CRISPR therapeutics (a LinkAGE webinar)

Hassan Shakeel: Hello everyone. My name’s Hassan Shakeel. I’m one of the Health Education England Genomics Education programme development leads. And it is my immense, immense pleasure today to introduce a talk by one of my very close friends and he has been since the start of medical school for me, Dr. James Paterson.

James was with me at Cambridge. Then he went to UCL to do his MB/PhD with Sir Paul Nurse, and he’s going to talk to us today about his work with CRISPR therapeutics. So without any further delay, James, you have the floor.

James Patterson: Thank you for that introduction, Hassan. Lemme just get the presentation shared. But thank you everyone for coming today.

Today I’m gonna be talking to you about CRISPR technologies, CRISPR therapeutics, and giving a sort of broad introduction to CRISPR.

Quick slide on my disclosures.

So an overview, just talk today. Today I’ll be talking to you about gene editing and CRISPR, I’m giving us sort of brief historical context on the development of this exciting technology, before going into a bit of an in-depth review into how the Cas9 gene editing system works, which is really the prototypical CRISPR system. I’ll then run through some alternative Cas systems in their potential use cases. Talk to you about novel ways of functionalizing CRISPR enzymes to enable novel therapeutics. Give you an overview of the CRISPR therapeutics landscape, and then chat briefly about some key challenges faced that facing CRISPR’s use in the clinic, and some ethical considerations as well.

So to start off with CRISPR, so CRISPR really is an amazing breakthrough for which two scientists and Emmanuelle Charpentier and Jennifer Doudna, received the Nobel Prize. And really I’d say Gene editing is the holy grail of gene medicines because it allows you to correct endogenous genes rather than inserting additional copies, which was really the topic of the majority of my talk last LinkAGE webinar.

And CRISPR has really increased the use in efficiency of genetic editing versus prime methods. These original methods include meganucleases, zinc finger nucleases (ZFNs) and TAL effector nucleases (TALENs).

And what’s really exciting to me is that they show the value of investing in research into fundamental science. So this work is all derived from work into the bacterial immune system, and now it’s having such a massive impact on the therapeutic space.

So just to put CRISPR into the context of the history of genome editing, because genome editing started long before CRISPR. And really it started off in the 1970s, 1980s with the demonstration that you could prevent gene replacement in yeast. It started even before that in bacteria in 1958, there was Nobel Prize for homologous recombination in bacteria.

But really it progressed throughout the 1980s and came to a front in the early 2000s with the expansion of zinc finger nucleases. And these have been used in therapeutics by a company called Sangamo. And I’ll go into them briefly as well.

In the late 2000s, TAL effector nucleases came onto the stage and has a brief moment, I’d say, before CRISPR came about. And these were, much easier to program than DNA sequence specific zinc finger nucleases for targeting specific genes.

And then really in 2012 is when CRISPR broke onto the scene. CRISPR’s been around also since the 1980s, so when it was first identified as weird repeat stretches of bacterial DNA.

It’s only in the 2000s when a lot of work by Emmanuelle Charpentier went into understanding what these repeat sequences were and the process of their importance for bacterial immunity to bacteriophage infection was discovered. And then this quickly progressed into a realization that this CRISPR tool could be used for site-specific genome editing.

And all these advancements are really resulting in CRISPR therapeutics starting to make their way into the clinic. So at present – this slide is a year old now – this data here, but there’s 31 CRISPR trials progressing through the clinic. And far outstrips some of the earlier technologies.

So you can see TALENs are seven; trials for zinc finger nuclease is five trials. And this is in ex vivo gene editing. So if anyone comes previous talk, you know this is where you edit cells outside the body and put them back into the patient.

While less trials ongoing in vivo gene editing, for all modalities, and this is really a result of the delivery issues that we touched upon last time as well.

But really today, CRISPR is the tool of choice for the therapeutic gene editing approaches. And this is because of: its high efficiency, the relative simplicity of targeting with it, it’s small size – so the stability to fit into various delivery vehicles and get into cells. And it’s really made a major impact. So today, 63% of gene editing trials utilize CRISPR as a tool of choice.

I keep talking about gene editing, but what really is gene editing at its core?

Gene editing is the modification of endogenous genome sequences through site-specific manipulation of DNA repair pathways.

And this is something that I think I’d like you to take away from this talk is that any gene editing or any genetics editing approach (whether it’s zinc fingers, TALENS or CRISPR) it’s really just a way of manipulating your cell in to correcting DNA damage in a way that results in it incorporating a novel nucleotide or correcting some sort of DNA sequence and returning it to it wild type, next state or whatever user-defined state you want it to be.

You want to integrate novel repaired sequences you need to introduce a DNA template, so that’s an additional component. And this can then be incorporated through homologous recombination. There’s some other exciting approaches called, for instance, prime editing that I’ll run through briefly as well.

Site-specific DNA breaks in the absence of template DNA result in random insertion-deletions due to the repair pathways you used, which can disrupt gene functions.

