Scientist Stories: Jeffrey Hubbell, Harnessing Cytokines for Cancer Immunotherapy

Okay, we can get started now. Thank you for coming to the Harnessing Cytokines for Cancer Immunotherapy Symposium. Welcome. I'm Jan Reichard, Executive Director of the Antibody Society, which is a nonprofit trade association for professionals engaged in antibody-related research and development. Our aim today is to inform and educate you about engineered cytokines, as well as the opportunities and challenges faced by scientists who are developing new cytokine-based therapies for cancer. This symposium was organized by Dr. Katherine Harris and Dr. Jonathan Sokolowski. Dr. Harris is the Vice President of Discovery at Amgen, where she leads oncology research. Dr. Sokolowski is the Director of Company Value Creation for CuriBio, a seed-stage biotechnology investing firm. They both serve on the Society's Meetings Committee, and we thank them very much for contributing their time and expertise. Today's program includes scientific presentations by 4 expert speakers. Following the presentations, at approximately 1:40, Eastern time, we will have a panel discussion specifically on those opportunities and challenges that arise when developing next-generation cytokine-based immunotherapies. Biographical information for all of the symposium participants is on the Society's website. Please do look in the chat for the URL. We've also reserved a 45-minute period at the end when you can break into small groups groups to further discuss the fascinating field of cytokine therapy and cytokine engineering in cancer therapy. Please note, technical aspects of the program are being ably managed behind the scenes by Pam Borghart, who is available to help if anyone has problems. Please use the Zoom chat feature to contact her. I'll now pass the mic, so to speak, to Dr. Dr. Harris, who will introduce the first speaker.

Wonderful. Thank you, Jan, for that great introduction and for the overview of the symposium we're running today. Again, I'm joined by my co-host, Jonathan Sodlaski. We've worked hard to put together what we hope will be an engaging symposium for you today. We, just as a preview, we do have 4 speakers, as Jan mentioned. That are coming up here. We're going to start with a keynote with Jeffrey Hubbell, followed by a presentation by Michelle Yen, Nathan West, and Jamie Spangler. We'll then go to a panel discussion with all of those speakers, with the addition of Andy Young from Asher Bio. And then we'll have some closing remarks and then an opportunity— we hope you'll stick around to meet those speakers in the breakout session that was mentioned previously. We're really excited to kick this off. I wanted to start with just a couple of minutes to kind of introduce the session and kind of our vision and reasoning behind picking this topic. And that really comes down to this new resurgence of cytokines for cancer therapy. And as, as you're all aware, for the last 40 years, you know, the role of cytokines in immunotherapy have been rigorously pursued, both preclinically and clinically, but we've only seen moderate success. In harnessing cytokines for cancer therapy over those many, many years. However, while we've been kind of slogging through understanding cytokines and their role in cancer biology, we have gained significant knowledge. And it's largely around the limitations of cytokines, right? And if we think about what cytokines are as being small, potent, short-lived signaling molecules, Those are really great attributes for cytokine signaling, but they actually make really poor drugs. So part of these limitations that we've had is trying to figure out how to harness those cytokines and make them better drugs. We love the biology when they're on target and doing those desired effects that we want, but the truth of the matter is that they have a very pleiotropic role that they play, especially when you administer them systemically, and we start to see off-target effects that are undesirable. So what we're really trying to do is learn the lessons from the 40 years of cytokine biology and clinical trials that have gone on and try to overcome some of these hurdles, such as, you know, a short half-life, limited efficacy as a monotherapy, and how to develop these as better drugs, you know, overcome low stability, poor expression, and other qualities of cytokines, again, that makes them good signaling molecules but not great drugs. And really, we've seen a resurgence in the last 5, 10 years of the use of cytokines for cancer therapy, and we're starting to see some, some successes. And this largely has to do with an expanding set of technologies that have been developed to help us engineer better cytokine drugs, as well as mimetics that can mimic the activity of those cytokines but have better properties for uses as therapeutics. We also have a growing understanding of cytokine biology and its role that it plays in the cancer immunity cycle. And we're getting better at using combinations to help leverage that cytokine activity to obtain optimal activity. With that, I would like to just pause a moment to introduce Jeffrey Hubbell, who is our first speaker this morning. Dr. Hubbell is the Eugene Bell Professor in Tissue Engineering at Pritzker School of Molecular Engineering at the University of Chicago. He was— previous to moving to Chicago, he was on the faculty of the Swiss Federal Institute of Technology, where he served as Director of the Institute of Bioengineering and Dean of the School of Life Sciences. We're happy to have him speaking today with us. The title of his talk is Molecular Engineering of IL-12 to enhance efficacy and reduce toxicity. And with that, I will hand it over to Dr. Hubble. Thank you very much.

