A Conversation With Systems Biologist, Philip Benfey, PhD
Developmental Dynamics: What initially provoked your interest in root development?
Philip Benfey: My scientific interests have always been on questions of developmental biology. My graduate work was with Phil Leder, and I was involved in cloning the IgE receptor as well as other genes involved in the allergic response. When it came time to look for a post-doc, I knew that I wanted to work on a developmental system. I also was looking for a genetically tractable organism. I was originally attracted to plants in large part due to the ease of introducing genes into the genome. The transition to plant research was quite straightforward. DNA is DNA, although it took me a while to master some of the terminology. I remember referring to the vascular cell types at one point as the “xylem and phylem” as opposed to “xylem and phloem.”
One of my post-doctoral projects with Nam-Hai Chua involved characterizing a complex promoter, the CaMV 35S promoter. My approach was to cut it up into subdomains and fuse each of them to a reporter gene. While analyzing the cellular-specificity of the expression patterns conferred by the subdomains, I came to appreciate the simplicity of root development.
Root development originates with stem cells at the root tip whose progeny remain in cell files emanating from specific stem cells (Fig. 2B). Thus the age and provenance of every cell is easily ascertained. Moreover, all developmental stages are present in the root at any time. One further simplification—the outer four cell layers are organized as concentric circles around the central vasculature. Taken together, these features reduce the developmental parameters from four dimensions to two dimensions
Dev Dyn: What is your lab's research focus today?
PB: The focus of our work is on root development, primarily in Arabidopsis. We use a combination of molecular genetics, genomics and systems biology approaches to address fundamental questions in developmental biology. We began our work on root development with a genetic screen for short or fat root mutants. This was really a matter of expediency as it was easier to screen for these than for long or thin mutants. Among the first mutants we characterized was one we not very imaginatively called “short root”. In addition to the feature for which it was named, it turned out to be entirely missing one of the four cell layers that surround the root vascular cylinder. Shortly after finding short-root we found another mutant missing a cell layer. By that time we became a bit more whimsical and called it “scarecrow” after the Wizard of Oz character who was missing a brain. We cloned both genes and found that they encode related transcription factors (Di Laurenzio et al.,1996; Helariutta et al.,2000).
The big surprise came when we performed expression studies. The SHORT-ROOT mRNA is made only in the vascular tissue, but its protein is found in the vascular tissue and the adjacent endodermal cell layer. Although this was not the first instance of a moving transcription factor in plants, it was the first case in which movement was closely tied to function. When SHORT-ROOT moves from the stele it physically interacts with SCARECROW in the endodermis. This interaction sequesters SHORT-ROOT in the nucleus and prevents further movement (Cui et al.,2007). The interaction also results in the activation of downstream target genes. To identify and characterize the downstream targets we have been employing genomics approaches. What is becoming clear is that SHORT-ROOT and SCARECROW not only interact with each other, but they also regulate a large number of transcription factors, which in turn regulate other transcription factors. Mapping these gene regulatory networks is a major aim of our current work.
Dev Dyn: What is your approach to mapping gene regulatory networks?
PB: Given the simplicity of the root, we realized that if we could profile expression in each of the cell types as well as sample all the developmental stages we could get a fairly complete picture of gene expression in an entire organ (Brady et al.,2007). To profile the cell types, we used GFP lines that were specific to individual cell lineages. We enzymatically digested the cell walls then passed the cells through a fluorescence activated cell sorter (FACS). We used the RNA from the sorted cells to hybridize to Affymetrix microarrays. Somewhat surprisingly we found that when the entire procedure was performed rapidly there was little change in the transcriptional profiles induced by these manipulations. For the developmental stages, we microdissected individual roots into 13 sections and used that material for microarray analysis. Recently, we have been able to use a mathematical approach to combine these datasets allowing us to assign expression levels to each of the cell types at different stages of development. We are now performing ChIP/chip and ChIP/seq in an effort to reconstruct the transcriptional networks that link the stem cells at the tip of the root to their differentiated progeny,
As we identify the regulatory networks, we are making use of the nearly complete library of insertional mutants in Arabidopsis as well as our extensive collection of tissue-specific promoters to determine the effects on cell specification, or loss-of-function and gain-of-function of key genes in the network.