This is how you can generate a knockout. And CRISPR use cases now really extend beyond this simple gene editing current paradigm. So CRISPR’s really evolved into something much broader. Sort of a platform to target specific bits of DNA and do whatever you want to do when you get them. And there’s a range of therapeutic applications that result from that.

But just to dive a bit more now. So the DNA. How DNA repair pathways are the key mediators of gene editing.

So any of the gene editing approaches we talk about, the classical gene editing approaches, are all about DNA site-specific cutting. So you pick sites where you want to target either gene deletion, or you want to target gene introduction, but you need to cut that site specifically.

This results in a double strand break. No different from the double strand breaks that you hear about with ionising radiation or any other mutagens. However you’ve introduced this breakout at one specific location. The double strand break has two ways to be corrected in a cell. They can either be corrected through mechanism called non-homologous end joining, which is low fidelity and can result in insertion-deletions, and therefore gene disruption. Or if you have a template which potentially has a new novel piece of DNA that you want to insert – so here highlighted, for instance, in red – this templated region can then be inserted into the genome at the site of which you cut, dictated by the homologous lines that are either side of your gene of interest that your trying to introduce.

So just going into a bit more detail, now, on how non-homologous end joining works. And this really again, is this, it’s a low fidelity way that your body uses to fix double strand breaks.

So it’s error prone because once you’ve cut and created a double strand break, there’s a range of enzymes that process the ends that are created, and they chew back nucleotides disrupting the gene sequence. They can introduce novel nucleotides into this at the cut site. And then it randomly glues these two ends back together with whatever nucleotides have been inserted or deleted. So the end product of just simply cutting, or doing a site-specific cut with your gene editing tool, is a random combination of deletions or insertions that can result in gene disruption.

Endogenously, really this is the only used to fix double strand bakes in the G1 phase of the cell cycle. Classically used then. And that’s because you don’t have a sister chromatid that you can use as a repair template. So obviously after S phase, you’re able to utilise another copy of your genome and then this can be used as a template for DNA repair.

And that DNA repair mechanism is known as homologous recombination. So homologous recombination is the DNA repair process that enables high fidelity DNA insertion.

So, for this type of gene editing you require both a site-specific double strand break, so your gene editor has to cut at a specific locus, and a template which will which features DNA sequences that are homologous to either side of your cut.

So, for instance, here you have your template. You can see there’s a left homology arm with a specific change that comes after this. And then there is a five-prime, or so, three-prime or right homology arm. And this is homologous i.e. the DNA sequence is the same as the DNA sequence in the genome immediately after your cut site.

What the cell will do is recognize this homology, and then fold those arms into the genome, introducing them, and then it gets tricked, introducing whatever site-specific change you have in them. So this is important for integrating genes into the genome or integrating particular modifications that you think the cell should have.

So for fixing things in a very high fidelity manner.

So what both these mechanisms have in common is the fact that you need to introduce a site-specific double strand break.

So the first approaches to doing this, I mentioned earlier, was the zinc finger nuclease. So I want to just take you through the history here so you can understand why CRISPR is so important. So it’s important to look back at really what we were using or what other people were using back in the early 2000s.

So zinc finger nuclease is really the original site-specific DNA modification enzyme.

And it’s a two component system, where zinc finger proteins are made of modular domains. So you can see here the P1, P2, P3 and P4 domain, highlighted in the left zinc finger protein. These domains are evolved to bind into specific 3-letter code. And when you chain these together, you get increasing specificity of findings. So P1, for instance, binds to the -GTT-, P2 binds to -GTA-, P3 to -CTG-, and P4 to -CTC-. And this gives you your actual full specificity of where you bind your zinc finger protein.

And you can modify these P1, P2, P3, P4 domains so that they have different sequence specificity.

On the end of these proteins you have a nuclease, which only works if you have two copies bound within a certain range of each other.

So again, you can increase the stringency of your cutting by having an additional zinc finger nuclease, which binds upstream of the left zinc finger protein. And then you reconstitute a functional cutting enzyme, so double strand break inducing enzyme, between the two. And this allows you to get a site-specific double strand break.

So zinc finger nucleases were originally popular, but just very difficult to engineer, because – as you can imagine – redesigning a protein to bind to specific three-letter DNA codes is quite timely, this all feels very hard to do, so didn’t really take off in sort of popular research use.

TALEN systems came next, and these are Tal effector nucleases actually derived from a bacteria that infects plants. Again, I think demonstrating the points of fundamental research.

And so these are also proteins that bind into specific bits of DNA, and they work through the same modular concept as chaining protein domains together that recognize particular nucleotides within the DNA.

And here we can see again a TALEN.

And each individual domain, now, has specific two amino acid difference, which dictates stability to bind either ‘A’, ‘T’, ‘G’ or ‘C’.

So by chaining together a range of these proteins with different amino acids at these X-marked positions, you’re able to generate a specificity of binding to a particular DNA sequence in the genome.

Again, these work through reconstitution of a DNA cutting enzyme, Fok1, which only cleaves when two copies are present in close vicinity. And that’s driven by having an additional TALEN that binds upstream to the first one.