Super, Katie. Thank you. Do you hear me and see my slides?

Yes, it looks great.

Great. Thanks. So yeah, in terms of key speakers here, our key participants in the study, I wanted to specifically call out Aslan Montserrat, who's driven much of the work that I'll present in a collaboration with Melody Swartz's lab, who's also at the University of Chicago. A cancer immunoengineer. Jun Ishihara was a senior scientist in my laboratory, still collaborates very closely in this project, and he's now running his own laboratory at the University College— sorry, at Imperial College London. And in terms of conflicts to declare, Aeroimmune is developing this technology, and those underlying are shareholders. So, we're all motivated, as Katie said, by improving these curves that everybody has seen, making tumors hot so as to make them more responsive to checkpoint immunotherapy. And as was said in those really nice introductory slides, our intention is to reduce the toxicity of cytokines. And I'll specifically focus on interleukin-12. And we've taken an approach by saying, how can we administer systemically yet enhance retention in the tumor. So, I wouldn't say we're targeting the tumor, but rather we are retaining in the tumor. Half-life outside the tumor is very quick. Half-life within the tumor, as I'll show you, is prolonged. And that allows us to enhance a therapeutic index by tipping that balance. That's the first approach that I'd like to show you. A second is in masking toxicity of interleukin-12 by changing the molecule itself. So as to have it be activated in the tumor microenvironment. So toward this first approach that I'd like to mention of, as this platform of retention, how would we retain in the tumor microenvironment but have clearance quickly elsewhere? We've exploited molecules that are present in essentially all solid tumors, like collagens I and collagen III. So we have, in order to address that, taken a small domain, about a 20-kilodalton domain from von Willebrand factor. So, VWF is a very large circulating blood protein, also a protein that's present in the subendothelium and in platelets. So, we've taken a— it's a very large multifunctional protein, but about 20 kilodaltons of which, called the A3 domain, binds collagen I and III with nanomolar Kd. So, we've taken that and fused that to cytokines in ways that I'll show you in just a minute, with the idea that the fused cytokine, when it extravasates in the tumor, just like any other biomolecule would extravasate in the tumor, is retained in the tumor microenvironment by binding to collagen. So, when we take that A3 domain of von Willebrand factor and inject it in the tail vein— I will show you a couple slides on intratumoral injection. But essentially everything else, the vast majority of my slides are with tail vein injected therapeutics. If we inject this A3 domain in the tail vein in a tumor-bearing mouse, this is a breast tumor-bearing mouse, orthotopic breast tumor-bearing mouse, we see that more than half of the fluorescence associated with that A3 domain is associated with the tumor 48 hours after injection. Whereas relatively small is in solid organs like the kidney and the stomach and the intestine, not shown here. There is some uptake in the liver, and I'll come back to that a little bit later. You will recall that the endothelia in the liver is fenestrated, and that leads to exposure to collagen in the liver, just like there's exposure to collagen in the tumor. There just happens to be much more exposed in the tumor. Even though the liver is much larger than the tumor under these conditions. So, again, we're not targeting it to the tumor. We're retaining it in the tumor. We're retaining it in the tumor through this biological affinity. So, we've turned to a variety of cytokines, and I'll show you some of them that focus in on IL-12 to ask, how can we enhance tumor inflammation in animal models of poorly inflamed tumors? Whether they're transplantable tumors or genetically engineered mouse models of tumors with inducible tumors that are engineered to be immunologically cold. How can we enhance dendritic cell infiltration? How can we enhance T cell infiltration? How could we enhance the activity of those cells? How can we activate T cells? How can we activate dendritic cells? Sorry. Just as to localization in the tumor, this is from an early work, not on a cytokine. Rather an antibody where we made these collagen binding domain, which I'll just call CBD. We made these fusions to an antibody anti-PDL1 and injected that in the tail vein. And just to illustrate where it goes, so one sees the vasculature here fluorescing in teal, and then one sees the antibody, that's the CBD antibody, throughout the whole tumor stroma, not just subendothelia, not just near the endothelial compartment, but rather throughout the tumor microenvironment. So the whole tumor becomes our depot for the administrated agent. So IL-12, let's focus in on IL-12. Lots of people have been pursuing IL-12. IL-12 