I do appreciate everybody coming here and logging in on Zoom and the introduction. I'm really excited to be here at NCATS and to sort of get that sense of the lay of the land and meet some of you and hear more about your science. And what I'm going to talk about today is a little bit of a peel-off from what I talked about yesterday, if anybody was there. And it has to do with some insights that we've developed around why it is that the cell surface glycocalyx -- which is the collection of complex carbohydrates on the surface of a cell -- like, why they change in disease settings, and particularly in cancer. And then, this being the sort of translational arm of the NIH, I'll talk about some new therapeutic modalities that we've developed that target this biology that we discovered.
Okay. So, just to get everyone on the same page, your cells are covered with complex carbohydrates. We also call them glycans. And the totality of those cell surface glycans forms this glycocalyx, which literally means the cage of sugar around the surface of all of our cells. I got interested in the cell surface glycocalyx early on in my training years. And in particular, because of observations that date back to the 1950s and 1960s, in fact, that had identified molecular changes in these glycans in the setting of cancer. And as I introduced yesterday, one of those changes is the observation that this well-manicured garden of sialoglycans -- which are glycans that terminate with this monosaccharide sialic acid -- these types of glycans tend to be overproduced in cancers. So, cancers have this hyper-sialylated phenotype.
And there's a related phenotype that we also study, which is the overproduction of a type of glycoprotein called a mucin on cancers. And later in the talk, I'll explain in more detail what defines a mucin. Sometimes these phenotypes are interlinked because mucins -- which, in this cartoon, are these large, kind of linear trees on the surface of cells -- they are heavily glycosylated and sometimes also heavily sialylated. So sometimes the hyper-sialylated phenotype is in part caused by a hyper-mucin expressing phenotype. So, these phenotypes have been known for some time. They're observed when people analyze biopsy samples from tumors. But until recently, we really didn't understand why these phenotypes occur. And so my lab started studying this in earnest maybe about 15, 17 years ago.
And it was a bit of a spin-out from the project that started with the development of imaging tools to characterize these phenotypes, which was the focus of yesterday's talk. And what we came to understand is that these phenotypes relate in part to the important role that the immune system plays in protecting us from cancer. And, you know, the interface between cancer and immunology really reached a peak back around the year 2010 or so, when the first generation of immune therapies were approved by the FDA, and which since have really kind of changed the practice of oncology. So, again, I think people here are probably well aware of cancer immune therapies, many of which were developed here at NIH, or at least in part at NIH.
We now know that our immune cells play a very important role in protecting us most of the time from cancer. So when things are going well, immune cells will form synapses with cancer cells and recognize that those cells are damaged and eliminate them from your body before they cause disease. And this is why most of us spend most of our lives cancer-free. But when cancers can kind of take root and form tumors and disseminate in the body, often what that means is they have figured out a way to undermine recognition by immune cells. Okay, so these immune cells -- which, in this scanning electron micrograph, are the green fuzzy balls -- they are forming a synapse with the cancer cells and sampling the merchandise and making a determination as to what to do next.
And a very oversimplified view of that synapse is on this cartoon. And the immune cell has receptors that engage ligands on the target cell. Some of those receptors play an activating role, and there's dozens of these activating receptors. Those kinds of receptors are looking for signatures on the surface of the cancer cell that would be signatures of damage. And without going into the gory details, those signatures often take the form of MHC molecules presenting peptides that have mutations. They're sort of neo-antigens. And this is a really clever way for a cell to basically send a signal to the outside world that something's wrong. And what's gone wrong, of course, at the early stages of cancer is that cell has accumulated genomic damage -- mutations in the chromosomes.
And some of those mutations land in open reading frames, which get transcribed and then translated into proteins that carry the same mutations. And when those proteins get degraded to peptides, and those peptides are presented on the cell surface in MHC molecules, the mutation in the peptide is a proxy for the damage to the chromosome. And so, if the immune cell has enough of that kind of engagement at the synapse, a signaling event will occur, which activates the immune cell, and it will kill that underlying target cell. It's just a beautiful system that protects us from having too much cell damage in our bodies. But there's a risk here because, you know, you have way more healthy cells in your body than you have cancer cells. And so, an immune cell might make a mistake, form a synapse with a healthy cell, react and kill it.
