GENE EDITING EXPLAINED! - A comprehensive guide to the principles, methods and technologies!

Hello and welcome to this long lecture on genome editing. This lecture is mostly intended for university students researchers clinicians and and the media who are quite new to genome editing and I want a good background in it. I cover things like zinc finger nucleases and TALENs all the way through to CRISPR and the latest CRISPR methods like prime editing. So it's quite a long lecture. If you look in the description, you'll see timestamps for each section - which you can skip to or play again. Also I've broken this lecture down into smaller pieces if you're just interested in a specific thing like CRISPR or prime editing or something like that. On the slides you'll see numbers in circles in the bottom right hand corner. Those are for the references that I've used to create this lecture so some of the graphics that you see come from those papers and reviews. I'm mostly giving you that list of references as the best starting point that I can see for you to continue your own reading. In the description you'll see all of those references with links to those papers and journal sites. Be aware that some of those will not be open access, so you'll need your institutional licenses to see those articles unless you want to pay. I hope you enjoy it and let's get started! In this lecture I'm going to cover the following sections. Each of these sections is timestamped in the description so you can jump straight to those sections or you can go back and watch a particular section again later on if you want to just revise what you've heard alternately to these sections will get broken up into very short videos later if you've just got an interest in one particular area so before I get into details on the different kinds of nucleases in the introduction i'll just revise hopefully what you already know about double strand DNA break repair and we'll look at the kinds of genome editing outcomes that you could achieve using sequence specific nucleases then I'll move on to describing zinc finger nucleases and the pros and cons of those and then I'll move on to talents and then we'll move on to CRISPR and before we get into how CRISPR is used for genome editing we'll describe the natural adaptive immune systems that CRISPR belongs to because their debt explains some of the issues of CRISPR specificity then we'll get into what CRISPR cash 9 is and how it's used and then the longest section will be spent on CRISPR specificity and the the advanced methods that are out there to overcome the limitations of CRISPR and in the last section we'll talk about precision editing where you want to make very specific edits and you need a absolute control over what you're doing and there's two areas of that homology directed repair and prime editing genome editing describes the ability to modify the DNA genomes of living organisms in controllable ways so that's modifying in living cells in living species rather than taking a piece of DNA out modifying it with some enzymes growing it up in bacteria then put it back again rats limb modifying in living cells so in the case of humans in human cells genome editing is critical to the study of genes and gene regulate regulatory element functions so rather than to understand the gene you take it out of a cell and you study it somewhere else or you're still studying a gene regulatory element like a promoter or an enhancer and you put that into some kind of reporter assay with genome editing you can study a gene or a gene regulatory element in its natural context it's chromosomal context so you can delete it mutate it exchange it modify it however you want to in its natural context this is very powerful genome editing also paves the way for the creation of models for human disease so you can take the specific mutations that occur in the human population and put them into cell lines or into animal models to try and mimic the disease and study very specific molecular aspects of that disease and gain some insight maybe develop some drugs and test them out genome editing also obviously allows for the advance correction of genetic diseases so these are advanced gene therapies so rather than when you've got a patient who's got a defective gene rather than overlaying that with a normal gene and saw those cells in that patient expressed both the normal and the mutant gene you can now actually try and correct the mutant gene and get rid of the problems of that mutant gene genome editing also allows for very powerful improvements in the biotechnology and food production sectors so you could modify microbes or plants to make better products for you or be resistant to diseases or easier to handle whatever it is that's limiting in those industries and genome editing editing can now be applied should it be safe to do so prior to genome editing Amitha called gene targeting was used to modify genomes this is what describes the genetic replacement based on cellular homologous recombination machinery so this is obviously used in yeast an awful lot it can be used in mammalian and Brianna Steph cells and in b-cells but in general it's very inefficient in most cell types of interest and for all those applications list listed above gene targeting just is just not useful in most situations so genome editing involves a variety of powerful synthetic biology strategies that have been developed to create sequence specific DNA endonucleases so these are enzymes that cut a specific DNA sequence of your choosing and you you place those under nucleuses to create the cut exactly where you want to so if you're trying to knock out a gene it's going to cut at your specific gene this then stimulates cellular DNA repair and DNA repair in fixing that break will either make mistakes or you can persuade it and trick it into making specific changes so the combination of your knowledge of DNA repair and your ability to create these enzymes that will cut where you want them to you can then modify a target genetic locus precisely and quickly this is a very very quick method so there are in three generations of nucleosis that have been created for the using gene of for their use in genome editing the first version involves ink finger nucleases so I shall take you through those then a better approach came along which are the tale nucleus or talons but in this most modern ear in the last few years CRISPR really has become the easy easiest and most applicable method out there so I'll take you through each of these three things it's important to consider zinc finger nucleases and tell nucleuses and understand them because those are the ones that have made it through two clinical trials and a lot of biotech applications so it's important to know what they actually are on what their value is so looking at double strand break repair pathways and so I'm mostly going to focus the talk on mammalian systems but a lot of this is true across eukaryotes the most dominant pathway of DNA break repair is called non-homologous end joining it's a very fast set of enzymatic processes which simply join breaks back together again and they ignore the sequence at those breaks so if you've got multiple breaks present potentially different ends will get joined to each other but also when joining the correct ends together it can do so in an error-prone way so on the left is a diagram here of the typical kinds of factors that recognize the breaks signal the breaks and then start to fill in any ends which aren't blunt and then ligate them together this kind of repair can be completed in minutes it's extremely IRA prone because no template copy is used and canonical non-homologous end joining can be accurate if the ends are our blunt and phosphorylated and clean it can join them back together the way they came apart and it's fine but if you're using one of these nucleases and genome editing it's gonna cut that site again because you've reinstated the target sequence and what will happen eventually is that non-homologous end joining will insert an extra base to facilitate the ligation of the ends and in which case you've got one base insertion you've changed the sequence that might inactivated gene alternative non homology an alternative enjoining pathways typically reset the five prime ends at the brakes so the two just that one strand back so five prime to three prime reception on what they're doing is exposing a short sequence a single standard sequence to look for local micro homology so it might just be a couple of a's or a couple of CS and it will join those two a couple of T's or a couple of g's on the on the other end and it will ligate through that so those will pair up and there may be bases that will need to be removed and so often you get these local deletions so collectively you get insertions or deletions from non-homologous end joining we call them in Del's for short and so the outcome of this kind of repair pathway of seemingly random so if you're trying to create a knock out of your favorite gene by targeting a nucleus at the beginning of the gene then this dominant non-homologous end joining pathways is excellent for creating all these seemingly random mutations and you'll get frameshift mutations in there and you'll get a loss of gene expression very very easily but actually when people look at lots and lots of targets and repeat genome editing in lots of cells they find that actually the kind of edits that occur from non-homologous end joining are not random and and to some degree that they're based on the local sequence context so you'll see at some sites you always get +1 insertions at other sites you always get deletions of a particular length because of the local micro homology a slower but accurate DNA break repair pathway is called homology directed repair and this is an enzymatic process that uses a in normal biology the sister chromatids template for accurate repair so officer this is normally associated with DNA replication so if you've got a stalled replication fork sometimes the system will will actually cut that fork create a DNA break and because you've got a local template available you can then copy that so a set of proteins that recognize those breaks reset so a key part of homology directed repair is a very extensive 5 prime to 3 prime reception at the break which exposes a long three prime overhang this is typically many kilobases in length this three prime break is then protected with protease like a replication protein a so they protect their single-stranded DNA binding proteins they protect these three prime ends they then get exchanged with route 51 this fantastic filament protein and in combination with the braket proteins these filaments are able to then invade and search a sister copy for a matching sequence and when it does so it base pairs up and a strand exchange occurs and so you get these Holliday junctions where you've got a crossover between one chromatid at another and then these are resolved and the gaps are filled in and