is quite a hot topic right now. And the question is, how do you tame its toxicity? So, and we hope that by localizing into the tumor and changing the therapeutic index, that we can change its toxicity profile. IL-12 interacts with T cells and NK cells so as to enhance interferon gamma production. And then there are many downstream genes from interferon gamma that are, for example, chemokines like CXCL9 and CXCL10. I'll come to some others of those, all of which should both activate cells in the tumor microenvironment as well as recruit cells in the tumor microenvironment. As to the molecule, so IL-12 is a heterodimer, a p35 and a p40 subunit. And we fused, and we make it heterodimeric. We have fused a collagen binding domain to each of those two subunits. That then heterodimer with the two CBDs gets a Kd of single-digit nanomolar to collagen I and collagen III. So it's not an antibody level, Katie, but in terms of a bioaffinity agent, it's not bad at all. So relatively high affinity, especially to such a dominant, in terms of mass, target molecule. This leads to enhancement of accumulation in the tumor microenvironment. So here is shown 1 hour with the IL-12, 1 hour after administration in the tail vein. Interestingly enough, we see enhanced clearance from the blood. So wild-type IL-12 being cleared from the blood in the mouse, and then the engineered CVD IL-12 being cleared much faster in the mouse. We think that's good news. So we're not such fans of long-circulating cytokines like interleukin-12. Rather, we want short blood exposure. In long tumor exposure. What we really care about with IL-12 is not actually where the IL-12 is, but we care about where the interferon gamma is that the IL-12 would have produced. So, in this experiment in a B16F10 melanoma-bearing mouse, so, make an injection at day 6 in the tumor and then bleed every day after that, bleed so as to get interferon gamma in the tumor microenvironment in another slide. And interferon gamma in the blood. So here, if we look at day 2 after injection when interferon gamma in the blood is highest— remember, interferon gamma in the blood is not helpful. It's merely harmful. But comparing the wild type at one dose of 10 micrograms versus 10 microgram molar equivalent— these are all molar equivalents always in my talk. So, you know, substantial diminution of IL-12 in the blood. Or at a higher dose of 25 micrograms, substantial diminution. Or at a rather high dose of 50 micrograms, also substantial. So we see that the dose-response curve for this off-target activation, which is primarily due to activation of NK cells resident in the blood or present in the blood, that's much reduced. So we would expect the toxicity to be improved in that case. Now, if that came with a diminution also of interferon gamma in the tumor, that would be bad news. So, we see it quickly cleared from the blood or reduced in the blood. But in the tumor, in fact, it's prolonged. So, here, if we're looking at your tumoral interferon gamma, so, which we get by homogenizing the tumor at different points in time, we see a prolonged interferon gamma signal in the tumor, whereas the wild type is relatively quickly decreased. And if we look at interferon gamma responsive genes, like the chemokine CXCL10 strongly elevated, and this is 4 days after single administration, or the chemokine CCL4 strongly elevated. Those are just some examples. So, we see both interferon gamma directly as well as its effects enhanced in the tumor over a function of time. So, what do we get in monotherapy here? So, monotherapy, working in the B16F10 melanoma model, This is a relatively small tumor, 60 millimeters cubed. In a 60-millimeter cubed, you can see that wild-type IL-12, so at this dose, does produce a survival benefit. But the CBD IL-12 produces a much better survival benefit, indeed with a whole bunch of complete responses. So under these conditions of the tumor, I'll show you a larger tumor in a few minutes, but we see a very strong efficacy benefit So I've shown you a toxicity benefit and an efficacy benefit of this collagen binding accumulation and retention in the tumor microenvironment. So staying in the B16F10 and asking about dose reduction, can we reduce dose and keep efficacy? So here you're looking at survival curves as shown where we were treating with a very high dose of wild-type interleukin-12 versus a much lower dose, 5-fold lower dose of the CBD IL-12, which I've shown you has on a mole per mole better toxicity. But now we're going to even 5-fold lower moles. We see a similar efficacy survival benefit of the CBD one. Also within really no body weight loss. Under those conditions. So we see really strong benefits and strong capability to reduce dose. So thinking about therapeutic index, and this is the only slide I think I'll show you with intratumoral administration, there is an interest in intratumoral collagen-binding cytokines. But if we compare IT wild-type at 5 micrograms to IT engineered at 5 micrograms, strong