 I'd like to thank the organizers for the invitation to be here. So, Anne's talk was focusing on the way genes and substrates and catalysts interact within cells, the way they create networks, the way you get the flow of carbon from one place to the other in plant cells. What I want to talk about is the next level up in the scale is cells themselves and how the different arrangements, the properties of cells can be used to create different forms and how we might be able to engineer things. Now, with plants we have a kind of close relationship with plants on a day-to-day basis and it's usually in the garden or in the supermarket and here we've got a picture of a range of vegetables, of course, cauliflower, broccoli, cabbages, etc. And you might be forgiven for seeing this picture and thinking, well, they have quite different plants, but rather these are all the same species. These are species or varieties that have been derived by breeding and selection from a common wild species. So, this is a wild mustard plant and through many millennia, often of breeding, all the common crop plants that humans use have been derived from natural antecedents by virtue of continual selection and breeding to create very different outcomes. And these plants are all derived from these same ancestral plant where you've got, say, floral meristems or vegetative meristems creating different arrangements of cells in the context of a growing system to create very different outcomes. And I show this picture to emphasize the plasticity. So, even as you see in metabolism, the different flow of genetically controlled flow of carbon into different substrates, at the cellular level, you see the rearrangement of cells during development, during growth to create different outcomes at the end point of these developmental processes. And I want to show this image from the BBC Natural History Unit, which shows here this is a leaf that's growing. And as you see in this accelerated time-lapse image, you've got this leaf and this stolen or stem-like arrangement, which is growing from the tip of the leaf. And as it grows, you'll see this small nub of tissue here which itself will continue to grow and expand. And I use this as an example to illustrate the dynamics of growth. This, of course, is a plant structure which is emerging, growing and creating this large structure. As you see it growing, you can see it forming the picture of a picture plant. And I think partly without human perception is not good at understanding or conceptualizing what happens at the scale of plant growth because plant growth is a little bit slow for us. So we're used to morphogenesis in or movement in animal systems. But in plants, of course, we take them for granted because they don't do this very quickly. But when you start to see things accelerated, crammed together, you can see the dynamics in a very concrete way. And I think it illustrates, for me anyway, the processes that are going on inside this process where there are literally millions of cells initiating essentially from a single cell to create this structure which is a very defined and constrained structure. And, of course, this process of morphogenesis is driven by growth and it's driven by the growth and dynamics of individual cells interacting as a population or society. And so at the root of this process is this in cartoon form. So we have a plant cell here which has grown and is about to divide to create two daughter cells. And I use this cartoon to illustrate a couple of the salient features of plant growth which make it different to animals and help explain why plants look different from animal systems. So plant cells are immobilized. So just like the macroscopic plant which is rooted in the ground, plant cells are immobilized. They have this extra cell in the matrix which locks them in place with their neighboring cells. So you have a single cell here which is attached to its neighbors. As it divides, it will form two daughter cells by essentially subdivision. So the new wall that's formed is actually built inside the existing cell to create this new cell wall structure. And these two daughter cells and all of their descendants will be locked together for the life of the plant. So this boundary here, these cells can continue to expand by division, but this boundary marks a clonal boundary which will not shift because these plants can't move. The plant cells can't move with respect to each other. So this also brings another consequence which is that because of the immobility of these cells, there's a relationship between the properties of cells and the local cellular anatomy. And if you imagine going back to that picture plant which starts at a single cell and grows and you've got millions of the cells all coordinating their behavior to create this final terminal structure. And it's composed of these individual elements in the population, individual cells which are of a particular type at a particular time. And we know that in the case of plants, plant fate or gene expression is largely controlled by interactions. It's the neighboring interactions which tell plant cells what they're to be and in return those cells will communicate with their neighbors and it's this social interaction which and the passage of genetic information between cells in a local fashion which controls gene expression. And so there's this network of exchange of information which takes place between cells within that growing picture plant. And it's that network, that population, that social interaction which describes the properties of that final structure. So here we've got a cell which has been obviously schematized cell which has been essentially told or promoted or programmed to divide. So as it divides it forms two new daughter cells. And those two daughter cells are in two different positions of course with respect to each other. So each cell will pick up different information from