Generative Biology- Learning to Program Cellular Machines

Transcript

• 00:00Thanks up and also wrapping
• 00:01up the workshop. It's the
• 00:03second keynote,
• 00:04doctor Wendell Lim,
• 00:06who's, visiting us from UCSF.
• 00:09He actually, did his postdoc
• 00:10at Yale,
• 00:11and then he got his,
• 00:13undergrad from Harvard and PhD,
• 00:15in biochemistry
• 00:16and biophysics in MIT.
• 00:19And Doctor Lim has made,
• 00:20pioneering contributions to multiple fields,
• 00:22including cell signaling,
• 00:24systems, synthetic biology, and more
• 00:26recently in immune cell engineering.
• 00:27And so he's currently the
• 00:28Bayer's distinguished professor,
• 00:31of cellular molecular pharmacology
• 00:33and the director of the
• 00:34Cell Design Institute at UCSF.
• 00:37So I'm gonna hand it
• 00:38over to Wendell. Thanks for
• 00:39coming.
• 00:40This works for me. Okay.
• 00:41Hi, everyone.
• 00:43So it's great to be
• 00:44here. And what I'm gonna
• 00:45do is tell you about
• 00:46our work,
• 00:48trying to engineer,
• 00:50new cellular behaviors.
• 00:52So what's shown here on
• 00:53this slide is a a
• 00:54really beautiful movie by Alex
• 00:56Ritter. It's a light sheet,
• 00:57microscopy movie of a, a
• 00:59t cell. And we have,
• 01:00as you know, these cells
• 01:02running around your body. They're
• 01:03patrolling your body, and they're
• 01:04able to defend you, from
• 01:05various infections and and diseases.
• 01:08And, you know, we are
• 01:10very interested in,
• 01:11harnessing those capabilities and asking,
• 01:13can we ask these cells
• 01:15to do new things that
• 01:16they don't normally do?
• 01:18And we're that's also a
• 01:19very fundamental question that we're
• 01:21interested in because,
• 01:22in general, you know,
• 01:24cells are the sort of
• 01:25smallest living unit of of
• 01:27life that really do complex
• 01:29level functions. And,
• 01:30they're able to sense lots
• 01:32of things, integrate that information,
• 01:34make lots of complex decisions,
• 01:35and they have this capability
• 01:36that molecular systems really by
• 01:38themselves don't do. They they
• 01:40are a set of molecules
• 01:41that work in concert together.
• 01:44So,
• 01:45when we are in this
• 01:46case, you know, traditionally, biology
• 01:49has been a field of
• 01:49of studying these, you know,
• 01:51complex evolved organisms and trying
• 01:53to take them apart. And
• 01:54we've gone through the era
• 01:55of really kind of now
• 01:57understanding the genomes,
• 01:58and the parts of all
• 01:59these things. But,
• 02:01for the test that we're
• 02:02talking about, really,
• 02:03what we need to do
• 02:04is,
• 02:05if we eventually want to
• 02:07be able to have, like,
• 02:07a chat PTP equivalent that
• 02:09says, we wanted to sell
• 02:11that can do x,
• 02:13and then hope that it
• 02:14would spit out some genetic
• 02:15information that we've upload into
• 02:17that cell. We really need
• 02:18to understand this the hierarchies
• 02:20of of biological language and
• 02:22encoding in a much
• 02:24deeper way.
• 02:25That is, you know, we
• 02:25know everything is encoded ultimately
• 02:27as sets of molecules and
• 02:29genes, but that these,
• 02:31molecules come together,
• 02:33in various cellular circuits and
• 02:34subroutines and then the cells,
• 02:36of course, have to talk
• 02:37to one another,
• 02:39and that much of the
• 02:40the the complex behavior that
• 02:41we see in real biology
• 02:43comes from many different layers
• 02:45like this. And so that's
• 02:46a lot like a very
• 02:47complex grammar.
• 02:48So I'm also gonna reference
• 02:50Hamlet,
• 02:51but we have, you know,
• 02:52these basic words that,
• 02:54we want to understand how
• 02:55we put them together to
• 02:56build sentences, to build essays,
• 02:58to make arguments, to write
• 02:59books. And we want to
• 03:01not just take apart classics,
• 03:03but we wanna be able
• 03:03to write our own, new
• 03:05books. So that's we're now
• 03:07thinking about this as more
• 03:08like generative biology
• 03:10that we wanna try to
• 03:11understand this hierarchical,
• 03:13sort of structure
• 03:14or grammar of biology.
