*This page is part of undergraduate assignment for Molecular Biology at Davidson College
BIO 306: Molecular Biology Course Webpage
A Genomic Regulatory Network for Development. Science 295 (2002).
Eric H. Davidson, Jonathan P. Rast, Paola Oliveri, Andrew Ransick, Cristina Calestani, Chiou-Hwa Yuh, Takuya Minokawa, Gabriele Amore, Veronica Hinman, Ce´sar Arenas-Mena, Ochan Otim, C. Titus Brown, Carolina B. Livi, Pei Yun Lee, Roger Revilla, Alistair G. Rust, Zheng jun Pan, Maria J. Schilstra, Peter J. C. Clarke, Maria I. Arnone, Lee Rowen, R. Andrew Cameron, David R. McClay, Leroy Hood, Hamid Bolouri.
Reviewed by Elizabeth Shafer.
Davidson et.al’s paper, “A Genomic Regulatory Network for Development” appeared in March 2002 issue of Science Magazine which focused on systems biology. This issue focused on systems biology because the field is becoming increasing popular and plausible with recent development of high throughput genomic technologies. Unlike more classical genetics, molecular, and developmental biology’s deconstructionist approaches to understand cellular processes, systems biology attempts to integrate information gathered in past experiments with new genomics expression data through the use computational mathematics to construct a fluid model of gene - protein interactions in a cell.
In this particular investigation, Davidson’s group wants to elucidate the genes involved in early (first 24 hours) mesoderm and endoderm development in sea urchins. Sea urchins are model organisms that have been traditionally studied by developmental biologists in an effort to understand embryogenesis, gastrulation, and tissue differentiation. Thus, a lot of literature exists for Davidson’s group to cite and build from. Davidson’s group was interested in two main questions: 1) What genes are involved in the process of early mesoderm/endoderm development? and 2) How is the unidirectional nature of this development maintained? Their goal was to create of web of interactions that would yield insight into these questions and serve to frame future questions and investigations.
The group describes two possible “views” presented by networks: one from the genome which includes all inputs and outputs of cis-regulating elements and genes in all cells at all times and one from the nucleus which includes only the inputs and outputs taking place in the nucleus at one particular time. The group felt that genes encoding transcription factors and cis-regulatory elements in the nucleus of different precursor cells formed the “heart” of communication in development networks. They purposed that each transcription factor and cis-regulatory element in a network incites expression/repression of another gene thereby initiating an unfolding web. In order to create a model of this network, they needed to determine which genes are expressed during each stage of the process and what genes underlie different morphological results from perturbation manipulations. To do this they incorporated the results of experimental tests such as perturbation analysis and cis-regulatory analysis with genomic data and computational strategies.
The progenitor cells of mesoderm originate in the vegetal pole of the embryo. Figure 1 is a traditional time elapse illustration of various structural stages of the embryo’s development. The figure illustrates the movement and changed morphology of veg1 precursors, veg2 precursors, micromere precursor, and skelegenic cells from 10 hours to 55 hours after fertilization. Table 1 summarizes previous knowledge involving steps along the development process and the genes that are attributed to them. Table 1 lists events in a linear, chronological order and few gene interactions are denoted.
As in most areas of science, the abnormal defines that which is normal. The goal of perturbation analysis is to chemically or experimentally manipulate gene expression in order to observe abnormal morphology and hypothesize the possible molecular or genetic mechanisms underlying the observation. They preformed four kinds of perturbations: introduction of Pmar1 morpholino antiisense, injection of krox-1 and engrailed fusion protein encoding mRNA, incorporation of mRNA coding for the intracellular domain of cadherin, and injection of mRNA encoding the extracellular domain of the N receptor. The perturbations produced a lack of translation of mRNA, conversion of a transcription factor into a dominant repressor, blocked Wnt/Tcf signaling, and Notch signaling in each experiment respectively. In each perturbation experiment, experimental embryos were subjected to quantitative real-time fluorescence polymerize chain reaction to detect underlying gene expression. The results are listed at this web site: http://www.its.caltech.edu/~mirsky/qpcr.htm. To attribute a change in a particular gene’s expression level as a result of perturbation, the quantity of the gene transcript detected by QPCR had to be 3 fold less or 3 fold greater than the control.
QPCR is a useful tool at this step to determine the genes underlying morphological changes, but it has its drawbacks. QPCR is beneficial because it is in real time and can detect quantitative changes over time. Additionally, it is multiplex and can detect multiple sequences at once from one source, however, each sequence most have its own unique fluorescence tag. Thus the limiting factor in the number of genes detected is the number of fluorescence tags available. The genes that QPCR can detect are also limited by the need for primers. Since researchers only have a limited number of dyes that they can use in each QPCR reaction, it is likely that they will not screen with the ORFs from an entire cDNA library and will preferentially use annotated genes or genes already implicated in the network. In this experiment they reported results of 43 genes. They do not include which primers they used or how they choose the primers they used. Since this information is the basis for the developmental network proposed in this paper, it is important to note that this was not a genome wide screen, and there is a possibility there are more genes involved in the pathway than reported in this network.
To address the issue of excluding genes previously not associated with endoderm and mesoderm development, they utilized subtractive hybridization on macroarrays to recover possible new regulatory genes. They report recovering new genes, but they do not highlight which genes they recovered and the results are reported to be in press.
As mentioned in this paper, perturbation analysis is unable to distinguish between direct and indirect effects. In order to determine direct interactions, cis-regulatory analysis or rescue experiments are required. This lab employed a computational species comparative methods of cis-regulatory analysis which involved a computer program that compared BAC composed of genes known to be involved in the mesoendoderm development process of S.Purpuratus (Sea Urchin) to Lytchinus variegatus (divergent cousin) looking for similarities that could represent cis-regulatory elements and score positively on gene transfer tests. They also were able to draw experimental based conclusions about cis-regulation from the exploratory macroarrays mentioned above.