so what you end up with is a perfect copy from one sister chromatid to another so this process overall can take over an hour depending on the sequence so it's quite slow it's limited to a very specific part of the cell cycle but potentially we could use this and genome editing to persuade the cell to make a specific change for us but instead of having a sister chromatid as the copy we would deliver a donor DNA and that donor could be a plasmid or a viral genome that we've introduced into cells and we do so in excess if we can and so the exchange of the sequence were interested in becomes the dominant event so a key thing about considering what's going to happen after you create a DNA break is that you've got these error-prone pathway and then you've got this accurate pathway and they're active at different times in the cell cycle so non-homologous end joining is is functional throughout the cell cycle certainly it's very dominant in g1 and s phase you know which is where most of your genome editing is going to be occurring HDR on the other hand is linked to replication so it's only active when DNA replication is occurring and throughout S phase but accumulates late in S phase and then in g2 particularly g2 is where a lot of these stalled fox are getting resolved so if your goal is to undertake destructive gene editing like creating a gene knockout then it's very easy because non-homologous end joining is dominant and you're going to get a lot of cells were which are going to have mutations which are useful to you in that the create frame shifts are now inactivate your gene if your goal is to undertake an accurate DNA edit so you doing gene correction or a specific mutation in a gene regulatory element you creating a specific model then somehow you need to deal with all the unwanted non-homologous end joining mutations and you've got a sift through all of this junk to find the cells which have the specific edits that you're interested in and this is a hidden problem of genome editing that a lot of people perhaps don't consider when they first get started so in this slide here I'm going to show you in very simplified terms the possible genome editing outcomes you can get just from creating DNA breaks using sequence specific nucleases so in these cartoons we've got two stands representing DNA and and then this this cut through here just shows a DNA break that we have introduced with our nucleus so if we just cut in a place of interest so that's the promoter early exons of a gene then we get insertions and deletions occur because of non-homologous end joining and so we can disrupt that genes function or that regulatory elements function if we provide a donor DNA at the same time so that could be a linear piece of DNA or it could be a plasmid which we are going to cut at either end with our nucleases at the same time as cutting Guard genomic target then without any special homology arms or special processes so this is just using non-homologous end joining you will get insertion of your sequence of interest it's nearly always a trans gene so I've colored it green here in cases it could be a GFP trans gene say and you will get insertion at that target site but the key thing is that you've got no control over the orientation of that insertion it can go in either way and also the ends either side of the insertion will be will have these in Dells on them so so you may lose all gained sequences so if it's important for you to for example in the case of perhaps you wanted to Tagg the c-terminus over of a gene and created gfp fusion or a fusion of another long tag in that case your insertion needs to be the correct orientation and it needs to be in the correct open reading frame as your gene and with this kind of approach very few of your cells are going to achieve that but potentially for example in the case of a GFP fusion you can you can easily select for those cells so if it's very inefficient knocking and very inefficient at creating the correct knocking those cells are still bright green so you'll be able to sort them out through flow cytometry for example so it depends on your application sometimes you can just do this very simple quick way alternatively you could make quite large genomic rearrangements just by cutting twice so let's say you've got a gene locus which you want to invert for some reason you can just cut either side of it and the cell when repairing those breaks may put that insert back in to its normal location but do so the wrong way around and obviously alternatively probably more dominantly you will lose that fragment and you'll end up with a deletion so that's just simply cutting with two nucleuses at the same time no other trips so small percentage of cells will make these big edits for you and we have done this we've we've deleted out several genes from a gene cluster and just by cutting twice and not doing anything else and you get a decent number of cells that have made that edit for you and you can do some experiments on them alternatively you may want to persuade the cell to use homology directed repair for you so for example let's say we want to knock in a gene at a specific location so in this case we're going to cut the target locus with our nucleus so that's the target locus here and then you'll have a donor fragment which contains the sequence you want to in but importantly that fragment has to be flanked by homology arms which are the same sequences as what's found at that either side of the break at the genomic target so this sequence here would be placed here this sequence here would be placed there and Stratton's strand exchange would be used would occur using these homology arms and in the case of sort of traditional HDR where you're knocking in large gene fragments you need at least 750 bases in lens for that homology arm and the longer the better alternatively you may just want to make precise edits so let's say you've got a defective gene and you're wanting to correct it with the real gene you would cut as close as possible to the way you want the change to occur and then you'd have a donor fragment again like you've seen on the left here with these homology arms so if you wanted to exchange a large gene sequence then you do need these long homology arms but let's say you're just making an edit of 1 2 or 3 bases as is typical for most human genetic diseases then it's been shown that your donor fragment can actually be very short indeed so actually no people use oligonucleotides so very short single-stranded DNA fragments you only need one strand which will have the edit which may be 1 2 or 3 bases and then that can just be flanked by 40 bases for zero basis of homology arm and that's sufficient to get an exchange to occur and so this has now become was until very recently the main way that you would make precise edits of a gene so in understanding zinc finger nucleases were perhaps need to look further back to what other enzymes cut DNA sequences specific DNA sequences and whether we could modify them or not so obviously restriction enzymes exist in lots of prokaryotes and we have a large catalogue of these enzymes which will cut a specific sequence and the idea was to try and modify those enzymes so by understanding them created crystal structures of them doing lots of mutagenesis on them potentially we can take an enzyme that cuts sequence X and persuade it to cut sequence Y specifically well actually most restriction enzymes cannot be readily adapted to cleave new sequences in general have got very short recognition sequences which is not helpful that they'll cut many many many times in the genome and not cut to unique sequences and when you persuade them to cut another sequence they're cut cut them very very weakly so really that wasn't a route to develop a genome editing however there's a class of enzymes the type 2's restriction enzymes which are quite interesting an example is the enzyme Fock one fok one and these type 2's restriction enzymes have separate cleavage and DNA recognition domains and importantly the cleavage domain has no specificity and can work on its own it just needs to be recruited to the DNA so for example here in the case of huaquan which functions as a dimer that's an important feature of Fock one is a dimeric protein so two molecules of the same protein in green and in blue here and this domain is the DNA recognition domain for combines this particular sequence GG 80 g and then the cleavage domain is here and a key feature would tattooist restriction enzymes they tend to bind in one place and cut a specific distance away and the sequence that they cut it doesn't matter it can cut up all the sequences here so type 2s reduce restriction enzymes are really useful in synthetic biology for lots of different techniques but in the case of genome editing we can take this cleavage domain which will only function as a dimer and add it on to another DNA binding domain so we've got the cleavage parts we need the DNA binding part and the concept was to fuse this catalytic domain of Fock one onto a different DNA recognition domain and when you look through the genomes of of mammals the most common DNA binding domain out there is the zinc finger domain so for example there's nearly 1500 human genes that have got zinc finger motifs and there's a large family of genes that contain a certain kind of zinc finger called the sis to hiss to offs or C to H to zinc finger domain and these domains which I'll show you on the next slide I've got a very simple Beta Beta Alpha fold and follow a very common amino acid motif so these were discovered in 1985 and zinc finger domains were heavily studied in the 1990s and from this very substantial amount of work a DNA recognition comeup code emerged from comparing the the protein amino acid sequence of zinc finger domains and the DNA but DNA sequence that they specifically bound to so if we look here i'm here's a crystal structure of zinc finger proteins bound to DNA and so they bind to normal b DNA and the zinc finger domains interdigitate or poke into the major groove of DNA and all these linked finger domains look like so you've got an alpha helix and to be two sheets here and it's all held together with a molecule of zinc hence the name zinc finger and then there are these specific residues on the Alpha helix and on this beta tone here which are the ones which bind to DNA and depending on on which amino acid in as you can see in this table is in which position in a triplet here will determine which nucleotide is bound and so you know you could actually look up through this table and and create on paper which ideal proteins would bind to your sequence of interest and indeed this is possible so you can then create zinc finger nucleases so zinc finger nucleases are where you've got an array of zinc finger proteins usually three or four each zinc finger binds to three bases so in the case of four zinc fingers is by mister twelve bases now as I said before fuckwad only functions as a dimer so you need to make two of these