Session 1: Emergent Scalar Technologies (Panel Discussion)

Transcript

I had a question for for Jim actually which is that of course you mentioned are getting new antigens I was wondering in any particular tumor what proportion of the t-cell response is directed against new antigens is not known in the tumors that we have looked at I can only tell you for the cd8 t-cells most the CD force will probably help or two so but some problems managing specificity it's about half now those are patients that have responding and so if you looked at baseline prior to therapy my guess is much smaller number maybe of just a couple percent sorry yeah it's a question also to gym you know a lot of mutations are that that can be used as new antigen are not driving mutations and so it's easy for sales to down regulate them and I think that could be a challenge right and how do you plan to address that oh so I think your question relates to the fact that you can have neo antigens from proteins that are not essential for cell function and so the cell just quits regulate to which that function as a way of escape which is what's called immune editing it is that your play so I think the way you do that is that you use half a dozen you don't just choose one I think if you're going to base a therapy on this kind of thing I think any single antigen or any single t-cell receptor unless it's accompanied by a very aggressive epitope spreading is likely to do immunity but that's okay you know you can pick a half a dozen and and go forward that way that's my I think right now for low mutation burden tumors that's going to be very tough to do but as our algorithms get better as we feed more data those algorithms we will be able to make those predictions more effectively I would be interested in knowing now how the panel with define the title of this session scalar technologies and how it applies across the three topics that you presented to us today I think they're definitely a vectorized technology I think it means scalable I think it does mean scalable from the from the point of view that it can be scaled to the immune system or it can be scaled to looking at ocean systems essentially if that answers your question yeah I had the same question now I answered and finally I don't think there is a bigger scale than that is it so all three of you today talk about rare population of cells or even single cells that can have the response of some of the clinical phenotypes we see my question is like still it is a collective response our physiology of all these unique events that because your when we talk to the ocean about the unique different functions in the macro organ so my question is how we can do you have any ideas moving forward essentially how we need to collect information to translate all these emergent technologies to the clinic or understand more higher levels of organization maybe you could expand your question a little bit did you not maybe there um so essentially how by finding how individual cells are working or for example the like very rare information we need to a sample in order to have a a translational to how to say for the clean as I said how a patient is gonna respond you have fine for example this nail antigen and you have create a vaccine how you know it's gonna respond if those are the correct ones or not for example for you Jim I think these are very early days yet I can take a best guess so my best guess is that these cellular did these rare cellular populations are also the best biomarkers of therapeutic response or failure I also believe that I mean that's a very simple question okay if I showed a piece of data early on in my talk looking at two men filtrating lymphocytes and their functional performance right after the start of pd-1 therapy and so these were also rare cells it was about a thousand cells that were analyzed if you looked at those functions they were not tumor killer they were all recruiting other immune cells which is probably why there's that lag time before a response so i think with something like the immune system where my teacher of min ology is here in the room it's alan so if i say something wrong i'm get slapped on the wrist and he's gonna now tell you the answer to that question I'll take a shot but tell me what you think so I think with the immune system we actually have a pretty reasonable insight into when you have a cell and it has a certain functional performance what that implies implies replication inflammation what have you and so I think as we get better at understanding these rare cell populations where they are physiologically what their functional performance is we will all that is telling you much much more about how to make these therapies work so I think it almost everything we see one can think about a clinical translation of that knowledge whether it translates into a therapy or whether translates into understanding how patients responded don't respond I don't know but I you almost can't avoid it right now it's such a rich rich area I'll just try to look a little bit more on that one um so I got in the two immune profiling talks you know those are really a big part of them is focused on discovering interactions new antigens or receptors whatever else that's one side of it but the integration of data side I think Jim rightly pointed towards the the functional profiling of the cells and understanding how various cell types in both time and space are communicated in orchestrating each other and I think it's not more complicated than knowing that in any population of cells at any given time there are different cells sending different cytokines interacting with each other and I think this will you know that kind of data in forms on targets and informs on how what the pathways