So again, while this is easier to program than a zinc finger nuclease, these TALEN proteins are extremely large so very difficult to deliver.

They’re also clunky in the sense that you have to have two enzymes, two proteins present and bind to achieve cutting. And so when CRISPR came along very rapidly, after the TALEN started to gain ground CRISPR/Cas9 system, shown here, it was really revolutionary.

So what are the constituents of the CRISPR system?

The CRISPR or Cas9 system, as hinted here, it features a Cas9 protein. So Cas9 is this protein highlighted by sort of beigey oval. And it also features an RNA component, a synthetic guide of RNA. Now the Cas9 itself features endogenous nuclease activities. There’s no need for the Fok1 nuclease that was present in both the TALEN and zinc finger nucleases.

And specificity of binding to DNA is dictated by your sgRNA sequence. Anyone [familiar knows] that complimentary base pairing is an important concept, for instance for siRNA therapeutics. And this effectively works in the same way.

There’s a region of your guide RNA that is complimentary to the region of the genome you want to cleave.

Upon binding to that region in the context of Cas9 the guide RNA, the Cas9, and the DNA stimulates cutting through two nuclease domains present in the Cas9 enzyme. And this is what allows you to achieve double strand breaks with Cas9.

So as you can see, this is a much simpler system to program because there’s no need to do protein engineering. One can simply synthesize guide RNAs with particular nucleotide sequences that exactly match up with your genome target sequence to achieve site-specific DNA cleavage within the human genome.

So where did this CRISPR/Cas9 system come from?

So CRISPR systems are really the bacterial immune system. And the way they work that a a virus might infect a bacteria. So shown up here at the top, so bacteria phage and will inject its DNA into the bacteria.

This DNA can then be randomly incorporated into the bacterial genome in regions that feature these clustered regularly interspersed short palendromic repeats (CRISPR). When these regions are inserted – so you can see here that the viral gene, in blue, has been inserted in the context of the black DNA, which is the bacterial genome – you can generate a transcript which features viral DNA interspersed with repeated bacterial gene DNA. This is transcribed, it’s a long RNA featuring both viral genes and also bacterial genes, which is then processed releasing small RNAs that feature viral RNA. (So now we’re at region C). That feature viral RNA, a component of the bacterial RNA, and they bind to another RNA called the tracker RNA. This is just a free-floating constant RNA that has complementarity to the black region of RNA that’s present in this bacteriophage plus bacterial crRNA.

And I know it gets a bit confusing with the number of different RNAs here to keep track of, but I promise it’ll get simpler on the next slide when people figured out how to fuse these into one single RNA.

Now the importance of this is that you’ve effectively generated an enzyme that is now specific to bacteriophage DNA. So that means that when your bacteriophage comes back, and infects this bacterial cell, you have a Cas9 bound to RNA that directs it to cutting that bacteriophage DNA. So, now when the bacteriophage infects, its genome gets chopped up and it’s unable to infect the bacterial cell.

So by randomly integrating bacteriophage DNA into the bacterial DNA, the bacteria’s able to establish long-term memories of infections and protect itself from future infections by the same bacteriophage.

So this two-RNA system that requires both the bacteriophage RNA and this constant tracker RNA, is slightly too complex to be used in a sort of efficient therapeutic context. We want to have one RNA, one protein, and we don’t want to have to worry about mixing together the tracker RNA, some other element of RNA and then forming a functional tripartite complex.

So work was done to fuse these two RNAs together, and this is really a seminal paper of the Doudna and Charpentier labs.

And what they did was they looked at how the crRNA, so this is the RNA with the sequence complimentary to target DNA and a bit of constant region, and the tracker RNA, which is the RNA that binds that constant region the crRNA. And they fuse them together with a small hairpin loop sequencer of a G and three As (-GAAA-).

This produces the single guide RNA, synthetic guide RNA. And what they showed is that this RNA is able to direct Cas9 to cleave particular plasmids, shown on this gel blot here.

I thought such a seminal paper is worth showing the actual figure. I think it’s now been cited around 15,500 times.

Able to show cutting. So, for instance here, this is the situation with just Cas9, this first lane and no RNA. You can see the plasmid is fully intact.

When they add in a crRNA – again remember this is the RNA containing a bit of a constant region and gene specific region and the track rRNA – they see cleavage.

And when they add in the chimera-A RNA, with a full length tracker region, they see cleavage of the band.

So this is really the paper, which aside from all of the other work, I think this is the key paper that contributed to their Nobel Prize.

So how do we use CRISPR in practice?

There’s two components here now: there’s a single guide RNA, which features a target specific region that you can program, and there’s a Cas9 protein. These are the two important components.

So when you’re in the lab and you want to knock out a gene, what do you do? How do you go about doing this?

So, first off you have to identify your target gene – so what gene are you going to knock out? And then you can use online tools to identify guides.

So it’s a range of online tools now that you will paste in the target sequence you want to hit, and they throw up a range of different guide RNA or potential target sequences that you can go after. And what’s really useful about these is they look for potential off-target sites.

