Synthetic Biology: Building Cell Signaling Networks

Top

00:00:07.19 Hi. My name is Wendell Lim. I'm a professor
00:00:09.16 at the University of California San Francisco and an investigator at
00:00:14.25 the Howard Hughes Medical Institute.
00:00:16.28 Today I'd like to tell you about using synthetic biology to build
00:00:18.26 self-signaling networks.
00:00:23.09 One of the most amazing things about living systems
00:00:23.28 is that they are able to monitor their environment and make complex decisions.
00:00:28.12 The example we're showing here, this is a classic film of the human
00:00:33.05 neutrophil, a white blood cell, and you can see
00:00:35.03 it's able to detect and chase that small little black dot
00:00:38.13 which is a bacteria.
00:00:40.11 So somehow it's detecting signals given off by that bacteria
00:00:43.29 and it's using it to coordinate
00:00:44.12 its entire cytoskeleton to give you this coordinated movement
00:00:49.27 and eventually the phagocytosis of that
00:00:50.09 bacterium. And so what we're interested in is trying to understand
00:00:55.20 how is it that cells like this are able to
00:00:58.25 read in many environmental inputs and then make these complex decisions about what they're
00:01:00.24 going to do.
00:01:03.10 What's particularly interesting for us is
00:01:04.18 the idea that these decisions that cells make are
00:01:08.24 coordinated by networks made of molecules.
00:01:11.24 And so we're interested in trying to understand how
00:01:13.22 is that a system of molecules can actually work together
00:01:17.00 to make these sorts of complex decisions.
00:01:19.04 In addition, cells show many, many different types of
00:01:22.16 responses to their environment.
00:01:26.17 And so how is it that so many different regulatory behaviors and
00:01:29.02 signaling behaviors have been able to
00:01:30.20 evolve. How it has been able to use this to get the
00:01:33.06 kinds of diverse cellular behaviors that
00:01:34.28 we see.
00:01:38.29 Traditionally the way that
00:01:40.22 the field of biology has attacked
00:01:41.05 these sorts of problems is to take an
00:01:44.29 interesting system like a cell that does something interesting and then try to dissect it.
00:01:48.13 That is using things like a microscope or other tools
00:01:50.05 to observe, describe and classify what's going on.
00:01:54.27 This could be using a microscope to look at the
00:01:58.27 structures that you see in the cell,
00:01:59.01 using genetics or genomics to look at the DNA sequences
00:02:04.02 that are and genes that are involved in the process or
00:02:06.15 structural methods to look at
00:02:09.22 the structures of these molecules. And network methods to look at how they're arranged in networks
00:02:13.08 as shown down here.
00:02:15.13 But one of the problems that we have in today's post genomic
00:02:18.28 era of biology is the fact that these approaches really help
00:02:22.27 us to take a complex system and take it apart.
00:02:25.04 So that's essentially like taking something like a radio
00:02:28.13 and then disassembling it. And what results is that we
00:02:32.02 now find that we often have a really detailed and complete
00:02:35.17 molecular parts list. But what we lack is we still have a
00:02:40.04 relatively poor understanding of how one can assemble
00:02:42.25 these sorts of parts into functional systems.
00:02:47.13 And so one of the things that
00:02:49.03 we're interested in is trying to complement traditional approaches
00:02:51.08 of taking complex systems and breaking them down into their
00:02:53.23 molecular parts. The traditional deconstruction
00:02:57.01 suppression biology with what we consider a synthetic biology approach.
00:03:04.05 That is the idea of using, trying to understand how
00:03:06.28 simpler molecular parts can be assembled and synthesized
00:03:11.00 to create more complex systems. This is a
00:03:13.22 lot like how an engineer for example, thinks about
00:03:17.19 how to use simple electronic parts to put them together
00:03:20.17 to assemble devices and systems
00:03:22.10 like a computer or, or they
00:03:25.25 can put them together in different ways
00:03:26.23 to make other kinds of devices with different behaviors.
00:03:28.18 So really this is an approach that really is inspired by
00:03:32.23 engineering but really is what we think goes on
00:03:35.03 during the course of evolution.
00:03:36.06 Where pre-existing molecular parts are put together
00:03:39.23 in different ways that might generate new behaviors.
00:03:41.17 And the goal here is that perhaps trying to understand
00:03:47.16 this complimentary synthetic approach of how
00:03:49.04 biological systems are put together
00:03:50.22 could help us not just understand
00:03:52.24 the details of any one system
00:03:55.16 but really more about the logic and rules about how
00:03:58.21 in general cellular systems are constructed.
00:04:04.13 Really this is about asking the
00:04:05.27 inverse question rather than
00:04:07.16 trying to take apart a cell.
00:04:09.08 We're really trying to ask can we view cells for example
00:04:10.26 as complex devices or robots that execute new
00:04:15.18 decision making behaviors and can we learn how to
00:04:17.16 program them much like this robot here which was
00:04:20.11 programmed to actually look for and search for red balls
00:04:22.29 In a field of other blue balls.