benefit there, a strong survival benefit there, and even in the case of IT administration. So we can use these molecules for intratumoral or for intravenous administration. So very low doses of the IL-12, the CBD IL-12, can then be effective. And the MC38 model is a model that's relatively responsive to immunotherapies. So in this case, even down to 2 micrograms of— with just 2 doses shown here, 2 micrograms, just 2 doses, and we see strong survival benefit with complete responses in the entire group size. So 5 micrograms is really very similarly effective as 2 micrograms, allowing us to explore dose reduction. What about in cold models? So or colder models. The B16 model is very poorly responsive to immunotherapy. It doesn't really respond at all to checkpoint blockade with PD-1, for example. Another model that's immune-excluded is the EMT-6 model. So, this is a triple-negative breast cancer model, so transplanted orthotopically. So, it is somewhat responsive to checkpoint blockade, although it's quite a cold tumor. So here, seeing survival with nothing in black, survival with wild-type IL-12, or survival with the CBD IL-12. So we see complete responses in nearly the entire group. Importantly, if we think about memory, if we take those 13 survivors from the treatment group, and we rechallenge them on the contralateral side after a period— this 60-day period, 12 of those 13 fail with no additional therapy. 12 of those 13 fail to develop any tumor. So they're showing strong memory after that 60-day period. You can really see the enhanced inflammation as well. So this is with the CD8 being stained in brown. So this immune-excluded tumor really is immune-excluded. It's really hard to find CD8 T cells in the tumor microenvironment. So whereas then there's a lot of CD8 infiltration after treatment with the CBD IL-12. So this tumor localization has a benefit in cold tumors as well as other tumors that are just immune suppressive, like the B16F10. Now, the immune suppressiveness of a B16F10 depends on the size of the tumor. So in this case, we've gone to really large tumors. 100 to 120 millimeters cubed before we've started treatment. And we're looking at treatment in combination with checkpoint blockade with anti-PD-1 plus anti-CTLA-4 in the magenta line, or treatment with just the monotherapy CBD IL-12, just 2 doses. So we could probably do better if we continue dosing. There's no, there's no necessity to limit to 2 doses. So as I mentioned, checkpoint alone, even this double therapy, is not effective. The CBD-IL-12 was very effective in giving a survival benefit. But we got like half complete responses when we combined those two. So we were able to accentuate the effects of checkpoint blockade by cytokine priming of the tumor microenvironment. It was interesting if we looked at those survivors, these 7 survivors, and did antigen-specific restimulations to ask, to what are they responding? How broad is the immune response. So for example, if we restimulate T cells from the spleen, these are systemic restimulations to exosomes produced by V16F10. So we see a strong interferon gamma response, strong IL-2 response. We can see some of the endogenous tumor antigens here. So these are two endogenous mutations, strong immunity. And then others that are really just public, so not specific tumor mutations, like the TRP-1 or TRP-2 peptide, also strong immunity. But it's interesting to see, I thought, how heterogeneous it was in these different inbred mice. So we have done this technology, used this technology for other cytokines. We actually started out with interleukin-2 and published on that in 2019, I think. But I just wanted to point out that on a mole-per-mole basis, mole-per-mole basis, that the CBD-IL-12 is a lot more effective than CBD-IL-2. We can go to higher doses. So, here, this 25 micrograms of IL-12 is the same molar dose as 8 micrograms of IL-2. If we go to higher molar doses of IL-2, as we've showed in our previous publication, we can get efficacy, but on a mole-per-mole basis, the IL-12 is a lot more potent of a molecule. Now, those are with large tumors, large transplantable tumors. And recall that extravasation in the tumor microenvironment is what led to exposed collagen. We also wanted to look at a metastatic model to ask, like, what about small tumors? Do they have enough exposed collagen? In order to get a favorable response. So in this case, we use the, the tail vein injection model of metastasis for B16F10. And so just looking at tumor burden, as we saw a substantial reduction in tumor burden, but I wanted to show you more flow cytometry on the lung. I think it's, it's more interesting. And here we've also measured tumor burden. So note that the, the group sizes are pretty large here. So, we see a strong increase in total immune cells and in total CD8s, a decrease in terms of percentages of Tregs. And if we look at effector CD8s, so a strong increase in effector CD8s, sort of as the best harbinger and one of the best in my opinions is the