Dev Dyn: Which papers have most impacted your research?
PB: (Coen and Meyerowitz,1991) This study describes a remarkably elegant model for how floral organs are specified. It was derived entirely from genetics and has, by and large, stood the test of time.
(Lee et al.,2002) One of the first papers to combine microarray expression analysis with ChIP/chip, they were able to identify a large number of known and many unknown connections between transcription factors and target genes. This was a key early paper in the field of systems biology.
Dev Dyn: How might your work impact animal developmental biology research?
PB: It is important to realize that development in multicellular organisms only evolved twice. By comparing plant and animal development one can compare and contrast mechanisms and begin to understand the underlying constraints. How many different ways are there to solve a particular biological problem? If we see the same solution being used in independently evolved organisms, this could point to a small number or perhaps even a single possible solution. Understanding why the search space of solutions is limited could reveal fundamental constants and/or laws.
More specifically, our work points to the need to perform genome-wide expression analyses at the level of individual cell types. We have shown that response to environmental stimuli can differ dramatically among cell types within an organ. In our initial study, we subjected plants to salt stress and to growth on iron deficient media. We then sorted cells and analyzed expression in different cell types. In both cases we found a remarkable amount of cell-type specific response to these environmental stimuli. Most of these responses had been missed in previous analyses, presumably because they were diluted out when the assay was performed on the entire plant or on an entire organ such as a leaf or root. There may well be a corollary within animals that goes beyond the obvious cellular specializations such as nerve cells.
Dev Dyn: What are some important questions in your field that remain to be answered?
PB: A question that we are particularly interested in is, “What happens as cells progress along a developmental pathway?” Are there abrupt changes in regulatory networks that mark developmental changes or is development more a series of incremental steps?
Dev Dyn: What exciting ideas are emerging in your field?
PB: Plants are playing a leading role in efforts to link phenotype and genotype. It is striking that we still have such a poor ability to predict phenotype from genotype, except in the case of a small number of Mendelian mutants. What we lack is a good knowledge of how different alleles interact to produce traits. In humans, most of the work is directed toward understanding disease states. In plants we can focus primarily on adaptive traits. Plants also have the advantage of being nonmobile and we can control their interactions both among themselves and with the environment. With high throughput automated phenotyping and next-generation sequencing there is a real opportunity to use natural variation to understand the regulatory networks controlling plant development. The basic idea is that by understanding the way different alleles interact to refine traits will provide insight into the way regulatory networks are tuned.
Dev Dyn: Do you think there is a lack of communication between plant and animal biologists? If so, what are your ideas for bridging the gap?
PB: I think this varies greatly. There certainly are some animal scientists who feel that they can ignore findings in plant biology. This is less true for those working on animal model organisms. With new imperatives of food and energy security, it seems all the more short sighted for animal biologists to think that results from plants are not relevant. At the same time, I think it is incumbent upon plant scientists to be as clear as possible in describing our work so that it is comprehensible to scientists from all backgrounds.
I would note that we have an ongoing and very pleasant collaboration with Olivier Pourquie whose aim is to compare developmental processes in plant root and vertebrate tail development. In particular we are comparing the repetitive processes that result in somites in the tail and lateral root formation in the root. To date, we have found some very intriguing similarities that suggest that there might be biological constraints that have led to the use of similar processes in both cases.
A Conversation With Regeneration Biologist, Kenneth Birnbaum, Ph.D.
Dev Dyn: What is your lab's research focus?
Kenneth Birnbaum: We are interested in the question of self-organization during regeneration and what gives certain cells the ability to reshape their fates during regeneration. We recently used genomic techniques together with live imaging and genetics to show that the root tip can regenerate without an active stem cell niche (Sena et al.,2009). We also used cell-specific profiling techniques to examine the genetic circuitry of specific cell types exhibiting developmental plasticity (Gifford et al.,2008).
Dev Dyn: What initially provoked your interest in this field?