proteins so you've got to assemble two proteins so R for zinc fingers with a fuc one cleavage domain at the c-terminus and Fuquan needs six bases of space to bind to and cleave so your target sites are these twelve bases there's a gap and then these twelve bases and there are tools to help you design these so key features of zinc finger nucleases is that the cleavage domain has no sequence specificity so you can cut what whatever you want to you do need to make two zfn proteins but this does increase sequence specificity so you've now got 24 bases of specificity which is pretty good there are mutants of the Fock one domain that would votes which are obligate heterodimers so there's a left fuckwad mutant and there's a right fuck well mutant and what this means is that if one of these let's say this left ZF Xenophon protein bound at an off target site through its 12 bases it wouldn't be able to homodimer eyes with itself and therefore cut that off target site so by creating obligate heterodimers you you're really restricting the possibility of Xenophon's cutting off target you do still get some off target active activity but it's greatly reduced the four times in finger domains is about as good as it gets longer race generally don't work so you're looking at 24 bases of specificity at best assembling zinc finger arrays now is quite straightforward because there are wonderful synthetic biology tools out there to to join fragments of DNA together quickly in cells the real problem with zinc finger nucleases is the vast majority of zinc finger domains do not function very well when assembled together in these arrays but for reasons that are not fully understood so the best approach is to take a sort of mass action approach and so the company sangamore based in california and and you can buy products license from sigma they had found zinc finger arrays that function well and I think these are pairs of zinc fingers and they've got pretty much every possible combination of pairs and they found that these pairs are happy to be joined to each other so they can create sets of 4 zinc finger domains for most sequences very quickly on a robotic platform and therefore create functional a defense quite quickly so a lot of those function well but for ordinary groups assembling them themselves and labs it really was quite a painful process because the majority of that offends you made just didn't function anywhere near as well as this should have done on paper however if you've got a sink thing in you clears that binds well and is specific it's a it's a superb tool for genome editing there's nothing fundamentally wrong with zinc finger nucleases it's just that they're difficult to make if you don't have one of these huge assembly platforms available to you a key advantage of zinc finger nucleases over all the methods that follow which I describe in this lecture is that the fans are quite small so they're much easier to deliver into cells and tissues than some of the more advanced tools that are used later on so they certainly do have their place in genome editing although you'll find in the modern era not many people use them so cell phones were first successfully used for genome editing in 2003 they've been used in many species for a very wide variety vadik editing applications and a lot of the genome editing strategies and approaches that that we know news today were developed using zetas fans at first so some of the foundation papers that are out there all use their defenses so another reason for knowing about them there are over 20 zfn based and genetic therapies that are going through different stages of clinical trials so they have their place but they were slow to make there was a low chance of them functioning well particularly when created in in in most standard research laboratories and a big breakthrough came in 2009 when the DNA recognition code of a different protein domain family was solved and that leads us on to learning about talyn's in the next section transcription activator like effector proteins or tail proteins I find in plant pathogenic bacteria of the genus Xanthomonas and there are class of DNA binding proteins which have a predictable DNA specificity and these different effectively transcription factor genes that their role in Xanthomonas is to activate specific host plant genes to support the virulence of Xanthomonas and a key reason for these domains evolving is that they have a very simple code and they're very easy to mutate and adapt so if a plant changes the the sequence of the promoter of these important host genes to try and stop Xanthomonas functioning and infecting them then the Xanthomonas can change the sequence of their tail proteins very quickly so what do these proteins look like well they have these large repeats in them called repeat variable domains and they all have the very same sequence protein sequence with the exception of these two residues at position 12 and 13 out of this 34 amino acid repeat and the structure of the repeats looks like this which is helix-turn-helix motif and these two residues are what specify which individual base each domain binds to so the zinc finger domains were good because one domain just binds to three bases but the tails are even better because one domain binds to one base so you only need a library of four proteins and then you can assemble them in any order you want to to bind to any sequence you want to so all the natural tail proteins have very long arrays of these are V DS and all these are V DS are very happy very good neighbors with each other so they're very happy to be assembled and in to longer race to create proteins with longer more specific DNA binding sequences so here's the RVD code so depending on what these two amino acids are here in the turn and I will specify bind to an a base h DC base and so on and so on we've also got a couple here which are nonspecific or bind to G or na so these are incredibly powerful tools so in the lab you just got fragments of DNA that encode each one of these four domains or these other specific domains here and you just need to assemble them together into long arrays and then put them into a longer protein there's very nice crystal structures of them which just show how they bind in that they just again interdigitate into the major groove of DNA they don't modify the DNA sequence at all the DNA structure at all so you've got very nice B DNA very straight when you look down the end of it and you just sort of get like this propeller of Procope of protein and sticking out from it so very very elegant protein and you can use tail proteins much like you do with zinc fingers and but instead of having is that a fan now you have something called a tail nucleus or Talon and again like before you have left and right proteins and you create this dimer so fuck one or will only cleave as a dimer and the difference in the architecture of talyn's is that more space is required between the DNA binding domains for the fuck want to work so you have a gap of between 14 and 20 bases unfuck one cuts in the middle you've just got to assemble two proteins just like before but a key feature now is that you've got much more specificity in the complex you've got typically thirty to forty basis of specificity and the majority of tail nucleus proteins that you make do function very well I would say about 75% of the talents that you make will cleave very well and they're quite straightforward to assemble in the northern area lab Anna and I have done sort of myself the kinds of breaks it creates both zinc finger nucleases and talons create breaks with the four base five prime overhang and that's just what fuck Wan does it doesn't create a blunt end it creates a five-prime overhang and you could use those overhang sequences if you want to in some strategies but that does mean that those breaks are a bit more mutagenic because non-homologous end joining in general will get rid of that five prime overhang and so you'll end up with deletions so talons were first used for genome editing back in 2011 they've become widely used in many species for a whole variety of editing applications that can do everything that seda fans can do it takes one to two weeks to make them in a standard molecular biology lab using the tools that are out there I've seen here half of talons I've got good cleavage activity I think it's more than that's about 75% there's a wide the wild application of talents has led to many improvements to reduce off target activity so there's been new RV d--'s have been developed which we've got better based specificity there's mutations and the fuc one domain to make it function as an obligate heterodimer as I've mentioned there are other mutations that improve the cleavage activity of Fock one and there's been lots of changes to the architecture of talents too to make sure that it will only cleave when the space is a certain length which again improves its specificity because on off target where two proteins might bind by chance the the the spacings not the same so it's a very powerful robust technology for genome editing again there's really nothing wrong with talyn's whatsoever they definitely have their place and talents are being pursued to final applications and a lot of different strategies however like was there defense you've still got to make these engineered proteins there's a certain commitment to to making these engineered proteins testing them and getting them working before you do any genome editing and this is the reason why they're not used so much these days because there's an easier approach out there so in summary I think the way you could consider zinc finger nucleases and talons is that they represent versions 1 & 2 of genome editing these technologies are very well established they're proven we know what the limits of use are these enter nucleus is these engineered nucleases can be very active and very specific and the quite a small payload when you're considering how to deliver them to cells and tissues but there's a sort of barrier to entry for a lot of people which is that that initial protein engineering can be time-consuming and and expensive you need to make two proteins per target and only a fraction of what you make will will perform well and and perhaps a another issues that if you're wanting to create more than one break or edit at the same time so let's say you're trying to mutate two genes at once or you trying to delete a genomic region by cutting twice in those situations for 2 cups you'll need to make 4 proteins and deliver four proteins to the settle at the same time 3 cups you up to 6 proteins it becomes very difficult very quickly so what if there was an approach out there where you could avoid having to make engineered proteins and that's what we'll find out about when we look at CRISPR clustered regularly interspaced short palindromic repeats is a very long acronym that is abbreviated to CRISPR which is certainly a CRISPR way of talking about the adaptive immune system