Thank you so much, Jessica. It's really an honor to be introduced by a former coworker, now superstar here at MIT. It's also a tremendous honor to be here celebrating with the Koch Institute, an amazing institution. And it's especially an honor for me to do the warm up for Peter Senter. So this is a really great opportunity for me. And I thought what I would do is talk about bioorthogonal chemistry, starting with a little bit of the origin story but then kind of morph the talk more into what's pertinent for this symposium subject, which is how we apply these chemistries to make next generation antibody drug conjugates. So with that, let me just start with the backstory.
I was trained as an organic synthetic chemist as a PhD student. And during those years, I spent a lot of time working in fume hoods with contraptions like this, which is the round bottom flask, and it's the standard vessel for performing organic reactions. And what this vessel does is it allows us to have complete control over all of the parameters of a chemical reaction, so that when we take reagent and reagent B, we can control the concentrations of those reagents in this vessel. We can choose the solvent that's the best suited for the chemistry, and we can rigorously exclude chemical functionalities that would interfere with the reaction. And sometimes, that means we're purging the atmosphere with an inert gas like nitrogen or argon because we want to exclude moisture or oxygen. And we have all this control. We can warm the reaction up and cool it down and spin it at different rates and add things and subtract things. And that's why there's thousands of organic chemical reactions in the compendium that we can choose from to form bonds between two different reactants.
And so that was my life in graduate school, and then I transitioned to a postdoctoral position in an immunology lab. And that's when it became clear that there would be some really interesting opportunities to use chemistry to study processes having to do with cells and how they interact with each other. But it's really hard to do classic chemical reactions in these less controlled biological settings. When you work with cells, you really don't have much control at all over the environment. The solvent has to be water. The temperature has to be physiological, and the cellular environment is full of different chemical reactivities that would interfere with the majority of conventional chemical reactions from the organic synthesis world. And so it was the opportunity that chemistry in a living system would offer that kind of drove me, as a young professor, to start thinking about this new brand of chemistry. We called it bioorthogonal chemistry.
And if doing chemistry in cells was sort of daunting enough at the time, it was kind of even more daunting but also even more exciting to think about whether someday, we could develop chemical reactions that you could do actually in the body of a human being, with all kinds of interesting applications for diagnostics, drug development, clinical translation of chemistry into the most complicated world of biology. So that thought process led us to coin this term bioorthogonal chemistry, which literally means chemical reactions between reagents that neither interact nor interfere with a biological environment. And an abstract way to think about this is in the universe of all possible chemical reactivities, which is vast and still yet to be totally explored, there is a galaxy that we would call the biological reactivity space. So that's the chemistry that's defined by all the chemicals in the human body, for example. And there's a lot of stuff. There's a lot of functional groups and molecules, both big and small, in a biological setting, but it's not infinite. There is a boundary around it. And somewhere else in that universe, there's a different kind of chemistry between chemicals that don't exist in biology, in nature, chemicals that react with each other in ways that don't touch base at all with this other biological space. And if we could understand and maybe even define this bioorthogonal space, we could take these chemicals, import them into this other galaxy where they could do chemistry that was pretty much ignorant of the surrounding biology.
Now, the motivation wasn't totally abstract because at the time, there was a specific problem in the world of glycobiology that I was interested in taking on, and it had to do with observations dating back many decades that the pattern of cell surface glycans changes when cells undergo transformations of state. And there was literature, even back in the 1960s and '70s pointing towards some pretty stereotypical changes in glycosylation associated with cancer. One of those is the observation that glycans that terminate with a sugar called sialic acid—these are the sialic glycans. These occur in certain patterns on healthy cells. But when these cells become cancerous and when tumors form, you get this upregulation of these sialic glycans where this kind of well-manicured garden of the healthy cell transforms into this tropical jungle on the cancer cell. So the hypersialylated glycome of cancers was observed in, I'd say, artificial experimental systems, systems where you can take tumors out or grow them in mouse models and characterize them with laborious tools. You grind up the tissue, characterize the glycans using mass spectrometry, do very difficult analytical chemistry. That's how the phenotype was characterized. And I thought, wouldn't it be cool if you could actually visualize this phenotype in living systems, in live tumors, in live animals, maybe someday even in human patients, and be able to study how the cell surface sialic glycans change as a function of space and time? So the problem at issue was, how would you image in vivo these cell surface sialic glycans? And more generally, how would you image cell surface glycosylation beyond the glycans in the living system? And there wasn't a technology for that at the time. So when I applied for faculty jobs, I thought, if I could come up with a technology to image cell surface glycans, that might be something useful and exciting, and maybe I'll get a job and be able to hire people to do this.