And so, to mitigate the risk of autoimmune cell killing, we also have these inhibitory receptors. Now, these receptors are looking for ligands on the cell surface that are kind of healthy signals. And if there's enough of that type of engagement, there's also a signaling event that happens where the inhibitory signal quenches the activating signal. And so, through this sort of checks and balances, the immune cell will collect information, integrate the signaling, and make a decision. Okay? And what we've learned in the world of cancer immunology -- before I ever started paying attention -- is that when cancers are able to survive under pressure of immune surveillance, it's often because they have upregulated healthy signals so that they can deliver a really strong inhibitory signal to immune cells, which quenches the activating signal that might come from their damaged peptides in their MHC complexes.
And so accordingly, if you can block the inhibitory receptors or the ligands that they're engaging, you can prevent this undermining of the immune system and allow the activating signals to dominate at the synapse. And that's exactly what this first generation of immune therapies does. These are inhibitors of the so-called T-cell checkpoint receptors, such as the famous receptor PD-1, the other famous receptor CTLA-4, and antibodies that block those checkpoint receptors basically undermine this inhibitory signaling. So here's an example of one of the phase three clinical readouts from one of the early trials where antibodies against PD-1 and CTLA-4 were tried individually and in combination.
And this was in the setting of metastatic melanoma, which, prior to the advent of immune therapy, was a 100 percent lethal diagnosis. And in this trial, the stunning observation was that patients that were treated either with the antibody against CTLA-4 or PD-1 or the combination of the two -- some fraction of those patients basically had a durable remission. So you can see here in this Kaplan-Meier curve, patients were dying, but at some point, somewhere between 30 percent and 50 percent of the patients were still alive at the three-year mark, and they just kept living and living and living. And that had never been seen before for any treatment of metastatic melanoma, which, of course, unleashed a ton of activity in the pharmaceutical industry to generate drugs that target these checkpoint receptors and their ligands.
The great behemoth of all of them is Merck's Keytruda. It's probably the number one selling drug of all time at Merck in their portfolio. Bristol-Myers Squibb had the first anti-CTLA-4 antibody -- that's Yervoy -- and they also have a PD-1 blocker, which is Opdivo. And after that came the Me2, Me3, Me4, Me5, Me6, and so on. And every major pharmaceutical company has drugs like these in their arsenal that they use, usually in combination with other therapeutics, to treat a number of different cancers. So this is all well and good, except for the fact that while some fraction -- usually a minority -- of patients has this amazing cure for these cancers, most patients do not.
Even in the setting of metastatic melanoma, which is one of the most responsive types of tumors to this class of immune therapies, maybe half of the patients benefit. And then, in other indications where these drugs have been approved, like in lung cancers and bladder cancers, the fraction of responders is even lower. And so, for the last decade or so, the big question in the world has been, why don't all patients have these responses when you block the PD-1 or CTLA-4 axis of immune modulation? And for those tumors that don't respond, are there other pathways of immune suppression that are important and might also be drugged? So, the industry has poured probably half a trillion dollars into trying to find other pathways.
And without going through all of the research and the clinical studies, I'll just say every single receptor and ligand on this chart between a T cell and a cancer cell has been targeted with pretty much antibody-based blocking drugs. There have been a couple of FDA approvals, for example, drugs that block LAG-3 or TIM-3, but generally they're not particularly impressive. They generally don't work for anyone on their own. Some people might respond when they're used in combination with a PD-1 or PD-L1, the blocking drug, but it's been underwhelming. And so, what are we missing? Now, as you might see from this list, all the other targets people have explored for immune therapy are proteins, cell surface proteins. Where no one has really looked is at the cell surface glycocalyx.