found in about 40% of all bacteria and 90% of archaea bacteria and it was discovered from the sequencing of lots of prokaryotic genomes and we you kept seeing these repetitive regions so in black here are all of these different repeats and in front of the repeats were all these open reading frames and so groups I'm guarding to studying what these CRISPR loss I were what were these clustered repeats all about and what it is is a system with which to detect bacterial bacteria phage genomes detect and cut them and also save fragments of those phage genomes into these repetitive arrays as a record of previous infection and so you can target them again so there's a bit of a chicken and egg situation goes on here but if the CRISPR casts system has identified a viral genome and cleaved it and therefore inactivated that phage infection and cut it up into small pieces some of those small pieces which are called spacers gets inserted into this CRISPR array so all these different colors here are fragments of phage genomes that this particular bacterium as-as-as cleaved before in its evolution this repeat array this CRISPR array is then transcribed as one long pre CRISPR transcript the polymerase 2 transcript which has got all these repeats in it those repeats well each one of those repeats binds this other tracer RNA to form a double-stranded DNA sequence here and then RNA s3 then binds to this duplex of RNAs and Cleaves them so each one of these spacer and repeat regions called a crisper RNA gets separated from this pre crispr RNA transcript so you end up with lots and lots of crispr RNA tracer RNA fragments these are then bound by caste 9 and caste 9 is an RNA guided DNA and a nuclease so these double stranded RNAs bind to caste 9 and program it they activate it for search in the genome cast 9 searches the genome for short DNA sequences called pro to prote of space at adjacent motifs or Pam sequences so each one of these spaces that we talked about before these phage sequences these spaces when present in the crisper RNA is now called a proto spacer and cast line needs to find a sequence in the genome called the pam which centers it it allows it to sit on to DNA cast and then unwinds the strands and if there's a match between the proto space so it has in its loaded crispr RNA if that matches the genomic target then the unwinding continues on and cast iron will then cleave that target so we can look at this in more detail again where we're looking at the the natural CRISPR system and this is a class two CRISPR system so we have cast nine and it loads up this duplex of RNAs of the crispr RNA which contains the proto spacer which is twenty bases which what it is supposed to match a phage genome and then that's paired up with the tracer RNA calf nine then searches the genome for short pam sequences these are often just three years old bases long and it then unwinds the strands at these pam sequences and starts to see whether there's a match between the crispr RNA and the genomic target if there is this unwinding continues this bubble gets bigger and when it reaches a full length and you've got this long our loop forms or an hour loop describes this RNA DNA duplex here and cast lands holding the the other DNA stand separately when it's got far enough then the cleavage domains of cast nine are activated and the cast iron systems used in most laboratories create a double strand break which is blunt ended and it cuts three bases away from that pam sequence so very predictable cleavage so in 2012 Jennifer dude in this lab and in one world shop NCA showed that cast nine from the bacterium streptococcus pyogenes is a programmable RNA guided DNA and a nucleus very quickly the big synthetic biology labs out there jumped on this and modified it in quite a simple but powerful way to turn it into a tool to allow anybody out there to use short guide RNAs to edit their genome of interest so in the tool that we used in labs the CRISPR class 9 system now only has one RNA that's called a single guide RNA and it combines one end of a crispr RNA with the other end of a tracer RNA so these RNAs are typically 98 to 100 bases long at the five prime end the 20 bases are the pro 2 spacer that Maps your genomic target of interest and then you have the structural RNA that binds to and programs cast 9 and that's all you need you need cast iron protein and this short RNA and so you can very quickly cut your target of interest there is nothing more involved it's easy to multiplex Plex this so you could cut to many targets all at once or you could create a large library of guide RNAs so the only difficulty is expressing and delivering those guide RNAs into cells at the same time as cast 9 just to home in on this a little bit more there are two nucleus domains in cast 9 once called the H&H domain which cuts the target strand or the non pam strand and then there's a rough sea domain that cuts the pam strand or you can call it the non-target stand because it's that's not the one bound by the guy Darren E and they cut pretty much the same time the rough seiderman does indeed cut first but what you end up with is a blunt double strand break that's three bases away from your pan so to perform a crisper experiments very straightforward and you've got lots of options about how you would deliver crisper into cells and the choice of these is very much based on how easy it is to get started the amount of money it might cost you to get going but then also what are the limitations of the cell type use that you are using or what specific experiment are you trying to complete so the most simple way is to use plasmids that Express cast 9 and the guide RNA and you just deliver 2 plasmids into your cells of interest sometimes you can put the guide RNA expression on the same plasma this cast 9 and you just delivering one plasmid that's the simplest way to get going but you need a way of delivering plasmids into cells efficiently and for a lot of primary cells that is not so straightforward another approach is to make cast 9 messenger RNA in cell in the lab in the tube and make the guide RNA in the lab as well so through in vitro transcription and then you deliver those to RNAs into cells and that's got the advantage of these RNAs getting to work very quickly and in some situations they're easier to deliver than plasmids another approach is to use recombinant cast 9 protein that's been made in a laboratory and you've guided RNA that's been synthesized or transcribed in the lab you mix the RNA in the protein together and tube and then you deliver that Rabb or nuclear protein into cells and this is probably the most potent way of doing CRISPR it's the most active way and finally you could also use lentiviruses to deliver a cast nine expression vector and RNA expression vectors and the reason for using lentivirus is that you could have a library of different lenta viral genomes that encode lots and lots of different guide RNAs so you could have a whole genome a whole gene library or you could have a library of guides against every kinase that sit there in the genome and then you could do a mass CRISPR experiment and select for cells with specific properties of interest and then those cells that you selected you then sequence the Lenti viral genome to see which guide RNAs were present after your experiment and so now you know which genes might be important in your in your particular process that you're interested in so very powerful tool for for performing screens and an important thing to consider here is the level of castellón expression and how long it takes to become expressed and how long it lasts is very different depending on the different ways that you deliver your CRISPR into cells so if you use the sort of standard approach of using plasmids or let's say a viral vector to deliver a CRISPR expression vector and a guide expression vector into cells then it takes very many hours before that DNA is made its way into the nucleus it's been transcribed the messenger RNA forecast line has been exported to endoplasmic reticulum translated cast line is folded imported into nucleus and then accumulates that takes many hours so it's often 16 to 24 hours before you've got maximal last nine levels and then that plasmid does often go into cells and quite high copy number and sticks around for quite a while even though the cells are dividing they're still enough plasmid air to to make casts nine and so the CRISPR may go on for several days if you use messenger RNA it gets to work quite quickly so it can get translated and cassiejenkins can get to work in just a few hours another advantage of messenger RNA is you can really sort of titrate how much of it is you you're using in cells so you can prevent over expression of casts 9 which is a problem for off target activity of cast 9 and its expression at last for a shorter period than than what you will get with past mood DNA and then if you deliver protein into cells and obviously it gets to work straight away and it's the shortest-lived method of cast 9 delivery so that cast 9 protein cannot be expressed again because there's no messenger RNA and that protein will just degrade and dilute as the cells divide so depending on these sorts of four main approaches of delivering CRISPR into cells you've got sort of different pros and cons which you can sort of read through later but in principle if you deliver plasmids into cells this is also the cheapest and simplest ways of doing CRISPR and it's quite straightforward to express and deliver for guide RNAs into cells at the same time but depending on the cell you're using the transaction might be quite difficult in all these situation plasmid situations the cast 9 expression will persist for quite a long period of time and off targets mutagenesis by cast 9 will be at its highest with this kind of approach if I just jump down to the protein approach so using Cassadine protein which and typically you would be purchasing from a company so that will cost you money so you've got to buy that Cassadine protein and also you've got to prepare your guide RNAs through transcription in the lab or you're going to spend money buying those but the advantages of the RNP approach is that the transactions very efficient it's suitable for doing small screens and you've got the lowest level of off-target activity because it cost - is short-lived and you can titrate in exactly how much RMP you need just to get your target to be cleaved and don't use anymore cast 9 than that and then the lentiviral approach obviously is suitable for these large screens and and it can be lentiviruses can be a better way of delivering their CRISPR system into certain kinds of cells particularly non-dividing cells so just to point out that when you want to express guide RNAs and cells from a DNA template the way to do this is to use RNA polymerase 3 promoters and these promoters are very useful because they're quite short they've got very high activity and importantly they start at a defined base so for example the u6 