its immediate surroundings from the cells around it. It will also deliver information to those surrounding cells which will put those surrounding cells in a different context. So by this kind of binary fission you're creating a breaking of symmetry creating new information which can then be used to feedback on the process. So the whole process of building a picture plant is a highly parallel and feedback driven process. It's much more like the process of organization of a social network or an economic network or a political network is like constructing an airplane for example. So most of our conventional engineering paradigms are based on blueprints. You have an endpoint that you specify. You then have parts that you assemble to create the integral elements subsystems and final product of what you, you know, this final endpoint. In the case of our biological system we need a DNA program which is not just implemented in a cell but implemented in millions of cells and where the key is not so much what is encoded in that individual cell but how those cells interact with their neighbors dynamically across this process of emergence, building a population which is growing by division, proliferating and at the same time putting in place interactions which self reinforce and bootstrap this final process. And of course the process that we're talking about is a source of the most organized structures that we know. The human brains in this room each started from a single cell and end up with trillions of connections by virtue of this kind of bootstrapping process, this local process. And of course if you look outside the room if you look in terms of human constructs our social constructs are some of the most complicated things that we know and they're not designed as such. They're not based on a blueprint. They grow, they emerge from local interactions just like an economic network does based on local transactions and you've got this process of growth. So I think it's actually quite a deep problem and a very broad problem and I think in the case of a biological system you have the huge advantage that the kind of interactions that we can build here are based on parts which have molecular specificity. Of course the kind of biological parts that we can create are the level of molecular specificity and accuracy which is beyond almost anything else that we can machine or create and using evolved systems. This will hopefully become more clear as we go on here. And so this whole talk is about trying to get towards a process where we can engineer cell populations to create some kind of supercellular structure and as you'll see most of the talk is about microbes. It's not about plants. It's about taking systems which are as simple as possible and engineering at the lowest level DNA parts and other interactions which can come together to create more ordered structures. And so we adopt this kind of, it's a variant of the engineering model if you like where you've got a test or a build test cycle but in the case of a biological context the parts that we're using that we design are based on DNA, DNA elements that are inserted into cells by virtue of transformation. So we have a biological system which we then need to be able to interrogate and derive precise parameters from. And those parameters can then be used to parameterize computational models which can then be used to design improved circuits hopefully. And in our, broadly in the synthetic biology field there's this paradigm which has been referred to already which is I think is certainly correctly underpins the whole business of synthetic biology where you start with DNA elements in a biological context you can modularize or abstract DNA based functions as parts. These can be put together to create devices, larger scale devices, circuits and systems and implemented in a multicellular context. And these processes, this hierarchical view of systems I think works. It also comes with the kind of principles that are found in every form of human engineering the benefits of decoupling designed from fabrication and the abstraction that comes with creating standardized elements that can be used where you get the social aspect of building complicated systems by using abstraction as a way of going forward. And I think just segueing away from the problem in more precise biological terms to this more general aspect of how do you create a way of going forward here if you're trying to tackle this quite ambitious problem. And if you go back to 1958 this is a picture of Jack Kilby's first integrated circuit it's quite a famous image of the first circuit in what became Texas Instruments and five logic devices on that rather crude device. And as this is 1958 by the early 60s you had the formulation of precise physical and electronic standards for the way these kind of devices could be put together in larger scale circuits. And this diagram shows going from Jack Kilby's circuit through the first planar transistor through to the first integrated circuits and now a modern microprocessor integrated circuit will have four, five billion logic elements on it. And this process of moving from very simple architectures through to more complicated structures started of course the way we're sort of tackling things in biology which is often with hand design using direct interaction with the substrates and the design principles where you've got individuals who are designing processes where they have to know everything they have to understand the whole process of the biological components that they're using they start with raw nucleotide sequences they synthesize, they assemble this implement it in a biological system where we're certainly at this stage and certainly in the electronics systems you move very rapidly from manual design and assembly through to a computer aided design process which Patrick mentioned earlier and clearly we're in this process of trying