Elliot Meyerowitz (Caltech , HHMI) 1: Why we need to understand plant development

I'm Elliot Meyerowitz. I'm at the California Institute of Technology, on the faculty in the division of Biology and Biological Engineering. What I want to talk about, in three parts, is why we use plants in the laboratory, how we use the plants, and one set of things that we've learned from the plants that we learned for reasons of curiosity, but which may have implications for agriculture in the future. In this part, Part 1, I'm going to talk about the urgent need we have for learning more about plants, for a variety of reasons that have to do with human progress and human suffering, and I'm going to talk about a set of methods that we've developed for studying plants so that we can learn, in detail, how they grow and develop. And I'm also going to say a couple of things about why that's important practically, as well as to satisfy our curiosity. In Part 2, I'll then talk about the application of these methods, in our laboratory, to understand more about how the leaves and flowers of a plant are organized around its stem, what you can see in the pictures here. And in Part 3, I'll talk about some surprising aspects of the models we've developed for figuring out how it is that leaves and flowers arrange themselves around a stem that has led to a new way of looking at plant development, in which mechanical signalling is as important as chemical signalling in the communication from one cell to the other and the formation of developmental patterns. So, we'll always be talking about the formation of developmental patterns, but with a long preface, which is Part 1.

Why should we study plants? And that's what I'll start with, and then we'll get up to, how do plants use chemical signals to form developmental patterns? And in Part 2, we'll actually answer that question, in part, how do plants use chemical signals? In Part 3, we'll talk about physical signals. This is a map from the World Food Organization, which shows hunger in the world. The darker areas, here, in Africa and in Southeast Asia, back here, are places where people don't have enough to eat. Not enough calories and not enough of other nutrients, and if you can read the fine print on this map you'll see that it says that more people die of hunger every year than from AIDS, from Malaria, and Tuberculosis, combined. So, one problem that we have that relates to the study of plants is how to feed the world. Let's look at the numbers. According to the Food and Agriculture Organization of the United Nations, 805 million people suffer from chronic hunger, and this is the report of 2014. That's at an encouraging low level compared to some years in the past, but it's still almost one in eight people on Earth who don't have enough to eat. One in four children in the world are stunted by malnutrition. In the United States, we're not really familiar with the extent of hunger in the world, but it's serious. Poor nutrition, according to the World Food Program, causes nearly half of deaths in children under five years old. That's 3.1 million children each year. That's a remarkable number. And so here's the World Health Organization's conclusion on the same subject. Together, maternal and child undernutrition account for more than 10 percent of the global burden of disease. So, in a series of films that are largely medicine and the treatment of individual diseases, there's a good region to have talks about plants, such as this one, because 10 percent of all the disease in the world is caused by our inability to grow plants in the right places at the right times and the right amounts. And an enormous contribution to the health of people in the world would come from a better understanding of how to grow plants, and

of how plants grow. It's surprising, in fact, that so little attention is paid to this part of biology in our medical funding agencies. Now, why are plants so important? I don't have to tell you that it's because they're what we eat. Plants directly provide about 85% of human food. And that other 15% is almost entirely something like this buffalo that just ate a plant. That is, in the richer societies where we eat meat, we couldn't have meat either, unless there were plants growing, so that our food comes from plants. Now, right now, as we pointed out, 805 million people on Earth are not getting enough to eat in the past year. There are about 7.8 billion people on the Earth. So, the problem is bad, and it's about to get worse, because by 2040 there will be 9 billion people in the world, as you can see on this graph. That's almost 25% more than the number of people that we have on Earth now. And we have to learn to feed all of those people, not just the 805 million who were starving, but the additional 1.2 billion who will be added to the world's population by 2040, which is not very many years from now. Now, you might think that it would be easy to do this, we just expand the amount of agricultural land that's used by farmers so that we can grow more food, but in fact this isn't happening. All the agricultural land that's suitable for plant growth now is in use. Indeed, the amount of agricultural land that's used on Earth is slowly going down. Why? Because there's desertification in some parts of the world, where current agricultural land is being degraded to the point where it becomes desert. And there's urbanization in some of richest farming areas in the world, such as around Shanghai, in China, the expansion of the city is eating up the farmland. So, we can't simply expand the amount of land on which we grow the food. In fact, we have to face not only an increasing population, but a slowly declining amount of arable land that can be used for agriculture. There's another problem that's also going to make things worse, and this is climate change. We know that the climate is warming,