and 7sk promoters started to G base the h1 promoter starts at an a base this base here will be the first base of your pro 2 spacer so you can only express guide RNAs that begin with a G or an A and then another key feature of polymerase 3 promoters is that they are terminated very easily with a row of T's so five T's or more terminates the transcription so you can easily get these short transcripts that start at a defined base and are a defined length and you can express them in high levels so how do we measure mutagenesis after performing CRISPR how do we know that the CRISPR has worked so all applications generally involve performing a genomic PCR of your target so you got oligonucleotide primers either side of your genomic target and you amplify that from genomic DNA from cells your parental cells before you perform CRISPR and then from your crisper treated cells and then you've got different ways of seeing whether mutations have occurred following Cassadine cleavage and DNA repair so one way is the surveyor assay and the surveyor I say involves performing those genomic PCRs of your parental cells or wild-type cells and then of your crisper treated cells and the key trick in this assay is that these PCR products are melted so they heat it up and then cooled down very quickly and so the all those different DNA strands don't necessarily pair up with the Strand it came with the the sister strand they came from so if you've got a mixture of wild-type and mutant sequences then you'll end up with mismatches between the annealed strands you'll get these little bubbles in the DNA and then you can cut those bubbles with an enzyme like t7 and a nucleus one which cuts mismatches within a double stranded DNA sequence so it'll cut any mismatched fragments into into pieces and then you can run these digestion products out on an agarose gel and through looking at the digestion pattern if you've got fragments of fallouts then that's because CRISPR has has occurred cation is bound cleaved and every Pro and repair has happened and the sequence has changed so this is a very easy universal assay you don't need to have a specific asset designed for a specific target but I think one thing you can often find when performing these surveyor assays is that you can get nonspecific cutting of your PCR products dependent on how clean they are how you prepare things in the lab and one thing certainly about the surveyor I say is that it does underreport mutagenesis when you compare it to other more sensitive methods but it is quite a quick way of screening whether the CRISPR is working on whether one guide is better than another for example so it's a commonly used assay another approach which is far more sensitive as the RFLPs a so the same as a but as I've described before so you're doing a genomic PCR of your parental cells and of your CRISPR treated cells but in this case you're going to digest your PCR fragments and you're not doing this melting reeling thing you just taking the PCR fragments and digesting them with the restriction enzyme which cuts your CRISPR target so when you're designing your guide RNAs you are looking at what restriction sites overlap that site that's three bases away from the Pam for that particular guide that restriction enzyme will obviously cut the wild-type PCR fragment but if CRISPR has occurred and it's changed that sequence in any way then that restriction enzyme site will then be lost and so that fragment won't occur and so by comparing the digestion pattern of your parental cells and your CRISPR treated cells you'll be able to see whether any restriction sites have been lost therefore whether Cassadine has cut and led to mutagenesis this is a really easy to perform I say and it's highly sensitive any base change at all will inactivate a restriction site and change the digest and pattern which you can then quantify so this is the most sensitive assay I think that's out there in terms of a sort of a gel based assay however you do need to design a specific RFLPs a for each target that you're going to do so if let's say you wanted to compare the performance of four different guides then you will need to test four different restriction enzyme digestion patterns so it can get quite complicated and quite hard work pretty quickly also the interpretation of these gels can can be difficult for some it's something that I found so the easiest assay out there that everyone's pretty much everyone's using is called the tied assay and so again this is just a genomic PCR of your parental cells and a genomic PCR of your CRISPR treated cells and you don't do any nucleus digestion or any agarose gels you just simply send them for Sanger sequencing so normally you should get to put this nice clean sequence if cells are treated with CRISPR then obviously at some point close to your Pam sequence the sequence is going to fall apart and become seemingly random and because you've got all those different insertions and deletions and you often have to tell your secrets in company that you're expecting the sequence to sort of degrade at a particular position and that this is an experiment and you want that sequence because often the sequence company will think they've done a bad job and won't send you this chromatogram and we'll just sequence your your fragment again and again and again and again so you need to tell them that you're expecting it to look bad you then take the two chromatograms here your parental and your CRISPR treated one and you upload them at the tide website the nki in Amsterdam so this is from bas van stencils lab laboratory it's a tool is developed for everybody and what this tool does is compare these two Kamata grams and it's able to be convolute the differences between them and and call whether insertions or deletions have occurred so this is obviously a very easy Universal assay for standard CRISPR and what I mean is CRISPR that just uses a single guide it does underreport mutagenesis slightly I would say 20 to 30 percent underreporting when compared with RFLP essays and it certainly can't really deal with complex CRISPR approaches using julna cases and to guides which I get to talk about later on so in summary genome editing with CRISPR you could consider this now the third generation of genome editing it's incredibly simple and easy to use because you do not need to engineer any proteins and test them out you just need to create these different RNAs to program the cos 9 cassadines are very efficient nuclease and it seems to work in all the species tested so far it's very easy to make multiple cuts and edits at the same time just by expressing more than one guide RNA the problem with CRISPR is that it comes from an adaptive immune system and by that the CRISPR must make mistakes so for example if a bacteriophage is normally been attacked by CRISPR if it mutates its genome such that the proto space that does not match anymore that back to your phage strain could then take over and win the battle against the bacterium that is infecting so that bacterium needs to cut that new phage sequence to some degree so it needs to have a low degree of error and when it cuts that phage genome by a mistake that mismatched genome by mistake it will then record it and now that new bacterial strain is then resistant to that new phage strain so there's this constant battle between the phage and the bacterium and both need to be error-prone to keep duking it out with each other so that's great for CRISPR in prokaryotes but that's not very good for us molecular biologists where we want caste and specificity to be perfect and it certainly is not gypsy studies have found that cast nine stabili interacts with hundreds to thousands of Pam sequences across the genome with any given guide RNA so not just the arm target but lots and lots of off targets cast ein tends to not cleave most of these targets but a small number of them will be cleaved and we'll end up getting mutated and sometimes they will get mutated just as efficiently as the on target sometimes more these off target sites typically have one two three mismatches in the proto spacer region so there can be tens to hundreds of off target sites for any given guide we've only got 20 of bases of specificity to begin with so the fact that it accepts it tolerates mismatches of between one and three bases does mean that there are lots of potential off target sites out there large-scale studies of mutagenesis I found that while some guides are better than others there's no really reliable rules for predicting what will be a more specific guide than another unfortunately so there's there's no real computational way of saying this is going to be a better guide than another other than to say one guides got less potential off targets for then another what has been found is that the first three bases of the protis beds proto spacer surprisingly are dispensable for on target cleavage so there's only 17 basis of specificity required and the first three of these can be mismatched and in general it's been found that cache line tolerates more mismatches at the five prime end of the protis base and these twelve bases here with some some groups have called the seat region are more specific and are more important and certainly are going to be involved in those early stages of of unwinding the genomic duplex and informing the CRISPR Complex so there are lots of tools out there for measuring off-target mutagenesis so groups have done chip seek to see where cast nine binds in cells they've been done in vitro site selection experiments or they prepare genomic DNA and cleave it with cast iron in vitro to find our targets and that does indeed stack up with what actually happens in vivo or you can do very clever methods which capture DNA breaks and cells and you can pair prepare DNA libraries of those breaks and then do high spirit sequencing and therefore sequence what's actually happening in cells and all of these methods have been really powerful and helping to describe off target events in general but i would argue it's not practical to perform off target assays like this in most of your experiments you really need to avoid off top off target activity in the first place if you can so there are design tools out there for the defense talons and CRISPR web tools out there that like generate lists of potential off target sites and so some reach researchers usually because reviewers of in certain journals have asked them to do it will perform target specific assays on a small number of those off target sites typically those that reside in protein coding exons to confirm whether the the CRISPR mutated cells don't have mutations at some other off target alternatively you could do lots of genomic PCR assays on all this different oft are sighs peel them and send them for deep sequencing and then profile that mutagenesis on quite a large scale so as I said I don't think this is practical for most people I think if you're developing medical strategies so you do in some of gene therapy style