Fighting Cancer with CRISPR

Published May 23, 2024, 4:30 AM

Last year, the FDA approved a treatment for sickle cell disease using a revolutionary new gene editing technology called CRISPR. Rachel Haurwitz conducted pioneering research on CRISPR as a graduate student. Now she’s the co-founder and CEO of Caribou Biosciences. Rachel's problem is this: How can you improve CRISPR and use it to engineer human immune cells to fight cancer? 

Pushkin. One of the most important technological breakthroughs so far this century is CRISPER aka Clustered regularly spaced palindromic repeats aka the extraordinary gene editing tool that is right now making its way to actual human patience. The FDA approved the first CRISPER produced drug last December, and now scientists are trying to improve on the original Crisper to bring more treatments to market. I'm Jacob Goldstein and this is What's Your Problem, the show where I talk to people who are trying to make technological progress. My guest today is Rachel Horowitz, the co founder and CEO of Caribou Biosciences. Rachel's problem is this, how can you make CRISPER work better? And how can you use it to engineer human immune cells to fight cancer. We started off talking about Rachel's graduate work at UC Berkeley. She studied with Jennifer DOWDNA, who would go on to win the Nobel Prize for her work on crisper. At the time, Rachel's work was focused on a protein called CAST six. Is it right that you spent five years studying one protein?

I spent five years studying one small protein composed of only one hundred and eighty seven amino acids, So I was pretty far down the road hole.

I mean, are you the world expert in that protein? Is there? Do you know more about that than anyone who has ever lived?

There are probably three of us who know more than we ever wanted two about that protein.

Just give me a little hit of that protein? Well, like, what is it? Why'd you spend five years studying it?

It was my entry point to Crisper. I joined Jennifer dowdna's lab as a brand new baby PhD student in two thousand and seven. This was the dark ages of Crisper. There were three peer reviewed manuscripts that had been published at the time, so it took me about forty five minutes to get up to speed on the field. It was great, and I was joining a project headed up by a post doctoral fellow in the lab, and he had identified these Crisper associated or CAST proteins and he was trying to study all of them. Now he was able to make and study all but one. One was proving difficult in the lab, so he gave that one to me to see if I could sorted out. We did eventually sort it out, and in the end it turned out to be a very important little protein. It's actually responsible for making these small Crisper RNAs that are at the heart of Crisper biology. And so I had a lot of fun for many years really understand how that particular protein functioned, what it did, how it did it on a molecular level, and then ultimately zooming far far out how it fits into the broader use of Crisper systems.

Yeah. I mean, if you're going to spend five years studying one protein, studying a protein that's essential to crisper, and doing it in like twenty ten, is as good as good as.

It gets right right place, right time.

And just to be clear briefly, just so we have it, what is crisper.

Crisper is a technology for editing the genome. Crisper allows us to do a few different things to change genomes. We can hit the delete key. We can get rid of a gene that we don't want to express anymore. We can make a small change, maybe even as simple as a single nucleotide of DNA, and we can insert one or multiple new genes to actually give a cell new capabilities it didn't have. Before.

So just in the last months, right order of magnitude months, there have been I guess the first drug approvals sort of based on Crisper, right, tell me about those.

It's incredibly exciting. At the end of last year, the first ever Crisper edited therapy was approved by the FDA. It's now but approved by other regulatory agencies outside the US too. So this is a cellular therapy for the treatment of sickle cell and beta thalacemia. So this is the use case where you take cells, you use Crisper to change them, and then you deliver the cells as the therapy back to these patients and the vision is to try to actually cure sickle cell disease.

It's quite remarkable and really fast from when you were in grad school and this kind of wasn't quite the original work, but this early work was happening. Right. It's twelve years, which for go from a lab and kind of just basic proof of concept to a thing in the world seems wildly fast.

It's lightning speed. I'm not aware of any other life science technology that went from really important publication in science magazine to approved therapy anywhere near that fast. There are probably a few things to thank for That. One is Crisper's actually not the first genomediting technology. Genomediting has been around for a while, but the other approaches are much harder to use, and so this really unlocked a much faster, broader scale of genomediting. So there was a lot of resident expertise and capability that could be turbocharged by the introduction of chris per Gino mediting.

It's like there were people who sort of knew how to do it already and then this incredible tool kind of fell out of the sky and was like, Oh, we can just do the thing. We're doing way better exactly we're saying there were a couple of reasons. Was that one reason was there another reason.

That's one, and I think another is that there were things developed for other fields or biology well understood that could quickly be taken advantage of. So, for example, the genetic cause of sickle cell disease has been known for decades, and yet there hasn't been the right tool to do much of anything about it. And so this was sort of the perfect marriage of this incredible enabling technology and its ability to solve a biology problem that's been well understood for a very long time.

