INTRODUCTION, FIGURES, CAVEATS, FUTURE EXPERIMENTS
Perhaps one of the most distinct and exciting aspects of the “genomic revolution” is the sheer quantity of data available to analyzed and distilled. Eric Davidson, in this review paper, synthesizes a plethora of embryological development data into a concise and thorough review of the known and previously unknown developmental pathways of the sea urchin, S.purpuratus. Perhaps the most resounding theme elicited from the paper is the dynamic nature of transcriptional regulatory elements. The charts and, most strikingly, the circuit diagrams, display a tightly integrated system that we can only assume is the result of “painstaking” evolutionary adaptations that have withstood the test of development and continued the success of the species.
Davidson begins the paper with an anecdotal discussion of what causes “cats to beget cats and fish to beget fish.” His ultimate focus, then, is the requisite “hardwiring” of development. Biologists have always been fascinated by the transition of single cell to organized adult. Traditional developmental and molecular techniques have yielded case studies of gene regulation and in depth information on specific systems within the cacophony of development. But, by using computational software and experimental cis-regulatory analysis of transcription factors and their target sites, functional relationships can be resolved and analyzed.
These relationships determine cell specification and, ultimately, define body plan. The prototype for Davidson’s work is endo16, a gene which encodes a “large poly functional protein that is secreted into the lumen of the embryonic and larval midgut.” The varying expression of endo16 determines the differentiation between endoderm and mesoderm, and, subsequently, gut and midgut. Summarily, they determined that nine different transcription factors regulate this single gene, and, are themselves, regulated by other developmental and maternal factors. But, endo16 is only a single gene within this network. With voluminous sea urchin development biology in hand, they chose their next step: to determine the “..network of regulatory events of endomesoderm specification during the first 24 hours…” of sea urchin development.
Figure 1 outlines the playing field for all of us sea urchin development non-experts. It shows four time points (10h, 15h, 24h, and 55h) during embryonic morphogenesis. The progression of cell lineages and their relative locations are made clear here. This figure can become an important reference point during future dissection of the circuit diagrams and perturbation analysis. Below figure 1, table 1 serves a similar function. It is a chronological outline of the sea urchin development process from maternal cues through invagination. The table summarizes the highlights of development on the inter- and intra-cellular level that, we will soon see, are determined by a network of transcription factors.
Figure 2 is Davidson group’s original experimental perturbation analysis. Interestingly, it is located within a review paper which normally summarizes only previous work, but, appropriately, confirms some of the information contained within Table and Figure 1. Part A shows the effect of MASO, morpholino antisense oligonucleotide, that binds to the leader sequence of pmar1 mRNA compared to a control of a fusion protein coding for the pmar1 transcription factor (sense instead of antisense) and GFP. As the fluorescence and microscopy images portray, MASO effectively prevents translation of the pmar1 mRNA by binding to the leader sequence. Part B shows the effect of Krox1, a known obligate repressor, and a control fusion protein that consists of the DNA binding domain of the Krox1 transcription factor and Drosophila Engrailed repressor domain, which we expect to bind to the DNA sequence appropriately, but, functionally, not repress transcription of the target genes And, in fact, the fusion protein does not repress as shown in the right image, while Krox 1 does repress the target genes appropriately and allow for normal development of the embryo, as shown in the image on the left. Part C is a similar perturbation. The right image shows the effect of blocking B-catenin nuclearization, by using an antisense mRNA, causing “endomesodermal specification [to be] completely wiped out. Part D continues this perturbation analysis by converting a positive transcription factor into a negative transcription factor by altering the N receptor. The effect of this alteration is limited to the development of the veg2 cell line which is greatly diminished. Generally, these four sets of images show a few of the experimental effects represented in the network analysis.