And here we are, sitting on a 60-year-old story of changes in sialoglycans associated with cancer. Well, guess what? There's a family of immune modulatory sialic acid binding receptors encoded in the genome. They're called the sialic acid binding immunoglobulin-like lectins, which is a big mouthful, so we just call them the Siglec family of receptors. There's 14 of them in humans. They're expressed often in combinations on every type of immune cell. Unlike the T cell checkpoint receptors, the CTLA-4 and PD-1, which are mostly expressed on the T cell, Siglecs are expressed on T cells, macrophages, neutrophils, NK cells, dendritic cells, all of the cell types that are thought to be important in cancer immune treatment, you know, the immune reaction against cancers, not just T cells, but macrophages and NK cells too.
And looking at this family of receptors, I've kind of broken it down to the extracellular domain up top. This crescent moon is meant to reflect what's called a V-set or an IgV domain. That's the domain that binds the sialoglycan. They have various numbers of Ig domains, which are these ovals, transmembrane domain across this yellow line. But then cytosolically, what's so intriguing is they have signaling motifs that are well known to immunologists and which are exactly the same motifs used by the T cell checkpoint receptor PD-1. The so-called ITIM motif and ITSM motif. And without going through the details, these are the motifs that allow this inhibitory signaling.
So, because PD-1, clinically validated target for immune therapy, has the same signaling biochemistry as several members of the Siglec family, we couldn't help but speculate that just as a T cell might use PD-1 to engage a cancer cell that has upregulated its ligand PD-L1, the consequence being inhibitory signaling because of recruitment of so-called phosphatases to the synapse for people who like to get into the signaling biochemistry, we thought, well, maybe the reason so many tumors have been found to be hyper-sialylated is because those sialoglycans can engage Siglec receptors on T cells, macrophages, and NK cells across the entire immune compartment, which could elicit the exact same inhibitory signaling.
And if that's true, maybe we should be drugging Siglec or their ligands, either on its own or together in combination with inhibitors of the PD-1, PD-L1 axis. So that was the hypothesis. And of course, you can't test that hypothesis without the money. So, back in the kind of late 2000s, I submitted an application to the NCI, and that grant got funded, thank you very much, and thus began the project in my lab focused on Siglec and Siglec ligand biology. So, to make a long story short, we started with a focus on NK cells and macrophages.
And NK cells and macrophages have an activating receptor called Fc-gamma-R3, which has actually the same signaling biochemistry as the T cell receptor on the T cell. But this is how NK cells and macrophages get activated. That Fc receptor is looking for target cells, like cancer cells, that are coated with antibodies. And this is something else that happens in cancer patients. When all of these like weird MHC peptides and other antigens start to get expressed on the surface of a cancer cell, our body often will generate antibodies against those antigens. That also contributes to our protection against cancer. But when cancers are surviving, it's because, you know, they've become immune to antibody-mediated immune pathways.
And so, you know, we asked the question, why don't endogenous anti-cancer antibodies elicit effective immune responses against NK cells and macrophages? And we hypothesized that maybe it's because those cancers acquire a lot of sialic acid, they become hyper-sialylated, which would then allow them to engage Siglec receptors that are on these types of immune cells. And the most prominent is Siglec-7. It's on every NK cell and it's on many macrophages. And so, we tested this hypothesis experimentally. All of these data are published long ago, more than 10 years ago now. But to make a long story short, this Siglec-7 sialoglycan axis is a resistance mechanism against anti-cancer antibodies.
And not just the antibodies our body might make as a cancer patient, also antibodies that we treat cancer patients with. For example, the drug Herceptin, that's an antibody against HER2. It used to be the frontline treatment for a HER2-positive tumor, a breast tumor usually. And patients that have a lot of sialic acid and sialomucins on their breast cancer cells are resistant to Herceptin through this mechanism. Likewise, Rituxan, that's an antibody that is used to treat people with a B-cell lymphoma, and hyper-sialylated B-cell lymphomas are resistant to killing mechanisms that are promoted by this antibody because of Siglec and sialoglycan engagement.