And I didn't really know how that might happen until I just serendipitously happened to go to a conference in Southampton, England, in 1992, and I was still a postdoc at the time. And I met a German chemist named Werner Reutter, who was talking about the biosynthesis of sialic acids. And he had made an observation that you could feed modified precursors to sialic acid, which were derivatives of n-acetylmannosamine. This is kind of the dedicated intermediate in metabolism that ultimately gets transformed to sialic acids. And what he talked about in his presentation was the fact that you could replace the methyl group of n-acetylmannosamine with conservatively extended, unnatural side chains. And if you fed these modified n-acetylmannosamines to cells, they would get taken up. There's about six enzymatic steps that transform this precursor to sialic acid. And then eventually, this gets converted to a nucleotide sugar. It goes into the Golgi compartment and gets loaded as a sialic acid onto growing glycans, which ultimately end up presented on the cell surface. So basically, you feed a cell this. A couple hours later, the cell is making this, and it's sitting there on the surface of the cell. And I thought, wouldn't it be cool if you could put a chemically reactive, functional group, like a handle, into that side chain? Simply feed the sugar to a cell, to an animal, maybe to a patient, and let the cells do all the hard work of delivering that reactive functional group to the cell surface, which would then give you a site to attach an imaging probe.
And so I wrote out my job applications about a month later and made this cartoon using PowerPoint 1994 and laid out the vision, which was a 2-step process to image cell surface glycans, first by making an unnatural, chemically-modified, sort of reactive sugar, metabolically labeled the cells with the sugar, and then, when the product is available on the cell surface, add a probe molecule with a complementary functional group where x and y would react to form a bond. And therein lies the problem because for this to work, you have to have a chemical reaction that neither interacts or interferes with all the other biological functionality. Otherwise, x and y will get distracted, and they won't react with each other. So this was the use case to develop what we later coined as bioorthogonal chemistry. So, amazingly enough, I sent this application out to a couple of universities looking for a job, and I interviewed at Berkeley. And people thought, oh, this is an interesting idea. And no one actually asked me, what are x and y? And thank God for that because I didn't actually have an idea yet as to what x and y might be, but it became clear when I started my job that we had better figure that out. So I recruited a handful of new graduate students, and we got to work developing chemical reactions that would fit all the criteria of bioorthogonality.
And we tried a bunch of things, but the one that stuck initially was the reaction between a functional group called the azide, which is three nitrogen atoms in a row—it's a tiny little functional group that has a radius on par with a methyl group. It's pretty conservative. If you wanted to add that to a biological molecule, you could probably do it quite discretely. And it's known that azides will react as electrophiles with soft nucleophiles like triphenylphosphine. And that was a chemistry well-known to the organic reaction community. It was first discovered by a famous German chemist named Hermann Staudinger, and it was used as a mechanism to convert azides to primary amines. And I myself had performed many of these Staudinger reactions when I was a graduate student. And what we did is we made a little tweak to triphenylphosphine, introducing this ester. And the consequence of that is when the phosphine and the azide react with each other, there's kind of an intramolecular rearrangement that occurs where the part linked to the azide and the part linked to the phosphine are now joined together through the intermediacy of an amide bond. So we converted a classic old Staudinger reaction to a ligation reaction. So we called this the Staudinger ligation. And this was the first bioorthogonal reaction that we were playing with back around the year 2000. And we developed this with the hope that we could use this chemistry for these in vivo imaging applications. But we didn't get all that far, and the fundamental problem was, although this reaction is beautifully well-tolerated on cells, in animals—you can do it in vivo. It's a rather slow-moving reaction. And when we started performing these chemistries in mouse models, we found that the reagents were getting cleared out of the body of the animal too quickly compared to the rate of the chemistry. And so after a couple of years of playing with the Staudinger ligation, we kind of moved on and tried to develop a faster bioorthogonal reaction, still using the azide as a component.