Table 2 summarizes the results of a gene discovery effort, undertaken to “…clothe with real genes the armature of interactions implied by the embryology…” Or, in plainer terms, they were trying to make sure that they weren’t missing any genes before they started building a large network, only to then discover two genes that dramatically alter the picture and implicate a phenomenal waste of effort and time. The gene discovery effort used a macroarray screening process that attempted to isolate very rare transcripts that could potentially be involved in regulation. Using the “Driver” (lacking sequence) and “Selectate” (sequence of interest) methodology, it appears they were able to subtractively hybridize out about four regulatory genes (36, 19, 39. 37) that can now be included in the upcoming network diagram.
Figure 3 is just plain scary – to the untrained eye that is. To the trained, ambitious eye, figure 3 is a marvel. I have one of each eye, and with trembling and trepidation embarked on the dissection of the circuits. The elaboration of every interaction would neither be efficient or practical, so the next logical presentation of the information is to describe how they constructed these circuits graphically. Key questions when building a diagram like this are: How do I convey a wealth of biological information on a 2-D space? What sort of arrows and relationships can be shown and how dynamic can the map be without being confusing? In my opinion, Davidson does a good job with this circuit. It is a relatively simple task to select a single gene and trace the interactions and “governing” transcription factors by simply following the legend with accuracy and intelligence. The network displayed here is the “view from the genome” of endomesoderm specification. The technical information conveyed in the circuit can be categorized into several categories: chronology (top to bottom), spatial orientation (color and middle), function/structure (color-coding), downstream differentiation genes (at bottom), activation/repression (solid/barred), maternal and zygotic influences (upper left), inferred relationships (dotted lines) and possible biochemical cytoplasmic interactions (big ovals.) In short, the circuit diagram is both a story and a map. It serves to direct, summarize and explain, which, in conjunction with experimental evidence, sheds light into the black box of genes turning on and off throughout development.
Figure 4 is a smaller slice of figure 3, namely, the top tier and left column, or, initial events in endomesoderm specification, combined with in situ hybridization evidence of genes described in this network. Part A is the specific network interactions, while B-G depict related interactions in situ. One of the more fascinating interactions, which is described in the text as a “surprise”, is the initiation of the repressive subnetworks in both the micromeres (pmar1) and veg2 endomesoderm (krox, krl). These subnetworks are turned on to stabilize differentiation of these areas by preventing additional transcriptional activity. Essentially, once cell lineage has been activated by these three genes, repressors are turned on, so their activation can’t be “undone” by any stray transcriptional factors. What a great biological idea! It makes very good evolutionary sense to definitively and permanently define cell types and disallow intercellular messages to disrupt the intra-cellular (or intra-cell type) game plan. Figure 5 continues to depict this sort of regulation via “lock-down devices.” These interactions demonstrate not only the highly complex regulatory networks, but, also, the amazing degree of stabilization “effort” made during development. It seems that turning genes on and off is not sufficient even with 9 transcription factors. It is also important to control and monitor these interactions within localized areas. Cell types become committed and can’t whimsically alter their expression patterns. The entire regulatory network thematically concentrates on forcing differentiation to occur appropriately and permanently. It is safe to say that this sort of regulatory network will, ultimately, produce very biologically successful individuals.
In the conclusion of the paper this statement catches my eye: "It seems no more possible to understand development from an informational point of view without unraveling the underlying regulatory networks than to understand where protein sequence comes from without knowing about the triplet code." Boom. Davidson is on to something here. Dissecting regulatory networks in this manner is the beginning of a new field. Endomesodermal differentiation in the sea urchin is a prototype. Imagine the implications of similar discoveries in genes governing homo sapiens' intelligence. Imagine the evolutionary implications of comparative gene network studies. Imagine the gene therapy opportunities of discovering tumor suppressor gene regulatory networks on this scale. The story is not over yet.
Yet, with all of the actual and potential glory woven between these circuits, it is important to preserve the perspective of critical observer. Of course, one of the more obvious cautions is to remember that hypothetical inferences into genomic data always ought to be backed up by more traditional molecular techniques to definitively convince most readers of their validity. Davidson does a good job using perturbation analysis and in situ hybridization to verify some aspects of the network, but, does not (and could not be expected to) display experimental evidence for all elements of the network.