00:04:25.19 Ok so what we're interested in is trying to build
00:04:30.01 and rewire new cellular responses.
00:04:33.24 Why do we want to do this. First of all, we really feel that we can learn
00:04:37.02 by building, that by tinkering with these systems
00:04:40.17 and putting them together in different ways and in systematic ways, we can reach a deeper
00:04:43.11 understanding of the molecular logic
00:04:47.09 of how cell signaling systems are put together.
00:04:47.26 The second thing is that if we really understand these rules
00:04:51.20 and these principles of how evolution has
00:04:54.02 used molecular parts to build new kinds of
00:04:57.06 cellular responses and behaviors,
00:04:58.28 then maybe we can actually apply this to build
00:05:02.05 designer cells that have new customized behaviors.
00:05:06.17 This includes the idea that maybe we could
00:05:09.08 build the cells that carry out very
00:05:10.18 sophisticated therapeutic actions, like
00:05:12.08 being able to seek out and kill cancer cells.
00:05:18.01 More broadly this really represents a kind of
00:05:20.28 a new way of trying to think about biology
00:05:22.07 mostly and historically biology has focused on
00:05:24.27 trying to study the biological systems that
00:05:27.09 exist and in a way synthetic biology has
00:05:31.02 given us the capability of not just
00:05:32.29 studying what exists but making
00:05:34.23 different variants of these and asking
00:05:36.17 what can exist or what cannot exist
00:05:38.18 and trying to understand the boundaries
00:05:41.10 between these.
00:05:42.12 Even though this area of sort of
00:05:47.14 synthetic systems, biological like systems may seem
00:05:48.13 like they're not truly biological, what I would argue
00:05:52.04 is that this systems still have to obey
00:05:54.13 the same molecular and physical laws
00:05:56.29 as evolved systems. And so
00:06:00.03 really trying to understand the
00:06:02.00 entirety of these really can perhaps
00:06:04.17 help us to understand those principles better and
00:06:06.07 in addition it might very well be
00:06:09.04 that within the sphere of what can exist
00:06:11.27 there may be cells that show
00:06:14.13 very useful behaviors that we can harness.
00:06:16.12 So how do we go about trying to rewire
00:06:21.02 cellular signaling systems.
00:06:24.08 One of the key principles that I want to raise
00:06:26.04 here is the principle of modularity.
00:06:27.18 So back in the 80s and 90s
00:06:29.03 when we were really starting to sequence
00:06:31.28 all of these genomes,
00:06:33.21 one of the really striking findings was that many signaling proteins
00:06:35.21 were built from simpler parts.
00:06:37.23 That is if you look at a lot of the
00:06:40.00 molecules involved in signaling we found that
00:06:42.01 they had a lot of modular
00:06:44.00 domains that were easily identifiable
00:06:45.27 by sequence.
00:06:47.12 And moreover that many of these domains
00:06:49.18 were found repeated in different proteins
00:06:51.10 and in different combinations.
00:06:53.11 These domains really come in two flavors
00:06:56.19 which I'll talk about in a second in more detail.
00:06:58.18 But one is catalytic domains
00:07:01.18 things like kianases that carry out
00:07:03.07 catalytic information and post
00:07:05.01 translational modifications like phosphorylation.
00:07:07.13 as well as interaction domains
00:07:09.02 that are involved in binding and recognition.
00:07:11.06 But the basic hypothesis that emerged
00:07:13.13 at this time was that perhaps
00:07:14.25 nature has evolved complex
00:07:16.26 regulatory circuits by using these modules
00:07:21.08 as kind of a tool kit to put together
00:07:24.19 different types of devices.
00:07:26.24 So let me tell you about the different types
00:07:27.17 of domains that we see.
00:07:29.06 As I said one class are catalytic
00:07:31.07 modules that really are involved
00:07:32.20 in directly transmitting information.
00:07:34.23 So a good example of this are the
00:07:37.05 enzymes that are involved in phosphorylation.
00:07:39.11 Kianases and phosphatases for example.
00:07:42.26 So kianases will put on
00:07:44.13 a phosphate onto a target protein that
00:07:47.03 shown in green here and that
00:07:48.19 phosphate might change the
00:07:50.28 conformation or the activity of
00:07:52.20 that protein leading to some sort of downstream
00:07:54.24 effect. Now that
00:07:55.29 phosphate can be taken off by of course
00:07:59.27 corresponding complementary eraser
00:08:00.05 catalytic function, in this case a
00:08:02.00 phosphatase. In addition to this
00:08:05.00 kind of writer and eraser catalytic activity
00:08:07.13 oftentimes there will also be a reader domains
00:08:09.25 that can recognize and bind to say the
00:08:12.22 phosphorylated moiety.
00:08:14.20 ok so you have this ensemble of
00:08:16.19 readers, writers and erasers.
00:08:20.01 Another example of that that's exactly
00:08:21.28 parallel are the enzymes that
00:08:26.19 regulate GTPases.
