This is a machine transcription and therefore it may contain inaccuracies, errors, or mispronunciations. Notice an error you think needs changing? Please contact the Bitesize Bio team using this form: https://bit.ly/bsbtranscriptions

Welcome to CRISPR Unedited,

a bite-sized bio podcast hosted by an Anthony Adamson.

In this episode of Crisp Put Unedited, I'm joined by p Hanson,

scientific coordinator at the Cell and gene therapy core at Lu University.

And we answer your practical questions about crispr learning about the

differences between CRISPR and RNAi.

Often the on target efficiency is higher. With CRISPR Eye,

we've actually had several, uh,

occasions where we've had complete turning off.

We discuss how to show if a gene is really knocked out, you

Need some functional tests.

So it could be that the protein is absent or some downstream. Uh,

regulators, if you have some genes that you know are regulated by this,

you can look at that.

And we talk about off targets

Making the same edit,

but having two different guide iron because then they completely negate the

problems. You have to look at both of them. They have to get the same phenotype,

and if you get the same phenotype, that cannot come from an off target effect.

All this and more in this episode are CRISPR edited.

Hello everyone and welcome to this podcast from Bite-Size Bio.

My name is Anthony Adamson and I'm run a core facility called the Genome

Medicine Unit, where as you might imagine,

we use CRISPR Cas nine an awful lot to engineer cultured cells.

We make novel mouse models,

and more recently we've been using this technology to make Jetta modified flies

as well. Uh, I'm joined today by Pier Johansson,

who is a fellow member of the CRISPR Core facility community. Um, so Pia,

would you just like to take a minute to introduce yourself, please?

Yes. Hello everyone. Uh, my name is Pira. I am in Lund in Sweden,

and we have a core here called, uh, cell and gene therapy core.

And we are actually producing tools for such things,

and most of it is research based.

So we make CRISPR edits in i p s cells,

and I also design CRISPR red for all other cell types.

We do cloning and vectors as well. So we also produce, uh,

lentivirus and AAVs. And soon we might also set up mRNA, uh,

production in our core.

Fantastic, thank you. Uh, well, a couple of months ago,

bite-sized bio hosted an online CRISPR method symposium, which pier and I both,

uh, presented at. Uh, now let's face it,

CRISPR is a bit of a game changer in discovery science. It's, well,

it's enabled cause like ours to be established,

and it allows us to build better and more representative models biology and

disease. So,

no surprise that there's enormous interest in this online symposium. And,

you know,

lots and lots of people out there were really excited to learn more about the

technology and many people wanted to get CRISPR up and running in their own

research. Uh, and many more of you were looking for advice and tips, uh,

and hints on maximizing your success with the technique.

So as a result on the day, we had loads and loads of brilliant questions, uh,

but ultimately far, far too many to answer in that, uh, online symposium.

So we thought we don't wanna disappoint everyone. Um,

and it might be a really good idea for peer and I to get together and follow up.

And in this CRISPR Clinic podcast, we're gonna go through some,

some of the many questions you submitted, uh,

both during the conference and afterwards and try and offer some advice and

guidance. You know,

the type of things that we might actually do if we were approaching, um, uh,

your projects in the laboratory. Uh, just a quick reminder,

there's loads of really useful resources and hints and tips and tricks and blogs

over it. Biosis bios, crispr hub as well, uh,

for you to look at in your own time.

And for those of you who didn't attend the symposium, I,

I believe the talks were were recorded and you'll be able to access them over

there too at some point if you're interested. So, uh,

I hope we find this CRISPR clinic useful. Uh, if it's successful,

there's enough interest,

we'll think about putting some more in over time as well.

So there's a constant source of expertise for you to get in touch and ask for

help about. Uh,

but I think at that point in time it might be a great idea to go to the

questions. Um, so Pierre,

I'm gonna come to you first and you have to excuse me to look to the side all

the time 'cause I've got a,

an alternative screen to be working from with all the questions

On that. I do too. I'm sorry.

Uh, so we've got a question from Claudia who's saying,

I'm gonna start my knockout experiments. And, um, from the look of the question,

it looks like she's trying to make her cell lines constantly express Cass nine

using the lentiviral construct. And she's asking, uh,

how can this be done and does the Castine need to be in activated at some point

as well?

Well, it's a good question. Uh, the,

I'm gonna say something very often during this, uh, q and A session.

It's like one of all, it depends on the locus and yeah, the one is,

it depends on the cell a little bit. But, um, so here it also depends like,

you know, sort of what type of knockout experiments. So, but generally, um,

you can, let's put it this way, you can inactivate it.

So there is this Kama Cai kakai or something.

It's called KA Cass nine, uh, that was done where you also,

at the same time as you introduce your other guide R N A and the Cass nine,

you also introduce a guide r n a against Cass nine.

So that means that it will actually stop being expressed in the cells.

There is however,

not so much evidence maybe that Cass nine is actually toxic or bad

for the cells.

So I mean this is more if you want to take it therapeutically later,

but for research, often it's not a huge problem,

but it does of course keep cutting. So you know,

you might not end up with the same population originally as you had a little bit

later. So, uh, if you can't go clonal, uh, or even if you can go clonal,

you might wanna know that this is what you've got and it stops here.

In that case you can inactivate it with a guide n a and you can of course also

if you just want to do a knockout experiment,

you might not have to use a LTI virus, then you just use, uh,

another type like a plasmid or mRNA based where you have transient,

uh, a transient effect because with a knockout you don't need it to be there for

longer.

That's like different if you want to do CRISPR or CRISPR eye often you want to

have it there all the time. Uh, but when it comes to this,

you could actually potentially, uh,

choose to use another one or even use r and p via,

via nuclear affection for example.