So, at this point, we convinced ourselves that this was a really important biology, but not just us. After we published a paper on this mechanism back in 2014, a number of labs got in there and looked at other Siglec family members on other immune cell types and other different experimental systems, including human clinical sample tests. And all these papers are published now, but between that five-year period of like 2014 and 2019, the literature kind of blew up. And you can get a nice summary of all of it if you read that little review that was published by C&EN back in 2019, provocatively declaring that glycoscience is fueling a new breed of cancer immune therapy. So, that gauntlet got thrown by C&EN, and we're like, all right, now we better develop an immune therapy.
Okay, so we thought about how would we do that. This is the translational part of the story. On the one hand, you might think, okay, take a page out of the playbook of Keytruda and Opdivo, generate an antibody that blocks this glycoimmune checkpoint receptor and prevents its ability to bind a sialoglycan on a target. But the problem is, this is a whole family of receptors, not just one. There's 14 Siglec family members. They're expressed on all immune cell types, usually more than one per immune cell. So, which one do you block, or do you have to block a bunch of them all at once? And we didn't know the answer to that question, and people are still trying to answer that question. But without knowing which one or ones to block, it's really hard to start a drug development campaign.
So, instead, we focused on the ligands. Now, the ligand story is also a book that's being written. We know that the Siglecs are not redundant with each other. They each have specific types of ligands that they engage. But one thing all of their ligands share in common is they need to have a sialic acid as part of the epitope. Because there's a conserved site in all of the Siglecs that engages the sialic acid residue. So, although it's not sufficient to define the specificity of a ligand, it is required for high affinity binding. So, we decided we would go after the heart of the beast and develop a therapeutic that basically targets sialic acids on cancer cells.
And we likened our thought process to kind of designing a lawnmower, right? A therapeutic that would park on the surface of a cancer cell and you turn it on and mow the sialic acids off. And if you could strip a cancer cell of its sialic acids, you deprive it of the ability to bind any and all Siglec receptors. So, we thought this is the nuclear option. Let's just try to do this. So, what does this Siglec-agnostic lawnmower look like as a molecule? What we did is we made targeted sialidases. And a sialidase is an enzyme that hydrolyzes the bond between the sialic acid monosaccharide and everything that lies beneath it.
And as a prototype, we made a fusion between the sialidase from Salmonella, it's a bacterial enzyme, and the HER2 antibody trastuzumab, the active ingredient in Herceptin. And we call this a trastuzumab-sialidase conjugate -- T-Sia. This is a proof of concept molecule. And as I'll show you later, some adaptations had to be made to actually make a human clinical candidate. We actually made this conjugate using chemistry. There was a very practical reason for this. It turns out to be a lot easier, at least if you're in my lab, to click big proteins together than to make them as the more conventional genetic fusions that biologists would pursue. Because when you fuse two completely different proteins to each other, often you have expression problems and issues with stability.
And we figured down the line, we probably have to solve those problems, but we didn't have to solve them in our prototype. Instead, we just used some technologies we already had in the lab. Yesterday, I mentioned the aldehyde tag technology, which we had developed to make site-specific antibody drug conjugates. And we just deployed that same approach to make site-specific antibody enzyme conjugates in this project. So we made Herceptin with aldehydes and then used our Pictet-Spengler ligation chemistry to click on a little linker group with an azide at the end of it. And then in parallel, we took the recombinant bacterial sialidase. We did some engineering on it, which I won't go into, but it's all published if you care.
And we attached through some cysteine conjugation chemistry a little spacer with this dibenzocyclooctyne. And that pre-functionalization allowed us to just do click chemistry between the enzyme and the antibody to attach one copy of the enzyme to each heavy chain near the C-terminal region, which basically kept the enzyme away from the Fab domains so as not to interfere with HER2 binding, and also away from the parts of the Fc domain that engage with Fc receptors and other immune effector functions. And we tested these antibody-sialidase conjugates in a number of preclinical models.