• 03:16And then can that really
• 03:18help us,
• 03:19to design cells that do,
• 03:21really complex and important things?
• 03:25So the, let's see. Okay.
• 03:29Okay. So, we're working on
• 03:31a couple different problems, but,
• 03:32you know, in in all
• 03:33cases, you need to kind
• 03:34of rephrase a traditional problem
• 03:36like in immunology. You might
• 03:37ask how to how to
• 03:38immune cells recognize and kill,
• 03:41disease causing foreign cells that,
• 03:43without causing broad damage.
• 03:45We're also working on development.
• 03:47I'm not gonna talk about
• 03:48that today, but in the
• 03:49case of of, immunology,
• 03:52to rephrase this, as a
• 03:54generative design question,
• 03:56we want to ask if
• 03:57we understand the design logic
• 03:59of biological systems, how can
• 04:01we, for example, engineer immune
• 04:02cells to precisely recognize,
• 04:05and kill solid tumors that
• 04:06normally that evade the natural
• 04:07immune system,
• 04:09or other sorts of, complex
• 04:11disease,
• 04:12tissue based diseases like autoimmunity,
• 04:15fibrosis, etcetera.
• 04:17So as I said before,
• 04:19you know, right now, the
• 04:20way that we interface with
• 04:21disease is largely, not always,
• 04:23but, through molecules, small molecules
• 04:25or biologics.
• 04:27And these are very, very,
• 04:29obviously,
• 04:30amazing,
• 04:31entities, but they tend to,
• 04:33again, have these systemic activities,
• 04:35and that and whereas, you
• 04:37know, what we're hoping is
• 04:38that cells have this ability
• 04:40to migrate, to sense different
• 04:42things at these different scales,
• 04:44and and decide when and
• 04:45where they will function,
• 04:47and, that they can, as
• 04:49I said, migrate, they can
• 04:50adhere, they can decide to
• 04:51stay somewhere, they can proliferate,
• 04:52they can talk to other
• 04:53cells. So we think it
• 04:54is, possibly a much more
• 04:56powerful way to interface, especially
• 04:59with complex diseases.
• 05:01And so when we wanna
• 05:02try to program cells, I
• 05:04mean, many people, of course,
• 05:05are familiar with,
• 05:06the great success of CAR
• 05:07T cells, chimeric antigen receptors,
• 05:10T cells that are able
• 05:11to redirect a T cell
• 05:12killing response
• 05:15to a a a a
• 05:15specific tumor antigen bearing cell.
• 05:18And,
• 05:19but, you know, although that's,
• 05:20you know, recognizing one thing,
• 05:22in many ways, we know
• 05:23that the CAR T is
• 05:24really about interacting with a
• 05:26network that's in the tissue
• 05:27in the body. They have
• 05:28to interact with the tumor,
• 05:29the stroma, other immune cells,
• 05:31and really, so, you know,
• 05:33I think in many cases
• 05:34in normal biology,
• 05:36physiology, as well as things
• 05:37that we'd like to do
• 05:38in terms of remodeling or
• 05:40treating disease. This is about
• 05:41kind of trying to rewire
• 05:43these cellular conversations and circuits.
• 05:47And so what is it
• 05:48that we want to do?
• 05:49If if we wanted to,
• 05:50like, draw in new new
• 05:51circuit connections,
• 05:52how do we connect these
• 05:53cells? And so,
• 05:55there are obviously a lot
• 05:56of different ways, but, I
• 05:58guess, one of the simplifications
• 05:59we're trying to make is
• 06:00to say that really there
• 06:01there are just a few
• 06:02types of state changes that
• 06:03you see when one cell
• 06:04talks to another cell.
• 06:06So if this particular cell
• 06:07here in node saw x,
• 06:09y, or z from another
• 06:10cell, it could turn on
• 06:12new new signals. It could
• 06:13turn on receptors that allow
• 06:15it to sense things. It
• 06:16could move or change its
• 06:17shape. It could adhere to
• 06:19things and stay in one
• 06:19place or could divide and
• 06:21grow,
• 06:22or die.