And one of the inspirations came from the work of Morten Meldal and Barry Sharpless. Morten is at University of Copenhagen, Denmark, and Barry is at Scripps in La Jolla. And independently, in their own separate labs, they stumbled upon a reaction that's very fast between azides and a different reagent called an alkyne. And that's what this triple bond is. So this is another chemistry that was kind of derived from old German precedents. There was a German chemist named Rolf Huisgen who characterized how azides and alkynes get together in a kind of a cycloaddition chemistry to form a product called a triazole. And what Barry and Morten discovered independently is that this also very slow reaction can be dramatically accelerated by adding a copper catalyst. And this is a chemistry that is very selective, that you can perform in the presence of biological molecules and biological functional groups without interference. So this was something that we looked at and thought, I wonder if we could use this chemistry for our imaging applications in cells and animals. But the answer was no for the simple reason that that absolutely essential copper catalyst is also a cytotoxin. So it will kill cells and animals at the concentrations that you need to do this chemistry. Nonetheless, this has turned out to be a really important type of chemistry in the field of chemical biology, in medicinal chemistry, and certainly in the world of biotechnology. And the Nobel Prize in 2022 was shared with these two chemists. And by the way, if you don't know, Barry Sharpless, for him, this was his second Nobel Prize. He's the only living human with two Nobel Prizes, arguably the greatest chemist on Earth. But this so-called click chemistry as it's now known—and one thing I've learned after the years is that biologists don't know many named chemical reactions. It's not something that you spend a lot of time worrying about. But you probably heard of click chemistry. And here's a little factoid that I learned hanging out with Barry and Morten. The term click chemistry, which was in the literature coined by Barry Sharpless, actually was truly coined by his wife. So that was her branding, and I thought that was really brilliant. But unfortunately, as powerful as this chemistry is, it wasn't going to solve our problems because we wanted to do the chemistry in living systems without harming the cells. So we came up with a kind of different idea for how to speed up the reaction between alkynes and azides. And instead of using a heavy metal catalyst, we used a well-known principle in physical organic chemistry called ring strain. And it turns out if you put a triple bond into an eight-membered ring, it's uncomfortable. The alkyne doesn't like that. It likes to be linear. It doesn't like to have its two bonds sort of like bent. And when you allow this strained cyclooctyne to react with an azide to form this triazole, this is a lot less strained. And the release of that strain makes the reaction go a lot faster. So we call this fast reaction of azides and alkynes the strain-promoted cycloaddition or, more acutely, copper-free click chemistry. And that's the chemistry we ended up using for in vivo imaging applications.
So the idea, again, was if you can get the azide into your biomolecule of interest—and for us, it was these cell surface glycans—then you should be able to click on a fluorescent or some other imaging probe using this copper-free click chemistry. And in principle, you should be able to do this in living cells or organisms. And putting azides into the sugars turned out to be the easy part of the project because we already had this precedent from Werner Reutter's work that I had learned about at that conference in England. So we knew that the acyl side chain of n-acetylmannosamine could tolerate these extensions and still be metabolized and converted to the sialic acid. So we simply put an azide here. We made this sugar. We called it ManNAz. And when cells take this up and it gets processed in the cell, at the end of the day, you'll get an azide-modified sialic acid on the cell surface. And that's not the only sugar you can decorate with azides. You can do it with other important human sugars like fucose and acetylgalactosamine. And we know from the work of Laura Kiessling here at MIT that you can put azides into sugars that bacteria will metabolize. And her lab has shown quite beautifully that azides can be used to decorate the cell surface glycolipids and polysaccharides in mycobacterium tuberculosis, which her group is now converting to a diagnostic for the point of care. So it seems like it's a pretty general principle.
OK, now we played around with some interesting imaging systems, and I thought I would just show this particular slide because this is some work we did imaging changes in cell surface glycosylation during zebrafish development. And I show this because the PhD student who did this work when he was in my lab, Jeremy Baskin, was an undergraduate from MIT. So, thank you very much. Now he's a professor at Cornell, and he did this really beautiful work showing that you could label glycans throughout the course of zebrafish development and profile their distribution not only on mature zebrafish, like this 5-day-old fish—and this is the head. You can see different sugars and different tissue types. But we were also able to look at how cell surface glycans redistribute at the level of individual cells in the very early embryos, like this 24 hour old embryo. And this was a really fun experiment because we learned something new. But also, The New York Times picked this up and put this image in the Tuesday science section. And Jeremy's parents are not scientists, and they never really understood what he did. And they're professional musicians, actually. But when this came out in The New York Times, I mean, his parents read The New York Times. So they saw this and they saw this image and they cut it out and put it on the refrigerator. And I thought, the best thing I could do as a mentor is give my student a project that ends up on their parents' refrigerator, right? How great is that? So this kind of defined the work that we were doing in our lab throughout those first maybe 10 years or so when I was on the Berkeley faculty.