And here's one example. We ran this experiment in collaboration with Heinz Läubli, who's a physician scientist at University Hospital in Basel, in the hands of his former grad student, Michal Stanczak. And Michal had set up in that lab a syngeneic orthotopic breast cancer model in the mouse, in which the mouse breast cancer cell line EMT6 was engineered to express human HER2. And then those cancer cells get grafted into the mammary fat pad, the natural tissue environment. And so the tumors start to grow in those environments. And then at different time points, we injected into these animals different reagents. So this model is known to be resistant to naked trastuzumab. It's considered a model of a cold tumor. And most sort of first-generation immune therapies, including naked antibodies, don't really have efficacy in this model.
And we observed that trastuzumab at 15 mg per kg was unable to retard tumor growth and was comparable to tumor growth in the presence of the saline injection. But when we introduced trastuzumab with a sialidase on it, those tumors had a much harder time engrafting. And when we harvested these little kind of anemic tumors, we found them to be riddled with immune cells. There really aren't cancer cells in there. They're just balls of immune cells. So the cancers went from cold to hot, and our immune therapy was much more successful than a naked antibody. And so with those kinds of data, we decided it was time to make something closer to what a human therapeutic might look like.
And for that, we formed a company, which we spun out of Stanford, which is actually located in Massachusetts, in the Cambridge area, called Palleon Pharmaceuticals. And the scientists at Palleon made some important changes to make this a more developable drug. First, they swapped out the bacterial sialidase we had used with a human sialidase. And they did some engineering on that human sialidase to make its kinetic parameters more comparable to what we had done, the bacterial enzyme. They exchanged the chemical conjugation approach that we had used to a kind of more manufacturer-friendly genetic fusion.
And they also, to keep the molecule's molecular weight more similar to a natural antibody, instead of having two enzymes on a full antibody, they just put one enzyme in place of an Fab domain. And this molecule is now in the manufacturing stage, and it's going into a clinical trial, second half of this year. So probably sometime in the fall, we'll be dosing the first patient in partnership with a Chinese company called Henlius Pharmaceuticals. So excited to see how that goes. Now, in parallel -- oh sorry, by the way, this is -- I forgot I added this. I want to point out, we can potentiate the efficacy of an antibody therapeutic by attaching a sialidase and souping up the immune system.
But since we published on that, other labs have been bolting sialidases onto all kinds of modalities that are also immune therapies to potentiate them. And this includes T-cell engagers. So there's a couple papers on this list where people took a T-cell engager and attached a sialidase, and it becomes more potent, because again, turning off the suppressive signals. People have loaded sialidases onto the surface of CAR T-cells, CAR M-cells, CAR NK-cells. And so keep your eye out, you know, in the world of other therapeutic modalities for sialidases as kind of an additional functionality that could be added to the equation.
Okay. Now, I also mentioned -- whoops -- mucins. That's another area that we've been studying, because, like the sialoglycan story, the story of mucin overexpression is as old as oncology, in a way. People have noted that many different cancers, especially solid tumors, have a hypermucin phenotype. So, mucin is, in fact, that some of these mucins get shed into circulation, and you can use them as biomarkers of disease. Case in point, the ovarian cancer marker, which is called CA-125, it's actually a shed form of a mucin called MUC-16. Okay. And there are other examples like that. And again, we've known that mucins tend to overpopulate tumor cell surfaces, but no one really has had a good indication as to why that might happen.
So we started studying this as well. And here's a little background for people who don't know the details of mucins. What defines a mucin is the fact that it is a glycoprotein rich in clusters of serine and threonine residues, where the side chains of those serines and threonines are glycosylated with glycans that start with N-acetylgalactosamine, sort of O-GalNAc glycans. The consequence of having these alpha-linked O-GalNAc glycans all clustered together in space and in sequence is it forces the underlying polypeptide into a rigid extended conformation.
And here's a little fragment of a mucin whose structure was solved by NMR. And you can kind of see the backbone. It's very gently helical. Sometimes the phi and psi angles are like a polyproline helix, but it's very extended compared to more globular, folded-up proteins. And there's a physical consequence to this. This rigidity actually can affect the mechanical properties of the glycocalyx. And just to give you a sense of how rigid these things are, my lab did some work ten years ago where we chemically synthesized mucin polypeptides with different densities of glycosylation. And if you make a polypeptide that is 100 percent like glycosylated serine and threonine, it is so rigid, it's like a toothpick.