00:08:27.10 So GTPases can exist in two forms
00:08:29.04 the GDP bound state and
00:08:31.09 then the GTP bound state which is
00:08:33.06 really the active one.
00:08:35.02 The transitions between these states
00:08:36.26 are mediated by another set
00:08:38.17 of writer and eraser catalytic activities.
00:08:41.26 Gaps are a quantonucleotided
00:08:45.27 change factors lead to putting
00:08:47.13 on and activating the GTPase
00:08:49.24 protein whereas gap proteins are GTPase
00:08:52.14 activated protein lead to the hydrolysis
00:08:54.03 of GTP and turn it off
00:08:57.15 the system. And once again there are
00:08:58.04 also often times reader
00:08:59.25 domains that can recognize
00:09:01.21 the specific activated form of the
00:09:04.07 GTP. So these sorts of writer
00:09:08.20 and eraser enzymes are
00:09:09.21 often put together. You can get
00:09:11.11 cascades or other network linkages where
00:09:14.07 for example one phospholation event will activate
00:09:15.23 another enzyme
00:09:17.28 that will in turn activate another enzyme
00:09:19.14 leading to the sorts of networks that
00:09:21.29 and circuits that we see in cells.
00:09:24.25 Now the second important class of modules that we see
00:09:28.26 in signaling domains in signaling protein
00:09:30.29 are interaction modules that really
00:09:32.24 play a more crucial role in directing
00:09:35.15 and controlling the flow
00:09:36.00 of information. so these catalytic
00:09:38.25 domains that I talked about are often
00:09:40.08 combined with recognition domains,
00:09:43.12 many of them are protein
00:09:44.18 interaction domains.
00:09:46.14 Like this SH 3 domain that
00:09:48.21 recognizes a cognate linear
00:09:50.07 peptide motif, in this case a
00:09:51.23 proline enrichment type. there are many other sorts
00:09:54.01 of domains that recognize different
00:09:55.24 distinct types of linear motifs.
00:09:57.16 In addition there are also
00:09:59.28 recognition or interaction domains that
00:10:02.12 bind to specific lipid species
00:10:04.11 at the plasma membrane.
00:10:05.27 So these are combined with the catalytic
00:10:08.23 domain control where these
00:10:10.16 catalytic domains go and who
00:10:12.12 they interact with.
00:10:13.26 So how is it that catalytic and
00:10:16.08 interaction domains can be put together
00:10:18.03 to yield the diverse circuitry that we
00:10:20.03 see. This slide shows
00:10:22.07 two very simple but very prevalent
00:10:24.19 mechanisms by which these domains can be
00:10:27.00 put together to give you
00:10:28.16 complex circuits. One is
00:10:30.00 really based on the simple principle of
00:10:31.29 recruitment assembly, that
00:10:33.11 the cell is full of these different types of
00:10:35.24 catalytic modules. and really a key question
00:10:40.09 is how do they efficiently interact with the
00:10:42.10 right ones. And so what you'll find
00:10:43.19 is often let's say protein X
00:10:45.20 is a kianase and Y is its
00:10:48.18 substrate. Then that, that
00:10:49.05 they might have a modular interaction domain
00:10:52.17 and recognition ligand fused to these partners
00:10:56.20 which allows protein X to specifically
00:10:58.27 interact with and modify protein
00:11:01.15 Y. The importance of these sorts of
00:11:05.08 recruitment and assembly interactions
00:11:07.12 is really exemplified at the
00:11:09.21 extreme by a class of proteins known as
00:11:11.24 scaffold proteins. These are proteins that have
00:11:14.02 multiple interaction domains and they
00:11:16.04 often combine to multiple signaling proteins
00:11:17.29 like these shown here, X, Y and Z.
00:11:19.21 They really help to coordinate
00:11:22.00 a set of catalytic components to be at the
00:11:24.16 right place in the right time, allowing
00:11:26.02 them to interact in a very specific
00:11:27.29 cascade. A second way that these modules
00:11:32.08 can be used to lead to
00:11:35.06 interesting behavior, is the construction of
00:11:37.20 allosteric signaling switches using modular auto
00:11:39.19 inhibition but often times you'll find a catalytic
00:11:43.08 domain say like a kianase that has
00:11:45.02 other interaction domains within it.
00:11:46.29 But often times in the basal state
00:11:49.23 these interaction events will actually auto
00:11:51.29 inhibit or interact with
00:11:53.00 the catalytic domain itself or with other
00:11:55.01 parts of the protein.
00:11:56.21 and that can lead to inhibition either through
00:11:59.27 directly sterically blocking
00:12:02.07 the active site of the catalytic function or by
00:12:06.27 conformationally contorting it so that it's inactive.
00:12:09.05 and if you get this sort of situation then oftentimes
00:12:11.24 there are cases where
00:12:14.00 a competing ligand that this disrupts
00:12:15.18 this auto inhibitory interaction can then result
00:12:18.23 in the allosteric activation of
00:12:21.09 that catalytic function. in this case for
00:12:23.03 example activation of a kianase.