But then I moved to Stanford in 2015, and I had the great fortune of recruiting a postdoctoral fellow who was an expert in RNA biochemistry. And just quickly, I'll tell you, we were able to use metabolic labeling of sialic acids to discover a new type of glycoconjugate, which is kind of mind blowing. So, just to give you a little bit of context, if you go to a textbook and read the chapter on glycosylation and cell surface glycans, you'll see a cartoon like this which shows you that on the plasma membrane of a cell, there's essentially two different classes of glycoconjugates. There's glycoproteins that come in different shapes and sizes, and there's glycolipids. And those are the two. And what Ryan discovered when he was in my lab is that there's a third class of glycoconjugates on the cell surface. They're glycosylated RNA molecules that are being presented in RNA-binding proteins. So this was the work of Ryan Flynn. And by the way, Ryan was a PhD student in Howard Chang's lab at Stanford, which is how he became a deep expert in RNA biochemistry. But before that, he was an undergraduate at MIT working for Phil Sharp. So, Phil, Ryan is your scientific son. He's kind of my scientific son. So does that make us scientific, like, siblings, or maybe you're my scientific father in some weird way? Anyways, so, thank you, Phil, another MIT graduate.
So what Ryan did is he labeled cells with an azide-modified sialic acid, and then he was fishing out sialoglycans attached to different conjugates, thinking that they should be mostly on proteins and lipids. But with his RNA skills, he was able to identify that actually, these sugars with azides were associated with RNAs in a weird way. And we went to great lengths to try and convince ourselves that this was true and not just some contaminant or some artifact. And as an example of one of Ryan's experiments, when he separated out RNAs of different sizes using a sucrose gradient and basically spinning at high speeds, he identified azide-labeled, therefore glycan-associated, RNAs that had this anomalous migration on a gel where they were running like they were large RNAs. But on the sucrose gradient, they were separating out with the small RNAs. So he purified these exquisitely and discovered that what they are are small structured RNAs which include some Y RNAs, some snoRNAs, and a lot of tRNAs, and that these small structured RNAs were covalently modified with the same kinds of n-linked glycans that you find on asparagine residues on glycoproteins. And furthermore, more recently, he's been able to identify some of the chemical details by which the N glycan is attached to these small-structured RNAs. And one of the attachment sites is a known modified base called ACP3U. And this is a modified base that is highly conserved in all domains of life on tRNAs. And what happens is this side chain, which is attached to the Uracil base, gets glycosylated at an amide, just like an asparagine side chain, to attach n-linked glycans, which then are further processed. And all of this is happening in the secretory pathway. So these small RNAs have a mechanism to get delivered from the cytosol into the lumen of the ER, where they are then recognized by the glycosylation machinery that normally would act on asparagine side chains. And if you keep your eyes on the literature, he has a paper in press right now. Here's a little preview in which he shows that this modification and the glycans associated with it have immune-modulatory activities. By the way, Ryan is now a professor at Harvard across the street. So he's a local, and hopefully, you'll get a chance to meet him if you haven't already.
OK, so that's the backstory and the ways in which we used bioorthogonal chemistries to image glycans and profile them and discover them as a basic research tool. But as we know, the theme of this symposium is antibody drug conjugates. So let's talk about how bioorthogonal and click chemistries are being used to make these really amazing molecules. Of all the translational applications, antibody drug conjugates and related antibody conjugates are really, I think, where the heart is for applications of bioorthogonal chemistry. Towards the end, I'll show you an example where people are using the chemistries to make cell therapies and even to do drug delivery. But as you can see, most of the work is happening here. These are just examples of biopharma companies who are kind of using bioorthogonal chemistries as a core part of their technology platform, but I wouldn't be surprised if you took inventory of all the big pharma companies here in Kendall Square, and you'd find somebody using these chemistries to do bioconjugation.