And we took some AFM images of these single molecules of these glycopolypeptides. And basically, based on what we call the persistence length, like the distance between the two ends on average, when these things are so rigid that they're like toothpicks, the persistence length is stiffer than double-stranded DNA. They're like carbon nanotubes almost in their stiffness. And as you back down on the density of these things, they get more and more soft and curly and flexible. Now, when mucins are densely populating the surface of a cell, they can actually create a very thick and stiff glycocalyx as a collective.
It's almost like, you know, a giant redwood forest. You know, their trees are tall and they're overshadowing everything beneath them. And if you pack them really close together, good luck, like, landing a helicopter there. You can't get through all of that. All right. And that's what mucins do. They create a giant redwood forest on the surface of the cell. Now, we have been publishing some work on studying mucins and synthesizing them and putting them on cells. And Matt Paszek, who at the time was a postdoc in Valerie Weaver's lab at UCSF, he saw some of our papers and he reached out to me.
And he said, you know, I have this computational work I've been doing to try to predict the consequences of having a thick, stiff mucinous glycocalyx. And my prediction has led me in a funny direction, because you might think having a thick, stiff glycocalyx that you couldn't land a helicopter on might prevent cells from attaching to other cells in the environment. But he had this computational prediction that, in fact, having this kind of a thick, stiff glycocalyx, although it might slow down the interactions of cells with their environment, it actually might have the paradoxical effect of promoting the formation of large adhesion plaques.
And without going into the biophysics here, his prediction was that if you had a thick glycocalyx -- which, this sort of structures here, and here's a cell that expresses that glycocalyx -- and the cell wants to attach to the extracellular matrix, normally the way cells interact with the ECM is through integrins that bind in clusters and form focal adhesions. His prediction was that having a thick, stiff glycocalyx would force the membrane of a cell to deform in order to get integrins engaged with the ECM. And as soon as that deformation occurred, there would be a kinetic trapping mechanism, wherein integrins right next to an engaged integrin would be more likely to attach as well, versus integrins in more distal regions of the cell that were still buried under the canopy of the glycocalyx.
And this kinetic trap would lead to the formation of larger plaques, which everyone kind of knew at the time would give stronger signaling. And signaling through focal adhesions is really important for cell survival, cell migration, some of the properties that we associate with aggressive cancers. But he didn't have the means to test this experimentally. This was all a theoretical prediction. So we teamed up with the Weaver group and with Matt at that time, and then he continued this work in his own lab at Cornell, where he currently is. And basically, to make a long story again short, we synthesized mucins of different lengths, we put them on cells, and then we looked at the effects. And what we found is when you have long, stiff mucins on a cell, you get large focal adhesions, enhanced survival and proliferation.
And furthermore, if you take cells that you've engineered with long, stiff mucins versus short, curly mucins, and put those cells into an animal, the cells with the large, stiff mucins will metastasize to the lungs and form these nodules, and you get a lot of lung sort of cancer growth. And by contrast, the cells with short mucins are not able to take up residence in a metastatic niche, and they're not able to survive and proliferate and form these nodules. And so, you know, our take on this was that having a thick mucinous glycocalyx, simply through biophysical mechanisms, is oncogenic, you know, it promotes cancer growth and dissemination. And correspondingly, if you could somehow drug those mucins, then maybe you could reverse this biology and make these cancer cells sort of less viable, less survivable, and more easily eradicated from the body. That was the premise.
So, drugging mucins has also been a theme in my lab. And it's not an easy concept, because we're not talking about a receptor-ligand biology here, which, you know, you could block with an antibody, potentially. This is a biophysical effect. So how do you deal with that? You can't just stick an antibody in these mucins and change their biology. They're still thick, they're still stiff, they're still there. You really want to get rid of them. So, more landscaping. Now, someone pointed out to me just before the talk that everyone now sees the chainsaw and thinks that Elon Musk is hacking at the federal government. I really apologize. My chainsaw predated his chainsaw. My chainsaw is a good guy, not a bad guy. And this chainsaw is the chainsaw that we want to develop to cut down the giant redwood trees in this forest. And we're not just taking the leaves off or the grass off. We really have to go at the trunk. So, where do you get a chainsaw from? Well, once again, in the natural world.