ratio of effector CD8s, meaning CTLs, to Tregs. So here we see, like, we can go from a very poor ratio in the untreated tumor to a very favorable ratio of about 10 to 15 T effectors per Treg. And that's much more effective than with an equimolar dose of the wild type. Interestingly, if we correlate that to metastatic burden— so metastatic burden here correlated with the ratio of CTLs to Tregs, there's a correlation coefficient of 0.02 there. So, that's saying that this difference that we made here in Treg ratios, that made a big difference in terms of tumor outcome. Now, the IL-12 will affect the myeloid population as well. And so, in particularly interesting is these cross-presenting CD103 DCs, so capable of priming CD8 T cells. Tumor antigen. So also a strong benefit there. Or if you look at, at the number of, of M1 macrophages or the fraction of M1 macrophages, strong ratios there. If we look at and ask by virtue of correlation with metastatic burden, does this number of CD103-positive DCs matter? P is 0.0007. So yeah, there's a strong correlation there. It matters a lot. And even the number of M1 macrophages, a strong correlation with metastatic burden, or rather reduction in metastatic burden. Those are all transplantable tumors, rather, whether systemically or orthotopically. So we also wanted to look at genetically engineered mouse models. And so this is the BRAF V600E model. A mutation in human melanomas, with a PTEN deletion, also characteristic of human melanomas. And here we're treating with combination therapy, either checkpoint blockade, PD-1 and CTLA-4 antibodies, or that in combination with the CBD IL-12. So you can see a much longer period of progression-free survival and much improved survival benefit. One could have argued that you need the hypervascularity of a transplanted tumor in order to see this collagen deposition that I mentioned. That's not so. That's why we did this experiment. It shows that that's not so, that even these tamoxifen-induced genetically engineered mouse models show a survival benefit and progression-free survivors. You can even see it in terms of the quality of the tumor. Note that in checkpoint blockade, the tumors are really 3-dimensional. So whereas in the combination therapies, the tumors are much more 2-dimensional. We can't do flow cytometry on these tumors. We can only do histology. So an even more challenging model than that is the BRAF V600E P10 deletion that expresses stably a beta-catenin. That's also characteristic of human cold tumors. Based on Tom Gajewski's work also at the University of Chicago. And we got these mice from him. So here we also see in this really cold tumor, good progression-free survival. As I mentioned, we can't do flow on these tumors, but we can do microscopy. And so if we count the number of CD8 T cells in the tumor, so we see that increased by a factor of 2 or 3. So here looking at CD8, see here in red. So, you know, you can really perceive and quantify that there's a strong elevation in immunogenicity in the tumor. So, we have explored the CBD platform with other cytokines, and I'll put that in my closing slide, as well as chemokines. But I mentioned— wanted to mention in my remaining 10 minutes or so, other approaches to engineering interleukin-12. Aslan Mansurov led this work as well. We asked the question, IL-12 is a heterodimer and its receptor is a heterodimer of a beta-2 and a beta-1 subunit. We asked the question whether we could use a domain of the beta-1 subunit fused to the p35 subunit, to come along and bind and block receptor ligation. Now this binding is a lower affinity— sorry, a poor affinity, yes, a lower affinity, higher Kd— than is the intact receptor. We only use a domain, or rather two domains, of the beta-1 subunit to form the mask. So when the full subunit comes along, it displaces the mask. Moreover, the mask is released by proteases. And we use both urokinase plasminogen activator substrates and MMP substrates within the masks. The mask is sensitive to both enzyme classes through— or rather, the linker of the mask. A good measurement for activity of IL-12 is downstream signaling STAT4. That's really the most proximal signaling. So, STAT4 measurements give us a way to characterize the EC50 of these molecules. And so, if you look at the masked molecule here, it has an EC50 that's, you know, say 650 picomolar. But if the unmasked wild-type IL-12 is about 8 picomolar, and if we take the masked one and we expose it to UPA, we totally regain the EC50 in terms of STAT4 phosphorylation. Likewise, if we expose it to MMP-2, we totally regain it, its activity. So, we show that the enzyme can regenerate the activity. So, what about the toxicity here? So, that was the whole point of the masking, was to improve the toxicity by limiting activation of NK cells in the blood and enhancing activation of tumor cells. So, in this case, we were injecting and then making bleeds So you see 3 doses, day 0, day 3, day 6, and bleeding 2 or 3 days after each injection. And