to assemble the tools which allow us to handle complexity in a reasonable way and I think this slide in a way encapsulates both the challenge and the opportunity for physicists and mathematicians and biologists to come together in a way that I'm sure we haven't quite plumbed the depth of the kinds of interactions that are feasible here but clearly the opportunity is there but there's one thing that I want to emphasize also that the superficial similarities between say semiconductor assemblies and design and biological systems breaks down very quickly and if you think of individual circuits in an electronic context compared to the kind of circuits that exist in biological systems you have insulation and a top-down design process for electronic circuits whereas in a biological system and this could be either at the genetic scale or the cellular scale or population scale you have networks of interacting elements which are often not insulated that the kind of you know that's for example a genetic circuit inside a cell all of the components have access to all of the other components so that you require molecular specificity to derive and separate elements from each other and so for example here we can have a circuit which is essentially comprised of identical elements which are simply wired up differently and here in the biological context we need a different element for each of the logic components that you're using inside that cell and I think this is in a way well both obvious but also important but because it also highlights the potential benefits of working with cells rather than genes so if that's a genetic circuit then you're constrained and those logical elements need to be insulated by molecular specificity if you're working in a cellular context there's a natural form of insulation where cells are the unit of gene expression and the interactions between cells are necessarily more limited and it's those connections which provide the edges between these nodes in this kind of network so they're much simpler networks because of the physical arrangements of cells and this will become very clear as we go on the question is if you have a network of interactions so you carry a cell it might have 30,000 genes you want to install a circuit where you want to examine the behaviour of one of your synthetic genes in that circuit what do you do? Well of course there's a revolution in cell biology with the use of these fluorescent proteins which come from corals and jellyfish which now have quite an elaborate palette of different colours and these directly visualisable proteins of course are genes that can be used to insert into a circuit similar to a flag in a computer program so you can look at the behaviour of individual genes inside a cell and so if you imagine that well you can do that of course you can put your gene in we would commonly use look at for example if you wanted to look at a population of cells and this is a real example I'm going to show you with bacteria you can insert a single gene for a fluorescent protein you can have these population of bacteria growing inside a small container being observed in an automatic fashion in say a microplate reader where you can get multi parameter measurements very accurately continually you can get very accurate measurements of say for example increase in fluorescence or increase in the optical density of the culture as the cells proliferate the more cells there are the more light scattering there is the more the optical density goes up so you can and you can have multiple fluorescent proteins for example so you can use this to measure say for example two genes and the relationship between those genes and what they might be doing in a particular circuit as these cells are growing and one of the things that we've been trying to do in this sort of effort to try and create computational systems which create cell interactions is to get at the lowest level understand what individual parts are doing and to get intrinsic values or estimates and measurements of what say promoters are doing or other circuit elements are doing inside a cell of the context but here we're not looking at individual cells we're looking at a population of things and the process of creating fluorescent proteins schematized here so here's our gene here's fluorescence at the end of the day but of course if you're looking at a biological system like this which is about as simple as you can find in that cuvette or that well the process of getting fluorescence and you're looking at the transcription rate which is actually the intrinsic value that we want to get out of this system is what of these promoters are doing and what the rate of that transcription rate is creating in terms of amount of RNA which is then translated to make a protein precursor which then needs to be matured and folded to create the fluorescent signal that we can measure in the machine so in order to get to this point we have to also deal with various side issues these genes are in cells on chromosomes in cells the chromosomes are dividing are proliferating the cells are dividing, they're also proliferating you've got transcripts which are being made which are also being diluted by cell growth they're creating proteins which are being diluted by cell growth and there's degradation rates so we've been struggling with how to express these processes in a formal way but also in a way that allows one to make measurements that allows one to interpret the data and combines a description of the process of gene expression and production of something you can measure and allows us to get to an intrinsic value at the end here so one of the first things that we've discovered both experimentally and theoretically the product of fluorescence the thing that you're actually measuring in these devices there's a linear relationship between the rate of accumulation of the intensity and the absorbance the ratio