and the growth of our crop plants depends very strongly on the ambient temperature. This is just one graph from many studies, which shows that the change in the yield for each degree warming can be very serious. In drought management that is in dry years we could lose 40% of our crops with just a change from 22 to 24 or 26 degrees Centigrade. 2 or 3 degrees Centigrade, an increase in temperature that we have to expect, given the level of carbon dioxide in the atmosphere, so that we're... have people who are starving now, we'll have more people to feed in the future, we'll have less land with which to feed them, and the warming of the climate will reduce the yields on the land that we do have. And so we have a daunting set of challenges to grow enough food to feed the world in the future. There are other reasons we would want to learn more about how plants grow. We've mentioned climate change and the increasing carbon dioxide in the atmosphere. Plants take up the carbon dioxide in the atmosphere and replace it with oxygen, as part of their photosynthesis. In fact, plants take up a net of 60 billion tons of carbon each year via photosynthesis, and the burning of fossil fuels, the anthropogenic contribution to global warming, adds 9 billions tons. And so the plants are capable of taking out far more carbon dioxide from the atmosphere than we humans put in in excess of what happens from normal, natural processes. The carbon in plant biomass on land is about 550 billion tons. The entire atmosphere holds 800 billion tons to carbon as carbon dioxide, so the plants play a major role in carbon balance in the atmosphere, and they can play it either way. If we can encourage plants to photosynthesize more, or grow more plants on the land that we have, they'll take up net carbon and help up to reduce the effects of the excess carbon that the burning of fossil fuels causes. If, on the other hand, we denude the land of plants and the carbon that's stored in the wood and other parts of those plants is released back to the atmosphere, we'll exacerbate the problems that we're causing by burning

fossil fuels. And so plants play a major role in the carbon cycle, and to mollify the effects of human activities on the carbon cycle we also have to learn more about plants. So, what do we need? We need a better understanding of how plants grow, a better understanding of how plants interact with their environments, including the abiotic environment, that is, the non-living environment of light, temperature, water, and nutrients like nitrogen and phosphorous, which are the two that are most often lacking in agricultural land, and we also have to learn how plants interact with the biotic environment, the pathogens that kill the crops while they're growing in the fields, or that destroy the yield after the harvest. I'm going to be talking about a developmental aspect of plants, the position of which the leaves and flowers are found, which has to do with the abiotic environment, and I hope there will be other lectures in this series on the interactions of plants with pathogens and with the biotic environment. So, what is it that we don't know that we need to know, to be able to understand how the plants genome turns into the living, 3-dimensional plant? We don't know how the genetic information of the plant becomes a plant, and so we can't predictively model plant growth and changes in plant growth that occur in response to changes in the genome, or in changes in the environment. We can't predict the effects of genetic changes, and so we can't breed plants for plant size, for plant architecture, for growth rate, or yield. We can only plant plants and look at random for ones that seem better, which is a laborious and very time-consuming process. Let's make an analogy to a product of human industry that gives us a direction that we might be able to begin taking to better understand how plants grow. A Boeing 777 aircraft, which was designed in the early 1990s, and was one of the first airplanes to be designed entirely in computers, has about 130,000 different types of parts, and given that some of these parts are found in greater numbers than one, 3,000,000 total parts. That degree of complexity could be encompassed in computer-aided design programs and in the computers that existed 20 years ago.