• 06:23And so we're interested in
• 06:25trying to build sort of,
• 06:27in a sense, domesticated modules
• 06:29that we can utilize to
• 06:30execute these sorts of functions,
• 06:32genetically encoded elements that we
• 06:33can put in. We're inspired
• 06:35by the the the cars,
• 06:36as I said, which is
• 06:37taking,
• 06:38an an antibody that recognizes
• 06:40an antigen of the user's
• 06:41choice,
• 06:42and fuses it to elements
• 06:44from the t cell receptor,
• 06:45which now allows when that
• 06:47t cell recognizes that target
• 06:49antigen, it now launches this
• 06:51complex t cell response,
• 06:53to proliferate,
• 06:54kill, and secrete. And that's
• 06:56the basis of our, CAR
• 06:57T therapies.
• 06:58We've been building a number
• 06:59of other things. One of
• 07:00them is the the synthetic
• 07:02NASH or syn NASH receptor.
• 07:03This is a, another chimeric
• 07:05type receptor that is actually,
• 07:07we think, much more flexible,
• 07:09allows us to connect,
• 07:10almost any input to any
• 07:12output. The idea here is
• 07:13that,
• 07:14based on the notch receptor,
• 07:16the,
• 07:17you can put a, extracellular
• 07:19antibody on the outside for
• 07:20an antigen of choice. And
• 07:22then the middle part of
• 07:23it, actually, when this binding
• 07:24is engaged, it cleaves the
• 07:26receptor and releases
• 07:28an intracellular transcription factor that
• 07:30can go into the nucleus
• 07:31and turn on a target
• 07:32gene that's driven by by
• 07:34the recognized the cognate promoter.
• 07:36And so what's great is
• 07:37you can change what the
• 07:39cell senses, and you can
• 07:40plug in any genetically encoded
• 07:42element here in the payload
• 07:43or multiple ones and create
• 07:44your own programs of x
• 07:46turns to y.
• 07:47So that's very flexible. We
• 07:48can do things like we
• 07:49can turn on a car
• 07:50in series after a Synash
• 07:52and actually have two different
• 07:53antigens that are required in
• 07:55sequence
• 07:56to, give you much more
• 07:57control.
• 07:58Another thing is, the synthetic
• 08:00adhesion molecules. We found that
• 08:01you can take a antibody,
• 08:03a tunable antibody, and then
• 08:04link it to different intracellular
• 08:06domains that are associated with
• 08:07cell adhesion. These engage with
• 08:09the cytoskeleton and create force
• 08:11and can create really strong
• 08:13and different kinds of attachments.
• 08:14And that's another important thing
• 08:16is that cells,
• 08:17they physically organize into tissues
• 08:19or, they bind to partners,
• 08:21recognize partners. And so really
• 08:23this,
• 08:24being able to both tune
• 08:25their physical organization kinda how
• 08:27they're physically wired with how
• 08:28they're biochemically wired is, I
• 08:30think, a really powerful thing.
• 08:31And then another example is
• 08:32we have recently gotten some,
• 08:34nice results on some synthetic,
• 08:35chemokines. This is very important
• 08:37for the immune system because,
• 08:38of course, as well as
• 08:39in development because,
• 08:41a lot of what a
• 08:42cell does is is determined
• 08:43by, where it's told to
• 08:45go. So these chemokine receptors,
• 08:47specify that cells to, for
• 08:48example, go to the lymph
• 08:49nodes and talk to other
• 08:50cells that have the same
• 08:51receptors. So it's a way
• 08:52for to mediate at this
• 08:54sort of high level, large
• 08:56scale,
• 08:57coordination and communication between cells.
• 09:00Okay. So oh, okay. This
• 09:02is screwed up. Sorry. So
• 09:04I'm gonna tell you about
• 09:05two things very briefly today,
• 09:07just as examples,
• 09:08of things that we're we're
• 09:10trying to do and have
• 09:11had had some success in.
• 09:13One is actually the idea
• 09:14of trying to,
• 09:16engineer cells to recognize,
• 09:19a a tissue, in this
• 09:20case, the brain. The idea
• 09:21is that can we actually
• 09:23combine kind of molecular scale
• 09:25recognition
• 09:26with,
• 09:27anatomical recognition. So I'll tell
• 09:29you about developing this kind
• 09:30of tissue GPS
• 09:32sensor, that can deliver,
• 09:34cellular actions to the brain
• 09:35and then how we can
• 09:36use that in different directions
• 09:38to either attack brain cancers
• 09:40or to, for example, attack,
• 09:42or treat,
• 09:43neuroinflammation.