actually are and how to interview with them and I think that you know the ability to look at populations that are not synchronized and see what's happening within a cell and the ability to look at populations that are not functionally I won't say equivalent but similar though those two levels of detail lenok are needed for a long day a lot of these mechanisms and immunology is a great examples that so it's a question here in the back for Garrett one of the things that we begin to appreciate in numbers of microbes over the last few years is the capacity to process very large amounts of nutrients carbon nitrogen phosphorus and so on through the oceans and while we have to study at a single cell level how these microbes function they do work in communities and microbial ecology is for the first time now tractable at a single cell level and also at scale and I think that's exciting but what are your thoughts in terms of the 99 percent of microbes that cannot be cultured either because they're dependent on each other or live in conditions that we don't fully understand and with environmental context could your technology help in getting over some of these challenges in culturing these microbes yeah i would i would hope so I mean I mean this is why why we start this and what I like about the Hawaiian system is that it's so simple you know at least you think it's simple because we can only recognize two or three of the major players or two seem to be only two or three major players but what we hope to do in the end is to be able to synthesize these three players into a real community how are they interact what are the responses what are the mechanisms that they specialize in and then maybe apply that to the more complex places like the Arctic and the coast of Peru but yeah you're right where it's a little along a little ambitious to think that we we can understand how communities work by by studying the individuals but we at least need to have an indication of who's there but then we also have to have effective methods of coming in and measure how the community functions as a whole and you're one of the nice systems that with the metal back there and the D so far where you actually see that that role play or the roles that these organisms play is not always fixed that they have a flexibility and can assign each other roles and and form a more perfect community yeah I think that's very important can I uh can i address a question of a mic here in front of me so I want to switch slightly the perspective on these on this question this this morning I've been just overly impressed maybe did the extent that I should have come in better prepared with the extreme development of the technology it's allowing us to interrogate the immune system at levels that if you're not in this field you probably didn't fully understand it's truly remarkable what you share with us this morning so I'd like to shift it a little bit into the other side of the equation which relates to this autoimmune spectrum of diseases which we are seeing with increasing frequency in our society and there's a big question about why where it's coming from and what's the nature of these conditions you when in medical school we learned that these immune responses were to single and double stranded DNA that the suggestion was that these were intact native forms of DNA but as you all have helped us to understand when we start to interrogate at a greater degree of specificity maybe we are looking at other than native forms of antigenic determinants that then relate to an immune response so I'd like to knowing that I'm kind of deflecting the conversation slightly get your opinion about how what you're learning helps to unearth are better understanding of the etiology of autoimmune disease okay so I suppose the first thing I would say is that um you know it's been in human patients very difficult to come up with at least on the antibody side and probably even more so on the t-cell side to identify immunogenic epitopes and autoimmune disease and that's been a tricky thing to do and it's been tricky to connect it also with with the actual antibody sequences there are some diseases for instance that we know there are big antibody antibody responses in and yet we found very few targets like ms an example where there's been very little done so I think as you know largely driven by amino oncology as these technologies are pushed to have more throughput and more precision it opens up opportunities to better dissect just what's driving that and so that you know that's I I think there's hope that we're going to have a much better understanding of that in the near future and certainly everything we're learning from you know oncology in terms of manipulating T cells and NK cells and macrophages and dendritic cells and the million different cell types that exist if you believe a cell type is really a thing all of this I think is going to put some some color on it and Alan's got his hand up so he should definitely say something I'll add to that from a different perspective much of the conversation here has been about the the the inducing side of autoimmunity with respect to antibodies with respect to their various changes that happen so that self is recognized but the other half of this is the effective mechanisms so for example we know that if you could interrogate properly inflammatory cells that are circulating in the blood you would have a very good chance I think of discovering the effective mechanisms that are induced by the autoimmune insults as they begin one good example of that is tnf tumor necrosis factor which in the case of rheumatoid arthritis I mean enbrel which is an antagonistic tnf resulted in engine squandering sixteen billion dollars on immune X it's a it's it's