So we’re talking here about some 20-nucleotide target site. Your genome is billions of nucleotides long. There’s a chance that those 20 nucleotides come up again and you don’t want that because you want to introduce one site-specific double strand break. What this program will do is hunt for anywhere that this site comes up again and tell you. It also now gives you some prediction of the efficiency of cutting because obviously you want to put this in and see most of your cells have cut in them or double strand break.

And so once you’ve done this design, you go online shopping, basically. You can copy the guide RNA sequence that you want to order into an online web store and order that guide RNA. And then you can also buy separately the Cas9 protein. So these are as purified components.

Once you’ve received your Cas9 and your guide RNA, you mix them together in the tube, and then you have to deliver this into your cell population.

So you’ll have your cells growing. Maybe they’re hematopoietic stem cells and you’re about to knock out a particular gene within them. Maybe they’re T cells. But you are effectively mixed together those cells. And one way of getting the CRISPR complex, then into those cells is through electroporation, which again I touched on the last talk.

So this literally ‘zaps’ your cells and forces the CRISPR complex into them, into the nucleus where it gains access to DNA and can perform cutting.

You then allow yourselves to recover and you can harvest your DNA from your cells and check for editing by sequencing.

So this plot is Sanger sequencing. You receive a trace. You also perform a control sequence, where you sequence the same region, but it hasn’t been exposed to CRISPR.

And then you look for cutting. So here, for instance, you can see that we’ve chopped out a T and a C resulting in a two base pair deletion in the majority of the cells by looking at the Sanger sequencing trace (in this particular genome editing experiment) .

One thing to highlight is that this deletion would’ve been caused by non-homologous end joining because we didn’t introduce a repair template. So we’re relying on the cells endogenous repair mechanism to erroneously correct the double strand break you’ve introduced and introduced some sort of small insertion or deletion at that site. So really DNA repair is doing most of the work after you’ve introduced your double strand break.

So what do you do next?

You have a population of cells that feature some sort of modification that’s been introduced into them through a double strand break, and now you want to use them.

You have your pool, which has has knockouts for within it, and again, if you were doing this experiment on primary T-cells trying to engineer a T-cell therapeutic, you might be an engineering natural killer cells as a therapeutic, or you might be engineering CD34 hematopoietic stem cells. And in these cell types you really need high efficiency of knockout because you can’t take single ones to those cells and grow them up. You have to deliver that entire population to your patient.

So you determine your knockout frequency through sequencing. And then you can introduce these cells into your patient, assuming that you’ve achieved a high knock out frequency.

However, there are some cell types and one that’s really exciting are induced pluripotent stem cells.

So imagine you have an induced pluripotent stem cell now from a patient and you’ve used CRISPR to knock out let’s say a dominant genetic mutation that’s causing the trouble. And now you have a stem cell that you’ve made that does no longer feature expression of that protein.

You can single-cell clone from your pool, your mixed bag of different modifications in your different cells, isolate a single cell that does have knockouts (i.e. it has a deletion or insertion of a non-multiple-of-three number of nucleotides, introducing a frame shift) and expand and differentiate using the downstream cell types.

Just want to highlight that induced pluripotent stem cells really leverage this advantage in being able to do clonal cell isolation and produce defined pure population of cells that all feature whatever knockouts or gene editing modification that you’ve wanted to introduce.

So as mentioned here, what you would do is take your induced pluripotent stem cells, do your precise genetic engineering, end up with a human pluripotent cell line that now features your engineered modification, and then the great thing about these is because they’re stem cells you can then differentiate them into whatever the downstream cell type you want.

So this is a great way of producing ex vivo engineered cell therapies at scale.

So people are turning IPS cells into T-cells, TNK cells, CD34 cells, as well as using them for more regenerative medicine approaches by making neurons to treat Parkinson’s, making cardiomyocytes to potentially treat cardiac damage.

Some limiting factors just to mention about bringing these into the clinic is just the safety of the cells. Obviously they’ve been engineered, they’ve had some double strand breaks. They’re potentially oncogenic themselves. The differentiation protocols used to turn them from a stem cell into a functional cell type are all being worked out at the moment. Because you really want to make sure high functioning differentiated cells.

And transplantation methods needs to be worked on. So in that cardiomyocyte example I gave, it’s not immediately obvious how you graft on some cardiomyocytes to a damaged heart and expect them to graft and then correct that damage.

So there’s a lot of work on ongoing in these fields. And really CRISPR technologies being used in all of them to produce high functioning differentiating cells.

And there really are many use cases of CRISPR even beyond healthcare.

People are using it to engineer crops to increase nutritional value and engineer pest resistance. So engineered mosquitoes for malaria control, so it things like gene drives. There’s engineered animal models to improve drug evaluation strategies, making better mouse models so we can get better predicting whether a drug is going to work before we enter the clinic.

And they’re really specific to human health and practice.

People are generating personalized medicine models with IPS cells, again, using these stem cells that come from patients for engineering stem cells, from patients. There’s mass improvements ongoing in the antibody and biologics manufacturing space, so making cell lines that produce antibodies better and cheaper, potentially expanding patient access to these.