• 09:44And then,
• 09:45related to that, I'll also
• 09:46talk about our our efforts
• 09:47to actually create cells that
• 09:49generate,
• 09:51customized
• 09:52multifactor,
• 09:53immunosuppressive
• 09:54programs,
• 09:55that, for example, can protect,
• 09:57against neuro inflammation
• 09:58or can protect transplanted
• 10:02organs, for example, in this
• 10:03case, beta islets from, immune
• 10:05rejection.
• 10:06So let me talk first
• 10:07about the brain, this kind
• 10:09of idea of a GPS
• 10:11in the cells that they
• 10:12can know where they have
• 10:13to go and and turn
• 10:14on specific responses.
• 10:16And this is we we
• 10:18were really interested in trying
• 10:19to do this in conceptually
• 10:20because, as I said, one
• 10:22of the things about molecular
• 10:23therapeutics
• 10:24is that even if you
• 10:25target a CAR T with
• 10:26just, you know, one antigen,
• 10:28is that those,
• 10:29that we have the same
• 10:31molecules, they operate in many
• 10:33different places in the body.
• 10:34So inherently, that's why you
• 10:35get a lot of cross
• 10:36reactions and toxicities.
• 10:38What we would love to
• 10:39be able to do is
• 10:40to be able to restrict
• 10:41a drug to act only
• 10:43in a target tissue, say,
• 10:44like the brain, so that
• 10:45you get much more specificity.
• 10:47And this is really kind
• 10:48of like saying, well, if
• 10:49you only had a street
• 10:50address to mail a letter,
• 10:51it could go to many
• 10:52different cities.
• 10:53But if you combine a
• 10:54street address with this higher
• 10:56scale thing like a ZIP
• 10:57code, you get the it
• 10:58gets to the right place.
• 11:00And so, this kind of
• 11:02thing is very difficult for
• 11:03a molecule to do, but
• 11:04a living cell, this is
• 11:05really what they do for
• 11:06a living.
• 11:07They can integrate information at
• 11:09multiple scales. Okay?
• 11:11So, Milos Simic is a
• 11:13a fellow, in in our
• 11:14institute that really took this
• 11:15on, and he asked, how
• 11:16can we try to do
• 11:18this? And the idea was
• 11:19to,
• 11:20use bioinformatics
• 11:21to screen for,
• 11:22BRAIN or CNS specific extracellular
• 11:25antigens, some kind of marker
• 11:26that we could recognize
• 11:28and then design a synapse
• 11:29receptor that could detect that
• 11:30and then use that to
• 11:31induce,
• 11:33in t cells,
• 11:34expression of a therapeutic payload,
• 11:36either a car that could
• 11:37attack a brain tumor or
• 11:39say a suppressive
• 11:40cytokine that could suppress neuroinflammation.
• 11:43So, we worked with Olga
• 11:45Tronskaya,
• 11:46a bioinformaticsist
• 11:47colleague at Princeton,
• 11:49and,
• 11:50looked for what were good
• 11:51candidates.
• 11:52And, and then we also
• 11:53worked with, Deb Sidu,
• 11:56a colleague who who's who,
• 11:57pans for, antibodies.
• 12:00And, what we found is
• 12:01that there are a couple
• 12:02different things that you could
• 12:03recognize in the brain, unique
• 12:04molecule markers. There were markers
• 12:05that were unique on neurons
• 12:07like this, neuro neural specific,
• 12:09cadherin. There are various, molecules
• 12:11that are specific to the
• 12:13myelin.
• 12:14But then, but one thing
• 12:15that I didn't realize at
• 12:16times that the brain has
• 12:17a very unique extracellular matrix.
• 12:19It forms, for example, the
• 12:20perineal nets around synapses,
• 12:22very important for that. And
• 12:23there are a bunch of
• 12:24molecules that are quite unique.
• 12:26One of them is Brevacan
• 12:27or BCAN,
• 12:29and we were able to
• 12:30find that this was we
• 12:31raised the Synash
• 12:33receptor against this and found
• 12:35that, in the end, this
• 12:36was one of the best
• 12:36ones that we we had.
• 12:38So I'll tell you about
• 12:38that.