a very effective drug it's almost almost a miracle drug and that you can tell from looking at the using exactly the same technology that was described here and looking at the effector cells as opposed to the ones that caused the initially I could mutter ok this is a question for dr. Heath in your talk use you showed a graph where you saw increased mutational baggage and a tumor and correlated that to an increased chance of that tumor being successfully treated with immunotherapy and I understand the connection between you know increased mutational baggage and increased infiltrating lymphocytes thereby making those tumors more susceptible to neo antigen treatments but I also see how increased mutational baggage could completely change the landscape of a cell surface repertoire thereby making those tumors less effectively treated by immunotherapies so you know with that in mind do you think that there's another area beyond mutational baggage with which to look at tumors and say that cancer would be a good one to think about for immunotherapy yeah absolutely i think if you look at the well so when you look at T cells there in a tumor you um you would like to know why they're there ok so some of them may just be spectators some may be helpers some may have been called there from some soup reason for the tumor the ones that are there specifically because they're trying to attack the tumor weather shutdown or not those cells probably have upregulated checkpoint markers because it being turned off and so I think if you want to understand let's take it very hard to treat tumor it could be that we simply don't know the checkpoints and so these very rare cell types can be mined to identify exactly what checkpoints you want to block to enable that patient to respond um but I think it does come down to you know if you did this randomly for t cells no tumor you're gonna you're never going to figure out anything you need to do a very very specific profiling but I think the information content is we're just tapping it and we're just beginning to understand the simplest crude istics I mean the simplest crudest thing is it find it so that's going to attack a to make more of it what we're saying but there's much more to do with the deep analysis of these populations just maybe after that there is an immense focus on T cells but they're not the only cells right so what about dendritic cells are with NK cells what about what about tumors that have down-regulated MHC so T cells probably work so great what are the other cell types I think people are picking up on that but it's pretty really so 200 meters down off of Hawaii what's the flux of photons per organism telemeters it's well the light intensity is point one percent there okay I don't know what the flux is for organism but then you need also the absorption cross cross section of the molecules I just wondering whether it made sense that organisms you know that that energy input level they could they could survive at that lovely it's generally accepted that it pays up to 100 meters deep to be photosynthetically active so the ones that are living between 100 and 200 meters probably have alternative oh do we know ways no no so is there a genomic difference although they may be waiting to be welled up in the dorm or something I i actually think they're in a in a recent stairs yeah just a resting face and so is there no genomic difference between the surface members of population and the hundred fight all for clays at between 100 and 200 meters in about equal numbers so is there a you know are the ones that are deeper better does their genome show that they've got well they can't they can't select where they are gonna hang out the right so so so you it's it's all in my view it's impossible for them to adapt to the environment because they're there by accident and whatever environment they are adapted to problem each more light and they need to be welled up during a storm or something in order to to proliferate and then go down with the in a passive manner to deeper but but the the thinking about it is not well developed at all if you ask oceanographers they really haven't thought that much about how how these LG distributes themselves over over these layers max thank you this questions for the one on the on the left there from my perspective your left side neither have you thought about trying to apply your micro fluidics to making tumor-specific detectors so for examining maybe if you had a person who has melanoma and so you could easily get a tumor sample in a normal sample finding marker you know marker antibodies for that so that they could know how much of the skin to excise I so yes we have we've made me done some some experiments along that that vein is quite interesting so when when hybridoma technology was first invented 40 years ago people weren't looking for antibodies they were looking for antigens they were looking for tumor specific antigens they would immunize with tumors and then they would counter screen antibodies against normal cells and tumor cells and that's how a lot of the tumors food engines were found and that kind of passed away when we got into the era of genomics but I would say that you know the immune system in addition to you know being good at making reagents and finding the antigen with the reagent in one shot also recognizes all kinds of things you'll never see with genomics you know post-translational modifications like oscillations everything else and technologies that have now brought like we can screen 5 million cells that that is brought hybridoma you know forward three or fours of magnitude and speed and and throughput I think it's it's very conceivable particularly go to other species as well where we haven't combed over all the antibodies