Then you don't have to worry about those things.

Yeah. So I suppose if you change the delivery modality, like you say, and uh,

and that's something that you'll hear me say an awful lot,

all all about delivery. If you change that modality,

do you need to have the Cass nine being expressed all the time?

'cause that's what lentivirus will do. Uh,

lentivirus will integrate in the genome and will give you constant production of

the Cass nine protein. I suppose just to touch on what you said there, you know,

Cass nine by itself relatively inert,

it doesn't really do any damage in the absence of guide r n A.

So suppose one thing you could do is you could lentiviral express the Cass nine,

but just transitively deliver the guide iron names to the cells as well.

And that might be a way to minimize constant activity of the,

of the Cass nine in those cells. Mm-hmm. That's a good idea. Yeah,

so just mix and match that kind of thing.

I wasn't aware of this kamikaze Cass nine, but that, that makes perfect sense.

You know, you could knock out Cass nine itself. Uh, so that, that's,

that's a really nice approach.

Yeah, I liked it a lot actually. Yeah.

Uh, I think, you know,

cloudy action has a couple of follow up questions as well. Um, and, uh,

one of them all, they all really relate to lentivirus essentially. Mm-hmm.

You know, this idea of an off switch, um, this idea of off targeting,

if you've got the Cass nine, the guide and f too long. So yeah, I,

I think it is a good idea to minimize the window of the actual editing,

but the Cass nine can stay expressed and probably won't have too many effects on

your cells. Um, in terms of that window of expression of the, of the guide,

R N a 24 hours is generally speaking enough to get editing.

So you could transfect in the guide r n A as an r n a molecule into the Cass

nine expressing cells, and that will, that will achieve the editing. Um,

I don't know if you get experienced this pier, but there is,

there is definitely risk sometimes with lentiviral transduced cells that you can

get silencing of the transgene.

So you may make cells that are expressing Cass nine all the time,

but if you keep culturing them, passaging them,

they may not be suitable for gene editing after a few weeks where depending on

the locus, that's the one of the things you've already said.

But depending the locus of integration, uh,

and the cell types you're working with,

any kind of selection you put on the Yeah, the cas the CASS nine might turn off.

Have you, you've experienced that in the past?

Yeah, so I, I haven't used so much of the normal Cass nine, we dead Cass nine.

I've used a lot of vir and there we have the CRISPR eye.

So then like it's in that particular viral vector is super good.

It doesn't get very much silenced. So at all,

we've used it for like four months, you know, into organoids,

all that sort of things.

But we also used the lentivirus for CRISPR a and uh,

there it got silenced.

We used two different ones and they both got silenced really quickly or like

sort of disappearing generally. Even the next sort, they were not po uh,

not positive anymore. And then in particular,

we found that when we added the guide R N A, because crisp bras are big,

you can't have, it's difficult to have all in one plasmid.

So we first made a little cell line with, with the Cass nine expressing cells,

and then we added the guide r n a in a different one a little bit later.

And then we found that despite the fact that these were sorted and supposedly

pure, uh, they had already dropped a lot just by itself.

And then adding another virus actually made the silencing even more actually,

of the original virus, not the second one. 'cause that's a small,

very good virus. So it really depends on the virus itself also it seems,

and maybe what actually is in it. Um,

so that was a big difference between things that are actually quite similar so,

and in the same cells. So, so that always varies a little bit. And,

and lentiviral constructs are, are quite different actually from each other.

So one should try and dig around.

Yeah, that's fascinating. So I mean,

I know this is a q and A for people that questions,

but I've got a question for you right now. What were the,

what were the effectors on the, on the dead cast nine, uh, for the,

the eye and the a, you know, one, obviously one silence and one didn't

Yeah, exactly. So yeah, exactly. So it had the crab on the, uh,

on the crispr eye and on the crispr a we had VP 64 and, uh,

one that was VP R and both of them were, um, were silenced.

And also it made pretty bad viruses, actually, because they're huge.

They were just on the border and maybe a little bit over.

So it was quite difficult actually to, to do that experiment.

Okay. That's really interesting because we, we've done a bit of crispr eye,

a little bit of crispie. Yeah. Not, not as much as you guys.

So that's a really interesting, um, observation.

Something I'll be looking out for as well in our own work. Um, okay.

So I'll jump onto the next question now. So this one, um, is, uh, for you again,

Pierre. Uh, what are your thoughts on the effect of off targets? Um, are you,

are you concerned about, uh, hitting other genes, genomic instability? Uh,

and especially with re with a perspective here by look of it as far as the

clinic's concerned as well?

Yes, that's a very clinic,

this is clearly a problem and that's something that you really have to

investigate like full on later. So off targets are a difficult thing. Uh,

I've been in several discussion rooms lately where people feel like, yes,

it's probably something we should consider, but it's hard to do.

So I have two, two viewpoints here a little bit, which I feel both of them.

One of them is that maybe we don't worry too much about them because if you,

unless you have in a specific gene, you know, actually inside the gene,

that seems to be important for your cell type.

And obviously we try to avoid that. But otherwise,

a small indel somewhere in the genome is probably nothing really to worry about.

Because the thing is that when you do clonal expansion of,

I work mostly with I p s, when you do the clonal expansion there,

you get really quite a lot of changes in the genome just because of that.

So when we do this molecular karyotyping,

there are really big things that are different. First of all,

from the reference genome, of course, you know,

that we are not the reference genome,

but also like a little bit from the parent clone.

And what we do there is that anything that is like a deletion or a loss of

heterozygosity or something like that, that is up to 400,000 base pairs,

you sort of accept. So, so knowing that, I feel a bit like, okay,

maybe we don't care so much about like a little indel in a random place,

but of course you never know.

