISPR control. And what we saw, so that was interesting, is that for some of them, the one where I put a star, the colonies that we recover are actually some pretty small colonies, indicating that the bacteria here are pretty stressed. And this points to the ideas that probably, first, not all targets are equal, some targets are better than others, and some target position might be tolerated by E. coli, and E. coli might be able to constantly repair the breaks introduced by Cas9 at these targets. And we could confirm this idea by simply repeating this experiment in a RecA mutant, so I think you might not see very well the white bar here, but that's a control if we don't target anything. Of course we cover a lot of transformants, but all of those, there is no white bar, basically, because if you're in a mutant of the repair systems of E. coli, RecA, then all these targets are going to kill the cells. But that still doesn't tell us why does it kill the cell when it actually does. And the idea is that it probably does so because when CRISPR is going to cut in the genome, it's going to cut all copies of the genome at the same position at the same time. And it turns out that most bacteria strictly rely on homologous recombination to repair breaks. So if you don't have a sister chromosome to do homologous recombination with, then basically you're dead. And we believe that this is the kind of damage that CRISPR does, and to demonstrate this what we did is very simply putting a template for recombination on a plasmid that is not going to be cut by the CRISPR. And the idea is that if our hypothesis is right, this should rescue the cells. And then when we transform our CRISPR system that targets this like the gene here, what we see is that without the repair template we do see a lot of deaths when we target the gene. But no way if we add the repair template we see that we rescue a lot of the cells. This confirms that the cells die because they don't have a template for repair. Then what we also noticed is that whenever we transform this CRISPR system to target in the position, whether it kills or not, we were wondering if this could trigger mutation at the target. And because when it kills it usually does not kill all the bacteria. You might have some that survive. And this is the case for this, like the two targets here where you have some bacteria surviving. But here when we played this on the exgal plate that allows us to see if the like the gene is intact or not, we see that we recover about half blue colonies and half white colonies. So that suggests that the CRISPR actually led to the introduction of mutation at this target position. And when we checked what kind of mutation we obtained, what we see is that we obtain a lot of very large deletions around the targets. Deletions up to about 40 kb. And in most cases, what was interesting is that these deletions involve these rep elements of E. coli. So I had no idea what these rep elements were before we studied these mutants. But those are basically repeat elements. So there are a lot of them in the genome of E. coli. I don't remember the exact number between 200 and 300. They are scattered throughout the genome. And basically the idea is that they have enough homology between them that if you introduce the Bostrand break, they can recombine together to repair the break and the cells might survive. If, of course, there were no essential genes in the deleted region. So then you can try to make the hypothesis. Why do eukaryotic cells survive CRISPR breaks so much better than bacteria? And one idea you could have is that, okay, bacteria, most bacteria, don't have non-homologous enjoining. But eukaryotic cells do. So maybe they are able to repair the CRISPR break with NHEJ, and that allows them to survive. So then we thought, okay, so let's try to introduce an NHEJ system in E. coli and see if that can rescue the cells. So that's what we did, and we took the NHEJ system from mycobacterium tuberculosis. So I told you that most bacteria don't have NHEJ. Actually there are a few bacteria species that do. And then we repeat the same experiment where we transform a CRISPR system that's going to cut in the chromosome. So in the control that doesn't target anything, we recover a lot of bacteria, but here, whether we have or we don't have the CRISPR system, the CRISPR system still kills most of the bacteria. So it does still kill most of the bacteria, but what we noticed is that when we played on this exgal plate that allows us to see there was a mutation introduced in like Z, we see that we start to recover more white colonies with the NHEJ system than without. And if we PCR the target position, we start to see small variation in size as a target position that are very typical of repairs made by NHEJ. And so we can map this small deletion and we see that when we have NHEJ present in the cell, we obtain a lot of deletion ranging from six bases to about 300 bases that are very variable in size but that are still very different from the big deletions we obtain without NHEJ in the cells. And something that was interesting is that these deletions were always very variable on one side but very... always within basically three bases on the other side. And that basically led to the idea that probably Cas9 might remain bound more strongly to this end of the break and might protect it from nucleases which might not be the case of the other side explaining the asymmetry in the repair. So, to sum up this part, basically, Cas9, when you introduce bacteria, if it's able to cut all copies of the chromosome at the same position at the same time as in the cell's cannodomalgoscommonation, it's going to die. But some bacteria might be able to survive by making, for instance, big deletions. And the other thing that we found out is that at least with the setup we had NHEJs not able to rescue the cells. It can make some repairs but very, very low efficiency. And efficiency is basically even too low to be able to use it really as a tool to introduce indels in bacteria. So now I want to talk about a slightly different thing you can do with this CRISPR system, which is to use the catalytic dead mutant of the Cas9 protein. And that's a very interesting mutant because it's still able to find its target position and bind to it very strongly, but it doesn't cut anymore. And the thing is that it buys strongly enough, for instance, to block transcription and so that you can use it to silence gene. So that's something that we demonstrated, the group of Sandley Chi, also published a very similar thing to this work. What you see here is just a GFP reporter gene and this is a relative fluorescence measure and then it's just a position we target with this guide RNAs, either within the gene or in the promoter of the gene and the fluorescent level we obtain. So you see that if you target in the promoter sequence you obtain very strong repression, in some cases really barely detectable, expression of GFP. If you target inside of the gene you can still get very