There’s really interesting diagnostics work now using CRISPR to detect specific nucleic acid sequences. And this is actually one of the first covid tests that was developed was by a company called Sherlock Biosciences in the US.

Obviously there’s a lot of work on going into ex vivo engineered cell therapeutics, and then really the dream future is in vivo gene medicine, being able to do site specific, genetic modifications in patients to fix whatever genetic ailment they might have.

So I just want to run through why this is important in the direct discovery system.

And here you can imagine that if you have a patient with a severe genetic disease and you want to screen for a drug that can help treat that genetic disease you can generate stem cells from that patient. So induced pluripotent stem cells. And seems a bit counter-intuitive that you can then turn that that at stem cell normal in a dish.

And this creates a controlled cell line. So it’s effectively your dream case. What would your patient look like if they had a normal genotype at the diseased or mutated site? And it gives you the upper bar for how good your cells can be. And it’s a perfect comparison to your diseased or mutating cells.

So you can now differentiate these into a cell type and screen drugs. And you can look to see how much your drugs make your disease cell models look like your fixed cell model. It’s a great way to screen for different compounds might assist these patients. You can also compare the transcriptomes of these different cells and look to see if there’s any potential drug targets that you could go after.

In cell therapies one particular area that’s really heavily under investigation at the moment is engineering T-cells to target cancers. And the key benefit of keeping that CRISPR enables is that it allows you to Use donor T-cells to engineer these cancer killing T-cells.

So the first generation of these CAR T-cell products are all based on autologous therapeutics, i.e. you harvest T-cells from the patient, they get shipped out to a factory somewhere, they get transduced with a virus expressing this chimeric antigen receptor that targets them to a cancer, and then they get reinfused back into the patient and go off and kill the patient’s tumour.

And you can imagine that there’s some problems around a patient might have been through many stages of chemotherapy. And their T-cells might be a bit knackered and therefore not be as efficient at killing the tumours as as a healthy person’s would.

Additionally, these patients are often really sick and every week counts. And if you have to send away those T-cells to a processing plant before you can get them back into the patient that time can be an important. So if you have off-the-shelf access to engineered T-cells that’d be great.

The problem is usually healthy donor T cells get rejected by the patient. Healthy donor T-cells can also go on and attack the patient because they themselves will recognize antigens in that patient that aren’t present in themselves.

So what you can do using CRISPR and for instance the nonhomologous end joining approach is knock out the genes responsible or that T-cell mediated killing for patients, so graft-versus-host disease. And you can also knock out markers that the patient’s immune system will recognize on these donor T-cells. So you can engineer them to be effectively immune inert or immune silent as possible. And this means that in a hospital you could have a bank of frozen allogeneic CAR-T cells ready to go for the patient, and to be infused as quickly as possible.

Using CRISPR as a drug itself, in vivo editing – this touches on quite a lot of what we went through last week – the key issue is delivery.

How do you get your CRISPR enzyme to the right place in the patient?

We now have to get the CRISPR protein cutting in the right place in the genome but getting that to the right cell in a patient is very difficult. And people are working on a range of different delivery vehicles and nanoparticles, like the lipid nanoparticle – covid vaccine comes in – and different viral vectors.

This would be really revolutionary once we can target these really beyond the liver for on demand once-and-done therapeutics to restore genes to their wild type functionality.

Some of the limitations of CRISPR come into come into targeting specificity. So we said that you have a 20-nucleotide region for which you can direct your CRISPR enzyme. As mentioned your genome is really big. There might be regions of the genome that slightly match your target site. You can get off-target cutting. It’s your delivery of CRISPR. Putting RNA into cells that doesn’t look particularly natural, can trigger immune responses. We spoke about that last time a bit as well. And there’s also some issues around just the toxicity of the immunogenicity of Cas proteins. Obviously their bacterial protein is recognized as foreign by the immune system. It’s not particularly a surprising benefit of the Cas system, there is a need to act very quickly. Once you’ve given it once you don’t need to give it again if you had to.

Just want to touch on something else, which is an inherent problem of all the gene-editing systems we’ve discussed so far. And this is the fact that you have to introduce a double strand break. Your cell really doesn’t like double strand breaks, and that’s why has all these repair mechanisms to fix them.

But these repair mechanisms aren’t completely perfect and any free chromosome ends can end up fusing somewhere they shouldn’t. You can get exceptionally large deletions, in some cases, really disrupting a lot of the chromosome. You get odd chromosomal inversions. There’s a lot of things that can go wrong if you’re introducing double strand breaks. So people are really trying to investigate CRISPR beyond being able to generate double strand breaks and do gene editing. I’ll touch on this shortly.

You can imagine this is especially a problem if you’re trying to do multiplex gene editing, you’re trying to delete multiple genes in the target cell .