But this is also the reason why you use first of all, more than one clone.

So that, you know,

they all have slightly different variations in the molecular karyotype.

But also why I always recommend, but this is not always used,

is why you would use another guide r n a. So you make two different lines,

for example, in i p s one making the same edit,

but having two different guide RNAs because then they completely negate the

problems. You have to look at both of them, they have to get the same phenotype,

and if you get the same phenotype, that cannot come from an off-target effect.

So, so that is like, I think,

better ways to sort of work around it in a research setting.

Yeah, absolutely. Yeah.

Yeah. In a clinical setting, uh,

I mean obviously there you have to test it much more and maybe whole genome

sequencing is the way to go. And then once again, you have to, you know,

you have to do the parent cell and then you have to do the edited cell,

and then you have to see what the differences are in the whole genome

sequencing. And then somehow you have to be able to assess what does that mean,

you know, these little changes, does it mean anything, you know,

so that will be a more complex task. Um,

but of course, also we have to remember that when it comes into the clinic,

very unlikely is it gonna be pluripotent cells,

it's going to be in a specific model system,

specific cell type where you can sort of go, well, this gene is not even on, uh,

in these cells, so it's nothing really to worry about.

So then you can also be more specific. Whereas in i p s, we look at all of them.

So that's my thoughts.

Yeah. You, you, you, you touched upon something there,

which I think is a bit of a, kind of a, a dirty secret in crispr,

that we do these off-target predictions all the time using fantastic web tools

that have been set up. Mm-hmm. But they're all against the reference genome. Um,

and the reference genome is not the genome that you've got in your IPCs or in

your stem cells are are, or the cells or the mouse you are working with. Mm-hmm.

Um, so those predictions are gonna be largely inaccurate. Um, you know,

to a degree inaccurate, I should say, not largely inaccurate. Uh,

and I suppose the case of dig looking for them, um,

and you can spend an awful of time effort going looking for them.

So those kind of suggestions you just made there about making your knockout

cells with one guide r n a, and then repeating the experiment,

but using a different guide down there to make the same knockout.

That's a really nice control. Uh, and I think that's, that,

that will solve a lot of those problems. Mm-hmm. I do disagree with you.

Is this something we should be worrying about largely because these days,

you know, we can design the guides pretty well. Mm-hmm. Uh,

we can deliver them using r and p. We touched on this in the first question.

So that window of expression is really limited.

That's been proven to reduce the likelihood of off targeting as well. Uh,

and as you say, whenever you culture your cells, every time you massage them,

you'll have more mutations in there.

And that background de novo mutation rate will be ha probably higher

than the induced mutation rate you're gonna get from Crispr Cass nine. Exactly.

Um, it's something that we never used to worry about. Um, and these days we do,

I mean, speaking from the perspective of the mouse community mm-hmm.

People've been making mouse models using mouse enbr stem cells for 30 years.

And we know they acquired mutations in culture. Um, but at the end of it,

we hope to get the mouse model we wanted and then, you know, work from there.

Mm-hmm. And, you know, no one really concerned themselves with, with those, um,

issues, I suppose where mouse concerned,

if you had a chromosomal rearrangement and then an abnormal karyotype,

then you would not get a germline transmission.

The mouse would not be able to breathe forward. Mm-hmm. Uh,

but generally speaking, you know,

it wasn't something people worried about at the editing stage. Uh,

it was something people worried about at the breeding stage. So yeah,

I think on balance off tag,

it's a probably not as big a problem as people first worried about, you know,

things have improved an awful lot in the techniques and the way we approach

things. But I completely agree that when you're talking with the clinic,

you're talking about, you know, patients, you know,

what may be a really good guide r n a for one person in a therapeutic setting.

It may not be a good guide down there for someone else.

And maybe in the future we'll see, uh, pre-screening,

like say whole genome sequencing,

that kind of thing established to make sure that, uh,

a treatment for one person is also gonna be safe in a second person as well.

Mm-hmm.

Yeah, absolutely.

Excellent. Uh, okay, so let's have a look at the next questions. Uh,

so we've done the off targets, now we've done a bit of delivery. Um,

yeah, so I suppose this question here, pier, um,

what can you do about guidewire air design if you cannot decrease off targets?

Now?

I suppose it probably relates more to maybe a gene knocking that kind of thing.

Yeah, yeah, exactly.

'cause that's where you are much more limited if you want to knock something in,

in specific place, or if you want Yeah, exactly.

Do a snip or something like that. Uh, again, um, it's basically the same.

Then you basically have to do it. If you're really concerned,

then you have to do it with another guide as well. So,

and if there are not another guy, let's, there isn't another one,

there is just one and that one has high of target risks,

well then, then you have to see where they are and you have to look,

uh, in like maybe via P C R or some other method,

you have to actually look in your cells if you have an edit in those, uh,

in those locations or not.

And then you have to decide whether or not those locations according to

your knowledge, could cause a problem. Of course, uh,

you don't really know that. I mean, people are like, oh, it's not in the Exxon,

it's fine. But I mean, I used to,

I worked a little bit with transposable elements and things, so you know, it,

it, there is no such thing as a safe locus really, but I mean,

so you just have to make an, an assumption like this is going to be okay.

But also just because there are off predictions doesn't mean that it's cuts

there. Uh, so you can look at the ones that are look seems most scary and,

and look there if it actually has cut. But the thing here also is to,

as Anthony also said,

to limit the amount of time that the Cass nine is in there. So here,

r and p is, is really a good way to go because then the less time it's in there,

uh, the less of target, the fewer of targets you're gonna have.

So I think that's the, the two ways about that.