strong repression up to 100-fold repression and it really depends on the orientation of the target in the gene. If you target the coding strand you can get good repression, if you target the template strand you can only have much weaker repression. And to understand what's happening here you can do a simple northern blot to see this is basically in the control of the full-length transcript of the GFP. If you target the promoter region you don't see any transcripts so basically you block the initiation of transcription. If you target within the gene you start to see a smaller transcript here appearing that the size of that transcript matches exactly what you would expect if the transcript is just stopped as a Cas9 binding position and if you target now the coding strand here you don't see any more the full-length transcript so you have very good repression and you produce this shorter transcript here. So basically this D-Cas9 protein is either to block initiation of transcription or to block even the elongation of transcription and it can be very easily reprogrammed to bind any place you want in the genome so you can use that to silence genes in a very convenient way. Then you can start to do even more things with this D-Cas9. You can fuse protein domains to it to introduce functions at a specific position in the genome. And here for instance what we did is to fuse it with the omega subunit of the RNA polymerase in order to turn it into a transcription activator. And then we target it in upstream of weak promoter sequence and we should see activation of the downstream gene. And so we targeted it to a different position upstream of the weak promoter, either on one strand or on the other strand, and we see that at a very specific distance from the promoter and in the right orientation we can get up to a 23-fold activation of the GFP. So I also just want to mention work by other groups related to these very interesting applications of D-Cas9, for instance including in imaging if you fuse a GFP to D-Cas9 and here for instance we targeted it to bind to telomere sequences. You can see here in this picture every little dot is a different telomere in the cells and here you don't have to fix the cell or anything, the cells are live so you can follow the dynamics of different loci in the cell and it's a very powerful tool for people studying the architecture chromosome organization and dynamics. So the last application I want to talk about is something that we're really focusing at the moment is to develop screens, functional screens based on CRISPR. And this comes from the fact that you can so easily reprogram this CRISPR system that you can construct libraries of CRISPR or of guide RNAs that are going for instance to target up to 10 to the 5 position here in the genome of E. coli and then we can introduce this library of guide RNAs in our cells and then you can perform a functional assay of interest. You might be interested in studying sub-MIC antibiotic concentrations, some different stresses you might want to combine it with some query gene knockout, etc. And the readout of this experiment is going to be done through deep sequencing. What you do is you sequence the CRISPR library both before you perform the experiment and after you perform the experiment and by comparing the proportion of each guide RNA changes during the course of the evolution you can basically compute the thickness of each mutant in the population. In series there are two ways you can try to perform these CRISPR screens. You can either think of doing them using CAS9 to introduce knockouts in combination with NHEJ and that's something that people have already published and are doing in eukaryotic systems where you have this endogenous NHEJ in the cells and these screens work very well. And you can also think about doing screens based on D-CAS9 where here you're not going to introduce knockouts but to knock down and silence the target position. And I explained to you that NHEJ repair at least in E. coli is very inefficient so we cannot do this but we can do this type of screens. But ideally we might be interested in both and both screens have actually different properties that might be complementary. So for D-CAS9 screens you can think that you would only have partial silence and polar effect but you actually know what the polar effect are. For instance in an operon if you knock down target a gene early in the operon it will also block expression of older downstream genes. A big advantage of D-CAS9 knockdown is that it's actually both inducible and reversible. So knockouts with CAS9 and NHEJ here the good thing is that you can have complete deletion but the problem is that the edits made are going to be quite unpredictable and you don't really understand what the polar effects will be. I just want to give you a flavor of what kind of data we can get with this type of screen and this is just a very raw data showing you the number of reads we get for each guide in the library both before the experiment and after the experiment. So this is a control where we don't induce the expression of D-CAS9 so everything falls more or less nicely along the diagonal meaning that the number of guides in the library didn't change between before and after the experiment. But now if you start to induce the expression of D-CAS9 you see a lot of points either going down or points going up and points going down here means that basically those guides provide a fitness defect to the cells in the population and when guides going up we provide a fitness advantage to guides in the population. And so then you can of course focus on some specific points and try to see what's going on and here for instance it's the exact same figure except that I just kept the targets in this mirror A gene that's an essential gene in E. coli involved in cell wall synthesis and what you see is that the control experiment everything is still nicely along the diagonal but now when you induce you see that here's all these pink points basically go down and the green one, blue one stays there and what these points highlight is these blue points actually target the template strain and if you remember from a few slides before I explained that targeting the template strain gets you only very weak repression but if you target the coding strain to that those are these red points here you have very good repression and you see depletion of this guide in the library which is expected because it's an essential gene. So like this you can very easily figure out what are all the essential genes in the cells and of course if you do that in a specific condition where you want to do functional screens you can also get more interesting information something that we think is also very interesting and that's linked to the fact that this is something that you can induce at a precise moment you can actually follow the evolution of each guide over time and here we sequence for instance this library after 20 generations of culture or after 40 generations of culture and