of those two is a constant product so if you have for example a construct, a plasmid that you make with one of these fluorescent proteins on board growing inside a cell population which is then proliferating inside a cell you can see for example that over time the absorbance, that is the amount of cells in that cuvette will increase with time and then flatten off as the growth rate finishes at the same time the amount of fluorescence will also accumulate with time and it will follow a similar kind of trajectory not quite the same but similar but if you then express these two functions essentially the rate of accumulation of fluorescence and the rate of accumulation of absorbance and plot them against each other you get a straight line relationship which makes it easy to measure and this provides essentially a parameter that we call alpha which is essentially an estimate of for a particular condition what proportion of the cell's effort is going to make this fluorescent protein and the rate at which it accumulates so what is being devoted essentially to the assembly of this fluorescent product which is a reflection of the intrinsic value of the promoter that drives this expression since that's the main variable in these experiments and that gives you a view of the value for the intrinsic ability of this promoter to create or drive expression but when you start looking at these values and then change the conditions you get a lot of variation which is due to the difference in load on the cells both because you're creating some kind of output you're now using here either different carbon sources or using cytostatic agents to create different loads and the promoters themselves respond differently according to the kind of perturbation that you see inside the system and so for example the amount of variation which is due to say in this case for the media or for experimental other experimental features is that the mass majority of the noise if you like the variation in the experiments comes from extrinsic processes things which are outside and not related to the way the promoter is working but rather to the way the cells are operating the way that the cells are feeding themselves off the media or the way other extrinsic processes are moderating this process and so what we've done is to take another process where we take these values for the intrinsic value of the promoter and then have two genes now looking at a test gene with respect to a standard gene where a standard gene is a promoter here is a fixed reference point and that we can correct for other extrinsic variables by comparing the alpha values for each of these two genes so we have this factor row which now where you're looking at very different media properties you still retain these straight line relationships by using this ratio metric estimate which essentially corrects for the transcriptional load and essentially the extrinsic factors which affect cell growth as opposed to the intrinsic value of the promoter and so if you look at the variation now where you correct for these extrinsic factors you can remove a lot of the variation which is associated with this external say for example the different media which is shown up here as well is now smoothed and this is our standard promoter up here of course it's a fixed value but you can see also for the other promoters you've got this intrinsic value you can use for circuit design and so you have this relationships now which you can have a value for the intrinsic properties of the promoters which can be used for design and so when you're looking at this level of say the nanometer scale if you like where you've got genes, you've got gene elements which are interacting, gene regulatory networks you can start to develop ways of measuring parameters that can be used for modeling and the rest of this talk is really about trying to move up the scale so it's all very well to have a gene regulatory network which will operate inside a cell but then what happens after this? How do we get to a reasonable way of modeling or understanding cell behavior and then think about patting those processes on a larger scale? So you can see here the scales here which run from molecular to cellular to population run from nanometers to microns to millimeters and run from seconds to minutes to hours as a rough estimate of the kind of scale of elements that we're trying to model so in the case of the cellular scale this is an image from a confocal microscope which shows a field of bacteria so they're just simply taken out of a shaking flask put onto a microscope slide and visualized where they're yellow because we've got both of these two different colors of fluorescent proteins that we're measuring inside these cells and you can see even in this population of cells there are variants and size and color depending on the different behavior, genetic behaviors inside the systems and essentially this idea of stochasticity is embodied in this slide but there's also the spatial aspect because of course bacterial cells in many of the circumstances that we want to work with them are on a solid substrate or in some other non-homogeneous system so this shows a couple of colonies that have grown on an agar plate and have merged at the end here and you can see here the individual cells of gene expression and the physical arrangement of cells inside the colonies and in a way this is a very simple analog of what happens in plants as well so if you're thinking about bacterial cell growth you have this slight quandary because those of you who have done any microbiology will know that if you plate individual cells on a agar plate and come back the next morning that single cell will have grown to make this very regular hemispherical colony geometrically very regular but the way it gets there the way it produces that regular colony is anything but radially symmetric so for example bacterial cells like E. coli