I'll,

I'll just expand on that by r and p because both of you and I have said it quite

a lot. And just for anyone watching that's Oh, yes, familiar. Uh,

this is the rib nucleo protein method. Um,

we talk about delivery modalities quite a lot.

This is essentially rather than genetically encode the Cass nine and guide r a

and say a plasmid or a virus, we simply buy Cass nine protein from a company.

We buy the guide r n a from a company, we mix 'em together in a test tube,

and that's what we transfect. So it's just the r n A and the protein.

And as PIs highlighted, really, uh, uh, efficient, um, delivery method,

really small window of expression and very high ONT tag activity and

reduced off tag activity as well. And I'll, I'll be honest with you,

we use this method across the board. I, I,

I don't think there's many circumstances at all these days where we use some of

the other delivery methods. It's pretty much r and p all the time for us. Um,

I'm gonna go back to what you said about, you know, detecting those off targets.

So you mentioned, you know, you can do P C r, you can do sequencing,

that kind of thing.

One of the challenges I think that's out there is there are some types of off

targets that might be a chromosome translocation or a big deletion,

that kind of thing. And they're a lot more difficult to detect Aren. I mean,

how would you go about looking for those?

Or would you even bother looking for those?

Hmm, yeah. I mean, I think, uh, in the end, uh, once you're done, like,

you know, you have, you, you need to check the karyotype, uh, and, uh,

you can do that by molecular karyotyping or, or GB banding.

And I think, uh,

a lot of people suggest that you need to do both because they're not detecting

the same things. Uh, I mean,

this is in particular with I P s where we're very concerned about the fact that

they have to, to stay completely the same. Um, so,

so that one, it's one way to, to look for it. Um,

what was the second part of your question?

Uh, things like chromosome, well, obviously you've launched deletion,

the chromosome translocations essentially. Oh, yes.

Yeah, yeah, that's right. Yeah. So a lot of things you will see there.

But what you also need to check, actually, uh,

or maybe this is something that's come up recently, is this, uh,

there was a paper showing that a lot of the things that we think are homozygous

edits are actually hemizygous edits. So one part,

like one let's just be cut out.

And obviously for a lot of things that could cause problems,

for some things maybe not, maybe it doesn't matter, it depends on where it is.

But then what you can do there is that you have to find a way to,

to look for that as well.

And often this type of snip arrays that molecular karyotyping, uh,

will find that, uh, the, the carrier,

like G GB banding will probably not find it, but the other ones might find it.

But if you're specifically, you know,

you can look around your edit site and you have to do some sort of serial

quantitative P C R,

either digital droplet or some other quantitative way or some genome

sequencing. If you do N G s,

you can do paired sequencing and then you should see the difference. Um,

we don't know how common it is anymore. I mean,

in that paper they showed that it was quite common,

but the majority of that was happening when you delivered Cass nine with the

plasmid. But I mean, uh,

a couple of of these networks are like the core U stem, for example.

We have different core, uh, facilities that work with I P Ss, uh,

have gathered in Europe. And,

and what we're gonna do there is that we're gonna look at all our lines and see

if we can actually detect this in any of them. And the, uh,

with the ones that are made with r and p, for example, if it's a small change,

um, then we will see if we can detect it anywhere. So may,

if it is a concern or something that needs to be introduced to the standard, uh,

quality control, uh, sort of, uh, panel or not.

So that's something we are gonna look into a little bit. When you use plasmid,

there's a bigger risk for sure,

but often that means that you often want to introduce something that is

he heterozygous anyway. Uh, and in that case, you of course sort of,

yeah, often when you see that it's heterozygous, that's good,

that means that you have the other alleles still there.

So maybe that's the way to go also with, uh,

with big insertions that you go for the heterozygous 'cause then you know that

the other one is still there.

It's a little bit tricky because normal sequencing,

like a normal sango sequencing will not detect it.

It just looks like it's homozygous, but in reality it's, uh, it's homozygous.

So that's a big new thing that has happened.

But that's like a weird thing 'cause it's sort of not really an off target.

It's an toxic on target. Yeah,

Absolutely. Yeah. No, it's, it's fascinating to hear you say that.

Obviously this is still a relatively new field and mm-hmm.

It's great to hear that, you know,

these communities are getting together to establish what is the quality of

control, what should we be looking for? Yes.

And I certainly don't think that's set in stone yet. Um,

and just to go back to the, this on target deletion, you know,

this big deletion and wait, obviously it wipes out a primer site.

So if you do a A P C I, you think, oh, I've got a homozygous, um,

colony or homo homozygous mouse. And we've seen this a lot in mice as well.

Mm-hmm. And of course,

in mice it's actually easier to detect because when we have our founder mouse

and we think, oh, we've got the homozygous point mutation, um,

we breed it forward with the wild type mouse,

and we only see the point mutation in half the, the litter.

So clearly we know that, um, it was not homozygous founder,

it was probably heterozygous,

and we just were not detecting that change as well. Mm-hmm. Uh,

so I think actually in, in the mouse community, this has been, uh,

well observed for, for a number of years now as well.

And now we are seeing it as well that in people that working cell lines,

I have to say, you know,

sometimes it's easier to make a mouse than it's to make a cell line because we

can breed out a lot. You know, if we have potential words for off targets,

we can breed them away. If we want to have, um, homozygous or heterozygous,

we can establish our breeding pattern in such a way to get those genotypes. Uh,

and obviously we can't do that with cells in a dish. So Yeah,

in some ways it's actually easier these days to make a mouse than it's to make a

modified cell line. It's a bit counterintuitive. Yeah. Um, yeah. Uh, but yeah,

it's fantastic to hear about, you know, the, these, um,

new drives to make sure the quality controls there.

'cause I think ultimately we talk about CRISPR allowing us to make better

models,

we have to make sure that those better models are what we think they are in the

first place.