• 12:40Okay. So how do we
• 12:42design a brain primed glioblastoma,
• 12:45cell therapy? There's a lethal,
• 12:47brain cancer.
• 12:49So
• 12:50it's been known for a
• 12:51long time that, a lot
• 12:52of,
• 12:53glioblastomas and other brain tumors,
• 12:55and in fact, many tumors
• 12:56have these common,
• 12:58tumor antigens,
• 12:59mostly embryonic sort of proteins
• 13:01that are expressed, improperly.
• 13:04And f a two and
• 13:05I l thirteen r a
• 13:05two are examples of antibody
• 13:07of antigens
• 13:08that are commonly expressed on
• 13:10many
• 13:11gliomas,
• 13:12but, but the but the
• 13:13problem is that these are
• 13:14also expressed in a lot
• 13:16of normal tissues in in
• 13:17lower levels maybe elsewhere,
• 13:19not in the brain, but
• 13:20elsewhere. So the idea here
• 13:22was could we improve on
• 13:23these by combining them and
• 13:25integrating multiple antigens? The idea
• 13:27being that let's take a
• 13:28SynNotch that recognizes a BCAN,
• 13:30and that's gonna be the
• 13:31priming interaction that will now
• 13:33turn on the expression of
• 13:34a car. Now this case,
• 13:35the car is two headed,
• 13:38for both of these things.
• 13:39So we one of the
• 13:40things that a lot of
• 13:41these tumors do is they
• 13:42escape if you need to
• 13:44kill one of those things.
• 13:45So we're gonna use the
• 13:47the brain to trigger everything.
• 13:49And the great thing is
• 13:50that, like, you the the
• 13:51the the tumor can't get
• 13:52a grant advantage by mutating
• 13:54BKAN in the brain. There's
• 13:56no selectable advantage.
• 13:57But then we're gonna cast
• 13:58this more complete net of
• 14:00killing two common antigens. But
• 14:02that what, and and importantly
• 14:04to know is that that,
• 14:05when a synosh when a
• 14:07t cell gets activated by
• 14:08synosh, there's kind of this
• 14:09blast radius of about a
• 14:10hundred microns where it can
• 14:12operate and start killing once
• 14:13the CAR
• 14:14is expressed.
• 14:15And as I said, these
• 14:16two antigens, the killing ones
• 14:18actually are are not expressed
• 14:19in the normal brain. So
• 14:20lobosoma is the only place
• 14:22where brain plus these two
• 14:24antigens,
• 14:25works.
• 14:26So, what's shown here is,
• 14:29hopefully
• 14:30okay. Yeah. Is a movie
• 14:31of a a t cell
• 14:32with this SynNotch and with
• 14:34a green reporter that turns
• 14:35on when the SynNotch is
• 14:36activated and it's interacting with,
• 14:38the surroundings of an astrocyte,
• 14:40which expresses BCAN in this
• 14:42ECM and it turns on,
• 14:43goes from green blue to
• 14:45green. I'm sorry.
• 14:46And so we can take,
• 14:47these,
• 14:48this kind of cell and,
• 14:50for example, turn on a
• 14:51car,
• 14:52and we can look at
• 14:54the retrieve the cells from
• 14:55the brain as well as
• 14:56the spleen spleen in the
• 14:57blood, and we only see
• 14:58strong activation,
• 15:00by a GFP marker,
• 15:02in the brain. So it's
• 15:03selectively,
• 15:04primed in the brain. And
• 15:05then when we when we,
• 15:07put a brain tumor in
• 15:08here
• 15:09and then we, give them
• 15:10the the these cells, you
• 15:11can see they, are able
• 15:13to clear that tumor,
• 15:14completely and give you really
• 15:16great survival.
• 15:17This is one of the
• 15:18best, results that we've seen
• 15:20in this kind of animal
• 15:21model. I should say also
• 15:22we see,
• 15:24a hundred days later, we
• 15:25still see after the tumor
• 15:26is cleared, we still see,
• 15:29a resident memory like cells,
• 15:31of these these CAR Ts
• 15:32in the brain, and they
• 15:33are the mice are resistant
• 15:35to rechallenge
• 15:36with even in the contralateral
• 15:38hemisphere.
• 15:39So so, it seems to
• 15:41be working really well.