Can you give me a sense of the landscape of how crisper is being used in drug therapies Now, broadly.

Crisper is being used in two very fundamental ways for drug development. The first is basic research and the second is actually designing and doing new therapies, and that falls largely into two categories. One is the kind of work that we are doing here at Caribou, where we use crisper to actually modify or engineer cells, and the cells are the therapy. So by the time we deliver, for example, our Carte cell therapy CEB tend to patients, there's no Crisper inside of those cells anymore. Crisper is gone. It has modified the genome in multiple ways, and the cell is the therapeutic. The other strategy that some companies are using is to actually deliver Crisper inside the human body, and the idea is to try to correct a gene that causes a rare genetic disorder, and so in that case, crisper itself is the therapy.

So in that latter case, I mean that is gene therapy essentially what people have therapy, and what's what seems to be next in line what's farthest along anyways in terms of other crisper derived therapies.

Yeah, there's some very exciting work coming out of a company called Intellia Therapeutics where they're actually using crisper as the drug. So they are delivering it packaged inside these little fat particles to go directly to a patient's liver to correct a gene that causes a disease. And they are running what's called a phase three trial for one of those medicines right now.

So I feel like this is a dumb question, But as I imagine that, like, does that mean that the therapy has to get to like every cell in the liver? Like is it going to change the genome of every cell in your liver? Is that the way that works?

Thank goodness, No, that's not requite.

It couldn't be that, right, It couldn't be that most of them like what like what? But it's sell by cell. It's like that the particle hits one liver cell and changes the genome, and then another one hits another one and then is there some kipping point? Like how does it work?

It's a wonderful question, and I think there are a lot of people who sit in a lot of conference rooms staring at whiteboards trying to understand what is that tipping point? Because I think it's biologically unrealistic to think you can edit one hundred percent of cells in the liver, and if that's what's needed for a therapy, you're probably out of luck, and instead focusing on diseases where there's some model or suggestion that you know, maybe editing ten percent of the cells or fifteen or twenty percent of the cells would be enough, and there's confidence that the technology might be able to accomplish that.

Well, what you mentioned that there's a therapy and did you say phase three in the final stage of clinical trials? What disease is that targeting?

So Intellia is working on a disease called transthyretin amyloidosis or a TTR. For sure. It's a disease caused by misfolded proteins and it leads to neurodegeneration and cardiomyopathies.

That's the one in the liver.

They are editing liver cells because the liver produces the misfolded protein that causes problems elsewhere in the body.

So, okay, clearly Crisper is this wildly useful breakthrough, but it's not perfect. And your company was founded in a way to address this key weakness of Crisper as originally developed. So what is the weakness in particular that your company is focusing on?

Specificity? When I say specificity, I mean editing the one site in the genome that we intend to and not accidentally making changes anywhere else. Right in Microsoft Word, you put the cursor exactly where you want to write new text, not a mystery where the new text is going to land. Using a biological tool like Crisper, more often than not, you edit the site that you intend to. But biology is noisy, and sometimes the system lands in places you didn't expect and can make changes in places you didn't want. That could be a problem for what you're trying to do. And so our team for years has been focused on the challenge of specificity and ultimately developing new technologies to address this head on.

What percent of the time does Crisper get it wrong? It's the question I want to ask, and I'm sure that's too broad a question, But how do you think about that? How should I think about that?

It varies dramatically so the way Crisper actually works, it's usually a specific protein called CAST nine that cuts the genome at the site that you're trying to edit. But CAST nine on its own can't do anything. It's inert. If you will, it needs an RNA, a piece of RNA that's actually specifically designed to match the sequence of the genome that you're trying to modify. It partners with this RNA and the RNA takes it to the right place. So depending on which RNA you've designed, the edits could be more or less specific. There are plenty of examples of first generation Crisper cast nine where you could get really efficient editing at the site you want, and really efficient editing at several other sites as well that you did not want. And then there are many of us, my company Caribou bios Sciences included, who have invented, developed access to new technologies that can overcome some of these specificity challenges.

I mean, it seems like in your case that particular technology is sort of the core proposition that the company is founded on. Right, Can we take crisper and make it work more reliably?

Absolutely?

So, what do you do to make it work better?

So? At the heart of our company is what we call the Shardona technology. Now, Shardona is an acronym. Cchr DNA stands for a mouthful crisper hybrid RNA DNA technology. You now see why we use an acronym.