Oh, yes, absolutely. Hmm.

Okay. Uh, so let's move on to the next question. So, um,

we've got a question here asking, um,

how is CRISPR different from R and AI as a technique? Um, and you know,

whether or not you should choose CRISPR over RNAi?

Yeah. Should I,

Uh, yeah, please go for it. Yeah.

Um, so the general idea is that, uh,

there are fewer of targets with CRISPR eye than there is with, uh,

RNAi or or other r n a, the other r n A type, uh,

small heins and so on. Um,

there've been papers that show that the off targets are, uh, fewer.

And also often the on target efficiency is higher. With crispr, I,

we've actually had several, uh,

occasions where we've had complete turning off. I mean,

that depends if that's something you want or not, you know,

if you just want reduction or if you want it off.

But we've had it off several times.

Like actually we see that the activation peak when we look for, um,

stone modifications is actually disappearing. Uh,

and also we don't see any R n A and RNA-seq,

but I thought the activation peak is gone, uh, is, is pretty cool actually.

So sometimes it's completely off. So it depends on what you want a little bit.

But, um, generally, uh, yeah,

maybe higher efficiency and fewer of targets with CRISPR eye.

So in terms of CRISPR eye, um, if you tag,

say an essential gene and you want to remove that, uh, with crispy,

you can't really do that, can you? Because you,

if you knock out an essential gene, you're gonna kill the cells. Yeah.

But with crisp, but crisper,

I have you experiences where you're able to reduce the level that gene so that

the cell can tolerate this and survive, but it still drives a phenotype,

something you can investigate.

Yes. Crisp rise actually really,

really good like that because you can actually dose it,

you can do it with the amount of virus that you put in. Uh,

so it's sort of dose dependent all, and also different guides like,

so we decide the guide design the guides within a certain region,

and often there's one that just doesn't work as well.

So then you can use that one. So it's actually quite good.

You choose maybe three different ones and you can choose different amounts of

your, um, of your virus as well. And then you can totally dose it.

Excellent. No, it's fascinating. So, you know,

RNA AI was around for a long time. Um, it's still a useful technology.

I do tend to agree with you that it is being superseded a little bit by the

CRISPR and CRISPR base technology that CRISPR eye. Um,

but it is a useful complimentary assay to perform as well as the crispr. Um,

so you know, if you get the same result from the,

your RNA AI experiment as if from your CRISPR experiment, then you know,

you are, you are onto a winning result basically.

I think that's a good way of controlling

For it. Exactly.

Um, okay, the next question,

can we use primary cells like human monocytes or is this only for cell lines?

So I'll answer this because we've been doing a little bit of this. Um, yes,

you absolutely can do. Um,

there are obviously different challenges associated with different cell types.

So with a cell line, generally speaking,

you can deliver everything to that cell, then you can easily,

for most cell lines,

isolate a clone from that cell line and grow it up in assay.

Has that got the change?

You're not gonna be able to do that in the same way with primary cells.

And our experience is, generally speaking,

what we like to try and do is first of all, optimize the delivery, so get crisp,

but working as well as we possibly can do in that primary cell line. Mm-hmm.

And then once we've got that, we'll maybe screen guide RNAs and we'll say, well,

this g a is really active and knocking the gene out, maybe this one isn't.

And we'll do that kind of similar thing that you just said there.

We get a knockdown population of cells where we may reduce the protein

expression level overall to nine, you know, to, to less than, um,

10% of the original amount of protein. And again,

if that's enough to drive a phenotype,

if that's enough to make the cells change their behaviors,

and you can measure that change in behavior, then that's sufficient.

But absolutely,

it is more challenging to work in primary cells than it's in cell lines. Uh,

but you can do it. Um, if you are new to CRISPR and you haven't done it before,

I would not recommend jumping straight into primary cells. I would recommend,

you know, let's get the technique established in a surrogate model system first.

You know, you mentioned human monocytes. There,

there are monocyte cell lines out there, like TP one. Mm-hmm. Uh,

U nine 30 sevens as well, I think. Uh,

and we've successfully applied crispr both these cell lines and there's

published protocols out there. So that would be my recommendation. Um,

p i I don't hear you got any experience with, with primary cells as well?

Yes. Not hands-on, but I have designed, uh,

experiments for people with primary cells and,

and there is really important to have a good, uh, communication first.

So you ask them exactly what is it that you want and what is the timeline here?

I mean, I had somebody's like, yeah, we take the cells out of the,

out of the mouth and then we have like seven days and then that's it, you know,

something like that. So I'm like, okay, so we have to use r and p.

There's nothing else that has enough time.

You have to maybe be able to sort them because you can also use Cass nine that

is like tagged or something like that.

So at least you increase the ones that actually has the,

the components in there, even if it hasn't necessarily cut. And, and, uh,

and so you have to really talk about what's possible.

And then exactly like you said, they have to set up the experiment first,

like sort of trial version, you know,

like try it and see if you can get the cells in there.

And also look in the literature a lot, what works for your cells?

And this is always the thing I'm like,

I cannot really tell you how to live deliver it.

You have to look into is it best with nuclear affection or is it lipo perfection

we are gonna use and, and see if other people have used it.

Then you take that protocol and, and you, you try it and exactly.

Try different guides and things like that. But, but that particular one,

it worked really well. They just sort of of did it and, and, uh, it worked. And,

and that's the thing also, sometimes you just have to try. Yeah, absolutely.

And be prepared for then do method development if it doesn't work, because this,

I often say if it doesn't work, just come back to me.

There are so many other options, like,

there are so many other things you can do, so we can try something else,

but this is what I, you know, sort of think might be the best one to try first.