Of more specific curiosity is the lack of detailed methods describing the differential gene discovery screens. We are told that, "Several screens were carried out (Table 2) in which endomesoderm specification was perturbed so as to generate material for use with a very sensitive hybridization technology designed for use with large-scale arrays of ~105 clone cDNA libraries." The analysis of this data was performed by a new software, BioArray, but it is still unclear to me how the screens were done or what sort of sensitive hybridization technology might be employed. The references, 3419, 36, and 37 are in press or preparation, but we are essentially asked to take this methodology on face value. Also, on a biological level, the regulatory value of very rarely expressed mRNA transcripts is questionable. Davidson et al. should be complimented on attempting to be as thorough as possible, but we must remember that rarely expressed genes are just that, and might/ought to have less startling implications than major, previously defined players.
A potential point of confusion within the paper are the implications of perturbation analysis and cis-regulatory analysis. On page 162, the authors discuss which experiments can distinguish between direct and indirect effects and which cannot (because of immediate and downstream effects.) Summarily, traditional perturbation analysis cannot distinguish direct/indirect effects, whereas cis-regulatory analysis can. However, rescue of a perturbation effect can confirm direct effects. The authors, though, are relatively clear about these distinctions and state, "Where a rescue experiment indicates an indirect effect, or where the effect must be indirect because the affected and the perturbed gene are expressed in different cells or at different times, the implied relationships are omitted from the network models." This is important because, although indirect effects could have potentially interesting biological implications, they do not belong in the models because they decrease both the accuracy, validity, and readability of the already complex models.
On a final note, many downstream effects still remain in the proverbial
"black box." As the authors state, "...but for many of the
regulatory gene the downstream targets are still unknown." (1675)
This reminds us that the work here, even in this well-defined and highly-studied
system is not complete. Thus, any future work on other regulatory
networks, especially on less-studied systems will have a long, steep up-hill
battle ahead of them.
FUTURE EXPERIMENTS AND IMPLICATIONS
The preceding paragraph is a great lead-in for this section and, generally, defines the next logical area of study: clarifying downstream effects. Additionally, along this same vein, dissecting known indirect effects into direct effects would further bolster this regulatory network. Combining traditional molecular and developmental laboratory techniques and the three new computational software packages, Netbuilder, FamilyRelations, and BioArray, should prove fruitful in completing the network in the coming months and years.
In a slightly different arena, the discovery of "repressive subnetworks" that stabilize cell differentiation and developmental processes is quite interesting. A search for similar repressive subnetworks in other organisms would prove exceptionally interesting. To my knowledge, this is the first concrete example of this specific type of sub-regulation and developmental stabilization. Because it is such a good "biological (adaptive) idea", it is assumedly used in many, if not most, organisms; verification of this theory would prove quite helpful.
In my opinion, the most fascinating implications for future experimentation revolve around one of the authors' concluding statements, "It will be necessary to consider regulatory gene networks as evolutionary palimpsests -- patterns of regulatory interactions that are successively overlain with new regulatory patterns." (1677) The traditional genetic approach to evolution is to use a single gene, normally ribosomal of mitochondrial, determine its rate of mutation through time, and determine phylogenetic relationships amongst species according to single gene relatedness. But, if two different evolutionary biologists use different genes, they will sometimes get similar results, but, often, they return very disparate trees. Ultimately, the dissection of many regulatory networks in different species will provide a "bird's eye view" of evolutionary processes. Factors such as timing, variation amongst transcription factors, spatial orientation, and regulatory subnetworks might provide genomic tools for evolutionary biologists in the coming years.
Return to the Davidson College Biology Home Page
Return to the Molecular Biology Home Page
Return to Amber's molecular home page
email the author with questions or comments: amhartman@davidson.edu
View other web pages designed by this student