Figure 3 for all practical purposes is a flow chart, integrating data derived from the perturbation experiments, in situ hybridizations exploring location and chronology of gene expression, QPCR, cis-regulation analysis, and previous information on embryology to model interactions involved in endomesoderm development. The purpose of the figure is to the reveal the complexity of these interactions, explain why certain genes are expressed at certain times, illuminate the striking autonomy of each regulatory gene, and expose the predominance of transcription factors among the genes represented.
The challenge of representing this model stems form the number of genes involved, the multiple effects of many genes, and the desire to indicate both where and how the interaction information was discovered experimentally and the desire to include any additional and relevant information. To examine the figure it is important to realize that they genes are grouped by the cells that they are expressed (colored blocks) in and can be represented two or more times. There is a correlation between a genes location relative to the top and the bottom of the figure and the time in which the gene is initially expressed, but it is not an exact hierarchy of events. Certain genes are sections of the network are discussed in the text, but others are ignored. Perhaps, the presentation of this figure could be greatly improved by the addition of a table that briefly annotates each gene. The description of symbols in the figure legend could be clarified by the inclusion of a legend similar to those provided for road maps in which the symbol is directly beside the description. Additionally, the clutter of the figure could be greatly reduced if the figure was not on paper but in an interactive form. Dialog boxes that appear when you mouse over a gene or a line indicating a pathway could supply information about the gene, the interaction or any information represented by arrow symbols, ovals, circles, or hatch marks in this figure. A moveable or three-dimensional model would be helpful as well.
Having elucidated a number of genes involved in endomesoderm development with the aforementioned experiments, the last objective of the researches was to understand why development is always progressive. Thus, they split the data from Figure 3 into two successive stages.
Figure 4 features the maternal and early interactions network from Figure 3 and the micromere development network in conjunction with in situ hybridizations of select genes in cells subjected to over expression of pmar1 mRNA or pmar-gfp fusion protein mRNA. The portion of the schematic network includes genes that are expressed first in the developmental pathway. The corresponding in situ hybridization results stain for genes expressed over time in the skelegenic pathway. Both the maternal/initial transcription sequence and the micromere sequence are initiated by activation of a control network and a repressor network. The control networks start a signal cascade that leads to differentiation; while the repressor networks yield stabilization of initiation events in endomesodermal cells and prevent the expression of endomesodermal specific genes in non-endomesoderm cells. The krl repressor pathway involved in the initial zygotic transcription response was previously accepted. The repressive action of pmar1, however, is new to this study. Thus, they have included the in situ hybridization data to reveal that global expression of pmar1 yields global endomesoderm.
Figure 5 features the developmental pathways of the endomesoderm from 7-12 hours and from 20-24 hours. Part A, highlights three pathways the Wnt8/Tcf loop, zygotic auto- and cross- regulation pathways, and the N- signal transduction of the gcm gene pathway. They describe these pathways as “lock-down functions” whose positive feedback loops and auto-regulation yield an expanded, differentiated endomesodermal state. The pathways illuminated in Figure 5B represent the downstream input of the pathways described in 4A. From the figure it is evident that the initial pathways are sufficient to incite every downstream gene and fully activate the endomesodermal regulatory system. They report in text that at this point regulatory inputs are stabilized, which they are unable to fully represent in the schematic network.
Figure 4 and 5 are included to explain the unidirectional movement of development. The evidence of auto-regulation and repression in the early zygotic network strongly supports this argument by eliminating the possibility of backflow of signals through the system. The stabilization of regulatory factors is also strong evidence for the system ability to combat regression, because it at least implies that expression of regulatory factors will not fall below a threshold necessary for further endomesodermal differentiation.
In their conclusion, Davidson’s group admits the difficulty of representing network data based on multiple data sources. They emphasize the importance of interdisciplinary affiliations. Advances in computational algorithms would improve the capabilities of integration software. Representation of network circuits could be improved by including more informative keys explaining symbols, listing a “gene bank” which includes a brief description of the genes included, or creating an interactive, moveable model.
They point to the other features involved in this network, such as multi-component systems involved in the biochemical and signal transduction pathways underlying the cis-regulations. The data reported in the network figure outlined mainly transcription factors that influenced other transcription factors but are they able to transcribe other genes? To answer this question, DNA microarrays of total genomic cDNA libraries of embryos with experimental conditions including the over expression of the regulatory factors described could elucidate what other genes they could be transcribing or repressing.
Since the genes included in these networks are involved in early development, as time passes they will no longer be necessary. For the majority of the sea urchin’s life these genes will be inactivated. For future study it would interesting to investigate the how these genes are inactivated. Is the process gradual? Is it mediated by downstream regulatory genes that repress their expression? Do epigenetic factors contribute to their complete inactivation? To address these questions, it would be beneficial to cluster expression data of these genes with the expression of genes involved later in development. In those genes with converse expression, functional test such as perturbations, immunoprecipiations, or FRET could be preformed to determine if they actually interact. Methylation of chromatid surrounding different genes during the different stages of development could provide clues to epigenetic regulation.
Of course the ultimate goal of systems biology and research such as this investigation is to model the interactions of all parts of the cell and amongst cells in an entire organism over a period of time. Until enough experimental information is complied and computation techniques improved, smaller networks such as the one presented in this paper can serve as insight to improve systems approaches and to provide a framework for experimental investigations.
Created by: Elizabeth Shafer. Email questions to firstname.lastname@example.org