KB: In my graduate work, I worked on the population genetics of avocados. My work now is quite different, but it all evolved in a few logical steps. My shift to Arabidopsis came from a gradual realization during graduate work that a powerful system was needed for the question in which I was most interested—What was the connection between the molecular attributes (which were simply markers for me at the time) and the tangible traits that gave a plant its specific attributes? Arabidopsis was just the most powerful system. If there was one “a-ha” moment in that transition, I would probably attribute it to my first glimpse of organisms expressing green fluorescent protein (GFP). These images really demonstrated to me that it was becoming possible to make the direct, live connection between the molecular level and the morphology of the organism. It is an incredibly versatile tool and remains an important part of our current arsenal in my lab. After my avocado work, I then shifted gears, doing my postdoc with Philip Benfey and using GFP techniques to take a genomics approach to development.
My interest in regeneration grew out of that moment when you start your own lab and realize you can work on anything you want. I recall one influential conversation with a physicist colleague who posed the question simply, what is so special about plant development? Regeneration was a good fit on many levels.
Regeneration captures many of the central questions in development similar to the problem of creating patterns during embryogenesis. But, in addition, unlike embryogenesis or adult organogenesis in plants, this pattern reestablishes itself from an unpredictable starting point, typically a damaged piece of tissue. So, regeneration is a great experimental system to examine the feedback mechanisms behind a self-organizing developmental system that can adapt to a different cellular layout or body plan for each “repair job.” Plants are particularly adept at regeneration so they are a nice system to unravel some of the basic principles that guide regeneration.
I also felt we had some new tools to attack this classic question. Those came largely from the work we had done in isolating specific cells for global transcriptional profiling. The premise was that a global view of regeneration at the cellular level would not only point us to the genetic players but also reveal some fundamental properties about the order of events and the return of cell identities during regeneration. The Arabidopsis root is also a nice system for imaging because of its transparency, highly ordered cellular arrangement, and array of available GFP markers. This gives us an opportunity to examine the coordination between molecular and morphological events at high resolution during regeneration.
Dev Dyn: Which papers have most impacted your research?
KB: (Skoog and Miller,1957). This was seminal work on a few levels. First, it showed one mechanism by which plants can create their basic polarity, roots vs. shoots, by varying a ratio between the phytohormones auxin and cytokinin. In a sense, this was an early illustration of how organisms use combinatorial signals to lock in specific developmental trajectories. Skoog and Miller are not directly credited with demonstrating the totipotency of plant cells but their finding was one of the critical pieces of knowledge that enabled several other groups to accomplish that feat and I would argue one of the most critical puzzle pieces of the decades long quest.
(van den Berg et al.,1997) and (Kidner et al.,2000). I put these two together because they use relatively simple tools, cell ablation and lineage marking, to provide firm evidence on the role of positional information in the root where the highly organized tissue patterning makes these techniques particularly informative. From my perspective, we are still exploring the implications of the van den Berg et al. experiments, which provided evidence of a stem cell niche in which quiescent center cells prevented differentiation of surrounding cells. This was important, timely work that continues to frame our view of root patterning and growth.
(Benkova et al.,2003) and (Nakajima et al.,2001). These papers provided critical examples on the ground rules for tissue organization at different scales. On a global scale, the Friml lab, in a series of elegant papers, has essentially revealed the plant's hidden transport system for auxin, which feeds back to shape its own transport and influences tissue development. On a local scale, the Nakajima et al. paper was the definitive work that showed how plants use the movement of transcription factors to establish pattern by communicating cell fate decisions from an inner to an outer tissue.
Dev Dyn: Your group recently reported that Arabidopsis organ regeneration can occur without a functional stem cell niche. You also show that young, but not older stage leaves can regenerate. Do you think there are dispersed specialized cells that are called to action? Or rather does any young cell have the capacity to fill the void?
KB: Other groups have shown that, in mature tissue, certain pericycle cells, which exhibit stem-cell like properties during lateral root formation, are the likely origin of callus, which is a blastema-like structure capable of regenerating the entire plant. So, in regeneration from callus, pericycle does seem to act like a stem cell reserve that can regenerate organs at both poles. In our leaf and root experiments, it looks like any young cell has the capacity to reshape its fate according to the demands of the new, regenerating organ coordinate system. This may tell us that part of the plant's high capacity to regenerate may be tied to the pluripotency of at least two classes of cells, those that are poised to exhibit developmental plasticity in response to the environment like pericycle cells and those that are developmentally young or kept that way like the meristematic cells in our experiments.