00:12:25.12 So we see these sorts of mechanisms over and over
00:12:31.10 again in many different signaling projects
00:12:33.16 with different types of catalytic interaction
00:12:34.19 domains. but it really lends itself to the possibility
00:12:37.21 that these modules and motifs
00:12:39.17 really might form a toolkit or a set of
00:12:43.09 building blocks or a cellular
00:12:44.27 language that we can use to create
00:12:47.17 new functions and that evolution
00:12:49.29 seems to have used this. we see this
00:12:52.03 of the different combinations of domains that we see
00:12:54.03 in evolved proteins.
00:12:56.04 But also another piece of evidence comes from
00:12:57.25 disease. so in diseases
00:12:59.15 like cancer or in viral or other pathogenic diseases
00:13:02.17 there's a lot of evidence that these sorts of
00:13:04.10 modules have been hijacked and used
00:13:06.11 to redirect catalytic functions to new
00:13:08.29 targets and other things like that.
00:13:11.15 And it's in other words really generating
00:13:13.29 signaling behaviors that actually are associated with
00:13:15.15 a pathology or a disease.
00:13:17.14 So a corollary of this is that
00:13:20.18 if this is really true, then
00:13:21.27 it argues that we should be able to
00:13:23.24 use these modular words to
00:13:26.10 say new things, that perhaps we can
00:13:28.25 actually use them to engineer
00:13:33.03 ourselves in a somewhat systematic and logical way.
00:13:33.06 So if we can do this, what is it that we want to say
00:13:37.15 and how do we say it.
00:13:39.03 So about ten years ago my group started
00:13:44.08 to try to explore this question.
00:13:46.01 and ask really could we start building
00:13:49.03 new types of signaling
00:13:49.16 proteins as well as signaling networks
00:13:52.08 and cellular behaviors.
00:13:53.26 so one of the first things that we did was actually
00:13:56.16 ask could we build synthetic signaling switch
00:13:59.02 proteins and we took a group of
00:14:02.02 different catalytic functions
00:14:03.20 and then a group of different interaction
00:14:06.02 domains and then made a
00:14:07.26 random combinatory libraries in which we
00:14:10.23 combined interaction domains with catalytic
00:14:12.18 domains as well as the
00:14:14.19 inter molecular ligands of these
00:14:15.24 domains with the idea that
00:14:17.15 we wanted to find out. Would any
00:14:19.07 of these show inhibition and
00:14:21.21 in a way that would allow them to now
00:14:24.02 if we added competing ligands
00:14:26.03 become activated and would they show
00:14:27.28 interesting behaviors, regulatory behaviors
00:14:30.03 like natural signaling switches.
00:14:32.17 And in fact what we found is that
00:14:34.14 when we did this a fairly large proportion
00:14:37.09 probably about 30% of them actually showed
00:14:39.04 some form of auto inhibition and regulation
00:14:42.22 and moreover when we looked at these in detail
00:14:44.23 and looked at how the types of behaviors of
00:14:47.07 integration between multiple inputs
00:14:48.23 we saw a lot of different types of
00:14:50.16 bulian behaviors, really we saw the
00:14:53.22 diversity that mimicked the types of
00:14:55.28 regulation that we usually see in
00:14:58.20 natural signaling proteins. Arguing we think that this is, this type of
00:15:01.14 domain shuffling really forms a
00:15:03.04 way to get an evolutionary starting point
00:15:05.08 for building new types of signaling
00:15:07.00 switches.
00:15:10.11 Another example of
00:15:10.21 looking at the modularity of signaling proteins
00:15:13.03 comes from looking at
00:15:14.05 signaling behaviors in the model cell
00:15:16.19 system of yeast.
00:15:18.08 So yeast is a single celled organism that has
00:15:20.11 to make many decisions based on
00:15:22.20 what is in its environment.
00:15:24.23 So one of the things that yeasts do is they actually
00:15:26.16 have two cell types and they can mate with
00:15:28.16 each other but to induce that mating
00:15:30.17 response, they have to detect a
00:15:31.24 pheromone that comes from
00:15:33.17 their partner and so there's a pathway
00:15:36.19 called the mating response that actually detects
00:15:36.27 a mating pheromone.
00:15:39.16 In this pathway shown in red, and then that
00:15:41.08 leads to the activation of a set of mating
00:15:43.17 response genes.
00:15:45.06 Now a completely different pathway that
00:15:47.07 yeasts have is a pathway that deals with
00:15:51.00 hyperosmotic stress. So if you
00:15:52.22 put yeasts into high salt environment
00:15:55.14 it detects that
00:15:57.20 change and then expresses a bunch of
00:15:59.16 genes that help protect it against that
00:16:01.28 osmolarity.
00:16:03.12 So although these pathways are completely different, what
00:16:06.28 does link them is that they are both
00:16:08.18 mediated by a set of kianases known as
00:16:11.10 map kianases.
00:16:12.22 The problem is that the cell is filled with
00:16:17.13 many different types of kianases
00:16:17.22 and quite a few of these mapped kianases
00:16:20.08 that are very closely related.