But it is also important, like you say,

that you might not need a hundred percent of the cells being like edited either.

Like sometimes people want to put back and, uh, cells into the animal and stuff,

and I'm like, maybe it doesn't matter if like,

as long as like maybe 80% of the cells show this phenotype,

you will see a difference from wild type transplants, for example. So, uh,

you know, you don't have to be a hundred percent, a lot of things is,

is good enough, you know, and you will see the, the change anyway.

So, and that's a, that's a really interesting point you made there. 'cause,

and I, and I don't know how I feel about this,

but I spoke to some people in the cancer community who when they're doing

xenografts and they're editing cells, the dish,

they don't want clonal cell lines.

They don't want a hundred percent knockout because cancer is a heterogeneous

disease. So they sometimes prefer to have, you know,

50 or 60% knockout and then graph that, uh, into, into the mice,

and then they can see what happens over time with,

with the respective populations of know the wild type cells versus the mutated

cells. Mm-hmm. So yeah, it, it's, it's,

it's that experimental design all the time and you design the experiment to suit

the model you're working with and what you're actually trying to find out.

Mm-hmm. And you made the part about delivery there as well. The,

the following question really nicely feeds into that regarding the delivery.

Do you recommend transfect lipectomy or nuclei affection instead of vial

transduction? Uh, and I think you've already answered it.

You have to do it depending on your cell. You know,

you optimize this based on what you're working with. Uh,

typically I would say we,

we generally speaking use nuclear affection rather than lip perfection. Mm-hmm.

But there are certain, some cell lines we work with that prefer lip perfection.

I assume if you are working in stem cells,

you are all nuclear fiction all the time.

We're Yeah, not always.

May may, maybe not, maybe not. Well,

We are, we are not very good at, uh,

at using like with big plasmids to put in and if nucle infection is working.

So, so it is working, but we're trying to optimize that.

So what we're trying to do now is actually optimize the system where we need

both R and p and a plasmid,

because it's the big construct to maybe do lipo perfection first,

and then we'll wait and then do the nuclear affection for the r p uh,

'cause the, the timeline is different between those components and, uh, yeah.

So it, but mostly nuclear affection,

but with a little bit of lipo perfection as well.

So you stagger the delivery, so you made you, well, yeah, so you,

you'll put the, the d n a donor in at different times to the Cass nine,

the guide r n a,

This is our plan,

this is my key plan because we did it all together and it worked. But, uh,

an extremely low efficiency. Okay. So I'm like, if it, if they're not green,

then we cannot use this, right? Like if,

and it's of course not all edits that turn on, uh, immediately in the I P S,

so we have to find a better system. And that was my thinking was that, you know,

if you nuclear effect something, the R N P is ready to go,

the plasmid needs to be amplified first, so by the time it's amplified,

the r p is gone. So I'm thinking if we put in the plasmid first, uh,

but we'll see how that goes.

No, we did that in one cell line several years ago, and it did work,

but I'm not gonna claim it was the perfect setup because we didn't do a

comparison with it all at the same time. So Yeah. Yeah, it definitely does work.

It's interesting you say that though, about staggering the delivery,

because one thing we're currently trying in, in our mouse embryos,

which we're a little bit behind the curve on, I'll admit,

is to change the d n a donor from being d n A into being adeno associated virus.

So we're actually using virus as the donor, and in that situation,

we are staggering. So we infect the embryos, um,

with the virus that contains the homology, insert homology,

and then we leave that infection to, um, pursue for about six hours or so.

And then we ate in the cast iron and the guide r n a.

So it is a staggered delivery,

of course with a a v The big advantage is that it, it's almost like a,

a self delivering modality. You know, you don't need lip affection,

you don't need nucle perfection.

It's job as a virus is to infect cells and the ITRs and the a AAV drag

it straight to the nucleus so the donor ends up in the nucleus where it needs to

be as well. And that appears to be really, um, you know, our,

our preliminary data is successful, so, you know,

we're looking to expand it on more projects.

I've heard from other people in the MA community that works really nicely.

I think you, you can do it in cell lines as well. It does work in cells and,

and, and stem cells. So that could be an alternative,

still staggering the delivery,

but just changing it from plasma D n A into being, uh, a viral D n A instead.

Yeah,

No, absolutely. And I think as, as you say, lots of people have used that,

and I think it is a very nice method. The thing is, of course,

you have to make also the AAV in between. So, and we do that,

so it's okay, but I think when we are trying to produce quite many cell lines,

uh, as a service, if we can get away from not doing that,

and also because, I mean, the a v still stays in there also for a while. So,

so we, I try to get away from that, but it is there as an opportunity to,

to deal if we have something that turns very difficult to, to deliver or so,

because it's a, it's a very smart way actually to have it.

And then it's there at high levels and uh, as you say,

it goes straight into the new. So it's, it's very clever in many ways, but yeah.

Oh, I mean, you raised a really good point about,

about those practical considerations that a, if you want to use AAV as a donor,

is that a whole new range of skill sets you have to develop in the laboratory?

And you guys are well set to do that? We, we are not, we don't package a aav,

we make Len virus, but we don't make a a v. Mm-hmm.

So we've ended up outsourcing our a AAV production for the these applications.

Yeah. And for the mouse projects, it seems to be cost effective. Um,

there is a big question mark of the whether or not we're gonna use this

technology in our cell projects,

but I strongly suspect it will not be cost effective there.

We'll need more a A V and there'll be a greater investment. And like you say,

if your plasma based approach works at, let's just say, you know,

5% efficiency and the AAV boost after to 10%,

well actually 5% you could probably work with, um, yeah. And you know,

especially selection markers and that kind of thing in there as well.