approach and you're doing gene correction in in a patient stem cells then certainly there's a lot of pressure on you to to look at off target activity and confirm that it's not occurred in your experiments but I think for most people this is it's too time-consuming so we need to avoid off target activity as much as possible so how can we make CRISPR more specific there's a lot of different ways so the first and most obvious one is do not express caste 9 for too long or for to higher-level the off target sites generally are inefficient sites the caste 9 complex might not be very stable at these off targets or the cleavage activity may not be very high at these targets so obviously the more caste 9 you have and the longer you express it you increase the chance of cleaving and mutating these weaker sites so if you can use kattiline mrna our protein in in your most important experiments another quite clever approach is you could use so in the sort of DNA vector based approaches you could use an additional guide RNA in addition to the ones that you need for your experiment you can use one against the caste 9 expression vector itself and so obviously caste 9 will be expressed and will function for a few hours but when it is functional that vector itself gets cleaved and may well get lost but certainly you won't be making any more caste 9 messenger RNA so after that initial burst of caste 9 expression and function casting will be lost and so what you end up with is quite low levels of and low persistence of of CRISPR when you use a self targeting guide that's quite clever another approach is to change the guide RNAs so as I mentioned the first three bases of the proto spacer are not essential for on target activity so you can use what's called a truncated guide RNA where the guides just have 17 18 or 19 nucleotide proto spaces the vast majority of these work at your on target but those off targets are substantially weakened now and so you tend to find off target mutagenesis falls quite considerably at almost all of your off target sites so that's quite a clever approach another approach is to use the dual nikkei strategy which I'll should go through and describe or you could use an enhanced specificity mutant of caste 9 so this is an engineered version of caste 9 where they've engineered out is it's loose specificity and again I shall describe that later on the dual nikkei's approach for CRISPR takes advantage of the fact that caste 9 has these two separate nucleus domains that cleave either the top or the bottom strand of your genomic targets and mutation of either one of these nucleus domains would lead to caste 9 being in the case so for example the d-10 a mutant which mutates the catalytic domain of the rough sea nucleus domain means that caste 9 can only Nick the target strand and a wok won't cut the Pam strand alternatively the H egg 40 a mutant takes out the catalytic domain of the H&H nucleus and so now this enzyme can only Nick the pam strand so now we have a Nick Nick's are repaired quite faithfully in mammalian cells so we're nick in itself doesn't do very much so if you Nick your genomic target or enough target it will get repaired faithfully you shouldn't get any mutagenesis however if you use this strategy where you use to guide RNAs and I've just called them a and B and they need to be arranged in this orientation so the pram sites are facing out and there are not as it stands these two guide RNAs will recruit the nikkei's at the same time hopefully so in this case we're using the d-10 a mutant and it will Nick the target stand in each case so you've got Nick's on opposite strands if those two casts tiny cases bind a Nick at the same time then what you end up with is a double strand break but instead of it being a blunt break it's now gots a long overhang between those two Nick points so in this case we'll have a long five-prime overhang between this point here all the way to there and on the opposite strand to fire prime overhang here and so if used the d-10 a mutant you create five prime overhangs if use the H a forty a mutant you create three prime overhangs using the same guide RNAs so you have this flexibility now so I'm just going to show this again but in a more realistic scenario where the strands are unwound so we have coincident binding of two casts nines that are by next to each other and they Nick at the same time because these strands are already unwound by caste 9 the complex will fall apart creating a double strand break with these overhangs so it's been found that the tail to tail orientation as it were of these complexes is or Pam out orientation of to guide RNAs is the most reliable way of creating double strand breaks using the dual knickers strategy and this offsets quite important if the two guides are too close to each other the cast nines cannot bind at the same time because they're getting in each other's way if the offsets too large then the chance of the the two bubbles the the unwound bases in one come in let's say you're a complex and you be complex merging into one large bubble will reduce and so you again you won't get a double strand break you'll just get two Nick's so there's an optimal window in which this works but there are tools out there that have allow you to go and design these guides these dual Nikkei strategies which are quite straightforward so when you use the dual Nikkei strategy you're on target cleavage is often quite similar to using cast nine wild type cast name with a single guide it tends to be far more mutagenic than or conventional crisper because those overhangs often get lost so you often get larger deletions so in the case of making knockouts it's a little bit easier with the dual Nikkei strategy and the key thing is you're off target activity is near zero when people have looked because the chance of these two guides having off targets that are in this arrangement somewhere else is is miniscule so here I'm going to show you a typical approach to making knock out cell lines using a dual neck ace and that's this is something that an undergraduate student did in our laboratory so it's very easy to do so firstly design your mutation strategy then this very much depends on your gene of interest in this case this was a transcription factor gene so we targeted the first Exxon which had the DNA binding domain encoded within it so this was Exxon - if it's an enzyme you might target the catalytic domain the only thing to really consider is whether your your gene of interest has multiple promoters or multiple splice forms because if you target the ATG or star site of one splice form you may not knock out another spice form which might take over so you might not get a knock out that you expected so all you need to design up initially is where to place your guide so you have an a nav guide and you've used a design tool to help pick those out for you you design you genomic PCR primers that flank that and you should always do your genomic PCR assay first and just make sure that you're able to amplify that region cleanly and perform your your assays and in this case for dual Nikko's we prefer to use an RFLP ese to look at the efficiency of mutagenesis you then need to so this is a plasmid based approach you then need to clone other girls that encode the pro 2 spaces for your guide RNAs and you typically clone them into two separate expression vectors which have the polymerase three promoters in them you can then choose to if you want to so then lift over these guide RNA expression units that you've created so for your a and your B guide and sub clone them into your cass nine expression vector to make an all-in-one vector the reason you would do that is that the efficiency of transfection of one plasmid is far higher than to three or four plasmids so with every additional plasmid you you add to a transfection your transaction efficiency will fall quite considerably so there are quick cloning tools to clone indoors protospace the sequences and to move over polymerase 3 promoter units into another vector so this takes a couple of weeks to do all together and I'm obviously it's only a small portion of your time in those couple of weeks you then transfect those into cells we're using electroporation in this case and our cast iron expression vector Co expresses GFP and so we can look for the fluorescence of GFP and so this tells us what our transaction efficiency is and in this case it's about 90% delivery into the cells so we know that at most 90% of the alleles of our targets could get mutated we can't get a hundred percent of most it's 90 percent so if you're only getting say 60% transfection you need to then adjust your mutagenesis that you'll see later on to that sixty percent number you need to normalize the transaction efficiency to understand your CRISPR efficiency and then we do these RFLP essays and as I warned before they can be sometimes confusing to understand but in this particular case we have this PCR product and we're cutting it with one of two different enzymes so individual digests this one cuts three times some crates for fragments this one costs once and grace to fragments and we can see in the case of here that guide a with wild-type casts nine works well because we've lost this restriction site that cut these two fragments up so these two bands disappear and they lost here and you sometimes see the join of of these two fragments been joined together as in here but often this is mutated as well so it will be a smear it won't be a clean band so the way to quantify this is to quantify the loss of these bands here and then alternatively guide B works because the production of these two fragments here is reduced and so we've got lots of uncut product and then you can compare those digestion patterns when you're using guides a and B together with the cast nine nikkei's and so from all of this you can then quantify what percentage of mutagenesis you've got so 90% of the guide a target site was mutated when using guide a so in other words this guide was a hundred percent efficient guide B we got 75% mutagenesis with guide B as compared with 90% transfection so this was a pretty good guide but not perfect and then when we used guides a and B together with the nikkei's we saw 70% mutagenesis of both alleles so basically the dual nikkei's is as good as your worst guide and then to find your knockout cells so if you can you try and clone your polyfill on a mix of cells so you do the crisp you do the CRISPR transfection three days later you then dilute out the cells to make a single cell per well or if colony-forming cells you would then plate out in a dilute way such that individual colonies will grow up from individual cells you then pick those colonies or lines and then you screen them and it's easiest to screen them for what you're interested in first so in this case we're interested in the knockout of the protein that were looking at which was there's F 1 in this case and we can see that of these 9 lines tested in this first run here 2 of