Dev Dyn: Are there certain molecular mechanisms in place that might enable young organs to regenerate?
KB: This is a critical question for us that goes to the core of what enables some tissues to regenerate and others not. Many developmental transitions occur close to the zone where cells lose their competence to regenerate a root tip. The number of dividing cells decreases along a sharp gradient as cells enter endoreduplication cycles and begin to display morphological signs of differentiation, including rapid elongation. We also have indirect evidence that chromatin structure is also changing along this gradient. And, recently, there has been some nice work in the root showing the role of hormones, in particular cytokinin, auxin, and ethylene in the transition from dividing to elongating cells. Interestingly, different treatments can affect different aspects of the developmental transition. We are trying to disentangle which of these different developmental transitions is most closely correlated with pluripotency and the competence to regenerate. We are hoping that this will give us clues to the specific phenomena, including signaling mechanisms that control a cell's ability to alter its fate.
Dev Dyn: Where do you go from here?
KB: Beside our work on the mechanisms that regulate competency in young cells, much of our efforts are focused on a better understanding of the mechanisms that restore pattern and order to the regenerating root. We believe that our demonstration of regeneration without an active stem cell niche told us something important about the source of signals that generate new patterns during regeneration and perhaps normal adult organogenesis. We would have thought previously that cell identities were specified at the reconstituted niche and maintained by other mechanisms. Our thinking has now shifted to the possibility that patterning mechanisms could in fact be controlled by a less centralized and more dispersed organizing system. There may not be a distinction between establishment and maintenance of pattern but rather one system that elaborates the entire pattern of the root, including the stem cell niche, at once. This is analogous to the difference between a perpetuating mechanism like a pebble thrown in a pond from which a ripple pattern emanates vs. a template mechanism like a stencil that elaborates the complete pattern at one time. In the template model, the stem cell niche, which we still believe is a critical regulator of growth, is still just one part of the entire pattern. It's a hypothesis that we want to address.
So, to me, this changes the kind of signals we may be looking for. One straightforward way to look for those signals is among the factors induced in the early stages of regeneration in our global analysis of regenerating roots. Some have roles in embryonic patterning and this makes them good candidates. We are also developing new techniques to ask about the role of cell–cell communication during regeneration, not just within the stem cell niche but among the tissues of the root. Regeneration looks like a process of sequential refinement so we hypothesize that there is extensive communication between tissues during the process. We are also adapting techniques that assay how the global transcriptional status of one cell type depends upon other cell types during regeneration.
We are also developing more high throughput methods to test gene function given the high level of redundancy of plants. One area where we are currently at the exploration phase is in cell-based assays that we can adapt to report on critical events during regeneration.
Dev Dyn: How might your work impact animal developmental biology research?
KB: Since plants and animals seem to have evolved multicellularity separately, the mechanisms of regeneration are likely to be quite divergent. However, I think there are two fertile areas where mechanisms or organizing principles may be shared across Kingdoms (Birnbaum and Sanchez Alvarado,2008). First, regeneration calls on some basic cellular machinery like reactivation of the cell cycle and the likely restructuring of chromatin in order to reshape cell fates. These are among the mechanisms most conserved in Eukaryotes. For example, we know already that the homolog of RETINOBLASTOMA in plants has an analogous role to that in humans, regulating developmental transitions and cell cycle control. What's also intriguing to me is the potential for common organizing principles in regeneration. In a general sense, developing tissues may have certain design constraints that force plants and animals to converge on similar solutions. For example, in general tissue growth, the existence of a stem cell niche to control the growth of tissue appears to be a convergent characteristic that operates across Kingdoms. In regeneration, the task of repatterning an existing organ from a damaged piece of tissue may be another problem whose solution is dictated by some shared constraints. Our questions about the existence of a perpetuating vs. template mechanism may be relevant in certain regeneration systems in animals. One can think of some basic mechanical problems in regeneration that are common across Kingdoms like sensing the edges of the new damaged tissue and re-establishing tissue symmetry. I think that if we keep an open mind and not always look to specific mechanisms in heterologous systems, we could benefit from thinking about how models of regulation and design principles from one Kingdom fit in the other Kingdom. This might give us ideas about how we design experiments in regeneration and interpret mutant phenotypes.