00:16:23.07 And so the question is
00:16:24.25 if these two pathways which are physiologically
00:16:27.00 distinct use shared or similar
00:16:29.10 components, how is that they actually
00:16:31.13 pass the information, transmit it
00:16:33.22 in the right way and you don't get kind of
00:16:37.00 cross talk so you get the wrong response.
00:16:37.14 And so one of the things that was discovered
00:16:39.26 around this time was that there existed
00:16:42.03 these scaffold proteins. For example,
00:16:44.02 there's a scaffold protein that has
00:16:45.20 multiple interactions remains that
00:16:48.06 binds to the mating receptor as well as the set of
00:16:50.23 kianases that are involved in the mating
00:16:52.27 response. They essentially act
00:16:54.16 almost as a bread board to
00:16:56.08 wire together the circuits so that you get
00:16:57.26 specific transmission of information from
00:17:00.04 the mating receptor to
00:17:02.09 the mating genes.
00:17:04.10 Correspondingly, there is also
00:17:07.07 a scaffold protein associated with the
00:17:09.14 osmotic response pathway which again
00:17:11.24 binds to that receptor and then also
00:17:14.01 assembles a distinct set of kianases
00:17:16.13 that are involved in transmitting the
00:17:18.27 information to the osmotic response genes.
00:17:21.27 So in other words both of these pathways
00:17:23.25 can coexist and be specific
00:17:25.20 because there is a scaffold protein that
00:17:28.15 essentially wires them together
00:17:30.22 in a specific way.
00:17:32.10 And so one of the hypotheses that came
00:17:34.08 from this was that perhaps then
00:17:36.13 evolution can create new pathways
00:17:39.10 and by wiring kianses and other catalytic
00:17:42.11 functions together in different ways
00:17:43.29 by generating new chimeric scaffold
00:17:46.19 proteins. So Sanhing Park a post doc in the lab
00:17:49.05 at the time decided to try to test this by
00:17:51.23 asking could we make a chimeric scaffold
00:17:53.24 protein that would assemble different
00:17:56.10 kianses in these two
00:17:58.07 distinct pathways. And what he
00:17:59.20 found is that when he did this and made
00:18:01.09 the right sort of combination of pathways
00:18:03.09 of components, he was able actually to
00:18:05.15 make a scaffold that would
00:18:07.09 when put into a living cell would
00:18:09.27 now cause the cell to induce
00:18:12.01 the osmolarity response when
00:18:14.02 the cell was stimulated by the mating
00:18:16.07 pheromone. So we could actually create a
00:18:18.01 new input output pathway
00:18:19.24 really mimicking evolution. In this case we
00:18:22.10 had a pathway where the cells would only survive
00:18:24.12 at high osmotic stress when we stimulated them
00:18:27.01 with mating pheromones. So really supporting
00:18:29.27 the idea that without
00:18:31.13 changing actually the catalytic component
00:18:33.14 and really only changing the factors that were
00:18:35.26 assembling them, we could create
00:18:37.20 new types of input output
00:18:39.19 wiring and behavior in the cell.
00:18:41.18 Here's another example of rewiring
00:18:46.10 that we have been able to do. So
00:18:48.01 Ensom Levskaya recognized that in plants
00:18:52.29 there are these really interesting protein
00:18:54.05 interaction domains that where the
00:18:56.29 interaction is regulated by light.
00:18:59.19 So if you shine a certain wavelength of light
00:19:01.03 750 nanometers on this
00:19:02.29 protein it will change conformations
00:19:04.28 and now tightly bind to its partner.
00:19:08.10 And so what we realize is that again many
00:19:11.01 signaling behaviors in mammalian cells
00:19:14.04 are controlled by induced recruitment.
00:19:17.01 So we asked could we actually replace many
00:19:19.17 of those interactions. oftentimes a catalytic
00:19:22.00 domain will be recruited to an activated receptor
00:19:24.00 but could we replace that receptor now
00:19:25.29 with this light responsive protein from
00:19:28.09 plants and recruit a catalytic domain.
00:19:31.11 For example, this GET to the plasma membrane
00:19:34.06 where it could act on its target GTPase and signaling
00:19:37.27 output. could we actually control the
00:19:39.23 cell now with light.
00:19:41.16 And so this is an example of
00:19:45.17 linking this light activated protein
00:19:47.24 to a Rac GET so that's going to activate
00:19:51.00 the protein Rac GTPase which itself
00:19:55.00 activates actin polymerization.
00:19:57.09 As you see here, what happens is that
00:19:59.00 when we shine the light now on
00:20:01.10 which is the red circle, on
00:20:03.11 the cell, you see recruitment of that Rac GET
00:20:06.09 and actually protrusion of the cell at that
00:20:08.21 point so we can actually make the cell
00:20:10.08 follow the light by putting this
00:20:13.00 novel chimeric protein that combines
00:20:16.19 light controlled protein interaction with this
00:20:19.05 catalytic activity.