It turns out that microbes that inhabit highly mucinous environments, like the gut -- the microbiome -- they often produce proteases that are very selectively active on mucins, mucin-specific proteases. And you can imagine why, if you're living in a thick forest of mucus, it might be of benefit to have these enzymes, because it's like you can machete your way through the jungle with them if you're a microbe that needs to get around. And if you can cut them into little tiny bits and pieces, maybe you can also use them as a food source. And that's what's happening in our microbiome. So, Stacy Malaker, who's now on the faculty at Yale, she was a postdoc in my lab. And Stacey is a world-class mass spectrometrist by training. And she said, well, let's get our hands on these microbial mucinases, let's clone them and express them and see what do they do, where do they cut, and why do they cut as they do. And you have to be kind of like a mucin aficionado. Are there any in the room? No. But you have them at the NIH, by the way. So, Larry Tabak was one of them, and Kelly Tenhagen is one of them. These are world-leading mucin biologists.
And the mucin people know that in the world of proteolysis, mucins are the cockroaches from the proteolytic apocalypse, right? You can't cut mucins with any normal protease. You can, you know, throw trypsin, chymotrypsin, pepsin, like papain, all the proteases that you use to digest everything else on the planet -- they won't touch the mucins because they're so densely glycosylated, nothing can get in there. They're really protected. And yet, these microbes have evolved proteases that only cut mucins. You can put a denatured, naked protein on a platter for mucinases, and they're like, not interested. It has to be a mucin. It's incredible how evolved these things are. So we've been characterizing them. We've been crystallizing them. We're trying to understand their structures and their specificities. And Stacy did some beautiful work with a bunch of different mucin proteases, showing like how they cut and why they cut. And some of them, like, they're recognizing certain glyco structures and parking on the glyco structure and then cutting right next door, right? So we've created like a library of motifs that describe these mucinases. And the one that we thought might be useful to deploy in a therapeutic context is the mucinase from E. coli in the microbiome that goes by the name of StcE. And we call it Sticky for short. There's another one called Slimy, by the way. So there's endless humor in the world of mucin biology. Because if you study snot, you better have a sense of humor.
Okay, so Sticky is a protein that has three domains. There's a catalytic domain, which is kind of like a conventional metalloprotease type domain. And then there's two other domains -- the C domain and the so-called INS domain. These two domains actually allow Sticky to interact with cell surfaces rather non-discriminately. And so you can think of the pink and the green domains as two hands that grab on to a cell. And then the catalytic domain is the mouth that chews off the stuff. Okay. And so in order to use Sticky in a targeted way -- because we don't want to chew every mucin off of every cell in your body -- we want to target this chainsaw the same way we did with the sialidase. We want to hook it up to a binder that targets the tumor cell surface. So we had to detune the enzyme so it wouldn't munch on off-target cells and would only munch on the cells that we parked it on explicitly. To do that, we removed the INS and the C domain. So we took off the hands. So now it's just a mouth, but it needs somebody to point it towards its food, right? And we also made a point mutation in the active site, this tryptophan 366 mutation, to increase the KM value a little bit more. And the reason we like increasing KM -- because you might remember from your Michaelis-Menten lectures, right? When you increase KM, that's basically like reducing the substrate binding affinity. That's good. We don't want it just binding substrates. We only want it to bind the substrates that we force into its mouth by virtue of proximity, okay? And so to get that proximity, we took this detuned Sticky and fused it to an antibody. And in this case, an antibody that binds HER2, which again is that same cancer antigen that Herceptin binds to. And the idea is that HER2 will park on the cells of interest, the HER2-positive cancer cells, and put the mouth right next to the mucins that we want to get rid of, and then hack, hack away at those mucins. And it works beautifully. And here's an experiment that illustrates that.