So is it worthwhile that investment? Uh,

and there's also the other aspects of using virus and that you'll have to

probably adhere to some local, uh,

GM requirements and put some application for health and safety,

that kind of thing. Um, and, you know, uh,

that could be a bit of a pain at times for people if you just wanna do one

experiment. So yeah, it's a great option to have, um, again,

tailor it to your experiment and what you're trying to achieve. Okay.

So the next, um, uh, question.

We've taught an awful lot about knocking out genes here,

but this is a question here. Is it possible to knock out a promoter?

So,

Yeah. Yeah. Pierre, do you wanna answer this one?

No, um, uh, I mean, yeah, sure.

Uh, I think the easiest way with, uh, promoter is to, uh,

you have to use two guides. So you knock out,

like you cut out the section so you don't try to make, uh,

an indel because often it's much more complicated with promoters, right?

It's not just one or two little things that changes it,

but you have to cut with two guides,

cut out the region that you think is the promoter and uh, and that should work.

Yeah. So I mean, it's that, you know, it's that how long is a piece of string,

uh, how big is the promoter? Uh,

and so the where you're gonna position those gather ass, I agree with you,

cut out the entire thing using two guide ass is, is probably the way to do it.

Um, we've cut our enhancers before, but mm-hmm.

Those enhancers are quite clear boundaries. You know, we knew from, um,

the genome browser dataset where the enhancer very likely start and ended and

'cause it was in a intergenic region,

we could be a little bit relaxed about where our guide RNAs were designed.

So that made it relatively simple.

We also did it by putting in a removable selection marker. Uh,

so we had a homology flanked to repair template that removed the entire three or

four kilobase of enhancer and it had a pg k cure mycin gene in there

that was all flanked by locks p science.

So we could select for the cells that had the deletion. Mm-hmm.

And then afterwards we could put pre recombination in those cells and remove

that selection marker afterwards.

So basically gluing the genome back together again, um, after the, the,

the recombination to get rid of that enhancer. That worked really effectively.

But in terms of promoters,

if we're thinking in terms of the sequences that regulate gene, uh,

gene activation that close to the gene of interest, you know,

right next to the gene, then you're gonna be a bit careful about, you know,

where you design those guide RNAs. 'cause you might mm-hmm. You know,

you might make other disruptions as well. Yeah. Uh, that,

that you don't wanna make. I suppose in, in that circumstance,

would you prefer to go down the route using CRISPR eye where you can modulate

the promoter rather than knock it out?

Yeah, exactly. I'm thinking that also like how often, I mean,

it's not so often that you just want to get rid of the promoter,

but then also have the rest of the gene intact. I mean,

I suppose that could happen, but, uh, then if you, I don't know.

I'm trying to think of a reason to do that, but I mean,

if you want to like exchange to promoter,

then that's sort of easier in a way actually.

'cause then you just do homology directed repair there. But, uh, exactly.

If you wanna keep the, keep it intact, uh, then it's quite hard. Exactly.

Then you have to just cut and be hoping that your guide cuts in a very

predictable way, uh, around there. Um, but yeah,

crispr eye for sure there as well. Then, then you just sort of turn it off.

But if, if it's just a turn off that you want,

then you can do that in other ways.

You can just cut out the transcription start site and, and things like that.

That is much easier. So it depends on if, if you're studying promoters yes,

then you have to do it and then,

but then one can also do it homology repair thing actually there where you cut

and then you put in a nonsense sequence or something like that. So

Yeah, absolutely disappear As, and again, I,

the question isn't phrased it this way,

but suppose maybe you're interested in a particular transcription factor bin.

Any site in the promoter you've already determined and you want to know how the

gene turns are off, then yeah, you could do maybe, uh,

a single guide r a in indel to disrupt that binary sequence to stop that, uh,

working. So you're not knocking out the promoter per se,

but you are changing the way the promoter will respond to different

stimulations. And that's a really nice technique.

That's something you could do quite easily, just a single guide, N H E J,

you know, disruption of the transcription factor binary site. Uh, so yeah,

we've done that as, as well, and that that will work quite effectively. Hmm. Um,

okay. So the next question, uh, I'm just conscious of time right now, so we'll,

we'll, we'll try and wrap through the last few questions here. We,

we did have a lot. Um, so what is the percentage of cells that, that, uh,

are gonna get an indel on both alleles? You know, I mean,

this is again a bit of a difficult question to answer, but in your experience,

how frequently do you get editing on both alleles?

Yeah, so that totally depends on the locus, but when we do r and p

And yeah, it, it's, it's pretty common actually.

In some ways it's more common to have homo homozygous

deletions than it is to have heterozygous.

Heterozygous can actually be harder to get.

So sometimes if we have a very high editing efficiency when we look in the bulk,

so we look at like all alleles and then we get maybe like 90%,

then it's more common. I would say that you get wild types and,

and then the others being homozygous and actually getting a heterozygous in

there. So there seems to be some level of all or nothing in there. Yeah.

And we found that one way to do it, if you get too many homozygous,

is to introduce a wild type template as well at the same time to,

to sort of reduce the amount of, uh, homozygous clients that comes. So,

so that's normally less of a problem than the other way around.

Yeah. So it, it, I mean,

it's great to hear you say things like that because that's exactly what we've

been doing. And same our experience too,

if a cell receives the CAS nine and the guide r n a, generally speaking,

both alleles will get caught. Mm-hmm. Um, and if you're trying to make, say,

point mutations,

we've had some guys come to and say we've got a disease develop point mutation

that is heterozygous in the patients heterozygous cell line.

What we don't want to do is get that mutation we want on allele number one,

and then knock out allele number two. We don't want that.

And we've done exactly the same thing you've said there.