00:20:20.20 So what these experiments hopefully illustrate
00:20:24.28 is that we really can use these sorts of signal
00:20:29.06 modules to flexibly rewire and reprogram
00:20:32.18 so they're signaling proteins as well
00:20:34.28 as the networks and to generate
00:20:37.29 novel cellular behavior.
00:20:40.10 So one of the things that we become more interested
00:20:42.13 in is asking the second question
00:20:44.02 about applications. Can we
00:20:46.00 actually start harnessing this capability
00:20:47.26 to rewire cells to, and put it to
00:20:49.28 use. And one of the most exciting
00:20:51.28 things for us is the idea of using this
00:20:54.03 in therapeutics, in medicine.
00:20:56.21 that mostly today when we think about medicine
00:20:59.28 and therapeutics, we think about small molecules
00:21:03.25 or biologics, which are very powerful and
00:21:06.07 wonderful agents. But one of the things that limits them
00:21:10.03 is they largely if you put them into the body
00:21:13.07 they act in a systemic way.
00:21:14.25 And then also they
00:21:16.08 largely can just either block or activate something
00:21:18.15 and they can't actually make very
00:21:20.08 complicated decisions.
00:21:21.24 And that contrast with for example the cell, so a
00:21:24.29 cell of the immune system for example which
00:21:26.15 also essentially is a therapeutic. it fights
00:21:28.22 disease, is able to move
00:21:30.17 through the body and find where
00:21:32.26 there is a problem. so it can
00:21:34.10 be spatially targeted.
00:21:36.03 It can also integrate sources of information
00:21:38.04 from that environment and make
00:21:40.00 complex decisions.
00:21:41.12 But it can also execute complex actions
00:21:43.08 like this kind of migration or
00:21:45.09 phagracytosis that we see
00:21:46.23 with the neutrophil.
00:21:49.08 And so one of the visions that we have is if
00:21:51.25 we really understood these principles of how
00:21:54.11 we can rewire self-signaling behaviors
00:21:57.07 then perhaps we could actually start to
00:21:59.15 customize cells so that they can actually
00:22:01.11 carry out functions that are
00:22:03.06 therapeutic. This would entail
00:22:06.08 for example trying to change what sensors
00:22:08.14 the cell has so that it can
00:22:09.29 detect either disease or user
00:22:12.13 provided inputs that we want
00:22:14.10 to also be able to change the
00:22:16.00 self-signaling networks to control
00:22:18.25 what's, how this information is processed.
00:22:20.26 And how decisions are made. And then link
00:22:23.07 these to the many interesting and powerful
00:22:25.19 payloads or actions that cells are capable
00:22:28.11 of to really yield
00:22:29.29 now what we consider a much smarter
00:22:31.28 therapeutic response.
00:22:35.16 So one of the great places
00:22:39.28 test beds to really explore this
00:22:42.10 today is in the area of adoptive
00:22:44.09 immunotherapy so immune cells, particularly
00:22:46.23 a type of immune cell known as
00:22:49.13 key lymphosites or T cells
00:22:50.18 are an ideal test bed for
00:22:52.22 this kind of therapeutic engineering.
00:22:54.17 These are cells that are able to
00:22:57.05 for example recognize and kill
00:22:59.19 a cell that's infected by a virus. so they're a
00:23:01.24 very powerful activities.
00:23:03.06 More importantly we know that we can actually
00:23:05.23 remove T cells from a patient.
00:23:07.21 We can genetically modify them ex vivo.
00:23:09.21 We can expand them ex vivo.
00:23:11.02 And then we can transfer them back into the
00:23:13.13 patient. So this is pretty well established.
00:23:15.19 And so we can do a lot of different things that
00:23:19.02 don't have to do with trying to modify things
00:23:22.03 In situ.
00:23:23.17 In addition, T cells can
00:23:27.18 be used for many potential targets.
00:23:30.02 They could be used to recognize cancer.
00:23:32.03 They could be used to recognize chronic
00:23:34.16 infection. And also they could be used to
00:23:36.25 combat auto immunity.
00:23:38.20 So how could we for example
00:23:42.00 redirect T cells to recognize cancer.
00:23:44.08 So if you look at a native T cell , the key receptor, a
00:23:48.18 molecule that's used to mediate its action
00:23:50.26 is the T cell receptor. So
00:23:52.28 if you look closely at that, that has a
00:23:54.16 extra cellular domain and then key intra cellular
00:23:56.06 signaling motifs. And so
00:23:58.20 for example when the cognate peptide
00:24:01.00 antigen is bound by that extra cellular domain
00:24:04.17 it will lead to phosphoralation of
00:24:06.21 these intra cellular motifs which leads
00:24:08.15 to the recruitment of various downstream
00:24:11.19 signaling proteins that then in turn
00:24:12.13 lead to the activation
00:24:14.11 of that T cell. And for example
00:24:16.06 killing of the target cell.