So we took a cell line -- it's the K562 cell line -- and it normally does not express HER2. And we made a variant that does express HER2, okay? And then we tested reagents to see how good they were at cutting the mucins off of the surface of the K562 cell. The major mucin on that cell is called CD43, and we have an antibody against CD43. So, by flow cytometry, we can see whether we've removed CD43, okay? And you can see over here on the left, the blue dots -- that's wild-type Sticky. And this is a dose-dependent cleavage of CD43. And you can see it's very active at sub-nanomolar concentrations. It quantitatively strips CD43 off the surface of the K562 cell. The red curve and the green curve are variants of our engineered Sticky. So the green dots are just the dead Sticky -- the detuned Sticky with no nanobody, no targeting. And the red dots are the nanobody-fused version of the dead Sticky. And both of those are similarly much less active on K562 cells. You have to go all the way out to, you know, kind of hundreds of nanomolar to micromolar concentrations before you start to get cleavage. And so the difference between the blue curve and these red and green curves is like the therapeutic window. So you'd want to keep the dose of your chainsaw down here in this sort of low nanomolar to sub-nanomolar range in order to get the cleavage you want without the off-target cleavage you don't want, okay? Now, going over here -- actually, sorry, I have some cartoons that show this. Yeah, so this is wild type Sticky and dead Sticky and nanobody dead Sticky. But if the K562 cells are expressing HER2, now they have the antigen, the nanobody dead Sticky is just as active as wild type, right? So we've basically recovered the activity by targeting with the nanobody, but the dead Sticky on its own, without the nanobody, is still dead until you get to high concentrations. So that's exactly the profile that we wanted to achieve.
And then we took our dead Sticky nanobody into a couple of mouse models of HER2-positive cancers. And what we observed is that in experiments where we're looking at the lung mass -- the metastases that form when you inject MUC1-positive, HER2-positive breast cancer cells into these mice -- those, you get sort of retardation in the formation of these masses, but you get a very significant reduction in the total area by H&E staining that is defining these metastases. So this is an indication that we're kind of on the right track and that we're able to target the mucin cleavage off of these cells, and that might have a therapeutic benefit. But we still have a lot of optimization to do, not the least of which is the fact that Sticky is a really great tool enzyme, but it's a bacterial enzyme. And for a human therapeutic, it's always a benefit to use a human enzyme just because you lower risk of immunogenicity and anti-drug antibody formation. And so right now what we're doing is trying to find a human mucin protease. None has ever been described, so we're fishing around from scratch, and we have a couple of ideas. We've identified a couple of cathepsin family members that are not specific for mucins, but they will digest mucins, unlike every other protease we've looked at. So we might be able to repurpose a cathepsin for human therapeutic use, but it's still kind of early and that work is very much work in progress. But in any event, hopefully what I've told you today will get you thinking more about the glycocalyx as a universe of targets for cancer immune therapy and other types of cancer therapies.
The take-home messages are, first of all, tumor sialoglycans, which have been known to be overly abundant since the 1960s. We now have a hypothesis as to the functional significance of that. We think that's a phenotype that arises by natural selection because hyper-sialylated tumors can evade the immune system through the siglec axis. We also showed that if you target a sialidase to tumors using sialidase-antibody fusions, that you can strip those sialoglycans and render tumor cells more sensitive to immune therapy mechanisms. And we're going to test that hypothesis in humans starting later this year. Then, in more recent work, we studied the biological significance of mucin overexpression on cancers and discovered, in collaboration with Val Weaver and Matt Passick, that there's a biophysical mechanism by which mucin overexpression can promote adhesion, focal adhesion signaling, survival, proliferation. And we're kind of in the early stages of trying to develop a new gardening tool -- the chainsaw -- to go after these mucins as a new axis of therapy. And hopefully someday you'll invite me back and I could talk about an upcoming clinical trial of a targeted mucin protease. So with that, I'll thank all the people that contributed to this work. It's 15 years of work now, at least, but many people in my lab are still working on this problem. None of it would happen without sustained federal funding. This is true for everybody in academic research. And for that, I have to thank the National Cancer Institute. So thank you very much. Happy to take questions.