this is targets in the mercy or secd gene and here I kept the same color code the blue ones target the coding strain so good repression, the red one target template strain, weak repression so you see that all the red lines are basically straight horizontal lines and the blue lines here both these genes are essential genes but it turns out that if you target mercy the guides are depleted much much faster in the population than if you target secd so both genes are essential but maybe you can start to see some degrees that's what I want to say so where we're going with this basically is that this single knockdown or knockout screens are actually very cool and in some ways are actually also pretty similar to what people have been doing already in the past for instance with transposon, tag metagenesis TN-seq kind of approaches but what we think we can do that's very exciting with this is now go after genetic interaction by doing multiple knockdown simultaneously in the cells and so that's something that we're really excited in doing right now and we're just starting to be able to construct these double knockdown screens and study them in the same way so with this I'd like to thank people in my group so we have a very nice lab space at the institute Pasteur, we just started a year and a half ago and some of my collaborators and funding, thank you very much we have time for a couple questions about the curiosity what's known about CRISPR and denocoxide denocoxide constitute a film bacteria that are able to sustain thousands of DNA double grades and repair them they typically have no less than four copies of the genome per cell so I was wondering whether some studies have been done on these lines it would be super easy to check whether denocoxide tend to have or not CRISPR system it turns out that's something like 50% of bacteria have CRISPR systems so there is this very nice database CRISPR database on lines that we could look into I never checked for denocoxide any questions? a few slides back in your mercy data for the different generation times some of the lines on the graph went up and 40 generations is that just an artifact of the way you're measuring or is there some recovery? so we're not quite sure yet we're just starting to look into this type of things so the true answer is I don't know I suspect that it might actually be some mutant guide RNAs that might survive and then have a regular fitness or recover a bit or something like that but I honestly I'm not quite sure so you showed that the CRISPR is killing these cells unless you provide a template for repair that is which means it would be an even better mutagenesis to a targeted insertion so you're putting your finger on a very good point and the answer is that we first believed so as well and it actually turns out not to be the case we were lucky to have this experiment where actually the template uses cells efficiently but it turns out that this works only if what you introduce is a point mutation and only for a specific homology the length of your homology arms needs to be very specific and as soon as you want to for instance introduce a small deletion efficiency drops dramatically and etc so we don't understand really why yet and this is something that we're investigating as well David, I was really nice but I think I'm just not quite understanding the CRISPR screen so it's similar to TM-seq or transposon-seq in some ways so one I was wondering if you had a comparing contrast in methods and then the second is how are you preventing death in this case so here it's using the D-Cas9 so we're not cutting the genome it's just going to silence expression of the target positions but you're perfectly right that it's very interesting to compare this type of method with TM-seq and we're actually doing that at the moment so we're conducting in parallel a D-Cas9 and a TM-seq screen and we're going to sequence them and analyze the results very soon following up on that how do you construct the library do you design your 20 base pair RNAs and then or is that random so you can design them so now you have these companies including probably the ones that is most commonly used is oligo arrays so I mean I don't have any shares in them whatsoever just think the ones that we're using they provide, you can order a pool of oligo and you can precisely determine the sequence of up to 10 to 5 different oligos you want to be in this pool and they just generate it for you so the technologies that they're using are very similar to the technology used to generate micro arrays except that then they just cleave it off the chip and recover it in a pool and you get that and then you have this pool of oligo and then you just use it in a standard cloning procedure so what's the role of the PAM sequence you showed this one promoter where you had nicely spaced target RNAs but then what is so you absolutely need a PAM motif to be able to Cas9 to bind so Cas9 is really first looking for the PAM motif there is no PAM motif it's not going to bind and so in the experiment where I showed that for instance we had tested many different positions along the promoter we specifically engineered a promoter to contain many PAM motifs so that we can test precisely different positions would that be a problem if you want to target a gene and then you would want to target the Cas1 or the minus 10 or something but there is no PAM so the answer to that is that you actually have quite a broad range of position you can target even if what you want to block is the initiation of transcription you see here it's a 100 base pair range where you are still very good blocking the initiation so GG motif is very frequent it's randomly every 8 bases one last question you were talking about immunizing with that CRISPR when you are treating the population to kill the bad guys how long does that immunization last so it should last as long as the cells keep the CRISPR system then how long will they keep it will depend on the stability of the plasmids that you inject and many other things so that we haven't really investigated the cells themselves how long do they persist the cells which keep active your resistance the cells were I'm not sure I understand which cells you are talking about well previous question was about the period of the existence of the resistance and you refer this period to the cell survival so cells how long do they survive in your system I mean if they are not killed by the CRISPR systems they are going to survive indefinitely there is apparently one question over here just a question about the natural CRISPR system how many repeats does the natural bacterium maintain it how big is it so that's very very very variable it's basically from 1 to 600 and it depends on different bacterial species and this is really not understood why some species of bacteria tend to have shorter CRISPR while some have larger CRISPR arrays I cannot really speak to that it's really still a mystery we need to ask more questions during the discussion thank you David thanks