OK, so since I'm the first talk and this is an ADC symposium, I thought I should give a little bit of background for people who don't spend a lot of time thinking about these molecules. They are exactly as they sound. They're conjugates between antibodies and payloads that a person wants to deliver specifically to cell types of interest. They are the manifestation of the 100-year-old magic bullet concept that Paul Ehrlich coined brought to life. And the idea is that you develop an antibody that binds an antigen that is overexpressed on the target of interest. Typically, it's a cancer cell. And that target has to internalize the antibody drug conjugate. And then it enters the endosome-lysosome pathway in which the antibody drug conjugate gets deconstructed. And I'm not going to talk about the details of what happens in here because you're going to hear a lot more about that from Peter Senter and other speakers later today. But through whatever mechanism, the drug gets released. It makes its way to its target. And right now, the two types of drugs that people are using are—they're very toxic, cytotoxic drugs. They either disrupt DNA replication or damage the DNA, or they interfere with microtubules. So those are the two mechanisms by which the toxins on the ADCs would poison the cell, leading to cell death. For some drugs, it's not just the target cell that will get killed. It might be a bystander. And sometimes, that's a benefit, especially since tumors can be very heterogeneous, where some cells might overexpress the antigen that you're targeting, but other cells might not. But those cells that don't overexpress the antigen might get killed through the bystander effect. And that's another design principle. I won't spend time talking about that.
ADCs, their history had a lot of fits and starts. People were working on them in the 1970s, trying again in the 1980s, trying again in the 1990s. But it wasn't until the year 2000 that the very first ADC got its FDA approval, and that was the drug Mylotarg. And there was a really quiet decade where no other ADC managed to get across the finish line, although again, there were lots of trials, lots of efforts. And then in 2010, there was a setback in the field where Mylotarg actually got pulled off the market. So there was a point in time in this century where there were zero ADCs that were FDA-approved. But again, problems were solved. People kept trying. Peter Senter's group at Seattle Genetics is a big part of this story. And eventually, the Seagen drug Adcetris got approved in 2011. And for me, that kind of marks the point at which there was a rebirth of interest in antibody drug conjugates. Two years later, Kadcyla was approved. This came from a partnership of Genentech and Immunogen. And then a couple of years after that, the field really took off, not just in the industry and with FDA approvals, but the investment community started really going in hard and doubling down to the point where now—I don't think I have a slide here. Where is it? Oh, sorry. There it is. There are 13 FDA-approved ADCs, and they are considered one of the most important class of molecules in the practice of oncology. Many of these drugs are now first line therapies. It's the first thing a patient gets. So you can look through these molecules in all of their glory. They differ, of course, in the antigens they target. But they also differ in the nature of the antibody. They differ in the way the linkers are designed, and there's some differences in the payloads and the ratio of payload to antibody that they carry. But all of them are the product of decades of optimization of all the different parameters of the ADC.
So just breaking it down a bit, in the early days of ADCs, a lot of work went into figuring out what's the right antibody. How do you get high affinity, high specificity? How do you humanize? And of course, a lot of that was worked out in the 1990s. Then later, there was a lot of focus on what should the drug be. One thing we learned is that the drug has to be really potent because you don't get a lot of it into a cell through the ADC mechanism. And so what little you get in there had better be very, very potent. So that was an insight that came kind of from the first decade of the 2000s. Then a lot of work went into the linker. It has to be stable in circulation. You don't want shedding of the drug non-specifically. But it does have to release that payload in the tumor microenvironment. So a lot of work went into designing and optimizing these beautiful linkers with a lot of very clever chemistry.
And that was the work of the first decade of the 2000s. But nowadays, there's a lot of attention being paid to the attachment site. How do you form literally the chemistry-biology interface, the interface between the antibody which you made in cell culture, that's biology, with this linker payload molecule, which you made from chemistry? And you have to figure out how do you make that junction. And that's the work I'm going to talk about where bioorthogonal chemistry has been an enabling tool.
OK, so just to give you a concept of old school chemistry, old school bioconjugation, and old is like my age, Peter's age. We're the old school here. The way that you would conjugate a synthetic molecule to a protein in the old school was either to target lysine side chains, which are amino groups, or cysteine side chains, which are thiols. And so far, every FDA-approved ADC has been made with one or the other of those two old school chemistries. So as a kind of case in point, Kadcyla. Kadcyla is a conjugate between the HER2 targeting antibody trastuzumab and the toxic drug emtansine. And there's a linker joining the two. And you can see that ultimately, the chemistry was performed by acylating the lysine amino groups. And an NHS ester is the typical acylating reagent where you have your drug payload as the R group. You do this chemistry, and now you've attached something onto lysine. So one of the first two approved ADCs leveraged this chemistry.