Now, a plant cell is not of much greater complexity in terms of its part numbers than a 777 aircraft. There are 27,000 different types of proteins that are found in all the cells of the plant together, each individual cell has some smaller number, and perhaps 10,000,000 protein molecules of these 27,000 different types. So, to have a computer model of all the proteins of a plant cell, and of all the different types of proteins of a plant cell, is something that would have been possible with the computers that we had 20 years ago, and easily possible with the types of computers that we have today. Now, this analogy between airplanes and plants isn't exact. Plant cell proteins move and interact with different partners, at speeds so that they have different partners every few milliseconds, while the parts of an airplane don't move with respect to each other, except for the control knobs and dials, and if they do move it's a very bad thing. A plant cell has at least equal complexity, in terms of part numbers, as a 777 aircraft, but in linear dimension it's 10^7 times smaller, so more than 10^20 times smaller in volume. So, it's a nanopackage that contains the complexity of an airplane. And a plant cell can reproduce itself, which an airplane can't, it has to be made in a factory. And a plant cell can make its own fuel through photosynthesis, which an airplane can't do. So, there's some principles of construction that a plant cell has that an airplane doesn't have, but nonetheless the computer technology of 20 years ago was enough to encompass all the parts of something as complicated as a plant cell, though not something as dynamic as a plant cell. Computers of today are up to the task of a full understanding and description of how a plant cell works. So, we have a possibility to develop a new approach to the study of plants, one that we call, half-jokingly, computational morphodynamics, in which we access the developmental information of a plant in a dynamic way by making real-time movies on a cell by cell basis of what each cell is doing in terms of its expansion and its division, what each gene is doing in terms of its expression pattern, and

what each protein is doing in terms of its movement within the cells. Our computers can handle that information and we've developed microscope methods that enable to access that information, and if we put that into the biological equivalent of a computer-aided design program, then we have a beginning of a complete definition of the developmental program of a living plant. The example that we're going to take is the example of phyllotaxis, the pattern of leaves and flowers around a stem in a living plant. Why are we going to take that example? Because the pattern of the leaves determines the photosynthetic efficiency and therefore the yield in a crop plant, and also because the phyllotactic pattern is simply fundamentally interesting, to understand how it is that something like a pineapple or a pinecone or a sunflower can have that lovely pattern of organs, forming one after the other around its stem. How does something as stupid as a plant measure an angle one time after the other to make a spiral phyllotaxis in which each subsequent organ comes out about 130 or 140 degrees from the previous organ, one time after the other after the other, to make this spiral phyllotactic pattern. So, the practical aspect: canopy structure depends on phyllotactic pattern, and therefore the photosynthesis that can be done by a field of plants depends on the position at which the leaves are found, and that models of soybean growth, for example, in which different positions of the leaves, angles of the leaves, and reflectivity of the leaves are put in, show that if we could change those qualities of a soybean plant, we could predictively create 8.5% more productivity in a field of soybeans, and at the same time use 13% less water. So... that what we understand now about the growth of soybeans tells us that if we could change their development to change the pattern of leaves around the stem, we could increase the yields of soybeans and consequently feed people. So, there's a practical aspect to the phyllotactic question, but there's also a theoretical aspect that's been of interest to mathematicians and to

mathematically-minded biologists for a long time. How does something as dumb as a plant measure an angle of 130 to 140 degrees around its stem over and over again? They don't have protractors. The first description of phyllotaxis was the ancient Greeks. Theophrastus described it in this book that you can't read, but there's the title page, "Historia Plantarum". The word phyllotaxis originated from the 18th century, Charles Bonnet, who felt that by categorizing the phyllotactic patterns, the different phyllotactic patterns of plants, you could come to some sort of natural classification of plants, which turned out to be wrong. The first mathematical considerations and mechanistic model, the first causal model for phyllotaxis, was done by a great 19th century botanist named Hofmeister, whose picture I've shown there. And Hofmeister created a model for how a plant might generate that organ time and time again for the formation of new leaves and new flowers. Not a model that we'll discuss, because it no longer pertains. There were many experiments done on phyllotaxis, and we'll come back to the some of them in the 1930s by Robin and Mary Snow, and in the 1950s by Ian Sussex, and these experiments will come back later in this collection of talks, because the observations that they made form the basis for the computational models that I'll describe. Phyllotaxis has also been of great interest to mathematical biologists and to mathematicians because they're interested in the natural origins of patterns like this spiral phyllotactic pattern, and one that I'll point out is Alan Turing, about whom there's been a recent popular movie, as he was one of the British mathematicians who decoded the German secret codes during World War II, was working at the time of his death on trying to come to a mathematical solution to how it is that plants create phyllotactic pattern, and we'll come back to his model in Part 2 and in Part 3. So, we have an urgent need to understand how plants grow and develop because humans are starving, and more humans will starve if we don't figure it out. We have one possible mechanism by which we can learn enough about plants to change the way they grow in a predictive and