But each of those words, I mean, it's like a relatively sort of you know, comprehensible acronym, right, like crisper hybrid RNA DNA. It's like, that's not wildly complicated.

Fair, I appreciate that, And to be fair, it does actually describe what the technology is. So I just told you usually CAST nine or other crisper proteins need an RNA partner to get to the right side in the genome. What some of my colleagues did is actually develop hybrid guides, guides that are part RNA and part DNA. And it turns out the inclusion of DNA improves the specificity dramatically. We can measure this in a very quantitative way and see that it improves the specificity of editing by many orders of magnitude.

A huh, So it's not like ten percent better, it's like one hundred times better.

A thousand times better, even more. In some cases.

Is there a sort of layperson's answer to why.

Absolutely. It all has to do with what we would call it biochemistry affinity, meaning what is the binding tightness of the crisper system for the target genome? And it might intuitively feel like higher binding, higher affinity is better, but it actually turns out the opposite is true, huh, And that by including DNA we actually decrease the affinity of the complex for the target. And the reason you want to decrease the affinity is that really the entire human genome resents a laundry list of potential off target sites we don't want to edit. So you want low enough affinity that you're not accidentally grabbing all these other pieces of the genome and instead grabbing the one site that you actually want to modify.

So is the challenge then to see how low you can get the affinity and have it still work. I mean, I get that you don't want it to not bind things that it's not supposed to bind to, or not cut things that it's not supposed to cut, but you do want it to bind to or cut the thing that it is supposed to cut. So presumably there's some optimal spot or maybe what's optimal depends on the use case. But how do you strike that balance.

It's a very careful balancing act. You're absolutely right. Our research team has spent a huge amount of time working on this and has found ways to really develop the appropriate way to balance these two needs for each time. We need to make an.

Edit still to come on the show. How Rachel and her colleagues are using this new kind of Crisper technology to create new treatments for cancer. There's this promising new kind of cancer treatment called car T cell therapy. T cells are a key part of the immune system, and the basic idea here is to engineer T cells to attack cancer cells. As you'll hear, a few car T cell therapies have been approved, but they're complicated and expensive. So Rachel and her colleagues are using Crisper to try to come up with a new kind of car T cell therapy that is both simpler and cheaper. So let's talk about some of the some of the projects you're working on with the technology. We'll start with what's farthest along clinically.

Furthest along is a cell therapy that we call CB ten, and we are developing this to treat relapsed or refractory B cell non Hodgkin lymphoma. So that's a kind of blood cancer, and it's when B cells, part of the immune system, become diseased. And so we are using our Crisper genomeediting, our Chardonnay technology to actually take healthy T cells from healthy donors and then modify them through Crisper to teach them how to find and kill these kinds of diseased B cell cancers. These therapies are called car T cell therapies. We're not the first to work on them. There are many who are advancing these kinds of therapies. And CAR again is an acronym. It stands for chimeric antigen receptor, and it describes a special protein that we can encourage the T cells to express that gives them the ability to specifically recognize and kill these B cells.

As I understand it, other companies have developed car T cell therapies that take an individual patient's own immune cells and develop them in the lab essentially, and then put them back into the patient to target cancer. Right. That is unsurprisingly, very very very expensive, right, because it's sort of like you're developing a custom drug for each patient, which is great in a way, but also it's like this bespoke, sort of individually tailored drug that it just costs a lot to make and it costs a lot to buy. And so my understanding is that you're trying to develop a version that is more like a traditional drug that doesn't have to be customized to every patient.

Is that right, That's exactly correct. So there are today in the United States six approved car te cell therapies, and each one of them is what's called anatologous cell therapy, which is scientific fancy word for patient specific.

And so those drugs, just to be clear, those are approved, they're in use, they exist in the world.

Those are approved commercial products today.

And they cost like hundreds of thousands of dollars per patient.

Correct. Yeah, So they are proof positive that the immune system, specifically T cells, can be incredibly powerful anti cancer agents. They also demonstrate how challenging it is to develop one batch of therapy for each and every patient. That is not scalable. That is not going to be something that delivers this kind of therapy to broader and broader patient populations, and it's also restricted to cancer patients who have sufficiently good T cells to make the product in the first place.

Oh, that's interesting, Like if you're a patient in your immune system is just totally beat down by having cancer or being treated for cancer, then you don't have the T cells to generate this therapy.

That's exactly correct. There's also quite a lot of complexity and almost handholding, if you will, necessary to make these patient specific therapies, and so cancer centers like MD Anderson or the University of Pennsylvania, they have tremendous expertise and they have the staff to really work with patients to shepherd them through this process to ensure that they can actually support them provide any additional therapies they need while they're waiting for their therapy to be manufactured, to give them access to this kind of therapy. But that's not where the majority of patients are treated. The majority of patients are treated in community hospitals and community clinics that don't have the resources to shepherd patients through this kind of very complex stuff.