But the price you pay for the simplicity of modifying lysines is that you have very little control over the sites, the lysine residues, that the payload is attached to. A typical antibody can have upwards of 50 solvent-exposed lysine residues, and this is true for trastuzumab. And if you modified every single one of those lysines with a drug, you would have a precipitate because the drugs that people use are generally hydrophobic. And if you cover a hydrophilic protein surface with a hydrophobic molecule, too many copy numbers, this molecule will not behave like a protein anymore. And you can't develop a drug that has those properties. So you have to modulate how many drugs are attached to an antibody to make a developable drug. And a lot of this is trial and error. But in the early days of making ADCs, people were kind of targeting around three or four. So three or four drugs per antibody is a molecule that you can still manage. You can still work with. It still dissolves. You can make a drug OK. And that's what was done with Kadcyla. The so-called drug antibody ratio, or DAR, is 3.5. But what does that mean? That's an average of a population of molecules. Because if you want 3 and 1/2 drugs attached to 50 lysines and you can't control anything about which site gets reacted, you're going to end up with a very heterogeneous mixture. So when you profile the molecules in a vial of Kadcyla, you end up with a distribution of drug antibody ratios which averages 3 and 1/2 to 4 but includes some antibodies with 0, some antibodies with more than 8, and scattered all over the surface of the protein. So I did a back of the envelope calculation once to figure out how many molecules are in a vial of Kadcyla, how many distinct chemical structures. It's more than 50,000. So imagine making a drug, quote unquote, that's a mixture of 50,000 molecules. No small molecule medicinal chemist would think about that as a good idea. You can't optimize. You can't systematically alter the structure and find what's the optimum because somewhere in the 50,000 Kadcylas, there's one which is the best. But you'll never know which one that is. OK, but that was the reality of having lysine chemistry as your only or one of your few options.
Now, Seattle Genetics was more clever here because they realized that if you target cysteine residues, you have a little bit more control over the situation. Antibodies do not have 50 surface-exposed cysteines. They have usually very few exposed cysteines, if any. What they do have are disulfide bonds at the hinge region that can be selectively reduced. And there's four of those disulfides. And you can selectively reduce them to make eight free cysteine side chains. And so what Seagen did is they used a different chemistry. It was Michael addition on these free thiol groups with maleimides. And since there's only eight free cysteines after you reduce the hinge disulfides, the most you can get is 8. But if you do chemistry to try to target a drug antibody ratio of 3.5, which is what C-Gen did in some of their molecules like Polivy, Adcetris, Padcev, again, you have to use a substoichiometric amount of drug. You don't want all eight modified. You only want four. So you're going to end up with a distribution not quite as massively heterogeneous as is the case of lysine chemistry, but still, there will be a mixture. There won't just be one homogeneous structure. Now, the only way you can get a homogeneous structure is if you drive the cysteine hinge modification chemistry all the way to completion to max it out at a drug antibody ratio of eight. And that's what Daiichi-Sankyo did when they developed Enhertu, which is a different trastuzumab conjugate, where the drug is now this DXd, which is a topoisomerase inhibitor. They have a DAR of eight, and this drug has physical properties that still make it kind of an OK-behaved molecule. So this is really the first ADC that I know of that was truly homogeneous as it was approved by the FDA. But is it a blessing or a curse that you have to have eight drugs? It really depends. Not all drugs are tolerated at the drug antibody ratio of eight. Some of them are too toxic to have that much on an antibody going to a tissue site. And these ADCs have off-target toxicities that might relate to the fact that they have such a high ratio of drug to antibody. So I'd say this is a step in the right direction of having a single molecule. But again, you don't have much optionality here in optimizing. The hinge cysteines are what they are, and you either modify them all or you modify some of them. And that's it.