• 15:43And then, in addition,
• 15:45we have done experiments where
• 15:47we put,
• 15:48the same tumors in the
• 15:49brain or in the flank,
• 15:51and what you can see
• 15:51is that on and then
• 15:53inject the cells and these
• 15:55tumors are in the same
• 15:56animal, but only the ones
• 15:57in the brain are cleared.
• 15:58The ones in the flank
• 15:59are not. And I should
• 16:00say that these are are
• 16:02are BKAN,
• 16:03sensors are responsive to both
• 16:05human and mouse. So it's
• 16:06really priming based on the
• 16:08endogenous mouse,
• 16:09BKAN.
• 16:12So, yeah, we see, brain
• 16:13restricted activity. Now,
• 16:15Milosz, wanted to also say
• 16:17if we have this kind
• 16:18of general,
• 16:19module that can say this
• 16:21is where you're gonna act,
• 16:23anatomically, could we use it
• 16:24to produce different kinds of
• 16:25payloads that would maybe push
• 16:27things in the opposite direction
• 16:28like in, neuroinflammation?
• 16:30And one,
• 16:32cytokine that's been shown to
• 16:33to have some effects if
• 16:35you, for example, express it
• 16:36by AAV
• 16:37in the brain is aisle
• 16:38ten.
• 16:39And but it can't be
• 16:40systemically injected because it's not
• 16:42stable enough to half life
• 16:43is really long. It doesn't
• 16:44get into the brain very
• 16:44well. So he worked with,
• 16:47several of our neurology
• 16:49colleagues and used the EAE
• 16:50model for multiple sclerosis. This
• 16:52is something where you induce
• 16:54an autoimmune response against a
• 16:55myelin protein, and then asked
• 16:57if we dose them with
• 16:59these,
• 16:59suppressor cells, these designer suppressor
• 17:02cells, could we reduce the
• 17:03kind of paralysis that you
• 17:04see in these,
• 17:05neurological like exams?
• 17:07And so what's shown here
• 17:09is that we, in fact,
• 17:10see a significant suppression of
• 17:11this is essentially paralysis in
• 17:13a longer life,
• 17:15survival. So hopefully oh,
• 17:18okay.
• 17:21Well, I can't figure out
• 17:22how to do that.
• 17:24Okay. I'll just actually stick
• 17:25with that. Anyway, you'll you
• 17:26would see that you'll see
• 17:27that the mice, by twelve
• 17:28days that were not treated
• 17:29were really pretty much paralyzed,
• 17:31but the ones that were
• 17:32treated,
• 17:33were were not.
• 17:35So,
• 17:36and then in another, related,
• 17:38intersecting paper, Nish Reddy, a
• 17:39former post student graduate student
• 17:42in the lab,
• 17:43Ashley asked, could we instead
• 17:44of just looking at aisle
• 17:45ten
• 17:52Okay. K.
• 17:54Cool.
• 17:56Could we now kind of
• 17:57make custom programs that have
• 17:58different suppressive cytokines,
• 18:00antibodies, etcetera?
• 18:02And,
• 18:04he then screened these for
• 18:05for how effective they were
• 18:07at suppressing,
• 18:08a t cell killing response.
• 18:10And, this is just summarizing
• 18:12this plot here. Basically, he,
• 18:14in the middle there, he
• 18:15saw that that the best
• 18:16payloads were these specific combinations.
• 18:18They turn out to be
• 18:19things that look that in
• 18:21which a normal Treg would
• 18:22fit. They are a combination
• 18:23of a suppressive cytokine
• 18:25or a suppressor agent, even
• 18:26anti PD one,
• 18:29PDL one. I'm sorry. And
• 18:30then,
• 18:32with a a sync for
• 18:33inflammatory cytokines like a like,
• 18:35CD twenty five, which is
• 18:36a sync for for IL
• 18:38two, the the required,
• 18:40inflammatory cytokine,
• 18:42which also leads to enhanced
• 18:44proliferation of these cells, the
• 18:45suppressor cells. And then he
• 18:47was able to show with,
• 18:48in collaboration
• 18:49with Matthias Heebrock's lab,
• 18:51that we could transplant,
• 18:53beta beta cell,
• 18:55islet, organoids,
• 18:57into mice and that these
• 18:58would be normally killed by,
• 19:00T cells, but that we
• 19:01could protect them for a
• 19:02number of days, with these
• 19:04these enhanced programs. And we're
• 19:05hoping to to,
• 19:07improve these and, improve these
• 19:09and and and apply them
• 19:11towards, neuroinflammation
• 19:12also.