or bacillus have a bacilliform shape and they grow uniaxially they don't go in a radially symmetric way they go uniaxially so the growth of a cell is mediated by this extension in a single axis followed by septation and this septation is mediated by Turing-like systems, the min proteins which will calculate the midpoint for the cell and so you end up with a capsule-like process where these daughter cells are created and then undergo the same process again after the cells separate so if you think of modelling that and what we've done is to create physical models for this division process you can start with single cells this shows part of a few sequences in a time-lapse sequence but individual cells growing and in a friction-free environment of course if you plated a single cell like this and came back the next morning there'd be a single colony one cell wide and several metres long which would be the result of this cell just undergoing its axial extension but of course in reality you can't extend indefinitely the frictional forces build up so you generate buckling processes which are well described in many physical systems these kind of processes which are growth, growth of individual cells and as these cells grow and form these columns of cells they buckle, they buckle very quickly and they create these effectively rafts of cells and you can make out as these mini or micro colonies are growing this is a simulation at this point but we see this same thing under the microscope you get these buckling processes that create cells in opposition cells which have easier or more difficult ways to grow and you build up physical tensions or forces inside these small micro colonies which create inhomogeneities and create fractal like patterns as we'll see in a minute so this model for those who are interested it's a Github open source software package called CellModeler which is online and that's the website there, cellmodeler.org and you can create a computational model and it's based on rigid body kinetics this shows the microscopy data this is real cells growing under a microscope under a cover slip on soft agar so you get the cells growing to form these micro colonies this shows a computational model showing similar kind of process where the cells are growing and pushing against each other to create these constrained three dimensional arrangements and so you can grow colonies you can grow hundreds of thousands of cells if you want I think this one has about 50,000 cells on it and you can simply grow those processes what's more you can also mark the dynamics of the process you can visualize what happens during the growth process so if you take this little micro colony and tag after the first two divisions you start from a single cell two cells, four cells and at the four cell stage you progress through the growth process this is what those four cells have created the clones of those four cells have created this fractal-like arrangement in these early colonies because of the underpinning uniaxial growth of the cells and because of the physical dynamics and the competition between cells you end up with these different directions and nature of growth sorry I didn't hear the first bit it's a nutrient situation so these are tiny colonies I think in these cases where they've got on a rich nutrient medium we don't see any limitation of growth at this early stage these are all possible but we go to the real stuff so we go into the experimental data and we always tie together theoretical observations with direct experimental observations sure but I think at a first approximation we see similar phenotypes where we don't have nutrient limitation in the small colonies which are only an hour or two old and so for yeah sure so this shows for example an image of mature colonies after an overnight growth which have been densely seeded on a plate and we've got three different species of bacteria which have red, blue or green proteins being expressed and even here you can see where the colonies have collided early you get these fractal like boundaries being formed and the nature of confocal microscopy is that you can go and look at these quite in detail so you can zoom in very effectively identify subsets of the field and zoom in to look at the individual details and as you zoom in you can start to see individual features here and you can see the kind of behavior and arrangement of cells due to these fractal boundaries that are being produced so these are not a model this is real data, real cells and we have colonies well methods now for creating split colonies this is a model, not data so you see these fractal boundaries being produced as a result of two cells segregating a plasmid actually two plasmids, segregating two daughter cells at the first step in division and we can also do that in vivo so I think I've got the next slide but we can use not just the different colony assays but also start to use different bacterial strains so this is a mutant of E. coli it's called the rodae mutant which creates spherical cells so as a defect in the nature of cell wall growth it forms spherical cells it doesn't have the same kind of uniaxial growth and this shows a colony, a small colony growing inside some normal wild type cells which are marked in the blue here and so you can create these split colonies this is again as real data this is not a model we've got plasmids one bearing a red fluorescent protein one a green fluorescent protein which is segregating at the first point in division and the consequent daughter cells are interacting at the boundaries to create these fractal like boundaries in the case of the rodae mutant we've got these roughly spherical cells you get a much smoother boundary and see the fractal dimension is much lower in this case and this shows again colliding colonies so this is a clonal sector and these are colonies that are growing together and you see the fractal