then— and we were comparing doses that had similar efficacy. So 15 micrograms of the mass IL-12 on a molar equivalent of IL-12 versus 5 of the wild-type, similar efficacy there. So, but if we look at toxicity, interferon gamma is what you care about the most. So the wild type induces quite an interferon gamma response in the blood after that third dose. But here I'm just showing you the third dose, but substantially less in the blood. Or if you look at other inflammatory cytokines like TNF-alpha or IL-6, or chemokines like CCL2, we see strong reductions in the tox profile. Or if you look more interestingly, or maybe more clinically, at tox profiles, like liver tox with ALT and AST, strong benefits from masking even though the dose is higher. But remember, the efficacy of those two doses in terms of tumor killing is fairly similar. Do they kill tumors though? If it doesn't kill the tumor, then it's not interesting. It means I just killed the molecule. Indeed not. Here, if we take 5 micrograms of wild-type IL-12, it's very effective in this MC38 model. 15 micrograms, but it's toxic. 15 micrograms of the masked one, it's not toxic, but it's just as effective in the MC38 model. We're able to really substantially ameliorate toxicity without a hit on efficacy. Now, if we go back to this EMT6 model, the triple-negative breast cancer that's immune-excluded. Here, checkpoint blockade does something a little bit. But comparing checkpoint blockade to the mast cell IL-12, you see that we get a whole lot of complete responders with the mast cell IL-12 in a genuinely immune-excluded tumor that's based on its TGF-beta expression. If you look at other tumors, so back to B16F10, so also looking at the masking at that 15-microgram dose, in red, or the masking in combination with PD-1. So we do see complete responders in a very challenging tumor at a pretty low dose and a pretty nontoxic dose of the mast cell IL-12. So what about, you know, human tissue? So can we show that this cleavage really occurs and the unmasking really occurs upon exposure to human tumors? So, here we took a fresh frozen breast tumor and made homogenous of that tumor. And this is back to our EC50 curve, looking at STAT4 phosphorylation. So, one sees that the masked cytokine is still masked. The unmasked cytokine, just the wild-type IL-12 in blue, well, that's a lower EC50. So, we changed the EC50. By almost 2 orders of magnitude. If we expose it to the tumor lysate in the red dots, we completely unmask it. If we expose it to an adjacent normal tissue, ANT, adjacent normal tissue, we see that there's only a small amount of unmasking. So, the unmasking really is pretty tumor-specific based on the proteases that we chose for the unmasking. You can see that by blotting as well. Here, looking at the cleavable mask, exposure to the tumor lysate leads to cleavage and unmasking of substantial fraction of the molecule, even after 5 minutes exposure. We see some unmasking, whereas with the adjacent normal tissue, we don't see much unmasking at all. So, the proteases, even after homogenization, the proteases from the AMT are not very effective. This NC is just a not cleavable control, and we don't even see cleavage in the— of course, in the tumor after 6 hours. What about serum? Because you could posit that there would be UPA and MMPs escaping from the tumor microenvironment into the circulation. And so, in this case, we took serum from those patients. So, here, 6 hours exposure to the tumor leads to substantial unmasking in the protease-sensitive mastocyte IL-12. Whereas 6 hours exposure to the serum from that same individual, same patient, led to just a wee tiny bit of unmasking. So the, the, the masking and unmasking is quite tumor-specific, not to the adjacent normal tissue and not to serum from the same patient. So we are playing and have played this game with a number of molecules. As I mentioned, we, we started out with, uh, with IL-2, and we published on that if you want to go have a look at that. We did show that you could reduce the toxicity of IL-2 meaning so that the pulmonary edema that comes from capillary leak syndrome, we could reduce that with IL-2 and enhance its efficacy. So, we have worked on, but not yet published on, interleukin-15. We think these activating agents, like IL-12, are more interesting than the proliferating agents, like IL-2 and IL-15. I would like to show you some work on a chemokine, on CCL4. Very briefly. I think I'm still well on time to leave time for discussion. I just want to show you a couple slides on, on CCL4. There was a very— I should have cited it, but there was a very interesting study that published about 3 years ago from George Koukos's laboratory in Lausanne, where he correlated chemokine expression in tumors with sensitivity to checkpoint blockade. And the two that came out on the top of his list were CCL4 and CCL5, meaning tumors expressing one or the other of those two chemokines were much more sensitive to checkpoint blockade immunotherapy.