them have lost their f1 protein expression and then we've probably got two that are quite normal and then we've got another 5 where we seem to have either heterozygous mutation or sort of frame or mutations which haven't created a frameshift knockout but certainly have really screwed up expression of those f1 so it's best to identify cells which have lost your expression first and then you do the genotyping on those so you do your RF P and you do your Sanger sequencing on those so indeed these were knockouts on both alleles and these these were our sir also mutants on both alleles just not knockouts on both it's a knockout and mutant heterozygous so another approach and very sensible approach now how to do in CRISPR is to used an enhance specificity mutant so as I said before class 9 has evolved to be error-prone and that enables adaptive immunity to new challenges and the way it does this is that Caston makes many backbone DNA contacts to stabilize complexes prior to cleavage so cassadine uses many positively charged residues like lysines and arginines to bind to the negatively charged phosphate backbone of your genomic DNA target stand and of the non target strand to hold them in place and researchers used the crystal structure and sort of mass mutagenesis assays to find out which ones of those contacts could be mutated so you substitute that lysine with an alanine for example and see where the cast iron still functions or not and then of all those were mutants which still functionally then combine the mutations together to see whether they could get 2 with 3 & 4 residue mutations and still have cast iron activity and indeed this is what they have so there are 4 different and hand specificity mutants of cast iron out there now that have recreated in different ways either through sort of rational design or through a sort of screening based approach and this slide just shows two of them so for example high fidelity casts 9 disrupts 4 contacts between casts 9 and the non target strand sorry and the target stand this one here and enhance specificity cast 9es cast nine has three mutations that disrupt contacts with the with the non targets tandem we have used this one so a key feature of using these enhance specificity mutants is now they do not tolerate any mismatches whatsoever and they need a full 20 base proto spacer so all 20 bit basis of that protospace are key to on target activity so truncated guides do not work extended guides if you've made a longer guide for some reason 21 22 bases they don't work and mismatch guides obviously don't work so our target activity is almost completely reduced there are some off targets in the sort of worst designed guides out there which might still have some activity but in general off target activity is lost when you use these mutants so if you can get hold of expression vectors for these enhance specificity mutants or use recombinant proteins with these mutants then certainly try them there are a small number of target sites where Catalan activity will be reduced when using these enhanced specificity mutants but by and large this is still a very potent system so in summary CRISPR has rapidly overtaken Zetas fans and talent technologies as new targets can be programmed without making any new proteins at all so that that barrier to entry of doing genome editing is now gone and really anybody in any field even people with quite limited molecular biology expertise can now perform genome editing CRISPR is becoming widely adopted in most molecular and cell biology fields crispers very efficient but specificity is a major concern especially if you use wild-type Castner so high specificity casts 9 meters are now available and they're also some very clever strategies to do in crisper as well which mean that your specificity is greatly improved it's possible to modify multiple targets in parallel and of course you can perform functional genomic screens so this is an extremely powerful approach for CRISPR there are a number of CRISPR applications that require precise edits so we're not just talking about knocking out and and destroying and disrupting according region of a gene or the important parts of a gene regulatory element so there are situations where you want to integrate trans genes or short sequences like epitope tags so these are very specific sequences and you will want to put them in specific places also if you're trying to correct human genetic disorders you wanted to make very specific single double or triple based changes and you don't want random mutagenesis to occur and a key thing to point out is if you look at the NCBI clinvar database and download the 75,000 or so human disease variants that are out there the vast majority of them are very small in length so a lot of them are so thirty percent of them are transition point mutations so single DNA base changes a transition mutation just just for reference is where you've swapped to pyrimidines for a pyrimidines so for example in a for a tea a tea for an A or a G for a C or C for a G a transversion point mutation occurs in about 20 percent of human genetic disease and that involves an exchange of a pyrimidine for a purine base or vice versa so for example an a for a T or a G for an a 20.per 626 percent of human genetic diseases involve short deletions and then the reduplication zone obviously angular larger changes such as copy number changes and large insertions and deletions when you look at the deletions and duplications and insertions the vast majority of them are less than 25 bases in length so most human disease associated mutations are very short in length and so if you've got a method to make small precise edits potentially you could use that to correct the majority of human genetic disorders so how do we make precise edits with CRISPR so the first option and one that most groups are using to date is to use homology directed repair of the cell so this involves creating a double strand break and obviously you can do very large insertions when you use someone's you directed repair but you can make small edits as well the second option is to use a new method called base editing we're using an enzyme to swap basis over the advantage of that is that there's no DNA break created and then the third option is to use a brand new method which came out at the end of 2019 so there's only a couple of months old it's called prime editing and there there's no double strand DNA breaks created and it is unlike all the other approaches is capable of making all kinds vedat edits of less than 80 bases in length but a very high efficiency and a very high specificity so I should take you through each of these three so as I mentioned in the overview at the beginning homology directed repair requires the donor DNA template and so for the in the case of a large alteration like inserting a trans gene that's often a plasmid or a viral vector with your transgene flanked by quite long homology arms of 750 bases or more if you wanted to make point mutations or very small edits then your HDR experiment would involve a single stranded DNA donor template so just an old agony that I'd were the homology arms are just 40 bases in length so both of these systems work and you can get precise editing of a target the problem is is that error-prone non-homologous end joining repair come finds these HDR strategies because it is the most dominant repair pathway for the double strand breaks that you create so the majority of cells that you that you create after doing an HDR experiment will be random mutants and in the background is a minority of accurate HDR Corrections and a lot of your correct HD ourselves will be correct on one allele and randomly mutated on another so this can be extremely frustrating and requires a lot of work to filter through each of these cell clones that you will make to find your yourselves with the edits that you're interested in so it's important again to look right back at the beginning where I that non-homologous end joining occurs throughout the cell cycle but is also very high in g1 and s phase whereas HDR occurs primarily in in g2 so one trick that can be used with great effect is to make a fusion of casts 9 with part of a protein called jamming in jamie-lynn is approach that is cell-cycle regulated its transcription an expression is not cell cycle regulated so it's expressed throughout the cell cycle but it is targeted for degradation by the proteasome but in g2 jamming in becomes phosphorylated and this domain that gets targeted for degradation is now resistant and so jamming in can then accumulate and function in g2 and M phase so this is a post translational regulation of cell cycle regulation of Germany so you can take that domain of jamming in that gets targeted for degradation but is protected by phosphorylation and stick that on to cast nine and this now means that the cache line jamming in fusion again is expressed throughout the cell cycle but gets degraded can't function apart from in g2 when it becomes phosphorylated and protected so when you use this you get a big shift in the bias of hgr experiments towards HDR events non-homologous end joining still happens but no perhaps is a minority event so now maybe two-thirds HDR on one third non-homologous end joining so this is a very powerful modification to HDR strategies so this is just showing an example from one recent paper of researchers who are very good at HDR and these are the kinds of outcomes that you might get so here that targeting a variety of genes with single base changes or yeah I think yeah or here's a triple base insertion so different HDR edits and in blue bars you can see this is a percentage of cells that have got their HDR edits and then an in gray are the percentage of last the percentage of alleles that have got random insertions and deletions on them because of non-homologous end joining so this shows the extent of the problem so base editing is an approach where you can make very small changes often just individual bases using an enzyme that will exchange that will modify the DNA base and when it's replicated it would lead to that that that sequence being changed so in this case they're using a fusion of Castine and often leaves and the nikkei's mutant d-10 a which I'll describe in a minute as to why they do that but the they use cast nine fused to a deaminase and there are different denominators out there so for example there's aid and there's Apple Beck and what these D rnases or ABC 7 and what these do lasers do is deaminate for example an adenosine base and convert it to an inner scene base when in a scene is replicated it gets replicated as guanosine base so what you end up with here then is an A to G change alternatively you could deaminate a cytosine base to a D you and that gets replicated as a T so then you get a c to t change these the d-10 any case and they're very clever about the design of which strand they put their guide RNA on and this is all to make sure that only the base edits on the strand that they're