boundaries in one case and smooth boundaries in the other and again the models back this up as well as well as this you can also start to explore these models by looking at adhesion so this is a wild type micro colony this is a bacterial strain that is now expressing under a rabbinose control one of these adhering genes this is an antigen 43 which is a gene that is or protein that is exported to the outside surface of the cell and it allows aggregation between cells and this aggregation creates these kind of extended processes where the cells once they contact each other form an aggregate and then are pushed away from each other and that's both found in the physical model and in the microscopy data the experimental data so we've got all these different potential contexts and we will try to use these in fact to create a plant-like context where I mentioned plant cells where they aggregate and form an extracellular matrix and we're trying to use the combination of spherical cells and cells that adhere to each other to create more plant-like processes but these are quite difficult to work with because they're very sick cells but that was our partly our goal in this so we have these models for cellular interaction and that allows us to look at the scale of the sort of micron scale interactions between cells but if we want to move towards population-based interactions we have to move up in scale to sort of millimeter scales so we've developed this I think quite interesting system for experiments which is very simple but it's based on use of membranes which have this black stuff here which is printed onto these filters and the hydrophobic ink allows the bacterial populations to be inoculated and to grow on the hydrophilic patches between the ink and the quadrants of homogenous bacterial populations which can then interact from population to population with the geometries of the bacteria highly constrained so they can only grow within that geometrically constrained quadrant and you can generate longer distance singling interactions and we've used a number of different singling processes but focused mainly on the quorum sensing signals which is about as simple as singling system as you can make between two cells we've got these homocerein lactone producing enzymes and receiving enzymes so you've got a two enzyme system one of which is the catalyst that produces the signal the signal will diffuse across membranes and it moves directly from one cell to the next by diffusion and there's a protein that you can express a cognate protein that you can express inside the cells which will recognize this dimerize and then bind to DNA on the basis of that dimerization catalyzed by the interaction with the signaling molecule so you can start to create systems where you combine these quadrant based filters where you can inoculate different cell types onto the filter and you can then use the genetic circuits to condition the signaling across these systems so this is actually the control over here where we have cells that are receiving a signal in this case it's a particular homocerein lactone I'm not going into details but that signal then diffuses from a source across the membrane and triggers a response in the adjacent cell quadrants and these you can digitize these and so you see the signal being conditioned as it diffuses across the system and this is in these are about 20mm about 10mm from there to there across this quadrant and here we've got a similar kind of experiment where we've got the same kind of signal diffusing from one end of the filter into the other but here we're looking at a system where we've got a signal that's being responded these cells are conditioned to respond to this but there's a negative feedback interaction with an alternate signaling process here which I'll describe in a minute and it gives you a conditioned signal where you get a much sharper cutoff between the cells that are receiving the signal and there's a feedback relationship there so the kind of signals that you can make and we've got now in this diagram we've got two different signals so we've got these AHL signaling systems but now there's two of them and they're hooked up, they're connected so we've got a signal system here which is producing a signal which cannot be received by the same cell but it can be received by the other side of the circuit if you like the signaling systems call them A and B where A needs to be received by the B conditioning cells and the B conditioning cells produces the A signal so you have a relationship which is governed by propagation of a signal if you like where A can signal to B B can signal to A but they can't signal to themselves so you have a process where you can create sort of a leapfrog arrangement where this cell type will signal to those and vice versa but not to themselves so if you start out with a system like this if you've got a short range interaction these cells can interact with each other if they're close to each other because they can then interact and feedback on each other create an excited state by virtual mutual induction if they're separated that doesn't happen so here we've got essentially a checkerboard arrangement on this quadrant arrangement we've got cell types A and B across the whole field they're too far apart to excite each other whereas here we've seeded the process with a mixture of the center portion of this checkerboard these are all mixed with A and B signals and here we've got this this process amplifying and then spreading as cells signal to the adjacent quadrant adjacent quadrant produce the opposite signal which then signals to the next one along so you have this feedback regulated propagation of a signal across centimeters across the filter based on these local interactions which are then propagated across the material and of course that's using a sleepfrog based process you can also do the other instead of having mutual induction