That—

we actually started this work before we published that, but that quite confirms our selection to ask, can we make a CBD CCL4? And if so, can the tumor retention create a gradient? Okay, a gradient of the chemokine. To allow it. So the, the chemokine that's in the blood is cleared super quickly. The chemokine that's in the tumor is cleared rather slowly. That gives us a gradient. Now, we don't expect monotherapy efficacy with CCL4, and indeed we don't see it. I'm not showing you that here, but this is in the B16F10. So, uh, with, uh, just checkpoint blockade in black or wild-type chemokine, A, it's cleared very quickly, and B, it doesn't form a gradient. So of course it doesn't do anything. That's a control where you certainly expect the answer that it doesn't do anything, and you just do it for a reviewer 3 who's going to demand it. Whereas the CBD CCL4, and this is with a single administration of the chemokine, really accentuates the responsiveness to the checkpoint blockade antibody. And we even got 2 complete responders. And we would rechallenge those 2 complete responders. They failed to grow palpable tumors. If we look at immune profiling in these tumors and ask— so just tumor volume, we see it's reduced. But look at total inflammation. This is on cells per mg basis. So total inflammation is up. If you look at these cross-presenting DCs, they're up by a factor of 2 or 3. CD8 frequency is up by a factor of, say, 2 or maybe almost 3. So, you know, strong benefits in the cellular readouts that correlate with the tumor growth. And so here, if we're looking at correlations between tumor volume and CD103 DCs, which are sensitive to CCL4— CCL4 makes them chemotact, But here we see 0.0001. So there's a strong correlation between the desirable cross-presenting DC and tumor volume that's associated with CCL4 treatment. The same thing was true with CD8 frequency. So our ability to increase— to chemotact for those cell populations, we believe, give a strong correlation. Interestingly, if you look at the correlation between CD8 frequency and CD103 DC frequency, there's a strong correlation there too, 0.001. So, you know, some of these tumors are really very sensitive to this CCL4 treatment. So, this is looking at the MMTD-PyMT, a transplantable one tumor. So orthotopic. So where we see that checkpoint blockade gave 1 out of 10 complete responders. So whereas the combination with the chemokine, just a single dose of the chemokine, mind you, gave 5 out of 10 complete responders. And of those complete responders, upon re-challenge on the contralateral side, all of them failed to develop palpable tumors. So all of them resisted tumor challenge. So, we can play this game of collagen binding on a number of molecules. In the last slide I'm going to show you, just to mention about toxicity, we talked about tox of IL-12 a lot. But here we were looking at tox of combination checkpoint plus CCL4. And we really saw really nothing interesting in terms of lymphocytic infiltrates in the liver or the kidney or the lung. Meaning all good news. We think the tox profile of this chemokine CCL4, even with checkpoint blockade, because only combination therapy makes sense, we think that tox profile is very favorable. So with that, I'll wind up and leave 5 minutes for questions. So as I said, we've played this little game a few times on different chemokines and cytokines, and we're quite interested in the idea of tumor retention. So with that, I'll turn it back over to the organizers for, for questions.