• 19:13So,
• 19:14the,
• 19:16hopefully, I've shown you that
• 19:17we can engineer,
• 19:18immune cells that use multi
• 19:20receptor circuits,
• 19:22to,
• 19:23to integrate information,
• 19:25at different scales and that
• 19:26can make very precise
• 19:28disease specific decisions.
• 19:30In the example of glioblastoma,
• 19:32we've been able to engineer,
• 19:35precision brain cancer therapies in
• 19:36which we program a cell
• 19:38that one recognizes that it's
• 19:39in the brain and two
• 19:40that induces
• 19:41a killing response,
• 19:43locally.
• 19:44And it's a powerful combination
• 19:45of kind of anatomical molecular
• 19:47specificity. And I think that
• 19:49kind of multi scale functionality
• 19:50is really part of the
• 19:51key of what living,
• 19:53systems can do and then
• 19:54and the challenge of how
• 19:55how we understand biological function,
• 19:57of course.
• 19:58And then these tissue sensing
• 19:59cells can be used in
• 20:00a disease agnostic manner to
• 20:02deliver,
• 20:03immune suppressive payloads,
• 20:05for neuroinflammation,
• 20:07as well as potentially
• 20:09regenerative payloads for things like
• 20:10neurodegeneration,
• 20:11and we can create customized
• 20:13multifactor programs.
• 20:14So I wanna just end
• 20:16by giving you some update
• 20:17on some of the the
• 20:18clinical things. We're we're, very
• 20:19excited to try to really
• 20:20push these through to the
• 20:22clinic, as soon as possible.
• 20:25And,
• 20:26we have, one one, phase
• 20:28one trial that we've already
• 20:29opened, which is called eSync.
• 20:30This is actually a synapse
• 20:31of our circuit that is
• 20:33actually primed by a,
• 20:35tumor specific
• 20:37GBM specific neoantigen. So it's
• 20:39absolutely unique. The problem is
• 20:41it's
• 20:42it's very heterogeneous. So if
• 20:43you only attack that, you
• 20:45get escape,
• 20:46because of the heterogeneity. But
• 20:47in this case, we're only
• 20:48using it for to flag
• 20:50the location and then killing
• 20:51more broadly. So that, is,
• 20:53already, down three patients have
• 20:55been dosed. And then this
• 20:56other one, the b sync
• 20:57is the brain priming using
• 20:59BECAN. That one we're gonna
• 21:00file, hopefully, by the end
• 21:02of this year, and start
• 21:03the trial next year. But
• 21:05this is we're really excited
• 21:06by it because,
• 21:07in this case, this is
• 21:08one of the first cases
• 21:09where you're actually using
• 21:10a non tumor antigen to
• 21:13as part of the recognition.
• 21:15And so that means what's
• 21:16exciting is, like, in the
• 21:17first one, we have to
• 21:17screen the patients to find
• 21:19which subpopulation
• 21:20has that neoantigen.
• 21:22But in this case, everyone
• 21:23has BCAN, so everyone can
• 21:25can is there can can
• 21:26be part of this. In
• 21:28addition,
• 21:28these these, killing antigens are
• 21:30found in many different tumors.
• 21:32So this this these look
• 21:34like they're they could work
• 21:35for a lot of pediatric
• 21:36gliomas,
• 21:37many brain cancers, including,
• 21:39brain mets from things like
• 21:41breast and lung, etcetera. So
• 21:43it's really interesting that that,
• 21:44you know, we're we've focused
• 21:46a lot of kind of
• 21:47targeting things to very specific
• 21:48molecular, sort of, individuals and
• 21:51these personalized things. But there
• 21:52is the capability in in
• 21:54this case to kind of
• 21:55cast the net at different
• 21:56levels and then get something
• 21:57that really could be very
• 21:59precise but still used for
• 22:00a large number of patients.
• 22:04And so let me end
• 22:05it's going back to this.