So this is where we thought maybe bioorthogonal chemistry strategies could be used to give you more control over how many drugs you put on your antibody but also where you put them. What if the hinge cysteine region is not the best place for the drug? What if you want them somewhere else? But you just don't have that optionality with old school chemistry. So what you really want is a mechanism to introduce somewhere, anywhere you want, into your antibody molecule a new chemical group that neither interacts nor interferes with the other side chains on the surface of the protein. And if you could do that, now you can target the chemistry specifically to those new functional groups and have as many drugs where you want them, so that you can optimize the ADC for your particular application. So we thought about what technologies could we use to introduce new side chains site-specifically into a protein. And ideally, it would be a technology that was simple, scalable, something practical, not too complicated. And we came up with an idea that was inspired by our unrelated work, really, on an enzyme called the Formylglycine-Generating Enzyme, or FGE. This is a natural enzyme. It's in all of your cells. It's in the ER lumen acting on proteins in the secretory pathway. It has a normal biological function I won't go into. But biochemically, what it does is it oxidizes the side chain of a cysteine residue within a short five amino acid consensus motif and converts it to the corresponding aldehyde. And this residue is called formylglycine, hence the name of the enzyme, FGE. And so we thought this is a really simple motif. It's just five residues. Could you clone that into any protein of interest that is going through the secretory pathway and basically get an aldehyde group on your protein at the site encoded by that cysteine? And that turns out to be quite general and works quite well. So what we would do is clone that sequence into the gene encoding either the heavy chain or the light chain. We would transduce CHO cells with these plasmids along with a little boost in FGE so we could have a lot of activity in the cells and just express the antibody. And it would get secreted into the media, and we would collect it. And what we would find is that those cysteine residues, wherever they were encoded, would be replaced with formylglycine quantitatively. And we can express these antibodies at 5, 10 grams per liter with quantitative conversion of cysteine to formylglycine inside the cell. So all we do is harvest the antibodies, and then we do chemistry. And we developed a specific ligation chemistry that works well on proteins that targets the aldehyde to form a carbon-carbon bond with a linker payload. We call that the HIPS ligation. It's kind of a take off on another old school German reaction from the early 1900s called the Pictet-Spengler reaction. And we made a few little tweaks to that classic old chemistry. What we do is we take an indole, modify it with a dimethylhydrazine. The hydrazine reacts with the aldehyde to form an iminium ion intermediate, and it's set up for an intramolecular electrophilic aromatic substitution to form the carbon-carbon bond. And that ultimately links whatever we attach to the indole onto the antibody explicitly and nowhere else, just on these aldehyde groups.
So we spun that out into a company called Redwood Bioscience that was basically led by a former grad student, David Rabuka. And eventually that company was acquired by Catalent, and now it's part of the Catalent portfolio as a technology that we partner with pharmaceutical companies who want to convert an antibody that they give us to an ADC. And so here's some of the PR off the Catalent website. The whole idea is you send your heavy chain and light chain sequences to Redwood. They will clone in this motif in different locations so that they can do structure activity relationship studies. They'll express the antibody in their engineered cells so that they get high conversion of the cysteine to the formylglycine. And then they'll do the Pictet-Spengler chemistry to attach linker payloads. And they have a whole suite of options in that regard. And what's cool is you can basically move the linker payload around. You can put it on the light chain, the heavy chain, way down at the bottom, up near the top. You can have linkers that have more than one payload if you want to have two drugs at the same time through two different mechanisms. And you can even combine technologies. For example, they have made ADCs of these shapes where they use the cysteine hinge chemistry, the Seattle Genetics chemistry, to put some of the drugs in this zone. But then you can use the aldehyde tag, the so-called smart tag, as Catalent branded it, to put other drugs elsewhere. So it just gives you a lot of optionality to make ADCs with different shapes and sizes with homogeneous, site-specific, very stable chemistry.
But ours is not the only company that is leveraging the principles of bioorthogonality to make site specific ADCs. Here's two other really clever examples that I'm excited about. A company called Ambrx, which Peter Schultz spun out from the Scripps Institute, uses the unnatural amino acid mutagenesis strategy, the kind of recoding strategy, in order to introduce clickable, unnatural amino acids at specific sites. And that's how they arm the antibody for site-specific antibody or payload linker attachment. And then there's another company called Synaffix, which is a European company that leverages the conserved N-linked glycans that are near the hinge region of all antibodies, all IgG1s at least. And what they do is they attach an azide-modified glycan onto these existing glycans. And that becomes the site for a copper-free click chemistry to make site-specific ADCs. And these are just two or three of now many examples that introduce the clickable handle onto the antibody using a variety of different strategies. But at the end of the day, it's the selectivity and the biocompatibility of the chemistry that gets you to the site-specific ADC.