So this is a figure from a paper where we’ve been trying to to modify three different sites in a T-cell. So they’re knocking out MHC class 1 expression through B2M. This makes the T-cell unrecognizable by the patient’s T-cells, so they can’t kill it. It’s knocking out PDCD-1, which is the receptor that recognizes the protein PD-L1. So it makes it makes your T-cells immune to checkpoint inhibition. And, these authors were also trying to knock out the TCR locus, so this prevents the donor T-cells from going on and attacking the patient.

These are three things that you really want to have if you’re having an off-the-shelf T cell therapy.

You can imagine this is that if you’re generating a lot of free chromosome ends, you can end up with a range of different translocations. Again, because non-homologous end joining isn’t perfect and it just glues together any free chromosome ends around. So you can end up with some really, really deleterious chromosomal translocations at relatively high frequency. All around 1% there’s still a risk.

So how are people trying to get around that?

And one way of doing this is by looking at the Cas9 enzyme itself and modifying it.

People have started to think about CRISPR/Cas9 now as a gene-homing vehicle rather than just as a double strand break inducing. And this is really now moving beyond classical gene editing .

So as I mentioned at the start your Cas9 enzyme has two nucleases, has two sites that cut each strand of your double strand of DNA. People have now engineered mutants of Cas9 that feature no nuclease activity. So these ones are able to bind site specifically in the DNA, but they don’t cut. And people are doing really exciting things here to fuse these ‘dead’ Cas9s to epigenome modifiers, to alter the epigenome at site-specific locations.

I think it’s also really interesting is people are making mutants of single domains of these DNA cutting sites, to generate Cas9 nickase enzymes.

What Cas9 nickase enzymes enable you to do is to trigger alternate DNA repair pathways that can substitute in and swap particular nucleotides near to that cut site. And these have enabled approaches such as base editing, prime editing – which I will come onto in a second – both approaches that don’t generate double strand breaks, and therefore rid you of any risk of DNA translocations.

Base editing works through generating a nick, and then having an enzyme which does a chemical modification to nuclides to change them. For instance, so cytidine deaminase base editor, and this is a nickase Cas9 that it’s attached to. So this is a big fusion protein.

And it can convert a CG nucleotide to a TA.

So this is particularly interesting if you’re trying to treat patients with single base pair mutations in a genome. You don’t need to do a full double strand break or introduce a homologous recombination template. You can directly modify a genome.

Prime editing is slightly more complex. It involves the fusion of a Cas9 nickase to a reverse transcriptase.

And a modified guide RNA now that has a template which features whatever gene or whatever DNA you want to insert into the patient’s genome.

So now you have your guide, which brings Cas9 to the right place in the genome where you want to introduce small sequence to.

When it binds the reverse transcriptase, this is very similar to ones present in the HIV virus, transcribes from the guide RNA templates and introduces whatever DNA you want to introduce. So this way you can do site-specific modifications in insertions of DNA up to about 15-nucleotides in length without having to generate the double strand break, which is really powerful.

Using the dead Cas9 enzyme people are fusing these dead Cas9 enzymes to epigenetic modifying enzymes. So these are things that can bind to histones and alter histone locks. For instance, binding dead Cas9 to P300 in the context of the guide RNA brings your Cas9 and P300 to where-ever in the genome you want to increase gene expression.

So you can switch on expression in particular genes.

You can also imagine fusing these to histone repressors, which can deposit repressive marks such as histone external methylation marks or even DNA methylation marks. You can fuse dead Cas9 into DNA methyltransferases and deposits methyl groups on DNA directly, shutting off that transcription.

Again, another way of very, very long term manner switching off expression of particular genes without having to worry about double strand breaks.

Now I’m gonna move to the clinic in this sort of final moments of time. I really want to give you an overview of the technology and where it is in the research stage, but everything I’ve spoken about so far is progressing in the clinic.

And now these 31 CRISPR clinical trials, that I mentioned, are ongoing in oncology for various leukaemias, lymphomas, renal carcinoma, solid tumours. There’s a sickle cell disease and B-thalassemia programs from CRISPR Therapeutics, which are probably nearing approval this year. Ophthalmology, metabolic disease like mucopolysaccharidoses and type one diabetes. There’s monogenic conditions hereditary angioedema and TTR amyloidosis, as well as familial hyper cholesterolemia and people are even using CRISPR to treat infectious disease.

So I’m going to take you note through some of the some of the upcoming clinical programs.

So first off, and this is the most advanced one, is the treatment of sickle cell disease. So this is an autologous therapy where patient’s own sickle haematopoietic stem cells are extracted, gene edited outside the patient, and then given back to the patient in the form of bone marrow transplants. And currently over 50 patients have been treated – in so far, I think – but one is completely transfusion independent, truly revolutionizing the treatment of sickle cell disease and beta thalassemia.

TTR amyloidosis is another disease actually, that we went through in the previous talk which is currently treated through a small interfering RNA drug called Onpattro (patisiran) from Alnylam. However there’s now a more permanent approach that’s coming through in the clinic, which is to knock out the TTR gene using CRISPR.