design-oriented way, which is computational morphodynamics, and I've interested the problem, one small aspect of plant growth, that's of some relevance to agriculture, that we will explore by using computational models and dynamic acquisition of information from the growing plants themselves. So, how do we study, understand, and learn to change the phyllotactic pattern? We take the problem to the laboratory, and in the laboratory we are using a plant that's called Arabidopsis thaliana. It's a member of the mustard family and although this is a small plant, that's a normally-sized hand, and one of the reasons that we use it in the laboratory is just because of it's of small size, so that even in a moderately sized laboratory like mine, we can grow something like a million plants a year, and that's very important for doing a lot of genetic experiments. It has other properties. The small size I've mentioned... it has a rapid lifecycle. You can grow the plant from seed to seed in about 8 weeks, and so the sorts of genetic experiments that we'll be discussing can be done with great facility using this plant, while with the more traditional crop plants there's one or two generations a year, and so the same amount of work takes a very much longer period of time to do. It grows very easily. It's a little weed that grows mostly throughout Europe and western Asia, also in parts of eastern Asia. It grows well under lights at room temperature. You can get about 20,000 seeds per plant; they're very fecund. And as of the year 2000, the genome was completely sequenced, so we know what the amino acid sequence of all of the proteins coded in the genome are, what all the small RNAs coded in the genome are, and we know their relatives positions to each other in the genome. So, we have an enormous amount of genetic information about this plant, and so it serves as a model system for a very large number of laboratories that study plant growth and development. And then, what do we do with Arabidopsis to find out how the phyllotactic pattern occurs, and therefore how we can change it in a predictive way? We find a way to watch the phyllotactic pattern

happen as the cells divide, as the cells grow, and create the new position for each leaf and for each flower. So, we look at the shoot apical meristem, which is where the action is, the collection of cells at the end of every shoot that's creating the primordia for leaves and flowers, and pouring out behind it the cells that will become the stem, so that the shoot apical meristem is a collection of stem cells, both literally and in the sense that the cells provide for differentiated cells in the mature organism. It forms during embryogenesis in the plant and it persists through the life of a shoot, pouring out behind it the cells that make the stem and on its sides the cells that will make leaves and flowers. How, you may ask, does this go on forever? It doesn't. Once the shoot stops growing it's already made some leaves, and in the base of each leaf a new meristem forms and that makes a branch, so that one step after the other the meristem, the primary embryonic meristem, creates a primary shoot, then there's secondary shoots, tertiary shoots... until you have the whole oak tree, all coming from the activity of a single shoot apical meristem and its progeny over long periods of time. So, we take the shoot apical meristem, where the phyllotactic action is happening, and we developed a series of microscope methods so that we could look at them in real-time and see what's actually happening as each organ forms, look at the patterns of gene expression, at the places in the cells where the proteins are found, how the cells divide, and what directions they expand before they divide, and then we created a set of computational image processing methods that enable us to put into our computers all these activities of the plants and then apply our models to the computer models in silico. So, we developed methods to watch gene action in these live images. This is a shoot apical meristem of an Arabidopsis plant at a time when it's making flowers. The plant here is... this is the shoot apical meristem in the middle. Each of these cells is about five micrometers across, very small, and those cells all look the same as each other in the shoot apical meristem,