you get a mutual competition so for example if you have a system where we have now our two states A and B but now repressing each other rather than activating each other you create a very different logic where if you have cells which can be preconditioned to be in state A they will inhibit state B so they'll promote themselves and repress the opposite state similar to this one is here so the B state will repress the A state but will excite itself and this is the circuit I won't go into details of the circuit down here and all of these are visualized by virtue of these fluorescent proteins that we're using and so in this case if you have one of these filters in the nature of the system they're isogenic strains but you can either push them into state A or B by giving them the signal and so these cells are preconditioned into state A or B decorated on through the filter along with a field of cells that are unconditioned so they are neutral in their process in this particular circuit they tend towards the green state we'll call that the A state but here we're starting out with these preconditioned cells and as they are allowed to grow across the system these cells in red, the B cells will continue to propagate will have a signal that then spreads to adjacent quadrants those cells will be recruited to the B state and so you get this recruitment of cells to this state and you have this boundary formation as they're competing with cells of the opposite state and so you have this population based effect with a very simple circuit this is the kind of testbed that we want to take and use to create more complicated circuits to create the kind of systems that describe schematically up here so for example if you have an AND gate expressed in this population context so an AND gate of course you can use within a cell have two different molecular functions that can create a state as a consequence of the interaction in a population context where you've got states A and B coming together to create a new transcriptional state by virtue of an AND function you can create state C for example as a new transcriptional state and the idea that A plus B equals C plays out in a population context to create the kind of creation of a novel band of cell types which of course could then be also used to bootstrap another set of interactions based on the interaction between C and B and C and A so the whole process can reiterate itself in a way that's quite familiar to say segmentation or other patterning processes that developmental biologists would be familiar with and I think this idea that you can start to think of quite simple ways of dealing with patterning is quite interesting to us at least and one of the other implications here is that these patterns which I think it comes to the way humans deal with things this is the kind of element that you can deal with because a lot of the complexity is underneath the process so you can start to think of these being described phenotypically in a way that's hierarchical and if you start to put those simple systems together in different order you get different outcomes and this is just a very simple example here we've got two patterning processes which are completely theoretical and arbitrary so one is a radial patterning process and the other is a bilateral asymmetric process and if you take those processes A and B in different order so the idea that you can capture this idea and have a transcriptional state as a result of one type of patterning and use that to trigger the next form of patterning so here we have radial patterning where the outer state here is now undergoes bilateral patterning and here we've got the bilateral patterning followed by radial patterning where the outer state in each case is active for the next step in the process and of course you end up with very different outcomes depending on the order with which you apply those different patterning processes and so I think this is quite an interesting test bed for ideas that could be implemented in plants and of course if you think about plants you've got issues here plants are slow, the generation times are slow they've got complex genomes there's often a lot of redundancy in systems they've got diploid or polyploid genetics tissue culture and regenerations quite slow and difficult and complex tissue morphologies which makes it really quite difficult to look at early simple stages and they're quite difficult to analyse at the cellular scale so one of the themes of our work and I think it's through the microbial stuff as well is that we're interested in simpler processes and almost to the exclusion of almost anything else that would go for simple systems biological systems where you can get direct measurements and so we've chosen to go with a lower plant called Marcantia polymorpha and the lower plants that is that if you go back 500 million years the first plants that emerged into the trestle environment were not dissimilar to these lower plants which are still with us and you've got these very flat arrangements there's no shoots there's no roots, there's no flowers there's no seeds you have a very different mode of growth but they are packed with all of the genetic equipment that you see in higher plants pretty much so these are haploid they're not diploid, they're haploid like E. coli and yeast they have male and female plants this is the male plant here, this is the female plant if you cross them you end up with not flowers and seeds etc you end up with these as a product of the crossing sporangia and so they make not seeds but spores there's about a quarter of a million spores in it so you have from a single cross you can make millions of progeny those progeny look like this they look roughly like yeast cells they have all these energy containing vacuoles full of oil and they can be stored at minus 80 indefinitely if you put them on to agar media they start growing so they grow