Now this is a lipid nanoparticle that features CRISPR mRNA and a TTR-gene-specific guide RNA. When this LNP enters the liver and hepatocytes within the liver the Cas9 mRNA is translated into protein into the Cas9 protein. It binds to the TTR specific guide RNA, cuts the TTR locus, non-homologous end joining occurs, and this disrupts the TTR gene decreasing its expression and really treating these patients.

So patients who had this, as you can see in panel B on this figure, have seen almost up to 80% to 90% reduction in the serum concentration of TTR. And this is really revolutionary because this is a once-and-done treatment. The patients don’t have to return for a follow up and they never need to take the drug again.

In the final minute of this talk now, I think I’ll just move on to thinking about some ethical considerations.

So I want to talk about the economics of these once-and-done therapies.

So any novel therapy usually commands a really high price tag. It’s usually in summation that in R&D, most drugs fail. And the one that works has to pay off for all the programs that didn’t work.

But, when you have a once-and-done therapy the economics are become of even more even more scary or have a really high sticker price. For instance, some of these once-and-done therapies, people talking about prices in the region of two to five million dollars. You can put this in the context of the lifetime cost of a sickle cell patient in the US: might be around $2 million. So you can argue potentially that they’re worth the money, just at sticker value.

But really people are trying to come up with innovative models for paying for these treatments.

So some companies are thinking, spreading these payments over a number of years with any payment of subsequent instalments dependent on the therapy fulfilling that promise of actually being a once-and-done therapeutic.

And cost issues are really a real problem to access. For instance, a company called Bluebird Bio has an approved beta thalassemia treat. But it’s not being commercialized in Europe because they couldn’t find anyone to pay for it, which is a real shame.

Some ethical considerations here, just that you have to think about the fact that many monogenic diseases often are prevalent in areas with poor access to healthcare.

So ex vivo therapeutics can be very difficult to deliver, and it’s really, there’s a big push to get LMP and in vivo therapeutics in, for instance, to treat sickle cell so that they can reach patients who need it the most.

I think one I’ve been thinking about a bit is that once-and-done therapies, it may be impossible to a upgrade your treatment. So how good is good enough? If you have an AAV therapeutic it’ll often make a subsequent dose less effective because you generated antibodies to that viral vector. You want to be sure that the treatment you are getting is the best possible treatment, which you have to put balance against – as a patient – against the fact how rapidly you get that treatment. You can’t just keep waiting until the next model of therapy like you would with an iPhone.

Additionally, you’re not gonna get repeat bone marrow transplants with novel therapeutics. People advance with treatment of various haematopoietic stem cell diseases.

So there really, I think there’s some thinking needs to go into how often you can get these therapies.

And finally, just to touch on the ethical issues of germline modification and something you might have seen is He Jiankui, an individual in China who under the radar performed a CRISPR deletion of CCR5. As you might know, CCR5 was the co-receptor HIV infection. And he did this in IVF embryos of patients where one of the parents was HIV positive. He was fined three million yuan, sentenced two years in prison. But he’s actually recently been released and actually unfortunately has been given the lab of his own.

Germline modification in general is fraught with ethical issues for various reasons to designer babies, or what can you modify? In the case of CCR5 deletion it’s really unethical because the safety’s unknown and the risk of transmission to children from parents who are HIV positive is truly minimal.

And in this case, it wasn’t even the mother that was HIV positive, it’s the father. So really minimal risk for transferred during birth and really does not filter the do-no-harm criteria.

After this event, a group of world-wide experts in CRISPR genome editing published a sort of moratorium, what’s called calling for a moratorium on gene editing for germ line modifications.

And with that, I’d like to end the talk.

Thank you for listening and happy to take any questions.

Hassan Shakeel: Thank you very much James for that talk. Extremely thorough as always. Going through all the sort of modern uses of CRISPR right up to, I guess the bleeding edge.

So I’ve got a couple of questions to get us started and if anyone does have any more, please feel free to post them in the chat and we’ll filter them as needed.

So the first question for you, James, is as follows.

One problem that we can foresee with, sort of, a CRISPR/Cas9 system is off target effects. Because there are certain areas of the genome that have a degree of complementarity or a degree of repetition, especially high repeat regions.

What sort of techniques do we use apart from just, making longer, more complimentary sequences, to target these and to work around this problem?

James Patterson: Yeah, it’s a good question. Targeting the repetitive regions, I think is, it’s gonna be difficult for a long time. I don’t think there’s a way to get CRISPR to work on regions that are truly repetitive.

There’s a company called eGenesis that is actually targeting repetitive in regions in pig genomes to delete endogenous retroviruses in pigs and [they’ve seen] CRISPR enzyme will go after and cut 65,000 sites in the genome.

It’s very difficult to get site-specific if you have repeats.

One approach is that people have made Fok1 versions of the CRISPR/Cas9 enzyme. And this, the end would double your specificity. So you’d have 40-nucleotide specificity instead of 20-nucleotide specificity.

Going beyond that, if the sequence is repetitive, I think it’s, even you would have a hard time figuring out the right one unless you could look at, figure out exactly what chromosome it on, but that’s sifting through genome sequence, you wouldn’t be able to.