We supply two repair templates, one that will encode the, uh,

mutated sequence and the other one that will basically rebuild the wild type

sequence. And quite often we put you, I dunno if you put like a, a,

a synonymous mutation there to stop the pite, that kind of thing,

so that you get a HDR R event on both alleles. Mm-hmm.

But the template used on each allele is different,

so you end up with a heterozygous cell at the end of it, and that that's,

that's very effectively for us as well. Uh, but I completely agree, you know,

if you are looking for a heterozygous change,

that's actually more difficult to achieve than a homozygous change quite often.

I know, I know.

Now, again, to come back to the example of using mice,

it's great because we mice, if we, we, you know, we will often see,

say we're trying to knock in G F P onto an on one allele,

we'll get that knock to work in allele number one,

and we'll have adle on allele number two, doesn't matter in mice, we breed the,

uh,

knocking allele forward and we gen attack the pups for the ones that have got

the G F P and we just ignore the other ones. Yeah. So yeah,

definitely more of a challenge in sales. Yeah. Okay. So,

um, that's the next question.

Is it enough to prove a knockout via P C R?

No,

No.

I would, I would, I you need some functional tests,

so it could be that the protein is absent or some downstream, uh,

regular later. If you have some genes that you know are regulated by this,

you can look at that, uh, some in vitro assay, something like that. Yeah,

absolutely. I would say, what do you think, Anthony? Yeah.

Oh, absolutely. Yeah. Uh, I mean I think we're,

I think there's a bit of a concern of mine is that we're gonna get people and if

a couple of years time publishing knockout cell lines and they're not knockouts,

um, they think the knockouts because the genetic disruption's been there,

but that just a disruption does not necessarily mean the protein has been lost.

No. We've seen a few studies over the last few years now where, you know,

we've got these genes in our cells through evolution because they're useful to

us and it looks like the cells are desperately trying to hold onto it in,

in some circumstances. And they can do some really strange things like,

you know, skip over the, the exome that contains the indel. Um,

they can make artificial exons in the middle of introns that will allow 'em to

start producing the protein products. Mm-hmm. So yeah, functional validation is,

is vital. You know,

Western BLO may be using antibodies directed against different region of

protein, especially N N C terminus.

If you see a truncated products on your western block,

that could be a residual protein being expressed that retains some function.

So your cells are not a bonafide knockout. So yeah,

it is not enough to just do PCR alone and sequencing the PCR and sequencing is a

great indication to, to, um, let you know your CRISPR working,

but the end product, you want the knockout cells,

it's not an indication of that at all. You know,

functional tests are really important. I agree. Uh,

and there's a follow up question from the same individual who's asking,

you know, how do you, uh, what,

how do you approach cells that have got different aneuploidies, different, um,

you know, K types, that kind of thing. So there's multiple alleles to target.

Oh, you mean like in some sort of hala cells or something?

Yeah, exactly. Yeah.

I dunno, I, no, I mean, I suppose it's just the,

the same there somehow you just have to like, uh, get all four of them.

Uh, you know, you just have to make sure that it's, it's all gone. Like,

it doesn't matter where it comes from really,

like which allele it is and all that. You just have to,

it should all work the same. So if there's more than one copy or two,

then it should work the same. And uh,

and then you just have to check that it's gone.

Yeah, absolutely. So if you've got four alleles and on one allele,

one you get a minus two deletion, allele two, you get a plus five, you know,

as long as you've got frameshift mutations on, on each allele. Mm-hmm.

And then you validate it as we've just said.

Then with the protein level expression as well as, you know,

make sure you do that functional validation, then it is possible.

And we've already highlighted that if you get the Cass nine and the guide iron

into the cells, generally speaking, every LE gets targeted. Mm-hmm. Um,

so it is, um, I,

I wouldn't actually say it's more challenging to make a knockout in cells with

abnormal karyotype. I think, you know, it works reasonably well, um,

for knocking, obviously, you know, you are less likely to get that knocking, uh,

to happen every single allele. Um mm-hmm.

Do you need an every allele if you don't have an every allele,

do you need to think about what has happened to the other ones as well? So,

you know, these are, these are bigger questions actually,

and ones that you may need to factor into, again, that functional validation,

what you're doing afterwards. Okay. So, uh,

I'm gonna move to this last question now. Um, and this is about Cass 13,

and someone's asking, when would you use Cass 13 instead of Cass nine? And, um,

in, I'll answer this, it is a great opportunity for me to say, um,

this is a subject beyond what this, uh,

q and A is set up for about CAS nine g and editing. Um,

we've been using CAS 13 ourselves and it's great for r n a targeting. Uh,

and there's so many derivative applications of CRISPR based technologies, um,

for, for different approaches. And you, we've talked about crispr, Ike,

and how we can knock down gene expression, uh,

using crispr i Cas 13 can be used in a similar kind of way, you know,

so we degrade the transcripts being produced. You may not full knockout,

but you may get a nice reduction. Um,

there is a fantastic blog on the bite-size bio website on using CAST 13 and

how you approach that with some extra information, um, um, about, uh,

how you might set those experiments up.

But I actually think that actually brings us to a great endpoint and allows us

to, to stop this q and a now. Um, I'm pretty tired. I about you, Pia.

Uh, but it's brilliant. It's been some fantastic questions.

It been really great to, uh,

to try and answer them for you and if there's enough interest, you know,

we're happy to put on more of these kind of events as well. Um, so for now, Pia,

I'm gonna say thank you for joining me today. Really enjoyed this. Thank you.

Thanks for having

Me.

Thank you very much and see you y'all later. Bye everyone.

You've been listening to CRISPR Unedited. To access more thoughts,

help and advice on crispr,

visit bitesize bio.com/crispr-unedited.