we've done that pretty well with mice but what about dogs what about you know going to different libraries and with higher throughput I think it's a really interesting prospect that you could use natural immune systems to find novel antigens and of course the beauty of it is that when you do it you've got the reagent in hand and and for things like you know immunohistochemistry and such it be quite interesting we just have we haven't had time to pursue that but that's been kicked around I actually were to grant into DoD hated it so I didn't get money for it we have wanted the back question for GM and Carl so you know India you both talk about isolating T cell subtypes from tumor and to guide as therapy so i was wondering there any effort that you try to capture the air the interacting partners of this t cells or you know to preserve the spatial information in solid tumor to identify new target um I'll take a shot at that so in principle our method is amenable to doing the entire analysis in an intact tumor the challenge is that as soon as you fix the tumor all your peptide MHC molecules fall apart and say you have to do it on very freshly real sector tumor and we started doing that and the goal is that you can actually see what is clonely expanded in the tumor as what tells you something about immunogenicity and I think just the more you can image the physiological environment the more you're going to learn so we can't do that yet but but we have early data that says it should be possible and I think it's a very valuable direction to go into yeah it goes just dad no just a little bit more so I think in the t-cell side there's you know two important problems and Jim hit one of them which is what is everything reacting to what are what are the antigens and then once you have the antigens what's the T cell and fishing that out those are sort of complementary I'm quite curious to see how mass spec approach has also come forward in terms of you know determining MHC presentation and what peptides are actually there and I think it's still early days for that but you can imagine that it's not just going to be the types of technologies we talked about there's going to be genomic methods layered on this and proteomic and that's all going to come together in ways that we necessarily can't or that we may not be able to predict right now Rob where did the T cells come from this is all based on a premise that their t cells and the T cells are prevented from going after the tumor because it's checkpoints and things like that but of course the process of making it the T cell that is prepared to go after the tumor if there were no checkpoint inhibitors says they came from somewhere and where did they come from what what's the basis that finest they came from evolution nobody passing by the details what what what process is producing the stuff that goes into the APC and leads to the cascade degenerates eventually a t-cell that is focused in that direction whether the T cells have come through combinatorial recombination of BD pdj and other genes and then counter selection for recognition itself so you you have to get lucky that you've got one that recognizes so you're saying it's just random motion that's this there well it's not people shown it's not perfectly random but it's highly stochastic let's it is that rather than then tumor cell death by some other mechanism or by weak interaction with some other T cell that isn't enough to kill off major part what you were whether the fundamental clones need to be there but then they will respond in the context of random cell death and inflammation and you know they need to get triggered I think it's probably a little more deterministic than that but I don't know because you a tumor is in many ways like an open wound right and so you do have innate immune cells that are there that are chewing up stuff and presenting antigen and they float away and so there is some recruitment that happens at a site through such actions like that we also tumors are often they have necrotic parts to them and parts in which cells are dying for who knows what reason information that's nonspecific and anoxia around the edges of the tumor mass but the question is do you have to have that I asked it because it raises the question of whether you in fact want to do another one of those non-intuitive things of doing something that would seem to damage cells non-specifically in order absolutely fragments that absolutely laughs I think that is without question one of the ways people are thinking about how to get a non-responsive tumor to generate to catalyze an immune response absolutely so you have physical damage you have radiation what else could you do to generate a lot of pieces that would start generating t-cells directed toward that repertoire chemo you can starve them of you know there's glycolysis inhibitors that caused massive cell death but perhaps cancer cells tend to want to die a little more than healthy cells so there's some very non selective approach is one could do to catalyze if you titrated them correctly there are clinical trials where people are doing a much more you know deliberate approach where they're injecting viruses you know that encode CMV peptides and they're deliberately pine and lacquer adjuvant to these things and a lot of their therapies we had in fact we look back they kind of worked that way so we may end up concluding that much of what has been conventional wisdom in in tumor treatment is in fact counterproductive or it's been working for reasons that we didn't understand when it does work got working for not working for reasons exactly right that's right well is there any evidence that under conditions of chemotherapy there is more access to these new antigens or that