Hassan Shakeel: Yeah, as I thought it’s, it’s going to be a massive problem going forward, but certainly there’s a lot of uses of CRISPR anyway.

Okay. My next question to you is one that I think has quite a lot of clinical implications.

Polymorphism.

So polymorphism will alter the binding site, right? So it can potentially not create an off-target effect. It just creates no-target effect. Is there a thought of having multi avid CRISPR Cas9 systems all around the same area to target a specific gene.

James Patterson: Yeah, so also good question. Actually that online tool I showed always highlights if there’s a single nucleotide polymorphism so present in your guide sequence so you can try and avoid them.

I think making it multi avid is difficult because usually then you’d be decreasing the specificity of your enzyme and then you get your off targets, which you want to avoid.

I think the key is that if you’re trying to knockout your gene, you actually have quite a wide expanse of genome to target. And you can find regions that don’t feature some nucleotide variants and target those. In the clinic, if you’re creating a clinical products in idea, these guide RNAs at high quality level cost in the region of billions to manufacture. I say if you had to make manufacture multiple copies from a cost-goods perspective, your therapeutic, it’s very expensive.

So usually try to find, make one, pick a really good one and move forward with that.

Hassan Shakeel: Yeah, absolutely. Okay, so another question.

So you mentioned a couple of delivery systems such as LMPs. But all of them I guess, share a somewhat of a common problem, and that is immunogenicity as you’ve highlighted already.

Now, immunogenicity, if you have a, the patient’s own endogenous system, that’s then cell that’s been modified is probably reduced.

But are there any other ways of reducing immunogenicity that people are working on?

James Patterson: So LMPs themselves, they’re a feature of protein components.

They’re inflammatory but they’re not necessarily immunogenic.

I’d say that their main issue is they at the moment all go to liver.

We personally are working on a non-immunogenic delivery vehicle example, which I’m excited to talk about at some point.

But you might have seen these recently, people are actually trying to turn endogenous retroviruses into delivery vehicles. So your genome is full of viruses, in principle those viruses are non-immunogenic, because you tolerate them.

So people are trying to harness endogenous viruses to make delivery vehicles that are non-immunogenic.

It’s a big new story yesterday or the day before about a company called Aria therapeutics that just launched with a very large funding grant to do just that.

So that’s one potential route.

There’s also exosome. So exosomes are small microvesicles derived from our cells, which are human. And they’re also critically non-immunogenic.

Hassan Shakeel: Thank you. And another question, and I think I already know the answer to this from our talks previously, but I think this is for the wider cohort.

As you mentioned, it’s difficult to get high yield in CRISPR. Really difficult. And that’s with one or two CRISPR modifications. But there are times where multiple modifications are needed to the same cell for multiple reasons: you may have a sort of tumour that needs multi, target, to combat. How does one increase, the transduction, one might say, of CRISPR/Cas9. The fidelity of it.

James Patterson: Yeah. So I’m gonna take this as a multiplex – “how do you get high enough efficiency to do multiplex?” question. I think one really nice example that came out recently – I had a slide on it but I just didn’t have the time to show it – was on, was engineering T-cells to target at T-cell leukaemias, and that was done at GOSH with UCL. It was in the news as well. And these authors deleted a gene on the T-cell that is present on normal T-cells. So they could target their T-cells with a chimeric antigen receptor to kill endogenous T-cells.

If you don’t do that, then your Car is going to target yourself, so you end up with something called fratricide you target yourself.

They also deleted NAC class 1 expression and endogenous TCR expression, as well as a gene called CD52, which is a target for an antibody commonly used to treat T-cell lymphomas.

And they use base editors. So because they used base editors they didn’t have to worry about the translocation effect because it only creates a single-strand nick. And this is why they’re able to do that and, with very high efficiencies, these enzymes now are really pushing out to 80%, 90% efficiency.

So if you multiply that through, you still happen with these sort of 60% to 70% of your cells featuring all the knockouts you wanted to include.

With stem cells, it’s a bit of a different story because you have longer time to culture them, and you can do your edits sequentially. So you can do a cut, wait, cut-gets-corrected or -disrupted, and then you can do another batch of CRISPR onto those cells and now already feature the original knockout.

And with that approach, honestly, for each knockout, most guides, you can get 95% to 100% knockout efficiency in your pool with modern approaches.

Hassan Shakeel: Perfect. Thank you very much again, Dr James Paterson. As always, it’s been a pleasure to have you talk to us for the second time for our LinkAGE webinar series.

These will be monthly talks. I’m pleased to announce the next one is going to be about SMA. Particularly SMA type 1. So that’s spinal muscular atrophy, and the exciting gene therapy nusinersen that is used in it. And that is going to be delivered by a paediatric neurology consultant from Great Ormond Street Hospital, Dr Louise Hartley.

Thank you very much again, James.

And if you would all be so kind as to fill in our feedback form, it’s really helpful for us to, design the series and make any alterations as needed going forward.

And we hope to see you all again soon.

Thank you very much.