Dev Dyn: What are some important questions in your field that remain to be answered?
KB: I think some critical questions in development are centered around knowing the details of the dynamic trajectories of specific cells during development. That is especially true of cells that switch their fates, and intermediate states of regenerating cells. Some motivating questions concern what happens inside the cell to allow this transition from one fate to another to take place. Do these cells pass through states that resemble stem cells in some ways?
Another set of questions revolves around the process of repatterning damaged tissue. It has become apparent that we have an increasing ability to control the pluripotency of cells in plants and animals. Our next and perhaps more challenging task is to understand how to invoke the developmental programs of specific organs or tissues, without completely reverting back to embryonic states. In plants, the Skoog and Miller experiments showed how hormone ratios could control the master regulatory switches for the two poles of the plants. A great deal of elegant work in the past 20 years has uncovered critical aspects of pattern formation in the root. And, we do have some mechanisms that link hormones to patterning circuits. But, I believe that one of our challenges is to provide models of how we get from hormone signaling to the backbone circuitry that can lay out an entire organ.
Dev Dyn: What exciting ideas are emerging in your field?
KB: First, the unfolding details of auxin transport, the ability of auxin to trigger specific cell fates, and its ability to shape its own transport has all the trappings of at least one major component of the self-organizing system shaping organ development in plants. There are several models of how auxin can elicit cell fate changes, for example, as a morphogen or a threshold trigger. These ideas have the potential to provide a mechanism that connects broad cues to local cell fate decisions.
Second, in the field of stem cells and regeneration, recent work that has shown the ability of tissues and organ-like structures to form outside of the classical niche. This work has parallels to our own system. These experiments could help us think broadly about the question of where the patterning information for specific organs comes from.
Third, I find intriguing models that examine cell identity as trajectories defined by expression of all genes in cell. In these models, there is the potential that cells have some freedom to embark on different fate trajectories but their potential is limited to specific paths. In one manifestation of these ideas, cell fates adhere to models of dynamic systems where small fluctuations can influence their ultimate fate or, in terms of dynamic systems, attractor state (Kashiwagi et al.,2006). I am still not entirely sure how well cells undergoing development will fit these models but I believe we will gain a new perspective by examining cell fate on a global level and asking whether the total contents of a cell are relevant to its potential. For example, cells may be able to reach certain states through multiple developmental routes and thus the concept of master regulator may be context dependent.
Dev Dyn: Do you think there is a lack of communication between plant and animal biologists?
KB: I think we would benefit from more communication between plant and animal biologists, largely because I find personally that I can make the most progress on tough problems by approaching them from a new perspective. It has been said by others that plants and animals represent nature's two independent experiments in multicellularity. The comparison of these two experiments represents a great opportunity to think about how developmental systems are designed as we dissect their mechanics. Our fields have influenced each other but our techniques and approaches have also developed independently to some extent. I think that those light bulb ideas often come by combining little tricks or even big concepts from other systems with familiar ideas to create entirely new approaches.
I think the most natural way to increase exposure to each other's ideas is through our current avenues of sharing information through meetings and journals. Such mixing already happens at many meetings and I know firsthand that journal editors often invite plant–animal comparative reviews. But we could probably increase these opportunities. Editors of broader journals are a critical resource for meaningful comparisons because they have an excellent view of the breadth of research in a field among many different model organisms. They are probably the first ones to recognize common themes across fields. In the specialty journals, even those specifically geared toward plants or animals could invite more comparative reviews on perhaps very specific topics, which the wider interest journals might not be likely to pursue.3
Figure 3. Left: Philip N. Benfey, PhD, Paul Kramer Distinguished Professor of Biology and Director, Duke Center for Systems Biology, Biology Department and Institute for Genome Science and Policy, Duke University, North Carolina. Right: Kenneth D. Birnbaum, PhD, Assistant Professor, Biology Department and Center for Genomics and Systems Biology, New York University, New York.
Download figure to PowerPoint