and then surrounding it we have flowers. So, here's a young flower that's forming on the flanks of the shoot apical meristem, there's an even younger flower right there that's just beginning to form, and here's one of the older flowers that's already beginning to make its sepals. So, one after the other, the flowers are made around the meristem at this angle of about 130 to 140 degrees, and we want to know how the plant chooses the angle. Let me give you an idea of what a lovely nanomachine this is. This is a photograph of a thread going through the eye of a needle, and this little tiny thing up in the corner is an Arabidopsis shoot apical meristem at the same scale. You can pack dozens of these through the eye of a needle, and yet those meristems will make the whole Arabidopsis plant, and very similar meristems, ones that aren't much bigger, will make an entire oak tree or a redwood tree if you just let the cells keep dividing long enough. And if we could understand that we could understand how the genome of a plant translates into the above-ground architecture of a plant and all its parts and all its functions. Here's a view of the phyllotactic pattern of Arabidopsis as it forms. On the left side, it's making leaves and they're numbered from oldest to younger, and on the right side the shoot apical meristem is making flowers, later in the life of a different plant, since to take these scanning electron micrographs we had to kill the plants. So, the pattern is quite regular and occurs in the leaves and the flowers. Now, the methods we developed for finding out what's going on as the phyllotactic pattern forms are live imaging methods in which we make parts of the plant fluorescent, either gene products as proteins, or parts of the cells, as in this case where we're looking at the plasma membranes of the individual cells, and we use a laser scanning confocal microscope to access full 3-dimensional information from the top of the plant, which we've computationally projected forward in this image and then stitched together one observation after the other that were made every two and a half hours for several days in an athletic event of considerable difficulty by a former postdoc, Venu Reddy. And you can see the division of every single cell

occurring on the surface of the plant here, and we can leaf through to see the cells below. The formation of the floral primordia, for example, this flower right here, labeled P1, starts out as a tiny primordium and the bulges out to make the sort of primordia that we saw on the scanning electron micrographs on the last slide. As a result of its establishment as a very young primordium, at this point in the flanks of the shoot apical meristem, and then rapid cell divisions, more rapid than those in the meristem, so that it begins to bulge out, create a separate primordium, and grow away from the shoot apical meristem. So, we can see that happening in this time lapse image, which is sped up about 3,000 times. Now, the shoot apical meristem, as I showed it to you, looked quite homogenous. All those cells in the meristem looked the same as each other in both of the types of the micrographs that I showed, but they're really not the same. Things are really a lot more complicated than that, and just to demonstrate that I'm showing here three different laser scanning confocal microscope pictures in which, up here, we're looking at expression of a gene called CUP-SHAPED COTYLEDONS 2, which has been stained with a yellow fluorescent protein, its promoter expresses a yellow fluorescent protein in those cells where the CUP-SHAPED COTYLEDONS protein 2 is normally expressed, and we've false-colored that red, and the place where the gene is expressed is in the boundary region between a flower primordium and the meristem. So, those cells are different from the others. This is a picture where the green is a different gene called KANADI, which is being expressed... KANADI 1... which is expressed in front of the primordia in those boundary regions, but also around the back of each primordium as it begins to form, then later jumps up to the front, so it's a very dynamic pattern of gene expression. It's different from CUP-SHAPED COTYLEDONS 2. Here are some other patterns of expression that we're showing. The red is SHOOT MERISTEMLESS, a different gene that's expressed in a completely different pattern in the meristem, and if we add together all of these different patterns of gene expression that we see in

the shoot apical meristem, there are dozens and perhaps hundreds of different types of cells in the shoot apical meristem, despite its uniform appearance. So, it's a very complicated object, but we have the methods to see where the genes are expressed and the methods to see how the dynamics of that gene expression changes as we intervene, genetically and in other ways, with the shoot apical meristem, so that we can begin to develop hypotheses of how it is that the meristem chooses the position for its new flower and leaf primordium, one time after the other, in this spiral phyllotactic pattern. So, this is where we end Part 1. Tune in for Part 2 and we'll say a little bit more about the phyllotactic pattern, a little bit more about the methods, and we'll show you how we use the methods to develop a theory of phyllotactic pattern that's tested and gives us predictive capability so that we can change the genotype of the plant in known ways, and change the pattern in which the leaves and in which the flowers come out around the plant. So, stay tuned for Part 2, and let me just finish by acknowledging my sources of funding in my laboratory, in particular the Howard Hughes Medical Institute and the Gordon and Betty Moore Foundation, who have been very generous, the Department of Energy, whose funding has led entirely to what will be Part 3, the Gatsby Foundation in England, and the National Institutes of Health, all of whom have made it possible for my laboratory to develop the methods I've talked about, to take the approach we've discussed, and to create the hypotheses and models that we'll talk about in the next two episodes. Video: https://www.youtube.com/watch?v=IyqCAQQ3xJk

Thanks iBiology

   

ibiology.org Facebook Twitter Google+