00:24:18.15 What was shown in the 80s
00:24:21.14 by a number of people including Zelig Esshar was
00:24:24.02 the idea that you could actually make synthetic
00:24:26.18 or chimeric T cell receptors so the
00:24:29.00 idea here was could you take
00:24:30.24 an antibody that recognized
00:24:33.02 a antigen that was only expressed on the surface
00:24:36.01 of a tumor cell. And
00:24:37.23 then remarkably they found that if they took
00:24:40.03 this and then added the intra cellular
00:24:42.18 portion of the T cell receptor
00:24:44.08 that this would actually get activated
00:24:46.16 when it recognized the cell with this
00:24:48.29 antigen and lead to T cell activation.
00:24:51.05 So you could actually make
00:24:53.27 these kinds of synthetic receptors
00:24:55.26 in this case known as chimeric antigen receptors
00:24:58.03 or CARs that are built of
00:25:00.10 different extra cellular recognition domains
00:25:03.16 things that recognize tumor antigens
00:25:05.17 But then link together different sorts of signaling
00:25:08.07 inter cellular signaling device that control
00:25:10.22 the activation of the T cell.
00:25:13.00 And now 20 years later this process
00:25:16.16 is refined somewhat and they're really quite powerful
00:25:18.10 versions of the chimeric antigen receptors
00:25:21.28 that can recognize tumor antigens
00:25:23.16 and stimulate T cell activation so we have these
00:25:26.18 modular synthetic sensors
00:25:29.07 that can recognize disease antigens and redirect
00:25:32.20 T cell activity towards these cells.
00:25:37.14 So the idea is that we can take these CAR modified T cells
00:25:41.09 put them back into the patient and they should be
00:25:43.01 able to move around the body and
00:25:45.11 seek out those cells that have
00:25:47.14 the cognate antigen and then kill them.
00:25:49.03 So although again this seems like science
00:25:51.21 fiction what we now know in
00:25:53.11 today's world, over the last several years
00:25:54.27 is that we actually can use this to
00:25:57.17 attack certain B cell cancers.
00:25:59.17 this has been remarkably successful
00:26:02.19 in a number of studies shown here.
00:26:02.26 And that there have been quite a few
00:26:05.14 patients that have had a form of B cell cancer
00:26:08.19 that is non responsive to chemotherapy, that
00:26:11.03 have been cured by using this
00:26:14.16 sort of approach with modified engineered T cells.
00:26:17.23 So this is tremendously exciting but
00:26:22.15 there are still problems with this particularly
00:26:24.24 what we have shown is that we can take
00:26:27.12 T cells, these powerful cells and redirect
00:26:29.11 them. But often times there are many
00:26:31.15 strong side effects because these
00:26:33.12 cells are very strong in terms of their immuno
00:26:35.17 response. They secrete a lot of
00:26:37.18 molecules called cytokines and
00:26:42.13 really leads to a big systemic immune response. So although
00:26:44.11 we can redirect them, right now
00:26:46.17 oftentimes they are too strong or
00:26:48.17 poorly controlled. So we really
00:26:49.25 need to learn not only how to redirect
00:26:52.00 this beast, but also how to tame the beast.
00:26:54.22 And so this is where we have
00:26:57.01 been trying to use synthetic biology
00:26:59.02 to generate engineered T cells that show much more
00:27:01.21 controlled behavior. And so one example
00:27:04.00 is the idea that could we actually make
00:27:06.06 T cells that are remote controlled.
00:27:08.28 T cells that only
00:27:11.01 switch on when for example a physician
00:27:14.01 provides a drug.
00:27:15.03 We'd still want these cells to recognize
00:27:17.15 the tumor cells but we want
00:27:19.03 the secondary switch to be layered over this.
00:27:21.04 That allows the user or the physician to
00:27:23.11 control when and to what extent they are
00:27:25.08 activated.
00:27:28.11 And so Chia Wu, a post doc in the lab,
00:27:30.01 has been doing studies at trying to build
00:27:32.07 switchable chimeric antigen receptors.
00:27:34.23 So this is the conventional chimeric antigen receptor
00:27:37.21 that of course has an extra cellular
00:27:39.27 antibody that recognizes the tumor antigen
00:27:42.22 and also has these intracellular signaling motifs.
00:27:45.06 What we decided to do was to ask
00:27:47.07 could we actually build an additional switch
00:27:50.04 into this receptor by creating a
00:27:52.18 split version of it. so
00:27:54.06 could we take the recognition component
00:27:56.05 but then take off some of the key
00:27:58.12 signaling modules and put those in a separate
00:27:59.28 molecule but then bring
00:28:01.25 them together using modules that
00:28:04.08 only heterodimerize when we add a drug.
00:28:08.10 Ok so the idea here is that the T cell
00:28:12.07 would only get activated when
00:28:15.05 it had this combination of both recognizing the cognitive antigen
00:28:18.20 as well as having the drug provided
00:28:20.11 by the user.
00:28:22.07 So fortunately we have been able to construct
00:28:25.23 a number of different versions of this
00:28:27.11 that work and one of them is shown here.
00:28:29.27 And so these are some actual assays
00:28:32.02 based on these cells so
00:28:33.18 what's plotted here is T cell
00:28:36.10 activation monitored by release of a cytokine
00:28:38.17 called IL 2.