OK, so that's what I'm going to say about ADCs. Before I finish, I'm just going to give you two more vignettes where bioorthogonal chemistry is used to make other kinds of conjugates starting with CAR Ts. So this is a universe that's very familiar with CAR Ts. You guys have invented some of the most impactful and innovative technologies that have been brought to bear on CAR Ts. But one thing we all know is that today's conventional autologous CAR Ts are expensive to deliver to patients and difficult to scale because of the complexity of the procedure. So the way it happens today is a patient will come in for their CAR Ts therapy, and it starts by delivering blood samples from which T cells get harvested. Then those T cells are genetically engineered ex vivo to introduce the CAR, the Chimeric Antigen Receptor. They have to be grown up and a lot of quality control. Make sure they're good and ready to go. And then they get delivered back into the patient. And then the patient has to be monitored because there's potentials for severe side effects. So the whole procedure can take several weeks, and it's very expensive and hard to scale. There's a growing interest in allogeneic CAR-T strategies where you could take T cells that are off the shelf, that don't have to get retrieved from the patient and genetically engineered ex vivo. And this is where bioorthogonal chemistry has also proven to be useful. There's a company called Acepodia, which is part in Taiwan and part in the Bay Area, formed by another one of my former PhD students. And what they do is they take off the shelf gamma delta T cells from pooled donors, not the standard alpha beta T cells of today's CAR Ts but the gamma delta T cells, which are the kind of early life T cells from when you're a baby. These T cells tend to be less prone to rejection. They're tolerated allogeneically, but they're difficult to genetically engineer, which is why Acepodia does not introduce a targeting CAR by genetic engineering. But rather, they do a chemical reaction using bioorthogonal chemistries with targeting antibodies directly onto their gamma delta T cell receptors. So no genetic engineering is necessary. Plus, these are off the shelf T cells, which means that they can make big batches of universal CAR Ts, they call these ACCs, Antibody Cell Conjugates, that can be delivered to a broader group of patients. They have three clinical programs. Last year, they had their first phase I readout of one of these chemically modified gamma delta CAR Ts. So I'm pretty excited to see where this goes. But it's a really clever application of the technology.
And then I thought I would end with the first example of this dream from 30 years ago now almost, which is, can you do these chemistries not just on cells in a dish or on antibodies from production? Can you do these chemistries inside the body of a human? And this has now been accomplished. And the chemistry that's been performed in humans is different from those chemistries I've already shown you. It's a chemistry called the tetrazine ligation that's been set up for this so-called click to release process. So the chemistry at issue here is one in which this molecule, which is called trans-cyclooctene, it's a strained alkene, so kind of the cousin of the strained alkyne that I showed earlier for copper-free click chemistry. This strained alkene will react with a tetrazine in a cycloaddition reaction that not only forms a bond between the two components but leads to the release of a functional group at this position. This gets kicked out. And so what you can do is basically use the trans-cyclooctene as a cage for a prodrug, where the chemistry causes the release of the prodrug. For example, this company, Shasqi Pharmaceuticals, made a prodrug version of the chemotherapy drug doxorubicin, where they introduced the TCO, the trans-cyclooctene cage, down here. And when you modify this amino group on doxorubicin, you deactivate it. It's not toxic at all. OK. And so now this is their platform. They take tumors of interest. They're starting with soft tissue sarcoma. They inject into the tumor microenvironment a hydrogel polymer based on hyaluronic acid, which is a totally safe natural polymer. And they chemically modify the hyaluronic acid with the tetrazine. So this is implanted right into the tumor microenvironment, intratumoral injection. And it just sits there, and it does nothing. It's totally inert. But then they infuse into the patient by IV that caged doxorubicin with that TCO modification. And this drug goes systemic, and it does nothing. It's harmless because it's caged. But when the drug encounters that polymer, which is in the tumor microenvironment, that click to release tetrazine ligation takes place, and the doxorubicin is set free. So now you have a very high local burst of active doxorubicin, which is toxic to the tumor cells. But you don't have that exposure system-wide. So you don't have as many side effects. You can get a much higher therapeutic window with this strategy. And they are in the clinic testing this out right now. So I'm really excited to see how bioorthogonal chemistry has not only enabled some of the therapeutic modalities here but is also now making its way into some truly translational settings where the chemistry happens in the human body. So with that, hopefully, I interested you in this concept of bioorthogonal chemistry and all the opportunities that it brings. And I'm happy to answer questions if there's time. Thank you.
Any questions?
The mics are on. I don't know, try it. Is there a button? Hello. Yeah.