Thank you, Dr. Hubbell. That was a wonderful talk, and we do have a couple of questions in the, in the Q&A. For those in the audience, if you could type any questions that you have in that question and answer box at the bottom, we'll be happy to answer those live, and those that we can't get to, we'll try to have our speakers respond during the next presentation. So the first question that we have in the chat is from— and I apologize if I don't get names right— Alireza. And the question is, what protease was present in the tumor microenvironment? Was this MMP?

Yeah, we made the linker sensitive to MMP-2, MMP-9, and urokinase plasminogen activator. We didn't actually characterize the tumors to see which protease was doing the release. But, you know, we just said, like, there are going to be some of those proteases present. So we targeted 2, 9, and UPA.

Great. And the second question is from Nicholas. This is, have you looked to see if there's cytokine accumulation or PD in collagen-rich tissues such as the skin or heart valve?

We looked at interferon responses in these tissues, not the heart valve. That would have been interesting. But it's— so you see inflammation induction as measured by interferon gamma, for example, or just accumulation of the IL-12, really only where there's vascular leakiness. So, you know, when the endothelium is intact, there is essentially— so we did look at the whole heart. You know, there's very, very, very little accumulation in, you know, collagenous tissues when the endothelium is intact. When it's disrupted, then we see accumulation. You could ask the question like, where would it be disrupted? Like in sites of chronic inflammation. I don't think we would want to do this, at least with IL-12, in a patient with active inflammatory bowel disease, for example, or active rheumatoid arthritis. Now fortunately for us, those patients aren't normally immunotherapy patients anyway. But I do think we would— we could expect a tox profile in cases of chronic inflammation and thus vascular leakiness.

Great. The next question is from Eric. Did you test the non-cleavable IL-12 compounds in vivo? Is it active?

We did test them in vivo. So, the activity of the non-cleavable one is much less. So you could ask, is it just attenuated or is cleavability important? But we did see a strong dependence on the details of the linker. So for example, I called that one L6. That's because it was the 6th damn generation of the linker. So we went through several generations of the linker to get the sensitivity right so that the Vmaxes or whatever, of the, of the linker sequences are important.

Yeah, the next question is from Jack, and it has to do with the protein expression and stability of the CBD-IL-12 fusion. So could you speak to maybe the kind of the developability of that?

Yeah, sure, that's a really good question, you know, and it's very fortuitous. I wish I could say we designed it that way, but the CBD expresses really well. So, you know, with the CBD IL-12, we get about expression at about a mg per liter. That's, you know, compared to antibodies, that's low, but compared to the dosing of a cytokine or a chemokine, that's a really big number. So yeah, we think the developability is very high. Moreover, like if we think about CCL4, CCL4 is a nightmare to express in its wild type. But expresses really well when you put the CBD on it. That's just dumb luck.

I think we have time for maybe one more question. I'm going to skip to Alexandra who asked, can you comment on possible clinical scheduling such as continuous treatment versus priming only?

That's a good question. I'm just a chemical engineer, so I'll comment, but I'll probably say something stupid as I want to do. But I was thinking like dosing every couple of weeks. So we do have a dog study that we have initiated with a collaborator, a veterinary dog study, I mean. And so we'll see something more about dosing from that. But we've characterized T-cell infiltration in the tumor as well as a function of time, not just the chemokines and cytokines, but also T-cells. And we see like a 1-week big bump up in terms of T-cells. And so I think that says that we'll think of repeated dosing, not just priming. Repeated dosing more or less on a 2-week timescale, said by a chemical engineer.

Great. Well, I'd like to stop questions there, but thank you very much for your, your talk, Jeff. And if you wouldn't mind, Jeff, going into the Q&A window and maybe you could type answers to the, the remaining questions that are in there. And then I'll also remind the audience, if you'd like to ask those questions or get more information, join Jeff's breakout session at the end of the symposium today and have an opportunity to, to further dive in.