So, so what do you have to do to make a one size fits most version of this, right you want to? It would be good for the world if we could move away from having to design this drug literally for each patient and have a drug that'll work for almost everybody. That's what you're trying to do. How do you do it?

Yeah, the vision is to develop what the field would call allogeneic or off the shelf car T cell therapies.

Which is what most drugs are, right, Like just a drug and the doctor gives you the drug and hopefully it makes you better.

Right right, So step one is we need to use healthy T cells from healthy donors instead of T cells from cancer patients.

Okay.

Step two is you have to make this safe. Right, Typically, if you take a T cell from one person and put it in another person, you are probably going to cause what is called graft versus host disease, which is where the other person's T cells attack parts of your.

Body, analogous to what is that transplant patients have. Basically, it's like yeah.

Correct, correct, So right off the bat, we have to prevent that, and we do that through genome editing by getting rid of something called the T cell receptor. It's the thing on the surface of the T cell that would usually empower it to cause graph versus host disease. So that's hitting the delete key once to get rid of that.

Okay.

The second step is you have to give the T cells the CAR so it knows what it's looking for on the surface of cancer cells to appropriately identify and kill them. And many of our peers stop there. Those two combined would be what they envision as a product, but our team looked at that and said, that will never be enough.

And so just to be clear, when you see people stop, there are people trying that, like, is that version of this drug in trials? Now?

Yes, multiple human clinical trials are being run with something that looks like.

That, and so if that works, that would be a off the shelf, one sized, fistmost normal drug kind of drug. You're skeptical that it's going to work.

Correct, We think you have to take it a step farther and I'll tell you why. Okay, Yeah, these off the shelf T cells, they are foreign to the patient's immune system, and the patient's immune system is going to figure that out fairly quickly and actually kill off the CAR T cells. And that's very different from when you've had a product made from your very own T cells. They can last for a very long time, and so we look at that differential in time and say that's a problem we have to solve.

Why doesn't everybody agree with you?

I would say more and more people agree with us if you look at what's happening in the field. In fact, many of the first off the shelf car T cells that have been tried in human clinical trials have been retired because they didn't work as well as people had hoped. And I think many are now going back to the drawing board to think about what are other things we can do to empower or enhance these T cells to overcome these challenges.

So what do you do, you were getting to the sort of next steps, what do you do to make it better tolerated by the patient's immune system.

Yeah, we think about how to bridge that gap, both very literally and in ways that are more about boosting the biology than necessarily entangling with the patient's immune system. So, for example, in some of our other cell therapies that we're developing for other blood cancers like multiple myeloma and AML, we actually deploy what we call immune cloaking, and so this is where we use our genomeediting to change what is or is not decorating the surface of the car T cells to try to slow down how the patient's immune system could recognize and clear the therapy. So that's a very literal way of addressing this challenge.

Cloaking is a cool name for it, to basically make the cell better at hiding from the immune.

System exactly exactly Is.

There an example of a particular change you make to that end.

Yes, So what we do is actually get rid of some of the proteins that would usually sit on the surface of a Carte cell. These are called HILA class one molecules, and it helps to prevent the patient's immune system from readily recognizing and clearing the therapy. It's a little more complex than that. There's actually a special kind that we then decorate the surface with to help ensure that all parts of the immune system cannot rapidly recognize it. I also want to be clear, we don't think this creates a perfect stealth cell that lasts forever. There are lots of things about these cells that we expect the patient's immune system to ultimately recognize and cause it to reject. This is about buying additional time with a hope that that allows additional anti tumor activity.

And is there is there a balance you have to strike there where, well, the cell still has to work, right, I mean, is there a universe where you do so much to try and cloak the cell that it can't whatever do its cellular business and persist as a cell until it finds the tumor?

Right? I think there's some extreme world where you try to put so many different genometics into the T cell that you break it, maybe both on a cell specific level, but also a population level. So if you think about it, we're trying to take this population of millions and millions of T cells and provide three, four five different genometics. Now, genomeediting is very efficient, but it's not one hundred percent efficient. You know, some medits might be eighty percent, otherre's ninety percent, maybe ninety five percent. So that means, as you now look at this whole population of T cells, every time you add a new edit, it means a fraction of a fraction of a fraction of the cells actually have all the edits that you desire. So we set a pretty high bar for ourselves. We only bring a program forward into the clinic if we can manufacture it in such a way that at least half of all of the T cells have all the edits that we're going for, and we've been able to do that with three different therapies so far.