00:28:40.03 And so what you see is when we look at the
00:28:41.25 conventional CAR cells,
00:28:42.22 CAR T cells that if we stimulate with the
00:28:47.16 cognate antigen which is shown in green
00:28:49.11 that the cells get activated.
00:28:51.16 And they actually don't care whether or not the
00:28:53.21 drug is present or not, which is indicated by the
00:28:55.20 orange square.
00:28:56.20 In contrast when we move to this on switch CAR
00:28:59.07 T cells bearing this on switch CAR we see
00:29:03.07 that they only get activated when there's a
00:29:04.12 combination of both the cognate
00:29:05.28 antigen in green as well as
00:29:07.25 the drug in orange. So
00:29:09.18 here is a case where the physician could actually
00:29:11.21 control when the cell is activated
00:29:13.22 as well as we found the timing
00:29:16.05 and the dosage or potency of that
00:29:19.00 response.
00:29:20.07 What's shown here now is time
00:29:23.00 lapse microscopy of
00:29:24.11 these switchable CAR T cells
00:29:26.18 attacking the tumor cells so
00:29:28.09 what you see here in the first movie is
00:29:30.24 the cells, the T cells are shown
00:29:32.22 in color and this is one recognizing
00:29:34.29 a cognate, a tumor cell with the
00:29:37.21 correct antigen. But this is
00:29:38.23 in the case without drug
00:29:40.15 and so you see that they interact but over the
00:29:43.10 course of this 30 minute movie
00:29:45.07 they don't actually kill that
00:29:47.27 target cell.
00:29:50.19 In contrast we now take the same cells and mix them with cancer
00:29:53.23 cells and then add the drug
00:29:55.07 what we see is this.
00:29:55.24 And we see the cells now engaged with
00:29:59.09 the target cells.
00:29:59.23 But then immediately you see this cell
00:30:01.03 bleding and then this blue dye is indicative
00:30:03.21 Of a cell undergoing afactosis.
00:30:06.02 So that cell immediately starts dying
00:30:07.10 and once that cell is on its way to death
00:30:09.21 this T cell reorients contacts the second cell,
00:30:14.02 reorients its receptors to that.
00:30:16.00 forming a different synapse and then
00:30:17.21 actually starts killing that. you see the blething
00:30:19.13 as well as the uptake of the
00:30:21.08 blue dye. So
00:30:22.24 within this 30 minute period where the
00:30:25.20 in the absence of drug the cell doesn't
00:30:28.00 do anything, doesn't kill anything
00:30:29.20 we see this, in the same time
00:30:32.04 when the drug is added they are able to
00:30:35.10 potently and serially
00:30:35.16 kill multiple target
00:30:37.17 cancer cells.
00:30:39.04 So we think this is tremendously exciting.
00:30:43.22 We think that in today's world, now that
00:30:47.03 we have the capabilities of engineering
00:30:49.03 and modifying cells and we understand much more
00:30:51.29 about the principles of how
00:30:53.03 Signaling networks are put together
00:30:55.29 that we're on the verge of really
00:30:57.22 exciting applications where we can
00:31:00.22 use this toolkit of modules
00:31:03.11 to engineer different types of cells that
00:31:05.20 can sense disease signals and
00:31:07.27 environmental or niche signals in the
00:31:09.22 body as well as user signals
00:31:11.14 and integrate them to give you customized
00:31:13.11 therapeutic action.
00:31:15.18 And so we think this is a really exciting period
00:31:17.17 for applying synthetic biology
00:31:19.16 to these sorts of mammalian cells, to immune cells
00:31:22.13 to create exciting and hopefully
00:31:25.07 very useful therapeutic functions.
00:31:27.29 And so that's the end of this lecture. I'd like to
00:31:31.18 thank all of the people from my lab who did
00:31:32.14 all of this work.
00:31:33.24 a very creative and inspiring group of
00:31:35.19 people as well as all of our
00:31:36.28 collaborators and colleagues.
00:31:38.14 But I’d like to thank you for watching
00:31:40.02 this video.
00:31:41.05 end

This Talk

Speaker: Wendell Lim

Audience:

Researcher

Recorded: August 2015

Talk Overview

Dr. Wendell Lim explains that many signaling proteins are built from simple modules arranged in different ways. Some modules are catalytic and transmit information (e.g., kinase or phosphatase domains), while others are interaction modules that regulate information flow (e.g., protein-protein interaction domains). By rearranging these modules, Lim’s lab has reprogrammed signaling pathways to generate novel cell behavior. They are now working to use these techniques to develop cell-based therapies.

Speaker Bio

Wendell Lim

Wendell Lim is a Professor in the Department of Cellular and Molecular Pharmacology and Director of the Center for Systems and Synthetic Biology at the University of California, San Francisco. He is also a Howard Hughes Medical Institute Investigator. George Cachianes has been teaching biotechnology at Abraham Lincoln High School in San Francisco for over… Continue Reading