• 22:06We are,
• 22:07we we are very interested
• 22:08in trying to
• 22:10apply AI and and predictive,
• 22:13methods,
• 22:14that allow us to design
• 22:16things. We have been working
• 22:17a lot on we worked
• 22:18with IBM on a number
• 22:19of, sort of modular motifs
• 22:22within CARs and other receptors
• 22:23to try to understand what
• 22:24their phenotypes would be, but
• 22:26we'd really
• 22:28like to be able to,
• 22:29do this and operate at
• 22:30these different scales and have
• 22:31predictions at that level. And
• 22:32part of our,
• 22:33sort of our strategy is
• 22:34to to simplify the the
• 22:36the alphabet of kind of
• 22:37components or words that we
• 22:39use and
• 22:40and that that we understand
• 22:41well and use these in
• 22:42big combinations,
• 22:44generate a lot of data
• 22:45from that, and then,
• 22:47try to, you know, predict
• 22:48what we can build, in
• 22:49that way. So, let me,
• 22:52also just thank, the people
• 22:53from my group and in
• 22:54particular,
• 22:55Milos who led the work
• 22:57on the brain targeting with
• 22:58our colleagues, Sudayo and Scott
• 23:00Zamvil, and then Nish Reddy
• 23:01who, led the work on
• 23:02the synthetic suppressor cells. Alright.
• 23:05Thank you.
• 23:12Thanks, Vandal. That was
• 23:14nominal.
• 23:18Yeah. Wonderful talk, Vandal, as
• 23:19always.
• 23:20What do you think, the
• 23:22knowledge gaps do we need
• 23:23for the AI to tell
• 23:24us which, the synthetic circuit
• 23:26field, that logic that allows
• 23:28you to do this, let
• 23:29us all do this? Yeah.
• 23:31Well,
• 23:31I mean, that's a great
• 23:32question. I'm open to lots
• 23:33of different ideas. I mean,
• 23:35look. I'm I mean, I'm
• 23:36a simple biochemist, so I
• 23:38think about these pieces and
• 23:39kind of how they're put
• 23:40together.
• 23:41You know, how to represent
• 23:42that information at these different
• 23:44scales, I think, is is,
• 23:45you know, something that I'd
• 23:46like to to explore more.
• 23:52Randall. Hi, Risa. Great great
• 23:54thoughts.
• 23:56Are you trying to find
• 23:57similar approach to identify this,
• 24:00glioblastoma
• 24:01specific antigens to find something
• 24:04that you can use on
• 24:06endothelial cells
• 24:08for organ and tissue specific
• 24:10targeting that would not be
• 24:11dependent just on inflammation and
• 24:13when t cells will go
• 24:14there anyway. Yeah. When you
• 24:17use this approach to induce
• 24:18extravasation
• 24:20by detecting. Yeah. Because there
• 24:21there are now,
• 24:23datasets
• 24:24available about
• 24:26organ specific endothelial. Yes. Yeah.
• 24:28So we're very excited about
• 24:29that. I mean, I think
• 24:30that,
• 24:31we came up with this
• 24:32ECM.
• 24:33We're looking into whether there's
• 24:34other tissue specific ECM. The
• 24:37and, yes, there's a lot
• 24:38of endothelial specificity,
• 24:40which is weird to me,
• 24:41but,
• 24:42it seems to be that
• 24:43way. And and we're actually
• 24:44excited because we some we
• 24:45think we have some ways
• 24:46to increase, transmigration,
• 24:51engineered interactions. So I think
• 24:52that could be interesting.
• 24:54And then there's also
• 24:56a lot of, information about,
• 24:58sort of combinations of proteases
• 25:00that are organ specific.
• 25:02So,
• 25:03you know, we're we're interested
• 25:05in looking at those and
• 25:06whether we can sense those.
• 25:07Yeah. And and quick related
• 25:09to that,
• 25:10speaking of VCM, the tenascin
• 25:12c is one of this
• 25:13ECM components. It's It's in
• 25:15brionic, but, in adults, it's
• 25:16mostly in tumors that will
• 25:17be in the main target.
• 25:19Yeah. Well, I mean, yes.
• 25:20A lot of these I
• 25:20mean, in in fibrotic tumors,
• 25:22that's another thing we that
• 25:23overlaps with this. There's a
• 25:25lot of,
• 25:27recognition of those, as a
• 25:28component,
• 25:29for, say, pancreatic, ovarian, etcetera,
• 25:32and fibrosis too.
• 25:33So that's the thing. This
• 25:34is sort of general flavors
• 25:36of things that are normal.
• 25:38Right? And and and but
• 25:39in the wrong combinations, they're
• 25:41disease.