interested in getting cooperated because NIC repair occurs on the non edited strand so NIC repair doesn't create obviously any mutations by itself but it's been used to bias how the cell either uses the edited or the or the non edited strand so the advantage of this approach is here using an enzyme so the percentage of base editing is very high indeed I guess a key disadvantage is that depending on the sequence that you're targeting it may not be possible just to edit one base only it might so for example we're making a specific a2g change there may be other a bases locally that will get converted to a G as well and so we'll get all the edits that we did not intend so so it does very much depend on the target is that how useful this base editing is but in some situations it's very very powerful indeed there's also a risk that this base editor will modify bases at off target sites and that those will be quite hard to to discover so I'm gonna end up with talking about prime editing prime editing is an incredibly powerful approach that uses a modified CRISPR system but there are no double strand breaks were created so we're not reliant on hosts double strand break repair it uses a cast 9 knee case to locate this prime editing enzyme to your target site and to prime the editing the intended edits is carried on a guide RNA template and reverse transcriptase is used to copy that guide RNA template onto your target and creates this edited strand and then that edited strand is then incorporated into your target so I shall take you through this in detail so the prime editor again involves a cast 9 a guide RNA and your target as before but there are three key changes from conventional CRISPR so firstly they're using the h 840 a new case of caste 9 so if this Knicks are tough targets it will create to Nick but doors won't lead to mutations and the Nick at the on target by itself again won't lead to any change either that in itself is just an entry point as a reminder the h 848 nikkei's nicks the non target strand or the palmar strand and so it will Nick here at this very specific location 3 bases away from the Pam the H 840 a new case is fused to a mutant version of the FML V reverse transcriptase domain this reverse transcriptase is very well used in molecular biology it's the one used in reverse transcription reactions when making C DNA's and molecular biology labs it's something that's been studied intensely there are lots of mutations of it have been made to look at its performance and so the mutant that they're using of the reverse transcriptase here they've selected for one that functions very well in this particular architecture here thirdly the guide RNA which is the same as before so normal guide RNA except that now it's much longer so surprise editing guide RNA or PEG RNA and this 3 prime extension carries the template for the change that you want to make to your genomic target so how does this work so as a reminder you're all crisper target sites consists of a Pam sequence in the genomic DNA next to 20 bases that match the pro 2 spacer in the guide RNA and cast line cooks three bases away from the Pam in this case it's just a nick and the nicks occurring on the Pam stone and that first base after there we call +1 and what the prime editor can do is change any sequence after this point so all the sequence here including the Pam is susceptible to the prime editing so the reaction occurs as follows the first step is the nicking of or binding to the genomic target opening up of the stands and the nicking of the Pam strand so what you end up with is a 3 prime flap here of your genomic target and it is pairs up with your peg RNA so the fact the very end of the peg RNA that 3 prime end pairs up with your proof to space of targets and generally this primer binding site is 12 bases long so again you know what the sequence is it's your proto spacer from the cut site on which those 12 bases are placed at the end of the Peck irony so all reverse transcription reactions require a primer of some kind and in this case this 3 prime overhang here is effectively a primer for 5 prime to 3 Prime reverse transcription so now in the next step a verse transcriptase adds on the edit and the edit just might be a single base insertion or deletion doesn't matter what it is here is this template so in this case I've shown a 3 base change in red so the original sequence in blue and black the three based change in red so it fills in that three base change and then a short sequence that matches the rest of your genomic targets which you want to be to stay the same so the the other side of your edit so this is the nod edited sequence and then a key thing is the reverse transcriptase cannot go any further into the peg RNA because it's bound up by caste 9 so the reverse transcriptase doesn't copy a guide RNA sequence into your genomic target it only copies in what you've intended in your edited region here so it fills that in and then that's it that's all this enzyme does so it binds opens nix primes and fills in and that's all it's doing so now we're hoping for the cell to incorporate this edited strand over the non edited strand so we have a 3 prime flap containing the edit in red and 3 Prime single strand Enzo a relatively stable in the genome there are no enzymes that will choose 3 prime to 5 prime so this free 3 prime flap is there and the NIC is there now there is going to be some breathing this Nick creates instability in this duplex so these bases will unzip at some to some level and so there will be some equilibrium between the 3 prime flap and a 5 prime flap now this will be the most dominant form this won't happen very often but when it does this 5 prime flap is now available for digestion there are many enzymes that choose single standard DNA 5 prime to 3 prime there are many enzymes that recognize 5 prime flaps and just cut the flap off so the idea is that host enzymes will excise this fire prime fly and your three prime flap your editor flap will they come on in and hybridize through the basis that you've kept the same but now you'll have a mismatch a mismatch between your edit and the original non edited sequence so you've got a 50/50 chance of the cell when repairing that mismatch of incorporating your edit now the clever thing in the prime meditating is that the have used a second guide RNA but now it's a conventional guide RNA doesn't have this three prime extension on so it's the second guide RNA to target your same prime editor enzyme to Nick the non edited stand okay so this guide RNA hopefully it's going to Nick after the primary things happened if it nicks before the primary things happened nothing really is going to happen then it gets filled in it's fine so some of the time this Nick is going to happen after the prime in it when it does the the its previously been find that by adding a Nick into a mismatch sequence so this was a why it's used in the the base editing system as I've just mentioned before is it biases the the repair of this mismatch to remove this strand and retain the edited strand so if you Nick the non edited strand you will lose the non edited sequence and retain your edit okay so in the best form of primary thing or to guide RNAs one of them is delivering your edit via reverse transcription and the second guide RNA is biasing the repair of the edit towards your edited sequence and so you don't end up going back to your original sequence so this is data from as Prime editing paper so as I've shown you before these are the kinds of ratios of specific edits you get from homology directed repair using single strand illegals so quite low levels of precise edits and high levels of background non-homologous end joining when used the prime editor you often get around 50% of yourselves contain your precise edit and it is precise there's no note in Dells associated with it when the primary that goes in and there's a very low background of indels associated with mismatch repair that that there hasn't been accurate but occur at your target and depending on how you design that second guide RNA to cut your target strand if you can design that towards your original target then and you can refer to the paper for this then this background goes down even lower so primary things a is a huge breakthrough in the CRISPR field it's got broad application because it's more efficient than conventional CRISPR for creating precision edits and there's got a very low level of error as it currently stands and and obviously it's people are going to work on this and probably optimize it even more a key thing is it can mediate all four transition point mutations and all eight transversion point mutations and base editing the enzymatic base editing I talked about before cannot do that it can only do transition mutations it's been possible so far to insert up to 45 bases using prime editing so adding epitope tags onto the ends of genes and also they've been able to delete up to 80 basis with prime editing and actually I think they've sort of discovered what the limit of it is yet so on paper this can correct around 90% of human pathogenic genetic variants in principle so it's very very exciting for everybody so in summary I've told you about Zelda fans and talents and showed you that nucleases can stimulate editing of genomes and living cells via host a double strand DNA break repair Zelda fans and talons are most useful when small proteins are required so when delivery to primary tissues is difficult but for most applications CRISPR is far easier faster and widely applicable and conventional CRISPR remains the choice for gene knockouts and performing screens but now CRISPR can be performed with high specificity and prime editing is probably the best choice for the precision editing of genomes well thanks for watching I hope you found that lecture useful please let us know in the comments whether whether it was good for you whether there's something else that you needed and please also definitely give us a like and consider subscribing because we've got a lot more material coming up and a lot of tutorials that we think you'll find very useful indeed please also consider following us on social media particularly Twitter and Facebook where you can engage with us and ask us questions and perhaps make suggestions of what we might change or what we might create in the future thanks very much bye

Channel: Genomics Gurus

Views: 1,468

Rating: 5 out of 5

Keywords: Genetics, Genes, University, College, Public Engagement, Science, Lecture, Tutorial, Glasgow, Scotland, University of Glasgow, Glasgow University, Adam West, Genome Medicine, Precision Medicine, CRISPR, CRISPR/Cas9, Cas9, gene, editing, gene editing, biology, natural, genome editing, genome, molecular biology, crispr cas9, crispr explained, crispr ted talk, crispr/cas9 technology, genetics biology, genetics documentary, genome engineering, genetic editing, crispr cas9 explained youtube

Id: _6ZBVf6H_IA

Length: 121min 20sec (7280 seconds)

Published: Wed Feb 19 2020