into these this is what they start out a day later or so they look like this they differentiate and they're much larger after another day or day and a half they start dividing you can see the first division here this is one of the ungeminated spores so after about a day and a half to two days this is the change already and the change accelerates so you get the formation of differentiated cell types this is what passes for a root in these lower plants, this is a rhizoid so they're just single cells and you can see this continuing cell division at the top here and as they continue growing the top part of the plant the photosynthetic plant elaborates to form a flattened sheet so after about five to seven days you have this flattened structure which will form this flattened sheet which is the body of the plant with these giant cells which are the roots of the plant underneath it and so you end up with this flattened structure which continues to grow this gives you an idea of scale so instead of having a bacterial or a yeast colony you've got a little baby plant to spread the spores and as it continues to grow you look at the top surface in detail and it has this very modulus architecture so you have repeating three-dimensional units which repeat one after the other in an identikit fashion and so each one of these units here has one of these donut-like structures this is an airport so this is the photosynthetic unit of these lower plants you could call it the equivalent of a leaf but it's separated into these small identical chambers which are spread across the top portion of the plant body and it has this very nice simple three-dimensional architecture a pour for gas exchange highly photosynthetically active cells inside a hollow chamber which undergo photosynthesis on the bottom of this green surface which has this on the top you have what passes for roots in these plants these single giant cells which emerge from the bottom surface and on the top you start to see specialized structures like this cup-like structure here you may notice that you see these small photosynthetic chambers on the top surface and inside here you can see these very unusual asexual propagules which form inside these these cup-like structures this is a cross-section of the cup-like structure so they start from single cells which grow inside these cups and create these little groups of propagules which grow to a certain size and then fix their position and you end up with a structure like this which is one of these propagules and inside you've got this cell specialization that takes place which is easily observable by microscopy but the key thing is that the process of growth is directly observable so similar to the E. coli colonies that I showed you you can observe growth this is from day two to day three of growth or germination of these asexual propagules that grow very quickly and you can use image processing techniques to take images from day two and day three here we've stretched day two image with a warp registration algorithm to make it fit over the top of day three and then you can match them overlay them and you can see that now the two images are overlaid on each other and what's different between them are the cell divisions that have taken place in the subsequent 24 hours and they show up as red lines here which you can see more clearly here so you can measure or visualize cell divisions quite clearly around one of these apical notches so you can visualize cell dynamics this way of course you can segment and quantify the process of growth and you can also map gene expression onto this and so you can see these small gemma these are some of the promoter fusion synthetic promoters that we've been using in the past and we've been using synthetic processes on top of the physical processes of growth and so we have, we think we're getting towards at least a plant-like system which embodies some of the benefits of working with simple bacterial and yeast-like processes not just with the physical and direct observation of the systems but also in terms of genetic modification and to summarize what we are aiming to do and what we think we're getting systems both for measurement of local cell properties of the way those local cell properties relate to interactions locally between cells the way gene fates and gene expression are established by local interactions and the way growth and metabolism results in tissue-wide interactions with the chemistry and physics of interaction and of course there's a feedback within this whole process where individual cell properties and say the geometry of cells are constrained by the physics of tissue arrangements, the shapes of cells are governed not just by genes themselves but their context inside the physical organism and shapes of cells we know constrain patterns of cell division therefore constrain arrangements so there's an unholy arrangement here where a lot of the complexity of the system emerges from local interactions playing out across these multi-scale linked processes and I think now with my county we have a system where we can encapsulate all of these processes in the same system in a simple way using the analytical techniques and models and I've finished there hopefully on a hopeful note and just to give thanks to the people who are doing the work these are people still in the lab and the green folk are working on plants and the ones in darker text are working on the microbial systems and the collaborators including things which I haven't talked about today which Anne alluded to in her talk which is the open plant initiative and you'll see this funny plant spanner thing around at times and that's really related to an effort to try and build open tools for engineering plants at different scales both at the genetic and the cellular scale and to have ways of distributing those tools in a more open fashion and thank you very much for your attention.