they're presented to a greater extent on antigen-presenting cells well chemotherapy tends to install genetic instability in tumors one thing anta mcquillen radiation has typically thought of as the way to expose a tumor to something like the immune system and there is evidence or that in the context of radiotherapy you would have more of these antigens presented to the immune system yeah i mean i would say it's a little sketchy because I mean you saw the state of the art management measurement just now and there's not much not much known out there but you know i was on a stand up to cancer review panel for pancreatic cancer a year ago and there you have this problem in that the pancreas as the tumor has a bit of a sheath around it and everybody was trying to promote tumor killing by radiation every single bullet that came in there were all immunotherapy proposals they all started with let's loosen up that that that sheath and now whether that's a good strategy or not you know maybe it's like lemmings going over a cliff it's just everybody says it's a good idea and that's what they do but it it made sense well so is the question for all three of you and if you could probably answer them separately it's more got to do with cost and translation and I think each of each of you have got the different technologies that you are applying to a different answer but I think one of the questions that we always ask is what is that end cost going to be / essay or identification so Gareth for you would be this would be continual measurements that would need to be taking to understand the health of the oceans and and what that cost would be given the current technology that you have and can you see getting cheaper and for Jim as far as its it's to look at the cost of identification of these rare events that are occurring for identification of the T cells and the t cell receptors and finally for Carla be with the scale that you have in the micro fluidics and the scalar component of being unlimited trying to try to push that through can you see your cost and identification of anybody's getting to a point where it's cheaper do that than what you can currently do by other methods well let me start yeah I I think it's useless to talk about cost if you don't talk about benefit and right now what I'm doing doesn't have much benefit so it's too costly anyway and that's what we call it research if research shows that it's good for something then it often doesn't matter how expensive an approach is so so comparing the cost of a novel technology to an existing one already shows you that the new technology hasn't anything to go for it and you're only going by price that's not for scientists do we should we should develop better technologies irrespective of cost show their usability their utility and then we hope that the engineers afters we'll make them cheaper and make them affordable to everyone but the real crux is that we have to prove utility first in in my case actually I haven't done that yet yeah so I'll answer this question too because it's it's an interesting question I mean so what an analysis that we do if we looked at 20 different patients we would see 20 different sets of neo antigens and we'd be lucky of one of them overlap one and the event not one patient and so taking that to its logical conclusion you would say you have to analyze everybody separately you're developing separate vaccines or separate T cell therapies okay and that's about the dumbest approach to thinking about that problem I can come up with the reason why and this is something that Lee is taught us all is that in technology advances exponentially even though we tend to thank linearly so let's say what happens that we identify these T cells as telling us the best checkpoints then suddenly we have an incredible treasure trove of t cells but is one common drug or let's say that we are able to use this data sets we develop to develop terrific MHC predict antigen prediction malagori algorithms so we no longer have to do the measurements which is not so ridiculous and in which case you know you rely on the technologies of Juno and titanal artists who are already dead set on on developing cell-based therapies so I think if you just look linearly I agree but its research and and the more you find out the better you equipped to solve a problem and you you're always going to take the most efficient way to solve that problem me yes um so how does I I would say that that the way to take our technology is right now it's already much cheaper on a per antibody basis than any other approach to generate antibodies there are cases where you don't care about getting a great antibody and it's pretty immunogenic you can do navio doma for a couple thousand bucks and then you should but in cases where you're looking for something that's high performance and we're having multiple candidates or we're going off your target that's difficult I think it's already there and I think it's going to get cheaper and cheaper by far the dominant cost right now is is labor and time associated with expression and characterization downstream the discovery is now very quick and when done at scale you know there is through some next-generation sequencing but you amortize that across a few hundred antibodies and ends up being a few dollars per antibody so from a cost of goods perspective it's not a lot from a labor and Technology from a labor perspective it's still we solve challenges on the downstream expression and ization and I there are good solutions to that they're industrialized robotics and you know I don't think need to be really innovative you just have to get mechanized to do it well well I'm afraid that's all the time we have so a big thank you to our speakers many times for questions