So you're saying that even with your improved version of Crisper, it's still sufficiently error prone. Not that it's highly error prone, but it still makes enough mistakes that something like half of the cells you're creating won't be exactly the way you want them to be.

I would say, it's not that it's making mistakes, meaning it's not making off target changes elsewhere that we didn't want. It's instead that in some fraction of the cells, they're just not getting.

The edit right of omission rather than a sinecommission. Yes, exactly. So it's the affinity, like you nailed the low affinity. So does that suggest just to sort of zoom out for a sec I mean it suggests that there is room for improvement on the kind of platform level. Presumably.

I think so. And I'll give an example of even the work we've done over the past few years. So our first program, which is for lymphoma, benefits from three edits. Fast forward now to our third program for AML. It has five different genomeedics in it. We're able to hit the delete key on three different genes in two different places. We can insert new genes to give new functionality to the T cells. And I think this already represents really pushing the envelope in terms of what you can accomplish. And I think there's further room to run with that as well.

If things go well, When when might you be, you know, submitting a drug for approval?

Fantastic question. We hope to start a phase three trial with CB ten next year. If you look at the kinds of phase three trials that have been run for these cell therapies before, they usually take two years or more to run, and then there's some time after that put all the documents together for the regulatory agencies. So a lot of work yet to be done. Very excited about what's coming next.

We'll be back in a minute with the lighting round. Let's have a lightning round. Let's finish with the lighting round. What's more frustrating pipetting or knitting?

Knitting?

What's the hardest thing you ever knit?

Mittens?

Tell me about pipe petting. I feel like you and pipetting have history.

You know. Pipeetting is about moving clear liquids from one to another. I spent many, many years where that is what I did every day, and obviously it gave me the ability to do hands on wet lab research. And there's a piece of it that I desperately miss, which is being the first to know the answer to an interesting biological question. Right, there's a magical aha moment when you see the results first. Now these days, I'm not that far away from the people who get to do the really cool work in our lab, So I'm at peace with the balancing act of not having to pipette and you know, being the third, fourth, tenth person who learns the cool news.

Worthwhile trade off at the end. Indeed, did you go to grad school assuming you would work in industry? Is there some moment when you are making a leap off of this path? You know you're getting a PhD? You clearly have been very good at school. Lots of people just stay in school all their lives and become professors and have wonderful careers. Was there some moment when you decided to step off of that path? Leap off of that path?

I was probably one of the few people in my PhD program who came to school knowing I didn't want to be a professor when I grew up. I actually thought I wanted to be a patent attorney when I grew up. However, we started working with patent attorneys because of all the cool technology that was coming out of the lab, and I pretty quickly realized that's not the job that I want to do.

Good figure figured that out.

Indeed before I went to three years of law school. So it created sort of this moment of well, I don't know what I'm going to do when I grew up. Now know a few things I don't want to do, and it meant I started thinking a lot about the industry side of science. I took a lot of business school classes at that point in time to try to learn and learn a new vocabulary. But I think that made it easier to take the entrepreneur or a leap, because I wasn't on a different path than I had to jump off to.

Go on basically your way out of going to law school. Indeed, so you were on the Forbes thirty under thirty, the Fortune forty under forty. As far as I know, there's no fifty under fifty. So do you have like a next move, Well.

There is a fifty over fifty, but I've got to wait a few more years for that.

I'm glad that you've got it lined up, though it's important to have a goal and you've got time. What's one thing that you wish people understood better about genes?

I think many people expect there's a very clear one to one connection that a gene means X. There are very few genes in our genome that result in one specific outcome. We as human beings are the product of this incredibly complicated cross signaling across every gene in our genome, and any one trait, even as simple as how tall we are, is the output of many, many, many different genes. And so I do wish there was a better understanding of just the rich complexity of our own biology, because then I think it feeds directly into how do you use a technology like crisper to change disease biology? And there are some examples, but not a ton of examples where one edit is enough.

It's like the the Mendelian pea plants maybe do more harm than good as a teaching to a like No, no, it's not usually like that.

Fair.

Yes. Rachel Horwitz is the co founder and CEO of Caribou Biosciences. Today's show was produced by Gabriel Hunter Chang. It was edited by Lyddy jeene Kott and engineered by Sarah Bruguier. You can email us at problem at Pushkin dot FM. I'm Jacob Goldstein and we'll be back next week with another episode of What's Your Problem.

What's Your Problem?

Every week on What’s Your Problem, entrepreneurs and engineers talk about the future they’re trying  
Social links
Follow podcast
Recent clips
Browse 138 clip(s)